diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,82393 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.7234775947912677,
+  "eval_steps": 250,
+  "global_step": 4500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00038299502106472615,
+      "grad_norm": 58.36933438464531,
+      "learning_rate": 5e-06,
+      "loss": 0.327,
+      "num_input_tokens_seen": 173192,
+      "step": 1
+    },
+    {
+      "epoch": 0.00038299502106472615,
+      "loss": 0.3826669752597809,
+      "loss_ce": 0.00034278002567589283,
+      "loss_iou": 1.0884798765182495,
+      "loss_num": 0.3828125,
+      "loss_xval": 0.3828125,
+      "num_input_tokens_seen": 173192,
+      "step": 1
+    },
+    {
+      "epoch": 0.0007659900421294523,
+      "grad_norm": 421.0828358936704,
+      "learning_rate": 5e-06,
+      "loss": 37.2345,
+      "num_input_tokens_seen": 346432,
+      "step": 2
+    },
+    {
+      "epoch": 0.0007659900421294523,
+      "loss": 37.343849182128906,
+      "loss_ce": 9.83613426797092e-05,
+      "loss_iou": 1.5366404056549072,
+      "loss_num": 37.25,
+      "loss_xval": 37.25,
+      "num_input_tokens_seen": 346432,
+      "step": 2
+    },
+    {
+      "epoch": 0.0011489850631941786,
+      "grad_norm": 246.44322210329707,
+      "learning_rate": 5e-06,
+      "loss": 14.7033,
+      "num_input_tokens_seen": 519784,
+      "step": 3
+    },
+    {
+      "epoch": 0.0011489850631941786,
+      "loss": 14.8048095703125,
+      "loss_ce": 0.00012145288201281801,
+      "loss_iou": 12.648301124572754,
+      "loss_num": 14.8125,
+      "loss_xval": 14.8125,
+      "num_input_tokens_seen": 519784,
+      "step": 3
+    },
+    {
+      "epoch": 0.0015319800842589046,
+      "grad_norm": 294.36104428006075,
+      "learning_rate": 5e-06,
+      "loss": 23.9066,
+      "num_input_tokens_seen": 692808,
+      "step": 4
+    },
+    {
+      "epoch": 0.0015319800842589046,
+      "loss": 23.937803268432617,
+      "loss_ce": 0.00030406011501327157,
+      "loss_iou": 1.5096111297607422,
+      "loss_num": 24.0,
+      "loss_xval": 24.0,
+      "num_input_tokens_seen": 692808,
+      "step": 4
+    },
+    {
+      "epoch": 0.0019149751053236309,
+      "grad_norm": 202.47763715173787,
+      "learning_rate": 5e-06,
+      "loss": 10.7347,
+      "num_input_tokens_seen": 865640,
+      "step": 5
+    },
+    {
+      "epoch": 0.0019149751053236309,
+      "eval_websight_new_CIoU": -0.01316814310848713,
+      "eval_websight_new_GIoU": 0.03720420226454735,
+      "eval_websight_new_IoU": 0.03720420226454735,
+      "eval_websight_new_MAE_all": 1.3634408712387085,
+      "eval_websight_new_MAE_h": 2.1734297275543213,
+      "eval_websight_new_MAE_w": 2.1817232370376587,
+      "eval_websight_new_MAE_x": 0.6177258789539337,
+      "eval_websight_new_MAE_y": 0.4808841943740845,
+      "eval_websight_new_NUM_probability": 0.9870634377002716,
+      "eval_websight_new_inside_bbox": 0.0,
+      "eval_websight_new_loss": 14.947561264038086,
+      "eval_websight_new_loss_ce": 6.12214753346052e-05,
+      "eval_websight_new_loss_iou": 1.5935487747192383,
+      "eval_websight_new_loss_num": 14.93359375,
+      "eval_websight_new_loss_xval": 14.93359375,
+      "eval_websight_new_runtime": 62.2711,
+      "eval_websight_new_samples_per_second": 0.803,
+      "eval_websight_new_steps_per_second": 0.032,
+      "num_input_tokens_seen": 865640,
+      "step": 5
+    },
+    {
+      "epoch": 0.0019149751053236309,
+      "eval_seeclick_CIoU": -0.019112106412649155,
+      "eval_seeclick_GIoU": 0.04438253119587898,
+      "eval_seeclick_IoU": 0.044382523745298386,
+      "eval_seeclick_MAE_all": 1.3827024102210999,
+      "eval_seeclick_MAE_h": 2.2133243083953857,
+      "eval_seeclick_MAE_w": 2.1218671798706055,
+      "eval_seeclick_MAE_x": 0.6495624482631683,
+      "eval_seeclick_MAE_y": 0.5460554659366608,
+      "eval_seeclick_NUM_probability": 0.9863913357257843,
+      "eval_seeclick_inside_bbox": 0.07465277798473835,
+      "eval_seeclick_loss": 14.71789836883545,
+      "eval_seeclick_loss_ce": 0.0003928400110453367,
+      "eval_seeclick_loss_iou": 1.6963666677474976,
+      "eval_seeclick_loss_num": 14.65625,
+      "eval_seeclick_loss_xval": 14.65625,
+      "eval_seeclick_runtime": 84.2849,
+      "eval_seeclick_samples_per_second": 0.593,
+      "eval_seeclick_steps_per_second": 0.024,
+      "num_input_tokens_seen": 865640,
+      "step": 5
+    },
+    {
+      "epoch": 0.0019149751053236309,
+      "eval_icons_CIoU": -0.009676441550254822,
+      "eval_icons_GIoU": 0.03315560147166252,
+      "eval_icons_IoU": 0.03315560333430767,
+      "eval_icons_MAE_all": 1.353821575641632,
+      "eval_icons_MAE_h": 2.2173551321029663,
+      "eval_icons_MAE_w": 2.198903441429138,
+      "eval_icons_MAE_x": 0.49197526276111603,
+      "eval_icons_MAE_y": 0.5070522427558899,
+      "eval_icons_NUM_probability": 0.9866805374622345,
+      "eval_icons_inside_bbox": 0.0711805559694767,
+      "eval_icons_loss": 15.255043983459473,
+      "eval_icons_loss_ce": 4.263626760803163e-05,
+      "eval_icons_loss_iou": 1.6046354174613953,
+      "eval_icons_loss_num": 15.28515625,
+      "eval_icons_loss_xval": 15.28515625,
+      "eval_icons_runtime": 82.1924,
+      "eval_icons_samples_per_second": 0.608,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 865640,
+      "step": 5
+    },
+    {
+      "epoch": 0.0019149751053236309,
+      "loss": 15.343791961669922,
+      "loss_ce": 4.252552753314376e-05,
+      "loss_iou": 1.606989860534668,
+      "loss_num": 15.375,
+      "loss_xval": 15.375,
+      "num_input_tokens_seen": 865640,
+      "step": 5
+    },
+    {
+      "epoch": 0.002297970126388357,
+      "grad_norm": 188.94230579429873,
+      "learning_rate": 5e-06,
+      "loss": 14.5862,
+      "num_input_tokens_seen": 1038128,
+      "step": 6
+    },
+    {
+      "epoch": 0.002297970126388357,
+      "loss": 14.539365768432617,
+      "loss_ce": 0.00030374148627743125,
+      "loss_iou": 1.5851291418075562,
+      "loss_num": 14.5625,
+      "loss_xval": 14.5625,
+      "num_input_tokens_seen": 1038128,
+      "step": 6
+    },
+    {
+      "epoch": 0.002680965147453083,
+      "grad_norm": 168.09641651439844,
+      "learning_rate": 5e-06,
+      "loss": 8.2737,
+      "num_input_tokens_seen": 1210752,
+      "step": 7
+    },
+    {
+      "epoch": 0.002680965147453083,
+      "loss": 8.371365547180176,
+      "loss_ce": 0.00027169095119461417,
+      "loss_iou": 81.91307067871094,
+      "loss_num": 8.375,
+      "loss_xval": 8.375,
+      "num_input_tokens_seen": 1210752,
+      "step": 7
+    },
+    {
+      "epoch": 0.003063960168517809,
+      "grad_norm": 140.32501556176152,
+      "learning_rate": 5e-06,
+      "loss": 11.1605,
+      "num_input_tokens_seen": 1383168,
+      "step": 8
+    },
+    {
+      "epoch": 0.003063960168517809,
+      "loss": 11.00023078918457,
+      "loss_ce": 0.00023003615206107497,
+      "loss_iou": 1.6515984535217285,
+      "loss_num": 11.0,
+      "loss_xval": 11.0,
+      "num_input_tokens_seen": 1383168,
+      "step": 8
+    },
+    {
+      "epoch": 0.0034469551895825352,
+      "grad_norm": 129.22621630179205,
+      "learning_rate": 5e-06,
+      "loss": 6.6702,
+      "num_input_tokens_seen": 1556192,
+      "step": 9
+    },
+    {
+      "epoch": 0.0034469551895825352,
+      "loss": 6.598052978515625,
+      "loss_ce": 0.000396652176277712,
+      "loss_iou": 58.22084045410156,
+      "loss_num": 6.59375,
+      "loss_xval": 6.59375,
+      "num_input_tokens_seen": 1556192,
+      "step": 9
+    },
+    {
+      "epoch": 0.0038299502106472617,
+      "grad_norm": 143.99423695445878,
+      "learning_rate": 5e-06,
+      "loss": 12.387,
+      "num_input_tokens_seen": 1729120,
+      "step": 10
+    },
+    {
+      "epoch": 0.0038299502106472617,
+      "loss": 12.289302825927734,
+      "loss_ce": 0.00023921528190840036,
+      "loss_iou": 1.606602668762207,
+      "loss_num": 12.3125,
+      "loss_xval": 12.3125,
+      "num_input_tokens_seen": 1729120,
+      "step": 10
+    },
+    {
+      "epoch": 0.004212945231711987,
+      "grad_norm": 108.40687466795202,
+      "learning_rate": 5e-06,
+      "loss": 5.6898,
+      "num_input_tokens_seen": 1901856,
+      "step": 11
+    },
+    {
+      "epoch": 0.004212945231711987,
+      "loss": 5.812960147857666,
+      "loss_ce": 0.0004599654348567128,
+      "loss_iou": 259.53076171875,
+      "loss_num": 5.8125,
+      "loss_xval": 5.8125,
+      "num_input_tokens_seen": 1901856,
+      "step": 11
+    },
+    {
+      "epoch": 0.004595940252776714,
+      "grad_norm": 131.46512738561623,
+      "learning_rate": 5e-06,
+      "loss": 12.4495,
+      "num_input_tokens_seen": 2074432,
+      "step": 12
+    },
+    {
+      "epoch": 0.004595940252776714,
+      "loss": 12.453392028808594,
+      "loss_ce": 0.00026683963369578123,
+      "loss_iou": 1.5887775421142578,
+      "loss_num": 12.4375,
+      "loss_xval": 12.4375,
+      "num_input_tokens_seen": 2074432,
+      "step": 12
+    },
+    {
+      "epoch": 0.00497893527384144,
+      "grad_norm": 101.21069895303762,
+      "learning_rate": 5e-06,
+      "loss": 5.1644,
+      "num_input_tokens_seen": 2247256,
+      "step": 13
+    },
+    {
+      "epoch": 0.00497893527384144,
+      "loss": 5.449373245239258,
+      "loss_ce": 0.00015416568203363568,
+      "loss_iou": 17.10838508605957,
+      "loss_num": 5.4375,
+      "loss_xval": 5.4375,
+      "num_input_tokens_seen": 2247256,
+      "step": 13
+    },
+    {
+      "epoch": 0.005361930294906166,
+      "grad_norm": 109.310493071197,
+      "learning_rate": 5e-06,
+      "loss": 11.2854,
+      "num_input_tokens_seen": 2420488,
+      "step": 14
+    },
+    {
+      "epoch": 0.005361930294906166,
+      "loss": 11.437763214111328,
+      "loss_ce": 0.00026362822973169386,
+      "loss_iou": 1.66493821144104,
+      "loss_num": 11.4375,
+      "loss_xval": 11.4375,
+      "num_input_tokens_seen": 2420488,
+      "step": 14
+    },
+    {
+      "epoch": 0.005744925315970892,
+      "grad_norm": 94.34650368787317,
+      "learning_rate": 5e-06,
+      "loss": 4.4436,
+      "num_input_tokens_seen": 2593328,
+      "step": 15
+    },
+    {
+      "epoch": 0.005744925315970892,
+      "loss": 4.543169021606445,
+      "loss_ce": 0.00020011767628602684,
+      "loss_iou": 6.623126029968262,
+      "loss_num": 4.53125,
+      "loss_xval": 4.53125,
+      "num_input_tokens_seen": 2593328,
+      "step": 15
+    },
+    {
+      "epoch": 0.006127920337035618,
+      "grad_norm": 100.17025206843569,
+      "learning_rate": 5e-06,
+      "loss": 10.3714,
+      "num_input_tokens_seen": 2765696,
+      "step": 16
+    },
+    {
+      "epoch": 0.006127920337035618,
+      "loss": 10.398763656616211,
+      "loss_ce": 0.0003265046398155391,
+      "loss_iou": 1.6537723541259766,
+      "loss_num": 10.375,
+      "loss_xval": 10.375,
+      "num_input_tokens_seen": 2765696,
+      "step": 16
+    },
+    {
+      "epoch": 0.0065109153581003444,
+      "grad_norm": 92.14093040650044,
+      "learning_rate": 5e-06,
+      "loss": 4.4866,
+      "num_input_tokens_seen": 2938840,
+      "step": 17
+    },
+    {
+      "epoch": 0.0065109153581003444,
+      "loss": 4.25418758392334,
+      "loss_ce": 0.00028116017347201705,
+      "loss_iou": 8.455916404724121,
+      "loss_num": 4.25,
+      "loss_xval": 4.25,
+      "num_input_tokens_seen": 2938840,
+      "step": 17
+    },
+    {
+      "epoch": 0.0068939103791650705,
+      "grad_norm": 83.76771747909497,
+      "learning_rate": 5e-06,
+      "loss": 9.6722,
+      "num_input_tokens_seen": 3111656,
+      "step": 18
+    },
+    {
+      "epoch": 0.0068939103791650705,
+      "loss": 9.726861953735352,
+      "loss_ce": 0.00029987277230247855,
+      "loss_iou": 1.6638808250427246,
+      "loss_num": 9.75,
+      "loss_xval": 9.75,
+      "num_input_tokens_seen": 3111656,
+      "step": 18
+    },
+    {
+      "epoch": 0.007276905400229797,
+      "grad_norm": 85.59127019381263,
+      "learning_rate": 5e-06,
+      "loss": 4.392,
+      "num_input_tokens_seen": 3284496,
+      "step": 19
+    },
+    {
+      "epoch": 0.007276905400229797,
+      "loss": 4.572667121887207,
+      "loss_ce": 0.0004015713930130005,
+      "loss_iou": 23.15438461303711,
+      "loss_num": 4.5625,
+      "loss_xval": 4.5625,
+      "num_input_tokens_seen": 3284496,
+      "step": 19
+    },
+    {
+      "epoch": 0.0076599004212945234,
+      "grad_norm": 78.4184602471564,
+      "learning_rate": 5e-06,
+      "loss": 7.9458,
+      "num_input_tokens_seen": 3457816,
+      "step": 20
+    },
+    {
+      "epoch": 0.0076599004212945234,
+      "loss": 7.8091044425964355,
+      "loss_ce": 0.0005108587210997939,
+      "loss_iou": 1.90053391456604,
+      "loss_num": 7.8125,
+      "loss_xval": 7.8125,
+      "num_input_tokens_seen": 3457816,
+      "step": 20
+    },
+    {
+      "epoch": 0.00804289544235925,
+      "grad_norm": 88.46632333266166,
+      "learning_rate": 5e-06,
+      "loss": 3.1119,
+      "num_input_tokens_seen": 3631000,
+      "step": 21
+    },
+    {
+      "epoch": 0.00804289544235925,
+      "loss": 2.914778232574463,
+      "loss_ce": 0.000715868838597089,
+      "loss_iou": 20.56275177001953,
+      "loss_num": 2.90625,
+      "loss_xval": 2.90625,
+      "num_input_tokens_seen": 3631000,
+      "step": 21
+    },
+    {
+      "epoch": 0.008425890463423975,
+      "grad_norm": 73.58581565716081,
+      "learning_rate": 5e-06,
+      "loss": 6.5807,
+      "num_input_tokens_seen": 3803952,
+      "step": 22
+    },
+    {
+      "epoch": 0.008425890463423975,
+      "loss": 6.504635810852051,
+      "loss_ce": 0.0007294907118193805,
+      "loss_iou": 1.8562748432159424,
+      "loss_num": 6.5,
+      "loss_xval": 6.5,
+      "num_input_tokens_seen": 3803952,
+      "step": 22
+    },
+    {
+      "epoch": 0.008808885484488702,
+      "grad_norm": 40.484161003235144,
+      "learning_rate": 5e-06,
+      "loss": 2.4782,
+      "num_input_tokens_seen": 3976616,
+      "step": 23
+    },
+    {
+      "epoch": 0.008808885484488702,
+      "loss": 2.4908900260925293,
+      "loss_ce": 0.0006555955624207854,
+      "loss_iou": 37.24055099487305,
+      "loss_num": 2.484375,
+      "loss_xval": 2.484375,
+      "num_input_tokens_seen": 3976616,
+      "step": 23
+    },
+    {
+      "epoch": 0.009191880505553428,
+      "grad_norm": 97.22230887683381,
+      "learning_rate": 5e-06,
+      "loss": 2.0994,
+      "num_input_tokens_seen": 4149200,
+      "step": 24
+    },
+    {
+      "epoch": 0.009191880505553428,
+      "loss": 2.0976505279541016,
+      "loss_ce": 0.000971070141531527,
+      "loss_iou": 13.649656295776367,
+      "loss_num": 2.09375,
+      "loss_xval": 2.09375,
+      "num_input_tokens_seen": 4149200,
+      "step": 24
+    },
+    {
+      "epoch": 0.009574875526618154,
+      "grad_norm": 154.48061606660323,
+      "learning_rate": 5e-06,
+      "loss": 7.5555,
+      "num_input_tokens_seen": 4321712,
+      "step": 25
+    },
+    {
+      "epoch": 0.009574875526618154,
+      "loss": 7.571127891540527,
+      "loss_ce": 0.0008151677320711315,
+      "loss_iou": 13.631052017211914,
+      "loss_num": 7.5625,
+      "loss_xval": 7.5625,
+      "num_input_tokens_seen": 4321712,
+      "step": 25
+    },
+    {
+      "epoch": 0.00995787054768288,
+      "grad_norm": 29.090531216489648,
+      "learning_rate": 5e-06,
+      "loss": 1.6815,
+      "num_input_tokens_seen": 4494408,
+      "step": 26
+    },
+    {
+      "epoch": 0.00995787054768288,
+      "loss": 1.7185226678848267,
+      "loss_ce": 0.0007492044242098927,
+      "loss_iou": 4.346151828765869,
+      "loss_num": 1.71875,
+      "loss_xval": 1.71875,
+      "num_input_tokens_seen": 4494408,
+      "step": 26
+    },
+    {
+      "epoch": 0.010340865568747606,
+      "grad_norm": 100.30542578014845,
+      "learning_rate": 5e-06,
+      "loss": 1.7095,
+      "num_input_tokens_seen": 4667032,
+      "step": 27
+    },
+    {
+      "epoch": 0.010340865568747606,
+      "loss": 1.786179780960083,
+      "loss_ce": 0.0010234997607767582,
+      "loss_iou": 1.342555284500122,
+      "loss_num": 1.78125,
+      "loss_xval": 1.78125,
+      "num_input_tokens_seen": 4667032,
+      "step": 27
+    },
+    {
+      "epoch": 0.010723860589812333,
+      "grad_norm": 123.5520679984825,
+      "learning_rate": 5e-06,
+      "loss": 5.1258,
+      "num_input_tokens_seen": 4839960,
+      "step": 28
+    },
+    {
+      "epoch": 0.010723860589812333,
+      "loss": 5.078887939453125,
+      "loss_ce": 0.0007630588952451944,
+      "loss_iou": 4.0871148109436035,
+      "loss_num": 5.0625,
+      "loss_xval": 5.0625,
+      "num_input_tokens_seen": 4839960,
+      "step": 28
+    },
+    {
+      "epoch": 0.011106855610877058,
+      "grad_norm": 75.49799926725528,
+      "learning_rate": 5e-06,
+      "loss": 2.2301,
+      "num_input_tokens_seen": 5012824,
+      "step": 29
+    },
+    {
+      "epoch": 0.011106855610877058,
+      "loss": 2.1779701709747314,
+      "loss_ce": 0.0012123838532716036,
+      "loss_iou": 6.067341327667236,
+      "loss_num": 2.171875,
+      "loss_xval": 2.171875,
+      "num_input_tokens_seen": 5012824,
+      "step": 29
+    },
+    {
+      "epoch": 0.011489850631941785,
+      "grad_norm": 69.37472894493497,
+      "learning_rate": 5e-06,
+      "loss": 4.0042,
+      "num_input_tokens_seen": 5186096,
+      "step": 30
+    },
+    {
+      "epoch": 0.011489850631941785,
+      "loss": 4.1497392654418945,
+      "loss_ce": 0.001301644602790475,
+      "loss_iou": 4.804612636566162,
+      "loss_num": 4.15625,
+      "loss_xval": 4.15625,
+      "num_input_tokens_seen": 5186096,
+      "step": 30
+    },
+    {
+      "epoch": 0.011872845653006512,
+      "grad_norm": 70.67664919604134,
+      "learning_rate": 5e-06,
+      "loss": 1.655,
+      "num_input_tokens_seen": 5358576,
+      "step": 31
+    },
+    {
+      "epoch": 0.011872845653006512,
+      "loss": 1.7472056150436401,
+      "loss_ce": 0.0011118631809949875,
+      "loss_iou": 2.0371646881103516,
+      "loss_num": 1.75,
+      "loss_xval": 1.75,
+      "num_input_tokens_seen": 5358576,
+      "step": 31
+    },
+    {
+      "epoch": 0.012255840674071237,
+      "grad_norm": 123.2471229403419,
+      "learning_rate": 5e-06,
+      "loss": 4.8823,
+      "num_input_tokens_seen": 5531816,
+      "step": 32
+    },
+    {
+      "epoch": 0.012255840674071237,
+      "loss": 4.876434326171875,
+      "loss_ce": 0.001434716279618442,
+      "loss_iou": 3.3636035919189453,
+      "loss_num": 4.875,
+      "loss_xval": 4.875,
+      "num_input_tokens_seen": 5531816,
+      "step": 32
+    },
+    {
+      "epoch": 0.012638835695135964,
+      "grad_norm": 103.57924655765518,
+      "learning_rate": 5e-06,
+      "loss": 2.1105,
+      "num_input_tokens_seen": 5704840,
+      "step": 33
+    },
+    {
+      "epoch": 0.012638835695135964,
+      "loss": 2.044581651687622,
+      "loss_ce": 0.0016129963332787156,
+      "loss_iou": 2.3663644790649414,
+      "loss_num": 2.046875,
+      "loss_xval": 2.046875,
+      "num_input_tokens_seen": 5704840,
+      "step": 33
+    },
+    {
+      "epoch": 0.013021830716200689,
+      "grad_norm": 103.70727825235932,
+      "learning_rate": 5e-06,
+      "loss": 4.4179,
+      "num_input_tokens_seen": 5877712,
+      "step": 34
+    },
+    {
+      "epoch": 0.013021830716200689,
+      "loss": 4.494317054748535,
+      "loss_ce": 0.002129969885572791,
+      "loss_iou": 3.775545358657837,
+      "loss_num": 4.5,
+      "loss_xval": 4.5,
+      "num_input_tokens_seen": 5877712,
+      "step": 34
+    },
+    {
+      "epoch": 0.013404825737265416,
+      "grad_norm": 47.36531865531552,
+      "learning_rate": 5e-06,
+      "loss": 1.1801,
+      "num_input_tokens_seen": 6050960,
+      "step": 35
+    },
+    {
+      "epoch": 0.013404825737265416,
+      "loss": 1.2508718967437744,
+      "loss_ce": 0.0023367758840322495,
+      "loss_iou": 2.011030673980713,
+      "loss_num": 1.25,
+      "loss_xval": 1.25,
+      "num_input_tokens_seen": 6050960,
+      "step": 35
+    },
+    {
+      "epoch": 0.013787820758330141,
+      "grad_norm": 96.79092463858534,
+      "learning_rate": 5e-06,
+      "loss": 2.1503,
+      "num_input_tokens_seen": 6224208,
+      "step": 36
+    },
+    {
+      "epoch": 0.013787820758330141,
+      "loss": 2.1378912925720215,
+      "loss_ce": 0.002148999134078622,
+      "loss_iou": 8.005640029907227,
+      "loss_num": 2.140625,
+      "loss_xval": 2.140625,
+      "num_input_tokens_seen": 6224208,
+      "step": 36
+    },
+    {
+      "epoch": 0.014170815779394868,
+      "grad_norm": 104.65517919895153,
+      "learning_rate": 5e-06,
+      "loss": 2.4288,
+      "num_input_tokens_seen": 6397048,
+      "step": 37
+    },
+    {
+      "epoch": 0.014170815779394868,
+      "loss": 2.458104133605957,
+      "loss_ce": 0.0030261282809078693,
+      "loss_iou": 2.239044189453125,
+      "loss_num": 2.453125,
+      "loss_xval": 2.453125,
+      "num_input_tokens_seen": 6397048,
+      "step": 37
+    },
+    {
+      "epoch": 0.014553810800459595,
+      "grad_norm": 45.91550086531722,
+      "learning_rate": 5e-06,
+      "loss": 1.9428,
+      "num_input_tokens_seen": 6570120,
+      "step": 38
+    },
+    {
+      "epoch": 0.014553810800459595,
+      "loss": 2.0247607231140137,
+      "loss_ce": 0.003276278730481863,
+      "loss_iou": 11.768243789672852,
+      "loss_num": 2.015625,
+      "loss_xval": 2.015625,
+      "num_input_tokens_seen": 6570120,
+      "step": 38
+    },
+    {
+      "epoch": 0.01493680582152432,
+      "grad_norm": 78.02027171698998,
+      "learning_rate": 5e-06,
+      "loss": 0.5863,
+      "num_input_tokens_seen": 6742992,
+      "step": 39
+    },
+    {
+      "epoch": 0.01493680582152432,
+      "loss": 0.5959486961364746,
+      "loss_ce": 0.0031753242947161198,
+      "loss_iou": 1.0671956539154053,
+      "loss_num": 0.59375,
+      "loss_xval": 0.59375,
+      "num_input_tokens_seen": 6742992,
+      "step": 39
+    },
+    {
+      "epoch": 0.015319800842589047,
+      "grad_norm": 108.02693357884827,
+      "learning_rate": 5e-06,
+      "loss": 2.9515,
+      "num_input_tokens_seen": 6915768,
+      "step": 40
+    },
+    {
+      "epoch": 0.015319800842589047,
+      "loss": 2.9932217597961426,
+      "loss_ce": 0.002987401094287634,
+      "loss_iou": 22.81262969970703,
+      "loss_num": 2.984375,
+      "loss_xval": 2.984375,
+      "num_input_tokens_seen": 6915768,
+      "step": 40
+    },
+    {
+      "epoch": 0.015702795863653772,
+      "grad_norm": 72.35437301636028,
+      "learning_rate": 5e-06,
+      "loss": 0.9581,
+      "num_input_tokens_seen": 7084968,
+      "step": 41
+    },
+    {
+      "epoch": 0.015702795863653772,
+      "loss": 1.0195496082305908,
+      "loss_ce": 0.0044127958826720715,
+      "loss_iou": 1.2581861019134521,
+      "loss_num": 1.015625,
+      "loss_xval": 1.015625,
+      "num_input_tokens_seen": 7084968,
+      "step": 41
+    },
+    {
+      "epoch": 0.0160857908847185,
+      "grad_norm": 57.05189470945179,
+      "learning_rate": 5e-06,
+      "loss": 2.1534,
+      "num_input_tokens_seen": 7258080,
+      "step": 42
+    },
+    {
+      "epoch": 0.0160857908847185,
+      "loss": 2.1609060764312744,
+      "loss_ce": 0.004656002391129732,
+      "loss_iou": 25.787906646728516,
+      "loss_num": 2.15625,
+      "loss_xval": 2.15625,
+      "num_input_tokens_seen": 7258080,
+      "step": 42
+    },
+    {
+      "epoch": 0.016468785905783226,
+      "grad_norm": 28.17430307668298,
+      "learning_rate": 5e-06,
+      "loss": 0.6822,
+      "num_input_tokens_seen": 7430792,
+      "step": 43
+    },
+    {
+      "epoch": 0.016468785905783226,
+      "loss": 0.6581391096115112,
+      "loss_ce": 0.003842195961624384,
+      "loss_iou": 1.172561764717102,
+      "loss_num": 0.65625,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 7430792,
+      "step": 43
+    },
+    {
+      "epoch": 0.01685178092684795,
+      "grad_norm": 61.68393713568411,
+      "learning_rate": 5e-06,
+      "loss": 1.0347,
+      "num_input_tokens_seen": 7603568,
+      "step": 44
+    },
+    {
+      "epoch": 0.01685178092684795,
+      "loss": 0.9895001649856567,
+      "loss_ce": 0.00414864718914032,
+      "loss_iou": 1.0747069120407104,
+      "loss_num": 0.984375,
+      "loss_xval": 0.984375,
+      "num_input_tokens_seen": 7603568,
+      "step": 44
+    },
+    {
+      "epoch": 0.017234775947912676,
+      "grad_norm": 32.228713424789944,
+      "learning_rate": 5e-06,
+      "loss": 0.8874,
+      "num_input_tokens_seen": 7772568,
+      "step": 45
+    },
+    {
+      "epoch": 0.017234775947912676,
+      "loss": 0.9120867252349854,
+      "loss_ce": 0.00437189731746912,
+      "loss_iou": 1.1424987316131592,
+      "loss_num": 0.90625,
+      "loss_xval": 0.90625,
+      "num_input_tokens_seen": 7772568,
+      "step": 45
+    },
+    {
+      "epoch": 0.017617770968977403,
+      "grad_norm": 67.19238474935851,
+      "learning_rate": 5e-06,
+      "loss": 0.6135,
+      "num_input_tokens_seen": 7945216,
+      "step": 46
+    },
+    {
+      "epoch": 0.017617770968977403,
+      "loss": 0.6822792887687683,
+      "loss_ce": 0.0050332145765423775,
+      "loss_iou": 1.542587161064148,
+      "loss_num": 0.67578125,
+      "loss_xval": 0.67578125,
+      "num_input_tokens_seen": 7945216,
+      "step": 46
+    },
+    {
+      "epoch": 0.01800076599004213,
+      "grad_norm": 98.77020921873604,
+      "learning_rate": 5e-06,
+      "loss": 1.7709,
+      "num_input_tokens_seen": 8118400,
+      "step": 47
+    },
+    {
+      "epoch": 0.01800076599004213,
+      "loss": 1.759229063987732,
+      "loss_ce": 0.004346287809312344,
+      "loss_iou": 2.3266377449035645,
+      "loss_num": 1.7578125,
+      "loss_xval": 1.7578125,
+      "num_input_tokens_seen": 8118400,
+      "step": 47
+    },
+    {
+      "epoch": 0.018383761011106857,
+      "grad_norm": 63.114520354559936,
+      "learning_rate": 5e-06,
+      "loss": 0.9873,
+      "num_input_tokens_seen": 8291176,
+      "step": 48
+    },
+    {
+      "epoch": 0.018383761011106857,
+      "loss": 1.0272853374481201,
+      "loss_ce": 0.004336033947765827,
+      "loss_iou": 1.1968190670013428,
+      "loss_num": 1.0234375,
+      "loss_xval": 1.0234375,
+      "num_input_tokens_seen": 8291176,
+      "step": 48
+    },
+    {
+      "epoch": 0.01876675603217158,
+      "grad_norm": 44.02810373754691,
+      "learning_rate": 5e-06,
+      "loss": 1.1976,
+      "num_input_tokens_seen": 8463632,
+      "step": 49
+    },
+    {
+      "epoch": 0.01876675603217158,
+      "loss": 1.1249773502349854,
+      "loss_ce": 0.0048601580783724785,
+      "loss_iou": 1.3406559228897095,
+      "loss_num": 1.1171875,
+      "loss_xval": 1.1171875,
+      "num_input_tokens_seen": 8463632,
+      "step": 49
+    },
+    {
+      "epoch": 0.019149751053236307,
+      "grad_norm": 84.624531596747,
+      "learning_rate": 5e-06,
+      "loss": 0.9335,
+      "num_input_tokens_seen": 8636576,
+      "step": 50
+    },
+    {
+      "epoch": 0.019149751053236307,
+      "eval_websight_new_CIoU": 0.43197502195835114,
+      "eval_websight_new_GIoU": 0.4391936808824539,
+      "eval_websight_new_IoU": 0.4392305761575699,
+      "eval_websight_new_MAE_all": 0.13614656776189804,
+      "eval_websight_new_MAE_h": 0.23954762518405914,
+      "eval_websight_new_MAE_w": 0.2133445292711258,
+      "eval_websight_new_MAE_x": 0.04393843561410904,
+      "eval_websight_new_MAE_y": 0.047755710780620575,
+      "eval_websight_new_NUM_probability": 0.14433185011148453,
+      "eval_websight_new_inside_bbox": 0.8020833432674408,
+      "eval_websight_new_loss": 1.6407192945480347,
+      "eval_websight_new_loss_ce": 0.009637001436203718,
+      "eval_websight_new_loss_iou": 1.931229829788208,
+      "eval_websight_new_loss_num": 1.6201171875,
+      "eval_websight_new_loss_xval": 1.6201171875,
+      "eval_websight_new_runtime": 55.8378,
+      "eval_websight_new_samples_per_second": 0.895,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 8636576,
+      "step": 50
+    },
+    {
+      "epoch": 0.019149751053236307,
+      "eval_seeclick_CIoU": 0.31005920469760895,
+      "eval_seeclick_GIoU": 0.31324100494384766,
+      "eval_seeclick_IoU": 0.3503878265619278,
+      "eval_seeclick_MAE_all": 0.1857384666800499,
+      "eval_seeclick_MAE_h": 0.27291879057884216,
+      "eval_seeclick_MAE_w": 0.22510244697332382,
+      "eval_seeclick_MAE_x": 0.13958367705345154,
+      "eval_seeclick_MAE_y": 0.10534897819161415,
+      "eval_seeclick_NUM_probability": 0.21946162730455399,
+      "eval_seeclick_inside_bbox": 0.5590277910232544,
+      "eval_seeclick_loss": 1.8699127435684204,
+      "eval_seeclick_loss_ce": 0.006399712758138776,
+      "eval_seeclick_loss_iou": 4.592440366744995,
+      "eval_seeclick_loss_num": 1.8291015625,
+      "eval_seeclick_loss_xval": 1.8291015625,
+      "eval_seeclick_runtime": 82.084,
+      "eval_seeclick_samples_per_second": 0.609,
+      "eval_seeclick_steps_per_second": 0.024,
+      "num_input_tokens_seen": 8636576,
+      "step": 50
+    },
+    {
+      "epoch": 0.019149751053236307,
+      "eval_icons_CIoU": 0.36255019903182983,
+      "eval_icons_GIoU": 0.36736391484737396,
+      "eval_icons_IoU": 0.3720673471689224,
+      "eval_icons_MAE_all": 0.15631137788295746,
+      "eval_icons_MAE_h": 0.251584492623806,
+      "eval_icons_MAE_w": 0.2644573599100113,
+      "eval_icons_MAE_x": 0.05817447975277901,
+      "eval_icons_MAE_y": 0.05102917365729809,
+      "eval_icons_NUM_probability": 0.09473497048020363,
+      "eval_icons_inside_bbox": 0.8506944477558136,
+      "eval_icons_loss": 1.907942771911621,
+      "eval_icons_loss_ce": 0.007402913179248571,
+      "eval_icons_loss_iou": 1.7391288876533508,
+      "eval_icons_loss_num": 1.89208984375,
+      "eval_icons_loss_xval": 1.89208984375,
+      "eval_icons_runtime": 82.3134,
+      "eval_icons_samples_per_second": 0.607,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 8636576,
+      "step": 50
+    },
+    {
+      "epoch": 0.019149751053236307,
+      "loss": 1.9059419631958008,
+      "loss_ce": 0.007504510693252087,
+      "loss_iou": 1.7757078409194946,
+      "loss_num": 1.8984375,
+      "loss_xval": 1.8984375,
+      "num_input_tokens_seen": 8636576,
+      "step": 50
+    },
+    {
+      "epoch": 0.019532746074301034,
+      "grad_norm": 72.6783448644803,
+      "learning_rate": 5e-06,
+      "loss": 1.7384,
+      "num_input_tokens_seen": 8809256,
+      "step": 51
+    },
+    {
+      "epoch": 0.019532746074301034,
+      "loss": 1.7620909214019775,
+      "loss_ce": 0.0052549513056874275,
+      "loss_iou": 3.7303404808044434,
+      "loss_num": 1.7578125,
+      "loss_xval": 1.7578125,
+      "num_input_tokens_seen": 8809256,
+      "step": 51
+    },
+    {
+      "epoch": 0.01991574109536576,
+      "grad_norm": 27.706596397207054,
+      "learning_rate": 5e-06,
+      "loss": 0.891,
+      "num_input_tokens_seen": 8982032,
+      "step": 52
+    },
+    {
+      "epoch": 0.01991574109536576,
+      "loss": 0.8567217588424683,
+      "loss_ce": 0.006135815754532814,
+      "loss_iou": 1.1876529455184937,
+      "loss_num": 0.8515625,
+      "loss_xval": 0.8515625,
+      "num_input_tokens_seen": 8982032,
+      "step": 52
+    },
+    {
+      "epoch": 0.020298736116430488,
+      "grad_norm": 43.56380441147401,
+      "learning_rate": 5e-06,
+      "loss": 0.4726,
+      "num_input_tokens_seen": 9155144,
+      "step": 53
+    },
+    {
+      "epoch": 0.020298736116430488,
+      "loss": 0.4914742410182953,
+      "loss_ce": 0.005878560245037079,
+      "loss_iou": 1.143690586090088,
+      "loss_num": 0.486328125,
+      "loss_xval": 0.486328125,
+      "num_input_tokens_seen": 9155144,
+      "step": 53
+    },
+    {
+      "epoch": 0.02068173113749521,
+      "grad_norm": 28.591352396787148,
+      "learning_rate": 5e-06,
+      "loss": 0.8557,
+      "num_input_tokens_seen": 9327944,
+      "step": 54
+    },
+    {
+      "epoch": 0.02068173113749521,
+      "loss": 0.8597661256790161,
+      "loss_ce": 0.005273905117064714,
+      "loss_iou": 1.217780590057373,
+      "loss_num": 0.85546875,
+      "loss_xval": 0.85546875,
+      "num_input_tokens_seen": 9327944,
+      "step": 54
+    },
+    {
+      "epoch": 0.02106472615855994,
+      "grad_norm": 46.95323992873654,
+      "learning_rate": 5e-06,
+      "loss": 0.5126,
+      "num_input_tokens_seen": 9500872,
+      "step": 55
+    },
+    {
+      "epoch": 0.02106472615855994,
+      "loss": 0.5463019609451294,
+      "loss_ce": 0.005774635821580887,
+      "loss_iou": 1.072084665298462,
+      "loss_num": 0.5390625,
+      "loss_xval": 0.5390625,
+      "num_input_tokens_seen": 9500872,
+      "step": 55
+    },
+    {
+      "epoch": 0.021447721179624665,
+      "grad_norm": 38.03301004910845,
+      "learning_rate": 5e-06,
+      "loss": 0.8326,
+      "num_input_tokens_seen": 9673920,
+      "step": 56
+    },
+    {
+      "epoch": 0.021447721179624665,
+      "loss": 0.7999447584152222,
+      "loss_ce": 0.005022889003157616,
+      "loss_iou": 1.042433261871338,
+      "loss_num": 0.796875,
+      "loss_xval": 0.796875,
+      "num_input_tokens_seen": 9673920,
+      "step": 56
+    },
+    {
+      "epoch": 0.021830716200689392,
+      "grad_norm": 16.554509285804894,
+      "learning_rate": 5e-06,
+      "loss": 0.4558,
+      "num_input_tokens_seen": 9846568,
+      "step": 57
+    },
+    {
+      "epoch": 0.021830716200689392,
+      "loss": 0.426127552986145,
+      "loss_ce": 0.006205694749951363,
+      "loss_iou": 1.0183396339416504,
+      "loss_num": 0.419921875,
+      "loss_xval": 0.419921875,
+      "num_input_tokens_seen": 9846568,
+      "step": 57
+    },
+    {
+      "epoch": 0.022213711221754116,
+      "grad_norm": 31.72178541360483,
+      "learning_rate": 5e-06,
+      "loss": 0.3621,
+      "num_input_tokens_seen": 10019416,
+      "step": 58
+    },
+    {
+      "epoch": 0.022213711221754116,
+      "loss": 0.35577136278152466,
+      "loss_ce": 0.0067723277024924755,
+      "loss_iou": 1.0083670616149902,
+      "loss_num": 0.349609375,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 10019416,
+      "step": 58
+    },
+    {
+      "epoch": 0.022596706242818843,
+      "grad_norm": 37.92132274146497,
+      "learning_rate": 5e-06,
+      "loss": 0.663,
+      "num_input_tokens_seen": 10192152,
+      "step": 59
+    },
+    {
+      "epoch": 0.022596706242818843,
+      "loss": 0.6462411880493164,
+      "loss_ce": 0.0053720735013484955,
+      "loss_iou": 1.8834075927734375,
+      "loss_num": 0.640625,
+      "loss_xval": 0.640625,
+      "num_input_tokens_seen": 10192152,
+      "step": 59
+    },
+    {
+      "epoch": 0.02297970126388357,
+      "grad_norm": 42.481798631476686,
+      "learning_rate": 5e-06,
+      "loss": 0.3958,
+      "num_input_tokens_seen": 10365416,
+      "step": 60
+    },
+    {
+      "epoch": 0.02297970126388357,
+      "loss": 0.375872403383255,
+      "loss_ce": 0.006731784902513027,
+      "loss_iou": 1.0175365209579468,
+      "loss_num": 0.369140625,
+      "loss_xval": 0.369140625,
+      "num_input_tokens_seen": 10365416,
+      "step": 60
+    },
+    {
+      "epoch": 0.023362696284948296,
+      "grad_norm": 54.60705120241134,
+      "learning_rate": 5e-06,
+      "loss": 0.8181,
+      "num_input_tokens_seen": 10534848,
+      "step": 61
+    },
+    {
+      "epoch": 0.023362696284948296,
+      "loss": 0.8863776922225952,
+      "loss_ce": 0.005518313962966204,
+      "loss_iou": 1.1352825164794922,
+      "loss_num": 0.8828125,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 10534848,
+      "step": 61
+    },
+    {
+      "epoch": 0.023745691306013023,
+      "grad_norm": 35.219975611996276,
+      "learning_rate": 5e-06,
+      "loss": 0.4362,
+      "num_input_tokens_seen": 10707600,
+      "step": 62
+    },
+    {
+      "epoch": 0.023745691306013023,
+      "loss": 0.43948858976364136,
+      "loss_ce": 0.006383126601576805,
+      "loss_iou": 1.1062934398651123,
+      "loss_num": 0.43359375,
+      "loss_xval": 0.43359375,
+      "num_input_tokens_seen": 10707600,
+      "step": 62
+    },
+    {
+      "epoch": 0.024128686327077747,
+      "grad_norm": 56.84713250392648,
+      "learning_rate": 5e-06,
+      "loss": 0.5531,
+      "num_input_tokens_seen": 10881128,
+      "step": 63
+    },
+    {
+      "epoch": 0.024128686327077747,
+      "loss": 0.5669519901275635,
+      "loss_ce": 0.006405100226402283,
+      "loss_iou": 1.3195397853851318,
+      "loss_num": 0.5625,
+      "loss_xval": 0.5625,
+      "num_input_tokens_seen": 10881128,
+      "step": 63
+    },
+    {
+      "epoch": 0.024511681348142474,
+      "grad_norm": 69.52194454977185,
+      "learning_rate": 5e-06,
+      "loss": 0.8904,
+      "num_input_tokens_seen": 11053920,
+      "step": 64
+    },
+    {
+      "epoch": 0.024511681348142474,
+      "loss": 0.9384351968765259,
+      "loss_ce": 0.0067945728078484535,
+      "loss_iou": 1.0879559516906738,
+      "loss_num": 0.9296875,
+      "loss_xval": 0.9296875,
+      "num_input_tokens_seen": 11053920,
+      "step": 64
+    },
+    {
+      "epoch": 0.0248946763692072,
+      "grad_norm": 40.79322177764759,
+      "learning_rate": 5e-06,
+      "loss": 0.6829,
+      "num_input_tokens_seen": 11226752,
+      "step": 65
+    },
+    {
+      "epoch": 0.0248946763692072,
+      "loss": 0.7247698307037354,
+      "loss_ce": 0.006019807420670986,
+      "loss_iou": 1.2506904602050781,
+      "loss_num": 0.71875,
+      "loss_xval": 0.71875,
+      "num_input_tokens_seen": 11226752,
+      "step": 65
+    },
+    {
+      "epoch": 0.025277671390271927,
+      "grad_norm": 47.654912684020516,
+      "learning_rate": 5e-06,
+      "loss": 0.487,
+      "num_input_tokens_seen": 11399488,
+      "step": 66
+    },
+    {
+      "epoch": 0.025277671390271927,
+      "loss": 0.5268010497093201,
+      "loss_ce": 0.006781519390642643,
+      "loss_iou": 1.1212661266326904,
+      "loss_num": 0.51953125,
+      "loss_xval": 0.51953125,
+      "num_input_tokens_seen": 11399488,
+      "step": 66
+    },
+    {
+      "epoch": 0.025660666411336654,
+      "grad_norm": 54.30692560515609,
+      "learning_rate": 5e-06,
+      "loss": 0.7733,
+      "num_input_tokens_seen": 11572368,
+      "step": 67
+    },
+    {
+      "epoch": 0.025660666411336654,
+      "loss": 0.7901598811149597,
+      "loss_ce": 0.006468463689088821,
+      "loss_iou": 1.510016918182373,
+      "loss_num": 0.78515625,
+      "loss_xval": 0.78515625,
+      "num_input_tokens_seen": 11572368,
+      "step": 67
+    },
+    {
+      "epoch": 0.026043661432401378,
+      "grad_norm": 44.986881254476216,
+      "learning_rate": 5e-06,
+      "loss": 0.4963,
+      "num_input_tokens_seen": 11745096,
+      "step": 68
+    },
+    {
+      "epoch": 0.026043661432401378,
+      "loss": 0.48728063702583313,
+      "loss_ce": 0.00632362999022007,
+      "loss_iou": 1.1487774848937988,
+      "loss_num": 0.48046875,
+      "loss_xval": 0.48046875,
+      "num_input_tokens_seen": 11745096,
+      "step": 68
+    },
+    {
+      "epoch": 0.026426656453466105,
+      "grad_norm": 38.949899945039,
+      "learning_rate": 5e-06,
+      "loss": 0.6862,
+      "num_input_tokens_seen": 11917968,
+      "step": 69
+    },
+    {
+      "epoch": 0.026426656453466105,
+      "loss": 0.6249829530715942,
+      "loss_ce": 0.006330636329948902,
+      "loss_iou": 1.0569710731506348,
+      "loss_num": 0.6171875,
+      "loss_xval": 0.6171875,
+      "num_input_tokens_seen": 11917968,
+      "step": 69
+    },
+    {
+      "epoch": 0.02680965147453083,
+      "grad_norm": 43.54601105668646,
+      "learning_rate": 5e-06,
+      "loss": 0.3444,
+      "num_input_tokens_seen": 12090616,
+      "step": 70
+    },
+    {
+      "epoch": 0.02680965147453083,
+      "loss": 0.3571290969848633,
+      "loss_ce": 0.006543142721056938,
+      "loss_iou": 1.0208394527435303,
+      "loss_num": 0.3515625,
+      "loss_xval": 0.3515625,
+      "num_input_tokens_seen": 12090616,
+      "step": 70
+    },
+    {
+      "epoch": 0.02719264649559556,
+      "grad_norm": 50.61616085983298,
+      "learning_rate": 5e-06,
+      "loss": 0.6366,
+      "num_input_tokens_seen": 12263536,
+      "step": 71
+    },
+    {
+      "epoch": 0.02719264649559556,
+      "loss": 0.6586880683898926,
+      "loss_ce": 0.006832567509263754,
+      "loss_iou": 1.185067892074585,
+      "loss_num": 0.65234375,
+      "loss_xval": 0.65234375,
+      "num_input_tokens_seen": 12263536,
+      "step": 71
+    },
+    {
+      "epoch": 0.027575641516660282,
+      "grad_norm": 45.08389744664967,
+      "learning_rate": 5e-06,
+      "loss": 0.4513,
+      "num_input_tokens_seen": 12436576,
+      "step": 72
+    },
+    {
+      "epoch": 0.027575641516660282,
+      "loss": 0.45050758123397827,
+      "loss_ce": 0.00739235058426857,
+      "loss_iou": 1.0219526290893555,
+      "loss_num": 0.443359375,
+      "loss_xval": 0.443359375,
+      "num_input_tokens_seen": 12436576,
+      "step": 72
+    },
+    {
+      "epoch": 0.02795863653772501,
+      "grad_norm": 33.203652387586985,
+      "learning_rate": 5e-06,
+      "loss": 0.6655,
+      "num_input_tokens_seen": 12609096,
+      "step": 73
+    },
+    {
+      "epoch": 0.02795863653772501,
+      "loss": 0.6817867755889893,
+      "loss_ce": 0.006493816152215004,
+      "loss_iou": 1.0247434377670288,
+      "loss_num": 0.67578125,
+      "loss_xval": 0.67578125,
+      "num_input_tokens_seen": 12609096,
+      "step": 73
+    },
+    {
+      "epoch": 0.028341631558789736,
+      "grad_norm": 33.33240414801773,
+      "learning_rate": 5e-06,
+      "loss": 0.3952,
+      "num_input_tokens_seen": 12781776,
+      "step": 74
+    },
+    {
+      "epoch": 0.028341631558789736,
+      "loss": 0.40079963207244873,
+      "loss_ce": 0.00700080581009388,
+      "loss_iou": 1.2275466918945312,
+      "loss_num": 0.39453125,
+      "loss_xval": 0.39453125,
+      "num_input_tokens_seen": 12781776,
+      "step": 74
+    },
+    {
+      "epoch": 0.028724626579854463,
+      "grad_norm": 57.60823019996819,
+      "learning_rate": 5e-06,
+      "loss": 0.5417,
+      "num_input_tokens_seen": 12954968,
+      "step": 75
+    },
+    {
+      "epoch": 0.028724626579854463,
+      "loss": 0.5834289193153381,
+      "loss_ce": 0.006524628959596157,
+      "loss_iou": 1.1341683864593506,
+      "loss_num": 0.578125,
+      "loss_xval": 0.578125,
+      "num_input_tokens_seen": 12954968,
+      "step": 75
+    },
+    {
+      "epoch": 0.02910762160091919,
+      "grad_norm": 56.37315781796848,
+      "learning_rate": 5e-06,
+      "loss": 0.6461,
+      "num_input_tokens_seen": 13127536,
+      "step": 76
+    },
+    {
+      "epoch": 0.02910762160091919,
+      "loss": 0.6031838059425354,
+      "loss_ce": 0.005771694239228964,
+      "loss_iou": 0.763561487197876,
+      "loss_num": 0.59765625,
+      "loss_xval": 0.59765625,
+      "num_input_tokens_seen": 13127536,
+      "step": 76
+    },
+    {
+      "epoch": 0.029490616621983913,
+      "grad_norm": 33.02286060044843,
+      "learning_rate": 5e-06,
+      "loss": 0.498,
+      "num_input_tokens_seen": 13300688,
+      "step": 77
+    },
+    {
+      "epoch": 0.029490616621983913,
+      "loss": 0.511459231376648,
+      "loss_ce": 0.007064702454954386,
+      "loss_iou": 1.1034919023513794,
+      "loss_num": 0.50390625,
+      "loss_xval": 0.50390625,
+      "num_input_tokens_seen": 13300688,
+      "step": 77
+    },
+    {
+      "epoch": 0.02987361164304864,
+      "grad_norm": 50.1946807362945,
+      "learning_rate": 5e-06,
+      "loss": 0.5003,
+      "num_input_tokens_seen": 13470424,
+      "step": 78
+    },
+    {
+      "epoch": 0.02987361164304864,
+      "loss": 0.4736681580543518,
+      "loss_ce": 0.0068712737411260605,
+      "loss_iou": 1.0265188217163086,
+      "loss_num": 0.466796875,
+      "loss_xval": 0.466796875,
+      "num_input_tokens_seen": 13470424,
+      "step": 78
+    },
+    {
+      "epoch": 0.030256606664113367,
+      "grad_norm": 44.010831517031136,
+      "learning_rate": 5e-06,
+      "loss": 0.4536,
+      "num_input_tokens_seen": 13643424,
+      "step": 79
+    },
+    {
+      "epoch": 0.030256606664113367,
+      "loss": 0.44002223014831543,
+      "loss_ce": 0.005940227769315243,
+      "loss_iou": 0.7701394557952881,
+      "loss_num": 0.43359375,
+      "loss_xval": 0.43359375,
+      "num_input_tokens_seen": 13643424,
+      "step": 79
+    },
+    {
+      "epoch": 0.030639601685178094,
+      "grad_norm": 44.464194440871175,
+      "learning_rate": 5e-06,
+      "loss": 0.522,
+      "num_input_tokens_seen": 13816072,
+      "step": 80
+    },
+    {
+      "epoch": 0.030639601685178094,
+      "loss": 0.42485499382019043,
+      "loss_ce": 0.006397980265319347,
+      "loss_iou": 1.0180633068084717,
+      "loss_num": 0.41796875,
+      "loss_xval": 0.41796875,
+      "num_input_tokens_seen": 13816072,
+      "step": 80
+    },
+    {
+      "epoch": 0.031022596706242817,
+      "grad_norm": 45.20686847170137,
+      "learning_rate": 5e-06,
+      "loss": 0.561,
+      "num_input_tokens_seen": 13988800,
+      "step": 81
+    },
+    {
+      "epoch": 0.031022596706242817,
+      "loss": 0.5155953168869019,
+      "loss_ce": 0.007050366140902042,
+      "loss_iou": 1.033956527709961,
+      "loss_num": 0.5078125,
+      "loss_xval": 0.5078125,
+      "num_input_tokens_seen": 13988800,
+      "step": 81
+    },
+    {
+      "epoch": 0.031405591727307544,
+      "grad_norm": 40.388364285379396,
+      "learning_rate": 5e-06,
+      "loss": 0.4487,
+      "num_input_tokens_seen": 14162104,
+      "step": 82
+    },
+    {
+      "epoch": 0.031405591727307544,
+      "loss": 0.44451308250427246,
+      "loss_ce": 0.007013088092207909,
+      "loss_iou": 1.0524094104766846,
+      "loss_num": 0.4375,
+      "loss_xval": 0.4375,
+      "num_input_tokens_seen": 14162104,
+      "step": 82
+    },
+    {
+      "epoch": 0.031788586748372274,
+      "grad_norm": 29.735627600750053,
+      "learning_rate": 5e-06,
+      "loss": 0.5666,
+      "num_input_tokens_seen": 14334784,
+      "step": 83
+    },
+    {
+      "epoch": 0.031788586748372274,
+      "loss": 0.5328165888786316,
+      "loss_ce": 0.006693531759083271,
+      "loss_iou": 1.1166609525680542,
+      "loss_num": 0.52734375,
+      "loss_xval": 0.52734375,
+      "num_input_tokens_seen": 14334784,
+      "step": 83
+    },
+    {
+      "epoch": 0.032171581769437,
+      "grad_norm": 27.722040585539666,
+      "learning_rate": 5e-06,
+      "loss": 0.3232,
+      "num_input_tokens_seen": 14507224,
+      "step": 84
+    },
+    {
+      "epoch": 0.032171581769437,
+      "loss": 0.3890770971775055,
+      "loss_ce": 0.006874939426779747,
+      "loss_iou": 1.0110466480255127,
+      "loss_num": 0.3828125,
+      "loss_xval": 0.3828125,
+      "num_input_tokens_seen": 14507224,
+      "step": 84
+    },
+    {
+      "epoch": 0.03255457679050172,
+      "grad_norm": 31.390461757254684,
+      "learning_rate": 5e-06,
+      "loss": 0.5257,
+      "num_input_tokens_seen": 14680288,
+      "step": 85
+    },
+    {
+      "epoch": 0.03255457679050172,
+      "loss": 0.5557568073272705,
+      "loss_ce": 0.006440396420657635,
+      "loss_iou": 1.0433627367019653,
+      "loss_num": 0.55078125,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 14680288,
+      "step": 85
+    },
+    {
+      "epoch": 0.03293757181156645,
+      "grad_norm": 32.839011842359895,
+      "learning_rate": 5e-06,
+      "loss": 0.4123,
+      "num_input_tokens_seen": 14853464,
+      "step": 86
+    },
+    {
+      "epoch": 0.03293757181156645,
+      "loss": 0.3949050307273865,
+      "loss_ce": 0.006965578068047762,
+      "loss_iou": 1.0332398414611816,
+      "loss_num": 0.388671875,
+      "loss_xval": 0.388671875,
+      "num_input_tokens_seen": 14853464,
+      "step": 86
+    },
+    {
+      "epoch": 0.033320566832631175,
+      "grad_norm": 25.639189178603175,
+      "learning_rate": 5e-06,
+      "loss": 0.5734,
+      "num_input_tokens_seen": 15026264,
+      "step": 87
+    },
+    {
+      "epoch": 0.033320566832631175,
+      "loss": 0.6040345430374146,
+      "loss_ce": 0.006378264632076025,
+      "loss_iou": 10.921671867370605,
+      "loss_num": 0.59765625,
+      "loss_xval": 0.59765625,
+      "num_input_tokens_seen": 15026264,
+      "step": 87
+    },
+    {
+      "epoch": 0.0337035618536959,
+      "grad_norm": 37.876886363808794,
+      "learning_rate": 5e-06,
+      "loss": 0.3964,
+      "num_input_tokens_seen": 15198920,
+      "step": 88
+    },
+    {
+      "epoch": 0.0337035618536959,
+      "loss": 0.4036218225955963,
+      "loss_ce": 0.00603879801928997,
+      "loss_iou": 1.032512903213501,
+      "loss_num": 0.3984375,
+      "loss_xval": 0.3984375,
+      "num_input_tokens_seen": 15198920,
+      "step": 88
+    },
+    {
+      "epoch": 0.03408655687476063,
+      "grad_norm": 22.333466483174806,
+      "learning_rate": 5e-06,
+      "loss": 0.4302,
+      "num_input_tokens_seen": 15371928,
+      "step": 89
+    },
+    {
+      "epoch": 0.03408655687476063,
+      "loss": 0.460678368806839,
+      "loss_ce": 0.006088521331548691,
+      "loss_iou": 1.0937793254852295,
+      "loss_num": 0.455078125,
+      "loss_xval": 0.455078125,
+      "num_input_tokens_seen": 15371928,
+      "step": 89
+    },
+    {
+      "epoch": 0.03446955189582535,
+      "grad_norm": 29.069531495758998,
+      "learning_rate": 5e-06,
+      "loss": 0.3827,
+      "num_input_tokens_seen": 15544760,
+      "step": 90
+    },
+    {
+      "epoch": 0.03446955189582535,
+      "loss": 0.35321101546287537,
+      "loss_ce": 0.0066533940844237804,
+      "loss_iou": 1.0216233730316162,
+      "loss_num": 0.345703125,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 15544760,
+      "step": 90
+    },
+    {
+      "epoch": 0.03485254691689008,
+      "grad_norm": 26.540816273996406,
+      "learning_rate": 5e-06,
+      "loss": 0.4596,
+      "num_input_tokens_seen": 15717512,
+      "step": 91
+    },
+    {
+      "epoch": 0.03485254691689008,
+      "loss": 0.4705299139022827,
+      "loss_ce": 0.006174429785460234,
+      "loss_iou": 1.0322036743164062,
+      "loss_num": 0.46484375,
+      "loss_xval": 0.46484375,
+      "num_input_tokens_seen": 15717512,
+      "step": 91
+    },
+    {
+      "epoch": 0.035235541937954806,
+      "grad_norm": 43.30091509778155,
+      "learning_rate": 5e-06,
+      "loss": 0.3605,
+      "num_input_tokens_seen": 15890272,
+      "step": 92
+    },
+    {
+      "epoch": 0.035235541937954806,
+      "loss": 0.3640331029891968,
+      "loss_ce": 0.00648917630314827,
+      "loss_iou": 1.5203404426574707,
+      "loss_num": 0.357421875,
+      "loss_xval": 0.357421875,
+      "num_input_tokens_seen": 15890272,
+      "step": 92
+    },
+    {
+      "epoch": 0.03561853695901953,
+      "grad_norm": 26.875492455183462,
+      "learning_rate": 5e-06,
+      "loss": 0.5214,
+      "num_input_tokens_seen": 16063008,
+      "step": 93
+    },
+    {
+      "epoch": 0.03561853695901953,
+      "loss": 0.5096790194511414,
+      "loss_ce": 0.006016903556883335,
+      "loss_iou": 1.1511123180389404,
+      "loss_num": 0.50390625,
+      "loss_xval": 0.50390625,
+      "num_input_tokens_seen": 16063008,
+      "step": 93
+    },
+    {
+      "epoch": 0.03600153198008426,
+      "grad_norm": 29.866848393811175,
+      "learning_rate": 5e-06,
+      "loss": 0.3029,
+      "num_input_tokens_seen": 16235992,
+      "step": 94
+    },
+    {
+      "epoch": 0.03600153198008426,
+      "loss": 0.30797234177589417,
+      "loss_ce": 0.006702814716845751,
+      "loss_iou": 1.0129989385604858,
+      "loss_num": 0.30078125,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 16235992,
+      "step": 94
+    },
+    {
+      "epoch": 0.036384527001148984,
+      "grad_norm": 28.283781291947857,
+      "learning_rate": 5e-06,
+      "loss": 0.5238,
+      "num_input_tokens_seen": 16408664,
+      "step": 95
+    },
+    {
+      "epoch": 0.036384527001148984,
+      "loss": 0.5198010206222534,
+      "loss_ce": 0.006861596368253231,
+      "loss_iou": 1.0395960807800293,
+      "loss_num": 0.51171875,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 16408664,
+      "step": 95
+    },
+    {
+      "epoch": 0.036767522022213714,
+      "grad_norm": 41.26521151645431,
+      "learning_rate": 5e-06,
+      "loss": 0.333,
+      "num_input_tokens_seen": 16581656,
+      "step": 96
+    },
+    {
+      "epoch": 0.036767522022213714,
+      "loss": 0.3221352696418762,
+      "loss_ce": 0.006339376792311668,
+      "loss_iou": 1.0053198337554932,
+      "loss_num": 0.31640625,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 16581656,
+      "step": 96
+    },
+    {
+      "epoch": 0.03715051704327844,
+      "grad_norm": 44.74181037480452,
+      "learning_rate": 5e-06,
+      "loss": 0.5314,
+      "num_input_tokens_seen": 16754648,
+      "step": 97
+    },
+    {
+      "epoch": 0.03715051704327844,
+      "loss": 0.5276448726654053,
+      "loss_ce": 0.006160501390695572,
+      "loss_iou": 1.0349318981170654,
+      "loss_num": 0.5234375,
+      "loss_xval": 0.5234375,
+      "num_input_tokens_seen": 16754648,
+      "step": 97
+    },
+    {
+      "epoch": 0.03753351206434316,
+      "grad_norm": 36.167358019916335,
+      "learning_rate": 5e-06,
+      "loss": 0.4675,
+      "num_input_tokens_seen": 16927656,
+      "step": 98
+    },
+    {
+      "epoch": 0.03753351206434316,
+      "loss": 0.45619115233421326,
+      "loss_ce": 0.0064841341227293015,
+      "loss_iou": 1.2112081050872803,
+      "loss_num": 0.44921875,
+      "loss_xval": 0.44921875,
+      "num_input_tokens_seen": 16927656,
+      "step": 98
+    },
+    {
+      "epoch": 0.03791650708540789,
+      "grad_norm": 37.21054003809216,
+      "learning_rate": 5e-06,
+      "loss": 0.3829,
+      "num_input_tokens_seen": 17100912,
+      "step": 99
+    },
+    {
+      "epoch": 0.03791650708540789,
+      "loss": 0.36574530601501465,
+      "loss_ce": 0.006126187741756439,
+      "loss_iou": 1.0228235721588135,
+      "loss_num": 0.359375,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 17100912,
+      "step": 99
+    },
+    {
+      "epoch": 0.038299502106472615,
+      "grad_norm": 56.04909535425301,
+      "learning_rate": 5e-06,
+      "loss": 0.4844,
+      "num_input_tokens_seen": 17273920,
+      "step": 100
+    },
+    {
+      "epoch": 0.038299502106472615,
+      "loss": 0.5044032335281372,
+      "loss_ce": 0.0068446556106209755,
+      "loss_iou": 1.1203261613845825,
+      "loss_num": 0.498046875,
+      "loss_xval": 0.498046875,
+      "num_input_tokens_seen": 17273920,
+      "step": 100
+    },
+    {
+      "epoch": 0.038682497127537345,
+      "grad_norm": 35.64530879543614,
+      "learning_rate": 5e-06,
+      "loss": 0.5197,
+      "num_input_tokens_seen": 17446584,
+      "step": 101
+    },
+    {
+      "epoch": 0.038682497127537345,
+      "loss": 0.5517237186431885,
+      "loss_ce": 0.0068018026649951935,
+      "loss_iou": 1.022824764251709,
+      "loss_num": 0.546875,
+      "loss_xval": 0.546875,
+      "num_input_tokens_seen": 17446584,
+      "step": 101
+    },
+    {
+      "epoch": 0.03906549214860207,
+      "grad_norm": 27.756838380117056,
+      "learning_rate": 5e-06,
+      "loss": 0.3293,
+      "num_input_tokens_seen": 17619448,
+      "step": 102
+    },
+    {
+      "epoch": 0.03906549214860207,
+      "loss": 0.30716967582702637,
+      "loss_ce": 0.006388426758348942,
+      "loss_iou": 1.035182237625122,
+      "loss_num": 0.30078125,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 17619448,
+      "step": 102
+    },
+    {
+      "epoch": 0.03944848716966679,
+      "grad_norm": 32.46338554186638,
+      "learning_rate": 5e-06,
+      "loss": 0.4103,
+      "num_input_tokens_seen": 17791928,
+      "step": 103
+    },
+    {
+      "epoch": 0.03944848716966679,
+      "loss": 0.4780769944190979,
+      "loss_ce": 0.0061531588435173035,
+      "loss_iou": 1.180977463722229,
+      "loss_num": 0.47265625,
+      "loss_xval": 0.47265625,
+      "num_input_tokens_seen": 17791928,
+      "step": 103
+    },
+    {
+      "epoch": 0.03983148219073152,
+      "grad_norm": 35.52457985288199,
+      "learning_rate": 5e-06,
+      "loss": 0.4137,
+      "num_input_tokens_seen": 17965016,
+      "step": 104
+    },
+    {
+      "epoch": 0.03983148219073152,
+      "loss": 0.3601241707801819,
+      "loss_ce": 0.0063643730245530605,
+      "loss_iou": 1.0742692947387695,
+      "loss_num": 0.353515625,
+      "loss_xval": 0.353515625,
+      "num_input_tokens_seen": 17965016,
+      "step": 104
+    },
+    {
+      "epoch": 0.040214477211796246,
+      "grad_norm": 41.866249068434854,
+      "learning_rate": 5e-06,
+      "loss": 0.4377,
+      "num_input_tokens_seen": 18137736,
+      "step": 105
+    },
+    {
+      "epoch": 0.040214477211796246,
+      "loss": 0.4701138734817505,
+      "loss_ce": 0.006490812171250582,
+      "loss_iou": 1.0544739961624146,
+      "loss_num": 0.462890625,
+      "loss_xval": 0.462890625,
+      "num_input_tokens_seen": 18137736,
+      "step": 105
+    },
+    {
+      "epoch": 0.040597472232860976,
+      "grad_norm": 30.01917950301748,
+      "learning_rate": 5e-06,
+      "loss": 0.3308,
+      "num_input_tokens_seen": 18310744,
+      "step": 106
+    },
+    {
+      "epoch": 0.040597472232860976,
+      "loss": 0.31964176893234253,
+      "loss_ce": 0.006531395949423313,
+      "loss_iou": 1.0224277973175049,
+      "loss_num": 0.3125,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 18310744,
+      "step": 106
+    },
+    {
+      "epoch": 0.0409804672539257,
+      "grad_norm": 36.06766890362076,
+      "learning_rate": 5e-06,
+      "loss": 0.4414,
+      "num_input_tokens_seen": 18483568,
+      "step": 107
+    },
+    {
+      "epoch": 0.0409804672539257,
+      "loss": 0.3927014470100403,
+      "loss_ce": 0.006715108640491962,
+      "loss_iou": 1.047964334487915,
+      "loss_num": 0.38671875,
+      "loss_xval": 0.38671875,
+      "num_input_tokens_seen": 18483568,
+      "step": 107
+    },
+    {
+      "epoch": 0.04136346227499042,
+      "grad_norm": 42.79337571439956,
+      "learning_rate": 5e-06,
+      "loss": 0.4251,
+      "num_input_tokens_seen": 18656480,
+      "step": 108
+    },
+    {
+      "epoch": 0.04136346227499042,
+      "loss": 0.4319663345813751,
+      "loss_ce": 0.006429227534681559,
+      "loss_iou": 1.017852783203125,
+      "loss_num": 0.42578125,
+      "loss_xval": 0.42578125,
+      "num_input_tokens_seen": 18656480,
+      "step": 108
+    },
+    {
+      "epoch": 0.04174645729605515,
+      "grad_norm": 47.21985875144009,
+      "learning_rate": 5e-06,
+      "loss": 0.5227,
+      "num_input_tokens_seen": 18829808,
+      "step": 109
+    },
+    {
+      "epoch": 0.04174645729605515,
+      "loss": 0.5298129320144653,
+      "loss_ce": 0.006863709073513746,
+      "loss_iou": 1.0822196006774902,
+      "loss_num": 0.5234375,
+      "loss_xval": 0.5234375,
+      "num_input_tokens_seen": 18829808,
+      "step": 109
+    },
+    {
+      "epoch": 0.04212945231711988,
+      "grad_norm": 30.911890471985224,
+      "learning_rate": 5e-06,
+      "loss": 0.3904,
+      "num_input_tokens_seen": 19002360,
+      "step": 110
+    },
+    {
+      "epoch": 0.04212945231711988,
+      "loss": 0.4123959243297577,
+      "loss_ce": 0.007000410463660955,
+      "loss_iou": 1.0159223079681396,
+      "loss_num": 0.40625,
+      "loss_xval": 0.40625,
+      "num_input_tokens_seen": 19002360,
+      "step": 110
+    },
+    {
+      "epoch": 0.0425124473381846,
+      "grad_norm": 27.433482377568076,
+      "learning_rate": 5e-06,
+      "loss": 0.6606,
+      "num_input_tokens_seen": 19175216,
+      "step": 111
+    },
+    {
+      "epoch": 0.0425124473381846,
+      "loss": 0.6263796091079712,
+      "loss_ce": 0.006750749424099922,
+      "loss_iou": 1.340945839881897,
+      "loss_num": 0.62109375,
+      "loss_xval": 0.62109375,
+      "num_input_tokens_seen": 19175216,
+      "step": 111
+    },
+    {
+      "epoch": 0.04289544235924933,
+      "grad_norm": 15.005011675758164,
+      "learning_rate": 5e-06,
+      "loss": 0.3459,
+      "num_input_tokens_seen": 19347720,
+      "step": 112
+    },
+    {
+      "epoch": 0.04289544235924933,
+      "loss": 0.4036482274532318,
+      "loss_ce": 0.006675555370748043,
+      "loss_iou": 1.0429775714874268,
+      "loss_num": 0.396484375,
+      "loss_xval": 0.396484375,
+      "num_input_tokens_seen": 19347720,
+      "step": 112
+    },
+    {
+      "epoch": 0.043278437380314054,
+      "grad_norm": 28.057970281066805,
+      "learning_rate": 5e-06,
+      "loss": 0.397,
+      "num_input_tokens_seen": 19520544,
+      "step": 113
+    },
+    {
+      "epoch": 0.043278437380314054,
+      "loss": 0.45218348503112793,
+      "loss_ce": 0.006504817865788937,
+      "loss_iou": 1.1148324012756348,
+      "loss_num": 0.4453125,
+      "loss_xval": 0.4453125,
+      "num_input_tokens_seen": 19520544,
+      "step": 113
+    },
+    {
+      "epoch": 0.043661432401378784,
+      "grad_norm": 36.79099990985957,
+      "learning_rate": 5e-06,
+      "loss": 0.5451,
+      "num_input_tokens_seen": 19693208,
+      "step": 114
+    },
+    {
+      "epoch": 0.043661432401378784,
+      "loss": 0.5250085592269897,
+      "loss_ce": 0.0064538768492639065,
+      "loss_iou": 1.0836858749389648,
+      "loss_num": 0.51953125,
+      "loss_xval": 0.51953125,
+      "num_input_tokens_seen": 19693208,
+      "step": 114
+    },
+    {
+      "epoch": 0.04404442742244351,
+      "grad_norm": 32.01939345082849,
+      "learning_rate": 5e-06,
+      "loss": 0.4893,
+      "num_input_tokens_seen": 19865688,
+      "step": 115
+    },
+    {
+      "epoch": 0.04404442742244351,
+      "loss": 0.5263857841491699,
+      "loss_ce": 0.006610426120460033,
+      "loss_iou": 7.912219047546387,
+      "loss_num": 0.51953125,
+      "loss_xval": 0.51953125,
+      "num_input_tokens_seen": 19865688,
+      "step": 115
+    },
+    {
+      "epoch": 0.04442742244350823,
+      "grad_norm": 21.26505252168516,
+      "learning_rate": 5e-06,
+      "loss": 0.4792,
+      "num_input_tokens_seen": 20038768,
+      "step": 116
+    },
+    {
+      "epoch": 0.04442742244350823,
+      "loss": 0.4978628158569336,
+      "loss_ce": 0.006407726556062698,
+      "loss_iou": 0.9237784147262573,
+      "loss_num": 0.4921875,
+      "loss_xval": 0.4921875,
+      "num_input_tokens_seen": 20038768,
+      "step": 116
+    },
+    {
+      "epoch": 0.04481041746457296,
+      "grad_norm": 31.743950455792852,
+      "learning_rate": 5e-06,
+      "loss": 0.2938,
+      "num_input_tokens_seen": 20211744,
+      "step": 117
+    },
+    {
+      "epoch": 0.04481041746457296,
+      "loss": 0.3334852159023285,
+      "loss_ce": 0.007313337177038193,
+      "loss_iou": 1.0548374652862549,
+      "loss_num": 0.326171875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 20211744,
+      "step": 117
+    },
+    {
+      "epoch": 0.045193412485637685,
+      "grad_norm": 37.95325305399651,
+      "learning_rate": 5e-06,
+      "loss": 0.5847,
+      "num_input_tokens_seen": 20384600,
+      "step": 118
+    },
+    {
+      "epoch": 0.045193412485637685,
+      "loss": 0.6075345277786255,
+      "loss_ce": 0.006216216832399368,
+      "loss_iou": 1.069112777709961,
+      "loss_num": 0.6015625,
+      "loss_xval": 0.6015625,
+      "num_input_tokens_seen": 20384600,
+      "step": 118
+    },
+    {
+      "epoch": 0.045576407506702415,
+      "grad_norm": 27.727475727388736,
+      "learning_rate": 5e-06,
+      "loss": 0.3302,
+      "num_input_tokens_seen": 20557288,
+      "step": 119
+    },
+    {
+      "epoch": 0.045576407506702415,
+      "loss": 0.37002646923065186,
+      "loss_ce": 0.006379032041877508,
+      "loss_iou": 1.0891635417938232,
+      "loss_num": 0.36328125,
+      "loss_xval": 0.36328125,
+      "num_input_tokens_seen": 20557288,
+      "step": 119
+    },
+    {
+      "epoch": 0.04595940252776714,
+      "grad_norm": 29.26227934580431,
+      "learning_rate": 5e-06,
+      "loss": 0.56,
+      "num_input_tokens_seen": 20730232,
+      "step": 120
+    },
+    {
+      "epoch": 0.04595940252776714,
+      "loss": 0.5652130842208862,
+      "loss_ce": 0.006131095811724663,
+      "loss_iou": 1.1232898235321045,
+      "loss_num": 0.55859375,
+      "loss_xval": 0.55859375,
+      "num_input_tokens_seen": 20730232,
+      "step": 120
+    },
+    {
+      "epoch": 0.04634239754883186,
+      "grad_norm": 38.70357615083572,
+      "learning_rate": 5e-06,
+      "loss": 0.3099,
+      "num_input_tokens_seen": 20903112,
+      "step": 121
+    },
+    {
+      "epoch": 0.04634239754883186,
+      "loss": 0.3055613338947296,
+      "loss_ce": 0.006244930438697338,
+      "loss_iou": 1.0352020263671875,
+      "loss_num": 0.298828125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 20903112,
+      "step": 121
+    },
+    {
+      "epoch": 0.04672539256989659,
+      "grad_norm": 38.84593131483085,
+      "learning_rate": 5e-06,
+      "loss": 0.5919,
+      "num_input_tokens_seen": 21076384,
+      "step": 122
+    },
+    {
+      "epoch": 0.04672539256989659,
+      "loss": 0.5931558609008789,
+      "loss_ce": 0.0062418365851044655,
+      "loss_iou": 1.0618486404418945,
+      "loss_num": 0.5859375,
+      "loss_xval": 0.5859375,
+      "num_input_tokens_seen": 21076384,
+      "step": 122
+    },
+    {
+      "epoch": 0.047108387590961316,
+      "grad_norm": 23.731658503817986,
+      "learning_rate": 5e-06,
+      "loss": 0.3136,
+      "num_input_tokens_seen": 21249376,
+      "step": 123
+    },
+    {
+      "epoch": 0.047108387590961316,
+      "loss": 0.3049842119216919,
+      "loss_ce": 0.006278182379901409,
+      "loss_iou": 0.8924912810325623,
+      "loss_num": 0.298828125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 21249376,
+      "step": 123
+    },
+    {
+      "epoch": 0.04749138261202605,
+      "grad_norm": 23.92192871406492,
+      "learning_rate": 5e-06,
+      "loss": 0.5108,
+      "num_input_tokens_seen": 21422104,
+      "step": 124
+    },
+    {
+      "epoch": 0.04749138261202605,
+      "loss": 0.5700915455818176,
+      "loss_ce": 0.006126689724624157,
+      "loss_iou": 1.1228581666946411,
+      "loss_num": 0.5625,
+      "loss_xval": 0.5625,
+      "num_input_tokens_seen": 21422104,
+      "step": 124
+    },
+    {
+      "epoch": 0.04787437763309077,
+      "grad_norm": 32.0527686315411,
+      "learning_rate": 5e-06,
+      "loss": 0.4048,
+      "num_input_tokens_seen": 21595152,
+      "step": 125
+    },
+    {
+      "epoch": 0.04787437763309077,
+      "loss": 0.35190096497535706,
+      "loss_ce": 0.006564033217728138,
+      "loss_iou": 1.0184372663497925,
+      "loss_num": 0.345703125,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 21595152,
+      "step": 125
+    },
+    {
+      "epoch": 0.04825737265415549,
+      "grad_norm": 32.19051576104468,
+      "learning_rate": 5e-06,
+      "loss": 0.5357,
+      "num_input_tokens_seen": 21767960,
+      "step": 126
+    },
+    {
+      "epoch": 0.04825737265415549,
+      "loss": 0.5383220314979553,
+      "loss_ce": 0.005851339548826218,
+      "loss_iou": 1.0118060111999512,
+      "loss_num": 0.53125,
+      "loss_xval": 0.53125,
+      "num_input_tokens_seen": 21767960,
+      "step": 126
+    },
+    {
+      "epoch": 0.048640367675220224,
+      "grad_norm": 30.852062165351455,
+      "learning_rate": 5e-06,
+      "loss": 0.3996,
+      "num_input_tokens_seen": 21940920,
+      "step": 127
+    },
+    {
+      "epoch": 0.048640367675220224,
+      "loss": 0.3716692328453064,
+      "loss_ce": 0.006434849463403225,
+      "loss_iou": 1.009103536605835,
+      "loss_num": 0.365234375,
+      "loss_xval": 0.365234375,
+      "num_input_tokens_seen": 21940920,
+      "step": 127
+    },
+    {
+      "epoch": 0.04902336269628495,
+      "grad_norm": 25.083165301601046,
+      "learning_rate": 5e-06,
+      "loss": 0.5802,
+      "num_input_tokens_seen": 22113824,
+      "step": 128
+    },
+    {
+      "epoch": 0.04902336269628495,
+      "loss": 0.6301028728485107,
+      "loss_ce": 0.006323575042188168,
+      "loss_iou": 1.0315409898757935,
+      "loss_num": 0.625,
+      "loss_xval": 0.625,
+      "num_input_tokens_seen": 22113824,
+      "step": 128
+    },
+    {
+      "epoch": 0.04940635771734968,
+      "grad_norm": 32.175327935154606,
+      "learning_rate": 5e-06,
+      "loss": 0.294,
+      "num_input_tokens_seen": 22286616,
+      "step": 129
+    },
+    {
+      "epoch": 0.04940635771734968,
+      "loss": 0.28241562843322754,
+      "loss_ce": 0.006292589008808136,
+      "loss_iou": 1.0409519672393799,
+      "loss_num": 0.275390625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 22286616,
+      "step": 129
+    },
+    {
+      "epoch": 0.0497893527384144,
+      "grad_norm": 28.477041376569616,
+      "learning_rate": 5e-06,
+      "loss": 0.5893,
+      "num_input_tokens_seen": 22459464,
+      "step": 130
+    },
+    {
+      "epoch": 0.0497893527384144,
+      "loss": 0.6028316020965576,
+      "loss_ce": 0.006151881068944931,
+      "loss_iou": 2.0542666912078857,
+      "loss_num": 0.59765625,
+      "loss_xval": 0.59765625,
+      "num_input_tokens_seen": 22459464,
+      "step": 130
+    },
+    {
+      "epoch": 0.050172347759479125,
+      "grad_norm": 16.26457165947007,
+      "learning_rate": 5e-06,
+      "loss": 0.311,
+      "num_input_tokens_seen": 22632328,
+      "step": 131
+    },
+    {
+      "epoch": 0.050172347759479125,
+      "loss": 0.3509281873703003,
+      "loss_ce": 0.005835396703332663,
+      "loss_iou": 1.0183525085449219,
+      "loss_num": 0.345703125,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 22632328,
+      "step": 131
+    },
+    {
+      "epoch": 0.050555342780543855,
+      "grad_norm": 14.366608602593727,
+      "learning_rate": 5e-06,
+      "loss": 0.2185,
+      "num_input_tokens_seen": 22805792,
+      "step": 132
+    },
+    {
+      "epoch": 0.050555342780543855,
+      "loss": 0.20429584383964539,
+      "loss_ce": 0.006541921757161617,
+      "loss_iou": 1.0044993162155151,
+      "loss_num": 0.197265625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 22805792,
+      "step": 132
+    },
+    {
+      "epoch": 0.05093833780160858,
+      "grad_norm": 26.165813380894868,
+      "learning_rate": 5e-06,
+      "loss": 0.3177,
+      "num_input_tokens_seen": 22978888,
+      "step": 133
+    },
+    {
+      "epoch": 0.05093833780160858,
+      "loss": 0.3332655429840088,
+      "loss_ce": 0.006727441679686308,
+      "loss_iou": 1.070448637008667,
+      "loss_num": 0.326171875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 22978888,
+      "step": 133
+    },
+    {
+      "epoch": 0.05132133282267331,
+      "grad_norm": 26.62556524899254,
+      "learning_rate": 5e-06,
+      "loss": 0.3302,
+      "num_input_tokens_seen": 23151696,
+      "step": 134
+    },
+    {
+      "epoch": 0.05132133282267331,
+      "loss": 0.35435616970062256,
+      "loss_ce": 0.006455780938267708,
+      "loss_iou": 1.1026301383972168,
+      "loss_num": 0.34765625,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 23151696,
+      "step": 134
+    },
+    {
+      "epoch": 0.05170432784373803,
+      "grad_norm": 46.30831954660043,
+      "learning_rate": 5e-06,
+      "loss": 0.4124,
+      "num_input_tokens_seen": 23324576,
+      "step": 135
+    },
+    {
+      "epoch": 0.05170432784373803,
+      "loss": 0.4030225872993469,
+      "loss_ce": 0.006294081453233957,
+      "loss_iou": 1.019537091255188,
+      "loss_num": 0.396484375,
+      "loss_xval": 0.396484375,
+      "num_input_tokens_seen": 23324576,
+      "step": 135
+    },
+    {
+      "epoch": 0.052087322864802756,
+      "grad_norm": 48.41750133971601,
+      "learning_rate": 5e-06,
+      "loss": 0.6584,
+      "num_input_tokens_seen": 23497536,
+      "step": 136
+    },
+    {
+      "epoch": 0.052087322864802756,
+      "loss": 0.6342858076095581,
+      "loss_ce": 0.0063560702838003635,
+      "loss_iou": 1.5107569694519043,
+      "loss_num": 0.62890625,
+      "loss_xval": 0.62890625,
+      "num_input_tokens_seen": 23497536,
+      "step": 136
+    },
+    {
+      "epoch": 0.052470317885867486,
+      "grad_norm": 28.673374952503366,
+      "learning_rate": 5e-06,
+      "loss": 0.5077,
+      "num_input_tokens_seen": 23670144,
+      "step": 137
+    },
+    {
+      "epoch": 0.052470317885867486,
+      "loss": 0.4749900698661804,
+      "loss_ce": 0.006972530856728554,
+      "loss_iou": 1.152420997619629,
+      "loss_num": 0.46875,
+      "loss_xval": 0.46875,
+      "num_input_tokens_seen": 23670144,
+      "step": 137
+    },
+    {
+      "epoch": 0.05285331290693221,
+      "grad_norm": 27.475418106466517,
+      "learning_rate": 5e-06,
+      "loss": 0.716,
+      "num_input_tokens_seen": 23843328,
+      "step": 138
+    },
+    {
+      "epoch": 0.05285331290693221,
+      "loss": 0.7018737196922302,
+      "loss_ce": 0.006072905845940113,
+      "loss_iou": 1.1133317947387695,
+      "loss_num": 0.6953125,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 23843328,
+      "step": 138
+    },
+    {
+      "epoch": 0.05323630792799693,
+      "grad_norm": 37.77224401074182,
+      "learning_rate": 5e-06,
+      "loss": 0.4237,
+      "num_input_tokens_seen": 24016376,
+      "step": 139
+    },
+    {
+      "epoch": 0.05323630792799693,
+      "loss": 0.41523200273513794,
+      "loss_ce": 0.006296470295637846,
+      "loss_iou": 1.0390156507492065,
+      "loss_num": 0.408203125,
+      "loss_xval": 0.408203125,
+      "num_input_tokens_seen": 24016376,
+      "step": 139
+    },
+    {
+      "epoch": 0.05361930294906166,
+      "grad_norm": 62.031647505417695,
+      "learning_rate": 5e-06,
+      "loss": 0.9266,
+      "num_input_tokens_seen": 24189160,
+      "step": 140
+    },
+    {
+      "epoch": 0.05361930294906166,
+      "loss": 0.9466855525970459,
+      "loss_ce": 0.005767581053078175,
+      "loss_iou": 1.0992369651794434,
+      "loss_num": 0.94140625,
+      "loss_xval": 0.94140625,
+      "num_input_tokens_seen": 24189160,
+      "step": 140
+    },
+    {
+      "epoch": 0.05400229797012639,
+      "grad_norm": 45.57677053576892,
+      "learning_rate": 5e-06,
+      "loss": 0.6621,
+      "num_input_tokens_seen": 24362424,
+      "step": 141
+    },
+    {
+      "epoch": 0.05400229797012639,
+      "loss": 0.6456308364868164,
+      "loss_ce": 0.0064706951379776,
+      "loss_iou": 1.0844511985778809,
+      "loss_num": 0.640625,
+      "loss_xval": 0.640625,
+      "num_input_tokens_seen": 24362424,
+      "step": 141
+    },
+    {
+      "epoch": 0.05438529299119112,
+      "grad_norm": 33.198264048981606,
+      "learning_rate": 5e-06,
+      "loss": 1.1533,
+      "num_input_tokens_seen": 24535224,
+      "step": 142
+    },
+    {
+      "epoch": 0.05438529299119112,
+      "loss": 1.1889057159423828,
+      "loss_ce": 0.0062885405495762825,
+      "loss_iou": 1.2113800048828125,
+      "loss_num": 1.1796875,
+      "loss_xval": 1.1796875,
+      "num_input_tokens_seen": 24535224,
+      "step": 142
+    },
+    {
+      "epoch": 0.05476828801225584,
+      "grad_norm": 23.691244793430347,
+      "learning_rate": 5e-06,
+      "loss": 0.5377,
+      "num_input_tokens_seen": 24708200,
+      "step": 143
+    },
+    {
+      "epoch": 0.05476828801225584,
+      "loss": 0.4705287516117096,
+      "loss_ce": 0.006417406257241964,
+      "loss_iou": 5.568227767944336,
+      "loss_num": 0.46484375,
+      "loss_xval": 0.46484375,
+      "num_input_tokens_seen": 24708200,
+      "step": 143
+    },
+    {
+      "epoch": 0.055151283033320564,
+      "grad_norm": 72.36887055732298,
+      "learning_rate": 5e-06,
+      "loss": 0.9508,
+      "num_input_tokens_seen": 24881232,
+      "step": 144
+    },
+    {
+      "epoch": 0.055151283033320564,
+      "loss": 0.9282786846160889,
+      "loss_ce": 0.006403679493814707,
+      "loss_iou": 1.1952100992202759,
+      "loss_num": 0.921875,
+      "loss_xval": 0.921875,
+      "num_input_tokens_seen": 24881232,
+      "step": 144
+    },
+    {
+      "epoch": 0.055534278054385294,
+      "grad_norm": 57.73536944055364,
+      "learning_rate": 5e-06,
+      "loss": 1.1748,
+      "num_input_tokens_seen": 25054168,
+      "step": 145
+    },
+    {
+      "epoch": 0.055534278054385294,
+      "loss": 1.1655254364013672,
+      "loss_ce": 0.006834015715867281,
+      "loss_iou": 1.104722023010254,
+      "loss_num": 1.15625,
+      "loss_xval": 1.15625,
+      "num_input_tokens_seen": 25054168,
+      "step": 145
+    },
+    {
+      "epoch": 0.05591727307545002,
+      "grad_norm": 47.047430292687814,
+      "learning_rate": 5e-06,
+      "loss": 1.2262,
+      "num_input_tokens_seen": 25223328,
+      "step": 146
+    },
+    {
+      "epoch": 0.05591727307545002,
+      "loss": 1.195068120956421,
+      "loss_ce": 0.006591492332518101,
+      "loss_iou": 1.2141139507293701,
+      "loss_num": 1.1875,
+      "loss_xval": 1.1875,
+      "num_input_tokens_seen": 25223328,
+      "step": 146
+    },
+    {
+      "epoch": 0.05630026809651475,
+      "grad_norm": 62.718817823310836,
+      "learning_rate": 5e-06,
+      "loss": 0.8621,
+      "num_input_tokens_seen": 25396024,
+      "step": 147
+    },
+    {
+      "epoch": 0.05630026809651475,
+      "loss": 0.8082544803619385,
+      "loss_ce": 0.006984918378293514,
+      "loss_iou": 1.0968055725097656,
+      "loss_num": 0.80078125,
+      "loss_xval": 0.80078125,
+      "num_input_tokens_seen": 25396024,
+      "step": 147
+    },
+    {
+      "epoch": 0.05668326311757947,
+      "grad_norm": 63.201855291575825,
+      "learning_rate": 5e-06,
+      "loss": 1.3091,
+      "num_input_tokens_seen": 25569216,
+      "step": 148
+    },
+    {
+      "epoch": 0.05668326311757947,
+      "loss": 1.2961779832839966,
+      "loss_ce": 0.006138939410448074,
+      "loss_iou": 1.3613967895507812,
+      "loss_num": 1.2890625,
+      "loss_xval": 1.2890625,
+      "num_input_tokens_seen": 25569216,
+      "step": 148
+    },
+    {
+      "epoch": 0.057066258138644195,
+      "grad_norm": 42.81570850698476,
+      "learning_rate": 5e-06,
+      "loss": 0.6737,
+      "num_input_tokens_seen": 25742280,
+      "step": 149
+    },
+    {
+      "epoch": 0.057066258138644195,
+      "loss": 0.6677260994911194,
+      "loss_ce": 0.006593281868845224,
+      "loss_iou": 1.3066319227218628,
+      "loss_num": 0.66015625,
+      "loss_xval": 0.66015625,
+      "num_input_tokens_seen": 25742280,
+      "step": 149
+    },
+    {
+      "epoch": 0.057449253159708925,
+      "grad_norm": 54.721571851084676,
+      "learning_rate": 5e-06,
+      "loss": 1.0214,
+      "num_input_tokens_seen": 25915544,
+      "step": 150
+    },
+    {
+      "epoch": 0.057449253159708925,
+      "loss": 0.9635263681411743,
+      "loss_ce": 0.006983435247093439,
+      "loss_iou": 1.1073896884918213,
+      "loss_num": 0.95703125,
+      "loss_xval": 0.95703125,
+      "num_input_tokens_seen": 25915544,
+      "step": 150
+    },
+    {
+      "epoch": 0.05783224818077365,
+      "grad_norm": 52.011697716773476,
+      "learning_rate": 5e-06,
+      "loss": 0.7022,
+      "num_input_tokens_seen": 26088520,
+      "step": 151
+    },
+    {
+      "epoch": 0.05783224818077365,
+      "loss": 0.7164946794509888,
+      "loss_ce": 0.007510306313633919,
+      "loss_iou": 1.124539852142334,
+      "loss_num": 0.7109375,
+      "loss_xval": 0.7109375,
+      "num_input_tokens_seen": 26088520,
+      "step": 151
+    },
+    {
+      "epoch": 0.05821524320183838,
+      "grad_norm": 49.124407306881245,
+      "learning_rate": 5e-06,
+      "loss": 1.022,
+      "num_input_tokens_seen": 26261656,
+      "step": 152
+    },
+    {
+      "epoch": 0.05821524320183838,
+      "loss": 1.027024269104004,
+      "loss_ce": 0.006516510155051947,
+      "loss_iou": 1.1156485080718994,
+      "loss_num": 1.0234375,
+      "loss_xval": 1.0234375,
+      "num_input_tokens_seen": 26261656,
+      "step": 152
+    },
+    {
+      "epoch": 0.0585982382229031,
+      "grad_norm": 20.596648525427405,
+      "learning_rate": 5e-06,
+      "loss": 0.476,
+      "num_input_tokens_seen": 26434768,
+      "step": 153
+    },
+    {
+      "epoch": 0.0585982382229031,
+      "loss": 0.4455060362815857,
+      "loss_ce": 0.007273610215634108,
+      "loss_iou": 1.0470380783081055,
+      "loss_num": 0.4375,
+      "loss_xval": 0.4375,
+      "num_input_tokens_seen": 26434768,
+      "step": 153
+    },
+    {
+      "epoch": 0.058981233243967826,
+      "grad_norm": 38.74292928899851,
+      "learning_rate": 5e-06,
+      "loss": 0.4473,
+      "num_input_tokens_seen": 26607624,
+      "step": 154
+    },
+    {
+      "epoch": 0.058981233243967826,
+      "loss": 0.4658415615558624,
+      "loss_ce": 0.0063688987866044044,
+      "loss_iou": 1.0424962043762207,
+      "loss_num": 0.458984375,
+      "loss_xval": 0.458984375,
+      "num_input_tokens_seen": 26607624,
+      "step": 154
+    },
+    {
+      "epoch": 0.059364228265032556,
+      "grad_norm": 44.82288396062709,
+      "learning_rate": 5e-06,
+      "loss": 0.6104,
+      "num_input_tokens_seen": 26780384,
+      "step": 155
+    },
+    {
+      "epoch": 0.059364228265032556,
+      "loss": 0.6404857635498047,
+      "loss_ce": 0.006940814666450024,
+      "loss_iou": 1.0927469730377197,
+      "loss_num": 0.6328125,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 26780384,
+      "step": 155
+    },
+    {
+      "epoch": 0.05974722328609728,
+      "grad_norm": 49.17972131845761,
+      "learning_rate": 5e-06,
+      "loss": 0.6254,
+      "num_input_tokens_seen": 26953456,
+      "step": 156
+    },
+    {
+      "epoch": 0.05974722328609728,
+      "loss": 0.5728522539138794,
+      "loss_ce": 0.007178430445492268,
+      "loss_iou": 1.025742530822754,
+      "loss_num": 0.56640625,
+      "loss_xval": 0.56640625,
+      "num_input_tokens_seen": 26953456,
+      "step": 156
+    },
+    {
+      "epoch": 0.06013021830716201,
+      "grad_norm": 40.51980029856512,
+      "learning_rate": 5e-06,
+      "loss": 0.6819,
+      "num_input_tokens_seen": 27126304,
+      "step": 157
+    },
+    {
+      "epoch": 0.06013021830716201,
+      "loss": 0.7225454449653625,
+      "loss_ce": 0.007701745256781578,
+      "loss_iou": 1.2270827293395996,
+      "loss_num": 0.71484375,
+      "loss_xval": 0.71484375,
+      "num_input_tokens_seen": 27126304,
+      "step": 157
+    },
+    {
+      "epoch": 0.060513213328226734,
+      "grad_norm": 54.900204917719726,
+      "learning_rate": 5e-06,
+      "loss": 0.6266,
+      "num_input_tokens_seen": 27299432,
+      "step": 158
+    },
+    {
+      "epoch": 0.060513213328226734,
+      "loss": 0.670357346534729,
+      "loss_ce": 0.007271396461874247,
+      "loss_iou": 1.1041251420974731,
+      "loss_num": 0.6640625,
+      "loss_xval": 0.6640625,
+      "num_input_tokens_seen": 27299432,
+      "step": 158
+    },
+    {
+      "epoch": 0.06089620834929146,
+      "grad_norm": 62.96238710810382,
+      "learning_rate": 5e-06,
+      "loss": 1.0806,
+      "num_input_tokens_seen": 27472512,
+      "step": 159
+    },
+    {
+      "epoch": 0.06089620834929146,
+      "loss": 1.0890766382217407,
+      "loss_ce": 0.00753366481512785,
+      "loss_iou": 1.5399413108825684,
+      "loss_num": 1.078125,
+      "loss_xval": 1.078125,
+      "num_input_tokens_seen": 27472512,
+      "step": 159
+    },
+    {
+      "epoch": 0.06127920337035619,
+      "grad_norm": 28.509577655399855,
+      "learning_rate": 5e-06,
+      "loss": 0.3219,
+      "num_input_tokens_seen": 27645800,
+      "step": 160
+    },
+    {
+      "epoch": 0.06127920337035619,
+      "loss": 0.3256051242351532,
+      "loss_ce": 0.007978152483701706,
+      "loss_iou": 1.0282046794891357,
+      "loss_num": 0.318359375,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 27645800,
+      "step": 160
+    },
+    {
+      "epoch": 0.06166219839142091,
+      "grad_norm": 36.40162438649297,
+      "learning_rate": 5e-06,
+      "loss": 0.5308,
+      "num_input_tokens_seen": 27819040,
+      "step": 161
+    },
+    {
+      "epoch": 0.06166219839142091,
+      "loss": 0.5527220964431763,
+      "loss_ce": 0.007800169289112091,
+      "loss_iou": 1.076228141784668,
+      "loss_num": 0.546875,
+      "loss_xval": 0.546875,
+      "num_input_tokens_seen": 27819040,
+      "step": 161
+    },
+    {
+      "epoch": 0.062045193412485634,
+      "grad_norm": 43.79767513197773,
+      "learning_rate": 5e-06,
+      "loss": 0.5429,
+      "num_input_tokens_seen": 27991984,
+      "step": 162
+    },
+    {
+      "epoch": 0.062045193412485634,
+      "loss": 0.5679165124893188,
+      "loss_ce": 0.0073696100153028965,
+      "loss_iou": 1.1133733987808228,
+      "loss_num": 0.5625,
+      "loss_xval": 0.5625,
+      "num_input_tokens_seen": 27991984,
+      "step": 162
+    },
+    {
+      "epoch": 0.062428188433550365,
+      "grad_norm": 43.332650487487484,
+      "learning_rate": 5e-06,
+      "loss": 0.5952,
+      "num_input_tokens_seen": 28164728,
+      "step": 163
+    },
+    {
+      "epoch": 0.062428188433550365,
+      "loss": 0.5726608633995056,
+      "loss_ce": 0.007475370075553656,
+      "loss_iou": 1.0769777297973633,
+      "loss_num": 0.56640625,
+      "loss_xval": 0.56640625,
+      "num_input_tokens_seen": 28164728,
+      "step": 163
+    },
+    {
+      "epoch": 0.06281118345461509,
+      "grad_norm": 44.61240035938271,
+      "learning_rate": 5e-06,
+      "loss": 0.5017,
+      "num_input_tokens_seen": 28337712,
+      "step": 164
+    },
+    {
+      "epoch": 0.06281118345461509,
+      "loss": 0.5077217817306519,
+      "loss_ce": 0.007721826899796724,
+      "loss_iou": 1.0379432439804077,
+      "loss_num": 0.5,
+      "loss_xval": 0.5,
+      "num_input_tokens_seen": 28337712,
+      "step": 164
+    },
+    {
+      "epoch": 0.06319417847567982,
+      "grad_norm": 40.832653551206405,
+      "learning_rate": 5e-06,
+      "loss": 0.749,
+      "num_input_tokens_seen": 28510480,
+      "step": 165
+    },
+    {
+      "epoch": 0.06319417847567982,
+      "loss": 0.6996516585350037,
+      "loss_ce": 0.007512982003390789,
+      "loss_iou": 1.1332221031188965,
+      "loss_num": 0.69140625,
+      "loss_xval": 0.69140625,
+      "num_input_tokens_seen": 28510480,
+      "step": 165
+    },
+    {
+      "epoch": 0.06357717349674455,
+      "grad_norm": 22.462002530260524,
+      "learning_rate": 5e-06,
+      "loss": 0.3961,
+      "num_input_tokens_seen": 28683168,
+      "step": 166
+    },
+    {
+      "epoch": 0.06357717349674455,
+      "loss": 0.4529934823513031,
+      "loss_ce": 0.007314770948141813,
+      "loss_iou": 1.0337318181991577,
+      "loss_num": 0.4453125,
+      "loss_xval": 0.4453125,
+      "num_input_tokens_seen": 28683168,
+      "step": 166
+    },
+    {
+      "epoch": 0.06396016851780927,
+      "grad_norm": 47.560377666933995,
+      "learning_rate": 5e-06,
+      "loss": 0.4786,
+      "num_input_tokens_seen": 28856272,
+      "step": 167
+    },
+    {
+      "epoch": 0.06396016851780927,
+      "loss": 0.42942649126052856,
+      "loss_ce": 0.007795642130076885,
+      "loss_iou": 1.039899468421936,
+      "loss_num": 0.421875,
+      "loss_xval": 0.421875,
+      "num_input_tokens_seen": 28856272,
+      "step": 167
+    },
+    {
+      "epoch": 0.064343163538874,
+      "grad_norm": 38.71805316720224,
+      "learning_rate": 5e-06,
+      "loss": 0.5277,
+      "num_input_tokens_seen": 29028808,
+      "step": 168
+    },
+    {
+      "epoch": 0.064343163538874,
+      "loss": 0.5406422019004822,
+      "loss_ce": 0.007683200761675835,
+      "loss_iou": 1.0520868301391602,
+      "loss_num": 0.53125,
+      "loss_xval": 0.53125,
+      "num_input_tokens_seen": 29028808,
+      "step": 168
+    },
+    {
+      "epoch": 0.06472615855993873,
+      "grad_norm": 34.09177727602261,
+      "learning_rate": 5e-06,
+      "loss": 0.6291,
+      "num_input_tokens_seen": 29201496,
+      "step": 169
+    },
+    {
+      "epoch": 0.06472615855993873,
+      "loss": 0.6294732093811035,
+      "loss_ce": 0.00691462866961956,
+      "loss_iou": 1.0863300561904907,
+      "loss_num": 0.62109375,
+      "loss_xval": 0.62109375,
+      "num_input_tokens_seen": 29201496,
+      "step": 169
+    },
+    {
+      "epoch": 0.06510915358100344,
+      "grad_norm": 30.894532213255346,
+      "learning_rate": 5e-06,
+      "loss": 0.3206,
+      "num_input_tokens_seen": 29374440,
+      "step": 170
+    },
+    {
+      "epoch": 0.06510915358100344,
+      "loss": 0.31897273659706116,
+      "loss_ce": 0.007571367546916008,
+      "loss_iou": 1.0226755142211914,
+      "loss_num": 0.310546875,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 29374440,
+      "step": 170
+    },
+    {
+      "epoch": 0.06549214860206817,
+      "grad_norm": 35.87379981893007,
+      "learning_rate": 5e-06,
+      "loss": 0.5715,
+      "num_input_tokens_seen": 29547296,
+      "step": 171
+    },
+    {
+      "epoch": 0.06549214860206817,
+      "loss": 0.5430014729499817,
+      "loss_ce": 0.007356944959610701,
+      "loss_iou": 1.0217440128326416,
+      "loss_num": 0.53515625,
+      "loss_xval": 0.53515625,
+      "num_input_tokens_seen": 29547296,
+      "step": 171
+    },
+    {
+      "epoch": 0.0658751436231329,
+      "grad_norm": 40.63341353480755,
+      "learning_rate": 5e-06,
+      "loss": 0.5034,
+      "num_input_tokens_seen": 29720576,
+      "step": 172
+    },
+    {
+      "epoch": 0.0658751436231329,
+      "loss": 0.4879244863986969,
+      "loss_ce": 0.007944025099277496,
+      "loss_iou": 1.0626220703125,
+      "loss_num": 0.48046875,
+      "loss_xval": 0.48046875,
+      "num_input_tokens_seen": 29720576,
+      "step": 172
+    },
+    {
+      "epoch": 0.06625813864419762,
+      "grad_norm": 33.30447626176966,
+      "learning_rate": 5e-06,
+      "loss": 0.572,
+      "num_input_tokens_seen": 29893584,
+      "step": 173
+    },
+    {
+      "epoch": 0.06625813864419762,
+      "loss": 0.5977349877357483,
+      "loss_ce": 0.00837952271103859,
+      "loss_iou": 1.091104507446289,
+      "loss_num": 0.58984375,
+      "loss_xval": 0.58984375,
+      "num_input_tokens_seen": 29893584,
+      "step": 173
+    },
+    {
+      "epoch": 0.06664113366526235,
+      "grad_norm": 27.137745305887503,
+      "learning_rate": 5e-06,
+      "loss": 0.3584,
+      "num_input_tokens_seen": 30066968,
+      "step": 174
+    },
+    {
+      "epoch": 0.06664113366526235,
+      "loss": 0.3292595148086548,
+      "loss_ce": 0.007970421575009823,
+      "loss_iou": 1.8459727764129639,
+      "loss_num": 0.3203125,
+      "loss_xval": 0.3203125,
+      "num_input_tokens_seen": 30066968,
+      "step": 174
+    },
+    {
+      "epoch": 0.06702412868632708,
+      "grad_norm": 20.19896643616232,
+      "learning_rate": 5e-06,
+      "loss": 0.4606,
+      "num_input_tokens_seen": 30239680,
+      "step": 175
+    },
+    {
+      "epoch": 0.06702412868632708,
+      "loss": 0.5244916677474976,
+      "loss_ce": 0.008622505702078342,
+      "loss_iou": 1.0415037870407104,
+      "loss_num": 0.515625,
+      "loss_xval": 0.515625,
+      "num_input_tokens_seen": 30239680,
+      "step": 175
+    },
+    {
+      "epoch": 0.0674071237073918,
+      "grad_norm": 15.519149833363251,
+      "learning_rate": 5e-06,
+      "loss": 0.3395,
+      "num_input_tokens_seen": 30412768,
+      "step": 176
+    },
+    {
+      "epoch": 0.0674071237073918,
+      "loss": 0.3429456353187561,
+      "loss_ce": 0.008228832855820656,
+      "loss_iou": 2.97749662399292,
+      "loss_num": 0.333984375,
+      "loss_xval": 0.333984375,
+      "num_input_tokens_seen": 30412768,
+      "step": 176
+    },
+    {
+      "epoch": 0.06779011872845653,
+      "grad_norm": 21.20051326178833,
+      "learning_rate": 5e-06,
+      "loss": 0.3479,
+      "num_input_tokens_seen": 30585968,
+      "step": 177
+    },
+    {
+      "epoch": 0.06779011872845653,
+      "loss": 0.3102911710739136,
+      "loss_ce": 0.007556804455816746,
+      "loss_iou": 1.0246970653533936,
+      "loss_num": 0.302734375,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 30585968,
+      "step": 177
+    },
+    {
+      "epoch": 0.06817311374952126,
+      "grad_norm": 33.047666010284104,
+      "learning_rate": 5e-06,
+      "loss": 0.6411,
+      "num_input_tokens_seen": 30758712,
+      "step": 178
+    },
+    {
+      "epoch": 0.06817311374952126,
+      "loss": 0.6166718006134033,
+      "loss_ce": 0.007785105612128973,
+      "loss_iou": 1.0879333019256592,
+      "loss_num": 0.609375,
+      "loss_xval": 0.609375,
+      "num_input_tokens_seen": 30758712,
+      "step": 178
+    },
+    {
+      "epoch": 0.06855610877058599,
+      "grad_norm": 23.013083233642437,
+      "learning_rate": 5e-06,
+      "loss": 0.2356,
+      "num_input_tokens_seen": 30931520,
+      "step": 179
+    },
+    {
+      "epoch": 0.06855610877058599,
+      "loss": 0.24693651497364044,
+      "loss_ce": 0.007678701542317867,
+      "loss_iou": 1.0095577239990234,
+      "loss_num": 0.2392578125,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 30931520,
+      "step": 179
+    },
+    {
+      "epoch": 0.0689391037916507,
+      "grad_norm": 32.276607454957585,
+      "learning_rate": 5e-06,
+      "loss": 0.57,
+      "num_input_tokens_seen": 31104128,
+      "step": 180
+    },
+    {
+      "epoch": 0.0689391037916507,
+      "loss": 0.5437986850738525,
+      "loss_ce": 0.0076659126207232475,
+      "loss_iou": 1.1042509078979492,
+      "loss_num": 0.53515625,
+      "loss_xval": 0.53515625,
+      "num_input_tokens_seen": 31104128,
+      "step": 180
+    },
+    {
+      "epoch": 0.06932209881271544,
+      "grad_norm": 30.696336614046267,
+      "learning_rate": 5e-06,
+      "loss": 0.4571,
+      "num_input_tokens_seen": 31277088,
+      "step": 181
+    },
+    {
+      "epoch": 0.06932209881271544,
+      "loss": 0.520167350769043,
+      "loss_ce": 0.008448591455817223,
+      "loss_iou": 1.1707911491394043,
+      "loss_num": 0.51171875,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 31277088,
+      "step": 181
+    },
+    {
+      "epoch": 0.06970509383378017,
+      "grad_norm": 21.549556583354015,
+      "learning_rate": 5e-06,
+      "loss": 0.5931,
+      "num_input_tokens_seen": 31450440,
+      "step": 182
+    },
+    {
+      "epoch": 0.06970509383378017,
+      "loss": 0.5975214242935181,
+      "loss_ce": 0.00816600676625967,
+      "loss_iou": 1.0781558752059937,
+      "loss_num": 0.58984375,
+      "loss_xval": 0.58984375,
+      "num_input_tokens_seen": 31450440,
+      "step": 182
+    },
+    {
+      "epoch": 0.07008808885484488,
+      "grad_norm": 32.2803987366505,
+      "learning_rate": 5e-06,
+      "loss": 0.3026,
+      "num_input_tokens_seen": 31623288,
+      "step": 183
+    },
+    {
+      "epoch": 0.07008808885484488,
+      "loss": 0.2824394106864929,
+      "loss_ce": 0.008147402666509151,
+      "loss_iou": 1.078061580657959,
+      "loss_num": 0.2734375,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 31623288,
+      "step": 183
+    },
+    {
+      "epoch": 0.07047108387590961,
+      "grad_norm": 29.490819955260243,
+      "learning_rate": 5e-06,
+      "loss": 0.5545,
+      "num_input_tokens_seen": 31796328,
+      "step": 184
+    },
+    {
+      "epoch": 0.07047108387590961,
+      "loss": 0.5676002502441406,
+      "loss_ce": 0.007053417153656483,
+      "loss_iou": 1.0296332836151123,
+      "loss_num": 0.5625,
+      "loss_xval": 0.5625,
+      "num_input_tokens_seen": 31796328,
+      "step": 184
+    },
+    {
+      "epoch": 0.07085407889697434,
+      "grad_norm": 27.546106552010087,
+      "learning_rate": 5e-06,
+      "loss": 0.2937,
+      "num_input_tokens_seen": 31968760,
+      "step": 185
+    },
+    {
+      "epoch": 0.07085407889697434,
+      "loss": 0.29248303174972534,
+      "loss_ce": 0.007448871619999409,
+      "loss_iou": 1.0106669664382935,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 31968760,
+      "step": 185
+    },
+    {
+      "epoch": 0.07123707391803906,
+      "grad_norm": 29.262020890055478,
+      "learning_rate": 5e-06,
+      "loss": 0.4562,
+      "num_input_tokens_seen": 32141472,
+      "step": 186
+    },
+    {
+      "epoch": 0.07123707391803906,
+      "loss": 0.4467780590057373,
+      "loss_ce": 0.007324915379285812,
+      "loss_iou": 1.034911870956421,
+      "loss_num": 0.439453125,
+      "loss_xval": 0.439453125,
+      "num_input_tokens_seen": 32141472,
+      "step": 186
+    },
+    {
+      "epoch": 0.07162006893910379,
+      "grad_norm": 30.348694777589845,
+      "learning_rate": 5e-06,
+      "loss": 0.3995,
+      "num_input_tokens_seen": 32314280,
+      "step": 187
+    },
+    {
+      "epoch": 0.07162006893910379,
+      "loss": 0.40412604808807373,
+      "loss_ce": 0.008618230000138283,
+      "loss_iou": 4.503968238830566,
+      "loss_num": 0.39453125,
+      "loss_xval": 0.39453125,
+      "num_input_tokens_seen": 32314280,
+      "step": 187
+    },
+    {
+      "epoch": 0.07200306396016852,
+      "grad_norm": 19.337885486132954,
+      "learning_rate": 5e-06,
+      "loss": 0.4688,
+      "num_input_tokens_seen": 32487032,
+      "step": 188
+    },
+    {
+      "epoch": 0.07200306396016852,
+      "loss": 0.43590885400772095,
+      "loss_ce": 0.007442049216479063,
+      "loss_iou": 1.0248808860778809,
+      "loss_num": 0.427734375,
+      "loss_xval": 0.427734375,
+      "num_input_tokens_seen": 32487032,
+      "step": 188
+    },
+    {
+      "epoch": 0.07238605898123325,
+      "grad_norm": 17.980072416669966,
+      "learning_rate": 5e-06,
+      "loss": 0.2825,
+      "num_input_tokens_seen": 32660208,
+      "step": 189
+    },
+    {
+      "epoch": 0.07238605898123325,
+      "loss": 0.26397398114204407,
+      "loss_ce": 0.008480816148221493,
+      "loss_iou": 1.0260534286499023,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 32660208,
+      "step": 189
+    },
+    {
+      "epoch": 0.07276905400229797,
+      "grad_norm": 23.282616528478933,
+      "learning_rate": 5e-06,
+      "loss": 0.4014,
+      "num_input_tokens_seen": 32833144,
+      "step": 190
+    },
+    {
+      "epoch": 0.07276905400229797,
+      "loss": 0.38087987899780273,
+      "loss_ce": 0.007832997478544712,
+      "loss_iou": 1.0457580089569092,
+      "loss_num": 0.373046875,
+      "loss_xval": 0.373046875,
+      "num_input_tokens_seen": 32833144,
+      "step": 190
+    },
+    {
+      "epoch": 0.0731520490233627,
+      "grad_norm": 36.93242731966085,
+      "learning_rate": 5e-06,
+      "loss": 0.4199,
+      "num_input_tokens_seen": 33006520,
+      "step": 191
+    },
+    {
+      "epoch": 0.0731520490233627,
+      "loss": 0.46541857719421387,
+      "loss_ce": 0.007899067364633083,
+      "loss_iou": 1.0965670347213745,
+      "loss_num": 0.45703125,
+      "loss_xval": 0.45703125,
+      "num_input_tokens_seen": 33006520,
+      "step": 191
+    },
+    {
+      "epoch": 0.07353504404442743,
+      "grad_norm": 75.6533228884653,
+      "learning_rate": 5e-06,
+      "loss": 0.4512,
+      "num_input_tokens_seen": 33179088,
+      "step": 192
+    },
+    {
+      "epoch": 0.07353504404442743,
+      "loss": 0.4501715898513794,
+      "loss_ce": 0.007056348491460085,
+      "loss_iou": 1.0614676475524902,
+      "loss_num": 0.443359375,
+      "loss_xval": 0.443359375,
+      "num_input_tokens_seen": 33179088,
+      "step": 192
+    },
+    {
+      "epoch": 0.07391803906549214,
+      "grad_norm": 25.310524245622176,
+      "learning_rate": 5e-06,
+      "loss": 0.3013,
+      "num_input_tokens_seen": 33351976,
+      "step": 193
+    },
+    {
+      "epoch": 0.07391803906549214,
+      "loss": 0.3208417594432831,
+      "loss_ce": 0.007609330117702484,
+      "loss_iou": 1.0761098861694336,
+      "loss_num": 0.3125,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 33351976,
+      "step": 193
+    },
+    {
+      "epoch": 0.07430103408655687,
+      "grad_norm": 25.649629507305033,
+      "learning_rate": 5e-06,
+      "loss": 0.4841,
+      "num_input_tokens_seen": 33525168,
+      "step": 194
+    },
+    {
+      "epoch": 0.07430103408655687,
+      "loss": 0.5254085063934326,
+      "loss_ce": 0.007342115510255098,
+      "loss_iou": 1.126143455505371,
+      "loss_num": 0.51953125,
+      "loss_xval": 0.51953125,
+      "num_input_tokens_seen": 33525168,
+      "step": 194
+    },
+    {
+      "epoch": 0.0746840291076216,
+      "grad_norm": 29.68138193030563,
+      "learning_rate": 5e-06,
+      "loss": 0.3128,
+      "num_input_tokens_seen": 33698232,
+      "step": 195
+    },
+    {
+      "epoch": 0.0746840291076216,
+      "loss": 0.3225231170654297,
+      "loss_ce": 0.007825830951333046,
+      "loss_iou": 1.0991780757904053,
+      "loss_num": 0.314453125,
+      "loss_xval": 0.314453125,
+      "num_input_tokens_seen": 33698232,
+      "step": 195
+    },
+    {
+      "epoch": 0.07506702412868632,
+      "grad_norm": 37.14601624704469,
+      "learning_rate": 5e-06,
+      "loss": 0.4957,
+      "num_input_tokens_seen": 33867800,
+      "step": 196
+    },
+    {
+      "epoch": 0.07506702412868632,
+      "loss": 0.4901979863643646,
+      "loss_ce": 0.007776104845106602,
+      "loss_iou": 1.0547126531600952,
+      "loss_num": 0.482421875,
+      "loss_xval": 0.482421875,
+      "num_input_tokens_seen": 33867800,
+      "step": 196
+    },
+    {
+      "epoch": 0.07545001914975105,
+      "grad_norm": 38.09017228687531,
+      "learning_rate": 5e-06,
+      "loss": 0.4282,
+      "num_input_tokens_seen": 34041192,
+      "step": 197
+    },
+    {
+      "epoch": 0.07545001914975105,
+      "loss": 0.4419156312942505,
+      "loss_ce": 0.008321871981024742,
+      "loss_iou": 1.1644515991210938,
+      "loss_num": 0.43359375,
+      "loss_xval": 0.43359375,
+      "num_input_tokens_seen": 34041192,
+      "step": 197
+    },
+    {
+      "epoch": 0.07583301417081578,
+      "grad_norm": 29.58474207677991,
+      "learning_rate": 5e-06,
+      "loss": 0.5765,
+      "num_input_tokens_seen": 34214016,
+      "step": 198
+    },
+    {
+      "epoch": 0.07583301417081578,
+      "loss": 0.5875325798988342,
+      "loss_ce": 0.00818687118589878,
+      "loss_iou": 1.0438282489776611,
+      "loss_num": 0.578125,
+      "loss_xval": 0.578125,
+      "num_input_tokens_seen": 34214016,
+      "step": 198
+    },
+    {
+      "epoch": 0.0762160091918805,
+      "grad_norm": 24.19820119095497,
+      "learning_rate": 5e-06,
+      "loss": 0.2809,
+      "num_input_tokens_seen": 34387072,
+      "step": 199
+    },
+    {
+      "epoch": 0.0762160091918805,
+      "loss": 0.2809727191925049,
+      "loss_ce": 0.007901410572230816,
+      "loss_iou": 1.0663940906524658,
+      "loss_num": 0.2734375,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 34387072,
+      "step": 199
+    },
+    {
+      "epoch": 0.07659900421294523,
+      "grad_norm": 49.97462063305508,
+      "learning_rate": 5e-06,
+      "loss": 0.3485,
+      "num_input_tokens_seen": 34559736,
+      "step": 200
+    },
+    {
+      "epoch": 0.07659900421294523,
+      "loss": 0.35517212748527527,
+      "loss_ce": 0.007760013919323683,
+      "loss_iou": 1.0157616138458252,
+      "loss_num": 0.34765625,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 34559736,
+      "step": 200
+    },
+    {
+      "epoch": 0.07698199923400996,
+      "grad_norm": 43.16095857177117,
+      "learning_rate": 5e-06,
+      "loss": 0.6135,
+      "num_input_tokens_seen": 34732608,
+      "step": 201
+    },
+    {
+      "epoch": 0.07698199923400996,
+      "loss": 0.5544754266738892,
+      "loss_ce": 0.007112160325050354,
+      "loss_iou": 1.0241763591766357,
+      "loss_num": 0.546875,
+      "loss_xval": 0.546875,
+      "num_input_tokens_seen": 34732608,
+      "step": 201
+    },
+    {
+      "epoch": 0.07736499425507469,
+      "grad_norm": 20.95385839326255,
+      "learning_rate": 5e-06,
+      "loss": 0.4632,
+      "num_input_tokens_seen": 34905488,
+      "step": 202
+    },
+    {
+      "epoch": 0.07736499425507469,
+      "loss": 0.48185425996780396,
+      "loss_ce": 0.007977311499416828,
+      "loss_iou": 1.038071632385254,
+      "loss_num": 0.474609375,
+      "loss_xval": 0.474609375,
+      "num_input_tokens_seen": 34905488,
+      "step": 202
+    },
+    {
+      "epoch": 0.0777479892761394,
+      "grad_norm": 56.181913423999944,
+      "learning_rate": 5e-06,
+      "loss": 0.6183,
+      "num_input_tokens_seen": 35078584,
+      "step": 203
+    },
+    {
+      "epoch": 0.0777479892761394,
+      "loss": 0.56095290184021,
+      "loss_ce": 0.00846262089908123,
+      "loss_iou": 1.0347716808319092,
+      "loss_num": 0.55078125,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 35078584,
+      "step": 203
+    },
+    {
+      "epoch": 0.07813098429720414,
+      "grad_norm": 56.88320874646355,
+      "learning_rate": 5e-06,
+      "loss": 0.892,
+      "num_input_tokens_seen": 35251872,
+      "step": 204
+    },
+    {
+      "epoch": 0.07813098429720414,
+      "loss": 0.9481920003890991,
+      "loss_ce": 0.0077622984535992146,
+      "loss_iou": 1.1150736808776855,
+      "loss_num": 0.94140625,
+      "loss_xval": 0.94140625,
+      "num_input_tokens_seen": 35251872,
+      "step": 204
+    },
+    {
+      "epoch": 0.07851397931826887,
+      "grad_norm": 47.13745838175376,
+      "learning_rate": 5e-06,
+      "loss": 0.5414,
+      "num_input_tokens_seen": 35424672,
+      "step": 205
+    },
+    {
+      "epoch": 0.07851397931826887,
+      "loss": 0.5352663397789001,
+      "loss_ce": 0.008166728541254997,
+      "loss_iou": 1.032301425933838,
+      "loss_num": 0.52734375,
+      "loss_xval": 0.52734375,
+      "num_input_tokens_seen": 35424672,
+      "step": 205
+    },
+    {
+      "epoch": 0.07889697433933358,
+      "grad_norm": 32.202532364665835,
+      "learning_rate": 5e-06,
+      "loss": 0.9978,
+      "num_input_tokens_seen": 35597600,
+      "step": 206
+    },
+    {
+      "epoch": 0.07889697433933358,
+      "loss": 1.0033421516418457,
+      "loss_ce": 0.008225006982684135,
+      "loss_iou": 1.0893646478652954,
+      "loss_num": 0.99609375,
+      "loss_xval": 0.99609375,
+      "num_input_tokens_seen": 35597600,
+      "step": 206
+    },
+    {
+      "epoch": 0.07927996936039831,
+      "grad_norm": 27.851259730721388,
+      "learning_rate": 5e-06,
+      "loss": 0.5275,
+      "num_input_tokens_seen": 35770688,
+      "step": 207
+    },
+    {
+      "epoch": 0.07927996936039831,
+      "loss": 0.505529522895813,
+      "loss_ce": 0.007726819254457951,
+      "loss_iou": 1.0436320304870605,
+      "loss_num": 0.498046875,
+      "loss_xval": 0.498046875,
+      "num_input_tokens_seen": 35770688,
+      "step": 207
+    },
+    {
+      "epoch": 0.07966296438146304,
+      "grad_norm": 67.75949609941692,
+      "learning_rate": 5e-06,
+      "loss": 0.7863,
+      "num_input_tokens_seen": 35943440,
+      "step": 208
+    },
+    {
+      "epoch": 0.07966296438146304,
+      "loss": 0.7841857671737671,
+      "loss_ce": 0.0073302993550896645,
+      "loss_iou": 1.0888667106628418,
+      "loss_num": 0.77734375,
+      "loss_xval": 0.77734375,
+      "num_input_tokens_seen": 35943440,
+      "step": 208
+    },
+    {
+      "epoch": 0.08004595940252776,
+      "grad_norm": 58.05924815659133,
+      "learning_rate": 5e-06,
+      "loss": 0.9796,
+      "num_input_tokens_seen": 36116504,
+      "step": 209
+    },
+    {
+      "epoch": 0.08004595940252776,
+      "loss": 0.9908746480941772,
+      "loss_ce": 0.00845281220972538,
+      "loss_iou": 1.0735149383544922,
+      "loss_num": 0.984375,
+      "loss_xval": 0.984375,
+      "num_input_tokens_seen": 36116504,
+      "step": 209
+    },
+    {
+      "epoch": 0.08042895442359249,
+      "grad_norm": 27.12379421689299,
+      "learning_rate": 5e-06,
+      "loss": 0.7469,
+      "num_input_tokens_seen": 36289384,
+      "step": 210
+    },
+    {
+      "epoch": 0.08042895442359249,
+      "loss": 0.7505313158035278,
+      "loss_ce": 0.008343802765011787,
+      "loss_iou": 1.218379020690918,
+      "loss_num": 0.7421875,
+      "loss_xval": 0.7421875,
+      "num_input_tokens_seen": 36289384,
+      "step": 210
+    },
+    {
+      "epoch": 0.08081194944465722,
+      "grad_norm": 45.08058931446742,
+      "learning_rate": 5e-06,
+      "loss": 0.6479,
+      "num_input_tokens_seen": 36462232,
+      "step": 211
+    },
+    {
+      "epoch": 0.08081194944465722,
+      "loss": 0.6172372102737427,
+      "loss_ce": 0.00835050456225872,
+      "loss_iou": 1.0919554233551025,
+      "loss_num": 0.609375,
+      "loss_xval": 0.609375,
+      "num_input_tokens_seen": 36462232,
+      "step": 211
+    },
+    {
+      "epoch": 0.08119494446572195,
+      "grad_norm": 68.53234771774173,
+      "learning_rate": 5e-06,
+      "loss": 1.0,
+      "num_input_tokens_seen": 36634992,
+      "step": 212
+    },
+    {
+      "epoch": 0.08119494446572195,
+      "loss": 0.9534499645233154,
+      "loss_ce": 0.008137499913573265,
+      "loss_iou": 1.1350417137145996,
+      "loss_num": 0.9453125,
+      "loss_xval": 0.9453125,
+      "num_input_tokens_seen": 36634992,
+      "step": 212
+    },
+    {
+      "epoch": 0.08157793948678667,
+      "grad_norm": 60.45343021219607,
+      "learning_rate": 5e-06,
+      "loss": 0.8913,
+      "num_input_tokens_seen": 36807480,
+      "step": 213
+    },
+    {
+      "epoch": 0.08157793948678667,
+      "loss": 0.8964667320251465,
+      "loss_ce": 0.008771365508437157,
+      "loss_iou": 1.0690150260925293,
+      "loss_num": 0.88671875,
+      "loss_xval": 0.88671875,
+      "num_input_tokens_seen": 36807480,
+      "step": 213
+    },
+    {
+      "epoch": 0.0819609345078514,
+      "grad_norm": 31.560015909623697,
+      "learning_rate": 5e-06,
+      "loss": 1.0086,
+      "num_input_tokens_seen": 36980544,
+      "step": 214
+    },
+    {
+      "epoch": 0.0819609345078514,
+      "loss": 0.9310482144355774,
+      "loss_ce": 0.00868492852896452,
+      "loss_iou": 1.2417666912078857,
+      "loss_num": 0.921875,
+      "loss_xval": 0.921875,
+      "num_input_tokens_seen": 36980544,
+      "step": 214
+    },
+    {
+      "epoch": 0.08234392952891613,
+      "grad_norm": 52.97214741803255,
+      "learning_rate": 5e-06,
+      "loss": 0.6637,
+      "num_input_tokens_seen": 37153688,
+      "step": 215
+    },
+    {
+      "epoch": 0.08234392952891613,
+      "loss": 0.6983016729354858,
+      "loss_ce": 0.009336812421679497,
+      "loss_iou": 1.079323172569275,
+      "loss_num": 0.6875,
+      "loss_xval": 0.6875,
+      "num_input_tokens_seen": 37153688,
+      "step": 215
+    },
+    {
+      "epoch": 0.08272692454998085,
+      "grad_norm": 60.17389726110234,
+      "learning_rate": 5e-06,
+      "loss": 0.843,
+      "num_input_tokens_seen": 37323184,
+      "step": 216
+    },
+    {
+      "epoch": 0.08272692454998085,
+      "loss": 0.8084574937820435,
+      "loss_ce": 0.00816454365849495,
+      "loss_iou": 1.069022297859192,
+      "loss_num": 0.80078125,
+      "loss_xval": 0.80078125,
+      "num_input_tokens_seen": 37323184,
+      "step": 216
+    },
+    {
+      "epoch": 0.08310991957104558,
+      "grad_norm": 57.111171054147256,
+      "learning_rate": 5e-06,
+      "loss": 0.5962,
+      "num_input_tokens_seen": 37495968,
+      "step": 217
+    },
+    {
+      "epoch": 0.08310991957104558,
+      "loss": 0.6317141056060791,
+      "loss_ce": 0.009155533276498318,
+      "loss_iou": 1.0423400402069092,
+      "loss_num": 0.62109375,
+      "loss_xval": 0.62109375,
+      "num_input_tokens_seen": 37495968,
+      "step": 217
+    },
+    {
+      "epoch": 0.0834929145921103,
+      "grad_norm": 50.41203884024478,
+      "learning_rate": 5e-06,
+      "loss": 1.0768,
+      "num_input_tokens_seen": 37669000,
+      "step": 218
+    },
+    {
+      "epoch": 0.0834929145921103,
+      "loss": 1.0677589178085327,
+      "loss_ce": 0.009653490036725998,
+      "loss_iou": 1.2282007932662964,
+      "loss_num": 1.0546875,
+      "loss_xval": 1.0546875,
+      "num_input_tokens_seen": 37669000,
+      "step": 218
+    },
+    {
+      "epoch": 0.08387590961317502,
+      "grad_norm": 43.272735637508,
+      "learning_rate": 5e-06,
+      "loss": 0.443,
+      "num_input_tokens_seen": 37842168,
+      "step": 219
+    },
+    {
+      "epoch": 0.08387590961317502,
+      "loss": 0.46490222215652466,
+      "loss_ce": 0.009335814043879509,
+      "loss_iou": 1.046978235244751,
+      "loss_num": 0.455078125,
+      "loss_xval": 0.455078125,
+      "num_input_tokens_seen": 37842168,
+      "step": 219
+    },
+    {
+      "epoch": 0.08425890463423975,
+      "grad_norm": 52.78682273165674,
+      "learning_rate": 5e-06,
+      "loss": 0.9541,
+      "num_input_tokens_seen": 38015080,
+      "step": 220
+    },
+    {
+      "epoch": 0.08425890463423975,
+      "loss": 0.9495786428451538,
+      "loss_ce": 0.008660627529025078,
+      "loss_iou": 1.1405905485153198,
+      "loss_num": 0.94140625,
+      "loss_xval": 0.94140625,
+      "num_input_tokens_seen": 38015080,
+      "step": 220
+    },
+    {
+      "epoch": 0.08464189965530448,
+      "grad_norm": 47.76023451785587,
+      "learning_rate": 5e-06,
+      "loss": 0.6282,
+      "num_input_tokens_seen": 38188104,
+      "step": 221
+    },
+    {
+      "epoch": 0.08464189965530448,
+      "loss": 0.6390358209609985,
+      "loss_ce": 0.009885435923933983,
+      "loss_iou": 1.0816030502319336,
+      "loss_num": 0.62890625,
+      "loss_xval": 0.62890625,
+      "num_input_tokens_seen": 38188104,
+      "step": 221
+    },
+    {
+      "epoch": 0.0850248946763692,
+      "grad_norm": 39.6378992368006,
+      "learning_rate": 5e-06,
+      "loss": 0.8432,
+      "num_input_tokens_seen": 38361440,
+      "step": 222
+    },
+    {
+      "epoch": 0.0850248946763692,
+      "loss": 0.8081187009811401,
+      "loss_ce": 0.009778868407011032,
+      "loss_iou": 1.0645579099655151,
+      "loss_num": 0.796875,
+      "loss_xval": 0.796875,
+      "num_input_tokens_seen": 38361440,
+      "step": 222
+    },
+    {
+      "epoch": 0.08540788969743393,
+      "grad_norm": 27.74761331621742,
+      "learning_rate": 5e-06,
+      "loss": 0.4048,
+      "num_input_tokens_seen": 38534360,
+      "step": 223
+    },
+    {
+      "epoch": 0.08540788969743393,
+      "loss": 0.42580628395080566,
+      "loss_ce": 0.00881408154964447,
+      "loss_iou": 1.0770916938781738,
+      "loss_num": 0.41796875,
+      "loss_xval": 0.41796875,
+      "num_input_tokens_seen": 38534360,
+      "step": 223
+    },
+    {
+      "epoch": 0.08579088471849866,
+      "grad_norm": 48.754843653913895,
+      "learning_rate": 5e-06,
+      "loss": 0.5333,
+      "num_input_tokens_seen": 38707240,
+      "step": 224
+    },
+    {
+      "epoch": 0.08579088471849866,
+      "loss": 0.520818829536438,
+      "loss_ce": 0.009588375687599182,
+      "loss_iou": 1.0615181922912598,
+      "loss_num": 0.51171875,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 38707240,
+      "step": 224
+    },
+    {
+      "epoch": 0.08617387973956339,
+      "grad_norm": 35.67101145239377,
+      "learning_rate": 5e-06,
+      "loss": 0.4983,
+      "num_input_tokens_seen": 38879976,
+      "step": 225
+    },
+    {
+      "epoch": 0.08617387973956339,
+      "loss": 0.5113679766654968,
+      "loss_ce": 0.009414846077561378,
+      "loss_iou": 1.0937516689300537,
+      "loss_num": 0.5,
+      "loss_xval": 0.5,
+      "num_input_tokens_seen": 38879976,
+      "step": 225
+    },
+    {
+      "epoch": 0.08655687476062811,
+      "grad_norm": 37.86353245054712,
+      "learning_rate": 5e-06,
+      "loss": 0.6152,
+      "num_input_tokens_seen": 39052584,
+      "step": 226
+    },
+    {
+      "epoch": 0.08655687476062811,
+      "loss": 0.5857442021369934,
+      "loss_ce": 0.01054888591170311,
+      "loss_iou": 1.019634485244751,
+      "loss_num": 0.57421875,
+      "loss_xval": 0.57421875,
+      "num_input_tokens_seen": 39052584,
+      "step": 226
+    },
+    {
+      "epoch": 0.08693986978169284,
+      "grad_norm": 41.492479478682746,
+      "learning_rate": 5e-06,
+      "loss": 0.6529,
+      "num_input_tokens_seen": 39225600,
+      "step": 227
+    },
+    {
+      "epoch": 0.08693986978169284,
+      "loss": 0.6377010941505432,
+      "loss_ce": 0.010015567764639854,
+      "loss_iou": 1.1252610683441162,
+      "loss_num": 0.62890625,
+      "loss_xval": 0.62890625,
+      "num_input_tokens_seen": 39225600,
+      "step": 227
+    },
+    {
+      "epoch": 0.08732286480275757,
+      "grad_norm": 36.925861584920455,
+      "learning_rate": 5e-06,
+      "loss": 0.6026,
+      "num_input_tokens_seen": 39398648,
+      "step": 228
+    },
+    {
+      "epoch": 0.08732286480275757,
+      "loss": 0.6346437335014343,
+      "loss_ce": 0.009643702767789364,
+      "loss_iou": 0.9308451414108276,
+      "loss_num": 0.625,
+      "loss_xval": 0.625,
+      "num_input_tokens_seen": 39398648,
+      "step": 228
+    },
+    {
+      "epoch": 0.08770585982382229,
+      "grad_norm": 33.41380847989801,
+      "learning_rate": 5e-06,
+      "loss": 0.331,
+      "num_input_tokens_seen": 39571560,
+      "step": 229
+    },
+    {
+      "epoch": 0.08770585982382229,
+      "loss": 0.31973588466644287,
+      "loss_ce": 0.010531795211136341,
+      "loss_iou": 1.2544775009155273,
+      "loss_num": 0.30859375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 39571560,
+      "step": 229
+    },
+    {
+      "epoch": 0.08808885484488702,
+      "grad_norm": 37.27004984198461,
+      "learning_rate": 5e-06,
+      "loss": 0.689,
+      "num_input_tokens_seen": 39741216,
+      "step": 230
+    },
+    {
+      "epoch": 0.08808885484488702,
+      "loss": 0.6989870071411133,
+      "loss_ce": 0.010510453954339027,
+      "loss_iou": 1.0752999782562256,
+      "loss_num": 0.6875,
+      "loss_xval": 0.6875,
+      "num_input_tokens_seen": 39741216,
+      "step": 230
+    },
+    {
+      "epoch": 0.08847184986595175,
+      "grad_norm": 34.237033316216014,
+      "learning_rate": 5e-06,
+      "loss": 0.3351,
+      "num_input_tokens_seen": 39914552,
+      "step": 231
+    },
+    {
+      "epoch": 0.08847184986595175,
+      "loss": 0.2838922142982483,
+      "loss_ce": 0.010820926167070866,
+      "loss_iou": 1.011523723602295,
+      "loss_num": 0.2734375,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 39914552,
+      "step": 231
+    },
+    {
+      "epoch": 0.08885484488701646,
+      "grad_norm": 28.558703603816248,
+      "learning_rate": 5e-06,
+      "loss": 0.6381,
+      "num_input_tokens_seen": 40087440,
+      "step": 232
+    },
+    {
+      "epoch": 0.08885484488701646,
+      "loss": 0.6444008946418762,
+      "loss_ce": 0.010123548097908497,
+      "loss_iou": 1.0606837272644043,
+      "loss_num": 0.6328125,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 40087440,
+      "step": 232
+    },
+    {
+      "epoch": 0.08923783990808119,
+      "grad_norm": 20.711593050355276,
+      "learning_rate": 5e-06,
+      "loss": 0.2953,
+      "num_input_tokens_seen": 40260120,
+      "step": 233
+    },
+    {
+      "epoch": 0.08923783990808119,
+      "loss": 0.26176732778549194,
+      "loss_ce": 0.009936293587088585,
+      "loss_iou": 0.9967314004898071,
+      "loss_num": 0.251953125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 40260120,
+      "step": 233
+    },
+    {
+      "epoch": 0.08962083492914592,
+      "grad_norm": 39.27223566087036,
+      "learning_rate": 5e-06,
+      "loss": 0.4499,
+      "num_input_tokens_seen": 40433352,
+      "step": 234
+    },
+    {
+      "epoch": 0.08962083492914592,
+      "loss": 0.37641066312789917,
+      "loss_ce": 0.009711459279060364,
+      "loss_iou": 1.0314005613327026,
+      "loss_num": 0.3671875,
+      "loss_xval": 0.3671875,
+      "num_input_tokens_seen": 40433352,
+      "step": 234
+    },
+    {
+      "epoch": 0.09000382995021065,
+      "grad_norm": 44.82076244879522,
+      "learning_rate": 5e-06,
+      "loss": 0.5356,
+      "num_input_tokens_seen": 40606288,
+      "step": 235
+    },
+    {
+      "epoch": 0.09000382995021065,
+      "loss": 0.514398992061615,
+      "loss_ce": 0.00951617956161499,
+      "loss_iou": 1.0371146202087402,
+      "loss_num": 0.50390625,
+      "loss_xval": 0.50390625,
+      "num_input_tokens_seen": 40606288,
+      "step": 235
+    },
+    {
+      "epoch": 0.09038682497127537,
+      "grad_norm": 23.730780034423823,
+      "learning_rate": 5e-06,
+      "loss": 0.5958,
+      "num_input_tokens_seen": 40779264,
+      "step": 236
+    },
+    {
+      "epoch": 0.09038682497127537,
+      "loss": 0.6494039297103882,
+      "loss_ce": 0.009755544364452362,
+      "loss_iou": 1.2160544395446777,
+      "loss_num": 0.640625,
+      "loss_xval": 0.640625,
+      "num_input_tokens_seen": 40779264,
+      "step": 236
+    },
+    {
+      "epoch": 0.0907698199923401,
+      "grad_norm": 39.65756158381604,
+      "learning_rate": 5e-06,
+      "loss": 0.3562,
+      "num_input_tokens_seen": 40948744,
+      "step": 237
+    },
+    {
+      "epoch": 0.0907698199923401,
+      "loss": 0.3717811405658722,
+      "loss_ce": 0.010453008115291595,
+      "loss_iou": 1.011744737625122,
+      "loss_num": 0.361328125,
+      "loss_xval": 0.361328125,
+      "num_input_tokens_seen": 40948744,
+      "step": 237
+    },
+    {
+      "epoch": 0.09115281501340483,
+      "grad_norm": 33.31493995502488,
+      "learning_rate": 5e-06,
+      "loss": 0.4441,
+      "num_input_tokens_seen": 41121688,
+      "step": 238
+    },
+    {
+      "epoch": 0.09115281501340483,
+      "loss": 0.47466912865638733,
+      "loss_ce": 0.009825348854064941,
+      "loss_iou": 1.0839557647705078,
+      "loss_num": 0.46484375,
+      "loss_xval": 0.46484375,
+      "num_input_tokens_seen": 41121688,
+      "step": 238
+    },
+    {
+      "epoch": 0.09153581003446955,
+      "grad_norm": 39.187070010958436,
+      "learning_rate": 5e-06,
+      "loss": 0.4312,
+      "num_input_tokens_seen": 41294488,
+      "step": 239
+    },
+    {
+      "epoch": 0.09153581003446955,
+      "loss": 0.4678073227405548,
+      "loss_ce": 0.009311222471296787,
+      "loss_iou": 1.0262682437896729,
+      "loss_num": 0.458984375,
+      "loss_xval": 0.458984375,
+      "num_input_tokens_seen": 41294488,
+      "step": 239
+    },
+    {
+      "epoch": 0.09191880505553428,
+      "grad_norm": 30.268551043091506,
+      "learning_rate": 5e-06,
+      "loss": 0.5959,
+      "num_input_tokens_seen": 41467400,
+      "step": 240
+    },
+    {
+      "epoch": 0.09191880505553428,
+      "loss": 0.6431682109832764,
+      "loss_ce": 0.010111584328114986,
+      "loss_iou": 1.150076150894165,
+      "loss_num": 0.6328125,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 41467400,
+      "step": 240
+    },
+    {
+      "epoch": 0.09230180007659901,
+      "grad_norm": 29.354480728493197,
+      "learning_rate": 5e-06,
+      "loss": 0.352,
+      "num_input_tokens_seen": 41640488,
+      "step": 241
+    },
+    {
+      "epoch": 0.09230180007659901,
+      "loss": 0.3974408805370331,
+      "loss_ce": 0.009989721700549126,
+      "loss_iou": 1.3911032676696777,
+      "loss_num": 0.38671875,
+      "loss_xval": 0.38671875,
+      "num_input_tokens_seen": 41640488,
+      "step": 241
+    },
+    {
+      "epoch": 0.09268479509766372,
+      "grad_norm": 57.10524842136358,
+      "learning_rate": 5e-06,
+      "loss": 0.6226,
+      "num_input_tokens_seen": 41813624,
+      "step": 242
+    },
+    {
+      "epoch": 0.09268479509766372,
+      "loss": 0.6125884056091309,
+      "loss_ce": 0.010049371980130672,
+      "loss_iou": 1.063937783241272,
+      "loss_num": 0.6015625,
+      "loss_xval": 0.6015625,
+      "num_input_tokens_seen": 41813624,
+      "step": 242
+    },
+    {
+      "epoch": 0.09306779011872846,
+      "grad_norm": 45.84410776062516,
+      "learning_rate": 5e-06,
+      "loss": 0.5885,
+      "num_input_tokens_seen": 41986624,
+      "step": 243
+    },
+    {
+      "epoch": 0.09306779011872846,
+      "loss": 0.5837477445602417,
+      "loss_ce": 0.00977311935275793,
+      "loss_iou": 1.0533053874969482,
+      "loss_num": 0.57421875,
+      "loss_xval": 0.57421875,
+      "num_input_tokens_seen": 41986624,
+      "step": 243
+    },
+    {
+      "epoch": 0.09345078513979319,
+      "grad_norm": 32.262416599010756,
+      "learning_rate": 5e-06,
+      "loss": 0.6973,
+      "num_input_tokens_seen": 42159784,
+      "step": 244
+    },
+    {
+      "epoch": 0.09345078513979319,
+      "loss": 0.7056179642677307,
+      "loss_ce": 0.010305466130375862,
+      "loss_iou": 1.114302396774292,
+      "loss_num": 0.6953125,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 42159784,
+      "step": 244
+    },
+    {
+      "epoch": 0.0938337801608579,
+      "grad_norm": 48.773859092039544,
+      "learning_rate": 5e-06,
+      "loss": 0.6242,
+      "num_input_tokens_seen": 42332736,
+      "step": 245
+    },
+    {
+      "epoch": 0.0938337801608579,
+      "loss": 0.668730616569519,
+      "loss_ce": 0.009306754916906357,
+      "loss_iou": 1.2523534297943115,
+      "loss_num": 0.66015625,
+      "loss_xval": 0.66015625,
+      "num_input_tokens_seen": 42332736,
+      "step": 245
+    },
+    {
+      "epoch": 0.09421677518192263,
+      "grad_norm": 60.37995512607071,
+      "learning_rate": 5e-06,
+      "loss": 0.738,
+      "num_input_tokens_seen": 42505896,
+      "step": 246
+    },
+    {
+      "epoch": 0.09421677518192263,
+      "loss": 0.7233730554580688,
+      "loss_ce": 0.00950586423277855,
+      "loss_iou": 1.0678024291992188,
+      "loss_num": 0.71484375,
+      "loss_xval": 0.71484375,
+      "num_input_tokens_seen": 42505896,
+      "step": 246
+    },
+    {
+      "epoch": 0.09459977020298736,
+      "grad_norm": 34.59265204812187,
+      "learning_rate": 5e-06,
+      "loss": 0.4118,
+      "num_input_tokens_seen": 42678720,
+      "step": 247
+    },
+    {
+      "epoch": 0.09459977020298736,
+      "loss": 0.4069308936595917,
+      "loss_ce": 0.009836145676672459,
+      "loss_iou": 1.0240042209625244,
+      "loss_num": 0.396484375,
+      "loss_xval": 0.396484375,
+      "num_input_tokens_seen": 42678720,
+      "step": 247
+    },
+    {
+      "epoch": 0.0949827652240521,
+      "grad_norm": 25.429298668066338,
+      "learning_rate": 5e-06,
+      "loss": 0.5488,
+      "num_input_tokens_seen": 42851568,
+      "step": 248
+    },
+    {
+      "epoch": 0.0949827652240521,
+      "loss": 0.5835607051849365,
+      "loss_ce": 0.010806784965097904,
+      "loss_iou": 1.0707110166549683,
+      "loss_num": 0.57421875,
+      "loss_xval": 0.57421875,
+      "num_input_tokens_seen": 42851568,
+      "step": 248
+    },
+    {
+      "epoch": 0.09536576024511681,
+      "grad_norm": 33.49257412615864,
+      "learning_rate": 5e-06,
+      "loss": 0.5205,
+      "num_input_tokens_seen": 43024512,
+      "step": 249
+    },
+    {
+      "epoch": 0.09536576024511681,
+      "loss": 0.5233336687088013,
+      "loss_ce": 0.01063838042318821,
+      "loss_iou": 1.0241611003875732,
+      "loss_num": 0.51171875,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 43024512,
+      "step": 249
+    },
+    {
+      "epoch": 0.09574875526618154,
+      "grad_norm": 35.51917126089131,
+      "learning_rate": 5e-06,
+      "loss": 0.7009,
+      "num_input_tokens_seen": 43193736,
+      "step": 250
+    },
+    {
+      "epoch": 0.09574875526618154,
+      "eval_websight_new_CIoU": 0.7970602214336395,
+      "eval_websight_new_GIoU": 0.7953665852546692,
+      "eval_websight_new_IoU": 0.7986245453357697,
+      "eval_websight_new_MAE_all": 0.01940658688545227,
+      "eval_websight_new_MAE_h": 0.02148013934493065,
+      "eval_websight_new_MAE_w": 0.023069137707352638,
+      "eval_websight_new_MAE_x": 0.014143784064799547,
+      "eval_websight_new_MAE_y": 0.01893328595906496,
+      "eval_websight_new_NUM_probability": 0.012394892051815987,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.24325791001319885,
+      "eval_websight_new_loss_ce": 0.02025469858199358,
+      "eval_websight_new_loss_iou": 1.0012649893760681,
+      "eval_websight_new_loss_num": 0.2210693359375,
+      "eval_websight_new_loss_xval": 0.2210693359375,
+      "eval_websight_new_runtime": 59.1202,
+      "eval_websight_new_samples_per_second": 0.846,
+      "eval_websight_new_steps_per_second": 0.034,
+      "num_input_tokens_seen": 43193736,
+      "step": 250
+    },
+    {
+      "epoch": 0.09574875526618154,
+      "eval_seeclick_CIoU": 0.4883647561073303,
+      "eval_seeclick_GIoU": 0.48301491141319275,
+      "eval_seeclick_IoU": 0.5227425396442413,
+      "eval_seeclick_MAE_all": 0.08011750504374504,
+      "eval_seeclick_MAE_h": 0.07117656245827675,
+      "eval_seeclick_MAE_w": 0.09652542322874069,
+      "eval_seeclick_MAE_x": 0.08502181246876717,
+      "eval_seeclick_MAE_y": 0.067746227607131,
+      "eval_seeclick_NUM_probability": 0.019341954961419106,
+      "eval_seeclick_inside_bbox": 0.8541666567325592,
+      "eval_seeclick_loss": 0.6910960674285889,
+      "eval_seeclick_loss_ce": 0.014814576599746943,
+      "eval_seeclick_loss_iou": 1.142272412776947,
+      "eval_seeclick_loss_num": 0.6513671875,
+      "eval_seeclick_loss_xval": 0.6513671875,
+      "eval_seeclick_runtime": 82.318,
+      "eval_seeclick_samples_per_second": 0.607,
+      "eval_seeclick_steps_per_second": 0.024,
+      "num_input_tokens_seen": 43193736,
+      "step": 250
+    },
+    {
+      "epoch": 0.09574875526618154,
+      "eval_icons_CIoU": 0.7694402039051056,
+      "eval_icons_GIoU": 0.7592164576053619,
+      "eval_icons_IoU": 0.7789261639118195,
+      "eval_icons_MAE_all": 0.029270297847688198,
+      "eval_icons_MAE_h": 0.03659070935100317,
+      "eval_icons_MAE_w": 0.02758670412003994,
+      "eval_icons_MAE_x": 0.024439156521111727,
+      "eval_icons_MAE_y": 0.02846463117748499,
+      "eval_icons_NUM_probability": 0.008785805199295282,
+      "eval_icons_inside_bbox": 0.9722222089767456,
+      "eval_icons_loss": 0.24321779608726501,
+      "eval_icons_loss_ce": 0.014401643071323633,
+      "eval_icons_loss_iou": 1.022590696811676,
+      "eval_icons_loss_num": 0.21490478515625,
+      "eval_icons_loss_xval": 0.21490478515625,
+      "eval_icons_runtime": 82.7673,
+      "eval_icons_samples_per_second": 0.604,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 43193736,
+      "step": 250
+    },
+    {
+      "epoch": 0.09574875526618154,
+      "loss": 0.2483743131160736,
+      "loss_ce": 0.014487605541944504,
+      "loss_iou": 1.037798285484314,
+      "loss_num": 0.234375,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 43193736,
+      "step": 250
+    },
+    {
+      "epoch": 0.09613175028724627,
+      "grad_norm": 20.25484850817279,
+      "learning_rate": 5e-06,
+      "loss": 0.2744,
+      "num_input_tokens_seen": 43366592,
+      "step": 251
+    },
+    {
+      "epoch": 0.09613175028724627,
+      "loss": 0.2883433699607849,
+      "loss_ce": 0.01063341461122036,
+      "loss_iou": 1.0252397060394287,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 43366592,
+      "step": 251
+    },
+    {
+      "epoch": 0.09651474530831099,
+      "grad_norm": 25.324768920980308,
+      "learning_rate": 5e-06,
+      "loss": 0.6112,
+      "num_input_tokens_seen": 43539632,
+      "step": 252
+    },
+    {
+      "epoch": 0.09651474530831099,
+      "loss": 0.6387765407562256,
+      "loss_ce": 0.009626125916838646,
+      "loss_iou": 1.0179805755615234,
+      "loss_num": 0.62890625,
+      "loss_xval": 0.62890625,
+      "num_input_tokens_seen": 43539632,
+      "step": 252
+    },
+    {
+      "epoch": 0.09689774032937572,
+      "grad_norm": 25.420029732622826,
+      "learning_rate": 5e-06,
+      "loss": 0.3988,
+      "num_input_tokens_seen": 43712048,
+      "step": 253
+    },
+    {
+      "epoch": 0.09689774032937572,
+      "loss": 0.45570528507232666,
+      "loss_ce": 0.008927949704229832,
+      "loss_iou": 1.9686532020568848,
+      "loss_num": 0.447265625,
+      "loss_xval": 0.447265625,
+      "num_input_tokens_seen": 43712048,
+      "step": 253
+    },
+    {
+      "epoch": 0.09728073535044045,
+      "grad_norm": 47.35429105933384,
+      "learning_rate": 5e-06,
+      "loss": 0.5065,
+      "num_input_tokens_seen": 43885328,
+      "step": 254
+    },
+    {
+      "epoch": 0.09728073535044045,
+      "loss": 0.4618152976036072,
+      "loss_ce": 0.010887575335800648,
+      "loss_iou": 1.2346810102462769,
+      "loss_num": 0.451171875,
+      "loss_xval": 0.451171875,
+      "num_input_tokens_seen": 43885328,
+      "step": 254
+    },
+    {
+      "epoch": 0.09766373037150516,
+      "grad_norm": 32.21036374102037,
+      "learning_rate": 5e-06,
+      "loss": 0.3966,
+      "num_input_tokens_seen": 44058120,
+      "step": 255
+    },
+    {
+      "epoch": 0.09766373037150516,
+      "loss": 0.3424299359321594,
+      "loss_ce": 0.009666264057159424,
+      "loss_iou": 1.009257197380066,
+      "loss_num": 0.33203125,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 44058120,
+      "step": 255
+    },
+    {
+      "epoch": 0.0980467253925699,
+      "grad_norm": 31.43279394338899,
+      "learning_rate": 5e-06,
+      "loss": 0.5264,
+      "num_input_tokens_seen": 44231032,
+      "step": 256
+    },
+    {
+      "epoch": 0.0980467253925699,
+      "loss": 0.5027552843093872,
+      "loss_ce": 0.010323646478354931,
+      "loss_iou": 1.0422286987304688,
+      "loss_num": 0.4921875,
+      "loss_xval": 0.4921875,
+      "num_input_tokens_seen": 44231032,
+      "step": 256
+    },
+    {
+      "epoch": 0.09842972041363462,
+      "grad_norm": 40.30321116368787,
+      "learning_rate": 5e-06,
+      "loss": 0.4123,
+      "num_input_tokens_seen": 44404192,
+      "step": 257
+    },
+    {
+      "epoch": 0.09842972041363462,
+      "loss": 0.40247195959091187,
+      "loss_ce": 0.010382117703557014,
+      "loss_iou": 1.0118200778961182,
+      "loss_num": 0.392578125,
+      "loss_xval": 0.392578125,
+      "num_input_tokens_seen": 44404192,
+      "step": 257
+    },
+    {
+      "epoch": 0.09881271543469936,
+      "grad_norm": 34.974454284072436,
+      "learning_rate": 5e-06,
+      "loss": 0.5083,
+      "num_input_tokens_seen": 44573512,
+      "step": 258
+    },
+    {
+      "epoch": 0.09881271543469936,
+      "loss": 0.5065228343009949,
+      "loss_ce": 0.011161506175994873,
+      "loss_iou": 1.0581029653549194,
+      "loss_num": 0.49609375,
+      "loss_xval": 0.49609375,
+      "num_input_tokens_seen": 44573512,
+      "step": 258
+    },
+    {
+      "epoch": 0.09919571045576407,
+      "grad_norm": 19.965723552087525,
+      "learning_rate": 5e-06,
+      "loss": 0.2637,
+      "num_input_tokens_seen": 44746088,
+      "step": 259
+    },
+    {
+      "epoch": 0.09919571045576407,
+      "loss": 0.28278666734695435,
+      "loss_ce": 0.010081607848405838,
+      "loss_iou": 1.0371840000152588,
+      "loss_num": 0.2734375,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 44746088,
+      "step": 259
+    },
+    {
+      "epoch": 0.0995787054768288,
+      "grad_norm": 22.547722992093767,
+      "learning_rate": 5e-06,
+      "loss": 0.3887,
+      "num_input_tokens_seen": 44919376,
+      "step": 260
+    },
+    {
+      "epoch": 0.0995787054768288,
+      "loss": 0.42721009254455566,
+      "loss_ce": 0.010828271508216858,
+      "loss_iou": 1.0751303434371948,
+      "loss_num": 0.416015625,
+      "loss_xval": 0.416015625,
+      "num_input_tokens_seen": 44919376,
+      "step": 260
+    },
+    {
+      "epoch": 0.09996170049789353,
+      "grad_norm": 38.337449044979444,
+      "learning_rate": 5e-06,
+      "loss": 0.5143,
+      "num_input_tokens_seen": 45092656,
+      "step": 261
+    },
+    {
+      "epoch": 0.09996170049789353,
+      "loss": 0.5790233016014099,
+      "loss_ce": 0.010175647214055061,
+      "loss_iou": 1.0254803895950317,
+      "loss_num": 0.5703125,
+      "loss_xval": 0.5703125,
+      "num_input_tokens_seen": 45092656,
+      "step": 261
+    },
+    {
+      "epoch": 0.10034469551895825,
+      "grad_norm": 21.79110441345305,
+      "learning_rate": 5e-06,
+      "loss": 0.4789,
+      "num_input_tokens_seen": 45265336,
+      "step": 262
+    },
+    {
+      "epoch": 0.10034469551895825,
+      "loss": 0.49044251441955566,
+      "loss_ce": 0.009973737411201,
+      "loss_iou": 1.088423252105713,
+      "loss_num": 0.48046875,
+      "loss_xval": 0.48046875,
+      "num_input_tokens_seen": 45265336,
+      "step": 262
+    },
+    {
+      "epoch": 0.10072769054002298,
+      "grad_norm": 30.89290505138009,
+      "learning_rate": 5e-06,
+      "loss": 0.3411,
+      "num_input_tokens_seen": 45438600,
+      "step": 263
+    },
+    {
+      "epoch": 0.10072769054002298,
+      "loss": 0.37563595175743103,
+      "loss_ce": 0.010523635894060135,
+      "loss_iou": 1.0144340991973877,
+      "loss_num": 0.365234375,
+      "loss_xval": 0.365234375,
+      "num_input_tokens_seen": 45438600,
+      "step": 263
+    },
+    {
+      "epoch": 0.10111068556108771,
+      "grad_norm": 23.257658751783733,
+      "learning_rate": 5e-06,
+      "loss": 0.5822,
+      "num_input_tokens_seen": 45608112,
+      "step": 264
+    },
+    {
+      "epoch": 0.10111068556108771,
+      "loss": 0.5752701163291931,
+      "loss_ce": 0.010328728705644608,
+      "loss_iou": 1.0628750324249268,
+      "loss_num": 0.56640625,
+      "loss_xval": 0.56640625,
+      "num_input_tokens_seen": 45608112,
+      "step": 264
+    },
+    {
+      "epoch": 0.10149368058215243,
+      "grad_norm": 23.542558977591938,
+      "learning_rate": 5e-06,
+      "loss": 0.3597,
+      "num_input_tokens_seen": 45781272,
+      "step": 265
+    },
+    {
+      "epoch": 0.10149368058215243,
+      "loss": 0.3570312261581421,
+      "loss_ce": 0.01059566717594862,
+      "loss_iou": 1.1102811098098755,
+      "loss_num": 0.345703125,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 45781272,
+      "step": 265
+    },
+    {
+      "epoch": 0.10187667560321716,
+      "grad_norm": 32.735593085781545,
+      "learning_rate": 5e-06,
+      "loss": 0.5042,
+      "num_input_tokens_seen": 45954352,
+      "step": 266
+    },
+    {
+      "epoch": 0.10187667560321716,
+      "loss": 0.4968581795692444,
+      "loss_ce": 0.010774178430438042,
+      "loss_iou": 1.0254132747650146,
+      "loss_num": 0.486328125,
+      "loss_xval": 0.486328125,
+      "num_input_tokens_seen": 45954352,
+      "step": 266
+    },
+    {
+      "epoch": 0.10225967062428189,
+      "grad_norm": 20.80385360127902,
+      "learning_rate": 5e-06,
+      "loss": 0.362,
+      "num_input_tokens_seen": 46127568,
+      "step": 267
+    },
+    {
+      "epoch": 0.10225967062428189,
+      "loss": 0.3527145981788635,
+      "loss_ce": 0.01116187870502472,
+      "loss_iou": 1.0225780010223389,
+      "loss_num": 0.341796875,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 46127568,
+      "step": 267
+    },
+    {
+      "epoch": 0.10264266564534662,
+      "grad_norm": 17.243438931835026,
+      "learning_rate": 5e-06,
+      "loss": 0.457,
+      "num_input_tokens_seen": 46300576,
+      "step": 268
+    },
+    {
+      "epoch": 0.10264266564534662,
+      "loss": 0.4650697708129883,
+      "loss_ce": 0.010724080726504326,
+      "loss_iou": 1.0136557817459106,
+      "loss_num": 0.455078125,
+      "loss_xval": 0.455078125,
+      "num_input_tokens_seen": 46300576,
+      "step": 268
+    },
+    {
+      "epoch": 0.10302566066641133,
+      "grad_norm": 46.74543932677949,
+      "learning_rate": 5e-06,
+      "loss": 0.3367,
+      "num_input_tokens_seen": 46474072,
+      "step": 269
+    },
+    {
+      "epoch": 0.10302566066641133,
+      "loss": 0.33559098839759827,
+      "loss_ce": 0.01149429939687252,
+      "loss_iou": 1.0826479196548462,
+      "loss_num": 0.32421875,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 46474072,
+      "step": 269
+    },
+    {
+      "epoch": 0.10340865568747606,
+      "grad_norm": 38.67504807042694,
+      "learning_rate": 5e-06,
+      "loss": 0.4949,
+      "num_input_tokens_seen": 46647136,
+      "step": 270
+    },
+    {
+      "epoch": 0.10340865568747606,
+      "loss": 0.5154224634170532,
+      "loss_ce": 0.011027935892343521,
+      "loss_iou": 1.0698435306549072,
+      "loss_num": 0.50390625,
+      "loss_xval": 0.50390625,
+      "num_input_tokens_seen": 46647136,
+      "step": 270
+    },
+    {
+      "epoch": 0.1037916507085408,
+      "grad_norm": 27.550225883706,
+      "learning_rate": 5e-06,
+      "loss": 0.3874,
+      "num_input_tokens_seen": 46819680,
+      "step": 271
+    },
+    {
+      "epoch": 0.1037916507085408,
+      "loss": 0.37122344970703125,
+      "loss_ce": 0.010627737268805504,
+      "loss_iou": 1.0211730003356934,
+      "loss_num": 0.361328125,
+      "loss_xval": 0.361328125,
+      "num_input_tokens_seen": 46819680,
+      "step": 271
+    },
+    {
+      "epoch": 0.10417464572960551,
+      "grad_norm": 22.806761619302367,
+      "learning_rate": 5e-06,
+      "loss": 0.5148,
+      "num_input_tokens_seen": 46992712,
+      "step": 272
+    },
+    {
+      "epoch": 0.10417464572960551,
+      "loss": 0.4890601634979248,
+      "loss_ce": 0.010056260973215103,
+      "loss_iou": 1.1803604364395142,
+      "loss_num": 0.478515625,
+      "loss_xval": 0.478515625,
+      "num_input_tokens_seen": 46992712,
+      "step": 272
+    },
+    {
+      "epoch": 0.10455764075067024,
+      "grad_norm": 15.847794001537517,
+      "learning_rate": 5e-06,
+      "loss": 0.254,
+      "num_input_tokens_seen": 47165808,
+      "step": 273
+    },
+    {
+      "epoch": 0.10455764075067024,
+      "loss": 0.2679173946380615,
+      "loss_ce": 0.010898372158408165,
+      "loss_iou": 1.0206106901168823,
+      "loss_num": 0.2578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 47165808,
+      "step": 273
+    },
+    {
+      "epoch": 0.10494063577173497,
+      "grad_norm": 34.31450912159891,
+      "learning_rate": 5e-06,
+      "loss": 0.3414,
+      "num_input_tokens_seen": 47339232,
+      "step": 274
+    },
+    {
+      "epoch": 0.10494063577173497,
+      "loss": 0.32526785135269165,
+      "loss_ce": 0.010570560581982136,
+      "loss_iou": 1.0116055011749268,
+      "loss_num": 0.314453125,
+      "loss_xval": 0.314453125,
+      "num_input_tokens_seen": 47339232,
+      "step": 274
+    },
+    {
+      "epoch": 0.10532363079279969,
+      "grad_norm": 27.716014130758875,
+      "learning_rate": 5e-06,
+      "loss": 0.4536,
+      "num_input_tokens_seen": 47511976,
+      "step": 275
+    },
+    {
+      "epoch": 0.10532363079279969,
+      "loss": 0.4193720519542694,
+      "loss_ce": 0.009704088792204857,
+      "loss_iou": 0.9634692668914795,
+      "loss_num": 0.41015625,
+      "loss_xval": 0.41015625,
+      "num_input_tokens_seen": 47511976,
+      "step": 275
+    },
+    {
+      "epoch": 0.10570662581386442,
+      "grad_norm": 17.7829828252544,
+      "learning_rate": 5e-06,
+      "loss": 0.3866,
+      "num_input_tokens_seen": 47685064,
+      "step": 276
+    },
+    {
+      "epoch": 0.10570662581386442,
+      "loss": 0.41627562046051025,
+      "loss_ce": 0.011002162471413612,
+      "loss_iou": 1.0582951307296753,
+      "loss_num": 0.40625,
+      "loss_xval": 0.40625,
+      "num_input_tokens_seen": 47685064,
+      "step": 276
+    },
+    {
+      "epoch": 0.10608962083492915,
+      "grad_norm": 31.24090989532103,
+      "learning_rate": 5e-06,
+      "loss": 0.3073,
+      "num_input_tokens_seen": 47858312,
+      "step": 277
+    },
+    {
+      "epoch": 0.10608962083492915,
+      "loss": 0.34194135665893555,
+      "loss_ce": 0.010398398153483868,
+      "loss_iou": 1.0338774919509888,
+      "loss_num": 0.33203125,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 47858312,
+      "step": 277
+    },
+    {
+      "epoch": 0.10647261585599387,
+      "grad_norm": 32.0296279715356,
+      "learning_rate": 5e-06,
+      "loss": 0.5668,
+      "num_input_tokens_seen": 48031616,
+      "step": 278
+    },
+    {
+      "epoch": 0.10647261585599387,
+      "loss": 0.5294106006622314,
+      "loss_ce": 0.010367615148425102,
+      "loss_iou": 1.145504355430603,
+      "loss_num": 0.51953125,
+      "loss_xval": 0.51953125,
+      "num_input_tokens_seen": 48031616,
+      "step": 278
+    },
+    {
+      "epoch": 0.1068556108770586,
+      "grad_norm": 21.181429801625686,
+      "learning_rate": 5e-06,
+      "loss": 0.3589,
+      "num_input_tokens_seen": 48204776,
+      "step": 279
+    },
+    {
+      "epoch": 0.1068556108770586,
+      "loss": 0.2801125645637512,
+      "loss_ce": 0.010459225624799728,
+      "loss_iou": 1.0051308870315552,
+      "loss_num": 0.26953125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 48204776,
+      "step": 279
+    },
+    {
+      "epoch": 0.10723860589812333,
+      "grad_norm": 19.094716754371948,
+      "learning_rate": 5e-06,
+      "loss": 0.4459,
+      "num_input_tokens_seen": 48377704,
+      "step": 280
+    },
+    {
+      "epoch": 0.10723860589812333,
+      "loss": 0.43079090118408203,
+      "loss_ce": 0.01038073468953371,
+      "loss_iou": 1.0597786903381348,
+      "loss_num": 0.419921875,
+      "loss_xval": 0.419921875,
+      "num_input_tokens_seen": 48377704,
+      "step": 280
+    },
+    {
+      "epoch": 0.10762160091918806,
+      "grad_norm": 23.839418127132326,
+      "learning_rate": 5e-06,
+      "loss": 0.3668,
+      "num_input_tokens_seen": 48550864,
+      "step": 281
+    },
+    {
+      "epoch": 0.10762160091918806,
+      "loss": 0.45468515157699585,
+      "loss_ce": 0.009494691155850887,
+      "loss_iou": 1.0974698066711426,
+      "loss_num": 0.4453125,
+      "loss_xval": 0.4453125,
+      "num_input_tokens_seen": 48550864,
+      "step": 281
+    },
+    {
+      "epoch": 0.10800459594025277,
+      "grad_norm": 29.281611849796462,
+      "learning_rate": 5e-06,
+      "loss": 0.4213,
+      "num_input_tokens_seen": 48724008,
+      "step": 282
+    },
+    {
+      "epoch": 0.10800459594025277,
+      "loss": 0.38292914628982544,
+      "loss_ce": 0.010126392357051373,
+      "loss_iou": 1.0047979354858398,
+      "loss_num": 0.373046875,
+      "loss_xval": 0.373046875,
+      "num_input_tokens_seen": 48724008,
+      "step": 282
+    },
+    {
+      "epoch": 0.1083875909613175,
+      "grad_norm": 22.852496320719037,
+      "learning_rate": 5e-06,
+      "loss": 0.3968,
+      "num_input_tokens_seen": 48897088,
+      "step": 283
+    },
+    {
+      "epoch": 0.1083875909613175,
+      "loss": 0.36059004068374634,
+      "loss_ce": 0.01049237884581089,
+      "loss_iou": 1.095686912536621,
+      "loss_num": 0.349609375,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 48897088,
+      "step": 283
+    },
+    {
+      "epoch": 0.10877058598238223,
+      "grad_norm": 16.194662237848114,
+      "learning_rate": 5e-06,
+      "loss": 0.4194,
+      "num_input_tokens_seen": 49070160,
+      "step": 284
+    },
+    {
+      "epoch": 0.10877058598238223,
+      "loss": 0.4404478669166565,
+      "loss_ce": 0.010516216047108173,
+      "loss_iou": 1.0644468069076538,
+      "loss_num": 0.4296875,
+      "loss_xval": 0.4296875,
+      "num_input_tokens_seen": 49070160,
+      "step": 284
+    },
+    {
+      "epoch": 0.10915358100344695,
+      "grad_norm": 27.283997636832314,
+      "learning_rate": 5e-06,
+      "loss": 0.2844,
+      "num_input_tokens_seen": 49243192,
+      "step": 285
+    },
+    {
+      "epoch": 0.10915358100344695,
+      "loss": 0.23761063814163208,
+      "loss_ce": 0.010437790304422379,
+      "loss_iou": 1.0049580335617065,
+      "loss_num": 0.2275390625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 49243192,
+      "step": 285
+    },
+    {
+      "epoch": 0.10953657602451168,
+      "grad_norm": 38.47563103402911,
+      "learning_rate": 5e-06,
+      "loss": 0.5385,
+      "num_input_tokens_seen": 49416184,
+      "step": 286
+    },
+    {
+      "epoch": 0.10953657602451168,
+      "loss": 0.5168796181678772,
+      "loss_ce": 0.011264389380812645,
+      "loss_iou": 1.1069316864013672,
+      "loss_num": 0.50390625,
+      "loss_xval": 0.50390625,
+      "num_input_tokens_seen": 49416184,
+      "step": 286
+    },
+    {
+      "epoch": 0.10991957104557641,
+      "grad_norm": 28.808207745099345,
+      "learning_rate": 5e-06,
+      "loss": 0.2741,
+      "num_input_tokens_seen": 49589080,
+      "step": 287
+    },
+    {
+      "epoch": 0.10991957104557641,
+      "loss": 0.2880805730819702,
+      "loss_ce": 0.01098096463829279,
+      "loss_iou": 1.044050693511963,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 49589080,
+      "step": 287
+    },
+    {
+      "epoch": 0.11030256606664113,
+      "grad_norm": 32.05739369285175,
+      "learning_rate": 5e-06,
+      "loss": 0.3938,
+      "num_input_tokens_seen": 49761760,
+      "step": 288
+    },
+    {
+      "epoch": 0.11030256606664113,
+      "loss": 0.41695815324783325,
+      "loss_ce": 0.010586072690784931,
+      "loss_iou": 1.3855371475219727,
+      "loss_num": 0.40625,
+      "loss_xval": 0.40625,
+      "num_input_tokens_seen": 49761760,
+      "step": 288
+    },
+    {
+      "epoch": 0.11068556108770586,
+      "grad_norm": 50.29793791048213,
+      "learning_rate": 5e-06,
+      "loss": 0.583,
+      "num_input_tokens_seen": 49934704,
+      "step": 289
+    },
+    {
+      "epoch": 0.11068556108770586,
+      "loss": 0.561424970626831,
+      "loss_ce": 0.01015543844550848,
+      "loss_iou": 1.0180261135101318,
+      "loss_num": 0.55078125,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 49934704,
+      "step": 289
+    },
+    {
+      "epoch": 0.11106855610877059,
+      "grad_norm": 35.66832463769736,
+      "learning_rate": 5e-06,
+      "loss": 0.5456,
+      "num_input_tokens_seen": 50107712,
+      "step": 290
+    },
+    {
+      "epoch": 0.11106855610877059,
+      "loss": 0.5598129630088806,
+      "loss_ce": 0.01049656979739666,
+      "loss_iou": 1.013216257095337,
+      "loss_num": 0.55078125,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 50107712,
+      "step": 290
+    },
+    {
+      "epoch": 0.11145155112983532,
+      "grad_norm": 31.568232222493787,
+      "learning_rate": 5e-06,
+      "loss": 0.4637,
+      "num_input_tokens_seen": 50280752,
+      "step": 291
+    },
+    {
+      "epoch": 0.11145155112983532,
+      "loss": 0.4419333338737488,
+      "loss_ce": 0.01102512702345848,
+      "loss_iou": 1.2086838483810425,
+      "loss_num": 0.431640625,
+      "loss_xval": 0.431640625,
+      "num_input_tokens_seen": 50280752,
+      "step": 291
+    },
+    {
+      "epoch": 0.11183454615090004,
+      "grad_norm": 32.23048022560672,
+      "learning_rate": 5e-06,
+      "loss": 0.5149,
+      "num_input_tokens_seen": 50453728,
+      "step": 292
+    },
+    {
+      "epoch": 0.11183454615090004,
+      "loss": 0.4906352162361145,
+      "loss_ce": 0.010654768906533718,
+      "loss_iou": 1.09377121925354,
+      "loss_num": 0.48046875,
+      "loss_xval": 0.48046875,
+      "num_input_tokens_seen": 50453728,
+      "step": 292
+    },
+    {
+      "epoch": 0.11221754117196477,
+      "grad_norm": 38.09973773117454,
+      "learning_rate": 5e-06,
+      "loss": 0.492,
+      "num_input_tokens_seen": 50626616,
+      "step": 293
+    },
+    {
+      "epoch": 0.11221754117196477,
+      "loss": 0.43555042147636414,
+      "loss_ce": 0.010257437825202942,
+      "loss_iou": 1.026540756225586,
+      "loss_num": 0.42578125,
+      "loss_xval": 0.42578125,
+      "num_input_tokens_seen": 50626616,
+      "step": 293
+    },
+    {
+      "epoch": 0.1126005361930295,
+      "grad_norm": 39.24787901065983,
+      "learning_rate": 5e-06,
+      "loss": 0.6363,
+      "num_input_tokens_seen": 50799848,
+      "step": 294
+    },
+    {
+      "epoch": 0.1126005361930295,
+      "loss": 0.6617828607559204,
+      "loss_ce": 0.011392205953598022,
+      "loss_iou": 1.0512713193893433,
+      "loss_num": 0.6484375,
+      "loss_xval": 0.6484375,
+      "num_input_tokens_seen": 50799848,
+      "step": 294
+    },
+    {
+      "epoch": 0.11298353121409421,
+      "grad_norm": 32.90943652131842,
+      "learning_rate": 5e-06,
+      "loss": 0.6314,
+      "num_input_tokens_seen": 50973248,
+      "step": 295
+    },
+    {
+      "epoch": 0.11298353121409421,
+      "loss": 0.6341913938522339,
+      "loss_ce": 0.011144520714879036,
+      "loss_iou": 1.3047211170196533,
+      "loss_num": 0.625,
+      "loss_xval": 0.625,
+      "num_input_tokens_seen": 50973248,
+      "step": 295
+    },
+    {
+      "epoch": 0.11336652623515894,
+      "grad_norm": 38.52268484842589,
+      "learning_rate": 5e-06,
+      "loss": 0.9833,
+      "num_input_tokens_seen": 51146048,
+      "step": 296
+    },
+    {
+      "epoch": 0.11336652623515894,
+      "loss": 0.9849939346313477,
+      "loss_ce": 0.010384575463831425,
+      "loss_iou": 1.2159556150436401,
+      "loss_num": 0.9765625,
+      "loss_xval": 0.9765625,
+      "num_input_tokens_seen": 51146048,
+      "step": 296
+    },
+    {
+      "epoch": 0.11374952125622367,
+      "grad_norm": 57.08290135274048,
+      "learning_rate": 5e-06,
+      "loss": 0.541,
+      "num_input_tokens_seen": 51318912,
+      "step": 297
+    },
+    {
+      "epoch": 0.11374952125622367,
+      "loss": 0.5805840492248535,
+      "loss_ce": 0.010027414187788963,
+      "loss_iou": 1.2200486660003662,
+      "loss_num": 0.5703125,
+      "loss_xval": 0.5703125,
+      "num_input_tokens_seen": 51318912,
+      "step": 297
+    },
+    {
+      "epoch": 0.11413251627728839,
+      "grad_norm": 72.8746542496279,
+      "learning_rate": 5e-06,
+      "loss": 1.3715,
+      "num_input_tokens_seen": 51491568,
+      "step": 298
+    },
+    {
+      "epoch": 0.11413251627728839,
+      "loss": 1.379122018814087,
+      "loss_ce": 0.010957980528473854,
+      "loss_iou": 1.378398060798645,
+      "loss_num": 1.3671875,
+      "loss_xval": 1.3671875,
+      "num_input_tokens_seen": 51491568,
+      "step": 298
+    },
+    {
+      "epoch": 0.11451551129835312,
+      "grad_norm": 25.52973147801263,
+      "learning_rate": 5e-06,
+      "loss": 0.5142,
+      "num_input_tokens_seen": 51664720,
+      "step": 299
+    },
+    {
+      "epoch": 0.11451551129835312,
+      "loss": 0.5176573395729065,
+      "loss_ce": 0.01228625699877739,
+      "loss_iou": 2.6109676361083984,
+      "loss_num": 0.50390625,
+      "loss_xval": 0.50390625,
+      "num_input_tokens_seen": 51664720,
+      "step": 299
+    },
+    {
+      "epoch": 0.11489850631941785,
+      "grad_norm": 25.631635843937243,
+      "learning_rate": 5e-06,
+      "loss": 0.3688,
+      "num_input_tokens_seen": 51837584,
+      "step": 300
+    },
+    {
+      "epoch": 0.11489850631941785,
+      "loss": 0.3400018811225891,
+      "loss_ce": 0.012365150265395641,
+      "loss_iou": 1.0193140506744385,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 51837584,
+      "step": 300
+    },
+    {
+      "epoch": 0.11528150134048257,
+      "grad_norm": 36.27654091995768,
+      "learning_rate": 5e-06,
+      "loss": 0.7293,
+      "num_input_tokens_seen": 52010008,
+      "step": 301
+    },
+    {
+      "epoch": 0.11528150134048257,
+      "loss": 0.7323391437530518,
+      "loss_ce": 0.009682867676019669,
+      "loss_iou": 0.9314297437667847,
+      "loss_num": 0.72265625,
+      "loss_xval": 0.72265625,
+      "num_input_tokens_seen": 52010008,
+      "step": 301
+    },
+    {
+      "epoch": 0.1156644963615473,
+      "grad_norm": 35.419766480705015,
+      "learning_rate": 5e-06,
+      "loss": 0.4071,
+      "num_input_tokens_seen": 52182672,
+      "step": 302
+    },
+    {
+      "epoch": 0.1156644963615473,
+      "loss": 0.42031729221343994,
+      "loss_ce": 0.012114165350794792,
+      "loss_iou": 1.031744360923767,
+      "loss_num": 0.408203125,
+      "loss_xval": 0.408203125,
+      "num_input_tokens_seen": 52182672,
+      "step": 302
+    },
+    {
+      "epoch": 0.11604749138261203,
+      "grad_norm": 42.847599300870044,
+      "learning_rate": 5e-06,
+      "loss": 0.6895,
+      "num_input_tokens_seen": 52355624,
+      "step": 303
+    },
+    {
+      "epoch": 0.11604749138261203,
+      "loss": 0.6960222721099854,
+      "loss_ce": 0.011696080677211285,
+      "loss_iou": 1.0434455871582031,
+      "loss_num": 0.68359375,
+      "loss_xval": 0.68359375,
+      "num_input_tokens_seen": 52355624,
+      "step": 303
+    },
+    {
+      "epoch": 0.11643048640367676,
+      "grad_norm": 32.86704853238209,
+      "learning_rate": 5e-06,
+      "loss": 0.4558,
+      "num_input_tokens_seen": 52528528,
+      "step": 304
+    },
+    {
+      "epoch": 0.11643048640367676,
+      "loss": 0.45993906259536743,
+      "loss_ce": 0.010964442044496536,
+      "loss_iou": 1.0512936115264893,
+      "loss_num": 0.44921875,
+      "loss_xval": 0.44921875,
+      "num_input_tokens_seen": 52528528,
+      "step": 304
+    },
+    {
+      "epoch": 0.11681348142474147,
+      "grad_norm": 40.050753637117815,
+      "learning_rate": 5e-06,
+      "loss": 0.4655,
+      "num_input_tokens_seen": 52701512,
+      "step": 305
+    },
+    {
+      "epoch": 0.11681348142474147,
+      "loss": 0.4385191798210144,
+      "loss_ce": 0.01078478991985321,
+      "loss_iou": 1.0094962120056152,
+      "loss_num": 0.427734375,
+      "loss_xval": 0.427734375,
+      "num_input_tokens_seen": 52701512,
+      "step": 305
+    },
+    {
+      "epoch": 0.1171964764458062,
+      "grad_norm": 54.80718049230888,
+      "learning_rate": 5e-06,
+      "loss": 0.9926,
+      "num_input_tokens_seen": 52874496,
+      "step": 306
+    },
+    {
+      "epoch": 0.1171964764458062,
+      "loss": 0.9117997884750366,
+      "loss_ce": 0.011897400952875614,
+      "loss_iou": 1.0914658308029175,
+      "loss_num": 0.8984375,
+      "loss_xval": 0.8984375,
+      "num_input_tokens_seen": 52874496,
+      "step": 306
+    },
+    {
+      "epoch": 0.11757947146687094,
+      "grad_norm": 29.68185097214178,
+      "learning_rate": 5e-06,
+      "loss": 0.6096,
+      "num_input_tokens_seen": 53047616,
+      "step": 307
+    },
+    {
+      "epoch": 0.11757947146687094,
+      "loss": 0.6523078680038452,
+      "loss_ce": 0.01217116229236126,
+      "loss_iou": 1.153963327407837,
+      "loss_num": 0.640625,
+      "loss_xval": 0.640625,
+      "num_input_tokens_seen": 53047616,
+      "step": 307
+    },
+    {
+      "epoch": 0.11796246648793565,
+      "grad_norm": 38.749918455185316,
+      "learning_rate": 5e-06,
+      "loss": 0.8524,
+      "num_input_tokens_seen": 53220816,
+      "step": 308
+    },
+    {
+      "epoch": 0.11796246648793565,
+      "loss": 0.8711147308349609,
+      "loss_ce": 0.01222802884876728,
+      "loss_iou": 1.238293170928955,
+      "loss_num": 0.859375,
+      "loss_xval": 0.859375,
+      "num_input_tokens_seen": 53220816,
+      "step": 308
+    },
+    {
+      "epoch": 0.11834546150900038,
+      "grad_norm": 56.841035253055814,
+      "learning_rate": 5e-06,
+      "loss": 0.6204,
+      "num_input_tokens_seen": 53393744,
+      "step": 309
+    },
+    {
+      "epoch": 0.11834546150900038,
+      "loss": 0.6435734629631042,
+      "loss_ce": 0.01100508775562048,
+      "loss_iou": 1.0825989246368408,
+      "loss_num": 0.6328125,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 53393744,
+      "step": 309
+    },
+    {
+      "epoch": 0.11872845653006511,
+      "grad_norm": 65.23315581624193,
+      "learning_rate": 5e-06,
+      "loss": 1.0426,
+      "num_input_tokens_seen": 53566608,
+      "step": 310
+    },
+    {
+      "epoch": 0.11872845653006511,
+      "loss": 1.0427701473236084,
+      "loss_ce": 0.011520168744027615,
+      "loss_iou": 1.2222198247909546,
+      "loss_num": 1.03125,
+      "loss_xval": 1.03125,
+      "num_input_tokens_seen": 53566608,
+      "step": 310
+    },
+    {
+      "epoch": 0.11911145155112983,
+      "grad_norm": 25.316963856011057,
+      "learning_rate": 5e-06,
+      "loss": 0.5238,
+      "num_input_tokens_seen": 53739936,
+      "step": 311
+    },
+    {
+      "epoch": 0.11911145155112983,
+      "loss": 0.5922715067863464,
+      "loss_ce": 0.013414058834314346,
+      "loss_iou": 1.1291487216949463,
+      "loss_num": 0.578125,
+      "loss_xval": 0.578125,
+      "num_input_tokens_seen": 53739936,
+      "step": 311
+    },
+    {
+      "epoch": 0.11949444657219456,
+      "grad_norm": 33.11489631674243,
+      "learning_rate": 5e-06,
+      "loss": 0.7113,
+      "num_input_tokens_seen": 53912920,
+      "step": 312
+    },
+    {
+      "epoch": 0.11949444657219456,
+      "loss": 0.6654304265975952,
+      "loss_ce": 0.013086654245853424,
+      "loss_iou": 1.065284252166748,
+      "loss_num": 0.65234375,
+      "loss_xval": 0.65234375,
+      "num_input_tokens_seen": 53912920,
+      "step": 312
+    },
+    {
+      "epoch": 0.11987744159325929,
+      "grad_norm": 34.16303791635373,
+      "learning_rate": 5e-06,
+      "loss": 0.5034,
+      "num_input_tokens_seen": 54085432,
+      "step": 313
+    },
+    {
+      "epoch": 0.11987744159325929,
+      "loss": 0.36735761165618896,
+      "loss_ce": 0.011644711717963219,
+      "loss_iou": 1.026916265487671,
+      "loss_num": 0.35546875,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 54085432,
+      "step": 313
+    },
+    {
+      "epoch": 0.12026043661432402,
+      "grad_norm": 38.22994920245021,
+      "learning_rate": 5e-06,
+      "loss": 0.7024,
+      "num_input_tokens_seen": 54258280,
+      "step": 314
+    },
+    {
+      "epoch": 0.12026043661432402,
+      "loss": 0.7669047117233276,
+      "loss_ce": 0.01251023355871439,
+      "loss_iou": 1.106337547302246,
+      "loss_num": 0.75390625,
+      "loss_xval": 0.75390625,
+      "num_input_tokens_seen": 54258280,
+      "step": 314
+    },
+    {
+      "epoch": 0.12064343163538874,
+      "grad_norm": 42.09697413056026,
+      "learning_rate": 5e-06,
+      "loss": 0.3733,
+      "num_input_tokens_seen": 54431280,
+      "step": 315
+    },
+    {
+      "epoch": 0.12064343163538874,
+      "loss": 0.35330480337142944,
+      "loss_ce": 0.01333899050951004,
+      "loss_iou": 1.0186564922332764,
+      "loss_num": 0.33984375,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 54431280,
+      "step": 315
+    },
+    {
+      "epoch": 0.12102642665645347,
+      "grad_norm": 35.74366625552186,
+      "learning_rate": 5e-06,
+      "loss": 0.6181,
+      "num_input_tokens_seen": 54604368,
+      "step": 316
+    },
+    {
+      "epoch": 0.12102642665645347,
+      "loss": 0.6472422480583191,
+      "loss_ce": 0.013453170657157898,
+      "loss_iou": 1.0799098014831543,
+      "loss_num": 0.6328125,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 54604368,
+      "step": 316
+    },
+    {
+      "epoch": 0.1214094216775182,
+      "grad_norm": 11.527198722935454,
+      "learning_rate": 5e-06,
+      "loss": 0.3938,
+      "num_input_tokens_seen": 54777456,
+      "step": 317
+    },
+    {
+      "epoch": 0.1214094216775182,
+      "loss": 0.36960890889167786,
+      "loss_ce": 0.012919455766677856,
+      "loss_iou": 1.0854665040969849,
+      "loss_num": 0.357421875,
+      "loss_xval": 0.357421875,
+      "num_input_tokens_seen": 54777456,
+      "step": 317
+    },
+    {
+      "epoch": 0.12179241669858291,
+      "grad_norm": 24.293199292377544,
+      "learning_rate": 5e-06,
+      "loss": 0.3295,
+      "num_input_tokens_seen": 54950048,
+      "step": 318
+    },
+    {
+      "epoch": 0.12179241669858291,
+      "loss": 0.35775184631347656,
+      "loss_ce": 0.011560434475541115,
+      "loss_iou": 1.0050764083862305,
+      "loss_num": 0.345703125,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 54950048,
+      "step": 318
+    },
+    {
+      "epoch": 0.12217541171964764,
+      "grad_norm": 33.30097884131126,
+      "learning_rate": 5e-06,
+      "loss": 0.4818,
+      "num_input_tokens_seen": 55123176,
+      "step": 319
+    },
+    {
+      "epoch": 0.12217541171964764,
+      "loss": 0.4854978322982788,
+      "loss_ce": 0.012841572985053062,
+      "loss_iou": 1.0677175521850586,
+      "loss_num": 0.47265625,
+      "loss_xval": 0.47265625,
+      "num_input_tokens_seen": 55123176,
+      "step": 319
+    },
+    {
+      "epoch": 0.12255840674071238,
+      "grad_norm": 30.662633429604817,
+      "learning_rate": 5e-06,
+      "loss": 0.4641,
+      "num_input_tokens_seen": 55295904,
+      "step": 320
+    },
+    {
+      "epoch": 0.12255840674071238,
+      "loss": 0.46251463890075684,
+      "loss_ce": 0.012563494965434074,
+      "loss_iou": 1.0785880088806152,
+      "loss_num": 0.44921875,
+      "loss_xval": 0.44921875,
+      "num_input_tokens_seen": 55295904,
+      "step": 320
+    },
+    {
+      "epoch": 0.12294140176177709,
+      "grad_norm": 36.52319132677881,
+      "learning_rate": 5e-06,
+      "loss": 0.4746,
+      "num_input_tokens_seen": 55468688,
+      "step": 321
+    },
+    {
+      "epoch": 0.12294140176177709,
+      "loss": 0.38861754536628723,
+      "loss_ce": 0.012885136529803276,
+      "loss_iou": 1.047745943069458,
+      "loss_num": 0.375,
+      "loss_xval": 0.375,
+      "num_input_tokens_seen": 55468688,
+      "step": 321
+    },
+    {
+      "epoch": 0.12332439678284182,
+      "grad_norm": 30.7933132232677,
+      "learning_rate": 5e-06,
+      "loss": 0.4801,
+      "num_input_tokens_seen": 55641392,
+      "step": 322
+    },
+    {
+      "epoch": 0.12332439678284182,
+      "loss": 0.42514312267303467,
+      "loss_ce": 0.014010298997163773,
+      "loss_iou": 1.0376410484313965,
+      "loss_num": 0.41015625,
+      "loss_xval": 0.41015625,
+      "num_input_tokens_seen": 55641392,
+      "step": 322
+    },
+    {
+      "epoch": 0.12370739180390655,
+      "grad_norm": 29.101293969380325,
+      "learning_rate": 5e-06,
+      "loss": 0.3619,
+      "num_input_tokens_seen": 55814200,
+      "step": 323
+    },
+    {
+      "epoch": 0.12370739180390655,
+      "loss": 0.40042075514793396,
+      "loss_ce": 0.012481309473514557,
+      "loss_iou": 1.045759677886963,
+      "loss_num": 0.388671875,
+      "loss_xval": 0.388671875,
+      "num_input_tokens_seen": 55814200,
+      "step": 323
+    },
+    {
+      "epoch": 0.12409038682497127,
+      "grad_norm": 32.08435616179997,
+      "learning_rate": 5e-06,
+      "loss": 0.4483,
+      "num_input_tokens_seen": 55987296,
+      "step": 324
+    },
+    {
+      "epoch": 0.12409038682497127,
+      "loss": 0.4775657653808594,
+      "loss_ce": 0.013454440981149673,
+      "loss_iou": 1.0528295040130615,
+      "loss_num": 0.46484375,
+      "loss_xval": 0.46484375,
+      "num_input_tokens_seen": 55987296,
+      "step": 324
+    },
+    {
+      "epoch": 0.124473381846036,
+      "grad_norm": 28.387154844066295,
+      "learning_rate": 5e-06,
+      "loss": 0.409,
+      "num_input_tokens_seen": 56159808,
+      "step": 325
+    },
+    {
+      "epoch": 0.124473381846036,
+      "loss": 0.43876785039901733,
+      "loss_ce": 0.012864558026194572,
+      "loss_iou": 1.0930651426315308,
+      "loss_num": 0.42578125,
+      "loss_xval": 0.42578125,
+      "num_input_tokens_seen": 56159808,
+      "step": 325
+    },
+    {
+      "epoch": 0.12485637686710073,
+      "grad_norm": 33.37057560894794,
+      "learning_rate": 5e-06,
+      "loss": 0.4507,
+      "num_input_tokens_seen": 56333032,
+      "step": 326
+    },
+    {
+      "epoch": 0.12485637686710073,
+      "loss": 0.4985562860965729,
+      "loss_ce": 0.012716436758637428,
+      "loss_iou": 1.147887110710144,
+      "loss_num": 0.486328125,
+      "loss_xval": 0.486328125,
+      "num_input_tokens_seen": 56333032,
+      "step": 326
+    },
+    {
+      "epoch": 0.12523937188816545,
+      "grad_norm": 37.32139130798763,
+      "learning_rate": 5e-06,
+      "loss": 0.4723,
+      "num_input_tokens_seen": 56502904,
+      "step": 327
+    },
+    {
+      "epoch": 0.12523937188816545,
+      "loss": 0.44586414098739624,
+      "loss_ce": 0.012514510191977024,
+      "loss_iou": 1.1612491607666016,
+      "loss_num": 0.43359375,
+      "loss_xval": 0.43359375,
+      "num_input_tokens_seen": 56502904,
+      "step": 327
+    },
+    {
+      "epoch": 0.12562236690923018,
+      "grad_norm": 28.902907172422736,
+      "learning_rate": 5e-06,
+      "loss": 0.4802,
+      "num_input_tokens_seen": 56675760,
+      "step": 328
+    },
+    {
+      "epoch": 0.12562236690923018,
+      "loss": 0.4507526159286499,
+      "loss_ce": 0.0130084790289402,
+      "loss_iou": 1.0351426601409912,
+      "loss_num": 0.4375,
+      "loss_xval": 0.4375,
+      "num_input_tokens_seen": 56675760,
+      "step": 328
+    },
+    {
+      "epoch": 0.1260053619302949,
+      "grad_norm": 38.32465303461408,
+      "learning_rate": 5e-06,
+      "loss": 0.4264,
+      "num_input_tokens_seen": 56849296,
+      "step": 329
+    },
+    {
+      "epoch": 0.1260053619302949,
+      "loss": 0.4553164541721344,
+      "loss_ce": 0.01342192105948925,
+      "loss_iou": 1.3379547595977783,
+      "loss_num": 0.44140625,
+      "loss_xval": 0.44140625,
+      "num_input_tokens_seen": 56849296,
+      "step": 329
+    },
+    {
+      "epoch": 0.12638835695135964,
+      "grad_norm": 31.373504744313774,
+      "learning_rate": 5e-06,
+      "loss": 0.5351,
+      "num_input_tokens_seen": 57022456,
+      "step": 330
+    },
+    {
+      "epoch": 0.12638835695135964,
+      "loss": 0.5396110415458679,
+      "loss_ce": 0.013732131570577621,
+      "loss_iou": 1.042524814605713,
+      "loss_num": 0.52734375,
+      "loss_xval": 0.52734375,
+      "num_input_tokens_seen": 57022456,
+      "step": 330
+    },
+    {
+      "epoch": 0.12677135197242437,
+      "grad_norm": 25.24014458006925,
+      "learning_rate": 5e-06,
+      "loss": 0.3363,
+      "num_input_tokens_seen": 57195656,
+      "step": 331
+    },
+    {
+      "epoch": 0.12677135197242437,
+      "loss": 0.30840808153152466,
+      "loss_ce": 0.013608288019895554,
+      "loss_iou": 1.026963710784912,
+      "loss_num": 0.294921875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 57195656,
+      "step": 331
+    },
+    {
+      "epoch": 0.1271543469934891,
+      "grad_norm": 20.587109923649756,
+      "learning_rate": 5e-06,
+      "loss": 0.5719,
+      "num_input_tokens_seen": 57368784,
+      "step": 332
+    },
+    {
+      "epoch": 0.1271543469934891,
+      "loss": 0.5962228775024414,
+      "loss_ce": 0.014435779303312302,
+      "loss_iou": 1.1006231307983398,
+      "loss_num": 0.58203125,
+      "loss_xval": 0.58203125,
+      "num_input_tokens_seen": 57368784,
+      "step": 332
+    },
+    {
+      "epoch": 0.1275373420145538,
+      "grad_norm": 19.42338701583887,
+      "learning_rate": 5e-06,
+      "loss": 0.3134,
+      "num_input_tokens_seen": 57541640,
+      "step": 333
+    },
+    {
+      "epoch": 0.1275373420145538,
+      "loss": 0.3170633018016815,
+      "loss_ce": 0.013718578033149242,
+      "loss_iou": 1.0257070064544678,
+      "loss_num": 0.302734375,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 57541640,
+      "step": 333
+    },
+    {
+      "epoch": 0.12792033703561853,
+      "grad_norm": 23.602551955156134,
+      "learning_rate": 5e-06,
+      "loss": 0.4829,
+      "num_input_tokens_seen": 57714872,
+      "step": 334
+    },
+    {
+      "epoch": 0.12792033703561853,
+      "loss": 0.5177361369132996,
+      "loss_ce": 0.01260918378829956,
+      "loss_iou": 1.1138770580291748,
+      "loss_num": 0.50390625,
+      "loss_xval": 0.50390625,
+      "num_input_tokens_seen": 57714872,
+      "step": 334
+    },
+    {
+      "epoch": 0.12830333205668326,
+      "grad_norm": 28.002566112166686,
+      "learning_rate": 5e-06,
+      "loss": 0.3301,
+      "num_input_tokens_seen": 57888000,
+      "step": 335
+    },
+    {
+      "epoch": 0.12830333205668326,
+      "loss": 0.3402954936027527,
+      "loss_ce": 0.012658771127462387,
+      "loss_iou": 1.0171738862991333,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 57888000,
+      "step": 335
+    },
+    {
+      "epoch": 0.128686327077748,
+      "grad_norm": 29.871569630960213,
+      "learning_rate": 5e-06,
+      "loss": 0.4392,
+      "num_input_tokens_seen": 58060760,
+      "step": 336
+    },
+    {
+      "epoch": 0.128686327077748,
+      "loss": 0.4566802382469177,
+      "loss_ce": 0.012344276532530785,
+      "loss_iou": 1.161993384361267,
+      "loss_num": 0.4453125,
+      "loss_xval": 0.4453125,
+      "num_input_tokens_seen": 58060760,
+      "step": 336
+    },
+    {
+      "epoch": 0.12906932209881272,
+      "grad_norm": 18.46477895053161,
+      "learning_rate": 5e-06,
+      "loss": 0.2781,
+      "num_input_tokens_seen": 58233368,
+      "step": 337
+    },
+    {
+      "epoch": 0.12906932209881272,
+      "loss": 0.2804095149040222,
+      "loss_ce": 0.012953441590070724,
+      "loss_iou": 1.0590258836746216,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 58233368,
+      "step": 337
+    },
+    {
+      "epoch": 0.12945231711987745,
+      "grad_norm": 19.423082442865574,
+      "learning_rate": 5e-06,
+      "loss": 0.3513,
+      "num_input_tokens_seen": 58405992,
+      "step": 338
+    },
+    {
+      "epoch": 0.12945231711987745,
+      "loss": 0.3404031991958618,
+      "loss_ce": 0.01227821595966816,
+      "loss_iou": 1.0197341442108154,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 58405992,
+      "step": 338
+    },
+    {
+      "epoch": 0.12983531214094216,
+      "grad_norm": 28.033606565177212,
+      "learning_rate": 5e-06,
+      "loss": 0.388,
+      "num_input_tokens_seen": 58578624,
+      "step": 339
+    },
+    {
+      "epoch": 0.12983531214094216,
+      "loss": 0.4279400408267975,
+      "loss_ce": 0.01265685074031353,
+      "loss_iou": 1.0694093704223633,
+      "loss_num": 0.416015625,
+      "loss_xval": 0.416015625,
+      "num_input_tokens_seen": 58578624,
+      "step": 339
+    },
+    {
+      "epoch": 0.13021830716200689,
+      "grad_norm": 36.54563621057312,
+      "learning_rate": 5e-06,
+      "loss": 0.5259,
+      "num_input_tokens_seen": 58751488,
+      "step": 340
+    },
+    {
+      "epoch": 0.13021830716200689,
+      "loss": 0.5527739524841309,
+      "loss_ce": 0.01371149905025959,
+      "loss_iou": 1.3441195487976074,
+      "loss_num": 0.5390625,
+      "loss_xval": 0.5390625,
+      "num_input_tokens_seen": 58751488,
+      "step": 340
+    },
+    {
+      "epoch": 0.13060130218307162,
+      "grad_norm": 25.999930582103996,
+      "learning_rate": 5e-06,
+      "loss": 0.3373,
+      "num_input_tokens_seen": 58924544,
+      "step": 341
+    },
+    {
+      "epoch": 0.13060130218307162,
+      "loss": 0.33388644456863403,
+      "loss_ce": 0.012719469144940376,
+      "loss_iou": 1.0462379455566406,
+      "loss_num": 0.3203125,
+      "loss_xval": 0.3203125,
+      "num_input_tokens_seen": 58924544,
+      "step": 341
+    },
+    {
+      "epoch": 0.13098429720413635,
+      "grad_norm": 20.40995864655311,
+      "learning_rate": 5e-06,
+      "loss": 0.4059,
+      "num_input_tokens_seen": 59097784,
+      "step": 342
+    },
+    {
+      "epoch": 0.13098429720413635,
+      "loss": 0.3902062177658081,
+      "loss_ce": 0.012032410129904747,
+      "loss_iou": 1.0180613994598389,
+      "loss_num": 0.37890625,
+      "loss_xval": 0.37890625,
+      "num_input_tokens_seen": 59097784,
+      "step": 342
+    },
+    {
+      "epoch": 0.13136729222520108,
+      "grad_norm": 30.970832407923083,
+      "learning_rate": 5e-06,
+      "loss": 0.3628,
+      "num_input_tokens_seen": 59270568,
+      "step": 343
+    },
+    {
+      "epoch": 0.13136729222520108,
+      "loss": 0.3716558814048767,
+      "loss_ce": 0.012402951717376709,
+      "loss_iou": 1.0250247716903687,
+      "loss_num": 0.359375,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 59270568,
+      "step": 343
+    },
+    {
+      "epoch": 0.1317502872462658,
+      "grad_norm": 33.1759437003962,
+      "learning_rate": 5e-06,
+      "loss": 0.5395,
+      "num_input_tokens_seen": 59443696,
+      "step": 344
+    },
+    {
+      "epoch": 0.1317502872462658,
+      "loss": 0.49835091829299927,
+      "loss_ce": 0.012022789567708969,
+      "loss_iou": 1.127155065536499,
+      "loss_num": 0.486328125,
+      "loss_xval": 0.486328125,
+      "num_input_tokens_seen": 59443696,
+      "step": 344
+    },
+    {
+      "epoch": 0.13213328226733054,
+      "grad_norm": 21.268179837572763,
+      "learning_rate": 5e-06,
+      "loss": 0.375,
+      "num_input_tokens_seen": 59616664,
+      "step": 345
+    },
+    {
+      "epoch": 0.13213328226733054,
+      "loss": 0.41703686118125916,
+      "loss_ce": 0.013228273950517178,
+      "loss_iou": 1.1385784149169922,
+      "loss_num": 0.404296875,
+      "loss_xval": 0.404296875,
+      "num_input_tokens_seen": 59616664,
+      "step": 345
+    },
+    {
+      "epoch": 0.13251627728839524,
+      "grad_norm": 22.353032834658464,
+      "learning_rate": 5e-06,
+      "loss": 0.3995,
+      "num_input_tokens_seen": 59789536,
+      "step": 346
+    },
+    {
+      "epoch": 0.13251627728839524,
+      "loss": 0.44128888845443726,
+      "loss_ce": 0.013066226616501808,
+      "loss_iou": 1.0135618448257446,
+      "loss_num": 0.427734375,
+      "loss_xval": 0.427734375,
+      "num_input_tokens_seen": 59789536,
+      "step": 346
+    },
+    {
+      "epoch": 0.13289927230945997,
+      "grad_norm": 31.340023846265844,
+      "learning_rate": 5e-06,
+      "loss": 0.3996,
+      "num_input_tokens_seen": 59962072,
+      "step": 347
+    },
+    {
+      "epoch": 0.13289927230945997,
+      "loss": 0.36322999000549316,
+      "loss_ce": 0.013132363557815552,
+      "loss_iou": 1.021392822265625,
+      "loss_num": 0.349609375,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 59962072,
+      "step": 347
+    },
+    {
+      "epoch": 0.1332822673305247,
+      "grad_norm": 30.760240558756742,
+      "learning_rate": 5e-06,
+      "loss": 0.3693,
+      "num_input_tokens_seen": 60134888,
+      "step": 348
+    },
+    {
+      "epoch": 0.1332822673305247,
+      "loss": 0.3806111812591553,
+      "loss_ce": 0.013179522007703781,
+      "loss_iou": 1.0316166877746582,
+      "loss_num": 0.3671875,
+      "loss_xval": 0.3671875,
+      "num_input_tokens_seen": 60134888,
+      "step": 348
+    },
+    {
+      "epoch": 0.13366526235158943,
+      "grad_norm": 27.02673377461334,
+      "learning_rate": 5e-06,
+      "loss": 0.3376,
+      "num_input_tokens_seen": 60307536,
+      "step": 349
+    },
+    {
+      "epoch": 0.13366526235158943,
+      "loss": 0.36044585704803467,
+      "loss_ce": 0.01364409551024437,
+      "loss_iou": 1.0658987760543823,
+      "loss_num": 0.34765625,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 60307536,
+      "step": 349
+    },
+    {
+      "epoch": 0.13404825737265416,
+      "grad_norm": 19.969380481440243,
+      "learning_rate": 5e-06,
+      "loss": 0.4478,
+      "num_input_tokens_seen": 60480200,
+      "step": 350
+    },
+    {
+      "epoch": 0.13404825737265416,
+      "loss": 0.4545081853866577,
+      "loss_ce": 0.012613639235496521,
+      "loss_iou": 1.1202318668365479,
+      "loss_num": 0.44140625,
+      "loss_xval": 0.44140625,
+      "num_input_tokens_seen": 60480200,
+      "step": 350
+    },
+    {
+      "epoch": 0.1344312523937189,
+      "grad_norm": 23.64571380270694,
+      "learning_rate": 5e-06,
+      "loss": 0.2641,
+      "num_input_tokens_seen": 60652832,
+      "step": 351
+    },
+    {
+      "epoch": 0.1344312523937189,
+      "loss": 0.24726730585098267,
+      "loss_ce": 0.012281963601708412,
+      "loss_iou": 1.019364595413208,
+      "loss_num": 0.2353515625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 60652832,
+      "step": 351
+    },
+    {
+      "epoch": 0.1348142474147836,
+      "grad_norm": 37.0483407213676,
+      "learning_rate": 5e-06,
+      "loss": 0.4657,
+      "num_input_tokens_seen": 60825920,
+      "step": 352
+    },
+    {
+      "epoch": 0.1348142474147836,
+      "loss": 0.4851919114589691,
+      "loss_ce": 0.011559093371033669,
+      "loss_iou": 1.0131995677947998,
+      "loss_num": 0.47265625,
+      "loss_xval": 0.47265625,
+      "num_input_tokens_seen": 60825920,
+      "step": 352
+    },
+    {
+      "epoch": 0.13519724243584832,
+      "grad_norm": 27.351549801074697,
+      "learning_rate": 5e-06,
+      "loss": 0.3466,
+      "num_input_tokens_seen": 60998592,
+      "step": 353
+    },
+    {
+      "epoch": 0.13519724243584832,
+      "loss": 0.2995198667049408,
+      "loss_ce": 0.013387059792876244,
+      "loss_iou": 1.0160716772079468,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 60998592,
+      "step": 353
+    },
+    {
+      "epoch": 0.13558023745691306,
+      "grad_norm": 19.41353116033888,
+      "learning_rate": 5e-06,
+      "loss": 0.5005,
+      "num_input_tokens_seen": 61171408,
+      "step": 354
+    },
+    {
+      "epoch": 0.13558023745691306,
+      "loss": 0.5307155251502991,
+      "loss_ce": 0.01289329119026661,
+      "loss_iou": 1.0501670837402344,
+      "loss_num": 0.51953125,
+      "loss_xval": 0.51953125,
+      "num_input_tokens_seen": 61171408,
+      "step": 354
+    },
+    {
+      "epoch": 0.13596323247797779,
+      "grad_norm": 17.48712667532762,
+      "learning_rate": 5e-06,
+      "loss": 0.27,
+      "num_input_tokens_seen": 61344744,
+      "step": 355
+    },
+    {
+      "epoch": 0.13596323247797779,
+      "loss": 0.30249089002609253,
+      "loss_ce": 0.013184243813157082,
+      "loss_iou": 1.1163172721862793,
+      "loss_num": 0.2890625,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 61344744,
+      "step": 355
+    },
+    {
+      "epoch": 0.13634622749904252,
+      "grad_norm": 23.233388365718046,
+      "learning_rate": 5e-06,
+      "loss": 0.3519,
+      "num_input_tokens_seen": 61517936,
+      "step": 356
+    },
+    {
+      "epoch": 0.13634622749904252,
+      "loss": 0.3717007339000702,
+      "loss_ce": 0.012569872662425041,
+      "loss_iou": 1.3388407230377197,
+      "loss_num": 0.359375,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 61517936,
+      "step": 356
+    },
+    {
+      "epoch": 0.13672922252010725,
+      "grad_norm": 26.758695720823354,
+      "learning_rate": 5e-06,
+      "loss": 0.3853,
+      "num_input_tokens_seen": 61691008,
+      "step": 357
+    },
+    {
+      "epoch": 0.13672922252010725,
+      "loss": 0.35257813334465027,
+      "loss_ce": 0.01236817054450512,
+      "loss_iou": 1.1086204051971436,
+      "loss_num": 0.33984375,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 61691008,
+      "step": 357
+    },
+    {
+      "epoch": 0.13711221754117198,
+      "grad_norm": 28.263344707617506,
+      "learning_rate": 5e-06,
+      "loss": 0.3578,
+      "num_input_tokens_seen": 61864416,
+      "step": 358
+    },
+    {
+      "epoch": 0.13711221754117198,
+      "loss": 0.3794240653514862,
+      "loss_ce": 0.013213129714131355,
+      "loss_iou": 1.0335272550582886,
+      "loss_num": 0.3671875,
+      "loss_xval": 0.3671875,
+      "num_input_tokens_seen": 61864416,
+      "step": 358
+    },
+    {
+      "epoch": 0.13749521256223668,
+      "grad_norm": 40.90973744330012,
+      "learning_rate": 5e-06,
+      "loss": 0.3265,
+      "num_input_tokens_seen": 62037032,
+      "step": 359
+    },
+    {
+      "epoch": 0.13749521256223668,
+      "loss": 0.3287930190563202,
+      "loss_ce": 0.013119187206029892,
+      "loss_iou": 1.017606258392334,
+      "loss_num": 0.31640625,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 62037032,
+      "step": 359
+    },
+    {
+      "epoch": 0.1378782075833014,
+      "grad_norm": 40.920368087423206,
+      "learning_rate": 5e-06,
+      "loss": 0.4415,
+      "num_input_tokens_seen": 62210144,
+      "step": 360
+    },
+    {
+      "epoch": 0.1378782075833014,
+      "loss": 0.3989262580871582,
+      "loss_ce": 0.012207522056996822,
+      "loss_iou": 1.0507839918136597,
+      "loss_num": 0.38671875,
+      "loss_xval": 0.38671875,
+      "num_input_tokens_seen": 62210144,
+      "step": 360
+    },
+    {
+      "epoch": 0.13826120260436614,
+      "grad_norm": 21.84843688634946,
+      "learning_rate": 5e-06,
+      "loss": 0.3067,
+      "num_input_tokens_seen": 62383016,
+      "step": 361
+    },
+    {
+      "epoch": 0.13826120260436614,
+      "loss": 0.3030824065208435,
+      "loss_ce": 0.012921245768666267,
+      "loss_iou": 1.1078020334243774,
+      "loss_num": 0.291015625,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 62383016,
+      "step": 361
+    },
+    {
+      "epoch": 0.13864419762543087,
+      "grad_norm": 25.303836611757855,
+      "learning_rate": 5e-06,
+      "loss": 0.3307,
+      "num_input_tokens_seen": 62555992,
+      "step": 362
+    },
+    {
+      "epoch": 0.13864419762543087,
+      "loss": 0.3620717227458954,
+      "loss_ce": 0.013927199877798557,
+      "loss_iou": 1.0925281047821045,
+      "loss_num": 0.34765625,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 62555992,
+      "step": 362
+    },
+    {
+      "epoch": 0.1390271926464956,
+      "grad_norm": 38.079841138035974,
+      "learning_rate": 5e-06,
+      "loss": 0.3423,
+      "num_input_tokens_seen": 62729008,
+      "step": 363
+    },
+    {
+      "epoch": 0.1390271926464956,
+      "loss": 0.34251147508621216,
+      "loss_ce": 0.01243335846811533,
+      "loss_iou": 1.069159746170044,
+      "loss_num": 0.330078125,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 62729008,
+      "step": 363
+    },
+    {
+      "epoch": 0.13941018766756033,
+      "grad_norm": 38.19344770186388,
+      "learning_rate": 5e-06,
+      "loss": 0.438,
+      "num_input_tokens_seen": 62902368,
+      "step": 364
+    },
+    {
+      "epoch": 0.13941018766756033,
+      "loss": 0.4055544137954712,
+      "loss_ce": 0.012976281344890594,
+      "loss_iou": 1.0455716848373413,
+      "loss_num": 0.392578125,
+      "loss_xval": 0.392578125,
+      "num_input_tokens_seen": 62902368,
+      "step": 364
+    },
+    {
+      "epoch": 0.13979318268862506,
+      "grad_norm": 25.96910003641756,
+      "learning_rate": 5e-06,
+      "loss": 0.3534,
+      "num_input_tokens_seen": 63075368,
+      "step": 365
+    },
+    {
+      "epoch": 0.13979318268862506,
+      "loss": 0.3414488434791565,
+      "loss_ce": 0.011614874936640263,
+      "loss_iou": 1.0137832164764404,
+      "loss_num": 0.330078125,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 63075368,
+      "step": 365
+    },
+    {
+      "epoch": 0.14017617770968976,
+      "grad_norm": 46.3796791167195,
+      "learning_rate": 5e-06,
+      "loss": 0.5287,
+      "num_input_tokens_seen": 63248064,
+      "step": 366
+    },
+    {
+      "epoch": 0.14017617770968976,
+      "loss": 0.5414717793464661,
+      "loss_ce": 0.011930737644433975,
+      "loss_iou": 1.0576518774032593,
+      "loss_num": 0.53125,
+      "loss_xval": 0.53125,
+      "num_input_tokens_seen": 63248064,
+      "step": 366
+    },
+    {
+      "epoch": 0.1405591727307545,
+      "grad_norm": 49.121315858050615,
+      "learning_rate": 5e-06,
+      "loss": 0.433,
+      "num_input_tokens_seen": 63420776,
+      "step": 367
+    },
+    {
+      "epoch": 0.1405591727307545,
+      "loss": 0.4409557580947876,
+      "loss_ce": 0.01175654400140047,
+      "loss_iou": 1.0536702871322632,
+      "loss_num": 0.4296875,
+      "loss_xval": 0.4296875,
+      "num_input_tokens_seen": 63420776,
+      "step": 367
+    },
+    {
+      "epoch": 0.14094216775181923,
+      "grad_norm": 64.61330303892356,
+      "learning_rate": 5e-06,
+      "loss": 0.8656,
+      "num_input_tokens_seen": 63590328,
+      "step": 368
+    },
+    {
+      "epoch": 0.14094216775181923,
+      "loss": 0.7870058417320251,
+      "loss_ce": 0.012591780163347721,
+      "loss_iou": 1.0539166927337646,
+      "loss_num": 0.7734375,
+      "loss_xval": 0.7734375,
+      "num_input_tokens_seen": 63590328,
+      "step": 368
+    },
+    {
+      "epoch": 0.14132516277288396,
+      "grad_norm": 24.734888518615882,
+      "learning_rate": 5e-06,
+      "loss": 0.4936,
+      "num_input_tokens_seen": 63763272,
+      "step": 369
+    },
+    {
+      "epoch": 0.14132516277288396,
+      "loss": 0.4711846113204956,
+      "loss_ce": 0.013420945033431053,
+      "loss_iou": 1.0639902353286743,
+      "loss_num": 0.45703125,
+      "loss_xval": 0.45703125,
+      "num_input_tokens_seen": 63763272,
+      "step": 369
+    },
+    {
+      "epoch": 0.14170815779394869,
+      "grad_norm": 27.347220635004692,
+      "learning_rate": 5e-06,
+      "loss": 0.6949,
+      "num_input_tokens_seen": 63936464,
+      "step": 370
+    },
+    {
+      "epoch": 0.14170815779394869,
+      "loss": 0.7150565385818481,
+      "loss_ce": 0.014861243776977062,
+      "loss_iou": 1.0729635953903198,
+      "loss_num": 0.69921875,
+      "loss_xval": 0.69921875,
+      "num_input_tokens_seen": 63936464,
+      "step": 370
+    },
+    {
+      "epoch": 0.14209115281501342,
+      "grad_norm": 51.38802471251687,
+      "learning_rate": 5e-06,
+      "loss": 0.6282,
+      "num_input_tokens_seen": 64109728,
+      "step": 371
+    },
+    {
+      "epoch": 0.14209115281501342,
+      "loss": 0.5930242538452148,
+      "loss_ce": 0.013922711834311485,
+      "loss_iou": 1.049805760383606,
+      "loss_num": 0.578125,
+      "loss_xval": 0.578125,
+      "num_input_tokens_seen": 64109728,
+      "step": 371
+    },
+    {
+      "epoch": 0.14247414783607812,
+      "grad_norm": 60.02099082142038,
+      "learning_rate": 5e-06,
+      "loss": 1.0242,
+      "num_input_tokens_seen": 64282720,
+      "step": 372
+    },
+    {
+      "epoch": 0.14247414783607812,
+      "loss": 1.0294208526611328,
+      "loss_ce": 0.01281927339732647,
+      "loss_iou": 1.1568076610565186,
+      "loss_num": 1.015625,
+      "loss_xval": 1.015625,
+      "num_input_tokens_seen": 64282720,
+      "step": 372
+    },
+    {
+      "epoch": 0.14285714285714285,
+      "grad_norm": 28.16591489985602,
+      "learning_rate": 5e-06,
+      "loss": 0.6113,
+      "num_input_tokens_seen": 64455304,
+      "step": 373
+    },
+    {
+      "epoch": 0.14285714285714285,
+      "loss": 0.5786975622177124,
+      "loss_ce": 0.014488564804196358,
+      "loss_iou": 1.6266827583312988,
+      "loss_num": 0.5625,
+      "loss_xval": 0.5625,
+      "num_input_tokens_seen": 64455304,
+      "step": 373
+    },
+    {
+      "epoch": 0.14324013787820758,
+      "grad_norm": 41.33892512644171,
+      "learning_rate": 5e-06,
+      "loss": 0.9286,
+      "num_input_tokens_seen": 64628736,
+      "step": 374
+    },
+    {
+      "epoch": 0.14324013787820758,
+      "loss": 0.9207422733306885,
+      "loss_ce": 0.014492250978946686,
+      "loss_iou": 1.0976927280426025,
+      "loss_num": 0.90625,
+      "loss_xval": 0.90625,
+      "num_input_tokens_seen": 64628736,
+      "step": 374
+    },
+    {
+      "epoch": 0.1436231328992723,
+      "grad_norm": 59.9188461455712,
+      "learning_rate": 5e-06,
+      "loss": 0.6966,
+      "num_input_tokens_seen": 64801456,
+      "step": 375
+    },
+    {
+      "epoch": 0.1436231328992723,
+      "loss": 0.6877446174621582,
+      "loss_ce": 0.012939905747771263,
+      "loss_iou": 1.0618836879730225,
+      "loss_num": 0.67578125,
+      "loss_xval": 0.67578125,
+      "num_input_tokens_seen": 64801456,
+      "step": 375
+    },
+    {
+      "epoch": 0.14400612792033704,
+      "grad_norm": 61.689073573426874,
+      "learning_rate": 5e-06,
+      "loss": 0.9012,
+      "num_input_tokens_seen": 64974208,
+      "step": 376
+    },
+    {
+      "epoch": 0.14400612792033704,
+      "loss": 0.9289538860321045,
+      "loss_ce": 0.012938244268298149,
+      "loss_iou": 1.1032987833023071,
+      "loss_num": 0.9140625,
+      "loss_xval": 0.9140625,
+      "num_input_tokens_seen": 64974208,
+      "step": 376
+    },
+    {
+      "epoch": 0.14438912294140177,
+      "grad_norm": 19.84003180347874,
+      "learning_rate": 5e-06,
+      "loss": 0.5017,
+      "num_input_tokens_seen": 65147280,
+      "step": 377
+    },
+    {
+      "epoch": 0.14438912294140177,
+      "loss": 0.524929404258728,
+      "loss_ce": 0.01614035852253437,
+      "loss_iou": 1.0543954372406006,
+      "loss_num": 0.5078125,
+      "loss_xval": 0.5078125,
+      "num_input_tokens_seen": 65147280,
+      "step": 377
+    },
+    {
+      "epoch": 0.1447721179624665,
+      "grad_norm": 24.906736164638446,
+      "learning_rate": 5e-06,
+      "loss": 0.671,
+      "num_input_tokens_seen": 65320232,
+      "step": 378
+    },
+    {
+      "epoch": 0.1447721179624665,
+      "loss": 0.6904242634773254,
+      "loss_ce": 0.015131345950067043,
+      "loss_iou": 1.0514883995056152,
+      "loss_num": 0.67578125,
+      "loss_xval": 0.67578125,
+      "num_input_tokens_seen": 65320232,
+      "step": 378
+    },
+    {
+      "epoch": 0.1451551129835312,
+      "grad_norm": 41.04443374679105,
+      "learning_rate": 5e-06,
+      "loss": 0.431,
+      "num_input_tokens_seen": 65493048,
+      "step": 379
+    },
+    {
+      "epoch": 0.1451551129835312,
+      "loss": 0.4022664427757263,
+      "loss_ce": 0.014082863926887512,
+      "loss_iou": 1.0451061725616455,
+      "loss_num": 0.388671875,
+      "loss_xval": 0.388671875,
+      "num_input_tokens_seen": 65493048,
+      "step": 379
+    },
+    {
+      "epoch": 0.14553810800459593,
+      "grad_norm": 50.32946176470696,
+      "learning_rate": 5e-06,
+      "loss": 0.8086,
+      "num_input_tokens_seen": 65665960,
+      "step": 380
+    },
+    {
+      "epoch": 0.14553810800459593,
+      "loss": 0.8396444916725159,
+      "loss_ce": 0.014937466010451317,
+      "loss_iou": 2.551274061203003,
+      "loss_num": 0.82421875,
+      "loss_xval": 0.82421875,
+      "num_input_tokens_seen": 65665960,
+      "step": 380
+    },
+    {
+      "epoch": 0.14592110302566066,
+      "grad_norm": 25.037409993565685,
+      "learning_rate": 5e-06,
+      "loss": 0.3744,
+      "num_input_tokens_seen": 65838928,
+      "step": 381
+    },
+    {
+      "epoch": 0.14592110302566066,
+      "loss": 0.378633975982666,
+      "loss_ce": 0.01535273902118206,
+      "loss_iou": 1.0505462884902954,
+      "loss_num": 0.36328125,
+      "loss_xval": 0.36328125,
+      "num_input_tokens_seen": 65838928,
+      "step": 381
+    },
+    {
+      "epoch": 0.1463040980467254,
+      "grad_norm": 31.282118782535907,
+      "learning_rate": 5e-06,
+      "loss": 0.5373,
+      "num_input_tokens_seen": 66011632,
+      "step": 382
+    },
+    {
+      "epoch": 0.1463040980467254,
+      "loss": 0.5232303142547607,
+      "loss_ce": 0.013464685529470444,
+      "loss_iou": 1.0495527982711792,
+      "loss_num": 0.5078125,
+      "loss_xval": 0.5078125,
+      "num_input_tokens_seen": 66011632,
+      "step": 382
+    },
+    {
+      "epoch": 0.14668709306779013,
+      "grad_norm": 27.643199490450233,
+      "learning_rate": 5e-06,
+      "loss": 0.4729,
+      "num_input_tokens_seen": 66184064,
+      "step": 383
+    },
+    {
+      "epoch": 0.14668709306779013,
+      "loss": 0.4467092454433441,
+      "loss_ce": 0.013481696136295795,
+      "loss_iou": 1.053185224533081,
+      "loss_num": 0.43359375,
+      "loss_xval": 0.43359375,
+      "num_input_tokens_seen": 66184064,
+      "step": 383
+    },
+    {
+      "epoch": 0.14707008808885486,
+      "grad_norm": 31.827409372569036,
+      "learning_rate": 5e-06,
+      "loss": 0.464,
+      "num_input_tokens_seen": 66356768,
+      "step": 384
+    },
+    {
+      "epoch": 0.14707008808885486,
+      "loss": 0.4203503429889679,
+      "loss_ce": 0.01434449665248394,
+      "loss_iou": 1.0519211292266846,
+      "loss_num": 0.40625,
+      "loss_xval": 0.40625,
+      "num_input_tokens_seen": 66356768,
+      "step": 384
+    },
+    {
+      "epoch": 0.14745308310991956,
+      "grad_norm": 34.496761129495205,
+      "learning_rate": 5e-06,
+      "loss": 0.4934,
+      "num_input_tokens_seen": 66529864,
+      "step": 385
+    },
+    {
+      "epoch": 0.14745308310991956,
+      "loss": 0.4703870415687561,
+      "loss_ce": 0.01433233730494976,
+      "loss_iou": 1.0265611410140991,
+      "loss_num": 0.45703125,
+      "loss_xval": 0.45703125,
+      "num_input_tokens_seen": 66529864,
+      "step": 385
+    },
+    {
+      "epoch": 0.1478360781309843,
+      "grad_norm": 26.041557954952363,
+      "learning_rate": 5e-06,
+      "loss": 0.4909,
+      "num_input_tokens_seen": 66702944,
+      "step": 386
+    },
+    {
+      "epoch": 0.1478360781309843,
+      "loss": 0.4845116138458252,
+      "loss_ce": 0.01673818938434124,
+      "loss_iou": 1.0745141506195068,
+      "loss_num": 0.46875,
+      "loss_xval": 0.46875,
+      "num_input_tokens_seen": 66702944,
+      "step": 386
+    },
+    {
+      "epoch": 0.14821907315204902,
+      "grad_norm": 32.47151434777721,
+      "learning_rate": 5e-06,
+      "loss": 0.3861,
+      "num_input_tokens_seen": 66876104,
+      "step": 387
+    },
+    {
+      "epoch": 0.14821907315204902,
+      "loss": 0.3625410497188568,
+      "loss_ce": 0.016105500981211662,
+      "loss_iou": 1.0794681310653687,
+      "loss_num": 0.345703125,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 66876104,
+      "step": 387
+    },
+    {
+      "epoch": 0.14860206817311375,
+      "grad_norm": 29.411348605919954,
+      "learning_rate": 5e-06,
+      "loss": 0.4896,
+      "num_input_tokens_seen": 67049216,
+      "step": 388
+    },
+    {
+      "epoch": 0.14860206817311375,
+      "loss": 0.4619877338409424,
+      "loss_ce": 0.014722118154168129,
+      "loss_iou": 1.6182482242584229,
+      "loss_num": 0.447265625,
+      "loss_xval": 0.447265625,
+      "num_input_tokens_seen": 67049216,
+      "step": 388
+    },
+    {
+      "epoch": 0.14898506319417848,
+      "grad_norm": 12.758152150747003,
+      "learning_rate": 5e-06,
+      "loss": 0.3211,
+      "num_input_tokens_seen": 67222088,
+      "step": 389
+    },
+    {
+      "epoch": 0.14898506319417848,
+      "loss": 0.36468395590782166,
+      "loss_ce": 0.013975952751934528,
+      "loss_iou": 1.072611689567566,
+      "loss_num": 0.3515625,
+      "loss_xval": 0.3515625,
+      "num_input_tokens_seen": 67222088,
+      "step": 389
+    },
+    {
+      "epoch": 0.1493680582152432,
+      "grad_norm": 33.42467835445704,
+      "learning_rate": 5e-06,
+      "loss": 0.2915,
+      "num_input_tokens_seen": 67394984,
+      "step": 390
+    },
+    {
+      "epoch": 0.1493680582152432,
+      "loss": 0.3005879819393158,
+      "loss_ce": 0.013722749426960945,
+      "loss_iou": 1.1104605197906494,
+      "loss_num": 0.287109375,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 67394984,
+      "step": 390
+    },
+    {
+      "epoch": 0.14975105323630794,
+      "grad_norm": 34.052127278713876,
+      "learning_rate": 5e-06,
+      "loss": 0.3981,
+      "num_input_tokens_seen": 67567776,
+      "step": 391
+    },
+    {
+      "epoch": 0.14975105323630794,
+      "loss": 0.42281055450439453,
+      "loss_ce": 0.013630880042910576,
+      "loss_iou": 1.039305329322815,
+      "loss_num": 0.41015625,
+      "loss_xval": 0.41015625,
+      "num_input_tokens_seen": 67567776,
+      "step": 391
+    },
+    {
+      "epoch": 0.15013404825737264,
+      "grad_norm": 24.50332959564291,
+      "learning_rate": 5e-06,
+      "loss": 0.3559,
+      "num_input_tokens_seen": 67740832,
+      "step": 392
+    },
+    {
+      "epoch": 0.15013404825737264,
+      "loss": 0.3538506031036377,
+      "loss_ce": 0.01486134435981512,
+      "loss_iou": 1.0434025526046753,
+      "loss_num": 0.33984375,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 67740832,
+      "step": 392
+    },
+    {
+      "epoch": 0.15051704327843737,
+      "grad_norm": 31.62897829173563,
+      "learning_rate": 5e-06,
+      "loss": 0.3742,
+      "num_input_tokens_seen": 67913800,
+      "step": 393
+    },
+    {
+      "epoch": 0.15051704327843737,
+      "loss": 0.37665513157844543,
+      "loss_ce": 0.01666979305446148,
+      "loss_iou": 1.0603320598602295,
+      "loss_num": 0.359375,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 67913800,
+      "step": 393
+    },
+    {
+      "epoch": 0.1509000382995021,
+      "grad_norm": 18.01387751534475,
+      "learning_rate": 5e-06,
+      "loss": 0.4846,
+      "num_input_tokens_seen": 68086800,
+      "step": 394
+    },
+    {
+      "epoch": 0.1509000382995021,
+      "loss": 0.4789716601371765,
+      "loss_ce": 0.013883786275982857,
+      "loss_iou": 1.030623435974121,
+      "loss_num": 0.46484375,
+      "loss_xval": 0.46484375,
+      "num_input_tokens_seen": 68086800,
+      "step": 394
+    },
+    {
+      "epoch": 0.15128303332056683,
+      "grad_norm": 22.891881277396507,
+      "learning_rate": 5e-06,
+      "loss": 0.269,
+      "num_input_tokens_seen": 68259872,
+      "step": 395
+    },
+    {
+      "epoch": 0.15128303332056683,
+      "loss": 0.2748351991176605,
+      "loss_ce": 0.015557845123112202,
+      "loss_iou": 1.0188287496566772,
+      "loss_num": 0.259765625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 68259872,
+      "step": 395
+    },
+    {
+      "epoch": 0.15166602834163156,
+      "grad_norm": 24.958916511184206,
+      "learning_rate": 5e-06,
+      "loss": 0.4305,
+      "num_input_tokens_seen": 68429376,
+      "step": 396
+    },
+    {
+      "epoch": 0.15166602834163156,
+      "loss": 0.4136584997177124,
+      "loss_ce": 0.015221013687551022,
+      "loss_iou": 1.0223636627197266,
+      "loss_num": 0.3984375,
+      "loss_xval": 0.3984375,
+      "num_input_tokens_seen": 68429376,
+      "step": 396
+    },
+    {
+      "epoch": 0.1520490233626963,
+      "grad_norm": 21.58147943119569,
+      "learning_rate": 5e-06,
+      "loss": 0.3588,
+      "num_input_tokens_seen": 68602144,
+      "step": 397
+    },
+    {
+      "epoch": 0.1520490233626963,
+      "loss": 0.3096686005592346,
+      "loss_ce": 0.015357056632637978,
+      "loss_iou": 1.025606393814087,
+      "loss_num": 0.294921875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 68602144,
+      "step": 397
+    },
+    {
+      "epoch": 0.152432018383761,
+      "grad_norm": 23.00317159852234,
+      "learning_rate": 5e-06,
+      "loss": 0.4268,
+      "num_input_tokens_seen": 68775072,
+      "step": 398
+    },
+    {
+      "epoch": 0.152432018383761,
+      "loss": 0.4410243034362793,
+      "loss_ce": 0.014998898841440678,
+      "loss_iou": 1.0346786975860596,
+      "loss_num": 0.42578125,
+      "loss_xval": 0.42578125,
+      "num_input_tokens_seen": 68775072,
+      "step": 398
+    },
+    {
+      "epoch": 0.15281501340482573,
+      "grad_norm": 22.980156215562417,
+      "learning_rate": 5e-06,
+      "loss": 0.319,
+      "num_input_tokens_seen": 68948048,
+      "step": 399
+    },
+    {
+      "epoch": 0.15281501340482573,
+      "loss": 0.3305703401565552,
+      "loss_ce": 0.014896501787006855,
+      "loss_iou": 1.210439920425415,
+      "loss_num": 0.31640625,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 68948048,
+      "step": 399
+    },
+    {
+      "epoch": 0.15319800842589046,
+      "grad_norm": 24.19557129912023,
+      "learning_rate": 5e-06,
+      "loss": 0.3657,
+      "num_input_tokens_seen": 69120840,
+      "step": 400
+    },
+    {
+      "epoch": 0.15319800842589046,
+      "loss": 0.3948410153388977,
+      "loss_ce": 0.014591965824365616,
+      "loss_iou": 0.9931958317756653,
+      "loss_num": 0.380859375,
+      "loss_xval": 0.380859375,
+      "num_input_tokens_seen": 69120840,
+      "step": 400
+    },
+    {
+      "epoch": 0.1535810034469552,
+      "grad_norm": 28.561895573863982,
+      "learning_rate": 5e-06,
+      "loss": 0.2773,
+      "num_input_tokens_seen": 69294032,
+      "step": 401
+    },
+    {
+      "epoch": 0.1535810034469552,
+      "loss": 0.23597465455532074,
+      "loss_ce": 0.015027390792965889,
+      "loss_iou": 1.0210425853729248,
+      "loss_num": 0.220703125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 69294032,
+      "step": 401
+    },
+    {
+      "epoch": 0.15396399846801992,
+      "grad_norm": 16.847300127233122,
+      "learning_rate": 5e-06,
+      "loss": 0.4095,
+      "num_input_tokens_seen": 69466776,
+      "step": 402
+    },
+    {
+      "epoch": 0.15396399846801992,
+      "loss": 0.4185844659805298,
+      "loss_ce": 0.01477588526904583,
+      "loss_iou": 1.0442343950271606,
+      "loss_num": 0.404296875,
+      "loss_xval": 0.404296875,
+      "num_input_tokens_seen": 69466776,
+      "step": 402
+    },
+    {
+      "epoch": 0.15434699348908465,
+      "grad_norm": 16.16141403050447,
+      "learning_rate": 5e-06,
+      "loss": 0.2522,
+      "num_input_tokens_seen": 69640120,
+      "step": 403
+    },
+    {
+      "epoch": 0.15434699348908465,
+      "loss": 0.2614048719406128,
+      "loss_ce": 0.01586044766008854,
+      "loss_iou": 1.0320557355880737,
+      "loss_num": 0.2451171875,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 69640120,
+      "step": 403
+    },
+    {
+      "epoch": 0.15472998851014938,
+      "grad_norm": 23.152463989977253,
+      "learning_rate": 5e-06,
+      "loss": 0.3147,
+      "num_input_tokens_seen": 69813408,
+      "step": 404
+    },
+    {
+      "epoch": 0.15472998851014938,
+      "loss": 0.2476850003004074,
+      "loss_ce": 0.013920358382165432,
+      "loss_iou": 1.002660870552063,
+      "loss_num": 0.2333984375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 69813408,
+      "step": 404
+    },
+    {
+      "epoch": 0.15511298353121408,
+      "grad_norm": 27.080087001699383,
+      "learning_rate": 5e-06,
+      "loss": 0.38,
+      "num_input_tokens_seen": 69986488,
+      "step": 405
+    },
+    {
+      "epoch": 0.15511298353121408,
+      "loss": 0.43006911873817444,
+      "loss_ce": 0.014541776850819588,
+      "loss_iou": 1.0204484462738037,
+      "loss_num": 0.416015625,
+      "loss_xval": 0.416015625,
+      "num_input_tokens_seen": 69986488,
+      "step": 405
+    },
+    {
+      "epoch": 0.1554959785522788,
+      "grad_norm": 27.415850120433483,
+      "learning_rate": 5e-06,
+      "loss": 0.3969,
+      "num_input_tokens_seen": 70159320,
+      "step": 406
+    },
+    {
+      "epoch": 0.1554959785522788,
+      "loss": 0.3625633120536804,
+      "loss_ce": 0.01588360406458378,
+      "loss_iou": 1.013737440109253,
+      "loss_num": 0.34765625,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 70159320,
+      "step": 406
+    },
+    {
+      "epoch": 0.15587897357334354,
+      "grad_norm": 23.287520317442905,
+      "learning_rate": 5e-06,
+      "loss": 0.3078,
+      "num_input_tokens_seen": 70332400,
+      "step": 407
+    },
+    {
+      "epoch": 0.15587897357334354,
+      "loss": 0.35009288787841797,
+      "loss_ce": 0.015376098453998566,
+      "loss_iou": 1.0652670860290527,
+      "loss_num": 0.333984375,
+      "loss_xval": 0.333984375,
+      "num_input_tokens_seen": 70332400,
+      "step": 407
+    },
+    {
+      "epoch": 0.15626196859440827,
+      "grad_norm": 21.034823162950822,
+      "learning_rate": 5e-06,
+      "loss": 0.4469,
+      "num_input_tokens_seen": 70505272,
+      "step": 408
+    },
+    {
+      "epoch": 0.15626196859440827,
+      "loss": 0.44711577892303467,
+      "loss_ce": 0.014986889436841011,
+      "loss_iou": 1.0202596187591553,
+      "loss_num": 0.431640625,
+      "loss_xval": 0.431640625,
+      "num_input_tokens_seen": 70505272,
+      "step": 408
+    },
+    {
+      "epoch": 0.156644963615473,
+      "grad_norm": 20.809851831139678,
+      "learning_rate": 5e-06,
+      "loss": 0.2758,
+      "num_input_tokens_seen": 70677928,
+      "step": 409
+    },
+    {
+      "epoch": 0.156644963615473,
+      "loss": 0.30263593792915344,
+      "loss_ce": 0.016014840453863144,
+      "loss_iou": 1.1829955577850342,
+      "loss_num": 0.287109375,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 70677928,
+      "step": 409
+    },
+    {
+      "epoch": 0.15702795863653773,
+      "grad_norm": 18.86130465184379,
+      "learning_rate": 5e-06,
+      "loss": 0.3039,
+      "num_input_tokens_seen": 70851096,
+      "step": 410
+    },
+    {
+      "epoch": 0.15702795863653773,
+      "loss": 0.2668569087982178,
+      "loss_ce": 0.014537591487169266,
+      "loss_iou": 1.0007319450378418,
+      "loss_num": 0.251953125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 70851096,
+      "step": 410
+    },
+    {
+      "epoch": 0.15741095365760246,
+      "grad_norm": 23.15169657964992,
+      "learning_rate": 5e-06,
+      "loss": 0.3236,
+      "num_input_tokens_seen": 71023624,
+      "step": 411
+    },
+    {
+      "epoch": 0.15741095365760246,
+      "loss": 0.26910457015037537,
+      "loss_ce": 0.014832114800810814,
+      "loss_iou": 1.0035874843597412,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 71023624,
+      "step": 411
+    },
+    {
+      "epoch": 0.15779394867866717,
+      "grad_norm": 29.247604662834046,
+      "learning_rate": 5e-06,
+      "loss": 0.4098,
+      "num_input_tokens_seen": 71196792,
+      "step": 412
+    },
+    {
+      "epoch": 0.15779394867866717,
+      "loss": 0.3962034583091736,
+      "loss_ce": 0.014245479367673397,
+      "loss_iou": 1.324742078781128,
+      "loss_num": 0.3828125,
+      "loss_xval": 0.3828125,
+      "num_input_tokens_seen": 71196792,
+      "step": 412
+    },
+    {
+      "epoch": 0.1581769436997319,
+      "grad_norm": 18.173080177086746,
+      "learning_rate": 5e-06,
+      "loss": 0.2962,
+      "num_input_tokens_seen": 71370040,
+      "step": 413
+    },
+    {
+      "epoch": 0.1581769436997319,
+      "loss": 0.2685660123825073,
+      "loss_ce": 0.015025995671749115,
+      "loss_iou": 1.0040959119796753,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 71370040,
+      "step": 413
+    },
+    {
+      "epoch": 0.15855993872079663,
+      "grad_norm": 24.228713979434175,
+      "learning_rate": 5e-06,
+      "loss": 0.4035,
+      "num_input_tokens_seen": 71542984,
+      "step": 414
+    },
+    {
+      "epoch": 0.15855993872079663,
+      "loss": 0.451690673828125,
+      "loss_ce": 0.014312762767076492,
+      "loss_iou": 2.112351894378662,
+      "loss_num": 0.4375,
+      "loss_xval": 0.4375,
+      "num_input_tokens_seen": 71542984,
+      "step": 414
+    },
+    {
+      "epoch": 0.15894293374186136,
+      "grad_norm": 28.466619697327705,
+      "learning_rate": 5e-06,
+      "loss": 0.3337,
+      "num_input_tokens_seen": 71715968,
+      "step": 415
+    },
+    {
+      "epoch": 0.15894293374186136,
+      "loss": 0.31649863719940186,
+      "loss_ce": 0.014374621212482452,
+      "loss_iou": 1.0054057836532593,
+      "loss_num": 0.302734375,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 71715968,
+      "step": 415
+    },
+    {
+      "epoch": 0.1593259287629261,
+      "grad_norm": 34.22332767495968,
+      "learning_rate": 5e-06,
+      "loss": 0.3714,
+      "num_input_tokens_seen": 71888720,
+      "step": 416
+    },
+    {
+      "epoch": 0.1593259287629261,
+      "loss": 0.3528294265270233,
+      "loss_ce": 0.014084311202168465,
+      "loss_iou": 1.0216398239135742,
+      "loss_num": 0.337890625,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 71888720,
+      "step": 416
+    },
+    {
+      "epoch": 0.15970892378399082,
+      "grad_norm": 24.9234179030379,
+      "learning_rate": 5e-06,
+      "loss": 0.3386,
+      "num_input_tokens_seen": 72061560,
+      "step": 417
+    },
+    {
+      "epoch": 0.15970892378399082,
+      "loss": 0.29265356063842773,
+      "loss_ce": 0.014943597838282585,
+      "loss_iou": 1.037266492843628,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 72061560,
+      "step": 417
+    },
+    {
+      "epoch": 0.16009191880505552,
+      "grad_norm": 26.353183066889333,
+      "learning_rate": 5e-06,
+      "loss": 0.3834,
+      "num_input_tokens_seen": 72234512,
+      "step": 418
+    },
+    {
+      "epoch": 0.16009191880505552,
+      "loss": 0.3900579810142517,
+      "loss_ce": 0.015057964250445366,
+      "loss_iou": 1.0375490188598633,
+      "loss_num": 0.375,
+      "loss_xval": 0.375,
+      "num_input_tokens_seen": 72234512,
+      "step": 418
+    },
+    {
+      "epoch": 0.16047491382612025,
+      "grad_norm": 24.393157376345613,
+      "learning_rate": 5e-06,
+      "loss": 0.2836,
+      "num_input_tokens_seen": 72407432,
+      "step": 419
+    },
+    {
+      "epoch": 0.16047491382612025,
+      "loss": 0.3062630295753479,
+      "loss_ce": 0.015979833900928497,
+      "loss_iou": 1.1124298572540283,
+      "loss_num": 0.291015625,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 72407432,
+      "step": 419
+    },
+    {
+      "epoch": 0.16085790884718498,
+      "grad_norm": 48.232176848351784,
+      "learning_rate": 5e-06,
+      "loss": 0.4717,
+      "num_input_tokens_seen": 72580496,
+      "step": 420
+    },
+    {
+      "epoch": 0.16085790884718498,
+      "loss": 0.45536208152770996,
+      "loss_ce": 0.014932382851839066,
+      "loss_iou": 1.0621998310089111,
+      "loss_num": 0.44140625,
+      "loss_xval": 0.44140625,
+      "num_input_tokens_seen": 72580496,
+      "step": 420
+    },
+    {
+      "epoch": 0.1612409038682497,
+      "grad_norm": 21.538410401949463,
+      "learning_rate": 5e-06,
+      "loss": 0.3602,
+      "num_input_tokens_seen": 72753192,
+      "step": 421
+    },
+    {
+      "epoch": 0.1612409038682497,
+      "loss": 0.3762405216693878,
+      "loss_ce": 0.01442410796880722,
+      "loss_iou": 1.0506818294525146,
+      "loss_num": 0.361328125,
+      "loss_xval": 0.361328125,
+      "num_input_tokens_seen": 72753192,
+      "step": 421
+    },
+    {
+      "epoch": 0.16162389888931444,
+      "grad_norm": 19.17179364958482,
+      "learning_rate": 5e-06,
+      "loss": 0.4197,
+      "num_input_tokens_seen": 72926376,
+      "step": 422
+    },
+    {
+      "epoch": 0.16162389888931444,
+      "loss": 0.4543018341064453,
+      "loss_ce": 0.014116259291768074,
+      "loss_iou": 1.1327141523361206,
+      "loss_num": 0.439453125,
+      "loss_xval": 0.439453125,
+      "num_input_tokens_seen": 72926376,
+      "step": 422
+    },
+    {
+      "epoch": 0.16200689391037917,
+      "grad_norm": 22.442863660160192,
+      "learning_rate": 5e-06,
+      "loss": 0.3071,
+      "num_input_tokens_seen": 73099552,
+      "step": 423
+    },
+    {
+      "epoch": 0.16200689391037917,
+      "loss": 0.33814510703086853,
+      "loss_ce": 0.014292564243078232,
+      "loss_iou": 1.0207548141479492,
+      "loss_num": 0.32421875,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 73099552,
+      "step": 423
+    },
+    {
+      "epoch": 0.1623898889314439,
+      "grad_norm": 24.46670225734681,
+      "learning_rate": 5e-06,
+      "loss": 0.4117,
+      "num_input_tokens_seen": 73272544,
+      "step": 424
+    },
+    {
+      "epoch": 0.1623898889314439,
+      "loss": 0.4213331639766693,
+      "loss_ce": 0.01398453488945961,
+      "loss_iou": 0.9296959638595581,
+      "loss_num": 0.408203125,
+      "loss_xval": 0.408203125,
+      "num_input_tokens_seen": 73272544,
+      "step": 424
+    },
+    {
+      "epoch": 0.1627728839525086,
+      "grad_norm": 22.183220184802156,
+      "learning_rate": 5e-06,
+      "loss": 0.3162,
+      "num_input_tokens_seen": 73445656,
+      "step": 425
+    },
+    {
+      "epoch": 0.1627728839525086,
+      "loss": 0.3421262800693512,
+      "loss_ce": 0.015466131269931793,
+      "loss_iou": 1.4988172054290771,
+      "loss_num": 0.326171875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 73445656,
+      "step": 425
+    },
+    {
+      "epoch": 0.16315587897357334,
+      "grad_norm": 20.306979484014494,
+      "learning_rate": 5e-06,
+      "loss": 0.3868,
+      "num_input_tokens_seen": 73618632,
+      "step": 426
+    },
+    {
+      "epoch": 0.16315587897357334,
+      "loss": 0.43750306963920593,
+      "loss_ce": 0.015872187912464142,
+      "loss_iou": 1.0614721775054932,
+      "loss_num": 0.421875,
+      "loss_xval": 0.421875,
+      "num_input_tokens_seen": 73618632,
+      "step": 426
+    },
+    {
+      "epoch": 0.16353887399463807,
+      "grad_norm": 33.29914225204966,
+      "learning_rate": 5e-06,
+      "loss": 0.3026,
+      "num_input_tokens_seen": 73791496,
+      "step": 427
+    },
+    {
+      "epoch": 0.16353887399463807,
+      "loss": 0.2640039324760437,
+      "loss_ce": 0.015712920576334,
+      "loss_iou": 1.0259437561035156,
+      "loss_num": 0.248046875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 73791496,
+      "step": 427
+    },
+    {
+      "epoch": 0.1639218690157028,
+      "grad_norm": 41.4513683196305,
+      "learning_rate": 5e-06,
+      "loss": 0.4092,
+      "num_input_tokens_seen": 73964424,
+      "step": 428
+    },
+    {
+      "epoch": 0.1639218690157028,
+      "loss": 0.36324575543403625,
+      "loss_ce": 0.014124661684036255,
+      "loss_iou": 1.0058783292770386,
+      "loss_num": 0.349609375,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 73964424,
+      "step": 428
+    },
+    {
+      "epoch": 0.16430486403676753,
+      "grad_norm": 18.832055606098304,
+      "learning_rate": 5e-06,
+      "loss": 0.2375,
+      "num_input_tokens_seen": 74137296,
+      "step": 429
+    },
+    {
+      "epoch": 0.16430486403676753,
+      "loss": 0.2725387215614319,
+      "loss_ce": 0.014482062309980392,
+      "loss_iou": 1.0127389430999756,
+      "loss_num": 0.2578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 74137296,
+      "step": 429
+    },
+    {
+      "epoch": 0.16468785905783226,
+      "grad_norm": 18.32240215523765,
+      "learning_rate": 5e-06,
+      "loss": 0.3873,
+      "num_input_tokens_seen": 74309776,
+      "step": 430
+    },
+    {
+      "epoch": 0.16468785905783226,
+      "loss": 0.35966914892196655,
+      "loss_ce": 0.013843945227563381,
+      "loss_iou": 2.9520955085754395,
+      "loss_num": 0.345703125,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 74309776,
+      "step": 430
+    },
+    {
+      "epoch": 0.16507085407889696,
+      "grad_norm": 17.887323166547127,
+      "learning_rate": 5e-06,
+      "loss": 0.2928,
+      "num_input_tokens_seen": 74483064,
+      "step": 431
+    },
+    {
+      "epoch": 0.16507085407889696,
+      "loss": 0.23529697954654694,
+      "loss_ce": 0.01593662239611149,
+      "loss_iou": 1.0336109399795532,
+      "loss_num": 0.2197265625,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 74483064,
+      "step": 431
+    },
+    {
+      "epoch": 0.1654538490999617,
+      "grad_norm": 21.04844653031928,
+      "learning_rate": 5e-06,
+      "loss": 0.3283,
+      "num_input_tokens_seen": 74656048,
+      "step": 432
+    },
+    {
+      "epoch": 0.1654538490999617,
+      "loss": 0.36195388436317444,
+      "loss_ce": 0.014297634363174438,
+      "loss_iou": 1.1591129302978516,
+      "loss_num": 0.34765625,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 74656048,
+      "step": 432
+    },
+    {
+      "epoch": 0.16583684412102642,
+      "grad_norm": 31.312609701196052,
+      "learning_rate": 5e-06,
+      "loss": 0.338,
+      "num_input_tokens_seen": 74829040,
+      "step": 433
+    },
+    {
+      "epoch": 0.16583684412102642,
+      "loss": 0.3524959683418274,
+      "loss_ce": 0.015215707942843437,
+      "loss_iou": 1.11476469039917,
+      "loss_num": 0.337890625,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 74829040,
+      "step": 433
+    },
+    {
+      "epoch": 0.16621983914209115,
+      "grad_norm": 19.451149418750603,
+      "learning_rate": 5e-06,
+      "loss": 0.3024,
+      "num_input_tokens_seen": 75001784,
+      "step": 434
+    },
+    {
+      "epoch": 0.16621983914209115,
+      "loss": 0.2773982286453247,
+      "loss_ce": 0.015191186219453812,
+      "loss_iou": 1.0119487047195435,
+      "loss_num": 0.26171875,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 75001784,
+      "step": 434
+    },
+    {
+      "epoch": 0.16660283416315588,
+      "grad_norm": 23.831570436092203,
+      "learning_rate": 5e-06,
+      "loss": 0.3795,
+      "num_input_tokens_seen": 75170824,
+      "step": 435
+    },
+    {
+      "epoch": 0.16660283416315588,
+      "loss": 0.4131200909614563,
+      "loss_ce": 0.01443847268819809,
+      "loss_iou": 1.014352798461914,
+      "loss_num": 0.3984375,
+      "loss_xval": 0.3984375,
+      "num_input_tokens_seen": 75170824,
+      "step": 435
+    },
+    {
+      "epoch": 0.1669858291842206,
+      "grad_norm": 29.056372361366613,
+      "learning_rate": 5e-06,
+      "loss": 0.3245,
+      "num_input_tokens_seen": 75343704,
+      "step": 436
+    },
+    {
+      "epoch": 0.1669858291842206,
+      "loss": 0.2704662084579468,
+      "loss_ce": 0.014973044395446777,
+      "loss_iou": 1.0120782852172852,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 75343704,
+      "step": 436
+    },
+    {
+      "epoch": 0.16736882420528534,
+      "grad_norm": 28.53054222649919,
+      "learning_rate": 5e-06,
+      "loss": 0.3515,
+      "num_input_tokens_seen": 75516808,
+      "step": 437
+    },
+    {
+      "epoch": 0.16736882420528534,
+      "loss": 0.37008148431777954,
+      "loss_ce": 0.014978909865021706,
+      "loss_iou": 1.0898067951202393,
+      "loss_num": 0.35546875,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 75516808,
+      "step": 437
+    },
+    {
+      "epoch": 0.16775181922635005,
+      "grad_norm": 29.32113828431669,
+      "learning_rate": 5e-06,
+      "loss": 0.3746,
+      "num_input_tokens_seen": 75689640,
+      "step": 438
+    },
+    {
+      "epoch": 0.16775181922635005,
+      "loss": 0.4055299162864685,
+      "loss_ce": 0.015271143987774849,
+      "loss_iou": 1.2104315757751465,
+      "loss_num": 0.390625,
+      "loss_xval": 0.390625,
+      "num_input_tokens_seen": 75689640,
+      "step": 438
+    },
+    {
+      "epoch": 0.16813481424741478,
+      "grad_norm": 37.7460884615262,
+      "learning_rate": 5e-06,
+      "loss": 0.3185,
+      "num_input_tokens_seen": 75862656,
+      "step": 439
+    },
+    {
+      "epoch": 0.16813481424741478,
+      "loss": 0.337505042552948,
+      "loss_ce": 0.015239392407238483,
+      "loss_iou": 1.0106298923492432,
+      "loss_num": 0.322265625,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 75862656,
+      "step": 439
+    },
+    {
+      "epoch": 0.1685178092684795,
+      "grad_norm": 53.12064814976083,
+      "learning_rate": 5e-06,
+      "loss": 0.4725,
+      "num_input_tokens_seen": 76035896,
+      "step": 440
+    },
+    {
+      "epoch": 0.1685178092684795,
+      "loss": 0.4797833561897278,
+      "loss_ce": 0.013474768027663231,
+      "loss_iou": 1.0127484798431396,
+      "loss_num": 0.466796875,
+      "loss_xval": 0.466796875,
+      "num_input_tokens_seen": 76035896,
+      "step": 440
+    },
+    {
+      "epoch": 0.16890080428954424,
+      "grad_norm": 28.55611723652522,
+      "learning_rate": 5e-06,
+      "loss": 0.3083,
+      "num_input_tokens_seen": 76208512,
+      "step": 441
+    },
+    {
+      "epoch": 0.16890080428954424,
+      "loss": 0.2897804379463196,
+      "loss_ce": 0.015854664146900177,
+      "loss_iou": 1.0168652534484863,
+      "loss_num": 0.2734375,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 76208512,
+      "step": 441
+    },
+    {
+      "epoch": 0.16928379931060897,
+      "grad_norm": 33.65285734014548,
+      "learning_rate": 5e-06,
+      "loss": 0.4504,
+      "num_input_tokens_seen": 76381496,
+      "step": 442
+    },
+    {
+      "epoch": 0.16928379931060897,
+      "loss": 0.45344850420951843,
+      "loss_ce": 0.01594851166009903,
+      "loss_iou": 1.0340341329574585,
+      "loss_num": 0.4375,
+      "loss_xval": 0.4375,
+      "num_input_tokens_seen": 76381496,
+      "step": 442
+    },
+    {
+      "epoch": 0.1696667943316737,
+      "grad_norm": 19.83607376978284,
+      "learning_rate": 5e-06,
+      "loss": 0.2024,
+      "num_input_tokens_seen": 76554016,
+      "step": 443
+    },
+    {
+      "epoch": 0.1696667943316737,
+      "loss": 0.2182447612285614,
+      "loss_ce": 0.01511977519840002,
+      "loss_iou": 1.00160813331604,
+      "loss_num": 0.203125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 76554016,
+      "step": 443
+    },
+    {
+      "epoch": 0.1700497893527384,
+      "grad_norm": 50.73054005217147,
+      "learning_rate": 5e-06,
+      "loss": 0.4392,
+      "num_input_tokens_seen": 76727072,
+      "step": 444
+    },
+    {
+      "epoch": 0.1700497893527384,
+      "loss": 0.4633931517601013,
+      "loss_ce": 0.013930247165262699,
+      "loss_iou": 1.4889075756072998,
+      "loss_num": 0.44921875,
+      "loss_xval": 0.44921875,
+      "num_input_tokens_seen": 76727072,
+      "step": 444
+    },
+    {
+      "epoch": 0.17043278437380313,
+      "grad_norm": 48.56503222833234,
+      "learning_rate": 5e-06,
+      "loss": 0.5239,
+      "num_input_tokens_seen": 76900224,
+      "step": 445
+    },
+    {
+      "epoch": 0.17043278437380313,
+      "loss": 0.5089391469955444,
+      "loss_ce": 0.015775062143802643,
+      "loss_iou": 1.053293228149414,
+      "loss_num": 0.4921875,
+      "loss_xval": 0.4921875,
+      "num_input_tokens_seen": 76900224,
+      "step": 445
+    },
+    {
+      "epoch": 0.17081577939486786,
+      "grad_norm": 20.773227639173367,
+      "learning_rate": 5e-06,
+      "loss": 0.4272,
+      "num_input_tokens_seen": 77073176,
+      "step": 446
+    },
+    {
+      "epoch": 0.17081577939486786,
+      "loss": 0.4128382205963135,
+      "loss_ce": 0.015865560621023178,
+      "loss_iou": 1.023116111755371,
+      "loss_num": 0.396484375,
+      "loss_xval": 0.396484375,
+      "num_input_tokens_seen": 77073176,
+      "step": 446
+    },
+    {
+      "epoch": 0.1711987744159326,
+      "grad_norm": 34.943028927085834,
+      "learning_rate": 5e-06,
+      "loss": 0.5892,
+      "num_input_tokens_seen": 77245872,
+      "step": 447
+    },
+    {
+      "epoch": 0.1711987744159326,
+      "loss": 0.5758577585220337,
+      "loss_ce": 0.0153108611702919,
+      "loss_iou": 1.0360980033874512,
+      "loss_num": 0.5625,
+      "loss_xval": 0.5625,
+      "num_input_tokens_seen": 77245872,
+      "step": 447
+    },
+    {
+      "epoch": 0.17158176943699732,
+      "grad_norm": 51.89654268093125,
+      "learning_rate": 5e-06,
+      "loss": 0.4811,
+      "num_input_tokens_seen": 77419048,
+      "step": 448
+    },
+    {
+      "epoch": 0.17158176943699732,
+      "loss": 0.45400649309158325,
+      "loss_ce": 0.014797521755099297,
+      "loss_iou": 1.245399832725525,
+      "loss_num": 0.439453125,
+      "loss_xval": 0.439453125,
+      "num_input_tokens_seen": 77419048,
+      "step": 448
+    },
+    {
+      "epoch": 0.17196476445806205,
+      "grad_norm": 61.78119051181622,
+      "learning_rate": 5e-06,
+      "loss": 0.8629,
+      "num_input_tokens_seen": 77591896,
+      "step": 449
+    },
+    {
+      "epoch": 0.17196476445806205,
+      "loss": 0.7947770357131958,
+      "loss_ce": 0.014503620564937592,
+      "loss_iou": 1.0696353912353516,
+      "loss_num": 0.78125,
+      "loss_xval": 0.78125,
+      "num_input_tokens_seen": 77591896,
+      "step": 449
+    },
+    {
+      "epoch": 0.17234775947912678,
+      "grad_norm": 28.53533006356215,
+      "learning_rate": 5e-06,
+      "loss": 0.7408,
+      "num_input_tokens_seen": 77761600,
+      "step": 450
+    },
+    {
+      "epoch": 0.17234775947912678,
+      "loss": 0.7695003747940063,
+      "loss_ce": 0.017059026286005974,
+      "loss_iou": 1.0799751281738281,
+      "loss_num": 0.75390625,
+      "loss_xval": 0.75390625,
+      "num_input_tokens_seen": 77761600,
+      "step": 450
+    },
+    {
+      "epoch": 0.17273075450019149,
+      "grad_norm": 39.79383529554913,
+      "learning_rate": 5e-06,
+      "loss": 0.7015,
+      "num_input_tokens_seen": 77934504,
+      "step": 451
+    },
+    {
+      "epoch": 0.17273075450019149,
+      "loss": 0.7374422550201416,
+      "loss_ce": 0.016739122569561005,
+      "loss_iou": 1.1498925685882568,
+      "loss_num": 0.71875,
+      "loss_xval": 0.71875,
+      "num_input_tokens_seen": 77934504,
+      "step": 451
+    },
+    {
+      "epoch": 0.17311374952125622,
+      "grad_norm": 52.263229433792276,
+      "learning_rate": 5e-06,
+      "loss": 0.7574,
+      "num_input_tokens_seen": 78107304,
+      "step": 452
+    },
+    {
+      "epoch": 0.17311374952125622,
+      "loss": 0.7194201350212097,
+      "loss_ce": 0.015318576246500015,
+      "loss_iou": 1.105474829673767,
+      "loss_num": 0.703125,
+      "loss_xval": 0.703125,
+      "num_input_tokens_seen": 78107304,
+      "step": 452
+    },
+    {
+      "epoch": 0.17349674454232095,
+      "grad_norm": 37.7961219675428,
+      "learning_rate": 5e-06,
+      "loss": 0.6433,
+      "num_input_tokens_seen": 78279896,
+      "step": 453
+    },
+    {
+      "epoch": 0.17349674454232095,
+      "loss": 0.6548479199409485,
+      "loss_ce": 0.01666434109210968,
+      "loss_iou": 1.0721873044967651,
+      "loss_num": 0.63671875,
+      "loss_xval": 0.63671875,
+      "num_input_tokens_seen": 78279896,
+      "step": 453
+    },
+    {
+      "epoch": 0.17387973956338568,
+      "grad_norm": 30.33310550199955,
+      "learning_rate": 5e-06,
+      "loss": 0.6961,
+      "num_input_tokens_seen": 78453088,
+      "step": 454
+    },
+    {
+      "epoch": 0.17387973956338568,
+      "loss": 0.6686890125274658,
+      "loss_ce": 0.016345294192433357,
+      "loss_iou": 1.0560839176177979,
+      "loss_num": 0.65234375,
+      "loss_xval": 0.65234375,
+      "num_input_tokens_seen": 78453088,
+      "step": 454
+    },
+    {
+      "epoch": 0.1742627345844504,
+      "grad_norm": 27.849510545790412,
+      "learning_rate": 5e-06,
+      "loss": 0.4943,
+      "num_input_tokens_seen": 78626048,
+      "step": 455
+    },
+    {
+      "epoch": 0.1742627345844504,
+      "loss": 0.4955611824989319,
+      "loss_ce": 0.01704554632306099,
+      "loss_iou": 1.1379518508911133,
+      "loss_num": 0.478515625,
+      "loss_xval": 0.478515625,
+      "num_input_tokens_seen": 78626048,
+      "step": 455
+    },
+    {
+      "epoch": 0.17464572960551514,
+      "grad_norm": 35.00992202568288,
+      "learning_rate": 5e-06,
+      "loss": 0.5929,
+      "num_input_tokens_seen": 78798792,
+      "step": 456
+    },
+    {
+      "epoch": 0.17464572960551514,
+      "loss": 0.6162586808204651,
+      "loss_ce": 0.016649283468723297,
+      "loss_iou": 1.0767818689346313,
+      "loss_num": 0.6015625,
+      "loss_xval": 0.6015625,
+      "num_input_tokens_seen": 78798792,
+      "step": 456
+    },
+    {
+      "epoch": 0.17502872462657987,
+      "grad_norm": 24.680822280520914,
+      "learning_rate": 5e-06,
+      "loss": 0.3934,
+      "num_input_tokens_seen": 78971744,
+      "step": 457
+    },
+    {
+      "epoch": 0.17502872462657987,
+      "loss": 0.3981289863586426,
+      "loss_ce": 0.01531648263335228,
+      "loss_iou": 1.0487823486328125,
+      "loss_num": 0.3828125,
+      "loss_xval": 0.3828125,
+      "num_input_tokens_seen": 78971744,
+      "step": 457
+    },
+    {
+      "epoch": 0.17541171964764457,
+      "grad_norm": 30.85913713643089,
+      "learning_rate": 5e-06,
+      "loss": 0.4519,
+      "num_input_tokens_seen": 79144760,
+      "step": 458
+    },
+    {
+      "epoch": 0.17541171964764457,
+      "loss": 0.46264809370040894,
+      "loss_ce": 0.01733560673892498,
+      "loss_iou": 1.066173791885376,
+      "loss_num": 0.4453125,
+      "loss_xval": 0.4453125,
+      "num_input_tokens_seen": 79144760,
+      "step": 458
+    },
+    {
+      "epoch": 0.1757947146687093,
+      "grad_norm": 32.07405643946138,
+      "learning_rate": 5e-06,
+      "loss": 0.413,
+      "num_input_tokens_seen": 79317632,
+      "step": 459
+    },
+    {
+      "epoch": 0.1757947146687093,
+      "loss": 0.43378493189811707,
+      "loss_ce": 0.017769280821084976,
+      "loss_iou": 1.0236985683441162,
+      "loss_num": 0.416015625,
+      "loss_xval": 0.416015625,
+      "num_input_tokens_seen": 79317632,
+      "step": 459
+    },
+    {
+      "epoch": 0.17617770968977403,
+      "grad_norm": 26.89925591070896,
+      "learning_rate": 5e-06,
+      "loss": 0.3841,
+      "num_input_tokens_seen": 79491000,
+      "step": 460
+    },
+    {
+      "epoch": 0.17617770968977403,
+      "loss": 0.37873604893684387,
+      "loss_ce": 0.016065148636698723,
+      "loss_iou": 1.0253320932388306,
+      "loss_num": 0.36328125,
+      "loss_xval": 0.36328125,
+      "num_input_tokens_seen": 79491000,
+      "step": 460
+    },
+    {
+      "epoch": 0.17656070471083876,
+      "grad_norm": 32.526180603764686,
+      "learning_rate": 5e-06,
+      "loss": 0.3972,
+      "num_input_tokens_seen": 79663888,
+      "step": 461
+    },
+    {
+      "epoch": 0.17656070471083876,
+      "loss": 0.4289093017578125,
+      "loss_ce": 0.015945417806506157,
+      "loss_iou": 1.243054986000061,
+      "loss_num": 0.412109375,
+      "loss_xval": 0.412109375,
+      "num_input_tokens_seen": 79663888,
+      "step": 461
+    },
+    {
+      "epoch": 0.1769436997319035,
+      "grad_norm": 29.189795519064553,
+      "learning_rate": 5e-06,
+      "loss": 0.42,
+      "num_input_tokens_seen": 79837168,
+      "step": 462
+    },
+    {
+      "epoch": 0.1769436997319035,
+      "loss": 0.4120938181877136,
+      "loss_ce": 0.01683015003800392,
+      "loss_iou": 1.0402803421020508,
+      "loss_num": 0.39453125,
+      "loss_xval": 0.39453125,
+      "num_input_tokens_seen": 79837168,
+      "step": 462
+    },
+    {
+      "epoch": 0.17732669475296822,
+      "grad_norm": 23.71756012685803,
+      "learning_rate": 5e-06,
+      "loss": 0.3112,
+      "num_input_tokens_seen": 80009976,
+      "step": 463
+    },
+    {
+      "epoch": 0.17732669475296822,
+      "loss": 0.32907718420028687,
+      "loss_ce": 0.016821343451738358,
+      "loss_iou": 1.041823148727417,
+      "loss_num": 0.3125,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 80009976,
+      "step": 463
+    },
+    {
+      "epoch": 0.17770968977403293,
+      "grad_norm": 27.954120116997547,
+      "learning_rate": 5e-06,
+      "loss": 0.3743,
+      "num_input_tokens_seen": 80182832,
+      "step": 464
+    },
+    {
+      "epoch": 0.17770968977403293,
+      "loss": 0.39653757214546204,
+      "loss_ce": 0.016898900270462036,
+      "loss_iou": 1.0429874658584595,
+      "loss_num": 0.37890625,
+      "loss_xval": 0.37890625,
+      "num_input_tokens_seen": 80182832,
+      "step": 464
+    },
+    {
+      "epoch": 0.17809268479509766,
+      "grad_norm": 38.597161628003036,
+      "learning_rate": 5e-06,
+      "loss": 0.3038,
+      "num_input_tokens_seen": 80355696,
+      "step": 465
+    },
+    {
+      "epoch": 0.17809268479509766,
+      "loss": 0.2565761208534241,
+      "loss_ce": 0.01573140360414982,
+      "loss_iou": 1.0030584335327148,
+      "loss_num": 0.2412109375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 80355696,
+      "step": 465
+    },
+    {
+      "epoch": 0.17847567981616239,
+      "grad_norm": 35.18100461780769,
+      "learning_rate": 5e-06,
+      "loss": 0.353,
+      "num_input_tokens_seen": 80528896,
+      "step": 466
+    },
+    {
+      "epoch": 0.17847567981616239,
+      "loss": 0.3511597514152527,
+      "loss_ce": 0.017419513314962387,
+      "loss_iou": 1.0240516662597656,
+      "loss_num": 0.333984375,
+      "loss_xval": 0.333984375,
+      "num_input_tokens_seen": 80528896,
+      "step": 466
+    },
+    {
+      "epoch": 0.17885867483722712,
+      "grad_norm": 31.213504805025785,
+      "learning_rate": 5e-06,
+      "loss": 0.3468,
+      "num_input_tokens_seen": 80702016,
+      "step": 467
+    },
+    {
+      "epoch": 0.17885867483722712,
+      "loss": 0.37608304619789124,
+      "loss_ce": 0.017562542110681534,
+      "loss_iou": 1.065513253211975,
+      "loss_num": 0.359375,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 80702016,
+      "step": 467
+    },
+    {
+      "epoch": 0.17924166985829185,
+      "grad_norm": 28.5635739429278,
+      "learning_rate": 5e-06,
+      "loss": 0.3127,
+      "num_input_tokens_seen": 80874832,
+      "step": 468
+    },
+    {
+      "epoch": 0.17924166985829185,
+      "loss": 0.3625808358192444,
+      "loss_ce": 0.01699979603290558,
+      "loss_iou": 1.1267366409301758,
+      "loss_num": 0.345703125,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 80874832,
+      "step": 468
+    },
+    {
+      "epoch": 0.17962466487935658,
+      "grad_norm": 21.934907619342116,
+      "learning_rate": 5e-06,
+      "loss": 0.298,
+      "num_input_tokens_seen": 81047712,
+      "step": 469
+    },
+    {
+      "epoch": 0.17962466487935658,
+      "loss": 0.2947612404823303,
+      "loss_ce": 0.016807135194540024,
+      "loss_iou": 1.0040606260299683,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 81047712,
+      "step": 469
+    },
+    {
+      "epoch": 0.1800076599004213,
+      "grad_norm": 18.3671290820128,
+      "learning_rate": 5e-06,
+      "loss": 0.4184,
+      "num_input_tokens_seen": 81220952,
+      "step": 470
+    },
+    {
+      "epoch": 0.1800076599004213,
+      "loss": 0.4173802435398102,
+      "loss_ce": 0.01698961667716503,
+      "loss_iou": 1.0343729257583618,
+      "loss_num": 0.400390625,
+      "loss_xval": 0.400390625,
+      "num_input_tokens_seen": 81220952,
+      "step": 470
+    },
+    {
+      "epoch": 0.180390654921486,
+      "grad_norm": 21.93973986022741,
+      "learning_rate": 5e-06,
+      "loss": 0.3134,
+      "num_input_tokens_seen": 81393944,
+      "step": 471
+    },
+    {
+      "epoch": 0.180390654921486,
+      "loss": 0.27204078435897827,
+      "loss_ce": 0.015326913446187973,
+      "loss_iou": 1.014489769935608,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 81393944,
+      "step": 471
+    },
+    {
+      "epoch": 0.18077364994255074,
+      "grad_norm": 29.11599156888172,
+      "learning_rate": 5e-06,
+      "loss": 0.3602,
+      "num_input_tokens_seen": 81563264,
+      "step": 472
+    },
+    {
+      "epoch": 0.18077364994255074,
+      "loss": 0.3316724896430969,
+      "loss_ce": 0.01624278910458088,
+      "loss_iou": 1.0116811990737915,
+      "loss_num": 0.31640625,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 81563264,
+      "step": 472
+    },
+    {
+      "epoch": 0.18115664496361547,
+      "grad_norm": 21.048303326103163,
+      "learning_rate": 5e-06,
+      "loss": 0.2952,
+      "num_input_tokens_seen": 81736328,
+      "step": 473
+    },
+    {
+      "epoch": 0.18115664496361547,
+      "loss": 0.26047569513320923,
+      "loss_ce": 0.016823377460241318,
+      "loss_iou": 1.0194365978240967,
+      "loss_num": 0.244140625,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 81736328,
+      "step": 473
+    },
+    {
+      "epoch": 0.1815396399846802,
+      "grad_norm": 18.105913293272476,
+      "learning_rate": 5e-06,
+      "loss": 0.3966,
+      "num_input_tokens_seen": 81909576,
+      "step": 474
+    },
+    {
+      "epoch": 0.1815396399846802,
+      "loss": 0.3882972002029419,
+      "loss_ce": 0.01671518012881279,
+      "loss_iou": 1.0324071645736694,
+      "loss_num": 0.37109375,
+      "loss_xval": 0.37109375,
+      "num_input_tokens_seen": 81909576,
+      "step": 474
+    },
+    {
+      "epoch": 0.18192263500574493,
+      "grad_norm": 27.69496716833715,
+      "learning_rate": 5e-06,
+      "loss": 0.2466,
+      "num_input_tokens_seen": 82082520,
+      "step": 475
+    },
+    {
+      "epoch": 0.18192263500574493,
+      "loss": 0.2682589292526245,
+      "loss_ce": 0.017648600041866302,
+      "loss_iou": 1.0662319660186768,
+      "loss_num": 0.25,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 82082520,
+      "step": 475
+    },
+    {
+      "epoch": 0.18230563002680966,
+      "grad_norm": 39.11406909071093,
+      "learning_rate": 5e-06,
+      "loss": 0.413,
+      "num_input_tokens_seen": 82255176,
+      "step": 476
+    },
+    {
+      "epoch": 0.18230563002680966,
+      "loss": 0.3718685507774353,
+      "loss_ce": 0.015667375177145004,
+      "loss_iou": 1.026766061782837,
+      "loss_num": 0.35546875,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 82255176,
+      "step": 476
+    },
+    {
+      "epoch": 0.18268862504787436,
+      "grad_norm": 32.52377364497424,
+      "learning_rate": 5e-06,
+      "loss": 0.2955,
+      "num_input_tokens_seen": 82428096,
+      "step": 477
+    },
+    {
+      "epoch": 0.18268862504787436,
+      "loss": 0.3095782995223999,
+      "loss_ce": 0.016853690147399902,
+      "loss_iou": 1.0867146253585815,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 82428096,
+      "step": 477
+    },
+    {
+      "epoch": 0.1830716200689391,
+      "grad_norm": 17.85674849229512,
+      "learning_rate": 5e-06,
+      "loss": 0.3755,
+      "num_input_tokens_seen": 82600960,
+      "step": 478
+    },
+    {
+      "epoch": 0.1830716200689391,
+      "loss": 0.4125042259693146,
+      "loss_ce": 0.016019873321056366,
+      "loss_iou": 1.0462924242019653,
+      "loss_num": 0.396484375,
+      "loss_xval": 0.396484375,
+      "num_input_tokens_seen": 82600960,
+      "step": 478
+    },
+    {
+      "epoch": 0.18345461509000383,
+      "grad_norm": 16.392710654275135,
+      "learning_rate": 5e-06,
+      "loss": 0.2168,
+      "num_input_tokens_seen": 82773936,
+      "step": 479
+    },
+    {
+      "epoch": 0.18345461509000383,
+      "loss": 0.19046780467033386,
+      "loss_ce": 0.015296910889446735,
+      "loss_iou": 1.0163350105285645,
+      "loss_num": 0.1748046875,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 82773936,
+      "step": 479
+    },
+    {
+      "epoch": 0.18383761011106856,
+      "grad_norm": 21.91953546551012,
+      "learning_rate": 5e-06,
+      "loss": 0.2539,
+      "num_input_tokens_seen": 82946864,
+      "step": 480
+    },
+    {
+      "epoch": 0.18383761011106856,
+      "loss": 0.2376902997493744,
+      "loss_ce": 0.016315776854753494,
+      "loss_iou": 1.0335052013397217,
+      "loss_num": 0.2216796875,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 82946864,
+      "step": 480
+    },
+    {
+      "epoch": 0.18422060513213329,
+      "grad_norm": 25.88124896687149,
+      "learning_rate": 5e-06,
+      "loss": 0.3046,
+      "num_input_tokens_seen": 83119272,
+      "step": 481
+    },
+    {
+      "epoch": 0.18422060513213329,
+      "loss": 0.29418930411338806,
+      "loss_ce": 0.013854829594492912,
+      "loss_iou": 1.0280160903930664,
+      "loss_num": 0.28125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 83119272,
+      "step": 481
+    },
+    {
+      "epoch": 0.18460360015319802,
+      "grad_norm": 25.5137148210727,
+      "learning_rate": 5e-06,
+      "loss": 0.3666,
+      "num_input_tokens_seen": 83292192,
+      "step": 482
+    },
+    {
+      "epoch": 0.18460360015319802,
+      "loss": 0.35649994015693665,
+      "loss_ce": 0.0181210245937109,
+      "loss_iou": 1.0152759552001953,
+      "loss_num": 0.337890625,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 83292192,
+      "step": 482
+    },
+    {
+      "epoch": 0.18498659517426275,
+      "grad_norm": 21.273354139372323,
+      "learning_rate": 5e-06,
+      "loss": 0.2382,
+      "num_input_tokens_seen": 83465240,
+      "step": 483
+    },
+    {
+      "epoch": 0.18498659517426275,
+      "loss": 0.2260705679655075,
+      "loss_ce": 0.01806274801492691,
+      "loss_iou": 1.017911672592163,
+      "loss_num": 0.2080078125,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 83465240,
+      "step": 483
+    },
+    {
+      "epoch": 0.18536959019532745,
+      "grad_norm": 22.990545344758896,
+      "learning_rate": 5e-06,
+      "loss": 0.3979,
+      "num_input_tokens_seen": 83638088,
+      "step": 484
+    },
+    {
+      "epoch": 0.18536959019532745,
+      "loss": 0.4391933083534241,
+      "loss_ce": 0.016097616404294968,
+      "loss_iou": 1.2297831773757935,
+      "loss_num": 0.423828125,
+      "loss_xval": 0.423828125,
+      "num_input_tokens_seen": 83638088,
+      "step": 484
+    },
+    {
+      "epoch": 0.18575258521639218,
+      "grad_norm": 15.283715848072962,
+      "learning_rate": 5e-06,
+      "loss": 0.1777,
+      "num_input_tokens_seen": 83810856,
+      "step": 485
+    },
+    {
+      "epoch": 0.18575258521639218,
+      "loss": 0.16487838327884674,
+      "loss_ce": 0.015708457678556442,
+      "loss_iou": 1.0036348104476929,
+      "loss_num": 0.1494140625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 83810856,
+      "step": 485
+    },
+    {
+      "epoch": 0.1861355802374569,
+      "grad_norm": 24.842294893652895,
+      "learning_rate": 5e-06,
+      "loss": 0.3012,
+      "num_input_tokens_seen": 83983872,
+      "step": 486
+    },
+    {
+      "epoch": 0.1861355802374569,
+      "loss": 0.2701072096824646,
+      "loss_ce": 0.017177525907754898,
+      "loss_iou": 1.0082621574401855,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 83983872,
+      "step": 486
+    },
+    {
+      "epoch": 0.18651857525852164,
+      "grad_norm": 30.158734932416355,
+      "learning_rate": 5e-06,
+      "loss": 0.3803,
+      "num_input_tokens_seen": 84156688,
+      "step": 487
+    },
+    {
+      "epoch": 0.18651857525852164,
+      "loss": 0.34104037284851074,
+      "loss_ce": 0.01572299748659134,
+      "loss_iou": 1.020815134048462,
+      "loss_num": 0.326171875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 84156688,
+      "step": 487
+    },
+    {
+      "epoch": 0.18690157027958637,
+      "grad_norm": 25.587891636501823,
+      "learning_rate": 5e-06,
+      "loss": 0.2957,
+      "num_input_tokens_seen": 84329472,
+      "step": 488
+    },
+    {
+      "epoch": 0.18690157027958637,
+      "loss": 0.26617318391799927,
+      "loss_ce": 0.016051091253757477,
+      "loss_iou": 1.0045745372772217,
+      "loss_num": 0.25,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 84329472,
+      "step": 488
+    },
+    {
+      "epoch": 0.1872845653006511,
+      "grad_norm": 22.26617325658685,
+      "learning_rate": 5e-06,
+      "loss": 0.3086,
+      "num_input_tokens_seen": 84498568,
+      "step": 489
+    },
+    {
+      "epoch": 0.1872845653006511,
+      "loss": 0.3364994525909424,
+      "loss_ce": 0.014966249465942383,
+      "loss_iou": 4.4488043785095215,
+      "loss_num": 0.322265625,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 84498568,
+      "step": 489
+    },
+    {
+      "epoch": 0.1876675603217158,
+      "grad_norm": 36.737892807519984,
+      "learning_rate": 5e-06,
+      "loss": 0.4123,
+      "num_input_tokens_seen": 84671392,
+      "step": 490
+    },
+    {
+      "epoch": 0.1876675603217158,
+      "loss": 0.41526561975479126,
+      "loss_ce": 0.016095727682113647,
+      "loss_iou": 1.0190294981002808,
+      "loss_num": 0.3984375,
+      "loss_xval": 0.3984375,
+      "num_input_tokens_seen": 84671392,
+      "step": 490
+    },
+    {
+      "epoch": 0.18805055534278053,
+      "grad_norm": 24.295039691533272,
+      "learning_rate": 5e-06,
+      "loss": 0.4046,
+      "num_input_tokens_seen": 84844384,
+      "step": 491
+    },
+    {
+      "epoch": 0.18805055534278053,
+      "loss": 0.43257319927215576,
+      "loss_ce": 0.01655757799744606,
+      "loss_iou": 1.3567330837249756,
+      "loss_num": 0.416015625,
+      "loss_xval": 0.416015625,
+      "num_input_tokens_seen": 84844384,
+      "step": 491
+    },
+    {
+      "epoch": 0.18843355036384526,
+      "grad_norm": 25.16055679726767,
+      "learning_rate": 5e-06,
+      "loss": 0.38,
+      "num_input_tokens_seen": 85017424,
+      "step": 492
+    },
+    {
+      "epoch": 0.18843355036384526,
+      "loss": 0.3606225848197937,
+      "loss_ce": 0.017849139869213104,
+      "loss_iou": 1.0175882577896118,
+      "loss_num": 0.34375,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 85017424,
+      "step": 492
+    },
+    {
+      "epoch": 0.18881654538491,
+      "grad_norm": 25.70617228138958,
+      "learning_rate": 5e-06,
+      "loss": 0.3107,
+      "num_input_tokens_seen": 85190704,
+      "step": 493
+    },
+    {
+      "epoch": 0.18881654538491,
+      "loss": 0.27605336904525757,
+      "loss_ce": 0.016776029020547867,
+      "loss_iou": 1.0141079425811768,
+      "loss_num": 0.259765625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 85190704,
+      "step": 493
+    },
+    {
+      "epoch": 0.18919954040597473,
+      "grad_norm": 29.598310923820303,
+      "learning_rate": 5e-06,
+      "loss": 0.4076,
+      "num_input_tokens_seen": 85363688,
+      "step": 494
+    },
+    {
+      "epoch": 0.18919954040597473,
+      "loss": 0.3656034469604492,
+      "loss_ce": 0.01672646962106228,
+      "loss_iou": 1.0188572406768799,
+      "loss_num": 0.349609375,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 85363688,
+      "step": 494
+    },
+    {
+      "epoch": 0.18958253542703946,
+      "grad_norm": 24.145249013799273,
+      "learning_rate": 5e-06,
+      "loss": 0.2913,
+      "num_input_tokens_seen": 85536736,
+      "step": 495
+    },
+    {
+      "epoch": 0.18958253542703946,
+      "loss": 0.28448039293289185,
+      "loss_ce": 0.016902253031730652,
+      "loss_iou": 1.0133237838745117,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 85536736,
+      "step": 495
+    },
+    {
+      "epoch": 0.1899655304481042,
+      "grad_norm": 20.78818998843579,
+      "learning_rate": 5e-06,
+      "loss": 0.4157,
+      "num_input_tokens_seen": 85709432,
+      "step": 496
+    },
+    {
+      "epoch": 0.1899655304481042,
+      "loss": 0.46383488178253174,
+      "loss_ce": 0.01559271290898323,
+      "loss_iou": 1.074376106262207,
+      "loss_num": 0.44921875,
+      "loss_xval": 0.44921875,
+      "num_input_tokens_seen": 85709432,
+      "step": 496
+    },
+    {
+      "epoch": 0.1903485254691689,
+      "grad_norm": 16.754483725629253,
+      "learning_rate": 5e-06,
+      "loss": 0.3812,
+      "num_input_tokens_seen": 85882104,
+      "step": 497
+    },
+    {
+      "epoch": 0.1903485254691689,
+      "loss": 0.3708678185939789,
+      "loss_ce": 0.017047010362148285,
+      "loss_iou": 2.423377752304077,
+      "loss_num": 0.353515625,
+      "loss_xval": 0.353515625,
+      "num_input_tokens_seen": 85882104,
+      "step": 497
+    },
+    {
+      "epoch": 0.19073152049023362,
+      "grad_norm": 18.086234825479977,
+      "learning_rate": 5e-06,
+      "loss": 0.2875,
+      "num_input_tokens_seen": 86055216,
+      "step": 498
+    },
+    {
+      "epoch": 0.19073152049023362,
+      "loss": 0.27319908142089844,
+      "loss_ce": 0.01660730317234993,
+      "loss_iou": 1.0002715587615967,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 86055216,
+      "step": 498
+    },
+    {
+      "epoch": 0.19111451551129835,
+      "grad_norm": 28.51022351505333,
+      "learning_rate": 5e-06,
+      "loss": 0.3366,
+      "num_input_tokens_seen": 86228040,
+      "step": 499
+    },
+    {
+      "epoch": 0.19111451551129835,
+      "loss": 0.41052719950675964,
+      "loss_ce": 0.017582854256033897,
+      "loss_iou": 1.1613000631332397,
+      "loss_num": 0.392578125,
+      "loss_xval": 0.392578125,
+      "num_input_tokens_seen": 86228040,
+      "step": 499
+    },
+    {
+      "epoch": 0.19149751053236308,
+      "grad_norm": 31.11421551827568,
+      "learning_rate": 5e-06,
+      "loss": 0.3285,
+      "num_input_tokens_seen": 86401016,
+      "step": 500
+    },
+    {
+      "epoch": 0.19149751053236308,
+      "eval_websight_new_CIoU": 0.8393459916114807,
+      "eval_websight_new_GIoU": 0.8380286693572998,
+      "eval_websight_new_IoU": 0.8403661549091339,
+      "eval_websight_new_MAE_all": 0.01726989820599556,
+      "eval_websight_new_MAE_h": 0.01403492921963334,
+      "eval_websight_new_MAE_w": 0.025333072058856487,
+      "eval_websight_new_MAE_x": 0.015599821228533983,
+      "eval_websight_new_MAE_y": 0.01411177171394229,
+      "eval_websight_new_NUM_probability": 0.0007267279142979532,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.21308483183383942,
+      "eval_websight_new_loss_ce": 0.03257010132074356,
+      "eval_websight_new_loss_iou": 1.0014038681983948,
+      "eval_websight_new_loss_num": 0.18133544921875,
+      "eval_websight_new_loss_xval": 0.18133544921875,
+      "eval_websight_new_runtime": 55.4241,
+      "eval_websight_new_samples_per_second": 0.902,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 86401016,
+      "step": 500
+    },
+    {
+      "epoch": 0.19149751053236308,
+      "eval_seeclick_CIoU": 0.5755606889724731,
+      "eval_seeclick_GIoU": 0.5703078657388687,
+      "eval_seeclick_IoU": 0.6038848459720612,
+      "eval_seeclick_MAE_all": 0.06362584792077541,
+      "eval_seeclick_MAE_h": 0.05255229026079178,
+      "eval_seeclick_MAE_w": 0.07664298266172409,
+      "eval_seeclick_MAE_x": 0.0703110322356224,
+      "eval_seeclick_MAE_y": 0.05499708279967308,
+      "eval_seeclick_NUM_probability": 0.001212626346386969,
+      "eval_seeclick_inside_bbox": 0.8854166567325592,
+      "eval_seeclick_loss": 0.5587007999420166,
+      "eval_seeclick_loss_ce": 0.02439273614436388,
+      "eval_seeclick_loss_iou": 1.0927678346633911,
+      "eval_seeclick_loss_num": 0.481201171875,
+      "eval_seeclick_loss_xval": 0.481201171875,
+      "eval_seeclick_runtime": 81.5555,
+      "eval_seeclick_samples_per_second": 0.613,
+      "eval_seeclick_steps_per_second": 0.025,
+      "num_input_tokens_seen": 86401016,
+      "step": 500
+    },
+    {
+      "epoch": 0.19149751053236308,
+      "eval_icons_CIoU": 0.7974367737770081,
+      "eval_icons_GIoU": 0.7924989759922028,
+      "eval_icons_IoU": 0.8080669343471527,
+      "eval_icons_MAE_all": 0.028948302380740643,
+      "eval_icons_MAE_h": 0.024239955469965935,
+      "eval_icons_MAE_w": 0.03740705642849207,
+      "eval_icons_MAE_x": 0.03228426072746515,
+      "eval_icons_MAE_y": 0.02186194295063615,
+      "eval_icons_NUM_probability": 0.0007074489549268037,
+      "eval_icons_inside_bbox": 0.9722222089767456,
+      "eval_icons_loss": 0.21479381620883942,
+      "eval_icons_loss_ce": 0.02179853804409504,
+      "eval_icons_loss_iou": 1.0462433099746704,
+      "eval_icons_loss_num": 0.179931640625,
+      "eval_icons_loss_xval": 0.179931640625,
+      "eval_icons_runtime": 81.8569,
+      "eval_icons_samples_per_second": 0.611,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 86401016,
+      "step": 500
+    },
+    {
+      "epoch": 0.19149751053236308,
+      "loss": 0.23735055327415466,
+      "loss_ce": 0.021835414692759514,
+      "loss_iou": 1.0912595987319946,
+      "loss_num": 0.2158203125,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 86401016,
+      "step": 500
+    },
+    {
+      "epoch": 0.1918805055534278,
+      "grad_norm": 31.044111730894407,
+      "learning_rate": 5e-06,
+      "loss": 0.2618,
+      "num_input_tokens_seen": 86574472,
+      "step": 501
+    },
+    {
+      "epoch": 0.1918805055534278,
+      "loss": 0.24594920873641968,
+      "loss_ce": 0.017677726224064827,
+      "loss_iou": 1.0402929782867432,
+      "loss_num": 0.228515625,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 86574472,
+      "step": 501
+    },
+    {
+      "epoch": 0.19226350057449254,
+      "grad_norm": 45.46735147246926,
+      "learning_rate": 5e-06,
+      "loss": 0.3916,
+      "num_input_tokens_seen": 86747880,
+      "step": 502
+    },
+    {
+      "epoch": 0.19226350057449254,
+      "loss": 0.3739714026451111,
+      "loss_ce": 0.018136456608772278,
+      "loss_iou": 1.0316057205200195,
+      "loss_num": 0.35546875,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 86747880,
+      "step": 502
+    },
+    {
+      "epoch": 0.19264649559555727,
+      "grad_norm": 42.728156691624314,
+      "learning_rate": 5e-06,
+      "loss": 0.3649,
+      "num_input_tokens_seen": 86921424,
+      "step": 503
+    },
+    {
+      "epoch": 0.19264649559555727,
+      "loss": 0.3752208352088928,
+      "loss_ce": 0.016578257083892822,
+      "loss_iou": 1.0332841873168945,
+      "loss_num": 0.359375,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 86921424,
+      "step": 503
+    },
+    {
+      "epoch": 0.19302949061662197,
+      "grad_norm": 39.50231245137594,
+      "learning_rate": 5e-06,
+      "loss": 0.3695,
+      "num_input_tokens_seen": 87094280,
+      "step": 504
+    },
+    {
+      "epoch": 0.19302949061662197,
+      "loss": 0.3911907374858856,
+      "loss_ce": 0.01643488183617592,
+      "loss_iou": 1.0340772867202759,
+      "loss_num": 0.375,
+      "loss_xval": 0.375,
+      "num_input_tokens_seen": 87094280,
+      "step": 504
+    },
+    {
+      "epoch": 0.1934124856376867,
+      "grad_norm": 17.28129230275025,
+      "learning_rate": 5e-06,
+      "loss": 0.2657,
+      "num_input_tokens_seen": 87267304,
+      "step": 505
+    },
+    {
+      "epoch": 0.1934124856376867,
+      "loss": 0.2689210772514343,
+      "loss_ce": 0.015747249126434326,
+      "loss_iou": 1.0793946981430054,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 87267304,
+      "step": 505
+    },
+    {
+      "epoch": 0.19379548065875143,
+      "grad_norm": 21.802907956959405,
+      "learning_rate": 5e-06,
+      "loss": 0.304,
+      "num_input_tokens_seen": 87440528,
+      "step": 506
+    },
+    {
+      "epoch": 0.19379548065875143,
+      "loss": 0.198129802942276,
+      "loss_ce": 0.016794349998235703,
+      "loss_iou": 1.0124921798706055,
+      "loss_num": 0.181640625,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 87440528,
+      "step": 506
+    },
+    {
+      "epoch": 0.19417847567981616,
+      "grad_norm": 39.47269099258926,
+      "learning_rate": 5e-06,
+      "loss": 0.358,
+      "num_input_tokens_seen": 87613536,
+      "step": 507
+    },
+    {
+      "epoch": 0.19417847567981616,
+      "loss": 0.39029720425605774,
+      "loss_ce": 0.016517896205186844,
+      "loss_iou": 1.0555205345153809,
+      "loss_num": 0.373046875,
+      "loss_xval": 0.373046875,
+      "num_input_tokens_seen": 87613536,
+      "step": 507
+    },
+    {
+      "epoch": 0.1945614707008809,
+      "grad_norm": 34.08258050384075,
+      "learning_rate": 5e-06,
+      "loss": 0.3682,
+      "num_input_tokens_seen": 87786504,
+      "step": 508
+    },
+    {
+      "epoch": 0.1945614707008809,
+      "loss": 0.33308690786361694,
+      "loss_ce": 0.016802724450826645,
+      "loss_iou": 1.0751721858978271,
+      "loss_num": 0.31640625,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 87786504,
+      "step": 508
+    },
+    {
+      "epoch": 0.19494446572194563,
+      "grad_norm": 22.0627334451357,
+      "learning_rate": 5e-06,
+      "loss": 0.3496,
+      "num_input_tokens_seen": 87959584,
+      "step": 509
+    },
+    {
+      "epoch": 0.19494446572194563,
+      "loss": 0.35061129927635193,
+      "loss_ce": 0.017237259075045586,
+      "loss_iou": 1.0571727752685547,
+      "loss_num": 0.333984375,
+      "loss_xval": 0.333984375,
+      "num_input_tokens_seen": 87959584,
+      "step": 509
+    },
+    {
+      "epoch": 0.19532746074301033,
+      "grad_norm": 35.033535187685835,
+      "learning_rate": 5e-06,
+      "loss": 0.3539,
+      "num_input_tokens_seen": 88132872,
+      "step": 510
+    },
+    {
+      "epoch": 0.19532746074301033,
+      "loss": 0.35290834307670593,
+      "loss_ce": 0.017092909663915634,
+      "loss_iou": 1.0979015827178955,
+      "loss_num": 0.3359375,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 88132872,
+      "step": 510
+    },
+    {
+      "epoch": 0.19571045576407506,
+      "grad_norm": 38.48371967756046,
+      "learning_rate": 5e-06,
+      "loss": 0.3068,
+      "num_input_tokens_seen": 88305984,
+      "step": 511
+    },
+    {
+      "epoch": 0.19571045576407506,
+      "loss": 0.3534683287143707,
+      "loss_ce": 0.017042554914951324,
+      "loss_iou": 1.103144645690918,
+      "loss_num": 0.3359375,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 88305984,
+      "step": 511
+    },
+    {
+      "epoch": 0.1960934507851398,
+      "grad_norm": 46.43906185076253,
+      "learning_rate": 5e-06,
+      "loss": 0.4626,
+      "num_input_tokens_seen": 88479016,
+      "step": 512
+    },
+    {
+      "epoch": 0.1960934507851398,
+      "loss": 0.5185445547103882,
+      "loss_ce": 0.017812136560678482,
+      "loss_iou": 1.2927305698394775,
+      "loss_num": 0.5,
+      "loss_xval": 0.5,
+      "num_input_tokens_seen": 88479016,
+      "step": 512
+    },
+    {
+      "epoch": 0.19647644580620452,
+      "grad_norm": 29.630687927421953,
+      "learning_rate": 5e-06,
+      "loss": 0.3324,
+      "num_input_tokens_seen": 88651728,
+      "step": 513
+    },
+    {
+      "epoch": 0.19647644580620452,
+      "loss": 0.3619629144668579,
+      "loss_ce": 0.01638186164200306,
+      "loss_iou": 1.2043843269348145,
+      "loss_num": 0.345703125,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 88651728,
+      "step": 513
+    },
+    {
+      "epoch": 0.19685944082726925,
+      "grad_norm": 24.973558380069395,
+      "learning_rate": 5e-06,
+      "loss": 0.3532,
+      "num_input_tokens_seen": 88824632,
+      "step": 514
+    },
+    {
+      "epoch": 0.19685944082726925,
+      "loss": 0.3557354807853699,
+      "loss_ce": 0.016990378499031067,
+      "loss_iou": 1.0580698251724243,
+      "loss_num": 0.337890625,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 88824632,
+      "step": 514
+    },
+    {
+      "epoch": 0.19724243584833398,
+      "grad_norm": 39.994809863245415,
+      "learning_rate": 5e-06,
+      "loss": 0.306,
+      "num_input_tokens_seen": 88997800,
+      "step": 515
+    },
+    {
+      "epoch": 0.19724243584833398,
+      "loss": 0.3369375765323639,
+      "loss_ce": 0.017235442996025085,
+      "loss_iou": 1.0511664152145386,
+      "loss_num": 0.3203125,
+      "loss_xval": 0.3203125,
+      "num_input_tokens_seen": 88997800,
+      "step": 515
+    },
+    {
+      "epoch": 0.1976254308693987,
+      "grad_norm": 36.78717112901744,
+      "learning_rate": 5e-06,
+      "loss": 0.5048,
+      "num_input_tokens_seen": 89170832,
+      "step": 516
+    },
+    {
+      "epoch": 0.1976254308693987,
+      "loss": 0.5017112493515015,
+      "loss_ce": 0.017092112451791763,
+      "loss_iou": 1.0794587135314941,
+      "loss_num": 0.484375,
+      "loss_xval": 0.484375,
+      "num_input_tokens_seen": 89170832,
+      "step": 516
+    },
+    {
+      "epoch": 0.1980084258904634,
+      "grad_norm": 12.642036382767236,
+      "learning_rate": 5e-06,
+      "loss": 0.2768,
+      "num_input_tokens_seen": 89343856,
+      "step": 517
+    },
+    {
+      "epoch": 0.1980084258904634,
+      "loss": 0.25486576557159424,
+      "loss_ce": 0.018537653610110283,
+      "loss_iou": 1.0257585048675537,
+      "loss_num": 0.236328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 89343856,
+      "step": 517
+    },
+    {
+      "epoch": 0.19839142091152814,
+      "grad_norm": 23.205769289092615,
+      "learning_rate": 5e-06,
+      "loss": 0.3597,
+      "num_input_tokens_seen": 89517368,
+      "step": 518
+    },
+    {
+      "epoch": 0.19839142091152814,
+      "loss": 0.36347025632858276,
+      "loss_ce": 0.01581399515271187,
+      "loss_iou": 1.0005018711090088,
+      "loss_num": 0.34765625,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 89517368,
+      "step": 518
+    },
+    {
+      "epoch": 0.19877441593259287,
+      "grad_norm": 42.05917832773593,
+      "learning_rate": 5e-06,
+      "loss": 0.409,
+      "num_input_tokens_seen": 89690656,
+      "step": 519
+    },
+    {
+      "epoch": 0.19877441593259287,
+      "loss": 0.4221374988555908,
+      "loss_ce": 0.016864046454429626,
+      "loss_iou": 1.0211807489395142,
+      "loss_num": 0.40625,
+      "loss_xval": 0.40625,
+      "num_input_tokens_seen": 89690656,
+      "step": 519
+    },
+    {
+      "epoch": 0.1991574109536576,
+      "grad_norm": 42.61402662722592,
+      "learning_rate": 5e-06,
+      "loss": 0.4126,
+      "num_input_tokens_seen": 89863760,
+      "step": 520
+    },
+    {
+      "epoch": 0.1991574109536576,
+      "loss": 0.3987889587879181,
+      "loss_ce": 0.016953013837337494,
+      "loss_iou": 1.1257686614990234,
+      "loss_num": 0.3828125,
+      "loss_xval": 0.3828125,
+      "num_input_tokens_seen": 89863760,
+      "step": 520
+    },
+    {
+      "epoch": 0.19954040597472233,
+      "grad_norm": 21.30541945935348,
+      "learning_rate": 5e-06,
+      "loss": 0.3476,
+      "num_input_tokens_seen": 90036776,
+      "step": 521
+    },
+    {
+      "epoch": 0.19954040597472233,
+      "loss": 0.3476550877094269,
+      "loss_ce": 0.016600407660007477,
+      "loss_iou": 1.0270713567733765,
+      "loss_num": 0.33203125,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 90036776,
+      "step": 521
+    },
+    {
+      "epoch": 0.19992340099578707,
+      "grad_norm": 24.706978334124425,
+      "learning_rate": 5e-06,
+      "loss": 0.3107,
+      "num_input_tokens_seen": 90209568,
+      "step": 522
+    },
+    {
+      "epoch": 0.19992340099578707,
+      "loss": 0.27285027503967285,
+      "loss_ce": 0.016990892589092255,
+      "loss_iou": 1.004989743232727,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 90209568,
+      "step": 522
+    },
+    {
+      "epoch": 0.20030639601685177,
+      "grad_norm": 71.3725387294566,
+      "learning_rate": 5e-06,
+      "loss": 0.3188,
+      "num_input_tokens_seen": 90382840,
+      "step": 523
+    },
+    {
+      "epoch": 0.20030639601685177,
+      "loss": 0.2938591241836548,
+      "loss_ce": 0.01602707989513874,
+      "loss_iou": 1.00357985496521,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 90382840,
+      "step": 523
+    },
+    {
+      "epoch": 0.2006893910379165,
+      "grad_norm": 48.48247472073512,
+      "learning_rate": 5e-06,
+      "loss": 0.4242,
+      "num_input_tokens_seen": 90555888,
+      "step": 524
+    },
+    {
+      "epoch": 0.2006893910379165,
+      "loss": 0.45546627044677734,
+      "loss_ce": 0.016867617145180702,
+      "loss_iou": 1.0269975662231445,
+      "loss_num": 0.439453125,
+      "loss_xval": 0.439453125,
+      "num_input_tokens_seen": 90555888,
+      "step": 524
+    },
+    {
+      "epoch": 0.20107238605898123,
+      "grad_norm": 35.65064728873406,
+      "learning_rate": 5e-06,
+      "loss": 0.3202,
+      "num_input_tokens_seen": 90729064,
+      "step": 525
+    },
+    {
+      "epoch": 0.20107238605898123,
+      "loss": 0.283124715089798,
+      "loss_ce": 0.01774386689066887,
+      "loss_iou": 1.0232046842575073,
+      "loss_num": 0.265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 90729064,
+      "step": 525
+    },
+    {
+      "epoch": 0.20145538108004596,
+      "grad_norm": 23.407253873121075,
+      "learning_rate": 5e-06,
+      "loss": 0.4186,
+      "num_input_tokens_seen": 90901784,
+      "step": 526
+    },
+    {
+      "epoch": 0.20145538108004596,
+      "loss": 0.3516148328781128,
+      "loss_ce": 0.01604355499148369,
+      "loss_iou": 1.0675827264785767,
+      "loss_num": 0.3359375,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 90901784,
+      "step": 526
+    },
+    {
+      "epoch": 0.2018383761011107,
+      "grad_norm": 35.4392944721751,
+      "learning_rate": 5e-06,
+      "loss": 0.3056,
+      "num_input_tokens_seen": 91074584,
+      "step": 527
+    },
+    {
+      "epoch": 0.2018383761011107,
+      "loss": 0.2868385910987854,
+      "loss_ce": 0.019138380885124207,
+      "loss_iou": 1.0470924377441406,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 91074584,
+      "step": 527
+    },
+    {
+      "epoch": 0.20222137112217542,
+      "grad_norm": 42.69896005227747,
+      "learning_rate": 5e-06,
+      "loss": 0.4509,
+      "num_input_tokens_seen": 91247608,
+      "step": 528
+    },
+    {
+      "epoch": 0.20222137112217542,
+      "loss": 0.4155544340610504,
+      "loss_ce": 0.016994860023260117,
+      "loss_iou": 1.0653858184814453,
+      "loss_num": 0.3984375,
+      "loss_xval": 0.3984375,
+      "num_input_tokens_seen": 91247608,
+      "step": 528
+    },
+    {
+      "epoch": 0.20260436614324015,
+      "grad_norm": 23.307668553653006,
+      "learning_rate": 5e-06,
+      "loss": 0.3109,
+      "num_input_tokens_seen": 91420448,
+      "step": 529
+    },
+    {
+      "epoch": 0.20260436614324015,
+      "loss": 0.3076122999191284,
+      "loss_ce": 0.016474585980176926,
+      "loss_iou": 1.058153748512268,
+      "loss_num": 0.291015625,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 91420448,
+      "step": 529
+    },
+    {
+      "epoch": 0.20298736116430485,
+      "grad_norm": 20.113420299178443,
+      "learning_rate": 5e-06,
+      "loss": 0.2891,
+      "num_input_tokens_seen": 91593160,
+      "step": 530
+    },
+    {
+      "epoch": 0.20298736116430485,
+      "loss": 0.2598230242729187,
+      "loss_ce": 0.0186731219291687,
+      "loss_iou": 1.0503168106079102,
+      "loss_num": 0.2412109375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 91593160,
+      "step": 530
+    },
+    {
+      "epoch": 0.20337035618536958,
+      "grad_norm": 37.42454102058793,
+      "learning_rate": 5e-06,
+      "loss": 0.2736,
+      "num_input_tokens_seen": 91766320,
+      "step": 531
+    },
+    {
+      "epoch": 0.20337035618536958,
+      "loss": 0.2909454107284546,
+      "loss_ce": 0.017874110490083694,
+      "loss_iou": 1.0835816860198975,
+      "loss_num": 0.2734375,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 91766320,
+      "step": 531
+    },
+    {
+      "epoch": 0.2037533512064343,
+      "grad_norm": 33.95476265655711,
+      "learning_rate": 5e-06,
+      "loss": 0.2945,
+      "num_input_tokens_seen": 91939392,
+      "step": 532
+    },
+    {
+      "epoch": 0.2037533512064343,
+      "loss": 0.3024279773235321,
+      "loss_ce": 0.017515864223241806,
+      "loss_iou": 1.0425469875335693,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 91939392,
+      "step": 532
+    },
+    {
+      "epoch": 0.20413634622749904,
+      "grad_norm": 22.947047229652018,
+      "learning_rate": 5e-06,
+      "loss": 0.2747,
+      "num_input_tokens_seen": 92112552,
+      "step": 533
+    },
+    {
+      "epoch": 0.20413634622749904,
+      "loss": 0.3030683100223541,
+      "loss_ce": 0.017057567834854126,
+      "loss_iou": 1.0824424028396606,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 92112552,
+      "step": 533
+    },
+    {
+      "epoch": 0.20451934124856377,
+      "grad_norm": 21.074821286054245,
+      "learning_rate": 5e-06,
+      "loss": 0.263,
+      "num_input_tokens_seen": 92285512,
+      "step": 534
+    },
+    {
+      "epoch": 0.20451934124856377,
+      "loss": 0.26460492610931396,
+      "loss_ce": 0.01814495027065277,
+      "loss_iou": 1.002181887626648,
+      "loss_num": 0.24609375,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 92285512,
+      "step": 534
+    },
+    {
+      "epoch": 0.2049023362696285,
+      "grad_norm": 29.429532237334737,
+      "learning_rate": 5e-06,
+      "loss": 0.234,
+      "num_input_tokens_seen": 92458944,
+      "step": 535
+    },
+    {
+      "epoch": 0.2049023362696285,
+      "loss": 0.2297022044658661,
+      "loss_ce": 0.018398497253656387,
+      "loss_iou": 1.032349705696106,
+      "loss_num": 0.2109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 92458944,
+      "step": 535
+    },
+    {
+      "epoch": 0.20528533129069323,
+      "grad_norm": 36.55545150405923,
+      "learning_rate": 5e-06,
+      "loss": 0.4114,
+      "num_input_tokens_seen": 92631848,
+      "step": 536
+    },
+    {
+      "epoch": 0.20528533129069323,
+      "loss": 0.3448227345943451,
+      "loss_ce": 0.017186010256409645,
+      "loss_iou": 1.0075856447219849,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 92631848,
+      "step": 536
+    },
+    {
+      "epoch": 0.20566832631175794,
+      "grad_norm": 26.7778727902466,
+      "learning_rate": 5e-06,
+      "loss": 0.2932,
+      "num_input_tokens_seen": 92805192,
+      "step": 537
+    },
+    {
+      "epoch": 0.20566832631175794,
+      "loss": 0.24786943197250366,
+      "loss_ce": 0.01770586520433426,
+      "loss_iou": 1.0099345445632935,
+      "loss_num": 0.23046875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 92805192,
+      "step": 537
+    },
+    {
+      "epoch": 0.20605132133282267,
+      "grad_norm": 17.330358694147566,
+      "learning_rate": 5e-06,
+      "loss": 0.2875,
+      "num_input_tokens_seen": 92978328,
+      "step": 538
+    },
+    {
+      "epoch": 0.20605132133282267,
+      "loss": 0.25674402713775635,
+      "loss_ce": 0.01748622953891754,
+      "loss_iou": 1.0039182901382446,
+      "loss_num": 0.2392578125,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 92978328,
+      "step": 538
+    },
+    {
+      "epoch": 0.2064343163538874,
+      "grad_norm": 29.656036286325826,
+      "learning_rate": 5e-06,
+      "loss": 0.3293,
+      "num_input_tokens_seen": 93151264,
+      "step": 539
+    },
+    {
+      "epoch": 0.2064343163538874,
+      "loss": 0.29664346575737,
+      "loss_ce": 0.017102457582950592,
+      "loss_iou": 1.0337765216827393,
+      "loss_num": 0.279296875,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 93151264,
+      "step": 539
+    },
+    {
+      "epoch": 0.20681731137495213,
+      "grad_norm": 35.01041250985919,
+      "learning_rate": 5e-06,
+      "loss": 0.382,
+      "num_input_tokens_seen": 93324280,
+      "step": 540
+    },
+    {
+      "epoch": 0.20681731137495213,
+      "loss": 0.3642435669898987,
+      "loss_ce": 0.017563890665769577,
+      "loss_iou": 1.0405396223068237,
+      "loss_num": 0.34765625,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 93324280,
+      "step": 540
+    },
+    {
+      "epoch": 0.20720030639601686,
+      "grad_norm": 29.514321748938745,
+      "learning_rate": 5e-06,
+      "loss": 0.3654,
+      "num_input_tokens_seen": 93493904,
+      "step": 541
+    },
+    {
+      "epoch": 0.20720030639601686,
+      "loss": 0.4098755121231079,
+      "loss_ce": 0.015893571078777313,
+      "loss_iou": 1.1384707689285278,
+      "loss_num": 0.39453125,
+      "loss_xval": 0.39453125,
+      "num_input_tokens_seen": 93493904,
+      "step": 541
+    },
+    {
+      "epoch": 0.2075833014170816,
+      "grad_norm": 29.661642873200357,
+      "learning_rate": 5e-06,
+      "loss": 0.3565,
+      "num_input_tokens_seen": 93667240,
+      "step": 542
+    },
+    {
+      "epoch": 0.2075833014170816,
+      "loss": 0.3676626980304718,
+      "loss_ce": 0.01780918426811695,
+      "loss_iou": 1.025194764137268,
+      "loss_num": 0.349609375,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 93667240,
+      "step": 542
+    },
+    {
+      "epoch": 0.2079662964381463,
+      "grad_norm": 29.83369255589317,
+      "learning_rate": 5e-06,
+      "loss": 0.2658,
+      "num_input_tokens_seen": 93840192,
+      "step": 543
+    },
+    {
+      "epoch": 0.2079662964381463,
+      "loss": 0.28363192081451416,
+      "loss_ce": 0.016969311982393265,
+      "loss_iou": 1.3747169971466064,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 93840192,
+      "step": 543
+    },
+    {
+      "epoch": 0.20834929145921102,
+      "grad_norm": 35.945204353559014,
+      "learning_rate": 5e-06,
+      "loss": 0.4175,
+      "num_input_tokens_seen": 94013360,
+      "step": 544
+    },
+    {
+      "epoch": 0.20834929145921102,
+      "loss": 0.4123837947845459,
+      "loss_ce": 0.017364269122481346,
+      "loss_iou": 1.116492509841919,
+      "loss_num": 0.39453125,
+      "loss_xval": 0.39453125,
+      "num_input_tokens_seen": 94013360,
+      "step": 544
+    },
+    {
+      "epoch": 0.20873228648027575,
+      "grad_norm": 18.211661675213367,
+      "learning_rate": 5e-06,
+      "loss": 0.2616,
+      "num_input_tokens_seen": 94186696,
+      "step": 545
+    },
+    {
+      "epoch": 0.20873228648027575,
+      "loss": 0.259822815656662,
+      "loss_ce": 0.018062567338347435,
+      "loss_iou": 1.042894721031189,
+      "loss_num": 0.2421875,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 94186696,
+      "step": 545
+    },
+    {
+      "epoch": 0.20911528150134048,
+      "grad_norm": 23.812323088726416,
+      "learning_rate": 5e-06,
+      "loss": 0.3155,
+      "num_input_tokens_seen": 94359896,
+      "step": 546
+    },
+    {
+      "epoch": 0.20911528150134048,
+      "loss": 0.348230242729187,
+      "loss_ce": 0.017907964065670967,
+      "loss_iou": 1.0230144262313843,
+      "loss_num": 0.330078125,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 94359896,
+      "step": 546
+    },
+    {
+      "epoch": 0.2094982765224052,
+      "grad_norm": 32.28595258584155,
+      "learning_rate": 5e-06,
+      "loss": 0.3224,
+      "num_input_tokens_seen": 94532536,
+      "step": 547
+    },
+    {
+      "epoch": 0.2094982765224052,
+      "loss": 0.3262408375740051,
+      "loss_ce": 0.018745729699730873,
+      "loss_iou": 1.0238080024719238,
+      "loss_num": 0.306640625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 94532536,
+      "step": 547
+    },
+    {
+      "epoch": 0.20988127154346994,
+      "grad_norm": 37.37970388133116,
+      "learning_rate": 5e-06,
+      "loss": 0.2765,
+      "num_input_tokens_seen": 94705584,
+      "step": 548
+    },
+    {
+      "epoch": 0.20988127154346994,
+      "loss": 0.24450857937335968,
+      "loss_ce": 0.01819021813571453,
+      "loss_iou": 1.003622055053711,
+      "loss_num": 0.2265625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 94705584,
+      "step": 548
+    },
+    {
+      "epoch": 0.21026426656453467,
+      "grad_norm": 21.037453351974072,
+      "learning_rate": 5e-06,
+      "loss": 0.3306,
+      "num_input_tokens_seen": 94878360,
+      "step": 549
+    },
+    {
+      "epoch": 0.21026426656453467,
+      "loss": 0.3464343845844269,
+      "loss_ce": 0.01647833362221718,
+      "loss_iou": 2.519430160522461,
+      "loss_num": 0.330078125,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 94878360,
+      "step": 549
+    },
+    {
+      "epoch": 0.21064726158559938,
+      "grad_norm": 28.012647090995817,
+      "learning_rate": 5e-06,
+      "loss": 0.4348,
+      "num_input_tokens_seen": 95051360,
+      "step": 550
+    },
+    {
+      "epoch": 0.21064726158559938,
+      "loss": 0.44148677587509155,
+      "loss_ce": 0.015095168724656105,
+      "loss_iou": 1.05269455909729,
+      "loss_num": 0.42578125,
+      "loss_xval": 0.42578125,
+      "num_input_tokens_seen": 95051360,
+      "step": 550
+    },
+    {
+      "epoch": 0.2110302566066641,
+      "grad_norm": 30.856241730449078,
+      "learning_rate": 5e-06,
+      "loss": 0.2847,
+      "num_input_tokens_seen": 95224592,
+      "step": 551
+    },
+    {
+      "epoch": 0.2110302566066641,
+      "loss": 0.2634372115135193,
+      "loss_ce": 0.018320009112358093,
+      "loss_iou": 1.0076944828033447,
+      "loss_num": 0.2451171875,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 95224592,
+      "step": 551
+    },
+    {
+      "epoch": 0.21141325162772884,
+      "grad_norm": 48.2047248051334,
+      "learning_rate": 5e-06,
+      "loss": 0.4344,
+      "num_input_tokens_seen": 95398152,
+      "step": 552
+    },
+    {
+      "epoch": 0.21141325162772884,
+      "loss": 0.4549892544746399,
+      "loss_ce": 0.017977558076381683,
+      "loss_iou": 1.051296591758728,
+      "loss_num": 0.4375,
+      "loss_xval": 0.4375,
+      "num_input_tokens_seen": 95398152,
+      "step": 552
+    },
+    {
+      "epoch": 0.21179624664879357,
+      "grad_norm": 31.21265694216636,
+      "learning_rate": 5e-06,
+      "loss": 0.2824,
+      "num_input_tokens_seen": 95570656,
+      "step": 553
+    },
+    {
+      "epoch": 0.21179624664879357,
+      "loss": 0.307553231716156,
+      "loss_ce": 0.017880387604236603,
+      "loss_iou": 1.0418583154678345,
+      "loss_num": 0.2890625,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 95570656,
+      "step": 553
+    },
+    {
+      "epoch": 0.2121792416698583,
+      "grad_norm": 20.617360425417267,
+      "learning_rate": 5e-06,
+      "loss": 0.3892,
+      "num_input_tokens_seen": 95743560,
+      "step": 554
+    },
+    {
+      "epoch": 0.2121792416698583,
+      "loss": 0.4240306615829468,
+      "loss_ce": 0.017658598721027374,
+      "loss_iou": 1.1064558029174805,
+      "loss_num": 0.40625,
+      "loss_xval": 0.40625,
+      "num_input_tokens_seen": 95743560,
+      "step": 554
+    },
+    {
+      "epoch": 0.21256223669092303,
+      "grad_norm": 17.537898251891384,
+      "learning_rate": 5e-06,
+      "loss": 0.2828,
+      "num_input_tokens_seen": 95916504,
+      "step": 555
+    },
+    {
+      "epoch": 0.21256223669092303,
+      "loss": 0.284464567899704,
+      "loss_ce": 0.017130576074123383,
+      "loss_iou": 1.0634559392929077,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 95916504,
+      "step": 555
+    },
+    {
+      "epoch": 0.21294523171198773,
+      "grad_norm": 29.95540871787839,
+      "learning_rate": 5e-06,
+      "loss": 0.3323,
+      "num_input_tokens_seen": 96089144,
+      "step": 556
+    },
+    {
+      "epoch": 0.21294523171198773,
+      "loss": 0.3734918236732483,
+      "loss_ce": 0.0176568403840065,
+      "loss_iou": 1.041034460067749,
+      "loss_num": 0.35546875,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 96089144,
+      "step": 556
+    },
+    {
+      "epoch": 0.21332822673305246,
+      "grad_norm": 26.88730291109121,
+      "learning_rate": 5e-06,
+      "loss": 0.3439,
+      "num_input_tokens_seen": 96262456,
+      "step": 557
+    },
+    {
+      "epoch": 0.21332822673305246,
+      "loss": 0.34446585178375244,
+      "loss_ce": 0.016462940722703934,
+      "loss_iou": 1.0123193264007568,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 96262456,
+      "step": 557
+    },
+    {
+      "epoch": 0.2137112217541172,
+      "grad_norm": 20.921247055091808,
+      "learning_rate": 5e-06,
+      "loss": 0.3568,
+      "num_input_tokens_seen": 96435336,
+      "step": 558
+    },
+    {
+      "epoch": 0.2137112217541172,
+      "loss": 0.3399205207824707,
+      "loss_ce": 0.016312118619680405,
+      "loss_iou": 1.1698265075683594,
+      "loss_num": 0.32421875,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 96435336,
+      "step": 558
+    },
+    {
+      "epoch": 0.21409421677518192,
