diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,16332 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.6484149855907781,
+  "eval_steps": 250,
+  "global_step": 2250,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00028818443804034583,
+      "grad_norm": 342.1254510360843,
+      "learning_rate": 0.0,
+      "loss": 5.5528,
+      "step": 1
+    },
+    {
+      "epoch": 0.00028818443804034583,
+      "eval_websight_new_IoU": 0.0,
+      "eval_websight_new_MAE_x": 683.8400268554688,
+      "eval_websight_new_MAE_y": 817.5599975585938,
+      "eval_websight_new_coord_probability": 0.0006622532964684069,
+      "eval_websight_new_correct_token_probability": 3.301670119526534e-07,
+      "eval_websight_new_inside_bbox": 0.0,
+      "eval_websight_new_loss": 5.271888256072998,
+      "eval_websight_new_range_100_probability": 6.324639252852648e-05,
+      "eval_websight_new_range_20_probability": 1.297366907238029e-05,
+      "eval_websight_new_range_5_probability": 3.246676669732551e-06,
+      "eval_websight_new_runtime": 34.3296,
+      "eval_websight_new_samples_per_second": 1.456,
+      "eval_websight_new_steps_per_second": 0.204,
+      "step": 1
+    },
+    {
+      "epoch": 0.00028818443804034583,
+      "eval_seeclick_IoU": 0.029257936403155327,
+      "eval_seeclick_MAE_x": 728.1400146484375,
+      "eval_seeclick_MAE_y": 768.47998046875,
+      "eval_seeclick_coord_probability": 0.0003137658059131354,
+      "eval_seeclick_correct_token_probability": 1.5348230419931497e-07,
+      "eval_seeclick_inside_bbox": 0.10000000149011612,
+      "eval_seeclick_loss": 6.226981163024902,
+      "eval_seeclick_range_100_probability": 3.0610663088737056e-05,
+      "eval_seeclick_range_20_probability": 6.1479131545638666e-06,
+      "eval_seeclick_range_5_probability": 1.5359454437202658e-06,
+      "eval_seeclick_runtime": 43.0329,
+      "eval_seeclick_samples_per_second": 1.162,
+      "eval_seeclick_steps_per_second": 0.163,
+      "step": 1
+    },
+    {
+      "epoch": 0.00028818443804034583,
+      "eval_icons_IoU": 0.004958232864737511,
+      "eval_icons_MAE_x": 745.6400146484375,
+      "eval_icons_MAE_y": 752.8400268554688,
+      "eval_icons_coord_probability": 0.0001312530366703868,
+      "eval_icons_correct_token_probability": 6.412179232029303e-08,
+      "eval_icons_inside_bbox": 0.0,
+      "eval_icons_loss": 5.205017566680908,
+      "eval_icons_range_100_probability": 1.1899669516424183e-05,
+      "eval_icons_range_20_probability": 2.5593178634153446e-06,
+      "eval_icons_range_5_probability": 6.412645348063961e-07,
+      "eval_icons_runtime": 43.1355,
+      "eval_icons_samples_per_second": 1.159,
+      "eval_icons_steps_per_second": 0.162,
+      "step": 1
+    },
+    {
+      "epoch": 0.0005763688760806917,
+      "grad_norm": 374.14019160055807,
+      "learning_rate": 1.2553691692674561e-06,
+      "loss": 6.1136,
+      "step": 2
+    },
+    {
+      "epoch": 0.0008645533141210375,
+      "grad_norm": 347.4988078431115,
+      "learning_rate": 1.989713057850388e-06,
+      "loss": 5.0944,
+      "step": 3
+    },
+    {
+      "epoch": 0.0011527377521613833,
+      "grad_norm": 179.43459303960572,
+      "learning_rate": 2.5107383385349122e-06,
+      "loss": 5.0628,
+      "step": 4
+    },
+    {
+      "epoch": 0.001440922190201729,
+      "grad_norm": 264.3045842109189,
+      "learning_rate": 2.914876943577515e-06,
+      "loss": 5.98,
+      "step": 5
+    },
+    {
+      "epoch": 0.001729106628242075,
+      "grad_norm": 159.84065752881398,
+      "learning_rate": 3.245082227117844e-06,
+      "loss": 4.9271,
+      "step": 6
+    },
+    {
+      "epoch": 0.0020172910662824206,
+      "grad_norm": 93.20770627312804,
+      "learning_rate": 3.524266816342358e-06,
+      "loss": 4.8324,
+      "step": 7
+    },
+    {
+      "epoch": 0.0023054755043227667,
+      "grad_norm": 134.3055088466808,
+      "learning_rate": 3.7661075078023677e-06,
+      "loss": 5.6501,
+      "step": 8
+    },
+    {
+      "epoch": 0.0025936599423631124,
+      "grad_norm": 70.87239560354378,
+      "learning_rate": 3.979426115700776e-06,
+      "loss": 4.7434,
+      "step": 9
+    },
+    {
+      "epoch": 0.002881844380403458,
+      "grad_norm": 50.71713625909726,
+      "learning_rate": 4.170246112844972e-06,
+      "loss": 4.662,
+      "step": 10
+    },
+    {
+      "epoch": 0.003170028818443804,
+      "grad_norm": 57.917505846640466,
+      "learning_rate": 4.342863797226275e-06,
+      "loss": 5.3537,
+      "step": 11
+    },
+    {
+      "epoch": 0.00345821325648415,
+      "grad_norm": 42.88832778930361,
+      "learning_rate": 4.5004513963853e-06,
+      "loss": 4.4709,
+      "step": 12
+    },
+    {
+      "epoch": 0.0037463976945244955,
+      "grad_norm": 42.62371538537093,
+      "learning_rate": 4.645417934887083e-06,
+      "loss": 4.3543,
+      "step": 13
+    },
+    {
+      "epoch": 0.004034582132564841,
+      "grad_norm": 42.273228097799816,
+      "learning_rate": 4.779635985609814e-06,
+      "loss": 5.129,
+      "step": 14
+    },
+    {
+      "epoch": 0.004322766570605188,
+      "grad_norm": 41.22565825127231,
+      "learning_rate": 4.904590001427903e-06,
+      "loss": 4.2895,
+      "step": 15
+    },
+    {
+      "epoch": 0.004610951008645533,
+      "grad_norm": 36.4504773077764,
+      "learning_rate": 5.0214766770698244e-06,
+      "loss": 4.1565,
+      "step": 16
+    },
+    {
+      "epoch": 0.004899135446685879,
+      "grad_norm": 34.29091312887054,
+      "learning_rate": 5.131274831432035e-06,
+      "loss": 4.8424,
+      "step": 17
+    },
+    {
+      "epoch": 0.005187319884726225,
+      "grad_norm": 48.649272813076784,
+      "learning_rate": 5.234795284968231e-06