+      "grad_norm": 35.29154381337728,
+      "learning_rate": 5e-06,
+      "loss": 0.1949,
+      "num_input_tokens_seen": 96604584,
+      "step": 559
+    },
+    {
+      "epoch": 0.21409421677518192,
+      "loss": 0.19075900316238403,
+      "loss_ce": 0.01717502251267433,
+      "loss_iou": 1.0107824802398682,
+      "loss_num": 0.173828125,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 96604584,
+      "step": 559
+    },
+    {
+      "epoch": 0.21447721179624665,
+      "grad_norm": 34.094667343184106,
+      "learning_rate": 5e-06,
+      "loss": 0.3641,
+      "num_input_tokens_seen": 96777680,
+      "step": 560
+    },
+    {
+      "epoch": 0.21447721179624665,
+      "loss": 0.3789745569229126,
+      "loss_ce": 0.017280228435993195,
+      "loss_iou": 1.017249345779419,
+      "loss_num": 0.361328125,
+      "loss_xval": 0.361328125,
+      "num_input_tokens_seen": 96777680,
+      "step": 560
+    },
+    {
+      "epoch": 0.21486020681731138,
+      "grad_norm": 20.445151200248425,
+      "learning_rate": 5e-06,
+      "loss": 0.2588,
+      "num_input_tokens_seen": 96950376,
+      "step": 561
+    },
+    {
+      "epoch": 0.21486020681731138,
+      "loss": 0.2581597566604614,
+      "loss_ce": 0.01816953346133232,
+      "loss_iou": 1.057889699935913,
+      "loss_num": 0.240234375,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 96950376,
+      "step": 561
+    },
+    {
+      "epoch": 0.2152432018383761,
+      "grad_norm": 28.099325370464555,
+      "learning_rate": 5e-06,
+      "loss": 0.3773,
+      "num_input_tokens_seen": 97123608,
+      "step": 562
+    },
+    {
+      "epoch": 0.2152432018383761,
+      "loss": 0.34945181012153625,
+      "loss_ce": 0.018885396420955658,
+      "loss_iou": 1.0347191095352173,
+      "loss_num": 0.330078125,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 97123608,
+      "step": 562
+    },
+    {
+      "epoch": 0.21562619685944082,
+      "grad_norm": 31.315819000716427,
+      "learning_rate": 5e-06,
+      "loss": 0.2773,
+      "num_input_tokens_seen": 97296712,
+      "step": 563
+    },
+    {
+      "epoch": 0.21562619685944082,
+      "loss": 0.24266687035560608,
+      "loss_ce": 0.01708093471825123,
+      "loss_iou": 0.8753404021263123,
+      "loss_num": 0.2255859375,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 97296712,
+      "step": 563
+    },
+    {
+      "epoch": 0.21600919188050555,
+      "grad_norm": 36.60482388882205,
+      "learning_rate": 5e-06,
+      "loss": 0.2586,
+      "num_input_tokens_seen": 97469784,
+      "step": 564
+    },
+    {
+      "epoch": 0.21600919188050555,
+      "loss": 0.2576292157173157,
+      "loss_ce": 0.018493477255105972,
+      "loss_iou": 1.0151938199996948,
+      "loss_num": 0.2392578125,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 97469784,
+      "step": 564
+    },
+    {
+      "epoch": 0.21639218690157028,
+      "grad_norm": 28.37729372949139,
+      "learning_rate": 5e-06,
+      "loss": 0.313,
+      "num_input_tokens_seen": 97642760,
+      "step": 565
+    },
+    {
+      "epoch": 0.21639218690157028,
+      "loss": 0.3322356343269348,
+      "loss_ce": 0.01692803017795086,
+      "loss_iou": 1.0117013454437256,
+      "loss_num": 0.314453125,
+      "loss_xval": 0.314453125,
+      "num_input_tokens_seen": 97642760,
+      "step": 565
+    },
+    {
+      "epoch": 0.216775181922635,
+      "grad_norm": 22.15718872563187,
+      "learning_rate": 5e-06,
+      "loss": 0.3209,
+      "num_input_tokens_seen": 97815760,
+      "step": 566
+    },
+    {
+      "epoch": 0.216775181922635,
+      "loss": 0.32263487577438354,
+      "loss_ce": 0.017337003722786903,
+      "loss_iou": 1.0038520097732544,
+      "loss_num": 0.3046875,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 97815760,
+      "step": 566
+    },
+    {
+      "epoch": 0.21715817694369974,
+      "grad_norm": 36.35316914881609,
+      "learning_rate": 5e-06,
+      "loss": 0.3312,
+      "num_input_tokens_seen": 97988544,
+      "step": 567
+    },
+    {
+      "epoch": 0.21715817694369974,
+      "loss": 0.38007500767707825,
+      "loss_ce": 0.017892412841320038,
+      "loss_iou": 1.1017529964447021,
+      "loss_num": 0.361328125,
+      "loss_xval": 0.361328125,
+      "num_input_tokens_seen": 97988544,
+      "step": 567
+    },
+    {
+      "epoch": 0.21754117196476447,
+      "grad_norm": 25.120149700187525,
+      "learning_rate": 5e-06,
+      "loss": 0.2921,
+      "num_input_tokens_seen": 98157904,
+      "step": 568
+    },
+    {
+      "epoch": 0.21754117196476447,
+      "loss": 0.2887548804283142,
+      "loss_ce": 0.017148425802588463,
+      "loss_iou": 1.0177134275436401,
+      "loss_num": 0.271484375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 98157904,
+      "step": 568
+    },
+    {
+      "epoch": 0.21792416698582917,
+      "grad_norm": 45.31235160902148,
+      "learning_rate": 5e-06,
+      "loss": 0.3036,
+      "num_input_tokens_seen": 98330504,
+      "step": 569
+    },
+    {
+      "epoch": 0.21792416698582917,
+      "loss": 0.33930450677871704,
+      "loss_ce": 0.01752716302871704,
+      "loss_iou": 1.066689133644104,
+      "loss_num": 0.322265625,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 98330504,
+      "step": 569
+    },
+    {
+      "epoch": 0.2183071620068939,
+      "grad_norm": 27.820946939019912,
+      "learning_rate": 5e-06,
+      "loss": 0.4104,
+      "num_input_tokens_seen": 98503352,
+      "step": 570
+    },
+    {
+      "epoch": 0.2183071620068939,
+      "loss": 0.4757930636405945,
+      "loss_ce": 0.017541078850626945,
+      "loss_iou": 1.1921957731246948,
+      "loss_num": 0.458984375,
+      "loss_xval": 0.458984375,
+      "num_input_tokens_seen": 98503352,
+      "step": 570
+    },
+    {
+      "epoch": 0.21869015702795863,
+      "grad_norm": 26.298287161026817,
+      "learning_rate": 5e-06,
+      "loss": 0.2843,
+      "num_input_tokens_seen": 98676928,
+      "step": 571
+    },
+    {
+      "epoch": 0.21869015702795863,
+      "loss": 0.2664155960083008,
+      "loss_ce": 0.017514213919639587,
+      "loss_iou": 1.0240974426269531,
+      "loss_num": 0.2490234375,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 98676928,
+      "step": 571
+    },
+    {
+      "epoch": 0.21907315204902336,
+      "grad_norm": 22.399804899954194,
+      "learning_rate": 5e-06,
+      "loss": 0.4406,
+      "num_input_tokens_seen": 98849736,
+      "step": 572
+    },
+    {
+      "epoch": 0.21907315204902336,
+      "loss": 0.48206478357315063,
+      "loss_ce": 0.01673276349902153,
+      "loss_iou": 1.0492788553237915,
+      "loss_num": 0.46484375,
+      "loss_xval": 0.46484375,
+      "num_input_tokens_seen": 98849736,
+      "step": 572
+    },
+    {
+      "epoch": 0.2194561470700881,
+      "grad_norm": 14.489176092642333,
+      "learning_rate": 5e-06,
+      "loss": 0.2433,
+      "num_input_tokens_seen": 99022592,
+      "step": 573
+    },
+    {
+      "epoch": 0.2194561470700881,
+      "loss": 0.2183997631072998,
+      "loss_ce": 0.01747203804552555,
+      "loss_iou": 1.0570449829101562,
+      "loss_num": 0.201171875,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 99022592,
+      "step": 573
+    },
+    {
+      "epoch": 0.21983914209115282,
+      "grad_norm": 26.137905967601572,
+      "learning_rate": 5e-06,
+      "loss": 0.309,
+      "num_input_tokens_seen": 99195680,
+      "step": 574
+    },
+    {
+      "epoch": 0.21983914209115282,
+      "loss": 0.37402984499931335,
+      "loss_ce": 0.017950735986232758,
+      "loss_iou": 1.2666800022125244,
+      "loss_num": 0.35546875,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 99195680,
+      "step": 574
+    },
+    {
+      "epoch": 0.22022213711221755,
+      "grad_norm": 25.845565662558503,
+      "learning_rate": 5e-06,
+      "loss": 0.3863,
+      "num_input_tokens_seen": 99368472,
+      "step": 575
+    },
+    {
+      "epoch": 0.22022213711221755,
+      "loss": 0.35913804173469543,
+      "loss_ce": 0.017463242635130882,
+      "loss_iou": 1.1200839281082153,
+      "loss_num": 0.341796875,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 99368472,
+      "step": 575
+    },
+    {
+      "epoch": 0.22060513213328226,
+      "grad_norm": 22.702387579466556,
+      "learning_rate": 5e-06,
+      "loss": 0.3197,
+      "num_input_tokens_seen": 99541504,
+      "step": 576
+    },
+    {
+      "epoch": 0.22060513213328226,
+      "loss": 0.3078612983226776,
+      "loss_ce": 0.018310517072677612,
+      "loss_iou": 1.1169304847717285,
+      "loss_num": 0.2890625,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 99541504,
+      "step": 576
+    },
+    {
+      "epoch": 0.22098812715434699,
+      "grad_norm": 31.942708726688444,
+      "learning_rate": 5e-06,
+      "loss": 0.321,
+      "num_input_tokens_seen": 99714288,
+      "step": 577
+    },
+    {
+      "epoch": 0.22098812715434699,
+      "loss": 0.32894328236579895,
+      "loss_ce": 0.018152287229895592,
+      "loss_iou": 1.015682578086853,
+      "loss_num": 0.310546875,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 99714288,
+      "step": 577
+    },
+    {
+      "epoch": 0.22137112217541172,
+      "grad_norm": 42.0077178120743,
+      "learning_rate": 5e-06,
+      "loss": 0.2944,
+      "num_input_tokens_seen": 99887304,
+      "step": 578
+    },
+    {
+      "epoch": 0.22137112217541172,
+      "loss": 0.3271752595901489,
+      "loss_ce": 0.017482876777648926,
+      "loss_iou": 1.07436203956604,
+      "loss_num": 0.310546875,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 99887304,
+      "step": 578
+    },
+    {
+      "epoch": 0.22175411719647645,
+      "grad_norm": 30.29576574211237,
+      "learning_rate": 5e-06,
+      "loss": 0.3389,
+      "num_input_tokens_seen": 100060096,
+      "step": 579
+    },
+    {
+      "epoch": 0.22175411719647645,
+      "loss": 0.3681296110153198,
+      "loss_ce": 0.01668919436633587,
+      "loss_iou": 1.1137621402740479,
+      "loss_num": 0.3515625,
+      "loss_xval": 0.3515625,
+      "num_input_tokens_seen": 100060096,
+      "step": 579
+    },
+    {
+      "epoch": 0.22213711221754118,
+      "grad_norm": 42.32461320236619,
+      "learning_rate": 5e-06,
+      "loss": 0.3718,
+      "num_input_tokens_seen": 100232920,
+      "step": 580
+    },
+    {
+      "epoch": 0.22213711221754118,
+      "loss": 0.33574941754341125,
+      "loss_ce": 0.018244553357362747,
+      "loss_iou": 1.0672136545181274,
+      "loss_num": 0.318359375,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 100232920,
+      "step": 580
+    },
+    {
+      "epoch": 0.2225201072386059,
+      "grad_norm": 32.82938766636816,
+      "learning_rate": 5e-06,
+      "loss": 0.2831,
+      "num_input_tokens_seen": 100405672,
+      "step": 581
+    },
+    {
+      "epoch": 0.2225201072386059,
+      "loss": 0.2923923432826996,
+      "loss_ce": 0.017245858907699585,
+      "loss_iou": 1.0123100280761719,
+      "loss_num": 0.275390625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 100405672,
+      "step": 581
+    },
+    {
+      "epoch": 0.22290310225967064,
+      "grad_norm": 45.72324256936807,
+      "learning_rate": 5e-06,
+      "loss": 0.3289,
+      "num_input_tokens_seen": 100575224,
+      "step": 582
+    },
+    {
+      "epoch": 0.22290310225967064,
+      "loss": 0.33408015966415405,
+      "loss_ce": 0.018040090799331665,
+      "loss_iou": 1.0105905532836914,
+      "loss_num": 0.31640625,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 100575224,
+      "step": 582
+    },
+    {
+      "epoch": 0.22328609728073534,
+      "grad_norm": 27.20456589802419,
+      "learning_rate": 5e-06,
+      "loss": 0.3253,
+      "num_input_tokens_seen": 100747824,
+      "step": 583
+    },
+    {
+      "epoch": 0.22328609728073534,
+      "loss": 0.30941587686538696,
+      "loss_ce": 0.01632506772875786,
+      "loss_iou": 1.0055162906646729,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 100747824,
+      "step": 583
+    },
+    {
+      "epoch": 0.22366909230180007,
+      "grad_norm": 36.605260599158925,
+      "learning_rate": 5e-06,
+      "loss": 0.3869,
+      "num_input_tokens_seen": 100920728,
+      "step": 584
+    },
+    {
+      "epoch": 0.22366909230180007,
+      "loss": 0.34552252292633057,
+      "loss_ce": 0.017763737589120865,
+      "loss_iou": 1.0450878143310547,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 100920728,
+      "step": 584
+    },
+    {
+      "epoch": 0.2240520873228648,
+      "grad_norm": 41.24873530783497,
+      "learning_rate": 5e-06,
+      "loss": 0.3717,
+      "num_input_tokens_seen": 101093472,
+      "step": 585
+    },
+    {
+      "epoch": 0.2240520873228648,
+      "loss": 0.3666747212409973,
+      "loss_ce": 0.016699114814400673,
+      "loss_iou": 1.1095600128173828,
+      "loss_num": 0.349609375,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 101093472,
+      "step": 585
+    },
+    {
+      "epoch": 0.22443508234392953,
+      "grad_norm": 53.24347873920217,
+      "learning_rate": 5e-06,
+      "loss": 0.5541,
+      "num_input_tokens_seen": 101266568,
+      "step": 586
+    },
+    {
+      "epoch": 0.22443508234392953,
+      "loss": 0.5650396347045898,
+      "loss_ce": 0.01718807965517044,
+      "loss_iou": 1.0786261558532715,
+      "loss_num": 0.546875,
+      "loss_xval": 0.546875,
+      "num_input_tokens_seen": 101266568,
+      "step": 586
+    },
+    {
+      "epoch": 0.22481807736499426,
+      "grad_norm": 25.188296835415493,
+      "learning_rate": 5e-06,
+      "loss": 0.4724,
+      "num_input_tokens_seen": 101439928,
+      "step": 587
+    },
+    {
+      "epoch": 0.22481807736499426,
+      "loss": 0.4406231641769409,
+      "loss_ce": 0.01752748340368271,
+      "loss_iou": 1.0750447511672974,
+      "loss_num": 0.423828125,
+      "loss_xval": 0.423828125,
+      "num_input_tokens_seen": 101439928,
+      "step": 587
+    },
+    {
+      "epoch": 0.225201072386059,
+      "grad_norm": 33.085510112399156,
+      "learning_rate": 5e-06,
+      "loss": 0.5682,
+      "num_input_tokens_seen": 101613016,
+      "step": 588
+    },
+    {
+      "epoch": 0.225201072386059,
+      "loss": 0.5826241970062256,
+      "loss_ce": 0.01817106083035469,
+      "loss_iou": 1.0221844911575317,
+      "loss_num": 0.5625,
+      "loss_xval": 0.5625,
+      "num_input_tokens_seen": 101613016,
+      "step": 588
+    },
+    {
+      "epoch": 0.2255840674071237,
+      "grad_norm": 68.14967060371285,
+      "learning_rate": 5e-06,
+      "loss": 0.4348,
+      "num_input_tokens_seen": 101786296,
+      "step": 589
+    },
+    {
+      "epoch": 0.2255840674071237,
+      "loss": 0.4092572331428528,
+      "loss_ce": 0.016923263669013977,
+      "loss_iou": 1.0072649717330933,
+      "loss_num": 0.392578125,
+      "loss_xval": 0.392578125,
+      "num_input_tokens_seen": 101786296,
+      "step": 589
+    },
+    {
+      "epoch": 0.22596706242818843,
+      "grad_norm": 58.539577797948795,
+      "learning_rate": 5e-06,
+      "loss": 0.9328,
+      "num_input_tokens_seen": 101958720,
+      "step": 590
+    },
+    {
+      "epoch": 0.22596706242818843,
+      "loss": 0.9331662654876709,
+      "loss_ce": 0.01715068519115448,
+      "loss_iou": 1.5227229595184326,
+      "loss_num": 0.9140625,
+      "loss_xval": 0.9140625,
+      "num_input_tokens_seen": 101958720,
+      "step": 590
+    },
+    {
+      "epoch": 0.22635005744925316,
+      "grad_norm": 29.474031494480535,
+      "learning_rate": 5e-06,
+      "loss": 0.5005,
+      "num_input_tokens_seen": 102131728,
+      "step": 591
+    },
+    {
+      "epoch": 0.22635005744925316,
+      "loss": 0.4739513397216797,
+      "loss_ce": 0.018384922295808792,
+      "loss_iou": 1.0300683975219727,
+      "loss_num": 0.455078125,
+      "loss_xval": 0.455078125,
+      "num_input_tokens_seen": 102131728,
+      "step": 591
+    },
+    {
+      "epoch": 0.2267330524703179,
+      "grad_norm": 38.89440333259645,
+      "learning_rate": 5e-06,
+      "loss": 0.6574,
+      "num_input_tokens_seen": 102304592,
+      "step": 592
+    },
+    {
+      "epoch": 0.2267330524703179,
+      "loss": 0.640261709690094,
+      "loss_ce": 0.0186796635389328,
+      "loss_iou": 1.0453568696975708,
+      "loss_num": 0.62109375,
+      "loss_xval": 0.62109375,
+      "num_input_tokens_seen": 102304592,
+      "step": 592
+    },
+    {
+      "epoch": 0.22711604749138262,
+      "grad_norm": 41.35986263831221,
+      "learning_rate": 5e-06,
+      "loss": 0.4902,
+      "num_input_tokens_seen": 102477688,
+      "step": 593
+    },
+    {
+      "epoch": 0.22711604749138262,
+      "loss": 0.5372838973999023,
+      "loss_ce": 0.01921752840280533,
+      "loss_iou": 1.0303370952606201,
+      "loss_num": 0.51953125,
+      "loss_xval": 0.51953125,
+      "num_input_tokens_seen": 102477688,
+      "step": 593
+    },
+    {
+      "epoch": 0.22749904251244735,
+      "grad_norm": 37.08661584549743,
+      "learning_rate": 5e-06,
+      "loss": 0.5568,
+      "num_input_tokens_seen": 102650608,
+      "step": 594
+    },
+    {
+      "epoch": 0.22749904251244735,
+      "loss": 0.5767726898193359,
+      "loss_ce": 0.0179347712546587,
+      "loss_iou": 1.0251212120056152,
+      "loss_num": 0.55859375,
+      "loss_xval": 0.55859375,
+      "num_input_tokens_seen": 102650608,
+      "step": 594
+    },
+    {
+      "epoch": 0.22788203753351208,
+      "grad_norm": 16.649004377658006,
+      "learning_rate": 5e-06,
+      "loss": 0.3834,
+      "num_input_tokens_seen": 102823688,
+      "step": 595
+    },
+    {
+      "epoch": 0.22788203753351208,
+      "loss": 0.44109010696411133,
+      "loss_ce": 0.017872339114546776,
+      "loss_iou": 1.0665128231048584,
+      "loss_num": 0.423828125,
+      "loss_xval": 0.423828125,
+      "num_input_tokens_seen": 102823688,
+      "step": 595
+    },
+    {
+      "epoch": 0.22826503255457678,
+      "grad_norm": 16.37457310632788,
+      "learning_rate": 5e-06,
+      "loss": 0.2507,
+      "num_input_tokens_seen": 102996344,
+      "step": 596
+    },
+    {
+      "epoch": 0.22826503255457678,
+      "loss": 0.23729053139686584,
+      "loss_ce": 0.018662605434656143,
+      "loss_iou": 1.0226633548736572,
+      "loss_num": 0.21875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 102996344,
+      "step": 596
+    },
+    {
+      "epoch": 0.2286480275756415,
+      "grad_norm": 21.53348319267111,
+      "learning_rate": 5e-06,
+      "loss": 0.4048,
+      "num_input_tokens_seen": 103168896,
+      "step": 597
+    },
+    {
+      "epoch": 0.2286480275756415,
+      "loss": 0.39744117856025696,
+      "loss_ce": 0.018534936010837555,
+      "loss_iou": 1.139091968536377,
+      "loss_num": 0.37890625,
+      "loss_xval": 0.37890625,
+      "num_input_tokens_seen": 103168896,
+      "step": 597
+    },
+    {
+      "epoch": 0.22903102259670624,
+      "grad_norm": 24.527654157341697,
+      "learning_rate": 5e-06,
+      "loss": 0.2627,
+      "num_input_tokens_seen": 103341832,
+      "step": 598
+    },
+    {
+      "epoch": 0.22903102259670624,
+      "loss": 0.31931132078170776,
+      "loss_ce": 0.018407993018627167,
+      "loss_iou": 1.044081687927246,
+      "loss_num": 0.30078125,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 103341832,
+      "step": 598
+    },
+    {
+      "epoch": 0.22941401761777097,
+      "grad_norm": 25.416895779872892,
+      "learning_rate": 5e-06,
+      "loss": 0.3839,
+      "num_input_tokens_seen": 103514584,
+      "step": 599
+    },
+    {
+      "epoch": 0.22941401761777097,
+      "loss": 0.4270893633365631,
+      "loss_ce": 0.018031757324934006,
+      "loss_iou": 1.1228041648864746,
+      "loss_num": 0.408203125,
+      "loss_xval": 0.408203125,
+      "num_input_tokens_seen": 103514584,
+      "step": 599
+    },
+    {
+      "epoch": 0.2297970126388357,
+      "grad_norm": 25.281019377135085,
+      "learning_rate": 5e-06,
+      "loss": 0.2995,
+      "num_input_tokens_seen": 103687432,
+      "step": 600
+    },
+    {
+      "epoch": 0.2297970126388357,
+      "loss": 0.27492815256118774,
+      "loss_ce": 0.01723773032426834,
+      "loss_iou": 1.144974708557129,
+      "loss_num": 0.2578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 103687432,
+      "step": 600
+    },
+    {
+      "epoch": 0.23018000765990043,
+      "grad_norm": 26.228245945257914,
+      "learning_rate": 5e-06,
+      "loss": 0.3585,
+      "num_input_tokens_seen": 103856976,
+      "step": 601
+    },
+    {
+      "epoch": 0.23018000765990043,
+      "loss": 0.3667566180229187,
+      "loss_ce": 0.019344517961144447,
+      "loss_iou": 1.0777132511138916,
+      "loss_num": 0.34765625,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 103856976,
+      "step": 601
+    },
+    {
+      "epoch": 0.23056300268096513,
+      "grad_norm": 41.11046425286124,
+      "learning_rate": 5e-06,
+      "loss": 0.2708,
+      "num_input_tokens_seen": 104029936,
+      "step": 602
+    },
+    {
+      "epoch": 0.23056300268096513,
+      "loss": 0.22370639443397522,
+      "loss_ce": 0.018323099240660667,
+      "loss_iou": 1.0170555114746094,
+      "loss_num": 0.205078125,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 104029936,
+      "step": 602
+    },
+    {
+      "epoch": 0.23094599770202986,
+      "grad_norm": 41.591703374258714,
+      "learning_rate": 5e-06,
+      "loss": 0.3554,
+      "num_input_tokens_seen": 104202896,
+      "step": 603
+    },
+    {
+      "epoch": 0.23094599770202986,
+      "loss": 0.38865795731544495,
+      "loss_ce": 0.017564216628670692,
+      "loss_iou": 1.0423808097839355,
+      "loss_num": 0.37109375,
+      "loss_xval": 0.37109375,
+      "num_input_tokens_seen": 104202896,
+      "step": 603
+    },
+    {
+      "epoch": 0.2313289927230946,
+      "grad_norm": 30.252928274985575,
+      "learning_rate": 5e-06,
+      "loss": 0.2503,
+      "num_input_tokens_seen": 104376000,
+      "step": 604
+    },
+    {
+      "epoch": 0.2313289927230946,
+      "loss": 0.2540701627731323,
+      "loss_ce": 0.018352385610342026,
+      "loss_iou": 1.1092627048492432,
+      "loss_num": 0.2353515625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 104376000,
+      "step": 604
+    },
+    {
+      "epoch": 0.23171198774415933,
+      "grad_norm": 28.653239475474283,
+      "learning_rate": 5e-06,
+      "loss": 0.2782,
+      "num_input_tokens_seen": 104548936,
+      "step": 605
+    },
+    {
+      "epoch": 0.23171198774415933,
+      "loss": 0.284741073846817,
+      "loss_ce": 0.019116077572107315,
+      "loss_iou": 1.1011557579040527,
+      "loss_num": 0.265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 104548936,
+      "step": 605
+    },
+    {
+      "epoch": 0.23209498276522406,
+      "grad_norm": 30.154135033335002,
+      "learning_rate": 5e-06,
+      "loss": 0.2832,
+      "num_input_tokens_seen": 104721440,
+      "step": 606
+    },
+    {
+      "epoch": 0.23209498276522406,
+      "loss": 0.3105369806289673,
+      "loss_ce": 0.017690293490886688,
+      "loss_iou": 1.0124354362487793,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 104721440,
+      "step": 606
+    },
+    {
+      "epoch": 0.2324779777862888,
+      "grad_norm": 27.52309361013761,
+      "learning_rate": 5e-06,
+      "loss": 0.3772,
+      "num_input_tokens_seen": 104894456,
+      "step": 607
+    },
+    {
+      "epoch": 0.2324779777862888,
+      "loss": 0.413565456867218,
+      "loss_ce": 0.018179725855588913,
+      "loss_iou": 1.1098564863204956,
+      "loss_num": 0.39453125,
+      "loss_xval": 0.39453125,
+      "num_input_tokens_seen": 104894456,
+      "step": 607
+    },
+    {
+      "epoch": 0.23286097280735352,
+      "grad_norm": 39.99339674434345,
+      "learning_rate": 5e-06,
+      "loss": 0.3057,
+      "num_input_tokens_seen": 105067744,
+      "step": 608
+    },
+    {
+      "epoch": 0.23286097280735352,
+      "loss": 0.30731672048568726,
+      "loss_ce": 0.018010085448622704,
+      "loss_iou": 1.0802106857299805,
+      "loss_num": 0.2890625,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 105067744,
+      "step": 608
+    },
+    {
+      "epoch": 0.23324396782841822,
+      "grad_norm": 33.07776467900004,
+      "learning_rate": 5e-06,
+      "loss": 0.34,
+      "num_input_tokens_seen": 105240768,
+      "step": 609
+    },
+    {
+      "epoch": 0.23324396782841822,
+      "loss": 0.42627090215682983,
+      "loss_ce": 0.018800202757120132,
+      "loss_iou": 1.2617712020874023,
+      "loss_num": 0.408203125,
+      "loss_xval": 0.408203125,
+      "num_input_tokens_seen": 105240768,
+      "step": 609
+    },
+    {
+      "epoch": 0.23362696284948295,
+      "grad_norm": 27.82244250656852,
+      "learning_rate": 5e-06,
+      "loss": 0.2098,
+      "num_input_tokens_seen": 105414232,
+      "step": 610
+    },
+    {
+      "epoch": 0.23362696284948295,
+      "loss": 0.2339729517698288,
+      "loss_ce": 0.0188850536942482,
+      "loss_iou": 1.027437686920166,
+      "loss_num": 0.21484375,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 105414232,
+      "step": 610
+    },
+    {
+      "epoch": 0.23400995787054768,
+      "grad_norm": 28.753221166597708,
+      "learning_rate": 5e-06,
+      "loss": 0.2854,
+      "num_input_tokens_seen": 105587168,
+      "step": 611
+    },
+    {
+      "epoch": 0.23400995787054768,
+      "loss": 0.31163489818573,
+      "loss_ce": 0.017201315611600876,
+      "loss_iou": 1.035597562789917,
+      "loss_num": 0.294921875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 105587168,
+      "step": 611
+    },
+    {
+      "epoch": 0.2343929528916124,
+      "grad_norm": 30.02519181773402,
+      "learning_rate": 5e-06,
+      "loss": 0.2157,
+      "num_input_tokens_seen": 105760128,
+      "step": 612
+    },
+    {
+      "epoch": 0.2343929528916124,
+      "loss": 0.18196620047092438,
+      "loss_ce": 0.018941298127174377,
+      "loss_iou": 1.0053131580352783,
+      "loss_num": 0.1630859375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 105760128,
+      "step": 612
+    },
+    {
+      "epoch": 0.23477594791267714,
+      "grad_norm": 27.270895360600054,
+      "learning_rate": 5e-06,
+      "loss": 0.2809,
+      "num_input_tokens_seen": 105933064,
+      "step": 613
+    },
+    {
+      "epoch": 0.23477594791267714,
+      "loss": 0.2941964864730835,
+      "loss_ce": 0.018073420971632004,
+      "loss_iou": 1.1391850709915161,
+      "loss_num": 0.275390625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 105933064,
+      "step": 613
+    },
+    {
+      "epoch": 0.23515894293374187,
+      "grad_norm": 32.75790616386231,
+      "learning_rate": 5e-06,
+      "loss": 0.212,
+      "num_input_tokens_seen": 106106080,
+      "step": 614
+    },
+    {
+      "epoch": 0.23515894293374187,
+      "loss": 0.2357349693775177,
+      "loss_ce": 0.018022585660219193,
+      "loss_iou": 1.0676966905593872,
+      "loss_num": 0.2177734375,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 106106080,
+      "step": 614
+    },
+    {
+      "epoch": 0.23554193795480657,
+      "grad_norm": 35.36961784872434,
+      "learning_rate": 5e-06,
+      "loss": 0.2747,
+      "num_input_tokens_seen": 106278944,
+      "step": 615
+    },
+    {
+      "epoch": 0.23554193795480657,
+      "loss": 0.22893795371055603,
+      "loss_ce": 0.018122535198926926,
+      "loss_iou": 1.0038223266601562,
+      "loss_num": 0.2109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 106278944,
+      "step": 615
+    },
+    {
+      "epoch": 0.2359249329758713,
+      "grad_norm": 31.206872306799447,
+      "learning_rate": 5e-06,
+      "loss": 0.2448,
+      "num_input_tokens_seen": 106451968,
+      "step": 616
+    },
+    {
+      "epoch": 0.2359249329758713,
+      "loss": 0.2548496127128601,
+      "loss_ce": 0.017422843724489212,
+      "loss_iou": 1.0527714490890503,
+      "loss_num": 0.2373046875,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 106451968,
+      "step": 616
+    },
+    {
+      "epoch": 0.23630792799693603,
+      "grad_norm": 24.20033365261073,
+      "learning_rate": 5e-06,
+      "loss": 0.2879,
+      "num_input_tokens_seen": 106625288,
+      "step": 617
+    },
+    {
+      "epoch": 0.23630792799693603,
+      "loss": 0.2967357635498047,
+      "loss_ce": 0.018537502735853195,
+      "loss_iou": 1.0783098936080933,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 106625288,
+      "step": 617
+    },
+    {
+      "epoch": 0.23669092301800077,
+      "grad_norm": 23.365551448031216,
+      "learning_rate": 5e-06,
+      "loss": 0.2135,
+      "num_input_tokens_seen": 106797944,
+      "step": 618
+    },
+    {
+      "epoch": 0.23669092301800077,
+      "loss": 0.20067939162254333,
+      "loss_ce": 0.017329787835478783,
+      "loss_iou": 1.0012857913970947,
+      "loss_num": 0.18359375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 106797944,
+      "step": 618
+    },
+    {
+      "epoch": 0.2370739180390655,
+      "grad_norm": 22.931989760098126,
+      "learning_rate": 5e-06,
+      "loss": 0.3273,
+      "num_input_tokens_seen": 106970904,
+      "step": 619
+    },
+    {
+      "epoch": 0.2370739180390655,
+      "loss": 0.3136942386627197,
+      "loss_ce": 0.017795825377106667,
+      "loss_iou": 1.0238518714904785,
+      "loss_num": 0.296875,
+      "loss_xval": 0.296875,
+      "num_input_tokens_seen": 106970904,
+      "step": 619
+    },
+    {
+      "epoch": 0.23745691306013023,
+      "grad_norm": 28.813060915646872,
+      "learning_rate": 5e-06,
+      "loss": 0.2295,
+      "num_input_tokens_seen": 107144096,
+      "step": 620
+    },
+    {
+      "epoch": 0.23745691306013023,
+      "loss": 0.2313339114189148,
+      "loss_ce": 0.0186874121427536,
+      "loss_iou": 1.010434627532959,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 107144096,
+      "step": 620
+    },
+    {
+      "epoch": 0.23783990808119496,
+      "grad_norm": 25.08585719487234,
+      "learning_rate": 5e-06,
+      "loss": 0.3269,
+      "num_input_tokens_seen": 107316976,
+      "step": 621
+    },
+    {
+      "epoch": 0.23783990808119496,
+      "loss": 0.3422039747238159,
+      "loss_ce": 0.01798522099852562,
+      "loss_iou": 1.014849305152893,
+      "loss_num": 0.32421875,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 107316976,
+      "step": 621
+    },
+    {
+      "epoch": 0.23822290310225966,
+      "grad_norm": 30.021678511783954,
+      "learning_rate": 5e-06,
+      "loss": 0.2982,
+      "num_input_tokens_seen": 107489816,
+      "step": 622
+    },
+    {
+      "epoch": 0.23822290310225966,
+      "loss": 0.30236726999282837,
+      "loss_ce": 0.016600700095295906,
+      "loss_iou": 1.0681586265563965,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 107489816,
+      "step": 622
+    },
+    {
+      "epoch": 0.2386058981233244,
+      "grad_norm": 36.570672047053655,
+      "learning_rate": 5e-06,
+      "loss": 0.2883,
+      "num_input_tokens_seen": 107663056,
+      "step": 623
+    },
+    {
+      "epoch": 0.2386058981233244,
+      "loss": 0.2441415935754776,
+      "loss_ce": 0.0194101482629776,
+      "loss_iou": 1.016437292098999,
+      "loss_num": 0.224609375,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 107663056,
+      "step": 623
+    },
+    {
+      "epoch": 0.23898889314438912,
+      "grad_norm": 26.877624242395285,
+      "learning_rate": 5e-06,
+      "loss": 0.2423,
+      "num_input_tokens_seen": 107836160,
+      "step": 624
+    },
+    {
+      "epoch": 0.23898889314438912,
+      "loss": 0.21540260314941406,
+      "loss_ce": 0.01789284497499466,
+      "loss_iou": 1.008943796157837,
+      "loss_num": 0.197265625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 107836160,
+      "step": 624
+    },
+    {
+      "epoch": 0.23937188816545385,
+      "grad_norm": 25.400953958313494,
+      "learning_rate": 5e-06,
+      "loss": 0.3771,
+      "num_input_tokens_seen": 108009496,
+      "step": 625
+    },
+    {
+      "epoch": 0.23937188816545385,
+      "loss": 0.42162907123565674,
+      "loss_ce": 0.019529465585947037,
+      "loss_iou": 1.0877376794815063,
+      "loss_num": 0.40234375,
+      "loss_xval": 0.40234375,
+      "num_input_tokens_seen": 108009496,
+      "step": 625
+    },
+    {
+      "epoch": 0.23975488318651858,
+      "grad_norm": 19.350539524761597,
+      "learning_rate": 5e-06,
+      "loss": 0.2422,
+      "num_input_tokens_seen": 108182224,
+      "step": 626
+    },
+    {
+      "epoch": 0.23975488318651858,
+      "loss": 0.19275686144828796,
+      "loss_ce": 0.018196308985352516,
+      "loss_iou": 1.022761583328247,
+      "loss_num": 0.1748046875,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 108182224,
+      "step": 626
+    },
+    {
+      "epoch": 0.2401378782075833,
+      "grad_norm": 40.621139366261346,
+      "learning_rate": 5e-06,
+      "loss": 0.3272,
+      "num_input_tokens_seen": 108355280,
+      "step": 627
+    },
+    {
+      "epoch": 0.2401378782075833,
+      "loss": 0.38383835554122925,
+      "loss_ce": 0.01909228041768074,
+      "loss_iou": 1.0367767810821533,
+      "loss_num": 0.365234375,
+      "loss_xval": 0.365234375,
+      "num_input_tokens_seen": 108355280,
+      "step": 627
+    },
+    {
+      "epoch": 0.24052087322864804,
+      "grad_norm": 32.69516719464062,
+      "learning_rate": 5e-06,
+      "loss": 0.3202,
+      "num_input_tokens_seen": 108524448,
+      "step": 628
+    },
+    {
+      "epoch": 0.24052087322864804,
+      "loss": 0.3452039957046509,
+      "loss_ce": 0.017811425030231476,
+      "loss_iou": 1.049802303314209,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 108524448,
+      "step": 628
+    },
+    {
+      "epoch": 0.24090386824971274,
+      "grad_norm": 21.751452752988556,
+      "learning_rate": 5e-06,
+      "loss": 0.2627,
+      "num_input_tokens_seen": 108697320,
+      "step": 629
+    },
+    {
+      "epoch": 0.24090386824971274,
+      "loss": 0.2546536922454834,
+      "loss_ce": 0.019057990983128548,
+      "loss_iou": 1.059018850326538,
+      "loss_num": 0.2353515625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 108697320,
+      "step": 629
+    },
+    {
+      "epoch": 0.24128686327077747,
+      "grad_norm": 29.944794638579904,
+      "learning_rate": 5e-06,
+      "loss": 0.3938,
+      "num_input_tokens_seen": 108869976,
+      "step": 630
+    },
+    {
+      "epoch": 0.24128686327077747,
+      "loss": 0.4710794985294342,
+      "loss_ce": 0.018076570704579353,
+      "loss_iou": 1.235339641571045,
+      "loss_num": 0.453125,
+      "loss_xval": 0.453125,
+      "num_input_tokens_seen": 108869976,
+      "step": 630
+    },
+    {
+      "epoch": 0.2416698582918422,
+      "grad_norm": 32.84861963867553,
+      "learning_rate": 5e-06,
+      "loss": 0.2684,
+      "num_input_tokens_seen": 109043056,
+      "step": 631
+    },
+    {
+      "epoch": 0.2416698582918422,
+      "loss": 0.2961300015449524,
+      "loss_ce": 0.020251117646694183,
+      "loss_iou": 1.0154446363449097,
+      "loss_num": 0.275390625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 109043056,
+      "step": 631
+    },
+    {
+      "epoch": 0.24205285331290693,
+      "grad_norm": 37.01900979423141,
+      "learning_rate": 5e-06,
+      "loss": 0.3095,
+      "num_input_tokens_seen": 109215736,
+      "step": 632
+    },
+    {
+      "epoch": 0.24205285331290693,
+      "loss": 0.3424670100212097,
+      "loss_ce": 0.01745479181408882,
+      "loss_iou": 1.0393264293670654,
+      "loss_num": 0.32421875,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 109215736,
+      "step": 632
+    },
+    {
+      "epoch": 0.24243584833397167,
+      "grad_norm": 34.48651243305176,
+      "learning_rate": 5e-06,
+      "loss": 0.3381,
+      "num_input_tokens_seen": 109388664,
+      "step": 633
+    },
+    {
+      "epoch": 0.24243584833397167,
+      "loss": 0.32500481605529785,
+      "loss_ce": 0.018120069056749344,
+      "loss_iou": 1.0546152591705322,
+      "loss_num": 0.306640625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 109388664,
+      "step": 633
+    },
+    {
+      "epoch": 0.2428188433550364,
+      "grad_norm": 41.596791970752896,
+      "learning_rate": 5e-06,
+      "loss": 0.2942,
+      "num_input_tokens_seen": 109561584,
+      "step": 634
+    },
+    {
+      "epoch": 0.2428188433550364,
+      "loss": 0.31500840187072754,
+      "loss_ce": 0.01776721328496933,
+      "loss_iou": 1.0234500169754028,
+      "loss_num": 0.296875,
+      "loss_xval": 0.296875,
+      "num_input_tokens_seen": 109561584,
+      "step": 634
+    },
+    {
+      "epoch": 0.2432018383761011,
+      "grad_norm": 36.33209925241172,
+      "learning_rate": 5e-06,
+      "loss": 0.2624,
+      "num_input_tokens_seen": 109734984,
+      "step": 635
+    },
+    {
+      "epoch": 0.2432018383761011,
+      "loss": 0.27469685673713684,
+      "loss_ce": 0.01847127452492714,
+      "loss_iou": 1.02968168258667,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 109734984,
+      "step": 635
+    },
+    {
+      "epoch": 0.24358483339716583,
+      "grad_norm": 40.307370612902695,
+      "learning_rate": 5e-06,
+      "loss": 0.2789,
+      "num_input_tokens_seen": 109907896,
+      "step": 636
+    },
+    {
+      "epoch": 0.24358483339716583,
+      "loss": 0.294486403465271,
+      "loss_ce": 0.01787509024143219,
+      "loss_iou": 1.0207655429840088,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 109907896,
+      "step": 636
+    },
+    {
+      "epoch": 0.24396782841823056,
+      "grad_norm": 32.377875645670116,
+      "learning_rate": 5e-06,
+      "loss": 0.3075,
+      "num_input_tokens_seen": 110081008,
+      "step": 637
+    },
+    {
+      "epoch": 0.24396782841823056,
+      "loss": 0.3352136015892029,
+      "loss_ce": 0.018196996301412582,
+      "loss_iou": 1.7230377197265625,
+      "loss_num": 0.31640625,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 110081008,
+      "step": 637
+    },
+    {
+      "epoch": 0.2443508234392953,
+      "grad_norm": 31.849941337676345,
+      "learning_rate": 5e-06,
+      "loss": 0.2119,
+      "num_input_tokens_seen": 110254528,
+      "step": 638
+    },
+    {
+      "epoch": 0.2443508234392953,
+      "loss": 0.19193212687969208,
+      "loss_ce": 0.018165042623877525,
+      "loss_iou": 0.9950222969055176,
+      "loss_num": 0.173828125,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 110254528,
+      "step": 638
+    },
+    {
+      "epoch": 0.24473381846036002,
+      "grad_norm": 39.34174564072891,
+      "learning_rate": 5e-06,
+      "loss": 0.3409,
+      "num_input_tokens_seen": 110427536,
+      "step": 639
+    },
+    {
+      "epoch": 0.24473381846036002,
+      "loss": 0.3895414471626282,
+      "loss_ce": 0.018325595185160637,
+      "loss_iou": 1.1946970224380493,
+      "loss_num": 0.37109375,
+      "loss_xval": 0.37109375,
+      "num_input_tokens_seen": 110427536,
+      "step": 639
+    },
+    {
+      "epoch": 0.24511681348142475,
+      "grad_norm": 29.075125285500715,
+      "learning_rate": 5e-06,
+      "loss": 0.2193,
+      "num_input_tokens_seen": 110600384,
+      "step": 640
+    },
+    {
+      "epoch": 0.24511681348142475,
+      "loss": 0.21013250946998596,
+      "loss_ce": 0.018848339095711708,
+      "loss_iou": 1.004418134689331,
+      "loss_num": 0.19140625,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 110600384,
+      "step": 640
+    },
+    {
+      "epoch": 0.24549980850248948,
+      "grad_norm": 32.01044314154312,
+      "learning_rate": 5e-06,
+      "loss": 0.3213,
+      "num_input_tokens_seen": 110773248,
+      "step": 641
+    },
+    {
+      "epoch": 0.24549980850248948,
+      "loss": 0.35537731647491455,
+      "loss_ce": 0.018097028136253357,
+      "loss_iou": 1.1918798685073853,
+      "loss_num": 0.337890625,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 110773248,
+      "step": 641
+    },
+    {
+      "epoch": 0.24588280352355418,
+      "grad_norm": 31.34468948038098,
+      "learning_rate": 5e-06,
+      "loss": 0.1919,
+      "num_input_tokens_seen": 110946488,
+      "step": 642
+    },
+    {
+      "epoch": 0.24588280352355418,
+      "loss": 0.19471794366836548,
+      "loss_ce": 0.01979117840528488,
+      "loss_iou": 1.0147979259490967,
+      "loss_num": 0.1748046875,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 110946488,
+      "step": 642
+    },
+    {
+      "epoch": 0.2462657985446189,
+      "grad_norm": 36.15331658548177,
+      "learning_rate": 5e-06,
+      "loss": 0.2914,
+      "num_input_tokens_seen": 111119800,
+      "step": 643
+    },
+    {
+      "epoch": 0.2462657985446189,
+      "loss": 0.25731784105300903,
+      "loss_ce": 0.01769382134079933,
+      "loss_iou": 1.0024409294128418,
+      "loss_num": 0.2392578125,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 111119800,
+      "step": 643
+    },
+    {
+      "epoch": 0.24664879356568364,
+      "grad_norm": 30.42264469721433,
+      "learning_rate": 5e-06,
+      "loss": 0.3181,
+      "num_input_tokens_seen": 111293128,
+      "step": 644
+    },
+    {
+      "epoch": 0.24664879356568364,
+      "loss": 0.32497987151145935,
+      "loss_ce": 0.01821719855070114,
+      "loss_iou": 1.0627901554107666,
+      "loss_num": 0.306640625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 111293128,
+      "step": 644
+    },
+    {
+      "epoch": 0.24703178858674837,
+      "grad_norm": 26.060083123614064,
+      "learning_rate": 5e-06,
+      "loss": 0.2957,
+      "num_input_tokens_seen": 111465896,
+      "step": 645
+    },
+    {
+      "epoch": 0.24703178858674837,
+      "loss": 0.3195127844810486,
+      "loss_ce": 0.018121197819709778,
+      "loss_iou": 1.0172330141067505,
+      "loss_num": 0.30078125,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 111465896,
+      "step": 645
+    },
+    {
+      "epoch": 0.2474147836078131,
+      "grad_norm": 17.570419821790672,
+      "learning_rate": 5e-06,
+      "loss": 0.2209,
+      "num_input_tokens_seen": 111639200,
+      "step": 646
+    },
+    {
+      "epoch": 0.2474147836078131,
+      "loss": 0.19249209761619568,
+      "loss_ce": 0.018847081810235977,
+      "loss_iou": 1.0150803327560425,
+      "loss_num": 0.173828125,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 111639200,
+      "step": 646
+    },
+    {
+      "epoch": 0.24779777862887784,
+      "grad_norm": 32.037862727923766,
+      "learning_rate": 5e-06,
+      "loss": 0.3627,
+      "num_input_tokens_seen": 111812008,
+      "step": 647
+    },
+    {
+      "epoch": 0.24779777862887784,
+      "loss": 0.30420738458633423,
+      "loss_ce": 0.01892906054854393,
+      "loss_iou": 1.00568687915802,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 111812008,
+      "step": 647
+    },
+    {
+      "epoch": 0.24818077364994254,
+      "grad_norm": 26.462807343965828,
+      "learning_rate": 5e-06,
+      "loss": 0.3076,
+      "num_input_tokens_seen": 111984760,
+      "step": 648
+    },
+    {
+      "epoch": 0.24818077364994254,
+      "loss": 0.3766970932483673,
+      "loss_ce": 0.017200034111738205,
+      "loss_iou": 1.0598634481430054,
+      "loss_num": 0.359375,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 111984760,
+      "step": 648
+    },
+    {
+      "epoch": 0.24856376867100727,
+      "grad_norm": 24.18097412725299,
+      "learning_rate": 5e-06,
+      "loss": 0.2638,
+      "num_input_tokens_seen": 112157824,
+      "step": 649
+    },
+    {
+      "epoch": 0.24856376867100727,
+      "loss": 0.245339497923851,
+      "loss_ce": 0.019265277311205864,
+      "loss_iou": 1.0034204721450806,
+      "loss_num": 0.2265625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 112157824,
+      "step": 649
+    },
+    {
+      "epoch": 0.248946763692072,
+      "grad_norm": 26.809266064922014,
+      "learning_rate": 5e-06,
+      "loss": 0.3174,
+      "num_input_tokens_seen": 112330656,
+      "step": 650
+    },
+    {
+      "epoch": 0.248946763692072,
+      "loss": 0.37160152196884155,
+      "loss_ce": 0.018330035731196404,
+      "loss_iou": 1.0580711364746094,
+      "loss_num": 0.353515625,
+      "loss_xval": 0.353515625,
+      "num_input_tokens_seen": 112330656,
+      "step": 650
+    },
+    {
+      "epoch": 0.24932975871313673,
+      "grad_norm": 43.82264882663346,
+      "learning_rate": 5e-06,
+      "loss": 0.3323,
+      "num_input_tokens_seen": 112503760,
+      "step": 651
+    },
+    {
+      "epoch": 0.24932975871313673,
+      "loss": 0.350696861743927,
+      "loss_ce": 0.018787670880556107,
+      "loss_iou": 1.0264992713928223,
+      "loss_num": 0.33203125,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 112503760,
+      "step": 651
+    },
+    {
+      "epoch": 0.24971275373420146,
+      "grad_norm": 34.4973935234838,
+      "learning_rate": 5e-06,
+      "loss": 0.2624,
+      "num_input_tokens_seen": 112677072,
+      "step": 652
+    },
+    {
+      "epoch": 0.24971275373420146,
+      "loss": 0.2623905539512634,
+      "loss_ce": 0.017639588564634323,
+      "loss_iou": 0.9752378463745117,
+      "loss_num": 0.2451171875,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 112677072,
+      "step": 652
+    },
+    {
+      "epoch": 0.2500957487552662,
+      "grad_norm": 44.05348458776447,
+      "learning_rate": 5e-06,
+      "loss": 0.4013,
+      "num_input_tokens_seen": 112850512,
+      "step": 653
+    },
+    {
+      "epoch": 0.2500957487552662,
+      "loss": 0.4051348567008972,
+      "loss_ce": 0.01902644708752632,
+      "loss_iou": 1.0465726852416992,
+      "loss_num": 0.38671875,
+      "loss_xval": 0.38671875,
+      "num_input_tokens_seen": 112850512,
+      "step": 653
+    },
+    {
+      "epoch": 0.2504787437763309,
+      "grad_norm": 14.702397870918162,
+      "learning_rate": 5e-06,
+      "loss": 0.209,
+      "num_input_tokens_seen": 113023680,
+      "step": 654
+    },
+    {
+      "epoch": 0.2504787437763309,
+      "loss": 0.2019750326871872,
+      "loss_ce": 0.018625423312187195,
+      "loss_iou": 1.009512186050415,
+      "loss_num": 0.18359375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 113023680,
+      "step": 654
+    },
+    {
+      "epoch": 0.25086173879739565,
+      "grad_norm": 12.47110250304027,
+      "learning_rate": 5e-06,
+      "loss": 0.2805,
+      "num_input_tokens_seen": 113196272,
+      "step": 655
+    },
+    {
+      "epoch": 0.25086173879739565,
+      "loss": 0.24651247262954712,
+      "loss_ce": 0.018363066017627716,
+      "loss_iou": 1.0092339515686035,
+      "loss_num": 0.228515625,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 113196272,
+      "step": 655
+    },
+    {
+      "epoch": 0.25124473381846035,
+      "grad_norm": 24.930653773392503,
+      "learning_rate": 5e-06,
+      "loss": 0.294,
+      "num_input_tokens_seen": 113369432,
+      "step": 656
+    },
+    {
+      "epoch": 0.25124473381846035,
+      "loss": 0.3188003897666931,
+      "loss_ce": 0.01789705827832222,
+      "loss_iou": 1.006160020828247,
+      "loss_num": 0.30078125,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 113369432,
+      "step": 656
+    },
+    {
+      "epoch": 0.2516277288395251,
+      "grad_norm": 44.84435832027603,
+      "learning_rate": 5e-06,
+      "loss": 0.2957,
+      "num_input_tokens_seen": 113542424,
+      "step": 657
+    },
+    {
+      "epoch": 0.2516277288395251,
+      "loss": 0.3221891522407532,
+      "loss_ce": 0.019454795867204666,
+      "loss_iou": 1.0263134241104126,
+      "loss_num": 0.302734375,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 113542424,
+      "step": 657
+    },
+    {
+      "epoch": 0.2520107238605898,
+      "grad_norm": 51.24750214164252,
+      "learning_rate": 5e-06,
+      "loss": 0.3477,
+      "num_input_tokens_seen": 113715368,
+      "step": 658
+    },
+    {
+      "epoch": 0.2520107238605898,
+      "loss": 0.39300888776779175,
+      "loss_ce": 0.017886806279420853,
+      "loss_iou": 1.098710536956787,
+      "loss_num": 0.375,
+      "loss_xval": 0.375,
+      "num_input_tokens_seen": 113715368,
+      "step": 658
+    },
+    {
+      "epoch": 0.2523937188816545,
+      "grad_norm": 19.62696854079827,
+      "learning_rate": 5e-06,
+      "loss": 0.303,
+      "num_input_tokens_seen": 113888304,
+      "step": 659
+    },
+    {
+      "epoch": 0.2523937188816545,
+      "loss": 0.2846168577671051,
+      "loss_ce": 0.0204567052423954,
+      "loss_iou": 1.025336503982544,
+      "loss_num": 0.263671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 113888304,
+      "step": 659
+    },
+    {
+      "epoch": 0.2527767139027193,
+      "grad_norm": 25.6402536421642,
+      "learning_rate": 5e-06,
+      "loss": 0.2787,
+      "num_input_tokens_seen": 114061288,
+      "step": 660
+    },
+    {
+      "epoch": 0.2527767139027193,
+      "loss": 0.29709354043006897,
+      "loss_ce": 0.01730838418006897,
+      "loss_iou": 1.0613741874694824,
+      "loss_num": 0.279296875,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 114061288,
+      "step": 660
+    },
+    {
+      "epoch": 0.253159708923784,
+      "grad_norm": 31.309285262466563,
+      "learning_rate": 5e-06,
+      "loss": 0.314,
+      "num_input_tokens_seen": 114234264,
+      "step": 661
+    },
+    {
+      "epoch": 0.253159708923784,
+      "loss": 0.34084004163742065,
+      "loss_ce": 0.018574416637420654,
+      "loss_iou": 1.0365395545959473,
+      "loss_num": 0.322265625,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 114234264,
+      "step": 661
+    },
+    {
+      "epoch": 0.25354270394484874,
+      "grad_norm": 28.455632602900838,
+      "learning_rate": 5e-06,
+      "loss": 0.2957,
+      "num_input_tokens_seen": 114407216,
+      "step": 662
+    },
+    {
+      "epoch": 0.25354270394484874,
+      "loss": 0.3027651011943817,
+      "loss_ce": 0.018097132444381714,
+      "loss_iou": 1.0344314575195312,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 114407216,
+      "step": 662
+    },
+    {
+      "epoch": 0.25392569896591344,
+      "grad_norm": 20.7908543989196,
+      "learning_rate": 5e-06,
+      "loss": 0.3477,
+      "num_input_tokens_seen": 114579936,
+      "step": 663
+    },
+    {
+      "epoch": 0.25392569896591344,
+      "loss": 0.3468393087387085,
+      "loss_ce": 0.018470175564289093,
+      "loss_iou": 1.138266921043396,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 114579936,
+      "step": 663
+    },
+    {
+      "epoch": 0.2543086939869782,
+      "grad_norm": 29.421606992895377,
+      "learning_rate": 5e-06,
+      "loss": 0.3197,
+      "num_input_tokens_seen": 114752832,
+      "step": 664
+    },
+    {
+      "epoch": 0.2543086939869782,
+      "loss": 0.3257991075515747,
+      "loss_ce": 0.019036393612623215,
+      "loss_iou": 1.0928757190704346,
+      "loss_num": 0.306640625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 114752832,
+      "step": 664
+    },
+    {
+      "epoch": 0.2546916890080429,
+      "grad_norm": 36.50931809186374,
+      "learning_rate": 5e-06,
+      "loss": 0.3505,
+      "num_input_tokens_seen": 114922392,
+      "step": 665
+    },
+    {
+      "epoch": 0.2546916890080429,
+      "loss": 0.36523759365081787,
+      "loss_ce": 0.019900668412446976,
+      "loss_iou": 1.0270261764526367,
+      "loss_num": 0.345703125,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 114922392,
+      "step": 665
+    },
+    {
+      "epoch": 0.2550746840291076,
+      "grad_norm": 29.69985389568634,
+      "learning_rate": 5e-06,
+      "loss": 0.3025,
+      "num_input_tokens_seen": 115095088,
+      "step": 666
+    },
+    {
+      "epoch": 0.2550746840291076,
+      "loss": 0.2568197548389435,
+      "loss_ce": 0.016890564933419228,
+      "loss_iou": 1.021826982498169,
+      "loss_num": 0.240234375,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 115095088,
+      "step": 666
+    },
+    {
+      "epoch": 0.25545767905017236,
+      "grad_norm": 34.42256022799013,
+      "learning_rate": 5e-06,
+      "loss": 0.2657,
+      "num_input_tokens_seen": 115268008,
+      "step": 667
+    },
+    {
+      "epoch": 0.25545767905017236,
+      "loss": 0.25539347529411316,
+      "loss_ce": 0.01876017078757286,
+      "loss_iou": 1.009194254875183,
+      "loss_num": 0.236328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 115268008,
+      "step": 667
+    },
+    {
+      "epoch": 0.25584067407123706,
+      "grad_norm": 46.15606482411902,
+      "learning_rate": 5e-06,
+      "loss": 0.3934,
+      "num_input_tokens_seen": 115441136,
+      "step": 668
+    },
+    {
+      "epoch": 0.25584067407123706,
+      "loss": 0.4060421586036682,
+      "loss_ce": 0.01798064261674881,
+      "loss_iou": 1.1865606307983398,
+      "loss_num": 0.388671875,
+      "loss_xval": 0.388671875,
+      "num_input_tokens_seen": 115441136,
+      "step": 668
+    },
+    {
+      "epoch": 0.2562236690923018,
+      "grad_norm": 37.288022716329095,
+      "learning_rate": 5e-06,
+      "loss": 0.2792,
+      "num_input_tokens_seen": 115614112,
+      "step": 669
+    },
+    {
+      "epoch": 0.2562236690923018,
+      "loss": 0.2116086184978485,
+      "loss_ce": 0.017883025109767914,
+      "loss_iou": 1.0063433647155762,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 115614112,
+      "step": 669
+    },
+    {
+      "epoch": 0.2566066641133665,
+      "grad_norm": 37.04308286123183,
+      "learning_rate": 5e-06,
+      "loss": 0.3174,
+      "num_input_tokens_seen": 115787224,
+      "step": 670
+    },
+    {
+      "epoch": 0.2566066641133665,
+      "loss": 0.3867374658584595,
+      "loss_ce": 0.017352700233459473,
+      "loss_iou": 1.0728284120559692,
+      "loss_num": 0.369140625,
+      "loss_xval": 0.369140625,
+      "num_input_tokens_seen": 115787224,
+      "step": 670
+    },
+    {
+      "epoch": 0.2569896591344312,
+      "grad_norm": 36.98854574218791,
+      "learning_rate": 5e-06,
+      "loss": 0.2975,
+      "num_input_tokens_seen": 115960176,
+      "step": 671
+    },
+    {
+      "epoch": 0.2569896591344312,
+      "loss": 0.256584107875824,
+      "loss_ce": 0.01866905391216278,
+      "loss_iou": 1.019071102142334,
+      "loss_num": 0.23828125,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 115960176,
+      "step": 671
+    },
+    {
+      "epoch": 0.257372654155496,
+      "grad_norm": 38.27793314515744,
+      "learning_rate": 5e-06,
+      "loss": 0.2906,
+      "num_input_tokens_seen": 116132992,
+      "step": 672
+    },
+    {
+      "epoch": 0.257372654155496,
+      "loss": 0.26381030678749084,
+      "loss_ce": 0.018326912075281143,
+      "loss_iou": 1.0126886367797852,
+      "loss_num": 0.2451171875,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 116132992,
+      "step": 672
+    },
+    {
+      "epoch": 0.2577556491765607,
+      "grad_norm": 34.093893551307985,
+      "learning_rate": 5e-06,
+      "loss": 0.4332,
+      "num_input_tokens_seen": 116306232,
+      "step": 673
+    },
+    {
+      "epoch": 0.2577556491765607,
+      "loss": 0.34484460949897766,
+      "loss_ce": 0.020503807812929153,
+      "loss_iou": 1.0959699153900146,
+      "loss_num": 0.32421875,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 116306232,
+      "step": 673
+    },
+    {
+      "epoch": 0.25813864419762544,
+      "grad_norm": 28.261585383287276,
+      "learning_rate": 5e-06,
+      "loss": 0.3172,
+      "num_input_tokens_seen": 116479264,
+      "step": 674
+    },
+    {
+      "epoch": 0.25813864419762544,
+      "loss": 0.3202231228351593,
+      "loss_ce": 0.018892554566264153,
+      "loss_iou": 1.0380526781082153,
+      "loss_num": 0.30078125,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 116479264,
+      "step": 674
+    },
+    {
+      "epoch": 0.25852163921869015,
+      "grad_norm": 42.29597171093613,
+      "learning_rate": 5e-06,
+      "loss": 0.2876,
+      "num_input_tokens_seen": 116652608,
+      "step": 675
+    },
+    {
+      "epoch": 0.25852163921869015,
+      "loss": 0.2780504822731018,
+      "loss_ce": 0.019627634435892105,
+      "loss_iou": 1.023488163948059,
+      "loss_num": 0.2578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 116652608,
+      "step": 675
+    },
+    {
+      "epoch": 0.2589046342397549,
+      "grad_norm": 39.768614094798515,
+      "learning_rate": 5e-06,
+      "loss": 0.3281,
+      "num_input_tokens_seen": 116825904,
+      "step": 676
+    },
+    {
+      "epoch": 0.2589046342397549,
+      "loss": 0.3559473156929016,
+      "loss_ce": 0.01878909207880497,
+      "loss_iou": 1.0924371480941772,
+      "loss_num": 0.337890625,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 116825904,
+      "step": 676
+    },
+    {
+      "epoch": 0.2592876292608196,
+      "grad_norm": 30.636294971818533,
+      "learning_rate": 5e-06,
+      "loss": 0.4043,
+      "num_input_tokens_seen": 116998536,
+      "step": 677
+    },
+    {
+      "epoch": 0.2592876292608196,
+      "loss": 0.32248398661613464,
+      "loss_ce": 0.018284771591424942,
+      "loss_iou": 1.0102880001068115,
+      "loss_num": 0.3046875,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 116998536,
+      "step": 677
+    },
+    {
+      "epoch": 0.2596706242818843,
+      "grad_norm": 28.465009096568117,
+      "learning_rate": 5e-06,
+      "loss": 0.2849,
+      "num_input_tokens_seen": 117171440,
+      "step": 678
+    },
+    {
+      "epoch": 0.2596706242818843,
+      "loss": 0.3148580491542816,
+      "loss_ce": 0.016762344166636467,
+      "loss_iou": 1.0402758121490479,
+      "loss_num": 0.298828125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 117171440,
+      "step": 678
+    },
+    {
+      "epoch": 0.26005361930294907,
+      "grad_norm": 39.52164357894705,
+      "learning_rate": 5e-06,
+      "loss": 0.3934,
+      "num_input_tokens_seen": 117344528,
+      "step": 679
+    },
+    {
+      "epoch": 0.26005361930294907,
+      "loss": 0.39193177223205566,
+      "loss_ce": 0.018884912133216858,
+      "loss_iou": 1.0810632705688477,
+      "loss_num": 0.373046875,
+      "loss_xval": 0.373046875,
+      "num_input_tokens_seen": 117344528,
+      "step": 679
+    },
+    {
+      "epoch": 0.26043661432401377,
+      "grad_norm": 36.19760168487612,
+      "learning_rate": 5e-06,
+      "loss": 0.271,
+      "num_input_tokens_seen": 117514016,
+      "step": 680
+    },
+    {
+      "epoch": 0.26043661432401377,
+      "loss": 0.2822980284690857,
+      "loss_ce": 0.018992377445101738,
+      "loss_iou": 1.0114190578460693,
+      "loss_num": 0.263671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 117514016,
+      "step": 680
+    },
+    {
+      "epoch": 0.26081960934507853,
+      "grad_norm": 28.301497706787487,
+      "learning_rate": 5e-06,
+      "loss": 0.38,
+      "num_input_tokens_seen": 117686864,
+      "step": 681
+    },
+    {
+      "epoch": 0.26081960934507853,
+      "loss": 0.3316190242767334,
+      "loss_ce": 0.01765417493879795,
+      "loss_iou": 1.019181251525879,
+      "loss_num": 0.314453125,
+      "loss_xval": 0.314453125,
+      "num_input_tokens_seen": 117686864,
+      "step": 681
+    },
+    {
+      "epoch": 0.26120260436614323,
+      "grad_norm": 29.967158323525172,
+      "learning_rate": 5e-06,
+      "loss": 0.3462,
+      "num_input_tokens_seen": 117859824,
+      "step": 682
+    },
+    {
+      "epoch": 0.26120260436614323,
+      "loss": 0.35776767134666443,
+      "loss_ce": 0.01890048012137413,
+      "loss_iou": 1.0225400924682617,
+      "loss_num": 0.33984375,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 117859824,
+      "step": 682
+    },
+    {
+      "epoch": 0.261585599387208,
+      "grad_norm": 45.81503050852084,
+      "learning_rate": 5e-06,
+      "loss": 0.4081,
+      "num_input_tokens_seen": 118032872,
+      "step": 683
+    },
+    {
+      "epoch": 0.261585599387208,
+      "loss": 0.3803051710128784,
+      "loss_ce": 0.018732909113168716,
+      "loss_iou": 1.0778758525848389,
+      "loss_num": 0.361328125,
+      "loss_xval": 0.361328125,
+      "num_input_tokens_seen": 118032872,
+      "step": 683
+    },
+    {
+      "epoch": 0.2619685944082727,
+      "grad_norm": 41.75583928817496,
+      "learning_rate": 5e-06,
+      "loss": 0.3874,
+      "num_input_tokens_seen": 118205704,
+      "step": 684
+    },
+    {
+      "epoch": 0.2619685944082727,
+      "loss": 0.36764055490493774,
+      "loss_ce": 0.016078053042292595,
+      "loss_iou": 1.0661917924880981,
+      "loss_num": 0.3515625,
+      "loss_xval": 0.3515625,
+      "num_input_tokens_seen": 118205704,
+      "step": 684
+    },
+    {
+      "epoch": 0.2623515894293374,
+      "grad_norm": 34.19024593736843,
+      "learning_rate": 5e-06,
+      "loss": 0.3418,
+      "num_input_tokens_seen": 118378656,
+      "step": 685
+    },
+    {
+      "epoch": 0.2623515894293374,
+      "loss": 0.3572046756744385,
+      "loss_ce": 0.018947839736938477,
+      "loss_iou": 1.0371094942092896,
+      "loss_num": 0.337890625,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 118378656,
+      "step": 685
+    },
+    {
+      "epoch": 0.26273458445040215,
+      "grad_norm": 32.897054384579484,
+      "learning_rate": 5e-06,
+      "loss": 0.4488,
+      "num_input_tokens_seen": 118551760,
+      "step": 686
+    },
+    {
+      "epoch": 0.26273458445040215,
+      "loss": 0.4040653109550476,
+      "loss_ce": 0.018078990280628204,
+      "loss_iou": 1.0203449726104736,
+      "loss_num": 0.38671875,
+      "loss_xval": 0.38671875,
+      "num_input_tokens_seen": 118551760,
+      "step": 686
+    },
+    {
+      "epoch": 0.26311757947146686,
+      "grad_norm": 52.02144318734267,
+      "learning_rate": 5e-06,
+      "loss": 0.4311,
+      "num_input_tokens_seen": 118724664,
+      "step": 687
+    },
+    {
+      "epoch": 0.26311757947146686,
+      "loss": 0.42668282985687256,
+      "loss_ce": 0.018845928832888603,
+      "loss_iou": 1.1407008171081543,
+      "loss_num": 0.408203125,
+      "loss_xval": 0.408203125,
+      "num_input_tokens_seen": 118724664,
+      "step": 687
+    },
+    {
+      "epoch": 0.2635005744925316,
+      "grad_norm": 51.556657707207776,
+      "learning_rate": 5e-06,
+      "loss": 0.4464,
+      "num_input_tokens_seen": 118897632,
+      "step": 688
+    },
+    {
+      "epoch": 0.2635005744925316,
+      "loss": 0.4025242328643799,
+      "loss_ce": 0.019223466515541077,
+      "loss_iou": 1.0114976167678833,
+      "loss_num": 0.3828125,
+      "loss_xval": 0.3828125,
+      "num_input_tokens_seen": 118897632,
+      "step": 688
+    },
+    {
+      "epoch": 0.2638835695135963,
+      "grad_norm": 31.65263708218579,
+      "learning_rate": 5e-06,
+      "loss": 0.4013,
+      "num_input_tokens_seen": 119070440,
+      "step": 689
+    },
+    {
+      "epoch": 0.2638835695135963,
+      "loss": 0.40110763907432556,
+      "loss_ce": 0.01805099844932556,
+      "loss_iou": 1.0903512239456177,
+      "loss_num": 0.3828125,
+      "loss_xval": 0.3828125,
+      "num_input_tokens_seen": 119070440,
+      "step": 689
+    },
+    {
+      "epoch": 0.2642665645346611,
+      "grad_norm": 36.34307834204635,
+      "learning_rate": 5e-06,
+      "loss": 0.2601,
+      "num_input_tokens_seen": 119243032,
+      "step": 690
+    },
+    {
+      "epoch": 0.2642665645346611,
+      "loss": 0.257997989654541,
+      "loss_ce": 0.018251921981573105,
+      "loss_iou": 1.0377609729766846,
+      "loss_num": 0.240234375,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 119243032,
+      "step": 690
+    },
+    {
+      "epoch": 0.2646495595557258,
+      "grad_norm": 47.70778236126298,
+      "learning_rate": 5e-06,
+      "loss": 0.4739,
+      "num_input_tokens_seen": 119416104,
+      "step": 691
+    },
+    {
+      "epoch": 0.2646495595557258,
+      "loss": 0.4721728563308716,
+      "loss_ce": 0.019536111503839493,
+      "loss_iou": 1.0260049104690552,
+      "loss_num": 0.453125,
+      "loss_xval": 0.453125,
+      "num_input_tokens_seen": 119416104,
+      "step": 691
+    },
+    {
+      "epoch": 0.2650325545767905,
+      "grad_norm": 29.64229602834131,
+      "learning_rate": 5e-06,
+      "loss": 0.4062,
+      "num_input_tokens_seen": 119589064,
+      "step": 692
+    },
+    {
+      "epoch": 0.2650325545767905,
+      "loss": 0.3716251254081726,
+      "loss_ce": 0.018292631953954697,
+      "loss_iou": 0.9940216541290283,
+      "loss_num": 0.353515625,
+      "loss_xval": 0.353515625,
+      "num_input_tokens_seen": 119589064,
+      "step": 692
+    },
+    {
+      "epoch": 0.26541554959785524,
+      "grad_norm": 32.23904198008546,
+      "learning_rate": 5e-06,
+      "loss": 0.3232,
+      "num_input_tokens_seen": 119761960,
+      "step": 693
+    },
+    {
+      "epoch": 0.26541554959785524,
+      "loss": 0.2772480249404907,
+      "loss_ce": 0.018092773854732513,
+      "loss_iou": 1.0166699886322021,
+      "loss_num": 0.259765625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 119761960,
+      "step": 693
+    },
+    {
+      "epoch": 0.26579854461891994,
+      "grad_norm": 31.76869273383469,
+      "learning_rate": 5e-06,
+      "loss": 0.258,
+      "num_input_tokens_seen": 119935400,
+      "step": 694
+    },
+    {
+      "epoch": 0.26579854461891994,
+      "loss": 0.22908979654312134,
+      "loss_ce": 0.017175735905766487,
+      "loss_iou": 1.0116028785705566,
+      "loss_num": 0.2119140625,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 119935400,
+      "step": 694
+    },
+    {
+      "epoch": 0.2661815396399847,
+      "grad_norm": 26.302439294025707,
+      "learning_rate": 5e-06,
+      "loss": 0.2365,
+      "num_input_tokens_seen": 120108416,
+      "step": 695
+    },
+    {
+      "epoch": 0.2661815396399847,
+      "loss": 0.27466124296188354,
+      "loss_ce": 0.018435679376125336,
+      "loss_iou": 1.0388991832733154,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 120108416,
+      "step": 695
+    },
+    {
+      "epoch": 0.2665645346610494,
+      "grad_norm": 30.018702947426444,
+      "learning_rate": 5e-06,
+      "loss": 0.2747,
+      "num_input_tokens_seen": 120281088,
+      "step": 696
+    },
+    {
+      "epoch": 0.2665645346610494,
+      "loss": 0.2880048155784607,
+      "loss_ce": 0.018351484090089798,
+      "loss_iou": 1.003197431564331,
+      "loss_num": 0.26953125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 120281088,
+      "step": 696
+    },
+    {
+      "epoch": 0.26694752968211416,
+      "grad_norm": 51.20962525091946,
+      "learning_rate": 5e-06,
+      "loss": 0.3086,
+      "num_input_tokens_seen": 120454352,
+      "step": 697
+    },
+    {
+      "epoch": 0.26694752968211416,
+      "loss": 0.26430341601371765,
+      "loss_ce": 0.02138349786400795,
+      "loss_iou": 1.0231826305389404,
+      "loss_num": 0.2431640625,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 120454352,
+      "step": 697
+    },
+    {
+      "epoch": 0.26733052470317886,
+      "grad_norm": 32.316880168240395,
+      "learning_rate": 5e-06,
+      "loss": 0.2535,
+      "num_input_tokens_seen": 120627288,
+      "step": 698
+    },
+    {
+      "epoch": 0.26733052470317886,
+      "loss": 0.2823450267314911,
+      "loss_ce": 0.017818670719861984,
+      "loss_iou": 1.0081267356872559,
+      "loss_num": 0.263671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 120627288,
+      "step": 698
+    },
+    {
+      "epoch": 0.26771351972424356,
+      "grad_norm": 28.501295156169963,
+      "learning_rate": 5e-06,
+      "loss": 0.3048,
+      "num_input_tokens_seen": 120800696,
+      "step": 699
+    },
+    {
+      "epoch": 0.26771351972424356,
+      "loss": 0.2860148549079895,
+      "loss_ce": 0.019779490306973457,
+      "loss_iou": 1.0732073783874512,
+      "loss_num": 0.265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 120800696,
+      "step": 699
+    },
+    {
+      "epoch": 0.2680965147453083,
+      "grad_norm": 38.19213868256434,
+      "learning_rate": 5e-06,
+      "loss": 0.3346,
+      "num_input_tokens_seen": 120973608,
+      "step": 700
+    },
+    {
+      "epoch": 0.2680965147453083,
+      "loss": 0.33420705795288086,
+      "loss_ce": 0.019143562763929367,
+      "loss_iou": 1.0907044410705566,
+      "loss_num": 0.314453125,
+      "loss_xval": 0.314453125,
+      "num_input_tokens_seen": 120973608,
+      "step": 700
+    },
+    {
+      "epoch": 0.268479509766373,
+      "grad_norm": 25.42889342675567,
+      "learning_rate": 5e-06,
+      "loss": 0.4255,
+      "num_input_tokens_seen": 121146512,
+      "step": 701
+    },
+    {
+      "epoch": 0.268479509766373,
+      "loss": 0.4972996115684509,
+      "loss_ce": 0.017807435244321823,
+      "loss_iou": 1.0941216945648193,
+      "loss_num": 0.48046875,
+      "loss_xval": 0.48046875,
+      "num_input_tokens_seen": 121146512,
+      "step": 701
+    },
+    {
+      "epoch": 0.2688625047874378,
+      "grad_norm": 19.2893195257886,
+      "learning_rate": 5e-06,
+      "loss": 0.2672,
+      "num_input_tokens_seen": 121319352,
+      "step": 702
+    },
+    {
+      "epoch": 0.2688625047874378,
+      "loss": 0.22158023715019226,
+      "loss_ce": 0.019187651574611664,
+      "loss_iou": 1.0422577857971191,
+      "loss_num": 0.2021484375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 121319352,
+      "step": 702
+    },
+    {
+      "epoch": 0.2692454998085025,
+      "grad_norm": 25.817232074067945,
+      "learning_rate": 5e-06,
+      "loss": 0.3147,
+      "num_input_tokens_seen": 121492040,
+      "step": 703
+    },
+    {
+      "epoch": 0.2692454998085025,
+      "loss": 0.33373576402664185,
+      "loss_ce": 0.01842815801501274,
+      "loss_iou": 1.0259368419647217,
+      "loss_num": 0.314453125,
+      "loss_xval": 0.314453125,
+      "num_input_tokens_seen": 121492040,
+      "step": 703
+    },
+    {
+      "epoch": 0.2696284948295672,
+      "grad_norm": 24.595472462666415,
+      "learning_rate": 5e-06,
+      "loss": 0.2248,
+      "num_input_tokens_seen": 121664760,
+      "step": 704
+    },
+    {
+      "epoch": 0.2696284948295672,
+      "loss": 0.2683330774307251,
+      "loss_ce": 0.0193096324801445,
+      "loss_iou": 1.021988868713379,
+      "loss_num": 0.2490234375,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 121664760,
+      "step": 704
+    },
+    {
+      "epoch": 0.27001148985063195,
+      "grad_norm": 31.826092860934732,
+      "learning_rate": 5e-06,
+      "loss": 0.2983,
+      "num_input_tokens_seen": 121837976,
+      "step": 705
+    },
+    {
+      "epoch": 0.27001148985063195,
+      "loss": 0.3716413080692291,
+      "loss_ce": 0.019468463957309723,
+      "loss_iou": 1.185765027999878,
+      "loss_num": 0.3515625,
+      "loss_xval": 0.3515625,
+      "num_input_tokens_seen": 121837976,
+      "step": 705
+    },
+    {
+      "epoch": 0.27039448487169665,
+      "grad_norm": 27.515951260871756,
+      "learning_rate": 5e-06,
+      "loss": 0.2401,
+      "num_input_tokens_seen": 122010944,
+      "step": 706
+    },
+    {
+      "epoch": 0.27039448487169665,
+      "loss": 0.21705345809459686,
+      "loss_ce": 0.018994377925992012,
+      "loss_iou": 1.081105351448059,
+      "loss_num": 0.1982421875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 122010944,
+      "step": 706
+    },
+    {
+      "epoch": 0.2707774798927614,
+      "grad_norm": 20.71376535604538,
+      "learning_rate": 5e-06,
+      "loss": 0.3434,
+      "num_input_tokens_seen": 122184408,
+      "step": 707
+    },
+    {
+      "epoch": 0.2707774798927614,
+      "loss": 0.3039228916168213,
+      "loss_ce": 0.018766645342111588,
+      "loss_iou": 1.0301353931427002,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 122184408,
+      "step": 707
+    },
+    {
+      "epoch": 0.2711604749138261,
+      "grad_norm": 25.93834274200068,
+      "learning_rate": 5e-06,
+      "loss": 0.2279,
+      "num_input_tokens_seen": 122357696,
+      "step": 708
+    },
+    {
+      "epoch": 0.2711604749138261,
+      "loss": 0.18875019252300262,
+      "loss_ce": 0.01858418434858322,
+      "loss_iou": 1.0017107725143433,
+      "loss_num": 0.169921875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 122357696,
+      "step": 708
+    },
+    {
+      "epoch": 0.27154346993489087,
+      "grad_norm": 29.21920944343291,
+      "learning_rate": 5e-06,
+      "loss": 0.2946,
+      "num_input_tokens_seen": 122531104,
+      "step": 709
+    },
+    {
+      "epoch": 0.27154346993489087,
+      "loss": 0.25699907541275024,
+      "loss_ce": 0.019938526675105095,
+      "loss_iou": 1.0130640268325806,
+      "loss_num": 0.2373046875,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 122531104,
+      "step": 709
+    },
+    {
+      "epoch": 0.27192646495595557,
+      "grad_norm": 27.400755121419984,
+      "learning_rate": 5e-06,
+      "loss": 0.2769,
+      "num_input_tokens_seen": 122703288,
+      "step": 710
+    },
+    {
+      "epoch": 0.27192646495595557,
+      "loss": 0.30395543575286865,
+      "loss_ce": 0.018310904502868652,
+      "loss_iou": 1.0546436309814453,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 122703288,
+      "step": 710
+    },
+    {
+      "epoch": 0.2723094599770203,
+      "grad_norm": 28.7883081324073,
+      "learning_rate": 5e-06,
+      "loss": 0.2618,
+      "num_input_tokens_seen": 122876424,
+      "step": 711
+    },
+    {
+      "epoch": 0.2723094599770203,
+      "loss": 0.20747819542884827,
+      "loss_ce": 0.02022233046591282,
+      "loss_iou": 1.0015647411346436,
+      "loss_num": 0.1875,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 122876424,
+      "step": 711
+    },
+    {
+      "epoch": 0.27269245499808503,
+      "grad_norm": 33.229815548894656,
+      "learning_rate": 5e-06,
+      "loss": 0.2744,
+      "num_input_tokens_seen": 123049448,
+      "step": 712
+    },
+    {
+      "epoch": 0.27269245499808503,
+      "loss": 0.23115840554237366,
+      "loss_ce": 0.019732624292373657,
+      "loss_iou": 1.0099749565124512,
+      "loss_num": 0.2109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 123049448,
+      "step": 712
+    },
+    {
+      "epoch": 0.27307545001914973,
+      "grad_norm": 28.783978451444586,
+      "learning_rate": 5e-06,
+      "loss": 0.2834,
+      "num_input_tokens_seen": 123221864,
+      "step": 713
+    },
+    {
+      "epoch": 0.27307545001914973,
+      "loss": 0.30808210372924805,
+      "loss_ce": 0.0177988912910223,
+      "loss_iou": 0.89212566614151,
+      "loss_num": 0.291015625,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 123221864,
+      "step": 713
+    },
+    {
+      "epoch": 0.2734584450402145,
+      "grad_norm": 20.492719049018284,
+      "learning_rate": 5e-06,
+      "loss": 0.288,
+      "num_input_tokens_seen": 123395040,
+      "step": 714
+    },
+    {
+      "epoch": 0.2734584450402145,
+      "loss": 0.26109039783477783,
+      "loss_ce": 0.01865876093506813,
+      "loss_iou": 1.037696123123169,
+      "loss_num": 0.2421875,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 123395040,
+      "step": 714
+    },
+    {
+      "epoch": 0.2738414400612792,
+      "grad_norm": 27.808046903025076,
+      "learning_rate": 5e-06,
+      "loss": 0.2707,
+      "num_input_tokens_seen": 123568200,
+      "step": 715
+    },
+    {
+      "epoch": 0.2738414400612792,
+      "loss": 0.24071624875068665,
+      "loss_ce": 0.018914487212896347,
+      "loss_iou": 1.0531505346298218,
+      "loss_num": 0.2216796875,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 123568200,
+      "step": 715
+    },
+    {
+      "epoch": 0.27422443508234395,
+      "grad_norm": 27.648231101059913,
+      "learning_rate": 5e-06,
+      "loss": 0.3143,
+      "num_input_tokens_seen": 123740960,
+      "step": 716
+    },
+    {
+      "epoch": 0.27422443508234395,
+      "loss": 0.3292272090911865,
+      "loss_ce": 0.017703771591186523,
+      "loss_iou": 1.1081669330596924,
+      "loss_num": 0.3125,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 123740960,
+      "step": 716
+    },
+    {
+      "epoch": 0.27460743010340866,
+      "grad_norm": 44.2006124857482,
+      "learning_rate": 5e-06,
+      "loss": 0.3903,
+      "num_input_tokens_seen": 123913648,
+      "step": 717
+    },
+    {
+      "epoch": 0.27460743010340866,
+      "loss": 0.3485841155052185,
+      "loss_ce": 0.019848760217428207,
+      "loss_iou": 1.073115348815918,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 123913648,
+      "step": 717
+    },
+    {
+      "epoch": 0.27499042512447336,
+      "grad_norm": 36.791535567538524,
+      "learning_rate": 5e-06,
+      "loss": 0.2921,
+      "num_input_tokens_seen": 124086576,
+      "step": 718
+    },
+    {
+      "epoch": 0.27499042512447336,
+      "loss": 0.3415762782096863,
+      "loss_ce": 0.020287182182073593,
+      "loss_iou": 1.0090223550796509,
+      "loss_num": 0.3203125,
+      "loss_xval": 0.3203125,
+      "num_input_tokens_seen": 124086576,
+      "step": 718
+    },
+    {
+      "epoch": 0.2753734201455381,
+      "grad_norm": 25.83204726799368,
+      "learning_rate": 5e-06,
+      "loss": 0.3082,
+      "num_input_tokens_seen": 124259544,
+      "step": 719
+    },
+    {
+      "epoch": 0.2753734201455381,
+      "loss": 0.316400945186615,
+      "loss_ce": 0.018854565918445587,
+      "loss_iou": 1.0233391523361206,
+      "loss_num": 0.296875,
+      "loss_xval": 0.296875,
+      "num_input_tokens_seen": 124259544,
+      "step": 719
+    },
+    {
+      "epoch": 0.2757564151666028,
+      "grad_norm": 39.520129299910394,
+      "learning_rate": 5e-06,
+      "loss": 0.2282,
+      "num_input_tokens_seen": 124432128,
+      "step": 720
+    },
+    {
+      "epoch": 0.2757564151666028,
+      "loss": 0.19742247462272644,
+      "loss_ce": 0.018772579729557037,
+      "loss_iou": 1.0239720344543457,
+      "loss_num": 0.1787109375,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 124432128,
+      "step": 720
+    },
+    {
+      "epoch": 0.2761394101876676,
+      "grad_norm": 25.857038261453084,
+      "learning_rate": 5e-06,
+      "loss": 0.2858,
+      "num_input_tokens_seen": 124604608,
+      "step": 721
+    },
+    {
+      "epoch": 0.2761394101876676,
+      "loss": 0.3113729953765869,
+      "loss_ce": 0.01718352548778057,
+      "loss_iou": 1.0240404605865479,
+      "loss_num": 0.294921875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 124604608,
+      "step": 721
+    },
+    {
+      "epoch": 0.2765224052087323,
+      "grad_norm": 25.072404068900614,
+      "learning_rate": 5e-06,
+      "loss": 0.2671,
+      "num_input_tokens_seen": 124777552,
+      "step": 722
+    },
+    {
+      "epoch": 0.2765224052087323,
+      "loss": 0.29605644941329956,
+      "loss_ce": 0.01920097880065441,
+      "loss_iou": 1.0405561923980713,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 124777552,
+      "step": 722
+    },
+    {
+      "epoch": 0.27690540022979704,
+      "grad_norm": 34.50832431840458,
+      "learning_rate": 5e-06,
+      "loss": 0.3078,
+      "num_input_tokens_seen": 124950512,
+      "step": 723
+    },
+    {
+      "epoch": 0.27690540022979704,
+      "loss": 0.32187792658805847,
+      "loss_ce": 0.017922839149832726,
+      "loss_iou": 1.0352263450622559,
+      "loss_num": 0.3046875,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 124950512,
+      "step": 723
+    },
+    {
+      "epoch": 0.27728839525086174,
+      "grad_norm": 25.935439943930284,
+      "learning_rate": 5e-06,
+      "loss": 0.3003,
+      "num_input_tokens_seen": 125123896,
+      "step": 724
+    },
+    {
+      "epoch": 0.27728839525086174,
+      "loss": 0.3337087631225586,
+      "loss_ce": 0.01925564929842949,
+      "loss_iou": 1.0178894996643066,
+      "loss_num": 0.314453125,
+      "loss_xval": 0.314453125,
+      "num_input_tokens_seen": 125123896,
+      "step": 724
+    },
+    {
+      "epoch": 0.27767139027192644,
+      "grad_norm": 24.152246508505147,
+      "learning_rate": 5e-06,
+      "loss": 0.3073,
+      "num_input_tokens_seen": 125296896,
+      "step": 725
+    },
+    {
+      "epoch": 0.27767139027192644,
+      "loss": 0.2829008102416992,
+      "loss_ce": 0.018984809517860413,
+      "loss_iou": 1.0293943881988525,
+      "loss_num": 0.263671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 125296896,
+      "step": 725
+    },
+    {
+      "epoch": 0.2780543852929912,
+      "grad_norm": 18.329942555997228,
+      "learning_rate": 5e-06,
+      "loss": 0.2388,
+      "num_input_tokens_seen": 125469584,
+      "step": 726
+    },
+    {
+      "epoch": 0.2780543852929912,
+      "loss": 0.23900489509105682,
+      "loss_ce": 0.019156260415911674,
+      "loss_iou": 1.197575330734253,
+      "loss_num": 0.2197265625,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 125469584,
+      "step": 726
+    },
+    {
+      "epoch": 0.2784373803140559,
+      "grad_norm": 21.43154241942784,
+      "learning_rate": 5e-06,
+      "loss": 0.3144,
+      "num_input_tokens_seen": 125642448,
+      "step": 727
+    },
+    {
+      "epoch": 0.2784373803140559,
+      "loss": 0.34105658531188965,
+      "loss_ce": 0.018913013860583305,
+      "loss_iou": 1.100663185119629,
+      "loss_num": 0.322265625,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 125642448,
+      "step": 727
+    },
+    {
+      "epoch": 0.27882037533512066,
+      "grad_norm": 27.170960597013888,
+      "learning_rate": 5e-06,
+      "loss": 0.3035,
+      "num_input_tokens_seen": 125815328,
+      "step": 728
+    },
+    {
+      "epoch": 0.27882037533512066,
+      "loss": 0.3259619474411011,
+      "loss_ce": 0.01773439720273018,
+      "loss_iou": 1.0813090801239014,
+      "loss_num": 0.30859375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 125815328,
+      "step": 728
+    },
+    {
+      "epoch": 0.27920337035618537,
+      "grad_norm": 27.85329724527193,
+      "learning_rate": 5e-06,
+      "loss": 0.3797,
+      "num_input_tokens_seen": 125988392,
+      "step": 729
+    },
+    {
+      "epoch": 0.27920337035618537,
+      "loss": 0.4470553696155548,
+      "loss_ce": 0.01822238229215145,
+      "loss_iou": 1.1166222095489502,
+      "loss_num": 0.4296875,
+      "loss_xval": 0.4296875,
+      "num_input_tokens_seen": 125988392,
+      "step": 729
+    },
+    {
+      "epoch": 0.2795863653772501,
+      "grad_norm": 27.964041799621693,
+      "learning_rate": 5e-06,
+      "loss": 0.2793,
+      "num_input_tokens_seen": 126161264,
+      "step": 730
+    },
+    {
+      "epoch": 0.2795863653772501,
+      "loss": 0.3219105005264282,
+      "loss_ce": 0.01954232156276703,
+      "loss_iou": 1.0590667724609375,
+      "loss_num": 0.302734375,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 126161264,
+      "step": 730
+    },
+    {
+      "epoch": 0.2799693603983148,
+      "grad_norm": 27.277962054056832,
+      "learning_rate": 5e-06,
+      "loss": 0.3276,
+      "num_input_tokens_seen": 126334072,
+      "step": 731
+    },
+    {
+      "epoch": 0.2799693603983148,
+      "loss": 0.29830729961395264,
+      "loss_ce": 0.01864420250058174,
+      "loss_iou": 1.041162371635437,
+      "loss_num": 0.279296875,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 126334072,
+      "step": 731
+    },
+    {
+      "epoch": 0.28035235541937953,
+      "grad_norm": 29.51935930158991,
+      "learning_rate": 5e-06,
+      "loss": 0.2424,
+      "num_input_tokens_seen": 126507056,
+      "step": 732
+    },
+    {
+      "epoch": 0.28035235541937953,
+      "loss": 0.24611225724220276,
+      "loss_ce": 0.020038019865751266,
+      "loss_iou": 1.0084537267684937,
+      "loss_num": 0.2265625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 126507056,
+      "step": 732
+    },
+    {
+      "epoch": 0.2807353504404443,
+      "grad_norm": 25.375005606094607,
+      "learning_rate": 5e-06,
+      "loss": 0.3309,
+      "num_input_tokens_seen": 126680400,
+      "step": 733
+    },
+    {
+      "epoch": 0.2807353504404443,
+      "loss": 0.3649820387363434,
+      "loss_ce": 0.018180258572101593,
+      "loss_iou": 1.1179325580596924,
+      "loss_num": 0.34765625,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 126680400,
+      "step": 733
+    },
+    {
+      "epoch": 0.281118345461509,
+      "grad_norm": 26.700877001288095,
+      "learning_rate": 5e-06,
+      "loss": 0.2212,
+      "num_input_tokens_seen": 126853240,
+      "step": 734
+    },
+    {
+      "epoch": 0.281118345461509,
+      "loss": 0.19777071475982666,
+      "loss_ce": 0.018693555146455765,
+      "loss_iou": 1.0232832431793213,
+      "loss_num": 0.1787109375,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 126853240,
+      "step": 734
+    },
+    {
+      "epoch": 0.28150134048257375,
+      "grad_norm": 39.93003839780164,
+      "learning_rate": 5e-06,
+      "loss": 0.3455,
+      "num_input_tokens_seen": 127026200,
+      "step": 735
+    },
+    {
+      "epoch": 0.28150134048257375,
+      "loss": 0.3599078357219696,
+      "loss_ce": 0.019331656396389008,
+      "loss_iou": 1.0984218120574951,
+      "loss_num": 0.33984375,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 127026200,
+      "step": 735
+    },
+    {
+      "epoch": 0.28188433550363845,
+      "grad_norm": 29.79316016423221,
+      "learning_rate": 5e-06,
+      "loss": 0.2482,
+      "num_input_tokens_seen": 127199528,
+      "step": 736
+    },
+    {
+      "epoch": 0.28188433550363845,
+      "loss": 0.2514440417289734,
+      "loss_ce": 0.019022157415747643,
+      "loss_iou": 1.055418610572815,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 127199528,
+      "step": 736
+    },
+    {
+      "epoch": 0.28226733052470315,
+      "grad_norm": 26.06499652291949,
+      "learning_rate": 5e-06,
+      "loss": 0.2795,
+      "num_input_tokens_seen": 127372704,
+      "step": 737
+    },
+    {
+      "epoch": 0.28226733052470315,
+      "loss": 0.28856316208839417,
+      "loss_ce": 0.019276045262813568,
+      "loss_iou": 1.009185791015625,
+      "loss_num": 0.26953125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 127372704,
+      "step": 737
+    },
+    {
+      "epoch": 0.2826503255457679,
+      "grad_norm": 15.524540017190272,
+      "learning_rate": 5e-06,
+      "loss": 0.26,
+      "num_input_tokens_seen": 127545696,
+      "step": 738
+    },
+    {
+      "epoch": 0.2826503255457679,
+      "loss": 0.21141722798347473,
+      "loss_ce": 0.01915648952126503,
+      "loss_iou": 1.0185959339141846,
+      "loss_num": 0.1923828125,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 127545696,
+      "step": 738
+    },
+    {
+      "epoch": 0.2830333205668326,
+      "grad_norm": 27.915596055871134,
+      "learning_rate": 5e-06,
+      "loss": 0.2655,
+      "num_input_tokens_seen": 127718712,
+      "step": 739
+    },
+    {
+      "epoch": 0.2830333205668326,
+      "loss": 0.26364731788635254,
+      "loss_ce": 0.01779773086309433,
+      "loss_iou": 1.0782397985458374,
+      "loss_num": 0.24609375,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 127718712,
+      "step": 739
+    },
+    {
+      "epoch": 0.28341631558789737,
+      "grad_norm": 37.87314443952695,
+      "learning_rate": 5e-06,
+      "loss": 0.3398,
+      "num_input_tokens_seen": 127891736,
+      "step": 740
+    },
+    {
+      "epoch": 0.28341631558789737,
+      "loss": 0.28291386365890503,
+      "loss_ce": 0.01924198493361473,
+      "loss_iou": 1.0338754653930664,
+      "loss_num": 0.263671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 127891736,
+      "step": 740
+    },
+    {
+      "epoch": 0.2837993106089621,
+      "grad_norm": 31.10362567038928,
+      "learning_rate": 5e-06,
+      "loss": 0.3584,
+      "num_input_tokens_seen": 128065056,
+      "step": 741
+    },
+    {
+      "epoch": 0.2837993106089621,
+      "loss": 0.407207190990448,
+      "loss_ce": 0.018779434263706207,
+      "loss_iou": 1.0766929388046265,
+      "loss_num": 0.388671875,
+      "loss_xval": 0.388671875,
+      "num_input_tokens_seen": 128065056,
+      "step": 741
+    },
+    {
+      "epoch": 0.28418230563002683,
+      "grad_norm": 31.388919914621702,
+      "learning_rate": 5e-06,
+      "loss": 0.2627,
+      "num_input_tokens_seen": 128238128,
+      "step": 742
+    },
+    {
+      "epoch": 0.28418230563002683,
+      "loss": 0.3214864134788513,
+      "loss_ce": 0.01966756023466587,
+      "loss_iou": 1.1697998046875,
+      "loss_num": 0.302734375,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 128238128,
+      "step": 742
+    },
+    {
+      "epoch": 0.28456530065109154,
+      "grad_norm": 36.49353980146307,
+      "learning_rate": 5e-06,
+      "loss": 0.3421,
+      "num_input_tokens_seen": 128411368,
+      "step": 743
+    },
+    {
+      "epoch": 0.28456530065109154,
+      "loss": 0.3016979694366455,
+      "loss_ce": 0.01959346793591976,
+      "loss_iou": 1.0393989086151123,
+      "loss_num": 0.28125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 128411368,
+      "step": 743
+    },
+    {
+      "epoch": 0.28494829567215624,
+      "grad_norm": 31.74476571965495,
+      "learning_rate": 5e-06,
+      "loss": 0.26,
+      "num_input_tokens_seen": 128584456,
+      "step": 744
+    },
+    {
+      "epoch": 0.28494829567215624,
+      "loss": 0.268583744764328,
+      "loss_ce": 0.01919407956302166,
+      "loss_iou": 1.0217280387878418,
+      "loss_num": 0.2490234375,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 128584456,
+      "step": 744
+    },
+    {
+      "epoch": 0.285331290693221,
+      "grad_norm": 25.740209441166378,
+      "learning_rate": 5e-06,
+      "loss": 0.3807,
+      "num_input_tokens_seen": 128756616,
+      "step": 745
+    },
+    {
+      "epoch": 0.285331290693221,
+      "loss": 0.47400274872779846,
+      "loss_ce": 0.018070142716169357,
+      "loss_iou": 1.2395962476730347,
+      "loss_num": 0.455078125,
+      "loss_xval": 0.455078125,
+      "num_input_tokens_seen": 128756616,
+      "step": 745
+    },
+    {
+      "epoch": 0.2857142857142857,
+      "grad_norm": 22.444852305414027,
+      "learning_rate": 5e-06,
+      "loss": 0.2382,
+      "num_input_tokens_seen": 128929456,
+      "step": 746
+    },
+    {
+      "epoch": 0.2857142857142857,
+      "loss": 0.24783599376678467,
+      "loss_ce": 0.02029692381620407,
+      "loss_iou": 1.0339313745498657,
+      "loss_num": 0.2275390625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 128929456,
+      "step": 746
+    },
+    {
+      "epoch": 0.28609728073535046,
+      "grad_norm": 31.30095918600143,
+      "learning_rate": 5e-06,
+      "loss": 0.2983,
+      "num_input_tokens_seen": 129102544,
+      "step": 747
+    },
+    {
+      "epoch": 0.28609728073535046,
+      "loss": 0.30754023790359497,
+      "loss_ce": 0.01835567131638527,
+      "loss_iou": 1.047607183456421,
+      "loss_num": 0.2890625,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 129102544,
+      "step": 747
+    },
+    {
+      "epoch": 0.28648027575641516,
+      "grad_norm": 35.098898699133656,
+      "learning_rate": 5e-06,
+      "loss": 0.3248,
+      "num_input_tokens_seen": 129275432,
+      "step": 748
+    },
+    {
+      "epoch": 0.28648027575641516,
+      "loss": 0.2544245719909668,
+      "loss_ce": 0.018340574577450752,
+      "loss_iou": 1.0043387413024902,
+      "loss_num": 0.236328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 129275432,
+      "step": 748
+    },
+    {
+      "epoch": 0.2868632707774799,
+      "grad_norm": 34.0628159275406,
+      "learning_rate": 5e-06,
+      "loss": 0.3995,
+      "num_input_tokens_seen": 129448312,
+      "step": 749
+    },
+    {
+      "epoch": 0.2868632707774799,
+      "loss": 0.37420105934143066,
+      "loss_ce": 0.018976427614688873,
+      "loss_iou": 1.2287843227386475,
+      "loss_num": 0.35546875,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 129448312,
+      "step": 749
+    },
+    {
+      "epoch": 0.2872462657985446,
+      "grad_norm": 28.776119314117416,
+      "learning_rate": 5e-06,
+      "loss": 0.2512,
+      "num_input_tokens_seen": 129620848,
+      "step": 750
+    },
+    {
+      "epoch": 0.2872462657985446,
+      "eval_websight_new_CIoU": 0.8060732185840607,
+      "eval_websight_new_GIoU": 0.8037431240081787,
+      "eval_websight_new_IoU": 0.8077706694602966,
+      "eval_websight_new_MAE_all": 0.02072321344166994,
+      "eval_websight_new_MAE_h": 0.01762983202934265,
+      "eval_websight_new_MAE_w": 0.02194000780582428,
+      "eval_websight_new_MAE_x": 0.020607253536581993,
+      "eval_websight_new_MAE_y": 0.02271575853228569,
+      "eval_websight_new_NUM_probability": 0.00020467791182454675,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.2679460048675537,
+      "eval_websight_new_loss_ce": 0.037874773144721985,
+      "eval_websight_new_loss_iou": 1.0019870400428772,
+      "eval_websight_new_loss_num": 0.221923828125,
+      "eval_websight_new_loss_xval": 0.221923828125,
+      "eval_websight_new_runtime": 55.4103,
+      "eval_websight_new_samples_per_second": 0.902,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 129620848,
+      "step": 750
+    },
+    {
+      "epoch": 0.2872462657985446,
+      "eval_seeclick_CIoU": 0.6069816946983337,
+      "eval_seeclick_GIoU": 0.6073736250400543,
+      "eval_seeclick_IoU": 0.6343135833740234,
+      "eval_seeclick_MAE_all": 0.06425964087247849,
+      "eval_seeclick_MAE_h": 0.046736148186028004,
+      "eval_seeclick_MAE_w": 0.08796555548906326,
+      "eval_seeclick_MAE_x": 0.07397038117051125,
+      "eval_seeclick_MAE_y": 0.04836645536124706,
+      "eval_seeclick_NUM_probability": 0.0002977608673973009,
+      "eval_seeclick_inside_bbox": 0.8975694477558136,
+      "eval_seeclick_loss": 0.5584555864334106,
+      "eval_seeclick_loss_ce": 0.029193194583058357,
+      "eval_seeclick_loss_iou": 1.1114333868026733,
+      "eval_seeclick_loss_num": 0.4971923828125,
+      "eval_seeclick_loss_xval": 0.4971923828125,
+      "eval_seeclick_runtime": 81.4777,
+      "eval_seeclick_samples_per_second": 0.614,
+      "eval_seeclick_steps_per_second": 0.025,
+      "num_input_tokens_seen": 129620848,
+      "step": 750
+    },
+    {
+      "epoch": 0.2872462657985446,
+      "eval_icons_CIoU": 0.7668884098529816,
+      "eval_icons_GIoU": 0.7602577805519104,
+      "eval_icons_IoU": 0.7727258205413818,
+      "eval_icons_MAE_all": 0.028916318900883198,
+      "eval_icons_MAE_h": 0.02443860098719597,
+      "eval_icons_MAE_w": 0.03288193140178919,
+      "eval_icons_MAE_x": 0.029078337363898754,
+      "eval_icons_MAE_y": 0.02926640771329403,
+      "eval_icons_NUM_probability": 0.00020542916900012642,
+      "eval_icons_inside_bbox": 0.9722222089767456,
+      "eval_icons_loss": 0.30261683464050293,
+      "eval_icons_loss_ce": 0.025362316519021988,
+      "eval_icons_loss_iou": 1.0172606706619263,
+      "eval_icons_loss_num": 0.27044677734375,
+      "eval_icons_loss_xval": 0.27044677734375,
+      "eval_icons_runtime": 82.4836,
+      "eval_icons_samples_per_second": 0.606,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 129620848,
+      "step": 750
+    },
+    {
+      "epoch": 0.2872462657985446,
+      "loss": 0.3204486072063446,
+      "loss_ce": 0.025404643267393112,
+      "loss_iou": 1.0329668521881104,
+      "loss_num": 0.294921875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 129620848,
+      "step": 750
+    },
+    {
+      "epoch": 0.2876292608196093,
+      "grad_norm": 31.849347898039117,
+      "learning_rate": 5e-06,
+      "loss": 0.3413,
+      "num_input_tokens_seen": 129793888,
+      "step": 751
+    },
+    {
+      "epoch": 0.2876292608196093,
+      "loss": 0.3468954861164093,
+      "loss_ce": 0.0186484232544899,
+      "loss_iou": 1.0144870281219482,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 129793888,
+      "step": 751
+    },
+    {
+      "epoch": 0.2880122558406741,
+      "grad_norm": 37.12241440467024,
+      "learning_rate": 5e-06,
+      "loss": 0.2601,
+      "num_input_tokens_seen": 129966816,
+      "step": 752
+    },
+    {
+      "epoch": 0.2880122558406741,
+      "loss": 0.25784942507743835,
+      "loss_ce": 0.017126768827438354,
+      "loss_iou": 1.0310940742492676,
+      "loss_num": 0.240234375,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 129966816,
+      "step": 752
+    },
+    {
+      "epoch": 0.2883952508617388,
+      "grad_norm": 24.471411990030894,
+      "learning_rate": 5e-06,
+      "loss": 0.2579,
+      "num_input_tokens_seen": 130139432,
+      "step": 753
+    },
+    {
+      "epoch": 0.2883952508617388,
+      "loss": 0.27429860830307007,
+      "loss_ce": 0.017706789076328278,
+      "loss_iou": 1.004685401916504,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 130139432,
+      "step": 753
+    },
+    {
+      "epoch": 0.28877824588280354,
+      "grad_norm": 25.500871018993212,
+      "learning_rate": 5e-06,
+      "loss": 0.2831,
+      "num_input_tokens_seen": 130312408,
+      "step": 754
+    },
+    {
+      "epoch": 0.28877824588280354,
+      "loss": 0.241877943277359,
+      "loss_ce": 0.020137224346399307,
+      "loss_iou": 1.0630152225494385,
+      "loss_num": 0.2216796875,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 130312408,
+      "step": 754
+    },
+    {
+      "epoch": 0.28916124090386824,
+      "grad_norm": 29.8151233620471,
+      "learning_rate": 5e-06,
+      "loss": 0.3143,
+      "num_input_tokens_seen": 130485576,
+      "step": 755
+    },
+    {
+      "epoch": 0.28916124090386824,
+      "loss": 0.30824464559555054,
+      "loss_ce": 0.018815917894244194,
+      "loss_iou": 1.1340521574020386,
+      "loss_num": 0.2890625,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 130485576,
+      "step": 755
+    },
+    {
+      "epoch": 0.289544235924933,
+      "grad_norm": 28.68845671130155,
+      "learning_rate": 5e-06,
+      "loss": 0.2707,
+      "num_input_tokens_seen": 130658200,
+      "step": 756
+    },
+    {
+      "epoch": 0.289544235924933,
+      "loss": 0.2569507956504822,
+      "loss_ce": 0.01903577148914337,
+      "loss_iou": 1.1343841552734375,
+      "loss_num": 0.23828125,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 130658200,
+      "step": 756
+    },
+    {
+      "epoch": 0.2899272309459977,
+      "grad_norm": 31.452580410051898,
+      "learning_rate": 5e-06,
+      "loss": 0.2497,
+      "num_input_tokens_seen": 130830656,
+      "step": 757
+    },
+    {
+      "epoch": 0.2899272309459977,
+      "loss": 0.25728702545166016,
+      "loss_ce": 0.018273361027240753,
+      "loss_iou": 0.8802897930145264,
+      "loss_num": 0.2392578125,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 130830656,
+      "step": 757
+    },
+    {
+      "epoch": 0.2903102259670624,
+      "grad_norm": 18.518207113485683,
+      "learning_rate": 5e-06,
+      "loss": 0.2537,
+      "num_input_tokens_seen": 131003400,
+      "step": 758
+    },
+    {
+      "epoch": 0.2903102259670624,
+      "loss": 0.2733742296695709,
+      "loss_ce": 0.020322469994425774,
+      "loss_iou": 1.0106251239776611,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 131003400,
+      "step": 758
+    },
+    {
+      "epoch": 0.29069322098812717,
+      "grad_norm": 29.93569981388942,
+      "learning_rate": 5e-06,
+      "loss": 0.3701,
+      "num_input_tokens_seen": 131176440,
+      "step": 759
+    },
+    {
+      "epoch": 0.29069322098812717,
+      "loss": 0.3478788733482361,
+      "loss_ce": 0.020242154598236084,
+      "loss_iou": 1.3264551162719727,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 131176440,
+      "step": 759
+    },
+    {
+      "epoch": 0.29107621600919187,
+      "grad_norm": 34.59169012607738,
+      "learning_rate": 5e-06,
+      "loss": 0.3407,
+      "num_input_tokens_seen": 131349016,
+      "step": 760
+    },
+    {
+      "epoch": 0.29107621600919187,
+      "loss": 0.2980373203754425,
+      "loss_ce": 0.019045621156692505,
+      "loss_iou": 1.052343487739563,
+      "loss_num": 0.279296875,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 131349016,
+      "step": 760
+    },
+    {
+      "epoch": 0.2914592110302566,
+      "grad_norm": 29.391192734076682,
+      "learning_rate": 5e-06,
+      "loss": 0.2276,
+      "num_input_tokens_seen": 131521896,
+      "step": 761
+    },
+    {
+      "epoch": 0.2914592110302566,
+      "loss": 0.2416970133781433,
+      "loss_ce": 0.018308322876691818,
+      "loss_iou": 1.0112210512161255,
+      "loss_num": 0.2236328125,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 131521896,
+      "step": 761
+    },
+    {
+      "epoch": 0.29184220605132133,
+      "grad_norm": 25.01338260168904,
+      "learning_rate": 5e-06,
+      "loss": 0.3266,
+      "num_input_tokens_seen": 131694616,
+      "step": 762
+    },
+    {
+      "epoch": 0.29184220605132133,
+      "loss": 0.3483281135559082,
+      "loss_ce": 0.018982430920004845,
+      "loss_iou": 1.1844432353973389,
+      "loss_num": 0.330078125,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 131694616,
+      "step": 762
+    },
+    {
+      "epoch": 0.29222520107238603,
+      "grad_norm": 24.561687064793826,
+      "learning_rate": 5e-06,
+      "loss": 0.2714,
+      "num_input_tokens_seen": 131867904,
+      "step": 763
+    },
+    {
+      "epoch": 0.29222520107238603,
+      "loss": 0.2406422197818756,
+      "loss_ce": 0.018596317619085312,
+      "loss_iou": 1.0086827278137207,
+      "loss_num": 0.2216796875,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 131867904,
+      "step": 763
+    },
+    {
+      "epoch": 0.2926081960934508,
+      "grad_norm": 29.799083783300127,
+      "learning_rate": 5e-06,
+      "loss": 0.3068,
+      "num_input_tokens_seen": 132040896,
+      "step": 764
+    },
+    {
+      "epoch": 0.2926081960934508,
+      "loss": 0.26021048426628113,
+      "loss_ce": 0.019060572609305382,
+      "loss_iou": 1.0764524936676025,
+      "loss_num": 0.2412109375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 132040896,
+      "step": 764
+    },
+    {
+      "epoch": 0.2929911911145155,
+      "grad_norm": 40.476805309914155,
+      "learning_rate": 5e-06,
+      "loss": 0.3188,
+      "num_input_tokens_seen": 132213992,
+      "step": 765
+    },
+    {
+      "epoch": 0.2929911911145155,
+      "loss": 0.31016725301742554,
+      "loss_ce": 0.01768677681684494,
+      "loss_iou": 1.0198488235473633,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 132213992,
+      "step": 765
+    },
+    {
+      "epoch": 0.29337418613558025,
+      "grad_norm": 21.384278854128862,
+      "learning_rate": 5e-06,
+      "loss": 0.2179,
+      "num_input_tokens_seen": 132386728,
+      "step": 766
+    },
+    {
+      "epoch": 0.29337418613558025,
+      "loss": 0.25086313486099243,
+      "loss_ce": 0.019906092435121536,
+      "loss_iou": 1.0787184238433838,
+      "loss_num": 0.23046875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 132386728,
+      "step": 766
+    },
+    {
+      "epoch": 0.29375718115664495,
+      "grad_norm": 22.781816524931717,
+      "learning_rate": 5e-06,
+      "loss": 0.3074,
+      "num_input_tokens_seen": 132559584,
+      "step": 767
+    },
+    {
+      "epoch": 0.29375718115664495,
+      "loss": 0.24777740240097046,
+      "loss_ce": 0.01913970522582531,
+      "loss_iou": 1.0119497776031494,
+      "loss_num": 0.228515625,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 132559584,
+      "step": 767
+    },
+    {
+      "epoch": 0.2941401761777097,
+      "grad_norm": 25.50660002640219,
+      "learning_rate": 5e-06,
+      "loss": 0.2968,
+      "num_input_tokens_seen": 132732688,
+      "step": 768
+    },
+    {
+      "epoch": 0.2941401761777097,
+      "loss": 0.2769160866737366,
+      "loss_ce": 0.017760805785655975,
+      "loss_iou": 0.9788704514503479,
+      "loss_num": 0.259765625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 132732688,
+      "step": 768
+    },
+    {
+      "epoch": 0.2945231711987744,
+      "grad_norm": 27.785423557385965,
+      "learning_rate": 5e-06,
+      "loss": 0.307,
+      "num_input_tokens_seen": 132905752,
+      "step": 769
+    },
+    {
+      "epoch": 0.2945231711987744,
+      "loss": 0.3069145083427429,
+      "loss_ce": 0.02017132379114628,
+      "loss_iou": 1.0028711557388306,
+      "loss_num": 0.287109375,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 132905752,
+      "step": 769
+    },
+    {
+      "epoch": 0.2949061662198391,
+      "grad_norm": 32.232262113044406,
+      "learning_rate": 5e-06,
+      "loss": 0.2295,
+      "num_input_tokens_seen": 133078728,
+      "step": 770
+    },
+    {
+      "epoch": 0.2949061662198391,
+      "loss": 0.2590344548225403,
+      "loss_ce": 0.01977665163576603,
+      "loss_iou": 1.0257396697998047,
+      "loss_num": 0.2392578125,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 133078728,
+      "step": 770
+    },
+    {
+      "epoch": 0.2952891612409039,
+      "grad_norm": 25.545717091142183,
+      "learning_rate": 5e-06,
+      "loss": 0.309,
+      "num_input_tokens_seen": 133252048,
+      "step": 771
+    },
+    {
+      "epoch": 0.2952891612409039,
+      "loss": 0.20813432335853577,
+      "loss_ce": 0.018681202083826065,
+      "loss_iou": 1.003133773803711,
+      "loss_num": 0.189453125,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 133252048,
+      "step": 771
+    },
+    {
+      "epoch": 0.2956721562619686,
+      "grad_norm": 30.70035907777987,
+      "learning_rate": 5e-06,
+      "loss": 0.3112,
+      "num_input_tokens_seen": 133425080,
+      "step": 772
+    },
+    {
+      "epoch": 0.2956721562619686,
+      "loss": 0.307090699672699,
+      "loss_ce": 0.01912684738636017,
+      "loss_iou": 1.2478001117706299,
+      "loss_num": 0.287109375,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 133425080,
+      "step": 772
+    },
+    {
+      "epoch": 0.29605515128303334,
+      "grad_norm": 36.01061968484035,
+      "learning_rate": 5e-06,
+      "loss": 0.2957,
+      "num_input_tokens_seen": 133598400,
+      "step": 773
+    },
+    {
+      "epoch": 0.29605515128303334,
+      "loss": 0.2688688635826111,
+      "loss_ce": 0.019113002344965935,
+      "loss_iou": 1.0169953107833862,
+      "loss_num": 0.25,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 133598400,
+      "step": 773
+    },
+    {
+      "epoch": 0.29643814630409804,
+      "grad_norm": 41.38894880087424,
+      "learning_rate": 5e-06,
+      "loss": 0.2532,
+      "num_input_tokens_seen": 133771272,
+      "step": 774
+    },
+    {
+      "epoch": 0.29643814630409804,
+      "loss": 0.2356131374835968,
+      "loss_ce": 0.0201590396463871,
+      "loss_iou": 1.0057514905929565,
+      "loss_num": 0.2158203125,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 133771272,
+      "step": 774
+    },
+    {
+      "epoch": 0.2968211413251628,
+      "grad_norm": 32.49854990049222,
+      "learning_rate": 5e-06,
+      "loss": 0.3348,
+      "num_input_tokens_seen": 133944504,
+      "step": 775
+    },
+    {
+      "epoch": 0.2968211413251628,
+      "loss": 0.3293522596359253,
+      "loss_ce": 0.01965990662574768,
+      "loss_iou": 1.0076303482055664,
+      "loss_num": 0.310546875,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 133944504,
+      "step": 775
+    },
+    {
+      "epoch": 0.2972041363462275,
+      "grad_norm": 22.050109669731302,
+      "learning_rate": 5e-06,
+      "loss": 0.2193,
+      "num_input_tokens_seen": 134117504,
+      "step": 776
+    },
+    {
+      "epoch": 0.2972041363462275,
+      "loss": 0.1762588620185852,
+      "loss_ce": 0.018971281126141548,
+      "loss_iou": 1.001531720161438,
+      "loss_num": 0.1572265625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 134117504,
+      "step": 776
+    },
+    {
+      "epoch": 0.2975871313672922,
+      "grad_norm": 35.12163201054286,
+      "learning_rate": 5e-06,
+      "loss": 0.3317,
+      "num_input_tokens_seen": 134290544,
+      "step": 777
+    },
+    {
+      "epoch": 0.2975871313672922,
+      "loss": 0.31508395075798035,
+      "loss_ce": 0.019307564944028854,
+      "loss_iou": 0.992944598197937,
+      "loss_num": 0.294921875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 134290544,
+      "step": 777
+    },
+    {
+      "epoch": 0.29797012638835696,
+      "grad_norm": 31.412389288023597,
+      "learning_rate": 5e-06,
+      "loss": 0.2951,
+      "num_input_tokens_seen": 134463448,
+      "step": 778
+    },
+    {
+      "epoch": 0.29797012638835696,
+      "loss": 0.302811861038208,
+      "loss_ce": 0.0210736021399498,
+      "loss_iou": 1.0193989276885986,
+      "loss_num": 0.28125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 134463448,
+      "step": 778
+    },
+    {
+      "epoch": 0.29835312140942166,
+      "grad_norm": 38.631078016427374,
+      "learning_rate": 5e-06,
+      "loss": 0.3086,
+      "num_input_tokens_seen": 134636536,
+      "step": 779
+    },
+    {
+      "epoch": 0.29835312140942166,
+      "loss": 0.26435214281082153,
+      "loss_ce": 0.018502552062273026,
+      "loss_iou": 1.0163710117340088,
+      "loss_num": 0.24609375,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 134636536,
+      "step": 779
+    },
+    {
+      "epoch": 0.2987361164304864,
+      "grad_norm": 39.02751751222748,
+      "learning_rate": 5e-06,
+      "loss": 0.2279,
+      "num_input_tokens_seen": 134809632,
+      "step": 780
+    },
+    {
+      "epoch": 0.2987361164304864,
+      "loss": 0.2375146448612213,
+      "loss_ce": 0.01913086511194706,
+      "loss_iou": 1.0060218572616577,
+      "loss_num": 0.21875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 134809632,
+      "step": 780
+    },
+    {
+      "epoch": 0.2991191114515511,
+      "grad_norm": 23.628170294683574,
+      "learning_rate": 5e-06,
+      "loss": 0.3677,
+      "num_input_tokens_seen": 134982376,
+      "step": 781
+    },
+    {
+      "epoch": 0.2991191114515511,
+      "loss": 0.37691134214401245,
+      "loss_ce": 0.016437701880931854,
+      "loss_iou": 1.006775975227356,
+      "loss_num": 0.361328125,
+      "loss_xval": 0.361328125,
+      "num_input_tokens_seen": 134982376,
+      "step": 781
+    },
+    {
+      "epoch": 0.2995021064726159,
+      "grad_norm": 24.990854764855325,
+      "learning_rate": 5e-06,
+      "loss": 0.2969,
+      "num_input_tokens_seen": 135155400,
+      "step": 782
+    },
+    {
+      "epoch": 0.2995021064726159,
+      "loss": 0.2875215411186218,
+      "loss_ce": 0.01896686479449272,
+      "loss_iou": 1.0138508081436157,
+      "loss_num": 0.26953125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 135155400,
+      "step": 782
+    },
+    {
+      "epoch": 0.2998851014936806,
+      "grad_norm": 27.848857295651754,
+      "learning_rate": 5e-06,
+      "loss": 0.3309,
+      "num_input_tokens_seen": 135328296,
+      "step": 783
+    },
+    {
+      "epoch": 0.2998851014936806,
+      "loss": 0.3307138681411743,
+      "loss_ce": 0.017847657203674316,
+      "loss_iou": 1.03713858127594,
+      "loss_num": 0.3125,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 135328296,
+      "step": 783
+    },
+    {
+      "epoch": 0.3002680965147453,
+      "grad_norm": 27.869984789088633,
+      "learning_rate": 5e-06,
+      "loss": 0.3094,
+      "num_input_tokens_seen": 135501560,
+      "step": 784
+    },
+    {
+      "epoch": 0.3002680965147453,
+      "loss": 0.30624768137931824,
+      "loss_ce": 0.018405893817543983,
+      "loss_iou": 1.1890881061553955,
+      "loss_num": 0.287109375,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 135501560,
+      "step": 784
+    },
+    {
+      "epoch": 0.30065109153581004,
+      "grad_norm": 31.86857918382479,
+      "learning_rate": 5e-06,
+      "loss": 0.358,
+      "num_input_tokens_seen": 135674512,
+      "step": 785
+    },
+    {
+      "epoch": 0.30065109153581004,
+      "loss": 0.32729578018188477,
+      "loss_ce": 0.021021369844675064,
+      "loss_iou": 1.0098835229873657,
+      "loss_num": 0.306640625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 135674512,
+      "step": 785
+    },
+    {
+      "epoch": 0.30103408655687475,
+      "grad_norm": 27.724965314133073,
+      "learning_rate": 5e-06,
+      "loss": 0.2584,
+      "num_input_tokens_seen": 135847352,
+      "step": 786
+    },
+    {
+      "epoch": 0.30103408655687475,
+      "loss": 0.2529115378856659,
+      "loss_ce": 0.020794840529561043,
+      "loss_iou": 1.0112671852111816,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 135847352,
+      "step": 786
+    },
+    {
+      "epoch": 0.3014170815779395,
+      "grad_norm": 33.8414322542126,
+      "learning_rate": 5e-06,
+      "loss": 0.4155,
+      "num_input_tokens_seen": 136020400,
+      "step": 787
+    },
+    {
+      "epoch": 0.3014170815779395,
+      "loss": 0.3727918267250061,
+      "loss_ce": 0.017078936100006104,
+      "loss_iou": 1.0145914554595947,
+      "loss_num": 0.35546875,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 136020400,
+      "step": 787
+    },
+    {
+      "epoch": 0.3018000765990042,
+      "grad_norm": 29.200512375102747,
+      "learning_rate": 5e-06,
+      "loss": 0.2671,
+      "num_input_tokens_seen": 136193440,
+      "step": 788
+    },
+    {
+      "epoch": 0.3018000765990042,
+      "loss": 0.3048609495162964,
+      "loss_ce": 0.020986445248126984,
+      "loss_iou": 1.0601730346679688,
+      "loss_num": 0.283203125,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 136193440,
+      "step": 788
+    },
+    {
+      "epoch": 0.30218307162006897,
+      "grad_norm": 28.613521881724473,
+      "learning_rate": 5e-06,
+      "loss": 0.3944,
+      "num_input_tokens_seen": 136366792,
+      "step": 789
+    },
+    {
+      "epoch": 0.30218307162006897,
+      "loss": 0.4070029854774475,
+      "loss_ce": 0.019429761916399002,
+      "loss_iou": 1.359428882598877,
+      "loss_num": 0.38671875,
+      "loss_xval": 0.38671875,
+      "num_input_tokens_seen": 136366792,
+      "step": 789
+    },
+    {
+      "epoch": 0.30256606664113367,
+      "grad_norm": 25.204892331329287,
+      "learning_rate": 5e-06,
+      "loss": 0.2787,
+      "num_input_tokens_seen": 136539808,
+      "step": 790
+    },
+    {
+      "epoch": 0.30256606664113367,
+      "loss": 0.28682586550712585,
+      "loss_ce": 0.01936980150640011,
+      "loss_iou": 1.015669345855713,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 136539808,
+      "step": 790
+    },
+    {
+      "epoch": 0.30294906166219837,
+      "grad_norm": 39.99448952895715,
+      "learning_rate": 5e-06,
+      "loss": 0.3394,
+      "num_input_tokens_seen": 136712568,
+      "step": 791
+    },
+    {
+      "epoch": 0.30294906166219837,
+      "loss": 0.31899553537368774,
+      "loss_ce": 0.019434988498687744,
+      "loss_iou": 1.0062270164489746,
+      "loss_num": 0.298828125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 136712568,
+      "step": 791
+    },
+    {
+      "epoch": 0.30333205668326313,
+      "grad_norm": 29.618683629226744,
+      "learning_rate": 5e-06,
+      "loss": 0.3589,
+      "num_input_tokens_seen": 136885576,
+      "step": 792
+    },
+    {
+      "epoch": 0.30333205668326313,
+      "loss": 0.29877734184265137,
+      "loss_ce": 0.019114281982183456,
+      "loss_iou": 1.3513352870941162,
+      "loss_num": 0.279296875,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 136885576,
+      "step": 792
+    },
+    {
+      "epoch": 0.30371505170432783,
+      "grad_norm": 23.868257252018164,
+      "learning_rate": 5e-06,
+      "loss": 0.3871,
+      "num_input_tokens_seen": 137058208,
+      "step": 793
+    },
+    {
+      "epoch": 0.30371505170432783,
+      "loss": 0.3736051023006439,
+      "loss_ce": 0.019967414438724518,
+      "loss_iou": 1.1055314540863037,
+      "loss_num": 0.353515625,
+      "loss_xval": 0.353515625,
+      "num_input_tokens_seen": 137058208,
+      "step": 793
+    },
+    {
+      "epoch": 0.3040980467253926,
+      "grad_norm": 28.612832145842777,
+      "learning_rate": 5e-06,
+      "loss": 0.2283,
+      "num_input_tokens_seen": 137231120,
+      "step": 794
+    },
+    {
+      "epoch": 0.3040980467253926,
+      "loss": 0.2663195729255676,
+      "loss_ce": 0.02077515423297882,
+      "loss_iou": 1.237123727798462,
+      "loss_num": 0.2451171875,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 137231120,
+      "step": 794
+    },
+    {
+      "epoch": 0.3044810417464573,
+      "grad_norm": 21.584350122426212,
+      "learning_rate": 5e-06,
+      "loss": 0.2675,
+      "num_input_tokens_seen": 137404216,
+      "step": 795
+    },
+    {
+      "epoch": 0.3044810417464573,
+      "loss": 0.2667836546897888,
+      "loss_ce": 0.018980927765369415,
+      "loss_iou": 1.0179177522659302,
+      "loss_num": 0.248046875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 137404216,
+      "step": 795
+    },
+    {
+      "epoch": 0.304864036767522,
+      "grad_norm": 17.3797470257764,
+      "learning_rate": 5e-06,
+      "loss": 0.2406,
+      "num_input_tokens_seen": 137577296,
+      "step": 796
+    },
+    {
+      "epoch": 0.304864036767522,
+      "loss": 0.21924707293510437,
+      "loss_ce": 0.017647963017225266,
+      "loss_iou": 1.01650071144104,
+      "loss_num": 0.201171875,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 137577296,
+      "step": 796
+    },
+    {
+      "epoch": 0.30524703178858675,
+      "grad_norm": 19.681467637573444,
+      "learning_rate": 5e-06,
+      "loss": 0.2729,
+      "num_input_tokens_seen": 137746704,
+      "step": 797
+    },
+    {
+      "epoch": 0.30524703178858675,
+      "loss": 0.3018743395805359,
+      "loss_ce": 0.017633602023124695,
+      "loss_iou": 1.007925033569336,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 137746704,
+      "step": 797
+    },
+    {
+      "epoch": 0.30563002680965146,
+      "grad_norm": 27.588431696498727,
+      "learning_rate": 5e-06,
+      "loss": 0.2344,
+      "num_input_tokens_seen": 137920112,
+      "step": 798
+    },
+    {
+      "epoch": 0.30563002680965146,
+      "loss": 0.22935327887535095,
+      "loss_ce": 0.018659912049770355,
+      "loss_iou": 1.0037200450897217,
+      "loss_num": 0.2109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 137920112,
+      "step": 798
+    },
+    {
+      "epoch": 0.3060130218307162,
+      "grad_norm": 31.444068762978855,
+      "learning_rate": 5e-06,
+      "loss": 0.3038,
+      "num_input_tokens_seen": 138092696,
+      "step": 799
+    },
+    {
+      "epoch": 0.3060130218307162,
+      "loss": 0.2891741991043091,
+      "loss_ce": 0.018300168216228485,
+      "loss_iou": 1.0806390047073364,
+      "loss_num": 0.271484375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 138092696,
+      "step": 799
+    },
+    {
+      "epoch": 0.3063960168517809,
+      "grad_norm": 25.85811328706055,
+      "learning_rate": 5e-06,
+      "loss": 0.2525,
+      "num_input_tokens_seen": 138265656,
+      "step": 800
+    },
+    {
+      "epoch": 0.3063960168517809,
+      "loss": 0.2749815881252289,
+      "loss_ce": 0.018389806151390076,
+      "loss_iou": 1.0961915254592896,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 138265656,
+      "step": 800
+    },
+    {
+      "epoch": 0.3067790118728457,
+      "grad_norm": 19.998868990620483,
+      "learning_rate": 5e-06,
+      "loss": 0.3475,
+      "num_input_tokens_seen": 138438640,
+      "step": 801
+    },
+    {
+      "epoch": 0.3067790118728457,
+      "loss": 0.2885781526565552,
+      "loss_ce": 0.01941310614347458,
+      "loss_iou": 1.0149998664855957,
+      "loss_num": 0.26953125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 138438640,
+      "step": 801
+    },
+    {
+      "epoch": 0.3071620068939104,
+      "grad_norm": 20.921510079443348,
+      "learning_rate": 5e-06,
+      "loss": 0.2429,
+      "num_input_tokens_seen": 138611832,
+      "step": 802
+    },
+    {
+      "epoch": 0.3071620068939104,
+      "loss": 0.24668265879154205,
+      "loss_ce": 0.018289102241396904,
+      "loss_iou": 1.1573847532272339,
+      "loss_num": 0.228515625,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 138611832,
+      "step": 802
+    },
+    {
+      "epoch": 0.3075450019149751,
+      "grad_norm": 27.405235343015118,
+      "learning_rate": 5e-06,
+      "loss": 0.2824,
+      "num_input_tokens_seen": 138784816,
+      "step": 803
+    },
+    {
+      "epoch": 0.3075450019149751,
+      "loss": 0.3569422960281372,
+      "loss_ce": 0.01905165985226631,
+      "loss_iou": 1.0295093059539795,
+      "loss_num": 0.337890625,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 138784816,
+      "step": 803
+    },
+    {
+      "epoch": 0.30792799693603984,
+      "grad_norm": 32.963114559822415,
+      "learning_rate": 5e-06,
+      "loss": 0.3284,
+      "num_input_tokens_seen": 138958144,
+      "step": 804
+    },
+    {
+      "epoch": 0.30792799693603984,
+      "loss": 0.316406786441803,
+      "loss_ce": 0.021118704229593277,
+      "loss_iou": 1.2123587131500244,
+      "loss_num": 0.294921875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 138958144,
+      "step": 804
+    },
+    {
+      "epoch": 0.30831099195710454,
+      "grad_norm": 34.84300997985131,
+      "learning_rate": 5e-06,
+      "loss": 0.3417,
+      "num_input_tokens_seen": 139131544,
+      "step": 805
+    },
+    {
+      "epoch": 0.30831099195710454,
+      "loss": 0.3351994752883911,
+      "loss_ce": 0.019403565675020218,
+      "loss_iou": 1.6786623001098633,
+      "loss_num": 0.31640625,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 139131544,
+      "step": 805
+    },
+    {
+      "epoch": 0.3086939869781693,
+      "grad_norm": 35.45536605472206,
+      "learning_rate": 5e-06,
+      "loss": 0.2027,
+      "num_input_tokens_seen": 139304696,
+      "step": 806
+    },
+    {
+      "epoch": 0.3086939869781693,
+      "loss": 0.2326219379901886,
+      "loss_ce": 0.02009751833975315,
+      "loss_iou": 1.007890224456787,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 139304696,
+      "step": 806
+    },
+    {
+      "epoch": 0.309076981999234,
+      "grad_norm": 38.75706380675298,
+      "learning_rate": 5e-06,
+      "loss": 0.3064,
+      "num_input_tokens_seen": 139477816,
+      "step": 807
+    },
+    {
+      "epoch": 0.309076981999234,
+      "loss": 0.29244673252105713,
+      "loss_ce": 0.019375447183847427,
+      "loss_iou": 1.0221316814422607,
+      "loss_num": 0.2734375,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 139477816,
+      "step": 807
+    },
+    {
+      "epoch": 0.30945997702029876,
+      "grad_norm": 41.831056947268316,
+      "learning_rate": 5e-06,
+      "loss": 0.3134,
+      "num_input_tokens_seen": 139650344,
+      "step": 808
+    },
+    {
+      "epoch": 0.30945997702029876,
+      "loss": 0.26006972789764404,
+      "loss_ce": 0.01885880157351494,
+      "loss_iou": 1.0169495344161987,
+      "loss_num": 0.2412109375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 139650344,
+      "step": 808
+    },
+    {
+      "epoch": 0.30984297204136346,
+      "grad_norm": 50.82809307784458,
+      "learning_rate": 5e-06,
+      "loss": 0.3306,
+      "num_input_tokens_seen": 139823688,
+      "step": 809
+    },
+    {
+      "epoch": 0.30984297204136346,
+      "loss": 0.2908038794994354,
+      "loss_ce": 0.018831226974725723,
+      "loss_iou": 1.0181925296783447,
+      "loss_num": 0.271484375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 139823688,
+      "step": 809
+    },
+    {
+      "epoch": 0.31022596706242817,
+      "grad_norm": 38.353784907422515,
+      "learning_rate": 5e-06,
+      "loss": 0.3149,
+      "num_input_tokens_seen": 139996600,
+      "step": 810
+    },
+    {
+      "epoch": 0.31022596706242817,
+      "loss": 0.30111557245254517,
+      "loss_ce": 0.018644865602254868,
+      "loss_iou": 1.0038089752197266,
+      "loss_num": 0.283203125,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 139996600,
+      "step": 810
+    },
+    {
+      "epoch": 0.3106089620834929,
+      "grad_norm": 28.165365237091542,
+      "learning_rate": 5e-06,
+      "loss": 0.4093,
+      "num_input_tokens_seen": 140169192,
+      "step": 811
+    },
+    {
+      "epoch": 0.3106089620834929,
+      "loss": 0.4380716383457184,
+      "loss_ce": 0.016684912145137787,
+      "loss_iou": 1.016355037689209,
+      "loss_num": 0.421875,
+      "loss_xval": 0.421875,
+      "num_input_tokens_seen": 140169192,
+      "step": 811
+    },
+    {
+      "epoch": 0.3109919571045576,
+      "grad_norm": 36.22693262684854,
+      "learning_rate": 5e-06,
+      "loss": 0.3467,
+      "num_input_tokens_seen": 140342288,
+      "step": 812
+    },
+    {
+      "epoch": 0.3109919571045576,
+      "loss": 0.40430572628974915,
+      "loss_ce": 0.021249089390039444,
+      "loss_iou": 1.0298683643341064,
+      "loss_num": 0.3828125,
+      "loss_xval": 0.3828125,
+      "num_input_tokens_seen": 140342288,
+      "step": 812
+    },
+    {
+      "epoch": 0.3113749521256224,
+      "grad_norm": 68.34275654606967,
+      "learning_rate": 5e-06,
+      "loss": 0.439,
+      "num_input_tokens_seen": 140515280,
+      "step": 813
+    },
+    {
+      "epoch": 0.3113749521256224,
+      "loss": 0.4387940466403961,
+      "loss_ce": 0.01862802915275097,
+      "loss_iou": 1.0701375007629395,
+      "loss_num": 0.419921875,
+      "loss_xval": 0.419921875,
+      "num_input_tokens_seen": 140515280,
+      "step": 813
+    },
+    {
+      "epoch": 0.3117579471466871,
+      "grad_norm": 30.29609036013505,
+      "learning_rate": 5e-06,
+      "loss": 0.3877,
+      "num_input_tokens_seen": 140687808,
+      "step": 814
+    },
+    {
+      "epoch": 0.3117579471466871,
+      "loss": 0.41614609956741333,
+      "loss_ce": 0.017464444041252136,
+      "loss_iou": 1.0300133228302002,
+      "loss_num": 0.3984375,
+      "loss_xval": 0.3984375,
+      "num_input_tokens_seen": 140687808,
+      "step": 814
+    },
+    {
+      "epoch": 0.31214094216775184,
+      "grad_norm": 46.158913572281875,
+      "learning_rate": 5e-06,
+      "loss": 0.4618,
+      "num_input_tokens_seen": 140860720,
+      "step": 815
+    },
+    {
+      "epoch": 0.31214094216775184,
+      "loss": 0.45985519886016846,
+      "loss_ce": 0.019181368872523308,
+      "loss_iou": 1.0219111442565918,
+      "loss_num": 0.44140625,
+      "loss_xval": 0.44140625,
+      "num_input_tokens_seen": 140860720,
+      "step": 815
+    },
+    {
+      "epoch": 0.31252393718881655,
+      "grad_norm": 48.979671790420646,
+      "learning_rate": 5e-06,
+      "loss": 0.4771,
+      "num_input_tokens_seen": 141033824,
+      "step": 816
+    },
+    {
+      "epoch": 0.31252393718881655,
+      "loss": 0.44618505239486694,
+      "loss_ce": 0.01918311044573784,
+      "loss_iou": 1.0167924165725708,
+      "loss_num": 0.427734375,
+      "loss_xval": 0.427734375,
+      "num_input_tokens_seen": 141033824,
+      "step": 816
+    },
+    {
+      "epoch": 0.31290693220988125,
+      "grad_norm": 79.25767496978227,
+      "learning_rate": 5e-06,
+      "loss": 0.6913,
+      "num_input_tokens_seen": 141206736,
+      "step": 817
+    },
+    {
+      "epoch": 0.31290693220988125,
+      "loss": 0.6464214324951172,
+      "loss_ce": 0.02044481784105301,
+      "loss_iou": 1.0313572883605957,
+      "loss_num": 0.625,
+      "loss_xval": 0.625,
+      "num_input_tokens_seen": 141206736,
+      "step": 817
+    },
+    {
+      "epoch": 0.313289927230946,
+      "grad_norm": 59.700056047035524,
+      "learning_rate": 5e-06,
+      "loss": 0.486,
+      "num_input_tokens_seen": 141379880,
+      "step": 818
+    },
+    {
+      "epoch": 0.313289927230946,
+      "loss": 0.5164886116981506,
+      "loss_ce": 0.020150743424892426,
+      "loss_iou": 1.0376331806182861,
+      "loss_num": 0.49609375,
+      "loss_xval": 0.49609375,
+      "num_input_tokens_seen": 141379880,
+      "step": 818
+    },
+    {
+      "epoch": 0.3136729222520107,
+      "grad_norm": 29.73039959881463,
+      "learning_rate": 5e-06,
+      "loss": 0.4492,
+      "num_input_tokens_seen": 141552624,
+      "step": 819
+    },
+    {
+      "epoch": 0.3136729222520107,
+      "loss": 0.49371808767318726,
+      "loss_ce": 0.01959698647260666,
+      "loss_iou": 1.1965837478637695,
+      "loss_num": 0.474609375,
+      "loss_xval": 0.474609375,
+      "num_input_tokens_seen": 141552624,
+      "step": 819
+    },
+    {
+      "epoch": 0.31405591727307547,
+      "grad_norm": 29.265472210379624,
+      "learning_rate": 5e-06,
+      "loss": 0.4304,
+      "num_input_tokens_seen": 141725424,
+      "step": 820
+    },
+    {
+      "epoch": 0.31405591727307547,
+      "loss": 0.4803914427757263,
+      "loss_ce": 0.01945391856133938,
+      "loss_iou": 0.9033697843551636,
+      "loss_num": 0.4609375,
+      "loss_xval": 0.4609375,
+      "num_input_tokens_seen": 141725424,
+      "step": 820
+    },
+    {
+      "epoch": 0.31443891229414017,
+      "grad_norm": 37.26799133330325,
+      "learning_rate": 5e-06,
+      "loss": 0.3368,
+      "num_input_tokens_seen": 141898224,
+      "step": 821
+    },
+    {
+      "epoch": 0.31443891229414017,
+      "loss": 0.3688926696777344,
+      "loss_ce": 0.019039157778024673,
+      "loss_iou": 1.0247838497161865,
+      "loss_num": 0.349609375,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 141898224,
+      "step": 821
+    },
+    {
+      "epoch": 0.31482190731520493,
+      "grad_norm": 42.61123158430906,
+      "learning_rate": 5e-06,
+      "loss": 0.3873,
+      "num_input_tokens_seen": 142071184,
+      "step": 822
+    },
+    {
+      "epoch": 0.31482190731520493,
+      "loss": 0.36723488569259644,
+      "loss_ce": 0.017869655042886734,
+      "loss_iou": 1.1218441724777222,
+      "loss_num": 0.349609375,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 142071184,
+      "step": 822
+    },
+    {
+      "epoch": 0.31520490233626963,
+      "grad_norm": 33.9693213636612,
+      "learning_rate": 5e-06,
+      "loss": 0.3381,
+      "num_input_tokens_seen": 142244352,
+      "step": 823
+    },
+    {
+      "epoch": 0.31520490233626963,
+      "loss": 0.35591232776641846,
+      "loss_ce": 0.01789965108036995,
+      "loss_iou": 1.0056263208389282,
+      "loss_num": 0.337890625,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 142244352,
+      "step": 823
+    },
+    {
+      "epoch": 0.31558789735733433,
+      "grad_norm": 18.745058443742334,
+      "learning_rate": 5e-06,
+      "loss": 0.3706,
+      "num_input_tokens_seen": 142417576,
+      "step": 824
+    },
+    {
+      "epoch": 0.31558789735733433,
+      "loss": 0.38489848375320435,
+      "loss_ce": 0.01978621818125248,
+      "loss_iou": 1.0301867723464966,
+      "loss_num": 0.365234375,
+      "loss_xval": 0.365234375,
+      "num_input_tokens_seen": 142417576,
+      "step": 824
+    },
+    {
+      "epoch": 0.3159708923783991,
+      "grad_norm": 33.33675359937986,
+      "learning_rate": 5e-06,
+      "loss": 0.2685,
+      "num_input_tokens_seen": 142590552,
+      "step": 825
+    },
+    {
+      "epoch": 0.3159708923783991,
+      "loss": 0.27861785888671875,
+      "loss_ce": 0.0215377788990736,
+      "loss_iou": 1.0413572788238525,
+      "loss_num": 0.2578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 142590552,
+      "step": 825
+    },
+    {
+      "epoch": 0.3163538873994638,
+      "grad_norm": 41.838821770096324,
+      "learning_rate": 5e-06,
+      "loss": 0.2168,
+      "num_input_tokens_seen": 142763592,
+      "step": 826
+    },
+    {
+      "epoch": 0.3163538873994638,
+      "loss": 0.2080744355916977,
+      "loss_ce": 0.020574431866407394,
+      "loss_iou": 1.0031614303588867,
+      "loss_num": 0.1875,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 142763592,
+      "step": 826
+    },
+    {
+      "epoch": 0.31673688242052855,
+      "grad_norm": 52.73453060565837,
+      "learning_rate": 5e-06,
+      "loss": 0.3533,
+      "num_input_tokens_seen": 142936720,
+      "step": 827
+    },
+    {
+      "epoch": 0.31673688242052855,
+      "loss": 0.36775538325309753,
+      "loss_ce": 0.019854985177516937,
+      "loss_iou": 0.9030606150627136,
+      "loss_num": 0.34765625,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 142936720,
+      "step": 827
+    },
+    {
+      "epoch": 0.31711987744159326,
+      "grad_norm": 20.633616115057396,
+      "learning_rate": 5e-06,
+      "loss": 0.2597,
+      "num_input_tokens_seen": 143109240,
+      "step": 828
+    },
+    {
+      "epoch": 0.31711987744159326,
+      "loss": 0.27895843982696533,
+      "loss_ce": 0.019559025764465332,
+      "loss_iou": 1.019107460975647,
+      "loss_num": 0.259765625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 143109240,
+      "step": 828
+    },
+    {
+      "epoch": 0.31750287246265796,
+      "grad_norm": 28.53891989940894,
+      "learning_rate": 5e-06,
+      "loss": 0.3038,
+      "num_input_tokens_seen": 143282320,
+      "step": 829
+    },
+    {
+      "epoch": 0.31750287246265796,
+      "loss": 0.3055585026741028,
+      "loss_ce": 0.02015809714794159,
+      "loss_iou": 1.046279788017273,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 143282320,
+      "step": 829
+    },
+    {
+      "epoch": 0.3178858674837227,
+      "grad_norm": 27.62069897087929,
+      "learning_rate": 5e-06,
+      "loss": 0.3134,
+      "num_input_tokens_seen": 143454920,
+      "step": 830
+    },
+    {
+      "epoch": 0.3178858674837227,
+      "loss": 0.31350505352020264,
+      "loss_ce": 0.017972828820347786,
+      "loss_iou": 1.0730494260787964,
+      "loss_num": 0.294921875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 143454920,
+      "step": 830
+    },
+    {
+      "epoch": 0.3182688625047874,
+      "grad_norm": 31.893708533509034,
+      "learning_rate": 5e-06,
+      "loss": 0.2461,
+      "num_input_tokens_seen": 143627840,
+      "step": 831
+    },
+    {
+      "epoch": 0.3182688625047874,
+      "loss": 0.3111610412597656,
+      "loss_ce": 0.01794816553592682,
+      "loss_iou": 1.034367561340332,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 143627840,
+      "step": 831
+    },
+    {
+      "epoch": 0.3186518575258522,
+      "grad_norm": 45.06367253498661,
+      "learning_rate": 5e-06,
+      "loss": 0.2579,
+      "num_input_tokens_seen": 143801048,
+      "step": 832
+    },
+    {
+      "epoch": 0.3186518575258522,
+      "loss": 0.2307465523481369,
+      "loss_ce": 0.020541470497846603,
+      "loss_iou": 1.044113278388977,
+      "loss_num": 0.2099609375,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 143801048,
+      "step": 832
+    },
+    {
+      "epoch": 0.3190348525469169,
+      "grad_norm": 34.95905333874307,
+      "learning_rate": 5e-06,
+      "loss": 0.268,
+      "num_input_tokens_seen": 143973608,
+      "step": 833
+    },
+    {
+      "epoch": 0.3190348525469169,
+      "loss": 0.296118825674057,
+      "loss_ce": 0.019141294062137604,
+      "loss_iou": 1.2847325801849365,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 143973608,
+      "step": 833
+    },
+    {
+      "epoch": 0.31941784756798164,
+      "grad_norm": 37.812370808504184,
+      "learning_rate": 5e-06,
+      "loss": 0.2669,
+      "num_input_tokens_seen": 144146480,
+      "step": 834
+    },
+    {
+      "epoch": 0.31941784756798164,
+      "loss": 0.251359224319458,
+      "loss_ce": 0.02040218561887741,
+      "loss_iou": 1.1682796478271484,
+      "loss_num": 0.23046875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 144146480,
+      "step": 834
+    },
+    {
+      "epoch": 0.31980084258904634,
+      "grad_norm": 28.85114403789628,
+      "learning_rate": 5e-06,
+      "loss": 0.2441,
+      "num_input_tokens_seen": 144319496,
+      "step": 835
+    },
+    {
+      "epoch": 0.31980084258904634,
+      "loss": 0.2518581748008728,
+      "loss_ce": 0.019558370113372803,
+      "loss_iou": 1.008854866027832,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 144319496,
+      "step": 835
+    },
+    {
+      "epoch": 0.32018383761011104,
+      "grad_norm": 29.92956153922238,
+      "learning_rate": 5e-06,
+      "loss": 0.2753,
+      "num_input_tokens_seen": 144492864,
+      "step": 836
+    },
+    {
+      "epoch": 0.32018383761011104,
+      "loss": 0.28272461891174316,
+      "loss_ce": 0.02027345821261406,
+      "loss_iou": 1.1115727424621582,
+      "loss_num": 0.26171875,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 144492864,
+      "step": 836
+    },
+    {
+      "epoch": 0.3205668326311758,
+      "grad_norm": 21.881929259341096,
+      "learning_rate": 5e-06,
+      "loss": 0.2654,
+      "num_input_tokens_seen": 144665848,
+      "step": 837
+    },
+    {
+      "epoch": 0.3205668326311758,
+      "loss": 0.25894656777381897,
+      "loss_ce": 0.01895633339881897,
+      "loss_iou": 1.0287256240844727,
+      "loss_num": 0.240234375,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 144665848,
+      "step": 837
+    },
+    {
+      "epoch": 0.3209498276522405,
+      "grad_norm": 21.29026285801437,
+      "learning_rate": 5e-06,
+      "loss": 0.2512,
+      "num_input_tokens_seen": 144838976,
+      "step": 838
+    },
+    {
+      "epoch": 0.3209498276522405,
+      "loss": 0.22368401288986206,
+      "loss_ce": 0.018544865772128105,
+      "loss_iou": 1.0287506580352783,
+      "loss_num": 0.205078125,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 144838976,
+      "step": 838
+    },
+    {
+      "epoch": 0.32133282267330526,
+      "grad_norm": 26.849038724261195,
+      "learning_rate": 5e-06,
+      "loss": 0.2713,
+      "num_input_tokens_seen": 145011864,
+      "step": 839
+    },
+    {
+      "epoch": 0.32133282267330526,
+      "loss": 0.3127955198287964,
+      "loss_ce": 0.020070888102054596,
+      "loss_iou": 1.1161601543426514,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 145011864,
+      "step": 839
+    },
+    {
+      "epoch": 0.32171581769436997,
+      "grad_norm": 33.585060968994426,
+      "learning_rate": 5e-06,
+      "loss": 0.2158,
+      "num_input_tokens_seen": 145184768,
+      "step": 840
+    },
+    {
+      "epoch": 0.32171581769436997,
+      "loss": 0.19624000787734985,
+      "loss_ce": 0.020641859620809555,
+      "loss_iou": 1.0071518421173096,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 145184768,
+      "step": 840
+    },
+    {
+      "epoch": 0.3220988127154347,
+      "grad_norm": 31.301923816413904,
+      "learning_rate": 5e-06,
+      "loss": 0.2532,
+      "num_input_tokens_seen": 145357272,
+      "step": 841
+    },
+    {
+      "epoch": 0.3220988127154347,
+      "loss": 0.30623412132263184,
+      "loss_ce": 0.020162362605333328,
+      "loss_iou": 1.032745361328125,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 145357272,
+      "step": 841
+    },
+    {
+      "epoch": 0.3224818077364994,
+      "grad_norm": 25.348758789640222,
+      "learning_rate": 5e-06,
+      "loss": 0.22,
+      "num_input_tokens_seen": 145530344,
+      "step": 842
+    },
+    {
+      "epoch": 0.3224818077364994,
+      "loss": 0.19132661819458008,
+      "loss_ce": 0.019512636587023735,
+      "loss_iou": 1.0022637844085693,
+      "loss_num": 0.171875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 145530344,
+      "step": 842
+    },
+    {
+      "epoch": 0.32286480275756413,
+      "grad_norm": 24.638971635835045,
+      "learning_rate": 5e-06,
+      "loss": 0.2721,
+      "num_input_tokens_seen": 145703064,
+      "step": 843
+    },
+    {
+      "epoch": 0.32286480275756413,
+      "loss": 0.22540715336799622,
+      "loss_ce": 0.019230403006076813,
+      "loss_iou": 1.0078539848327637,
+      "loss_num": 0.2060546875,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 145703064,
+      "step": 843
+    },
+    {
+      "epoch": 0.3232477977786289,
+      "grad_norm": 44.55137979394336,
+      "learning_rate": 5e-06,
+      "loss": 0.2526,
+      "num_input_tokens_seen": 145876192,
+      "step": 844
+    },
+    {
+      "epoch": 0.3232477977786289,
+      "loss": 0.24295461177825928,
+      "loss_ce": 0.02127492055296898,
+      "loss_iou": 1.0485103130340576,
+      "loss_num": 0.2216796875,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 145876192,
+      "step": 844
+    },
+    {
+      "epoch": 0.3236307927996936,
+      "grad_norm": 24.890352961647036,
+      "learning_rate": 5e-06,
+      "loss": 0.2727,
+      "num_input_tokens_seen": 146049144,
+      "step": 845
+    },
+    {
+      "epoch": 0.3236307927996936,
+      "loss": 0.23165491223335266,
+      "loss_ce": 0.019496705383062363,
+      "loss_iou": 1.009504795074463,
+      "loss_num": 0.2119140625,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 146049144,
+      "step": 845
+    },
+    {
+      "epoch": 0.32401378782075835,
+      "grad_norm": 25.007161446809004,
+      "learning_rate": 5e-06,
+      "loss": 0.2209,
+      "num_input_tokens_seen": 146222448,
+      "step": 846
+    },
+    {
+      "epoch": 0.32401378782075835,
+      "loss": 0.20678061246871948,
+      "loss_ce": 0.018731294199824333,
+      "loss_iou": 1.0130515098571777,
+      "loss_num": 0.1884765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 146222448,
+      "step": 846
+    },
+    {
+      "epoch": 0.32439678284182305,
+      "grad_norm": 25.81087641029904,
+      "learning_rate": 5e-06,
+      "loss": 0.2753,
+      "num_input_tokens_seen": 146395592,
+      "step": 847
+    },
+    {
+      "epoch": 0.32439678284182305,
+      "loss": 0.27316826581954956,
+      "loss_ce": 0.01950616016983986,
+      "loss_iou": 1.010009527206421,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 146395592,
+      "step": 847
+    },
+    {
+      "epoch": 0.3247797778628878,
+      "grad_norm": 27.322352176353736,
+      "learning_rate": 5e-06,
+      "loss": 0.2061,
+      "num_input_tokens_seen": 146568568,
+      "step": 848
+    },
+    {
+      "epoch": 0.3247797778628878,
+      "loss": 0.18458572030067444,
+      "loss_ce": 0.018814239650964737,
+      "loss_iou": 1.0022218227386475,
+      "loss_num": 0.166015625,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 146568568,
+      "step": 848
+    },
+    {
+      "epoch": 0.3251627728839525,
+      "grad_norm": 24.522639268896462,
+      "learning_rate": 5e-06,
+      "loss": 0.2667,
+      "num_input_tokens_seen": 146741328,
+      "step": 849
+    },
+    {
+      "epoch": 0.3251627728839525,
+      "loss": 0.27562767267227173,
+      "loss_ce": 0.020378652960062027,
+      "loss_iou": 1.0394654273986816,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 146741328,
+      "step": 849
+    },
+    {
+      "epoch": 0.3255457679050172,
+      "grad_norm": 15.574133460724482,
+      "learning_rate": 5e-06,
+      "loss": 0.1955,
+      "num_input_tokens_seen": 146914040,
+      "step": 850
+    },
+    {
+      "epoch": 0.3255457679050172,
+      "loss": 0.21060767769813538,
+      "loss_ce": 0.018285904079675674,
+      "loss_iou": 1.0735305547714233,
+      "loss_num": 0.1923828125,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 146914040,
+      "step": 850
+    },
+    {
+      "epoch": 0.32592876292608197,
+      "grad_norm": 18.379237703982493,
+      "learning_rate": 5e-06,
+      "loss": 0.2648,
+      "num_input_tokens_seen": 147087104,
+      "step": 851
+    },
+    {
+      "epoch": 0.32592876292608197,
+      "loss": 0.24579739570617676,
+      "loss_ce": 0.01929592713713646,
+      "loss_iou": 1.0109920501708984,
+      "loss_num": 0.2265625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 147087104,
+      "step": 851
+    },
+    {
+      "epoch": 0.3263117579471467,
+      "grad_norm": 28.601822137055382,
+      "learning_rate": 5e-06,
+      "loss": 0.2195,
+      "num_input_tokens_seen": 147259904,
+      "step": 852
+    },
+    {
+      "epoch": 0.3263117579471467,
+      "loss": 0.2619357109069824,
+      "loss_ce": 0.019992347806692123,
+      "loss_iou": 1.066663384437561,
+      "loss_num": 0.2421875,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 147259904,
+      "step": 852
+    },
+    {
+      "epoch": 0.32669475296821143,
+      "grad_norm": 27.97334658629691,
+      "learning_rate": 5e-06,
+      "loss": 0.2733,
+      "num_input_tokens_seen": 147433096,
+      "step": 853
+    },
+    {
+      "epoch": 0.32669475296821143,
+      "loss": 0.27185162901878357,
+      "loss_ce": 0.02038678154349327,
+      "loss_iou": 0.9701149463653564,
+      "loss_num": 0.251953125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 147433096,
+      "step": 853
+    },
+    {
+      "epoch": 0.32707774798927614,
+      "grad_norm": 32.71851833006168,
+      "learning_rate": 5e-06,
+      "loss": 0.255,
+      "num_input_tokens_seen": 147605824,
+      "step": 854
+    },
+    {
+      "epoch": 0.32707774798927614,
+      "loss": 0.2587239146232605,
+      "loss_ce": 0.017024705186486244,
+      "loss_iou": 0.9123989343643188,
+      "loss_num": 0.2421875,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 147605824,
+      "step": 854
+    },
+    {
+      "epoch": 0.32746074301034084,
+      "grad_norm": 22.92444775658955,
+      "learning_rate": 5e-06,
+      "loss": 0.2654,
+      "num_input_tokens_seen": 147778928,
+      "step": 855
+    },
+    {
+      "epoch": 0.32746074301034084,
+      "loss": 0.2387467622756958,
+      "loss_ce": 0.018898125737905502,
+      "loss_iou": 1.0224438905715942,
+      "loss_num": 0.2197265625,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 147778928,
+      "step": 855
+    },
+    {
+      "epoch": 0.3278437380314056,
+      "grad_norm": 25.280672116603366,
+      "learning_rate": 5e-06,
+      "loss": 0.2834,
+      "num_input_tokens_seen": 147951744,
+      "step": 856
+    },
+    {
+      "epoch": 0.3278437380314056,
+      "loss": 0.21689927577972412,
+      "loss_ce": 0.020121928304433823,
+      "loss_iou": 0.9918257594108582,
+      "loss_num": 0.197265625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 147951744,
+      "step": 856
+    },
+    {
+      "epoch": 0.3282267330524703,
+      "grad_norm": 28.654276497458522,
+      "learning_rate": 5e-06,
+      "loss": 0.2463,
+      "num_input_tokens_seen": 148124912,
+      "step": 857
+    },
+    {
+      "epoch": 0.3282267330524703,
+      "loss": 0.27426350116729736,
+      "loss_ce": 0.020479325205087662,
+      "loss_iou": 1.02805495262146,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 148124912,
+      "step": 857
+    },
+    {
+      "epoch": 0.32860972807353506,
+      "grad_norm": 33.29380106819679,
+      "learning_rate": 5e-06,
+      "loss": 0.2217,
+      "num_input_tokens_seen": 148298080,
+      "step": 858
+    },
+    {
+      "epoch": 0.32860972807353506,
+      "loss": 0.21757923066616058,
+      "loss_ce": 0.020557750016450882,
+      "loss_iou": 1.0261526107788086,
+      "loss_num": 0.197265625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 148298080,
+      "step": 858
+    },
+    {
+      "epoch": 0.32899272309459976,
+      "grad_norm": 33.55535640775121,
+      "learning_rate": 5e-06,
+      "loss": 0.2818,
+      "num_input_tokens_seen": 148471016,
+      "step": 859
+    },
+    {
+      "epoch": 0.32899272309459976,
+      "loss": 0.21814215183258057,
+      "loss_ce": 0.018923403695225716,
+      "loss_iou": 1.0018761157989502,
+      "loss_num": 0.19921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 148471016,
+      "step": 859
+    },
+    {
+      "epoch": 0.3293757181156645,
+      "grad_norm": 26.54593700223138,
+      "learning_rate": 5e-06,
+      "loss": 0.2736,
+      "num_input_tokens_seen": 148644016,
+      "step": 860
+    },
+    {
+      "epoch": 0.3293757181156645,
+      "loss": 0.31072235107421875,
+      "loss_ce": 0.019523631781339645,
+      "loss_iou": 1.087246298789978,
+      "loss_num": 0.291015625,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 148644016,
+      "step": 860
+    },
+    {
+      "epoch": 0.3297587131367292,
+      "grad_norm": 22.016117691997643,
+      "learning_rate": 5e-06,
+      "loss": 0.2776,
+      "num_input_tokens_seen": 148816904,
+      "step": 861
+    },
+    {
+      "epoch": 0.3297587131367292,
+      "loss": 0.2748348116874695,
+      "loss_ce": 0.01958579383790493,
+      "loss_iou": 1.0558685064315796,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 148816904,
+      "step": 861
+    },
+    {
+      "epoch": 0.3301417081577939,
+      "grad_norm": 27.05117190953951,
+      "learning_rate": 5e-06,
+      "loss": 0.2274,
+      "num_input_tokens_seen": 148989896,
+      "step": 862
+    },
+    {
+      "epoch": 0.3301417081577939,
+      "loss": 0.20970948040485382,
+      "loss_ce": 0.020439468324184418,
+      "loss_iou": 1.0353190898895264,
+      "loss_num": 0.189453125,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 148989896,
+      "step": 862
+    },
+    {
+      "epoch": 0.3305247031788587,
+      "grad_norm": 35.11845782318814,
+      "learning_rate": 5e-06,
+      "loss": 0.2434,
+      "num_input_tokens_seen": 149163184,
+      "step": 863
+    },
+    {
+      "epoch": 0.3305247031788587,
+      "loss": 0.271926611661911,
+      "loss_ce": 0.019851425662636757,
+      "loss_iou": 1.0185465812683105,
+      "loss_num": 0.251953125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 149163184,
+      "step": 863
+    },
+    {
+      "epoch": 0.3309076981999234,
+      "grad_norm": 26.226221844466277,
+      "learning_rate": 5e-06,
+      "loss": 0.3065,
+      "num_input_tokens_seen": 149336352,
+      "step": 864
+    },
+    {
+      "epoch": 0.3309076981999234,
+      "loss": 0.28416454792022705,
+      "loss_ce": 0.019516095519065857,
+      "loss_iou": 1.0216724872589111,
+      "loss_num": 0.265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 149336352,
+      "step": 864
+    },
+    {
+      "epoch": 0.33129069322098814,
+      "grad_norm": 23.19677223917612,
+      "learning_rate": 5e-06,
+      "loss": 0.2548,
+      "num_input_tokens_seen": 149509128,
+      "step": 865
+    },
+    {
+      "epoch": 0.33129069322098814,
+      "loss": 0.3117525577545166,
+      "loss_ce": 0.019394144415855408,
+      "loss_iou": 1.027750849723816,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 149509128,
+      "step": 865
+    },
+    {
+      "epoch": 0.33167368824205284,
+      "grad_norm": 22.600602819517086,
+      "learning_rate": 5e-06,
+      "loss": 0.2817,
+      "num_input_tokens_seen": 149682136,
+      "step": 866
+    },
+    {
+      "epoch": 0.33167368824205284,
+      "loss": 0.23056963086128235,
+      "loss_ce": 0.01926591619849205,
+      "loss_iou": 0.9869059324264526,
+      "loss_num": 0.2109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 149682136,
+      "step": 866
+    },
+    {
+      "epoch": 0.3320566832631176,
+      "grad_norm": 34.16005746791917,
+      "learning_rate": 5e-06,
+      "loss": 0.2815,
+      "num_input_tokens_seen": 149855344,
+      "step": 867
+    },
+    {
+      "epoch": 0.3320566832631176,
+      "loss": 0.28401294350624084,
+      "loss_ce": 0.020463157445192337,
+      "loss_iou": 1.9210349321365356,
+      "loss_num": 0.263671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 149855344,
+      "step": 867
+    },
+    {
+      "epoch": 0.3324396782841823,
+      "grad_norm": 29.907646899782975,
+      "learning_rate": 5e-06,
+      "loss": 0.2703,
+      "num_input_tokens_seen": 150028720,
+      "step": 868
+    },
+    {
+      "epoch": 0.3324396782841823,
+      "loss": 0.2970985174179077,
+      "loss_ce": 0.019998911768198013,
+      "loss_iou": 1.049875259399414,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 150028720,
+      "step": 868
+    },
+    {
+      "epoch": 0.332822673305247,
+      "grad_norm": 29.99771694528357,
+      "learning_rate": 5e-06,
+      "loss": 0.3178,
+      "num_input_tokens_seen": 150201784,
+      "step": 869
+    },
+    {
+      "epoch": 0.332822673305247,
+      "loss": 0.3909516930580139,
+      "loss_ce": 0.018515150994062424,
+      "loss_iou": 1.0345795154571533,
+      "loss_num": 0.373046875,
+      "loss_xval": 0.373046875,
+      "num_input_tokens_seen": 150201784,
+      "step": 869
+    },
+    {
+      "epoch": 0.33320566832631177,
+      "grad_norm": 33.004047019120286,
+      "learning_rate": 5e-06,
+      "loss": 0.2027,
+      "num_input_tokens_seen": 150374760,
+      "step": 870
+    },
+    {
+      "epoch": 0.33320566832631177,
+      "loss": 0.180461585521698,
+      "loss_ce": 0.021342938765883446,
+      "loss_iou": 1.0032970905303955,
+      "loss_num": 0.1591796875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 150374760,
+      "step": 870
+    },
+    {
+      "epoch": 0.33358866334737647,
+      "grad_norm": 45.27232694325867,
+      "learning_rate": 5e-06,
+      "loss": 0.3349,
+      "num_input_tokens_seen": 150547536,
+      "step": 871
+    },
+    {
+      "epoch": 0.33358866334737647,
+      "loss": 0.35715845227241516,
+      "loss_ce": 0.018901610746979713,
+      "loss_iou": 1.0309621095657349,
+      "loss_num": 0.337890625,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 150547536,
+      "step": 871
+    },
+    {
+      "epoch": 0.3339716583684412,
+      "grad_norm": 31.843471049279653,
+      "learning_rate": 5e-06,
+      "loss": 0.2672,
+      "num_input_tokens_seen": 150720560,
+      "step": 872
+    },
+    {
+      "epoch": 0.3339716583684412,
+      "loss": 0.23046442866325378,
+      "loss_ce": 0.01879449561238289,
+      "loss_iou": 1.0066142082214355,
+      "loss_num": 0.2119140625,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 150720560,
+      "step": 872
+    },
+    {
+      "epoch": 0.33435465338950593,
+      "grad_norm": 33.7652737493065,
+      "learning_rate": 5e-06,
+      "loss": 0.262,
+      "num_input_tokens_seen": 150893480,
+      "step": 873
+    },
+    {
+      "epoch": 0.33435465338950593,
+      "loss": 0.22619307041168213,
+      "loss_ce": 0.02123701572418213,
+      "loss_iou": 1.004620909690857,
+      "loss_num": 0.205078125,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 150893480,
+      "step": 873
+    },
+    {
+      "epoch": 0.3347376484105707,
+      "grad_norm": 33.864588355012586,
+      "learning_rate": 5e-06,
+      "loss": 0.2216,
+      "num_input_tokens_seen": 151063080,
+      "step": 874
+    },
+    {
+      "epoch": 0.3347376484105707,
+      "loss": 0.2183486968278885,
+      "loss_ce": 0.02132720872759819,
+      "loss_iou": 1.015843391418457,
+      "loss_num": 0.197265625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 151063080,
+      "step": 874
+    },
+    {
+      "epoch": 0.3351206434316354,
+      "grad_norm": 27.357396846645152,
+      "learning_rate": 5e-06,
+      "loss": 0.3265,
+      "num_input_tokens_seen": 151235760,
+      "step": 875
+    },
+    {
+      "epoch": 0.3351206434316354,
+      "loss": 0.3101559281349182,
+      "loss_ce": 0.019018245860934258,
+      "loss_iou": 1.107253909111023,
+      "loss_num": 0.291015625,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 151235760,
+      "step": 875
+    },
+    {
+      "epoch": 0.3355036384527001,
+      "grad_norm": 13.869717694026997,
+      "learning_rate": 5e-06,
+      "loss": 0.2955,
+      "num_input_tokens_seen": 151408968,
+      "step": 876
+    },
+    {
+      "epoch": 0.3355036384527001,
+      "loss": 0.3140192925930023,
+      "loss_ce": 0.01970777101814747,
+      "loss_iou": 1.2262418270111084,
+      "loss_num": 0.294921875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 151408968,
+      "step": 876
+    },
+    {
+      "epoch": 0.33588663347376485,
+      "grad_norm": 14.395864679785616,
+      "learning_rate": 5e-06,
+      "loss": 0.2199,
+      "num_input_tokens_seen": 151581912,
+      "step": 877
+    },
+    {
+      "epoch": 0.33588663347376485,
+      "loss": 0.25289204716682434,
+      "loss_ce": 0.01924947090446949,
+      "loss_iou": 1.035520076751709,
+      "loss_num": 0.2333984375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 151581912,
+      "step": 877
+    },
+    {
+      "epoch": 0.33626962849482955,
+      "grad_norm": 28.536779401306894,
+      "learning_rate": 5e-06,
+      "loss": 0.2443,
+      "num_input_tokens_seen": 151754712,
+      "step": 878
+    },
+    {
+      "epoch": 0.33626962849482955,
+      "loss": 0.25656014680862427,
+      "loss_ce": 0.017912685871124268,
+      "loss_iou": 1.0007470846176147,
+      "loss_num": 0.23828125,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 151754712,
+      "step": 878
+    },
+    {
+      "epoch": 0.3366526235158943,
+      "grad_norm": 31.989938231085112,
+      "learning_rate": 5e-06,
+      "loss": 0.205,
+      "num_input_tokens_seen": 151927232,
+      "step": 879
+    },
+    {
+      "epoch": 0.3366526235158943,
+      "loss": 0.17606347799301147,
+      "loss_ce": 0.018714847043156624,
+      "loss_iou": 1.0011568069458008,
+      "loss_num": 0.1572265625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 151927232,
+      "step": 879
+    },
+    {
+      "epoch": 0.337035618536959,
+      "grad_norm": 33.091155354932745,
+      "learning_rate": 5e-06,
+      "loss": 0.1916,
+      "num_input_tokens_seen": 152100176,
+      "step": 880
+    },
+    {
+      "epoch": 0.337035618536959,
+      "loss": 0.16588345170021057,
+      "loss_ce": 0.019704259932041168,
+      "loss_iou": 1.0007708072662354,
+      "loss_num": 0.146484375,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 152100176,
+      "step": 880
+    },
+    {
+      "epoch": 0.33741861355802377,
+      "grad_norm": 35.506378999432776,
+      "learning_rate": 5e-06,
+      "loss": 0.2679,
+      "num_input_tokens_seen": 152272968,
+      "step": 881
+    },
+    {
+      "epoch": 0.33741861355802377,
+      "loss": 0.2750643491744995,
+      "loss_ce": 0.019143931567668915,
+      "loss_iou": 1.0157005786895752,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 152272968,
+      "step": 881
+    },
+    {
+      "epoch": 0.3378016085790885,
+      "grad_norm": 29.668205512706834,
+      "learning_rate": 5e-06,
+      "loss": 0.2678,
+      "num_input_tokens_seen": 152446016,
+      "step": 882
+    },
+    {
+      "epoch": 0.3378016085790885,
+      "loss": 0.311052531003952,
+      "loss_ce": 0.01869412139058113,
+      "loss_iou": 1.227002739906311,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 152446016,
+      "step": 882
+    },
+    {
+      "epoch": 0.3381846036001532,
+      "grad_norm": 27.83597592754868,
+      "learning_rate": 5e-06,
+      "loss": 0.2616,
+      "num_input_tokens_seen": 152618800,
+      "step": 883
+    },
+    {
+      "epoch": 0.3381846036001532,
+      "loss": 0.2664565145969391,
+      "loss_ce": 0.01950826123356819,
+      "loss_iou": 1.0522820949554443,
+      "loss_num": 0.2470703125,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 152618800,
+      "step": 883
+    },
+    {
+      "epoch": 0.33856759862121794,
+      "grad_norm": 31.058857810008302,
+      "learning_rate": 5e-06,
+      "loss": 0.2731,
+      "num_input_tokens_seen": 152792032,
+      "step": 884
+    },
+    {
+      "epoch": 0.33856759862121794,
+      "loss": 0.3341100811958313,
+      "loss_ce": 0.019229726865887642,
+      "loss_iou": 1.10325288772583,
+      "loss_num": 0.314453125,
+      "loss_xval": 0.314453125,
+      "num_input_tokens_seen": 152792032,
+      "step": 884
+    },
+    {
+      "epoch": 0.33895059364228264,
+      "grad_norm": 22.04291331215238,
+      "learning_rate": 5e-06,
+      "loss": 0.2674,
+      "num_input_tokens_seen": 152965144,
+      "step": 885
+    },
+    {
+      "epoch": 0.33895059364228264,
+      "loss": 0.24300506711006165,
+      "loss_ce": 0.018273629248142242,
+      "loss_iou": 1.0013980865478516,
+      "loss_num": 0.224609375,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 152965144,
+      "step": 885
+    },
+    {
+      "epoch": 0.3393335886633474,
+      "grad_norm": 22.290170546120653,
+      "learning_rate": 5e-06,
+      "loss": 0.2737,
+      "num_input_tokens_seen": 153137984,
+      "step": 886
+    },
+    {
+      "epoch": 0.3393335886633474,
+      "loss": 0.2488933652639389,
+      "loss_ce": 0.019523244351148605,
+      "loss_iou": 2.009737968444824,
+      "loss_num": 0.2294921875,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 153137984,
+      "step": 886
+    },
+    {
+      "epoch": 0.3397165836844121,
+      "grad_norm": 32.707924870130725,
+      "learning_rate": 5e-06,
+      "loss": 0.2528,
+      "num_input_tokens_seen": 153311136,
+      "step": 887
+    },
+    {
+      "epoch": 0.3397165836844121,
+      "loss": 0.2415219247341156,
+      "loss_ce": 0.019476018846035004,
+      "loss_iou": 1.016276478767395,
+      "loss_num": 0.2216796875,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 153311136,
+      "step": 887
+    },
+    {
+      "epoch": 0.3400995787054768,
+      "grad_norm": 34.992697891143266,
+      "learning_rate": 5e-06,
+      "loss": 0.2079,
+      "num_input_tokens_seen": 153483800,
+      "step": 888
+    },
+    {
+      "epoch": 0.3400995787054768,
+      "loss": 0.2507489323616028,
+      "loss_ce": 0.018876362591981888,
+      "loss_iou": 1.0302425622940063,
+      "loss_num": 0.2314453125,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 153483800,
+      "step": 888
+    },
+    {
+      "epoch": 0.34048257372654156,
+      "grad_norm": 41.754336205978944,
+      "learning_rate": 5e-06,
+      "loss": 0.2532,
+      "num_input_tokens_seen": 153656496,
+      "step": 889
+    },
+    {
+      "epoch": 0.34048257372654156,
+      "loss": 0.23359987139701843,
+      "loss_ce": 0.020953377708792686,
+      "loss_iou": 1.0017032623291016,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 153656496,
+      "step": 889
+    },
+    {
+      "epoch": 0.34086556874760626,
+      "grad_norm": 30.40687800194179,
+      "learning_rate": 5e-06,
+      "loss": 0.2286,
+      "num_input_tokens_seen": 153829608,
+      "step": 890
+    },
+    {
+      "epoch": 0.34086556874760626,
+      "loss": 0.25328874588012695,
+      "loss_ce": 0.019218916073441505,
+      "loss_iou": 1.0172107219696045,
+      "loss_num": 0.234375,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 153829608,
+      "step": 890
+    },
+    {
+      "epoch": 0.341248563768671,
+      "grad_norm": 31.11285113679715,
+      "learning_rate": 5e-06,
+      "loss": 0.3226,
+      "num_input_tokens_seen": 154002776,
+      "step": 891
+    },
+    {
+      "epoch": 0.341248563768671,
+      "loss": 0.2829424738883972,
+      "loss_ce": 0.020247183740139008,
+      "loss_iou": 1.0092939138412476,
+      "loss_num": 0.26171875,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 154002776,
+      "step": 891
+    },
+    {
+      "epoch": 0.3416315587897357,
+      "grad_norm": 47.443313103501666,
+      "learning_rate": 5e-06,
+      "loss": 0.2477,
+      "num_input_tokens_seen": 154176184,
+      "step": 892
+    },
+    {
+      "epoch": 0.3416315587897357,
+      "loss": 0.16495199501514435,
+      "loss_ce": 0.02164144441485405,
+      "loss_iou": 1.0047979354858398,
+      "loss_num": 0.1435546875,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 154176184,
+      "step": 892
+    },
+    {
+      "epoch": 0.3420145538108005,
+      "grad_norm": 60.4096998334582,
+      "learning_rate": 5e-06,
+      "loss": 0.3608,
+      "num_input_tokens_seen": 154349120,
+      "step": 893
+    },
+    {
+      "epoch": 0.3420145538108005,
+      "loss": 0.3488176465034485,
+      "loss_ce": 0.01996021717786789,
+      "loss_iou": 1.0872275829315186,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 154349120,
+      "step": 893
+    },
+    {
+      "epoch": 0.3423975488318652,
+      "grad_norm": 48.137410809081906,
+      "learning_rate": 5e-06,
+      "loss": 0.2708,
+      "num_input_tokens_seen": 154522336,
+      "step": 894
+    },
+    {
+      "epoch": 0.3423975488318652,
+      "loss": 0.30305665731430054,
+      "loss_ce": 0.01973147690296173,
+      "loss_iou": 1.0144085884094238,
+      "loss_num": 0.283203125,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 154522336,
+      "step": 894
+    },
+    {
+      "epoch": 0.3427805438529299,
+      "grad_norm": 38.15102960082304,
+      "learning_rate": 5e-06,
+      "loss": 0.2898,
+      "num_input_tokens_seen": 154695152,
+      "step": 895
+    },
+    {
+      "epoch": 0.3427805438529299,
+      "loss": 0.3265007734298706,
+      "loss_ce": 0.019371863454580307,
+      "loss_iou": 1.1647114753723145,
+      "loss_num": 0.306640625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 154695152,
+      "step": 895
+    },
+    {
+      "epoch": 0.34316353887399464,
+      "grad_norm": 39.93498885433648,
+      "learning_rate": 5e-06,
+      "loss": 0.247,
+      "num_input_tokens_seen": 154868480,
+      "step": 896
+    },
+    {
+      "epoch": 0.34316353887399464,
+      "loss": 0.2490977942943573,
+      "loss_ce": 0.019605603069067,
+      "loss_iou": 1.0045669078826904,
+      "loss_num": 0.2294921875,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 154868480,
+      "step": 896
+    },
+    {
+      "epoch": 0.34354653389505935,
+      "grad_norm": 49.53133487615153,
+      "learning_rate": 5e-06,
+      "loss": 0.3085,
+      "num_input_tokens_seen": 155041736,
+      "step": 897
+    },
+    {
+      "epoch": 0.34354653389505935,
+      "loss": 0.28766751289367676,
+      "loss_ce": 0.0194790530949831,
+      "loss_iou": 1.0107684135437012,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 155041736,
+      "step": 897
+    },
+    {
+      "epoch": 0.3439295289161241,
+      "grad_norm": 43.41462576045239,
+      "learning_rate": 5e-06,
+      "loss": 0.2817,
+      "num_input_tokens_seen": 155214920,
+      "step": 898
+    },
+    {
+      "epoch": 0.3439295289161241,
+      "loss": 0.29736804962158203,
+      "loss_ce": 0.020695675164461136,
+      "loss_iou": 1.0137401819229126,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 155214920,
+      "step": 898
+    },
+    {
+      "epoch": 0.3443125239371888,
+      "grad_norm": 29.071497778161586,
+      "learning_rate": 5e-06,
+      "loss": 0.2676,
+      "num_input_tokens_seen": 155387760,
+      "step": 899
+    },
+    {
+      "epoch": 0.3443125239371888,
+      "loss": 0.2834978699684143,
+      "loss_ce": 0.019337717443704605,
+      "loss_iou": 1.027559518814087,
+      "loss_num": 0.263671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 155387760,
+      "step": 899
+    },
+    {
+      "epoch": 0.34469551895825357,
+      "grad_norm": 40.99302143203642,
+      "learning_rate": 5e-06,
+      "loss": 0.3124,
+      "num_input_tokens_seen": 155557064,
+      "step": 900
+    },
+    {
+      "epoch": 0.34469551895825357,
+      "loss": 0.33803874254226685,
+      "loss_ce": 0.019801409915089607,
+      "loss_iou": 1.0807641744613647,
+      "loss_num": 0.318359375,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 155557064,
+      "step": 900
+    },
+    {
+      "epoch": 0.34507851397931827,
+      "grad_norm": 42.64613280035388,
+      "learning_rate": 5e-06,
+      "loss": 0.3903,
+      "num_input_tokens_seen": 155729992,
+      "step": 901
+    },
+    {
+      "epoch": 0.34507851397931827,
+      "loss": 0.46559372544288635,
+      "loss_ce": 0.020037095993757248,
+      "loss_iou": 1.1090466976165771,
+      "loss_num": 0.4453125,
+      "loss_xval": 0.4453125,
+      "num_input_tokens_seen": 155729992,
+      "step": 901
+    },
+    {
+      "epoch": 0.34546150900038297,
+      "grad_norm": 25.745717808459887,
+      "learning_rate": 5e-06,
+      "loss": 0.2969,
+      "num_input_tokens_seen": 155902824,
+      "step": 902
+    },
+    {
+      "epoch": 0.34546150900038297,
+      "loss": 0.3537680506706238,
+      "loss_ce": 0.01917332038283348,
+      "loss_iou": 1.0919166803359985,
+      "loss_num": 0.333984375,
+      "loss_xval": 0.333984375,
+      "num_input_tokens_seen": 155902824,
+      "step": 902
+    },
+    {
+      "epoch": 0.34584450402144773,
+      "grad_norm": 38.51981430922358,
+      "learning_rate": 5e-06,
+      "loss": 0.3012,
+      "num_input_tokens_seen": 156075952,
+      "step": 903
+    },
+    {
+      "epoch": 0.34584450402144773,
+      "loss": 0.31221455335617065,
+      "loss_ce": 0.02022237703204155,
+      "loss_iou": 1.0048507452011108,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 156075952,
+      "step": 903
+    },
+    {
+      "epoch": 0.34622749904251243,
+      "grad_norm": 42.086478438408925,
+      "learning_rate": 5e-06,
+      "loss": 0.3136,
+      "num_input_tokens_seen": 156248960,
+      "step": 904
+    },
+    {
+      "epoch": 0.34622749904251243,
+      "loss": 0.3787992000579834,
+      "loss_ce": 0.020522834733128548,
+      "loss_iou": 1.0666258335113525,
+      "loss_num": 0.357421875,
+      "loss_xval": 0.357421875,
+      "num_input_tokens_seen": 156248960,
+      "step": 904
+    },
+    {
+      "epoch": 0.3466104940635772,
+      "grad_norm": 31.18795904043135,
+      "learning_rate": 5e-06,
+      "loss": 0.3417,
+      "num_input_tokens_seen": 156422032,
+      "step": 905
+    },
+    {
+      "epoch": 0.3466104940635772,
+      "loss": 0.3480222821235657,
+      "loss_ce": 0.01892072893679142,
+      "loss_iou": 1.045752763748169,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 156422032,
+      "step": 905
+    },
+    {
+      "epoch": 0.3469934890846419,
+      "grad_norm": 19.604571605873183,
+      "learning_rate": 5e-06,
+      "loss": 0.2438,
+      "num_input_tokens_seen": 156594896,
+      "step": 906
+    },
+    {
+      "epoch": 0.3469934890846419,
+      "loss": 0.2784842252731323,
+      "loss_ce": 0.019878283143043518,
+      "loss_iou": 1.0400936603546143,
+      "loss_num": 0.2578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 156594896,
+      "step": 906
+    },
+    {
+      "epoch": 0.34737648410570665,
+      "grad_norm": 27.065435221215285,
+      "learning_rate": 5e-06,
+      "loss": 0.3149,
+      "num_input_tokens_seen": 156767496,
+      "step": 907
+    },
+    {
+      "epoch": 0.34737648410570665,
+      "loss": 0.3503131866455078,
+      "loss_ce": 0.02182195335626602,
+      "loss_iou": 1.635839819908142,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 156767496,
+      "step": 907
+    },
+    {
+      "epoch": 0.34775947912677135,
+      "grad_norm": 34.892979096375406,
+      "learning_rate": 5e-06,
+      "loss": 0.2396,
+      "num_input_tokens_seen": 156940544,
+      "step": 908
+    },
+    {
+      "epoch": 0.34775947912677135,
+      "loss": 0.2266232967376709,
+      "loss_ce": 0.020629655569791794,
+      "loss_iou": 1.0288121700286865,
+      "loss_num": 0.2060546875,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 156940544,
+      "step": 908
+    },
+    {
+      "epoch": 0.34814247414783606,
+      "grad_norm": 37.6920864330686,
+      "learning_rate": 5e-06,
+      "loss": 0.2843,
+      "num_input_tokens_seen": 157113304,
+      "step": 909
+    },
+    {
+      "epoch": 0.34814247414783606,
+      "loss": 0.29115408658981323,
+      "loss_ce": 0.020402126014232635,
+      "loss_iou": 1.0215134620666504,
+      "loss_num": 0.271484375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 157113304,
+      "step": 909
+    },
+    {
+      "epoch": 0.3485254691689008,
+      "grad_norm": 33.315665964062156,
+      "learning_rate": 5e-06,
+      "loss": 0.2281,
+      "num_input_tokens_seen": 157286088,
+      "step": 910
+    },
+    {
+      "epoch": 0.3485254691689008,
+      "loss": 0.2402557134628296,
+      "loss_ce": 0.01888120174407959,
+      "loss_iou": 1.0174598693847656,
+      "loss_num": 0.2216796875,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 157286088,
+      "step": 910
+    },
+    {
+      "epoch": 0.3489084641899655,
+      "grad_norm": 28.743804817260262,
+      "learning_rate": 5e-06,
+      "loss": 0.3037,
+      "num_input_tokens_seen": 157459064,
+      "step": 911
+    },
+    {
+      "epoch": 0.3489084641899655,
+      "loss": 0.24579393863677979,
+      "loss_ce": 0.019231446087360382,
+      "loss_iou": 1.5633665323257446,
+      "loss_num": 0.2265625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 157459064,
+      "step": 911
+    },
+    {
+      "epoch": 0.3492914592110303,
+      "grad_norm": 32.52219973162685,
+      "learning_rate": 5e-06,
+      "loss": 0.2491,
+      "num_input_tokens_seen": 157631888,
+      "step": 912
+    },
+    {
+      "epoch": 0.3492914592110303,
+      "loss": 0.2055998146533966,
+      "loss_ce": 0.020419150590896606,
+      "loss_iou": 1.0066945552825928,
+      "loss_num": 0.185546875,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 157631888,
+      "step": 912
+    },
+    {
+      "epoch": 0.349674454232095,
+      "grad_norm": 43.165692368411264,
+      "learning_rate": 5e-06,
+      "loss": 0.246,
+      "num_input_tokens_seen": 157801072,
+      "step": 913
+    },
+    {
+      "epoch": 0.349674454232095,
+      "loss": 0.2419818788766861,
+      "loss_ce": 0.019142530858516693,
+      "loss_iou": 1.0147067308425903,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 157801072,
+      "step": 913
+    },
+    {
+      "epoch": 0.35005744925315974,
+      "grad_norm": 39.8351373397396,
+      "learning_rate": 5e-06,
+      "loss": 0.2307,
+      "num_input_tokens_seen": 157973896,
+      "step": 914
+    },
+    {
+      "epoch": 0.35005744925315974,
+      "loss": 0.23154376447200775,
+      "loss_ce": 0.01853107661008835,
+      "loss_iou": 1.0227255821228027,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 157973896,
+      "step": 914
+    },
+    {
+      "epoch": 0.35044044427422444,
+      "grad_norm": 30.211194241758076,
+      "learning_rate": 5e-06,
+      "loss": 0.2847,
+      "num_input_tokens_seen": 158147080,
+      "step": 915
+    },
+    {
+      "epoch": 0.35044044427422444,
+      "loss": 0.3091517686843872,
+      "loss_ce": 0.021309982985258102,
+      "loss_iou": 1.5687074661254883,
+      "loss_num": 0.287109375,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 158147080,
+      "step": 915
+    },
+    {
+      "epoch": 0.35082343929528914,
+      "grad_norm": 37.68092787773169,
+      "learning_rate": 5e-06,
+      "loss": 0.3188,
+      "num_input_tokens_seen": 158320312,
+      "step": 916
+    },
+    {
+      "epoch": 0.35082343929528914,
+      "loss": 0.21438775956630707,
+      "loss_ce": 0.020173899829387665,
+      "loss_iou": 1.0433349609375,
+      "loss_num": 0.1943359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 158320312,
+      "step": 916
+    },
+    {
+      "epoch": 0.3512064343163539,
+      "grad_norm": 44.983632818646086,
+      "learning_rate": 5e-06,
+      "loss": 0.2567,
+      "num_input_tokens_seen": 158493192,
+      "step": 917
+    },
+    {
+      "epoch": 0.3512064343163539,
+      "loss": 0.2533189654350281,
+      "loss_ce": 0.020469840615987778,
+      "loss_iou": 1.1834526062011719,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 158493192,
+      "step": 917
+    },
+    {
+      "epoch": 0.3515894293374186,
+      "grad_norm": 32.39357558877045,
+      "learning_rate": 5e-06,
+      "loss": 0.2413,
+      "num_input_tokens_seen": 158666176,
+      "step": 918
+    },
+    {
+      "epoch": 0.3515894293374186,
+      "loss": 0.17894810438156128,
+      "loss_ce": 0.01915806531906128,
+      "loss_iou": 1.0037028789520264,
+      "loss_num": 0.16015625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 158666176,
+      "step": 918
+    },
+    {
+      "epoch": 0.35197242435848336,
+      "grad_norm": 27.62549462577862,
+      "learning_rate": 5e-06,
+      "loss": 0.3017,
+      "num_input_tokens_seen": 158839064,
+      "step": 919
+    },
+    {
+      "epoch": 0.35197242435848336,
+      "loss": 0.3494510352611542,
+      "loss_ce": 0.01851840503513813,
+      "loss_iou": 1.0774567127227783,
+      "loss_num": 0.330078125,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 158839064,
+      "step": 919
+    },
+    {
+      "epoch": 0.35235541937954806,
+      "grad_norm": 26.037858753567132,
+      "learning_rate": 5e-06,
+      "loss": 0.2658,
+      "num_input_tokens_seen": 159012456,
+      "step": 920
+    },
+    {
+      "epoch": 0.35235541937954806,
+      "loss": 0.2600114047527313,
+      "loss_ce": 0.02099773660302162,
+      "loss_iou": 1.0828306674957275,
+      "loss_num": 0.2392578125,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 159012456,
+      "step": 920
+    },
+    {
+      "epoch": 0.35273841440061277,
+      "grad_norm": 33.2862942978842,
+      "learning_rate": 5e-06,
+      "loss": 0.2344,
+      "num_input_tokens_seen": 159185584,
+      "step": 921
+    },
+    {
+      "epoch": 0.35273841440061277,
+      "loss": 0.24772392213344574,
+      "loss_ce": 0.018842089921236038,
+      "loss_iou": 1.0024566650390625,
+      "loss_num": 0.228515625,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 159185584,
+      "step": 921
+    },
+    {
+      "epoch": 0.3531214094216775,
+      "grad_norm": 42.377583632534844,
+      "learning_rate": 5e-06,
+      "loss": 0.2958,
+      "num_input_tokens_seen": 159358416,
+      "step": 922
+    },
+    {
+      "epoch": 0.3531214094216775,
+      "loss": 0.27739599347114563,
+      "loss_ce": 0.01958349347114563,
+      "loss_iou": 1.0809643268585205,
+      "loss_num": 0.2578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 159358416,
+      "step": 922
+    },
+    {
+      "epoch": 0.3535044044427422,
+      "grad_norm": 42.38230605939041,
+      "learning_rate": 5e-06,
+      "loss": 0.3198,
+      "num_input_tokens_seen": 159531552,
+      "step": 923
+    },
+    {
+      "epoch": 0.3535044044427422,
+      "loss": 0.34952735900878906,
+      "loss_ce": 0.021219246089458466,
+      "loss_iou": 1.0483672618865967,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 159531552,
+      "step": 923
+    },
+    {
+      "epoch": 0.353887399463807,
+      "grad_norm": 31.06636704009796,
+      "learning_rate": 5e-06,
+      "loss": 0.2219,
+      "num_input_tokens_seen": 159704552,
+      "step": 924
+    },
+    {
+      "epoch": 0.353887399463807,
+      "loss": 0.21600431203842163,
+      "loss_ce": 0.01941007934510708,
+      "loss_iou": 1.0015068054199219,
+      "loss_num": 0.1962890625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 159704552,
+      "step": 924
+    },
+    {
+      "epoch": 0.3542703944848717,
+      "grad_norm": 27.49356051820737,
+      "learning_rate": 5e-06,
+      "loss": 0.2664,
+      "num_input_tokens_seen": 159877752,
+      "step": 925
+    },
+    {
+      "epoch": 0.3542703944848717,
+      "loss": 0.21517300605773926,
+      "loss_ce": 0.01973842643201351,
+      "loss_iou": 1.0043420791625977,
+      "loss_num": 0.1953125,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 159877752,
+      "step": 925
+    },
+    {
+      "epoch": 0.35465338950593644,
+      "grad_norm": 26.41876018406636,
+      "learning_rate": 5e-06,
+      "loss": 0.2788,
+      "num_input_tokens_seen": 160051032,
+      "step": 926
+    },
+    {
+      "epoch": 0.35465338950593644,
+      "loss": 0.2924797534942627,
+      "loss_ce": 0.0201408751308918,
+      "loss_iou": 1.050068974494934,
+      "loss_num": 0.271484375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 160051032,
+      "step": 926
+    },
+    {
+      "epoch": 0.35503638452700115,
+      "grad_norm": 29.833120520130493,
+      "learning_rate": 5e-06,
+      "loss": 0.3168,
+      "num_input_tokens_seen": 160224368,
+      "step": 927
+    },
+    {
+      "epoch": 0.35503638452700115,
+      "loss": 0.2220856249332428,
+      "loss_ce": 0.02042545937001705,
+      "loss_iou": 1.002797245979309,
+      "loss_num": 0.201171875,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 160224368,
+      "step": 927
+    },
+    {
+      "epoch": 0.35541937954806585,
+      "grad_norm": 27.609107074489003,
+      "learning_rate": 5e-06,
+      "loss": 0.2503,
+      "num_input_tokens_seen": 160397280,
+      "step": 928
+    },
+    {
+      "epoch": 0.35541937954806585,
+      "loss": 0.2418726682662964,
+      "loss_ce": 0.02049814537167549,
+      "loss_iou": 1.1102688312530518,
+      "loss_num": 0.2216796875,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 160397280,
+      "step": 928
+    },
+    {
+      "epoch": 0.3558023745691306,
+      "grad_norm": 29.57272954416723,
+      "learning_rate": 5e-06,
+      "loss": 0.2923,
+      "num_input_tokens_seen": 160569992,
+      "step": 929
+    },
+    {
+      "epoch": 0.3558023745691306,
+      "loss": 0.2873765826225281,
+      "loss_ce": 0.02004261687397957,
+      "loss_iou": 1.100219964981079,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 160569992,
+      "step": 929
+    },
+    {
+      "epoch": 0.3561853695901953,
+      "grad_norm": 29.217580457996434,
+      "learning_rate": 5e-06,
+      "loss": 0.2273,
+      "num_input_tokens_seen": 160743128,
+      "step": 930
+    },
+    {
+      "epoch": 0.3561853695901953,
+      "loss": 0.22324742376804352,
+      "loss_ce": 0.019634131342172623,
+      "loss_iou": 1.0148777961730957,
+      "loss_num": 0.203125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 160743128,
+      "step": 930
+    },
+    {
+      "epoch": 0.35656836461126007,
+      "grad_norm": 36.5321377163354,
+      "learning_rate": 5e-06,
+      "loss": 0.3072,
+      "num_input_tokens_seen": 160916272,
+      "step": 931
+    },
+    {
+      "epoch": 0.35656836461126007,
+      "loss": 0.2983270287513733,
+      "loss_ce": 0.021471569314599037,
+      "loss_iou": 1.0132410526275635,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 160916272,
+      "step": 931
+    },
+    {
+      "epoch": 0.35695135963232477,
+      "grad_norm": 27.108967710770894,
+      "learning_rate": 5e-06,
+      "loss": 0.176,
+      "num_input_tokens_seen": 161089152,
+      "step": 932
+    },
+    {
+      "epoch": 0.35695135963232477,
+      "loss": 0.18797306716442108,
+      "loss_ce": 0.019638102501630783,
+      "loss_iou": 1.0014984607696533,
+      "loss_num": 0.16796875,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 161089152,
+      "step": 932
+    },
+    {
+      "epoch": 0.35733435465338953,
+      "grad_norm": 24.419209023342862,
+      "learning_rate": 5e-06,
+      "loss": 0.2917,
+      "num_input_tokens_seen": 161262408,
+      "step": 933
+    },
+    {
+      "epoch": 0.35733435465338953,
+      "loss": 0.3132162094116211,
+      "loss_ce": 0.0211019366979599,
+      "loss_iou": 1.067854642868042,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 161262408,
+      "step": 933
+    },
+    {
+      "epoch": 0.35771734967445423,
+      "grad_norm": 32.067596818972596,
+      "learning_rate": 5e-06,
+      "loss": 0.2357,
+      "num_input_tokens_seen": 161435616,
+      "step": 934
+    },
+    {
+      "epoch": 0.35771734967445423,
+      "loss": 0.20871080458164215,
+      "loss_ce": 0.020478377118706703,
+      "loss_iou": 1.0352756977081299,
+      "loss_num": 0.1884765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 161435616,
+      "step": 934
+    },
+    {
+      "epoch": 0.35810034469551894,
+      "grad_norm": 36.608693997731365,
+      "learning_rate": 5e-06,
+      "loss": 0.2748,
+      "num_input_tokens_seen": 161608768,
+      "step": 935
+    },
+    {
+      "epoch": 0.35810034469551894,
+      "loss": 0.29362761974334717,
+      "loss_ce": 0.020190125331282616,
+      "loss_iou": 1.0156643390655518,
+      "loss_num": 0.2734375,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 161608768,
+      "step": 935
+    },
+    {
+      "epoch": 0.3584833397165837,
+      "grad_norm": 24.677787951327748,
+      "learning_rate": 5e-06,
+      "loss": 0.1886,
+      "num_input_tokens_seen": 161781680,
+      "step": 936
+    },
+    {
+      "epoch": 0.3584833397165837,
+      "loss": 0.17715074121952057,
+      "loss_ce": 0.020351428538560867,
+      "loss_iou": 0.9857987761497498,
+      "loss_num": 0.1572265625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 161781680,
+      "step": 936
+    },
+    {
+      "epoch": 0.3588663347376484,
+      "grad_norm": 22.374263056377668,
+      "learning_rate": 5e-06,
+      "loss": 0.305,
+      "num_input_tokens_seen": 161954624,
+      "step": 937
+    },
+    {
+      "epoch": 0.3588663347376484,
+      "loss": 0.21580612659454346,
+      "loss_ce": 0.01921188458800316,
+      "loss_iou": 0.886029839515686,
+      "loss_num": 0.1962890625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 161954624,
+      "step": 937
+    },
+    {
+      "epoch": 0.35924932975871315,
+      "grad_norm": 27.39624654170563,
+      "learning_rate": 5e-06,
+      "loss": 0.3217,
+      "num_input_tokens_seen": 162127632,
+      "step": 938
+    },
+    {
+      "epoch": 0.35924932975871315,
+      "loss": 0.32278984785079956,
+      "loss_ce": 0.020543742924928665,
+      "loss_iou": 1.0597132444381714,
+      "loss_num": 0.302734375,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 162127632,
+      "step": 938
+    },
+    {
+      "epoch": 0.35963232477977786,
+      "grad_norm": 35.36408821697388,
+      "learning_rate": 5e-06,
+      "loss": 0.3753,
+      "num_input_tokens_seen": 162300440,
+      "step": 939
+    },
+    {
+      "epoch": 0.35963232477977786,
+      "loss": 0.35825634002685547,
+      "loss_ce": 0.02134224772453308,
+      "loss_iou": 1.1178550720214844,
+      "loss_num": 0.3359375,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 162300440,
+      "step": 939
+    },
+    {
+      "epoch": 0.3600153198008426,
+      "grad_norm": 32.27612707323375,
+      "learning_rate": 5e-06,
+      "loss": 0.2525,
+      "num_input_tokens_seen": 162473368,
+      "step": 940
+    },
+    {
+      "epoch": 0.3600153198008426,
+      "loss": 0.2508542835712433,
+      "loss_ce": 0.018920686095952988,
+      "loss_iou": 1.0336934328079224,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 162473368,
+      "step": 940
+    },
+    {
+      "epoch": 0.3603983148219073,
+      "grad_norm": 26.328718140221437,
+      "learning_rate": 5e-06,
+      "loss": 0.3067,
+      "num_input_tokens_seen": 162646824,
+      "step": 941
+    },
+    {
+      "epoch": 0.3603983148219073,
+      "loss": 0.34090521931648254,
+      "loss_ce": 0.021813416853547096,
+      "loss_iou": 1.027925729751587,
+      "loss_num": 0.318359375,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 162646824,
+      "step": 941
+    },
+    {
+      "epoch": 0.360781309842972,
+      "grad_norm": 36.2819084239107,
+      "learning_rate": 5e-06,
+      "loss": 0.2025,
+      "num_input_tokens_seen": 162819664,
+      "step": 942
+    },
+    {
+      "epoch": 0.360781309842972,
+      "loss": 0.20151524245738983,
+      "loss_ce": 0.02054600790143013,
+      "loss_iou": 1.0331640243530273,
+      "loss_num": 0.1806640625,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 162819664,
+      "step": 942
+    },
+    {
+      "epoch": 0.3611643048640368,
+      "grad_norm": 45.38164385254464,
+      "learning_rate": 5e-06,
+      "loss": 0.2846,
+      "num_input_tokens_seen": 162992528,
+      "step": 943
+    },
+    {
+      "epoch": 0.3611643048640368,
+      "loss": 0.2944408357143402,
+      "loss_ce": 0.01892814412713051,
+      "loss_iou": 1.0125874280929565,
+      "loss_num": 0.275390625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 162992528,
+      "step": 943
+    },
+    {
+      "epoch": 0.3615472998851015,
+      "grad_norm": 25.859829660421223,
+      "learning_rate": 5e-06,
+      "loss": 0.2095,
+      "num_input_tokens_seen": 163165904,
+      "step": 944
+    },
+    {
+      "epoch": 0.3615472998851015,
+      "loss": 0.25529569387435913,
+      "loss_ce": 0.019242238253355026,
+      "loss_iou": 1.001347303390503,
+      "loss_num": 0.236328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 163165904,
+      "step": 944
+    },
+    {
+      "epoch": 0.36193029490616624,
+      "grad_norm": 28.6788539541314,
+      "learning_rate": 5e-06,
+      "loss": 0.3256,
+      "num_input_tokens_seen": 163339152,
+      "step": 945
+    },
+    {
+      "epoch": 0.36193029490616624,
+      "loss": 0.3408098816871643,
+      "loss_ce": 0.019154615700244904,
+      "loss_iou": 1.025606393814087,
+      "loss_num": 0.322265625,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 163339152,
+      "step": 945
+    },
+    {
+      "epoch": 0.36231328992723094,
+      "grad_norm": 21.10913626917847,
+      "learning_rate": 5e-06,
+      "loss": 0.2721,
+      "num_input_tokens_seen": 163511832,
+      "step": 946
+    },
+    {
+      "epoch": 0.36231328992723094,
+      "loss": 0.259053111076355,
+      "loss_ce": 0.020161528140306473,
+      "loss_iou": 1.0623010396957397,
+      "loss_num": 0.2392578125,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 163511832,
+      "step": 946
+    },
+    {
+      "epoch": 0.3626962849482957,
+      "grad_norm": 44.33962548458138,
+      "learning_rate": 5e-06,
+      "loss": 0.2804,
+      "num_input_tokens_seen": 163684696,
+      "step": 947
+    },
+    {
+      "epoch": 0.3626962849482957,
+      "loss": 0.27441030740737915,
+      "loss_ce": 0.02087029069662094,
+      "loss_iou": 1.0241224765777588,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 163684696,
+      "step": 947
+    },
+    {
+      "epoch": 0.3630792799693604,
+      "grad_norm": 51.65153418796951,
+      "learning_rate": 5e-06,
+      "loss": 0.2737,
+      "num_input_tokens_seen": 163857912,
+      "step": 948
+    },
+    {
+      "epoch": 0.3630792799693604,
+      "loss": 0.291856974363327,
+      "loss_ce": 0.02025051787495613,
+      "loss_iou": 1.0185136795043945,
+      "loss_num": 0.271484375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 163857912,
+      "step": 948
+    },
+    {
+      "epoch": 0.3634622749904251,
+      "grad_norm": 29.070846081236027,
+      "learning_rate": 5e-06,
+      "loss": 0.2419,
+      "num_input_tokens_seen": 164031016,
+      "step": 949
+    },
+    {
+      "epoch": 0.3634622749904251,
+      "loss": 0.26171785593032837,
+      "loss_ce": 0.019286222755908966,
+      "loss_iou": 1.0417400598526,
+      "loss_num": 0.2421875,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 164031016,
+      "step": 949
+    },
+    {
+      "epoch": 0.36384527001148986,
+      "grad_norm": 23.19025535792982,
+      "learning_rate": 5e-06,
+      "loss": 0.2316,
+      "num_input_tokens_seen": 164203672,
+      "step": 950
+    },
+    {
+      "epoch": 0.36384527001148986,
+      "loss": 0.23933814465999603,
+      "loss_ce": 0.01802467182278633,
+      "loss_iou": 1.0510921478271484,
+      "loss_num": 0.2216796875,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 164203672,
+      "step": 950
+    },
+    {
+      "epoch": 0.36422826503255457,
+      "grad_norm": 38.49785846617074,
+      "learning_rate": 5e-06,
+      "loss": 0.2897,
+      "num_input_tokens_seen": 164376880,
+      "step": 951
+    },
+    {
+      "epoch": 0.36422826503255457,
+      "loss": 0.28677451610565186,
+      "loss_ce": 0.019684676080942154,
+      "loss_iou": 1.0165079832077026,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 164376880,
+      "step": 951
+    },
+    {
+      "epoch": 0.3646112600536193,
+      "grad_norm": 31.328935383781246,
+      "learning_rate": 5e-06,
+      "loss": 0.3101,
+      "num_input_tokens_seen": 164550000,
+      "step": 952
+    },
+    {
+      "epoch": 0.3646112600536193,
+      "loss": 0.3002566993236542,
+      "loss_ce": 0.018640486523509026,
+      "loss_iou": 1.0196378231048584,
+      "loss_num": 0.28125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 164550000,
+      "step": 952
+    },
+    {
+      "epoch": 0.364994255074684,
+      "grad_norm": 31.244400096108578,
+      "learning_rate": 5e-06,
+      "loss": 0.3367,
+      "num_input_tokens_seen": 164722936,
+      "step": 953
+    },
+    {
+      "epoch": 0.364994255074684,
+      "loss": 0.29647910594940186,
+      "loss_ce": 0.02023400366306305,
+      "loss_iou": 1.0197391510009766,
+      "loss_num": 0.275390625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 164722936,
+      "step": 953
+    },
+    {
+      "epoch": 0.36537725009574873,
+      "grad_norm": 36.71793962885349,
+      "learning_rate": 5e-06,
+      "loss": 0.2579,
+      "num_input_tokens_seen": 164896328,
+      "step": 954
+    },
+    {
+      "epoch": 0.36537725009574873,
+      "loss": 0.262016624212265,
+      "loss_ce": 0.02129395306110382,
+      "loss_iou": 1.035705804824829,
+      "loss_num": 0.240234375,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 164896328,
+      "step": 954
+    },
+    {
+      "epoch": 0.3657602451168135,
+      "grad_norm": 37.574584377052794,
+      "learning_rate": 5e-06,
+      "loss": 0.3782,
+      "num_input_tokens_seen": 165069352,
+      "step": 955
+    },
+    {
+      "epoch": 0.3657602451168135,
+      "loss": 0.4396094083786011,
+      "loss_ce": 0.020053725689649582,
+      "loss_iou": 3.2485127449035645,
+      "loss_num": 0.419921875,
+      "loss_xval": 0.419921875,
+      "num_input_tokens_seen": 165069352,
+      "step": 955
+    },
+    {
+      "epoch": 0.3661432401378782,
+      "grad_norm": 34.42679518876968,
+      "learning_rate": 5e-06,
+      "loss": 0.3147,
+      "num_input_tokens_seen": 165242528,
+      "step": 956
+    },
+    {
+      "epoch": 0.3661432401378782,
+      "loss": 0.2801949381828308,
+      "loss_ce": 0.02042931132018566,
+      "loss_iou": 1.007472276687622,
+      "loss_num": 0.259765625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 165242528,
+      "step": 956
+    },
+    {
+      "epoch": 0.36652623515894295,
+      "grad_norm": 27.517040916418587,
+      "learning_rate": 5e-06,
+      "loss": 0.3127,
+      "num_input_tokens_seen": 165411808,
+      "step": 957
+    },
+    {
+      "epoch": 0.36652623515894295,
+      "loss": 0.29710865020751953,
+      "loss_ce": 0.019764892756938934,
+      "loss_iou": 1.1402711868286133,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 165411808,
+      "step": 957
+    },
+    {
+      "epoch": 0.36690923018000765,
+      "grad_norm": 25.34036629920117,
+      "learning_rate": 5e-06,
+      "loss": 0.2089,
+      "num_input_tokens_seen": 165584792,
+      "step": 958
+    },
+    {
+      "epoch": 0.36690923018000765,
+      "loss": 0.19531746208667755,
+      "loss_ce": 0.019475182518363,
+      "loss_iou": 1.0212349891662598,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 165584792,
+      "step": 958
+    },
+    {
+      "epoch": 0.3672922252010724,
+      "grad_norm": 31.0376230069948,
+      "learning_rate": 5e-06,
+      "loss": 0.3084,
+      "num_input_tokens_seen": 165757536,
+      "step": 959
+    },
+    {
+      "epoch": 0.3672922252010724,
+      "loss": 0.23202432692050934,
+      "loss_ce": 0.01901163160800934,
+      "loss_iou": 1.0255075693130493,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 165757536,
+      "step": 959
+    },
+    {
+      "epoch": 0.3676752202221371,
+      "grad_norm": 32.45066158774197,
+      "learning_rate": 5e-06,
+      "loss": 0.1919,
+      "num_input_tokens_seen": 165930696,
+      "step": 960
+    },
+    {
+      "epoch": 0.3676752202221371,
+      "loss": 0.18574444949626923,
+      "loss_ce": 0.019423654302954674,
+      "loss_iou": 1.0056147575378418,
+      "loss_num": 0.166015625,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 165930696,
+      "step": 960
+    },
+    {
+      "epoch": 0.3680582152432018,
+      "grad_norm": 24.452173233060154,
+      "learning_rate": 5e-06,
+      "loss": 0.2577,
+      "num_input_tokens_seen": 166103496,
+      "step": 961
+    },
+    {
+      "epoch": 0.3680582152432018,
+      "loss": 0.2453029602766037,
+      "loss_ce": 0.019167695194482803,
+      "loss_iou": 1.0065820217132568,
+      "loss_num": 0.2265625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 166103496,
+      "step": 961
+    },
+    {
+      "epoch": 0.36844121026426657,
+      "grad_norm": 31.50527693117424,
+      "learning_rate": 5e-06,
+      "loss": 0.196,
+      "num_input_tokens_seen": 166276544,
+      "step": 962
+    },
+    {
+      "epoch": 0.36844121026426657,
+      "loss": 0.18020428717136383,
+      "loss_ce": 0.02047528699040413,
+      "loss_iou": 1.0018154382705688,
+      "loss_num": 0.16015625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 166276544,
+      "step": 962
+    },
+    {
+      "epoch": 0.3688242052853313,
+      "grad_norm": 35.310253666246936,
+      "learning_rate": 5e-06,
+      "loss": 0.236,
+      "num_input_tokens_seen": 166449960,
+      "step": 963
+    },
+    {
+      "epoch": 0.3688242052853313,
+      "loss": 0.2314654141664505,
+      "loss_ce": 0.020955156534910202,
+      "loss_iou": 1.0441038608551025,
+      "loss_num": 0.2109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 166449960,
+      "step": 963
+    },
+    {
+      "epoch": 0.36920720030639603,
+      "grad_norm": 34.72278989600349,
+      "learning_rate": 5e-06,
+      "loss": 0.2881,
+      "num_input_tokens_seen": 166623200,
+      "step": 964
+    },
+    {
+      "epoch": 0.36920720030639603,
+      "loss": 0.32761016488075256,
+      "loss_ce": 0.01919952780008316,
+      "loss_iou": 1.1988705396652222,
+      "loss_num": 0.30859375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 166623200,
+      "step": 964
+    },
+    {
+      "epoch": 0.36959019532746074,
+      "grad_norm": 35.99209920451824,
+      "learning_rate": 5e-06,
+      "loss": 0.2629,
+      "num_input_tokens_seen": 166796008,
+      "step": 965
+    },
+    {
+      "epoch": 0.36959019532746074,
+      "loss": 0.2871081233024597,
+      "loss_ce": 0.01953001320362091,
+      "loss_iou": 1.0272150039672852,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 166796008,
+      "step": 965
+    },
+    {
+      "epoch": 0.3699731903485255,
+      "grad_norm": 34.54876742428441,
+      "learning_rate": 5e-06,
+      "loss": 0.224,
+      "num_input_tokens_seen": 166968864,
+      "step": 966
+    },
+    {
+      "epoch": 0.3699731903485255,
+      "loss": 0.23276183009147644,
+      "loss_ce": 0.02017638087272644,
+      "loss_iou": 1.0384440422058105,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 166968864,
+      "step": 966
+    },
+    {
+      "epoch": 0.3703561853695902,
+      "grad_norm": 31.537146617666913,
+      "learning_rate": 5e-06,
+      "loss": 0.2377,
+      "num_input_tokens_seen": 167142048,
+      "step": 967
+    },
+    {
+      "epoch": 0.3703561853695902,
+      "loss": 0.2386997640132904,
+      "loss_ce": 0.018484920263290405,
+      "loss_iou": 1.0021443367004395,
+      "loss_num": 0.220703125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 167142048,
+      "step": 967
+    },
+    {
+      "epoch": 0.3707391803906549,
+      "grad_norm": 28.718798089964558,
+      "learning_rate": 5e-06,
+      "loss": 0.2094,
+      "num_input_tokens_seen": 167315016,
+      "step": 968
+    },
+    {
+      "epoch": 0.3707391803906549,
+      "loss": 0.1817106306552887,
+      "loss_ce": 0.019662294536828995,
+      "loss_iou": 1.0258190631866455,
+      "loss_num": 0.162109375,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 167315016,
+      "step": 968
+    },
+    {
+      "epoch": 0.37112217541171966,
+      "grad_norm": 40.4887294725314,
+      "learning_rate": 5e-06,
+      "loss": 0.2375,
+      "num_input_tokens_seen": 167488120,
+      "step": 969
+    },
+    {
+      "epoch": 0.37112217541171966,
+      "loss": 0.2075423300266266,
+      "loss_ce": 0.021629242226481438,
+      "loss_iou": 1.0055090188980103,
+      "loss_num": 0.185546875,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 167488120,
+      "step": 969
+    },
+    {
+      "epoch": 0.37150517043278436,
+      "grad_norm": 42.83308812336441,
+      "learning_rate": 5e-06,
+      "loss": 0.2509,
+      "num_input_tokens_seen": 167657296,
+      "step": 970
+    },
+    {
+      "epoch": 0.37150517043278436,
+      "loss": 0.24004191160202026,
+      "loss_ce": 0.021413978189229965,
+      "loss_iou": 1.0232372283935547,
+      "loss_num": 0.21875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 167657296,
+      "step": 970
+    },
+    {
+      "epoch": 0.3718881654538491,
+      "grad_norm": 27.347654952228595,
+      "learning_rate": 5e-06,
+      "loss": 0.3097,
+      "num_input_tokens_seen": 167830448,
+      "step": 971
+    },
+    {
+      "epoch": 0.3718881654538491,
+      "loss": 0.30894187092781067,
+      "loss_ce": 0.02036764845252037,
+      "loss_iou": 1.0451247692108154,
+      "loss_num": 0.2890625,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 167830448,
+      "step": 971
+    },
+    {
+      "epoch": 0.3722711604749138,
+      "grad_norm": 37.39125530343754,
+      "learning_rate": 5e-06,
+      "loss": 0.2068,
+      "num_input_tokens_seen": 168003512,
+      "step": 972
+    },
+    {
+      "epoch": 0.3722711604749138,
+      "loss": 0.2168152779340744,
+      "loss_ce": 0.018878266215324402,
+      "loss_iou": 1.0135071277618408,
+      "loss_num": 0.1982421875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 168003512,
+      "step": 972
+    },
+    {
+      "epoch": 0.3726541554959786,
+      "grad_norm": 49.908204652296874,
+      "learning_rate": 5e-06,
+      "loss": 0.2756,
+      "num_input_tokens_seen": 168176136,
+      "step": 973
+    },
+    {
+      "epoch": 0.3726541554959786,
+      "loss": 0.29072046279907227,
+      "loss_ce": 0.019480209797620773,
+      "loss_iou": 1.0204592943191528,
+      "loss_num": 0.271484375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 168176136,
+      "step": 973
+    },
+    {
+      "epoch": 0.3730371505170433,
+      "grad_norm": 31.16355749647373,
+      "learning_rate": 5e-06,
+      "loss": 0.2951,
+      "num_input_tokens_seen": 168349208,
+      "step": 974
+    },
+    {
+      "epoch": 0.3730371505170433,
+      "loss": 0.30966952443122864,
+      "loss_ce": 0.02072908729314804,
+      "loss_iou": 1.0145212411880493,
+      "loss_num": 0.2890625,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 168349208,
+      "step": 974
+    },
+    {
+      "epoch": 0.373420145538108,
+      "grad_norm": 25.088461228350013,
+      "learning_rate": 5e-06,
+      "loss": 0.3297,
+      "num_input_tokens_seen": 168522392,
+      "step": 975
+    },
+    {
+      "epoch": 0.373420145538108,
+      "loss": 0.3033157289028168,
+      "loss_ce": 0.01999053731560707,
+      "loss_iou": 1.001996636390686,
+      "loss_num": 0.283203125,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 168522392,
+      "step": 975
+    },
+    {
+      "epoch": 0.37380314055917274,
+      "grad_norm": 40.386975520720455,
+      "learning_rate": 5e-06,
+      "loss": 0.3908,
+      "num_input_tokens_seen": 168695216,
+      "step": 976
+    },
+    {
+      "epoch": 0.37380314055917274,
+      "loss": 0.33824622631073,
+      "loss_ce": 0.019764790311455727,
+      "loss_iou": 1.0352516174316406,
+      "loss_num": 0.318359375,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 168695216,
+      "step": 976
+    },
+    {
+      "epoch": 0.37418613558023744,
+      "grad_norm": 57.71792347077378,
+      "learning_rate": 5e-06,
+      "loss": 0.4084,
+      "num_input_tokens_seen": 168868456,
+      "step": 977
+    },
+    {
+      "epoch": 0.37418613558023744,
+      "loss": 0.42148301005363464,
+      "loss_ce": 0.020115818828344345,
+      "loss_iou": 1.053954839706421,
+      "loss_num": 0.40234375,
+      "loss_xval": 0.40234375,
+      "num_input_tokens_seen": 168868456,
+      "step": 977
+    },
+    {
+      "epoch": 0.3745691306013022,
+      "grad_norm": 48.833072329454026,
+      "learning_rate": 5e-06,
+      "loss": 0.3615,
+      "num_input_tokens_seen": 169041200,
+      "step": 978
+    },
+    {
+      "epoch": 0.3745691306013022,
+      "loss": 0.3466894030570984,
+      "loss_ce": 0.022104421630501747,
+      "loss_iou": 1.0623598098754883,
+      "loss_num": 0.32421875,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 169041200,
+      "step": 978
+    },
+    {
+      "epoch": 0.3749521256223669,
+      "grad_norm": 37.36730018726415,
+      "learning_rate": 5e-06,
+      "loss": 0.4141,
+      "num_input_tokens_seen": 169213848,
+      "step": 979
+    },
+    {
+      "epoch": 0.3749521256223669,
+      "loss": 0.4149620234966278,
+      "loss_ce": 0.020186619833111763,
+      "loss_iou": 1.0112701654434204,
+      "loss_num": 0.39453125,
+      "loss_xval": 0.39453125,
+      "num_input_tokens_seen": 169213848,
+      "step": 979
+    },
+    {
+      "epoch": 0.3753351206434316,
+      "grad_norm": 46.6991870180064,
+      "learning_rate": 5e-06,
+      "loss": 0.4204,
+      "num_input_tokens_seen": 169386720,
+      "step": 980
+    },
+    {
+      "epoch": 0.3753351206434316,
+      "loss": 0.4489254355430603,
+      "loss_ce": 0.02241174876689911,
+      "loss_iou": 1.0938462018966675,
+      "loss_num": 0.42578125,
+      "loss_xval": 0.42578125,
+      "num_input_tokens_seen": 169386720,
+      "step": 980
+    },
+    {
+      "epoch": 0.37571811566449637,
+      "grad_norm": 68.94776622161878,
+      "learning_rate": 5e-06,
+      "loss": 0.6308,
+      "num_input_tokens_seen": 169559888,
+      "step": 981
+    },
+    {
+      "epoch": 0.37571811566449637,
+      "loss": 0.6707226037979126,
+      "loss_ce": 0.021308545023202896,
+      "loss_iou": 1.3989322185516357,
+      "loss_num": 0.6484375,
+      "loss_xval": 0.6484375,
+      "num_input_tokens_seen": 169559888,
+      "step": 981
+    },
+    {
+      "epoch": 0.37610111068556107,
+      "grad_norm": 55.08751091480485,
+      "learning_rate": 5e-06,
+      "loss": 0.4322,
+      "num_input_tokens_seen": 169733104,
+      "step": 982
+    },
+    {
+      "epoch": 0.37610111068556107,
+      "loss": 0.42639780044555664,
+      "loss_ce": 0.020391955971717834,
+      "loss_iou": 1.0782774686813354,
+      "loss_num": 0.40625,
+      "loss_xval": 0.40625,
+      "num_input_tokens_seen": 169733104,
+      "step": 982
+    },
+    {
+      "epoch": 0.3764841057066258,
+      "grad_norm": 38.44895143776549,
+      "learning_rate": 5e-06,
+      "loss": 0.4994,
+      "num_input_tokens_seen": 169906048,
+      "step": 983
+    },
+    {
+      "epoch": 0.3764841057066258,
+      "loss": 0.4617816209793091,
+      "loss_ce": 0.018422234803438187,
+      "loss_iou": 0.9563711285591125,
+      "loss_num": 0.443359375,
+      "loss_xval": 0.443359375,
+      "num_input_tokens_seen": 169906048,
+      "step": 983
+    },
+    {
+      "epoch": 0.37686710072769053,
+      "grad_norm": 18.926519580396565,
+      "learning_rate": 5e-06,
+      "loss": 0.3627,
+      "num_input_tokens_seen": 170078736,
+      "step": 984
+    },
+    {
+      "epoch": 0.37686710072769053,
+      "loss": 0.32097262144088745,
+      "loss_ce": 0.019825173541903496,
+      "loss_iou": 1.0138651132583618,
+      "loss_num": 0.30078125,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 170078736,
+      "step": 984
+    },
+    {
+      "epoch": 0.3772500957487553,
+      "grad_norm": 13.845202092258038,
+      "learning_rate": 5e-06,
+      "loss": 0.2875,
+      "num_input_tokens_seen": 170252032,
+      "step": 985
+    },
+    {
+      "epoch": 0.3772500957487553,
+      "loss": 0.31729811429977417,
+      "loss_ce": 0.02091141790151596,
+      "loss_iou": 1.0532989501953125,
+      "loss_num": 0.296875,
+      "loss_xval": 0.296875,
+      "num_input_tokens_seen": 170252032,
+      "step": 985
+    },
+    {
+      "epoch": 0.37763309076982,
+      "grad_norm": 29.6947170824272,
+      "learning_rate": 5e-06,
+      "loss": 0.3102,
+      "num_input_tokens_seen": 170425024,
+      "step": 986
+    },
+    {
+      "epoch": 0.37763309076982,
+      "loss": 0.3302353024482727,
+      "loss_ce": 0.02335052751004696,
+      "loss_iou": 1.0863749980926514,
+      "loss_num": 0.306640625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 170425024,
+      "step": 986
+    },
+    {
+      "epoch": 0.3780160857908847,
+      "grad_norm": 12.067861650195477,
+      "learning_rate": 5e-06,
+      "loss": 0.2124,
+      "num_input_tokens_seen": 170598016,
+      "step": 987
+    },
+    {
+      "epoch": 0.3780160857908847,
+      "loss": 0.2587829530239105,
+      "loss_ce": 0.020379632711410522,
+      "loss_iou": 1.0715415477752686,
+      "loss_num": 0.23828125,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 170598016,
+      "step": 987
+    },
+    {
+      "epoch": 0.37839908081194945,
+      "grad_norm": 15.67376693155364,
+      "learning_rate": 5e-06,
+      "loss": 0.2562,
+      "num_input_tokens_seen": 170770912,
+      "step": 988
+    },
+    {
+      "epoch": 0.37839908081194945,
+      "loss": 0.24709764122962952,
+      "loss_ce": 0.020352039486169815,
+      "loss_iou": 1.096572756767273,
+      "loss_num": 0.2265625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 170770912,
+      "step": 988
+    },
+    {
+      "epoch": 0.37878207583301415,
+      "grad_norm": 32.01541760856049,
+      "learning_rate": 5e-06,
+      "loss": 0.2612,
+      "num_input_tokens_seen": 170943712,
+      "step": 989
+    },
+    {
+      "epoch": 0.37878207583301415,
+      "loss": 0.254817396402359,
+      "loss_ce": 0.020930690690875053,
+      "loss_iou": 1.0075466632843018,
+      "loss_num": 0.234375,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 170943712,
+      "step": 989
+    },
+    {
+      "epoch": 0.3791650708540789,
+      "grad_norm": 30.10256252096975,
+      "learning_rate": 5e-06,
+      "loss": 0.2557,
+      "num_input_tokens_seen": 171116320,
+      "step": 990
+    },
+    {
+      "epoch": 0.3791650708540789,
+      "loss": 0.24919039011001587,
+      "loss_ce": 0.021285109221935272,
+      "loss_iou": 1.0024375915527344,
+      "loss_num": 0.2275390625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 171116320,
+      "step": 990
+    },
+    {
+      "epoch": 0.3795480658751436,
+      "grad_norm": 36.33932731783155,
+      "learning_rate": 5e-06,
+      "loss": 0.2792,
+      "num_input_tokens_seen": 171289368,
+      "step": 991
+    },
+    {
+      "epoch": 0.3795480658751436,
+      "loss": 0.36623942852020264,
+      "loss_ce": 0.01846110261976719,
+      "loss_iou": 1.0178117752075195,
+      "loss_num": 0.34765625,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 171289368,
+      "step": 991
+    },
+    {
+      "epoch": 0.3799310608962084,
+      "grad_norm": 23.561074635829417,
+      "learning_rate": 5e-06,
+      "loss": 0.2706,
+      "num_input_tokens_seen": 171461888,
+      "step": 992
+    },
+    {
+      "epoch": 0.3799310608962084,
+      "loss": 0.22868096828460693,
+      "loss_ce": 0.021405581384897232,
+      "loss_iou": 1.0184024572372437,
+      "loss_num": 0.20703125,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 171461888,
+      "step": 992
+    },
+    {
+      "epoch": 0.3803140559172731,
+      "grad_norm": 69.65429230110702,
+      "learning_rate": 5e-06,
+      "loss": 0.2282,
+      "num_input_tokens_seen": 171634728,
+      "step": 993
+    },
+    {
+      "epoch": 0.3803140559172731,
+      "loss": 0.1933428943157196,
+      "loss_ce": 0.020674441009759903,
+      "loss_iou": 1.048812985420227,
+      "loss_num": 0.1728515625,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 171634728,
+      "step": 993
+    },
+    {
+      "epoch": 0.3806970509383378,
+      "grad_norm": 47.02556831642122,
+      "learning_rate": 5e-06,
+      "loss": 0.2991,
+      "num_input_tokens_seen": 171807632,
+      "step": 994
+    },
+    {
+      "epoch": 0.3806970509383378,
+      "loss": 0.25592872500419617,
+      "loss_ce": 0.02253028377890587,
+      "loss_iou": 1.0186436176300049,
+      "loss_num": 0.2333984375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 171807632,
+      "step": 994
+    },
+    {
+      "epoch": 0.38108004595940254,
+      "grad_norm": 26.723063321849086,
+      "learning_rate": 5e-06,
+      "loss": 0.2258,
+      "num_input_tokens_seen": 171980448,
+      "step": 995
+    },
+    {
+      "epoch": 0.38108004595940254,
+      "loss": 0.19038596749305725,
+      "loss_ce": 0.018755123019218445,
+      "loss_iou": 1.0027767419815063,
+      "loss_num": 0.171875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 171980448,
+      "step": 995
+    },
+    {
+      "epoch": 0.38146304098046724,
+      "grad_norm": 22.90353645456606,
+      "learning_rate": 5e-06,
+      "loss": 0.2788,
+      "num_input_tokens_seen": 172153312,
+      "step": 996
+    },
+    {
+      "epoch": 0.38146304098046724,
+      "loss": 0.357512503862381,
+      "loss_ce": 0.020720507949590683,
+      "loss_iou": 1.1017611026763916,
+      "loss_num": 0.3359375,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 172153312,
+      "step": 996
+    },
+    {
+      "epoch": 0.381846036001532,
+      "grad_norm": 38.48221669166512,
+      "learning_rate": 5e-06,
+      "loss": 0.2013,
+      "num_input_tokens_seen": 172326240,
+      "step": 997
+    },
+    {
+      "epoch": 0.381846036001532,
+      "loss": 0.18817144632339478,
+      "loss_ce": 0.01983649656176567,
+      "loss_iou": 1.0173728466033936,
+      "loss_num": 0.16796875,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 172326240,
+      "step": 997
+    },
+    {
+      "epoch": 0.3822290310225967,
+      "grad_norm": 36.67878110625945,
+      "learning_rate": 5e-06,
+      "loss": 0.2996,
+      "num_input_tokens_seen": 172499128,
+      "step": 998
+    },
+    {
+      "epoch": 0.3822290310225967,
+      "loss": 0.3369196355342865,
+      "loss_ce": 0.01880439557135105,
+      "loss_iou": 1.6968703269958496,
+      "loss_num": 0.318359375,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 172499128,
+      "step": 998
+    },
+    {
+      "epoch": 0.38261202604366146,
+      "grad_norm": 28.376835361028178,
+      "learning_rate": 5e-06,
+      "loss": 0.2075,
+      "num_input_tokens_seen": 172671712,
+      "step": 999
+    },
+    {
+      "epoch": 0.38261202604366146,
+      "loss": 0.23104539513587952,
+      "loss_ce": 0.020107891410589218,
+      "loss_iou": 1.0930787324905396,
+      "loss_num": 0.2109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 172671712,
+      "step": 999
+    },
+    {
+      "epoch": 0.38299502106472616,
+      "grad_norm": 34.65065629131397,
+      "learning_rate": 5e-06,
+      "loss": 0.2881,
+      "num_input_tokens_seen": 172844576,
+      "step": 1000
+    },
+    {
+      "epoch": 0.38299502106472616,
+      "eval_websight_new_CIoU": 0.88560551404953,
+      "eval_websight_new_GIoU": 0.885301798582077,
+      "eval_websight_new_IoU": 0.886048287153244,
+      "eval_websight_new_MAE_all": 0.011711293365806341,
+      "eval_websight_new_MAE_h": 0.012148695066571236,
+      "eval_websight_new_MAE_w": 0.015516783576458693,
+      "eval_websight_new_MAE_x": 0.011325123254209757,
+      "eval_websight_new_MAE_y": 0.00785457226447761,
+      "eval_websight_new_NUM_probability": 0.0001039152812154498,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.1705828160047531,
+      "eval_websight_new_loss_ce": 0.040730688720941544,
+      "eval_websight_new_loss_iou": 1.0006657838821411,
+      "eval_websight_new_loss_num": 0.13189697265625,
+      "eval_websight_new_loss_xval": 0.13189697265625,
+      "eval_websight_new_runtime": 60.6628,
+      "eval_websight_new_samples_per_second": 0.824,
+      "eval_websight_new_steps_per_second": 0.033,
+      "num_input_tokens_seen": 172844576,
+      "step": 1000
+    },
+    {
+      "epoch": 0.38299502106472616,
+      "eval_seeclick_CIoU": 0.6596595346927643,
+      "eval_seeclick_GIoU": 0.6570102870464325,
+      "eval_seeclick_IoU": 0.6813171207904816,
+      "eval_seeclick_MAE_all": 0.04912090487778187,
+      "eval_seeclick_MAE_h": 0.04407734237611294,
+      "eval_seeclick_MAE_w": 0.0640980452299118,
+      "eval_seeclick_MAE_x": 0.04707919806241989,
+      "eval_seeclick_MAE_y": 0.041229039430618286,
+      "eval_seeclick_NUM_probability": 0.00013923477672506124,
+      "eval_seeclick_inside_bbox": 0.9409722089767456,
+      "eval_seeclick_loss": 0.4747886657714844,
+      "eval_seeclick_loss_ce": 0.03171798028051853,
+      "eval_seeclick_loss_iou": 1.0753774046897888,
+      "eval_seeclick_loss_num": 0.4178466796875,
+      "eval_seeclick_loss_xval": 0.4178466796875,
+      "eval_seeclick_runtime": 87.6366,
+      "eval_seeclick_samples_per_second": 0.571,
+      "eval_seeclick_steps_per_second": 0.023,
+      "num_input_tokens_seen": 172844576,
+      "step": 1000
+    },
+    {
+      "epoch": 0.38299502106472616,
+      "eval_icons_CIoU": 0.8593625128269196,
+      "eval_icons_GIoU": 0.8555968403816223,
+      "eval_icons_IoU": 0.8631813824176788,
+      "eval_icons_MAE_all": 0.01792115345597267,
+      "eval_icons_MAE_h": 0.020976515486836433,
+      "eval_icons_MAE_w": 0.02032420039176941,
+      "eval_icons_MAE_x": 0.015029226895421743,
+      "eval_icons_MAE_y": 0.015354669652879238,
+      "eval_icons_NUM_probability": 0.00010661216583685018,
+      "eval_icons_inside_bbox": 1.0,
+      "eval_icons_loss": 0.16257581114768982,
+      "eval_icons_loss_ce": 0.02727161906659603,
+      "eval_icons_loss_iou": 1.0135286450386047,
+      "eval_icons_loss_num": 0.126068115234375,
+      "eval_icons_loss_xval": 0.126068115234375,
+      "eval_icons_runtime": 90.5649,
+      "eval_icons_samples_per_second": 0.552,
+      "eval_icons_steps_per_second": 0.022,
+      "num_input_tokens_seen": 172844576,
+      "step": 1000
+    },
+    {
+      "epoch": 0.38299502106472616,
+      "loss": 0.17055588960647583,
+      "loss_ce": 0.02730637788772583,
+      "loss_iou": 1.0265284776687622,
+      "loss_num": 0.1435546875,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 172844576,
+      "step": 1000
+    },
+    {
+      "epoch": 0.38337801608579086,
+      "grad_norm": 34.91923648615059,
+      "learning_rate": 5e-06,
+      "loss": 0.1858,
+      "num_input_tokens_seen": 173017632,
+      "step": 1001
+    },
+    {
+      "epoch": 0.38337801608579086,
+      "loss": 0.18923646211624146,
+      "loss_ce": 0.020779425278306007,
+      "loss_iou": 1.0047352313995361,
+      "loss_num": 0.16796875,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 173017632,
+      "step": 1001
+    },
+    {
+      "epoch": 0.3837610111068556,
+      "grad_norm": 35.66388745742093,
+      "learning_rate": 5e-06,
+      "loss": 0.2295,
+      "num_input_tokens_seen": 173190344,
+      "step": 1002
+    },
+    {
+      "epoch": 0.3837610111068556,
+      "loss": 0.2730959951877594,
+      "loss_ce": 0.019678032025694847,
+      "loss_iou": 1.074703335762024,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 173190344,
+      "step": 1002
+    },
+    {
+      "epoch": 0.3841440061279203,
+      "grad_norm": 37.025768383792325,
+      "learning_rate": 5e-06,
+      "loss": 0.24,
+      "num_input_tokens_seen": 173363256,
+      "step": 1003
+    },
+    {
+      "epoch": 0.3841440061279203,
+      "loss": 0.25298523902893066,
+      "loss_ce": 0.020868537947535515,
+      "loss_iou": 1.1796908378601074,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 173363256,
+      "step": 1003
+    },
+    {
+      "epoch": 0.3845270011489851,
+      "grad_norm": 38.128815178417504,
+      "learning_rate": 5e-06,
+      "loss": 0.3525,
+      "num_input_tokens_seen": 173536208,
+      "step": 1004
+    },
+    {
+      "epoch": 0.3845270011489851,
+      "loss": 0.43987375497817993,
+      "loss_ce": 0.020318102091550827,
+      "loss_iou": 1.224792242050171,
+      "loss_num": 0.419921875,
+      "loss_xval": 0.419921875,
+      "num_input_tokens_seen": 173536208,
+      "step": 1004
+    },
+    {
+      "epoch": 0.3849099961700498,
+      "grad_norm": 39.33223346405527,
+      "learning_rate": 5e-06,
+      "loss": 0.1855,
+      "num_input_tokens_seen": 173709080,
+      "step": 1005
+    },
+    {
+      "epoch": 0.3849099961700498,
+      "loss": 0.21229764819145203,
+      "loss_ce": 0.020342081785202026,
+      "loss_iou": 1.0204285383224487,
+      "loss_num": 0.1923828125,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 173709080,
+      "step": 1005
+    },
+    {
+      "epoch": 0.38529299119111454,
+      "grad_norm": 32.986916212887074,
+      "learning_rate": 5e-06,
+      "loss": 0.2482,
+      "num_input_tokens_seen": 173881672,
+      "step": 1006
+    },
+    {
+      "epoch": 0.38529299119111454,
+      "loss": 0.2718982696533203,
+      "loss_ce": 0.019334791228175163,
+      "loss_iou": 0.9743728637695312,
+      "loss_num": 0.251953125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 173881672,
+      "step": 1006
+    },
+    {
+      "epoch": 0.38567598621217924,
+      "grad_norm": 33.071792353423476,
+      "learning_rate": 5e-06,
+      "loss": 0.2528,
+      "num_input_tokens_seen": 174054688,
+      "step": 1007
+    },
+    {
+      "epoch": 0.38567598621217924,
+      "loss": 0.21027815341949463,
+      "loss_ce": 0.019482256844639778,
+      "loss_iou": 1.053891658782959,
+      "loss_num": 0.1904296875,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 174054688,
+      "step": 1007
+    },
+    {
+      "epoch": 0.38605898123324395,
+      "grad_norm": 45.872281069364874,
+      "learning_rate": 5e-06,
+      "loss": 0.2233,
+      "num_input_tokens_seen": 174227720,
+      "step": 1008
+    },
+    {
+      "epoch": 0.38605898123324395,
+      "loss": 0.24927006661891937,
+      "loss_ce": 0.019350633025169373,
+      "loss_iou": 1.0087388753890991,
+      "loss_num": 0.2294921875,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 174227720,
+      "step": 1008
+    },
+    {
+      "epoch": 0.3864419762543087,
+      "grad_norm": 28.142361946977083,
+      "learning_rate": 5e-06,
+      "loss": 0.2527,
+      "num_input_tokens_seen": 174400504,
+      "step": 1009
+    },
+    {
+      "epoch": 0.3864419762543087,
+      "loss": 0.26419174671173096,
+      "loss_ce": 0.019806988537311554,
+      "loss_iou": 1.026214361190796,
+      "loss_num": 0.244140625,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 174400504,
+      "step": 1009
+    },
+    {
+      "epoch": 0.3868249712753734,
+      "grad_norm": 26.611276161604945,
+      "learning_rate": 5e-06,
+      "loss": 0.241,
+      "num_input_tokens_seen": 174573072,
+      "step": 1010
+    },
+    {
+      "epoch": 0.3868249712753734,
+      "loss": 0.2983134984970093,
+      "loss_ce": 0.01968800649046898,
+      "loss_iou": 1.1066608428955078,
+      "loss_num": 0.279296875,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 174573072,
+      "step": 1010
+    },
+    {
+      "epoch": 0.38720796629643817,
+      "grad_norm": 30.461948311421125,
+      "learning_rate": 5e-06,
+      "loss": 0.2167,
+      "num_input_tokens_seen": 174745704,
+      "step": 1011
+    },
+    {
+      "epoch": 0.38720796629643817,
+      "loss": 0.24395182728767395,
+      "loss_ce": 0.019708681851625443,
+      "loss_iou": 1.0080559253692627,
+      "loss_num": 0.224609375,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 174745704,
+      "step": 1011
+    },
+    {
+      "epoch": 0.38759096131750287,
+      "grad_norm": 32.797647663022616,
+      "learning_rate": 5e-06,
+      "loss": 0.1919,
+      "num_input_tokens_seen": 174919040,
+      "step": 1012
+    },
+    {
+      "epoch": 0.38759096131750287,
+      "loss": 0.2338232696056366,
+      "loss_ce": 0.019650910049676895,
+      "loss_iou": 1.0448970794677734,
+      "loss_num": 0.2138671875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 174919040,
+      "step": 1012
+    },
+    {
+      "epoch": 0.38797395633856757,
+      "grad_norm": 30.065078518528374,
+      "learning_rate": 5e-06,
+      "loss": 0.278,
+      "num_input_tokens_seen": 175092104,
+      "step": 1013
+    },
+    {
+      "epoch": 0.38797395633856757,
+      "loss": 0.32156360149383545,
+      "loss_ce": 0.020110974088311195,
+      "loss_iou": 1.0617436170578003,
+      "loss_num": 0.30078125,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 175092104,
+      "step": 1013
+    },
+    {
+      "epoch": 0.38835695135963233,
+      "grad_norm": 28.867010484729118,
+      "learning_rate": 5e-06,
+      "loss": 0.2178,
+      "num_input_tokens_seen": 175264816,
+      "step": 1014
+    },
+    {
+      "epoch": 0.38835695135963233,
+      "loss": 0.18044480681419373,
+      "loss_ce": 0.018396452069282532,
+      "loss_iou": 1.0876814126968384,
+      "loss_num": 0.162109375,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 175264816,
+      "step": 1014
+    },
+    {
+      "epoch": 0.38873994638069703,
+      "grad_norm": 28.792328412821917,
+      "learning_rate": 5e-06,
+      "loss": 0.2482,
+      "num_input_tokens_seen": 175437672,
+      "step": 1015
+    },
+    {
+      "epoch": 0.38873994638069703,
+      "loss": 0.2683184742927551,
+      "loss_ce": 0.018562620505690575,
+      "loss_iou": 1.0557260513305664,
+      "loss_num": 0.25,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 175437672,
+      "step": 1015
+    },
+    {
+      "epoch": 0.3891229414017618,
+      "grad_norm": 24.516995066221835,
+      "learning_rate": 5e-06,
+      "loss": 0.2651,
+      "num_input_tokens_seen": 175610552,
+      "step": 1016
+    },
+    {
+      "epoch": 0.3891229414017618,
+      "loss": 0.20224417746067047,
+      "loss_ce": 0.020542524755001068,
+      "loss_iou": 1.0099539756774902,
+      "loss_num": 0.181640625,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 175610552,
+      "step": 1016
+    },
+    {
+      "epoch": 0.3895059364228265,
+      "grad_norm": 29.986105631087796,
+      "learning_rate": 5e-06,
+      "loss": 0.3219,
+      "num_input_tokens_seen": 175783464,
+      "step": 1017
+    },
+    {
+      "epoch": 0.3895059364228265,
+      "loss": 0.3537331223487854,
+      "loss_ce": 0.02060324139893055,
+      "loss_iou": 1.188672661781311,
+      "loss_num": 0.333984375,
+      "loss_xval": 0.333984375,
+      "num_input_tokens_seen": 175783464,
+      "step": 1017
+    },
+    {
+      "epoch": 0.38988893144389125,
+      "grad_norm": 37.60632462935342,
+      "learning_rate": 5e-06,
+      "loss": 0.2641,
+      "num_input_tokens_seen": 175956424,
+      "step": 1018
+    },
+    {
+      "epoch": 0.38988893144389125,
+      "loss": 0.28005295991897583,
+      "loss_ce": 0.021630097180604935,
+      "loss_iou": 1.034566879272461,
+      "loss_num": 0.2578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 175956424,
+      "step": 1018
+    },
+    {
+      "epoch": 0.39027192646495595,
+      "grad_norm": 26.55092653731164,
+      "learning_rate": 5e-06,
+      "loss": 0.299,
+      "num_input_tokens_seen": 176125440,
+      "step": 1019
+    },
+    {
+      "epoch": 0.39027192646495595,
+      "loss": 0.2950381338596344,
+      "loss_ce": 0.020502010360360146,
+      "loss_iou": 1.079911470413208,
+      "loss_num": 0.275390625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 176125440,
+      "step": 1019
+    },
+    {
+      "epoch": 0.39065492148602066,
+      "grad_norm": 24.91511831938555,
+      "learning_rate": 5e-06,
+      "loss": 0.2435,
+      "num_input_tokens_seen": 176298344,
+      "step": 1020
+    },
+    {
+      "epoch": 0.39065492148602066,
+      "loss": 0.2348160445690155,
+      "loss_ce": 0.0199722982943058,
+      "loss_iou": 0.9842574596405029,
+      "loss_num": 0.21484375,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 176298344,
+      "step": 1020
+    },
+    {
+      "epoch": 0.3910379165070854,
+      "grad_norm": 36.35647838615096,
+      "learning_rate": 5e-06,
+      "loss": 0.2235,
+      "num_input_tokens_seen": 176471520,
+      "step": 1021
+    },
+    {
+      "epoch": 0.3910379165070854,
+      "loss": 0.20562788844108582,
+      "loss_ce": 0.02185102179646492,
+      "loss_iou": 0.9844660758972168,
+      "loss_num": 0.18359375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 176471520,
+      "step": 1021
+    },
+    {
+      "epoch": 0.3914209115281501,
+      "grad_norm": 34.59426683529261,
+      "learning_rate": 5e-06,
+      "loss": 0.2816,
+      "num_input_tokens_seen": 176644144,
+      "step": 1022
+    },
+    {
+      "epoch": 0.3914209115281501,
+      "loss": 0.2733522653579712,
+      "loss_ce": 0.01700459234416485,
+      "loss_iou": 1.3063082695007324,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 176644144,
+      "step": 1022
+    },
+    {
+      "epoch": 0.3918039065492149,
+      "grad_norm": 13.813531849418283,
+      "learning_rate": 5e-06,
+      "loss": 0.2646,
+      "num_input_tokens_seen": 176817016,
+      "step": 1023
+    },
+    {
+      "epoch": 0.3918039065492149,
+      "loss": 0.2954465448856354,
+      "loss_ce": 0.018224865198135376,
+      "loss_iou": 1.582000732421875,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 176817016,
+      "step": 1023
+    },
+    {
+      "epoch": 0.3921869015702796,
+      "grad_norm": 28.728838579840332,
+      "learning_rate": 5e-06,
+      "loss": 0.2812,
+      "num_input_tokens_seen": 176989848,
+      "step": 1024
+    },
+    {
+      "epoch": 0.3921869015702796,
+      "loss": 0.3526727557182312,
+      "loss_ce": 0.019359752535820007,
+      "loss_iou": 1.1450241804122925,
+      "loss_num": 0.333984375,
+      "loss_xval": 0.333984375,
+      "num_input_tokens_seen": 176989848,
+      "step": 1024
+    },
+    {
+      "epoch": 0.39256989659134434,
+      "grad_norm": 17.574054634642955,
+      "learning_rate": 5e-06,
+      "loss": 0.2239,
+      "num_input_tokens_seen": 177162880,
+      "step": 1025
+    },
+    {
+      "epoch": 0.39256989659134434,
+      "loss": 0.25614941120147705,
+      "loss_ce": 0.02116405963897705,
+      "loss_iou": 1.0470411777496338,
+      "loss_num": 0.2353515625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 177162880,
+      "step": 1025
+    },
+    {
+      "epoch": 0.39295289161240904,
+      "grad_norm": 31.87346085801817,
+      "learning_rate": 5e-06,
+      "loss": 0.2257,
+      "num_input_tokens_seen": 177335720,
+      "step": 1026
+    },
+    {
+      "epoch": 0.39295289161240904,
+      "loss": 0.2296457290649414,
+      "loss_ce": 0.02029513008892536,
+      "loss_iou": 1.0434712171554565,
+      "loss_num": 0.208984375,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 177335720,
+      "step": 1026
+    },
+    {
+      "epoch": 0.39333588663347374,
+      "grad_norm": 34.07962343740927,
+      "learning_rate": 5e-06,
+      "loss": 0.2748,
+      "num_input_tokens_seen": 177508480,
+      "step": 1027
+    },
+    {
+      "epoch": 0.39333588663347374,
+      "loss": 0.2981407046318054,
+      "loss_ce": 0.020308684557676315,
+      "loss_iou": 1.0449728965759277,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 177508480,
+      "step": 1027
+    },
+    {
+      "epoch": 0.3937188816545385,
+      "grad_norm": 21.19770177391623,
+      "learning_rate": 5e-06,
+      "loss": 0.2962,
+      "num_input_tokens_seen": 177681640,
+      "step": 1028
+    },
+    {
+      "epoch": 0.3937188816545385,
+      "loss": 0.32453253865242004,
+      "loss_ce": 0.019722970202565193,
+      "loss_iou": 1.0778220891952515,
+      "loss_num": 0.3046875,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 177681640,
+      "step": 1028
+    },
+    {
+      "epoch": 0.3941018766756032,
+      "grad_norm": 33.706028660777044,
+      "learning_rate": 5e-06,
+      "loss": 0.231,
+      "num_input_tokens_seen": 177854520,
+      "step": 1029
+    },
+    {
+      "epoch": 0.3941018766756032,
+      "loss": 0.2419479936361313,
+      "loss_ce": 0.020390372723340988,
+      "loss_iou": 1.0999131202697754,
+      "loss_num": 0.2216796875,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 177854520,
+      "step": 1029
+    },
+    {
+      "epoch": 0.39448487169666796,
+      "grad_norm": 35.236379668839774,
+      "learning_rate": 5e-06,
+      "loss": 0.2638,
+      "num_input_tokens_seen": 178020376,
+      "step": 1030
+    },
+    {
+      "epoch": 0.39448487169666796,
+      "loss": 0.23927438259124756,
+      "loss_ce": 0.022111300379037857,
+      "loss_iou": 1.0072345733642578,
+      "loss_num": 0.216796875,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 178020376,
+      "step": 1030
+    },
+    {
+      "epoch": 0.39486786671773266,
+      "grad_norm": 32.1285055354119,
+      "learning_rate": 5e-06,
+      "loss": 0.245,
+      "num_input_tokens_seen": 178193424,
+      "step": 1031
+    },
+    {
+      "epoch": 0.39486786671773266,
+      "loss": 0.29661643505096436,
+      "loss_ce": 0.019333723932504654,
+      "loss_iou": 1.0214552879333496,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 178193424,
+      "step": 1031
+    },
+    {
+      "epoch": 0.3952508617387974,
+      "grad_norm": 25.32324873100988,
+      "learning_rate": 5e-06,
+      "loss": 0.3053,
+      "num_input_tokens_seen": 178366336,
+      "step": 1032
+    },
+    {
+      "epoch": 0.3952508617387974,
+      "loss": 0.34949928522109985,
+      "loss_ce": 0.019421163946390152,
+      "loss_iou": 1.5528998374938965,
+      "loss_num": 0.330078125,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 178366336,
+      "step": 1032
+    },
+    {
+      "epoch": 0.3956338567598621,
+      "grad_norm": 27.588990325820333,
+      "learning_rate": 5e-06,
+      "loss": 0.2433,
+      "num_input_tokens_seen": 178539032,
+      "step": 1033
+    },
+    {
+      "epoch": 0.3956338567598621,
+      "loss": 0.22613750398159027,
+      "loss_ce": 0.019716612994670868,
+      "loss_iou": 1.0760841369628906,
+      "loss_num": 0.2060546875,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 178539032,
+      "step": 1033
+    },
+    {
+      "epoch": 0.3960168517809268,
+      "grad_norm": 35.87362400033478,
+      "learning_rate": 5e-06,
+      "loss": 0.3129,
+      "num_input_tokens_seen": 178711576,
+      "step": 1034
+    },
+    {
+      "epoch": 0.3960168517809268,
+      "loss": 0.31826305389404297,
+      "loss_ce": 0.01943489909172058,
+      "loss_iou": 1.1315784454345703,
+      "loss_num": 0.298828125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 178711576,
+      "step": 1034
+    },
+    {
+      "epoch": 0.3963998468019916,
+      "grad_norm": 28.41816637255368,
+      "learning_rate": 5e-06,
+      "loss": 0.2566,
+      "num_input_tokens_seen": 178884296,
+      "step": 1035
+    },
+    {
+      "epoch": 0.3963998468019916,
+      "loss": 0.1878519058227539,
+      "loss_ce": 0.019883153960108757,
+      "loss_iou": 1.001306176185608,
+      "loss_num": 0.16796875,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 178884296,
+      "step": 1035
+    },
+    {
+      "epoch": 0.3967828418230563,
+      "grad_norm": 28.305814305305066,
+      "learning_rate": 5e-06,
+      "loss": 0.2616,
+      "num_input_tokens_seen": 179057336,
+      "step": 1036
+    },
+    {
+      "epoch": 0.3967828418230563,
+      "loss": 0.24126334488391876,
+      "loss_ce": 0.018362944945693016,
+      "loss_iou": 1.0048975944519043,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 179057336,
+      "step": 1036
+    },
+    {
+      "epoch": 0.39716583684412105,
+      "grad_norm": 27.021356199118504,
+      "learning_rate": 5e-06,
+      "loss": 0.228,
+      "num_input_tokens_seen": 179230528,
+      "step": 1037
+    },
+    {
+      "epoch": 0.39716583684412105,
+      "loss": 0.2893359661102295,
+      "loss_ce": 0.020354025065898895,
+      "loss_iou": 1.10371732711792,
+      "loss_num": 0.26953125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 179230528,
+      "step": 1037
+    },
+    {
+      "epoch": 0.39754883186518575,
+      "grad_norm": 34.33560302695971,
+      "learning_rate": 5e-06,
+      "loss": 0.2778,
+      "num_input_tokens_seen": 179403520,
+      "step": 1038
+    },
+    {
+      "epoch": 0.39754883186518575,
+      "loss": 0.256139874458313,
+      "loss_ce": 0.018468990921974182,
+      "loss_iou": 1.0579614639282227,
+      "loss_num": 0.2373046875,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 179403520,
+      "step": 1038
+    },
+    {
+      "epoch": 0.3979318268862505,
+      "grad_norm": 34.447395900685464,
+      "learning_rate": 5e-06,
+      "loss": 0.2138,
+      "num_input_tokens_seen": 179576640,
+      "step": 1039
+    },
+    {
+      "epoch": 0.3979318268862505,
+      "loss": 0.24738307297229767,
+      "loss_ce": 0.02027124911546707,
+      "loss_iou": 1.0358725786209106,
+      "loss_num": 0.2275390625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 179576640,
+      "step": 1039
+    },
+    {
+      "epoch": 0.3983148219073152,
+      "grad_norm": 25.621526396280814,
+      "learning_rate": 5e-06,
+      "loss": 0.2689,
+      "num_input_tokens_seen": 179749928,
+      "step": 1040
+    },
+    {
+      "epoch": 0.3983148219073152,
+      "loss": 0.2736883759498596,
+      "loss_ce": 0.019904205575585365,
+      "loss_iou": 1.008162498474121,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 179749928,
+      "step": 1040
+    },
+    {
+      "epoch": 0.3986978169283799,
+      "grad_norm": 24.693164082863355,
+      "learning_rate": 5e-06,
+      "loss": 0.21,
+      "num_input_tokens_seen": 179922960,
+      "step": 1041
+    },
+    {
+      "epoch": 0.3986978169283799,
+      "loss": 0.22558265924453735,
+      "loss_ce": 0.020565571263432503,
+      "loss_iou": 1.0665814876556396,
+      "loss_num": 0.205078125,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 179922960,
+      "step": 1041
+    },
+    {
+      "epoch": 0.39908081194944467,
+      "grad_norm": 27.018742048274287,
+      "learning_rate": 5e-06,
+      "loss": 0.2807,
+      "num_input_tokens_seen": 180095264,
+      "step": 1042
+    },
+    {
+      "epoch": 0.39908081194944467,
+      "loss": 0.2431066781282425,
+      "loss_ce": 0.020145248621702194,
+      "loss_iou": 1.0317829847335815,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 180095264,
+      "step": 1042
+    },
+    {
+      "epoch": 0.39946380697050937,
+      "grad_norm": 28.78757936901894,
+      "learning_rate": 5e-06,
+      "loss": 0.3031,
+      "num_input_tokens_seen": 180268200,
+      "step": 1043
+    },
+    {
+      "epoch": 0.39946380697050937,
+      "loss": 0.25353485345840454,
+      "loss_ce": 0.021296095103025436,
+      "loss_iou": 1.0509358644485474,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 180268200,
+      "step": 1043
+    },
+    {
+      "epoch": 0.39984680199157413,
+      "grad_norm": 28.53706200030292,
+      "learning_rate": 5e-06,
+      "loss": 0.2743,
+      "num_input_tokens_seen": 180441240,
+      "step": 1044
+    },
+    {
+      "epoch": 0.39984680199157413,
+      "loss": 0.27864956855773926,
+      "loss_ce": 0.019067034125328064,
+      "loss_iou": 1.0073164701461792,
+      "loss_num": 0.259765625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 180441240,
+      "step": 1044
+    },
+    {
+      "epoch": 0.40022979701263883,
+      "grad_norm": 28.757712763800257,
+      "learning_rate": 5e-06,
+      "loss": 0.2324,
+      "num_input_tokens_seen": 180614320,
+      "step": 1045
+    },
+    {
+      "epoch": 0.40022979701263883,
+      "loss": 0.21844510734081268,
+      "loss_ce": 0.020508088171482086,
+      "loss_iou": 1.0111403465270996,
+      "loss_num": 0.1982421875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 180614320,
+      "step": 1045
+    },
+    {
+      "epoch": 0.40061279203370354,
+      "grad_norm": 36.304606140461196,
+      "learning_rate": 5e-06,
+      "loss": 0.2559,
+      "num_input_tokens_seen": 180787416,
+      "step": 1046
+    },
+    {
+      "epoch": 0.40061279203370354,
+      "loss": 0.2342616319656372,
+      "loss_ce": 0.019906163215637207,
+      "loss_iou": 1.0033844709396362,
+      "loss_num": 0.21484375,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 180787416,
+      "step": 1046
+    },
+    {
+      "epoch": 0.4009957870547683,
+      "grad_norm": 24.178970934702544,
+      "learning_rate": 5e-06,
+      "loss": 0.2158,
+      "num_input_tokens_seen": 180960432,
+      "step": 1047
+    },
+    {
+      "epoch": 0.4009957870547683,
+      "loss": 0.24342913925647736,
+      "loss_ce": 0.018758729100227356,
+      "loss_iou": 1.027082085609436,
+      "loss_num": 0.224609375,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 180960432,
+      "step": 1047
+    },
+    {
+      "epoch": 0.401378782075833,
+      "grad_norm": 23.55980725206559,
+      "learning_rate": 5e-06,
+      "loss": 0.2725,
+      "num_input_tokens_seen": 181132816,
+      "step": 1048
+    },
+    {
+      "epoch": 0.401378782075833,
+      "loss": 0.24806642532348633,
+      "loss_ce": 0.019550804048776627,
+      "loss_iou": 1.0464789867401123,
+      "loss_num": 0.228515625,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 181132816,
+      "step": 1048
+    },
+    {
+      "epoch": 0.40176177709689775,
+      "grad_norm": 15.16717833608982,
+      "learning_rate": 5e-06,
+      "loss": 0.2624,
+      "num_input_tokens_seen": 181305688,
+      "step": 1049
+    },
+    {
+      "epoch": 0.40176177709689775,
+      "loss": 0.2303747683763504,
+      "loss_ce": 0.021390395238995552,
+      "loss_iou": 1.0903661251068115,
+      "loss_num": 0.208984375,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 181305688,
+      "step": 1049
+    },
+    {
+      "epoch": 0.40214477211796246,
+      "grad_norm": 19.53155959200495,
+      "learning_rate": 5e-06,
+      "loss": 0.2944,
+      "num_input_tokens_seen": 181477952,
+      "step": 1050
+    },
+    {
+      "epoch": 0.40214477211796246,
+      "loss": 0.3223775029182434,
+      "loss_ce": 0.019276905804872513,
+      "loss_iou": 1.0366268157958984,
+      "loss_num": 0.302734375,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 181477952,
+      "step": 1050
+    },
+    {
+      "epoch": 0.4025277671390272,
+      "grad_norm": 28.92356228415835,
+      "learning_rate": 5e-06,
+      "loss": 0.2129,
+      "num_input_tokens_seen": 181650616,
+      "step": 1051
+    },
+    {
+      "epoch": 0.4025277671390272,
+      "loss": 0.2253831923007965,
+      "loss_ce": 0.020427130162715912,
+      "loss_iou": 1.0213334560394287,
+      "loss_num": 0.205078125,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 181650616,
+      "step": 1051
+    },
+    {
+      "epoch": 0.4029107621600919,
+      "grad_norm": 37.07564566175605,
+      "learning_rate": 5e-06,
+      "loss": 0.2537,
+      "num_input_tokens_seen": 181823720,
+      "step": 1052
+    },
+    {
+      "epoch": 0.4029107621600919,
+      "loss": 0.23519429564476013,
+      "loss_ce": 0.01882467046380043,
+      "loss_iou": 1.0054073333740234,
+      "loss_num": 0.216796875,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 181823720,
+      "step": 1052
+    },
+    {
+      "epoch": 0.4032937571811566,
+      "grad_norm": 31.201015046885313,
+      "learning_rate": 5e-06,
+      "loss": 0.2164,
+      "num_input_tokens_seen": 181996632,
+      "step": 1053
+    },
+    {
+      "epoch": 0.4032937571811566,
+      "loss": 0.20811310410499573,
+      "loss_ce": 0.01975860819220543,
+      "loss_iou": 1.007824182510376,
+      "loss_num": 0.1884765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 181996632,
+      "step": 1053
+    },
+    {
+      "epoch": 0.4036767522022214,
+      "grad_norm": 29.227328404145602,
+      "learning_rate": 5e-06,
+      "loss": 0.2954,
+      "num_input_tokens_seen": 182169560,
+      "step": 1054
+    },
+    {
+      "epoch": 0.4036767522022214,
+      "loss": 0.27629297971725464,
+      "loss_ce": 0.019334979355335236,
+      "loss_iou": 1.0095736980438232,
+      "loss_num": 0.2578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 182169560,
+      "step": 1054
+    },
+    {
+      "epoch": 0.4040597472232861,
+      "grad_norm": 17.552602461062378,
+      "learning_rate": 5e-06,
+      "loss": 0.2578,
+      "num_input_tokens_seen": 182342216,
+      "step": 1055
+    },
+    {
+      "epoch": 0.4040597472232861,
+      "loss": 0.33272290229797363,
+      "loss_ce": 0.019917726516723633,
+      "loss_iou": 1.024327039718628,
+      "loss_num": 0.3125,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 182342216,
+      "step": 1055
+    },
+    {
+      "epoch": 0.40444274224435084,
+      "grad_norm": 29.33588487581412,
+      "learning_rate": 5e-06,
+      "loss": 0.2677,
+      "num_input_tokens_seen": 182515128,
+      "step": 1056
+    },
+    {
+      "epoch": 0.40444274224435084,
+      "loss": 0.2414645403623581,
+      "loss_ce": 0.020639337599277496,
+      "loss_iou": 1.0498523712158203,
+      "loss_num": 0.220703125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 182515128,
+      "step": 1056
+    },
+    {
+      "epoch": 0.40482573726541554,
+      "grad_norm": 31.24753138828455,
+      "learning_rate": 5e-06,
+      "loss": 0.2141,
+      "num_input_tokens_seen": 182688608,
+      "step": 1057
+    },
+    {
+      "epoch": 0.40482573726541554,
+      "loss": 0.2143682837486267,
+      "loss_ce": 0.02027648501098156,
+      "loss_iou": 1.0015480518341064,
+      "loss_num": 0.1943359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 182688608,
+      "step": 1057
+    },
+    {
+      "epoch": 0.4052087322864803,
+      "grad_norm": 30.426761376657293,
+      "learning_rate": 5e-06,
+      "loss": 0.2727,
+      "num_input_tokens_seen": 182861744,
+      "step": 1058
+    },
+    {
+      "epoch": 0.4052087322864803,
+      "loss": 0.21315500140190125,
+      "loss_ce": 0.02089424803853035,
+      "loss_iou": 1.0028371810913086,
+      "loss_num": 0.1923828125,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 182861744,
+      "step": 1058
+    },
+    {
+      "epoch": 0.405591727307545,
+      "grad_norm": 40.38200187104438,
+      "learning_rate": 5e-06,
+      "loss": 0.2406,
+      "num_input_tokens_seen": 183034680,
+      "step": 1059
+    },
+    {
+      "epoch": 0.405591727307545,
+      "loss": 0.2552831172943115,
+      "loss_ce": 0.02084709331393242,
+      "loss_iou": 1.0171141624450684,
+      "loss_num": 0.234375,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 183034680,
+      "step": 1059
+    },
+    {
+      "epoch": 0.4059747223286097,
+      "grad_norm": 46.59560384369054,
+      "learning_rate": 5e-06,
+      "loss": 0.3161,
+      "num_input_tokens_seen": 183207984,
+      "step": 1060
+    },
+    {
+      "epoch": 0.4059747223286097,
+      "loss": 0.32938194274902344,
+      "loss_ce": 0.019811611622571945,
+      "loss_iou": 1.0379105806350708,
+      "loss_num": 0.30859375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 183207984,
+      "step": 1060
+    },
+    {
+      "epoch": 0.40635771734967446,
+      "grad_norm": 37.40181740370994,
+      "learning_rate": 5e-06,
+      "loss": 0.3163,
+      "num_input_tokens_seen": 183381376,
+      "step": 1061
+    },
+    {
+      "epoch": 0.40635771734967446,
+      "loss": 0.33082133531570435,
+      "loss_ce": 0.020884817466139793,
+      "loss_iou": 1.0592249631881714,
+      "loss_num": 0.310546875,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 183381376,
+      "step": 1061
+    },
+    {
+      "epoch": 0.40674071237073917,
+      "grad_norm": 29.14551515879792,
+      "learning_rate": 5e-06,
+      "loss": 0.2702,
+      "num_input_tokens_seen": 183554152,
+      "step": 1062
+    },
+    {
+      "epoch": 0.40674071237073917,
+      "loss": 0.2896021604537964,
+      "loss_ce": 0.019826781004667282,
+      "loss_iou": 1.015427827835083,
+      "loss_num": 0.26953125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 183554152,
+      "step": 1062
+    },
+    {
+      "epoch": 0.4071237073918039,
+      "grad_norm": 28.24061135290386,
+      "learning_rate": 5e-06,
+      "loss": 0.2408,
+      "num_input_tokens_seen": 183727000,
+      "step": 1063
+    },
+    {
+      "epoch": 0.4071237073918039,
+      "loss": 0.2897298038005829,
+      "loss_ce": 0.020686831325292587,
+      "loss_iou": 1.2125548124313354,
+      "loss_num": 0.26953125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 183727000,
+      "step": 1063
+    },
+    {
+      "epoch": 0.4075067024128686,
+      "grad_norm": 33.52604644553763,
+      "learning_rate": 5e-06,
+      "loss": 0.3711,
+      "num_input_tokens_seen": 183899816,
+      "step": 1064
+    },
+    {
+      "epoch": 0.4075067024128686,
+      "loss": 0.4948132634162903,
+      "loss_ce": 0.019593555480241776,
+      "loss_iou": 1.2182378768920898,
+      "loss_num": 0.474609375,
+      "loss_xval": 0.474609375,
+      "num_input_tokens_seen": 183899816,
+      "step": 1064
+    },
+    {
+      "epoch": 0.4078896974339334,
+      "grad_norm": 27.88108930750279,
+      "learning_rate": 5e-06,
+      "loss": 0.2385,
+      "num_input_tokens_seen": 184072696,
+      "step": 1065
+    },
+    {
+      "epoch": 0.4078896974339334,
+      "loss": 0.21623888611793518,
+      "loss_ce": 0.020682232454419136,
+      "loss_iou": 1.0074313879013062,
+      "loss_num": 0.1953125,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 184072696,
+      "step": 1065
+    },
+    {
+      "epoch": 0.4082726924549981,
+      "grad_norm": 37.1247900528633,
+      "learning_rate": 5e-06,
+      "loss": 0.2424,
+      "num_input_tokens_seen": 184245136,
+      "step": 1066
+    },
+    {
+      "epoch": 0.4082726924549981,
+      "loss": 0.22287967801094055,
+      "loss_ce": 0.019205372780561447,
+      "loss_iou": 1.0023624897003174,
+      "loss_num": 0.2041015625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 184245136,
+      "step": 1066
+    },
+    {
+      "epoch": 0.4086556874760628,
+      "grad_norm": 26.491031650768395,
+      "learning_rate": 5e-06,
+      "loss": 0.3627,
+      "num_input_tokens_seen": 184418320,
+      "step": 1067
+    },
+    {
+      "epoch": 0.4086556874760628,
+      "loss": 0.3207070827484131,
+      "loss_ce": 0.021024484187364578,
+      "loss_iou": 1.0953075885772705,
+      "loss_num": 0.298828125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 184418320,
+      "step": 1067
+    },
+    {
+      "epoch": 0.40903868249712755,
+      "grad_norm": 29.791257760310767,
+      "learning_rate": 5e-06,
+      "loss": 0.2265,
+      "num_input_tokens_seen": 184591320,
+      "step": 1068
+    },
+    {
+      "epoch": 0.40903868249712755,
+      "loss": 0.24238575994968414,
+      "loss_ce": 0.019729506224393845,
+      "loss_iou": 4.7769131098520414e+35,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 184591320,
+      "step": 1068
+    },
+    {
+      "epoch": 0.40942167751819225,
+      "grad_norm": 37.98331683212382,
+      "learning_rate": 5e-06,
+      "loss": 0.2225,
+      "num_input_tokens_seen": 184764320,
+      "step": 1069
+    },
+    {
+      "epoch": 0.40942167751819225,
+      "loss": 0.22537072002887726,
+      "loss_ce": 0.02084190584719181,
+      "loss_iou": 1.002504587173462,
+      "loss_num": 0.2041015625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 184764320,
+      "step": 1069
+    },
+    {
+      "epoch": 0.409804672539257,
+      "grad_norm": 35.89718177327458,
+      "learning_rate": 5e-06,
+      "loss": 0.375,
+      "num_input_tokens_seen": 184937048,
+      "step": 1070
+    },
+    {
+      "epoch": 0.409804672539257,
+      "loss": 0.3266220688819885,
+      "loss_ce": 0.01985936611890793,
+      "loss_iou": 1.1617192029953003,
+      "loss_num": 0.306640625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 184937048,
+      "step": 1070
+    },
+    {
+      "epoch": 0.4101876675603217,
+      "grad_norm": 21.651350807757197,
+      "learning_rate": 5e-06,
+      "loss": 0.1698,
+      "num_input_tokens_seen": 185110016,
+      "step": 1071
+    },
+    {
+      "epoch": 0.4101876675603217,
+      "loss": 0.15666621923446655,
+      "loss_ce": 0.020985066890716553,
+      "loss_iou": 1.0026936531066895,
+      "loss_num": 0.1357421875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 185110016,
+      "step": 1071
+    },
+    {
+      "epoch": 0.41057066258138647,
+      "grad_norm": 48.532328506967026,
+      "learning_rate": 5e-06,
+      "loss": 0.2417,
+      "num_input_tokens_seen": 185283120,
+      "step": 1072
+    },
+    {
+      "epoch": 0.41057066258138647,
+      "loss": 0.24449491500854492,
+      "loss_ce": 0.019214145839214325,
+      "loss_iou": 1.0388288497924805,
+      "loss_num": 0.2255859375,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 185283120,
+      "step": 1072
+    },
+    {
+      "epoch": 0.4109536576024512,
+      "grad_norm": 32.26843347837672,
+      "learning_rate": 5e-06,
+      "loss": 0.2295,
+      "num_input_tokens_seen": 185455952,
+      "step": 1073
+    },
+    {
+      "epoch": 0.4109536576024512,
+      "loss": 0.21386288106441498,
+      "loss_ce": 0.020625578239560127,
+      "loss_iou": 1.003318190574646,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 185455952,
+      "step": 1073
+    },
+    {
+      "epoch": 0.4113366526235159,
+      "grad_norm": 25.19425817777292,
+      "learning_rate": 5e-06,
+      "loss": 0.2858,
+      "num_input_tokens_seen": 185628432,
+      "step": 1074
+    },
+    {
+      "epoch": 0.4113366526235159,
+      "loss": 0.2122112363576889,
+      "loss_ce": 0.018485646694898605,
+      "loss_iou": 1.0039255619049072,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 185628432,
+      "step": 1074
+    },
+    {
+      "epoch": 0.41171964764458063,
+      "grad_norm": 13.217673976589335,
+      "learning_rate": 5e-06,
+      "loss": 0.1704,
+      "num_input_tokens_seen": 185801592,
+      "step": 1075
+    },
+    {
+      "epoch": 0.41171964764458063,
+      "loss": 0.17114916443824768,
+      "loss_ce": 0.020758548751473427,
+      "loss_iou": 1.0100396871566772,
+      "loss_num": 0.150390625,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 185801592,
+      "step": 1075
+    },
+    {
+      "epoch": 0.41210264266564534,
+      "grad_norm": 14.260577675080459,
+      "learning_rate": 5e-06,
+      "loss": 0.2129,
+      "num_input_tokens_seen": 185974600,
+      "step": 1076
+    },
+    {
+      "epoch": 0.41210264266564534,
+      "loss": 0.24450165033340454,
+      "loss_ce": 0.018915709108114243,
+      "loss_iou": 1.0633727312088013,
+      "loss_num": 0.2255859375,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 185974600,
+      "step": 1076
+    },
+    {
+      "epoch": 0.4124856376867101,
+      "grad_norm": 28.824251015629905,
+      "learning_rate": 5e-06,
+      "loss": 0.2476,
+      "num_input_tokens_seen": 186147568,
+      "step": 1077
+    },
+    {
+      "epoch": 0.4124856376867101,
+      "loss": 0.23406794667243958,
+      "loss_ce": 0.01995660364627838,
+      "loss_iou": 1.0066158771514893,
+      "loss_num": 0.2138671875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 186147568,
+      "step": 1077
+    },
+    {
+      "epoch": 0.4128686327077748,
+      "grad_norm": 36.032062956592505,
+      "learning_rate": 5e-06,
+      "loss": 0.3205,
+      "num_input_tokens_seen": 186320584,
+      "step": 1078
+    },
+    {
+      "epoch": 0.4128686327077748,
+      "loss": 0.3176344633102417,
+      "loss_ce": 0.019416697323322296,
+      "loss_iou": 1.0374646186828613,
+      "loss_num": 0.298828125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 186320584,
+      "step": 1078
+    },
+    {
+      "epoch": 0.4132516277288395,
+      "grad_norm": 30.008842799861185,
+      "learning_rate": 5e-06,
+      "loss": 0.2067,
+      "num_input_tokens_seen": 186493616,
+      "step": 1079
+    },
+    {
+      "epoch": 0.4132516277288395,
+      "loss": 0.20071427524089813,
+      "loss_ce": 0.01980607956647873,
+      "loss_iou": 1.0139362812042236,
+      "loss_num": 0.1806640625,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 186493616,
+      "step": 1079
+    },
+    {
+      "epoch": 0.41363462274990426,
+      "grad_norm": 25.83195018400052,
+      "learning_rate": 5e-06,
+      "loss": 0.3091,
+      "num_input_tokens_seen": 186666664,
+      "step": 1080
+    },
+    {
+      "epoch": 0.41363462274990426,
+      "loss": 0.25226709246635437,
+      "loss_ce": 0.019417980685830116,
+      "loss_iou": 1.0057278871536255,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 186666664,
+      "step": 1080
+    },
+    {
+      "epoch": 0.41401761777096896,
+      "grad_norm": 25.59488449260052,
+      "learning_rate": 5e-06,
+      "loss": 0.2476,
+      "num_input_tokens_seen": 186839648,
+      "step": 1081
+    },
+    {
+      "epoch": 0.41401761777096896,
+      "loss": 0.24771007895469666,
+      "loss_ce": 0.02007945626974106,
+      "loss_iou": 1.0573749542236328,
+      "loss_num": 0.2275390625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 186839648,
+      "step": 1081
+    },
+    {
+      "epoch": 0.4144006127920337,
+      "grad_norm": 36.715223570817095,
+      "learning_rate": 5e-06,
+      "loss": 0.2796,
+      "num_input_tokens_seen": 187012432,
+      "step": 1082
+    },
+    {
+      "epoch": 0.4144006127920337,
+      "loss": 0.3474213182926178,
+      "loss_ce": 0.019174234941601753,
+      "loss_iou": 0.8901275396347046,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 187012432,
+      "step": 1082
+    },
+    {
+      "epoch": 0.4147836078130984,
+      "grad_norm": 36.67370334258155,
+      "learning_rate": 5e-06,
+      "loss": 0.3268,
+      "num_input_tokens_seen": 187185240,
+      "step": 1083
+    },
+    {
+      "epoch": 0.4147836078130984,
+      "loss": 0.2963693141937256,
+      "loss_ce": 0.018720384687185287,
+      "loss_iou": 1.019195318222046,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 187185240,
+      "step": 1083
+    },
+    {
+      "epoch": 0.4151666028341632,
+      "grad_norm": 30.295862631894526,
+      "learning_rate": 5e-06,
+      "loss": 0.32,
+      "num_input_tokens_seen": 187356408,
+      "step": 1084
+    },
+    {
+      "epoch": 0.4151666028341632,
+      "loss": 0.3452816605567932,
+      "loss_ce": 0.019353941082954407,
+      "loss_iou": 1.0064892768859863,
+      "loss_num": 0.326171875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 187356408,
+      "step": 1084
+    },
+    {
+      "epoch": 0.4155495978552279,
+      "grad_norm": 36.726952729545765,
+      "learning_rate": 5e-06,
+      "loss": 0.186,
+      "num_input_tokens_seen": 187529288,
+      "step": 1085
+    },
+    {
+      "epoch": 0.4155495978552279,
+      "loss": 0.1999548077583313,
+      "loss_ce": 0.02014523558318615,
+      "loss_iou": 1.0213425159454346,
+      "loss_num": 0.1796875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 187529288,
+      "step": 1085
+    },
+    {
+      "epoch": 0.4159325928762926,
+      "grad_norm": 41.84504012812776,
+      "learning_rate": 5e-06,
+      "loss": 0.3025,
+      "num_input_tokens_seen": 187702224,
+      "step": 1086
+    },
+    {
+      "epoch": 0.4159325928762926,
+      "loss": 0.30090370774269104,
+      "loss_ce": 0.019653702154755592,
+      "loss_iou": 1.0435850620269775,
+      "loss_num": 0.28125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 187702224,
+      "step": 1086
+    },
+    {
+      "epoch": 0.41631558789735734,
+      "grad_norm": 35.58417190733154,
+      "learning_rate": 5e-06,
+      "loss": 0.2655,
+      "num_input_tokens_seen": 187875016,
+      "step": 1087
+    },
+    {
+      "epoch": 0.41631558789735734,
+      "loss": 0.34806880354881287,
+      "loss_ce": 0.018478956073522568,
+      "loss_iou": 1.1121827363967896,
+      "loss_num": 0.330078125,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 187875016,
+      "step": 1087
+    },
+    {
+      "epoch": 0.41669858291842204,
+      "grad_norm": 57.372595580243946,
+      "learning_rate": 5e-06,
+      "loss": 0.2721,
+      "num_input_tokens_seen": 188048056,
+      "step": 1088
+    },
+    {
+      "epoch": 0.41669858291842204,
+      "loss": 0.24286219477653503,
+      "loss_ce": 0.018741104751825333,
+      "loss_iou": 1.0283300876617432,
+      "loss_num": 0.224609375,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 188048056,
+      "step": 1088
+    },
+    {
+      "epoch": 0.4170815779394868,
+      "grad_norm": 35.09600697549438,
+      "learning_rate": 5e-06,
+      "loss": 0.2756,
+      "num_input_tokens_seen": 188221192,
+      "step": 1089
+    },
+    {
+      "epoch": 0.4170815779394868,
+      "loss": 0.32690510153770447,
+      "loss_ce": 0.019532054662704468,
+      "loss_iou": 0.9916969537734985,
+      "loss_num": 0.306640625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 188221192,
+      "step": 1089
+    },
+    {
+      "epoch": 0.4174645729605515,
+      "grad_norm": 35.8157320365255,
+      "learning_rate": 5e-06,
+      "loss": 0.2852,
+      "num_input_tokens_seen": 188394200,
+      "step": 1090
+    },
+    {
+      "epoch": 0.4174645729605515,
+      "loss": 0.287636935710907,
+      "loss_ce": 0.018838126212358475,
+      "loss_iou": 1.0336689949035645,
+      "loss_num": 0.26953125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 188394200,
+      "step": 1090
+    },
+    {
+      "epoch": 0.41784756798161626,
+      "grad_norm": 37.08378410038575,
+      "learning_rate": 5e-06,
+      "loss": 0.2278,
+      "num_input_tokens_seen": 188567032,
+      "step": 1091
+    },
+    {
+      "epoch": 0.41784756798161626,
+      "loss": 0.24460598826408386,
+      "loss_ce": 0.020790068432688713,
+      "loss_iou": 1.0342530012130737,
+      "loss_num": 0.2236328125,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 188567032,
+      "step": 1091
+    },
+    {
+      "epoch": 0.41823056300268097,
+      "grad_norm": 37.41860078335122,
+      "learning_rate": 5e-06,
+      "loss": 0.2608,
+      "num_input_tokens_seen": 188739480,
+      "step": 1092
+    },
+    {
+      "epoch": 0.41823056300268097,
+      "loss": 0.2734837532043457,
+      "loss_ce": 0.019089236855506897,
+      "loss_iou": 0.9870978593826294,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 188739480,
+      "step": 1092
+    },
+    {
+      "epoch": 0.41861355802374567,
+      "grad_norm": 20.193221961767293,
+      "learning_rate": 5e-06,
+      "loss": 0.2244,
+      "num_input_tokens_seen": 188912824,
+      "step": 1093
+    },
+    {
+      "epoch": 0.41861355802374567,
+      "loss": 0.20880837738513947,
+      "loss_ce": 0.02075905352830887,
+      "loss_iou": 1.0575461387634277,
+      "loss_num": 0.1884765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 188912824,
+      "step": 1093
+    },
+    {
+      "epoch": 0.4189965530448104,
+      "grad_norm": 26.32238347120246,
+      "learning_rate": 5e-06,
+      "loss": 0.2898,
+      "num_input_tokens_seen": 189085880,
+      "step": 1094
+    },
+    {
+      "epoch": 0.4189965530448104,
+      "loss": 0.2794140577316284,
+      "loss_ce": 0.020563945174217224,
+      "loss_iou": 1.1031012535095215,
+      "loss_num": 0.259765625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 189085880,
+      "step": 1094
+    },
+    {
+      "epoch": 0.41937954806587513,
+      "grad_norm": 37.26817002701847,
+      "learning_rate": 5e-06,
+      "loss": 0.2058,
+      "num_input_tokens_seen": 189258864,
+      "step": 1095
+    },
+    {
+      "epoch": 0.41937954806587513,
+      "loss": 0.1816214621067047,
+      "loss_ce": 0.019695177674293518,
+      "loss_iou": 1.0097739696502686,
+      "loss_num": 0.162109375,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 189258864,
+      "step": 1095
+    },
+    {
+      "epoch": 0.4197625430869399,
+      "grad_norm": 34.95408758265353,
+      "learning_rate": 5e-06,
+      "loss": 0.2835,
+      "num_input_tokens_seen": 189431648,
+      "step": 1096
+    },
+    {
+      "epoch": 0.4197625430869399,
+      "loss": 0.29773491621017456,
+      "loss_ce": 0.01990288496017456,
+      "loss_iou": 1.141841173171997,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 189431648,
+      "step": 1096
+    },
+    {
+      "epoch": 0.4201455381080046,
+      "grad_norm": 35.4006169941124,
+      "learning_rate": 5e-06,
+      "loss": 0.2949,
+      "num_input_tokens_seen": 189604664,
+      "step": 1097
+    },
+    {
+      "epoch": 0.4201455381080046,
+      "loss": 0.2436686009168625,
+      "loss_ce": 0.021439600735902786,
+      "loss_iou": 1.0211408138275146,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 189604664,
+      "step": 1097
+    },
+    {
+      "epoch": 0.42052853312906935,
+      "grad_norm": 34.428038870408685,
+      "learning_rate": 5e-06,
+      "loss": 0.2832,
+      "num_input_tokens_seen": 189778208,
+      "step": 1098
+    },
+    {
+      "epoch": 0.42052853312906935,
+      "loss": 0.2590641677379608,
+      "loss_ce": 0.020294636487960815,
+      "loss_iou": 1.025743007659912,
+      "loss_num": 0.23828125,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 189778208,
+      "step": 1098
+    },
+    {
+      "epoch": 0.42091152815013405,
+      "grad_norm": 35.0534447485883,
+      "learning_rate": 5e-06,
+      "loss": 0.2622,
+      "num_input_tokens_seen": 189950968,
+      "step": 1099
+    },
+    {
+      "epoch": 0.42091152815013405,
+      "loss": 0.2537471055984497,
+      "loss_ce": 0.019250061362981796,
+      "loss_iou": 1.0193878412246704,
+      "loss_num": 0.234375,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 189950968,
+      "step": 1099
+    },
+    {
+      "epoch": 0.42129452317119875,
+      "grad_norm": 35.467707534962194,
+      "learning_rate": 5e-06,
+      "loss": 0.3142,
+      "num_input_tokens_seen": 190123976,
+      "step": 1100
+    },
+    {
+      "epoch": 0.42129452317119875,
+      "loss": 0.3694373369216919,
+      "loss_ce": 0.019095528870821,
+      "loss_iou": 1.0063408613204956,
+      "loss_num": 0.349609375,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 190123976,
+      "step": 1100
+    },
+    {
+      "epoch": 0.4216775181922635,
+      "grad_norm": 30.95793796142663,
+      "learning_rate": 5e-06,
+      "loss": 0.2187,
+      "num_input_tokens_seen": 190297032,
+      "step": 1101
+    },
+    {
+      "epoch": 0.4216775181922635,
+      "loss": 0.23031185567378998,
+      "loss_ce": 0.02138851210474968,
+      "loss_iou": 1.0118106603622437,
+      "loss_num": 0.208984375,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 190297032,
+      "step": 1101
+    },
+    {
+      "epoch": 0.4220605132133282,
+      "grad_norm": 35.56362632816114,
+      "learning_rate": 5e-06,
+      "loss": 0.281,
+      "num_input_tokens_seen": 190470000,
+      "step": 1102
+    },
+    {
+      "epoch": 0.4220605132133282,
+      "loss": 0.22374743223190308,
+      "loss_ce": 0.019890017807483673,
+      "loss_iou": 1.0248589515686035,
+      "loss_num": 0.2041015625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 190470000,
+      "step": 1102
+    },
+    {
+      "epoch": 0.422443508234393,
+      "grad_norm": 48.711027174685476,
+      "learning_rate": 5e-06,
+      "loss": 0.2976,
+      "num_input_tokens_seen": 190643112,
+      "step": 1103
+    },
+    {
+      "epoch": 0.422443508234393,
+      "loss": 0.29828524589538574,
+      "loss_ce": 0.021185625344514847,
+      "loss_iou": 1.0907062292099,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 190643112,
+      "step": 1103
+    },
+    {
+      "epoch": 0.4228265032554577,
+      "grad_norm": 37.48421738750284,
+      "learning_rate": 5e-06,
+      "loss": 0.2993,
+      "num_input_tokens_seen": 190815712,
+      "step": 1104
+    },
+    {
+      "epoch": 0.4228265032554577,
+      "loss": 0.2960641384124756,
+      "loss_ce": 0.019452795386314392,
+      "loss_iou": 1.012925624847412,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 190815712,
+      "step": 1104
+    },
+    {
+      "epoch": 0.4232094982765224,
+      "grad_norm": 30.2858820801016,
+      "learning_rate": 5e-06,
+      "loss": 0.2165,
+      "num_input_tokens_seen": 190988688,
+      "step": 1105
+    },
+    {
+      "epoch": 0.4232094982765224,
+      "loss": 0.22336909174919128,
+      "loss_ce": 0.020122017711400986,
+      "loss_iou": 1.015170931816101,
+      "loss_num": 0.203125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 190988688,
+      "step": 1105
+    },
+    {
+      "epoch": 0.42359249329758714,
+      "grad_norm": 36.10718236901174,
+      "learning_rate": 5e-06,
+      "loss": 0.2552,
+      "num_input_tokens_seen": 191161624,
+      "step": 1106
+    },
+    {
+      "epoch": 0.42359249329758714,
+      "loss": 0.2612323760986328,
+      "loss_ce": 0.02002144604921341,
+      "loss_iou": 1.019909143447876,
+      "loss_num": 0.2412109375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 191161624,
+      "step": 1106
+    },
+    {
+      "epoch": 0.42397548831865184,
+      "grad_norm": 43.646195988064775,
+      "learning_rate": 5e-06,
+      "loss": 0.3127,
+      "num_input_tokens_seen": 191331472,
+      "step": 1107
+    },
+    {
+      "epoch": 0.42397548831865184,
+      "loss": 0.3050917088985443,
+      "loss_ce": 0.02323136292397976,
+      "loss_iou": 1.1774249076843262,
+      "loss_num": 0.28125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 191331472,
+      "step": 1107
+    },
+    {
+      "epoch": 0.4243584833397166,
+      "grad_norm": 28.77716085555061,
+      "learning_rate": 5e-06,
+      "loss": 0.316,
+      "num_input_tokens_seen": 191504464,
+      "step": 1108
+    },
+    {
+      "epoch": 0.4243584833397166,
+      "loss": 0.3153734803199768,
+      "loss_ce": 0.021306123584508896,
+      "loss_iou": 1.0083816051483154,
+      "loss_num": 0.294921875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 191504464,
+      "step": 1108
+    },
+    {
+      "epoch": 0.4247414783607813,
+      "grad_norm": 23.328156987166356,
+      "learning_rate": 5e-06,
+      "loss": 0.2327,
+      "num_input_tokens_seen": 191677864,
+      "step": 1109
+    },
+    {
+      "epoch": 0.4247414783607813,
+      "loss": 0.17955052852630615,
+      "loss_ce": 0.020676013082265854,
+      "loss_iou": 1.0050337314605713,
+      "loss_num": 0.1591796875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 191677864,
+      "step": 1109
+    },
+    {
+      "epoch": 0.42512447338184606,
+      "grad_norm": 44.71531405551023,
+      "learning_rate": 5e-06,
+      "loss": 0.2593,
+      "num_input_tokens_seen": 191850688,
+      "step": 1110
+    },
+    {
+      "epoch": 0.42512447338184606,
+      "loss": 0.2549092173576355,
+      "loss_ce": 0.019313503056764603,
+      "loss_iou": 1.0502129793167114,
+      "loss_num": 0.2353515625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 191850688,
+      "step": 1110
+    },
+    {
+      "epoch": 0.42550746840291076,
+      "grad_norm": 45.485111903636096,
+      "learning_rate": 5e-06,
+      "loss": 0.2858,
+      "num_input_tokens_seen": 192023720,
+      "step": 1111
+    },
+    {
+      "epoch": 0.42550746840291076,
+      "loss": 0.30052047967910767,
+      "loss_ce": 0.019636696204543114,
+      "loss_iou": 0.8777676820755005,
+      "loss_num": 0.28125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 192023720,
+      "step": 1111
+    },
+    {
+      "epoch": 0.42589046342397546,
+      "grad_norm": 30.37565301000588,
+      "learning_rate": 5e-06,
+      "loss": 0.321,
+      "num_input_tokens_seen": 192196616,
+      "step": 1112
+    },
+    {
+      "epoch": 0.42589046342397546,
+      "loss": 0.26790064573287964,
+      "loss_ce": 0.01985376700758934,
+      "loss_iou": 1.0034968852996826,
+      "loss_num": 0.248046875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 192196616,
+      "step": 1112
+    },
+    {
+      "epoch": 0.4262734584450402,
+      "grad_norm": 23.429370774409772,
+      "learning_rate": 5e-06,
+      "loss": 0.3124,
+      "num_input_tokens_seen": 192369312,
+      "step": 1113
+    },
+    {
+      "epoch": 0.4262734584450402,
+      "loss": 0.25368160009384155,
+      "loss_ce": 0.018452120944857597,
+      "loss_iou": 1.0222837924957275,
+      "loss_num": 0.2353515625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 192369312,
+      "step": 1113
+    },
+    {
+      "epoch": 0.4266564534661049,
+      "grad_norm": 49.81728447039158,
+      "learning_rate": 5e-06,
+      "loss": 0.2437,
+      "num_input_tokens_seen": 192541976,
+      "step": 1114
+    },
+    {
+      "epoch": 0.4266564534661049,
+      "loss": 0.19863948225975037,
+      "loss_ce": 0.01913510262966156,
+      "loss_iou": 1.0239115953445435,
+      "loss_num": 0.1796875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 192541976,
+      "step": 1114
+    },
+    {
+      "epoch": 0.4270394484871697,
+      "grad_norm": 53.96170294298491,
+      "learning_rate": 5e-06,
+      "loss": 0.3647,
+      "num_input_tokens_seen": 192715040,
+      "step": 1115
+    },
+    {
+      "epoch": 0.4270394484871697,
+      "loss": 0.38576173782348633,
+      "loss_ce": 0.021381856873631477,
+      "loss_iou": 1.0469461679458618,
+      "loss_num": 0.365234375,
+      "loss_xval": 0.365234375,
+      "num_input_tokens_seen": 192715040,
+      "step": 1115
+    },
+    {
+      "epoch": 0.4274224435082344,
+      "grad_norm": 30.996940663427463,
+      "learning_rate": 5e-06,
+      "loss": 0.3304,
+      "num_input_tokens_seen": 192887808,
+      "step": 1116
+    },
+    {
+      "epoch": 0.4274224435082344,
+      "loss": 0.32507526874542236,
+      "loss_ce": 0.018556740134954453,
+      "loss_iou": 1.0313948392868042,
+      "loss_num": 0.306640625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 192887808,
+      "step": 1116
+    },
+    {
+      "epoch": 0.42780543852929914,
+      "grad_norm": 23.36501609384908,
+      "learning_rate": 5e-06,
+      "loss": 0.1921,
+      "num_input_tokens_seen": 193060816,
+      "step": 1117
+    },
+    {
+      "epoch": 0.42780543852929914,
+      "loss": 0.17319723963737488,
+      "loss_ce": 0.022196270525455475,
+      "loss_iou": 1.0087032318115234,
+      "loss_num": 0.1513671875,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 193060816,
+      "step": 1117
+    },
+    {
+      "epoch": 0.42818843355036385,
+      "grad_norm": 43.03438693532102,
+      "learning_rate": 5e-06,
+      "loss": 0.3621,
+      "num_input_tokens_seen": 193233944,
+      "step": 1118
+    },
+    {
+      "epoch": 0.42818843355036385,
+      "loss": 0.39938050508499146,
+      "loss_ce": 0.022183235734701157,
+      "loss_iou": 1.0878607034683228,
+      "loss_num": 0.376953125,
+      "loss_xval": 0.376953125,
+      "num_input_tokens_seen": 193233944,
+      "step": 1118
+    },
+    {
+      "epoch": 0.42857142857142855,
+      "grad_norm": 31.289737397930423,
+      "learning_rate": 5e-06,
+      "loss": 0.2237,
+      "num_input_tokens_seen": 193407104,
+      "step": 1119
+    },
+    {
+      "epoch": 0.42857142857142855,
+      "loss": 0.2626212537288666,
+      "loss_ce": 0.02067788690328598,
+      "loss_iou": 1.1765764951705933,
+      "loss_num": 0.2421875,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 193407104,
+      "step": 1119
+    },
+    {
+      "epoch": 0.4289544235924933,
+      "grad_norm": 28.985363759382338,
+      "learning_rate": 5e-06,
+      "loss": 0.2763,
+      "num_input_tokens_seen": 193576656,
+      "step": 1120
+    },
+    {
+      "epoch": 0.4289544235924933,
+      "loss": 0.29900774359703064,
+      "loss_ce": 0.019955020397901535,
+      "loss_iou": 1.0166741609573364,
+      "loss_num": 0.279296875,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 193576656,
+      "step": 1120
+    },
+    {
+      "epoch": 0.429337418613558,
+      "grad_norm": 22.460035617900825,
+      "learning_rate": 5e-06,
+      "loss": 0.1649,
+      "num_input_tokens_seen": 193749608,
+      "step": 1121
+    },
+    {
+      "epoch": 0.429337418613558,
+      "loss": 0.15083445608615875,
+      "loss_ce": 0.019730933010578156,
+      "loss_iou": 1.0127614736557007,
+      "loss_num": 0.130859375,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 193749608,
+      "step": 1121
+    },
+    {
+      "epoch": 0.42972041363462277,
+      "grad_norm": 38.12969874093416,
+      "learning_rate": 5e-06,
+      "loss": 0.2681,
+      "num_input_tokens_seen": 193922552,
+      "step": 1122
+    },
+    {
+      "epoch": 0.42972041363462277,
+      "loss": 0.3112063407897949,
+      "loss_ce": 0.019946586340665817,
+      "loss_iou": 0.9104921817779541,
+      "loss_num": 0.291015625,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 193922552,
+      "step": 1122
+    },
+    {
+      "epoch": 0.43010340865568747,
+      "grad_norm": 35.74217380806701,
+      "learning_rate": 5e-06,
+      "loss": 0.2094,
+      "num_input_tokens_seen": 194095616,
+      "step": 1123
+    },
+    {
+      "epoch": 0.43010340865568747,
+      "loss": 0.2030068337917328,
+      "loss_ce": 0.02014549821615219,
+      "loss_iou": 1.064756155014038,
+      "loss_num": 0.1826171875,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 194095616,
+      "step": 1123
+    },
+    {
+      "epoch": 0.4304864036767522,
+      "grad_norm": 37.04493085369391,
+      "learning_rate": 5e-06,
+      "loss": 0.2416,
+      "num_input_tokens_seen": 194268272,
+      "step": 1124
+    },
+    {
+      "epoch": 0.4304864036767522,
+      "loss": 0.2464175522327423,
+      "loss_ce": 0.020343340933322906,
+      "loss_iou": 1.0350528955459595,
+      "loss_num": 0.2265625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 194268272,
+      "step": 1124
+    },
+    {
+      "epoch": 0.43086939869781693,
+      "grad_norm": 38.20645250797387,
+      "learning_rate": 5e-06,
+      "loss": 0.2287,
+      "num_input_tokens_seen": 194441376,
+      "step": 1125
+    },
+    {
+      "epoch": 0.43086939869781693,
+      "loss": 0.21355994045734406,
+      "loss_ce": 0.020688854157924652,
+      "loss_iou": 1.0069462060928345,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 194441376,
+      "step": 1125
+    },
+    {
+      "epoch": 0.43125239371888163,
+      "grad_norm": 28.52901901475493,
+      "learning_rate": 5e-06,
+      "loss": 0.2658,
+      "num_input_tokens_seen": 194614168,
+      "step": 1126
+    },
+    {
+      "epoch": 0.43125239371888163,
+      "loss": 0.2590653896331787,
+      "loss_ce": 0.020051725208759308,
+      "loss_iou": 1.050733208656311,
+      "loss_num": 0.2392578125,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 194614168,
+      "step": 1126
+    },
+    {
+      "epoch": 0.4316353887399464,
+      "grad_norm": 23.461995361491727,
+      "learning_rate": 5e-06,
+      "loss": 0.2763,
+      "num_input_tokens_seen": 194787328,
+      "step": 1127
+    },
+    {
+      "epoch": 0.4316353887399464,
+      "loss": 0.2598300576210022,
+      "loss_ce": 0.020206015557050705,
+      "loss_iou": 1.007830262184143,
+      "loss_num": 0.2392578125,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 194787328,
+      "step": 1127
+    },
+    {
+      "epoch": 0.4320183837610111,
+      "grad_norm": 29.33121127397308,
+      "learning_rate": 5e-06,
+      "loss": 0.2646,
+      "num_input_tokens_seen": 194959912,
+      "step": 1128
+    },
+    {
+      "epoch": 0.4320183837610111,
+      "loss": 0.23208530247211456,
+      "loss_ce": 0.017912931740283966,
+      "loss_iou": 1.0141150951385498,
+      "loss_num": 0.2138671875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 194959912,
+      "step": 1128
+    },
+    {
+      "epoch": 0.43240137878207585,
+      "grad_norm": 39.34486748573164,
+      "learning_rate": 5e-06,
+      "loss": 0.2952,
+      "num_input_tokens_seen": 195132584,
+      "step": 1129
+    },
+    {
+      "epoch": 0.43240137878207585,
+      "loss": 0.23368357121944427,
+      "loss_ce": 0.02024361863732338,
+      "loss_iou": 1.0584793090820312,
+      "loss_num": 0.2138671875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 195132584,
+      "step": 1129
+    },
+    {
+      "epoch": 0.43278437380314055,
+      "grad_norm": 22.854477573257608,
+      "learning_rate": 5e-06,
+      "loss": 0.2324,
+      "num_input_tokens_seen": 195305600,
+      "step": 1130
+    },
+    {
+      "epoch": 0.43278437380314055,
+      "loss": 0.31261980533599854,
+      "loss_ce": 0.020383477210998535,
+      "loss_iou": 1.2783775329589844,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 195305600,
+      "step": 1130
+    },
+    {
+      "epoch": 0.4331673688242053,
+      "grad_norm": 33.972667461606015,
+      "learning_rate": 5e-06,
+      "loss": 0.227,
+      "num_input_tokens_seen": 195478744,
+      "step": 1131
+    },
+    {
+      "epoch": 0.4331673688242053,
+      "loss": 0.19473661482334137,
+      "loss_ce": 0.02170194685459137,
+      "loss_iou": 1.0100171566009521,
+      "loss_num": 0.1728515625,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 195478744,
+      "step": 1131
+    },
+    {
+      "epoch": 0.43355036384527,
+      "grad_norm": 33.56408684153845,
+      "learning_rate": 5e-06,
+      "loss": 0.3593,
+      "num_input_tokens_seen": 195651632,
+      "step": 1132
+    },
+    {
+      "epoch": 0.43355036384527,
+      "loss": 0.280134379863739,
+      "loss_ce": 0.01988050900399685,
+      "loss_iou": 1.0125410556793213,
+      "loss_num": 0.259765625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 195651632,
+      "step": 1132
+    },
+    {
+      "epoch": 0.4339333588663347,
+      "grad_norm": 27.666942950446835,
+      "learning_rate": 5e-06,
+      "loss": 0.2947,
+      "num_input_tokens_seen": 195824592,
+      "step": 1133
+    },
+    {
+      "epoch": 0.4339333588663347,
+      "loss": 0.35064661502838135,
+      "loss_ce": 0.02154507488012314,
+      "loss_iou": 1.8595023155212402,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 195824592,
+      "step": 1133
+    },
+    {
+      "epoch": 0.4343163538873995,
+      "grad_norm": 26.545706769596773,
+      "learning_rate": 5e-06,
+      "loss": 0.2278,
+      "num_input_tokens_seen": 195997792,
+      "step": 1134
+    },
+    {
+      "epoch": 0.4343163538873995,
+      "loss": 0.23813310265541077,
+      "loss_ce": 0.021336231380701065,
+      "loss_iou": 1.0221257209777832,
+      "loss_num": 0.216796875,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 195997792,
+      "step": 1134
+    },
+    {
+      "epoch": 0.4346993489084642,
+      "grad_norm": 31.21511075386116,
+      "learning_rate": 5e-06,
+      "loss": 0.2317,
+      "num_input_tokens_seen": 196170952,
+      "step": 1135
+    },
+    {
+      "epoch": 0.4346993489084642,
+      "loss": 0.23391905426979065,
+      "loss_ce": 0.021943967789411545,
+      "loss_iou": 1.0050550699234009,
+      "loss_num": 0.2119140625,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 196170952,
+      "step": 1135
+    },
+    {
+      "epoch": 0.43508234392952894,
+      "grad_norm": 34.490804779413054,
+      "learning_rate": 5e-06,
+      "loss": 0.3355,
+      "num_input_tokens_seen": 196343768,
+      "step": 1136
+    },
+    {
+      "epoch": 0.43508234392952894,
+      "loss": 0.3720143437385559,
+      "loss_ce": 0.021550467237830162,
+      "loss_iou": 1.0862687826156616,
+      "loss_num": 0.349609375,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 196343768,
+      "step": 1136
+    },
+    {
+      "epoch": 0.43546533895059364,
+      "grad_norm": 32.27694300787197,
+      "learning_rate": 5e-06,
+      "loss": 0.2392,
+      "num_input_tokens_seen": 196516592,
+      "step": 1137
+    },
+    {
+      "epoch": 0.43546533895059364,
+      "loss": 0.25271502137184143,
+      "loss_ce": 0.019682809710502625,
+      "loss_iou": 1.098147988319397,
+      "loss_num": 0.2333984375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 196516592,
+      "step": 1137
+    },
+    {
+      "epoch": 0.43584833397165834,
+      "grad_norm": 29.930265659285894,
+      "learning_rate": 5e-06,
+      "loss": 0.275,
+      "num_input_tokens_seen": 196689856,
+      "step": 1138
+    },
+    {
+      "epoch": 0.43584833397165834,
+      "loss": 0.256893128156662,
+      "loss_ce": 0.01958843693137169,
+      "loss_iou": 1.010021448135376,
+      "loss_num": 0.2373046875,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 196689856,
+      "step": 1138
+    },
+    {
+      "epoch": 0.4362313289927231,
+      "grad_norm": 40.63195635884255,
+      "learning_rate": 5e-06,
+      "loss": 0.2234,
+      "num_input_tokens_seen": 196862832,
+      "step": 1139
+    },
+    {
+      "epoch": 0.4362313289927231,
+      "loss": 0.20894160866737366,
+      "loss_ce": 0.020404014736413956,
+      "loss_iou": 1.120874285697937,
+      "loss_num": 0.1884765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 196862832,
+      "step": 1139
+    },
+    {
+      "epoch": 0.4366143240137878,
+      "grad_norm": 50.467821313162986,
+      "learning_rate": 5e-06,
+      "loss": 0.2889,
+      "num_input_tokens_seen": 197035712,
+      "step": 1140
+    },
+    {
+      "epoch": 0.4366143240137878,
+      "loss": 0.2801207900047302,
+      "loss_ce": 0.020721379667520523,
+      "loss_iou": 1.0571398735046387,
+      "loss_num": 0.259765625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 197035712,
+      "step": 1140
+    },
+    {
+      "epoch": 0.43699731903485256,
+      "grad_norm": 35.97720727841275,
+      "learning_rate": 5e-06,
+      "loss": 0.2381,
+      "num_input_tokens_seen": 197208320,
+      "step": 1141
+    },
+    {
+      "epoch": 0.43699731903485256,
+      "loss": 0.22430667281150818,
+      "loss_ce": 0.022646527737379074,
+      "loss_iou": 1.0075345039367676,
+      "loss_num": 0.201171875,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 197208320,
+      "step": 1141
+    },
+    {
+      "epoch": 0.43738031405591726,
+      "grad_norm": 34.950341336361674,
+      "learning_rate": 5e-06,
+      "loss": 0.2609,
+      "num_input_tokens_seen": 197381760,
+      "step": 1142
+    },
+    {
+      "epoch": 0.43738031405591726,
+      "loss": 0.24490168690681458,
+      "loss_ce": 0.020902663469314575,
+      "loss_iou": 1.0394790172576904,
+      "loss_num": 0.2236328125,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 197381760,
+      "step": 1142
+    },
+    {
+      "epoch": 0.437763309076982,
+      "grad_norm": 25.107315876505393,
+      "learning_rate": 5e-06,
+      "loss": 0.2079,
+      "num_input_tokens_seen": 197554856,
+      "step": 1143
+    },
+    {
+      "epoch": 0.437763309076982,
+      "loss": 0.24060098826885223,
+      "loss_ce": 0.020508214831352234,
+      "loss_iou": 1.3005890846252441,
+      "loss_num": 0.2197265625,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 197554856,
+      "step": 1143
+    },
+    {
+      "epoch": 0.4381463040980467,
+      "grad_norm": 41.9365348406118,
+      "learning_rate": 5e-06,
+      "loss": 0.3391,
+      "num_input_tokens_seen": 197727792,
+      "step": 1144
+    },
+    {
+      "epoch": 0.4381463040980467,
+      "loss": 0.3874055743217468,
+      "loss_ce": 0.01972980797290802,
+      "loss_iou": 1.177030324935913,
+      "loss_num": 0.3671875,
+      "loss_xval": 0.3671875,
+      "num_input_tokens_seen": 197727792,
+      "step": 1144
+    },
+    {
+      "epoch": 0.4385292991191114,
+      "grad_norm": 41.52103336616536,
+      "learning_rate": 5e-06,
+      "loss": 0.2384,
+      "num_input_tokens_seen": 197900872,
+      "step": 1145
+    },
+    {
+      "epoch": 0.4385292991191114,
+      "loss": 0.2604897618293762,
+      "loss_ce": 0.022025402635335922,
+      "loss_iou": 1.2552176713943481,
+      "loss_num": 0.23828125,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 197900872,
+      "step": 1145
+    },
+    {
+      "epoch": 0.4389122941401762,
+      "grad_norm": 29.365917048394262,
+      "learning_rate": 5e-06,
+      "loss": 0.2574,
+      "num_input_tokens_seen": 198073696,
+      "step": 1146
+    },
+    {
+      "epoch": 0.4389122941401762,
+      "loss": 0.2711096405982971,
+      "loss_ce": 0.01976683922111988,
+      "loss_iou": 1.0339949131011963,
+      "loss_num": 0.251953125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 198073696,
+      "step": 1146
+    },
+    {
+      "epoch": 0.4392952891612409,
+      "grad_norm": 24.33312632490962,
+      "learning_rate": 5e-06,
+      "loss": 0.2307,
+      "num_input_tokens_seen": 198246672,
+      "step": 1147
+    },
+    {
+      "epoch": 0.4392952891612409,
+      "loss": 0.22538858652114868,
+      "loss_ce": 0.02092081494629383,
+      "loss_iou": 1.0422980785369873,
+      "loss_num": 0.2041015625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 198246672,
+      "step": 1147
+    },
+    {
+      "epoch": 0.43967828418230565,
+      "grad_norm": 27.943519811091573,
+      "learning_rate": 5e-06,
+      "loss": 0.2625,
+      "num_input_tokens_seen": 198419504,
+      "step": 1148
+    },
+    {
+      "epoch": 0.43967828418230565,
+      "loss": 0.22157150506973267,
+      "loss_ce": 0.020765848457813263,
+      "loss_iou": 1.0049946308135986,
+      "loss_num": 0.201171875,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 198419504,
+      "step": 1148
+    },
+    {
+      "epoch": 0.44006127920337035,
+      "grad_norm": 27.988290899295784,
+      "learning_rate": 5e-06,
+      "loss": 0.275,
+      "num_input_tokens_seen": 198592888,
+      "step": 1149
+    },
+    {
+      "epoch": 0.44006127920337035,
+      "loss": 0.23316389322280884,
+      "loss_ce": 0.019784986972808838,
+      "loss_iou": 1.017301082611084,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 198592888,
+      "step": 1149
+    },
+    {
+      "epoch": 0.4404442742244351,
+      "grad_norm": 30.494644210178567,
+      "learning_rate": 5e-06,
+      "loss": 0.3304,
+      "num_input_tokens_seen": 198761976,
+      "step": 1150
+    },
+    {
+      "epoch": 0.4404442742244351,
+      "loss": 0.4319886863231659,
+      "loss_ce": 0.023297281935811043,
+      "loss_iou": 1.8764491081237793,
+      "loss_num": 0.408203125,
+      "loss_xval": 0.408203125,
+      "num_input_tokens_seen": 198761976,
+      "step": 1150
+    },
+    {
+      "epoch": 0.4408272692454998,
+      "grad_norm": 26.69361483665594,
+      "learning_rate": 5e-06,
+      "loss": 0.254,
+      "num_input_tokens_seen": 198934792,
+      "step": 1151
+    },
+    {
+      "epoch": 0.4408272692454998,
+      "loss": 0.23793041706085205,
+      "loss_ce": 0.021011468023061752,
+      "loss_iou": 1.084611177444458,
+      "loss_num": 0.216796875,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 198934792,
+      "step": 1151
+    },
+    {
+      "epoch": 0.4412102642665645,
+      "grad_norm": 39.02427073739685,
+      "learning_rate": 5e-06,
+      "loss": 0.2906,
+      "num_input_tokens_seen": 199108008,
+      "step": 1152
+    },
+    {
+      "epoch": 0.4412102642665645,
+      "loss": 0.3364917039871216,
+      "loss_ce": 0.020329583436250687,
+      "loss_iou": 1.074775218963623,
+      "loss_num": 0.31640625,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 199108008,
+      "step": 1152
+    },
+    {
+      "epoch": 0.44159325928762927,
+      "grad_norm": 35.111010496577286,
+      "learning_rate": 5e-06,
+      "loss": 0.2216,
+      "num_input_tokens_seen": 199280944,
+      "step": 1153
+    },
+    {
+      "epoch": 0.44159325928762927,
+      "loss": 0.2469591498374939,
+      "loss_ce": 0.02064078487455845,
+      "loss_iou": 1.0694262981414795,
+      "loss_num": 0.2265625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 199280944,
+      "step": 1153
+    },
+    {
+      "epoch": 0.44197625430869397,
+      "grad_norm": 24.67561900275964,
+      "learning_rate": 5e-06,
+      "loss": 0.2682,
+      "num_input_tokens_seen": 199453416,
+      "step": 1154
+    },
+    {
+      "epoch": 0.44197625430869397,
+      "loss": 0.2632145583629608,
+      "loss_ce": 0.018585659563541412,
+      "loss_iou": 1.0391876697540283,
+      "loss_num": 0.244140625,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 199453416,
+      "step": 1154
+    },
+    {
+      "epoch": 0.44235924932975873,
+      "grad_norm": 17.922251172974647,
+      "learning_rate": 5e-06,
+      "loss": 0.2408,
+      "num_input_tokens_seen": 199626208,
+      "step": 1155
+    },
+    {
+      "epoch": 0.44235924932975873,
+      "loss": 0.24920615553855896,
+      "loss_ce": 0.018859483301639557,
+      "loss_iou": 1.166675090789795,
+      "loss_num": 0.23046875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 199626208,
+      "step": 1155
+    },
+    {
+      "epoch": 0.44274224435082343,
+      "grad_norm": 26.41346711504118,
+      "learning_rate": 5e-06,
+      "loss": 0.2129,
+      "num_input_tokens_seen": 199799312,
+      "step": 1156
+    },
+    {
+      "epoch": 0.44274224435082343,
+      "loss": 0.24005916714668274,
+      "loss_ce": 0.020271575078368187,
+      "loss_iou": 1.0121383666992188,
+      "loss_num": 0.2197265625,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 199799312,
+      "step": 1156
+    },
+    {
+      "epoch": 0.4431252393718882,
+      "grad_norm": 32.11700929479766,
+      "learning_rate": 5e-06,
+      "loss": 0.2136,
+      "num_input_tokens_seen": 199972096,
+      "step": 1157
+    },
+    {
+      "epoch": 0.4431252393718882,
+      "loss": 0.21976438164710999,
+      "loss_ce": 0.020057355985045433,
+      "loss_iou": 1.0306508541107178,
+      "loss_num": 0.19921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 199972096,
+      "step": 1157
+    },
+    {
+      "epoch": 0.4435082343929529,
+      "grad_norm": 30.826752734022154,
+      "learning_rate": 5e-06,
+      "loss": 0.2375,
+      "num_input_tokens_seen": 200145200,
+      "step": 1158
+    },
+    {
+      "epoch": 0.4435082343929529,
+      "loss": 0.20936432480812073,
+      "loss_ce": 0.019606031477451324,
+      "loss_iou": 1.0025780200958252,
+      "loss_num": 0.189453125,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 200145200,
+      "step": 1158
+    },
+    {
+      "epoch": 0.4438912294140176,
+      "grad_norm": 25.947265547466927,
+      "learning_rate": 5e-06,
+      "loss": 0.2716,
+      "num_input_tokens_seen": 200318200,
+      "step": 1159
+    },
+    {
+      "epoch": 0.4438912294140176,
+      "loss": 0.2963043451309204,
+      "loss_ce": 0.02054748684167862,
+      "loss_iou": 1.0963495969772339,
+      "loss_num": 0.275390625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 200318200,
+      "step": 1159
+    },
+    {
+      "epoch": 0.44427422443508235,
+      "grad_norm": 24.302641455107704,
+      "learning_rate": 5e-06,
+      "loss": 0.2201,
+      "num_input_tokens_seen": 200491392,
+      "step": 1160
+    },
+    {
+      "epoch": 0.44427422443508235,
+      "loss": 0.18968833982944489,
+      "loss_ce": 0.020620953291654587,
+      "loss_iou": 1.037597894668579,
+      "loss_num": 0.1689453125,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 200491392,
+      "step": 1160
+    },
+    {
+      "epoch": 0.44465721945614706,
+      "grad_norm": 28.242728423249723,
+      "learning_rate": 5e-06,
+      "loss": 0.3088,
+      "num_input_tokens_seen": 200660680,
+      "step": 1161
+    },
+    {
+      "epoch": 0.44465721945614706,
+      "loss": 0.32645246386528015,
+      "loss_ce": 0.01968976855278015,
+      "loss_iou": 1.0893526077270508,
+      "loss_num": 0.306640625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 200660680,
+      "step": 1161
+    },
+    {
+      "epoch": 0.4450402144772118,
+      "grad_norm": 31.27411261250214,
+      "learning_rate": 5e-06,
+      "loss": 0.311,
+      "num_input_tokens_seen": 200833776,
+      "step": 1162
+    },
+    {
+      "epoch": 0.4450402144772118,
+      "loss": 0.3027549982070923,
+      "loss_ce": 0.019734999164938927,
+      "loss_iou": 1.0339843034744263,
+      "loss_num": 0.283203125,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 200833776,
+      "step": 1162
+    },
+    {
+      "epoch": 0.4454232094982765,
+      "grad_norm": 28.83713172926187,
+      "learning_rate": 5e-06,
+      "loss": 0.2088,
+      "num_input_tokens_seen": 201007016,
+      "step": 1163
+    },
+    {
+      "epoch": 0.4454232094982765,
+      "loss": 0.1710694581270218,
+      "loss_ce": 0.021411247551441193,
+      "loss_iou": 1.0376994609832764,
+      "loss_num": 0.1494140625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 201007016,
+      "step": 1163
+    },
+    {
+      "epoch": 0.4458062045193413,
+      "grad_norm": 24.68230334788439,
+      "learning_rate": 5e-06,
+      "loss": 0.2782,
+      "num_input_tokens_seen": 201180064,
+      "step": 1164
+    },
+    {
+      "epoch": 0.4458062045193413,
+      "loss": 0.3124706745147705,
+      "loss_ce": 0.019440874457359314,
+      "loss_iou": 1.0577409267425537,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 201180064,
+      "step": 1164
+    },
+    {
+      "epoch": 0.446189199540406,
+      "grad_norm": 27.25612711491665,
+      "learning_rate": 5e-06,
+      "loss": 0.1769,
+      "num_input_tokens_seen": 201353120,
+      "step": 1165
+    },
+    {
+      "epoch": 0.446189199540406,
+      "loss": 0.18721093237400055,
+      "loss_ce": 0.02034081518650055,
+      "loss_iou": 1.023526906967163,
+      "loss_num": 0.1669921875,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 201353120,
+      "step": 1165
+    },
+    {
+      "epoch": 0.4465721945614707,
+      "grad_norm": 43.52114898506472,
+      "learning_rate": 5e-06,
+      "loss": 0.2885,
+      "num_input_tokens_seen": 201526272,
+      "step": 1166
+    },
+    {
+      "epoch": 0.4465721945614707,
+      "loss": 0.3110736906528473,
+      "loss_ce": 0.02005806751549244,
+      "loss_iou": 1.0133700370788574,
+      "loss_num": 0.291015625,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 201526272,
+      "step": 1166
+    },
+    {
+      "epoch": 0.44695518958253544,
+      "grad_norm": 23.933824697037963,
+      "learning_rate": 5e-06,
+      "loss": 0.2168,
+      "num_input_tokens_seen": 201699400,
+      "step": 1167
+    },
+    {
+      "epoch": 0.44695518958253544,
+      "loss": 0.23789691925048828,
+      "loss_ce": 0.018963806331157684,
+      "loss_iou": 1.0314826965332031,
+      "loss_num": 0.21875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 201699400,
+      "step": 1167
+    },
+    {
+      "epoch": 0.44733818460360014,
+      "grad_norm": 24.68663628579167,
+      "learning_rate": 5e-06,
+      "loss": 0.3253,
+      "num_input_tokens_seen": 201872616,
+      "step": 1168
+    },
+    {
+      "epoch": 0.44733818460360014,
+      "loss": 0.3806021809577942,
+      "loss_ce": 0.021593384444713593,
+      "loss_iou": 1.2345356941223145,
+      "loss_num": 0.359375,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 201872616,
+      "step": 1168
+    },
+    {
+      "epoch": 0.4477211796246649,
+      "grad_norm": 28.08736080430726,
+      "learning_rate": 5e-06,
+      "loss": 0.2538,
+      "num_input_tokens_seen": 202045792,
+      "step": 1169
+    },
+    {
+      "epoch": 0.4477211796246649,
+      "loss": 0.2737443447113037,
+      "loss_ce": 0.019593939185142517,
+      "loss_iou": 1.0906858444213867,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 202045792,
+      "step": 1169
+    },
+    {
+      "epoch": 0.4481041746457296,
+      "grad_norm": 39.28701170625111,
+      "learning_rate": 5e-06,
+      "loss": 0.29,
+      "num_input_tokens_seen": 202218752,
+      "step": 1170
+    },
+    {
+      "epoch": 0.4481041746457296,
+      "loss": 0.28742527961730957,
+      "loss_ce": 0.019969239830970764,
+      "loss_iou": 1.0095866918563843,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 202218752,
+      "step": 1170
+    },
+    {
+      "epoch": 0.4484871696667943,
+      "grad_norm": 33.28356494069342,
+      "learning_rate": 5e-06,
+      "loss": 0.2831,
+      "num_input_tokens_seen": 202391632,
+      "step": 1171
+    },
+    {
+      "epoch": 0.4484871696667943,
+      "loss": 0.2931366562843323,
+      "loss_ce": 0.020248495042324066,
+      "loss_iou": 1.1357561349868774,
+      "loss_num": 0.2734375,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 202391632,
+      "step": 1171
+    },
+    {
+      "epoch": 0.44887016468785906,
+      "grad_norm": 27.589215773927542,
+      "learning_rate": 5e-06,
+      "loss": 0.2546,
+      "num_input_tokens_seen": 202564520,
+      "step": 1172
+    },
+    {
+      "epoch": 0.44887016468785906,
+      "loss": 0.25870171189308167,
+      "loss_ce": 0.01993216946721077,
+      "loss_iou": 1.0024595260620117,
+      "loss_num": 0.23828125,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 202564520,
+      "step": 1172
+    },
+    {
+      "epoch": 0.44925315970892377,
+      "grad_norm": 35.92483723023636,
+      "learning_rate": 5e-06,
+      "loss": 0.2835,
+      "num_input_tokens_seen": 202737576,
+      "step": 1173
+    },
+    {
+      "epoch": 0.44925315970892377,
+      "loss": 0.2748756408691406,
+      "loss_ce": 0.020236970856785774,
+      "loss_iou": 1.0638651847839355,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 202737576,
+      "step": 1173
+    },
+    {
+      "epoch": 0.4496361547299885,
+      "grad_norm": 30.022342219911604,
+      "learning_rate": 5e-06,
+      "loss": 0.285,
+      "num_input_tokens_seen": 202910528,
+      "step": 1174
+    },
+    {
+      "epoch": 0.4496361547299885,
+      "loss": 0.2753847539424896,
+      "loss_ce": 0.020013663917779922,
+      "loss_iou": 1.0376520156860352,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 202910528,
+      "step": 1174
+    },
+    {
+      "epoch": 0.4500191497510532,
+      "grad_norm": 32.25365793654511,
+      "learning_rate": 5e-06,
+      "loss": 0.2748,
+      "num_input_tokens_seen": 203083688,
+      "step": 1175
+    },
+    {
+      "epoch": 0.4500191497510532,
+      "loss": 0.3484739065170288,
+      "loss_ce": 0.018884049728512764,
+      "loss_iou": 1.121877908706665,
+      "loss_num": 0.330078125,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 203083688,
+      "step": 1175
+    },
+    {
+      "epoch": 0.450402144772118,
+      "grad_norm": 28.833320703308424,
+      "learning_rate": 5e-06,
+      "loss": 0.2602,
+      "num_input_tokens_seen": 203257080,
+      "step": 1176
+    },
+    {
+      "epoch": 0.450402144772118,
+      "loss": 0.23676294088363647,
+      "loss_ce": 0.021675053983926773,
+      "loss_iou": 1.1116487979888916,
+      "loss_num": 0.21484375,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 203257080,
+      "step": 1176
+    },
+    {
+      "epoch": 0.4507851397931827,
+      "grad_norm": 31.152714343773237,
+      "learning_rate": 5e-06,
+      "loss": 0.2325,
+      "num_input_tokens_seen": 203430104,
+      "step": 1177
+    },
+    {
+      "epoch": 0.4507851397931827,
+      "loss": 0.2718905508518219,
+      "loss_ce": 0.0206698477268219,
+      "loss_iou": 1.0138851404190063,
+      "loss_num": 0.251953125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 203430104,
+      "step": 1177
+    },
+    {
+      "epoch": 0.4511681348142474,
+      "grad_norm": 32.222635202745295,
+      "learning_rate": 5e-06,
+      "loss": 0.2379,
+      "num_input_tokens_seen": 203603160,
+      "step": 1178
+    },
+    {
+      "epoch": 0.4511681348142474,
+      "loss": 0.2524706721305847,
+      "loss_ce": 0.01931639388203621,
+      "loss_iou": 1.0933088064193726,
+      "loss_num": 0.2333984375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 203603160,
+      "step": 1178
+    },
+    {
+      "epoch": 0.45155112983531215,
+      "grad_norm": 28.513552595286786,
+      "learning_rate": 5e-06,
+      "loss": 0.2808,
+      "num_input_tokens_seen": 203776208,
+      "step": 1179
+    },
+    {
+      "epoch": 0.45155112983531215,
+      "loss": 0.24923914670944214,
+      "loss_ce": 0.01986902952194214,
+      "loss_iou": 1.0238929986953735,
+      "loss_num": 0.2294921875,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 203776208,
+      "step": 1179
+    },
+    {
+      "epoch": 0.45193412485637685,
+      "grad_norm": 21.249686000705484,
+      "learning_rate": 5e-06,
+      "loss": 0.2621,
+      "num_input_tokens_seen": 203949184,
+      "step": 1180
+    },
+    {
+      "epoch": 0.45193412485637685,
+      "loss": 0.25357022881507874,
+      "loss_ce": 0.02023281529545784,
+      "loss_iou": 1.0669682025909424,
+      "loss_num": 0.2333984375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 203949184,
+      "step": 1180
+    },
+    {
+      "epoch": 0.4523171198774416,
+      "grad_norm": 26.917478057523944,
+      "learning_rate": 5e-06,
+      "loss": 0.2158,
+      "num_input_tokens_seen": 204122024,
+      "step": 1181
+    },
+    {
+      "epoch": 0.4523171198774416,
+      "loss": 0.22131076455116272,
+      "loss_ce": 0.01873507723212242,
+      "loss_iou": 1.0175056457519531,
+      "loss_num": 0.2021484375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 204122024,
+      "step": 1181
+    },
+    {
+      "epoch": 0.4527001148985063,
+      "grad_norm": 30.28490368060961,
+      "learning_rate": 5e-06,
+      "loss": 0.2525,
+      "num_input_tokens_seen": 204294960,
+      "step": 1182
+    },
+    {
+      "epoch": 0.4527001148985063,
+      "loss": 0.2743193209171295,
+      "loss_ce": 0.020291002467274666,
+      "loss_iou": 1.040724754333496,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 204294960,
+      "step": 1182
+    },
+    {
+      "epoch": 0.45308310991957107,
+      "grad_norm": 22.87707927828686,
+      "learning_rate": 5e-06,
+      "loss": 0.2321,
+      "num_input_tokens_seen": 204467368,
+      "step": 1183
+    },
+    {
+      "epoch": 0.45308310991957107,
+      "loss": 0.2445080280303955,
+      "loss_ce": 0.018800020217895508,
+      "loss_iou": 1.1185842752456665,
+      "loss_num": 0.2255859375,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 204467368,
+      "step": 1183
+    },
+    {
+      "epoch": 0.4534661049406358,
+      "grad_norm": 20.696879568895486,
+      "learning_rate": 5e-06,
+      "loss": 0.2679,
+      "num_input_tokens_seen": 204640280,
+      "step": 1184
+    },
+    {
+      "epoch": 0.4534661049406358,
+      "loss": 0.2555941045284271,
+      "loss_ce": 0.021341171115636826,
+      "loss_iou": 1.0109895467758179,
+      "loss_num": 0.234375,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 204640280,
+      "step": 1184
+    },
+    {
+      "epoch": 0.4538490999617005,
+      "grad_norm": 26.268558075458888,
+      "learning_rate": 5e-06,
+      "loss": 0.2105,
+      "num_input_tokens_seen": 204813240,
+      "step": 1185
+    },
+    {
+      "epoch": 0.4538490999617005,
+      "loss": 0.25668853521347046,
+      "loss_ce": 0.021214904263615608,
+      "loss_iou": 1.0567761659622192,
+      "loss_num": 0.2353515625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 204813240,
+      "step": 1185
+    },
+    {
+      "epoch": 0.45423209498276523,
+      "grad_norm": 32.61383844153683,
+      "learning_rate": 5e-06,
+      "loss": 0.2827,
+      "num_input_tokens_seen": 204986088,
+      "step": 1186
+    },
+    {
+      "epoch": 0.45423209498276523,
+      "loss": 0.33852893114089966,
+      "loss_ce": 0.019315045326948166,
+      "loss_iou": 1.038220763206482,
+      "loss_num": 0.318359375,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 204986088,
+      "step": 1186
+    },
+    {
+      "epoch": 0.45461509000382994,
+      "grad_norm": 28.72239339332055,
+      "learning_rate": 5e-06,
+      "loss": 0.2099,
+      "num_input_tokens_seen": 205159464,
+      "step": 1187
+    },
+    {
+      "epoch": 0.45461509000382994,
+      "loss": 0.14363889396190643,
+      "loss_ce": 0.02016477659344673,
+      "loss_iou": 1.0007100105285645,
+      "loss_num": 0.12353515625,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 205159464,
+      "step": 1187
+    },
+    {
+      "epoch": 0.4549980850248947,
+      "grad_norm": 29.17153599753967,
+      "learning_rate": 5e-06,
+      "loss": 0.2577,
+      "num_input_tokens_seen": 205332696,
+      "step": 1188
+    },
+    {
+      "epoch": 0.4549980850248947,
+      "loss": 0.266764760017395,
+      "loss_ce": 0.020182739943265915,
+      "loss_iou": 1.0024018287658691,
+      "loss_num": 0.24609375,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 205332696,
+      "step": 1188
+    },
+    {
+      "epoch": 0.4553810800459594,
+      "grad_norm": 28.532039716512504,
+      "learning_rate": 5e-06,
+      "loss": 0.1976,
+      "num_input_tokens_seen": 205505880,
+      "step": 1189
+    },
+    {
+      "epoch": 0.4553810800459594,
+      "loss": 0.2205185443162918,
+      "loss_ce": 0.02087254822254181,
+      "loss_iou": 1.0607324838638306,
+      "loss_num": 0.19921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 205505880,
+      "step": 1189
+    },
+    {
+      "epoch": 0.45576407506702415,
+      "grad_norm": 37.53982498463436,
+      "learning_rate": 5e-06,
+      "loss": 0.2872,
+      "num_input_tokens_seen": 205678504,
+      "step": 1190
+    },
+    {
+      "epoch": 0.45576407506702415,
+      "loss": 0.31850773096084595,
+      "loss_ce": 0.02065618336200714,
+      "loss_iou": 1.0170081853866577,
+      "loss_num": 0.296875,
+      "loss_xval": 0.296875,
+      "num_input_tokens_seen": 205678504,
+      "step": 1190
+    },
+    {
+      "epoch": 0.45614707008808886,
+      "grad_norm": 29.4571782160019,
+      "learning_rate": 5e-06,
+      "loss": 0.2473,
+      "num_input_tokens_seen": 205851752,
+      "step": 1191
+    },
+    {
+      "epoch": 0.45614707008808886,
+      "loss": 0.26461970806121826,
+      "loss_ce": 0.020173925906419754,
+      "loss_iou": 1.0164811611175537,
+      "loss_num": 0.244140625,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 205851752,
+      "step": 1191
+    },
+    {
+      "epoch": 0.45653006510915356,
+      "grad_norm": 30.10232987030623,
+      "learning_rate": 5e-06,
+      "loss": 0.3273,
+      "num_input_tokens_seen": 206024416,
+      "step": 1192
+    },
+    {
+      "epoch": 0.45653006510915356,
+      "loss": 0.2999013662338257,
+      "loss_ce": 0.020177241414785385,
+      "loss_iou": 1.0649040937423706,
+      "loss_num": 0.279296875,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 206024416,
+      "step": 1192
+    },
+    {
+      "epoch": 0.4569130601302183,
+      "grad_norm": 46.115333340752976,
+      "learning_rate": 5e-06,
+      "loss": 0.2377,
+      "num_input_tokens_seen": 206197440,
+      "step": 1193
+    },
+    {
+      "epoch": 0.4569130601302183,
+      "loss": 0.23769903182983398,
+      "loss_ce": 0.021268364042043686,
+      "loss_iou": 1.0191850662231445,
+      "loss_num": 0.216796875,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 206197440,
+      "step": 1193
+    },
+    {
+      "epoch": 0.457296055151283,
+      "grad_norm": 43.7152421239042,
+      "learning_rate": 5e-06,
+      "loss": 0.2804,
+      "num_input_tokens_seen": 206370312,
+      "step": 1194
+    },
+    {
+      "epoch": 0.457296055151283,
+      "loss": 0.3282549977302551,
+      "loss_ce": 0.020637793466448784,
+      "loss_iou": 0.9924590587615967,
+      "loss_num": 0.30859375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 206370312,
+      "step": 1194
+    },
+    {
+      "epoch": 0.4576790501723478,
+      "grad_norm": 33.74389350078248,
+      "learning_rate": 5e-06,
+      "loss": 0.2559,
+      "num_input_tokens_seen": 206543424,
+      "step": 1195
+    },
+    {
+      "epoch": 0.4576790501723478,
+      "loss": 0.3125259578227997,
+      "loss_ce": 0.020289622247219086,
+      "loss_iou": 1.0566866397857666,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 206543424,
+      "step": 1195
+    },
+    {
+      "epoch": 0.4580620451934125,
+      "grad_norm": 39.31242299560097,
+      "learning_rate": 5e-06,
+      "loss": 0.2625,
+      "num_input_tokens_seen": 206716040,
+      "step": 1196
+    },
+    {
+      "epoch": 0.4580620451934125,
+      "loss": 0.21807287633419037,
+      "loss_ce": 0.019098270684480667,
+      "loss_iou": 1.010589838027954,
+      "loss_num": 0.19921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 206716040,
+      "step": 1196
+    },
+    {
+      "epoch": 0.4584450402144772,
+      "grad_norm": 49.293738253540674,
+      "learning_rate": 5e-06,
+      "loss": 0.2648,
+      "num_input_tokens_seen": 206888584,
+      "step": 1197
+    },
+    {
+      "epoch": 0.4584450402144772,
+      "loss": 0.2763878107070923,
+      "loss_ce": 0.02113880030810833,
+      "loss_iou": 1.055002212524414,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 206888584,
+      "step": 1197
+    },
+    {
+      "epoch": 0.45882803523554194,
+      "grad_norm": 33.281452562256966,
+      "learning_rate": 5e-06,
+      "loss": 0.3585,
+      "num_input_tokens_seen": 207061368,
+      "step": 1198
+    },
+    {
+      "epoch": 0.45882803523554194,
+      "loss": 0.4192292094230652,
+      "loss_ce": 0.02018134854733944,
+      "loss_iou": 1.1942784786224365,
+      "loss_num": 0.3984375,
+      "loss_xval": 0.3984375,
+      "num_input_tokens_seen": 207061368,
+      "step": 1198
+    },
+    {
+      "epoch": 0.45921103025660664,
+      "grad_norm": 27.32779327998403,
+      "learning_rate": 5e-06,
+      "loss": 0.233,
+      "num_input_tokens_seen": 207234448,
+      "step": 1199
+    },
+    {
+      "epoch": 0.45921103025660664,
+      "loss": 0.26918455958366394,
+      "loss_ce": 0.020100083202123642,
+      "loss_iou": 1.0386629104614258,
+      "loss_num": 0.2490234375,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 207234448,
+      "step": 1199
+    },
+    {
+      "epoch": 0.4595940252776714,
+      "grad_norm": 34.42167959598355,
+      "learning_rate": 5e-06,
+      "loss": 0.2061,
+      "num_input_tokens_seen": 207407368,
+      "step": 1200
+    },
+    {
+      "epoch": 0.4595940252776714,
+      "loss": 0.2239762842655182,
+      "loss_ce": 0.019752655178308487,
+      "loss_iou": 1.047470211982727,
+      "loss_num": 0.2041015625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 207407368,
+      "step": 1200
+    },
+    {
+      "epoch": 0.4599770202987361,
+      "grad_norm": 33.34176773218534,
+      "learning_rate": 5e-06,
+      "loss": 0.2566,
+      "num_input_tokens_seen": 207580336,
+      "step": 1201
+    },
+    {
+      "epoch": 0.4599770202987361,
+      "loss": 0.25941014289855957,
+      "loss_ce": 0.019908176735043526,
+      "loss_iou": 1.0563254356384277,
+      "loss_num": 0.2392578125,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 207580336,
+      "step": 1201
+    },
+    {
+      "epoch": 0.46036001531980086,
+      "grad_norm": 29.238023332602307,
+      "learning_rate": 5e-06,
+      "loss": 0.2481,
+      "num_input_tokens_seen": 207753496,
+      "step": 1202
+    },
+    {
+      "epoch": 0.46036001531980086,
+      "loss": 0.19793447852134705,
+      "loss_ce": 0.020932529121637344,
+      "loss_iou": 1.0017145872116089,
+      "loss_num": 0.1767578125,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 207753496,
+      "step": 1202
+    },
+    {
+      "epoch": 0.46074301034086557,
+      "grad_norm": 37.123953292975,
+      "learning_rate": 5e-06,
+      "loss": 0.3079,
+      "num_input_tokens_seen": 207926736,
+      "step": 1203
+    },
+    {
+      "epoch": 0.46074301034086557,
+      "loss": 0.3396281599998474,
+      "loss_ce": 0.019437748938798904,
+      "loss_iou": 1.0252783298492432,
+      "loss_num": 0.3203125,
+      "loss_xval": 0.3203125,
+      "num_input_tokens_seen": 207926736,
+      "step": 1203
+    },
+    {
+      "epoch": 0.46112600536193027,
+      "grad_norm": 49.708482696429726,
+      "learning_rate": 5e-06,
+      "loss": 0.3337,
+      "num_input_tokens_seen": 208099768,
+      "step": 1204
+    },
+    {
+      "epoch": 0.46112600536193027,
+      "loss": 0.30779531598091125,
+      "loss_ce": 0.020075593143701553,
+      "loss_iou": 1.01540207862854,
+      "loss_num": 0.287109375,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 208099768,
+      "step": 1204
+    },
+    {
+      "epoch": 0.461509000382995,
+      "grad_norm": 33.38269900294354,
+      "learning_rate": 5e-06,
+      "loss": 0.3222,
+      "num_input_tokens_seen": 208273040,
+      "step": 1205
+    },
+    {
+      "epoch": 0.461509000382995,
+      "loss": 0.32141298055648804,
+      "loss_ce": 0.020265530794858932,
+      "loss_iou": 1.094458818435669,
+      "loss_num": 0.30078125,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 208273040,
+      "step": 1205
+    },
+    {
+      "epoch": 0.46189199540405973,
+      "grad_norm": 32.721773895186935,
+      "learning_rate": 5e-06,
+      "loss": 0.2943,
+      "num_input_tokens_seen": 208445712,
+      "step": 1206
+    },
+    {
+      "epoch": 0.46189199540405973,
+      "loss": 0.30638575553894043,
+      "loss_ce": 0.020374994724988937,
+      "loss_iou": 1.0706769227981567,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 208445712,
+      "step": 1206
+    },
+    {
+      "epoch": 0.4622749904251245,
+      "grad_norm": 62.49189654564097,
+      "learning_rate": 5e-06,
+      "loss": 0.3114,
+      "num_input_tokens_seen": 208618688,
+      "step": 1207
+    },
+    {
+      "epoch": 0.4622749904251245,
+      "loss": 0.29980647563934326,
+      "loss_ce": 0.020387519150972366,
+      "loss_iou": 1.1299349069595337,
+      "loss_num": 0.279296875,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 208618688,
+      "step": 1207
+    },
+    {
+      "epoch": 0.4626579854461892,
+      "grad_norm": 83.33686582974181,
+      "learning_rate": 5e-06,
+      "loss": 0.5361,
+      "num_input_tokens_seen": 208791936,
+      "step": 1208
+    },
+    {
+      "epoch": 0.4626579854461892,
+      "loss": 0.5459185838699341,
+      "loss_ce": 0.021748628467321396,
+      "loss_iou": 1.1664130687713623,
+      "loss_num": 0.5234375,
+      "loss_xval": 0.5234375,
+      "num_input_tokens_seen": 208791936,
+      "step": 1208
+    },
+    {
+      "epoch": 0.46304098046725395,
+      "grad_norm": 32.498828012614354,
+      "learning_rate": 5e-06,
+      "loss": 0.3044,
+      "num_input_tokens_seen": 208965216,
+      "step": 1209
+    },
+    {
+      "epoch": 0.46304098046725395,
+      "loss": 0.3312862813472748,
+      "loss_ce": 0.02061733417212963,
+      "loss_iou": 1.144300103187561,
+      "loss_num": 0.310546875,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 208965216,
+      "step": 1209
+    },
+    {
+      "epoch": 0.46342397548831865,
+      "grad_norm": 38.343275183035935,
+      "learning_rate": 5e-06,
+      "loss": 0.37,
+      "num_input_tokens_seen": 209137952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.46342397548831865,
+      "loss": 0.34652167558670044,
+      "loss_ce": 0.019739460200071335,
+      "loss_iou": 0.8828321695327759,
+      "loss_num": 0.326171875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 209137952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.46380697050938335,
+      "grad_norm": 51.65954090616292,
+      "learning_rate": 5e-06,
+      "loss": 0.3785,
+      "num_input_tokens_seen": 209310688,
+      "step": 1211
+    },
+    {
+      "epoch": 0.46380697050938335,
+      "loss": 0.3275665044784546,
+      "loss_ce": 0.02055967040359974,
+      "loss_iou": 1.0090607404708862,
+      "loss_num": 0.306640625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 209310688,
+      "step": 1211
+    },
+    {
+      "epoch": 0.4641899655304481,
+      "grad_norm": 43.65917138990877,
+      "learning_rate": 5e-06,
+      "loss": 0.3131,
+      "num_input_tokens_seen": 209483632,
+      "step": 1212
+    },
+    {
+      "epoch": 0.4641899655304481,
+      "loss": 0.319579541683197,
+      "loss_ce": 0.02014104649424553,
+      "loss_iou": 1.007224202156067,
+      "loss_num": 0.298828125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 209483632,
+      "step": 1212
+    },
+    {
+      "epoch": 0.4645729605515128,
+      "grad_norm": 21.208706371704558,
+      "learning_rate": 5e-06,
+      "loss": 0.2875,
+      "num_input_tokens_seen": 209656520,
+      "step": 1213
+    },
+    {
+      "epoch": 0.4645729605515128,
+      "loss": 0.23015263676643372,
+      "loss_ce": 0.018482713028788567,
+      "loss_iou": 1.0930778980255127,
+      "loss_num": 0.2119140625,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 209656520,
+      "step": 1213
+    },
+    {
+      "epoch": 0.4649559555725776,
+      "grad_norm": 22.846096589976895,
+      "learning_rate": 5e-06,
+      "loss": 0.2401,
+      "num_input_tokens_seen": 209829808,
+      "step": 1214
+    },
+    {
+      "epoch": 0.4649559555725776,
+      "loss": 0.2387043833732605,
+      "loss_ce": 0.021419227123260498,
+      "loss_iou": 1.1203703880310059,
+      "loss_num": 0.216796875,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 209829808,
+      "step": 1214
+    },
+    {
+      "epoch": 0.4653389505936423,
+      "grad_norm": 21.76714792430826,
+      "learning_rate": 5e-06,
+      "loss": 0.2076,
+      "num_input_tokens_seen": 210002408,
+      "step": 1215
+    },
+    {
+      "epoch": 0.4653389505936423,
+      "loss": 0.19137930870056152,
+      "loss_ce": 0.020480871200561523,
+      "loss_iou": 1.027408480644226,
+      "loss_num": 0.1708984375,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 210002408,
+      "step": 1215
+    },
+    {
+      "epoch": 0.46572194561470703,
+      "grad_norm": 42.78407183247811,
+      "learning_rate": 5e-06,
+      "loss": 0.2315,
+      "num_input_tokens_seen": 210175520,
+      "step": 1216
+    },
+    {
+      "epoch": 0.46572194561470703,
+      "loss": 0.2105085402727127,
+      "loss_ce": 0.021726813167333603,
+      "loss_iou": 1.0148282051086426,
+      "loss_num": 0.1884765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 210175520,
+      "step": 1216
+    },
+    {
+      "epoch": 0.46610494063577174,
+      "grad_norm": 42.92081573795019,
+      "learning_rate": 5e-06,
+      "loss": 0.2418,
+      "num_input_tokens_seen": 210348544,
+      "step": 1217
+    },
+    {
+      "epoch": 0.46610494063577174,
+      "loss": 0.24414533376693726,
+      "loss_ce": 0.021000802516937256,
+      "loss_iou": 1.0263220071792603,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 210348544,
+      "step": 1217
+    },
+    {
+      "epoch": 0.46648793565683644,
+      "grad_norm": 25.986169050505218,
+      "learning_rate": 5e-06,
+      "loss": 0.2528,
+      "num_input_tokens_seen": 210521272,
+      "step": 1218
+    },
+    {
+      "epoch": 0.46648793565683644,
+      "loss": 0.30587664246559143,
+      "loss_ce": 0.02169695310294628,
+      "loss_iou": 1.0758498907089233,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 210521272,
+      "step": 1218
+    },
+    {
+      "epoch": 0.4668709306779012,
+      "grad_norm": 26.45532019578287,
+      "learning_rate": 5e-06,
+      "loss": 0.2786,
+      "num_input_tokens_seen": 210694296,
+      "step": 1219
+    },
+    {
+      "epoch": 0.4668709306779012,
+      "loss": 0.3204804062843323,
+      "loss_ce": 0.020248491317033768,
+      "loss_iou": 1.3104071617126465,
+      "loss_num": 0.30078125,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 210694296,
+      "step": 1219
+    },
+    {
+      "epoch": 0.4672539256989659,
+      "grad_norm": 29.044389132790407,
+      "learning_rate": 5e-06,
+      "loss": 0.1978,
+      "num_input_tokens_seen": 210867208,
+      "step": 1220
+    },
+    {
+      "epoch": 0.4672539256989659,
+      "loss": 0.2194499671459198,
+      "loss_ce": 0.019559821113944054,
+      "loss_iou": 0.885863184928894,
+      "loss_num": 0.2001953125,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 210867208,
+      "step": 1220
+    },
+    {
+      "epoch": 0.46763692072003066,
+      "grad_norm": 32.577022795819815,
+      "learning_rate": 5e-06,
+      "loss": 0.2507,
+      "num_input_tokens_seen": 211040032,
+      "step": 1221
+    },
+    {
+      "epoch": 0.46763692072003066,
+      "loss": 0.2515682876110077,
+      "loss_ce": 0.021221602335572243,
+      "loss_iou": 1.0183402299880981,
+      "loss_num": 0.23046875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 211040032,
+      "step": 1221
+    },
+    {
+      "epoch": 0.46801991574109536,
+      "grad_norm": 29.22541175312178,
+      "learning_rate": 5e-06,
+      "loss": 0.279,
+      "num_input_tokens_seen": 211213408,
+      "step": 1222
+    },
+    {
+      "epoch": 0.46801991574109536,
+      "loss": 0.23782938718795776,
+      "loss_ce": 0.020239055156707764,
+      "loss_iou": 1.0102102756500244,
+      "loss_num": 0.2177734375,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 211213408,
+      "step": 1222
+    },
+    {
+      "epoch": 0.4684029107621601,
+      "grad_norm": 21.64071661144197,
+      "learning_rate": 5e-06,
+      "loss": 0.3053,
+      "num_input_tokens_seen": 211386344,
+      "step": 1223
+    },
+    {
+      "epoch": 0.4684029107621601,
+      "loss": 0.27443018555641174,
+      "loss_ce": 0.02027980238199234,
+      "loss_iou": 0.9637537598609924,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 211386344,
+      "step": 1223
+    },
+    {
+      "epoch": 0.4687859057832248,
+      "grad_norm": 28.11459993389933,
+      "learning_rate": 5e-06,
+      "loss": 0.2068,
+      "num_input_tokens_seen": 211559856,
+      "step": 1224
+    },
+    {
+      "epoch": 0.4687859057832248,
+      "loss": 0.2193112075328827,
+      "loss_ce": 0.020580727607011795,
+      "loss_iou": 1.0139524936676025,
+      "loss_num": 0.19921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 211559856,
+      "step": 1224
+    },
+    {
+      "epoch": 0.4691689008042895,
+      "grad_norm": 36.428761035536986,
+      "learning_rate": 5e-06,
+      "loss": 0.2531,
+      "num_input_tokens_seen": 211733248,
+      "step": 1225
+    },
+    {
+      "epoch": 0.4691689008042895,
+      "loss": 0.3093939423561096,
+      "loss_ce": 0.01972106657922268,
+      "loss_iou": 1.015163779258728,
+      "loss_num": 0.2890625,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 211733248,
+      "step": 1225
+    },
+    {
+      "epoch": 0.4695518958253543,
+      "grad_norm": 39.01721544263482,
+      "learning_rate": 5e-06,
+      "loss": 0.2558,
+      "num_input_tokens_seen": 211906336,
+      "step": 1226
+    },
+    {
+      "epoch": 0.4695518958253543,
+      "loss": 0.26606816053390503,
+      "loss_ce": 0.021134087815880775,
+      "loss_iou": -1.7795388064727341e+28,
+      "loss_num": 0.2451171875,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 211906336,
+      "step": 1226
+    },
+    {
+      "epoch": 0.469934890846419,
+      "grad_norm": 27.60863119072085,
+      "learning_rate": 5e-06,
+      "loss": 0.2386,
+      "num_input_tokens_seen": 212078952,
+      "step": 1227
+    },
+    {
+      "epoch": 0.469934890846419,
+      "loss": 0.20051205158233643,
+      "loss_ce": 0.020397303625941277,
+      "loss_iou": 1.0308964252471924,
+      "loss_num": 0.1796875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 212078952,
+      "step": 1227
+    },
+    {
+      "epoch": 0.47031788586748374,
+      "grad_norm": 25.995949679468232,
+      "learning_rate": 5e-06,
+      "loss": 0.2314,
+      "num_input_tokens_seen": 212251880,
+      "step": 1228
+    },
+    {
+      "epoch": 0.47031788586748374,
+      "loss": 0.19911321997642517,
+      "loss_ce": 0.02076849900186062,
+      "loss_iou": 1.0025396347045898,
+      "loss_num": 0.1787109375,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 212251880,
+      "step": 1228
+    },
+    {
+      "epoch": 0.47070088088854845,
+      "grad_norm": 36.32822038372454,
+      "learning_rate": 5e-06,
+      "loss": 0.2005,
+      "num_input_tokens_seen": 212424800,
+      "step": 1229
+    },
+    {
+      "epoch": 0.47070088088854845,
+      "loss": 0.2403562068939209,
+      "loss_ce": 0.0217282734811306,
+      "loss_iou": 1.041759729385376,
+      "loss_num": 0.21875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 212424800,
+      "step": 1229
+    },
+    {
+      "epoch": 0.47108387590961315,
+      "grad_norm": 42.008764834368776,
+      "learning_rate": 5e-06,
+      "loss": 0.257,
+      "num_input_tokens_seen": 212597832,
+      "step": 1230
+    },
+    {
+      "epoch": 0.47108387590961315,
+      "loss": 0.26775553822517395,
+      "loss_ce": 0.0201359074562788,
+      "loss_iou": 1.010007619857788,
+      "loss_num": 0.248046875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 212597832,
+      "step": 1230
+    },
+    {
+      "epoch": 0.4714668709306779,
+      "grad_norm": 35.65222787113635,
+      "learning_rate": 5e-06,
+      "loss": 0.1874,
+      "num_input_tokens_seen": 212770320,
+      "step": 1231
+    },
+    {
+      "epoch": 0.4714668709306779,
+      "loss": 0.19671890139579773,
+      "loss_ce": 0.02063247561454773,
+      "loss_iou": 1.0493252277374268,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 212770320,
+      "step": 1231
+    },
+    {
+      "epoch": 0.4718498659517426,
+      "grad_norm": 31.942214418097883,
+      "learning_rate": 5e-06,
+      "loss": 0.2551,
+      "num_input_tokens_seen": 212943368,
+      "step": 1232
+    },
+    {
+      "epoch": 0.4718498659517426,
+      "loss": 0.3143588602542877,
+      "loss_ce": 0.022122547030448914,
+      "loss_iou": 1.219099760055542,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 212943368,
+      "step": 1232
+    },
+    {
+      "epoch": 0.47223286097280737,
+      "grad_norm": 33.31657235703236,
+      "learning_rate": 5e-06,
+      "loss": 0.2829,
+      "num_input_tokens_seen": 213116424,
+      "step": 1233
+    },
+    {
+      "epoch": 0.47223286097280737,
+      "loss": 0.30633944272994995,
+      "loss_ce": 0.021000079810619354,
+      "loss_iou": 1.0062294006347656,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 213116424,
+      "step": 1233
+    },
+    {
+      "epoch": 0.47261585599387207,
+      "grad_norm": 33.776484152274676,
+      "learning_rate": 5e-06,
+      "loss": 0.2491,
+      "num_input_tokens_seen": 213289256,
+      "step": 1234
+    },
+    {
+      "epoch": 0.47261585599387207,
+      "loss": 0.2458813637495041,
+      "loss_ce": 0.02114991284906864,
+      "loss_iou": 1.0139302015304565,
+      "loss_num": 0.224609375,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 213289256,
+      "step": 1234
+    },
+    {
+      "epoch": 0.4729988510149368,
+      "grad_norm": 23.591833356456384,
+      "learning_rate": 5e-06,
+      "loss": 0.226,
+      "num_input_tokens_seen": 213462384,
+      "step": 1235
+    },
+    {
+      "epoch": 0.4729988510149368,
+      "loss": 0.2640923261642456,
+      "loss_ce": 0.02001270279288292,
+      "loss_iou": 1.134010910987854,
+      "loss_num": 0.244140625,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 213462384,
+      "step": 1235
+    },
+    {
+      "epoch": 0.47338184603600153,
+      "grad_norm": 23.575302914366794,
+      "learning_rate": 5e-06,
+      "loss": 0.2856,
+      "num_input_tokens_seen": 213635320,
+      "step": 1236
+    },
+    {
+      "epoch": 0.47338184603600153,
+      "loss": 0.3351038694381714,
+      "loss_ce": 0.021139021962881088,
+      "loss_iou": 1.0476080179214478,
+      "loss_num": 0.314453125,
+      "loss_xval": 0.314453125,
+      "num_input_tokens_seen": 213635320,
+      "step": 1236
+    },
+    {
+      "epoch": 0.47376484105706623,
+      "grad_norm": 27.247542152040623,
+      "learning_rate": 5e-06,
+      "loss": 0.2751,
+      "num_input_tokens_seen": 213808440,
+      "step": 1237
+    },
+    {
+      "epoch": 0.47376484105706623,
+      "loss": 0.33616071939468384,
+      "loss_ce": 0.02085307240486145,
+      "loss_iou": 1.0902354717254639,
+      "loss_num": 0.314453125,
+      "loss_xval": 0.314453125,
+      "num_input_tokens_seen": 213808440,
+      "step": 1237
+    },
+    {
+      "epoch": 0.474147836078131,
+      "grad_norm": 29.056772296303375,
+      "learning_rate": 5e-06,
+      "loss": 0.2855,
+      "num_input_tokens_seen": 213981424,
+      "step": 1238
+    },
+    {
+      "epoch": 0.474147836078131,
+      "loss": 0.2904357314109802,
+      "loss_ce": 0.020721357315778732,
+      "loss_iou": 1.0729703903198242,
+      "loss_num": 0.26953125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 213981424,
+      "step": 1238
+    },
+    {
+      "epoch": 0.4745308310991957,
+      "grad_norm": 25.19565289524144,
+      "learning_rate": 5e-06,
+      "loss": 0.2261,
+      "num_input_tokens_seen": 214154408,
+      "step": 1239
+    },
+    {
+      "epoch": 0.4745308310991957,
+      "loss": 0.19168391823768616,
+      "loss_ce": 0.01980891078710556,
+      "loss_iou": 1.0037028789520264,
+      "loss_num": 0.171875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 214154408,
+      "step": 1239
+    },
+    {
+      "epoch": 0.47491382612026045,
+      "grad_norm": 28.591882792030717,
+      "learning_rate": 5e-06,
+      "loss": 0.3062,
+      "num_input_tokens_seen": 214327432,
+      "step": 1240
+    },
+    {
+      "epoch": 0.47491382612026045,
+      "loss": 0.2543603777885437,
+      "loss_ce": 0.022243663668632507,
+      "loss_iou": 1.098926305770874,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 214327432,
+      "step": 1240
+    },
+    {
+      "epoch": 0.47529682114132515,
+      "grad_norm": 27.2066659745211,
+      "learning_rate": 5e-06,
+      "loss": 0.202,
+      "num_input_tokens_seen": 214500088,
+      "step": 1241
+    },
+    {
+      "epoch": 0.47529682114132515,
+      "loss": 0.2134188413619995,
+      "loss_ce": 0.019815323874354362,
+      "loss_iou": 0.9447759389877319,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 214500088,
+      "step": 1241
+    },
+    {
+      "epoch": 0.4756798161623899,
+      "grad_norm": 31.714890019626516,
+      "learning_rate": 5e-06,
+      "loss": 0.2822,
+      "num_input_tokens_seen": 214673072,
+      "step": 1242
+    },
+    {
+      "epoch": 0.4756798161623899,
+      "loss": 0.3018686771392822,
+      "loss_ce": 0.020069342106580734,
+      "loss_iou": 1.040618658065796,
+      "loss_num": 0.28125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 214673072,
+      "step": 1242
+    },
+    {
+      "epoch": 0.4760628111834546,
+      "grad_norm": 41.539759519045866,
+      "learning_rate": 5e-06,
+      "loss": 0.2599,
+      "num_input_tokens_seen": 214845760,
+      "step": 1243
+    },
+    {
+      "epoch": 0.4760628111834546,
+      "loss": 0.2806388735771179,
+      "loss_ce": 0.019042205065488815,
+      "loss_iou": 1.0721485614776611,
+      "loss_num": 0.26171875,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 214845760,
+      "step": 1243
+    },
+    {
+      "epoch": 0.4764458062045193,
+      "grad_norm": 26.00145575425882,
+      "learning_rate": 5e-06,
+      "loss": 0.2321,
+      "num_input_tokens_seen": 215018680,
+      "step": 1244
+    },
+    {
+      "epoch": 0.4764458062045193,
+      "loss": 0.22874602675437927,
+      "loss_ce": 0.01939544454216957,
+      "loss_iou": 1.001299500465393,
+      "loss_num": 0.208984375,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 215018680,
+      "step": 1244
+    },
+    {
+      "epoch": 0.4768288012255841,
+      "grad_norm": 41.614299138596365,
+      "learning_rate": 5e-06,
+      "loss": 0.2361,
+      "num_input_tokens_seen": 215191304,
+      "step": 1245
+    },
+    {
+      "epoch": 0.4768288012255841,
+      "loss": 0.2315431535243988,
+      "loss_ce": 0.0205446258187294,
+      "loss_iou": 1.0208760499954224,
+      "loss_num": 0.2109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 215191304,
+      "step": 1245
+    },
+    {
+      "epoch": 0.4772117962466488,
+      "grad_norm": 47.040980481672904,
+      "learning_rate": 5e-06,
+      "loss": 0.2758,
+      "num_input_tokens_seen": 215364248,
+      "step": 1246
+    },
+    {
+      "epoch": 0.4772117962466488,
+      "loss": 0.28870445489883423,
+      "loss_ce": 0.020393896847963333,
+      "loss_iou": 1.137253999710083,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 215364248,
+      "step": 1246
+    },
+    {
+      "epoch": 0.47759479126771354,
+      "grad_norm": 36.53723029226407,
+      "learning_rate": 5e-06,
+      "loss": 0.2395,
+      "num_input_tokens_seen": 215537144,
+      "step": 1247
+    },
+    {
+      "epoch": 0.47759479126771354,
+      "loss": 0.24715638160705566,
+      "loss_ce": 0.020166628062725067,
+      "loss_iou": 1.118346929550171,
+      "loss_num": 0.2265625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 215537144,
+      "step": 1247
+    },
+    {
+      "epoch": 0.47797778628877824,
+      "grad_norm": 28.875781579919764,
+      "learning_rate": 5e-06,
+      "loss": 0.1856,
+      "num_input_tokens_seen": 215710080,
+      "step": 1248
+    },
+    {
+      "epoch": 0.47797778628877824,
+      "loss": 0.18068018555641174,
+      "loss_ce": 0.02168361470103264,
+      "loss_iou": 1.0261390209197998,
+      "loss_num": 0.1591796875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 215710080,
+      "step": 1248
+    },
+    {
+      "epoch": 0.478360781309843,
+      "grad_norm": 34.72003795966805,
+      "learning_rate": 5e-06,
+      "loss": 0.205,
+      "num_input_tokens_seen": 215883128,
+      "step": 1249
+    },
+    {
+      "epoch": 0.478360781309843,
+      "loss": 0.16687241196632385,
+      "loss_ce": 0.021730797365307808,
+      "loss_iou": 1.0021653175354004,
+      "loss_num": 0.1455078125,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 215883128,
+      "step": 1249
+    },
+    {
+      "epoch": 0.4787437763309077,
+      "grad_norm": 55.81362605658789,
+      "learning_rate": 5e-06,
+      "loss": 0.2416,
+      "num_input_tokens_seen": 216056208,
+      "step": 1250
+    },
+    {
+      "epoch": 0.4787437763309077,
+      "eval_websight_new_CIoU": 0.8993282616138458,
+      "eval_websight_new_GIoU": 0.8991754651069641,
+      "eval_websight_new_IoU": 0.8996633589267731,
+      "eval_websight_new_MAE_all": 0.011344741564244032,
+      "eval_websight_new_MAE_h": 0.013371880631893873,
+      "eval_websight_new_MAE_w": 0.01601374614983797,
+      "eval_websight_new_MAE_x": 0.009378236252814531,
+      "eval_websight_new_MAE_y": 0.006615103920921683,
+      "eval_websight_new_NUM_probability": 9.162682181340642e-05,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.16499315202236176,
+      "eval_websight_new_loss_ce": 0.0413089245557785,
+      "eval_websight_new_loss_iou": 1.0006011128425598,
+      "eval_websight_new_loss_num": 0.125,
+      "eval_websight_new_loss_xval": 0.125,
+      "eval_websight_new_runtime": 56.5573,
+      "eval_websight_new_samples_per_second": 0.884,
+      "eval_websight_new_steps_per_second": 0.035,
+      "num_input_tokens_seen": 216056208,
+      "step": 1250
+    },
+    {
+      "epoch": 0.4787437763309077,
+      "eval_seeclick_CIoU": 0.625552624464035,
+      "eval_seeclick_GIoU": 0.6292989253997803,
+      "eval_seeclick_IoU": 0.6577684879302979,
+      "eval_seeclick_MAE_all": 0.06135014072060585,
+      "eval_seeclick_MAE_h": 0.05675686150789261,
+      "eval_seeclick_MAE_w": 0.07418486848473549,
+      "eval_seeclick_MAE_x": 0.06162244640290737,
+      "eval_seeclick_MAE_y": 0.05283636972308159,
+      "eval_seeclick_NUM_probability": 0.0001246245992660988,
+      "eval_seeclick_inside_bbox": 0.8975694477558136,
+      "eval_seeclick_loss": 0.5222324132919312,
+      "eval_seeclick_loss_ce": 0.032219236716628075,
+      "eval_seeclick_loss_iou": 1.1460830569267273,
+      "eval_seeclick_loss_num": 0.46142578125,
+      "eval_seeclick_loss_xval": 0.46142578125,
+      "eval_seeclick_runtime": 85.2578,
+      "eval_seeclick_samples_per_second": 0.586,
+      "eval_seeclick_steps_per_second": 0.023,
+      "num_input_tokens_seen": 216056208,
+      "step": 1250
+    },
+    {
+      "epoch": 0.4787437763309077,
+      "eval_icons_CIoU": 0.8609585464000702,
+      "eval_icons_GIoU": 0.8591577708721161,
+      "eval_icons_IoU": 0.8646650612354279,
+      "eval_icons_MAE_all": 0.01873237080872059,
+      "eval_icons_MAE_h": 0.02262669149786234,
+      "eval_icons_MAE_w": 0.022232558578252792,
+      "eval_icons_MAE_x": 0.014843590091913939,
+      "eval_icons_MAE_y": 0.015226639807224274,
+      "eval_icons_NUM_probability": 9.073268302017823e-05,
+      "eval_icons_inside_bbox": 1.0,
+      "eval_icons_loss": 0.17036914825439453,
+      "eval_icons_loss_ce": 0.027757744304835796,
+      "eval_icons_loss_iou": 1.0137335062026978,
+      "eval_icons_loss_num": 0.133453369140625,
+      "eval_icons_loss_xval": 0.133453369140625,
+      "eval_icons_runtime": 86.4594,
+      "eval_icons_samples_per_second": 0.578,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 216056208,
+      "step": 1250
+    },
+    {
+      "epoch": 0.4787437763309077,
+      "loss": 0.17677251994609833,
+      "loss_ce": 0.02778569608926773,
+      "loss_iou": 1.0268021821975708,
+      "loss_num": 0.1494140625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 216056208,
+      "step": 1250
+    },
+    {
+      "epoch": 0.4791267713519724,
+      "grad_norm": 40.001223857015134,
+      "learning_rate": 5e-06,
+      "loss": 0.1946,
+      "num_input_tokens_seen": 216229520,
+      "step": 1251
+    },
+    {
+      "epoch": 0.4791267713519724,
+      "loss": 0.18243470788002014,
+      "loss_ce": 0.02111878991127014,
+      "loss_iou": 1.01613187789917,
+      "loss_num": 0.1611328125,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 216229520,
+      "step": 1251
+    },
+    {
+      "epoch": 0.47950976637303716,
+      "grad_norm": 31.521366519005536,
+      "learning_rate": 5e-06,
+      "loss": 0.2673,
+      "num_input_tokens_seen": 216402744,
+      "step": 1252
+    },
+    {
+      "epoch": 0.47950976637303716,
+      "loss": 0.26666757464408875,
+      "loss_ce": 0.021550390869379044,
+      "loss_iou": 1.0124704837799072,
+      "loss_num": 0.2451171875,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 216402744,
+      "step": 1252
+    },
+    {
+      "epoch": 0.47989276139410186,
+      "grad_norm": 28.036570071899565,
+      "learning_rate": 5e-06,
+      "loss": 0.2371,
+      "num_input_tokens_seen": 216575744,
+      "step": 1253
+    },
+    {
+      "epoch": 0.47989276139410186,
+      "loss": 0.3153555393218994,
+      "loss_ce": 0.020647302269935608,
+      "loss_iou": 1.0578529834747314,
+      "loss_num": 0.294921875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 216575744,
+      "step": 1253
+    },
+    {
+      "epoch": 0.4802757564151666,
+      "grad_norm": 33.72330941098161,
+      "learning_rate": 5e-06,
+      "loss": 0.2868,
+      "num_input_tokens_seen": 216748136,
+      "step": 1254
+    },
+    {
+      "epoch": 0.4802757564151666,
+      "loss": 0.28596433997154236,
+      "loss_ce": 0.020095188170671463,
+      "loss_iou": 1.083143711090088,
+      "loss_num": 0.265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 216748136,
+      "step": 1254
+    },
+    {
+      "epoch": 0.4806587514362313,
+      "grad_norm": 28.545714072290114,
+      "learning_rate": 5e-06,
+      "loss": 0.2298,
+      "num_input_tokens_seen": 216921392,
+      "step": 1255
+    },
+    {
+      "epoch": 0.4806587514362313,
+      "loss": 0.25406983494758606,
+      "loss_ce": 0.020671386271715164,
+      "loss_iou": 1.2555147409439087,
+      "loss_num": 0.2333984375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 216921392,
+      "step": 1255
+    },
+    {
+      "epoch": 0.4810417464572961,
+      "grad_norm": 25.958929048140725,
+      "learning_rate": 5e-06,
+      "loss": 0.2512,
+      "num_input_tokens_seen": 217094440,
+      "step": 1256
+    },
+    {
+      "epoch": 0.4810417464572961,
+      "loss": 0.29601573944091797,
+      "loss_ce": 0.020197857171297073,
+      "loss_iou": 1.0392547845840454,
+      "loss_num": 0.275390625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 217094440,
+      "step": 1256
+    },
+    {
+      "epoch": 0.4814247414783608,
+      "grad_norm": 29.56119118045122,
+      "learning_rate": 5e-06,
+      "loss": 0.3023,
+      "num_input_tokens_seen": 217266896,
+      "step": 1257
+    },
+    {
+      "epoch": 0.4814247414783608,
+      "loss": 0.37968093156814575,
+      "loss_ce": 0.019634529948234558,
+      "loss_iou": 1.1065092086791992,
+      "loss_num": 0.359375,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 217266896,
+      "step": 1257
+    },
+    {
+      "epoch": 0.4818077364994255,
+      "grad_norm": 28.549049276922037,
+      "learning_rate": 5e-06,
+      "loss": 0.2112,
+      "num_input_tokens_seen": 217439728,
+      "step": 1258
+    },
+    {
+      "epoch": 0.4818077364994255,
+      "loss": 0.20042143762111664,
+      "loss_ce": 0.020856007933616638,
+      "loss_iou": 1.0481743812561035,
+      "loss_num": 0.1796875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 217439728,
+      "step": 1258
+    },
+    {
+      "epoch": 0.48219073152049025,
+      "grad_norm": 39.25289950745477,
+      "learning_rate": 5e-06,
+      "loss": 0.2395,
+      "num_input_tokens_seen": 217612664,
+      "step": 1259
+    },
+    {
+      "epoch": 0.48219073152049025,
+      "loss": 0.26584044098854065,
+      "loss_ce": 0.020723257213830948,
+      "loss_iou": 0.9868621826171875,
+      "loss_num": 0.2451171875,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 217612664,
+      "step": 1259
+    },
+    {
+      "epoch": 0.48257372654155495,
+      "grad_norm": 43.435645856671705,
+      "learning_rate": 5e-06,
+      "loss": 0.2626,
+      "num_input_tokens_seen": 217785856,
+      "step": 1260
+    },
+    {
+      "epoch": 0.48257372654155495,
+      "loss": 0.21399447321891785,
+      "loss_ce": 0.02173374593257904,
+      "loss_iou": 1.0139672756195068,
+      "loss_num": 0.1923828125,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 217785856,
+      "step": 1260
+    },
+    {
+      "epoch": 0.4829567215626197,
+      "grad_norm": 37.75141670203048,
+      "learning_rate": 5e-06,
+      "loss": 0.2419,
+      "num_input_tokens_seen": 217958800,
+      "step": 1261
+    },
+    {
+      "epoch": 0.4829567215626197,
+      "loss": 0.21330128610134125,
+      "loss_ce": 0.02012501284480095,
+      "loss_iou": 1.0445476770401,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 217958800,
+      "step": 1261
+    },
+    {
+      "epoch": 0.4833397165836844,
+      "grad_norm": 26.454020869883134,
+      "learning_rate": 5e-06,
+      "loss": 0.2543,
+      "num_input_tokens_seen": 218131784,
+      "step": 1262
+    },
+    {
+      "epoch": 0.4833397165836844,
+      "loss": 0.2772575914859772,
+      "loss_ce": 0.02182544767856598,
+      "loss_iou": 1.0503876209259033,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 218131784,
+      "step": 1262
+    },
+    {
+      "epoch": 0.4837227116047491,
+      "grad_norm": 26.083432961672667,
+      "learning_rate": 5e-06,
+      "loss": 0.1786,
+      "num_input_tokens_seen": 218304288,
+      "step": 1263
+    },
+    {
+      "epoch": 0.4837227116047491,
+      "loss": 0.1882094442844391,
+      "loss_ce": 0.02054586447775364,
+      "loss_iou": 1.0622601509094238,
+      "loss_num": 0.16796875,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 218304288,
+      "step": 1263
+    },
+    {
+      "epoch": 0.48410570662581387,
+      "grad_norm": 30.387732450393234,
+      "learning_rate": 5e-06,
+      "loss": 0.2227,
+      "num_input_tokens_seen": 218477696,
+      "step": 1264
+    },
+    {
+      "epoch": 0.48410570662581387,
+      "loss": 0.23629167675971985,
+      "loss_ce": 0.021631035953760147,
+      "loss_iou": 1.0289371013641357,
+      "loss_num": 0.21484375,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 218477696,
+      "step": 1264
+    },
+    {
+      "epoch": 0.4844887016468786,
+      "grad_norm": 26.831065114890695,
+      "learning_rate": 5e-06,
+      "loss": 0.2217,
+      "num_input_tokens_seen": 218650680,
+      "step": 1265
+    },
+    {
+      "epoch": 0.4844887016468786,
+      "loss": 0.2116309255361557,
+      "loss_ce": 0.02126227878034115,
+      "loss_iou": 1.0325360298156738,
+      "loss_num": 0.1904296875,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 218650680,
+      "step": 1265
+    },
+    {
+      "epoch": 0.48487169666794333,
+      "grad_norm": 26.661143836911545,
+      "learning_rate": 5e-06,
+      "loss": 0.2495,
+      "num_input_tokens_seen": 218823744,
+      "step": 1266
+    },
+    {
+      "epoch": 0.48487169666794333,
+      "loss": 0.2849521040916443,
+      "loss_ce": 0.021524369716644287,
+      "loss_iou": 1.0682734251022339,
+      "loss_num": 0.263671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 218823744,
+      "step": 1266
+    },
+    {
+      "epoch": 0.48525469168900803,
+      "grad_norm": 33.55121540611247,
+      "learning_rate": 5e-06,
+      "loss": 0.2134,
+      "num_input_tokens_seen": 218996744,
+      "step": 1267
+    },
+    {
+      "epoch": 0.48525469168900803,
+      "loss": 0.21019728481769562,
+      "loss_ce": 0.02031690999865532,
+      "loss_iou": 1.0088659524917603,
+      "loss_num": 0.189453125,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 218996744,
+      "step": 1267
+    },
+    {
+      "epoch": 0.4856376867100728,
+      "grad_norm": 38.76012561855069,
+      "learning_rate": 5e-06,
+      "loss": 0.2625,
+      "num_input_tokens_seen": 219169520,
+      "step": 1268
+    },
+    {
+      "epoch": 0.4856376867100728,
+      "loss": 0.266610324382782,
+      "loss_ce": 0.021493151783943176,
+      "loss_iou": 1.0532801151275635,
+      "loss_num": 0.2451171875,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 219169520,
+      "step": 1268
+    },
+    {
+      "epoch": 0.4860206817311375,
+      "grad_norm": 25.69733505577472,
+      "learning_rate": 5e-06,
+      "loss": 0.2445,
+      "num_input_tokens_seen": 219342160,
+      "step": 1269
+    },
+    {
+      "epoch": 0.4860206817311375,
+      "loss": 0.27903541922569275,
+      "loss_ce": 0.01908668503165245,
+      "loss_iou": 0.9650876522064209,
+      "loss_num": 0.259765625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 219342160,
+      "step": 1269
+    },
+    {
+      "epoch": 0.4864036767522022,
+      "grad_norm": 28.647249426243246,
+      "learning_rate": 5e-06,
+      "loss": 0.2317,
+      "num_input_tokens_seen": 219514504,
+      "step": 1270
+    },
+    {
+      "epoch": 0.4864036767522022,
+      "loss": 0.23824116587638855,
+      "loss_ce": 0.0194911677390337,
+      "loss_iou": 1.0316154956817627,
+      "loss_num": 0.21875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 219514504,
+      "step": 1270
+    },
+    {
+      "epoch": 0.48678667177326695,
+      "grad_norm": 24.09040725347035,
+      "learning_rate": 5e-06,
+      "loss": 0.3011,
+      "num_input_tokens_seen": 219686880,
+      "step": 1271
+    },
+    {
+      "epoch": 0.48678667177326695,
+      "loss": 0.3540859520435333,
+      "loss_ce": 0.020101584494113922,
+      "loss_iou": 1.0527961254119873,
+      "loss_num": 0.333984375,
+      "loss_xval": 0.333984375,
+      "num_input_tokens_seen": 219686880,
+      "step": 1271
+    },
+    {
+      "epoch": 0.48716966679433166,
+      "grad_norm": 41.66327530287003,
+      "learning_rate": 5e-06,
+      "loss": 0.2688,
+      "num_input_tokens_seen": 219859928,
+      "step": 1272
+    },
+    {
+      "epoch": 0.48716966679433166,
+      "loss": 0.34175288677215576,
+      "loss_ce": 0.02064690738916397,
+      "loss_iou": 1.2458839416503906,
+      "loss_num": 0.3203125,
+      "loss_xval": 0.3203125,
+      "num_input_tokens_seen": 219859928,
+      "step": 1272
+    },
+    {
+      "epoch": 0.4875526618153964,
+      "grad_norm": 43.8863915282584,
+      "learning_rate": 5e-06,
+      "loss": 0.3065,
+      "num_input_tokens_seen": 220032792,
+      "step": 1273
+    },
+    {
+      "epoch": 0.4875526618153964,
+      "loss": 0.2963223457336426,
+      "loss_ce": 0.021297920495271683,
+      "loss_iou": 1.021531105041504,
+      "loss_num": 0.275390625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 220032792,
+      "step": 1273
+    },
+    {
+      "epoch": 0.4879356568364611,
+      "grad_norm": 31.15985261756482,
+      "learning_rate": 5e-06,
+      "loss": 0.2837,
+      "num_input_tokens_seen": 220205480,
+      "step": 1274
+    },
+    {
+      "epoch": 0.4879356568364611,
+      "loss": 0.2977232038974762,
+      "loss_ce": 0.02135601080954075,
+      "loss_iou": 1.0700514316558838,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 220205480,
+      "step": 1274
+    },
+    {
+      "epoch": 0.4883186518575259,
+      "grad_norm": 27.083749619057436,
+      "learning_rate": 5e-06,
+      "loss": 0.289,
+      "num_input_tokens_seen": 220378616,
+      "step": 1275
+    },
+    {
+      "epoch": 0.4883186518575259,
+      "loss": 0.3561253845691681,
+      "loss_ce": 0.020248912274837494,
+      "loss_iou": 1.0839688777923584,
+      "loss_num": 0.3359375,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 220378616,
+      "step": 1275
+    },
+    {
+      "epoch": 0.4887016468785906,
+      "grad_norm": 37.52991517794262,
+      "learning_rate": 5e-06,
+      "loss": 0.2556,
+      "num_input_tokens_seen": 220551680,
+      "step": 1276
+    },
+    {
+      "epoch": 0.4887016468785906,
+      "loss": 0.23188307881355286,
+      "loss_ce": 0.021922152489423752,
+      "loss_iou": 1.0254385471343994,
+      "loss_num": 0.2099609375,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 220551680,
+      "step": 1276
+    },
+    {
+      "epoch": 0.4890846418996553,
+      "grad_norm": 31.922314748751266,
+      "learning_rate": 5e-06,
+      "loss": 0.2336,
+      "num_input_tokens_seen": 220724960,
+      "step": 1277
+    },
+    {
+      "epoch": 0.4890846418996553,
+      "loss": 0.19547098875045776,
+      "loss_ce": 0.020055953413248062,
+      "loss_iou": 1.0054305791854858,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 220724960,
+      "step": 1277
+    },
+    {
+      "epoch": 0.48946763692072004,
+      "grad_norm": 20.772756783892586,
+      "learning_rate": 5e-06,
+      "loss": 0.3165,
+      "num_input_tokens_seen": 220897696,
+      "step": 1278
+    },
+    {
+      "epoch": 0.48946763692072004,
+      "loss": 0.2660837769508362,
+      "loss_ce": 0.021210746839642525,
+      "loss_iou": 1.0261434316635132,
+      "loss_num": 0.2451171875,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 220897696,
+      "step": 1278
+    },
+    {
+      "epoch": 0.48985063194178474,
+      "grad_norm": 21.80453036032498,
+      "learning_rate": 5e-06,
+      "loss": 0.1961,
+      "num_input_tokens_seen": 221070320,
+      "step": 1279
+    },
+    {
+      "epoch": 0.48985063194178474,
+      "loss": 0.20578120648860931,
+      "loss_ce": 0.019807081669569016,
+      "loss_iou": 1.0507761240005493,
+      "loss_num": 0.185546875,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 221070320,
+      "step": 1279
+    },
+    {
+      "epoch": 0.4902336269628495,
+      "grad_norm": 34.31817030457487,
+      "learning_rate": 5e-06,
+      "loss": 0.2552,
+      "num_input_tokens_seen": 221243296,
+      "step": 1280
+    },
+    {
+      "epoch": 0.4902336269628495,
+      "loss": 0.3230183720588684,
+      "loss_ce": 0.01949053257703781,
+      "loss_iou": 1.6514267921447754,
+      "loss_num": 0.302734375,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 221243296,
+      "step": 1280
+    },
+    {
+      "epoch": 0.4906166219839142,
+      "grad_norm": 29.44297168945806,
+      "learning_rate": 5e-06,
+      "loss": 0.2868,
+      "num_input_tokens_seen": 221416328,
+      "step": 1281
+    },
+    {
+      "epoch": 0.4906166219839142,
+      "loss": 0.31145864725112915,
+      "loss_ce": 0.02044302597641945,
+      "loss_iou": 1.084681749343872,
+      "loss_num": 0.291015625,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 221416328,
+      "step": 1281
+    },
+    {
+      "epoch": 0.49099961700497896,
+      "grad_norm": 24.28100217361298,
+      "learning_rate": 5e-06,
+      "loss": 0.241,
+      "num_input_tokens_seen": 221589264,
+      "step": 1282
+    },
+    {
+      "epoch": 0.49099961700497896,
+      "loss": 0.27191275358200073,
+      "loss_ce": 0.02069203555583954,
+      "loss_iou": 1.0331673622131348,
+      "loss_num": 0.251953125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 221589264,
+      "step": 1282
+    },
+    {
+      "epoch": 0.49138261202604366,
+      "grad_norm": 30.994349085547384,
+      "learning_rate": 5e-06,
+      "loss": 0.2326,
+      "num_input_tokens_seen": 221762320,
+      "step": 1283
+    },
+    {
+      "epoch": 0.49138261202604366,
+      "loss": 0.2575764060020447,
+      "loss_ce": 0.01978345587849617,
+      "loss_iou": 1.0269644260406494,
+      "loss_num": 0.23828125,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 221762320,
+      "step": 1283
+    },
+    {
+      "epoch": 0.49176560704710837,
+      "grad_norm": 38.6801635000788,
+      "learning_rate": 5e-06,
+      "loss": 0.201,
+      "num_input_tokens_seen": 221935392,
+      "step": 1284
+    },
+    {
+      "epoch": 0.49176560704710837,
+      "loss": 0.18782469630241394,
+      "loss_ce": 0.021137692034244537,
+      "loss_iou": 1.0069403648376465,
+      "loss_num": 0.1669921875,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 221935392,
+      "step": 1284
+    },
+    {
+      "epoch": 0.4921486020681731,
+      "grad_norm": 27.639627105321036,
+      "learning_rate": 5e-06,
+      "loss": 0.2377,
+      "num_input_tokens_seen": 222108128,
+      "step": 1285
+    },
+    {
+      "epoch": 0.4921486020681731,
+      "loss": 0.2548784017562866,
+      "loss_ce": 0.019404759630560875,
+      "loss_iou": 1.0391649007797241,
+      "loss_num": 0.2353515625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 222108128,
+      "step": 1285
+    },
+    {
+      "epoch": 0.4925315970892378,
+      "grad_norm": 22.161921926812532,
+      "learning_rate": 5e-06,
+      "loss": 0.1802,
+      "num_input_tokens_seen": 222281120,
+      "step": 1286
+    },
+    {
+      "epoch": 0.4925315970892378,
+      "loss": 0.17910084128379822,
+      "loss_ce": 0.01979907602071762,
+      "loss_iou": 1.028035044670105,
+      "loss_num": 0.1591796875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 222281120,
+      "step": 1286
+    },
+    {
+      "epoch": 0.4929145921103026,
+      "grad_norm": 26.32633461607658,
+      "learning_rate": 5e-06,
+      "loss": 0.2956,
+      "num_input_tokens_seen": 222453960,
+      "step": 1287
+    },
+    {
+      "epoch": 0.4929145921103026,
+      "loss": 0.3282597064971924,
+      "loss_ce": 0.021008752286434174,
+      "loss_iou": 1.0230724811553955,
+      "loss_num": 0.306640625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 222453960,
+      "step": 1287
+    },
+    {
+      "epoch": 0.4932975871313673,
+      "grad_norm": 50.046689203103455,
+      "learning_rate": 5e-06,
+      "loss": 0.2674,
+      "num_input_tokens_seen": 222626888,
+      "step": 1288
+    },
+    {
+      "epoch": 0.4932975871313673,
+      "loss": 0.22985592484474182,
+      "loss_ce": 0.02111569046974182,
+      "loss_iou": 1.0050923824310303,
+      "loss_num": 0.208984375,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 222626888,
+      "step": 1288
+    },
+    {
+      "epoch": 0.49368058215243205,
+      "grad_norm": 34.9913963459762,
+      "learning_rate": 5e-06,
+      "loss": 0.2121,
+      "num_input_tokens_seen": 222799736,
+      "step": 1289
+    },
+    {
+      "epoch": 0.49368058215243205,
+      "loss": 0.2531659007072449,
+      "loss_ce": 0.020255738869309425,
+      "loss_iou": 1.0212188959121704,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 222799736,
+      "step": 1289
+    },
+    {
+      "epoch": 0.49406357717349675,
+      "grad_norm": 32.45673806117689,
+      "learning_rate": 5e-06,
+      "loss": 0.2929,
+      "num_input_tokens_seen": 222972904,
+      "step": 1290
+    },
+    {
+      "epoch": 0.49406357717349675,
+      "loss": 0.2759690284729004,
+      "loss_ce": 0.02120828628540039,
+      "loss_iou": 1.0240442752838135,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 222972904,
+      "step": 1290
+    },
+    {
+      "epoch": 0.49444657219456145,
+      "grad_norm": 47.547632682028144,
+      "learning_rate": 5e-06,
+      "loss": 0.3203,
+      "num_input_tokens_seen": 223145920,
+      "step": 1291
+    },
+    {
+      "epoch": 0.49444657219456145,
+      "loss": 0.2650194466114044,
+      "loss_ce": 0.020512614399194717,
+      "loss_iou": 1.0113552808761597,
+      "loss_num": 0.244140625,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 223145920,
+      "step": 1291
+    },
+    {
+      "epoch": 0.4948295672156262,
+      "grad_norm": 44.28677941576457,
+      "learning_rate": 5e-06,
+      "loss": 0.2589,
+      "num_input_tokens_seen": 223318880,
+      "step": 1292
+    },
+    {
+      "epoch": 0.4948295672156262,
+      "loss": 0.2196100503206253,
+      "loss_ce": 0.021734073758125305,
+      "loss_iou": 0.9984222650527954,
+      "loss_num": 0.1982421875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 223318880,
+      "step": 1292
+    },
+    {
+      "epoch": 0.4952125622366909,
+      "grad_norm": 36.16609817287782,
+      "learning_rate": 5e-06,
+      "loss": 0.3403,
+      "num_input_tokens_seen": 223491800,
+      "step": 1293
+    },
+    {
+      "epoch": 0.4952125622366909,
+      "loss": 0.3044971525669098,
+      "loss_ce": 0.022392667829990387,
+      "loss_iou": 1.1346598863601685,
+      "loss_num": 0.28125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 223491800,
+      "step": 1293
+    },
+    {
+      "epoch": 0.49559555725775567,
+      "grad_norm": 21.054867054897002,
+      "learning_rate": 5e-06,
+      "loss": 0.2747,
+      "num_input_tokens_seen": 223664648,
+      "step": 1294
+    },
+    {
+      "epoch": 0.49559555725775567,
+      "loss": 0.29920294880867004,
+      "loss_ce": 0.019051577895879745,
+      "loss_iou": 1.1052517890930176,
+      "loss_num": 0.279296875,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 223664648,
+      "step": 1294
+    },
+    {
+      "epoch": 0.4959785522788204,
+      "grad_norm": 16.948986363038664,
+      "learning_rate": 5e-06,
+      "loss": 0.3082,
+      "num_input_tokens_seen": 223837328,
+      "step": 1295
+    },
+    {
+      "epoch": 0.4959785522788204,
+      "loss": 0.3230059742927551,
+      "loss_ce": 0.02002747543156147,
+      "loss_iou": 1.1161134243011475,
+      "loss_num": 0.302734375,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 223837328,
+      "step": 1295
+    },
+    {
+      "epoch": 0.4963615472998851,
+      "grad_norm": 16.96679685389391,
+      "learning_rate": 5e-06,
+      "loss": 0.2246,
+      "num_input_tokens_seen": 224010416,
+      "step": 1296
+    },
+    {
+      "epoch": 0.4963615472998851,
+      "loss": 0.23388779163360596,
+      "loss_ce": 0.01770125702023506,
+      "loss_iou": 0.8875874876976013,
+      "loss_num": 0.2158203125,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 224010416,
+      "step": 1296
+    },
+    {
+      "epoch": 0.49674454232094983,
+      "grad_norm": 31.47918902702664,
+      "learning_rate": 5e-06,
+      "loss": 0.3196,
+      "num_input_tokens_seen": 224183672,
+      "step": 1297
+    },
+    {
+      "epoch": 0.49674454232094983,
+      "loss": 0.34980836510658264,
+      "loss_ce": 0.021195102483034134,
+      "loss_iou": 1.0460903644561768,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 224183672,
+      "step": 1297
+    },
+    {
+      "epoch": 0.49712753734201454,
+      "grad_norm": 41.42384157620713,
+      "learning_rate": 5e-06,
+      "loss": 0.2303,
+      "num_input_tokens_seen": 224357024,
+      "step": 1298
+    },
+    {
+      "epoch": 0.49712753734201454,
+      "loss": 0.21717743575572968,
+      "loss_ce": 0.022414255887269974,
+      "loss_iou": 1.001346230506897,
+      "loss_num": 0.1943359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 224357024,
+      "step": 1298
+    },
+    {
+      "epoch": 0.4975105323630793,
+      "grad_norm": 36.173696628924326,
+      "learning_rate": 5e-06,
+      "loss": 0.3034,
+      "num_input_tokens_seen": 224529872,
+      "step": 1299
+    },
+    {
+      "epoch": 0.4975105323630793,
+      "loss": 0.23724958300590515,
+      "loss_ce": 0.02020856738090515,
+      "loss_iou": 1.0051367282867432,
+      "loss_num": 0.216796875,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 224529872,
+      "step": 1299
+    },
+    {
+      "epoch": 0.497893527384144,
+      "grad_norm": 22.15717286074569,
+      "learning_rate": 5e-06,
+      "loss": 0.2266,
+      "num_input_tokens_seen": 224702944,
+      "step": 1300
+    },
+    {
+      "epoch": 0.497893527384144,
+      "loss": 0.22423772513866425,
+      "loss_ce": 0.02129582315683365,
+      "loss_iou": 1.030242919921875,
+      "loss_num": 0.203125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 224702944,
+      "step": 1300
+    },
+    {
+      "epoch": 0.49827652240520875,
+      "grad_norm": 35.5021631522705,
+      "learning_rate": 5e-06,
+      "loss": 0.2559,
+      "num_input_tokens_seen": 224875800,
+      "step": 1301
+    },
+    {
+      "epoch": 0.49827652240520875,
+      "loss": 0.29484647512435913,
+      "loss_ce": 0.021775171160697937,
+      "loss_iou": 1.0023797750473022,
+      "loss_num": 0.2734375,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 224875800,
+      "step": 1301
+    },
+    {
+      "epoch": 0.49865951742627346,
+      "grad_norm": 46.35512650727594,
+      "learning_rate": 5e-06,
+      "loss": 0.2939,
+      "num_input_tokens_seen": 225048824,
+      "step": 1302
+    },
+    {
+      "epoch": 0.49865951742627346,
+      "loss": 0.32741639018058777,
+      "loss_ce": 0.02126404270529747,
+      "loss_iou": 1.0482133626937866,
+      "loss_num": 0.306640625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 225048824,
+      "step": 1302
+    },
+    {
+      "epoch": 0.49904251244733816,
+      "grad_norm": 34.31661781023796,
+      "learning_rate": 5e-06,
+      "loss": 0.1727,
+      "num_input_tokens_seen": 225221712,
+      "step": 1303
+    },
+    {
+      "epoch": 0.49904251244733816,
+      "loss": 0.16094151139259338,
+      "loss_ce": 0.018607541918754578,
+      "loss_iou": 1.000558614730835,
+      "loss_num": 0.142578125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 225221712,
+      "step": 1303
+    },
+    {
+      "epoch": 0.4994255074684029,
+      "grad_norm": 29.129981132600637,
+      "learning_rate": 5e-06,
+      "loss": 0.2064,
+      "num_input_tokens_seen": 225394392,
+      "step": 1304
+    },
+    {
+      "epoch": 0.4994255074684029,
+      "loss": 0.19164007902145386,
+      "loss_ce": 0.019520942121744156,
+      "loss_iou": 1.0027861595153809,
+      "loss_num": 0.171875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 225394392,
+      "step": 1304
+    },
+    {
+      "epoch": 0.4998085024894676,
+      "grad_norm": 39.55374574961186,
+      "learning_rate": 5e-06,
+      "loss": 0.248,
+      "num_input_tokens_seen": 225567568,
+      "step": 1305
+    },
+    {
+      "epoch": 0.4998085024894676,
+      "loss": 0.2537155747413635,
+      "loss_ce": 0.021904051303863525,
+      "loss_iou": 1.28212308883667,
+      "loss_num": 0.2314453125,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 225567568,
+      "step": 1305
+    },
+    {
+      "epoch": 0.5001914975105324,
+      "grad_norm": 40.71644853830682,
+      "learning_rate": 5e-06,
+      "loss": 0.2348,
+      "num_input_tokens_seen": 225740720,
+      "step": 1306
+    },
+    {
+      "epoch": 0.5001914975105324,
+      "loss": 0.19909319281578064,
+      "loss_ce": 0.02093156985938549,
+      "loss_iou": 0.9498213529586792,
+      "loss_num": 0.177734375,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 225740720,
+      "step": 1306
+    },
+    {
+      "epoch": 0.5005744925315971,
+      "grad_norm": 29.06221694350055,
+      "learning_rate": 5e-06,
+      "loss": 0.1784,
+      "num_input_tokens_seen": 225913584,
+      "step": 1307
+    },
+    {
+      "epoch": 0.5005744925315971,
+      "loss": 0.1767686903476715,
+      "loss_ce": 0.018321428447961807,
+      "loss_iou": 1.0008810758590698,
+      "loss_num": 0.158203125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 225913584,
+      "step": 1307
+    },
+    {
+      "epoch": 0.5009574875526618,
+      "grad_norm": 50.6076976981039,
+      "learning_rate": 5e-06,
+      "loss": 0.1954,
+      "num_input_tokens_seen": 226086816,
+      "step": 1308
+    },
+    {
+      "epoch": 0.5009574875526618,
+      "loss": 0.19694334268569946,
+      "loss_ce": 0.020795879885554314,
+      "loss_iou": 1.0267889499664307,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 226086816,
+      "step": 1308
+    },
+    {
+      "epoch": 0.5013404825737265,
+      "grad_norm": 35.17112276616945,
+      "learning_rate": 5e-06,
+      "loss": 0.2929,
+      "num_input_tokens_seen": 226259584,
+      "step": 1309
+    },
+    {
+      "epoch": 0.5013404825737265,
+      "loss": 0.2977929413318634,
+      "loss_ce": 0.0208764486014843,
+      "loss_iou": 1.084275484085083,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 226259584,
+      "step": 1309
+    },
+    {
+      "epoch": 0.5017234775947913,
+      "grad_norm": 28.65337481435919,
+      "learning_rate": 5e-06,
+      "loss": 0.2208,
+      "num_input_tokens_seen": 226432240,
+      "step": 1310
+    },
+    {
+      "epoch": 0.5017234775947913,
+      "loss": 0.2865043878555298,
+      "loss_ce": 0.019536608830094337,
+      "loss_iou": 1.0236896276474,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 226432240,
+      "step": 1310
+    },
+    {
+      "epoch": 0.502106472615856,
+      "grad_norm": 31.646856094478498,
+      "learning_rate": 5e-06,
+      "loss": 0.2176,
+      "num_input_tokens_seen": 226601512,
+      "step": 1311
+    },
+    {
+      "epoch": 0.502106472615856,
+      "loss": 0.24922175705432892,
+      "loss_ce": 0.01890559494495392,
+      "loss_iou": 0.88271164894104,
+      "loss_num": 0.23046875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 226601512,
+      "step": 1311
+    },
+    {
+      "epoch": 0.5024894676369207,
+      "grad_norm": 36.226073173748745,
+      "learning_rate": 5e-06,
+      "loss": 0.2344,
+      "num_input_tokens_seen": 226774608,
+      "step": 1312
+    },
+    {
+      "epoch": 0.5024894676369207,
+      "loss": 0.2274923473596573,
+      "loss_ce": 0.02155972458422184,
+      "loss_iou": 1.152855634689331,
+      "loss_num": 0.2060546875,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 226774608,
+      "step": 1312
+    },
+    {
+      "epoch": 0.5028724626579855,
+      "grad_norm": 38.4014820846328,
+      "learning_rate": 5e-06,
+      "loss": 0.2594,
+      "num_input_tokens_seen": 226947576,
+      "step": 1313
+    },
+    {
+      "epoch": 0.5028724626579855,
+      "loss": 0.32784831523895264,
+      "loss_ce": 0.01901041530072689,
+      "loss_iou": 1.092888593673706,
+      "loss_num": 0.30859375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 226947576,
+      "step": 1313
+    },
+    {
+      "epoch": 0.5032554576790502,
+      "grad_norm": 38.89804778530378,
+      "learning_rate": 5e-06,
+      "loss": 0.2806,
+      "num_input_tokens_seen": 227120832,
+      "step": 1314
+    },
+    {
+      "epoch": 0.5032554576790502,
+      "loss": 0.28596794605255127,
+      "loss_ce": 0.02144157886505127,
+      "loss_iou": 1.021527886390686,
+      "loss_num": 0.263671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 227120832,
+      "step": 1314
+    },
+    {
+      "epoch": 0.5036384527001149,
+      "grad_norm": 31.807060485433798,
+      "learning_rate": 5e-06,
+      "loss": 0.2344,
+      "num_input_tokens_seen": 227293688,
+      "step": 1315
+    },
+    {
+      "epoch": 0.5036384527001149,
+      "loss": 0.21565531194210052,
+      "loss_ce": 0.020037636160850525,
+      "loss_iou": 1.0489392280578613,
+      "loss_num": 0.1953125,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 227293688,
+      "step": 1315
+    },
+    {
+      "epoch": 0.5040214477211796,
+      "grad_norm": 23.146613469858394,
+      "learning_rate": 5e-06,
+      "loss": 0.2209,
+      "num_input_tokens_seen": 227466600,
+      "step": 1316
+    },
+    {
+      "epoch": 0.5040214477211796,
+      "loss": 0.2014552652835846,
+      "loss_ce": 0.02018084190785885,
+      "loss_iou": 1.0091123580932617,
+      "loss_num": 0.181640625,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 227466600,
+      "step": 1316
+    },
+    {
+      "epoch": 0.5044044427422444,
+      "grad_norm": 22.32627210915022,
+      "learning_rate": 5e-06,
+      "loss": 0.21,
+      "num_input_tokens_seen": 227639752,
+      "step": 1317
+    },
+    {
+      "epoch": 0.5044044427422444,
+      "loss": 0.21035532653331757,
+      "loss_ce": 0.019437363371253014,
+      "loss_iou": 1.0013794898986816,
+      "loss_num": 0.19140625,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 227639752,
+      "step": 1317
+    },
+    {
+      "epoch": 0.504787437763309,
+      "grad_norm": 24.765486554199338,
+      "learning_rate": 5e-06,
+      "loss": 0.2454,
+      "num_input_tokens_seen": 227812776,
+      "step": 1318
+    },
+    {
+      "epoch": 0.504787437763309,
+      "loss": 0.2919178009033203,
+      "loss_ce": 0.021043788641691208,
+      "loss_iou": 1.0157856941223145,
+      "loss_num": 0.271484375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 227812776,
+      "step": 1318
+    },
+    {
+      "epoch": 0.5051704327843738,
+      "grad_norm": 36.61425807120079,
+      "learning_rate": 5e-06,
+      "loss": 0.1963,
+      "num_input_tokens_seen": 227985864,
+      "step": 1319
+    },
+    {
+      "epoch": 0.5051704327843738,
+      "loss": 0.18542779982089996,
+      "loss_ce": 0.018618717789649963,
+      "loss_iou": 1.0022015571594238,
+      "loss_num": 0.1669921875,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 227985864,
+      "step": 1319
+    },
+    {
+      "epoch": 0.5055534278054385,
+      "grad_norm": 33.75917521038577,
+      "learning_rate": 5e-06,
+      "loss": 0.2716,
+      "num_input_tokens_seen": 228158688,
+      "step": 1320
+    },
+    {
+      "epoch": 0.5055534278054385,
+      "loss": 0.23160414397716522,
+      "loss_ce": 0.020117323845624924,
+      "loss_iou": 1.0108810663223267,
+      "loss_num": 0.2119140625,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 228158688,
+      "step": 1320
+    },
+    {
+      "epoch": 0.5059364228265033,
+      "grad_norm": 34.89146044779558,
+      "learning_rate": 5e-06,
+      "loss": 0.2391,
+      "num_input_tokens_seen": 228331520,
+      "step": 1321
+    },
+    {
+      "epoch": 0.5059364228265033,
+      "loss": 0.23049719631671906,
+      "loss_ce": 0.020047983154654503,
+      "loss_iou": 1.028801441192627,
+      "loss_num": 0.2109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 228331520,
+      "step": 1321
+    },
+    {
+      "epoch": 0.506319417847568,
+      "grad_norm": 29.487383714282466,
+      "learning_rate": 5e-06,
+      "loss": 0.2963,
+      "num_input_tokens_seen": 228504720,
+      "step": 1322
+    },
+    {
+      "epoch": 0.506319417847568,
+      "loss": 0.24549441039562225,
+      "loss_ce": 0.02247193455696106,
+      "loss_iou": 1.0049664974212646,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 228504720,
+      "step": 1322
+    },
+    {
+      "epoch": 0.5067024128686327,
+      "grad_norm": 24.43148361606767,
+      "learning_rate": 5e-06,
+      "loss": 0.2366,
+      "num_input_tokens_seen": 228677624,
+      "step": 1323
+    },
+    {
+      "epoch": 0.5067024128686327,
+      "loss": 0.17037566006183624,
+      "loss_ce": 0.01998504251241684,
+      "loss_iou": 1.0022523403167725,
+      "loss_num": 0.150390625,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 228677624,
+      "step": 1323
+    },
+    {
+      "epoch": 0.5070854078896975,
+      "grad_norm": 25.358514927064693,
+      "learning_rate": 5e-06,
+      "loss": 0.2574,
+      "num_input_tokens_seen": 228850800,
+      "step": 1324
+    },
+    {
+      "epoch": 0.5070854078896975,
+      "loss": 0.2549811005592346,
+      "loss_ce": 0.021094385534524918,
+      "loss_iou": 1.0110515356063843,
+      "loss_num": 0.234375,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 228850800,
+      "step": 1324
+    },
+    {
+      "epoch": 0.5074684029107621,
+      "grad_norm": 26.514133382129888,
+      "learning_rate": 5e-06,
+      "loss": 0.2547,
+      "num_input_tokens_seen": 229023664,
+      "step": 1325
+    },
+    {
+      "epoch": 0.5074684029107621,
+      "loss": 0.2352285385131836,
+      "loss_ce": 0.021727576851844788,
+      "loss_iou": 1.0286118984222412,
+      "loss_num": 0.2138671875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 229023664,
+      "step": 1325
+    },
+    {
+      "epoch": 0.5078513979318269,
+      "grad_norm": 38.798782986456516,
+      "learning_rate": 5e-06,
+      "loss": 0.264,
+      "num_input_tokens_seen": 229196784,
+      "step": 1326
+    },
+    {
+      "epoch": 0.5078513979318269,
+      "loss": 0.3162633180618286,
+      "loss_ce": 0.022195931524038315,
+      "loss_iou": 1.3013497591018677,
+      "loss_num": 0.294921875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 229196784,
+      "step": 1326
+    },
+    {
+      "epoch": 0.5082343929528916,
+      "grad_norm": 34.50600779885732,
+      "learning_rate": 5e-06,
+      "loss": 0.2081,
+      "num_input_tokens_seen": 229369408,
+      "step": 1327
+    },
+    {
+      "epoch": 0.5082343929528916,
+      "loss": 0.1770828366279602,
+      "loss_ce": 0.018391430377960205,
+      "loss_iou": 0.8795397877693176,
+      "loss_num": 0.158203125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 229369408,
+      "step": 1327
+    },
+    {
+      "epoch": 0.5086173879739564,
+      "grad_norm": 28.688164962242777,
+      "learning_rate": 5e-06,
+      "loss": 0.2512,
+      "num_input_tokens_seen": 229542488,
+      "step": 1328
+    },
+    {
+      "epoch": 0.5086173879739564,
+      "loss": 0.2442503720521927,
+      "loss_ce": 0.020800668746232986,
+      "loss_iou": 1.0248935222625732,
+      "loss_num": 0.2236328125,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 229542488,
+      "step": 1328
+    },
+    {
+      "epoch": 0.509000382995021,
+      "grad_norm": 33.64137325261358,
+      "learning_rate": 5e-06,
+      "loss": 0.2543,
+      "num_input_tokens_seen": 229715656,
+      "step": 1329
+    },
+    {
+      "epoch": 0.509000382995021,
+      "loss": 0.2702787518501282,
+      "loss_ce": 0.020400822162628174,
+      "loss_iou": 1.170792818069458,
+      "loss_num": 0.25,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 229715656,
+      "step": 1329
+    },
+    {
+      "epoch": 0.5093833780160858,
+      "grad_norm": 36.39428144567596,
+      "learning_rate": 5e-06,
+      "loss": 0.2902,
+      "num_input_tokens_seen": 229888712,
+      "step": 1330
+    },
+    {
+      "epoch": 0.5093833780160858,
+      "loss": 0.2342301458120346,
+      "loss_ce": 0.021461594849824905,
+      "loss_iou": 1.0727958679199219,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 229888712,
+      "step": 1330
+    },
+    {
+      "epoch": 0.5097663730371506,
+      "grad_norm": 25.070009992914553,
+      "learning_rate": 5e-06,
+      "loss": 0.2314,
+      "num_input_tokens_seen": 230061896,
+      "step": 1331
+    },
+    {
+      "epoch": 0.5097663730371506,
+      "loss": 0.2624282240867615,
+      "loss_ce": 0.021034184843301773,
+      "loss_iou": 1.0383816957473755,
+      "loss_num": 0.2412109375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 230061896,
+      "step": 1331
+    },
+    {
+      "epoch": 0.5101493680582152,
+      "grad_norm": 35.76388904354218,
+      "learning_rate": 5e-06,
+      "loss": 0.2072,
+      "num_input_tokens_seen": 230234944,
+      "step": 1332
+    },
+    {
+      "epoch": 0.5101493680582152,
+      "loss": 0.2063194364309311,
+      "loss_ce": 0.019979100674390793,
+      "loss_iou": 1.0199110507965088,
+      "loss_num": 0.1865234375,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 230234944,
+      "step": 1332
+    },
+    {
+      "epoch": 0.51053236307928,
+      "grad_norm": 20.77696301352915,
+      "learning_rate": 5e-06,
+      "loss": 0.2245,
+      "num_input_tokens_seen": 230407960,
+      "step": 1333
+    },
+    {
+      "epoch": 0.51053236307928,
+      "loss": 0.23807215690612793,
+      "loss_ce": 0.021946672350168228,
+      "loss_iou": 1.0557270050048828,
+      "loss_num": 0.2158203125,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 230407960,
+      "step": 1333
+    },
+    {
+      "epoch": 0.5109153581003447,
+      "grad_norm": 43.40236449501893,
+      "learning_rate": 5e-06,
+      "loss": 0.2583,
+      "num_input_tokens_seen": 230581128,
+      "step": 1334
+    },
+    {
+      "epoch": 0.5109153581003447,
+      "loss": 0.26795485615730286,
+      "loss_ce": 0.021372821182012558,
+      "loss_iou": 1.0251628160476685,
+      "loss_num": 0.24609375,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 230581128,
+      "step": 1334
+    },
+    {
+      "epoch": 0.5112983531214095,
+      "grad_norm": 38.36400782556495,
+      "learning_rate": 5e-06,
+      "loss": 0.2806,
+      "num_input_tokens_seen": 230753384,
+      "step": 1335
+    },
+    {
+      "epoch": 0.5112983531214095,
+      "loss": 0.2859666645526886,
+      "loss_ce": 0.017717158421874046,
+      "loss_iou": 0.9628697633743286,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 230753384,
+      "step": 1335
+    },
+    {
+      "epoch": 0.5116813481424741,
+      "grad_norm": 22.614666083386055,
+      "learning_rate": 5e-06,
+      "loss": 0.2495,
+      "num_input_tokens_seen": 230926144,
+      "step": 1336
+    },
+    {
+      "epoch": 0.5116813481424741,
+      "loss": 0.2618767023086548,
+      "loss_ce": 0.02017749845981598,
+      "loss_iou": 1.1013388633728027,
+      "loss_num": 0.2421875,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 230926144,
+      "step": 1336
+    },
+    {
+      "epoch": 0.5120643431635389,
+      "grad_norm": 31.241732332051885,
+      "learning_rate": 5e-06,
+      "loss": 0.2089,
+      "num_input_tokens_seen": 231098912,
+      "step": 1337
+    },
+    {
+      "epoch": 0.5120643431635389,
+      "loss": 0.22717610001564026,
+      "loss_ce": 0.020266912877559662,
+      "loss_iou": 1.0211718082427979,
+      "loss_num": 0.20703125,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 231098912,
+      "step": 1337
+    },
+    {
+      "epoch": 0.5124473381846036,
+      "grad_norm": 40.09849691785989,
+      "learning_rate": 5e-06,
+      "loss": 0.2688,
+      "num_input_tokens_seen": 231271840,
+      "step": 1338
+    },
+    {
+      "epoch": 0.5124473381846036,
+      "loss": 0.26360654830932617,
+      "loss_ce": 0.01983211562037468,
+      "loss_iou": 1.0164717435836792,
+      "loss_num": 0.244140625,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 231271840,
+      "step": 1338
+    },
+    {
+      "epoch": 0.5128303332056683,
+      "grad_norm": 32.845511557733886,
+      "learning_rate": 5e-06,
+      "loss": 0.273,
+      "num_input_tokens_seen": 231444504,
+      "step": 1339
+    },
+    {
+      "epoch": 0.5128303332056683,
+      "loss": 0.27383947372436523,
+      "loss_ce": 0.01956700161099434,
+      "loss_iou": 1.0331006050109863,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 231444504,
+      "step": 1339
+    },
+    {
+      "epoch": 0.513213328226733,
+      "grad_norm": 36.65452646587695,
+      "learning_rate": 5e-06,
+      "loss": 0.2347,
+      "num_input_tokens_seen": 231617464,
+      "step": 1340
+    },
+    {
+      "epoch": 0.513213328226733,
+      "loss": 0.21411117911338806,
+      "loss_ce": 0.022338712587952614,
+      "loss_iou": 1.02278733253479,
+      "loss_num": 0.19140625,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 231617464,
+      "step": 1340
+    },
+    {
+      "epoch": 0.5135963232477978,
+      "grad_norm": 42.29017363996495,
+      "learning_rate": 5e-06,
+      "loss": 0.2139,
+      "num_input_tokens_seen": 231790368,
+      "step": 1341
+    },
+    {
+      "epoch": 0.5135963232477978,
+      "loss": 0.22402310371398926,
+      "loss_ce": 0.021752599626779556,
+      "loss_iou": 1.0306389331817627,
+      "loss_num": 0.2021484375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 231790368,
+      "step": 1341
+    },
+    {
+      "epoch": 0.5139793182688625,
+      "grad_norm": 47.29629926879741,
+      "learning_rate": 5e-06,
+      "loss": 0.2668,
+      "num_input_tokens_seen": 231962936,
+      "step": 1342
+    },
+    {
+      "epoch": 0.5139793182688625,
+      "loss": 0.28897830843925476,
+      "loss_ce": 0.02097294107079506,
+      "loss_iou": 1.057763695716858,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 231962936,
+      "step": 1342
+    },
+    {
+      "epoch": 0.5143623132899272,
+      "grad_norm": 24.648160299512856,
+      "learning_rate": 5e-06,
+      "loss": 0.2407,
+      "num_input_tokens_seen": 232135720,
+      "step": 1343
+    },
+    {
+      "epoch": 0.5143623132899272,
+      "loss": 0.25257134437561035,
+      "loss_ce": 0.01844049245119095,
+      "loss_iou": 0.9149987101554871,
+      "loss_num": 0.234375,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 232135720,
+      "step": 1343
+    },
+    {
+      "epoch": 0.514745308310992,
+      "grad_norm": 23.948393528127145,
+      "learning_rate": 5e-06,
+      "loss": 0.1945,
+      "num_input_tokens_seen": 232308168,
+      "step": 1344
+    },
+    {
+      "epoch": 0.514745308310992,
+      "loss": 0.15678608417510986,
+      "loss_ce": 0.018785586580634117,
+      "loss_iou": 0.8803042769432068,
+      "loss_num": 0.1376953125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 232308168,
+      "step": 1344
+    },
+    {
+      "epoch": 0.5151283033320567,
+      "grad_norm": 16.1456748348062,
+      "learning_rate": 5e-06,
+      "loss": 0.2041,
+      "num_input_tokens_seen": 232480952,
+      "step": 1345
+    },
+    {
+      "epoch": 0.5151283033320567,
+      "loss": 0.17203077673912048,
+      "loss_ce": 0.020480498671531677,
+      "loss_iou": 1.002892255783081,
+      "loss_num": 0.1513671875,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 232480952,
+      "step": 1345
+    },
+    {
+      "epoch": 0.5155112983531214,
+      "grad_norm": 10.864190384455942,
+      "learning_rate": 5e-06,
+      "loss": 0.1912,
+      "num_input_tokens_seen": 232654008,
+      "step": 1346
+    },
+    {
+      "epoch": 0.5155112983531214,
+      "loss": 0.203628808259964,
+      "loss_ce": 0.022781629115343094,
+      "loss_iou": 1.1514043807983398,
+      "loss_num": 0.1806640625,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 232654008,
+      "step": 1346
+    },
+    {
+      "epoch": 0.5158942933741861,
+      "grad_norm": 16.390795807993168,
+      "learning_rate": 5e-06,
+      "loss": 0.1773,
+      "num_input_tokens_seen": 232826856,
+      "step": 1347
+    },
+    {
+      "epoch": 0.5158942933741861,
+      "loss": 0.1825709044933319,
+      "loss_ce": 0.02021739073097706,
+      "loss_iou": 1.0611228942871094,
+      "loss_num": 0.162109375,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 232826856,
+      "step": 1347
+    },
+    {
+      "epoch": 0.5162772883952509,
+      "grad_norm": 54.976880532330874,
+      "learning_rate": 5e-06,
+      "loss": 0.2725,
+      "num_input_tokens_seen": 232999616,
+      "step": 1348
+    },
+    {
+      "epoch": 0.5162772883952509,
+      "loss": 0.29398179054260254,
+      "loss_ce": 0.020727401599287987,
+      "loss_iou": 1.1835370063781738,
+      "loss_num": 0.2734375,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 232999616,
+      "step": 1348
+    },
+    {
+      "epoch": 0.5166602834163155,
+      "grad_norm": 32.01183554460165,
+      "learning_rate": 5e-06,
+      "loss": 0.2167,
+      "num_input_tokens_seen": 233172416,
+      "step": 1349
+    },
+    {
+      "epoch": 0.5166602834163155,
+      "loss": 0.19804415106773376,
+      "loss_ce": 0.020859088748693466,
+      "loss_iou": 1.0127989053726196,
+      "loss_num": 0.1767578125,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 233172416,
+      "step": 1349
+    },
+    {
+      "epoch": 0.5170432784373803,
+      "grad_norm": 28.674047907310722,
+      "learning_rate": 5e-06,
+      "loss": 0.2378,
+      "num_input_tokens_seen": 233345096,
+      "step": 1350
+    },
+    {
+      "epoch": 0.5170432784373803,
+      "loss": 0.2748212218284607,
+      "loss_ce": 0.019999457523226738,
+      "loss_iou": 1.0336179733276367,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 233345096,
+      "step": 1350
+    },
+    {
+      "epoch": 0.517426273458445,
+      "grad_norm": 35.15392862620632,
+      "learning_rate": 5e-06,
+      "loss": 0.1947,
+      "num_input_tokens_seen": 233517904,
+      "step": 1351
+    },
+    {
+      "epoch": 0.517426273458445,
+      "loss": 0.1975768804550171,
+      "loss_ce": 0.01978147029876709,
+      "loss_iou": 1.0122634172439575,
+      "loss_num": 0.177734375,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 233517904,
+      "step": 1351
+    },
+    {
+      "epoch": 0.5178092684795098,
+      "grad_norm": 33.684261942202,
+      "learning_rate": 5e-06,
+      "loss": 0.2148,
+      "num_input_tokens_seen": 233690808,
+      "step": 1352
+    },
+    {
+      "epoch": 0.5178092684795098,
+      "loss": 0.240241140127182,
+      "loss_ce": 0.022589776664972305,
+      "loss_iou": 1.0366296768188477,
+      "loss_num": 0.2177734375,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 233690808,
+      "step": 1352
+    },
+    {
+      "epoch": 0.5181922635005745,
+      "grad_norm": 31.301514112753125,
+      "learning_rate": 5e-06,
+      "loss": 0.2582,
+      "num_input_tokens_seen": 233863688,
+      "step": 1353
+    },
+    {
+      "epoch": 0.5181922635005745,
+      "loss": 0.20045393705368042,
+      "loss_ce": 0.01960677094757557,
+      "loss_iou": 1.0051530599594116,
+      "loss_num": 0.1806640625,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 233863688,
+      "step": 1353
+    },
+    {
+      "epoch": 0.5185752585216392,
+      "grad_norm": 33.153027066248825,
+      "learning_rate": 5e-06,
+      "loss": 0.2091,
+      "num_input_tokens_seen": 234036264,
+      "step": 1354
+    },
+    {
+      "epoch": 0.5185752585216392,
+      "loss": 0.20700308680534363,
+      "loss_ce": 0.021090004593133926,
+      "loss_iou": 1.0467784404754639,
+      "loss_num": 0.185546875,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 234036264,
+      "step": 1354
+    },
+    {
+      "epoch": 0.518958253542704,
+      "grad_norm": 38.335258783336045,
+      "learning_rate": 5e-06,
+      "loss": 0.2545,
+      "num_input_tokens_seen": 234209368,
+      "step": 1355
+    },
+    {
+      "epoch": 0.518958253542704,
+      "loss": 0.2540934085845947,
+      "loss_ce": 0.02008461207151413,
+      "loss_iou": 1.0373358726501465,
+      "loss_num": 0.234375,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 234209368,
+      "step": 1355
+    },
+    {
+      "epoch": 0.5193412485637686,
+      "grad_norm": 33.080406690400075,
+      "learning_rate": 5e-06,
+      "loss": 0.1793,
+      "num_input_tokens_seen": 234382384,
+      "step": 1356
+    },
+    {
+      "epoch": 0.5193412485637686,
+      "loss": 0.19170862436294556,
+      "loss_ce": 0.02196984738111496,
+      "loss_iou": 1.003777027130127,
+      "loss_num": 0.169921875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 234382384,
+      "step": 1356
+    },
+    {
+      "epoch": 0.5197242435848334,
+      "grad_norm": 29.393792328760647,
+      "learning_rate": 5e-06,
+      "loss": 0.2662,
+      "num_input_tokens_seen": 234555088,
+      "step": 1357
+    },
+    {
+      "epoch": 0.5197242435848334,
+      "loss": 0.3434233069419861,
+      "loss_ce": 0.020547322928905487,
+      "loss_iou": 1.283623456954956,
+      "loss_num": 0.322265625,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 234555088,
+      "step": 1357
+    },
+    {
+      "epoch": 0.5201072386058981,
+      "grad_norm": 29.68426568876046,
+      "learning_rate": 5e-06,
+      "loss": 0.1527,
+      "num_input_tokens_seen": 234728136,
+      "step": 1358
+    },
+    {
+      "epoch": 0.5201072386058981,
+      "loss": 0.1551058292388916,
+      "loss_ce": 0.02223229594528675,
+      "loss_iou": 1.0033645629882812,
+      "loss_num": 0.1328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 234728136,
+      "step": 1358
+    },
+    {
+      "epoch": 0.5204902336269629,
+      "grad_norm": 41.63303639801133,
+      "learning_rate": 5e-06,
+      "loss": 0.3478,
+      "num_input_tokens_seen": 234900752,
+      "step": 1359
+    },
+    {
+      "epoch": 0.5204902336269629,
+      "loss": 0.2942076325416565,
+      "loss_ce": 0.020770136266946793,
+      "loss_iou": 1.0545668601989746,
+      "loss_num": 0.2734375,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 234900752,
+      "step": 1359
+    },
+    {
+      "epoch": 0.5208732286480275,
+      "grad_norm": 38.08271812322465,
+      "learning_rate": 5e-06,
+      "loss": 0.2165,
+      "num_input_tokens_seen": 235073496,
+      "step": 1360
+    },
+    {
+      "epoch": 0.5208732286480275,
+      "loss": 0.2514550983905792,
+      "loss_ce": 0.020253915339708328,
+      "loss_iou": 1.036755084991455,
+      "loss_num": 0.2314453125,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 235073496,
+      "step": 1360
+    },
+    {
+      "epoch": 0.5212562236690923,
+      "grad_norm": 48.7808133245496,
+      "learning_rate": 5e-06,
+      "loss": 0.3216,
+      "num_input_tokens_seen": 235246760,
+      "step": 1361
+    },
+    {
+      "epoch": 0.5212562236690923,
+      "loss": 0.2707056403160095,
+      "loss_ce": 0.02033940888941288,
+      "loss_iou": 1.0714623928070068,
+      "loss_num": 0.25,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 235246760,
+      "step": 1361
+    },
+    {
+      "epoch": 0.5216392186901571,
+      "grad_norm": 43.49549644385844,
+      "learning_rate": 5e-06,
+      "loss": 0.2786,
+      "num_input_tokens_seen": 235419320,
+      "step": 1362
+    },
+    {
+      "epoch": 0.5216392186901571,
+      "loss": 0.26690515875816345,
+      "loss_ce": 0.022093143314123154,
+      "loss_iou": 1.0660336017608643,
+      "loss_num": 0.2451171875,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 235419320,
+      "step": 1362
+    },
+    {
+      "epoch": 0.5220222137112217,
+      "grad_norm": 29.674097574665602,
+      "learning_rate": 5e-06,
+      "loss": 0.3186,
+      "num_input_tokens_seen": 235589272,
+      "step": 1363
+    },
+    {
+      "epoch": 0.5220222137112217,
+      "loss": 0.3059327006340027,
+      "loss_ce": 0.019799869507551193,
+      "loss_iou": 1.036852478981018,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 235589272,
+      "step": 1363
+    },
+    {
+      "epoch": 0.5224052087322865,
+      "grad_norm": 20.00999596008343,
+      "learning_rate": 5e-06,
+      "loss": 0.1837,
+      "num_input_tokens_seen": 235762120,
+      "step": 1364
+    },
+    {
+      "epoch": 0.5224052087322865,
+      "loss": 0.1387934386730194,
+      "loss_ce": 0.02075144834816456,
+      "loss_iou": 1.0177533626556396,
+      "loss_num": 0.1181640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 235762120,
+      "step": 1364
+    },
+    {
+      "epoch": 0.5227882037533512,
+      "grad_norm": 35.35985928537426,
+      "learning_rate": 5e-06,
+      "loss": 0.2153,
+      "num_input_tokens_seen": 235934800,
+      "step": 1365
+    },
+    {
+      "epoch": 0.5227882037533512,
+      "loss": 0.24349024891853333,
+      "loss_ce": 0.021322278305888176,
+      "loss_iou": 1.007620096206665,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 235934800,
+      "step": 1365
+    },
+    {
+      "epoch": 0.523171198774416,
+      "grad_norm": 37.38076587715216,
+      "learning_rate": 5e-06,
+      "loss": 0.2603,
+      "num_input_tokens_seen": 236107896,
+      "step": 1366
+    },
+    {
+      "epoch": 0.523171198774416,
+      "loss": 0.2549581229686737,
+      "loss_ce": 0.021681763231754303,
+      "loss_iou": 1.0397226810455322,
+      "loss_num": 0.2333984375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 236107896,
+      "step": 1366
+    },
+    {
+      "epoch": 0.5235541937954806,
+      "grad_norm": 28.733650081221064,
+      "learning_rate": 5e-06,
+      "loss": 0.2669,
+      "num_input_tokens_seen": 236280776,
+      "step": 1367
+    },
+    {
+      "epoch": 0.5235541937954806,
+      "loss": 0.26827141642570496,
+      "loss_ce": 0.022421810775995255,
+      "loss_iou": 1.0088081359863281,
+      "loss_num": 0.24609375,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 236280776,
+      "step": 1367
+    },
+    {
+      "epoch": 0.5239371888165454,
+      "grad_norm": 36.763076262248894,
+      "learning_rate": 5e-06,
+      "loss": 0.2298,
+      "num_input_tokens_seen": 236453704,
+      "step": 1368
+    },
+    {
+      "epoch": 0.5239371888165454,
+      "loss": 0.24433454871177673,
+      "loss_ce": 0.022349681705236435,
+      "loss_iou": 1.0200729370117188,
+      "loss_num": 0.2216796875,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 236453704,
+      "step": 1368
+    },
+    {
+      "epoch": 0.5243201838376101,
+      "grad_norm": 54.352311727641876,
+      "learning_rate": 5e-06,
+      "loss": 0.2858,
+      "num_input_tokens_seen": 236626752,
+      "step": 1369
+    },
+    {
+      "epoch": 0.5243201838376101,
+      "loss": 0.2584747076034546,
+      "loss_ce": 0.021780386567115784,
+      "loss_iou": 1.018236756324768,
+      "loss_num": 0.236328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 236626752,
+      "step": 1369
+    },
+    {
+      "epoch": 0.5247031788586748,
+      "grad_norm": 39.28697472365648,
+      "learning_rate": 5e-06,
+      "loss": 0.2362,
+      "num_input_tokens_seen": 236799888,
+      "step": 1370
+    },
+    {
+      "epoch": 0.5247031788586748,
+      "loss": 0.24799107015132904,
+      "loss_ce": 0.022649260237812996,
+      "loss_iou": 1.01529860496521,
+      "loss_num": 0.2255859375,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 236799888,
+      "step": 1370
+    },
+    {
+      "epoch": 0.5250861738797395,
+      "grad_norm": 39.85297075105549,
+      "learning_rate": 5e-06,
+      "loss": 0.2601,
+      "num_input_tokens_seen": 236972848,
+      "step": 1371
+    },
+    {
+      "epoch": 0.5250861738797395,
+      "loss": 0.2791779339313507,
+      "loss_ce": 0.020510949194431305,
+      "loss_iou": 1.0840009450912476,
+      "loss_num": 0.2578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 236972848,
+      "step": 1371
+    },
+    {
+      "epoch": 0.5254691689008043,
+      "grad_norm": 23.883892009766925,
+      "learning_rate": 5e-06,
+      "loss": 0.2045,
+      "num_input_tokens_seen": 237145664,
+      "step": 1372
+    },
+    {
+      "epoch": 0.5254691689008043,
+      "loss": 0.21139544248580933,
+      "loss_ce": 0.020050212740898132,
+      "loss_iou": 1.0639290809631348,
+      "loss_num": 0.19140625,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 237145664,
+      "step": 1372
+    },
+    {
+      "epoch": 0.5258521639218691,
+      "grad_norm": 48.69033583444745,
+      "learning_rate": 5e-06,
+      "loss": 0.2613,
+      "num_input_tokens_seen": 237318480,
+      "step": 1373
+    },
+    {
+      "epoch": 0.5258521639218691,
+      "loss": 0.23347464203834534,
+      "loss_ce": 0.019729536026716232,
+      "loss_iou": 1.0228080749511719,
+      "loss_num": 0.2138671875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 237318480,
+      "step": 1373
+    },
+    {
+      "epoch": 0.5262351589429337,
+      "grad_norm": 33.13288312093415,
+      "learning_rate": 5e-06,
+      "loss": 0.2333,
+      "num_input_tokens_seen": 237491280,
+      "step": 1374
+    },
+    {
+      "epoch": 0.5262351589429337,
+      "loss": 0.1887282282114029,
+      "loss_ce": 0.020820513367652893,
+      "loss_iou": 1.0077931880950928,
+      "loss_num": 0.16796875,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 237491280,
+      "step": 1374
+    },
+    {
+      "epoch": 0.5266181539639985,
+      "grad_norm": 21.103176966025885,
+      "learning_rate": 5e-06,
+      "loss": 0.2045,
+      "num_input_tokens_seen": 237664680,
+      "step": 1375
+    },
+    {
+      "epoch": 0.5266181539639985,
+      "loss": 0.2112509310245514,
+      "loss_ce": 0.02008882910013199,
+      "loss_iou": 1.010251760482788,
+      "loss_num": 0.19140625,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 237664680,
+      "step": 1375
+    },
+    {
+      "epoch": 0.5270011489850632,
+      "grad_norm": 34.49932971770472,
+      "learning_rate": 5e-06,
+      "loss": 0.1732,
+      "num_input_tokens_seen": 237837416,
+      "step": 1376
+    },
+    {
+      "epoch": 0.5270011489850632,
+      "loss": 0.16933594644069672,
+      "loss_ce": 0.020043957978487015,
+      "loss_iou": 1.0145107507705688,
+      "loss_num": 0.1494140625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 237837416,
+      "step": 1376
+    },
+    {
+      "epoch": 0.5273841440061279,
+      "grad_norm": 47.41067228375037,
+      "learning_rate": 5e-06,
+      "loss": 0.2119,
+      "num_input_tokens_seen": 238010416,
+      "step": 1377
+    },
+    {
+      "epoch": 0.5273841440061279,
+      "loss": 0.2054481953382492,
+      "loss_ce": 0.020755819976329803,
+      "loss_iou": 1.0072085857391357,
+      "loss_num": 0.1845703125,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 238010416,
+      "step": 1377
+    },
+    {
+      "epoch": 0.5277671390271926,
+      "grad_norm": 39.93611250459804,
+      "learning_rate": 5e-06,
+      "loss": 0.2428,
+      "num_input_tokens_seen": 238183624,
+      "step": 1378
+    },
+    {
+      "epoch": 0.5277671390271926,
+      "loss": 0.22903317213058472,
+      "loss_ce": 0.02218502014875412,
+      "loss_iou": 1.0179922580718994,
+      "loss_num": 0.20703125,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 238183624,
+      "step": 1378
+    },
+    {
+      "epoch": 0.5281501340482574,
+      "grad_norm": 27.4562870797303,
+      "learning_rate": 5e-06,
+      "loss": 0.2511,
+      "num_input_tokens_seen": 238356984,
+      "step": 1379
+    },
+    {
+      "epoch": 0.5281501340482574,
+      "loss": 0.2706752419471741,
+      "loss_ce": 0.021285587921738625,
+      "loss_iou": 1.0386700630187988,
+      "loss_num": 0.2490234375,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 238356984,
+      "step": 1379
+    },
+    {
+      "epoch": 0.5285331290693221,
+      "grad_norm": 27.974109028105044,
+      "learning_rate": 5e-06,
+      "loss": 0.2066,
+      "num_input_tokens_seen": 238530016,
+      "step": 1380
+    },
+    {
+      "epoch": 0.5285331290693221,
+      "loss": 0.2735742926597595,
+      "loss_ce": 0.020034248009324074,
+      "loss_iou": 1.0495893955230713,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 238530016,
+      "step": 1380
+    },
+    {
+      "epoch": 0.5289161240903868,
+      "grad_norm": 45.21358194144243,
+      "learning_rate": 5e-06,
+      "loss": 0.1966,
+      "num_input_tokens_seen": 238703256,
+      "step": 1381
+    },
+    {
+      "epoch": 0.5289161240903868,
+      "loss": 0.20240136981010437,
+      "loss_ce": 0.021065907552838326,
+      "loss_iou": 1.013717770576477,
+      "loss_num": 0.181640625,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 238703256,
+      "step": 1381
+    },
+    {
+      "epoch": 0.5292991191114516,
+      "grad_norm": 42.50025687762121,
+      "learning_rate": 5e-06,
+      "loss": 0.3085,
+      "num_input_tokens_seen": 238876048,
+      "step": 1382
+    },
+    {
+      "epoch": 0.5292991191114516,
+      "loss": 0.29392093420028687,
+      "loss_ce": 0.021826203912496567,
+      "loss_iou": 1.034827709197998,
+      "loss_num": 0.271484375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 238876048,
+      "step": 1382
+    },
+    {
+      "epoch": 0.5296821141325163,
+      "grad_norm": 30.425252298057856,
+      "learning_rate": 5e-06,
+      "loss": 0.2059,
+      "num_input_tokens_seen": 239049176,
+      "step": 1383
+    },
+    {
+      "epoch": 0.5296821141325163,
+      "loss": 0.2247311770915985,
+      "loss_ce": 0.02093479037284851,
+      "loss_iou": 1.028517723083496,
+      "loss_num": 0.2041015625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 239049176,
+      "step": 1383
+    },
+    {
+      "epoch": 0.530065109153581,
+      "grad_norm": 33.61457679214779,
+      "learning_rate": 5e-06,
+      "loss": 0.2666,
+      "num_input_tokens_seen": 239222136,
+      "step": 1384
+    },
+    {
+      "epoch": 0.530065109153581,
+      "loss": 0.2373058795928955,
+      "loss_ce": 0.022034896537661552,
+      "loss_iou": 1.0021536350250244,
+      "loss_num": 0.21484375,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 239222136,
+      "step": 1384
+    },
+    {
+      "epoch": 0.5304481041746457,
+      "grad_norm": 39.7211041095127,
+      "learning_rate": 5e-06,
+      "loss": 0.2564,
+      "num_input_tokens_seen": 239395320,
+      "step": 1385
+    },
+    {
+      "epoch": 0.5304481041746457,
+      "loss": 0.2629600167274475,
+      "loss_ce": 0.01991802640259266,
+      "loss_iou": 1.1126071214675903,
+      "loss_num": 0.2431640625,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 239395320,
+      "step": 1385
+    },
+    {
+      "epoch": 0.5308310991957105,
+      "grad_norm": 46.416829273045806,
+      "learning_rate": 5e-06,
+      "loss": 0.33,
+      "num_input_tokens_seen": 239568768,
+      "step": 1386
+    },
+    {
+      "epoch": 0.5308310991957105,
+      "loss": 0.27796030044555664,
+      "loss_ce": 0.022345073521137238,
+      "loss_iou": 1.0250229835510254,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 239568768,
+      "step": 1386
+    },
+    {
+      "epoch": 0.5312140942167752,
+      "grad_norm": 31.069103071591833,
+      "learning_rate": 5e-06,
+      "loss": 0.2698,
+      "num_input_tokens_seen": 239741832,
+      "step": 1387
+    },
+    {
+      "epoch": 0.5312140942167752,
+      "loss": 0.3004428446292877,
+      "loss_ce": 0.02065768837928772,
+      "loss_iou": 1.1097912788391113,
+      "loss_num": 0.279296875,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 239741832,
+      "step": 1387
+    },
+    {
+      "epoch": 0.5315970892378399,
+      "grad_norm": 26.256177961290657,
+      "learning_rate": 5e-06,
+      "loss": 0.2724,
+      "num_input_tokens_seen": 239914976,
+      "step": 1388
+    },
+    {
+      "epoch": 0.5315970892378399,
+      "loss": 0.2281334102153778,
+      "loss_ce": 0.021834567189216614,
+      "loss_iou": 1.005028486251831,
+      "loss_num": 0.2060546875,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 239914976,
+      "step": 1388
+    },
+    {
+      "epoch": 0.5319800842589046,
+      "grad_norm": 28.352365425330024,
+      "learning_rate": 5e-06,
+      "loss": 0.3002,
+      "num_input_tokens_seen": 240087576,
+      "step": 1389
+    },
+    {
+      "epoch": 0.5319800842589046,
+      "loss": 0.24115517735481262,
+      "loss_ce": 0.020146869122982025,
+      "loss_iou": 1.0026452541351318,
+      "loss_num": 0.220703125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 240087576,
+      "step": 1389
+    },
+    {
+      "epoch": 0.5323630792799694,
+      "grad_norm": 33.1114051606871,
+      "learning_rate": 5e-06,
+      "loss": 0.2046,
+      "num_input_tokens_seen": 240260296,
+      "step": 1390
+    },
+    {
+      "epoch": 0.5323630792799694,
+      "loss": 0.237400084733963,
+      "loss_ce": 0.02023700624704361,
+      "loss_iou": 1.0121262073516846,
+      "loss_num": 0.216796875,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 240260296,
+      "step": 1390
+    },
+    {
+      "epoch": 0.532746074301034,
+      "grad_norm": 43.21014613584952,
+      "learning_rate": 5e-06,
+      "loss": 0.271,
+      "num_input_tokens_seen": 240433432,
+      "step": 1391
+    },
+    {
+      "epoch": 0.532746074301034,
+      "loss": 0.3160042464733124,
+      "loss_ce": 0.02034994773566723,
+      "loss_iou": 1.059490442276001,
+      "loss_num": 0.294921875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 240433432,
+      "step": 1391
+    },
+    {
+      "epoch": 0.5331290693220988,
+      "grad_norm": 28.60585062243727,
+      "learning_rate": 5e-06,
+      "loss": 0.2541,
+      "num_input_tokens_seen": 240606648,
+      "step": 1392
+    },
+    {
+      "epoch": 0.5331290693220988,
+      "loss": 0.2064480185508728,
+      "loss_ce": 0.021145276725292206,
+      "loss_iou": 1.0031088590621948,
+      "loss_num": 0.185546875,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 240606648,
+      "step": 1392
+    },
+    {
+      "epoch": 0.5335120643431636,
+      "grad_norm": 24.29340048658995,
+      "learning_rate": 5e-06,
+      "loss": 0.2703,
+      "num_input_tokens_seen": 240779056,
+      "step": 1393
+    },
+    {
+      "epoch": 0.5335120643431636,
+      "loss": 0.24477875232696533,
+      "loss_ce": 0.021756291389465332,
+      "loss_iou": 1.0061001777648926,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 240779056,
+      "step": 1393
+    },
+    {
+      "epoch": 0.5338950593642283,
+      "grad_norm": 34.78546575491859,
+      "learning_rate": 5e-06,
+      "loss": 0.2491,
+      "num_input_tokens_seen": 240952344,
+      "step": 1394
+    },
+    {
+      "epoch": 0.5338950593642283,
+      "loss": 0.23100616037845612,
+      "loss_ce": 0.021655568853020668,
+      "loss_iou": 1.0422629117965698,
+      "loss_num": 0.208984375,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 240952344,
+      "step": 1394
+    },
+    {
+      "epoch": 0.534278054385293,
+      "grad_norm": 42.96399669803133,
+      "learning_rate": 5e-06,
+      "loss": 0.3026,
+      "num_input_tokens_seen": 241125424,
+      "step": 1395
+    },
+    {
+      "epoch": 0.534278054385293,
+      "loss": 0.3512245714664459,
+      "loss_ce": 0.022367149591445923,
+      "loss_iou": 1.081865668296814,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 241125424,
+      "step": 1395
+    },
+    {
+      "epoch": 0.5346610494063577,
+      "grad_norm": 31.602337959979078,
+      "learning_rate": 5e-06,
+      "loss": 0.2021,
+      "num_input_tokens_seen": 241298240,
+      "step": 1396
+    },
+    {
+      "epoch": 0.5346610494063577,
+      "loss": 0.20335182547569275,
+      "loss_ce": 0.020673586055636406,
+      "loss_iou": 1.04007089138031,
+      "loss_num": 0.1826171875,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 241298240,
+      "step": 1396
+    },
+    {
+      "epoch": 0.5350440444274225,
+      "grad_norm": 35.38432134211127,
+      "learning_rate": 5e-06,
+      "loss": 0.2385,
+      "num_input_tokens_seen": 241471400,
+      "step": 1397
+    },
+    {
+      "epoch": 0.5350440444274225,
+      "loss": 0.26554152369499207,
+      "loss_ce": 0.021034665405750275,
+      "loss_iou": 1.047926664352417,
+      "loss_num": 0.244140625,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 241471400,
+      "step": 1397
+    },
+    {
+      "epoch": 0.5354270394484871,
+      "grad_norm": 37.54265941196159,
+      "learning_rate": 5e-06,
+      "loss": 0.2632,
+      "num_input_tokens_seen": 241644472,
+      "step": 1398
+    },
+    {
+      "epoch": 0.5354270394484871,
+      "loss": 0.230668842792511,
+      "loss_ce": 0.020768944174051285,
+      "loss_iou": 1.0046364068984985,
+      "loss_num": 0.2099609375,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 241644472,
+      "step": 1398
+    },
+    {
+      "epoch": 0.5358100344695519,
+      "grad_norm": 36.60518748131619,
+      "learning_rate": 5e-06,
+      "loss": 0.2057,
+      "num_input_tokens_seen": 241817608,
+      "step": 1399
+    },
+    {
+      "epoch": 0.5358100344695519,
+      "loss": 0.20266515016555786,
+      "loss_ce": 0.020231062546372414,
+      "loss_iou": 1.0437780618667603,
+      "loss_num": 0.1826171875,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 241817608,
+      "step": 1399
+    },
+    {
+      "epoch": 0.5361930294906166,
+      "grad_norm": 30.71190131618502,
+      "learning_rate": 5e-06,
+      "loss": 0.1983,
+      "num_input_tokens_seen": 241990056,
+      "step": 1400
+    },
+    {
+      "epoch": 0.5361930294906166,
+      "loss": 0.24845024943351746,
+      "loss_ce": 0.019843079149723053,
+      "loss_iou": 1.0894261598587036,
+      "loss_num": 0.228515625,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 241990056,
+      "step": 1400
+    },
+    {
+      "epoch": 0.5365760245116813,
+      "grad_norm": 34.43627927450469,
+      "learning_rate": 5e-06,
+      "loss": 0.1957,
+      "num_input_tokens_seen": 242163112,
+      "step": 1401
+    },
+    {
+      "epoch": 0.5365760245116813,
+      "loss": 0.222532257437706,
+      "loss_ce": 0.022520046681165695,
+      "loss_iou": 1.0313280820846558,
+      "loss_num": 0.2001953125,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 242163112,
+      "step": 1401
+    },
+    {
+      "epoch": 0.536959019532746,
+      "grad_norm": 38.91070144458712,
+      "learning_rate": 5e-06,
+      "loss": 0.2387,
+      "num_input_tokens_seen": 242336248,
+      "step": 1402
+    },
+    {
+      "epoch": 0.536959019532746,
+      "loss": 0.24522612988948822,
+      "loss_ce": 0.020555716007947922,
+      "loss_iou": 1.0357367992401123,
+      "loss_num": 0.224609375,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 242336248,
+      "step": 1402
+    },
+    {
+      "epoch": 0.5373420145538108,
+      "grad_norm": 42.22580467490461,
+      "learning_rate": 5e-06,
+      "loss": 0.2548,
+      "num_input_tokens_seen": 242509344,
+      "step": 1403
+    },
+    {
+      "epoch": 0.5373420145538108,
+      "loss": 0.2751389145851135,
+      "loss_ce": 0.021232664585113525,
+      "loss_iou": 1.1418867111206055,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 242509344,
+      "step": 1403
+    },
+    {
+      "epoch": 0.5377250095748756,
+      "grad_norm": 25.80160854409078,
+      "learning_rate": 5e-06,
+      "loss": 0.2568,
+      "num_input_tokens_seen": 242682240,
+      "step": 1404
+    },
+    {
+      "epoch": 0.5377250095748756,
+      "loss": 0.21290144324302673,
+      "loss_ce": 0.01978621445596218,
+      "loss_iou": 1.0695327520370483,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 242682240,
+      "step": 1404
+    },
+    {
+      "epoch": 0.5381080045959402,
+      "grad_norm": 29.013355076490416,
+      "learning_rate": 5e-06,
+      "loss": 0.1944,
+      "num_input_tokens_seen": 242855424,
+      "step": 1405
+    },
+    {
+      "epoch": 0.5381080045959402,
+      "loss": 0.18262693285942078,
+      "loss_ce": 0.021372050046920776,
+      "loss_iou": 1.0200414657592773,
+      "loss_num": 0.1611328125,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 242855424,
+      "step": 1405
+    },
+    {
+      "epoch": 0.538490999617005,
+      "grad_norm": 34.489133825620655,
+      "learning_rate": 5e-06,
+      "loss": 0.2304,
+      "num_input_tokens_seen": 243024776,
+      "step": 1406
+    },
+    {
+      "epoch": 0.538490999617005,
+      "loss": 0.21656718850135803,
+      "loss_ce": 0.01978985220193863,
+      "loss_iou": 1.0040063858032227,
+      "loss_num": 0.197265625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 243024776,
+      "step": 1406
+    },
+    {
+      "epoch": 0.5388739946380697,
+      "grad_norm": 40.11347155892702,
+      "learning_rate": 5e-06,
+      "loss": 0.2875,
+      "num_input_tokens_seen": 243197944,
+      "step": 1407
+    },
+    {
+      "epoch": 0.5388739946380697,
+      "loss": 0.24362048506736755,
+      "loss_ce": 0.020109744742512703,
+      "loss_iou": 1.0843185186386108,
+      "loss_num": 0.2236328125,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 243197944,
+      "step": 1407
+    },
+    {
+      "epoch": 0.5392569896591344,
+      "grad_norm": 29.6068668671716,
+      "learning_rate": 5e-06,
+      "loss": 0.2132,
+      "num_input_tokens_seen": 243371016,
+      "step": 1408
+    },
+    {
+      "epoch": 0.5392569896591344,
+      "loss": 0.25290000438690186,
+      "loss_ce": 0.0213326346129179,
+      "loss_iou": 1.2228739261627197,
+      "loss_num": 0.2314453125,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 243371016,
+      "step": 1408
+    },
+    {
+      "epoch": 0.5396399846801991,
+      "grad_norm": 25.28544123695245,
+      "learning_rate": 5e-06,
+      "loss": 0.2863,
+      "num_input_tokens_seen": 243544064,
+      "step": 1409
+    },
+    {
+      "epoch": 0.5396399846801991,
+      "loss": 0.24945750832557678,
+      "loss_ce": 0.019232889637351036,
+      "loss_iou": 1.1065994501113892,
+      "loss_num": 0.23046875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 243544064,
+      "step": 1409
+    },
+    {
+      "epoch": 0.5400229797012639,
+      "grad_norm": 35.31818721965278,
+      "learning_rate": 5e-06,
+      "loss": 0.281,
+      "num_input_tokens_seen": 243717064,
+      "step": 1410
+    },
+    {
+      "epoch": 0.5400229797012639,
+      "loss": 0.29192155599594116,
+      "loss_ce": 0.020437180995941162,
+      "loss_iou": 1.0520268678665161,
+      "loss_num": 0.271484375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 243717064,
+      "step": 1410
+    },
+    {
+      "epoch": 0.5404059747223287,
+      "grad_norm": 36.389040810693295,
+      "learning_rate": 5e-06,
+      "loss": 0.2786,
+      "num_input_tokens_seen": 243889960,
+      "step": 1411
+    },
+    {
+      "epoch": 0.5404059747223287,
+      "loss": 0.2415573000907898,
+      "loss_ce": 0.021037276834249496,
+      "loss_iou": 1.1709904670715332,
+      "loss_num": 0.220703125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 243889960,
+      "step": 1411
+    },
+    {
+      "epoch": 0.5407889697433933,
+      "grad_norm": 24.93525037623869,
+      "learning_rate": 5e-06,
+      "loss": 0.2767,
+      "num_input_tokens_seen": 244062832,
+      "step": 1412
+    },
+    {
+      "epoch": 0.5407889697433933,
+      "loss": 0.3619447946548462,
+      "loss_ce": 0.021856890991330147,
+      "loss_iou": 1.0857049226760864,
+      "loss_num": 0.33984375,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 244062832,
+      "step": 1412
+    },
+    {
+      "epoch": 0.5411719647644581,
+      "grad_norm": 34.52752507359736,
+      "learning_rate": 5e-06,
+      "loss": 0.2621,
+      "num_input_tokens_seen": 244235680,
+      "step": 1413
+    },
+    {
+      "epoch": 0.5411719647644581,
+      "loss": 0.2576747536659241,
+      "loss_ce": 0.01963762938976288,
+      "loss_iou": 1.0238568782806396,
+      "loss_num": 0.23828125,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 244235680,
+      "step": 1413
+    },
+    {
+      "epoch": 0.5415549597855228,
+      "grad_norm": 22.679668038653407,
+      "learning_rate": 5e-06,
+      "loss": 0.2084,
+      "num_input_tokens_seen": 244408328,
+      "step": 1414
+    },
+    {
+      "epoch": 0.5415549597855228,
+      "loss": 0.1996358335018158,
+      "loss_ce": 0.018605569377541542,
+      "loss_iou": 1.0567339658737183,
+      "loss_num": 0.1806640625,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 244408328,
+      "step": 1414
+    },
+    {
+      "epoch": 0.5419379548065875,
+      "grad_norm": 31.1917342671544,
+      "learning_rate": 5e-06,
+      "loss": 0.2115,
+      "num_input_tokens_seen": 244581744,
+      "step": 1415
+    },
+    {
+      "epoch": 0.5419379548065875,
+      "loss": 0.21254712343215942,
+      "loss_ce": 0.02150709182024002,
+      "loss_iou": 1.0056824684143066,
+      "loss_num": 0.19140625,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 244581744,
+      "step": 1415
+    },
+    {
+      "epoch": 0.5423209498276522,
+      "grad_norm": 26.335178729234826,
+      "learning_rate": 5e-06,
+      "loss": 0.2533,
+      "num_input_tokens_seen": 244754712,
+      "step": 1416
+    },
+    {
+      "epoch": 0.5423209498276522,
+      "loss": 0.27759096026420593,
+      "loss_ce": 0.020693987607955933,
+      "loss_iou": 1.1383533477783203,
+      "loss_num": 0.2578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 244754712,
+      "step": 1416
+    },
+    {
+      "epoch": 0.542703944848717,
+      "grad_norm": 25.083193859683952,
+      "learning_rate": 5e-06,
+      "loss": 0.2966,
+      "num_input_tokens_seen": 244927200,
+      "step": 1417
+    },
+    {
+      "epoch": 0.542703944848717,
+      "loss": 0.18403224647045135,
+      "loss_ce": 0.020763196051120758,
+      "loss_iou": -3.9631676720860365e+17,
+      "loss_num": 0.1630859375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 244927200,
+      "step": 1417
+    },
+    {
+      "epoch": 0.5430869398697817,
+      "grad_norm": 26.016439424328823,
+      "learning_rate": 5e-06,
+      "loss": 0.2138,
+      "num_input_tokens_seen": 245100368,
+      "step": 1418
+    },
+    {
+      "epoch": 0.5430869398697817,
+      "loss": 0.1996343731880188,
+      "loss_ce": 0.018909268081188202,
+      "loss_iou": 1.039953351020813,
+      "loss_num": 0.1806640625,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 245100368,
+      "step": 1418
+    },
+    {
+      "epoch": 0.5434699348908464,
+      "grad_norm": 28.21882526564885,
+      "learning_rate": 5e-06,
+      "loss": 0.2513,
+      "num_input_tokens_seen": 245272672,
+      "step": 1419
+    },
+    {
+      "epoch": 0.5434699348908464,
+      "loss": 0.22828856110572815,
+      "loss_ce": 0.021013177931308746,
+      "loss_iou": 1.0036940574645996,
+      "loss_num": 0.20703125,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 245272672,
+      "step": 1419
+    },
+    {
+      "epoch": 0.5438529299119111,
+      "grad_norm": 26.678920714651735,
+      "learning_rate": 5e-06,
+      "loss": 0.2133,
+      "num_input_tokens_seen": 245445504,
+      "step": 1420
+    },
+    {
+      "epoch": 0.5438529299119111,
+      "loss": 0.24139827489852905,
+      "loss_ce": 0.019413433969020844,
+      "loss_iou": 1.034382700920105,
+      "loss_num": 0.2216796875,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 245445504,
+      "step": 1420
+    },
+    {
+      "epoch": 0.5442359249329759,
+      "grad_norm": 41.33102167850886,
+      "learning_rate": 5e-06,
+      "loss": 0.2339,
+      "num_input_tokens_seen": 245618696,
+      "step": 1421
+    },
+    {
+      "epoch": 0.5442359249329759,
+      "loss": 0.23862941563129425,
+      "loss_ce": 0.019513199105858803,
+      "loss_iou": 1.044738531112671,
+      "loss_num": 0.21875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 245618696,
+      "step": 1421
+    },
+    {
+      "epoch": 0.5446189199540405,
+      "grad_norm": 37.35848943267274,
+      "learning_rate": 5e-06,
+      "loss": 0.2347,
+      "num_input_tokens_seen": 245791336,
+      "step": 1422
+    },
+    {
+      "epoch": 0.5446189199540405,
+      "loss": 0.2276022732257843,
+      "loss_ce": 0.020326875150203705,
+      "loss_iou": 1.112799048423767,
+      "loss_num": 0.20703125,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 245791336,
+      "step": 1422
+    },
+    {
+      "epoch": 0.5450019149751053,
+      "grad_norm": 34.4538962509511,
+      "learning_rate": 5e-06,
+      "loss": 0.2323,
+      "num_input_tokens_seen": 245964384,
+      "step": 1423
+    },
+    {
+      "epoch": 0.5450019149751053,
+      "loss": 0.2902798652648926,
+      "loss_ce": 0.02306796796619892,
+      "loss_iou": 1.0727440118789673,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 245964384,
+      "step": 1423
+    },
+    {
+      "epoch": 0.5453849099961701,
+      "grad_norm": 29.81776513589979,
+      "learning_rate": 5e-06,
+      "loss": 0.1872,
+      "num_input_tokens_seen": 246137000,
+      "step": 1424
+    },
+    {
+      "epoch": 0.5453849099961701,
+      "loss": 0.1663355976343155,
+      "loss_ce": 0.019851218909025192,
+      "loss_iou": 1.002077579498291,
+      "loss_num": 0.146484375,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 246137000,
+      "step": 1424
+    },
+    {
+      "epoch": 0.5457679050172348,
+      "grad_norm": 47.50351566884468,
+      "learning_rate": 5e-06,
+      "loss": 0.2471,
+      "num_input_tokens_seen": 246309744,
+      "step": 1425
+    },
+    {
+      "epoch": 0.5457679050172348,
+      "loss": 0.26415950059890747,
+      "loss_ce": 0.021972009912133217,
+      "loss_iou": 1.1133008003234863,
+      "loss_num": 0.2421875,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 246309744,
+      "step": 1425
+    },
+    {
+      "epoch": 0.5461509000382995,
+      "grad_norm": 35.42252092555776,
+      "learning_rate": 5e-06,
+      "loss": 0.2452,
+      "num_input_tokens_seen": 246482224,
+      "step": 1426
+    },
+    {
+      "epoch": 0.5461509000382995,
+      "loss": 0.22726401686668396,
+      "loss_ce": 0.01931723766028881,
+      "loss_iou": 1.0251805782318115,
+      "loss_num": 0.2080078125,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 246482224,
+      "step": 1426
+    },
+    {
+      "epoch": 0.5465338950593642,
+      "grad_norm": 26.07618375093433,
+      "learning_rate": 5e-06,
+      "loss": 0.2365,
+      "num_input_tokens_seen": 246655184,
+      "step": 1427
+    },
+    {
+      "epoch": 0.5465338950593642,
+      "loss": 0.22888973355293274,
+      "loss_ce": 0.020820889621973038,
+      "loss_iou": 1.027405023574829,
+      "loss_num": 0.2080078125,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 246655184,
+      "step": 1427
+    },
+    {
+      "epoch": 0.546916890080429,
+      "grad_norm": 27.74269664167861,
+      "learning_rate": 5e-06,
+      "loss": 0.1659,
+      "num_input_tokens_seen": 246828488,
+      "step": 1428
+    },
+    {
+      "epoch": 0.546916890080429,
+      "loss": 0.1319091022014618,
+      "loss_ce": 0.02119133621454239,
+      "loss_iou": 1.0033646821975708,
+      "loss_num": 0.11083984375,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 246828488,
+      "step": 1428
+    },
+    {
+      "epoch": 0.5472998851014936,
+      "grad_norm": 47.618605413464145,
+      "learning_rate": 5e-06,
+      "loss": 0.28,
+      "num_input_tokens_seen": 247001456,
+      "step": 1429
+    },
+    {
+      "epoch": 0.5472998851014936,
+      "loss": 0.24275851249694824,
+      "loss_ce": 0.0203463826328516,
+      "loss_iou": 1.0035829544067383,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 247001456,
+      "step": 1429
+    },
+    {
+      "epoch": 0.5476828801225584,
+      "grad_norm": 95.07469045236165,
+      "learning_rate": 5e-06,
+      "loss": 0.2775,
+      "num_input_tokens_seen": 247174344,
+      "step": 1430
+    },
+    {
+      "epoch": 0.5476828801225584,
+      "loss": 0.275626540184021,
+      "loss_ce": 0.019523024559020996,
+      "loss_iou": 1.013465166091919,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 247174344,
+      "step": 1430
+    },
+    {
+      "epoch": 0.5480658751436231,
+      "grad_norm": 39.65837448926188,
+      "learning_rate": 5e-06,
+      "loss": 0.2261,
+      "num_input_tokens_seen": 247347136,
+      "step": 1431
+    },
+    {
+      "epoch": 0.5480658751436231,
+      "loss": 0.1636388599872589,
+      "loss_ce": 0.019656913354992867,
+      "loss_iou": 1.0006221532821655,
+      "loss_num": 0.1435546875,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 247347136,
+      "step": 1431
+    },
+    {
+      "epoch": 0.5484488701646879,
+      "grad_norm": 26.931540948457975,
+      "learning_rate": 5e-06,
+      "loss": 0.2191,
+      "num_input_tokens_seen": 247520504,
+      "step": 1432
+    },
+    {
+      "epoch": 0.5484488701646879,
+      "loss": 0.20612883567810059,
+      "loss_ce": 0.02143644355237484,
+      "loss_iou": 1.0078787803649902,
+      "loss_num": 0.1845703125,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 247520504,
+      "step": 1432
+    },
+    {
+      "epoch": 0.5488318651857526,
+      "grad_norm": 28.208258312522524,
+      "learning_rate": 5e-06,
+      "loss": 0.2821,
+      "num_input_tokens_seen": 247693176,
+      "step": 1433
+    },
+    {
+      "epoch": 0.5488318651857526,
+      "loss": 0.3653810918331146,
+      "loss_ce": 0.019983142614364624,
+      "loss_iou": 1.0598523616790771,
+      "loss_num": 0.345703125,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 247693176,
+      "step": 1433
+    },
+    {
+      "epoch": 0.5492148602068173,
+      "grad_norm": 29.833269339360154,
+      "learning_rate": 5e-06,
+      "loss": 0.219,
+      "num_input_tokens_seen": 247866072,
+      "step": 1434
+    },
+    {
+      "epoch": 0.5492148602068173,
+      "loss": 0.1977815181016922,
+      "loss_ce": 0.019009539857506752,
+      "loss_iou": 1.0558912754058838,
+      "loss_num": 0.1787109375,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 247866072,
+      "step": 1434
+    },
+    {
+      "epoch": 0.5495978552278821,
+      "grad_norm": 33.24822640975801,
+      "learning_rate": 5e-06,
+      "loss": 0.2243,
+      "num_input_tokens_seen": 248038728,
+      "step": 1435
+    },
+    {
+      "epoch": 0.5495978552278821,
+      "loss": 0.2502945363521576,
+      "loss_ce": 0.01909336820244789,
+      "loss_iou": 1.1917321681976318,
+      "loss_num": 0.2314453125,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 248038728,
+      "step": 1435
+    },
+    {
+      "epoch": 0.5499808502489467,
+      "grad_norm": 21.834141828576332,
+      "learning_rate": 5e-06,
+      "loss": 0.2581,
+      "num_input_tokens_seen": 248211784,
+      "step": 1436
+    },
+    {
+      "epoch": 0.5499808502489467,
+      "loss": 0.1583734154701233,
+      "loss_ce": 0.0202508382499218,
+      "loss_iou": 1.0127925872802734,
+      "loss_num": 0.1376953125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 248211784,
+      "step": 1436
+    },
+    {
+      "epoch": 0.5503638452700115,
+      "grad_norm": 23.653747309977522,
+      "learning_rate": 5e-06,
+      "loss": 0.2427,
+      "num_input_tokens_seen": 248384840,
+      "step": 1437
+    },
+    {
+      "epoch": 0.5503638452700115,
+      "loss": 0.186224102973938,
+      "loss_ce": 0.02020847424864769,
+      "loss_iou": 1.003072738647461,
+      "loss_num": 0.166015625,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 248384840,
+      "step": 1437
+    },
+    {
+      "epoch": 0.5507468402910762,
+      "grad_norm": 26.591948475025305,
+      "learning_rate": 5e-06,
+      "loss": 0.1648,
+      "num_input_tokens_seen": 248558136,
+      "step": 1438
+    },
+    {
+      "epoch": 0.5507468402910762,
+      "loss": 0.1620870679616928,
+      "loss_ce": 0.02115688845515251,
+      "loss_iou": 1.0057390928268433,
+      "loss_num": 0.140625,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 248558136,
+      "step": 1438
+    },
+    {
+      "epoch": 0.551129835312141,
+      "grad_norm": 44.742800345656015,
+      "learning_rate": 5e-06,
+      "loss": 0.245,
+      "num_input_tokens_seen": 248730896,
+      "step": 1439
+    },
+    {
+      "epoch": 0.551129835312141,
+      "loss": 0.2673144042491913,
+      "loss_ce": 0.020488237962126732,
+      "loss_iou": 1.0043540000915527,
+      "loss_num": 0.2470703125,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 248730896,
+      "step": 1439
+    },
+    {
+      "epoch": 0.5515128303332056,
+      "grad_norm": 37.738405534998805,
+      "learning_rate": 5e-06,
+      "loss": 0.2234,
+      "num_input_tokens_seen": 248900032,
+      "step": 1440
+    },
+    {
+      "epoch": 0.5515128303332056,
+      "loss": 0.2288774847984314,
+      "loss_ce": 0.0203813835978508,
+      "loss_iou": 1.0389633178710938,
+      "loss_num": 0.208984375,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 248900032,
+      "step": 1440
+    },
+    {
+      "epoch": 0.5518958253542704,
+      "grad_norm": 28.895011673028538,
+      "learning_rate": 5e-06,
+      "loss": 0.253,
+      "num_input_tokens_seen": 249073008,
+      "step": 1441
+    },
+    {
+      "epoch": 0.5518958253542704,
+      "loss": 0.18834465742111206,
+      "loss_ce": 0.019765563309192657,
+      "loss_iou": 0.974697470664978,
+      "loss_num": 0.1689453125,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 249073008,
+      "step": 1441
+    },
+    {
+      "epoch": 0.5522788203753352,
+      "grad_norm": 18.81039078862195,
+      "learning_rate": 5e-06,
+      "loss": 0.244,
+      "num_input_tokens_seen": 249246200,
+      "step": 1442
+    },
+    {
+      "epoch": 0.5522788203753352,
+      "loss": 0.2613523006439209,
+      "loss_ce": 0.018371347337961197,
+      "loss_iou": 1.048504114151001,
+      "loss_num": 0.2431640625,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 249246200,
+      "step": 1442
+    },
+    {
+      "epoch": 0.5526618153963998,
+      "grad_norm": 30.81313041119159,
+      "learning_rate": 5e-06,
+      "loss": 0.2541,
+      "num_input_tokens_seen": 249418984,
+      "step": 1443
+    },
+    {
+      "epoch": 0.5526618153963998,
+      "loss": 0.3366760313510895,
+      "loss_ce": 0.020269788801670074,
+      "loss_iou": 1.0565412044525146,
+      "loss_num": 0.31640625,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 249418984,
+      "step": 1443
+    },
+    {
+      "epoch": 0.5530448104174646,
+      "grad_norm": 20.92797494782328,
+      "learning_rate": 5e-06,
+      "loss": 0.2445,
+      "num_input_tokens_seen": 249591752,
+      "step": 1444
+    },
+    {
+      "epoch": 0.5530448104174646,
+      "loss": 0.25018686056137085,
+      "loss_ce": 0.020450517535209656,
+      "loss_iou": 1.016252040863037,
+      "loss_num": 0.2294921875,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 249591752,
+      "step": 1444
+    },
+    {
+      "epoch": 0.5534278054385293,
+      "grad_norm": 33.838335380949694,
+      "learning_rate": 5e-06,
+      "loss": 0.2051,
+      "num_input_tokens_seen": 249764864,
+      "step": 1445
+    },
+    {
+      "epoch": 0.5534278054385293,
+      "loss": 0.22452612221240997,
+      "loss_ce": 0.022499756887555122,
+      "loss_iou": 1.0481488704681396,
+      "loss_num": 0.2021484375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 249764864,
+      "step": 1445
+    },
+    {
+      "epoch": 0.5538108004595941,
+      "grad_norm": 28.991763308037577,
+      "learning_rate": 5e-06,
+      "loss": 0.2625,
+      "num_input_tokens_seen": 249937608,
+      "step": 1446
+    },
+    {
+      "epoch": 0.5538108004595941,
+      "loss": 0.2906506061553955,
+      "loss_ce": 0.020142812281847,
+      "loss_iou": 1.282198429107666,
+      "loss_num": 0.26953125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 249937608,
+      "step": 1446
+    },
+    {
+      "epoch": 0.5541937954806587,
+      "grad_norm": 29.089797794294615,
+      "learning_rate": 5e-06,
+      "loss": 0.26,
+      "num_input_tokens_seen": 250110232,
+      "step": 1447
+    },
+    {
+      "epoch": 0.5541937954806587,
+      "loss": 0.3229195773601532,
+      "loss_ce": 0.020551415160298347,
+      "loss_iou": 1.019213080406189,
+      "loss_num": 0.302734375,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 250110232,
+      "step": 1447
+    },
+    {
+      "epoch": 0.5545767905017235,
+      "grad_norm": 26.8909519813235,
+      "learning_rate": 5e-06,
+      "loss": 0.1862,
+      "num_input_tokens_seen": 250283600,
+      "step": 1448
+    },
+    {
+      "epoch": 0.5545767905017235,
+      "loss": 0.16345219314098358,
+      "loss_ce": 0.02044682390987873,
+      "loss_iou": 1.0251049995422363,
+      "loss_num": 0.142578125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 250283600,
+      "step": 1448
+    },
+    {
+      "epoch": 0.5549597855227882,
+      "grad_norm": 30.580572657257935,
+      "learning_rate": 5e-06,
+      "loss": 0.2356,
+      "num_input_tokens_seen": 250455976,
+      "step": 1449
+    },
+    {
+      "epoch": 0.5549597855227882,
+      "loss": 0.26306387782096863,
+      "loss_ce": 0.020815353840589523,
+      "loss_iou": 1.0646202564239502,
+      "loss_num": 0.2421875,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 250455976,
+      "step": 1449
+    },
+    {
+      "epoch": 0.5553427805438529,
+      "grad_norm": 27.359346762967068,
+      "learning_rate": 5e-06,
+      "loss": 0.2284,
+      "num_input_tokens_seen": 250628712,
+      "step": 1450
+    },
+    {
+      "epoch": 0.5553427805438529,
+      "loss": 0.2822895050048828,
+      "loss_ce": 0.019166937097907066,
+      "loss_iou": 1.0252337455749512,
+      "loss_num": 0.263671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 250628712,
+      "step": 1450
+    },
+    {
+      "epoch": 0.5557257755649176,
+      "grad_norm": 24.65924185646727,
+      "learning_rate": 5e-06,
+      "loss": 0.2387,
+      "num_input_tokens_seen": 250801504,
+      "step": 1451
+    },
+    {
+      "epoch": 0.5557257755649176,
+      "loss": 0.28496530652046204,
+      "loss_ce": 0.020194800570607185,
+      "loss_iou": 1.07538640499115,
+      "loss_num": 0.265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 250801504,
+      "step": 1451
+    },
+    {
+      "epoch": 0.5561087705859824,
+      "grad_norm": 16.90062559129393,
+      "learning_rate": 5e-06,
+      "loss": 0.1497,
+      "num_input_tokens_seen": 250974672,
+      "step": 1452
+    },
+    {
+      "epoch": 0.5561087705859824,
+      "loss": 0.13314805924892426,
+      "loss_ce": 0.021819941699504852,
+      "loss_iou": 1.0009329319000244,
+      "loss_num": 0.111328125,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 250974672,
+      "step": 1452
+    },
+    {
+      "epoch": 0.5564917656070472,
+      "grad_norm": 27.94039267193011,
+      "learning_rate": 5e-06,
+      "loss": 0.2496,
+      "num_input_tokens_seen": 251148072,
+      "step": 1453
+    },
+    {
+      "epoch": 0.5564917656070472,
+      "loss": 0.2612951397895813,
+      "loss_ce": 0.019901107996702194,
+      "loss_iou": 1.042182445526123,
+      "loss_num": 0.2412109375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 251148072,
+      "step": 1453
+    },
+    {
+      "epoch": 0.5568747606281118,
+      "grad_norm": 36.678368035743375,
+      "learning_rate": 5e-06,
+      "loss": 0.243,
+      "num_input_tokens_seen": 251321304,
+      "step": 1454
+    },
+    {
+      "epoch": 0.5568747606281118,
+      "loss": 0.2739933729171753,
+      "loss_ce": 0.02124677784740925,
+      "loss_iou": 1.0768338441848755,
+      "loss_num": 0.251953125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 251321304,
+      "step": 1454
+    },
+    {
+      "epoch": 0.5572577556491766,
+      "grad_norm": 24.70238774410247,
+      "learning_rate": 5e-06,
+      "loss": 0.2202,
+      "num_input_tokens_seen": 251494048,
+      "step": 1455
+    },
+    {
+      "epoch": 0.5572577556491766,
+      "loss": 0.23047927021980286,
+      "loss_ce": 0.017466576769948006,
+      "loss_iou": 1.013255000114441,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 251494048,
+      "step": 1455
+    },
+    {
+      "epoch": 0.5576407506702413,
+      "grad_norm": 23.99901276583434,
+      "learning_rate": 5e-06,
+      "loss": 0.1982,
+      "num_input_tokens_seen": 251667488,
+      "step": 1456
+    },
+    {
+      "epoch": 0.5576407506702413,
+      "loss": 0.22782182693481445,
+      "loss_ce": 0.020729543641209602,
+      "loss_iou": 1.0163581371307373,
+      "loss_num": 0.20703125,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 251667488,
+      "step": 1456
+    },
+    {
+      "epoch": 0.558023745691306,
+      "grad_norm": 20.10364549368622,
+      "learning_rate": 5e-06,
+      "loss": 0.2347,
+      "num_input_tokens_seen": 251840240,
+      "step": 1457
+    },
+    {
+      "epoch": 0.558023745691306,
+      "loss": 0.3191094994544983,
+      "loss_ce": 0.0212579183280468,
+      "loss_iou": 0.8945208787918091,
+      "loss_num": 0.296875,
+      "loss_xval": 0.296875,
+      "num_input_tokens_seen": 251840240,
+      "step": 1457
+    },
+    {
+      "epoch": 0.5584067407123707,
+      "grad_norm": 22.94158214370783,
+      "learning_rate": 5e-06,
+      "loss": 0.1973,
+      "num_input_tokens_seen": 252013032,
+      "step": 1458
+    },
+    {
+      "epoch": 0.5584067407123707,
+      "loss": 0.1668839454650879,
+      "loss_ce": 0.019545067101716995,
+      "loss_iou": 1.0783774852752686,
+      "loss_num": 0.1474609375,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 252013032,
+      "step": 1458
+    },
+    {
+      "epoch": 0.5587897357334355,
+      "grad_norm": 27.846996712471338,
+      "learning_rate": 5e-06,
+      "loss": 0.2153,
+      "num_input_tokens_seen": 252186040,
+      "step": 1459
+    },
+    {
+      "epoch": 0.5587897357334355,
+      "loss": 0.23262223601341248,
+      "loss_ce": 0.01912127062678337,
+      "loss_iou": 1.2774409055709839,
+      "loss_num": 0.2138671875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 252186040,
+      "step": 1459
+    },
+    {
+      "epoch": 0.5591727307545002,
+      "grad_norm": 29.953167089807298,
+      "learning_rate": 5e-06,
+      "loss": 0.2447,
+      "num_input_tokens_seen": 252358920,
+      "step": 1460
+    },
+    {
+      "epoch": 0.5591727307545002,
+      "loss": 0.2146613597869873,
+      "loss_ce": 0.0197761133313179,
+      "loss_iou": 1.064612627029419,
+      "loss_num": 0.1953125,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 252358920,
+      "step": 1460
+    },
+    {
+      "epoch": 0.5595557257755649,
+      "grad_norm": 24.440971150018694,
+      "learning_rate": 5e-06,
+      "loss": 0.2266,
+      "num_input_tokens_seen": 252531824,
+      "step": 1461
+    },
+    {
+      "epoch": 0.5595557257755649,
+      "loss": 0.2598302960395813,
+      "loss_ce": 0.021304896101355553,
+      "loss_iou": 1.0376880168914795,
+      "loss_num": 0.23828125,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 252531824,
+      "step": 1461
+    },
+    {
+      "epoch": 0.5599387207966297,
+      "grad_norm": 27.876724687870354,
+      "learning_rate": 5e-06,
+      "loss": 0.2594,
+      "num_input_tokens_seen": 252704720,
+      "step": 1462
+    },
+    {
+      "epoch": 0.5599387207966297,
+      "loss": 0.28872719407081604,
+      "loss_ce": 0.019318006932735443,
+      "loss_iou": 1.0152583122253418,
+      "loss_num": 0.26953125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 252704720,
+      "step": 1462
+    },
+    {
+      "epoch": 0.5603217158176944,
+      "grad_norm": 32.41463048543971,
+      "learning_rate": 5e-06,
+      "loss": 0.2042,
+      "num_input_tokens_seen": 252877360,
+      "step": 1463
+    },
+    {
+      "epoch": 0.5603217158176944,
+      "loss": 0.19869321584701538,
+      "loss_ce": 0.02071470394730568,
+      "loss_iou": 0.9860658049583435,
+      "loss_num": 0.177734375,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 252877360,
+      "step": 1463
+    },
+    {
+      "epoch": 0.5607047108387591,
+      "grad_norm": 31.18462843405718,
+      "learning_rate": 5e-06,
+      "loss": 0.2878,
+      "num_input_tokens_seen": 253050392,
+      "step": 1464
+    },
+    {
+      "epoch": 0.5607047108387591,
+      "loss": 0.29182857275009155,
+      "loss_ce": 0.020771419629454613,
+      "loss_iou": 1.0111753940582275,
+      "loss_num": 0.271484375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 253050392,
+      "step": 1464
+    },
+    {
+      "epoch": 0.5610877058598238,
+      "grad_norm": 23.093980086732667,
+      "learning_rate": 5e-06,
+      "loss": 0.2851,
+      "num_input_tokens_seen": 253223208,
+      "step": 1465
+    },
+    {
+      "epoch": 0.5610877058598238,
+      "loss": 0.29920488595962524,
+      "loss_ce": 0.019663862884044647,
+      "loss_iou": 2.007282257080078,
+      "loss_num": 0.279296875,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 253223208,
+      "step": 1465
+    },
+    {
+      "epoch": 0.5614707008808886,
+      "grad_norm": 21.495583545383287,
+      "learning_rate": 5e-06,
+      "loss": 0.2176,
+      "num_input_tokens_seen": 253396032,
+      "step": 1466
+    },
+    {
+      "epoch": 0.5614707008808886,
+      "loss": 0.16312876343727112,
+      "loss_ce": 0.019818205386400223,
+      "loss_iou": 1.0369665622711182,
+      "loss_num": 0.1435546875,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 253396032,
+      "step": 1466
+    },
+    {
+      "epoch": 0.5618536959019532,
+      "grad_norm": 32.0839754181178,
+      "learning_rate": 5e-06,
+      "loss": 0.223,
+      "num_input_tokens_seen": 253568952,
+      "step": 1467
+    },
+    {
+      "epoch": 0.5618536959019532,
+      "loss": 0.24454674124717712,
+      "loss_ce": 0.020608752965927124,
+      "loss_iou": 1.0019643306732178,
+      "loss_num": 0.2236328125,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 253568952,
+      "step": 1467
+    },
+    {
+      "epoch": 0.562236690923018,
+      "grad_norm": 30.212997661988872,
+      "learning_rate": 5e-06,
+      "loss": 0.1683,
+      "num_input_tokens_seen": 253741792,
+      "step": 1468
+    },
+    {
+      "epoch": 0.562236690923018,
+      "loss": 0.17628955841064453,
+      "loss_ce": 0.02058887481689453,
+      "loss_iou": 1.0021042823791504,
+      "loss_num": 0.1552734375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 253741792,
+      "step": 1468
+    },
+    {
+      "epoch": 0.5626196859440827,
+      "grad_norm": 37.497116361590386,
+      "learning_rate": 5e-06,
+      "loss": 0.2354,
+      "num_input_tokens_seen": 253914840,
+      "step": 1469
+    },
+    {
+      "epoch": 0.5626196859440827,
+      "loss": 0.24797192215919495,
+      "loss_ce": 0.020432859659194946,
+      "loss_iou": 0.9836539626121521,
+      "loss_num": 0.2275390625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 253914840,
+      "step": 1469
+    },
+    {
+      "epoch": 0.5630026809651475,
+      "grad_norm": 47.64589918729428,
+      "learning_rate": 5e-06,
+      "loss": 0.2739,
+      "num_input_tokens_seen": 254088168,
+      "step": 1470
+    },
+    {
+      "epoch": 0.5630026809651475,
+      "loss": 0.22957342863082886,
+      "loss_ce": 0.021443545818328857,
+      "loss_iou": 1.033742904663086,
+      "loss_num": 0.2080078125,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 254088168,
+      "step": 1470
+    },
+    {
+      "epoch": 0.5633856759862121,
+      "grad_norm": 52.447713593223874,
+      "learning_rate": 5e-06,
+      "loss": 0.2729,
+      "num_input_tokens_seen": 254260984,
+      "step": 1471
+    },
+    {
+      "epoch": 0.5633856759862121,
+      "loss": 0.3072156012058258,
+      "loss_ce": 0.020716574043035507,
+      "loss_iou": 1.0662565231323242,
+      "loss_num": 0.287109375,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 254260984,
+      "step": 1471
+    },
+    {
+      "epoch": 0.5637686710072769,
+      "grad_norm": 27.162718678453754,
+      "learning_rate": 5e-06,
+      "loss": 0.208,
+      "num_input_tokens_seen": 254434440,
+      "step": 1472
+    },
+    {
+      "epoch": 0.5637686710072769,
+      "loss": 0.21323244273662567,
+      "loss_ce": 0.02030031383037567,
+      "loss_iou": 1.198883295059204,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 254434440,
+      "step": 1472
+    },
+    {
+      "epoch": 0.5641516660283417,
+      "grad_norm": 30.729533469850466,
+      "learning_rate": 5e-06,
+      "loss": 0.2193,
+      "num_input_tokens_seen": 254607128,
+      "step": 1473
+    },
+    {
+      "epoch": 0.5641516660283417,
+      "loss": 0.2197684943675995,
+      "loss_ce": 0.021160103380680084,
+      "loss_iou": 1.0076149702072144,
+      "loss_num": 0.1982421875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 254607128,
+      "step": 1473
+    },
+    {
+      "epoch": 0.5645346610494063,
+      "grad_norm": 41.532002399408285,
+      "learning_rate": 5e-06,
+      "loss": 0.2282,
+      "num_input_tokens_seen": 254780352,
+      "step": 1474
+    },
+    {
+      "epoch": 0.5645346610494063,
+      "loss": 0.21799220144748688,
+      "loss_ce": 0.020238302648067474,
+      "loss_iou": 1.0031582117080688,
+      "loss_num": 0.197265625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 254780352,
+      "step": 1474
+    },
+    {
+      "epoch": 0.5649176560704711,
+      "grad_norm": 40.10343286688188,
+      "learning_rate": 5e-06,
+      "loss": 0.2617,
+      "num_input_tokens_seen": 254953736,
+      "step": 1475
+    },
+    {
+      "epoch": 0.5649176560704711,
+      "loss": 0.30094754695892334,
+      "loss_ce": 0.019758574664592743,
+      "loss_iou": 1.1032297611236572,
+      "loss_num": 0.28125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 254953736,
+      "step": 1475
+    },
+    {
+      "epoch": 0.5653006510915358,
+      "grad_norm": 29.5335538702408,
+      "learning_rate": 5e-06,
+      "loss": 0.242,
+      "num_input_tokens_seen": 255126792,
+      "step": 1476
+    },
+    {
+      "epoch": 0.5653006510915358,
+      "loss": 0.23481246829032898,
+      "loss_ce": 0.018015600740909576,
+      "loss_iou": -0.00921630859375,
+      "loss_num": 0.216796875,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 255126792,
+      "step": 1476
+    },
+    {
+      "epoch": 0.5656836461126006,
+      "grad_norm": 30.262688256639134,
+      "learning_rate": 5e-06,
+      "loss": 0.2554,
+      "num_input_tokens_seen": 255299688,
+      "step": 1477
+    },
+    {
+      "epoch": 0.5656836461126006,
+      "loss": 0.2591654360294342,
+      "loss_ce": 0.01978554204106331,
+      "loss_iou": 1.044213056564331,
+      "loss_num": 0.2392578125,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 255299688,
+      "step": 1477
+    },
+    {
+      "epoch": 0.5660666411336652,
+      "grad_norm": 31.018493474825984,
+      "learning_rate": 5e-06,
+      "loss": 0.2725,
+      "num_input_tokens_seen": 255472976,
+      "step": 1478
+    },
+    {
+      "epoch": 0.5660666411336652,
+      "loss": 0.29889142513275146,
+      "loss_ce": 0.01965557411313057,
+      "loss_iou": 1.062753677368164,
+      "loss_num": 0.279296875,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 255472976,
+      "step": 1478
+    },
+    {
+      "epoch": 0.56644963615473,
+      "grad_norm": 51.46796253515065,
+      "learning_rate": 5e-06,
+      "loss": 0.2663,
+      "num_input_tokens_seen": 255645936,
+      "step": 1479
+    },
+    {
+      "epoch": 0.56644963615473,
+      "loss": 0.24959470331668854,
+      "loss_ce": 0.02156735584139824,
+      "loss_iou": 1.0347603559494019,
+      "loss_num": 0.228515625,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 255645936,
+      "step": 1479
+    },
+    {
+      "epoch": 0.5668326311757947,
+      "grad_norm": 30.522244174594228,
+      "learning_rate": 5e-06,
+      "loss": 0.1875,
+      "num_input_tokens_seen": 255818808,
+      "step": 1480
+    },
+    {
+      "epoch": 0.5668326311757947,
+      "loss": 0.2001829445362091,
+      "loss_ce": 0.020800607278943062,
+      "loss_iou": 1.0021522045135498,
+      "loss_num": 0.1796875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 255818808,
+      "step": 1480
+    },
+    {
+      "epoch": 0.5672156261968594,
+      "grad_norm": 30.645199803497867,
+      "learning_rate": 5e-06,
+      "loss": 0.2606,
+      "num_input_tokens_seen": 255991800,
+      "step": 1481
+    },
+    {
+      "epoch": 0.5672156261968594,
+      "loss": 0.2134249210357666,
+      "loss_ce": 0.018966909497976303,
+      "loss_iou": 0.8776056170463562,
+      "loss_num": 0.1943359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 255991800,
+      "step": 1481
+    },
+    {
+      "epoch": 0.5675986212179241,
+      "grad_norm": 30.114094340279998,
+      "learning_rate": 5e-06,
+      "loss": 0.2152,
+      "num_input_tokens_seen": 256164736,
+      "step": 1482
+    },
+    {
+      "epoch": 0.5675986212179241,
+      "loss": 0.24286949634552002,
+      "loss_ce": 0.021311860531568527,
+      "loss_iou": 1.0650312900543213,
+      "loss_num": 0.2216796875,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 256164736,
+      "step": 1482
+    },
+    {
+      "epoch": 0.5679816162389889,
+      "grad_norm": 60.050880997864404,
+      "learning_rate": 5e-06,
+      "loss": 0.2796,
+      "num_input_tokens_seen": 256337552,
+      "step": 1483
+    },
+    {
+      "epoch": 0.5679816162389889,
+      "loss": 0.23949606716632843,
+      "loss_ce": 0.02001364901661873,
+      "loss_iou": 1.0076258182525635,
+      "loss_num": 0.2197265625,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 256337552,
+      "step": 1483
+    },
+    {
+      "epoch": 0.5683646112600537,
+      "grad_norm": 41.05984639796678,
+      "learning_rate": 5e-06,
+      "loss": 0.286,
+      "num_input_tokens_seen": 256510528,
+      "step": 1484
+    },
+    {
+      "epoch": 0.5683646112600537,
+      "loss": 0.3509371280670166,
+      "loss_ce": 0.019516244530677795,
+      "loss_iou": 1.0642974376678467,
+      "loss_num": 0.33203125,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 256510528,
+      "step": 1484
+    },
+    {
+      "epoch": 0.5687476062811183,
+      "grad_norm": 46.347360230691486,
+      "learning_rate": 5e-06,
+      "loss": 0.371,
+      "num_input_tokens_seen": 256683304,
+      "step": 1485
+    },
+    {
+      "epoch": 0.5687476062811183,
+      "loss": 0.3466867208480835,
+      "loss_ce": 0.019294165074825287,
+      "loss_iou": 1.0048203468322754,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 256683304,
+      "step": 1485
+    },
+    {
+      "epoch": 0.5691306013021831,
+      "grad_norm": 23.912583644311166,
+      "learning_rate": 5e-06,
+      "loss": 0.3312,
+      "num_input_tokens_seen": 256856448,
+      "step": 1486
+    },
+    {
+      "epoch": 0.5691306013021831,
+      "loss": 0.38610681891441345,
+      "loss_ce": 0.02038414590060711,
+      "loss_iou": 1.1153357028961182,
+      "loss_num": 0.365234375,
+      "loss_xval": 0.365234375,
+      "num_input_tokens_seen": 256856448,
+      "step": 1486
+    },
+    {
+      "epoch": 0.5695135963232478,
+      "grad_norm": 67.72882364069386,
+      "learning_rate": 5e-06,
+      "loss": 0.3698,
+      "num_input_tokens_seen": 257029184,
+      "step": 1487
+    },
+    {
+      "epoch": 0.5695135963232478,
+      "loss": 0.35023269057273865,
+      "loss_ce": 0.021497325971722603,
+      "loss_iou": 1.0250391960144043,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 257029184,
+      "step": 1487
+    },
+    {
+      "epoch": 0.5698965913443125,
+      "grad_norm": 48.08334423675242,
+      "learning_rate": 5e-06,
+      "loss": 0.4014,
+      "num_input_tokens_seen": 257202240,
+      "step": 1488
+    },
+    {
+      "epoch": 0.5698965913443125,
+      "loss": 0.3962937593460083,
+      "loss_ce": 0.020561344921588898,
+      "loss_iou": 1.0564351081848145,
+      "loss_num": 0.375,
+      "loss_xval": 0.375,
+      "num_input_tokens_seen": 257202240,
+      "step": 1488
+    },
+    {
+      "epoch": 0.5702795863653772,
+      "grad_norm": 79.31275399258222,
+      "learning_rate": 5e-06,
+      "loss": 0.3377,
+      "num_input_tokens_seen": 257375232,
+      "step": 1489
+    },
+    {
+      "epoch": 0.5702795863653772,
+      "loss": 0.3447765111923218,
+      "loss_ce": 0.021656420081853867,
+      "loss_iou": 1.2226455211639404,
+      "loss_num": 0.322265625,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 257375232,
+      "step": 1489
+    },
+    {
+      "epoch": 0.570662581386442,
+      "grad_norm": 30.31768962775951,
+      "learning_rate": 5e-06,
+      "loss": 0.2937,
+      "num_input_tokens_seen": 257548264,
+      "step": 1490
+    },
+    {
+      "epoch": 0.570662581386442,
+      "loss": 0.2713618278503418,
+      "loss_ce": 0.020995626226067543,
+      "loss_iou": 1.0252916812896729,
+      "loss_num": 0.25,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 257548264,
+      "step": 1490
+    },
+    {
+      "epoch": 0.5710455764075067,
+      "grad_norm": 55.12673938107984,
+      "learning_rate": 5e-06,
+      "loss": 0.2587,
+      "num_input_tokens_seen": 257721768,
+      "step": 1491
+    },
+    {
+      "epoch": 0.5710455764075067,
+      "loss": 0.2455759048461914,
+      "loss_ce": 0.022675510495901108,
+      "loss_iou": 1.016119360923767,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 257721768,
+      "step": 1491
+    },
+    {
+      "epoch": 0.5714285714285714,
+      "grad_norm": 41.4510872847938,
+      "learning_rate": 5e-06,
+      "loss": 0.2507,
+      "num_input_tokens_seen": 257894736,
+      "step": 1492
+    },
+    {
+      "epoch": 0.5714285714285714,
+      "loss": 0.2653059959411621,
+      "loss_ce": 0.02165365219116211,
+      "loss_iou": 1.0070408582687378,
+      "loss_num": 0.244140625,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 257894736,
+      "step": 1492
+    },
+    {
+      "epoch": 0.5718115664496362,
+      "grad_norm": 30.448018451643417,
+      "learning_rate": 5e-06,
+      "loss": 0.2485,
+      "num_input_tokens_seen": 258067632,
+      "step": 1493
+    },
+    {
+      "epoch": 0.5718115664496362,
+      "loss": 0.23420105874538422,
+      "loss_ce": 0.02155456691980362,
+      "loss_iou": 1.0096468925476074,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 258067632,
+      "step": 1493
+    },
+    {
+      "epoch": 0.5721945614707009,
+      "grad_norm": 22.206726089495916,
+      "learning_rate": 5e-06,
+      "loss": 0.244,
+      "num_input_tokens_seen": 258240832,
+      "step": 1494
+    },
+    {
+      "epoch": 0.5721945614707009,
+      "loss": 0.23333390057086945,
+      "loss_ce": 0.02026016265153885,
+      "loss_iou": 1.0437605381011963,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 258240832,
+      "step": 1494
+    },
+    {
+      "epoch": 0.5725775564917656,
+      "grad_norm": 33.7110286932408,
+      "learning_rate": 5e-06,
+      "loss": 0.1897,
+      "num_input_tokens_seen": 258413480,
+      "step": 1495
+    },
+    {
+      "epoch": 0.5725775564917656,
+      "loss": 0.1620618999004364,
+      "loss_ce": 0.021009644493460655,
+      "loss_iou": 1.0012726783752441,
+      "loss_num": 0.140625,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 258413480,
+      "step": 1495
+    },
+    {
+      "epoch": 0.5729605515128303,
+      "grad_norm": 40.76843772553818,
+      "learning_rate": 5e-06,
+      "loss": 0.2507,
+      "num_input_tokens_seen": 258586424,
+      "step": 1496
+    },
+    {
+      "epoch": 0.5729605515128303,
+      "loss": 0.28173011541366577,
+      "loss_ce": 0.020560679957270622,
+      "loss_iou": 1.0176188945770264,
+      "loss_num": 0.26171875,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 258586424,
+      "step": 1496
+    },
+    {
+      "epoch": 0.5733435465338951,
+      "grad_norm": 43.95036831704065,
+      "learning_rate": 5e-06,
+      "loss": 0.2553,
+      "num_input_tokens_seen": 258759152,
+      "step": 1497
+    },
+    {
+      "epoch": 0.5733435465338951,
+      "loss": 0.26648202538490295,
+      "loss_ce": 0.019899986684322357,
+      "loss_iou": 1.0377943515777588,
+      "loss_num": 0.24609375,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 258759152,
+      "step": 1497
+    },
+    {
+      "epoch": 0.5737265415549598,
+      "grad_norm": 33.315325278592105,
+      "learning_rate": 5e-06,
+      "loss": 0.2254,
+      "num_input_tokens_seen": 258932120,
+      "step": 1498
+    },
+    {
+      "epoch": 0.5737265415549598,
+      "loss": 0.21317031979560852,
+      "loss_ce": 0.02219131588935852,
+      "loss_iou": 1.0846459865570068,
+      "loss_num": 0.19140625,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 258932120,
+      "step": 1498
+    },
+    {
+      "epoch": 0.5741095365760245,
+      "grad_norm": 30.609871109394707,
+      "learning_rate": 5e-06,
+      "loss": 0.3001,
+      "num_input_tokens_seen": 259105056,
+      "step": 1499
+    },
+    {
+      "epoch": 0.5741095365760245,
+      "loss": 0.23890420794487,
+      "loss_ce": 0.021008696407079697,
+      "loss_iou": 1.0146489143371582,
+      "loss_num": 0.2177734375,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 259105056,
+      "step": 1499
+    },
+    {
+      "epoch": 0.5744925315970892,
+      "grad_norm": 34.31191492677639,
+      "learning_rate": 5e-06,
+      "loss": 0.2872,
+      "num_input_tokens_seen": 259277880,
+      "step": 1500
+    },
+    {
+      "epoch": 0.5744925315970892,
+      "eval_websight_new_CIoU": 0.8998479247093201,
+      "eval_websight_new_GIoU": 0.8992944955825806,
+      "eval_websight_new_IoU": 0.9003492593765259,
+      "eval_websight_new_MAE_all": 0.009700013790279627,
+      "eval_websight_new_MAE_h": 0.008959516882896423,
+      "eval_websight_new_MAE_w": 0.007738122018054128,
+      "eval_websight_new_MAE_x": 0.009648083243519068,
+      "eval_websight_new_MAE_y": 0.012454329989850521,
+      "eval_websight_new_NUM_probability": 7.878957694629207e-05,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.15105408430099487,
+      "eval_websight_new_loss_ce": 0.04203689843416214,
+      "eval_websight_new_loss_iou": 1.000507116317749,
+      "eval_websight_new_loss_num": 0.107513427734375,
+      "eval_websight_new_loss_xval": 0.107513427734375,
+      "eval_websight_new_runtime": 55.5311,
+      "eval_websight_new_samples_per_second": 0.9,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 259277880,
+      "step": 1500
+    },
+    {
+      "epoch": 0.5744925315970892,
+      "eval_seeclick_CIoU": 0.6475102007389069,
+      "eval_seeclick_GIoU": 0.6547337472438812,
+      "eval_seeclick_IoU": 0.6767177283763885,
+      "eval_seeclick_MAE_all": 0.05796422250568867,
+      "eval_seeclick_MAE_h": 0.04909752868115902,
+      "eval_seeclick_MAE_w": 0.07029224187135696,
+      "eval_seeclick_MAE_x": 0.058849262073636055,
+      "eval_seeclick_MAE_y": 0.05361786112189293,
+      "eval_seeclick_NUM_probability": 0.00010826160723809153,
+      "eval_seeclick_inside_bbox": 0.8975694477558136,
+      "eval_seeclick_loss": 0.4864910840988159,
+      "eval_seeclick_loss_ce": 0.032762957736849785,
+      "eval_seeclick_loss_iou": 1.1455382108688354,
+      "eval_seeclick_loss_num": 0.43011474609375,
+      "eval_seeclick_loss_xval": 0.43011474609375,
+      "eval_seeclick_runtime": 84.9697,
+      "eval_seeclick_samples_per_second": 0.588,
+      "eval_seeclick_steps_per_second": 0.024,
+      "num_input_tokens_seen": 259277880,
+      "step": 1500
+    },
+    {
+      "epoch": 0.5744925315970892,
+      "eval_icons_CIoU": 0.8272820115089417,
+      "eval_icons_GIoU": 0.8224707245826721,
+      "eval_icons_IoU": 0.8362298011779785,
+      "eval_icons_MAE_all": 0.02518743323162198,
+      "eval_icons_MAE_h": 0.018266789615154266,
+      "eval_icons_MAE_w": 0.03073503915220499,
+      "eval_icons_MAE_x": 0.030140070244669914,
+      "eval_icons_MAE_y": 0.021607825998216867,
+      "eval_icons_NUM_probability": 7.407073280774057e-05,
+      "eval_icons_inside_bbox": 0.9722222089767456,
+      "eval_icons_loss": 0.21765857934951782,
+      "eval_icons_loss_ce": 0.028413159772753716,
+      "eval_icons_loss_iou": 1.04171884059906,
+      "eval_icons_loss_num": 0.178863525390625,
+      "eval_icons_loss_xval": 0.178863525390625,
+      "eval_icons_runtime": 84.1218,
+      "eval_icons_samples_per_second": 0.594,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 259277880,
+      "step": 1500
+    },
+    {
+      "epoch": 0.5744925315970892,
+      "loss": 0.24968013167381287,
+      "loss_ce": 0.028427692130208015,
+      "loss_iou": 1.0830289125442505,
+      "loss_num": 0.2216796875,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 259277880,
+      "step": 1500
+    },
+    {
+      "epoch": 0.574875526618154,
+      "grad_norm": 55.80232622645771,
+      "learning_rate": 5e-06,
+      "loss": 0.3,
+      "num_input_tokens_seen": 259450864,
+      "step": 1501
+    },
+    {
+      "epoch": 0.574875526618154,
+      "loss": 0.2990880012512207,
+      "loss_ce": 0.020279401913285255,
+      "loss_iou": 1.3054325580596924,
+      "loss_num": 0.279296875,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 259450864,
+      "step": 1501
+    },
+    {
+      "epoch": 0.5752585216392186,
+      "grad_norm": 24.290580399677125,
+      "learning_rate": 5e-06,
+      "loss": 0.2629,
+      "num_input_tokens_seen": 259623904,
+      "step": 1502
+    },
+    {
+      "epoch": 0.5752585216392186,
+      "loss": 0.20929402112960815,
+      "loss_ce": 0.01947469264268875,
+      "loss_iou": 1.027704119682312,
+      "loss_num": 0.189453125,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 259623904,
+      "step": 1502
+    },
+    {
+      "epoch": 0.5756415166602834,
+      "grad_norm": 30.846707526882263,
+      "learning_rate": 5e-06,
+      "loss": 0.1998,
+      "num_input_tokens_seen": 259797160,
+      "step": 1503
+    },
+    {
+      "epoch": 0.5756415166602834,
+      "loss": 0.22802825272083282,
+      "loss_ce": 0.021912530064582825,
+      "loss_iou": 1.0090848207473755,
+      "loss_num": 0.2060546875,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 259797160,
+      "step": 1503
+    },
+    {
+      "epoch": 0.5760245116813482,
+      "grad_norm": 34.683843547461564,
+      "learning_rate": 5e-06,
+      "loss": 0.3247,
+      "num_input_tokens_seen": 259970080,
+      "step": 1504
+    },
+    {
+      "epoch": 0.5760245116813482,
+      "loss": 0.31092390418052673,
+      "loss_ce": 0.019786197692155838,
+      "loss_iou": 1.0614619255065918,
+      "loss_num": 0.291015625,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 259970080,
+      "step": 1504
+    },
+    {
+      "epoch": 0.5764075067024129,
+      "grad_norm": 26.00262513867948,
+      "learning_rate": 5e-06,
+      "loss": 0.2268,
+      "num_input_tokens_seen": 260142872,
+      "step": 1505
+    },
+    {
+      "epoch": 0.5764075067024129,
+      "loss": 0.2198001891374588,
+      "loss_ce": 0.021130749955773354,
+      "loss_iou": 2.49227237701416,
+      "loss_num": 0.1982421875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 260142872,
+      "step": 1505
+    },
+    {
+      "epoch": 0.5767905017234776,
+      "grad_norm": 25.905227039404274,
+      "learning_rate": 5e-06,
+      "loss": 0.3218,
+      "num_input_tokens_seen": 260315896,
+      "step": 1506
+    },
+    {
+      "epoch": 0.5767905017234776,
+      "loss": 0.25244709849357605,
+      "loss_ce": 0.0195369403809309,
+      "loss_iou": 1.0567975044250488,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 260315896,
+      "step": 1506
+    },
+    {
+      "epoch": 0.5771734967445423,
+      "grad_norm": 32.73442958754808,
+      "learning_rate": 5e-06,
+      "loss": 0.2985,
+      "num_input_tokens_seen": 260485488,
+      "step": 1507
+    },
+    {
+      "epoch": 0.5771734967445423,
+      "loss": 0.2852766215801239,
+      "loss_ce": 0.021299580112099648,
+      "loss_iou": 1.0657081604003906,
+      "loss_num": 0.263671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 260485488,
+      "step": 1507
+    },
+    {
+      "epoch": 0.5775564917656071,
+      "grad_norm": 27.04774616807621,
+      "learning_rate": 5e-06,
+      "loss": 0.2106,
+      "num_input_tokens_seen": 260658664,
+      "step": 1508
+    },
+    {
+      "epoch": 0.5775564917656071,
+      "loss": 0.18731904029846191,
+      "loss_ce": 0.02032686024904251,
+      "loss_iou": 1.0016627311706543,
+      "loss_num": 0.1669921875,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 260658664,
+      "step": 1508
+    },
+    {
+      "epoch": 0.5779394867866717,
+      "grad_norm": 23.776641591909762,
+      "learning_rate": 5e-06,
+      "loss": 0.2227,
+      "num_input_tokens_seen": 260831824,
+      "step": 1509
+    },
+    {
+      "epoch": 0.5779394867866717,
+      "loss": 0.22671952843666077,
+      "loss_ce": 0.01981034129858017,
+      "loss_iou": 0.8762431144714355,
+      "loss_num": 0.20703125,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 260831824,
+      "step": 1509
+    },
+    {
+      "epoch": 0.5783224818077365,
+      "grad_norm": 23.614096820361407,
+      "learning_rate": 5e-06,
+      "loss": 0.1847,
+      "num_input_tokens_seen": 261004752,
+      "step": 1510
+    },
+    {
+      "epoch": 0.5783224818077365,
+      "loss": 0.19045805931091309,
+      "loss_ce": 0.020597223192453384,
+      "loss_iou": 1.0042232275009155,
+      "loss_num": 0.169921875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 261004752,
+      "step": 1510
+    },
+    {
+      "epoch": 0.5787054768288012,
+      "grad_norm": 28.50545548830771,
+      "learning_rate": 5e-06,
+      "loss": 0.2731,
+      "num_input_tokens_seen": 261177704,
+      "step": 1511
+    },
+    {
+      "epoch": 0.5787054768288012,
+      "loss": 0.31981053948402405,
+      "loss_ce": 0.021592766046524048,
+      "loss_iou": 1.026820421218872,
+      "loss_num": 0.298828125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 261177704,
+      "step": 1511
+    },
+    {
+      "epoch": 0.579088471849866,
+      "grad_norm": 26.1929496348344,
+      "learning_rate": 5e-06,
+      "loss": 0.186,
+      "num_input_tokens_seen": 261351232,
+      "step": 1512
+    },
+    {
+      "epoch": 0.579088471849866,
+      "loss": 0.20613685250282288,
+      "loss_ce": 0.02217688038945198,
+      "loss_iou": 1.0080938339233398,
+      "loss_num": 0.18359375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 261351232,
+      "step": 1512
+    },
+    {
+      "epoch": 0.5794714668709307,
+      "grad_norm": 36.631132215354285,
+      "learning_rate": 5e-06,
+      "loss": 0.2809,
+      "num_input_tokens_seen": 261524224,
+      "step": 1513
+    },
+    {
+      "epoch": 0.5794714668709307,
+      "loss": 0.246841698884964,
+      "loss_ce": 0.01985194906592369,
+      "loss_iou": 1.3500326871871948,
+      "loss_num": 0.2265625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 261524224,
+      "step": 1513
+    },
+    {
+      "epoch": 0.5798544618919954,
+      "grad_norm": 29.077264193554647,
+      "learning_rate": 5e-06,
+      "loss": 0.2444,
+      "num_input_tokens_seen": 261696800,
+      "step": 1514
+    },
+    {
+      "epoch": 0.5798544618919954,
+      "loss": 0.2662728428840637,
+      "loss_ce": 0.020911507308483124,
+      "loss_iou": 1.1145825386047363,
+      "loss_num": 0.2451171875,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 261696800,
+      "step": 1514
+    },
+    {
+      "epoch": 0.5802374569130602,
+      "grad_norm": 29.768499005814423,
+      "learning_rate": 5e-06,
+      "loss": 0.2606,
+      "num_input_tokens_seen": 261869992,
+      "step": 1515
+    },
+    {
+      "epoch": 0.5802374569130602,
+      "loss": 0.3129635453224182,
+      "loss_ce": 0.019567549228668213,
+      "loss_iou": 1.077012538909912,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 261869992,
+      "step": 1515
+    },
+    {
+      "epoch": 0.5806204519341248,
+      "grad_norm": 30.278667752168946,
+      "learning_rate": 5e-06,
+      "loss": 0.1945,
+      "num_input_tokens_seen": 262042752,
+      "step": 1516
+    },
+    {
+      "epoch": 0.5806204519341248,
+      "loss": 0.17716622352600098,
+      "loss_ce": 0.020122766494750977,
+      "loss_iou": 1.0018476247787476,
+      "loss_num": 0.1572265625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 262042752,
+      "step": 1516
+    },
+    {
+      "epoch": 0.5810034469551896,
+      "grad_norm": 30.94662522269035,
+      "learning_rate": 5e-06,
+      "loss": 0.2048,
+      "num_input_tokens_seen": 262215528,
+      "step": 1517
+    },
+    {
+      "epoch": 0.5810034469551896,
+      "loss": 0.2343868613243103,
+      "loss_ce": 0.021069001406431198,
+      "loss_iou": 1.023953914642334,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 262215528,
+      "step": 1517
+    },
+    {
+      "epoch": 0.5813864419762543,
+      "grad_norm": 29.744591758048816,
+      "learning_rate": 5e-06,
+      "loss": 0.2075,
+      "num_input_tokens_seen": 262388576,
+      "step": 1518
+    },
+    {
+      "epoch": 0.5813864419762543,
+      "loss": 0.23691639304161072,
+      "loss_ce": 0.020119523629546165,
+      "loss_iou": 1.0106154680252075,
+      "loss_num": 0.216796875,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 262388576,
+      "step": 1518
+    },
+    {
+      "epoch": 0.5817694369973191,
+      "grad_norm": 24.549477449191258,
+      "learning_rate": 5e-06,
+      "loss": 0.2543,
+      "num_input_tokens_seen": 262561072,
+      "step": 1519
+    },
+    {
+      "epoch": 0.5817694369973191,
+      "loss": 0.28415757417678833,
+      "loss_ce": 0.01987534761428833,
+      "loss_iou": 1.010164737701416,
+      "loss_num": 0.263671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 262561072,
+      "step": 1519
+    },
+    {
+      "epoch": 0.5821524320183837,
+      "grad_norm": 20.161012045135564,
+      "learning_rate": 5e-06,
+      "loss": 0.2069,
+      "num_input_tokens_seen": 262733624,
+      "step": 1520
+    },
+    {
+      "epoch": 0.5821524320183837,
+      "loss": 0.2429553121328354,
+      "loss_ce": 0.019627690315246582,
+      "loss_iou": 1.100738525390625,
+      "loss_num": 0.2236328125,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 262733624,
+      "step": 1520
+    },
+    {
+      "epoch": 0.5825354270394485,
+      "grad_norm": 25.1038676233111,
+      "learning_rate": 5e-06,
+      "loss": 0.1574,
+      "num_input_tokens_seen": 262906864,
+      "step": 1521
+    },
+    {
+      "epoch": 0.5825354270394485,
+      "loss": 0.15586310625076294,
+      "loss_ce": 0.021036438643932343,
+      "loss_iou": 1.0012270212173462,
+      "loss_num": 0.134765625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 262906864,
+      "step": 1521
+    },
+    {
+      "epoch": 0.5829184220605133,
+      "grad_norm": 23.49886140963644,
+      "learning_rate": 5e-06,
+      "loss": 0.2296,
+      "num_input_tokens_seen": 263079752,
+      "step": 1522
+    },
+    {
+      "epoch": 0.5829184220605133,
+      "loss": 0.26926282048225403,
+      "loss_ce": 0.02030041441321373,
+      "loss_iou": 1.013770341873169,
+      "loss_num": 0.2490234375,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 263079752,
+      "step": 1522
+    },
+    {
+      "epoch": 0.5833014170815779,
+      "grad_norm": 26.60289452188516,
+      "learning_rate": 5e-06,
+      "loss": 0.2097,
+      "num_input_tokens_seen": 263252944,
+      "step": 1523
+    },
+    {
+      "epoch": 0.5833014170815779,
+      "loss": 0.15148669481277466,
+      "loss_ce": 0.022763557732105255,
+      "loss_iou": 1.0016040802001953,
+      "loss_num": 0.12890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 263252944,
+      "step": 1523
+    },
+    {
+      "epoch": 0.5836844121026427,
+      "grad_norm": 34.65592953030426,
+      "learning_rate": 5e-06,
+      "loss": 0.2001,
+      "num_input_tokens_seen": 263425840,
+      "step": 1524
+    },
+    {
+      "epoch": 0.5836844121026427,
+      "loss": 0.20139773190021515,
+      "loss_ce": 0.020306427031755447,
+      "loss_iou": 1.0493004322052002,
+      "loss_num": 0.1806640625,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 263425840,
+      "step": 1524
+    },
+    {
+      "epoch": 0.5840674071237074,
+      "grad_norm": 46.70347242404038,
+      "learning_rate": 5e-06,
+      "loss": 0.2972,
+      "num_input_tokens_seen": 263598696,
+      "step": 1525
+    },
+    {
+      "epoch": 0.5840674071237074,
+      "loss": 0.2707063555717468,
+      "loss_ce": 0.02174396812915802,
+      "loss_iou": 1.0237774848937988,
+      "loss_num": 0.2490234375,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 263598696,
+      "step": 1525
+    },
+    {
+      "epoch": 0.5844504021447721,
+      "grad_norm": 31.70010982744844,
+      "learning_rate": 5e-06,
+      "loss": 0.1844,
+      "num_input_tokens_seen": 263771520,
+      "step": 1526
+    },
+    {
+      "epoch": 0.5844504021447721,
+      "loss": 0.1498769372701645,
+      "loss_ce": 0.021458961069583893,
+      "loss_iou": 1.0089287757873535,
+      "loss_num": 0.12890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 263771520,
+      "step": 1526
+    },
+    {
+      "epoch": 0.5848333971658368,
+      "grad_norm": 30.65245007818194,
+      "learning_rate": 5e-06,
+      "loss": 0.283,
+      "num_input_tokens_seen": 263944952,
+      "step": 1527
+    },
+    {
+      "epoch": 0.5848333971658368,
+      "loss": 0.2526056170463562,
+      "loss_ce": 0.020794082432985306,
+      "loss_iou": 1.1226778030395508,
+      "loss_num": 0.2314453125,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 263944952,
+      "step": 1527
+    },
+    {
+      "epoch": 0.5852163921869016,
+      "grad_norm": 35.71132886886885,
+      "learning_rate": 5e-06,
+      "loss": 0.228,
+      "num_input_tokens_seen": 264117704,
+      "step": 1528
+    },
+    {
+      "epoch": 0.5852163921869016,
+      "loss": 0.2943909168243408,
+      "loss_ce": 0.023089634254574776,
+      "loss_iou": 1.307476282119751,
+      "loss_num": 0.271484375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 264117704,
+      "step": 1528
+    },
+    {
+      "epoch": 0.5855993872079663,
+      "grad_norm": 46.6442063736582,
+      "learning_rate": 5e-06,
+      "loss": 0.2299,
+      "num_input_tokens_seen": 264287544,
+      "step": 1529
+    },
+    {
+      "epoch": 0.5855993872079663,
+      "loss": 0.28918662667274475,
+      "loss_ce": 0.021120227873325348,
+      "loss_iou": 1.409267783164978,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 264287544,
+      "step": 1529
+    },
+    {
+      "epoch": 0.585982382229031,
+      "grad_norm": 33.45443583117367,
+      "learning_rate": 5e-06,
+      "loss": 0.2288,
+      "num_input_tokens_seen": 264460256,
+      "step": 1530
+    },
+    {
+      "epoch": 0.585982382229031,
+      "loss": 0.23621076345443726,
+      "loss_ce": 0.019535966217517853,
+      "loss_iou": 0.8824535608291626,
+      "loss_num": 0.216796875,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 264460256,
+      "step": 1530
+    },
+    {
+      "epoch": 0.5863653772500957,
+      "grad_norm": 25.02279285582714,
+      "learning_rate": 5e-06,
+      "loss": 0.2264,
+      "num_input_tokens_seen": 264633048,
+      "step": 1531
+    },
+    {
+      "epoch": 0.5863653772500957,
+      "loss": 0.27118995785713196,
+      "loss_ce": 0.020762711763381958,
+      "loss_iou": 1.0662559270858765,
+      "loss_num": 0.25,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 264633048,
+      "step": 1531
+    },
+    {
+      "epoch": 0.5867483722711605,
+      "grad_norm": 31.053513002202813,
+      "learning_rate": 5e-06,
+      "loss": 0.1991,
+      "num_input_tokens_seen": 264806088,
+      "step": 1532
+    },
+    {
+      "epoch": 0.5867483722711605,
+      "loss": 0.20815600454807281,
+      "loss_ce": 0.022426022216677666,
+      "loss_iou": 1.0312124490737915,
+      "loss_num": 0.185546875,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 264806088,
+      "step": 1532
+    },
+    {
+      "epoch": 0.5871313672922251,
+      "grad_norm": 28.228856148742768,
+      "learning_rate": 5e-06,
+      "loss": 0.2329,
+      "num_input_tokens_seen": 264979104,
+      "step": 1533
+    },
+    {
+      "epoch": 0.5871313672922251,
+      "loss": 0.2584834098815918,
+      "loss_ce": 0.02032424695789814,
+      "loss_iou": 1.1087110042572021,
+      "loss_num": 0.23828125,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 264979104,
+      "step": 1533
+    },
+    {
+      "epoch": 0.5875143623132899,
+      "grad_norm": 22.208791524816075,
+      "learning_rate": 5e-06,
+      "loss": 0.2332,
+      "num_input_tokens_seen": 265152016,
+      "step": 1534
+    },
+    {
+      "epoch": 0.5875143623132899,
+      "loss": 0.27769356966018677,
+      "loss_ce": 0.019636942073702812,
+      "loss_iou": 1.3713454008102417,
+      "loss_num": 0.2578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 265152016,
+      "step": 1534
+    },
+    {
+      "epoch": 0.5878973573343547,
+      "grad_norm": 26.176153806246376,
+      "learning_rate": 5e-06,
+      "loss": 0.1873,
+      "num_input_tokens_seen": 265325016,
+      "step": 1535
+    },
+    {
+      "epoch": 0.5878973573343547,
+      "loss": 0.2010667324066162,
+      "loss_ce": 0.021684397011995316,
+      "loss_iou": 1.0136194229125977,
+      "loss_num": 0.1796875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 265325016,
+      "step": 1535
+    },
+    {
+      "epoch": 0.5882803523554194,
+      "grad_norm": 28.91329659880314,
+      "learning_rate": 5e-06,
+      "loss": 0.2185,
+      "num_input_tokens_seen": 265497768,
+      "step": 1536
+    },
+    {
+      "epoch": 0.5882803523554194,
+      "loss": 0.21207498013973236,
+      "loss_ce": 0.019631125032901764,
+      "loss_iou": 1.0956573486328125,
+      "loss_num": 0.1923828125,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 265497768,
+      "step": 1536
+    },
+    {
+      "epoch": 0.5886633473764841,
+      "grad_norm": 29.28604948881464,
+      "learning_rate": 5e-06,
+      "loss": 0.2919,
+      "num_input_tokens_seen": 265670608,
+      "step": 1537
+    },
+    {
+      "epoch": 0.5886633473764841,
+      "loss": 0.36655059456825256,
+      "loss_ce": 0.01877225562930107,
+      "loss_iou": 1.0107009410858154,
+      "loss_num": 0.34765625,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 265670608,
+      "step": 1537
+    },
+    {
+      "epoch": 0.5890463423975488,
+      "grad_norm": 27.81439886334659,
+      "learning_rate": 5e-06,
+      "loss": 0.22,
+      "num_input_tokens_seen": 265842936,
+      "step": 1538
+    },
+    {
+      "epoch": 0.5890463423975488,
+      "loss": 0.25227534770965576,
+      "loss_ce": 0.019792431965470314,
+      "loss_iou": 1.0118613243103027,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 265842936,
+      "step": 1538
+    },
+    {
+      "epoch": 0.5894293374186136,
+      "grad_norm": 36.55628004442563,
+      "learning_rate": 5e-06,
+      "loss": 0.2325,
+      "num_input_tokens_seen": 266015696,
+      "step": 1539
+    },
+    {
+      "epoch": 0.5894293374186136,
+      "loss": 0.26872682571411133,
+      "loss_ce": 0.020802026614546776,
+      "loss_iou": 1.065087914466858,
+      "loss_num": 0.248046875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 266015696,
+      "step": 1539
+    },
+    {
+      "epoch": 0.5898123324396782,
+      "grad_norm": 39.728022637273845,
+      "learning_rate": 5e-06,
+      "loss": 0.2392,
+      "num_input_tokens_seen": 266189008,
+      "step": 1540
+    },
+    {
+      "epoch": 0.5898123324396782,
+      "loss": 0.18677133321762085,
+      "loss_ce": 0.022037452086806297,
+      "loss_iou": 1.0023694038391113,
+      "loss_num": 0.1650390625,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 266189008,
+      "step": 1540
+    },
+    {
+      "epoch": 0.590195327460743,
+      "grad_norm": 30.577603448642417,
+      "learning_rate": 5e-06,
+      "loss": 0.2204,
+      "num_input_tokens_seen": 266362008,
+      "step": 1541
+    },
+    {
+      "epoch": 0.590195327460743,
+      "loss": 0.20363269746303558,
+      "loss_ce": 0.020283091813325882,
+      "loss_iou": 1.0037730932235718,
+      "loss_num": 0.18359375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 266362008,
+      "step": 1541
+    },
+    {
+      "epoch": 0.5905783224818077,
+      "grad_norm": 31.569422166715896,
+      "learning_rate": 5e-06,
+      "loss": 0.1983,
+      "num_input_tokens_seen": 266535504,
+      "step": 1542
+    },
+    {
+      "epoch": 0.5905783224818077,
+      "loss": 0.25024089217185974,
+      "loss_ce": 0.02178630605340004,
+      "loss_iou": 1.2143375873565674,
+      "loss_num": 0.228515625,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 266535504,
+      "step": 1542
+    },
+    {
+      "epoch": 0.5909613175028725,
+      "grad_norm": 35.75139674065863,
+      "learning_rate": 5e-06,
+      "loss": 0.2342,
+      "num_input_tokens_seen": 266708360,
+      "step": 1543
+    },
+    {
+      "epoch": 0.5909613175028725,
+      "loss": 0.1854146420955658,
+      "loss_ce": 0.019337981939315796,
+      "loss_iou": 1.0077852010726929,
+      "loss_num": 0.166015625,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 266708360,
+      "step": 1543
+    },
+    {
+      "epoch": 0.5913443125239372,
+      "grad_norm": 28.310511466260408,
+      "learning_rate": 5e-06,
+      "loss": 0.2223,
+      "num_input_tokens_seen": 266881232,
+      "step": 1544
+    },
+    {
+      "epoch": 0.5913443125239372,
+      "loss": 0.1748485565185547,
+      "loss_ce": 0.019269946962594986,
+      "loss_iou": 1.0812819004058838,
+      "loss_num": 0.1552734375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 266881232,
+      "step": 1544
+    },
+    {
+      "epoch": 0.5917273075450019,
+      "grad_norm": 32.615718350974944,
+      "learning_rate": 5e-06,
+      "loss": 0.2253,
+      "num_input_tokens_seen": 267054064,
+      "step": 1545
+    },
+    {
+      "epoch": 0.5917273075450019,
+      "loss": 0.2351870834827423,
+      "loss_ce": 0.020587481558322906,
+      "loss_iou": 1.2616922855377197,
+      "loss_num": 0.21484375,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 267054064,
+      "step": 1545
+    },
+    {
+      "epoch": 0.5921103025660667,
+      "grad_norm": 33.03801999893034,
+      "learning_rate": 5e-06,
+      "loss": 0.2478,
+      "num_input_tokens_seen": 267226800,
+      "step": 1546
+    },
+    {
+      "epoch": 0.5921103025660667,
+      "loss": 0.2021588236093521,
+      "loss_ce": 0.020975960418581963,
+      "loss_iou": 1.0036303997039795,
+      "loss_num": 0.181640625,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 267226800,
+      "step": 1546
+    },
+    {
+      "epoch": 0.5924932975871313,
+      "grad_norm": 29.573668790736626,
+      "learning_rate": 5e-06,
+      "loss": 0.3187,
+      "num_input_tokens_seen": 267399576,
+      "step": 1547
+    },
+    {
+      "epoch": 0.5924932975871313,
+      "loss": 0.28870028257369995,
+      "loss_ce": 0.019474195316433907,
+      "loss_iou": 1.0450565814971924,
+      "loss_num": 0.26953125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 267399576,
+      "step": 1547
+    },
+    {
+      "epoch": 0.5928762926081961,
+      "grad_norm": 34.80479538562652,
+      "learning_rate": 5e-06,
+      "loss": 0.2033,
+      "num_input_tokens_seen": 267572288,
+      "step": 1548
+    },
+    {
+      "epoch": 0.5928762926081961,
+      "loss": 0.240036278963089,
+      "loss_ce": 0.02067592367529869,
+      "loss_iou": 1.066023349761963,
+      "loss_num": 0.2197265625,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 267572288,
+      "step": 1548
+    },
+    {
+      "epoch": 0.5932592876292608,
+      "grad_norm": 33.705767028333696,
+      "learning_rate": 5e-06,
+      "loss": 0.2367,
+      "num_input_tokens_seen": 267745432,
+      "step": 1549
+    },
+    {
+      "epoch": 0.5932592876292608,
+      "loss": 0.24160504341125488,
+      "loss_ce": 0.02181744948029518,
+      "loss_iou": 1.0746049880981445,
+      "loss_num": 0.2197265625,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 267745432,
+      "step": 1549
+    },
+    {
+      "epoch": 0.5936422826503256,
+      "grad_norm": 27.237508864290163,
+      "learning_rate": 5e-06,
+      "loss": 0.2539,
+      "num_input_tokens_seen": 267918424,
+      "step": 1550
+    },
+    {
+      "epoch": 0.5936422826503256,
+      "loss": 0.22879761457443237,
+      "loss_ce": 0.020240485668182373,
+      "loss_iou": 1.034693717956543,
+      "loss_num": 0.208984375,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 267918424,
+      "step": 1550
+    },
+    {
+      "epoch": 0.5940252776713902,
+      "grad_norm": 25.724312141977965,
+      "learning_rate": 5e-06,
+      "loss": 0.2716,
+      "num_input_tokens_seen": 268087952,
+      "step": 1551
+    },
+    {
+      "epoch": 0.5940252776713902,
+      "loss": 0.2817750573158264,
+      "loss_ce": 0.02207046002149582,
+      "loss_iou": 1.0592975616455078,
+      "loss_num": 0.259765625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 268087952,
+      "step": 1551
+    },
+    {
+      "epoch": 0.594408272692455,
+      "grad_norm": 26.99685174643569,
+      "learning_rate": 5e-06,
+      "loss": 0.2457,
+      "num_input_tokens_seen": 268260736,
+      "step": 1552
+    },
+    {
+      "epoch": 0.594408272692455,
+      "loss": 0.26946550607681274,
+      "loss_ce": 0.021540693938732147,
+      "loss_iou": 1.043034315109253,
+      "loss_num": 0.248046875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 268260736,
+      "step": 1552
+    },
+    {
+      "epoch": 0.5947912677135198,
+      "grad_norm": 34.15153816987403,
+      "learning_rate": 5e-06,
+      "loss": 0.2533,
+      "num_input_tokens_seen": 268433768,
+      "step": 1553
+    },
+    {
+      "epoch": 0.5947912677135198,
+      "loss": 0.27357256412506104,
+      "loss_ce": 0.020703919231891632,
+      "loss_iou": 1.0922720432281494,
+      "loss_num": 0.251953125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 268433768,
+      "step": 1553
+    },
+    {
+      "epoch": 0.5951742627345844,
+      "grad_norm": 30.296502570549386,
+      "learning_rate": 5e-06,
+      "loss": 0.1694,
+      "num_input_tokens_seen": 268606672,
+      "step": 1554
+    },
+    {
+      "epoch": 0.5951742627345844,
+      "loss": 0.1596354842185974,
+      "loss_ce": 0.01974291354417801,
+      "loss_iou": 1.0062255859375,
+      "loss_num": 0.1396484375,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 268606672,
+      "step": 1554
+    },
+    {
+      "epoch": 0.5955572577556492,
+      "grad_norm": 30.945768386800445,
+      "learning_rate": 5e-06,
+      "loss": 0.1955,
+      "num_input_tokens_seen": 268779456,
+      "step": 1555
+    },
+    {
+      "epoch": 0.5955572577556492,
+      "loss": 0.19111934304237366,
+      "loss_ce": 0.01985469087958336,
+      "loss_iou": 1.0321484804153442,
+      "loss_num": 0.1708984375,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 268779456,
+      "step": 1555
+    },
+    {
+      "epoch": 0.5959402527767139,
+      "grad_norm": 29.196716116468462,
+      "learning_rate": 5e-06,
+      "loss": 0.2325,
+      "num_input_tokens_seen": 268952560,
+      "step": 1556
+    },
+    {
+      "epoch": 0.5959402527767139,
+      "loss": 0.24636295437812805,
+      "loss_ce": 0.020838063210248947,
+      "loss_iou": 1.1337873935699463,
+      "loss_num": 0.2255859375,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 268952560,
+      "step": 1556
+    },
+    {
+      "epoch": 0.5963232477977787,
+      "grad_norm": 33.06575460533335,
+      "learning_rate": 5e-06,
+      "loss": 0.271,
+      "num_input_tokens_seen": 269125648,
+      "step": 1557
+    },
+    {
+      "epoch": 0.5963232477977787,
+      "loss": 0.2610067129135132,
+      "loss_ce": 0.022725481539964676,
+      "loss_iou": 1.0387709140777588,
+      "loss_num": 0.23828125,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 269125648,
+      "step": 1557
+    },
+    {
+      "epoch": 0.5967062428188433,
+      "grad_norm": 36.4405642598575,
+      "learning_rate": 5e-06,
+      "loss": 0.2514,
+      "num_input_tokens_seen": 269298584,
+      "step": 1558
+    },
+    {
+      "epoch": 0.5967062428188433,
+      "loss": 0.23544277250766754,
+      "loss_ce": 0.02011074870824814,
+      "loss_iou": 1.0255565643310547,
+      "loss_num": 0.21484375,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 269298584,
+      "step": 1558
+    },
+    {
+      "epoch": 0.5970892378399081,
+      "grad_norm": 39.24894248601204,
+      "learning_rate": 5e-06,
+      "loss": 0.3001,
+      "num_input_tokens_seen": 269471696,
+      "step": 1559
+    },
+    {
+      "epoch": 0.5970892378399081,
+      "loss": 0.28608623147010803,
+      "loss_ce": 0.021193664520978928,
+      "loss_iou": 1.002530574798584,
+      "loss_num": 0.265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 269471696,
+      "step": 1559
+    },
+    {
+      "epoch": 0.5974722328609728,
+      "grad_norm": 33.06850598554862,
+      "learning_rate": 5e-06,
+      "loss": 0.2206,
+      "num_input_tokens_seen": 269644656,
+      "step": 1560
+    },
+    {
+      "epoch": 0.5974722328609728,
+      "loss": 0.22993288934230804,
+      "loss_ce": 0.019605742767453194,
+      "loss_iou": 1.0346921682357788,
+      "loss_num": 0.2099609375,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 269644656,
+      "step": 1560
+    },
+    {
+      "epoch": 0.5978552278820375,
+      "grad_norm": 29.91969492518461,
+      "learning_rate": 5e-06,
+      "loss": 0.2925,
+      "num_input_tokens_seen": 269817520,
+      "step": 1561
+    },
+    {
+      "epoch": 0.5978552278820375,
+      "loss": 0.2565000057220459,
+      "loss_ce": 0.019561532884836197,
+      "loss_iou": 1.0348634719848633,
+      "loss_num": 0.2373046875,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 269817520,
+      "step": 1561
+    },
+    {
+      "epoch": 0.5982382229031022,
+      "grad_norm": 21.177720280584925,
+      "learning_rate": 5e-06,
+      "loss": 0.2179,
+      "num_input_tokens_seen": 269990584,
+      "step": 1562
+    },
+    {
+      "epoch": 0.5982382229031022,
+      "loss": 0.19144877791404724,
+      "loss_ce": 0.02073344960808754,
+      "loss_iou": 1.016440987586975,
+      "loss_num": 0.1708984375,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 269990584,
+      "step": 1562
+    },
+    {
+      "epoch": 0.598621217924167,
+      "grad_norm": 23.906628156849884,
+      "learning_rate": 5e-06,
+      "loss": 0.2542,
+      "num_input_tokens_seen": 270163040,
+      "step": 1563
+    },
+    {
+      "epoch": 0.598621217924167,
+      "loss": 0.2534239590167999,
+      "loss_ce": 0.020025519654154778,
+      "loss_iou": 1.0246292352676392,
+      "loss_num": 0.2333984375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 270163040,
+      "step": 1563
+    },
+    {
+      "epoch": 0.5990042129452318,
+      "grad_norm": 26.215274489795025,
+      "learning_rate": 5e-06,
+      "loss": 0.2049,
+      "num_input_tokens_seen": 270336280,
+      "step": 1564
+    },
+    {
+      "epoch": 0.5990042129452318,
+      "loss": 0.1972009241580963,
+      "loss_ce": 0.02178589627146721,
+      "loss_iou": 1.0354797840118408,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 270336280,
+      "step": 1564
+    },
+    {
+      "epoch": 0.5993872079662964,
+      "grad_norm": 36.62401599598024,
+      "learning_rate": 5e-06,
+      "loss": 0.2472,
+      "num_input_tokens_seen": 270509168,
+      "step": 1565
+    },
+    {
+      "epoch": 0.5993872079662964,
+      "loss": 0.19089263677597046,
+      "loss_ce": 0.02231353148818016,
+      "loss_iou": 1.0060560703277588,
+      "loss_num": 0.1689453125,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 270509168,
+      "step": 1565
+    },
+    {
+      "epoch": 0.5997702029873612,
+      "grad_norm": 30.118964926716398,
+      "learning_rate": 5e-06,
+      "loss": 0.2621,
+      "num_input_tokens_seen": 270681968,
+      "step": 1566
+    },
+    {
+      "epoch": 0.5997702029873612,
+      "loss": 0.259109228849411,
+      "loss_ce": 0.02119421400129795,
+      "loss_iou": 1.0127711296081543,
+      "loss_num": 0.23828125,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 270681968,
+      "step": 1566
+    },
+    {
+      "epoch": 0.6001531980084259,
+      "grad_norm": 21.00401339010055,
+      "learning_rate": 5e-06,
+      "loss": 0.2207,
+      "num_input_tokens_seen": 270854608,
+      "step": 1567
+    },
+    {
+      "epoch": 0.6001531980084259,
+      "loss": 0.22783751785755157,
+      "loss_ce": 0.020623154938220978,
+      "loss_iou": 1.0646750926971436,
+      "loss_num": 0.20703125,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 270854608,
+      "step": 1567
+    },
+    {
+      "epoch": 0.6005361930294906,
+      "grad_norm": 17.71810463096343,
+      "learning_rate": 5e-06,
+      "loss": 0.1758,
+      "num_input_tokens_seen": 271027576,
+      "step": 1568
+    },
+    {
+      "epoch": 0.6005361930294906,
+      "loss": 0.12108717858791351,
+      "loss_ce": 0.021081075072288513,
+      "loss_iou": 1.0009524822235107,
+      "loss_num": 0.10009765625,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 271027576,
+      "step": 1568
+    },
+    {
+      "epoch": 0.6009191880505553,
+      "grad_norm": 41.2732816626539,
+      "learning_rate": 5e-06,
+      "loss": 0.2311,
+      "num_input_tokens_seen": 271200608,
+      "step": 1569
+    },
+    {
+      "epoch": 0.6009191880505553,
+      "loss": 0.2101738303899765,
+      "loss_ce": 0.021270014345645905,
+      "loss_iou": 1.0262513160705566,
+      "loss_num": 0.1884765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 271200608,
+      "step": 1569
+    },
+    {
+      "epoch": 0.6013021830716201,
+      "grad_norm": 37.32284388122061,
+      "learning_rate": 5e-06,
+      "loss": 0.3139,
+      "num_input_tokens_seen": 271373896,
+      "step": 1570
+    },
+    {
+      "epoch": 0.6013021830716201,
+      "loss": 0.32823628187179565,
+      "loss_ce": 0.01970357820391655,
+      "loss_iou": 1.0195873975753784,
+      "loss_num": 0.30859375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 271373896,
+      "step": 1570
+    },
+    {
+      "epoch": 0.6016851780926848,
+      "grad_norm": 28.475165993161532,
+      "learning_rate": 5e-06,
+      "loss": 0.2955,
+      "num_input_tokens_seen": 271546480,
+      "step": 1571
+    },
+    {
+      "epoch": 0.6016851780926848,
+      "loss": 0.2166106402873993,
+      "loss_ce": 0.021786419674754143,
+      "loss_iou": 1.0070724487304688,
+      "loss_num": 0.1953125,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 271546480,
+      "step": 1571
+    },
+    {
+      "epoch": 0.6020681731137495,
+      "grad_norm": 25.18692711010138,
+      "learning_rate": 5e-06,
+      "loss": 0.2575,
+      "num_input_tokens_seen": 271718888,
+      "step": 1572
+    },
+    {
+      "epoch": 0.6020681731137495,
+      "loss": 0.29178929328918457,
+      "loss_ce": 0.020854221656918526,
+      "loss_iou": 1.1269915103912354,
+      "loss_num": 0.271484375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 271718888,
+      "step": 1572
+    },
+    {
+      "epoch": 0.6024511681348143,
+      "grad_norm": 46.49956786920276,
+      "learning_rate": 5e-06,
+      "loss": 0.2366,
+      "num_input_tokens_seen": 271890128,
+      "step": 1573
+    },
+    {
+      "epoch": 0.6024511681348143,
+      "loss": 0.2369789481163025,
+      "loss_ce": 0.021891064941883087,
+      "loss_iou": 1.002115249633789,
+      "loss_num": 0.21484375,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 271890128,
+      "step": 1573
+    },
+    {
+      "epoch": 0.602834163155879,
+      "grad_norm": 32.18644200082336,
+      "learning_rate": 5e-06,
+      "loss": 0.309,
+      "num_input_tokens_seen": 272063224,
+      "step": 1574
+    },
+    {
+      "epoch": 0.602834163155879,
+      "loss": 0.2830401659011841,
+      "loss_ce": 0.021748654544353485,
+      "loss_iou": 1.1032469272613525,
+      "loss_num": 0.26171875,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 272063224,
+      "step": 1574
+    },
+    {
+      "epoch": 0.6032171581769437,
+      "grad_norm": 47.72776838553225,
+      "learning_rate": 5e-06,
+      "loss": 0.2869,
+      "num_input_tokens_seen": 272235864,
+      "step": 1575
+    },
+    {
+      "epoch": 0.6032171581769437,
+      "loss": 0.3127654194831848,
+      "loss_ce": 0.019064251333475113,
+      "loss_iou": 1.054098129272461,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 272235864,
+      "step": 1575
+    },
+    {
+      "epoch": 0.6036001531980084,
+      "grad_norm": 28.609362106101166,
+      "learning_rate": 5e-06,
+      "loss": 0.2425,
+      "num_input_tokens_seen": 272408776,
+      "step": 1576
+    },
+    {
+      "epoch": 0.6036001531980084,
+      "loss": 0.1467621922492981,
+      "loss_ce": 0.019137680530548096,
+      "loss_iou": 1.0034462213516235,
+      "loss_num": 0.1279296875,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 272408776,
+      "step": 1576
+    },
+    {
+      "epoch": 0.6039831482190732,
+      "grad_norm": 28.13433494976978,
+      "learning_rate": 5e-06,
+      "loss": 0.2013,
+      "num_input_tokens_seen": 272581528,
+      "step": 1577
+    },
+    {
+      "epoch": 0.6039831482190732,
+      "loss": 0.2090604156255722,
+      "loss_ce": 0.018996933475136757,
+      "loss_iou": 1.0215201377868652,
+      "loss_num": 0.1904296875,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 272581528,
+      "step": 1577
+    },
+    {
+      "epoch": 0.6043661432401379,
+      "grad_norm": 30.14117326051236,
+      "learning_rate": 5e-06,
+      "loss": 0.2265,
+      "num_input_tokens_seen": 272754568,
+      "step": 1578
+    },
+    {
+      "epoch": 0.6043661432401379,
+      "loss": 0.2674601674079895,
+      "loss_ce": 0.021793659776449203,
+      "loss_iou": 1.015618085861206,
+      "loss_num": 0.24609375,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 272754568,
+      "step": 1578
+    },
+    {
+      "epoch": 0.6047491382612026,
+      "grad_norm": 28.68259479834075,
+      "learning_rate": 5e-06,
+      "loss": 0.2768,
+      "num_input_tokens_seen": 272927408,
+      "step": 1579
+    },
+    {
+      "epoch": 0.6047491382612026,
+      "loss": 0.2753274738788605,
+      "loss_ce": 0.021055009216070175,
+      "loss_iou": 0.9962226152420044,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 272927408,
+      "step": 1579
+    },
+    {
+      "epoch": 0.6051321332822673,
+      "grad_norm": 31.988660693403762,
+      "learning_rate": 5e-06,
+      "loss": 0.234,
+      "num_input_tokens_seen": 273100464,
+      "step": 1580
+    },
+    {
+      "epoch": 0.6051321332822673,
+      "loss": 0.22599288821220398,
+      "loss_ce": 0.023051002994179726,
+      "loss_iou": 1.0454695224761963,
+      "loss_num": 0.203125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 273100464,
+      "step": 1580
+    },
+    {
+      "epoch": 0.6055151283033321,
+      "grad_norm": 34.11420750058403,
+      "learning_rate": 5e-06,
+      "loss": 0.3195,
+      "num_input_tokens_seen": 273273432,
+      "step": 1581
+    },
+    {
+      "epoch": 0.6055151283033321,
+      "loss": 0.34745293855667114,
+      "loss_ce": 0.022135566920042038,
+      "loss_iou": 1.0547513961791992,
+      "loss_num": 0.326171875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 273273432,
+      "step": 1581
+    },
+    {
+      "epoch": 0.6058981233243967,
+      "grad_norm": 29.537134166160698,
+      "learning_rate": 5e-06,
+      "loss": 0.2829,
+      "num_input_tokens_seen": 273446384,
+      "step": 1582
+    },
+    {
+      "epoch": 0.6058981233243967,
+      "loss": 0.26747170090675354,
+      "loss_ce": 0.02088966965675354,
+      "loss_iou": 1.044858455657959,
+      "loss_num": 0.24609375,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 273446384,
+      "step": 1582
+    },
+    {
+      "epoch": 0.6062811183454615,
+      "grad_norm": 27.618517631204,
+      "learning_rate": 5e-06,
+      "loss": 0.2604,
+      "num_input_tokens_seen": 273619232,
+      "step": 1583
+    },
+    {
+      "epoch": 0.6062811183454615,
+      "loss": 0.25198063254356384,
+      "loss_ce": 0.020962558686733246,
+      "loss_iou": 1.0346366167068481,
+      "loss_num": 0.2314453125,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 273619232,
+      "step": 1583
+    },
+    {
+      "epoch": 0.6066641133665263,
+      "grad_norm": 25.74671603837596,
+      "learning_rate": 5e-06,
+      "loss": 0.2338,
+      "num_input_tokens_seen": 273792080,
+      "step": 1584
+    },
+    {
+      "epoch": 0.6066641133665263,
+      "loss": 0.26737356185913086,
+      "loss_ce": 0.021035686135292053,
+      "loss_iou": 1.031913161277771,
+      "loss_num": 0.24609375,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 273792080,
+      "step": 1584
+    },
+    {
+      "epoch": 0.607047108387591,
+      "grad_norm": 27.69611185926586,
+      "learning_rate": 5e-06,
+      "loss": 0.2177,
+      "num_input_tokens_seen": 273964672,
+      "step": 1585
+    },
+    {
+      "epoch": 0.607047108387591,
+      "loss": 0.17387646436691284,
+      "loss_ce": 0.02037305384874344,
+      "loss_iou": 1.0016062259674072,
+      "loss_num": 0.1533203125,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 273964672,
+      "step": 1585
+    },
+    {
+      "epoch": 0.6074301034086557,
+      "grad_norm": 22.44781098462438,
+      "learning_rate": 5e-06,
+      "loss": 0.1801,
+      "num_input_tokens_seen": 274136968,
+      "step": 1586
+    },
+    {
+      "epoch": 0.6074301034086557,
+      "loss": 0.1811303198337555,
+      "loss_ce": 0.020180605351924896,
+      "loss_iou": 1.0062899589538574,
+      "loss_num": 0.1611328125,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 274136968,
+      "step": 1586
+    },
+    {
+      "epoch": 0.6078130984297204,
+      "grad_norm": 28.575654203393004,
+      "learning_rate": 5e-06,
+      "loss": 0.2037,
+      "num_input_tokens_seen": 274309800,
+      "step": 1587
+    },
+    {
+      "epoch": 0.6078130984297204,
+      "loss": 0.19381070137023926,
+      "loss_ce": 0.021447420120239258,
+      "loss_iou": 1.0267584323883057,
+      "loss_num": 0.171875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 274309800,
+      "step": 1587
+    },
+    {
+      "epoch": 0.6081960934507852,
+      "grad_norm": 29.977417165715824,
+      "learning_rate": 5e-06,
+      "loss": 0.234,
+      "num_input_tokens_seen": 274479440,
+      "step": 1588
+    },
+    {
+      "epoch": 0.6081960934507852,
+      "loss": 0.30903321504592896,
+      "loss_ce": 0.021252445876598358,
+      "loss_iou": 1.0187793970108032,
+      "loss_num": 0.287109375,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 274479440,
+      "step": 1588
+    },
+    {
+      "epoch": 0.6085790884718498,
+      "grad_norm": 27.635098052466525,
+      "learning_rate": 5e-06,
+      "loss": 0.1795,
+      "num_input_tokens_seen": 274652112,
+      "step": 1589
+    },
+    {
+      "epoch": 0.6085790884718498,
+      "loss": 0.21000096201896667,
+      "loss_ce": 0.021585427224636078,
+      "loss_iou": 1.0599991083145142,
+      "loss_num": 0.1884765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 274652112,
+      "step": 1589
+    },
+    {
+      "epoch": 0.6089620834929146,
+      "grad_norm": 33.60852914184709,
+      "learning_rate": 5e-06,
+      "loss": 0.2577,
+      "num_input_tokens_seen": 274825168,
+      "step": 1590
+    },
+    {
+      "epoch": 0.6089620834929146,
+      "loss": 0.2673984467983246,
+      "loss_ce": 0.02301366999745369,
+      "loss_iou": 1.0333936214447021,
+      "loss_num": 0.244140625,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 274825168,
+      "step": 1590
+    },
+    {
+      "epoch": 0.6093450785139793,
+      "grad_norm": 31.5686827262231,
+      "learning_rate": 5e-06,
+      "loss": 0.2695,
+      "num_input_tokens_seen": 274998272,
+      "step": 1591
+    },
+    {
+      "epoch": 0.6093450785139793,
+      "loss": 0.2690613865852356,
+      "loss_ce": 0.020587250590324402,
+      "loss_iou": 1.126404047012329,
+      "loss_num": 0.248046875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 274998272,
+      "step": 1591
+    },
+    {
+      "epoch": 0.609728073535044,
+      "grad_norm": 24.76210316977389,
+      "learning_rate": 5e-06,
+      "loss": 0.2361,
+      "num_input_tokens_seen": 275171152,
+      "step": 1592
+    },
+    {
+      "epoch": 0.609728073535044,
+      "loss": 0.18169772624969482,
+      "loss_ce": 0.02050388790667057,
+      "loss_iou": 1.0014326572418213,
+      "loss_num": 0.1611328125,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 275171152,
+      "step": 1592
+    },
+    {
+      "epoch": 0.6101110685561087,
+      "grad_norm": 32.5769524023531,
+      "learning_rate": 5e-06,
+      "loss": 0.2315,
+      "num_input_tokens_seen": 275344120,
+      "step": 1593
+    },
+    {
+      "epoch": 0.6101110685561087,
+      "loss": 0.2832563519477844,
+      "loss_ce": 0.019584478810429573,
+      "loss_iou": 1.044764518737793,
+      "loss_num": 0.263671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 275344120,
+      "step": 1593
+    },
+    {
+      "epoch": 0.6104940635771735,
+      "grad_norm": 39.1097092552448,
+      "learning_rate": 5e-06,
+      "loss": 0.2219,
+      "num_input_tokens_seen": 275517296,
+      "step": 1594
+    },
+    {
+      "epoch": 0.6104940635771735,
+      "loss": 0.2574824392795563,
+      "loss_ce": 0.02152051031589508,
+      "loss_iou": 1.0240764617919922,
+      "loss_num": 0.236328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 275517296,
+      "step": 1594
+    },
+    {
+      "epoch": 0.6108770585982383,
+      "grad_norm": 34.747334844192,
+      "learning_rate": 5e-06,
+      "loss": 0.1953,
+      "num_input_tokens_seen": 275689976,
+      "step": 1595
+    },
+    {
+      "epoch": 0.6108770585982383,
+      "loss": 0.20638251304626465,
+      "loss_ce": 0.02126288414001465,
+      "loss_iou": 1.0271966457366943,
+      "loss_num": 0.185546875,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 275689976,
+      "step": 1595
+    },
+    {
+      "epoch": 0.6112600536193029,
+      "grad_norm": 31.4174384960578,
+      "learning_rate": 5e-06,
+      "loss": 0.2529,
+      "num_input_tokens_seen": 275863072,
+      "step": 1596
+    },
+    {
+      "epoch": 0.6112600536193029,
+      "loss": 0.3017125725746155,
+      "loss_ce": 0.021439146250486374,
+      "loss_iou": 1.0911595821380615,
+      "loss_num": 0.28125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 275863072,
+      "step": 1596
+    },
+    {
+      "epoch": 0.6116430486403677,
+      "grad_norm": 24.44942504133733,
+      "learning_rate": 5e-06,
+      "loss": 0.2421,
+      "num_input_tokens_seen": 276035896,
+      "step": 1597
+    },
+    {
+      "epoch": 0.6116430486403677,
+      "loss": 0.2739262282848358,
+      "loss_ce": 0.02203413099050522,
+      "loss_iou": 1.0876541137695312,
+      "loss_num": 0.251953125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 276035896,
+      "step": 1597
+    },
+    {
+      "epoch": 0.6120260436614324,
+      "grad_norm": 32.93735089841536,
+      "learning_rate": 5e-06,
+      "loss": 0.1792,
+      "num_input_tokens_seen": 276209040,
+      "step": 1598
+    },
+    {
+      "epoch": 0.6120260436614324,
+      "loss": 0.1790980100631714,
+      "loss_ce": 0.02065073698759079,
+      "loss_iou": 1.0007457733154297,
+      "loss_num": 0.158203125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 276209040,
+      "step": 1598
+    },
+    {
+      "epoch": 0.6124090386824971,
+      "grad_norm": 37.411721853665675,
+      "learning_rate": 5e-06,
+      "loss": 0.27,
+      "num_input_tokens_seen": 276378432,
+      "step": 1599
+    },
+    {
+      "epoch": 0.6124090386824971,
+      "loss": 0.1956443190574646,
+      "loss_ce": 0.020107213407754898,
+      "loss_iou": 1.0098230838775635,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 276378432,
+      "step": 1599
+    },
+    {
+      "epoch": 0.6127920337035618,
+      "grad_norm": 18.335439105215123,
+      "learning_rate": 5e-06,
+      "loss": 0.2042,
+      "num_input_tokens_seen": 276551216,
+      "step": 1600
+    },
+    {
+      "epoch": 0.6127920337035618,
+      "loss": 0.1819685399532318,
+      "loss_ce": 0.021262984722852707,
+      "loss_iou": 1.0139724016189575,
+      "loss_num": 0.1611328125,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 276551216,
+      "step": 1600
+    },
+    {
+      "epoch": 0.6131750287246266,
+      "grad_norm": 20.397255184454824,
+      "learning_rate": 5e-06,
+      "loss": 0.2506,
+      "num_input_tokens_seen": 276724144,
+      "step": 1601
+    },
+    {
+      "epoch": 0.6131750287246266,
+      "loss": 0.29360121488571167,
+      "loss_ce": 0.02052992209792137,
+      "loss_iou": 1.0553431510925293,
+      "loss_num": 0.2734375,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 276724144,
+      "step": 1601
+    },
+    {
+      "epoch": 0.6135580237456913,
+      "grad_norm": 29.533116116854934,
+      "learning_rate": 5e-06,
+      "loss": 0.2152,
+      "num_input_tokens_seen": 276897128,
+      "step": 1602
+    },
+    {
+      "epoch": 0.6135580237456913,
+      "loss": 0.2100106179714203,
+      "loss_ce": 0.021228883415460587,
+      "loss_iou": 1.032307744026184,
+      "loss_num": 0.1884765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 276897128,
+      "step": 1602
+    },
+    {
+      "epoch": 0.613941018766756,
+      "grad_norm": 34.48299475173718,
+      "learning_rate": 5e-06,
+      "loss": 0.1998,
+      "num_input_tokens_seen": 277070168,
+      "step": 1603
+    },
+    {
+      "epoch": 0.613941018766756,
+      "loss": 0.22132205963134766,
+      "loss_ce": 0.021798141300678253,
+      "loss_iou": 1.0356637239456177,
+      "loss_num": 0.19921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 277070168,
+      "step": 1603
+    },
+    {
+      "epoch": 0.6143240137878208,
+      "grad_norm": 35.99404331891088,
+      "learning_rate": 5e-06,
+      "loss": 0.2527,
+      "num_input_tokens_seen": 277242848,
+      "step": 1604
+    },
+    {
+      "epoch": 0.6143240137878208,
+      "loss": 0.2681909203529358,
+      "loss_ce": 0.02179197408258915,
+      "loss_iou": 1.2864093780517578,
+      "loss_num": 0.24609375,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 277242848,
+      "step": 1604
+    },
+    {
+      "epoch": 0.6147070088088855,
+      "grad_norm": 28.881208952887658,
+      "learning_rate": 5e-06,
+      "loss": 0.2144,
+      "num_input_tokens_seen": 277415664,
+      "step": 1605
+    },
+    {
+      "epoch": 0.6147070088088855,
+      "loss": 0.22464069724082947,
+      "loss_ce": 0.020356036722660065,
+      "loss_iou": 0.9713693857192993,
+      "loss_num": 0.2041015625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 277415664,
+      "step": 1605
+    },
+    {
+      "epoch": 0.6150900038299502,
+      "grad_norm": 27.440955567741156,
+      "learning_rate": 5e-06,
+      "loss": 0.1921,
+      "num_input_tokens_seen": 277588912,
+      "step": 1606
+    },
+    {
+      "epoch": 0.6150900038299502,
+      "loss": 0.14175783097743988,
+      "loss_ce": 0.022281505167484283,
+      "loss_iou": 1.0071306228637695,
+      "loss_num": 0.11962890625,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 277588912,
+      "step": 1606
+    },
+    {
+      "epoch": 0.6154729988510149,
+      "grad_norm": 37.156480360771,
+      "learning_rate": 5e-06,
+      "loss": 0.255,
+      "num_input_tokens_seen": 277761752,
+      "step": 1607
+    },
+    {
+      "epoch": 0.6154729988510149,
+      "loss": 0.26891425251960754,
+      "loss_ce": 0.01909736543893814,
+      "loss_iou": 1.0075769424438477,
+      "loss_num": 0.25,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 277761752,
+      "step": 1607
+    },
+    {
+      "epoch": 0.6158559938720797,
+      "grad_norm": 25.095222555526888,
+      "learning_rate": 5e-06,
+      "loss": 0.2492,
+      "num_input_tokens_seen": 277934936,
+      "step": 1608
+    },
+    {
+      "epoch": 0.6158559938720797,
+      "loss": 0.3241659998893738,
+      "loss_ce": 0.02149268612265587,
+      "loss_iou": 1.0435982942581177,
+      "loss_num": 0.302734375,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 277934936,
+      "step": 1608
+    },
+    {
+      "epoch": 0.6162389888931444,
+      "grad_norm": 27.34854352904686,
+      "learning_rate": 5e-06,
+      "loss": 0.1863,
+      "num_input_tokens_seen": 278107928,
+      "step": 1609
+    },
+    {
+      "epoch": 0.6162389888931444,
+      "loss": 0.1921370029449463,
+      "loss_ce": 0.02007889747619629,
+      "loss_iou": 1.0090808868408203,
+      "loss_num": 0.171875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 278107928,
+      "step": 1609
+    },
+    {
+      "epoch": 0.6166219839142091,
+      "grad_norm": 37.360157431991865,
+      "learning_rate": 5e-06,
+      "loss": 0.2326,
+      "num_input_tokens_seen": 278280776,
+      "step": 1610
+    },
+    {
+      "epoch": 0.6166219839142091,
+      "loss": 0.2784643769264221,
+      "loss_ce": 0.020224634557962418,
+      "loss_iou": 1.1624329090118408,
+      "loss_num": 0.2578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 278280776,
+      "step": 1610
+    },
+    {
+      "epoch": 0.6170049789352738,
+      "grad_norm": 54.19153033092486,
+      "learning_rate": 5e-06,
+      "loss": 0.2161,
+      "num_input_tokens_seen": 278453728,
+      "step": 1611
+    },
+    {
+      "epoch": 0.6170049789352738,
+      "loss": 0.22191108763217926,
+      "loss_ce": 0.022753378376364708,
+      "loss_iou": 1.0086132287979126,
+      "loss_num": 0.19921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 278453728,
+      "step": 1611
+    },
+    {
+      "epoch": 0.6173879739563386,
+      "grad_norm": 33.036250720636794,
+      "learning_rate": 5e-06,
+      "loss": 0.266,
+      "num_input_tokens_seen": 278626800,
+      "step": 1612
+    },
+    {
+      "epoch": 0.6173879739563386,
+      "loss": 0.26405468583106995,
+      "loss_ce": 0.02125684544444084,
+      "loss_iou": 1.0224087238311768,
+      "loss_num": 0.2431640625,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 278626800,
+      "step": 1612
+    },
+    {
+      "epoch": 0.6177709689774032,
+      "grad_norm": 75.22629515793881,
+      "learning_rate": 5e-06,
+      "loss": 0.3944,
+      "num_input_tokens_seen": 278799464,
+      "step": 1613
+    },
+    {
+      "epoch": 0.6177709689774032,
+      "loss": 0.3575478494167328,
+      "loss_ce": 0.022464849054813385,
+      "loss_iou": 1.0131324529647827,
+      "loss_num": 0.3359375,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 278799464,
+      "step": 1613
+    },
+    {
+      "epoch": 0.618153963998468,
+      "grad_norm": 38.635798603481994,
+      "learning_rate": 5e-06,
+      "loss": 0.2966,
+      "num_input_tokens_seen": 278972672,
+      "step": 1614
+    },
+    {
+      "epoch": 0.618153963998468,
+      "loss": 0.30380117893218994,
+      "loss_ce": 0.021452531218528748,
+      "loss_iou": 1.0792896747589111,
+      "loss_num": 0.283203125,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 278972672,
+      "step": 1614
+    },
+    {
+      "epoch": 0.6185369590195328,
+      "grad_norm": 60.024768302190616,
+      "learning_rate": 5e-06,
+      "loss": 0.3133,
+      "num_input_tokens_seen": 279145368,
+      "step": 1615
+    },
+    {
+      "epoch": 0.6185369590195328,
+      "loss": 0.305372953414917,
+      "loss_ce": 0.020704954862594604,
+      "loss_iou": 1.1135112047195435,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 279145368,
+      "step": 1615
+    },
+    {
+      "epoch": 0.6189199540405975,
+      "grad_norm": 76.05768078594984,
+      "learning_rate": 5e-06,
+      "loss": 0.2857,
+      "num_input_tokens_seen": 279318816,
+      "step": 1616
+    },
+    {
+      "epoch": 0.6189199540405975,
+      "loss": 0.27048641443252563,
+      "loss_ce": 0.020974673330783844,
+      "loss_iou": 1.005634069442749,
+      "loss_num": 0.25,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 279318816,
+      "step": 1616
+    },
+    {
+      "epoch": 0.6193029490616622,
+      "grad_norm": 29.828337919803086,
+      "learning_rate": 5e-06,
+      "loss": 0.3173,
+      "num_input_tokens_seen": 279488520,
+      "step": 1617
+    },
+    {
+      "epoch": 0.6193029490616622,
+      "loss": 0.32703596353530884,
+      "loss_ce": 0.022470548748970032,
+      "loss_iou": 1.0088353157043457,
+      "loss_num": 0.3046875,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 279488520,
+      "step": 1617
+    },
+    {
+      "epoch": 0.6196859440827269,
+      "grad_norm": 59.639405587106204,
+      "learning_rate": 5e-06,
+      "loss": 0.4584,
+      "num_input_tokens_seen": 279661328,
+      "step": 1618
+    },
+    {
+      "epoch": 0.6196859440827269,
+      "loss": 0.4526071548461914,
+      "loss_ce": 0.021454807370901108,
+      "loss_iou": 1.0371252298355103,
+      "loss_num": 0.431640625,
+      "loss_xval": 0.431640625,
+      "num_input_tokens_seen": 279661328,
+      "step": 1618
+    },
+    {
+      "epoch": 0.6200689391037917,
+      "grad_norm": 74.61035306623141,
+      "learning_rate": 5e-06,
+      "loss": 0.3122,
+      "num_input_tokens_seen": 279834216,
+      "step": 1619
+    },
+    {
+      "epoch": 0.6200689391037917,
+      "loss": 0.2902352213859558,
+      "loss_ce": 0.021314337849617004,
+      "loss_iou": 1.079186201095581,
+      "loss_num": 0.26953125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 279834216,
+      "step": 1619
+    },
+    {
+      "epoch": 0.6204519341248563,
+      "grad_norm": 102.86484359917267,
+      "learning_rate": 5e-06,
+      "loss": 0.8238,
+      "num_input_tokens_seen": 280007224,
+      "step": 1620
+    },
+    {
+      "epoch": 0.6204519341248563,
+      "loss": 0.8232241868972778,
+      "loss_ce": 0.020001530647277832,
+      "loss_iou": 1.0411213636398315,
+      "loss_num": 0.8046875,
+      "loss_xval": 0.8046875,
+      "num_input_tokens_seen": 280007224,
+      "step": 1620
+    },
+    {
+      "epoch": 0.6208349291459211,
+      "grad_norm": 40.756967969013964,
+      "learning_rate": 5e-06,
+      "loss": 0.3399,
+      "num_input_tokens_seen": 280180296,
+      "step": 1621
+    },
+    {
+      "epoch": 0.6208349291459211,
+      "loss": 0.3413882255554199,
+      "loss_ce": 0.02095363475382328,
+      "loss_iou": 1.1663868427276611,
+      "loss_num": 0.3203125,
+      "loss_xval": 0.3203125,
+      "num_input_tokens_seen": 280180296,
+      "step": 1621
+    },
+    {
+      "epoch": 0.6212179241669858,
+      "grad_norm": 62.14917524690402,
+      "learning_rate": 5e-06,
+      "loss": 0.6053,
+      "num_input_tokens_seen": 280352808,
+      "step": 1622
+    },
+    {
+      "epoch": 0.6212179241669858,
+      "loss": 0.592242956161499,
+      "loss_ce": 0.021442197263240814,
+      "loss_iou": 1.039393663406372,
+      "loss_num": 0.5703125,
+      "loss_xval": 0.5703125,
+      "num_input_tokens_seen": 280352808,
+      "step": 1622
+    },
+    {
+      "epoch": 0.6216009191880506,
+      "grad_norm": 37.529966296005945,
+      "learning_rate": 5e-06,
+      "loss": 0.3617,
+      "num_input_tokens_seen": 280525816,
+      "step": 1623
+    },
+    {
+      "epoch": 0.6216009191880506,
+      "loss": 0.36118265986442566,
+      "loss_ce": 0.020606480538845062,
+      "loss_iou": 1.027377963066101,
+      "loss_num": 0.33984375,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 280525816,
+      "step": 1623
+    },
+    {
+      "epoch": 0.6219839142091153,
+      "grad_norm": 24.375650783154086,
+      "learning_rate": 5e-06,
+      "loss": 0.3293,
+      "num_input_tokens_seen": 280698568,
+      "step": 1624
+    },
+    {
+      "epoch": 0.6219839142091153,
+      "loss": 0.29702889919281006,
+      "loss_ce": 0.02029551938176155,
+      "loss_iou": 1.015905499458313,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 280698568,
+      "step": 1624
+    },
+    {
+      "epoch": 0.62236690923018,
+      "grad_norm": 47.15583885616379,
+      "learning_rate": 5e-06,
+      "loss": 0.4098,
+      "num_input_tokens_seen": 280871560,
+      "step": 1625
+    },
+    {
+      "epoch": 0.62236690923018,
+      "loss": 0.3579084873199463,
+      "loss_ce": 0.021116478368639946,
+      "loss_iou": 1.030195951461792,
+      "loss_num": 0.3359375,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 280871560,
+      "step": 1625
+    },
+    {
+      "epoch": 0.6227499042512448,
+      "grad_norm": 35.927884118207174,
+      "learning_rate": 5e-06,
+      "loss": 0.2328,
+      "num_input_tokens_seen": 281044376,
+      "step": 1626
+    },
+    {
+      "epoch": 0.6227499042512448,
+      "loss": 0.24369803071022034,
+      "loss_ce": 0.02079763263463974,
+      "loss_iou": 1.0101929903030396,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 281044376,
+      "step": 1626
+    },
+    {
+      "epoch": 0.6231328992723094,
+      "grad_norm": 38.372933248487016,
+      "learning_rate": 5e-06,
+      "loss": 0.3057,
+      "num_input_tokens_seen": 281217632,
+      "step": 1627
+    },
+    {
+      "epoch": 0.6231328992723094,
+      "loss": 0.34230750799179077,
+      "loss_ce": 0.022361237555742264,
+      "loss_iou": 1.1838070154190063,
+      "loss_num": 0.3203125,
+      "loss_xval": 0.3203125,
+      "num_input_tokens_seen": 281217632,
+      "step": 1627
+    },
+    {
+      "epoch": 0.6235158942933742,
+      "grad_norm": 36.05030553734535,
+      "learning_rate": 5e-06,
+      "loss": 0.2974,
+      "num_input_tokens_seen": 281390384,
+      "step": 1628
+    },
+    {
+      "epoch": 0.6235158942933742,
+      "loss": 0.23143544793128967,
+      "loss_ce": 0.021474510431289673,
+      "loss_iou": 1.023942470550537,
+      "loss_num": 0.2099609375,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 281390384,
+      "step": 1628
+    },
+    {
+      "epoch": 0.6238988893144389,
+      "grad_norm": 33.046635329876665,
+      "learning_rate": 5e-06,
+      "loss": 0.2272,
+      "num_input_tokens_seen": 281563440,
+      "step": 1629
+    },
+    {
+      "epoch": 0.6238988893144389,
+      "loss": 0.18551652133464813,
+      "loss_ce": 0.021942298859357834,
+      "loss_iou": 1.0375853776931763,
+      "loss_num": 0.1640625,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 281563440,
+      "step": 1629
+    },
+    {
+      "epoch": 0.6242818843355037,
+      "grad_norm": 32.33134558376756,
+      "learning_rate": 5e-06,
+      "loss": 0.234,
+      "num_input_tokens_seen": 281736640,
+      "step": 1630
+    },
+    {
+      "epoch": 0.6242818843355037,
+      "loss": 0.2025236189365387,
+      "loss_ce": 0.0236295685172081,
+      "loss_iou": 1.0388567447662354,
+      "loss_num": 0.1787109375,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 281736640,
+      "step": 1630
+    },
+    {
+      "epoch": 0.6246648793565683,
+      "grad_norm": 38.36817956911151,
+      "learning_rate": 5e-06,
+      "loss": 0.2175,
+      "num_input_tokens_seen": 281909704,
+      "step": 1631
+    },
+    {
+      "epoch": 0.6246648793565683,
+      "loss": 0.18636493384838104,
+      "loss_ce": 0.022302431985735893,
+      "loss_iou": 1.0021567344665527,
+      "loss_num": 0.1640625,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 281909704,
+      "step": 1631
+    },
+    {
+      "epoch": 0.6250478743776331,
+      "grad_norm": 30.744052702819378,
+      "learning_rate": 5e-06,
+      "loss": 0.2626,
+      "num_input_tokens_seen": 282082792,
+      "step": 1632
+    },
+    {
+      "epoch": 0.6250478743776331,
+      "loss": 0.254894882440567,
+      "loss_ce": 0.021313339471817017,
+      "loss_iou": 1.093733549118042,
+      "loss_num": 0.2333984375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 282082792,
+      "step": 1632
+    },
+    {
+      "epoch": 0.6254308693986979,
+      "grad_norm": 16.26735525242088,
+      "learning_rate": 5e-06,
+      "loss": 0.2333,
+      "num_input_tokens_seen": 282252504,
+      "step": 1633
+    },
+    {
+      "epoch": 0.6254308693986979,
+      "loss": 0.24430033564567566,
+      "loss_ce": 0.021766144782304764,
+      "loss_iou": 1.007523775100708,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 282252504,
+      "step": 1633
+    },
+    {
+      "epoch": 0.6258138644197625,
+      "grad_norm": 26.344668297926265,
+      "learning_rate": 5e-06,
+      "loss": 0.2552,
+      "num_input_tokens_seen": 282425272,
+      "step": 1634
+    },
+    {
+      "epoch": 0.6258138644197625,
+      "loss": 0.2294924259185791,
+      "loss_ce": 0.0196535587310791,
+      "loss_iou": 1.0868526697158813,
+      "loss_num": 0.2099609375,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 282425272,
+      "step": 1634
+    },
+    {
+      "epoch": 0.6261968594408273,
+      "grad_norm": 25.93744116888221,
+      "learning_rate": 5e-06,
+      "loss": 0.2068,
+      "num_input_tokens_seen": 282598016,
+      "step": 1635
+    },
+    {
+      "epoch": 0.6261968594408273,
+      "loss": 0.17579901218414307,
+      "loss_ce": 0.021258002147078514,
+      "loss_iou": 1.0365262031555176,
+      "loss_num": 0.154296875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 282598016,
+      "step": 1635
+    },
+    {
+      "epoch": 0.626579854461892,
+      "grad_norm": 26.145387506131154,
+      "learning_rate": 5e-06,
+      "loss": 0.2339,
+      "num_input_tokens_seen": 282770952,
+      "step": 1636
+    },
+    {
+      "epoch": 0.626579854461892,
+      "loss": 0.19785335659980774,
+      "loss_ce": 0.021827977150678635,
+      "loss_iou": 1.0252678394317627,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 282770952,
+      "step": 1636
+    },
+    {
+      "epoch": 0.6269628494829568,
+      "grad_norm": 30.47869796105112,
+      "learning_rate": 5e-06,
+      "loss": 0.2366,
+      "num_input_tokens_seen": 282944160,
+      "step": 1637
+    },
+    {
+      "epoch": 0.6269628494829568,
+      "loss": 0.21427859365940094,
+      "loss_ce": 0.02043093368411064,
+      "loss_iou": 1.0064401626586914,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 282944160,
+      "step": 1637
+    },
+    {
+      "epoch": 0.6273458445040214,
+      "grad_norm": 32.82926948027278,
+      "learning_rate": 5e-06,
+      "loss": 0.2111,
+      "num_input_tokens_seen": 283116960,
+      "step": 1638
+    },
+    {
+      "epoch": 0.6273458445040214,
+      "loss": 0.2528603971004486,
+      "loss_ce": 0.02202543243765831,
+      "loss_iou": 1.0393564701080322,
+      "loss_num": 0.23046875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 283116960,
+      "step": 1638
+    },
+    {
+      "epoch": 0.6277288395250862,
+      "grad_norm": 25.447593959033043,
+      "learning_rate": 5e-06,
+      "loss": 0.2319,
+      "num_input_tokens_seen": 283290080,
+      "step": 1639
+    },
+    {
+      "epoch": 0.6277288395250862,
+      "loss": 0.20689032971858978,
+      "loss_ce": 0.019756542518734932,
+      "loss_iou": 0.8765753507614136,
+      "loss_num": 0.1875,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 283290080,
+      "step": 1639
+    },
+    {
+      "epoch": 0.6281118345461509,
+      "grad_norm": 22.74771257097244,
+      "learning_rate": 5e-06,
+      "loss": 0.2312,
+      "num_input_tokens_seen": 283463224,
+      "step": 1640
+    },
+    {
+      "epoch": 0.6281118345461509,
+      "loss": 0.19958898425102234,
+      "loss_ce": 0.021183211356401443,
+      "loss_iou": 1.0254478454589844,
+      "loss_num": 0.1787109375,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 283463224,
+      "step": 1640
+    },
+    {
+      "epoch": 0.6284948295672156,
+      "grad_norm": 26.714942902917763,
+      "learning_rate": 5e-06,
+      "loss": 0.1812,
+      "num_input_tokens_seen": 283636248,
+      "step": 1641
+    },
+    {
+      "epoch": 0.6284948295672156,
+      "loss": 0.1777523159980774,
+      "loss_ce": 0.021990593522787094,
+      "loss_iou": 1.0114291906356812,
+      "loss_num": 0.15625,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 283636248,
+      "step": 1641
+    },
+    {
+      "epoch": 0.6288778245882803,
+      "grad_norm": 22.714348404617184,
+      "learning_rate": 5e-06,
+      "loss": 0.2647,
+      "num_input_tokens_seen": 283809472,
+      "step": 1642
+    },
+    {
+      "epoch": 0.6288778245882803,
+      "loss": 0.26512426137924194,
+      "loss_ce": 0.020861556753516197,
+      "loss_iou": 1.3547765016555786,
+      "loss_num": 0.244140625,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 283809472,
+      "step": 1642
+    },
+    {
+      "epoch": 0.6292608196093451,
+      "grad_norm": 23.311314197616092,
+      "learning_rate": 5e-06,
+      "loss": 0.2186,
+      "num_input_tokens_seen": 283982360,
+      "step": 1643
+    },
+    {
+      "epoch": 0.6292608196093451,
+      "loss": 0.24859797954559326,
+      "loss_ce": 0.02255428209900856,
+      "loss_iou": 1.008880376815796,
+      "loss_num": 0.2255859375,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 283982360,
+      "step": 1643
+    },
+    {
+      "epoch": 0.6296438146304099,
+      "grad_norm": 23.892541682634135,
+      "learning_rate": 5e-06,
+      "loss": 0.2183,
+      "num_input_tokens_seen": 284155680,
+      "step": 1644
+    },
+    {
+      "epoch": 0.6296438146304099,
+      "loss": 0.23071157932281494,
+      "loss_ce": 0.022337544709444046,
+      "loss_iou": 3.700438976287842,
+      "loss_num": 0.2080078125,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 284155680,
+      "step": 1644
+    },
+    {
+      "epoch": 0.6300268096514745,
+      "grad_norm": 27.523460965028566,
+      "learning_rate": 5e-06,
+      "loss": 0.1842,
+      "num_input_tokens_seen": 284325152,
+      "step": 1645
+    },
+    {
+      "epoch": 0.6300268096514745,
+      "loss": 0.16071315109729767,
+      "loss_ce": 0.021614033728837967,
+      "loss_iou": 1.0021328926086426,
+      "loss_num": 0.138671875,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 284325152,
+      "step": 1645
+    },
+    {
+      "epoch": 0.6304098046725393,
+      "grad_norm": 25.977153537205968,
+      "learning_rate": 5e-06,
+      "loss": 0.182,
+      "num_input_tokens_seen": 284498248,
+      "step": 1646
+    },
+    {
+      "epoch": 0.6304098046725393,
+      "loss": 0.18294882774353027,
+      "loss_ce": 0.021693941205739975,
+      "loss_iou": 0.9968886375427246,
+      "loss_num": 0.1611328125,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 284498248,
+      "step": 1646
+    },
+    {
+      "epoch": 0.630792799693604,
+      "grad_norm": 17.89546757301777,
+      "learning_rate": 5e-06,
+      "loss": 0.2154,
+      "num_input_tokens_seen": 284671096,
+      "step": 1647
+    },
+    {
+      "epoch": 0.630792799693604,
+      "loss": 0.18741793930530548,
+      "loss_ce": 0.01932712085545063,
+      "loss_iou": 0.891245424747467,
+      "loss_num": 0.16796875,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 284671096,
+      "step": 1647
+    },
+    {
+      "epoch": 0.6311757947146687,
+      "grad_norm": 24.095337657626278,
+      "learning_rate": 5e-06,
+      "loss": 0.1844,
+      "num_input_tokens_seen": 284844192,
+      "step": 1648
+    },
+    {
+      "epoch": 0.6311757947146687,
+      "loss": 0.20032504200935364,
+      "loss_ce": 0.02191927656531334,
+      "loss_iou": 1.0907624959945679,
+      "loss_num": 0.1787109375,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 284844192,
+      "step": 1648
+    },
+    {
+      "epoch": 0.6315587897357334,
+      "grad_norm": 36.76351481502342,
+      "learning_rate": 5e-06,
+      "loss": 0.2067,
+      "num_input_tokens_seen": 285016912,
+      "step": 1649
+    },
+    {
+      "epoch": 0.6315587897357334,
+      "loss": 0.2016732096672058,
+      "loss_ce": 0.021680543199181557,
+      "loss_iou": 4.320743083953857,
+      "loss_num": 0.1796875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 285016912,
+      "step": 1649
+    },
+    {
+      "epoch": 0.6319417847567982,
+      "grad_norm": 27.485742269959434,
+      "learning_rate": 5e-06,
+      "loss": 0.207,
+      "num_input_tokens_seen": 285189800,
+      "step": 1650
+    },
+    {
+      "epoch": 0.6319417847567982,
+      "loss": 0.2237074077129364,
+      "loss_ce": 0.020399296656250954,
+      "loss_iou": 1.0881264209747314,
+      "loss_num": 0.203125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 285189800,
+      "step": 1650
+    },
+    {
+      "epoch": 0.6323247797778628,
+      "grad_norm": 16.255869079745406,
+      "learning_rate": 5e-06,
+      "loss": 0.1965,
+      "num_input_tokens_seen": 285363000,
+      "step": 1651
+    },
+    {
+      "epoch": 0.6323247797778628,
+      "loss": 0.17966032028198242,
+      "loss_ce": 0.022006506100296974,
+      "loss_iou": 1.1113789081573486,
+      "loss_num": 0.1572265625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 285363000,
+      "step": 1651
+    },
+    {
+      "epoch": 0.6327077747989276,
+      "grad_norm": 16.03206054029721,
+      "learning_rate": 5e-06,
+      "loss": 0.207,
+      "num_input_tokens_seen": 285536200,
+      "step": 1652
+    },
+    {
+      "epoch": 0.6327077747989276,
+      "loss": 0.21859735250473022,
+      "loss_ce": 0.021697930991649628,
+      "loss_iou": 1.0353161096572876,
+      "loss_num": 0.197265625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 285536200,
+      "step": 1652
+    },
+    {
+      "epoch": 0.6330907698199923,
+      "grad_norm": 24.418809592359533,
+      "learning_rate": 5e-06,
+      "loss": 0.1921,
+      "num_input_tokens_seen": 285708704,
+      "step": 1653
+    },
+    {
+      "epoch": 0.6330907698199923,
+      "loss": 0.25326675176620483,
+      "loss_ce": 0.02151626907289028,
+      "loss_iou": 1.0436711311340332,
+      "loss_num": 0.2314453125,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 285708704,
+      "step": 1653
+    },
+    {
+      "epoch": 0.6334737648410571,
+      "grad_norm": 37.11570855501896,
+      "learning_rate": 5e-06,
+      "loss": 0.3043,
+      "num_input_tokens_seen": 285881616,
+      "step": 1654
+    },
+    {
+      "epoch": 0.6334737648410571,
+      "loss": 0.3383890390396118,
+      "loss_ce": 0.020762072876095772,
+      "loss_iou": 1.2734043598175049,
+      "loss_num": 0.318359375,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 285881616,
+      "step": 1654
+    },
+    {
+      "epoch": 0.6338567598621218,
+      "grad_norm": 25.86617718679928,
+      "learning_rate": 5e-06,
+      "loss": 0.1773,
+      "num_input_tokens_seen": 286054784,
+      "step": 1655
+    },
+    {
+      "epoch": 0.6338567598621218,
+      "loss": 0.12519463896751404,
+      "loss_ce": 0.019359681755304337,
+      "loss_iou": 1.00177001953125,
+      "loss_num": 0.10595703125,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 286054784,
+      "step": 1655
+    },
+    {
+      "epoch": 0.6342397548831865,
+      "grad_norm": 28.462609982513673,
+      "learning_rate": 5e-06,
+      "loss": 0.2588,
+      "num_input_tokens_seen": 286227912,
+      "step": 1656
+    },
+    {
+      "epoch": 0.6342397548831865,
+      "loss": 0.25157588720321655,
+      "loss_ce": 0.02220574952661991,
+      "loss_iou": 1.0223424434661865,
+      "loss_num": 0.2294921875,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 286227912,
+      "step": 1656
+    },
+    {
+      "epoch": 0.6346227499042513,
+      "grad_norm": 30.96868167947015,
+      "learning_rate": 5e-06,
+      "loss": 0.207,
+      "num_input_tokens_seen": 286401312,
+      "step": 1657
+    },
+    {
+      "epoch": 0.6346227499042513,
+      "loss": 0.23401233553886414,
+      "loss_ce": 0.020938601344823837,
+      "loss_iou": 1.0035679340362549,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 286401312,
+      "step": 1657
+    },
+    {
+      "epoch": 0.6350057449253159,
+      "grad_norm": 44.1654473720544,
+      "learning_rate": 5e-06,
+      "loss": 0.2216,
+      "num_input_tokens_seen": 286574152,
+      "step": 1658
+    },
+    {
+      "epoch": 0.6350057449253159,
+      "loss": 0.26512664556503296,
+      "loss_ce": 0.018361497670412064,
+      "loss_iou": 1.0043344497680664,
+      "loss_num": 0.2470703125,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 286574152,
+      "step": 1658
+    },
+    {
+      "epoch": 0.6353887399463807,
+      "grad_norm": 16.50700974628968,
+      "learning_rate": 5e-06,
+      "loss": 0.1528,
+      "num_input_tokens_seen": 286746752,
+      "step": 1659
+    },
+    {
+      "epoch": 0.6353887399463807,
+      "loss": 0.1298852562904358,
+      "loss_ce": 0.02121216617524624,
+      "loss_iou": 1.0161259174346924,
+      "loss_num": 0.10888671875,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 286746752,
+      "step": 1659
+    },
+    {
+      "epoch": 0.6357717349674454,
+      "grad_norm": 21.526947475498805,
+      "learning_rate": 5e-06,
+      "loss": 0.2173,
+      "num_input_tokens_seen": 286919896,
+      "step": 1660
+    },
+    {
+      "epoch": 0.6357717349674454,
+      "loss": 0.1910654902458191,
+      "loss_ce": 0.02114361897110939,
+      "loss_iou": 1.0035732984542847,
+      "loss_num": 0.169921875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 286919896,
+      "step": 1660
+    },
+    {
+      "epoch": 0.6361547299885102,
+      "grad_norm": 23.07709534576469,
+      "learning_rate": 5e-06,
+      "loss": 0.2024,
+      "num_input_tokens_seen": 287092832,
+      "step": 1661
+    },
+    {
+      "epoch": 0.6361547299885102,
+      "loss": 0.20819546282291412,
+      "loss_ce": 0.021122703328728676,
+      "loss_iou": 1.021871566772461,
+      "loss_num": 0.1875,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 287092832,
+      "step": 1661
+    },
+    {
+      "epoch": 0.6365377250095748,
+      "grad_norm": 31.348648113259728,
+      "learning_rate": 5e-06,
+      "loss": 0.2227,
+      "num_input_tokens_seen": 287266000,
+      "step": 1662
+    },
+    {
+      "epoch": 0.6365377250095748,
+      "loss": 0.1994139552116394,
+      "loss_ce": 0.020336810499429703,
+      "loss_iou": 1.0821785926818848,
+      "loss_num": 0.1787109375,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 287266000,
+      "step": 1662
+    },
+    {
+      "epoch": 0.6369207200306396,
+      "grad_norm": 39.429898446331514,
+      "learning_rate": 5e-06,
+      "loss": 0.1862,
+      "num_input_tokens_seen": 287438760,
+      "step": 1663
+    },
+    {
+      "epoch": 0.6369207200306396,
+      "loss": 0.19204115867614746,
+      "loss_ce": 0.021081678569316864,
+      "loss_iou": 0.9045433402061462,
+      "loss_num": 0.1708984375,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 287438760,
+      "step": 1663
+    },
+    {
+      "epoch": 0.6373037150517044,
+      "grad_norm": 24.579819652640182,
+      "learning_rate": 5e-06,
+      "loss": 0.2217,
+      "num_input_tokens_seen": 287611488,
+      "step": 1664
+    },
+    {
+      "epoch": 0.6373037150517044,
+      "loss": 0.20055729150772095,
+      "loss_ce": 0.020930826663970947,
+      "loss_iou": 1.0237832069396973,
+      "loss_num": 0.1796875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 287611488,
+      "step": 1664
+    },
+    {
+      "epoch": 0.637686710072769,
+      "grad_norm": 18.916837951616397,
+      "learning_rate": 5e-06,
+      "loss": 0.1694,
+      "num_input_tokens_seen": 287784648,
+      "step": 1665
+    },
+    {
+      "epoch": 0.637686710072769,
+      "loss": 0.18160784244537354,
+      "loss_ce": 0.021115906536579132,
+      "loss_iou": 1.0030603408813477,
+      "loss_num": 0.16015625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 287784648,
+      "step": 1665
+    },
+    {
+      "epoch": 0.6380697050938338,
+      "grad_norm": 28.85645059721232,
+      "learning_rate": 5e-06,
+      "loss": 0.2832,
+      "num_input_tokens_seen": 287957632,
+      "step": 1666
+    },
+    {
+      "epoch": 0.6380697050938338,
+      "loss": 0.37453964352607727,
+      "loss_ce": 0.01998642459511757,
+      "loss_iou": 2.200084686279297,
+      "loss_num": 0.35546875,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 287957632,
+      "step": 1666
+    },
+    {
+      "epoch": 0.6384527001148985,
+      "grad_norm": 28.787803855370317,
+      "learning_rate": 5e-06,
+      "loss": 0.1736,
+      "num_input_tokens_seen": 288130336,
+      "step": 1667
+    },
+    {
+      "epoch": 0.6384527001148985,
+      "loss": 0.1775747835636139,
+      "loss_ce": 0.02156892418861389,
+      "loss_iou": 1.0389750003814697,
+      "loss_num": 0.15625,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 288130336,
+      "step": 1667
+    },
+    {
+      "epoch": 0.6388356951359633,
+      "grad_norm": 29.94586825692482,
+      "learning_rate": 5e-06,
+      "loss": 0.2187,
+      "num_input_tokens_seen": 288303472,
+      "step": 1668
+    },
+    {
+      "epoch": 0.6388356951359633,
+      "loss": 0.23204870522022247,
+      "loss_ce": 0.020561883226037025,
+      "loss_iou": 0.9820910692214966,
+      "loss_num": 0.2119140625,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 288303472,
+      "step": 1668
+    },
+    {
+      "epoch": 0.6392186901570279,
+      "grad_norm": 27.772451317795284,
+      "learning_rate": 5e-06,
+      "loss": 0.2079,
+      "num_input_tokens_seen": 288476312,
+      "step": 1669
+    },
+    {
+      "epoch": 0.6392186901570279,
+      "loss": 0.1951129138469696,
+      "loss_ce": 0.020918574184179306,
+      "loss_iou": 1.0338059663772583,
+      "loss_num": 0.173828125,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 288476312,
+      "step": 1669
+    },
+    {
+      "epoch": 0.6396016851780927,
+      "grad_norm": 27.280457085926844,
+      "learning_rate": 5e-06,
+      "loss": 0.2288,
+      "num_input_tokens_seen": 288649312,
+      "step": 1670
+    },
+    {
+      "epoch": 0.6396016851780927,
+      "loss": 0.27583760023117065,
+      "loss_ce": 0.020832736045122147,
+      "loss_iou": 1.078195333480835,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 288649312,
+      "step": 1670
+    },
+    {
+      "epoch": 0.6399846801991574,
+      "grad_norm": 32.09006951719356,
+      "learning_rate": 5e-06,
+      "loss": 0.2056,
+      "num_input_tokens_seen": 288821832,
+      "step": 1671
+    },
+    {
+      "epoch": 0.6399846801991574,
+      "loss": 0.23407715559005737,
+      "loss_ce": 0.019111350178718567,
+      "loss_iou": 0.9695178270339966,
+      "loss_num": 0.21484375,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 288821832,
+      "step": 1671
+    },
+    {
+      "epoch": 0.6403676752202221,
+      "grad_norm": 41.748672694407006,
+      "learning_rate": 5e-06,
+      "loss": 0.2483,
+      "num_input_tokens_seen": 288991464,
+      "step": 1672
+    },
+    {
+      "epoch": 0.6403676752202221,
+      "loss": 0.21558049321174622,
+      "loss_ce": 0.02185489796102047,
+      "loss_iou": 1.0301659107208252,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 288991464,
+      "step": 1672
+    },
+    {
+      "epoch": 0.6407506702412868,
+      "grad_norm": 39.87815093937564,
+      "learning_rate": 5e-06,
+      "loss": 0.2109,
+      "num_input_tokens_seen": 289164728,
+      "step": 1673
+    },
+    {
+      "epoch": 0.6407506702412868,
+      "loss": 0.2459956407546997,
+      "loss_ce": 0.021935580298304558,
+      "loss_iou": 1.057338833808899,
+      "loss_num": 0.2236328125,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 289164728,
+      "step": 1673
+    },
+    {
+      "epoch": 0.6411336652623516,
+      "grad_norm": 35.7630799527807,
+      "learning_rate": 5e-06,
+      "loss": 0.2092,
+      "num_input_tokens_seen": 289337528,
+      "step": 1674
+    },
+    {
+      "epoch": 0.6411336652623516,
+      "loss": 0.22301757335662842,
+      "loss_ce": 0.020319823175668716,
+      "loss_iou": 1.0427360534667969,
+      "loss_num": 0.203125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 289337528,
+      "step": 1674
+    },
+    {
+      "epoch": 0.6415166602834164,
+      "grad_norm": 28.352158649635488,
+      "learning_rate": 5e-06,
+      "loss": 0.191,
+      "num_input_tokens_seen": 289510168,
+      "step": 1675
+    },
+    {
+      "epoch": 0.6415166602834164,
+      "loss": 0.18749476969242096,
+      "loss_ce": 0.021814826875925064,
+      "loss_iou": 1.0057984590530396,
+      "loss_num": 0.166015625,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 289510168,
+      "step": 1675
+    },
+    {
+      "epoch": 0.641899655304481,
+      "grad_norm": 33.214192015695716,
+      "learning_rate": 5e-06,
+      "loss": 0.1927,
+      "num_input_tokens_seen": 289683288,
+      "step": 1676
+    },
+    {
+      "epoch": 0.641899655304481,
+      "loss": 0.19039714336395264,
+      "loss_ce": 0.019376643002033234,
+      "loss_iou": 0.9672365784645081,
+      "loss_num": 0.1708984375,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 289683288,
+      "step": 1676
+    },
+    {
+      "epoch": 0.6422826503255458,
+      "grad_norm": 27.11509876909704,
+      "learning_rate": 5e-06,
+      "loss": 0.1772,
+      "num_input_tokens_seen": 289856024,
+      "step": 1677
+    },
+    {
+      "epoch": 0.6422826503255458,
+      "loss": 0.22653698921203613,
+      "loss_ce": 0.020909534767270088,
+      "loss_iou": 1.023066759109497,
+      "loss_num": 0.2060546875,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 289856024,
+      "step": 1677
+    },
+    {
+      "epoch": 0.6426656453466105,
+      "grad_norm": 33.94245069034829,
+      "learning_rate": 5e-06,
+      "loss": 0.2166,
+      "num_input_tokens_seen": 290028792,
+      "step": 1678
+    },
+    {
+      "epoch": 0.6426656453466105,
+      "loss": 0.21347205340862274,
+      "loss_ce": 0.02047889307141304,
+      "loss_iou": 1.0252187252044678,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 290028792,
+      "step": 1678
+    },
+    {
+      "epoch": 0.6430486403676752,
+      "grad_norm": 36.118953789569794,
+      "learning_rate": 5e-06,
+      "loss": 0.256,
+      "num_input_tokens_seen": 290202104,
+      "step": 1679
+    },
+    {
+      "epoch": 0.6430486403676752,
+      "loss": 0.35787707567214966,
+      "loss_ce": 0.01974230445921421,
+      "loss_iou": 1.1241517066955566,
+      "loss_num": 0.337890625,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 290202104,
+      "step": 1679
+    },
+    {
+      "epoch": 0.6434316353887399,
+      "grad_norm": 34.75509918571788,
+      "learning_rate": 5e-06,
+      "loss": 0.2249,
+      "num_input_tokens_seen": 290374728,
+      "step": 1680
+    },
+    {
+      "epoch": 0.6434316353887399,
+      "loss": 0.21840673685073853,
+      "loss_ce": 0.021049557253718376,
+      "loss_iou": 1.0462462902069092,
+      "loss_num": 0.197265625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 290374728,
+      "step": 1680
+    },
+    {
+      "epoch": 0.6438146304098047,
+      "grad_norm": 29.673059916538776,
+      "learning_rate": 5e-06,
+      "loss": 0.2113,
+      "num_input_tokens_seen": 290547696,
+      "step": 1681
+    },
+    {
+      "epoch": 0.6438146304098047,
+      "loss": 0.23977741599082947,
+      "loss_ce": 0.023041583597660065,
+      "loss_iou": 1.0123975276947021,
+      "loss_num": 0.216796875,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 290547696,
+      "step": 1681
+    },
+    {
+      "epoch": 0.6441976254308694,
+      "grad_norm": 29.471975842247414,
+      "learning_rate": 5e-06,
+      "loss": 0.2328,
+      "num_input_tokens_seen": 290717232,
+      "step": 1682
+    },
+    {
+      "epoch": 0.6441976254308694,
+      "loss": 0.18632182478904724,
+      "loss_ce": 0.020916547626256943,
+      "loss_iou": 1.036426305770874,
+      "loss_num": 0.1650390625,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 290717232,
+      "step": 1682
+    },
+    {
+      "epoch": 0.6445806204519341,
+      "grad_norm": 22.77480070318721,
+      "learning_rate": 5e-06,
+      "loss": 0.1895,
+      "num_input_tokens_seen": 290890248,
+      "step": 1683
+    },
+    {
+      "epoch": 0.6445806204519341,
+      "loss": 0.13573291897773743,
+      "loss_ce": 0.022634778171777725,
+      "loss_iou": 1.00262451171875,
+      "loss_num": 0.11328125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 290890248,
+      "step": 1683
+    },
+    {
+      "epoch": 0.6449636154729989,
+      "grad_norm": 35.551238250964886,
+      "learning_rate": 5e-06,
+      "loss": 0.2259,
+      "num_input_tokens_seen": 291063376,
+      "step": 1684
+    },
+    {
+      "epoch": 0.6449636154729989,
+      "loss": 0.25219807028770447,
+      "loss_ce": 0.020508628338575363,
+      "loss_iou": 1.0799319744110107,
+      "loss_num": 0.2314453125,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 291063376,
+      "step": 1684
+    },
+    {
+      "epoch": 0.6453466104940636,
+      "grad_norm": 29.99433453945507,
+      "learning_rate": 5e-06,
+      "loss": 0.1988,
+      "num_input_tokens_seen": 291236424,
+      "step": 1685
+    },
+    {
+      "epoch": 0.6453466104940636,
+      "loss": 0.18930909037590027,
+      "loss_ce": 0.02018067240715027,
+      "loss_iou": 1.0057191848754883,
+      "loss_num": 0.1689453125,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 291236424,
+      "step": 1685
+    },
+    {
+      "epoch": 0.6457296055151283,
+      "grad_norm": 37.8492513803729,
+      "learning_rate": 5e-06,
+      "loss": 0.2212,
+      "num_input_tokens_seen": 291409312,
+      "step": 1686
+    },
+    {
+      "epoch": 0.6457296055151283,
+      "loss": 0.24024012684822083,
+      "loss_ce": 0.020757704973220825,
+      "loss_iou": 0.892715334892273,
+      "loss_num": 0.2197265625,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 291409312,
+      "step": 1686
+    },
+    {
+      "epoch": 0.646112600536193,
+      "grad_norm": 30.779299488203282,
+      "learning_rate": 5e-06,
+      "loss": 0.2733,
+      "num_input_tokens_seen": 291582360,
+      "step": 1687
+    },
+    {
+      "epoch": 0.646112600536193,
+      "loss": 0.32749754190444946,
+      "loss_ce": 0.02110106498003006,
+      "loss_iou": 1.1312453746795654,
+      "loss_num": 0.306640625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 291582360,
+      "step": 1687
+    },
+    {
+      "epoch": 0.6464955955572578,
+      "grad_norm": 32.70923023423609,
+      "learning_rate": 5e-06,
+      "loss": 0.2588,
+      "num_input_tokens_seen": 291755264,
+      "step": 1688
+    },
+    {
+      "epoch": 0.6464955955572578,
+      "loss": 0.3143477141857147,
+      "loss_ce": 0.02040240727365017,
+      "loss_iou": 1.493055820465088,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 291755264,
+      "step": 1688
+    },
+    {
+      "epoch": 0.6468785905783225,
+      "grad_norm": 27.306927819272115,
+      "learning_rate": 5e-06,
+      "loss": 0.2093,
+      "num_input_tokens_seen": 291928456,
+      "step": 1689
+    },
+    {
+      "epoch": 0.6468785905783225,
+      "loss": 0.20340877771377563,
+      "loss_ce": 0.022073332220315933,
+      "loss_iou": 1.0076968669891357,
+      "loss_num": 0.181640625,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 291928456,
+      "step": 1689
+    },
+    {
+      "epoch": 0.6472615855993872,
+      "grad_norm": 23.322346668501687,
+      "learning_rate": 5e-06,
+      "loss": 0.2521,
+      "num_input_tokens_seen": 292101520,
+      "step": 1690
+    },
+    {
+      "epoch": 0.6472615855993872,
+      "loss": 0.20817424356937408,
+      "loss_ce": 0.02049114555120468,
+      "loss_iou": 1.0999984741210938,
+      "loss_num": 0.1875,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 292101520,
+      "step": 1690
+    },
+    {
+      "epoch": 0.6476445806204519,
+      "grad_norm": 23.571239380389898,
+      "learning_rate": 5e-06,
+      "loss": 0.201,
+      "num_input_tokens_seen": 292274584,
+      "step": 1691
+    },
+    {
+      "epoch": 0.6476445806204519,
+      "loss": 0.2165507972240448,
+      "loss_ce": 0.022703150287270546,
+      "loss_iou": 1.0310851335525513,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 292274584,
+      "step": 1691
+    },
+    {
+      "epoch": 0.6480275756415167,
+      "grad_norm": 28.042909165345502,
+      "learning_rate": 5e-06,
+      "loss": 0.1859,
+      "num_input_tokens_seen": 292448056,
+      "step": 1692
+    },
+    {
+      "epoch": 0.6480275756415167,
+      "loss": 0.16891542077064514,
+      "loss_ce": 0.02072206139564514,
+      "loss_iou": 1.0096759796142578,
+      "loss_num": 0.1484375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 292448056,
+      "step": 1692
+    },
+    {
+      "epoch": 0.6484105706625813,
+      "grad_norm": 26.77857807766684,
+      "learning_rate": 5e-06,
+      "loss": 0.2242,
+      "num_input_tokens_seen": 292621320,
+      "step": 1693
+    },
+    {
+      "epoch": 0.6484105706625813,
+      "loss": 0.2604060173034668,
+      "loss_ce": 0.020598899573087692,
+      "loss_iou": 1.0400781631469727,
+      "loss_num": 0.240234375,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 292621320,
+      "step": 1693
+    },
+    {
+      "epoch": 0.6487935656836461,
+      "grad_norm": 29.361275525562398,
+      "learning_rate": 5e-06,
+      "loss": 0.2229,
+      "num_input_tokens_seen": 292794472,
+      "step": 1694
+    },
+    {
+      "epoch": 0.6487935656836461,
+      "loss": 0.21219977736473083,
+      "loss_ce": 0.02085457369685173,
+      "loss_iou": 1.0477708578109741,
+      "loss_num": 0.19140625,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 292794472,
+      "step": 1694
+    },
+    {
+      "epoch": 0.6491765607047109,
+      "grad_norm": 37.07073880078034,
+      "learning_rate": 5e-06,
+      "loss": 0.2105,
+      "num_input_tokens_seen": 292967456,
+      "step": 1695
+    },
+    {
+      "epoch": 0.6491765607047109,
+      "loss": 0.27898040413856506,
+      "loss_ce": 0.02061857283115387,
+      "loss_iou": 1.0436944961547852,
+      "loss_num": 0.2578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 292967456,
+      "step": 1695
+    },
+    {
+      "epoch": 0.6495595557257756,
+      "grad_norm": 45.18737609991261,
+      "learning_rate": 5e-06,
+      "loss": 0.3145,
+      "num_input_tokens_seen": 293140272,
+      "step": 1696
+    },
+    {
+      "epoch": 0.6495595557257756,
+      "loss": 0.38537001609802246,
+      "loss_ce": 0.020746011286973953,
+      "loss_iou": 1.120348572731018,
+      "loss_num": 0.365234375,
+      "loss_xval": 0.365234375,
+      "num_input_tokens_seen": 293140272,
+      "step": 1696
+    },
+    {
+      "epoch": 0.6499425507468403,
+      "grad_norm": 23.833577214188534,
+      "learning_rate": 5e-06,
+      "loss": 0.2291,
+      "num_input_tokens_seen": 293313096,
+      "step": 1697
+    },
+    {
+      "epoch": 0.6499425507468403,
+      "loss": 0.1783304065465927,
+      "loss_ce": 0.021714193746447563,
+      "loss_iou": 1.012166142463684,
+      "loss_num": 0.15625,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 293313096,
+      "step": 1697
+    },
+    {
+      "epoch": 0.650325545767905,
+      "grad_norm": 28.94721750502272,
+      "learning_rate": 5e-06,
+      "loss": 0.1924,
+      "num_input_tokens_seen": 293485568,
+      "step": 1698
+    },
+    {
+      "epoch": 0.650325545767905,
+      "loss": 0.20731818675994873,
+      "loss_ce": 0.020092826336622238,
+      "loss_iou": 1.0383963584899902,
+      "loss_num": 0.1875,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 293485568,
+      "step": 1698
+    },
+    {
+      "epoch": 0.6507085407889698,
+      "grad_norm": 26.39736723406935,
+      "learning_rate": 5e-06,
+      "loss": 0.1738,
+      "num_input_tokens_seen": 293658400,
+      "step": 1699
+    },
+    {
+      "epoch": 0.6507085407889698,
+      "loss": 0.1380467712879181,
+      "loss_ce": 0.019028211012482643,
+      "loss_iou": 1.0051178932189941,
+      "loss_num": 0.119140625,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 293658400,
+      "step": 1699
+    },
+    {
+      "epoch": 0.6510915358100344,
+      "grad_norm": 34.994208015125395,
+      "learning_rate": 5e-06,
+      "loss": 0.1486,
+      "num_input_tokens_seen": 293831368,
+      "step": 1700
+    },
+    {
+      "epoch": 0.6510915358100344,
+      "loss": 0.1436193585395813,
+      "loss_ce": 0.021365948021411896,
+      "loss_iou": 1.000901222229004,
+      "loss_num": 0.1220703125,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 293831368,
+      "step": 1700
+    },
+    {
+      "epoch": 0.6514745308310992,
+      "grad_norm": 40.207505559738514,
+      "learning_rate": 5e-06,
+      "loss": 0.2223,
+      "num_input_tokens_seen": 294004096,
+      "step": 1701
+    },
+    {
+      "epoch": 0.6514745308310992,
+      "loss": 0.1822366863489151,
+      "loss_ce": 0.0209817998111248,
+      "loss_iou": 1.0126309394836426,
+      "loss_num": 0.1611328125,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 294004096,
+      "step": 1701
+    },
+    {
+      "epoch": 0.6518575258521639,
+      "grad_norm": 29.243923476544346,
+      "learning_rate": 5e-06,
+      "loss": 0.256,
+      "num_input_tokens_seen": 294176928,
+      "step": 1702
+    },
+    {
+      "epoch": 0.6518575258521639,
+      "loss": 0.2195555418729782,
+      "loss_ce": 0.020886098966002464,
+      "loss_iou": 1.1945915222167969,
+      "loss_num": 0.1982421875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 294176928,
+      "step": 1702
+    },
+    {
+      "epoch": 0.6522405208732287,
+      "grad_norm": 32.061288386345836,
+      "learning_rate": 5e-06,
+      "loss": 0.1868,
+      "num_input_tokens_seen": 294349888,
+      "step": 1703
+    },
+    {
+      "epoch": 0.6522405208732287,
+      "loss": 0.19357256591320038,
+      "loss_ce": 0.02139239013195038,
+      "loss_iou": 1.1044819355010986,
+      "loss_num": 0.171875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 294349888,
+      "step": 1703
+    },
+    {
+      "epoch": 0.6526235158942933,
+      "grad_norm": 37.757048849163496,
+      "learning_rate": 5e-06,
+      "loss": 0.2269,
+      "num_input_tokens_seen": 294522608,
+      "step": 1704
+    },
+    {
+      "epoch": 0.6526235158942933,
+      "loss": 0.21185964345932007,
+      "loss_ce": 0.019415799528360367,
+      "loss_iou": 0.9338486194610596,
+      "loss_num": 0.1923828125,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 294522608,
+      "step": 1704
+    },
+    {
+      "epoch": 0.6530065109153581,
+      "grad_norm": 38.167257992835175,
+      "learning_rate": 5e-06,
+      "loss": 0.172,
+      "num_input_tokens_seen": 294695712,
+      "step": 1705
+    },
+    {
+      "epoch": 0.6530065109153581,
+      "loss": 0.17286206781864166,
+      "loss_ce": 0.02143385075032711,
+      "loss_iou": 1.0026297569274902,
+      "loss_num": 0.1513671875,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 294695712,
+      "step": 1705
+    },
+    {
+      "epoch": 0.6533895059364229,
+      "grad_norm": 36.23637394923736,
+      "learning_rate": 5e-06,
+      "loss": 0.2488,
+      "num_input_tokens_seen": 294868800,
+      "step": 1706
+    },
+    {
+      "epoch": 0.6533895059364229,
+      "loss": 0.22707700729370117,
+      "loss_ce": 0.02193785458803177,
+      "loss_iou": 1.0421545505523682,
+      "loss_num": 0.205078125,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 294868800,
+      "step": 1706
+    },
+    {
+      "epoch": 0.6537725009574875,
+      "grad_norm": 37.870493352470206,
+      "learning_rate": 5e-06,
+      "loss": 0.2235,
+      "num_input_tokens_seen": 295041824,
+      "step": 1707
+    },
+    {
+      "epoch": 0.6537725009574875,
+      "loss": 0.21987639367580414,
+      "loss_ce": 0.02016936056315899,
+      "loss_iou": 0.8914568424224854,
+      "loss_num": 0.19921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 295041824,
+      "step": 1707
+    },
+    {
+      "epoch": 0.6541554959785523,
+      "grad_norm": 38.45133141639371,
+      "learning_rate": 5e-06,
+      "loss": 0.189,
+      "num_input_tokens_seen": 295215032,
+      "step": 1708
+    },
+    {
+      "epoch": 0.6541554959785523,
+      "loss": 0.2147710770368576,
+      "loss_ce": 0.020496182143688202,
+      "loss_iou": 1.0865932703018188,
+      "loss_num": 0.1943359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 295215032,
+      "step": 1708
+    },
+    {
+      "epoch": 0.654538490999617,
+      "grad_norm": 34.235338399320725,
+      "learning_rate": 5e-06,
+      "loss": 0.1696,
+      "num_input_tokens_seen": 295387952,
+      "step": 1709
+    },
+    {
+      "epoch": 0.654538490999617,
+      "loss": 0.21084356307983398,
+      "loss_ce": 0.022977357730269432,
+      "loss_iou": 1.0059623718261719,
+      "loss_num": 0.1875,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 295387952,
+      "step": 1709
+    },
+    {
+      "epoch": 0.6549214860206817,
+      "grad_norm": 33.04359780642886,
+      "learning_rate": 5e-06,
+      "loss": 0.2155,
+      "num_input_tokens_seen": 295560728,
+      "step": 1710
+    },
+    {
+      "epoch": 0.6549214860206817,
+      "loss": 0.23250362277030945,
+      "loss_ce": 0.020650606602430344,
+      "loss_iou": 1.0279673337936401,
+      "loss_num": 0.2119140625,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 295560728,
+      "step": 1710
+    },
+    {
+      "epoch": 0.6553044810417464,
+      "grad_norm": 19.966859329183624,
+      "learning_rate": 5e-06,
+      "loss": 0.1738,
+      "num_input_tokens_seen": 295733856,
+      "step": 1711
+    },
+    {
+      "epoch": 0.6553044810417464,
+      "loss": 0.1837073266506195,
+      "loss_ce": 0.021781060844659805,
+      "loss_iou": 1.016855001449585,
+      "loss_num": 0.162109375,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 295733856,
+      "step": 1711
+    },
+    {
+      "epoch": 0.6556874760628112,
+      "grad_norm": 23.880401350337625,
+      "learning_rate": 5e-06,
+      "loss": 0.2026,
+      "num_input_tokens_seen": 295906992,
+      "step": 1712
+    },
+    {
+      "epoch": 0.6556874760628112,
+      "loss": 0.21519222855567932,
+      "loss_ce": 0.022138018161058426,
+      "loss_iou": 1.0332608222961426,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 295906992,
+      "step": 1712
+    },
+    {
+      "epoch": 0.656070471083876,
+      "grad_norm": 31.534495763656256,
+      "learning_rate": 5e-06,
+      "loss": 0.1444,
+      "num_input_tokens_seen": 296080216,
+      "step": 1713
+    },
+    {
+      "epoch": 0.656070471083876,
+      "loss": 0.1463220715522766,
+      "loss_ce": 0.020345501601696014,
+      "loss_iou": 1.0005614757537842,
+      "loss_num": 0.1259765625,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 296080216,
+      "step": 1713
+    },
+    {
+      "epoch": 0.6564534661049406,
+      "grad_norm": 34.2174501424663,
+      "learning_rate": 5e-06,
+      "loss": 0.2318,
+      "num_input_tokens_seen": 296253312,
+      "step": 1714
+    },
+    {
+      "epoch": 0.6564534661049406,
+      "loss": 0.23360732197761536,
+      "loss_ce": 0.020960837602615356,
+      "loss_iou": 1.0654317140579224,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 296253312,
+      "step": 1714
+    },
+    {
+      "epoch": 0.6568364611260054,
+      "grad_norm": 31.381640399255122,
+      "learning_rate": 5e-06,
+      "loss": 0.1829,
+      "num_input_tokens_seen": 296426312,
+      "step": 1715
+    },
+    {
+      "epoch": 0.6568364611260054,
+      "loss": 0.16834989190101624,
+      "loss_ce": 0.021438272669911385,
+      "loss_iou": 1.0068178176879883,
+      "loss_num": 0.146484375,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 296426312,
+      "step": 1715
+    },
+    {
+      "epoch": 0.6572194561470701,
+      "grad_norm": 27.4816830684573,
+      "learning_rate": 5e-06,
+      "loss": 0.2295,
+      "num_input_tokens_seen": 296599056,
+      "step": 1716
+    },
+    {
+      "epoch": 0.6572194561470701,
+      "loss": 0.17428439855575562,
+      "loss_ce": 0.020780974999070168,
+      "loss_iou": 1.0212924480438232,
+      "loss_num": 0.1533203125,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 296599056,
+      "step": 1716
+    },
+    {
+      "epoch": 0.6576024511681348,
+      "grad_norm": 33.80017547041262,
+      "learning_rate": 5e-06,
+      "loss": 0.2593,
+      "num_input_tokens_seen": 296771928,
+      "step": 1717
+    },
+    {
+      "epoch": 0.6576024511681348,
+      "loss": 0.2885706126689911,
+      "loss_ce": 0.022701462730765343,
+      "loss_iou": 1.0551213026046753,
+      "loss_num": 0.265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 296771928,
+      "step": 1717
+    },
+    {
+      "epoch": 0.6579854461891995,
+      "grad_norm": 31.938425239438647,
+      "learning_rate": 5e-06,
+      "loss": 0.2438,
+      "num_input_tokens_seen": 296945000,
+      "step": 1718
+    },
+    {
+      "epoch": 0.6579854461891995,
+      "loss": 0.28085488080978394,
+      "loss_ce": 0.019624410197138786,
+      "loss_iou": 1.2784092426300049,
+      "loss_num": 0.26171875,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 296945000,
+      "step": 1718
+    },
+    {
+      "epoch": 0.6583684412102643,
+      "grad_norm": 26.634670937491236,
+      "learning_rate": 5e-06,
+      "loss": 0.1825,
+      "num_input_tokens_seen": 297118256,
+      "step": 1719
+    },
+    {
+      "epoch": 0.6583684412102643,
+      "loss": 0.20411787927150726,
+      "loss_ce": 0.019486526027321815,
+      "loss_iou": 1.0216999053955078,
+      "loss_num": 0.1845703125,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 297118256,
+      "step": 1719
+    },
+    {
+      "epoch": 0.658751436231329,
+      "grad_norm": 31.39031539117525,
+      "learning_rate": 5e-06,
+      "loss": 0.2815,
+      "num_input_tokens_seen": 297291424,
+      "step": 1720
+    },
+    {
+      "epoch": 0.658751436231329,
+      "loss": 0.3041594922542572,
+      "loss_ce": 0.0213225856423378,
+      "loss_iou": 1.1091408729553223,
+      "loss_num": 0.283203125,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 297291424,
+      "step": 1720
+    },
+    {
+      "epoch": 0.6591344312523937,
+      "grad_norm": 20.24297703046322,
+      "learning_rate": 5e-06,
+      "loss": 0.2664,
+      "num_input_tokens_seen": 297464400,
+      "step": 1721
+    },
+    {
+      "epoch": 0.6591344312523937,
+      "loss": 0.22056660056114197,
+      "loss_ce": 0.02061541937291622,
+      "loss_iou": 3.551558494567871,
+      "loss_num": 0.2001953125,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 297464400,
+      "step": 1721
+    },
+    {
+      "epoch": 0.6595174262734584,
+      "grad_norm": 50.324486409063304,
+      "learning_rate": 5e-06,
+      "loss": 0.2629,
+      "num_input_tokens_seen": 297637456,
+      "step": 1722
+    },
+    {
+      "epoch": 0.6595174262734584,
+      "loss": 0.28015631437301636,
+      "loss_ce": 0.021733462810516357,
+      "loss_iou": 1.0348681211471558,
+      "loss_num": 0.2578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 297637456,
+      "step": 1722
+    },
+    {
+      "epoch": 0.6599004212945232,
+      "grad_norm": 37.87642873521686,
+      "learning_rate": 5e-06,
+      "loss": 0.2542,
+      "num_input_tokens_seen": 297810136,
+      "step": 1723
+    },
+    {
+      "epoch": 0.6599004212945232,
+      "loss": 0.25040507316589355,
+      "loss_ce": 0.018837690353393555,
+      "loss_iou": 1.0068309307098389,
+      "loss_num": 0.2314453125,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 297810136,
+      "step": 1723
+    },
+    {
+      "epoch": 0.6602834163155878,
+      "grad_norm": 31.731085646022194,
+      "learning_rate": 5e-06,
+      "loss": 0.2083,
+      "num_input_tokens_seen": 297983224,
+      "step": 1724
+    },
+    {
+      "epoch": 0.6602834163155878,
+      "loss": 0.22054725885391235,
+      "loss_ce": 0.0201078113168478,
+      "loss_iou": 1.0050134658813477,
+      "loss_num": 0.2001953125,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 297983224,
+      "step": 1724
+    },
+    {
+      "epoch": 0.6606664113366526,
+      "grad_norm": 51.012362404574965,
+      "learning_rate": 5e-06,
+      "loss": 0.2488,
+      "num_input_tokens_seen": 298156328,
+      "step": 1725
+    },
+    {
+      "epoch": 0.6606664113366526,
+      "loss": 0.21641242504119873,
+      "loss_ce": 0.021832339465618134,
+      "loss_iou": 1.051340103149414,
+      "loss_num": 0.1943359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 298156328,
+      "step": 1725
+    },
+    {
+      "epoch": 0.6610494063577174,
+      "grad_norm": 49.488018407958606,
+      "learning_rate": 5e-06,
+      "loss": 0.2671,
+      "num_input_tokens_seen": 298329096,
+      "step": 1726
+    },
+    {
+      "epoch": 0.6610494063577174,
+      "loss": 0.2786455750465393,
+      "loss_ce": 0.0202227421104908,
+      "loss_iou": 1.1039822101593018,
+      "loss_num": 0.2578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 298329096,
+      "step": 1726
+    },
+    {
+      "epoch": 0.6614324013787821,
+      "grad_norm": 33.18477592994935,
+      "learning_rate": 5e-06,
+      "loss": 0.2186,
+      "num_input_tokens_seen": 298501624,
+      "step": 1727
+    },
+    {
+      "epoch": 0.6614324013787821,
+      "loss": 0.19704484939575195,
+      "loss_ce": 0.020348068326711655,
+      "loss_iou": 1.369502067565918,
+      "loss_num": 0.1767578125,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 298501624,
+      "step": 1727
+    },
+    {
+      "epoch": 0.6618153963998468,
+      "grad_norm": 27.470186172630914,
+      "learning_rate": 5e-06,
+      "loss": 0.2499,
+      "num_input_tokens_seen": 298674488,
+      "step": 1728
+    },
+    {
+      "epoch": 0.6618153963998468,
+      "loss": 0.29464203119277954,
+      "loss_ce": 0.020594192668795586,
+      "loss_iou": 1.1022378206253052,
+      "loss_num": 0.2734375,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 298674488,
+      "step": 1728
+    },
+    {
+      "epoch": 0.6621983914209115,
+      "grad_norm": 26.470082484477828,
+      "learning_rate": 5e-06,
+      "loss": 0.1768,
+      "num_input_tokens_seen": 298847728,
+      "step": 1729
+    },
+    {
+      "epoch": 0.6621983914209115,
+      "loss": 0.19022804498672485,
+      "loss_ce": 0.02067238837480545,
+      "loss_iou": 1.0489208698272705,
+      "loss_num": 0.169921875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 298847728,
+      "step": 1729
+    },
+    {
+      "epoch": 0.6625813864419763,
+      "grad_norm": 34.58069753353553,
+      "learning_rate": 5e-06,
+      "loss": 0.2653,
+      "num_input_tokens_seen": 299020904,
+      "step": 1730
+    },
+    {
+      "epoch": 0.6625813864419763,
+      "loss": 0.23365744948387146,
+      "loss_ce": 0.019424058496952057,
+      "loss_iou": 1.0863093137741089,
+      "loss_num": 0.2138671875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 299020904,
+      "step": 1730
+    },
+    {
+      "epoch": 0.6629643814630409,
+      "grad_norm": 32.29339221098594,
+      "learning_rate": 5e-06,
+      "loss": 0.2302,
+      "num_input_tokens_seen": 299193568,
+      "step": 1731
+    },
+    {
+      "epoch": 0.6629643814630409,
+      "loss": 0.24851149320602417,
+      "loss_ce": 0.021949004381895065,
+      "loss_iou": 1.064180850982666,
+      "loss_num": 0.2265625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 299193568,
+      "step": 1731
+    },
+    {
+      "epoch": 0.6633473764841057,
+      "grad_norm": 26.165793245855838,
+      "learning_rate": 5e-06,
+      "loss": 0.2111,
+      "num_input_tokens_seen": 299366744,
+      "step": 1732
+    },
+    {
+      "epoch": 0.6633473764841057,
+      "loss": 0.16233590245246887,
+      "loss_ce": 0.02000192180275917,
+      "loss_iou": 1.0067273378372192,
+      "loss_num": 0.142578125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 299366744,
+      "step": 1732
+    },
+    {
+      "epoch": 0.6637303715051704,
+      "grad_norm": 52.60015171294837,
+      "learning_rate": 5e-06,
+      "loss": 0.2458,
+      "num_input_tokens_seen": 299539784,
+      "step": 1733
+    },
+    {
+      "epoch": 0.6637303715051704,
+      "loss": 0.20560500025749207,
+      "loss_ce": 0.020485367625951767,
+      "loss_iou": 1.0419907569885254,
+      "loss_num": 0.185546875,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 299539784,
+      "step": 1733
+    },
+    {
+      "epoch": 0.6641133665262352,
+      "grad_norm": 41.139530338835335,
+      "learning_rate": 5e-06,
+      "loss": 0.239,
+      "num_input_tokens_seen": 299712544,
+      "step": 1734
+    },
+    {
+      "epoch": 0.6641133665262352,
+      "loss": 0.2755686044692993,
+      "loss_ce": 0.020990949124097824,
+      "loss_iou": 1.0008246898651123,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 299712544,
+      "step": 1734
+    },
+    {
+      "epoch": 0.6644963615472999,
+      "grad_norm": 29.248925135000487,
+      "learning_rate": 5e-06,
+      "loss": 0.2162,
+      "num_input_tokens_seen": 299885552,
+      "step": 1735
+    },
+    {
+      "epoch": 0.6644963615472999,
+      "loss": 0.2706708312034607,
+      "loss_ce": 0.02024359256029129,
+      "loss_iou": 1.039218544960022,
+      "loss_num": 0.25,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 299885552,
+      "step": 1735
+    },
+    {
+      "epoch": 0.6648793565683646,
+      "grad_norm": 40.82120532816002,
+      "learning_rate": 5e-06,
+      "loss": 0.2969,
+      "num_input_tokens_seen": 300058296,
+      "step": 1736
+    },
+    {
+      "epoch": 0.6648793565683646,
+      "loss": 0.37651216983795166,
+      "loss_ce": 0.02116549387574196,
+      "loss_iou": 1.0084853172302246,
+      "loss_num": 0.35546875,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 300058296,
+      "step": 1736
+    },
+    {
+      "epoch": 0.6652623515894294,
+      "grad_norm": 31.57196412613779,
+      "learning_rate": 5e-06,
+      "loss": 0.2361,
+      "num_input_tokens_seen": 300231216,
+      "step": 1737
+    },
+    {
+      "epoch": 0.6652623515894294,
+      "loss": 0.2580101490020752,
+      "loss_ce": 0.02064444310963154,
+      "loss_iou": 1.0133150815963745,
+      "loss_num": 0.2373046875,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 300231216,
+      "step": 1737
+    },
+    {
+      "epoch": 0.665645346610494,
+      "grad_norm": 48.78518103902854,
+      "learning_rate": 5e-06,
+      "loss": 0.2229,
+      "num_input_tokens_seen": 300404144,
+      "step": 1738
+    },
+    {
+      "epoch": 0.665645346610494,
+      "loss": 0.19864070415496826,
+      "loss_ce": 0.02267635054886341,
+      "loss_iou": 1.0027598142623901,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 300404144,
+      "step": 1738
+    },
+    {
+      "epoch": 0.6660283416315588,
+      "grad_norm": 39.83128700582122,
+      "learning_rate": 5e-06,
+      "loss": 0.2173,
+      "num_input_tokens_seen": 300577064,
+      "step": 1739
+    },
+    {
+      "epoch": 0.6660283416315588,
+      "loss": 0.21679170429706573,
+      "loss_ce": 0.022089559584856033,
+      "loss_iou": 1.00639009475708,
+      "loss_num": 0.1943359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 300577064,
+      "step": 1739
+    },
+    {
+      "epoch": 0.6664113366526235,
+      "grad_norm": 32.20998987442295,
+      "learning_rate": 5e-06,
+      "loss": 0.225,
+      "num_input_tokens_seen": 300750048,
+      "step": 1740
+    },
+    {
+      "epoch": 0.6664113366526235,
+      "loss": 0.1972222626209259,
+      "loss_ce": 0.020952735096216202,
+      "loss_iou": 1.0016233921051025,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 300750048,
+      "step": 1740
+    },
+    {
+      "epoch": 0.6667943316736883,
+      "grad_norm": 39.70792162268065,
+      "learning_rate": 5e-06,
+      "loss": 0.2469,
+      "num_input_tokens_seen": 300923072,
+      "step": 1741
+    },
+    {
+      "epoch": 0.6667943316736883,
+      "loss": 0.25135573744773865,
+      "loss_ce": 0.020886968821287155,
+      "loss_iou": 1.003562092781067,
+      "loss_num": 0.23046875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 300923072,
+      "step": 1741
+    },
+    {
+      "epoch": 0.6671773266947529,
+      "grad_norm": 44.71350835915628,
+      "learning_rate": 5e-06,
+      "loss": 0.2468,
+      "num_input_tokens_seen": 301092496,
+      "step": 1742
+    },
+    {
+      "epoch": 0.6671773266947529,
+      "loss": 0.2708122730255127,
+      "loss_ce": 0.02093435451388359,
+      "loss_iou": 1.0258967876434326,
+      "loss_num": 0.25,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 301092496,
+      "step": 1742
+    },
+    {
+      "epoch": 0.6675603217158177,
+      "grad_norm": 33.32851894888501,
+      "learning_rate": 5e-06,
+      "loss": 0.2162,
+      "num_input_tokens_seen": 301265456,
+      "step": 1743
+    },
+    {
+      "epoch": 0.6675603217158177,
+      "loss": 0.22700703144073486,
+      "loss_ce": 0.02247823402285576,
+      "loss_iou": 1.108150601387024,
+      "loss_num": 0.2041015625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 301265456,
+      "step": 1743
+    },
+    {
+      "epoch": 0.6679433167368825,
+      "grad_norm": 31.718458304650298,
+      "learning_rate": 5e-06,
+      "loss": 0.1828,
+      "num_input_tokens_seen": 301438424,
+      "step": 1744
+    },
+    {
+      "epoch": 0.6679433167368825,
+      "loss": 0.186773881316185,
+      "loss_ce": 0.019659623503684998,
+      "loss_iou": 1.0158426761627197,
+      "loss_num": 0.1669921875,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 301438424,
+      "step": 1744
+    },
+    {
+      "epoch": 0.6683263117579471,
+      "grad_norm": 26.422891777734012,
+      "learning_rate": 5e-06,
+      "loss": 0.2487,
+      "num_input_tokens_seen": 301610784,
+      "step": 1745
+    },
+    {
+      "epoch": 0.6683263117579471,
+      "loss": 0.25292688608169556,
+      "loss_ce": 0.020199844613671303,
+      "loss_iou": 1.0773735046386719,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 301610784,
+      "step": 1745
+    },
+    {
+      "epoch": 0.6687093067790119,
+      "grad_norm": 30.388233032232165,
+      "learning_rate": 5e-06,
+      "loss": 0.2648,
+      "num_input_tokens_seen": 301783448,
+      "step": 1746
+    },
+    {
+      "epoch": 0.6687093067790119,
+      "loss": 0.267360121011734,
+      "loss_ce": 0.020533941686153412,
+      "loss_iou": 1.0193747282028198,
+      "loss_num": 0.2470703125,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 301783448,
+      "step": 1746
+    },
+    {
+      "epoch": 0.6690923018000766,
+      "grad_norm": 32.88619745615411,
+      "learning_rate": 5e-06,
+      "loss": 0.2584,
+      "num_input_tokens_seen": 301956288,
+      "step": 1747
+    },
+    {
+      "epoch": 0.6690923018000766,
+      "loss": 0.24807390570640564,
+      "loss_ce": 0.02029069885611534,
+      "loss_iou": 1.0906078815460205,
+      "loss_num": 0.2275390625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 301956288,
+      "step": 1747
+    },
+    {
+      "epoch": 0.6694752968211414,
+      "grad_norm": 33.873327785376866,
+      "learning_rate": 5e-06,
+      "loss": 0.2666,
+      "num_input_tokens_seen": 302129344,
+      "step": 1748
+    },
+    {
+      "epoch": 0.6694752968211414,
+      "loss": 0.30207377672195435,
+      "loss_ce": 0.021312061697244644,
+      "loss_iou": 1.071867823600769,
+      "loss_num": 0.28125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 302129344,
+      "step": 1748
+    },
+    {
+      "epoch": 0.669858291842206,
+      "grad_norm": 29.252939318589306,
+      "learning_rate": 5e-06,
+      "loss": 0.1732,
+      "num_input_tokens_seen": 302302592,
+      "step": 1749
+    },
+    {
+      "epoch": 0.669858291842206,
+      "loss": 0.1982949674129486,
+      "loss_ce": 0.019339896738529205,
+      "loss_iou": 1.044935703277588,
+      "loss_num": 0.1787109375,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 302302592,
+      "step": 1749
+    },
+    {
+      "epoch": 0.6702412868632708,
+      "grad_norm": 36.6035992879886,
+      "learning_rate": 5e-06,
+      "loss": 0.2687,
+      "num_input_tokens_seen": 302475376,
+      "step": 1750
+    },
+    {
+      "epoch": 0.6702412868632708,
+      "eval_websight_new_CIoU": 0.9130485653877258,
+      "eval_websight_new_GIoU": 0.9125950634479523,
+      "eval_websight_new_IoU": 0.9134105741977692,
+      "eval_websight_new_MAE_all": 0.009155952837318182,
+      "eval_websight_new_MAE_h": 0.008650638163089752,
+      "eval_websight_new_MAE_w": 0.009684689342975616,
+      "eval_websight_new_MAE_x": 0.008301699068397284,
+      "eval_websight_new_MAE_y": 0.009986785240471363,
+      "eval_websight_new_NUM_probability": 8.473627895000391e-05,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.1299513578414917,
+      "eval_websight_new_loss_ce": 0.04177057556807995,
+      "eval_websight_new_loss_iou": 1.0004342794418335,
+      "eval_websight_new_loss_num": 0.087677001953125,
+      "eval_websight_new_loss_xval": 0.087677001953125,
+      "eval_websight_new_runtime": 55.7981,
+      "eval_websight_new_samples_per_second": 0.896,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 302475376,
+      "step": 1750
+    },
+    {
+      "epoch": 0.6702412868632708,
+      "eval_seeclick_CIoU": 0.6328222751617432,
+      "eval_seeclick_GIoU": 0.637648195028305,
+      "eval_seeclick_IoU": 0.6658692359924316,
+      "eval_seeclick_MAE_all": 0.060492176562547684,
+      "eval_seeclick_MAE_h": 0.038623055443167686,
+      "eval_seeclick_MAE_w": 0.08360513672232628,
+      "eval_seeclick_MAE_x": 0.07366372644901276,
+      "eval_seeclick_MAE_y": 0.046076804399490356,
+      "eval_seeclick_NUM_probability": 0.00011501446715556085,
+      "eval_seeclick_inside_bbox": 0.9097222089767456,
+      "eval_seeclick_loss": 0.5198314189910889,
+      "eval_seeclick_loss_ce": 0.0325775733217597,
+      "eval_seeclick_loss_iou": 1.1606248021125793,
+      "eval_seeclick_loss_num": 0.4901123046875,
+      "eval_seeclick_loss_xval": 0.4901123046875,
+      "eval_seeclick_runtime": 82.4764,
+      "eval_seeclick_samples_per_second": 0.606,
+      "eval_seeclick_steps_per_second": 0.024,
+      "num_input_tokens_seen": 302475376,
+      "step": 1750
+    },
+    {
+      "epoch": 0.6702412868632708,
+      "eval_icons_CIoU": 0.8662566840648651,
+      "eval_icons_GIoU": 0.8614764511585236,
+      "eval_icons_IoU": 0.8702732026576996,
+      "eval_icons_MAE_all": 0.01815019128844142,
+      "eval_icons_MAE_h": 0.018266551662236452,
+      "eval_icons_MAE_w": 0.018474142998456955,
+      "eval_icons_MAE_x": 0.016881443560123444,
+      "eval_icons_MAE_y": 0.018978629261255264,
+      "eval_icons_NUM_probability": 7.645552977919579e-05,
+      "eval_icons_inside_bbox": 1.0,
+      "eval_icons_loss": 0.14464829862117767,
+      "eval_icons_loss_ce": 0.028360147960484028,
+      "eval_icons_loss_iou": 1.0153546333312988,
+      "eval_icons_loss_num": 0.1070709228515625,
+      "eval_icons_loss_xval": 0.1070709228515625,
+      "eval_icons_runtime": 82.735,
+      "eval_icons_samples_per_second": 0.604,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 302475376,
+      "step": 1750
+    },
+    {
+      "epoch": 0.6702412868632708,
+      "loss": 0.1576666533946991,
+      "loss_ce": 0.028394190594553947,
+      "loss_iou": 1.0300023555755615,
+      "loss_num": 0.12890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 302475376,
+      "step": 1750
+    },
+    {
+      "epoch": 0.6706242818843355,
+      "grad_norm": 51.95143696163088,
+      "learning_rate": 5e-06,
+      "loss": 0.2168,
+      "num_input_tokens_seen": 302648304,
+      "step": 1751
+    },
+    {
+      "epoch": 0.6706242818843355,
+      "loss": 0.2516476809978485,
+      "loss_ce": 0.020568573847413063,
+      "loss_iou": 1.066545009613037,
+      "loss_num": 0.2314453125,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 302648304,
+      "step": 1751
+    },
+    {
+      "epoch": 0.6710072769054002,
+      "grad_norm": 34.17640835813798,
+      "learning_rate": 5e-06,
+      "loss": 0.2336,
+      "num_input_tokens_seen": 302821224,
+      "step": 1752
+    },
+    {
+      "epoch": 0.6710072769054002,
+      "loss": 0.2308710515499115,
+      "loss_ce": 0.022252880036830902,
+      "loss_iou": 1.0408929586410522,
+      "loss_num": 0.208984375,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 302821224,
+      "step": 1752
+    },
+    {
+      "epoch": 0.6713902719264649,
+      "grad_norm": 30.280846550618126,
+      "learning_rate": 5e-06,
+      "loss": 0.2257,
+      "num_input_tokens_seen": 302994432,
+      "step": 1753
+    },
+    {
+      "epoch": 0.6713902719264649,
+      "loss": 0.18668514490127563,
+      "loss_ce": 0.02304989844560623,
+      "loss_iou": 1.007064700126648,
+      "loss_num": 0.1640625,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 302994432,
+      "step": 1753
+    },
+    {
+      "epoch": 0.6717732669475297,
+      "grad_norm": 24.04276657890047,
+      "learning_rate": 5e-06,
+      "loss": 0.2479,
+      "num_input_tokens_seen": 303167336,
+      "step": 1754
+    },
+    {
+      "epoch": 0.6717732669475297,
+      "loss": 0.3154633045196533,
+      "loss_ce": 0.021456940099596977,
+      "loss_iou": 1.0964784622192383,
+      "loss_num": 0.294921875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 303167336,
+      "step": 1754
+    },
+    {
+      "epoch": 0.6721562619685945,
+      "grad_norm": 28.63546381930945,
+      "learning_rate": 5e-06,
+      "loss": 0.1799,
+      "num_input_tokens_seen": 303339800,
+      "step": 1755
+    },
+    {
+      "epoch": 0.6721562619685945,
+      "loss": 0.1940738409757614,
+      "loss_ce": 0.019940536469221115,
+      "loss_iou": 1.013685703277588,
+      "loss_num": 0.173828125,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 303339800,
+      "step": 1755
+    },
+    {
+      "epoch": 0.6725392569896591,
+      "grad_norm": 43.70532154112608,
+      "learning_rate": 5e-06,
+      "loss": 0.2719,
+      "num_input_tokens_seen": 303512416,
+      "step": 1756
+    },
+    {
+      "epoch": 0.6725392569896591,
+      "loss": 0.2490992546081543,
+      "loss_ce": 0.019240861758589745,
+      "loss_iou": 1.0142097473144531,
+      "loss_num": 0.2294921875,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 303512416,
+      "step": 1756
+    },
+    {
+      "epoch": 0.6729222520107239,
+      "grad_norm": 31.59577451586341,
+      "learning_rate": 5e-06,
+      "loss": 0.253,
+      "num_input_tokens_seen": 303685704,
+      "step": 1757
+    },
+    {
+      "epoch": 0.6729222520107239,
+      "loss": 0.2875283360481262,
+      "loss_ce": 0.022086460143327713,
+      "loss_iou": 1.224463701248169,
+      "loss_num": 0.265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 303685704,
+      "step": 1757
+    },
+    {
+      "epoch": 0.6733052470317886,
+      "grad_norm": 32.15160803956687,
+      "learning_rate": 5e-06,
+      "loss": 0.269,
+      "num_input_tokens_seen": 303858832,
+      "step": 1758
+    },
+    {
+      "epoch": 0.6733052470317886,
+      "loss": 0.22685261070728302,
+      "loss_ce": 0.02073688805103302,
+      "loss_iou": 1.0503966808319092,
+      "loss_num": 0.2060546875,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 303858832,
+      "step": 1758
+    },
+    {
+      "epoch": 0.6736882420528533,
+      "grad_norm": 24.766245863577627,
+      "learning_rate": 5e-06,
+      "loss": 0.2847,
+      "num_input_tokens_seen": 304027704,
+      "step": 1759
+    },
+    {
+      "epoch": 0.6736882420528533,
+      "loss": 0.2911217510700226,
+      "loss_ce": 0.020736005157232285,
+      "loss_iou": 1.1352735757827759,
+      "loss_num": 0.26953125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 304027704,
+      "step": 1759
+    },
+    {
+      "epoch": 0.674071237073918,
+      "grad_norm": 21.91037300207571,
+      "learning_rate": 5e-06,
+      "loss": 0.2097,
+      "num_input_tokens_seen": 304200096,
+      "step": 1760
+    },
+    {
+      "epoch": 0.674071237073918,
+      "loss": 0.20687098801136017,
+      "loss_ce": 0.02034755051136017,
+      "loss_iou": 1.0257363319396973,
+      "loss_num": 0.1865234375,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 304200096,
+      "step": 1760
+    },
+    {
+      "epoch": 0.6744542320949828,
+      "grad_norm": 27.580579349103214,
+      "learning_rate": 5e-06,
+      "loss": 0.2412,
+      "num_input_tokens_seen": 304372944,
+      "step": 1761
+    },
+    {
+      "epoch": 0.6744542320949828,
+      "loss": 0.23300498723983765,
+      "loss_ce": 0.020236432552337646,
+      "loss_iou": 1.0436434745788574,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 304372944,
+      "step": 1761
+    },
+    {
+      "epoch": 0.6748372271160475,
+      "grad_norm": 38.406674583378454,
+      "learning_rate": 5e-06,
+      "loss": 0.2576,
+      "num_input_tokens_seen": 304545992,
+      "step": 1762
+    },
+    {
+      "epoch": 0.6748372271160475,
+      "loss": 0.27290505170822144,
+      "loss_ce": 0.02363748662173748,
+      "loss_iou": 1.0639228820800781,
+      "loss_num": 0.2490234375,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 304545992,
+      "step": 1762
+    },
+    {
+      "epoch": 0.6752202221371122,
+      "grad_norm": 21.226288009192686,
+      "learning_rate": 5e-06,
+      "loss": 0.1636,
+      "num_input_tokens_seen": 304718952,
+      "step": 1763
+    },
+    {
+      "epoch": 0.6752202221371122,
+      "loss": 0.17204737663269043,
+      "loss_ce": 0.018757598474621773,
+      "loss_iou": 0.8931306600570679,
+      "loss_num": 0.1533203125,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 304718952,
+      "step": 1763
+    },
+    {
+      "epoch": 0.675603217158177,
+      "grad_norm": 44.21116227709698,
+      "learning_rate": 5e-06,
+      "loss": 0.3058,
+      "num_input_tokens_seen": 304892248,
+      "step": 1764
+    },
+    {
+      "epoch": 0.675603217158177,
+      "loss": 0.2746030390262604,
+      "loss_ce": 0.01990334317088127,
+      "loss_iou": 1.0796000957489014,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 304892248,
+      "step": 1764
+    },
+    {
+      "epoch": 0.6759862121792417,
+      "grad_norm": 21.00194825165669,
+      "learning_rate": 5e-06,
+      "loss": 0.1898,
+      "num_input_tokens_seen": 305065088,
+      "step": 1765
+    },
+    {
+      "epoch": 0.6759862121792417,
+      "loss": 0.15532422065734863,
+      "loss_ce": 0.02043652907013893,
+      "loss_iou": 1.0036628246307373,
+      "loss_num": 0.134765625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 305065088,
+      "step": 1765
+    },
+    {
+      "epoch": 0.6763692072003064,
+      "grad_norm": 25.8206740495571,
+      "learning_rate": 5e-06,
+      "loss": 0.1807,
+      "num_input_tokens_seen": 305238112,
+      "step": 1766
+    },
+    {
+      "epoch": 0.6763692072003064,
+      "loss": 0.1638648957014084,
+      "loss_ce": 0.020859526470303535,
+      "loss_iou": 1.0076106786727905,
+      "loss_num": 0.142578125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 305238112,
+      "step": 1766
+    },
+    {
+      "epoch": 0.6767522022213711,
+      "grad_norm": 29.409402510935802,
+      "learning_rate": 5e-06,
+      "loss": 0.253,
+      "num_input_tokens_seen": 305411176,
+      "step": 1767
+    },
+    {
+      "epoch": 0.6767522022213711,
+      "loss": 0.26368898153305054,
+      "loss_ce": 0.02095215767621994,
+      "loss_iou": 0.9582514762878418,
+      "loss_num": 0.2431640625,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 305411176,
+      "step": 1767
+    },
+    {
+      "epoch": 0.6771351972424359,
+      "grad_norm": 33.04338935083643,
+      "learning_rate": 5e-06,
+      "loss": 0.2338,
+      "num_input_tokens_seen": 305584496,
+      "step": 1768
+    },
+    {
+      "epoch": 0.6771351972424359,
+      "loss": 0.280699759721756,
+      "loss_ce": 0.021483462303876877,
+      "loss_iou": 1.0993882417678833,
+      "loss_num": 0.259765625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 305584496,
+      "step": 1768
+    },
+    {
+      "epoch": 0.6775181922635006,
+      "grad_norm": 26.930870875600633,
+      "learning_rate": 5e-06,
+      "loss": 0.2234,
+      "num_input_tokens_seen": 305753888,
+      "step": 1769
+    },
+    {
+      "epoch": 0.6775181922635006,
+      "loss": 0.16383543610572815,
+      "loss_ce": 0.02211179956793785,
+      "loss_iou": 1.0060840845108032,
+      "loss_num": 0.1416015625,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 305753888,
+      "step": 1769
+    },
+    {
+      "epoch": 0.6779011872845653,
+      "grad_norm": 31.539665686264513,
+      "learning_rate": 5e-06,
+      "loss": 0.1786,
+      "num_input_tokens_seen": 305926968,
+      "step": 1770
+    },
+    {
+      "epoch": 0.6779011872845653,
+      "loss": 0.14972564578056335,
+      "loss_ce": 0.021551813930273056,
+      "loss_iou": 1.006534457206726,
+      "loss_num": 0.1279296875,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 305926968,
+      "step": 1770
+    },
+    {
+      "epoch": 0.67828418230563,
+      "grad_norm": 23.749091023528408,
+      "learning_rate": 5e-06,
+      "loss": 0.2213,
+      "num_input_tokens_seen": 306099816,
+      "step": 1771
+    },
+    {
+      "epoch": 0.67828418230563,
+      "loss": 0.1796785145998001,
+      "loss_ce": 0.01933915913105011,
+      "loss_iou": 1.0010827779769897,
+      "loss_num": 0.16015625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 306099816,
+      "step": 1771
+    },
+    {
+      "epoch": 0.6786671773266948,
+      "grad_norm": 30.60906493905963,
+      "learning_rate": 5e-06,
+      "loss": 0.1519,
+      "num_input_tokens_seen": 306272848,
+      "step": 1772
+    },
+    {
+      "epoch": 0.6786671773266948,
+      "loss": 0.16486088931560516,
+      "loss_ce": 0.020329643040895462,
+      "loss_iou": 1.0240137577056885,
+      "loss_num": 0.14453125,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 306272848,
+      "step": 1772
+    },
+    {
+      "epoch": 0.6790501723477594,
+      "grad_norm": 26.415161692070637,
+      "learning_rate": 5e-06,
+      "loss": 0.1976,
+      "num_input_tokens_seen": 306446264,
+      "step": 1773
+    },
+    {
+      "epoch": 0.6790501723477594,
+      "loss": 0.2072274088859558,
+      "loss_ce": 0.02161949872970581,
+      "loss_iou": 1.0546596050262451,
+      "loss_num": 0.185546875,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 306446264,
+      "step": 1773
+    },
+    {
+      "epoch": 0.6794331673688242,
+      "grad_norm": 25.473193131889023,
+      "learning_rate": 5e-06,
+      "loss": 0.2043,
+      "num_input_tokens_seen": 306619528,
+      "step": 1774
+    },
+    {
+      "epoch": 0.6794331673688242,
+      "loss": 0.18194493651390076,
+      "loss_ce": 0.02142249047756195,
+      "loss_iou": 1.021148681640625,
+      "loss_num": 0.16015625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 306619528,
+      "step": 1774
+    },
+    {
+      "epoch": 0.679816162389889,
+      "grad_norm": 31.990611083987442,
+      "learning_rate": 5e-06,
+      "loss": 0.1896,
+      "num_input_tokens_seen": 306792296,
+      "step": 1775
+    },
+    {
+      "epoch": 0.679816162389889,
+      "loss": 0.16622373461723328,
+      "loss_ce": 0.021021097898483276,
+      "loss_iou": 1.001260757446289,
+      "loss_num": 0.1455078125,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 306792296,
+      "step": 1775
+    },
+    {
+      "epoch": 0.6801991574109536,
+      "grad_norm": 31.886559171456238,
+      "learning_rate": 5e-06,
+      "loss": 0.1991,
+      "num_input_tokens_seen": 306965216,
+      "step": 1776
+    },
+    {
+      "epoch": 0.6801991574109536,
+      "loss": 0.22478890419006348,
+      "loss_ce": 0.02123665250837803,
+      "loss_iou": 1.0109344720840454,
+      "loss_num": 0.203125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 306965216,
+      "step": 1776
+    },
+    {
+      "epoch": 0.6805821524320184,
+      "grad_norm": 36.18691494698098,
+      "learning_rate": 5e-06,
+      "loss": 0.1627,
+      "num_input_tokens_seen": 307138248,
+      "step": 1777
+    },
+    {
+      "epoch": 0.6805821524320184,
+      "loss": 0.18779145181179047,
+      "loss_ce": 0.020921334624290466,
+      "loss_iou": 1.028134822845459,
+      "loss_num": 0.1669921875,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 307138248,
+      "step": 1777
+    },
+    {
+      "epoch": 0.6809651474530831,
+      "grad_norm": 23.832373663803043,
+      "learning_rate": 5e-06,
+      "loss": 0.2239,
+      "num_input_tokens_seen": 307311440,
+      "step": 1778
+    },
+    {
+      "epoch": 0.6809651474530831,
+      "loss": 0.2649584412574768,
+      "loss_ce": 0.020573675632476807,
+      "loss_iou": 1.0320253372192383,
+      "loss_num": 0.244140625,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 307311440,
+      "step": 1778
+    },
+    {
+      "epoch": 0.6813481424741479,
+      "grad_norm": 23.33458947613337,
+      "learning_rate": 5e-06,
+      "loss": 0.1962,
+      "num_input_tokens_seen": 307484264,
+      "step": 1779
+    },
+    {
+      "epoch": 0.6813481424741479,
+      "loss": 0.1672830432653427,
+      "loss_ce": 0.018998129293322563,
+      "loss_iou": 1.0017093420028687,
+      "loss_num": 0.1484375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 307484264,
+      "step": 1779
+    },
+    {
+      "epoch": 0.6817311374952125,
+      "grad_norm": 31.416724601681047,
+      "learning_rate": 5e-06,
+      "loss": 0.2722,
+      "num_input_tokens_seen": 307657208,
+      "step": 1780
+    },
+    {
+      "epoch": 0.6817311374952125,
+      "loss": 0.23875892162322998,
+      "loss_ce": 0.020314078778028488,
+      "loss_iou": 1.0517665147781372,
+      "loss_num": 0.21875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 307657208,
+      "step": 1780
+    },
+    {
+      "epoch": 0.6821141325162773,
+      "grad_norm": 33.7553905878189,
+      "learning_rate": 5e-06,
+      "loss": 0.2234,
+      "num_input_tokens_seen": 307830152,
+      "step": 1781
+    },
+    {
+      "epoch": 0.6821141325162773,
+      "loss": 0.23483659327030182,
+      "loss_ce": 0.02084733173251152,
+      "loss_iou": 1.0256426334381104,
+      "loss_num": 0.2138671875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 307830152,
+      "step": 1781
+    },
+    {
+      "epoch": 0.682497127537342,
+      "grad_norm": 21.318674986674704,
+      "learning_rate": 5e-06,
+      "loss": 0.1875,
+      "num_input_tokens_seen": 308003104,
+      "step": 1782
+    },
+    {
+      "epoch": 0.682497127537342,
+      "loss": 0.1763216257095337,
+      "loss_ce": 0.021353356540203094,
+      "loss_iou": 1.34816575050354,
+      "loss_num": 0.1552734375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 308003104,
+      "step": 1782
+    },
+    {
+      "epoch": 0.6828801225584067,
+      "grad_norm": 29.362432741919577,
+      "learning_rate": 5e-06,
+      "loss": 0.2344,
+      "num_input_tokens_seen": 308176144,
+      "step": 1783
+    },
+    {
+      "epoch": 0.6828801225584067,
+      "loss": 0.2393755167722702,
+      "loss_ce": 0.021480005234479904,
+      "loss_iou": 0.979744553565979,
+      "loss_num": 0.2177734375,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 308176144,
+      "step": 1783
+    },
+    {
+      "epoch": 0.6832631175794714,
+      "grad_norm": 34.696348663593994,
+      "learning_rate": 5e-06,
+      "loss": 0.2452,
+      "num_input_tokens_seen": 308348824,
+      "step": 1784
+    },
+    {
+      "epoch": 0.6832631175794714,
+      "loss": 0.2556586265563965,
+      "loss_ce": 0.02244327962398529,
+      "loss_iou": 1.0269641876220703,
+      "loss_num": 0.2333984375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 308348824,
+      "step": 1784
+    },
+    {
+      "epoch": 0.6836461126005362,
+      "grad_norm": 20.509339795245733,
+      "learning_rate": 5e-06,
+      "loss": 0.237,
+      "num_input_tokens_seen": 308521984,
+      "step": 1785
+    },
+    {
+      "epoch": 0.6836461126005362,
+      "loss": 0.15286551415920258,
+      "loss_ce": 0.019808875396847725,
+      "loss_iou": 1.003145694732666,
+      "loss_num": 0.1328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 308521984,
+      "step": 1785
+    },
+    {
+      "epoch": 0.684029107621601,
+      "grad_norm": 26.791570992031154,
+      "learning_rate": 5e-06,
+      "loss": 0.2871,
+      "num_input_tokens_seen": 308694760,
+      "step": 1786
+    },
+    {
+      "epoch": 0.684029107621601,
+      "loss": 0.3113164007663727,
+      "loss_ce": 0.021155260503292084,
+      "loss_iou": 1.0845476388931274,
+      "loss_num": 0.291015625,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 308694760,
+      "step": 1786
+    },
+    {
+      "epoch": 0.6844121026426656,
+      "grad_norm": 9.353929146451705,
+      "learning_rate": 5e-06,
+      "loss": 0.2244,
+      "num_input_tokens_seen": 308867624,
+      "step": 1787
+    },
+    {
+      "epoch": 0.6844121026426656,
+      "loss": 0.20658767223358154,
+      "loss_ce": 0.022017350420355797,
+      "loss_iou": 1.0860481262207031,
+      "loss_num": 0.1845703125,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 308867624,
+      "step": 1787
+    },
+    {
+      "epoch": 0.6847950976637304,
+      "grad_norm": 20.58304051139585,
+      "learning_rate": 5e-06,
+      "loss": 0.2084,
+      "num_input_tokens_seen": 309040664,
+      "step": 1788
+    },
+    {
+      "epoch": 0.6847950976637304,
+      "loss": 0.1884603202342987,
+      "loss_ce": 0.02058313600718975,
+      "loss_iou": 1.0026249885559082,
+      "loss_num": 0.16796875,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 309040664,
+      "step": 1788
+    },
+    {
+      "epoch": 0.6851780926847951,
+      "grad_norm": 23.51325793939438,
+      "learning_rate": 5e-06,
+      "loss": 0.2274,
+      "num_input_tokens_seen": 309214056,
+      "step": 1789
+    },
+    {
+      "epoch": 0.6851780926847951,
+      "loss": 0.17779070138931274,
+      "loss_ce": 0.02209002524614334,
+      "loss_iou": 1.0156054496765137,
+      "loss_num": 0.1552734375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 309214056,
+      "step": 1789
+    },
+    {
+      "epoch": 0.6855610877058598,
+      "grad_norm": 30.523427600375143,
+      "learning_rate": 5e-06,
+      "loss": 0.199,
+      "num_input_tokens_seen": 309386984,
+      "step": 1790
+    },
+    {
+      "epoch": 0.6855610877058598,
+      "loss": 0.2529047131538391,
+      "loss_ce": 0.02023870125412941,
+      "loss_iou": 1.097543478012085,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 309386984,
+      "step": 1790
+    },
+    {
+      "epoch": 0.6859440827269245,
+      "grad_norm": 29.3815753506915,
+      "learning_rate": 5e-06,
+      "loss": 0.1741,
+      "num_input_tokens_seen": 309560096,
+      "step": 1791
+    },
+    {
+      "epoch": 0.6859440827269245,
+      "loss": 0.18151524662971497,
+      "loss_ce": 0.019344843924045563,
+      "loss_iou": 1.019117832183838,
+      "loss_num": 0.162109375,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 309560096,
+      "step": 1791
+    },
+    {
+      "epoch": 0.6863270777479893,
+      "grad_norm": 24.416869287090638,
+      "learning_rate": 5e-06,
+      "loss": 0.2331,
+      "num_input_tokens_seen": 309732576,
+      "step": 1792
+    },
+    {
+      "epoch": 0.6863270777479893,
+      "loss": 0.22634467482566833,
+      "loss_ce": 0.019923774525523186,
+      "loss_iou": 1.1221668720245361,
+      "loss_num": 0.2060546875,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 309732576,
+      "step": 1792
+    },
+    {
+      "epoch": 0.686710072769054,
+      "grad_norm": 20.21171192055158,
+      "learning_rate": 5e-06,
+      "loss": 0.1818,
+      "num_input_tokens_seen": 309905544,
+      "step": 1793
+    },
+    {
+      "epoch": 0.686710072769054,
+      "loss": 0.19531100988388062,
+      "loss_ce": 0.022032195702195168,
+      "loss_iou": 1.0145184993743896,
+      "loss_num": 0.1728515625,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 309905544,
+      "step": 1793
+    },
+    {
+      "epoch": 0.6870930677901187,
+      "grad_norm": 29.931546959836826,
+      "learning_rate": 5e-06,
+      "loss": 0.2151,
+      "num_input_tokens_seen": 310078512,
+      "step": 1794
+    },
+    {
+      "epoch": 0.6870930677901187,
+      "loss": 0.2022952437400818,
+      "loss_ce": 0.02041049674153328,
+      "loss_iou": 1.0460946559906006,
+      "loss_num": 0.181640625,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 310078512,
+      "step": 1794
+    },
+    {
+      "epoch": 0.6874760628111835,
+      "grad_norm": 27.17543643661829,
+      "learning_rate": 5e-06,
+      "loss": 0.1886,
+      "num_input_tokens_seen": 310251408,
+      "step": 1795
+    },
+    {
+      "epoch": 0.6874760628111835,
+      "loss": 0.17125827074050903,
+      "loss_ce": 0.02123386040329933,
+      "loss_iou": 1.0009284019470215,
+      "loss_num": 0.150390625,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 310251408,
+      "step": 1795
+    },
+    {
+      "epoch": 0.6878590578322482,
+      "grad_norm": 25.98153473046399,
+      "learning_rate": 5e-06,
+      "loss": 0.2399,
+      "num_input_tokens_seen": 310424544,
+      "step": 1796
+    },
+    {
+      "epoch": 0.6878590578322482,
+      "loss": 0.23470944166183472,
+      "loss_ce": 0.019804662093520164,
+      "loss_iou": 1.0204904079437256,
+      "loss_num": 0.21484375,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 310424544,
+      "step": 1796
+    },
+    {
+      "epoch": 0.6882420528533129,
+      "grad_norm": 36.81538122837507,
+      "learning_rate": 5e-06,
+      "loss": 0.2462,
+      "num_input_tokens_seen": 310597288,
+      "step": 1797
+    },
+    {
+      "epoch": 0.6882420528533129,
+      "loss": 0.271190345287323,
+      "loss_ce": 0.020335841923952103,
+      "loss_iou": 1.472503900527954,
+      "loss_num": 0.25,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 310597288,
+      "step": 1797
+    },
+    {
+      "epoch": 0.6886250478743776,
+      "grad_norm": 47.644600023114414,
+      "learning_rate": 5e-06,
+      "loss": 0.2223,
+      "num_input_tokens_seen": 310770488,
+      "step": 1798
+    },
+    {
+      "epoch": 0.6886250478743776,
+      "loss": 0.2012127935886383,
+      "loss_ce": 0.0221966914832592,
+      "loss_iou": 1.0174134969711304,
+      "loss_num": 0.1787109375,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 310770488,
+      "step": 1798
+    },
+    {
+      "epoch": 0.6890080428954424,
+      "grad_norm": 39.08401946318235,
+      "learning_rate": 5e-06,
+      "loss": 0.174,
+      "num_input_tokens_seen": 310943440,
+      "step": 1799
+    },
+    {
+      "epoch": 0.6890080428954424,
+      "loss": 0.15548494458198547,
+      "loss_ce": 0.020902413874864578,
+      "loss_iou": 1.0032386779785156,
+      "loss_num": 0.134765625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 310943440,
+      "step": 1799
+    },
+    {
+      "epoch": 0.6893910379165071,
+      "grad_norm": 42.43240130592587,
+      "learning_rate": 5e-06,
+      "loss": 0.25,
+      "num_input_tokens_seen": 311116304,
+      "step": 1800
+    },
+    {
+      "epoch": 0.6893910379165071,
+      "loss": 0.2081068456172943,
+      "loss_ce": 0.018653731793165207,
+      "loss_iou": 1.005466103553772,
+      "loss_num": 0.189453125,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 311116304,
+      "step": 1800
+    },
+    {
+      "epoch": 0.6897740329375718,
+      "grad_norm": 16.92622312624107,
+      "learning_rate": 5e-06,
+      "loss": 0.1629,
+      "num_input_tokens_seen": 311289352,
+      "step": 1801
+    },
+    {
+      "epoch": 0.6897740329375718,
+      "loss": 0.15594685077667236,
+      "loss_ce": 0.022188305854797363,
+      "loss_iou": 1.0031026601791382,
+      "loss_num": 0.1337890625,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 311289352,
+      "step": 1801
+    },
+    {
+      "epoch": 0.6901570279586365,
+      "grad_norm": 43.27051799407796,
+      "learning_rate": 5e-06,
+      "loss": 0.2678,
+      "num_input_tokens_seen": 311462320,
+      "step": 1802
+    },
+    {
+      "epoch": 0.6901570279586365,
+      "loss": 0.2979928255081177,
+      "loss_ce": 0.021747712045907974,
+      "loss_iou": 1.0230367183685303,
+      "loss_num": 0.275390625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 311462320,
+      "step": 1802
+    },
+    {
+      "epoch": 0.6905400229797013,
+      "grad_norm": 32.77732264245229,
+      "learning_rate": 5e-06,
+      "loss": 0.2891,
+      "num_input_tokens_seen": 311634952,
+      "step": 1803
+    },
+    {
+      "epoch": 0.6905400229797013,
+      "loss": 0.30508124828338623,
+      "loss_ce": 0.02175602689385414,
+      "loss_iou": 1.0113251209259033,
+      "loss_num": 0.283203125,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 311634952,
+      "step": 1803
+    },
+    {
+      "epoch": 0.6909230180007659,
+      "grad_norm": 25.16295713300046,
+      "learning_rate": 5e-06,
+      "loss": 0.2448,
+      "num_input_tokens_seen": 311807592,
+      "step": 1804
+    },
+    {
+      "epoch": 0.6909230180007659,
+      "loss": 0.2753583490848541,
+      "loss_ce": 0.02114691585302353,
+      "loss_iou": 1.1480212211608887,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 311807592,
+      "step": 1804
+    },
+    {
+      "epoch": 0.6913060130218307,
+      "grad_norm": 33.56581977308756,
+      "learning_rate": 5e-06,
+      "loss": 0.2076,
+      "num_input_tokens_seen": 311980744,
+      "step": 1805
+    },
+    {
+      "epoch": 0.6913060130218307,
+      "loss": 0.2249014675617218,
+      "loss_ce": 0.0222647525370121,
+      "loss_iou": 1.1057695150375366,
+      "loss_num": 0.203125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 311980744,
+      "step": 1805
+    },
+    {
+      "epoch": 0.6916890080428955,
+      "grad_norm": 40.95398966890601,
+      "learning_rate": 5e-06,
+      "loss": 0.2382,
+      "num_input_tokens_seen": 312153968,
+      "step": 1806
+    },
+    {
+      "epoch": 0.6916890080428955,
+      "loss": 0.20139385759830475,
+      "loss_ce": 0.02146221324801445,
+      "loss_iou": 1.0483115911483765,
+      "loss_num": 0.1796875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 312153968,
+      "step": 1806
+    },
+    {
+      "epoch": 0.6920720030639602,
+      "grad_norm": 33.06774960828792,
+      "learning_rate": 5e-06,
+      "loss": 0.3074,
+      "num_input_tokens_seen": 312327064,
+      "step": 1807
+    },
+    {
+      "epoch": 0.6920720030639602,
+      "loss": 0.2533792555332184,
+      "loss_ce": 0.020102884620428085,
+      "loss_iou": 1.029458999633789,
+      "loss_num": 0.2333984375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 312327064,
+      "step": 1807
+    },
+    {
+      "epoch": 0.6924549980850249,
+      "grad_norm": 20.167685865150624,
+      "learning_rate": 5e-06,
+      "loss": 0.2446,
+      "num_input_tokens_seen": 312499512,
+      "step": 1808
+    },
+    {
+      "epoch": 0.6924549980850249,
+      "loss": 0.2071533352136612,
+      "loss_ce": 0.020019549876451492,
+      "loss_iou": 1.005505084991455,
+      "loss_num": 0.1875,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 312499512,
+      "step": 1808
+    },
+    {
+      "epoch": 0.6928379931060896,
+      "grad_norm": 24.88942268196023,
+      "learning_rate": 5e-06,
+      "loss": 0.1927,
+      "num_input_tokens_seen": 312672160,
+      "step": 1809
+    },
+    {
+      "epoch": 0.6928379931060896,
+      "loss": 0.2139686644077301,
+      "loss_ce": 0.020944979041814804,
+      "loss_iou": 1.0078048706054688,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 312672160,
+      "step": 1809
+    },
+    {
+      "epoch": 0.6932209881271544,
+      "grad_norm": 54.577292414133005,
+      "learning_rate": 5e-06,
+      "loss": 0.2115,
+      "num_input_tokens_seen": 312845208,
+      "step": 1810
+    },
+    {
+      "epoch": 0.6932209881271544,
+      "loss": 0.2643311619758606,
+      "loss_ce": 0.021838489919900894,
+      "loss_iou": 1.1271297931671143,
+      "loss_num": 0.2421875,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 312845208,
+      "step": 1810
+    },
+    {
+      "epoch": 0.693603983148219,
+      "grad_norm": 34.267886907261975,
+      "learning_rate": 5e-06,
+      "loss": 0.2612,
+      "num_input_tokens_seen": 313018064,
+      "step": 1811
+    },
+    {
+      "epoch": 0.693603983148219,
+      "loss": 0.26771610975265503,
+      "loss_ce": 0.020889950916171074,
+      "loss_iou": 1.0351500511169434,
+      "loss_num": 0.2470703125,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 313018064,
+      "step": 1811
+    },
+    {
+      "epoch": 0.6939869781692838,
+      "grad_norm": 27.288532665173932,
+      "learning_rate": 5e-06,
+      "loss": 0.2585,
+      "num_input_tokens_seen": 313191096,
+      "step": 1812
+    },
+    {
+      "epoch": 0.6939869781692838,
+      "loss": 0.2542651295661926,
+      "loss_ce": 0.021538063883781433,
+      "loss_iou": 1.0879662036895752,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 313191096,
+      "step": 1812
+    },
+    {
+      "epoch": 0.6943699731903485,
+      "grad_norm": 29.010078146165668,
+      "learning_rate": 5e-06,
+      "loss": 0.2039,
+      "num_input_tokens_seen": 313364240,
+      "step": 1813
+    },
+    {
+      "epoch": 0.6943699731903485,
+      "loss": 0.24573346972465515,
+      "loss_ce": 0.020941000431776047,
+      "loss_iou": 1.0150185823440552,
+      "loss_num": 0.224609375,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 313364240,
+      "step": 1813
+    },
+    {
+      "epoch": 0.6947529682114133,
+      "grad_norm": 53.61927297149612,
+      "learning_rate": 5e-06,
+      "loss": 0.2497,
+      "num_input_tokens_seen": 313536896,
+      "step": 1814
+    },
+    {
+      "epoch": 0.6947529682114133,
+      "loss": 0.25504958629608154,
+      "loss_ce": 0.02018630877137184,
+      "loss_iou": 1.1167068481445312,
+      "loss_num": 0.234375,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 313536896,
+      "step": 1814
+    },
+    {
+      "epoch": 0.695135963232478,
+      "grad_norm": 38.93817465803134,
+      "learning_rate": 5e-06,
+      "loss": 0.2548,
+      "num_input_tokens_seen": 313710240,
+      "step": 1815
+    },
+    {
+      "epoch": 0.695135963232478,
+      "loss": 0.24499168992042542,
+      "loss_ce": 0.02245749533176422,
+      "loss_iou": 1.0315313339233398,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 313710240,
+      "step": 1815
+    },
+    {
+      "epoch": 0.6955189582535427,
+      "grad_norm": 31.376630992015357,
+      "learning_rate": 5e-06,
+      "loss": 0.2778,
+      "num_input_tokens_seen": 313882832,
+      "step": 1816
+    },
+    {
+      "epoch": 0.6955189582535427,
+      "loss": 0.3155968189239502,
+      "loss_ce": 0.018355632200837135,
+      "loss_iou": NaN,
+      "loss_num": 0.296875,
+      "loss_xval": 0.296875,
+      "num_input_tokens_seen": 313882832,
+      "step": 1816
+    },
+    {
+      "epoch": 0.6959019532746075,
+      "grad_norm": 25.29873565702416,
+      "learning_rate": 5e-06,
+      "loss": 0.2453,
+      "num_input_tokens_seen": 314055344,
+      "step": 1817
+    },
+    {
+      "epoch": 0.6959019532746075,
+      "loss": 0.25351792573928833,
+      "loss_ce": 0.022560909390449524,
+      "loss_iou": 1.0375151634216309,
+      "loss_num": 0.23046875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 314055344,
+      "step": 1817
+    },
+    {
+      "epoch": 0.6962849482956721,
+      "grad_norm": 26.80187544021142,
+      "learning_rate": 5e-06,
+      "loss": 0.1737,
+      "num_input_tokens_seen": 314228136,
+      "step": 1818
+    },
+    {
+      "epoch": 0.6962849482956721,
+      "loss": 0.14994150400161743,
+      "loss_ce": 0.02188975363969803,
+      "loss_iou": 1.000547170639038,
+      "loss_num": 0.1279296875,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 314228136,
+      "step": 1818
+    },
+    {
+      "epoch": 0.6966679433167369,
+      "grad_norm": 28.61940095339164,
+      "learning_rate": 5e-06,
+      "loss": 0.2436,
+      "num_input_tokens_seen": 314401296,
+      "step": 1819
+    },
+    {
+      "epoch": 0.6966679433167369,
+      "loss": 0.21377557516098022,
+      "loss_ce": 0.020782414823770523,
+      "loss_iou": 1.0035618543624878,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 314401296,
+      "step": 1819
+    },
+    {
+      "epoch": 0.6970509383378016,
+      "grad_norm": 17.3206459206108,
+      "learning_rate": 5e-06,
+      "loss": 0.2735,
+      "num_input_tokens_seen": 314574384,
+      "step": 1820
+    },
+    {
+      "epoch": 0.6970509383378016,
+      "loss": 0.2638574540615082,
+      "loss_ce": 0.02240237034857273,
+      "loss_iou": 1.1314351558685303,
+      "loss_num": 0.2412109375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 314574384,
+      "step": 1820
+    },
+    {
+      "epoch": 0.6974339333588664,
+      "grad_norm": 27.36915259376651,
+      "learning_rate": 5e-06,
+      "loss": 0.2481,
+      "num_input_tokens_seen": 314747208,
+      "step": 1821
+    },
+    {
+      "epoch": 0.6974339333588664,
+      "loss": 0.24881714582443237,
+      "loss_ce": 0.020728763192892075,
+      "loss_iou": 1.0145410299301147,
+      "loss_num": 0.228515625,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 314747208,
+      "step": 1821
+    },
+    {
+      "epoch": 0.697816928379931,
+      "grad_norm": 22.691723151541613,
+      "learning_rate": 5e-06,
+      "loss": 0.1891,
+      "num_input_tokens_seen": 314920240,
+      "step": 1822
+    },
+    {
+      "epoch": 0.697816928379931,
+      "loss": 0.16145747900009155,
+      "loss_ce": 0.0218700859695673,
+      "loss_iou": 1.024095058441162,
+      "loss_num": 0.1396484375,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 314920240,
+      "step": 1822
+    },
+    {
+      "epoch": 0.6981999234009958,
+      "grad_norm": 22.46967825224015,
+      "learning_rate": 5e-06,
+      "loss": 0.278,
+      "num_input_tokens_seen": 315093088,
+      "step": 1823
+    },
+    {
+      "epoch": 0.6981999234009958,
+      "loss": 0.3342197835445404,
+      "loss_ce": 0.021292559802532196,
+      "loss_iou": 1.044093370437622,
+      "loss_num": 0.3125,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 315093088,
+      "step": 1823
+    },
+    {
+      "epoch": 0.6985829184220606,
+      "grad_norm": 24.523147278668453,
+      "learning_rate": 5e-06,
+      "loss": 0.2306,
+      "num_input_tokens_seen": 315266040,
+      "step": 1824
+    },
+    {
+      "epoch": 0.6985829184220606,
+      "loss": 0.21536186337471008,
+      "loss_ce": 0.02139214053750038,
+      "loss_iou": 1.0824241638183594,
+      "loss_num": 0.1943359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 315266040,
+      "step": 1824
+    },
+    {
+      "epoch": 0.6989659134431252,
+      "grad_norm": 33.30640836205073,
+      "learning_rate": 5e-06,
+      "loss": 0.2435,
+      "num_input_tokens_seen": 315439136,
+      "step": 1825
+    },
+    {
+      "epoch": 0.6989659134431252,
+      "loss": 0.28745371103286743,
+      "loss_ce": 0.02207283303141594,
+      "loss_iou": 1.0710875988006592,
+      "loss_num": 0.265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 315439136,
+      "step": 1825
+    },
+    {
+      "epoch": 0.69934890846419,
+      "grad_norm": 25.15528350612063,
+      "learning_rate": 5e-06,
+      "loss": 0.2096,
+      "num_input_tokens_seen": 315611704,
+      "step": 1826
+    },
+    {
+      "epoch": 0.69934890846419,
+      "loss": 0.21007584035396576,
+      "loss_ce": 0.021233057603240013,
+      "loss_iou": 1.0091652870178223,
+      "loss_num": 0.1884765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 315611704,
+      "step": 1826
+    },
+    {
+      "epoch": 0.6997319034852547,
+      "grad_norm": 30.373965426663016,
+      "learning_rate": 5e-06,
+      "loss": 0.2322,
+      "num_input_tokens_seen": 315781056,
+      "step": 1827
+    },
+    {
+      "epoch": 0.6997319034852547,
+      "loss": 0.2667866349220276,
+      "loss_ce": 0.022279784083366394,
+      "loss_iou": 1.012694001197815,
+      "loss_num": 0.244140625,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 315781056,
+      "step": 1827
+    },
+    {
+      "epoch": 0.7001148985063195,
+      "grad_norm": 23.609229512710332,
+      "learning_rate": 5e-06,
+      "loss": 0.2367,
+      "num_input_tokens_seen": 315954008,
+      "step": 1828
+    },
+    {
+      "epoch": 0.7001148985063195,
+      "loss": 0.20945066213607788,
+      "loss_ce": 0.020180631428956985,
+      "loss_iou": 1.009779453277588,
+      "loss_num": 0.189453125,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 315954008,
+      "step": 1828
+    },
+    {
+      "epoch": 0.7004978935273841,
+      "grad_norm": 61.88931626384124,
+      "learning_rate": 5e-06,
+      "loss": 0.2644,
+      "num_input_tokens_seen": 316126888,
+      "step": 1829
+    },
+    {
+      "epoch": 0.7004978935273841,
+      "loss": 0.2504710257053375,
+      "loss_ce": 0.020490553230047226,
+      "loss_iou": 1.0115447044372559,
+      "loss_num": 0.23046875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 316126888,
+      "step": 1829
+    },
+    {
+      "epoch": 0.7008808885484489,
+      "grad_norm": 33.63699569721759,
+      "learning_rate": 5e-06,
+      "loss": 0.2755,
+      "num_input_tokens_seen": 316299888,
+      "step": 1830
+    },
+    {
+      "epoch": 0.7008808885484489,
+      "loss": 0.31528377532958984,
+      "loss_ce": 0.02097224071621895,
+      "loss_iou": 1.0388249158859253,
+      "loss_num": 0.294921875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 316299888,
+      "step": 1830
+    },
+    {
+      "epoch": 0.7012638835695136,
+      "grad_norm": 36.98355302940735,
+      "learning_rate": 5e-06,
+      "loss": 0.3499,
+      "num_input_tokens_seen": 316473136,
+      "step": 1831
+    },
+    {
+      "epoch": 0.7012638835695136,
+      "loss": 0.40260884165763855,
+      "loss_ce": 0.023702584207057953,
+      "loss_iou": 1.1157183647155762,
+      "loss_num": 0.37890625,
+      "loss_xval": 0.37890625,
+      "num_input_tokens_seen": 316473136,
+      "step": 1831
+    },
+    {
+      "epoch": 0.7016468785905783,
+      "grad_norm": 56.72892788926891,
+      "learning_rate": 5e-06,
+      "loss": 0.2425,
+      "num_input_tokens_seen": 316645952,
+      "step": 1832
+    },
+    {
+      "epoch": 0.7016468785905783,
+      "loss": 0.22687043249607086,
+      "loss_ce": 0.020205387845635414,
+      "loss_iou": 1.0851229429244995,
+      "loss_num": 0.20703125,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 316645952,
+      "step": 1832
+    },
+    {
+      "epoch": 0.702029873611643,
+      "grad_norm": 64.42947426499191,
+      "learning_rate": 5e-06,
+      "loss": 0.3272,
+      "num_input_tokens_seen": 316818576,
+      "step": 1833
+    },
+    {
+      "epoch": 0.702029873611643,
+      "loss": 0.3310290575027466,
+      "loss_ce": 0.020970463752746582,
+      "loss_iou": 1.009353518486023,
+      "loss_num": 0.310546875,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 316818576,
+      "step": 1833
+    },
+    {
+      "epoch": 0.7024128686327078,
+      "grad_norm": 33.30955134153118,
+      "learning_rate": 5e-06,
+      "loss": 0.2896,
+      "num_input_tokens_seen": 316991216,
+      "step": 1834
+    },
+    {
+      "epoch": 0.7024128686327078,
+      "loss": 0.2690102458000183,
+      "loss_ce": 0.022061984986066818,
+      "loss_iou": 1.0346126556396484,
+      "loss_num": 0.2470703125,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 316991216,
+      "step": 1834
+    },
+    {
+      "epoch": 0.7027958636537724,
+      "grad_norm": 45.28812474396357,
+      "learning_rate": 5e-06,
+      "loss": 0.3274,
+      "num_input_tokens_seen": 317164296,
+      "step": 1835
+    },
+    {
+      "epoch": 0.7027958636537724,
+      "loss": 0.3138014078140259,
+      "loss_ce": 0.021076807752251625,
+      "loss_iou": 1.0114707946777344,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 317164296,
+      "step": 1835
+    },
+    {
+      "epoch": 0.7031788586748372,
+      "grad_norm": 16.852578736279813,
+      "learning_rate": 5e-06,
+      "loss": 0.2851,
+      "num_input_tokens_seen": 317337200,
+      "step": 1836
+    },
+    {
+      "epoch": 0.7031788586748372,
+      "loss": 0.29935815930366516,
+      "loss_ce": 0.020427484065294266,
+      "loss_iou": 1.0910439491271973,
+      "loss_num": 0.279296875,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 317337200,
+      "step": 1836
+    },
+    {
+      "epoch": 0.703561853695902,
+      "grad_norm": 41.01459174752276,
+      "learning_rate": 5e-06,
+      "loss": 0.1961,
+      "num_input_tokens_seen": 317510032,
+      "step": 1837
+    },
+    {
+      "epoch": 0.703561853695902,
+      "loss": 0.2029310017824173,
+      "loss_ce": 0.021595552563667297,
+      "loss_iou": 1.0131685733795166,
+      "loss_num": 0.181640625,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 317510032,
+      "step": 1837
+    },
+    {
+      "epoch": 0.7039448487169667,
+      "grad_norm": 41.26333970809996,
+      "learning_rate": 5e-06,
+      "loss": 0.2812,
+      "num_input_tokens_seen": 317683112,
+      "step": 1838
+    },
+    {
+      "epoch": 0.7039448487169667,
+      "loss": 0.2858147621154785,
+      "loss_ce": 0.021288391202688217,
+      "loss_iou": 1.279576063156128,
+      "loss_num": 0.263671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 317683112,
+      "step": 1838
+    },
+    {
+      "epoch": 0.7043278437380314,
+      "grad_norm": 32.12157863098567,
+      "learning_rate": 5e-06,
+      "loss": 0.2464,
+      "num_input_tokens_seen": 317856480,
+      "step": 1839
+    },
+    {
+      "epoch": 0.7043278437380314,
+      "loss": 0.1840820014476776,
+      "loss_ce": 0.020568817853927612,
+      "loss_iou": 0.8792381882667542,
+      "loss_num": 0.1630859375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 317856480,
+      "step": 1839
+    },
+    {
+      "epoch": 0.7047108387590961,
+      "grad_norm": 24.574644804661407,
+      "learning_rate": 5e-06,
+      "loss": 0.2033,
+      "num_input_tokens_seen": 318029536,
+      "step": 1840
+    },
+    {
+      "epoch": 0.7047108387590961,
+      "loss": 0.1571667194366455,
+      "loss_ce": 0.020936254411935806,
+      "loss_iou": 1.0036258697509766,
+      "loss_num": 0.13671875,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 318029536,
+      "step": 1840
+    },
+    {
+      "epoch": 0.7050938337801609,
+      "grad_norm": 26.059741678743233,
+      "learning_rate": 5e-06,
+      "loss": 0.1947,
+      "num_input_tokens_seen": 318202800,
+      "step": 1841
+    },
+    {
+      "epoch": 0.7050938337801609,
+      "loss": 0.18011049926280975,
+      "loss_ce": 0.022456692531704903,
+      "loss_iou": 1.0436346530914307,
+      "loss_num": 0.1572265625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 318202800,
+      "step": 1841
+    },
+    {
+      "epoch": 0.7054768288012255,
+      "grad_norm": 33.16696966465949,
+      "learning_rate": 5e-06,
+      "loss": 0.17,
+      "num_input_tokens_seen": 318375920,
+      "step": 1842
+    },
+    {
+      "epoch": 0.7054768288012255,
+      "loss": 0.15799319744110107,
+      "loss_ce": 0.021640656515955925,
+      "loss_iou": 1.0010097026824951,
+      "loss_num": 0.13671875,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 318375920,
+      "step": 1842
+    },
+    {
+      "epoch": 0.7058598238222903,
+      "grad_norm": 37.9307528576101,
+      "learning_rate": 5e-06,
+      "loss": 0.2035,
+      "num_input_tokens_seen": 318548752,
+      "step": 1843
+    },
+    {
+      "epoch": 0.7058598238222903,
+      "loss": 0.20799294114112854,
+      "loss_ce": 0.020737089216709137,
+      "loss_iou": 1.0322823524475098,
+      "loss_num": 0.1875,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 318548752,
+      "step": 1843
+    },
+    {
+      "epoch": 0.706242818843355,
+      "grad_norm": 28.351593090408336,
+      "learning_rate": 5e-06,
+      "loss": 0.2193,
+      "num_input_tokens_seen": 318722048,
+      "step": 1844
+    },
+    {
+      "epoch": 0.706242818843355,
+      "loss": 0.24400952458381653,
+      "loss_ce": 0.020498784258961678,
+      "loss_iou": 1.0954004526138306,
+      "loss_num": 0.2236328125,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 318722048,
+      "step": 1844
+    },
+    {
+      "epoch": 0.7066258138644198,
+      "grad_norm": 16.71906040148047,
+      "learning_rate": 5e-06,
+      "loss": 0.2042,
+      "num_input_tokens_seen": 318895464,
+      "step": 1845
+    },
+    {
+      "epoch": 0.7066258138644198,
+      "loss": 0.18017390370368958,
+      "loss_ce": 0.01830865815281868,
+      "loss_iou": 1.0080845355987549,
+      "loss_num": 0.162109375,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 318895464,
+      "step": 1845
+    },
+    {
+      "epoch": 0.7070088088854845,
+      "grad_norm": 25.977744715712003,
+      "learning_rate": 5e-06,
+      "loss": 0.2046,
+      "num_input_tokens_seen": 319068416,
+      "step": 1846
+    },
+    {
+      "epoch": 0.7070088088854845,
+      "loss": 0.25116094946861267,
+      "loss_ce": 0.020692206919193268,
+      "loss_iou": 1.0209722518920898,
+      "loss_num": 0.23046875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 319068416,
+      "step": 1846
+    },
+    {
+      "epoch": 0.7073918039065492,
+      "grad_norm": 34.94968383525336,
+      "learning_rate": 5e-06,
+      "loss": 0.2244,
+      "num_input_tokens_seen": 319241712,
+      "step": 1847
+    },
+    {
+      "epoch": 0.7073918039065492,
+      "loss": 0.2732456922531128,
+      "loss_ce": 0.022208085283637047,
+      "loss_iou": 1.0894484519958496,
+      "loss_num": 0.251953125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 319241712,
+      "step": 1847
+    },
+    {
+      "epoch": 0.707774798927614,
+      "grad_norm": 38.78415559673452,
+      "learning_rate": 5e-06,
+      "loss": 0.1989,
+      "num_input_tokens_seen": 319414632,
+      "step": 1848
+    },
+    {
+      "epoch": 0.707774798927614,
+      "loss": 0.22200219333171844,
+      "loss_ce": 0.021074455231428146,
+      "loss_iou": 1.0549986362457275,
+      "loss_num": 0.201171875,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 319414632,
+      "step": 1848
+    },
+    {
+      "epoch": 0.7081577939486786,
+      "grad_norm": 37.47710799659906,
+      "learning_rate": 5e-06,
+      "loss": 0.19,
+      "num_input_tokens_seen": 319587448,
+      "step": 1849
+    },
+    {
+      "epoch": 0.7081577939486786,
+      "loss": 0.1949232965707779,
+      "loss_ce": 0.01975240930914879,
+      "loss_iou": 1.0029751062393188,
+      "loss_num": 0.1748046875,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 319587448,
+      "step": 1849
+    },
+    {
+      "epoch": 0.7085407889697434,
+      "grad_norm": 24.402401517247196,
+      "learning_rate": 5e-06,
+      "loss": 0.2155,
+      "num_input_tokens_seen": 319760048,
+      "step": 1850
+    },
+    {
+      "epoch": 0.7085407889697434,
+      "loss": 0.17069105803966522,
+      "loss_ce": 0.02188735269010067,
+      "loss_iou": 1.0251952409744263,
+      "loss_num": 0.1484375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 319760048,
+      "step": 1850
+    },
+    {
+      "epoch": 0.7089237839908081,
+      "grad_norm": 29.130151725933946,
+      "learning_rate": 5e-06,
+      "loss": 0.2035,
+      "num_input_tokens_seen": 319932864,
+      "step": 1851
+    },
+    {
+      "epoch": 0.7089237839908081,
+      "loss": 0.18228460848331451,
+      "loss_ce": 0.022067327052354813,
+      "loss_iou": 1.0114346742630005,
+      "loss_num": 0.16015625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 319932864,
+      "step": 1851
+    },
+    {
+      "epoch": 0.7093067790118729,
+      "grad_norm": 45.23779309242795,
+      "learning_rate": 5e-06,
+      "loss": 0.2543,
+      "num_input_tokens_seen": 320105680,
+      "step": 1852
+    },
+    {
+      "epoch": 0.7093067790118729,
+      "loss": 0.24232217669487,
+      "loss_ce": 0.02186320349574089,
+      "loss_iou": 1.1184134483337402,
+      "loss_num": 0.220703125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 320105680,
+      "step": 1852
+    },
+    {
+      "epoch": 0.7096897740329375,
+      "grad_norm": 33.19348338625306,
+      "learning_rate": 5e-06,
+      "loss": 0.2521,
+      "num_input_tokens_seen": 320278488,
+      "step": 1853
+    },
+    {
+      "epoch": 0.7096897740329375,
+      "loss": 0.2258143424987793,
+      "loss_ce": 0.021102434024214745,
+      "loss_iou": 1.023998737335205,
+      "loss_num": 0.205078125,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 320278488,
+      "step": 1853
+    },
+    {
+      "epoch": 0.7100727690540023,
+      "grad_norm": 30.516140167944577,
+      "learning_rate": 5e-06,
+      "loss": 0.1968,
+      "num_input_tokens_seen": 320451320,
+      "step": 1854
+    },
+    {
+      "epoch": 0.7100727690540023,
+      "loss": 0.23307326436042786,
+      "loss_ce": 0.020854037255048752,
+      "loss_iou": 1.1666905879974365,
+      "loss_num": 0.2119140625,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 320451320,
+      "step": 1854
+    },
+    {
+      "epoch": 0.710455764075067,
+      "grad_norm": 29.262022286823285,
+      "learning_rate": 5e-06,
+      "loss": 0.2492,
+      "num_input_tokens_seen": 320624088,
+      "step": 1855
+    },
+    {
+      "epoch": 0.710455764075067,
+      "loss": 0.21036887168884277,
+      "loss_ce": 0.01987816020846367,
+      "loss_iou": 1.0528923273086548,
+      "loss_num": 0.1904296875,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 320624088,
+      "step": 1855
+    },
+    {
+      "epoch": 0.7108387590961317,
+      "grad_norm": 22.6309814105481,
+      "learning_rate": 5e-06,
+      "loss": 0.2209,
+      "num_input_tokens_seen": 320796304,
+      "step": 1856
+    },
+    {
+      "epoch": 0.7108387590961317,
+      "loss": 0.1989680528640747,
+      "loss_ce": 0.019402626901865005,
+      "loss_iou": 1.003929615020752,
+      "loss_num": 0.1796875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 320796304,
+      "step": 1856
+    },
+    {
+      "epoch": 0.7112217541171965,
+      "grad_norm": 27.431224869874097,
+      "learning_rate": 5e-06,
+      "loss": 0.1652,
+      "num_input_tokens_seen": 320969272,
+      "step": 1857
+    },
+    {
+      "epoch": 0.7112217541171965,
+      "loss": 0.1824040412902832,
+      "loss_ce": 0.0225529782474041,
+      "loss_iou": 1.026893973350525,
+      "loss_num": 0.16015625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 320969272,
+      "step": 1857
+    },
+    {
+      "epoch": 0.7116047491382612,
+      "grad_norm": 33.80062001300053,
+      "learning_rate": 5e-06,
+      "loss": 0.2347,
+      "num_input_tokens_seen": 321142200,
+      "step": 1858
+    },
+    {
+      "epoch": 0.7116047491382612,
+      "loss": 0.20400670170783997,
+      "loss_ce": 0.021511578932404518,
+      "loss_iou": 1.0098191499710083,
+      "loss_num": 0.1826171875,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 321142200,
+      "step": 1858
+    },
+    {
+      "epoch": 0.711987744159326,
+      "grad_norm": 28.330625937192586,
+      "learning_rate": 5e-06,
+      "loss": 0.2163,
+      "num_input_tokens_seen": 321314720,
+      "step": 1859
+    },
+    {
+      "epoch": 0.711987744159326,
+      "loss": 0.2437838613986969,
+      "loss_ce": 0.020700372755527496,
+      "loss_iou": 1.153388500213623,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 321314720,
+      "step": 1859
+    },
+    {
+      "epoch": 0.7123707391803906,
+      "grad_norm": 40.836253157253,
+      "learning_rate": 5e-06,
+      "loss": 0.1979,
+      "num_input_tokens_seen": 321487200,
+      "step": 1860
+    },
+    {
+      "epoch": 0.7123707391803906,
+      "loss": 0.22483478486537933,
+      "loss_ce": 0.02140461653470993,
+      "loss_iou": 1.0260980129241943,
+      "loss_num": 0.203125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 321487200,
+      "step": 1860
+    },
+    {
+      "epoch": 0.7127537342014554,
+      "grad_norm": 30.395535199677283,
+      "learning_rate": 5e-06,
+      "loss": 0.2491,
+      "num_input_tokens_seen": 321659920,
+      "step": 1861
+    },
+    {
+      "epoch": 0.7127537342014554,
+      "loss": 0.2886390686035156,
+      "loss_ce": 0.02032851055264473,
+      "loss_iou": 1.3753141164779663,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 321659920,
+      "step": 1861
+    },
+    {
+      "epoch": 0.7131367292225201,
+      "grad_norm": 22.248818786729604,
+      "learning_rate": 5e-06,
+      "loss": 0.2034,
+      "num_input_tokens_seen": 321832856,
+      "step": 1862
+    },
+    {
+      "epoch": 0.7131367292225201,
+      "loss": 0.15091675519943237,
+      "loss_ce": 0.021400148048996925,
+      "loss_iou": 1.0252573490142822,
+      "loss_num": 0.1298828125,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 321832856,
+      "step": 1862
+    },
+    {
+      "epoch": 0.7135197242435848,
+      "grad_norm": 28.365412603846075,
+      "learning_rate": 5e-06,
+      "loss": 0.2122,
+      "num_input_tokens_seen": 322005928,
+      "step": 1863
+    },
+    {
+      "epoch": 0.7135197242435848,
+      "loss": 0.2460293173789978,
+      "loss_ce": 0.021419944241642952,
+      "loss_iou": 1.1003503799438477,
+      "loss_num": 0.224609375,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 322005928,
+      "step": 1863
+    },
+    {
+      "epoch": 0.7139027192646495,
+      "grad_norm": 28.21811543717518,
+      "learning_rate": 5e-06,
+      "loss": 0.2566,
+      "num_input_tokens_seen": 322179088,
+      "step": 1864
+    },
+    {
+      "epoch": 0.7139027192646495,
+      "loss": 0.2528541684150696,
+      "loss_ce": 0.02250749245285988,
+      "loss_iou": 3.2350127696990967,
+      "loss_num": 0.23046875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 322179088,
+      "step": 1864
+    },
+    {
+      "epoch": 0.7142857142857143,
+      "grad_norm": 31.58919824994602,
+      "learning_rate": 5e-06,
+      "loss": 0.3049,
+      "num_input_tokens_seen": 322352536,
+      "step": 1865
+    },
+    {
+      "epoch": 0.7142857142857143,
+      "loss": 0.2377358078956604,
+      "loss_ce": 0.021976524963974953,
+      "loss_iou": 1.0187640190124512,
+      "loss_num": 0.2158203125,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 322352536,
+      "step": 1865
+    },
+    {
+      "epoch": 0.7146687093067791,
+      "grad_norm": 23.940216788480484,
+      "learning_rate": 5e-06,
+      "loss": 0.1936,
+      "num_input_tokens_seen": 322525600,
+      "step": 1866
+    },
+    {
+      "epoch": 0.7146687093067791,
+      "loss": 0.17934811115264893,
+      "loss_ce": 0.02083982154726982,
+      "loss_iou": 1.0212841033935547,
+      "loss_num": 0.158203125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 322525600,
+      "step": 1866
+    },
+    {
+      "epoch": 0.7150517043278437,
+      "grad_norm": 23.618919057048874,
+      "learning_rate": 5e-06,
+      "loss": 0.1859,
+      "num_input_tokens_seen": 322698872,
+      "step": 1867
+    },
+    {
+      "epoch": 0.7150517043278437,
+      "loss": 0.20185860991477966,
+      "loss_ce": 0.021988004446029663,
+      "loss_iou": 1.0116420984268188,
+      "loss_num": 0.1796875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 322698872,
+      "step": 1867
+    },
+    {
+      "epoch": 0.7154346993489085,
+      "grad_norm": 26.95212945883852,
+      "learning_rate": 5e-06,
+      "loss": 0.2307,
+      "num_input_tokens_seen": 322871712,
+      "step": 1868
+    },
+    {
+      "epoch": 0.7154346993489085,
+      "loss": 0.19863513112068176,
+      "loss_ce": 0.021999388933181763,
+      "loss_iou": 1.0071381330490112,
+      "loss_num": 0.1767578125,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 322871712,
+      "step": 1868
+    },
+    {
+      "epoch": 0.7158176943699732,
+      "grad_norm": 36.59885840226958,
+      "learning_rate": 5e-06,
+      "loss": 0.1883,
+      "num_input_tokens_seen": 323044928,
+      "step": 1869
+    },
+    {
+      "epoch": 0.7158176943699732,
+      "loss": 0.17399346828460693,
+      "loss_ce": 0.020306944847106934,
+      "loss_iou": 1.0215654373168945,
+      "loss_num": 0.1533203125,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 323044928,
+      "step": 1869
+    },
+    {
+      "epoch": 0.7162006893910379,
+      "grad_norm": 40.02371575368256,
+      "learning_rate": 5e-06,
+      "loss": 0.2284,
+      "num_input_tokens_seen": 323217896,
+      "step": 1870
+    },
+    {
+      "epoch": 0.7162006893910379,
+      "loss": 0.2157425433397293,
+      "loss_ce": 0.02000279538333416,
+      "loss_iou": 1.035560131072998,
+      "loss_num": 0.1953125,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 323217896,
+      "step": 1870
+    },
+    {
+      "epoch": 0.7165836844121026,
+      "grad_norm": 21.78368644754905,
+      "learning_rate": 5e-06,
+      "loss": 0.1449,
+      "num_input_tokens_seen": 323390552,
+      "step": 1871
+    },
+    {
+      "epoch": 0.7165836844121026,
+      "loss": 0.1488848626613617,
+      "loss_ce": 0.020039651542901993,
+      "loss_iou": 1.0301177501678467,
+      "loss_num": 0.12890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 323390552,
+      "step": 1871
+    },
+    {
+      "epoch": 0.7169666794331674,
+      "grad_norm": 30.65881465222365,
+      "learning_rate": 5e-06,
+      "loss": 0.2183,
+      "num_input_tokens_seen": 323563552,
+      "step": 1872
+    },
+    {
+      "epoch": 0.7169666794331674,
+      "loss": 0.17817732691764832,
+      "loss_ce": 0.021622147411108017,
+      "loss_iou": 1.0330085754394531,
+      "loss_num": 0.15625,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 323563552,
+      "step": 1872
+    },
+    {
+      "epoch": 0.7173496744542321,
+      "grad_norm": 37.830110764306,
+      "learning_rate": 5e-06,
+      "loss": 0.2421,
+      "num_input_tokens_seen": 323736632,
+      "step": 1873
+    },
+    {
+      "epoch": 0.7173496744542321,
+      "loss": 0.2265729159116745,
+      "loss_ce": 0.021250661462545395,
+      "loss_iou": 0.9831060171127319,
+      "loss_num": 0.205078125,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 323736632,
+      "step": 1873
+    },
+    {
+      "epoch": 0.7177326694752968,
+      "grad_norm": 30.63227223544062,
+      "learning_rate": 5e-06,
+      "loss": 0.1998,
+      "num_input_tokens_seen": 323909224,
+      "step": 1874
+    },
+    {
+      "epoch": 0.7177326694752968,
+      "loss": 0.2136291265487671,
+      "loss_ce": 0.021734602749347687,
+      "loss_iou": 1.0560487508773804,
+      "loss_num": 0.19140625,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 323909224,
+      "step": 1874
+    },
+    {
+      "epoch": 0.7181156644963615,
+      "grad_norm": 27.650254576505585,
+      "learning_rate": 5e-06,
+      "loss": 0.3444,
+      "num_input_tokens_seen": 324082240,
+      "step": 1875
+    },
+    {
+      "epoch": 0.7181156644963615,
+      "loss": 0.4103952646255493,
+      "loss_ce": 0.02208959311246872,
+      "loss_iou": 1.0925495624542236,
+      "loss_num": 0.388671875,
+      "loss_xval": 0.388671875,
+      "num_input_tokens_seen": 324082240,
+      "step": 1875
+    },
+    {
+      "epoch": 0.7184986595174263,
+      "grad_norm": 21.903677973358782,
+      "learning_rate": 5e-06,
+      "loss": 0.2315,
+      "num_input_tokens_seen": 324252064,
+      "step": 1876
+    },
+    {
+      "epoch": 0.7184986595174263,
+      "loss": 0.24588297307491302,
+      "loss_ce": 0.022738447412848473,
+      "loss_iou": 1.1731048822402954,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 324252064,
+      "step": 1876
+    },
+    {
+      "epoch": 0.718881654538491,
+      "grad_norm": 42.13102796256304,
+      "learning_rate": 5e-06,
+      "loss": 0.2257,
+      "num_input_tokens_seen": 324425096,
+      "step": 1877
+    },
+    {
+      "epoch": 0.718881654538491,
+      "loss": 0.2502211928367615,
+      "loss_ce": 0.019874490797519684,
+      "loss_iou": 1.0776121616363525,
+      "loss_num": 0.23046875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 324425096,
+      "step": 1877
+    },
+    {
+      "epoch": 0.7192646495595557,
+      "grad_norm": 34.698582698143234,
+      "learning_rate": 5e-06,
+      "loss": 0.1942,
+      "num_input_tokens_seen": 324597704,
+      "step": 1878
+    },
+    {
+      "epoch": 0.7192646495595557,
+      "loss": 0.2114652544260025,
+      "loss_ce": 0.019265543669462204,
+      "loss_iou": 1.0166583061218262,
+      "loss_num": 0.1923828125,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 324597704,
+      "step": 1878
+    },
+    {
+      "epoch": 0.7196476445806205,
+      "grad_norm": 27.7722452880696,
+      "learning_rate": 5e-06,
+      "loss": 0.2278,
+      "num_input_tokens_seen": 324770808,
+      "step": 1879
+    },
+    {
+      "epoch": 0.7196476445806205,
+      "loss": 0.2010706663131714,
+      "loss_ce": 0.02144422009587288,
+      "loss_iou": 1.0378485918045044,
+      "loss_num": 0.1796875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 324770808,
+      "step": 1879
+    },
+    {
+      "epoch": 0.7200306396016852,
+      "grad_norm": 25.019436539960555,
+      "learning_rate": 5e-06,
+      "loss": 0.1939,
+      "num_input_tokens_seen": 324943704,
+      "step": 1880
+    },
+    {
+      "epoch": 0.7200306396016852,
+      "loss": 0.14633317291736603,
+      "loss_ce": 0.021089037880301476,
+      "loss_iou": 1.0011621713638306,
+      "loss_num": 0.125,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 324943704,
+      "step": 1880
+    },
+    {
+      "epoch": 0.7204136346227499,
+      "grad_norm": 39.26587788359916,
+      "learning_rate": 5e-06,
+      "loss": 0.2408,
+      "num_input_tokens_seen": 325116648,
+      "step": 1881
+    },
+    {
+      "epoch": 0.7204136346227499,
+      "loss": 0.244556725025177,
+      "loss_ce": 0.020496664568781853,
+      "loss_iou": 1.0621919631958008,
+      "loss_num": 0.2236328125,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 325116648,
+      "step": 1881
+    },
+    {
+      "epoch": 0.7207966296438146,
+      "grad_norm": 21.890345170337966,
+      "learning_rate": 5e-06,
+      "loss": 0.2042,
+      "num_input_tokens_seen": 325289192,
+      "step": 1882
+    },
+    {
+      "epoch": 0.7207966296438146,
+      "loss": 0.2093326300382614,
+      "loss_ce": 0.02018468827009201,
+      "loss_iou": 1.0757040977478027,
+      "loss_num": 0.189453125,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 325289192,
+      "step": 1882
+    },
+    {
+      "epoch": 0.7211796246648794,
+      "grad_norm": 26.583255122561585,
+      "learning_rate": 5e-06,
+      "loss": 0.1908,
+      "num_input_tokens_seen": 325462360,
+      "step": 1883
+    },
+    {
+      "epoch": 0.7211796246648794,
+      "loss": 0.21546167135238647,
+      "loss_ce": 0.022163324058055878,
+      "loss_iou": 1.0645512342453003,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 325462360,
+      "step": 1883
+    },
+    {
+      "epoch": 0.721562619685944,
+      "grad_norm": 29.33114335313516,
+      "learning_rate": 5e-06,
+      "loss": 0.236,
+      "num_input_tokens_seen": 325635792,
+      "step": 1884
+    },
+    {
+      "epoch": 0.721562619685944,
+      "loss": 0.3069278299808502,
+      "loss_ce": 0.020367780700325966,
+      "loss_iou": 1.0742062330245972,
+      "loss_num": 0.287109375,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 325635792,
+      "step": 1884
+    },
+    {
+      "epoch": 0.7219456147070088,
+      "grad_norm": 57.63312598078634,
+      "learning_rate": 5e-06,
+      "loss": 0.2272,
+      "num_input_tokens_seen": 325808752,
+      "step": 1885
+    },
+    {
+      "epoch": 0.7219456147070088,
+      "loss": 0.20381572842597961,
+      "loss_ce": 0.020405080169439316,
+      "loss_iou": 1.0015591382980347,
+      "loss_num": 0.18359375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 325808752,
+      "step": 1885
+    },
+    {
+      "epoch": 0.7223286097280736,
+      "grad_norm": 19.83901232484663,
+      "learning_rate": 5e-06,
+      "loss": 0.2271,
+      "num_input_tokens_seen": 325981624,
+      "step": 1886
+    },
+    {
+      "epoch": 0.7223286097280736,
+      "loss": 0.17987343668937683,
+      "loss_ce": 0.01959512010216713,
+      "loss_iou": 1.0048749446868896,
+      "loss_num": 0.16015625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 325981624,
+      "step": 1886
+    },
+    {
+      "epoch": 0.7227116047491383,
+      "grad_norm": 23.234290999388904,
+      "learning_rate": 5e-06,
+      "loss": 0.2187,
+      "num_input_tokens_seen": 326154704,
+      "step": 1887
+    },
+    {
+      "epoch": 0.7227116047491383,
+      "loss": 0.22305038571357727,
+      "loss_ce": 0.021939542144536972,
+      "loss_iou": 1.0524241924285889,
+      "loss_num": 0.201171875,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 326154704,
+      "step": 1887
+    },
+    {
+      "epoch": 0.723094599770203,
+      "grad_norm": 36.07468391266884,
+      "learning_rate": 5e-06,
+      "loss": 0.2834,
+      "num_input_tokens_seen": 326327992,
+      "step": 1888
+    },
+    {
+      "epoch": 0.723094599770203,
+      "loss": 0.3624667525291443,
+      "loss_ce": 0.020059533417224884,
+      "loss_iou": 2.5894272327423096,
+      "loss_num": 0.341796875,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 326327992,
+      "step": 1888
+    },
+    {
+      "epoch": 0.7234775947912677,
+      "grad_norm": 37.85224542460828,
+      "learning_rate": 5e-06,
+      "loss": 0.232,
+      "num_input_tokens_seen": 326501080,
+      "step": 1889
+    },
+    {
+      "epoch": 0.7234775947912677,
+      "loss": 0.25145411491394043,
+      "loss_ce": 0.021900873631238937,
+      "loss_iou": 1.0371770858764648,
+      "loss_num": 0.2294921875,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 326501080,
+      "step": 1889
+    },
+    {
+      "epoch": 0.7238605898123325,
+      "grad_norm": 29.64032047382248,
+      "learning_rate": 5e-06,
+      "loss": 0.1923,
+      "num_input_tokens_seen": 326674160,
+      "step": 1890
+    },
+    {
+      "epoch": 0.7238605898123325,
+      "loss": 0.17930102348327637,
+      "loss_ce": 0.019877180457115173,
+      "loss_iou": 1.0070887804031372,
+      "loss_num": 0.1591796875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 326674160,
+      "step": 1890
+    },
+    {
+      "epoch": 0.7242435848333971,
+      "grad_norm": 22.305280094687475,
+      "learning_rate": 5e-06,
+      "loss": 0.2314,
+      "num_input_tokens_seen": 326847016,
+      "step": 1891
+    },
+    {
+      "epoch": 0.7242435848333971,
+      "loss": 0.2490444779396057,
+      "loss_ce": 0.01973540522158146,
+      "loss_iou": 1.1225407123565674,
+      "loss_num": 0.2294921875,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 326847016,
+      "step": 1891
+    },
+    {
+      "epoch": 0.7246265798544619,
+      "grad_norm": 13.35396760968828,
+      "learning_rate": 5e-06,
+      "loss": 0.2064,
+      "num_input_tokens_seen": 327019808,
+      "step": 1892
+    },
+    {
+      "epoch": 0.7246265798544619,
+      "loss": 0.2213289737701416,
+      "loss_ce": 0.0208895206451416,
+      "loss_iou": 1.1386830806732178,
+      "loss_num": 0.2001953125,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 327019808,
+      "step": 1892
+    },
+    {
+      "epoch": 0.7250095748755266,
+      "grad_norm": 17.63824320302357,
+      "learning_rate": 5e-06,
+      "loss": 0.1918,
+      "num_input_tokens_seen": 327192560,
+      "step": 1893
+    },
+    {
+      "epoch": 0.7250095748755266,
+      "loss": 0.15080958604812622,
+      "loss_ce": 0.01982813887298107,
+      "loss_iou": 1.00714111328125,
+      "loss_num": 0.130859375,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 327192560,
+      "step": 1893
+    },
+    {
+      "epoch": 0.7253925698965914,
+      "grad_norm": 14.81202091255723,
+      "learning_rate": 5e-06,
+      "loss": 0.1812,
+      "num_input_tokens_seen": 327365376,
+      "step": 1894
+    },
+    {
+      "epoch": 0.7253925698965914,
+      "loss": 0.14532124996185303,
+      "loss_ce": 0.020229686051607132,
+      "loss_iou": 1.0008666515350342,
+      "loss_num": 0.125,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 327365376,
+      "step": 1894
+    },
+    {
+      "epoch": 0.725775564917656,
+      "grad_norm": 32.98841455217743,
+      "learning_rate": 5e-06,
+      "loss": 0.1703,
+      "num_input_tokens_seen": 327538480,
+      "step": 1895
+    },
+    {
+      "epoch": 0.725775564917656,
+      "loss": 0.17655575275421143,
+      "loss_ce": 0.020641446113586426,
+      "loss_iou": 1.0238258838653564,
+      "loss_num": 0.15625,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 327538480,
+      "step": 1895
+    },
+    {
+      "epoch": 0.7261585599387208,
+      "grad_norm": 27.088272969106512,
+      "learning_rate": 5e-06,
+      "loss": 0.2131,
+      "num_input_tokens_seen": 327711344,
+      "step": 1896
+    },
+    {
+      "epoch": 0.7261585599387208,
+      "loss": 0.2655426263809204,
+      "loss_ce": 0.019876133650541306,
+      "loss_iou": 1.0179781913757324,
+      "loss_num": 0.24609375,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 327711344,
+      "step": 1896
+    },
+    {
+      "epoch": 0.7265415549597856,
+      "grad_norm": 24.266018201396527,
+      "learning_rate": 5e-06,
+      "loss": 0.2471,
+      "num_input_tokens_seen": 327884464,
+      "step": 1897
+    },
+    {
+      "epoch": 0.7265415549597856,
+      "loss": 0.23851393163204193,
+      "loss_ce": 0.01976393163204193,
+      "loss_iou": 1.0578659772872925,
+      "loss_num": 0.21875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 327884464,
+      "step": 1897
+    },
+    {
+      "epoch": 0.7269245499808502,
+      "grad_norm": 32.69504654197707,
+      "learning_rate": 5e-06,
+      "loss": 0.2307,
+      "num_input_tokens_seen": 328057616,
+      "step": 1898
+    },
+    {
+      "epoch": 0.7269245499808502,
+      "loss": 0.28195953369140625,
+      "loss_ce": 0.02188875898718834,
+      "loss_iou": 1.0293887853622437,
+      "loss_num": 0.259765625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 328057616,
+      "step": 1898
+    },
+    {
+      "epoch": 0.727307545001915,
+      "grad_norm": 31.398790680254006,
+      "learning_rate": 5e-06,
+      "loss": 0.2314,
+      "num_input_tokens_seen": 328230576,
+      "step": 1899
+    },
+    {
+      "epoch": 0.727307545001915,
+      "loss": 0.25293177366256714,
+      "loss_ce": 0.020387839525938034,
+      "loss_iou": 1.1067075729370117,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 328230576,
+      "step": 1899
+    },
+    {
+      "epoch": 0.7276905400229797,
+      "grad_norm": 31.319907783873976,
+      "learning_rate": 5e-06,
+      "loss": 0.2518,
+      "num_input_tokens_seen": 328403432,
+      "step": 1900
+    },
+    {
+      "epoch": 0.7276905400229797,
+      "loss": 0.23616424202919006,
+      "loss_ce": 0.02168671041727066,
+      "loss_iou": 1.0616157054901123,
+      "loss_num": 0.21484375,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 328403432,
+      "step": 1900
+    },
+    {
+      "epoch": 0.7280735350440444,
+      "grad_norm": 22.96159105155981,
+      "learning_rate": 5e-06,
+      "loss": 0.2045,
+      "num_input_tokens_seen": 328576248,
+      "step": 1901
+    },
+    {
+      "epoch": 0.7280735350440444,
+      "loss": 0.1811392605304718,
+      "loss_ce": 0.0214712955057621,
+      "loss_iou": 1.0072622299194336,
+      "loss_num": 0.16015625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 328576248,
+      "step": 1901
+    },
+    {
+      "epoch": 0.7284565300651091,
+      "grad_norm": 20.793484682660253,
+      "learning_rate": 5e-06,
+      "loss": 0.2069,
+      "num_input_tokens_seen": 328748744,
+      "step": 1902
+    },
+    {
+      "epoch": 0.7284565300651091,
+      "loss": 0.22401505708694458,
+      "loss_ce": 0.019913481548428535,
+      "loss_iou": 1.0367412567138672,
+      "loss_num": 0.2041015625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 328748744,
+      "step": 1902
+    },
+    {
+      "epoch": 0.7288395250861739,
+      "grad_norm": 24.684177600987404,
+      "learning_rate": 5e-06,
+      "loss": 0.2624,
+      "num_input_tokens_seen": 328921688,
+      "step": 1903
+    },
+    {
+      "epoch": 0.7288395250861739,
+      "loss": 0.298135906457901,
+      "loss_ce": 0.022501137107610703,
+      "loss_iou": 1.0930315256118774,
+      "loss_num": 0.275390625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 328921688,
+      "step": 1903
+    },
+    {
+      "epoch": 0.7292225201072386,
+      "grad_norm": 33.94493969385388,
+      "learning_rate": 5e-06,
+      "loss": 0.1772,
+      "num_input_tokens_seen": 329094792,
+      "step": 1904
+    },
+    {
+      "epoch": 0.7292225201072386,
+      "loss": 0.18944339454174042,
+      "loss_ce": 0.021596720442175865,
+      "loss_iou": 1.0274012088775635,
+      "loss_num": 0.16796875,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 329094792,
+      "step": 1904
+    },
+    {
+      "epoch": 0.7296055151283033,
+      "grad_norm": 39.59546038720797,
+      "learning_rate": 5e-06,
+      "loss": 0.2316,
+      "num_input_tokens_seen": 329267496,
+      "step": 1905
+    },
+    {
+      "epoch": 0.7296055151283033,
+      "loss": 0.21068190038204193,
+      "loss_ce": 0.02177809551358223,
+      "loss_iou": 1.0081746578216553,
+      "loss_num": 0.1884765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 329267496,
+      "step": 1905
+    },
+    {
+      "epoch": 0.729988510149368,
+      "grad_norm": 43.32140474351385,
+      "learning_rate": 5e-06,
+      "loss": 0.284,
+      "num_input_tokens_seen": 329440488,
+      "step": 1906
+    },
+    {
+      "epoch": 0.729988510149368,
+      "loss": 0.23827320337295532,
+      "loss_ce": 0.02019459754228592,
+      "loss_iou": 1.0973848104476929,
+      "loss_num": 0.2177734375,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 329440488,
+      "step": 1906
+    },
+    {
+      "epoch": 0.7303715051704328,
+      "grad_norm": 30.18629047529966,
+      "learning_rate": 5e-06,
+      "loss": 0.2078,
+      "num_input_tokens_seen": 329613352,
+      "step": 1907
+    },
+    {
+      "epoch": 0.7303715051704328,
+      "loss": 0.20912249386310577,
+      "loss_ce": 0.020279716700315475,
+      "loss_iou": 1.0031514167785645,
+      "loss_num": 0.1884765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 329613352,
+      "step": 1907
+    },
+    {
+      "epoch": 0.7307545001914975,
+      "grad_norm": 18.667252988879593,
+      "learning_rate": 5e-06,
+      "loss": 0.2115,
+      "num_input_tokens_seen": 329786320,
+      "step": 1908
+    },
+    {
+      "epoch": 0.7307545001914975,
+      "loss": 0.19140136241912842,
+      "loss_ce": 0.020319821313023567,
+      "loss_iou": 1.0130813121795654,
+      "loss_num": 0.1708984375,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 329786320,
+      "step": 1908
+    },
+    {
+      "epoch": 0.7311374952125622,
+      "grad_norm": 18.289709185667753,
+      "learning_rate": 5e-06,
+      "loss": 0.1699,
+      "num_input_tokens_seen": 329958960,
+      "step": 1909
+    },
+    {
+      "epoch": 0.7311374952125622,
+      "loss": 0.21079006791114807,
+      "loss_ce": 0.019933123141527176,
+      "loss_iou": 1.0034202337265015,
+      "loss_num": 0.1904296875,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 329958960,
+      "step": 1909
+    },
+    {
+      "epoch": 0.731520490233627,
+      "grad_norm": 35.93743772055923,
+      "learning_rate": 5e-06,
+      "loss": 0.1935,
+      "num_input_tokens_seen": 330131696,
+      "step": 1910
+    },
+    {
+      "epoch": 0.731520490233627,
+      "loss": 0.1739504188299179,
+      "loss_ce": 0.01861594431102276,
+      "loss_iou": 0.8773469924926758,
+      "loss_num": 0.1552734375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 330131696,
+      "step": 1910
+    },
+    {
+      "epoch": 0.7319034852546917,
+      "grad_norm": 48.448821400523876,
+      "learning_rate": 5e-06,
+      "loss": 0.2859,
+      "num_input_tokens_seen": 330303048,
+      "step": 1911
+    },
+    {
+      "epoch": 0.7319034852546917,
+      "loss": 0.23737116158008575,
+      "loss_ce": 0.020085997879505157,
+      "loss_iou": 1.020316243171692,
+      "loss_num": 0.216796875,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 330303048,
+      "step": 1911
+    },
+    {
+      "epoch": 0.7322864802757564,
+      "grad_norm": 23.65524386702177,
+      "learning_rate": 5e-06,
+      "loss": 0.1487,
+      "num_input_tokens_seen": 330476160,
+      "step": 1912
+    },
+    {
+      "epoch": 0.7322864802757564,
+      "loss": 0.14482766389846802,
+      "loss_ce": 0.021628201007843018,
+      "loss_iou": 1.0011117458343506,
+      "loss_num": 0.123046875,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 330476160,
+      "step": 1912
+    },
+    {
+      "epoch": 0.7326694752968211,
+      "grad_norm": 50.27152919345598,
+      "learning_rate": 5e-06,
+      "loss": 0.2478,
+      "num_input_tokens_seen": 330649200,
+      "step": 1913
+    },
+    {
+      "epoch": 0.7326694752968211,
+      "loss": 0.28704240918159485,
+      "loss_ce": 0.022027764469385147,
+      "loss_iou": 1.0086801052093506,
+      "loss_num": 0.265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 330649200,
+      "step": 1913
+    },
+    {
+      "epoch": 0.7330524703178859,
+      "grad_norm": 28.28467450339639,
+      "learning_rate": 5e-06,
+      "loss": 0.1824,
+      "num_input_tokens_seen": 330822152,
+      "step": 1914
+    },
+    {
+      "epoch": 0.7330524703178859,
+      "loss": 0.18633687496185303,
+      "loss_ce": 0.020565396174788475,
+      "loss_iou": 1.0290091037750244,
+      "loss_num": 0.166015625,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 330822152,
+      "step": 1914
+    },
+    {
+      "epoch": 0.7334354653389505,
+      "grad_norm": 74.82632620370264,
+      "learning_rate": 5e-06,
+      "loss": 0.3198,
+      "num_input_tokens_seen": 330995384,
+      "step": 1915
+    },
+    {
+      "epoch": 0.7334354653389505,
+      "loss": 0.3004084825515747,
+      "loss_ce": 0.01976882293820381,
+      "loss_iou": 1.0028400421142578,
+      "loss_num": 0.28125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 330995384,
+      "step": 1915
+    },
+    {
+      "epoch": 0.7338184603600153,
+      "grad_norm": 36.30834955459456,
+      "learning_rate": 5e-06,
+      "loss": 0.2722,
+      "num_input_tokens_seen": 331168296,
+      "step": 1916
+    },
+    {
+      "epoch": 0.7338184603600153,
+      "loss": 0.24254938960075378,
+      "loss_ce": 0.022334547713398933,
+      "loss_iou": 1.0026772022247314,
+      "loss_num": 0.220703125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 331168296,
+      "step": 1916
+    },
+    {
+      "epoch": 0.7342014553810801,
+      "grad_norm": 29.075128758724365,
+      "learning_rate": 5e-06,
+      "loss": 0.2522,
+      "num_input_tokens_seen": 331341696,
+      "step": 1917
+    },
+    {
+      "epoch": 0.7342014553810801,
+      "loss": 0.25283104181289673,
+      "loss_ce": 0.021019525825977325,
+      "loss_iou": 1.015965461730957,
+      "loss_num": 0.2314453125,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 331341696,
+      "step": 1917
+    },
+    {
+      "epoch": 0.7345844504021448,
+      "grad_norm": 51.62003870462262,
+      "learning_rate": 5e-06,
+      "loss": 0.2467,
+      "num_input_tokens_seen": 331514488,
+      "step": 1918
+    },
+    {
+      "epoch": 0.7345844504021448,
+      "loss": 0.23413214087486267,
+      "loss_ce": 0.02258429303765297,
+      "loss_iou": 1.0972702503204346,
+      "loss_num": 0.2119140625,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 331514488,
+      "step": 1918
+    },
+    {
+      "epoch": 0.7349674454232095,
+      "grad_norm": 68.56112628782104,
+      "learning_rate": 5e-06,
+      "loss": 0.419,
+      "num_input_tokens_seen": 331687288,
+      "step": 1919
+    },
+    {
+      "epoch": 0.7349674454232095,
+      "loss": 0.40661534667015076,
+      "loss_ce": 0.01916416361927986,
+      "loss_iou": 0.9524319171905518,
+      "loss_num": 0.38671875,
+      "loss_xval": 0.38671875,
+      "num_input_tokens_seen": 331687288,
+      "step": 1919
+    },
+    {
+      "epoch": 0.7353504404442742,
+      "grad_norm": 30.10748045309126,
+      "learning_rate": 5e-06,
+      "loss": 0.2931,
+      "num_input_tokens_seen": 331860520,
+      "step": 1920
+    },
+    {
+      "epoch": 0.7353504404442742,
+      "loss": 0.3277454078197479,
+      "loss_ce": 0.022325461730360985,
+      "loss_iou": 1.275111436843872,
+      "loss_num": 0.3046875,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 331860520,
+      "step": 1920
+    },
+    {
+      "epoch": 0.735733435465339,
+      "grad_norm": 51.05229778968739,
+      "learning_rate": 5e-06,
+      "loss": 0.4039,
+      "num_input_tokens_seen": 332033792,
+      "step": 1921
+    },
+    {
+      "epoch": 0.735733435465339,
+      "loss": 0.37663334608078003,
+      "loss_ce": 0.022629449144005775,
+      "loss_iou": 1.0174347162246704,
+      "loss_num": 0.353515625,
+      "loss_xval": 0.353515625,
+      "num_input_tokens_seen": 332033792,
+      "step": 1921
+    },
+    {
+      "epoch": 0.7361164304864036,
+      "grad_norm": 56.819231729650525,
+      "learning_rate": 5e-06,
+      "loss": 0.3581,
+      "num_input_tokens_seen": 332206664,
+      "step": 1922
+    },
+    {
+      "epoch": 0.7361164304864036,
+      "loss": 0.37505754828453064,
+      "loss_ce": 0.020443283021450043,
+      "loss_iou": 1.0403156280517578,
+      "loss_num": 0.35546875,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 332206664,
+      "step": 1922
+    },
+    {
+      "epoch": 0.7364994255074684,
+      "grad_norm": 67.59315787006065,
+      "learning_rate": 5e-06,
+      "loss": 0.5422,
+      "num_input_tokens_seen": 332379312,
+      "step": 1923
+    },
+    {
+      "epoch": 0.7364994255074684,
+      "loss": 0.4992562234401703,
+      "loss_ce": 0.02025231346487999,
+      "loss_iou": 1.0131621360778809,
+      "loss_num": 0.478515625,
+      "loss_xval": 0.478515625,
+      "num_input_tokens_seen": 332379312,
+      "step": 1923
+    },
+    {
+      "epoch": 0.7368824205285331,
+      "grad_norm": 41.83252474139575,
+      "learning_rate": 5e-06,
+      "loss": 0.3726,
+      "num_input_tokens_seen": 332552456,
+      "step": 1924
+    },
+    {
+      "epoch": 0.7368824205285331,
+      "loss": 0.3864128291606903,
+      "loss_ce": 0.021666740998625755,
+      "loss_iou": 0.998197615146637,
+      "loss_num": 0.365234375,
+      "loss_xval": 0.365234375,
+      "num_input_tokens_seen": 332552456,
+      "step": 1924
+    },
+    {
+      "epoch": 0.7372654155495979,
+      "grad_norm": 43.10657188034638,
+      "learning_rate": 5e-06,
+      "loss": 0.3381,
+      "num_input_tokens_seen": 332725344,
+      "step": 1925
+    },
+    {
+      "epoch": 0.7372654155495979,
+      "loss": 0.3305189609527588,
+      "loss_ce": 0.022779714316129684,
+      "loss_iou": 1.039806842803955,
+      "loss_num": 0.30859375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 332725344,
+      "step": 1925
+    },
+    {
+      "epoch": 0.7376484105706625,
+      "grad_norm": 180.95190202755572,
+      "learning_rate": 5e-06,
+      "loss": 1.355,
+      "num_input_tokens_seen": 332898656,
+      "step": 1926
+    },
+    {
+      "epoch": 0.7376484105706625,
+      "loss": 1.3647723197937012,
+      "loss_ce": 0.021022357046604156,
+      "loss_iou": 1.2992215156555176,
+      "loss_num": 1.34375,
+      "loss_xval": 1.34375,
+      "num_input_tokens_seen": 332898656,
+      "step": 1926
+    },
+    {
+      "epoch": 0.7380314055917273,
+      "grad_norm": 35.60054618008287,
+      "learning_rate": 5e-06,
+      "loss": 0.2432,
+      "num_input_tokens_seen": 333071768,
+      "step": 1927
+    },
+    {
+      "epoch": 0.7380314055917273,
+      "loss": 0.2595188021659851,
+      "loss_ce": 0.01904027909040451,
+      "loss_iou": 0.883613109588623,
+      "loss_num": 0.240234375,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 333071768,
+      "step": 1927
+    },
+    {
+      "epoch": 0.7384144006127921,
+      "grad_norm": 109.72271360750891,
+      "learning_rate": 5e-06,
+      "loss": 0.6054,
+      "num_input_tokens_seen": 333245000,
+      "step": 1928
+    },
+    {
+      "epoch": 0.7384144006127921,
+      "loss": 0.6654654741287231,
+      "loss_ce": 0.020445970818400383,
+      "loss_iou": 1.1310943365097046,
+      "loss_num": 0.64453125,
+      "loss_xval": 0.64453125,
+      "num_input_tokens_seen": 333245000,
+      "step": 1928
+    },
+    {
+      "epoch": 0.7387973956338567,
+      "grad_norm": 82.77122485843303,
+      "learning_rate": 5e-06,
+      "loss": 0.7423,
+      "num_input_tokens_seen": 333418072,
+      "step": 1929
+    },
+    {
+      "epoch": 0.7387973956338567,
+      "loss": 0.7344993948936462,
+      "loss_ce": 0.021120460703969002,
+      "loss_iou": 1.038887619972229,
+      "loss_num": 0.71484375,
+      "loss_xval": 0.71484375,
+      "num_input_tokens_seen": 333418072,
+      "step": 1929
+    },
+    {
+      "epoch": 0.7391803906549215,
+      "grad_norm": 23.992720364107736,
+      "learning_rate": 5e-06,
+      "loss": 0.3149,
+      "num_input_tokens_seen": 333591104,
+      "step": 1930
+    },
+    {
+      "epoch": 0.7391803906549215,
+      "loss": 0.3694052994251251,
+      "loss_ce": 0.021016621962189674,
+      "loss_iou": 1.0462157726287842,
+      "loss_num": 0.34765625,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 333591104,
+      "step": 1930
+    },
+    {
+      "epoch": 0.7395633856759862,
+      "grad_norm": 33.174950142789754,
+      "learning_rate": 5e-06,
+      "loss": 0.3172,
+      "num_input_tokens_seen": 333763832,
+      "step": 1931
+    },
+    {
+      "epoch": 0.7395633856759862,
+      "loss": 0.3304111957550049,
+      "loss_ce": 0.02010846883058548,
+      "loss_iou": 1.093118667602539,
+      "loss_num": 0.310546875,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 333763832,
+      "step": 1931
+    },
+    {
+      "epoch": 0.739946380697051,
+      "grad_norm": 27.50218950073205,
+      "learning_rate": 5e-06,
+      "loss": 0.2231,
+      "num_input_tokens_seen": 333936968,
+      "step": 1932
+    },
+    {
+      "epoch": 0.739946380697051,
+      "loss": 0.25011956691741943,
+      "loss_ce": 0.022702565416693687,
+      "loss_iou": 1.086707592010498,
+      "loss_num": 0.2275390625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 333936968,
+      "step": 1932
+    },
+    {
+      "epoch": 0.7403293757181156,
+      "grad_norm": 26.609634414458018,
+      "learning_rate": 5e-06,
+      "loss": 0.2598,
+      "num_input_tokens_seen": 334110008,
+      "step": 1933
+    },
+    {
+      "epoch": 0.7403293757181156,
+      "loss": 0.19299322366714478,
+      "loss_ce": 0.022277887910604477,
+      "loss_iou": 1.0121299028396606,
+      "loss_num": 0.1708984375,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 334110008,
+      "step": 1933
+    },
+    {
+      "epoch": 0.7407123707391804,
+      "grad_norm": 36.71801216282503,
+      "learning_rate": 5e-06,
+      "loss": 0.2352,
+      "num_input_tokens_seen": 334283192,
+      "step": 1934
+    },
+    {
+      "epoch": 0.7407123707391804,
+      "loss": 0.29141005873680115,
+      "loss_ce": 0.021146384999155998,
+      "loss_iou": 1.1183552742004395,
+      "loss_num": 0.26953125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 334283192,
+      "step": 1934
+    },
+    {
+      "epoch": 0.7410953657602452,
+      "grad_norm": 48.04796533209979,
+      "learning_rate": 5e-06,
+      "loss": 0.2181,
+      "num_input_tokens_seen": 334456104,
+      "step": 1935
+    },
+    {
+      "epoch": 0.7410953657602452,
+      "loss": 0.199800044298172,
+      "loss_ce": 0.022004637867212296,
+      "loss_iou": 1.0481548309326172,
+      "loss_num": 0.177734375,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 334456104,
+      "step": 1935
+    },
+    {
+      "epoch": 0.7414783607813098,
+      "grad_norm": 28.96683163057016,
+      "learning_rate": 5e-06,
+      "loss": 0.2257,
+      "num_input_tokens_seen": 334629256,
+      "step": 1936
+    },
+    {
+      "epoch": 0.7414783607813098,
+      "loss": 0.2581450939178467,
+      "loss_ce": 0.021939020603895187,
+      "loss_iou": 1.0524083375930786,
+      "loss_num": 0.236328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 334629256,
+      "step": 1936
+    },
+    {
+      "epoch": 0.7418613558023746,
+      "grad_norm": 23.239962749653266,
+      "learning_rate": 5e-06,
+      "loss": 0.1837,
+      "num_input_tokens_seen": 334802328,
+      "step": 1937
+    },
+    {
+      "epoch": 0.7418613558023746,
+      "loss": 0.1810568869113922,
+      "loss_ce": 0.02321997843682766,
+      "loss_iou": 1.0080535411834717,
+      "loss_num": 0.158203125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 334802328,
+      "step": 1937
+    },
+    {
+      "epoch": 0.7422443508234393,
+      "grad_norm": 22.8872095408136,
+      "learning_rate": 5e-06,
+      "loss": 0.1751,
+      "num_input_tokens_seen": 334975168,
+      "step": 1938
+    },
+    {
+      "epoch": 0.7422443508234393,
+      "loss": 0.17818833887577057,
+      "loss_ce": 0.02047349140048027,
+      "loss_iou": 1.0368616580963135,
+      "loss_num": 0.158203125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 334975168,
+      "step": 1938
+    },
+    {
+      "epoch": 0.7426273458445041,
+      "grad_norm": 27.716447217111558,
+      "learning_rate": 5e-06,
+      "loss": 0.1666,
+      "num_input_tokens_seen": 335144456,
+      "step": 1939
+    },
+    {
+      "epoch": 0.7426273458445041,
+      "loss": 0.17497143149375916,
+      "loss_ce": 0.020552489906549454,
+      "loss_iou": 1.0021817684173584,
+      "loss_num": 0.154296875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 335144456,
+      "step": 1939
+    },
+    {
+      "epoch": 0.7430103408655687,
+      "grad_norm": 25.457089685695703,
+      "learning_rate": 5e-06,
+      "loss": 0.2083,
+      "num_input_tokens_seen": 335317368,
+      "step": 1940
+    },
+    {
+      "epoch": 0.7430103408655687,
+      "loss": 0.18655753135681152,
+      "loss_ce": 0.02017568051815033,
+      "loss_iou": 1.0007985830307007,
+      "loss_num": 0.166015625,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 335317368,
+      "step": 1940
+    },
+    {
+      "epoch": 0.7433933358866335,
+      "grad_norm": 29.218416777853236,
+      "learning_rate": 5e-06,
+      "loss": 0.3297,
+      "num_input_tokens_seen": 335490392,
+      "step": 1941
+    },
+    {
+      "epoch": 0.7433933358866335,
+      "loss": 0.3510556221008301,
+      "loss_ce": 0.021038515493273735,
+      "loss_iou": 1.0548796653747559,
+      "loss_num": 0.330078125,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 335490392,
+      "step": 1941
+    },
+    {
+      "epoch": 0.7437763309076982,
+      "grad_norm": 26.701328298012317,
+      "learning_rate": 5e-06,
+      "loss": 0.1655,
+      "num_input_tokens_seen": 335663552,
+      "step": 1942
+    },
+    {
+      "epoch": 0.7437763309076982,
+      "loss": 0.13223397731781006,
+      "loss_ce": 0.02029549330472946,
+      "loss_iou": 1.0102285146713257,
+      "loss_num": 0.11181640625,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 335663552,
+      "step": 1942
+    },
+    {
+      "epoch": 0.7441593259287629,
+      "grad_norm": 32.80138127728905,
+      "learning_rate": 5e-06,
+      "loss": 0.1888,
+      "num_input_tokens_seen": 335836368,
+      "step": 1943
+    },
+    {
+      "epoch": 0.7441593259287629,
+      "loss": 0.2144605815410614,
+      "loss_ce": 0.021040182560682297,
+      "loss_iou": 1.0209829807281494,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 335836368,
+      "step": 1943
+    },
+    {
+      "epoch": 0.7445423209498276,
+      "grad_norm": 25.188772203269707,
+      "learning_rate": 5e-06,
+      "loss": 0.2231,
+      "num_input_tokens_seen": 336009616,
+      "step": 1944
+    },
+    {
+      "epoch": 0.7445423209498276,
+      "loss": 0.2847317159175873,
+      "loss_ce": 0.021670211106538773,
+      "loss_iou": 1.0225014686584473,
+      "loss_num": 0.263671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 336009616,
+      "step": 1944
+    },
+    {
+      "epoch": 0.7449253159708924,
+      "grad_norm": 27.33898594322862,
+      "learning_rate": 5e-06,
+      "loss": 0.1878,
+      "num_input_tokens_seen": 336182800,
+      "step": 1945
+    },
+    {
+      "epoch": 0.7449253159708924,
+      "loss": 0.18002618849277496,
+      "loss_ce": 0.022982735186815262,
+      "loss_iou": 1.0188391208648682,
+      "loss_num": 0.1572265625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 336182800,
+      "step": 1945
+    },
+    {
+      "epoch": 0.7453083109919572,
+      "grad_norm": 22.219886110299534,
+      "learning_rate": 5e-06,
+      "loss": 0.2346,
+      "num_input_tokens_seen": 336355760,
+      "step": 1946
+    },
+    {
+      "epoch": 0.7453083109919572,
+      "loss": 0.25230154395103455,
+      "loss_ce": 0.020367953926324844,
+      "loss_iou": 1.0885792970657349,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 336355760,
+      "step": 1946
+    },
+    {
+      "epoch": 0.7456913060130218,
+      "grad_norm": 19.908992415199794,
+      "learning_rate": 5e-06,
+      "loss": 0.1841,
+      "num_input_tokens_seen": 336528496,
+      "step": 1947
+    },
+    {
+      "epoch": 0.7456913060130218,
+      "loss": 0.22517076134681702,
+      "loss_ce": 0.021404892206192017,
+      "loss_iou": 1.009037971496582,
+      "loss_num": 0.2041015625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 336528496,
+      "step": 1947
+    },
+    {
+      "epoch": 0.7460743010340866,
+      "grad_norm": 19.781945656679195,
+      "learning_rate": 5e-06,
+      "loss": 0.1946,
+      "num_input_tokens_seen": 336701264,
+      "step": 1948
+    },
+    {
+      "epoch": 0.7460743010340866,
+      "loss": 0.2301415503025055,
+      "loss_ce": 0.021096128970384598,
+      "loss_iou": 3.1631762981414795,
+      "loss_num": 0.208984375,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 336701264,
+      "step": 1948
+    },
+    {
+      "epoch": 0.7464572960551513,
+      "grad_norm": 29.355093963100664,
+      "learning_rate": 5e-06,
+      "loss": 0.2014,
+      "num_input_tokens_seen": 336874368,
+      "step": 1949
+    },
+    {
+      "epoch": 0.7464572960551513,
+      "loss": 0.1448083519935608,
+      "loss_ce": 0.02145630121231079,
+      "loss_iou": 1.012711524963379,
+      "loss_num": 0.12353515625,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 336874368,
+      "step": 1949
+    },
+    {
+      "epoch": 0.746840291076216,
+      "grad_norm": 26.855786936722488,
+      "learning_rate": 5e-06,
+      "loss": 0.2183,
+      "num_input_tokens_seen": 337047448,
+      "step": 1950
+    },
+    {
+      "epoch": 0.746840291076216,
+      "loss": 0.21372899413108826,
+      "loss_ce": 0.021102052181959152,
+      "loss_iou": 1.0027873516082764,
+      "loss_num": 0.1923828125,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 337047448,
+      "step": 1950
+    },
+    {
+      "epoch": 0.7472232860972807,
+      "grad_norm": 29.78256819336044,
+      "learning_rate": 5e-06,
+      "loss": 0.2189,
+      "num_input_tokens_seen": 337220720,
+      "step": 1951
+    },
+    {
+      "epoch": 0.7472232860972807,
+      "loss": 0.22373650968074799,
+      "loss_ce": 0.02109978348016739,
+      "loss_iou": 1.032132863998413,
+      "loss_num": 0.203125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 337220720,
+      "step": 1951
+    },
+    {
+      "epoch": 0.7476062811183455,
+      "grad_norm": 29.64222327377733,
+      "learning_rate": 5e-06,
+      "loss": 0.2249,
+      "num_input_tokens_seen": 337394200,
+      "step": 1952
+    },
+    {
+      "epoch": 0.7476062811183455,
+      "loss": 0.2393241822719574,
+      "loss_ce": 0.019597632810473442,
+      "loss_iou": 1.1529085636138916,
+      "loss_num": 0.2197265625,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 337394200,
+      "step": 1952
+    },
+    {
+      "epoch": 0.7479892761394102,
+      "grad_norm": 29.486507672186317,
+      "learning_rate": 5e-06,
+      "loss": 0.3001,
+      "num_input_tokens_seen": 337567312,
+      "step": 1953
+    },
+    {
+      "epoch": 0.7479892761394102,
+      "loss": 0.2638113796710968,
+      "loss_ce": 0.02296665497124195,
+      "loss_iou": 3.5116450786590576,
+      "loss_num": 0.2412109375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 337567312,
+      "step": 1953
+    },
+    {
+      "epoch": 0.7483722711604749,
+      "grad_norm": 23.535065054104823,
+      "learning_rate": 5e-06,
+      "loss": 0.1708,
+      "num_input_tokens_seen": 337740152,
+      "step": 1954
+    },
+    {
+      "epoch": 0.7483722711604749,
+      "loss": 0.1761426478624344,
+      "loss_ce": 0.021296454593539238,
+      "loss_iou": 1.1424171924591064,
+      "loss_num": 0.1552734375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 337740152,
+      "step": 1954
+    },
+    {
+      "epoch": 0.7487552661815396,
+      "grad_norm": 21.769023134524247,
+      "learning_rate": 5e-06,
+      "loss": 0.1987,
+      "num_input_tokens_seen": 337912816,
+      "step": 1955
+    },
+    {
+      "epoch": 0.7487552661815396,
+      "loss": 0.23964551091194153,
+      "loss_ce": 0.01961377263069153,
+      "loss_iou": 2.957237720489502,
+      "loss_num": 0.2197265625,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 337912816,
+      "step": 1955
+    },
+    {
+      "epoch": 0.7491382612026044,
+      "grad_norm": 14.491886493471908,
+      "learning_rate": 5e-06,
+      "loss": 0.208,
+      "num_input_tokens_seen": 338082056,
+      "step": 1956
+    },
+    {
+      "epoch": 0.7491382612026044,
+      "loss": 0.2334989309310913,
+      "loss_ce": 0.023110752925276756,
+      "loss_iou": 1.1820414066314697,
+      "loss_num": 0.2099609375,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 338082056,
+      "step": 1956
+    },
+    {
+      "epoch": 0.749521256223669,
+      "grad_norm": 22.076057577851937,
+      "learning_rate": 5e-06,
+      "loss": 0.1821,
+      "num_input_tokens_seen": 338255256,
+      "step": 1957
+    },
+    {
+      "epoch": 0.749521256223669,
+      "loss": 0.1340140700340271,
+      "loss_ce": 0.020946452394127846,
+      "loss_iou": 1.0144014358520508,
+      "loss_num": 0.11328125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 338255256,
+      "step": 1957
+    },
+    {
+      "epoch": 0.7499042512447338,
+      "grad_norm": 24.821614162565513,
+      "learning_rate": 5e-06,
+      "loss": 0.1859,
+      "num_input_tokens_seen": 338428080,
+      "step": 1958
+    },
+    {
+      "epoch": 0.7499042512447338,
+      "loss": 0.16101813316345215,
+      "loss_ce": 0.020942451432347298,
+      "loss_iou": 1.0032565593719482,
+      "loss_num": 0.1396484375,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 338428080,
+      "step": 1958
+    },
+    {
+      "epoch": 0.7502872462657986,
+      "grad_norm": 20.563365404112613,
+      "learning_rate": 5e-06,
+      "loss": 0.2758,
+      "num_input_tokens_seen": 338597520,
+      "step": 1959
+    },
+    {
+      "epoch": 0.7502872462657986,
+      "loss": 0.25606077909469604,
+      "loss_ce": 0.0215636994689703,
+      "loss_iou": 1.018204927444458,
+      "loss_num": 0.234375,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 338597520,
+      "step": 1959
+    },
+    {
+      "epoch": 0.7506702412868632,
+      "grad_norm": 21.817141799337517,
+      "learning_rate": 5e-06,
+      "loss": 0.2089,
+      "num_input_tokens_seen": 338770312,
+      "step": 1960
+    },
+    {
+      "epoch": 0.7506702412868632,
+      "loss": 0.232526034116745,
+      "loss_ce": 0.020245764404535294,
+      "loss_iou": 1.0887852907180786,
+      "loss_num": 0.2119140625,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 338770312,
+      "step": 1960
+    },
+    {
+      "epoch": 0.751053236307928,
+      "grad_norm": 29.991984152111403,
+      "learning_rate": 5e-06,
+      "loss": 0.1692,
+      "num_input_tokens_seen": 338943232,
+      "step": 1961
+    },
+    {
+      "epoch": 0.751053236307928,
+      "loss": 0.17735590040683746,
+      "loss_ce": 0.021227972581982613,
+      "loss_iou": 1.0765886306762695,
+      "loss_num": 0.15625,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 338943232,
+      "step": 1961
+    },
+    {
+      "epoch": 0.7514362313289927,
+      "grad_norm": 23.697245978079234,
+      "learning_rate": 5e-06,
+      "loss": 0.1825,
+      "num_input_tokens_seen": 339116448,
+      "step": 1962
+    },
+    {
+      "epoch": 0.7514362313289927,
+      "loss": 0.15435320138931274,
+      "loss_ce": 0.021967953070998192,
+      "loss_iou": 1.001384973526001,
+      "loss_num": 0.1328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 339116448,
+      "step": 1962
+    },
+    {
+      "epoch": 0.7518192263500575,
+      "grad_norm": 18.029364515237333,
+      "learning_rate": 5e-06,
+      "loss": 0.1804,
+      "num_input_tokens_seen": 339289328,
+      "step": 1963
+    },
+    {
+      "epoch": 0.7518192263500575,
+      "loss": 0.13389170169830322,
+      "loss_ce": 0.01783336140215397,
+      "loss_iou": 1.0064783096313477,
+      "loss_num": 0.1162109375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 339289328,
+      "step": 1963
+    },
+    {
+      "epoch": 0.7522022213711221,
+      "grad_norm": 17.664782598935876,
+      "learning_rate": 5e-06,
+      "loss": 0.18,
+      "num_input_tokens_seen": 339462496,
+      "step": 1964
+    },
+    {
+      "epoch": 0.7522022213711221,
+      "loss": 0.17858535051345825,
+      "loss_ce": 0.02080945298075676,
+      "loss_iou": 1.0110020637512207,
+      "loss_num": 0.158203125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 339462496,
+      "step": 1964
+    },
+    {
+      "epoch": 0.7525852163921869,
+      "grad_norm": 28.50489486344704,
+      "learning_rate": 5e-06,
+      "loss": 0.2123,
+      "num_input_tokens_seen": 339635312,
+      "step": 1965
+    },
+    {
+      "epoch": 0.7525852163921869,
+      "loss": 0.2552037537097931,
+      "loss_ce": 0.020035291090607643,
+      "loss_iou": 1.068601131439209,
+      "loss_num": 0.2353515625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 339635312,
+      "step": 1965
+    },
+    {
+      "epoch": 0.7529682114132517,
+      "grad_norm": 25.976001639195506,
+      "learning_rate": 5e-06,
+      "loss": 0.2451,
+      "num_input_tokens_seen": 339808200,
+      "step": 1966
+    },
+    {
+      "epoch": 0.7529682114132517,
+      "loss": 0.2603006064891815,
+      "loss_ce": 0.019211728125810623,
+      "loss_iou": 1.0250359773635864,
+      "loss_num": 0.2412109375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 339808200,
+      "step": 1966
+    },
+    {
+      "epoch": 0.7533512064343163,
+      "grad_norm": 37.32805644716445,
+      "learning_rate": 5e-06,
+      "loss": 0.2289,
+      "num_input_tokens_seen": 339981232,
+      "step": 1967
+    },
+    {
+      "epoch": 0.7533512064343163,
+      "loss": 0.23148807883262634,
+      "loss_ce": 0.021282989531755447,
+      "loss_iou": 1.006373643875122,
+      "loss_num": 0.2099609375,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 339981232,
+      "step": 1967
+    },
+    {
+      "epoch": 0.7537342014553811,
+      "grad_norm": 26.530237429389697,
+      "learning_rate": 5e-06,
+      "loss": 0.2303,
+      "num_input_tokens_seen": 340154240,
+      "step": 1968
+    },
+    {
+      "epoch": 0.7537342014553811,
+      "loss": 0.24449728429317474,
+      "loss_ce": 0.021657921373844147,
+      "loss_iou": 1.009596347808838,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 340154240,
+      "step": 1968
+    },
+    {
+      "epoch": 0.7541171964764458,
+      "grad_norm": 24.563041510666604,
+      "learning_rate": 5e-06,
+      "loss": 0.2157,
+      "num_input_tokens_seen": 340327520,
+      "step": 1969
+    },
+    {
+      "epoch": 0.7541171964764458,
+      "loss": 0.20339317619800568,
+      "loss_ce": 0.020348750054836273,
+      "loss_iou": 1.0349680185317993,
+      "loss_num": 0.1826171875,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 340327520,
+      "step": 1969
+    },
+    {
+      "epoch": 0.7545001914975106,
+      "grad_norm": 26.82092541960977,
+      "learning_rate": 5e-06,
+      "loss": 0.1736,
+      "num_input_tokens_seen": 340500296,
+      "step": 1970
+    },
+    {
+      "epoch": 0.7545001914975106,
+      "loss": 0.19258993864059448,
+      "loss_ce": 0.020592864602804184,
+      "loss_iou": 1.0155099630355835,
+      "loss_num": 0.171875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 340500296,
+      "step": 1970
+    },
+    {
+      "epoch": 0.7548831865185752,
+      "grad_norm": 16.375679650085974,
+      "learning_rate": 5e-06,
+      "loss": 0.246,
+      "num_input_tokens_seen": 340673192,
+      "step": 1971
+    },
+    {
+      "epoch": 0.7548831865185752,
+      "loss": 0.24143922328948975,
+      "loss_ce": 0.018477799370884895,
+      "loss_iou": 1.0040318965911865,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 340673192,
+      "step": 1971
+    },
+    {
+      "epoch": 0.75526618153964,
+      "grad_norm": 17.621575018072924,
+      "learning_rate": 5e-06,
+      "loss": 0.1851,
+      "num_input_tokens_seen": 340846432,
+      "step": 1972
+    },
+    {
+      "epoch": 0.75526618153964,
+      "loss": 0.1742793172597885,
+      "loss_ce": 0.020531758666038513,
+      "loss_iou": 1.0360504388809204,
+      "loss_num": 0.1533203125,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 340846432,
+      "step": 1972
+    },
+    {
+      "epoch": 0.7556491765607047,
+      "grad_norm": 16.324200281366345,
+      "learning_rate": 5e-06,
+      "loss": 0.1853,
+      "num_input_tokens_seen": 341019616,
+      "step": 1973
+    },
+    {
+      "epoch": 0.7556491765607047,
+      "loss": 0.17437125742435455,
+      "loss_ce": 0.01989128440618515,
+      "loss_iou": 1.0057682991027832,
+      "loss_num": 0.154296875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 341019616,
+      "step": 1973
+    },
+    {
+      "epoch": 0.7560321715817694,
+      "grad_norm": 24.37760011723323,
+      "learning_rate": 5e-06,
+      "loss": 0.146,
+      "num_input_tokens_seen": 341192336,
+      "step": 1974
+    },
+    {
+      "epoch": 0.7560321715817694,
+      "loss": 0.13626563549041748,
+      "loss_ce": 0.021305926144123077,
+      "loss_iou": 1.0007812976837158,
+      "loss_num": 0.11474609375,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 341192336,
+      "step": 1974
+    },
+    {
+      "epoch": 0.7564151666028341,
+      "grad_norm": 32.72219329255361,
+      "learning_rate": 5e-06,
+      "loss": 0.2585,
+      "num_input_tokens_seen": 341365312,
+      "step": 1975
+    },
+    {
+      "epoch": 0.7564151666028341,
+      "loss": 0.23727071285247803,
+      "loss_ce": 0.020168669521808624,
+      "loss_iou": 1.013498306274414,
+      "loss_num": 0.216796875,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 341365312,
+      "step": 1975
+    },
+    {
+      "epoch": 0.7567981616238989,
+      "grad_norm": 31.79542505969276,
+      "learning_rate": 5e-06,
+      "loss": 0.2152,
+      "num_input_tokens_seen": 341538392,
+      "step": 1976
+    },
+    {
+      "epoch": 0.7567981616238989,
+      "loss": 0.2574518322944641,
+      "loss_ce": 0.020757492631673813,
+      "loss_iou": 1.0479531288146973,
+      "loss_num": 0.236328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 341538392,
+      "step": 1976
+    },
+    {
+      "epoch": 0.7571811566449637,
+      "grad_norm": 25.08359959058798,
+      "learning_rate": 5e-06,
+      "loss": 0.1767,
+      "num_input_tokens_seen": 341711288,
+      "step": 1977
+    },
+    {
+      "epoch": 0.7571811566449637,
+      "loss": 0.20602668821811676,
+      "loss_ce": 0.020907070487737656,
+      "loss_iou": 1.0533559322357178,
+      "loss_num": 0.185546875,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 341711288,
+      "step": 1977
+    },
+    {
+      "epoch": 0.7575641516660283,
+      "grad_norm": 32.50815703718899,
+      "learning_rate": 5e-06,
+      "loss": 0.2079,
+      "num_input_tokens_seen": 341884544,
+      "step": 1978
+    },
+    {
+      "epoch": 0.7575641516660283,
+      "loss": 0.24934254586696625,
+      "loss_ce": 0.02253590151667595,
+      "loss_iou": 1.0637845993041992,
+      "loss_num": 0.2265625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 341884544,
+      "step": 1978
+    },
+    {
+      "epoch": 0.7579471466870931,
+      "grad_norm": 43.25528112245329,
+      "learning_rate": 5e-06,
+      "loss": 0.2784,
+      "num_input_tokens_seen": 342057504,
+      "step": 1979
+    },
+    {
+      "epoch": 0.7579471466870931,
+      "loss": 0.3320387005805969,
+      "loss_ce": 0.022468380630016327,
+      "loss_iou": 1.1759567260742188,
+      "loss_num": 0.30859375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 342057504,
+      "step": 1979
+    },
+    {
+      "epoch": 0.7583301417081578,
+      "grad_norm": 24.578675392235837,
+      "learning_rate": 5e-06,
+      "loss": 0.1512,
+      "num_input_tokens_seen": 342230272,
+      "step": 1980
+    },
+    {
+      "epoch": 0.7583301417081578,
+      "loss": 0.14127089083194733,
+      "loss_ce": 0.021519910544157028,
+      "loss_iou": 1.013789415359497,
+      "loss_num": 0.11962890625,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 342230272,
+      "step": 1980
+    },
+    {
+      "epoch": 0.7587131367292225,
+      "grad_norm": 34.95370956751587,
+      "learning_rate": 5e-06,
+      "loss": 0.2554,
+      "num_input_tokens_seen": 342403064,
+      "step": 1981
+    },
+    {
+      "epoch": 0.7587131367292225,
+      "loss": 0.2870115637779236,
+      "loss_ce": 0.021996907889842987,
+      "loss_iou": 1.2670702934265137,
+      "loss_num": 0.265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 342403064,
+      "step": 1981
+    },
+    {
+      "epoch": 0.7590961317502872,
+      "grad_norm": 33.95058763744049,
+      "learning_rate": 5e-06,
+      "loss": 0.1714,
+      "num_input_tokens_seen": 342576152,
+      "step": 1982
+    },
+    {
+      "epoch": 0.7590961317502872,
+      "loss": 0.1686820089817047,
+      "loss_ce": 0.020366588607430458,
+      "loss_iou": 1.0044770240783691,
+      "loss_num": 0.1484375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 342576152,
+      "step": 1982
+    },
+    {
+      "epoch": 0.759479126771352,
+      "grad_norm": 37.854718918510926,
+      "learning_rate": 5e-06,
+      "loss": 0.1765,
+      "num_input_tokens_seen": 342748848,
+      "step": 1983
+    },
+    {
+      "epoch": 0.759479126771352,
+      "loss": 0.18910756707191467,
+      "loss_ce": 0.020467441529035568,
+      "loss_iou": 1.018295407295227,
+      "loss_num": 0.1689453125,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 342748848,
+      "step": 1983
+    },
+    {
+      "epoch": 0.7598621217924167,
+      "grad_norm": 42.20493584403363,
+      "learning_rate": 5e-06,
+      "loss": 0.1897,
+      "num_input_tokens_seen": 342921960,
+      "step": 1984
+    },
+    {
+      "epoch": 0.7598621217924167,
+      "loss": 0.17905503511428833,
+      "loss_ce": 0.02109605446457863,
+      "loss_iou": 1.0033435821533203,
+      "loss_num": 0.158203125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 342921960,
+      "step": 1984
+    },
+    {
+      "epoch": 0.7602451168134814,
+      "grad_norm": 34.01835564223233,
+      "learning_rate": 5e-06,
+      "loss": 0.2467,
+      "num_input_tokens_seen": 343094824,
+      "step": 1985
+    },
+    {
+      "epoch": 0.7602451168134814,
+      "loss": 0.27467986941337585,
+      "loss_ce": 0.021506035700440407,
+      "loss_iou": 1.085257649421692,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 343094824,
+      "step": 1985
+    },
+    {
+      "epoch": 0.7606281118345462,
+      "grad_norm": 35.93022189793799,
+      "learning_rate": 5e-06,
+      "loss": 0.2204,
+      "num_input_tokens_seen": 343268144,
+      "step": 1986
+    },
+    {
+      "epoch": 0.7606281118345462,
+      "loss": 0.26171666383743286,
+      "loss_ce": 0.022641940042376518,
+      "loss_iou": 1.0948108434677124,
+      "loss_num": 0.2392578125,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 343268144,
+      "step": 1986
+    },
+    {
+      "epoch": 0.7610111068556109,
+      "grad_norm": 28.91457837451079,
+      "learning_rate": 5e-06,
+      "loss": 0.2297,
+      "num_input_tokens_seen": 343440912,
+      "step": 1987
+    },
+    {
+      "epoch": 0.7610111068556109,
+      "loss": 0.22892248630523682,
+      "loss_ce": 0.021036745980381966,
+      "loss_iou": 1.0305060148239136,
+      "loss_num": 0.2080078125,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 343440912,
+      "step": 1987
+    },
+    {
+      "epoch": 0.7613941018766756,
+      "grad_norm": 24.05486499303163,
+      "learning_rate": 5e-06,
+      "loss": 0.1849,
+      "num_input_tokens_seen": 343613736,
+      "step": 1988
+    },
+    {
+      "epoch": 0.7613941018766756,
+      "loss": 0.22147849202156067,
+      "loss_ce": 0.02244284562766552,
+      "loss_iou": 1.0540740489959717,
+      "loss_num": 0.19921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 343613736,
+      "step": 1988
+    },
+    {
+      "epoch": 0.7617770968977403,
+      "grad_norm": 29.733899839657322,
+      "learning_rate": 5e-06,
+      "loss": 0.2641,
+      "num_input_tokens_seen": 343786608,
+      "step": 1989
+    },
+    {
+      "epoch": 0.7617770968977403,
+      "loss": 0.28331613540649414,
+      "loss_ce": 0.021353235468268394,
+      "loss_iou": 1.0573334693908691,
+      "loss_num": 0.26171875,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 343786608,
+      "step": 1989
+    },
+    {
+      "epoch": 0.7621600919188051,
+      "grad_norm": 27.944995791783562,
+      "learning_rate": 5e-06,
+      "loss": 0.17,
+      "num_input_tokens_seen": 343959224,
+      "step": 1990
+    },
+    {
+      "epoch": 0.7621600919188051,
+      "loss": 0.14480268955230713,
+      "loss_ce": 0.020901326090097427,
+      "loss_iou": 1.010115146636963,
+      "loss_num": 0.1240234375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 343959224,
+      "step": 1990
+    },
+    {
+      "epoch": 0.7625430869398698,
+      "grad_norm": 21.727588820142955,
+      "learning_rate": 5e-06,
+      "loss": 0.1742,
+      "num_input_tokens_seen": 344132128,
+      "step": 1991
+    },
+    {
+      "epoch": 0.7625430869398698,
+      "loss": 0.16735507547855377,
+      "loss_ce": 0.02123691514134407,
+      "loss_iou": 1.010033369064331,
+      "loss_num": 0.146484375,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 344132128,
+      "step": 1991
+    },
+    {
+      "epoch": 0.7629260819609345,
+      "grad_norm": 22.119752668454986,
+      "learning_rate": 5e-06,
+      "loss": 0.163,
+      "num_input_tokens_seen": 344305016,
+      "step": 1992
+    },
+    {
+      "epoch": 0.7629260819609345,
+      "loss": 0.20107056200504303,
+      "loss_ce": 0.02193237468600273,
+      "loss_iou": 1.1044776439666748,
+      "loss_num": 0.1787109375,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 344305016,
+      "step": 1992
+    },
+    {
+      "epoch": 0.7633090769819992,
+      "grad_norm": 25.744499024295337,
+      "learning_rate": 5e-06,
+      "loss": 0.2258,
+      "num_input_tokens_seen": 344478160,
+      "step": 1993
+    },
+    {
+      "epoch": 0.7633090769819992,
+      "loss": 0.23733490705490112,
+      "loss_ce": 0.02200287953019142,
+      "loss_iou": 1.0133190155029297,
+      "loss_num": 0.21484375,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 344478160,
+      "step": 1993
+    },
+    {
+      "epoch": 0.763692072003064,
+      "grad_norm": 19.137289495522907,
+      "learning_rate": 5e-06,
+      "loss": 0.2101,
+      "num_input_tokens_seen": 344651520,
+      "step": 1994
+    },
+    {
+      "epoch": 0.763692072003064,
+      "loss": 0.19120462238788605,
+      "loss_ce": 0.021893102675676346,
+      "loss_iou": 1.0239546298980713,
+      "loss_num": 0.1689453125,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 344651520,
+      "step": 1994
+    },
+    {
+      "epoch": 0.7640750670241286,
+      "grad_norm": 27.315501159437055,
+      "learning_rate": 5e-06,
+      "loss": 0.241,
+      "num_input_tokens_seen": 344824336,
+      "step": 1995
+    },
+    {
+      "epoch": 0.7640750670241286,
+      "loss": 0.2901032567024231,
+      "loss_ce": 0.021060287952423096,
+      "loss_iou": 1.0676565170288086,
+      "loss_num": 0.26953125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 344824336,
+      "step": 1995
+    },
+    {
+      "epoch": 0.7644580620451934,
+      "grad_norm": 20.198019155313965,
+      "learning_rate": 5e-06,
+      "loss": 0.2246,
+      "num_input_tokens_seen": 344997088,
+      "step": 1996
+    },
+    {
+      "epoch": 0.7644580620451934,
+      "loss": 0.2020232379436493,
+      "loss_ce": 0.020077433437108994,
+      "loss_iou": 1.0154491662979126,
+      "loss_num": 0.181640625,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 344997088,
+      "step": 1996
+    },
+    {
+      "epoch": 0.7648410570662582,
+      "grad_norm": 28.205076589215945,
+      "learning_rate": 5e-06,
+      "loss": 0.1778,
+      "num_input_tokens_seen": 345170424,
+      "step": 1997
+    },
+    {
+      "epoch": 0.7648410570662582,
+      "loss": 0.15298880636692047,
+      "loss_ce": 0.02078666165471077,
+      "loss_iou": 1.01637601852417,
+      "loss_num": 0.1318359375,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 345170424,
+      "step": 1997
+    },
+    {
+      "epoch": 0.7652240520873229,
+      "grad_norm": 30.740484651778416,
+      "learning_rate": 5e-06,
+      "loss": 0.2372,
+      "num_input_tokens_seen": 345343024,
+      "step": 1998
+    },
+    {
+      "epoch": 0.7652240520873229,
+      "loss": 0.28460055589675903,
+      "loss_ce": 0.01830417476594448,
+      "loss_iou": 1.0067775249481201,
+      "loss_num": 0.265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 345343024,
+      "step": 1998
+    },
+    {
+      "epoch": 0.7656070471083876,
+      "grad_norm": 39.224480963224835,
+      "learning_rate": 5e-06,
+      "loss": 0.233,
+      "num_input_tokens_seen": 345515992,
+      "step": 1999
+    },
+    {
+      "epoch": 0.7656070471083876,
+      "loss": 0.19093447923660278,
+      "loss_ce": 0.020585356280207634,
+      "loss_iou": 1.0244207382202148,
+      "loss_num": 0.169921875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 345515992,
+      "step": 1999
+    },
+    {
+      "epoch": 0.7659900421294523,
+      "grad_norm": 45.45632633343496,
+      "learning_rate": 5e-06,
+      "loss": 0.2306,
+      "num_input_tokens_seen": 345689096,
+      "step": 2000
+    },
+    {
+      "epoch": 0.7659900421294523,
+      "eval_websight_new_CIoU": 0.9228273034095764,
+      "eval_websight_new_GIoU": 0.9226952195167542,
+      "eval_websight_new_IoU": 0.9230187237262726,
+      "eval_websight_new_MAE_all": 0.009114941116422415,
+      "eval_websight_new_MAE_h": 0.013225205708295107,
+      "eval_websight_new_MAE_w": 0.011090578977018595,
+      "eval_websight_new_MAE_x": 0.007492676377296448,
+      "eval_websight_new_MAE_y": 0.004651304334402084,
+      "eval_websight_new_NUM_probability": 7.791836469550617e-05,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.14586272835731506,
+      "eval_websight_new_loss_ce": 0.04211839847266674,
+      "eval_websight_new_loss_iou": 1.0003713369369507,
+      "eval_websight_new_loss_num": 0.101348876953125,
+      "eval_websight_new_loss_xval": 0.101348876953125,
+      "eval_websight_new_runtime": 55.7017,
+      "eval_websight_new_samples_per_second": 0.898,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 345689096,
+      "step": 2000
+    },
+    {
+      "epoch": 0.7659900421294523,
+      "eval_seeclick_CIoU": 0.6462754011154175,
+      "eval_seeclick_GIoU": 0.6530387699604034,
+      "eval_seeclick_IoU": 0.6783847212791443,
+      "eval_seeclick_MAE_all": 0.06139025278389454,
+      "eval_seeclick_MAE_h": 0.03869203571230173,
+      "eval_seeclick_MAE_w": 0.0867950338870287,
+      "eval_seeclick_MAE_x": 0.08058405667543411,
+      "eval_seeclick_MAE_y": 0.03948989789932966,
+      "eval_seeclick_NUM_probability": 0.00010747522537712939,
+      "eval_seeclick_inside_bbox": 0.9097222089767456,
+      "eval_seeclick_loss": 0.5110470056533813,
+      "eval_seeclick_loss_ce": 0.03283202834427357,
+      "eval_seeclick_loss_iou": 1.152630627155304,
+      "eval_seeclick_loss_num": 0.4788818359375,
+      "eval_seeclick_loss_xval": 0.4788818359375,
+      "eval_seeclick_runtime": 83.792,
+      "eval_seeclick_samples_per_second": 0.597,
+      "eval_seeclick_steps_per_second": 0.024,
+      "num_input_tokens_seen": 345689096,
+      "step": 2000
+    },
+    {
+      "epoch": 0.7659900421294523,
+      "eval_icons_CIoU": 0.8534445464611053,
+      "eval_icons_GIoU": 0.848601758480072,
+      "eval_icons_IoU": 0.8605115115642548,
+      "eval_icons_MAE_all": 0.022490271367132664,
+      "eval_icons_MAE_h": 0.019543664529919624,
+      "eval_icons_MAE_w": 0.029191632755100727,
+      "eval_icons_MAE_x": 0.02440809179097414,
+      "eval_icons_MAE_y": 0.01681769685819745,
+      "eval_icons_NUM_probability": 7.248714246088639e-05,
+      "eval_icons_inside_bbox": 0.9722222089767456,
+      "eval_icons_loss": 0.1872413605451584,
+      "eval_icons_loss_ce": 0.028505710884928703,
+      "eval_icons_loss_iou": 1.0292596817016602,
+      "eval_icons_loss_num": 0.148284912109375,
+      "eval_icons_loss_xval": 0.148284912109375,
+      "eval_icons_runtime": 85.5313,
+      "eval_icons_samples_per_second": 0.585,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 345689096,
+      "step": 2000
+    },
+    {
+      "epoch": 0.7659900421294523,
+      "loss": 0.19107291102409363,
+      "loss_ce": 0.028536278754472733,
+      "loss_iou": 1.037780523300171,
+      "loss_num": 0.162109375,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 345689096,
+      "step": 2000
+    },
+    {
+      "epoch": 0.7663730371505171,
+      "grad_norm": 37.62851686661492,
+      "learning_rate": 5e-06,
+      "loss": 0.2714,
+      "num_input_tokens_seen": 345862128,
+      "step": 2001
+    },
+    {
+      "epoch": 0.7663730371505171,
+      "loss": 0.30593255162239075,
+      "loss_ce": 0.020165957510471344,
+      "loss_iou": 1.0260435342788696,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 345862128,
+      "step": 2001
+    },
+    {
+      "epoch": 0.7667560321715817,
+      "grad_norm": 30.529578676536875,
+      "learning_rate": 5e-06,
+      "loss": 0.2283,
+      "num_input_tokens_seen": 346034768,
+      "step": 2002
+    },
+    {
+      "epoch": 0.7667560321715817,
+      "loss": 0.22363267838954926,
+      "loss_ce": 0.021179085597395897,
+      "loss_iou": 1.083114504814148,
+      "loss_num": 0.2021484375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 346034768,
+      "step": 2002
+    },
+    {
+      "epoch": 0.7671390271926465,
+      "grad_norm": 28.619464310192097,
+      "learning_rate": 5e-06,
+      "loss": 0.1877,
+      "num_input_tokens_seen": 346203840,
+      "step": 2003
+    },
+    {
+      "epoch": 0.7671390271926465,
+      "loss": 0.23254099488258362,
+      "loss_ce": 0.020932119339704514,
+      "loss_iou": 1.1524338722229004,
+      "loss_num": 0.2119140625,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 346203840,
+      "step": 2003
+    },
+    {
+      "epoch": 0.7675220222137112,
+      "grad_norm": 29.2887391250122,
+      "learning_rate": 5e-06,
+      "loss": 0.1851,
+      "num_input_tokens_seen": 346377152,
+      "step": 2004
+    },
+    {
+      "epoch": 0.7675220222137112,
+      "loss": 0.1696953922510147,
+      "loss_ce": 0.02064754255115986,
+      "loss_iou": 1.009027123451233,
+      "loss_num": 0.1494140625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 346377152,
+      "step": 2004
+    },
+    {
+      "epoch": 0.767905017234776,
+      "grad_norm": 27.633679375149583,
+      "learning_rate": 5e-06,
+      "loss": 0.2442,
+      "num_input_tokens_seen": 346550168,
+      "step": 2005
+    },
+    {
+      "epoch": 0.767905017234776,
+      "loss": 0.22287316620349884,
+      "loss_ce": 0.019931267946958542,
+      "loss_iou": 1.0643115043640137,
+      "loss_num": 0.203125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 346550168,
+      "step": 2005
+    },
+    {
+      "epoch": 0.7682880122558406,
+      "grad_norm": 36.47280484065821,
+      "learning_rate": 5e-06,
+      "loss": 0.2381,
+      "num_input_tokens_seen": 346722848,
+      "step": 2006
+    },
+    {
+      "epoch": 0.7682880122558406,
+      "loss": 0.24743898212909698,
+      "loss_ce": 0.021273205056786537,
+      "loss_iou": 1.0317459106445312,
+      "loss_num": 0.2265625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 346722848,
+      "step": 2006
+    },
+    {
+      "epoch": 0.7686710072769054,
+      "grad_norm": 36.33389931984183,
+      "learning_rate": 5e-06,
+      "loss": 0.2083,
+      "num_input_tokens_seen": 346895904,
+      "step": 2007
+    },
+    {
+      "epoch": 0.7686710072769054,
+      "loss": 0.14966237545013428,
+      "loss_ce": 0.02185475081205368,
+      "loss_iou": 0.8812423348426819,
+      "loss_num": 0.1279296875,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 346895904,
+      "step": 2007
+    },
+    {
+      "epoch": 0.7690540022979702,
+      "grad_norm": 28.770279519362564,
+      "learning_rate": 5e-06,
+      "loss": 0.2267,
+      "num_input_tokens_seen": 347068800,
+      "step": 2008
+    },
+    {
+      "epoch": 0.7690540022979702,
+      "loss": 0.16842623054981232,
+      "loss_ce": 0.020751673728227615,
+      "loss_iou": 1.016047716140747,
+      "loss_num": 0.1474609375,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 347068800,
+      "step": 2008
+    },
+    {
+      "epoch": 0.7694369973190348,
+      "grad_norm": 25.881096443290257,
+      "learning_rate": 5e-06,
+      "loss": 0.1906,
+      "num_input_tokens_seen": 347241552,
+      "step": 2009
+    },
+    {
+      "epoch": 0.7694369973190348,
+      "loss": 0.22845833003520966,
+      "loss_ce": 0.021427083760499954,
+      "loss_iou": 1.0062979459762573,
+      "loss_num": 0.20703125,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 347241552,
+      "step": 2009
+    },
+    {
+      "epoch": 0.7698199923400996,
+      "grad_norm": 33.24424791559408,
+      "learning_rate": 5e-06,
+      "loss": 0.1866,
+      "num_input_tokens_seen": 347413224,
+      "step": 2010
+    },
+    {
+      "epoch": 0.7698199923400996,
+      "loss": 0.1563665270805359,
+      "loss_ce": 0.02102106809616089,
+      "loss_iou": 1.0179115533828735,
+      "loss_num": 0.1357421875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 347413224,
+      "step": 2010
+    },
+    {
+      "epoch": 0.7702029873611643,
+      "grad_norm": 38.63471375316766,
+      "learning_rate": 5e-06,
+      "loss": 0.2204,
+      "num_input_tokens_seen": 347586024,
+      "step": 2011
+    },
+    {
+      "epoch": 0.7702029873611643,
+      "loss": 0.2395392805337906,
+      "loss_ce": 0.021460656076669693,
+      "loss_iou": 1.1907134056091309,
+      "loss_num": 0.2177734375,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 347586024,
+      "step": 2011
+    },
+    {
+      "epoch": 0.7705859823822291,
+      "grad_norm": 25.37617773051809,
+      "learning_rate": 5e-06,
+      "loss": 0.1709,
+      "num_input_tokens_seen": 347758936,
+      "step": 2012
+    },
+    {
+      "epoch": 0.7705859823822291,
+      "loss": 0.17508289217948914,
+      "loss_ce": 0.02310534566640854,
+      "loss_iou": 1.030497670173645,
+      "loss_num": 0.15234375,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 347758936,
+      "step": 2012
+    },
+    {
+      "epoch": 0.7709689774032937,
+      "grad_norm": 29.09347029281265,
+      "learning_rate": 5e-06,
+      "loss": 0.1937,
+      "num_input_tokens_seen": 347931872,
+      "step": 2013
+    },
+    {
+      "epoch": 0.7709689774032937,
+      "loss": 0.19590593874454498,
+      "loss_ce": 0.020490895956754684,
+      "loss_iou": 0.8654123544692993,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 347931872,
+      "step": 2013
+    },
+    {
+      "epoch": 0.7713519724243585,
+      "grad_norm": 27.7951852319954,
+      "learning_rate": 5e-06,
+      "loss": 0.2602,
+      "num_input_tokens_seen": 348104600,
+      "step": 2014
+    },
+    {
+      "epoch": 0.7713519724243585,
+      "loss": 0.34026581048965454,
+      "loss_ce": 0.019648149609565735,
+      "loss_iou": 1.2110865116119385,
+      "loss_num": 0.3203125,
+      "loss_xval": 0.3203125,
+      "num_input_tokens_seen": 348104600,
+      "step": 2014
+    },
+    {
+      "epoch": 0.7717349674454232,
+      "grad_norm": 43.72003435417935,
+      "learning_rate": 5e-06,
+      "loss": 0.1909,
+      "num_input_tokens_seen": 348277448,
+      "step": 2015
+    },
+    {
+      "epoch": 0.7717349674454232,
+      "loss": 0.1730906367301941,
+      "loss_ce": 0.02184552513062954,
+      "loss_iou": 1.026153802871704,
+      "loss_num": 0.1513671875,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 348277448,
+      "step": 2015
+    },
+    {
+      "epoch": 0.7721179624664879,
+      "grad_norm": 32.331311683787206,
+      "learning_rate": 5e-06,
+      "loss": 0.1919,
+      "num_input_tokens_seen": 348450496,
+      "step": 2016
+    },
+    {
+      "epoch": 0.7721179624664879,
+      "loss": 0.19085338711738586,
+      "loss_ce": 0.02056529000401497,
+      "loss_iou": 1.0118297338485718,
+      "loss_num": 0.169921875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 348450496,
+      "step": 2016
+    },
+    {
+      "epoch": 0.7725009574875527,
+      "grad_norm": 44.85670015680483,
+      "learning_rate": 5e-06,
+      "loss": 0.2239,
+      "num_input_tokens_seen": 348623496,
+      "step": 2017
+    },
+    {
+      "epoch": 0.7725009574875527,
+      "loss": 0.22206342220306396,
+      "loss_ce": 0.021196719259023666,
+      "loss_iou": 1.0615859031677246,
+      "loss_num": 0.201171875,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 348623496,
+      "step": 2017
+    },
+    {
+      "epoch": 0.7728839525086174,
+      "grad_norm": 31.448395185881434,
+      "learning_rate": 5e-06,
+      "loss": 0.2297,
+      "num_input_tokens_seen": 348796480,
+      "step": 2018
+    },
+    {
+      "epoch": 0.7728839525086174,
+      "loss": 0.2339247763156891,
+      "loss_ce": 0.01975243166089058,
+      "loss_iou": 1.0011065006256104,
+      "loss_num": 0.2138671875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 348796480,
+      "step": 2018
+    },
+    {
+      "epoch": 0.7732669475296822,
+      "grad_norm": 58.93911242383194,
+      "learning_rate": 5e-06,
+      "loss": 0.2494,
+      "num_input_tokens_seen": 348968856,
+      "step": 2019
+    },
+    {
+      "epoch": 0.7732669475296822,
+      "loss": 0.24253229796886444,
+      "loss_ce": 0.019876059144735336,
+      "loss_iou": 1.0425891876220703,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 348968856,
+      "step": 2019
+    },
+    {
+      "epoch": 0.7736499425507468,
+      "grad_norm": 31.521528661475983,
+      "learning_rate": 5e-06,
+      "loss": 0.2652,
+      "num_input_tokens_seen": 349141984,
+      "step": 2020
+    },
+    {
+      "epoch": 0.7736499425507468,
+      "loss": 0.22655808925628662,
+      "loss_ce": 0.02221238613128662,
+      "loss_iou": 1.077653408050537,
+      "loss_num": 0.2041015625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 349141984,
+      "step": 2020
+    },
+    {
+      "epoch": 0.7740329375718116,
+      "grad_norm": 51.802398179186724,
+      "learning_rate": 5e-06,
+      "loss": 0.2872,
+      "num_input_tokens_seen": 349314688,
+      "step": 2021
+    },
+    {
+      "epoch": 0.7740329375718116,
+      "loss": 0.24481329321861267,
+      "loss_ce": 0.020692195743322372,
+      "loss_iou": 1.0158942937850952,
+      "loss_num": 0.224609375,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 349314688,
+      "step": 2021
+    },
+    {
+      "epoch": 0.7744159325928763,
+      "grad_norm": 30.043300193427182,
+      "learning_rate": 5e-06,
+      "loss": 0.2316,
+      "num_input_tokens_seen": 349487840,
+      "step": 2022
+    },
+    {
+      "epoch": 0.7744159325928763,
+      "loss": 0.28100672364234924,
+      "loss_ce": 0.020325560122728348,
+      "loss_iou": 1.015325665473938,
+      "loss_num": 0.259765625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 349487840,
+      "step": 2022
+    },
+    {
+      "epoch": 0.774798927613941,
+      "grad_norm": 58.21751211807337,
+      "learning_rate": 5e-06,
+      "loss": 0.2654,
+      "num_input_tokens_seen": 349660672,
+      "step": 2023
+    },
+    {
+      "epoch": 0.774798927613941,
+      "loss": 0.2263191044330597,
+      "loss_ce": 0.021729253232479095,
+      "loss_iou": 1.007361888885498,
+      "loss_num": 0.205078125,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 349660672,
+      "step": 2023
+    },
+    {
+      "epoch": 0.7751819226350057,
+      "grad_norm": 26.886955651983698,
+      "learning_rate": 5e-06,
+      "loss": 0.2666,
+      "num_input_tokens_seen": 349833424,
+      "step": 2024
+    },
+    {
+      "epoch": 0.7751819226350057,
+      "loss": 0.250285267829895,
+      "loss_ce": 0.021525517106056213,
+      "loss_iou": 1.1009631156921387,
+      "loss_num": 0.228515625,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 349833424,
+      "step": 2024
+    },
+    {
+      "epoch": 0.7755649176560705,
+      "grad_norm": 22.93500372346212,
+      "learning_rate": 5e-06,
+      "loss": 0.2116,
+      "num_input_tokens_seen": 350006432,
+      "step": 2025
+    },
+    {
+      "epoch": 0.7755649176560705,
+      "loss": 0.2316955029964447,
+      "loss_ce": 0.020147655159235,
+      "loss_iou": 1.4622738361358643,
+      "loss_num": 0.2119140625,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 350006432,
+      "step": 2025
+    },
+    {
+      "epoch": 0.7759479126771351,
+      "grad_norm": 15.292739097693845,
+      "learning_rate": 5e-06,
+      "loss": 0.1678,
+      "num_input_tokens_seen": 350179648,
+      "step": 2026
+    },
+    {
+      "epoch": 0.7759479126771351,
+      "loss": 0.15203899145126343,
+      "loss_ce": 0.021851008757948875,
+      "loss_iou": 0.9951585531234741,
+      "loss_num": 0.1298828125,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 350179648,
+      "step": 2026
+    },
+    {
+      "epoch": 0.7763309076981999,
+      "grad_norm": 15.43524855294656,
+      "learning_rate": 5e-06,
+      "loss": 0.1686,
+      "num_input_tokens_seen": 350352496,
+      "step": 2027
+    },
+    {
+      "epoch": 0.7763309076981999,
+      "loss": 0.17393441498279572,
+      "loss_ce": 0.020614106208086014,
+      "loss_iou": 1.019606590270996,
+      "loss_num": 0.1533203125,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 350352496,
+      "step": 2027
+    },
+    {
+      "epoch": 0.7767139027192647,
+      "grad_norm": 24.727026774937,
+      "learning_rate": 5e-06,
+      "loss": 0.2016,
+      "num_input_tokens_seen": 350525272,
+      "step": 2028
+    },
+    {
+      "epoch": 0.7767139027192647,
+      "loss": 0.16336774826049805,
+      "loss_ce": 0.021461009979248047,
+      "loss_iou": 1.018656611442566,
+      "loss_num": 0.1416015625,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 350525272,
+      "step": 2028
+    },
+    {
+      "epoch": 0.7770968977403294,
+      "grad_norm": 28.55673031986966,
+      "learning_rate": 5e-06,
+      "loss": 0.1844,
+      "num_input_tokens_seen": 350697960,
+      "step": 2029
+    },
+    {
+      "epoch": 0.7770968977403294,
+      "loss": 0.18055246770381927,
+      "loss_ce": 0.020335180684924126,
+      "loss_iou": 1.0059808492660522,
+      "loss_num": 0.16015625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 350697960,
+      "step": 2029
+    },
+    {
+      "epoch": 0.7774798927613941,
+      "grad_norm": 29.256057572891525,
+      "learning_rate": 5e-06,
+      "loss": 0.1845,
+      "num_input_tokens_seen": 350870944,
+      "step": 2030
+    },
+    {
+      "epoch": 0.7774798927613941,
+      "loss": 0.21915271878242493,
+      "loss_ce": 0.018896378576755524,
+      "loss_iou": 1.0013165473937988,
+      "loss_num": 0.2001953125,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 350870944,
+      "step": 2030
+    },
+    {
+      "epoch": 0.7778628877824588,
+      "grad_norm": 31.918065765061225,
+      "learning_rate": 5e-06,
+      "loss": 0.1945,
+      "num_input_tokens_seen": 351040520,
+      "step": 2031
+    },
+    {
+      "epoch": 0.7778628877824588,
+      "loss": 0.20982369780540466,
+      "loss_ce": 0.023117145523428917,
+      "loss_iou": 1.1483569145202637,
+      "loss_num": 0.1865234375,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 351040520,
+      "step": 2031
+    },
+    {
+      "epoch": 0.7782458828035236,
+      "grad_norm": 24.643181892068174,
+      "learning_rate": 5e-06,
+      "loss": 0.1785,
+      "num_input_tokens_seen": 351213032,
+      "step": 2032
+    },
+    {
+      "epoch": 0.7782458828035236,
+      "loss": 0.14677008986473083,
+      "loss_ce": 0.020976630970835686,
+      "loss_iou": 1.0016255378723145,
+      "loss_num": 0.1259765625,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 351213032,
+      "step": 2032
+    },
+    {
+      "epoch": 0.7786288778245882,
+      "grad_norm": 22.723302450830133,
+      "learning_rate": 5e-06,
+      "loss": 0.2351,
+      "num_input_tokens_seen": 351385744,
+      "step": 2033
+    },
+    {
+      "epoch": 0.7786288778245882,
+      "loss": 0.2747228741645813,
+      "loss_ce": 0.022098341956734657,
+      "loss_iou": 1.1827709674835205,
+      "loss_num": 0.251953125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 351385744,
+      "step": 2033
+    },
+    {
+      "epoch": 0.779011872845653,
+      "grad_norm": 28.561332086628987,
+      "learning_rate": 5e-06,
+      "loss": 0.2361,
+      "num_input_tokens_seen": 351558224,
+      "step": 2034
+    },
+    {
+      "epoch": 0.779011872845653,
+      "loss": 0.2604377567768097,
+      "loss_ce": 0.019959241151809692,
+      "loss_iou": 1.0186917781829834,
+      "loss_num": 0.240234375,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 351558224,
+      "step": 2034
+    },
+    {
+      "epoch": 0.7793948678667177,
+      "grad_norm": 32.677920741788206,
+      "learning_rate": 5e-06,
+      "loss": 0.1881,
+      "num_input_tokens_seen": 351731232,
+      "step": 2035
+    },
+    {
+      "epoch": 0.7793948678667177,
+      "loss": 0.1979053020477295,
+      "loss_ce": 0.021879900246858597,
+      "loss_iou": 1.046138048171997,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 351731232,
+      "step": 2035
+    },
+    {
+      "epoch": 0.7797778628877825,
+      "grad_norm": 30.640751263353486,
+      "learning_rate": 5e-06,
+      "loss": 0.2185,
+      "num_input_tokens_seen": 351904312,
+      "step": 2036
+    },
+    {
+      "epoch": 0.7797778628877825,
+      "loss": 0.1802019327878952,
+      "loss_ce": 0.0223650224506855,
+      "loss_iou": 1.0634071826934814,
+      "loss_num": 0.158203125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 351904312,
+      "step": 2036
+    },
+    {
+      "epoch": 0.7801608579088471,
+      "grad_norm": 27.624165621482252,
+      "learning_rate": 5e-06,
+      "loss": 0.2514,
+      "num_input_tokens_seen": 352077536,
+      "step": 2037
+    },
+    {
+      "epoch": 0.7801608579088471,
+      "loss": 0.28442516922950745,
+      "loss_ce": 0.0214246716350317,
+      "loss_iou": 1.0522079467773438,
+      "loss_num": 0.263671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 352077536,
+      "step": 2037
+    },
+    {
+      "epoch": 0.7805438529299119,
+      "grad_norm": 26.965646504091804,
+      "learning_rate": 5e-06,
+      "loss": 0.2443,
+      "num_input_tokens_seen": 352250400,
+      "step": 2038
+    },
+    {
+      "epoch": 0.7805438529299119,
+      "loss": 0.26357731223106384,
+      "loss_ce": 0.020199626684188843,
+      "loss_iou": 1.0834300518035889,
+      "loss_num": 0.2431640625,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 352250400,
+      "step": 2038
+    },
+    {
+      "epoch": 0.7809268479509767,
+      "grad_norm": 32.410844358147976,
+      "learning_rate": 5e-06,
+      "loss": 0.2331,
+      "num_input_tokens_seen": 352423376,
+      "step": 2039
+    },
+    {
+      "epoch": 0.7809268479509767,
+      "loss": 0.2548708915710449,
+      "loss_ce": 0.022326957434415817,
+      "loss_iou": 1.0806260108947754,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 352423376,
+      "step": 2039
+    },
+    {
+      "epoch": 0.7813098429720413,
+      "grad_norm": 25.32244849030144,
+      "learning_rate": 5e-06,
+      "loss": 0.177,
+      "num_input_tokens_seen": 352596808,
+      "step": 2040
+    },
+    {
+      "epoch": 0.7813098429720413,
+      "loss": 0.1374257504940033,
+      "loss_ce": 0.019536349922418594,
+      "loss_iou": 1.0070759057998657,
+      "loss_num": 0.11767578125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 352596808,
+      "step": 2040
+    },
+    {
+      "epoch": 0.7816928379931061,
+      "grad_norm": 26.46573959918601,
+      "learning_rate": 5e-06,
+      "loss": 0.2183,
+      "num_input_tokens_seen": 352770024,
+      "step": 2041
+    },
+    {
+      "epoch": 0.7816928379931061,
+      "loss": 0.22981318831443787,
+      "loss_ce": 0.02113397791981697,
+      "loss_iou": 1.1103980541229248,
+      "loss_num": 0.208984375,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 352770024,
+      "step": 2041
+    },
+    {
+      "epoch": 0.7820758330141708,
+      "grad_norm": 32.560681429372664,
+      "learning_rate": 5e-06,
+      "loss": 0.2158,
+      "num_input_tokens_seen": 352943200,
+      "step": 2042
+    },
+    {
+      "epoch": 0.7820758330141708,
+      "loss": 0.18732693791389465,
+      "loss_ce": 0.022104769945144653,
+      "loss_iou": 1.007914423942566,
+      "loss_num": 0.1650390625,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 352943200,
+      "step": 2042
+    },
+    {
+      "epoch": 0.7824588280352356,
+      "grad_norm": 43.25860910942051,
+      "learning_rate": 5e-06,
+      "loss": 0.2537,
+      "num_input_tokens_seen": 353116408,
+      "step": 2043
+    },
+    {
+      "epoch": 0.7824588280352356,
+      "loss": 0.22283262014389038,
+      "loss_ce": 0.02141660824418068,
+      "loss_iou": 1.0568044185638428,
+      "loss_num": 0.201171875,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 353116408,
+      "step": 2043
+    },
+    {
+      "epoch": 0.7828418230563002,
+      "grad_norm": 36.265030863036884,
+      "learning_rate": 5e-06,
+      "loss": 0.1648,
+      "num_input_tokens_seen": 353289184,
+      "step": 2044
+    },
+    {
+      "epoch": 0.7828418230563002,
+      "loss": 0.17458541691303253,
+      "loss_ce": 0.020532675087451935,
+      "loss_iou": 1.0050618648529053,
+      "loss_num": 0.154296875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 353289184,
+      "step": 2044
+    },
+    {
+      "epoch": 0.783224818077365,
+      "grad_norm": 25.566315369884663,
+      "learning_rate": 5e-06,
+      "loss": 0.2078,
+      "num_input_tokens_seen": 353462240,
+      "step": 2045
+    },
+    {
+      "epoch": 0.783224818077365,
+      "loss": 0.1959572583436966,
+      "loss_ce": 0.021274644881486893,
+      "loss_iou": 1.0417563915252686,
+      "loss_num": 0.1748046875,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 353462240,
+      "step": 2045
+    },
+    {
+      "epoch": 0.7836078130984298,
+      "grad_norm": 24.19159450809671,
+      "learning_rate": 5e-06,
+      "loss": 0.1916,
+      "num_input_tokens_seen": 353634736,
+      "step": 2046
+    },
+    {
+      "epoch": 0.7836078130984298,
+      "loss": 0.1993541270494461,
+      "loss_ce": 0.019727662205696106,
+      "loss_iou": 1.0452462434768677,
+      "loss_num": 0.1796875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 353634736,
+      "step": 2046
+    },
+    {
+      "epoch": 0.7839908081194944,
+      "grad_norm": 35.112046726772306,
+      "learning_rate": 5e-06,
+      "loss": 0.2197,
+      "num_input_tokens_seen": 353807432,
+      "step": 2047
+    },
+    {
+      "epoch": 0.7839908081194944,
+      "loss": 0.2643466293811798,
+      "loss_ce": 0.02258637547492981,
+      "loss_iou": 1.0596150159835815,
+      "loss_num": 0.2421875,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 353807432,
+      "step": 2047
+    },
+    {
+      "epoch": 0.7843738031405592,
+      "grad_norm": 38.27092369525028,
+      "learning_rate": 5e-06,
+      "loss": 0.1947,
+      "num_input_tokens_seen": 353980744,
+      "step": 2048
+    },
+    {
+      "epoch": 0.7843738031405592,
+      "loss": 0.19919967651367188,
+      "loss_ce": 0.022136688232421875,
+      "loss_iou": 1.01801598072052,
+      "loss_num": 0.1767578125,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 353980744,
+      "step": 2048
+    },
+    {
+      "epoch": 0.7847567981616239,
+      "grad_norm": 35.303257622853316,
+      "learning_rate": 5e-06,
+      "loss": 0.2135,
+      "num_input_tokens_seen": 354153920,
+      "step": 2049
+    },
+    {
+      "epoch": 0.7847567981616239,
+      "loss": 0.19922125339508057,
+      "loss_ce": 0.019594788551330566,
+      "loss_iou": 1.0008752346038818,
+      "loss_num": 0.1796875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 354153920,
+      "step": 2049
+    },
+    {
+      "epoch": 0.7851397931826887,
+      "grad_norm": 26.223173743701402,
+      "learning_rate": 5e-06,
+      "loss": 0.2507,
+      "num_input_tokens_seen": 354326824,
+      "step": 2050
+    },
+    {
+      "epoch": 0.7851397931826887,
+      "loss": 0.2493046373128891,
+      "loss_ce": 0.019080042839050293,
+      "loss_iou": 1.091526985168457,
+      "loss_num": 0.23046875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 354326824,
+      "step": 2050
+    },
+    {
+      "epoch": 0.7855227882037533,
+      "grad_norm": 26.155226545194463,
+      "learning_rate": 5e-06,
+      "loss": 0.1696,
+      "num_input_tokens_seen": 354499760,
+      "step": 2051
+    },
+    {
+      "epoch": 0.7855227882037533,
+      "loss": 0.13985827565193176,
+      "loss_ce": 0.021328000351786613,
+      "loss_iou": 1.001230239868164,
+      "loss_num": 0.11865234375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 354499760,
+      "step": 2051
+    },
+    {
+      "epoch": 0.7859057832248181,
+      "grad_norm": 33.22083130224732,
+      "learning_rate": 5e-06,
+      "loss": 0.1797,
+      "num_input_tokens_seen": 354672576,
+      "step": 2052
+    },
+    {
+      "epoch": 0.7859057832248181,
+      "loss": 0.16594764590263367,
+      "loss_ce": 0.020195700228214264,
+      "loss_iou": 1.0120246410369873,
+      "loss_num": 0.1455078125,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 354672576,
+      "step": 2052
+    },
+    {
+      "epoch": 0.7862887782458828,
+      "grad_norm": 31.91281088969455,
+      "learning_rate": 5e-06,
+      "loss": 0.2004,
+      "num_input_tokens_seen": 354845680,
+      "step": 2053
+    },
+    {
+      "epoch": 0.7862887782458828,
+      "loss": 0.13769575953483582,
+      "loss_ce": 0.020203087478876114,
+      "loss_iou": 1.0007909536361694,
+      "loss_num": 0.11767578125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 354845680,
+      "step": 2053
+    },
+    {
+      "epoch": 0.7866717732669475,
+      "grad_norm": 24.824588348523132,
+      "learning_rate": 5e-06,
+      "loss": 0.2602,
+      "num_input_tokens_seen": 355018536,
+      "step": 2054
+    },
+    {
+      "epoch": 0.7866717732669475,
+      "loss": 0.20731166005134583,
+      "loss_ce": 0.019506465643644333,
+      "loss_iou": 1.0803972482681274,
+      "loss_num": 0.1875,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 355018536,
+      "step": 2054
+    },
+    {
+      "epoch": 0.7870547682880122,
+      "grad_norm": 49.85580010214409,
+      "learning_rate": 5e-06,
+      "loss": 0.2189,
+      "num_input_tokens_seen": 355191824,
+      "step": 2055
+    },
+    {
+      "epoch": 0.7870547682880122,
+      "loss": 0.18205732107162476,
+      "loss_ce": 0.021840039640665054,
+      "loss_iou": 1.0319499969482422,
+      "loss_num": 0.16015625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 355191824,
+      "step": 2055
+    },
+    {
+      "epoch": 0.787437763309077,
+      "grad_norm": 35.185942298183626,
+      "learning_rate": 5e-06,
+      "loss": 0.26,
+      "num_input_tokens_seen": 355365016,
+      "step": 2056
+    },
+    {
+      "epoch": 0.787437763309077,
+      "loss": 0.31597667932510376,
+      "loss_ce": 0.022031398490071297,
+      "loss_iou": 1.257796287536621,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 355365016,
+      "step": 2056
+    },
+    {
+      "epoch": 0.7878207583301418,
+      "grad_norm": 34.48133769594558,
+      "learning_rate": 5e-06,
+      "loss": 0.2261,
+      "num_input_tokens_seen": 355538216,
+      "step": 2057
+    },
+    {
+      "epoch": 0.7878207583301418,
+      "loss": 0.2338372766971588,
+      "loss_ce": 0.022228389978408813,
+      "loss_iou": 1.0854668617248535,
+      "loss_num": 0.2119140625,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 355538216,
+      "step": 2057
+    },
+    {
+      "epoch": 0.7882037533512064,
+      "grad_norm": 27.905404587663607,
+      "learning_rate": 5e-06,
+      "loss": 0.2673,
+      "num_input_tokens_seen": 355711048,
+      "step": 2058
+    },
+    {
+      "epoch": 0.7882037533512064,
+      "loss": 0.26376307010650635,
+      "loss_ce": 0.019805539399385452,
+      "loss_iou": 1.094059944152832,
+      "loss_num": 0.244140625,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 355711048,
+      "step": 2058
+    },
+    {
+      "epoch": 0.7885867483722712,
+      "grad_norm": 37.08467618361751,
+      "learning_rate": 5e-06,
+      "loss": 0.2489,
+      "num_input_tokens_seen": 355884240,
+      "step": 2059
+    },
+    {
+      "epoch": 0.7885867483722712,
+      "loss": 0.2613140940666199,
+      "loss_ce": 0.02144593559205532,
+      "loss_iou": 1.0572009086608887,
+      "loss_num": 0.240234375,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 355884240,
+      "step": 2059
+    },
+    {
+      "epoch": 0.7889697433933359,
+      "grad_norm": 27.2685601043996,
+      "learning_rate": 5e-06,
+      "loss": 0.2062,
+      "num_input_tokens_seen": 356057592,
+      "step": 2060
+    },
+    {
+      "epoch": 0.7889697433933359,
+      "loss": 0.17139238119125366,
+      "loss_ce": 0.021245889365673065,
+      "loss_iou": 1.0331575870513916,
+      "loss_num": 0.150390625,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 356057592,
+      "step": 2060
+    },
+    {
+      "epoch": 0.7893527384144006,
+      "grad_norm": 40.15787509001266,
+      "learning_rate": 5e-06,
+      "loss": 0.2283,
+      "num_input_tokens_seen": 356230424,
+      "step": 2061
+    },
+    {
+      "epoch": 0.7893527384144006,
+      "loss": 0.2529626190662384,
+      "loss_ce": 0.020784892141819,
+      "loss_iou": 0.8782117366790771,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 356230424,
+      "step": 2061
+    },
+    {
+      "epoch": 0.7897357334354653,
+      "grad_norm": 27.111634342880546,
+      "learning_rate": 5e-06,
+      "loss": 0.1672,
+      "num_input_tokens_seen": 356403264,
+      "step": 2062
+    },
+    {
+      "epoch": 0.7897357334354653,
+      "loss": 0.18083465099334717,
+      "loss_ce": 0.021838072687387466,
+      "loss_iou": 1.0160695314407349,
+      "loss_num": 0.1591796875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 356403264,
+      "step": 2062
+    },
+    {
+      "epoch": 0.7901187284565301,
+      "grad_norm": 44.99530458943178,
+      "learning_rate": 5e-06,
+      "loss": 0.2912,
+      "num_input_tokens_seen": 356576056,
+      "step": 2063
+    },
+    {
+      "epoch": 0.7901187284565301,
+      "loss": 0.2838969826698303,
+      "loss_ce": 0.021445803344249725,
+      "loss_iou": 1.0098295211791992,
+      "loss_num": 0.26171875,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 356576056,
+      "step": 2063
+    },
+    {
+      "epoch": 0.7905017234775948,
+      "grad_norm": 18.00601333246881,
+      "learning_rate": 5e-06,
+      "loss": 0.2703,
+      "num_input_tokens_seen": 356749432,
+      "step": 2064
+    },
+    {
+      "epoch": 0.7905017234775948,
+      "loss": 0.22683188319206238,
+      "loss_ce": 0.021570663899183273,
+      "loss_iou": 1.008726716041565,
+      "loss_num": 0.205078125,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 356749432,
+      "step": 2064
+    },
+    {
+      "epoch": 0.7908847184986595,
+      "grad_norm": 14.109572408707434,
+      "learning_rate": 5e-06,
+      "loss": 0.2188,
+      "num_input_tokens_seen": 356922256,
+      "step": 2065
+    },
+    {
+      "epoch": 0.7908847184986595,
+      "loss": 0.20682987570762634,
+      "loss_ce": 0.02030644193291664,
+      "loss_iou": 1.1127333641052246,
+      "loss_num": 0.1865234375,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 356922256,
+      "step": 2065
+    },
+    {
+      "epoch": 0.7912677135197242,
+      "grad_norm": 14.511777793218755,
+      "learning_rate": 5e-06,
+      "loss": 0.1767,
+      "num_input_tokens_seen": 357095208,
+      "step": 2066
+    },
+    {
+      "epoch": 0.7912677135197242,
+      "loss": 0.15498453378677368,
+      "loss_ce": 0.02180583029985428,
+      "loss_iou": 1.0268007516860962,
+      "loss_num": 0.1328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 357095208,
+      "step": 2066
+    },
+    {
+      "epoch": 0.791650708540789,
+      "grad_norm": 34.14061707017323,
+      "learning_rate": 5e-06,
+      "loss": 0.2862,
+      "num_input_tokens_seen": 357268344,
+      "step": 2067
+    },
+    {
+      "epoch": 0.791650708540789,
+      "loss": 0.31500089168548584,
+      "loss_ce": 0.0212386604398489,
+      "loss_iou": 1.1864237785339355,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 357268344,
+      "step": 2067
+    },
+    {
+      "epoch": 0.7920337035618537,
+      "grad_norm": 47.77593428494504,
+      "learning_rate": 5e-06,
+      "loss": 0.2179,
+      "num_input_tokens_seen": 357441400,
+      "step": 2068
+    },
+    {
+      "epoch": 0.7920337035618537,
+      "loss": 0.17708002030849457,
+      "loss_ce": 0.02137933485209942,
+      "loss_iou": 1.0025368928909302,
+      "loss_num": 0.1552734375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 357441400,
+      "step": 2068
+    },
+    {
+      "epoch": 0.7924166985829184,
+      "grad_norm": 29.402961507585474,
+      "learning_rate": 5e-06,
+      "loss": 0.2119,
+      "num_input_tokens_seen": 357614232,
+      "step": 2069
+    },
+    {
+      "epoch": 0.7924166985829184,
+      "loss": 0.20056194067001343,
+      "loss_ce": 0.019470641389489174,
+      "loss_iou": 1.0144984722137451,
+      "loss_num": 0.1806640625,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 357614232,
+      "step": 2069
+    },
+    {
+      "epoch": 0.7927996936039832,
+      "grad_norm": 23.703830648499626,
+      "learning_rate": 5e-06,
+      "loss": 0.2828,
+      "num_input_tokens_seen": 357787064,
+      "step": 2070
+    },
+    {
+      "epoch": 0.7927996936039832,
+      "loss": 0.2905902862548828,
+      "loss_ce": 0.02014351263642311,
+      "loss_iou": 1.168410301208496,
+      "loss_num": 0.26953125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 357787064,
+      "step": 2070
+    },
+    {
+      "epoch": 0.7931826886250479,
+      "grad_norm": 25.10557124225831,
+      "learning_rate": 5e-06,
+      "loss": 0.19,
+      "num_input_tokens_seen": 357960160,
+      "step": 2071
+    },
+    {
+      "epoch": 0.7931826886250479,
+      "loss": 0.13955967128276825,
+      "loss_ce": 0.021517682820558548,
+      "loss_iou": 1.0058672428131104,
+      "loss_num": 0.1181640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 357960160,
+      "step": 2071
+    },
+    {
+      "epoch": 0.7935656836461126,
+      "grad_norm": 26.610721080718825,
+      "learning_rate": 5e-06,
+      "loss": 0.2413,
+      "num_input_tokens_seen": 358133040,
+      "step": 2072
+    },
+    {
+      "epoch": 0.7935656836461126,
+      "loss": 0.19871611893177032,
+      "loss_ce": 0.02147001028060913,
+      "loss_iou": 0.9230528473854065,
+      "loss_num": 0.177734375,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 358133040,
+      "step": 2072
+    },
+    {
+      "epoch": 0.7939486786671773,
+      "grad_norm": 30.32934432344583,
+      "learning_rate": 5e-06,
+      "loss": 0.2017,
+      "num_input_tokens_seen": 358305792,
+      "step": 2073
+    },
+    {
+      "epoch": 0.7939486786671773,
+      "loss": 0.23280559480190277,
+      "loss_ce": 0.019670821726322174,
+      "loss_iou": 1.2293424606323242,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 358305792,
+      "step": 2073
+    },
+    {
+      "epoch": 0.7943316736882421,
+      "grad_norm": 21.668661242135155,
+      "learning_rate": 5e-06,
+      "loss": 0.2759,
+      "num_input_tokens_seen": 358478600,
+      "step": 2074
+    },
+    {
+      "epoch": 0.7943316736882421,
+      "loss": 0.26689743995666504,
+      "loss_ce": 0.02068161964416504,
+      "loss_iou": 1.0439400672912598,
+      "loss_num": 0.24609375,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 358478600,
+      "step": 2074
+    },
+    {
+      "epoch": 0.7947146687093067,
+      "grad_norm": 29.057915372282242,
+      "learning_rate": 5e-06,
+      "loss": 0.2486,
+      "num_input_tokens_seen": 358651464,
+      "step": 2075
+    },
+    {
+      "epoch": 0.7947146687093067,
+      "loss": 0.23087254166603088,
+      "loss_ce": 0.023658178746700287,
+      "loss_iou": 1.0160874128341675,
+      "loss_num": 0.20703125,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 358651464,
+      "step": 2075
+    },
+    {
+      "epoch": 0.7950976637303715,
+      "grad_norm": 30.89577680325765,
+      "learning_rate": 5e-06,
+      "loss": 0.2869,
+      "num_input_tokens_seen": 358824696,
+      "step": 2076
+    },
+    {
+      "epoch": 0.7950976637303715,
+      "loss": 0.2944260835647583,
+      "loss_ce": 0.0224534310400486,
+      "loss_iou": 1.1428030729293823,
+      "loss_num": 0.271484375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 358824696,
+      "step": 2076
+    },
+    {
+      "epoch": 0.7954806587514363,
+      "grad_norm": 41.178257398994205,
+      "learning_rate": 5e-06,
+      "loss": 0.1971,
+      "num_input_tokens_seen": 358997896,
+      "step": 2077
+    },
+    {
+      "epoch": 0.7954806587514363,
+      "loss": 0.19884341955184937,
+      "loss_ce": 0.021231111139059067,
+      "loss_iou": 1.0368378162384033,
+      "loss_num": 0.177734375,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 358997896,
+      "step": 2077
+    },
+    {
+      "epoch": 0.795863653772501,
+      "grad_norm": 37.12009418828327,
+      "learning_rate": 5e-06,
+      "loss": 0.306,
+      "num_input_tokens_seen": 359170800,
+      "step": 2078
+    },
+    {
+      "epoch": 0.795863653772501,
+      "loss": 0.3145269453525543,
+      "loss_ce": 0.020886801183223724,
+      "loss_iou": 1.0644571781158447,
+      "loss_num": 0.29296875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 359170800,
+      "step": 2078
+    },
+    {
+      "epoch": 0.7962466487935657,
+      "grad_norm": 30.776086178949864,
+      "learning_rate": 5e-06,
+      "loss": 0.2334,
+      "num_input_tokens_seen": 359343760,
+      "step": 2079
+    },
+    {
+      "epoch": 0.7962466487935657,
+      "loss": 0.24104654788970947,
+      "loss_ce": 0.02095378190279007,
+      "loss_iou": 1.2850362062454224,
+      "loss_num": 0.2197265625,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 359343760,
+      "step": 2079
+    },
+    {
+      "epoch": 0.7966296438146304,
+      "grad_norm": 22.805719872927575,
+      "learning_rate": 5e-06,
+      "loss": 0.1952,
+      "num_input_tokens_seen": 359516672,
+      "step": 2080
+    },
+    {
+      "epoch": 0.7966296438146304,
+      "loss": 0.19519786536693573,
+      "loss_ce": 0.019965939223766327,
+      "loss_iou": 1.0284029245376587,
+      "loss_num": 0.1748046875,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 359516672,
+      "step": 2080
+    },
+    {
+      "epoch": 0.7970126388356952,
+      "grad_norm": 44.0592154790448,
+      "learning_rate": 5e-06,
+      "loss": 0.2271,
+      "num_input_tokens_seen": 359689872,
+      "step": 2081
+    },
+    {
+      "epoch": 0.7970126388356952,
+      "loss": 0.15045644342899323,
+      "loss_ce": 0.020024321973323822,
+      "loss_iou": 1.0021027326583862,
+      "loss_num": 0.130859375,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 359689872,
+      "step": 2081
+    },
+    {
+      "epoch": 0.7973956338567598,
+      "grad_norm": 23.506748884147438,
+      "learning_rate": 5e-06,
+      "loss": 0.2539,
+      "num_input_tokens_seen": 359862848,
+      "step": 2082
+    },
+    {
+      "epoch": 0.7973956338567598,
+      "loss": 0.2538887858390808,
+      "loss_ce": 0.019818957895040512,
+      "loss_iou": 1.2948635816574097,
+      "loss_num": 0.234375,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 359862848,
+      "step": 2082
+    },
+    {
+      "epoch": 0.7977786288778246,
+      "grad_norm": 28.647543722436414,
+      "learning_rate": 5e-06,
+      "loss": 0.2223,
+      "num_input_tokens_seen": 360035736,
+      "step": 2083
+    },
+    {
+      "epoch": 0.7977786288778246,
+      "loss": 0.21334119141101837,
+      "loss_ce": 0.01985974609851837,
+      "loss_iou": 1.0197844505310059,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 360035736,
+      "step": 2083
+    },
+    {
+      "epoch": 0.7981616238988893,
+      "grad_norm": 30.10333488122401,
+      "learning_rate": 5e-06,
+      "loss": 0.1708,
+      "num_input_tokens_seen": 360208664,
+      "step": 2084
+    },
+    {
+      "epoch": 0.7981616238988893,
+      "loss": 0.15990781784057617,
+      "loss_ce": 0.021449562162160873,
+      "loss_iou": 1.0397412776947021,
+      "loss_num": 0.138671875,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 360208664,
+      "step": 2084
+    },
+    {
+      "epoch": 0.798544618919954,
+      "grad_norm": 52.63208100675534,
+      "learning_rate": 5e-06,
+      "loss": 0.2085,
+      "num_input_tokens_seen": 360381704,
+      "step": 2085
+    },
+    {
+      "epoch": 0.798544618919954,
+      "loss": 0.20577174425125122,
+      "loss_ce": 0.02199489437043667,
+      "loss_iou": 1.0019832849502563,
+      "loss_num": 0.18359375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 360381704,
+      "step": 2085
+    },
+    {
+      "epoch": 0.7989276139410187,
+      "grad_norm": 29.04375222577297,
+      "learning_rate": 5e-06,
+      "loss": 0.1741,
+      "num_input_tokens_seen": 360554776,
+      "step": 2086
+    },
+    {
+      "epoch": 0.7989276139410187,
+      "loss": 0.2163219153881073,
+      "loss_ce": 0.0211925208568573,
+      "loss_iou": 1.025056004524231,
+      "loss_num": 0.1953125,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 360554776,
+      "step": 2086
+    },
+    {
+      "epoch": 0.7993106089620835,
+      "grad_norm": 29.037543914778514,
+      "learning_rate": 5e-06,
+      "loss": 0.2268,
+      "num_input_tokens_seen": 360727624,
+      "step": 2087
+    },
+    {
+      "epoch": 0.7993106089620835,
+      "loss": 0.2393513023853302,
+      "loss_ce": 0.021150611340999603,
+      "loss_iou": 1.039138674736023,
+      "loss_num": 0.2177734375,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 360727624,
+      "step": 2087
+    },
+    {
+      "epoch": 0.7996936039831483,
+      "grad_norm": 36.23397125700912,
+      "learning_rate": 5e-06,
+      "loss": 0.182,
+      "num_input_tokens_seen": 360900776,
+      "step": 2088
+    },
+    {
+      "epoch": 0.7996936039831483,
+      "loss": 0.17068441212177277,
+      "loss_ce": 0.020965175703167915,
+      "loss_iou": 1.0142332315444946,
+      "loss_num": 0.1494140625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 360900776,
+      "step": 2088
+    },
+    {
+      "epoch": 0.8000765990042129,
+      "grad_norm": 52.41293020388545,
+      "learning_rate": 5e-06,
+      "loss": 0.2191,
+      "num_input_tokens_seen": 361073576,
+      "step": 2089
+    },
+    {
+      "epoch": 0.8000765990042129,
+      "loss": 0.21941468119621277,
+      "loss_ce": 0.021477673202753067,
+      "loss_iou": 1.0017900466918945,
+      "loss_num": 0.1982421875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 361073576,
+      "step": 2089
+    },
+    {
+      "epoch": 0.8004595940252777,
+      "grad_norm": 30.35644681172131,
+      "learning_rate": 5e-06,
+      "loss": 0.1564,
+      "num_input_tokens_seen": 361246336,
+      "step": 2090
+    },
+    {
+      "epoch": 0.8004595940252777,
+      "loss": 0.16635313630104065,
+      "loss_ce": 0.020662210881710052,
+      "loss_iou": 1.0163631439208984,
+      "loss_num": 0.1455078125,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 361246336,
+      "step": 2090
+    },
+    {
+      "epoch": 0.8008425890463424,
+      "grad_norm": 33.49175637491388,
+      "learning_rate": 5e-06,
+      "loss": 0.2111,
+      "num_input_tokens_seen": 361419024,
+      "step": 2091
+    },
+    {
+      "epoch": 0.8008425890463424,
+      "loss": 0.23608174920082092,
+      "loss_ce": 0.021726282313466072,
+      "loss_iou": 1.0362515449523926,
+      "loss_num": 0.21484375,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 361419024,
+      "step": 2091
+    },
+    {
+      "epoch": 0.8012255840674071,
+      "grad_norm": 38.10898259614945,
+      "learning_rate": 5e-06,
+      "loss": 0.2263,
+      "num_input_tokens_seen": 361591832,
+      "step": 2092
+    },
+    {
+      "epoch": 0.8012255840674071,
+      "loss": 0.24866808950901031,
+      "loss_ce": 0.021129027009010315,
+      "loss_iou": 1.0237829685211182,
+      "loss_num": 0.2275390625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 361591832,
+      "step": 2092
+    },
+    {
+      "epoch": 0.8016085790884718,
+      "grad_norm": 60.87094645779356,
+      "learning_rate": 5e-06,
+      "loss": 0.2476,
+      "num_input_tokens_seen": 361764896,
+      "step": 2093
+    },
+    {
+      "epoch": 0.8016085790884718,
+      "loss": 0.2902504503726959,
+      "loss_ce": 0.01876608096063137,
+      "loss_iou": 1.036728858947754,
+      "loss_num": 0.271484375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 361764896,
+      "step": 2093
+    },
+    {
+      "epoch": 0.8019915741095366,
+      "grad_norm": 33.85406987960584,
+      "learning_rate": 5e-06,
+      "loss": 0.2625,
+      "num_input_tokens_seen": 361937568,
+      "step": 2094
+    },
+    {
+      "epoch": 0.8019915741095366,
+      "loss": 0.19497427344322205,
+      "loss_ce": 0.02126821130514145,
+      "loss_iou": 1.0377048254013062,
+      "loss_num": 0.173828125,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 361937568,
+      "step": 2094
+    },
+    {
+      "epoch": 0.8023745691306013,
+      "grad_norm": 22.97566587849187,
+      "learning_rate": 5e-06,
+      "loss": 0.2191,
+      "num_input_tokens_seen": 362110456,
+      "step": 2095
+    },
+    {
+      "epoch": 0.8023745691306013,
+      "loss": 0.2223237156867981,
+      "loss_ce": 0.02029736153781414,
+      "loss_iou": 1.0889081954956055,
+      "loss_num": 0.2021484375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 362110456,
+      "step": 2095
+    },
+    {
+      "epoch": 0.802757564151666,
+      "grad_norm": 22.655194333177427,
+      "learning_rate": 5e-06,
+      "loss": 0.2478,
+      "num_input_tokens_seen": 362283680,
+      "step": 2096
+    },
+    {
+      "epoch": 0.802757564151666,
+      "loss": 0.27080637216567993,
+      "loss_ce": 0.02065379172563553,
+      "loss_iou": 1.3040798902511597,
+      "loss_num": 0.25,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 362283680,
+      "step": 2096
+    },
+    {
+      "epoch": 0.8031405591727308,
+      "grad_norm": 23.83590037094105,
+      "learning_rate": 5e-06,
+      "loss": 0.2786,
+      "num_input_tokens_seen": 362456224,
+      "step": 2097
+    },
+    {
+      "epoch": 0.8031405591727308,
+      "loss": 0.2533208727836609,
+      "loss_ce": 0.020532775670289993,
+      "loss_iou": 1.056600570678711,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 362456224,
+      "step": 2097
+    },
+    {
+      "epoch": 0.8035235541937955,
+      "grad_norm": 27.137259830250642,
+      "learning_rate": 5e-06,
+      "loss": 0.1775,
+      "num_input_tokens_seen": 362628992,
+      "step": 2098
+    },
+    {
+      "epoch": 0.8035235541937955,
+      "loss": 0.15176838636398315,
+      "loss_ce": 0.020909003913402557,
+      "loss_iou": 0.9796233773231506,
+      "loss_num": 0.130859375,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 362628992,
+      "step": 2098
+    },
+    {
+      "epoch": 0.8039065492148602,
+      "grad_norm": 28.31381030618961,
+      "learning_rate": 5e-06,
+      "loss": 0.1969,
+      "num_input_tokens_seen": 362798384,
+      "step": 2099
+    },
+    {
+      "epoch": 0.8039065492148602,
+      "loss": 0.1582416445016861,
+      "loss_ce": 0.022682560607790947,
+      "loss_iou": 1.0035731792449951,
+      "loss_num": 0.1357421875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 362798384,
+      "step": 2099
+    },
+    {
+      "epoch": 0.8042895442359249,
+      "grad_norm": 33.032148385521666,
+      "learning_rate": 5e-06,
+      "loss": 0.2686,
+      "num_input_tokens_seen": 362971272,
+      "step": 2100
+    },
+    {
+      "epoch": 0.8042895442359249,
+      "loss": 0.28126421570777893,
+      "loss_ce": 0.01985062286257744,
+      "loss_iou": 1.083396553993225,
+      "loss_num": 0.26171875,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 362971272,
+      "step": 2100
+    },
+    {
+      "epoch": 0.8046725392569897,
+      "grad_norm": 22.602817795148876,
+      "learning_rate": 5e-06,
+      "loss": 0.1724,
+      "num_input_tokens_seen": 363144400,
+      "step": 2101
+    },
+    {
+      "epoch": 0.8046725392569897,
+      "loss": 0.1815548539161682,
+      "loss_ce": 0.02103240415453911,
+      "loss_iou": 1.030548334121704,
+      "loss_num": 0.16015625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 363144400,
+      "step": 2101
+    },
+    {
+      "epoch": 0.8050555342780544,
+      "grad_norm": 38.47049504384218,
+      "learning_rate": 5e-06,
+      "loss": 0.2397,
+      "num_input_tokens_seen": 363317320,
+      "step": 2102
+    },
+    {
+      "epoch": 0.8050555342780544,
+      "loss": 0.18445762991905212,
+      "loss_ce": 0.021615827456116676,
+      "loss_iou": 0.9913636445999146,
+      "loss_num": 0.1630859375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 363317320,
+      "step": 2102
+    },
+    {
+      "epoch": 0.8054385292991191,
+      "grad_norm": 25.713987475296403,
+      "learning_rate": 5e-06,
+      "loss": 0.1688,
+      "num_input_tokens_seen": 363490280,
+      "step": 2103
+    },
+    {
+      "epoch": 0.8054385292991191,
+      "loss": 0.18460124731063843,
+      "loss_ce": 0.021393220871686935,
+      "loss_iou": 1.0054594278335571,
+      "loss_num": 0.1630859375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 363490280,
+      "step": 2103
+    },
+    {
+      "epoch": 0.8058215243201838,
+      "grad_norm": 23.378121283979322,
+      "learning_rate": 5e-06,
+      "loss": 0.1959,
+      "num_input_tokens_seen": 363663168,
+      "step": 2104
+    },
+    {
+      "epoch": 0.8058215243201838,
+      "loss": 0.22951623797416687,
+      "loss_ce": 0.021142205223441124,
+      "loss_iou": 1.0498735904693604,
+      "loss_num": 0.2080078125,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 363663168,
+      "step": 2104
+    },
+    {
+      "epoch": 0.8062045193412486,
+      "grad_norm": 17.599769831356202,
+      "learning_rate": 5e-06,
+      "loss": 0.2367,
+      "num_input_tokens_seen": 363835856,
+      "step": 2105
+    },
+    {
+      "epoch": 0.8062045193412486,
+      "loss": 0.17348536849021912,
+      "loss_ce": 0.02077540010213852,
+      "loss_iou": 1.0428988933563232,
+      "loss_num": 0.15234375,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 363835856,
+      "step": 2105
+    },
+    {
+      "epoch": 0.8065875143623132,
+      "grad_norm": 19.769912404207556,
+      "learning_rate": 5e-06,
+      "loss": 0.1991,
+      "num_input_tokens_seen": 364009184,
+      "step": 2106
+    },
+    {
+      "epoch": 0.8065875143623132,
+      "loss": 0.17005111277103424,
+      "loss_ce": 0.021247392520308495,
+      "loss_iou": 1.009874701499939,
+      "loss_num": 0.1484375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 364009184,
+      "step": 2106
+    },
+    {
+      "epoch": 0.806970509383378,
+      "grad_norm": 26.649027915265457,
+      "learning_rate": 5e-06,
+      "loss": 0.1644,
+      "num_input_tokens_seen": 364182736,
+      "step": 2107
+    },
+    {
+      "epoch": 0.806970509383378,
+      "loss": 0.17446421086788177,
+      "loss_ce": 0.020228378474712372,
+      "loss_iou": 0.9851024150848389,
+      "loss_num": 0.154296875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 364182736,
+      "step": 2107
+    },
+    {
+      "epoch": 0.8073535044044428,
+      "grad_norm": 35.44751979570732,
+      "learning_rate": 5e-06,
+      "loss": 0.2026,
+      "num_input_tokens_seen": 364355976,
+      "step": 2108
+    },
+    {
+      "epoch": 0.8073535044044428,
+      "loss": 0.20467320084571838,
+      "loss_ce": 0.022361187264323235,
+      "loss_iou": 1.039179801940918,
+      "loss_num": 0.1826171875,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 364355976,
+      "step": 2108
+    },
+    {
+      "epoch": 0.8077364994255075,
+      "grad_norm": 28.703068016312443,
+      "learning_rate": 5e-06,
+      "loss": 0.2332,
+      "num_input_tokens_seen": 364528720,
+      "step": 2109
+    },
+    {
+      "epoch": 0.8077364994255075,
+      "loss": 0.21477672457695007,
+      "loss_ce": 0.019342154264450073,
+      "loss_iou": 1.0085649490356445,
+      "loss_num": 0.1953125,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 364528720,
+      "step": 2109
+    },
+    {
+      "epoch": 0.8081194944465722,
+      "grad_norm": 28.63109062864066,
+      "learning_rate": 5e-06,
+      "loss": 0.1813,
+      "num_input_tokens_seen": 364701560,
+      "step": 2110
+    },
+    {
+      "epoch": 0.8081194944465722,
+      "loss": 0.14917662739753723,
+      "loss_ce": 0.021033326163887978,
+      "loss_iou": 1.0138494968414307,
+      "loss_num": 0.1279296875,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 364701560,
+      "step": 2110
+    },
+    {
+      "epoch": 0.8085024894676369,
+      "grad_norm": 38.70015498466309,
+      "learning_rate": 5e-06,
+      "loss": 0.1954,
+      "num_input_tokens_seen": 364874104,
+      "step": 2111
+    },
+    {
+      "epoch": 0.8085024894676369,
+      "loss": 0.18069684505462646,
+      "loss_ce": 0.01962505653500557,
+      "loss_iou": 1.0061640739440918,
+      "loss_num": 0.1611328125,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 364874104,
+      "step": 2111
+    },
+    {
+      "epoch": 0.8088854844887017,
+      "grad_norm": 33.512475942128496,
+      "learning_rate": 5e-06,
+      "loss": 0.2515,
+      "num_input_tokens_seen": 365047352,
+      "step": 2112
+    },
+    {
+      "epoch": 0.8088854844887017,
+      "loss": 0.3174218237400055,
+      "loss_ce": 0.021706480532884598,
+      "loss_iou": 1.0606505870819092,
+      "loss_num": 0.294921875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 365047352,
+      "step": 2112
+    },
+    {
+      "epoch": 0.8092684795097663,
+      "grad_norm": 29.020052069927704,
+      "learning_rate": 5e-06,
+      "loss": 0.1748,
+      "num_input_tokens_seen": 365220232,
+      "step": 2113
+    },
+    {
+      "epoch": 0.8092684795097663,
+      "loss": 0.15384788811206818,
+      "loss_ce": 0.02005882002413273,
+      "loss_iou": 1.007267951965332,
+      "loss_num": 0.1337890625,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 365220232,
+      "step": 2113
+    },
+    {
+      "epoch": 0.8096514745308311,
+      "grad_norm": 24.396770407285977,
+      "learning_rate": 5e-06,
+      "loss": 0.2347,
+      "num_input_tokens_seen": 365393064,
+      "step": 2114
+    },
+    {
+      "epoch": 0.8096514745308311,
+      "loss": 0.2564903795719147,
+      "loss_ce": 0.020162250846624374,
+      "loss_iou": 1.3626725673675537,
+      "loss_num": 0.236328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 365393064,
+      "step": 2114
+    },
+    {
+      "epoch": 0.8100344695518958,
+      "grad_norm": 40.716933413128686,
+      "learning_rate": 5e-06,
+      "loss": 0.3466,
+      "num_input_tokens_seen": 365566056,
+      "step": 2115
+    },
+    {
+      "epoch": 0.8100344695518958,
+      "loss": 0.30868810415267944,
+      "loss_ce": 0.022250106558203697,
+      "loss_iou": 1.0419776439666748,
+      "loss_num": 0.287109375,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 365566056,
+      "step": 2115
+    },
+    {
+      "epoch": 0.8104174645729606,
+      "grad_norm": 28.101673655580367,
+      "learning_rate": 5e-06,
+      "loss": 0.2098,
+      "num_input_tokens_seen": 365739120,
+      "step": 2116
+    },
+    {
+      "epoch": 0.8104174645729606,
+      "loss": 0.24193474650382996,
+      "loss_ce": 0.02074335142970085,
+      "loss_iou": 1.0302741527557373,
+      "loss_num": 0.220703125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 365739120,
+      "step": 2116
+    },
+    {
+      "epoch": 0.8108004595940252,
+      "grad_norm": 27.215441670805387,
+      "learning_rate": 5e-06,
+      "loss": 0.2351,
+      "num_input_tokens_seen": 365912000,
+      "step": 2117
+    },
+    {
+      "epoch": 0.8108004595940252,
+      "loss": 0.1826138198375702,
+      "loss_ce": 0.020473934710025787,
+      "loss_iou": 1.0102052688598633,
+      "loss_num": 0.162109375,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 365912000,
+      "step": 2117
+    },
+    {
+      "epoch": 0.81118345461509,
+      "grad_norm": 30.17607812198289,
+      "learning_rate": 5e-06,
+      "loss": 0.2069,
+      "num_input_tokens_seen": 366084704,
+      "step": 2118
+    },
+    {
+      "epoch": 0.81118345461509,
+      "loss": 0.20413601398468018,
+      "loss_ce": 0.022800570353865623,
+      "loss_iou": 2.548659324645996,
+      "loss_num": 0.181640625,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 366084704,
+      "step": 2118
+    },
+    {
+      "epoch": 0.8115664496361548,
+      "grad_norm": 35.92505267130526,
+      "learning_rate": 5e-06,
+      "loss": 0.196,
+      "num_input_tokens_seen": 366257608,
+      "step": 2119
+    },
+    {
+      "epoch": 0.8115664496361548,
+      "loss": 0.20702782273292542,
+      "loss_ce": 0.020748518407344818,
+      "loss_iou": 1.0230991840362549,
+      "loss_num": 0.1865234375,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 366257608,
+      "step": 2119
+    },
+    {
+      "epoch": 0.8119494446572194,
+      "grad_norm": 29.807727312866955,
+      "learning_rate": 5e-06,
+      "loss": 0.2408,
+      "num_input_tokens_seen": 366430608,
+      "step": 2120
+    },
+    {
+      "epoch": 0.8119494446572194,
+      "loss": 0.2166225165128708,
+      "loss_ce": 0.021554159000515938,
+      "loss_iou": 1.045461654663086,
+      "loss_num": 0.1953125,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 366430608,
+      "step": 2120
+    },
+    {
+      "epoch": 0.8123324396782842,
+      "grad_norm": 27.749053225360697,
+      "learning_rate": 5e-06,
+      "loss": 0.2009,
+      "num_input_tokens_seen": 366603600,
+      "step": 2121
+    },
+    {
+      "epoch": 0.8123324396782842,
+      "loss": 0.21991193294525146,
+      "loss_ce": 0.021181467920541763,
+      "loss_iou": 1.0545415878295898,
+      "loss_num": 0.19921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 366603600,
+      "step": 2121
+    },
+    {
+      "epoch": 0.8127154346993489,
+      "grad_norm": 28.99954836164965,
+      "learning_rate": 5e-06,
+      "loss": 0.1613,
+      "num_input_tokens_seen": 366776760,
+      "step": 2122
+    },
+    {
+      "epoch": 0.8127154346993489,
+      "loss": 0.1873789131641388,
+      "loss_ce": 0.019562749192118645,
+      "loss_iou": 1.0477802753448486,
+      "loss_num": 0.16796875,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 366776760,
+      "step": 2122
+    },
+    {
+      "epoch": 0.8130984297204137,
+      "grad_norm": 29.983256245017014,
+      "learning_rate": 5e-06,
+      "loss": 0.2394,
+      "num_input_tokens_seen": 366949696,
+      "step": 2123
+    },
+    {
+      "epoch": 0.8130984297204137,
+      "loss": 0.28331607580184937,
+      "loss_ce": 0.02055971696972847,
+      "loss_iou": 1.051964282989502,
+      "loss_num": 0.263671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 366949696,
+      "step": 2123
+    },
+    {
+      "epoch": 0.8134814247414783,
+      "grad_norm": 18.967397198416126,
+      "learning_rate": 5e-06,
+      "loss": 0.2025,
+      "num_input_tokens_seen": 367122664,
+      "step": 2124
+    },
+    {
+      "epoch": 0.8134814247414783,
+      "loss": 0.1603628695011139,
+      "loss_ce": 0.02062288671731949,
+      "loss_iou": 1.014026165008545,
+      "loss_num": 0.1396484375,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 367122664,
+      "step": 2124
+    },
+    {
+      "epoch": 0.8138644197625431,
+      "grad_norm": 23.401417353363087,
+      "learning_rate": 5e-06,
+      "loss": 0.2419,
+      "num_input_tokens_seen": 367295616,
+      "step": 2125
+    },
+    {
+      "epoch": 0.8138644197625431,
+      "loss": 0.24524861574172974,
+      "loss_ce": 0.02088339626789093,
+      "loss_iou": 1.0667381286621094,
+      "loss_num": 0.224609375,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 367295616,
+      "step": 2125
+    },
+    {
+      "epoch": 0.8142474147836078,
+      "grad_norm": 18.681374620707405,
+      "learning_rate": 5e-06,
+      "loss": 0.1829,
+      "num_input_tokens_seen": 367468840,
+      "step": 2126
+    },
+    {
+      "epoch": 0.8142474147836078,
+      "loss": 0.13272728025913239,
+      "loss_ce": 0.020514145493507385,
+      "loss_iou": 1.0010244846343994,
+      "loss_num": 0.1123046875,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 367468840,
+      "step": 2126
+    },
+    {
+      "epoch": 0.8146304098046725,
+      "grad_norm": 24.468260270885658,
+      "learning_rate": 5e-06,
+      "loss": 0.2504,
+      "num_input_tokens_seen": 367641648,
+      "step": 2127
+    },
+    {
+      "epoch": 0.8146304098046725,
+      "loss": 0.25432175397872925,
+      "loss_ce": 0.021655723452568054,
+      "loss_iou": 1.0185184478759766,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 367641648,
+      "step": 2127
+    },
+    {
+      "epoch": 0.8150134048257373,
+      "grad_norm": 28.332489760750022,
+      "learning_rate": 5e-06,
+      "loss": 0.2322,
+      "num_input_tokens_seen": 367814864,
+      "step": 2128
+    },
+    {
+      "epoch": 0.8150134048257373,
+      "loss": 0.23658353090286255,
+      "loss_ce": 0.021068410947918892,
+      "loss_iou": 1.0070009231567383,
+      "loss_num": 0.2158203125,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 367814864,
+      "step": 2128
+    },
+    {
+      "epoch": 0.815396399846802,
+      "grad_norm": 20.92409368983065,
+      "learning_rate": 5e-06,
+      "loss": 0.1697,
+      "num_input_tokens_seen": 367987808,
+      "step": 2129
+    },
+    {
+      "epoch": 0.815396399846802,
+      "loss": 0.16304948925971985,
+      "loss_ce": 0.02010514587163925,
+      "loss_iou": 1.0108318328857422,
+      "loss_num": 0.142578125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 367987808,
+      "step": 2129
+    },
+    {
+      "epoch": 0.8157793948678668,
+      "grad_norm": 16.311196285773622,
+      "learning_rate": 5e-06,
+      "loss": 0.1442,
+      "num_input_tokens_seen": 368161048,
+      "step": 2130
+    },
+    {
+      "epoch": 0.8157793948678668,
+      "loss": 0.1444142460823059,
+      "loss_ce": 0.020573917776346207,
+      "loss_iou": 1.0141609907150269,
+      "loss_num": 0.1240234375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 368161048,
+      "step": 2130
+    },
+    {
+      "epoch": 0.8161623898889314,
+      "grad_norm": 19.817899330667746,
+      "learning_rate": 5e-06,
+      "loss": 0.2286,
+      "num_input_tokens_seen": 368330792,
+      "step": 2131
+    },
+    {
+      "epoch": 0.8161623898889314,
+      "loss": 0.24902987480163574,
+      "loss_ce": 0.022406332194805145,
+      "loss_iou": 1.040596604347229,
+      "loss_num": 0.2265625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 368330792,
+      "step": 2131
+    },
+    {
+      "epoch": 0.8165453849099962,
+      "grad_norm": 21.139751947844744,
+      "learning_rate": 5e-06,
+      "loss": 0.2042,
+      "num_input_tokens_seen": 368503360,
+      "step": 2132
+    },
+    {
+      "epoch": 0.8165453849099962,
+      "loss": 0.19341111183166504,
+      "loss_ce": 0.020437486469745636,
+      "loss_iou": 1.0012788772583008,
+      "loss_num": 0.1728515625,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 368503360,
+      "step": 2132
+    },
+    {
+      "epoch": 0.8169283799310609,
+      "grad_norm": 19.840473264358938,
+      "learning_rate": 5e-06,
+      "loss": 0.1921,
+      "num_input_tokens_seen": 368676456,
+      "step": 2133
+    },
+    {
+      "epoch": 0.8169283799310609,
+      "loss": 0.17164191603660583,
+      "loss_ce": 0.020824046805500984,
+      "loss_iou": 1.0003976821899414,
+      "loss_num": 0.150390625,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 368676456,
+      "step": 2133
+    },
+    {
+      "epoch": 0.8173113749521256,
+      "grad_norm": 29.21394391065354,
+      "learning_rate": 5e-06,
+      "loss": 0.1859,
+      "num_input_tokens_seen": 368849576,
+      "step": 2134
+    },
+    {
+      "epoch": 0.8173113749521256,
+      "loss": 0.22054269909858704,
+      "loss_ce": 0.021262917667627335,
+      "loss_iou": 1.0249555110931396,
+      "loss_num": 0.19921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 368849576,
+      "step": 2134
+    },
+    {
+      "epoch": 0.8176943699731903,
+      "grad_norm": 40.56123718651517,
+      "learning_rate": 5e-06,
+      "loss": 0.2921,
+      "num_input_tokens_seen": 369022440,
+      "step": 2135
+    },
+    {
+      "epoch": 0.8176943699731903,
+      "loss": 0.3429931402206421,
+      "loss_ce": 0.021398894488811493,
+      "loss_iou": 1.076008915901184,
+      "loss_num": 0.322265625,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 369022440,
+      "step": 2135
+    },
+    {
+      "epoch": 0.8180773649942551,
+      "grad_norm": 32.061146496298235,
+      "learning_rate": 5e-06,
+      "loss": 0.2334,
+      "num_input_tokens_seen": 369195448,
+      "step": 2136
+    },
+    {
+      "epoch": 0.8180773649942551,
+      "loss": 0.21450859308242798,
+      "loss_ce": 0.020538877695798874,
+      "loss_iou": 1.0362865924835205,
+      "loss_num": 0.1943359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 369195448,
+      "step": 2136
+    },
+    {
+      "epoch": 0.8184603600153199,
+      "grad_norm": 28.584843416958208,
+      "learning_rate": 5e-06,
+      "loss": 0.2059,
+      "num_input_tokens_seen": 369368440,
+      "step": 2137
+    },
+    {
+      "epoch": 0.8184603600153199,
+      "loss": 0.1834711879491806,
+      "loss_ce": 0.0206293947994709,
+      "loss_iou": 1.011899471282959,
+      "loss_num": 0.1630859375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 369368440,
+      "step": 2137
+    },
+    {
+      "epoch": 0.8188433550363845,
+      "grad_norm": 22.691707761012573,
+      "learning_rate": 5e-06,
+      "loss": 0.1715,
+      "num_input_tokens_seen": 369541216,
+      "step": 2138
+    },
+    {
+      "epoch": 0.8188433550363845,
+      "loss": 0.15554609894752502,
+      "loss_ce": 0.020963573828339577,
+      "loss_iou": 1.0016136169433594,
+      "loss_num": 0.134765625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 369541216,
+      "step": 2138
+    },
+    {
+      "epoch": 0.8192263500574493,
+      "grad_norm": 36.32095499337774,
+      "learning_rate": 5e-06,
+      "loss": 0.2489,
+      "num_input_tokens_seen": 369714304,
+      "step": 2139
+    },
+    {
+      "epoch": 0.8192263500574493,
+      "loss": 0.296573281288147,
+      "loss_ce": 0.020084049552679062,
+      "loss_iou": 1.2641435861587524,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 369714304,
+      "step": 2139
+    },
+    {
+      "epoch": 0.819609345078514,
+      "grad_norm": 26.687714119843587,
+      "learning_rate": 5e-06,
+      "loss": 0.2286,
+      "num_input_tokens_seen": 369887256,
+      "step": 2140
+    },
+    {
+      "epoch": 0.819609345078514,
+      "loss": 0.2310848981142044,
+      "loss_ce": 0.021062923595309258,
+      "loss_iou": 1.0150922536849976,
+      "loss_num": 0.2099609375,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 369887256,
+      "step": 2140
+    },
+    {
+      "epoch": 0.8199923400995787,
+      "grad_norm": 42.32563606563892,
+      "learning_rate": 5e-06,
+      "loss": 0.3101,
+      "num_input_tokens_seen": 370059944,
+      "step": 2141
+    },
+    {
+      "epoch": 0.8199923400995787,
+      "loss": 0.2709304392337799,
+      "loss_ce": 0.021418720483779907,
+      "loss_iou": 1.113772988319397,
+      "loss_num": 0.25,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 370059944,
+      "step": 2141
+    },
+    {
+      "epoch": 0.8203753351206434,
+      "grad_norm": 37.72077567350258,
+      "learning_rate": 5e-06,
+      "loss": 0.1873,
+      "num_input_tokens_seen": 370232800,
+      "step": 2142
+    },
+    {
+      "epoch": 0.8203753351206434,
+      "loss": 0.2032146453857422,
+      "loss_ce": 0.02212332934141159,
+      "loss_iou": 1.0222200155258179,
+      "loss_num": 0.1806640625,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 370232800,
+      "step": 2142
+    },
+    {
+      "epoch": 0.8207583301417082,
+      "grad_norm": 28.976553942483406,
+      "learning_rate": 5e-06,
+      "loss": 0.2758,
+      "num_input_tokens_seen": 370405640,
+      "step": 2143
+    },
+    {
+      "epoch": 0.8207583301417082,
+      "loss": 0.31651657819747925,
+      "loss_ce": 0.022144028916954994,
+      "loss_iou": 1.1050704717636108,
+      "loss_num": 0.294921875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 370405640,
+      "step": 2143
+    },
+    {
+      "epoch": 0.8211413251627729,
+      "grad_norm": 29.255901172619154,
+      "learning_rate": 5e-06,
+      "loss": 0.1941,
+      "num_input_tokens_seen": 370578624,
+      "step": 2144
+    },
+    {
+      "epoch": 0.8211413251627729,
+      "loss": 0.1887333244085312,
+      "loss_ce": 0.020520437508821487,
+      "loss_iou": 1.041619062423706,
+      "loss_num": 0.16796875,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 370578624,
+      "step": 2144
+    },
+    {
+      "epoch": 0.8215243201838376,
+      "grad_norm": 46.483402494126544,
+      "learning_rate": 5e-06,
+      "loss": 0.2374,
+      "num_input_tokens_seen": 370751592,
+      "step": 2145
+    },
+    {
+      "epoch": 0.8215243201838376,
+      "loss": 0.2568110525608063,
+      "loss_ce": 0.021459471434354782,
+      "loss_iou": 1.0386011600494385,
+      "loss_num": 0.2353515625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 370751592,
+      "step": 2145
+    },
+    {
+      "epoch": 0.8219073152049023,
+      "grad_norm": 29.947747066676154,
+      "learning_rate": 5e-06,
+      "loss": 0.2778,
+      "num_input_tokens_seen": 370924456,
+      "step": 2146
+    },
+    {
+      "epoch": 0.8219073152049023,
+      "loss": 0.2346021831035614,
+      "loss_ce": 0.020063605159521103,
+      "loss_iou": 1.0243932008743286,
+      "loss_num": 0.21484375,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 370924456,
+      "step": 2146
+    },
+    {
+      "epoch": 0.8222903102259671,
+      "grad_norm": 28.665856752031125,
+      "learning_rate": 5e-06,
+      "loss": 0.2505,
+      "num_input_tokens_seen": 371097616,
+      "step": 2147
+    },
+    {
+      "epoch": 0.8222903102259671,
+      "loss": 0.2775253653526306,
+      "loss_ce": 0.022703586146235466,
+      "loss_iou": 1.0256977081298828,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 371097616,
+      "step": 2147
+    },
+    {
+      "epoch": 0.8226733052470317,
+      "grad_norm": 42.77345054291103,
+      "learning_rate": 5e-06,
+      "loss": 0.204,
+      "num_input_tokens_seen": 371270672,
+      "step": 2148
+    },
+    {
+      "epoch": 0.8226733052470317,
+      "loss": 0.20909292995929718,
+      "loss_ce": 0.02183707244694233,
+      "loss_iou": 17.05697250366211,
+      "loss_num": 0.1875,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 371270672,
+      "step": 2148
+    },
+    {
+      "epoch": 0.8230563002680965,
+      "grad_norm": 64.91123199264825,
+      "learning_rate": 5e-06,
+      "loss": 0.2677,
+      "num_input_tokens_seen": 371443352,
+      "step": 2149
+    },
+    {
+      "epoch": 0.8230563002680965,
+      "loss": 0.24871593713760376,
+      "loss_ce": 0.019834104925394058,
+      "loss_iou": 1.0209572315216064,
+      "loss_num": 0.228515625,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 371443352,
+      "step": 2149
+    },
+    {
+      "epoch": 0.8234392952891613,
+      "grad_norm": 26.387438541462288,
+      "learning_rate": 5e-06,
+      "loss": 0.2468,
+      "num_input_tokens_seen": 371616552,
+      "step": 2150
+    },
+    {
+      "epoch": 0.8234392952891613,
+      "loss": 0.24685382843017578,
+      "loss_ce": 0.021817198023200035,
+      "loss_iou": 1.0309360027313232,
+      "loss_num": 0.224609375,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 371616552,
+      "step": 2150
+    },
+    {
+      "epoch": 0.8238222903102259,
+      "grad_norm": 63.38123772802693,
+      "learning_rate": 5e-06,
+      "loss": 0.3584,
+      "num_input_tokens_seen": 371785936,
+      "step": 2151
+    },
+    {
+      "epoch": 0.8238222903102259,
+      "loss": 0.3480418920516968,
+      "loss_ce": 0.01955069601535797,
+      "loss_iou": 1.041642665863037,
+      "loss_num": 0.328125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 371785936,
+      "step": 2151
+    },
+    {
+      "epoch": 0.8242052853312907,
+      "grad_norm": 29.862904214490893,
+      "learning_rate": 5e-06,
+      "loss": 0.2428,
+      "num_input_tokens_seen": 371958816,
+      "step": 2152
+    },
+    {
+      "epoch": 0.8242052853312907,
+      "loss": 0.22665396332740784,
+      "loss_ce": 0.018768219277262688,
+      "loss_iou": 0.9489197731018066,
+      "loss_num": 0.2080078125,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 371958816,
+      "step": 2152
+    },
+    {
+      "epoch": 0.8245882803523554,
+      "grad_norm": 91.69820226345597,
+      "learning_rate": 5e-06,
+      "loss": 0.4184,
+      "num_input_tokens_seen": 372131608,
+      "step": 2153
+    },
+    {
+      "epoch": 0.8245882803523554,
+      "loss": 0.3356097936630249,
+      "loss_ce": 0.021889086812734604,
+      "loss_iou": 1.030665397644043,
+      "loss_num": 0.314453125,
+      "loss_xval": 0.314453125,
+      "num_input_tokens_seen": 372131608,
+      "step": 2153
+    },
+    {
+      "epoch": 0.8249712753734202,
+      "grad_norm": 42.06860118666766,
+      "learning_rate": 5e-06,
+      "loss": 0.352,
+      "num_input_tokens_seen": 372304832,
+      "step": 2154
+    },
+    {
+      "epoch": 0.8249712753734202,
+      "loss": 0.35705554485321045,
+      "loss_ce": 0.022827016189694405,
+      "loss_iou": 1.0636298656463623,
+      "loss_num": 0.333984375,
+      "loss_xval": 0.333984375,
+      "num_input_tokens_seen": 372304832,
+      "step": 2154
+    },
+    {
+      "epoch": 0.8253542703944848,
+      "grad_norm": 61.00509001393007,
+      "learning_rate": 5e-06,
+      "loss": 0.3619,
+      "num_input_tokens_seen": 372477664,
+      "step": 2155
+    },
+    {
+      "epoch": 0.8253542703944848,
+      "loss": 0.36045730113983154,
+      "loss_ce": 0.020613551139831543,
+      "loss_iou": 1.0060477256774902,
+      "loss_num": 0.33984375,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 372477664,
+      "step": 2155
+    },
+    {
+      "epoch": 0.8257372654155496,
+      "grad_norm": 37.30844281452132,
+      "learning_rate": 5e-06,
+      "loss": 0.34,
+      "num_input_tokens_seen": 372650296,
+      "step": 2156
+    },
+    {
+      "epoch": 0.8257372654155496,
+      "loss": 0.31085580587387085,
+      "loss_ce": 0.020328452810645103,
+      "loss_iou": 1.030238389968872,
+      "loss_num": 0.291015625,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 372650296,
+      "step": 2156
+    },
+    {
+      "epoch": 0.8261202604366144,
+      "grad_norm": 82.53127759207469,
+      "learning_rate": 5e-06,
+      "loss": 0.4079,
+      "num_input_tokens_seen": 372822944,
+      "step": 2157
+    },
+    {
+      "epoch": 0.8261202604366144,
+      "loss": 0.376138836145401,
+      "loss_ce": 0.02005971595644951,
+      "loss_iou": 1.04275643825531,
+      "loss_num": 0.35546875,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 372822944,
+      "step": 2157
+    },
+    {
+      "epoch": 0.826503255457679,
+      "grad_norm": 60.30255454366365,
+      "learning_rate": 5e-06,
+      "loss": 0.4474,
+      "num_input_tokens_seen": 372995960,
+      "step": 2158
+    },
+    {
+      "epoch": 0.826503255457679,
+      "loss": 0.41652911901474,
+      "loss_ce": 0.020288866013288498,
+      "loss_iou": 1.0174659490585327,
+      "loss_num": 0.396484375,
+      "loss_xval": 0.396484375,
+      "num_input_tokens_seen": 372995960,
+      "step": 2158
+    },
+    {
+      "epoch": 0.8268862504787438,
+      "grad_norm": 28.334335210002113,
+      "learning_rate": 5e-06,
+      "loss": 0.2853,
+      "num_input_tokens_seen": 373169096,
+      "step": 2159
+    },
+    {
+      "epoch": 0.8268862504787438,
+      "loss": 0.3576209843158722,
+      "loss_ce": 0.02180556021630764,
+      "loss_iou": 19.851102828979492,
+      "loss_num": 0.3359375,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 373169096,
+      "step": 2159
+    },
+    {
+      "epoch": 0.8272692454998085,
+      "grad_norm": 49.53509686154801,
+      "learning_rate": 5e-06,
+      "loss": 0.4267,
+      "num_input_tokens_seen": 373341848,
+      "step": 2160
+    },
+    {
+      "epoch": 0.8272692454998085,
+      "loss": 0.41843438148498535,
+      "loss_ce": 0.0207293089479208,
+      "loss_iou": 1.0405924320220947,
+      "loss_num": 0.3984375,
+      "loss_xval": 0.3984375,
+      "num_input_tokens_seen": 373341848,
+      "step": 2160
+    },
+    {
+      "epoch": 0.8276522405208733,
+      "grad_norm": 20.37068228369542,
+      "learning_rate": 5e-06,
+      "loss": 0.2118,
+      "num_input_tokens_seen": 373514824,
+      "step": 2161
+    },
+    {
+      "epoch": 0.8276522405208733,
+      "loss": 0.24461804330348969,
+      "loss_ce": 0.019581424072384834,
+      "loss_iou": 0.8936472535133362,
+      "loss_num": 0.224609375,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 373514824,
+      "step": 2161
+    },
+    {
+      "epoch": 0.8280352355419379,
+      "grad_norm": 33.944329710520314,
+      "learning_rate": 5e-06,
+      "loss": 0.2168,
+      "num_input_tokens_seen": 373687448,
+      "step": 2162
+    },
+    {
+      "epoch": 0.8280352355419379,
+      "loss": 0.21628016233444214,
+      "loss_ce": 0.021333884447813034,
+      "loss_iou": 1.006072998046875,
+      "loss_num": 0.1953125,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 373687448,
+      "step": 2162
+    },
+    {
+      "epoch": 0.8284182305630027,
+      "grad_norm": 27.412374432146066,
+      "learning_rate": 5e-06,
+      "loss": 0.2433,
+      "num_input_tokens_seen": 373860320,
+      "step": 2163
+    },
+    {
+      "epoch": 0.8284182305630027,
+      "loss": 0.20806829631328583,
+      "loss_ce": 0.020690374076366425,
+      "loss_iou": 1.0283517837524414,
+      "loss_num": 0.1875,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 373860320,
+      "step": 2163
+    },
+    {
+      "epoch": 0.8288012255840674,
+      "grad_norm": 19.06495389278362,
+      "learning_rate": 5e-06,
+      "loss": 0.1999,
+      "num_input_tokens_seen": 374033112,
+      "step": 2164
+    },
+    {
+      "epoch": 0.8288012255840674,
+      "loss": 0.2409350574016571,
+      "loss_ce": 0.019987797364592552,
+      "loss_iou": 1.3699593544006348,
+      "loss_num": 0.220703125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 374033112,
+      "step": 2164
+    },
+    {
+      "epoch": 0.8291842206051321,
+      "grad_norm": 27.206349337833004,
+      "learning_rate": 5e-06,
+      "loss": 0.2062,
+      "num_input_tokens_seen": 374206168,
+      "step": 2165
+    },
+    {
+      "epoch": 0.8291842206051321,
+      "loss": 0.18531841039657593,
+      "loss_ce": 0.02137799561023712,
+      "loss_iou": 1.0548542737960815,
+      "loss_num": 0.1640625,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 374206168,
+      "step": 2165
+    },
+    {
+      "epoch": 0.8295672156261968,
+      "grad_norm": 25.131795982644185,
+      "learning_rate": 5e-06,
+      "loss": 0.1718,
+      "num_input_tokens_seen": 374379016,
+      "step": 2166
+    },
+    {
+      "epoch": 0.8295672156261968,
+      "loss": 0.13089054822921753,
+      "loss_ce": 0.022309008985757828,
+      "loss_iou": 1.0004377365112305,
+      "loss_num": 0.1083984375,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 374379016,
+      "step": 2166
+    },
+    {
+      "epoch": 0.8299502106472616,
+      "grad_norm": 29.096596528635764,
+      "learning_rate": 5e-06,
+      "loss": 0.1633,
+      "num_input_tokens_seen": 374552040,
+      "step": 2167
+    },
+    {
+      "epoch": 0.8299502106472616,
+      "loss": 0.16615352034568787,
+      "loss_ce": 0.021866418421268463,
+      "loss_iou": 1.0073778629302979,
+      "loss_num": 0.14453125,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 374552040,
+      "step": 2167
+    },
+    {
+      "epoch": 0.8303332056683264,
+      "grad_norm": 27.675018344639614,
+      "learning_rate": 5e-06,
+      "loss": 0.2255,
+      "num_input_tokens_seen": 374724824,
+      "step": 2168
+    },
+    {
+      "epoch": 0.8303332056683264,
+      "loss": 0.21904276311397552,
+      "loss_ce": 0.02134990133345127,
+      "loss_iou": 1.0683300495147705,
+      "loss_num": 0.197265625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 374724824,
+      "step": 2168
+    },
+    {
+      "epoch": 0.830716200689391,
+      "grad_norm": 28.67479113154888,
+      "learning_rate": 5e-06,
+      "loss": 0.1988,
+      "num_input_tokens_seen": 374897856,
+      "step": 2169
+    },
+    {
+      "epoch": 0.830716200689391,
+      "loss": 0.2088775783777237,
+      "loss_ce": 0.020645156502723694,
+      "loss_iou": 1.0369899272918701,
+      "loss_num": 0.1884765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 374897856,
+      "step": 2169
+    },
+    {
+      "epoch": 0.8310991957104558,
+      "grad_norm": 20.00694280309562,
+      "learning_rate": 5e-06,
+      "loss": 0.191,
+      "num_input_tokens_seen": 375070864,
+      "step": 2170
+    },
+    {
+      "epoch": 0.8310991957104558,
+      "loss": 0.16283264756202698,
+      "loss_ce": 0.02043761871755123,
+      "loss_iou": 1.0019893646240234,
+      "loss_num": 0.142578125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 375070864,
+      "step": 2170
+    },
+    {
+      "epoch": 0.8314821907315205,
+      "grad_norm": 20.537005457242522,
+      "learning_rate": 5e-06,
+      "loss": 0.2047,
+      "num_input_tokens_seen": 375243624,
+      "step": 2171
+    },
+    {
+      "epoch": 0.8314821907315205,
+      "loss": 0.14835287630558014,
+      "loss_ce": 0.02231527492403984,
+      "loss_iou": 1.0038883686065674,
+      "loss_num": 0.1259765625,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 375243624,
+      "step": 2171
+    },
+    {
+      "epoch": 0.8318651857525852,
+      "grad_norm": 24.981646224644035,
+      "learning_rate": 5e-06,
+      "loss": 0.2113,
+      "num_input_tokens_seen": 375416456,
+      "step": 2172
+    },
+    {
+      "epoch": 0.8318651857525852,
+      "loss": 0.23956190049648285,
+      "loss_ce": 0.019530151039361954,
+      "loss_iou": 1.0428447723388672,
+      "loss_num": 0.2197265625,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 375416456,
+      "step": 2172
+    },
+    {
+      "epoch": 0.8322481807736499,
+      "grad_norm": 26.02767440825196,
+      "learning_rate": 5e-06,
+      "loss": 0.1984,
+      "num_input_tokens_seen": 375589656,
+      "step": 2173
+    },
+    {
+      "epoch": 0.8322481807736499,
+      "loss": 0.19998681545257568,
+      "loss_ce": 0.021092761307954788,
+      "loss_iou": 1.0418498516082764,
+      "loss_num": 0.1787109375,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 375589656,
+      "step": 2173
+    },
+    {
+      "epoch": 0.8326311757947147,
+      "grad_norm": 22.477332073355125,
+      "learning_rate": 5e-06,
+      "loss": 0.1727,
+      "num_input_tokens_seen": 375762792,
+      "step": 2174
+    },
+    {
+      "epoch": 0.8326311757947147,
+      "loss": 0.15376584231853485,
+      "loss_ce": 0.021227996796369553,
+      "loss_iou": 1.002480387687683,
+      "loss_num": 0.1328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 375762792,
+      "step": 2174
+    },
+    {
+      "epoch": 0.8330141708157794,
+      "grad_norm": 26.873458656094986,
+      "learning_rate": 5e-06,
+      "loss": 0.2338,
+      "num_input_tokens_seen": 375935360,
+      "step": 2175
+    },
+    {
+      "epoch": 0.8330141708157794,
+      "loss": 0.22055332362651825,
+      "loss_ce": 0.021822858601808548,
+      "loss_iou": 1.0023974180221558,
+      "loss_num": 0.19921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 375935360,
+      "step": 2175
+    },
+    {
+      "epoch": 0.8333971658368441,
+      "grad_norm": 31.37132126335547,
+      "learning_rate": 5e-06,
+      "loss": 0.1972,
+      "num_input_tokens_seen": 376108328,
+      "step": 2176
+    },
+    {
+      "epoch": 0.8333971658368441,
+      "loss": 0.18834498524665833,
+      "loss_ce": 0.020925553515553474,
+      "loss_iou": 1.0201756954193115,
+      "loss_num": 0.1669921875,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 376108328,
+      "step": 2176
+    },
+    {
+      "epoch": 0.8337801608579088,
+      "grad_norm": 37.163570600558636,
+      "learning_rate": 5e-06,
+      "loss": 0.2749,
+      "num_input_tokens_seen": 376280864,
+      "step": 2177
+    },
+    {
+      "epoch": 0.8337801608579088,
+      "loss": 0.2612314820289612,
+      "loss_ce": 0.020569849759340286,
+      "loss_iou": 4.367435932159424,
+      "loss_num": 0.240234375,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 376280864,
+      "step": 2177
+    },
+    {
+      "epoch": 0.8341631558789736,
+      "grad_norm": 21.319412688034024,
+      "learning_rate": 5e-06,
+      "loss": 0.2065,
+      "num_input_tokens_seen": 376454120,
+      "step": 2178
+    },
+    {
+      "epoch": 0.8341631558789736,
+      "loss": 0.22772261500358582,
+      "loss_ce": 0.021118616685271263,
+      "loss_iou": 1.0112444162368774,
+      "loss_num": 0.20703125,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 376454120,
+      "step": 2178
+    },
+    {
+      "epoch": 0.8345461509000383,
+      "grad_norm": 21.729132395081642,
+      "learning_rate": 5e-06,
+      "loss": 0.2015,
+      "num_input_tokens_seen": 376626848,
+      "step": 2179
+    },
+    {
+      "epoch": 0.8345461509000383,
+      "loss": 0.23938342928886414,
+      "loss_ce": 0.020572395995259285,
+      "loss_iou": 1.0284411907196045,
+      "loss_num": 0.21875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 376626848,
+      "step": 2179
+    },
+    {
+      "epoch": 0.834929145921103,
+      "grad_norm": 20.853724556187558,
+      "learning_rate": 5e-06,
+      "loss": 0.2455,
+      "num_input_tokens_seen": 376799728,
+      "step": 2180
+    },
+    {
+      "epoch": 0.834929145921103,
+      "loss": 0.30870160460472107,
+      "loss_ce": 0.019578076899051666,
+      "loss_iou": 0.8902505040168762,
+      "loss_num": 0.2890625,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 376799728,
+      "step": 2180
+    },
+    {
+      "epoch": 0.8353121409421678,
+      "grad_norm": 28.389613043360605,
+      "learning_rate": 5e-06,
+      "loss": 0.1318,
+      "num_input_tokens_seen": 376972560,
+      "step": 2181
+    },
+    {
+      "epoch": 0.8353121409421678,
+      "loss": 0.13370594382286072,
+      "loss_ce": 0.020607803016901016,
+      "loss_iou": 1.0374066829681396,
+      "loss_num": 0.11328125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 376972560,
+      "step": 2181
+    },
+    {
+      "epoch": 0.8356951359632325,
+      "grad_norm": 29.161765278793478,
+      "learning_rate": 5e-06,
+      "loss": 0.2211,
+      "num_input_tokens_seen": 377145440,
+      "step": 2182
+    },
+    {
+      "epoch": 0.8356951359632325,
+      "loss": 0.2618965804576874,
+      "loss_ce": 0.021784283220767975,
+      "loss_iou": 1.1681959629058838,
+      "loss_num": 0.240234375,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 377145440,
+      "step": 2182
+    },
+    {
+      "epoch": 0.8360781309842972,
+      "grad_norm": 21.795808593225356,
+      "learning_rate": 5e-06,
+      "loss": 0.2082,
+      "num_input_tokens_seen": 377318280,
+      "step": 2183
+    },
+    {
+      "epoch": 0.8360781309842972,
+      "loss": 0.21658295392990112,
+      "loss_ce": 0.02066010981798172,
+      "loss_iou": 1.0167796611785889,
+      "loss_num": 0.1962890625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 377318280,
+      "step": 2183
+    },
+    {
+      "epoch": 0.8364611260053619,
+      "grad_norm": 22.245390840813947,
+      "learning_rate": 5e-06,
+      "loss": 0.159,
+      "num_input_tokens_seen": 377491168,
+      "step": 2184
+    },
+    {
+      "epoch": 0.8364611260053619,
+      "loss": 0.17089596390724182,
+      "loss_ce": 0.022031232714653015,
+      "loss_iou": 1.0310887098312378,
+      "loss_num": 0.1484375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 377491168,
+      "step": 2184
+    },
+    {
+      "epoch": 0.8368441210264267,
+      "grad_norm": 24.88724443711426,
+      "learning_rate": 5e-06,
+      "loss": 0.2646,
+      "num_input_tokens_seen": 377664032,
+      "step": 2185
+    },
+    {
+      "epoch": 0.8368441210264267,
+      "loss": 0.23490586876869202,
+      "loss_ce": 0.021710064262151718,
+      "loss_iou": 1.0489373207092285,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 377664032,
+      "step": 2185
+    },
+    {
+      "epoch": 0.8372271160474913,
+      "grad_norm": 22.16338435228188,
+      "learning_rate": 5e-06,
+      "loss": 0.1908,
+      "num_input_tokens_seen": 377837440,
+      "step": 2186
+    },
+    {
+      "epoch": 0.8372271160474913,
+      "loss": 0.2569584846496582,
+      "loss_ce": 0.020386239513754845,
+      "loss_iou": 1.0582001209259033,
+      "loss_num": 0.236328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 377837440,
+      "step": 2186
+    },
+    {
+      "epoch": 0.8376101110685561,
+      "grad_norm": 24.7277753271268,
+      "learning_rate": 5e-06,
+      "loss": 0.2387,
+      "num_input_tokens_seen": 378010264,
+      "step": 2187
+    },
+    {
+      "epoch": 0.8376101110685561,
+      "loss": 0.25835248827934265,
+      "loss_ce": 0.02092573791742325,
+      "loss_iou": 1.0623435974121094,
+      "loss_num": 0.2373046875,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 378010264,
+      "step": 2187
+    },
+    {
+      "epoch": 0.8379931060896209,
+      "grad_norm": 24.719697236965832,
+      "learning_rate": 5e-06,
+      "loss": 0.1879,
+      "num_input_tokens_seen": 378183416,
+      "step": 2188
+    },
+    {
+      "epoch": 0.8379931060896209,
+      "loss": 0.19210229814052582,
+      "loss_ce": 0.02059350535273552,
+      "loss_iou": 1.016941785812378,
+      "loss_num": 0.171875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 378183416,
+      "step": 2188
+    },
+    {
+      "epoch": 0.8383761011106856,
+      "grad_norm": 24.113227832932065,
+      "learning_rate": 5e-06,
+      "loss": 0.26,
+      "num_input_tokens_seen": 378356272,
+      "step": 2189
+    },
+    {
+      "epoch": 0.8383761011106856,
+      "loss": 0.3190307915210724,
+      "loss_ce": 0.020080596208572388,
+      "loss_iou": 1.0181410312652588,
+      "loss_num": 0.298828125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 378356272,
+      "step": 2189
+    },
+    {
+      "epoch": 0.8387590961317503,
+      "grad_norm": 24.62787157043042,
+      "learning_rate": 5e-06,
+      "loss": 0.2192,
+      "num_input_tokens_seen": 378529560,
+      "step": 2190
+    },
+    {
+      "epoch": 0.8387590961317503,
+      "loss": 0.1842830926179886,
+      "loss_ce": 0.021380264312028885,
+      "loss_iou": 1.012451410293579,
+      "loss_num": 0.1630859375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 378529560,
+      "step": 2190
+    },
+    {
+      "epoch": 0.839142091152815,
+      "grad_norm": 38.905196400234196,
+      "learning_rate": 5e-06,
+      "loss": 0.2227,
+      "num_input_tokens_seen": 378702368,
+      "step": 2191
+    },
+    {
+      "epoch": 0.839142091152815,
+      "loss": 0.2040984332561493,
+      "loss_ce": 0.022152645513415337,
+      "loss_iou": 1.0184812545776367,
+      "loss_num": 0.181640625,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 378702368,
+      "step": 2191
+    },
+    {
+      "epoch": 0.8395250861738798,
+      "grad_norm": 26.759518992974776,
+      "learning_rate": 5e-06,
+      "loss": 0.1746,
+      "num_input_tokens_seen": 378875176,
+      "step": 2192
+    },
+    {
+      "epoch": 0.8395250861738798,
+      "loss": 0.1554000973701477,
+      "loss_ce": 0.020146194845438004,
+      "loss_iou": 1.0175584554672241,
+      "loss_num": 0.134765625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 378875176,
+      "step": 2192
+    },
+    {
+      "epoch": 0.8399080811949444,
+      "grad_norm": 27.126001230207233,
+      "learning_rate": 5e-06,
+      "loss": 0.1853,
+      "num_input_tokens_seen": 379048104,
+      "step": 2193
+    },
+    {
+      "epoch": 0.8399080811949444,
+      "loss": 0.23117583990097046,
+      "loss_ce": 0.020116277039051056,
+      "loss_iou": 0.9910538792610168,
+      "loss_num": 0.2109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 379048104,
+      "step": 2193
+    },
+    {
+      "epoch": 0.8402910762160092,
+      "grad_norm": 20.014968760344406,
+      "learning_rate": 5e-06,
+      "loss": 0.2103,
+      "num_input_tokens_seen": 379220888,
+      "step": 2194
+    },
+    {
+      "epoch": 0.8402910762160092,
+      "loss": 0.20933499932289124,
+      "loss_ce": 0.021102581173181534,
+      "loss_iou": 1.0173618793487549,
+      "loss_num": 0.1884765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 379220888,
+      "step": 2194
+    },
+    {
+      "epoch": 0.8406740712370739,
+      "grad_norm": 31.80461049633313,
+      "learning_rate": 5e-06,
+      "loss": 0.1958,
+      "num_input_tokens_seen": 379394040,
+      "step": 2195
+    },
+    {
+      "epoch": 0.8406740712370739,
+      "loss": 0.18281367421150208,
+      "loss_ce": 0.021314652636647224,
+      "loss_iou": 1.0045641660690308,
+      "loss_num": 0.1611328125,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 379394040,
+      "step": 2195
+    },
+    {
+      "epoch": 0.8410570662581387,
+      "grad_norm": 29.73284418557474,
+      "learning_rate": 5e-06,
+      "loss": 0.208,
+      "num_input_tokens_seen": 379567176,
+      "step": 2196
+    },
+    {
+      "epoch": 0.8410570662581387,
+      "loss": 0.2742595374584198,
+      "loss_ce": 0.020475370809435844,
+      "loss_iou": 1.0036565065383911,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 379567176,
+      "step": 2196
+    },
+    {
+      "epoch": 0.8414400612792033,
+      "grad_norm": 21.963597026004095,
+      "learning_rate": 5e-06,
+      "loss": 0.2257,
+      "num_input_tokens_seen": 379740144,
+      "step": 2197
+    },
+    {
+      "epoch": 0.8414400612792033,
+      "loss": 0.25153568387031555,
+      "loss_ce": 0.020639687776565552,
+      "loss_iou": 1.0811197757720947,
+      "loss_num": 0.23046875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 379740144,
+      "step": 2197
+    },
+    {
+      "epoch": 0.8418230563002681,
+      "grad_norm": 25.053483094314313,
+      "learning_rate": 5e-06,
+      "loss": 0.2007,
+      "num_input_tokens_seen": 379913112,
+      "step": 2198
+    },
+    {
+      "epoch": 0.8418230563002681,
+      "loss": 0.222274512052536,
+      "loss_ce": 0.021346762776374817,
+      "loss_iou": 1.0332086086273193,
+      "loss_num": 0.201171875,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 379913112,
+      "step": 2198
+    },
+    {
+      "epoch": 0.8422060513213329,
+      "grad_norm": 33.65434983095775,
+      "learning_rate": 5e-06,
+      "loss": 0.2125,
+      "num_input_tokens_seen": 380085608,
+      "step": 2199
+    },
+    {
+      "epoch": 0.8422060513213329,
+      "loss": 0.17855077981948853,
+      "loss_ce": 0.018943851813673973,
+      "loss_iou": 1.009749174118042,
+      "loss_num": 0.1591796875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 380085608,
+      "step": 2199
+    },
+    {
+      "epoch": 0.8425890463423975,
+      "grad_norm": 24.803349461339845,
+      "learning_rate": 5e-06,
+      "loss": 0.2258,
+      "num_input_tokens_seen": 380258560,
+      "step": 2200
+    },
+    {
+      "epoch": 0.8425890463423975,
+      "loss": 0.27116209268569946,
+      "loss_ce": 0.020246557891368866,
+      "loss_iou": 1.1479721069335938,
+      "loss_num": 0.25,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 380258560,
+      "step": 2200
+    },
+    {
+      "epoch": 0.8429720413634623,
+      "grad_norm": 26.52332924299739,
+      "learning_rate": 5e-06,
+      "loss": 0.2184,
+      "num_input_tokens_seen": 380431568,
+      "step": 2201
+    },
+    {
+      "epoch": 0.8429720413634623,
+      "loss": 0.1903233826160431,
+      "loss_ce": 0.02119497023522854,
+      "loss_iou": 1.0225993394851685,
+      "loss_num": 0.1689453125,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 380431568,
+      "step": 2201
+    },
+    {
+      "epoch": 0.843355036384527,
+      "grad_norm": 19.306394250905814,
+      "learning_rate": 5e-06,
+      "loss": 0.1937,
+      "num_input_tokens_seen": 380604168,
+      "step": 2202
+    },
+    {
+      "epoch": 0.843355036384527,
+      "loss": 0.23338836431503296,
+      "loss_ce": 0.02245086431503296,
+      "loss_iou": 1.0257644653320312,
+      "loss_num": 0.2109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 380604168,
+      "step": 2202
+    },
+    {
+      "epoch": 0.8437380314055918,
+      "grad_norm": 33.28715800345254,
+      "learning_rate": 5e-06,
+      "loss": 0.1573,
+      "num_input_tokens_seen": 380776720,
+      "step": 2203
+    },
+    {
+      "epoch": 0.8437380314055918,
+      "loss": 0.15688501298427582,
+      "loss_ce": 0.020715579390525818,
+      "loss_iou": 1.0013618469238281,
+      "loss_num": 0.1357421875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 380776720,
+      "step": 2203
+    },
+    {
+      "epoch": 0.8441210264266564,
+      "grad_norm": 25.592195810622155,
+      "learning_rate": 5e-06,
+      "loss": 0.2243,
+      "num_input_tokens_seen": 380949600,
+      "step": 2204
+    },
+    {
+      "epoch": 0.8441210264266564,
+      "loss": 0.20534177124500275,
+      "loss_ce": 0.019001437351107597,
+      "loss_iou": 1.033729076385498,
+      "loss_num": 0.1865234375,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 380949600,
+      "step": 2204
+    },
+    {
+      "epoch": 0.8445040214477212,
+      "grad_norm": 17.2297216848919,
+      "learning_rate": 5e-06,
+      "loss": 0.2054,
+      "num_input_tokens_seen": 381122752,
+      "step": 2205
+    },
+    {
+      "epoch": 0.8445040214477212,
+      "loss": 0.18818581104278564,
+      "loss_ce": 0.018752217292785645,
+      "loss_iou": 1.0046367645263672,
+      "loss_num": 0.169921875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 381122752,
+      "step": 2205
+    },
+    {
+      "epoch": 0.844887016468786,
+      "grad_norm": 25.84307296924863,
+      "learning_rate": 5e-06,
+      "loss": 0.2294,
+      "num_input_tokens_seen": 381295568,
+      "step": 2206
+    },
+    {
+      "epoch": 0.844887016468786,
+      "loss": 0.2742776870727539,
+      "loss_ce": 0.02098177559673786,
+      "loss_iou": 1.0310285091400146,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 381295568,
+      "step": 2206
+    },
+    {
+      "epoch": 0.8452700114898506,
+      "grad_norm": 26.137643475532453,
+      "learning_rate": 5e-06,
+      "loss": 0.2608,
+      "num_input_tokens_seen": 381468568,
+      "step": 2207
+    },
+    {
+      "epoch": 0.8452700114898506,
+      "loss": 0.2337532341480255,
+      "loss_ce": 0.020130200311541557,
+      "loss_iou": 0.9583425521850586,
+      "loss_num": 0.2138671875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 381468568,
+      "step": 2207
+    },
+    {
+      "epoch": 0.8456530065109154,
+      "grad_norm": 22.739127563204576,
+      "learning_rate": 5e-06,
+      "loss": 0.1654,
+      "num_input_tokens_seen": 381641584,
+      "step": 2208
+    },
+    {
+      "epoch": 0.8456530065109154,
+      "loss": 0.19659042358398438,
+      "loss_ce": 0.019710546359419823,
+      "loss_iou": 1.021919846534729,
+      "loss_num": 0.1767578125,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 381641584,
+      "step": 2208
+    },
+    {
+      "epoch": 0.8460360015319801,
+      "grad_norm": 24.891123378386958,
+      "learning_rate": 5e-06,
+      "loss": 0.2735,
+      "num_input_tokens_seen": 381814368,
+      "step": 2209
+    },
+    {
+      "epoch": 0.8460360015319801,
+      "loss": 0.2113550901412964,
+      "loss_ce": 0.019826766103506088,
+      "loss_iou": 1.0132403373718262,
+      "loss_num": 0.19140625,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 381814368,
+      "step": 2209
+    },
+    {
+      "epoch": 0.8464189965530448,
+      "grad_norm": 28.400589982922263,
+      "learning_rate": 5e-06,
+      "loss": 0.1942,
+      "num_input_tokens_seen": 381987128,
+      "step": 2210
+    },
+    {
+      "epoch": 0.8464189965530448,
+      "loss": 0.18612128496170044,
+      "loss_ce": 0.020105667412281036,
+      "loss_iou": 1.0098161697387695,
+      "loss_num": 0.166015625,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 381987128,
+      "step": 2210
+    },
+    {
+      "epoch": 0.8468019915741095,
+      "grad_norm": 26.111639458418296,
+      "learning_rate": 5e-06,
+      "loss": 0.2345,
+      "num_input_tokens_seen": 382159824,
+      "step": 2211
+    },
+    {
+      "epoch": 0.8468019915741095,
+      "loss": 0.25814738869667053,
+      "loss_ce": 0.020049242302775383,
+      "loss_iou": 1.0347554683685303,
+      "loss_num": 0.23828125,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 382159824,
+      "step": 2211
+    },
+    {
+      "epoch": 0.8471849865951743,
+      "grad_norm": 26.875641179779645,
+      "learning_rate": 5e-06,
+      "loss": 0.2024,
+      "num_input_tokens_seen": 382332736,
+      "step": 2212
+    },
+    {
+      "epoch": 0.8471849865951743,
+      "loss": 0.20974765717983246,
+      "loss_ce": 0.02163730189204216,
+      "loss_iou": 1.0757482051849365,
+      "loss_num": 0.1884765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 382332736,
+      "step": 2212
+    },
+    {
+      "epoch": 0.847567981616239,
+      "grad_norm": 32.48165760489675,
+      "learning_rate": 5e-06,
+      "loss": 0.2173,
+      "num_input_tokens_seen": 382505648,
+      "step": 2213
+    },
+    {
+      "epoch": 0.847567981616239,
+      "loss": 0.23111853003501892,
+      "loss_ce": 0.020425166934728622,
+      "loss_iou": 1.0130873918533325,
+      "loss_num": 0.2109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 382505648,
+      "step": 2213
+    },
+    {
+      "epoch": 0.8479509766373037,
+      "grad_norm": 28.07529780996336,
+      "learning_rate": 5e-06,
+      "loss": 0.1389,
+      "num_input_tokens_seen": 382678792,
+      "step": 2214
+    },
+    {
+      "epoch": 0.8479509766373037,
+      "loss": 0.1350688338279724,
+      "loss_ce": 0.02206224389374256,
+      "loss_iou": 1.0005462169647217,
+      "loss_num": 0.11279296875,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 382678792,
+      "step": 2214
+    },
+    {
+      "epoch": 0.8483339716583684,
+      "grad_norm": 34.326411712989156,
+      "learning_rate": 5e-06,
+      "loss": 0.1991,
+      "num_input_tokens_seen": 382851968,
+      "step": 2215
+    },
+    {
+      "epoch": 0.8483339716583684,
+      "loss": 0.23377156257629395,
+      "loss_ce": 0.021491285413503647,
+      "loss_iou": 1.0419487953186035,
+      "loss_num": 0.2119140625,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 382851968,
+      "step": 2215
+    },
+    {
+      "epoch": 0.8487169666794332,
+      "grad_norm": 43.532088051136896,
+      "learning_rate": 5e-06,
+      "loss": 0.2152,
+      "num_input_tokens_seen": 383024904,
+      "step": 2216
+    },
+    {
+      "epoch": 0.8487169666794332,
+      "loss": 0.27027279138565063,
+      "loss_ce": 0.023446639999747276,
+      "loss_iou": 1.0436408519744873,
+      "loss_num": 0.2470703125,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 383024904,
+      "step": 2216
+    },
+    {
+      "epoch": 0.8490999617004978,
+      "grad_norm": 36.55397794869057,
+      "learning_rate": 5e-06,
+      "loss": 0.2482,
+      "num_input_tokens_seen": 383197920,
+      "step": 2217
+    },
+    {
+      "epoch": 0.8490999617004978,
+      "loss": 0.2088879495859146,
+      "loss_ce": 0.021448995918035507,
+      "loss_iou": 1.0622910261154175,
+      "loss_num": 0.1875,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 383197920,
+      "step": 2217
+    },
+    {
+      "epoch": 0.8494829567215626,
+      "grad_norm": 19.570506990689445,
+      "learning_rate": 5e-06,
+      "loss": 0.1712,
+      "num_input_tokens_seen": 383370320,
+      "step": 2218
+    },
+    {
+      "epoch": 0.8494829567215626,
+      "loss": 0.12578538060188293,
+      "loss_ce": 0.02117113023996353,
+      "loss_iou": 1.0086634159088135,
+      "loss_num": 0.1044921875,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 383370320,
+      "step": 2218
+    },
+    {
+      "epoch": 0.8498659517426274,
+      "grad_norm": 20.79907884711053,
+      "learning_rate": 5e-06,
+      "loss": 0.1775,
+      "num_input_tokens_seen": 383543840,
+      "step": 2219
+    },
+    {
+      "epoch": 0.8498659517426274,
+      "loss": 0.14958077669143677,
+      "loss_ce": 0.021315395832061768,
+      "loss_iou": 1.0119179487228394,
+      "loss_num": 0.1279296875,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 383543840,
+      "step": 2219
+    },
+    {
+      "epoch": 0.8502489467636921,
+      "grad_norm": 41.55804129473687,
+      "learning_rate": 5e-06,
+      "loss": 0.1978,
+      "num_input_tokens_seen": 383716784,
+      "step": 2220
+    },
+    {
+      "epoch": 0.8502489467636921,
+      "loss": 0.20395150780677795,
+      "loss_ce": 0.021029146388173103,
+      "loss_iou": 1.0557193756103516,
+      "loss_num": 0.1826171875,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 383716784,
+      "step": 2220
+    },
+    {
+      "epoch": 0.8506319417847568,
+      "grad_norm": 34.48967079210651,
+      "learning_rate": 5e-06,
+      "loss": 0.2426,
+      "num_input_tokens_seen": 383889664,
+      "step": 2221
+    },
+    {
+      "epoch": 0.8506319417847568,
+      "loss": 0.24308425188064575,
+      "loss_ce": 0.01981765404343605,
+      "loss_iou": 1.001485824584961,
+      "loss_num": 0.2236328125,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 383889664,
+      "step": 2221
+    },
+    {
+      "epoch": 0.8510149368058215,
+      "grad_norm": 24.43551228595156,
+      "learning_rate": 5e-06,
+      "loss": 0.1599,
+      "num_input_tokens_seen": 384062600,
+      "step": 2222
+    },
+    {
+      "epoch": 0.8510149368058215,
+      "loss": 0.17191261053085327,
+      "loss_ce": 0.02036231942474842,
+      "loss_iou": 1.019148349761963,
+      "loss_num": 0.1513671875,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 384062600,
+      "step": 2222
+    },
+    {
+      "epoch": 0.8513979318268863,
+      "grad_norm": 48.541317174882316,
+      "learning_rate": 5e-06,
+      "loss": 0.1604,
+      "num_input_tokens_seen": 384236016,
+      "step": 2223
+    },
+    {
+      "epoch": 0.8513979318268863,
+      "loss": 0.14895141124725342,
+      "loss_ce": 0.021021710708737373,
+      "loss_iou": 1.0104756355285645,
+      "loss_num": 0.1279296875,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 384236016,
+      "step": 2223
+    },
+    {
+      "epoch": 0.8517809268479509,
+      "grad_norm": 32.418865843011815,
+      "learning_rate": 5e-06,
+      "loss": 0.2399,
+      "num_input_tokens_seen": 384409240,
+      "step": 2224
+    },
+    {
+      "epoch": 0.8517809268479509,
+      "loss": 0.2681843638420105,
+      "loss_ce": 0.020930945873260498,
+      "loss_iou": 1.0598926544189453,
+      "loss_num": 0.2470703125,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 384409240,
+      "step": 2224
+    },
+    {
+      "epoch": 0.8521639218690157,
+      "grad_norm": 30.691232291627884,
+      "learning_rate": 5e-06,
+      "loss": 0.1607,
+      "num_input_tokens_seen": 384581896,
+      "step": 2225
+    },
+    {
+      "epoch": 0.8521639218690157,
+      "loss": 0.15713070333003998,
+      "loss_ce": 0.021480072289705276,
+      "loss_iou": 1.0184576511383057,
+      "loss_num": 0.1357421875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 384581896,
+      "step": 2225
+    },
+    {
+      "epoch": 0.8525469168900804,
+      "grad_norm": 26.887317307608107,
+      "learning_rate": 5e-06,
+      "loss": 0.1923,
+      "num_input_tokens_seen": 384754616,
+      "step": 2226
+    },
+    {
+      "epoch": 0.8525469168900804,
+      "loss": 0.13293921947479248,
+      "loss_ce": 0.021061770617961884,
+      "loss_iou": 1.002497673034668,
+      "loss_num": 0.11181640625,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 384754616,
+      "step": 2226
+    },
+    {
+      "epoch": 0.8529299119111452,
+      "grad_norm": 35.72703494227469,
+      "learning_rate": 5e-06,
+      "loss": 0.237,
+      "num_input_tokens_seen": 384927320,
+      "step": 2227
+    },
+    {
+      "epoch": 0.8529299119111452,
+      "loss": 0.2753600478172302,
+      "loss_ce": 0.019988957792520523,
+      "loss_iou": 1.6816809177398682,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 384927320,
+      "step": 2227
+    },
+    {
+      "epoch": 0.8533129069322098,
+      "grad_norm": 30.365398214819066,
+      "learning_rate": 5e-06,
+      "loss": 0.1783,
+      "num_input_tokens_seen": 385100520,
+      "step": 2228
+    },
+    {
+      "epoch": 0.8533129069322098,
+      "loss": 0.19887545704841614,
+      "loss_ce": 0.020225554704666138,
+      "loss_iou": 1.050445318222046,
+      "loss_num": 0.1787109375,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 385100520,
+      "step": 2228
+    },
+    {
+      "epoch": 0.8536959019532746,
+      "grad_norm": 22.968221588790158,
+      "learning_rate": 5e-06,
+      "loss": 0.2021,
+      "num_input_tokens_seen": 385273720,
+      "step": 2229
+    },
+    {
+      "epoch": 0.8536959019532746,
+      "loss": 0.24257753789424896,
+      "loss_ce": 0.020287496969103813,
+      "loss_iou": 1.0931062698364258,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 385273720,
+      "step": 2229
+    },
+    {
+      "epoch": 0.8540788969743394,
+      "grad_norm": 25.78179269820435,
+      "learning_rate": 5e-06,
+      "loss": 0.2538,
+      "num_input_tokens_seen": 385446512,
+      "step": 2230
+    },
+    {
+      "epoch": 0.8540788969743394,
+      "loss": 0.1526906043291092,
+      "loss_ce": 0.02210589125752449,
+      "loss_iou": 1.0236563682556152,
+      "loss_num": 0.130859375,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 385446512,
+      "step": 2230
+    },
+    {
+      "epoch": 0.854461891995404,
+      "grad_norm": 31.826088139453805,
+      "learning_rate": 5e-06,
+      "loss": 0.1534,
+      "num_input_tokens_seen": 385619712,
+      "step": 2231
+    },
+    {
+      "epoch": 0.854461891995404,
+      "loss": 0.13914582133293152,
+      "loss_ce": 0.022385576739907265,
+      "loss_iou": 0.9842702746391296,
+      "loss_num": 0.11669921875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 385619712,
+      "step": 2231
+    },
+    {
+      "epoch": 0.8548448870164688,
+      "grad_norm": 31.061881249812,
+      "learning_rate": 5e-06,
+      "loss": 0.1978,
+      "num_input_tokens_seen": 385792480,
+      "step": 2232
+    },
+    {
+      "epoch": 0.8548448870164688,
+      "loss": 0.21213898062705994,
+      "loss_ce": 0.021831370890140533,
+      "loss_iou": 1.017231822013855,
+      "loss_num": 0.1904296875,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 385792480,
+      "step": 2232
+    },
+    {
+      "epoch": 0.8552278820375335,
+      "grad_norm": 33.208733578507065,
+      "learning_rate": 5e-06,
+      "loss": 0.1814,
+      "num_input_tokens_seen": 385965608,
+      "step": 2233
+    },
+    {
+      "epoch": 0.8552278820375335,
+      "loss": 0.1736002117395401,
+      "loss_ce": 0.02168370969593525,
+      "loss_iou": 1.0410422086715698,
+      "loss_num": 0.15234375,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 385965608,
+      "step": 2233
+    },
+    {
+      "epoch": 0.8556108770585983,
+      "grad_norm": 24.36039320694214,
+      "learning_rate": 5e-06,
+      "loss": 0.169,
+      "num_input_tokens_seen": 386138736,
+      "step": 2234
+    },
+    {
+      "epoch": 0.8556108770585983,
+      "loss": 0.16091981530189514,
+      "loss_ce": 0.02218690887093544,
+      "loss_iou": 1.0135810375213623,
+      "loss_num": 0.138671875,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 386138736,
+      "step": 2234
+    },
+    {
+      "epoch": 0.8559938720796629,
+      "grad_norm": 26.630559602629717,
+      "learning_rate": 5e-06,
+      "loss": 0.1939,
+      "num_input_tokens_seen": 386311616,
+      "step": 2235
+    },
+    {
+      "epoch": 0.8559938720796629,
+      "loss": 0.20099741220474243,
+      "loss_ce": 0.019295761361718178,
+      "loss_iou": 1.00544011592865,
+      "loss_num": 0.181640625,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 386311616,
+      "step": 2235
+    },
+    {
+      "epoch": 0.8563768671007277,
+      "grad_norm": 29.432108707921554,
+      "learning_rate": 5e-06,
+      "loss": 0.1899,
+      "num_input_tokens_seen": 386485216,
+      "step": 2236
+    },
+    {
+      "epoch": 0.8563768671007277,
+      "loss": 0.2291969358921051,
+      "loss_ce": 0.022653959691524506,
+      "loss_iou": 1.0504539012908936,
+      "loss_num": 0.20703125,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 386485216,
+      "step": 2236
+    },
+    {
+      "epoch": 0.8567598621217924,
+      "grad_norm": 32.27350031667615,
+      "learning_rate": 5e-06,
+      "loss": 0.1902,
+      "num_input_tokens_seen": 386658064,
+      "step": 2237
+    },
+    {
+      "epoch": 0.8567598621217924,
+      "loss": 0.18089860677719116,
+      "loss_ce": 0.02165788784623146,
+      "loss_iou": 1.0156598091125488,
+      "loss_num": 0.1591796875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 386658064,
+      "step": 2237
+    },
+    {
+      "epoch": 0.8571428571428571,
+      "grad_norm": 16.848194771374665,
+      "learning_rate": 5e-06,
+      "loss": 0.166,
+      "num_input_tokens_seen": 386830816,
+      "step": 2238
+    },
+    {
+      "epoch": 0.8571428571428571,
+      "loss": 0.1765880137681961,
+      "loss_ce": 0.020704224705696106,
+      "loss_iou": 1.0056991577148438,
+      "loss_num": 0.15625,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 386830816,
+      "step": 2238
+    },
+    {
+      "epoch": 0.8575258521639219,
+      "grad_norm": 38.74609977662123,
+      "learning_rate": 5e-06,
+      "loss": 0.1959,
+      "num_input_tokens_seen": 387003656,
+      "step": 2239
+    },
+    {
+      "epoch": 0.8575258521639219,
+      "loss": 0.16399705410003662,
+      "loss_ce": 0.020076151937246323,
+      "loss_iou": 1.002182960510254,
+      "loss_num": 0.1435546875,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 387003656,
+      "step": 2239
+    },
+    {
+      "epoch": 0.8579088471849866,
+      "grad_norm": 30.62404269175233,
+      "learning_rate": 5e-06,
+      "loss": 0.2349,
+      "num_input_tokens_seen": 387176648,
+      "step": 2240
+    },
+    {
+      "epoch": 0.8579088471849866,
+      "loss": 0.23139256238937378,
+      "loss_ce": 0.021614745259284973,
+      "loss_iou": 1.033926248550415,
+      "loss_num": 0.2099609375,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 387176648,
+      "step": 2240
+    },
+    {
+      "epoch": 0.8582918422060514,
+      "grad_norm": 37.479556106391456,
+      "learning_rate": 5e-06,
+      "loss": 0.2319,
+      "num_input_tokens_seen": 387349160,
+      "step": 2241
+    },
+    {
+      "epoch": 0.8582918422060514,
+      "loss": 0.2569850981235504,
+      "loss_ce": 0.020168686285614967,
+      "loss_iou": 1.0369092226028442,
+      "loss_num": 0.236328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 387349160,
+      "step": 2241
+    },
+    {
+      "epoch": 0.858674837227116,
+      "grad_norm": 25.513422322811078,
+      "learning_rate": 5e-06,
+      "loss": 0.1938,
+      "num_input_tokens_seen": 387522208,
+      "step": 2242
+    },
+    {
+      "epoch": 0.858674837227116,
+      "loss": 0.2113761603832245,
+      "loss_ce": 0.020214062184095383,
+      "loss_iou": 1.0536246299743652,
+      "loss_num": 0.19140625,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 387522208,
+      "step": 2242
+    },
+    {
+      "epoch": 0.8590578322481808,
+      "grad_norm": 36.624122585785116,
+      "learning_rate": 5e-06,
+      "loss": 0.2894,
+      "num_input_tokens_seen": 387691656,
+      "step": 2243
+    },
+    {
+      "epoch": 0.8590578322481808,
+      "loss": 0.3449817895889282,
+      "loss_ce": 0.020396839827299118,
+      "loss_iou": 1.0083636045455933,
+      "loss_num": 0.32421875,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 387691656,
+      "step": 2243
+    },
+    {
+      "epoch": 0.8594408272692455,
+      "grad_norm": 26.62656131057298,
+      "learning_rate": 5e-06,
+      "loss": 0.2166,
+      "num_input_tokens_seen": 387864496,
+      "step": 2244
+    },
+    {
+      "epoch": 0.8594408272692455,
+      "loss": 0.20909324288368225,
+      "loss_ce": 0.021288076415657997,
+      "loss_iou": 1.0143465995788574,
+      "loss_num": 0.1875,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 387864496,
+      "step": 2244
+    },
+    {
+      "epoch": 0.8598238222903102,
+      "grad_norm": 28.81258415694795,
+      "learning_rate": 5e-06,
+      "loss": 0.2048,
+      "num_input_tokens_seen": 388037376,
+      "step": 2245
+    },
+    {
+      "epoch": 0.8598238222903102,
+      "loss": 0.22831951081752777,
+      "loss_ce": 0.02000652253627777,
+      "loss_iou": 1.0391998291015625,
+      "loss_num": 0.2080078125,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 388037376,
+      "step": 2245
+    },
+    {
+      "epoch": 0.8602068173113749,
+      "grad_norm": 26.798224782980558,
+      "learning_rate": 5e-06,
+      "loss": 0.1991,
+      "num_input_tokens_seen": 388210568,
+      "step": 2246
+    },
+    {
+      "epoch": 0.8602068173113749,
+      "loss": 0.19132068753242493,
+      "loss_ce": 0.020971566438674927,
+      "loss_iou": 1.0045247077941895,
+      "loss_num": 0.169921875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 388210568,
+      "step": 2246
+    },
+    {
+      "epoch": 0.8605898123324397,
+      "grad_norm": 29.27226331767748,
+      "learning_rate": 5e-06,
+      "loss": 0.2494,
+      "num_input_tokens_seen": 388383984,
+      "step": 2247
+    },
+    {
+      "epoch": 0.8605898123324397,
+      "loss": 0.2672705054283142,
+      "loss_ce": 0.020383302122354507,
+      "loss_iou": 1.04952871799469,
+      "loss_num": 0.2470703125,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 388383984,
+      "step": 2247
+    },
+    {
+      "epoch": 0.8609728073535045,
+      "grad_norm": 29.59762985020245,
+      "learning_rate": 5e-06,
+      "loss": 0.2152,
+      "num_input_tokens_seen": 388556784,
+      "step": 2248
+    },
+    {
+      "epoch": 0.8609728073535045,
+      "loss": 0.21293094754219055,
+      "loss_ce": 0.020059864968061447,
+      "loss_iou": 1.0007498264312744,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 388556784,
+      "step": 2248
+    },
+    {
+      "epoch": 0.8613558023745691,
+      "grad_norm": 34.46333613595097,
+      "learning_rate": 5e-06,
+      "loss": 0.278,
+      "num_input_tokens_seen": 388730048,
+      "step": 2249
+    },
+    {
+      "epoch": 0.8613558023745691,
+      "loss": 0.3135603964328766,
+      "loss_ce": 0.02254476770758629,
+      "loss_iou": 1.1369743347167969,
+      "loss_num": 0.291015625,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 388730048,
+      "step": 2249
+    },
+    {
+      "epoch": 0.8617387973956339,
+      "grad_norm": 18.148969965182555,
+      "learning_rate": 5e-06,
+      "loss": 0.1675,
+      "num_input_tokens_seen": 388902864,
+      "step": 2250
+    },
+    {
+      "epoch": 0.8617387973956339,
+      "eval_websight_new_CIoU": 0.925597608089447,
+      "eval_websight_new_GIoU": 0.925267904996872,
+      "eval_websight_new_IoU": 0.9258756935596466,
+      "eval_websight_new_MAE_all": 0.007574299583211541,
+      "eval_websight_new_MAE_h": 0.006647913018241525,
+      "eval_websight_new_MAE_w": 0.007652605883777142,
+      "eval_websight_new_MAE_x": 0.007741168374195695,
+      "eval_websight_new_MAE_y": 0.008255511056631804,
+      "eval_websight_new_NUM_probability": 8.322440044139512e-05,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.11716654151678085,
+      "eval_websight_new_loss_ce": 0.041889848187565804,
+      "eval_websight_new_loss_iou": 1.0004693865776062,
+      "eval_websight_new_loss_num": 0.07281494140625,
+      "eval_websight_new_loss_xval": 0.07281494140625,
+      "eval_websight_new_runtime": 56.5385,
+      "eval_websight_new_samples_per_second": 0.884,
+      "eval_websight_new_steps_per_second": 0.035,
+      "num_input_tokens_seen": 388902864,
+      "step": 2250
+    },
+    {
+      "epoch": 0.8617387973956339,
+      "eval_seeclick_CIoU": 0.6681079566478729,
+      "eval_seeclick_GIoU": 0.6744631826877594,
+      "eval_seeclick_IoU": 0.6976601779460907,
+      "eval_seeclick_MAE_all": 0.055332526564598083,
+      "eval_seeclick_MAE_h": 0.03067066613584757,
+      "eval_seeclick_MAE_w": 0.08008594438433647,
+      "eval_seeclick_MAE_x": 0.07529773563146591,
+      "eval_seeclick_MAE_y": 0.03527577221393585,
+      "eval_seeclick_NUM_probability": 0.00011169226127094589,
+      "eval_seeclick_inside_bbox": 0.9097222089767456,
+      "eval_seeclick_loss": 0.47150173783302307,
+      "eval_seeclick_loss_ce": 0.03271147049963474,
+      "eval_seeclick_loss_iou": 1.140836775302887,
+      "eval_seeclick_loss_num": 0.438751220703125,
+      "eval_seeclick_loss_xval": 0.438751220703125,
+      "eval_seeclick_runtime": 85.9892,
+      "eval_seeclick_samples_per_second": 0.581,
+      "eval_seeclick_steps_per_second": 0.023,
+      "num_input_tokens_seen": 388902864,
+      "step": 2250
+    },
+    {
+      "epoch": 0.8617387973956339,
+      "eval_icons_CIoU": 0.8652712106704712,
+      "eval_icons_GIoU": 0.8643836081027985,
+      "eval_icons_IoU": 0.8742239475250244,
+      "eval_icons_MAE_all": 0.020118060521781445,
+      "eval_icons_MAE_h": 0.01535157673060894,
+      "eval_icons_MAE_w": 0.023680626414716244,
+      "eval_icons_MAE_x": 0.024736194871366024,
+      "eval_icons_MAE_y": 0.016703845001757145,
+      "eval_icons_NUM_probability": 7.394958811346442e-05,
+      "eval_icons_inside_bbox": 0.984375,
+      "eval_icons_loss": 0.17382636666297913,
+      "eval_icons_loss_ce": 0.02849969081580639,
+      "eval_icons_loss_iou": 1.042946219444275,
+      "eval_icons_loss_num": 0.13238525390625,
+      "eval_icons_loss_xval": 0.13238525390625,
+      "eval_icons_runtime": 82.6232,
+      "eval_icons_samples_per_second": 0.605,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 388902864,
+      "step": 2250
+    },
+    {
+      "epoch": 0.8617387973956339,
+      "loss": 0.1544845551252365,
+      "loss_ce": 0.02850799635052681,
+      "loss_iou": 1.0282611846923828,
+      "loss_num": 0.1259765625,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 388902864,
+      "step": 2250
+    },
+    {
+      "epoch": 0.8621217924166986,
+      "grad_norm": 40.262798507762206,
+      "learning_rate": 5e-06,
+      "loss": 0.2046,
+      "num_input_tokens_seen": 389075768,
+      "step": 2251
+    },
+    {
+      "epoch": 0.8621217924166986,
+      "loss": 0.24195796251296997,
+      "loss_ce": 0.020766548812389374,
+      "loss_iou": 1.143406867980957,
+      "loss_num": 0.220703125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 389075768,
+      "step": 2251
+    },
+    {
+      "epoch": 0.8625047874377633,
+      "grad_norm": 23.771191181194094,
+      "learning_rate": 5e-06,
+      "loss": 0.2024,
+      "num_input_tokens_seen": 389248752,
+      "step": 2252
+    },
+    {
+      "epoch": 0.8625047874377633,
+      "loss": 0.1764417290687561,
+      "loss_ce": 0.021778643131256104,
+      "loss_iou": 1.0457133054733276,
+      "loss_num": 0.154296875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 389248752,
+      "step": 2252
+    },
+    {
+      "epoch": 0.862887782458828,
+      "grad_norm": 30.254986516485246,
+      "learning_rate": 5e-06,
+      "loss": 0.2046,
+      "num_input_tokens_seen": 389421704,
+      "step": 2253
+    },
+    {
+      "epoch": 0.862887782458828,
+      "loss": 0.18306171894073486,
+      "loss_ce": 0.01991475746035576,
+      "loss_iou": 1.9940533638000488,
+      "loss_num": 0.1630859375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 389421704,
+      "step": 2253
+    },
+    {
+      "epoch": 0.8632707774798928,
+      "grad_norm": 21.778127440440873,
+      "learning_rate": 5e-06,
+      "loss": 0.1694,
+      "num_input_tokens_seen": 389594840,
+      "step": 2254
+    },
+    {
+      "epoch": 0.8632707774798928,
+      "loss": 0.17349952459335327,
+      "loss_ce": 0.020484384149312973,
+      "loss_iou": 1.0874111652374268,
+      "loss_num": 0.1533203125,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 389594840,
+      "step": 2254
+    },
+    {
+      "epoch": 0.8636537725009575,
+      "grad_norm": 23.969587805732107,
+      "learning_rate": 5e-06,
+      "loss": 0.1503,
+      "num_input_tokens_seen": 389768176,
+      "step": 2255
+    },
+    {
+      "epoch": 0.8636537725009575,
+      "loss": 0.1377573311328888,
+      "loss_ce": 0.022095710039138794,
+      "loss_iou": 1.0036972761154175,
+      "loss_num": 0.11572265625,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 389768176,
+      "step": 2255
+    },
+    {
+      "epoch": 0.8640367675220222,
+      "grad_norm": 28.174938708153093,
+      "learning_rate": 5e-06,
+      "loss": 0.2382,
+      "num_input_tokens_seen": 389941656,
+      "step": 2256
+    },
+    {
+      "epoch": 0.8640367675220222,
+      "loss": 0.21343086659908295,
+      "loss_ce": 0.02281806617975235,
+      "loss_iou": 1.106879711151123,
+      "loss_num": 0.1904296875,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 389941656,
+      "step": 2256
+    },
+    {
+      "epoch": 0.864419762543087,
+      "grad_norm": 26.522715403720657,
+      "learning_rate": 5e-06,
+      "loss": 0.2505,
+      "num_input_tokens_seen": 390114520,
+      "step": 2257
+    },
+    {
+      "epoch": 0.864419762543087,
+      "loss": 0.2640177607536316,
+      "loss_ce": 0.020853698253631592,
+      "loss_iou": 1.0961253643035889,
+      "loss_num": 0.2431640625,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 390114520,
+      "step": 2257
+    },
+    {
+      "epoch": 0.8648027575641517,
+      "grad_norm": 28.008023588531298,
+      "learning_rate": 5e-06,
+      "loss": 0.1965,
+      "num_input_tokens_seen": 390285824,
+      "step": 2258
+    },
+    {
+      "epoch": 0.8648027575641517,
+      "loss": 0.2037627100944519,
+      "loss_ce": 0.021389653906226158,
+      "loss_iou": 0.9831381440162659,
+      "loss_num": 0.1826171875,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 390285824,
+      "step": 2258
+    },
+    {
+      "epoch": 0.8651857525852163,
+      "grad_norm": 34.32974022877522,
+      "learning_rate": 5e-06,
+      "loss": 0.2206,
+      "num_input_tokens_seen": 390458624,
+      "step": 2259
+    },
+    {
+      "epoch": 0.8651857525852163,
+      "loss": 0.2119344174861908,
+      "loss_ce": 0.02107749506831169,
+      "loss_iou": 1.0079381465911865,
+      "loss_num": 0.1904296875,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 390458624,
+      "step": 2259
+    },
+    {
+      "epoch": 0.8655687476062811,
+      "grad_norm": 28.77967027761165,
+      "learning_rate": 5e-06,
+      "loss": 0.2051,
+      "num_input_tokens_seen": 390631360,
+      "step": 2260
+    },
+    {
+      "epoch": 0.8655687476062811,
+      "loss": 0.19896094501018524,
+      "loss_ce": 0.022081052884459496,
+      "loss_iou": 1.0542672872543335,
+      "loss_num": 0.1767578125,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 390631360,
+      "step": 2260
+    },
+    {
+      "epoch": 0.8659517426273459,
+      "grad_norm": 23.3673415650207,
+      "learning_rate": 5e-06,
+      "loss": 0.2002,
+      "num_input_tokens_seen": 390804368,
+      "step": 2261
+    },
+    {
+      "epoch": 0.8659517426273459,
+      "loss": 0.20097553730010986,
+      "loss_ce": 0.021043909713625908,
+      "loss_iou": 1.0051054954528809,
+      "loss_num": 0.1796875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 390804368,
+      "step": 2261
+    },
+    {
+      "epoch": 0.8663347376484106,
+      "grad_norm": 28.53559371537568,
+      "learning_rate": 5e-06,
+      "loss": 0.254,
+      "num_input_tokens_seen": 390977120,
+      "step": 2262
+    },
+    {
+      "epoch": 0.8663347376484106,
+      "loss": 0.2656322419643402,
+      "loss_ce": 0.02045402117073536,
+      "loss_iou": 1.1557904481887817,
+      "loss_num": 0.2451171875,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 390977120,
+      "step": 2262
+    },
+    {
+      "epoch": 0.8667177326694753,
+      "grad_norm": 36.17334970556343,
+      "learning_rate": 5e-06,
+      "loss": 0.2581,
+      "num_input_tokens_seen": 391149976,
+      "step": 2263
+    },
+    {
+      "epoch": 0.8667177326694753,
+      "loss": 0.25398796796798706,
+      "loss_ce": 0.02156611531972885,
+      "loss_iou": 1.130397915840149,
+      "loss_num": 0.232421875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 391149976,
+      "step": 2263
+    },
+    {
+      "epoch": 0.86710072769054,
+      "grad_norm": 31.429505011009915,
+      "learning_rate": 5e-06,
+      "loss": 0.2676,
+      "num_input_tokens_seen": 391323128,
+      "step": 2264
+    },
+    {
+      "epoch": 0.86710072769054,
+      "loss": 0.27364373207092285,
+      "loss_ce": 0.021263331174850464,
+      "loss_iou": 1.0309957265853882,
+      "loss_num": 0.251953125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 391323128,
+      "step": 2264
+    },
+    {
+      "epoch": 0.8674837227116048,
+      "grad_norm": 31.740600094807085,
+      "learning_rate": 5e-06,
+      "loss": 0.209,
+      "num_input_tokens_seen": 391495984,
+      "step": 2265
+    },
+    {
+      "epoch": 0.8674837227116048,
+      "loss": 0.23950469493865967,
+      "loss_ce": 0.020571591332554817,
+      "loss_iou": 1.0285762548446655,
+      "loss_num": 0.21875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 391495984,
+      "step": 2265
+    },
+    {
+      "epoch": 0.8678667177326694,
+      "grad_norm": 26.633176792341132,
+      "learning_rate": 5e-06,
+      "loss": 0.232,
+      "num_input_tokens_seen": 391668984,
+      "step": 2266
+    },
+    {
+      "epoch": 0.8678667177326694,
+      "loss": 0.22047019004821777,
+      "loss_ce": 0.023204566910862923,
+      "loss_iou": 1.0347226858139038,
+      "loss_num": 0.197265625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 391668984,
+      "step": 2266
+    },
+    {
+      "epoch": 0.8682497127537342,
+      "grad_norm": 34.927207210440685,
+      "learning_rate": 5e-06,
+      "loss": 0.1771,
+      "num_input_tokens_seen": 391841904,
+      "step": 2267
+    },
+    {
+      "epoch": 0.8682497127537342,
+      "loss": 0.13843293488025665,
+      "loss_ce": 0.021123360842466354,
+      "loss_iou": 1.0038037300109863,
+      "loss_num": 0.1171875,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 391841904,
+      "step": 2267
+    },
+    {
+      "epoch": 0.868632707774799,
+      "grad_norm": 39.75350681767262,
+      "learning_rate": 5e-06,
+      "loss": 0.151,
+      "num_input_tokens_seen": 392014976,
+      "step": 2268
+    },
+    {
+      "epoch": 0.868632707774799,
+      "loss": 0.1553298830986023,
+      "loss_ce": 0.021418750286102295,
+      "loss_iou": 1.0097993612289429,
+      "loss_num": 0.1337890625,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 392014976,
+      "step": 2268
+    },
+    {
+      "epoch": 0.8690157027958637,
+      "grad_norm": 23.521416444063636,
+      "learning_rate": 5e-06,
+      "loss": 0.2839,
+      "num_input_tokens_seen": 392188224,
+      "step": 2269
+    },
+    {
+      "epoch": 0.8690157027958637,
+      "loss": 0.3249661326408386,
+      "loss_ce": 0.019973479211330414,
+      "loss_iou": 1.1880580186843872,
+      "loss_num": 0.3046875,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 392188224,
+      "step": 2269
+    },
+    {
+      "epoch": 0.8693986978169284,
+      "grad_norm": 24.155046461733335,
+      "learning_rate": 5e-06,
+      "loss": 0.1708,
+      "num_input_tokens_seen": 392361312,
+      "step": 2270
+    },
+    {
+      "epoch": 0.8693986978169284,
+      "loss": 0.12677741050720215,
+      "loss_ce": 0.021003492176532745,
+      "loss_iou": 1.0012356042861938,
+      "loss_num": 0.10595703125,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 392361312,
+      "step": 2270
+    },
+    {
+      "epoch": 0.8697816928379931,
+      "grad_norm": 30.501720723333715,
+      "learning_rate": 5e-06,
+      "loss": 0.223,
+      "num_input_tokens_seen": 392534216,
+      "step": 2271
+    },
+    {
+      "epoch": 0.8697816928379931,
+      "loss": 0.20977681875228882,
+      "loss_ce": 0.02099508047103882,
+      "loss_iou": 1.0238680839538574,
+      "loss_num": 0.1884765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 392534216,
+      "step": 2271
+    },
+    {
+      "epoch": 0.8701646878590579,
+      "grad_norm": 28.394669787689313,
+      "learning_rate": 5e-06,
+      "loss": 0.2039,
+      "num_input_tokens_seen": 392706952,
+      "step": 2272
+    },
+    {
+      "epoch": 0.8701646878590579,
+      "loss": 0.2121831327676773,
+      "loss_ce": 0.021082064136862755,
+      "loss_iou": 1.0486198663711548,
+      "loss_num": 0.19140625,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 392706952,
+      "step": 2272
+    },
+    {
+      "epoch": 0.8705476828801225,
+      "grad_norm": 30.8297616515793,
+      "learning_rate": 5e-06,
+      "loss": 0.2058,
+      "num_input_tokens_seen": 392880088,
+      "step": 2273
+    },
+    {
+      "epoch": 0.8705476828801225,
+      "loss": 0.20134717226028442,
+      "loss_ce": 0.02117139846086502,
+      "loss_iou": 1.0266752243041992,
+      "loss_num": 0.1796875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 392880088,
+      "step": 2273
+    },
+    {
+      "epoch": 0.8709306779011873,
+      "grad_norm": 35.38446460366821,
+      "learning_rate": 5e-06,
+      "loss": 0.2135,
+      "num_input_tokens_seen": 393053272,
+      "step": 2274
+    },
+    {
+      "epoch": 0.8709306779011873,
+      "loss": 0.24780648946762085,
+      "loss_ce": 0.021793287247419357,
+      "loss_iou": 1.1017720699310303,
+      "loss_num": 0.2255859375,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 393053272,
+      "step": 2274
+    },
+    {
+      "epoch": 0.871313672922252,
+      "grad_norm": 61.66856149553012,
+      "learning_rate": 5e-06,
+      "loss": 0.2248,
+      "num_input_tokens_seen": 393226288,
+      "step": 2275
+    },
+    {
+      "epoch": 0.871313672922252,
+      "loss": 0.2328328788280487,
+      "loss_ce": 0.02226157858967781,
+      "loss_iou": 1.0248265266418457,
+      "loss_num": 0.2109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 393226288,
+      "step": 2275
+    },
+    {
+      "epoch": 0.8716966679433167,
+      "grad_norm": 35.51432763180763,
+      "learning_rate": 5e-06,
+      "loss": 0.2252,
+      "num_input_tokens_seen": 393398824,
+      "step": 2276
+    },
+    {
+      "epoch": 0.8716966679433167,
+      "loss": 0.2406993806362152,
+      "loss_ce": 0.020484544336795807,
+      "loss_iou": 1.0057376623153687,
+      "loss_num": 0.220703125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 393398824,
+      "step": 2276
+    },
+    {
+      "epoch": 0.8720796629643814,
+      "grad_norm": 52.04962328569182,
+      "learning_rate": 5e-06,
+      "loss": 0.2097,
+      "num_input_tokens_seen": 393571680,
+      "step": 2277
+    },
+    {
+      "epoch": 0.8720796629643814,
+      "loss": 0.19266335666179657,
+      "loss_ce": 0.02145974338054657,
+      "loss_iou": 1.0567872524261475,
+      "loss_num": 0.1708984375,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 393571680,
+      "step": 2277
+    },
+    {
+      "epoch": 0.8724626579854462,
+      "grad_norm": 29.23762116128551,
+      "learning_rate": 5e-06,
+      "loss": 0.2211,
+      "num_input_tokens_seen": 393744744,
+      "step": 2278
+    },
+    {
+      "epoch": 0.8724626579854462,
+      "loss": 0.27350762486457825,
+      "loss_ce": 0.022042784839868546,
+      "loss_iou": 1.0850646495819092,
+      "loss_num": 0.251953125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 393744744,
+      "step": 2278
+    },
+    {
+      "epoch": 0.872845653006511,
+      "grad_norm": 78.34644998257741,
+      "learning_rate": 5e-06,
+      "loss": 0.2937,
+      "num_input_tokens_seen": 393918376,
+      "step": 2279
+    },
+    {
+      "epoch": 0.872845653006511,
+      "loss": 0.30728715658187866,
+      "loss_ce": 0.021642638370394707,
+      "loss_iou": 1.0545094013214111,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 393918376,
+      "step": 2279
+    },
+    {
+      "epoch": 0.8732286480275756,
+      "grad_norm": 31.58669547372777,
+      "learning_rate": 5e-06,
+      "loss": 0.2419,
+      "num_input_tokens_seen": 394091272,
+      "step": 2280
+    },
+    {
+      "epoch": 0.8732286480275756,
+      "loss": 0.21547313034534454,
+      "loss_ce": 0.020221669226884842,
+      "loss_iou": 1.0116897821426392,
+      "loss_num": 0.1953125,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 394091272,
+      "step": 2280
+    },
+    {
+      "epoch": 0.8736116430486404,
+      "grad_norm": 24.39006798014416,
+      "learning_rate": 5e-06,
+      "loss": 0.1925,
+      "num_input_tokens_seen": 394264520,
+      "step": 2281
+    },
+    {
+      "epoch": 0.8736116430486404,
+      "loss": 0.19357803463935852,
+      "loss_ce": 0.020970605313777924,
+      "loss_iou": 1.0049718618392944,
+      "loss_num": 0.1728515625,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 394264520,
+      "step": 2281
+    },
+    {
+      "epoch": 0.8739946380697051,
+      "grad_norm": 36.59112125689424,
+      "learning_rate": 5e-06,
+      "loss": 0.1889,
+      "num_input_tokens_seen": 394437224,
+      "step": 2282
+    },
+    {
+      "epoch": 0.8739946380697051,
+      "loss": 0.16868579387664795,
+      "loss_ce": 0.0213469248265028,
+      "loss_iou": 0.8756428956985474,
+      "loss_num": 0.1474609375,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 394437224,
+      "step": 2282
+    },
+    {
+      "epoch": 0.8743776330907698,
+      "grad_norm": 42.07793700100101,
+      "learning_rate": 5e-06,
+      "loss": 0.2283,
+      "num_input_tokens_seen": 394610136,
+      "step": 2283
+    },
+    {
+      "epoch": 0.8743776330907698,
+      "loss": 0.2361423671245575,
+      "loss_ce": 0.020322054624557495,
+      "loss_iou": 1.027820110321045,
+      "loss_num": 0.2158203125,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 394610136,
+      "step": 2283
+    },
+    {
+      "epoch": 0.8747606281118345,
+      "grad_norm": 30.9570465693495,
+      "learning_rate": 5e-06,
+      "loss": 0.2176,
+      "num_input_tokens_seen": 394783072,
+      "step": 2284
+    },
+    {
+      "epoch": 0.8747606281118345,
+      "loss": 0.21155185997486115,
+      "loss_ce": 0.019901473075151443,
+      "loss_iou": 1.0322974920272827,
+      "loss_num": 0.19140625,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 394783072,
+      "step": 2284
+    },
+    {
+      "epoch": 0.8751436231328993,
+      "grad_norm": 22.795918529470963,
+      "learning_rate": 5e-06,
+      "loss": 0.2171,
+      "num_input_tokens_seen": 394955864,
+      "step": 2285
+    },
+    {
+      "epoch": 0.8751436231328993,
+      "loss": 0.21094149351119995,
+      "loss_ce": 0.02167147397994995,
+      "loss_iou": 1.0987880229949951,
+      "loss_num": 0.189453125,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 394955864,
+      "step": 2285
+    },
+    {
+      "epoch": 0.875526618153964,
+      "grad_norm": 21.65343377133573,
+      "learning_rate": 5e-06,
+      "loss": 0.1927,
+      "num_input_tokens_seen": 395128664,
+      "step": 2286
+    },
+    {
+      "epoch": 0.875526618153964,
+      "loss": 0.21473294496536255,
+      "loss_ce": 0.020763229578733444,
+      "loss_iou": 1.014101505279541,
+      "loss_num": 0.1943359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 395128664,
+      "step": 2286
+    },
+    {
+      "epoch": 0.8759096131750287,
+      "grad_norm": 23.664828712220437,
+      "learning_rate": 5e-06,
+      "loss": 0.171,
+      "num_input_tokens_seen": 395301648,
+      "step": 2287
+    },
+    {
+      "epoch": 0.8759096131750287,
+      "loss": 0.19143562018871307,
+      "loss_ce": 0.020109932869672775,
+      "loss_iou": 1.0008352994918823,
+      "loss_num": 0.1708984375,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 395301648,
+      "step": 2287
+    },
+    {
+      "epoch": 0.8762926081960934,
+      "grad_norm": 32.53198796879764,
+      "learning_rate": 5e-06,
+      "loss": 0.165,
+      "num_input_tokens_seen": 395474368,
+      "step": 2288
+    },
+    {
+      "epoch": 0.8762926081960934,
+      "loss": 0.1494501382112503,
+      "loss_ce": 0.021642513573169708,
+      "loss_iou": 1.0112065076828003,
+      "loss_num": 0.1279296875,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 395474368,
+      "step": 2288
+    },
+    {
+      "epoch": 0.8766756032171582,
+      "grad_norm": 34.66682965373584,
+      "learning_rate": 5e-06,
+      "loss": 0.2111,
+      "num_input_tokens_seen": 395644016,
+      "step": 2289
+    },
+    {
+      "epoch": 0.8766756032171582,
+      "loss": 0.23358552157878876,
+      "loss_ce": 0.021305250003933907,
+      "loss_iou": 1.059319257736206,
+      "loss_num": 0.2119140625,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 395644016,
+      "step": 2289
+    },
+    {
+      "epoch": 0.8770585982382229,
+      "grad_norm": 18.081521143637627,
+      "learning_rate": 5e-06,
+      "loss": 0.211,
+      "num_input_tokens_seen": 395816712,
+      "step": 2290
+    },
+    {
+      "epoch": 0.8770585982382229,
+      "loss": 0.17363527417182922,
+      "loss_ce": 0.021596703678369522,
+      "loss_iou": 1.0326075553894043,
+      "loss_num": 0.15234375,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 395816712,
+      "step": 2290
+    },
+    {
+      "epoch": 0.8774415932592876,
+      "grad_norm": 18.210939402159955,
+      "learning_rate": 5e-06,
+      "loss": 0.2151,
+      "num_input_tokens_seen": 395989752,
+      "step": 2291
+    },
+    {
+      "epoch": 0.8774415932592876,
+      "loss": 0.228585883975029,
+      "loss_ce": 0.021676693111658096,
+      "loss_iou": 1.062607765197754,
+      "loss_num": 0.20703125,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 395989752,
+      "step": 2291
+    },
+    {
+      "epoch": 0.8778245882803524,
+      "grad_norm": 27.092469139972376,
+      "learning_rate": 5e-06,
+      "loss": 0.1521,
+      "num_input_tokens_seen": 396162512,
+      "step": 2292
+    },
+    {
+      "epoch": 0.8778245882803524,
+      "loss": 0.18751798570156097,
+      "loss_ce": 0.020647864788770676,
+      "loss_iou": 8.32317092605775e+22,
+      "loss_num": 0.1669921875,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 396162512,
+      "step": 2292
+    },
+    {
+      "epoch": 0.8782075833014171,
+      "grad_norm": 30.001011135359228,
+      "learning_rate": 5e-06,
+      "loss": 0.2066,
+      "num_input_tokens_seen": 396336072,
+      "step": 2293
+    },
+    {
+      "epoch": 0.8782075833014171,
+      "loss": 0.1969650387763977,
+      "loss_ce": 0.020817570388317108,
+      "loss_iou": 1.007326364517212,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 396336072,
+      "step": 2293
+    },
+    {
+      "epoch": 0.8785905783224818,
+      "grad_norm": 26.083469781572422,
+      "learning_rate": 5e-06,
+      "loss": 0.2127,
+      "num_input_tokens_seen": 396509304,
+      "step": 2294
+    },
+    {
+      "epoch": 0.8785905783224818,
+      "loss": 0.1951311081647873,
+      "loss_ce": 0.020448490977287292,
+      "loss_iou": 1.0216398239135742,
+      "loss_num": 0.1748046875,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 396509304,
+      "step": 2294
+    },
+    {
+      "epoch": 0.8789735733435465,
+      "grad_norm": 27.761394867677065,
+      "learning_rate": 5e-06,
+      "loss": 0.2432,
+      "num_input_tokens_seen": 396682400,
+      "step": 2295
+    },
+    {
+      "epoch": 0.8789735733435465,
+      "loss": 0.2696751356124878,
+      "loss_ce": 0.021384134888648987,
+      "loss_iou": 1.0838731527328491,
+      "loss_num": 0.248046875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 396682400,
+      "step": 2295
+    },
+    {
+      "epoch": 0.8793565683646113,
+      "grad_norm": 20.606424256937945,
+      "learning_rate": 5e-06,
+      "loss": 0.16,
+      "num_input_tokens_seen": 396854928,
+      "step": 2296
+    },
+    {
+      "epoch": 0.8793565683646113,
+      "loss": 0.1552027463912964,
+      "loss_ce": 0.020192988216876984,
+      "loss_iou": 1.001530647277832,
+      "loss_num": 0.134765625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 396854928,
+      "step": 2296
+    },
+    {
+      "epoch": 0.8797395633856759,
+      "grad_norm": 13.681449188907138,
+      "learning_rate": 5e-06,
+      "loss": 0.1166,
+      "num_input_tokens_seen": 397027488,
+      "step": 2297
+    },
+    {
+      "epoch": 0.8797395633856759,
+      "loss": 0.12320609390735626,
+      "loss_ce": 0.020117711275815964,
+      "loss_iou": 1.0067899227142334,
+      "loss_num": 0.10302734375,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 397027488,
+      "step": 2297
+    },
+    {
+      "epoch": 0.8801225584067407,
+      "grad_norm": 20.135722752597573,
+      "learning_rate": 5e-06,
+      "loss": 0.1691,
+      "num_input_tokens_seen": 397200152,
+      "step": 2298
+    },
+    {
+      "epoch": 0.8801225584067407,
+      "loss": 0.17363953590393066,
+      "loss_ce": 0.021173711866140366,
+      "loss_iou": 1.0135451555252075,
+      "loss_num": 0.15234375,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 397200152,
+      "step": 2298
+    },
+    {
+      "epoch": 0.8805055534278055,
+      "grad_norm": 30.52294808367406,
+      "learning_rate": 5e-06,
+      "loss": 0.1874,
+      "num_input_tokens_seen": 397372912,
+      "step": 2299
+    },
+    {
+      "epoch": 0.8805055534278055,
+      "loss": 0.2073453664779663,
+      "loss_ce": 0.02069985494017601,
+      "loss_iou": 1.1049034595489502,
+      "loss_num": 0.1865234375,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 397372912,
+      "step": 2299
+    },
+    {
+      "epoch": 0.8808885484488702,
+      "grad_norm": 32.87993080154758,
+      "learning_rate": 5e-06,
+      "loss": 0.1795,
+      "num_input_tokens_seen": 397546000,
+      "step": 2300
+    },
+    {
+      "epoch": 0.8808885484488702,
+      "loss": 0.16229727864265442,
+      "loss_ce": 0.02075674757361412,
+      "loss_iou": 1.0005006790161133,
+      "loss_num": 0.1416015625,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 397546000,
+      "step": 2300
+    },
+    {
+      "epoch": 0.8812715434699349,
+      "grad_norm": 25.04198557838158,
+      "learning_rate": 5e-06,
+      "loss": 0.1954,
+      "num_input_tokens_seen": 397718872,
+      "step": 2301
+    },
+    {
+      "epoch": 0.8812715434699349,
+      "loss": 0.1789538711309433,
+      "loss_ce": 0.02154420129954815,
+      "loss_iou": 1.0006424188613892,
+      "loss_num": 0.1572265625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 397718872,
+      "step": 2301
+    },
+    {
+      "epoch": 0.8816545384909996,
+      "grad_norm": 21.51257930812808,
+      "learning_rate": 5e-06,
+      "loss": 0.1907,
+      "num_input_tokens_seen": 397891928,
+      "step": 2302
+    },
+    {
+      "epoch": 0.8816545384909996,
+      "loss": 0.20330749452114105,
+      "loss_ce": 0.021727904677391052,
+      "loss_iou": 1.0173156261444092,
+      "loss_num": 0.181640625,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 397891928,
+      "step": 2302
+    },
+    {
+      "epoch": 0.8820375335120644,
+      "grad_norm": 23.850911095849863,
+      "learning_rate": 5e-06,
+      "loss": 0.2333,
+      "num_input_tokens_seen": 398064856,
+      "step": 2303
+    },
+    {
+      "epoch": 0.8820375335120644,
+      "loss": 0.23615139722824097,
+      "loss_ce": 0.020147982984781265,
+      "loss_iou": 1.3376621007919312,
+      "loss_num": 0.2158203125,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 398064856,
+      "step": 2303
+    },
+    {
+      "epoch": 0.882420528533129,
+      "grad_norm": 28.43514558943183,
+      "learning_rate": 5e-06,
+      "loss": 0.1375,
+      "num_input_tokens_seen": 398237736,
+      "step": 2304
+    },
+    {
+      "epoch": 0.882420528533129,
+      "loss": 0.1300610601902008,
+      "loss_ce": 0.019282255321741104,
+      "loss_iou": 1.0002906322479248,
+      "loss_num": 0.11083984375,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 398237736,
+      "step": 2304
+    },
+    {
+      "epoch": 0.8828035235541938,
+      "grad_norm": 23.604836666199777,
+      "learning_rate": 5e-06,
+      "loss": 0.1471,
+      "num_input_tokens_seen": 398410496,
+      "step": 2305
+    },
+    {
+      "epoch": 0.8828035235541938,
+      "loss": 0.1526983082294464,
+      "loss_ce": 0.020496167242527008,
+      "loss_iou": 1.0016909837722778,
+      "loss_num": 0.1318359375,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 398410496,
+      "step": 2305
+    },
+    {
+      "epoch": 0.8831865185752585,
+      "grad_norm": 28.21258043854851,
+      "learning_rate": 5e-06,
+      "loss": 0.1752,
+      "num_input_tokens_seen": 398583592,
+      "step": 2306
+    },
+    {
+      "epoch": 0.8831865185752585,
+      "loss": 0.19757401943206787,
+      "loss_ce": 0.020572058856487274,
+      "loss_iou": 1.0885980129241943,
+      "loss_num": 0.1767578125,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 398583592,
+      "step": 2306
+    },
+    {
+      "epoch": 0.8835695135963233,
+      "grad_norm": 24.9937839191693,
+      "learning_rate": 5e-06,
+      "loss": 0.1695,
+      "num_input_tokens_seen": 398756624,
+      "step": 2307
+    },
+    {
+      "epoch": 0.8835695135963233,
+      "loss": 0.14079371094703674,
+      "loss_ce": 0.020798582583665848,
+      "loss_iou": 1.0043283700942993,
+      "loss_num": 0.1201171875,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 398756624,
+      "step": 2307
+    },
+    {
+      "epoch": 0.8839525086173879,
+      "grad_norm": 22.69790201692408,
+      "learning_rate": 5e-06,
+      "loss": 0.2299,
+      "num_input_tokens_seen": 398929480,
+      "step": 2308
+    },
+    {
+      "epoch": 0.8839525086173879,
+      "loss": 0.19839608669281006,
+      "loss_ce": 0.02188241109251976,
+      "loss_iou": 1.0025566816329956,
+      "loss_num": 0.1767578125,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 398929480,
+      "step": 2308
+    },
+    {
+      "epoch": 0.8843355036384527,
+      "grad_norm": 22.367970040983835,
+      "learning_rate": 5e-06,
+      "loss": 0.2076,
+      "num_input_tokens_seen": 399102480,
+      "step": 2309
+    },
+    {
+      "epoch": 0.8843355036384527,
+      "loss": 0.24297428131103516,
+      "loss_ce": 0.021843906491994858,
+      "loss_iou": 1.0768591165542603,
+      "loss_num": 0.220703125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 399102480,
+      "step": 2309
+    },
+    {
+      "epoch": 0.8847184986595175,
+      "grad_norm": 23.681641625883703,
+      "learning_rate": 5e-06,
+      "loss": 0.2082,
+      "num_input_tokens_seen": 399275288,
+      "step": 2310
+    },
+    {
+      "epoch": 0.8847184986595175,
+      "loss": 0.2108621895313263,
+      "loss_ce": 0.020432502031326294,
+      "loss_iou": 1.0344903469085693,
+      "loss_num": 0.1904296875,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 399275288,
+      "step": 2310
+    },
+    {
+      "epoch": 0.8851014936805821,
+      "grad_norm": 24.10603166601443,
+      "learning_rate": 5e-06,
+      "loss": 0.2664,
+      "num_input_tokens_seen": 399447968,
+      "step": 2311
+    },
+    {
+      "epoch": 0.8851014936805821,
+      "loss": 0.2505452632904053,
+      "loss_ce": 0.020808910951018333,
+      "loss_iou": 1.1432623863220215,
+      "loss_num": 0.2294921875,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 399447968,
+      "step": 2311
+    },
+    {
+      "epoch": 0.8854844887016469,
+      "grad_norm": 22.023493830738364,
+      "learning_rate": 5e-06,
+      "loss": 0.1626,
+      "num_input_tokens_seen": 399620456,
+      "step": 2312
+    },
+    {
+      "epoch": 0.8854844887016469,
+      "loss": 0.17484048008918762,
+      "loss_ce": 0.020421534776687622,
+      "loss_iou": 1.0182372331619263,
+      "loss_num": 0.154296875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 399620456,
+      "step": 2312
+    },
+    {
+      "epoch": 0.8858674837227116,
+      "grad_norm": 25.7498391692391,
+      "learning_rate": 5e-06,
+      "loss": 0.2168,
+      "num_input_tokens_seen": 399793472,
+      "step": 2313
+    },
+    {
+      "epoch": 0.8858674837227116,
+      "loss": 0.1925235092639923,
+      "loss_ce": 0.02113679237663746,
+      "loss_iou": 1.0034724473953247,
+      "loss_num": 0.171875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 399793472,
+      "step": 2313
+    },
+    {
+      "epoch": 0.8862504787437764,
+      "grad_norm": 25.791077663218175,
+      "learning_rate": 5e-06,
+      "loss": 0.212,
+      "num_input_tokens_seen": 399966424,
+      "step": 2314
+    },
+    {
+      "epoch": 0.8862504787437764,
+      "loss": 0.17002347111701965,
+      "loss_ce": 0.019938018172979355,
+      "loss_iou": 1.0325002670288086,
+      "loss_num": 0.150390625,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 399966424,
+      "step": 2314
+    },
+    {
+      "epoch": 0.886633473764841,
+      "grad_norm": 26.5356666547837,
+      "learning_rate": 5e-06,
+      "loss": 0.2239,
+      "num_input_tokens_seen": 400139488,
+      "step": 2315
+    },
+    {
+      "epoch": 0.886633473764841,
+      "loss": 0.263716459274292,
+      "loss_ce": 0.020247209817171097,
+      "loss_iou": 1.0764899253845215,
+      "loss_num": 0.2431640625,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 400139488,
+      "step": 2315
+    },
+    {
+      "epoch": 0.8870164687859058,
+      "grad_norm": 23.15099092876043,
+      "learning_rate": 5e-06,
+      "loss": 0.2141,
+      "num_input_tokens_seen": 400312184,
+      "step": 2316
+    },
+    {
+      "epoch": 0.8870164687859058,
+      "loss": 0.22713661193847656,
+      "loss_ce": 0.01992225833237171,
+      "loss_iou": 1.0985429286956787,
+      "loss_num": 0.20703125,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 400312184,
+      "step": 2316
+    },
+    {
+      "epoch": 0.8873994638069705,
+      "grad_norm": 30.484934669020465,
+      "learning_rate": 5e-06,
+      "loss": 0.1671,
+      "num_input_tokens_seen": 400485120,
+      "step": 2317
+    },
+    {
+      "epoch": 0.8873994638069705,
+      "loss": 0.1689269244670868,
+      "loss_ce": 0.02097770944237709,
+      "loss_iou": 1.0121673345565796,
+      "loss_num": 0.1484375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 400485120,
+      "step": 2317
+    },
+    {
+      "epoch": 0.8877824588280352,
+      "grad_norm": 28.322684395014303,
+      "learning_rate": 5e-06,
+      "loss": 0.2812,
+      "num_input_tokens_seen": 400658064,
+      "step": 2318
+    },
+    {
+      "epoch": 0.8877824588280352,
+      "loss": 0.3154546916484833,
+      "loss_ce": 0.019800391048192978,
+      "loss_iou": 1.0829479694366455,
+      "loss_num": 0.294921875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 400658064,
+      "step": 2318
+    },
+    {
+      "epoch": 0.8881654538491,
+      "grad_norm": 30.649776439699774,
+      "learning_rate": 5e-06,
+      "loss": 0.3124,
+      "num_input_tokens_seen": 400830888,
+      "step": 2319
+    },
+    {
+      "epoch": 0.8881654538491,
+      "loss": 0.39187049865722656,
+      "loss_ce": 0.019617099314928055,
+      "loss_iou": 1.1072109937667847,
+      "loss_num": 0.373046875,
+      "loss_xval": 0.373046875,
+      "num_input_tokens_seen": 400830888,
+      "step": 2319
+    },
+    {
+      "epoch": 0.8885484488701647,
+      "grad_norm": 21.581339684611716,
+      "learning_rate": 5e-06,
+      "loss": 0.2397,
+      "num_input_tokens_seen": 401003832,
+      "step": 2320
+    },
+    {
+      "epoch": 0.8885484488701647,
+      "loss": 0.2565586566925049,
+      "loss_ce": 0.02132917195558548,
+      "loss_iou": 1.0717475414276123,
+      "loss_num": 0.2353515625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 401003832,
+      "step": 2320
+    },
+    {
+      "epoch": 0.8889314438912295,
+      "grad_norm": 37.18032737189331,
+      "learning_rate": 5e-06,
+      "loss": 0.2298,
+      "num_input_tokens_seen": 401176816,
+      "step": 2321
+    },
+    {
+      "epoch": 0.8889314438912295,
+      "loss": 0.24256186187267303,
+      "loss_ce": 0.02124839276075363,
+      "loss_iou": 1.0216059684753418,
+      "loss_num": 0.2216796875,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 401176816,
+      "step": 2321
+    },
+    {
+      "epoch": 0.8893144389122941,
+      "grad_norm": 28.287794798157343,
+      "learning_rate": 5e-06,
+      "loss": 0.1649,
+      "num_input_tokens_seen": 401349880,
+      "step": 2322
+    },
+    {
+      "epoch": 0.8893144389122941,
+      "loss": 0.1690763682126999,
+      "loss_ce": 0.02088300511240959,
+      "loss_iou": 1.0003900527954102,
+      "loss_num": 0.1484375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 401349880,
+      "step": 2322
+    },
+    {
+      "epoch": 0.8896974339333589,
+      "grad_norm": 29.561315514949193,
+      "learning_rate": 5e-06,
+      "loss": 0.2366,
+      "num_input_tokens_seen": 401522624,
+      "step": 2323
+    },
+    {
+      "epoch": 0.8896974339333589,
+      "loss": 0.2799972891807556,
+      "loss_ce": 0.018339548259973526,
+      "loss_iou": 0.9488890171051025,
+      "loss_num": 0.26171875,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 401522624,
+      "step": 2323
+    },
+    {
+      "epoch": 0.8900804289544236,
+      "grad_norm": 40.101712454574034,
+      "learning_rate": 5e-06,
+      "loss": 0.2302,
+      "num_input_tokens_seen": 401695440,
+      "step": 2324
+    },
+    {
+      "epoch": 0.8900804289544236,
+      "loss": 0.23104974627494812,
+      "loss_ce": 0.021516067907214165,
+      "loss_iou": 1.0083305835723877,
+      "loss_num": 0.2099609375,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 401695440,
+      "step": 2324
+    },
+    {
+      "epoch": 0.8904634239754883,
+      "grad_norm": 38.15009426096198,
+      "learning_rate": 5e-06,
+      "loss": 0.3425,
+      "num_input_tokens_seen": 401864912,
+      "step": 2325
+    },
+    {
+      "epoch": 0.8904634239754883,
+      "loss": 0.2780781090259552,
+      "loss_ce": 0.020875953137874603,
+      "loss_iou": 1.0925370454788208,
+      "loss_num": 0.2578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 401864912,
+      "step": 2325
+    },
+    {
+      "epoch": 0.890846418996553,
+      "grad_norm": 25.608395885078057,
+      "learning_rate": 5e-06,
+      "loss": 0.2329,
+      "num_input_tokens_seen": 402037568,
+      "step": 2326
+    },
+    {
+      "epoch": 0.890846418996553,
+      "loss": 0.2409341037273407,
+      "loss_ce": 0.022062024101614952,
+      "loss_iou": 1.021268367767334,
+      "loss_num": 0.21875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 402037568,
+      "step": 2326
+    },
+    {
+      "epoch": 0.8912294140176178,
+      "grad_norm": 23.19897108760206,
+      "learning_rate": 5e-06,
+      "loss": 0.208,
+      "num_input_tokens_seen": 402210232,
+      "step": 2327
+    },
+    {
+      "epoch": 0.8912294140176178,
+      "loss": 0.23976726830005646,
+      "loss_ce": 0.02107830159366131,
+      "loss_iou": 1.0104955434799194,
+      "loss_num": 0.21875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 402210232,
+      "step": 2327
+    },
+    {
+      "epoch": 0.8916124090386826,
+      "grad_norm": 27.320496113020106,
+      "learning_rate": 5e-06,
+      "loss": 0.2069,
+      "num_input_tokens_seen": 402382840,
+      "step": 2328
+    },
+    {
+      "epoch": 0.8916124090386826,
+      "loss": 0.2346423864364624,
+      "loss_ce": 0.020470030605793,
+      "loss_iou": 1.071311354637146,
+      "loss_num": 0.2138671875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 402382840,
+      "step": 2328
+    },
+    {
+      "epoch": 0.8919954040597472,
+      "grad_norm": 43.879260420265275,
+      "learning_rate": 5e-06,
+      "loss": 0.1992,
+      "num_input_tokens_seen": 402555744,
+      "step": 2329
+    },
+    {
+      "epoch": 0.8919954040597472,
+      "loss": 0.21297091245651245,
+      "loss_ce": 0.021320518106222153,
+      "loss_iou": 1.0636193752288818,
+      "loss_num": 0.19140625,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 402555744,
+      "step": 2329
+    },
+    {
+      "epoch": 0.892378399080812,
+      "grad_norm": 34.839557886600105,
+      "learning_rate": 5e-06,
+      "loss": 0.2513,
+      "num_input_tokens_seen": 402729064,
+      "step": 2330
+    },
+    {
+      "epoch": 0.892378399080812,
+      "loss": 0.29425114393234253,
+      "loss_ce": 0.020325375720858574,
+      "loss_iou": 1.0446100234985352,
+      "loss_num": 0.2734375,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 402729064,
+      "step": 2330
+    },
+    {
+      "epoch": 0.8927613941018767,
+      "grad_norm": 35.37658936273886,
+      "learning_rate": 5e-06,
+      "loss": 0.2368,
+      "num_input_tokens_seen": 402902120,
+      "step": 2331
+    },
+    {
+      "epoch": 0.8927613941018767,
+      "loss": 0.19274011254310608,
+      "loss_ce": 0.021658584475517273,
+      "loss_iou": 1.0351388454437256,
+      "loss_num": 0.1708984375,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 402902120,
+      "step": 2331
+    },
+    {
+      "epoch": 0.8931443891229414,
+      "grad_norm": 16.206575345603024,
+      "learning_rate": 5e-06,
+      "loss": 0.1766,
+      "num_input_tokens_seen": 403075048,
+      "step": 2332
+    },
+    {
+      "epoch": 0.8931443891229414,
+      "loss": 0.15430493652820587,
+      "loss_ce": 0.020760010927915573,
+      "loss_iou": 1.0188767910003662,
+      "loss_num": 0.1337890625,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 403075048,
+      "step": 2332
+    },
+    {
+      "epoch": 0.8935273841440061,
+      "grad_norm": 51.08128692755289,
+      "learning_rate": 5e-06,
+      "loss": 0.2559,
+      "num_input_tokens_seen": 403247920,
+      "step": 2333
+    },
+    {
+      "epoch": 0.8935273841440061,
+      "loss": 0.23070073127746582,
+      "loss_ce": 0.020251505076885223,
+      "loss_iou": 1.1131281852722168,
+      "loss_num": 0.2109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 403247920,
+      "step": 2333
+    },
+    {
+      "epoch": 0.8939103791650709,
+      "grad_norm": 37.256973862167605,
+      "learning_rate": 5e-06,
+      "loss": 0.2325,
+      "num_input_tokens_seen": 403420944,
+      "step": 2334
+    },
+    {
+      "epoch": 0.8939103791650709,
+      "loss": 0.2458370476961136,
+      "loss_ce": 0.020739397034049034,
+      "loss_iou": 1.0012136697769165,
+      "loss_num": 0.224609375,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 403420944,
+      "step": 2334
+    },
+    {
+      "epoch": 0.8942933741861355,
+      "grad_norm": 41.35793357453431,
+      "learning_rate": 5e-06,
+      "loss": 0.1832,
+      "num_input_tokens_seen": 403593960,
+      "step": 2335
+    },
+    {
+      "epoch": 0.8942933741861355,
+      "loss": 0.1998976767063141,
+      "loss_ce": 0.02149190567433834,
+      "loss_iou": 1.020889163017273,
+      "loss_num": 0.1787109375,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 403593960,
+      "step": 2335
+    },
+    {
+      "epoch": 0.8946763692072003,
+      "grad_norm": 28.510224293167948,
+      "learning_rate": 5e-06,
+      "loss": 0.2628,
+      "num_input_tokens_seen": 403766944,
+      "step": 2336
+    },
+    {
+      "epoch": 0.8946763692072003,
+      "loss": 0.27586835622787476,
+      "loss_ce": 0.021046578884124756,
+      "loss_iou": 1.0037611722946167,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 403766944,
+      "step": 2336
+    },
+    {
+      "epoch": 0.895059364228265,
+      "grad_norm": 52.29913170388032,
+      "learning_rate": 5e-06,
+      "loss": 0.2434,
+      "num_input_tokens_seen": 403939816,
+      "step": 2337
+    },
+    {
+      "epoch": 0.895059364228265,
+      "loss": 0.19048947095870972,
+      "loss_ce": 0.021238982677459717,
+      "loss_iou": 1.0058882236480713,
+      "loss_num": 0.1689453125,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 403939816,
+      "step": 2337
+    },
+    {
+      "epoch": 0.8954423592493298,
+      "grad_norm": 28.3177053650063,
+      "learning_rate": 5e-06,
+      "loss": 0.2138,
+      "num_input_tokens_seen": 404112640,
+      "step": 2338
+    },
+    {
+      "epoch": 0.8954423592493298,
+      "loss": 0.2756827473640442,
+      "loss_ce": 0.020616818219423294,
+      "loss_iou": 1.1016333103179932,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 404112640,
+      "step": 2338
+    },
+    {
+      "epoch": 0.8958253542703944,
+      "grad_norm": 45.586183419404904,
+      "learning_rate": 5e-06,
+      "loss": 0.2483,
+      "num_input_tokens_seen": 404285592,
+      "step": 2339
+    },
+    {
+      "epoch": 0.8958253542703944,
+      "loss": 0.19825810194015503,
+      "loss_ce": 0.021988561376929283,
+      "loss_iou": 0.97584068775177,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 404285592,
+      "step": 2339
+    },
+    {
+      "epoch": 0.8962083492914592,
+      "grad_norm": 35.55583502603937,
+      "learning_rate": 5e-06,
+      "loss": 0.2635,
+      "num_input_tokens_seen": 404458248,
+      "step": 2340
+    },
+    {
+      "epoch": 0.8962083492914592,
+      "loss": 0.22127856314182281,
+      "loss_ce": 0.019374258816242218,
+      "loss_iou": 1.1606025695800781,
+      "loss_num": 0.2021484375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 404458248,
+      "step": 2340
+    },
+    {
+      "epoch": 0.896591344312524,
+      "grad_norm": 50.96979750171572,
+      "learning_rate": 5e-06,
+      "loss": 0.2206,
+      "num_input_tokens_seen": 404631176,
+      "step": 2341
+    },
+    {
+      "epoch": 0.896591344312524,
+      "loss": 0.239015594124794,
+      "loss_ce": 0.021242156624794006,
+      "loss_iou": 1.0026321411132812,
+      "loss_num": 0.2177734375,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 404631176,
+      "step": 2341
+    },
+    {
+      "epoch": 0.8969743393335886,
+      "grad_norm": 34.36341472556792,
+      "learning_rate": 5e-06,
+      "loss": 0.2009,
+      "num_input_tokens_seen": 404804424,
+      "step": 2342
+    },
+    {
+      "epoch": 0.8969743393335886,
+      "loss": 0.19493183493614197,
+      "loss_ce": 0.020432332530617714,
+      "loss_iou": 1.0039664506912231,
+      "loss_num": 0.1748046875,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 404804424,
+      "step": 2342
+    },
+    {
+      "epoch": 0.8973573343546534,
+      "grad_norm": 36.966826818188856,
+      "learning_rate": 5e-06,
+      "loss": 0.2087,
+      "num_input_tokens_seen": 404977240,
+      "step": 2343
+    },
+    {
+      "epoch": 0.8973573343546534,
+      "loss": 0.2385326772928238,
+      "loss_ce": 0.02033199742436409,
+      "loss_iou": 1.0842691659927368,
+      "loss_num": 0.2177734375,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 404977240,
+      "step": 2343
+    },
+    {
+      "epoch": 0.8977403293757181,
+      "grad_norm": 27.400570672450467,
+      "learning_rate": 5e-06,
+      "loss": 0.1975,
+      "num_input_tokens_seen": 405150168,
+      "step": 2344
+    },
+    {
+      "epoch": 0.8977403293757181,
+      "loss": 0.16610579192638397,
+      "loss_ce": 0.01998763158917427,
+      "loss_iou": 1.0006134510040283,
+      "loss_num": 0.146484375,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 405150168,
+      "step": 2344
+    },
+    {
+      "epoch": 0.8981233243967829,
+      "grad_norm": 45.70484841006972,
+      "learning_rate": 5e-06,
+      "loss": 0.1995,
+      "num_input_tokens_seen": 405323056,
+      "step": 2345
+    },
+    {
+      "epoch": 0.8981233243967829,
+      "loss": 0.17091411352157593,
+      "loss_ce": 0.020889708772301674,
+      "loss_iou": 1.0392924547195435,
+      "loss_num": 0.150390625,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 405323056,
+      "step": 2345
+    },
+    {
+      "epoch": 0.8985063194178475,
+      "grad_norm": 39.637040555052735,
+      "learning_rate": 5e-06,
+      "loss": 0.2342,
+      "num_input_tokens_seen": 405495648,
+      "step": 2346
+    },
+    {
+      "epoch": 0.8985063194178475,
+      "loss": 0.17984412610530853,
+      "loss_ce": 0.019687866792082787,
+      "loss_iou": 1.0187370777130127,
+      "loss_num": 0.16015625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 405495648,
+      "step": 2346
+    },
+    {
+      "epoch": 0.8988893144389123,
+      "grad_norm": 30.57360473875093,
+      "learning_rate": 5e-06,
+      "loss": 0.1576,
+      "num_input_tokens_seen": 405668616,
+      "step": 2347
+    },
+    {
+      "epoch": 0.8988893144389123,
+      "loss": 0.1363660991191864,
+      "loss_ce": 0.0201551616191864,
+      "loss_iou": 1.0004048347473145,
+      "loss_num": 0.1162109375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 405668616,
+      "step": 2347
+    },
+    {
+      "epoch": 0.899272309459977,
+      "grad_norm": 32.006914027939956,
+      "learning_rate": 5e-06,
+      "loss": 0.2305,
+      "num_input_tokens_seen": 405841544,
+      "step": 2348
+    },
+    {
+      "epoch": 0.899272309459977,
+      "loss": 0.26787641644477844,
+      "loss_ce": 0.020745061337947845,
+      "loss_iou": 1.089160442352295,
+      "loss_num": 0.2470703125,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 405841544,
+      "step": 2348
+    },
+    {
+      "epoch": 0.8996553044810417,
+      "grad_norm": 72.12367727509918,
+      "learning_rate": 5e-06,
+      "loss": 0.2419,
+      "num_input_tokens_seen": 406014736,
+      "step": 2349
+    },
+    {
+      "epoch": 0.8996553044810417,
+      "loss": 0.2803584635257721,
+      "loss_ce": 0.022545959800481796,
+      "loss_iou": 2.920067548751831,
+      "loss_num": 0.2578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 406014736,
+      "step": 2349
+    },
+    {
+      "epoch": 0.9000382995021065,
+      "grad_norm": 20.010034228136895,
+      "learning_rate": 5e-06,
+      "loss": 0.2155,
+      "num_input_tokens_seen": 406182336,
+      "step": 2350
+    },
+    {
+      "epoch": 0.9000382995021065,
+      "loss": 0.187499538064003,
+      "loss_ce": 0.02062942273914814,
+      "loss_iou": 1.0197784900665283,
+      "loss_num": 0.1669921875,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 406182336,
+      "step": 2350
+    },
+    {
+      "epoch": 0.9004212945231712,
+      "grad_norm": 22.701776072233677,
+      "learning_rate": 5e-06,
+      "loss": 0.1878,
+      "num_input_tokens_seen": 406355256,
+      "step": 2351
+    },
+    {
+      "epoch": 0.9004212945231712,
+      "loss": 0.20990976691246033,
+      "loss_ce": 0.021860448643565178,
+      "loss_iou": 1.080086350440979,
+      "loss_num": 0.1884765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 406355256,
+      "step": 2351
+    },
+    {
+      "epoch": 0.900804289544236,
+      "grad_norm": 20.61023095429758,
+      "learning_rate": 5e-06,
+      "loss": 0.2031,
+      "num_input_tokens_seen": 406528296,
+      "step": 2352
+    },
+    {
+      "epoch": 0.900804289544236,
+      "loss": 0.2028670758008957,
+      "loss_ce": 0.01994471065700054,
+      "loss_iou": 1.007051706314087,
+      "loss_num": 0.1826171875,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 406528296,
+      "step": 2352
+    },
+    {
+      "epoch": 0.9011872845653006,
+      "grad_norm": 21.73740240986204,
+      "learning_rate": 5e-06,
+      "loss": 0.1929,
+      "num_input_tokens_seen": 406701456,
+      "step": 2353
+    },
+    {
+      "epoch": 0.9011872845653006,
+      "loss": 0.16568265855312347,
+      "loss_ce": 0.02170073240995407,
+      "loss_iou": 1.048267126083374,
+      "loss_num": 0.1435546875,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 406701456,
+      "step": 2353
+    },
+    {
+      "epoch": 0.9015702795863654,
+      "grad_norm": 22.73960217988902,
+      "learning_rate": 5e-06,
+      "loss": 0.1836,
+      "num_input_tokens_seen": 406874688,
+      "step": 2354
+    },
+    {
+      "epoch": 0.9015702795863654,
+      "loss": 0.18578344583511353,
+      "loss_ce": 0.02214818075299263,
+      "loss_iou": 1.0128567218780518,
+      "loss_num": 0.1640625,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 406874688,
+      "step": 2354
+    },
+    {
+      "epoch": 0.9019532746074301,
+      "grad_norm": 29.40987383685999,
+      "learning_rate": 5e-06,
+      "loss": 0.1872,
+      "num_input_tokens_seen": 407047728,
+      "step": 2355
+    },
+    {
+      "epoch": 0.9019532746074301,
+      "loss": 0.1829497069120407,
+      "loss_ce": 0.02132861688733101,
+      "loss_iou": 1.0231342315673828,
+      "loss_num": 0.162109375,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 407047728,
+      "step": 2355
+    },
+    {
+      "epoch": 0.9023362696284948,
+      "grad_norm": 40.44204858238806,
+      "learning_rate": 5e-06,
+      "loss": 0.235,
+      "num_input_tokens_seen": 407220592,
+      "step": 2356
+    },
+    {
+      "epoch": 0.9023362696284948,
+      "loss": 0.2262561023235321,
+      "loss_ce": 0.0208728164434433,
+      "loss_iou": 1.0273866653442383,
+      "loss_num": 0.205078125,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 407220592,
+      "step": 2356
+    },
+    {
+      "epoch": 0.9027192646495595,
+      "grad_norm": 24.87077660065261,
+      "learning_rate": 5e-06,
+      "loss": 0.1873,
+      "num_input_tokens_seen": 407393568,
+      "step": 2357
+    },
+    {
+      "epoch": 0.9027192646495595,
+      "loss": 0.19813472032546997,
+      "loss_ce": 0.022719688713550568,
+      "loss_iou": 1.0532939434051514,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 407393568,
+      "step": 2357
+    },
+    {
+      "epoch": 0.9031022596706243,
+      "grad_norm": 22.827788690630538,
+      "learning_rate": 5e-06,
+      "loss": 0.2113,
+      "num_input_tokens_seen": 407566344,
+      "step": 2358
+    },
+    {
+      "epoch": 0.9031022596706243,
+      "loss": 0.18151769042015076,
+      "loss_ce": 0.021544544026255608,
+      "loss_iou": 1.0150020122528076,
+      "loss_num": 0.16015625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 407566344,
+      "step": 2358
+    },
+    {
+      "epoch": 0.903485254691689,
+      "grad_norm": 22.379163306053044,
+      "learning_rate": 5e-06,
+      "loss": 0.2535,
+      "num_input_tokens_seen": 407738880,
+      "step": 2359
+    },
+    {
+      "epoch": 0.903485254691689,
+      "loss": 0.32208019495010376,
+      "loss_ce": 0.021298956125974655,
+      "loss_iou": 10.810672760009766,
+      "loss_num": 0.30078125,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 407738880,
+      "step": 2359
+    },
+    {
+      "epoch": 0.9038682497127537,
+      "grad_norm": 25.666087321083204,
+      "learning_rate": 5e-06,
+      "loss": 0.172,
+      "num_input_tokens_seen": 407911744,
+      "step": 2360
+    },
+    {
+      "epoch": 0.9038682497127537,
+      "loss": 0.19929426908493042,
+      "loss_ce": 0.01826399937272072,
+      "loss_iou": 1.1633094549179077,
+      "loss_num": 0.1806640625,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 407911744,
+      "step": 2360
+    },
+    {
+      "epoch": 0.9042512447338185,
+      "grad_norm": 16.089832093242897,
+      "learning_rate": 5e-06,
+      "loss": 0.28,
+      "num_input_tokens_seen": 408084720,
+      "step": 2361
+    },
+    {
+      "epoch": 0.9042512447338185,
+      "loss": 0.30379611253738403,
+      "loss_ce": 0.020470933988690376,
+      "loss_iou": 1.242593765258789,
+      "loss_num": 0.283203125,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 408084720,
+      "step": 2361
+    },
+    {
+      "epoch": 0.9046342397548832,
+      "grad_norm": 18.99963223087458,
+      "learning_rate": 5e-06,
+      "loss": 0.1829,
+      "num_input_tokens_seen": 408257656,
+      "step": 2362
+    },
+    {
+      "epoch": 0.9046342397548832,
+      "loss": 0.186909019947052,
+      "loss_ce": 0.0196116641163826,
+      "loss_iou": 1.025866150856018,
+      "loss_num": 0.1669921875,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 408257656,
+      "step": 2362
+    },
+    {
+      "epoch": 0.9050172347759479,
+      "grad_norm": 16.0476265285785,
+      "learning_rate": 5e-06,
+      "loss": 0.1888,
+      "num_input_tokens_seen": 408431064,
+      "step": 2363
+    },
+    {
+      "epoch": 0.9050172347759479,
+      "loss": 0.20397067070007324,
+      "loss_ce": 0.021414516493678093,
+      "loss_iou": 1.0676087141036987,
+      "loss_num": 0.1826171875,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 408431064,
+      "step": 2363
+    },
+    {
+      "epoch": 0.9054002297970126,
+      "grad_norm": 41.37737189647491,
+      "learning_rate": 5e-06,
+      "loss": 0.2086,
+      "num_input_tokens_seen": 408603920,
+      "step": 2364
+    },
+    {
+      "epoch": 0.9054002297970126,
+      "loss": 0.20587798953056335,
+      "loss_ce": 0.02161286026239395,
+      "loss_iou": 1.01041579246521,
+      "loss_num": 0.1845703125,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 408603920,
+      "step": 2364
+    },
+    {
+      "epoch": 0.9057832248180774,
+      "grad_norm": 35.201733243723055,
+      "learning_rate": 5e-06,
+      "loss": 0.2086,
+      "num_input_tokens_seen": 408776576,
+      "step": 2365
+    },
+    {
+      "epoch": 0.9057832248180774,
+      "loss": 0.22596822679042816,
+      "loss_ce": 0.021927697584033012,
+      "loss_iou": 1.0464751720428467,
+      "loss_num": 0.2041015625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 408776576,
+      "step": 2365
+    },
+    {
+      "epoch": 0.9061662198391421,
+      "grad_norm": 26.22058701727546,
+      "learning_rate": 5e-06,
+      "loss": 0.3156,
+      "num_input_tokens_seen": 408949328,
+      "step": 2366
+    },
+    {
+      "epoch": 0.9061662198391421,
+      "loss": 0.2362404614686966,
+      "loss_ce": 0.021213598549365997,
+      "loss_iou": 1.0060431957244873,
+      "loss_num": 0.21484375,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 408949328,
+      "step": 2366
+    },
+    {
+      "epoch": 0.9065492148602068,
+      "grad_norm": 28.728109473112713,
+      "learning_rate": 5e-06,
+      "loss": 0.1718,
+      "num_input_tokens_seen": 409122152,
+      "step": 2367
+    },
+    {
+      "epoch": 0.9065492148602068,
+      "loss": 0.18759480118751526,
+      "loss_ce": 0.019992249086499214,
+      "loss_iou": 1.0010476112365723,
+      "loss_num": 0.16796875,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 409122152,
+      "step": 2367
+    },
+    {
+      "epoch": 0.9069322098812715,
+      "grad_norm": 22.84123618774518,
+      "learning_rate": 5e-06,
+      "loss": 0.1925,
+      "num_input_tokens_seen": 409295160,
+      "step": 2368
+    },
+    {
+      "epoch": 0.9069322098812715,
+      "loss": 0.16528218984603882,
+      "loss_ce": 0.01965230703353882,
+      "loss_iou": 1.0095326900482178,
+      "loss_num": 0.1455078125,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 409295160,
+      "step": 2368
+    },
+    {
+      "epoch": 0.9073152049023363,
+      "grad_norm": 38.60956499979633,
+      "learning_rate": 5e-06,
+      "loss": 0.2157,
+      "num_input_tokens_seen": 409468280,
+      "step": 2369
+    },
+    {
+      "epoch": 0.9073152049023363,
+      "loss": 0.24069184064865112,
+      "loss_ce": 0.02023286744952202,
+      "loss_iou": 1.0126674175262451,
+      "loss_num": 0.220703125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 409468280,
+      "step": 2369
+    },
+    {
+      "epoch": 0.907698199923401,
+      "grad_norm": 23.961042618032646,
+      "learning_rate": 5e-06,
+      "loss": 0.2062,
+      "num_input_tokens_seen": 409641272,
+      "step": 2370
+    },
+    {
+      "epoch": 0.907698199923401,
+      "loss": 0.18946030735969543,
+      "loss_ce": 0.019965671002864838,
+      "loss_iou": 0.9818347096443176,
+      "loss_num": 0.169921875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 409641272,
+      "step": 2370
+    },
+    {
+      "epoch": 0.9080811949444657,
+      "grad_norm": 48.243077347673,
+      "learning_rate": 5e-06,
+      "loss": 0.1906,
+      "num_input_tokens_seen": 409814064,
+      "step": 2371
+    },
+    {
+      "epoch": 0.9080811949444657,
+      "loss": 0.16905364394187927,
+      "loss_ce": 0.021836843341588974,
+      "loss_iou": 1.0045949220657349,
+      "loss_num": 0.1474609375,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 409814064,
+      "step": 2371
+    },
+    {
+      "epoch": 0.9084641899655305,
+      "grad_norm": 26.146141822474032,
+      "learning_rate": 5e-06,
+      "loss": 0.2315,
+      "num_input_tokens_seen": 409987160,
+      "step": 2372
+    },
+    {
+      "epoch": 0.9084641899655305,
+      "loss": 0.25680333375930786,
+      "loss_ce": 0.022001095116138458,
+      "loss_iou": 1.0284111499786377,
+      "loss_num": 0.234375,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 409987160,
+      "step": 2372
+    },
+    {
+      "epoch": 0.9088471849865952,
+      "grad_norm": 22.502885624134297,
+      "learning_rate": 5e-06,
+      "loss": 0.2325,
+      "num_input_tokens_seen": 410160072,
+      "step": 2373
+    },
+    {
+      "epoch": 0.9088471849865952,
+      "loss": 0.20859001576900482,
+      "loss_ce": 0.02072381228208542,
+      "loss_iou": 1.0815236568450928,
+      "loss_num": 0.1875,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 410160072,
+      "step": 2373
+    },
+    {
+      "epoch": 0.9092301800076599,
+      "grad_norm": 19.128803175523963,
+      "learning_rate": 5e-06,
+      "loss": 0.1617,
+      "num_input_tokens_seen": 410332584,
+      "step": 2374
+    },
+    {
+      "epoch": 0.9092301800076599,
+      "loss": 0.190344899892807,
+      "loss_ce": 0.020239923149347305,
+      "loss_iou": 1.0056294202804565,
+      "loss_num": 0.169921875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 410332584,
+      "step": 2374
+    },
+    {
+      "epoch": 0.9096131750287246,
+      "grad_norm": 25.13174196757647,
+      "learning_rate": 5e-06,
+      "loss": 0.1785,
+      "num_input_tokens_seen": 410505744,
+      "step": 2375
+    },
+    {
+      "epoch": 0.9096131750287246,
+      "loss": 0.16508866846561432,
+      "loss_ce": 0.022083301097154617,
+      "loss_iou": 1.0263667106628418,
+      "loss_num": 0.142578125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 410505744,
+      "step": 2375
+    },
+    {
+      "epoch": 0.9099961700497894,
+      "grad_norm": 22.44469459578855,
+      "learning_rate": 5e-06,
+      "loss": 0.2237,
+      "num_input_tokens_seen": 410678744,
+      "step": 2376
+    },
+    {
+      "epoch": 0.9099961700497894,
+      "loss": 0.23425155878067017,
+      "loss_ce": 0.02056746929883957,
+      "loss_iou": 1.0533339977264404,
+      "loss_num": 0.2138671875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 410678744,
+      "step": 2376
+    },
+    {
+      "epoch": 0.910379165070854,
+      "grad_norm": 25.35229109462942,
+      "learning_rate": 5e-06,
+      "loss": 0.224,
+      "num_input_tokens_seen": 410851328,
+      "step": 2377
+    },
+    {
+      "epoch": 0.910379165070854,
+      "loss": 0.222754567861557,
+      "loss_ce": 0.019995778799057007,
+      "loss_iou": 1.049330472946167,
+      "loss_num": 0.203125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 410851328,
+      "step": 2377
+    },
+    {
+      "epoch": 0.9107621600919188,
+      "grad_norm": 21.17120095168434,
+      "learning_rate": 5e-06,
+      "loss": 0.1934,
+      "num_input_tokens_seen": 411024296,
+      "step": 2378
+    },
+    {
+      "epoch": 0.9107621600919188,
+      "loss": 0.14764659106731415,
+      "loss_ce": 0.020937608554959297,
+      "loss_iou": 1.0029979944229126,
+      "loss_num": 0.126953125,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 411024296,
+      "step": 2378
+    },
+    {
+      "epoch": 0.9111451551129836,
+      "grad_norm": 36.91029657137582,
+      "learning_rate": 5e-06,
+      "loss": 0.1684,
+      "num_input_tokens_seen": 411197248,
+      "step": 2379
+    },
+    {
+      "epoch": 0.9111451551129836,
+      "loss": 0.144285649061203,
+      "loss_ce": 0.021604998037219048,
+      "loss_iou": 1.0007081031799316,
+      "loss_num": 0.12255859375,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 411197248,
+      "step": 2379
+    },
+    {
+      "epoch": 0.9115281501340483,
+      "grad_norm": 27.327345254548007,
+      "learning_rate": 5e-06,
+      "loss": 0.2386,
+      "num_input_tokens_seen": 411370328,
+      "step": 2380
+    },
+    {
+      "epoch": 0.9115281501340483,
+      "loss": 0.2468884289264679,
+      "loss_ce": 0.01947145164012909,
+      "loss_iou": 1.0034431219100952,
+      "loss_num": 0.2275390625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 411370328,
+      "step": 2380
+    },
+    {
+      "epoch": 0.911911145155113,
+      "grad_norm": 34.977127587219535,
+      "learning_rate": 5e-06,
+      "loss": 0.2414,
+      "num_input_tokens_seen": 411543456,
+      "step": 2381
+    },
+    {
+      "epoch": 0.911911145155113,
+      "loss": 0.21146827936172485,
+      "loss_ce": 0.020672377198934555,
+      "loss_iou": 1.002312183380127,
+      "loss_num": 0.1904296875,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 411543456,
+      "step": 2381
+    },
+    {
+      "epoch": 0.9122941401761777,
+      "grad_norm": 15.96437352384054,
+      "learning_rate": 5e-06,
+      "loss": 0.1853,
+      "num_input_tokens_seen": 411716576,
+      "step": 2382
+    },
+    {
+      "epoch": 0.9122941401761777,
+      "loss": 0.20538511872291565,
+      "loss_ce": 0.021943964064121246,
+      "loss_iou": 1.0280776023864746,
+      "loss_num": 0.18359375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 411716576,
+      "step": 2382
+    },
+    {
+      "epoch": 0.9126771351972425,
+      "grad_norm": 35.51237483478961,
+      "learning_rate": 5e-06,
+      "loss": 0.2434,
+      "num_input_tokens_seen": 411889760,
+      "step": 2383
+    },
+    {
+      "epoch": 0.9126771351972425,
+      "loss": 0.1973738968372345,
+      "loss_ce": 0.021104369312524796,
+      "loss_iou": 1.0060076713562012,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 411889760,
+      "step": 2383
+    },
+    {
+      "epoch": 0.9130601302183071,
+      "grad_norm": 25.647787431160904,
+      "learning_rate": 5e-06,
+      "loss": 0.2108,
+      "num_input_tokens_seen": 412062928,
+      "step": 2384
+    },
+    {
+      "epoch": 0.9130601302183071,
+      "loss": 0.17348915338516235,
+      "loss_ce": 0.021877823397517204,
+      "loss_iou": 1.005397081375122,
+      "loss_num": 0.1513671875,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 412062928,
+      "step": 2384
+    },
+    {
+      "epoch": 0.9134431252393719,
+      "grad_norm": 23.053344913429793,
+      "learning_rate": 5e-06,
+      "loss": 0.2619,
+      "num_input_tokens_seen": 412235584,
+      "step": 2385
+    },
+    {
+      "epoch": 0.9134431252393719,
+      "loss": 0.22121822834014893,
+      "loss_ce": 0.02090083621442318,
+      "loss_iou": 1.191190242767334,
+      "loss_num": 0.2001953125,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 412235584,
+      "step": 2385
+    },
+    {
+      "epoch": 0.9138261202604366,
+      "grad_norm": 26.292265105545425,
+      "learning_rate": 5e-06,
+      "loss": 0.2143,
+      "num_input_tokens_seen": 412408272,
+      "step": 2386
+    },
+    {
+      "epoch": 0.9138261202604366,
+      "loss": 0.1930234730243683,
+      "loss_ce": 0.018768098205327988,
+      "loss_iou": 1.0049376487731934,
+      "loss_num": 0.173828125,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 412408272,
+      "step": 2386
+    },
+    {
+      "epoch": 0.9142091152815014,
+      "grad_norm": 24.98979994341704,
+      "learning_rate": 5e-06,
+      "loss": 0.1405,
+      "num_input_tokens_seen": 412581104,
+      "step": 2387
+    },
+    {
+      "epoch": 0.9142091152815014,
+      "loss": 0.15191525220870972,
+      "loss_ce": 0.020811742171645164,
+      "loss_iou": 1.0201034545898438,
+      "loss_num": 0.130859375,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 412581104,
+      "step": 2387
+    },
+    {
+      "epoch": 0.914592110302566,
+      "grad_norm": 23.354223453219895,
+      "learning_rate": 5e-06,
+      "loss": 0.2023,
+      "num_input_tokens_seen": 412754104,
+      "step": 2388
+    },
+    {
+      "epoch": 0.914592110302566,
+      "loss": 0.16768714785575867,
+      "loss_ce": 0.020897595211863518,
+      "loss_iou": 1.000795602798462,
+      "loss_num": 0.146484375,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 412754104,
+      "step": 2388
+    },
+    {
+      "epoch": 0.9149751053236308,
+      "grad_norm": 42.04023037420445,
+      "learning_rate": 5e-06,
+      "loss": 0.1797,
+      "num_input_tokens_seen": 412927000,
+      "step": 2389
+    },
+    {
+      "epoch": 0.9149751053236308,
+      "loss": 0.1643044352531433,
+      "loss_ce": 0.020688720047473907,
+      "loss_iou": 1.00919508934021,
+      "loss_num": 0.1435546875,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 412927000,
+      "step": 2389
+    },
+    {
+      "epoch": 0.9153581003446956,
+      "grad_norm": 38.683829128169535,
+      "learning_rate": 5e-06,
+      "loss": 0.2661,
+      "num_input_tokens_seen": 413099632,
+      "step": 2390
+    },
+    {
+      "epoch": 0.9153581003446956,
+      "loss": 0.25069767236709595,
+      "loss_ce": 0.021632760763168335,
+      "loss_iou": 1.0596530437469482,
+      "loss_num": 0.2294921875,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 413099632,
+      "step": 2390
+    },
+    {
+      "epoch": 0.9157410953657602,
+      "grad_norm": 17.870399262431235,
+      "learning_rate": 5e-06,
+      "loss": 0.1684,
+      "num_input_tokens_seen": 413272616,
+      "step": 2391
+    },
+    {
+      "epoch": 0.9157410953657602,
+      "loss": 0.1694970577955246,
+      "loss_ce": 0.023409409448504448,
+      "loss_iou": 1.0341999530792236,
+      "loss_num": 0.146484375,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 413272616,
+      "step": 2391
+    },
+    {
+      "epoch": 0.916124090386825,
+      "grad_norm": 83.70239355912715,
+      "learning_rate": 5e-06,
+      "loss": 0.2372,
+      "num_input_tokens_seen": 413445360,
+      "step": 2392
+    },
+    {
+      "epoch": 0.916124090386825,
+      "loss": 0.23829631507396698,
+      "loss_ce": 0.019668392837047577,
+      "loss_iou": 1.054909586906433,
+      "loss_num": 0.21875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 413445360,
+      "step": 2392
+    },
+    {
+      "epoch": 0.9165070854078897,
+      "grad_norm": 42.05504986560037,
+      "learning_rate": 5e-06,
+      "loss": 0.2355,
+      "num_input_tokens_seen": 413618760,
+      "step": 2393
+    },
+    {
+      "epoch": 0.9165070854078897,
+      "loss": 0.22386720776557922,
+      "loss_ce": 0.021657757461071014,
+      "loss_iou": 1.0026435852050781,
+      "loss_num": 0.2021484375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 413618760,
+      "step": 2393
+    },
+    {
+      "epoch": 0.9168900804289544,
+      "grad_norm": 37.73854328490997,
+      "learning_rate": 5e-06,
+      "loss": 0.2432,
+      "num_input_tokens_seen": 413791664,
+      "step": 2394
+    },
+    {
+      "epoch": 0.9168900804289544,
+      "loss": 0.22379669547080994,
+      "loss_ce": 0.02054961770772934,
+      "loss_iou": 1.001222848892212,
+      "loss_num": 0.203125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 413791664,
+      "step": 2394
+    },
+    {
+      "epoch": 0.9172730754500191,
+      "grad_norm": 18.763365736694332,
+      "learning_rate": 5e-06,
+      "loss": 0.1778,
+      "num_input_tokens_seen": 413964616,
+      "step": 2395
+    },
+    {
+      "epoch": 0.9172730754500191,
+      "loss": 0.20379221439361572,
+      "loss_ce": 0.020869851112365723,
+      "loss_iou": 1.0439505577087402,
+      "loss_num": 0.1826171875,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 413964616,
+      "step": 2395
+    },
+    {
+      "epoch": 0.9176560704710839,
+      "grad_norm": 21.442752576039208,
+      "learning_rate": 5e-06,
+      "loss": 0.1763,
+      "num_input_tokens_seen": 414138008,
+      "step": 2396
+    },
+    {
+      "epoch": 0.9176560704710839,
+      "loss": 0.18055325746536255,
+      "loss_ce": 0.0207021776586771,
+      "loss_iou": 1.007684350013733,
+      "loss_num": 0.16015625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 414138008,
+      "step": 2396
+    },
+    {
+      "epoch": 0.9180390654921486,
+      "grad_norm": 50.31807377112917,
+      "learning_rate": 5e-06,
+      "loss": 0.2828,
+      "num_input_tokens_seen": 414310904,
+      "step": 2397
+    },
+    {
+      "epoch": 0.9180390654921486,
+      "loss": 0.26484301686286926,
+      "loss_ce": 0.020580342039465904,
+      "loss_iou": 1.010157585144043,
+      "loss_num": 0.244140625,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 414310904,
+      "step": 2397
+    },
+    {
+      "epoch": 0.9184220605132133,
+      "grad_norm": 48.38373924778628,
+      "learning_rate": 5e-06,
+      "loss": 0.2294,
+      "num_input_tokens_seen": 414483704,
+      "step": 2398
+    },
+    {
+      "epoch": 0.9184220605132133,
+      "loss": 0.18663525581359863,
+      "loss_ce": 0.020619627088308334,
+      "loss_iou": 1.0017452239990234,
+      "loss_num": 0.166015625,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 414483704,
+      "step": 2398
+    },
+    {
+      "epoch": 0.918805055534278,
+      "grad_norm": 21.248214249983533,
+      "learning_rate": 5e-06,
+      "loss": 0.2166,
+      "num_input_tokens_seen": 414656816,
+      "step": 2399
+    },
+    {
+      "epoch": 0.918805055534278,
+      "loss": 0.2035892903804779,
+      "loss_ce": 0.021460391581058502,
+      "loss_iou": 1.0131632089614868,
+      "loss_num": 0.181640625,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 414656816,
+      "step": 2399
+    },
+    {
+      "epoch": 0.9191880505553428,
+      "grad_norm": 31.826929765619607,
+      "learning_rate": 5e-06,
+      "loss": 0.3054,
+      "num_input_tokens_seen": 414829968,
+      "step": 2400
+    },
+    {
+      "epoch": 0.9191880505553428,
+      "loss": 0.32269665598869324,
+      "loss_ce": 0.02069469541311264,
+      "loss_iou": 1.033881425857544,
+      "loss_num": 0.302734375,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 414829968,
+      "step": 2400
+    },
+    {
+      "epoch": 0.9195710455764075,
+      "grad_norm": 36.31243766548263,
+      "learning_rate": 5e-06,
+      "loss": 0.2791,
+      "num_input_tokens_seen": 415003472,
+      "step": 2401
+    },
+    {
+      "epoch": 0.9195710455764075,
+      "loss": 0.1605401337146759,
+      "loss_ce": 0.022234464064240456,
+      "loss_iou": 1.016472339630127,
+      "loss_num": 0.138671875,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 415003472,
+      "step": 2401
+    },
+    {
+      "epoch": 0.9199540405974722,
+      "grad_norm": 52.23825924030564,
+      "learning_rate": 5e-06,
+      "loss": 0.2401,
+      "num_input_tokens_seen": 415176336,
+      "step": 2402
+    },
+    {
+      "epoch": 0.9199540405974722,
+      "loss": 0.24016940593719482,
+      "loss_ce": 0.020320788025856018,
+      "loss_iou": 1.043022632598877,
+      "loss_num": 0.2197265625,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 415176336,
+      "step": 2402
+    },
+    {
+      "epoch": 0.920337035618537,
+      "grad_norm": 34.65960734586794,
+      "learning_rate": 5e-06,
+      "loss": 0.1916,
+      "num_input_tokens_seen": 415349016,
+      "step": 2403
+    },
+    {
+      "epoch": 0.920337035618537,
+      "loss": 0.186640664935112,
+      "loss_ce": 0.022578159347176552,
+      "loss_iou": 1.025832176208496,
+      "loss_num": 0.1640625,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 415349016,
+      "step": 2403
+    },
+    {
+      "epoch": 0.9207200306396017,
+      "grad_norm": 33.46523262659949,
+      "learning_rate": 5e-06,
+      "loss": 0.2845,
+      "num_input_tokens_seen": 415521760,
+      "step": 2404
+    },
+    {
+      "epoch": 0.9207200306396017,
+      "loss": 0.25068753957748413,
+      "loss_ce": 0.02180570550262928,
+      "loss_iou": 1.0948412418365479,
+      "loss_num": 0.228515625,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 415521760,
+      "step": 2404
+    },
+    {
+      "epoch": 0.9211030256606664,
+      "grad_norm": 20.460913905721025,
+      "learning_rate": 5e-06,
+      "loss": 0.1433,
+      "num_input_tokens_seen": 415694848,
+      "step": 2405
+    },
+    {
+      "epoch": 0.9211030256606664,
+      "loss": 0.14977923035621643,
+      "loss_ce": 0.02184954844415188,
+      "loss_iou": 1.0027718544006348,
+      "loss_num": 0.1279296875,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 415694848,
+      "step": 2405
+    },
+    {
+      "epoch": 0.9214860206817311,
+      "grad_norm": 50.72670269751362,
+      "learning_rate": 5e-06,
+      "loss": 0.1744,
+      "num_input_tokens_seen": 415867912,
+      "step": 2406
+    },
+    {
+      "epoch": 0.9214860206817311,
+      "loss": 0.19672946631908417,
+      "loss_ce": 0.021985813975334167,
+      "loss_iou": 1.00660240650177,
+      "loss_num": 0.1748046875,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 415867912,
+      "step": 2406
+    },
+    {
+      "epoch": 0.9218690157027959,
+      "grad_norm": 29.623117232777613,
+      "learning_rate": 5e-06,
+      "loss": 0.2317,
+      "num_input_tokens_seen": 416040912,
+      "step": 2407
+    },
+    {
+      "epoch": 0.9218690157027959,
+      "loss": 0.14956334233283997,
+      "loss_ce": 0.021755734458565712,
+      "loss_iou": 1.0036413669586182,
+      "loss_num": 0.1279296875,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 416040912,
+      "step": 2407
+    },
+    {
+      "epoch": 0.9222520107238605,
+      "grad_norm": 35.23146260409525,
+      "learning_rate": 5e-06,
+      "loss": 0.2066,
+      "num_input_tokens_seen": 416213568,
+      "step": 2408
+    },
+    {
+      "epoch": 0.9222520107238605,
+      "loss": 0.17996053397655487,
+      "loss_ce": 0.02139120362699032,
+      "loss_iou": 1.0081908702850342,
+      "loss_num": 0.158203125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 416213568,
+      "step": 2408
+    },
+    {
+      "epoch": 0.9226350057449253,
+      "grad_norm": 66.48201776577294,
+      "learning_rate": 5e-06,
+      "loss": 0.239,
+      "num_input_tokens_seen": 416386272,
+      "step": 2409
+    },
+    {
+      "epoch": 0.9226350057449253,
+      "loss": 0.2576059401035309,
+      "loss_ce": 0.021521952003240585,
+      "loss_iou": 1.1739799976348877,
+      "loss_num": 0.236328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 416386272,
+      "step": 2409
+    },
+    {
+      "epoch": 0.92301800076599,
+      "grad_norm": 32.69023473660693,
+      "learning_rate": 5e-06,
+      "loss": 0.2489,
+      "num_input_tokens_seen": 416559384,
+      "step": 2410
+    },
+    {
+      "epoch": 0.92301800076599,
+      "loss": 0.19133315980434418,
+      "loss_ce": 0.01982436515390873,
+      "loss_iou": 1.0112589597702026,
+      "loss_num": 0.171875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 416559384,
+      "step": 2410
+    },
+    {
+      "epoch": 0.9234009957870548,
+      "grad_norm": 24.341024657896455,
+      "learning_rate": 5e-06,
+      "loss": 0.2366,
+      "num_input_tokens_seen": 416732424,
+      "step": 2411
+    },
+    {
+      "epoch": 0.9234009957870548,
+      "loss": 0.21062758564949036,
+      "loss_ce": 0.020319968461990356,
+      "loss_iou": 1.11098051071167,
+      "loss_num": 0.1904296875,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 416732424,
+      "step": 2411
+    },
+    {
+      "epoch": 0.9237839908081195,
+      "grad_norm": 23.874798414372748,
+      "learning_rate": 5e-06,
+      "loss": 0.2051,
+      "num_input_tokens_seen": 416905416,
+      "step": 2412
+    },
+    {
+      "epoch": 0.9237839908081195,
+      "loss": 0.17523883283138275,
+      "loss_ce": 0.020392633974552155,
+      "loss_iou": 1.0309886932373047,
+      "loss_num": 0.1552734375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 416905416,
+      "step": 2412
+    },
+    {
+      "epoch": 0.9241669858291842,
+      "grad_norm": 24.93969323539543,
+      "learning_rate": 5e-06,
+      "loss": 0.1812,
+      "num_input_tokens_seen": 417078432,
+      "step": 2413
+    },
+    {
+      "epoch": 0.9241669858291842,
+      "loss": 0.16211159527301788,
+      "loss_ce": 0.021120378747582436,
+      "loss_iou": 1.0268622636795044,
+      "loss_num": 0.140625,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 417078432,
+      "step": 2413
+    },
+    {
+      "epoch": 0.924549980850249,
+      "grad_norm": 35.228836376996796,
+      "learning_rate": 5e-06,
+      "loss": 0.2267,
+      "num_input_tokens_seen": 417251472,
+      "step": 2414
+    },
+    {
+      "epoch": 0.924549980850249,
+      "loss": 0.23107093572616577,
+      "loss_ce": 0.02049964852631092,
+      "loss_iou": 1.0622409582138062,
+      "loss_num": 0.2109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 417251472,
+      "step": 2414
+    },
+    {
+      "epoch": 0.9249329758713136,
+      "grad_norm": 24.71475558465295,
+      "learning_rate": 5e-06,
+      "loss": 0.2123,
+      "num_input_tokens_seen": 417424664,
+      "step": 2415
+    },
+    {
+      "epoch": 0.9249329758713136,
+      "loss": 0.17129939794540405,
+      "loss_ce": 0.02045101299881935,
+      "loss_iou": 1.0016270875930786,
+      "loss_num": 0.150390625,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 417424664,
+      "step": 2415
+    },
+    {
+      "epoch": 0.9253159708923784,
+      "grad_norm": 34.31229272225131,
+      "learning_rate": 5e-06,
+      "loss": 0.2047,
+      "num_input_tokens_seen": 417597624,
+      "step": 2416
+    },
+    {
+      "epoch": 0.9253159708923784,
+      "loss": 0.20795701444149017,
+      "loss_ce": 0.021494608372449875,
+      "loss_iou": 1.0274615287780762,
+      "loss_num": 0.1865234375,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 417597624,
+      "step": 2416
+    },
+    {
+      "epoch": 0.9256989659134431,
+      "grad_norm": 25.928366534167534,
+      "learning_rate": 5e-06,
+      "loss": 0.2272,
+      "num_input_tokens_seen": 417770784,
+      "step": 2417
+    },
+    {
+      "epoch": 0.9256989659134431,
+      "loss": 0.23110869526863098,
+      "loss_ce": 0.021636024117469788,
+      "loss_iou": 1.0217278003692627,
+      "loss_num": 0.208984375,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 417770784,
+      "step": 2417
+    },
+    {
+      "epoch": 0.9260819609345079,
+      "grad_norm": 51.19708150594874,
+      "learning_rate": 5e-06,
+      "loss": 0.2705,
+      "num_input_tokens_seen": 417944048,
+      "step": 2418
+    },
+    {
+      "epoch": 0.9260819609345079,
+      "loss": 0.2860315144062042,
+      "loss_ce": 0.02101687341928482,
+      "loss_iou": 1.1807692050933838,
+      "loss_num": 0.265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 417944048,
+      "step": 2418
+    },
+    {
+      "epoch": 0.9264649559555725,
+      "grad_norm": 38.28563578186094,
+      "learning_rate": 5e-06,
+      "loss": 0.2517,
+      "num_input_tokens_seen": 418116832,
+      "step": 2419
+    },
+    {
+      "epoch": 0.9264649559555725,
+      "loss": 0.21095630526542664,
+      "loss_ce": 0.020282475277781487,
+      "loss_iou": 1.0341272354125977,
+      "loss_num": 0.1904296875,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 418116832,
+      "step": 2419
+    },
+    {
+      "epoch": 0.9268479509766373,
+      "grad_norm": 36.43215353743023,
+      "learning_rate": 5e-06,
+      "loss": 0.2319,
+      "num_input_tokens_seen": 418289800,
+      "step": 2420
+    },
+    {
+      "epoch": 0.9268479509766373,
+      "loss": 0.23096629977226257,
+      "loss_ce": 0.020028796046972275,
+      "loss_iou": 1.0649144649505615,
+      "loss_num": 0.2109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 418289800,
+      "step": 2420
+    },
+    {
+      "epoch": 0.9272309459977021,
+      "grad_norm": 32.00124685256224,
+      "learning_rate": 5e-06,
+      "loss": 0.1744,
+      "num_input_tokens_seen": 418462696,
+      "step": 2421
+    },
+    {
+      "epoch": 0.9272309459977021,
+      "loss": 0.18664014339447021,
+      "loss_ce": 0.02172316424548626,
+      "loss_iou": 1.004173755645752,
+      "loss_num": 0.1650390625,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 418462696,
+      "step": 2421
+    },
+    {
+      "epoch": 0.9276139410187667,
+      "grad_norm": 57.02463023651104,
+      "learning_rate": 5e-06,
+      "loss": 0.2429,
+      "num_input_tokens_seen": 418635648,
+      "step": 2422
+    },
+    {
+      "epoch": 0.9276139410187667,
+      "loss": 0.23585951328277588,
+      "loss_ce": 0.02174818143248558,
+      "loss_iou": 1.022843360900879,
+      "loss_num": 0.2138671875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 418635648,
+      "step": 2422
+    },
+    {
+      "epoch": 0.9279969360398315,
+      "grad_norm": 28.207117397336745,
+      "learning_rate": 5e-06,
+      "loss": 0.2404,
+      "num_input_tokens_seen": 418808968,
+      "step": 2423
+    },
+    {
+      "epoch": 0.9279969360398315,
+      "loss": 0.26323407888412476,
+      "loss_ce": 0.02184002473950386,
+      "loss_iou": 1.1213366985321045,
+      "loss_num": 0.2412109375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 418808968,
+      "step": 2423
+    },
+    {
+      "epoch": 0.9283799310608962,
+      "grad_norm": 44.109485209680216,
+      "learning_rate": 5e-06,
+      "loss": 0.2911,
+      "num_input_tokens_seen": 418982176,
+      "step": 2424
+    },
+    {
+      "epoch": 0.9283799310608962,
+      "loss": 0.25499510765075684,
+      "loss_ce": 0.019460441544651985,
+      "loss_iou": 1.2126247882843018,
+      "loss_num": 0.2353515625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 418982176,
+      "step": 2424
+    },
+    {
+      "epoch": 0.928762926081961,
+      "grad_norm": 38.387519998801864,
+      "learning_rate": 5e-06,
+      "loss": 0.2096,
+      "num_input_tokens_seen": 419155328,
+      "step": 2425
+    },
+    {
+      "epoch": 0.928762926081961,
+      "loss": 0.19838780164718628,
+      "loss_ce": 0.021996185183525085,
+      "loss_iou": 1.0736583471298218,
+      "loss_num": 0.1767578125,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 419155328,
+      "step": 2425
+    },
+    {
+      "epoch": 0.9291459211030256,
+      "grad_norm": 28.7370779052248,
+      "learning_rate": 5e-06,
+      "loss": 0.1685,
+      "num_input_tokens_seen": 419328512,
+      "step": 2426
+    },
+    {
+      "epoch": 0.9291459211030256,
+      "loss": 0.17336629331111908,
+      "loss_ce": 0.02138875052332878,
+      "loss_iou": 0.9574381113052368,
+      "loss_num": 0.15234375,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 419328512,
+      "step": 2426
+    },
+    {
+      "epoch": 0.9295289161240904,
+      "grad_norm": 31.995777644262525,
+      "learning_rate": 5e-06,
+      "loss": 0.2352,
+      "num_input_tokens_seen": 419501688,
+      "step": 2427
+    },
+    {
+      "epoch": 0.9295289161240904,
+      "loss": 0.24376463890075684,
+      "loss_ce": 0.021840814501047134,
+      "loss_iou": 1.0403281450271606,
+      "loss_num": 0.2216796875,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 419501688,
+      "step": 2427
+    },
+    {
+      "epoch": 0.9299119111451551,
+      "grad_norm": 43.915227363341806,
+      "learning_rate": 5e-06,
+      "loss": 0.1724,
+      "num_input_tokens_seen": 419674408,
+      "step": 2428
+    },
+    {
+      "epoch": 0.9299119111451551,
+      "loss": 0.16767174005508423,
+      "loss_ce": 0.019753042608499527,
+      "loss_iou": 1.0534336566925049,
+      "loss_num": 0.1474609375,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 419674408,
+      "step": 2428
+    },
+    {
+      "epoch": 0.9302949061662198,
+      "grad_norm": 34.99340342966463,
+      "learning_rate": 5e-06,
+      "loss": 0.2348,
+      "num_input_tokens_seen": 419847200,
+      "step": 2429
+    },
+    {
+      "epoch": 0.9302949061662198,
+      "loss": 0.24999958276748657,
+      "loss_ce": 0.020019114017486572,
+      "loss_iou": 1.1393516063690186,
+      "loss_num": 0.23046875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 419847200,
+      "step": 2429
+    },
+    {
+      "epoch": 0.9306779011872846,
+      "grad_norm": 27.957835315109037,
+      "learning_rate": 5e-06,
+      "loss": 0.2015,
+      "num_input_tokens_seen": 420020184,
+      "step": 2430
+    },
+    {
+      "epoch": 0.9306779011872846,
+      "loss": 0.2168235331773758,
+      "loss_ce": 0.021633097901940346,
+      "loss_iou": 1.0590548515319824,
+      "loss_num": 0.1953125,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 420020184,
+      "step": 2430
+    },
+    {
+      "epoch": 0.9310608962083493,
+      "grad_norm": 30.479759737812316,
+      "learning_rate": 5e-06,
+      "loss": 0.1696,
+      "num_input_tokens_seen": 420192984,
+      "step": 2431
+    },
+    {
+      "epoch": 0.9310608962083493,
+      "loss": 0.2150260955095291,
+      "loss_ce": 0.01977463811635971,
+      "loss_iou": 1.0108896493911743,
+      "loss_num": 0.1953125,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 420192984,
+      "step": 2431
+    },
+    {
+      "epoch": 0.9314438912294141,
+      "grad_norm": 29.95663225231501,
+      "learning_rate": 5e-06,
+      "loss": 0.2143,
+      "num_input_tokens_seen": 420365680,
+      "step": 2432
+    },
+    {
+      "epoch": 0.9314438912294141,
+      "loss": 0.18899086117744446,
+      "loss_ce": 0.019984519109129906,
+      "loss_iou": 1.0001256465911865,
+      "loss_num": 0.1689453125,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 420365680,
+      "step": 2432
+    },
+    {
+      "epoch": 0.9318268862504787,
+      "grad_norm": 28.49949267539661,
+      "learning_rate": 5e-06,
+      "loss": 0.1758,
+      "num_input_tokens_seen": 420538536,
+      "step": 2433
+    },
+    {
+      "epoch": 0.9318268862504787,
+      "loss": 0.1715395152568817,
+      "loss_ce": 0.02011130005121231,
+      "loss_iou": 1.0529817342758179,
+      "loss_num": 0.1513671875,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 420538536,
+      "step": 2433
+    },
+    {
+      "epoch": 0.9322098812715435,
+      "grad_norm": 23.356378197389986,
+      "learning_rate": 5e-06,
+      "loss": 0.2315,
+      "num_input_tokens_seen": 420711392,
+      "step": 2434
+    },
+    {
+      "epoch": 0.9322098812715435,
+      "loss": 0.2523745894432068,
+      "loss_ce": 0.02257724106311798,
+      "loss_iou": 1.0776373147964478,
+      "loss_num": 0.2294921875,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 420711392,
+      "step": 2434
+    },
+    {
+      "epoch": 0.9325928762926082,
+      "grad_norm": 27.878087081766314,
+      "learning_rate": 5e-06,
+      "loss": 0.1885,
+      "num_input_tokens_seen": 420883976,
+      "step": 2435
+    },
+    {
+      "epoch": 0.9325928762926082,
+      "loss": 0.16960689425468445,
+      "loss_ce": 0.022023893892765045,
+      "loss_iou": 90.7451171875,
+      "loss_num": 0.1474609375,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 420883976,
+      "step": 2435
+    },
+    {
+      "epoch": 0.9329758713136729,
+      "grad_norm": 21.591229060591598,
+      "learning_rate": 5e-06,
+      "loss": 0.1899,
+      "num_input_tokens_seen": 421056688,
+      "step": 2436
+    },
+    {
+      "epoch": 0.9329758713136729,
+      "loss": 0.1906215250492096,
+      "loss_ce": 0.020638613030314445,
+      "loss_iou": 1.0376107692718506,
+      "loss_num": 0.169921875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 421056688,
+      "step": 2436
+    },
+    {
+      "epoch": 0.9333588663347376,
+      "grad_norm": 27.294068486612392,
+      "learning_rate": 5e-06,
+      "loss": 0.1863,
+      "num_input_tokens_seen": 421229760,
+      "step": 2437
+    },
+    {
+      "epoch": 0.9333588663347376,
+      "loss": 0.18855778872966766,
+      "loss_ce": 0.02214544080197811,
+      "loss_iou": 1.0555882453918457,
+      "loss_num": 0.166015625,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 421229760,
+      "step": 2437
+    },
+    {
+      "epoch": 0.9337418613558024,
+      "grad_norm": 32.587310319264084,
+      "learning_rate": 5e-06,
+      "loss": 0.189,
+      "num_input_tokens_seen": 421402960,
+      "step": 2438
+    },
+    {
+      "epoch": 0.9337418613558024,
+      "loss": 0.19774208962917328,
+      "loss_ce": 0.02147255465388298,
+      "loss_iou": 1.0686249732971191,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 421402960,
+      "step": 2438
+    },
+    {
+      "epoch": 0.9341248563768672,
+      "grad_norm": 24.589879187900202,
+      "learning_rate": 5e-06,
+      "loss": 0.237,
+      "num_input_tokens_seen": 421576272,
+      "step": 2439
+    },
+    {
+      "epoch": 0.9341248563768672,
+      "loss": 0.2968590557575226,
+      "loss_ce": 0.02152945101261139,
+      "loss_iou": 1.2773818969726562,
+      "loss_num": 0.275390625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 421576272,
+      "step": 2439
+    },
+    {
+      "epoch": 0.9345078513979318,
+      "grad_norm": 34.9548196804876,
+      "learning_rate": 5e-06,
+      "loss": 0.2036,
+      "num_input_tokens_seen": 421749328,
+      "step": 2440
+    },
+    {
+      "epoch": 0.9345078513979318,
+      "loss": 0.2245294451713562,
+      "loss_ce": 0.021831698715686798,
+      "loss_iou": 1.032029390335083,
+      "loss_num": 0.203125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 421749328,
+      "step": 2440
+    },
+    {
+      "epoch": 0.9348908464189966,
+      "grad_norm": 28.641058877782275,
+      "learning_rate": 5e-06,
+      "loss": 0.1641,
+      "num_input_tokens_seen": 421922496,
+      "step": 2441
+    },
+    {
+      "epoch": 0.9348908464189966,
+      "loss": 0.1931767463684082,
+      "loss_ce": 0.021057605743408203,
+      "loss_iou": 1.0118610858917236,
+      "loss_num": 0.171875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 421922496,
+      "step": 2441
+    },
+    {
+      "epoch": 0.9352738414400613,
+      "grad_norm": 28.024374780013545,
+      "learning_rate": 5e-06,
+      "loss": 0.2141,
+      "num_input_tokens_seen": 422095488,
+      "step": 2442
+    },
+    {
+      "epoch": 0.9352738414400613,
+      "loss": 0.18159523606300354,
+      "loss_ce": 0.01881447806954384,
+      "loss_iou": 1.083962321281433,
+      "loss_num": 0.1630859375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 422095488,
+      "step": 2442
+    },
+    {
+      "epoch": 0.935656836461126,
+      "grad_norm": 18.739231319484176,
+      "learning_rate": 5e-06,
+      "loss": 0.1999,
+      "num_input_tokens_seen": 422268552,
+      "step": 2443
+    },
+    {
+      "epoch": 0.935656836461126,
+      "loss": 0.21589979529380798,
+      "loss_ce": 0.019702300429344177,
+      "loss_iou": 1.099543571472168,
+      "loss_num": 0.1962890625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 422268552,
+      "step": 2443
+    },
+    {
+      "epoch": 0.9360398314821907,
+      "grad_norm": 42.80397169834053,
+      "learning_rate": 5e-06,
+      "loss": 0.2415,
+      "num_input_tokens_seen": 422441688,
+      "step": 2444
+    },
+    {
+      "epoch": 0.9360398314821907,
+      "loss": 0.2545335292816162,
+      "loss_ce": 0.022660961374640465,
+      "loss_iou": 1.0548175573349,
+      "loss_num": 0.2314453125,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 422441688,
+      "step": 2444
+    },
+    {
+      "epoch": 0.9364228265032555,
+      "grad_norm": 26.968170587405464,
+      "learning_rate": 5e-06,
+      "loss": 0.2625,
+      "num_input_tokens_seen": 422614744,
+      "step": 2445
+    },
+    {
+      "epoch": 0.9364228265032555,
+      "loss": 0.2984175682067871,
+      "loss_ce": 0.0207686685025692,
+      "loss_iou": 1.228839635848999,
+      "loss_num": 0.27734375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 422614744,
+      "step": 2445
+    },
+    {
+      "epoch": 0.9368058215243202,
+      "grad_norm": 21.29776441160989,
+      "learning_rate": 5e-06,
+      "loss": 0.1967,
+      "num_input_tokens_seen": 422787888,
+      "step": 2446
+    },
+    {
+      "epoch": 0.9368058215243202,
+      "loss": 0.15890049934387207,
+      "loss_ce": 0.022547969594597816,
+      "loss_iou": 1.0457978248596191,
+      "loss_num": 0.13671875,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 422787888,
+      "step": 2446
+    },
+    {
+      "epoch": 0.9371888165453849,
+      "grad_norm": 27.680384244658374,
+      "learning_rate": 5e-06,
+      "loss": 0.1658,
+      "num_input_tokens_seen": 422961024,
+      "step": 2447
+    },
+    {
+      "epoch": 0.9371888165453849,
+      "loss": 0.20146608352661133,
+      "loss_ce": 0.021625995635986328,
+      "loss_iou": 1.2909531593322754,
+      "loss_num": 0.1796875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 422961024,
+      "step": 2447
+    },
+    {
+      "epoch": 0.9375718115664496,
+      "grad_norm": 36.61282361352501,
+      "learning_rate": 5e-06,
+      "loss": 0.2415,
+      "num_input_tokens_seen": 423133776,
+      "step": 2448
+    },
+    {
+      "epoch": 0.9375718115664496,
+      "loss": 0.22485896944999695,
+      "loss_ce": 0.021245690062642097,
+      "loss_iou": 1.0258800983428955,
+      "loss_num": 0.203125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 423133776,
+      "step": 2448
+    },
+    {
+      "epoch": 0.9379548065875144,
+      "grad_norm": 26.416596468709095,
+      "learning_rate": 5e-06,
+      "loss": 0.1476,
+      "num_input_tokens_seen": 423306536,
+      "step": 2449
+    },
+    {
+      "epoch": 0.9379548065875144,
+      "loss": 0.14405319094657898,
+      "loss_ce": 0.02247115597128868,
+      "loss_iou": 0.9872841835021973,
+      "loss_num": 0.12158203125,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 423306536,
+      "step": 2449
+    },
+    {
+      "epoch": 0.938337801608579,
+      "grad_norm": 27.007111721030263,
+      "learning_rate": 5e-06,
+      "loss": 0.1971,
+      "num_input_tokens_seen": 423479520,
+      "step": 2450
+    },
+    {
+      "epoch": 0.938337801608579,
+      "loss": 0.2385256439447403,
+      "loss_ce": 0.021728765219449997,
+      "loss_iou": 1.0041799545288086,
+      "loss_num": 0.216796875,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 423479520,
+      "step": 2450
+    },
+    {
+      "epoch": 0.9387207966296438,
+      "grad_norm": 31.69896577460898,
+      "learning_rate": 5e-06,
+      "loss": 0.2496,
+      "num_input_tokens_seen": 423652704,
+      "step": 2451
+    },
+    {
+      "epoch": 0.9387207966296438,
+      "loss": 0.21565771102905273,
+      "loss_ce": 0.021749012172222137,
+      "loss_iou": 1.0183502435684204,
+      "loss_num": 0.1943359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 423652704,
+      "step": 2451
+    },
+    {
+      "epoch": 0.9391037916507086,
+      "grad_norm": 31.989382374706516,
+      "learning_rate": 5e-06,
+      "loss": 0.2617,
+      "num_input_tokens_seen": 423825736,
+      "step": 2452
+    },
+    {
+      "epoch": 0.9391037916507086,
+      "loss": 0.2783448100090027,
+      "loss_ce": 0.01998301036655903,
+      "loss_iou": 1.0539875030517578,
+      "loss_num": 0.2578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 423825736,
+      "step": 2452
+    },
+    {
+      "epoch": 0.9394867866717733,
+      "grad_norm": 22.154819849787714,
+      "learning_rate": 5e-06,
+      "loss": 0.1582,
+      "num_input_tokens_seen": 423998760,
+      "step": 2453
+    },
+    {
+      "epoch": 0.9394867866717733,
+      "loss": 0.13484029471874237,
+      "loss_ce": 0.02216940000653267,
+      "loss_iou": 1.0007212162017822,
+      "loss_num": 0.11279296875,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 423998760,
+      "step": 2453
+    },
+    {
+      "epoch": 0.939869781692838,
+      "grad_norm": 31.694484928270235,
+      "learning_rate": 5e-06,
+      "loss": 0.2342,
+      "num_input_tokens_seen": 424171520,
+      "step": 2454
+    },
+    {
+      "epoch": 0.939869781692838,
+      "loss": 0.22024646401405334,
+      "loss_ce": 0.022370483726263046,
+      "loss_iou": 1.0082926750183105,
+      "loss_num": 0.1982421875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 424171520,
+      "step": 2454
+    },
+    {
+      "epoch": 0.9402527767139027,
+      "grad_norm": 27.88397412282971,
+      "learning_rate": 5e-06,
+      "loss": 0.2001,
+      "num_input_tokens_seen": 424344736,
+      "step": 2455
+    },
+    {
+      "epoch": 0.9402527767139027,
+      "loss": 0.17761653661727905,
+      "loss_ce": 0.02136654406785965,
+      "loss_iou": 1.0251858234405518,
+      "loss_num": 0.15625,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 424344736,
+      "step": 2455
+    },
+    {
+      "epoch": 0.9406357717349675,
+      "grad_norm": 48.855361931225794,
+      "learning_rate": 5e-06,
+      "loss": 0.295,
+      "num_input_tokens_seen": 424516104,
+      "step": 2456
+    },
+    {
+      "epoch": 0.9406357717349675,
+      "loss": 0.2889312505722046,
+      "loss_ce": 0.022146565839648247,
+      "loss_iou": 1.675811529159546,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 424516104,
+      "step": 2456
+    },
+    {
+      "epoch": 0.9410187667560321,
+      "grad_norm": 30.33568671434178,
+      "learning_rate": 5e-06,
+      "loss": 0.2114,
+      "num_input_tokens_seen": 424689408,
+      "step": 2457
+    },
+    {
+      "epoch": 0.9410187667560321,
+      "loss": 0.17778730392456055,
+      "loss_ce": 0.02098798379302025,
+      "loss_iou": 1.013716697692871,
+      "loss_num": 0.1572265625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 424689408,
+      "step": 2457
+    },
+    {
+      "epoch": 0.9414017617770969,
+      "grad_norm": 38.34361219398389,
+      "learning_rate": 5e-06,
+      "loss": 0.2282,
+      "num_input_tokens_seen": 424862320,
+      "step": 2458
+    },
+    {
+      "epoch": 0.9414017617770969,
+      "loss": 0.2448033094406128,
+      "loss_ce": 0.021658774465322495,
+      "loss_iou": 1.003727674484253,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 424862320,
+      "step": 2458
+    },
+    {
+      "epoch": 0.9417847567981616,
+      "grad_norm": 21.74090728512593,
+      "learning_rate": 5e-06,
+      "loss": 0.1913,
+      "num_input_tokens_seen": 425035480,
+      "step": 2459
+    },
+    {
+      "epoch": 0.9417847567981616,
+      "loss": 0.2162160873413086,
+      "loss_ce": 0.02041531540453434,
+      "loss_iou": 1.0237390995025635,
+      "loss_num": 0.1953125,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 425035480,
+      "step": 2459
+    },
+    {
+      "epoch": 0.9421677518192263,
+      "grad_norm": 36.583530825826195,
+      "learning_rate": 5e-06,
+      "loss": 0.2563,
+      "num_input_tokens_seen": 425208728,
+      "step": 2460
+    },
+    {
+      "epoch": 0.9421677518192263,
+      "loss": 0.23762497305870056,
+      "loss_ce": 0.02137741819024086,
+      "loss_iou": 1.1218187808990479,
+      "loss_num": 0.2158203125,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 425208728,
+      "step": 2460
+    },
+    {
+      "epoch": 0.942550746840291,
+      "grad_norm": 22.02133972261915,
+      "learning_rate": 5e-06,
+      "loss": 0.1593,
+      "num_input_tokens_seen": 425381832,
+      "step": 2461
+    },
+    {
+      "epoch": 0.942550746840291,
+      "loss": 0.1533421277999878,
+      "loss_ce": 0.021078944206237793,
+      "loss_iou": 1.0010343790054321,
+      "loss_num": 0.1318359375,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 425381832,
+      "step": 2461
+    },
+    {
+      "epoch": 0.9429337418613558,
+      "grad_norm": 20.812638383616566,
+      "learning_rate": 5e-06,
+      "loss": 0.1997,
+      "num_input_tokens_seen": 425555104,
+      "step": 2462
+    },
+    {
+      "epoch": 0.9429337418613558,
+      "loss": 0.1622682511806488,
+      "loss_ce": 0.021368587389588356,
+      "loss_iou": 1.0101661682128906,
+      "loss_num": 0.140625,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 425555104,
+      "step": 2462
+    },
+    {
+      "epoch": 0.9433167368824206,
+      "grad_norm": 23.13334762573491,
+      "learning_rate": 5e-06,
+      "loss": 0.1648,
+      "num_input_tokens_seen": 425727848,
+      "step": 2463
+    },
+    {
+      "epoch": 0.9433167368824206,
+      "loss": 0.214143767952919,
+      "loss_ce": 0.022371307015419006,
+      "loss_iou": 1.0665621757507324,
+      "loss_num": 0.19140625,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 425727848,
+      "step": 2463
+    },
+    {
+      "epoch": 0.9436997319034852,
+      "grad_norm": 28.141571096769116,
+      "learning_rate": 5e-06,
+      "loss": 0.1967,
+      "num_input_tokens_seen": 425900896,
+      "step": 2464
+    },
+    {
+      "epoch": 0.9436997319034852,
+      "loss": 0.17636796832084656,
+      "loss_ce": 0.021460752934217453,
+      "loss_iou": 1.0157206058502197,
+      "loss_num": 0.1552734375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 425900896,
+      "step": 2464
+    },
+    {
+      "epoch": 0.94408272692455,
+      "grad_norm": 28.273967629327604,
+      "learning_rate": 5e-06,
+      "loss": 0.1526,
+      "num_input_tokens_seen": 426074000,
+      "step": 2465
+    },
+    {
+      "epoch": 0.94408272692455,
+      "loss": 0.14048194885253906,
+      "loss_ce": 0.020334240049123764,
+      "loss_iou": 0.9708118438720703,
+      "loss_num": 0.1201171875,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 426074000,
+      "step": 2465
+    },
+    {
+      "epoch": 0.9444657219456147,
+      "grad_norm": 26.325953505132922,
+      "learning_rate": 5e-06,
+      "loss": 0.207,
+      "num_input_tokens_seen": 426247016,
+      "step": 2466
+    },
+    {
+      "epoch": 0.9444657219456147,
+      "loss": 0.242500901222229,
+      "loss_ce": 0.019539479166269302,
+      "loss_iou": 1.0102567672729492,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 426247016,
+      "step": 2466
+    },
+    {
+      "epoch": 0.9448487169666794,
+      "grad_norm": 20.862450952264904,
+      "learning_rate": 5e-06,
+      "loss": 0.1364,
+      "num_input_tokens_seen": 426416488,
+      "step": 2467
+    },
+    {
+      "epoch": 0.9448487169666794,
+      "loss": 0.14106445014476776,
+      "loss_ce": 0.022198490798473358,
+      "loss_iou": 1.0052566528320312,
+      "loss_num": 0.11865234375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 426416488,
+      "step": 2467
+    },
+    {
+      "epoch": 0.9452317119877441,
+      "grad_norm": 38.57116748053257,
+      "learning_rate": 5e-06,
+      "loss": 0.2787,
+      "num_input_tokens_seen": 426589128,
+      "step": 2468
+    },
+    {
+      "epoch": 0.9452317119877441,
+      "loss": 0.26711416244506836,
+      "loss_ce": 0.019067276269197464,
+      "loss_iou": 1.2741436958312988,
+      "loss_num": 0.248046875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 426589128,
+      "step": 2468
+    },
+    {
+      "epoch": 0.9456147070088089,
+      "grad_norm": 37.05523511090258,
+      "learning_rate": 5e-06,
+      "loss": 0.2053,
+      "num_input_tokens_seen": 426762136,
+      "step": 2469
+    },
+    {
+      "epoch": 0.9456147070088089,
+      "loss": 0.18002773821353912,
+      "loss_ce": 0.021000638604164124,
+      "loss_iou": 1.007516622543335,
+      "loss_num": 0.1591796875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 426762136,
+      "step": 2469
+    },
+    {
+      "epoch": 0.9459977020298737,
+      "grad_norm": 22.47467923645663,
+      "learning_rate": 5e-06,
+      "loss": 0.2386,
+      "num_input_tokens_seen": 426934976,
+      "step": 2470
+    },
+    {
+      "epoch": 0.9459977020298737,
+      "loss": 0.22461026906967163,
+      "loss_ce": 0.01965421438217163,
+      "loss_iou": 0.9954475164413452,
+      "loss_num": 0.205078125,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 426934976,
+      "step": 2470
+    },
+    {
+      "epoch": 0.9463806970509383,
+      "grad_norm": 35.96795631207314,
+      "learning_rate": 5e-06,
+      "loss": 0.2329,
+      "num_input_tokens_seen": 427107808,
+      "step": 2471
+    },
+    {
+      "epoch": 0.9463806970509383,
+      "loss": 0.2343979775905609,
+      "loss_ce": 0.02199564129114151,
+      "loss_iou": 1.1027498245239258,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 427107808,
+      "step": 2471
+    },
+    {
+      "epoch": 0.9467636920720031,
+      "grad_norm": 32.507799383535485,
+      "learning_rate": 5e-06,
+      "loss": 0.1832,
+      "num_input_tokens_seen": 427280736,
+      "step": 2472
+    },
+    {
+      "epoch": 0.9467636920720031,
+      "loss": 0.17686253786087036,
+      "loss_ce": 0.01963598094880581,
+      "loss_iou": 1.0173301696777344,
+      "loss_num": 0.1572265625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 427280736,
+      "step": 2472
+    },
+    {
+      "epoch": 0.9471466870930678,
+      "grad_norm": 24.407314706980245,
+      "learning_rate": 5e-06,
+      "loss": 0.227,
+      "num_input_tokens_seen": 427453840,
+      "step": 2473
+    },
+    {
+      "epoch": 0.9471466870930678,
+      "loss": 0.18174517154693604,
+      "loss_ce": 0.022748585790395737,
+      "loss_iou": 1.0485135316848755,
+      "loss_num": 0.1591796875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 427453840,
+      "step": 2473
+    },
+    {
+      "epoch": 0.9475296821141325,
+      "grad_norm": 26.028871470573005,
+      "learning_rate": 5e-06,
+      "loss": 0.2004,
+      "num_input_tokens_seen": 427627000,
+      "step": 2474
+    },
+    {
+      "epoch": 0.9475296821141325,
+      "loss": 0.15802933275699615,
+      "loss_ce": 0.021127481013536453,
+      "loss_iou": 1.0013928413391113,
+      "loss_num": 0.13671875,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 427627000,
+      "step": 2474
+    },
+    {
+      "epoch": 0.9479126771351972,
+      "grad_norm": 33.81848883195221,
+      "learning_rate": 5e-06,
+      "loss": 0.2085,
+      "num_input_tokens_seen": 427800448,
+      "step": 2475
+    },
+    {
+      "epoch": 0.9479126771351972,
+      "loss": 0.16287113726139069,
+      "loss_ce": 0.021513711661100388,
+      "loss_iou": 1.0379126071929932,
+      "loss_num": 0.1416015625,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 427800448,
+      "step": 2475
+    },
+    {
+      "epoch": 0.948295672156262,
+      "grad_norm": 49.474482420045476,
+      "learning_rate": 5e-06,
+      "loss": 0.2371,
+      "num_input_tokens_seen": 427973656,
+      "step": 2476
+    },
+    {
+      "epoch": 0.948295672156262,
+      "loss": 0.26268815994262695,
+      "loss_ce": 0.021416183561086655,
+      "loss_iou": 1.0379433631896973,
+      "loss_num": 0.2412109375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 427973656,
+      "step": 2476
+    },
+    {
+      "epoch": 0.9486786671773267,
+      "grad_norm": 28.928450317786854,
+      "learning_rate": 5e-06,
+      "loss": 0.2998,
+      "num_input_tokens_seen": 428146792,
+      "step": 2477
+    },
+    {
+      "epoch": 0.9486786671773267,
+      "loss": 0.2834222912788391,
+      "loss_ce": 0.020604893565177917,
+      "loss_iou": 1.092254400253296,
+      "loss_num": 0.263671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 428146792,
+      "step": 2477
+    },
+    {
+      "epoch": 0.9490616621983914,
+      "grad_norm": 23.9767630999348,
+      "learning_rate": 5e-06,
+      "loss": 0.2223,
+      "num_input_tokens_seen": 428319744,
+      "step": 2478
+    },
+    {
+      "epoch": 0.9490616621983914,
+      "loss": 0.22838380932807922,
+      "loss_ce": 0.021047383546829224,
+      "loss_iou": 1.040341854095459,
+      "loss_num": 0.20703125,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 428319744,
+      "step": 2478
+    },
+    {
+      "epoch": 0.9494446572194561,
+      "grad_norm": 27.230865238357527,
+      "learning_rate": 5e-06,
+      "loss": 0.2295,
+      "num_input_tokens_seen": 428492616,
+      "step": 2479
+    },
+    {
+      "epoch": 0.9494446572194561,
+      "loss": 0.1958887279033661,
+      "loss_ce": 0.01968022808432579,
+      "loss_iou": 0.9880644083023071,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 428492616,
+      "step": 2479
+    },
+    {
+      "epoch": 0.9498276522405209,
+      "grad_norm": 42.87427055455577,
+      "learning_rate": 5e-06,
+      "loss": 0.248,
+      "num_input_tokens_seen": 428666016,
+      "step": 2480
+    },
+    {
+      "epoch": 0.9498276522405209,
+      "loss": 0.25098004937171936,
+      "loss_ce": 0.02032819390296936,
+      "loss_iou": 1.064856767654419,
+      "loss_num": 0.23046875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 428666016,
+      "step": 2480
+    },
+    {
+      "epoch": 0.9502106472615856,
+      "grad_norm": 32.41868064726956,
+      "learning_rate": 5e-06,
+      "loss": 0.2358,
+      "num_input_tokens_seen": 428838920,
+      "step": 2481
+    },
+    {
+      "epoch": 0.9502106472615856,
+      "loss": 0.19582206010818481,
+      "loss_ce": 0.02101735584437847,
+      "loss_iou": 1.0015753507614136,
+      "loss_num": 0.1748046875,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 428838920,
+      "step": 2481
+    },
+    {
+      "epoch": 0.9505936422826503,
+      "grad_norm": 24.455258994810496,
+      "learning_rate": 5e-06,
+      "loss": 0.3003,
+      "num_input_tokens_seen": 429012136,
+      "step": 2482
+    },
+    {
+      "epoch": 0.9505936422826503,
+      "loss": 0.31088584661483765,
+      "loss_ce": 0.020663654431700706,
+      "loss_iou": 1.1978161334991455,
+      "loss_num": 0.291015625,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 429012136,
+      "step": 2482
+    },
+    {
+      "epoch": 0.9509766373037151,
+      "grad_norm": 24.558980927253444,
+      "learning_rate": 5e-06,
+      "loss": 0.2565,
+      "num_input_tokens_seen": 429184968,
+      "step": 2483
+    },
+    {
+      "epoch": 0.9509766373037151,
+      "loss": 0.22562754154205322,
+      "loss_ce": 0.020793568342924118,
+      "loss_iou": 1.0763115882873535,
+      "loss_num": 0.205078125,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 429184968,
+      "step": 2483
+    },
+    {
+      "epoch": 0.9513596323247798,
+      "grad_norm": 57.7302153478801,
+      "learning_rate": 5e-06,
+      "loss": 0.2699,
+      "num_input_tokens_seen": 429358384,
+      "step": 2484
+    },
+    {
+      "epoch": 0.9513596323247798,
+      "loss": 0.27880072593688965,
+      "loss_ce": 0.022575143724679947,
+      "loss_iou": 1.048409342765808,
+      "loss_num": 0.255859375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 429358384,
+      "step": 2484
+    },
+    {
+      "epoch": 0.9517426273458445,
+      "grad_norm": 41.352072095365656,
+      "learning_rate": 5e-06,
+      "loss": 0.3364,
+      "num_input_tokens_seen": 429531624,
+      "step": 2485
+    },
+    {
+      "epoch": 0.9517426273458445,
+      "loss": 0.25817739963531494,
+      "loss_ce": 0.02197132632136345,
+      "loss_iou": 1.0042860507965088,
+      "loss_num": 0.236328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 429531624,
+      "step": 2485
+    },
+    {
+      "epoch": 0.9521256223669092,
+      "grad_norm": 127.31234523264732,
+      "learning_rate": 5e-06,
+      "loss": 0.5231,
+      "num_input_tokens_seen": 429704504,
+      "step": 2486
+    },
+    {
+      "epoch": 0.9521256223669092,
+      "loss": 0.4990769028663635,
+      "loss_ce": 0.02153782919049263,
+      "loss_iou": 1.0146496295928955,
+      "loss_num": 0.4765625,
+      "loss_xval": 0.4765625,
+      "num_input_tokens_seen": 429704504,
+      "step": 2486
+    },
+    {
+      "epoch": 0.952508617387974,
+      "grad_norm": 37.48308836335706,
+      "learning_rate": 5e-06,
+      "loss": 0.1841,
+      "num_input_tokens_seen": 429877656,
+      "step": 2487
+    },
+    {
+      "epoch": 0.952508617387974,
+      "loss": 0.23241500556468964,
+      "loss_ce": 0.019768521189689636,
+      "loss_iou": 1.0113465785980225,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 429877656,
+      "step": 2487
+    },
+    {
+      "epoch": 0.9528916124090386,
+      "grad_norm": 67.83485410278199,
+      "learning_rate": 5e-06,
+      "loss": 0.6158,
+      "num_input_tokens_seen": 430050504,
+      "step": 2488
+    },
+    {
+      "epoch": 0.9528916124090386,
+      "loss": 0.6186951398849487,
+      "loss_ce": 0.019085798412561417,
+      "loss_iou": 1.0142709016799927,
+      "loss_num": 0.6015625,
+      "loss_xval": 0.6015625,
+      "num_input_tokens_seen": 430050504,
+      "step": 2488
+    },
+    {
+      "epoch": 0.9532746074301034,
+      "grad_norm": 43.22575450579638,
+      "learning_rate": 5e-06,
+      "loss": 0.271,
+      "num_input_tokens_seen": 430223656,
+      "step": 2489
+    },
+    {
+      "epoch": 0.9532746074301034,
+      "loss": 0.28194326162338257,
+      "loss_ce": 0.019858306273818016,
+      "loss_iou": 1.0584187507629395,
+      "loss_num": 0.26171875,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 430223656,
+      "step": 2489
+    },
+    {
+      "epoch": 0.9536576024511682,
+      "grad_norm": 48.527551822606995,
+      "learning_rate": 5e-06,
+      "loss": 0.3893,
+      "num_input_tokens_seen": 430396144,
+      "step": 2490
+    },
+    {
+      "epoch": 0.9536576024511682,
+      "loss": 0.3745991587638855,
+      "loss_ce": 0.021083537489175797,
+      "loss_iou": 1.0061142444610596,
+      "loss_num": 0.353515625,
+      "loss_xval": 0.353515625,
+      "num_input_tokens_seen": 430396144,
+      "step": 2490
+    },
+    {
+      "epoch": 0.9540405974722329,
+      "grad_norm": 38.51900506222416,
+      "learning_rate": 5e-06,
+      "loss": 0.2578,
+      "num_input_tokens_seen": 430568832,
+      "step": 2491
+    },
+    {
+      "epoch": 0.9540405974722329,
+      "loss": 0.29192161560058594,
+      "loss_ce": 0.01946067623794079,
+      "loss_iou": 1.0029220581054688,
+      "loss_num": 0.2734375,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 430568832,
+      "step": 2491
+    },
+    {
+      "epoch": 0.9544235924932976,
+      "grad_norm": 33.67553454531333,
+      "learning_rate": 5e-06,
+      "loss": 0.21,
+      "num_input_tokens_seen": 430741736,
+      "step": 2492
+    },
+    {
+      "epoch": 0.9544235924932976,
+      "loss": 0.2243577241897583,
+      "loss_ce": 0.020378228276968002,
+      "loss_iou": 1.1379162073135376,
+      "loss_num": 0.2041015625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 430741736,
+      "step": 2492
+    },
+    {
+      "epoch": 0.9548065875143623,
+      "grad_norm": 40.622192134653055,
+      "learning_rate": 5e-06,
+      "loss": 0.2468,
+      "num_input_tokens_seen": 430914824,
+      "step": 2493
+    },
+    {
+      "epoch": 0.9548065875143623,
+      "loss": 0.25958168506622314,
+      "loss_ce": 0.02160559967160225,
+      "loss_iou": 1.0875476598739624,
+      "loss_num": 0.23828125,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 430914824,
+      "step": 2493
+    },
+    {
+      "epoch": 0.9551895825354271,
+      "grad_norm": 32.91565123938936,
+      "learning_rate": 5e-06,
+      "loss": 0.2176,
+      "num_input_tokens_seen": 431087672,
+      "step": 2494
+    },
+    {
+      "epoch": 0.9551895825354271,
+      "loss": 0.21976381540298462,
+      "loss_ce": 0.021460603922605515,
+      "loss_iou": 1.0525672435760498,
+      "loss_num": 0.1982421875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 431087672,
+      "step": 2494
+    },
+    {
+      "epoch": 0.9555725775564917,
+      "grad_norm": 30.327394649264097,
+      "learning_rate": 5e-06,
+      "loss": 0.2404,
+      "num_input_tokens_seen": 431260584,
+      "step": 2495
+    },
+    {
+      "epoch": 0.9555725775564917,
+      "loss": 0.22192560136318207,
+      "loss_ce": 0.021242011338472366,
+      "loss_iou": 1.0752569437026978,
+      "loss_num": 0.201171875,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 431260584,
+      "step": 2495
+    },
+    {
+      "epoch": 0.9559555725775565,
+      "grad_norm": 28.437906400554276,
+      "learning_rate": 5e-06,
+      "loss": 0.1712,
+      "num_input_tokens_seen": 431433752,
+      "step": 2496
+    },
+    {
+      "epoch": 0.9559555725775565,
+      "loss": 0.15172705054283142,
+      "loss_ce": 0.020867664366960526,
+      "loss_iou": 1.0025279521942139,
+      "loss_num": 0.130859375,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 431433752,
+      "step": 2496
+    },
+    {
+      "epoch": 0.9563385675986212,
+      "grad_norm": 28.213415813890716,
+      "learning_rate": 5e-06,
+      "loss": 0.1673,
+      "num_input_tokens_seen": 431607072,
+      "step": 2497
+    },
+    {
+      "epoch": 0.9563385675986212,
+      "loss": 0.17967821657657623,
+      "loss_ce": 0.020254386588931084,
+      "loss_iou": 1.0270193815231323,
+      "loss_num": 0.1591796875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 431607072,
+      "step": 2497
+    },
+    {
+      "epoch": 0.956721562619686,
+      "grad_norm": 31.790672881180054,
+      "learning_rate": 5e-06,
+      "loss": 0.1963,
+      "num_input_tokens_seen": 431779568,
+      "step": 2498
+    },
+    {
+      "epoch": 0.956721562619686,
+      "loss": 0.22004179656505585,
+      "loss_ce": 0.02082304283976555,
+      "loss_iou": 1.3927192687988281,
+      "loss_num": 0.19921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 431779568,
+      "step": 2498
+    },
+    {
+      "epoch": 0.9571045576407506,
+      "grad_norm": 32.857413340516864,
+      "learning_rate": 5e-06,
+      "loss": 0.2231,
+      "num_input_tokens_seen": 431952488,
+      "step": 2499
+    },
+    {
+      "epoch": 0.9571045576407506,
+      "loss": 0.18316613137722015,
+      "loss_ce": 0.02185022458434105,
+      "loss_iou": 1.0009148120880127,
+      "loss_num": 0.1611328125,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 431952488,
+      "step": 2499
+    },
+    {
+      "epoch": 0.9574875526618154,
+      "grad_norm": 29.62708282557836,
+      "learning_rate": 5e-06,
+      "loss": 0.2037,
+      "num_input_tokens_seen": 432125408,
+      "step": 2500
+    },
+    {
+      "epoch": 0.9574875526618154,
+      "eval_websight_new_CIoU": 0.9313753843307495,
+      "eval_websight_new_GIoU": 0.9312478005886078,
+      "eval_websight_new_IoU": 0.9315329194068909,
+      "eval_websight_new_MAE_all": 0.006993612740188837,
+      "eval_websight_new_MAE_h": 0.009050879394635558,
+      "eval_websight_new_MAE_w": 0.007409252226352692,
+      "eval_websight_new_MAE_x": 0.007410084828734398,
+      "eval_websight_new_MAE_y": 0.0041042345110327005,
+      "eval_websight_new_NUM_probability": 8.344057278009132e-05,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.11439123004674911,
+      "eval_websight_new_loss_ce": 0.04191363602876663,
+      "eval_websight_new_loss_iou": 1.0002729892730713,
+      "eval_websight_new_loss_num": 0.06982421875,
+      "eval_websight_new_loss_xval": 0.06982421875,
+      "eval_websight_new_runtime": 55.5425,
+      "eval_websight_new_samples_per_second": 0.9,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 432125408,
+      "step": 2500
+    },
+    {
+      "epoch": 0.9574875526618154,
+      "eval_seeclick_CIoU": 0.6130296289920807,
+      "eval_seeclick_GIoU": 0.6226363182067871,
+      "eval_seeclick_IoU": 0.6598637700080872,
+      "eval_seeclick_MAE_all": 0.07341223955154419,
+      "eval_seeclick_MAE_h": 0.02738715149462223,
+      "eval_seeclick_MAE_w": 0.1214686743915081,
+      "eval_seeclick_MAE_x": 0.11567392945289612,
+      "eval_seeclick_MAE_y": 0.029119192622601986,
+      "eval_seeclick_NUM_probability": 0.00011037691729143262,
+      "eval_seeclick_inside_bbox": 0.8697916567325592,
+      "eval_seeclick_loss": 0.5349428653717041,
+      "eval_seeclick_loss_ce": 0.032741205766797066,
+      "eval_seeclick_loss_iou": 1.181551992893219,
+      "eval_seeclick_loss_num": 0.489105224609375,
+      "eval_seeclick_loss_xval": 0.489105224609375,
+      "eval_seeclick_runtime": 92.3956,
+      "eval_seeclick_samples_per_second": 0.541,
+      "eval_seeclick_steps_per_second": 0.022,
+      "num_input_tokens_seen": 432125408,
+      "step": 2500
+    },
+    {
+      "epoch": 0.9574875526618154,
+      "eval_icons_CIoU": 0.8740069270133972,
+      "eval_icons_GIoU": 0.8716206848621368,
+      "eval_icons_IoU": 0.8827140927314758,
+      "eval_icons_MAE_all": 0.02025833074003458,
+      "eval_icons_MAE_h": 0.01597935613244772,
+      "eval_icons_MAE_w": 0.027698833495378494,
+      "eval_icons_MAE_x": 0.023655809462070465,
+      "eval_icons_MAE_y": 0.013699323870241642,
+      "eval_icons_NUM_probability": 7.427079981425777e-05,
+      "eval_icons_inside_bbox": 0.984375,
+      "eval_icons_loss": 0.16799156367778778,
+      "eval_icons_loss_ce": 0.028499008156359196,
+      "eval_icons_loss_iou": 1.042014479637146,
+      "eval_icons_loss_num": 0.127960205078125,
+      "eval_icons_loss_xval": 0.127960205078125,
+      "eval_icons_runtime": 82.8347,
+      "eval_icons_samples_per_second": 0.604,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 432125408,
+      "step": 2500
+    },
+    {
+      "epoch": 0.9574875526618154,
+      "loss": 0.15654532611370087,
+      "loss_ce": 0.028493572026491165,
+      "loss_iou": 1.0288493633270264,
+      "loss_num": 0.1279296875,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 432125408,
+      "step": 2500
+    },
+    {
+      "epoch": 0.9578705476828802,
+      "grad_norm": 34.23645201056457,
+      "learning_rate": 5e-06,
+      "loss": 0.1892,
+      "num_input_tokens_seen": 432298384,
+      "step": 2501
+    },
+    {
+      "epoch": 0.9578705476828802,
+      "loss": 0.16409295797348022,
+      "loss_ce": 0.02001948095858097,
+      "loss_iou": 1.0105680227279663,
+      "loss_num": 0.14453125,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 432298384,
+      "step": 2501
+    },
+    {
+      "epoch": 0.9582535427039448,
+      "grad_norm": 28.206097624638492,
+      "learning_rate": 5e-06,
+      "loss": 0.1565,
+      "num_input_tokens_seen": 432471304,
+      "step": 2502
+    },
+    {
+      "epoch": 0.9582535427039448,
+      "loss": 0.1427156925201416,
+      "loss_ce": 0.019851919263601303,
+      "loss_iou": 1.00319504737854,
+      "loss_num": 0.123046875,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 432471304,
+      "step": 2502
+    },
+    {
+      "epoch": 0.9586365377250096,
+      "grad_norm": 23.28794682294048,
+      "learning_rate": 5e-06,
+      "loss": 0.1888,
+      "num_input_tokens_seen": 432644128,
+      "step": 2503
+    },
+    {
+      "epoch": 0.9586365377250096,
+      "loss": 0.18110212683677673,
+      "loss_ce": 0.019450515508651733,
+      "loss_iou": 1.0770268440246582,
+      "loss_num": 0.162109375,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 432644128,
+      "step": 2503
+    },
+    {
+      "epoch": 0.9590195327460743,
+      "grad_norm": 28.321850181258878,
+      "learning_rate": 5e-06,
+      "loss": 0.1913,
+      "num_input_tokens_seen": 432817176,
+      "step": 2504
+    },
+    {
+      "epoch": 0.9590195327460743,
+      "loss": 0.15393656492233276,
+      "loss_ce": 0.019933875650167465,
+      "loss_iou": 1.052881121635437,
+      "loss_num": 0.1337890625,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 432817176,
+      "step": 2504
+    },
+    {
+      "epoch": 0.9594025277671391,
+      "grad_norm": 21.905521174487735,
+      "learning_rate": 5e-06,
+      "loss": 0.1719,
+      "num_input_tokens_seen": 432990048,
+      "step": 2505
+    },
+    {
+      "epoch": 0.9594025277671391,
+      "loss": 0.1624247133731842,
+      "loss_ce": 0.022226953878998756,
+      "loss_iou": 1.0590568780899048,
+      "loss_num": 0.140625,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 432990048,
+      "step": 2505
+    },
+    {
+      "epoch": 0.9597855227882037,
+      "grad_norm": 39.579709731044446,
+      "learning_rate": 5e-06,
+      "loss": 0.1799,
+      "num_input_tokens_seen": 433163088,
+      "step": 2506
+    },
+    {
+      "epoch": 0.9597855227882037,
+      "loss": 0.19605445861816406,
+      "loss_ce": 0.020212167873978615,
+      "loss_iou": 1.0139884948730469,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 433163088,
+      "step": 2506
+    },
+    {
+      "epoch": 0.9601685178092685,
+      "grad_norm": 29.228807115410213,
+      "learning_rate": 5e-06,
+      "loss": 0.2188,
+      "num_input_tokens_seen": 433335904,
+      "step": 2507
+    },
+    {
+      "epoch": 0.9601685178092685,
+      "loss": 0.15629726648330688,
+      "loss_ce": 0.02250819280743599,
+      "loss_iou": 1.0016660690307617,
+      "loss_num": 0.1337890625,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 433335904,
+      "step": 2507
+    },
+    {
+      "epoch": 0.9605515128303332,
+      "grad_norm": 31.417712113405898,
+      "learning_rate": 5e-06,
+      "loss": 0.1762,
+      "num_input_tokens_seen": 433509104,
+      "step": 2508
+    },
+    {
+      "epoch": 0.9605515128303332,
+      "loss": 0.16565579175949097,
+      "loss_ce": 0.020636267960071564,
+      "loss_iou": 1.0083417892456055,
+      "loss_num": 0.14453125,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 433509104,
+      "step": 2508
+    },
+    {
+      "epoch": 0.9609345078513979,
+      "grad_norm": 23.38483090549218,
+      "learning_rate": 5e-06,
+      "loss": 0.1705,
+      "num_input_tokens_seen": 433682216,
+      "step": 2509
+    },
+    {
+      "epoch": 0.9609345078513979,
+      "loss": 0.16561777889728546,
+      "loss_ce": 0.02194102294743061,
+      "loss_iou": 1.0171895027160645,
+      "loss_num": 0.1435546875,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 433682216,
+      "step": 2509
+    },
+    {
+      "epoch": 0.9613175028724626,
+      "grad_norm": 29.654963437029014,
+      "learning_rate": 5e-06,
+      "loss": 0.2501,
+      "num_input_tokens_seen": 433854968,
+      "step": 2510
+    },
+    {
+      "epoch": 0.9613175028724626,
+      "loss": 0.23729385435581207,
+      "loss_ce": 0.021046288311481476,
+      "loss_iou": 1.159039855003357,
+      "loss_num": 0.2158203125,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 433854968,
+      "step": 2510
+    },
+    {
+      "epoch": 0.9617004978935274,
+      "grad_norm": 23.730611767262584,
+      "learning_rate": 5e-06,
+      "loss": 0.1672,
+      "num_input_tokens_seen": 434027888,
+      "step": 2511
+    },
+    {
+      "epoch": 0.9617004978935274,
+      "loss": 0.1595851480960846,
+      "loss_ce": 0.021157419309020042,
+      "loss_iou": 1.0851759910583496,
+      "loss_num": 0.138671875,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 434027888,
+      "step": 2511
+    },
+    {
+      "epoch": 0.9620834929145922,
+      "grad_norm": 21.111844972999215,
+      "learning_rate": 5e-06,
+      "loss": 0.1558,
+      "num_input_tokens_seen": 434200968,
+      "step": 2512
+    },
+    {
+      "epoch": 0.9620834929145922,
+      "loss": 0.1313307285308838,
+      "loss_ce": 0.02110123634338379,
+      "loss_iou": 1.0032182931900024,
+      "loss_num": 0.1103515625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 434200968,
+      "step": 2512
+    },
+    {
+      "epoch": 0.9624664879356568,
+      "grad_norm": 22.055065060453778,
+      "learning_rate": 5e-06,
+      "loss": 0.1905,
+      "num_input_tokens_seen": 434373864,
+      "step": 2513
+    },
+    {
+      "epoch": 0.9624664879356568,
+      "loss": 0.22645601630210876,
+      "loss_ce": 0.021560989320278168,
+      "loss_iou": 1.0219660997390747,
+      "loss_num": 0.205078125,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 434373864,
+      "step": 2513
+    },
+    {
+      "epoch": 0.9628494829567216,
+      "grad_norm": 23.41958901805516,
+      "learning_rate": 5e-06,
+      "loss": 0.1681,
+      "num_input_tokens_seen": 434547056,
+      "step": 2514
+    },
+    {
+      "epoch": 0.9628494829567216,
+      "loss": 0.20366087555885315,
+      "loss_ce": 0.020433329045772552,
+      "loss_iou": 1.0604270696640015,
+      "loss_num": 0.18359375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 434547056,
+      "step": 2514
+    },
+    {
+      "epoch": 0.9632324779777863,
+      "grad_norm": 25.817644090311354,
+      "learning_rate": 5e-06,
+      "loss": 0.2051,
+      "num_input_tokens_seen": 434720016,
+      "step": 2515
+    },
+    {
+      "epoch": 0.9632324779777863,
+      "loss": 0.19977524876594543,
+      "loss_ce": 0.02118638902902603,
+      "loss_iou": 1.0252264738082886,
+      "loss_num": 0.1787109375,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 434720016,
+      "step": 2515
+    },
+    {
+      "epoch": 0.963615472998851,
+      "grad_norm": 21.44654044806338,
+      "learning_rate": 5e-06,
+      "loss": 0.2181,
+      "num_input_tokens_seen": 434892784,
+      "step": 2516
+    },
+    {
+      "epoch": 0.963615472998851,
+      "loss": 0.17512735724449158,
+      "loss_ce": 0.021501857787370682,
+      "loss_iou": 1.0227500200271606,
+      "loss_num": 0.1533203125,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 434892784,
+      "step": 2516
+    },
+    {
+      "epoch": 0.9639984680199157,
+      "grad_norm": 29.243444115137073,
+      "learning_rate": 5e-06,
+      "loss": 0.2127,
+      "num_input_tokens_seen": 435065904,
+      "step": 2517
+    },
+    {
+      "epoch": 0.9639984680199157,
+      "loss": 0.16162803769111633,
+      "loss_ce": 0.020270608365535736,
+      "loss_iou": 0.9842989444732666,
+      "loss_num": 0.1416015625,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 435065904,
+      "step": 2517
+    },
+    {
+      "epoch": 0.9643814630409805,
+      "grad_norm": 19.00245730141102,
+      "learning_rate": 5e-06,
+      "loss": 0.1528,
+      "num_input_tokens_seen": 435239144,
+      "step": 2518
+    },
+    {
+      "epoch": 0.9643814630409805,
+      "loss": 0.13501927256584167,
+      "loss_ce": 0.02204321324825287,
+      "loss_iou": 1.0068590641021729,
+      "loss_num": 0.11279296875,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 435239144,
+      "step": 2518
+    },
+    {
+      "epoch": 0.9647644580620451,
+      "grad_norm": 27.28207204256603,
+      "learning_rate": 5e-06,
+      "loss": 0.2288,
+      "num_input_tokens_seen": 435411984,
+      "step": 2519
+    },
+    {
+      "epoch": 0.9647644580620451,
+      "loss": 0.22544914484024048,
+      "loss_ce": 0.021469654515385628,
+      "loss_iou": 1.056739091873169,
+      "loss_num": 0.2041015625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 435411984,
+      "step": 2519
+    },
+    {
+      "epoch": 0.9651474530831099,
+      "grad_norm": 23.159181319312985,
+      "learning_rate": 5e-06,
+      "loss": 0.1664,
+      "num_input_tokens_seen": 435584632,
+      "step": 2520
+    },
+    {
+      "epoch": 0.9651474530831099,
+      "loss": 0.1685417741537094,
+      "loss_ce": 0.020409442484378815,
+      "loss_iou": 1.0068516731262207,
+      "loss_num": 0.1484375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 435584632,
+      "step": 2520
+    },
+    {
+      "epoch": 0.9655304481041747,
+      "grad_norm": 21.30143045733993,
+      "learning_rate": 5e-06,
+      "loss": 0.1843,
+      "num_input_tokens_seen": 435757240,
+      "step": 2521
+    },
+    {
+      "epoch": 0.9655304481041747,
+      "loss": 0.24511337280273438,
+      "loss_ce": 0.019344326108694077,
+      "loss_iou": 1.0494751930236816,
+      "loss_num": 0.2255859375,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 435757240,
+      "step": 2521
+    },
+    {
+      "epoch": 0.9659134431252394,
+      "grad_norm": 25.198745190813224,
+      "learning_rate": 5e-06,
+      "loss": 0.1939,
+      "num_input_tokens_seen": 435930544,
+      "step": 2522
+    },
+    {
+      "epoch": 0.9659134431252394,
+      "loss": 0.18986818194389343,
+      "loss_ce": 0.019885260611772537,
+      "loss_iou": 1.0129027366638184,
+      "loss_num": 0.169921875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 435930544,
+      "step": 2522
+    },
+    {
+      "epoch": 0.9662964381463041,
+      "grad_norm": 29.056411759244,
+      "learning_rate": 5e-06,
+      "loss": 0.1893,
+      "num_input_tokens_seen": 436103592,
+      "step": 2523
+    },
+    {
+      "epoch": 0.9662964381463041,
+      "loss": 0.21499845385551453,
+      "loss_ce": 0.02170010842382908,
+      "loss_iou": 1.0486135482788086,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 436103592,
+      "step": 2523
+    },
+    {
+      "epoch": 0.9666794331673688,
+      "grad_norm": 30.64720355993145,
+      "learning_rate": 5e-06,
+      "loss": 0.2418,
+      "num_input_tokens_seen": 436276536,
+      "step": 2524
+    },
+    {
+      "epoch": 0.9666794331673688,
+      "loss": 0.2639557719230652,
+      "loss_ce": 0.02103586122393608,
+      "loss_iou": 1.0420210361480713,
+      "loss_num": 0.2431640625,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 436276536,
+      "step": 2524
+    },
+    {
+      "epoch": 0.9670624281884336,
+      "grad_norm": 25.76626566408051,
+      "learning_rate": 5e-06,
+      "loss": 0.216,
+      "num_input_tokens_seen": 436449248,
+      "step": 2525
+    },
+    {
+      "epoch": 0.9670624281884336,
+      "loss": 0.24285364151000977,
+      "loss_ce": 0.021357063204050064,
+      "loss_iou": 1.0765326023101807,
+      "loss_num": 0.2216796875,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 436449248,
+      "step": 2525
+    },
+    {
+      "epoch": 0.9674454232094982,
+      "grad_norm": 27.500392147356273,
+      "learning_rate": 5e-06,
+      "loss": 0.1725,
+      "num_input_tokens_seen": 436622304,
+      "step": 2526
+    },
+    {
+      "epoch": 0.9674454232094982,
+      "loss": 0.16017383337020874,
+      "loss_ce": 0.02189869061112404,
+      "loss_iou": 1.0175732374191284,
+      "loss_num": 0.138671875,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 436622304,
+      "step": 2526
+    },
+    {
+      "epoch": 0.967828418230563,
+      "grad_norm": 25.47773672192388,
+      "learning_rate": 5e-06,
+      "loss": 0.1636,
+      "num_input_tokens_seen": 436795032,
+      "step": 2527
+    },
+    {
+      "epoch": 0.967828418230563,
+      "loss": 0.12045948952436447,
+      "loss_ce": 0.02060597762465477,
+      "loss_iou": 1.0010665655136108,
+      "loss_num": 0.099609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 436795032,
+      "step": 2527
+    },
+    {
+      "epoch": 0.9682114132516277,
+      "grad_norm": 22.648055947928828,
+      "learning_rate": 5e-06,
+      "loss": 0.1987,
+      "num_input_tokens_seen": 436968056,
+      "step": 2528
+    },
+    {
+      "epoch": 0.9682114132516277,
+      "loss": 0.23438477516174316,
+      "loss_ce": 0.02057860977947712,
+      "loss_iou": 1.0746012926101685,
+      "loss_num": 0.2138671875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 436968056,
+      "step": 2528
+    },
+    {
+      "epoch": 0.9685944082726925,
+      "grad_norm": 23.70877903619858,
+      "learning_rate": 5e-06,
+      "loss": 0.2398,
+      "num_input_tokens_seen": 437141232,
+      "step": 2529
+    },
+    {
+      "epoch": 0.9685944082726925,
+      "loss": 0.2138303965330124,
+      "loss_ce": 0.02138654887676239,
+      "loss_iou": 1.0094234943389893,
+      "loss_num": 0.1923828125,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 437141232,
+      "step": 2529
+    },
+    {
+      "epoch": 0.9689774032937571,
+      "grad_norm": 22.86932490243878,
+      "learning_rate": 5e-06,
+      "loss": 0.1436,
+      "num_input_tokens_seen": 437314296,
+      "step": 2530
+    },
+    {
+      "epoch": 0.9689774032937571,
+      "loss": 0.1513497680425644,
+      "loss_ce": 0.018689867109060287,
+      "loss_iou": 1.0209299325942993,
+      "loss_num": 0.1328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 437314296,
+      "step": 2530
+    },
+    {
+      "epoch": 0.9693603983148219,
+      "grad_norm": 26.33529692364274,
+      "learning_rate": 5e-06,
+      "loss": 0.2707,
+      "num_input_tokens_seen": 437486992,
+      "step": 2531
+    },
+    {
+      "epoch": 0.9693603983148219,
+      "loss": 0.3272448182106018,
+      "loss_ce": 0.021214541047811508,
+      "loss_iou": 1.067396640777588,
+      "loss_num": 0.306640625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 437486992,
+      "step": 2531
+    },
+    {
+      "epoch": 0.9697433933358867,
+      "grad_norm": 27.425904108762616,
+      "learning_rate": 5e-06,
+      "loss": 0.18,
+      "num_input_tokens_seen": 437659840,
+      "step": 2532
+    },
+    {
+      "epoch": 0.9697433933358867,
+      "loss": 0.17225991189479828,
+      "loss_ce": 0.021686187013983727,
+      "loss_iou": 1.0260432958602905,
+      "loss_num": 0.150390625,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 437659840,
+      "step": 2532
+    },
+    {
+      "epoch": 0.9701263883569513,
+      "grad_norm": 21.04805762064608,
+      "learning_rate": 5e-06,
+      "loss": 0.1602,
+      "num_input_tokens_seen": 437832872,
+      "step": 2533
+    },
+    {
+      "epoch": 0.9701263883569513,
+      "loss": 0.1586833894252777,
+      "loss_ce": 0.020072538405656815,
+      "loss_iou": 1.0158920288085938,
+      "loss_num": 0.138671875,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 437832872,
+      "step": 2533
+    },
+    {
+      "epoch": 0.9705093833780161,
+      "grad_norm": 54.9970946542518,
+      "learning_rate": 5e-06,
+      "loss": 0.2388,
+      "num_input_tokens_seen": 438005496,
+      "step": 2534
+    },
+    {
+      "epoch": 0.9705093833780161,
+      "loss": 0.22775228321552277,
+      "loss_ce": 0.020507417619228363,
+      "loss_iou": 1.063530445098877,
+      "loss_num": 0.20703125,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 438005496,
+      "step": 2534
+    },
+    {
+      "epoch": 0.9708923783990808,
+      "grad_norm": 27.47018740555721,
+      "learning_rate": 5e-06,
+      "loss": 0.1288,
+      "num_input_tokens_seen": 438178304,
+      "step": 2535
+    },
+    {
+      "epoch": 0.9708923783990808,
+      "loss": 0.12139128148555756,
+      "loss_ce": 0.021446216851472855,
+      "loss_iou": 1.0063189268112183,
+      "loss_num": 0.10009765625,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 438178304,
+      "step": 2535
+    },
+    {
+      "epoch": 0.9712753734201456,
+      "grad_norm": 24.763260067649266,
+      "learning_rate": 5e-06,
+      "loss": 0.151,
+      "num_input_tokens_seen": 438351224,
+      "step": 2536
+    },
+    {
+      "epoch": 0.9712753734201456,
+      "loss": 0.1290130615234375,
+      "loss_ce": 0.019638072699308395,
+      "loss_iou": 1.0127249956130981,
+      "loss_num": 0.109375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 438351224,
+      "step": 2536
+    },
+    {
+      "epoch": 0.9716583684412102,
+      "grad_norm": 20.584487121470104,
+      "learning_rate": 5e-06,
+      "loss": 0.1623,
+      "num_input_tokens_seen": 438524552,
+      "step": 2537
+    },
+    {
+      "epoch": 0.9716583684412102,
+      "loss": 0.17038914561271667,
+      "loss_ce": 0.020730936899781227,
+      "loss_iou": 1.0084894895553589,
+      "loss_num": 0.1494140625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 438524552,
+      "step": 2537
+    },
+    {
+      "epoch": 0.972041363462275,
+      "grad_norm": 22.849692926258793,
+      "learning_rate": 5e-06,
+      "loss": 0.1942,
+      "num_input_tokens_seen": 438698088,
+      "step": 2538
+    },
+    {
+      "epoch": 0.972041363462275,
+      "loss": 0.19734948873519897,
+      "loss_ce": 0.02098839357495308,
+      "loss_iou": 1.0326783657073975,
+      "loss_num": 0.1767578125,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 438698088,
+      "step": 2538
+    },
+    {
+      "epoch": 0.9724243584833397,
+      "grad_norm": 42.844511246710226,
+      "learning_rate": 5e-06,
+      "loss": 0.2173,
+      "num_input_tokens_seen": 438870816,
+      "step": 2539
+    },
+    {
+      "epoch": 0.9724243584833397,
+      "loss": 0.27259647846221924,
+      "loss_ce": 0.01948370225727558,
+      "loss_iou": 1.029143214225769,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 438870816,
+      "step": 2539
+    },
+    {
+      "epoch": 0.9728073535044044,
+      "grad_norm": 21.005623578912854,
+      "learning_rate": 5e-06,
+      "loss": 0.1626,
+      "num_input_tokens_seen": 439043440,
+      "step": 2540
+    },
+    {
+      "epoch": 0.9728073535044044,
+      "loss": 0.1620333343744278,
+      "loss_ce": 0.019912976771593094,
+      "loss_iou": 1.0526223182678223,
+      "loss_num": 0.142578125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 439043440,
+      "step": 2540
+    },
+    {
+      "epoch": 0.9731903485254692,
+      "grad_norm": 17.165563583152803,
+      "learning_rate": 5e-06,
+      "loss": 0.1666,
+      "num_input_tokens_seen": 439216064,
+      "step": 2541
+    },
+    {
+      "epoch": 0.9731903485254692,
+      "loss": 0.18346142768859863,
+      "loss_ce": 0.020924817770719528,
+      "loss_iou": 1.0263832807540894,
+      "loss_num": 0.162109375,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 439216064,
+      "step": 2541
+    },
+    {
+      "epoch": 0.9735733435465339,
+      "grad_norm": 13.940829930362852,
+      "learning_rate": 5e-06,
+      "loss": 0.1615,
+      "num_input_tokens_seen": 439389288,
+      "step": 2542
+    },
+    {
+      "epoch": 0.9735733435465339,
+      "loss": 0.15158751606941223,
+      "loss_ce": 0.020789172500371933,
+      "loss_iou": 1.0165517330169678,
+      "loss_num": 0.130859375,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 439389288,
+      "step": 2542
+    },
+    {
+      "epoch": 0.9739563385675987,
+      "grad_norm": 28.670148954166926,
+      "learning_rate": 5e-06,
+      "loss": 0.1747,
+      "num_input_tokens_seen": 439562032,
+      "step": 2543
+    },
+    {
+      "epoch": 0.9739563385675987,
+      "loss": 0.18583689630031586,
+      "loss_ce": 0.02183542773127556,
+      "loss_iou": 1.0023903846740723,
+      "loss_num": 0.1640625,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 439562032,
+      "step": 2543
+    },
+    {
+      "epoch": 0.9743393335886633,
+      "grad_norm": 25.67718614600306,
+      "learning_rate": 5e-06,
+      "loss": 0.1972,
+      "num_input_tokens_seen": 439735200,
+      "step": 2544
+    },
+    {
+      "epoch": 0.9743393335886633,
+      "loss": 0.2517070770263672,
+      "loss_ce": 0.02160453237593174,
+      "loss_iou": 1.0602805614471436,
+      "loss_num": 0.23046875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 439735200,
+      "step": 2544
+    },
+    {
+      "epoch": 0.9747223286097281,
+      "grad_norm": 35.183277933179696,
+      "learning_rate": 5e-06,
+      "loss": 0.1793,
+      "num_input_tokens_seen": 439908008,
+      "step": 2545
+    },
+    {
+      "epoch": 0.9747223286097281,
+      "loss": 0.19800016283988953,
+      "loss_ce": 0.02038784883916378,
+      "loss_iou": 1.0084583759307861,
+      "loss_num": 0.177734375,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 439908008,
+      "step": 2545
+    },
+    {
+      "epoch": 0.9751053236307928,
+      "grad_norm": 25.51645935562071,
+      "learning_rate": 5e-06,
+      "loss": 0.2091,
+      "num_input_tokens_seen": 440080880,
+      "step": 2546
+    },
+    {
+      "epoch": 0.9751053236307928,
+      "loss": 0.23107486963272095,
+      "loss_ce": 0.02038150653243065,
+      "loss_iou": 1.071962594985962,
+      "loss_num": 0.2109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 440080880,
+      "step": 2546
+    },
+    {
+      "epoch": 0.9754883186518575,
+      "grad_norm": 27.49943928004506,
+      "learning_rate": 5e-06,
+      "loss": 0.2209,
+      "num_input_tokens_seen": 440253680,
+      "step": 2547
+    },
+    {
+      "epoch": 0.9754883186518575,
+      "loss": 0.16668814420700073,
+      "loss_ce": 0.021180346608161926,
+      "loss_iou": 1.0026423931121826,
+      "loss_num": 0.1455078125,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 440253680,
+      "step": 2547
+    },
+    {
+      "epoch": 0.9758713136729222,
+      "grad_norm": 31.990871381822163,
+      "learning_rate": 5e-06,
+      "loss": 0.2257,
+      "num_input_tokens_seen": 440426568,
+      "step": 2548
+    },
+    {
+      "epoch": 0.9758713136729222,
+      "loss": 0.19652478396892548,
+      "loss_ce": 0.020896129310131073,
+      "loss_iou": 1.0123958587646484,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 440426568,
+      "step": 2548
+    },
+    {
+      "epoch": 0.976254308693987,
+      "grad_norm": 29.020982825762303,
+      "learning_rate": 5e-06,
+      "loss": 0.2053,
+      "num_input_tokens_seen": 440599144,
+      "step": 2549
+    },
+    {
+      "epoch": 0.976254308693987,
+      "loss": 0.20318135619163513,
+      "loss_ce": 0.01955709606409073,
+      "loss_iou": 1.0952755212783813,
+      "loss_num": 0.18359375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 440599144,
+      "step": 2549
+    },
+    {
+      "epoch": 0.9766373037150518,
+      "grad_norm": 24.228724393713808,
+      "learning_rate": 5e-06,
+      "loss": 0.1567,
+      "num_input_tokens_seen": 440772152,
+      "step": 2550
+    },
+    {
+      "epoch": 0.9766373037150518,
+      "loss": 0.17433422803878784,
+      "loss_ce": 0.019182849675416946,
+      "loss_iou": 1.0140941143035889,
+      "loss_num": 0.1552734375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 440772152,
+      "step": 2550
+    },
+    {
+      "epoch": 0.9770202987361164,
+      "grad_norm": 25.817141077281935,
+      "learning_rate": 5e-06,
+      "loss": 0.2186,
+      "num_input_tokens_seen": 440945232,
+      "step": 2551
+    },
+    {
+      "epoch": 0.9770202987361164,
+      "loss": 0.1617930382490158,
+      "loss_ce": 0.02104596421122551,
+      "loss_iou": 1.002140998840332,
+      "loss_num": 0.140625,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 440945232,
+      "step": 2551
+    },
+    {
+      "epoch": 0.9774032937571812,
+      "grad_norm": 17.31820511238446,
+      "learning_rate": 5e-06,
+      "loss": 0.2124,
+      "num_input_tokens_seen": 441116728,
+      "step": 2552
+    },
+    {
+      "epoch": 0.9774032937571812,
+      "loss": 0.18396136164665222,
+      "loss_ce": 0.020295601338148117,
+      "loss_iou": 1.076249361038208,
+      "loss_num": 0.1640625,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 441116728,
+      "step": 2552
+    },
+    {
+      "epoch": 0.9777862887782459,
+      "grad_norm": 42.41528006791648,
+      "learning_rate": 5e-06,
+      "loss": 0.2129,
+      "num_input_tokens_seen": 441289624,
+      "step": 2553
+    },
+    {
+      "epoch": 0.9777862887782459,
+      "loss": 0.23856016993522644,
+      "loss_ce": 0.021702256053686142,
+      "loss_iou": 1.062626600265503,
+      "loss_num": 0.216796875,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 441289624,
+      "step": 2553
+    },
+    {
+      "epoch": 0.9781692837993106,
+      "grad_norm": 31.993248906786572,
+      "learning_rate": 5e-06,
+      "loss": 0.2483,
+      "num_input_tokens_seen": 441462816,
+      "step": 2554
+    },
+    {
+      "epoch": 0.9781692837993106,
+      "loss": 0.26399481296539307,
+      "loss_ce": 0.021196939051151276,
+      "loss_iou": 1.0644161701202393,
+      "loss_num": 0.2431640625,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 441462816,
+      "step": 2554
+    },
+    {
+      "epoch": 0.9785522788203753,
+      "grad_norm": 24.620900857402145,
+      "learning_rate": 5e-06,
+      "loss": 0.2296,
+      "num_input_tokens_seen": 441635728,
+      "step": 2555
+    },
+    {
+      "epoch": 0.9785522788203753,
+      "loss": 0.20849530398845673,
+      "loss_ce": 0.021178405731916428,
+      "loss_iou": 1.034769892692566,
+      "loss_num": 0.1875,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 441635728,
+      "step": 2555
+    },
+    {
+      "epoch": 0.9789352738414401,
+      "grad_norm": 33.034510362091545,
+      "learning_rate": 5e-06,
+      "loss": 0.18,
+      "num_input_tokens_seen": 441808936,
+      "step": 2556
+    },
+    {
+      "epoch": 0.9789352738414401,
+      "loss": 0.1863858699798584,
+      "loss_ce": 0.022323377430438995,
+      "loss_iou": 1.0175793170928955,
+      "loss_num": 0.1640625,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 441808936,
+      "step": 2556
+    },
+    {
+      "epoch": 0.9793182688625048,
+      "grad_norm": 41.240118072591805,
+      "learning_rate": 5e-06,
+      "loss": 0.2394,
+      "num_input_tokens_seen": 441981856,
+      "step": 2557
+    },
+    {
+      "epoch": 0.9793182688625048,
+      "loss": 0.28613021969795227,
+      "loss_ce": 0.020566251128911972,
+      "loss_iou": 1.1313062906265259,
+      "loss_num": 0.265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 441981856,
+      "step": 2557
+    },
+    {
+      "epoch": 0.9797012638835695,
+      "grad_norm": 22.57894844857494,
+      "learning_rate": 5e-06,
+      "loss": 0.1881,
+      "num_input_tokens_seen": 442154616,
+      "step": 2558
+    },
+    {
+      "epoch": 0.9797012638835695,
+      "loss": 0.2461327314376831,
+      "loss_ce": 0.019692298024892807,
+      "loss_iou": 1.016852617263794,
+      "loss_num": 0.2265625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 442154616,
+      "step": 2558
+    },
+    {
+      "epoch": 0.9800842589046342,
+      "grad_norm": 28.10806253106997,
+      "learning_rate": 5e-06,
+      "loss": 0.2214,
+      "num_input_tokens_seen": 442327432,
+      "step": 2559
+    },
+    {
+      "epoch": 0.9800842589046342,
+      "loss": 0.22265824675559998,
+      "loss_ce": 0.02301226556301117,
+      "loss_iou": 1.0936853885650635,
+      "loss_num": 0.19921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 442327432,
+      "step": 2559
+    },
+    {
+      "epoch": 0.980467253925699,
+      "grad_norm": 25.40437390070875,
+      "learning_rate": 5e-06,
+      "loss": 0.2176,
+      "num_input_tokens_seen": 442500600,
+      "step": 2560
+    },
+    {
+      "epoch": 0.980467253925699,
+      "loss": 0.1724667251110077,
+      "loss_ce": 0.020977461710572243,
+      "loss_iou": 1.0222909450531006,
+      "loss_num": 0.1513671875,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 442500600,
+      "step": 2560
+    },
+    {
+      "epoch": 0.9808502489467636,
+      "grad_norm": 28.458831937750002,
+      "learning_rate": 5e-06,
+      "loss": 0.1892,
+      "num_input_tokens_seen": 442673584,
+      "step": 2561
+    },
+    {
+      "epoch": 0.9808502489467636,
+      "loss": 0.20998351275920868,
+      "loss_ce": 0.02034728042781353,
+      "loss_iou": 1.0627766847610474,
+      "loss_num": 0.189453125,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 442673584,
+      "step": 2561
+    },
+    {
+      "epoch": 0.9812332439678284,
+      "grad_norm": 25.891144878268932,
+      "learning_rate": 5e-06,
+      "loss": 0.1874,
+      "num_input_tokens_seen": 442846120,
+      "step": 2562
+    },
+    {
+      "epoch": 0.9812332439678284,
+      "loss": 0.1612968146800995,
+      "loss_ce": 0.02061077207326889,
+      "loss_iou": 1.0302972793579102,
+      "loss_num": 0.140625,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 442846120,
+      "step": 2562
+    },
+    {
+      "epoch": 0.9816162389888932,
+      "grad_norm": 27.943424729093966,
+      "learning_rate": 5e-06,
+      "loss": 0.1868,
+      "num_input_tokens_seen": 443019264,
+      "step": 2563
+    },
+    {
+      "epoch": 0.9816162389888932,
+      "loss": 0.18382543325424194,
+      "loss_ce": 0.020617440342903137,
+      "loss_iou": 1.006981372833252,
+      "loss_num": 0.1630859375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 443019264,
+      "step": 2563
+    },
+    {
+      "epoch": 0.9819992340099579,
+      "grad_norm": 24.980205216177644,
+      "learning_rate": 5e-06,
+      "loss": 0.2163,
+      "num_input_tokens_seen": 443192032,
+      "step": 2564
+    },
+    {
+      "epoch": 0.9819992340099579,
+      "loss": 0.16928371787071228,
+      "loss_ce": 0.021151382476091385,
+      "loss_iou": 1.0022011995315552,
+      "loss_num": 0.1484375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 443192032,
+      "step": 2564
+    },
+    {
+      "epoch": 0.9823822290310226,
+      "grad_norm": 33.59013075971121,
+      "learning_rate": 5e-06,
+      "loss": 0.138,
+      "num_input_tokens_seen": 443365096,
+      "step": 2565
+    },
+    {
+      "epoch": 0.9823822290310226,
+      "loss": 0.12297128140926361,
+      "loss_ce": 0.020798426121473312,
+      "loss_iou": 1.00077486038208,
+      "loss_num": 0.10205078125,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 443365096,
+      "step": 2565
+    },
+    {
+      "epoch": 0.9827652240520873,
+      "grad_norm": 41.872255241259964,
+      "learning_rate": 5e-06,
+      "loss": 0.2303,
+      "num_input_tokens_seen": 443538264,
+      "step": 2566
+    },
+    {
+      "epoch": 0.9827652240520873,
+      "loss": 0.2431665062904358,
+      "loss_ce": 0.02093750610947609,
+      "loss_iou": 1.0195742845535278,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 443538264,
+      "step": 2566
+    },
+    {
+      "epoch": 0.9831482190731521,
+      "grad_norm": 39.45379360719268,
+      "learning_rate": 5e-06,
+      "loss": 0.1935,
+      "num_input_tokens_seen": 443711328,
+      "step": 2567
+    },
+    {
+      "epoch": 0.9831482190731521,
+      "loss": 0.22139954566955566,
+      "loss_ce": 0.021143201738595963,
+      "loss_iou": 1.073280692100525,
+      "loss_num": 0.2001953125,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 443711328,
+      "step": 2567
+    },
+    {
+      "epoch": 0.9835312140942167,
+      "grad_norm": 23.73753841797015,
+      "learning_rate": 5e-06,
+      "loss": 0.1913,
+      "num_input_tokens_seen": 443884392,
+      "step": 2568
+    },
+    {
+      "epoch": 0.9835312140942167,
+      "loss": 0.19663631916046143,
+      "loss_ce": 0.02070249244570732,
+      "loss_iou": 1.0560352802276611,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 443884392,
+      "step": 2568
+    },
+    {
+      "epoch": 0.9839142091152815,
+      "grad_norm": 24.083009439187595,
+      "learning_rate": 5e-06,
+      "loss": 0.2446,
+      "num_input_tokens_seen": 444057472,
+      "step": 2569
+    },
+    {
+      "epoch": 0.9839142091152815,
+      "loss": 0.2394472062587738,
+      "loss_ce": 0.019598571583628654,
+      "loss_iou": 1.0479086637496948,
+      "loss_num": 0.2197265625,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 444057472,
+      "step": 2569
+    },
+    {
+      "epoch": 0.9842972041363462,
+      "grad_norm": 28.394723621777164,
+      "learning_rate": 5e-06,
+      "loss": 0.1834,
+      "num_input_tokens_seen": 444230328,
+      "step": 2570
+    },
+    {
+      "epoch": 0.9842972041363462,
+      "loss": 0.14444415271282196,
+      "loss_ce": 0.02176349051296711,
+      "loss_iou": 1.0061496496200562,
+      "loss_num": 0.12255859375,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 444230328,
+      "step": 2570
+    },
+    {
+      "epoch": 0.984680199157411,
+      "grad_norm": 34.516951239161614,
+      "learning_rate": 5e-06,
+      "loss": 0.2691,
+      "num_input_tokens_seen": 444403144,
+      "step": 2571
+    },
+    {
+      "epoch": 0.984680199157411,
+      "loss": 0.2842923104763031,
+      "loss_ce": 0.021047698333859444,
+      "loss_iou": 1.0349102020263672,
+      "loss_num": 0.263671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 444403144,
+      "step": 2571
+    },
+    {
+      "epoch": 0.9850631941784757,
+      "grad_norm": 22.553912274036357,
+      "learning_rate": 5e-06,
+      "loss": 0.1653,
+      "num_input_tokens_seen": 444576072,
+      "step": 2572
+    },
+    {
+      "epoch": 0.9850631941784757,
+      "loss": 0.1860370934009552,
+      "loss_ce": 0.022279776632785797,
+      "loss_iou": 1.0185402631759644,
+      "loss_num": 0.1640625,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 444576072,
+      "step": 2572
+    },
+    {
+      "epoch": 0.9854461891995404,
+      "grad_norm": 28.51811051258652,
+      "learning_rate": 5e-06,
+      "loss": 0.18,
+      "num_input_tokens_seen": 444748992,
+      "step": 2573
+    },
+    {
+      "epoch": 0.9854461891995404,
+      "loss": 0.18080393970012665,
+      "loss_ce": 0.019671138375997543,
+      "loss_iou": 1.0438616275787354,
+      "loss_num": 0.1611328125,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 444748992,
+      "step": 2573
+    },
+    {
+      "epoch": 0.9858291842206052,
+      "grad_norm": 21.75087601397511,
+      "learning_rate": 5e-06,
+      "loss": 0.2008,
+      "num_input_tokens_seen": 444922000,
+      "step": 2574
+    },
+    {
+      "epoch": 0.9858291842206052,
+      "loss": 0.21475309133529663,
+      "loss_ce": 0.02060026302933693,
+      "loss_iou": 1.039602518081665,
+      "loss_num": 0.1943359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 444922000,
+      "step": 2574
+    },
+    {
+      "epoch": 0.9862121792416698,
+      "grad_norm": 31.05637531059788,
+      "learning_rate": 5e-06,
+      "loss": 0.2755,
+      "num_input_tokens_seen": 445095000,
+      "step": 2575
+    },
+    {
+      "epoch": 0.9862121792416698,
+      "loss": 0.21724437177181244,
+      "loss_ce": 0.01833079382777214,
+      "loss_iou": 4.5821070671081543e-07,
+      "loss_num": 0.19921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 445095000,
+      "step": 2575
+    },
+    {
+      "epoch": 0.9865951742627346,
+      "grad_norm": 31.039617786281315,
+      "learning_rate": 5e-06,
+      "loss": 0.1922,
+      "num_input_tokens_seen": 445267944,
+      "step": 2576
+    },
+    {
+      "epoch": 0.9865951742627346,
+      "loss": 0.20234927535057068,
+      "loss_ce": 0.018694482743740082,
+      "loss_iou": 1.0001885890960693,
+      "loss_num": 0.18359375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 445267944,
+      "step": 2576
+    },
+    {
+      "epoch": 0.9869781692837993,
+      "grad_norm": 27.339199641718203,
+      "learning_rate": 5e-06,
+      "loss": 0.2087,
+      "num_input_tokens_seen": 445440944,
+      "step": 2577
+    },
+    {
+      "epoch": 0.9869781692837993,
+      "loss": 0.17758116126060486,
+      "loss_ce": 0.02041563019156456,
+      "loss_iou": 1.0020227432250977,
+      "loss_num": 0.1572265625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 445440944,
+      "step": 2577
+    },
+    {
+      "epoch": 0.9873611643048641,
+      "grad_norm": 25.9871456221162,
+      "learning_rate": 5e-06,
+      "loss": 0.1379,
+      "num_input_tokens_seen": 445614016,
+      "step": 2578
+    },
+    {
+      "epoch": 0.9873611643048641,
+      "loss": 0.13205954432487488,
+      "loss_ce": 0.02195211499929428,
+      "loss_iou": 1.0053297281265259,
+      "loss_num": 0.1103515625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 445614016,
+      "step": 2578
+    },
+    {
+      "epoch": 0.9877441593259287,
+      "grad_norm": 29.412608075507368,
+      "learning_rate": 5e-06,
+      "loss": 0.2376,
+      "num_input_tokens_seen": 445786792,
+      "step": 2579
+    },
+    {
+      "epoch": 0.9877441593259287,
+      "loss": 0.1717311441898346,
+      "loss_ce": 0.021157432347536087,
+      "loss_iou": 1.031954050064087,
+      "loss_num": 0.150390625,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 445786792,
+      "step": 2579
+    },
+    {
+      "epoch": 0.9881271543469935,
+      "grad_norm": 24.35383108838625,
+      "learning_rate": 5e-06,
+      "loss": 0.1826,
+      "num_input_tokens_seen": 445959464,
+      "step": 2580
+    },
+    {
+      "epoch": 0.9881271543469935,
+      "loss": 0.1787584125995636,
+      "loss_ce": 0.020250104367733,
+      "loss_iou": 1.003459095954895,
+      "loss_num": 0.158203125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 445959464,
+      "step": 2580
+    },
+    {
+      "epoch": 0.9885101493680583,
+      "grad_norm": 26.71496003431928,
+      "learning_rate": 5e-06,
+      "loss": 0.2517,
+      "num_input_tokens_seen": 446132280,
+      "step": 2581
+    },
+    {
+      "epoch": 0.9885101493680583,
+      "loss": 0.2053559273481369,
+      "loss_ce": 0.020541464909911156,
+      "loss_iou": 1.047257900238037,
+      "loss_num": 0.1845703125,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 446132280,
+      "step": 2581
+    },
+    {
+      "epoch": 0.9888931443891229,
+      "grad_norm": 22.444135524638458,
+      "learning_rate": 5e-06,
+      "loss": 0.1933,
+      "num_input_tokens_seen": 446305624,
+      "step": 2582
+    },
+    {
+      "epoch": 0.9888931443891229,
+      "loss": 0.13972869515419006,
+      "loss_ce": 0.01991669088602066,
+      "loss_iou": 1.0086948871612549,
+      "loss_num": 0.11962890625,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 446305624,
+      "step": 2582
+    },
+    {
+      "epoch": 0.9892761394101877,
+      "grad_norm": 45.64098443506414,
+      "learning_rate": 5e-06,
+      "loss": 0.2166,
+      "num_input_tokens_seen": 446478768,
+      "step": 2583
+    },
+    {
+      "epoch": 0.9892761394101877,
+      "loss": 0.17628300189971924,
+      "loss_ce": 0.02168095111846924,
+      "loss_iou": 1.0157346725463867,
+      "loss_num": 0.154296875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 446478768,
+      "step": 2583
+    },
+    {
+      "epoch": 0.9896591344312524,
+      "grad_norm": 39.25585080550029,
+      "learning_rate": 5e-06,
+      "loss": 0.2946,
+      "num_input_tokens_seen": 446652000,
+      "step": 2584
+    },
+    {
+      "epoch": 0.9896591344312524,
+      "loss": 0.27448806166648865,
+      "loss_ce": 0.022565443068742752,
+      "loss_iou": 1.0934661626815796,
+      "loss_num": 0.251953125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 446652000,
+      "step": 2584
+    },
+    {
+      "epoch": 0.9900421294523171,
+      "grad_norm": 43.83002101259563,
+      "learning_rate": 5e-06,
+      "loss": 0.2272,
+      "num_input_tokens_seen": 446824616,
+      "step": 2585
+    },
+    {
+      "epoch": 0.9900421294523171,
+      "loss": 0.24637830257415771,
+      "loss_ce": 0.02048720419406891,
+      "loss_iou": 0.9968891739845276,
+      "loss_num": 0.2255859375,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 446824616,
+      "step": 2585
+    },
+    {
+      "epoch": 0.9904251244733818,
+      "grad_norm": 20.64342158416282,
+      "learning_rate": 5e-06,
+      "loss": 0.2007,
+      "num_input_tokens_seen": 446997200,
+      "step": 2586
+    },
+    {
+      "epoch": 0.9904251244733818,
+      "loss": 0.2109462469816208,
+      "loss_ce": 0.020943813025951385,
+      "loss_iou": 1.0423662662506104,
+      "loss_num": 0.1904296875,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 446997200,
+      "step": 2586
+    },
+    {
+      "epoch": 0.9908081194944466,
+      "grad_norm": 56.14326889968074,
+      "learning_rate": 5e-06,
+      "loss": 0.2198,
+      "num_input_tokens_seen": 447170112,
+      "step": 2587
+    },
+    {
+      "epoch": 0.9908081194944466,
+      "loss": 0.23894494771957397,
+      "loss_ce": 0.02111048437654972,
+      "loss_iou": 1.1187589168548584,
+      "loss_num": 0.2177734375,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 447170112,
+      "step": 2587
+    },
+    {
+      "epoch": 0.9911911145155113,
+      "grad_norm": 35.00610524319135,
+      "learning_rate": 5e-06,
+      "loss": 0.2362,
+      "num_input_tokens_seen": 447343136,
+      "step": 2588
+    },
+    {
+      "epoch": 0.9911911145155113,
+      "loss": 0.21689264476299286,
+      "loss_ce": 0.022190498188138008,
+      "loss_iou": 1.0384045839309692,
+      "loss_num": 0.1943359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 447343136,
+      "step": 2588
+    },
+    {
+      "epoch": 0.991574109536576,
+      "grad_norm": 28.2697251072908,
+      "learning_rate": 5e-06,
+      "loss": 0.1733,
+      "num_input_tokens_seen": 447516080,
+      "step": 2589
+    },
+    {
+      "epoch": 0.991574109536576,
+      "loss": 0.15879778563976288,
+      "loss_ce": 0.02085833251476288,
+      "loss_iou": 1.036719799041748,
+      "loss_num": 0.1376953125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 447516080,
+      "step": 2589
+    },
+    {
+      "epoch": 0.9919571045576407,
+      "grad_norm": 39.47081502445282,
+      "learning_rate": 5e-06,
+      "loss": 0.2003,
+      "num_input_tokens_seen": 447689096,
+      "step": 2590
+    },
+    {
+      "epoch": 0.9919571045576407,
+      "loss": 0.17974254488945007,
+      "loss_ce": 0.022943221032619476,
+      "loss_iou": 1.0426125526428223,
+      "loss_num": 0.1572265625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 447689096,
+      "step": 2590
+    },
+    {
+      "epoch": 0.9923400995787055,
+      "grad_norm": 63.24975708634823,
+      "learning_rate": 5e-06,
+      "loss": 0.2663,
+      "num_input_tokens_seen": 447862232,
+      "step": 2591
+    },
+    {
+      "epoch": 0.9923400995787055,
+      "loss": 0.33286064863204956,
+      "loss_ce": 0.02280203439295292,
+      "loss_iou": 1.0508623123168945,
+      "loss_num": 0.310546875,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 447862232,
+      "step": 2591
+    },
+    {
+      "epoch": 0.9927230945997702,
+      "grad_norm": 24.562342961105927,
+      "learning_rate": 5e-06,
+      "loss": 0.192,
+      "num_input_tokens_seen": 448035240,
+      "step": 2592
+    },
+    {
+      "epoch": 0.9927230945997702,
+      "loss": 0.2081054151058197,
+      "loss_ce": 0.021124217659235,
+      "loss_iou": 1.079519271850586,
+      "loss_num": 0.1865234375,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 448035240,
+      "step": 2592
+    },
+    {
+      "epoch": 0.9931060896208349,
+      "grad_norm": 39.36316765832567,
+      "learning_rate": 5e-06,
+      "loss": 0.2356,
+      "num_input_tokens_seen": 448207928,
+      "step": 2593
+    },
+    {
+      "epoch": 0.9931060896208349,
+      "loss": 0.223251074552536,
+      "loss_ce": 0.019576773047447205,
+      "loss_iou": 1.0996551513671875,
+      "loss_num": 0.2041015625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 448207928,
+      "step": 2593
+    },
+    {
+      "epoch": 0.9934890846418997,
+      "grad_norm": 34.030114413781845,
+      "learning_rate": 5e-06,
+      "loss": 0.2907,
+      "num_input_tokens_seen": 448380904,
+      "step": 2594
+    },
+    {
+      "epoch": 0.9934890846418997,
+      "loss": 0.31716522574424744,
+      "loss_ce": 0.021724559366703033,
+      "loss_iou": 1.1569843292236328,
+      "loss_num": 0.294921875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 448380904,
+      "step": 2594
+    },
+    {
+      "epoch": 0.9938720796629644,
+      "grad_norm": 45.04046472214304,
+      "learning_rate": 5e-06,
+      "loss": 0.2346,
+      "num_input_tokens_seen": 448553768,
+      "step": 2595
+    },
+    {
+      "epoch": 0.9938720796629644,
+      "loss": 0.21021856367588043,
+      "loss_ce": 0.01966679096221924,
+      "loss_iou": 1.0189051628112793,
+      "loss_num": 0.1904296875,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 448553768,
+      "step": 2595
+    },
+    {
+      "epoch": 0.9942550746840291,
+      "grad_norm": 25.264268495162163,
+      "learning_rate": 5e-06,
+      "loss": 0.2099,
+      "num_input_tokens_seen": 448726416,
+      "step": 2596
+    },
+    {
+      "epoch": 0.9942550746840291,
+      "loss": 0.21929392218589783,
+      "loss_ce": 0.01989205926656723,
+      "loss_iou": 0.8835408687591553,
+      "loss_num": 0.19921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 448726416,
+      "step": 2596
+    },
+    {
+      "epoch": 0.9946380697050938,
+      "grad_norm": 20.300640982375388,
+      "learning_rate": 5e-06,
+      "loss": 0.1662,
+      "num_input_tokens_seen": 448899352,
+      "step": 2597
+    },
+    {
+      "epoch": 0.9946380697050938,
+      "loss": 0.16648760437965393,
+      "loss_ce": 0.01927081122994423,
+      "loss_iou": 1.029667854309082,
+      "loss_num": 0.1474609375,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 448899352,
+      "step": 2597
+    },
+    {
+      "epoch": 0.9950210647261586,
+      "grad_norm": 43.651510203906085,
+      "learning_rate": 5e-06,
+      "loss": 0.2064,
+      "num_input_tokens_seen": 449072120,
+      "step": 2598
+    },
+    {
+      "epoch": 0.9950210647261586,
+      "loss": 0.23701754212379456,
+      "loss_ce": 0.020403757691383362,
+      "loss_iou": 1.1127095222473145,
+      "loss_num": 0.216796875,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 449072120,
+      "step": 2598
+    },
+    {
+      "epoch": 0.9954040597472232,
+      "grad_norm": 22.81379268659071,
+      "learning_rate": 5e-06,
+      "loss": 0.2165,
+      "num_input_tokens_seen": 449244872,
+      "step": 2599
+    },
+    {
+      "epoch": 0.9954040597472232,
+      "loss": 0.21868206560611725,
+      "loss_ce": 0.019890563562512398,
+      "loss_iou": 0.9505128860473633,
+      "loss_num": 0.19921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 449244872,
+      "step": 2599
+    },
+    {
+      "epoch": 0.995787054768288,
+      "grad_norm": 29.275378725887922,
+      "learning_rate": 5e-06,
+      "loss": 0.194,
+      "num_input_tokens_seen": 449417752,
+      "step": 2600
+    },
+    {
+      "epoch": 0.995787054768288,
+      "loss": 0.15811039507389069,
+      "loss_ce": 0.021635789424180984,
+      "loss_iou": 1.0020790100097656,
+      "loss_num": 0.13671875,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 449417752,
+      "step": 2600
+    },
+    {
+      "epoch": 0.9961700497893528,
+      "grad_norm": 23.962375684845348,
+      "learning_rate": 5e-06,
+      "loss": 0.1902,
+      "num_input_tokens_seen": 449590616,
+      "step": 2601
+    },
+    {
+      "epoch": 0.9961700497893528,
+      "loss": 0.14931534230709076,
+      "loss_ce": 0.02065323479473591,
+      "loss_iou": 1.0113236904144287,
+      "loss_num": 0.12890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 449590616,
+      "step": 2601
+    },
+    {
+      "epoch": 0.9965530448104175,
+      "grad_norm": 21.851142549349564,
+      "learning_rate": 5e-06,
+      "loss": 0.1965,
+      "num_input_tokens_seen": 449763296,
+      "step": 2602
+    },
+    {
+      "epoch": 0.9965530448104175,
+      "loss": 0.17328669130802155,
+      "loss_ce": 0.019142404198646545,
+      "loss_iou": 1.076406717300415,
+      "loss_num": 0.154296875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 449763296,
+      "step": 2602
+    },
+    {
+      "epoch": 0.9969360398314822,
+      "grad_norm": 20.320872643821353,
+      "learning_rate": 5e-06,
+      "loss": 0.2369,
+      "num_input_tokens_seen": 449936216,
+      "step": 2603
+    },
+    {
+      "epoch": 0.9969360398314822,
+      "loss": 0.2127036452293396,
+      "loss_ce": 0.022579146549105644,
+      "loss_iou": 1.0166010856628418,
+      "loss_num": 0.1904296875,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 449936216,
+      "step": 2603
+    },
+    {
+      "epoch": 0.9973190348525469,
+      "grad_norm": 21.00771430933762,
+      "learning_rate": 5e-06,
+      "loss": 0.1887,
+      "num_input_tokens_seen": 450109200,
+      "step": 2604
+    },
+    {
+      "epoch": 0.9973190348525469,
+      "loss": 0.1232326477766037,
+      "loss_ce": 0.023073958232998848,
+      "loss_iou": 1.0016324520111084,
+      "loss_num": 0.10009765625,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 450109200,
+      "step": 2604
+    },
+    {
+      "epoch": 0.9977020298736117,
+      "grad_norm": 26.40707134566965,
+      "learning_rate": 5e-06,
+      "loss": 0.1996,
+      "num_input_tokens_seen": 450282088,
+      "step": 2605
+    },
+    {
+      "epoch": 0.9977020298736117,
+      "loss": 0.16095593571662903,
+      "loss_ce": 0.01874401792883873,
+      "loss_iou": 1.0023956298828125,
+      "loss_num": 0.142578125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 450282088,
+      "step": 2605
+    },
+    {
+      "epoch": 0.9980850248946763,
+      "grad_norm": 22.127976472193634,
+      "learning_rate": 5e-06,
+      "loss": 0.2074,
+      "num_input_tokens_seen": 450455496,
+      "step": 2606
+    },
+    {
+      "epoch": 0.9980850248946763,
+      "loss": 0.23797369003295898,
+      "loss_ce": 0.020688533782958984,
+      "loss_iou": 1.4147803783416748,
+      "loss_num": 0.216796875,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 450455496,
+      "step": 2606
+    },
+    {
+      "epoch": 0.9984680199157411,
+      "grad_norm": 24.30602607449792,
+      "learning_rate": 5e-06,
+      "loss": 0.1747,
+      "num_input_tokens_seen": 450628328,
+      "step": 2607
+    },
+    {
+      "epoch": 0.9984680199157411,
+      "loss": 0.20958831906318665,
+      "loss_ce": 0.020501410588622093,
+      "loss_iou": 1.335038185119629,
+      "loss_num": 0.189453125,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 450628328,
+      "step": 2607
+    },
+    {
+      "epoch": 0.9988510149368058,
+      "grad_norm": 24.772227090561294,
+      "learning_rate": 5e-06,
+      "loss": 0.2028,
+      "num_input_tokens_seen": 450801640,
+      "step": 2608
+    },
+    {
+      "epoch": 0.9988510149368058,
+      "loss": 0.2230081856250763,
+      "loss_ce": 0.021714244037866592,
+      "loss_iou": 1.085056185722351,
+      "loss_num": 0.201171875,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 450801640,
+      "step": 2608
+    },
+    {
+      "epoch": 0.9992340099578706,
+      "grad_norm": 31.85535411790147,
+      "learning_rate": 5e-06,
+      "loss": 0.191,
+      "num_input_tokens_seen": 450974184,
+      "step": 2609
+    },
+    {
+      "epoch": 0.9992340099578706,
+      "loss": 0.2219005823135376,
+      "loss_ce": 0.019446972757577896,
+      "loss_iou": 1.210686445236206,
+      "loss_num": 0.2021484375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 450974184,
+      "step": 2609
+    },
+    {
+      "epoch": 0.9996170049789352,
+      "grad_norm": 27.50167474855051,
+      "learning_rate": 5e-06,
+      "loss": 0.2368,
+      "num_input_tokens_seen": 451146848,
+      "step": 2610
+    },
+    {
+      "epoch": 0.9996170049789352,
+      "loss": 0.2298506498336792,
+      "loss_ce": 0.020255915820598602,
+      "loss_iou": 1.0357897281646729,
+      "loss_num": 0.2099609375,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 451146848,
+      "step": 2610
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 23.82388520113061,
+      "learning_rate": 5e-06,
+      "loss": 0.1789,
+      "num_input_tokens_seen": 451319744,
+      "step": 2611
+    },
+    {
+      "epoch": 1.0,
+      "loss": 0.19941318035125732,
+      "loss_ce": 0.021495692431926727,
+      "loss_iou": 1.0034189224243164,
+      "loss_num": 0.177734375,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 451319744,
+      "step": 2611
+    },
+    {
+      "epoch": 1.0003829950210648,
+      "grad_norm": 20.103405756766083,
+      "learning_rate": 5e-06,
+      "loss": 0.1692,
+      "num_input_tokens_seen": 451492768,
+      "step": 2612
+    },
+    {
+      "epoch": 1.0003829950210648,
+      "loss": 0.16814805567264557,
+      "loss_ce": 0.019527453929185867,
+      "loss_iou": 1.0039558410644531,
+      "loss_num": 0.1484375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 451492768,
+      "step": 2612
+    },
+    {
+      "epoch": 1.0007659900421295,
+      "grad_norm": 30.33221650252089,
+      "learning_rate": 5e-06,
+      "loss": 0.1691,
+      "num_input_tokens_seen": 451665952,
+      "step": 2613
+    },
+    {
+      "epoch": 1.0007659900421295,
+      "loss": 0.16631048917770386,
+      "loss_ce": 0.022145451977849007,
+      "loss_iou": 1.0690546035766602,
+      "loss_num": 0.14453125,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 451665952,
+      "step": 2613
+    },
+    {
+      "epoch": 1.0011489850631943,
+      "grad_norm": 29.113994599823556,
+      "learning_rate": 5e-06,
+      "loss": 0.1324,
+      "num_input_tokens_seen": 451838648,
+      "step": 2614
+    },
+    {
+      "epoch": 1.0011489850631943,
+      "loss": 0.14101362228393555,
+      "loss_ce": 0.018210891634225845,
+      "loss_iou": 1.007625699043274,
+      "loss_num": 0.123046875,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 451838648,
+      "step": 2614
+    },
+    {
+      "epoch": 1.0015319800842588,
+      "grad_norm": 26.460448235571473,
+      "learning_rate": 5e-06,
+      "loss": 0.169,
+      "num_input_tokens_seen": 452011384,
+      "step": 2615
+    },
+    {
+      "epoch": 1.0015319800842588,
+      "loss": 0.17002080380916595,
+      "loss_ce": 0.019508112221956253,
+      "loss_iou": 1.0012292861938477,
+      "loss_num": 0.150390625,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 452011384,
+      "step": 2615
+    },
+    {
+      "epoch": 1.0019149751053236,
+      "grad_norm": 22.107495489047828,
+      "learning_rate": 5e-06,
+      "loss": 0.1728,
+      "num_input_tokens_seen": 452183680,
+      "step": 2616
+    },
+    {
+      "epoch": 1.0019149751053236,
+      "loss": 0.14123472571372986,
+      "loss_ce": 0.02011045441031456,
+      "loss_iou": 1.0368467569351196,
+      "loss_num": 0.12109375,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 452183680,
+      "step": 2616
+    },
+    {
+      "epoch": 1.0022979701263883,
+      "grad_norm": 24.84546531113666,
+      "learning_rate": 5e-06,
+      "loss": 0.1521,
+      "num_input_tokens_seen": 452356632,
+      "step": 2617
+    },
+    {
+      "epoch": 1.0022979701263883,
+      "loss": 0.13976626098155975,
+      "loss_ce": 0.01964908093214035,
+      "loss_iou": 1.001399040222168,
+      "loss_num": 0.1201171875,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 452356632,
+      "step": 2617
+    },
+    {
+      "epoch": 1.002680965147453,
+      "grad_norm": 21.598581182574627,
+      "learning_rate": 5e-06,
+      "loss": 0.1558,
+      "num_input_tokens_seen": 452529648,
+      "step": 2618
+    },
+    {
+      "epoch": 1.002680965147453,
+      "loss": 0.14163951575756073,
+      "loss_ce": 0.01844004914164543,
+      "loss_iou": 1.0433669090270996,
+      "loss_num": 0.123046875,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 452529648,
+      "step": 2618
+    },
+    {
+      "epoch": 1.0030639601685178,
+      "grad_norm": 25.008835471138518,
+      "learning_rate": 5e-06,
+      "loss": 0.2027,
+      "num_input_tokens_seen": 452702528,
+      "step": 2619
+    },
+    {
+      "epoch": 1.0030639601685178,
+      "loss": 0.14462541043758392,
+      "loss_ce": 0.018099535256624222,
+      "loss_iou": 1.0006023645401,
+      "loss_num": 0.126953125,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 452702528,
+      "step": 2619
+    },
+    {
+      "epoch": 1.0034469551895826,
+      "grad_norm": 15.996517487548525,
+      "learning_rate": 5e-06,
+      "loss": 0.203,
+      "num_input_tokens_seen": 452875272,
+      "step": 2620
+    },
+    {
+      "epoch": 1.0034469551895826,
+      "loss": 0.22338154911994934,
+      "loss_ce": 0.02019551210105419,
+      "loss_iou": 1.002182960510254,
+      "loss_num": 0.203125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 452875272,
+      "step": 2620
+    },
+    {
+      "epoch": 1.0038299502106474,
+      "grad_norm": 26.077149324673446,
+      "learning_rate": 5e-06,
+      "loss": 0.1983,
+      "num_input_tokens_seen": 453048480,
+      "step": 2621
+    },
+    {
+      "epoch": 1.0038299502106474,
+      "loss": 0.266177237033844,
+      "loss_ce": 0.02118212729692459,
+      "loss_iou": 1.065863847732544,
+      "loss_num": 0.2451171875,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 453048480,
+      "step": 2621
+    },
+    {
+      "epoch": 1.004212945231712,
+      "grad_norm": 24.889998973498113,
+      "learning_rate": 5e-06,
+      "loss": 0.1617,
+      "num_input_tokens_seen": 453221336,
+      "step": 2622
+    },
+    {
+      "epoch": 1.004212945231712,
+      "loss": 0.20550508797168732,
+      "loss_ce": 0.02093476988375187,
+      "loss_iou": 1.0505411624908447,
+      "loss_num": 0.1845703125,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 453221336,
+      "step": 2622
+    },
+    {
+      "epoch": 1.0045959402527767,
+      "grad_norm": 18.0054008237673,
+      "learning_rate": 5e-06,
+      "loss": 0.1792,
+      "num_input_tokens_seen": 453394408,
+      "step": 2623
+    },
+    {
+      "epoch": 1.0045959402527767,
+      "loss": 0.13319995999336243,
+      "loss_ce": 0.020590096712112427,
+      "loss_iou": 1.0009472370147705,
+      "loss_num": 0.11279296875,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 453394408,
+      "step": 2623
+    },
+    {
+      "epoch": 1.0049789352738414,
+      "grad_norm": 21.48118129655174,
+      "learning_rate": 5e-06,
+      "loss": 0.168,
+      "num_input_tokens_seen": 453567472,
+      "step": 2624
+    },
+    {
+      "epoch": 1.0049789352738414,
+      "loss": 0.16904892027378082,
+      "loss_ce": 0.02048935554921627,
+      "loss_iou": 1.0616116523742676,
+      "loss_num": 0.1484375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 453567472,
+      "step": 2624
+    },
+    {
+      "epoch": 1.0053619302949062,
+      "grad_norm": 24.030889297447565,
+      "learning_rate": 5e-06,
+      "loss": 0.1929,
+      "num_input_tokens_seen": 453740280,
+      "step": 2625
+    },
+    {
+      "epoch": 1.0053619302949062,
+      "loss": 0.16246747970581055,
+      "loss_ce": 0.019828319549560547,
+      "loss_iou": 1.0066397190093994,
+      "loss_num": 0.142578125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 453740280,
+      "step": 2625
+    },
+    {
+      "epoch": 1.005744925315971,
+      "grad_norm": 29.232199595915233,
+      "learning_rate": 5e-06,
+      "loss": 0.1712,
+      "num_input_tokens_seen": 453913384,
+      "step": 2626
+    },
+    {
+      "epoch": 1.005744925315971,
+      "loss": 0.20079506933689117,
+      "loss_ce": 0.020314104855060577,
+      "loss_iou": 1.0599197149276733,
+      "loss_num": 0.1806640625,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 453913384,
+      "step": 2626
+    },
+    {
+      "epoch": 1.0061279203370357,
+      "grad_norm": 31.868805292436047,
+      "learning_rate": 5e-06,
+      "loss": 0.2713,
+      "num_input_tokens_seen": 454086320,
+      "step": 2627
+    },
+    {
+      "epoch": 1.0061279203370357,
+      "loss": 0.254930704832077,
+      "loss_ce": 0.020067419856786728,
+      "loss_iou": 1.2221109867095947,
+      "loss_num": 0.234375,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 454086320,
+      "step": 2627
+    },
+    {
+      "epoch": 1.0065109153581004,
+      "grad_norm": 25.275369950358368,
+      "learning_rate": 5e-06,
+      "loss": 0.1447,
+      "num_input_tokens_seen": 454259512,
+      "step": 2628
+    },
+    {
+      "epoch": 1.0065109153581004,
+      "loss": 0.16165083646774292,
+      "loss_ce": 0.021117383614182472,
+      "loss_iou": 1.0255322456359863,
+      "loss_num": 0.140625,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 454259512,
+      "step": 2628
+    },
+    {
+      "epoch": 1.006893910379165,
+      "grad_norm": 28.217706011517446,
+      "learning_rate": 5e-06,
+      "loss": 0.2765,
+      "num_input_tokens_seen": 454432672,
+      "step": 2629
+    },
+    {
+      "epoch": 1.006893910379165,
+      "loss": 0.22138825058937073,
+      "loss_ce": 0.02009430341422558,
+      "loss_iou": 1.1180342435836792,
+      "loss_num": 0.201171875,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 454432672,
+      "step": 2629
+    },
+    {
+      "epoch": 1.0072769054002297,
+      "grad_norm": 23.197617638130787,
+      "learning_rate": 5e-06,
+      "loss": 0.2146,
+      "num_input_tokens_seen": 454605624,
+      "step": 2630
+    },
+    {
+      "epoch": 1.0072769054002297,
+      "loss": 0.2131013423204422,
+      "loss_ce": 0.0229157917201519,
+      "loss_iou": 1.0519930124282837,
+      "loss_num": 0.1904296875,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 454605624,
+      "step": 2630
+    },
+    {
+      "epoch": 1.0076599004212945,
+      "grad_norm": 23.264612619001955,
+      "learning_rate": 5e-06,
+      "loss": 0.1406,
+      "num_input_tokens_seen": 454778656,
+      "step": 2631
+    },
+    {
+      "epoch": 1.0076599004212945,
+      "loss": 0.12333838641643524,
+      "loss_ce": 0.01970069482922554,
+      "loss_iou": 1.0003728866577148,
+      "loss_num": 0.103515625,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 454778656,
+      "step": 2631
+    },
+    {
+      "epoch": 1.0080428954423593,
+      "grad_norm": 27.308085755809195,
+      "learning_rate": 5e-06,
+      "loss": 0.1873,
+      "num_input_tokens_seen": 454952344,
+      "step": 2632
+    },
+    {
+      "epoch": 1.0080428954423593,
+      "loss": 0.18797634541988373,
+      "loss_ce": 0.02202175185084343,
+      "loss_iou": 1.0183154344558716,
+      "loss_num": 0.166015625,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 454952344,
+      "step": 2632
+    },
+    {
+      "epoch": 1.008425890463424,
+      "grad_norm": 51.13928430698843,
+      "learning_rate": 5e-06,
+      "loss": 0.1993,
+      "num_input_tokens_seen": 455125024,
+      "step": 2633
+    },
+    {
+      "epoch": 1.008425890463424,
+      "loss": 0.23166146874427795,
+      "loss_ce": 0.020723965018987656,
+      "loss_iou": 1.5644888877868652,
+      "loss_num": 0.2109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 455125024,
+      "step": 2633
+    },
+    {
+      "epoch": 1.0088088854844888,
+      "grad_norm": 36.3187290682828,
+      "learning_rate": 5e-06,
+      "loss": 0.1898,
+      "num_input_tokens_seen": 455298184,
+      "step": 2634
+    },
+    {
+      "epoch": 1.0088088854844888,
+      "loss": 0.1815185546875,
+      "loss_ce": 0.020202644169330597,
+      "loss_iou": 1.0271393060684204,
+      "loss_num": 0.1611328125,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 455298184,
+      "step": 2634
+    },
+    {
+      "epoch": 1.0091918805055535,
+      "grad_norm": 23.522381250220548,
+      "learning_rate": 5e-06,
+      "loss": 0.165,
+      "num_input_tokens_seen": 455471224,
+      "step": 2635
+    },
+    {
+      "epoch": 1.0091918805055535,
+      "loss": 0.20535528659820557,
+      "loss_ce": 0.02023564837872982,
+      "loss_iou": 1.0670104026794434,
+      "loss_num": 0.185546875,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 455471224,
+      "step": 2635
+    },
+    {
+      "epoch": 1.009574875526618,
+      "grad_norm": 30.0462320621947,
+      "learning_rate": 5e-06,
+      "loss": 0.1998,
+      "num_input_tokens_seen": 455643912,
+      "step": 2636
+    },
+    {
+      "epoch": 1.009574875526618,
+      "loss": 0.17439775168895721,
+      "loss_ce": 0.019734669476747513,
+      "loss_iou": 1.0009647607803345,
+      "loss_num": 0.154296875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 455643912,
+      "step": 2636
+    },
+    {
+      "epoch": 1.0099578705476828,
+      "grad_norm": 39.616333037041514,
+      "learning_rate": 5e-06,
+      "loss": 0.2247,
+      "num_input_tokens_seen": 455817056,
+      "step": 2637
+    },
+    {
+      "epoch": 1.0099578705476828,
+      "loss": 0.21939142048358917,
+      "loss_ce": 0.021210266277194023,
+      "loss_iou": 1.058443307876587,
+      "loss_num": 0.1982421875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 455817056,
+      "step": 2637
+    },
+    {
+      "epoch": 1.0103408655687476,
+      "grad_norm": 28.92030922603685,
+      "learning_rate": 5e-06,
+      "loss": 0.2297,
+      "num_input_tokens_seen": 455990240,
+      "step": 2638
+    },
+    {
+      "epoch": 1.0103408655687476,
+      "loss": 0.2293231040239334,
+      "loss_ce": 0.022902213037014008,
+      "loss_iou": 1.330636739730835,
+      "loss_num": 0.2060546875,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 455990240,
+      "step": 2638
+    },
+    {
+      "epoch": 1.0107238605898123,
+      "grad_norm": 21.164605163067154,
+      "learning_rate": 5e-06,
+      "loss": 0.1839,
+      "num_input_tokens_seen": 456162992,
+      "step": 2639
+    },
+    {
+      "epoch": 1.0107238605898123,
+      "loss": 0.18436919152736664,
+      "loss_ce": 0.018292535096406937,
+      "loss_iou": 1.0153162479400635,
+      "loss_num": 0.166015625,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 456162992,
+      "step": 2639
+    },
+    {
+      "epoch": 1.011106855610877,
+      "grad_norm": 30.72034630005925,
+      "learning_rate": 5e-06,
+      "loss": 0.2067,
+      "num_input_tokens_seen": 456336144,
+      "step": 2640
+    },
+    {
+      "epoch": 1.011106855610877,
+      "loss": 0.20957262814044952,
+      "loss_ce": 0.021706413477659225,
+      "loss_iou": 1.0251965522766113,
+      "loss_num": 0.1875,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 456336144,
+      "step": 2640
+    },
+    {
+      "epoch": 1.0114898506319419,
+      "grad_norm": 24.02837295298551,
+      "learning_rate": 5e-06,
+      "loss": 0.1673,
+      "num_input_tokens_seen": 456508832,
+      "step": 2641
+    },
+    {
+      "epoch": 1.0114898506319419,
+      "loss": 0.15824568271636963,
+      "loss_ce": 0.019970526918768883,
+      "loss_iou": 1.0397586822509766,
+      "loss_num": 0.138671875,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 456508832,
+      "step": 2641
+    },
+    {
+      "epoch": 1.0118728456530066,
+      "grad_norm": 46.078856687371086,
+      "learning_rate": 5e-06,
+      "loss": 0.2773,
+      "num_input_tokens_seen": 456681824,
+      "step": 2642
+    },
+    {
+      "epoch": 1.0118728456530066,
+      "loss": 0.2912445068359375,
+      "loss_ce": 0.020553607493638992,
+      "loss_iou": 1.063234806060791,
+      "loss_num": 0.271484375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 456681824,
+      "step": 2642
+    },
+    {
+      "epoch": 1.0122558406740712,
+      "grad_norm": 28.80071795580853,
+      "learning_rate": 5e-06,
+      "loss": 0.2123,
+      "num_input_tokens_seen": 456854728,
+      "step": 2643
+    },
+    {
+      "epoch": 1.0122558406740712,
+      "loss": 0.2342531681060791,
+      "loss_ce": 0.01867700181901455,
+      "loss_iou": 1.106345772743225,
+      "loss_num": 0.2158203125,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 456854728,
+      "step": 2643
+    },
+    {
+      "epoch": 1.012638835695136,
+      "grad_norm": 27.452590449993657,
+      "learning_rate": 5e-06,
+      "loss": 0.2459,
+      "num_input_tokens_seen": 457027496,
+      "step": 2644
+    },
+    {
+      "epoch": 1.012638835695136,
+      "loss": 0.2144833505153656,
+      "loss_ce": 0.021246036514639854,
+      "loss_iou": 1.0015583038330078,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 457027496,
+      "step": 2644
+    },
+    {
+      "epoch": 1.0130218307162007,
+      "grad_norm": 27.130762085346163,
+      "learning_rate": 5e-06,
+      "loss": 0.2251,
+      "num_input_tokens_seen": 457200544,
+      "step": 2645
+    },
+    {
+      "epoch": 1.0130218307162007,
+      "loss": 0.20875662565231323,
+      "loss_ce": 0.020829375833272934,
+      "loss_iou": 1.0030250549316406,
+      "loss_num": 0.1875,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 457200544,
+      "step": 2645
+    },
+    {
+      "epoch": 1.0134048257372654,
+      "grad_norm": 47.90456318918338,
+      "learning_rate": 5e-06,
+      "loss": 0.2206,
+      "num_input_tokens_seen": 457373624,
+      "step": 2646
+    },
+    {
+      "epoch": 1.0134048257372654,
+      "loss": 0.26886725425720215,
+      "loss_ce": 0.022163134068250656,
+      "loss_iou": 1.0137746334075928,
+      "loss_num": 0.2470703125,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 457373624,
+      "step": 2646
+    },
+    {
+      "epoch": 1.0137878207583302,
+      "grad_norm": 45.40839199207175,
+      "learning_rate": 5e-06,
+      "loss": 0.2435,
+      "num_input_tokens_seen": 457546344,
+      "step": 2647
+    },
+    {
+      "epoch": 1.0137878207583302,
+      "loss": 0.2117205560207367,
+      "loss_ce": 0.020436381921172142,
+      "loss_iou": 1.1137763261795044,
+      "loss_num": 0.19140625,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 457546344,
+      "step": 2647
+    },
+    {
+      "epoch": 1.014170815779395,
+      "grad_norm": 31.369161286808033,
+      "learning_rate": 5e-06,
+      "loss": 0.1555,
+      "num_input_tokens_seen": 457719240,
+      "step": 2648
+    },
+    {
+      "epoch": 1.014170815779395,
+      "loss": 0.13760226964950562,
+      "loss_ce": 0.02322237938642502,
+      "loss_iou": 1.0234071016311646,
+      "loss_num": 0.1142578125,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 457719240,
+      "step": 2648
+    },
+    {
+      "epoch": 1.0145538108004597,
+      "grad_norm": 19.891610461402454,
+      "learning_rate": 5e-06,
+      "loss": 0.1579,
+      "num_input_tokens_seen": 457892416,
+      "step": 2649
+    },
+    {
+      "epoch": 1.0145538108004597,
+      "loss": 0.18401141464710236,
+      "loss_ce": 0.022756528109312057,
+      "loss_iou": 1.0163376331329346,
+      "loss_num": 0.1611328125,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 457892416,
+      "step": 2649
+    },
+    {
+      "epoch": 1.0149368058215242,
+      "grad_norm": 23.259849293220743,
+      "learning_rate": 5e-06,
+      "loss": 0.1973,
+      "num_input_tokens_seen": 458065152,
+      "step": 2650
+    },
+    {
+      "epoch": 1.0149368058215242,
+      "loss": 0.1957165151834488,
+      "loss_ce": 0.021766329184174538,
+      "loss_iou": 1.0006461143493652,
+      "loss_num": 0.173828125,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 458065152,
+      "step": 2650
+    },
+    {
+      "epoch": 1.015319800842589,
+      "grad_norm": 43.91331783389807,
+      "learning_rate": 5e-06,
+      "loss": 0.1989,
+      "num_input_tokens_seen": 458237888,
+      "step": 2651
+    },
+    {
+      "epoch": 1.015319800842589,
+      "loss": 0.2078561782836914,
+      "loss_ce": 0.021271690726280212,
+      "loss_iou": 1.003035068511963,
+      "loss_num": 0.1865234375,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 458237888,
+      "step": 2651
+    },
+    {
+      "epoch": 1.0157027958636538,
+      "grad_norm": 38.57047268554671,
+      "learning_rate": 5e-06,
+      "loss": 0.1556,
+      "num_input_tokens_seen": 458411048,
+      "step": 2652
+    },
+    {
+      "epoch": 1.0157027958636538,
+      "loss": 0.1338464617729187,
+      "loss_ce": 0.020778842270374298,
+      "loss_iou": 1.0011705160140991,
+      "loss_num": 0.11328125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 458411048,
+      "step": 2652
+    },
+    {
+      "epoch": 1.0160857908847185,
+      "grad_norm": 21.82677951625571,
+      "learning_rate": 5e-06,
+      "loss": 0.2378,
+      "num_input_tokens_seen": 458584168,
+      "step": 2653
+    },
+    {
+      "epoch": 1.0160857908847185,
+      "loss": 0.2706211805343628,
+      "loss_ce": 0.020132914185523987,
+      "loss_iou": 1.1052721738815308,
+      "loss_num": 0.25,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 458584168,
+      "step": 2653
+    },
+    {
+      "epoch": 1.0164687859057833,
+      "grad_norm": 20.04536041659893,
+      "learning_rate": 5e-06,
+      "loss": 0.2075,
+      "num_input_tokens_seen": 458757376,
+      "step": 2654
+    },
+    {
+      "epoch": 1.0164687859057833,
+      "loss": 0.20592357218265533,
+      "loss_ce": 0.020742908120155334,
+      "loss_iou": 1.0075843334197998,
+      "loss_num": 0.185546875,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 458757376,
+      "step": 2654
+    },
+    {
+      "epoch": 1.016851780926848,
+      "grad_norm": 44.789371331460984,
+      "learning_rate": 5e-06,
+      "loss": 0.1781,
+      "num_input_tokens_seen": 458930216,
+      "step": 2655
+    },
+    {
+      "epoch": 1.016851780926848,
+      "loss": 0.16285867989063263,
+      "loss_ce": 0.02147073857486248,
+      "loss_iou": 1.0407545566558838,
+      "loss_num": 0.1416015625,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 458930216,
+      "step": 2655
+    },
+    {
+      "epoch": 1.0172347759479128,
+      "grad_norm": 33.14025348410339,
+      "learning_rate": 5e-06,
+      "loss": 0.1942,
+      "num_input_tokens_seen": 459103216,
+      "step": 2656
+    },
+    {
+      "epoch": 1.0172347759479128,
+      "loss": 0.19790545105934143,
+      "loss_ce": 0.02188006415963173,
+      "loss_iou": 1.0012357234954834,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 459103216,
+      "step": 2656
+    },
+    {
+      "epoch": 1.0176177709689773,
+      "grad_norm": 29.809596683884156,
+      "learning_rate": 5e-06,
+      "loss": 0.2239,
+      "num_input_tokens_seen": 459276448,
+      "step": 2657
+    },
+    {
+      "epoch": 1.0176177709689773,
+      "loss": 0.23329783976078033,
+      "loss_ce": 0.01979687437415123,
+      "loss_iou": 1.0008468627929688,
+      "loss_num": 0.2138671875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 459276448,
+      "step": 2657
+    },
+    {
+      "epoch": 1.018000765990042,
+      "grad_norm": 30.763285231618685,
+      "learning_rate": 5e-06,
+      "loss": 0.156,
+      "num_input_tokens_seen": 459449216,
+      "step": 2658
+    },
+    {
+      "epoch": 1.018000765990042,
+      "loss": 0.1709698736667633,
+      "loss_ce": 0.019419565796852112,
+      "loss_iou": 1.0088496208190918,
+      "loss_num": 0.1513671875,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 459449216,
+      "step": 2658
+    },
+    {
+      "epoch": 1.0183837610111068,
+      "grad_norm": 41.45551240380358,
+      "learning_rate": 5e-06,
+      "loss": 0.1866,
+      "num_input_tokens_seen": 459622264,
+      "step": 2659
+    },
+    {
+      "epoch": 1.0183837610111068,
+      "loss": 0.16339613497257233,
+      "loss_ce": 0.019719377160072327,
+      "loss_iou": 1.0058441162109375,
+      "loss_num": 0.1435546875,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 459622264,
+      "step": 2659
+    },
+    {
+      "epoch": 1.0187667560321716,
+      "grad_norm": 24.041455167168042,
+      "learning_rate": 5e-06,
+      "loss": 0.2229,
+      "num_input_tokens_seen": 459794920,
+      "step": 2660
+    },
+    {
+      "epoch": 1.0187667560321716,
+      "loss": 0.22818854451179504,
+      "loss_ce": 0.019875554367899895,
+      "loss_iou": 1.0752671957015991,
+      "loss_num": 0.2080078125,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 459794920,
+      "step": 2660
+    },
+    {
+      "epoch": 1.0191497510532364,
+      "grad_norm": 40.4623954131525,
+      "learning_rate": 5e-06,
+      "loss": 0.1492,
+      "num_input_tokens_seen": 459968016,
+      "step": 2661
+    },
+    {
+      "epoch": 1.0191497510532364,
+      "loss": 0.12593373656272888,
+      "loss_ce": 0.02003774419426918,
+      "loss_iou": 1.0006287097930908,
+      "loss_num": 0.10595703125,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 459968016,
+      "step": 2661
+    },
+    {
+      "epoch": 1.019532746074301,
+      "grad_norm": 23.331262836860812,
+      "learning_rate": 5e-06,
+      "loss": 0.2209,
+      "num_input_tokens_seen": 460140872,
+      "step": 2662
+    },
+    {
+      "epoch": 1.019532746074301,
+      "loss": 0.22041067481040955,
+      "loss_ce": 0.02052055299282074,
+      "loss_iou": 1.0110864639282227,
+      "loss_num": 0.2001953125,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 460140872,
+      "step": 2662
+    },
+    {
+      "epoch": 1.0199157410953659,
+      "grad_norm": 53.680906002538734,
+      "learning_rate": 5e-06,
+      "loss": 0.227,
+      "num_input_tokens_seen": 460314040,
+      "step": 2663
+    },
+    {
+      "epoch": 1.0199157410953659,
+      "loss": 0.245751291513443,
+      "loss_ce": 0.019982237368822098,
+      "loss_iou": 1.0646202564239502,
+      "loss_num": 0.2255859375,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 460314040,
+      "step": 2663
+    },
+    {
+      "epoch": 1.0202987361164304,
+      "grad_norm": 31.59724384859943,
+      "learning_rate": 5e-06,
+      "loss": 0.2249,
+      "num_input_tokens_seen": 460486744,
+      "step": 2664
+    },
+    {
+      "epoch": 1.0202987361164304,
+      "loss": 0.22602880001068115,
+      "loss_ce": 0.02326999604701996,
+      "loss_iou": 1.0191833972930908,
+      "loss_num": 0.203125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 460486744,
+      "step": 2664
+    },
+    {
+      "epoch": 1.0206817311374952,
+      "grad_norm": 55.58539040679261,
+      "learning_rate": 5e-06,
+      "loss": 0.2816,
+      "num_input_tokens_seen": 460659728,
+      "step": 2665
+    },
+    {
+      "epoch": 1.0206817311374952,
+      "loss": 0.3014325797557831,
+      "loss_ce": 0.020182587206363678,
+      "loss_iou": 1.0474450588226318,
+      "loss_num": 0.28125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 460659728,
+      "step": 2665
+    },
+    {
+      "epoch": 1.02106472615856,
+      "grad_norm": 23.924529770016672,
+      "learning_rate": 5e-06,
+      "loss": 0.2308,
+      "num_input_tokens_seen": 460832664,
+      "step": 2666
+    },
+    {
+      "epoch": 1.02106472615856,
+      "loss": 0.23173928260803223,
+      "loss_ce": 0.019703146070241928,
+      "loss_iou": 1.03975510597229,
+      "loss_num": 0.2119140625,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 460832664,
+      "step": 2666
+    },
+    {
+      "epoch": 1.0214477211796247,
+      "grad_norm": 61.7786700457965,
+      "learning_rate": 5e-06,
+      "loss": 0.1617,
+      "num_input_tokens_seen": 461005496,
+      "step": 2667
+    },
+    {
+      "epoch": 1.0214477211796247,
+      "loss": 0.1547263115644455,
+      "loss_ce": 0.019594479352235794,
+      "loss_iou": 1.0006184577941895,
+      "loss_num": 0.134765625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 461005496,
+      "step": 2667
+    },
+    {
+      "epoch": 1.0218307162006894,
+      "grad_norm": 23.446864568879292,
+      "learning_rate": 5e-06,
+      "loss": 0.2579,
+      "num_input_tokens_seen": 461178544,
+      "step": 2668
+    },
+    {
+      "epoch": 1.0218307162006894,
+      "loss": 0.27943965792655945,
+      "loss_ce": 0.020040256902575493,
+      "loss_iou": 1.0679092407226562,
+      "loss_num": 0.259765625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 461178544,
+      "step": 2668
+    },
+    {
+      "epoch": 1.0222137112217542,
+      "grad_norm": 37.334407917544844,
+      "learning_rate": 5e-06,
+      "loss": 0.2227,
+      "num_input_tokens_seen": 461351560,
+      "step": 2669
+    },
+    {
+      "epoch": 1.0222137112217542,
+      "loss": 0.23498889803886414,
+      "loss_ce": 0.020877568051218987,
+      "loss_iou": 1.036283016204834,
+      "loss_num": 0.2138671875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 461351560,
+      "step": 2669
+    },
+    {
+      "epoch": 1.022596706242819,
+      "grad_norm": 24.23724171944618,
+      "learning_rate": 5e-06,
+      "loss": 0.2083,
+      "num_input_tokens_seen": 461524936,
+      "step": 2670
+    },
+    {
+      "epoch": 1.022596706242819,
+      "loss": 0.19315654039382935,
+      "loss_ce": 0.02280740812420845,
+      "loss_iou": 1.0224149227142334,
+      "loss_num": 0.169921875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 461524936,
+      "step": 2670
+    },
+    {
+      "epoch": 1.0229797012638835,
+      "grad_norm": 43.99273166540144,
+      "learning_rate": 5e-06,
+      "loss": 0.1895,
+      "num_input_tokens_seen": 461697568,
+      "step": 2671
+    },
+    {
+      "epoch": 1.0229797012638835,
+      "loss": 0.18223366141319275,
+      "loss_ce": 0.021985851228237152,
+      "loss_iou": 1.0184319019317627,
+      "loss_num": 0.16015625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 461697568,
+      "step": 2671
+    },
+    {
+      "epoch": 1.0233626962849482,
+      "grad_norm": 27.24794231197852,
+      "learning_rate": 5e-06,
+      "loss": 0.2238,
+      "num_input_tokens_seen": 461870352,
+      "step": 2672
+    },
+    {
+      "epoch": 1.0233626962849482,
+      "loss": 0.24211221933364868,
+      "loss_ce": 0.02305705100297928,
+      "loss_iou": 1.0264095067977905,
+      "loss_num": 0.21875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 461870352,
+      "step": 2672
+    },
+    {
+      "epoch": 1.023745691306013,
+      "grad_norm": 33.861199537220216,
+      "learning_rate": 5e-06,
+      "loss": 0.196,
+      "num_input_tokens_seen": 462043528,
+      "step": 2673
+    },
+    {
+      "epoch": 1.023745691306013,
+      "loss": 0.17982439696788788,
+      "loss_ce": 0.020705748349428177,
+      "loss_iou": 1.0384835004806519,
+      "loss_num": 0.1591796875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 462043528,
+      "step": 2673
+    },
+    {
+      "epoch": 1.0241286863270778,
+      "grad_norm": 22.053638294927843,
+      "learning_rate": 5e-06,
+      "loss": 0.2145,
+      "num_input_tokens_seen": 462216592,
+      "step": 2674
+    },
+    {
+      "epoch": 1.0241286863270778,
+      "loss": 0.23037189245224,
+      "loss_ce": 0.02327960543334484,
+      "loss_iou": 1.0616304874420166,
+      "loss_num": 0.20703125,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 462216592,
+      "step": 2674
+    },
+    {
+      "epoch": 1.0245116813481425,
+      "grad_norm": 34.33366681501337,
+      "learning_rate": 5e-06,
+      "loss": 0.2301,
+      "num_input_tokens_seen": 462389864,
+      "step": 2675
+    },
+    {
+      "epoch": 1.0245116813481425,
+      "loss": 0.2765474319458008,
+      "loss_ce": 0.021786676719784737,
+      "loss_iou": 1.0377389192581177,
+      "loss_num": 0.25390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 462389864,
+      "step": 2675
+    },
+    {
+      "epoch": 1.0248946763692073,
+      "grad_norm": 27.150370061380826,
+      "learning_rate": 5e-06,
+      "loss": 0.1892,
+      "num_input_tokens_seen": 462562808,
+      "step": 2676
+    },
+    {
+      "epoch": 1.0248946763692073,
+      "loss": 0.18681272864341736,
+      "loss_ce": 0.020858149975538254,
+      "loss_iou": 1.019316554069519,
+      "loss_num": 0.166015625,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 462562808,
+      "step": 2676
+    },
+    {
+      "epoch": 1.0252776713902718,
+      "grad_norm": 23.518013552057838,
+      "learning_rate": 5e-06,
+      "loss": 0.1951,
+      "num_input_tokens_seen": 462735944,
+      "step": 2677
+    },
+    {
+      "epoch": 1.0252776713902718,
+      "loss": 0.15821880102157593,
+      "loss_ce": 0.01973002590239048,
+      "loss_iou": 1.0040994882583618,
+      "loss_num": 0.138671875,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 462735944,
+      "step": 2677
+    },
+    {
+      "epoch": 1.0256606664113366,
+      "grad_norm": 14.889740274456658,
+      "learning_rate": 5e-06,
+      "loss": 0.1483,
+      "num_input_tokens_seen": 462908728,
+      "step": 2678
+    },
+    {
+      "epoch": 1.0256606664113366,
+      "loss": 0.1388009786605835,
+      "loss_ce": 0.0204537995159626,
+      "loss_iou": 1.004044771194458,
+      "loss_num": 0.1181640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 462908728,
+      "step": 2678
+    },
+    {
+      "epoch": 1.0260436614324013,
+      "grad_norm": 23.297625717045992,
+      "learning_rate": 5e-06,
+      "loss": 0.1822,
+      "num_input_tokens_seen": 463081704,
+      "step": 2679
+    },
+    {
+      "epoch": 1.0260436614324013,
+      "loss": 0.19627460837364197,
+      "loss_ce": 0.020249227061867714,
+      "loss_iou": 1.0534923076629639,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 463081704,
+      "step": 2679
+    },
+    {
+      "epoch": 1.026426656453466,
+      "grad_norm": 19.64664560796669,
+      "learning_rate": 5e-06,
+      "loss": 0.155,
+      "num_input_tokens_seen": 463254840,
+      "step": 2680
+    },
+    {
+      "epoch": 1.026426656453466,
+      "loss": 0.143958181142807,
+      "loss_ce": 0.021491151303052902,
+      "loss_iou": 1.0011191368103027,
+      "loss_num": 0.12255859375,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 463254840,
+      "step": 2680
+    },
+    {
+      "epoch": 1.0268096514745308,
+      "grad_norm": 20.310161499712436,
+      "learning_rate": 5e-06,
+      "loss": 0.2153,
+      "num_input_tokens_seen": 463427896,
+      "step": 2681
+    },
+    {
+      "epoch": 1.0268096514745308,
+      "loss": 0.2070397436618805,
+      "loss_ce": 0.020577343180775642,
+      "loss_iou": 1.0429177284240723,
+      "loss_num": 0.1865234375,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 463427896,
+      "step": 2681
+    },
+    {
+      "epoch": 1.0271926464955956,
+      "grad_norm": 17.325785775797268,
+      "learning_rate": 5e-06,
+      "loss": 0.2248,
+      "num_input_tokens_seen": 463600600,
+      "step": 2682
+    },
+    {
+      "epoch": 1.0271926464955956,
+      "loss": 0.205996572971344,
+      "loss_ce": 0.022708002477884293,
+      "loss_iou": 1.047377586364746,
+      "loss_num": 0.18359375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 463600600,
+      "step": 2682
+    },
+    {
+      "epoch": 1.0275756415166604,
+      "grad_norm": 18.49515030440836,
+      "learning_rate": 5e-06,
+      "loss": 0.1786,
+      "num_input_tokens_seen": 463773552,
+      "step": 2683
+    },
+    {
+      "epoch": 1.0275756415166604,
+      "loss": 0.23582297563552856,
+      "loss_ce": 0.019453343003988266,
+      "loss_iou": 1.2463963031768799,
+      "loss_num": 0.216796875,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 463773552,
+      "step": 2683
+    },
+    {
+      "epoch": 1.027958636537725,
+      "grad_norm": 21.69668716747673,
+      "learning_rate": 5e-06,
+      "loss": 0.1462,
+      "num_input_tokens_seen": 463946792,
+      "step": 2684
+    },
+    {
+      "epoch": 1.027958636537725,
+      "loss": 0.12513580918312073,
+      "loss_ce": 0.022016923874616623,
+      "loss_iou": 1.0008490085601807,
+      "loss_num": 0.10302734375,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 463946792,
+      "step": 2684
+    },
+    {
+      "epoch": 1.0283416315587897,
+      "grad_norm": 27.265711410161398,
+      "learning_rate": 5e-06,
+      "loss": 0.1631,
+      "num_input_tokens_seen": 464119312,
+      "step": 2685
+    },
+    {
+      "epoch": 1.0283416315587897,
+      "loss": 0.19970130920410156,
+      "loss_ce": 0.01970863714814186,
+      "loss_iou": 1.0485548973083496,
+      "loss_num": 0.1796875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 464119312,
+      "step": 2685
+    },
+    {
+      "epoch": 1.0287246265798544,
+      "grad_norm": 24.029693810443284,
+      "learning_rate": 5e-06,
+      "loss": 0.1921,
+      "num_input_tokens_seen": 464292416,
+      "step": 2686
+    },
+    {
+      "epoch": 1.0287246265798544,
+      "loss": 0.22213320434093475,
+      "loss_ce": 0.020717188715934753,
+      "loss_iou": 1.0227363109588623,
+      "loss_num": 0.201171875,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 464292416,
+      "step": 2686
+    },
+    {
+      "epoch": 1.0291076216009192,
+      "grad_norm": 21.937065930896914,
+      "learning_rate": 5e-06,
+      "loss": 0.1715,
+      "num_input_tokens_seen": 464465192,
+      "step": 2687
+    },
+    {
+      "epoch": 1.0291076216009192,
+      "loss": 0.1793980896472931,
+      "loss_ce": 0.020157355815172195,
+      "loss_iou": 1.0054259300231934,
+      "loss_num": 0.1591796875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 464465192,
+      "step": 2687
+    },
+    {
+      "epoch": 1.029490616621984,
+      "grad_norm": 23.353506224694193,
+      "learning_rate": 5e-06,
+      "loss": 0.143,
+      "num_input_tokens_seen": 464638496,
+      "step": 2688
+    },
+    {
+      "epoch": 1.029490616621984,
+      "loss": 0.13977614045143127,
+      "loss_ce": 0.020849131047725677,
+      "loss_iou": 1.0167276859283447,
+      "loss_num": 0.119140625,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 464638496,
+      "step": 2688
+    },
+    {
+      "epoch": 1.0298736116430487,
+      "grad_norm": 23.77239661642061,
+      "learning_rate": 5e-06,
+      "loss": 0.1798,
+      "num_input_tokens_seen": 464810936,
+      "step": 2689
+    },
+    {
+      "epoch": 1.0298736116430487,
+      "loss": 0.21237342059612274,
+      "loss_ce": 0.02011268399655819,
+      "loss_iou": 1.2245774269104004,
+      "loss_num": 0.1923828125,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 464810936,
+      "step": 2689
+    },
+    {
+      "epoch": 1.0302566066641135,
+      "grad_norm": 17.674660483168775,
+      "learning_rate": 5e-06,
+      "loss": 0.1708,
+      "num_input_tokens_seen": 464983800,
+      "step": 2690
+    },
+    {
+      "epoch": 1.0302566066641135,
+      "loss": 0.14379066228866577,
+      "loss_ce": 0.020926905795931816,
+      "loss_iou": 1.0017061233520508,
+      "loss_num": 0.123046875,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 464983800,
+      "step": 2690
+    },
+    {
+      "epoch": 1.030639601685178,
+      "grad_norm": 19.617520136686487,
+      "learning_rate": 5e-06,
+      "loss": 0.1952,
+      "num_input_tokens_seen": 465157104,
+      "step": 2691
+    },
+    {
+      "epoch": 1.030639601685178,
+      "loss": 0.16387206315994263,
+      "loss_ce": 0.01964598335325718,
+      "loss_iou": 1.0038049221038818,
+      "loss_num": 0.14453125,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 465157104,
+      "step": 2691
+    },
+    {
+      "epoch": 1.0310225967062427,
+      "grad_norm": 17.7155011736127,
+      "learning_rate": 5e-06,
+      "loss": 0.1993,
+      "num_input_tokens_seen": 465329920,
+      "step": 2692
+    },
+    {
+      "epoch": 1.0310225967062427,
+      "loss": 0.16195493936538696,
+      "loss_ce": 0.019468368962407112,
+      "loss_iou": 1.0223468542099,
+      "loss_num": 0.142578125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 465329920,
+      "step": 2692
+    },
+    {
+      "epoch": 1.0314055917273075,
+      "grad_norm": 17.70671620632501,
+      "learning_rate": 5e-06,
+      "loss": 0.1711,
+      "num_input_tokens_seen": 465502472,
+      "step": 2693
+    },
+    {
+      "epoch": 1.0314055917273075,
+      "loss": 0.12763410806655884,
+      "loss_ce": 0.01972394622862339,
+      "loss_iou": 1.0092477798461914,
+      "loss_num": 0.10791015625,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 465502472,
+      "step": 2693
+    },
+    {
+      "epoch": 1.0317885867483723,
+      "grad_norm": 22.52502334501282,
+      "learning_rate": 5e-06,
+      "loss": 0.1693,
+      "num_input_tokens_seen": 465675480,
+      "step": 2694
+    },
+    {
+      "epoch": 1.0317885867483723,
+      "loss": 0.19143357872962952,
+      "loss_ce": 0.020901355892419815,
+      "loss_iou": 0.988635778427124,
+      "loss_num": 0.1708984375,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 465675480,
+      "step": 2694
+    },
+    {
+      "epoch": 1.032171581769437,
+      "grad_norm": 23.850111143473104,
+      "learning_rate": 5e-06,
+      "loss": 0.2177,
+      "num_input_tokens_seen": 465848320,
+      "step": 2695
+    },
+    {
+      "epoch": 1.032171581769437,
+      "loss": 0.27090975642204285,
+      "loss_ce": 0.018712498247623444,
+      "loss_iou": 1.1247084140777588,
+      "loss_num": 0.251953125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 465848320,
+      "step": 2695
+    },
+    {
+      "epoch": 1.0325545767905018,
+      "grad_norm": 24.100004859986235,
+      "learning_rate": 5e-06,
+      "loss": 0.1467,
+      "num_input_tokens_seen": 466021600,
+      "step": 2696
+    },
+    {
+      "epoch": 1.0325545767905018,
+      "loss": 0.1535719335079193,
+      "loss_ce": 0.02072891592979431,
+      "loss_iou": 1.008996605873108,
+      "loss_num": 0.1328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 466021600,
+      "step": 2696
+    },
+    {
+      "epoch": 1.0329375718115665,
+      "grad_norm": 39.686082647434645,
+      "learning_rate": 5e-06,
+      "loss": 0.1759,
+      "num_input_tokens_seen": 466194168,
+      "step": 2697
+    },
+    {
+      "epoch": 1.0329375718115665,
+      "loss": 0.18978765606880188,
+      "loss_ce": 0.02105596661567688,
+      "loss_iou": 1.017608642578125,
+      "loss_num": 0.1689453125,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 466194168,
+      "step": 2697
+    },
+    {
+      "epoch": 1.033320566832631,
+      "grad_norm": 32.17313421412968,
+      "learning_rate": 5e-06,
+      "loss": 0.139,
+      "num_input_tokens_seen": 466366984,
+      "step": 2698
+    },
+    {
+      "epoch": 1.033320566832631,
+      "loss": 0.13241815567016602,
+      "loss_ce": 0.020967965945601463,
+      "loss_iou": 1.0103569030761719,
+      "loss_num": 0.111328125,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 466366984,
+      "step": 2698
+    },
+    {
+      "epoch": 1.0337035618536958,
+      "grad_norm": 28.81487627958747,
+      "learning_rate": 5e-06,
+      "loss": 0.2004,
+      "num_input_tokens_seen": 466540200,
+      "step": 2699
+    },
+    {
+      "epoch": 1.0337035618536958,
+      "loss": 0.2008867859840393,
+      "loss_ce": 0.01924615353345871,
+      "loss_iou": 0.9303417205810547,
+      "loss_num": 0.181640625,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 466540200,
+      "step": 2699
+    },
+    {
+      "epoch": 1.0340865568747606,
+      "grad_norm": 21.392598667153916,
+      "learning_rate": 5e-06,
+      "loss": 0.1486,
+      "num_input_tokens_seen": 466713632,
+      "step": 2700
+    },
+    {
+      "epoch": 1.0340865568747606,
+      "loss": 0.14038464426994324,
+      "loss_ce": 0.01922985538840294,
+      "loss_iou": 1.0356340408325195,
+      "loss_num": 0.12109375,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 466713632,
+      "step": 2700
+    },
+    {
+      "epoch": 1.0344695518958253,
+      "grad_norm": 29.166508226110697,
+      "learning_rate": 5e-06,
+      "loss": 0.2176,
+      "num_input_tokens_seen": 466885960,
+      "step": 2701
+    },
+    {
+      "epoch": 1.0344695518958253,
+      "loss": 0.27286502718925476,
+      "loss_ce": 0.02011844702064991,
+      "loss_iou": 1.4925570487976074,
+      "loss_num": 0.251953125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 466885960,
+      "step": 2701
+    },
+    {
+      "epoch": 1.03485254691689,
+      "grad_norm": 31.334617619095408,
+      "learning_rate": 5e-06,
+      "loss": 0.1873,
+      "num_input_tokens_seen": 467058840,
+      "step": 2702
+    },
+    {
+      "epoch": 1.03485254691689,
+      "loss": 0.217850923538208,
+      "loss_ce": 0.02040218934416771,
+      "loss_iou": 1.0190616846084595,
+      "loss_num": 0.197265625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 467058840,
+      "step": 2702
+    },
+    {
+      "epoch": 1.0352355419379549,
+      "grad_norm": 27.30942700897604,
+      "learning_rate": 5e-06,
+      "loss": 0.1852,
+      "num_input_tokens_seen": 467232000,
+      "step": 2703
+    },
+    {
+      "epoch": 1.0352355419379549,
+      "loss": 0.18020913004875183,
+      "loss_ce": 0.022494282573461533,
+      "loss_iou": 1.0234897136688232,
+      "loss_num": 0.158203125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 467232000,
+      "step": 2703
+    },
+    {
+      "epoch": 1.0356185369590196,
+      "grad_norm": 23.122492456689073,
+      "learning_rate": 5e-06,
+      "loss": 0.1686,
+      "num_input_tokens_seen": 467404456,
+      "step": 2704
+    },
+    {
+      "epoch": 1.0356185369590196,
+      "loss": 0.13935062289237976,
+      "loss_ce": 0.01920292340219021,
+      "loss_iou": 0.9711761474609375,
+      "loss_num": 0.1201171875,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 467404456,
+      "step": 2704
+    },
+    {
+      "epoch": 1.0360015319800842,
+      "grad_norm": 32.89013374464874,
+      "learning_rate": 5e-06,
+      "loss": 0.2026,
+      "num_input_tokens_seen": 467577312,
+      "step": 2705
+    },
+    {
+      "epoch": 1.0360015319800842,
+      "loss": 0.24207282066345215,
+      "loss_ce": 0.019843831658363342,
+      "loss_iou": 1.0349886417388916,
+      "loss_num": 0.22265625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 467577312,
+      "step": 2705
+    },
+    {
+      "epoch": 1.036384527001149,
+      "grad_norm": 21.513507526762115,
+      "learning_rate": 5e-06,
+      "loss": 0.2125,
+      "num_input_tokens_seen": 467749976,
+      "step": 2706
+    },
+    {
+      "epoch": 1.036384527001149,
+      "loss": 0.18888524174690247,
+      "loss_ce": 0.019329577684402466,
+      "loss_iou": 1.011885643005371,
+      "loss_num": 0.169921875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 467749976,
+      "step": 2706
+    },
+    {
+      "epoch": 1.0367675220222137,
+      "grad_norm": 21.080481269575095,
+      "learning_rate": 5e-06,
+      "loss": 0.1922,
+      "num_input_tokens_seen": 467922808,
+      "step": 2707
+    },
+    {
+      "epoch": 1.0367675220222137,
+      "loss": 0.17099253833293915,
+      "loss_ce": 0.020846057683229446,
+      "loss_iou": 1.0104583501815796,
+      "loss_num": 0.150390625,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 467922808,
+      "step": 2707
+    },
+    {
+      "epoch": 1.0371505170432784,
+      "grad_norm": 28.250440462278117,
+      "learning_rate": 5e-06,
+      "loss": 0.217,
+      "num_input_tokens_seen": 468095312,
+      "step": 2708
+    },
+    {
+      "epoch": 1.0371505170432784,
+      "loss": 0.2283402681350708,
+      "loss_ce": 0.0191117562353611,
+      "loss_iou": 1.0676853656768799,
+      "loss_num": 0.208984375,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 468095312,
+      "step": 2708
+    },
+    {
+      "epoch": 1.0375335120643432,
+      "grad_norm": 20.106679232759276,
+      "learning_rate": 5e-06,
+      "loss": 0.1596,
+      "num_input_tokens_seen": 468268128,
+      "step": 2709
+    },
+    {
+      "epoch": 1.0375335120643432,
+      "loss": 0.1397133767604828,
+      "loss_ce": 0.018985841423273087,
+      "loss_iou": 1.023972749710083,
+      "loss_num": 0.12060546875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 468268128,
+      "step": 2709
+    },
+    {
+      "epoch": 1.037916507085408,
+      "grad_norm": 18.545317480370425,
+      "learning_rate": 5e-06,
+      "loss": 0.1516,
+      "num_input_tokens_seen": 468441168,
+      "step": 2710
+    },
+    {
+      "epoch": 1.037916507085408,
+      "loss": 0.17443513870239258,
+      "loss_ce": 0.021481044590473175,
+      "loss_iou": 1.008813738822937,
+      "loss_num": 0.1533203125,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 468441168,
+      "step": 2710
+    },
+    {
+      "epoch": 1.0382995021064727,
+      "grad_norm": 31.698141043016687,
+      "learning_rate": 5e-06,
+      "loss": 0.1895,
+      "num_input_tokens_seen": 468614712,
+      "step": 2711
+    },
+    {
+      "epoch": 1.0382995021064727,
+      "loss": 0.15882930159568787,
+      "loss_ce": 0.021683303639292717,
+      "loss_iou": 1.0105422735214233,
+      "loss_num": 0.13671875,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 468614712,
+      "step": 2711
+    },
+    {
+      "epoch": 1.0386824971275372,
+      "grad_norm": 25.409130913828093,
+      "learning_rate": 5e-06,
+      "loss": 0.2247,
+      "num_input_tokens_seen": 468787384,
+      "step": 2712
+    },
+    {
+      "epoch": 1.0386824971275372,
+      "loss": 0.2565978467464447,
+      "loss_ce": 0.01978144608438015,
+      "loss_iou": 1.1115121841430664,
+      "loss_num": 0.236328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 468787384,
+      "step": 2712
+    },
+    {
+      "epoch": 1.039065492148602,
+      "grad_norm": 23.513043474147057,
+      "learning_rate": 5e-06,
+      "loss": 0.1912,
+      "num_input_tokens_seen": 468960824,
+      "step": 2713
+    },
+    {
+      "epoch": 1.039065492148602,
+      "loss": 0.16786177456378937,
+      "loss_ce": 0.020431354641914368,
+      "loss_iou": 1.161240577697754,
+      "loss_num": 0.1474609375,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 468960824,
+      "step": 2713
+    },
+    {
+      "epoch": 1.0394484871696668,
+      "grad_norm": 26.585693680176337,
+      "learning_rate": 5e-06,
+      "loss": 0.1743,
+      "num_input_tokens_seen": 469133664,
+      "step": 2714
+    },
+    {
+      "epoch": 1.0394484871696668,
+      "loss": 0.19016367197036743,
+      "loss_ce": 0.020791107788681984,
+      "loss_iou": 1.0482709407806396,
+      "loss_num": 0.1689453125,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 469133664,
+      "step": 2714
+    },
+    {
+      "epoch": 1.0398314821907315,
+      "grad_norm": 23.614034613887178,
+      "learning_rate": 5e-06,
+      "loss": 0.172,
+      "num_input_tokens_seen": 469306784,
+      "step": 2715
+    },
+    {
+      "epoch": 1.0398314821907315,
+      "loss": 0.20761451125144958,
+      "loss_ce": 0.018893800675868988,
+      "loss_iou": 0.9347192049026489,
+      "loss_num": 0.1884765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 469306784,
+      "step": 2715
+    },
+    {
+      "epoch": 1.0402144772117963,
+      "grad_norm": 16.3842713035619,
+      "learning_rate": 5e-06,
+      "loss": 0.2064,
+      "num_input_tokens_seen": 469479608,
+      "step": 2716
+    },
+    {
+      "epoch": 1.0402144772117963,
+      "loss": 0.20429082214832306,
+      "loss_ce": 0.01956792175769806,
+      "loss_iou": 1.0194833278656006,
+      "loss_num": 0.1845703125,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 469479608,
+      "step": 2716
+    },
+    {
+      "epoch": 1.040597472232861,
+      "grad_norm": 22.725221382510657,
+      "learning_rate": 5e-06,
+      "loss": 0.1585,
+      "num_input_tokens_seen": 469652632,
+      "step": 2717
+    },
+    {
+      "epoch": 1.040597472232861,
+      "loss": 0.14823034405708313,
+      "loss_ce": 0.021460331976413727,
+      "loss_iou": 1.013222336769104,
+      "loss_num": 0.126953125,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 469652632,
+      "step": 2717
+    },
+    {
+      "epoch": 1.0409804672539258,
+      "grad_norm": 59.15850805484919,
+      "learning_rate": 5e-06,
+      "loss": 0.2061,
+      "num_input_tokens_seen": 469825560,
+      "step": 2718
+    },
+    {
+      "epoch": 1.0409804672539258,
+      "loss": 0.15422753989696503,
+      "loss_ce": 0.021292969584465027,
+      "loss_iou": 1.0036109685897827,
+      "loss_num": 0.1328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 469825560,
+      "step": 2718
+    },
+    {
+      "epoch": 1.0413634622749903,
+      "grad_norm": 32.60760835613013,
+      "learning_rate": 5e-06,
+      "loss": 0.1614,
+      "num_input_tokens_seen": 469998592,
+      "step": 2719
+    },
+    {
+      "epoch": 1.0413634622749903,
+      "loss": 0.13662421703338623,
+      "loss_ce": 0.02145087718963623,
+      "loss_iou": 1.0014822483062744,
+      "loss_num": 0.115234375,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 469998592,
+      "step": 2719
+    },
+    {
+      "epoch": 1.041746457296055,
+      "grad_norm": 22.858226704792663,
+      "learning_rate": 5e-06,
+      "loss": 0.155,
+      "num_input_tokens_seen": 470171240,
+      "step": 2720
+    },
+    {
+      "epoch": 1.041746457296055,
+      "loss": 0.13193772733211517,
+      "loss_ce": 0.021708227694034576,
+      "loss_iou": 1.0113179683685303,
+      "loss_num": 0.1103515625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 470171240,
+      "step": 2720
+    },
+    {
+      "epoch": 1.0421294523171198,
+      "grad_norm": 32.50632891454952,
+      "learning_rate": 5e-06,
+      "loss": 0.1947,
+      "num_input_tokens_seen": 470344384,
+      "step": 2721
+    },
+    {
+      "epoch": 1.0421294523171198,
+      "loss": 0.16579297184944153,
+      "loss_ce": 0.01973584294319153,
+      "loss_iou": 0.9779366254806519,
+      "loss_num": 0.146484375,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 470344384,
+      "step": 2721
+    },
+    {
+      "epoch": 1.0425124473381846,
+      "grad_norm": 22.57412908985093,
+      "learning_rate": 5e-06,
+      "loss": 0.1527,
+      "num_input_tokens_seen": 470517656,
+      "step": 2722
+    },
+    {
+      "epoch": 1.0425124473381846,
+      "loss": 0.1500321626663208,
+      "loss_ce": 0.020942814648151398,
+      "loss_iou": 1.0070710182189941,
+      "loss_num": 0.12890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 470517656,
+      "step": 2722
+    },
+    {
+      "epoch": 1.0428954423592494,
+      "grad_norm": 48.54972079558057,
+      "learning_rate": 5e-06,
+      "loss": 0.2444,
+      "num_input_tokens_seen": 470690352,
+      "step": 2723
+    },
+    {
+      "epoch": 1.0428954423592494,
+      "loss": 0.25511783361434937,
+      "loss_ce": 0.02068181149661541,
+      "loss_iou": 1.0030978918075562,
+      "loss_num": 0.234375,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 470690352,
+      "step": 2723
+    },
+    {
+      "epoch": 1.0432784373803141,
+      "grad_norm": 35.89554204645707,
+      "learning_rate": 5e-06,
+      "loss": 0.175,
+      "num_input_tokens_seen": 470862976,
+      "step": 2724
+    },
+    {
+      "epoch": 1.0432784373803141,
+      "loss": 0.16105236113071442,
+      "loss_ce": 0.02122081257402897,
+      "loss_iou": 1.000620722770691,
+      "loss_num": 0.1396484375,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 470862976,
+      "step": 2724
+    },
+    {
+      "epoch": 1.0436614324013789,
+      "grad_norm": 39.07446599870957,
+      "learning_rate": 5e-06,
+      "loss": 0.2192,
+      "num_input_tokens_seen": 471035872,
+      "step": 2725
+    },
+    {
+      "epoch": 1.0436614324013789,
+      "loss": 0.22075331211090088,
+      "loss_ce": 0.022144924849271774,
+      "loss_iou": 1.0329551696777344,
+      "loss_num": 0.1982421875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 471035872,
+      "step": 2725
+    },
+    {
+      "epoch": 1.0440444274224434,
+      "grad_norm": 28.588671946034065,
+      "learning_rate": 5e-06,
+      "loss": 0.2162,
+      "num_input_tokens_seen": 471208616,
+      "step": 2726
+    },
+    {
+      "epoch": 1.0440444274224434,
+      "loss": 0.20154617726802826,
+      "loss_ce": 0.019661404192447662,
+      "loss_iou": 1.0136446952819824,
+      "loss_num": 0.181640625,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 471208616,
+      "step": 2726
+    },
+    {
+      "epoch": 1.0444274224435082,
+      "grad_norm": 40.00772693807137,
+      "learning_rate": 5e-06,
+      "loss": 0.1639,
+      "num_input_tokens_seen": 471381824,
+      "step": 2727
+    },
+    {
+      "epoch": 1.0444274224435082,
+      "loss": 0.16869238018989563,
+      "loss_ce": 0.02336767688393593,
+      "loss_iou": 1.0468491315841675,
+      "loss_num": 0.1455078125,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 471381824,
+      "step": 2727
+    },
+    {
+      "epoch": 1.044810417464573,
+      "grad_norm": 28.13426340486647,
+      "learning_rate": 5e-06,
+      "loss": 0.1684,
+      "num_input_tokens_seen": 471554920,
+      "step": 2728
+    },
+    {
+      "epoch": 1.044810417464573,
+      "loss": 0.12310892343521118,
+      "loss_ce": 0.021546421572566032,
+      "loss_iou": 1.0003392696380615,
+      "loss_num": 0.1015625,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 471554920,
+      "step": 2728
+    },
+    {
+      "epoch": 1.0451934124856377,
+      "grad_norm": 29.710748314738474,
+      "learning_rate": 5e-06,
+      "loss": 0.1646,
+      "num_input_tokens_seen": 471728248,
+      "step": 2729
+    },
+    {
+      "epoch": 1.0451934124856377,
+      "loss": 0.14808785915374756,
+      "loss_ce": 0.023087866604328156,
+      "loss_iou": 1.005138874053955,
+      "loss_num": 0.125,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 471728248,
+      "step": 2729
+    },
+    {
+      "epoch": 1.0455764075067024,
+      "grad_norm": 39.41039975761352,
+      "learning_rate": 5e-06,
+      "loss": 0.177,
+      "num_input_tokens_seen": 471901096,
+      "step": 2730
+    },
+    {
+      "epoch": 1.0455764075067024,
+      "loss": 0.1642495095729828,
+      "loss_ce": 0.019199464470148087,
+      "loss_iou": 1.1425344944000244,
+      "loss_num": 0.1455078125,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 471901096,
+      "step": 2730
+    },
+    {
+      "epoch": 1.0459594025277672,
+      "grad_norm": 79.7902468080741,
+      "learning_rate": 5e-06,
+      "loss": 0.2167,
+      "num_input_tokens_seen": 472074096,
+      "step": 2731
+    },
+    {
+      "epoch": 1.0459594025277672,
+      "loss": 0.2294425517320633,
+      "loss_ce": 0.023570969700813293,
+      "loss_iou": 1.0148272514343262,
+      "loss_num": 0.2060546875,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 472074096,
+      "step": 2731
+    },
+    {
+      "epoch": 1.046342397548832,
+      "grad_norm": 27.929160338654125,
+      "learning_rate": 5e-06,
+      "loss": 0.2061,
+      "num_input_tokens_seen": 472246752,
+      "step": 2732
+    },
+    {
+      "epoch": 1.046342397548832,
+      "loss": 0.22269487380981445,
+      "loss_ce": 0.01993608847260475,
+      "loss_iou": 1.0086251497268677,
+      "loss_num": 0.203125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 472246752,
+      "step": 2732
+    },
+    {
+      "epoch": 1.0467253925698965,
+      "grad_norm": 66.44112395580295,
+      "learning_rate": 5e-06,
+      "loss": 0.2601,
+      "num_input_tokens_seen": 472419664,
+      "step": 2733
+    },
+    {
+      "epoch": 1.0467253925698965,
+      "loss": 0.3042086362838745,
+      "loss_ce": 0.02039516344666481,
+      "loss_iou": 1.005509376525879,
+      "loss_num": 0.283203125,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 472419664,
+      "step": 2733
+    },
+    {
+      "epoch": 1.0471083875909613,
+      "grad_norm": 39.12380861449442,
+      "learning_rate": 5e-06,
+      "loss": 0.2179,
+      "num_input_tokens_seen": 472592408,
+      "step": 2734
+    },
+    {
+      "epoch": 1.0471083875909613,
+      "loss": 0.16589589416980743,
+      "loss_ce": 0.021212061867117882,
+      "loss_iou": 0.8990572094917297,
+      "loss_num": 0.14453125,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 472592408,
+      "step": 2734
+    },
+    {
+      "epoch": 1.047491382612026,
+      "grad_norm": 87.35692921222793,
+      "learning_rate": 5e-06,
+      "loss": 0.3555,
+      "num_input_tokens_seen": 472765632,
+      "step": 2735
+    },
+    {
+      "epoch": 1.047491382612026,
+      "loss": 0.33498415350914,
+      "loss_ce": 0.021385516971349716,
+      "loss_iou": 1.032576560974121,
+      "loss_num": 0.314453125,
+      "loss_xval": 0.314453125,
+      "num_input_tokens_seen": 472765632,
+      "step": 2735
+    },
+    {
+      "epoch": 1.0478743776330908,
+      "grad_norm": 35.16914924326974,
+      "learning_rate": 5e-06,
+      "loss": 0.2094,
+      "num_input_tokens_seen": 472938440,
+      "step": 2736
+    },
+    {
+      "epoch": 1.0478743776330908,
+      "loss": 0.19124934077262878,
+      "loss_ce": 0.019740547984838486,
+      "loss_iou": 1.0017101764678955,
+      "loss_num": 0.171875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 472938440,
+      "step": 2736
+    },
+    {
+      "epoch": 1.0482573726541555,
+      "grad_norm": 26.90994084526525,
+      "learning_rate": 5e-06,
+      "loss": 0.2423,
+      "num_input_tokens_seen": 473111736,
+      "step": 2737
+    },
+    {
+      "epoch": 1.0482573726541555,
+      "loss": 0.24850554764270782,
+      "loss_ce": 0.020356133580207825,
+      "loss_iou": 1.0285611152648926,
+      "loss_num": 0.228515625,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 473111736,
+      "step": 2737
+    },
+    {
+      "epoch": 1.0486403676752203,
+      "grad_norm": 26.95539026300767,
+      "learning_rate": 5e-06,
+      "loss": 0.212,
+      "num_input_tokens_seen": 473284680,
+      "step": 2738
+    },
+    {
+      "epoch": 1.0486403676752203,
+      "loss": 0.1881379336118698,
+      "loss_ce": 0.019558828324079514,
+      "loss_iou": 1.0077383518218994,
+      "loss_num": 0.1689453125,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 473284680,
+      "step": 2738
+    },
+    {
+      "epoch": 1.049023362696285,
+      "grad_norm": 57.78363821465061,
+      "learning_rate": 5e-06,
+      "loss": 0.272,
+      "num_input_tokens_seen": 473457608,
+      "step": 2739
+    },
+    {
+      "epoch": 1.049023362696285,
+      "loss": 0.23863255977630615,
+      "loss_ce": 0.021103277802467346,
+      "loss_iou": 1.0214968919754028,
+      "loss_num": 0.2177734375,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 473457608,
+      "step": 2739
+    },
+    {
+      "epoch": 1.0494063577173496,
+      "grad_norm": 48.252095532275206,
+      "learning_rate": 5e-06,
+      "loss": 0.2814,
+      "num_input_tokens_seen": 473630624,
+      "step": 2740
+    },
+    {
+      "epoch": 1.0494063577173496,
+      "loss": 0.27010446786880493,
+      "loss_ce": 0.02132517099380493,
+      "loss_iou": 1.046252965927124,
+      "loss_num": 0.2490234375,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 473630624,
+      "step": 2740
+    },
+    {
+      "epoch": 1.0497893527384143,
+      "grad_norm": 48.47258890101169,
+      "learning_rate": 5e-06,
+      "loss": 0.2966,
+      "num_input_tokens_seen": 473803488,
+      "step": 2741
+    },
+    {
+      "epoch": 1.0497893527384143,
+      "loss": 0.28845489025115967,
+      "loss_ce": 0.021242991089820862,
+      "loss_iou": 1.019432783126831,
+      "loss_num": 0.267578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 473803488,
+      "step": 2741
+    },
+    {
+      "epoch": 1.050172347759479,
+      "grad_norm": 34.68729107241668,
+      "learning_rate": 5e-06,
+      "loss": 0.2335,
+      "num_input_tokens_seen": 473976120,
+      "step": 2742
+    },
+    {
+      "epoch": 1.050172347759479,
+      "loss": 0.26116693019866943,
+      "loss_ce": 0.021237719804048538,
+      "loss_iou": 1.1353564262390137,
+      "loss_num": 0.240234375,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 473976120,
+      "step": 2742
+    },
+    {
+      "epoch": 1.0505553427805439,
+      "grad_norm": 91.08869170976169,
+      "learning_rate": 5e-06,
+      "loss": 0.3836,
+      "num_input_tokens_seen": 474149240,
+      "step": 2743
+    },
+    {
+      "epoch": 1.0505553427805439,
+      "loss": 0.3859593868255615,
+      "loss_ce": 0.021701589226722717,
+      "loss_iou": 1.0187512636184692,
+      "loss_num": 0.36328125,
+      "loss_xval": 0.36328125,
+      "num_input_tokens_seen": 474149240,
+      "step": 2743
+    },
+    {
+      "epoch": 1.0509383378016086,
+      "grad_norm": 37.43984099159935,
+      "learning_rate": 5e-06,
+      "loss": 0.2488,
+      "num_input_tokens_seen": 474322184,
+      "step": 2744
+    },
+    {
+      "epoch": 1.0509383378016086,
+      "loss": 0.2325676679611206,
+      "loss_ce": 0.019677046686410904,
+      "loss_iou": 1.0216319561004639,
+      "loss_num": 0.212890625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 474322184,
+      "step": 2744
+    },
+    {
+      "epoch": 1.0513213328226734,
+      "grad_norm": 96.43731864902831,
+      "learning_rate": 5e-06,
+      "loss": 0.5828,
+      "num_input_tokens_seen": 474495384,
+      "step": 2745
+    },
+    {
+      "epoch": 1.0513213328226734,
+      "loss": 0.6128767728805542,
+      "loss_ce": 0.021079927682876587,
+      "loss_iou": 1.0349185466766357,
+      "loss_num": 0.59375,
+      "loss_xval": 0.59375,
+      "num_input_tokens_seen": 474495384,
+      "step": 2745
+    },
+    {
+      "epoch": 1.0517043278437381,
+      "grad_norm": 33.287081245733056,
+      "learning_rate": 5e-06,
+      "loss": 0.1964,
+      "num_input_tokens_seen": 474668184,
+      "step": 2746
+    },
+    {
+      "epoch": 1.0517043278437381,
+      "loss": 0.22697071731090546,
+      "loss_ce": 0.019268084317445755,
+      "loss_iou": 1.056471347808838,
+      "loss_num": 0.2080078125,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 474668184,
+      "step": 2746
+    },
+    {
+      "epoch": 1.0520873228648027,
+      "grad_norm": 98.93006472064867,
+      "learning_rate": 5e-06,
+      "loss": 0.6221,
+      "num_input_tokens_seen": 474841064,
+      "step": 2747
+    },
+    {
+      "epoch": 1.0520873228648027,
+      "loss": 0.6143052577972412,
+      "loss_ce": 0.01860210858285427,
+      "loss_iou": 1.0286753177642822,
+      "loss_num": 0.59375,
+      "loss_xval": 0.59375,
+      "num_input_tokens_seen": 474841064,
+      "step": 2747
+    },
+    {
+      "epoch": 1.0524703178858674,
+      "grad_norm": 32.246487920245904,
+      "learning_rate": 5e-06,
+      "loss": 0.2712,
+      "num_input_tokens_seen": 475014160,
+      "step": 2748
+    },
+    {
+      "epoch": 1.0524703178858674,
+      "loss": 0.30843839049339294,
+      "loss_ce": 0.023648329079151154,
+      "loss_iou": 1.0537768602371216,
+      "loss_num": 0.28515625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 475014160,
+      "step": 2748
+    },
+    {
+      "epoch": 1.0528533129069322,
+      "grad_norm": 36.094590295797396,
+      "learning_rate": 5e-06,
+      "loss": 0.2888,
+      "num_input_tokens_seen": 475186936,
+      "step": 2749
+    },
+    {
+      "epoch": 1.0528533129069322,
+      "loss": 0.3027374744415283,
+      "loss_ce": 0.019778510555624962,
+      "loss_iou": 1.0816279649734497,
+      "loss_num": 0.283203125,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 475186936,
+      "step": 2749
+    },
+    {
+      "epoch": 1.053236307927997,
+      "grad_norm": 31.006086293118372,
+      "learning_rate": 5e-06,
+      "loss": 0.2068,
+      "num_input_tokens_seen": 475359608,
+      "step": 2750
+    },
+    {
+      "epoch": 1.053236307927997,
+      "eval_websight_new_CIoU": 0.9340087175369263,
+      "eval_websight_new_GIoU": 0.9337825775146484,
+      "eval_websight_new_IoU": 0.9342544078826904,
+      "eval_websight_new_MAE_all": 0.006378704681992531,
+      "eval_websight_new_MAE_h": 0.005829754751175642,
+      "eval_websight_new_MAE_w": 0.007576425559818745,
+      "eval_websight_new_MAE_x": 0.00746025238186121,
+      "eval_websight_new_MAE_y": 0.004648384638130665,
+      "eval_websight_new_NUM_probability": 6.985414438531734e-05,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.10419899225234985,
+      "eval_websight_new_loss_ce": 0.04258443042635918,
+      "eval_websight_new_loss_iou": 1.0004280805587769,
+      "eval_websight_new_loss_num": 0.06005859375,
+      "eval_websight_new_loss_xval": 0.06005859375,
+      "eval_websight_new_runtime": 55.01,
+      "eval_websight_new_samples_per_second": 0.909,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 475359608,
+      "step": 2750
+    },
+    {
+      "epoch": 1.053236307927997,
+      "eval_seeclick_CIoU": 0.7091011106967926,
+      "eval_seeclick_GIoU": 0.715798556804657,
+      "eval_seeclick_IoU": 0.736909419298172,
+      "eval_seeclick_MAE_all": 0.04782724380493164,
+      "eval_seeclick_MAE_h": 0.025104526430368423,
+      "eval_seeclick_MAE_w": 0.07058498449623585,
+      "eval_seeclick_MAE_x": 0.06600936688482761,
+      "eval_seeclick_MAE_y": 0.029610102996230125,
+      "eval_seeclick_NUM_probability": 9.428920384380035e-05,
+      "eval_seeclick_inside_bbox": 0.8975694477558136,
+      "eval_seeclick_loss": 0.3992612361907959,
+      "eval_seeclick_loss_ce": 0.03321938402950764,
+      "eval_seeclick_loss_iou": 1.1248716711997986,
+      "eval_seeclick_loss_num": 0.3755340576171875,
+      "eval_seeclick_loss_xval": 0.3755340576171875,
+      "eval_seeclick_runtime": 89.4809,
+      "eval_seeclick_samples_per_second": 0.559,
+      "eval_seeclick_steps_per_second": 0.022,
+      "num_input_tokens_seen": 475359608,
+      "step": 2750
+    },
+    {
+      "epoch": 1.053236307927997,
+      "eval_icons_CIoU": 0.8451197743415833,
+      "eval_icons_GIoU": 0.8467774093151093,
+      "eval_icons_IoU": 0.8570410311222076,
+      "eval_icons_MAE_all": 0.0221947580575943,
+      "eval_icons_MAE_h": 0.013928203843533993,
+      "eval_icons_MAE_w": 0.03198568522930145,
+      "eval_icons_MAE_x": 0.03160202130675316,
+      "eval_icons_MAE_y": 0.011263125110417604,
+      "eval_icons_NUM_probability": 6.373344513121992e-05,
+      "eval_icons_inside_bbox": 0.9288194477558136,
+      "eval_icons_loss": 0.16468946635723114,
+      "eval_icons_loss_ce": 0.028911873698234558,
+      "eval_icons_loss_iou": 1.045575201511383,
+      "eval_icons_loss_num": 0.122589111328125,
+      "eval_icons_loss_xval": 0.122589111328125,
+      "eval_icons_runtime": 84.3458,
+      "eval_icons_samples_per_second": 0.593,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 475359608,
+      "step": 2750
+    },
+    {
+      "epoch": 1.0536193029490617,
+      "grad_norm": 20.972395160202332,
+      "learning_rate": 5e-06,
+      "loss": 4.9244,
+      "num_input_tokens_seen": 475532696,
+      "step": 2751
+    },
+    {
+      "epoch": 1.0536193029490617,
+      "loss": 4.94212532043457,
+      "loss_ce": 0.7298550605773926,
+      "loss_iou": 1.8575999736785889,
+      "loss_num": 0.099609375,
+      "loss_xval": 4.212270259857178,
+      "num_input_tokens_seen": 475532696,
+      "step": 2751
+    },
+    {
+      "epoch": 1.0540022979701265,
+      "grad_norm": 22.579961359595984,
+      "learning_rate": 5e-06,
+      "loss": 4.8401,
+      "num_input_tokens_seen": 475706144,
+      "step": 2752
+    },
+    {
+      "epoch": 1.0540022979701265,
+      "loss": 4.863560199737549,
+      "loss_ce": 0.7389425039291382,
+      "loss_iou": 1.8568644523620605,
+      "loss_num": 0.08203125,
+      "loss_xval": 4.124617576599121,
+      "num_input_tokens_seen": 475706144,
+      "step": 2752
+    },
+    {
+      "epoch": 1.0543852929911912,
+      "grad_norm": 38.90659473774426,
+      "learning_rate": 5e-06,
+      "loss": 4.5824,
+      "num_input_tokens_seen": 475878984,
+      "step": 2753
+    },
+    {
+      "epoch": 1.0543852929911912,
+      "loss": 4.571383476257324,
+      "loss_ce": 0.713236391544342,
+      "loss_iou": 1.7164270877838135,
+      "loss_num": 0.0849609375,
+      "loss_xval": 3.858147144317627,
+      "num_input_tokens_seen": 475878984,
+      "step": 2753
+    },
+    {
+      "epoch": 1.0547682880122558,
+      "grad_norm": 51.251617120540054,
+      "learning_rate": 5e-06,
+      "loss": 4.74,
+      "num_input_tokens_seen": 476051824,
+      "step": 2754
+    },
+    {
+      "epoch": 1.0547682880122558,
+      "loss": 4.697440147399902,
+      "loss_ce": 0.6825171709060669,
+      "loss_iou": 1.7874908447265625,
+      "loss_num": 0.087890625,
+      "loss_xval": 4.014923095703125,
+      "num_input_tokens_seen": 476051824,
+      "step": 2754
+    },
+    {
+      "epoch": 1.0551512830333205,
+      "grad_norm": 86.9999704302525,
+      "learning_rate": 5e-06,
+      "loss": 4.2924,
+      "num_input_tokens_seen": 476224392,
+      "step": 2755
+    },
+    {
+      "epoch": 1.0551512830333205,
+      "loss": 4.226644992828369,
+      "loss_ce": 0.586954653263092,
+      "loss_iou": 1.6256312131881714,
+      "loss_num": 0.07763671875,
+      "loss_xval": 3.6396901607513428,
+      "num_input_tokens_seen": 476224392,
+      "step": 2755
+    },
+    {
+      "epoch": 1.0555342780543853,
+      "grad_norm": 89.00515259187776,
+      "learning_rate": 5e-06,
+      "loss": 4.7428,
+      "num_input_tokens_seen": 476397344,
+      "step": 2756
+    },
+    {
+      "epoch": 1.0555342780543853,
+      "loss": 4.77473258972168,
+      "loss_ce": 0.7308616042137146,
+      "loss_iou": 1.7889032363891602,
+      "loss_num": 0.09326171875,
+      "loss_xval": 4.04387092590332,
+      "num_input_tokens_seen": 476397344,
+      "step": 2756
+    },
+    {
+      "epoch": 1.05591727307545,
+      "grad_norm": 185.9105092239929,
+      "learning_rate": 5e-06,
+      "loss": 4.0115,
+      "num_input_tokens_seen": 476566464,
+      "step": 2757
+    },
+    {
+      "epoch": 1.05591727307545,
+      "loss": 3.987050771713257,
+      "loss_ce": 0.747867226600647,
+      "loss_iou": 1.4663325548171997,
+      "loss_num": 0.061279296875,
+      "loss_xval": 3.2391836643218994,
+      "num_input_tokens_seen": 476566464,
+      "step": 2757
+    },
+    {
+      "epoch": 1.0563002680965148,
+      "grad_norm": 94.49240018056493,
+      "learning_rate": 5e-06,
+      "loss": 4.7167,
+      "num_input_tokens_seen": 476739216,
+      "step": 2758
+    },
+    {
+      "epoch": 1.0563002680965148,
+      "loss": 4.594268798828125,
+      "loss_ce": 0.6055147647857666,
+      "loss_iou": 1.3898848295211792,
+      "loss_num": 0.2421875,
+      "loss_xval": 3.9887540340423584,
+      "num_input_tokens_seen": 476739216,
+      "step": 2758
+    },
+    {
+      "epoch": 1.0566832631175795,
+      "grad_norm": 237.09172010163056,
+      "learning_rate": 5e-06,
+      "loss": 3.9164,
+      "num_input_tokens_seen": 476912080,
+      "step": 2759
+    },
+    {
+      "epoch": 1.0566832631175795,
+      "loss": 3.877072334289551,
+      "loss_ce": 0.6420649886131287,
+      "loss_iou": 1.0923571586608887,
+      "loss_num": 0.2099609375,
+      "loss_xval": 3.2350072860717773,
+      "num_input_tokens_seen": 476912080,
+      "step": 2759
+    },
+    {
+      "epoch": 1.0570662581386443,
+      "grad_norm": 150.1091672828241,
+      "learning_rate": 5e-06,
+      "loss": 3.3728,
+      "num_input_tokens_seen": 477085408,
+      "step": 2760
+    },
+    {
+      "epoch": 1.0570662581386443,
+      "loss": 3.46321177482605,
+      "loss_ce": 0.6880149841308594,
+      "loss_iou": 1.1341803073883057,
+      "loss_num": 0.1015625,
+      "loss_xval": 2.7751965522766113,
+      "num_input_tokens_seen": 477085408,
+      "step": 2760
+    },
+    {
+      "epoch": 1.0574492531597088,
+      "grad_norm": 285.7596434576294,
+      "learning_rate": 5e-06,
+      "loss": 3.9921,
+      "num_input_tokens_seen": 477257920,
+      "step": 2761
+    },
+    {
+      "epoch": 1.0574492531597088,
+      "loss": 3.7242329120635986,
+      "loss_ce": 0.5976026654243469,
+      "loss_iou": 1.0127780437469482,
+      "loss_num": 0.220703125,
+      "loss_xval": 3.1266303062438965,
+      "num_input_tokens_seen": 477257920,
+      "step": 2761
+    },
+    {
+      "epoch": 1.0578322481807736,
+      "grad_norm": 133.89055976999657,
+      "learning_rate": 5e-06,
+      "loss": 3.3863,
+      "num_input_tokens_seen": 477431136,
+      "step": 2762
+    },
+    {
+      "epoch": 1.0578322481807736,
+      "loss": 3.3770768642425537,
+      "loss_ce": 0.7506231069564819,
+      "loss_iou": 1.2131903171539307,
+      "loss_num": 0.0400390625,
+      "loss_xval": 2.6264538764953613,
+      "num_input_tokens_seen": 477431136,
+      "step": 2762
+    },
+    {
+      "epoch": 1.0582152432018384,
+      "grad_norm": 79.24134849478921,
+      "learning_rate": 5e-06,
+      "loss": 3.0953,
+      "num_input_tokens_seen": 477604256,
+      "step": 2763
+    },
+    {
+      "epoch": 1.0582152432018384,
+      "loss": 2.9831225872039795,
+      "loss_ce": 0.6824136972427368,
+      "loss_iou": 0.9375858902931213,
+      "loss_num": 0.0849609375,
+      "loss_xval": 2.3007090091705322,
+      "num_input_tokens_seen": 477604256,
+      "step": 2763
+    },
+    {
+      "epoch": 1.058598238222903,
+      "grad_norm": 92.58299551211535,
+      "learning_rate": 5e-06,
+      "loss": 2.6359,
+      "num_input_tokens_seen": 477776768,
+      "step": 2764
+    },
+    {
+      "epoch": 1.058598238222903,
+      "loss": 2.6042308807373047,
+      "loss_ce": 0.6946655511856079,
+      "loss_iou": 0.8491918444633484,
+      "loss_num": 0.042236328125,
+      "loss_xval": 1.9095653295516968,
+      "num_input_tokens_seen": 477776768,
+      "step": 2764
+    },
+    {
+      "epoch": 1.0589812332439679,
+      "grad_norm": 216.32596930989925,
+      "learning_rate": 5e-06,
+      "loss": 3.9107,
+      "num_input_tokens_seen": 477949632,
+      "step": 2765
+    },
+    {
+      "epoch": 1.0589812332439679,
+      "loss": 3.8467416763305664,
+      "loss_ce": 0.7250851392745972,
+      "loss_iou": 1.4247809648513794,
+      "loss_num": 0.054443359375,
+      "loss_xval": 3.121656656265259,
+      "num_input_tokens_seen": 477949632,
+      "step": 2765
+    },
+    {
+      "epoch": 1.0593642282650326,
+      "grad_norm": 39.851080910056396,
+      "learning_rate": 5e-06,
+      "loss": 2.8466,
+      "num_input_tokens_seen": 478121952,
+      "step": 2766
+    },
+    {
+      "epoch": 1.0593642282650326,
+      "loss": 3.068800687789917,
+      "loss_ce": 0.5916388034820557,
+      "loss_iou": 1.0230047702789307,
+      "loss_num": 0.08642578125,
+      "loss_xval": 2.4771618843078613,
+      "num_input_tokens_seen": 478121952,
+      "step": 2766
+    },
+    {
+      "epoch": 1.0597472232860974,
+      "grad_norm": 65.348712385135,
+      "learning_rate": 5e-06,
+      "loss": 2.6881,
+      "num_input_tokens_seen": 478294696,
+      "step": 2767
+    },
+    {
+      "epoch": 1.0597472232860974,
+      "loss": 2.7627148628234863,
+      "loss_ce": 0.7295560836791992,
+      "loss_iou": 0.8837059736251831,
+      "loss_num": 0.05322265625,
+      "loss_xval": 2.033158779144287,
+      "num_input_tokens_seen": 478294696,
+      "step": 2767
+    },
+    {
+      "epoch": 1.060130218307162,
+      "grad_norm": 167.5137190983422,
+      "learning_rate": 5e-06,
+      "loss": 2.9012,
+      "num_input_tokens_seen": 478467216,
+      "step": 2768
+    },
+    {
+      "epoch": 1.060130218307162,
+      "loss": 2.764270544052124,
+      "loss_ce": 0.6801129579544067,
+      "loss_iou": 0.9556529521942139,
+      "loss_num": 0.03466796875,
+      "loss_xval": 2.0841574668884277,
+      "num_input_tokens_seen": 478467216,
+      "step": 2768
+    },
+    {
+      "epoch": 1.0605132133282267,
+      "grad_norm": 97.4310789988781,
+      "learning_rate": 5e-06,
+      "loss": 3.3915,
+      "num_input_tokens_seen": 478640048,
+      "step": 2769
+    },
+    {
+      "epoch": 1.0605132133282267,
+      "loss": 3.468452215194702,
+      "loss_ce": 0.6962400674819946,
+      "loss_iou": 1.134763240814209,
+      "loss_num": 0.1005859375,
+      "loss_xval": 2.772212028503418,
+      "num_input_tokens_seen": 478640048,
+      "step": 2769
+    },
+    {
+      "epoch": 1.0608962083492914,
+      "grad_norm": 82.34273348062686,
+      "learning_rate": 5e-06,
+      "loss": 3.1041,
+      "num_input_tokens_seen": 478813136,
+      "step": 2770
+    },
+    {
+      "epoch": 1.0608962083492914,
+      "loss": 3.1314048767089844,
+      "loss_ce": 0.7152508497238159,
+      "loss_iou": 0.9957966804504395,
+      "loss_num": 0.0849609375,
+      "loss_xval": 2.416153907775879,
+      "num_input_tokens_seen": 478813136,
+      "step": 2770
+    },
+    {
+      "epoch": 1.0612792033703562,
+      "grad_norm": 60.91499911640653,
+      "learning_rate": 5e-06,
+      "loss": 2.7954,
+      "num_input_tokens_seen": 478986136,
+      "step": 2771
+    },
+    {
+      "epoch": 1.0612792033703562,
+      "loss": 2.714437961578369,
+      "loss_ce": 0.7429600954055786,
+      "loss_iou": 0.8766692280769348,
+      "loss_num": 0.043701171875,
+      "loss_xval": 1.9714781045913696,
+      "num_input_tokens_seen": 478986136,
+      "step": 2771
+    },
+    {
+      "epoch": 1.061662198391421,
+      "grad_norm": 220.12619331832371,
+      "learning_rate": 5e-06,
+      "loss": 2.967,
+      "num_input_tokens_seen": 479158960,
+      "step": 2772
+    },
+    {
+      "epoch": 1.061662198391421,
+      "loss": 3.0788824558258057,
+      "loss_ce": 0.6809020638465881,
+      "loss_iou": 1.1042025089263916,
+      "loss_num": 0.037841796875,
+      "loss_xval": 2.397980213165283,
+      "num_input_tokens_seen": 479158960,
+      "step": 2772
+    },
+    {
+      "epoch": 1.0620451934124857,
+      "grad_norm": 147.62426189121615,
+      "learning_rate": 5e-06,
+      "loss": 3.4391,
+      "num_input_tokens_seen": 479331592,
+      "step": 2773
+    },
+    {
+      "epoch": 1.0620451934124857,
+      "loss": 3.3189139366149902,
+      "loss_ce": 0.695880115032196,
+      "loss_iou": 1.0483334064483643,
+      "loss_num": 0.10498046875,
+      "loss_xval": 2.6230340003967285,
+      "num_input_tokens_seen": 479331592,
+      "step": 2773
+    },
+    {
+      "epoch": 1.0624281884335505,
+      "grad_norm": 81.79723496561954,
+      "learning_rate": 5e-06,
+      "loss": 3.1469,
+      "num_input_tokens_seen": 479504992,
+      "step": 2774
+    },
+    {
+      "epoch": 1.0624281884335505,
+      "loss": 3.269740104675293,
+      "loss_ce": 0.7365933656692505,
+      "loss_iou": 0.9709189534187317,
+      "loss_num": 0.1181640625,
+      "loss_xval": 2.533146381378174,
+      "num_input_tokens_seen": 479504992,
+      "step": 2774
+    },
+    {
+      "epoch": 1.062811183454615,
+      "grad_norm": 77.8703028378014,
+      "learning_rate": 5e-06,
+      "loss": 2.7768,
+      "num_input_tokens_seen": 479677592,
+      "step": 2775
+    },
+    {
+      "epoch": 1.062811183454615,
+      "loss": 2.8037338256835938,
+      "loss_ce": 0.69895339012146,
+      "loss_iou": 0.9126807451248169,
+      "loss_num": 0.055908203125,
+      "loss_xval": 2.104780435562134,
+      "num_input_tokens_seen": 479677592,
+      "step": 2775
+    },
+    {
+      "epoch": 1.0631941784756798,
+      "grad_norm": 125.95945099555713,
+      "learning_rate": 5e-06,
+      "loss": 2.9744,
+      "num_input_tokens_seen": 479850744,
+      "step": 2776
+    },
+    {
+      "epoch": 1.0631941784756798,
+      "loss": 2.8986823558807373,
+      "loss_ce": 0.6944018602371216,
+      "loss_iou": 0.9481484889984131,
+      "loss_num": 0.0615234375,
+      "loss_xval": 2.204280376434326,
+      "num_input_tokens_seen": 479850744,
+      "step": 2776
+    },
+    {
+      "epoch": 1.0635771734967445,
+      "grad_norm": 70.51303175752362,
+      "learning_rate": 5e-06,
+      "loss": 2.6296,
+      "num_input_tokens_seen": 480023552,
+      "step": 2777
+    },
+    {
+      "epoch": 1.0635771734967445,
+      "loss": 2.730379581451416,
+      "loss_ce": 0.7116711139678955,
+      "loss_iou": 0.8874670267105103,
+      "loss_num": 0.048828125,
+      "loss_xval": 2.0187084674835205,
+      "num_input_tokens_seen": 480023552,
+      "step": 2777
+    },
+    {
+      "epoch": 1.0639601685178093,
+      "grad_norm": 42.99076277028715,
+      "learning_rate": 5e-06,
+      "loss": 2.5426,
+      "num_input_tokens_seen": 480196704,
+      "step": 2778
+    },
+    {
+      "epoch": 1.0639601685178093,
+      "loss": 2.469575881958008,
+      "loss_ce": 0.718497097492218,
+      "loss_iou": 0.8037315607070923,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 1.7510788440704346,
+      "num_input_tokens_seen": 480196704,
+      "step": 2778
+    },
+    {
+      "epoch": 1.064343163538874,
+      "grad_norm": 39.271392985638734,
+      "learning_rate": 5e-06,
+      "loss": 2.2897,
+      "num_input_tokens_seen": 480369528,
+      "step": 2779
+    },
+    {
+      "epoch": 1.064343163538874,
+      "loss": 2.2908480167388916,
+      "loss_ce": 0.7359123229980469,
+      "loss_iou": 0.7067586183547974,
+      "loss_num": 0.0283203125,
+      "loss_xval": 1.5549356937408447,
+      "num_input_tokens_seen": 480369528,
+      "step": 2779
+    },
+    {
+      "epoch": 1.0647261585599388,
+      "grad_norm": 131.43859578584966,
+      "learning_rate": 5e-06,
+      "loss": 2.5243,
+      "num_input_tokens_seen": 480542824,
+      "step": 2780
+    },
+    {
+      "epoch": 1.0647261585599388,
+      "loss": 2.5856826305389404,
+      "loss_ce": 0.6940747499465942,
+      "loss_iou": 0.8606293201446533,
+      "loss_num": 0.033935546875,
+      "loss_xval": 1.8916077613830566,
+      "num_input_tokens_seen": 480542824,
+      "step": 2780
+    },
+    {
+      "epoch": 1.0651091535810036,
+      "grad_norm": 87.11997550244749,
+      "learning_rate": 5e-06,
+      "loss": 3.6927,
+      "num_input_tokens_seen": 480715432,
+      "step": 2781
+    },
+    {
+      "epoch": 1.0651091535810036,
+      "loss": 3.9174413681030273,
+      "loss_ce": 0.6469007730484009,
+      "loss_iou": 1.4599162340164185,
+      "loss_num": 0.0703125,
+      "loss_xval": 3.270540475845337,
+      "num_input_tokens_seen": 480715432,
+      "step": 2781
+    },
+    {
+      "epoch": 1.065492148602068,
+      "grad_norm": 108.72209188483511,
+      "learning_rate": 5e-06,
+      "loss": 2.7006,
+      "num_input_tokens_seen": 480888360,
+      "step": 2782
+    },
+    {
+      "epoch": 1.065492148602068,
+      "loss": 2.7656283378601074,
+      "loss_ce": 0.7435428500175476,
+      "loss_iou": 0.8747512698173523,
+      "loss_num": 0.054443359375,
+      "loss_xval": 2.022085666656494,
+      "num_input_tokens_seen": 480888360,
+      "step": 2782
+    },
+    {
+      "epoch": 1.0658751436231328,
+      "grad_norm": 246.58809910604464,
+      "learning_rate": 5e-06,
+      "loss": 2.5774,
+      "num_input_tokens_seen": 481061448,
+      "step": 2783
+    },
+    {
+      "epoch": 1.0658751436231328,
+      "loss": 2.5640876293182373,
+      "loss_ce": 0.6850498914718628,
+      "loss_iou": 0.8536118865013123,
+      "loss_num": 0.034423828125,
+      "loss_xval": 1.8790377378463745,
+      "num_input_tokens_seen": 481061448,
+      "step": 2783
+    },
+    {
+      "epoch": 1.0662581386441976,
+      "grad_norm": 100.41665743419827,
+      "learning_rate": 5e-06,
+      "loss": 2.94,
+      "num_input_tokens_seen": 481234256,
+      "step": 2784
+    },
+    {
+      "epoch": 1.0662581386441976,
+      "loss": 2.8582053184509277,
+      "loss_ce": 0.6741833686828613,
+      "loss_iou": 0.9432071447372437,
+      "loss_num": 0.0595703125,
+      "loss_xval": 2.1840219497680664,
+      "num_input_tokens_seen": 481234256,
+      "step": 2784
+    },
+    {
+      "epoch": 1.0666411336652624,
+      "grad_norm": 130.65958261450226,
+      "learning_rate": 5e-06,
+      "loss": 2.6775,
+      "num_input_tokens_seen": 481407112,
+      "step": 2785
+    },
+    {
+      "epoch": 1.0666411336652624,
+      "loss": 2.661080837249756,
+      "loss_ce": 0.7737404108047485,
+      "loss_iou": 0.8334407210350037,
+      "loss_num": 0.0439453125,
+      "loss_xval": 1.8873404264450073,
+      "num_input_tokens_seen": 481407112,
+      "step": 2785
+    },
+    {
+      "epoch": 1.0670241286863271,
+      "grad_norm": 51.86095039881347,
+      "learning_rate": 5e-06,
+      "loss": 2.6063,
+      "num_input_tokens_seen": 481580584,
+      "step": 2786
+    },
+    {
+      "epoch": 1.0670241286863271,
+      "loss": 2.6839590072631836,
+      "loss_ce": 0.6764041185379028,
+      "loss_iou": 0.9079523086547852,
+      "loss_num": 0.038330078125,
+      "loss_xval": 2.0075550079345703,
+      "num_input_tokens_seen": 481580584,
+      "step": 2786
+    },
+    {
+      "epoch": 1.0674071237073919,
+      "grad_norm": 173.088161838326,
+      "learning_rate": 5e-06,
+      "loss": 2.3757,
+      "num_input_tokens_seen": 481753856,
+      "step": 2787
+    },
+    {
+      "epoch": 1.0674071237073919,
+      "loss": 2.3286516666412354,
+      "loss_ce": 0.6821632981300354,
+      "loss_iou": 0.7492085695266724,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 1.6464884281158447,
+      "num_input_tokens_seen": 481753856,
+      "step": 2787
+    },
+    {
+      "epoch": 1.0677901187284564,
+      "grad_norm": 82.01832109897481,
+      "learning_rate": 5e-06,
+      "loss": 2.5041,
+      "num_input_tokens_seen": 481926768,
+      "step": 2788
+    },
+    {
+      "epoch": 1.0677901187284564,
+      "loss": 2.5003244876861572,
+      "loss_ce": 0.6895257234573364,
+      "loss_iou": 0.8330117464065552,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 1.8107988834381104,
+      "num_input_tokens_seen": 481926768,
+      "step": 2788
+    },
+    {
+      "epoch": 1.0681731137495212,
+      "grad_norm": 289.3937243822588,
+      "learning_rate": 5e-06,
+      "loss": 2.7773,
+      "num_input_tokens_seen": 482099640,
+      "step": 2789
+    },
+    {
+      "epoch": 1.0681731137495212,
+      "loss": 2.9110474586486816,
+      "loss_ce": 0.6996558308601379,
+      "loss_iou": 1.0353833436965942,
+      "loss_num": 0.028076171875,
+      "loss_xval": 2.2113916873931885,
+      "num_input_tokens_seen": 482099640,
+      "step": 2789
+    },
+    {
+      "epoch": 1.068556108770586,
+      "grad_norm": 87.94871422248494,
+      "learning_rate": 5e-06,
+      "loss": 3.0462,
+      "num_input_tokens_seen": 482272576,
+      "step": 2790
+    },
+    {
+      "epoch": 1.068556108770586,
+      "loss": 3.0079867839813232,
+      "loss_ce": 0.7180944681167603,
+      "loss_iou": 0.9775877594947815,
+      "loss_num": 0.06689453125,
+      "loss_xval": 2.2898924350738525,
+      "num_input_tokens_seen": 482272576,
+      "step": 2790
+    },
+    {
+      "epoch": 1.0689391037916507,
+      "grad_norm": 67.37188580188071,
+      "learning_rate": 5e-06,
+      "loss": 2.6366,
+      "num_input_tokens_seen": 482445272,
+      "step": 2791
+    },
+    {
+      "epoch": 1.0689391037916507,
+      "loss": 2.634394645690918,
+      "loss_ce": 0.6736334562301636,
+      "loss_iou": 0.865878701210022,
+      "loss_num": 0.0458984375,
+      "loss_xval": 1.960761308670044,
+      "num_input_tokens_seen": 482445272,
+      "step": 2791
+    },
+    {
+      "epoch": 1.0693220988127154,
+      "grad_norm": 67.52657384742601,
+      "learning_rate": 5e-06,
+      "loss": 2.4575,
+      "num_input_tokens_seen": 482618344,
+      "step": 2792
+    },
+    {
+      "epoch": 1.0693220988127154,
+      "loss": 2.4036059379577637,
+      "loss_ce": 0.7781287431716919,
+      "loss_iou": 0.7347356081008911,
+      "loss_num": 0.03125,
+      "loss_xval": 1.6254770755767822,
+      "num_input_tokens_seen": 482618344,
+      "step": 2792
+    },
+    {
+      "epoch": 1.0697050938337802,
+      "grad_norm": 87.27920424026138,
+      "learning_rate": 5e-06,
+      "loss": 2.3177,
+      "num_input_tokens_seen": 482791072,
+      "step": 2793
+    },
+    {
+      "epoch": 1.0697050938337802,
+      "loss": 2.2667553424835205,
+      "loss_ce": 0.7141283750534058,
+      "loss_iou": 0.7067334651947021,
+      "loss_num": 0.02783203125,
+      "loss_xval": 1.5526270866394043,
+      "num_input_tokens_seen": 482791072,
+      "step": 2793
+    },
+    {
+      "epoch": 1.070088088854845,
+      "grad_norm": 192.24234268939296,
+      "learning_rate": 5e-06,
+      "loss": 3.2865,
+      "num_input_tokens_seen": 482964096,
+      "step": 2794
+    },
+    {
+      "epoch": 1.070088088854845,
+      "loss": 3.2384159564971924,
+      "loss_ce": 0.6809695959091187,
+      "loss_iou": 1.1539673805236816,
+      "loss_num": 0.0498046875,
+      "loss_xval": 2.5574464797973633,
+      "num_input_tokens_seen": 482964096,
+      "step": 2794
+    },
+    {
+      "epoch": 1.0704710838759097,
+      "grad_norm": 154.55315774871912,
+      "learning_rate": 5e-06,
+      "loss": 2.5979,
+      "num_input_tokens_seen": 483137496,
+      "step": 2795
+    },
+    {
+      "epoch": 1.0704710838759097,
+      "loss": 2.694046974182129,
+      "loss_ce": 0.7154624462127686,
+      "loss_iou": 0.898166835308075,
+      "loss_num": 0.036376953125,
+      "loss_xval": 1.97858464717865,
+      "num_input_tokens_seen": 483137496,
+      "step": 2795
+    },
+    {
+      "epoch": 1.0708540788969743,
+      "grad_norm": 241.487390491254,
+      "learning_rate": 5e-06,
+      "loss": 2.8126,
+      "num_input_tokens_seen": 483310328,
+      "step": 2796
+    },
+    {
+      "epoch": 1.0708540788969743,
+      "loss": 2.9573071002960205,
+      "loss_ce": 0.7261373996734619,
+      "loss_iou": 1.0228114128112793,
+      "loss_num": 0.037109375,
+      "loss_xval": 2.2311697006225586,
+      "num_input_tokens_seen": 483310328,
+      "step": 2796
+    },
+    {
+      "epoch": 1.071237073918039,
+      "grad_norm": 181.30192210869185,
+      "learning_rate": 5e-06,
+      "loss": 2.6225,
+      "num_input_tokens_seen": 483483296,
+      "step": 2797
+    },
+    {
+      "epoch": 1.071237073918039,
+      "loss": 2.587357521057129,
+      "loss_ce": 0.7516204118728638,
+      "loss_iou": 0.8230810761451721,
+      "loss_num": 0.037841796875,
+      "loss_xval": 1.8357373476028442,
+      "num_input_tokens_seen": 483483296,
+      "step": 2797
+    },
+    {
+      "epoch": 1.0716200689391038,
+      "grad_norm": 51.07158078303042,
+      "learning_rate": 5e-06,
+      "loss": 2.3856,
+      "num_input_tokens_seen": 483656160,
+      "step": 2798
+    },
+    {
+      "epoch": 1.0716200689391038,
+      "loss": 2.1889824867248535,
+      "loss_ce": 0.636174201965332,
+      "loss_iou": 0.7104403376579285,
+      "loss_num": 0.0263671875,
+      "loss_xval": 1.552808165550232,
+      "num_input_tokens_seen": 483656160,
+      "step": 2798
+    },
+    {
+      "epoch": 1.0720030639601685,
+      "grad_norm": 47.23474939291006,
+      "learning_rate": 5e-06,
+      "loss": 2.1921,
+      "num_input_tokens_seen": 483828888,
+      "step": 2799
+    },
+    {
+      "epoch": 1.0720030639601685,
+      "loss": 2.226820230484009,
+      "loss_ce": 0.6776589751243591,
+      "loss_iou": 0.715834379196167,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 1.549161434173584,
+      "num_input_tokens_seen": 483828888,
+      "step": 2799
+    },
+    {
+      "epoch": 1.0723860589812333,
+      "grad_norm": 53.960842263662435,
+      "learning_rate": 5e-06,
+      "loss": 2.1868,
+      "num_input_tokens_seen": 484002144,
+      "step": 2800
+    },
+    {
+      "epoch": 1.0723860589812333,
+      "loss": 2.277585506439209,
+      "loss_ce": 0.7320078611373901,
+      "loss_iou": 0.7191082835197449,
+      "loss_num": 0.021484375,
+      "loss_xval": 1.5455774068832397,
+      "num_input_tokens_seen": 484002144,
+      "step": 2800
+    },
+    {
+      "epoch": 1.072769054002298,
+      "grad_norm": 263.27582944108934,
+      "learning_rate": 5e-06,
+      "loss": 2.1796,
+      "num_input_tokens_seen": 484174920,
+      "step": 2801
+    },
+    {
+      "epoch": 1.072769054002298,
+      "loss": 2.3287782669067383,
+      "loss_ce": 0.6678345203399658,
+      "loss_iou": 0.7781646847724915,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 1.660943627357483,
+      "num_input_tokens_seen": 484174920,
+      "step": 2801
+    },
+    {
+      "epoch": 1.0731520490233626,
+      "grad_norm": 117.56310256583214,
+      "learning_rate": 5e-06,
+      "loss": 3.2097,
+      "num_input_tokens_seen": 484348112,
+      "step": 2802
+    },
+    {
+      "epoch": 1.0731520490233626,
+      "loss": 3.3388285636901855,
+      "loss_ce": 0.6726955771446228,
+      "loss_iou": 1.181455135345459,
+      "loss_num": 0.060546875,
+      "loss_xval": 2.666132926940918,
+      "num_input_tokens_seen": 484348112,
+      "step": 2802
+    },
+    {
+      "epoch": 1.0735350440444273,
+      "grad_norm": 104.58298648616196,
+      "learning_rate": 5e-06,
+      "loss": 2.5987,
+      "num_input_tokens_seen": 484521336,
+      "step": 2803
+    },
+    {
+      "epoch": 1.0735350440444273,
+      "loss": 2.521456003189087,
+      "loss_ce": 0.7201911211013794,
+      "loss_iou": 0.7994971871376038,
+      "loss_num": 0.04052734375,
+      "loss_xval": 1.8012648820877075,
+      "num_input_tokens_seen": 484521336,
+      "step": 2803
+    },
+    {
+      "epoch": 1.073918039065492,
+      "grad_norm": 96.78154982441428,
+      "learning_rate": 5e-06,
+      "loss": 2.4816,
+      "num_input_tokens_seen": 484694392,
+      "step": 2804
+    },
+    {
+      "epoch": 1.073918039065492,
+      "loss": 2.4801526069641113,
+      "loss_ce": 0.7741658091545105,
+      "loss_iou": 0.778072714805603,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 1.705986738204956,
+      "num_input_tokens_seen": 484694392,
+      "step": 2804
+    },
+    {
+      "epoch": 1.0743010340865569,
+      "grad_norm": 279.11723943979456,
+      "learning_rate": 5e-06,
+      "loss": 2.6919,
+      "num_input_tokens_seen": 484867720,
+      "step": 2805
+    },
+    {
+      "epoch": 1.0743010340865569,
+      "loss": 2.7241034507751465,
+      "loss_ce": 0.7051548361778259,
+      "loss_iou": 0.9290604591369629,
+      "loss_num": 0.0322265625,
+      "loss_xval": 2.018948554992676,
+      "num_input_tokens_seen": 484867720,
+      "step": 2805
+    },
+    {
+      "epoch": 1.0746840291076216,
+      "grad_norm": 91.67594166503147,
+      "learning_rate": 5e-06,
+      "loss": 3.1007,
+      "num_input_tokens_seen": 485040416,
+      "step": 2806
+    },
+    {
+      "epoch": 1.0746840291076216,
+      "loss": 3.09213924407959,
+      "loss_ce": 0.6244996190071106,
+      "loss_iou": 1.0971009731292725,
+      "loss_num": 0.0546875,
+      "loss_xval": 2.467639446258545,
+      "num_input_tokens_seen": 485040416,
+      "step": 2806
+    },
+    {
+      "epoch": 1.0750670241286864,
+      "grad_norm": 85.68762205479932,
+      "learning_rate": 5e-06,
+      "loss": 2.4442,
+      "num_input_tokens_seen": 485212800,
+      "step": 2807
+    },
+    {
+      "epoch": 1.0750670241286864,
+      "loss": 2.4304544925689697,
+      "loss_ce": 0.6337364912033081,
+      "loss_iou": 0.7999093532562256,
+      "loss_num": 0.039306640625,
+      "loss_xval": 1.7967181205749512,
+      "num_input_tokens_seen": 485212800,
+      "step": 2807
+    },
+    {
+      "epoch": 1.0754500191497511,
+      "grad_norm": 77.01171669214705,
+      "learning_rate": 5e-06,
+      "loss": 2.4726,
+      "num_input_tokens_seen": 485385712,
+      "step": 2808
+    },
+    {
+      "epoch": 1.0754500191497511,
+      "loss": 2.4367589950561523,
+      "loss_ce": 0.7392523288726807,
+      "loss_iou": 0.7764266133308411,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 1.6975065469741821,
+      "num_input_tokens_seen": 485385712,
+      "step": 2808
+    },
+    {
+      "epoch": 1.075833014170816,
+      "grad_norm": 94.84736549098368,
+      "learning_rate": 5e-06,
+      "loss": 2.5519,
+      "num_input_tokens_seen": 485558496,
+      "step": 2809
+    },
+    {
+      "epoch": 1.075833014170816,
+      "loss": 2.4850964546203613,
+      "loss_ce": 0.6541642546653748,
+      "loss_iou": 0.8250730633735657,
+      "loss_num": 0.0361328125,
+      "loss_xval": 1.8309322595596313,
+      "num_input_tokens_seen": 485558496,
+      "step": 2809
+    },
+    {
+      "epoch": 1.0762160091918804,
+      "grad_norm": 150.29738868003025,
+      "learning_rate": 5e-06,
+      "loss": 2.2246,
+      "num_input_tokens_seen": 485726712,
+      "step": 2810
+    },
+    {
+      "epoch": 1.0762160091918804,
+      "loss": 2.1678030490875244,
+      "loss_ce": 0.7016191482543945,
+      "loss_iou": 0.6786486506462097,
+      "loss_num": 0.021728515625,
+      "loss_xval": 1.4661840200424194,
+      "num_input_tokens_seen": 485726712,
+      "step": 2810
+    },
+    {
+      "epoch": 1.0765990042129452,
+      "grad_norm": 177.0528897847532,
+      "learning_rate": 5e-06,
+      "loss": 2.7007,
+      "num_input_tokens_seen": 485899624,
+      "step": 2811
+    },
+    {
+      "epoch": 1.0765990042129452,
+      "loss": 2.6600353717803955,
+      "loss_ce": 0.639045774936676,
+      "loss_iou": 0.9138761162757874,
+      "loss_num": 0.03857421875,
+      "loss_xval": 2.020989418029785,
+      "num_input_tokens_seen": 485899624,
+      "step": 2811
+    },
+    {
+      "epoch": 1.07698199923401,
+      "grad_norm": 102.07468339681954,
+      "learning_rate": 5e-06,
+      "loss": 2.2049,
+      "num_input_tokens_seen": 486072344,
+      "step": 2812
+    },
+    {
+      "epoch": 1.07698199923401,
+      "loss": 2.532315731048584,
+      "loss_ce": 0.6821858286857605,
+      "loss_iou": 0.8397378325462341,
+      "loss_num": 0.0341796875,
+      "loss_xval": 1.8501299619674683,
+      "num_input_tokens_seen": 486072344,
+      "step": 2812
+    },
+    {
+      "epoch": 1.0773649942550747,
+      "grad_norm": 131.57059991411325,
+      "learning_rate": 5e-06,
+      "loss": 2.3435,
+      "num_input_tokens_seen": 486245128,
+      "step": 2813
+    },
+    {
+      "epoch": 1.0773649942550747,
+      "loss": 2.2519588470458984,
+      "loss_ce": 0.7076823711395264,
+      "loss_iou": 0.714551568031311,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 1.544276475906372,
+      "num_input_tokens_seen": 486245128,
+      "step": 2813
+    },
+    {
+      "epoch": 1.0777479892761395,
+      "grad_norm": 278.5990370028417,
+      "learning_rate": 5e-06,
+      "loss": 2.7089,
+      "num_input_tokens_seen": 486417960,
+      "step": 2814
+    },
+    {
+      "epoch": 1.0777479892761395,
+      "loss": 2.645965576171875,
+      "loss_ce": 0.6513679027557373,
+      "loss_iou": 0.9244838953018188,
+      "loss_num": 0.029052734375,
+      "loss_xval": 1.9945976734161377,
+      "num_input_tokens_seen": 486417960,
+      "step": 2814
+    },
+    {
+      "epoch": 1.0781309842972042,
+      "grad_norm": 54.95344309622083,
+      "learning_rate": 5e-06,
+      "loss": 2.3731,
+      "num_input_tokens_seen": 486590672,
+      "step": 2815
+    },
+    {
+      "epoch": 1.0781309842972042,
+      "loss": 2.100264549255371,
+      "loss_ce": 0.6261555552482605,
+      "loss_iou": 0.6392761468887329,
+      "loss_num": 0.0390625,
+      "loss_xval": 1.4356567859649658,
+      "num_input_tokens_seen": 486590672,
+      "step": 2815
+    },
+    {
+      "epoch": 1.0785139793182688,
+      "grad_norm": 107.60883534856733,
+      "learning_rate": 5e-06,
+      "loss": 2.4516,
+      "num_input_tokens_seen": 486763376,
+      "step": 2816
+    },
+    {
+      "epoch": 1.0785139793182688,
+      "loss": 2.4624667167663574,
+      "loss_ce": 0.7012736797332764,
+      "loss_iou": 0.7974972128868103,
+      "loss_num": 0.033203125,
+      "loss_xval": 1.7611931562423706,
+      "num_input_tokens_seen": 486763376,
+      "step": 2816
+    },
+    {
+      "epoch": 1.0788969743393335,
+      "grad_norm": 120.23417852325592,
+      "learning_rate": 5e-06,
+      "loss": 2.1123,
+      "num_input_tokens_seen": 486932832,
+      "step": 2817
+    },
+    {
+      "epoch": 1.0788969743393335,
+      "loss": 2.0849769115448,
+      "loss_ce": 0.6629939675331116,
+      "loss_iou": 6.256191116505706e+25,
+      "loss_num": 0.0223388671875,
+      "loss_xval": -9.735556609752802e+33,
+      "num_input_tokens_seen": 486932832,
+      "step": 2817
+    },
+    {
+      "epoch": 1.0792799693603983,
+      "grad_norm": 116.6227009118233,
+      "learning_rate": 5e-06,
+      "loss": 2.4347,
+      "num_input_tokens_seen": 487105624,
+      "step": 2818
+    },
+    {
+      "epoch": 1.0792799693603983,
+      "loss": 2.4521543979644775,
+      "loss_ce": 0.6253559589385986,
+      "loss_iou": 0.8615193367004395,
+      "loss_num": 0.020751953125,
+      "loss_xval": 1.826798439025879,
+      "num_input_tokens_seen": 487105624,
+      "step": 2818
+    },
+    {
+      "epoch": 1.079662964381463,
+      "grad_norm": 149.29655584940656,
+      "learning_rate": 5e-06,
+      "loss": 2.1704,
+      "num_input_tokens_seen": 487278464,
+      "step": 2819
+    },
+    {
+      "epoch": 1.079662964381463,
+      "loss": 2.266498565673828,
+      "loss_ce": 0.6545871496200562,
+      "loss_iou": 0.7549304366111755,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 1.611911654472351,
+      "num_input_tokens_seen": 487278464,
+      "step": 2819
+    },
+    {
+      "epoch": 1.0800459594025278,
+      "grad_norm": 148.14156207961628,
+      "learning_rate": 5e-06,
+      "loss": 2.5219,
+      "num_input_tokens_seen": 487451344,
+      "step": 2820
+    },
+    {
+      "epoch": 1.0800459594025278,
+      "loss": 2.676058769226074,
+      "loss_ce": 0.6396024227142334,
+      "loss_iou": 0.925424337387085,
+      "loss_num": 0.037109375,
+      "loss_xval": 2.03645658493042,
+      "num_input_tokens_seen": 487451344,
+      "step": 2820
+    },
+    {
+      "epoch": 1.0804289544235925,
+      "grad_norm": 103.59302016767316,
+      "learning_rate": 5e-06,
+      "loss": 2.2078,
+      "num_input_tokens_seen": 487624448,
+      "step": 2821
+    },
+    {
+      "epoch": 1.0804289544235925,
+      "loss": 2.300532341003418,
+      "loss_ce": 0.649788498878479,
+      "loss_iou": 0.7663967609405518,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 1.6507439613342285,
+      "num_input_tokens_seen": 487624448,
+      "step": 2821
+    },
+    {
+      "epoch": 1.0808119494446573,
+      "grad_norm": 121.75501520149543,
+      "learning_rate": 5e-06,
+      "loss": 2.4232,
+      "num_input_tokens_seen": 487797296,
+      "step": 2822
+    },
+    {
+      "epoch": 1.0808119494446573,
+      "loss": 2.333770275115967,
+      "loss_ce": 0.6665495038032532,
+      "loss_iou": 0.7725751996040344,
+      "loss_num": 0.0244140625,
+      "loss_xval": 1.6672207117080688,
+      "num_input_tokens_seen": 487797296,
+      "step": 2822
+    },
+    {
+      "epoch": 1.081194944465722,
+      "grad_norm": 167.1612778930809,
+      "learning_rate": 5e-06,
+      "loss": 2.1984,
+      "num_input_tokens_seen": 487970016,
+      "step": 2823
+    },
+    {
+      "epoch": 1.081194944465722,
+      "loss": 2.09686279296875,
+      "loss_ce": 0.6161590814590454,
+      "loss_iou": 0.6905776262283325,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 1.480703592300415,
+      "num_input_tokens_seen": 487970016,
+      "step": 2823
+    },
+    {
+      "epoch": 1.0815779394867866,
+      "grad_norm": 92.4429662904512,
+      "learning_rate": 5e-06,
+      "loss": 2.616,
+      "num_input_tokens_seen": 488143296,
+      "step": 2824
+    },
+    {
+      "epoch": 1.0815779394867866,
+      "loss": 2.497222900390625,
+      "loss_ce": 0.669642984867096,
+      "loss_iou": 0.8369162082672119,
+      "loss_num": 0.03076171875,
+      "loss_xval": 1.8275799751281738,
+      "num_input_tokens_seen": 488143296,
+      "step": 2824
+    },
+    {
+      "epoch": 1.0819609345078514,
+      "grad_norm": 193.15992522059403,
+      "learning_rate": 5e-06,
+      "loss": 2.1213,
+      "num_input_tokens_seen": 488316384,
+      "step": 2825
+    },
+    {
+      "epoch": 1.0819609345078514,
+      "loss": 2.1571407318115234,
+      "loss_ce": 0.6244411468505859,
+      "loss_iou": 0.7029953002929688,
+      "loss_num": 0.025390625,
+      "loss_xval": 1.5326995849609375,
+      "num_input_tokens_seen": 488316384,
+      "step": 2825
+    },
+    {
+      "epoch": 1.0823439295289161,
+      "grad_norm": 313.7898566610402,
+      "learning_rate": 5e-06,
+      "loss": 2.3996,
+      "num_input_tokens_seen": 488489136,
+      "step": 2826
+    },
+    {
+      "epoch": 1.0823439295289161,
+      "loss": 2.307638645172119,
+      "loss_ce": 0.6069540977478027,
+      "loss_iou": 0.7825322151184082,
+      "loss_num": 0.027099609375,
+      "loss_xval": 1.7006845474243164,
+      "num_input_tokens_seen": 488489136,
+      "step": 2826
+    },
+    {
+      "epoch": 1.0827269245499809,
+      "grad_norm": 99.87707022706083,
+      "learning_rate": 5e-06,
+      "loss": 2.5446,
+      "num_input_tokens_seen": 488662056,
+      "step": 2827
+    },
+    {
+      "epoch": 1.0827269245499809,
+      "loss": 2.5995631217956543,
+      "loss_ce": 0.6866831183433533,
+      "loss_iou": 0.8688545227050781,
+      "loss_num": 0.034912109375,
+      "loss_xval": 1.9128799438476562,
+      "num_input_tokens_seen": 488662056,
+      "step": 2827
+    },
+    {
+      "epoch": 1.0831099195710456,
+      "grad_norm": 372.36293237635937,
+      "learning_rate": 5e-06,
+      "loss": 2.5754,
+      "num_input_tokens_seen": 488835248,
+      "step": 2828
+    },
+    {
+      "epoch": 1.0831099195710456,
+      "loss": 2.4599075317382812,
+      "loss_ce": 0.6324270963668823,
+      "loss_iou": 0.8389416337013245,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 1.827480435371399,
+      "num_input_tokens_seen": 488835248,
+      "step": 2828
+    },
+    {
+      "epoch": 1.0834929145921104,
+      "grad_norm": 123.33707719794329,
+      "learning_rate": 5e-06,
+      "loss": 2.2394,
+      "num_input_tokens_seen": 489008408,
+      "step": 2829
+    },
+    {
+      "epoch": 1.0834929145921104,
+      "loss": 2.0730512142181396,
+      "loss_ce": 0.6280667781829834,
+      "loss_iou": 0.6656379699707031,
+      "loss_num": 0.022705078125,
+      "loss_xval": 1.4449844360351562,
+      "num_input_tokens_seen": 489008408,
+      "step": 2829
+    },
+    {
+      "epoch": 1.083875909613175,
+      "grad_norm": 313.0034887809654,
+      "learning_rate": 5e-06,
+      "loss": 2.7079,
+      "num_input_tokens_seen": 489181368,
+      "step": 2830
+    },
+    {
+      "epoch": 1.083875909613175,
+      "loss": 2.8506133556365967,
+      "loss_ce": 0.6318695545196533,
+      "loss_iou": 1.0071990489959717,
+      "loss_num": 0.040771484375,
+      "loss_xval": 2.2187438011169434,
+      "num_input_tokens_seen": 489181368,
+      "step": 2830
+    },
+    {
+      "epoch": 1.0842589046342397,
+      "grad_norm": 168.8747494529447,
+      "learning_rate": 5e-06,
+      "loss": 2.7962,
+      "num_input_tokens_seen": 489354272,
+      "step": 2831
+    },
+    {
+      "epoch": 1.0842589046342397,
+      "loss": 2.8026838302612305,
+      "loss_ce": 0.6917127370834351,
+      "loss_iou": 0.9522751569747925,
+      "loss_num": 0.041259765625,
+      "loss_xval": 2.110971450805664,
+      "num_input_tokens_seen": 489354272,
+      "step": 2831
+    },
+    {
+      "epoch": 1.0846418996553044,
+      "grad_norm": 57.168527638211046,
+      "learning_rate": 5e-06,
+      "loss": 2.6492,
+      "num_input_tokens_seen": 489527768,
+      "step": 2832
+    },
+    {
+      "epoch": 1.0846418996553044,
+      "loss": 2.810802698135376,
+      "loss_ce": 0.68154376745224,
+      "loss_iou": 0.9660577774047852,
+      "loss_num": 0.039306640625,
+      "loss_xval": 2.1292591094970703,
+      "num_input_tokens_seen": 489527768,
+      "step": 2832
+    },
+    {
+      "epoch": 1.0850248946763692,
+      "grad_norm": 53.892156776320036,
+      "learning_rate": 5e-06,
+      "loss": 2.32,
+      "num_input_tokens_seen": 489700904,
+      "step": 2833
+    },
+    {
+      "epoch": 1.0850248946763692,
+      "loss": 2.2683029174804688,
+      "loss_ce": 0.6636078953742981,
+      "loss_iou": 0.7453100681304932,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 1.6046948432922363,
+      "num_input_tokens_seen": 489700904,
+      "step": 2833
+    },
+    {
+      "epoch": 1.085407889697434,
+      "grad_norm": 85.90525825970508,
+      "learning_rate": 5e-06,
+      "loss": 2.0927,
+      "num_input_tokens_seen": 489873864,
+      "step": 2834
+    },
+    {
+      "epoch": 1.085407889697434,
+      "loss": 2.0061228275299072,
+      "loss_ce": 0.603214681148529,
+      "loss_iou": -2.0070057552195992e+22,
+      "loss_num": 0.027099609375,
+      "loss_xval": 6.770755103529415e+36,
+      "num_input_tokens_seen": 489873864,
+      "step": 2834
+    },
+    {
+      "epoch": 1.0857908847184987,
+      "grad_norm": 171.33366141709405,
+      "learning_rate": 5e-06,
+      "loss": 2.0763,
+      "num_input_tokens_seen": 490046992,
+      "step": 2835
+    },
+    {
+      "epoch": 1.0857908847184987,
+      "loss": 2.1125564575195312,
+      "loss_ce": 0.6527152061462402,
+      "loss_iou": 0.680894136428833,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 1.459841251373291,
+      "num_input_tokens_seen": 490046992,
+      "step": 2835
+    },
+    {
+      "epoch": 1.0861738797395635,
+      "grad_norm": 160.93263017016588,
+      "learning_rate": 5e-06,
+      "loss": 2.6295,
+      "num_input_tokens_seen": 490220232,
+      "step": 2836
+    },
+    {
+      "epoch": 1.0861738797395635,
+      "loss": 2.500584602355957,
+      "loss_ce": 0.6741725206375122,
+      "loss_iou": 0.8312051296234131,
+      "loss_num": 0.03271484375,
+      "loss_xval": 1.8264117240905762,
+      "num_input_tokens_seen": 490220232,
+      "step": 2836
+    },
+    {
+      "epoch": 1.086556874760628,
+      "grad_norm": 98.89169004201597,
+      "learning_rate": 5e-06,
+      "loss": 2.0596,
+      "num_input_tokens_seen": 490393368,
+      "step": 2837
+    },
+    {
+      "epoch": 1.086556874760628,
+      "loss": 2.169426918029785,
+      "loss_ce": 0.6463626027107239,
+      "loss_iou": 0.7101861834526062,
+      "loss_num": 0.0205078125,
+      "loss_xval": 1.5230640172958374,
+      "num_input_tokens_seen": 490393368,
+      "step": 2837
+    },
+    {
+      "epoch": 1.0869398697816928,
+      "grad_norm": 171.29138796187056,
+      "learning_rate": 5e-06,
+      "loss": 2.0366,
+      "num_input_tokens_seen": 490565872,
+      "step": 2838
+    },
+    {
+      "epoch": 1.0869398697816928,
+      "loss": 1.9540889263153076,
+      "loss_ce": 0.5884835720062256,
+      "loss_iou": 0.6313195824623108,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 1.3656054735183716,
+      "num_input_tokens_seen": 490565872,
+      "step": 2838
+    },
+    {
+      "epoch": 1.0873228648027575,
+      "grad_norm": 294.0112567125096,
+      "learning_rate": 5e-06,
+      "loss": 2.7592,
+      "num_input_tokens_seen": 490739336,
+      "step": 2839
+    },
+    {
+      "epoch": 1.0873228648027575,
+      "loss": 2.7852325439453125,
+      "loss_ce": 0.6034911274909973,
+      "loss_iou": 1.039387583732605,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 2.18174147605896,
+      "num_input_tokens_seen": 490739336,
+      "step": 2839
+    },
+    {
+      "epoch": 1.0877058598238223,
+      "grad_norm": 138.16441247610052,
+      "learning_rate": 5e-06,
+      "loss": 2.692,
+      "num_input_tokens_seen": 490912320,
+      "step": 2840
+    },
+    {
+      "epoch": 1.0877058598238223,
+      "loss": 2.710242748260498,
+      "loss_ce": 0.6224827766418457,
+      "loss_iou": 0.9509843587875366,
+      "loss_num": 0.037109375,
+      "loss_xval": 2.0877597332000732,
+      "num_input_tokens_seen": 490912320,
+      "step": 2840
+    },
+    {
+      "epoch": 1.088088854844887,
+      "grad_norm": 82.79934485866873,
+      "learning_rate": 5e-06,
+      "loss": 2.2325,
+      "num_input_tokens_seen": 491085120,
+      "step": 2841
+    },
+    {
+      "epoch": 1.088088854844887,
+      "loss": 2.093496799468994,
+      "loss_ce": 0.6058523654937744,
+      "loss_iou": 0.6878225207328796,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 1.4876445531845093,
+      "num_input_tokens_seen": 491085120,
+      "step": 2841
+    },
+    {
+      "epoch": 1.0884718498659518,
+      "grad_norm": 112.61014009674933,
+      "learning_rate": 5e-06,
+      "loss": 2.3055,
+      "num_input_tokens_seen": 491258304,
+      "step": 2842
+    },
+    {
+      "epoch": 1.0884718498659518,
+      "loss": 2.3159525394439697,
+      "loss_ce": 0.6486879587173462,
+      "loss_iou": 0.7727192640304565,
+      "loss_num": 0.0244140625,
+      "loss_xval": 1.667264699935913,
+      "num_input_tokens_seen": 491258304,
+      "step": 2842
+    },
+    {
+      "epoch": 1.0888548448870166,
+      "grad_norm": 157.64913384604296,
+      "learning_rate": 5e-06,
+      "loss": 2.2396,
+      "num_input_tokens_seen": 491431200,
+      "step": 2843
+    },
+    {
+      "epoch": 1.0888548448870166,
+      "loss": 2.414534568786621,
+      "loss_ce": 0.630474328994751,
+      "loss_iou": 0.8160413503646851,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 1.7840602397918701,
+      "num_input_tokens_seen": 491431200,
+      "step": 2843
+    },
+    {
+      "epoch": 1.089237839908081,
+      "grad_norm": 128.54688564774014,
+      "learning_rate": 5e-06,
+      "loss": 2.1839,
+      "num_input_tokens_seen": 491604360,
+      "step": 2844
+    },
+    {
+      "epoch": 1.089237839908081,
+      "loss": 2.1136837005615234,
+      "loss_ce": 0.6559200882911682,
+      "loss_iou": 0.6847227811813354,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 1.457763433456421,
+      "num_input_tokens_seen": 491604360,
+      "step": 2844
+    },
+    {
+      "epoch": 1.0896208349291459,
+      "grad_norm": 133.58048026425544,
+      "learning_rate": 5e-06,
+      "loss": 2.3242,
+      "num_input_tokens_seen": 491777240,
+      "step": 2845
+    },
+    {
+      "epoch": 1.0896208349291459,
+      "loss": 2.1787991523742676,
+      "loss_ce": 0.5843753814697266,
+      "loss_iou": 0.7383129596710205,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 1.594423770904541,
+      "num_input_tokens_seen": 491777240,
+      "step": 2845
+    },
+    {
+      "epoch": 1.0900038299502106,
+      "grad_norm": 122.44268752260001,
+      "learning_rate": 5e-06,
+      "loss": 2.3311,
+      "num_input_tokens_seen": 491950152,
+      "step": 2846
+    },
+    {
+      "epoch": 1.0900038299502106,
+      "loss": 2.378659725189209,
+      "loss_ce": 0.6128594279289246,
+      "loss_iou": 0.823634922504425,
+      "loss_num": 0.023681640625,
+      "loss_xval": 1.76580011844635,
+      "num_input_tokens_seen": 491950152,
+      "step": 2846
+    },
+    {
+      "epoch": 1.0903868249712754,
+      "grad_norm": 207.86810599802814,
+      "learning_rate": 5e-06,
+      "loss": 2.4016,
+      "num_input_tokens_seen": 492123440,
+      "step": 2847
+    },
+    {
+      "epoch": 1.0903868249712754,
+      "loss": 2.386385917663574,
+      "loss_ce": 0.5922180414199829,
+      "loss_iou": 0.8199966549873352,
+      "loss_num": 0.03076171875,
+      "loss_xval": 1.7941681146621704,
+      "num_input_tokens_seen": 492123440,
+      "step": 2847
+    },
+    {
+      "epoch": 1.0907698199923401,
+      "grad_norm": 289.7377745310439,
+      "learning_rate": 5e-06,
+      "loss": 2.5178,
+      "num_input_tokens_seen": 492296328,
+      "step": 2848
+    },
+    {
+      "epoch": 1.0907698199923401,
+      "loss": 2.3781158924102783,
+      "loss_ce": 0.6338722109794617,
+      "loss_iou": 0.8073635101318359,
+      "loss_num": 0.02587890625,
+      "loss_xval": 1.7442436218261719,
+      "num_input_tokens_seen": 492296328,
+      "step": 2848
+    },
+    {
+      "epoch": 1.0911528150134049,
+      "grad_norm": 58.10374609007535,
+      "learning_rate": 5e-06,
+      "loss": 2.5232,
+      "num_input_tokens_seen": 492469600,
+      "step": 2849
+    },
+    {
+      "epoch": 1.0911528150134049,
+      "loss": 2.472416400909424,
+      "loss_ce": 0.6209355592727661,
+      "loss_iou": 0.8428851366043091,
+      "loss_num": 0.033203125,
+      "loss_xval": 1.8514807224273682,
+      "num_input_tokens_seen": 492469600,
+      "step": 2849
+    },
+    {
+      "epoch": 1.0915358100344696,
+      "grad_norm": 174.9568940032917,
+      "learning_rate": 5e-06,
+      "loss": 2.2801,
+      "num_input_tokens_seen": 492642400,
+      "step": 2850
+    },
+    {
+      "epoch": 1.0915358100344696,
+      "loss": 2.270138740539551,
+      "loss_ce": 0.6334530711174011,
+      "loss_iou": 0.752272367477417,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 1.636685848236084,
+      "num_input_tokens_seen": 492642400,
+      "step": 2850
+    },
+    {
+      "epoch": 1.0919188050555342,
+      "grad_norm": 126.78242624920647,
+      "learning_rate": 5e-06,
+      "loss": 2.1337,
+      "num_input_tokens_seen": 492815128,
+      "step": 2851
+    },
+    {
+      "epoch": 1.0919188050555342,
+      "loss": 2.2764225006103516,
+      "loss_ce": 0.6189845204353333,
+      "loss_iou": 0.7655476331710815,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 1.657438039779663,
+      "num_input_tokens_seen": 492815128,
+      "step": 2851
+    },
+    {
+      "epoch": 1.092301800076599,
+      "grad_norm": 261.650315207769,
+      "learning_rate": 5e-06,
+      "loss": 2.4944,
+      "num_input_tokens_seen": 492987896,
+      "step": 2852
+    },
+    {
+      "epoch": 1.092301800076599,
+      "loss": 2.5657906532287598,
+      "loss_ce": 0.6617599725723267,
+      "loss_iou": 0.8752331733703613,
+      "loss_num": 0.03076171875,
+      "loss_xval": 1.9040307998657227,
+      "num_input_tokens_seen": 492987896,
+      "step": 2852
+    },
+    {
+      "epoch": 1.0926847950976637,
+      "grad_norm": 118.42784116999931,
+      "learning_rate": 5e-06,
+      "loss": 2.1608,
+      "num_input_tokens_seen": 493160384,
+      "step": 2853
+    },
+    {
+      "epoch": 1.0926847950976637,
+      "loss": 1.999885082244873,
+      "loss_ce": 0.5869404077529907,
+      "loss_iou": 0.6425074338912964,
+      "loss_num": 0.025634765625,
+      "loss_xval": 1.4001271724700928,
+      "num_input_tokens_seen": 493160384,
+      "step": 2853
+    },
+    {
+      "epoch": 1.0930677901187285,
+      "grad_norm": 150.2787262374642,
+      "learning_rate": 5e-06,
+      "loss": 2.3293,
+      "num_input_tokens_seen": 493333056,
+      "step": 2854
+    },
+    {
+      "epoch": 1.0930677901187285,
+      "loss": 2.2261369228363037,
+      "loss_ce": 0.5819826126098633,
+      "loss_iou": 0.7748359441757202,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 1.6441543102264404,
+      "num_input_tokens_seen": 493333056,
+      "step": 2854
+    },
+    {
+      "epoch": 1.0934507851397932,
+      "grad_norm": 336.4674460277969,
+      "learning_rate": 5e-06,
+      "loss": 2.317,
+      "num_input_tokens_seen": 493506240,
+      "step": 2855
+    },
+    {
+      "epoch": 1.0934507851397932,
+      "loss": 2.2473292350769043,
+      "loss_ce": 0.6243661642074585,
+      "loss_iou": 0.7441293001174927,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 1.6229631900787354,
+      "num_input_tokens_seen": 493506240,
+      "step": 2855
+    },
+    {
+      "epoch": 1.093833780160858,
+      "grad_norm": 135.67138766139968,
+      "learning_rate": 5e-06,
+      "loss": 2.1961,
+      "num_input_tokens_seen": 493679272,
+      "step": 2856
+    },
+    {
+      "epoch": 1.093833780160858,
+      "loss": 2.183864116668701,
+      "loss_ce": 0.6299283504486084,
+      "loss_iou": 0.7201136350631714,
+      "loss_num": 0.022705078125,
+      "loss_xval": 1.5539357662200928,
+      "num_input_tokens_seen": 493679272,
+      "step": 2856
+    },
+    {
+      "epoch": 1.0942167751819227,
+      "grad_norm": 148.68457907104002,
+      "learning_rate": 5e-06,
+      "loss": 2.204,
+      "num_input_tokens_seen": 493851736,
+      "step": 2857
+    },
+    {
+      "epoch": 1.0942167751819227,
+      "loss": 2.527132511138916,
+      "loss_ce": 0.6028381586074829,
+      "loss_iou": 0.8985790610313416,
+      "loss_num": 0.025390625,
+      "loss_xval": 1.924294352531433,
+      "num_input_tokens_seen": 493851736,
+      "step": 2857
+    },
+    {
+      "epoch": 1.0945997702029873,
+      "grad_norm": 110.51339810601723,
+      "learning_rate": 5e-06,
+      "loss": 2.6957,
+      "num_input_tokens_seen": 494024760,
+      "step": 2858
+    },
+    {
+      "epoch": 1.0945997702029873,
+      "loss": 2.516145706176758,
+      "loss_ce": 0.649133563041687,
+      "loss_iou": 0.8449745178222656,
+      "loss_num": 0.035400390625,
+      "loss_xval": 1.8670120239257812,
+      "num_input_tokens_seen": 494024760,
+      "step": 2858
+    },
+    {
+      "epoch": 1.094982765224052,
+      "grad_norm": 99.21795225591899,
+      "learning_rate": 5e-06,
+      "loss": 2.016,
+      "num_input_tokens_seen": 494197848,
+      "step": 2859
+    },
+    {
+      "epoch": 1.094982765224052,
+      "loss": 2.1738383769989014,
+      "loss_ce": 0.6236727833747864,
+      "loss_iou": 0.7145969867706299,
+      "loss_num": 0.024169921875,
+      "loss_xval": 1.5501655340194702,
+      "num_input_tokens_seen": 494197848,
+      "step": 2859
+    },
+    {
+      "epoch": 1.0953657602451168,
+      "grad_norm": 197.07936172925068,
+      "learning_rate": 5e-06,
+      "loss": 1.9726,
+      "num_input_tokens_seen": 494371280,
+      "step": 2860
+    },
+    {
+      "epoch": 1.0953657602451168,
+      "loss": 2.044356346130371,
+      "loss_ce": 0.6590442061424255,
+      "loss_iou": 0.6417223215103149,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 1.3853123188018799,
+      "num_input_tokens_seen": 494371280,
+      "step": 2860
+    },
+    {
+      "epoch": 1.0957487552661815,
+      "grad_norm": 108.94740994693298,
+      "learning_rate": 5e-06,
+      "loss": 2.6536,
+      "num_input_tokens_seen": 494544024,
+      "step": 2861
+    },
+    {
+      "epoch": 1.0957487552661815,
+      "loss": 2.578054189682007,
+      "loss_ce": 0.5826956629753113,
+      "loss_iou": 0.9052110314369202,
+      "loss_num": 0.036865234375,
+      "loss_xval": 1.9953584671020508,
+      "num_input_tokens_seen": 494544024,
+      "step": 2861
+    },
+    {
+      "epoch": 1.0961317502872463,
+      "grad_norm": 167.67119584141844,
+      "learning_rate": 5e-06,
+      "loss": 2.0504,
+      "num_input_tokens_seen": 494717120,
+      "step": 2862
+    },
+    {
+      "epoch": 1.0961317502872463,
+      "loss": 1.9566445350646973,
+      "loss_ce": 0.5846694707870483,
+      "loss_iou": 0.6443310379981995,
+      "loss_num": 0.0166015625,
+      "loss_xval": 1.371975064277649,
+      "num_input_tokens_seen": 494717120,
+      "step": 2862
+    },
+    {
+      "epoch": 1.096514745308311,
+      "grad_norm": 332.4719782769773,
+      "learning_rate": 5e-06,
+      "loss": 2.8235,
+      "num_input_tokens_seen": 494890240,
+      "step": 2863
+    },
+    {
+      "epoch": 1.096514745308311,
+      "loss": 2.830749988555908,
+      "loss_ce": 0.5865357518196106,
+      "loss_iou": 1.0550904273986816,
+      "loss_num": 0.02685546875,
+      "loss_xval": 2.2442140579223633,
+      "num_input_tokens_seen": 494890240,
+      "step": 2863
+    },
+    {
+      "epoch": 1.0968977403293758,
+      "grad_norm": 136.13828964738934,
+      "learning_rate": 5e-06,
+      "loss": 2.4489,
+      "num_input_tokens_seen": 495063232,
+      "step": 2864
+    },
+    {
+      "epoch": 1.0968977403293758,
+      "loss": 2.4033021926879883,
+      "loss_ce": 0.6511270999908447,
+      "loss_iou": 0.8048596382141113,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 1.752175211906433,
+      "num_input_tokens_seen": 495063232,
+      "step": 2864
+    },
+    {
+      "epoch": 1.0972807353504404,
+      "grad_norm": 384.33973063577145,
+      "learning_rate": 5e-06,
+      "loss": 2.7995,
+      "num_input_tokens_seen": 495236240,
+      "step": 2865
+    },
+    {
+      "epoch": 1.0972807353504404,
+      "loss": 2.769808292388916,
+      "loss_ce": 0.6287542581558228,
+      "loss_iou": 0.9799509048461914,
+      "loss_num": 0.0361328125,
+      "loss_xval": 2.141054153442383,
+      "num_input_tokens_seen": 495236240,
+      "step": 2865
+    },
+    {
+      "epoch": 1.097663730371505,
+      "grad_norm": 188.15476483245635,
+      "learning_rate": 5e-06,
+      "loss": 2.6408,
+      "num_input_tokens_seen": 495409144,
+      "step": 2866
+    },
+    {
+      "epoch": 1.097663730371505,
+      "loss": 2.7341835498809814,
+      "loss_ce": 0.5780057311058044,
+      "loss_iou": 0.9697515368461609,
+      "loss_num": 0.04345703125,
+      "loss_xval": 2.1561777591705322,
+      "num_input_tokens_seen": 495409144,
+      "step": 2866
+    },
+    {
+      "epoch": 1.0980467253925699,
+      "grad_norm": 298.01300219408984,
+      "learning_rate": 5e-06,
+      "loss": 2.7389,
+      "num_input_tokens_seen": 495582432,
+      "step": 2867
+    },
+    {
+      "epoch": 1.0980467253925699,
+      "loss": 2.7545623779296875,
+      "loss_ce": 0.6666834950447083,
+      "loss_iou": 0.9440858364105225,
+      "loss_num": 0.0400390625,
+      "loss_xval": 2.087878704071045,
+      "num_input_tokens_seen": 495582432,
+      "step": 2867
+    },
+    {
+      "epoch": 1.0984297204136346,
+      "grad_norm": 62.530187986737516,
+      "learning_rate": 5e-06,
+      "loss": 2.5512,
+      "num_input_tokens_seen": 495755416,
+      "step": 2868
+    },
+    {
+      "epoch": 1.0984297204136346,
+      "loss": 2.5481717586517334,
+      "loss_ce": 0.6293148398399353,
+      "loss_iou": 0.8566452264785767,
+      "loss_num": 0.041015625,
+      "loss_xval": 1.9188568592071533,
+      "num_input_tokens_seen": 495755416,
+      "step": 2868
+    },
+    {
+      "epoch": 1.0988127154346994,
+      "grad_norm": 76.01019855681703,
+      "learning_rate": 5e-06,
+      "loss": 2.2085,
+      "num_input_tokens_seen": 495928232,
+      "step": 2869
+    },
+    {
+      "epoch": 1.0988127154346994,
+      "loss": 2.2282767295837402,
+      "loss_ce": 0.6378158330917358,
+      "loss_iou": 0.7301059365272522,
+      "loss_num": 0.026123046875,
+      "loss_xval": 1.5904608964920044,
+      "num_input_tokens_seen": 495928232,
+      "step": 2869
+    },
+    {
+      "epoch": 1.0991957104557641,
+      "grad_norm": 136.03440512249384,
+      "learning_rate": 5e-06,
+      "loss": 2.3114,
+      "num_input_tokens_seen": 496100928,
+      "step": 2870
+    },
+    {
+      "epoch": 1.0991957104557641,
+      "loss": 2.460556983947754,
+      "loss_ce": 0.6350305676460266,
+      "loss_iou": 0.8189521431922913,
+      "loss_num": 0.03759765625,
+      "loss_xval": 1.8255263566970825,
+      "num_input_tokens_seen": 496100928,
+      "step": 2870
+    },
+    {
+      "epoch": 1.099578705476829,
+      "grad_norm": 160.81883728934292,
+      "learning_rate": 5e-06,
+      "loss": 2.185,
+      "num_input_tokens_seen": 496274040,
+      "step": 2871
+    },
+    {
+      "epoch": 1.099578705476829,
+      "loss": 2.116295099258423,
+      "loss_ce": 0.5956770181655884,
+      "loss_iou": 0.7045534253120422,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 1.5206180810928345,
+      "num_input_tokens_seen": 496274040,
+      "step": 2871
+    },
+    {
+      "epoch": 1.0999617004978934,
+      "grad_norm": 127.7551622154229,
+      "learning_rate": 5e-06,
+      "loss": 2.1363,
+      "num_input_tokens_seen": 496447016,
+      "step": 2872
+    },
+    {
+      "epoch": 1.0999617004978934,
+      "loss": 2.1486854553222656,
+      "loss_ce": 0.5900260210037231,
+      "loss_iou": 0.7154869437217712,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 1.5586594343185425,
+      "num_input_tokens_seen": 496447016,
+      "step": 2872
+    },
+    {
+      "epoch": 1.1003446955189582,
+      "grad_norm": 99.36040266112705,
+      "learning_rate": 5e-06,
+      "loss": 2.1568,
+      "num_input_tokens_seen": 496620016,
+      "step": 2873
+    },
+    {
+      "epoch": 1.1003446955189582,
+      "loss": 2.2553670406341553,
+      "loss_ce": 0.6284530162811279,
+      "loss_iou": 0.7608752250671387,
+      "loss_num": 0.02099609375,
+      "loss_xval": 1.6269140243530273,
+      "num_input_tokens_seen": 496620016,
+      "step": 2873
+    },
+    {
+      "epoch": 1.100727690540023,
+      "grad_norm": 113.41540259100084,
+      "learning_rate": 5e-06,
+      "loss": 1.851,
+      "num_input_tokens_seen": 496793232,
+      "step": 2874
+    },
+    {
+      "epoch": 1.100727690540023,
+      "loss": 2.0813405513763428,
+      "loss_ce": 0.5790365934371948,
+      "loss_iou": 0.7055282592773438,
+      "loss_num": 0.018310546875,
+      "loss_xval": 1.5023040771484375,
+      "num_input_tokens_seen": 496793232,
+      "step": 2874
+    },
+    {
+      "epoch": 1.1011106855610877,
+      "grad_norm": 204.24479564880605,
+      "learning_rate": 5e-06,
+      "loss": 2.4847,
+      "num_input_tokens_seen": 496966232,
+      "step": 2875
+    },
+    {
+      "epoch": 1.1011106855610877,
+      "loss": 2.5521883964538574,
+      "loss_ce": 0.6590052247047424,
+      "loss_iou": 0.8944981098175049,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 1.8931832313537598,
+      "num_input_tokens_seen": 496966232,
+      "step": 2875
+    },
+    {
+      "epoch": 1.1014936805821525,
+      "grad_norm": 89.06032181286207,
+      "learning_rate": 5e-06,
+      "loss": 2.5199,
+      "num_input_tokens_seen": 497139064,
+      "step": 2876
+    },
+    {
+      "epoch": 1.1014936805821525,
+      "loss": 2.587961435317993,
+      "loss_ce": 0.5859717130661011,
+      "loss_iou": 0.9097473621368408,
+      "loss_num": 0.03662109375,
+      "loss_xval": 2.0019898414611816,
+      "num_input_tokens_seen": 497139064,
+      "step": 2876
+    },
+    {
+      "epoch": 1.1018766756032172,
+      "grad_norm": 53.350404368714756,
+      "learning_rate": 5e-06,
+      "loss": 2.1928,
+      "num_input_tokens_seen": 497312136,
+      "step": 2877
+    },
+    {
+      "epoch": 1.1018766756032172,
+      "loss": 2.1687560081481934,
+      "loss_ce": 0.6387738585472107,
+      "loss_iou": 0.7056039571762085,
+      "loss_num": 0.023681640625,
+      "loss_xval": 1.529982328414917,
+      "num_input_tokens_seen": 497312136,
+      "step": 2877
+    },
+    {
+      "epoch": 1.102259670624282,
+      "grad_norm": 87.39247123786168,
+      "learning_rate": 5e-06,
+      "loss": 1.912,
+      "num_input_tokens_seen": 497481128,
+      "step": 2878
+    },
+    {
+      "epoch": 1.102259670624282,
+      "loss": 1.871387243270874,
+      "loss_ce": 0.6387965083122253,
+      "loss_iou": 0.573814868927002,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 1.232590675354004,
+      "num_input_tokens_seen": 497481128,
+      "step": 2878
+    },
+    {
+      "epoch": 1.1026426656453465,
+      "grad_norm": 192.25773845240278,
+      "learning_rate": 5e-06,
+      "loss": 2.3301,
+      "num_input_tokens_seen": 497654184,
+      "step": 2879
+    },
+    {
+      "epoch": 1.1026426656453465,
+      "loss": 2.184957981109619,
+      "loss_ce": 0.6186071634292603,
+      "loss_iou": 0.7164029479026794,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 1.5663508176803589,
+      "num_input_tokens_seen": 497654184,
+      "step": 2879
+    },
+    {
+      "epoch": 1.1030256606664113,
+      "grad_norm": 313.25544215760783,
+      "learning_rate": 5e-06,
+      "loss": 2.887,
+      "num_input_tokens_seen": 497827048,
+      "step": 2880
+    },
+    {
+      "epoch": 1.1030256606664113,
+      "loss": 2.810183048248291,
+      "loss_ce": 0.6071299314498901,
+      "loss_iou": 1.0216926336288452,
+      "loss_num": 0.031982421875,
+      "loss_xval": 2.2030529975891113,
+      "num_input_tokens_seen": 497827048,
+      "step": 2880
+    },
+    {
+      "epoch": 1.103408655687476,
+      "grad_norm": 72.68253705884098,
+      "learning_rate": 5e-06,
+      "loss": 2.4853,
+      "num_input_tokens_seen": 497999984,
+      "step": 2881
+    },
+    {
+      "epoch": 1.103408655687476,
+      "loss": 2.429750919342041,
+      "loss_ce": 0.6344648599624634,
+      "loss_iou": 0.7948598861694336,
+      "loss_num": 0.041015625,
+      "loss_xval": 1.7819806337356567,
+      "num_input_tokens_seen": 497999984,
+      "step": 2881
+    },
+    {
+      "epoch": 1.1037916507085408,
+      "grad_norm": 99.60009967625733,
+      "learning_rate": 5e-06,
+      "loss": 2.1784,
+      "num_input_tokens_seen": 498172688,
+      "step": 2882
+    },
+    {
+      "epoch": 1.1037916507085408,
+      "loss": 2.157282829284668,
+      "loss_ce": 0.6364111304283142,
+      "loss_iou": 0.7046801447868347,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 1.5208715200424194,
+      "num_input_tokens_seen": 498172688,
+      "step": 2882
+    },
+    {
+      "epoch": 1.1041746457296056,
+      "grad_norm": 104.85088022303032,
+      "learning_rate": 5e-06,
+      "loss": 1.9184,
+      "num_input_tokens_seen": 498345624,
+      "step": 2883
+    },
+    {
+      "epoch": 1.1041746457296056,
+      "loss": 2.0137667655944824,
+      "loss_ce": 0.657676100730896,
+      "loss_iou": 0.623815655708313,
+      "loss_num": 0.021728515625,
+      "loss_xval": 1.356090784072876,
+      "num_input_tokens_seen": 498345624,
+      "step": 2883
+    },
+    {
+      "epoch": 1.1045576407506703,
+      "grad_norm": 260.0011582682272,
+      "learning_rate": 5e-06,
+      "loss": 2.3415,
+      "num_input_tokens_seen": 498518072,
+      "step": 2884
+    },
+    {
+      "epoch": 1.1045576407506703,
+      "loss": 2.4038548469543457,
+      "loss_ce": 0.6130247116088867,
+      "loss_iou": 0.8475940227508545,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 1.790830135345459,
+      "num_input_tokens_seen": 498518072,
+      "step": 2884
+    },
+    {
+      "epoch": 1.104940635771735,
+      "grad_norm": 84.77041030695203,
+      "learning_rate": 5e-06,
+      "loss": 2.7126,
+      "num_input_tokens_seen": 498690768,
+      "step": 2885
+    },
+    {
+      "epoch": 1.104940635771735,
+      "loss": 2.818016529083252,
+      "loss_ce": 0.6027964949607849,
+      "loss_iou": 1.0119068622589111,
+      "loss_num": 0.038330078125,
+      "loss_xval": 2.2152199745178223,
+      "num_input_tokens_seen": 498690768,
+      "step": 2885
+    },
+    {
+      "epoch": 1.1053236307927996,
+      "grad_norm": 118.91458595205364,
+      "learning_rate": 5e-06,
+      "loss": 2.4023,
+      "num_input_tokens_seen": 498863592,
+      "step": 2886
+    },
+    {
+      "epoch": 1.1053236307927996,
+      "loss": 2.473665714263916,
+      "loss_ce": 0.6676845550537109,
+      "loss_iou": 0.8347227573394775,
+      "loss_num": 0.02734375,
+      "loss_xval": 1.805981159210205,
+      "num_input_tokens_seen": 498863592,
+      "step": 2886
+    },
+    {
+      "epoch": 1.1057066258138644,
+      "grad_norm": 65.16102474275446,
+      "learning_rate": 5e-06,
+      "loss": 2.2125,
+      "num_input_tokens_seen": 499036496,
+      "step": 2887
+    },
+    {
+      "epoch": 1.1057066258138644,
+      "loss": 2.2524232864379883,
+      "loss_ce": 0.6581348776817322,
+      "loss_iou": 0.7283729314804077,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 1.5942885875701904,
+      "num_input_tokens_seen": 499036496,
+      "step": 2887
+    },
+    {
+      "epoch": 1.1060896208349291,
+      "grad_norm": 139.32643251555038,
+      "learning_rate": 5e-06,
+      "loss": 2.1848,
+      "num_input_tokens_seen": 499209312,
+      "step": 2888
+    },
+    {
+      "epoch": 1.1060896208349291,
+      "loss": 2.2043979167938232,
+      "loss_ce": 0.6207559704780579,
+      "loss_iou": 0.7400783896446228,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 1.5836418867111206,
+      "num_input_tokens_seen": 499209312,
+      "step": 2888
+    },
+    {
+      "epoch": 1.1064726158559939,
+      "grad_norm": 134.19208541204503,
+      "learning_rate": 5e-06,
+      "loss": 2.5185,
+      "num_input_tokens_seen": 499382184,
+      "step": 2889
+    },
+    {
+      "epoch": 1.1064726158559939,
+      "loss": 2.557345390319824,
+      "loss_ce": 0.610958456993103,
+      "loss_iou": 0.8938783407211304,
+      "loss_num": 0.03173828125,
+      "loss_xval": 1.9463870525360107,
+      "num_input_tokens_seen": 499382184,
+      "step": 2889
+    },
+    {
+      "epoch": 1.1068556108770586,
+      "grad_norm": 76.11039531379505,
+      "learning_rate": 5e-06,
+      "loss": 1.9668,
+      "num_input_tokens_seen": 499555192,
+      "step": 2890
+    },
+    {
+      "epoch": 1.1068556108770586,
+      "loss": 1.968073844909668,
+      "loss_ce": 0.6216538548469543,
+      "loss_iou": 0.6257856488227844,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 1.3464199304580688,
+      "num_input_tokens_seen": 499555192,
+      "step": 2890
+    },
+    {
+      "epoch": 1.1072386058981234,
+      "grad_norm": 154.2424569238608,
+      "learning_rate": 5e-06,
+      "loss": 2.2009,
+      "num_input_tokens_seen": 499728504,
+      "step": 2891
+    },
+    {
+      "epoch": 1.1072386058981234,
+      "loss": 2.0825328826904297,
+      "loss_ce": 0.6169934272766113,
+      "loss_iou": 0.6926695704460144,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 1.4655393362045288,
+      "num_input_tokens_seen": 499728504,
+      "step": 2891
+    },
+    {
+      "epoch": 1.1076216009191882,
+      "grad_norm": 226.64459263529557,
+      "learning_rate": 5e-06,
+      "loss": 2.4927,
+      "num_input_tokens_seen": 499901288,
+      "step": 2892
+    },
+    {
+      "epoch": 1.1076216009191882,
+      "loss": 2.600008487701416,
+      "loss_ce": 0.6066851615905762,
+      "loss_iou": 0.9301334023475647,
+      "loss_num": 0.026611328125,
+      "loss_xval": 1.9933234453201294,
+      "num_input_tokens_seen": 499901288,
+      "step": 2892
+    },
+    {
+      "epoch": 1.1080045959402527,
+      "grad_norm": 137.34916796942628,
+      "learning_rate": 5e-06,
+      "loss": 2.3934,
+      "num_input_tokens_seen": 500074440,
+      "step": 2893
+    },
+    {
+      "epoch": 1.1080045959402527,
+      "loss": 2.4239723682403564,
+      "loss_ce": 0.6311205625534058,
+      "loss_iou": 0.8262354135513306,
+      "loss_num": 0.028076171875,
+      "loss_xval": 1.7928516864776611,
+      "num_input_tokens_seen": 500074440,
+      "step": 2893
+    },
+    {
+      "epoch": 1.1083875909613174,
+      "grad_norm": 87.32613970301273,
+      "learning_rate": 5e-06,
+      "loss": 2.1623,
+      "num_input_tokens_seen": 500247240,
+      "step": 2894
+    },
+    {
+      "epoch": 1.1083875909613174,
+      "loss": 2.1889734268188477,
+      "loss_ce": 0.5862493515014648,
+      "loss_iou": 0.7315835952758789,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 1.6027240753173828,
+      "num_input_tokens_seen": 500247240,
+      "step": 2894
+    },
+    {
+      "epoch": 1.1087705859823822,
+      "grad_norm": 131.23641455431007,
+      "learning_rate": 5e-06,
+      "loss": 2.1098,
+      "num_input_tokens_seen": 500420064,
+      "step": 2895
+    },
+    {
+      "epoch": 1.1087705859823822,
+      "loss": 2.164780855178833,
+      "loss_ce": 0.6054773330688477,
+      "loss_iou": 0.7021981477737427,
+      "loss_num": 0.031005859375,
+      "loss_xval": 1.5593035221099854,
+      "num_input_tokens_seen": 500420064,
+      "step": 2895
+    },
+    {
+      "epoch": 1.109153581003447,
+      "grad_norm": 149.03837118405545,
+      "learning_rate": 5e-06,
+      "loss": 1.924,
+      "num_input_tokens_seen": 500593088,
+      "step": 2896
+    },
+    {
+      "epoch": 1.109153581003447,
+      "loss": 1.8993990421295166,
+      "loss_ce": 0.5676180124282837,
+      "loss_iou": 0.619732677936554,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 1.331781029701233,
+      "num_input_tokens_seen": 500593088,
+      "step": 2896
+    },
+    {
+      "epoch": 1.1095365760245117,
+      "grad_norm": 122.9939154883094,
+      "learning_rate": 5e-06,
+      "loss": 2.3066,
+      "num_input_tokens_seen": 500765976,
+      "step": 2897
+    },
+    {
+      "epoch": 1.1095365760245117,
+      "loss": 2.4041805267333984,
+      "loss_ce": 0.645162045955658,
+      "loss_iou": 0.8066943287849426,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 1.7590185403823853,
+      "num_input_tokens_seen": 500765976,
+      "step": 2897
+    },
+    {
+      "epoch": 1.1099195710455765,
+      "grad_norm": 88.42111700464596,
+      "learning_rate": 5e-06,
+      "loss": 2.1336,
+      "num_input_tokens_seen": 500938864,
+      "step": 2898
+    },
+    {
+      "epoch": 1.1099195710455765,
+      "loss": 2.1812314987182617,
+      "loss_ce": 0.5808380842208862,
+      "loss_iou": 0.7426711320877075,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 1.6003934144973755,
+      "num_input_tokens_seen": 500938864,
+      "step": 2898
+    },
+    {
+      "epoch": 1.110302566066641,
+      "grad_norm": 108.04490992874197,
+      "learning_rate": 5e-06,
+      "loss": 1.9805,
+      "num_input_tokens_seen": 501111808,
+      "step": 2899
+    },
+    {
+      "epoch": 1.110302566066641,
+      "loss": 1.9913827180862427,
+      "loss_ce": 0.6420392394065857,
+      "loss_iou": 0.6279035210609436,
+      "loss_num": 0.018798828125,
+      "loss_xval": 1.3493434190750122,
+      "num_input_tokens_seen": 501111808,
+      "step": 2899
+    },
+    {
+      "epoch": 1.1106855610877058,
+      "grad_norm": 179.89219576201393,
+      "learning_rate": 5e-06,
+      "loss": 2.3296,
+      "num_input_tokens_seen": 501285208,
+      "step": 2900
+    },
+    {
+      "epoch": 1.1106855610877058,
+      "loss": 2.362100124359131,
+      "loss_ce": 0.6323884725570679,
+      "loss_iou": 0.8084898591041565,
+      "loss_num": 0.0224609375,
+      "loss_xval": 1.729711651802063,
+      "num_input_tokens_seen": 501285208,
+      "step": 2900
+    },
+    {
+      "epoch": 1.1110685561087705,
+      "grad_norm": 125.27576242226647,
+      "learning_rate": 5e-06,
+      "loss": 2.3567,
+      "num_input_tokens_seen": 501457992,
+      "step": 2901
+    },
+    {
+      "epoch": 1.1110685561087705,
+      "loss": 2.4861788749694824,
+      "loss_ce": 0.6386929750442505,
+      "loss_iou": 0.8328006267547607,
+      "loss_num": 0.036376953125,
+      "loss_xval": 1.8474860191345215,
+      "num_input_tokens_seen": 501457992,
+      "step": 2901
+    },
+    {
+      "epoch": 1.1114515511298353,
+      "grad_norm": 79.61821369387268,
+      "learning_rate": 5e-06,
+      "loss": 1.9526,
+      "num_input_tokens_seen": 501631104,
+      "step": 2902
+    },
+    {
+      "epoch": 1.1114515511298353,
+      "loss": 2.082009792327881,
+      "loss_ce": 0.6238939762115479,
+      "loss_iou": 0.6812369227409363,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 1.4581159353256226,
+      "num_input_tokens_seen": 501631104,
+      "step": 2902
+    },
+    {
+      "epoch": 1.1118345461509,
+      "grad_norm": 99.69354139503271,
+      "learning_rate": 5e-06,
+      "loss": 1.9434,
+      "num_input_tokens_seen": 501804296,
+      "step": 2903
+    },
+    {
+      "epoch": 1.1118345461509,
+      "loss": 1.9701728820800781,
+      "loss_ce": 0.6007962226867676,
+      "loss_iou": 0.6232259273529053,
+      "loss_num": 0.024658203125,
+      "loss_xval": 1.3693766593933105,
+      "num_input_tokens_seen": 501804296,
+      "step": 2903
+    },
+    {
+      "epoch": 1.1122175411719648,
+      "grad_norm": 183.53801831093696,
+      "learning_rate": 5e-06,
+      "loss": 2.2145,
+      "num_input_tokens_seen": 501977424,
+      "step": 2904
+    },
+    {
+      "epoch": 1.1122175411719648,
+      "loss": 2.293793201446533,
+      "loss_ce": 0.5702643990516663,
+      "loss_iou": 0.7880339622497559,
+      "loss_num": 0.029541015625,
+      "loss_xval": 1.7235288619995117,
+      "num_input_tokens_seen": 501977424,
+      "step": 2904
+    },
+    {
+      "epoch": 1.1126005361930296,
+      "grad_norm": 201.893057671295,
+      "learning_rate": 5e-06,
+      "loss": 2.1023,
+      "num_input_tokens_seen": 502150080,
+      "step": 2905
+    },
+    {
+      "epoch": 1.1126005361930296,
+      "loss": 2.3387532234191895,
+      "loss_ce": 0.6087428331375122,
+      "loss_iou": 0.8072353601455688,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 1.7300102710723877,
+      "num_input_tokens_seen": 502150080,
+      "step": 2905
+    },
+    {
+      "epoch": 1.1129835312140943,
+      "grad_norm": 190.64929744087027,
+      "learning_rate": 5e-06,
+      "loss": 2.4365,
+      "num_input_tokens_seen": 502323136,
+      "step": 2906
+    },
+    {
+      "epoch": 1.1129835312140943,
+      "loss": 2.474616050720215,
+      "loss_ce": 0.6042324304580688,
+      "loss_iou": 0.8702505230903625,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 1.8703839778900146,
+      "num_input_tokens_seen": 502323136,
+      "step": 2906
+    },
+    {
+      "epoch": 1.1133665262351589,
+      "grad_norm": 202.96116314184982,
+      "learning_rate": 5e-06,
+      "loss": 2.3781,
+      "num_input_tokens_seen": 502496312,
+      "step": 2907
+    },
+    {
+      "epoch": 1.1133665262351589,
+      "loss": 2.477080821990967,
+      "loss_ce": 0.5928637385368347,
+      "loss_iou": 0.8861697912216187,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 1.8842170238494873,
+      "num_input_tokens_seen": 502496312,
+      "step": 2907
+    },
+    {
+      "epoch": 1.1137495212562236,
+      "grad_norm": 247.36583450430214,
+      "learning_rate": 5e-06,
+      "loss": 2.714,
+      "num_input_tokens_seen": 502669216,
+      "step": 2908
+    },
+    {
+      "epoch": 1.1137495212562236,
+      "loss": 2.628657341003418,
+      "loss_ce": 0.5726035833358765,
+      "loss_iou": 0.9327510595321655,
+      "loss_num": 0.0380859375,
+      "loss_xval": 2.035790205001831,
+      "num_input_tokens_seen": 502669216,
+      "step": 2908
+    },
+    {
+      "epoch": 1.1141325162772884,
+      "grad_norm": 297.24115073489264,
+      "learning_rate": 5e-06,
+      "loss": 2.3879,
+      "num_input_tokens_seen": 502842104,
+      "step": 2909
+    },
+    {
+      "epoch": 1.1141325162772884,
+      "loss": 2.3422248363494873,
+      "loss_ce": 0.610176682472229,
+      "loss_iou": 0.8055077195167542,
+      "loss_num": 0.024169921875,
+      "loss_xval": 1.7320481538772583,
+      "num_input_tokens_seen": 502842104,
+      "step": 2909
+    },
+    {
+      "epoch": 1.1145155112983531,
+      "grad_norm": 75.55251896683606,
+      "learning_rate": 5e-06,
+      "loss": 2.6398,
+      "num_input_tokens_seen": 503015640,
+      "step": 2910
+    },
+    {
+      "epoch": 1.1145155112983531,
+      "loss": 2.694908857345581,
+      "loss_ce": 0.6324771642684937,
+      "loss_iou": 0.9035302400588989,
+      "loss_num": 0.051025390625,
+      "loss_xval": 2.062431812286377,
+      "num_input_tokens_seen": 503015640,
+      "step": 2910
+    },
+    {
+      "epoch": 1.114898506319418,
+      "grad_norm": 103.0143953814475,
+      "learning_rate": 5e-06,
+      "loss": 2.1158,
+      "num_input_tokens_seen": 503188752,
+      "step": 2911
+    },
+    {
+      "epoch": 1.114898506319418,
+      "loss": 2.159280776977539,
+      "loss_ce": 0.5753985643386841,
+      "loss_iou": 0.7022194862365723,
+      "loss_num": 0.035888671875,
+      "loss_xval": 1.5838823318481445,
+      "num_input_tokens_seen": 503188752,
+      "step": 2911
+    },
+    {
+      "epoch": 1.1152815013404827,
+      "grad_norm": 172.0503308676565,
+      "learning_rate": 5e-06,
+      "loss": 2.1211,
+      "num_input_tokens_seen": 503361272,
+      "step": 2912
+    },
+    {
+      "epoch": 1.1152815013404827,
+      "loss": 1.9591606855392456,
+      "loss_ce": 0.5694947242736816,
+      "loss_iou": 0.6471339464187622,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 1.3896658420562744,
+      "num_input_tokens_seen": 503361272,
+      "step": 2912
+    },
+    {
+      "epoch": 1.1156644963615472,
+      "grad_norm": 146.1986484175921,
+      "learning_rate": 5e-06,
+      "loss": 2.661,
+      "num_input_tokens_seen": 503533984,
+      "step": 2913
+    },
+    {
+      "epoch": 1.1156644963615472,
+      "loss": 2.7327136993408203,
+      "loss_ce": 0.6500381231307983,
+      "loss_iou": 0.9599170088768005,
+      "loss_num": 0.03271484375,
+      "loss_xval": 2.0826759338378906,
+      "num_input_tokens_seen": 503533984,
+      "step": 2913
+    },
+    {
+      "epoch": 1.116047491382612,
+      "grad_norm": 102.3933499563087,
+      "learning_rate": 5e-06,
+      "loss": 2.4212,
+      "num_input_tokens_seen": 503706936,
+      "step": 2914
+    },
+    {
+      "epoch": 1.116047491382612,
+      "loss": 2.3756508827209473,
+      "loss_ce": 0.6869063377380371,
+      "loss_iou": 0.7599606513977051,
+      "loss_num": 0.03369140625,
+      "loss_xval": 1.6887445449829102,
+      "num_input_tokens_seen": 503706936,
+      "step": 2914
+    },
+    {
+      "epoch": 1.1164304864036767,
+      "grad_norm": 135.0876229271413,
+      "learning_rate": 5e-06,
+      "loss": 2.1516,
+      "num_input_tokens_seen": 503879768,
+      "step": 2915
+    },
+    {
+      "epoch": 1.1164304864036767,
+      "loss": 2.2691879272460938,
+      "loss_ce": 0.6411617994308472,
+      "loss_iou": 0.7393975257873535,
+      "loss_num": 0.02978515625,
+      "loss_xval": 1.628026008605957,
+      "num_input_tokens_seen": 503879768,
+      "step": 2915
+    },
+    {
+      "epoch": 1.1168134814247415,
+      "grad_norm": 302.1638750177775,
+      "learning_rate": 5e-06,
+      "loss": 2.2163,
+      "num_input_tokens_seen": 504052560,
+      "step": 2916
+    },
+    {
+      "epoch": 1.1168134814247415,
+      "loss": 2.273881435394287,
+      "loss_ce": 0.6011604070663452,
+      "loss_iou": 0.7756763100624084,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 1.672721028327942,
+      "num_input_tokens_seen": 504052560,
+      "step": 2916
+    },
+    {
+      "epoch": 1.1171964764458062,
+      "grad_norm": 67.07883381518754,
+      "learning_rate": 5e-06,
+      "loss": 2.2384,
+      "num_input_tokens_seen": 504225232,
+      "step": 2917
+    },
+    {
+      "epoch": 1.1171964764458062,
+      "loss": 2.1237258911132812,
+      "loss_ce": 0.6211868524551392,
+      "loss_iou": 0.680865466594696,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 1.502539038658142,
+      "num_input_tokens_seen": 504225232,
+      "step": 2917
+    },
+    {
+      "epoch": 1.117579471466871,
+      "grad_norm": 159.24011522402344,
+      "learning_rate": 5e-06,
+      "loss": 2.0324,
+      "num_input_tokens_seen": 504398160,
+      "step": 2918
+    },
+    {
+      "epoch": 1.117579471466871,
+      "loss": 2.0517094135284424,
+      "loss_ce": 0.603214681148529,
+      "loss_iou": 0.667332112789154,
+      "loss_num": 0.022705078125,
+      "loss_xval": 1.448494791984558,
+      "num_input_tokens_seen": 504398160,
+      "step": 2918
+    },
+    {
+      "epoch": 1.1179624664879357,
+      "grad_norm": 93.94077699502306,
+      "learning_rate": 5e-06,
+      "loss": 2.1048,
+      "num_input_tokens_seen": 504571184,
+      "step": 2919
+    },
+    {
+      "epoch": 1.1179624664879357,
+      "loss": 2.038072109222412,
+      "loss_ce": 0.5945290327072144,
+      "loss_iou": 0.6829226016998291,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 1.4435429573059082,
+      "num_input_tokens_seen": 504571184,
+      "step": 2919
+    },
+    {
+      "epoch": 1.1183454615090005,
+      "grad_norm": 120.5352238902658,
+      "learning_rate": 5e-06,
+      "loss": 2.2054,
+      "num_input_tokens_seen": 504744248,
+      "step": 2920
+    },
+    {
+      "epoch": 1.1183454615090005,
+      "loss": 2.1758975982666016,
+      "loss_ce": 0.5827404856681824,
+      "loss_iou": 0.7432339191436768,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 1.5931572914123535,
+      "num_input_tokens_seen": 504744248,
+      "step": 2920
+    },
+    {
+      "epoch": 1.118728456530065,
+      "grad_norm": 147.69704849799925,
+      "learning_rate": 5e-06,
+      "loss": 2.3654,
+      "num_input_tokens_seen": 504916992,
+      "step": 2921
+    },
+    {
+      "epoch": 1.118728456530065,
+      "loss": 2.4102468490600586,
+      "loss_ce": 0.605229377746582,
+      "loss_iou": 0.8305176496505737,
+      "loss_num": 0.02880859375,
+      "loss_xval": 1.8050172328948975,
+      "num_input_tokens_seen": 504916992,
+      "step": 2921
+    },
+    {
+      "epoch": 1.1191114515511298,
+      "grad_norm": 150.70132009449856,
+      "learning_rate": 5e-06,
+      "loss": 2.4769,
+      "num_input_tokens_seen": 505090440,
+      "step": 2922
+    },
+    {
+      "epoch": 1.1191114515511298,
+      "loss": 2.4750609397888184,
+      "loss_ce": 0.6045688390731812,
+      "loss_iou": 0.8626751899719238,
+      "loss_num": 0.029052734375,
+      "loss_xval": 1.8704919815063477,
+      "num_input_tokens_seen": 505090440,
+      "step": 2922
+    },
+    {
+      "epoch": 1.1194944465721945,
+      "grad_norm": 50.74361178457912,
+      "learning_rate": 5e-06,
+      "loss": 2.0561,
+      "num_input_tokens_seen": 505263352,
+      "step": 2923
+    },
+    {
+      "epoch": 1.1194944465721945,
+      "loss": 2.148237705230713,
+      "loss_ce": 0.6184177398681641,
+      "loss_iou": 0.7168753147125244,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.5298199653625488,
+      "num_input_tokens_seen": 505263352,
+      "step": 2923
+    },
+    {
+      "epoch": 1.1198774415932593,
+      "grad_norm": 59.80759277741302,
+      "learning_rate": 5e-06,
+      "loss": 1.8143,
+      "num_input_tokens_seen": 505436224,
+      "step": 2924
+    },
+    {
+      "epoch": 1.1198774415932593,
+      "loss": 1.801448106765747,
+      "loss_ce": 0.5888853669166565,
+      "loss_iou": 0.5577735900878906,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 1.2125625610351562,
+      "num_input_tokens_seen": 505436224,
+      "step": 2924
+    },
+    {
+      "epoch": 1.120260436614324,
+      "grad_norm": 130.05618151174428,
+      "learning_rate": 5e-06,
+      "loss": 2.0072,
+      "num_input_tokens_seen": 505609024,
+      "step": 2925
+    },
+    {
+      "epoch": 1.120260436614324,
+      "loss": 2.052910566329956,
+      "loss_ce": 0.5666494369506836,
+      "loss_iou": 0.7000855207443237,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 1.4862611293792725,
+      "num_input_tokens_seen": 505609024,
+      "step": 2925
+    },
+    {
+      "epoch": 1.1206434316353888,
+      "grad_norm": 133.19126693332592,
+      "learning_rate": 5e-06,
+      "loss": 2.3704,
+      "num_input_tokens_seen": 505781896,
+      "step": 2926
+    },
+    {
+      "epoch": 1.1206434316353888,
+      "loss": 2.4285738468170166,
+      "loss_ce": 0.5899032354354858,
+      "loss_iou": 0.8476800918579102,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 1.8386707305908203,
+      "num_input_tokens_seen": 505781896,
+      "step": 2926
+    },
+    {
+      "epoch": 1.1210264266564534,
+      "grad_norm": 115.55355336966625,
+      "learning_rate": 5e-06,
+      "loss": 1.7624,
+      "num_input_tokens_seen": 505955000,
+      "step": 2927
+    },
+    {
+      "epoch": 1.1210264266564534,
+      "loss": 1.7960158586502075,
+      "loss_ce": 0.5944313406944275,
+      "loss_iou": 0.5624850392341614,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 1.2015844583511353,
+      "num_input_tokens_seen": 505955000,
+      "step": 2927
+    },
+    {
+      "epoch": 1.1214094216775181,
+      "grad_norm": 167.47206572003535,
+      "learning_rate": 5e-06,
+      "loss": 2.2812,
+      "num_input_tokens_seen": 506127864,
+      "step": 2928
+    },
+    {
+      "epoch": 1.1214094216775181,
+      "loss": 2.132957935333252,
+      "loss_ce": 0.5568255186080933,
+      "loss_iou": 0.7221788167953491,
+      "loss_num": 0.0263671875,
+      "loss_xval": 1.5761325359344482,
+      "num_input_tokens_seen": 506127864,
+      "step": 2928
+    },
+    {
+      "epoch": 1.1217924166985829,
+      "grad_norm": 283.35226630248883,
+      "learning_rate": 5e-06,
+      "loss": 2.0465,
+      "num_input_tokens_seen": 506300784,
+      "step": 2929
+    },
+    {
+      "epoch": 1.1217924166985829,
+      "loss": 2.1811602115631104,
+      "loss_ce": 0.5846886038780212,
+      "loss_iou": 0.7596005201339722,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 1.5964715480804443,
+      "num_input_tokens_seen": 506300784,
+      "step": 2929
+    },
+    {
+      "epoch": 1.1221754117196476,
+      "grad_norm": 87.36076360074662,
+      "learning_rate": 5e-06,
+      "loss": 2.4209,
+      "num_input_tokens_seen": 506473792,
+      "step": 2930
+    },
+    {
+      "epoch": 1.1221754117196476,
+      "loss": 2.3417320251464844,
+      "loss_ce": 0.5851880311965942,
+      "loss_iou": 0.8060064315795898,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 1.7565441131591797,
+      "num_input_tokens_seen": 506473792,
+      "step": 2930
+    },
+    {
+      "epoch": 1.1225584067407124,
+      "grad_norm": 134.59615822917155,
+      "learning_rate": 5e-06,
+      "loss": 2.0875,
+      "num_input_tokens_seen": 506646888,
+      "step": 2931
+    },
+    {
+      "epoch": 1.1225584067407124,
+      "loss": 2.0470709800720215,
+      "loss_ce": 0.6375364065170288,
+      "loss_iou": 0.6499576568603516,
+      "loss_num": 0.02197265625,
+      "loss_xval": 1.4095344543457031,
+      "num_input_tokens_seen": 506646888,
+      "step": 2931
+    },
+    {
+      "epoch": 1.1229414017617771,
+      "grad_norm": 183.85111989359334,
+      "learning_rate": 5e-06,
+      "loss": 2.3452,
+      "num_input_tokens_seen": 506820112,
+      "step": 2932
+    },
+    {
+      "epoch": 1.1229414017617771,
+      "loss": 2.2805278301239014,
+      "loss_ce": 0.6472984552383423,
+      "loss_iou": 0.7543282508850098,
+      "loss_num": 0.02490234375,
+      "loss_xval": 1.6332292556762695,
+      "num_input_tokens_seen": 506820112,
+      "step": 2932
+    },
+    {
+      "epoch": 1.123324396782842,
+      "grad_norm": 100.58133962340794,
+      "learning_rate": 5e-06,
+      "loss": 2.2745,
+      "num_input_tokens_seen": 506993024,
+      "step": 2933
+    },
+    {
+      "epoch": 1.123324396782842,
+      "loss": 2.497943162918091,
+      "loss_ce": 0.5707536935806274,
+      "loss_iou": 0.8855613470077515,
+      "loss_num": 0.03125,
+      "loss_xval": 1.927189588546753,
+      "num_input_tokens_seen": 506993024,
+      "step": 2933
+    },
+    {
+      "epoch": 1.1237073918039067,
+      "grad_norm": 65.21433535769,
+      "learning_rate": 5e-06,
+      "loss": 2.0247,
+      "num_input_tokens_seen": 507165872,
+      "step": 2934
+    },
+    {
+      "epoch": 1.1237073918039067,
+      "loss": 1.8829970359802246,
+      "loss_ce": 0.569800853729248,
+      "loss_iou": 0.6100282073020935,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 1.313196063041687,
+      "num_input_tokens_seen": 507165872,
+      "step": 2934
+    },
+    {
+      "epoch": 1.1240903868249712,
+      "grad_norm": 102.68592955931913,
+      "learning_rate": 5e-06,
+      "loss": 1.9043,
+      "num_input_tokens_seen": 507338480,
+      "step": 2935
+    },
+    {
+      "epoch": 1.1240903868249712,
+      "loss": 1.7282692193984985,
+      "loss_ce": 0.5764355063438416,
+      "loss_iou": 0.5376782417297363,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 1.1518335342407227,
+      "num_input_tokens_seen": 507338480,
+      "step": 2935
+    },
+    {
+      "epoch": 1.124473381846036,
+      "grad_norm": 141.3760785132849,
+      "learning_rate": 5e-06,
+      "loss": 2.6952,
+      "num_input_tokens_seen": 507511264,
+      "step": 2936
+    },
+    {
+      "epoch": 1.124473381846036,
+      "loss": 2.7415976524353027,
+      "loss_ce": 0.5869454145431519,
+      "loss_iou": 0.9786018133163452,
+      "loss_num": 0.03955078125,
+      "loss_xval": 2.1546523571014404,
+      "num_input_tokens_seen": 507511264,
+      "step": 2936
+    },
+    {
+      "epoch": 1.1248563768671007,
+      "grad_norm": 111.42610780173028,
+      "learning_rate": 5e-06,
+      "loss": 2.0939,
+      "num_input_tokens_seen": 507684232,
+      "step": 2937
+    },
+    {
+      "epoch": 1.1248563768671007,
+      "loss": 2.3174006938934326,
+      "loss_ce": 0.5758513808250427,
+      "loss_iou": 0.7888349294662476,
+      "loss_num": 0.03271484375,
+      "loss_xval": 1.7415492534637451,
+      "num_input_tokens_seen": 507684232,
+      "step": 2937
+    },
+    {
+      "epoch": 1.1252393718881655,
+      "grad_norm": 61.77266775174526,
+      "learning_rate": 5e-06,
+      "loss": 1.9409,
+      "num_input_tokens_seen": 507856928,
+      "step": 2938
+    },
+    {
+      "epoch": 1.1252393718881655,
+      "loss": 1.7175027132034302,
+      "loss_ce": 0.5949825644493103,
+      "loss_iou": 0.524501621723175,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 1.122520089149475,
+      "num_input_tokens_seen": 507856928,
+      "step": 2938
+    },
+    {
+      "epoch": 1.1256223669092302,
+      "grad_norm": 103.4123729731931,
+      "learning_rate": 5e-06,
+      "loss": 1.9969,
+      "num_input_tokens_seen": 508029744,
+      "step": 2939
+    },
+    {
+      "epoch": 1.1256223669092302,
+      "loss": 2.0798449516296387,
+      "loss_ce": 0.5971314311027527,
+      "loss_iou": 0.6966027617454529,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 1.4827135801315308,
+      "num_input_tokens_seen": 508029744,
+      "step": 2939
+    },
+    {
+      "epoch": 1.126005361930295,
+      "grad_norm": 168.68270993108666,
+      "learning_rate": 5e-06,
+      "loss": 2.1583,
+      "num_input_tokens_seen": 508202720,
+      "step": 2940
+    },
+    {
+      "epoch": 1.126005361930295,
+      "loss": 2.2846107482910156,
+      "loss_ce": 0.6081647872924805,
+      "loss_iou": 0.7893948554992676,
+      "loss_num": 0.01953125,
+      "loss_xval": 1.6764459609985352,
+      "num_input_tokens_seen": 508202720,
+      "step": 2940
+    },
+    {
+      "epoch": 1.1263883569513595,
+      "grad_norm": 162.22939937828494,
+      "learning_rate": 5e-06,
+      "loss": 2.2816,
+      "num_input_tokens_seen": 508375792,
+      "step": 2941
+    },
+    {
+      "epoch": 1.1263883569513595,
+      "loss": 2.4022090435028076,
+      "loss_ce": 0.5834155082702637,
+      "loss_iou": 0.842105507850647,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 1.818793535232544,
+      "num_input_tokens_seen": 508375792,
+      "step": 2941
+    },
+    {
+      "epoch": 1.1267713519724243,
+      "grad_norm": 133.3995652593231,
+      "learning_rate": 5e-06,
+      "loss": 2.007,
+      "num_input_tokens_seen": 508548848,
+      "step": 2942
+    },
+    {
+      "epoch": 1.1267713519724243,
+      "loss": 2.081279754638672,
+      "loss_ce": 0.5720000267028809,
+      "loss_iou": 0.7090771794319153,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 1.5092798471450806,
+      "num_input_tokens_seen": 508548848,
+      "step": 2942
+    },
+    {
+      "epoch": 1.127154346993489,
+      "grad_norm": 138.05559501511522,
+      "learning_rate": 5e-06,
+      "loss": 2.197,
+      "num_input_tokens_seen": 508721712,
+      "step": 2943
+    },
+    {
+      "epoch": 1.127154346993489,
+      "loss": 2.1268904209136963,
+      "loss_ce": 0.5533802509307861,
+      "loss_iou": 0.7246366143226624,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 1.5735102891921997,
+      "num_input_tokens_seen": 508721712,
+      "step": 2943
+    },
+    {
+      "epoch": 1.1275373420145538,
+      "grad_norm": 162.67668917262557,
+      "learning_rate": 5e-06,
+      "loss": 1.9532,
+      "num_input_tokens_seen": 508895048,
+      "step": 2944
+    },
+    {
+      "epoch": 1.1275373420145538,
+      "loss": 2.0485429763793945,
+      "loss_ce": 0.5689271092414856,
+      "loss_iou": 0.6940621137619019,
+      "loss_num": 0.018310546875,
+      "loss_xval": 1.4796159267425537,
+      "num_input_tokens_seen": 508895048,
+      "step": 2944
+    },
+    {
+      "epoch": 1.1279203370356186,
+      "grad_norm": 183.67846924849545,
+      "learning_rate": 5e-06,
+      "loss": 2.0898,
+      "num_input_tokens_seen": 509067848,
+      "step": 2945
+    },
+    {
+      "epoch": 1.1279203370356186,
+      "loss": 2.27622127532959,
+      "loss_ce": 0.5981290340423584,
+      "loss_iou": 0.7724873423576355,
+      "loss_num": 0.026611328125,
+      "loss_xval": 1.678092360496521,
+      "num_input_tokens_seen": 509067848,
+      "step": 2945
+    },
+    {
+      "epoch": 1.1283033320566833,
+      "grad_norm": 90.26400795267458,
+      "learning_rate": 5e-06,
+      "loss": 2.2719,
+      "num_input_tokens_seen": 509241072,
+      "step": 2946
+    },
+    {
+      "epoch": 1.1283033320566833,
+      "loss": 2.257774591445923,
+      "loss_ce": 0.5660457611083984,
+      "loss_iou": 0.7746363878250122,
+      "loss_num": 0.028564453125,
+      "loss_xval": 1.6917288303375244,
+      "num_input_tokens_seen": 509241072,
+      "step": 2946
+    },
+    {
+      "epoch": 1.128686327077748,
+      "grad_norm": 94.63258031697269,
+      "learning_rate": 5e-06,
+      "loss": 1.8902,
+      "num_input_tokens_seen": 509414184,
+      "step": 2947
+    },
+    {
+      "epoch": 1.128686327077748,
+      "loss": 1.777647852897644,
+      "loss_ce": 0.5984277725219727,
+      "loss_iou": 0.5606946349143982,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 1.1792200803756714,
+      "num_input_tokens_seen": 509414184,
+      "step": 2947
+    },
+    {
+      "epoch": 1.1290693220988128,
+      "grad_norm": 134.63031029028897,
+      "learning_rate": 5e-06,
+      "loss": 2.0517,
+      "num_input_tokens_seen": 509587032,
+      "step": 2948
+    },
+    {
+      "epoch": 1.1290693220988128,
+      "loss": 2.034257650375366,
+      "loss_ce": 0.5827445387840271,
+      "loss_iou": 0.6827267408370972,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 1.4515130519866943,
+      "num_input_tokens_seen": 509587032,
+      "step": 2948
+    },
+    {
+      "epoch": 1.1294523171198774,
+      "grad_norm": 171.3631026885139,
+      "learning_rate": 5e-06,
+      "loss": 2.1489,
+      "num_input_tokens_seen": 509760200,
+      "step": 2949
+    },
+    {
+      "epoch": 1.1294523171198774,
+      "loss": 2.174747943878174,
+      "loss_ce": 0.5836296081542969,
+      "loss_iou": 0.7315332293510437,
+      "loss_num": 0.025634765625,
+      "loss_xval": 1.5911182165145874,
+      "num_input_tokens_seen": 509760200,
+      "step": 2949
+    },
+    {
+      "epoch": 1.1298353121409421,
+      "grad_norm": 130.8085905636241,
+      "learning_rate": 5e-06,
+      "loss": 2.3057,
+      "num_input_tokens_seen": 509933456,
+      "step": 2950
+    },
+    {
+      "epoch": 1.1298353121409421,
+      "loss": 2.243117332458496,
+      "loss_ce": 0.53421950340271,
+      "loss_iou": 0.7882257699966431,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 1.7088978290557861,
+      "num_input_tokens_seen": 509933456,
+      "step": 2950
+    },
+    {
+      "epoch": 1.1302183071620069,
+      "grad_norm": 83.9293575351945,
+      "learning_rate": 5e-06,
+      "loss": 2.0456,
+      "num_input_tokens_seen": 510106688,
+      "step": 2951
+    },
+    {
+      "epoch": 1.1302183071620069,
+      "loss": 2.049988031387329,
+      "loss_ce": 0.551052451133728,
+      "loss_iou": 0.6941241025924683,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 1.4989354610443115,
+      "num_input_tokens_seen": 510106688,
+      "step": 2951
+    },
+    {
+      "epoch": 1.1306013021830716,
+      "grad_norm": 216.60638569578,
+      "learning_rate": 5e-06,
+      "loss": 1.9574,
+      "num_input_tokens_seen": 510279328,
+      "step": 2952
+    },
+    {
+      "epoch": 1.1306013021830716,
+      "loss": 2.0311362743377686,
+      "loss_ce": 0.5471563339233398,
+      "loss_iou": 0.6727455854415894,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 1.4839799404144287,
+      "num_input_tokens_seen": 510279328,
+      "step": 2952
+    },
+    {
+      "epoch": 1.1309842972041364,
+      "grad_norm": 132.50570065785007,
+      "learning_rate": 5e-06,
+      "loss": 2.0608,
+      "num_input_tokens_seen": 510452544,
+      "step": 2953
+    },
+    {
+      "epoch": 1.1309842972041364,
+      "loss": 2.091780662536621,
+      "loss_ce": 0.5471699237823486,
+      "loss_iou": 0.7210510969161987,
+      "loss_num": 0.0205078125,
+      "loss_xval": 1.5446107387542725,
+      "num_input_tokens_seen": 510452544,
+      "step": 2953
+    },
+    {
+      "epoch": 1.1313672922252012,
+      "grad_norm": 124.17635472163106,
+      "learning_rate": 5e-06,
+      "loss": 2.0695,
+      "num_input_tokens_seen": 510625568,
+      "step": 2954
+    },
+    {
+      "epoch": 1.1313672922252012,
+      "loss": 2.0901994705200195,
+      "loss_ce": 0.5760436654090881,
+      "loss_iou": 0.6860941052436829,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 1.5141559839248657,
+      "num_input_tokens_seen": 510625568,
+      "step": 2954
+    },
+    {
+      "epoch": 1.1317502872462657,
+      "grad_norm": 120.58443867857027,
+      "learning_rate": 5e-06,
+      "loss": 2.1215,
+      "num_input_tokens_seen": 510798616,
+      "step": 2955
+    },
+    {
+      "epoch": 1.1317502872462657,
+      "loss": 2.036712884902954,
+      "loss_ce": 0.6060461401939392,
+      "loss_iou": 0.6764082908630371,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 1.4306669235229492,
+      "num_input_tokens_seen": 510798616,
+      "step": 2955
+    },
+    {
+      "epoch": 1.1321332822673305,
+      "grad_norm": 207.7984739476321,
+      "learning_rate": 5e-06,
+      "loss": 2.0438,
+      "num_input_tokens_seen": 510971456,
+      "step": 2956
+    },
+    {
+      "epoch": 1.1321332822673305,
+      "loss": 1.9081439971923828,
+      "loss_ce": 0.5633724927902222,
+      "loss_iou": 0.6264721155166626,
+      "loss_num": 0.018310546875,
+      "loss_xval": 1.3447716236114502,
+      "num_input_tokens_seen": 510971456,
+      "step": 2956
+    },
+    {
+      "epoch": 1.1325162772883952,
+      "grad_norm": 135.05079194814368,
+      "learning_rate": 5e-06,
+      "loss": 2.0915,
+      "num_input_tokens_seen": 511144496,
+      "step": 2957
+    },
+    {
+      "epoch": 1.1325162772883952,
+      "loss": 2.1809394359588623,
+      "loss_ce": 0.5449835658073425,
+      "loss_iou": 0.762374997138977,
+      "loss_num": 0.022216796875,
+      "loss_xval": 1.635956048965454,
+      "num_input_tokens_seen": 511144496,
+      "step": 2957
+    },
+    {
+      "epoch": 1.13289927230946,
+      "grad_norm": 146.6064746236638,
+      "learning_rate": 5e-06,
+      "loss": 1.8096,
+      "num_input_tokens_seen": 511317528,
+      "step": 2958
+    },
+    {
+      "epoch": 1.13289927230946,
+      "loss": 1.8042373657226562,
+      "loss_ce": 0.5380661487579346,
+      "loss_iou": 0.5805953741073608,
+      "loss_num": 0.02099609375,
+      "loss_xval": 1.2661712169647217,
+      "num_input_tokens_seen": 511317528,
+      "step": 2958
+    },
+    {
+      "epoch": 1.1332822673305247,
+      "grad_norm": 182.20626784676352,
+      "learning_rate": 5e-06,
+      "loss": 2.1662,
+      "num_input_tokens_seen": 511486416,
+      "step": 2959
+    },
+    {
+      "epoch": 1.1332822673305247,
+      "loss": 2.0279202461242676,
+      "loss_ce": 0.5270369052886963,
+      "loss_iou": 0.6989279389381409,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 1.5008832216262817,
+      "num_input_tokens_seen": 511486416,
+      "step": 2959
+    },
+    {
+      "epoch": 1.1336652623515895,
+      "grad_norm": 130.73128800985407,
+      "learning_rate": 5e-06,
+      "loss": 2.0859,
+      "num_input_tokens_seen": 511659280,
+      "step": 2960
+    },
+    {
+      "epoch": 1.1336652623515895,
+      "loss": 2.081346035003662,
+      "loss_ce": 0.6038487553596497,
+      "loss_iou": 0.6952458620071411,
+      "loss_num": 0.017333984375,
+      "loss_xval": 1.4774973392486572,
+      "num_input_tokens_seen": 511659280,
+      "step": 2960
+    },
+    {
+      "epoch": 1.1340482573726542,
+      "grad_norm": 104.33194520531575,
+      "learning_rate": 5e-06,
+      "loss": 2.1437,
+      "num_input_tokens_seen": 511832640,
+      "step": 2961
+    },
+    {
+      "epoch": 1.1340482573726542,
+      "loss": 2.028927803039551,
+      "loss_ce": 0.5392760038375854,
+      "loss_iou": 0.6836686730384827,
+      "loss_num": 0.0244140625,
+      "loss_xval": 1.4896517992019653,
+      "num_input_tokens_seen": 511832640,
+      "step": 2961
+    },
+    {
+      "epoch": 1.134431252393719,
+      "grad_norm": 185.51423538122006,
+      "learning_rate": 5e-06,
+      "loss": 2.0979,
+      "num_input_tokens_seen": 512006024,
+      "step": 2962
+    },
+    {
+      "epoch": 1.134431252393719,
+      "loss": 2.060588836669922,
+      "loss_ce": 0.5479211807250977,
+      "loss_iou": 0.7005019187927246,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 1.5126676559448242,
+      "num_input_tokens_seen": 512006024,
+      "step": 2962
+    },
+    {
+      "epoch": 1.1348142474147835,
+      "grad_norm": 122.34527085825678,
+      "learning_rate": 5e-06,
+      "loss": 2.5203,
+      "num_input_tokens_seen": 512179048,
+      "step": 2963
+    },
+    {
+      "epoch": 1.1348142474147835,
+      "loss": 2.5278701782226562,
+      "loss_ce": 0.5822385549545288,
+      "loss_iou": 0.8813852071762085,
+      "loss_num": 0.03662109375,
+      "loss_xval": 1.945631742477417,
+      "num_input_tokens_seen": 512179048,
+      "step": 2963
+    },
+    {
+      "epoch": 1.1351972424358483,
+      "grad_norm": 78.56436353360586,
+      "learning_rate": 5e-06,
+      "loss": 2.0548,
+      "num_input_tokens_seen": 512352192,
+      "step": 2964
+    },
+    {
+      "epoch": 1.1351972424358483,
+      "loss": 1.862177848815918,
+      "loss_ce": 0.5344089269638062,
+      "loss_iou": 0.6073201894760132,
+      "loss_num": 0.022705078125,
+      "loss_xval": 1.3277690410614014,
+      "num_input_tokens_seen": 512352192,
+      "step": 2964
+    },
+    {
+      "epoch": 1.135580237456913,
+      "grad_norm": 279.6255196265593,
+      "learning_rate": 5e-06,
+      "loss": 2.1784,
+      "num_input_tokens_seen": 512524936,
+      "step": 2965
+    },
+    {
+      "epoch": 1.135580237456913,
+      "loss": 2.185105800628662,
+      "loss_ce": 0.5289831161499023,
+      "loss_iou": 0.7837650179862976,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 1.6561225652694702,
+      "num_input_tokens_seen": 512524936,
+      "step": 2965
+    },
+    {
+      "epoch": 1.1359632324779778,
+      "grad_norm": 98.68948307232363,
+      "learning_rate": 5e-06,
+      "loss": 2.532,
+      "num_input_tokens_seen": 512697464,
+      "step": 2966
+    },
+    {
+      "epoch": 1.1359632324779778,
+      "loss": 2.500180959701538,
+      "loss_ce": 0.560333251953125,
+      "loss_iou": 0.8793476819992065,
+      "loss_num": 0.0361328125,
+      "loss_xval": 1.9398475885391235,
+      "num_input_tokens_seen": 512697464,
+      "step": 2966
+    },
+    {
+      "epoch": 1.1363462274990426,
+      "grad_norm": 201.91759851011244,
+      "learning_rate": 5e-06,
+      "loss": 2.4546,
+      "num_input_tokens_seen": 512870456,
+      "step": 2967
+    },
+    {
+      "epoch": 1.1363462274990426,
+      "loss": 2.580899953842163,
+      "loss_ce": 0.5556600093841553,
+      "loss_iou": 0.9261636734008789,
+      "loss_num": 0.03466796875,
+      "loss_xval": 2.025239944458008,
+      "num_input_tokens_seen": 512870456,
+      "step": 2967
+    },
+    {
+      "epoch": 1.1367292225201073,
+      "grad_norm": 90.42583979033348,
+      "learning_rate": 5e-06,
+      "loss": 2.2886,
+      "num_input_tokens_seen": 513043744,
+      "step": 2968
+    },
+    {
+      "epoch": 1.1367292225201073,
+      "loss": 2.3291778564453125,
+      "loss_ce": 0.5995110273361206,
+      "loss_iou": 0.8052021265029907,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 1.7296669483184814,
+      "num_input_tokens_seen": 513043744,
+      "step": 2968
+    },
+    {
+      "epoch": 1.1371122175411719,
+      "grad_norm": 86.52715858494919,
+      "learning_rate": 5e-06,
+      "loss": 1.8593,
+      "num_input_tokens_seen": 513216520,
+      "step": 2969
+    },
+    {
+      "epoch": 1.1371122175411719,
+      "loss": 1.7987768650054932,
+      "loss_ce": 0.5653985142707825,
+      "loss_iou": 0.5767874717712402,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 1.2333784103393555,
+      "num_input_tokens_seen": 513216520,
+      "step": 2969
+    },
+    {
+      "epoch": 1.1374952125622366,
+      "grad_norm": 122.87749878125817,
+      "learning_rate": 5e-06,
+      "loss": 2.0181,
+      "num_input_tokens_seen": 513389552,
+      "step": 2970
+    },
+    {
+      "epoch": 1.1374952125622366,
+      "loss": 1.8785089254379272,
+      "loss_ce": 0.5847567319869995,
+      "loss_iou": 0.6094768047332764,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 1.2937521934509277,
+      "num_input_tokens_seen": 513389552,
+      "step": 2970
+    },
+    {
+      "epoch": 1.1378782075833014,
+      "grad_norm": 226.85791358507996,
+      "learning_rate": 5e-06,
+      "loss": 2.3399,
+      "num_input_tokens_seen": 513561832,
+      "step": 2971
+    },
+    {
+      "epoch": 1.1378782075833014,
+      "loss": 2.312203884124756,
+      "loss_ce": 0.5954175591468811,
+      "loss_iou": 0.8090767860412598,
+      "loss_num": 0.019775390625,
+      "loss_xval": 1.7167863845825195,
+      "num_input_tokens_seen": 513561832,
+      "step": 2971
+    },
+    {
+      "epoch": 1.1382612026043661,
+      "grad_norm": 90.29982701520944,
+      "learning_rate": 5e-06,
+      "loss": 2.2469,
+      "num_input_tokens_seen": 513734936,
+      "step": 2972
+    },
+    {
+      "epoch": 1.1382612026043661,
+      "loss": 2.257134199142456,
+      "loss_ce": 0.5403452515602112,
+      "loss_iou": 0.7982443571090698,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 1.7167890071868896,
+      "num_input_tokens_seen": 513734936,
+      "step": 2972
+    },
+    {
+      "epoch": 1.138644197625431,
+      "grad_norm": 155.1021040353951,
+      "learning_rate": 5e-06,
+      "loss": 1.9722,
+      "num_input_tokens_seen": 513907872,
+      "step": 2973
+    },
+    {
+      "epoch": 1.138644197625431,
+      "loss": 1.933345079421997,
+      "loss_ce": 0.5490626096725464,
+      "loss_iou": 0.6550623178482056,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 1.3842823505401611,
+      "num_input_tokens_seen": 513907872,
+      "step": 2973
+    },
+    {
+      "epoch": 1.1390271926464957,
+      "grad_norm": 112.20735893892689,
+      "learning_rate": 5e-06,
+      "loss": 2.0814,
+      "num_input_tokens_seen": 514080728,
+      "step": 2974
+    },
+    {
+      "epoch": 1.1390271926464957,
+      "loss": 2.0183908939361572,
+      "loss_ce": 0.5023434162139893,
+      "loss_iou": 0.693021297454834,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 1.510798454284668,
+      "num_input_tokens_seen": 514080728,
+      "step": 2974
+    },
+    {
+      "epoch": 1.1394101876675604,
+      "grad_norm": 271.87382046629205,
+      "learning_rate": 5e-06,
+      "loss": 2.0258,
+      "num_input_tokens_seen": 514253856,
+      "step": 2975
+    },
+    {
+      "epoch": 1.1394101876675604,
+      "loss": 2.0958802700042725,
+      "loss_ce": 0.5488342046737671,
+      "loss_iou": 0.7046447992324829,
+      "loss_num": 0.027587890625,
+      "loss_xval": 1.5470459461212158,
+      "num_input_tokens_seen": 514253856,
+      "step": 2975
+    },
+    {
+      "epoch": 1.1397931826886252,
+      "grad_norm": 127.40662963833958,
+      "learning_rate": 5e-06,
+      "loss": 2.3439,
+      "num_input_tokens_seen": 514427240,
+      "step": 2976
+    },
+    {
+      "epoch": 1.1397931826886252,
+      "loss": 2.33837890625,
+      "loss_ce": 0.5436375141143799,
+      "loss_iou": 0.8391736149787903,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 1.7947412729263306,
+      "num_input_tokens_seen": 514427240,
+      "step": 2976
+    },
+    {
+      "epoch": 1.1401761777096897,
+      "grad_norm": 265.98599893325144,
+      "learning_rate": 5e-06,
+      "loss": 2.076,
+      "num_input_tokens_seen": 514599880,
+      "step": 2977
+    },
+    {
+      "epoch": 1.1401761777096897,
+      "loss": 2.0318119525909424,
+      "loss_ce": 0.5489307641983032,
+      "loss_iou": 0.6901710033416748,
+      "loss_num": 0.0205078125,
+      "loss_xval": 1.4828810691833496,
+      "num_input_tokens_seen": 514599880,
+      "step": 2977
+    },
+    {
+      "epoch": 1.1405591727307545,
+      "grad_norm": 171.01161957565827,
+      "learning_rate": 5e-06,
+      "loss": 2.2905,
+      "num_input_tokens_seen": 514772440,
+      "step": 2978
+    },
+    {
+      "epoch": 1.1405591727307545,
+      "loss": 2.3822574615478516,
+      "loss_ce": 0.5285988450050354,
+      "loss_iou": 0.8416242003440857,
+      "loss_num": 0.0341796875,
+      "loss_xval": 1.8536585569381714,
+      "num_input_tokens_seen": 514772440,
+      "step": 2978
+    },
+    {
+      "epoch": 1.1409421677518192,
+      "grad_norm": 278.93089062665047,
+      "learning_rate": 5e-06,
+      "loss": 2.2799,
+      "num_input_tokens_seen": 514945344,
+      "step": 2979
+    },
+    {
+      "epoch": 1.1409421677518192,
+      "loss": 2.5735392570495605,
+      "loss_ce": 0.5378756523132324,
+      "loss_iou": 0.9553776383399963,
+      "loss_num": 0.02490234375,
+      "loss_xval": 2.035663604736328,
+      "num_input_tokens_seen": 514945344,
+      "step": 2979
+    },
+    {
+      "epoch": 1.141325162772884,
+      "grad_norm": 127.14942804815144,
+      "learning_rate": 5e-06,
+      "loss": 2.3917,
+      "num_input_tokens_seen": 515118176,
+      "step": 2980
+    },
+    {
+      "epoch": 1.141325162772884,
+      "loss": 2.3706796169281006,
+      "loss_ce": 0.5371127724647522,
+      "loss_iou": 0.836888313293457,
+      "loss_num": 0.031982421875,
+      "loss_xval": 1.833566665649414,
+      "num_input_tokens_seen": 515118176,
+      "step": 2980
+    },
+    {
+      "epoch": 1.1417081577939487,
+      "grad_norm": 98.17838159288843,
+      "learning_rate": 5e-06,
+      "loss": 1.6324,
+      "num_input_tokens_seen": 515290928,
+      "step": 2981
+    },
+    {
+      "epoch": 1.1417081577939487,
+      "loss": 1.5124866962432861,
+      "loss_ce": 0.5301163196563721,
+      "loss_iou": 0.45512863993644714,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.9823703169822693,
+      "num_input_tokens_seen": 515290928,
+      "step": 2981
+    },
+    {
+      "epoch": 1.1420911528150135,
+      "grad_norm": 231.40948469137263,
+      "learning_rate": 5e-06,
+      "loss": 2.637,
+      "num_input_tokens_seen": 515463840,
+      "step": 2982
+    },
+    {
+      "epoch": 1.1420911528150135,
+      "loss": 2.742429256439209,
+      "loss_ce": 0.5224322080612183,
+      "loss_iou": 1.048703908920288,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 2.219996929168701,
+      "num_input_tokens_seen": 515463840,
+      "step": 2982
+    },
+    {
+      "epoch": 1.142474147836078,
+      "grad_norm": 94.91045147812316,
+      "learning_rate": 5e-06,
+      "loss": 2.4669,
+      "num_input_tokens_seen": 515636528,
+      "step": 2983
+    },
+    {
+      "epoch": 1.142474147836078,
+      "loss": 2.3929977416992188,
+      "loss_ce": 0.5427125692367554,
+      "loss_iou": 0.8329795598983765,
+      "loss_num": 0.036865234375,
+      "loss_xval": 1.8502854108810425,
+      "num_input_tokens_seen": 515636528,
+      "step": 2983
+    },
+    {
+      "epoch": 1.1428571428571428,
+      "grad_norm": 77.5044490935408,
+      "learning_rate": 5e-06,
+      "loss": 2.1173,
+      "num_input_tokens_seen": 515809480,
+      "step": 2984
+    },
+    {
+      "epoch": 1.1428571428571428,
+      "loss": 2.1471107006073,
+      "loss_ce": 0.5289281606674194,
+      "loss_iou": 0.7355438470840454,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 1.6181824207305908,
+      "num_input_tokens_seen": 515809480,
+      "step": 2984
+    },
+    {
+      "epoch": 1.1432401378782076,
+      "grad_norm": 170.05717011259233,
+      "learning_rate": 5e-06,
+      "loss": 2.1984,
+      "num_input_tokens_seen": 515982408,
+      "step": 2985
+    },
+    {
+      "epoch": 1.1432401378782076,
+      "loss": 2.1551971435546875,
+      "loss_ce": 0.5516361594200134,
+      "loss_iou": 0.7555311918258667,
+      "loss_num": 0.0185546875,
+      "loss_xval": 1.6035611629486084,
+      "num_input_tokens_seen": 515982408,
+      "step": 2985
+    },
+    {
+      "epoch": 1.1436231328992723,
+      "grad_norm": 160.51003341679348,
+      "learning_rate": 5e-06,
+      "loss": 3.0738,
+      "num_input_tokens_seen": 516155320,
+      "step": 2986
+    },
+    {
+      "epoch": 1.1436231328992723,
+      "loss": 3.399182081222534,
+      "loss_ce": 0.5560451745986938,
+      "loss_iou": 1.3256211280822754,
+      "loss_num": 0.038330078125,
+      "loss_xval": 2.843136787414551,
+      "num_input_tokens_seen": 516155320,
+      "step": 2986
+    },
+    {
+      "epoch": 1.144006127920337,
+      "grad_norm": 128.32679133374626,
+      "learning_rate": 5e-06,
+      "loss": 2.2714,
+      "num_input_tokens_seen": 516328432,
+      "step": 2987
+    },
+    {
+      "epoch": 1.144006127920337,
+      "loss": 2.3910999298095703,
+      "loss_ce": 0.5498851537704468,
+      "loss_iou": 0.8372029066085815,
+      "loss_num": 0.033447265625,
+      "loss_xval": 1.841214895248413,
+      "num_input_tokens_seen": 516328432,
+      "step": 2987
+    },
+    {
+      "epoch": 1.1443891229414018,
+      "grad_norm": 64.3963790748517,
+      "learning_rate": 5e-06,
+      "loss": 2.0932,
+      "num_input_tokens_seen": 516498432,
+      "step": 2988
+    },
+    {
+      "epoch": 1.1443891229414018,
+      "loss": 2.0400850772857666,
+      "loss_ce": 0.5410438776016235,
+      "loss_iou": 0.6955808401107788,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 1.4990413188934326,
+      "num_input_tokens_seen": 516498432,
+      "step": 2988
+    },
+    {
+      "epoch": 1.1447721179624666,
+      "grad_norm": 124.91494391282485,
+      "learning_rate": 5e-06,
+      "loss": 1.9027,
+      "num_input_tokens_seen": 516671136,
+      "step": 2989
+    },
+    {
+      "epoch": 1.1447721179624666,
+      "loss": 2.0308449268341064,
+      "loss_ce": 0.5373611450195312,
+      "loss_iou": 0.6931225061416626,
+      "loss_num": 0.021484375,
+      "loss_xval": 1.4934837818145752,
+      "num_input_tokens_seen": 516671136,
+      "step": 2989
+    },
+    {
+      "epoch": 1.1451551129835311,
+      "grad_norm": 183.98754523757103,
+      "learning_rate": 5e-06,
+      "loss": 2.2069,
+      "num_input_tokens_seen": 516843880,
+      "step": 2990
+    },
+    {
+      "epoch": 1.1451551129835311,
+      "loss": 2.306507110595703,
+      "loss_ce": 0.5649071931838989,
+      "loss_iou": 0.8115959167480469,
+      "loss_num": 0.023681640625,
+      "loss_xval": 1.7415999174118042,
+      "num_input_tokens_seen": 516843880,
+      "step": 2990
+    },
+    {
+      "epoch": 1.1455381080045959,
+      "grad_norm": 88.9670323543909,
+      "learning_rate": 5e-06,
+      "loss": 2.4037,
+      "num_input_tokens_seen": 517017168,
+      "step": 2991
+    },
+    {
+      "epoch": 1.1455381080045959,
+      "loss": 2.4599194526672363,
+      "loss_ce": 0.5973893404006958,
+      "loss_iou": 0.842336893081665,
+      "loss_num": 0.03564453125,
+      "loss_xval": 1.86253023147583,
+      "num_input_tokens_seen": 517017168,
+      "step": 2991
+    },
+    {
+      "epoch": 1.1459211030256606,
+      "grad_norm": 86.98039124727654,
+      "learning_rate": 5e-06,
+      "loss": 2.0201,
+      "num_input_tokens_seen": 517190192,
+      "step": 2992
+    },
+    {
+      "epoch": 1.1459211030256606,
+      "loss": 2.0168302059173584,
+      "loss_ce": 0.5655362606048584,
+      "loss_iou": 0.6668243408203125,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 1.4512939453125,
+      "num_input_tokens_seen": 517190192,
+      "step": 2992
+    },
+    {
+      "epoch": 1.1463040980467254,
+      "grad_norm": 134.43718663507704,
+      "learning_rate": 5e-06,
+      "loss": 1.9381,
+      "num_input_tokens_seen": 517363216,
+      "step": 2993
+    },
+    {
+      "epoch": 1.1463040980467254,
+      "loss": 1.8722496032714844,
+      "loss_ce": 0.5516296625137329,
+      "loss_iou": 0.6261608600616455,
+      "loss_num": 0.013671875,
+      "loss_xval": 1.320620059967041,
+      "num_input_tokens_seen": 517363216,
+      "step": 2993
+    },
+    {
+      "epoch": 1.1466870930677902,
+      "grad_norm": 110.95566969817125,
+      "learning_rate": 5e-06,
+      "loss": 2.1158,
+      "num_input_tokens_seen": 517536136,
+      "step": 2994
+    },
+    {
+      "epoch": 1.1466870930677902,
+      "loss": 2.022862195968628,
+      "loss_ce": 0.5184419751167297,
+      "loss_iou": 0.6844000816345215,
+      "loss_num": 0.027099609375,
+      "loss_xval": 1.504420280456543,
+      "num_input_tokens_seen": 517536136,
+      "step": 2994
+    },
+    {
+      "epoch": 1.147070088088855,
+      "grad_norm": 88.68752519561478,
+      "learning_rate": 5e-06,
+      "loss": 1.9335,
+      "num_input_tokens_seen": 517708928,
+      "step": 2995
+    },
+    {
+      "epoch": 1.147070088088855,
+      "loss": 2.0943524837493896,
+      "loss_ce": 0.5301757454872131,
+      "loss_iou": 0.7113485932350159,
+      "loss_num": 0.0283203125,
+      "loss_xval": 1.5641766786575317,
+      "num_input_tokens_seen": 517708928,
+      "step": 2995
+    },
+    {
+      "epoch": 1.1474530831099194,
+      "grad_norm": 96.74101352871402,
+      "learning_rate": 5e-06,
+      "loss": 1.8792,
+      "num_input_tokens_seen": 517882120,
+      "step": 2996
+    },
+    {
+      "epoch": 1.1474530831099194,
+      "loss": 1.9112402200698853,
+      "loss_ce": 0.5335820317268372,
+      "loss_iou": 0.6322038173675537,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 1.3776583671569824,
+      "num_input_tokens_seen": 517882120,
+      "step": 2996
+    },
+    {
+      "epoch": 1.1478360781309842,
+      "grad_norm": 157.84643782714286,
+      "learning_rate": 5e-06,
+      "loss": 1.8198,
+      "num_input_tokens_seen": 518054944,
+      "step": 2997
+    },
+    {
+      "epoch": 1.1478360781309842,
+      "loss": 2.0277700424194336,
+      "loss_ce": 0.5270143747329712,
+      "loss_iou": 0.7133447527885437,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 1.5007556676864624,
+      "num_input_tokens_seen": 518054944,
+      "step": 2997
+    },
+    {
+      "epoch": 1.148219073152049,
+      "grad_norm": 128.03047628782417,
+      "learning_rate": 5e-06,
+      "loss": 2.1358,
+      "num_input_tokens_seen": 518227800,
+      "step": 2998
+    },
+    {
+      "epoch": 1.148219073152049,
+      "loss": 2.072675943374634,
+      "loss_ce": 0.5211225748062134,
+      "loss_iou": 0.7174270749092102,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 1.5515533685684204,
+      "num_input_tokens_seen": 518227800,
+      "step": 2998
+    },
+    {
+      "epoch": 1.1486020681731137,
+      "grad_norm": 58.07235021252467,
+      "learning_rate": 5e-06,
+      "loss": 1.7931,
+      "num_input_tokens_seen": 518400528,
+      "step": 2999
+    },
+    {
+      "epoch": 1.1486020681731137,
+      "loss": 1.7334604263305664,
+      "loss_ce": 0.5397881269454956,
+      "loss_iou": 0.5571175217628479,
+      "loss_num": 0.015869140625,
+      "loss_xval": 1.1936722993850708,
+      "num_input_tokens_seen": 518400528,
+      "step": 2999
+    },
+    {
+      "epoch": 1.1489850631941785,
+      "grad_norm": 135.98954753552283,
+      "learning_rate": 5e-06,
+      "loss": 1.7787,
+      "num_input_tokens_seen": 518573152,
+      "step": 3000
+    },
+    {
+      "epoch": 1.1489850631941785,
+      "eval_websight_new_CIoU": 0.9289008677005768,
+      "eval_websight_new_GIoU": 0.9288975298404694,
+      "eval_websight_new_IoU": 0.9291035234928131,
+      "eval_websight_new_MAE_all": 0.009797717444598675,
+      "eval_websight_new_MAE_h": 0.01205116230994463,
+      "eval_websight_new_MAE_w": 0.014374295715242624,
+      "eval_websight_new_MAE_x": 0.00619434705004096,
+      "eval_websight_new_MAE_y": 0.006571063073351979,
+      "eval_websight_new_NUM_probability": 0.0017544926377013326,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 1.5373013019561768,
+      "eval_websight_new_loss_ce": 0.7034120559692383,
+      "eval_websight_new_loss_iou": 0.37897779047489166,
+      "eval_websight_new_loss_num": 0.009069442749023438,
+      "eval_websight_new_loss_xval": 0.8032665550708771,
+      "eval_websight_new_runtime": 59.7409,
+      "eval_websight_new_samples_per_second": 0.837,
+      "eval_websight_new_steps_per_second": 0.033,
+      "num_input_tokens_seen": 518573152,
+      "step": 3000
+    },
+    {
+      "epoch": 1.1489850631941785,
+      "eval_seeclick_CIoU": 0.683236688375473,
+      "eval_seeclick_GIoU": 0.6938706934452057,
+      "eval_seeclick_IoU": 0.713070422410965,
+      "eval_seeclick_MAE_all": 0.05435188487172127,
+      "eval_seeclick_MAE_h": 0.040214281529188156,
+      "eval_seeclick_MAE_w": 0.07368129119277,
+      "eval_seeclick_MAE_x": 0.06478147581219673,
+      "eval_seeclick_MAE_y": 0.03873049933463335,
+      "eval_seeclick_NUM_probability": 0.002822827664203942,
+      "eval_seeclick_inside_bbox": 0.9131944477558136,
+      "eval_seeclick_loss": 2.2081284523010254,
+      "eval_seeclick_loss_ce": 0.7061522006988525,
+      "eval_seeclick_loss_iou": 0.6415136754512787,
+      "eval_seeclick_loss_num": 0.0447235107421875,
+      "eval_seeclick_loss_xval": 1.5065380930900574,
+      "eval_seeclick_runtime": 81.9172,
+      "eval_seeclick_samples_per_second": 0.61,
+      "eval_seeclick_steps_per_second": 0.024,
+      "num_input_tokens_seen": 518573152,
+      "step": 3000
+    },
+    {
+      "epoch": 1.1489850631941785,
+      "eval_icons_CIoU": 0.8049757182598114,
+      "eval_icons_GIoU": 0.8091687560081482,
+      "eval_icons_IoU": 0.820679098367691,
+      "eval_icons_MAE_all": 0.03267125133424997,
+      "eval_icons_MAE_h": 0.025516432709991932,
+      "eval_icons_MAE_w": 0.05081920512020588,
+      "eval_icons_MAE_x": 0.04043934307992458,
+      "eval_icons_MAE_y": 0.013910012319684029,
+      "eval_icons_NUM_probability": 0.005454682279378176,
+      "eval_icons_inside_bbox": 0.9288194477558136,
+      "eval_icons_loss": 2.3851919174194336,
+      "eval_icons_loss_ce": 0.6270142197608948,
+      "eval_icons_loss_iou": 0.812638521194458,
+      "eval_icons_loss_num": 0.02117156982421875,
+      "eval_icons_loss_xval": 1.731142520904541,
+      "eval_icons_runtime": 83.6012,
+      "eval_icons_samples_per_second": 0.598,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 518573152,
+      "step": 3000
+    },
+    {
+      "epoch": 1.1489850631941785,
+      "loss": 2.4711475372314453,
+      "loss_ce": 0.6271020174026489,
+      "loss_iou": 0.863520622253418,
+      "loss_num": 0.0234375,
+      "loss_xval": 1.844045639038086,
+      "num_input_tokens_seen": 518573152,
+      "step": 3000
+    },
+    {
+      "epoch": 1.1493680582152432,
+      "grad_norm": 139.09242166658748,
+      "learning_rate": 5e-06,
+      "loss": 2.0149,
+      "num_input_tokens_seen": 518746240,
+      "step": 3001
+    },
+    {
+      "epoch": 1.1493680582152432,
+      "loss": 2.01564621925354,
+      "loss_ce": 0.5103839635848999,
+      "loss_iou": 0.6980351805686951,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 1.5052622556686401,
+      "num_input_tokens_seen": 518746240,
+      "step": 3001
+    },
+    {
+      "epoch": 1.149751053236308,
+      "grad_norm": 131.59173575821038,
+      "learning_rate": 5e-06,
+      "loss": 1.8836,
+      "num_input_tokens_seen": 518919272,
+      "step": 3002
+    },
+    {
+      "epoch": 1.149751053236308,
+      "loss": 1.8277509212493896,
+      "loss_ce": 0.5466691255569458,
+      "loss_iou": 0.5962751507759094,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 1.2810817956924438,
+      "num_input_tokens_seen": 518919272,
+      "step": 3002
+    },
+    {
+      "epoch": 1.1501340482573728,
+      "grad_norm": 108.20560324442643,
+      "learning_rate": 5e-06,
+      "loss": 2.3203,
+      "num_input_tokens_seen": 519092152,
+      "step": 3003
+    },
+    {
+      "epoch": 1.1501340482573728,
+      "loss": 2.4972357749938965,
+      "loss_ce": 0.5610809326171875,
+      "loss_iou": 0.8992298245429993,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 1.9361549615859985,
+      "num_input_tokens_seen": 519092152,
+      "step": 3003
+    },
+    {
+      "epoch": 1.1505170432784373,
+      "grad_norm": 121.10687922785246,
+      "learning_rate": 5e-06,
+      "loss": 1.8939,
+      "num_input_tokens_seen": 519265104,
+      "step": 3004
+    },
+    {
+      "epoch": 1.1505170432784373,
+      "loss": 1.9453966617584229,
+      "loss_ce": 0.5247212648391724,
+      "loss_iou": 0.654505729675293,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 1.420675277709961,
+      "num_input_tokens_seen": 519265104,
+      "step": 3004
+    },
+    {
+      "epoch": 1.150900038299502,
+      "grad_norm": 114.66267510067331,
+      "learning_rate": 5e-06,
+      "loss": 2.1012,
+      "num_input_tokens_seen": 519437728,
+      "step": 3005
+    },
+    {
+      "epoch": 1.150900038299502,
+      "loss": 2.1453614234924316,
+      "loss_ce": 0.5125018358230591,
+      "loss_iou": 0.7602317333221436,
+      "loss_num": 0.0224609375,
+      "loss_xval": 1.632859706878662,
+      "num_input_tokens_seen": 519437728,
+      "step": 3005
+    },
+    {
+      "epoch": 1.1512830333205668,
+      "grad_norm": 102.5620072458507,
+      "learning_rate": 5e-06,
+      "loss": 1.9573,
+      "num_input_tokens_seen": 519610488,
+      "step": 3006
+    },
+    {
+      "epoch": 1.1512830333205668,
+      "loss": 1.7833776473999023,
+      "loss_ce": 0.5496975183486938,
+      "loss_iou": 0.563983678817749,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 1.233680248260498,
+      "num_input_tokens_seen": 519610488,
+      "step": 3006
+    },
+    {
+      "epoch": 1.1516660283416316,
+      "grad_norm": 278.0204483214526,
+      "learning_rate": 5e-06,
+      "loss": 1.9901,
+      "num_input_tokens_seen": 519783440,
+      "step": 3007
+    },
+    {
+      "epoch": 1.1516660283416316,
+      "loss": 1.9606174230575562,
+      "loss_ce": 0.5405219197273254,
+      "loss_iou": 0.6772871017456055,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 1.420095443725586,
+      "num_input_tokens_seen": 519783440,
+      "step": 3007
+    },
+    {
+      "epoch": 1.1520490233626963,
+      "grad_norm": 100.00293202770162,
+      "learning_rate": 5e-06,
+      "loss": 2.4169,
+      "num_input_tokens_seen": 519956480,
+      "step": 3008
+    },
+    {
+      "epoch": 1.1520490233626963,
+      "loss": 2.4275078773498535,
+      "loss_ce": 0.5300484895706177,
+      "loss_iou": 0.8840935230255127,
+      "loss_num": 0.02587890625,
+      "loss_xval": 1.8974595069885254,
+      "num_input_tokens_seen": 519956480,
+      "step": 3008
+    },
+    {
+      "epoch": 1.152432018383761,
+      "grad_norm": 175.15205276773116,
+      "learning_rate": 5e-06,
+      "loss": 2.0254,
+      "num_input_tokens_seen": 520129472,
+      "step": 3009
+    },
+    {
+      "epoch": 1.152432018383761,
+      "loss": 1.9535837173461914,
+      "loss_ce": 0.5508226156234741,
+      "loss_iou": 0.6521862149238586,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 1.4027611017227173,
+      "num_input_tokens_seen": 520129472,
+      "step": 3009
+    },
+    {
+      "epoch": 1.1528150134048256,
+      "grad_norm": 93.02316642342778,
+      "learning_rate": 5e-06,
+      "loss": 2.0402,
+      "num_input_tokens_seen": 520302256,
+      "step": 3010
+    },
+    {
+      "epoch": 1.1528150134048256,
+      "loss": 1.934401273727417,
+      "loss_ce": 0.5041775703430176,
+      "loss_iou": 0.6717768311500549,
+      "loss_num": 0.017333984375,
+      "loss_xval": 1.4177113771438599,
+      "num_input_tokens_seen": 520302256,
+      "step": 3010
+    },
+    {
+      "epoch": 1.1531980084258904,
+      "grad_norm": 142.03973974345791,
+      "learning_rate": 5e-06,
+      "loss": 1.8762,
+      "num_input_tokens_seen": 520475544,
+      "step": 3011
+    },
+    {
+      "epoch": 1.1531980084258904,
+      "loss": 1.8028934001922607,
+      "loss_ce": 0.5216626524925232,
+      "loss_iou": 0.6030939817428589,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 1.2812306880950928,
+      "num_input_tokens_seen": 520475544,
+      "step": 3011
+    },
+    {
+      "epoch": 1.1535810034469551,
+      "grad_norm": 105.14038043841408,
+      "learning_rate": 5e-06,
+      "loss": 2.0907,
+      "num_input_tokens_seen": 520648496,
+      "step": 3012
+    },
+    {
+      "epoch": 1.1535810034469551,
+      "loss": 1.9208060503005981,
+      "loss_ce": 0.5305529832839966,
+      "loss_iou": 0.6518068313598633,
+      "loss_num": 0.017333984375,
+      "loss_xval": 1.3902530670166016,
+      "num_input_tokens_seen": 520648496,
+      "step": 3012
+    },
+    {
+      "epoch": 1.15396399846802,
+      "grad_norm": 112.28781251389441,
+      "learning_rate": 5e-06,
+      "loss": 1.9978,
+      "num_input_tokens_seen": 520821256,
+      "step": 3013
+    },
+    {
+      "epoch": 1.15396399846802,
+      "loss": 2.028007984161377,
+      "loss_ce": 0.5007326006889343,
+      "loss_iou": 0.7139551043510437,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 1.5272754430770874,
+      "num_input_tokens_seen": 520821256,
+      "step": 3013
+    },
+    {
+      "epoch": 1.1543469934890846,
+      "grad_norm": 110.42064520217507,
+      "learning_rate": 5e-06,
+      "loss": 1.9384,
+      "num_input_tokens_seen": 520994744,
+      "step": 3014
+    },
+    {
+      "epoch": 1.1543469934890846,
+      "loss": 1.97543203830719,
+      "loss_ce": 0.5129232406616211,
+      "loss_iou": 0.6709364652633667,
+      "loss_num": 0.024169921875,
+      "loss_xval": 1.4625089168548584,
+      "num_input_tokens_seen": 520994744,
+      "step": 3014
+    },
+    {
+      "epoch": 1.1547299885101494,
+      "grad_norm": 146.10573365097127,
+      "learning_rate": 5e-06,
+      "loss": 2.1011,
+      "num_input_tokens_seen": 521167760,
+      "step": 3015
+    },
+    {
+      "epoch": 1.1547299885101494,
+      "loss": 2.122317314147949,
+      "loss_ce": 0.4962806701660156,
+      "loss_iou": 0.7608027458190918,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 1.6260366439819336,
+      "num_input_tokens_seen": 521167760,
+      "step": 3015
+    },
+    {
+      "epoch": 1.1551129835312142,
+      "grad_norm": 224.75354314705422,
+      "learning_rate": 5e-06,
+      "loss": 2.1782,
+      "num_input_tokens_seen": 521340496,
+      "step": 3016
+    },
+    {
+      "epoch": 1.1551129835312142,
+      "loss": 2.319140911102295,
+      "loss_ce": 0.4974133372306824,
+      "loss_iou": 0.8658657073974609,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 1.8217277526855469,
+      "num_input_tokens_seen": 521340496,
+      "step": 3016
+    },
+    {
+      "epoch": 1.155495978552279,
+      "grad_norm": 72.06937341953629,
+      "learning_rate": 5e-06,
+      "loss": 2.2732,
+      "num_input_tokens_seen": 521513288,
+      "step": 3017
+    },
+    {
+      "epoch": 1.155495978552279,
+      "loss": 2.2797303199768066,
+      "loss_ce": 0.5103136301040649,
+      "loss_iou": 0.8009376525878906,
+      "loss_num": 0.033447265625,
+      "loss_xval": 1.7694168090820312,
+      "num_input_tokens_seen": 521513288,
+      "step": 3017
+    },
+    {
+      "epoch": 1.1558789735733435,
+      "grad_norm": 97.42804989793225,
+      "learning_rate": 5e-06,
+      "loss": 1.844,
+      "num_input_tokens_seen": 521686168,
+      "step": 3018
+    },
+    {
+      "epoch": 1.1558789735733435,
+      "loss": 1.8331570625305176,
+      "loss_ce": 0.5346876382827759,
+      "loss_iou": 0.6024208068847656,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 1.2984695434570312,
+      "num_input_tokens_seen": 521686168,
+      "step": 3018
+    },
+    {
+      "epoch": 1.1562619685944082,
+      "grad_norm": 224.27826224442722,
+      "learning_rate": 5e-06,
+      "loss": 2.6632,
+      "num_input_tokens_seen": 521859192,
+      "step": 3019
+    },
+    {
+      "epoch": 1.1562619685944082,
+      "loss": 2.9461395740509033,
+      "loss_ce": 0.4964212477207184,
+      "loss_iou": 1.1607111692428589,
+      "loss_num": 0.025634765625,
+      "loss_xval": 2.4497182369232178,
+      "num_input_tokens_seen": 521859192,
+      "step": 3019
+    },
+    {
+      "epoch": 1.156644963615473,
+      "grad_norm": 65.17701512819956,
+      "learning_rate": 5e-06,
+      "loss": 2.5659,
+      "num_input_tokens_seen": 522031760,
+      "step": 3020
+    },
+    {
+      "epoch": 1.156644963615473,
+      "loss": 2.4863595962524414,
+      "loss_ce": 0.5229304432868958,
+      "loss_iou": 0.8717292547225952,
+      "loss_num": 0.0439453125,
+      "loss_xval": 1.9634292125701904,
+      "num_input_tokens_seen": 522031760,
+      "step": 3020
+    },
+    {
+      "epoch": 1.1570279586365377,
+      "grad_norm": 121.37290926126369,
+      "learning_rate": 5e-06,
+      "loss": 2.1516,
+      "num_input_tokens_seen": 522204552,
+      "step": 3021
+    },
+    {
+      "epoch": 1.1570279586365377,
+      "loss": 2.1316518783569336,
+      "loss_ce": 0.5568913817405701,
+      "loss_iou": 0.7265586853027344,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 1.5747604370117188,
+      "num_input_tokens_seen": 522204552,
+      "step": 3021
+    },
+    {
+      "epoch": 1.1574109536576025,
+      "grad_norm": 229.35583791473852,
+      "learning_rate": 5e-06,
+      "loss": 2.5877,
+      "num_input_tokens_seen": 522377560,
+      "step": 3022
+    },
+    {
+      "epoch": 1.1574109536576025,
+      "loss": 2.5526371002197266,
+      "loss_ce": 0.5315263271331787,
+      "loss_iou": 0.9546470642089844,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 2.0211105346679688,
+      "num_input_tokens_seen": 522377560,
+      "step": 3022
+    },
+    {
+      "epoch": 1.1577939486786673,
+      "grad_norm": 167.43325952523725,
+      "learning_rate": 5e-06,
+      "loss": 2.6765,
+      "num_input_tokens_seen": 522550464,
+      "step": 3023
+    },
+    {
+      "epoch": 1.1577939486786673,
+      "loss": 2.6495301723480225,
+      "loss_ce": 0.5795937180519104,
+      "loss_iou": 0.9373120069503784,
+      "loss_num": 0.0390625,
+      "loss_xval": 2.069936513900757,
+      "num_input_tokens_seen": 522550464,
+      "step": 3023
+    },
+    {
+      "epoch": 1.1581769436997318,
+      "grad_norm": 157.84680622680904,
+      "learning_rate": 5e-06,
+      "loss": 2.5017,
+      "num_input_tokens_seen": 522723520,
+      "step": 3024
+    },
+    {
+      "epoch": 1.1581769436997318,
+      "loss": 2.427170753479004,
+      "loss_ce": 0.5697979927062988,
+      "loss_iou": 0.8340818881988525,
+      "loss_num": 0.037841796875,
+      "loss_xval": 1.857372760772705,
+      "num_input_tokens_seen": 522723520,
+      "step": 3024
+    },
+    {
+      "epoch": 1.1585599387207965,
+      "grad_norm": 82.11481296858165,
+      "learning_rate": 5e-06,
+      "loss": 2.2963,
+      "num_input_tokens_seen": 522896568,
+      "step": 3025
+    },
+    {
+      "epoch": 1.1585599387207965,
+      "loss": 2.3341455459594727,
+      "loss_ce": 0.583962082862854,
+      "loss_iou": 0.7866516709327698,
+      "loss_num": 0.035400390625,
+      "loss_xval": 1.7501832246780396,
+      "num_input_tokens_seen": 522896568,
+      "step": 3025
+    },
+    {
+      "epoch": 1.1589429337418613,
+      "grad_norm": 103.61180247622377,
+      "learning_rate": 5e-06,
+      "loss": 1.9232,
+      "num_input_tokens_seen": 523069224,
+      "step": 3026
+    },
+    {
+      "epoch": 1.1589429337418613,
+      "loss": 1.9566779136657715,
+      "loss_ce": 0.5542987585067749,
+      "loss_iou": 0.6405511498451233,
+      "loss_num": 0.024169921875,
+      "loss_xval": 1.4023791551589966,
+      "num_input_tokens_seen": 523069224,
+      "step": 3026
+    },
+    {
+      "epoch": 1.159325928762926,
+      "grad_norm": 287.0764509963293,
+      "learning_rate": 5e-06,
+      "loss": 2.9193,
+      "num_input_tokens_seen": 523242384,
+      "step": 3027
+    },
+    {
+      "epoch": 1.159325928762926,
+      "loss": 2.9638757705688477,
+      "loss_ce": 0.5367143154144287,
+      "loss_iou": 1.1446415185928345,
+      "loss_num": 0.027587890625,
+      "loss_xval": 2.427161455154419,
+      "num_input_tokens_seen": 523242384,
+      "step": 3027
+    },
+    {
+      "epoch": 1.1597089237839908,
+      "grad_norm": 99.91399363308518,
+      "learning_rate": 5e-06,
+      "loss": 2.3371,
+      "num_input_tokens_seen": 523415472,
+      "step": 3028
+    },
+    {
+      "epoch": 1.1597089237839908,
+      "loss": 2.3318090438842773,
+      "loss_ce": 0.6090725660324097,
+      "loss_iou": 0.7692052125930786,
+      "loss_num": 0.036865234375,
+      "loss_xval": 1.7227365970611572,
+      "num_input_tokens_seen": 523415472,
+      "step": 3028
+    },
+    {
+      "epoch": 1.1600919188050556,
+      "grad_norm": 95.4680485424965,
+      "learning_rate": 5e-06,
+      "loss": 2.0019,
+      "num_input_tokens_seen": 523588216,
+      "step": 3029
+    },
+    {
+      "epoch": 1.1600919188050556,
+      "loss": 1.8098641633987427,
+      "loss_ce": 0.5736472606658936,
+      "loss_iou": 0.5691583156585693,
+      "loss_num": 0.01953125,
+      "loss_xval": 1.2362170219421387,
+      "num_input_tokens_seen": 523588216,
+      "step": 3029
+    },
+    {
+      "epoch": 1.1604749138261203,
+      "grad_norm": 117.48196063963505,
+      "learning_rate": 5e-06,
+      "loss": 2.0943,
+      "num_input_tokens_seen": 523761080,
+      "step": 3030
+    },
+    {
+      "epoch": 1.1604749138261203,
+      "loss": 2.036898612976074,
+      "loss_ce": 0.582169234752655,
+      "loss_iou": 0.6922236680984497,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 1.4547293186187744,
+      "num_input_tokens_seen": 523761080,
+      "step": 3030
+    },
+    {
+      "epoch": 1.160857908847185,
+      "grad_norm": 201.2175189580315,
+      "learning_rate": 5e-06,
+      "loss": 2.9544,
+      "num_input_tokens_seen": 523934072,
+      "step": 3031
+    },
+    {
+      "epoch": 1.160857908847185,
+      "loss": 2.9619741439819336,
+      "loss_ce": 0.6040058135986328,
+      "loss_iou": 1.1072983741760254,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 2.357968330383301,
+      "num_input_tokens_seen": 523934072,
+      "step": 3031
+    },
+    {
+      "epoch": 1.1612409038682496,
+      "grad_norm": 146.70510454521533,
+      "learning_rate": 5e-06,
+      "loss": 2.2877,
+      "num_input_tokens_seen": 524107192,
+      "step": 3032
+    },
+    {
+      "epoch": 1.1612409038682496,
+      "loss": 2.3786630630493164,
+      "loss_ce": 0.5580625534057617,
+      "loss_iou": 0.8286963105201721,
+      "loss_num": 0.03271484375,
+      "loss_xval": 1.8206006288528442,
+      "num_input_tokens_seen": 524107192,
+      "step": 3032
+    },
+    {
+      "epoch": 1.1616238988893144,
+      "grad_norm": 232.97661371340706,
+      "learning_rate": 5e-06,
+      "loss": 1.93,
+      "num_input_tokens_seen": 524280192,
+      "step": 3033
+    },
+    {
+      "epoch": 1.1616238988893144,
+      "loss": 1.9253848791122437,
+      "loss_ce": 0.5511703491210938,
+      "loss_iou": 0.6458474397659302,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 1.3742144107818604,
+      "num_input_tokens_seen": 524280192,
+      "step": 3033
+    },
+    {
+      "epoch": 1.1620068939103791,
+      "grad_norm": 103.81499572093095,
+      "learning_rate": 5e-06,
+      "loss": 2.2417,
+      "num_input_tokens_seen": 524453144,
+      "step": 3034
+    },
+    {
+      "epoch": 1.1620068939103791,
+      "loss": 2.3110437393188477,
+      "loss_ce": 0.5635131597518921,
+      "loss_iou": 0.8246625065803528,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 1.7475305795669556,
+      "num_input_tokens_seen": 524453144,
+      "step": 3034
+    },
+    {
+      "epoch": 1.162389888931444,
+      "grad_norm": 193.80084892592967,
+      "learning_rate": 5e-06,
+      "loss": 2.2382,
+      "num_input_tokens_seen": 524625928,
+      "step": 3035
+    },
+    {
+      "epoch": 1.162389888931444,
+      "loss": 2.1590077877044678,
+      "loss_ce": 0.5422309637069702,
+      "loss_iou": 0.7368246912956238,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 1.6167768239974976,
+      "num_input_tokens_seen": 524625928,
+      "step": 3035
+    },
+    {
+      "epoch": 1.1627728839525087,
+      "grad_norm": 223.50433193857575,
+      "learning_rate": 5e-06,
+      "loss": 1.9806,
+      "num_input_tokens_seen": 524798696,
+      "step": 3036
+    },
+    {
+      "epoch": 1.1627728839525087,
+      "loss": 2.0209784507751465,
+      "loss_ce": 0.5899786949157715,
+      "loss_iou": 0.6683807373046875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 1.430999755859375,
+      "num_input_tokens_seen": 524798696,
+      "step": 3036
+    },
+    {
+      "epoch": 1.1631558789735734,
+      "grad_norm": 127.91361924924034,
+      "learning_rate": 5e-06,
+      "loss": 2.0166,
+      "num_input_tokens_seen": 524971760,
+      "step": 3037
+    },
+    {
+      "epoch": 1.1631558789735734,
+      "loss": 1.9538748264312744,
+      "loss_ce": 0.5548742413520813,
+      "loss_iou": 0.6514350771903992,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 1.3990005254745483,
+      "num_input_tokens_seen": 524971760,
+      "step": 3037
+    },
+    {
+      "epoch": 1.163538873994638,
+      "grad_norm": 108.51388885330405,
+      "learning_rate": 5e-06,
+      "loss": 2.089,
+      "num_input_tokens_seen": 525144952,
+      "step": 3038
+    },
+    {
+      "epoch": 1.163538873994638,
+      "loss": 2.0692315101623535,
+      "loss_ce": 0.5200036764144897,
+      "loss_iou": 0.7181258201599121,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 1.5492277145385742,
+      "num_input_tokens_seen": 525144952,
+      "step": 3038
+    },
+    {
+      "epoch": 1.1639218690157027,
+      "grad_norm": 112.63384375864676,
+      "learning_rate": 5e-06,
+      "loss": 1.8902,
+      "num_input_tokens_seen": 525317936,
+      "step": 3039
+    },
+    {
+      "epoch": 1.1639218690157027,
+      "loss": 1.8628909587860107,
+      "loss_ce": 0.5142827033996582,
+      "loss_iou": 0.6378051042556763,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 1.3486082553863525,
+      "num_input_tokens_seen": 525317936,
+      "step": 3039
+    },
+    {
+      "epoch": 1.1643048640367675,
+      "grad_norm": 134.6131501116591,
+      "learning_rate": 5e-06,
+      "loss": 1.7304,
+      "num_input_tokens_seen": 525490816,
+      "step": 3040
+    },
+    {
+      "epoch": 1.1643048640367675,
+      "loss": 1.8034213781356812,
+      "loss_ce": 0.5231952667236328,
+      "loss_iou": 0.6067954897880554,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 1.2802261114120483,
+      "num_input_tokens_seen": 525490816,
+      "step": 3040
+    },
+    {
+      "epoch": 1.1646878590578322,
+      "grad_norm": 114.7384873154603,
+      "learning_rate": 5e-06,
+      "loss": 2.1501,
+      "num_input_tokens_seen": 525663824,
+      "step": 3041
+    },
+    {
+      "epoch": 1.1646878590578322,
+      "loss": 2.2172675132751465,
+      "loss_ce": 0.5395190715789795,
+      "loss_iou": 0.7881540060043335,
+      "loss_num": 0.020263671875,
+      "loss_xval": 1.677748441696167,
+      "num_input_tokens_seen": 525663824,
+      "step": 3041
+    },
+    {
+      "epoch": 1.165070854078897,
+      "grad_norm": 86.38432166069843,
+      "learning_rate": 5e-06,
+      "loss": 1.8492,
+      "num_input_tokens_seen": 525836488,
+      "step": 3042
+    },
+    {
+      "epoch": 1.165070854078897,
+      "loss": 1.7687615156173706,
+      "loss_ce": 0.553092360496521,
+      "loss_iou": 0.56175297498703,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 1.19821298122406,
+      "num_input_tokens_seen": 525836488,
+      "step": 3042
+    },
+    {
+      "epoch": 1.1654538490999617,
+      "grad_norm": 85.05003253130162,
+      "learning_rate": 5e-06,
+      "loss": 1.8564,
+      "num_input_tokens_seen": 526009024,
+      "step": 3043
+    },
+    {
+      "epoch": 1.1654538490999617,
+      "loss": 2.0090861320495605,
+      "loss_ce": 0.5071377158164978,
+      "loss_iou": 0.7066322565078735,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 1.501948595046997,
+      "num_input_tokens_seen": 526009024,
+      "step": 3043
+    },
+    {
+      "epoch": 1.1658368441210265,
+      "grad_norm": 197.92156942534768,
+      "learning_rate": 5e-06,
+      "loss": 2.0245,
+      "num_input_tokens_seen": 526181768,
+      "step": 3044
+    },
+    {
+      "epoch": 1.1658368441210265,
+      "loss": 1.8668968677520752,
+      "loss_ce": 0.5239740610122681,
+      "loss_iou": 0.6356947422027588,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 1.3429226875305176,
+      "num_input_tokens_seen": 526181768,
+      "step": 3044
+    },
+    {
+      "epoch": 1.1662198391420913,
+      "grad_norm": 99.4182298535617,
+      "learning_rate": 5e-06,
+      "loss": 2.4788,
+      "num_input_tokens_seen": 526354976,
+      "step": 3045
+    },
+    {
+      "epoch": 1.1662198391420913,
+      "loss": 2.455658435821533,
+      "loss_ce": 0.5290402173995972,
+      "loss_iou": 0.8945225477218628,
+      "loss_num": 0.027587890625,
+      "loss_xval": 1.9266183376312256,
+      "num_input_tokens_seen": 526354976,
+      "step": 3045
+    },
+    {
+      "epoch": 1.1666028341631558,
+      "grad_norm": 86.57337881425134,
+      "learning_rate": 5e-06,
+      "loss": 2.1731,
+      "num_input_tokens_seen": 526527656,
+      "step": 3046
+    },
+    {
+      "epoch": 1.1666028341631558,
+      "loss": 2.1833906173706055,
+      "loss_ce": 0.4955127239227295,
+      "loss_iou": 0.774236798286438,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 1.687877893447876,
+      "num_input_tokens_seen": 526527656,
+      "step": 3046
+    },
+    {
+      "epoch": 1.1669858291842206,
+      "grad_norm": 139.61278862580969,
+      "learning_rate": 5e-06,
+      "loss": 1.9832,
+      "num_input_tokens_seen": 526700408,
+      "step": 3047
+    },
+    {
+      "epoch": 1.1669858291842206,
+      "loss": 1.8087213039398193,
+      "loss_ce": 0.5244953632354736,
+      "loss_iou": 0.5974199771881104,
+      "loss_num": 0.017822265625,
+      "loss_xval": 1.2842259407043457,
+      "num_input_tokens_seen": 526700408,
+      "step": 3047
+    },
+    {
+      "epoch": 1.1673688242052853,
+      "grad_norm": 137.95949675032745,
+      "learning_rate": 5e-06,
+      "loss": 2.1025,
+      "num_input_tokens_seen": 526873352,
+      "step": 3048
+    },
+    {
+      "epoch": 1.1673688242052853,
+      "loss": 2.092449188232422,
+      "loss_ce": 0.5285845994949341,
+      "loss_iou": 0.7295947074890137,
+      "loss_num": 0.02099609375,
+      "loss_xval": 1.5638647079467773,
+      "num_input_tokens_seen": 526873352,
+      "step": 3048
+    },
+    {
+      "epoch": 1.16775181922635,
+      "grad_norm": 91.15195807997374,
+      "learning_rate": 5e-06,
+      "loss": 1.9721,
+      "num_input_tokens_seen": 527046472,
+      "step": 3049
+    },
+    {
+      "epoch": 1.16775181922635,
+      "loss": 2.130580186843872,
+      "loss_ce": 0.5217145085334778,
+      "loss_iou": 0.7446946501731873,
+      "loss_num": 0.02392578125,
+      "loss_xval": 1.6088656187057495,
+      "num_input_tokens_seen": 527046472,
+      "step": 3049
+    },
+    {
+      "epoch": 1.1681348142474148,
+      "grad_norm": 91.44638543843973,
+      "learning_rate": 5e-06,
+      "loss": 1.9778,
+      "num_input_tokens_seen": 527219440,
+      "step": 3050
+    },
+    {
+      "epoch": 1.1681348142474148,
+      "loss": 1.8737220764160156,
+      "loss_ce": 0.5123358964920044,
+      "loss_iou": 0.6441329717636108,
+      "loss_num": 0.0146484375,
+      "loss_xval": 1.3613860607147217,
+      "num_input_tokens_seen": 527219440,
+      "step": 3050
+    },
+    {
+      "epoch": 1.1685178092684796,
+      "grad_norm": 176.32139285541254,
+      "learning_rate": 5e-06,
+      "loss": 2.0724,
+      "num_input_tokens_seen": 527392064,
+      "step": 3051
+    },
+    {
+      "epoch": 1.1685178092684796,
+      "loss": 2.222956657409668,
+      "loss_ce": 0.5154544711112976,
+      "loss_iou": 0.7708501815795898,
+      "loss_num": 0.033203125,
+      "loss_xval": 1.7075023651123047,
+      "num_input_tokens_seen": 527392064,
+      "step": 3051
+    },
+    {
+      "epoch": 1.1689008042895441,
+      "grad_norm": 99.75705022812105,
+      "learning_rate": 5e-06,
+      "loss": 2.3729,
+      "num_input_tokens_seen": 527565152,
+      "step": 3052
+    },
+    {
+      "epoch": 1.1689008042895441,
+      "loss": 2.274721384048462,
+      "loss_ce": 0.5184080600738525,
+      "loss_iou": 0.8150157928466797,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 1.7563133239746094,
+      "num_input_tokens_seen": 527565152,
+      "step": 3052
+    },
+    {
+      "epoch": 1.1692837993106089,
+      "grad_norm": 115.9429975894583,
+      "learning_rate": 5e-06,
+      "loss": 2.1547,
+      "num_input_tokens_seen": 527738584,
+      "step": 3053
+    },
+    {
+      "epoch": 1.1692837993106089,
+      "loss": 2.240044355392456,
+      "loss_ce": 0.5178654789924622,
+      "loss_iou": 0.7991082668304443,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 1.7221789360046387,
+      "num_input_tokens_seen": 527738584,
+      "step": 3053
+    },
+    {
+      "epoch": 1.1696667943316736,
+      "grad_norm": 146.86466547060593,
+      "learning_rate": 5e-06,
+      "loss": 1.9783,
+      "num_input_tokens_seen": 527911520,
+      "step": 3054
+    },
+    {
+      "epoch": 1.1696667943316736,
+      "loss": 1.9764864444732666,
+      "loss_ce": 0.4814373850822449,
+      "loss_iou": 0.6908839344978333,
+      "loss_num": 0.022705078125,
+      "loss_xval": 1.4950491189956665,
+      "num_input_tokens_seen": 527911520,
+      "step": 3054
+    },
+    {
+      "epoch": 1.1700497893527384,
+      "grad_norm": 161.57457181911457,
+      "learning_rate": 5e-06,
+      "loss": 2.138,
+      "num_input_tokens_seen": 528084400,
+      "step": 3055
+    },
+    {
+      "epoch": 1.1700497893527384,
+      "loss": 2.122333288192749,
+      "loss_ce": 0.4733904004096985,
+      "loss_iou": 0.7757652997970581,
+      "loss_num": 0.01953125,
+      "loss_xval": 1.6489427089691162,
+      "num_input_tokens_seen": 528084400,
+      "step": 3055
+    },
+    {
+      "epoch": 1.1704327843738032,
+      "grad_norm": 166.2849525229474,
+      "learning_rate": 5e-06,
+      "loss": 2.0086,
+      "num_input_tokens_seen": 528257360,
+      "step": 3056
+    },
+    {
+      "epoch": 1.1704327843738032,
+      "loss": 1.9678535461425781,
+      "loss_ce": 0.5155329704284668,
+      "loss_iou": 0.6903784275054932,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 1.4523205757141113,
+      "num_input_tokens_seen": 528257360,
+      "step": 3056
+    },
+    {
+      "epoch": 1.170815779394868,
+      "grad_norm": 134.07302002271393,
+      "learning_rate": 5e-06,
+      "loss": 2.0798,
+      "num_input_tokens_seen": 528430256,
+      "step": 3057
+    },
+    {
+      "epoch": 1.170815779394868,
+      "loss": 2.149479866027832,
+      "loss_ce": 0.4850102365016937,
+      "loss_iou": 0.7758992910385132,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 1.6644694805145264,
+      "num_input_tokens_seen": 528430256,
+      "step": 3057
+    },
+    {
+      "epoch": 1.1711987744159327,
+      "grad_norm": 121.44923981604792,
+      "learning_rate": 5e-06,
+      "loss": 2.1475,
+      "num_input_tokens_seen": 528603048,
+      "step": 3058
+    },
+    {
+      "epoch": 1.1711987744159327,
+      "loss": 2.1266353130340576,
+      "loss_ce": 0.489056795835495,
+      "loss_iou": 0.7569606900215149,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 1.6375786066055298,
+      "num_input_tokens_seen": 528603048,
+      "step": 3058
+    },
+    {
+      "epoch": 1.1715817694369974,
+      "grad_norm": 107.73166820449426,
+      "learning_rate": 5e-06,
+      "loss": 1.7975,
+      "num_input_tokens_seen": 528776240,
+      "step": 3059
+    },
+    {
+      "epoch": 1.1715817694369974,
+      "loss": 1.757315993309021,
+      "loss_ce": 0.5063617825508118,
+      "loss_iou": 0.591266930103302,
+      "loss_num": 0.013671875,
+      "loss_xval": 1.250954270362854,
+      "num_input_tokens_seen": 528776240,
+      "step": 3059
+    },
+    {
+      "epoch": 1.171964764458062,
+      "grad_norm": 142.87362295853583,
+      "learning_rate": 5e-06,
+      "loss": 1.6761,
+      "num_input_tokens_seen": 528949016,
+      "step": 3060
+    },
+    {
+      "epoch": 1.171964764458062,
+      "loss": 1.6243350505828857,
+      "loss_ce": 0.4753933548927307,
+      "loss_iou": 0.5394213795661926,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 1.1489416360855103,
+      "num_input_tokens_seen": 528949016,
+      "step": 3060
+    },
+    {
+      "epoch": 1.1723477594791267,
+      "grad_norm": 173.64181803167602,
+      "learning_rate": 5e-06,
+      "loss": 2.124,
+      "num_input_tokens_seen": 529122504,
+      "step": 3061
+    },
+    {
+      "epoch": 1.1723477594791267,
+      "loss": 2.029665946960449,
+      "loss_ce": 0.4836769104003906,
+      "loss_iou": 0.7265772819519043,
+      "loss_num": 0.0185546875,
+      "loss_xval": 1.5459890365600586,
+      "num_input_tokens_seen": 529122504,
+      "step": 3061
+    },
+    {
+      "epoch": 1.1727307545001915,
+      "grad_norm": 125.77741119958662,
+      "learning_rate": 5e-06,
+      "loss": 2.4215,
+      "num_input_tokens_seen": 529295208,
+      "step": 3062
+    },
+    {
+      "epoch": 1.1727307545001915,
+      "loss": 2.460716724395752,
+      "loss_ce": 0.47949209809303284,
+      "loss_iou": 0.8830683827400208,
+      "loss_num": 0.04296875,
+      "loss_xval": 1.9812246561050415,
+      "num_input_tokens_seen": 529295208,
+      "step": 3062
+    },
+    {
+      "epoch": 1.1731137495212562,
+      "grad_norm": 109.1398471362209,
+      "learning_rate": 5e-06,
+      "loss": 2.137,
+      "num_input_tokens_seen": 529467976,
+      "step": 3063
+    },
+    {
+      "epoch": 1.1731137495212562,
+      "loss": 2.1569390296936035,
+      "loss_ce": 0.4654855728149414,
+      "loss_iou": 0.7971580028533936,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 1.691453456878662,
+      "num_input_tokens_seen": 529467976,
+      "step": 3063
+    },
+    {
+      "epoch": 1.173496744542321,
+      "grad_norm": 196.19940314823938,
+      "learning_rate": 5e-06,
+      "loss": 2.4079,
+      "num_input_tokens_seen": 529640848,
+      "step": 3064
+    },
+    {
+      "epoch": 1.173496744542321,
+      "loss": 2.412062644958496,
+      "loss_ce": 0.49375009536743164,
+      "loss_iou": 0.9106332659721375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 1.918312430381775,
+      "num_input_tokens_seen": 529640848,
+      "step": 3064
+    },
+    {
+      "epoch": 1.1738797395633858,
+      "grad_norm": 120.60717314557823,
+      "learning_rate": 5e-06,
+      "loss": 2.4104,
+      "num_input_tokens_seen": 529813776,
+      "step": 3065
+    },
+    {
+      "epoch": 1.1738797395633858,
+      "loss": 2.3200154304504395,
+      "loss_ce": 0.5093563795089722,
+      "loss_iou": 0.8057507276535034,
+      "loss_num": 0.039794921875,
+      "loss_xval": 1.8106591701507568,
+      "num_input_tokens_seen": 529813776,
+      "step": 3065
+    },
+    {
+      "epoch": 1.1742627345844503,
+      "grad_norm": 88.43849490081806,
+      "learning_rate": 5e-06,
+      "loss": 2.2045,
+      "num_input_tokens_seen": 529986888,
+      "step": 3066
+    },
+    {
+      "epoch": 1.1742627345844503,
+      "loss": 2.1570796966552734,
+      "loss_ce": 0.5062353014945984,
+      "loss_iou": 0.7442454695701599,
+      "loss_num": 0.032470703125,
+      "loss_xval": 1.6508444547653198,
+      "num_input_tokens_seen": 529986888,
+      "step": 3066
+    },
+    {
+      "epoch": 1.174645729605515,
+      "grad_norm": 259.62984874746115,
+      "learning_rate": 5e-06,
+      "loss": 2.0067,
+      "num_input_tokens_seen": 530159496,
+      "step": 3067
+    },
+    {
+      "epoch": 1.174645729605515,
+      "loss": 1.8444569110870361,
+      "loss_ce": 0.49705448746681213,
+      "loss_iou": 0.6307781934738159,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 1.3474023342132568,
+      "num_input_tokens_seen": 530159496,
+      "step": 3067
+    },
+    {
+      "epoch": 1.1750287246265798,
+      "grad_norm": 72.49031287732348,
+      "learning_rate": 5e-06,
+      "loss": 2.1604,
+      "num_input_tokens_seen": 530328504,
+      "step": 3068
+    },
+    {
+      "epoch": 1.1750287246265798,
+      "loss": 2.269585132598877,
+      "loss_ce": 0.4805722236633301,
+      "loss_iou": 0.8140620589256287,
+      "loss_num": 0.0322265625,
+      "loss_xval": 1.7890127897262573,
+      "num_input_tokens_seen": 530328504,
+      "step": 3068
+    },
+    {
+      "epoch": 1.1754117196476446,
+      "grad_norm": 110.81024774100332,
+      "learning_rate": 5e-06,
+      "loss": 2.1665,
+      "num_input_tokens_seen": 530501384,
+      "step": 3069
+    },
+    {
+      "epoch": 1.1754117196476446,
+      "loss": 2.2385213375091553,
+      "loss_ce": 0.4997120797634125,
+      "loss_iou": 0.81157386302948,
+      "loss_num": 0.023193359375,
+      "loss_xval": 1.73880934715271,
+      "num_input_tokens_seen": 530501384,
+      "step": 3069
+    },
+    {
+      "epoch": 1.1757947146687093,
+      "grad_norm": 292.41156610271406,
+      "learning_rate": 5e-06,
+      "loss": 2.4211,
+      "num_input_tokens_seen": 530674736,
+      "step": 3070
+    },
+    {
+      "epoch": 1.1757947146687093,
+      "loss": 2.433666944503784,
+      "loss_ce": 0.5467498302459717,
+      "loss_iou": 0.8623733520507812,
+      "loss_num": 0.032470703125,
+      "loss_xval": 1.8869171142578125,
+      "num_input_tokens_seen": 530674736,
+      "step": 3070
+    },
+    {
+      "epoch": 1.176177709689774,
+      "grad_norm": 181.99659514108347,
+      "learning_rate": 5e-06,
+      "loss": 2.5487,
+      "num_input_tokens_seen": 530847776,
+      "step": 3071
+    },
+    {
+      "epoch": 1.176177709689774,
+      "loss": 2.5342164039611816,
+      "loss_ce": 0.5647098422050476,
+      "loss_iou": 0.901562511920929,
+      "loss_num": 0.033203125,
+      "loss_xval": 1.969506859779358,
+      "num_input_tokens_seen": 530847776,
+      "step": 3071
+    },
+    {
+      "epoch": 1.1765607047108388,
+      "grad_norm": 330.34257726631733,
+      "learning_rate": 5e-06,
+      "loss": 2.5627,
+      "num_input_tokens_seen": 531020992,
+      "step": 3072
+    },
+    {
+      "epoch": 1.1765607047108388,
+      "loss": 2.55936336517334,
+      "loss_ce": 0.5791891813278198,
+      "loss_iou": 0.9034783244132996,
+      "loss_num": 0.03466796875,
+      "loss_xval": 1.9801744222640991,
+      "num_input_tokens_seen": 531020992,
+      "step": 3072
+    },
+    {
+      "epoch": 1.1769436997319036,
+      "grad_norm": 59.011670424016785,
+      "learning_rate": 5e-06,
+      "loss": 2.4984,
+      "num_input_tokens_seen": 531194504,
+      "step": 3073
+    },
+    {
+      "epoch": 1.1769436997319036,
+      "loss": 2.4518914222717285,
+      "loss_ce": 0.5527679920196533,
+      "loss_iou": 0.8246227502822876,
+      "loss_num": 0.050048828125,
+      "loss_xval": 1.8991234302520752,
+      "num_input_tokens_seen": 531194504,
+      "step": 3073
+    },
+    {
+      "epoch": 1.1773266947529681,
+      "grad_norm": 101.3018967292303,
+      "learning_rate": 5e-06,
+      "loss": 2.2879,
+      "num_input_tokens_seen": 531367744,
+      "step": 3074
+    },
+    {
+      "epoch": 1.1773266947529681,
+      "loss": 2.198519468307495,
+      "loss_ce": 0.582373857498169,
+      "loss_iou": 0.7422463893890381,
+      "loss_num": 0.0263671875,
+      "loss_xval": 1.6161456108093262,
+      "num_input_tokens_seen": 531367744,
+      "step": 3074
+    },
+    {
+      "epoch": 1.177709689774033,
+      "grad_norm": 85.82880948447881,
+      "learning_rate": 5e-06,
+      "loss": 1.9941,
+      "num_input_tokens_seen": 531540552,
+      "step": 3075
+    },
+    {
+      "epoch": 1.177709689774033,
+      "loss": 1.8359382152557373,
+      "loss_ce": 0.5748423337936401,
+      "loss_iou": 0.5849394798278809,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 1.2610960006713867,
+      "num_input_tokens_seen": 531540552,
+      "step": 3075
+    },
+    {
+      "epoch": 1.1780926847950977,
+      "grad_norm": 127.92420307754955,
+      "learning_rate": 5e-06,
+      "loss": 2.4393,
+      "num_input_tokens_seen": 531713392,
+      "step": 3076
+    },
+    {
+      "epoch": 1.1780926847950977,
+      "loss": 2.404845952987671,
+      "loss_ce": 0.545185387134552,
+      "loss_iou": 0.8741356134414673,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 1.8596603870391846,
+      "num_input_tokens_seen": 531713392,
+      "step": 3076
+    },
+    {
+      "epoch": 1.1784756798161624,
+      "grad_norm": 55.50579941902585,
+      "learning_rate": 5e-06,
+      "loss": 2.0792,
+      "num_input_tokens_seen": 531885920,
+      "step": 3077
+    },
+    {
+      "epoch": 1.1784756798161624,
+      "loss": 1.95820951461792,
+      "loss_ce": 0.546619713306427,
+      "loss_iou": 0.6498562097549438,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 1.4115898609161377,
+      "num_input_tokens_seen": 531885920,
+      "step": 3077
+    },
+    {
+      "epoch": 1.1788586748372272,
+      "grad_norm": 118.59406570043875,
+      "learning_rate": 5e-06,
+      "loss": 1.84,
+      "num_input_tokens_seen": 532058696,
+      "step": 3078
+    },
+    {
+      "epoch": 1.1788586748372272,
+      "loss": 1.9588350057601929,
+      "loss_ce": 0.5443446040153503,
+      "loss_iou": 0.6634677052497864,
+      "loss_num": 0.017578125,
+      "loss_xval": 1.4144903421401978,
+      "num_input_tokens_seen": 532058696,
+      "step": 3078
+    },
+    {
+      "epoch": 1.179241669858292,
+      "grad_norm": 127.6107443593757,
+      "learning_rate": 5e-06,
+      "loss": 2.0947,
+      "num_input_tokens_seen": 532231656,
+      "step": 3079
+    },
+    {
+      "epoch": 1.179241669858292,
+      "loss": 1.9777263402938843,
+      "loss_ce": 0.5301843285560608,
+      "loss_iou": 0.6778877973556519,
+      "loss_num": 0.018310546875,
+      "loss_xval": 1.4475419521331787,
+      "num_input_tokens_seen": 532231656,
+      "step": 3079
+    },
+    {
+      "epoch": 1.1796246648793565,
+      "grad_norm": 208.10514886009696,
+      "learning_rate": 5e-06,
+      "loss": 1.9466,
+      "num_input_tokens_seen": 532404432,
+      "step": 3080
+    },
+    {
+      "epoch": 1.1796246648793565,
+      "loss": 1.9196019172668457,
+      "loss_ce": 0.541554868221283,
+      "loss_iou": 0.644559383392334,
+      "loss_num": 0.017822265625,
+      "loss_xval": 1.378046989440918,
+      "num_input_tokens_seen": 532404432,
+      "step": 3080
+    },
+    {
+      "epoch": 1.1800076599004212,
+      "grad_norm": 138.20639601401902,
+      "learning_rate": 5e-06,
+      "loss": 2.2284,
+      "num_input_tokens_seen": 532577672,
+      "step": 3081
+    },
+    {
+      "epoch": 1.1800076599004212,
+      "loss": 2.2285828590393066,
+      "loss_ce": 0.5263062119483948,
+      "loss_iou": 0.7832977771759033,
+      "loss_num": 0.027099609375,
+      "loss_xval": 1.7022767066955566,
+      "num_input_tokens_seen": 532577672,
+      "step": 3081
+    },
+    {
+      "epoch": 1.180390654921486,
+      "grad_norm": 106.42366775498245,
+      "learning_rate": 5e-06,
+      "loss": 1.7875,
+      "num_input_tokens_seen": 532750640,
+      "step": 3082
+    },
+    {
+      "epoch": 1.180390654921486,
+      "loss": 1.6181203126907349,
+      "loss_ce": 0.5389182567596436,
+      "loss_iou": 0.5006301403045654,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 1.0792021751403809,
+      "num_input_tokens_seen": 532750640,
+      "step": 3082
+    },
+    {
+      "epoch": 1.1807736499425507,
+      "grad_norm": 174.05157276395775,
+      "learning_rate": 5e-06,
+      "loss": 1.9299,
+      "num_input_tokens_seen": 532923888,
+      "step": 3083
+    },
+    {
+      "epoch": 1.1807736499425507,
+      "loss": 1.84665846824646,
+      "loss_ce": 0.5498446226119995,
+      "loss_iou": 0.6193541884422302,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 1.2968138456344604,
+      "num_input_tokens_seen": 532923888,
+      "step": 3083
+    },
+    {
+      "epoch": 1.1811566449636155,
+      "grad_norm": 78.5711205346916,
+      "learning_rate": 5e-06,
+      "loss": 2.1162,
+      "num_input_tokens_seen": 533096768,
+      "step": 3084
+    },
+    {
+      "epoch": 1.1811566449636155,
+      "loss": 2.3008179664611816,
+      "loss_ce": 0.5162010788917542,
+      "loss_iou": 0.8209277987480164,
+      "loss_num": 0.028564453125,
+      "loss_xval": 1.7846168279647827,
+      "num_input_tokens_seen": 533096768,
+      "step": 3084
+    },
+    {
+      "epoch": 1.1815396399846803,
+      "grad_norm": 69.41066415665922,
+      "learning_rate": 5e-06,
+      "loss": 1.9532,
+      "num_input_tokens_seen": 533269672,
+      "step": 3085
+    },
+    {
+      "epoch": 1.1815396399846803,
+      "loss": 1.8059289455413818,
+      "loss_ce": 0.5179780721664429,
+      "loss_iou": 0.600259006023407,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 1.2879507541656494,
+      "num_input_tokens_seen": 533269672,
+      "step": 3085
+    },
+    {
+      "epoch": 1.181922635005745,
+      "grad_norm": 213.4351583881707,
+      "learning_rate": 5e-06,
+      "loss": 2.1173,
+      "num_input_tokens_seen": 533442752,
+      "step": 3086
+    },
+    {
+      "epoch": 1.181922635005745,
+      "loss": 2.108553886413574,
+      "loss_ce": 0.5023859143257141,
+      "loss_iou": 0.7601457238197327,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 1.6061679124832153,
+      "num_input_tokens_seen": 533442752,
+      "step": 3086
+    },
+    {
+      "epoch": 1.1823056300268098,
+      "grad_norm": 170.57106458149943,
+      "learning_rate": 5e-06,
+      "loss": 2.1106,
+      "num_input_tokens_seen": 533615552,
+      "step": 3087
+    },
+    {
+      "epoch": 1.1823056300268098,
+      "loss": 2.032696008682251,
+      "loss_ce": 0.5327855348587036,
+      "loss_iou": 0.7113352417945862,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 1.4999104738235474,
+      "num_input_tokens_seen": 533615552,
+      "step": 3087
+    },
+    {
+      "epoch": 1.1826886250478743,
+      "grad_norm": 422.2572321374377,
+      "learning_rate": 5e-06,
+      "loss": 2.6274,
+      "num_input_tokens_seen": 533788624,
+      "step": 3088
+    },
+    {
+      "epoch": 1.1826886250478743,
+      "loss": 2.3464598655700684,
+      "loss_ce": 0.49449825286865234,
+      "loss_iou": 0.8754742741584778,
+      "loss_num": 0.020263671875,
+      "loss_xval": 1.8519617319107056,
+      "num_input_tokens_seen": 533788624,
+      "step": 3088
+    },
+    {
+      "epoch": 1.183071620068939,
+      "grad_norm": 95.19896558821257,
+      "learning_rate": 5e-06,
+      "loss": 2.3712,
+      "num_input_tokens_seen": 533961888,
+      "step": 3089
+    },
+    {
+      "epoch": 1.183071620068939,
+      "loss": 2.4336447715759277,
+      "loss_ce": 0.5236128568649292,
+      "loss_iou": 0.871397852897644,
+      "loss_num": 0.033447265625,
+      "loss_xval": 1.910032033920288,
+      "num_input_tokens_seen": 533961888,
+      "step": 3089
+    },
+    {
+      "epoch": 1.1834546150900038,
+      "grad_norm": 141.36348073633923,
+      "learning_rate": 5e-06,
+      "loss": 2.1838,
+      "num_input_tokens_seen": 534134648,
+      "step": 3090
+    },
+    {
+      "epoch": 1.1834546150900038,
+      "loss": 2.1448426246643066,
+      "loss_ce": 0.503603458404541,
+      "loss_iou": 0.7510090470314026,
+      "loss_num": 0.02783203125,
+      "loss_xval": 1.6412392854690552,
+      "num_input_tokens_seen": 534134648,
+      "step": 3090
+    },
+    {
+      "epoch": 1.1838376101110686,
+      "grad_norm": 185.078878828955,
+      "learning_rate": 5e-06,
+      "loss": 2.0241,
+      "num_input_tokens_seen": 534307288,
+      "step": 3091
+    },
+    {
+      "epoch": 1.1838376101110686,
+      "loss": 1.9381513595581055,
+      "loss_ce": 0.49507689476013184,
+      "loss_iou": 0.6708323359489441,
+      "loss_num": 0.020263671875,
+      "loss_xval": 1.4430745840072632,
+      "num_input_tokens_seen": 534307288,
+      "step": 3091
+    },
+    {
+      "epoch": 1.1842206051321333,
+      "grad_norm": 185.54291656001425,
+      "learning_rate": 5e-06,
+      "loss": 2.2354,
+      "num_input_tokens_seen": 534476920,
+      "step": 3092
+    },
+    {
+      "epoch": 1.1842206051321333,
+      "loss": 2.2484748363494873,
+      "loss_ce": 0.4822235107421875,
+      "loss_iou": 0.8301472067832947,
+      "loss_num": 0.021240234375,
+      "loss_xval": 1.7514809370040894,
+      "num_input_tokens_seen": 534476920,
+      "step": 3092
+    },
+    {
+      "epoch": 1.184603600153198,
+      "grad_norm": 100.56233131470334,
+      "learning_rate": 5e-06,
+      "loss": 2.2592,
+      "num_input_tokens_seen": 534650168,
+      "step": 3093
+    },
+    {
+      "epoch": 1.184603600153198,
+      "loss": 2.173462390899658,
+      "loss_ce": 0.49785223603248596,
+      "loss_iou": 0.7757627964019775,
+      "loss_num": 0.02490234375,
+      "loss_xval": 1.675610065460205,
+      "num_input_tokens_seen": 534650168,
+      "step": 3093
+    },
+    {
+      "epoch": 1.1849865951742626,
+      "grad_norm": 156.1794590780056,
+      "learning_rate": 5e-06,
+      "loss": 1.7972,
+      "num_input_tokens_seen": 534822960,
+      "step": 3094
+    },
+    {
+      "epoch": 1.1849865951742626,
+      "loss": 1.6737004518508911,
+      "loss_ce": 0.47707849740982056,
+      "loss_iou": 0.5625748634338379,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 1.1966218948364258,
+      "num_input_tokens_seen": 534822960,
+      "step": 3094
+    },
+    {
+      "epoch": 1.1853695901953274,
+      "grad_norm": 331.3179014216725,
+      "learning_rate": 5e-06,
+      "loss": 2.1746,
+      "num_input_tokens_seen": 534995776,
+      "step": 3095
+    },
+    {
+      "epoch": 1.1853695901953274,
+      "loss": 2.3747072219848633,
+      "loss_ce": 0.4959421753883362,
+      "loss_iou": 0.8826808929443359,
+      "loss_num": 0.022705078125,
+      "loss_xval": 1.8787651062011719,
+      "num_input_tokens_seen": 534995776,
+      "step": 3095
+    },
+    {
+      "epoch": 1.1857525852163922,
+      "grad_norm": 95.659466390315,
+      "learning_rate": 5e-06,
+      "loss": 2.1805,
+      "num_input_tokens_seen": 535168896,
+      "step": 3096
+    },
+    {
+      "epoch": 1.1857525852163922,
+      "loss": 2.071247100830078,
+      "loss_ce": 0.5040996670722961,
+      "loss_iou": 0.7256512641906738,
+      "loss_num": 0.023193359375,
+      "loss_xval": 1.5671472549438477,
+      "num_input_tokens_seen": 535168896,
+      "step": 3096
+    },
+    {
+      "epoch": 1.186135580237457,
+      "grad_norm": 98.1924804357043,
+      "learning_rate": 5e-06,
+      "loss": 1.7807,
+      "num_input_tokens_seen": 535338752,
+      "step": 3097
+    },
+    {
+      "epoch": 1.186135580237457,
+      "loss": 1.6573166847229004,
+      "loss_ce": 0.5046724677085876,
+      "loss_iou": 0.540708065032959,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 1.152644157409668,
+      "num_input_tokens_seen": 535338752,
+      "step": 3097
+    },
+    {
+      "epoch": 1.1865185752585217,
+      "grad_norm": 135.8681467987988,
+      "learning_rate": 5e-06,
+      "loss": 2.236,
+      "num_input_tokens_seen": 535511952,
+      "step": 3098
+    },
+    {
+      "epoch": 1.1865185752585217,
+      "loss": 2.317953109741211,
+      "loss_ce": 0.5204952955245972,
+      "loss_iou": 0.8491530418395996,
+      "loss_num": 0.019775390625,
+      "loss_xval": 1.7974576950073242,
+      "num_input_tokens_seen": 535511952,
+      "step": 3098
+    },
+    {
+      "epoch": 1.1869015702795864,
+      "grad_norm": 106.61760525812039,
+      "learning_rate": 5e-06,
+      "loss": 2.204,
+      "num_input_tokens_seen": 535685040,
+      "step": 3099
+    },
+    {
+      "epoch": 1.1869015702795864,
+      "loss": 2.198577880859375,
+      "loss_ce": 0.49872907996177673,
+      "loss_iou": 0.7996619939804077,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 1.6998488903045654,
+      "num_input_tokens_seen": 535685040,
+      "step": 3099
+    },
+    {
+      "epoch": 1.1872845653006512,
+      "grad_norm": 77.8865254037316,
+      "learning_rate": 5e-06,
+      "loss": 1.9618,
+      "num_input_tokens_seen": 535858192,
+      "step": 3100
+    },
+    {
+      "epoch": 1.1872845653006512,
+      "loss": 1.7767446041107178,
+      "loss_ce": 0.5023010969161987,
+      "loss_iou": 0.600615918636322,
+      "loss_num": 0.0146484375,
+      "loss_xval": 1.274443507194519,
+      "num_input_tokens_seen": 535858192,
+      "step": 3100
+    },
+    {
+      "epoch": 1.1876675603217157,
+      "grad_norm": 155.840972582791,
+      "learning_rate": 5e-06,
+      "loss": 1.9774,
+      "num_input_tokens_seen": 536031320,
+      "step": 3101
+    },
+    {
+      "epoch": 1.1876675603217157,
+      "loss": 1.8352344036102295,
+      "loss_ce": 0.5101897716522217,
+      "loss_iou": 0.6163797378540039,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 1.3250446319580078,
+      "num_input_tokens_seen": 536031320,
+      "step": 3101
+    },
+    {
+      "epoch": 1.1880505553427805,
+      "grad_norm": 110.38495657579385,
+      "learning_rate": 5e-06,
+      "loss": 2.0977,
+      "num_input_tokens_seen": 536204088,
+      "step": 3102
+    },
+    {
+      "epoch": 1.1880505553427805,
+      "loss": 2.2395238876342773,
+      "loss_ce": 0.49135062098503113,
+      "loss_iou": 0.8049337863922119,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 1.7481732368469238,
+      "num_input_tokens_seen": 536204088,
+      "step": 3102
+    },
+    {
+      "epoch": 1.1884335503638452,
+      "grad_norm": 59.5160470004755,
+      "learning_rate": 5e-06,
+      "loss": 1.8086,
+      "num_input_tokens_seen": 536377472,
+      "step": 3103
+    },
+    {
+      "epoch": 1.1884335503638452,
+      "loss": 1.804865837097168,
+      "loss_ce": 0.51838219165802,
+      "loss_iou": 0.6018751859664917,
+      "loss_num": 0.0166015625,
+      "loss_xval": 1.2864835262298584,
+      "num_input_tokens_seen": 536377472,
+      "step": 3103
+    },
+    {
+      "epoch": 1.18881654538491,
+      "grad_norm": 89.1968533612831,
+      "learning_rate": 5e-06,
+      "loss": 1.6553,
+      "num_input_tokens_seen": 536550112,
+      "step": 3104
+    },
+    {
+      "epoch": 1.18881654538491,
+      "loss": 1.7249507904052734,
+      "loss_ce": 0.4785350561141968,
+      "loss_iou": 0.589348554611206,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 1.246415615081787,
+      "num_input_tokens_seen": 536550112,
+      "step": 3104
+    },
+    {
+      "epoch": 1.1891995404059748,
+      "grad_norm": 137.08758924315913,
+      "learning_rate": 5e-06,
+      "loss": 1.7983,
+      "num_input_tokens_seen": 536723160,
+      "step": 3105
+    },
+    {
+      "epoch": 1.1891995404059748,
+      "loss": 1.8237223625183105,
+      "loss_ce": 0.493984192609787,
+      "loss_iou": 0.6302469372749329,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 1.3297382593154907,
+      "num_input_tokens_seen": 536723160,
+      "step": 3105
+    },
+    {
+      "epoch": 1.1895825354270395,
+      "grad_norm": 145.7121057801843,
+      "learning_rate": 5e-06,
+      "loss": 2.0669,
+      "num_input_tokens_seen": 536896160,
+      "step": 3106
+    },
+    {
+      "epoch": 1.1895825354270395,
+      "loss": 2.0431172847747803,
+      "loss_ce": 0.47122979164123535,
+      "loss_iou": 0.7367188930511475,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 1.571887493133545,
+      "num_input_tokens_seen": 536896160,
+      "step": 3106
+    },
+    {
+      "epoch": 1.1899655304481043,
+      "grad_norm": 75.90376758680951,
+      "learning_rate": 5e-06,
+      "loss": 2.3048,
+      "num_input_tokens_seen": 537069344,
+      "step": 3107
+    },
+    {
+      "epoch": 1.1899655304481043,
+      "loss": 2.289397716522217,
+      "loss_ce": 0.44671905040740967,
+      "loss_iou": 0.8297255039215088,
+      "loss_num": 0.03662109375,
+      "loss_xval": 1.8426785469055176,
+      "num_input_tokens_seen": 537069344,
+      "step": 3107
+    },
+    {
+      "epoch": 1.1903485254691688,
+      "grad_norm": 99.21312086510689,
+      "learning_rate": 5e-06,
+      "loss": 1.719,
+      "num_input_tokens_seen": 537242456,
+      "step": 3108
+    },
+    {
+      "epoch": 1.1903485254691688,
+      "loss": 1.5367971658706665,
+      "loss_ce": 0.4708802402019501,
+      "loss_iou": 0.5069498419761658,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 1.0659170150756836,
+      "num_input_tokens_seen": 537242456,
+      "step": 3108
+    },
+    {
+      "epoch": 1.1907315204902336,
+      "grad_norm": 113.36016933580596,
+      "learning_rate": 5e-06,
+      "loss": 1.8786,
+      "num_input_tokens_seen": 537415464,
+      "step": 3109
+    },
+    {
+      "epoch": 1.1907315204902336,
+      "loss": 1.7586313486099243,
+      "loss_ce": 0.4701662063598633,
+      "loss_iou": 0.5941989421844482,
+      "loss_num": 0.02001953125,
+      "loss_xval": 1.2819952964782715,
+      "num_input_tokens_seen": 537415464,
+      "step": 3109
+    },
+    {
+      "epoch": 1.1911145155112983,
+      "grad_norm": 119.55250576519718,
+      "learning_rate": 5e-06,
+      "loss": 1.9036,
+      "num_input_tokens_seen": 537588368,
+      "step": 3110
+    },
+    {
+      "epoch": 1.1911145155112983,
+      "loss": 1.8303320407867432,
+      "loss_ce": 0.4702279269695282,
+      "loss_iou": 0.6384260654449463,
+      "loss_num": 0.0166015625,
+      "loss_xval": 1.3601040840148926,
+      "num_input_tokens_seen": 537588368,
+      "step": 3110
+    },
+    {
+      "epoch": 1.191497510532363,
+      "grad_norm": 156.538618103619,
+      "learning_rate": 5e-06,
+      "loss": 1.9782,
+      "num_input_tokens_seen": 537761344,
+      "step": 3111
+    },
+    {
+      "epoch": 1.191497510532363,
+      "loss": 1.9333223104476929,
+      "loss_ce": 0.4669739603996277,
+      "loss_iou": 0.6894271969795227,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 1.4663482904434204,
+      "num_input_tokens_seen": 537761344,
+      "step": 3111
+    },
+    {
+      "epoch": 1.1918805055534278,
+      "grad_norm": 162.18242779358536,
+      "learning_rate": 5e-06,
+      "loss": 2.2692,
+      "num_input_tokens_seen": 537934552,
+      "step": 3112
+    },
+    {
+      "epoch": 1.1918805055534278,
+      "loss": 2.299771308898926,
+      "loss_ce": 0.4665790796279907,
+      "loss_iou": 0.8423773050308228,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 1.8331921100616455,
+      "num_input_tokens_seen": 537934552,
+      "step": 3112
+    },
+    {
+      "epoch": 1.1922635005744926,
+      "grad_norm": 47.94872101437066,
+      "learning_rate": 5e-06,
+      "loss": 1.7896,
+      "num_input_tokens_seen": 538107488,
+      "step": 3113
+    },
+    {
+      "epoch": 1.1922635005744926,
+      "loss": 1.7490005493164062,
+      "loss_ce": 0.4844675064086914,
+      "loss_iou": 0.5784487724304199,
+      "loss_num": 0.021484375,
+      "loss_xval": 1.2645330429077148,
+      "num_input_tokens_seen": 538107488,
+      "step": 3113
+    },
+    {
+      "epoch": 1.1926464955955574,
+      "grad_norm": 105.30828538266552,
+      "learning_rate": 5e-06,
+      "loss": 1.6832,
+      "num_input_tokens_seen": 538280568,
+      "step": 3114
+    },
+    {
+      "epoch": 1.1926464955955574,
+      "loss": 1.5757901668548584,
+      "loss_ce": 0.43801149725914,
+      "loss_iou": 0.5380741357803345,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 1.1377785205841064,
+      "num_input_tokens_seen": 538280568,
+      "step": 3114
+    },
+    {
+      "epoch": 1.193029490616622,
+      "grad_norm": 158.47482049089467,
+      "learning_rate": 5e-06,
+      "loss": 2.0204,
+      "num_input_tokens_seen": 538453360,
+      "step": 3115
+    },
+    {
+      "epoch": 1.193029490616622,
+      "loss": 2.0267176628112793,
+      "loss_ce": 0.4336962103843689,
+      "loss_iou": 0.7289599776268005,
+      "loss_num": 0.027099609375,
+      "loss_xval": 1.593021273612976,
+      "num_input_tokens_seen": 538453360,
+      "step": 3115
+    },
+    {
+      "epoch": 1.1934124856376866,
+      "grad_norm": 168.57864556594072,
+      "learning_rate": 5e-06,
+      "loss": 2.0213,
+      "num_input_tokens_seen": 538626680,
+      "step": 3116
+    },
+    {
+      "epoch": 1.1934124856376866,
+      "loss": 2.125065326690674,
+      "loss_ce": 0.4352427124977112,
+      "loss_iou": 0.7829911708831787,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 1.689822793006897,
+      "num_input_tokens_seen": 538626680,
+      "step": 3116
+    },
+    {
+      "epoch": 1.1937954806587514,
+      "grad_norm": 165.0405788645187,
+      "learning_rate": 5e-06,
+      "loss": 1.7747,
+      "num_input_tokens_seen": 538799744,
+      "step": 3117
+    },
+    {
+      "epoch": 1.1937954806587514,
+      "loss": 1.824463129043579,
+      "loss_ce": 0.4427624046802521,
+      "loss_iou": 0.6334620714187622,
+      "loss_num": 0.02294921875,
+      "loss_xval": 1.3817007541656494,
+      "num_input_tokens_seen": 538799744,
+      "step": 3117
+    },
+    {
+      "epoch": 1.1941784756798162,
+      "grad_norm": 131.74273003933374,
+      "learning_rate": 5e-06,
+      "loss": 2.3743,
+      "num_input_tokens_seen": 538972856,
+      "step": 3118
+    },
+    {
+      "epoch": 1.1941784756798162,
+      "loss": 2.4387192726135254,
+      "loss_ce": 0.4320226311683655,
+      "loss_iou": 0.9001989364624023,
+      "loss_num": 0.041259765625,
+      "loss_xval": 2.0066967010498047,
+      "num_input_tokens_seen": 538972856,
+      "step": 3118
+    },
+    {
+      "epoch": 1.194561470700881,
+      "grad_norm": 87.15329040106326,
+      "learning_rate": 5e-06,
+      "loss": 1.8024,
+      "num_input_tokens_seen": 539145752,
+      "step": 3119
+    },
+    {
+      "epoch": 1.194561470700881,
+      "loss": 1.725273847579956,
+      "loss_ce": 0.44722792506217957,
+      "loss_iou": 0.5879517793655396,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 1.278045892715454,
+      "num_input_tokens_seen": 539145752,
+      "step": 3119
+    },
+    {
+      "epoch": 1.1949444657219457,
+      "grad_norm": 112.77290771039542,
+      "learning_rate": 5e-06,
+      "loss": 1.8274,
+      "num_input_tokens_seen": 539318872,
+      "step": 3120
+    },
+    {
+      "epoch": 1.1949444657219457,
+      "loss": 1.7642662525177002,
+      "loss_ce": 0.46796268224716187,
+      "loss_iou": 0.607609212398529,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 1.296303629875183,
+      "num_input_tokens_seen": 539318872,
+      "step": 3120
+    },
+    {
+      "epoch": 1.1953274607430102,
+      "grad_norm": 98.35715045614509,
+      "learning_rate": 5e-06,
+      "loss": 1.9458,
+      "num_input_tokens_seen": 539491696,
+      "step": 3121
+    },
+    {
+      "epoch": 1.1953274607430102,
+      "loss": 1.9283684492111206,
+      "loss_ce": 0.46976467967033386,
+      "loss_iou": 0.685448169708252,
+      "loss_num": 0.017578125,
+      "loss_xval": 1.458603858947754,
+      "num_input_tokens_seen": 539491696,
+      "step": 3121
+    },
+    {
+      "epoch": 1.195710455764075,
+      "grad_norm": 85.9145857232359,
+      "learning_rate": 5e-06,
+      "loss": 1.8642,
+      "num_input_tokens_seen": 539664320,
+      "step": 3122
+    },
+    {
+      "epoch": 1.195710455764075,
+      "loss": 1.9947984218597412,
+      "loss_ce": 0.4589182138442993,
+      "loss_iou": 0.6967273950576782,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 1.5358803272247314,
+      "num_input_tokens_seen": 539664320,
+      "step": 3122
+    },
+    {
+      "epoch": 1.1960934507851397,
+      "grad_norm": 178.0041822743624,
+      "learning_rate": 5e-06,
+      "loss": 2.1027,
+      "num_input_tokens_seen": 539836968,
+      "step": 3123
+    },
+    {
+      "epoch": 1.1960934507851397,
+      "loss": 2.0192744731903076,
+      "loss_ce": 0.4354323744773865,
+      "loss_iou": 0.7428792119026184,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 1.5838419198989868,
+      "num_input_tokens_seen": 539836968,
+      "step": 3123
+    },
+    {
+      "epoch": 1.1964764458062045,
+      "grad_norm": 102.29299889009614,
+      "learning_rate": 5e-06,
+      "loss": 2.2874,
+      "num_input_tokens_seen": 540010136,
+      "step": 3124
+    },
+    {
+      "epoch": 1.1964764458062045,
+      "loss": 2.20661997795105,
+      "loss_ce": 0.4187849164009094,
+      "loss_iou": 0.8059658408164978,
+      "loss_num": 0.03515625,
+      "loss_xval": 1.7878351211547852,
+      "num_input_tokens_seen": 540010136,
+      "step": 3124
+    },
+    {
+      "epoch": 1.1968594408272692,
+      "grad_norm": 169.16879881047464,
+      "learning_rate": 5e-06,
+      "loss": 1.7049,
+      "num_input_tokens_seen": 540182912,
+      "step": 3125
+    },
+    {
+      "epoch": 1.1968594408272692,
+      "loss": 1.7866005897521973,
+      "loss_ce": 0.38834109902381897,
+      "loss_iou": 0.6257350444793701,
+      "loss_num": 0.029296875,
+      "loss_xval": 1.3587088584899902,
+      "num_input_tokens_seen": 540182912,
+      "step": 3125
+    },
+    {
+      "epoch": 1.197242435848334,
+      "grad_norm": 214.9198131394962,
+      "learning_rate": 5e-06,
+      "loss": 3.465,
+      "num_input_tokens_seen": 540355872,
+      "step": 3126
+    },
+    {
+      "epoch": 1.197242435848334,
+      "loss": 3.44808292388916,
+      "loss_ce": 0.42785316705703735,
+      "loss_iou": 1.4119094610214233,
+      "loss_num": 0.039306640625,
+      "loss_xval": 3.0202300548553467,
+      "num_input_tokens_seen": 540355872,
+      "step": 3126
+    },
+    {
+      "epoch": 1.1976254308693988,
+      "grad_norm": 244.08982046201126,
+      "learning_rate": 5e-06,
+      "loss": 2.077,
+      "num_input_tokens_seen": 540529208,
+      "step": 3127
+    },
+    {
+      "epoch": 1.1976254308693988,
+      "loss": 2.0087804794311523,
+      "loss_ce": 0.48206713795661926,
+      "loss_iou": 0.7112632989883423,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 1.5267136096954346,
+      "num_input_tokens_seen": 540529208,
+      "step": 3127
+    },
+    {
+      "epoch": 1.1980084258904635,
+      "grad_norm": 250.80855549566525,
+      "learning_rate": 5e-06,
+      "loss": 2.5378,
+      "num_input_tokens_seen": 540698496,
+      "step": 3128
+    },
+    {
+      "epoch": 1.1980084258904635,
+      "loss": 2.458965301513672,
+      "loss_ce": 0.4827730655670166,
+      "loss_iou": 0.9210793972015381,
+      "loss_num": 0.02685546875,
+      "loss_xval": 1.9761919975280762,
+      "num_input_tokens_seen": 540698496,
+      "step": 3128
+    },
+    {
+      "epoch": 1.198391420911528,
+      "grad_norm": 205.82804991445155,
+      "learning_rate": 5e-06,
+      "loss": 2.2837,
+      "num_input_tokens_seen": 540872048,
+      "step": 3129
+    },
+    {
+      "epoch": 1.198391420911528,
+      "loss": 2.347994327545166,
+      "loss_ce": 0.4887653887271881,
+      "loss_iou": 0.8249697685241699,
+      "loss_num": 0.041748046875,
+      "loss_xval": 1.8592290878295898,
+      "num_input_tokens_seen": 540872048,
+      "step": 3129
+    },
+    {
+      "epoch": 1.1987744159325928,
+      "grad_norm": 193.933614492199,
+      "learning_rate": 5e-06,
+      "loss": 1.9608,
+      "num_input_tokens_seen": 541044768,
+      "step": 3130
+    },
+    {
+      "epoch": 1.1987744159325928,
+      "loss": 2.005110263824463,
+      "loss_ce": 0.46384477615356445,
+      "loss_iou": 0.7266265153884888,
+      "loss_num": 0.017578125,
+      "loss_xval": 1.5412657260894775,
+      "num_input_tokens_seen": 541044768,
+      "step": 3130
+    },
+    {
+      "epoch": 1.1991574109536576,
+      "grad_norm": 172.24400401577392,
+      "learning_rate": 5e-06,
+      "loss": 1.9607,
+      "num_input_tokens_seen": 541217544,
+      "step": 3131
+    },
+    {
+      "epoch": 1.1991574109536576,
+      "loss": 1.8374714851379395,
+      "loss_ce": 0.46363934874534607,
+      "loss_iou": 0.6391102075576782,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 1.3738319873809814,
+      "num_input_tokens_seen": 541217544,
+      "step": 3131
+    },
+    {
+      "epoch": 1.1995404059747223,
+      "grad_norm": 81.50056155127288,
+      "learning_rate": 5e-06,
+      "loss": 1.6268,
+      "num_input_tokens_seen": 541390576,
+      "step": 3132
+    },
+    {
+      "epoch": 1.1995404059747223,
+      "loss": 1.5544099807739258,
+      "loss_ce": 0.44078072905540466,
+      "loss_iou": 0.5041108131408691,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 1.1033754348754883,
+      "num_input_tokens_seen": 541390576,
+      "step": 3132
+    },
+    {
+      "epoch": 1.199923400995787,
+      "grad_norm": 211.04545142130024,
+      "learning_rate": 5e-06,
+      "loss": 1.6326,
+      "num_input_tokens_seen": 541563320,
+      "step": 3133
+    },
+    {
+      "epoch": 1.199923400995787,
+      "loss": 1.7176035642623901,
+      "loss_ce": 0.42498356103897095,
+      "loss_iou": 0.5992518663406372,
+      "loss_num": 0.018798828125,
+      "loss_xval": 1.2807180881500244,
+      "num_input_tokens_seen": 541563320,
+      "step": 3133
+    },
+    {
+      "epoch": 1.2003063960168519,
+      "grad_norm": 122.70851376184869,
+      "learning_rate": 5e-06,
+      "loss": 1.9539,
+      "num_input_tokens_seen": 541736184,
+      "step": 3134
+    },
+    {
+      "epoch": 1.2003063960168519,
+      "loss": 2.0332226753234863,
+      "loss_ce": 0.4139161705970764,
+      "loss_iou": 0.7215796113014221,
+      "loss_num": 0.03515625,
+      "loss_xval": 1.6193065643310547,
+      "num_input_tokens_seen": 541736184,
+      "step": 3134
+    },
+    {
+      "epoch": 1.2006893910379164,
+      "grad_norm": 101.12918973362021,
+      "learning_rate": 5e-06,
+      "loss": 1.7302,
+      "num_input_tokens_seen": 541909384,
+      "step": 3135
+    },
+    {
+      "epoch": 1.2006893910379164,
+      "loss": 1.8653372526168823,
+      "loss_ce": 0.47176891565322876,
+      "loss_iou": 0.6533271670341492,
+      "loss_num": 0.017333984375,
+      "loss_xval": 1.3935683965682983,
+      "num_input_tokens_seen": 541909384,
+      "step": 3135
+    },
+    {
+      "epoch": 1.2010723860589811,
+      "grad_norm": 85.17518217819423,
+      "learning_rate": 5e-06,
+      "loss": 1.932,
+      "num_input_tokens_seen": 542082280,
+      "step": 3136
+    },
+    {
+      "epoch": 1.2010723860589811,
+      "loss": 2.052091121673584,
+      "loss_ce": 0.44024819135665894,
+      "loss_iou": 0.7444589734077454,
+      "loss_num": 0.024658203125,
+      "loss_xval": 1.6118427515029907,
+      "num_input_tokens_seen": 542082280,
+      "step": 3136
+    },
+    {
+      "epoch": 1.201455381080046,
+      "grad_norm": 109.44411326728972,
+      "learning_rate": 5e-06,
+      "loss": 1.6436,
+      "num_input_tokens_seen": 542254880,
+      "step": 3137
+    },
+    {
+      "epoch": 1.201455381080046,
+      "loss": 1.718399167060852,
+      "loss_ce": 0.42179369926452637,
+      "loss_iou": 0.6057153940200806,
+      "loss_num": 0.01708984375,
+      "loss_xval": 1.2966053485870361,
+      "num_input_tokens_seen": 542254880,
+      "step": 3137
+    },
+    {
+      "epoch": 1.2018383761011107,
+      "grad_norm": 84.40625767941842,
+      "learning_rate": 5e-06,
+      "loss": 1.9448,
+      "num_input_tokens_seen": 542428208,
+      "step": 3138
+    },
+    {
+      "epoch": 1.2018383761011107,
+      "loss": 1.921809196472168,
+      "loss_ce": 0.3864181637763977,
+      "loss_iou": 0.7132673263549805,
+      "loss_num": 0.021728515625,
+      "loss_xval": 1.535390853881836,
+      "num_input_tokens_seen": 542428208,
+      "step": 3138
+    },
+    {
+      "epoch": 1.2022213711221754,
+      "grad_norm": 74.28199122633579,
+      "learning_rate": 5e-06,
+      "loss": 1.509,
+      "num_input_tokens_seen": 542601032,
+      "step": 3139
+    },
+    {
+      "epoch": 1.2022213711221754,
+      "loss": 1.3731598854064941,
+      "loss_ce": 0.4295925498008728,
+      "loss_iou": 0.4371461868286133,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.7442958354949951,
+      "num_input_tokens_seen": 542601032,
+      "step": 3139
+    },
+    {
+      "epoch": 1.2026043661432402,
+      "grad_norm": 166.1930064220397,
+      "learning_rate": 5e-06,
+      "loss": 1.6838,
+      "num_input_tokens_seen": 542774376,
+      "step": 3140
+    },
+    {
+      "epoch": 1.2026043661432402,
+      "loss": 1.6118892431259155,
+      "loss_ce": 0.38677966594696045,
+      "loss_iou": 0.566519021987915,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 1.211620807647705,
+      "num_input_tokens_seen": 542774376,
+      "step": 3140
+    },
+    {
+      "epoch": 1.202987361164305,
+      "grad_norm": 190.0469889866551,
+      "learning_rate": 5e-06,
+      "loss": 1.9974,
+      "num_input_tokens_seen": 542947216,
+      "step": 3141
+    },
+    {
+      "epoch": 1.202987361164305,
+      "loss": 1.7716901302337646,
+      "loss_ce": 0.40608906745910645,
+      "loss_iou": 0.6318361759185791,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 1.3499760627746582,
+      "num_input_tokens_seen": 542947216,
+      "step": 3141
+    },
+    {
+      "epoch": 1.2033703561853697,
+      "grad_norm": 276.2473347389736,
+      "learning_rate": 5e-06,
+      "loss": 2.0013,
+      "num_input_tokens_seen": 543120248,
+      "step": 3142
+    },
+    {
+      "epoch": 1.2033703561853697,
+      "loss": 1.96783447265625,
+      "loss_ce": 0.40639060735702515,
+      "loss_iou": 0.7039092183113098,
+      "loss_num": 0.03076171875,
+      "loss_xval": 1.56144380569458,
+      "num_input_tokens_seen": 543120248,
+      "step": 3142
+    },
+    {
+      "epoch": 1.2037533512064342,
+      "grad_norm": 110.19184001956351,
+      "learning_rate": 5e-06,
+      "loss": 2.093,
+      "num_input_tokens_seen": 543293088,
+      "step": 3143
+    },
+    {
+      "epoch": 1.2037533512064342,
+      "loss": 2.0981013774871826,
+      "loss_ce": 0.42810600996017456,
+      "loss_iou": 0.7602143883705139,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 1.6699954271316528,
+      "num_input_tokens_seen": 543293088,
+      "step": 3143
+    },
+    {
+      "epoch": 1.204136346227499,
+      "grad_norm": 161.9044402146107,
+      "learning_rate": 5e-06,
+      "loss": 1.6954,
+      "num_input_tokens_seen": 543465880,
+      "step": 3144
+    },
+    {
+      "epoch": 1.204136346227499,
+      "loss": 1.7552735805511475,
+      "loss_ce": 0.41109591722488403,
+      "loss_iou": 0.6269075870513916,
+      "loss_num": 0.01806640625,
+      "loss_xval": 1.3441777229309082,
+      "num_input_tokens_seen": 543465880,
+      "step": 3144
+    },
+    {
+      "epoch": 1.2045193412485637,
+      "grad_norm": 137.31052434918928,
+      "learning_rate": 5e-06,
+      "loss": 2.0543,
+      "num_input_tokens_seen": 543638880,
+      "step": 3145
+    },
+    {
+      "epoch": 1.2045193412485637,
+      "loss": 2.0171594619750977,
+      "loss_ce": 0.42851918935775757,
+      "loss_iou": 0.7467432022094727,
+      "loss_num": 0.01904296875,
+      "loss_xval": 1.5886402130126953,
+      "num_input_tokens_seen": 543638880,
+      "step": 3145
+    },
+    {
+      "epoch": 1.2049023362696285,
+      "grad_norm": 180.0421227264355,
+      "learning_rate": 5e-06,
+      "loss": 2.6378,
+      "num_input_tokens_seen": 543811792,
+      "step": 3146
+    },
+    {
+      "epoch": 1.2049023362696285,
+      "loss": 2.5615615844726562,
+      "loss_ce": 0.41080349683761597,
+      "loss_iou": 0.8848272562026978,
+      "loss_num": 0.076171875,
+      "loss_xval": 2.1507580280303955,
+      "num_input_tokens_seen": 543811792,
+      "step": 3146
+    },
+    {
+      "epoch": 1.2052853312906933,
+      "grad_norm": 186.06827639905708,
+      "learning_rate": 5e-06,
+      "loss": 1.9071,
+      "num_input_tokens_seen": 543984848,
+      "step": 3147
+    },
+    {
+      "epoch": 1.2052853312906933,
+      "loss": 1.9688708782196045,
+      "loss_ce": 0.37465977668762207,
+      "loss_iou": 0.7266404628753662,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 1.5942111015319824,
+      "num_input_tokens_seen": 543984848,
+      "step": 3147
+    },
+    {
+      "epoch": 1.205668326311758,
+      "grad_norm": 177.21960815543142,
+      "learning_rate": 5e-06,
+      "loss": 2.8569,
+      "num_input_tokens_seen": 544157968,
+      "step": 3148
+    },
+    {
+      "epoch": 1.205668326311758,
+      "loss": 2.867978096008301,
+      "loss_ce": 0.3810361623764038,
+      "loss_iou": 1.1110856533050537,
+      "loss_num": 0.052978515625,
+      "loss_xval": 2.4869418144226074,
+      "num_input_tokens_seen": 544157968,
+      "step": 3148
+    },
+    {
+      "epoch": 1.2060513213328226,
+      "grad_norm": 104.67555522556067,
+      "learning_rate": 5e-06,
+      "loss": 2.2487,
+      "num_input_tokens_seen": 544331024,
+      "step": 3149
+    },
+    {
+      "epoch": 1.2060513213328226,
+      "loss": 2.21226167678833,
+      "loss_ce": 0.4219363331794739,
+      "loss_iou": 0.7929898500442505,
+      "loss_num": 0.040771484375,
+      "loss_xval": 1.790325403213501,
+      "num_input_tokens_seen": 544331024,
+      "step": 3149
+    },
+    {
+      "epoch": 1.2064343163538873,
+      "grad_norm": 111.45546340906715,
+      "learning_rate": 5e-06,
+      "loss": 2.2152,
+      "num_input_tokens_seen": 544504400,
+      "step": 3150
+    },
+    {
+      "epoch": 1.2064343163538873,
+      "loss": 2.238651752471924,
+      "loss_ce": 0.4570075273513794,
+      "loss_iou": 0.8178850412368774,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 1.7816441059112549,
+      "num_input_tokens_seen": 544504400,
+      "step": 3150
+    },
+    {
+      "epoch": 1.206817311374952,
+      "grad_norm": 126.96258382781926,
+      "learning_rate": 5e-06,
+      "loss": 1.6517,
+      "num_input_tokens_seen": 544677280,
+      "step": 3151
+    },
+    {
+      "epoch": 1.206817311374952,
+      "loss": 1.7380001544952393,
+      "loss_ce": 0.4228131175041199,
+      "loss_iou": 0.6230170726776123,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 1.3151869773864746,
+      "num_input_tokens_seen": 544677280,
+      "step": 3151
+    },
+    {
+      "epoch": 1.2072003063960168,
+      "grad_norm": 173.14029111978968,
+      "learning_rate": 5e-06,
+      "loss": 2.1909,
+      "num_input_tokens_seen": 544850264,
+      "step": 3152
+    },
+    {
+      "epoch": 1.2072003063960168,
+      "loss": 2.182771921157837,
+      "loss_ce": 0.4373222887516022,
+      "loss_iou": 0.8156265020370483,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 1.7454497814178467,
+      "num_input_tokens_seen": 544850264,
+      "step": 3152
+    },
+    {
+      "epoch": 1.2075833014170816,
+      "grad_norm": 109.54421461126226,
+      "learning_rate": 5e-06,
+      "loss": 1.9302,
+      "num_input_tokens_seen": 545023304,
+      "step": 3153
+    },
+    {
+      "epoch": 1.2075833014170816,
+      "loss": 1.8964380025863647,
+      "loss_ce": 0.4667872488498688,
+      "loss_iou": 0.6590697765350342,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 1.4296507835388184,
+      "num_input_tokens_seen": 545023304,
+      "step": 3153
+    },
+    {
+      "epoch": 1.2079662964381463,
+      "grad_norm": 155.12626056609227,
+      "learning_rate": 5e-06,
+      "loss": 1.7024,
+      "num_input_tokens_seen": 545196296,
+      "step": 3154
+    },
+    {
+      "epoch": 1.2079662964381463,
+      "loss": 1.7114551067352295,
+      "loss_ce": 0.3582027554512024,
+      "loss_iou": 0.6047877669334412,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 1.3532522916793823,
+      "num_input_tokens_seen": 545196296,
+      "step": 3154
+    },
+    {
+      "epoch": 1.208349291459211,
+      "grad_norm": 85.11109123507192,
+      "learning_rate": 5e-06,
+      "loss": 2.179,
+      "num_input_tokens_seen": 545369256,
+      "step": 3155
+    },
+    {
+      "epoch": 1.208349291459211,
+      "loss": 2.168123722076416,
+      "loss_ce": 0.4210514724254608,
+      "loss_iou": 0.7872323393821716,
+      "loss_num": 0.03466796875,
+      "loss_xval": 1.7470721006393433,
+      "num_input_tokens_seen": 545369256,
+      "step": 3155
+    },
+    {
+      "epoch": 1.2087322864802759,
+      "grad_norm": 93.31662199160257,
+      "learning_rate": 5e-06,
+      "loss": 1.8172,
+      "num_input_tokens_seen": 545542424,
+      "step": 3156
+    },
+    {
+      "epoch": 1.2087322864802759,
+      "loss": 1.997159481048584,
+      "loss_ce": 0.4138537049293518,
+      "loss_iou": 0.7168543338775635,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 1.583305835723877,
+      "num_input_tokens_seen": 545542424,
+      "step": 3156
+    },
+    {
+      "epoch": 1.2091152815013404,
+      "grad_norm": 194.46825793740766,
+      "learning_rate": 5e-06,
+      "loss": 2.0425,
+      "num_input_tokens_seen": 545715432,
+      "step": 3157
+    },
+    {
+      "epoch": 1.2091152815013404,
+      "loss": 2.0387141704559326,
+      "loss_ce": 0.40486350655555725,
+      "loss_iou": 0.7460329532623291,
+      "loss_num": 0.0283203125,
+      "loss_xval": 1.6338505744934082,
+      "num_input_tokens_seen": 545715432,
+      "step": 3157
+    },
+    {
+      "epoch": 1.2094982765224052,
+      "grad_norm": 102.17582157733656,
+      "learning_rate": 5e-06,
+      "loss": 2.4329,
+      "num_input_tokens_seen": 545888312,
+      "step": 3158
+    },
+    {
+      "epoch": 1.2094982765224052,
+      "loss": 2.5089845657348633,
+      "loss_ce": 0.3819023072719574,
+      "loss_iou": 0.9404637813568115,
+      "loss_num": 0.04931640625,
+      "loss_xval": 2.127082347869873,
+      "num_input_tokens_seen": 545888312,
+      "step": 3158
+    },
+    {
+      "epoch": 1.20988127154347,
+      "grad_norm": 128.1332777760175,
+      "learning_rate": 5e-06,
+      "loss": 1.9519,
+      "num_input_tokens_seen": 546061184,
+      "step": 3159
+    },
+    {
+      "epoch": 1.20988127154347,
+      "loss": 2.0174331665039062,
+      "loss_ce": 0.43313467502593994,
+      "loss_iou": 0.7280622720718384,
+      "loss_num": 0.025634765625,
+      "loss_xval": 1.5842983722686768,
+      "num_input_tokens_seen": 546061184,
+      "step": 3159
+    },
+    {
+      "epoch": 1.2102642665645347,
+      "grad_norm": 166.00311725221368,
+      "learning_rate": 5e-06,
+      "loss": 2.016,
+      "num_input_tokens_seen": 546234064,
+      "step": 3160
+    },
+    {
+      "epoch": 1.2102642665645347,
+      "loss": 2.0598645210266113,
+      "loss_ce": 0.3841882646083832,
+      "loss_iou": 0.7913904190063477,
+      "loss_num": 0.0185546875,
+      "loss_xval": 1.6756762266159058,
+      "num_input_tokens_seen": 546234064,
+      "step": 3160
+    },
+    {
+      "epoch": 1.2106472615855994,
+      "grad_norm": 75.9356801665817,
+      "learning_rate": 5e-06,
+      "loss": 2.3616,
+      "num_input_tokens_seen": 546407040,
+      "step": 3161
+    },
+    {
+      "epoch": 1.2106472615855994,
+      "loss": 2.368368148803711,
+      "loss_ce": 0.40133553743362427,
+      "loss_iou": 0.8806720972061157,
+      "loss_num": 0.041259765625,
+      "loss_xval": 1.967032551765442,
+      "num_input_tokens_seen": 546407040,
+      "step": 3161
+    },
+    {
+      "epoch": 1.2110302566066642,
+      "grad_norm": 80.20061715058009,
+      "learning_rate": 5e-06,
+      "loss": 1.8253,
+      "num_input_tokens_seen": 546580256,
+      "step": 3162
+    },
+    {
+      "epoch": 1.2110302566066642,
+      "loss": 1.9942257404327393,
+      "loss_ce": 0.3905842900276184,
+      "loss_iou": 0.705446183681488,
+      "loss_num": 0.03857421875,
+      "loss_xval": 1.603641390800476,
+      "num_input_tokens_seen": 546580256,
+      "step": 3162
+    },
+    {
+      "epoch": 1.2114132516277287,
+      "grad_norm": 139.36934028075223,
+      "learning_rate": 5e-06,
+      "loss": 1.7623,
+      "num_input_tokens_seen": 546752784,
+      "step": 3163
+    },
+    {
+      "epoch": 1.2114132516277287,
+      "loss": 1.612166404724121,
+      "loss_ce": 0.4187697172164917,
+      "loss_iou": 0.5422397255897522,
+      "loss_num": 0.021728515625,
+      "loss_xval": 1.1659308671951294,
+      "num_input_tokens_seen": 546752784,
+      "step": 3163
+    },
+    {
+      "epoch": 1.2117962466487935,
+      "grad_norm": 89.05301644728904,
+      "learning_rate": 5e-06,
+      "loss": 2.116,
+      "num_input_tokens_seen": 546925888,
+      "step": 3164
+    },
+    {
+      "epoch": 1.2117962466487935,
+      "loss": 2.1580970287323,
+      "loss_ce": 0.3667049705982208,
+      "loss_iou": 0.7959645986557007,
+      "loss_num": 0.039794921875,
+      "loss_xval": 1.7913920879364014,
+      "num_input_tokens_seen": 546925888,
+      "step": 3164
+    },
+    {
+      "epoch": 1.2121792416698582,
+      "grad_norm": 105.80627408837645,
+      "learning_rate": 5e-06,
+      "loss": 1.9025,
+      "num_input_tokens_seen": 547098880,
+      "step": 3165
+    },
+    {
+      "epoch": 1.2121792416698582,
+      "loss": 2.0705251693725586,
+      "loss_ce": 0.39548683166503906,
+      "loss_iou": 0.7514901161193848,
+      "loss_num": 0.034423828125,
+      "loss_xval": 1.6750383377075195,
+      "num_input_tokens_seen": 547098880,
+      "step": 3165
+    },
+    {
+      "epoch": 1.212562236690923,
+      "grad_norm": 176.6958791685355,
+      "learning_rate": 5e-06,
+      "loss": 1.9749,
+      "num_input_tokens_seen": 547271656,
+      "step": 3166
+    },
+    {
+      "epoch": 1.212562236690923,
+      "loss": 1.965118408203125,
+      "loss_ce": 0.41928714513778687,
+      "loss_iou": 0.7106139063835144,
+      "loss_num": 0.02490234375,
+      "loss_xval": 1.5458310842514038,
+      "num_input_tokens_seen": 547271656,
+      "step": 3166
+    },
+    {
+      "epoch": 1.2129452317119878,
+      "grad_norm": 86.74503746663704,
+      "learning_rate": 5e-06,
+      "loss": 2.4458,
+      "num_input_tokens_seen": 547444392,
+      "step": 3167
+    },
+    {
+      "epoch": 1.2129452317119878,
+      "loss": 2.434870481491089,
+      "loss_ce": 0.3902822732925415,
+      "loss_iou": 0.9276590347290039,
+      "loss_num": 0.037841796875,
+      "loss_xval": 2.044588088989258,
+      "num_input_tokens_seen": 547444392,
+      "step": 3167
+    },
+    {
+      "epoch": 1.2133282267330525,
+      "grad_norm": 109.50378315927813,
+      "learning_rate": 5e-06,
+      "loss": 2.0427,
+      "num_input_tokens_seen": 547617224,
+      "step": 3168
+    },
+    {
+      "epoch": 1.2133282267330525,
+      "loss": 2.069617509841919,
+      "loss_ce": 0.4061136543750763,
+      "loss_iou": 0.7618972063064575,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 1.663503885269165,
+      "num_input_tokens_seen": 547617224,
+      "step": 3168
+    },
+    {
+      "epoch": 1.2137112217541173,
+      "grad_norm": 102.55478575623683,
+      "learning_rate": 5e-06,
+      "loss": 1.7574,
+      "num_input_tokens_seen": 547790144,
+      "step": 3169
+    },
+    {
+      "epoch": 1.2137112217541173,
+      "loss": 1.8150783777236938,
+      "loss_ce": 0.41318345069885254,
+      "loss_iou": 0.6508070230484009,
+      "loss_num": 0.02001953125,
+      "loss_xval": 1.4018948078155518,
+      "num_input_tokens_seen": 547790144,
+      "step": 3169
+    },
+    {
+      "epoch": 1.214094216775182,
+      "grad_norm": 103.97034145766871,
+      "learning_rate": 5e-06,
+      "loss": 2.5278,
+      "num_input_tokens_seen": 547962632,
+      "step": 3170
+    },
+    {
+      "epoch": 1.214094216775182,
+      "loss": 2.4732987880706787,
+      "loss_ce": 0.365742564201355,
+      "loss_iou": 0.9795289039611816,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 2.1075563430786133,
+      "num_input_tokens_seen": 547962632,
+      "step": 3170
+    },
+    {
+      "epoch": 1.2144772117962466,
+      "grad_norm": 91.75053945426667,
+      "learning_rate": 5e-06,
+      "loss": 2.3501,
+      "num_input_tokens_seen": 548135768,
+      "step": 3171
+    },
+    {
+      "epoch": 1.2144772117962466,
+      "loss": 2.3920657634735107,
+      "loss_ce": 0.3515428304672241,
+      "loss_iou": 0.8424050807952881,
+      "loss_num": 0.0712890625,
+      "loss_xval": 2.040523052215576,
+      "num_input_tokens_seen": 548135768,
+      "step": 3171
+    },
+    {
+      "epoch": 1.2148602068173113,
+      "grad_norm": 61.849301496515274,
+      "learning_rate": 5e-06,
+      "loss": 1.7051,
+      "num_input_tokens_seen": 548308728,
+      "step": 3172
+    },
+    {
+      "epoch": 1.2148602068173113,
+      "loss": 1.6450124979019165,
+      "loss_ce": 0.3722168803215027,
+      "loss_iou": 0.5765833854675293,
+      "loss_num": 0.02392578125,
+      "loss_xval": 1.2727956771850586,
+      "num_input_tokens_seen": 548308728,
+      "step": 3172
+    },
+    {
+      "epoch": 1.215243201838376,
+      "grad_norm": 95.09659762996829,
+      "learning_rate": 5e-06,
+      "loss": 1.6952,
+      "num_input_tokens_seen": 548481448,
+      "step": 3173
+    },
+    {
+      "epoch": 1.215243201838376,
+      "loss": 1.82415771484375,
+      "loss_ce": 0.38953959941864014,
+      "loss_iou": 0.6570368409156799,
+      "loss_num": 0.024169921875,
+      "loss_xval": 1.4346181154251099,
+      "num_input_tokens_seen": 548481448,
+      "step": 3173
+    },
+    {
+      "epoch": 1.2156261968594408,
+      "grad_norm": 101.60642646717736,
+      "learning_rate": 5e-06,
+      "loss": 2.0832,
+      "num_input_tokens_seen": 548654184,
+      "step": 3174
+    },
+    {
+      "epoch": 1.2156261968594408,
+      "loss": 2.0614495277404785,
+      "loss_ce": 0.3467678427696228,
+      "loss_iou": 0.8017683029174805,
+      "loss_num": 0.022216796875,
+      "loss_xval": 1.714681625366211,
+      "num_input_tokens_seen": 548654184,
+      "step": 3174
+    },
+    {
+      "epoch": 1.2160091918805056,
+      "grad_norm": 90.59437001960373,
+      "learning_rate": 5e-06,
+      "loss": 1.9203,
+      "num_input_tokens_seen": 548827624,
+      "step": 3175
+    },
+    {
+      "epoch": 1.2160091918805056,
+      "loss": 2.055656909942627,
+      "loss_ce": 0.3785133361816406,
+      "loss_iou": 0.7676794528961182,
+      "loss_num": 0.0283203125,
+      "loss_xval": 1.6771435737609863,
+      "num_input_tokens_seen": 548827624,
+      "step": 3175
+    },
+    {
+      "epoch": 1.2163921869015704,
+      "grad_norm": 165.76325074868544,
+      "learning_rate": 5e-06,
+      "loss": 1.5723,
+      "num_input_tokens_seen": 549000272,
+      "step": 3176
+    },
+    {
+      "epoch": 1.2163921869015704,
+      "loss": 1.5530579090118408,
+      "loss_ce": 0.39820683002471924,
+      "loss_iou": 0.5333581566810608,
+      "loss_num": 0.017578125,
+      "loss_xval": 1.1548510789871216,
+      "num_input_tokens_seen": 549000272,
+      "step": 3176
+    },
+    {
+      "epoch": 1.216775181922635,
+      "grad_norm": 128.32266130053333,
+      "learning_rate": 5e-06,
+      "loss": 2.0198,
+      "num_input_tokens_seen": 549173096,
+      "step": 3177
+    },
+    {
+      "epoch": 1.216775181922635,
+      "loss": 2.107774019241333,
+      "loss_ce": 0.3247598111629486,
+      "loss_iou": 0.8118257522583008,
+      "loss_num": 0.03173828125,
+      "loss_xval": 1.7830142974853516,
+      "num_input_tokens_seen": 549173096,
+      "step": 3177
+    },
+    {
+      "epoch": 1.2171581769436997,
+      "grad_norm": 90.04220984141216,
+      "learning_rate": 5e-06,
+      "loss": 1.5218,
+      "num_input_tokens_seen": 549345928,
+      "step": 3178
+    },
+    {
+      "epoch": 1.2171581769436997,
+      "loss": 1.4979586601257324,
+      "loss_ce": 0.29987651109695435,
+      "loss_iou": 0.5426750779151917,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 1.1980820894241333,
+      "num_input_tokens_seen": 549345928,
+      "step": 3178
+    },
+    {
+      "epoch": 1.2175411719647644,
+      "grad_norm": 133.8270785888188,
+      "learning_rate": 5e-06,
+      "loss": 1.7561,
+      "num_input_tokens_seen": 549518784,
+      "step": 3179
+    },
+    {
+      "epoch": 1.2175411719647644,
+      "loss": 1.7928619384765625,
+      "loss_ce": 0.3510966897010803,
+      "loss_iou": 0.661266565322876,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 1.441765308380127,
+      "num_input_tokens_seen": 549518784,
+      "step": 3179
+    },
+    {
+      "epoch": 1.2179241669858292,
+      "grad_norm": 71.61817012756259,
+      "learning_rate": 5e-06,
+      "loss": 1.7947,
+      "num_input_tokens_seen": 549691616,
+      "step": 3180
+    },
+    {
+      "epoch": 1.2179241669858292,
+      "loss": 1.7454323768615723,
+      "loss_ce": 0.28611862659454346,
+      "loss_iou": 0.6721922755241394,
+      "loss_num": 0.02294921875,
+      "loss_xval": 1.4593137502670288,
+      "num_input_tokens_seen": 549691616,
+      "step": 3180
+    },
+    {
+      "epoch": 1.218307162006894,
+      "grad_norm": 174.49661188752026,
+      "learning_rate": 5e-06,
+      "loss": 1.523,
+      "num_input_tokens_seen": 549864648,
+      "step": 3181
+    },
+    {
+      "epoch": 1.218307162006894,
+      "loss": 1.7097549438476562,
+      "loss_ce": 0.31086260080337524,
+      "loss_iou": 0.6387467384338379,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 1.3988924026489258,
+      "num_input_tokens_seen": 549864648,
+      "step": 3181
+    },
+    {
+      "epoch": 1.2186901570279587,
+      "grad_norm": 249.66159079891125,
+      "learning_rate": 5e-06,
+      "loss": 2.0089,
+      "num_input_tokens_seen": 550037816,
+      "step": 3182
+    },
+    {
+      "epoch": 1.2186901570279587,
+      "loss": 2.0109968185424805,
+      "loss_ce": 0.27985674142837524,
+      "loss_iou": 0.7917481660842896,
+      "loss_num": 0.029541015625,
+      "loss_xval": 1.731140375137329,
+      "num_input_tokens_seen": 550037816,
+      "step": 3182
+    },
+    {
+      "epoch": 1.2190731520490234,
+      "grad_norm": 101.54537421742987,
+      "learning_rate": 5e-06,
+      "loss": 1.7719,
+      "num_input_tokens_seen": 550210952,
+      "step": 3183
+    },
+    {
+      "epoch": 1.2190731520490234,
+      "loss": 1.8788511753082275,
+      "loss_ce": 0.31219401955604553,
+      "loss_iou": 0.6646761894226074,
+      "loss_num": 0.04736328125,
+      "loss_xval": 1.5666570663452148,
+      "num_input_tokens_seen": 550210952,
+      "step": 3183
+    },
+    {
+      "epoch": 1.2194561470700882,
+      "grad_norm": 143.63311814092074,
+      "learning_rate": 5e-06,
+      "loss": 1.6131,
+      "num_input_tokens_seen": 550384064,
+      "step": 3184
+    },
+    {
+      "epoch": 1.2194561470700882,
+      "loss": 1.5912206172943115,
+      "loss_ce": 0.30355510115623474,
+      "loss_iou": 0.5518527626991272,
+      "loss_num": 0.036865234375,
+      "loss_xval": 1.2376166582107544,
+      "num_input_tokens_seen": 550384064,
+      "step": 3184
+    },
+    {
+      "epoch": 1.2198391420911527,
+      "grad_norm": 74.71351576334541,
+      "learning_rate": 5e-06,
+      "loss": 2.4814,
+      "num_input_tokens_seen": 550557072,
+      "step": 3185
+    },
+    {
+      "epoch": 1.2198391420911527,
+      "loss": 2.515709400177002,
+      "loss_ce": 0.3064751923084259,
+      "loss_iou": 0.8775662183761597,
+      "loss_num": 0.0908203125,
+      "loss_xval": 2.2092342376708984,
+      "num_input_tokens_seen": 550557072,
+      "step": 3185
+    },
+    {
+      "epoch": 1.2202221371122175,
+      "grad_norm": 106.75768485415921,
+      "learning_rate": 5e-06,
+      "loss": 1.9261,
+      "num_input_tokens_seen": 550730416,
+      "step": 3186
+    },
+    {
+      "epoch": 1.2202221371122175,
+      "loss": 1.9721856117248535,
+      "loss_ce": 0.35608041286468506,
+      "loss_iou": 0.7219014763832092,
+      "loss_num": 0.034423828125,
+      "loss_xval": 1.616105318069458,
+      "num_input_tokens_seen": 550730416,
+      "step": 3186
+    },
+    {
+      "epoch": 1.2206051321332823,
+      "grad_norm": 125.6871573178647,
+      "learning_rate": 5e-06,
+      "loss": 2.1257,
+      "num_input_tokens_seen": 550903288,
+      "step": 3187
+    },
+    {
+      "epoch": 1.2206051321332823,
+      "loss": 2.1850404739379883,
+      "loss_ce": 0.3669430613517761,
+      "loss_iou": 0.8350129723548889,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 1.8180972337722778,
+      "num_input_tokens_seen": 550903288,
+      "step": 3187
+    },
+    {
+      "epoch": 1.220988127154347,
+      "grad_norm": 106.55127289380127,
+      "learning_rate": 5e-06,
+      "loss": 1.6297,
+      "num_input_tokens_seen": 551076160,
+      "step": 3188
+    },
+    {
+      "epoch": 1.220988127154347,
+      "loss": 1.5373339653015137,
+      "loss_ce": 0.3313254714012146,
+      "loss_iou": 0.5609052181243896,
+      "loss_num": 0.016845703125,
+      "loss_xval": 1.2060084342956543,
+      "num_input_tokens_seen": 551076160,
+      "step": 3188
+    },
+    {
+      "epoch": 1.2213711221754118,
+      "grad_norm": 332.88685463480067,
+      "learning_rate": 5e-06,
+      "loss": 2.1437,
+      "num_input_tokens_seen": 551248976,
+      "step": 3189
+    },
+    {
+      "epoch": 1.2213711221754118,
+      "loss": 2.1342363357543945,
+      "loss_ce": 0.35409724712371826,
+      "loss_iou": 0.8193603754043579,
+      "loss_num": 0.0283203125,
+      "loss_xval": 1.7801392078399658,
+      "num_input_tokens_seen": 551248976,
+      "step": 3189
+    },
+    {
+      "epoch": 1.2217541171964765,
+      "grad_norm": 111.80509309635501,
+      "learning_rate": 5e-06,
+      "loss": 2.0058,
+      "num_input_tokens_seen": 551422032,
+      "step": 3190
+    },
+    {
+      "epoch": 1.2217541171964765,
+      "loss": 2.0466792583465576,
+      "loss_ce": 0.3329498767852783,
+      "loss_iou": 0.7603681087493896,
+      "loss_num": 0.03857421875,
+      "loss_xval": 1.7137293815612793,
+      "num_input_tokens_seen": 551422032,
+      "step": 3190
+    },
+    {
+      "epoch": 1.222137112217541,
+      "grad_norm": 243.96988241511502,
+      "learning_rate": 5e-06,
+      "loss": 1.7906,
+      "num_input_tokens_seen": 551594824,
+      "step": 3191
+    },
+    {
+      "epoch": 1.222137112217541,
+      "loss": 1.803696870803833,
+      "loss_ce": 0.3025438189506531,
+      "loss_iou": 0.6702237725257874,
+      "loss_num": 0.0322265625,
+      "loss_xval": 1.5011531114578247,
+      "num_input_tokens_seen": 551594824,
+      "step": 3191
+    },
+    {
+      "epoch": 1.2225201072386058,
+      "grad_norm": 143.00161490878253,
+      "learning_rate": 5e-06,
+      "loss": 2.1548,
+      "num_input_tokens_seen": 551767880,
+      "step": 3192
+    },
+    {
+      "epoch": 1.2225201072386058,
+      "loss": 2.170114517211914,
+      "loss_ce": 0.30388131737709045,
+      "loss_iou": 0.859386146068573,
+      "loss_num": 0.029541015625,
+      "loss_xval": 1.866233229637146,
+      "num_input_tokens_seen": 551767880,
+      "step": 3192
+    },
+    {
+      "epoch": 1.2229031022596706,
+      "grad_norm": 87.0770299486737,
+      "learning_rate": 5e-06,
+      "loss": 1.6753,
+      "num_input_tokens_seen": 551940552,
+      "step": 3193
+    },
+    {
+      "epoch": 1.2229031022596706,
+      "loss": 1.6283884048461914,
+      "loss_ce": 0.3293622136116028,
+      "loss_iou": 0.5896986722946167,
+      "loss_num": 0.02392578125,
+      "loss_xval": 1.2990262508392334,
+      "num_input_tokens_seen": 551940552,
+      "step": 3193
+    },
+    {
+      "epoch": 1.2232860972807353,
+      "grad_norm": 164.9114922094196,
+      "learning_rate": 5e-06,
+      "loss": 1.5698,
+      "num_input_tokens_seen": 552113672,
+      "step": 3194
+    },
+    {
+      "epoch": 1.2232860972807353,
+      "loss": 1.3979899883270264,
+      "loss_ce": 0.3084055185317993,
+      "loss_iou": 0.4936447739601135,
+      "loss_num": 0.0205078125,
+      "loss_xval": 1.089584469795227,
+      "num_input_tokens_seen": 552113672,
+      "step": 3194
+    },
+    {
+      "epoch": 1.2236690923018,
+      "grad_norm": 163.39307516080493,
+      "learning_rate": 5e-06,
+      "loss": 1.7532,
+      "num_input_tokens_seen": 552286464,
+      "step": 3195
+    },
+    {
+      "epoch": 1.2236690923018,
+      "loss": 1.7099965810775757,
+      "loss_ce": 0.2678207755088806,
+      "loss_iou": 0.6490053534507751,
+      "loss_num": 0.02880859375,
+      "loss_xval": 1.4421757459640503,
+      "num_input_tokens_seen": 552286464,
+      "step": 3195
+    },
+    {
+      "epoch": 1.2240520873228649,
+      "grad_norm": 145.08628084374266,
+      "learning_rate": 5e-06,
+      "loss": 2.174,
+      "num_input_tokens_seen": 552459520,
+      "step": 3196
+    },
+    {
+      "epoch": 1.2240520873228649,
+      "loss": 2.1587939262390137,
+      "loss_ce": 0.31603682041168213,
+      "loss_iou": 0.8135904669761658,
+      "loss_num": 0.043212890625,
+      "loss_xval": 1.8427571058273315,
+      "num_input_tokens_seen": 552459520,
+      "step": 3196
+    },
+    {
+      "epoch": 1.2244350823439296,
+      "grad_norm": 62.70519028469807,
+      "learning_rate": 5e-06,
+      "loss": 1.656,
+      "num_input_tokens_seen": 552632120,
+      "step": 3197
+    },
+    {
+      "epoch": 1.2244350823439296,
+      "loss": 1.7491471767425537,
+      "loss_ce": 0.30180686712265015,
+      "loss_iou": 0.6547919511795044,
+      "loss_num": 0.027587890625,
+      "loss_xval": 1.4473402500152588,
+      "num_input_tokens_seen": 552632120,
+      "step": 3197
+    },
+    {
+      "epoch": 1.2248180773649944,
+      "grad_norm": 123.27105193638035,
+      "learning_rate": 5e-06,
+      "loss": 1.5216,
+      "num_input_tokens_seen": 552805312,
+      "step": 3198
+    },
+    {
+      "epoch": 1.2248180773649944,
+      "loss": 1.5703985691070557,
+      "loss_ce": 0.25883591175079346,
+      "loss_iou": 0.5804944634437561,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 1.3115626573562622,
+      "num_input_tokens_seen": 552805312,
+      "step": 3198
+    },
+    {
+      "epoch": 1.225201072386059,
+      "grad_norm": 256.7896731375906,
+      "learning_rate": 5e-06,
+      "loss": 2.2714,
+      "num_input_tokens_seen": 552978072,
+      "step": 3199
+    },
+    {
+      "epoch": 1.225201072386059,
+      "loss": 2.1277894973754883,
+      "loss_ce": 0.25805285573005676,
+      "loss_iou": 0.811699390411377,
+      "loss_num": 0.04931640625,
+      "loss_xval": 1.825791358947754,
+      "num_input_tokens_seen": 552978072,
+      "step": 3199
+    },
+    {
+      "epoch": 1.2255840674071237,
+      "grad_norm": 85.84432593178089,
+      "learning_rate": 5e-06,
+      "loss": 1.858,
+      "num_input_tokens_seen": 553150808,
+      "step": 3200
+    },
+    {
+      "epoch": 1.2255840674071237,
+      "loss": 1.7213298082351685,
+      "loss_ce": 0.3204355835914612,
+      "loss_iou": 0.617317259311676,
+      "loss_num": 0.033203125,
+      "loss_xval": 1.400894284248352,
+      "num_input_tokens_seen": 553150808,
+      "step": 3200
+    },
+    {
+      "epoch": 1.2259670624281884,
+      "grad_norm": 238.3452424198299,
+      "learning_rate": 5e-06,
+      "loss": 1.943,
+      "num_input_tokens_seen": 553323536,
+      "step": 3201
+    },
+    {
+      "epoch": 1.2259670624281884,
+      "loss": 1.7847257852554321,
+      "loss_ce": 0.3421986997127533,
+      "loss_iou": 0.646281898021698,
+      "loss_num": 0.030029296875,
+      "loss_xval": 1.442527174949646,
+      "num_input_tokens_seen": 553323536,
+      "step": 3201
+    },
+    {
+      "epoch": 1.2263500574492532,
+      "grad_norm": 107.85156694617933,
+      "learning_rate": 5e-06,
+      "loss": 2.0012,
+      "num_input_tokens_seen": 553496432,
+      "step": 3202
+    },
+    {
+      "epoch": 1.2263500574492532,
+      "loss": 1.8426074981689453,
+      "loss_ce": 0.30851882696151733,
+      "loss_iou": 0.6797640323638916,
+      "loss_num": 0.034912109375,
+      "loss_xval": 1.5340886116027832,
+      "num_input_tokens_seen": 553496432,
+      "step": 3202
+    },
+    {
+      "epoch": 1.226733052470318,
+      "grad_norm": 132.43472861822866,
+      "learning_rate": 5e-06,
+      "loss": 1.821,
+      "num_input_tokens_seen": 553669560,
+      "step": 3203
+    },
+    {
+      "epoch": 1.226733052470318,
+      "loss": 1.8719840049743652,
+      "loss_ce": 0.2932372987270355,
+      "loss_iou": 0.6697444915771484,
+      "loss_num": 0.0478515625,
+      "loss_xval": 1.5787467956542969,
+      "num_input_tokens_seen": 553669560,
+      "step": 3203
+    },
+    {
+      "epoch": 1.2271160474913827,
+      "grad_norm": 122.07163065642287,
+      "learning_rate": 5e-06,
+      "loss": 1.9149,
+      "num_input_tokens_seen": 553841960,
+      "step": 3204
+    },
+    {
+      "epoch": 1.2271160474913827,
+      "loss": 2.065173864364624,
+      "loss_ce": 0.31117385625839233,
+      "loss_iou": 0.7777567505836487,
+      "loss_num": 0.039794921875,
+      "loss_xval": 1.7539998292922974,
+      "num_input_tokens_seen": 553841960,
+      "step": 3204
+    },
+    {
+      "epoch": 1.2274990425124472,
+      "grad_norm": 117.89645509200975,
+      "learning_rate": 5e-06,
+      "loss": 1.599,
+      "num_input_tokens_seen": 554015008,
+      "step": 3205
+    },
+    {
+      "epoch": 1.2274990425124472,
+      "loss": 1.7536414861679077,
+      "loss_ce": 0.3128109574317932,
+      "loss_iou": 0.6543446779251099,
+      "loss_num": 0.0263671875,
+      "loss_xval": 1.4408304691314697,
+      "num_input_tokens_seen": 554015008,
+      "step": 3205
+    },
+    {
+      "epoch": 1.227882037533512,
+      "grad_norm": 103.8880128453902,
+      "learning_rate": 5e-06,
+      "loss": 1.9502,
+      "num_input_tokens_seen": 554188192,
+      "step": 3206
+    },
+    {
+      "epoch": 1.227882037533512,
+      "loss": 1.9497625827789307,
+      "loss_ce": 0.20692452788352966,
+      "loss_iou": 0.8029986023902893,
+      "loss_num": 0.02734375,
+      "loss_xval": 1.7428380250930786,
+      "num_input_tokens_seen": 554188192,
+      "step": 3206
+    },
+    {
+      "epoch": 1.2282650325545768,
+      "grad_norm": 96.83289943117904,
+      "learning_rate": 5e-06,
+      "loss": 1.5813,
+      "num_input_tokens_seen": 554361096,
+      "step": 3207
+    },
+    {
+      "epoch": 1.2282650325545768,
+      "loss": 1.4702672958374023,
+      "loss_ce": 0.2342482954263687,
+      "loss_iou": 0.5651378631591797,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 1.2360191345214844,
+      "num_input_tokens_seen": 554361096,
+      "step": 3207
+    },
+    {
+      "epoch": 1.2286480275756415,
+      "grad_norm": 119.81877608285815,
+      "learning_rate": 5e-06,
+      "loss": 1.59,
+      "num_input_tokens_seen": 554534456,
+      "step": 3208
+    },
+    {
+      "epoch": 1.2286480275756415,
+      "loss": 1.7240769863128662,
+      "loss_ce": 0.244428813457489,
+      "loss_iou": 0.6755694150924683,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 1.479648232460022,
+      "num_input_tokens_seen": 554534456,
+      "step": 3208
+    },
+    {
+      "epoch": 1.2290310225967063,
+      "grad_norm": 499.2687277835983,
+      "learning_rate": 5e-06,
+      "loss": 2.3241,
+      "num_input_tokens_seen": 554707416,
+      "step": 3209
+    },
+    {
+      "epoch": 1.2290310225967063,
+      "loss": 2.3630847930908203,
+      "loss_ce": 0.31900089979171753,
+      "loss_iou": 0.938149094581604,
+      "loss_num": 0.033447265625,
+      "loss_xval": 2.044083833694458,
+      "num_input_tokens_seen": 554707416,
+      "step": 3209
+    },
+    {
+      "epoch": 1.229414017617771,
+      "grad_norm": 112.24615547428523,
+      "learning_rate": 5e-06,
+      "loss": 2.1574,
+      "num_input_tokens_seen": 554880504,
+      "step": 3210
+    },
+    {
+      "epoch": 1.229414017617771,
+      "loss": 2.1365976333618164,
+      "loss_ce": 0.2381032407283783,
+      "loss_iou": 0.8168008327484131,
+      "loss_num": 0.052978515625,
+      "loss_xval": 1.8984942436218262,
+      "num_input_tokens_seen": 554880504,
+      "step": 3210
+    },
+    {
+      "epoch": 1.2297970126388358,
+      "grad_norm": 152.87935583291184,
+      "learning_rate": 5e-06,
+      "loss": 1.9168,
+      "num_input_tokens_seen": 555053472,
+      "step": 3211
+    },
+    {
+      "epoch": 1.2297970126388358,
+      "loss": 1.924161434173584,
+      "loss_ce": 0.2370298206806183,
+      "loss_iou": 0.706541895866394,
+      "loss_num": 0.054931640625,
+      "loss_xval": 1.687131643295288,
+      "num_input_tokens_seen": 555053472,
+      "step": 3211
+    },
+    {
+      "epoch": 1.2301800076599005,
+      "grad_norm": 130.56666898176,
+      "learning_rate": 5e-06,
+      "loss": 1.758,
+      "num_input_tokens_seen": 555226824,
+      "step": 3212
+    },
+    {
+      "epoch": 1.2301800076599005,
+      "loss": 1.8666596412658691,
+      "loss_ce": 0.3369540572166443,
+      "loss_iou": 0.70448899269104,
+      "loss_num": 0.024169921875,
+      "loss_xval": 1.52970552444458,
+      "num_input_tokens_seen": 555226824,
+      "step": 3212
+    },
+    {
+      "epoch": 1.230563002680965,
+      "grad_norm": 165.93343641993468,
+      "learning_rate": 5e-06,
+      "loss": 2.7272,
+      "num_input_tokens_seen": 555395992,
+      "step": 3213
+    },
+    {
+      "epoch": 1.230563002680965,
+      "loss": 2.9067001342773438,
+      "loss_ce": 0.251071035861969,
+      "loss_iou": 1.2319283485412598,
+      "loss_num": 0.038330078125,
+      "loss_xval": 2.6556291580200195,
+      "num_input_tokens_seen": 555395992,
+      "step": 3213
+    },
+    {
+      "epoch": 1.2309459977020298,
+      "grad_norm": 69.42459640699623,
+      "learning_rate": 5e-06,
+      "loss": 2.2747,
+      "num_input_tokens_seen": 555568824,
+      "step": 3214
+    },
+    {
+      "epoch": 1.2309459977020298,
+      "loss": 2.240429162979126,
+      "loss_ce": 0.26815319061279297,
+      "loss_iou": 0.8390432596206665,
+      "loss_num": 0.058837890625,
+      "loss_xval": 1.972275972366333,
+      "num_input_tokens_seen": 555568824,
+      "step": 3214
+    },
+    {
+      "epoch": 1.2313289927230946,
+      "grad_norm": 130.9276975457229,
+      "learning_rate": 5e-06,
+      "loss": 1.65,
+      "num_input_tokens_seen": 555741224,
+      "step": 3215
+    },
+    {
+      "epoch": 1.2313289927230946,
+      "loss": 1.7605862617492676,
+      "loss_ce": 0.2360796332359314,
+      "loss_iou": 0.6469578742980957,
+      "loss_num": 0.046142578125,
+      "loss_xval": 1.5245065689086914,
+      "num_input_tokens_seen": 555741224,
+      "step": 3215
+    },
+    {
+      "epoch": 1.2317119877441594,
+      "grad_norm": 178.53664350802376,
+      "learning_rate": 5e-06,
+      "loss": 1.8495,
+      "num_input_tokens_seen": 555914120,
+      "step": 3216
+    },
+    {
+      "epoch": 1.2317119877441594,
+      "loss": 1.8208178281784058,
+      "loss_ce": 0.33206358551979065,
+      "loss_iou": 0.6912460327148438,
+      "loss_num": 0.021240234375,
+      "loss_xval": 1.4887542724609375,
+      "num_input_tokens_seen": 555914120,
+      "step": 3216
+    },
+    {
+      "epoch": 1.2320949827652241,
+      "grad_norm": 69.78103362951437,
+      "learning_rate": 5e-06,
+      "loss": 2.0902,
+      "num_input_tokens_seen": 556087096,
+      "step": 3217
+    },
+    {
+      "epoch": 1.2320949827652241,
+      "loss": 2.0778441429138184,
+      "loss_ce": 0.3265916705131531,
+      "loss_iou": 0.7904821634292603,
+      "loss_num": 0.0341796875,
+      "loss_xval": 1.7512524127960205,
+      "num_input_tokens_seen": 556087096,
+      "step": 3217
+    },
+    {
+      "epoch": 1.2324779777862889,
+      "grad_norm": 78.92699984343082,
+      "learning_rate": 5e-06,
+      "loss": 1.6877,
+      "num_input_tokens_seen": 556259840,
+      "step": 3218
+    },
+    {
+      "epoch": 1.2324779777862889,
+      "loss": 1.6588375568389893,
+      "loss_ce": 0.21913450956344604,
+      "loss_iou": 0.612734854221344,
+      "loss_num": 0.042724609375,
+      "loss_xval": 1.439703106880188,
+      "num_input_tokens_seen": 556259840,
+      "step": 3218
+    },
+    {
+      "epoch": 1.2328609728073534,
+      "grad_norm": 72.62503991498532,
+      "learning_rate": 5e-06,
+      "loss": 1.7251,
+      "num_input_tokens_seen": 556433024,
+      "step": 3219
+    },
+    {
+      "epoch": 1.2328609728073534,
+      "loss": 1.6549322605133057,
+      "loss_ce": 0.27027806639671326,
+      "loss_iou": 0.6480461359024048,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 1.3846542835235596,
+      "num_input_tokens_seen": 556433024,
+      "step": 3219
+    },
+    {
+      "epoch": 1.2332439678284182,
+      "grad_norm": 139.84716285832107,
+      "learning_rate": 5e-06,
+      "loss": 1.6814,
+      "num_input_tokens_seen": 556605968,
+      "step": 3220
+    },
+    {
+      "epoch": 1.2332439678284182,
+      "loss": 1.5921721458435059,
+      "loss_ce": 0.23918727040290833,
+      "loss_iou": 0.6415956020355225,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 1.35298490524292,
+      "num_input_tokens_seen": 556605968,
+      "step": 3220
+    },
+    {
+      "epoch": 1.233626962849483,
+      "grad_norm": 80.87651371428406,
+      "learning_rate": 5e-06,
+      "loss": 1.6152,
+      "num_input_tokens_seen": 556779232,
+      "step": 3221
+    },
+    {
+      "epoch": 1.233626962849483,
+      "loss": 1.5735828876495361,
+      "loss_ce": 0.21316048502922058,
+      "loss_iou": 0.6230670213699341,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 1.3604223728179932,
+      "num_input_tokens_seen": 556779232,
+      "step": 3221
+    },
+    {
+      "epoch": 1.2340099578705477,
+      "grad_norm": 120.65197264472344,
+      "learning_rate": 5e-06,
+      "loss": 1.458,
+      "num_input_tokens_seen": 556951912,
+      "step": 3222
+    },
+    {
+      "epoch": 1.2340099578705477,
+      "loss": 1.3882265090942383,
+      "loss_ce": 0.24410924315452576,
+      "loss_iou": 0.5242375731468201,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 1.1441172361373901,
+      "num_input_tokens_seen": 556951912,
+      "step": 3222
+    },
+    {
+      "epoch": 1.2343929528916124,
+      "grad_norm": 224.3041869268561,
+      "learning_rate": 5e-06,
+      "loss": 1.9567,
+      "num_input_tokens_seen": 557124416,
+      "step": 3223
+    },
+    {
+      "epoch": 1.2343929528916124,
+      "loss": 1.889853596687317,
+      "loss_ce": 0.21171405911445618,
+      "loss_iou": 0.7566112875938416,
+      "loss_num": 0.032958984375,
+      "loss_xval": 1.678139567375183,
+      "num_input_tokens_seen": 557124416,
+      "step": 3223
+    },
+    {
+      "epoch": 1.2347759479126772,
+      "grad_norm": 79.16541663282247,
+      "learning_rate": 5e-06,
+      "loss": 1.6871,
+      "num_input_tokens_seen": 557297456,
+      "step": 3224
+    },
+    {
+      "epoch": 1.2347759479126772,
+      "loss": 1.6730278730392456,
+      "loss_ce": 0.27667927742004395,
+      "loss_iou": 0.6475761532783508,
+      "loss_num": 0.020263671875,
+      "loss_xval": 1.3963485956192017,
+      "num_input_tokens_seen": 557297456,
+      "step": 3224
+    },
+    {
+      "epoch": 1.235158942933742,
+      "grad_norm": 135.92474746173508,
+      "learning_rate": 5e-06,
+      "loss": 1.6281,
+      "num_input_tokens_seen": 557470304,
+      "step": 3225
+    },
+    {
+      "epoch": 1.235158942933742,
+      "loss": 1.6265156269073486,
+      "loss_ce": 0.3339425027370453,
+      "loss_iou": 0.6044469475746155,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 1.292573094367981,
+      "num_input_tokens_seen": 557470304,
+      "step": 3225
+    },
+    {
+      "epoch": 1.2355419379548065,
+      "grad_norm": 177.22506327512397,
+      "learning_rate": 5e-06,
+      "loss": 1.8915,
+      "num_input_tokens_seen": 557643144,
+      "step": 3226
+    },
+    {
+      "epoch": 1.2355419379548065,
+      "loss": 1.8765814304351807,
+      "loss_ce": 0.2532196640968323,
+      "loss_iou": 0.7239123582839966,
+      "loss_num": 0.03515625,
+      "loss_xval": 1.6233618259429932,
+      "num_input_tokens_seen": 557643144,
+      "step": 3226
+    },
+    {
+      "epoch": 1.2359249329758712,
+      "grad_norm": 172.39927254746493,
+      "learning_rate": 5e-06,
+      "loss": 1.5542,
+      "num_input_tokens_seen": 557816280,
+      "step": 3227
+    },
+    {
+      "epoch": 1.2359249329758712,
+      "loss": 1.4366008043289185,
+      "loss_ce": 0.23155264556407928,
+      "loss_iou": 0.5630648136138916,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 1.2050480842590332,
+      "num_input_tokens_seen": 557816280,
+      "step": 3227
+    },
+    {
+      "epoch": 1.236307927996936,
+      "grad_norm": 119.64625104571982,
+      "learning_rate": 5e-06,
+      "loss": 1.8928,
+      "num_input_tokens_seen": 557989160,
+      "step": 3228
+    },
+    {
+      "epoch": 1.236307927996936,
+      "loss": 1.8278614282608032,
+      "loss_ce": 0.22566193342208862,
+      "loss_iou": 0.6986521482467651,
+      "loss_num": 0.041015625,
+      "loss_xval": 1.6021993160247803,
+      "num_input_tokens_seen": 557989160,
+      "step": 3228
+    },
+    {
+      "epoch": 1.2366909230180008,
+      "grad_norm": 163.34388612453532,
+      "learning_rate": 5e-06,
+      "loss": 1.6782,
+      "num_input_tokens_seen": 558161800,
+      "step": 3229
+    },
+    {
+      "epoch": 1.2366909230180008,
+      "loss": 1.6708768606185913,
+      "loss_ce": 0.2302100658416748,
+      "loss_iou": 0.6086390614509583,
+      "loss_num": 0.044677734375,
+      "loss_xval": 1.4406667947769165,
+      "num_input_tokens_seen": 558161800,
+      "step": 3229
+    },
+    {
+      "epoch": 1.2370739180390655,
+      "grad_norm": 127.47224003097482,
+      "learning_rate": 5e-06,
+      "loss": 1.5003,
+      "num_input_tokens_seen": 558331208,
+      "step": 3230
+    },
+    {
+      "epoch": 1.2370739180390655,
+      "loss": 1.5581037998199463,
+      "loss_ce": 0.27291542291641235,
+      "loss_iou": 0.5680092573165894,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 1.2605302333831787,
+      "num_input_tokens_seen": 558331208,
+      "step": 3230
+    },
+    {
+      "epoch": 1.2374569130601303,
+      "grad_norm": 305.39788329021883,
+      "learning_rate": 5e-06,
+      "loss": 1.9204,
+      "num_input_tokens_seen": 558504008,
+      "step": 3231
+    },
+    {
+      "epoch": 1.2374569130601303,
+      "loss": 1.757807970046997,
+      "loss_ce": 0.24562327563762665,
+      "loss_iou": 0.669269859790802,
+      "loss_num": 0.03466796875,
+      "loss_xval": 1.512184739112854,
+      "num_input_tokens_seen": 558504008,
+      "step": 3231
+    },
+    {
+      "epoch": 1.237839908081195,
+      "grad_norm": 115.22644563825207,
+      "learning_rate": 5e-06,
+      "loss": 2.0037,
+      "num_input_tokens_seen": 558677088,
+      "step": 3232
+    },
+    {
+      "epoch": 1.237839908081195,
+      "loss": 2.114217758178711,
+      "loss_ce": 0.27020809054374695,
+      "loss_iou": 0.8052446246147156,
+      "loss_num": 0.046630859375,
+      "loss_xval": 1.8440097570419312,
+      "num_input_tokens_seen": 558677088,
+      "step": 3232
+    },
+    {
+      "epoch": 1.2382229031022596,
+      "grad_norm": 129.5269224026582,
+      "learning_rate": 5e-06,
+      "loss": 1.6434,
+      "num_input_tokens_seen": 558850040,
+      "step": 3233
+    },
+    {
+      "epoch": 1.2382229031022596,
+      "loss": 1.7667224407196045,
+      "loss_ce": 0.31137874722480774,
+      "loss_iou": 0.6632645130157471,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 1.4553437232971191,
+      "num_input_tokens_seen": 558850040,
+      "step": 3233
+    },
+    {
+      "epoch": 1.2386058981233243,
+      "grad_norm": 109.59507394125446,
+      "learning_rate": 5e-06,
+      "loss": 2.0253,
+      "num_input_tokens_seen": 559022976,
+      "step": 3234
+    },
+    {
+      "epoch": 1.2386058981233243,
+      "loss": 1.931747555732727,
+      "loss_ce": 0.2941380739212036,
+      "loss_iou": 0.7258481979370117,
+      "loss_num": 0.037109375,
+      "loss_xval": 1.6376094818115234,
+      "num_input_tokens_seen": 559022976,
+      "step": 3234
+    },
+    {
+      "epoch": 1.238988893144389,
+      "grad_norm": 120.38702571982651,
+      "learning_rate": 5e-06,
+      "loss": 1.7104,
+      "num_input_tokens_seen": 559195960,
+      "step": 3235
+    },
+    {
+      "epoch": 1.238988893144389,
+      "loss": 1.7832870483398438,
+      "loss_ce": 0.23368623852729797,
+      "loss_iou": 0.6858416795730591,
+      "loss_num": 0.03564453125,
+      "loss_xval": 1.5215246677398682,
+      "num_input_tokens_seen": 559195960,
+      "step": 3235
+    },
+    {
+      "epoch": 1.2393718881654538,
+      "grad_norm": 122.12165189773208,
+      "learning_rate": 5e-06,
+      "loss": 1.8396,
+      "num_input_tokens_seen": 559369008,
+      "step": 3236
+    },
+    {
+      "epoch": 1.2393718881654538,
+      "loss": 1.6692432165145874,
+      "loss_ce": 0.2788049578666687,
+      "loss_iou": 0.636472761631012,
+      "loss_num": 0.0234375,
+      "loss_xval": 1.390438199043274,
+      "num_input_tokens_seen": 559369008,
+      "step": 3236
+    },
+    {
+      "epoch": 1.2397548831865186,
+      "grad_norm": 87.73278494127817,
+      "learning_rate": 5e-06,
+      "loss": 1.9152,
+      "num_input_tokens_seen": 559541816,
+      "step": 3237
+    },
+    {
+      "epoch": 1.2397548831865186,
+      "loss": 1.9559450149536133,
+      "loss_ce": 0.25496596097946167,
+      "loss_iou": 0.7817944884300232,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 1.7009791135787964,
+      "num_input_tokens_seen": 559541816,
+      "step": 3237
+    },
+    {
+      "epoch": 1.2401378782075834,
+      "grad_norm": 121.25869855775306,
+      "learning_rate": 5e-06,
+      "loss": 1.4474,
+      "num_input_tokens_seen": 559714584,
+      "step": 3238
+    },
+    {
+      "epoch": 1.2401378782075834,
+      "loss": 1.5788640975952148,
+      "loss_ce": 0.24121490120887756,
+      "loss_iou": 0.6191266775131226,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 1.3376491069793701,
+      "num_input_tokens_seen": 559714584,
+      "step": 3238
+    },
+    {
+      "epoch": 1.2405208732286481,
+      "grad_norm": 156.6318429034414,
+      "learning_rate": 5e-06,
+      "loss": 1.6067,
+      "num_input_tokens_seen": 559887576,
+      "step": 3239
+    },
+    {
+      "epoch": 1.2405208732286481,
+      "loss": 1.8070673942565918,
+      "loss_ce": 0.21646954119205475,
+      "loss_iou": 0.7184556722640991,
+      "loss_num": 0.03076171875,
+      "loss_xval": 1.5905978679656982,
+      "num_input_tokens_seen": 559887576,
+      "step": 3239
+    },
+    {
+      "epoch": 1.2409038682497127,
+      "grad_norm": 124.03688040514857,
+      "learning_rate": 5e-06,
+      "loss": 1.4161,
+      "num_input_tokens_seen": 560060584,
+      "step": 3240
+    },
+    {
+      "epoch": 1.2409038682497127,
+      "loss": 1.395925760269165,
+      "loss_ce": 0.2303113490343094,
+      "loss_iou": 0.5364052653312683,
+      "loss_num": 0.0185546875,
+      "loss_xval": 1.1656144857406616,
+      "num_input_tokens_seen": 560060584,
+      "step": 3240
+    },
+    {
+      "epoch": 1.2412868632707774,
+      "grad_norm": 106.66204792652965,
+      "learning_rate": 5e-06,
+      "loss": 1.6025,
+      "num_input_tokens_seen": 560233576,
+      "step": 3241
+    },
+    {
+      "epoch": 1.2412868632707774,
+      "loss": 1.7795209884643555,
+      "loss_ce": 0.22966763377189636,
+      "loss_iou": 0.6916441917419434,
+      "loss_num": 0.033447265625,
+      "loss_xval": 1.5498533248901367,
+      "num_input_tokens_seen": 560233576,
+      "step": 3241
+    },
+    {
+      "epoch": 1.2416698582918422,
+      "grad_norm": 154.84085934143445,
+      "learning_rate": 5e-06,
+      "loss": 1.5263,
+      "num_input_tokens_seen": 560406488,
+      "step": 3242
+    },
+    {
+      "epoch": 1.2416698582918422,
+      "loss": 1.4058425426483154,
+      "loss_ce": 0.22820329666137695,
+      "loss_iou": 0.513380229473114,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 1.1776392459869385,
+      "num_input_tokens_seen": 560406488,
+      "step": 3242
+    },
+    {
+      "epoch": 1.242052853312907,
+      "grad_norm": 115.73958391415185,
+      "learning_rate": 5e-06,
+      "loss": 2.1085,
+      "num_input_tokens_seen": 560579680,
+      "step": 3243
+    },
+    {
+      "epoch": 1.242052853312907,
+      "loss": 2.232184410095215,
+      "loss_ce": 0.25877219438552856,
+      "loss_iou": 0.8758053183555603,
+      "loss_num": 0.04443359375,
+      "loss_xval": 1.9734123945236206,
+      "num_input_tokens_seen": 560579680,
+      "step": 3243
+    },
+    {
+      "epoch": 1.2424358483339717,
+      "grad_norm": 136.69204076216332,
+      "learning_rate": 5e-06,
+      "loss": 1.5209,
+      "num_input_tokens_seen": 560752720,
+      "step": 3244
+    },
+    {
+      "epoch": 1.2424358483339717,
+      "loss": 1.6917695999145508,
+      "loss_ce": 0.3209766447544098,
+      "loss_iou": 0.6305564045906067,
+      "loss_num": 0.02197265625,
+      "loss_xval": 1.3707929849624634,
+      "num_input_tokens_seen": 560752720,
+      "step": 3244
+    },
+    {
+      "epoch": 1.2428188433550365,
+      "grad_norm": 135.49102560240664,
+      "learning_rate": 5e-06,
+      "loss": 1.8771,
+      "num_input_tokens_seen": 560925560,
+      "step": 3245
+    },
+    {
+      "epoch": 1.2428188433550365,
+      "loss": 1.9464309215545654,
+      "loss_ce": 0.2025822401046753,
+      "loss_iou": 0.7859258055686951,
+      "loss_num": 0.034423828125,
+      "loss_xval": 1.7438486814498901,
+      "num_input_tokens_seen": 560925560,
+      "step": 3245
+    },
+    {
+      "epoch": 1.243201838376101,
+      "grad_norm": 135.75860197753198,
+      "learning_rate": 5e-06,
+      "loss": 1.6662,
+      "num_input_tokens_seen": 561098104,
+      "step": 3246
+    },
+    {
+      "epoch": 1.243201838376101,
+      "loss": 1.75368332862854,
+      "loss_ce": 0.23724700510501862,
+      "loss_iou": 0.6847623586654663,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 1.5164363384246826,
+      "num_input_tokens_seen": 561098104,
+      "step": 3246
+    },
+    {
+      "epoch": 1.2435848333971657,
+      "grad_norm": 165.5008227522788,
+      "learning_rate": 5e-06,
+      "loss": 1.658,
+      "num_input_tokens_seen": 561271096,
+      "step": 3247
+    },
+    {
+      "epoch": 1.2435848333971657,
+      "loss": 1.527578353881836,
+      "loss_ce": 0.22737956047058105,
+      "loss_iou": 0.6089922785758972,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 1.3001989126205444,
+      "num_input_tokens_seen": 561271096,
+      "step": 3247
+    },
+    {
+      "epoch": 1.2439678284182305,
+      "grad_norm": 174.5125200656963,
+      "learning_rate": 5e-06,
+      "loss": 2.3443,
+      "num_input_tokens_seen": 561443736,
+      "step": 3248
+    },
+    {
+      "epoch": 1.2439678284182305,
+      "loss": 2.3821632862091064,
+      "loss_ce": 0.24301809072494507,
+      "loss_iou": 0.9132615327835083,
+      "loss_num": 0.0625,
+      "loss_xval": 2.1391451358795166,
+      "num_input_tokens_seen": 561443736,
+      "step": 3248
+    },
+    {
+      "epoch": 1.2443508234392953,
+      "grad_norm": 87.023104835008,
+      "learning_rate": 5e-06,
+      "loss": 1.9802,
+      "num_input_tokens_seen": 561616816,
+      "step": 3249
+    },
+    {
+      "epoch": 1.2443508234392953,
+      "loss": 2.027251720428467,
+      "loss_ce": 0.2520645260810852,
+      "loss_iou": 0.8110250234603882,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 1.7751872539520264,
+      "num_input_tokens_seen": 561616816,
+      "step": 3249
+    },
+    {
+      "epoch": 1.24473381846036,
+      "grad_norm": 134.0405492023822,
+      "learning_rate": 5e-06,
+      "loss": 1.5023,
+      "num_input_tokens_seen": 561789592,
+      "step": 3250
+    },
+    {
+      "epoch": 1.24473381846036,
+      "eval_websight_new_CIoU": 0.8580086827278137,
+      "eval_websight_new_GIoU": 0.8578130304813385,
+      "eval_websight_new_IoU": 0.85853511095047,
+      "eval_websight_new_MAE_all": 0.017037231475114822,
+      "eval_websight_new_MAE_h": 0.01280592754483223,
+      "eval_websight_new_MAE_w": 0.035745494067668915,
+      "eval_websight_new_MAE_x": 0.012769540306180716,
+      "eval_websight_new_MAE_y": 0.0068279586266726255,
+      "eval_websight_new_NUM_probability": 0.022801623679697514,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 1.7849124670028687,
+      "eval_websight_new_loss_ce": 0.5348544716835022,
+      "eval_websight_new_loss_iou": 0.5845544040203094,
+      "eval_websight_new_loss_num": 0.020751953125,
+      "eval_websight_new_loss_xval": 1.272868573665619,
+      "eval_websight_new_runtime": 59.4868,
+      "eval_websight_new_samples_per_second": 0.841,
+      "eval_websight_new_steps_per_second": 0.034,
+      "num_input_tokens_seen": 561789592,
+      "step": 3250
+    },
+    {
+      "epoch": 1.24473381846036,
+      "eval_seeclick_CIoU": 0.5733348727226257,
+      "eval_seeclick_GIoU": 0.5819355547428131,
+      "eval_seeclick_IoU": 0.6091125011444092,
+      "eval_seeclick_MAE_all": 0.06609445624053478,
+      "eval_seeclick_MAE_h": 0.0467582605779171,
+      "eval_seeclick_MAE_w": 0.1045493446290493,
+      "eval_seeclick_MAE_x": 0.0733401719480753,
+      "eval_seeclick_MAE_y": 0.0397300636395812,
+      "eval_seeclick_NUM_probability": 0.06769441440701485,
+      "eval_seeclick_inside_bbox": 0.8819444477558136,
+      "eval_seeclick_loss": 2.5507171154022217,
+      "eval_seeclick_loss_ce": 0.4472598135471344,
+      "eval_seeclick_loss_iou": 0.9121057689189911,
+      "eval_seeclick_loss_num": 0.0566864013671875,
+      "eval_seeclick_loss_xval": 2.107689380645752,
+      "eval_seeclick_runtime": 83.6077,
+      "eval_seeclick_samples_per_second": 0.598,
+      "eval_seeclick_steps_per_second": 0.024,
+      "num_input_tokens_seen": 561789592,
+      "step": 3250
+    },
+    {
+      "epoch": 1.24473381846036,
+      "eval_icons_CIoU": 0.8252148926258087,
+      "eval_icons_GIoU": 0.8298048079013824,
+      "eval_icons_IoU": 0.8399739861488342,
+      "eval_icons_MAE_all": 0.028438608162105083,
+      "eval_icons_MAE_h": 0.017280051484704018,
+      "eval_icons_MAE_w": 0.04606501758098602,
+      "eval_icons_MAE_x": 0.037893764674663544,
+      "eval_icons_MAE_y": 0.012515589827671647,
+      "eval_icons_NUM_probability": 0.23579960316419601,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 2.475114345550537,
+      "eval_icons_loss_ce": 0.41409504413604736,
+      "eval_icons_loss_iou": 0.9459198415279388,
+      "eval_icons_loss_num": 0.016490936279296875,
+      "eval_icons_loss_xval": 1.9742524027824402,
+      "eval_icons_runtime": 82.7171,
+      "eval_icons_samples_per_second": 0.604,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 561789592,
+      "step": 3250
+    },
+    {
+      "epoch": 1.24473381846036,
+      "loss": 2.4445314407348633,
+      "loss_ce": 0.41270190477371216,
+      "loss_iou": 0.9729459285736084,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 2.031829357147217,
+      "num_input_tokens_seen": 561789592,
+      "step": 3250
+    },
+    {
+      "epoch": 1.2451168134814248,
+      "grad_norm": 228.77569955935664,
+      "learning_rate": 5e-06,
+      "loss": 2.2008,
+      "num_input_tokens_seen": 561963064,
+      "step": 3251
+    },
+    {
+      "epoch": 1.2451168134814248,
+      "loss": 2.1139419078826904,
+      "loss_ce": 0.2600467801094055,
+      "loss_iou": 0.8336858749389648,
+      "loss_num": 0.037353515625,
+      "loss_xval": 1.8538951873779297,
+      "num_input_tokens_seen": 561963064,
+      "step": 3251
+    },
+    {
+      "epoch": 1.2454998085024895,
+      "grad_norm": 110.26617648983085,
+      "learning_rate": 5e-06,
+      "loss": 2.0558,
+      "num_input_tokens_seen": 562136384,
+      "step": 3252
+    },
+    {
+      "epoch": 1.2454998085024895,
+      "loss": 2.0282206535339355,
+      "loss_ce": 0.2218882441520691,
+      "loss_iou": 0.7741989493370056,
+      "loss_num": 0.0517578125,
+      "loss_xval": 1.8063324689865112,
+      "num_input_tokens_seen": 562136384,
+      "step": 3252
+    },
+    {
+      "epoch": 1.2458828035235543,
+      "grad_norm": 92.51480806821526,
+      "learning_rate": 5e-06,
+      "loss": 1.7312,
+      "num_input_tokens_seen": 562309504,
+      "step": 3253
+    },
+    {
+      "epoch": 1.2458828035235543,
+      "loss": 1.6316841840744019,
+      "loss_ce": 0.27195367217063904,
+      "loss_iou": 0.5550483465194702,
+      "loss_num": 0.0498046875,
+      "loss_xval": 1.3597304821014404,
+      "num_input_tokens_seen": 562309504,
+      "step": 3253
+    },
+    {
+      "epoch": 1.2462657985446188,
+      "grad_norm": 145.45499441622874,
+      "learning_rate": 5e-06,
+      "loss": 1.8745,
+      "num_input_tokens_seen": 562482232,
+      "step": 3254
+    },
+    {
+      "epoch": 1.2462657985446188,
+      "loss": 1.952561616897583,
+      "loss_ce": 0.26895102858543396,
+      "loss_iou": 0.7191856503486633,
+      "loss_num": 0.049072265625,
+      "loss_xval": 1.6836105585098267,
+      "num_input_tokens_seen": 562482232,
+      "step": 3254
+    },
+    {
+      "epoch": 1.2466487935656836,
+      "grad_norm": 161.5191585467992,
+      "learning_rate": 5e-06,
+      "loss": 2.3314,
+      "num_input_tokens_seen": 562655464,
+      "step": 3255
+    },
+    {
+      "epoch": 1.2466487935656836,
+      "loss": 2.4464144706726074,
+      "loss_ce": 0.22417005896568298,
+      "loss_iou": 1.0176469087600708,
+      "loss_num": 0.037353515625,
+      "loss_xval": 2.2222445011138916,
+      "num_input_tokens_seen": 562655464,
+      "step": 3255
+    },
+    {
+      "epoch": 1.2470317885867483,
+      "grad_norm": 89.43310852326586,
+      "learning_rate": 5e-06,
+      "loss": 2.0123,
+      "num_input_tokens_seen": 562828528,
+      "step": 3256
+    },
+    {
+      "epoch": 1.2470317885867483,
+      "loss": 1.9258875846862793,
+      "loss_ce": 0.25422924757003784,
+      "loss_iou": 0.7314590215682983,
+      "loss_num": 0.041748046875,
+      "loss_xval": 1.6355254650115967,
+      "num_input_tokens_seen": 562828528,
+      "step": 3256
+    },
+    {
+      "epoch": 1.247414783607813,
+      "grad_norm": 84.93228700629363,
+      "learning_rate": 5e-06,
+      "loss": 1.9054,
+      "num_input_tokens_seen": 563001416,
+      "step": 3257
+    },
+    {
+      "epoch": 1.247414783607813,
+      "loss": 1.9564383029937744,
+      "loss_ce": 0.28483855724334717,
+      "loss_iou": 0.7379299402236938,
+      "loss_num": 0.0390625,
+      "loss_xval": 1.6715996265411377,
+      "num_input_tokens_seen": 563001416,
+      "step": 3257
+    },
+    {
+      "epoch": 1.2477977786288779,
+      "grad_norm": 97.46715560652545,
+      "learning_rate": 5e-06,
+      "loss": 1.9426,
+      "num_input_tokens_seen": 563174360,
+      "step": 3258
+    },
+    {
+      "epoch": 1.2477977786288779,
+      "loss": 2.0292422771453857,
+      "loss_ce": 0.318520724773407,
+      "loss_iou": 0.749098539352417,
+      "loss_num": 0.04248046875,
+      "loss_xval": 1.710721492767334,
+      "num_input_tokens_seen": 563174360,
+      "step": 3258
+    },
+    {
+      "epoch": 1.2481807736499426,
+      "grad_norm": 91.15667847541916,
+      "learning_rate": 5e-06,
+      "loss": 1.6545,
+      "num_input_tokens_seen": 563347680,
+      "step": 3259
+    },
+    {
+      "epoch": 1.2481807736499426,
+      "loss": 1.6798195838928223,
+      "loss_ce": 0.2844923138618469,
+      "loss_iou": 0.6500409245491028,
+      "loss_num": 0.01904296875,
+      "loss_xval": 1.3953272104263306,
+      "num_input_tokens_seen": 563347680,
+      "step": 3259
+    },
+    {
+      "epoch": 1.2485637686710072,
+      "grad_norm": 95.97606407986159,
+      "learning_rate": 5e-06,
+      "loss": 1.5331,
+      "num_input_tokens_seen": 563520864,
+      "step": 3260
+    },
+    {
+      "epoch": 1.2485637686710072,
+      "loss": 1.5802524089813232,
+      "loss_ce": 0.25654444098472595,
+      "loss_iou": 0.6218760013580322,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 1.3237080574035645,
+      "num_input_tokens_seen": 563520864,
+      "step": 3260
+    },
+    {
+      "epoch": 1.248946763692072,
+      "grad_norm": 165.89110254922696,
+      "learning_rate": 5e-06,
+      "loss": 1.9111,
+      "num_input_tokens_seen": 563693664,
+      "step": 3261
+    },
+    {
+      "epoch": 1.248946763692072,
+      "loss": 1.8390283584594727,
+      "loss_ce": 0.2378949224948883,
+      "loss_iou": 0.7517080307006836,
+      "loss_num": 0.01953125,
+      "loss_xval": 1.6011333465576172,
+      "num_input_tokens_seen": 563693664,
+      "step": 3261
+    },
+    {
+      "epoch": 1.2493297587131367,
+      "grad_norm": 189.87879349929213,
+      "learning_rate": 5e-06,
+      "loss": 1.706,
+      "num_input_tokens_seen": 563866592,
+      "step": 3262
+    },
+    {
+      "epoch": 1.2493297587131367,
+      "loss": 1.602787971496582,
+      "loss_ce": 0.239871546626091,
+      "loss_iou": 0.6360023021697998,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 1.3629164695739746,
+      "num_input_tokens_seen": 563866592,
+      "step": 3262
+    },
+    {
+      "epoch": 1.2497127537342014,
+      "grad_norm": 136.87018797305842,
+      "learning_rate": 5e-06,
+      "loss": 1.7997,
+      "num_input_tokens_seen": 564039712,
+      "step": 3263
+    },
+    {
+      "epoch": 1.2497127537342014,
+      "loss": 1.919539213180542,
+      "loss_ce": 0.24399109184741974,
+      "loss_iou": 0.7718561291694641,
+      "loss_num": 0.0263671875,
+      "loss_xval": 1.6755481958389282,
+      "num_input_tokens_seen": 564039712,
+      "step": 3263
+    },
+    {
+      "epoch": 1.2500957487552662,
+      "grad_norm": 145.61809969199768,
+      "learning_rate": 5e-06,
+      "loss": 1.5724,
+      "num_input_tokens_seen": 564212800,
+      "step": 3264
+    },
+    {
+      "epoch": 1.2500957487552662,
+      "loss": 1.554048776626587,
+      "loss_ce": 0.28291055560112,
+      "loss_iou": 0.5928444862365723,
+      "loss_num": 0.01708984375,
+      "loss_xval": 1.2711381912231445,
+      "num_input_tokens_seen": 564212800,
+      "step": 3264
+    },
+    {
+      "epoch": 1.250478743776331,
+      "grad_norm": 304.4632897757661,
+      "learning_rate": 5e-06,
+      "loss": 2.0111,
+      "num_input_tokens_seen": 564385776,
+      "step": 3265
+    },
+    {
+      "epoch": 1.250478743776331,
+      "loss": 2.1204137802124023,
+      "loss_ce": 0.21181172132492065,
+      "loss_iou": 0.8808146715164185,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 1.908601999282837,
+      "num_input_tokens_seen": 564385776,
+      "step": 3265
+    },
+    {
+      "epoch": 1.2508617387973957,
+      "grad_norm": 120.73757773626873,
+      "learning_rate": 5e-06,
+      "loss": 2.014,
+      "num_input_tokens_seen": 564558968,
+      "step": 3266
+    },
+    {
+      "epoch": 1.2508617387973957,
+      "loss": 1.9903008937835693,
+      "loss_ce": 0.27592355012893677,
+      "loss_iou": 0.7686266899108887,
+      "loss_num": 0.035400390625,
+      "loss_xval": 1.7143774032592773,
+      "num_input_tokens_seen": 564558968,
+      "step": 3266
+    },
+    {
+      "epoch": 1.2512447338184605,
+      "grad_norm": 118.76541949402854,
+      "learning_rate": 5e-06,
+      "loss": 1.5929,
+      "num_input_tokens_seen": 564732080,
+      "step": 3267
+    },
+    {
+      "epoch": 1.2512447338184605,
+      "loss": 1.749008297920227,
+      "loss_ce": 0.26510366797447205,
+      "loss_iou": 0.6826719045639038,
+      "loss_num": 0.023681640625,
+      "loss_xval": 1.4839046001434326,
+      "num_input_tokens_seen": 564732080,
+      "step": 3267
+    },
+    {
+      "epoch": 1.2516277288395252,
+      "grad_norm": 166.23823548579298,
+      "learning_rate": 5e-06,
+      "loss": 1.8225,
+      "num_input_tokens_seen": 564905120,
+      "step": 3268
+    },
+    {
+      "epoch": 1.2516277288395252,
+      "loss": 1.8056085109710693,
+      "loss_ce": 0.2737082839012146,
+      "loss_iou": 0.683491587638855,
+      "loss_num": 0.032958984375,
+      "loss_xval": 1.53190016746521,
+      "num_input_tokens_seen": 564905120,
+      "step": 3268
+    },
+    {
+      "epoch": 1.2520107238605898,
+      "grad_norm": 192.34750726691107,
+      "learning_rate": 5e-06,
+      "loss": 1.6959,
+      "num_input_tokens_seen": 565078176,
+      "step": 3269
+    },
+    {
+      "epoch": 1.2520107238605898,
+      "loss": 1.9059031009674072,
+      "loss_ce": 0.29611802101135254,
+      "loss_iou": 0.7298498153686523,
+      "loss_num": 0.030029296875,
+      "loss_xval": 1.6097850799560547,
+      "num_input_tokens_seen": 565078176,
+      "step": 3269
+    },
+    {
+      "epoch": 1.2523937188816545,
+      "grad_norm": 61.299248319254446,
+      "learning_rate": 5e-06,
+      "loss": 1.9484,
+      "num_input_tokens_seen": 565250432,
+      "step": 3270
+    },
+    {
+      "epoch": 1.2523937188816545,
+      "loss": 2.06618070602417,
+      "loss_ce": 0.22253888845443726,
+      "loss_iou": 0.789618730545044,
+      "loss_num": 0.052978515625,
+      "loss_xval": 1.843641757965088,
+      "num_input_tokens_seen": 565250432,
+      "step": 3270
+    },
+    {
+      "epoch": 1.2527767139027193,
+      "grad_norm": 156.37870387188582,
+      "learning_rate": 5e-06,
+      "loss": 2.0245,
+      "num_input_tokens_seen": 565423752,
+      "step": 3271
+    },
+    {
+      "epoch": 1.2527767139027193,
+      "loss": 2.005537509918213,
+      "loss_ce": 0.19682714343070984,
+      "loss_iou": 0.8074923157691956,
+      "loss_num": 0.038818359375,
+      "loss_xval": 1.8087102174758911,
+      "num_input_tokens_seen": 565423752,
+      "step": 3271
+    },
+    {
+      "epoch": 1.253159708923784,
+      "grad_norm": 196.4282815454342,
+      "learning_rate": 5e-06,
+      "loss": 1.6765,
+      "num_input_tokens_seen": 565596736,
+      "step": 3272
+    },
+    {
+      "epoch": 1.253159708923784,
+      "loss": 1.6703040599822998,
+      "loss_ce": 0.20657575130462646,
+      "loss_iou": 0.6687232255935669,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 1.4637281894683838,
+      "num_input_tokens_seen": 565596736,
+      "step": 3272
+    },
+    {
+      "epoch": 1.2535427039448488,
+      "grad_norm": 336.0777691554218,
+      "learning_rate": 5e-06,
+      "loss": 1.6791,
+      "num_input_tokens_seen": 565769752,
+      "step": 3273
+    },
+    {
+      "epoch": 1.2535427039448488,
+      "loss": 1.6296412944793701,
+      "loss_ce": 0.2718214988708496,
+      "loss_iou": 0.6263586282730103,
+      "loss_num": 0.02099609375,
+      "loss_xval": 1.3578197956085205,
+      "num_input_tokens_seen": 565769752,
+      "step": 3273
+    },
+    {
+      "epoch": 1.2539256989659133,
+      "grad_norm": 125.3995037519553,
+      "learning_rate": 5e-06,
+      "loss": 1.7241,
+      "num_input_tokens_seen": 565942672,
+      "step": 3274
+    },
+    {
+      "epoch": 1.2539256989659133,
+      "loss": 1.6929802894592285,
+      "loss_ce": 0.2351137101650238,
+      "loss_iou": 0.6609401702880859,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 1.4578666687011719,
+      "num_input_tokens_seen": 565942672,
+      "step": 3274
+    },
+    {
+      "epoch": 1.254308693986978,
+      "grad_norm": 159.5506450773098,
+      "learning_rate": 5e-06,
+      "loss": 1.5594,
+      "num_input_tokens_seen": 566115584,
+      "step": 3275
+    },
+    {
+      "epoch": 1.254308693986978,
+      "loss": 1.4766712188720703,
+      "loss_ce": 0.17369934916496277,
+      "loss_iou": 0.5988736152648926,
+      "loss_num": 0.02099609375,
+      "loss_xval": 1.2925348281860352,
+      "num_input_tokens_seen": 566115584,
+      "step": 3275
+    },
+    {
+      "epoch": 1.2546916890080428,
+      "grad_norm": 169.4832582524888,
+      "learning_rate": 5e-06,
+      "loss": 1.7684,
+      "num_input_tokens_seen": 566288688,
+      "step": 3276
+    },
+    {
+      "epoch": 1.2546916890080428,
+      "loss": 1.7170007228851318,
+      "loss_ce": 0.2073092758655548,
+      "loss_iou": 0.6677485704421997,
+      "loss_num": 0.034912109375,
+      "loss_xval": 1.5096914768218994,
+      "num_input_tokens_seen": 566288688,
+      "step": 3276
+    },
+    {
+      "epoch": 1.2550746840291076,
+      "grad_norm": 442.3508245704993,
+      "learning_rate": 5e-06,
+      "loss": 2.0786,
+      "num_input_tokens_seen": 566461920,
+      "step": 3277
+    },
+    {
+      "epoch": 1.2550746840291076,
+      "loss": 1.9707210063934326,
+      "loss_ce": 0.29288074374198914,
+      "loss_iou": 0.7639079093933105,
+      "loss_num": 0.030029296875,
+      "loss_xval": 1.677840232849121,
+      "num_input_tokens_seen": 566461920,
+      "step": 3277
+    },
+    {
+      "epoch": 1.2554576790501724,
+      "grad_norm": 76.01274657392398,
+      "learning_rate": 5e-06,
+      "loss": 2.1564,
+      "num_input_tokens_seen": 566634792,
+      "step": 3278
+    },
+    {
+      "epoch": 1.2554576790501724,
+      "loss": 2.133183240890503,
+      "loss_ce": 0.20118530094623566,
+      "loss_iou": 0.8348344564437866,
+      "loss_num": 0.052490234375,
+      "loss_xval": 1.9319980144500732,
+      "num_input_tokens_seen": 566634792,
+      "step": 3278
+    },
+    {
+      "epoch": 1.2558406740712371,
+      "grad_norm": 237.51501335104388,
+      "learning_rate": 5e-06,
+      "loss": 2.0103,
+      "num_input_tokens_seen": 566808032,
+      "step": 3279
+    },
+    {
+      "epoch": 1.2558406740712371,
+      "loss": 1.9694063663482666,
+      "loss_ce": 0.22570320963859558,
+      "loss_iou": 0.6653085947036743,
+      "loss_num": 0.08251953125,
+      "loss_xval": 1.7437031269073486,
+      "num_input_tokens_seen": 566808032,
+      "step": 3279
+    },
+    {
+      "epoch": 1.2562236690923019,
+      "grad_norm": 154.0387832752544,
+      "learning_rate": 5e-06,
+      "loss": 1.588,
+      "num_input_tokens_seen": 566980776,
+      "step": 3280
+    },
+    {
+      "epoch": 1.2562236690923019,
+      "loss": 1.5974631309509277,
+      "loss_ce": 0.25681233406066895,
+      "loss_iou": 0.6179877519607544,
+      "loss_num": 0.02099609375,
+      "loss_xval": 1.3284437656402588,
+      "num_input_tokens_seen": 566980776,
+      "step": 3280
+    },
+    {
+      "epoch": 1.2566066641133666,
+      "grad_norm": 228.31667108048424,
+      "learning_rate": 5e-06,
+      "loss": 2.1985,
+      "num_input_tokens_seen": 567153464,
+      "step": 3281
+    },
+    {
+      "epoch": 1.2566066641133666,
+      "loss": 2.1468119621276855,
+      "loss_ce": 0.2339622676372528,
+      "loss_iou": 0.8826028108596802,
+      "loss_num": 0.029541015625,
+      "loss_xval": 1.9128496646881104,
+      "num_input_tokens_seen": 567153464,
+      "step": 3281
+    },
+    {
+      "epoch": 1.2569896591344312,
+      "grad_norm": 95.64891904610043,
+      "learning_rate": 5e-06,
+      "loss": 2.2372,
+      "num_input_tokens_seen": 567326832,
+      "step": 3282
+    },
+    {
+      "epoch": 1.2569896591344312,
+      "loss": 2.37243390083313,
+      "loss_ce": 0.25682705640792847,
+      "loss_iou": 0.9149811267852783,
+      "loss_num": 0.05712890625,
+      "loss_xval": 2.1156067848205566,
+      "num_input_tokens_seen": 567326832,
+      "step": 3282
+    },
+    {
+      "epoch": 1.257372654155496,
+      "grad_norm": 91.41657487146114,
+      "learning_rate": 5e-06,
+      "loss": 1.6062,
+      "num_input_tokens_seen": 567499528,
+      "step": 3283
+    },
+    {
+      "epoch": 1.257372654155496,
+      "loss": 1.4798704385757446,
+      "loss_ce": 0.25309568643569946,
+      "loss_iou": 0.54359370470047,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 1.22677481174469,
+      "num_input_tokens_seen": 567499528,
+      "step": 3283
+    },
+    {
+      "epoch": 1.2577556491765607,
+      "grad_norm": 238.74094096290244,
+      "learning_rate": 5e-06,
+      "loss": 1.9775,
+      "num_input_tokens_seen": 567672440,
+      "step": 3284
+    },
+    {
+      "epoch": 1.2577556491765607,
+      "loss": 1.8096680641174316,
+      "loss_ce": 0.24874097108840942,
+      "loss_iou": 0.7278817892074585,
+      "loss_num": 0.02099609375,
+      "loss_xval": 1.560927152633667,
+      "num_input_tokens_seen": 567672440,
+      "step": 3284
+    },
+    {
+      "epoch": 1.2581386441976254,
+      "grad_norm": 159.01997976887932,
+      "learning_rate": 5e-06,
+      "loss": 1.7992,
+      "num_input_tokens_seen": 567845352,
+      "step": 3285
+    },
+    {
+      "epoch": 1.2581386441976254,
+      "loss": 1.9142980575561523,
+      "loss_ce": 0.26271456480026245,
+      "loss_iou": 0.7631086111068726,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 1.6515834331512451,
+      "num_input_tokens_seen": 567845352,
+      "step": 3285
+    },
+    {
+      "epoch": 1.2585216392186902,
+      "grad_norm": 181.6217377870929,
+      "learning_rate": 5e-06,
+      "loss": 1.6094,
+      "num_input_tokens_seen": 568018296,
+      "step": 3286
+    },
+    {
+      "epoch": 1.2585216392186902,
+      "loss": 1.5701494216918945,
+      "loss_ce": 0.26261407136917114,
+      "loss_iou": 0.5949907898902893,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 1.3075352907180786,
+      "num_input_tokens_seen": 568018296,
+      "step": 3286
+    },
+    {
+      "epoch": 1.258904634239755,
+      "grad_norm": 106.04391570828672,
+      "learning_rate": 5e-06,
+      "loss": 1.8732,
+      "num_input_tokens_seen": 568191448,
+      "step": 3287
+    },
+    {
+      "epoch": 1.258904634239755,
+      "loss": 1.8496975898742676,
+      "loss_ce": 0.2655804753303528,
+      "loss_iou": 0.7483574151992798,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 1.5841171741485596,
+      "num_input_tokens_seen": 568191448,
+      "step": 3287
+    },
+    {
+      "epoch": 1.2592876292608195,
+      "grad_norm": 77.93709908838132,
+      "learning_rate": 5e-06,
+      "loss": 1.4416,
+      "num_input_tokens_seen": 568364240,
+      "step": 3288
+    },
+    {
+      "epoch": 1.2592876292608195,
+      "loss": 1.5831642150878906,
+      "loss_ce": 0.28663915395736694,
+      "loss_iou": 0.5807881355285645,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 1.296525001525879,
+      "num_input_tokens_seen": 568364240,
+      "step": 3288
+    },
+    {
+      "epoch": 1.2596706242818843,
+      "grad_norm": 92.30320114278938,
+      "learning_rate": 5e-06,
+      "loss": 1.5375,
+      "num_input_tokens_seen": 568537336,
+      "step": 3289
+    },
+    {
+      "epoch": 1.2596706242818843,
+      "loss": 1.4855799674987793,
+      "loss_ce": 0.258095383644104,
+      "loss_iou": 0.5723146796226501,
+      "loss_num": 0.0166015625,
+      "loss_xval": 1.2274845838546753,
+      "num_input_tokens_seen": 568537336,
+      "step": 3289
+    },
+    {
+      "epoch": 1.260053619302949,
+      "grad_norm": 82.06844638151883,
+      "learning_rate": 5e-06,
+      "loss": 1.6525,
+      "num_input_tokens_seen": 568710128,
+      "step": 3290
+    },
+    {
+      "epoch": 1.260053619302949,
+      "loss": 1.7507693767547607,
+      "loss_ce": 0.2482771873474121,
+      "loss_iou": 0.6878000497817993,
+      "loss_num": 0.025390625,
+      "loss_xval": 1.5024921894073486,
+      "num_input_tokens_seen": 568710128,
+      "step": 3290
+    },
+    {
+      "epoch": 1.2604366143240138,
+      "grad_norm": 222.3930668374789,
+      "learning_rate": 5e-06,
+      "loss": 1.6031,
+      "num_input_tokens_seen": 568883000,
+      "step": 3291
+    },
+    {
+      "epoch": 1.2604366143240138,
+      "loss": 1.6394113302230835,
+      "loss_ce": 0.18138174712657928,
+      "loss_iou": 0.6749376654624939,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 1.4580296277999878,
+      "num_input_tokens_seen": 568883000,
+      "step": 3291
+    },
+    {
+      "epoch": 1.2608196093450785,
+      "grad_norm": 114.01998218747941,
+      "learning_rate": 5e-06,
+      "loss": 1.867,
+      "num_input_tokens_seen": 569056168,
+      "step": 3292
+    },
+    {
+      "epoch": 1.2608196093450785,
+      "loss": 1.9098315238952637,
+      "loss_ce": 0.24824270606040955,
+      "loss_iou": 0.7440940141677856,
+      "loss_num": 0.03466796875,
+      "loss_xval": 1.6615889072418213,
+      "num_input_tokens_seen": 569056168,
+      "step": 3292
+    },
+    {
+      "epoch": 1.2612026043661433,
+      "grad_norm": 92.28780702393767,
+      "learning_rate": 5e-06,
+      "loss": 1.5238,
+      "num_input_tokens_seen": 569229304,
+      "step": 3293
+    },
+    {
+      "epoch": 1.2612026043661433,
+      "loss": 1.4576985836029053,
+      "loss_ce": 0.23735719919204712,
+      "loss_iou": 0.549181342124939,
+      "loss_num": 0.0244140625,
+      "loss_xval": 1.220341444015503,
+      "num_input_tokens_seen": 569229304,
+      "step": 3293
+    },
+    {
+      "epoch": 1.261585599387208,
+      "grad_norm": 158.6336337533615,
+      "learning_rate": 5e-06,
+      "loss": 1.5514,
+      "num_input_tokens_seen": 569402488,
+      "step": 3294
+    },
+    {
+      "epoch": 1.261585599387208,
+      "loss": 1.596184492111206,
+      "loss_ce": 0.21201471984386444,
+      "loss_iou": 0.6427226662635803,
+      "loss_num": 0.019775390625,
+      "loss_xval": 1.3841696977615356,
+      "num_input_tokens_seen": 569402488,
+      "step": 3294
+    },
+    {
+      "epoch": 1.2619685944082728,
+      "grad_norm": 157.9328062451583,
+      "learning_rate": 5e-06,
+      "loss": 1.9171,
+      "num_input_tokens_seen": 569575648,
+      "step": 3295
+    },
+    {
+      "epoch": 1.2619685944082728,
+      "loss": 1.813136100769043,
+      "loss_ce": 0.2165515124797821,
+      "loss_iou": 0.739820659160614,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 1.596584677696228,
+      "num_input_tokens_seen": 569575648,
+      "step": 3295
+    },
+    {
+      "epoch": 1.2623515894293373,
+      "grad_norm": 71.1308307861241,
+      "learning_rate": 5e-06,
+      "loss": 2.1265,
+      "num_input_tokens_seen": 569748648,
+      "step": 3296
+    },
+    {
+      "epoch": 1.2623515894293373,
+      "loss": 2.096177339553833,
+      "loss_ce": 0.23352748155593872,
+      "loss_iou": 0.80034339427948,
+      "loss_num": 0.05224609375,
+      "loss_xval": 1.86264967918396,
+      "num_input_tokens_seen": 569748648,
+      "step": 3296
+    },
+    {
+      "epoch": 1.262734584450402,
+      "grad_norm": 307.3840565641425,
+      "learning_rate": 5e-06,
+      "loss": 1.9823,
+      "num_input_tokens_seen": 569921568,
+      "step": 3297
+    },
+    {
+      "epoch": 1.262734584450402,
+      "loss": 2.0598111152648926,
+      "loss_ce": 0.2661011815071106,
+      "loss_iou": 0.8003278374671936,
+      "loss_num": 0.03857421875,
+      "loss_xval": 1.7937098741531372,
+      "num_input_tokens_seen": 569921568,
+      "step": 3297
+    },
+    {
+      "epoch": 1.2631175794714669,
+      "grad_norm": 116.75184479233158,
+      "learning_rate": 5e-06,
+      "loss": 1.894,
+      "num_input_tokens_seen": 570094896,
+      "step": 3298
+    },
+    {
+      "epoch": 1.2631175794714669,
+      "loss": 1.9589283466339111,
+      "loss_ce": 0.22966188192367554,
+      "loss_iou": 0.7824494242668152,
+      "loss_num": 0.032958984375,
+      "loss_xval": 1.7292665243148804,
+      "num_input_tokens_seen": 570094896,
+      "step": 3298
+    },
+    {
+      "epoch": 1.2635005744925316,
+      "grad_norm": 202.01678384155286,
+      "learning_rate": 5e-06,
+      "loss": 1.8331,
+      "num_input_tokens_seen": 570267528,
+      "step": 3299
+    },
+    {
+      "epoch": 1.2635005744925316,
+      "loss": 1.9015049934387207,
+      "loss_ce": 0.2262372225522995,
+      "loss_iou": 0.7774227261543274,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 1.6752678155899048,
+      "num_input_tokens_seen": 570267528,
+      "step": 3299
+    },
+    {
+      "epoch": 1.2638835695135964,
+      "grad_norm": 152.66790830149523,
+      "learning_rate": 5e-06,
+      "loss": 1.6035,
+      "num_input_tokens_seen": 570440568,
+      "step": 3300
+    },
+    {
+      "epoch": 1.2638835695135964,
+      "loss": 1.636314034461975,
+      "loss_ce": 0.24354144930839539,
+      "loss_iou": 0.6540737152099609,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 1.3927726745605469,
+      "num_input_tokens_seen": 570440568,
+      "step": 3300
+    },
+    {
+      "epoch": 1.2642665645346611,
+      "grad_norm": 425.4893270942091,
+      "learning_rate": 5e-06,
+      "loss": 1.8108,
+      "num_input_tokens_seen": 570613776,
+      "step": 3301
+    },
+    {
+      "epoch": 1.2642665645346611,
+      "loss": 1.7876484394073486,
+      "loss_ce": 0.22738194465637207,
+      "loss_iou": 0.7047243118286133,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 1.5602664947509766,
+      "num_input_tokens_seen": 570613776,
+      "step": 3301
+    },
+    {
+      "epoch": 1.2646495595557257,
+      "grad_norm": 79.07818183845008,
+      "learning_rate": 5e-06,
+      "loss": 2.1274,
+      "num_input_tokens_seen": 570787008,
+      "step": 3302
+    },
+    {
+      "epoch": 1.2646495595557257,
+      "loss": 2.1811628341674805,
+      "loss_ce": 0.20051544904708862,
+      "loss_iou": 0.8656898736953735,
+      "loss_num": 0.0498046875,
+      "loss_xval": 1.980647325515747,
+      "num_input_tokens_seen": 570787008,
+      "step": 3302
+    },
+    {
+      "epoch": 1.2650325545767904,
+      "grad_norm": 112.437851888758,
+      "learning_rate": 5e-06,
+      "loss": 1.5488,
+      "num_input_tokens_seen": 570956728,
+      "step": 3303
+    },
+    {
+      "epoch": 1.2650325545767904,
+      "loss": 1.6475396156311035,
+      "loss_ce": 0.2554776966571808,
+      "loss_iou": 0.6174481511116028,
+      "loss_num": 0.031494140625,
+      "loss_xval": 1.3920618295669556,
+      "num_input_tokens_seen": 570956728,
+      "step": 3303
+    },
+    {
+      "epoch": 1.2654155495978552,
+      "grad_norm": 169.52040758555097,
+      "learning_rate": 5e-06,
+      "loss": 2.4792,
+      "num_input_tokens_seen": 571130008,
+      "step": 3304
+    },
+    {
+      "epoch": 1.2654155495978552,
+      "loss": 2.4361348152160645,
+      "loss_ce": 0.28582724928855896,
+      "loss_iou": 0.9875988960266113,
+      "loss_num": 0.03515625,
+      "loss_xval": 2.1503076553344727,
+      "num_input_tokens_seen": 571130008,
+      "step": 3304
+    },
+    {
+      "epoch": 1.26579854461892,
+      "grad_norm": 83.45002335813479,
+      "learning_rate": 5e-06,
+      "loss": 2.1598,
+      "num_input_tokens_seen": 571302904,
+      "step": 3305
+    },
+    {
+      "epoch": 1.26579854461892,
+      "loss": 2.2095844745635986,
+      "loss_ce": 0.222993403673172,
+      "loss_iou": 0.8178194761276245,
+      "loss_num": 0.0703125,
+      "loss_xval": 1.9865909814834595,
+      "num_input_tokens_seen": 571302904,
+      "step": 3305
+    },
+    {
+      "epoch": 1.2661815396399847,
+      "grad_norm": 75.82920681965383,
+      "learning_rate": 5e-06,
+      "loss": 1.6185,
+      "num_input_tokens_seen": 571475680,
+      "step": 3306
+    },
+    {
+      "epoch": 1.2661815396399847,
+      "loss": 1.6119019985198975,
+      "loss_ce": 0.2990841865539551,
+      "loss_iou": 0.5760866403579712,
+      "loss_num": 0.031982421875,
+      "loss_xval": 1.3128178119659424,
+      "num_input_tokens_seen": 571475680,
+      "step": 3306
+    },
+    {
+      "epoch": 1.2665645346610495,
+      "grad_norm": 143.68126024410492,
+      "learning_rate": 5e-06,
+      "loss": 1.8082,
+      "num_input_tokens_seen": 571648600,
+      "step": 3307
+    },
+    {
+      "epoch": 1.2665645346610495,
+      "loss": 2.061361789703369,
+      "loss_ce": 0.3056067228317261,
+      "loss_iou": 0.8106932044029236,
+      "loss_num": 0.02685546875,
+      "loss_xval": 1.7557553052902222,
+      "num_input_tokens_seen": 571648600,
+      "step": 3307
+    },
+    {
+      "epoch": 1.2669475296821142,
+      "grad_norm": 179.8805817421556,
+      "learning_rate": 5e-06,
+      "loss": 1.7957,
+      "num_input_tokens_seen": 571821520,
+      "step": 3308
+    },
+    {
+      "epoch": 1.2669475296821142,
+      "loss": 1.7377934455871582,
+      "loss_ce": 0.19629667699337006,
+      "loss_iou": 0.709514856338501,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 1.506584644317627,
+      "num_input_tokens_seen": 571821520,
+      "step": 3308
+    },
+    {
+      "epoch": 1.267330524703179,
+      "grad_norm": 84.08926843473174,
+      "learning_rate": 5e-06,
+      "loss": 1.6271,
+      "num_input_tokens_seen": 571994152,
+      "step": 3309
+    },
+    {
+      "epoch": 1.267330524703179,
+      "loss": 1.6579076051712036,
+      "loss_ce": 0.2033272236585617,
+      "loss_iou": 0.6503553986549377,
+      "loss_num": 0.03076171875,
+      "loss_xval": 1.4545804262161255,
+      "num_input_tokens_seen": 571994152,
+      "step": 3309
+    },
+    {
+      "epoch": 1.2677135197242435,
+      "grad_norm": 216.91039107143519,
+      "learning_rate": 5e-06,
+      "loss": 1.6531,
+      "num_input_tokens_seen": 572166848,
+      "step": 3310
+    },
+    {
+      "epoch": 1.2677135197242435,
+      "loss": 1.5964694023132324,
+      "loss_ce": 0.20418967306613922,
+      "loss_iou": 0.658084511756897,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 1.392279863357544,
+      "num_input_tokens_seen": 572166848,
+      "step": 3310
+    },
+    {
+      "epoch": 1.2680965147453083,
+      "grad_norm": 109.1279957353923,
+      "learning_rate": 5e-06,
+      "loss": 2.0814,
+      "num_input_tokens_seen": 572340224,
+      "step": 3311
+    },
+    {
+      "epoch": 1.2680965147453083,
+      "loss": 1.9804760217666626,
+      "loss_ce": 0.26399368047714233,
+      "loss_iou": 0.762598991394043,
+      "loss_num": 0.038330078125,
+      "loss_xval": 1.716482162475586,
+      "num_input_tokens_seen": 572340224,
+      "step": 3311
+    },
+    {
+      "epoch": 1.268479509766373,
+      "grad_norm": 107.37991499709248,
+      "learning_rate": 5e-06,
+      "loss": 1.7606,
+      "num_input_tokens_seen": 572513376,
+      "step": 3312
+    },
+    {
+      "epoch": 1.268479509766373,
+      "loss": 1.807293176651001,
+      "loss_ce": 0.22050227224826813,
+      "loss_iou": 0.7271722555160522,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 1.5867908000946045,
+      "num_input_tokens_seen": 572513376,
+      "step": 3312
+    },
+    {
+      "epoch": 1.2688625047874378,
+      "grad_norm": 91.16603981539338,
+      "learning_rate": 5e-06,
+      "loss": 1.9189,
+      "num_input_tokens_seen": 572686184,
+      "step": 3313
+    },
+    {
+      "epoch": 1.2688625047874378,
+      "loss": 2.00659441947937,
+      "loss_ce": 0.21515393257141113,
+      "loss_iou": 0.8299853801727295,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 1.791440486907959,
+      "num_input_tokens_seen": 572686184,
+      "step": 3313
+    },
+    {
+      "epoch": 1.2692454998085025,
+      "grad_norm": 101.56326843579461,
+      "learning_rate": 5e-06,
+      "loss": 1.8273,
+      "num_input_tokens_seen": 572859016,
+      "step": 3314
+    },
+    {
+      "epoch": 1.2692454998085025,
+      "loss": 1.8420684337615967,
+      "loss_ce": 0.21937984228134155,
+      "loss_iou": 0.7437478303909302,
+      "loss_num": 0.027099609375,
+      "loss_xval": 1.6226885318756104,
+      "num_input_tokens_seen": 572859016,
+      "step": 3314
+    },
+    {
+      "epoch": 1.269628494829567,
+      "grad_norm": 102.3069855191529,
+      "learning_rate": 5e-06,
+      "loss": 1.3796,
+      "num_input_tokens_seen": 573031664,
+      "step": 3315
+    },
+    {
+      "epoch": 1.269628494829567,
+      "loss": 1.4531071186065674,
+      "loss_ce": 0.252878874540329,
+      "loss_iou": 0.5692150592803955,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 1.200228214263916,
+      "num_input_tokens_seen": 573031664,
+      "step": 3315
+    },
+    {
+      "epoch": 1.2700114898506318,
+      "grad_norm": 126.41855887441481,
+      "learning_rate": 5e-06,
+      "loss": 1.44,
+      "num_input_tokens_seen": 573204672,
+      "step": 3316
+    },
+    {
+      "epoch": 1.2700114898506318,
+      "loss": 1.4367523193359375,
+      "loss_ce": 0.21155793964862823,
+      "loss_iou": 0.5507381558418274,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 1.2251945734024048,
+      "num_input_tokens_seen": 573204672,
+      "step": 3316
+    },
+    {
+      "epoch": 1.2703944848716966,
+      "grad_norm": 133.43197200117166,
+      "learning_rate": 5e-06,
+      "loss": 1.6787,
+      "num_input_tokens_seen": 573377680,
+      "step": 3317
+    },
+    {
+      "epoch": 1.2703944848716966,
+      "loss": 1.887868046760559,
+      "loss_ce": 0.2615564167499542,
+      "loss_iou": 0.7499233484268188,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 1.6263115406036377,
+      "num_input_tokens_seen": 573377680,
+      "step": 3317
+    },
+    {
+      "epoch": 1.2707774798927614,
+      "grad_norm": 112.2828628960354,
+      "learning_rate": 5e-06,
+      "loss": 2.0386,
+      "num_input_tokens_seen": 573550792,
+      "step": 3318
+    },
+    {
+      "epoch": 1.2707774798927614,
+      "loss": 2.0267248153686523,
+      "loss_ce": 0.15854592621326447,
+      "loss_iou": 0.8536450862884521,
+      "loss_num": 0.0322265625,
+      "loss_xval": 1.8681788444519043,
+      "num_input_tokens_seen": 573550792,
+      "step": 3318
+    },
+    {
+      "epoch": 1.271160474913826,
+      "grad_norm": 80.14634241609927,
+      "learning_rate": 5e-06,
+      "loss": 1.6105,
+      "num_input_tokens_seen": 573723624,
+      "step": 3319
+    },
+    {
+      "epoch": 1.271160474913826,
+      "loss": 1.638106346130371,
+      "loss_ce": 0.24392704665660858,
+      "loss_iou": 0.616645336151123,
+      "loss_num": 0.0322265625,
+      "loss_xval": 1.394179344177246,
+      "num_input_tokens_seen": 573723624,
+      "step": 3319
+    },
+    {
+      "epoch": 1.2715434699348909,
+      "grad_norm": 231.3540922648395,
+      "learning_rate": 5e-06,
+      "loss": 1.6655,
+      "num_input_tokens_seen": 573896544,
+      "step": 3320
+    },
+    {
+      "epoch": 1.2715434699348909,
+      "loss": 1.5085893869400024,
+      "loss_ce": 0.2365088164806366,
+      "loss_iou": 0.595192551612854,
+      "loss_num": 0.016357421875,
+      "loss_xval": 1.272080659866333,
+      "num_input_tokens_seen": 573896544,
+      "step": 3320
+    },
+    {
+      "epoch": 1.2719264649559556,
+      "grad_norm": 72.31407858217699,
+      "learning_rate": 5e-06,
+      "loss": 1.9471,
+      "num_input_tokens_seen": 574069736,
+      "step": 3321
+    },
+    {
+      "epoch": 1.2719264649559556,
+      "loss": 1.8264158964157104,
+      "loss_ce": 0.23415064811706543,
+      "loss_iou": 0.7034813165664673,
+      "loss_num": 0.037109375,
+      "loss_xval": 1.5922653675079346,
+      "num_input_tokens_seen": 574069736,
+      "step": 3321
+    },
+    {
+      "epoch": 1.2723094599770204,
+      "grad_norm": 100.37074009002036,
+      "learning_rate": 5e-06,
+      "loss": 1.6824,
+      "num_input_tokens_seen": 574242672,
+      "step": 3322
+    },
+    {
+      "epoch": 1.2723094599770204,
+      "loss": 1.4854986667633057,
+      "loss_ce": 0.24368014931678772,
+      "loss_iou": 0.5532212257385254,
+      "loss_num": 0.027099609375,
+      "loss_xval": 1.2418184280395508,
+      "num_input_tokens_seen": 574242672,
+      "step": 3322
+    },
+    {
+      "epoch": 1.2726924549980851,
+      "grad_norm": 122.5845039333213,
+      "learning_rate": 5e-06,
+      "loss": 2.0082,
+      "num_input_tokens_seen": 574415792,
+      "step": 3323
+    },
+    {
+      "epoch": 1.2726924549980851,
+      "loss": 2.0704798698425293,
+      "loss_ce": 0.2315565049648285,
+      "loss_iou": 0.858945369720459,
+      "loss_num": 0.024169921875,
+      "loss_xval": 1.838923454284668,
+      "num_input_tokens_seen": 574415792,
+      "step": 3323
+    },
+    {
+      "epoch": 1.2730754500191497,
+      "grad_norm": 90.9971326969193,
+      "learning_rate": 5e-06,
+      "loss": 1.9792,
+      "num_input_tokens_seen": 574588576,
+      "step": 3324
+    },
+    {
+      "epoch": 1.2730754500191497,
+      "loss": 2.1510653495788574,
+      "loss_ce": 0.21998129785060883,
+      "loss_iou": 0.8659937381744385,
+      "loss_num": 0.039794921875,
+      "loss_xval": 1.9310842752456665,
+      "num_input_tokens_seen": 574588576,
+      "step": 3324
+    },
+    {
+      "epoch": 1.2734584450402144,
+      "grad_norm": 161.08256716104233,
+      "learning_rate": 5e-06,
+      "loss": 1.4981,
+      "num_input_tokens_seen": 574761496,
+      "step": 3325
+    },
+    {
+      "epoch": 1.2734584450402144,
+      "loss": 1.5489988327026367,
+      "loss_ce": 0.26540493965148926,
+      "loss_iou": 0.5177735090255737,
+      "loss_num": 0.049560546875,
+      "loss_xval": 1.2835938930511475,
+      "num_input_tokens_seen": 574761496,
+      "step": 3325
+    },
+    {
+      "epoch": 1.2738414400612792,
+      "grad_norm": 81.783766369408,
+      "learning_rate": 5e-06,
+      "loss": 1.9854,
+      "num_input_tokens_seen": 574934272,
+      "step": 3326
+    },
+    {
+      "epoch": 1.2738414400612792,
+      "loss": 1.9096903800964355,
+      "loss_ce": 0.1965281367301941,
+      "loss_iou": 0.7584366202354431,
+      "loss_num": 0.039306640625,
+      "loss_xval": 1.7131623029708862,
+      "num_input_tokens_seen": 574934272,
+      "step": 3326
+    },
+    {
+      "epoch": 1.274224435082344,
+      "grad_norm": 161.39022465012826,
+      "learning_rate": 5e-06,
+      "loss": 1.5728,
+      "num_input_tokens_seen": 575107376,
+      "step": 3327
+    },
+    {
+      "epoch": 1.274224435082344,
+      "loss": 1.6352365016937256,
+      "loss_ce": 0.22083421051502228,
+      "loss_iou": 0.6329213380813599,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 1.4144022464752197,
+      "num_input_tokens_seen": 575107376,
+      "step": 3327
+    },
+    {
+      "epoch": 1.2746074301034087,
+      "grad_norm": 66.53598461945988,
+      "learning_rate": 5e-06,
+      "loss": 1.9659,
+      "num_input_tokens_seen": 575280264,
+      "step": 3328
+    },
+    {
+      "epoch": 1.2746074301034087,
+      "loss": 2.036944627761841,
+      "loss_ce": 0.22707851231098175,
+      "loss_iou": 0.7948256731033325,
+      "loss_num": 0.0439453125,
+      "loss_xval": 1.809866189956665,
+      "num_input_tokens_seen": 575280264,
+      "step": 3328
+    },
+    {
+      "epoch": 1.2749904251244732,
+      "grad_norm": 77.72813265415147,
+      "learning_rate": 5e-06,
+      "loss": 1.4719,
+      "num_input_tokens_seen": 575453096,
+      "step": 3329
+    },
+    {
+      "epoch": 1.2749904251244732,
+      "loss": 1.4825770854949951,
+      "loss_ce": 0.22733338177204132,
+      "loss_iou": 0.5647861957550049,
+      "loss_num": 0.025146484375,
+      "loss_xval": 1.2451729774475098,
+      "num_input_tokens_seen": 575453096,
+      "step": 3329
+    },
+    {
+      "epoch": 1.275373420145538,
+      "grad_norm": 211.22215426560996,
+      "learning_rate": 5e-06,
+      "loss": 2.1258,
+      "num_input_tokens_seen": 575626136,
+      "step": 3330
+    },
+    {
+      "epoch": 1.275373420145538,
+      "loss": 2.1490392684936523,
+      "loss_ce": 0.27700331807136536,
+      "loss_iou": 0.8870066404342651,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 1.8720357418060303,
+      "num_input_tokens_seen": 575626136,
+      "step": 3330
+    },
+    {
+      "epoch": 1.2757564151666028,
+      "grad_norm": 109.66363871140655,
+      "learning_rate": 5e-06,
+      "loss": 2.2011,
+      "num_input_tokens_seen": 575799360,
+      "step": 3331
+    },
+    {
+      "epoch": 1.2757564151666028,
+      "loss": 2.222154378890991,
+      "loss_ce": 0.2608831822872162,
+      "loss_iou": 0.8851766586303711,
+      "loss_num": 0.038330078125,
+      "loss_xval": 1.9612712860107422,
+      "num_input_tokens_seen": 575799360,
+      "step": 3331
+    },
+    {
+      "epoch": 1.2761394101876675,
+      "grad_norm": 115.06326247553235,
+      "learning_rate": 5e-06,
+      "loss": 1.7082,
+      "num_input_tokens_seen": 575971848,
+      "step": 3332
+    },
+    {
+      "epoch": 1.2761394101876675,
+      "loss": 1.5471818447113037,
+      "loss_ce": 0.21208080649375916,
+      "loss_iou": 0.5933317542076111,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 1.3138607740402222,
+      "num_input_tokens_seen": 575971848,
+      "step": 3332
+    },
+    {
+      "epoch": 1.2765224052087323,
+      "grad_norm": 125.10481966929216,
+      "learning_rate": 5e-06,
+      "loss": 1.718,
+      "num_input_tokens_seen": 576144984,
+      "step": 3333
+    },
+    {
+      "epoch": 1.2765224052087323,
+      "loss": 1.7597615718841553,
+      "loss_ce": 0.22273939847946167,
+      "loss_iou": 0.7130453586578369,
+      "loss_num": 0.022216796875,
+      "loss_xval": 1.5370221138000488,
+      "num_input_tokens_seen": 576144984,
+      "step": 3333
+    },
+    {
+      "epoch": 1.276905400229797,
+      "grad_norm": 98.92262775558656,
+      "learning_rate": 5e-06,
+      "loss": 1.7493,
+      "num_input_tokens_seen": 576318296,
+      "step": 3334
+    },
+    {
+      "epoch": 1.276905400229797,
+      "loss": 1.7959173917770386,
+      "loss_ce": 0.25960248708724976,
+      "loss_iou": 0.694945752620697,
+      "loss_num": 0.029296875,
+      "loss_xval": 1.536314845085144,
+      "num_input_tokens_seen": 576318296,
+      "step": 3334
+    },
+    {
+      "epoch": 1.2772883952508618,
+      "grad_norm": 98.34931008554716,
+      "learning_rate": 5e-06,
+      "loss": 1.5485,
+      "num_input_tokens_seen": 576491080,
+      "step": 3335
+    },
+    {
+      "epoch": 1.2772883952508618,
+      "loss": 1.4789702892303467,
+      "loss_ce": 0.2304389774799347,
+      "loss_iou": 0.5721111297607422,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 1.2485313415527344,
+      "num_input_tokens_seen": 576491080,
+      "step": 3335
+    },
+    {
+      "epoch": 1.2776713902719266,
+      "grad_norm": 105.37604160624177,
+      "learning_rate": 5e-06,
+      "loss": 1.9593,
+      "num_input_tokens_seen": 576663952,
+      "step": 3336
+    },
+    {
+      "epoch": 1.2776713902719266,
+      "loss": 2.047152280807495,
+      "loss_ce": 0.26648885011672974,
+      "loss_iou": 0.8120236396789551,
+      "loss_num": 0.03125,
+      "loss_xval": 1.7806634902954102,
+      "num_input_tokens_seen": 576663952,
+      "step": 3336
+    },
+    {
+      "epoch": 1.2780543852929913,
+      "grad_norm": 98.1583425890488,
+      "learning_rate": 5e-06,
+      "loss": 1.3958,
+      "num_input_tokens_seen": 576837200,
+      "step": 3337
+    },
+    {
+      "epoch": 1.2780543852929913,
+      "loss": 1.3965688943862915,
+      "loss_ce": 0.24463441967964172,
+      "loss_iou": 0.5312895178794861,
+      "loss_num": 0.017822265625,
+      "loss_xval": 1.1519345045089722,
+      "num_input_tokens_seen": 576837200,
+      "step": 3337
+    },
+    {
+      "epoch": 1.2784373803140558,
+      "grad_norm": 178.8829176594653,
+      "learning_rate": 5e-06,
+      "loss": 1.7012,
+      "num_input_tokens_seen": 577010496,
+      "step": 3338
+    },
+    {
+      "epoch": 1.2784373803140558,
+      "loss": 1.8009527921676636,
+      "loss_ce": 0.2655678987503052,
+      "loss_iou": 0.7156294584274292,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 1.5353848934173584,
+      "num_input_tokens_seen": 577010496,
+      "step": 3338
+    },
+    {
+      "epoch": 1.2788203753351206,
+      "grad_norm": 99.67392432700711,
+      "learning_rate": 5e-06,
+      "loss": 1.878,
+      "num_input_tokens_seen": 577183768,
+      "step": 3339
+    },
+    {
+      "epoch": 1.2788203753351206,
+      "loss": 1.8287030458450317,
+      "loss_ce": 0.244386687874794,
+      "loss_iou": 0.6971874237060547,
+      "loss_num": 0.0380859375,
+      "loss_xval": 1.5843162536621094,
+      "num_input_tokens_seen": 577183768,
+      "step": 3339
+    },
+    {
+      "epoch": 1.2792033703561854,
+      "grad_norm": 107.38374897880072,
+      "learning_rate": 5e-06,
+      "loss": 1.6041,
+      "num_input_tokens_seen": 577356544,
+      "step": 3340
+    },
+    {
+      "epoch": 1.2792033703561854,
+      "loss": 1.550626277923584,
+      "loss_ce": 0.26853227615356445,
+      "loss_iou": 0.5984139442443848,
+      "loss_num": 0.01708984375,
+      "loss_xval": 1.2820940017700195,
+      "num_input_tokens_seen": 577356544,
+      "step": 3340
+    },
+    {
+      "epoch": 1.2795863653772501,
+      "grad_norm": 123.60648613766934,
+      "learning_rate": 5e-06,
+      "loss": 1.9738,
+      "num_input_tokens_seen": 577529592,
+      "step": 3341
+    },
+    {
+      "epoch": 1.2795863653772501,
+      "loss": 2.078113555908203,
+      "loss_ce": 0.2186749279499054,
+      "loss_iou": 0.8510449528694153,
+      "loss_num": 0.031494140625,
+      "loss_xval": 1.859438419342041,
+      "num_input_tokens_seen": 577529592,
+      "step": 3341
+    },
+    {
+      "epoch": 1.2799693603983149,
+      "grad_norm": 93.47068258699876,
+      "learning_rate": 5e-06,
+      "loss": 1.9349,
+      "num_input_tokens_seen": 577702624,
+      "step": 3342
+    },
+    {
+      "epoch": 1.2799693603983149,
+      "loss": 1.8729562759399414,
+      "loss_ce": 0.2133171111345291,
+      "loss_iou": 0.7257546782493591,
+      "loss_num": 0.041748046875,
+      "loss_xval": 1.6596392393112183,
+      "num_input_tokens_seen": 577702624,
+      "step": 3342
+    },
+    {
+      "epoch": 1.2803523554193794,
+      "grad_norm": 94.29295943043381,
+      "learning_rate": 5e-06,
+      "loss": 1.3165,
+      "num_input_tokens_seen": 577875264,
+      "step": 3343
+    },
+    {
+      "epoch": 1.2803523554193794,
+      "loss": 1.3596270084381104,
+      "loss_ce": 0.26244881749153137,
+      "loss_iou": 0.5014546513557434,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 1.0971781015396118,
+      "num_input_tokens_seen": 577875264,
+      "step": 3343
+    },
+    {
+      "epoch": 1.2807353504404442,
+      "grad_norm": 124.71385888265097,
+      "learning_rate": 5e-06,
+      "loss": 1.6864,
+      "num_input_tokens_seen": 578047864,
+      "step": 3344
+    },
+    {
+      "epoch": 1.2807353504404442,
+      "loss": 1.8556559085845947,
+      "loss_ce": 0.21935562789440155,
+      "loss_iou": 0.7627912759780884,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 1.6363003253936768,
+      "num_input_tokens_seen": 578047864,
+      "step": 3344
+    },
+    {
+      "epoch": 1.281118345461509,
+      "grad_norm": 151.5891353046151,
+      "learning_rate": 5e-06,
+      "loss": 1.8347,
+      "num_input_tokens_seen": 578220520,
+      "step": 3345
+    },
+    {
+      "epoch": 1.281118345461509,
+      "loss": 1.7800143957138062,
+      "loss_ce": 0.19940626621246338,
+      "loss_iou": 0.7452296018600464,
+      "loss_num": 0.01806640625,
+      "loss_xval": 1.5806081295013428,
+      "num_input_tokens_seen": 578220520,
+      "step": 3345
+    },
+    {
+      "epoch": 1.2815013404825737,
+      "grad_norm": 105.68129864653623,
+      "learning_rate": 5e-06,
+      "loss": 1.9297,
+      "num_input_tokens_seen": 578393304,
+      "step": 3346
+    },
+    {
+      "epoch": 1.2815013404825737,
+      "loss": 1.8328579664230347,
+      "loss_ce": 0.23113152384757996,
+      "loss_iou": 0.7155971527099609,
+      "loss_num": 0.0341796875,
+      "loss_xval": 1.6017265319824219,
+      "num_input_tokens_seen": 578393304,
+      "step": 3346
+    },
+    {
+      "epoch": 1.2818843355036385,
+      "grad_norm": 106.32825988455923,
+      "learning_rate": 5e-06,
+      "loss": 1.4914,
+      "num_input_tokens_seen": 578566488,
+      "step": 3347
+    },
+    {
+      "epoch": 1.2818843355036385,
+      "loss": 1.635674238204956,
+      "loss_ce": 0.21635141968727112,
+      "loss_iou": 0.6693781614303589,
+      "loss_num": 0.01611328125,
+      "loss_xval": 1.4193227291107178,
+      "num_input_tokens_seen": 578566488,
+      "step": 3347
+    },
+    {
+      "epoch": 1.2822673305247032,
+      "grad_norm": 119.87350863033323,
+      "learning_rate": 5e-06,
+      "loss": 1.5509,
+      "num_input_tokens_seen": 578739608,
+      "step": 3348
+    },
+    {
+      "epoch": 1.2822673305247032,
+      "loss": 1.4294211864471436,
+      "loss_ce": 0.180038183927536,
+      "loss_iou": 0.5733305215835571,
+      "loss_num": 0.0205078125,
+      "loss_xval": 1.2493832111358643,
+      "num_input_tokens_seen": 578739608,
+      "step": 3348
+    },
+    {
+      "epoch": 1.282650325545768,
+      "grad_norm": 102.46857820913375,
+      "learning_rate": 5e-06,
+      "loss": 1.5167,
+      "num_input_tokens_seen": 578912640,
+      "step": 3349
+    },
+    {
+      "epoch": 1.282650325545768,
+      "loss": 1.6376855373382568,
+      "loss_ce": 0.18948471546173096,
+      "loss_iou": 0.6668798923492432,
+      "loss_num": 0.02294921875,
+      "loss_xval": 1.4482008218765259,
+      "num_input_tokens_seen": 578912640,
+      "step": 3349
+    },
+    {
+      "epoch": 1.2830333205668327,
+      "grad_norm": 84.18815580132912,
+      "learning_rate": 5e-06,
+      "loss": 1.4077,
+      "num_input_tokens_seen": 579085336,
+      "step": 3350
+    },
+    {
+      "epoch": 1.2830333205668327,
+      "loss": 1.3755912780761719,
+      "loss_ce": 0.19753962755203247,
+      "loss_iou": 0.5309813618659973,
+      "loss_num": 0.023193359375,
+      "loss_xval": 1.1780515909194946,
+      "num_input_tokens_seen": 579085336,
+      "step": 3350
+    },
+    {
+      "epoch": 1.2834163155878975,
+      "grad_norm": 118.6917879357888,
+      "learning_rate": 5e-06,
+      "loss": 1.5693,
+      "num_input_tokens_seen": 579258376,
+      "step": 3351
+    },
+    {
+      "epoch": 1.2834163155878975,
+      "loss": 1.615788221359253,
+      "loss_ce": 0.24710959196090698,
+      "loss_iou": 0.6457650661468506,
+      "loss_num": 0.015380859375,
+      "loss_xval": 1.3686785697937012,
+      "num_input_tokens_seen": 579258376,
+      "step": 3351
+    },
+    {
+      "epoch": 1.283799310608962,
+      "grad_norm": 175.36351041767077,
+      "learning_rate": 5e-06,
+      "loss": 1.8253,
+      "num_input_tokens_seen": 579431512,
+      "step": 3352
+    },
+    {
+      "epoch": 1.283799310608962,
+      "loss": 1.7172226905822754,
+      "loss_ce": 0.24279974400997162,
+      "loss_iou": 0.6783125400543213,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 1.4744229316711426,
+      "num_input_tokens_seen": 579431512,
+      "step": 3352
+    },
+    {
+      "epoch": 1.2841823056300268,
+      "grad_norm": 138.4932788465756,
+      "learning_rate": 5e-06,
+      "loss": 2.0092,
+      "num_input_tokens_seen": 579604624,
+      "step": 3353
+    },
+    {
+      "epoch": 1.2841823056300268,
+      "loss": 1.8588067293167114,
+      "loss_ce": 0.2468595653772354,
+      "loss_iou": 0.7251935005187988,
+      "loss_num": 0.0322265625,
+      "loss_xval": 1.6119470596313477,
+      "num_input_tokens_seen": 579604624,
+      "step": 3353
+    },
+    {
+      "epoch": 1.2845653006510915,
+      "grad_norm": 81.77329426208514,
+      "learning_rate": 5e-06,
+      "loss": 1.5477,
+      "num_input_tokens_seen": 579777960,
+      "step": 3354
+    },
+    {
+      "epoch": 1.2845653006510915,
+      "loss": 1.4595190286636353,
+      "loss_ce": 0.24423342943191528,
+      "loss_iou": 0.5725475549697876,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 1.2152855396270752,
+      "num_input_tokens_seen": 579777960,
+      "step": 3354
+    },
+    {
+      "epoch": 1.2849482956721563,
+      "grad_norm": 152.47406169663037,
+      "learning_rate": 5e-06,
+      "loss": 1.5219,
+      "num_input_tokens_seen": 579950984,
+      "step": 3355
+    },
+    {
+      "epoch": 1.2849482956721563,
+      "loss": 1.4859492778778076,
+      "loss_ce": 0.27817896008491516,
+      "loss_iou": 0.5744203925132751,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 1.2077702283859253,
+      "num_input_tokens_seen": 579950984,
+      "step": 3355
+    },
+    {
+      "epoch": 1.285331290693221,
+      "grad_norm": 100.04572783802985,
+      "learning_rate": 5e-06,
+      "loss": 1.8967,
+      "num_input_tokens_seen": 580123816,
+      "step": 3356
+    },
+    {
+      "epoch": 1.285331290693221,
+      "loss": 1.830329179763794,
+      "loss_ce": 0.24495810270309448,
+      "loss_iou": 0.7337560653686523,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 1.5853710174560547,
+      "num_input_tokens_seen": 580123816,
+      "step": 3356
+    },
+    {
+      "epoch": 1.2857142857142856,
+      "grad_norm": 118.56857953203593,
+      "learning_rate": 5e-06,
+      "loss": 1.4661,
+      "num_input_tokens_seen": 580296592,
+      "step": 3357
+    },
+    {
+      "epoch": 1.2857142857142856,
+      "loss": 1.46898353099823,
+      "loss_ce": 0.2476564347743988,
+      "loss_iou": 0.5652838945388794,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 1.2213270664215088,
+      "num_input_tokens_seen": 580296592,
+      "step": 3357
+    },
+    {
+      "epoch": 1.2860972807353503,
+      "grad_norm": 385.29104544462143,
+      "learning_rate": 5e-06,
+      "loss": 1.7356,
+      "num_input_tokens_seen": 580469512,
+      "step": 3358
+    },
+    {
+      "epoch": 1.2860972807353503,
+      "loss": 1.827427625656128,
+      "loss_ce": 0.22973480820655823,
+      "loss_iou": 0.7317382097244263,
+      "loss_num": 0.02685546875,
+      "loss_xval": 1.5976927280426025,
+      "num_input_tokens_seen": 580469512,
+      "step": 3358
+    },
+    {
+      "epoch": 1.286480275756415,
+      "grad_norm": 117.23976088897443,
+      "learning_rate": 5e-06,
+      "loss": 2.1095,
+      "num_input_tokens_seen": 580642488,
+      "step": 3359
+    },
+    {
+      "epoch": 1.286480275756415,
+      "loss": 2.1356539726257324,
+      "loss_ce": 0.2708384692668915,
+      "loss_iou": 0.845737874507904,
+      "loss_num": 0.03466796875,
+      "loss_xval": 1.864815592765808,
+      "num_input_tokens_seen": 580642488,
+      "step": 3359
+    },
+    {
+      "epoch": 1.2868632707774799,
+      "grad_norm": 195.45524028811684,
+      "learning_rate": 5e-06,
+      "loss": 1.5106,
+      "num_input_tokens_seen": 580815576,
+      "step": 3360
+    },
+    {
+      "epoch": 1.2868632707774799,
+      "loss": 1.553452968597412,
+      "loss_ce": 0.23564180731773376,
+      "loss_iou": 0.6180273294448853,
+      "loss_num": 0.016357421875,
+      "loss_xval": 1.3178112506866455,
+      "num_input_tokens_seen": 580815576,
+      "step": 3360
+    },
+    {
+      "epoch": 1.2872462657985446,
+      "grad_norm": 183.7944958876849,
+      "learning_rate": 5e-06,
+      "loss": 1.8716,
+      "num_input_tokens_seen": 580988736,
+      "step": 3361
+    },
+    {
+      "epoch": 1.2872462657985446,
+      "loss": 1.8165233135223389,
+      "loss_ce": 0.23417851328849792,
+      "loss_iou": 0.7233012914657593,
+      "loss_num": 0.027099609375,
+      "loss_xval": 1.5823447704315186,
+      "num_input_tokens_seen": 580988736,
+      "step": 3361
+    },
+    {
+      "epoch": 1.2876292608196094,
+      "grad_norm": 320.58422365920035,
+      "learning_rate": 5e-06,
+      "loss": 1.9082,
+      "num_input_tokens_seen": 581161720,
+      "step": 3362
+    },
+    {
+      "epoch": 1.2876292608196094,
+      "loss": 2.008775234222412,
+      "loss_ce": 0.24376258254051208,
+      "loss_iou": 0.8073720335960388,
+      "loss_num": 0.030029296875,
+      "loss_xval": 1.7650126218795776,
+      "num_input_tokens_seen": 581161720,
+      "step": 3362
+    },
+    {
+      "epoch": 1.2880122558406741,
+      "grad_norm": 84.63246045541894,
+      "learning_rate": 5e-06,
+      "loss": 1.683,
+      "num_input_tokens_seen": 581334952,
+      "step": 3363
+    },
+    {
+      "epoch": 1.2880122558406741,
+      "loss": 1.6573615074157715,
+      "loss_ce": 0.25077182054519653,
+      "loss_iou": 0.6485462784767151,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 1.4065896272659302,
+      "num_input_tokens_seen": 581334952,
+      "step": 3363
+    },
+    {
+      "epoch": 1.288395250861739,
+      "grad_norm": 138.0620839987321,
+      "learning_rate": 5e-06,
+      "loss": 1.9789,
+      "num_input_tokens_seen": 581507880,
+      "step": 3364
+    },
+    {
+      "epoch": 1.288395250861739,
+      "loss": 1.7434158325195312,
+      "loss_ce": 0.21551012992858887,
+      "loss_iou": 0.6336733102798462,
+      "loss_num": 0.05224609375,
+      "loss_xval": 1.5279057025909424,
+      "num_input_tokens_seen": 581507880,
+      "step": 3364
+    },
+    {
+      "epoch": 1.2887782458828037,
+      "grad_norm": 88.13174218950272,
+      "learning_rate": 5e-06,
+      "loss": 2.0411,
+      "num_input_tokens_seen": 581680552,
+      "step": 3365
+    },
+    {
+      "epoch": 1.2887782458828037,
+      "loss": 2.142244577407837,
+      "loss_ce": 0.246106818318367,
+      "loss_iou": 0.8491919636726379,
+      "loss_num": 0.03955078125,
+      "loss_xval": 1.8961378335952759,
+      "num_input_tokens_seen": 581680552,
+      "step": 3365
+    },
+    {
+      "epoch": 1.2891612409038682,
+      "grad_norm": 106.17229374306666,
+      "learning_rate": 5e-06,
+      "loss": 1.5407,
+      "num_input_tokens_seen": 581853712,
+      "step": 3366
+    },
+    {
+      "epoch": 1.2891612409038682,
+      "loss": 1.7057932615280151,
+      "loss_ce": 0.174179345369339,
+      "loss_iou": 0.6264026165008545,
+      "loss_num": 0.0556640625,
+      "loss_xval": 1.531613826751709,
+      "num_input_tokens_seen": 581853712,
+      "step": 3366
+    },
+    {
+      "epoch": 1.289544235924933,
+      "grad_norm": 112.54860255656129,
+      "learning_rate": 5e-06,
+      "loss": 1.8566,
+      "num_input_tokens_seen": 582026704,
+      "step": 3367
+    },
+    {
+      "epoch": 1.289544235924933,
+      "loss": 1.845980167388916,
+      "loss_ce": 0.24146392941474915,
+      "loss_iou": 0.6961179971694946,
+      "loss_num": 0.04248046875,
+      "loss_xval": 1.6045162677764893,
+      "num_input_tokens_seen": 582026704,
+      "step": 3367
+    },
+    {
+      "epoch": 1.2899272309459977,
+      "grad_norm": 129.4485495674371,
+      "learning_rate": 5e-06,
+      "loss": 1.4022,
+      "num_input_tokens_seen": 582199240,
+      "step": 3368
+    },
+    {
+      "epoch": 1.2899272309459977,
+      "loss": 1.3163137435913086,
+      "loss_ce": 0.18781515955924988,
+      "loss_iou": 0.5210669040679932,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 1.1202588081359863,
+      "num_input_tokens_seen": 582199240,
+      "step": 3368
+    },
+    {
+      "epoch": 1.2903102259670625,
+      "grad_norm": 154.53419435302868,
+      "learning_rate": 5e-06,
+      "loss": 2.0915,
+      "num_input_tokens_seen": 582372680,
+      "step": 3369
+    },
+    {
+      "epoch": 1.2903102259670625,
+      "loss": 2.0044426918029785,
+      "loss_ce": 0.2393362671136856,
+      "loss_iou": 0.7790987491607666,
+      "loss_num": 0.04150390625,
+      "loss_xval": 1.7651066780090332,
+      "num_input_tokens_seen": 582372680,
+      "step": 3369
+    },
+    {
+      "epoch": 1.2906932209881272,
+      "grad_norm": 204.9802556331962,
+      "learning_rate": 5e-06,
+      "loss": 1.5666,
+      "num_input_tokens_seen": 582545696,
+      "step": 3370
+    },
+    {
+      "epoch": 1.2906932209881272,
+      "loss": 1.6414077281951904,
+      "loss_ce": 0.26562219858169556,
+      "loss_iou": 0.6418416500091553,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 1.3757853507995605,
+      "num_input_tokens_seen": 582545696,
+      "step": 3370
+    },
+    {
+      "epoch": 1.2910762160091918,
+      "grad_norm": 174.45829575674773,
+      "learning_rate": 5e-06,
+      "loss": 2.6819,
+      "num_input_tokens_seen": 582718384,
+      "step": 3371
+    },
+    {
+      "epoch": 1.2910762160091918,
+      "loss": 2.629523277282715,
+      "loss_ce": 0.21946796774864197,
+      "loss_iou": 0.9732161164283752,
+      "loss_num": 0.0927734375,
+      "loss_xval": 2.41005539894104,
+      "num_input_tokens_seen": 582718384,
+      "step": 3371
+    },
+    {
+      "epoch": 1.2914592110302565,
+      "grad_norm": 161.16437623061037,
+      "learning_rate": 5e-06,
+      "loss": 1.7993,
+      "num_input_tokens_seen": 582891496,
+      "step": 3372
+    },
+    {
+      "epoch": 1.2914592110302565,
+      "loss": 1.6719965934753418,
+      "loss_ce": 0.20556679368019104,
+      "loss_iou": 0.6563715934753418,
+      "loss_num": 0.03076171875,
+      "loss_xval": 1.4664297103881836,
+      "num_input_tokens_seen": 582891496,
+      "step": 3372
+    },
+    {
+      "epoch": 1.2918422060513213,
+      "grad_norm": 183.23312031668578,
+      "learning_rate": 5e-06,
+      "loss": 1.7636,
+      "num_input_tokens_seen": 583064400,
+      "step": 3373
+    },
+    {
+      "epoch": 1.2918422060513213,
+      "loss": 1.8405812978744507,
+      "loss_ce": 0.3046218454837799,
+      "loss_iou": 0.7224475145339966,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 1.5359594821929932,
+      "num_input_tokens_seen": 583064400,
+      "step": 3373
+    },
+    {
+      "epoch": 1.292225201072386,
+      "grad_norm": 281.53897382838215,
+      "learning_rate": 5e-06,
+      "loss": 1.7992,
+      "num_input_tokens_seen": 583237280,
+      "step": 3374
+    },
+    {
+      "epoch": 1.292225201072386,
+      "loss": 1.7347018718719482,
+      "loss_ce": 0.23590445518493652,
+      "loss_iou": 0.6990141868591309,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 1.4987974166870117,
+      "num_input_tokens_seen": 583237280,
+      "step": 3374
+    },
+    {
+      "epoch": 1.2926081960934508,
+      "grad_norm": 302.85637195824273,
+      "learning_rate": 5e-06,
+      "loss": 2.3102,
+      "num_input_tokens_seen": 583410136,
+      "step": 3375
+    },
+    {
+      "epoch": 1.2926081960934508,
+      "loss": 2.5609192848205566,
+      "loss_ce": 0.26040640473365784,
+      "loss_iou": 1.0801881551742554,
+      "loss_num": 0.028076171875,
+      "loss_xval": 2.3005130290985107,
+      "num_input_tokens_seen": 583410136,
+      "step": 3375
+    },
+    {
+      "epoch": 1.2929911911145155,
+      "grad_norm": 130.0003850790514,
+      "learning_rate": 5e-06,
+      "loss": 2.2914,
+      "num_input_tokens_seen": 583583152,
+      "step": 3376
+    },
+    {
+      "epoch": 1.2929911911145155,
+      "loss": 2.368964195251465,
+      "loss_ce": 0.2561070919036865,
+      "loss_iou": 0.8890701532363892,
+      "loss_num": 0.06689453125,
+      "loss_xval": 2.1128571033477783,
+      "num_input_tokens_seen": 583583152,
+      "step": 3376
+    },
+    {
+      "epoch": 1.2933741861355803,
+      "grad_norm": 73.73597806623299,
+      "learning_rate": 5e-06,
+      "loss": 1.6875,
+      "num_input_tokens_seen": 583756336,
+      "step": 3377
+    },
+    {
+      "epoch": 1.2933741861355803,
+      "loss": 1.8240755796432495,
+      "loss_ce": 0.21413224935531616,
+      "loss_iou": 0.7477816939353943,
+      "loss_num": 0.02294921875,
+      "loss_xval": 1.6099432706832886,
+      "num_input_tokens_seen": 583756336,
+      "step": 3377
+    },
+    {
+      "epoch": 1.293757181156645,
+      "grad_norm": 156.63117108922665,
+      "learning_rate": 5e-06,
+      "loss": 1.6359,
+      "num_input_tokens_seen": 583928928,
+      "step": 3378
+    },
+    {
+      "epoch": 1.293757181156645,
+      "loss": 1.6808727979660034,
+      "loss_ce": 0.27408987283706665,
+      "loss_iou": 0.650596022605896,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 1.406782865524292,
+      "num_input_tokens_seen": 583928928,
+      "step": 3378
+    },
+    {
+      "epoch": 1.2941401761777098,
+      "grad_norm": 157.78362428064494,
+      "learning_rate": 5e-06,
+      "loss": 1.6835,
+      "num_input_tokens_seen": 584101776,
+      "step": 3379
+    },
+    {
+      "epoch": 1.2941401761777098,
+      "loss": 1.5322535037994385,
+      "loss_ce": 0.24172645807266235,
+      "loss_iou": 0.6080930829048157,
+      "loss_num": 0.014892578125,
+      "loss_xval": 1.2905269861221313,
+      "num_input_tokens_seen": 584101776,
+      "step": 3379
+    },
+    {
+      "epoch": 1.2945231711987744,
+      "grad_norm": 340.747508174026,
+      "learning_rate": 5e-06,
+      "loss": 1.8243,
+      "num_input_tokens_seen": 584274624,
+      "step": 3380
+    },
+    {
+      "epoch": 1.2945231711987744,
+      "loss": 1.675611138343811,
+      "loss_ce": 0.22095349431037903,
+      "loss_iou": 0.6779208779335022,
+      "loss_num": 0.019775390625,
+      "loss_xval": 1.4546576738357544,
+      "num_input_tokens_seen": 584274624,
+      "step": 3380
+    },
+    {
+      "epoch": 1.2949061662198391,
+      "grad_norm": 129.25041796196015,
+      "learning_rate": 5e-06,
+      "loss": 1.75,
+      "num_input_tokens_seen": 584447240,
+      "step": 3381
+    },
+    {
+      "epoch": 1.2949061662198391,
+      "loss": 1.8822615146636963,
+      "loss_ce": 0.2786910831928253,
+      "loss_iou": 0.7368133068084717,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 1.6035704612731934,
+      "num_input_tokens_seen": 584447240,
+      "step": 3381
+    },
+    {
+      "epoch": 1.2952891612409039,
+      "grad_norm": 118.61187800027115,
+      "learning_rate": 5e-06,
+      "loss": 1.3366,
+      "num_input_tokens_seen": 584620000,
+      "step": 3382
+    },
+    {
+      "epoch": 1.2952891612409039,
+      "loss": 1.2266054153442383,
+      "loss_ce": 0.1901278793811798,
+      "loss_iou": 0.48373866081237793,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 1.0364775657653809,
+      "num_input_tokens_seen": 584620000,
+      "step": 3382
+    },
+    {
+      "epoch": 1.2956721562619686,
+      "grad_norm": 149.1785933503822,
+      "learning_rate": 5e-06,
+      "loss": 1.7609,
+      "num_input_tokens_seen": 584792928,
+      "step": 3383
+    },
+    {
+      "epoch": 1.2956721562619686,
+      "loss": 1.6310421228408813,
+      "loss_ce": 0.22532498836517334,
+      "loss_iou": 0.6436392068862915,
+      "loss_num": 0.023681640625,
+      "loss_xval": 1.405717134475708,
+      "num_input_tokens_seen": 584792928,
+      "step": 3383
+    },
+    {
+      "epoch": 1.2960551512830334,
+      "grad_norm": 103.98641929599162,
+      "learning_rate": 5e-06,
+      "loss": 1.3192,
+      "num_input_tokens_seen": 584965776,
+      "step": 3384
+    },
+    {
+      "epoch": 1.2960551512830334,
+      "loss": 1.3010835647583008,
+      "loss_ce": 0.20399756729602814,
+      "loss_iou": 0.5086259841918945,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 1.0970860719680786,
+      "num_input_tokens_seen": 584965776,
+      "step": 3384
+    },
+    {
+      "epoch": 1.296438146304098,
+      "grad_norm": 152.74040832798303,
+      "learning_rate": 5e-06,
+      "loss": 1.4629,
+      "num_input_tokens_seen": 585138608,
+      "step": 3385
+    },
+    {
+      "epoch": 1.296438146304098,
+      "loss": 1.4023289680480957,
+      "loss_ce": 0.2729048728942871,
+      "loss_iou": 0.5235438346862793,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 1.1294240951538086,
+      "num_input_tokens_seen": 585138608,
+      "step": 3385
+    },
+    {
+      "epoch": 1.2968211413251627,
+      "grad_norm": 189.98002443436982,
+      "learning_rate": 5e-06,
+      "loss": 1.6322,
+      "num_input_tokens_seen": 585311384,
+      "step": 3386
+    },
+    {
+      "epoch": 1.2968211413251627,
+      "loss": 1.6417934894561768,
+      "loss_ce": 0.2782427966594696,
+      "loss_iou": 0.6331761479377747,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 1.3635507822036743,
+      "num_input_tokens_seen": 585311384,
+      "step": 3386
+    },
+    {
+      "epoch": 1.2972041363462274,
+      "grad_norm": 125.16948817146444,
+      "learning_rate": 5e-06,
+      "loss": 1.5475,
+      "num_input_tokens_seen": 585484344,
+      "step": 3387
+    },
+    {
+      "epoch": 1.2972041363462274,
+      "loss": 1.6016695499420166,
+      "loss_ce": 0.18500235676765442,
+      "loss_iou": 0.6527305245399475,
+      "loss_num": 0.022216796875,
+      "loss_xval": 1.416667103767395,
+      "num_input_tokens_seen": 585484344,
+      "step": 3387
+    },
+    {
+      "epoch": 1.2975871313672922,
+      "grad_norm": 183.59067421613585,
+      "learning_rate": 5e-06,
+      "loss": 1.5959,
+      "num_input_tokens_seen": 585657600,
+      "step": 3388
+    },
+    {
+      "epoch": 1.2975871313672922,
+      "loss": 1.695542573928833,
+      "loss_ce": 0.19205079972743988,
+      "loss_iou": 0.7017734050750732,
+      "loss_num": 0.02001953125,
+      "loss_xval": 1.5034918785095215,
+      "num_input_tokens_seen": 585657600,
+      "step": 3388
+    },
+    {
+      "epoch": 1.297970126388357,
+      "grad_norm": 247.7912259115244,
+      "learning_rate": 5e-06,
+      "loss": 2.0144,
+      "num_input_tokens_seen": 585830392,
+      "step": 3389
+    },
+    {
+      "epoch": 1.297970126388357,
+      "loss": 2.0455682277679443,
+      "loss_ce": 0.2454589456319809,
+      "loss_iou": 0.8351743221282959,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 1.8001093864440918,
+      "num_input_tokens_seen": 585830392,
+      "step": 3389
+    },
+    {
+      "epoch": 1.2983531214094217,
+      "grad_norm": 76.46328648680853,
+      "learning_rate": 5e-06,
+      "loss": 1.7439,
+      "num_input_tokens_seen": 586003496,
+      "step": 3390
+    },
+    {
+      "epoch": 1.2983531214094217,
+      "loss": 1.6645066738128662,
+      "loss_ce": 0.22745877504348755,
+      "loss_iou": 0.6660947799682617,
+      "loss_num": 0.02099609375,
+      "loss_xval": 1.4370479583740234,
+      "num_input_tokens_seen": 586003496,
+      "step": 3390
+    },
+    {
+      "epoch": 1.2987361164304865,
+      "grad_norm": 153.52668657699834,
+      "learning_rate": 5e-06,
+      "loss": 1.4076,
+      "num_input_tokens_seen": 586176648,
+      "step": 3391
+    },
+    {
+      "epoch": 1.2987361164304865,
+      "loss": 1.3777530193328857,
+      "loss_ce": 0.22161175310611725,
+      "loss_iou": 0.5349493026733398,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 1.1561412811279297,
+      "num_input_tokens_seen": 586176648,
+      "step": 3391
+    },
+    {
+      "epoch": 1.2991191114515512,
+      "grad_norm": 195.1283230852289,
+      "learning_rate": 5e-06,
+      "loss": 1.6028,
+      "num_input_tokens_seen": 586349664,
+      "step": 3392
+    },
+    {
+      "epoch": 1.2991191114515512,
+      "loss": 1.6521891355514526,
+      "loss_ce": 0.2032175064086914,
+      "loss_iou": 0.6701949834823608,
+      "loss_num": 0.021728515625,
+      "loss_xval": 1.4489715099334717,
+      "num_input_tokens_seen": 586349664,
+      "step": 3392
+    },
+    {
+      "epoch": 1.299502106472616,
+      "grad_norm": 227.96607768799603,
+      "learning_rate": 5e-06,
+      "loss": 1.4066,
+      "num_input_tokens_seen": 586522800,
+      "step": 3393
+    },
+    {
+      "epoch": 1.299502106472616,
+      "loss": 1.3554717302322388,
+      "loss_ce": 0.24635626375675201,
+      "loss_iou": 0.5075300931930542,
+      "loss_num": 0.018798828125,
+      "loss_xval": 1.1091153621673584,
+      "num_input_tokens_seen": 586522800,
+      "step": 3393
+    },
+    {
+      "epoch": 1.2998851014936805,
+      "grad_norm": 77.66038341172509,
+      "learning_rate": 5e-06,
+      "loss": 1.7276,
+      "num_input_tokens_seen": 586695656,
+      "step": 3394
+    },
+    {
+      "epoch": 1.2998851014936805,
+      "loss": 1.7297232151031494,
+      "loss_ce": 0.1546655297279358,
+      "loss_iou": 0.7195356488227844,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 1.5750576257705688,
+      "num_input_tokens_seen": 586695656,
+      "step": 3394
+    },
+    {
+      "epoch": 1.3002680965147453,
+      "grad_norm": 94.67778664875233,
+      "learning_rate": 5e-06,
+      "loss": 1.3643,
+      "num_input_tokens_seen": 586868424,
+      "step": 3395
+    },
+    {
+      "epoch": 1.3002680965147453,
+      "loss": 1.3454577922821045,
+      "loss_ce": 0.242027148604393,
+      "loss_iou": 0.48811671137809753,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 1.1034307479858398,
+      "num_input_tokens_seen": 586868424,
+      "step": 3395
+    },
+    {
+      "epoch": 1.30065109153581,
+      "grad_norm": 107.85203258493827,
+      "learning_rate": 5e-06,
+      "loss": 1.7604,
+      "num_input_tokens_seen": 587041720,
+      "step": 3396
+    },
+    {
+      "epoch": 1.30065109153581,
+      "loss": 1.859063744544983,
+      "loss_ce": 0.29286590218544006,
+      "loss_iou": 0.7240779399871826,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 1.5661978721618652,
+      "num_input_tokens_seen": 587041720,
+      "step": 3396
+    },
+    {
+      "epoch": 1.3010340865568748,
+      "grad_norm": 139.59650402931635,
+      "learning_rate": 5e-06,
+      "loss": 1.5741,
+      "num_input_tokens_seen": 587214208,
+      "step": 3397
+    },
+    {
+      "epoch": 1.3010340865568748,
+      "loss": 1.5014113187789917,
+      "loss_ce": 0.22343674302101135,
+      "loss_iou": 0.5911662578582764,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 1.2779746055603027,
+      "num_input_tokens_seen": 587214208,
+      "step": 3397
+    },
+    {
+      "epoch": 1.3014170815779396,
+      "grad_norm": 138.43404940968887,
+      "learning_rate": 5e-06,
+      "loss": 1.6001,
+      "num_input_tokens_seen": 587387416,
+      "step": 3398
+    },
+    {
+      "epoch": 1.3014170815779396,
+      "loss": 1.6280248165130615,
+      "loss_ce": 0.28513529896736145,
+      "loss_iou": 0.6276062726974487,
+      "loss_num": 0.017578125,
+      "loss_xval": 1.3428895473480225,
+      "num_input_tokens_seen": 587387416,
+      "step": 3398
+    },
+    {
+      "epoch": 1.301800076599004,
+      "grad_norm": 169.81339489368256,
+      "learning_rate": 5e-06,
+      "loss": 1.3702,
+      "num_input_tokens_seen": 587560144,
+      "step": 3399
+    },
+    {
+      "epoch": 1.301800076599004,
+      "loss": 1.4205771684646606,
+      "loss_ce": 0.2325984686613083,
+      "loss_iou": 0.5562238693237305,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 1.187978744506836,
+      "num_input_tokens_seen": 587560144,
+      "step": 3399
+    },
+    {
+      "epoch": 1.3021830716200689,
+      "grad_norm": 272.7182510346355,
+      "learning_rate": 5e-06,
+      "loss": 2.0319,
+      "num_input_tokens_seen": 587733352,
+      "step": 3400
+    },
+    {
+      "epoch": 1.3021830716200689,
+      "loss": 2.091144561767578,
+      "loss_ce": 0.22405028343200684,
+      "loss_iou": 0.8771965503692627,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 1.8670945167541504,
+      "num_input_tokens_seen": 587733352,
+      "step": 3400
+    },
+    {
+      "epoch": 1.3025660666411336,
+      "grad_norm": 74.46317501421858,
+      "learning_rate": 5e-06,
+      "loss": 2.1574,
+      "num_input_tokens_seen": 587906696,
+      "step": 3401
+    },
+    {
+      "epoch": 1.3025660666411336,
+      "loss": 2.1836647987365723,
+      "loss_ce": 0.2143188863992691,
+      "loss_iou": 0.8512501120567322,
+      "loss_num": 0.053466796875,
+      "loss_xval": 1.9693459272384644,
+      "num_input_tokens_seen": 587906696,
+      "step": 3401
+    },
+    {
+      "epoch": 1.3029490616621984,
+      "grad_norm": 123.36731959699017,
+      "learning_rate": 5e-06,
+      "loss": 1.6449,
+      "num_input_tokens_seen": 588079416,
+      "step": 3402
+    },
+    {
+      "epoch": 1.3029490616621984,
+      "loss": 1.690931797027588,
+      "loss_ce": 0.22199325263500214,
+      "loss_iou": 0.6643704175949097,
+      "loss_num": 0.028076171875,
+      "loss_xval": 1.4689385890960693,
+      "num_input_tokens_seen": 588079416,
+      "step": 3402
+    },
+    {
+      "epoch": 1.3033320566832631,
+      "grad_norm": 99.49372580534332,
+      "learning_rate": 5e-06,
+      "loss": 1.7657,
+      "num_input_tokens_seen": 588251944,
+      "step": 3403
+    },
+    {
+      "epoch": 1.3033320566832631,
+      "loss": 1.8089815378189087,
+      "loss_ce": 0.20827698707580566,
+      "loss_iou": 0.7316877841949463,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 1.6007046699523926,
+      "num_input_tokens_seen": 588251944,
+      "step": 3403
+    },
+    {
+      "epoch": 1.3037150517043279,
+      "grad_norm": 95.52033263317679,
+      "learning_rate": 5e-06,
+      "loss": 1.3348,
+      "num_input_tokens_seen": 588424848,
+      "step": 3404
+    },
+    {
+      "epoch": 1.3037150517043279,
+      "loss": 1.454200267791748,
+      "loss_ce": 0.2225445806980133,
+      "loss_iou": 0.5643294453620911,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 1.2316557168960571,
+      "num_input_tokens_seen": 588424848,
+      "step": 3404
+    },
+    {
+      "epoch": 1.3040980467253926,
+      "grad_norm": 104.31693829184277,
+      "learning_rate": 5e-06,
+      "loss": 1.5788,
+      "num_input_tokens_seen": 588597792,
+      "step": 3405
+    },
+    {
+      "epoch": 1.3040980467253926,
+      "loss": 1.558342456817627,
+      "loss_ce": 0.18813736736774445,
+      "loss_iou": 0.6328411102294922,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 1.3702049255371094,
+      "num_input_tokens_seen": 588597792,
+      "step": 3405
+    },
+    {
+      "epoch": 1.3044810417464574,
+      "grad_norm": 131.06770403703226,
+      "learning_rate": 5e-06,
+      "loss": 1.8656,
+      "num_input_tokens_seen": 588770904,
+      "step": 3406
+    },
+    {
+      "epoch": 1.3044810417464574,
+      "loss": 1.9378714561462402,
+      "loss_ce": 0.23714855313301086,
+      "loss_iou": 0.8048138618469238,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 1.7007226943969727,
+      "num_input_tokens_seen": 588770904,
+      "step": 3406
+    },
+    {
+      "epoch": 1.304864036767522,
+      "grad_norm": 98.80346222293404,
+      "learning_rate": 5e-06,
+      "loss": 1.6852,
+      "num_input_tokens_seen": 588941032,
+      "step": 3407
+    },
+    {
+      "epoch": 1.304864036767522,
+      "loss": 1.6783905029296875,
+      "loss_ce": 0.24246475100517273,
+      "loss_iou": 0.6664491891860962,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 1.4359257221221924,
+      "num_input_tokens_seen": 588941032,
+      "step": 3407
+    },
+    {
+      "epoch": 1.3052470317885867,
+      "grad_norm": 70.35809587279714,
+      "learning_rate": 5e-06,
+      "loss": 1.8113,
+      "num_input_tokens_seen": 589113768,
+      "step": 3408
+    },
+    {
+      "epoch": 1.3052470317885867,
+      "loss": 1.8585872650146484,
+      "loss_ce": 0.19603464007377625,
+      "loss_iou": 0.7664875388145447,
+      "loss_num": 0.02587890625,
+      "loss_xval": 1.6625527143478394,
+      "num_input_tokens_seen": 589113768,
+      "step": 3408
+    },
+    {
+      "epoch": 1.3056300268096515,
+      "grad_norm": 98.41693683214676,
+      "learning_rate": 5e-06,
+      "loss": 1.4478,
+      "num_input_tokens_seen": 589286864,
+      "step": 3409
+    },
+    {
+      "epoch": 1.3056300268096515,
+      "loss": 1.317795991897583,
+      "loss_ce": 0.23353098332881927,
+      "loss_iou": 0.49277031421661377,
+      "loss_num": 0.019775390625,
+      "loss_xval": 1.0842649936676025,
+      "num_input_tokens_seen": 589286864,
+      "step": 3409
+    },
+    {
+      "epoch": 1.3060130218307162,
+      "grad_norm": 245.15268466888702,
+      "learning_rate": 5e-06,
+      "loss": 1.6307,
+      "num_input_tokens_seen": 589459896,
+      "step": 3410
+    },
+    {
+      "epoch": 1.3060130218307162,
+      "loss": 1.6969010829925537,
+      "loss_ce": 0.25010430812835693,
+      "loss_iou": 0.6713963747024536,
+      "loss_num": 0.020751953125,
+      "loss_xval": 1.4467966556549072,
+      "num_input_tokens_seen": 589459896,
+      "step": 3410
+    },
+    {
+      "epoch": 1.306396016851781,
+      "grad_norm": 108.56488167085968,
+      "learning_rate": 5e-06,
+      "loss": 1.4357,
+      "num_input_tokens_seen": 589632448,
+      "step": 3411
+    },
+    {
+      "epoch": 1.306396016851781,
+      "loss": 1.6918354034423828,
+      "loss_ce": 0.2020936906337738,
+      "loss_iou": 0.6973243951797485,
+      "loss_num": 0.01904296875,
+      "loss_xval": 1.489741563796997,
+      "num_input_tokens_seen": 589632448,
+      "step": 3411
+    },
+    {
+      "epoch": 1.3067790118728457,
+      "grad_norm": 93.64761251480384,
+      "learning_rate": 5e-06,
+      "loss": 1.6644,
+      "num_input_tokens_seen": 589805568,
+      "step": 3412
+    },
+    {
+      "epoch": 1.3067790118728457,
+      "loss": 1.5487308502197266,
+      "loss_ce": 0.21271872520446777,
+      "loss_iou": 0.6254035234451294,
+      "loss_num": 0.01708984375,
+      "loss_xval": 1.3360121250152588,
+      "num_input_tokens_seen": 589805568,
+      "step": 3412
+    },
+    {
+      "epoch": 1.3071620068939103,
+      "grad_norm": 165.7201224858923,
+      "learning_rate": 5e-06,
+      "loss": 1.5928,
+      "num_input_tokens_seen": 589978696,
+      "step": 3413
+    },
+    {
+      "epoch": 1.3071620068939103,
+      "loss": 1.587674856185913,
+      "loss_ce": 0.24202266335487366,
+      "loss_iou": 0.6225178837776184,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 1.3456522226333618,
+      "num_input_tokens_seen": 589978696,
+      "step": 3413
+    },
+    {
+      "epoch": 1.307545001914975,
+      "grad_norm": 128.49288898734875,
+      "learning_rate": 5e-06,
+      "loss": 1.9588,
+      "num_input_tokens_seen": 590151728,
+      "step": 3414
+    },
+    {
+      "epoch": 1.307545001914975,
+      "loss": 2.0160014629364014,
+      "loss_ce": 0.20754095911979675,
+      "loss_iou": 0.8112126588821411,
+      "loss_num": 0.037109375,
+      "loss_xval": 1.8084604740142822,
+      "num_input_tokens_seen": 590151728,
+      "step": 3414
+    },
+    {
+      "epoch": 1.3079279969360398,
+      "grad_norm": 81.97583984885038,
+      "learning_rate": 5e-06,
+      "loss": 1.5687,
+      "num_input_tokens_seen": 590324792,
+      "step": 3415
+    },
+    {
+      "epoch": 1.3079279969360398,
+      "loss": 1.5454689264297485,
+      "loss_ce": 0.2742193043231964,
+      "loss_iou": 0.592549204826355,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 1.271249532699585,
+      "num_input_tokens_seen": 590324792,
+      "step": 3415
+    },
+    {
+      "epoch": 1.3083109919571045,
+      "grad_norm": 89.25642109482489,
+      "learning_rate": 5e-06,
+      "loss": 1.6331,
+      "num_input_tokens_seen": 590497752,
+      "step": 3416
+    },
+    {
+      "epoch": 1.3083109919571045,
+      "loss": 1.6145857572555542,
+      "loss_ce": 0.2385154664516449,
+      "loss_iou": 0.6140300631523132,
+      "loss_num": 0.029541015625,
+      "loss_xval": 1.3760703802108765,
+      "num_input_tokens_seen": 590497752,
+      "step": 3416
+    },
+    {
+      "epoch": 1.3086939869781693,
+      "grad_norm": 99.03178678196664,
+      "learning_rate": 5e-06,
+      "loss": 1.5583,
+      "num_input_tokens_seen": 590670808,
+      "step": 3417
+    },
+    {
+      "epoch": 1.3086939869781693,
+      "loss": 1.5358097553253174,
+      "loss_ce": 0.2650596499443054,
+      "loss_iou": 0.5914602279663086,
+      "loss_num": 0.017578125,
+      "loss_xval": 1.2707500457763672,
+      "num_input_tokens_seen": 590670808,
+      "step": 3417
+    },
+    {
+      "epoch": 1.309076981999234,
+      "grad_norm": 128.51842962470516,
+      "learning_rate": 5e-06,
+      "loss": 1.3123,
+      "num_input_tokens_seen": 590843984,
+      "step": 3418
+    },
+    {
+      "epoch": 1.309076981999234,
+      "loss": 1.1670749187469482,
+      "loss_ce": 0.19401878118515015,
+      "loss_iou": 0.46426552534103394,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.9730561971664429,
+      "num_input_tokens_seen": 590843984,
+      "step": 3418
+    },
+    {
+      "epoch": 1.3094599770202988,
+      "grad_norm": 124.19813938301355,
+      "learning_rate": 5e-06,
+      "loss": 1.6318,
+      "num_input_tokens_seen": 591016968,
+      "step": 3419
+    },
+    {
+      "epoch": 1.3094599770202988,
+      "loss": 1.706656813621521,
+      "loss_ce": 0.1851901412010193,
+      "loss_iou": 0.7119662761688232,
+      "loss_num": 0.01953125,
+      "loss_xval": 1.5214667320251465,
+      "num_input_tokens_seen": 591016968,
+      "step": 3419
+    },
+    {
+      "epoch": 1.3098429720413636,
+      "grad_norm": 131.65719949900188,
+      "learning_rate": 5e-06,
+      "loss": 1.5282,
+      "num_input_tokens_seen": 591189720,
+      "step": 3420
+    },
+    {
+      "epoch": 1.3098429720413636,
+      "loss": 1.5373845100402832,
+      "loss_ce": 0.19379600882530212,
+      "loss_iou": 0.6266740560531616,
+      "loss_num": 0.01806640625,
+      "loss_xval": 1.3435885906219482,
+      "num_input_tokens_seen": 591189720,
+      "step": 3420
+    },
+    {
+      "epoch": 1.310225967062428,
+      "grad_norm": 128.56683467176094,
+      "learning_rate": 5e-06,
+      "loss": 1.9242,
+      "num_input_tokens_seen": 591362656,
+      "step": 3421
+    },
+    {
+      "epoch": 1.310225967062428,
+      "loss": 1.9553898572921753,
+      "loss_ce": 0.2091091424226761,
+      "loss_iou": 0.7950153350830078,
+      "loss_num": 0.03125,
+      "loss_xval": 1.7462806701660156,
+      "num_input_tokens_seen": 591362656,
+      "step": 3421
+    },
+    {
+      "epoch": 1.3106089620834929,
+      "grad_norm": 103.29049215670747,
+      "learning_rate": 5e-06,
+      "loss": 1.5314,
+      "num_input_tokens_seen": 591535368,
+      "step": 3422
+    },
+    {
+      "epoch": 1.3106089620834929,
+      "loss": 1.6691319942474365,
+      "loss_ce": 0.17890393733978271,
+      "loss_iou": 0.6813018321990967,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 1.4902281761169434,
+      "num_input_tokens_seen": 591535368,
+      "step": 3422
+    },
+    {
+      "epoch": 1.3109919571045576,
+      "grad_norm": 167.01470859700683,
+      "learning_rate": 5e-06,
+      "loss": 2.5699,
+      "num_input_tokens_seen": 591708368,
+      "step": 3423
+    },
+    {
+      "epoch": 1.3109919571045576,
+      "loss": 2.400451183319092,
+      "loss_ce": 0.251986563205719,
+      "loss_iou": 0.961805522441864,
+      "loss_num": 0.044921875,
+      "loss_xval": 2.1484644412994385,
+      "num_input_tokens_seen": 591708368,
+      "step": 3423
+    },
+    {
+      "epoch": 1.3113749521256224,
+      "grad_norm": 77.31027702011308,
+      "learning_rate": 5e-06,
+      "loss": 1.735,
+      "num_input_tokens_seen": 591881200,
+      "step": 3424
+    },
+    {
+      "epoch": 1.3113749521256224,
+      "loss": 1.7121155261993408,
+      "loss_ce": 0.23263107240200043,
+      "loss_iou": 0.6526145935058594,
+      "loss_num": 0.034912109375,
+      "loss_xval": 1.4794845581054688,
+      "num_input_tokens_seen": 591881200,
+      "step": 3424
+    },
+    {
+      "epoch": 1.3117579471466871,
+      "grad_norm": 111.6612291674844,
+      "learning_rate": 5e-06,
+      "loss": 1.6015,
+      "num_input_tokens_seen": 592054368,
+      "step": 3425
+    },
+    {
+      "epoch": 1.3117579471466871,
+      "loss": 1.5268713235855103,
+      "loss_ce": 0.30413222312927246,
+      "loss_iou": 0.5636705756187439,
+      "loss_num": 0.01904296875,
+      "loss_xval": 1.2227391004562378,
+      "num_input_tokens_seen": 592054368,
+      "step": 3425
+    },
+    {
+      "epoch": 1.312140942167752,
+      "grad_norm": 191.7757606302062,
+      "learning_rate": 5e-06,
+      "loss": 1.5402,
+      "num_input_tokens_seen": 592227528,
+      "step": 3426
+    },
+    {
+      "epoch": 1.312140942167752,
+      "loss": 1.368982195854187,
+      "loss_ce": 0.18858899176120758,
+      "loss_iou": 0.5364398956298828,
+      "loss_num": 0.021484375,
+      "loss_xval": 1.1803932189941406,
+      "num_input_tokens_seen": 592227528,
+      "step": 3426
+    },
+    {
+      "epoch": 1.3125239371888164,
+      "grad_norm": 261.1235860404239,
+      "learning_rate": 5e-06,
+      "loss": 2.0906,
+      "num_input_tokens_seen": 592400528,
+      "step": 3427
+    },
+    {
+      "epoch": 1.3125239371888164,
+      "loss": 1.9674935340881348,
+      "loss_ce": 0.20730724930763245,
+      "loss_iou": 0.8112759590148926,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 1.7601861953735352,
+      "num_input_tokens_seen": 592400528,
+      "step": 3427
+    },
+    {
+      "epoch": 1.3129069322098812,
+      "grad_norm": 108.63903059318454,
+      "learning_rate": 5e-06,
+      "loss": 1.7872,
+      "num_input_tokens_seen": 592573792,
+      "step": 3428
+    },
+    {
+      "epoch": 1.3129069322098812,
+      "loss": 1.8051846027374268,
+      "loss_ce": 0.20400851964950562,
+      "loss_iou": 0.7238973379135132,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 1.6011760234832764,
+      "num_input_tokens_seen": 592573792,
+      "step": 3428
+    },
+    {
+      "epoch": 1.313289927230946,
+      "grad_norm": 123.06938863619305,
+      "learning_rate": 5e-06,
+      "loss": 1.4081,
+      "num_input_tokens_seen": 592746480,
+      "step": 3429
+    },
+    {
+      "epoch": 1.313289927230946,
+      "loss": 1.3916993141174316,
+      "loss_ce": 0.19342002272605896,
+      "loss_iou": 0.5483736991882324,
+      "loss_num": 0.020263671875,
+      "loss_xval": 1.1982793807983398,
+      "num_input_tokens_seen": 592746480,
+      "step": 3429
+    },
+    {
+      "epoch": 1.3136729222520107,
+      "grad_norm": 103.83790948129145,
+      "learning_rate": 5e-06,
+      "loss": 1.8334,
+      "num_input_tokens_seen": 592919472,
+      "step": 3430
+    },
+    {
+      "epoch": 1.3136729222520107,
+      "loss": 1.760236144065857,
+      "loss_ce": 0.21684814989566803,
+      "loss_iou": 0.6988484859466553,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 1.5433878898620605,
+      "num_input_tokens_seen": 592919472,
+      "step": 3430
+    },
+    {
+      "epoch": 1.3140559172730755,
+      "grad_norm": 122.655449245257,
+      "learning_rate": 5e-06,
+      "loss": 1.3296,
+      "num_input_tokens_seen": 593092216,
+      "step": 3431
+    },
+    {
+      "epoch": 1.3140559172730755,
+      "loss": 1.393941879272461,
+      "loss_ce": 0.23210212588310242,
+      "loss_iou": 0.5376459956169128,
+      "loss_num": 0.017333984375,
+      "loss_xval": 1.1618398427963257,
+      "num_input_tokens_seen": 593092216,
+      "step": 3431
+    },
+    {
+      "epoch": 1.3144389122941402,
+      "grad_norm": 94.77904336321453,
+      "learning_rate": 5e-06,
+      "loss": 1.6082,
+      "num_input_tokens_seen": 593264800,
+      "step": 3432
+    },
+    {
+      "epoch": 1.3144389122941402,
+      "loss": 1.6623306274414062,
+      "loss_ce": 0.20671777427196503,
+      "loss_iou": 0.6692584753036499,
+      "loss_num": 0.0234375,
+      "loss_xval": 1.4556128978729248,
+      "num_input_tokens_seen": 593264800,
+      "step": 3432
+    },
+    {
+      "epoch": 1.314821907315205,
+      "grad_norm": 147.30436807819416,
+      "learning_rate": 5e-06,
+      "loss": 1.5344,
+      "num_input_tokens_seen": 593437800,
+      "step": 3433
+    },
+    {
+      "epoch": 1.314821907315205,
+      "loss": 1.4200994968414307,
+      "loss_ce": 0.231092631816864,
+      "loss_iou": 0.5567836761474609,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 1.1890068054199219,
+      "num_input_tokens_seen": 593437800,
+      "step": 3433
+    },
+    {
+      "epoch": 1.3152049023362697,
+      "grad_norm": 106.637022094441,
+      "learning_rate": 5e-06,
+      "loss": 1.9204,
+      "num_input_tokens_seen": 593610720,
+      "step": 3434
+    },
+    {
+      "epoch": 1.3152049023362697,
+      "loss": 2.031893730163574,
+      "loss_ce": 0.20427463948726654,
+      "loss_iou": 0.7924715876579285,
+      "loss_num": 0.048583984375,
+      "loss_xval": 1.8276190757751465,
+      "num_input_tokens_seen": 593610720,
+      "step": 3434
+    },
+    {
+      "epoch": 1.3155878973573343,
+      "grad_norm": 128.15138920085224,
+      "learning_rate": 5e-06,
+      "loss": 1.4556,
+      "num_input_tokens_seen": 593783784,
+      "step": 3435
+    },
+    {
+      "epoch": 1.3155878973573343,
+      "loss": 1.2844393253326416,
+      "loss_ce": 0.23994150757789612,
+      "loss_iou": 0.47250521183013916,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 1.0444977283477783,
+      "num_input_tokens_seen": 593783784,
+      "step": 3435
+    },
+    {
+      "epoch": 1.315970892378399,
+      "grad_norm": 199.4011930943759,
+      "learning_rate": 5e-06,
+      "loss": 1.8581,
+      "num_input_tokens_seen": 593956616,
+      "step": 3436
+    },
+    {
+      "epoch": 1.315970892378399,
+      "loss": 1.9263231754302979,
+      "loss_ce": 0.19064660370349884,
+      "loss_iou": 0.8233740925788879,
+      "loss_num": 0.017822265625,
+      "loss_xval": 1.7356765270233154,
+      "num_input_tokens_seen": 593956616,
+      "step": 3436
+    },
+    {
+      "epoch": 1.3163538873994638,
+      "grad_norm": 91.87156513692791,
+      "learning_rate": 5e-06,
+      "loss": 1.9686,
+      "num_input_tokens_seen": 594129520,
+      "step": 3437
+    },
+    {
+      "epoch": 1.3163538873994638,
+      "loss": 2.0934529304504395,
+      "loss_ce": 0.20075491070747375,
+      "loss_iou": 0.871367335319519,
+      "loss_num": 0.030029296875,
+      "loss_xval": 1.892698049545288,
+      "num_input_tokens_seen": 594129520,
+      "step": 3437
+    },
+    {
+      "epoch": 1.3167368824205286,
+      "grad_norm": 71.4230553026647,
+      "learning_rate": 5e-06,
+      "loss": 1.5046,
+      "num_input_tokens_seen": 594302296,
+      "step": 3438
+    },
+    {
+      "epoch": 1.3167368824205286,
+      "loss": 1.48215651512146,
+      "loss_ce": 0.2117040455341339,
+      "loss_iou": 0.5868406295776367,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 1.2704524993896484,
+      "num_input_tokens_seen": 594302296,
+      "step": 3438
+    },
+    {
+      "epoch": 1.3171198774415933,
+      "grad_norm": 232.85164408634435,
+      "learning_rate": 5e-06,
+      "loss": 1.4754,
+      "num_input_tokens_seen": 594475336,
+      "step": 3439
+    },
+    {
+      "epoch": 1.3171198774415933,
+      "loss": 1.3627231121063232,
+      "loss_ce": 0.22045598924160004,
+      "loss_iou": 0.5019044280052185,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 1.142267107963562,
+      "num_input_tokens_seen": 594475336,
+      "step": 3439
+    },
+    {
+      "epoch": 1.3175028724626578,
+      "grad_norm": 125.46307331108713,
+      "learning_rate": 5e-06,
+      "loss": 1.8846,
+      "num_input_tokens_seen": 594648088,
+      "step": 3440
+    },
+    {
+      "epoch": 1.3175028724626578,
+      "loss": 1.8807228803634644,
+      "loss_ce": 0.22862456738948822,
+      "loss_iou": 0.7542718052864075,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 1.652098298072815,
+      "num_input_tokens_seen": 594648088,
+      "step": 3440
+    },
+    {
+      "epoch": 1.3178858674837226,
+      "grad_norm": 140.29810763456942,
+      "learning_rate": 5e-06,
+      "loss": 1.6031,
+      "num_input_tokens_seen": 594821080,
+      "step": 3441
+    },
+    {
+      "epoch": 1.3178858674837226,
+      "loss": 1.528963327407837,
+      "loss_ce": 0.2620272934436798,
+      "loss_iou": 0.5841668844223022,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 1.2669360637664795,
+      "num_input_tokens_seen": 594821080,
+      "step": 3441
+    },
+    {
+      "epoch": 1.3182688625047874,
+      "grad_norm": 152.95364929288792,
+      "learning_rate": 5e-06,
+      "loss": 1.6388,
+      "num_input_tokens_seen": 594993744,
+      "step": 3442
+    },
+    {
+      "epoch": 1.3182688625047874,
+      "loss": 1.4266736507415771,
+      "loss_ce": 0.18095433712005615,
+      "loss_iou": 0.5637319087982178,
+      "loss_num": 0.023681640625,
+      "loss_xval": 1.2457194328308105,
+      "num_input_tokens_seen": 594993744,
+      "step": 3442
+    },
+    {
+      "epoch": 1.3186518575258521,
+      "grad_norm": 230.79285770032826,
+      "learning_rate": 5e-06,
+      "loss": 1.9245,
+      "num_input_tokens_seen": 595166472,
+      "step": 3443
+    },
+    {
+      "epoch": 1.3186518575258521,
+      "loss": 2.0874247550964355,
+      "loss_ce": 0.19721028208732605,
+      "loss_iou": 0.8709800243377686,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 1.890214443206787,
+      "num_input_tokens_seen": 595166472,
+      "step": 3443
+    },
+    {
+      "epoch": 1.3190348525469169,
+      "grad_norm": 165.79366498689848,
+      "learning_rate": 5e-06,
+      "loss": 1.5884,
+      "num_input_tokens_seen": 595339312,
+      "step": 3444
+    },
+    {
+      "epoch": 1.3190348525469169,
+      "loss": 1.4338977336883545,
+      "loss_ce": 0.14951470494270325,
+      "loss_iou": 0.5644022226333618,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 1.2704670429229736,
+      "num_input_tokens_seen": 595339312,
+      "step": 3444
+    },
+    {
+      "epoch": 1.3194178475679816,
+      "grad_norm": 159.2525834189813,
+      "learning_rate": 5e-06,
+      "loss": 1.8316,
+      "num_input_tokens_seen": 595512416,
+      "step": 3445
+    },
+    {
+      "epoch": 1.3194178475679816,
+      "loss": 1.8100316524505615,
+      "loss_ce": 0.19420024752616882,
+      "loss_iou": 0.7189875245094299,
+      "loss_num": 0.03564453125,
+      "loss_xval": 1.6158314943313599,
+      "num_input_tokens_seen": 595512416,
+      "step": 3445
+    },
+    {
+      "epoch": 1.3198008425890464,
+      "grad_norm": 150.4904805822265,
+      "learning_rate": 5e-06,
+      "loss": 1.4687,
+      "num_input_tokens_seen": 595685248,
+      "step": 3446
+    },
+    {
+      "epoch": 1.3198008425890464,
+      "loss": 1.5566127300262451,
+      "loss_ce": 0.22897005081176758,
+      "loss_iou": 0.6195403337478638,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 1.3276426792144775,
+      "num_input_tokens_seen": 595685248,
+      "step": 3446
+    },
+    {
+      "epoch": 1.3201838376101112,
+      "grad_norm": 337.42542280897067,
+      "learning_rate": 5e-06,
+      "loss": 1.8937,
+      "num_input_tokens_seen": 595858256,
+      "step": 3447
+    },
+    {
+      "epoch": 1.3201838376101112,
+      "loss": 1.8821277618408203,
+      "loss_ce": 0.19640478491783142,
+      "loss_iou": 0.792202353477478,
+      "loss_num": 0.020263671875,
+      "loss_xval": 1.685723066329956,
+      "num_input_tokens_seen": 595858256,
+      "step": 3447
+    },
+    {
+      "epoch": 1.320566832631176,
+      "grad_norm": 138.67140759043002,
+      "learning_rate": 5e-06,
+      "loss": 2.2424,
+      "num_input_tokens_seen": 596031208,
+      "step": 3448
+    },
+    {
+      "epoch": 1.320566832631176,
+      "loss": 2.335386276245117,
+      "loss_ce": 0.22995525598526,
+      "loss_iou": 0.913860559463501,
+      "loss_num": 0.0556640625,
+      "loss_xval": 2.105431079864502,
+      "num_input_tokens_seen": 596031208,
+      "step": 3448
+    },
+    {
+      "epoch": 1.3209498276522404,
+      "grad_norm": 133.25394069902987,
+      "learning_rate": 5e-06,
+      "loss": 1.7018,
+      "num_input_tokens_seen": 596204200,
+      "step": 3449
+    },
+    {
+      "epoch": 1.3209498276522404,
+      "loss": 1.6877968311309814,
+      "loss_ce": 0.22173988819122314,
+      "loss_iou": 0.6771659851074219,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 1.4660568237304688,
+      "num_input_tokens_seen": 596204200,
+      "step": 3449
+    },
+    {
+      "epoch": 1.3213328226733052,
+      "grad_norm": 155.64299683467604,
+      "learning_rate": 5e-06,
+      "loss": 2.1258,
+      "num_input_tokens_seen": 596377008,
+      "step": 3450
+    },
+    {
+      "epoch": 1.3213328226733052,
+      "loss": 1.9931901693344116,
+      "loss_ce": 0.1854073405265808,
+      "loss_iou": 0.8141697645187378,
+      "loss_num": 0.035888671875,
+      "loss_xval": 1.8077828884124756,
+      "num_input_tokens_seen": 596377008,
+      "step": 3450
+    },
+    {
+      "epoch": 1.32171581769437,
+      "grad_norm": 207.4083988067701,
+      "learning_rate": 5e-06,
+      "loss": 1.8708,
+      "num_input_tokens_seen": 596550160,
+      "step": 3451
+    },
+    {
+      "epoch": 1.32171581769437,
+      "loss": 1.7659146785736084,
+      "loss_ce": 0.22494441270828247,
+      "loss_iou": 0.6785661578178406,
+      "loss_num": 0.036865234375,
+      "loss_xval": 1.5409702062606812,
+      "num_input_tokens_seen": 596550160,
+      "step": 3451
+    },
+    {
+      "epoch": 1.3220988127154347,
+      "grad_norm": 187.2132726586493,
+      "learning_rate": 5e-06,
+      "loss": 1.6678,
+      "num_input_tokens_seen": 596723184,
+      "step": 3452
+    },
+    {
+      "epoch": 1.3220988127154347,
+      "loss": 1.6605591773986816,
+      "loss_ce": 0.20885691046714783,
+      "loss_iou": 0.6607265472412109,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 1.4517021179199219,
+      "num_input_tokens_seen": 596723184,
+      "step": 3452
+    },
+    {
+      "epoch": 1.3224818077364995,
+      "grad_norm": 90.99317747748806,
+      "learning_rate": 5e-06,
+      "loss": 1.5969,
+      "num_input_tokens_seen": 596896464,
+      "step": 3453
+    },
+    {
+      "epoch": 1.3224818077364995,
+      "loss": 1.510523796081543,
+      "loss_ce": 0.22982454299926758,
+      "loss_iou": 0.5661919713020325,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 1.280699372291565,
+      "num_input_tokens_seen": 596896464,
+      "step": 3453
+    },
+    {
+      "epoch": 1.322864802757564,
+      "grad_norm": 141.30134918580436,
+      "learning_rate": 5e-06,
+      "loss": 1.2841,
+      "num_input_tokens_seen": 597069352,
+      "step": 3454
+    },
+    {
+      "epoch": 1.322864802757564,
+      "loss": 1.2556250095367432,
+      "loss_ce": 0.24805593490600586,
+      "loss_iou": 0.46800267696380615,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.9946906566619873,
+      "num_input_tokens_seen": 597069352,
+      "step": 3454
+    },
+    {
+      "epoch": 1.3232477977786288,
+      "grad_norm": 148.12058201644876,
+      "learning_rate": 5e-06,
+      "loss": 1.5733,
+      "num_input_tokens_seen": 597242064,
+      "step": 3455
+    },
+    {
+      "epoch": 1.3232477977786288,
+      "loss": 1.5438635349273682,
+      "loss_ce": 0.22933542728424072,
+      "loss_iou": 0.590186357498169,
+      "loss_num": 0.02685546875,
+      "loss_xval": 1.314527988433838,
+      "num_input_tokens_seen": 597242064,
+      "step": 3455
+    },
+    {
+      "epoch": 1.3236307927996935,
+      "grad_norm": 86.24126150448357,
+      "learning_rate": 5e-06,
+      "loss": 1.415,
+      "num_input_tokens_seen": 597414864,
+      "step": 3456
+    },
+    {
+      "epoch": 1.3236307927996935,
+      "loss": 1.4794440269470215,
+      "loss_ce": 0.19419018924236298,
+      "loss_iou": 0.598406970500946,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 1.285253882408142,
+      "num_input_tokens_seen": 597414864,
+      "step": 3456
+    },
+    {
+      "epoch": 1.3240137878207583,
+      "grad_norm": 205.26520536273173,
+      "learning_rate": 5e-06,
+      "loss": 1.5919,
+      "num_input_tokens_seen": 597587928,
+      "step": 3457
+    },
+    {
+      "epoch": 1.3240137878207583,
+      "loss": 1.6292989253997803,
+      "loss_ce": 0.18911823630332947,
+      "loss_iou": 0.678128719329834,
+      "loss_num": 0.016845703125,
+      "loss_xval": 1.440180778503418,
+      "num_input_tokens_seen": 597587928,
+      "step": 3457
+    },
+    {
+      "epoch": 1.324396782841823,
+      "grad_norm": 148.95097772210877,
+      "learning_rate": 5e-06,
+      "loss": 1.5373,
+      "num_input_tokens_seen": 597760640,
+      "step": 3458
+    },
+    {
+      "epoch": 1.324396782841823,
+      "loss": 1.580047845840454,
+      "loss_ce": 0.1832701712846756,
+      "loss_iou": 0.6467683911323547,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 1.3967777490615845,
+      "num_input_tokens_seen": 597760640,
+      "step": 3458
+    },
+    {
+      "epoch": 1.3247797778628878,
+      "grad_norm": 212.87837915686478,
+      "learning_rate": 5e-06,
+      "loss": 2.0722,
+      "num_input_tokens_seen": 597933448,
+      "step": 3459
+    },
+    {
+      "epoch": 1.3247797778628878,
+      "loss": 2.0425939559936523,
+      "loss_ce": 0.26647183299064636,
+      "loss_iou": 0.8155207633972168,
+      "loss_num": 0.029052734375,
+      "loss_xval": 1.7761220932006836,
+      "num_input_tokens_seen": 597933448,
+      "step": 3459
+    },
+    {
+      "epoch": 1.3251627728839526,
+      "grad_norm": 164.9439321869532,
+      "learning_rate": 5e-06,
+      "loss": 1.7587,
+      "num_input_tokens_seen": 598106576,
+      "step": 3460
+    },
+    {
+      "epoch": 1.3251627728839526,
+      "loss": 1.8102011680603027,
+      "loss_ce": 0.23557034134864807,
+      "loss_iou": 0.7101669311523438,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 1.5746307373046875,
+      "num_input_tokens_seen": 598106576,
+      "step": 3460
+    },
+    {
+      "epoch": 1.3255457679050173,
+      "grad_norm": 216.54728902785226,
+      "learning_rate": 5e-06,
+      "loss": 1.4584,
+      "num_input_tokens_seen": 598279824,
+      "step": 3461
+    },
+    {
+      "epoch": 1.3255457679050173,
+      "loss": 1.3819681406021118,
+      "loss_ce": 0.19735556840896606,
+      "loss_iou": 0.558477520942688,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 1.184612512588501,
+      "num_input_tokens_seen": 598279824,
+      "step": 3461
+    },
+    {
+      "epoch": 1.325928762926082,
+      "grad_norm": 131.5791298294782,
+      "learning_rate": 5e-06,
+      "loss": 1.9053,
+      "num_input_tokens_seen": 598452728,
+      "step": 3462
+    },
+    {
+      "epoch": 1.325928762926082,
+      "loss": 1.9645386934280396,
+      "loss_ce": 0.1781918704509735,
+      "loss_iou": 0.8089753985404968,
+      "loss_num": 0.03369140625,
+      "loss_xval": 1.7863469123840332,
+      "num_input_tokens_seen": 598452728,
+      "step": 3462
+    },
+    {
+      "epoch": 1.3263117579471466,
+      "grad_norm": 102.69471251638501,
+      "learning_rate": 5e-06,
+      "loss": 1.3905,
+      "num_input_tokens_seen": 598625568,
+      "step": 3463
+    },
+    {
+      "epoch": 1.3263117579471466,
+      "loss": 1.3003264665603638,
+      "loss_ce": 0.1388418972492218,
+      "loss_iou": 0.5353016257286072,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 1.1614845991134644,
+      "num_input_tokens_seen": 598625568,
+      "step": 3463
+    },
+    {
+      "epoch": 1.3266947529682114,
+      "grad_norm": 285.2463484476111,
+      "learning_rate": 5e-06,
+      "loss": 1.6416,
+      "num_input_tokens_seen": 598798448,
+      "step": 3464
+    },
+    {
+      "epoch": 1.3266947529682114,
+      "loss": 1.7376794815063477,
+      "loss_ce": 0.2819916605949402,
+      "loss_iou": 0.6811367869377136,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 1.4556878805160522,
+      "num_input_tokens_seen": 598798448,
+      "step": 3464
+    },
+    {
+      "epoch": 1.3270777479892761,
+      "grad_norm": 86.81365492022437,
+      "learning_rate": 5e-06,
+      "loss": 2.2049,
+      "num_input_tokens_seen": 598971512,
+      "step": 3465
+    },
+    {
+      "epoch": 1.3270777479892761,
+      "loss": 2.11521053314209,
+      "loss_ce": 0.23786184191703796,
+      "loss_iou": 0.8401026129722595,
+      "loss_num": 0.03955078125,
+      "loss_xval": 1.877348780632019,
+      "num_input_tokens_seen": 598971512,
+      "step": 3465
+    },
+    {
+      "epoch": 1.327460743010341,
+      "grad_norm": 184.49532333398986,
+      "learning_rate": 5e-06,
+      "loss": 1.7217,
+      "num_input_tokens_seen": 599144320,
+      "step": 3466
+    },
+    {
+      "epoch": 1.327460743010341,
+      "loss": 1.6120097637176514,
+      "loss_ce": 0.2162342518568039,
+      "loss_iou": 0.6482966542243958,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 1.3957754373550415,
+      "num_input_tokens_seen": 599144320,
+      "step": 3466
+    },
+    {
+      "epoch": 1.3278437380314057,
+      "grad_norm": 198.4854520617887,
+      "learning_rate": 5e-06,
+      "loss": 1.721,
+      "num_input_tokens_seen": 599313952,
+      "step": 3467
+    },
+    {
+      "epoch": 1.3278437380314057,
+      "loss": 1.8033454418182373,
+      "loss_ce": 0.25241461396217346,
+      "loss_iou": 0.7142319679260254,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 1.5509309768676758,
+      "num_input_tokens_seen": 599313952,
+      "step": 3467
+    },
+    {
+      "epoch": 1.3282267330524702,
+      "grad_norm": 136.42639742425405,
+      "learning_rate": 5e-06,
+      "loss": 1.9464,
+      "num_input_tokens_seen": 599487080,
+      "step": 3468
+    },
+    {
+      "epoch": 1.3282267330524702,
+      "loss": 1.9296438694000244,
+      "loss_ce": 0.2463741898536682,
+      "loss_iou": 0.7575284242630005,
+      "loss_num": 0.03369140625,
+      "loss_xval": 1.683269739151001,
+      "num_input_tokens_seen": 599487080,
+      "step": 3468
+    },
+    {
+      "epoch": 1.328609728073535,
+      "grad_norm": 124.98895467778436,
+      "learning_rate": 5e-06,
+      "loss": 1.5272,
+      "num_input_tokens_seen": 599660120,
+      "step": 3469
+    },
+    {
+      "epoch": 1.328609728073535,
+      "loss": 1.6587231159210205,
+      "loss_ce": 0.24159637093544006,
+      "loss_iou": 0.6685395240783691,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 1.4171266555786133,
+      "num_input_tokens_seen": 599660120,
+      "step": 3469
+    },
+    {
+      "epoch": 1.3289927230945997,
+      "grad_norm": 180.30626110863463,
+      "learning_rate": 5e-06,
+      "loss": 1.9557,
+      "num_input_tokens_seen": 599833224,
+      "step": 3470
+    },
+    {
+      "epoch": 1.3289927230945997,
+      "loss": 1.9767966270446777,
+      "loss_ce": 0.22852396965026855,
+      "loss_iou": 0.8138030171394348,
+      "loss_num": 0.024169921875,
+      "loss_xval": 1.7482725381851196,
+      "num_input_tokens_seen": 599833224,
+      "step": 3470
+    },
+    {
+      "epoch": 1.3293757181156645,
+      "grad_norm": 89.62017521644816,
+      "learning_rate": 5e-06,
+      "loss": 1.7182,
+      "num_input_tokens_seen": 600005984,
+      "step": 3471
+    },
+    {
+      "epoch": 1.3293757181156645,
+      "loss": 1.600099802017212,
+      "loss_ce": 0.22685885429382324,
+      "loss_iou": 0.6239984035491943,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 1.3732409477233887,
+      "num_input_tokens_seen": 600005984,
+      "step": 3471
+    },
+    {
+      "epoch": 1.3297587131367292,
+      "grad_norm": 149.46918114341312,
+      "learning_rate": 5e-06,
+      "loss": 1.5313,
+      "num_input_tokens_seen": 600178728,
+      "step": 3472
+    },
+    {
+      "epoch": 1.3297587131367292,
+      "loss": 1.872377634048462,
+      "loss_ce": 0.21394246816635132,
+      "loss_iou": 0.7716308832168579,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 1.6584351062774658,
+      "num_input_tokens_seen": 600178728,
+      "step": 3472
+    },
+    {
+      "epoch": 1.330141708157794,
+      "grad_norm": 126.44145986486113,
+      "learning_rate": 5e-06,
+      "loss": 1.7678,
+      "num_input_tokens_seen": 600351824,
+      "step": 3473
+    },
+    {
+      "epoch": 1.330141708157794,
+      "loss": 1.6964324712753296,
+      "loss_ce": 0.20465189218521118,
+      "loss_iou": 0.6850992441177368,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 1.4917805194854736,
+      "num_input_tokens_seen": 600351824,
+      "step": 3473
+    },
+    {
+      "epoch": 1.3305247031788587,
+      "grad_norm": 128.1187390726551,
+      "learning_rate": 5e-06,
+      "loss": 1.6612,
+      "num_input_tokens_seen": 600524840,
+      "step": 3474
+    },
+    {
+      "epoch": 1.3305247031788587,
+      "loss": 1.6536750793457031,
+      "loss_ce": 0.2202017456293106,
+      "loss_iou": 0.6750801205635071,
+      "loss_num": 0.0166015625,
+      "loss_xval": 1.4334732294082642,
+      "num_input_tokens_seen": 600524840,
+      "step": 3474
+    },
+    {
+      "epoch": 1.3309076981999235,
+      "grad_norm": 149.5298696161023,
+      "learning_rate": 5e-06,
+      "loss": 1.7771,
+      "num_input_tokens_seen": 600697480,
+      "step": 3475
+    },
+    {
+      "epoch": 1.3309076981999235,
+      "loss": 1.8149094581604004,
+      "loss_ce": 0.1902473270893097,
+      "loss_iou": 0.7272480130195618,
+      "loss_num": 0.033935546875,
+      "loss_xval": 1.6246620416641235,
+      "num_input_tokens_seen": 600697480,
+      "step": 3475
+    },
+    {
+      "epoch": 1.3312906932209883,
+      "grad_norm": 113.263239695,
+      "learning_rate": 5e-06,
+      "loss": 1.6474,
+      "num_input_tokens_seen": 600870552,
+      "step": 3476
+    },
+    {
+      "epoch": 1.3312906932209883,
+      "loss": 1.6476244926452637,
+      "loss_ce": 0.2178780734539032,
+      "loss_iou": 0.668364405632019,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 1.429746389389038,
+      "num_input_tokens_seen": 600870552,
+      "step": 3476
+    },
+    {
+      "epoch": 1.3316736882420528,
+      "grad_norm": 215.23388526200247,
+      "learning_rate": 5e-06,
+      "loss": 1.65,
+      "num_input_tokens_seen": 601043640,
+      "step": 3477
+    },
+    {
+      "epoch": 1.3316736882420528,
+      "loss": 1.5874621868133545,
+      "loss_ce": 0.25152575969696045,
+      "loss_iou": 0.6199029684066772,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.3359363079071045,
+      "num_input_tokens_seen": 601043640,
+      "step": 3477
+    },
+    {
+      "epoch": 1.3320566832631175,
+      "grad_norm": 145.5701925704748,
+      "learning_rate": 5e-06,
+      "loss": 1.7505,
+      "num_input_tokens_seen": 601216648,
+      "step": 3478
+    },
+    {
+      "epoch": 1.3320566832631175,
+      "loss": 1.7719186544418335,
+      "loss_ce": 0.2581689655780792,
+      "loss_iou": 0.6935508251190186,
+      "loss_num": 0.025390625,
+      "loss_xval": 1.513749599456787,
+      "num_input_tokens_seen": 601216648,
+      "step": 3478
+    },
+    {
+      "epoch": 1.3324396782841823,
+      "grad_norm": 99.1289875141019,
+      "learning_rate": 5e-06,
+      "loss": 1.3603,
+      "num_input_tokens_seen": 601389632,
+      "step": 3479
+    },
+    {
+      "epoch": 1.3324396782841823,
+      "loss": 1.467305064201355,
+      "loss_ce": 0.17460595071315765,
+      "loss_iou": 0.6101099252700806,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 1.2926990985870361,
+      "num_input_tokens_seen": 601389632,
+      "step": 3479
+    },
+    {
+      "epoch": 1.332822673305247,
+      "grad_norm": 158.04506298297707,
+      "learning_rate": 5e-06,
+      "loss": 1.5577,
+      "num_input_tokens_seen": 601562648,
+      "step": 3480
+    },
+    {
+      "epoch": 1.332822673305247,
+      "loss": 1.596683144569397,
+      "loss_ce": 0.2407524287700653,
+      "loss_iou": 0.6262075901031494,
+      "loss_num": 0.020751953125,
+      "loss_xval": 1.3559308052062988,
+      "num_input_tokens_seen": 601562648,
+      "step": 3480
+    },
+    {
+      "epoch": 1.3332056683263118,
+      "grad_norm": 186.04044373935972,
+      "learning_rate": 5e-06,
+      "loss": 1.6341,
+      "num_input_tokens_seen": 601735416,
+      "step": 3481
+    },
+    {
+      "epoch": 1.3332056683263118,
+      "loss": 1.6800787448883057,
+      "loss_ce": 0.22159579396247864,
+      "loss_iou": 0.6769953966140747,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 1.4584829807281494,
+      "num_input_tokens_seen": 601735416,
+      "step": 3481
+    },
+    {
+      "epoch": 1.3335886633473764,
+      "grad_norm": 78.26524965179112,
+      "learning_rate": 5e-06,
+      "loss": 2.0157,
+      "num_input_tokens_seen": 601908736,
+      "step": 3482
+    },
+    {
+      "epoch": 1.3335886633473764,
+      "loss": 2.0095057487487793,
+      "loss_ce": 0.2427942454814911,
+      "loss_iou": 0.8008362054824829,
+      "loss_num": 0.032958984375,
+      "loss_xval": 1.7667114734649658,
+      "num_input_tokens_seen": 601908736,
+      "step": 3482
+    },
+    {
+      "epoch": 1.3339716583684411,
+      "grad_norm": 58.68459387284598,
+      "learning_rate": 5e-06,
+      "loss": 1.4854,
+      "num_input_tokens_seen": 602081568,
+      "step": 3483
+    },
+    {
+      "epoch": 1.3339716583684411,
+      "loss": 1.6807550191879272,
+      "loss_ce": 0.20905466377735138,
+      "loss_iou": 0.6858318448066711,
+      "loss_num": 0.02001953125,
+      "loss_xval": 1.4717003107070923,
+      "num_input_tokens_seen": 602081568,
+      "step": 3483
+    },
+    {
+      "epoch": 1.3343546533895059,
+      "grad_norm": 225.15046392943285,
+      "learning_rate": 5e-06,
+      "loss": 1.4289,
+      "num_input_tokens_seen": 602254184,
+      "step": 3484
+    },
+    {
+      "epoch": 1.3343546533895059,
+      "loss": 1.378389596939087,
+      "loss_ce": 0.2561156153678894,
+      "loss_iou": 0.5120189189910889,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 1.1222739219665527,
+      "num_input_tokens_seen": 602254184,
+      "step": 3484
+    },
+    {
+      "epoch": 1.3347376484105706,
+      "grad_norm": 149.07422323153412,
+      "learning_rate": 5e-06,
+      "loss": 1.6976,
+      "num_input_tokens_seen": 602426936,
+      "step": 3485
+    },
+    {
+      "epoch": 1.3347376484105706,
+      "loss": 1.577803134918213,
+      "loss_ce": 0.18569129705429077,
+      "loss_iou": 0.6507525444030762,
+      "loss_num": 0.01806640625,
+      "loss_xval": 1.3921117782592773,
+      "num_input_tokens_seen": 602426936,
+      "step": 3485
+    },
+    {
+      "epoch": 1.3351206434316354,
+      "grad_norm": 145.89055009931414,
+      "learning_rate": 5e-06,
+      "loss": 1.558,
+      "num_input_tokens_seen": 602599912,
+      "step": 3486
+    },
+    {
+      "epoch": 1.3351206434316354,
+      "loss": 1.4918200969696045,
+      "loss_ce": 0.2002352774143219,
+      "loss_iou": 0.5969338417053223,
+      "loss_num": 0.01953125,
+      "loss_xval": 1.2915849685668945,
+      "num_input_tokens_seen": 602599912,
+      "step": 3486
+    },
+    {
+      "epoch": 1.3355036384527001,
+      "grad_norm": 131.9695778630955,
+      "learning_rate": 5e-06,
+      "loss": 1.9589,
+      "num_input_tokens_seen": 602772952,
+      "step": 3487
+    },
+    {
+      "epoch": 1.3355036384527001,
+      "loss": 1.8470755815505981,
+      "loss_ce": 0.21932345628738403,
+      "loss_iou": 0.7450588941574097,
+      "loss_num": 0.027587890625,
+      "loss_xval": 1.6277520656585693,
+      "num_input_tokens_seen": 602772952,
+      "step": 3487
+    },
+    {
+      "epoch": 1.335886633473765,
+      "grad_norm": 94.05160146535559,
+      "learning_rate": 5e-06,
+      "loss": 1.4983,
+      "num_input_tokens_seen": 602945816,
+      "step": 3488
+    },
+    {
+      "epoch": 1.335886633473765,
+      "loss": 1.445399284362793,
+      "loss_ce": 0.20202450454235077,
+      "loss_iou": 0.5794968008995056,
+      "loss_num": 0.016845703125,
+      "loss_xval": 1.2433747053146362,
+      "num_input_tokens_seen": 602945816,
+      "step": 3488
+    },
+    {
+      "epoch": 1.3362696284948297,
+      "grad_norm": 102.22866160437519,
+      "learning_rate": 5e-06,
+      "loss": 1.3725,
+      "num_input_tokens_seen": 603118896,
+      "step": 3489
+    },
+    {
+      "epoch": 1.3362696284948297,
+      "loss": 1.4227652549743652,
+      "loss_ce": 0.19586127996444702,
+      "loss_iou": 0.5797452926635742,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 1.2269039154052734,
+      "num_input_tokens_seen": 603118896,
+      "step": 3489
+    },
+    {
+      "epoch": 1.3366526235158944,
+      "grad_norm": 111.9340649797218,
+      "learning_rate": 5e-06,
+      "loss": 1.565,
+      "num_input_tokens_seen": 603291552,
+      "step": 3490
+    },
+    {
+      "epoch": 1.3366526235158944,
+      "loss": 1.350496768951416,
+      "loss_ce": 0.20106390118598938,
+      "loss_iou": 0.526208758354187,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 1.149432897567749,
+      "num_input_tokens_seen": 603291552,
+      "step": 3490
+    },
+    {
+      "epoch": 1.337035618536959,
+      "grad_norm": 123.09725427670344,
+      "learning_rate": 5e-06,
+      "loss": 1.695,
+      "num_input_tokens_seen": 603460576,
+      "step": 3491
+    },
+    {
+      "epoch": 1.337035618536959,
+      "loss": 1.5950605869293213,
+      "loss_ce": 0.1703254133462906,
+      "loss_iou": 0.6679187417030334,
+      "loss_num": 0.017822265625,
+      "loss_xval": 1.424735188484192,
+      "num_input_tokens_seen": 603460576,
+      "step": 3491
+    },
+    {
+      "epoch": 1.3374186135580237,
+      "grad_norm": 260.9762500876592,
+      "learning_rate": 5e-06,
+      "loss": 2.0278,
+      "num_input_tokens_seen": 603633856,
+      "step": 3492
+    },
+    {
+      "epoch": 1.3374186135580237,
+      "loss": 1.98429274559021,
+      "loss_ce": 0.2618187665939331,
+      "loss_iou": 0.8138126134872437,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 1.7224738597869873,
+      "num_input_tokens_seen": 603633856,
+      "step": 3492
+    },
+    {
+      "epoch": 1.3378016085790885,
+      "grad_norm": 128.98469840714696,
+      "learning_rate": 5e-06,
+      "loss": 1.9122,
+      "num_input_tokens_seen": 603806712,
+      "step": 3493
+    },
+    {
+      "epoch": 1.3378016085790885,
+      "loss": 1.9915841817855835,
+      "loss_ce": 0.19809076189994812,
+      "loss_iou": 0.8212156891822815,
+      "loss_num": 0.0302734375,
+      "loss_xval": 1.793493390083313,
+      "num_input_tokens_seen": 603806712,
+      "step": 3493
+    },
+    {
+      "epoch": 1.3381846036001532,
+      "grad_norm": 112.9546501008244,
+      "learning_rate": 5e-06,
+      "loss": 1.5566,
+      "num_input_tokens_seen": 603979632,
+      "step": 3494
+    },
+    {
+      "epoch": 1.3381846036001532,
+      "loss": 1.6332077980041504,
+      "loss_ce": 0.1553523987531662,
+      "loss_iou": 0.6758478283882141,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 1.4778553247451782,
+      "num_input_tokens_seen": 603979632,
+      "step": 3494
+    },
+    {
+      "epoch": 1.338567598621218,
+      "grad_norm": 112.76004031268579,
+      "learning_rate": 5e-06,
+      "loss": 2.0918,
+      "num_input_tokens_seen": 604152776,
+      "step": 3495
+    },
+    {
+      "epoch": 1.338567598621218,
+      "loss": 2.122227191925049,
+      "loss_ce": 0.20239807665348053,
+      "loss_iou": 0.8473047018051147,
+      "loss_num": 0.044921875,
+      "loss_xval": 1.919829249382019,
+      "num_input_tokens_seen": 604152776,
+      "step": 3495
+    },
+    {
+      "epoch": 1.3389505936422825,
+      "grad_norm": 61.13943662348016,
+      "learning_rate": 5e-06,
+      "loss": 1.2617,
+      "num_input_tokens_seen": 604325528,
+      "step": 3496
+    },
+    {
+      "epoch": 1.3389505936422825,
+      "loss": 1.3334009647369385,
+      "loss_ce": 0.2177307903766632,
+      "loss_iou": 0.5026745796203613,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 1.1156702041625977,
+      "num_input_tokens_seen": 604325528,
+      "step": 3496
+    },
+    {
+      "epoch": 1.3393335886633473,
+      "grad_norm": 125.2813718635993,
+      "learning_rate": 5e-06,
+      "loss": 1.6588,
+      "num_input_tokens_seen": 604498376,
+      "step": 3497
+    },
+    {
+      "epoch": 1.3393335886633473,
+      "loss": 1.6947588920593262,
+      "loss_ce": 0.1737125813961029,
+      "loss_iou": 0.6697334051132202,
+      "loss_num": 0.036376953125,
+      "loss_xval": 1.5210464000701904,
+      "num_input_tokens_seen": 604498376,
+      "step": 3497
+    },
+    {
+      "epoch": 1.339716583684412,
+      "grad_norm": 110.09947659583729,
+      "learning_rate": 5e-06,
+      "loss": 1.9428,
+      "num_input_tokens_seen": 604671360,
+      "step": 3498
+    },
+    {
+      "epoch": 1.339716583684412,
+      "loss": 1.9898093938827515,
+      "loss_ce": 0.21323680877685547,
+      "loss_iou": 0.806773841381073,
+      "loss_num": 0.03271484375,
+      "loss_xval": 1.776572585105896,
+      "num_input_tokens_seen": 604671360,
+      "step": 3498
+    },
+    {
+      "epoch": 1.3400995787054768,
+      "grad_norm": 249.006305925835,
+      "learning_rate": 5e-06,
+      "loss": 1.37,
+      "num_input_tokens_seen": 604844368,
+      "step": 3499
+    },
+    {
+      "epoch": 1.3400995787054768,
+      "loss": 1.4726142883300781,
+      "loss_ce": 0.1997637003660202,
+      "loss_iou": 0.5933191776275635,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 1.272850513458252,
+      "num_input_tokens_seen": 604844368,
+      "step": 3499
+    },
+    {
+      "epoch": 1.3404825737265416,
+      "grad_norm": 140.0807669408355,
+      "learning_rate": 5e-06,
+      "loss": 2.0479,
+      "num_input_tokens_seen": 605017096,
+      "step": 3500
+    },
+    {
+      "epoch": 1.3404825737265416,
+      "eval_websight_new_CIoU": 0.9405237138271332,
+      "eval_websight_new_GIoU": 0.940310925245285,
+      "eval_websight_new_IoU": 0.9406976103782654,
+      "eval_websight_new_MAE_all": 0.006035956786945462,
+      "eval_websight_new_MAE_h": 0.004176952177658677,
+      "eval_websight_new_MAE_w": 0.009054332505911589,
+      "eval_websight_new_MAE_x": 0.005822476930916309,
+      "eval_websight_new_MAE_y": 0.005090065533295274,
+      "eval_websight_new_NUM_probability": 0.04790673404932022,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 1.0469005107879639,
+      "eval_websight_new_loss_ce": 0.49881765246391296,
+      "eval_websight_new_loss_iou": 0.2625816911458969,
+      "eval_websight_new_loss_num": 0.00641632080078125,
+      "eval_websight_new_loss_xval": 0.5572678744792938,
+      "eval_websight_new_runtime": 56.503,
+      "eval_websight_new_samples_per_second": 0.885,
+      "eval_websight_new_steps_per_second": 0.035,
+      "num_input_tokens_seen": 605017096,
+      "step": 3500
+    },
+    {
+      "epoch": 1.3404825737265416,
+      "eval_seeclick_CIoU": 0.5847286581993103,
+      "eval_seeclick_GIoU": 0.6049769222736359,
+      "eval_seeclick_IoU": 0.6383702158927917,
+      "eval_seeclick_MAE_all": 0.07808250188827515,
+      "eval_seeclick_MAE_h": 0.029664406552910805,
+      "eval_seeclick_MAE_w": 0.1290624961256981,
+      "eval_seeclick_MAE_x": 0.11963751167058945,
+      "eval_seeclick_MAE_y": 0.03396560251712799,
+      "eval_seeclick_NUM_probability": 0.24144858866930008,
+      "eval_seeclick_inside_bbox": 0.8142361044883728,
+      "eval_seeclick_loss": 1.9916496276855469,
+      "eval_seeclick_loss_ce": 0.33574724197387695,
+      "eval_seeclick_loss_iou": 0.6812882423400879,
+      "eval_seeclick_loss_num": 0.0514373779296875,
+      "eval_seeclick_loss_xval": 1.6200228333473206,
+      "eval_seeclick_runtime": 92.076,
+      "eval_seeclick_samples_per_second": 0.543,
+      "eval_seeclick_steps_per_second": 0.022,
+      "num_input_tokens_seen": 605017096,
+      "step": 3500
+    },
+    {
+      "epoch": 1.3404825737265416,
+      "eval_icons_CIoU": 0.8462036848068237,
+      "eval_icons_GIoU": 0.8470831513404846,
+      "eval_icons_IoU": 0.8618262708187103,
+      "eval_icons_MAE_all": 0.027814035303890705,
+      "eval_icons_MAE_h": 0.014042175374925137,
+      "eval_icons_MAE_w": 0.04157224856317043,
+      "eval_icons_MAE_x": 0.04165832698345184,
+      "eval_icons_MAE_y": 0.01398338982835412,
+      "eval_icons_NUM_probability": 0.32583674788475037,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 1.869412899017334,
+      "eval_icons_loss_ce": 0.40531717240810394,
+      "eval_icons_loss_iou": 0.6782236099243164,
+      "eval_icons_loss_num": 0.013891220092773438,
+      "eval_icons_loss_xval": 1.425851821899414,
+      "eval_icons_runtime": 84.4096,
+      "eval_icons_samples_per_second": 0.592,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 605017096,
+      "step": 3500
+    },
+    {
+      "epoch": 1.3404825737265416,
+      "loss": 1.8371400833129883,
+      "loss_ce": 0.40471842885017395,
+      "loss_iou": 0.6813368797302246,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 1.4324216842651367,
+      "num_input_tokens_seen": 605017096,
+      "step": 3500
+    },
+    {
+      "epoch": 1.3408655687476063,
+      "grad_norm": 112.9518999820373,
+      "learning_rate": 5e-06,
+      "loss": 1.4466,
+      "num_input_tokens_seen": 605189832,
+      "step": 3501
+    },
+    {
+      "epoch": 1.3408655687476063,
+      "loss": 1.4242026805877686,
+      "loss_ce": 0.2152501493692398,
+      "loss_iou": 0.5491173267364502,
+      "loss_num": 0.022216796875,
+      "loss_xval": 1.2089524269104004,
+      "num_input_tokens_seen": 605189832,
+      "step": 3501
+    },
+    {
+      "epoch": 1.341248563768671,
+      "grad_norm": 93.35543209952594,
+      "learning_rate": 5e-06,
+      "loss": 1.6746,
+      "num_input_tokens_seen": 605362848,
+      "step": 3502
+    },
+    {
+      "epoch": 1.341248563768671,
+      "loss": 1.530531883239746,
+      "loss_ce": 0.20783312618732452,
+      "loss_iou": 0.6111785173416138,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 1.3226988315582275,
+      "num_input_tokens_seen": 605362848,
+      "step": 3502
+    },
+    {
+      "epoch": 1.3416315587897358,
+      "grad_norm": 144.6148396653385,
+      "learning_rate": 5e-06,
+      "loss": 1.3291,
+      "num_input_tokens_seen": 605535600,
+      "step": 3503
+    },
+    {
+      "epoch": 1.3416315587897358,
+      "loss": 1.2425280809402466,
+      "loss_ce": 0.15439146757125854,
+      "loss_iou": 0.49774259328842163,
+      "loss_num": 0.0185546875,
+      "loss_xval": 1.0881365537643433,
+      "num_input_tokens_seen": 605535600,
+      "step": 3503
+    },
+    {
+      "epoch": 1.3420145538108006,
+      "grad_norm": 108.5974686234195,
+      "learning_rate": 5e-06,
+      "loss": 1.8874,
+      "num_input_tokens_seen": 605708712,
+      "step": 3504
+    },
+    {
+      "epoch": 1.3420145538108006,
+      "loss": 1.9352554082870483,
+      "loss_ce": 0.2253366857767105,
+      "loss_iou": 0.7844027280807495,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 1.709918737411499,
+      "num_input_tokens_seen": 605708712,
+      "step": 3504
+    },
+    {
+      "epoch": 1.3423975488318651,
+      "grad_norm": 87.69758816123331,
+      "learning_rate": 5e-06,
+      "loss": 1.356,
+      "num_input_tokens_seen": 605881576,
+      "step": 3505
+    },
+    {
+      "epoch": 1.3423975488318651,
+      "loss": 1.414637565612793,
+      "loss_ce": 0.2152813822031021,
+      "loss_iou": 0.544441282749176,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 1.199356198310852,
+      "num_input_tokens_seen": 605881576,
+      "step": 3505
+    },
+    {
+      "epoch": 1.3427805438529299,
+      "grad_norm": 187.31954750786883,
+      "learning_rate": 5e-06,
+      "loss": 1.5291,
+      "num_input_tokens_seen": 606054360,
+      "step": 3506
+    },
+    {
+      "epoch": 1.3427805438529299,
+      "loss": 1.408973217010498,
+      "loss_ce": 0.17048245668411255,
+      "loss_iou": 0.5761851072311401,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 1.2384908199310303,
+      "num_input_tokens_seen": 606054360,
+      "step": 3506
+    },
+    {
+      "epoch": 1.3431635388739946,
+      "grad_norm": 128.05965769265785,
+      "learning_rate": 5e-06,
+      "loss": 1.7629,
+      "num_input_tokens_seen": 606227248,
+      "step": 3507
+    },
+    {
+      "epoch": 1.3431635388739946,
+      "loss": 1.6909637451171875,
+      "loss_ce": 0.22392970323562622,
+      "loss_iou": 0.665676474571228,
+      "loss_num": 0.027099609375,
+      "loss_xval": 1.467034101486206,
+      "num_input_tokens_seen": 606227248,
+      "step": 3507
+    },
+    {
+      "epoch": 1.3435465338950594,
+      "grad_norm": 126.99911627210503,
+      "learning_rate": 5e-06,
+      "loss": 1.4575,
+      "num_input_tokens_seen": 606400064,
+      "step": 3508
+    },
+    {
+      "epoch": 1.3435465338950594,
+      "loss": 1.6847248077392578,
+      "loss_ce": 0.22748036682605743,
+      "loss_iou": 0.6699368953704834,
+      "loss_num": 0.0234375,
+      "loss_xval": 1.4572443962097168,
+      "num_input_tokens_seen": 606400064,
+      "step": 3508
+    },
+    {
+      "epoch": 1.3439295289161242,
+      "grad_norm": 140.8328950124996,
+      "learning_rate": 5e-06,
+      "loss": 1.9108,
+      "num_input_tokens_seen": 606573192,
+      "step": 3509
+    },
+    {
+      "epoch": 1.3439295289161242,
+      "loss": 1.89664626121521,
+      "loss_ce": 0.19021235406398773,
+      "loss_iou": 0.7913883328437805,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 1.706433892250061,
+      "num_input_tokens_seen": 606573192,
+      "step": 3509
+    },
+    {
+      "epoch": 1.3443125239371887,
+      "grad_norm": 101.83869826114956,
+      "learning_rate": 5e-06,
+      "loss": 1.6246,
+      "num_input_tokens_seen": 606745992,
+      "step": 3510
+    },
+    {
+      "epoch": 1.3443125239371887,
+      "loss": 1.769139051437378,
+      "loss_ce": 0.1718544363975525,
+      "loss_iou": 0.7396976351737976,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 1.5972846746444702,
+      "num_input_tokens_seen": 606745992,
+      "step": 3510
+    },
+    {
+      "epoch": 1.3446955189582535,
+      "grad_norm": 140.4556739205744,
+      "learning_rate": 5e-06,
+      "loss": 1.4881,
+      "num_input_tokens_seen": 606919272,
+      "step": 3511
+    },
+    {
+      "epoch": 1.3446955189582535,
+      "loss": 1.4090797901153564,
+      "loss_ce": 0.22030985355377197,
+      "loss_iou": 0.5447786450386047,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 1.1887699365615845,
+      "num_input_tokens_seen": 606919272,
+      "step": 3511
+    },
+    {
+      "epoch": 1.3450785139793182,
+      "grad_norm": 113.42709916370082,
+      "learning_rate": 5e-06,
+      "loss": 1.9464,
+      "num_input_tokens_seen": 607092528,
+      "step": 3512
+    },
+    {
+      "epoch": 1.3450785139793182,
+      "loss": 1.916111946105957,
+      "loss_ce": 0.25201815366744995,
+      "loss_iou": 0.7657628059387207,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 1.6640939712524414,
+      "num_input_tokens_seen": 607092528,
+      "step": 3512
+    },
+    {
+      "epoch": 1.345461509000383,
+      "grad_norm": 91.05770661450043,
+      "learning_rate": 5e-06,
+      "loss": 1.4648,
+      "num_input_tokens_seen": 607265624,
+      "step": 3513
+    },
+    {
+      "epoch": 1.345461509000383,
+      "loss": 1.5231677293777466,
+      "loss_ce": 0.2110317200422287,
+      "loss_iou": 0.5995036959648132,
+      "loss_num": 0.022705078125,
+      "loss_xval": 1.3121360540390015,
+      "num_input_tokens_seen": 607265624,
+      "step": 3513
+    },
+    {
+      "epoch": 1.3458445040214477,
+      "grad_norm": 191.65907520626683,
+      "learning_rate": 5e-06,
+      "loss": 1.4242,
+      "num_input_tokens_seen": 607438648,
+      "step": 3514
+    },
+    {
+      "epoch": 1.3458445040214477,
+      "loss": 1.2999374866485596,
+      "loss_ce": 0.1912992000579834,
+      "loss_iou": 0.5133645534515381,
+      "loss_num": 0.016357421875,
+      "loss_xval": 1.1086382865905762,
+      "num_input_tokens_seen": 607438648,
+      "step": 3514
+    },
+    {
+      "epoch": 1.3462274990425125,
+      "grad_norm": 115.8155073206714,
+      "learning_rate": 5e-06,
+      "loss": 2.0663,
+      "num_input_tokens_seen": 607611496,
+      "step": 3515
+    },
+    {
+      "epoch": 1.3462274990425125,
+      "loss": 1.9306249618530273,
+      "loss_ce": 0.15561552345752716,
+      "loss_iou": 0.7690355181694031,
+      "loss_num": 0.04736328125,
+      "loss_xval": 1.7486423254013062,
+      "num_input_tokens_seen": 607611496,
+      "step": 3515
+    },
+    {
+      "epoch": 1.3466104940635772,
+      "grad_norm": 185.59225077752072,
+      "learning_rate": 5e-06,
+      "loss": 1.6295,
+      "num_input_tokens_seen": 607784448,
+      "step": 3516
+    },
+    {
+      "epoch": 1.3466104940635772,
+      "loss": 1.791547417640686,
+      "loss_ce": 0.2342563271522522,
+      "loss_iou": 0.7256364822387695,
+      "loss_num": 0.021240234375,
+      "loss_xval": 1.557291030883789,
+      "num_input_tokens_seen": 607784448,
+      "step": 3516
+    },
+    {
+      "epoch": 1.346993489084642,
+      "grad_norm": 248.29660682207296,
+      "learning_rate": 5e-06,
+      "loss": 2.5236,
+      "num_input_tokens_seen": 607957392,
+      "step": 3517
+    },
+    {
+      "epoch": 1.346993489084642,
+      "loss": 2.4775795936584473,
+      "loss_ce": 0.18136361241340637,
+      "loss_iou": 1.0829529762268066,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 2.2962160110473633,
+      "num_input_tokens_seen": 607957392,
+      "step": 3517
+    },
+    {
+      "epoch": 1.3473764841057068,
+      "grad_norm": 125.59973971151662,
+      "learning_rate": 5e-06,
+      "loss": 1.7813,
+      "num_input_tokens_seen": 608130072,
+      "step": 3518
+    },
+    {
+      "epoch": 1.3473764841057068,
+      "loss": 1.811166524887085,
+      "loss_ce": 0.20386719703674316,
+      "loss_iou": 0.7005002498626709,
+      "loss_num": 0.041259765625,
+      "loss_xval": 1.6072993278503418,
+      "num_input_tokens_seen": 608130072,
+      "step": 3518
+    },
+    {
+      "epoch": 1.3477594791267713,
+      "grad_norm": 343.13462323871494,
+      "learning_rate": 5e-06,
+      "loss": 1.9953,
+      "num_input_tokens_seen": 608303304,
+      "step": 3519
+    },
+    {
+      "epoch": 1.3477594791267713,
+      "loss": 2.134481430053711,
+      "loss_ce": 0.213484525680542,
+      "loss_iou": 0.8866459727287292,
+      "loss_num": 0.029541015625,
+      "loss_xval": 1.9209970235824585,
+      "num_input_tokens_seen": 608303304,
+      "step": 3519
+    },
+    {
+      "epoch": 1.348142474147836,
+      "grad_norm": 124.55212862480187,
+      "learning_rate": 5e-06,
+      "loss": 1.9489,
+      "num_input_tokens_seen": 608476184,
+      "step": 3520
+    },
+    {
+      "epoch": 1.348142474147836,
+      "loss": 1.8672987222671509,
+      "loss_ce": 0.25148433446884155,
+      "loss_iou": 0.723404049873352,
+      "loss_num": 0.03369140625,
+      "loss_xval": 1.615814447402954,
+      "num_input_tokens_seen": 608476184,
+      "step": 3520
+    },
+    {
+      "epoch": 1.3485254691689008,
+      "grad_norm": 289.87028842325446,
+      "learning_rate": 5e-06,
+      "loss": 1.78,
+      "num_input_tokens_seen": 608649008,
+      "step": 3521
+    },
+    {
+      "epoch": 1.3485254691689008,
+      "loss": 1.7762246131896973,
+      "loss_ce": 0.18558359146118164,
+      "loss_iou": 0.7254962921142578,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 1.5906410217285156,
+      "num_input_tokens_seen": 608649008,
+      "step": 3521
+    },
+    {
+      "epoch": 1.3489084641899656,
+      "grad_norm": 88.59708559388797,
+      "learning_rate": 5e-06,
+      "loss": 1.5324,
+      "num_input_tokens_seen": 608821992,
+      "step": 3522
+    },
+    {
+      "epoch": 1.3489084641899656,
+      "loss": 1.530510663986206,
+      "loss_ce": 0.22984153032302856,
+      "loss_iou": 0.5942737460136414,
+      "loss_num": 0.0224609375,
+      "loss_xval": 1.3006690740585327,
+      "num_input_tokens_seen": 608821992,
+      "step": 3522
+    },
+    {
+      "epoch": 1.3492914592110303,
+      "grad_norm": 100.5229129843677,
+      "learning_rate": 5e-06,
+      "loss": 1.496,
+      "num_input_tokens_seen": 608994904,
+      "step": 3523
+    },
+    {
+      "epoch": 1.3492914592110303,
+      "loss": 1.6989405155181885,
+      "loss_ce": 0.18340282142162323,
+      "loss_iou": 0.6874868869781494,
+      "loss_num": 0.028076171875,
+      "loss_xval": 1.5155377388000488,
+      "num_input_tokens_seen": 608994904,
+      "step": 3523
+    },
+    {
+      "epoch": 1.3496744542320949,
+      "grad_norm": 112.6623939901179,
+      "learning_rate": 5e-06,
+      "loss": 1.8221,
+      "num_input_tokens_seen": 609168032,
+      "step": 3524
+    },
+    {
+      "epoch": 1.3496744542320949,
+      "loss": 1.9208285808563232,
+      "loss_ce": 0.21432477235794067,
+      "loss_iou": 0.7663684487342834,
+      "loss_num": 0.03466796875,
+      "loss_xval": 1.706503987312317,
+      "num_input_tokens_seen": 609168032,
+      "step": 3524
+    },
+    {
+      "epoch": 1.3500574492531596,
+      "grad_norm": 95.7038487178071,
+      "learning_rate": 5e-06,
+      "loss": 1.6282,
+      "num_input_tokens_seen": 609340992,
+      "step": 3525
+    },
+    {
+      "epoch": 1.3500574492531596,
+      "loss": 1.6250360012054443,
+      "loss_ce": 0.2213936150074005,
+      "loss_iou": 0.6471489667892456,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 1.4036424160003662,
+      "num_input_tokens_seen": 609340992,
+      "step": 3525
+    },
+    {
+      "epoch": 1.3504404442742244,
+      "grad_norm": 222.96820677305527,
+      "learning_rate": 5e-06,
+      "loss": 1.5,
+      "num_input_tokens_seen": 609513624,
+      "step": 3526
+    },
+    {
+      "epoch": 1.3504404442742244,
+      "loss": 1.4050097465515137,
+      "loss_ce": 0.17917197942733765,
+      "loss_iou": 0.5576211214065552,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 1.2258379459381104,
+      "num_input_tokens_seen": 609513624,
+      "step": 3526
+    },
+    {
+      "epoch": 1.3508234392952891,
+      "grad_norm": 156.76325591007046,
+      "learning_rate": 5e-06,
+      "loss": 1.7096,
+      "num_input_tokens_seen": 609686208,
+      "step": 3527
+    },
+    {
+      "epoch": 1.3508234392952891,
+      "loss": 1.935443639755249,
+      "loss_ce": 0.1976446509361267,
+      "loss_iou": 0.7885772585868835,
+      "loss_num": 0.0322265625,
+      "loss_xval": 1.737799048423767,
+      "num_input_tokens_seen": 609686208,
+      "step": 3527
+    },
+    {
+      "epoch": 1.351206434316354,
+      "grad_norm": 290.8758488378372,
+      "learning_rate": 5e-06,
+      "loss": 2.0351,
+      "num_input_tokens_seen": 609859128,
+      "step": 3528
+    },
+    {
+      "epoch": 1.351206434316354,
+      "loss": 1.888675332069397,
+      "loss_ce": 0.22578385472297668,
+      "loss_iou": 0.7497501373291016,
+      "loss_num": 0.03271484375,
+      "loss_xval": 1.6628913879394531,
+      "num_input_tokens_seen": 609859128,
+      "step": 3528
+    },
+    {
+      "epoch": 1.3515894293374187,
+      "grad_norm": 183.54240686583393,
+      "learning_rate": 5e-06,
+      "loss": 1.5176,
+      "num_input_tokens_seen": 610032096,
+      "step": 3529
+    },
+    {
+      "epoch": 1.3515894293374187,
+      "loss": 1.494391918182373,
+      "loss_ce": 0.1818937063217163,
+      "loss_iou": 0.6103659272193909,
+      "loss_num": 0.018310546875,
+      "loss_xval": 1.3124982118606567,
+      "num_input_tokens_seen": 610032096,
+      "step": 3529
+    },
+    {
+      "epoch": 1.3519724243584834,
+      "grad_norm": 416.0503668947494,
+      "learning_rate": 5e-06,
+      "loss": 2.2382,
+      "num_input_tokens_seen": 610201696,
+      "step": 3530
+    },
+    {
+      "epoch": 1.3519724243584834,
+      "loss": 2.271575450897217,
+      "loss_ce": 0.20846495032310486,
+      "loss_iou": 0.9695436954498291,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 2.063110828399658,
+      "num_input_tokens_seen": 610201696,
+      "step": 3530
+    },
+    {
+      "epoch": 1.3523554193795482,
+      "grad_norm": 130.61666940566212,
+      "learning_rate": 5e-06,
+      "loss": 1.8439,
+      "num_input_tokens_seen": 610374552,
+      "step": 3531
+    },
+    {
+      "epoch": 1.3523554193795482,
+      "loss": 1.9753737449645996,
+      "loss_ce": 0.18173204362392426,
+      "loss_iou": 0.8015450239181519,
+      "loss_num": 0.0380859375,
+      "loss_xval": 1.7936418056488037,
+      "num_input_tokens_seen": 610374552,
+      "step": 3531
+    },
+    {
+      "epoch": 1.3527384144006127,
+      "grad_norm": 194.02855345479304,
+      "learning_rate": 5e-06,
+      "loss": 1.6476,
+      "num_input_tokens_seen": 610547976,
+      "step": 3532
+    },
+    {
+      "epoch": 1.3527384144006127,
+      "loss": 1.765662670135498,
+      "loss_ce": 0.24570322036743164,
+      "loss_iou": 0.7102513313293457,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 1.5199594497680664,
+      "num_input_tokens_seen": 610547976,
+      "step": 3532
+    },
+    {
+      "epoch": 1.3531214094216775,
+      "grad_norm": 325.080996320109,
+      "learning_rate": 5e-06,
+      "loss": 2.3593,
+      "num_input_tokens_seen": 610721000,
+      "step": 3533
+    },
+    {
+      "epoch": 1.3531214094216775,
+      "loss": 2.304840326309204,
+      "loss_ce": 0.21709582209587097,
+      "loss_iou": 0.9870485067367554,
+      "loss_num": 0.022705078125,
+      "loss_xval": 2.0877444744110107,
+      "num_input_tokens_seen": 610721000,
+      "step": 3533
+    },
+    {
+      "epoch": 1.3535044044427422,
+      "grad_norm": 83.10097665107112,
+      "learning_rate": 5e-06,
+      "loss": 1.6988,
+      "num_input_tokens_seen": 610893856,
+      "step": 3534
+    },
+    {
+      "epoch": 1.3535044044427422,
+      "loss": 1.6446635723114014,
+      "loss_ce": 0.15993928909301758,
+      "loss_iou": 0.6867896318435669,
+      "loss_num": 0.022216796875,
+      "loss_xval": 1.4847242832183838,
+      "num_input_tokens_seen": 610893856,
+      "step": 3534
+    },
+    {
+      "epoch": 1.353887399463807,
+      "grad_norm": 101.77705256899391,
+      "learning_rate": 5e-06,
+      "loss": 1.4157,
+      "num_input_tokens_seen": 611066984,
+      "step": 3535
+    },
+    {
+      "epoch": 1.353887399463807,
+      "loss": 1.5131068229675293,
+      "loss_ce": 0.1964910626411438,
+      "loss_iou": 0.6061533093452454,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 1.3166157007217407,
+      "num_input_tokens_seen": 611066984,
+      "step": 3535
+    },
+    {
+      "epoch": 1.3542703944848717,
+      "grad_norm": 204.98252957985457,
+      "learning_rate": 5e-06,
+      "loss": 1.644,
+      "num_input_tokens_seen": 611240120,
+      "step": 3536
+    },
+    {
+      "epoch": 1.3542703944848717,
+      "loss": 1.608548641204834,
+      "loss_ce": 0.22869566082954407,
+      "loss_iou": 0.6464694738388062,
+      "loss_num": 0.017333984375,
+      "loss_xval": 1.3798530101776123,
+      "num_input_tokens_seen": 611240120,
+      "step": 3536
+    },
+    {
+      "epoch": 1.3546533895059365,
+      "grad_norm": 109.11836657217839,
+      "learning_rate": 5e-06,
+      "loss": 1.7624,
+      "num_input_tokens_seen": 611412856,
+      "step": 3537
+    },
+    {
+      "epoch": 1.3546533895059365,
+      "loss": 1.5386110544204712,
+      "loss_ce": 0.19527839124202728,
+      "loss_iou": 0.6114704608917236,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 1.3433327674865723,
+      "num_input_tokens_seen": 611412856,
+      "step": 3537
+    },
+    {
+      "epoch": 1.355036384527001,
+      "grad_norm": 108.76088799857489,
+      "learning_rate": 5e-06,
+      "loss": 1.5263,
+      "num_input_tokens_seen": 611585592,
+      "step": 3538
+    },
+    {
+      "epoch": 1.355036384527001,
+      "loss": 1.7183423042297363,
+      "loss_ce": 0.28218644857406616,
+      "loss_iou": 0.6774896383285522,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 1.4361560344696045,
+      "num_input_tokens_seen": 611585592,
+      "step": 3538
+    },
+    {
+      "epoch": 1.3554193795480658,
+      "grad_norm": 103.06026257785365,
+      "learning_rate": 5e-06,
+      "loss": 1.5056,
+      "num_input_tokens_seen": 611758768,
+      "step": 3539
+    },
+    {
+      "epoch": 1.3554193795480658,
+      "loss": 1.5339665412902832,
+      "loss_ce": 0.19743913412094116,
+      "loss_iou": 0.6244862079620361,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 1.3365273475646973,
+      "num_input_tokens_seen": 611758768,
+      "step": 3539
+    },
+    {
+      "epoch": 1.3558023745691306,
+      "grad_norm": 117.67632754300494,
+      "learning_rate": 5e-06,
+      "loss": 1.6944,
+      "num_input_tokens_seen": 611928664,
+      "step": 3540
+    },
+    {
+      "epoch": 1.3558023745691306,
+      "loss": 1.6984690427780151,
+      "loss_ce": 0.22155040502548218,
+      "loss_iou": 0.6877696514129639,
+      "loss_num": 0.020263671875,
+      "loss_xval": 1.4769186973571777,
+      "num_input_tokens_seen": 611928664,
+      "step": 3540
+    },
+    {
+      "epoch": 1.3561853695901953,
+      "grad_norm": 86.88476948794022,
+      "learning_rate": 5e-06,
+      "loss": 1.3772,
+      "num_input_tokens_seen": 612101776,
+      "step": 3541
+    },
+    {
+      "epoch": 1.3561853695901953,
+      "loss": 1.4097684621810913,
+      "loss_ce": 0.1925264596939087,
+      "loss_iou": 0.5699399709701538,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 1.2172420024871826,
+      "num_input_tokens_seen": 612101776,
+      "step": 3541
+    },
+    {
+      "epoch": 1.35656836461126,
+      "grad_norm": 104.28539292758965,
+      "learning_rate": 5e-06,
+      "loss": 1.4779,
+      "num_input_tokens_seen": 612275072,
+      "step": 3542
+    },
+    {
+      "epoch": 1.35656836461126,
+      "loss": 1.4341868162155151,
+      "loss_ce": 0.18426361680030823,
+      "loss_iou": 0.5773999691009521,
+      "loss_num": 0.01904296875,
+      "loss_xval": 1.2499232292175293,
+      "num_input_tokens_seen": 612275072,
+      "step": 3542
+    },
+    {
+      "epoch": 1.3569513596323248,
+      "grad_norm": 137.64803241159876,
+      "learning_rate": 5e-06,
+      "loss": 1.6254,
+      "num_input_tokens_seen": 612447672,
+      "step": 3543
+    },
+    {
+      "epoch": 1.3569513596323248,
+      "loss": 1.5099307298660278,
+      "loss_ce": 0.15565082430839539,
+      "loss_iou": 0.6298681497573853,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 1.3542797565460205,
+      "num_input_tokens_seen": 612447672,
+      "step": 3543
+    },
+    {
+      "epoch": 1.3573343546533896,
+      "grad_norm": 203.08852558116658,
+      "learning_rate": 5e-06,
+      "loss": 1.8637,
+      "num_input_tokens_seen": 612616752,
+      "step": 3544
+    },
+    {
+      "epoch": 1.3573343546533896,
+      "loss": 1.918691635131836,
+      "loss_ce": 0.1849869042634964,
+      "loss_iou": 0.7682195901870728,
+      "loss_num": 0.03955078125,
+      "loss_xval": 1.7337048053741455,
+      "num_input_tokens_seen": 612616752,
+      "step": 3544
+    },
+    {
+      "epoch": 1.3577173496744543,
+      "grad_norm": 158.2249014581109,
+      "learning_rate": 5e-06,
+      "loss": 1.4388,
+      "num_input_tokens_seen": 612789736,
+      "step": 3545
+    },
+    {
+      "epoch": 1.3577173496744543,
+      "loss": 1.4531426429748535,
+      "loss_ce": 0.12079142779111862,
+      "loss_iou": 0.6085736751556396,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 1.3323512077331543,
+      "num_input_tokens_seen": 612789736,
+      "step": 3545
+    },
+    {
+      "epoch": 1.3581003446955189,
+      "grad_norm": 152.06662030528923,
+      "learning_rate": 5e-06,
+      "loss": 1.4821,
+      "num_input_tokens_seen": 612962848,
+      "step": 3546
+    },
+    {
+      "epoch": 1.3581003446955189,
+      "loss": 1.4804325103759766,
+      "loss_ce": 0.20803119242191315,
+      "loss_iou": 0.5983589291572571,
+      "loss_num": 0.01513671875,
+      "loss_xval": 1.2724014520645142,
+      "num_input_tokens_seen": 612962848,
+      "step": 3546
+    },
+    {
+      "epoch": 1.3584833397165836,
+      "grad_norm": 275.75166928757216,
+      "learning_rate": 5e-06,
+      "loss": 1.5002,
+      "num_input_tokens_seen": 613135504,
+      "step": 3547
+    },
+    {
+      "epoch": 1.3584833397165836,
+      "loss": 1.507258653640747,
+      "loss_ce": 0.19036146998405457,
+      "loss_iou": 0.6261304616928101,
+      "loss_num": 0.012939453125,
+      "loss_xval": 1.3168971538543701,
+      "num_input_tokens_seen": 613135504,
+      "step": 3547
+    },
+    {
+      "epoch": 1.3588663347376484,
+      "grad_norm": 123.4670584102338,
+      "learning_rate": 5e-06,
+      "loss": 1.8353,
+      "num_input_tokens_seen": 613308312,
+      "step": 3548
+    },
+    {
+      "epoch": 1.3588663347376484,
+      "loss": 1.8402364253997803,
+      "loss_ce": 0.21213197708129883,
+      "loss_iou": 0.7385822534561157,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 1.6281044483184814,
+      "num_input_tokens_seen": 613308312,
+      "step": 3548
+    },
+    {
+      "epoch": 1.3592493297587132,
+      "grad_norm": 166.29377812510188,
+      "learning_rate": 5e-06,
+      "loss": 1.646,
+      "num_input_tokens_seen": 613481424,
+      "step": 3549
+    },
+    {
+      "epoch": 1.3592493297587132,
+      "loss": 1.6446444988250732,
+      "loss_ce": 0.15899218618869781,
+      "loss_iou": 0.6780067682266235,
+      "loss_num": 0.02587890625,
+      "loss_xval": 1.485652208328247,
+      "num_input_tokens_seen": 613481424,
+      "step": 3549
+    },
+    {
+      "epoch": 1.359632324779778,
+      "grad_norm": 160.32427315708424,
+      "learning_rate": 5e-06,
+      "loss": 1.5357,
+      "num_input_tokens_seen": 613654432,
+      "step": 3550
+    },
+    {
+      "epoch": 1.359632324779778,
+      "loss": 1.657616138458252,
+      "loss_ce": 0.2261035144329071,
+      "loss_iou": 0.6774415969848633,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 1.4315128326416016,
+      "num_input_tokens_seen": 613654432,
+      "step": 3550
+    },
+    {
+      "epoch": 1.3600153198008427,
+      "grad_norm": 130.92150090516574,
+      "learning_rate": 5e-06,
+      "loss": 1.5814,
+      "num_input_tokens_seen": 613827304,
+      "step": 3551
+    },
+    {
+      "epoch": 1.3600153198008427,
+      "loss": 1.6296170949935913,
+      "loss_ce": 0.2307167500257492,
+      "loss_iou": 0.6356074213981628,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 1.3989003896713257,
+      "num_input_tokens_seen": 613827304,
+      "step": 3551
+    },
+    {
+      "epoch": 1.3603983148219072,
+      "grad_norm": 145.42628896473127,
+      "learning_rate": 5e-06,
+      "loss": 1.4659,
+      "num_input_tokens_seen": 614000248,
+      "step": 3552
+    },
+    {
+      "epoch": 1.3603983148219072,
+      "loss": 1.3170982599258423,
+      "loss_ce": 0.16705623269081116,
+      "loss_iou": 0.5328152179718018,
+      "loss_num": 0.016845703125,
+      "loss_xval": 1.1439690589904785,
+      "num_input_tokens_seen": 614000248,
+      "step": 3552
+    },
+    {
+      "epoch": 1.360781309842972,
+      "grad_norm": 121.05505468908999,
+      "learning_rate": 5e-06,
+      "loss": 1.6945,
+      "num_input_tokens_seen": 614173160,
+      "step": 3553
+    },
+    {
+      "epoch": 1.360781309842972,
+      "loss": 1.5743227005004883,
+      "loss_ce": 0.16809436678886414,
+      "loss_iou": 0.6442152857780457,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 1.4062284231185913,
+      "num_input_tokens_seen": 614173160,
+      "step": 3553
+    },
+    {
+      "epoch": 1.3611643048640367,
+      "grad_norm": 112.7108650207361,
+      "learning_rate": 5e-06,
+      "loss": 1.3712,
+      "num_input_tokens_seen": 614346064,
+      "step": 3554
+    },
+    {
+      "epoch": 1.3611643048640367,
+      "loss": 1.3476276397705078,
+      "loss_ce": 0.20763450860977173,
+      "loss_iou": 0.5259749293327332,
+      "loss_num": 0.017578125,
+      "loss_xval": 1.1399930715560913,
+      "num_input_tokens_seen": 614346064,
+      "step": 3554
+    },
+    {
+      "epoch": 1.3615472998851015,
+      "grad_norm": 330.86850954956566,
+      "learning_rate": 5e-06,
+      "loss": 1.7237,
+      "num_input_tokens_seen": 614518840,
+      "step": 3555
+    },
+    {
+      "epoch": 1.3615472998851015,
+      "loss": 1.6065523624420166,
+      "loss_ce": 0.16759878396987915,
+      "loss_iou": 0.6765080690383911,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 1.4389536380767822,
+      "num_input_tokens_seen": 614518840,
+      "step": 3555
+    },
+    {
+      "epoch": 1.3619302949061662,
+      "grad_norm": 142.18084009906164,
+      "learning_rate": 5e-06,
+      "loss": 1.7091,
+      "num_input_tokens_seen": 614691528,
+      "step": 3556
+    },
+    {
+      "epoch": 1.3619302949061662,
+      "loss": 1.593044400215149,
+      "loss_ce": 0.16852545738220215,
+      "loss_iou": 0.6348363757133484,
+      "loss_num": 0.031005859375,
+      "loss_xval": 1.4245189428329468,
+      "num_input_tokens_seen": 614691528,
+      "step": 3556
+    },
+    {
+      "epoch": 1.362313289927231,
+      "grad_norm": 200.13824810397716,
+      "learning_rate": 5e-06,
+      "loss": 1.6684,
+      "num_input_tokens_seen": 614864280,
+      "step": 3557
+    },
+    {
+      "epoch": 1.362313289927231,
+      "loss": 1.684021234512329,
+      "loss_ce": 0.19837597012519836,
+      "loss_iou": 0.6903018951416016,
+      "loss_num": 0.02099609375,
+      "loss_xval": 1.4856452941894531,
+      "num_input_tokens_seen": 614864280,
+      "step": 3557
+    },
+    {
+      "epoch": 1.3626962849482958,
+      "grad_norm": 136.52859958092625,
+      "learning_rate": 5e-06,
+      "loss": 2.1012,
+      "num_input_tokens_seen": 615037440,
+      "step": 3558
+    },
+    {
+      "epoch": 1.3626962849482958,
+      "loss": 2.1489267349243164,
+      "loss_ce": 0.19121010601520538,
+      "loss_iou": 0.885474443435669,
+      "loss_num": 0.037353515625,
+      "loss_xval": 1.957716464996338,
+      "num_input_tokens_seen": 615037440,
+      "step": 3558
+    },
+    {
+      "epoch": 1.3630792799693605,
+      "grad_norm": 95.58481875319109,
+      "learning_rate": 5e-06,
+      "loss": 1.6488,
+      "num_input_tokens_seen": 615209168,
+      "step": 3559
+    },
+    {
+      "epoch": 1.3630792799693605,
+      "loss": 1.5920864343643188,
+      "loss_ce": 0.19161845743656158,
+      "loss_iou": 0.6444936394691467,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 1.4004679918289185,
+      "num_input_tokens_seen": 615209168,
+      "step": 3559
+    },
+    {
+      "epoch": 1.363462274990425,
+      "grad_norm": 199.02495995648124,
+      "learning_rate": 5e-06,
+      "loss": 2.1887,
+      "num_input_tokens_seen": 615382208,
+      "step": 3560
+    },
+    {
+      "epoch": 1.363462274990425,
+      "loss": 2.1514439582824707,
+      "loss_ce": 0.19695159792900085,
+      "loss_iou": 0.898175060749054,
+      "loss_num": 0.03173828125,
+      "loss_xval": 1.954492211341858,
+      "num_input_tokens_seen": 615382208,
+      "step": 3560
+    },
+    {
+      "epoch": 1.3638452700114898,
+      "grad_norm": 112.0891784952754,
+      "learning_rate": 5e-06,
+      "loss": 1.705,
+      "num_input_tokens_seen": 615555544,
+      "step": 3561
+    },
+    {
+      "epoch": 1.3638452700114898,
+      "loss": 1.7380800247192383,
+      "loss_ce": 0.20713254809379578,
+      "loss_iou": 0.7163404226303101,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 1.5309474468231201,
+      "num_input_tokens_seen": 615555544,
+      "step": 3561
+    },
+    {
+      "epoch": 1.3642282650325546,
+      "grad_norm": 192.29869113246352,
+      "learning_rate": 5e-06,
+      "loss": 1.8479,
+      "num_input_tokens_seen": 615728920,
+      "step": 3562
+    },
+    {
+      "epoch": 1.3642282650325546,
+      "loss": 1.7795076370239258,
+      "loss_ce": 0.2079392671585083,
+      "loss_iou": 0.7436088919639587,
+      "loss_num": 0.016845703125,
+      "loss_xval": 1.5715683698654175,
+      "num_input_tokens_seen": 615728920,
+      "step": 3562
+    },
+    {
+      "epoch": 1.3646112600536193,
+      "grad_norm": 68.36400261160614,
+      "learning_rate": 5e-06,
+      "loss": 1.9486,
+      "num_input_tokens_seen": 615901720,
+      "step": 3563
+    },
+    {
+      "epoch": 1.3646112600536193,
+      "loss": 1.926328182220459,
+      "loss_ce": 0.20747782289981842,
+      "loss_iou": 0.7692152261734009,
+      "loss_num": 0.0361328125,
+      "loss_xval": 1.7188503742218018,
+      "num_input_tokens_seen": 615901720,
+      "step": 3563
+    },
+    {
+      "epoch": 1.364994255074684,
+      "grad_norm": 112.75874750994122,
+      "learning_rate": 5e-06,
+      "loss": 1.5633,
+      "num_input_tokens_seen": 616074560,
+      "step": 3564
+    },
+    {
+      "epoch": 1.364994255074684,
+      "loss": 1.5410892963409424,
+      "loss_ce": 0.1854608654975891,
+      "loss_iou": 0.627856969833374,
+      "loss_num": 0.02001953125,
+      "loss_xval": 1.355628490447998,
+      "num_input_tokens_seen": 616074560,
+      "step": 3564
+    },
+    {
+      "epoch": 1.3653772500957486,
+      "grad_norm": 175.84719838322135,
+      "learning_rate": 5e-06,
+      "loss": 2.109,
+      "num_input_tokens_seen": 616247752,
+      "step": 3565
+    },
+    {
+      "epoch": 1.3653772500957486,
+      "loss": 1.9987949132919312,
+      "loss_ce": 0.18364642560482025,
+      "loss_iou": 0.8309751152992249,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 1.8151484727859497,
+      "num_input_tokens_seen": 616247752,
+      "step": 3565
+    },
+    {
+      "epoch": 1.3657602451168134,
+      "grad_norm": 71.38045533359247,
+      "learning_rate": 5e-06,
+      "loss": 1.7527,
+      "num_input_tokens_seen": 616420520,
+      "step": 3566
+    },
+    {
+      "epoch": 1.3657602451168134,
+      "loss": 1.66278076171875,
+      "loss_ce": 0.19346889853477478,
+      "loss_iou": 0.6478638648986816,
+      "loss_num": 0.03466796875,
+      "loss_xval": 1.4693117141723633,
+      "num_input_tokens_seen": 616420520,
+      "step": 3566
+    },
+    {
+      "epoch": 1.3661432401378781,
+      "grad_norm": 67.2068048753262,
+      "learning_rate": 5e-06,
+      "loss": 1.6118,
+      "num_input_tokens_seen": 616593520,
+      "step": 3567
+    },
+    {
+      "epoch": 1.3661432401378781,
+      "loss": 1.6439144611358643,
+      "loss_ce": 0.29087749123573303,
+      "loss_iou": 0.6355944275856018,
+      "loss_num": 0.016357421875,
+      "loss_xval": 1.3530369997024536,
+      "num_input_tokens_seen": 616593520,
+      "step": 3567
+    },
+    {
+      "epoch": 1.366526235158943,
+      "grad_norm": 89.2511966558542,
+      "learning_rate": 5e-06,
+      "loss": 1.4504,
+      "num_input_tokens_seen": 616766360,
+      "step": 3568
+    },
+    {
+      "epoch": 1.366526235158943,
+      "loss": 1.5952684879302979,
+      "loss_ce": 0.2199249565601349,
+      "loss_iou": 0.6187782883644104,
+      "loss_num": 0.027587890625,
+      "loss_xval": 1.3753434419631958,
+      "num_input_tokens_seen": 616766360,
+      "step": 3568
+    },
+    {
+      "epoch": 1.3669092301800077,
+      "grad_norm": 164.71363153291628,
+      "learning_rate": 5e-06,
+      "loss": 2.0781,
+      "num_input_tokens_seen": 616939432,
+      "step": 3569
+    },
+    {
+      "epoch": 1.3669092301800077,
+      "loss": 2.082994222640991,
+      "loss_ce": 0.23589718341827393,
+      "loss_iou": 0.8618725538253784,
+      "loss_num": 0.024658203125,
+      "loss_xval": 1.8470971584320068,
+      "num_input_tokens_seen": 616939432,
+      "step": 3569
+    },
+    {
+      "epoch": 1.3672922252010724,
+      "grad_norm": 144.59929639130402,
+      "learning_rate": 5e-06,
+      "loss": 1.7726,
+      "num_input_tokens_seen": 617112248,
+      "step": 3570
+    },
+    {
+      "epoch": 1.3672922252010724,
+      "loss": 1.602302074432373,
+      "loss_ce": 0.18006178736686707,
+      "loss_iou": 0.6458430290222168,
+      "loss_num": 0.026123046875,
+      "loss_xval": 1.4081411361694336,
+      "num_input_tokens_seen": 617112248,
+      "step": 3570
+    },
+    {
+      "epoch": 1.3676752202221372,
+      "grad_norm": 141.1230383930963,
+      "learning_rate": 5e-06,
+      "loss": 1.5116,
+      "num_input_tokens_seen": 617285568,
+      "step": 3571
+    },
+    {
+      "epoch": 1.3676752202221372,
+      "loss": 1.4005942344665527,
+      "loss_ce": 0.19415241479873657,
+      "loss_iou": 0.5592755675315857,
+      "loss_num": 0.017578125,
+      "loss_xval": 1.2064417600631714,
+      "num_input_tokens_seen": 617285568,
+      "step": 3571
+    },
+    {
+      "epoch": 1.368058215243202,
+      "grad_norm": 84.54980833831,
+      "learning_rate": 5e-06,
+      "loss": 1.6988,
+      "num_input_tokens_seen": 617458336,
+      "step": 3572
+    },
+    {
+      "epoch": 1.368058215243202,
+      "loss": 1.5296988487243652,
+      "loss_ce": 0.1943495273590088,
+      "loss_iou": 0.614039957523346,
+      "loss_num": 0.021484375,
+      "loss_xval": 1.335349202156067,
+      "num_input_tokens_seen": 617458336,
+      "step": 3572
+    },
+    {
+      "epoch": 1.3684412102642667,
+      "grad_norm": 184.85652314277982,
+      "learning_rate": 5e-06,
+      "loss": 1.6501,
+      "num_input_tokens_seen": 617631352,
+      "step": 3573
+    },
+    {
+      "epoch": 1.3684412102642667,
+      "loss": 1.5000786781311035,
+      "loss_ce": 0.2294812798500061,
+      "loss_iou": 0.5807638168334961,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 1.2705974578857422,
+      "num_input_tokens_seen": 617631352,
+      "step": 3573
+    },
+    {
+      "epoch": 1.3688242052853312,
+      "grad_norm": 183.29486314610384,
+      "learning_rate": 5e-06,
+      "loss": 1.6727,
+      "num_input_tokens_seen": 617804456,
+      "step": 3574
+    },
+    {
+      "epoch": 1.3688242052853312,
+      "loss": 1.6086540222167969,
+      "loss_ce": 0.21161183714866638,
+      "loss_iou": 0.6385082602500916,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 1.397042155265808,
+      "num_input_tokens_seen": 617804456,
+      "step": 3574
+    },
+    {
+      "epoch": 1.369207200306396,
+      "grad_norm": 84.22326668577249,
+      "learning_rate": 5e-06,
+      "loss": 2.0519,
+      "num_input_tokens_seen": 617977784,
+      "step": 3575
+    },
+    {
+      "epoch": 1.369207200306396,
+      "loss": 2.0773351192474365,
+      "loss_ce": 0.21203234791755676,
+      "loss_iou": 0.8519018888473511,
+      "loss_num": 0.0322265625,
+      "loss_xval": 1.8653028011322021,
+      "num_input_tokens_seen": 617977784,
+      "step": 3575
+    },
+    {
+      "epoch": 1.3695901953274607,
+      "grad_norm": 71.3680249380758,
+      "learning_rate": 5e-06,
+      "loss": 1.4871,
+      "num_input_tokens_seen": 618150600,
+      "step": 3576
+    },
+    {
+      "epoch": 1.3695901953274607,
+      "loss": 1.4333336353302002,
+      "loss_ce": 0.16744062304496765,
+      "loss_iou": -3.073839740252618e+36,
+      "loss_num": 0.020751953125,
+      "loss_xval": -7.272897730801609e+28,
+      "num_input_tokens_seen": 618150600,
+      "step": 3576
+    },
+    {
+      "epoch": 1.3699731903485255,
+      "grad_norm": 86.93418845876175,
+      "learning_rate": 5e-06,
+      "loss": 1.3279,
+      "num_input_tokens_seen": 618323584,
+      "step": 3577
+    },
+    {
+      "epoch": 1.3699731903485255,
+      "loss": 1.43703031539917,
+      "loss_ce": 0.22806599736213684,
+      "loss_iou": 0.5516409873962402,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 1.2089643478393555,
+      "num_input_tokens_seen": 618323584,
+      "step": 3577
+    },
+    {
+      "epoch": 1.3703561853695903,
+      "grad_norm": 157.4833698628764,
+      "learning_rate": 5e-06,
+      "loss": 1.4033,
+      "num_input_tokens_seen": 618496408,
+      "step": 3578
+    },
+    {
+      "epoch": 1.3703561853695903,
+      "loss": 1.3804898262023926,
+      "loss_ce": 0.22576244175434113,
+      "loss_iou": 0.5269334316253662,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 1.1547274589538574,
+      "num_input_tokens_seen": 618496408,
+      "step": 3578
+    },
+    {
+      "epoch": 1.3707391803906548,
+      "grad_norm": 138.00020927299406,
+      "learning_rate": 5e-06,
+      "loss": 1.4502,
+      "num_input_tokens_seen": 618668840,
+      "step": 3579
+    },
+    {
+      "epoch": 1.3707391803906548,
+      "loss": 1.4024038314819336,
+      "loss_ce": 0.19461020827293396,
+      "loss_iou": 0.5448606014251709,
+      "loss_num": 0.023681640625,
+      "loss_xval": 1.2077937126159668,
+      "num_input_tokens_seen": 618668840,
+      "step": 3579
+    },
+    {
+      "epoch": 1.3711221754117195,
+      "grad_norm": 113.76663493518731,
+      "learning_rate": 5e-06,
+      "loss": 1.6558,
+      "num_input_tokens_seen": 618841784,
+      "step": 3580
+    },
+    {
+      "epoch": 1.3711221754117195,
+      "loss": 1.6215949058532715,
+      "loss_ce": 0.16539452970027924,
+      "loss_iou": 0.6719478368759155,
+      "loss_num": 0.0224609375,
+      "loss_xval": 1.456200361251831,
+      "num_input_tokens_seen": 618841784,
+      "step": 3580
+    },
+    {
+      "epoch": 1.3715051704327843,
+      "grad_norm": 93.92814455177887,
+      "learning_rate": 5e-06,
+      "loss": 1.3708,
+      "num_input_tokens_seen": 619014496,
+      "step": 3581
+    },
+    {
+      "epoch": 1.3715051704327843,
+      "loss": 1.3997132778167725,
+      "loss_ce": 0.16591432690620422,
+      "loss_iou": 0.5728321075439453,
+      "loss_num": 0.017578125,
+      "loss_xval": 1.2337989807128906,
+      "num_input_tokens_seen": 619014496,
+      "step": 3581
+    },
+    {
+      "epoch": 1.371888165453849,
+      "grad_norm": 146.42374708408562,
+      "learning_rate": 5e-06,
+      "loss": 1.6331,
+      "num_input_tokens_seen": 619187808,
+      "step": 3582
+    },
+    {
+      "epoch": 1.371888165453849,
+      "loss": 1.6726750135421753,
+      "loss_ce": 0.18158861994743347,
+      "loss_iou": 0.6885516047477722,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 1.4910863637924194,
+      "num_input_tokens_seen": 619187808,
+      "step": 3582
+    },
+    {
+      "epoch": 1.3722711604749138,
+      "grad_norm": 120.95380911160353,
+      "learning_rate": 5e-06,
+      "loss": 1.6599,
+      "num_input_tokens_seen": 619360944,
+      "step": 3583
+    },
+    {
+      "epoch": 1.3722711604749138,
+      "loss": 1.6582934856414795,
+      "loss_ce": 0.20701220631599426,
+      "loss_iou": 0.6840451955795288,
+      "loss_num": 0.0166015625,
+      "loss_xval": 1.4512813091278076,
+      "num_input_tokens_seen": 619360944,
+      "step": 3583
+    },
+    {
+      "epoch": 1.3726541554959786,
+      "grad_norm": 78.01497378817737,
+      "learning_rate": 5e-06,
+      "loss": 1.3103,
+      "num_input_tokens_seen": 619534224,
+      "step": 3584
+    },
+    {
+      "epoch": 1.3726541554959786,
+      "loss": 1.2011638879776,
+      "loss_ce": 0.19282567501068115,
+      "loss_iou": 0.4743228852748871,
+      "loss_num": 0.011962890625,
+      "loss_xval": 1.008338212966919,
+      "num_input_tokens_seen": 619534224,
+      "step": 3584
+    },
+    {
+      "epoch": 1.3730371505170433,
+      "grad_norm": 92.40419626817528,
+      "learning_rate": 5e-06,
+      "loss": 1.5359,
+      "num_input_tokens_seen": 619707352,
+      "step": 3585
+    },
+    {
+      "epoch": 1.3730371505170433,
+      "loss": 1.704606056213379,
+      "loss_ce": 0.21170318126678467,
+      "loss_iou": 0.7000647783279419,
+      "loss_num": 0.0185546875,
+      "loss_xval": 1.4929029941558838,
+      "num_input_tokens_seen": 619707352,
+      "step": 3585
+    },
+    {
+      "epoch": 1.373420145538108,
+      "grad_norm": 124.85981223625015,
+      "learning_rate": 5e-06,
+      "loss": 1.7413,
+      "num_input_tokens_seen": 619880480,
+      "step": 3586
+    },
+    {
+      "epoch": 1.373420145538108,
+      "loss": 1.7765295505523682,
+      "loss_ce": 0.24645628035068512,
+      "loss_iou": 0.7011939287185669,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 1.5300734043121338,
+      "num_input_tokens_seen": 619880480,
+      "step": 3586
+    },
+    {
+      "epoch": 1.3738031405591729,
+      "grad_norm": 138.66393423856508,
+      "learning_rate": 5e-06,
+      "loss": 1.3644,
+      "num_input_tokens_seen": 620053512,
+      "step": 3587
+    },
+    {
+      "epoch": 1.3738031405591729,
+      "loss": 1.4122133255004883,
+      "loss_ce": 0.14615964889526367,
+      "loss_iou": 0.5936439037322998,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 1.2660536766052246,
+      "num_input_tokens_seen": 620053512,
+      "step": 3587
+    },
+    {
+      "epoch": 1.3741861355802374,
+      "grad_norm": 153.9009960217307,
+      "learning_rate": 5e-06,
+      "loss": 1.4305,
+      "num_input_tokens_seen": 620226264,
+      "step": 3588
+    },
+    {
+      "epoch": 1.3741861355802374,
+      "loss": 1.2958606481552124,
+      "loss_ce": 0.13410046696662903,
+      "loss_iou": 0.5219964385032654,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 1.1617602109909058,
+      "num_input_tokens_seen": 620226264,
+      "step": 3588
+    },
+    {
+      "epoch": 1.3745691306013021,
+      "grad_norm": 188.06566009943361,
+      "learning_rate": 5e-06,
+      "loss": 1.3848,
+      "num_input_tokens_seen": 620398928,
+      "step": 3589
+    },
+    {
+      "epoch": 1.3745691306013021,
+      "loss": 1.2425801753997803,
+      "loss_ce": 0.1674756109714508,
+      "loss_iou": 0.5067752599716187,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 1.0751044750213623,
+      "num_input_tokens_seen": 620398928,
+      "step": 3589
+    },
+    {
+      "epoch": 1.374952125622367,
+      "grad_norm": 82.30347593001349,
+      "learning_rate": 5e-06,
+      "loss": 1.8051,
+      "num_input_tokens_seen": 620571976,
+      "step": 3590
+    },
+    {
+      "epoch": 1.374952125622367,
+      "loss": 1.7292916774749756,
+      "loss_ce": 0.20649453997612,
+      "loss_iou": 0.6882479190826416,
+      "loss_num": 0.029296875,
+      "loss_xval": 1.5227971076965332,
+      "num_input_tokens_seen": 620571976,
+      "step": 3590
+    },
+    {
+      "epoch": 1.3753351206434317,
+      "grad_norm": 101.65268624823504,
+      "learning_rate": 5e-06,
+      "loss": 1.3789,
+      "num_input_tokens_seen": 620744872,
+      "step": 3591
+    },
+    {
+      "epoch": 1.3753351206434317,
+      "loss": 1.3920843601226807,
+      "loss_ce": 0.2281237542629242,
+      "loss_iou": 0.5459238290786743,
+      "loss_num": 0.014404296875,
+      "loss_xval": 1.1639606952667236,
+      "num_input_tokens_seen": 620744872,
+      "step": 3591
+    },
+    {
+      "epoch": 1.3757181156644964,
+      "grad_norm": 177.78046060644215,
+      "learning_rate": 5e-06,
+      "loss": 1.5136,
+      "num_input_tokens_seen": 620918000,
+      "step": 3592
+    },
+    {
+      "epoch": 1.3757181156644964,
+      "loss": 1.5106201171875,
+      "loss_ce": 0.17325910925865173,
+      "loss_iou": 0.6299536824226379,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 1.3373609781265259,
+      "num_input_tokens_seen": 620918000,
+      "step": 3592
+    },
+    {
+      "epoch": 1.376101110685561,
+      "grad_norm": 109.71608368807934,
+      "learning_rate": 5e-06,
+      "loss": 1.7797,
+      "num_input_tokens_seen": 621091064,
+      "step": 3593
+    },
+    {
+      "epoch": 1.376101110685561,
+      "loss": 1.7327550649642944,
+      "loss_ce": 0.22606757283210754,
+      "loss_iou": 0.6914541721343994,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 1.5066876411437988,
+      "num_input_tokens_seen": 621091064,
+      "step": 3593
+    },
+    {
+      "epoch": 1.3764841057066257,
+      "grad_norm": 97.89136025517432,
+      "learning_rate": 5e-06,
+      "loss": 1.3367,
+      "num_input_tokens_seen": 621263880,
+      "step": 3594
+    },
+    {
+      "epoch": 1.3764841057066257,
+      "loss": 1.2630410194396973,
+      "loss_ce": 0.2281639277935028,
+      "loss_iou": 0.47805559635162354,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 1.034877061843872,
+      "num_input_tokens_seen": 621263880,
+      "step": 3594
+    },
+    {
+      "epoch": 1.3768671007276905,
+      "grad_norm": 113.75836262361007,
+      "learning_rate": 5e-06,
+      "loss": 1.3874,
+      "num_input_tokens_seen": 621437032,
+      "step": 3595
+    },
+    {
+      "epoch": 1.3768671007276905,
+      "loss": 1.394885778427124,
+      "loss_ce": 0.18021541833877563,
+      "loss_iou": 0.5659228563308716,
+      "loss_num": 0.0166015625,
+      "loss_xval": 1.2146704196929932,
+      "num_input_tokens_seen": 621437032,
+      "step": 3595
+    },
+    {
+      "epoch": 1.3772500957487552,
+      "grad_norm": 132.61150641515562,
+      "learning_rate": 5e-06,
+      "loss": 1.466,
+      "num_input_tokens_seen": 621610152,
+      "step": 3596
+    },
+    {
+      "epoch": 1.3772500957487552,
+      "loss": 1.277604579925537,
+      "loss_ce": 0.20018568634986877,
+      "loss_iou": 0.5098550319671631,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 1.0774188041687012,
+      "num_input_tokens_seen": 621610152,
+      "step": 3596
+    },
+    {
+      "epoch": 1.37763309076982,
+      "grad_norm": 197.29756768755266,
+      "learning_rate": 5e-06,
+      "loss": 1.5008,
+      "num_input_tokens_seen": 621782832,
+      "step": 3597
+    },
+    {
+      "epoch": 1.37763309076982,
+      "loss": 1.2304048538208008,
+      "loss_ce": 0.15923373401165009,
+      "loss_iou": -2.3966519160564962e+30,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 1.2850934724873698e+35,
+      "num_input_tokens_seen": 621782832,
+      "step": 3597
+    },
+    {
+      "epoch": 1.3780160857908847,
+      "grad_norm": 132.8124837495467,
+      "learning_rate": 5e-06,
+      "loss": 1.6714,
+      "num_input_tokens_seen": 621955864,
+      "step": 3598
+    },
+    {
+      "epoch": 1.3780160857908847,
+      "loss": 1.5206103324890137,
+      "loss_ce": 0.22293275594711304,
+      "loss_iou": 0.5956466197967529,
+      "loss_num": 0.021240234375,
+      "loss_xval": 1.2976775169372559,
+      "num_input_tokens_seen": 621955864,
+      "step": 3598
+    },
+    {
+      "epoch": 1.3783990808119495,
+      "grad_norm": 143.7155048499848,
+      "learning_rate": 5e-06,
+      "loss": 1.5459,
+      "num_input_tokens_seen": 622128792,
+      "step": 3599
+    },
+    {
+      "epoch": 1.3783990808119495,
+      "loss": 1.546707272529602,
+      "loss_ce": 0.17243452370166779,
+      "loss_iou": 0.6349208950996399,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 1.3742729425430298,
+      "num_input_tokens_seen": 622128792,
+      "step": 3599
+    },
+    {
+      "epoch": 1.3787820758330143,
+      "grad_norm": 201.95855457358925,
+      "learning_rate": 5e-06,
+      "loss": 1.5652,
+      "num_input_tokens_seen": 622301776,
+      "step": 3600
+    },
+    {
+      "epoch": 1.3787820758330143,
+      "loss": 1.778625726699829,
+      "loss_ce": 0.2336582988500595,
+      "loss_iou": 0.7346419095993042,
+      "loss_num": 0.01513671875,
+      "loss_xval": 1.5449674129486084,
+      "num_input_tokens_seen": 622301776,
+      "step": 3600
+    },
+    {
+      "epoch": 1.379165070854079,
+      "grad_norm": 374.5716709419254,
+      "learning_rate": 5e-06,
+      "loss": 2.083,
+      "num_input_tokens_seen": 622474968,
+      "step": 3601
+    },
+    {
+      "epoch": 1.379165070854079,
+      "loss": 2.073340892791748,
+      "loss_ce": 0.2615928053855896,
+      "loss_iou": 0.8359583020210266,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 1.8117482662200928,
+      "num_input_tokens_seen": 622474968,
+      "step": 3601
+    },
+    {
+      "epoch": 1.3795480658751436,
+      "grad_norm": 88.0574071209228,
+      "learning_rate": 5e-06,
+      "loss": 2.4811,
+      "num_input_tokens_seen": 622648152,
+      "step": 3602
+    },
+    {
+      "epoch": 1.3795480658751436,
+      "loss": 2.474989891052246,
+      "loss_ce": 0.217107355594635,
+      "loss_iou": 0.9219099283218384,
+      "loss_num": 0.0830078125,
+      "loss_xval": 2.2578823566436768,
+      "num_input_tokens_seen": 622648152,
+      "step": 3602
+    },
+    {
+      "epoch": 1.3799310608962083,
+      "grad_norm": 214.45984808367047,
+      "learning_rate": 5e-06,
+      "loss": 1.7667,
+      "num_input_tokens_seen": 622821168,
+      "step": 3603
+    },
+    {
+      "epoch": 1.3799310608962083,
+      "loss": 1.8158915042877197,
+      "loss_ce": 0.2158391922712326,
+      "loss_iou": 0.7445147037506104,
+      "loss_num": 0.022216796875,
+      "loss_xval": 1.6000523567199707,
+      "num_input_tokens_seen": 622821168,
+      "step": 3603
+    },
+    {
+      "epoch": 1.380314055917273,
+      "grad_norm": 292.4304309739641,
+      "learning_rate": 5e-06,
+      "loss": 2.5309,
+      "num_input_tokens_seen": 622994408,
+      "step": 3604
+    },
+    {
+      "epoch": 1.380314055917273,
+      "loss": 2.4345271587371826,
+      "loss_ce": 0.23869533836841583,
+      "loss_iou": 1.0452425479888916,
+      "loss_num": 0.02099609375,
+      "loss_xval": 2.195831775665283,
+      "num_input_tokens_seen": 622994408,
+      "step": 3604
+    },
+    {
+      "epoch": 1.3806970509383378,
+      "grad_norm": 140.3299423413433,
+      "learning_rate": 5e-06,
+      "loss": 2.1769,
+      "num_input_tokens_seen": 623167592,
+      "step": 3605
+    },
+    {
+      "epoch": 1.3806970509383378,
+      "loss": 2.1421940326690674,
+      "loss_ce": 0.19230958819389343,
+      "loss_iou": 0.832730233669281,
+      "loss_num": 0.056884765625,
+      "loss_xval": 1.949884295463562,
+      "num_input_tokens_seen": 623167592,
+      "step": 3605
+    },
+    {
+      "epoch": 1.3810800459594026,
+      "grad_norm": 90.03475829845245,
+      "learning_rate": 5e-06,
+      "loss": 1.4828,
+      "num_input_tokens_seen": 623340672,
+      "step": 3606
+    },
+    {
+      "epoch": 1.3810800459594026,
+      "loss": 1.4579837322235107,
+      "loss_ce": 0.19193831086158752,
+      "loss_iou": 0.5905880331993103,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 1.2660454511642456,
+      "num_input_tokens_seen": 623340672,
+      "step": 3606
+    },
+    {
+      "epoch": 1.3814630409804671,
+      "grad_norm": 255.06613070780614,
+      "learning_rate": 5e-06,
+      "loss": 1.6012,
+      "num_input_tokens_seen": 623510016,
+      "step": 3607
+    },
+    {
+      "epoch": 1.3814630409804671,
+      "loss": 1.664123296737671,
+      "loss_ce": 0.18670949339866638,
+      "loss_iou": 0.6796249151229858,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 1.4774138927459717,
+      "num_input_tokens_seen": 623510016,
+      "step": 3607
+    },
+    {
+      "epoch": 1.3818460360015319,
+      "grad_norm": 97.47083310193024,
+      "learning_rate": 5e-06,
+      "loss": 1.7039,
+      "num_input_tokens_seen": 623682848,
+      "step": 3608
+    },
+    {
+      "epoch": 1.3818460360015319,
+      "loss": 1.6879403591156006,
+      "loss_ce": 0.20535466074943542,
+      "loss_iou": 0.6692103147506714,
+      "loss_num": 0.02880859375,
+      "loss_xval": 1.4825856685638428,
+      "num_input_tokens_seen": 623682848,
+      "step": 3608
+    },
+    {
+      "epoch": 1.3822290310225966,
+      "grad_norm": 335.93712642421195,
+      "learning_rate": 5e-06,
+      "loss": 1.8219,
+      "num_input_tokens_seen": 623855784,
+      "step": 3609
+    },
+    {
+      "epoch": 1.3822290310225966,
+      "loss": 1.798883080482483,
+      "loss_ce": 0.21966424584388733,
+      "loss_iou": 0.722043514251709,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 1.579218864440918,
+      "num_input_tokens_seen": 623855784,
+      "step": 3609
+    },
+    {
+      "epoch": 1.3826120260436614,
+      "grad_norm": 115.76381225993897,
+      "learning_rate": 5e-06,
+      "loss": 1.9713,
+      "num_input_tokens_seen": 624028672,
+      "step": 3610
+    },
+    {
+      "epoch": 1.3826120260436614,
+      "loss": 2.0470962524414062,
+      "loss_ce": 0.20051613450050354,
+      "loss_iou": 0.8092153668403625,
+      "loss_num": 0.045654296875,
+      "loss_xval": 1.846580147743225,
+      "num_input_tokens_seen": 624028672,
+      "step": 3610
+    },
+    {
+      "epoch": 1.3829950210647262,
+      "grad_norm": 289.03065419006526,
+      "learning_rate": 5e-06,
+      "loss": 1.8144,
+      "num_input_tokens_seen": 624201688,
+      "step": 3611
+    },
+    {
+      "epoch": 1.3829950210647262,
+      "loss": 1.8520580530166626,
+      "loss_ce": 0.19172219932079315,
+      "loss_iou": 0.7532635927200317,
+      "loss_num": 0.03076171875,
+      "loss_xval": 1.6603357791900635,
+      "num_input_tokens_seen": 624201688,
+      "step": 3611
+    },
+    {
+      "epoch": 1.383378016085791,
+      "grad_norm": 126.01295749728209,
+      "learning_rate": 5e-06,
+      "loss": 1.7867,
+      "num_input_tokens_seen": 624374544,
+      "step": 3612
+    },
+    {
+      "epoch": 1.383378016085791,
+      "loss": 1.8940067291259766,
+      "loss_ce": 0.18726450204849243,
+      "loss_iou": 0.8059163093566895,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 1.706742286682129,
+      "num_input_tokens_seen": 624374544,
+      "step": 3612
+    },
+    {
+      "epoch": 1.3837610111068557,
+      "grad_norm": 106.44765104110519,
+      "learning_rate": 5e-06,
+      "loss": 1.782,
+      "num_input_tokens_seen": 624547768,
+      "step": 3613
+    },
+    {
+      "epoch": 1.3837610111068557,
+      "loss": 1.7128829956054688,
+      "loss_ce": 0.20912262797355652,
+      "loss_iou": 0.6836427450180054,
+      "loss_num": 0.02734375,
+      "loss_xval": 1.5037600994110107,
+      "num_input_tokens_seen": 624547768,
+      "step": 3613
+    },
+    {
+      "epoch": 1.3841440061279204,
+      "grad_norm": 145.1517137796239,
+      "learning_rate": 5e-06,
+      "loss": 1.6311,
+      "num_input_tokens_seen": 624720400,
+      "step": 3614
+    },
+    {
+      "epoch": 1.3841440061279204,
+      "loss": 1.650829553604126,
+      "loss_ce": 0.22188735008239746,
+      "loss_iou": 0.6540767550468445,
+      "loss_num": 0.024169921875,
+      "loss_xval": 1.428942084312439,
+      "num_input_tokens_seen": 624720400,
+      "step": 3614
+    },
+    {
+      "epoch": 1.3845270011489852,
+      "grad_norm": 146.12296407478735,
+      "learning_rate": 5e-06,
+      "loss": 1.8438,
+      "num_input_tokens_seen": 624893424,
+      "step": 3615
+    },
+    {
+      "epoch": 1.3845270011489852,
+      "loss": 1.6571210622787476,
+      "loss_ce": 0.22394540905952454,
+      "loss_iou": 0.6682174205780029,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 1.4331756830215454,
+      "num_input_tokens_seen": 624893424,
+      "step": 3615
+    },
+    {
+      "epoch": 1.3849099961700497,
+      "grad_norm": 69.83785985774774,
+      "learning_rate": 5e-06,
+      "loss": 1.35,
+      "num_input_tokens_seen": 625066432,
+      "step": 3616
+    },
+    {
+      "epoch": 1.3849099961700497,
+      "loss": 1.428694248199463,
+      "loss_ce": 0.2676292359828949,
+      "loss_iou": 0.5431942343711853,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 1.161064863204956,
+      "num_input_tokens_seen": 625066432,
+      "step": 3616
+    },
+    {
+      "epoch": 1.3852929911911145,
+      "grad_norm": 75.65348777664539,
+      "learning_rate": 5e-06,
+      "loss": 1.2476,
+      "num_input_tokens_seen": 625239336,
+      "step": 3617
+    },
+    {
+      "epoch": 1.3852929911911145,
+      "loss": 1.1371839046478271,
+      "loss_ce": 0.18536092340946198,
+      "loss_iou": 0.4348958730697632,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.9518229961395264,
+      "num_input_tokens_seen": 625239336,
+      "step": 3617
+    },
+    {
+      "epoch": 1.3856759862121792,
+      "grad_norm": 67.4584138420159,
+      "learning_rate": 5e-06,
+      "loss": 1.3155,
+      "num_input_tokens_seen": 625412304,
+      "step": 3618
+    },
+    {
+      "epoch": 1.3856759862121792,
+      "loss": 1.2084017992019653,
+      "loss_ce": 0.18417063355445862,
+      "loss_iou": 0.4751892685890198,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 1.0173646211624146,
+      "num_input_tokens_seen": 625412304,
+      "step": 3618
+    },
+    {
+      "epoch": 1.386058981233244,
+      "grad_norm": 151.97864710302989,
+      "learning_rate": 5e-06,
+      "loss": 1.5277,
+      "num_input_tokens_seen": 625585048,
+      "step": 3619
+    },
+    {
+      "epoch": 1.386058981233244,
+      "loss": 1.6334717273712158,
+      "loss_ce": 0.22242796421051025,
+      "loss_iou": 0.6616528630256653,
+      "loss_num": 0.017578125,
+      "loss_xval": 1.4110437631607056,
+      "num_input_tokens_seen": 625585048,
+      "step": 3619
+    },
+    {
+      "epoch": 1.3864419762543088,
+      "grad_norm": 114.49035974980248,
+      "learning_rate": 5e-06,
+      "loss": 1.7804,
+      "num_input_tokens_seen": 625758288,
+      "step": 3620
+    },
+    {
+      "epoch": 1.3864419762543088,
+      "loss": 1.7300214767456055,
+      "loss_ce": 0.19582955539226532,
+      "loss_iou": 0.6872619986534119,
+      "loss_num": 0.031982421875,
+      "loss_xval": 1.5341919660568237,
+      "num_input_tokens_seen": 625758288,
+      "step": 3620
+    },
+    {
+      "epoch": 1.3868249712753733,
+      "grad_norm": 97.59979423682461,
+      "learning_rate": 5e-06,
+      "loss": 1.4962,
+      "num_input_tokens_seen": 625931048,
+      "step": 3621
+    },
+    {
+      "epoch": 1.3868249712753733,
+      "loss": 1.4186863899230957,
+      "loss_ce": 0.19230541586875916,
+      "loss_iou": 0.5793540477752686,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 1.2263808250427246,
+      "num_input_tokens_seen": 625931048,
+      "step": 3621
+    },
+    {
+      "epoch": 1.387207966296438,
+      "grad_norm": 146.54098911248724,
+      "learning_rate": 5e-06,
+      "loss": 1.3584,
+      "num_input_tokens_seen": 626103680,
+      "step": 3622
+    },
+    {
+      "epoch": 1.387207966296438,
+      "loss": 1.3940868377685547,
+      "loss_ce": 0.18014013767242432,
+      "loss_iou": 0.5690094232559204,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 1.2139465808868408,
+      "num_input_tokens_seen": 626103680,
+      "step": 3622
+    },
+    {
+      "epoch": 1.3875909613175028,
+      "grad_norm": 120.98971428742999,
+      "learning_rate": 5e-06,
+      "loss": 1.4945,
+      "num_input_tokens_seen": 626276424,
+      "step": 3623
+    },
+    {
+      "epoch": 1.3875909613175028,
+      "loss": 1.5319600105285645,
+      "loss_ce": 0.2022242546081543,
+      "loss_iou": 0.6283536553382874,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 1.3297358751296997,
+      "num_input_tokens_seen": 626276424,
+      "step": 3623
+    },
+    {
+      "epoch": 1.3879739563385676,
+      "grad_norm": 97.84442574480609,
+      "learning_rate": 5e-06,
+      "loss": 1.5906,
+      "num_input_tokens_seen": 626449648,
+      "step": 3624
+    },
+    {
+      "epoch": 1.3879739563385676,
+      "loss": 1.5129203796386719,
+      "loss_ce": 0.21519820392131805,
+      "loss_iou": 0.6155816316604614,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 1.2977221012115479,
+      "num_input_tokens_seen": 626449648,
+      "step": 3624
+    },
+    {
+      "epoch": 1.3883569513596323,
+      "grad_norm": 152.8080708366177,
+      "learning_rate": 5e-06,
+      "loss": 1.6142,
+      "num_input_tokens_seen": 626622808,
+      "step": 3625
+    },
+    {
+      "epoch": 1.3883569513596323,
+      "loss": 1.673858880996704,
+      "loss_ce": 0.17594227194786072,
+      "loss_iou": 0.6999318599700928,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 1.4979166984558105,
+      "num_input_tokens_seen": 626622808,
+      "step": 3625
+    },
+    {
+      "epoch": 1.388739946380697,
+      "grad_norm": 149.94144582539803,
+      "learning_rate": 5e-06,
+      "loss": 1.4942,
+      "num_input_tokens_seen": 626795728,
+      "step": 3626
+    },
+    {
+      "epoch": 1.388739946380697,
+      "loss": 1.3925130367279053,
+      "loss_ce": 0.20805421471595764,
+      "loss_iou": 0.5429129600524902,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 1.1844587326049805,
+      "num_input_tokens_seen": 626795728,
+      "step": 3626
+    },
+    {
+      "epoch": 1.3891229414017618,
+      "grad_norm": 113.10008803350293,
+      "learning_rate": 5e-06,
+      "loss": 1.9643,
+      "num_input_tokens_seen": 626969064,
+      "step": 3627
+    },
+    {
+      "epoch": 1.3891229414017618,
+      "loss": 1.9852036237716675,
+      "loss_ce": 0.1441543996334076,
+      "loss_iou": 0.8232345581054688,
+      "loss_num": 0.0390625,
+      "loss_xval": 1.8410491943359375,
+      "num_input_tokens_seen": 626969064,
+      "step": 3627
+    },
+    {
+      "epoch": 1.3895059364228266,
+      "grad_norm": 76.91817878452832,
+      "learning_rate": 5e-06,
+      "loss": 1.4131,
+      "num_input_tokens_seen": 627141552,
+      "step": 3628
+    },
+    {
+      "epoch": 1.3895059364228266,
+      "loss": 1.406415343284607,
+      "loss_ce": 0.15866446495056152,
+      "loss_iou": 0.5848433971405029,
+      "loss_num": 0.015625,
+      "loss_xval": 1.2477507591247559,
+      "num_input_tokens_seen": 627141552,
+      "step": 3628
+    },
+    {
+      "epoch": 1.3898889314438914,
+      "grad_norm": 96.04041022337205,
+      "learning_rate": 5e-06,
+      "loss": 1.426,
+      "num_input_tokens_seen": 627314736,
+      "step": 3629
+    },
+    {
+      "epoch": 1.3898889314438914,
+      "loss": 1.3025336265563965,
+      "loss_ce": 0.19101479649543762,
+      "loss_iou": 0.5096931457519531,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 1.1115188598632812,
+      "num_input_tokens_seen": 627314736,
+      "step": 3629
+    },
+    {
+      "epoch": 1.390271926464956,
+      "grad_norm": 145.9791382708225,
+      "learning_rate": 5e-06,
+      "loss": 1.6838,
+      "num_input_tokens_seen": 627487544,
+      "step": 3630
+    },
+    {
+      "epoch": 1.390271926464956,
+      "loss": 1.740018606185913,
+      "loss_ce": 0.15928755700588226,
+      "loss_iou": 0.7386229634284973,
+      "loss_num": 0.020751953125,
+      "loss_xval": 1.5807310342788696,
+      "num_input_tokens_seen": 627487544,
+      "step": 3630
+    },
+    {
+      "epoch": 1.3906549214860207,
+      "grad_norm": 128.38685082067695,
+      "learning_rate": 5e-06,
+      "loss": 1.6129,
+      "num_input_tokens_seen": 627660392,
+      "step": 3631
+    },
+    {
+      "epoch": 1.3906549214860207,
+      "loss": 1.6399121284484863,
+      "loss_ce": 0.18036915361881256,
+      "loss_iou": 0.6751756072044373,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 1.4595431089401245,
+      "num_input_tokens_seen": 627660392,
+      "step": 3631
+    },
+    {
+      "epoch": 1.3910379165070854,
+      "grad_norm": 114.96278717059289,
+      "learning_rate": 5e-06,
+      "loss": 1.7292,
+      "num_input_tokens_seen": 627833672,
+      "step": 3632
+    },
+    {
+      "epoch": 1.3910379165070854,
+      "loss": 1.6817591190338135,
+      "loss_ce": 0.15832598507404327,
+      "loss_iou": 0.7013833522796631,
+      "loss_num": 0.024169921875,
+      "loss_xval": 1.5234332084655762,
+      "num_input_tokens_seen": 627833672,
+      "step": 3632
+    },
+    {
+      "epoch": 1.3914209115281502,
+      "grad_norm": 61.58322538749416,
+      "learning_rate": 5e-06,
+      "loss": 1.4345,
+      "num_input_tokens_seen": 628006720,
+      "step": 3633
+    },
+    {
+      "epoch": 1.3914209115281502,
+      "loss": 1.4161045551300049,
+      "loss_ce": 0.19610156118869781,
+      "loss_iou": 0.5703591108322144,
+      "loss_num": 0.015869140625,
+      "loss_xval": 1.2200028896331787,
+      "num_input_tokens_seen": 628006720,
+      "step": 3633
+    },
+    {
+      "epoch": 1.391803906549215,
+      "grad_norm": 115.21706585870986,
+      "learning_rate": 5e-06,
+      "loss": 1.3091,
+      "num_input_tokens_seen": 628179688,
+      "step": 3634
+    },
+    {
+      "epoch": 1.391803906549215,
+      "loss": 1.316110610961914,
+      "loss_ce": 0.19886082410812378,
+      "loss_iou": 0.5165106058120728,
+      "loss_num": 0.016845703125,
+      "loss_xval": 1.1172497272491455,
+      "num_input_tokens_seen": 628179688,
+      "step": 3634
+    },
+    {
+      "epoch": 1.3921869015702795,
+      "grad_norm": 149.91107409355843,
+      "learning_rate": 5e-06,
+      "loss": 1.6016,
+      "num_input_tokens_seen": 628352616,
+      "step": 3635
+    },
+    {
+      "epoch": 1.3921869015702795,
+      "loss": 1.632288932800293,
+      "loss_ce": 0.15338543057441711,
+      "loss_iou": 0.6840776205062866,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 1.4789035320281982,
+      "num_input_tokens_seen": 628352616,
+      "step": 3635
+    },
+    {
+      "epoch": 1.3925698965913442,
+      "grad_norm": 122.59793314083974,
+      "learning_rate": 5e-06,
+      "loss": 1.7816,
+      "num_input_tokens_seen": 628525488,
+      "step": 3636
+    },
+    {
+      "epoch": 1.3925698965913442,
+      "loss": 1.639886736869812,
+      "loss_ce": 0.1811191737651825,
+      "loss_iou": 0.6472305059432983,
+      "loss_num": 0.032958984375,
+      "loss_xval": 1.4587676525115967,
+      "num_input_tokens_seen": 628525488,
+      "step": 3636
+    },
+    {
+      "epoch": 1.392952891612409,
+      "grad_norm": 74.43094317206129,
+      "learning_rate": 5e-06,
+      "loss": 1.4292,
+      "num_input_tokens_seen": 628694824,
+      "step": 3637
+    },
+    {
+      "epoch": 1.392952891612409,
+      "loss": 1.1312984228134155,
+      "loss_ce": 0.17416593432426453,
+      "loss_iou": 0.4436388611793518,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.9571324586868286,
+      "num_input_tokens_seen": 628694824,
+      "step": 3637
+    },
+    {
+      "epoch": 1.3933358866334737,
+      "grad_norm": 127.56365057135433,
+      "learning_rate": 5e-06,
+      "loss": 1.7343,
+      "num_input_tokens_seen": 628867976,
+      "step": 3638
+    },
+    {
+      "epoch": 1.3933358866334737,
+      "loss": 1.6866893768310547,
+      "loss_ce": 0.1625714898109436,
+      "loss_iou": 0.7057845592498779,
+      "loss_num": 0.0224609375,
+      "loss_xval": 1.5241179466247559,
+      "num_input_tokens_seen": 628867976,
+      "step": 3638
+    },
+    {
+      "epoch": 1.3937188816545385,
+      "grad_norm": 94.22051678330423,
+      "learning_rate": 5e-06,
+      "loss": 1.9939,
+      "num_input_tokens_seen": 629041152,
+      "step": 3639
+    },
+    {
+      "epoch": 1.3937188816545385,
+      "loss": 1.8643584251403809,
+      "loss_ce": 0.2131306231021881,
+      "loss_iou": 0.7520970106124878,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 1.6512277126312256,
+      "num_input_tokens_seen": 629041152,
+      "step": 3639
+    },
+    {
+      "epoch": 1.3941018766756033,
+      "grad_norm": 165.30328183868085,
+      "learning_rate": 5e-06,
+      "loss": 1.412,
+      "num_input_tokens_seen": 629213752,
+      "step": 3640
+    },
+    {
+      "epoch": 1.3941018766756033,
+      "loss": 1.2959129810333252,
+      "loss_ce": 0.18002738058567047,
+      "loss_iou": 0.5097403526306152,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.1040449142456055,
+      "num_input_tokens_seen": 629213752,
+      "step": 3640
+    },
+    {
+      "epoch": 1.394484871696668,
+      "grad_norm": 117.91491116115839,
+      "learning_rate": 5e-06,
+      "loss": 1.7211,
+      "num_input_tokens_seen": 629387048,
+      "step": 3641
+    },
+    {
+      "epoch": 1.394484871696668,
+      "loss": 1.7913999557495117,
+      "loss_ce": 0.21221789717674255,
+      "loss_iou": 0.7316076159477234,
+      "loss_num": 0.023193359375,
+      "loss_xval": 1.5791820287704468,
+      "num_input_tokens_seen": 629387048,
+      "step": 3641
+    },
+    {
+      "epoch": 1.3948678667177328,
+      "grad_norm": 205.35782887366204,
+      "learning_rate": 5e-06,
+      "loss": 1.6886,
+      "num_input_tokens_seen": 629560248,
+      "step": 3642
+    },
+    {
+      "epoch": 1.3948678667177328,
+      "loss": 1.7776062488555908,
+      "loss_ce": 0.20876279473304749,
+      "loss_iou": 0.7054269909858704,
+      "loss_num": 0.031494140625,
+      "loss_xval": 1.5688434839248657,
+      "num_input_tokens_seen": 629560248,
+      "step": 3642
+    },
+    {
+      "epoch": 1.3952508617387975,
+      "grad_norm": 146.09051892094294,
+      "learning_rate": 5e-06,
+      "loss": 1.6681,
+      "num_input_tokens_seen": 629733592,
+      "step": 3643
+    },
+    {
+      "epoch": 1.3952508617387975,
+      "loss": 1.5654737949371338,
+      "loss_ce": 0.21972210705280304,
+      "loss_iou": 0.6193785071372986,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 1.3457516431808472,
+      "num_input_tokens_seen": 629733592,
+      "step": 3643
+    },
+    {
+      "epoch": 1.395633856759862,
+      "grad_norm": 111.08368485993188,
+      "learning_rate": 5e-06,
+      "loss": 1.5608,
+      "num_input_tokens_seen": 629906992,
+      "step": 3644
+    },
+    {
+      "epoch": 1.395633856759862,
+      "loss": 1.566825032234192,
+      "loss_ce": 0.1464221030473709,
+      "loss_iou": 0.6561853885650635,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 1.420403003692627,
+      "num_input_tokens_seen": 629906992,
+      "step": 3644
+    },
+    {
+      "epoch": 1.3960168517809268,
+      "grad_norm": 70.18813202929198,
+      "learning_rate": 5e-06,
+      "loss": 1.4565,
+      "num_input_tokens_seen": 630079776,
+      "step": 3645
+    },
+    {
+      "epoch": 1.3960168517809268,
+      "loss": 1.5499849319458008,
+      "loss_ce": 0.2004162073135376,
+      "loss_iou": 0.6262614130973816,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 1.3495687246322632,
+      "num_input_tokens_seen": 630079776,
+      "step": 3645
+    },
+    {
+      "epoch": 1.3963998468019916,
+      "grad_norm": 171.39538329360198,
+      "learning_rate": 5e-06,
+      "loss": 1.2866,
+      "num_input_tokens_seen": 630252768,
+      "step": 3646
+    },
+    {
+      "epoch": 1.3963998468019916,
+      "loss": 1.5590405464172363,
+      "loss_ce": 0.20193514227867126,
+      "loss_iou": 0.6220951676368713,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 1.3571053743362427,
+      "num_input_tokens_seen": 630252768,
+      "step": 3646
+    },
+    {
+      "epoch": 1.3967828418230563,
+      "grad_norm": 126.50631427677583,
+      "learning_rate": 5e-06,
+      "loss": 1.707,
+      "num_input_tokens_seen": 630425480,
+      "step": 3647
+    },
+    {
+      "epoch": 1.3967828418230563,
+      "loss": 1.6378086805343628,
+      "loss_ce": 0.17789973318576813,
+      "loss_iou": 0.6760299205780029,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 1.4599089622497559,
+      "num_input_tokens_seen": 630425480,
+      "step": 3647
+    },
+    {
+      "epoch": 1.397165836844121,
+      "grad_norm": 123.02433156337872,
+      "learning_rate": 5e-06,
+      "loss": 1.5238,
+      "num_input_tokens_seen": 630598432,
+      "step": 3648
+    },
+    {
+      "epoch": 1.397165836844121,
+      "loss": 1.542478084564209,
+      "loss_ce": 0.20391885936260223,
+      "loss_iou": 0.6159501075744629,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 1.3385591506958008,
+      "num_input_tokens_seen": 630598432,
+      "step": 3648
+    },
+    {
+      "epoch": 1.3975488318651856,
+      "grad_norm": 109.74008233158122,
+      "learning_rate": 5e-06,
+      "loss": 1.4739,
+      "num_input_tokens_seen": 630771424,
+      "step": 3649
+    },
+    {
+      "epoch": 1.3975488318651856,
+      "loss": 1.6322606801986694,
+      "loss_ce": 0.20045846700668335,
+      "loss_iou": 0.6789138317108154,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 1.4318022727966309,
+      "num_input_tokens_seen": 630771424,
+      "step": 3649
+    },
+    {
+      "epoch": 1.3979318268862504,
+      "grad_norm": 176.64068054854914,
+      "learning_rate": 5e-06,
+      "loss": 1.5302,
+      "num_input_tokens_seen": 630944536,
+      "step": 3650
+    },
+    {
+      "epoch": 1.3979318268862504,
+      "loss": 1.3865649700164795,
+      "loss_ce": 0.18322482705116272,
+      "loss_iou": 0.5652015209197998,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 1.2033400535583496,
+      "num_input_tokens_seen": 630944536,
+      "step": 3650
+    },
+    {
+      "epoch": 1.3983148219073152,
+      "grad_norm": 127.73682160398597,
+      "learning_rate": 5e-06,
+      "loss": 1.79,
+      "num_input_tokens_seen": 631117648,
+      "step": 3651
+    },
+    {
+      "epoch": 1.3983148219073152,
+      "loss": 1.6050747632980347,
+      "loss_ce": 0.15001869201660156,
+      "loss_iou": 0.6679272651672363,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 1.4550561904907227,
+      "num_input_tokens_seen": 631117648,
+      "step": 3651
+    },
+    {
+      "epoch": 1.39869781692838,
+      "grad_norm": 121.84412608839591,
+      "learning_rate": 5e-06,
+      "loss": 1.3806,
+      "num_input_tokens_seen": 631290560,
+      "step": 3652
+    },
+    {
+      "epoch": 1.39869781692838,
+      "loss": 1.3234338760375977,
+      "loss_ce": 0.16607999801635742,
+      "loss_iou": 0.5445430278778076,
+      "loss_num": 0.013671875,
+      "loss_xval": 1.1573538780212402,
+      "num_input_tokens_seen": 631290560,
+      "step": 3652
+    },
+    {
+      "epoch": 1.3990808119494447,
+      "grad_norm": 155.74749740528833,
+      "learning_rate": 5e-06,
+      "loss": 1.5854,
+      "num_input_tokens_seen": 631463608,
+      "step": 3653
+    },
+    {
+      "epoch": 1.3990808119494447,
+      "loss": 1.6742525100708008,
+      "loss_ce": 0.2013673633337021,
+      "loss_iou": 0.6678390502929688,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 1.4728851318359375,
+      "num_input_tokens_seen": 631463608,
+      "step": 3653
+    },
+    {
+      "epoch": 1.3994638069705094,
+      "grad_norm": 132.56573499627265,
+      "learning_rate": 5e-06,
+      "loss": 1.3746,
+      "num_input_tokens_seen": 631636760,
+      "step": 3654
+    },
+    {
+      "epoch": 1.3994638069705094,
+      "loss": 1.409779667854309,
+      "loss_ce": 0.23161810636520386,
+      "loss_iou": 0.5528411269187927,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 1.1781615018844604,
+      "num_input_tokens_seen": 631636760,
+      "step": 3654
+    },
+    {
+      "epoch": 1.3998468019915742,
+      "grad_norm": 228.82114588109818,
+      "learning_rate": 5e-06,
+      "loss": 1.7761,
+      "num_input_tokens_seen": 631809640,
+      "step": 3655
+    },
+    {
+      "epoch": 1.3998468019915742,
+      "loss": 1.7355780601501465,
+      "loss_ce": 0.16319149732589722,
+      "loss_iou": 0.7151482701301575,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 1.572386384010315,
+      "num_input_tokens_seen": 631809640,
+      "step": 3655
+    },
+    {
+      "epoch": 1.400229797012639,
+      "grad_norm": 149.41649831300995,
+      "learning_rate": 5e-06,
+      "loss": 1.5748,
+      "num_input_tokens_seen": 631982616,
+      "step": 3656
+    },
+    {
+      "epoch": 1.400229797012639,
+      "loss": 1.7118803262710571,
+      "loss_ce": 0.2118891477584839,
+      "loss_iou": 0.6864274740219116,
+      "loss_num": 0.025390625,
+      "loss_xval": 1.4999911785125732,
+      "num_input_tokens_seen": 631982616,
+      "step": 3656
+    },
+    {
+      "epoch": 1.4006127920337035,
+      "grad_norm": 186.10776374164269,
+      "learning_rate": 5e-06,
+      "loss": 1.814,
+      "num_input_tokens_seen": 632155624,
+      "step": 3657
+    },
+    {
+      "epoch": 1.4006127920337035,
+      "loss": 1.858457088470459,
+      "loss_ce": 0.21200542151927948,
+      "loss_iou": 0.7538898587226868,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 1.6464515924453735,
+      "num_input_tokens_seen": 632155624,
+      "step": 3657
+    },
+    {
+      "epoch": 1.4009957870547682,
+      "grad_norm": 133.89308676528069,
+      "learning_rate": 5e-06,
+      "loss": 2.081,
+      "num_input_tokens_seen": 632328096,
+      "step": 3658
+    },
+    {
+      "epoch": 1.4009957870547682,
+      "loss": 2.1265530586242676,
+      "loss_ce": 0.16438747942447662,
+      "loss_iou": 0.8673743009567261,
+      "loss_num": 0.04541015625,
+      "loss_xval": 1.9621655941009521,
+      "num_input_tokens_seen": 632328096,
+      "step": 3658
+    },
+    {
+      "epoch": 1.401378782075833,
+      "grad_norm": 117.00820295619212,
+      "learning_rate": 5e-06,
+      "loss": 1.6316,
+      "num_input_tokens_seen": 632501288,
+      "step": 3659
+    },
+    {
+      "epoch": 1.401378782075833,
+      "loss": 1.6250028610229492,
+      "loss_ce": 0.19171608984470367,
+      "loss_iou": 0.6683492660522461,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.4332866668701172,
+      "num_input_tokens_seen": 632501288,
+      "step": 3659
+    },
+    {
+      "epoch": 1.4017617770968978,
+      "grad_norm": 148.2435165289862,
+      "learning_rate": 5e-06,
+      "loss": 1.5913,
+      "num_input_tokens_seen": 632674584,
+      "step": 3660
+    },
+    {
+      "epoch": 1.4017617770968978,
+      "loss": 1.769160270690918,
+      "loss_ce": 0.2604833245277405,
+      "loss_iou": 0.7003681659698486,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 1.5086770057678223,
+      "num_input_tokens_seen": 632674584,
+      "step": 3660
+    },
+    {
+      "epoch": 1.4021447721179625,
+      "grad_norm": 102.0281381022801,
+      "learning_rate": 5e-06,
+      "loss": 1.7656,
+      "num_input_tokens_seen": 632847376,
+      "step": 3661
+    },
+    {
+      "epoch": 1.4021447721179625,
+      "loss": 1.818744421005249,
+      "loss_ce": 0.20584852993488312,
+      "loss_iou": 0.7327175140380859,
+      "loss_num": 0.029541015625,
+      "loss_xval": 1.6128959655761719,
+      "num_input_tokens_seen": 632847376,
+      "step": 3661
+    },
+    {
+      "epoch": 1.4025277671390273,
+      "grad_norm": 97.21557210103879,
+      "learning_rate": 5e-06,
+      "loss": 1.3706,
+      "num_input_tokens_seen": 633020272,
+      "step": 3662
+    },
+    {
+      "epoch": 1.4025277671390273,
+      "loss": 1.3934794664382935,
+      "loss_ce": 0.21816977858543396,
+      "loss_iou": 0.5200889110565186,
+      "loss_num": 0.027099609375,
+      "loss_xval": 1.175309658050537,
+      "num_input_tokens_seen": 633020272,
+      "step": 3662
+    },
+    {
+      "epoch": 1.4029107621600918,
+      "grad_norm": 134.3171957222966,
+      "learning_rate": 5e-06,
+      "loss": 1.5424,
+      "num_input_tokens_seen": 633193384,
+      "step": 3663
+    },
+    {
+      "epoch": 1.4029107621600918,
+      "loss": 1.334301233291626,
+      "loss_ce": 0.1728307008743286,
+      "loss_iou": 0.5449382066726685,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 1.161470651626587,
+      "num_input_tokens_seen": 633193384,
+      "step": 3663
+    },
+    {
+      "epoch": 1.4032937571811566,
+      "grad_norm": 105.19535281678613,
+      "learning_rate": 5e-06,
+      "loss": 1.6734,
+      "num_input_tokens_seen": 633366136,
+      "step": 3664
+    },
+    {
+      "epoch": 1.4032937571811566,
+      "loss": 1.624187707901001,
+      "loss_ce": 0.15212851762771606,
+      "loss_iou": 0.6565923690795898,
+      "loss_num": 0.03173828125,
+      "loss_xval": 1.4720592498779297,
+      "num_input_tokens_seen": 633366136,
+      "step": 3664
+    },
+    {
+      "epoch": 1.4036767522022213,
+      "grad_norm": 68.19636593146579,
+      "learning_rate": 5e-06,
+      "loss": 1.5386,
+      "num_input_tokens_seen": 633538880,
+      "step": 3665
+    },
+    {
+      "epoch": 1.4036767522022213,
+      "loss": 1.4059292078018188,
+      "loss_ce": 0.18063823878765106,
+      "loss_iou": 0.5717671513557434,
+      "loss_num": 0.016357421875,
+      "loss_xval": 1.2252908945083618,
+      "num_input_tokens_seen": 633538880,
+      "step": 3665
+    },
+    {
+      "epoch": 1.404059747223286,
+      "grad_norm": 141.4386747847512,
+      "learning_rate": 5e-06,
+      "loss": 1.4879,
+      "num_input_tokens_seen": 633712216,
+      "step": 3666
+    },
+    {
+      "epoch": 1.404059747223286,
+      "loss": 1.5945827960968018,
+      "loss_ce": 0.20796307921409607,
+      "loss_iou": 0.6225091218948364,
+      "loss_num": 0.0283203125,
+      "loss_xval": 1.3866198062896729,
+      "num_input_tokens_seen": 633712216,
+      "step": 3666
+    },
+    {
+      "epoch": 1.4044427422443508,
+      "grad_norm": 124.74239464839431,
+      "learning_rate": 5e-06,
+      "loss": 1.6612,
+      "num_input_tokens_seen": 633884912,
+      "step": 3667
+    },
+    {
+      "epoch": 1.4044427422443508,
+      "loss": 1.5374257564544678,
+      "loss_ce": 0.1949932724237442,
+      "loss_iou": 0.6300327181816101,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 1.3424323797225952,
+      "num_input_tokens_seen": 633884912,
+      "step": 3667
+    },
+    {
+      "epoch": 1.4048257372654156,
+      "grad_norm": 144.58184486992087,
+      "learning_rate": 5e-06,
+      "loss": 1.5791,
+      "num_input_tokens_seen": 634057744,
+      "step": 3668
+    },
+    {
+      "epoch": 1.4048257372654156,
+      "loss": 1.656828761100769,
+      "loss_ce": 0.1751592755317688,
+      "loss_iou": 0.69180828332901,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 1.481669545173645,
+      "num_input_tokens_seen": 634057744,
+      "step": 3668
+    },
+    {
+      "epoch": 1.4052087322864804,
+      "grad_norm": 108.73694816194813,
+      "learning_rate": 5e-06,
+      "loss": 1.4652,
+      "num_input_tokens_seen": 634230824,
+      "step": 3669
+    },
+    {
+      "epoch": 1.4052087322864804,
+      "loss": 1.6303739547729492,
+      "loss_ce": 0.20090125501155853,
+      "loss_iou": 0.6692651510238647,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 1.4294726848602295,
+      "num_input_tokens_seen": 634230824,
+      "step": 3669
+    },
+    {
+      "epoch": 1.4055917273075451,
+      "grad_norm": 116.11450724984425,
+      "learning_rate": 5e-06,
+      "loss": 1.6504,
+      "num_input_tokens_seen": 634403640,
+      "step": 3670
+    },
+    {
+      "epoch": 1.4055917273075451,
+      "loss": 1.6293880939483643,
+      "loss_ce": 0.17838090658187866,
+      "loss_iou": 0.6587921380996704,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 1.4510071277618408,
+      "num_input_tokens_seen": 634403640,
+      "step": 3670
+    },
+    {
+      "epoch": 1.4059747223286096,
+      "grad_norm": 76.9912662304742,
+      "learning_rate": 5e-06,
+      "loss": 1.4265,
+      "num_input_tokens_seen": 634576848,
+      "step": 3671
+    },
+    {
+      "epoch": 1.4059747223286096,
+      "loss": 1.4073896408081055,
+      "loss_ce": 0.157753124833107,
+      "loss_iou": 0.5690321326255798,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 1.2496365308761597,
+      "num_input_tokens_seen": 634576848,
+      "step": 3671
+    },
+    {
+      "epoch": 1.4063577173496744,
+      "grad_norm": 149.5367189108514,
+      "learning_rate": 5e-06,
+      "loss": 1.5006,
+      "num_input_tokens_seen": 634750160,
+      "step": 3672
+    },
+    {
+      "epoch": 1.4063577173496744,
+      "loss": 1.475905418395996,
+      "loss_ce": 0.2115403562784195,
+      "loss_iou": 0.6025347113609314,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 1.2643650770187378,
+      "num_input_tokens_seen": 634750160,
+      "step": 3672
+    },
+    {
+      "epoch": 1.4067407123707392,
+      "grad_norm": 122.68840219726172,
+      "learning_rate": 5e-06,
+      "loss": 1.647,
+      "num_input_tokens_seen": 634922744,
+      "step": 3673
+    },
+    {
+      "epoch": 1.4067407123707392,
+      "loss": 1.7613252401351929,
+      "loss_ce": 0.1777595579624176,
+      "loss_iou": 0.7400555610656738,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 1.5835657119750977,
+      "num_input_tokens_seen": 634922744,
+      "step": 3673
+    },
+    {
+      "epoch": 1.407123707391804,
+      "grad_norm": 141.85612757119978,
+      "learning_rate": 5e-06,
+      "loss": 1.7861,
+      "num_input_tokens_seen": 635095576,
+      "step": 3674
+    },
+    {
+      "epoch": 1.407123707391804,
+      "loss": 1.7490111589431763,
+      "loss_ce": 0.1951323002576828,
+      "loss_iou": 0.7126999497413635,
+      "loss_num": 0.025634765625,
+      "loss_xval": 1.553878903388977,
+      "num_input_tokens_seen": 635095576,
+      "step": 3674
+    },
+    {
+      "epoch": 1.4075067024128687,
+      "grad_norm": 117.45742225619787,
+      "learning_rate": 5e-06,
+      "loss": 1.3144,
+      "num_input_tokens_seen": 635268088,
+      "step": 3675
+    },
+    {
+      "epoch": 1.4075067024128687,
+      "loss": 1.2747701406478882,
+      "loss_ce": 0.15608471632003784,
+      "loss_iou": 0.47549569606781006,
+      "loss_num": 0.033447265625,
+      "loss_xval": 1.1186854839324951,
+      "num_input_tokens_seen": 635268088,
+      "step": 3675
+    },
+    {
+      "epoch": 1.4078896974339334,
+      "grad_norm": 127.83676373850088,
+      "learning_rate": 5e-06,
+      "loss": 1.6213,
+      "num_input_tokens_seen": 635441208,
+      "step": 3676
+    },
+    {
+      "epoch": 1.4078896974339334,
+      "loss": 1.6106691360473633,
+      "loss_ce": 0.1536344438791275,
+      "loss_iou": 0.6680924892425537,
+      "loss_num": 0.024169921875,
+      "loss_xval": 1.4570345878601074,
+      "num_input_tokens_seen": 635441208,
+      "step": 3676
+    },
+    {
+      "epoch": 1.408272692454998,
+      "grad_norm": 123.81896112475852,
+      "learning_rate": 5e-06,
+      "loss": 1.3013,
+      "num_input_tokens_seen": 635614088,
+      "step": 3677
+    },
+    {
+      "epoch": 1.408272692454998,
+      "loss": 1.3278595209121704,
+      "loss_ce": 0.2005331963300705,
+      "loss_iou": 0.5221745371818542,
+      "loss_num": 0.0166015625,
+      "loss_xval": 1.1273263692855835,
+      "num_input_tokens_seen": 635614088,
+      "step": 3677
+    },
+    {
+      "epoch": 1.4086556874760627,
+      "grad_norm": 150.54447805865803,
+      "learning_rate": 5e-06,
+      "loss": 1.6948,
+      "num_input_tokens_seen": 635787536,
+      "step": 3678
+    },
+    {
+      "epoch": 1.4086556874760627,
+      "loss": 1.6913161277770996,
+      "loss_ce": 0.2019982933998108,
+      "loss_iou": 0.6712336540222168,
+      "loss_num": 0.029296875,
+      "loss_xval": 1.4893178939819336,
+      "num_input_tokens_seen": 635787536,
+      "step": 3678
+    },
+    {
+      "epoch": 1.4090386824971275,
+      "grad_norm": 177.2569650026807,
+      "learning_rate": 5e-06,
+      "loss": 1.5617,
+      "num_input_tokens_seen": 635960624,
+      "step": 3679
+    },
+    {
+      "epoch": 1.4090386824971275,
+      "loss": 1.7486238479614258,
+      "loss_ce": 0.20002582669258118,
+      "loss_iou": 0.7126840353012085,
+      "loss_num": 0.024658203125,
+      "loss_xval": 1.548598051071167,
+      "num_input_tokens_seen": 635960624,
+      "step": 3679
+    },
+    {
+      "epoch": 1.4094216775181923,
+      "grad_norm": 91.34441728373196,
+      "learning_rate": 5e-06,
+      "loss": 1.7018,
+      "num_input_tokens_seen": 636133752,
+      "step": 3680
+    },
+    {
+      "epoch": 1.4094216775181923,
+      "loss": 1.7471097707748413,
+      "loss_ce": 0.2088487446308136,
+      "loss_iou": 0.7028005123138428,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 1.5382609367370605,
+      "num_input_tokens_seen": 636133752,
+      "step": 3680
+    },
+    {
+      "epoch": 1.409804672539257,
+      "grad_norm": 118.20382505311971,
+      "learning_rate": 5e-06,
+      "loss": 1.2404,
+      "num_input_tokens_seen": 636306808,
+      "step": 3681
+    },
+    {
+      "epoch": 1.409804672539257,
+      "loss": 1.1724213361740112,
+      "loss_ce": 0.23402735590934753,
+      "loss_iou": 0.42903590202331543,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.9383940696716309,
+      "num_input_tokens_seen": 636306808,
+      "step": 3681
+    },
+    {
+      "epoch": 1.4101876675603218,
+      "grad_norm": 155.07307351540132,
+      "learning_rate": 5e-06,
+      "loss": 1.5644,
+      "num_input_tokens_seen": 636479888,
+      "step": 3682
+    },
+    {
+      "epoch": 1.4101876675603218,
+      "loss": 1.5142014026641846,
+      "loss_ce": 0.16153010725975037,
+      "loss_iou": 0.6491597890853882,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 1.3526713848114014,
+      "num_input_tokens_seen": 636479888,
+      "step": 3682
+    },
+    {
+      "epoch": 1.4105706625813865,
+      "grad_norm": 106.23004908580938,
+      "learning_rate": 5e-06,
+      "loss": 1.7452,
+      "num_input_tokens_seen": 636653024,
+      "step": 3683
+    },
+    {
+      "epoch": 1.4105706625813865,
+      "loss": 1.6639301776885986,
+      "loss_ce": 0.15145039558410645,
+      "loss_iou": 0.6784200668334961,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 1.4862346649169922,
+      "num_input_tokens_seen": 636653024,
+      "step": 3683
+    },
+    {
+      "epoch": 1.4109536576024513,
+      "grad_norm": 92.66027418804897,
+      "learning_rate": 5e-06,
+      "loss": 1.2667,
+      "num_input_tokens_seen": 636825744,
+      "step": 3684
+    },
+    {
+      "epoch": 1.4109536576024513,
+      "loss": 1.0458866357803345,
+      "loss_ce": 0.15764087438583374,
+      "loss_iou": 0.420815110206604,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.8803722858428955,
+      "num_input_tokens_seen": 636825744,
+      "step": 3684
+    },
+    {
+      "epoch": 1.4113366526235158,
+      "grad_norm": 110.78237425208994,
+      "learning_rate": 5e-06,
+      "loss": 1.3655,
+      "num_input_tokens_seen": 636998592,
+      "step": 3685
+    },
+    {
+      "epoch": 1.4113366526235158,
+      "loss": 1.4087257385253906,
+      "loss_ce": 0.18637892603874207,
+      "loss_iou": 0.5656717419624329,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 1.2223469018936157,
+      "num_input_tokens_seen": 636998592,
+      "step": 3685
+    },
+    {
+      "epoch": 1.4117196476445806,
+      "grad_norm": 119.81532075716193,
+      "learning_rate": 5e-06,
+      "loss": 1.6419,
+      "num_input_tokens_seen": 637171880,
+      "step": 3686
+    },
+    {
+      "epoch": 1.4117196476445806,
+      "loss": 1.6677496433258057,
+      "loss_ce": 0.2320818305015564,
+      "loss_iou": 0.670623242855072,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 1.435667872428894,
+      "num_input_tokens_seen": 637171880,
+      "step": 3686
+    },
+    {
+      "epoch": 1.4121026426656453,
+      "grad_norm": 131.88755969852434,
+      "learning_rate": 5e-06,
+      "loss": 1.4355,
+      "num_input_tokens_seen": 637341968,
+      "step": 3687
+    },
+    {
+      "epoch": 1.4121026426656453,
+      "loss": 1.3841338157653809,
+      "loss_ce": 0.17219631373882294,
+      "loss_iou": 0.5581018924713135,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 1.211937427520752,
+      "num_input_tokens_seen": 637341968,
+      "step": 3687
+    },
+    {
+      "epoch": 1.41248563768671,
+      "grad_norm": 151.8059388176097,
+      "learning_rate": 5e-06,
+      "loss": 1.8072,
+      "num_input_tokens_seen": 637514528,
+      "step": 3688
+    },
+    {
+      "epoch": 1.41248563768671,
+      "loss": 1.9605388641357422,
+      "loss_ce": 0.19696488976478577,
+      "loss_iou": 0.8153807520866394,
+      "loss_num": 0.026611328125,
+      "loss_xval": 1.7635740041732788,
+      "num_input_tokens_seen": 637514528,
+      "step": 3688
+    },
+    {
+      "epoch": 1.4128686327077749,
+      "grad_norm": 135.8035768969732,
+      "learning_rate": 5e-06,
+      "loss": 1.6703,
+      "num_input_tokens_seen": 637687248,
+      "step": 3689
+    },
+    {
+      "epoch": 1.4128686327077749,
+      "loss": 1.5969858169555664,
+      "loss_ce": 0.19671300053596497,
+      "loss_iou": 0.6430990695953369,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 1.4002728462219238,
+      "num_input_tokens_seen": 637687248,
+      "step": 3689
+    },
+    {
+      "epoch": 1.4132516277288394,
+      "grad_norm": 120.6836359932711,
+      "learning_rate": 5e-06,
+      "loss": 1.5456,
+      "num_input_tokens_seen": 637860232,
+      "step": 3690
+    },
+    {
+      "epoch": 1.4132516277288394,
+      "loss": 1.6012014150619507,
+      "loss_ce": 0.20785996317863464,
+      "loss_iou": 0.6385194659233093,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 1.3933414220809937,
+      "num_input_tokens_seen": 637860232,
+      "step": 3690
+    },
+    {
+      "epoch": 1.4136346227499041,
+      "grad_norm": 92.45627459761134,
+      "learning_rate": 5e-06,
+      "loss": 1.5551,
+      "num_input_tokens_seen": 638033256,
+      "step": 3691
+    },
+    {
+      "epoch": 1.4136346227499041,
+      "loss": 1.7157249450683594,
+      "loss_ce": 0.22539356350898743,
+      "loss_iou": 0.6931637525558472,
+      "loss_num": 0.020751953125,
+      "loss_xval": 1.4903314113616943,
+      "num_input_tokens_seen": 638033256,
+      "step": 3691
+    },
+    {
+      "epoch": 1.414017617770969,
+      "grad_norm": 136.57183184362495,
+      "learning_rate": 5e-06,
+      "loss": 1.2977,
+      "num_input_tokens_seen": 638205808,
+      "step": 3692
+    },
+    {
+      "epoch": 1.414017617770969,
+      "loss": 1.3344258069992065,
+      "loss_ce": 0.18167954683303833,
+      "loss_iou": 0.5222960114479065,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 1.152746319770813,
+      "num_input_tokens_seen": 638205808,
+      "step": 3692
+    },
+    {
+      "epoch": 1.4144006127920337,
+      "grad_norm": 136.52495627908385,
+      "learning_rate": 5e-06,
+      "loss": 1.9508,
+      "num_input_tokens_seen": 638378784,
+      "step": 3693
+    },
+    {
+      "epoch": 1.4144006127920337,
+      "loss": 1.865358829498291,
+      "loss_ce": 0.16259488463401794,
+      "loss_iou": 0.7511622309684753,
+      "loss_num": 0.0400390625,
+      "loss_xval": 1.7027639150619507,
+      "num_input_tokens_seen": 638378784,
+      "step": 3693
+    },
+    {
+      "epoch": 1.4147836078130984,
+      "grad_norm": 238.08665224260275,
+      "learning_rate": 5e-06,
+      "loss": 1.4559,
+      "num_input_tokens_seen": 638551760,
+      "step": 3694
+    },
+    {
+      "epoch": 1.4147836078130984,
+      "loss": 1.5375279188156128,
+      "loss_ce": 0.1788027137517929,
+      "loss_iou": 0.6223556995391846,
+      "loss_num": 0.022705078125,
+      "loss_xval": 1.3587250709533691,
+      "num_input_tokens_seen": 638551760,
+      "step": 3694
+    },
+    {
+      "epoch": 1.4151666028341632,
+      "grad_norm": 86.71693668470905,
+      "learning_rate": 5e-06,
+      "loss": 1.897,
+      "num_input_tokens_seen": 638724952,
+      "step": 3695
+    },
+    {
+      "epoch": 1.4151666028341632,
+      "loss": 1.8192293643951416,
+      "loss_ce": 0.20848700404167175,
+      "loss_iou": 0.7126893401145935,
+      "loss_num": 0.037109375,
+      "loss_xval": 1.6107425689697266,
+      "num_input_tokens_seen": 638724952,
+      "step": 3695
+    },
+    {
+      "epoch": 1.415549597855228,
+      "grad_norm": 138.74805554876042,
+      "learning_rate": 5e-06,
+      "loss": 1.3908,
+      "num_input_tokens_seen": 638897848,
+      "step": 3696
+    },
+    {
+      "epoch": 1.415549597855228,
+      "loss": 1.4511733055114746,
+      "loss_ce": 0.1650530993938446,
+      "loss_iou": 0.5951322317123413,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 1.2861201763153076,
+      "num_input_tokens_seen": 638897848,
+      "step": 3696
+    },
+    {
+      "epoch": 1.4159325928762927,
+      "grad_norm": 306.11785917938215,
+      "learning_rate": 5e-06,
+      "loss": 1.7482,
+      "num_input_tokens_seen": 639071096,
+      "step": 3697
+    },
+    {
+      "epoch": 1.4159325928762927,
+      "loss": 1.8727173805236816,
+      "loss_ce": 0.21603775024414062,
+      "loss_iou": 0.7770397663116455,
+      "loss_num": 0.0205078125,
+      "loss_xval": 1.656679630279541,
+      "num_input_tokens_seen": 639071096,
+      "step": 3697
+    },
+    {
+      "epoch": 1.4163155878973575,
+      "grad_norm": 85.81341137112987,
+      "learning_rate": 5e-06,
+      "loss": 1.8047,
+      "num_input_tokens_seen": 639243728,
+      "step": 3698
+    },
+    {
+      "epoch": 1.4163155878973575,
+      "loss": 1.8856124877929688,
+      "loss_ce": 0.2053176760673523,
+      "loss_iou": 0.7709640264511108,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 1.6802948713302612,
+      "num_input_tokens_seen": 639243728,
+      "step": 3698
+    },
+    {
+      "epoch": 1.416698582918422,
+      "grad_norm": 336.9428656679885,
+      "learning_rate": 5e-06,
+      "loss": 2.2322,
+      "num_input_tokens_seen": 639416896,
+      "step": 3699
+    },
+    {
+      "epoch": 1.416698582918422,
+      "loss": 2.191256523132324,
+      "loss_ce": 0.19940683245658875,
+      "loss_iou": 0.9435871839523315,
+      "loss_num": 0.02099609375,
+      "loss_xval": 1.991849660873413,
+      "num_input_tokens_seen": 639416896,
+      "step": 3699
+    },
+    {
+      "epoch": 1.4170815779394867,
+      "grad_norm": 92.35111266977904,
+      "learning_rate": 5e-06,
+      "loss": 1.8368,
+      "num_input_tokens_seen": 639589640,
+      "step": 3700
+    },
+    {
+      "epoch": 1.4170815779394867,
+      "loss": 1.8230584859848022,
+      "loss_ce": 0.16814501583576202,
+      "loss_iou": 0.7557098865509033,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 1.6549134254455566,
+      "num_input_tokens_seen": 639589640,
+      "step": 3700
+    },
+    {
+      "epoch": 1.4174645729605515,
+      "grad_norm": 327.4133991835956,
+      "learning_rate": 5e-06,
+      "loss": 1.6466,
+      "num_input_tokens_seen": 639762656,
+      "step": 3701
+    },
+    {
+      "epoch": 1.4174645729605515,
+      "loss": 1.5542529821395874,
+      "loss_ce": 0.13617312908172607,
+      "loss_iou": 0.6641790866851807,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 1.4180798530578613,
+      "num_input_tokens_seen": 639762656,
+      "step": 3701
+    },
+    {
+      "epoch": 1.4178475679816163,
+      "grad_norm": 62.16172672217173,
+      "learning_rate": 5e-06,
+      "loss": 1.957,
+      "num_input_tokens_seen": 639931984,
+      "step": 3702
+    },
+    {
+      "epoch": 1.4178475679816163,
+      "loss": 1.9768989086151123,
+      "loss_ce": 0.1790628433227539,
+      "loss_iou": 0.7930219769477844,
+      "loss_num": 0.04248046875,
+      "loss_xval": 1.7978359460830688,
+      "num_input_tokens_seen": 639931984,
+      "step": 3702
+    },
+    {
+      "epoch": 1.418230563002681,
+      "grad_norm": 250.41063902161702,
+      "learning_rate": 5e-06,
+      "loss": 1.9669,
+      "num_input_tokens_seen": 640105168,
+      "step": 3703
+    },
+    {
+      "epoch": 1.418230563002681,
+      "loss": 2.0117201805114746,
+      "loss_ce": 0.22301751375198364,
+      "loss_iou": 0.8056672215461731,
+      "loss_num": 0.035400390625,
+      "loss_xval": 1.7887026071548462,
+      "num_input_tokens_seen": 640105168,
+      "step": 3703
+    },
+    {
+      "epoch": 1.4186135580237456,
+      "grad_norm": 163.9289972514936,
+      "learning_rate": 5e-06,
+      "loss": 1.7896,
+      "num_input_tokens_seen": 640278344,
+      "step": 3704
+    },
+    {
+      "epoch": 1.4186135580237456,
+      "loss": 1.970999002456665,
+      "loss_ce": 0.2278166115283966,
+      "loss_iou": 0.8001800775527954,
+      "loss_num": 0.028564453125,
+      "loss_xval": 1.7431824207305908,
+      "num_input_tokens_seen": 640278344,
+      "step": 3704
+    },
+    {
+      "epoch": 1.4189965530448103,
+      "grad_norm": 249.6520301998006,
+      "learning_rate": 5e-06,
+      "loss": 2.0686,
+      "num_input_tokens_seen": 640451592,
+      "step": 3705
+    },
+    {
+      "epoch": 1.4189965530448103,
+      "loss": 2.0172922611236572,
+      "loss_ce": 0.19945912063121796,
+      "loss_iou": 0.8343011140823364,
+      "loss_num": 0.02978515625,
+      "loss_xval": 1.8178331851959229,
+      "num_input_tokens_seen": 640451592,
+      "step": 3705
+    },
+    {
+      "epoch": 1.419379548065875,
+      "grad_norm": 67.82720968672345,
+      "learning_rate": 5e-06,
+      "loss": 1.8445,
+      "num_input_tokens_seen": 640624408,
+      "step": 3706
+    },
+    {
+      "epoch": 1.419379548065875,
+      "loss": 1.8409152030944824,
+      "loss_ce": 0.15923503041267395,
+      "loss_iou": 0.7513015270233154,
+      "loss_num": 0.035888671875,
+      "loss_xval": 1.6816802024841309,
+      "num_input_tokens_seen": 640624408,
+      "step": 3706
+    },
+    {
+      "epoch": 1.4197625430869398,
+      "grad_norm": 132.1660312975049,
+      "learning_rate": 5e-06,
+      "loss": 1.4361,
+      "num_input_tokens_seen": 640797160,
+      "step": 3707
+    },
+    {
+      "epoch": 1.4197625430869398,
+      "loss": 1.3635417222976685,
+      "loss_ce": 0.19868165254592896,
+      "loss_iou": 0.5315419435501099,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 1.1648600101470947,
+      "num_input_tokens_seen": 640797160,
+      "step": 3707
+    },
+    {
+      "epoch": 1.4201455381080046,
+      "grad_norm": 159.00627526237912,
+      "learning_rate": 5e-06,
+      "loss": 1.7623,
+      "num_input_tokens_seen": 640970472,
+      "step": 3708
+    },
+    {
+      "epoch": 1.4201455381080046,
+      "loss": 1.839440941810608,
+      "loss_ce": 0.23471230268478394,
+      "loss_iou": 0.7542381286621094,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.6047286987304688,
+      "num_input_tokens_seen": 640970472,
+      "step": 3708
+    },
+    {
+      "epoch": 1.4205285331290693,
+      "grad_norm": 91.5233965072586,
+      "learning_rate": 5e-06,
+      "loss": 1.5655,
+      "num_input_tokens_seen": 641143488,
+      "step": 3709
+    },
+    {
+      "epoch": 1.4205285331290693,
+      "loss": 1.4377386569976807,
+      "loss_ce": 0.18732315301895142,
+      "loss_iou": 0.5748385190963745,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 1.250415563583374,
+      "num_input_tokens_seen": 641143488,
+      "step": 3709
+    },
+    {
+      "epoch": 1.420911528150134,
+      "grad_norm": 110.57612094579471,
+      "learning_rate": 5e-06,
+      "loss": 1.4309,
+      "num_input_tokens_seen": 641316656,
+      "step": 3710
+    },
+    {
+      "epoch": 1.420911528150134,
+      "loss": 1.3285162448883057,
+      "loss_ce": 0.1683727502822876,
+      "loss_iou": 0.5430691838264465,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 1.160143494606018,
+      "num_input_tokens_seen": 641316656,
+      "step": 3710
+    },
+    {
+      "epoch": 1.4212945231711989,
+      "grad_norm": 212.41504148692457,
+      "learning_rate": 5e-06,
+      "loss": 1.6513,
+      "num_input_tokens_seen": 641489936,
+      "step": 3711
+    },
+    {
+      "epoch": 1.4212945231711989,
+      "loss": 1.6178189516067505,
+      "loss_ce": 0.1876608431339264,
+      "loss_iou": 0.6751620769500732,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 1.4301581382751465,
+      "num_input_tokens_seen": 641489936,
+      "step": 3711
+    },
+    {
+      "epoch": 1.4216775181922636,
+      "grad_norm": 103.55674639460399,
+      "learning_rate": 5e-06,
+      "loss": 1.7225,
+      "num_input_tokens_seen": 641662936,
+      "step": 3712
+    },
+    {
+      "epoch": 1.4216775181922636,
+      "loss": 1.9159207344055176,
+      "loss_ce": 0.21713170409202576,
+      "loss_iou": 0.7882983684539795,
+      "loss_num": 0.0244140625,
+      "loss_xval": 1.698789119720459,
+      "num_input_tokens_seen": 641662936,
+      "step": 3712
+    },
+    {
+      "epoch": 1.4220605132133282,
+      "grad_norm": 83.6160439494949,
+      "learning_rate": 5e-06,
+      "loss": 1.5368,
+      "num_input_tokens_seen": 641835768,
+      "step": 3713
+    },
+    {
+      "epoch": 1.4220605132133282,
+      "loss": 1.5422358512878418,
+      "loss_ce": 0.19803684949874878,
+      "loss_iou": 0.6211351156234741,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 1.3441989421844482,
+      "num_input_tokens_seen": 641835768,
+      "step": 3713
+    },
+    {
+      "epoch": 1.422443508234393,
+      "grad_norm": 156.9246201501115,
+      "learning_rate": 5e-06,
+      "loss": 1.6804,
+      "num_input_tokens_seen": 642008592,
+      "step": 3714
+    },
+    {
+      "epoch": 1.422443508234393,
+      "loss": 1.8472943305969238,
+      "loss_ce": 0.16469363868236542,
+      "loss_iou": 0.770102858543396,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 1.682600736618042,
+      "num_input_tokens_seen": 642008592,
+      "step": 3714
+    },
+    {
+      "epoch": 1.4228265032554577,
+      "grad_norm": 91.33352353280192,
+      "learning_rate": 5e-06,
+      "loss": 1.774,
+      "num_input_tokens_seen": 642181552,
+      "step": 3715
+    },
+    {
+      "epoch": 1.4228265032554577,
+      "loss": 1.6739634275436401,
+      "loss_ce": 0.17115110158920288,
+      "loss_iou": 0.6929649710655212,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 1.4878586530685425,
+      "num_input_tokens_seen": 642181552,
+      "step": 3715
+    },
+    {
+      "epoch": 1.4232094982765224,
+      "grad_norm": 91.89252200885332,
+      "learning_rate": 5e-06,
+      "loss": 1.4894,
+      "num_input_tokens_seen": 642354568,
+      "step": 3716
+    },
+    {
+      "epoch": 1.4232094982765224,
+      "loss": 1.5184659957885742,
+      "loss_ce": 0.1681097000837326,
+      "loss_iou": 0.6297527551651001,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 1.3503563404083252,
+      "num_input_tokens_seen": 642354568,
+      "step": 3716
+    },
+    {
+      "epoch": 1.4235924932975872,
+      "grad_norm": 185.32181837068592,
+      "learning_rate": 5e-06,
+      "loss": 1.4635,
+      "num_input_tokens_seen": 642527512,
+      "step": 3717
+    },
+    {
+      "epoch": 1.4235924932975872,
+      "loss": 1.4110028743743896,
+      "loss_ce": 0.15500202775001526,
+      "loss_iou": 0.5823765993118286,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 1.2560007572174072,
+      "num_input_tokens_seen": 642527512,
+      "step": 3717
+    },
+    {
+      "epoch": 1.4239754883186517,
+      "grad_norm": 95.47791705566677,
+      "learning_rate": 5e-06,
+      "loss": 1.7641,
+      "num_input_tokens_seen": 642700576,
+      "step": 3718
+    },
+    {
+      "epoch": 1.4239754883186517,
+      "loss": 1.8137187957763672,
+      "loss_ce": 0.17136678099632263,
+      "loss_iou": 0.7560819387435913,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 1.6423518657684326,
+      "num_input_tokens_seen": 642700576,
+      "step": 3718
+    },
+    {
+      "epoch": 1.4243584833397165,
+      "grad_norm": 130.32200948376027,
+      "learning_rate": 5e-06,
+      "loss": 1.5196,
+      "num_input_tokens_seen": 642873912,
+      "step": 3719
+    },
+    {
+      "epoch": 1.4243584833397165,
+      "loss": 1.578095555305481,
+      "loss_ce": 0.22507821023464203,
+      "loss_iou": 0.6102397441864014,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 1.3530173301696777,
+      "num_input_tokens_seen": 642873912,
+      "step": 3719
+    },
+    {
+      "epoch": 1.4247414783607812,
+      "grad_norm": 164.4305841901078,
+      "learning_rate": 5e-06,
+      "loss": 1.7802,
+      "num_input_tokens_seen": 643047464,
+      "step": 3720
+    },
+    {
+      "epoch": 1.4247414783607812,
+      "loss": 1.7335370779037476,
+      "loss_ce": 0.18880623579025269,
+      "loss_iou": 0.7317770719528198,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 1.5447309017181396,
+      "num_input_tokens_seen": 643047464,
+      "step": 3720
+    },
+    {
+      "epoch": 1.425124473381846,
+      "grad_norm": 111.37483587714223,
+      "learning_rate": 5e-06,
+      "loss": 1.8079,
+      "num_input_tokens_seen": 643220608,
+      "step": 3721
+    },
+    {
+      "epoch": 1.425124473381846,
+      "loss": 1.8084741830825806,
+      "loss_ce": 0.16545546054840088,
+      "loss_iou": 0.7520818710327148,
+      "loss_num": 0.02783203125,
+      "loss_xval": 1.6430187225341797,
+      "num_input_tokens_seen": 643220608,
+      "step": 3721
+    },
+    {
+      "epoch": 1.4255074684029108,
+      "grad_norm": 70.5761506236639,
+      "learning_rate": 5e-06,
+      "loss": 1.3081,
+      "num_input_tokens_seen": 643393384,
+      "step": 3722
+    },
+    {
+      "epoch": 1.4255074684029108,
+      "loss": 1.4663512706756592,
+      "loss_ce": 0.1774357110261917,
+      "loss_iou": 0.5970029234886169,
+      "loss_num": 0.01904296875,
+      "loss_xval": 1.2889155149459839,
+      "num_input_tokens_seen": 643393384,
+      "step": 3722
+    },
+    {
+      "epoch": 1.4258904634239755,
+      "grad_norm": 174.3773016576876,
+      "learning_rate": 5e-06,
+      "loss": 1.3919,
+      "num_input_tokens_seen": 643565864,
+      "step": 3723
+    },
+    {
+      "epoch": 1.4258904634239755,
+      "loss": 1.5935523509979248,
+      "loss_ce": 0.12954799830913544,
+      "loss_iou": 0.6833571195602417,
+      "loss_num": 0.01953125,
+      "loss_xval": 1.4640042781829834,
+      "num_input_tokens_seen": 643565864,
+      "step": 3723
+    },
+    {
+      "epoch": 1.4262734584450403,
+      "grad_norm": 157.3914023402223,
+      "learning_rate": 5e-06,
+      "loss": 1.7692,
+      "num_input_tokens_seen": 643738968,
+      "step": 3724
+    },
+    {
+      "epoch": 1.4262734584450403,
+      "loss": 1.7137362957000732,
+      "loss_ce": 0.19263222813606262,
+      "loss_iou": 0.6818166971206665,
+      "loss_num": 0.031494140625,
+      "loss_xval": 1.521104097366333,
+      "num_input_tokens_seen": 643738968,
+      "step": 3724
+    },
+    {
+      "epoch": 1.426656453466105,
+      "grad_norm": 103.36857082377544,
+      "learning_rate": 5e-06,
+      "loss": 1.3669,
+      "num_input_tokens_seen": 643912112,
+      "step": 3725
+    },
+    {
+      "epoch": 1.426656453466105,
+      "loss": 1.345698356628418,
+      "loss_ce": 0.18965527415275574,
+      "loss_iou": 0.5394168496131897,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 1.1560431718826294,
+      "num_input_tokens_seen": 643912112,
+      "step": 3725
+    },
+    {
+      "epoch": 1.4270394484871698,
+      "grad_norm": 93.56588767405124,
+      "learning_rate": 5e-06,
+      "loss": 1.4712,
+      "num_input_tokens_seen": 644085240,
+      "step": 3726
+    },
+    {
+      "epoch": 1.4270394484871698,
+      "loss": 1.3277511596679688,
+      "loss_ce": 0.16545361280441284,
+      "loss_iou": 0.5361658334732056,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 1.1622974872589111,
+      "num_input_tokens_seen": 644085240,
+      "step": 3726
+    },
+    {
+      "epoch": 1.4274224435082343,
+      "grad_norm": 112.12544501076535,
+      "learning_rate": 5e-06,
+      "loss": 1.6564,
+      "num_input_tokens_seen": 644254872,
+      "step": 3727
+    },
+    {
+      "epoch": 1.4274224435082343,
+      "loss": 1.8178613185882568,
+      "loss_ce": 0.20174267888069153,
+      "loss_iou": 0.7438503503799438,
+      "loss_num": 0.025634765625,
+      "loss_xval": 1.6161186695098877,
+      "num_input_tokens_seen": 644254872,
+      "step": 3727
+    },
+    {
+      "epoch": 1.427805438529299,
+      "grad_norm": 73.91467844184959,
+      "learning_rate": 5e-06,
+      "loss": 1.205,
+      "num_input_tokens_seen": 644427824,
+      "step": 3728
+    },
+    {
+      "epoch": 1.427805438529299,
+      "loss": 1.2962121963500977,
+      "loss_ce": 0.24076081812381744,
+      "loss_iou": 0.4939274191856384,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 1.0554512739181519,
+      "num_input_tokens_seen": 644427824,
+      "step": 3728
+    },
+    {
+      "epoch": 1.4281884335503638,
+      "grad_norm": 152.54727985440198,
+      "learning_rate": 5e-06,
+      "loss": 1.7261,
+      "num_input_tokens_seen": 644601480,
+      "step": 3729
+    },
+    {
+      "epoch": 1.4281884335503638,
+      "loss": 1.8054468631744385,
+      "loss_ce": 0.1608993113040924,
+      "loss_iou": 0.7548604607582092,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 1.6445475816726685,
+      "num_input_tokens_seen": 644601480,
+      "step": 3729
+    },
+    {
+      "epoch": 1.4285714285714286,
+      "grad_norm": 82.0129507976678,
+      "learning_rate": 5e-06,
+      "loss": 2.032,
+      "num_input_tokens_seen": 644774432,
+      "step": 3730
+    },
+    {
+      "epoch": 1.4285714285714286,
+      "loss": 1.9827988147735596,
+      "loss_ce": 0.18624532222747803,
+      "loss_iou": 0.8012003898620605,
+      "loss_num": 0.038818359375,
+      "loss_xval": 1.796553611755371,
+      "num_input_tokens_seen": 644774432,
+      "step": 3730
+    },
+    {
+      "epoch": 1.4289544235924934,
+      "grad_norm": 152.8632448487534,
+      "learning_rate": 5e-06,
+      "loss": 1.5189,
+      "num_input_tokens_seen": 644947120,
+      "step": 3731
+    },
+    {
+      "epoch": 1.4289544235924934,
+      "loss": 1.4444881677627563,
+      "loss_ce": 0.16636133193969727,
+      "loss_iou": 0.5974984765052795,
+      "loss_num": 0.0166015625,
+      "loss_xval": 1.278126835823059,
+      "num_input_tokens_seen": 644947120,
+      "step": 3731
+    },
+    {
+      "epoch": 1.429337418613558,
+      "grad_norm": 283.4599789297149,
+      "learning_rate": 5e-06,
+      "loss": 2.0318,
+      "num_input_tokens_seen": 645120224,
+      "step": 3732
+    },
+    {
+      "epoch": 1.429337418613558,
+      "loss": 1.9308521747589111,
+      "loss_ce": 0.21116694808006287,
+      "loss_iou": 0.8041785359382629,
+      "loss_num": 0.022216796875,
+      "loss_xval": 1.7196851968765259,
+      "num_input_tokens_seen": 645120224,
+      "step": 3732
+    },
+    {
+      "epoch": 1.4297204136346227,
+      "grad_norm": 130.5335208854067,
+      "learning_rate": 5e-06,
+      "loss": 2.0281,
+      "num_input_tokens_seen": 645293208,
+      "step": 3733
+    },
+    {
+      "epoch": 1.4297204136346227,
+      "loss": 1.900571584701538,
+      "loss_ce": 0.21438062191009521,
+      "loss_iou": 0.7400986552238464,
+      "loss_num": 0.041259765625,
+      "loss_xval": 1.6861909627914429,
+      "num_input_tokens_seen": 645293208,
+      "step": 3733
+    },
+    {
+      "epoch": 1.4301034086556874,
+      "grad_norm": 77.59650536748586,
+      "learning_rate": 5e-06,
+      "loss": 1.7141,
+      "num_input_tokens_seen": 645466184,
+      "step": 3734
+    },
+    {
+      "epoch": 1.4301034086556874,
+      "loss": 1.670695185661316,
+      "loss_ce": 0.18912170827388763,
+      "loss_iou": 0.6798431277275085,
+      "loss_num": 0.0244140625,
+      "loss_xval": 1.481573462486267,
+      "num_input_tokens_seen": 645466184,
+      "step": 3734
+    },
+    {
+      "epoch": 1.4304864036767522,
+      "grad_norm": 210.19312482574938,
+      "learning_rate": 5e-06,
+      "loss": 1.9514,
+      "num_input_tokens_seen": 645639128,
+      "step": 3735
+    },
+    {
+      "epoch": 1.4304864036767522,
+      "loss": 2.1336963176727295,
+      "loss_ce": 0.22373518347740173,
+      "loss_iou": 0.8884216547012329,
+      "loss_num": 0.026611328125,
+      "loss_xval": 1.9099611043930054,
+      "num_input_tokens_seen": 645639128,
+      "step": 3735
+    },
+    {
+      "epoch": 1.430869398697817,
+      "grad_norm": 97.08987338223963,
+      "learning_rate": 5e-06,
+      "loss": 1.7825,
+      "num_input_tokens_seen": 645811568,
+      "step": 3736
+    },
+    {
+      "epoch": 1.430869398697817,
+      "loss": 1.7851035594940186,
+      "loss_ce": 0.21304568648338318,
+      "loss_iou": 0.7253295183181763,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 1.5720579624176025,
+      "num_input_tokens_seen": 645811568,
+      "step": 3736
+    },
+    {
+      "epoch": 1.4312523937188817,
+      "grad_norm": 268.82543845535594,
+      "learning_rate": 5e-06,
+      "loss": 1.4205,
+      "num_input_tokens_seen": 645984384,
+      "step": 3737
+    },
+    {
+      "epoch": 1.4312523937188817,
+      "loss": 1.3121609687805176,
+      "loss_ce": 0.18860487639904022,
+      "loss_iou": 0.5270795822143555,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 1.123556137084961,
+      "num_input_tokens_seen": 645984384,
+      "step": 3737
+    },
+    {
+      "epoch": 1.4316353887399464,
+      "grad_norm": 83.84882696903573,
+      "learning_rate": 5e-06,
+      "loss": 1.8783,
+      "num_input_tokens_seen": 646157008,
+      "step": 3738
+    },
+    {
+      "epoch": 1.4316353887399464,
+      "loss": 1.8196080923080444,
+      "loss_ce": 0.17939066886901855,
+      "loss_iou": 0.7516272664070129,
+      "loss_num": 0.02734375,
+      "loss_xval": 1.6402174234390259,
+      "num_input_tokens_seen": 646157008,
+      "step": 3738
+    },
+    {
+      "epoch": 1.4320183837610112,
+      "grad_norm": 190.26463881770857,
+      "learning_rate": 5e-06,
+      "loss": 1.6885,
+      "num_input_tokens_seen": 646329528,
+      "step": 3739
+    },
+    {
+      "epoch": 1.4320183837610112,
+      "loss": 1.6450743675231934,
+      "loss_ce": 0.23552067577838898,
+      "loss_iou": 0.6517372727394104,
+      "loss_num": 0.021240234375,
+      "loss_xval": 1.4095536470413208,
+      "num_input_tokens_seen": 646329528,
+      "step": 3739
+    },
+    {
+      "epoch": 1.432401378782076,
+      "grad_norm": 221.4220954318138,
+      "learning_rate": 5e-06,
+      "loss": 1.5916,
+      "num_input_tokens_seen": 646502176,
+      "step": 3740
+    },
+    {
+      "epoch": 1.432401378782076,
+      "loss": 1.6941814422607422,
+      "loss_ce": 0.1842448115348816,
+      "loss_iou": 0.7121521234512329,
+      "loss_num": 0.01708984375,
+      "loss_xval": 1.5099365711212158,
+      "num_input_tokens_seen": 646502176,
+      "step": 3740
+    },
+    {
+      "epoch": 1.4327843738031405,
+      "grad_norm": 97.99613947599644,
+      "learning_rate": 5e-06,
+      "loss": 1.9946,
+      "num_input_tokens_seen": 646675432,
+      "step": 3741
+    },
+    {
+      "epoch": 1.4327843738031405,
+      "loss": 2.0199780464172363,
+      "loss_ce": 0.22191393375396729,
+      "loss_iou": 0.8097376227378845,
+      "loss_num": 0.03564453125,
+      "loss_xval": 1.798064112663269,
+      "num_input_tokens_seen": 646675432,
+      "step": 3741
+    },
+    {
+      "epoch": 1.4331673688242053,
+      "grad_norm": 132.07464574854623,
+      "learning_rate": 5e-06,
+      "loss": 1.3352,
+      "num_input_tokens_seen": 646848008,
+      "step": 3742
+    },
+    {
+      "epoch": 1.4331673688242053,
+      "loss": 1.2995100021362305,
+      "loss_ce": 0.19524481892585754,
+      "loss_iou": 0.5238122940063477,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 1.1042652130126953,
+      "num_input_tokens_seen": 646848008,
+      "step": 3742
+    },
+    {
+      "epoch": 1.43355036384527,
+      "grad_norm": 245.3325410425919,
+      "learning_rate": 5e-06,
+      "loss": 1.9307,
+      "num_input_tokens_seen": 647020960,
+      "step": 3743
+    },
+    {
+      "epoch": 1.43355036384527,
+      "loss": 1.9817793369293213,
+      "loss_ce": 0.20634689927101135,
+      "loss_iou": 0.8121546506881714,
+      "loss_num": 0.0302734375,
+      "loss_xval": 1.7754323482513428,
+      "num_input_tokens_seen": 647020960,
+      "step": 3743
+    },
+    {
+      "epoch": 1.4339333588663348,
+      "grad_norm": 166.79214130718145,
+      "learning_rate": 5e-06,
+      "loss": 1.7984,
+      "num_input_tokens_seen": 647194368,
+      "step": 3744
+    },
+    {
+      "epoch": 1.4339333588663348,
+      "loss": 1.826289415359497,
+      "loss_ce": 0.20446836948394775,
+      "loss_iou": 0.7626163959503174,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.6218209266662598,
+      "num_input_tokens_seen": 647194368,
+      "step": 3744
+    },
+    {
+      "epoch": 1.4343163538873995,
+      "grad_norm": 101.34890471679762,
+      "learning_rate": 5e-06,
+      "loss": 1.6824,
+      "num_input_tokens_seen": 647367128,
+      "step": 3745
+    },
+    {
+      "epoch": 1.4343163538873995,
+      "loss": 1.6728715896606445,
+      "loss_ce": 0.14931800961494446,
+      "loss_iou": 0.6829499006271362,
+      "loss_num": 0.031494140625,
+      "loss_xval": 1.5235536098480225,
+      "num_input_tokens_seen": 647367128,
+      "step": 3745
+    },
+    {
+      "epoch": 1.434699348908464,
+      "grad_norm": 101.27894614002867,
+      "learning_rate": 5e-06,
+      "loss": 1.4336,
+      "num_input_tokens_seen": 647540472,
+      "step": 3746
+    },
+    {
+      "epoch": 1.434699348908464,
+      "loss": 1.4333503246307373,
+      "loss_ce": 0.21338771283626556,
+      "loss_iou": 0.5676991939544678,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 1.2199625968933105,
+      "num_input_tokens_seen": 647540472,
+      "step": 3746
+    },
+    {
+      "epoch": 1.4350823439295288,
+      "grad_norm": 135.69922529206178,
+      "learning_rate": 5e-06,
+      "loss": 1.8911,
+      "num_input_tokens_seen": 647713472,
+      "step": 3747
+    },
+    {
+      "epoch": 1.4350823439295288,
+      "loss": 1.9443538188934326,
+      "loss_ce": 0.22142541408538818,
+      "loss_iou": 0.7996660470962524,
+      "loss_num": 0.024658203125,
+      "loss_xval": 1.7229282855987549,
+      "num_input_tokens_seen": 647713472,
+      "step": 3747
+    },
+    {
+      "epoch": 1.4354653389505936,
+      "grad_norm": 101.93708093168205,
+      "learning_rate": 5e-06,
+      "loss": 1.7479,
+      "num_input_tokens_seen": 647886560,
+      "step": 3748
+    },
+    {
+      "epoch": 1.4354653389505936,
+      "loss": 1.609716773033142,
+      "loss_ce": 0.15427130460739136,
+      "loss_iou": 0.6694341897964478,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 1.4554455280303955,
+      "num_input_tokens_seen": 647886560,
+      "step": 3748
+    },
+    {
+      "epoch": 1.4358483339716583,
+      "grad_norm": 75.44481130968893,
+      "learning_rate": 5e-06,
+      "loss": 1.4704,
+      "num_input_tokens_seen": 648059536,
+      "step": 3749
+    },
+    {
+      "epoch": 1.4358483339716583,
+      "loss": 1.5968767404556274,
+      "loss_ce": 0.18753957748413086,
+      "loss_iou": 0.6483941078186035,
+      "loss_num": 0.0224609375,
+      "loss_xval": 1.409337043762207,
+      "num_input_tokens_seen": 648059536,
+      "step": 3749
+    },
+    {
+      "epoch": 1.436231328992723,
+      "grad_norm": 99.53147962818682,
+      "learning_rate": 5e-06,
+      "loss": 1.3952,
+      "num_input_tokens_seen": 648232472,
+      "step": 3750
+    },
+    {
+      "epoch": 1.436231328992723,
+      "eval_websight_new_CIoU": 0.9285513758659363,
+      "eval_websight_new_GIoU": 0.9281909763813019,
+      "eval_websight_new_IoU": 0.9289136528968811,
+      "eval_websight_new_MAE_all": 0.006849759723991156,
+      "eval_websight_new_MAE_h": 0.004385469830594957,
+      "eval_websight_new_MAE_w": 0.008191783213987947,
+      "eval_websight_new_MAE_x": 0.008513058768585324,
+      "eval_websight_new_MAE_y": 0.006308725103735924,
+      "eval_websight_new_NUM_probability": 0.12843996286392212,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 1.0500247478485107,
+      "eval_websight_new_loss_ce": 0.4122988283634186,
+      "eval_websight_new_loss_iou": 0.30277392268180847,
+      "eval_websight_new_loss_num": 0.0061969757080078125,
+      "eval_websight_new_loss_xval": 0.6365155577659607,
+      "eval_websight_new_runtime": 55.9458,
+      "eval_websight_new_samples_per_second": 0.894,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 648232472,
+      "step": 3750
+    },
+    {
+      "epoch": 1.436231328992723,
+      "eval_seeclick_CIoU": 0.6186380684375763,
+      "eval_seeclick_GIoU": 0.6335676312446594,
+      "eval_seeclick_IoU": 0.6683100163936615,
+      "eval_seeclick_MAE_all": 0.07598558999598026,
+      "eval_seeclick_MAE_h": 0.03271046606823802,
+      "eval_seeclick_MAE_w": 0.11810741201043129,
+      "eval_seeclick_MAE_x": 0.11608804017305374,
+      "eval_seeclick_MAE_y": 0.03703642264008522,
+      "eval_seeclick_NUM_probability": 0.42897067964076996,
+      "eval_seeclick_inside_bbox": 0.8576388955116272,
+      "eval_seeclick_loss": 1.7334070205688477,
+      "eval_seeclick_loss_ce": 0.24777323752641678,
+      "eval_seeclick_loss_iou": 0.6064954698085785,
+      "eval_seeclick_loss_num": 0.049999237060546875,
+      "eval_seeclick_loss_xval": 1.463082492351532,
+      "eval_seeclick_runtime": 85.2465,
+      "eval_seeclick_samples_per_second": 0.587,
+      "eval_seeclick_steps_per_second": 0.023,
+      "num_input_tokens_seen": 648232472,
+      "step": 3750
+    },
+    {
+      "epoch": 1.436231328992723,
+      "eval_icons_CIoU": 0.803604245185852,
+      "eval_icons_GIoU": 0.8062641024589539,
+      "eval_icons_IoU": 0.8186492621898651,
+      "eval_icons_MAE_all": 0.028514167293906212,
+      "eval_icons_MAE_h": 0.014705082401633263,
+      "eval_icons_MAE_w": 0.039799148216843605,
+      "eval_icons_MAE_x": 0.043168528005480766,
+      "eval_icons_MAE_y": 0.016383902169764042,
+      "eval_icons_NUM_probability": 0.3404311537742615,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 1.9245898723602295,
+      "eval_icons_loss_ce": 0.3984464108943939,
+      "eval_icons_loss_iou": 0.7120104730129242,
+      "eval_icons_loss_num": 0.01808929443359375,
+      "eval_icons_loss_xval": 1.5145055651664734,
+      "eval_icons_runtime": 93.4996,
+      "eval_icons_samples_per_second": 0.535,
+      "eval_icons_steps_per_second": 0.021,
+      "num_input_tokens_seen": 648232472,
+      "step": 3750
+    },
+    {
+      "epoch": 1.436231328992723,
+      "loss": 1.9871809482574463,
+      "loss_ce": 0.3983060121536255,
+      "loss_iou": 0.7439765930175781,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 1.5888748168945312,
+      "num_input_tokens_seen": 648232472,
+      "step": 3750
+    },
+    {
+      "epoch": 1.4366143240137879,
+      "grad_norm": 85.30457316621232,
+      "learning_rate": 5e-06,
+      "loss": 1.5212,
+      "num_input_tokens_seen": 648405336,
+      "step": 3751
+    },
+    {
+      "epoch": 1.4366143240137879,
+      "loss": 1.478672981262207,
+      "loss_ce": 0.20677736401557922,
+      "loss_iou": 0.5964428186416626,
+      "loss_num": 0.015869140625,
+      "loss_xval": 1.2718956470489502,
+      "num_input_tokens_seen": 648405336,
+      "step": 3751
+    },
+    {
+      "epoch": 1.4369973190348526,
+      "grad_norm": 113.75891963551362,
+      "learning_rate": 5e-06,
+      "loss": 1.572,
+      "num_input_tokens_seen": 648578376,
+      "step": 3752
+    },
+    {
+      "epoch": 1.4369973190348526,
+      "loss": 1.4738179445266724,
+      "loss_ce": 0.2258295714855194,
+      "loss_iou": 0.5844433903694153,
+      "loss_num": 0.015869140625,
+      "loss_xval": 1.2479883432388306,
+      "num_input_tokens_seen": 648578376,
+      "step": 3752
+    },
+    {
+      "epoch": 1.4373803140559174,
+      "grad_norm": 196.98913144134312,
+      "learning_rate": 5e-06,
+      "loss": 1.6073,
+      "num_input_tokens_seen": 648751344,
+      "step": 3753
+    },
+    {
+      "epoch": 1.4373803140559174,
+      "loss": 1.7447113990783691,
+      "loss_ce": 0.17518246173858643,
+      "loss_iou": 0.7425433397293091,
+      "loss_num": 0.016845703125,
+      "loss_xval": 1.5695288181304932,
+      "num_input_tokens_seen": 648751344,
+      "step": 3753
+    },
+    {
+      "epoch": 1.4377633090769821,
+      "grad_norm": 274.07027465792294,
+      "learning_rate": 5e-06,
+      "loss": 1.9837,
+      "num_input_tokens_seen": 648924448,
+      "step": 3754
+    },
+    {
+      "epoch": 1.4377633090769821,
+      "loss": 1.9940869808197021,
+      "loss_ce": 0.16950249671936035,
+      "loss_iou": 0.8452146053314209,
+      "loss_num": 0.02685546875,
+      "loss_xval": 1.8245844841003418,
+      "num_input_tokens_seen": 648924448,
+      "step": 3754
+    },
+    {
+      "epoch": 1.4381463040980467,
+      "grad_norm": 150.30767114356615,
+      "learning_rate": 5e-06,
+      "loss": 1.7165,
+      "num_input_tokens_seen": 649097424,
+      "step": 3755
+    },
+    {
+      "epoch": 1.4381463040980467,
+      "loss": 1.673203945159912,
+      "loss_ce": 0.18848004937171936,
+      "loss_iou": 0.6845005750656128,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 1.4847238063812256,
+      "num_input_tokens_seen": 649097424,
+      "step": 3755
+    },
+    {
+      "epoch": 1.4385292991191114,
+      "grad_norm": 178.9262208328964,
+      "learning_rate": 5e-06,
+      "loss": 1.6534,
+      "num_input_tokens_seen": 649270264,
+      "step": 3756
+    },
+    {
+      "epoch": 1.4385292991191114,
+      "loss": 1.4100279808044434,
+      "loss_ce": 0.132390558719635,
+      "loss_iou": 0.5667972564697266,
+      "loss_num": 0.02880859375,
+      "loss_xval": 1.2776374816894531,
+      "num_input_tokens_seen": 649270264,
+      "step": 3756
+    },
+    {
+      "epoch": 1.4389122941401762,
+      "grad_norm": 185.56122412169145,
+      "learning_rate": 5e-06,
+      "loss": 2.2031,
+      "num_input_tokens_seen": 649443120,
+      "step": 3757
+    },
+    {
+      "epoch": 1.4389122941401762,
+      "loss": 2.201828718185425,
+      "loss_ce": 0.18113598227500916,
+      "loss_iou": 0.8843087553977966,
+      "loss_num": 0.05029296875,
+      "loss_xval": 2.020692825317383,
+      "num_input_tokens_seen": 649443120,
+      "step": 3757
+    },
+    {
+      "epoch": 1.439295289161241,
+      "grad_norm": 156.22916748801134,
+      "learning_rate": 5e-06,
+      "loss": 1.6584,
+      "num_input_tokens_seen": 649616224,
+      "step": 3758
+    },
+    {
+      "epoch": 1.439295289161241,
+      "loss": 1.7143478393554688,
+      "loss_ce": 0.2034144401550293,
+      "loss_iou": 0.6921732425689697,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 1.5109333992004395,
+      "num_input_tokens_seen": 649616224,
+      "step": 3758
+    },
+    {
+      "epoch": 1.4396782841823057,
+      "grad_norm": 134.4854109687959,
+      "learning_rate": 5e-06,
+      "loss": 1.4323,
+      "num_input_tokens_seen": 649789256,
+      "step": 3759
+    },
+    {
+      "epoch": 1.4396782841823057,
+      "loss": 1.3913935422897339,
+      "loss_ce": 0.20808057487010956,
+      "loss_iou": 0.531201183795929,
+      "loss_num": 0.024169921875,
+      "loss_xval": 1.183313012123108,
+      "num_input_tokens_seen": 649789256,
+      "step": 3759
+    },
+    {
+      "epoch": 1.4400612792033702,
+      "grad_norm": 67.81917494606034,
+      "learning_rate": 5e-06,
+      "loss": 1.9246,
+      "num_input_tokens_seen": 649962464,
+      "step": 3760
+    },
+    {
+      "epoch": 1.4400612792033702,
+      "loss": 1.9014415740966797,
+      "loss_ce": 0.25480008125305176,
+      "loss_iou": 0.772661566734314,
+      "loss_num": 0.020263671875,
+      "loss_xval": 1.646641492843628,
+      "num_input_tokens_seen": 649962464,
+      "step": 3760
+    },
+    {
+      "epoch": 1.440444274224435,
+      "grad_norm": 101.11515987968119,
+      "learning_rate": 5e-06,
+      "loss": 1.4676,
+      "num_input_tokens_seen": 650135224,
+      "step": 3761
+    },
+    {
+      "epoch": 1.440444274224435,
+      "loss": 1.2977476119995117,
+      "loss_ce": 0.17146068811416626,
+      "loss_iou": 0.5049769282341003,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 1.1262868642807007,
+      "num_input_tokens_seen": 650135224,
+      "step": 3761
+    },
+    {
+      "epoch": 1.4408272692454998,
+      "grad_norm": 165.17808664996986,
+      "learning_rate": 5e-06,
+      "loss": 1.5256,
+      "num_input_tokens_seen": 650308608,
+      "step": 3762
+    },
+    {
+      "epoch": 1.4408272692454998,
+      "loss": 1.5596181154251099,
+      "loss_ce": 0.19343899190425873,
+      "loss_iou": 0.633162796497345,
+      "loss_num": 0.02001953125,
+      "loss_xval": 1.36617910861969,
+      "num_input_tokens_seen": 650308608,
+      "step": 3762
+    },
+    {
+      "epoch": 1.4412102642665645,
+      "grad_norm": 120.37777033006283,
+      "learning_rate": 5e-06,
+      "loss": 1.6499,
+      "num_input_tokens_seen": 650481568,
+      "step": 3763
+    },
+    {
+      "epoch": 1.4412102642665645,
+      "loss": 1.596879482269287,
+      "loss_ce": 0.16848647594451904,
+      "loss_iou": 0.6603940725326538,
+      "loss_num": 0.021484375,
+      "loss_xval": 1.4283931255340576,
+      "num_input_tokens_seen": 650481568,
+      "step": 3763
+    },
+    {
+      "epoch": 1.4415932592876293,
+      "grad_norm": 76.98411155572275,
+      "learning_rate": 5e-06,
+      "loss": 1.2079,
+      "num_input_tokens_seen": 650654240,
+      "step": 3764
+    },
+    {
+      "epoch": 1.4415932592876293,
+      "loss": 1.2885901927947998,
+      "loss_ce": 0.1566745638847351,
+      "loss_iou": 0.515710711479187,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 1.131915807723999,
+      "num_input_tokens_seen": 650654240,
+      "step": 3764
+    },
+    {
+      "epoch": 1.441976254308694,
+      "grad_norm": 105.05592545315972,
+      "learning_rate": 5e-06,
+      "loss": 1.2087,
+      "num_input_tokens_seen": 650827232,
+      "step": 3765
+    },
+    {
+      "epoch": 1.441976254308694,
+      "loss": 1.3063874244689941,
+      "loss_ce": 0.18088823556900024,
+      "loss_iou": 0.5067803263664246,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 1.1254992485046387,
+      "num_input_tokens_seen": 650827232,
+      "step": 3765
+    },
+    {
+      "epoch": 1.4423592493297588,
+      "grad_norm": 84.51373154328186,
+      "learning_rate": 5e-06,
+      "loss": 1.7202,
+      "num_input_tokens_seen": 651000288,
+      "step": 3766
+    },
+    {
+      "epoch": 1.4423592493297588,
+      "loss": 1.9406160116195679,
+      "loss_ce": 0.2464371621608734,
+      "loss_iou": 0.7758309245109558,
+      "loss_num": 0.028564453125,
+      "loss_xval": 1.6941789388656616,
+      "num_input_tokens_seen": 651000288,
+      "step": 3766
+    },
+    {
+      "epoch": 1.4427422443508235,
+      "grad_norm": 70.52425073354833,
+      "learning_rate": 5e-06,
+      "loss": 1.261,
+      "num_input_tokens_seen": 651173392,
+      "step": 3767
+    },
+    {
+      "epoch": 1.4427422443508235,
+      "loss": 1.3253724575042725,
+      "loss_ce": 0.20682013034820557,
+      "loss_iou": 0.5058094263076782,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 1.1185524463653564,
+      "num_input_tokens_seen": 651173392,
+      "step": 3767
+    },
+    {
+      "epoch": 1.4431252393718883,
+      "grad_norm": 247.24629844383443,
+      "learning_rate": 5e-06,
+      "loss": 1.5622,
+      "num_input_tokens_seen": 651346336,
+      "step": 3768
+    },
+    {
+      "epoch": 1.4431252393718883,
+      "loss": 1.6428744792938232,
+      "loss_ce": 0.19238018989562988,
+      "loss_iou": 0.6811187267303467,
+      "loss_num": 0.017578125,
+      "loss_xval": 1.4504942893981934,
+      "num_input_tokens_seen": 651346336,
+      "step": 3768
+    },
+    {
+      "epoch": 1.4435082343929528,
+      "grad_norm": 95.39385812224009,
+      "learning_rate": 5e-06,
+      "loss": 1.8846,
+      "num_input_tokens_seen": 651519320,
+      "step": 3769
+    },
+    {
+      "epoch": 1.4435082343929528,
+      "loss": 1.7897255420684814,
+      "loss_ce": 0.22061772644519806,
+      "loss_iou": 0.7338947653770447,
+      "loss_num": 0.020263671875,
+      "loss_xval": 1.5691078901290894,
+      "num_input_tokens_seen": 651519320,
+      "step": 3769
+    },
+    {
+      "epoch": 1.4438912294140176,
+      "grad_norm": 144.53921328265437,
+      "learning_rate": 5e-06,
+      "loss": 1.5361,
+      "num_input_tokens_seen": 651692672,
+      "step": 3770
+    },
+    {
+      "epoch": 1.4438912294140176,
+      "loss": 1.5159696340560913,
+      "loss_ce": 0.21220770478248596,
+      "loss_iou": 0.5900828838348389,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 1.3037619590759277,
+      "num_input_tokens_seen": 651692672,
+      "step": 3770
+    },
+    {
+      "epoch": 1.4442742244350824,
+      "grad_norm": 135.80791640212837,
+      "learning_rate": 5e-06,
+      "loss": 1.7292,
+      "num_input_tokens_seen": 651865456,
+      "step": 3771
+    },
+    {
+      "epoch": 1.4442742244350824,
+      "loss": 1.7545547485351562,
+      "loss_ce": 0.184688538312912,
+      "loss_iou": 0.7362881302833557,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 1.5698662996292114,
+      "num_input_tokens_seen": 651865456,
+      "step": 3771
+    },
+    {
+      "epoch": 1.4446572194561471,
+      "grad_norm": 94.44691510131742,
+      "learning_rate": 5e-06,
+      "loss": 1.91,
+      "num_input_tokens_seen": 652038360,
+      "step": 3772
+    },
+    {
+      "epoch": 1.4446572194561471,
+      "loss": 1.745355248451233,
+      "loss_ce": 0.18886971473693848,
+      "loss_iou": 0.7118974924087524,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 1.5564854145050049,
+      "num_input_tokens_seen": 652038360,
+      "step": 3772
+    },
+    {
+      "epoch": 1.4450402144772119,
+      "grad_norm": 109.58872241046265,
+      "learning_rate": 5e-06,
+      "loss": 1.4302,
+      "num_input_tokens_seen": 652211344,
+      "step": 3773
+    },
+    {
+      "epoch": 1.4450402144772119,
+      "loss": 1.5201438665390015,
+      "loss_ce": 0.167641282081604,
+      "loss_iou": 0.6205872297286987,
+      "loss_num": 0.022216796875,
+      "loss_xval": 1.3525025844573975,
+      "num_input_tokens_seen": 652211344,
+      "step": 3773
+    },
+    {
+      "epoch": 1.4454232094982764,
+      "grad_norm": 126.77775998202408,
+      "learning_rate": 5e-06,
+      "loss": 1.5639,
+      "num_input_tokens_seen": 652384360,
+      "step": 3774
+    },
+    {
+      "epoch": 1.4454232094982764,
+      "loss": 1.648437261581421,
+      "loss_ce": 0.1750509887933731,
+      "loss_iou": 0.6851489543914795,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 1.473386287689209,
+      "num_input_tokens_seen": 652384360,
+      "step": 3774
+    },
+    {
+      "epoch": 1.4458062045193412,
+      "grad_norm": 106.01905857449188,
+      "learning_rate": 5e-06,
+      "loss": 1.7414,
+      "num_input_tokens_seen": 652557240,
+      "step": 3775
+    },
+    {
+      "epoch": 1.4458062045193412,
+      "loss": 1.8190821409225464,
+      "loss_ce": 0.2274116724729538,
+      "loss_iou": 0.7480751872062683,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 1.5916703939437866,
+      "num_input_tokens_seen": 652557240,
+      "step": 3775
+    },
+    {
+      "epoch": 1.446189199540406,
+      "grad_norm": 72.76302246621044,
+      "learning_rate": 5e-06,
+      "loss": 1.4197,
+      "num_input_tokens_seen": 652730304,
+      "step": 3776
+    },
+    {
+      "epoch": 1.446189199540406,
+      "loss": 1.3549346923828125,
+      "loss_ce": 0.2012617588043213,
+      "loss_iou": 0.5414360761642456,
+      "loss_num": 0.01416015625,
+      "loss_xval": 1.1536729335784912,
+      "num_input_tokens_seen": 652730304,
+      "step": 3776
+    },
+    {
+      "epoch": 1.4465721945614707,
+      "grad_norm": 121.25168514069287,
+      "learning_rate": 5e-06,
+      "loss": 1.5176,
+      "num_input_tokens_seen": 652903680,
+      "step": 3777
+    },
+    {
+      "epoch": 1.4465721945614707,
+      "loss": 1.3323194980621338,
+      "loss_ce": 0.21395091712474823,
+      "loss_iou": 0.5245010852813721,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 1.1183686256408691,
+      "num_input_tokens_seen": 652903680,
+      "step": 3777
+    },
+    {
+      "epoch": 1.4469551895825354,
+      "grad_norm": 256.18857843742103,
+      "learning_rate": 5e-06,
+      "loss": 1.6278,
+      "num_input_tokens_seen": 653076376,
+      "step": 3778
+    },
+    {
+      "epoch": 1.4469551895825354,
+      "loss": 1.4770128726959229,
+      "loss_ce": 0.16997739672660828,
+      "loss_iou": 0.6138449311256409,
+      "loss_num": 0.015869140625,
+      "loss_xval": 1.3070355653762817,
+      "num_input_tokens_seen": 653076376,
+      "step": 3778
+    },
+    {
+      "epoch": 1.4473381846036002,
+      "grad_norm": 385.7979687662338,
+      "learning_rate": 5e-06,
+      "loss": 1.7794,
+      "num_input_tokens_seen": 653249376,
+      "step": 3779
+    },
+    {
+      "epoch": 1.4473381846036002,
+      "loss": 1.6642146110534668,
+      "loss_ce": 0.19940748810768127,
+      "loss_iou": 0.658551037311554,
+      "loss_num": 0.029541015625,
+      "loss_xval": 1.464807152748108,
+      "num_input_tokens_seen": 653249376,
+      "step": 3779
+    },
+    {
+      "epoch": 1.447721179624665,
+      "grad_norm": 206.30552232507333,
+      "learning_rate": 5e-06,
+      "loss": 1.7149,
+      "num_input_tokens_seen": 653422032,
+      "step": 3780
+    },
+    {
+      "epoch": 1.447721179624665,
+      "loss": 1.7484610080718994,
+      "loss_ce": 0.18937820196151733,
+      "loss_iou": 0.7355655431747437,
+      "loss_num": 0.017578125,
+      "loss_xval": 1.5590827465057373,
+      "num_input_tokens_seen": 653422032,
+      "step": 3780
+    },
+    {
+      "epoch": 1.4481041746457297,
+      "grad_norm": 134.49787199330845,
+      "learning_rate": 5e-06,
+      "loss": 1.8694,
+      "num_input_tokens_seen": 653595408,
+      "step": 3781
+    },
+    {
+      "epoch": 1.4481041746457297,
+      "loss": 1.9886581897735596,
+      "loss_ce": 0.18671278655529022,
+      "loss_iou": 0.8289817571640015,
+      "loss_num": 0.02880859375,
+      "loss_xval": 1.801945447921753,
+      "num_input_tokens_seen": 653595408,
+      "step": 3781
+    },
+    {
+      "epoch": 1.4484871696667942,
+      "grad_norm": 76.68047915267991,
+      "learning_rate": 5e-06,
+      "loss": 1.3536,
+      "num_input_tokens_seen": 653768352,
+      "step": 3782
+    },
+    {
+      "epoch": 1.4484871696667942,
+      "loss": 1.3922433853149414,
+      "loss_ce": 0.18917998671531677,
+      "loss_iou": 0.5656887292861938,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 1.2030632495880127,
+      "num_input_tokens_seen": 653768352,
+      "step": 3782
+    },
+    {
+      "epoch": 1.448870164687859,
+      "grad_norm": 124.78964133467781,
+      "learning_rate": 5e-06,
+      "loss": 1.3352,
+      "num_input_tokens_seen": 653941408,
+      "step": 3783
+    },
+    {
+      "epoch": 1.448870164687859,
+      "loss": 1.2876272201538086,
+      "loss_ce": 0.19031816720962524,
+      "loss_iou": 0.5140933990478516,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 1.0973091125488281,
+      "num_input_tokens_seen": 653941408,
+      "step": 3783
+    },
+    {
+      "epoch": 1.4492531597089238,
+      "grad_norm": 120.9594175394944,
+      "learning_rate": 5e-06,
+      "loss": 1.6848,
+      "num_input_tokens_seen": 654114296,
+      "step": 3784
+    },
+    {
+      "epoch": 1.4492531597089238,
+      "loss": 1.6132863759994507,
+      "loss_ce": 0.2119094878435135,
+      "loss_iou": 0.6478319764137268,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 1.4013768434524536,
+      "num_input_tokens_seen": 654114296,
+      "step": 3784
+    },
+    {
+      "epoch": 1.4496361547299885,
+      "grad_norm": 105.09965336934216,
+      "learning_rate": 5e-06,
+      "loss": 1.374,
+      "num_input_tokens_seen": 654287312,
+      "step": 3785
+    },
+    {
+      "epoch": 1.4496361547299885,
+      "loss": 1.2698428630828857,
+      "loss_ce": 0.22151289880275726,
+      "loss_iou": 0.48902401328086853,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 1.0483300685882568,
+      "num_input_tokens_seen": 654287312,
+      "step": 3785
+    },
+    {
+      "epoch": 1.4500191497510533,
+      "grad_norm": 119.20706807447374,
+      "learning_rate": 5e-06,
+      "loss": 1.3342,
+      "num_input_tokens_seen": 654456576,
+      "step": 3786
+    },
+    {
+      "epoch": 1.4500191497510533,
+      "loss": 1.2609329223632812,
+      "loss_ce": 0.11578431725502014,
+      "loss_iou": 0.5418888330459595,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 1.145148515701294,
+      "num_input_tokens_seen": 654456576,
+      "step": 3786
+    },
+    {
+      "epoch": 1.450402144772118,
+      "grad_norm": 131.5739674704228,
+      "learning_rate": 5e-06,
+      "loss": 1.4568,
+      "num_input_tokens_seen": 654629664,
+      "step": 3787
+    },
+    {
+      "epoch": 1.450402144772118,
+      "loss": 1.3627843856811523,
+      "loss_ce": 0.22565358877182007,
+      "loss_iou": 0.5240402817726135,
+      "loss_num": 0.017822265625,
+      "loss_xval": 1.137130856513977,
+      "num_input_tokens_seen": 654629664,
+      "step": 3787
+    },
+    {
+      "epoch": 1.4507851397931826,
+      "grad_norm": 123.91765590689454,
+      "learning_rate": 5e-06,
+      "loss": 1.4768,
+      "num_input_tokens_seen": 654802256,
+      "step": 3788
+    },
+    {
+      "epoch": 1.4507851397931826,
+      "loss": 1.4885624647140503,
+      "loss_ce": 0.17179864645004272,
+      "loss_iou": 0.6020312309265137,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 1.2755041122436523,
+      "num_input_tokens_seen": 654802256,
+      "step": 3788
+    },
+    {
+      "epoch": 1.4511681348142473,
+      "grad_norm": 120.84765963073131,
+      "learning_rate": 5e-06,
+      "loss": 1.411,
+      "num_input_tokens_seen": 654975000,
+      "step": 3789
+    },
+    {
+      "epoch": 1.4511681348142473,
+      "loss": 1.4020578861236572,
+      "loss_ce": 0.15986859798431396,
+      "loss_iou": 0.5749979019165039,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 1.2421894073486328,
+      "num_input_tokens_seen": 654975000,
+      "step": 3789
+    },
+    {
+      "epoch": 1.451551129835312,
+      "grad_norm": 103.58734996622945,
+      "learning_rate": 5e-06,
+      "loss": 1.6362,
+      "num_input_tokens_seen": 655147560,
+      "step": 3790
+    },
+    {
+      "epoch": 1.451551129835312,
+      "loss": 1.6650502681732178,
+      "loss_ce": 0.18911117315292358,
+      "loss_iou": 0.6893855333328247,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 1.4759390354156494,
+      "num_input_tokens_seen": 655147560,
+      "step": 3790
+    },
+    {
+      "epoch": 1.4519341248563769,
+      "grad_norm": 118.18186315359165,
+      "learning_rate": 5e-06,
+      "loss": 1.548,
+      "num_input_tokens_seen": 655320184,
+      "step": 3791
+    },
+    {
+      "epoch": 1.4519341248563769,
+      "loss": 1.671811819076538,
+      "loss_ce": 0.16698873043060303,
+      "loss_iou": 0.6983343362808228,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 1.504823088645935,
+      "num_input_tokens_seen": 655320184,
+      "step": 3791
+    },
+    {
+      "epoch": 1.4523171198774416,
+      "grad_norm": 168.50277127678518,
+      "learning_rate": 5e-06,
+      "loss": 2.0662,
+      "num_input_tokens_seen": 655492912,
+      "step": 3792
+    },
+    {
+      "epoch": 1.4523171198774416,
+      "loss": 2.023236036300659,
+      "loss_ce": 0.15831105411052704,
+      "loss_iou": 0.8838174343109131,
+      "loss_num": 0.01953125,
+      "loss_xval": 1.8649249076843262,
+      "num_input_tokens_seen": 655492912,
+      "step": 3792
+    },
+    {
+      "epoch": 1.4527001148985064,
+      "grad_norm": 89.94589993359895,
+      "learning_rate": 5e-06,
+      "loss": 1.8428,
+      "num_input_tokens_seen": 655666056,
+      "step": 3793
+    },
+    {
+      "epoch": 1.4527001148985064,
+      "loss": 1.965169906616211,
+      "loss_ce": 0.21689045429229736,
+      "loss_iou": 0.791223406791687,
+      "loss_num": 0.033203125,
+      "loss_xval": 1.748279333114624,
+      "num_input_tokens_seen": 655666056,
+      "step": 3793
+    },
+    {
+      "epoch": 1.4530831099195711,
+      "grad_norm": 195.35134482147066,
+      "learning_rate": 5e-06,
+      "loss": 1.5465,
+      "num_input_tokens_seen": 655839000,
+      "step": 3794
+    },
+    {
+      "epoch": 1.4530831099195711,
+      "loss": 1.5583913326263428,
+      "loss_ce": 0.14241421222686768,
+      "loss_iou": 0.6645772457122803,
+      "loss_num": 0.017333984375,
+      "loss_xval": 1.4159770011901855,
+      "num_input_tokens_seen": 655839000,
+      "step": 3794
+    },
+    {
+      "epoch": 1.4534661049406359,
+      "grad_norm": 138.16568988915304,
+      "learning_rate": 5e-06,
+      "loss": 1.9311,
+      "num_input_tokens_seen": 656012144,
+      "step": 3795
+    },
+    {
+      "epoch": 1.4534661049406359,
+      "loss": 1.902432918548584,
+      "loss_ce": 0.16578739881515503,
+      "loss_iou": 0.8244079351425171,
+      "loss_num": 0.017578125,
+      "loss_xval": 1.7366454601287842,
+      "num_input_tokens_seen": 656012144,
+      "step": 3795
+    },
+    {
+      "epoch": 1.4538490999617004,
+      "grad_norm": 91.95252792391383,
+      "learning_rate": 5e-06,
+      "loss": 1.9875,
+      "num_input_tokens_seen": 656184864,
+      "step": 3796
+    },
+    {
+      "epoch": 1.4538490999617004,
+      "loss": 1.8245433568954468,
+      "loss_ce": 0.1364639848470688,
+      "loss_iou": 0.7680509090423584,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 1.6880793571472168,
+      "num_input_tokens_seen": 656184864,
+      "step": 3796
+    },
+    {
+      "epoch": 1.4542320949827652,
+      "grad_norm": 93.74971054419059,
+      "learning_rate": 5e-06,
+      "loss": 1.529,
+      "num_input_tokens_seen": 656357800,
+      "step": 3797
+    },
+    {
+      "epoch": 1.4542320949827652,
+      "loss": 1.3864085674285889,
+      "loss_ce": 0.1551656872034073,
+      "loss_iou": 0.5659693479537964,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 1.2312428951263428,
+      "num_input_tokens_seen": 656357800,
+      "step": 3797
+    },
+    {
+      "epoch": 1.45461509000383,
+      "grad_norm": 211.2465508525641,
+      "learning_rate": 5e-06,
+      "loss": 1.9917,
+      "num_input_tokens_seen": 656530216,
+      "step": 3798
+    },
+    {
+      "epoch": 1.45461509000383,
+      "loss": 1.9216835498809814,
+      "loss_ce": 0.14129409193992615,
+      "loss_iou": 0.8456543684005737,
+      "loss_num": 0.017822265625,
+      "loss_xval": 1.7803895473480225,
+      "num_input_tokens_seen": 656530216,
+      "step": 3798
+    },
+    {
+      "epoch": 1.4549980850248947,
+      "grad_norm": 111.48053081234072,
+      "learning_rate": 5e-06,
+      "loss": 1.7296,
+      "num_input_tokens_seen": 656703144,
+      "step": 3799
+    },
+    {
+      "epoch": 1.4549980850248947,
+      "loss": 1.7023835182189941,
+      "loss_ce": 0.17317184805870056,
+      "loss_iou": 0.6857789754867554,
+      "loss_num": 0.031494140625,
+      "loss_xval": 1.5292117595672607,
+      "num_input_tokens_seen": 656703144,
+      "step": 3799
+    },
+    {
+      "epoch": 1.4553810800459595,
+      "grad_norm": 111.11626330585243,
+      "learning_rate": 5e-06,
+      "loss": 1.2795,
+      "num_input_tokens_seen": 656875856,
+      "step": 3800
+    },
+    {
+      "epoch": 1.4553810800459595,
+      "loss": 1.0074357986450195,
+      "loss_ce": 0.1625213772058487,
+      "loss_iou": 0.391291081905365,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.8449143767356873,
+      "num_input_tokens_seen": 656875856,
+      "step": 3800
+    },
+    {
+      "epoch": 1.4557640750670242,
+      "grad_norm": 104.89400882774832,
+      "learning_rate": 5e-06,
+      "loss": 1.4837,
+      "num_input_tokens_seen": 657048928,
+      "step": 3801
+    },
+    {
+      "epoch": 1.4557640750670242,
+      "loss": 1.5924118757247925,
+      "loss_ce": 0.19961793720722198,
+      "loss_iou": 0.6431438326835632,
+      "loss_num": 0.021240234375,
+      "loss_xval": 1.3927940130233765,
+      "num_input_tokens_seen": 657048928,
+      "step": 3801
+    },
+    {
+      "epoch": 1.4561470700880887,
+      "grad_norm": 129.35849983658812,
+      "learning_rate": 5e-06,
+      "loss": 1.6337,
+      "num_input_tokens_seen": 657222032,
+      "step": 3802
+    },
+    {
+      "epoch": 1.4561470700880887,
+      "loss": 1.6484363079071045,
+      "loss_ce": 0.1564095914363861,
+      "loss_iou": 0.6908985376358032,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 1.4920265674591064,
+      "num_input_tokens_seen": 657222032,
+      "step": 3802
+    },
+    {
+      "epoch": 1.4565300651091535,
+      "grad_norm": 158.4448251316895,
+      "learning_rate": 5e-06,
+      "loss": 1.6882,
+      "num_input_tokens_seen": 657395000,
+      "step": 3803
+    },
+    {
+      "epoch": 1.4565300651091535,
+      "loss": 1.8415119647979736,
+      "loss_ce": 0.23011425137519836,
+      "loss_iou": 0.749149739742279,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 1.611397624015808,
+      "num_input_tokens_seen": 657395000,
+      "step": 3803
+    },
+    {
+      "epoch": 1.4569130601302183,
+      "grad_norm": 90.37963164548083,
+      "learning_rate": 5e-06,
+      "loss": 1.7413,
+      "num_input_tokens_seen": 657568096,
+      "step": 3804
+    },
+    {
+      "epoch": 1.4569130601302183,
+      "loss": 1.6751604080200195,
+      "loss_ce": 0.18919594585895538,
+      "loss_iou": 0.6686719655990601,
+      "loss_num": 0.02978515625,
+      "loss_xval": 1.4859645366668701,
+      "num_input_tokens_seen": 657568096,
+      "step": 3804
+    },
+    {
+      "epoch": 1.457296055151283,
+      "grad_norm": 97.08764336384468,
+      "learning_rate": 5e-06,
+      "loss": 1.6194,
+      "num_input_tokens_seen": 657741072,
+      "step": 3805
+    },
+    {
+      "epoch": 1.457296055151283,
+      "loss": 1.3500893115997314,
+      "loss_ce": 0.17926084995269775,
+      "loss_iou": 0.5503036975860596,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 1.1708283424377441,
+      "num_input_tokens_seen": 657741072,
+      "step": 3805
+    },
+    {
+      "epoch": 1.4576790501723478,
+      "grad_norm": 92.41936563895115,
+      "learning_rate": 5e-06,
+      "loss": 1.3274,
+      "num_input_tokens_seen": 657914192,
+      "step": 3806
+    },
+    {
+      "epoch": 1.4576790501723478,
+      "loss": 1.2393465042114258,
+      "loss_ce": 0.18367427587509155,
+      "loss_iou": 0.4906809329986572,
+      "loss_num": 0.014892578125,
+      "loss_xval": 1.0556721687316895,
+      "num_input_tokens_seen": 657914192,
+      "step": 3806
+    },
+    {
+      "epoch": 1.4580620451934125,
+      "grad_norm": 272.39600150539786,
+      "learning_rate": 5e-06,
+      "loss": 1.5496,
+      "num_input_tokens_seen": 658087064,
+      "step": 3807
+    },
+    {
+      "epoch": 1.4580620451934125,
+      "loss": 1.6365811824798584,
+      "loss_ce": 0.20125389099121094,
+      "loss_iou": 0.6523712277412415,
+      "loss_num": 0.026123046875,
+      "loss_xval": 1.435327172279358,
+      "num_input_tokens_seen": 658087064,
+      "step": 3807
+    },
+    {
+      "epoch": 1.4584450402144773,
+      "grad_norm": 145.24395504305105,
+      "learning_rate": 5e-06,
+      "loss": 1.7756,
+      "num_input_tokens_seen": 658259712,
+      "step": 3808
+    },
+    {
+      "epoch": 1.4584450402144773,
+      "loss": 1.8197343349456787,
+      "loss_ce": 0.1754903495311737,
+      "loss_iou": 0.7533658742904663,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 1.6442439556121826,
+      "num_input_tokens_seen": 658259712,
+      "step": 3808
+    },
+    {
+      "epoch": 1.458828035235542,
+      "grad_norm": 193.5180404049713,
+      "learning_rate": 5e-06,
+      "loss": 1.7664,
+      "num_input_tokens_seen": 658432736,
+      "step": 3809
+    },
+    {
+      "epoch": 1.458828035235542,
+      "loss": 1.5885456800460815,
+      "loss_ce": 0.20038504898548126,
+      "loss_iou": 0.6512947082519531,
+      "loss_num": 0.01708984375,
+      "loss_xval": 1.3881607055664062,
+      "num_input_tokens_seen": 658432736,
+      "step": 3809
+    },
+    {
+      "epoch": 1.4592110302566066,
+      "grad_norm": 184.14992587635396,
+      "learning_rate": 5e-06,
+      "loss": 1.5467,
+      "num_input_tokens_seen": 658605616,
+      "step": 3810
+    },
+    {
+      "epoch": 1.4592110302566066,
+      "loss": 1.6167919635772705,
+      "loss_ce": 0.16174444556236267,
+      "loss_iou": 0.6547240614891052,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 1.4550474882125854,
+      "num_input_tokens_seen": 658605616,
+      "step": 3810
+    },
+    {
+      "epoch": 1.4595940252776713,
+      "grad_norm": 212.16964484122292,
+      "learning_rate": 5e-06,
+      "loss": 2.2109,
+      "num_input_tokens_seen": 658778496,
+      "step": 3811
+    },
+    {
+      "epoch": 1.4595940252776713,
+      "loss": 2.2612743377685547,
+      "loss_ce": 0.19108566641807556,
+      "loss_iou": 0.9135123491287231,
+      "loss_num": 0.048583984375,
+      "loss_xval": 2.0701887607574463,
+      "num_input_tokens_seen": 658778496,
+      "step": 3811
+    },
+    {
+      "epoch": 1.459977020298736,
+      "grad_norm": 81.01815564486779,
+      "learning_rate": 5e-06,
+      "loss": 1.8385,
+      "num_input_tokens_seen": 658951296,
+      "step": 3812
+    },
+    {
+      "epoch": 1.459977020298736,
+      "loss": 1.8274188041687012,
+      "loss_ce": 0.18849460780620575,
+      "loss_iou": 0.7424662113189697,
+      "loss_num": 0.03076171875,
+      "loss_xval": 1.6389241218566895,
+      "num_input_tokens_seen": 658951296,
+      "step": 3812
+    },
+    {
+      "epoch": 1.4603600153198009,
+      "grad_norm": 49.95469091414546,
+      "learning_rate": 5e-06,
+      "loss": 1.5201,
+      "num_input_tokens_seen": 659124368,
+      "step": 3813
+    },
+    {
+      "epoch": 1.4603600153198009,
+      "loss": 1.5202065706253052,
+      "loss_ce": 0.13053268194198608,
+      "loss_iou": 0.6430333852767944,
+      "loss_num": 0.020751953125,
+      "loss_xval": 1.3740489482879639,
+      "num_input_tokens_seen": 659124368,
+      "step": 3813
+    },
+    {
+      "epoch": 1.4607430103408656,
+      "grad_norm": 107.13977637727947,
+      "learning_rate": 5e-06,
+      "loss": 1.3482,
+      "num_input_tokens_seen": 659297336,
+      "step": 3814
+    },
+    {
+      "epoch": 1.4607430103408656,
+      "loss": 1.1944173574447632,
+      "loss_ce": 0.1378280520439148,
+      "loss_iou": 0.4841967523097992,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 1.0565893650054932,
+      "num_input_tokens_seen": 659297336,
+      "step": 3814
+    },
+    {
+      "epoch": 1.4611260053619302,
+      "grad_norm": 138.38002923188432,
+      "learning_rate": 5e-06,
+      "loss": 1.7644,
+      "num_input_tokens_seen": 659470184,
+      "step": 3815
+    },
+    {
+      "epoch": 1.4611260053619302,
+      "loss": 1.7486977577209473,
+      "loss_ce": 0.20691266655921936,
+      "loss_iou": 0.6980165243148804,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 1.5417850017547607,
+      "num_input_tokens_seen": 659470184,
+      "step": 3815
+    },
+    {
+      "epoch": 1.461509000382995,
+      "grad_norm": 126.91918089264476,
+      "learning_rate": 5e-06,
+      "loss": 1.701,
+      "num_input_tokens_seen": 659642752,
+      "step": 3816
+    },
+    {
+      "epoch": 1.461509000382995,
+      "loss": 1.7082602977752686,
+      "loss_ce": 0.16621346771717072,
+      "loss_iou": 0.7019315958023071,
+      "loss_num": 0.027587890625,
+      "loss_xval": 1.5420467853546143,
+      "num_input_tokens_seen": 659642752,
+      "step": 3816
+    },
+    {
+      "epoch": 1.4618919954040597,
+      "grad_norm": 160.46795186791917,
+      "learning_rate": 5e-06,
+      "loss": 1.2791,
+      "num_input_tokens_seen": 659815784,
+      "step": 3817
+    },
+    {
+      "epoch": 1.4618919954040597,
+      "loss": 1.3117930889129639,
+      "loss_ce": 0.15506237745285034,
+      "loss_iou": 0.5235862731933594,
+      "loss_num": 0.02197265625,
+      "loss_xval": 1.1567306518554688,
+      "num_input_tokens_seen": 659815784,
+      "step": 3817
+    },
+    {
+      "epoch": 1.4622749904251244,
+      "grad_norm": 143.27679106100427,
+      "learning_rate": 5e-06,
+      "loss": 1.4454,
+      "num_input_tokens_seen": 659988680,
+      "step": 3818
+    },
+    {
+      "epoch": 1.4622749904251244,
+      "loss": 1.5526273250579834,
+      "loss_ce": 0.15987583994865417,
+      "loss_iou": 0.6397961378097534,
+      "loss_num": 0.022705078125,
+      "loss_xval": 1.3927514553070068,
+      "num_input_tokens_seen": 659988680,
+      "step": 3818
+    },
+    {
+      "epoch": 1.4626579854461892,
+      "grad_norm": 240.47910652884292,
+      "learning_rate": 5e-06,
+      "loss": 2.0018,
+      "num_input_tokens_seen": 660161832,
+      "step": 3819
+    },
+    {
+      "epoch": 1.4626579854461892,
+      "loss": 1.9347071647644043,
+      "loss_ce": 0.2272402048110962,
+      "loss_iou": 0.7853130102157593,
+      "loss_num": 0.02734375,
+      "loss_xval": 1.7074668407440186,
+      "num_input_tokens_seen": 660161832,
+      "step": 3819
+    },
+    {
+      "epoch": 1.463040980467254,
+      "grad_norm": 77.54631845218896,
+      "learning_rate": 5e-06,
+      "loss": 1.425,
+      "num_input_tokens_seen": 660334304,
+      "step": 3820
+    },
+    {
+      "epoch": 1.463040980467254,
+      "loss": 1.4100160598754883,
+      "loss_ce": 0.12929484248161316,
+      "loss_iou": 0.5822703838348389,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 1.2807211875915527,
+      "num_input_tokens_seen": 660334304,
+      "step": 3820
+    },
+    {
+      "epoch": 1.4634239754883187,
+      "grad_norm": 82.10061315820577,
+      "learning_rate": 5e-06,
+      "loss": 1.4941,
+      "num_input_tokens_seen": 660507456,
+      "step": 3821
+    },
+    {
+      "epoch": 1.4634239754883187,
+      "loss": 1.5278706550598145,
+      "loss_ce": 0.17169851064682007,
+      "loss_iou": 0.6022193431854248,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 1.3561720848083496,
+      "num_input_tokens_seen": 660507456,
+      "step": 3821
+    },
+    {
+      "epoch": 1.4638069705093835,
+      "grad_norm": 152.86590021249606,
+      "learning_rate": 5e-06,
+      "loss": 1.5364,
+      "num_input_tokens_seen": 660680816,
+      "step": 3822
+    },
+    {
+      "epoch": 1.4638069705093835,
+      "loss": 1.4629637002944946,
+      "loss_ce": 0.1718747466802597,
+      "loss_iou": 0.6119294166564941,
+      "loss_num": 0.013427734375,
+      "loss_xval": 1.2910890579223633,
+      "num_input_tokens_seen": 660680816,
+      "step": 3822
+    },
+    {
+      "epoch": 1.4641899655304482,
+      "grad_norm": 89.8082383751445,
+      "learning_rate": 5e-06,
+      "loss": 1.3508,
+      "num_input_tokens_seen": 660853888,
+      "step": 3823
+    },
+    {
+      "epoch": 1.4641899655304482,
+      "loss": 1.3696964979171753,
+      "loss_ce": 0.19920630753040314,
+      "loss_iou": 0.5320376753807068,
+      "loss_num": 0.021240234375,
+      "loss_xval": 1.1704901456832886,
+      "num_input_tokens_seen": 660853888,
+      "step": 3823
+    },
+    {
+      "epoch": 1.4645729605515128,
+      "grad_norm": 176.87759831014753,
+      "learning_rate": 5e-06,
+      "loss": 1.3726,
+      "num_input_tokens_seen": 661026400,
+      "step": 3824
+    },
+    {
+      "epoch": 1.4645729605515128,
+      "loss": 1.6656455993652344,
+      "loss_ce": 0.11513148248195648,
+      "loss_iou": 0.7068977355957031,
+      "loss_num": 0.02734375,
+      "loss_xval": 1.5505142211914062,
+      "num_input_tokens_seen": 661026400,
+      "step": 3824
+    },
+    {
+      "epoch": 1.4649559555725775,
+      "grad_norm": 120.89378195695188,
+      "learning_rate": 5e-06,
+      "loss": 1.8561,
+      "num_input_tokens_seen": 661199488,
+      "step": 3825
+    },
+    {
+      "epoch": 1.4649559555725775,
+      "loss": 1.9418671131134033,
+      "loss_ce": 0.18025416135787964,
+      "loss_iou": 0.7908101081848145,
+      "loss_num": 0.035888671875,
+      "loss_xval": 1.761612892150879,
+      "num_input_tokens_seen": 661199488,
+      "step": 3825
+    },
+    {
+      "epoch": 1.4653389505936423,
+      "grad_norm": 122.12627774650512,
+      "learning_rate": 5e-06,
+      "loss": 1.3821,
+      "num_input_tokens_seen": 661372544,
+      "step": 3826
+    },
+    {
+      "epoch": 1.4653389505936423,
+      "loss": 1.5193982124328613,
+      "loss_ce": 0.1425769031047821,
+      "loss_iou": 0.6450299024581909,
+      "loss_num": 0.017333984375,
+      "loss_xval": 1.3768212795257568,
+      "num_input_tokens_seen": 661372544,
+      "step": 3826
+    },
+    {
+      "epoch": 1.465721945614707,
+      "grad_norm": 230.93679128858588,
+      "learning_rate": 5e-06,
+      "loss": 2.1023,
+      "num_input_tokens_seen": 661545160,
+      "step": 3827
+    },
+    {
+      "epoch": 1.465721945614707,
+      "loss": 2.205559253692627,
+      "loss_ce": 0.15959420800209045,
+      "loss_iou": 0.9437894225120544,
+      "loss_num": 0.03173828125,
+      "loss_xval": 2.0459651947021484,
+      "num_input_tokens_seen": 661545160,
+      "step": 3827
+    },
+    {
+      "epoch": 1.4661049406357718,
+      "grad_norm": 77.35587336059181,
+      "learning_rate": 5e-06,
+      "loss": 1.7795,
+      "num_input_tokens_seen": 661718152,
+      "step": 3828
+    },
+    {
+      "epoch": 1.4661049406357718,
+      "loss": 1.8080837726593018,
+      "loss_ce": 0.18644201755523682,
+      "loss_iou": 0.7486259937286377,
+      "loss_num": 0.02490234375,
+      "loss_xval": 1.6216416358947754,
+      "num_input_tokens_seen": 661718152,
+      "step": 3828
+    },
+    {
+      "epoch": 1.4664879356568363,
+      "grad_norm": 144.25956094759215,
+      "learning_rate": 5e-06,
+      "loss": 1.4247,
+      "num_input_tokens_seen": 661891280,
+      "step": 3829
+    },
+    {
+      "epoch": 1.4664879356568363,
+      "loss": 1.2728171348571777,
+      "loss_ce": 0.15106286108493805,
+      "loss_iou": 0.5125526189804077,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.1217544078826904,
+      "num_input_tokens_seen": 661891280,
+      "step": 3829
+    },
+    {
+      "epoch": 1.466870930677901,
+      "grad_norm": 170.3971935136544,
+      "learning_rate": 5e-06,
+      "loss": 1.6923,
+      "num_input_tokens_seen": 662064112,
+      "step": 3830
+    },
+    {
+      "epoch": 1.466870930677901,
+      "loss": 1.7541427612304688,
+      "loss_ce": 0.1572430282831192,
+      "loss_iou": 0.7572511434555054,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 1.5968997478485107,
+      "num_input_tokens_seen": 662064112,
+      "step": 3830
+    },
+    {
+      "epoch": 1.4672539256989658,
+      "grad_norm": 97.01879089579793,
+      "learning_rate": 5e-06,
+      "loss": 1.9848,
+      "num_input_tokens_seen": 662237120,
+      "step": 3831
+    },
+    {
+      "epoch": 1.4672539256989658,
+      "loss": 1.905449390411377,
+      "loss_ce": 0.20804432034492493,
+      "loss_iou": 0.7706385254859924,
+      "loss_num": 0.03125,
+      "loss_xval": 1.6974049806594849,
+      "num_input_tokens_seen": 662237120,
+      "step": 3831
+    },
+    {
+      "epoch": 1.4676369207200306,
+      "grad_norm": 78.1377968147627,
+      "learning_rate": 5e-06,
+      "loss": 1.5092,
+      "num_input_tokens_seen": 662410112,
+      "step": 3832
+    },
+    {
+      "epoch": 1.4676369207200306,
+      "loss": 1.3702943325042725,
+      "loss_ce": 0.1734234243631363,
+      "loss_iou": 0.5395823121070862,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 1.1968708038330078,
+      "num_input_tokens_seen": 662410112,
+      "step": 3832
+    },
+    {
+      "epoch": 1.4680199157410954,
+      "grad_norm": 138.67596927650027,
+      "learning_rate": 5e-06,
+      "loss": 1.4924,
+      "num_input_tokens_seen": 662583464,
+      "step": 3833
+    },
+    {
+      "epoch": 1.4680199157410954,
+      "loss": 1.5313942432403564,
+      "loss_ce": 0.1910170614719391,
+      "loss_iou": 0.6064525842666626,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 1.3403772115707397,
+      "num_input_tokens_seen": 662583464,
+      "step": 3833
+    },
+    {
+      "epoch": 1.4684029107621601,
+      "grad_norm": 94.77838700294897,
+      "learning_rate": 5e-06,
+      "loss": 1.7593,
+      "num_input_tokens_seen": 662753336,
+      "step": 3834
+    },
+    {
+      "epoch": 1.4684029107621601,
+      "loss": 1.6682029962539673,
+      "loss_ce": 0.18056021630764008,
+      "loss_iou": 0.692002534866333,
+      "loss_num": 0.020751953125,
+      "loss_xval": 1.487642765045166,
+      "num_input_tokens_seen": 662753336,
+      "step": 3834
+    },
+    {
+      "epoch": 1.4687859057832249,
+      "grad_norm": 140.78705327462356,
+      "learning_rate": 5e-06,
+      "loss": 1.3048,
+      "num_input_tokens_seen": 662926400,
+      "step": 3835
+    },
+    {
+      "epoch": 1.4687859057832249,
+      "loss": 1.4433050155639648,
+      "loss_ce": 0.20680516958236694,
+      "loss_iou": 0.5778599977493286,
+      "loss_num": 0.01611328125,
+      "loss_xval": 1.2365000247955322,
+      "num_input_tokens_seen": 662926400,
+      "step": 3835
+    },
+    {
+      "epoch": 1.4691689008042896,
+      "grad_norm": 162.05062889591497,
+      "learning_rate": 5e-06,
+      "loss": 1.7313,
+      "num_input_tokens_seen": 663099248,
+      "step": 3836
+    },
+    {
+      "epoch": 1.4691689008042896,
+      "loss": 1.8132050037384033,
+      "loss_ce": 0.17610056698322296,
+      "loss_iou": 0.7324011325836182,
+      "loss_num": 0.034423828125,
+      "loss_xval": 1.6371045112609863,
+      "num_input_tokens_seen": 663099248,
+      "step": 3836
+    },
+    {
+      "epoch": 1.4695518958253544,
+      "grad_norm": 114.63206588501914,
+      "learning_rate": 5e-06,
+      "loss": 1.2958,
+      "num_input_tokens_seen": 663272248,
+      "step": 3837
+    },
+    {
+      "epoch": 1.4695518958253544,
+      "loss": 1.2157354354858398,
+      "loss_ce": 0.18062147498130798,
+      "loss_iou": 0.4794405400753021,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 1.035114049911499,
+      "num_input_tokens_seen": 663272248,
+      "step": 3837
+    },
+    {
+      "epoch": 1.469934890846419,
+      "grad_norm": 143.54249613964905,
+      "learning_rate": 5e-06,
+      "loss": 1.4071,
+      "num_input_tokens_seen": 663445008,
+      "step": 3838
+    },
+    {
+      "epoch": 1.469934890846419,
+      "loss": 1.571700096130371,
+      "loss_ce": 0.15280543267726898,
+      "loss_iou": 0.6679586172103882,
+      "loss_num": 0.0166015625,
+      "loss_xval": 1.4188945293426514,
+      "num_input_tokens_seen": 663445008,
+      "step": 3838
+    },
+    {
+      "epoch": 1.4703178858674837,
+      "grad_norm": 119.08983301678559,
+      "learning_rate": 5e-06,
+      "loss": 1.3561,
+      "num_input_tokens_seen": 663617696,
+      "step": 3839
+    },
+    {
+      "epoch": 1.4703178858674837,
+      "loss": 1.3493022918701172,
+      "loss_ce": 0.16583970189094543,
+      "loss_iou": 0.5522873401641846,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 1.1834626197814941,
+      "num_input_tokens_seen": 663617696,
+      "step": 3839
+    },
+    {
+      "epoch": 1.4707008808885484,
+      "grad_norm": 335.4058954399076,
+      "learning_rate": 5e-06,
+      "loss": 1.6527,
+      "num_input_tokens_seen": 663790488,
+      "step": 3840
+    },
+    {
+      "epoch": 1.4707008808885484,
+      "loss": 1.5318005084991455,
+      "loss_ce": 0.1281571388244629,
+      "loss_iou": 0.6288542151451111,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 1.4036434888839722,
+      "num_input_tokens_seen": 663790488,
+      "step": 3840
+    },
+    {
+      "epoch": 1.4710838759096132,
+      "grad_norm": 187.15833963722213,
+      "learning_rate": 5e-06,
+      "loss": 1.8186,
+      "num_input_tokens_seen": 663963648,
+      "step": 3841
+    },
+    {
+      "epoch": 1.4710838759096132,
+      "loss": 1.7523595094680786,
+      "loss_ce": 0.1846073418855667,
+      "loss_iou": 0.7219864130020142,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 1.5677522420883179,
+      "num_input_tokens_seen": 663963648,
+      "step": 3841
+    },
+    {
+      "epoch": 1.471466870930678,
+      "grad_norm": 276.0166430233373,
+      "learning_rate": 5e-06,
+      "loss": 2.0253,
+      "num_input_tokens_seen": 664136872,
+      "step": 3842
+    },
+    {
+      "epoch": 1.471466870930678,
+      "loss": 2.061936616897583,
+      "loss_ce": 0.1929805725812912,
+      "loss_iou": 0.8287040591239929,
+      "loss_num": 0.042236328125,
+      "loss_xval": 1.8689559698104858,
+      "num_input_tokens_seen": 664136872,
+      "step": 3842
+    },
+    {
+      "epoch": 1.4718498659517425,
+      "grad_norm": 182.32725780040136,
+      "learning_rate": 5e-06,
+      "loss": 1.4631,
+      "num_input_tokens_seen": 664310000,
+      "step": 3843
+    },
+    {
+      "epoch": 1.4718498659517425,
+      "loss": 1.393494963645935,
+      "loss_ce": 0.22394207119941711,
+      "loss_iou": 0.5386490821838379,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 1.1695528030395508,
+      "num_input_tokens_seen": 664310000,
+      "step": 3843
+    },
+    {
+      "epoch": 1.4722328609728073,
+      "grad_norm": 325.14936297944143,
+      "learning_rate": 5e-06,
+      "loss": 1.7968,
+      "num_input_tokens_seen": 664483112,
+      "step": 3844
+    },
+    {
+      "epoch": 1.4722328609728073,
+      "loss": 1.7259163856506348,
+      "loss_ce": 0.16397422552108765,
+      "loss_iou": 0.7327837944030762,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.5619421005249023,
+      "num_input_tokens_seen": 664483112,
+      "step": 3844
+    },
+    {
+      "epoch": 1.472615855993872,
+      "grad_norm": 154.19209840188947,
+      "learning_rate": 5e-06,
+      "loss": 1.8539,
+      "num_input_tokens_seen": 664656376,
+      "step": 3845
+    },
+    {
+      "epoch": 1.472615855993872,
+      "loss": 1.9076762199401855,
+      "loss_ce": 0.20823287963867188,
+      "loss_iou": 0.7454125881195068,
+      "loss_num": 0.041748046875,
+      "loss_xval": 1.6994433403015137,
+      "num_input_tokens_seen": 664656376,
+      "step": 3845
+    },
+    {
+      "epoch": 1.4729988510149368,
+      "grad_norm": 167.35411849209817,
+      "learning_rate": 5e-06,
+      "loss": 1.4539,
+      "num_input_tokens_seen": 664829280,
+      "step": 3846
+    },
+    {
+      "epoch": 1.4729988510149368,
+      "loss": 1.5264216661453247,
+      "loss_ce": 0.1563391089439392,
+      "loss_iou": 0.6442545056343079,
+      "loss_num": 0.016357421875,
+      "loss_xval": 1.3700824975967407,
+      "num_input_tokens_seen": 664829280,
+      "step": 3846
+    },
+    {
+      "epoch": 1.4733818460360015,
+      "grad_norm": 146.77845846627713,
+      "learning_rate": 5e-06,
+      "loss": 2.0207,
+      "num_input_tokens_seen": 665002352,
+      "step": 3847
+    },
+    {
+      "epoch": 1.4733818460360015,
+      "loss": 2.0562148094177246,
+      "loss_ce": 0.21677184104919434,
+      "loss_iou": 0.8762645125389099,
+      "loss_num": 0.017333984375,
+      "loss_xval": 1.8394430875778198,
+      "num_input_tokens_seen": 665002352,
+      "step": 3847
+    },
+    {
+      "epoch": 1.4737648410570663,
+      "grad_norm": 93.92841077468523,
+      "learning_rate": 5e-06,
+      "loss": 1.8418,
+      "num_input_tokens_seen": 665174992,
+      "step": 3848
+    },
+    {
+      "epoch": 1.4737648410570663,
+      "loss": 1.8205523490905762,
+      "loss_ce": 0.14205557107925415,
+      "loss_iou": 0.7516324520111084,
+      "loss_num": 0.03515625,
+      "loss_xval": 1.6784968376159668,
+      "num_input_tokens_seen": 665174992,
+      "step": 3848
+    },
+    {
+      "epoch": 1.474147836078131,
+      "grad_norm": 157.33118163806148,
+      "learning_rate": 5e-06,
+      "loss": 1.5346,
+      "num_input_tokens_seen": 665348536,
+      "step": 3849
+    },
+    {
+      "epoch": 1.474147836078131,
+      "loss": 1.5996110439300537,
+      "loss_ce": 0.25675928592681885,
+      "loss_iou": 0.6246882677078247,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 1.3428518772125244,
+      "num_input_tokens_seen": 665348536,
+      "step": 3849
+    },
+    {
+      "epoch": 1.4745308310991958,
+      "grad_norm": 113.91028531435974,
+      "learning_rate": 5e-06,
+      "loss": 1.7914,
+      "num_input_tokens_seen": 665521480,
+      "step": 3850
+    },
+    {
+      "epoch": 1.4745308310991958,
+      "loss": 1.7607181072235107,
+      "loss_ce": 0.19359637796878815,
+      "loss_iou": 0.7268286943435669,
+      "loss_num": 0.022705078125,
+      "loss_xval": 1.5671217441558838,
+      "num_input_tokens_seen": 665521480,
+      "step": 3850
+    },
+    {
+      "epoch": 1.4749138261202606,
+      "grad_norm": 126.63833983812442,
+      "learning_rate": 5e-06,
+      "loss": 1.667,
+      "num_input_tokens_seen": 665694456,
+      "step": 3851
+    },
+    {
+      "epoch": 1.4749138261202606,
+      "loss": 1.6462596654891968,
+      "loss_ce": 0.22160466015338898,
+      "loss_iou": 0.6712127327919006,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 1.4246550798416138,
+      "num_input_tokens_seen": 665694456,
+      "step": 3851
+    },
+    {
+      "epoch": 1.475296821141325,
+      "grad_norm": 127.9277838472454,
+      "learning_rate": 5e-06,
+      "loss": 1.6844,
+      "num_input_tokens_seen": 665867328,
+      "step": 3852
+    },
+    {
+      "epoch": 1.475296821141325,
+      "loss": 1.7342861890792847,
+      "loss_ce": 0.16108569502830505,
+      "loss_iou": 0.7260838747024536,
+      "loss_num": 0.024169921875,
+      "loss_xval": 1.5732004642486572,
+      "num_input_tokens_seen": 665867328,
+      "step": 3852
+    },
+    {
+      "epoch": 1.4756798161623899,
+      "grad_norm": 82.94437218184599,
+      "learning_rate": 5e-06,
+      "loss": 1.4305,
+      "num_input_tokens_seen": 666040384,
+      "step": 3853
+    },
+    {
+      "epoch": 1.4756798161623899,
+      "loss": 1.351753830909729,
+      "loss_ce": 0.1556341052055359,
+      "loss_iou": 0.537299394607544,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 1.196119785308838,
+      "num_input_tokens_seen": 666040384,
+      "step": 3853
+    },
+    {
+      "epoch": 1.4760628111834546,
+      "grad_norm": 125.65292723452029,
+      "learning_rate": 5e-06,
+      "loss": 1.4625,
+      "num_input_tokens_seen": 666213184,
+      "step": 3854
+    },
+    {
+      "epoch": 1.4760628111834546,
+      "loss": 1.3487107753753662,
+      "loss_ce": 0.14079010486602783,
+      "loss_iou": 0.5578787326812744,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 1.2079205513000488,
+      "num_input_tokens_seen": 666213184,
+      "step": 3854
+    },
+    {
+      "epoch": 1.4764458062045194,
+      "grad_norm": 131.79253341079308,
+      "learning_rate": 5e-06,
+      "loss": 1.6839,
+      "num_input_tokens_seen": 666386208,
+      "step": 3855
+    },
+    {
+      "epoch": 1.4764458062045194,
+      "loss": 1.6640814542770386,
+      "loss_ce": 0.19890397787094116,
+      "loss_iou": 0.6766195297241211,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 1.4651775360107422,
+      "num_input_tokens_seen": 666386208,
+      "step": 3855
+    },
+    {
+      "epoch": 1.4768288012255841,
+      "grad_norm": 136.18593025029074,
+      "learning_rate": 5e-06,
+      "loss": 1.6107,
+      "num_input_tokens_seen": 666559128,
+      "step": 3856
+    },
+    {
+      "epoch": 1.4768288012255841,
+      "loss": 1.6052563190460205,
+      "loss_ce": 0.17612627148628235,
+      "loss_iou": 0.6662404537200928,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.4291300773620605,
+      "num_input_tokens_seen": 666559128,
+      "step": 3856
+    },
+    {
+      "epoch": 1.4772117962466487,
+      "grad_norm": 150.1747582565557,
+      "learning_rate": 5e-06,
+      "loss": 1.6752,
+      "num_input_tokens_seen": 666731936,
+      "step": 3857
+    },
+    {
+      "epoch": 1.4772117962466487,
+      "loss": 1.6757440567016602,
+      "loss_ce": 0.1703186333179474,
+      "loss_iou": 0.6937527656555176,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 1.5054254531860352,
+      "num_input_tokens_seen": 666731936,
+      "step": 3857
+    },
+    {
+      "epoch": 1.4775947912677134,
+      "grad_norm": 114.09954680579511,
+      "learning_rate": 5e-06,
+      "loss": 1.8521,
+      "num_input_tokens_seen": 666904496,
+      "step": 3858
+    },
+    {
+      "epoch": 1.4775947912677134,
+      "loss": 1.7862820625305176,
+      "loss_ce": 0.17286039888858795,
+      "loss_iou": 0.7237640619277954,
+      "loss_num": 0.033203125,
+      "loss_xval": 1.6134216785430908,
+      "num_input_tokens_seen": 666904496,
+      "step": 3858
+    },
+    {
+      "epoch": 1.4779777862887782,
+      "grad_norm": 169.8345223772268,
+      "learning_rate": 5e-06,
+      "loss": 1.3313,
+      "num_input_tokens_seen": 667077768,
+      "step": 3859
+    },
+    {
+      "epoch": 1.4779777862887782,
+      "loss": 1.3333956003189087,
+      "loss_ce": 0.1704695224761963,
+      "loss_iou": 0.5455590486526489,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 1.1629259586334229,
+      "num_input_tokens_seen": 667077768,
+      "step": 3859
+    },
+    {
+      "epoch": 1.478360781309843,
+      "grad_norm": 147.86795164609876,
+      "learning_rate": 5e-06,
+      "loss": 1.6171,
+      "num_input_tokens_seen": 667250776,
+      "step": 3860
+    },
+    {
+      "epoch": 1.478360781309843,
+      "loss": 1.5118118524551392,
+      "loss_ce": 0.19477102160453796,
+      "loss_iou": 0.5991026759147644,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 1.3170408010482788,
+      "num_input_tokens_seen": 667250776,
+      "step": 3860
+    },
+    {
+      "epoch": 1.4787437763309077,
+      "grad_norm": 102.01628569584015,
+      "learning_rate": 5e-06,
+      "loss": 1.2921,
+      "num_input_tokens_seen": 667423832,
+      "step": 3861
+    },
+    {
+      "epoch": 1.4787437763309077,
+      "loss": 1.2340381145477295,
+      "loss_ce": 0.16711857914924622,
+      "loss_iou": 0.49796777963638306,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 1.0669194459915161,
+      "num_input_tokens_seen": 667423832,
+      "step": 3861
+    },
+    {
+      "epoch": 1.4791267713519725,
+      "grad_norm": 104.32723047255132,
+      "learning_rate": 5e-06,
+      "loss": 1.5295,
+      "num_input_tokens_seen": 667596472,
+      "step": 3862
+    },
+    {
+      "epoch": 1.4791267713519725,
+      "loss": 1.3800314664840698,
+      "loss_ce": 0.16500824689865112,
+      "loss_iou": 0.5674268007278442,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 1.2150232791900635,
+      "num_input_tokens_seen": 667596472,
+      "step": 3862
+    },
+    {
+      "epoch": 1.4795097663730372,
+      "grad_norm": 132.21034295835113,
+      "learning_rate": 5e-06,
+      "loss": 1.4402,
+      "num_input_tokens_seen": 667769472,
+      "step": 3863
+    },
+    {
+      "epoch": 1.4795097663730372,
+      "loss": 1.5435000658035278,
+      "loss_ce": 0.1579607129096985,
+      "loss_iou": 0.6303611993789673,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 1.3855392932891846,
+      "num_input_tokens_seen": 667769472,
+      "step": 3863
+    },
+    {
+      "epoch": 1.479892761394102,
+      "grad_norm": 184.56186037129535,
+      "learning_rate": 5e-06,
+      "loss": 1.6246,
+      "num_input_tokens_seen": 667942584,
+      "step": 3864
+    },
+    {
+      "epoch": 1.479892761394102,
+      "loss": 1.652564287185669,
+      "loss_ce": 0.18969173729419708,
+      "loss_iou": 0.6824555397033691,
+      "loss_num": 0.01953125,
+      "loss_xval": 1.4628725051879883,
+      "num_input_tokens_seen": 667942584,
+      "step": 3864
+    },
+    {
+      "epoch": 1.4802757564151667,
+      "grad_norm": 170.45213369292847,
+      "learning_rate": 5e-06,
+      "loss": 1.5748,
+      "num_input_tokens_seen": 668115696,
+      "step": 3865
+    },
+    {
+      "epoch": 1.4802757564151667,
+      "loss": 1.5227084159851074,
+      "loss_ce": 0.19459585845470428,
+      "loss_iou": 0.6224150657653809,
+      "loss_num": 0.0166015625,
+      "loss_xval": 1.3281126022338867,
+      "num_input_tokens_seen": 668115696,
+      "step": 3865
+    },
+    {
+      "epoch": 1.4806587514362313,
+      "grad_norm": 103.4924956997599,
+      "learning_rate": 5e-06,
+      "loss": 1.6014,
+      "num_input_tokens_seen": 668288672,
+      "step": 3866
+    },
+    {
+      "epoch": 1.4806587514362313,
+      "loss": 1.5472420454025269,
+      "loss_ce": 0.2081267535686493,
+      "loss_iou": 0.6122151613235474,
+      "loss_num": 0.02294921875,
+      "loss_xval": 1.3391153812408447,
+      "num_input_tokens_seen": 668288672,
+      "step": 3866
+    },
+    {
+      "epoch": 1.481041746457296,
+      "grad_norm": 175.38685505076648,
+      "learning_rate": 5e-06,
+      "loss": 1.7048,
+      "num_input_tokens_seen": 668461584,
+      "step": 3867
+    },
+    {
+      "epoch": 1.481041746457296,
+      "loss": 1.6872774362564087,
+      "loss_ce": 0.1981140375137329,
+      "loss_iou": 0.6807389259338379,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 1.4891633987426758,
+      "num_input_tokens_seen": 668461584,
+      "step": 3867
+    },
+    {
+      "epoch": 1.4814247414783608,
+      "grad_norm": 136.4284280379348,
+      "learning_rate": 5e-06,
+      "loss": 1.7501,
+      "num_input_tokens_seen": 668634728,
+      "step": 3868
+    },
+    {
+      "epoch": 1.4814247414783608,
+      "loss": 1.8383429050445557,
+      "loss_ce": 0.13189396262168884,
+      "loss_iou": 0.7788836359977722,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 1.7064489126205444,
+      "num_input_tokens_seen": 668634728,
+      "step": 3868
+    },
+    {
+      "epoch": 1.4818077364994255,
+      "grad_norm": 132.9321594877913,
+      "learning_rate": 5e-06,
+      "loss": 1.372,
+      "num_input_tokens_seen": 668808128,
+      "step": 3869
+    },
+    {
+      "epoch": 1.4818077364994255,
+      "loss": 1.389357328414917,
+      "loss_ce": 0.19434954226016998,
+      "loss_iou": 0.5608369708061218,
+      "loss_num": 0.0146484375,
+      "loss_xval": 1.1950076818466187,
+      "num_input_tokens_seen": 668808128,
+      "step": 3869
+    },
+    {
+      "epoch": 1.4821907315204903,
+      "grad_norm": 188.2283810651053,
+      "learning_rate": 5e-06,
+      "loss": 1.4836,
+      "num_input_tokens_seen": 668980768,
+      "step": 3870
+    },
+    {
+      "epoch": 1.4821907315204903,
+      "loss": 1.577918291091919,
+      "loss_ce": 0.14871646463871002,
+      "loss_iou": 0.6660932302474976,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 1.4292018413543701,
+      "num_input_tokens_seen": 668980768,
+      "step": 3870
+    },
+    {
+      "epoch": 1.4825737265415548,
+      "grad_norm": 141.23940856964686,
+      "learning_rate": 5e-06,
+      "loss": 1.4481,
+      "num_input_tokens_seen": 669153744,
+      "step": 3871
+    },
+    {
+      "epoch": 1.4825737265415548,
+      "loss": 1.255262851715088,
+      "loss_ce": 0.15619684755802155,
+      "loss_iou": 0.5179625153541565,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 1.099065899848938,
+      "num_input_tokens_seen": 669153744,
+      "step": 3871
+    },
+    {
+      "epoch": 1.4829567215626196,
+      "grad_norm": 90.8889940085996,
+      "learning_rate": 5e-06,
+      "loss": 1.6327,
+      "num_input_tokens_seen": 669326544,
+      "step": 3872
+    },
+    {
+      "epoch": 1.4829567215626196,
+      "loss": 1.4989891052246094,
+      "loss_ce": 0.19657351076602936,
+      "loss_iou": 0.5932244062423706,
+      "loss_num": 0.023193359375,
+      "loss_xval": 1.3024156093597412,
+      "num_input_tokens_seen": 669326544,
+      "step": 3872
+    },
+    {
+      "epoch": 1.4833397165836844,
+      "grad_norm": 106.92784849994153,
+      "learning_rate": 5e-06,
+      "loss": 1.199,
+      "num_input_tokens_seen": 669499320,
+      "step": 3873
+    },
+    {
+      "epoch": 1.4833397165836844,
+      "loss": 1.090671181678772,
+      "loss_ce": 0.17284780740737915,
+      "loss_iou": 0.424091100692749,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.917823314666748,
+      "num_input_tokens_seen": 669499320,
+      "step": 3873
+    },
+    {
+      "epoch": 1.4837227116047491,
+      "grad_norm": 119.3902921505456,
+      "learning_rate": 5e-06,
+      "loss": 1.6888,
+      "num_input_tokens_seen": 669672344,
+      "step": 3874
+    },
+    {
+      "epoch": 1.4837227116047491,
+      "loss": 1.724827766418457,
+      "loss_ce": 0.1560395061969757,
+      "loss_iou": 0.7323310375213623,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 1.5687880516052246,
+      "num_input_tokens_seen": 669672344,
+      "step": 3874
+    },
+    {
+      "epoch": 1.4841057066258139,
+      "grad_norm": 141.76058799087716,
+      "learning_rate": 5e-06,
+      "loss": 1.6218,
+      "num_input_tokens_seen": 669845368,
+      "step": 3875
+    },
+    {
+      "epoch": 1.4841057066258139,
+      "loss": 1.6312702894210815,
+      "loss_ce": 0.17275990545749664,
+      "loss_iou": 0.6746134757995605,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 1.458510398864746,
+      "num_input_tokens_seen": 669845368,
+      "step": 3875
+    },
+    {
+      "epoch": 1.4844887016468786,
+      "grad_norm": 129.6314489565196,
+      "learning_rate": 5e-06,
+      "loss": 1.5693,
+      "num_input_tokens_seen": 670018440,
+      "step": 3876
+    },
+    {
+      "epoch": 1.4844887016468786,
+      "loss": 1.6121318340301514,
+      "loss_ce": 0.21013782918453217,
+      "loss_iou": 0.6578909754753113,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 1.4019941091537476,
+      "num_input_tokens_seen": 670018440,
+      "step": 3876
+    },
+    {
+      "epoch": 1.4848716966679434,
+      "grad_norm": 197.97396657663876,
+      "learning_rate": 5e-06,
+      "loss": 1.6026,
+      "num_input_tokens_seen": 670191504,
+      "step": 3877
+    },
+    {
+      "epoch": 1.4848716966679434,
+      "loss": 1.46147620677948,
+      "loss_ce": 0.1412167251110077,
+      "loss_iou": 0.6187021136283875,
+      "loss_num": 0.0166015625,
+      "loss_xval": 1.32025945186615,
+      "num_input_tokens_seen": 670191504,
+      "step": 3877
+    },
+    {
+      "epoch": 1.4852546916890081,
+      "grad_norm": 107.1568237957383,
+      "learning_rate": 5e-06,
+      "loss": 1.6205,
+      "num_input_tokens_seen": 670364448,
+      "step": 3878
+    },
+    {
+      "epoch": 1.4852546916890081,
+      "loss": 1.6485404968261719,
+      "loss_ce": 0.18954724073410034,
+      "loss_iou": 0.6764570474624634,
+      "loss_num": 0.021240234375,
+      "loss_xval": 1.4589931964874268,
+      "num_input_tokens_seen": 670364448,
+      "step": 3878
+    },
+    {
+      "epoch": 1.485637686710073,
+      "grad_norm": 131.64257329031273,
+      "learning_rate": 5e-06,
+      "loss": 1.325,
+      "num_input_tokens_seen": 670537432,
+      "step": 3879
+    },
+    {
+      "epoch": 1.485637686710073,
+      "loss": 1.3019726276397705,
+      "loss_ce": 0.15859565138816833,
+      "loss_iou": 0.5418881177902222,
+      "loss_num": 0.011962890625,
+      "loss_xval": 1.1433770656585693,
+      "num_input_tokens_seen": 670537432,
+      "step": 3879
+    },
+    {
+      "epoch": 1.4860206817311374,
+      "grad_norm": 95.66175125333709,
+      "learning_rate": 5e-06,
+      "loss": 1.5738,
+      "num_input_tokens_seen": 670710448,
+      "step": 3880
+    },
+    {
+      "epoch": 1.4860206817311374,
+      "loss": 1.641021490097046,
+      "loss_ce": 0.14951558411121368,
+      "loss_iou": 0.6779733896255493,
+      "loss_num": 0.027099609375,
+      "loss_xval": 1.4915058612823486,
+      "num_input_tokens_seen": 670710448,
+      "step": 3880
+    },
+    {
+      "epoch": 1.4864036767522022,
+      "grad_norm": 116.52670753883206,
+      "learning_rate": 5e-06,
+      "loss": 1.3799,
+      "num_input_tokens_seen": 670883072,
+      "step": 3881
+    },
+    {
+      "epoch": 1.4864036767522022,
+      "loss": 1.361303687095642,
+      "loss_ce": 0.15120264887809753,
+      "loss_iou": 0.5624327063560486,
+      "loss_num": 0.01708984375,
+      "loss_xval": 1.2101010084152222,
+      "num_input_tokens_seen": 670883072,
+      "step": 3881
+    },
+    {
+      "epoch": 1.486786671773267,
+      "grad_norm": 324.4629411639793,
+      "learning_rate": 5e-06,
+      "loss": 1.8486,
+      "num_input_tokens_seen": 671055992,
+      "step": 3882
+    },
+    {
+      "epoch": 1.486786671773267,
+      "loss": 1.8590505123138428,
+      "loss_ce": 0.17376533150672913,
+      "loss_iou": 0.784445583820343,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 1.685285210609436,
+      "num_input_tokens_seen": 671055992,
+      "step": 3882
+    },
+    {
+      "epoch": 1.4871696667943317,
+      "grad_norm": 105.6266431676682,
+      "learning_rate": 5e-06,
+      "loss": 1.8311,
+      "num_input_tokens_seen": 671229032,
+      "step": 3883
+    },
+    {
+      "epoch": 1.4871696667943317,
+      "loss": 1.7354176044464111,
+      "loss_ce": 0.1748410016298294,
+      "loss_iou": 0.7163845300674438,
+      "loss_num": 0.025634765625,
+      "loss_xval": 1.5605766773223877,
+      "num_input_tokens_seen": 671229032,
+      "step": 3883
+    },
+    {
+      "epoch": 1.4875526618153965,
+      "grad_norm": 264.2298469536217,
+      "learning_rate": 5e-06,
+      "loss": 1.7788,
+      "num_input_tokens_seen": 671402112,
+      "step": 3884
+    },
+    {
+      "epoch": 1.4875526618153965,
+      "loss": 1.6878291368484497,
+      "loss_ce": 0.22226132452487946,
+      "loss_iou": 0.6703296899795532,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 1.4655678272247314,
+      "num_input_tokens_seen": 671402112,
+      "step": 3884
+    },
+    {
+      "epoch": 1.487935656836461,
+      "grad_norm": 225.60078800531593,
+      "learning_rate": 5e-06,
+      "loss": 1.6987,
+      "num_input_tokens_seen": 671575032,
+      "step": 3885
+    },
+    {
+      "epoch": 1.487935656836461,
+      "loss": 1.4834569692611694,
+      "loss_ce": 0.17097559571266174,
+      "loss_iou": 0.60823655128479,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 1.31248140335083,
+      "num_input_tokens_seen": 671575032,
+      "step": 3885
+    },
+    {
+      "epoch": 1.4883186518575258,
+      "grad_norm": 167.5257703107799,
+      "learning_rate": 5e-06,
+      "loss": 2.206,
+      "num_input_tokens_seen": 671747976,
+      "step": 3886
+    },
+    {
+      "epoch": 1.4883186518575258,
+      "loss": 2.038886070251465,
+      "loss_ce": 0.16198047995567322,
+      "loss_iou": 0.8290777206420898,
+      "loss_num": 0.043701171875,
+      "loss_xval": 1.8769054412841797,
+      "num_input_tokens_seen": 671747976,
+      "step": 3886
+    },
+    {
+      "epoch": 1.4887016468785905,
+      "grad_norm": 86.88630888341918,
+      "learning_rate": 5e-06,
+      "loss": 1.5637,
+      "num_input_tokens_seen": 671920688,
+      "step": 3887
+    },
+    {
+      "epoch": 1.4887016468785905,
+      "loss": 1.4317007064819336,
+      "loss_ce": 0.14628341794013977,
+      "loss_iou": 0.5837181806564331,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 1.2731492519378662,
+      "num_input_tokens_seen": 671920688,
+      "step": 3887
+    },
+    {
+      "epoch": 1.4890846418996553,
+      "grad_norm": 158.54654743374675,
+      "learning_rate": 5e-06,
+      "loss": 1.408,
+      "num_input_tokens_seen": 672093936,
+      "step": 3888
+    },
+    {
+      "epoch": 1.4890846418996553,
+      "loss": 1.4867074489593506,
+      "loss_ce": 0.16030298173427582,
+      "loss_iou": 0.6173266768455505,
+      "loss_num": 0.018310546875,
+      "loss_xval": 1.3264044523239136,
+      "num_input_tokens_seen": 672093936,
+      "step": 3888
+    },
+    {
+      "epoch": 1.48946763692072,
+      "grad_norm": 125.28459812812459,
+      "learning_rate": 5e-06,
+      "loss": 1.4976,
+      "num_input_tokens_seen": 672267136,
+      "step": 3889
+    },
+    {
+      "epoch": 1.48946763692072,
+      "loss": 1.3624978065490723,
+      "loss_ce": 0.17049774527549744,
+      "loss_iou": 0.555045485496521,
+      "loss_num": 0.016357421875,
+      "loss_xval": 1.192000150680542,
+      "num_input_tokens_seen": 672267136,
+      "step": 3889
+    },
+    {
+      "epoch": 1.4898506319417848,
+      "grad_norm": 149.58744976740113,
+      "learning_rate": 5e-06,
+      "loss": 1.372,
+      "num_input_tokens_seen": 672440328,
+      "step": 3890
+    },
+    {
+      "epoch": 1.4898506319417848,
+      "loss": 1.3499786853790283,
+      "loss_ce": 0.1211739107966423,
+      "loss_iou": 0.5589977502822876,
+      "loss_num": 0.022216796875,
+      "loss_xval": 1.2288048267364502,
+      "num_input_tokens_seen": 672440328,
+      "step": 3890
+    },
+    {
+      "epoch": 1.4902336269628496,
+      "grad_norm": 152.0818351162049,
+      "learning_rate": 5e-06,
+      "loss": 1.5785,
+      "num_input_tokens_seen": 672613368,
+      "step": 3891
+    },
+    {
+      "epoch": 1.4902336269628496,
+      "loss": 1.3897818326950073,
+      "loss_ce": 0.17107608914375305,
+      "loss_iou": 0.5633171200752258,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 1.2187057733535767,
+      "num_input_tokens_seen": 672613368,
+      "step": 3891
+    },
+    {
+      "epoch": 1.4906166219839143,
+      "grad_norm": 108.03528672017505,
+      "learning_rate": 5e-06,
+      "loss": 1.5662,
+      "num_input_tokens_seen": 672786296,
+      "step": 3892
+    },
+    {
+      "epoch": 1.4906166219839143,
+      "loss": 1.725993275642395,
+      "loss_ce": 0.17192400991916656,
+      "loss_iou": 0.7165030241012573,
+      "loss_num": 0.024169921875,
+      "loss_xval": 1.5540692806243896,
+      "num_input_tokens_seen": 672786296,
+      "step": 3892
+    },
+    {
+      "epoch": 1.490999617004979,
+      "grad_norm": 120.93625765029971,
+      "learning_rate": 5e-06,
+      "loss": 1.6048,
+      "num_input_tokens_seen": 672958984,
+      "step": 3893
+    },
+    {
+      "epoch": 1.490999617004979,
+      "loss": 1.6273632049560547,
+      "loss_ce": 0.20472627878189087,
+      "loss_iou": 0.6684107780456543,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 1.4226369857788086,
+      "num_input_tokens_seen": 672958984,
+      "step": 3893
+    },
+    {
+      "epoch": 1.4913826120260436,
+      "grad_norm": 152.1381935646056,
+      "learning_rate": 5e-06,
+      "loss": 1.0851,
+      "num_input_tokens_seen": 673132112,
+      "step": 3894
+    },
+    {
+      "epoch": 1.4913826120260436,
+      "loss": 0.9724470376968384,
+      "loss_ce": 0.15467657148838043,
+      "loss_iou": 0.3776809573173523,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.8177703619003296,
+      "num_input_tokens_seen": 673132112,
+      "step": 3894
+    },
+    {
+      "epoch": 1.4917656070471084,
+      "grad_norm": 119.91263870959762,
+      "learning_rate": 5e-06,
+      "loss": 1.7792,
+      "num_input_tokens_seen": 673304944,
+      "step": 3895
+    },
+    {
+      "epoch": 1.4917656070471084,
+      "loss": 1.760056972503662,
+      "loss_ce": 0.15842179954051971,
+      "loss_iou": 0.7236691117286682,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 1.6016350984573364,
+      "num_input_tokens_seen": 673304944,
+      "step": 3895
+    },
+    {
+      "epoch": 1.4921486020681731,
+      "grad_norm": 81.29026001799502,
+      "learning_rate": 5e-06,
+      "loss": 1.4743,
+      "num_input_tokens_seen": 673477536,
+      "step": 3896
+    },
+    {
+      "epoch": 1.4921486020681731,
+      "loss": 1.5240058898925781,
+      "loss_ce": 0.195234015583992,
+      "loss_iou": 0.6291077136993408,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 1.3287720680236816,
+      "num_input_tokens_seen": 673477536,
+      "step": 3896
+    },
+    {
+      "epoch": 1.4925315970892379,
+      "grad_norm": 153.74166088899742,
+      "learning_rate": 5e-06,
+      "loss": 1.6929,
+      "num_input_tokens_seen": 673650760,
+      "step": 3897
+    },
+    {
+      "epoch": 1.4925315970892379,
+      "loss": 1.6741621494293213,
+      "loss_ce": 0.16368725895881653,
+      "loss_iou": 0.7229956388473511,
+      "loss_num": 0.012939453125,
+      "loss_xval": 1.5104749202728271,
+      "num_input_tokens_seen": 673650760,
+      "step": 3897
+    },
+    {
+      "epoch": 1.4929145921103026,
+      "grad_norm": 133.10121478712492,
+      "learning_rate": 5e-06,
+      "loss": 1.7809,
+      "num_input_tokens_seen": 673823768,
+      "step": 3898
+    },
+    {
+      "epoch": 1.4929145921103026,
+      "loss": 1.89362633228302,
+      "loss_ce": 0.16937804222106934,
+      "loss_iou": 0.7788111567497253,
+      "loss_num": 0.033447265625,
+      "loss_xval": 1.7242482900619507,
+      "num_input_tokens_seen": 673823768,
+      "step": 3898
+    },
+    {
+      "epoch": 1.4932975871313672,
+      "grad_norm": 162.6035247166546,
+      "learning_rate": 5e-06,
+      "loss": 1.4498,
+      "num_input_tokens_seen": 673993712,
+      "step": 3899
+    },
+    {
+      "epoch": 1.4932975871313672,
+      "loss": 1.5272722244262695,
+      "loss_ce": 0.17096881568431854,
+      "loss_iou": 0.6098533868789673,
+      "loss_num": 0.02734375,
+      "loss_xval": 1.3563034534454346,
+      "num_input_tokens_seen": 673993712,
+      "step": 3899
+    },
+    {
+      "epoch": 1.493680582152432,
+      "grad_norm": 153.67429353808546,
+      "learning_rate": 5e-06,
+      "loss": 1.867,
+      "num_input_tokens_seen": 674166952,
+      "step": 3900
+    },
+    {
+      "epoch": 1.493680582152432,
+      "loss": 1.883059024810791,
+      "loss_ce": 0.20422764122486115,
+      "loss_iou": 0.7705985307693481,
+      "loss_num": 0.027587890625,
+      "loss_xval": 1.6788313388824463,
+      "num_input_tokens_seen": 674166952,
+      "step": 3900
+    },
+    {
+      "epoch": 1.4940635771734967,
+      "grad_norm": 60.57757712105375,
+      "learning_rate": 5e-06,
+      "loss": 1.6093,
+      "num_input_tokens_seen": 674340048,
+      "step": 3901
+    },
+    {
+      "epoch": 1.4940635771734967,
+      "loss": 1.677382469177246,
+      "loss_ce": 0.15197619795799255,
+      "loss_iou": 0.6813737750053406,
+      "loss_num": 0.032470703125,
+      "loss_xval": 1.5254062414169312,
+      "num_input_tokens_seen": 674340048,
+      "step": 3901
+    },
+    {
+      "epoch": 1.4944465721945615,
+      "grad_norm": 165.47416506905398,
+      "learning_rate": 5e-06,
+      "loss": 1.3134,
+      "num_input_tokens_seen": 674513152,
+      "step": 3902
+    },
+    {
+      "epoch": 1.4944465721945615,
+      "loss": 1.3706345558166504,
+      "loss_ce": 0.18077810108661652,
+      "loss_iou": 0.5623659491539001,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 1.1898564100265503,
+      "num_input_tokens_seen": 674513152,
+      "step": 3902
+    },
+    {
+      "epoch": 1.4948295672156262,
+      "grad_norm": 141.53108056287766,
+      "learning_rate": 5e-06,
+      "loss": 1.9451,
+      "num_input_tokens_seen": 674685936,
+      "step": 3903
+    },
+    {
+      "epoch": 1.4948295672156262,
+      "loss": 2.033285617828369,
+      "loss_ce": 0.15381646156311035,
+      "loss_iou": 0.8622809648513794,
+      "loss_num": 0.031005859375,
+      "loss_xval": 1.8794691562652588,
+      "num_input_tokens_seen": 674685936,
+      "step": 3903
+    },
+    {
+      "epoch": 1.495212562236691,
+      "grad_norm": 154.93278934688692,
+      "learning_rate": 5e-06,
+      "loss": 1.5626,
+      "num_input_tokens_seen": 674859344,
+      "step": 3904
+    },
+    {
+      "epoch": 1.495212562236691,
+      "loss": 1.5289535522460938,
+      "loss_ce": 0.13078534603118896,
+      "loss_iou": 0.6564816236495972,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 1.3981683254241943,
+      "num_input_tokens_seen": 674859344,
+      "step": 3904
+    },
+    {
+      "epoch": 1.4955955572577557,
+      "grad_norm": 199.77810636348377,
+      "learning_rate": 5e-06,
+      "loss": 1.3503,
+      "num_input_tokens_seen": 675032312,
+      "step": 3905
+    },
+    {
+      "epoch": 1.4955955572577557,
+      "loss": 1.3020457029342651,
+      "loss_ce": 0.2171669900417328,
+      "loss_iou": 0.51427161693573,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 1.08487868309021,
+      "num_input_tokens_seen": 675032312,
+      "step": 3905
+    },
+    {
+      "epoch": 1.4959785522788205,
+      "grad_norm": 120.0415211054411,
+      "learning_rate": 5e-06,
+      "loss": 1.3756,
+      "num_input_tokens_seen": 675205224,
+      "step": 3906
+    },
+    {
+      "epoch": 1.4959785522788205,
+      "loss": 1.4973325729370117,
+      "loss_ce": 0.1475808471441269,
+      "loss_iou": 0.6320596933364868,
+      "loss_num": 0.01708984375,
+      "loss_xval": 1.3497517108917236,
+      "num_input_tokens_seen": 675205224,
+      "step": 3906
+    },
+    {
+      "epoch": 1.496361547299885,
+      "grad_norm": 116.03800266511163,
+      "learning_rate": 5e-06,
+      "loss": 1.2903,
+      "num_input_tokens_seen": 675378000,
+      "step": 3907
+    },
+    {
+      "epoch": 1.496361547299885,
+      "loss": 1.3025251626968384,
+      "loss_ce": 0.2025635540485382,
+      "loss_iou": 0.5094839930534363,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 1.0999616384506226,
+      "num_input_tokens_seen": 675378000,
+      "step": 3907
+    },
+    {
+      "epoch": 1.4967445423209498,
+      "grad_norm": 124.87824766868395,
+      "learning_rate": 5e-06,
+      "loss": 1.4478,
+      "num_input_tokens_seen": 675551176,
+      "step": 3908
+    },
+    {
+      "epoch": 1.4967445423209498,
+      "loss": 1.451635479927063,
+      "loss_ce": 0.184466153383255,
+      "loss_iou": 0.5844513177871704,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 1.2671692371368408,
+      "num_input_tokens_seen": 675551176,
+      "step": 3908
+    },
+    {
+      "epoch": 1.4971275373420145,
+      "grad_norm": 99.83496962957669,
+      "learning_rate": 5e-06,
+      "loss": 1.3585,
+      "num_input_tokens_seen": 675724448,
+      "step": 3909
+    },
+    {
+      "epoch": 1.4971275373420145,
+      "loss": 1.4900596141815186,
+      "loss_ce": 0.1743171215057373,
+      "loss_iou": 0.6139869689941406,
+      "loss_num": 0.017578125,
+      "loss_xval": 1.3157424926757812,
+      "num_input_tokens_seen": 675724448,
+      "step": 3909
+    },
+    {
+      "epoch": 1.4975105323630793,
+      "grad_norm": 118.73129195967844,
+      "learning_rate": 5e-06,
+      "loss": 1.6089,
+      "num_input_tokens_seen": 675897048,
+      "step": 3910
+    },
+    {
+      "epoch": 1.4975105323630793,
+      "loss": 1.420870065689087,
+      "loss_ce": 0.1739872694015503,
+      "loss_iou": 0.5579507350921631,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 1.2468829154968262,
+      "num_input_tokens_seen": 675897048,
+      "step": 3910
+    },
+    {
+      "epoch": 1.497893527384144,
+      "grad_norm": 122.12571647799199,
+      "learning_rate": 5e-06,
+      "loss": 1.1813,
+      "num_input_tokens_seen": 676069904,
+      "step": 3911
+    },
+    {
+      "epoch": 1.497893527384144,
+      "loss": 1.2542847394943237,
+      "loss_ce": 0.19382768869400024,
+      "loss_iou": 0.48590174317359924,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 1.0604569911956787,
+      "num_input_tokens_seen": 676069904,
+      "step": 3911
+    },
+    {
+      "epoch": 1.4982765224052088,
+      "grad_norm": 139.25997298449545,
+      "learning_rate": 5e-06,
+      "loss": 1.5783,
+      "num_input_tokens_seen": 676242864,
+      "step": 3912
+    },
+    {
+      "epoch": 1.4982765224052088,
+      "loss": 1.51690673828125,
+      "loss_ce": 0.1545480489730835,
+      "loss_iou": 0.6304590702056885,
+      "loss_num": 0.020263671875,
+      "loss_xval": 1.362358570098877,
+      "num_input_tokens_seen": 676242864,
+      "step": 3912
+    },
+    {
+      "epoch": 1.4986595174262733,
+      "grad_norm": 123.25310142689585,
+      "learning_rate": 5e-06,
+      "loss": 1.8552,
+      "num_input_tokens_seen": 676416160,
+      "step": 3913
+    },
+    {
+      "epoch": 1.4986595174262733,
+      "loss": 1.8524422645568848,
+      "loss_ce": 0.21841105818748474,
+      "loss_iou": 0.754271388053894,
+      "loss_num": 0.025146484375,
+      "loss_xval": 1.634031057357788,
+      "num_input_tokens_seen": 676416160,
+      "step": 3913
+    },
+    {
+      "epoch": 1.499042512447338,
+      "grad_norm": 107.64315769116206,
+      "learning_rate": 5e-06,
+      "loss": 1.1797,
+      "num_input_tokens_seen": 676588952,
+      "step": 3914
+    },
+    {
+      "epoch": 1.499042512447338,
+      "loss": 1.0533498525619507,
+      "loss_ce": 0.1467789113521576,
+      "loss_iou": 0.4129336476325989,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.9025121927261353,
+      "num_input_tokens_seen": 676588952,
+      "step": 3914
+    },
+    {
+      "epoch": 1.4994255074684029,
+      "grad_norm": 189.16867982905197,
+      "learning_rate": 5e-06,
+      "loss": 1.399,
+      "num_input_tokens_seen": 676761864,
+      "step": 3915
+    },
+    {
+      "epoch": 1.4994255074684029,
+      "loss": 1.5084915161132812,
+      "loss_ce": 0.16149842739105225,
+      "loss_iou": 0.6406902074813843,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 1.3469932079315186,
+      "num_input_tokens_seen": 676761864,
+      "step": 3915
+    },
+    {
+      "epoch": 1.4998085024894676,
+      "grad_norm": 93.63628383723204,
+      "learning_rate": 5e-06,
+      "loss": 1.755,
+      "num_input_tokens_seen": 676934536,
+      "step": 3916
+    },
+    {
+      "epoch": 1.4998085024894676,
+      "loss": 1.7024974822998047,
+      "loss_ce": 0.17724838852882385,
+      "loss_iou": 0.6955468654632568,
+      "loss_num": 0.02685546875,
+      "loss_xval": 1.5252490043640137,
+      "num_input_tokens_seen": 676934536,
+      "step": 3916
+    },
+    {
+      "epoch": 1.5001914975105324,
+      "grad_norm": 110.46402062423184,
+      "learning_rate": 5e-06,
+      "loss": 1.1692,
+      "num_input_tokens_seen": 677107328,
+      "step": 3917
+    },
+    {
+      "epoch": 1.5001914975105324,
+      "loss": 1.185386300086975,
+      "loss_ce": 0.1652449518442154,
+      "loss_iou": 0.4723433256149292,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 1.020141363143921,
+      "num_input_tokens_seen": 677107328,
+      "step": 3917
+    },
+    {
+      "epoch": 1.5005744925315971,
+      "grad_norm": 177.81936394082368,
+      "learning_rate": 5e-06,
+      "loss": 1.772,
+      "num_input_tokens_seen": 677280320,
+      "step": 3918
+    },
+    {
+      "epoch": 1.5005744925315971,
+      "loss": 1.663917899131775,
+      "loss_ce": 0.1926594078540802,
+      "loss_iou": 0.6747772097587585,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 1.471258521080017,
+      "num_input_tokens_seen": 677280320,
+      "step": 3918
+    },
+    {
+      "epoch": 1.500957487552662,
+      "grad_norm": 113.27744686178022,
+      "learning_rate": 5e-06,
+      "loss": 1.5075,
+      "num_input_tokens_seen": 677453384,
+      "step": 3919
+    },
+    {
+      "epoch": 1.500957487552662,
+      "loss": 1.3370167016983032,
+      "loss_ce": 0.19517093896865845,
+      "loss_iou": 0.5225220322608948,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.1418458223342896,
+      "num_input_tokens_seen": 677453384,
+      "step": 3919
+    },
+    {
+      "epoch": 1.5013404825737267,
+      "grad_norm": 147.21865546380945,
+      "learning_rate": 5e-06,
+      "loss": 1.7438,
+      "num_input_tokens_seen": 677626224,
+      "step": 3920
+    },
+    {
+      "epoch": 1.5013404825737267,
+      "loss": 1.8057869672775269,
+      "loss_ce": 0.11913647502660751,
+      "loss_iou": 0.7625452280044556,
+      "loss_num": 0.0322265625,
+      "loss_xval": 1.6866505146026611,
+      "num_input_tokens_seen": 677626224,
+      "step": 3920
+    },
+    {
+      "epoch": 1.5017234775947914,
+      "grad_norm": 143.36489983173033,
+      "learning_rate": 5e-06,
+      "loss": 1.3252,
+      "num_input_tokens_seen": 677799232,
+      "step": 3921
+    },
+    {
+      "epoch": 1.5017234775947914,
+      "loss": 1.3773850202560425,
+      "loss_ce": 0.15254107117652893,
+      "loss_iou": 0.5581769943237305,
+      "loss_num": 0.021728515625,
+      "loss_xval": 1.224843978881836,
+      "num_input_tokens_seen": 677799232,
+      "step": 3921
+    },
+    {
+      "epoch": 1.502106472615856,
+      "grad_norm": 299.8088889611129,
+      "learning_rate": 5e-06,
+      "loss": 1.6504,
+      "num_input_tokens_seen": 677972304,
+      "step": 3922
+    },
+    {
+      "epoch": 1.502106472615856,
+      "loss": 1.6555671691894531,
+      "loss_ce": 0.21011576056480408,
+      "loss_iou": 0.6768578290939331,
+      "loss_num": 0.018310546875,
+      "loss_xval": 1.4454514980316162,
+      "num_input_tokens_seen": 677972304,
+      "step": 3922
+    },
+    {
+      "epoch": 1.5024894676369207,
+      "grad_norm": 140.5422263206731,
+      "learning_rate": 5e-06,
+      "loss": 1.7557,
+      "num_input_tokens_seen": 678145656,
+      "step": 3923
+    },
+    {
+      "epoch": 1.5024894676369207,
+      "loss": 1.9341998100280762,
+      "loss_ce": 0.2066763937473297,
+      "loss_iou": 0.7899091839790344,
+      "loss_num": 0.029541015625,
+      "loss_xval": 1.7275234460830688,
+      "num_input_tokens_seen": 678145656,
+      "step": 3923
+    },
+    {
+      "epoch": 1.5028724626579855,
+      "grad_norm": 168.56528740382507,
+      "learning_rate": 5e-06,
+      "loss": 1.2538,
+      "num_input_tokens_seen": 678318784,
+      "step": 3924
+    },
+    {
+      "epoch": 1.5028724626579855,
+      "loss": 1.2720829248428345,
+      "loss_ce": 0.18417221307754517,
+      "loss_iou": 0.5023903846740723,
+      "loss_num": 0.0166015625,
+      "loss_xval": 1.0879106521606445,
+      "num_input_tokens_seen": 678318784,
+      "step": 3924
+    },
+    {
+      "epoch": 1.5032554576790502,
+      "grad_norm": 146.38262092112728,
+      "learning_rate": 5e-06,
+      "loss": 1.3102,
+      "num_input_tokens_seen": 678492024,
+      "step": 3925
+    },
+    {
+      "epoch": 1.5032554576790502,
+      "loss": 1.4724841117858887,
+      "loss_ce": 0.16766434907913208,
+      "loss_iou": 0.6028035879135132,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 1.3048198223114014,
+      "num_input_tokens_seen": 678492024,
+      "step": 3925
+    },
+    {
+      "epoch": 1.5036384527001148,
+      "grad_norm": 301.52686169574025,
+      "learning_rate": 5e-06,
+      "loss": 1.8261,
+      "num_input_tokens_seen": 678665008,
+      "step": 3926
+    },
+    {
+      "epoch": 1.5036384527001148,
+      "loss": 1.8101704120635986,
+      "loss_ce": 0.16355374455451965,
+      "loss_iou": 0.7610829472541809,
+      "loss_num": 0.02490234375,
+      "loss_xval": 1.6466165781021118,
+      "num_input_tokens_seen": 678665008,
+      "step": 3926
+    },
+    {
+      "epoch": 1.5040214477211795,
+      "grad_norm": 151.62069510106224,
+      "learning_rate": 5e-06,
+      "loss": 1.5278,
+      "num_input_tokens_seen": 678838120,
+      "step": 3927
+    },
+    {
+      "epoch": 1.5040214477211795,
+      "loss": 1.6126360893249512,
+      "loss_ce": 0.17145290970802307,
+      "loss_iou": 0.676081657409668,
+      "loss_num": 0.017822265625,
+      "loss_xval": 1.441183090209961,
+      "num_input_tokens_seen": 678838120,
+      "step": 3927
+    },
+    {
+      "epoch": 1.5044044427422443,
+      "grad_norm": 151.24627586443248,
+      "learning_rate": 5e-06,
+      "loss": 1.5012,
+      "num_input_tokens_seen": 679010648,
+      "step": 3928
+    },
+    {
+      "epoch": 1.5044044427422443,
+      "loss": 1.468981146812439,
+      "loss_ce": 0.1473751664161682,
+      "loss_iou": 0.6055051684379578,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 1.3216060400009155,
+      "num_input_tokens_seen": 679010648,
+      "step": 3928
+    },
+    {
+      "epoch": 1.504787437763309,
+      "grad_norm": 117.36174820735435,
+      "learning_rate": 5e-06,
+      "loss": 1.3066,
+      "num_input_tokens_seen": 679183704,
+      "step": 3929
+    },
+    {
+      "epoch": 1.504787437763309,
+      "loss": 1.3060994148254395,
+      "loss_ce": 0.19131767749786377,
+      "loss_iou": 0.5271174311637878,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 1.1147817373275757,
+      "num_input_tokens_seen": 679183704,
+      "step": 3929
+    },
+    {
+      "epoch": 1.5051704327843738,
+      "grad_norm": 159.3159302948904,
+      "learning_rate": 5e-06,
+      "loss": 1.6127,
+      "num_input_tokens_seen": 679356424,
+      "step": 3930
+    },
+    {
+      "epoch": 1.5051704327843738,
+      "loss": 1.5889123678207397,
+      "loss_ce": 0.18345588445663452,
+      "loss_iou": 0.659011721611023,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 1.405456304550171,
+      "num_input_tokens_seen": 679356424,
+      "step": 3930
+    },
+    {
+      "epoch": 1.5055534278054385,
+      "grad_norm": 141.7121383308052,
+      "learning_rate": 5e-06,
+      "loss": 1.2912,
+      "num_input_tokens_seen": 679529184,
+      "step": 3931
+    },
+    {
+      "epoch": 1.5055534278054385,
+      "loss": 1.127032995223999,
+      "loss_ce": 0.1291084736585617,
+      "loss_iou": 0.45585617423057556,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.9979245066642761,
+      "num_input_tokens_seen": 679529184,
+      "step": 3931
+    },
+    {
+      "epoch": 1.5059364228265033,
+      "grad_norm": 193.61432466420874,
+      "learning_rate": 5e-06,
+      "loss": 1.7357,
+      "num_input_tokens_seen": 679701784,
+      "step": 3932
+    },
+    {
+      "epoch": 1.5059364228265033,
+      "loss": 1.6995859146118164,
+      "loss_ce": 0.14858072996139526,
+      "loss_iou": 0.706441342830658,
+      "loss_num": 0.027587890625,
+      "loss_xval": 1.551005244255066,
+      "num_input_tokens_seen": 679701784,
+      "step": 3932
+    },
+    {
+      "epoch": 1.506319417847568,
+      "grad_norm": 148.5073978214111,
+      "learning_rate": 5e-06,
+      "loss": 1.3603,
+      "num_input_tokens_seen": 679874576,
+      "step": 3933
+    },
+    {
+      "epoch": 1.506319417847568,
+      "loss": 1.3367570638656616,
+      "loss_ce": 0.15908914804458618,
+      "loss_iou": 0.5338717699050903,
+      "loss_num": 0.02197265625,
+      "loss_xval": 1.1776678562164307,
+      "num_input_tokens_seen": 679874576,
+      "step": 3933
+    },
+    {
+      "epoch": 1.5067024128686328,
+      "grad_norm": 125.59857770780995,
+      "learning_rate": 5e-06,
+      "loss": 1.2445,
+      "num_input_tokens_seen": 680047504,
+      "step": 3934
+    },
+    {
+      "epoch": 1.5067024128686328,
+      "loss": 1.1900660991668701,
+      "loss_ce": 0.1693650484085083,
+      "loss_iou": 0.4734853208065033,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 1.0207011699676514,
+      "num_input_tokens_seen": 680047504,
+      "step": 3934
+    },
+    {
+      "epoch": 1.5070854078896976,
+      "grad_norm": 216.85466740687056,
+      "learning_rate": 5e-06,
+      "loss": 1.8651,
+      "num_input_tokens_seen": 680220768,
+      "step": 3935
+    },
+    {
+      "epoch": 1.5070854078896976,
+      "loss": 1.8050951957702637,
+      "loss_ce": 0.20160922408103943,
+      "loss_iou": 0.7552036643028259,
+      "loss_num": 0.0185546875,
+      "loss_xval": 1.6034859418869019,
+      "num_input_tokens_seen": 680220768,
+      "step": 3935
+    },
+    {
+      "epoch": 1.5074684029107621,
+      "grad_norm": 104.59036594951397,
+      "learning_rate": 5e-06,
+      "loss": 1.5949,
+      "num_input_tokens_seen": 680393960,
+      "step": 3936
+    },
+    {
+      "epoch": 1.5074684029107621,
+      "loss": 1.5934373140335083,
+      "loss_ce": 0.18331730365753174,
+      "loss_iou": 0.6564760208129883,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 1.4101200103759766,
+      "num_input_tokens_seen": 680393960,
+      "step": 3936
+    },
+    {
+      "epoch": 1.5078513979318269,
+      "grad_norm": 94.69427317633223,
+      "learning_rate": 5e-06,
+      "loss": 1.2119,
+      "num_input_tokens_seen": 680566496,
+      "step": 3937
+    },
+    {
+      "epoch": 1.5078513979318269,
+      "loss": 1.2829943895339966,
+      "loss_ce": 0.11970832198858261,
+      "loss_iou": 0.5344933867454529,
+      "loss_num": 0.018798828125,
+      "loss_xval": 1.1632860898971558,
+      "num_input_tokens_seen": 680566496,
+      "step": 3937
+    },
+    {
+      "epoch": 1.5082343929528916,
+      "grad_norm": 153.91357361344652,
+      "learning_rate": 5e-06,
+      "loss": 1.31,
+      "num_input_tokens_seen": 680739464,
+      "step": 3938
+    },
+    {
+      "epoch": 1.5082343929528916,
+      "loss": 1.3061113357543945,
+      "loss_ce": 0.13433563709259033,
+      "loss_iou": 0.5441702604293823,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 1.1717755794525146,
+      "num_input_tokens_seen": 680739464,
+      "step": 3938
+    },
+    {
+      "epoch": 1.5086173879739564,
+      "grad_norm": 103.84133103064748,
+      "learning_rate": 5e-06,
+      "loss": 1.615,
+      "num_input_tokens_seen": 680912048,
+      "step": 3939
+    },
+    {
+      "epoch": 1.5086173879739564,
+      "loss": 1.4809836149215698,
+      "loss_ce": 0.1380067616701126,
+      "loss_iou": 0.6224467158317566,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 1.3429769277572632,
+      "num_input_tokens_seen": 680912048,
+      "step": 3939
+    },
+    {
+      "epoch": 1.509000382995021,
+      "grad_norm": 121.17580221839172,
+      "learning_rate": 5e-06,
+      "loss": 1.4507,
+      "num_input_tokens_seen": 681084920,
+      "step": 3940
+    },
+    {
+      "epoch": 1.509000382995021,
+      "loss": 1.5148675441741943,
+      "loss_ce": 0.15818607807159424,
+      "loss_iou": 0.6258504986763,
+      "loss_num": 0.02099609375,
+      "loss_xval": 1.3566814661026,
+      "num_input_tokens_seen": 681084920,
+      "step": 3940
+    },
+    {
+      "epoch": 1.5093833780160857,
+      "grad_norm": 226.64884905297725,
+      "learning_rate": 5e-06,
+      "loss": 2.0001,
+      "num_input_tokens_seen": 681257592,
+      "step": 3941
+    },
+    {
+      "epoch": 1.5093833780160857,
+      "loss": 2.0569167137145996,
+      "loss_ce": 0.1741567999124527,
+      "loss_iou": 0.8931165337562561,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.8827601671218872,
+      "num_input_tokens_seen": 681257592,
+      "step": 3941
+    },
+    {
+      "epoch": 1.5097663730371504,
+      "grad_norm": 136.4010765764811,
+      "learning_rate": 5e-06,
+      "loss": 1.9591,
+      "num_input_tokens_seen": 681430552,
+      "step": 3942
+    },
+    {
+      "epoch": 1.5097663730371504,
+      "loss": 1.8821616172790527,
+      "loss_ce": 0.2154098004102707,
+      "loss_iou": 0.7494220733642578,
+      "loss_num": 0.03369140625,
+      "loss_xval": 1.6667518615722656,
+      "num_input_tokens_seen": 681430552,
+      "step": 3942
+    },
+    {
+      "epoch": 1.5101493680582152,
+      "grad_norm": 67.87533683784773,
+      "learning_rate": 5e-06,
+      "loss": 1.3768,
+      "num_input_tokens_seen": 681603568,
+      "step": 3943
+    },
+    {
+      "epoch": 1.5101493680582152,
+      "loss": 1.332188367843628,
+      "loss_ce": 0.1632273942232132,
+      "loss_iou": 0.5403978824615479,
+      "loss_num": 0.017578125,
+      "loss_xval": 1.1689610481262207,
+      "num_input_tokens_seen": 681603568,
+      "step": 3943
+    },
+    {
+      "epoch": 1.51053236307928,
+      "grad_norm": 111.59310271327914,
+      "learning_rate": 5e-06,
+      "loss": 1.3828,
+      "num_input_tokens_seen": 681776256,
+      "step": 3944
+    },
+    {
+      "epoch": 1.51053236307928,
+      "loss": 1.3522703647613525,
+      "loss_ce": 0.16833075881004333,
+      "loss_iou": 0.5519307255744934,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 1.1839395761489868,
+      "num_input_tokens_seen": 681776256,
+      "step": 3944
+    },
+    {
+      "epoch": 1.5109153581003447,
+      "grad_norm": 205.43888918345615,
+      "learning_rate": 5e-06,
+      "loss": 1.5487,
+      "num_input_tokens_seen": 681949584,
+      "step": 3945
+    },
+    {
+      "epoch": 1.5109153581003447,
+      "loss": 1.6860454082489014,
+      "loss_ce": 0.18768858909606934,
+      "loss_iou": 0.699862003326416,
+      "loss_num": 0.019775390625,
+      "loss_xval": 1.498356819152832,
+      "num_input_tokens_seen": 681949584,
+      "step": 3945
+    },
+    {
+      "epoch": 1.5112983531214095,
+      "grad_norm": 127.98848796269809,
+      "learning_rate": 5e-06,
+      "loss": 1.5937,
+      "num_input_tokens_seen": 682122688,
+      "step": 3946
+    },
+    {
+      "epoch": 1.5112983531214095,
+      "loss": 1.4739629030227661,
+      "loss_ce": 0.16732311248779297,
+      "loss_iou": 0.6131281852722168,
+      "loss_num": 0.01611328125,
+      "loss_xval": 1.3001699447631836,
+      "num_input_tokens_seen": 682122688,
+      "step": 3946
+    },
+    {
+      "epoch": 1.5116813481424742,
+      "grad_norm": 154.43631041264297,
+      "learning_rate": 5e-06,
+      "loss": 1.8958,
+      "num_input_tokens_seen": 682295648,
+      "step": 3947
+    },
+    {
+      "epoch": 1.5116813481424742,
+      "loss": 1.7483105659484863,
+      "loss_ce": 0.13244858384132385,
+      "loss_iou": 0.7305383682250977,
+      "loss_num": 0.031005859375,
+      "loss_xval": 1.6158618927001953,
+      "num_input_tokens_seen": 682295648,
+      "step": 3947
+    },
+    {
+      "epoch": 1.512064343163539,
+      "grad_norm": 108.00771571125344,
+      "learning_rate": 5e-06,
+      "loss": 1.3608,
+      "num_input_tokens_seen": 682468688,
+      "step": 3948
+    },
+    {
+      "epoch": 1.512064343163539,
+      "loss": 1.4411499500274658,
+      "loss_ce": 0.18800832331180573,
+      "loss_iou": 0.5710746049880981,
+      "loss_num": 0.022216796875,
+      "loss_xval": 1.2531416416168213,
+      "num_input_tokens_seen": 682468688,
+      "step": 3948
+    },
+    {
+      "epoch": 1.5124473381846038,
+      "grad_norm": 223.0197942802195,
+      "learning_rate": 5e-06,
+      "loss": 1.6571,
+      "num_input_tokens_seen": 682639792,
+      "step": 3949
+    },
+    {
+      "epoch": 1.5124473381846038,
+      "loss": 1.5951857566833496,
+      "loss_ce": 0.18513739109039307,
+      "loss_iou": 0.6478037238121033,
+      "loss_num": 0.02294921875,
+      "loss_xval": 1.4100483655929565,
+      "num_input_tokens_seen": 682639792,
+      "step": 3949
+    },
+    {
+      "epoch": 1.5128303332056683,
+      "grad_norm": 83.44992441252616,
+      "learning_rate": 5e-06,
+      "loss": 2.0769,
+      "num_input_tokens_seen": 682813040,
+      "step": 3950
+    },
+    {
+      "epoch": 1.5128303332056683,
+      "loss": 2.0230278968811035,
+      "loss_ce": 0.21663933992385864,
+      "loss_iou": 0.8293722867965698,
+      "loss_num": 0.029541015625,
+      "loss_xval": 1.8063886165618896,
+      "num_input_tokens_seen": 682813040,
+      "step": 3950
+    },
+    {
+      "epoch": 1.513213328226733,
+      "grad_norm": 109.7736289437539,
+      "learning_rate": 5e-06,
+      "loss": 1.8202,
+      "num_input_tokens_seen": 682985920,
+      "step": 3951
+    },
+    {
+      "epoch": 1.513213328226733,
+      "loss": 1.589260220527649,
+      "loss_ce": 0.1508396863937378,
+      "loss_iou": 0.6564966440200806,
+      "loss_num": 0.025146484375,
+      "loss_xval": 1.4384205341339111,
+      "num_input_tokens_seen": 682985920,
+      "step": 3951
+    },
+    {
+      "epoch": 1.5135963232477978,
+      "grad_norm": 113.40872734827907,
+      "learning_rate": 5e-06,
+      "loss": 1.4412,
+      "num_input_tokens_seen": 683158960,
+      "step": 3952
+    },
+    {
+      "epoch": 1.5135963232477978,
+      "loss": 1.4061739444732666,
+      "loss_ce": 0.16902853548526764,
+      "loss_iou": 0.5697598457336426,
+      "loss_num": 0.01953125,
+      "loss_xval": 1.2096796035766602,
+      "num_input_tokens_seen": 683158960,
+      "step": 3952
+    },
+    {
+      "epoch": 1.5139793182688623,
+      "grad_norm": 82.42682002201437,
+      "learning_rate": 5e-06,
+      "loss": 1.8388,
+      "num_input_tokens_seen": 683331592,
+      "step": 3953
+    },
+    {
+      "epoch": 1.5139793182688623,
+      "loss": 2.0235116481781006,
+      "loss_ce": 0.14281706511974335,
+      "loss_iou": 0.8556303977966309,
+      "loss_num": 0.033935546875,
+      "loss_xval": 1.8806943893432617,
+      "num_input_tokens_seen": 683331592,
+      "step": 3953
+    },
+    {
+      "epoch": 1.514362313289927,
+      "grad_norm": 91.4057633169997,
+      "learning_rate": 5e-06,
+      "loss": 1.5258,
+      "num_input_tokens_seen": 683504528,
+      "step": 3954
+    },
+    {
+      "epoch": 1.514362313289927,
+      "loss": 1.494154691696167,
+      "loss_ce": 0.1196802407503128,
+      "loss_iou": 0.6352352499961853,
+      "loss_num": 0.020751953125,
+      "loss_xval": 1.3662956953048706,
+      "num_input_tokens_seen": 683504528,
+      "step": 3954
+    },
+    {
+      "epoch": 1.5147453083109919,
+      "grad_norm": 107.3594175956104,
+      "learning_rate": 5e-06,
+      "loss": 1.5227,
+      "num_input_tokens_seen": 683677384,
+      "step": 3955
+    },
+    {
+      "epoch": 1.5147453083109919,
+      "loss": 1.478843331336975,
+      "loss_ce": 0.16039426624774933,
+      "loss_iou": 0.6079397201538086,
+      "loss_num": 0.0205078125,
+      "loss_xval": 1.3184490203857422,
+      "num_input_tokens_seen": 683677384,
+      "step": 3955
+    },
+    {
+      "epoch": 1.5151283033320566,
+      "grad_norm": 93.84858563554947,
+      "learning_rate": 5e-06,
+      "loss": 1.3261,
+      "num_input_tokens_seen": 683850496,
+      "step": 3956
+    },
+    {
+      "epoch": 1.5151283033320566,
+      "loss": 1.3284571170806885,
+      "loss_ce": 0.22089354693889618,
+      "loss_iou": 0.5197699069976807,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 1.1075634956359863,
+      "num_input_tokens_seen": 683850496,
+      "step": 3956
+    },
+    {
+      "epoch": 1.5155112983531214,
+      "grad_norm": 211.77244019223895,
+      "learning_rate": 5e-06,
+      "loss": 1.9287,
+      "num_input_tokens_seen": 684023272,
+      "step": 3957
+    },
+    {
+      "epoch": 1.5155112983531214,
+      "loss": 1.8714628219604492,
+      "loss_ce": 0.14375276863574982,
+      "loss_iou": 0.8195739984512329,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 1.7277100086212158,
+      "num_input_tokens_seen": 684023272,
+      "step": 3957
+    },
+    {
+      "epoch": 1.5158942933741861,
+      "grad_norm": 254.66520592087497,
+      "learning_rate": 5e-06,
+      "loss": 1.9458,
+      "num_input_tokens_seen": 684196400,
+      "step": 3958
+    },
+    {
+      "epoch": 1.5158942933741861,
+      "loss": 1.9426088333129883,
+      "loss_ce": 0.13388551771640778,
+      "loss_iou": 0.8289831876754761,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 1.8087232112884521,
+      "num_input_tokens_seen": 684196400,
+      "step": 3958
+    },
+    {
+      "epoch": 1.5162772883952509,
+      "grad_norm": 152.56221898504054,
+      "learning_rate": 5e-06,
+      "loss": 2.9065,
+      "num_input_tokens_seen": 684369808,
+      "step": 3959
+    },
+    {
+      "epoch": 1.5162772883952509,
+      "loss": 2.9309732913970947,
+      "loss_ce": 0.14469051361083984,
+      "loss_iou": 1.317763090133667,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 2.786283016204834,
+      "num_input_tokens_seen": 684369808,
+      "step": 3959
+    },
+    {
+      "epoch": 1.5166602834163156,
+      "grad_norm": 81.62702327862401,
+      "learning_rate": 5e-06,
+      "loss": 2.354,
+      "num_input_tokens_seen": 684542680,
+      "step": 3960
+    },
+    {
+      "epoch": 1.5166602834163156,
+      "loss": 2.4589433670043945,
+      "loss_ce": 0.1825285255908966,
+      "loss_iou": 0.9466791749000549,
+      "loss_num": 0.07666015625,
+      "loss_xval": 2.2764148712158203,
+      "num_input_tokens_seen": 684542680,
+      "step": 3960
+    },
+    {
+      "epoch": 1.5170432784373804,
+      "grad_norm": 82.43059801788286,
+      "learning_rate": 5e-06,
+      "loss": 1.492,
+      "num_input_tokens_seen": 684715264,
+      "step": 3961
+    },
+    {
+      "epoch": 1.5170432784373804,
+      "loss": 1.3565406799316406,
+      "loss_ce": 0.10414262115955353,
+      "loss_iou": 0.5798122882843018,
+      "loss_num": 0.0185546875,
+      "loss_xval": 1.2523980140686035,
+      "num_input_tokens_seen": 684715264,
+      "step": 3961
+    },
+    {
+      "epoch": 1.5174262734584452,
+      "grad_norm": 337.9323224462057,
+      "learning_rate": 5e-06,
+      "loss": 2.3093,
+      "num_input_tokens_seen": 684887952,
+      "step": 3962
+    },
+    {
+      "epoch": 1.5174262734584452,
+      "loss": 2.363466262817383,
+      "loss_ce": 0.1540198028087616,
+      "loss_iou": 1.040392279624939,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 2.209446668624878,
+      "num_input_tokens_seen": 684887952,
+      "step": 3962
+    },
+    {
+      "epoch": 1.51780926847951,
+      "grad_norm": 121.1484778544416,
+      "learning_rate": 5e-06,
+      "loss": 2.2396,
+      "num_input_tokens_seen": 685061208,
+      "step": 3963
+    },
+    {
+      "epoch": 1.51780926847951,
+      "loss": 2.2443549633026123,
+      "loss_ce": 0.18490296602249146,
+      "loss_iou": 0.8711566925048828,
+      "loss_num": 0.0634765625,
+      "loss_xval": 2.0594520568847656,
+      "num_input_tokens_seen": 685061208,
+      "step": 3963
+    },
+    {
+      "epoch": 1.5181922635005745,
+      "grad_norm": 83.59056278522132,
+      "learning_rate": 5e-06,
+      "loss": 1.9785,
+      "num_input_tokens_seen": 685233880,
+      "step": 3964
+    },
+    {
+      "epoch": 1.5181922635005745,
+      "loss": 1.9430819749832153,
+      "loss_ce": 0.16593565046787262,
+      "loss_iou": 0.824882984161377,
+      "loss_num": 0.025390625,
+      "loss_xval": 1.777146339416504,
+      "num_input_tokens_seen": 685233880,
+      "step": 3964
+    },
+    {
+      "epoch": 1.5185752585216392,
+      "grad_norm": 120.30319911187016,
+      "learning_rate": 5e-06,
+      "loss": 2.475,
+      "num_input_tokens_seen": 685406640,
+      "step": 3965
+    },
+    {
+      "epoch": 1.5185752585216392,
+      "loss": 2.295506477355957,
+      "loss_ce": 0.17737790942192078,
+      "loss_iou": 1.001630187034607,
+      "loss_num": 0.02294921875,
+      "loss_xval": 2.118128538131714,
+      "num_input_tokens_seen": 685406640,
+      "step": 3965
+    },
+    {
+      "epoch": 1.518958253542704,
+      "grad_norm": 55.87681187685328,
+      "learning_rate": 5e-06,
+      "loss": 2.176,
+      "num_input_tokens_seen": 685579960,
+      "step": 3966
+    },
+    {
+      "epoch": 1.518958253542704,
+      "loss": 2.278275728225708,
+      "loss_ce": 0.24889490008354187,
+      "loss_iou": 0.8973199129104614,
+      "loss_num": 0.046875,
+      "loss_xval": 2.029381036758423,
+      "num_input_tokens_seen": 685579960,
+      "step": 3966
+    },
+    {
+      "epoch": 1.5193412485637685,
+      "grad_norm": 71.51627517987505,
+      "learning_rate": 5e-06,
+      "loss": 1.6915,
+      "num_input_tokens_seen": 685749456,
+      "step": 3967
+    },
+    {
+      "epoch": 1.5193412485637685,
+      "loss": 1.5556166172027588,
+      "loss_ce": 0.19647814333438873,
+      "loss_iou": 0.6312598586082458,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.3591383695602417,
+      "num_input_tokens_seen": 685749456,
+      "step": 3967
+    },
+    {
+      "epoch": 1.5197242435848333,
+      "grad_norm": 199.55954187293344,
+      "learning_rate": 5e-06,
+      "loss": 1.5538,
+      "num_input_tokens_seen": 685922568,
+      "step": 3968
+    },
+    {
+      "epoch": 1.5197242435848333,
+      "loss": 1.4557030200958252,
+      "loss_ce": 0.17327332496643066,
+      "loss_iou": 0.5964608192443848,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 1.2824296951293945,
+      "num_input_tokens_seen": 685922568,
+      "step": 3968
+    },
+    {
+      "epoch": 1.520107238605898,
+      "grad_norm": 214.03904571066968,
+      "learning_rate": 5e-06,
+      "loss": 1.8144,
+      "num_input_tokens_seen": 686095808,
+      "step": 3969
+    },
+    {
+      "epoch": 1.520107238605898,
+      "loss": 1.8164312839508057,
+      "loss_ce": 0.15939146280288696,
+      "loss_iou": 0.7656842470169067,
+      "loss_num": 0.025146484375,
+      "loss_xval": 1.6570398807525635,
+      "num_input_tokens_seen": 686095808,
+      "step": 3969
+    },
+    {
+      "epoch": 1.5204902336269628,
+      "grad_norm": 224.43256981929963,
+      "learning_rate": 5e-06,
+      "loss": 1.9742,
+      "num_input_tokens_seen": 686268848,
+      "step": 3970
+    },
+    {
+      "epoch": 1.5204902336269628,
+      "loss": 2.041718006134033,
+      "loss_ce": 0.14143480360507965,
+      "loss_iou": 0.8721081018447876,
+      "loss_num": 0.03125,
+      "loss_xval": 1.9002830982208252,
+      "num_input_tokens_seen": 686268848,
+      "step": 3970
+    },
+    {
+      "epoch": 1.5208732286480275,
+      "grad_norm": 172.600474175764,
+      "learning_rate": 5e-06,
+      "loss": 1.8223,
+      "num_input_tokens_seen": 686441752,
+      "step": 3971
+    },
+    {
+      "epoch": 1.5208732286480275,
+      "loss": 1.8466782569885254,
+      "loss_ce": 0.18908868730068207,
+      "loss_iou": 0.7544234991073608,
+      "loss_num": 0.02978515625,
+      "loss_xval": 1.6575896739959717,
+      "num_input_tokens_seen": 686441752,
+      "step": 3971
+    },
+    {
+      "epoch": 1.5212562236690923,
+      "grad_norm": 97.08139887337438,
+      "learning_rate": 5e-06,
+      "loss": 1.5846,
+      "num_input_tokens_seen": 686614816,
+      "step": 3972
+    },
+    {
+      "epoch": 1.5212562236690923,
+      "loss": 1.645388126373291,
+      "loss_ce": 0.1794380396604538,
+      "loss_iou": 0.6563454866409302,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 1.4659502506256104,
+      "num_input_tokens_seen": 686614816,
+      "step": 3972
+    },
+    {
+      "epoch": 1.521639218690157,
+      "grad_norm": 118.64503531925449,
+      "learning_rate": 5e-06,
+      "loss": 1.4486,
+      "num_input_tokens_seen": 686787720,
+      "step": 3973
+    },
+    {
+      "epoch": 1.521639218690157,
+      "loss": 1.5404670238494873,
+      "loss_ce": 0.16298604011535645,
+      "loss_iou": 0.6268203854560852,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 1.3774811029434204,
+      "num_input_tokens_seen": 686787720,
+      "step": 3973
+    },
+    {
+      "epoch": 1.5220222137112218,
+      "grad_norm": 153.00926754387302,
+      "learning_rate": 5e-06,
+      "loss": 1.7508,
+      "num_input_tokens_seen": 686960536,
+      "step": 3974
+    },
+    {
+      "epoch": 1.5220222137112218,
+      "loss": 1.7331939935684204,
+      "loss_ce": 0.17863628268241882,
+      "loss_iou": 0.7111777663230896,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 1.5545576810836792,
+      "num_input_tokens_seen": 686960536,
+      "step": 3974
+    },
+    {
+      "epoch": 1.5224052087322866,
+      "grad_norm": 99.33252420757358,
+      "learning_rate": 5e-06,
+      "loss": 1.4462,
+      "num_input_tokens_seen": 687133816,
+      "step": 3975
+    },
+    {
+      "epoch": 1.5224052087322866,
+      "loss": 1.349708080291748,
+      "loss_ce": 0.1548483967781067,
+      "loss_iou": 0.5491968393325806,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.1948597431182861,
+      "num_input_tokens_seen": 687133816,
+      "step": 3975
+    },
+    {
+      "epoch": 1.5227882037533513,
+      "grad_norm": 164.65765648653823,
+      "learning_rate": 5e-06,
+      "loss": 1.4702,
+      "num_input_tokens_seen": 687306776,
+      "step": 3976
+    },
+    {
+      "epoch": 1.5227882037533513,
+      "loss": 1.6038367748260498,
+      "loss_ce": 0.18100813031196594,
+      "loss_iou": 0.6422615051269531,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 1.4228286743164062,
+      "num_input_tokens_seen": 687306776,
+      "step": 3976
+    },
+    {
+      "epoch": 1.523171198774416,
+      "grad_norm": 85.26483364804754,
+      "learning_rate": 5e-06,
+      "loss": 1.3608,
+      "num_input_tokens_seen": 687479800,
+      "step": 3977
+    },
+    {
+      "epoch": 1.523171198774416,
+      "loss": 1.4947741031646729,
+      "loss_ce": 0.1709693968296051,
+      "loss_iou": 0.6206426620483398,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 1.3238048553466797,
+      "num_input_tokens_seen": 687479800,
+      "step": 3977
+    },
+    {
+      "epoch": 1.5235541937954806,
+      "grad_norm": 95.08004515964755,
+      "learning_rate": 5e-06,
+      "loss": 1.4765,
+      "num_input_tokens_seen": 687652672,
+      "step": 3978
+    },
+    {
+      "epoch": 1.5235541937954806,
+      "loss": 1.6495649814605713,
+      "loss_ce": 0.19735872745513916,
+      "loss_iou": 0.6748487949371338,
+      "loss_num": 0.0205078125,
+      "loss_xval": 1.4522061347961426,
+      "num_input_tokens_seen": 687652672,
+      "step": 3978
+    },
+    {
+      "epoch": 1.5239371888165454,
+      "grad_norm": 108.5698208549644,
+      "learning_rate": 5e-06,
+      "loss": 1.4894,
+      "num_input_tokens_seen": 687825624,
+      "step": 3979
+    },
+    {
+      "epoch": 1.5239371888165454,
+      "loss": 1.5349109172821045,
+      "loss_ce": 0.14889198541641235,
+      "loss_iou": 0.6471720337867737,
+      "loss_num": 0.018310546875,
+      "loss_xval": 1.386018991470337,
+      "num_input_tokens_seen": 687825624,
+      "step": 3979
+    },
+    {
+      "epoch": 1.5243201838376101,
+      "grad_norm": 123.08096805229017,
+      "learning_rate": 5e-06,
+      "loss": 1.6522,
+      "num_input_tokens_seen": 687998728,
+      "step": 3980
+    },
+    {
+      "epoch": 1.5243201838376101,
+      "loss": 1.725043535232544,
+      "loss_ce": 0.17687951028347015,
+      "loss_iou": 0.7139319181442261,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 1.5481641292572021,
+      "num_input_tokens_seen": 687998728,
+      "step": 3980
+    },
+    {
+      "epoch": 1.5247031788586747,
+      "grad_norm": 142.74838888561476,
+      "learning_rate": 5e-06,
+      "loss": 1.3553,
+      "num_input_tokens_seen": 688171400,
+      "step": 3981
+    },
+    {
+      "epoch": 1.5247031788586747,
+      "loss": 1.1920249462127686,
+      "loss_ce": 0.12986698746681213,
+      "loss_iou": 0.4850126802921295,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 1.0621578693389893,
+      "num_input_tokens_seen": 688171400,
+      "step": 3981
+    },
+    {
+      "epoch": 1.5250861738797394,
+      "grad_norm": 108.84666223775466,
+      "learning_rate": 5e-06,
+      "loss": 1.3054,
+      "num_input_tokens_seen": 688344192,
+      "step": 3982
+    },
+    {
+      "epoch": 1.5250861738797394,
+      "loss": 1.3433678150177002,
+      "loss_ce": 0.16022491455078125,
+      "loss_iou": 0.555637001991272,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 1.183142900466919,
+      "num_input_tokens_seen": 688344192,
+      "step": 3982
+    },
+    {
+      "epoch": 1.5254691689008042,
+      "grad_norm": 144.08404942888112,
+      "learning_rate": 5e-06,
+      "loss": 1.468,
+      "num_input_tokens_seen": 688517040,
+      "step": 3983
+    },
+    {
+      "epoch": 1.5254691689008042,
+      "loss": 1.5013774633407593,
+      "loss_ce": 0.14980170130729675,
+      "loss_iou": 0.6245488524436951,
+      "loss_num": 0.0205078125,
+      "loss_xval": 1.3515757322311401,
+      "num_input_tokens_seen": 688517040,
+      "step": 3983
+    },
+    {
+      "epoch": 1.525852163921869,
+      "grad_norm": 176.312711176238,
+      "learning_rate": 5e-06,
+      "loss": 1.6051,
+      "num_input_tokens_seen": 688689992,
+      "step": 3984
+    },
+    {
+      "epoch": 1.525852163921869,
+      "loss": 1.482277512550354,
+      "loss_ce": 0.16307860612869263,
+      "loss_iou": 0.6153489351272583,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 1.3191988468170166,
+      "num_input_tokens_seen": 688689992,
+      "step": 3984
+    },
+    {
+      "epoch": 1.5262351589429337,
+      "grad_norm": 121.81468625578677,
+      "learning_rate": 5e-06,
+      "loss": 1.3592,
+      "num_input_tokens_seen": 688862904,
+      "step": 3985
+    },
+    {
+      "epoch": 1.5262351589429337,
+      "loss": 1.2518173456192017,
+      "loss_ce": 0.15622524917125702,
+      "loss_iou": 0.5110986828804016,
+      "loss_num": 0.0146484375,
+      "loss_xval": 1.0804554224014282,
+      "num_input_tokens_seen": 688862904,
+      "step": 3985
+    },
+    {
+      "epoch": 1.5266181539639985,
+      "grad_norm": 298.4313072516741,
+      "learning_rate": 5e-06,
+      "loss": 1.7817,
+      "num_input_tokens_seen": 689035448,
+      "step": 3986
+    },
+    {
+      "epoch": 1.5266181539639985,
+      "loss": 1.6795730590820312,
+      "loss_ce": 0.13823232054710388,
+      "loss_iou": 0.6964516043663025,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 1.541340708732605,
+      "num_input_tokens_seen": 689035448,
+      "step": 3986
+    },
+    {
+      "epoch": 1.5270011489850632,
+      "grad_norm": 242.10426549026667,
+      "learning_rate": 5e-06,
+      "loss": 1.4475,
+      "num_input_tokens_seen": 689204640,
+      "step": 3987
+    },
+    {
+      "epoch": 1.5270011489850632,
+      "loss": 1.5300861597061157,
+      "loss_ce": 0.16784243285655975,
+      "loss_iou": 0.631607174873352,
+      "loss_num": 0.019775390625,
+      "loss_xval": 1.362243890762329,
+      "num_input_tokens_seen": 689204640,
+      "step": 3987
+    },
+    {
+      "epoch": 1.527384144006128,
+      "grad_norm": 192.50921147839483,
+      "learning_rate": 5e-06,
+      "loss": 1.7361,
+      "num_input_tokens_seen": 689377632,
+      "step": 3988
+    },
+    {
+      "epoch": 1.527384144006128,
+      "loss": 1.644757628440857,
+      "loss_ce": 0.19601956009864807,
+      "loss_iou": 0.6878089904785156,
+      "loss_num": 0.0146484375,
+      "loss_xval": 1.4487380981445312,
+      "num_input_tokens_seen": 689377632,
+      "step": 3988
+    },
+    {
+      "epoch": 1.5277671390271927,
+      "grad_norm": 111.88305398399335,
+      "learning_rate": 5e-06,
+      "loss": 1.6252,
+      "num_input_tokens_seen": 689550544,
+      "step": 3989
+    },
+    {
+      "epoch": 1.5277671390271927,
+      "loss": 1.6201847791671753,
+      "loss_ce": 0.1457895040512085,
+      "loss_iou": 0.6800992488861084,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 1.4743952751159668,
+      "num_input_tokens_seen": 689550544,
+      "step": 3989
+    },
+    {
+      "epoch": 1.5281501340482575,
+      "grad_norm": 97.06904162293944,
+      "learning_rate": 5e-06,
+      "loss": 1.3497,
+      "num_input_tokens_seen": 689723640,
+      "step": 3990
+    },
+    {
+      "epoch": 1.5281501340482575,
+      "loss": 1.361687183380127,
+      "loss_ce": 0.17057424783706665,
+      "loss_iou": 0.5443174839019775,
+      "loss_num": 0.0205078125,
+      "loss_xval": 1.191112995147705,
+      "num_input_tokens_seen": 689723640,
+      "step": 3990
+    },
+    {
+      "epoch": 1.5285331290693223,
+      "grad_norm": 90.30874889339908,
+      "learning_rate": 5e-06,
+      "loss": 1.5368,
+      "num_input_tokens_seen": 689896488,
+      "step": 3991
+    },
+    {
+      "epoch": 1.5285331290693223,
+      "loss": 1.5996413230895996,
+      "loss_ce": 0.17233280837535858,
+      "loss_iou": 0.6681984066963196,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 1.4273086786270142,
+      "num_input_tokens_seen": 689896488,
+      "step": 3991
+    },
+    {
+      "epoch": 1.5289161240903868,
+      "grad_norm": 147.39944798644706,
+      "learning_rate": 5e-06,
+      "loss": 1.6209,
+      "num_input_tokens_seen": 690069520,
+      "step": 3992
+    },
+    {
+      "epoch": 1.5289161240903868,
+      "loss": 1.534162998199463,
+      "loss_ce": 0.18425728380680084,
+      "loss_iou": 0.6326555609703064,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 1.3499058485031128,
+      "num_input_tokens_seen": 690069520,
+      "step": 3992
+    },
+    {
+      "epoch": 1.5292991191114516,
+      "grad_norm": 167.28159313669147,
+      "learning_rate": 5e-06,
+      "loss": 1.6969,
+      "num_input_tokens_seen": 690242560,
+      "step": 3993
+    },
+    {
+      "epoch": 1.5292991191114516,
+      "loss": 1.727402925491333,
+      "loss_ce": 0.15593892335891724,
+      "loss_iou": 0.7414052486419678,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 1.5714640617370605,
+      "num_input_tokens_seen": 690242560,
+      "step": 3993
+    },
+    {
+      "epoch": 1.5296821141325163,
+      "grad_norm": 216.9949331892913,
+      "learning_rate": 5e-06,
+      "loss": 1.7428,
+      "num_input_tokens_seen": 690415248,
+      "step": 3994
+    },
+    {
+      "epoch": 1.5296821141325163,
+      "loss": 1.61613929271698,
+      "loss_ce": 0.1425824761390686,
+      "loss_iou": 0.6704636812210083,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 1.4735567569732666,
+      "num_input_tokens_seen": 690415248,
+      "step": 3994
+    },
+    {
+      "epoch": 1.5300651091535808,
+      "grad_norm": 93.0324383986881,
+      "learning_rate": 5e-06,
+      "loss": 1.4356,
+      "num_input_tokens_seen": 690588264,
+      "step": 3995
+    },
+    {
+      "epoch": 1.5300651091535808,
+      "loss": 1.479750156402588,
+      "loss_ce": 0.15825852751731873,
+      "loss_iou": 0.6298162937164307,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 1.3214917182922363,
+      "num_input_tokens_seen": 690588264,
+      "step": 3995
+    },
+    {
+      "epoch": 1.5304481041746456,
+      "grad_norm": 106.38967492820314,
+      "learning_rate": 5e-06,
+      "loss": 1.4427,
+      "num_input_tokens_seen": 690761440,
+      "step": 3996
+    },
+    {
+      "epoch": 1.5304481041746456,
+      "loss": 1.488261103630066,
+      "loss_ce": 0.16846603155136108,
+      "loss_iou": 0.6170355677604675,
+      "loss_num": 0.01708984375,
+      "loss_xval": 1.31979501247406,
+      "num_input_tokens_seen": 690761440,
+      "step": 3996
+    },
+    {
+      "epoch": 1.5308310991957104,
+      "grad_norm": 120.89542447178839,
+      "learning_rate": 5e-06,
+      "loss": 1.7495,
+      "num_input_tokens_seen": 690934416,
+      "step": 3997
+    },
+    {
+      "epoch": 1.5308310991957104,
+      "loss": 1.8797813653945923,
+      "loss_ce": 0.18472015857696533,
+      "loss_iou": 0.782970666885376,
+      "loss_num": 0.02587890625,
+      "loss_xval": 1.6950613260269165,
+      "num_input_tokens_seen": 690934416,
+      "step": 3997
+    },
+    {
+      "epoch": 1.5312140942167751,
+      "grad_norm": 109.55889150361551,
+      "learning_rate": 5e-06,
+      "loss": 1.5846,
+      "num_input_tokens_seen": 691107376,
+      "step": 3998
+    },
+    {
+      "epoch": 1.5312140942167751,
+      "loss": 1.6063764095306396,
+      "loss_ce": 0.2207745611667633,
+      "loss_iou": 0.6444305777549744,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 1.3856018781661987,
+      "num_input_tokens_seen": 691107376,
+      "step": 3998
+    },
+    {
+      "epoch": 1.5315970892378399,
+      "grad_norm": 134.8528535173368,
+      "learning_rate": 5e-06,
+      "loss": 1.4115,
+      "num_input_tokens_seen": 691280352,
+      "step": 3999
+    },
+    {
+      "epoch": 1.5315970892378399,
+      "loss": 1.5181658267974854,
+      "loss_ce": 0.18865513801574707,
+      "loss_iou": 0.6115785241127014,
+      "loss_num": 0.021240234375,
+      "loss_xval": 1.3295108079910278,
+      "num_input_tokens_seen": 691280352,
+      "step": 3999
+    },
+    {
+      "epoch": 1.5319800842589046,
+      "grad_norm": 108.508561955465,
+      "learning_rate": 5e-06,
+      "loss": 1.7889,
+      "num_input_tokens_seen": 691453432,
+      "step": 4000
+    },
+    {
+      "epoch": 1.5319800842589046,
+      "eval_websight_new_CIoU": 0.9476378560066223,
+      "eval_websight_new_GIoU": 0.9474721252918243,
+      "eval_websight_new_IoU": 0.9477640390396118,
+      "eval_websight_new_MAE_all": 0.004908971022814512,
+      "eval_websight_new_MAE_h": 0.0029728240333497524,
+      "eval_websight_new_MAE_w": 0.005848645232617855,
+      "eval_websight_new_MAE_x": 0.004454620066098869,
+      "eval_websight_new_MAE_y": 0.006359795108437538,
+      "eval_websight_new_NUM_probability": 0.2463129609823227,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.8320034742355347,
+      "eval_websight_new_loss_ce": 0.3229493796825409,
+      "eval_websight_new_loss_iou": 0.23782658576965332,
+      "eval_websight_new_loss_num": 0.005114555358886719,
+      "eval_websight_new_loss_xval": 0.5012307018041611,
+      "eval_websight_new_runtime": 56.7352,
+      "eval_websight_new_samples_per_second": 0.881,
+      "eval_websight_new_steps_per_second": 0.035,
+      "num_input_tokens_seen": 691453432,
+      "step": 4000
+    },
+    {
+      "epoch": 1.5319800842589046,
+      "eval_seeclick_CIoU": 0.6127658784389496,
+      "eval_seeclick_GIoU": 0.6294102370738983,
+      "eval_seeclick_IoU": 0.6650410592556,
+      "eval_seeclick_MAE_all": 0.07940063998103142,
+      "eval_seeclick_MAE_h": 0.037692490965127945,
+      "eval_seeclick_MAE_w": 0.12155372649431229,
+      "eval_seeclick_MAE_x": 0.11558730155229568,
+      "eval_seeclick_MAE_y": 0.04276901297271252,
+      "eval_seeclick_NUM_probability": 0.5139389932155609,
+      "eval_seeclick_inside_bbox": 0.8420138955116272,
+      "eval_seeclick_loss": 1.731054425239563,
+      "eval_seeclick_loss_ce": 0.2207578793168068,
+      "eval_seeclick_loss_iou": 0.6101343333721161,
+      "eval_seeclick_loss_num": 0.050933837890625,
+      "eval_seeclick_loss_xval": 1.4750294089317322,
+      "eval_seeclick_runtime": 86.5426,
+      "eval_seeclick_samples_per_second": 0.578,
+      "eval_seeclick_steps_per_second": 0.023,
+      "num_input_tokens_seen": 691453432,
+      "step": 4000
+    },
+    {
+      "epoch": 1.5319800842589046,
+      "eval_icons_CIoU": 0.8217889368534088,
+      "eval_icons_GIoU": 0.8226584494113922,
+      "eval_icons_IoU": 0.838074266910553,
+      "eval_icons_MAE_all": 0.03133867587894201,
+      "eval_icons_MAE_h": 0.0188470296561718,
+      "eval_icons_MAE_w": 0.041911233216524124,
+      "eval_icons_MAE_x": 0.04401544854044914,
+      "eval_icons_MAE_y": 0.020580998621881008,
+      "eval_icons_NUM_probability": 0.3711031377315521,
+      "eval_icons_inside_bbox": 0.9409722089767456,
+      "eval_icons_loss": 1.8929972648620605,
+      "eval_icons_loss_ce": 0.39243292808532715,
+      "eval_icons_loss_iou": 0.6932045519351959,
+      "eval_icons_loss_num": 0.02086639404296875,
+      "eval_icons_loss_xval": 1.4908402562141418,
+      "eval_icons_runtime": 86.0582,
+      "eval_icons_samples_per_second": 0.581,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 691453432,
+      "step": 4000
+    },
+    {
+      "epoch": 1.5319800842589046,
+      "loss": 1.930707573890686,
+      "loss_ce": 0.39242303371429443,
+      "loss_iou": 0.7135698199272156,
+      "loss_num": 0.022216796875,
+      "loss_xval": 1.5382846593856812,
+      "num_input_tokens_seen": 691453432,
+      "step": 4000
+    },
+    {
+      "epoch": 1.5323630792799694,
+      "grad_norm": 46.19858695867596,
+      "learning_rate": 5e-06,
+      "loss": 1.35,
+      "num_input_tokens_seen": 691626112,
+      "step": 4001
+    },
+    {
+      "epoch": 1.5323630792799694,
+      "loss": 1.2584831714630127,
+      "loss_ce": 0.1658729612827301,
+      "loss_iou": 0.4762977957725525,
+      "loss_num": 0.028076171875,
+      "loss_xval": 1.092610239982605,
+      "num_input_tokens_seen": 691626112,
+      "step": 4001
+    },
+    {
+      "epoch": 1.5327460743010342,
+      "grad_norm": 64.63637224563966,
+      "learning_rate": 5e-06,
+      "loss": 1.3761,
+      "num_input_tokens_seen": 691798888,
+      "step": 4002
+    },
+    {
+      "epoch": 1.5327460743010342,
+      "loss": 1.1979390382766724,
+      "loss_ce": 0.20775623619556427,
+      "loss_iou": 0.4690675437450409,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.9901828169822693,
+      "num_input_tokens_seen": 691798888,
+      "step": 4002
+    },
+    {
+      "epoch": 1.533129069322099,
+      "grad_norm": 115.07410986903477,
+      "learning_rate": 5e-06,
+      "loss": 1.193,
+      "num_input_tokens_seen": 691971736,
+      "step": 4003
+    },
+    {
+      "epoch": 1.533129069322099,
+      "loss": 1.0701923370361328,
+      "loss_ce": 0.12962406873703003,
+      "loss_iou": 0.4435812532901764,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.9405682682991028,
+      "num_input_tokens_seen": 691971736,
+      "step": 4003
+    },
+    {
+      "epoch": 1.5335120643431637,
+      "grad_norm": 170.75227991480216,
+      "learning_rate": 5e-06,
+      "loss": 1.5122,
+      "num_input_tokens_seen": 692144880,
+      "step": 4004
+    },
+    {
+      "epoch": 1.5335120643431637,
+      "loss": 1.5192091464996338,
+      "loss_ce": 0.15954117476940155,
+      "loss_iou": 0.6162812113761902,
+      "loss_num": 0.025390625,
+      "loss_xval": 1.3596681356430054,
+      "num_input_tokens_seen": 692144880,
+      "step": 4004
+    },
+    {
+      "epoch": 1.5338950593642284,
+      "grad_norm": 153.79374738053806,
+      "learning_rate": 5e-06,
+      "loss": 1.7093,
+      "num_input_tokens_seen": 692317432,
+      "step": 4005
+    },
+    {
+      "epoch": 1.5338950593642284,
+      "loss": 1.8067653179168701,
+      "loss_ce": 0.18908385932445526,
+      "loss_iou": 0.7398710250854492,
+      "loss_num": 0.027587890625,
+      "loss_xval": 1.6176815032958984,
+      "num_input_tokens_seen": 692317432,
+      "step": 4005
+    },
+    {
+      "epoch": 1.534278054385293,
+      "grad_norm": 99.87547366584295,
+      "learning_rate": 5e-06,
+      "loss": 1.2958,
+      "num_input_tokens_seen": 692490368,
+      "step": 4006
+    },
+    {
+      "epoch": 1.534278054385293,
+      "loss": 1.2684710025787354,
+      "loss_ce": 0.1644982248544693,
+      "loss_iou": 0.5198361873626709,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 1.1039729118347168,
+      "num_input_tokens_seen": 692490368,
+      "step": 4006
+    },
+    {
+      "epoch": 1.5346610494063577,
+      "grad_norm": 97.69049595070976,
+      "learning_rate": 5e-06,
+      "loss": 1.3975,
+      "num_input_tokens_seen": 692663400,
+      "step": 4007
+    },
+    {
+      "epoch": 1.5346610494063577,
+      "loss": 1.3261613845825195,
+      "loss_ce": 0.19734099507331848,
+      "loss_iou": 0.5318784713745117,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 1.1288204193115234,
+      "num_input_tokens_seen": 692663400,
+      "step": 4007
+    },
+    {
+      "epoch": 1.5350440444274225,
+      "grad_norm": 198.11477350605034,
+      "learning_rate": 5e-06,
+      "loss": 1.4736,
+      "num_input_tokens_seen": 692836616,
+      "step": 4008
+    },
+    {
+      "epoch": 1.5350440444274225,
+      "loss": 1.4241814613342285,
+      "loss_ce": 0.19302211701869965,
+      "loss_iou": 0.5766849517822266,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 1.2311592102050781,
+      "num_input_tokens_seen": 692836616,
+      "step": 4008
+    },
+    {
+      "epoch": 1.535427039448487,
+      "grad_norm": 157.87159707061483,
+      "learning_rate": 5e-06,
+      "loss": 1.612,
+      "num_input_tokens_seen": 693009600,
+      "step": 4009
+    },
+    {
+      "epoch": 1.535427039448487,
+      "loss": 1.4267079830169678,
+      "loss_ce": 0.17157082259655,
+      "loss_iou": 0.5778554081916809,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 1.2551370859146118,
+      "num_input_tokens_seen": 693009600,
+      "step": 4009
+    },
+    {
+      "epoch": 1.5358100344695518,
+      "grad_norm": 113.47818812880557,
+      "learning_rate": 5e-06,
+      "loss": 1.4546,
+      "num_input_tokens_seen": 693182488,
+      "step": 4010
+    },
+    {
+      "epoch": 1.5358100344695518,
+      "loss": 1.485398530960083,
+      "loss_ce": 0.1909884810447693,
+      "loss_iou": 0.6073033213615417,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 1.2944101095199585,
+      "num_input_tokens_seen": 693182488,
+      "step": 4010
+    },
+    {
+      "epoch": 1.5361930294906165,
+      "grad_norm": 111.18114453476046,
+      "learning_rate": 5e-06,
+      "loss": 1.5617,
+      "num_input_tokens_seen": 693355216,
+      "step": 4011
+    },
+    {
+      "epoch": 1.5361930294906165,
+      "loss": 1.6103756427764893,
+      "loss_ce": 0.17208832502365112,
+      "loss_iou": 0.662030041217804,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 1.438287377357483,
+      "num_input_tokens_seen": 693355216,
+      "step": 4011
+    },
+    {
+      "epoch": 1.5365760245116813,
+      "grad_norm": 78.26411281282405,
+      "learning_rate": 5e-06,
+      "loss": 1.7468,
+      "num_input_tokens_seen": 693527760,
+      "step": 4012
+    },
+    {
+      "epoch": 1.5365760245116813,
+      "loss": 1.6645243167877197,
+      "loss_ce": 0.15518814325332642,
+      "loss_iou": 0.6866138577461243,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 1.5093361139297485,
+      "num_input_tokens_seen": 693527760,
+      "step": 4012
+    },
+    {
+      "epoch": 1.536959019532746,
+      "grad_norm": 63.912879554020364,
+      "learning_rate": 5e-06,
+      "loss": 1.2575,
+      "num_input_tokens_seen": 693700288,
+      "step": 4013
+    },
+    {
+      "epoch": 1.536959019532746,
+      "loss": 1.220572829246521,
+      "loss_ce": 0.14960825443267822,
+      "loss_iou": 0.4911249876022339,
+      "loss_num": 0.017822265625,
+      "loss_xval": 1.0709645748138428,
+      "num_input_tokens_seen": 693700288,
+      "step": 4013
+    },
+    {
+      "epoch": 1.5373420145538108,
+      "grad_norm": 88.04675407486323,
+      "learning_rate": 5e-06,
+      "loss": 1.2159,
+      "num_input_tokens_seen": 693872880,
+      "step": 4014
+    },
+    {
+      "epoch": 1.5373420145538108,
+      "loss": 1.228543996810913,
+      "loss_ce": 0.1612216681241989,
+      "loss_iou": 0.5036013126373291,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 1.0673222541809082,
+      "num_input_tokens_seen": 693872880,
+      "step": 4014
+    },
+    {
+      "epoch": 1.5377250095748756,
+      "grad_norm": 156.12297498999627,
+      "learning_rate": 5e-06,
+      "loss": 1.5896,
+      "num_input_tokens_seen": 694045672,
+      "step": 4015
+    },
+    {
+      "epoch": 1.5377250095748756,
+      "loss": 1.554947853088379,
+      "loss_ce": 0.20863324403762817,
+      "loss_iou": 0.6180731058120728,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 1.3463146686553955,
+      "num_input_tokens_seen": 694045672,
+      "step": 4015
+    },
+    {
+      "epoch": 1.5381080045959403,
+      "grad_norm": 115.45476900828294,
+      "learning_rate": 5e-06,
+      "loss": 1.743,
+      "num_input_tokens_seen": 694218832,
+      "step": 4016
+    },
+    {
+      "epoch": 1.5381080045959403,
+      "loss": 1.7367010116577148,
+      "loss_ce": 0.2145659327507019,
+      "loss_iou": 0.6940509676933289,
+      "loss_num": 0.02685546875,
+      "loss_xval": 1.5221351385116577,
+      "num_input_tokens_seen": 694218832,
+      "step": 4016
+    },
+    {
+      "epoch": 1.538490999617005,
+      "grad_norm": 57.166597421566486,
+      "learning_rate": 5e-06,
+      "loss": 1.273,
+      "num_input_tokens_seen": 694391936,
+      "step": 4017
+    },
+    {
+      "epoch": 1.538490999617005,
+      "loss": 1.1809477806091309,
+      "loss_ce": 0.15838342905044556,
+      "loss_iou": 0.46304917335510254,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.02256441116333,
+      "num_input_tokens_seen": 694391936,
+      "step": 4017
+    },
+    {
+      "epoch": 1.5388739946380698,
+      "grad_norm": 93.45333582588987,
+      "learning_rate": 5e-06,
+      "loss": 1.2499,
+      "num_input_tokens_seen": 694565152,
+      "step": 4018
+    },
+    {
+      "epoch": 1.5388739946380698,
+      "loss": 1.3203922510147095,
+      "loss_ce": 0.17759379744529724,
+      "loss_iou": 0.5262026786804199,
+      "loss_num": 0.01806640625,
+      "loss_xval": 1.1427984237670898,
+      "num_input_tokens_seen": 694565152,
+      "step": 4018
+    },
+    {
+      "epoch": 1.5392569896591344,
+      "grad_norm": 95.92580225372755,
+      "learning_rate": 5e-06,
+      "loss": 1.6361,
+      "num_input_tokens_seen": 694738304,
+      "step": 4019
+    },
+    {
+      "epoch": 1.5392569896591344,
+      "loss": 1.5220376253128052,
+      "loss_ce": 0.15661263465881348,
+      "loss_iou": 0.6216772794723511,
+      "loss_num": 0.0244140625,
+      "loss_xval": 1.3654248714447021,
+      "num_input_tokens_seen": 694738304,
+      "step": 4019
+    },
+    {
+      "epoch": 1.5396399846801991,
+      "grad_norm": 67.68619604488842,
+      "learning_rate": 5e-06,
+      "loss": 1.1718,
+      "num_input_tokens_seen": 694911200,
+      "step": 4020
+    },
+    {
+      "epoch": 1.5396399846801991,
+      "loss": 1.244327187538147,
+      "loss_ce": 0.17027240991592407,
+      "loss_iou": 0.5018863677978516,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 1.0740547180175781,
+      "num_input_tokens_seen": 694911200,
+      "step": 4020
+    },
+    {
+      "epoch": 1.540022979701264,
+      "grad_norm": 182.7957511965751,
+      "learning_rate": 5e-06,
+      "loss": 1.504,
+      "num_input_tokens_seen": 695084080,
+      "step": 4021
+    },
+    {
+      "epoch": 1.540022979701264,
+      "loss": 1.5927867889404297,
+      "loss_ce": 0.1959582269191742,
+      "loss_iou": 0.6407970786094666,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 1.396828532218933,
+      "num_input_tokens_seen": 695084080,
+      "step": 4021
+    },
+    {
+      "epoch": 1.5404059747223287,
+      "grad_norm": 107.9979394300168,
+      "learning_rate": 5e-06,
+      "loss": 1.705,
+      "num_input_tokens_seen": 695257040,
+      "step": 4022
+    },
+    {
+      "epoch": 1.5404059747223287,
+      "loss": 1.6298996210098267,
+      "loss_ce": 0.13876935839653015,
+      "loss_iou": 0.6566979289054871,
+      "loss_num": 0.03564453125,
+      "loss_xval": 1.4549974203109741,
+      "num_input_tokens_seen": 695257040,
+      "step": 4022
+    },
+    {
+      "epoch": 1.5407889697433932,
+      "grad_norm": 109.67021807219069,
+      "learning_rate": 5e-06,
+      "loss": 1.2811,
+      "num_input_tokens_seen": 695429928,
+      "step": 4023
+    },
+    {
+      "epoch": 1.5407889697433932,
+      "loss": 1.1913628578186035,
+      "loss_ce": 0.16078069806098938,
+      "loss_iou": 0.48027217388153076,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 1.0305821895599365,
+      "num_input_tokens_seen": 695429928,
+      "step": 4023
+    },
+    {
+      "epoch": 1.541171964764458,
+      "grad_norm": 152.34037272652972,
+      "learning_rate": 5e-06,
+      "loss": 1.4865,
+      "num_input_tokens_seen": 695602808,
+      "step": 4024
+    },
+    {
+      "epoch": 1.541171964764458,
+      "loss": 1.4521257877349854,
+      "loss_ce": 0.1641998291015625,
+      "loss_iou": 0.6030999422073364,
+      "loss_num": 0.016357421875,
+      "loss_xval": 1.2879259586334229,
+      "num_input_tokens_seen": 695602808,
+      "step": 4024
+    },
+    {
+      "epoch": 1.5415549597855227,
+      "grad_norm": 119.86376995028226,
+      "learning_rate": 5e-06,
+      "loss": 1.4631,
+      "num_input_tokens_seen": 695775456,
+      "step": 4025
+    },
+    {
+      "epoch": 1.5415549597855227,
+      "loss": 1.4829367399215698,
+      "loss_ce": 0.17699642479419708,
+      "loss_iou": 0.6118019223213196,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 1.3059402704238892,
+      "num_input_tokens_seen": 695775456,
+      "step": 4025
+    },
+    {
+      "epoch": 1.5419379548065875,
+      "grad_norm": 199.6008949325988,
+      "learning_rate": 5e-06,
+      "loss": 1.6634,
+      "num_input_tokens_seen": 695948720,
+      "step": 4026
+    },
+    {
+      "epoch": 1.5419379548065875,
+      "loss": 1.6517527103424072,
+      "loss_ce": 0.1972631812095642,
+      "loss_iou": 0.6919969320297241,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 1.4544894695281982,
+      "num_input_tokens_seen": 695948720,
+      "step": 4026
+    },
+    {
+      "epoch": 1.5423209498276522,
+      "grad_norm": 173.84512058132393,
+      "learning_rate": 5e-06,
+      "loss": 2.1056,
+      "num_input_tokens_seen": 696121840,
+      "step": 4027
+    },
+    {
+      "epoch": 1.5423209498276522,
+      "loss": 2.1429200172424316,
+      "loss_ce": 0.20784983038902283,
+      "loss_iou": 0.9154415726661682,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 1.9350701570510864,
+      "num_input_tokens_seen": 696121840,
+      "step": 4027
+    },
+    {
+      "epoch": 1.542703944848717,
+      "grad_norm": 137.5637997552979,
+      "learning_rate": 5e-06,
+      "loss": 1.9552,
+      "num_input_tokens_seen": 696294640,
+      "step": 4028
+    },
+    {
+      "epoch": 1.542703944848717,
+      "loss": 2.0105209350585938,
+      "loss_ce": 0.1763545572757721,
+      "loss_iou": 0.8391107320785522,
+      "loss_num": 0.03125,
+      "loss_xval": 1.8341662883758545,
+      "num_input_tokens_seen": 696294640,
+      "step": 4028
+    },
+    {
+      "epoch": 1.5430869398697817,
+      "grad_norm": 95.63297736079416,
+      "learning_rate": 5e-06,
+      "loss": 1.7056,
+      "num_input_tokens_seen": 696468072,
+      "step": 4029
+    },
+    {
+      "epoch": 1.5430869398697817,
+      "loss": 1.6168577671051025,
+      "loss_ce": 0.19520872831344604,
+      "loss_iou": 0.6636748313903809,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 1.4216489791870117,
+      "num_input_tokens_seen": 696468072,
+      "step": 4029
+    },
+    {
+      "epoch": 1.5434699348908465,
+      "grad_norm": 104.9830239639552,
+      "learning_rate": 5e-06,
+      "loss": 1.4851,
+      "num_input_tokens_seen": 696641408,
+      "step": 4030
+    },
+    {
+      "epoch": 1.5434699348908465,
+      "loss": 1.4427449703216553,
+      "loss_ce": 0.1917538344860077,
+      "loss_iou": 0.5754314661026001,
+      "loss_num": 0.02001953125,
+      "loss_xval": 1.2509911060333252,
+      "num_input_tokens_seen": 696641408,
+      "step": 4030
+    },
+    {
+      "epoch": 1.5438529299119113,
+      "grad_norm": 112.33921601219592,
+      "learning_rate": 5e-06,
+      "loss": 1.777,
+      "num_input_tokens_seen": 696814304,
+      "step": 4031
+    },
+    {
+      "epoch": 1.5438529299119113,
+      "loss": 1.7711230516433716,
+      "loss_ce": 0.18419550359249115,
+      "loss_iou": 0.7331000566482544,
+      "loss_num": 0.024169921875,
+      "loss_xval": 1.5869276523590088,
+      "num_input_tokens_seen": 696814304,
+      "step": 4031
+    },
+    {
+      "epoch": 1.544235924932976,
+      "grad_norm": 223.81233438764835,
+      "learning_rate": 5e-06,
+      "loss": 1.3109,
+      "num_input_tokens_seen": 696987152,
+      "step": 4032
+    },
+    {
+      "epoch": 1.544235924932976,
+      "loss": 1.3282899856567383,
+      "loss_ce": 0.14142560958862305,
+      "loss_iou": 0.5531795024871826,
+      "loss_num": 0.01611328125,
+      "loss_xval": 1.1868643760681152,
+      "num_input_tokens_seen": 696987152,
+      "step": 4032
+    },
+    {
+      "epoch": 1.5446189199540405,
+      "grad_norm": 121.71460532194978,
+      "learning_rate": 5e-06,
+      "loss": 1.6271,
+      "num_input_tokens_seen": 697159976,
+      "step": 4033
+    },
+    {
+      "epoch": 1.5446189199540405,
+      "loss": 1.4645191431045532,
+      "loss_ce": 0.15616536140441895,
+      "loss_iou": 0.5996571779251099,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 1.3083536624908447,
+      "num_input_tokens_seen": 697159976,
+      "step": 4033
+    },
+    {
+      "epoch": 1.5450019149751053,
+      "grad_norm": 224.39751296901443,
+      "learning_rate": 5e-06,
+      "loss": 1.9026,
+      "num_input_tokens_seen": 697333144,
+      "step": 4034
+    },
+    {
+      "epoch": 1.5450019149751053,
+      "loss": 1.8723915815353394,
+      "loss_ce": 0.18856292963027954,
+      "loss_iou": 0.7223464846611023,
+      "loss_num": 0.0478515625,
+      "loss_xval": 1.6838287115097046,
+      "num_input_tokens_seen": 697333144,
+      "step": 4034
+    },
+    {
+      "epoch": 1.54538490999617,
+      "grad_norm": 86.38781709948974,
+      "learning_rate": 5e-06,
+      "loss": 1.5271,
+      "num_input_tokens_seen": 697505984,
+      "step": 4035
+    },
+    {
+      "epoch": 1.54538490999617,
+      "loss": 1.4957278966903687,
+      "loss_ce": 0.1962394416332245,
+      "loss_iou": 0.5999701023101807,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 1.2994885444641113,
+      "num_input_tokens_seen": 697505984,
+      "step": 4035
+    },
+    {
+      "epoch": 1.5457679050172348,
+      "grad_norm": 59.55226856790362,
+      "learning_rate": 5e-06,
+      "loss": 1.4834,
+      "num_input_tokens_seen": 697679176,
+      "step": 4036
+    },
+    {
+      "epoch": 1.5457679050172348,
+      "loss": 1.5904284715652466,
+      "loss_ce": 0.16325372457504272,
+      "loss_iou": 0.6481729745864868,
+      "loss_num": 0.026123046875,
+      "loss_xval": 1.4271748065948486,
+      "num_input_tokens_seen": 697679176,
+      "step": 4036
+    },
+    {
+      "epoch": 1.5461509000382994,
+      "grad_norm": 123.63629350762896,
+      "learning_rate": 5e-06,
+      "loss": 1.2917,
+      "num_input_tokens_seen": 697852312,
+      "step": 4037
+    },
+    {
+      "epoch": 1.5461509000382994,
+      "loss": 1.2972408533096313,
+      "loss_ce": 0.17061948776245117,
+      "loss_iou": 0.5271702408790588,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 1.1266213655471802,
+      "num_input_tokens_seen": 697852312,
+      "step": 4037
+    },
+    {
+      "epoch": 1.5465338950593641,
+      "grad_norm": 137.7346544970482,
+      "learning_rate": 5e-06,
+      "loss": 1.3046,
+      "num_input_tokens_seen": 698025168,
+      "step": 4038
+    },
+    {
+      "epoch": 1.5465338950593641,
+      "loss": 1.2661736011505127,
+      "loss_ce": 0.14972297847270966,
+      "loss_iou": 0.5060555338859558,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 1.1164506673812866,
+      "num_input_tokens_seen": 698025168,
+      "step": 4038
+    },
+    {
+      "epoch": 1.5469168900804289,
+      "grad_norm": 159.85474924022307,
+      "learning_rate": 5e-06,
+      "loss": 1.5059,
+      "num_input_tokens_seen": 698198272,
+      "step": 4039
+    },
+    {
+      "epoch": 1.5469168900804289,
+      "loss": 1.4872212409973145,
+      "loss_ce": 0.1583525836467743,
+      "loss_iou": 0.6273249387741089,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 1.3288686275482178,
+      "num_input_tokens_seen": 698198272,
+      "step": 4039
+    },
+    {
+      "epoch": 1.5472998851014936,
+      "grad_norm": 103.27065357471398,
+      "learning_rate": 5e-06,
+      "loss": 1.3301,
+      "num_input_tokens_seen": 698371192,
+      "step": 4040
+    },
+    {
+      "epoch": 1.5472998851014936,
+      "loss": 1.3318990468978882,
+      "loss_ce": 0.20116597414016724,
+      "loss_iou": 0.5277841091156006,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 1.1307330131530762,
+      "num_input_tokens_seen": 698371192,
+      "step": 4040
+    },
+    {
+      "epoch": 1.5476828801225584,
+      "grad_norm": 142.5808620696021,
+      "learning_rate": 5e-06,
+      "loss": 1.3475,
+      "num_input_tokens_seen": 698543920,
+      "step": 4041
+    },
+    {
+      "epoch": 1.5476828801225584,
+      "loss": 1.4051227569580078,
+      "loss_ce": 0.1359701156616211,
+      "loss_iou": 0.5977416038513184,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 1.2691526412963867,
+      "num_input_tokens_seen": 698543920,
+      "step": 4041
+    },
+    {
+      "epoch": 1.5480658751436231,
+      "grad_norm": 118.78991913862968,
+      "learning_rate": 5e-06,
+      "loss": 1.5009,
+      "num_input_tokens_seen": 698716568,
+      "step": 4042
+    },
+    {
+      "epoch": 1.5480658751436231,
+      "loss": 1.5289782285690308,
+      "loss_ce": 0.17239195108413696,
+      "loss_iou": 0.6370944380760193,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 1.3565863370895386,
+      "num_input_tokens_seen": 698716568,
+      "step": 4042
+    },
+    {
+      "epoch": 1.548448870164688,
+      "grad_norm": 161.89217274162206,
+      "learning_rate": 5e-06,
+      "loss": 1.6185,
+      "num_input_tokens_seen": 698889800,
+      "step": 4043
+    },
+    {
+      "epoch": 1.548448870164688,
+      "loss": 1.551445484161377,
+      "loss_ce": 0.16810211539268494,
+      "loss_iou": 0.6364959478378296,
+      "loss_num": 0.02197265625,
+      "loss_xval": 1.3833434581756592,
+      "num_input_tokens_seen": 698889800,
+      "step": 4043
+    },
+    {
+      "epoch": 1.5488318651857527,
+      "grad_norm": 532.9776921227485,
+      "learning_rate": 5e-06,
+      "loss": 1.5141,
+      "num_input_tokens_seen": 699062904,
+      "step": 4044
+    },
+    {
+      "epoch": 1.5488318651857527,
+      "loss": 1.4322932958602905,
+      "loss_ce": 0.1805468648672104,
+      "loss_iou": 0.574313759803772,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 1.251746416091919,
+      "num_input_tokens_seen": 699062904,
+      "step": 4044
+    },
+    {
+      "epoch": 1.5492148602068174,
+      "grad_norm": 150.11870950442585,
+      "learning_rate": 5e-06,
+      "loss": 1.698,
+      "num_input_tokens_seen": 699235832,
+      "step": 4045
+    },
+    {
+      "epoch": 1.5492148602068174,
+      "loss": 1.642263412475586,
+      "loss_ce": 0.15046949684619904,
+      "loss_iou": 0.6751571893692017,
+      "loss_num": 0.0283203125,
+      "loss_xval": 1.4917938709259033,
+      "num_input_tokens_seen": 699235832,
+      "step": 4045
+    },
+    {
+      "epoch": 1.5495978552278822,
+      "grad_norm": 191.6479875586437,
+      "learning_rate": 5e-06,
+      "loss": 1.3066,
+      "num_input_tokens_seen": 699408488,
+      "step": 4046
+    },
+    {
+      "epoch": 1.5495978552278822,
+      "loss": 1.3165333271026611,
+      "loss_ce": 0.13863667845726013,
+      "loss_iou": 0.5390978455543518,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 1.1778966188430786,
+      "num_input_tokens_seen": 699408488,
+      "step": 4046
+    },
+    {
+      "epoch": 1.5499808502489467,
+      "grad_norm": 79.09853614584158,
+      "learning_rate": 5e-06,
+      "loss": 1.8071,
+      "num_input_tokens_seen": 699581464,
+      "step": 4047
+    },
+    {
+      "epoch": 1.5499808502489467,
+      "loss": 1.7550569772720337,
+      "loss_ce": 0.14108368754386902,
+      "loss_iou": 0.7232463955879211,
+      "loss_num": 0.033447265625,
+      "loss_xval": 1.6139732599258423,
+      "num_input_tokens_seen": 699581464,
+      "step": 4047
+    },
+    {
+      "epoch": 1.5503638452700115,
+      "grad_norm": 86.21733289719229,
+      "learning_rate": 5e-06,
+      "loss": 1.2929,
+      "num_input_tokens_seen": 699754040,
+      "step": 4048
+    },
+    {
+      "epoch": 1.5503638452700115,
+      "loss": 1.265880823135376,
+      "loss_ce": 0.16111326217651367,
+      "loss_iou": 0.5023578405380249,
+      "loss_num": 0.02001953125,
+      "loss_xval": 1.1047675609588623,
+      "num_input_tokens_seen": 699754040,
+      "step": 4048
+    },
+    {
+      "epoch": 1.5507468402910762,
+      "grad_norm": 163.46510712305485,
+      "learning_rate": 5e-06,
+      "loss": 1.5917,
+      "num_input_tokens_seen": 699926840,
+      "step": 4049
+    },
+    {
+      "epoch": 1.5507468402910762,
+      "loss": 1.6322741508483887,
+      "loss_ce": 0.16537222266197205,
+      "loss_iou": 0.6964025497436523,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 1.4669017791748047,
+      "num_input_tokens_seen": 699926840,
+      "step": 4049
+    },
+    {
+      "epoch": 1.551129835312141,
+      "grad_norm": 96.6453890523453,
+      "learning_rate": 5e-06,
+      "loss": 1.631,
+      "num_input_tokens_seen": 700099976,
+      "step": 4050
+    },
+    {
+      "epoch": 1.551129835312141,
+      "loss": 1.5915098190307617,
+      "loss_ce": 0.17163625359535217,
+      "loss_iou": 0.6488100290298462,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 1.4198734760284424,
+      "num_input_tokens_seen": 700099976,
+      "step": 4050
+    },
+    {
+      "epoch": 1.5515128303332055,
+      "grad_norm": 62.40844671149482,
+      "learning_rate": 5e-06,
+      "loss": 1.3397,
+      "num_input_tokens_seen": 700272632,
+      "step": 4051
+    },
+    {
+      "epoch": 1.5515128303332055,
+      "loss": 1.4549493789672852,
+      "loss_ce": 0.19186116755008698,
+      "loss_iou": 0.5998668074607849,
+      "loss_num": 0.0126953125,
+      "loss_xval": 1.2630881071090698,
+      "num_input_tokens_seen": 700272632,
+      "step": 4051
+    },
+    {
+      "epoch": 1.5518958253542703,
+      "grad_norm": 126.44252936551133,
+      "learning_rate": 5e-06,
+      "loss": 1.3642,
+      "num_input_tokens_seen": 700445696,
+      "step": 4052
+    },
+    {
+      "epoch": 1.5518958253542703,
+      "loss": 1.4157166481018066,
+      "loss_ce": 0.1703338921070099,
+      "loss_iou": 0.5618698596954346,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 1.2453827857971191,
+      "num_input_tokens_seen": 700445696,
+      "step": 4052
+    },
+    {
+      "epoch": 1.552278820375335,
+      "grad_norm": 122.92057155810977,
+      "learning_rate": 5e-06,
+      "loss": 1.3993,
+      "num_input_tokens_seen": 700618784,
+      "step": 4053
+    },
+    {
+      "epoch": 1.552278820375335,
+      "loss": 1.3907073736190796,
+      "loss_ce": 0.17245326936244965,
+      "loss_iou": 0.5662040710449219,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 1.2182540893554688,
+      "num_input_tokens_seen": 700618784,
+      "step": 4053
+    },
+    {
+      "epoch": 1.5526618153963998,
+      "grad_norm": 148.87953002659964,
+      "learning_rate": 5e-06,
+      "loss": 1.6484,
+      "num_input_tokens_seen": 700792176,
+      "step": 4054
+    },
+    {
+      "epoch": 1.5526618153963998,
+      "loss": 1.5574525594711304,
+      "loss_ce": 0.10910694301128387,
+      "loss_iou": 0.6690123081207275,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 1.44834566116333,
+      "num_input_tokens_seen": 700792176,
+      "step": 4054
+    },
+    {
+      "epoch": 1.5530448104174646,
+      "grad_norm": 106.30489665811756,
+      "learning_rate": 5e-06,
+      "loss": 1.9592,
+      "num_input_tokens_seen": 700964728,
+      "step": 4055
+    },
+    {
+      "epoch": 1.5530448104174646,
+      "loss": 1.8856539726257324,
+      "loss_ce": 0.16048921644687653,
+      "loss_iou": 0.7678557634353638,
+      "loss_num": 0.037841796875,
+      "loss_xval": 1.7251646518707275,
+      "num_input_tokens_seen": 700964728,
+      "step": 4055
+    },
+    {
+      "epoch": 1.5534278054385293,
+      "grad_norm": 192.2419175697598,
+      "learning_rate": 5e-06,
+      "loss": 1.5801,
+      "num_input_tokens_seen": 701138128,
+      "step": 4056
+    },
+    {
+      "epoch": 1.5534278054385293,
+      "loss": 1.5525187253952026,
+      "loss_ce": 0.22270707786083221,
+      "loss_iou": 0.6143839955329895,
+      "loss_num": 0.020263671875,
+      "loss_xval": 1.329811692237854,
+      "num_input_tokens_seen": 701138128,
+      "step": 4056
+    },
+    {
+      "epoch": 1.553810800459594,
+      "grad_norm": 166.7682208271423,
+      "learning_rate": 5e-06,
+      "loss": 1.7475,
+      "num_input_tokens_seen": 701310848,
+      "step": 4057
+    },
+    {
+      "epoch": 1.553810800459594,
+      "loss": 1.866917610168457,
+      "loss_ce": 0.14354538917541504,
+      "loss_iou": 0.807303786277771,
+      "loss_num": 0.021728515625,
+      "loss_xval": 1.723372220993042,
+      "num_input_tokens_seen": 701310848,
+      "step": 4057
+    },
+    {
+      "epoch": 1.5541937954806588,
+      "grad_norm": 142.03188244107125,
+      "learning_rate": 5e-06,
+      "loss": 2.334,
+      "num_input_tokens_seen": 701483760,
+      "step": 4058
+    },
+    {
+      "epoch": 1.5541937954806588,
+      "loss": 2.191911220550537,
+      "loss_ce": 0.22595535218715668,
+      "loss_iou": 0.8638982772827148,
+      "loss_num": 0.047607421875,
+      "loss_xval": 1.9659557342529297,
+      "num_input_tokens_seen": 701483760,
+      "step": 4058
+    },
+    {
+      "epoch": 1.5545767905017236,
+      "grad_norm": 88.74749877146184,
+      "learning_rate": 5e-06,
+      "loss": 1.3027,
+      "num_input_tokens_seen": 701656888,
+      "step": 4059
+    },
+    {
+      "epoch": 1.5545767905017236,
+      "loss": 1.2640423774719238,
+      "loss_ce": 0.17497318983078003,
+      "loss_iou": 0.5103548765182495,
+      "loss_num": 0.013671875,
+      "loss_xval": 1.089069128036499,
+      "num_input_tokens_seen": 701656888,
+      "step": 4059
+    },
+    {
+      "epoch": 1.5549597855227884,
+      "grad_norm": 155.0285265772275,
+      "learning_rate": 5e-06,
+      "loss": 1.5255,
+      "num_input_tokens_seen": 701826400,
+      "step": 4060
+    },
+    {
+      "epoch": 1.5549597855227884,
+      "loss": 1.3491778373718262,
+      "loss_ce": 0.15473467111587524,
+      "loss_iou": 0.5659410953521729,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 1.1944432258605957,
+      "num_input_tokens_seen": 701826400,
+      "step": 4060
+    },
+    {
+      "epoch": 1.5553427805438529,
+      "grad_norm": 132.81623454348352,
+      "learning_rate": 5e-06,
+      "loss": 1.6386,
+      "num_input_tokens_seen": 701999288,
+      "step": 4061
+    },
+    {
+      "epoch": 1.5553427805438529,
+      "loss": 1.6399247646331787,
+      "loss_ce": 0.15263910591602325,
+      "loss_iou": 0.6787319183349609,
+      "loss_num": 0.02587890625,
+      "loss_xval": 1.4872856140136719,
+      "num_input_tokens_seen": 701999288,
+      "step": 4061
+    },
+    {
+      "epoch": 1.5557257755649176,
+      "grad_norm": 132.6690892951261,
+      "learning_rate": 5e-06,
+      "loss": 1.2982,
+      "num_input_tokens_seen": 702172496,
+      "step": 4062
+    },
+    {
+      "epoch": 1.5557257755649176,
+      "loss": 1.3524725437164307,
+      "loss_ce": 0.18340986967086792,
+      "loss_iou": 0.5478798151016235,
+      "loss_num": 0.0146484375,
+      "loss_xval": 1.169062852859497,
+      "num_input_tokens_seen": 702172496,
+      "step": 4062
+    },
+    {
+      "epoch": 1.5561087705859824,
+      "grad_norm": 144.18941232089801,
+      "learning_rate": 5e-06,
+      "loss": 1.7399,
+      "num_input_tokens_seen": 702345424,
+      "step": 4063
+    },
+    {
+      "epoch": 1.5561087705859824,
+      "loss": 1.7615002393722534,
+      "loss_ce": 0.1590670645236969,
+      "loss_iou": 0.749382495880127,
+      "loss_num": 0.020751953125,
+      "loss_xval": 1.602433204650879,
+      "num_input_tokens_seen": 702345424,
+      "step": 4063
+    },
+    {
+      "epoch": 1.5564917656070472,
+      "grad_norm": 168.94267630592827,
+      "learning_rate": 5e-06,
+      "loss": 1.8126,
+      "num_input_tokens_seen": 702518808,
+      "step": 4064
+    },
+    {
+      "epoch": 1.5564917656070472,
+      "loss": 1.8514591455459595,
+      "loss_ce": 0.2286132425069809,
+      "loss_iou": 0.7452913522720337,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 1.6228458881378174,
+      "num_input_tokens_seen": 702518808,
+      "step": 4064
+    },
+    {
+      "epoch": 1.5568747606281117,
+      "grad_norm": 144.1567964190179,
+      "learning_rate": 5e-06,
+      "loss": 1.5602,
+      "num_input_tokens_seen": 702688384,
+      "step": 4065
+    },
+    {
+      "epoch": 1.5568747606281117,
+      "loss": 1.4671590328216553,
+      "loss_ce": 0.1535789668560028,
+      "loss_iou": 0.6142637729644775,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 1.31358003616333,
+      "num_input_tokens_seen": 702688384,
+      "step": 4065
+    },
+    {
+      "epoch": 1.5572577556491765,
+      "grad_norm": 248.88925403808216,
+      "learning_rate": 5e-06,
+      "loss": 1.7117,
+      "num_input_tokens_seen": 702861544,
+      "step": 4066
+    },
+    {
+      "epoch": 1.5572577556491765,
+      "loss": 1.5558196306228638,
+      "loss_ce": 0.18789410591125488,
+      "loss_iou": 0.6358518600463867,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.3679256439208984,
+      "num_input_tokens_seen": 702861544,
+      "step": 4066
+    },
+    {
+      "epoch": 1.5576407506702412,
+      "grad_norm": 148.6132169700537,
+      "learning_rate": 5e-06,
+      "loss": 1.9878,
+      "num_input_tokens_seen": 703034504,
+      "step": 4067
+    },
+    {
+      "epoch": 1.5576407506702412,
+      "loss": 2.099031925201416,
+      "loss_ce": 0.17101332545280457,
+      "loss_iou": 0.8806352615356445,
+      "loss_num": 0.033447265625,
+      "loss_xval": 1.928018569946289,
+      "num_input_tokens_seen": 703034504,
+      "step": 4067
+    },
+    {
+      "epoch": 1.558023745691306,
+      "grad_norm": 244.41205692724475,
+      "learning_rate": 5e-06,
+      "loss": 2.0029,
+      "num_input_tokens_seen": 703207136,
+      "step": 4068
+    },
+    {
+      "epoch": 1.558023745691306,
+      "loss": 2.001335620880127,
+      "loss_ce": 0.2015422284603119,
+      "loss_iou": 0.8007146120071411,
+      "loss_num": 0.03955078125,
+      "loss_xval": 1.7997934818267822,
+      "num_input_tokens_seen": 703207136,
+      "step": 4068
+    },
+    {
+      "epoch": 1.5584067407123707,
+      "grad_norm": 265.45685510750087,
+      "learning_rate": 5e-06,
+      "loss": 1.5545,
+      "num_input_tokens_seen": 703380376,
+      "step": 4069
+    },
+    {
+      "epoch": 1.5584067407123707,
+      "loss": 1.5851879119873047,
+      "loss_ce": 0.16373172402381897,
+      "loss_iou": 0.6626628637313843,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.4214560985565186,
+      "num_input_tokens_seen": 703380376,
+      "step": 4069
+    },
+    {
+      "epoch": 1.5587897357334355,
+      "grad_norm": 127.7041350680125,
+      "learning_rate": 5e-06,
+      "loss": 1.4033,
+      "num_input_tokens_seen": 703553280,
+      "step": 4070
+    },
+    {
+      "epoch": 1.5587897357334355,
+      "loss": 1.444726586341858,
+      "loss_ce": 0.1692478358745575,
+      "loss_iou": 0.5924741625785828,
+      "loss_num": 0.01806640625,
+      "loss_xval": 1.275478720664978,
+      "num_input_tokens_seen": 703553280,
+      "step": 4070
+    },
+    {
+      "epoch": 1.5591727307545002,
+      "grad_norm": 142.4465541088045,
+      "learning_rate": 5e-06,
+      "loss": 1.594,
+      "num_input_tokens_seen": 703726168,
+      "step": 4071
+    },
+    {
+      "epoch": 1.5591727307545002,
+      "loss": 1.6197677850723267,
+      "loss_ce": 0.1557142734527588,
+      "loss_iou": 0.6873642802238464,
+      "loss_num": 0.017822265625,
+      "loss_xval": 1.4640535116195679,
+      "num_input_tokens_seen": 703726168,
+      "step": 4071
+    },
+    {
+      "epoch": 1.559555725775565,
+      "grad_norm": 127.87412145966195,
+      "learning_rate": 5e-06,
+      "loss": 1.663,
+      "num_input_tokens_seen": 703899496,
+      "step": 4072
+    },
+    {
+      "epoch": 1.559555725775565,
+      "loss": 1.7447503805160522,
+      "loss_ce": 0.23163306713104248,
+      "loss_iou": 0.7167637348175049,
+      "loss_num": 0.015869140625,
+      "loss_xval": 1.5131173133850098,
+      "num_input_tokens_seen": 703899496,
+      "step": 4072
+    },
+    {
+      "epoch": 1.5599387207966298,
+      "grad_norm": 204.54413428405158,
+      "learning_rate": 5e-06,
+      "loss": 1.6549,
+      "num_input_tokens_seen": 704072232,
+      "step": 4073
+    },
+    {
+      "epoch": 1.5599387207966298,
+      "loss": 1.8466553688049316,
+      "loss_ce": 0.1531267762184143,
+      "loss_iou": 0.8081443309783936,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 1.693528652191162,
+      "num_input_tokens_seen": 704072232,
+      "step": 4073
+    },
+    {
+      "epoch": 1.5603217158176945,
+      "grad_norm": 155.9762483650282,
+      "learning_rate": 5e-06,
+      "loss": 1.7382,
+      "num_input_tokens_seen": 704245248,
+      "step": 4074
+    },
+    {
+      "epoch": 1.5603217158176945,
+      "loss": 1.7038124799728394,
+      "loss_ce": 0.16794651746749878,
+      "loss_iou": 0.7059518098831177,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 1.5358660221099854,
+      "num_input_tokens_seen": 704245248,
+      "step": 4074
+    },
+    {
+      "epoch": 1.560704710838759,
+      "grad_norm": 142.34730879508822,
+      "learning_rate": 5e-06,
+      "loss": 2.0228,
+      "num_input_tokens_seen": 704418272,
+      "step": 4075
+    },
+    {
+      "epoch": 1.560704710838759,
+      "loss": 1.9907587766647339,
+      "loss_ce": 0.14460095763206482,
+      "loss_iou": 0.8328078985214233,
+      "loss_num": 0.0361328125,
+      "loss_xval": 1.8461577892303467,
+      "num_input_tokens_seen": 704418272,
+      "step": 4075
+    },
+    {
+      "epoch": 1.5610877058598238,
+      "grad_norm": 88.85904020323856,
+      "learning_rate": 5e-06,
+      "loss": 1.655,
+      "num_input_tokens_seen": 704591168,
+      "step": 4076
+    },
+    {
+      "epoch": 1.5610877058598238,
+      "loss": 1.9035416841506958,
+      "loss_ce": 0.18996544182300568,
+      "loss_iou": 0.777259349822998,
+      "loss_num": 0.03173828125,
+      "loss_xval": 1.713576316833496,
+      "num_input_tokens_seen": 704591168,
+      "step": 4076
+    },
+    {
+      "epoch": 1.5614707008808886,
+      "grad_norm": 120.86826739205884,
+      "learning_rate": 5e-06,
+      "loss": 1.2974,
+      "num_input_tokens_seen": 704764128,
+      "step": 4077
+    },
+    {
+      "epoch": 1.5614707008808886,
+      "loss": 1.17726469039917,
+      "loss_ce": 0.1245112419128418,
+      "loss_iou": 0.48397257924079895,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 1.0527534484863281,
+      "num_input_tokens_seen": 704764128,
+      "step": 4077
+    },
+    {
+      "epoch": 1.561853695901953,
+      "grad_norm": 116.76087367670702,
+      "learning_rate": 5e-06,
+      "loss": 2.2354,
+      "num_input_tokens_seen": 704937136,
+      "step": 4078
+    },
+    {
+      "epoch": 1.561853695901953,
+      "loss": 2.2421231269836426,
+      "loss_ce": 0.17673224210739136,
+      "loss_iou": 0.9810445308685303,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 2.0602335929870605,
+      "num_input_tokens_seen": 704937136,
+      "step": 4078
+    },
+    {
+      "epoch": 1.5622366909230179,
+      "grad_norm": 129.41712769878112,
+      "learning_rate": 5e-06,
+      "loss": 1.6554,
+      "num_input_tokens_seen": 705109776,
+      "step": 4079
+    },
+    {
+      "epoch": 1.5622366909230179,
+      "loss": 1.6967781782150269,
+      "loss_ce": 0.12966777384281158,
+      "loss_iou": 0.7334758639335632,
+      "loss_num": 0.02001953125,
+      "loss_xval": 1.5671104192733765,
+      "num_input_tokens_seen": 705109776,
+      "step": 4079
+    },
+    {
+      "epoch": 1.5626196859440826,
+      "grad_norm": 147.87306043107048,
+      "learning_rate": 5e-06,
+      "loss": 1.4896,
+      "num_input_tokens_seen": 705282960,
+      "step": 4080
+    },
+    {
+      "epoch": 1.5626196859440826,
+      "loss": 1.5170713663101196,
+      "loss_ce": 0.14922286570072174,
+      "loss_iou": 0.646219789981842,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 1.367848515510559,
+      "num_input_tokens_seen": 705282960,
+      "step": 4080
+    },
+    {
+      "epoch": 1.5630026809651474,
+      "grad_norm": 227.72109178509598,
+      "learning_rate": 5e-06,
+      "loss": 2.0063,
+      "num_input_tokens_seen": 705455600,
+      "step": 4081
+    },
+    {
+      "epoch": 1.5630026809651474,
+      "loss": 2.1332335472106934,
+      "loss_ce": 0.13369309902191162,
+      "loss_iou": 0.9299460053443909,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 1.9995403289794922,
+      "num_input_tokens_seen": 705455600,
+      "step": 4081
+    },
+    {
+      "epoch": 1.5633856759862121,
+      "grad_norm": 70.03848448318948,
+      "learning_rate": 5e-06,
+      "loss": 1.2402,
+      "num_input_tokens_seen": 705628872,
+      "step": 4082
+    },
+    {
+      "epoch": 1.5633856759862121,
+      "loss": 1.375079870223999,
+      "loss_ce": 0.17386046051979065,
+      "loss_iou": 0.5470055937767029,
+      "loss_num": 0.021484375,
+      "loss_xval": 1.2012194395065308,
+      "num_input_tokens_seen": 705628872,
+      "step": 4082
+    },
+    {
+      "epoch": 1.563768671007277,
+      "grad_norm": 135.75341219091302,
+      "learning_rate": 5e-06,
+      "loss": 1.1816,
+      "num_input_tokens_seen": 705802112,
+      "step": 4083
+    },
+    {
+      "epoch": 1.563768671007277,
+      "loss": 1.241959571838379,
+      "loss_ce": 0.1726619303226471,
+      "loss_iou": 0.5051383376121521,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 1.0692976713180542,
+      "num_input_tokens_seen": 705802112,
+      "step": 4083
+    },
+    {
+      "epoch": 1.5641516660283417,
+      "grad_norm": 224.00194830761635,
+      "learning_rate": 5e-06,
+      "loss": 1.7552,
+      "num_input_tokens_seen": 705975256,
+      "step": 4084
+    },
+    {
+      "epoch": 1.5641516660283417,
+      "loss": 1.9128923416137695,
+      "loss_ce": 0.15855154395103455,
+      "loss_iou": 0.8008764982223511,
+      "loss_num": 0.030517578125,
+      "loss_xval": 1.7543408870697021,
+      "num_input_tokens_seen": 705975256,
+      "step": 4084
+    },
+    {
+      "epoch": 1.5645346610494064,
+      "grad_norm": 88.2324151297439,
+      "learning_rate": 5e-06,
+      "loss": 1.3487,
+      "num_input_tokens_seen": 706148112,
+      "step": 4085
+    },
+    {
+      "epoch": 1.5645346610494064,
+      "loss": 1.2271016836166382,
+      "loss_ce": 0.11906194686889648,
+      "loss_iou": 0.5063360929489136,
+      "loss_num": 0.01904296875,
+      "loss_xval": 1.1018750667572021,
+      "num_input_tokens_seen": 706148112,
+      "step": 4085
+    },
+    {
+      "epoch": 1.5649176560704712,
+      "grad_norm": 129.6185017710986,
+      "learning_rate": 5e-06,
+      "loss": 1.295,
+      "num_input_tokens_seen": 706320832,
+      "step": 4086
+    },
+    {
+      "epoch": 1.5649176560704712,
+      "loss": 1.3663123846054077,
+      "loss_ce": 0.135855033993721,
+      "loss_iou": 0.5584354400634766,
+      "loss_num": 0.022705078125,
+      "loss_xval": 1.2304573059082031,
+      "num_input_tokens_seen": 706320832,
+      "step": 4086
+    },
+    {
+      "epoch": 1.565300651091536,
+      "grad_norm": 189.63905035964822,
+      "learning_rate": 5e-06,
+      "loss": 1.812,
+      "num_input_tokens_seen": 706490760,
+      "step": 4087
+    },
+    {
+      "epoch": 1.565300651091536,
+      "loss": 1.9216313362121582,
+      "loss_ce": 0.15427327156066895,
+      "loss_iou": 0.8071714639663696,
+      "loss_num": 0.030517578125,
+      "loss_xval": 1.7673580646514893,
+      "num_input_tokens_seen": 706490760,
+      "step": 4087
+    },
+    {
+      "epoch": 1.5656836461126007,
+      "grad_norm": 112.03550902850772,
+      "learning_rate": 5e-06,
+      "loss": 1.3883,
+      "num_input_tokens_seen": 706663896,
+      "step": 4088
+    },
+    {
+      "epoch": 1.5656836461126007,
+      "loss": 1.2862303256988525,
+      "loss_ce": 0.1765056997537613,
+      "loss_iou": 0.5050118565559387,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 1.1097246408462524,
+      "num_input_tokens_seen": 706663896,
+      "step": 4088
+    },
+    {
+      "epoch": 1.5660666411336652,
+      "grad_norm": 264.38863985724606,
+      "learning_rate": 5e-06,
+      "loss": 1.3277,
+      "num_input_tokens_seen": 706836520,
+      "step": 4089
+    },
+    {
+      "epoch": 1.5660666411336652,
+      "loss": 1.313084602355957,
+      "loss_ce": 0.13801221549510956,
+      "loss_iou": 0.5486568212509155,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 1.175072431564331,
+      "num_input_tokens_seen": 706836520,
+      "step": 4089
+    },
+    {
+      "epoch": 1.56644963615473,
+      "grad_norm": 231.41789883023728,
+      "learning_rate": 5e-06,
+      "loss": 1.5801,
+      "num_input_tokens_seen": 707009448,
+      "step": 4090
+    },
+    {
+      "epoch": 1.56644963615473,
+      "loss": 1.5929927825927734,
+      "loss_ce": 0.17305263876914978,
+      "loss_iou": 0.6497131586074829,
+      "loss_num": 0.024169921875,
+      "loss_xval": 1.4199402332305908,
+      "num_input_tokens_seen": 707009448,
+      "step": 4090
+    },
+    {
+      "epoch": 1.5668326311757947,
+      "grad_norm": 358.92833373521785,
+      "learning_rate": 5e-06,
+      "loss": 2.3622,
+      "num_input_tokens_seen": 707182544,
+      "step": 4091
+    },
+    {
+      "epoch": 1.5668326311757947,
+      "loss": 2.406782865524292,
+      "loss_ce": 0.17092815041542053,
+      "loss_iou": 0.9906079769134521,
+      "loss_num": 0.051025390625,
+      "loss_xval": 2.2358546257019043,
+      "num_input_tokens_seen": 707182544,
+      "step": 4091
+    },
+    {
+      "epoch": 1.5672156261968593,
+      "grad_norm": 141.66751245145667,
+      "learning_rate": 5e-06,
+      "loss": 2.0203,
+      "num_input_tokens_seen": 707355544,
+      "step": 4092
+    },
+    {
+      "epoch": 1.5672156261968593,
+      "loss": 2.183703899383545,
+      "loss_ce": 0.1479608416557312,
+      "loss_iou": 0.8810917139053345,
+      "loss_num": 0.0546875,
+      "loss_xval": 2.035742998123169,
+      "num_input_tokens_seen": 707355544,
+      "step": 4092
+    },
+    {
+      "epoch": 1.567598621217924,
+      "grad_norm": 123.3936171504265,
+      "learning_rate": 5e-06,
+      "loss": 1.5455,
+      "num_input_tokens_seen": 707528248,
+      "step": 4093
+    },
+    {
+      "epoch": 1.567598621217924,
+      "loss": 1.55316162109375,
+      "loss_ce": 0.10258376598358154,
+      "loss_iou": 0.6512075662612915,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 1.450577974319458,
+      "num_input_tokens_seen": 707528248,
+      "step": 4093
+    },
+    {
+      "epoch": 1.5679816162389888,
+      "grad_norm": 225.26098894949774,
+      "learning_rate": 5e-06,
+      "loss": 1.4629,
+      "num_input_tokens_seen": 707700984,
+      "step": 4094
+    },
+    {
+      "epoch": 1.5679816162389888,
+      "loss": 1.34732985496521,
+      "loss_ce": 0.1244833841919899,
+      "loss_iou": 0.5621830821037292,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 1.2030709981918335,
+      "num_input_tokens_seen": 707700984,
+      "step": 4094
+    },
+    {
+      "epoch": 1.5683646112600536,
+      "grad_norm": 209.7959169229572,
+      "learning_rate": 5e-06,
+      "loss": 1.8999,
+      "num_input_tokens_seen": 707873912,
+      "step": 4095
+    },
+    {
+      "epoch": 1.5683646112600536,
+      "loss": 1.8382642269134521,
+      "loss_ce": 0.16478237509727478,
+      "loss_iou": 0.7574868202209473,
+      "loss_num": 0.03173828125,
+      "loss_xval": 1.6734819412231445,
+      "num_input_tokens_seen": 707873912,
+      "step": 4095
+    },
+    {
+      "epoch": 1.5687476062811183,
+      "grad_norm": 70.59792028901389,
+      "learning_rate": 5e-06,
+      "loss": 1.3797,
+      "num_input_tokens_seen": 708046896,
+      "step": 4096
+    },
+    {
+      "epoch": 1.5687476062811183,
+      "loss": 1.4648287296295166,
+      "loss_ce": 0.19003352522850037,
+      "loss_iou": 0.5990675091743469,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 1.2747951745986938,
+      "num_input_tokens_seen": 708046896,
+      "step": 4096
+    },
+    {
+      "epoch": 1.569130601302183,
+      "grad_norm": 189.0194476326106,
+      "learning_rate": 5e-06,
+      "loss": 1.3509,
+      "num_input_tokens_seen": 708219976,
+      "step": 4097
+    },
+    {
+      "epoch": 1.569130601302183,
+      "loss": 1.4974284172058105,
+      "loss_ce": 0.16130970418453217,
+      "loss_iou": 0.6132650375366211,
+      "loss_num": 0.02197265625,
+      "loss_xval": 1.3361186981201172,
+      "num_input_tokens_seen": 708219976,
+      "step": 4097
+    },
+    {
+      "epoch": 1.5695135963232478,
+      "grad_norm": 178.25234115410692,
+      "learning_rate": 5e-06,
+      "loss": 1.6199,
+      "num_input_tokens_seen": 708393144,
+      "step": 4098
+    },
+    {
+      "epoch": 1.5695135963232478,
+      "loss": 1.5991699695587158,
+      "loss_ce": 0.23091337084770203,
+      "loss_iou": 0.6257786750793457,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 1.3682565689086914,
+      "num_input_tokens_seen": 708393144,
+      "step": 4098
+    },
+    {
+      "epoch": 1.5698965913443126,
+      "grad_norm": 203.68978487678905,
+      "learning_rate": 5e-06,
+      "loss": 1.4014,
+      "num_input_tokens_seen": 708565960,
+      "step": 4099
+    },
+    {
+      "epoch": 1.5698965913443126,
+      "loss": 1.286609172821045,
+      "loss_ce": 0.15518397092819214,
+      "loss_iou": 0.5227895975112915,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 1.131425142288208,
+      "num_input_tokens_seen": 708565960,
+      "step": 4099
+    },
+    {
+      "epoch": 1.5702795863653773,
+      "grad_norm": 119.15708207594575,
+      "learning_rate": 5e-06,
+      "loss": 1.3562,
+      "num_input_tokens_seen": 708738696,
+      "step": 4100
+    },
+    {
+      "epoch": 1.5702795863653773,
+      "loss": 1.2358953952789307,
+      "loss_ce": 0.15120267868041992,
+      "loss_iou": 0.5067323446273804,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 1.0795657634735107,
+      "num_input_tokens_seen": 708738696,
+      "step": 4100
+    },
+    {
+      "epoch": 1.570662581386442,
+      "grad_norm": 127.40129626622009,
+      "learning_rate": 5e-06,
+      "loss": 1.3663,
+      "num_input_tokens_seen": 708911680,
+      "step": 4101
+    },
+    {
+      "epoch": 1.570662581386442,
+      "loss": 1.2586824893951416,
+      "loss_ce": 0.14269477128982544,
+      "loss_iou": 0.5022382140159607,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 1.1159876585006714,
+      "num_input_tokens_seen": 708911680,
+      "step": 4101
+    },
+    {
+      "epoch": 1.5710455764075069,
+      "grad_norm": 160.54159834166944,
+      "learning_rate": 5e-06,
+      "loss": 1.4902,
+      "num_input_tokens_seen": 709084704,
+      "step": 4102
+    },
+    {
+      "epoch": 1.5710455764075069,
+      "loss": 1.2457003593444824,
+      "loss_ce": 0.15577131509780884,
+      "loss_iou": 0.5157363414764404,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 1.0899291038513184,
+      "num_input_tokens_seen": 709084704,
+      "step": 4102
+    },
+    {
+      "epoch": 1.5714285714285714,
+      "grad_norm": 155.78345187403025,
+      "learning_rate": 5e-06,
+      "loss": 1.7929,
+      "num_input_tokens_seen": 709257568,
+      "step": 4103
+    },
+    {
+      "epoch": 1.5714285714285714,
+      "loss": 1.8913211822509766,
+      "loss_ce": 0.1973917931318283,
+      "loss_iou": 0.7813519239425659,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 1.6939294338226318,
+      "num_input_tokens_seen": 709257568,
+      "step": 4103
+    },
+    {
+      "epoch": 1.5718115664496362,
+      "grad_norm": 81.54697218385583,
+      "learning_rate": 5e-06,
+      "loss": 1.667,
+      "num_input_tokens_seen": 709430528,
+      "step": 4104
+    },
+    {
+      "epoch": 1.5718115664496362,
+      "loss": 1.6795519590377808,
+      "loss_ce": 0.178236186504364,
+      "loss_iou": 0.7043932676315308,
+      "loss_num": 0.0185546875,
+      "loss_xval": 1.501315712928772,
+      "num_input_tokens_seen": 709430528,
+      "step": 4104
+    },
+    {
+      "epoch": 1.572194561470701,
+      "grad_norm": 97.0537341349617,
+      "learning_rate": 5e-06,
+      "loss": 1.2385,
+      "num_input_tokens_seen": 709603584,
+      "step": 4105
+    },
+    {
+      "epoch": 1.572194561470701,
+      "loss": 1.094093680381775,
+      "loss_ce": 0.1425839066505432,
+      "loss_iou": 0.43873709440231323,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.9408286809921265,
+      "num_input_tokens_seen": 709603584,
+      "step": 4105
+    },
+    {
+      "epoch": 1.5725775564917654,
+      "grad_norm": 171.79064175114,
+      "learning_rate": 5e-06,
+      "loss": 1.3332,
+      "num_input_tokens_seen": 709776600,
+      "step": 4106
+    },
+    {
+      "epoch": 1.5725775564917654,
+      "loss": 1.4016001224517822,
+      "loss_ce": 0.14572454988956451,
+      "loss_iou": 0.5824055075645447,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 1.2558754682540894,
+      "num_input_tokens_seen": 709776600,
+      "step": 4106
+    },
+    {
+      "epoch": 1.5729605515128302,
+      "grad_norm": 104.86836487044407,
+      "learning_rate": 5e-06,
+      "loss": 1.5366,
+      "num_input_tokens_seen": 709949312,
+      "step": 4107
+    },
+    {
+      "epoch": 1.5729605515128302,
+      "loss": 1.4513180255889893,
+      "loss_ce": 0.16940246522426605,
+      "loss_iou": 0.6044130325317383,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 1.2819156646728516,
+      "num_input_tokens_seen": 709949312,
+      "step": 4107
+    },
+    {
+      "epoch": 1.573343546533895,
+      "grad_norm": 99.64302601390047,
+      "learning_rate": 5e-06,
+      "loss": 1.176,
+      "num_input_tokens_seen": 710122168,
+      "step": 4108
+    },
+    {
+      "epoch": 1.573343546533895,
+      "loss": 1.3021981716156006,
+      "loss_ce": 0.17345380783081055,
+      "loss_iou": 0.5259047746658325,
+      "loss_num": 0.015380859375,
+      "loss_xval": 1.12874436378479,
+      "num_input_tokens_seen": 710122168,
+      "step": 4108
+    },
+    {
+      "epoch": 1.5737265415549597,
+      "grad_norm": 165.9908599513972,
+      "learning_rate": 5e-06,
+      "loss": 1.6324,
+      "num_input_tokens_seen": 710294968,
+      "step": 4109
+    },
+    {
+      "epoch": 1.5737265415549597,
+      "loss": 1.7453044652938843,
+      "loss_ce": 0.12673524022102356,
+      "loss_iou": 0.7542919516563416,
+      "loss_num": 0.02197265625,
+      "loss_xval": 1.618569254875183,
+      "num_input_tokens_seen": 710294968,
+      "step": 4109
+    },
+    {
+      "epoch": 1.5741095365760245,
+      "grad_norm": 131.39107236290215,
+      "learning_rate": 5e-06,
+      "loss": 1.5451,
+      "num_input_tokens_seen": 710467552,
+      "step": 4110
+    },
+    {
+      "epoch": 1.5741095365760245,
+      "loss": 1.5330334901809692,
+      "loss_ce": 0.17671683430671692,
+      "loss_iou": 0.6343045830726624,
+      "loss_num": 0.017578125,
+      "loss_xval": 1.3563166856765747,
+      "num_input_tokens_seen": 710467552,
+      "step": 4110
+    },
+    {
+      "epoch": 1.5744925315970892,
+      "grad_norm": 95.10638950465642,
+      "learning_rate": 5e-06,
+      "loss": 1.2058,
+      "num_input_tokens_seen": 710640328,
+      "step": 4111
+    },
+    {
+      "epoch": 1.5744925315970892,
+      "loss": 1.2235019207000732,
+      "loss_ce": 0.13141286373138428,
+      "loss_iou": 0.5170376300811768,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 1.0920891761779785,
+      "num_input_tokens_seen": 710640328,
+      "step": 4111
+    },
+    {
+      "epoch": 1.574875526618154,
+      "grad_norm": 112.59656671003388,
+      "learning_rate": 5e-06,
+      "loss": 1.2657,
+      "num_input_tokens_seen": 710813304,
+      "step": 4112
+    },
+    {
+      "epoch": 1.574875526618154,
+      "loss": 1.2622308731079102,
+      "loss_ce": 0.14622503519058228,
+      "loss_iou": 0.5209240913391113,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 1.1160058975219727,
+      "num_input_tokens_seen": 710813304,
+      "step": 4112
+    },
+    {
+      "epoch": 1.5752585216392188,
+      "grad_norm": 121.46169215186669,
+      "learning_rate": 5e-06,
+      "loss": 1.563,
+      "num_input_tokens_seen": 710986528,
+      "step": 4113
+    },
+    {
+      "epoch": 1.5752585216392188,
+      "loss": 1.690927267074585,
+      "loss_ce": 0.1960771679878235,
+      "loss_iou": 0.7036018371582031,
+      "loss_num": 0.017578125,
+      "loss_xval": 1.4948501586914062,
+      "num_input_tokens_seen": 710986528,
+      "step": 4113
+    },
+    {
+      "epoch": 1.5756415166602835,
+      "grad_norm": 96.98294240494276,
+      "learning_rate": 5e-06,
+      "loss": 1.6662,
+      "num_input_tokens_seen": 711159728,
+      "step": 4114
+    },
+    {
+      "epoch": 1.5756415166602835,
+      "loss": 1.6186466217041016,
+      "loss_ce": 0.19410072267055511,
+      "loss_iou": 0.6570056080818176,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 1.4245460033416748,
+      "num_input_tokens_seen": 711159728,
+      "step": 4114
+    },
+    {
+      "epoch": 1.5760245116813483,
+      "grad_norm": 95.67673641953289,
+      "learning_rate": 5e-06,
+      "loss": 1.4595,
+      "num_input_tokens_seen": 711332608,
+      "step": 4115
+    },
+    {
+      "epoch": 1.5760245116813483,
+      "loss": 1.4235999584197998,
+      "loss_ce": 0.16213397681713104,
+      "loss_iou": 0.5831255316734314,
+      "loss_num": 0.01904296875,
+      "loss_xval": 1.2614659070968628,
+      "num_input_tokens_seen": 711332608,
+      "step": 4115
+    },
+    {
+      "epoch": 1.576407506702413,
+      "grad_norm": 92.77410602492115,
+      "learning_rate": 5e-06,
+      "loss": 1.417,
+      "num_input_tokens_seen": 711505976,
+      "step": 4116
+    },
+    {
+      "epoch": 1.576407506702413,
+      "loss": 1.5653266906738281,
+      "loss_ce": 0.2093820571899414,
+      "loss_iou": 0.6309142112731934,
+      "loss_num": 0.018798828125,
+      "loss_xval": 1.3559446334838867,
+      "num_input_tokens_seen": 711505976,
+      "step": 4116
+    },
+    {
+      "epoch": 1.5767905017234776,
+      "grad_norm": 112.31256374166821,
+      "learning_rate": 5e-06,
+      "loss": 1.4461,
+      "num_input_tokens_seen": 711678688,
+      "step": 4117
+    },
+    {
+      "epoch": 1.5767905017234776,
+      "loss": 1.3936405181884766,
+      "loss_ce": 0.10780879855155945,
+      "loss_iou": 0.5804463624954224,
+      "loss_num": 0.02490234375,
+      "loss_xval": 1.2477457523345947,
+      "num_input_tokens_seen": 711678688,
+      "step": 4117
+    },
+    {
+      "epoch": 1.5771734967445423,
+      "grad_norm": 119.24830662489816,
+      "learning_rate": 5e-06,
+      "loss": 1.6079,
+      "num_input_tokens_seen": 711851200,
+      "step": 4118
+    },
+    {
+      "epoch": 1.5771734967445423,
+      "loss": 1.512297511100769,
+      "loss_ce": 0.15411420166492462,
+      "loss_iou": 0.6256248950958252,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 1.3581833839416504,
+      "num_input_tokens_seen": 711851200,
+      "step": 4118
+    },
+    {
+      "epoch": 1.577556491765607,
+      "grad_norm": 136.61281502636845,
+      "learning_rate": 5e-06,
+      "loss": 1.2418,
+      "num_input_tokens_seen": 712024120,
+      "step": 4119
+    },
+    {
+      "epoch": 1.577556491765607,
+      "loss": 1.2185297012329102,
+      "loss_ce": 0.1887328326702118,
+      "loss_iou": 0.4820462167263031,
+      "loss_num": 0.01318359375,
+      "loss_xval": 1.029796838760376,
+      "num_input_tokens_seen": 712024120,
+      "step": 4119
+    },
+    {
+      "epoch": 1.5779394867866716,
+      "grad_norm": 161.70568656741355,
+      "learning_rate": 5e-06,
+      "loss": 1.4603,
+      "num_input_tokens_seen": 712197288,
+      "step": 4120
+    },
+    {
+      "epoch": 1.5779394867866716,
+      "loss": 1.4775609970092773,
+      "loss_ce": 0.1906827837228775,
+      "loss_iou": 0.6030033230781555,
+      "loss_num": 0.01611328125,
+      "loss_xval": 1.286878228187561,
+      "num_input_tokens_seen": 712197288,
+      "step": 4120
+    },
+    {
+      "epoch": 1.5783224818077364,
+      "grad_norm": 125.21855298836583,
+      "learning_rate": 5e-06,
+      "loss": 1.4454,
+      "num_input_tokens_seen": 712370424,
+      "step": 4121
+    },
+    {
+      "epoch": 1.5783224818077364,
+      "loss": 1.4849789142608643,
+      "loss_ce": 0.1686321496963501,
+      "loss_iou": 0.6066292524337769,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 1.3163468837738037,
+      "num_input_tokens_seen": 712370424,
+      "step": 4121
+    },
+    {
+      "epoch": 1.5787054768288011,
+      "grad_norm": 202.20400727751363,
+      "learning_rate": 5e-06,
+      "loss": 1.707,
+      "num_input_tokens_seen": 712543424,
+      "step": 4122
+    },
+    {
+      "epoch": 1.5787054768288011,
+      "loss": 1.7018821239471436,
+      "loss_ce": 0.15344122052192688,
+      "loss_iou": 0.7219133377075195,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 1.548440933227539,
+      "num_input_tokens_seen": 712543424,
+      "step": 4122
+    },
+    {
+      "epoch": 1.579088471849866,
+      "grad_norm": 90.25571528996664,
+      "learning_rate": 5e-06,
+      "loss": 1.5607,
+      "num_input_tokens_seen": 712716384,
+      "step": 4123
+    },
+    {
+      "epoch": 1.579088471849866,
+      "loss": 1.656998872756958,
+      "loss_ce": 0.16306447982788086,
+      "loss_iou": 0.7058295011520386,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 1.4939343929290771,
+      "num_input_tokens_seen": 712716384,
+      "step": 4123
+    },
+    {
+      "epoch": 1.5794714668709307,
+      "grad_norm": 146.265974471142,
+      "learning_rate": 5e-06,
+      "loss": 1.2993,
+      "num_input_tokens_seen": 712889216,
+      "step": 4124
+    },
+    {
+      "epoch": 1.5794714668709307,
+      "loss": 1.3837032318115234,
+      "loss_ce": 0.13485397398471832,
+      "loss_iou": 0.5820052623748779,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 1.2488493919372559,
+      "num_input_tokens_seen": 712889216,
+      "step": 4124
+    },
+    {
+      "epoch": 1.5798544618919954,
+      "grad_norm": 180.46149214467812,
+      "learning_rate": 5e-06,
+      "loss": 1.5767,
+      "num_input_tokens_seen": 713061896,
+      "step": 4125
+    },
+    {
+      "epoch": 1.5798544618919954,
+      "loss": 1.7555794715881348,
+      "loss_ce": 0.1521853655576706,
+      "loss_iou": 0.7463535070419312,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 1.6033942699432373,
+      "num_input_tokens_seen": 713061896,
+      "step": 4125
+    },
+    {
+      "epoch": 1.5802374569130602,
+      "grad_norm": 144.27489392024512,
+      "learning_rate": 5e-06,
+      "loss": 1.7659,
+      "num_input_tokens_seen": 713234720,
+      "step": 4126
+    },
+    {
+      "epoch": 1.5802374569130602,
+      "loss": 1.6287106275558472,
+      "loss_ce": 0.16540609300136566,
+      "loss_iou": 0.6711969375610352,
+      "loss_num": 0.024169921875,
+      "loss_xval": 1.4633045196533203,
+      "num_input_tokens_seen": 713234720,
+      "step": 4126
+    },
+    {
+      "epoch": 1.580620451934125,
+      "grad_norm": 142.99195314920425,
+      "learning_rate": 5e-06,
+      "loss": 1.2814,
+      "num_input_tokens_seen": 713407728,
+      "step": 4127
+    },
+    {
+      "epoch": 1.580620451934125,
+      "loss": 1.34083092212677,
+      "loss_ce": 0.19812652468681335,
+      "loss_iou": 0.5264760851860046,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 1.1427043676376343,
+      "num_input_tokens_seen": 713407728,
+      "step": 4127
+    },
+    {
+      "epoch": 1.5810034469551897,
+      "grad_norm": 165.02152487307984,
+      "learning_rate": 5e-06,
+      "loss": 1.5148,
+      "num_input_tokens_seen": 713580632,
+      "step": 4128
+    },
+    {
+      "epoch": 1.5810034469551897,
+      "loss": 1.6564905643463135,
+      "loss_ce": 0.21012263000011444,
+      "loss_iou": 0.6827177405357361,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 1.4463680982589722,
+      "num_input_tokens_seen": 713580632,
+      "step": 4128
+    },
+    {
+      "epoch": 1.5813864419762544,
+      "grad_norm": 215.9535446466486,
+      "learning_rate": 5e-06,
+      "loss": 1.6682,
+      "num_input_tokens_seen": 713753704,
+      "step": 4129
+    },
+    {
+      "epoch": 1.5813864419762544,
+      "loss": 1.7163509130477905,
+      "loss_ce": 0.15969273447990417,
+      "loss_iou": 0.7143948078155518,
+      "loss_num": 0.025634765625,
+      "loss_xval": 1.5566582679748535,
+      "num_input_tokens_seen": 713753704,
+      "step": 4129
+    },
+    {
+      "epoch": 1.5817694369973192,
+      "grad_norm": 111.27153619210029,
+      "learning_rate": 5e-06,
+      "loss": 1.3561,
+      "num_input_tokens_seen": 713926688,
+      "step": 4130
+    },
+    {
+      "epoch": 1.5817694369973192,
+      "loss": 1.3754353523254395,
+      "loss_ce": 0.13145066797733307,
+      "loss_iou": 0.5646650195121765,
+      "loss_num": 0.02294921875,
+      "loss_xval": 1.243984580039978,
+      "num_input_tokens_seen": 713926688,
+      "step": 4130
+    },
+    {
+      "epoch": 1.5821524320183837,
+      "grad_norm": 136.51067513872,
+      "learning_rate": 5e-06,
+      "loss": 1.4769,
+      "num_input_tokens_seen": 714099512,
+      "step": 4131
+    },
+    {
+      "epoch": 1.5821524320183837,
+      "loss": 1.2997477054595947,
+      "loss_ce": 0.15273964405059814,
+      "loss_iou": 0.5227227210998535,
+      "loss_num": 0.020263671875,
+      "loss_xval": 1.141392707824707,
+      "num_input_tokens_seen": 714099512,
+      "step": 4131
+    },
+    {
+      "epoch": 1.5825354270394485,
+      "grad_norm": 155.1805329899844,
+      "learning_rate": 5e-06,
+      "loss": 1.4896,
+      "num_input_tokens_seen": 714272424,
+      "step": 4132
+    },
+    {
+      "epoch": 1.5825354270394485,
+      "loss": 1.3481863737106323,
+      "loss_ce": 0.14173448085784912,
+      "loss_iou": 0.5591127872467041,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 1.2064518928527832,
+      "num_input_tokens_seen": 714272424,
+      "step": 4132
+    },
+    {
+      "epoch": 1.5829184220605133,
+      "grad_norm": 93.67945409196703,
+      "learning_rate": 5e-06,
+      "loss": 1.4453,
+      "num_input_tokens_seen": 714445512,
+      "step": 4133
+    },
+    {
+      "epoch": 1.5829184220605133,
+      "loss": 1.3629486560821533,
+      "loss_ce": 0.1463022530078888,
+      "loss_iou": 0.567628026008606,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 1.216646432876587,
+      "num_input_tokens_seen": 714445512,
+      "step": 4133
+    },
+    {
+      "epoch": 1.5833014170815778,
+      "grad_norm": 176.43149146789915,
+      "learning_rate": 5e-06,
+      "loss": 1.3805,
+      "num_input_tokens_seen": 714618000,
+      "step": 4134
+    },
+    {
+      "epoch": 1.5833014170815778,
+      "loss": 1.3133476972579956,
+      "loss_ce": 0.1425570845603943,
+      "loss_iou": 0.539893627166748,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 1.170790672302246,
+      "num_input_tokens_seen": 714618000,
+      "step": 4134
+    },
+    {
+      "epoch": 1.5836844121026425,
+      "grad_norm": 100.57607901972257,
+      "learning_rate": 5e-06,
+      "loss": 1.361,
+      "num_input_tokens_seen": 714790840,
+      "step": 4135
+    },
+    {
+      "epoch": 1.5836844121026425,
+      "loss": 1.1960337162017822,
+      "loss_ce": 0.20448628067970276,
+      "loss_iou": 0.46229591965675354,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.9915474057197571,
+      "num_input_tokens_seen": 714790840,
+      "step": 4135
+    },
+    {
+      "epoch": 1.5840674071237073,
+      "grad_norm": 198.05811479622452,
+      "learning_rate": 5e-06,
+      "loss": 1.3672,
+      "num_input_tokens_seen": 714963856,
+      "step": 4136
+    },
+    {
+      "epoch": 1.5840674071237073,
+      "loss": 1.2496293783187866,
+      "loss_ce": 0.1796765923500061,
+      "loss_iou": 0.4971880316734314,
+      "loss_num": 0.01513671875,
+      "loss_xval": 1.0699528455734253,
+      "num_input_tokens_seen": 714963856,
+      "step": 4136
+    },
+    {
+      "epoch": 1.584450402144772,
+      "grad_norm": 218.06086572369287,
+      "learning_rate": 5e-06,
+      "loss": 1.684,
+      "num_input_tokens_seen": 715136784,
+      "step": 4137
+    },
+    {
+      "epoch": 1.584450402144772,
+      "loss": 1.6358394622802734,
+      "loss_ce": 0.18271474540233612,
+      "loss_iou": 0.6757200956344604,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 1.453124761581421,
+      "num_input_tokens_seen": 715136784,
+      "step": 4137
+    },
+    {
+      "epoch": 1.5848333971658368,
+      "grad_norm": 139.27848894680739,
+      "learning_rate": 5e-06,
+      "loss": 1.4464,
+      "num_input_tokens_seen": 715310048,
+      "step": 4138
+    },
+    {
+      "epoch": 1.5848333971658368,
+      "loss": 1.282130241394043,
+      "loss_ce": 0.17187875509262085,
+      "loss_iou": 0.5068163871765137,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.1102514266967773,
+      "num_input_tokens_seen": 715310048,
+      "step": 4138
+    },
+    {
+      "epoch": 1.5852163921869016,
+      "grad_norm": 117.16573760868806,
+      "learning_rate": 5e-06,
+      "loss": 1.4076,
+      "num_input_tokens_seen": 715482880,
+      "step": 4139
+    },
+    {
+      "epoch": 1.5852163921869016,
+      "loss": 1.5313211679458618,
+      "loss_ce": 0.163076251745224,
+      "loss_iou": 0.6271766424179077,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 1.3682448863983154,
+      "num_input_tokens_seen": 715482880,
+      "step": 4139
+    },
+    {
+      "epoch": 1.5855993872079663,
+      "grad_norm": 167.40106243290677,
+      "learning_rate": 5e-06,
+      "loss": 1.582,
+      "num_input_tokens_seen": 715655872,
+      "step": 4140
+    },
+    {
+      "epoch": 1.5855993872079663,
+      "loss": 1.6569092273712158,
+      "loss_ce": 0.15909257531166077,
+      "loss_iou": 0.7087472677230835,
+      "loss_num": 0.01611328125,
+      "loss_xval": 1.497816801071167,
+      "num_input_tokens_seen": 715655872,
+      "step": 4140
+    },
+    {
+      "epoch": 1.585982382229031,
+      "grad_norm": 242.45741736718762,
+      "learning_rate": 5e-06,
+      "loss": 1.5618,
+      "num_input_tokens_seen": 715825728,
+      "step": 4141
+    },
+    {
+      "epoch": 1.585982382229031,
+      "loss": 1.4228453636169434,
+      "loss_ce": 0.17762215435504913,
+      "loss_iou": 0.5792308449745178,
+      "loss_num": 0.017333984375,
+      "loss_xval": 1.2452231645584106,
+      "num_input_tokens_seen": 715825728,
+      "step": 4141
+    },
+    {
+      "epoch": 1.5863653772500959,
+      "grad_norm": 215.18232499518476,
+      "learning_rate": 5e-06,
+      "loss": 2.1397,
+      "num_input_tokens_seen": 715998824,
+      "step": 4142
+    },
+    {
+      "epoch": 1.5863653772500959,
+      "loss": 2.1633801460266113,
+      "loss_ce": 0.15684956312179565,
+      "loss_iou": 0.9566954374313354,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 2.006530523300171,
+      "num_input_tokens_seen": 715998824,
+      "step": 4142
+    },
+    {
+      "epoch": 1.5867483722711606,
+      "grad_norm": 196.24639417525137,
+      "learning_rate": 5e-06,
+      "loss": 2.2569,
+      "num_input_tokens_seen": 716171632,
+      "step": 4143
+    },
+    {
+      "epoch": 1.5867483722711606,
+      "loss": 2.2805094718933105,
+      "loss_ce": 0.17804448306560516,
+      "loss_iou": 0.9064571261405945,
+      "loss_num": 0.057861328125,
+      "loss_xval": 2.1024651527404785,
+      "num_input_tokens_seen": 716171632,
+      "step": 4143
+    },
+    {
+      "epoch": 1.5871313672922251,
+      "grad_norm": 212.68308123228155,
+      "learning_rate": 5e-06,
+      "loss": 2.7161,
+      "num_input_tokens_seen": 716344312,
+      "step": 4144
+    },
+    {
+      "epoch": 1.5871313672922251,
+      "loss": 2.824314594268799,
+      "loss_ce": 0.2492833435535431,
+      "loss_iou": 1.209207534790039,
+      "loss_num": 0.03125,
+      "loss_xval": 2.575031280517578,
+      "num_input_tokens_seen": 716344312,
+      "step": 4144
+    },
+    {
+      "epoch": 1.58751436231329,
+      "grad_norm": 97.51620370993898,
+      "learning_rate": 5e-06,
+      "loss": 1.5787,
+      "num_input_tokens_seen": 716517296,
+      "step": 4145
+    },
+    {
+      "epoch": 1.58751436231329,
+      "loss": 1.413367509841919,
+      "loss_ce": 0.13391654193401337,
+      "loss_iou": 0.5669410228729248,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 1.2604079246520996,
+      "num_input_tokens_seen": 716517296,
+      "step": 4145
+    },
+    {
+      "epoch": 1.5878973573343547,
+      "grad_norm": 84.09415924064281,
+      "learning_rate": 5e-06,
+      "loss": 1.6434,
+      "num_input_tokens_seen": 716690456,
+      "step": 4146
+    },
+    {
+      "epoch": 1.5878973573343547,
+      "loss": 1.9350314140319824,
+      "loss_ce": 0.12022538483142853,
+      "loss_iou": 0.8216485977172852,
+      "loss_num": 0.034423828125,
+      "loss_xval": 1.8148059844970703,
+      "num_input_tokens_seen": 716690456,
+      "step": 4146
+    },
+    {
+      "epoch": 1.5882803523554194,
+      "grad_norm": 131.85175873613758,
+      "learning_rate": 5e-06,
+      "loss": 2.0919,
+      "num_input_tokens_seen": 716863408,
+      "step": 4147
+    },
+    {
+      "epoch": 1.5882803523554194,
+      "loss": 2.241295576095581,
+      "loss_ce": 0.14061076939105988,
+      "loss_iou": 0.9912604093551636,
+      "loss_num": 0.023681640625,
+      "loss_xval": 2.100684881210327,
+      "num_input_tokens_seen": 716863408,
+      "step": 4147
+    },
+    {
+      "epoch": 1.588663347376484,
+      "grad_norm": 78.5322331711656,
+      "learning_rate": 5e-06,
+      "loss": 1.7065,
+      "num_input_tokens_seen": 717036496,
+      "step": 4148
+    },
+    {
+      "epoch": 1.588663347376484,
+      "loss": 1.801315188407898,
+      "loss_ce": 0.11861561238765717,
+      "loss_iou": 0.7705795764923096,
+      "loss_num": 0.0283203125,
+      "loss_xval": 1.6826996803283691,
+      "num_input_tokens_seen": 717036496,
+      "step": 4148
+    },
+    {
+      "epoch": 1.5890463423975487,
+      "grad_norm": 108.38839175352602,
+      "learning_rate": 5e-06,
+      "loss": 1.4917,
+      "num_input_tokens_seen": 717209536,
+      "step": 4149
+    },
+    {
+      "epoch": 1.5890463423975487,
+      "loss": 1.5406887531280518,
+      "loss_ce": 0.19515490531921387,
+      "loss_iou": 0.625128984451294,
+      "loss_num": 0.01904296875,
+      "loss_xval": 1.337751865386963,
+      "num_input_tokens_seen": 717209536,
+      "step": 4149
+    },
+    {
+      "epoch": 1.5894293374186135,
+      "grad_norm": 82.13063350594624,
+      "learning_rate": 5e-06,
+      "loss": 1.7176,
+      "num_input_tokens_seen": 717382800,
+      "step": 4150
+    },
+    {
+      "epoch": 1.5894293374186135,
+      "loss": 1.8092962503433228,
+      "loss_ce": 0.1871616542339325,
+      "loss_iou": 0.7534501552581787,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 1.6221346855163574,
+      "num_input_tokens_seen": 717382800,
+      "step": 4150
+    },
+    {
+      "epoch": 1.5898123324396782,
+      "grad_norm": 120.62412092658283,
+      "learning_rate": 5e-06,
+      "loss": 1.5449,
+      "num_input_tokens_seen": 717555752,
+      "step": 4151
+    },
+    {
+      "epoch": 1.5898123324396782,
+      "loss": 1.7137246131896973,
+      "loss_ce": 0.21029409766197205,
+      "loss_iou": 0.6971803903579712,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 1.5034306049346924,
+      "num_input_tokens_seen": 717555752,
+      "step": 4151
+    },
+    {
+      "epoch": 1.590195327460743,
+      "grad_norm": 231.08103008179881,
+      "learning_rate": 5e-06,
+      "loss": 1.8221,
+      "num_input_tokens_seen": 717728968,
+      "step": 4152
+    },
+    {
+      "epoch": 1.590195327460743,
+      "loss": 1.7823009490966797,
+      "loss_ce": 0.22162017226219177,
+      "loss_iou": 0.7400267124176025,
+      "loss_num": 0.01611328125,
+      "loss_xval": 1.560680866241455,
+      "num_input_tokens_seen": 717728968,
+      "step": 4152
+    },
+    {
+      "epoch": 1.5905783224818077,
+      "grad_norm": 148.41194264439528,
+      "learning_rate": 5e-06,
+      "loss": 1.4061,
+      "num_input_tokens_seen": 717901888,
+      "step": 4153
+    },
+    {
+      "epoch": 1.5905783224818077,
+      "loss": 1.4149425029754639,
+      "loss_ce": 0.15381485223770142,
+      "loss_iou": 0.5752658843994141,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 1.2611274719238281,
+      "num_input_tokens_seen": 717901888,
+      "step": 4153
+    },
+    {
+      "epoch": 1.5909613175028725,
+      "grad_norm": 172.7051509765074,
+      "learning_rate": 5e-06,
+      "loss": 1.4777,
+      "num_input_tokens_seen": 718074976,
+      "step": 4154
+    },
+    {
+      "epoch": 1.5909613175028725,
+      "loss": 1.486426830291748,
+      "loss_ce": 0.1446625143289566,
+      "loss_iou": 0.6393422484397888,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 1.3417643308639526,
+      "num_input_tokens_seen": 718074976,
+      "step": 4154
+    },
+    {
+      "epoch": 1.5913443125239373,
+      "grad_norm": 107.68109281663867,
+      "learning_rate": 5e-06,
+      "loss": 1.7517,
+      "num_input_tokens_seen": 718247808,
+      "step": 4155
+    },
+    {
+      "epoch": 1.5913443125239373,
+      "loss": 1.8165727853775024,
+      "loss_ce": 0.16311918199062347,
+      "loss_iou": 0.7450922727584839,
+      "loss_num": 0.03271484375,
+      "loss_xval": 1.6024281978607178,
+      "num_input_tokens_seen": 718247808,
+      "step": 4155
+    },
+    {
+      "epoch": 1.591727307545002,
+      "grad_norm": 109.20059029580428,
+      "learning_rate": 5e-06,
+      "loss": 1.4292,
+      "num_input_tokens_seen": 718420800,
+      "step": 4156
+    },
+    {
+      "epoch": 1.591727307545002,
+      "loss": 1.460224986076355,
+      "loss_ce": 0.13261058926582336,
+      "loss_iou": 0.6007121205329895,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 1.327614426612854,
+      "num_input_tokens_seen": 718420800,
+      "step": 4156
+    },
+    {
+      "epoch": 1.5921103025660668,
+      "grad_norm": 96.94773639437696,
+      "learning_rate": 5e-06,
+      "loss": 1.2066,
+      "num_input_tokens_seen": 718594040,
+      "step": 4157
+    },
+    {
+      "epoch": 1.5921103025660668,
+      "loss": 1.2734310626983643,
+      "loss_ce": 0.1546371579170227,
+      "loss_iou": 0.5315343737602234,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 1.1187938451766968,
+      "num_input_tokens_seen": 718594040,
+      "step": 4157
+    },
+    {
+      "epoch": 1.5924932975871313,
+      "grad_norm": 152.5733884973267,
+      "learning_rate": 5e-06,
+      "loss": 1.6566,
+      "num_input_tokens_seen": 718767408,
+      "step": 4158
+    },
+    {
+      "epoch": 1.5924932975871313,
+      "loss": 1.4709134101867676,
+      "loss_ce": 0.16556543111801147,
+      "loss_iou": 0.6193030476570129,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 1.3053480386734009,
+      "num_input_tokens_seen": 718767408,
+      "step": 4158
+    },
+    {
+      "epoch": 1.592876292608196,
+      "grad_norm": 151.32500543632838,
+      "learning_rate": 5e-06,
+      "loss": 1.5158,
+      "num_input_tokens_seen": 718940264,
+      "step": 4159
+    },
+    {
+      "epoch": 1.592876292608196,
+      "loss": 1.4600410461425781,
+      "loss_ce": 0.18419691920280457,
+      "loss_iou": 0.6048715710639954,
+      "loss_num": 0.01318359375,
+      "loss_xval": 1.2758442163467407,
+      "num_input_tokens_seen": 718940264,
+      "step": 4159
+    },
+    {
+      "epoch": 1.5932592876292608,
+      "grad_norm": 155.07072042887276,
+      "learning_rate": 5e-06,
+      "loss": 2.0443,
+      "num_input_tokens_seen": 719113560,
+      "step": 4160
+    },
+    {
+      "epoch": 1.5932592876292608,
+      "loss": 2.1296849250793457,
+      "loss_ce": 0.15815117955207825,
+      "loss_iou": 0.8649172782897949,
+      "loss_num": 0.04833984375,
+      "loss_xval": 1.9715337753295898,
+      "num_input_tokens_seen": 719113560,
+      "step": 4160
+    },
+    {
+      "epoch": 1.5936422826503256,
+      "grad_norm": 140.57978385602607,
+      "learning_rate": 5e-06,
+      "loss": 1.2904,
+      "num_input_tokens_seen": 719286400,
+      "step": 4161
+    },
+    {
+      "epoch": 1.5936422826503256,
+      "loss": 1.3131370544433594,
+      "loss_ce": 0.16869494318962097,
+      "loss_iou": 0.543198823928833,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 1.144442081451416,
+      "num_input_tokens_seen": 719286400,
+      "step": 4161
+    },
+    {
+      "epoch": 1.5940252776713901,
+      "grad_norm": 136.09479291602526,
+      "learning_rate": 5e-06,
+      "loss": 1.2916,
+      "num_input_tokens_seen": 719459200,
+      "step": 4162
+    },
+    {
+      "epoch": 1.5940252776713901,
+      "loss": 1.2238564491271973,
+      "loss_ce": 0.16852031648159027,
+      "loss_iou": 0.4966316819190979,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 1.0553361177444458,
+      "num_input_tokens_seen": 719459200,
+      "step": 4162
+    },
+    {
+      "epoch": 1.5944082726924549,
+      "grad_norm": 157.00291345601957,
+      "learning_rate": 5e-06,
+      "loss": 1.6166,
+      "num_input_tokens_seen": 719632456,
+      "step": 4163
+    },
+    {
+      "epoch": 1.5944082726924549,
+      "loss": 1.6060909032821655,
+      "loss_ce": 0.18971838057041168,
+      "loss_iou": 0.6511489152908325,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 1.416372537612915,
+      "num_input_tokens_seen": 719632456,
+      "step": 4163
+    },
+    {
+      "epoch": 1.5947912677135196,
+      "grad_norm": 145.2367620609835,
+      "learning_rate": 5e-06,
+      "loss": 1.6286,
+      "num_input_tokens_seen": 719805552,
+      "step": 4164
+    },
+    {
+      "epoch": 1.5947912677135196,
+      "loss": 1.5709772109985352,
+      "loss_ce": 0.13955217599868774,
+      "loss_iou": 0.6321249008178711,
+      "loss_num": 0.033447265625,
+      "loss_xval": 1.4314250946044922,
+      "num_input_tokens_seen": 719805552,
+      "step": 4164
+    },
+    {
+      "epoch": 1.5951742627345844,
+      "grad_norm": 108.52186220452043,
+      "learning_rate": 5e-06,
+      "loss": 1.4935,
+      "num_input_tokens_seen": 719978200,
+      "step": 4165
+    },
+    {
+      "epoch": 1.5951742627345844,
+      "loss": 1.5908496379852295,
+      "loss_ce": 0.18345247209072113,
+      "loss_iou": 0.6416716575622559,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 1.4073972702026367,
+      "num_input_tokens_seen": 719978200,
+      "step": 4165
+    },
+    {
+      "epoch": 1.5955572577556492,
+      "grad_norm": 156.18950512363426,
+      "learning_rate": 5e-06,
+      "loss": 1.6529,
+      "num_input_tokens_seen": 720151256,
+      "step": 4166
+    },
+    {
+      "epoch": 1.5955572577556492,
+      "loss": 1.686915397644043,
+      "loss_ce": 0.1795039176940918,
+      "loss_iou": 0.7060678601264954,
+      "loss_num": 0.01904296875,
+      "loss_xval": 1.5074115991592407,
+      "num_input_tokens_seen": 720151256,
+      "step": 4166
+    },
+    {
+      "epoch": 1.595940252776714,
+      "grad_norm": 125.43865878654431,
+      "learning_rate": 5e-06,
+      "loss": 1.5317,
+      "num_input_tokens_seen": 720320352,
+      "step": 4167
+    },
+    {
+      "epoch": 1.595940252776714,
+      "loss": 1.6104614734649658,
+      "loss_ce": 0.13296779990196228,
+      "loss_iou": 0.6841661930084229,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 1.4774937629699707,
+      "num_input_tokens_seen": 720320352,
+      "step": 4167
+    },
+    {
+      "epoch": 1.5963232477977787,
+      "grad_norm": 136.12731473282872,
+      "learning_rate": 5e-06,
+      "loss": 1.3777,
+      "num_input_tokens_seen": 720493408,
+      "step": 4168
+    },
+    {
+      "epoch": 1.5963232477977787,
+      "loss": 1.5521327257156372,
+      "loss_ce": 0.16964644193649292,
+      "loss_iou": 0.6411638259887695,
+      "loss_num": 0.02001953125,
+      "loss_xval": 1.382486343383789,
+      "num_input_tokens_seen": 720493408,
+      "step": 4168
+    },
+    {
+      "epoch": 1.5967062428188434,
+      "grad_norm": 132.3225138177353,
+      "learning_rate": 5e-06,
+      "loss": 1.4554,
+      "num_input_tokens_seen": 720666600,
+      "step": 4169
+    },
+    {
+      "epoch": 1.5967062428188434,
+      "loss": 1.4904382228851318,
+      "loss_ce": 0.12942801415920258,
+      "loss_iou": 0.6297086477279663,
+      "loss_num": 0.020263671875,
+      "loss_xval": 1.3610103130340576,
+      "num_input_tokens_seen": 720666600,
+      "step": 4169
+    },
+    {
+      "epoch": 1.5970892378399082,
+      "grad_norm": 115.20173630301336,
+      "learning_rate": 5e-06,
+      "loss": 1.4385,
+      "num_input_tokens_seen": 720839496,
+      "step": 4170
+    },
+    {
+      "epoch": 1.5970892378399082,
+      "loss": 1.5010660886764526,
+      "loss_ce": 0.1956491619348526,
+      "loss_iou": 0.5966476798057556,
+      "loss_num": 0.0224609375,
+      "loss_xval": 1.3054169416427612,
+      "num_input_tokens_seen": 720839496,
+      "step": 4170
+    },
+    {
+      "epoch": 1.597472232860973,
+      "grad_norm": 142.6082978665598,
+      "learning_rate": 5e-06,
+      "loss": 1.3468,
+      "num_input_tokens_seen": 721012352,
+      "step": 4171
+    },
+    {
+      "epoch": 1.597472232860973,
+      "loss": 1.3007220029830933,
+      "loss_ce": 0.1823827028274536,
+      "loss_iou": 0.5270956754684448,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 1.1183393001556396,
+      "num_input_tokens_seen": 721012352,
+      "step": 4171
+    },
+    {
+      "epoch": 1.5978552278820375,
+      "grad_norm": 168.93688929215168,
+      "learning_rate": 5e-06,
+      "loss": 1.9489,
+      "num_input_tokens_seen": 721185152,
+      "step": 4172
+    },
+    {
+      "epoch": 1.5978552278820375,
+      "loss": 1.736716389656067,
+      "loss_ce": 0.16597309708595276,
+      "loss_iou": 0.7086504697799683,
+      "loss_num": 0.03076171875,
+      "loss_xval": 1.5707433223724365,
+      "num_input_tokens_seen": 721185152,
+      "step": 4172
+    },
+    {
+      "epoch": 1.5982382229031022,
+      "grad_norm": 115.08275550344064,
+      "learning_rate": 5e-06,
+      "loss": 1.5576,
+      "num_input_tokens_seen": 721358280,
+      "step": 4173
+    },
+    {
+      "epoch": 1.5982382229031022,
+      "loss": 1.4300172328948975,
+      "loss_ce": 0.16085779666900635,
+      "loss_iou": 0.5783511400222778,
+      "loss_num": 0.0224609375,
+      "loss_xval": 1.2691595554351807,
+      "num_input_tokens_seen": 721358280,
+      "step": 4173
+    },
+    {
+      "epoch": 1.598621217924167,
+      "grad_norm": 151.18544005799114,
+      "learning_rate": 5e-06,
+      "loss": 1.5803,
+      "num_input_tokens_seen": 721531232,
+      "step": 4174
+    },
+    {
+      "epoch": 1.598621217924167,
+      "loss": 1.5780599117279053,
+      "loss_ce": 0.18436744809150696,
+      "loss_iou": 0.6431353092193604,
+      "loss_num": 0.021484375,
+      "loss_xval": 1.3936924934387207,
+      "num_input_tokens_seen": 721531232,
+      "step": 4174
+    },
+    {
+      "epoch": 1.5990042129452318,
+      "grad_norm": 95.07224439462104,
+      "learning_rate": 5e-06,
+      "loss": 1.8869,
+      "num_input_tokens_seen": 721704312,
+      "step": 4175
+    },
+    {
+      "epoch": 1.5990042129452318,
+      "loss": 1.758817434310913,
+      "loss_ce": 0.13629360496997833,
+      "loss_iou": 0.7348763942718506,
+      "loss_num": 0.030517578125,
+      "loss_xval": 1.6225237846374512,
+      "num_input_tokens_seen": 721704312,
+      "step": 4175
+    },
+    {
+      "epoch": 1.5993872079662963,
+      "grad_norm": 86.37329986341791,
+      "learning_rate": 5e-06,
+      "loss": 1.2869,
+      "num_input_tokens_seen": 721877168,
+      "step": 4176
+    },
+    {
+      "epoch": 1.5993872079662963,
+      "loss": 1.2956174612045288,
+      "loss_ce": 0.17034663259983063,
+      "loss_iou": 0.5340862274169922,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 1.1252708435058594,
+      "num_input_tokens_seen": 721877168,
+      "step": 4176
+    },
+    {
+      "epoch": 1.599770202987361,
+      "grad_norm": 141.03852514451134,
+      "learning_rate": 5e-06,
+      "loss": 1.2412,
+      "num_input_tokens_seen": 722050448,
+      "step": 4177
+    },
+    {
+      "epoch": 1.599770202987361,
+      "loss": 1.272310733795166,
+      "loss_ce": 0.19302532076835632,
+      "loss_iou": 0.5057986974716187,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 1.0792853832244873,
+      "num_input_tokens_seen": 722050448,
+      "step": 4177
+    },
+    {
+      "epoch": 1.6001531980084258,
+      "grad_norm": 151.4463610938401,
+      "learning_rate": 5e-06,
+      "loss": 1.4196,
+      "num_input_tokens_seen": 722223512,
+      "step": 4178
+    },
+    {
+      "epoch": 1.6001531980084258,
+      "loss": 1.5337145328521729,
+      "loss_ce": 0.1342497169971466,
+      "loss_iou": 0.6537882089614868,
+      "loss_num": 0.018310546875,
+      "loss_xval": 1.3994648456573486,
+      "num_input_tokens_seen": 722223512,
+      "step": 4178
+    },
+    {
+      "epoch": 1.6005361930294906,
+      "grad_norm": 124.17179064680883,
+      "learning_rate": 5e-06,
+      "loss": 1.3181,
+      "num_input_tokens_seen": 722396256,
+      "step": 4179
+    },
+    {
+      "epoch": 1.6005361930294906,
+      "loss": 1.4280534982681274,
+      "loss_ce": 0.16537800431251526,
+      "loss_iou": 0.5884453058242798,
+      "loss_num": 0.01708984375,
+      "loss_xval": 1.2626755237579346,
+      "num_input_tokens_seen": 722396256,
+      "step": 4179
+    },
+    {
+      "epoch": 1.6009191880505553,
+      "grad_norm": 371.16790633061066,
+      "learning_rate": 5e-06,
+      "loss": 1.4541,
+      "num_input_tokens_seen": 722569264,
+      "step": 4180
+    },
+    {
+      "epoch": 1.6009191880505553,
+      "loss": 1.4142394065856934,
+      "loss_ce": 0.1499151587486267,
+      "loss_iou": 0.5764980912208557,
+      "loss_num": 0.022216796875,
+      "loss_xval": 1.2643243074417114,
+      "num_input_tokens_seen": 722569264,
+      "step": 4180
+    },
+    {
+      "epoch": 1.60130218307162,
+      "grad_norm": 158.38630178691437,
+      "learning_rate": 5e-06,
+      "loss": 1.9153,
+      "num_input_tokens_seen": 722742344,
+      "step": 4181
+    },
+    {
+      "epoch": 1.60130218307162,
+      "loss": 1.7647039890289307,
+      "loss_ce": 0.19260618090629578,
+      "loss_iou": 0.728462278842926,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 1.572097897529602,
+      "num_input_tokens_seen": 722742344,
+      "step": 4181
+    },
+    {
+      "epoch": 1.6016851780926848,
+      "grad_norm": 251.06629771242092,
+      "learning_rate": 5e-06,
+      "loss": 1.7332,
+      "num_input_tokens_seen": 722915640,
+      "step": 4182
+    },
+    {
+      "epoch": 1.6016851780926848,
+      "loss": 1.745754599571228,
+      "loss_ce": 0.15105757117271423,
+      "loss_iou": 0.7322850227355957,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 1.5946969985961914,
+      "num_input_tokens_seen": 722915640,
+      "step": 4182
+    },
+    {
+      "epoch": 1.6020681731137496,
+      "grad_norm": 131.27761042079712,
+      "learning_rate": 5e-06,
+      "loss": 1.6333,
+      "num_input_tokens_seen": 723088552,
+      "step": 4183
+    },
+    {
+      "epoch": 1.6020681731137496,
+      "loss": 1.581559658050537,
+      "loss_ce": 0.15056690573692322,
+      "loss_iou": 0.6644862294197083,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 1.4309927225112915,
+      "num_input_tokens_seen": 723088552,
+      "step": 4183
+    },
+    {
+      "epoch": 1.6024511681348144,
+      "grad_norm": 146.93881199696452,
+      "learning_rate": 5e-06,
+      "loss": 1.6093,
+      "num_input_tokens_seen": 723261424,
+      "step": 4184
+    },
+    {
+      "epoch": 1.6024511681348144,
+      "loss": 1.342673897743225,
+      "loss_ce": 0.10359810292720795,
+      "loss_iou": 0.562012255191803,
+      "loss_num": 0.02294921875,
+      "loss_xval": 1.229676365852356,
+      "num_input_tokens_seen": 723261424,
+      "step": 4184
+    },
+    {
+      "epoch": 1.6028341631558791,
+      "grad_norm": 165.6412352146461,
+      "learning_rate": 5e-06,
+      "loss": 1.3122,
+      "num_input_tokens_seen": 723433976,
+      "step": 4185
+    },
+    {
+      "epoch": 1.6028341631558791,
+      "loss": 1.2699909210205078,
+      "loss_ce": 0.1705361157655716,
+      "loss_iou": 0.5034626722335815,
+      "loss_num": 0.0185546875,
+      "loss_xval": 1.099454641342163,
+      "num_input_tokens_seen": 723433976,
+      "step": 4185
+    },
+    {
+      "epoch": 1.6032171581769437,
+      "grad_norm": 124.46698113894391,
+      "learning_rate": 5e-06,
+      "loss": 1.4424,
+      "num_input_tokens_seen": 723606664,
+      "step": 4186
+    },
+    {
+      "epoch": 1.6032171581769437,
+      "loss": 1.5509071350097656,
+      "loss_ce": 0.16748195886611938,
+      "loss_iou": 0.646195650100708,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 1.383425235748291,
+      "num_input_tokens_seen": 723606664,
+      "step": 4186
+    },
+    {
+      "epoch": 1.6036001531980084,
+      "grad_norm": 99.2164667549704,
+      "learning_rate": 5e-06,
+      "loss": 1.342,
+      "num_input_tokens_seen": 723779400,
+      "step": 4187
+    },
+    {
+      "epoch": 1.6036001531980084,
+      "loss": 1.476794719696045,
+      "loss_ce": 0.1615278273820877,
+      "loss_iou": 0.6092020273208618,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 1.3152668476104736,
+      "num_input_tokens_seen": 723779400,
+      "step": 4187
+    },
+    {
+      "epoch": 1.6039831482190732,
+      "grad_norm": 150.00525951319077,
+      "learning_rate": 5e-06,
+      "loss": 1.396,
+      "num_input_tokens_seen": 723952440,
+      "step": 4188
+    },
+    {
+      "epoch": 1.6039831482190732,
+      "loss": 1.4182698726654053,
+      "loss_ce": 0.17111729085445404,
+      "loss_iou": 0.58997642993927,
+      "loss_num": 0.013427734375,
+      "loss_xval": 1.24715256690979,
+      "num_input_tokens_seen": 723952440,
+      "step": 4188
+    },
+    {
+      "epoch": 1.604366143240138,
+      "grad_norm": 193.94537393184032,
+      "learning_rate": 5e-06,
+      "loss": 1.5548,
+      "num_input_tokens_seen": 724125104,
+      "step": 4189
+    },
+    {
+      "epoch": 1.604366143240138,
+      "loss": 1.5976792573928833,
+      "loss_ce": 0.1452297568321228,
+      "loss_iou": 0.6719491481781006,
+      "loss_num": 0.021728515625,
+      "loss_xval": 1.4524493217468262,
+      "num_input_tokens_seen": 724125104,
+      "step": 4189
+    },
+    {
+      "epoch": 1.6047491382612025,
+      "grad_norm": 175.33307876378808,
+      "learning_rate": 5e-06,
+      "loss": 1.5052,
+      "num_input_tokens_seen": 724297976,
+      "step": 4190
+    },
+    {
+      "epoch": 1.6047491382612025,
+      "loss": 1.55488920211792,
+      "loss_ce": 0.18336597084999084,
+      "loss_iou": 0.645859956741333,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 1.371523380279541,
+      "num_input_tokens_seen": 724297976,
+      "step": 4190
+    },
+    {
+      "epoch": 1.6051321332822672,
+      "grad_norm": 145.4086847251544,
+      "learning_rate": 5e-06,
+      "loss": 1.6218,
+      "num_input_tokens_seen": 724470944,
+      "step": 4191
+    },
+    {
+      "epoch": 1.6051321332822672,
+      "loss": 1.651831030845642,
+      "loss_ce": 0.20629402995109558,
+      "loss_iou": 0.6676690578460693,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 1.4455370903015137,
+      "num_input_tokens_seen": 724470944,
+      "step": 4191
+    },
+    {
+      "epoch": 1.605515128303332,
+      "grad_norm": 83.23085303981833,
+      "learning_rate": 5e-06,
+      "loss": 1.4423,
+      "num_input_tokens_seen": 724643840,
+      "step": 4192
+    },
+    {
+      "epoch": 1.605515128303332,
+      "loss": 1.6921238899230957,
+      "loss_ce": 0.18480592966079712,
+      "loss_iou": 0.6849792003631592,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 1.5073180198669434,
+      "num_input_tokens_seen": 724643840,
+      "step": 4192
+    },
+    {
+      "epoch": 1.6058981233243967,
+      "grad_norm": 235.94366271150906,
+      "learning_rate": 5e-06,
+      "loss": 1.3723,
+      "num_input_tokens_seen": 724816920,
+      "step": 4193
+    },
+    {
+      "epoch": 1.6058981233243967,
+      "loss": 1.2548590898513794,
+      "loss_ce": 0.14185810089111328,
+      "loss_iou": 0.5108004212379456,
+      "loss_num": 0.018310546875,
+      "loss_xval": 1.1130009889602661,
+      "num_input_tokens_seen": 724816920,
+      "step": 4193
+    },
+    {
+      "epoch": 1.6062811183454615,
+      "grad_norm": 153.38321902166095,
+      "learning_rate": 5e-06,
+      "loss": 1.6748,
+      "num_input_tokens_seen": 724989960,
+      "step": 4194
+    },
+    {
+      "epoch": 1.6062811183454615,
+      "loss": 1.7610971927642822,
+      "loss_ce": 0.17520743608474731,
+      "loss_iou": 0.7269964218139648,
+      "loss_num": 0.0263671875,
+      "loss_xval": 1.5858898162841797,
+      "num_input_tokens_seen": 724989960,
+      "step": 4194
+    },
+    {
+      "epoch": 1.6066641133665263,
+      "grad_norm": 251.33073815191366,
+      "learning_rate": 5e-06,
+      "loss": 1.998,
+      "num_input_tokens_seen": 725163144,
+      "step": 4195
+    },
+    {
+      "epoch": 1.6066641133665263,
+      "loss": 2.229393720626831,
+      "loss_ce": 0.21598659455776215,
+      "loss_iou": 0.9182636141777039,
+      "loss_num": 0.035400390625,
+      "loss_xval": 2.0134072303771973,
+      "num_input_tokens_seen": 725163144,
+      "step": 4195
+    },
+    {
+      "epoch": 1.607047108387591,
+      "grad_norm": 195.8249379939445,
+      "learning_rate": 5e-06,
+      "loss": 1.4789,
+      "num_input_tokens_seen": 725335928,
+      "step": 4196
+    },
+    {
+      "epoch": 1.607047108387591,
+      "loss": 1.4367557764053345,
+      "loss_ce": 0.15262383222579956,
+      "loss_iou": 0.5933904051780701,
+      "loss_num": 0.01953125,
+      "loss_xval": 1.2841318845748901,
+      "num_input_tokens_seen": 725335928,
+      "step": 4196
+    },
+    {
+      "epoch": 1.6074301034086558,
+      "grad_norm": 124.93389924958187,
+      "learning_rate": 5e-06,
+      "loss": 1.4569,
+      "num_input_tokens_seen": 725508936,
+      "step": 4197
+    },
+    {
+      "epoch": 1.6074301034086558,
+      "loss": 1.4103691577911377,
+      "loss_ce": 0.15445317327976227,
+      "loss_iou": 0.5805336833000183,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 1.2559159994125366,
+      "num_input_tokens_seen": 725508936,
+      "step": 4197
+    },
+    {
+      "epoch": 1.6078130984297205,
+      "grad_norm": 134.4865770057758,
+      "learning_rate": 5e-06,
+      "loss": 1.6511,
+      "num_input_tokens_seen": 725682192,
+      "step": 4198
+    },
+    {
+      "epoch": 1.6078130984297205,
+      "loss": 1.6141129732131958,
+      "loss_ce": 0.17460237443447113,
+      "loss_iou": 0.6625043153762817,
+      "loss_num": 0.02294921875,
+      "loss_xval": 1.4395105838775635,
+      "num_input_tokens_seen": 725682192,
+      "step": 4198
+    },
+    {
+      "epoch": 1.6081960934507853,
+      "grad_norm": 124.76184771201105,
+      "learning_rate": 5e-06,
+      "loss": 1.4239,
+      "num_input_tokens_seen": 725855096,
+      "step": 4199
+    },
+    {
+      "epoch": 1.6081960934507853,
+      "loss": 1.4835871458053589,
+      "loss_ce": 0.17362385988235474,
+      "loss_iou": 0.6245250701904297,
+      "loss_num": 0.01220703125,
+      "loss_xval": 1.3099632263183594,
+      "num_input_tokens_seen": 725855096,
+      "step": 4199
+    },
+    {
+      "epoch": 1.6085790884718498,
+      "grad_norm": 113.66008530444972,
+      "learning_rate": 5e-06,
+      "loss": 1.668,
+      "num_input_tokens_seen": 726027752,
+      "step": 4200
+    },
+    {
+      "epoch": 1.6085790884718498,
+      "loss": 1.6983070373535156,
+      "loss_ce": 0.17619121074676514,
+      "loss_iou": 0.71070396900177,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 1.52211594581604,
+      "num_input_tokens_seen": 726027752,
+      "step": 4200
+    },
+    {
+      "epoch": 1.6089620834929146,
+      "grad_norm": 79.37422880142995,
+      "learning_rate": 5e-06,
+      "loss": 1.1525,
+      "num_input_tokens_seen": 726200240,
+      "step": 4201
+    },
+    {
+      "epoch": 1.6089620834929146,
+      "loss": 1.1343028545379639,
+      "loss_ce": 0.1569584608078003,
+      "loss_iou": 0.4649295508861542,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.9773444533348083,
+      "num_input_tokens_seen": 726200240,
+      "step": 4201
+    },
+    {
+      "epoch": 1.6093450785139793,
+      "grad_norm": 121.0359814229709,
+      "learning_rate": 5e-06,
+      "loss": 1.1703,
+      "num_input_tokens_seen": 726372488,
+      "step": 4202
+    },
+    {
+      "epoch": 1.6093450785139793,
+      "loss": 1.114872694015503,
+      "loss_ce": 0.13682061433792114,
+      "loss_iou": 0.447262704372406,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.978052020072937,
+      "num_input_tokens_seen": 726372488,
+      "step": 4202
+    },
+    {
+      "epoch": 1.6097280735350439,
+      "grad_norm": 152.19469348382117,
+      "learning_rate": 5e-06,
+      "loss": 1.2371,
+      "num_input_tokens_seen": 726545560,
+      "step": 4203
+    },
+    {
+      "epoch": 1.6097280735350439,
+      "loss": 1.3882474899291992,
+      "loss_ce": 0.1839291751384735,
+      "loss_iou": 0.5568863749504089,
+      "loss_num": 0.01806640625,
+      "loss_xval": 1.2043184041976929,
+      "num_input_tokens_seen": 726545560,
+      "step": 4203
+    },
+    {
+      "epoch": 1.6101110685561086,
+      "grad_norm": 152.85058899432732,
+      "learning_rate": 5e-06,
+      "loss": 1.3286,
+      "num_input_tokens_seen": 726718288,
+      "step": 4204
+    },
+    {
+      "epoch": 1.6101110685561086,
+      "loss": 1.3046131134033203,
+      "loss_ce": 0.15318046510219574,
+      "loss_iou": 0.5195334553718567,
+      "loss_num": 0.0224609375,
+      "loss_xval": 1.1514326333999634,
+      "num_input_tokens_seen": 726718288,
+      "step": 4204
+    },
+    {
+      "epoch": 1.6104940635771734,
+      "grad_norm": 102.86555761747161,
+      "learning_rate": 5e-06,
+      "loss": 1.4742,
+      "num_input_tokens_seen": 726891280,
+      "step": 4205
+    },
+    {
+      "epoch": 1.6104940635771734,
+      "loss": 1.4549144506454468,
+      "loss_ce": 0.18275438249111176,
+      "loss_iou": 0.5906088352203369,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 1.2721600532531738,
+      "num_input_tokens_seen": 726891280,
+      "step": 4205
+    },
+    {
+      "epoch": 1.6108770585982382,
+      "grad_norm": 124.48242543593666,
+      "learning_rate": 5e-06,
+      "loss": 1.4113,
+      "num_input_tokens_seen": 727064296,
+      "step": 4206
+    },
+    {
+      "epoch": 1.6108770585982382,
+      "loss": 1.40886390209198,
+      "loss_ce": 0.17270991206169128,
+      "loss_iou": 0.5808913707733154,
+      "loss_num": 0.014892578125,
+      "loss_xval": 1.2361540794372559,
+      "num_input_tokens_seen": 727064296,
+      "step": 4206
+    },
+    {
+      "epoch": 1.611260053619303,
+      "grad_norm": 152.93814477278573,
+      "learning_rate": 5e-06,
+      "loss": 1.4134,
+      "num_input_tokens_seen": 727237248,
+      "step": 4207
+    },
+    {
+      "epoch": 1.611260053619303,
+      "loss": 1.248939037322998,
+      "loss_ce": 0.16090553998947144,
+      "loss_iou": 0.5178936719894409,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 1.0880334377288818,
+      "num_input_tokens_seen": 727237248,
+      "step": 4207
+    },
+    {
+      "epoch": 1.6116430486403677,
+      "grad_norm": 161.91302956545493,
+      "learning_rate": 5e-06,
+      "loss": 1.6091,
+      "num_input_tokens_seen": 727410024,
+      "step": 4208
+    },
+    {
+      "epoch": 1.6116430486403677,
+      "loss": 1.6586511135101318,
+      "loss_ce": 0.1684921234846115,
+      "loss_iou": 0.7031330466270447,
+      "loss_num": 0.016845703125,
+      "loss_xval": 1.4901589155197144,
+      "num_input_tokens_seen": 727410024,
+      "step": 4208
+    },
+    {
+      "epoch": 1.6120260436614324,
+      "grad_norm": 127.17948579750434,
+      "learning_rate": 5e-06,
+      "loss": 1.2158,
+      "num_input_tokens_seen": 727583024,
+      "step": 4209
+    },
+    {
+      "epoch": 1.6120260436614324,
+      "loss": 1.3533885478973389,
+      "loss_ce": 0.1721852868795395,
+      "loss_iou": 0.5454966425895691,
+      "loss_num": 0.01806640625,
+      "loss_xval": 1.1812032461166382,
+      "num_input_tokens_seen": 727583024,
+      "step": 4209
+    },
+    {
+      "epoch": 1.6124090386824972,
+      "grad_norm": 156.97799307329777,
+      "learning_rate": 5e-06,
+      "loss": 1.4997,
+      "num_input_tokens_seen": 727755576,
+      "step": 4210
+    },
+    {
+      "epoch": 1.6124090386824972,
+      "loss": 1.493351697921753,
+      "loss_ce": 0.11826685070991516,
+      "loss_iou": 0.6387753486633301,
+      "loss_num": 0.01953125,
+      "loss_xval": 1.3750848770141602,
+      "num_input_tokens_seen": 727755576,
+      "step": 4210
+    },
+    {
+      "epoch": 1.612792033703562,
+      "grad_norm": 101.78998317340869,
+      "learning_rate": 5e-06,
+      "loss": 1.3822,
+      "num_input_tokens_seen": 727928264,
+      "step": 4211
+    },
+    {
+      "epoch": 1.612792033703562,
+      "loss": 1.5854644775390625,
+      "loss_ce": 0.1664758026599884,
+      "loss_iou": 0.6659457683563232,
+      "loss_num": 0.017333984375,
+      "loss_xval": 1.4189887046813965,
+      "num_input_tokens_seen": 727928264,
+      "step": 4211
+    },
+    {
+      "epoch": 1.6131750287246267,
+      "grad_norm": 162.3333980065569,
+      "learning_rate": 5e-06,
+      "loss": 1.6166,
+      "num_input_tokens_seen": 728100824,
+      "step": 4212
+    },
+    {
+      "epoch": 1.6131750287246267,
+      "loss": 1.9017621278762817,
+      "loss_ce": 0.1310673952102661,
+      "loss_iou": 0.8206501007080078,
+      "loss_num": 0.02587890625,
+      "loss_xval": 1.7706948518753052,
+      "num_input_tokens_seen": 728100824,
+      "step": 4212
+    },
+    {
+      "epoch": 1.6135580237456915,
+      "grad_norm": 167.7287404606955,
+      "learning_rate": 5e-06,
+      "loss": 1.478,
+      "num_input_tokens_seen": 728273472,
+      "step": 4213
+    },
+    {
+      "epoch": 1.6135580237456915,
+      "loss": 1.4391961097717285,
+      "loss_ce": 0.16294166445732117,
+      "loss_iou": 0.6000260710716248,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 1.2762545347213745,
+      "num_input_tokens_seen": 728273472,
+      "step": 4213
+    },
+    {
+      "epoch": 1.613941018766756,
+      "grad_norm": 193.79280839460625,
+      "learning_rate": 5e-06,
+      "loss": 1.7893,
+      "num_input_tokens_seen": 728446560,
+      "step": 4214
+    },
+    {
+      "epoch": 1.613941018766756,
+      "loss": 1.788272738456726,
+      "loss_ce": 0.14585144817829132,
+      "loss_iou": 0.7796915173530579,
+      "loss_num": 0.0166015625,
+      "loss_xval": 1.6424213647842407,
+      "num_input_tokens_seen": 728446560,
+      "step": 4214
+    },
+    {
+      "epoch": 1.6143240137878208,
+      "grad_norm": 107.96259250437333,
+      "learning_rate": 5e-06,
+      "loss": 1.3204,
+      "num_input_tokens_seen": 728619168,
+      "step": 4215
+    },
+    {
+      "epoch": 1.6143240137878208,
+      "loss": 1.391005039215088,
+      "loss_ce": 0.16493898630142212,
+      "loss_iou": 0.5781972408294678,
+      "loss_num": 0.013916015625,
+      "loss_xval": 1.2260661125183105,
+      "num_input_tokens_seen": 728619168,
+      "step": 4215
+    },
+    {
+      "epoch": 1.6147070088088855,
+      "grad_norm": 187.0352787801529,
+      "learning_rate": 5e-06,
+      "loss": 1.4137,
+      "num_input_tokens_seen": 728792184,
+      "step": 4216
+    },
+    {
+      "epoch": 1.6147070088088855,
+      "loss": 1.4656509160995483,
+      "loss_ce": 0.14935088157653809,
+      "loss_iou": 0.6199725270271301,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 1.3163000345230103,
+      "num_input_tokens_seen": 728792184,
+      "step": 4216
+    },
+    {
+      "epoch": 1.61509000382995,
+      "grad_norm": 169.04310416732002,
+      "learning_rate": 5e-06,
+      "loss": 1.663,
+      "num_input_tokens_seen": 728964840,
+      "step": 4217
+    },
+    {
+      "epoch": 1.61509000382995,
+      "loss": 1.6586328744888306,
+      "loss_ce": 0.1359909176826477,
+      "loss_iou": 0.699309229850769,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 1.522641897201538,
+      "num_input_tokens_seen": 728964840,
+      "step": 4217
+    },
+    {
+      "epoch": 1.6154729988510148,
+      "grad_norm": 152.06177212909685,
+      "learning_rate": 5e-06,
+      "loss": 1.5059,
+      "num_input_tokens_seen": 729137720,
+      "step": 4218
+    },
+    {
+      "epoch": 1.6154729988510148,
+      "loss": 1.583996295928955,
+      "loss_ce": 0.17172470688819885,
+      "loss_iou": 0.6719104051589966,
+      "loss_num": 0.013671875,
+      "loss_xval": 1.4122717380523682,
+      "num_input_tokens_seen": 729137720,
+      "step": 4218
+    },
+    {
+      "epoch": 1.6158559938720796,
+      "grad_norm": 200.02509033924764,
+      "learning_rate": 5e-06,
+      "loss": 1.6775,
+      "num_input_tokens_seen": 729310504,
+      "step": 4219
+    },
+    {
+      "epoch": 1.6158559938720796,
+      "loss": 1.5794682502746582,
+      "loss_ce": 0.16939860582351685,
+      "loss_iou": 0.6682611703872681,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 1.4100697040557861,
+      "num_input_tokens_seen": 729310504,
+      "step": 4219
+    },
+    {
+      "epoch": 1.6162389888931443,
+      "grad_norm": 277.57050945481717,
+      "learning_rate": 5e-06,
+      "loss": 1.9702,
+      "num_input_tokens_seen": 729483616,
+      "step": 4220
+    },
+    {
+      "epoch": 1.6162389888931443,
+      "loss": 1.8970720767974854,
+      "loss_ce": 0.13330280780792236,
+      "loss_iou": 0.8094968795776367,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 1.7637691497802734,
+      "num_input_tokens_seen": 729483616,
+      "step": 4220
+    },
+    {
+      "epoch": 1.616621983914209,
+      "grad_norm": 83.35662605388745,
+      "learning_rate": 5e-06,
+      "loss": 1.9715,
+      "num_input_tokens_seen": 729656672,
+      "step": 4221
+    },
+    {
+      "epoch": 1.616621983914209,
+      "loss": 1.9594939947128296,
+      "loss_ce": 0.16559158265590668,
+      "loss_iou": 0.8036589622497559,
+      "loss_num": 0.037353515625,
+      "loss_xval": 1.7939023971557617,
+      "num_input_tokens_seen": 729656672,
+      "step": 4221
+    },
+    {
+      "epoch": 1.6170049789352738,
+      "grad_norm": 156.0426568366871,
+      "learning_rate": 5e-06,
+      "loss": 1.7302,
+      "num_input_tokens_seen": 729829664,
+      "step": 4222
+    },
+    {
+      "epoch": 1.6170049789352738,
+      "loss": 1.7628661394119263,
+      "loss_ce": 0.14224308729171753,
+      "loss_iou": 0.743722140789032,
+      "loss_num": 0.026611328125,
+      "loss_xval": 1.620622992515564,
+      "num_input_tokens_seen": 729829664,
+      "step": 4222
+    },
+    {
+      "epoch": 1.6173879739563386,
+      "grad_norm": 160.89069524338066,
+      "learning_rate": 5e-06,
+      "loss": 1.6602,
+      "num_input_tokens_seen": 730002456,
+      "step": 4223
+    },
+    {
+      "epoch": 1.6173879739563386,
+      "loss": 1.6575422286987305,
+      "loss_ce": 0.16317525506019592,
+      "loss_iou": 0.6911532282829285,
+      "loss_num": 0.0224609375,
+      "loss_xval": 1.494367003440857,
+      "num_input_tokens_seen": 730002456,
+      "step": 4223
+    },
+    {
+      "epoch": 1.6177709689774034,
+      "grad_norm": 176.03727937909886,
+      "learning_rate": 5e-06,
+      "loss": 1.9298,
+      "num_input_tokens_seen": 730175536,
+      "step": 4224
+    },
+    {
+      "epoch": 1.6177709689774034,
+      "loss": 1.876756191253662,
+      "loss_ce": 0.18384908139705658,
+      "loss_iou": 0.8029355406761169,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 1.6929072141647339,
+      "num_input_tokens_seen": 730175536,
+      "step": 4224
+    },
+    {
+      "epoch": 1.6181539639984681,
+      "grad_norm": 77.70857364021485,
+      "learning_rate": 5e-06,
+      "loss": 1.91,
+      "num_input_tokens_seen": 730348768,
+      "step": 4225
+    },
+    {
+      "epoch": 1.6181539639984681,
+      "loss": 1.9864534139633179,
+      "loss_ce": 0.1839562952518463,
+      "loss_iou": 0.822848916053772,
+      "loss_num": 0.03125,
+      "loss_xval": 1.802497148513794,
+      "num_input_tokens_seen": 730348768,
+      "step": 4225
+    },
+    {
+      "epoch": 1.6185369590195329,
+      "grad_norm": 98.3553739132101,
+      "learning_rate": 5e-06,
+      "loss": 1.4728,
+      "num_input_tokens_seen": 730521824,
+      "step": 4226
+    },
+    {
+      "epoch": 1.6185369590195329,
+      "loss": 1.4634766578674316,
+      "loss_ce": 0.144941046833992,
+      "loss_iou": 0.6283076405525208,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 1.3185354471206665,
+      "num_input_tokens_seen": 730521824,
+      "step": 4226
+    },
+    {
+      "epoch": 1.6189199540405976,
+      "grad_norm": 179.6512128786453,
+      "learning_rate": 5e-06,
+      "loss": 1.7384,
+      "num_input_tokens_seen": 730694672,
+      "step": 4227
+    },
+    {
+      "epoch": 1.6189199540405976,
+      "loss": 1.8198180198669434,
+      "loss_ce": 0.1349303126335144,
+      "loss_iou": 0.8038392066955566,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 1.6848878860473633,
+      "num_input_tokens_seen": 730694672,
+      "step": 4227
+    },
+    {
+      "epoch": 1.6193029490616622,
+      "grad_norm": 223.12057888652353,
+      "learning_rate": 5e-06,
+      "loss": 2.0535,
+      "num_input_tokens_seen": 730867736,
+      "step": 4228
+    },
+    {
+      "epoch": 1.6193029490616622,
+      "loss": 2.180980920791626,
+      "loss_ce": 0.1341322958469391,
+      "loss_iou": 0.9227772951126099,
+      "loss_num": 0.040283203125,
+      "loss_xval": 2.0468485355377197,
+      "num_input_tokens_seen": 730867736,
+      "step": 4228
+    },
+    {
+      "epoch": 1.619685944082727,
+      "grad_norm": 114.72192326790878,
+      "learning_rate": 5e-06,
+      "loss": 1.9036,
+      "num_input_tokens_seen": 731041016,
+      "step": 4229
+    },
+    {
+      "epoch": 1.619685944082727,
+      "loss": 1.9049959182739258,
+      "loss_ce": 0.18319116532802582,
+      "loss_iou": 0.7841202020645142,
+      "loss_num": 0.03076171875,
+      "loss_xval": 1.7218048572540283,
+      "num_input_tokens_seen": 731041016,
+      "step": 4229
+    },
+    {
+      "epoch": 1.6200689391037917,
+      "grad_norm": 61.207408232360386,
+      "learning_rate": 5e-06,
+      "loss": 1.2428,
+      "num_input_tokens_seen": 731213896,
+      "step": 4230
+    },
+    {
+      "epoch": 1.6200689391037917,
+      "loss": 1.3117761611938477,
+      "loss_ce": 0.15150532126426697,
+      "loss_iou": 0.5487176179885864,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 1.1602709293365479,
+      "num_input_tokens_seen": 731213896,
+      "step": 4230
+    },
+    {
+      "epoch": 1.6204519341248562,
+      "grad_norm": 98.28703306409523,
+      "learning_rate": 5e-06,
+      "loss": 1.1018,
+      "num_input_tokens_seen": 731386744,
+      "step": 4231
+    },
+    {
+      "epoch": 1.6204519341248562,
+      "loss": 1.0563000440597534,
+      "loss_ce": 0.0884590819478035,
+      "loss_iou": 0.45318925380706787,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.9494082927703857,
+      "num_input_tokens_seen": 731386744,
+      "step": 4231
+    },
+    {
+      "epoch": 1.620834929145921,
+      "grad_norm": 165.15158453212973,
+      "learning_rate": 5e-06,
+      "loss": 1.4172,
+      "num_input_tokens_seen": 731559808,
+      "step": 4232
+    },
+    {
+      "epoch": 1.620834929145921,
+      "loss": 1.3487050533294678,
+      "loss_ce": 0.171361044049263,
+      "loss_iou": 0.5394319295883179,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 1.1773440837860107,
+      "num_input_tokens_seen": 731559808,
+      "step": 4232
+    },
+    {
+      "epoch": 1.6212179241669857,
+      "grad_norm": 209.7155277261917,
+      "learning_rate": 5e-06,
+      "loss": 1.7705,
+      "num_input_tokens_seen": 731732840,
+      "step": 4233
+    },
+    {
+      "epoch": 1.6212179241669857,
+      "loss": 1.8195605278015137,
+      "loss_ce": 0.14862287044525146,
+      "loss_iou": 0.7685896158218384,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 1.6709377765655518,
+      "num_input_tokens_seen": 731732840,
+      "step": 4233
+    },
+    {
+      "epoch": 1.6216009191880505,
+      "grad_norm": 135.34385611849774,
+      "learning_rate": 5e-06,
+      "loss": 1.5091,
+      "num_input_tokens_seen": 731905344,
+      "step": 4234
+    },
+    {
+      "epoch": 1.6216009191880505,
+      "loss": 1.689784288406372,
+      "loss_ce": 0.14371293783187866,
+      "loss_iou": 0.7203012704849243,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 1.5460712909698486,
+      "num_input_tokens_seen": 731905344,
+      "step": 4234
+    },
+    {
+      "epoch": 1.6219839142091153,
+      "grad_norm": 297.4578435313298,
+      "learning_rate": 5e-06,
+      "loss": 1.4793,
+      "num_input_tokens_seen": 732078448,
+      "step": 4235
+    },
+    {
+      "epoch": 1.6219839142091153,
+      "loss": 1.3251738548278809,
+      "loss_ce": 0.18002194166183472,
+      "loss_iou": 0.529378354549408,
+      "loss_num": 0.017333984375,
+      "loss_xval": 1.145151972770691,
+      "num_input_tokens_seen": 732078448,
+      "step": 4235
+    },
+    {
+      "epoch": 1.62236690923018,
+      "grad_norm": 86.83817184916266,
+      "learning_rate": 5e-06,
+      "loss": 1.2281,
+      "num_input_tokens_seen": 732251432,
+      "step": 4236
+    },
+    {
+      "epoch": 1.62236690923018,
+      "loss": 1.221873164176941,
+      "loss_ce": 0.16788625717163086,
+      "loss_iou": 0.47956913709640503,
+      "loss_num": 0.01904296875,
+      "loss_xval": 1.05398690700531,
+      "num_input_tokens_seen": 732251432,
+      "step": 4236
+    },
+    {
+      "epoch": 1.6227499042512448,
+      "grad_norm": 113.9655956875708,
+      "learning_rate": 5e-06,
+      "loss": 1.5004,
+      "num_input_tokens_seen": 732424424,
+      "step": 4237
+    },
+    {
+      "epoch": 1.6227499042512448,
+      "loss": 1.6975293159484863,
+      "loss_ce": 0.14779433608055115,
+      "loss_iou": 0.7001604437828064,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 1.5497349500656128,
+      "num_input_tokens_seen": 732424424,
+      "step": 4237
+    },
+    {
+      "epoch": 1.6231328992723095,
+      "grad_norm": 131.99938853002732,
+      "learning_rate": 5e-06,
+      "loss": 1.5101,
+      "num_input_tokens_seen": 732597216,
+      "step": 4238
+    },
+    {
+      "epoch": 1.6231328992723095,
+      "loss": 1.4549403190612793,
+      "loss_ce": 0.1477314680814743,
+      "loss_iou": 0.5885409712791443,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 1.2988470792770386,
+      "num_input_tokens_seen": 732597216,
+      "step": 4238
+    },
+    {
+      "epoch": 1.6235158942933743,
+      "grad_norm": 116.74818789125096,
+      "learning_rate": 5e-06,
+      "loss": 1.8328,
+      "num_input_tokens_seen": 732770272,
+      "step": 4239
+    },
+    {
+      "epoch": 1.6235158942933743,
+      "loss": 1.7853672504425049,
+      "loss_ce": 0.17761091887950897,
+      "loss_iou": 0.7398828268051147,
+      "loss_num": 0.025634765625,
+      "loss_xval": 1.6077563762664795,
+      "num_input_tokens_seen": 732770272,
+      "step": 4239
+    },
+    {
+      "epoch": 1.623898889314439,
+      "grad_norm": 115.16556321758176,
+      "learning_rate": 5e-06,
+      "loss": 1.4008,
+      "num_input_tokens_seen": 732943208,
+      "step": 4240
+    },
+    {
+      "epoch": 1.623898889314439,
+      "loss": 1.5703788995742798,
+      "loss_ce": 0.1624567210674286,
+      "loss_iou": 0.6411864161491394,
+      "loss_num": 0.025146484375,
+      "loss_xval": 1.4079221487045288,
+      "num_input_tokens_seen": 732943208,
+      "step": 4240
+    },
+    {
+      "epoch": 1.6242818843355038,
+      "grad_norm": 160.31123185536194,
+      "learning_rate": 5e-06,
+      "loss": 1.5956,
+      "num_input_tokens_seen": 733116376,
+      "step": 4241
+    },
+    {
+      "epoch": 1.6242818843355038,
+      "loss": 1.5039113759994507,
+      "loss_ce": 0.14939121901988983,
+      "loss_iou": 0.6216417551040649,
+      "loss_num": 0.022216796875,
+      "loss_xval": 1.3545200824737549,
+      "num_input_tokens_seen": 733116376,
+      "step": 4241
+    },
+    {
+      "epoch": 1.6246648793565683,
+      "grad_norm": 149.68244304531984,
+      "learning_rate": 5e-06,
+      "loss": 1.7246,
+      "num_input_tokens_seen": 733289152,
+      "step": 4242
+    },
+    {
+      "epoch": 1.6246648793565683,
+      "loss": 1.6408194303512573,
+      "loss_ce": 0.14709943532943726,
+      "loss_iou": 0.6603731513023376,
+      "loss_num": 0.03466796875,
+      "loss_xval": 1.4937199354171753,
+      "num_input_tokens_seen": 733289152,
+      "step": 4242
+    },
+    {
+      "epoch": 1.625047874377633,
+      "grad_norm": 64.99801486935895,
+      "learning_rate": 5e-06,
+      "loss": 1.3936,
+      "num_input_tokens_seen": 733462248,
+      "step": 4243
+    },
+    {
+      "epoch": 1.625047874377633,
+      "loss": 1.3872981071472168,
+      "loss_ce": 0.15859389305114746,
+      "loss_iou": 0.5587490797042847,
+      "loss_num": 0.022216796875,
+      "loss_xval": 1.2287042140960693,
+      "num_input_tokens_seen": 733462248,
+      "step": 4243
+    },
+    {
+      "epoch": 1.6254308693986979,
+      "grad_norm": 104.31327966536591,
+      "learning_rate": 5e-06,
+      "loss": 1.1563,
+      "num_input_tokens_seen": 733635496,
+      "step": 4244
+    },
+    {
+      "epoch": 1.6254308693986979,
+      "loss": 1.0872504711151123,
+      "loss_ce": 0.16980624198913574,
+      "loss_iou": 0.4264497756958008,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.9174442291259766,
+      "num_input_tokens_seen": 733635496,
+      "step": 4244
+    },
+    {
+      "epoch": 1.6258138644197624,
+      "grad_norm": 120.54582933263765,
+      "learning_rate": 5e-06,
+      "loss": 1.3233,
+      "num_input_tokens_seen": 733808528,
+      "step": 4245
+    },
+    {
+      "epoch": 1.6258138644197624,
+      "loss": 1.3540951013565063,
+      "loss_ce": 0.19186602532863617,
+      "loss_iou": 0.5447909832000732,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 1.162229061126709,
+      "num_input_tokens_seen": 733808528,
+      "step": 4245
+    },
+    {
+      "epoch": 1.6261968594408271,
+      "grad_norm": 155.19784512722597,
+      "learning_rate": 5e-06,
+      "loss": 1.5014,
+      "num_input_tokens_seen": 733981488,
+      "step": 4246
+    },
+    {
+      "epoch": 1.6261968594408271,
+      "loss": 1.5388509035110474,
+      "loss_ce": 0.1324869990348816,
+      "loss_iou": 0.6651875972747803,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 1.4063639640808105,
+      "num_input_tokens_seen": 733981488,
+      "step": 4246
+    },
+    {
+      "epoch": 1.626579854461892,
+      "grad_norm": 166.99671583118356,
+      "learning_rate": 5e-06,
+      "loss": 1.5675,
+      "num_input_tokens_seen": 734154400,
+      "step": 4247
+    },
+    {
+      "epoch": 1.626579854461892,
+      "loss": 1.4532661437988281,
+      "loss_ce": 0.13205987215042114,
+      "loss_iou": 0.6123547554016113,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.3212060928344727,
+      "num_input_tokens_seen": 734154400,
+      "step": 4247
+    },
+    {
+      "epoch": 1.6269628494829567,
+      "grad_norm": 153.52289951618562,
+      "learning_rate": 5e-06,
+      "loss": 1.4302,
+      "num_input_tokens_seen": 734326904,
+      "step": 4248
+    },
+    {
+      "epoch": 1.6269628494829567,
+      "loss": 1.346290946006775,
+      "loss_ce": 0.15899604558944702,
+      "loss_iou": 0.5563092231750488,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 1.1872949600219727,
+      "num_input_tokens_seen": 734326904,
+      "step": 4248
+    },
+    {
+      "epoch": 1.6273458445040214,
+      "grad_norm": 138.05420791491989,
+      "learning_rate": 5e-06,
+      "loss": 1.4822,
+      "num_input_tokens_seen": 734499984,
+      "step": 4249
+    },
+    {
+      "epoch": 1.6273458445040214,
+      "loss": 1.5413062572479248,
+      "loss_ce": 0.21249623596668243,
+      "loss_iou": 0.6194525957107544,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 1.3288099765777588,
+      "num_input_tokens_seen": 734499984,
+      "step": 4249
+    },
+    {
+      "epoch": 1.6277288395250862,
+      "grad_norm": 100.62585858377918,
+      "learning_rate": 5e-06,
+      "loss": 1.36,
+      "num_input_tokens_seen": 734672824,
+      "step": 4250
+    },
+    {
+      "epoch": 1.6277288395250862,
+      "eval_websight_new_CIoU": 0.9162223041057587,
+      "eval_websight_new_GIoU": 0.9161066710948944,
+      "eval_websight_new_IoU": 0.9164418876171112,
+      "eval_websight_new_MAE_all": 0.009378555696457624,
+      "eval_websight_new_MAE_h": 0.007857543416321278,
+      "eval_websight_new_MAE_w": 0.015783830545842648,
+      "eval_websight_new_MAE_x": 0.008010117802768946,
+      "eval_websight_new_MAE_y": 0.005862730089575052,
+      "eval_websight_new_NUM_probability": 0.401885524392128,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 1.018053650856018,
+      "eval_websight_new_loss_ce": 0.25705206394195557,
+      "eval_websight_new_loss_iou": 0.3521401435136795,
+      "eval_websight_new_loss_num": 0.01025390625,
+      "eval_websight_new_loss_xval": 0.7555345594882965,
+      "eval_websight_new_runtime": 56.1371,
+      "eval_websight_new_samples_per_second": 0.891,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 734672824,
+      "step": 4250
+    },
+    {
+      "epoch": 1.6277288395250862,
+      "eval_seeclick_CIoU": 0.6385174989700317,
+      "eval_seeclick_GIoU": 0.652334988117218,
+      "eval_seeclick_IoU": 0.6844546496868134,
+      "eval_seeclick_MAE_all": 0.06909987516701221,
+      "eval_seeclick_MAE_h": 0.022533646784722805,
+      "eval_seeclick_MAE_w": 0.11771515384316444,
+      "eval_seeclick_MAE_x": 0.11195151880383492,
+      "eval_seeclick_MAE_y": 0.02419918216764927,
+      "eval_seeclick_NUM_probability": 0.5888088047504425,
+      "eval_seeclick_inside_bbox": 0.8854166567325592,
+      "eval_seeclick_loss": 1.6230900287628174,
+      "eval_seeclick_loss_ce": 0.19231893867254257,
+      "eval_seeclick_loss_iou": 0.5816226601600647,
+      "eval_seeclick_loss_num": 0.0477447509765625,
+      "eval_seeclick_loss_xval": 1.401816487312317,
+      "eval_seeclick_runtime": 88.333,
+      "eval_seeclick_samples_per_second": 0.566,
+      "eval_seeclick_steps_per_second": 0.023,
+      "num_input_tokens_seen": 734672824,
+      "step": 4250
+    },
+    {
+      "epoch": 1.6277288395250862,
+      "eval_icons_CIoU": 0.8221758902072906,
+      "eval_icons_GIoU": 0.8194511830806732,
+      "eval_icons_IoU": 0.8379125595092773,
+      "eval_icons_MAE_all": 0.03192135691642761,
+      "eval_icons_MAE_h": 0.02202616073191166,
+      "eval_icons_MAE_w": 0.04361097142100334,
+      "eval_icons_MAE_x": 0.042733339592814445,
+      "eval_icons_MAE_y": 0.01931495126336813,
+      "eval_icons_NUM_probability": 0.40384694933891296,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 1.9730126857757568,
+      "eval_icons_loss_ce": 0.38168513774871826,
+      "eval_icons_loss_iou": 0.7400006651878357,
+      "eval_icons_loss_num": 0.022357940673828125,
+      "eval_icons_loss_xval": 1.5917567014694214,
+      "eval_icons_runtime": 87.2141,
+      "eval_icons_samples_per_second": 0.573,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 734672824,
+      "step": 4250
+    },
+    {
+      "epoch": 1.6277288395250862,
+      "loss": 2.055295944213867,
+      "loss_ce": 0.3824533224105835,
+      "loss_iou": 0.7747604846954346,
+      "loss_num": 0.024658203125,
+      "loss_xval": 1.6728425025939941,
+      "num_input_tokens_seen": 734672824,
+      "step": 4250
+    },
+    {
+      "epoch": 1.628111834546151,
+      "grad_norm": 79.00335484482997,
+      "learning_rate": 5e-06,
+      "loss": 1.4629,
+      "num_input_tokens_seen": 734846360,
+      "step": 4251
+    },
+    {
+      "epoch": 1.628111834546151,
+      "loss": 1.3319406509399414,
+      "loss_ce": 0.14555197954177856,
+      "loss_iou": 0.5480893850326538,
+      "loss_num": 0.01806640625,
+      "loss_xval": 1.1863887310028076,
+      "num_input_tokens_seen": 734846360,
+      "step": 4251
+    },
+    {
+      "epoch": 1.6284948295672157,
+      "grad_norm": 175.7979061714301,
+      "learning_rate": 5e-06,
+      "loss": 1.355,
+      "num_input_tokens_seen": 735019488,
+      "step": 4252
+    },
+    {
+      "epoch": 1.6284948295672157,
+      "loss": 1.2803380489349365,
+      "loss_ce": 0.15754714608192444,
+      "loss_iou": 0.5031984448432922,
+      "loss_num": 0.023193359375,
+      "loss_xval": 1.1227909326553345,
+      "num_input_tokens_seen": 735019488,
+      "step": 4252
+    },
+    {
+      "epoch": 1.6288778245882805,
+      "grad_norm": 177.45530485706,
+      "learning_rate": 5e-06,
+      "loss": 1.925,
+      "num_input_tokens_seen": 735192456,
+      "step": 4253
+    },
+    {
+      "epoch": 1.6288778245882805,
+      "loss": 1.9332362413406372,
+      "loss_ce": 0.2007823884487152,
+      "loss_iou": 0.7958838939666748,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 1.7324538230895996,
+      "num_input_tokens_seen": 735192456,
+      "step": 4253
+    },
+    {
+      "epoch": 1.6292608196093452,
+      "grad_norm": 88.9595650995408,
+      "learning_rate": 5e-06,
+      "loss": 1.2875,
+      "num_input_tokens_seen": 735365296,
+      "step": 4254
+    },
+    {
+      "epoch": 1.6292608196093452,
+      "loss": 1.3784162998199463,
+      "loss_ce": 0.13960078358650208,
+      "loss_iou": 0.5806809663772583,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 1.2388155460357666,
+      "num_input_tokens_seen": 735365296,
+      "step": 4254
+    },
+    {
+      "epoch": 1.62964381463041,
+      "grad_norm": 236.99011132617477,
+      "learning_rate": 5e-06,
+      "loss": 1.3328,
+      "num_input_tokens_seen": 735538528,
+      "step": 4255
+    },
+    {
+      "epoch": 1.62964381463041,
+      "loss": 1.2332539558410645,
+      "loss_ce": 0.12416472285985947,
+      "loss_iou": 0.5107366442680359,
+      "loss_num": 0.017578125,
+      "loss_xval": 1.1090892553329468,
+      "num_input_tokens_seen": 735538528,
+      "step": 4255
+    },
+    {
+      "epoch": 1.6300268096514745,
+      "grad_norm": 173.3068144320145,
+      "learning_rate": 5e-06,
+      "loss": 1.8645,
+      "num_input_tokens_seen": 735711648,
+      "step": 4256
+    },
+    {
+      "epoch": 1.6300268096514745,
+      "loss": 1.8725402355194092,
+      "loss_ce": 0.1486969292163849,
+      "loss_iou": 0.7692092657089233,
+      "loss_num": 0.037109375,
+      "loss_xval": 1.7238433361053467,
+      "num_input_tokens_seen": 735711648,
+      "step": 4256
+    },
+    {
+      "epoch": 1.6304098046725393,
+      "grad_norm": 129.64462652022746,
+      "learning_rate": 5e-06,
+      "loss": 1.2881,
+      "num_input_tokens_seen": 735884368,
+      "step": 4257
+    },
+    {
+      "epoch": 1.6304098046725393,
+      "loss": 1.2674479484558105,
+      "loss_ce": 0.13030312955379486,
+      "loss_iou": 0.514647901058197,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 1.137144923210144,
+      "num_input_tokens_seen": 735884368,
+      "step": 4257
+    },
+    {
+      "epoch": 1.630792799693604,
+      "grad_norm": 179.32605556331572,
+      "learning_rate": 5e-06,
+      "loss": 1.542,
+      "num_input_tokens_seen": 736057584,
+      "step": 4258
+    },
+    {
+      "epoch": 1.630792799693604,
+      "loss": 1.4749085903167725,
+      "loss_ce": 0.15508228540420532,
+      "loss_iou": 0.6110392808914185,
+      "loss_num": 0.01953125,
+      "loss_xval": 1.319826364517212,
+      "num_input_tokens_seen": 736057584,
+      "step": 4258
+    },
+    {
+      "epoch": 1.6311757947146686,
+      "grad_norm": 120.53814416384229,
+      "learning_rate": 5e-06,
+      "loss": 1.5756,
+      "num_input_tokens_seen": 736230544,
+      "step": 4259
+    },
+    {
+      "epoch": 1.6311757947146686,
+      "loss": 1.6546740531921387,
+      "loss_ce": 0.16424831748008728,
+      "loss_iou": 0.6956065893173218,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 1.4904258251190186,
+      "num_input_tokens_seen": 736230544,
+      "step": 4259
+    },
+    {
+      "epoch": 1.6315587897357333,
+      "grad_norm": 74.18145854487854,
+      "learning_rate": 5e-06,
+      "loss": 1.153,
+      "num_input_tokens_seen": 736403896,
+      "step": 4260
+    },
+    {
+      "epoch": 1.6315587897357333,
+      "loss": 1.2199480533599854,
+      "loss_ce": 0.14292341470718384,
+      "loss_iou": 0.5010062456130981,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 1.0770246982574463,
+      "num_input_tokens_seen": 736403896,
+      "step": 4260
+    },
+    {
+      "epoch": 1.631941784756798,
+      "grad_norm": 81.70348520697706,
+      "learning_rate": 5e-06,
+      "loss": 1.3334,
+      "num_input_tokens_seen": 736576688,
+      "step": 4261
+    },
+    {
+      "epoch": 1.631941784756798,
+      "loss": 1.3570826053619385,
+      "loss_ce": 0.159705251455307,
+      "loss_iou": 0.5623422265052795,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 1.197377324104309,
+      "num_input_tokens_seen": 736576688,
+      "step": 4261
+    },
+    {
+      "epoch": 1.6323247797778628,
+      "grad_norm": 255.40016348708113,
+      "learning_rate": 5e-06,
+      "loss": 1.4284,
+      "num_input_tokens_seen": 736750192,
+      "step": 4262
+    },
+    {
+      "epoch": 1.6323247797778628,
+      "loss": 1.5462151765823364,
+      "loss_ce": 0.18443627655506134,
+      "loss_iou": 0.6099055409431458,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 1.3617788553237915,
+      "num_input_tokens_seen": 736750192,
+      "step": 4262
+    },
+    {
+      "epoch": 1.6327077747989276,
+      "grad_norm": 169.70891739035935,
+      "learning_rate": 5e-06,
+      "loss": 1.7396,
+      "num_input_tokens_seen": 736923136,
+      "step": 4263
+    },
+    {
+      "epoch": 1.6327077747989276,
+      "loss": 1.6810753345489502,
+      "loss_ce": 0.13653725385665894,
+      "loss_iou": 0.6868503093719482,
+      "loss_num": 0.0341796875,
+      "loss_xval": 1.5445380210876465,
+      "num_input_tokens_seen": 736923136,
+      "step": 4263
+    },
+    {
+      "epoch": 1.6330907698199923,
+      "grad_norm": 80.76270682928212,
+      "learning_rate": 5e-06,
+      "loss": 1.3823,
+      "num_input_tokens_seen": 737096320,
+      "step": 4264
+    },
+    {
+      "epoch": 1.6330907698199923,
+      "loss": 1.3377436399459839,
+      "loss_ce": 0.20186583697795868,
+      "loss_iou": 0.5355597734451294,
+      "loss_num": 0.012939453125,
+      "loss_xval": 1.1358778476715088,
+      "num_input_tokens_seen": 737096320,
+      "step": 4264
+    },
+    {
+      "epoch": 1.633473764841057,
+      "grad_norm": 136.92582532641694,
+      "learning_rate": 5e-06,
+      "loss": 1.2552,
+      "num_input_tokens_seen": 737269128,
+      "step": 4265
+    },
+    {
+      "epoch": 1.633473764841057,
+      "loss": 1.2514030933380127,
+      "loss_ce": 0.16696761548519135,
+      "loss_iou": 0.5057186484336853,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 1.0844353437423706,
+      "num_input_tokens_seen": 737269128,
+      "step": 4265
+    },
+    {
+      "epoch": 1.6338567598621219,
+      "grad_norm": 186.46948201714852,
+      "learning_rate": 5e-06,
+      "loss": 1.3126,
+      "num_input_tokens_seen": 737441792,
+      "step": 4266
+    },
+    {
+      "epoch": 1.6338567598621219,
+      "loss": 1.2710222005844116,
+      "loss_ce": 0.14832234382629395,
+      "loss_iou": 0.5136738419532776,
+      "loss_num": 0.01904296875,
+      "loss_xval": 1.1226998567581177,
+      "num_input_tokens_seen": 737441792,
+      "step": 4266
+    },
+    {
+      "epoch": 1.6342397548831866,
+      "grad_norm": 163.69091903635953,
+      "learning_rate": 5e-06,
+      "loss": 2.1152,
+      "num_input_tokens_seen": 737614592,
+      "step": 4267
+    },
+    {
+      "epoch": 1.6342397548831866,
+      "loss": 2.1176929473876953,
+      "loss_ce": 0.15432396531105042,
+      "loss_iou": 0.8610180616378784,
+      "loss_num": 0.04833984375,
+      "loss_xval": 1.9633691310882568,
+      "num_input_tokens_seen": 737614592,
+      "step": 4267
+    },
+    {
+      "epoch": 1.6346227499042514,
+      "grad_norm": 116.22006071019236,
+      "learning_rate": 5e-06,
+      "loss": 1.5089,
+      "num_input_tokens_seen": 737787440,
+      "step": 4268
+    },
+    {
+      "epoch": 1.6346227499042514,
+      "loss": 1.5042097568511963,
+      "loss_ce": 0.15562763810157776,
+      "loss_iou": 0.6205800771713257,
+      "loss_num": 0.021484375,
+      "loss_xval": 1.3485820293426514,
+      "num_input_tokens_seen": 737787440,
+      "step": 4268
+    },
+    {
+      "epoch": 1.635005744925316,
+      "grad_norm": 119.21878105702457,
+      "learning_rate": 5e-06,
+      "loss": 1.4444,
+      "num_input_tokens_seen": 737960080,
+      "step": 4269
+    },
+    {
+      "epoch": 1.635005744925316,
+      "loss": 1.3537085056304932,
+      "loss_ce": 0.1828244924545288,
+      "loss_iou": 0.5500721335411072,
+      "loss_num": 0.01416015625,
+      "loss_xval": 1.1708840131759644,
+      "num_input_tokens_seen": 737960080,
+      "step": 4269
+    },
+    {
+      "epoch": 1.6353887399463807,
+      "grad_norm": 109.5187969297078,
+      "learning_rate": 5e-06,
+      "loss": 1.7138,
+      "num_input_tokens_seen": 738133296,
+      "step": 4270
+    },
+    {
+      "epoch": 1.6353887399463807,
+      "loss": 1.6863322257995605,
+      "loss_ce": 0.14331993460655212,
+      "loss_iou": 0.7184818387031555,
+      "loss_num": 0.021240234375,
+      "loss_xval": 1.5430123805999756,
+      "num_input_tokens_seen": 738133296,
+      "step": 4270
+    },
+    {
+      "epoch": 1.6357717349674454,
+      "grad_norm": 74.76696792186888,
+      "learning_rate": 5e-06,
+      "loss": 1.264,
+      "num_input_tokens_seen": 738306352,
+      "step": 4271
+    },
+    {
+      "epoch": 1.6357717349674454,
+      "loss": 1.2260103225708008,
+      "loss_ce": 0.18773356080055237,
+      "loss_iou": 0.49386221170425415,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 1.0382767915725708,
+      "num_input_tokens_seen": 738306352,
+      "step": 4271
+    },
+    {
+      "epoch": 1.6361547299885102,
+      "grad_norm": 178.4165857708238,
+      "learning_rate": 5e-06,
+      "loss": 1.4123,
+      "num_input_tokens_seen": 738479312,
+      "step": 4272
+    },
+    {
+      "epoch": 1.6361547299885102,
+      "loss": 1.3048121929168701,
+      "loss_ce": 0.1136963814496994,
+      "loss_iou": 0.5423809885978699,
+      "loss_num": 0.021240234375,
+      "loss_xval": 1.367012858390808,
+      "num_input_tokens_seen": 738479312,
+      "step": 4272
+    },
+    {
+      "epoch": 1.6365377250095747,
+      "grad_norm": 231.64930660933527,
+      "learning_rate": 5e-06,
+      "loss": 1.3932,
+      "num_input_tokens_seen": 738652208,
+      "step": 4273
+    },
+    {
+      "epoch": 1.6365377250095747,
+      "loss": 1.337618112564087,
+      "loss_ce": 0.1696353554725647,
+      "loss_iou": 0.5430062413215637,
+      "loss_num": 0.016357421875,
+      "loss_xval": 1.1679826974868774,
+      "num_input_tokens_seen": 738652208,
+      "step": 4273
+    },
+    {
+      "epoch": 1.6369207200306395,
+      "grad_norm": 269.29606389009,
+      "learning_rate": 5e-06,
+      "loss": 2.1258,
+      "num_input_tokens_seen": 738825200,
+      "step": 4274
+    },
+    {
+      "epoch": 1.6369207200306395,
+      "loss": 2.2030158042907715,
+      "loss_ce": 0.16961421072483063,
+      "loss_iou": 0.9405593872070312,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 2.0334014892578125,
+      "num_input_tokens_seen": 738825200,
+      "step": 4274
+    },
+    {
+      "epoch": 1.6373037150517042,
+      "grad_norm": 148.0039806718881,
+      "learning_rate": 5e-06,
+      "loss": 1.665,
+      "num_input_tokens_seen": 738998392,
+      "step": 4275
+    },
+    {
+      "epoch": 1.6373037150517042,
+      "loss": 1.5252833366394043,
+      "loss_ce": 0.17528021335601807,
+      "loss_iou": 0.6302933692932129,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 1.3500032424926758,
+      "num_input_tokens_seen": 738998392,
+      "step": 4275
+    },
+    {
+      "epoch": 1.637686710072769,
+      "grad_norm": 173.16527791033351,
+      "learning_rate": 5e-06,
+      "loss": 1.2718,
+      "num_input_tokens_seen": 739171272,
+      "step": 4276
+    },
+    {
+      "epoch": 1.637686710072769,
+      "loss": 1.0502935647964478,
+      "loss_ce": 0.14169564843177795,
+      "loss_iou": 0.4223317503929138,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.9085978269577026,
+      "num_input_tokens_seen": 739171272,
+      "step": 4276
+    },
+    {
+      "epoch": 1.6380697050938338,
+      "grad_norm": 147.36560063091952,
+      "learning_rate": 5e-06,
+      "loss": 1.5918,
+      "num_input_tokens_seen": 739344272,
+      "step": 4277
+    },
+    {
+      "epoch": 1.6380697050938338,
+      "loss": 1.7093251943588257,
+      "loss_ce": 0.20406115055084229,
+      "loss_iou": 0.7129286527633667,
+      "loss_num": 0.015869140625,
+      "loss_xval": 1.5052640438079834,
+      "num_input_tokens_seen": 739344272,
+      "step": 4277
+    },
+    {
+      "epoch": 1.6384527001148985,
+      "grad_norm": 184.70601390030873,
+      "learning_rate": 5e-06,
+      "loss": 1.7006,
+      "num_input_tokens_seen": 739517024,
+      "step": 4278
+    },
+    {
+      "epoch": 1.6384527001148985,
+      "loss": 1.6746817827224731,
+      "loss_ce": 0.14700838923454285,
+      "loss_iou": 0.687878429889679,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 1.527673363685608,
+      "num_input_tokens_seen": 739517024,
+      "step": 4278
+    },
+    {
+      "epoch": 1.6388356951359633,
+      "grad_norm": 172.35186396439505,
+      "learning_rate": 5e-06,
+      "loss": 1.3699,
+      "num_input_tokens_seen": 739689888,
+      "step": 4279
+    },
+    {
+      "epoch": 1.6388356951359633,
+      "loss": 1.3808223009109497,
+      "loss_ce": 0.17082251608371735,
+      "loss_iou": 0.5690044164657593,
+      "loss_num": 0.014404296875,
+      "loss_xval": 1.2099997997283936,
+      "num_input_tokens_seen": 739689888,
+      "step": 4279
+    },
+    {
+      "epoch": 1.639218690157028,
+      "grad_norm": 291.7120410585724,
+      "learning_rate": 5e-06,
+      "loss": 1.8506,
+      "num_input_tokens_seen": 739862648,
+      "step": 4280
+    },
+    {
+      "epoch": 1.639218690157028,
+      "loss": 1.9401651620864868,
+      "loss_ce": 0.15000417828559875,
+      "loss_iou": 0.8300780653953552,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 1.7901610136032104,
+      "num_input_tokens_seen": 739862648,
+      "step": 4280
+    },
+    {
+      "epoch": 1.6396016851780928,
+      "grad_norm": 145.98182649385603,
+      "learning_rate": 5e-06,
+      "loss": 2.0004,
+      "num_input_tokens_seen": 740035488,
+      "step": 4281
+    },
+    {
+      "epoch": 1.6396016851780928,
+      "loss": 1.8754746913909912,
+      "loss_ce": 0.13285154104232788,
+      "loss_iou": 0.800449788570404,
+      "loss_num": 0.0283203125,
+      "loss_xval": 1.742623209953308,
+      "num_input_tokens_seen": 740035488,
+      "step": 4281
+    },
+    {
+      "epoch": 1.6399846801991576,
+      "grad_norm": 165.6245313970752,
+      "learning_rate": 5e-06,
+      "loss": 1.9451,
+      "num_input_tokens_seen": 740208512,
+      "step": 4282
+    },
+    {
+      "epoch": 1.6399846801991576,
+      "loss": 1.8908205032348633,
+      "loss_ce": 0.16656585037708282,
+      "loss_iou": 0.7740535736083984,
+      "loss_num": 0.03515625,
+      "loss_xval": 1.7242546081542969,
+      "num_input_tokens_seen": 740208512,
+      "step": 4282
+    },
+    {
+      "epoch": 1.640367675220222,
+      "grad_norm": 66.89123684629259,
+      "learning_rate": 5e-06,
+      "loss": 1.1958,
+      "num_input_tokens_seen": 740381368,
+      "step": 4283
+    },
+    {
+      "epoch": 1.640367675220222,
+      "loss": 1.2351512908935547,
+      "loss_ce": 0.19138486683368683,
+      "loss_iou": 0.4932425022125244,
+      "loss_num": 0.011474609375,
+      "loss_xval": 1.0437664985656738,
+      "num_input_tokens_seen": 740381368,
+      "step": 4283
+    },
+    {
+      "epoch": 1.6407506702412868,
+      "grad_norm": 117.65326730680832,
+      "learning_rate": 5e-06,
+      "loss": 1.5424,
+      "num_input_tokens_seen": 740554144,
+      "step": 4284
+    },
+    {
+      "epoch": 1.6407506702412868,
+      "loss": 1.3629440069198608,
+      "loss_ce": 0.13619455695152283,
+      "loss_iou": 0.5643940567970276,
+      "loss_num": 0.01953125,
+      "loss_xval": 1.2267495393753052,
+      "num_input_tokens_seen": 740554144,
+      "step": 4284
+    },
+    {
+      "epoch": 1.6411336652623516,
+      "grad_norm": 172.1231842524589,
+      "learning_rate": 5e-06,
+      "loss": 1.4857,
+      "num_input_tokens_seen": 740726976,
+      "step": 4285
+    },
+    {
+      "epoch": 1.6411336652623516,
+      "loss": 1.3251349925994873,
+      "loss_ce": 0.16517385840415955,
+      "loss_iou": 0.55124831199646,
+      "loss_num": 0.011474609375,
+      "loss_xval": 1.151782512664795,
+      "num_input_tokens_seen": 740726976,
+      "step": 4285
+    },
+    {
+      "epoch": 1.6415166602834164,
+      "grad_norm": 183.24079291365823,
+      "learning_rate": 5e-06,
+      "loss": 1.7292,
+      "num_input_tokens_seen": 740896352,
+      "step": 4286
+    },
+    {
+      "epoch": 1.6415166602834164,
+      "loss": 1.864220380783081,
+      "loss_ce": 0.13857772946357727,
+      "loss_iou": 0.7636086940765381,
+      "loss_num": 0.039794921875,
+      "loss_xval": 1.7256428003311157,
+      "num_input_tokens_seen": 740896352,
+      "step": 4286
+    },
+    {
+      "epoch": 1.641899655304481,
+      "grad_norm": 75.01338252661343,
+      "learning_rate": 5e-06,
+      "loss": 1.5956,
+      "num_input_tokens_seen": 741069256,
+      "step": 4287
+    },
+    {
+      "epoch": 1.641899655304481,
+      "loss": 1.4006035327911377,
+      "loss_ce": 0.1531718373298645,
+      "loss_iou": 0.5650457739830017,
+      "loss_num": 0.0234375,
+      "loss_xval": 1.2474316358566284,
+      "num_input_tokens_seen": 741069256,
+      "step": 4287
+    },
+    {
+      "epoch": 1.6422826503255457,
+      "grad_norm": 113.8831730921573,
+      "learning_rate": 5e-06,
+      "loss": 1.4835,
+      "num_input_tokens_seen": 741242472,
+      "step": 4288
+    },
+    {
+      "epoch": 1.6422826503255457,
+      "loss": 1.2966430187225342,
+      "loss_ce": 0.17165182530879974,
+      "loss_iou": 0.5388215780258179,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 1.1249911785125732,
+      "num_input_tokens_seen": 741242472,
+      "step": 4288
+    },
+    {
+      "epoch": 1.6426656453466104,
+      "grad_norm": 317.2341601707341,
+      "learning_rate": 5e-06,
+      "loss": 1.3591,
+      "num_input_tokens_seen": 741415208,
+      "step": 4289
+    },
+    {
+      "epoch": 1.6426656453466104,
+      "loss": 1.3367316722869873,
+      "loss_ce": 0.14232824742794037,
+      "loss_iou": 0.5580628514289856,
+      "loss_num": 0.015625,
+      "loss_xval": 1.1944032907485962,
+      "num_input_tokens_seen": 741415208,
+      "step": 4289
+    },
+    {
+      "epoch": 1.6430486403676752,
+      "grad_norm": 131.1750460487662,
+      "learning_rate": 5e-06,
+      "loss": 1.6847,
+      "num_input_tokens_seen": 741588256,
+      "step": 4290
+    },
+    {
+      "epoch": 1.6430486403676752,
+      "loss": 1.8473162651062012,
+      "loss_ce": 0.15907804667949677,
+      "loss_iou": 0.7536649703979492,
+      "loss_num": 0.0361328125,
+      "loss_xval": 1.6882381439208984,
+      "num_input_tokens_seen": 741588256,
+      "step": 4290
+    },
+    {
+      "epoch": 1.64343163538874,
+      "grad_norm": 175.40988601634265,
+      "learning_rate": 5e-06,
+      "loss": 1.6343,
+      "num_input_tokens_seen": 741761416,
+      "step": 4291
+    },
+    {
+      "epoch": 1.64343163538874,
+      "loss": 1.556229591369629,
+      "loss_ce": 0.19933602213859558,
+      "loss_iou": 0.6340437531471252,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 1.3568936586380005,
+      "num_input_tokens_seen": 741761416,
+      "step": 4291
+    },
+    {
+      "epoch": 1.6438146304098047,
+      "grad_norm": 188.56269865478197,
+      "learning_rate": 5e-06,
+      "loss": 1.5454,
+      "num_input_tokens_seen": 741934456,
+      "step": 4292
+    },
+    {
+      "epoch": 1.6438146304098047,
+      "loss": 1.4899033308029175,
+      "loss_ce": 0.20824198424816132,
+      "loss_iou": 0.6150586009025574,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 1.2816613912582397,
+      "num_input_tokens_seen": 741934456,
+      "step": 4292
+    },
+    {
+      "epoch": 1.6441976254308694,
+      "grad_norm": 293.7354363788219,
+      "learning_rate": 5e-06,
+      "loss": 1.9317,
+      "num_input_tokens_seen": 742107264,
+      "step": 4293
+    },
+    {
+      "epoch": 1.6441976254308694,
+      "loss": 1.8700616359710693,
+      "loss_ce": 0.1377236396074295,
+      "loss_iou": 0.8095741271972656,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 1.7323379516601562,
+      "num_input_tokens_seen": 742107264,
+      "step": 4293
+    },
+    {
+      "epoch": 1.6445806204519342,
+      "grad_norm": 155.56475849248338,
+      "learning_rate": 5e-06,
+      "loss": 1.9621,
+      "num_input_tokens_seen": 742280000,
+      "step": 4294
+    },
+    {
+      "epoch": 1.6445806204519342,
+      "loss": 1.9358739852905273,
+      "loss_ce": 0.16428327560424805,
+      "loss_iou": 0.8030315637588501,
+      "loss_num": 0.033203125,
+      "loss_xval": 1.7715904712677002,
+      "num_input_tokens_seen": 742280000,
+      "step": 4294
+    },
+    {
+      "epoch": 1.644963615472999,
+      "grad_norm": 44.49111387657106,
+      "learning_rate": 5e-06,
+      "loss": 1.7541,
+      "num_input_tokens_seen": 742452792,
+      "step": 4295
+    },
+    {
+      "epoch": 1.644963615472999,
+      "loss": 1.712242603302002,
+      "loss_ce": 0.14141297340393066,
+      "loss_iou": -20753281974272.0,
+      "loss_num": 0.02734375,
+      "loss_xval": 1.790556472822374e+31,
+      "num_input_tokens_seen": 742452792,
+      "step": 4295
+    },
+    {
+      "epoch": 1.6453466104940637,
+      "grad_norm": 140.2333146591295,
+      "learning_rate": 5e-06,
+      "loss": 1.5158,
+      "num_input_tokens_seen": 742625760,
+      "step": 4296
+    },
+    {
+      "epoch": 1.6453466104940637,
+      "loss": 1.5926063060760498,
+      "loss_ce": 0.177087664604187,
+      "loss_iou": 0.6723132133483887,
+      "loss_num": 0.01416015625,
+      "loss_xval": 1.4155187606811523,
+      "num_input_tokens_seen": 742625760,
+      "step": 4296
+    },
+    {
+      "epoch": 1.6457296055151283,
+      "grad_norm": 172.70051532883474,
+      "learning_rate": 5e-06,
+      "loss": 2.3346,
+      "num_input_tokens_seen": 742798680,
+      "step": 4297
+    },
+    {
+      "epoch": 1.6457296055151283,
+      "loss": 2.3730525970458984,
+      "loss_ce": 0.20516502857208252,
+      "loss_iou": 1.0251668691635132,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 2.1678876876831055,
+      "num_input_tokens_seen": 742798680,
+      "step": 4297
+    },
+    {
+      "epoch": 1.646112600536193,
+      "grad_norm": 96.58527355825707,
+      "learning_rate": 5e-06,
+      "loss": 1.7464,
+      "num_input_tokens_seen": 742972056,
+      "step": 4298
+    },
+    {
+      "epoch": 1.646112600536193,
+      "loss": 1.7323307991027832,
+      "loss_ce": 0.16866838932037354,
+      "loss_iou": 0.7081312537193298,
+      "loss_num": 0.029541015625,
+      "loss_xval": 1.5636624097824097,
+      "num_input_tokens_seen": 742972056,
+      "step": 4298
+    },
+    {
+      "epoch": 1.6464955955572578,
+      "grad_norm": 76.79804901297256,
+      "learning_rate": 5e-06,
+      "loss": 1.3906,
+      "num_input_tokens_seen": 743144872,
+      "step": 4299
+    },
+    {
+      "epoch": 1.6464955955572578,
+      "loss": 1.330512523651123,
+      "loss_ce": 0.2305942177772522,
+      "loss_iou": 0.514253556728363,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 1.099918246269226,
+      "num_input_tokens_seen": 743144872,
+      "step": 4299
+    },
+    {
+      "epoch": 1.6468785905783225,
+      "grad_norm": 122.44180625472737,
+      "learning_rate": 5e-06,
+      "loss": 1.7444,
+      "num_input_tokens_seen": 743318064,
+      "step": 4300
+    },
+    {
+      "epoch": 1.6468785905783225,
+      "loss": 1.6140046119689941,
+      "loss_ce": 0.1705130934715271,
+      "loss_iou": 0.6736500859260559,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.4434915781021118,
+      "num_input_tokens_seen": 743318064,
+      "step": 4300
+    },
+    {
+      "epoch": 1.647261585599387,
+      "grad_norm": 84.41639006897243,
+      "learning_rate": 5e-06,
+      "loss": 1.7019,
+      "num_input_tokens_seen": 743491320,
+      "step": 4301
+    },
+    {
+      "epoch": 1.647261585599387,
+      "loss": 1.6535019874572754,
+      "loss_ce": 0.19724097847938538,
+      "loss_iou": 0.6837274432182312,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 1.4562610387802124,
+      "num_input_tokens_seen": 743491320,
+      "step": 4301
+    },
+    {
+      "epoch": 1.6476445806204518,
+      "grad_norm": 85.11099246075345,
+      "learning_rate": 5e-06,
+      "loss": 1.2476,
+      "num_input_tokens_seen": 743664160,
+      "step": 4302
+    },
+    {
+      "epoch": 1.6476445806204518,
+      "loss": 1.3284763097763062,
+      "loss_ce": 0.16261108219623566,
+      "loss_iou": 0.5181132555007935,
+      "loss_num": 0.02587890625,
+      "loss_xval": 1.165865182876587,
+      "num_input_tokens_seen": 743664160,
+      "step": 4302
+    },
+    {
+      "epoch": 1.6480275756415166,
+      "grad_norm": 100.86795750806613,
+      "learning_rate": 5e-06,
+      "loss": 1.5445,
+      "num_input_tokens_seen": 743837032,
+      "step": 4303
+    },
+    {
+      "epoch": 1.6480275756415166,
+      "loss": 1.605276107788086,
+      "loss_ce": 0.145803302526474,
+      "loss_iou": 0.6753846406936646,
+      "loss_num": 0.021728515625,
+      "loss_xval": 1.459472894668579,
+      "num_input_tokens_seen": 743837032,
+      "step": 4303
+    },
+    {
+      "epoch": 1.6484105706625813,
+      "grad_norm": 124.00043972417424,
+      "learning_rate": 5e-06,
+      "loss": 1.6567,
+      "num_input_tokens_seen": 744009816,
+      "step": 4304
+    },
+    {
+      "epoch": 1.6484105706625813,
+      "loss": 1.62785804271698,
+      "loss_ce": 0.16736261546611786,
+      "loss_iou": 0.6700212359428406,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 1.4604953527450562,
+      "num_input_tokens_seen": 744009816,
+      "step": 4304
+    },
+    {
+      "epoch": 1.648793565683646,
+      "grad_norm": 266.6892693451771,
+      "learning_rate": 5e-06,
+      "loss": 1.5544,
+      "num_input_tokens_seen": 744182792,
+      "step": 4305
+    },
+    {
+      "epoch": 1.648793565683646,
+      "loss": 1.5796635150909424,
+      "loss_ce": 0.12979306280612946,
+      "loss_iou": 0.6984153985977173,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 1.4498703479766846,
+      "num_input_tokens_seen": 744182792,
+      "step": 4305
+    },
+    {
+      "epoch": 1.6491765607047109,
+      "grad_norm": 120.91812762698265,
+      "learning_rate": 5e-06,
+      "loss": 1.5587,
+      "num_input_tokens_seen": 744355864,
+      "step": 4306
+    },
+    {
+      "epoch": 1.6491765607047109,
+      "loss": 1.5446631908416748,
+      "loss_ce": 0.17455412447452545,
+      "loss_iou": 0.6263539791107178,
+      "loss_num": 0.0234375,
+      "loss_xval": 1.3701090812683105,
+      "num_input_tokens_seen": 744355864,
+      "step": 4306
+    },
+    {
+      "epoch": 1.6495595557257756,
+      "grad_norm": 99.2804550777207,
+      "learning_rate": 5e-06,
+      "loss": 1.1902,
+      "num_input_tokens_seen": 744528744,
+      "step": 4307
+    },
+    {
+      "epoch": 1.6495595557257756,
+      "loss": 1.2010624408721924,
+      "loss_ce": 0.19453051686286926,
+      "loss_iou": 0.46843019127845764,
+      "loss_num": 0.013916015625,
+      "loss_xval": 1.0065319538116455,
+      "num_input_tokens_seen": 744528744,
+      "step": 4307
+    },
+    {
+      "epoch": 1.6499425507468404,
+      "grad_norm": 139.7206590368086,
+      "learning_rate": 5e-06,
+      "loss": 1.4275,
+      "num_input_tokens_seen": 744701952,
+      "step": 4308
+    },
+    {
+      "epoch": 1.6499425507468404,
+      "loss": 1.4538816213607788,
+      "loss_ce": 0.14945608377456665,
+      "loss_iou": 0.5948854684829712,
+      "loss_num": 0.02294921875,
+      "loss_xval": 1.3044254779815674,
+      "num_input_tokens_seen": 744701952,
+      "step": 4308
+    },
+    {
+      "epoch": 1.6503255457679051,
+      "grad_norm": 125.34830773908683,
+      "learning_rate": 5e-06,
+      "loss": 1.3716,
+      "num_input_tokens_seen": 744874792,
+      "step": 4309
+    },
+    {
+      "epoch": 1.6503255457679051,
+      "loss": 1.357515811920166,
+      "loss_ce": 0.1182800680398941,
+      "loss_iou": 0.589146077632904,
+      "loss_num": 0.01220703125,
+      "loss_xval": 1.239235758781433,
+      "num_input_tokens_seen": 744874792,
+      "step": 4309
+    },
+    {
+      "epoch": 1.65070854078897,
+      "grad_norm": 134.44926680400766,
+      "learning_rate": 5e-06,
+      "loss": 1.4559,
+      "num_input_tokens_seen": 745047936,
+      "step": 4310
+    },
+    {
+      "epoch": 1.65070854078897,
+      "loss": 1.4882766008377075,
+      "loss_ce": 0.1506461203098297,
+      "loss_iou": 0.6241985559463501,
+      "loss_num": 0.017822265625,
+      "loss_xval": 1.3376305103302002,
+      "num_input_tokens_seen": 745047936,
+      "step": 4310
+    },
+    {
+      "epoch": 1.6510915358100344,
+      "grad_norm": 180.53997717268547,
+      "learning_rate": 5e-06,
+      "loss": 1.3624,
+      "num_input_tokens_seen": 745220864,
+      "step": 4311
+    },
+    {
+      "epoch": 1.6510915358100344,
+      "loss": 1.1268970966339111,
+      "loss_ce": 0.15746453404426575,
+      "loss_iou": 0.4212244749069214,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.9694325923919678,
+      "num_input_tokens_seen": 745220864,
+      "step": 4311
+    },
+    {
+      "epoch": 1.6514745308310992,
+      "grad_norm": 235.10424515817198,
+      "learning_rate": 5e-06,
+      "loss": 1.6102,
+      "num_input_tokens_seen": 745393592,
+      "step": 4312
+    },
+    {
+      "epoch": 1.6514745308310992,
+      "loss": 1.5685279369354248,
+      "loss_ce": 0.14919352531433105,
+      "loss_iou": 0.6671409606933594,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 1.4193344116210938,
+      "num_input_tokens_seen": 745393592,
+      "step": 4312
+    },
+    {
+      "epoch": 1.651857525852164,
+      "grad_norm": 169.14939483295666,
+      "learning_rate": 5e-06,
+      "loss": 1.5582,
+      "num_input_tokens_seen": 745566184,
+      "step": 4313
+    },
+    {
+      "epoch": 1.651857525852164,
+      "loss": 1.528897762298584,
+      "loss_ce": 0.15899689495563507,
+      "loss_iou": 0.6291643381118774,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 1.3699009418487549,
+      "num_input_tokens_seen": 745566184,
+      "step": 4313
+    },
+    {
+      "epoch": 1.6522405208732287,
+      "grad_norm": 101.44406082473364,
+      "learning_rate": 5e-06,
+      "loss": 1.4022,
+      "num_input_tokens_seen": 745739000,
+      "step": 4314
+    },
+    {
+      "epoch": 1.6522405208732287,
+      "loss": 1.4315533638000488,
+      "loss_ce": 0.17240798473358154,
+      "loss_iou": 0.5926769375801086,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 1.2591453790664673,
+      "num_input_tokens_seen": 745739000,
+      "step": 4314
+    },
+    {
+      "epoch": 1.6526235158942932,
+      "grad_norm": 153.54532476078182,
+      "learning_rate": 5e-06,
+      "loss": 1.4521,
+      "num_input_tokens_seen": 745912176,
+      "step": 4315
+    },
+    {
+      "epoch": 1.6526235158942932,
+      "loss": 1.4062680006027222,
+      "loss_ce": 0.19291892647743225,
+      "loss_iou": 0.5603793859481812,
+      "loss_num": 0.0185546875,
+      "loss_xval": 1.2133491039276123,
+      "num_input_tokens_seen": 745912176,
+      "step": 4315
+    },
+    {
+      "epoch": 1.653006510915358,
+      "grad_norm": 126.15423208683085,
+      "learning_rate": 5e-06,
+      "loss": 1.3409,
+      "num_input_tokens_seen": 746085200,
+      "step": 4316
+    },
+    {
+      "epoch": 1.653006510915358,
+      "loss": 1.410337209701538,
+      "loss_ce": 0.2024148404598236,
+      "loss_iou": 0.5630066394805908,
+      "loss_num": 0.016357421875,
+      "loss_xval": 1.2079224586486816,
+      "num_input_tokens_seen": 746085200,
+      "step": 4316
+    },
+    {
+      "epoch": 1.6533895059364228,
+      "grad_norm": 113.4981658484504,
+      "learning_rate": 5e-06,
+      "loss": 1.4783,
+      "num_input_tokens_seen": 746258392,
+      "step": 4317
+    },
+    {
+      "epoch": 1.6533895059364228,
+      "loss": 1.333397626876831,
+      "loss_ce": 0.17618131637573242,
+      "loss_iou": 0.5437875986099243,
+      "loss_num": 0.013916015625,
+      "loss_xval": 1.1572163105010986,
+      "num_input_tokens_seen": 746258392,
+      "step": 4317
+    },
+    {
+      "epoch": 1.6537725009574875,
+      "grad_norm": 106.16823314010469,
+      "learning_rate": 5e-06,
+      "loss": 1.5228,
+      "num_input_tokens_seen": 746431328,
+      "step": 4318
+    },
+    {
+      "epoch": 1.6537725009574875,
+      "loss": 1.705797791481018,
+      "loss_ce": 0.16248458623886108,
+      "loss_iou": 0.6962934732437134,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 1.5433132648468018,
+      "num_input_tokens_seen": 746431328,
+      "step": 4318
+    },
+    {
+      "epoch": 1.6541554959785523,
+      "grad_norm": 64.50687135872865,
+      "learning_rate": 5e-06,
+      "loss": 1.3769,
+      "num_input_tokens_seen": 746604360,
+      "step": 4319
+    },
+    {
+      "epoch": 1.6541554959785523,
+      "loss": 1.2541346549987793,
+      "loss_ce": 0.14104785025119781,
+      "loss_iou": 0.5109654068946838,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 1.1130868196487427,
+      "num_input_tokens_seen": 746604360,
+      "step": 4319
+    },
+    {
+      "epoch": 1.654538490999617,
+      "grad_norm": 120.06465185779766,
+      "learning_rate": 5e-06,
+      "loss": 1.4102,
+      "num_input_tokens_seen": 746777384,
+      "step": 4320
+    },
+    {
+      "epoch": 1.654538490999617,
+      "loss": 1.188852071762085,
+      "loss_ce": 0.12857110798358917,
+      "loss_iou": 0.4838911294937134,
+      "loss_num": 0.0185546875,
+      "loss_xval": 1.0471584796905518,
+      "num_input_tokens_seen": 746777384,
+      "step": 4320
+    },
+    {
+      "epoch": 1.6549214860206818,
+      "grad_norm": 99.73864319221155,
+      "learning_rate": 5e-06,
+      "loss": 1.2341,
+      "num_input_tokens_seen": 746950424,
+      "step": 4321
+    },
+    {
+      "epoch": 1.6549214860206818,
+      "loss": 1.3568239212036133,
+      "loss_ce": 0.1628800332546234,
+      "loss_iou": 0.5363030433654785,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 1.193943977355957,
+      "num_input_tokens_seen": 746950424,
+      "step": 4321
+    },
+    {
+      "epoch": 1.6553044810417465,
+      "grad_norm": 113.5453243126911,
+      "learning_rate": 5e-06,
+      "loss": 1.6964,
+      "num_input_tokens_seen": 747123408,
+      "step": 4322
+    },
+    {
+      "epoch": 1.6553044810417465,
+      "loss": 1.706247329711914,
+      "loss_ce": 0.18428021669387817,
+      "loss_iou": 0.6889925003051758,
+      "loss_num": 0.02880859375,
+      "loss_xval": 1.5219669342041016,
+      "num_input_tokens_seen": 747123408,
+      "step": 4322
+    },
+    {
+      "epoch": 1.6556874760628113,
+      "grad_norm": 128.88972982006158,
+      "learning_rate": 5e-06,
+      "loss": 1.4075,
+      "num_input_tokens_seen": 747296576,
+      "step": 4323
+    },
+    {
+      "epoch": 1.6556874760628113,
+      "loss": 1.39468252658844,
+      "loss_ce": 0.16338348388671875,
+      "loss_iou": 0.5695374608039856,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 1.2312990427017212,
+      "num_input_tokens_seen": 747296576,
+      "step": 4323
+    },
+    {
+      "epoch": 1.656070471083876,
+      "grad_norm": 162.94147982802778,
+      "learning_rate": 5e-06,
+      "loss": 1.7359,
+      "num_input_tokens_seen": 747469320,
+      "step": 4324
+    },
+    {
+      "epoch": 1.656070471083876,
+      "loss": 2.078629732131958,
+      "loss_ce": 0.17305155098438263,
+      "loss_iou": 0.8767850399017334,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 1.9055781364440918,
+      "num_input_tokens_seen": 747469320,
+      "step": 4324
+    },
+    {
+      "epoch": 1.6564534661049406,
+      "grad_norm": 124.06849237730506,
+      "learning_rate": 5e-06,
+      "loss": 1.3565,
+      "num_input_tokens_seen": 747642304,
+      "step": 4325
+    },
+    {
+      "epoch": 1.6564534661049406,
+      "loss": 1.3706270456314087,
+      "loss_ce": 0.1738179475069046,
+      "loss_iou": 0.5613104104995728,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 1.1968090534210205,
+      "num_input_tokens_seen": 747642304,
+      "step": 4325
+    },
+    {
+      "epoch": 1.6568364611260054,
+      "grad_norm": 195.91109856325355,
+      "learning_rate": 5e-06,
+      "loss": 1.6801,
+      "num_input_tokens_seen": 747814768,
+      "step": 4326
+    },
+    {
+      "epoch": 1.6568364611260054,
+      "loss": 1.7964779138565063,
+      "loss_ce": 0.1447073221206665,
+      "loss_iou": 0.7687869071960449,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 1.6517705917358398,
+      "num_input_tokens_seen": 747814768,
+      "step": 4326
+    },
+    {
+      "epoch": 1.6572194561470701,
+      "grad_norm": 121.30326584270553,
+      "learning_rate": 5e-06,
+      "loss": 1.4,
+      "num_input_tokens_seen": 747987728,
+      "step": 4327
+    },
+    {
+      "epoch": 1.6572194561470701,
+      "loss": 1.5141677856445312,
+      "loss_ce": 0.13041839003562927,
+      "loss_iou": 0.62269127368927,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 1.38374924659729,
+      "num_input_tokens_seen": 747987728,
+      "step": 4327
+    },
+    {
+      "epoch": 1.6576024511681346,
+      "grad_norm": 190.22677907897673,
+      "learning_rate": 5e-06,
+      "loss": 1.3527,
+      "num_input_tokens_seen": 748160808,
+      "step": 4328
+    },
+    {
+      "epoch": 1.6576024511681346,
+      "loss": 1.2638424634933472,
+      "loss_ce": 0.17623110115528107,
+      "loss_iou": 0.517835259437561,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 1.087611436843872,
+      "num_input_tokens_seen": 748160808,
+      "step": 4328
+    },
+    {
+      "epoch": 1.6579854461891994,
+      "grad_norm": 171.87342369575066,
+      "learning_rate": 5e-06,
+      "loss": 1.4965,
+      "num_input_tokens_seen": 748333496,
+      "step": 4329
+    },
+    {
+      "epoch": 1.6579854461891994,
+      "loss": 1.554666519165039,
+      "loss_ce": 0.1455255150794983,
+      "loss_iou": 0.638286292552948,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 1.4091410636901855,
+      "num_input_tokens_seen": 748333496,
+      "step": 4329
+    },
+    {
+      "epoch": 1.6583684412102642,
+      "grad_norm": 109.43188793076496,
+      "learning_rate": 5e-06,
+      "loss": 1.4313,
+      "num_input_tokens_seen": 748506256,
+      "step": 4330
+    },
+    {
+      "epoch": 1.6583684412102642,
+      "loss": 1.4558132886886597,
+      "loss_ce": 0.18514424562454224,
+      "loss_iou": 0.5906720757484436,
+      "loss_num": 0.017822265625,
+      "loss_xval": 1.2706691026687622,
+      "num_input_tokens_seen": 748506256,
+      "step": 4330
+    },
+    {
+      "epoch": 1.658751436231329,
+      "grad_norm": 108.72071196031955,
+      "learning_rate": 5e-06,
+      "loss": 1.4831,
+      "num_input_tokens_seen": 748679320,
+      "step": 4331
+    },
+    {
+      "epoch": 1.658751436231329,
+      "loss": 1.451843500137329,
+      "loss_ce": 0.20245873928070068,
+      "loss_iou": 0.5653510093688965,
+      "loss_num": 0.023681640625,
+      "loss_xval": 1.249384880065918,
+      "num_input_tokens_seen": 748679320,
+      "step": 4331
+    },
+    {
+      "epoch": 1.6591344312523937,
+      "grad_norm": 107.55488067482345,
+      "learning_rate": 5e-06,
+      "loss": 1.3853,
+      "num_input_tokens_seen": 748852344,
+      "step": 4332
+    },
+    {
+      "epoch": 1.6591344312523937,
+      "loss": 1.4387247562408447,
+      "loss_ce": 0.18469186127185822,
+      "loss_iou": 0.5855430364608765,
+      "loss_num": 0.0166015625,
+      "loss_xval": 1.254032850265503,
+      "num_input_tokens_seen": 748852344,
+      "step": 4332
+    },
+    {
+      "epoch": 1.6595174262734584,
+      "grad_norm": 138.97777822884672,
+      "learning_rate": 5e-06,
+      "loss": 1.5506,
+      "num_input_tokens_seen": 749025504,
+      "step": 4333
+    },
+    {
+      "epoch": 1.6595174262734584,
+      "loss": 1.5302248001098633,
+      "loss_ce": 0.13889721035957336,
+      "loss_iou": 0.6157077550888062,
+      "loss_num": 0.031982421875,
+      "loss_xval": 1.3913276195526123,
+      "num_input_tokens_seen": 749025504,
+      "step": 4333
+    },
+    {
+      "epoch": 1.6599004212945232,
+      "grad_norm": 127.38663465582381,
+      "learning_rate": 5e-06,
+      "loss": 1.4353,
+      "num_input_tokens_seen": 749198056,
+      "step": 4334
+    },
+    {
+      "epoch": 1.6599004212945232,
+      "loss": 1.3108580112457275,
+      "loss_ce": 0.19315487146377563,
+      "loss_iou": 0.5166457891464233,
+      "loss_num": 0.016845703125,
+      "loss_xval": 1.1177031993865967,
+      "num_input_tokens_seen": 749198056,
+      "step": 4334
+    },
+    {
+      "epoch": 1.660283416315588,
+      "grad_norm": 121.81982195951672,
+      "learning_rate": 5e-06,
+      "loss": 1.6543,
+      "num_input_tokens_seen": 749371064,
+      "step": 4335
+    },
+    {
+      "epoch": 1.660283416315588,
+      "loss": 1.6364576816558838,
+      "loss_ce": 0.16774462163448334,
+      "loss_iou": 0.6846739053726196,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 1.4687130451202393,
+      "num_input_tokens_seen": 749371064,
+      "step": 4335
+    },
+    {
+      "epoch": 1.6606664113366527,
+      "grad_norm": 89.15176029224519,
+      "learning_rate": 5e-06,
+      "loss": 1.4432,
+      "num_input_tokens_seen": 749544112,
+      "step": 4336
+    },
+    {
+      "epoch": 1.6606664113366527,
+      "loss": 1.460259199142456,
+      "loss_ce": 0.1678803563117981,
+      "loss_iou": 0.6032969951629639,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 1.2923789024353027,
+      "num_input_tokens_seen": 749544112,
+      "step": 4336
+    },
+    {
+      "epoch": 1.6610494063577175,
+      "grad_norm": 148.71456479347296,
+      "learning_rate": 5e-06,
+      "loss": 1.0917,
+      "num_input_tokens_seen": 749716984,
+      "step": 4337
+    },
+    {
+      "epoch": 1.6610494063577175,
+      "loss": 1.0275170803070068,
+      "loss_ce": 0.11111672967672348,
+      "loss_iou": 0.4249817728996277,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.9164003133773804,
+      "num_input_tokens_seen": 749716984,
+      "step": 4337
+    },
+    {
+      "epoch": 1.6614324013787822,
+      "grad_norm": 115.50154730509183,
+      "learning_rate": 5e-06,
+      "loss": 1.7234,
+      "num_input_tokens_seen": 749890016,
+      "step": 4338
+    },
+    {
+      "epoch": 1.6614324013787822,
+      "loss": 1.7582378387451172,
+      "loss_ce": 0.17511804401874542,
+      "loss_iou": 0.7226817607879639,
+      "loss_num": 0.027587890625,
+      "loss_xval": 1.5831198692321777,
+      "num_input_tokens_seen": 749890016,
+      "step": 4338
+    },
+    {
+      "epoch": 1.6618153963998468,
+      "grad_norm": 109.33496661169397,
+      "learning_rate": 5e-06,
+      "loss": 1.3089,
+      "num_input_tokens_seen": 750063056,
+      "step": 4339
+    },
+    {
+      "epoch": 1.6618153963998468,
+      "loss": 1.4102544784545898,
+      "loss_ce": 0.15036270022392273,
+      "loss_iou": 0.5823689699172974,
+      "loss_num": 0.01904296875,
+      "loss_xval": 1.2598917484283447,
+      "num_input_tokens_seen": 750063056,
+      "step": 4339
+    },
+    {
+      "epoch": 1.6621983914209115,
+      "grad_norm": 98.19377264488183,
+      "learning_rate": 5e-06,
+      "loss": 1.3632,
+      "num_input_tokens_seen": 750235864,
+      "step": 4340
+    },
+    {
+      "epoch": 1.6621983914209115,
+      "loss": 1.4982824325561523,
+      "loss_ce": 0.19171017408370972,
+      "loss_iou": 0.5953027009963989,
+      "loss_num": 0.023193359375,
+      "loss_xval": 1.3065721988677979,
+      "num_input_tokens_seen": 750235864,
+      "step": 4340
+    },
+    {
+      "epoch": 1.6625813864419763,
+      "grad_norm": 159.9248649242272,
+      "learning_rate": 5e-06,
+      "loss": 1.2652,
+      "num_input_tokens_seen": 750408744,
+      "step": 4341
+    },
+    {
+      "epoch": 1.6625813864419763,
+      "loss": 1.2764732837677002,
+      "loss_ce": 0.1473279893398285,
+      "loss_iou": 0.5065434575080872,
+      "loss_num": 0.023193359375,
+      "loss_xval": 1.1291451454162598,
+      "num_input_tokens_seen": 750408744,
+      "step": 4341
+    },
+    {
+      "epoch": 1.6629643814630408,
+      "grad_norm": 191.00419749506867,
+      "learning_rate": 5e-06,
+      "loss": 1.5639,
+      "num_input_tokens_seen": 750581496,
+      "step": 4342
+    },
+    {
+      "epoch": 1.6629643814630408,
+      "loss": 1.575960636138916,
+      "loss_ce": 0.20127132534980774,
+      "loss_iou": 0.6380282640457153,
+      "loss_num": 0.019775390625,
+      "loss_xval": 1.3746893405914307,
+      "num_input_tokens_seen": 750581496,
+      "step": 4342
+    },
+    {
+      "epoch": 1.6633473764841056,
+      "grad_norm": 137.39907596519382,
+      "learning_rate": 5e-06,
+      "loss": 1.4354,
+      "num_input_tokens_seen": 750754536,
+      "step": 4343
+    },
+    {
+      "epoch": 1.6633473764841056,
+      "loss": 1.6266783475875854,
+      "loss_ce": 0.17832821607589722,
+      "loss_iou": 0.672340989112854,
+      "loss_num": 0.020751953125,
+      "loss_xval": 1.448350191116333,
+      "num_input_tokens_seen": 750754536,
+      "step": 4343
+    },
+    {
+      "epoch": 1.6637303715051703,
+      "grad_norm": 109.33498349661797,
+      "learning_rate": 5e-06,
+      "loss": 1.6813,
+      "num_input_tokens_seen": 750927400,
+      "step": 4344
+    },
+    {
+      "epoch": 1.6637303715051703,
+      "loss": 1.6764832735061646,
+      "loss_ce": 0.16076403856277466,
+      "loss_iou": 0.7087873220443726,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 1.5157191753387451,
+      "num_input_tokens_seen": 750927400,
+      "step": 4344
+    },
+    {
+      "epoch": 1.664113366526235,
+      "grad_norm": 132.21153686371912,
+      "learning_rate": 5e-06,
+      "loss": 1.4743,
+      "num_input_tokens_seen": 751100208,
+      "step": 4345
+    },
+    {
+      "epoch": 1.664113366526235,
+      "loss": 1.6191737651824951,
+      "loss_ce": 0.1825704425573349,
+      "loss_iou": 0.651269793510437,
+      "loss_num": 0.02685546875,
+      "loss_xval": 1.436603307723999,
+      "num_input_tokens_seen": 751100208,
+      "step": 4345
+    },
+    {
+      "epoch": 1.6644963615472999,
+      "grad_norm": 143.87108557151714,
+      "learning_rate": 5e-06,
+      "loss": 1.5951,
+      "num_input_tokens_seen": 751267976,
+      "step": 4346
+    },
+    {
+      "epoch": 1.6644963615472999,
+      "loss": 1.6740915775299072,
+      "loss_ce": 0.179146409034729,
+      "loss_iou": 0.6795557141304016,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 1.4949451684951782,
+      "num_input_tokens_seen": 751267976,
+      "step": 4346
+    },
+    {
+      "epoch": 1.6648793565683646,
+      "grad_norm": 106.6810436195744,
+      "learning_rate": 5e-06,
+      "loss": 1.5367,
+      "num_input_tokens_seen": 751440792,
+      "step": 4347
+    },
+    {
+      "epoch": 1.6648793565683646,
+      "loss": 1.4976184368133545,
+      "loss_ce": 0.13274140655994415,
+      "loss_iou": 0.6166731119155884,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 1.3648769855499268,
+      "num_input_tokens_seen": 751440792,
+      "step": 4347
+    },
+    {
+      "epoch": 1.6652623515894294,
+      "grad_norm": 103.23602112630653,
+      "learning_rate": 5e-06,
+      "loss": 1.398,
+      "num_input_tokens_seen": 751613856,
+      "step": 4348
+    },
+    {
+      "epoch": 1.6652623515894294,
+      "loss": 1.3447130918502808,
+      "loss_ce": 0.1880890130996704,
+      "loss_iou": 0.5271035432815552,
+      "loss_num": 0.0205078125,
+      "loss_xval": 1.1566240787506104,
+      "num_input_tokens_seen": 751613856,
+      "step": 4348
+    },
+    {
+      "epoch": 1.6656453466104941,
+      "grad_norm": 155.02316345437134,
+      "learning_rate": 5e-06,
+      "loss": 1.3166,
+      "num_input_tokens_seen": 751786368,
+      "step": 4349
+    },
+    {
+      "epoch": 1.6656453466104941,
+      "loss": 1.3376691341400146,
+      "loss_ce": 0.14817002415657043,
+      "loss_iou": 0.5526505708694458,
+      "loss_num": 0.016845703125,
+      "loss_xval": 1.1894991397857666,
+      "num_input_tokens_seen": 751786368,
+      "step": 4349
+    },
+    {
+      "epoch": 1.6660283416315589,
+      "grad_norm": 138.28528744846454,
+      "learning_rate": 5e-06,
+      "loss": 1.6043,
+      "num_input_tokens_seen": 751959088,
+      "step": 4350
+    },
+    {
+      "epoch": 1.6660283416315589,
+      "loss": 1.6053296327590942,
+      "loss_ce": 0.13853321969509125,
+      "loss_iou": 0.6715390682220459,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 1.4667963981628418,
+      "num_input_tokens_seen": 751959088,
+      "step": 4350
+    },
+    {
+      "epoch": 1.6664113366526236,
+      "grad_norm": 241.1625900477056,
+      "learning_rate": 5e-06,
+      "loss": 1.5604,
+      "num_input_tokens_seen": 752132040,
+      "step": 4351
+    },
+    {
+      "epoch": 1.6664113366526236,
+      "loss": 1.562677264213562,
+      "loss_ce": 0.16798928380012512,
+      "loss_iou": 0.6683523654937744,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 1.3946881294250488,
+      "num_input_tokens_seen": 752132040,
+      "step": 4351
+    },
+    {
+      "epoch": 1.6667943316736884,
+      "grad_norm": 127.74427420012312,
+      "learning_rate": 5e-06,
+      "loss": 1.8056,
+      "num_input_tokens_seen": 752305016,
+      "step": 4352
+    },
+    {
+      "epoch": 1.6667943316736884,
+      "loss": 1.7611594200134277,
+      "loss_ce": 0.1871078461408615,
+      "loss_iou": 0.6976702809333801,
+      "loss_num": 0.035888671875,
+      "loss_xval": 1.5740514993667603,
+      "num_input_tokens_seen": 752305016,
+      "step": 4352
+    },
+    {
+      "epoch": 1.667177326694753,
+      "grad_norm": 114.88384420280931,
+      "learning_rate": 5e-06,
+      "loss": 1.0786,
+      "num_input_tokens_seen": 752478144,
+      "step": 4353
+    },
+    {
+      "epoch": 1.667177326694753,
+      "loss": 1.088322639465332,
+      "loss_ce": 0.17090633511543274,
+      "loss_iou": 0.4382079839706421,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.9174163341522217,
+      "num_input_tokens_seen": 752478144,
+      "step": 4353
+    },
+    {
+      "epoch": 1.6675603217158177,
+      "grad_norm": 214.57938217468157,
+      "learning_rate": 5e-06,
+      "loss": 1.4141,
+      "num_input_tokens_seen": 752651088,
+      "step": 4354
+    },
+    {
+      "epoch": 1.6675603217158177,
+      "loss": 1.4140644073486328,
+      "loss_ce": 0.15576159954071045,
+      "loss_iou": 0.568452000617981,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 1.258302927017212,
+      "num_input_tokens_seen": 752651088,
+      "step": 4354
+    },
+    {
+      "epoch": 1.6679433167368825,
+      "grad_norm": 95.12374924026861,
+      "learning_rate": 5e-06,
+      "loss": 1.4094,
+      "num_input_tokens_seen": 752823896,
+      "step": 4355
+    },
+    {
+      "epoch": 1.6679433167368825,
+      "loss": 1.4674533605575562,
+      "loss_ce": 0.13537859916687012,
+      "loss_iou": 8.569318057542831e+32,
+      "loss_num": 0.013671875,
+      "loss_xval": 8.203829036485028e+35,
+      "num_input_tokens_seen": 752823896,
+      "step": 4355
+    },
+    {
+      "epoch": 1.668326311757947,
+      "grad_norm": 90.87505988937167,
+      "learning_rate": 5e-06,
+      "loss": 1.558,
+      "num_input_tokens_seen": 752996504,
+      "step": 4356
+    },
+    {
+      "epoch": 1.668326311757947,
+      "loss": 1.5766923427581787,
+      "loss_ce": 0.16272133588790894,
+      "loss_iou": 0.6496124267578125,
+      "loss_num": 0.02294921875,
+      "loss_xval": 1.413970947265625,
+      "num_input_tokens_seen": 752996504,
+      "step": 4356
+    },
+    {
+      "epoch": 1.6687093067790117,
+      "grad_norm": 141.6348893494254,
+      "learning_rate": 5e-06,
+      "loss": 1.3547,
+      "num_input_tokens_seen": 753169632,
+      "step": 4357
+    },
+    {
+      "epoch": 1.6687093067790117,
+      "loss": 1.3793840408325195,
+      "loss_ce": 0.16662092506885529,
+      "loss_iou": 0.5807315111160278,
+      "loss_num": 0.01025390625,
+      "loss_xval": 1.2127630710601807,
+      "num_input_tokens_seen": 753169632,
+      "step": 4357
+    },
+    {
+      "epoch": 1.6690923018000765,
+      "grad_norm": 153.06472665831689,
+      "learning_rate": 5e-06,
+      "loss": 1.5865,
+      "num_input_tokens_seen": 753342568,
+      "step": 4358
+    },
+    {
+      "epoch": 1.6690923018000765,
+      "loss": 1.5346765518188477,
+      "loss_ce": 0.16870664060115814,
+      "loss_iou": 0.6440293192863464,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 1.3659700155258179,
+      "num_input_tokens_seen": 753342568,
+      "step": 4358
+    },
+    {
+      "epoch": 1.6694752968211413,
+      "grad_norm": 138.2465074214682,
+      "learning_rate": 5e-06,
+      "loss": 1.6133,
+      "num_input_tokens_seen": 753515872,
+      "step": 4359
+    },
+    {
+      "epoch": 1.6694752968211413,
+      "loss": 1.493098258972168,
+      "loss_ce": 0.18320278823375702,
+      "loss_iou": 0.6241936087608337,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 1.309895396232605,
+      "num_input_tokens_seen": 753515872,
+      "step": 4359
+    },
+    {
+      "epoch": 1.669858291842206,
+      "grad_norm": 98.27732783563468,
+      "learning_rate": 5e-06,
+      "loss": 1.5559,
+      "num_input_tokens_seen": 753688824,
+      "step": 4360
+    },
+    {
+      "epoch": 1.669858291842206,
+      "loss": 1.5809683799743652,
+      "loss_ce": 0.12172377109527588,
+      "loss_iou": 0.6667560935020447,
+      "loss_num": 0.025146484375,
+      "loss_xval": 1.4592446088790894,
+      "num_input_tokens_seen": 753688824,
+      "step": 4360
+    },
+    {
+      "epoch": 1.6702412868632708,
+      "grad_norm": 102.88437547538797,
+      "learning_rate": 5e-06,
+      "loss": 1.2817,
+      "num_input_tokens_seen": 753861536,
+      "step": 4361
+    },
+    {
+      "epoch": 1.6702412868632708,
+      "loss": 1.1647738218307495,
+      "loss_ce": 0.13653725385665894,
+      "loss_iou": 0.4613076150417328,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 1.0282366275787354,
+      "num_input_tokens_seen": 753861536,
+      "step": 4361
+    },
+    {
+      "epoch": 1.6706242818843355,
+      "grad_norm": 155.73607378861095,
+      "learning_rate": 5e-06,
+      "loss": 1.4204,
+      "num_input_tokens_seen": 754034784,
+      "step": 4362
+    },
+    {
+      "epoch": 1.6706242818843355,
+      "loss": 1.3382277488708496,
+      "loss_ce": 0.1748562753200531,
+      "loss_iou": 0.5535790920257568,
+      "loss_num": 0.01123046875,
+      "loss_xval": 1.1633715629577637,
+      "num_input_tokens_seen": 754034784,
+      "step": 4362
+    },
+    {
+      "epoch": 1.6710072769054003,
+      "grad_norm": 136.3262772158089,
+      "learning_rate": 5e-06,
+      "loss": 1.2393,
+      "num_input_tokens_seen": 754207320,
+      "step": 4363
+    },
+    {
+      "epoch": 1.6710072769054003,
+      "loss": 1.2228909730911255,
+      "loss_ce": 0.1309235692024231,
+      "loss_iou": 0.5086911916732788,
+      "loss_num": 0.014892578125,
+      "loss_xval": 1.0919673442840576,
+      "num_input_tokens_seen": 754207320,
+      "step": 4363
+    },
+    {
+      "epoch": 1.671390271926465,
+      "grad_norm": 268.5851211662383,
+      "learning_rate": 5e-06,
+      "loss": 1.7028,
+      "num_input_tokens_seen": 754379912,
+      "step": 4364
+    },
+    {
+      "epoch": 1.671390271926465,
+      "loss": 1.8059022426605225,
+      "loss_ce": 0.1507665067911148,
+      "loss_iou": 0.7666090726852417,
+      "loss_num": 0.0244140625,
+      "loss_xval": 1.6551358699798584,
+      "num_input_tokens_seen": 754379912,
+      "step": 4364
+    },
+    {
+      "epoch": 1.6717732669475298,
+      "grad_norm": 194.19436886467636,
+      "learning_rate": 5e-06,
+      "loss": 1.5381,
+      "num_input_tokens_seen": 754552584,
+      "step": 4365
+    },
+    {
+      "epoch": 1.6717732669475298,
+      "loss": 1.6535385847091675,
+      "loss_ce": 0.14655561745166779,
+      "loss_iou": 0.6829653978347778,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 1.5069830417633057,
+      "num_input_tokens_seen": 754552584,
+      "step": 4365
+    },
+    {
+      "epoch": 1.6721562619685946,
+      "grad_norm": 363.1074922804633,
+      "learning_rate": 5e-06,
+      "loss": 1.633,
+      "num_input_tokens_seen": 754725384,
+      "step": 4366
+    },
+    {
+      "epoch": 1.6721562619685946,
+      "loss": 1.7063180208206177,
+      "loss_ce": 0.16991159319877625,
+      "loss_iou": 0.7248987555503845,
+      "loss_num": 0.017333984375,
+      "loss_xval": 1.536406397819519,
+      "num_input_tokens_seen": 754725384,
+      "step": 4366
+    },
+    {
+      "epoch": 1.672539256989659,
+      "grad_norm": 333.9474035795017,
+      "learning_rate": 5e-06,
+      "loss": 2.5983,
+      "num_input_tokens_seen": 754898616,
+      "step": 4367
+    },
+    {
+      "epoch": 1.672539256989659,
+      "loss": 2.56595516204834,
+      "loss_ce": 0.19206538796424866,
+      "loss_iou": 1.064325213432312,
+      "loss_num": 0.049072265625,
+      "loss_xval": 2.373889684677124,
+      "num_input_tokens_seen": 754898616,
+      "step": 4367
+    },
+    {
+      "epoch": 1.6729222520107239,
+      "grad_norm": 89.54363251972225,
+      "learning_rate": 5e-06,
+      "loss": 2.1643,
+      "num_input_tokens_seen": 755071528,
+      "step": 4368
+    },
+    {
+      "epoch": 1.6729222520107239,
+      "loss": 2.049285888671875,
+      "loss_ce": 0.1552119255065918,
+      "loss_iou": 0.8515169620513916,
+      "loss_num": 0.0380859375,
+      "loss_xval": 1.8940738439559937,
+      "num_input_tokens_seen": 755071528,
+      "step": 4368
+    },
+    {
+      "epoch": 1.6733052470317886,
+      "grad_norm": 288.5888479185838,
+      "learning_rate": 5e-06,
+      "loss": 1.4525,
+      "num_input_tokens_seen": 755244576,
+      "step": 4369
+    },
+    {
+      "epoch": 1.6733052470317886,
+      "loss": 1.488768458366394,
+      "loss_ce": 0.19600477814674377,
+      "loss_iou": 0.6057324409484863,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 1.2927637100219727,
+      "num_input_tokens_seen": 755244576,
+      "step": 4369
+    },
+    {
+      "epoch": 1.6736882420528532,
+      "grad_norm": 175.09232969517532,
+      "learning_rate": 5e-06,
+      "loss": 1.8828,
+      "num_input_tokens_seen": 755417560,
+      "step": 4370
+    },
+    {
+      "epoch": 1.6736882420528532,
+      "loss": 1.8737170696258545,
+      "loss_ce": 0.18630965054035187,
+      "loss_iou": 0.7724452018737793,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 1.6874074935913086,
+      "num_input_tokens_seen": 755417560,
+      "step": 4370
+    },
+    {
+      "epoch": 1.674071237073918,
+      "grad_norm": 157.99918288709196,
+      "learning_rate": 5e-06,
+      "loss": 1.747,
+      "num_input_tokens_seen": 755590256,
+      "step": 4371
+    },
+    {
+      "epoch": 1.674071237073918,
+      "loss": 1.6813737154006958,
+      "loss_ce": 0.19650748372077942,
+      "loss_iou": 0.6870436668395996,
+      "loss_num": 0.022216796875,
+      "loss_xval": 1.4848661422729492,
+      "num_input_tokens_seen": 755590256,
+      "step": 4371
+    },
+    {
+      "epoch": 1.6744542320949827,
+      "grad_norm": 152.377673994168,
+      "learning_rate": 5e-06,
+      "loss": 1.7661,
+      "num_input_tokens_seen": 755763224,
+      "step": 4372
+    },
+    {
+      "epoch": 1.6744542320949827,
+      "loss": 1.7150580883026123,
+      "loss_ce": 0.1900370866060257,
+      "loss_iou": 0.6973097324371338,
+      "loss_num": 0.026123046875,
+      "loss_xval": 1.5250210762023926,
+      "num_input_tokens_seen": 755763224,
+      "step": 4372
+    },
+    {
+      "epoch": 1.6748372271160474,
+      "grad_norm": 207.45305571019634,
+      "learning_rate": 5e-06,
+      "loss": 1.6515,
+      "num_input_tokens_seen": 755936408,
+      "step": 4373
+    },
+    {
+      "epoch": 1.6748372271160474,
+      "loss": 1.5031293630599976,
+      "loss_ce": 0.1421300172805786,
+      "loss_iou": 0.6314579248428345,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 1.360999345779419,
+      "num_input_tokens_seen": 755936408,
+      "step": 4373
+    },
+    {
+      "epoch": 1.6752202221371122,
+      "grad_norm": 114.78800827220954,
+      "learning_rate": 5e-06,
+      "loss": 1.5075,
+      "num_input_tokens_seen": 756109432,
+      "step": 4374
+    },
+    {
+      "epoch": 1.6752202221371122,
+      "loss": 1.4894447326660156,
+      "loss_ce": 0.1648956835269928,
+      "loss_iou": 0.6026736497879028,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 1.3245489597320557,
+      "num_input_tokens_seen": 756109432,
+      "step": 4374
+    },
+    {
+      "epoch": 1.675603217158177,
+      "grad_norm": 154.2731282607183,
+      "learning_rate": 5e-06,
+      "loss": 1.4073,
+      "num_input_tokens_seen": 756282152,
+      "step": 4375
+    },
+    {
+      "epoch": 1.675603217158177,
+      "loss": 1.337599754333496,
+      "loss_ce": 0.1523275226354599,
+      "loss_iou": 0.5486602783203125,
+      "loss_num": 0.017578125,
+      "loss_xval": 1.185272216796875,
+      "num_input_tokens_seen": 756282152,
+      "step": 4375
+    },
+    {
+      "epoch": 1.6759862121792417,
+      "grad_norm": 148.39229143516698,
+      "learning_rate": 5e-06,
+      "loss": 1.3415,
+      "num_input_tokens_seen": 756455248,
+      "step": 4376
+    },
+    {
+      "epoch": 1.6759862121792417,
+      "loss": 1.2090262174606323,
+      "loss_ce": 0.15305881202220917,
+      "loss_iou": 0.47675997018814087,
+      "loss_num": 0.0205078125,
+      "loss_xval": 1.0485211610794067,
+      "num_input_tokens_seen": 756455248,
+      "step": 4376
+    },
+    {
+      "epoch": 1.6763692072003065,
+      "grad_norm": 118.59006548468923,
+      "learning_rate": 5e-06,
+      "loss": 1.4575,
+      "num_input_tokens_seen": 756628136,
+      "step": 4377
+    },
+    {
+      "epoch": 1.6763692072003065,
+      "loss": 1.5994715690612793,
+      "loss_ce": 0.16459201276302338,
+      "loss_iou": 0.6657277345657349,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 1.4348795413970947,
+      "num_input_tokens_seen": 756628136,
+      "step": 4377
+    },
+    {
+      "epoch": 1.6767522022213712,
+      "grad_norm": 126.86195805763037,
+      "learning_rate": 5e-06,
+      "loss": 1.5015,
+      "num_input_tokens_seen": 756801352,
+      "step": 4378
+    },
+    {
+      "epoch": 1.6767522022213712,
+      "loss": 1.5466210842132568,
+      "loss_ce": 0.15083849430084229,
+      "loss_iou": 0.6531525254249573,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 1.3957825899124146,
+      "num_input_tokens_seen": 756801352,
+      "step": 4378
+    },
+    {
+      "epoch": 1.677135197242436,
+      "grad_norm": 78.38021438088244,
+      "learning_rate": 5e-06,
+      "loss": 1.2303,
+      "num_input_tokens_seen": 756974280,
+      "step": 4379
+    },
+    {
+      "epoch": 1.677135197242436,
+      "loss": 1.1930649280548096,
+      "loss_ce": 0.16140449047088623,
+      "loss_iou": 0.47673720121383667,
+      "loss_num": 0.015625,
+      "loss_xval": 1.0316604375839233,
+      "num_input_tokens_seen": 756974280,
+      "step": 4379
+    },
+    {
+      "epoch": 1.6775181922635007,
+      "grad_norm": 107.58407424952352,
+      "learning_rate": 5e-06,
+      "loss": 1.4725,
+      "num_input_tokens_seen": 757147352,
+      "step": 4380
+    },
+    {
+      "epoch": 1.6775181922635007,
+      "loss": 1.5374525785446167,
+      "loss_ce": 0.1311381757259369,
+      "loss_iou": 0.6446855068206787,
+      "loss_num": 0.0234375,
+      "loss_xval": 1.4063143730163574,
+      "num_input_tokens_seen": 757147352,
+      "step": 4380
+    },
+    {
+      "epoch": 1.6779011872845653,
+      "grad_norm": 190.88086679141256,
+      "learning_rate": 5e-06,
+      "loss": 1.4762,
+      "num_input_tokens_seen": 757320336,
+      "step": 4381
+    },
+    {
+      "epoch": 1.6779011872845653,
+      "loss": 1.4439189434051514,
+      "loss_ce": 0.17356400191783905,
+      "loss_iou": 0.5909727811813354,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 1.270354986190796,
+      "num_input_tokens_seen": 757320336,
+      "step": 4381
+    },
+    {
+      "epoch": 1.67828418230563,
+      "grad_norm": 116.10727496920556,
+      "learning_rate": 5e-06,
+      "loss": 1.742,
+      "num_input_tokens_seen": 757492872,
+      "step": 4382
+    },
+    {
+      "epoch": 1.67828418230563,
+      "loss": 1.6558839082717896,
+      "loss_ce": 0.20242519676685333,
+      "loss_iou": 0.6721638441085815,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 1.453458547592163,
+      "num_input_tokens_seen": 757492872,
+      "step": 4382
+    },
+    {
+      "epoch": 1.6786671773266948,
+      "grad_norm": 141.76299308346458,
+      "learning_rate": 5e-06,
+      "loss": 1.4861,
+      "num_input_tokens_seen": 757665824,
+      "step": 4383
+    },
+    {
+      "epoch": 1.6786671773266948,
+      "loss": 1.4049732685089111,
+      "loss_ce": 0.184755340218544,
+      "loss_iou": 0.5702530145645142,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 1.2202179431915283,
+      "num_input_tokens_seen": 757665824,
+      "step": 4383
+    },
+    {
+      "epoch": 1.6790501723477593,
+      "grad_norm": 135.79000566704417,
+      "learning_rate": 5e-06,
+      "loss": 1.5194,
+      "num_input_tokens_seen": 757838728,
+      "step": 4384
+    },
+    {
+      "epoch": 1.6790501723477593,
+      "loss": 1.4411894083023071,
+      "loss_ce": 0.17360654473304749,
+      "loss_iou": 0.5963310599327087,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 1.2675827741622925,
+      "num_input_tokens_seen": 757838728,
+      "step": 4384
+    },
+    {
+      "epoch": 1.679433167368824,
+      "grad_norm": 146.4705927634235,
+      "learning_rate": 5e-06,
+      "loss": 1.541,
+      "num_input_tokens_seen": 758011640,
+      "step": 4385
+    },
+    {
+      "epoch": 1.679433167368824,
+      "loss": 1.5647530555725098,
+      "loss_ce": 0.16160601377487183,
+      "loss_iou": 0.6532489061355591,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.4031469821929932,
+      "num_input_tokens_seen": 758011640,
+      "step": 4385
+    },
+    {
+      "epoch": 1.6798161623898888,
+      "grad_norm": 177.39243836807586,
+      "learning_rate": 5e-06,
+      "loss": 2.0011,
+      "num_input_tokens_seen": 758184696,
+      "step": 4386
+    },
+    {
+      "epoch": 1.6798161623898888,
+      "loss": 2.0785374641418457,
+      "loss_ce": 0.16670063138008118,
+      "loss_iou": 0.8813335299491882,
+      "loss_num": 0.02978515625,
+      "loss_xval": 1.9118369817733765,
+      "num_input_tokens_seen": 758184696,
+      "step": 4386
+    },
+    {
+      "epoch": 1.6801991574109536,
+      "grad_norm": 167.1618213411226,
+      "learning_rate": 5e-06,
+      "loss": 1.4763,
+      "num_input_tokens_seen": 758357448,
+      "step": 4387
+    },
+    {
+      "epoch": 1.6801991574109536,
+      "loss": 1.451549768447876,
+      "loss_ce": 0.1958613395690918,
+      "loss_iou": 0.5861572027206421,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 1.2556884288787842,
+      "num_input_tokens_seen": 758357448,
+      "step": 4387
+    },
+    {
+      "epoch": 1.6805821524320184,
+      "grad_norm": 120.3318231030432,
+      "learning_rate": 5e-06,
+      "loss": 1.3509,
+      "num_input_tokens_seen": 758530552,
+      "step": 4388
+    },
+    {
+      "epoch": 1.6805821524320184,
+      "loss": 1.3233647346496582,
+      "loss_ce": 0.12391694635152817,
+      "loss_iou": 0.555976927280426,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 1.199447751045227,
+      "num_input_tokens_seen": 758530552,
+      "step": 4388
+    },
+    {
+      "epoch": 1.6809651474530831,
+      "grad_norm": 133.2513019471835,
+      "learning_rate": 5e-06,
+      "loss": 1.3939,
+      "num_input_tokens_seen": 758703744,
+      "step": 4389
+    },
+    {
+      "epoch": 1.6809651474530831,
+      "loss": 1.524483561515808,
+      "loss_ce": 0.18629640340805054,
+      "loss_iou": 0.6356615424156189,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 1.3381870985031128,
+      "num_input_tokens_seen": 758703744,
+      "step": 4389
+    },
+    {
+      "epoch": 1.6813481424741479,
+      "grad_norm": 126.09216545969078,
+      "learning_rate": 5e-06,
+      "loss": 1.3425,
+      "num_input_tokens_seen": 758876632,
+      "step": 4390
+    },
+    {
+      "epoch": 1.6813481424741479,
+      "loss": 1.4416831731796265,
+      "loss_ce": 0.17311879992485046,
+      "loss_iou": 0.5921832323074341,
+      "loss_num": 0.016845703125,
+      "loss_xval": 1.2685644626617432,
+      "num_input_tokens_seen": 758876632,
+      "step": 4390
+    },
+    {
+      "epoch": 1.6817311374952126,
+      "grad_norm": 119.3339984017816,
+      "learning_rate": 5e-06,
+      "loss": 1.6733,
+      "num_input_tokens_seen": 759049768,
+      "step": 4391
+    },
+    {
+      "epoch": 1.6817311374952126,
+      "loss": 1.6312274932861328,
+      "loss_ce": 0.1816640943288803,
+      "loss_iou": 0.6654555201530457,
+      "loss_num": 0.023681640625,
+      "loss_xval": 1.4495635032653809,
+      "num_input_tokens_seen": 759049768,
+      "step": 4391
+    },
+    {
+      "epoch": 1.6821141325162774,
+      "grad_norm": 87.3722891188307,
+      "learning_rate": 5e-06,
+      "loss": 1.2888,
+      "num_input_tokens_seen": 759222352,
+      "step": 4392
+    },
+    {
+      "epoch": 1.6821141325162774,
+      "loss": 1.141264796257019,
+      "loss_ce": 0.13132750988006592,
+      "loss_iou": 0.47223857045173645,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 1.0099372863769531,
+      "num_input_tokens_seen": 759222352,
+      "step": 4392
+    },
+    {
+      "epoch": 1.6824971275373422,
+      "grad_norm": 94.5590831376501,
+      "learning_rate": 5e-06,
+      "loss": 1.4809,
+      "num_input_tokens_seen": 759395056,
+      "step": 4393
+    },
+    {
+      "epoch": 1.6824971275373422,
+      "loss": 1.6476218700408936,
+      "loss_ce": 0.1588333249092102,
+      "loss_iou": 0.6961917877197266,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.4887886047363281,
+      "num_input_tokens_seen": 759395056,
+      "step": 4393
+    },
+    {
+      "epoch": 1.6828801225584067,
+      "grad_norm": 87.55062154600387,
+      "learning_rate": 5e-06,
+      "loss": 1.4034,
+      "num_input_tokens_seen": 759568416,
+      "step": 4394
+    },
+    {
+      "epoch": 1.6828801225584067,
+      "loss": 1.526280403137207,
+      "loss_ce": 0.19728177785873413,
+      "loss_iou": 0.6122837662696838,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 1.3289986848831177,
+      "num_input_tokens_seen": 759568416,
+      "step": 4394
+    },
+    {
+      "epoch": 1.6832631175794714,
+      "grad_norm": 169.4888444747027,
+      "learning_rate": 5e-06,
+      "loss": 1.4735,
+      "num_input_tokens_seen": 759741552,
+      "step": 4395
+    },
+    {
+      "epoch": 1.6832631175794714,
+      "loss": 1.3289074897766113,
+      "loss_ce": 0.16519808769226074,
+      "loss_iou": 0.5489262342453003,
+      "loss_num": 0.01318359375,
+      "loss_xval": 1.1637094020843506,
+      "num_input_tokens_seen": 759741552,
+      "step": 4395
+    },
+    {
+      "epoch": 1.6836461126005362,
+      "grad_norm": 171.116956366303,
+      "learning_rate": 5e-06,
+      "loss": 1.5816,
+      "num_input_tokens_seen": 759914608,
+      "step": 4396
+    },
+    {
+      "epoch": 1.6836461126005362,
+      "loss": 1.6261197328567505,
+      "loss_ce": 0.1836853176355362,
+      "loss_iou": 0.6743574738502502,
+      "loss_num": 0.018798828125,
+      "loss_xval": 1.4424344301223755,
+      "num_input_tokens_seen": 759914608,
+      "step": 4396
+    },
+    {
+      "epoch": 1.684029107621601,
+      "grad_norm": 108.58593223828433,
+      "learning_rate": 5e-06,
+      "loss": 1.4723,
+      "num_input_tokens_seen": 760087568,
+      "step": 4397
+    },
+    {
+      "epoch": 1.684029107621601,
+      "loss": 1.4462080001831055,
+      "loss_ce": 0.18597783148288727,
+      "loss_iou": 0.5661959648132324,
+      "loss_num": 0.025634765625,
+      "loss_xval": 1.2602300643920898,
+      "num_input_tokens_seen": 760087568,
+      "step": 4397
+    },
+    {
+      "epoch": 1.6844121026426655,
+      "grad_norm": 158.90865656353407,
+      "learning_rate": 5e-06,
+      "loss": 1.3799,
+      "num_input_tokens_seen": 760260616,
+      "step": 4398
+    },
+    {
+      "epoch": 1.6844121026426655,
+      "loss": 1.4212050437927246,
+      "loss_ce": 0.15659356117248535,
+      "loss_iou": 0.5904814004898071,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 1.2646114826202393,
+      "num_input_tokens_seen": 760260616,
+      "step": 4398
+    },
+    {
+      "epoch": 1.6847950976637303,
+      "grad_norm": 166.7730290021102,
+      "learning_rate": 5e-06,
+      "loss": 1.3809,
+      "num_input_tokens_seen": 760433520,
+      "step": 4399
+    },
+    {
+      "epoch": 1.6847950976637303,
+      "loss": 1.2549279928207397,
+      "loss_ce": 0.12187740951776505,
+      "loss_iou": 0.524761974811554,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 1.133050560951233,
+      "num_input_tokens_seen": 760433520,
+      "step": 4399
+    },
+    {
+      "epoch": 1.685178092684795,
+      "grad_norm": 221.98706377249738,
+      "learning_rate": 5e-06,
+      "loss": 1.8523,
+      "num_input_tokens_seen": 760606408,
+      "step": 4400
+    },
+    {
+      "epoch": 1.685178092684795,
+      "loss": 1.804720163345337,
+      "loss_ce": 0.17628535628318787,
+      "loss_iou": 0.7449730634689331,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 1.6284348964691162,
+      "num_input_tokens_seen": 760606408,
+      "step": 4400
+    },
+    {
+      "epoch": 1.6855610877058598,
+      "grad_norm": 112.42241183740752,
+      "learning_rate": 5e-06,
+      "loss": 1.4014,
+      "num_input_tokens_seen": 760779048,
+      "step": 4401
+    },
+    {
+      "epoch": 1.6855610877058598,
+      "loss": 1.5267062187194824,
+      "loss_ce": 0.23780015110969543,
+      "loss_iou": 0.5960673093795776,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.2889058589935303,
+      "num_input_tokens_seen": 760779048,
+      "step": 4401
+    },
+    {
+      "epoch": 1.6859440827269245,
+      "grad_norm": 143.74853706173124,
+      "learning_rate": 5e-06,
+      "loss": 1.4157,
+      "num_input_tokens_seen": 760951968,
+      "step": 4402
+    },
+    {
+      "epoch": 1.6859440827269245,
+      "loss": 1.2913093566894531,
+      "loss_ce": 0.1852709800004959,
+      "loss_iou": 0.5213266611099243,
+      "loss_num": 0.0126953125,
+      "loss_xval": 1.1060383319854736,
+      "num_input_tokens_seen": 760951968,
+      "step": 4402
+    },
+    {
+      "epoch": 1.6863270777479893,
+      "grad_norm": 138.56000546703723,
+      "learning_rate": 5e-06,
+      "loss": 1.6607,
+      "num_input_tokens_seen": 761125120,
+      "step": 4403
+    },
+    {
+      "epoch": 1.6863270777479893,
+      "loss": 1.567737102508545,
+      "loss_ce": 0.15642426908016205,
+      "loss_iou": 0.6509536504745483,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 1.4113128185272217,
+      "num_input_tokens_seen": 761125120,
+      "step": 4403
+    },
+    {
+      "epoch": 1.686710072769054,
+      "grad_norm": 95.27413689996389,
+      "learning_rate": 5e-06,
+      "loss": 1.371,
+      "num_input_tokens_seen": 761297968,
+      "step": 4404
+    },
+    {
+      "epoch": 1.686710072769054,
+      "loss": 1.440436840057373,
+      "loss_ce": 0.19408228993415833,
+      "loss_iou": 0.5803000330924988,
+      "loss_num": 0.01708984375,
+      "loss_xval": 1.2463544607162476,
+      "num_input_tokens_seen": 761297968,
+      "step": 4404
+    },
+    {
+      "epoch": 1.6870930677901188,
+      "grad_norm": 103.479557156628,
+      "learning_rate": 5e-06,
+      "loss": 1.2875,
+      "num_input_tokens_seen": 761470936,
+      "step": 4405
+    },
+    {
+      "epoch": 1.6870930677901188,
+      "loss": 1.2585489749908447,
+      "loss_ce": 0.16511748731136322,
+      "loss_iou": 0.5093622207641602,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 1.0934314727783203,
+      "num_input_tokens_seen": 761470936,
+      "step": 4405
+    },
+    {
+      "epoch": 1.6874760628111836,
+      "grad_norm": 347.94817848014566,
+      "learning_rate": 5e-06,
+      "loss": 1.6833,
+      "num_input_tokens_seen": 761643976,
+      "step": 4406
+    },
+    {
+      "epoch": 1.6874760628111836,
+      "loss": 1.9606256484985352,
+      "loss_ce": 0.1655784398317337,
+      "loss_iou": 0.8379838466644287,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 1.7950472831726074,
+      "num_input_tokens_seen": 761643976,
+      "step": 4406
+    },
+    {
+      "epoch": 1.6878590578322483,
+      "grad_norm": 123.26309271138102,
+      "learning_rate": 5e-06,
+      "loss": 1.7184,
+      "num_input_tokens_seen": 761817392,
+      "step": 4407
+    },
+    {
+      "epoch": 1.6878590578322483,
+      "loss": 1.6759836673736572,
+      "loss_ce": 0.17374470829963684,
+      "loss_iou": 0.6895960569381714,
+      "loss_num": 0.024658203125,
+      "loss_xval": 1.5022389888763428,
+      "num_input_tokens_seen": 761817392,
+      "step": 4407
+    },
+    {
+      "epoch": 1.6882420528533129,
+      "grad_norm": 99.28764694934321,
+      "learning_rate": 5e-06,
+      "loss": 1.3062,
+      "num_input_tokens_seen": 761989928,
+      "step": 4408
+    },
+    {
+      "epoch": 1.6882420528533129,
+      "loss": 1.1667100191116333,
+      "loss_ce": 0.140516996383667,
+      "loss_iou": 0.47569724917411804,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 1.0261931419372559,
+      "num_input_tokens_seen": 761989928,
+      "step": 4408
+    },
+    {
+      "epoch": 1.6886250478743776,
+      "grad_norm": 171.47767896132896,
+      "learning_rate": 5e-06,
+      "loss": 1.56,
+      "num_input_tokens_seen": 762162840,
+      "step": 4409
+    },
+    {
+      "epoch": 1.6886250478743776,
+      "loss": 1.5050292015075684,
+      "loss_ce": 0.1594279408454895,
+      "loss_iou": 0.6307169198989868,
+      "loss_num": 0.016845703125,
+      "loss_xval": 1.3456013202667236,
+      "num_input_tokens_seen": 762162840,
+      "step": 4409
+    },
+    {
+      "epoch": 1.6890080428954424,
+      "grad_norm": 121.89540539163043,
+      "learning_rate": 5e-06,
+      "loss": 1.5027,
+      "num_input_tokens_seen": 762335752,
+      "step": 4410
+    },
+    {
+      "epoch": 1.6890080428954424,
+      "loss": 1.3650343418121338,
+      "loss_ce": 0.13624507188796997,
+      "loss_iou": 0.574462354183197,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 1.228789210319519,
+      "num_input_tokens_seen": 762335752,
+      "step": 4410
+    },
+    {
+      "epoch": 1.6893910379165071,
+      "grad_norm": 86.30101791986955,
+      "learning_rate": 5e-06,
+      "loss": 1.3677,
+      "num_input_tokens_seen": 762508736,
+      "step": 4411
+    },
+    {
+      "epoch": 1.6893910379165071,
+      "loss": 1.0739622116088867,
+      "loss_ce": 0.15682095289230347,
+      "loss_iou": 0.4264737367630005,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.9171411991119385,
+      "num_input_tokens_seen": 762508736,
+      "step": 4411
+    },
+    {
+      "epoch": 1.6897740329375717,
+      "grad_norm": 130.31307203258814,
+      "learning_rate": 5e-06,
+      "loss": 1.3792,
+      "num_input_tokens_seen": 762681840,
+      "step": 4412
+    },
+    {
+      "epoch": 1.6897740329375717,
+      "loss": 1.4163744449615479,
+      "loss_ce": 0.16193100810050964,
+      "loss_iou": 0.5822387933731079,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 1.2544434070587158,
+      "num_input_tokens_seen": 762681840,
+      "step": 4412
+    },
+    {
+      "epoch": 1.6901570279586364,
+      "grad_norm": 217.68686266869614,
+      "learning_rate": 5e-06,
+      "loss": 1.6925,
+      "num_input_tokens_seen": 762854712,
+      "step": 4413
+    },
+    {
+      "epoch": 1.6901570279586364,
+      "loss": 1.7711422443389893,
+      "loss_ce": 0.11256309598684311,
+      "loss_iou": 0.7717334032058716,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 1.6585791110992432,
+      "num_input_tokens_seen": 762854712,
+      "step": 4413
+    },
+    {
+      "epoch": 1.6905400229797012,
+      "grad_norm": 219.31080060334318,
+      "learning_rate": 5e-06,
+      "loss": 1.9431,
+      "num_input_tokens_seen": 763027792,
+      "step": 4414
+    },
+    {
+      "epoch": 1.6905400229797012,
+      "loss": 1.839816689491272,
+      "loss_ce": 0.14460596442222595,
+      "loss_iou": 0.7895609140396118,
+      "loss_num": 0.023193359375,
+      "loss_xval": 1.6952106952667236,
+      "num_input_tokens_seen": 763027792,
+      "step": 4414
+    },
+    {
+      "epoch": 1.690923018000766,
+      "grad_norm": 111.31137192535812,
+      "learning_rate": 5e-06,
+      "loss": 1.3674,
+      "num_input_tokens_seen": 763197408,
+      "step": 4415
+    },
+    {
+      "epoch": 1.690923018000766,
+      "loss": 1.3934731483459473,
+      "loss_ce": 0.17336517572402954,
+      "loss_iou": 0.5678481459617615,
+      "loss_num": 0.016845703125,
+      "loss_xval": 1.220107913017273,
+      "num_input_tokens_seen": 763197408,
+      "step": 4415
+    },
+    {
+      "epoch": 1.6913060130218307,
+      "grad_norm": 199.10840495806485,
+      "learning_rate": 5e-06,
+      "loss": 1.4781,
+      "num_input_tokens_seen": 763370200,
+      "step": 4416
+    },
+    {
+      "epoch": 1.6913060130218307,
+      "loss": 1.648474931716919,
+      "loss_ce": 0.18365013599395752,
+      "loss_iou": 0.6617642045021057,
+      "loss_num": 0.0283203125,
+      "loss_xval": 1.4648247957229614,
+      "num_input_tokens_seen": 763370200,
+      "step": 4416
+    },
+    {
+      "epoch": 1.6916890080428955,
+      "grad_norm": 172.62679094094702,
+      "learning_rate": 5e-06,
+      "loss": 1.6154,
+      "num_input_tokens_seen": 763543024,
+      "step": 4417
+    },
+    {
+      "epoch": 1.6916890080428955,
+      "loss": 1.767137885093689,
+      "loss_ce": 0.20010000467300415,
+      "loss_iou": 0.731486439704895,
+      "loss_num": 0.020751953125,
+      "loss_xval": 1.56703782081604,
+      "num_input_tokens_seen": 763543024,
+      "step": 4417
+    },
+    {
+      "epoch": 1.6920720030639602,
+      "grad_norm": 155.6231546622594,
+      "learning_rate": 5e-06,
+      "loss": 1.5274,
+      "num_input_tokens_seen": 763715832,
+      "step": 4418
+    },
+    {
+      "epoch": 1.6920720030639602,
+      "loss": 1.490821361541748,
+      "loss_ce": 0.16336703300476074,
+      "loss_iou": 0.6236423254013062,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 1.3274543285369873,
+      "num_input_tokens_seen": 763715832,
+      "step": 4418
+    },
+    {
+      "epoch": 1.692454998085025,
+      "grad_norm": 136.93419745344565,
+      "learning_rate": 5e-06,
+      "loss": 1.5552,
+      "num_input_tokens_seen": 763888800,
+      "step": 4419
+    },
+    {
+      "epoch": 1.692454998085025,
+      "loss": 1.453904628753662,
+      "loss_ce": 0.16904865205287933,
+      "loss_iou": 0.5940271615982056,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 1.2848560810089111,
+      "num_input_tokens_seen": 763888800,
+      "step": 4419
+    },
+    {
+      "epoch": 1.6928379931060897,
+      "grad_norm": 119.27385611344276,
+      "learning_rate": 5e-06,
+      "loss": 1.4746,
+      "num_input_tokens_seen": 764061912,
+      "step": 4420
+    },
+    {
+      "epoch": 1.6928379931060897,
+      "loss": 1.4965870380401611,
+      "loss_ce": 0.16408222913742065,
+      "loss_iou": 0.6219103336334229,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 1.3325047492980957,
+      "num_input_tokens_seen": 764061912,
+      "step": 4420
+    },
+    {
+      "epoch": 1.6932209881271545,
+      "grad_norm": 148.22914127729123,
+      "learning_rate": 5e-06,
+      "loss": 1.2519,
+      "num_input_tokens_seen": 764234864,
+      "step": 4421
+    },
+    {
+      "epoch": 1.6932209881271545,
+      "loss": 1.1870217323303223,
+      "loss_ce": 0.1505342274904251,
+      "loss_iou": 0.48453712463378906,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 1.0364875793457031,
+      "num_input_tokens_seen": 764234864,
+      "step": 4421
+    },
+    {
+      "epoch": 1.693603983148219,
+      "grad_norm": 217.42243318030833,
+      "learning_rate": 5e-06,
+      "loss": 1.4346,
+      "num_input_tokens_seen": 764407416,
+      "step": 4422
+    },
+    {
+      "epoch": 1.693603983148219,
+      "loss": 1.5068445205688477,
+      "loss_ce": 0.1651231348514557,
+      "loss_iou": 0.6330951452255249,
+      "loss_num": 0.01513671875,
+      "loss_xval": 1.3417212963104248,
+      "num_input_tokens_seen": 764407416,
+      "step": 4422
+    },
+    {
+      "epoch": 1.6939869781692838,
+      "grad_norm": 144.58956498694056,
+      "learning_rate": 5e-06,
+      "loss": 1.5736,
+      "num_input_tokens_seen": 764580328,
+      "step": 4423
+    },
+    {
+      "epoch": 1.6939869781692838,
+      "loss": 1.440114974975586,
+      "loss_ce": 0.17704446613788605,
+      "loss_iou": 0.5746352672576904,
+      "loss_num": 0.022705078125,
+      "loss_xval": 1.2630705833435059,
+      "num_input_tokens_seen": 764580328,
+      "step": 4423
+    },
+    {
+      "epoch": 1.6943699731903485,
+      "grad_norm": 144.27529345784765,
+      "learning_rate": 5e-06,
+      "loss": 1.1634,
+      "num_input_tokens_seen": 764753280,
+      "step": 4424
+    },
+    {
+      "epoch": 1.6943699731903485,
+      "loss": 1.0947041511535645,
+      "loss_ce": 0.1289558708667755,
+      "loss_iou": -2.2310650564016837e+32,
+      "loss_num": 0.0108642578125,
+      "loss_xval": -1.5687176177824339e+31,
+      "num_input_tokens_seen": 764753280,
+      "step": 4424
+    },
+    {
+      "epoch": 1.6947529682114133,
+      "grad_norm": 133.67172424205728,
+      "learning_rate": 5e-06,
+      "loss": 1.4623,
+      "num_input_tokens_seen": 764926280,
+      "step": 4425
+    },
+    {
+      "epoch": 1.6947529682114133,
+      "loss": 1.499297857284546,
+      "loss_ce": 0.19930747151374817,
+      "loss_iou": 0.6100171804428101,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 1.2999904155731201,
+      "num_input_tokens_seen": 764926280,
+      "step": 4425
+    },
+    {
+      "epoch": 1.6951359632324778,
+      "grad_norm": 107.91597408181849,
+      "learning_rate": 5e-06,
+      "loss": 1.1663,
+      "num_input_tokens_seen": 765098848,
+      "step": 4426
+    },
+    {
+      "epoch": 1.6951359632324778,
+      "loss": 1.2694299221038818,
+      "loss_ce": 0.13105368614196777,
+      "loss_iou": 0.5242815017700195,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 1.138376235961914,
+      "num_input_tokens_seen": 765098848,
+      "step": 4426
+    },
+    {
+      "epoch": 1.6955189582535426,
+      "grad_norm": 147.9407199283946,
+      "learning_rate": 5e-06,
+      "loss": 1.6608,
+      "num_input_tokens_seen": 765271832,
+      "step": 4427
+    },
+    {
+      "epoch": 1.6955189582535426,
+      "loss": 1.682665228843689,
+      "loss_ce": 0.16860201954841614,
+      "loss_iou": 0.6825686693191528,
+      "loss_num": 0.02978515625,
+      "loss_xval": 1.5140631198883057,
+      "num_input_tokens_seen": 765271832,
+      "step": 4427
+    },
+    {
+      "epoch": 1.6959019532746074,
+      "grad_norm": 96.45699860235075,
+      "learning_rate": 5e-06,
+      "loss": 1.2963,
+      "num_input_tokens_seen": 765444376,
+      "step": 4428
+    },
+    {
+      "epoch": 1.6959019532746074,
+      "loss": 1.4512474536895752,
+      "loss_ce": 0.110992431640625,
+      "loss_iou": 0.6127086877822876,
+      "loss_num": 0.02294921875,
+      "loss_xval": 1.3402550220489502,
+      "num_input_tokens_seen": 765444376,
+      "step": 4428
+    },
+    {
+      "epoch": 1.6962849482956721,
+      "grad_norm": 111.69842313932696,
+      "learning_rate": 5e-06,
+      "loss": 1.3578,
+      "num_input_tokens_seen": 765617160,
+      "step": 4429
+    },
+    {
+      "epoch": 1.6962849482956721,
+      "loss": 1.37142014503479,
+      "loss_ce": 0.12713995575904846,
+      "loss_iou": 0.5754330158233643,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 1.2442803382873535,
+      "num_input_tokens_seen": 765617160,
+      "step": 4429
+    },
+    {
+      "epoch": 1.6966679433167369,
+      "grad_norm": 126.36557345297781,
+      "learning_rate": 5e-06,
+      "loss": 1.1393,
+      "num_input_tokens_seen": 765789968,
+      "step": 4430
+    },
+    {
+      "epoch": 1.6966679433167369,
+      "loss": 1.105392575263977,
+      "loss_ce": 0.14927037060260773,
+      "loss_iou": 0.44925248622894287,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.9561222195625305,
+      "num_input_tokens_seen": 765789968,
+      "step": 4430
+    },
+    {
+      "epoch": 1.6970509383378016,
+      "grad_norm": 205.77526077065198,
+      "learning_rate": 5e-06,
+      "loss": 1.8327,
+      "num_input_tokens_seen": 765962976,
+      "step": 4431
+    },
+    {
+      "epoch": 1.6970509383378016,
+      "loss": 1.962249517440796,
+      "loss_ce": 0.14602002501487732,
+      "loss_iou": 0.8052705526351929,
+      "loss_num": 0.041015625,
+      "loss_xval": 1.8162295818328857,
+      "num_input_tokens_seen": 765962976,
+      "step": 4431
+    },
+    {
+      "epoch": 1.6974339333588664,
+      "grad_norm": 150.59541879231966,
+      "learning_rate": 5e-06,
+      "loss": 1.1428,
+      "num_input_tokens_seen": 766135560,
+      "step": 4432
+    },
+    {
+      "epoch": 1.6974339333588664,
+      "loss": 1.0464577674865723,
+      "loss_ce": 0.1264677792787552,
+      "loss_iou": -1011550697553920.0,
+      "loss_num": 0.01708984375,
+      "loss_xval": -963172185931776.0,
+      "num_input_tokens_seen": 766135560,
+      "step": 4432
+    },
+    {
+      "epoch": 1.6978169283799311,
+      "grad_norm": 119.31109674841161,
+      "learning_rate": 5e-06,
+      "loss": 1.7125,
+      "num_input_tokens_seen": 766308248,
+      "step": 4433
+    },
+    {
+      "epoch": 1.6978169283799311,
+      "loss": 1.832090973854065,
+      "loss_ce": 0.17394044995307922,
+      "loss_iou": 0.7552837133407593,
+      "loss_num": 0.029541015625,
+      "loss_xval": 1.6581504344940186,
+      "num_input_tokens_seen": 766308248,
+      "step": 4433
+    },
+    {
+      "epoch": 1.698199923400996,
+      "grad_norm": 109.41324524999341,
+      "learning_rate": 5e-06,
+      "loss": 1.8317,
+      "num_input_tokens_seen": 766481184,
+      "step": 4434
+    },
+    {
+      "epoch": 1.698199923400996,
+      "loss": 1.5660760402679443,
+      "loss_ce": 0.17505106329917908,
+      "loss_iou": 0.6377580165863037,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 1.3832430839538574,
+      "num_input_tokens_seen": 766481184,
+      "step": 4434
+    },
+    {
+      "epoch": 1.6985829184220607,
+      "grad_norm": 128.3683062068298,
+      "learning_rate": 5e-06,
+      "loss": 1.2891,
+      "num_input_tokens_seen": 766653712,
+      "step": 4435
+    },
+    {
+      "epoch": 1.6985829184220607,
+      "loss": 1.3084986209869385,
+      "loss_ce": 0.15213462710380554,
+      "loss_iou": 0.5399739742279053,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 1.1563639640808105,
+      "num_input_tokens_seen": 766653712,
+      "step": 4435
+    },
+    {
+      "epoch": 1.6989659134431252,
+      "grad_norm": 147.53416318551177,
+      "learning_rate": 5e-06,
+      "loss": 1.5314,
+      "num_input_tokens_seen": 766826512,
+      "step": 4436
+    },
+    {
+      "epoch": 1.6989659134431252,
+      "loss": 1.686354160308838,
+      "loss_ce": 0.16853588819503784,
+      "loss_iou": 0.6877726316452026,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 1.5178182125091553,
+      "num_input_tokens_seen": 766826512,
+      "step": 4436
+    },
+    {
+      "epoch": 1.69934890846419,
+      "grad_norm": 128.1561609852027,
+      "learning_rate": 5e-06,
+      "loss": 1.4629,
+      "num_input_tokens_seen": 766999496,
+      "step": 4437
+    },
+    {
+      "epoch": 1.69934890846419,
+      "loss": 1.5417002439498901,
+      "loss_ce": 0.16466131806373596,
+      "loss_iou": 0.6501741409301758,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 1.3770389556884766,
+      "num_input_tokens_seen": 766999496,
+      "step": 4437
+    },
+    {
+      "epoch": 1.6997319034852547,
+      "grad_norm": 147.28453253910888,
+      "learning_rate": 5e-06,
+      "loss": 1.6051,
+      "num_input_tokens_seen": 767172384,
+      "step": 4438
+    },
+    {
+      "epoch": 1.6997319034852547,
+      "loss": 1.5956776142120361,
+      "loss_ce": 0.1512976437807083,
+      "loss_iou": 0.6851721405982971,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 1.4443799257278442,
+      "num_input_tokens_seen": 767172384,
+      "step": 4438
+    },
+    {
+      "epoch": 1.7001148985063195,
+      "grad_norm": 127.73815537430345,
+      "learning_rate": 5e-06,
+      "loss": 1.8774,
+      "num_input_tokens_seen": 767345432,
+      "step": 4439
+    },
+    {
+      "epoch": 1.7001148985063195,
+      "loss": 1.9447025060653687,
+      "loss_ce": 0.19345439970493317,
+      "loss_iou": 0.7954543828964233,
+      "loss_num": 0.0322265625,
+      "loss_xval": 1.7512481212615967,
+      "num_input_tokens_seen": 767345432,
+      "step": 4439
+    },
+    {
+      "epoch": 1.700497893527384,
+      "grad_norm": 146.03481917365204,
+      "learning_rate": 5e-06,
+      "loss": 1.3528,
+      "num_input_tokens_seen": 767518664,
+      "step": 4440
+    },
+    {
+      "epoch": 1.700497893527384,
+      "loss": 1.2968599796295166,
+      "loss_ce": 0.15924659371376038,
+      "loss_iou": 0.5294238328933716,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 1.1376135349273682,
+      "num_input_tokens_seen": 767518664,
+      "step": 4440
+    },
+    {
+      "epoch": 1.7008808885484488,
+      "grad_norm": 174.71735126736093,
+      "learning_rate": 5e-06,
+      "loss": 1.4372,
+      "num_input_tokens_seen": 767691856,
+      "step": 4441
+    },
+    {
+      "epoch": 1.7008808885484488,
+      "loss": 1.5270781517028809,
+      "loss_ce": 0.17142990231513977,
+      "loss_iou": 0.627073347568512,
+      "loss_num": 0.020263671875,
+      "loss_xval": 1.3556482791900635,
+      "num_input_tokens_seen": 767691856,
+      "step": 4441
+    },
+    {
+      "epoch": 1.7012638835695135,
+      "grad_norm": 142.54225699355837,
+      "learning_rate": 5e-06,
+      "loss": 1.8282,
+      "num_input_tokens_seen": 767864664,
+      "step": 4442
+    },
+    {
+      "epoch": 1.7012638835695135,
+      "loss": 1.8250694274902344,
+      "loss_ce": 0.1725281924009323,
+      "loss_iou": 0.7562022805213928,
+      "loss_num": 0.028076171875,
+      "loss_xval": 1.6525412797927856,
+      "num_input_tokens_seen": 767864664,
+      "step": 4442
+    },
+    {
+      "epoch": 1.7016468785905783,
+      "grad_norm": 190.05889194422352,
+      "learning_rate": 5e-06,
+      "loss": 1.4428,
+      "num_input_tokens_seen": 768034504,
+      "step": 4443
+    },
+    {
+      "epoch": 1.7016468785905783,
+      "loss": 1.4877607822418213,
+      "loss_ce": 0.17869636416435242,
+      "loss_iou": 0.6047275066375732,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 1.3090643882751465,
+      "num_input_tokens_seen": 768034504,
+      "step": 4443
+    },
+    {
+      "epoch": 1.702029873611643,
+      "grad_norm": 220.0245345181745,
+      "learning_rate": 5e-06,
+      "loss": 1.4822,
+      "num_input_tokens_seen": 768207584,
+      "step": 4444
+    },
+    {
+      "epoch": 1.702029873611643,
+      "loss": 1.5720257759094238,
+      "loss_ce": 0.14049610495567322,
+      "loss_iou": 0.6731317043304443,
+      "loss_num": 0.01708984375,
+      "loss_xval": 1.4315295219421387,
+      "num_input_tokens_seen": 768207584,
+      "step": 4444
+    },
+    {
+      "epoch": 1.7024128686327078,
+      "grad_norm": 163.44446468669128,
+      "learning_rate": 5e-06,
+      "loss": 1.454,
+      "num_input_tokens_seen": 768380576,
+      "step": 4445
+    },
+    {
+      "epoch": 1.7024128686327078,
+      "loss": 1.4086887836456299,
+      "loss_ce": 0.16745227575302124,
+      "loss_iou": 0.5547307729721069,
+      "loss_num": 0.0263671875,
+      "loss_xval": 1.2412365674972534,
+      "num_input_tokens_seen": 768380576,
+      "step": 4445
+    },
+    {
+      "epoch": 1.7027958636537726,
+      "grad_norm": 177.0732586462458,
+      "learning_rate": 5e-06,
+      "loss": 1.5041,
+      "num_input_tokens_seen": 768553848,
+      "step": 4446
+    },
+    {
+      "epoch": 1.7027958636537726,
+      "loss": 1.3350529670715332,
+      "loss_ce": 0.16157157719135284,
+      "loss_iou": 0.5390570163726807,
+      "loss_num": 0.01904296875,
+      "loss_xval": 1.1734814643859863,
+      "num_input_tokens_seen": 768553848,
+      "step": 4446
+    },
+    {
+      "epoch": 1.7031788586748373,
+      "grad_norm": 324.704177597589,
+      "learning_rate": 5e-06,
+      "loss": 1.5901,
+      "num_input_tokens_seen": 768727168,
+      "step": 4447
+    },
+    {
+      "epoch": 1.7031788586748373,
+      "loss": 1.524158239364624,
+      "loss_ce": 0.19488538801670074,
+      "loss_iou": 0.617181658744812,
+      "loss_num": 0.01904296875,
+      "loss_xval": 1.329272985458374,
+      "num_input_tokens_seen": 768727168,
+      "step": 4447
+    },
+    {
+      "epoch": 1.703561853695902,
+      "grad_norm": 149.51563464904,
+      "learning_rate": 5e-06,
+      "loss": 1.3408,
+      "num_input_tokens_seen": 768900200,
+      "step": 4448
+    },
+    {
+      "epoch": 1.703561853695902,
+      "loss": 1.549727439880371,
+      "loss_ce": 0.19345209002494812,
+      "loss_iou": 0.631743311882019,
+      "loss_num": 0.0185546875,
+      "loss_xval": 1.3562753200531006,
+      "num_input_tokens_seen": 768900200,
+      "step": 4448
+    },
+    {
+      "epoch": 1.7039448487169668,
+      "grad_norm": 173.92162894221724,
+      "learning_rate": 5e-06,
+      "loss": 1.3456,
+      "num_input_tokens_seen": 769073760,
+      "step": 4449
+    },
+    {
+      "epoch": 1.7039448487169668,
+      "loss": 1.4524028301239014,
+      "loss_ce": 0.2201455682516098,
+      "loss_iou": 0.5847184062004089,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 1.2322572469711304,
+      "num_input_tokens_seen": 769073760,
+      "step": 4449
+    },
+    {
+      "epoch": 1.7043278437380314,
+      "grad_norm": 111.37129237785055,
+      "learning_rate": 5e-06,
+      "loss": 2.125,
+      "num_input_tokens_seen": 769246672,
+      "step": 4450
+    },
+    {
+      "epoch": 1.7043278437380314,
+      "loss": 2.1455016136169434,
+      "loss_ce": 0.19774608314037323,
+      "loss_iou": 0.9000557661056519,
+      "loss_num": 0.029541015625,
+      "loss_xval": 1.9477556943893433,
+      "num_input_tokens_seen": 769246672,
+      "step": 4450
+    },
+    {
+      "epoch": 1.7047108387590961,
+      "grad_norm": 119.46190874424306,
+      "learning_rate": 5e-06,
+      "loss": 1.5371,
+      "num_input_tokens_seen": 769419440,
+      "step": 4451
+    },
+    {
+      "epoch": 1.7047108387590961,
+      "loss": 1.6254570484161377,
+      "loss_ce": 0.20357799530029297,
+      "loss_iou": 0.6606770753860474,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 1.4218790531158447,
+      "num_input_tokens_seen": 769419440,
+      "step": 4451
+    },
+    {
+      "epoch": 1.7050938337801609,
+      "grad_norm": 178.46957738083788,
+      "learning_rate": 5e-06,
+      "loss": 1.4418,
+      "num_input_tokens_seen": 769592536,
+      "step": 4452
+    },
+    {
+      "epoch": 1.7050938337801609,
+      "loss": 1.4509649276733398,
+      "loss_ce": 0.20302559435367584,
+      "loss_iou": 0.5866924524307251,
+      "loss_num": 0.014892578125,
+      "loss_xval": 1.2479393482208252,
+      "num_input_tokens_seen": 769592536,
+      "step": 4452
+    },
+    {
+      "epoch": 1.7054768288012254,
+      "grad_norm": 122.64579033541503,
+      "learning_rate": 5e-06,
+      "loss": 1.6957,
+      "num_input_tokens_seen": 769765208,
+      "step": 4453
+    },
+    {
+      "epoch": 1.7054768288012254,
+      "loss": 1.6926445960998535,
+      "loss_ce": 0.16972097754478455,
+      "loss_iou": 0.7031121253967285,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 1.522923469543457,
+      "num_input_tokens_seen": 769765208,
+      "step": 4453
+    },
+    {
+      "epoch": 1.7058598238222902,
+      "grad_norm": 117.43935994232162,
+      "learning_rate": 5e-06,
+      "loss": 1.2035,
+      "num_input_tokens_seen": 769938488,
+      "step": 4454
+    },
+    {
+      "epoch": 1.7058598238222902,
+      "loss": 1.2287380695343018,
+      "loss_ce": 0.17198912799358368,
+      "loss_iou": 0.49802476167678833,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 1.0567489862442017,
+      "num_input_tokens_seen": 769938488,
+      "step": 4454
+    },
+    {
+      "epoch": 1.706242818843355,
+      "grad_norm": 171.4136323693914,
+      "learning_rate": 5e-06,
+      "loss": 1.4278,
+      "num_input_tokens_seen": 770111528,
+      "step": 4455
+    },
+    {
+      "epoch": 1.706242818843355,
+      "loss": 1.3430722951889038,
+      "loss_ce": 0.17970164120197296,
+      "loss_iou": 0.5561878681182861,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 1.1633706092834473,
+      "num_input_tokens_seen": 770111528,
+      "step": 4455
+    },
+    {
+      "epoch": 1.7066258138644197,
+      "grad_norm": 112.23669738981084,
+      "learning_rate": 5e-06,
+      "loss": 1.3412,
+      "num_input_tokens_seen": 770284456,
+      "step": 4456
+    },
+    {
+      "epoch": 1.7066258138644197,
+      "loss": 1.4289932250976562,
+      "loss_ce": 0.1448153406381607,
+      "loss_iou": 0.5987387299537659,
+      "loss_num": 0.017333984375,
+      "loss_xval": 1.2841778993606567,
+      "num_input_tokens_seen": 770284456,
+      "step": 4456
+    },
+    {
+      "epoch": 1.7070088088854845,
+      "grad_norm": 159.1568544580478,
+      "learning_rate": 5e-06,
+      "loss": 1.6756,
+      "num_input_tokens_seen": 770457600,
+      "step": 4457
+    },
+    {
+      "epoch": 1.7070088088854845,
+      "loss": 1.526301383972168,
+      "loss_ce": 0.15513348579406738,
+      "loss_iou": 0.6388309597969055,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 1.371167778968811,
+      "num_input_tokens_seen": 770457600,
+      "step": 4457
+    },
+    {
+      "epoch": 1.7073918039065492,
+      "grad_norm": 119.78269604509411,
+      "learning_rate": 5e-06,
+      "loss": 1.3578,
+      "num_input_tokens_seen": 770630944,
+      "step": 4458
+    },
+    {
+      "epoch": 1.7073918039065492,
+      "loss": 1.4750096797943115,
+      "loss_ce": 0.20000287890434265,
+      "loss_iou": 0.6068332195281982,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 1.2750067710876465,
+      "num_input_tokens_seen": 770630944,
+      "step": 4458
+    },
+    {
+      "epoch": 1.707774798927614,
+      "grad_norm": 174.9362652954177,
+      "learning_rate": 5e-06,
+      "loss": 2.8323,
+      "num_input_tokens_seen": 770803792,
+      "step": 4459
+    },
+    {
+      "epoch": 1.707774798927614,
+      "loss": 3.0283281803131104,
+      "loss_ce": 0.21793930232524872,
+      "loss_iou": 1.3216679096221924,
+      "loss_num": 0.033447265625,
+      "loss_xval": 2.8103890419006348,
+      "num_input_tokens_seen": 770803792,
+      "step": 4459
+    },
+    {
+      "epoch": 1.7081577939486787,
+      "grad_norm": 217.87076352949362,
+      "learning_rate": 5e-06,
+      "loss": 1.7368,
+      "num_input_tokens_seen": 770976568,
+      "step": 4460
+    },
+    {
+      "epoch": 1.7081577939486787,
+      "loss": 1.71844482421875,
+      "loss_ce": 0.1270090639591217,
+      "loss_iou": 0.7199426889419556,
+      "loss_num": 0.0302734375,
+      "loss_xval": 1.5914356708526611,
+      "num_input_tokens_seen": 770976568,
+      "step": 4460
+    },
+    {
+      "epoch": 1.7085407889697435,
+      "grad_norm": 206.31400409004686,
+      "learning_rate": 5e-06,
+      "loss": 1.8282,
+      "num_input_tokens_seen": 771149464,
+      "step": 4461
+    },
+    {
+      "epoch": 1.7085407889697435,
+      "loss": 1.7612568140029907,
+      "loss_ce": 0.13727818429470062,
+      "loss_iou": 0.774467945098877,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 1.623978614807129,
+      "num_input_tokens_seen": 771149464,
+      "step": 4461
+    },
+    {
+      "epoch": 1.7089237839908082,
+      "grad_norm": 75.67437294320627,
+      "learning_rate": 5e-06,
+      "loss": 3.3843,
+      "num_input_tokens_seen": 771322536,
+      "step": 4462
+    },
+    {
+      "epoch": 1.7089237839908082,
+      "loss": 3.3501906394958496,
+      "loss_ce": 0.1202506422996521,
+      "loss_iou": 1.4440104961395264,
+      "loss_num": 0.068359375,
+      "loss_xval": 3.2066245079040527,
+      "num_input_tokens_seen": 771322536,
+      "step": 4462
+    },
+    {
+      "epoch": 1.709306779011873,
+      "grad_norm": 51.130343034001655,
+      "learning_rate": 5e-06,
+      "loss": 2.3109,
+      "num_input_tokens_seen": 771495240,
+      "step": 4463
+    },
+    {
+      "epoch": 1.709306779011873,
+      "loss": 2.279101848602295,
+      "loss_ce": 0.1349378526210785,
+      "loss_iou": 0.9278560280799866,
+      "loss_num": 0.0576171875,
+      "loss_xval": 2.1441640853881836,
+      "num_input_tokens_seen": 771495240,
+      "step": 4463
+    },
+    {
+      "epoch": 1.7096897740329375,
+      "grad_norm": 207.40094931787996,
+      "learning_rate": 5e-06,
+      "loss": 2.0242,
+      "num_input_tokens_seen": 771668448,
+      "step": 4464
+    },
+    {
+      "epoch": 1.7096897740329375,
+      "loss": 2.151801586151123,
+      "loss_ce": 0.12937594950199127,
+      "loss_iou": 0.9341254830360413,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 2.022425889968872,
+      "num_input_tokens_seen": 771668448,
+      "step": 4464
+    },
+    {
+      "epoch": 1.7100727690540023,
+      "grad_norm": 31.24866368096372,
+      "learning_rate": 5e-06,
+      "loss": 3.8595,
+      "num_input_tokens_seen": 771841696,
+      "step": 4465
+    },
+    {
+      "epoch": 1.7100727690540023,
+      "loss": 3.599266767501831,
+      "loss_ce": 0.1363958716392517,
+      "loss_iou": 1.5269067287445068,
+      "loss_num": 0.08154296875,
+      "loss_xval": 3.4512743949890137,
+      "num_input_tokens_seen": 771841696,
+      "step": 4465
+    },
+    {
+      "epoch": 1.710455764075067,
+      "grad_norm": 69.05065205129283,
+      "learning_rate": 5e-06,
+      "loss": 3.5328,
+      "num_input_tokens_seen": 772014960,
+      "step": 4466
+    },
+    {
+      "epoch": 1.710455764075067,
+      "loss": 3.5418059825897217,
+      "loss_ce": 0.1020163744688034,
+      "loss_iou": 1.5677952766418457,
+      "loss_num": 0.060791015625,
+      "loss_xval": 3.4397897720336914,
+      "num_input_tokens_seen": 772014960,
+      "step": 4466
+    },
+    {
+      "epoch": 1.7108387590961316,
+      "grad_norm": 84.27304567971493,
+      "learning_rate": 5e-06,
+      "loss": 2.3251,
+      "num_input_tokens_seen": 772188136,
+      "step": 4467
+    },
+    {
+      "epoch": 1.7108387590961316,
+      "loss": 2.3423097133636475,
+      "loss_ce": 0.03949767351150513,
+      "loss_iou": 0.9586570262908936,
+      "loss_num": 0.0771484375,
+      "loss_xval": 2.302812099456787,
+      "num_input_tokens_seen": 772188136,
+      "step": 4467
+    },
+    {
+      "epoch": 1.7112217541171963,
+      "grad_norm": 82.79243727200686,
+      "learning_rate": 5e-06,
+      "loss": 1.6614,
+      "num_input_tokens_seen": 772360792,
+      "step": 4468
+    },
+    {
+      "epoch": 1.7112217541171963,
+      "loss": 1.734809398651123,
+      "loss_ce": 0.02447887510061264,
+      "loss_iou": 0.7800614833831787,
+      "loss_num": 0.030029296875,
+      "loss_xval": 1.7103304862976074,
+      "num_input_tokens_seen": 772360792,
+      "step": 4468
+    },
+    {
+      "epoch": 1.711604749138261,
+      "grad_norm": 186.4089718887717,
+      "learning_rate": 5e-06,
+      "loss": 1.9182,
+      "num_input_tokens_seen": 772533304,
+      "step": 4469
+    },
+    {
+      "epoch": 1.711604749138261,
+      "loss": 1.9668108224868774,
+      "loss_ce": 0.02520131692290306,
+      "loss_iou": 0.894724428653717,
+      "loss_num": 0.030517578125,
+      "loss_xval": 1.9416093826293945,
+      "num_input_tokens_seen": 772533304,
+      "step": 4469
+    },
+    {
+      "epoch": 1.7119877441593259,
+      "grad_norm": 110.13222302725462,
+      "learning_rate": 5e-06,
+      "loss": 2.0086,
+      "num_input_tokens_seen": 772706024,
+      "step": 4470
+    },
+    {
+      "epoch": 1.7119877441593259,
+      "loss": 2.069946527481079,
+      "loss_ce": 0.0550527423620224,
+      "loss_iou": 0.896729052066803,
+      "loss_num": 0.044189453125,
+      "loss_xval": 2.0148935317993164,
+      "num_input_tokens_seen": 772706024,
+      "step": 4470
+    },
+    {
+      "epoch": 1.7123707391803906,
+      "grad_norm": 181.69641199426349,
+      "learning_rate": 5e-06,
+      "loss": 2.0872,
+      "num_input_tokens_seen": 772879144,
+      "step": 4471
+    },
+    {
+      "epoch": 1.7123707391803906,
+      "loss": 1.9669781923294067,
+      "loss_ce": 0.03427477926015854,
+      "loss_iou": 0.8851749300956726,
+      "loss_num": 0.032470703125,
+      "loss_xval": 1.9327033758163452,
+      "num_input_tokens_seen": 772879144,
+      "step": 4471
+    },
+    {
+      "epoch": 1.7127537342014554,
+      "grad_norm": 114.76042625727442,
+      "learning_rate": 5e-06,
+      "loss": 2.89,
+      "num_input_tokens_seen": 773052304,
+      "step": 4472
+    },
+    {
+      "epoch": 1.7127537342014554,
+      "loss": 2.91530704498291,
+      "loss_ce": 0.0744759738445282,
+      "loss_iou": 1.0402885675430298,
+      "loss_num": 0.15234375,
+      "loss_xval": 2.8408310413360596,
+      "num_input_tokens_seen": 773052304,
+      "step": 4472
+    },
+    {
+      "epoch": 1.7131367292225201,
+      "grad_norm": 42.576118664516486,
+      "learning_rate": 5e-06,
+      "loss": 2.3271,
+      "num_input_tokens_seen": 773224992,
+      "step": 4473
+    },
+    {
+      "epoch": 1.7131367292225201,
+      "loss": 2.4316883087158203,
+      "loss_ce": 0.0571332685649395,
+      "loss_iou": 1.0326144695281982,
+      "loss_num": 0.061767578125,
+      "loss_xval": 2.3745551109313965,
+      "num_input_tokens_seen": 773224992,
+      "step": 4473
+    },
+    {
+      "epoch": 1.713519724243585,
+      "grad_norm": 93.32687384805274,
+      "learning_rate": 5e-06,
+      "loss": 1.6449,
+      "num_input_tokens_seen": 773397960,
+      "step": 4474
+    },
+    {
+      "epoch": 1.713519724243585,
+      "loss": 1.6859182119369507,
+      "loss_ce": 0.07083331048488617,
+      "loss_iou": 0.7422043085098267,
+      "loss_num": 0.026123046875,
+      "loss_xval": 1.6150848865509033,
+      "num_input_tokens_seen": 773397960,
+      "step": 4474
+    },
+    {
+      "epoch": 1.7139027192646497,
+      "grad_norm": 192.48370805098128,
+      "learning_rate": 5e-06,
+      "loss": 3.3229,
+      "num_input_tokens_seen": 773570760,
+      "step": 4475
+    },
+    {
+      "epoch": 1.7139027192646497,
+      "loss": 3.3088417053222656,
+      "loss_ce": 0.07787613570690155,
+      "loss_iou": 1.5147746801376343,
+      "loss_num": 0.040283203125,
+      "loss_xval": 3.2309653759002686,
+      "num_input_tokens_seen": 773570760,
+      "step": 4475
+    },
+    {
+      "epoch": 1.7142857142857144,
+      "grad_norm": 135.82201524103337,
+      "learning_rate": 5e-06,
+      "loss": 2.8948,
+      "num_input_tokens_seen": 773743496,
+      "step": 4476
+    },
+    {
+      "epoch": 1.7142857142857144,
+      "loss": 2.903146266937256,
+      "loss_ce": 0.08123888820409775,
+      "loss_iou": 1.2310221195220947,
+      "loss_num": 0.072265625,
+      "loss_xval": 2.8219075202941895,
+      "num_input_tokens_seen": 773743496,
+      "step": 4476
+    },
+    {
+      "epoch": 1.7146687093067792,
+      "grad_norm": 228.59322919300953,
+      "learning_rate": 5e-06,
+      "loss": 2.7464,
+      "num_input_tokens_seen": 773916120,
+      "step": 4477
+    },
+    {
+      "epoch": 1.7146687093067792,
+      "loss": 2.861143112182617,
+      "loss_ce": 0.059802476316690445,
+      "loss_iou": 0.9922231435775757,
+      "loss_num": 0.1630859375,
+      "loss_xval": 2.8013405799865723,
+      "num_input_tokens_seen": 773916120,
+      "step": 4477
+    },
+    {
+      "epoch": 1.7150517043278437,
+      "grad_norm": 116.40818586712737,
+      "learning_rate": 5e-06,
+      "loss": 1.8228,
+      "num_input_tokens_seen": 774089200,
+      "step": 4478
+    },
+    {
+      "epoch": 1.7150517043278437,
+      "loss": 1.9288240671157837,
+      "loss_ce": 0.08956010639667511,
+      "loss_iou": 0.8184051513671875,
+      "loss_num": 0.04052734375,
+      "loss_xval": 1.839263916015625,
+      "num_input_tokens_seen": 774089200,
+      "step": 4478
+    },
+    {
+      "epoch": 1.7154346993489085,
+      "grad_norm": 230.81461555684135,
+      "learning_rate": 5e-06,
+      "loss": 2.1421,
+      "num_input_tokens_seen": 774262152,
+      "step": 4479
+    },
+    {
+      "epoch": 1.7154346993489085,
+      "loss": 2.154937267303467,
+      "loss_ce": 0.08769381791353226,
+      "loss_iou": 0.9814976453781128,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 2.0672433376312256,
+      "num_input_tokens_seen": 774262152,
+      "step": 4479
+    },
+    {
+      "epoch": 1.7158176943699732,
+      "grad_norm": 148.6980282717763,
+      "learning_rate": 5e-06,
+      "loss": 2.4154,
+      "num_input_tokens_seen": 774435080,
+      "step": 4480
+    },
+    {
+      "epoch": 1.7158176943699732,
+      "loss": 2.1077704429626465,
+      "loss_ce": 0.0960702896118164,
+      "loss_iou": 0.908865213394165,
+      "loss_num": 0.038818359375,
+      "loss_xval": 2.01170015335083,
+      "num_input_tokens_seen": 774435080,
+      "step": 4480
+    },
+    {
+      "epoch": 1.7162006893910378,
+      "grad_norm": 252.26938985562964,
+      "learning_rate": 5e-06,
+      "loss": 2.7989,
+      "num_input_tokens_seen": 774608088,
+      "step": 4481
+    },
+    {
+      "epoch": 1.7162006893910378,
+      "loss": 3.045273542404175,
+      "loss_ce": 0.06512296199798584,
+      "loss_iou": 1.2238399982452393,
+      "loss_num": 0.1064453125,
+      "loss_xval": 2.9801506996154785,
+      "num_input_tokens_seen": 774608088,
+      "step": 4481
+    },
+    {
+      "epoch": 1.7165836844121025,
+      "grad_norm": 85.5479843529618,
+      "learning_rate": 5e-06,
+      "loss": 2.2,
+      "num_input_tokens_seen": 774781128,
+      "step": 4482
+    },
+    {
+      "epoch": 1.7165836844121025,
+      "loss": 2.168067693710327,
+      "loss_ce": 0.13942614197731018,
+      "loss_iou": 0.8768696784973145,
+      "loss_num": 0.054931640625,
+      "loss_xval": 2.028641700744629,
+      "num_input_tokens_seen": 774781128,
+      "step": 4482
+    },
+    {
+      "epoch": 1.7169666794331673,
+      "grad_norm": 74.43292598569569,
+      "learning_rate": 5e-06,
+      "loss": 1.8072,
+      "num_input_tokens_seen": 774954192,
+      "step": 4483
+    },
+    {
+      "epoch": 1.7169666794331673,
+      "loss": 1.8323304653167725,
+      "loss_ce": 0.11497731506824493,
+      "loss_iou": 0.796664834022522,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 1.717353105545044,
+      "num_input_tokens_seen": 774954192,
+      "step": 4483
+    },
+    {
+      "epoch": 1.717349674454232,
+      "grad_norm": 105.396020354874,
+      "learning_rate": 5e-06,
+      "loss": 1.5868,
+      "num_input_tokens_seen": 775127248,
+      "step": 4484
+    },
+    {
+      "epoch": 1.717349674454232,
+      "loss": 1.6237494945526123,
+      "loss_ce": 0.0984693318605423,
+      "loss_iou": 0.7000790238380432,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 1.5252801179885864,
+      "num_input_tokens_seen": 775127248,
+      "step": 4484
+    },
+    {
+      "epoch": 1.7177326694752968,
+      "grad_norm": 143.44924050936237,
+      "learning_rate": 5e-06,
+      "loss": 2.091,
+      "num_input_tokens_seen": 775300320,
+      "step": 4485
+    },
+    {
+      "epoch": 1.7177326694752968,
+      "loss": 2.083596706390381,
+      "loss_ce": 0.09264512360095978,
+      "loss_iou": 0.9233322143554688,
+      "loss_num": 0.02880859375,
+      "loss_xval": 1.9909515380859375,
+      "num_input_tokens_seen": 775300320,
+      "step": 4485
+    },
+    {
+      "epoch": 1.7181156644963615,
+      "grad_norm": 98.5734193934433,
+      "learning_rate": 5e-06,
+      "loss": 1.6979,
+      "num_input_tokens_seen": 775473160,
+      "step": 4486
+    },
+    {
+      "epoch": 1.7181156644963615,
+      "loss": 1.7515714168548584,
+      "loss_ce": 0.06850959360599518,
+      "loss_iou": 0.7677699327468872,
+      "loss_num": 0.029541015625,
+      "loss_xval": 1.6830618381500244,
+      "num_input_tokens_seen": 775473160,
+      "step": 4486
+    },
+    {
+      "epoch": 1.7184986595174263,
+      "grad_norm": 134.16194814590582,
+      "learning_rate": 5e-06,
+      "loss": 2.0556,
+      "num_input_tokens_seen": 775646392,
+      "step": 4487
+    },
+    {
+      "epoch": 1.7184986595174263,
+      "loss": 2.051379919052124,
+      "loss_ce": 0.0793566182255745,
+      "loss_iou": 0.9005929231643677,
+      "loss_num": 0.0341796875,
+      "loss_xval": 1.9720232486724854,
+      "num_input_tokens_seen": 775646392,
+      "step": 4487
+    },
+    {
+      "epoch": 1.718881654538491,
+      "grad_norm": 101.5943321708956,
+      "learning_rate": 5e-06,
+      "loss": 2.0247,
+      "num_input_tokens_seen": 775818960,
+      "step": 4488
+    },
+    {
+      "epoch": 1.718881654538491,
+      "loss": 2.083380699157715,
+      "loss_ce": 0.08968771994113922,
+      "loss_iou": 0.8956502676010132,
+      "loss_num": 0.04052734375,
+      "loss_xval": 1.9936931133270264,
+      "num_input_tokens_seen": 775818960,
+      "step": 4488
+    },
+    {
+      "epoch": 1.7192646495595558,
+      "grad_norm": 73.59504718495192,
+      "learning_rate": 5e-06,
+      "loss": 1.5223,
+      "num_input_tokens_seen": 775992016,
+      "step": 4489
+    },
+    {
+      "epoch": 1.7192646495595558,
+      "loss": 1.4159590005874634,
+      "loss_ce": 0.08734661340713501,
+      "loss_iou": 0.5951838493347168,
+      "loss_num": 0.027587890625,
+      "loss_xval": 1.3286123275756836,
+      "num_input_tokens_seen": 775992016,
+      "step": 4489
+    },
+    {
+      "epoch": 1.7196476445806206,
+      "grad_norm": 106.59675438874885,
+      "learning_rate": 5e-06,
+      "loss": 1.6202,
+      "num_input_tokens_seen": 776164744,
+      "step": 4490
+    },
+    {
+      "epoch": 1.7196476445806206,
+      "loss": 1.59060537815094,
+      "loss_ce": 0.09607075154781342,
+      "loss_iou": 0.6930070519447327,
+      "loss_num": 0.021728515625,
+      "loss_xval": 1.4945346117019653,
+      "num_input_tokens_seen": 776164744,
+      "step": 4490
+    },
+    {
+      "epoch": 1.7200306396016853,
+      "grad_norm": 139.04872605535303,
+      "learning_rate": 5e-06,
+      "loss": 2.0484,
+      "num_input_tokens_seen": 776337136,
+      "step": 4491
+    },
+    {
+      "epoch": 1.7200306396016853,
+      "loss": 2.171727180480957,
+      "loss_ce": 0.10225944966077805,
+      "loss_iou": 0.9552661180496216,
+      "loss_num": 0.03173828125,
+      "loss_xval": 2.069467782974243,
+      "num_input_tokens_seen": 776337136,
+      "step": 4491
+    },
+    {
+      "epoch": 1.7204136346227499,
+      "grad_norm": 122.62049336158304,
+      "learning_rate": 5e-06,
+      "loss": 1.547,
+      "num_input_tokens_seen": 776510008,
+      "step": 4492
+    },
+    {
+      "epoch": 1.7204136346227499,
+      "loss": 1.5319805145263672,
+      "loss_ce": 0.11845797300338745,
+      "loss_iou": 0.6547898054122925,
+      "loss_num": 0.020751953125,
+      "loss_xval": 1.413522481918335,
+      "num_input_tokens_seen": 776510008,
+      "step": 4492
+    },
+    {
+      "epoch": 1.7207966296438146,
+      "grad_norm": 231.88840384635802,
+      "learning_rate": 5e-06,
+      "loss": 1.7515,
+      "num_input_tokens_seen": 776682808,
+      "step": 4493
+    },
+    {
+      "epoch": 1.7207966296438146,
+      "loss": 1.772214651107788,
+      "loss_ce": 0.08848626911640167,
+      "loss_iou": 0.7935243844985962,
+      "loss_num": 0.019287109375,
+      "loss_xval": 1.6837284564971924,
+      "num_input_tokens_seen": 776682808,
+      "step": 4493
+    },
+    {
+      "epoch": 1.7211796246648794,
+      "grad_norm": 160.42903791757934,
+      "learning_rate": 5e-06,
+      "loss": 2.0512,
+      "num_input_tokens_seen": 776856032,
+      "step": 4494
+    },
+    {
+      "epoch": 1.7211796246648794,
+      "loss": 2.05763840675354,
+      "loss_ce": 0.08359895646572113,
+      "loss_iou": 0.9068195819854736,
+      "loss_num": 0.0322265625,
+      "loss_xval": 1.9740395545959473,
+      "num_input_tokens_seen": 776856032,
+      "step": 4494
+    },
+    {
+      "epoch": 1.721562619685944,
+      "grad_norm": 261.3765985052225,
+      "learning_rate": 5e-06,
+      "loss": 2.5035,
+      "num_input_tokens_seen": 777028608,
+      "step": 4495
+    },
+    {
+      "epoch": 1.721562619685944,
+      "loss": 2.637826681137085,
+      "loss_ce": 0.07435683161020279,
+      "loss_iou": 1.1283535957336426,
+      "loss_num": 0.061279296875,
+      "loss_xval": 2.563469886779785,
+      "num_input_tokens_seen": 777028608,
+      "step": 4495
+    },
+    {
+      "epoch": 1.7219456147070087,
+      "grad_norm": 105.80215052792512,
+      "learning_rate": 5e-06,
+      "loss": 1.8306,
+      "num_input_tokens_seen": 777197728,
+      "step": 4496
+    },
+    {
+      "epoch": 1.7219456147070087,
+      "loss": 1.7702192068099976,
+      "loss_ce": 0.10497844219207764,
+      "loss_iou": 0.757333517074585,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 1.66524076461792,
+      "num_input_tokens_seen": 777197728,
+      "step": 4496
+    },
+    {
+      "epoch": 1.7223286097280734,
+      "grad_norm": 212.14864221149261,
+      "learning_rate": 5e-06,
+      "loss": 1.6199,
+      "num_input_tokens_seen": 777370368,
+      "step": 4497
+    },
+    {
+      "epoch": 1.7223286097280734,
+      "loss": 1.4185817241668701,
+      "loss_ce": 0.07815849035978317,
+      "loss_iou": 0.6156766414642334,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 1.3404231071472168,
+      "num_input_tokens_seen": 777370368,
+      "step": 4497
+    },
+    {
+      "epoch": 1.7227116047491382,
+      "grad_norm": 96.49319602480695,
+      "learning_rate": 5e-06,
+      "loss": 1.8624,
+      "num_input_tokens_seen": 777543328,
+      "step": 4498
+    },
+    {
+      "epoch": 1.7227116047491382,
+      "loss": 1.8736438751220703,
+      "loss_ce": 0.11589843779802322,
+      "loss_iou": 0.8065766096115112,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 1.7577455043792725,
+      "num_input_tokens_seen": 777543328,
+      "step": 4498
+    },
+    {
+      "epoch": 1.723094599770203,
+      "grad_norm": 97.74607710610101,
+      "learning_rate": 5e-06,
+      "loss": 1.5515,
+      "num_input_tokens_seen": 777716200,
+      "step": 4499
+    },
+    {
+      "epoch": 1.723094599770203,
+      "loss": 1.6198670864105225,
+      "loss_ce": 0.10625797510147095,
+      "loss_iou": 0.6743460893630981,
+      "loss_num": 0.032958984375,
+      "loss_xval": 1.5136091709136963,
+      "num_input_tokens_seen": 777716200,
+      "step": 4499
+    },
+    {
+      "epoch": 1.7234775947912677,
+      "grad_norm": 115.47292848843725,
+      "learning_rate": 5e-06,
+      "loss": 1.3939,
+      "num_input_tokens_seen": 777889224,
+      "step": 4500
+    },
+    {
+      "epoch": 1.7234775947912677,
+      "eval_websight_new_CIoU": 0.8651993870735168,
+      "eval_websight_new_GIoU": 0.8643690347671509,
+      "eval_websight_new_IoU": 0.866310328245163,
+      "eval_websight_new_MAE_all": 0.014750218950212002,
+      "eval_websight_new_MAE_h": 0.01531203230842948,
+      "eval_websight_new_MAE_w": 0.01181055186316371,
+      "eval_websight_new_MAE_x": 0.014414471574127674,
+      "eval_websight_new_MAE_y": 0.01746381539851427,
+      "eval_websight_new_NUM_probability": 0.42599648237228394,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 1.4053547382354736,
+      "eval_websight_new_loss_ce": 0.15365064144134521,
+      "eval_websight_new_loss_iou": 0.6049605309963226,
+      "eval_websight_new_loss_num": 0.012073516845703125,
+      "eval_websight_new_loss_xval": 1.2702237963676453,
+      "eval_websight_new_runtime": 56.6952,
+      "eval_websight_new_samples_per_second": 0.882,
+      "eval_websight_new_steps_per_second": 0.035,
+      "num_input_tokens_seen": 777889224,
+      "step": 4500
+    },
+    {
+      "epoch": 1.7234775947912677,
+      "eval_seeclick_CIoU": 0.6288062632083893,
+      "eval_seeclick_GIoU": 0.6406533718109131,
+      "eval_seeclick_IoU": 0.6674403548240662,
+      "eval_seeclick_MAE_all": 0.06479492411017418,
+      "eval_seeclick_MAE_h": 0.027880861423909664,
+      "eval_seeclick_MAE_w": 0.10141636058688164,
+      "eval_seeclick_MAE_x": 0.09699774533510208,
+      "eval_seeclick_MAE_y": 0.03288473468273878,
+      "eval_seeclick_NUM_probability": 0.5766470730304718,
+      "eval_seeclick_inside_bbox": 0.8541666567325592,
+      "eval_seeclick_loss": 1.6871815919876099,
+      "eval_seeclick_loss_ce": 0.13062653690576553,
+      "eval_seeclick_loss_iou": 0.6280854344367981,
+      "eval_seeclick_loss_num": 0.04927825927734375,
+      "eval_seeclick_loss_xval": 1.5024629831314087,
+      "eval_seeclick_runtime": 88.2076,
+      "eval_seeclick_samples_per_second": 0.567,
+      "eval_seeclick_steps_per_second": 0.023,
+      "num_input_tokens_seen": 777889224,
+      "step": 4500
+    },
+    {
+      "epoch": 1.7234775947912677,
+      "eval_icons_CIoU": 0.7683820724487305,
+      "eval_icons_GIoU": 0.7657085359096527,
+      "eval_icons_IoU": 0.7841051816940308,
+      "eval_icons_MAE_all": 0.03661998547613621,
+      "eval_icons_MAE_h": 0.026071634143590927,
+      "eval_icons_MAE_w": 0.04999036341905594,
+      "eval_icons_MAE_x": 0.048626719042658806,
+      "eval_icons_MAE_y": 0.021791222505271435,
+      "eval_icons_NUM_probability": 0.40404000878334045,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 2.2493743896484375,
+      "eval_icons_loss_ce": 0.23850609362125397,
+      "eval_icons_loss_iou": 0.9390899240970612,
+      "eval_icons_loss_num": 0.0259246826171875,
+      "eval_icons_loss_xval": 2.0077268481254578,
+      "eval_icons_runtime": 87.9887,
+      "eval_icons_samples_per_second": 0.568,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 777889224,
+      "step": 4500
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 7833,
+  "num_input_tokens_seen": 777889224,
+  "num_train_epochs": 3,
+  "save_steps": 250,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4869346964406272.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}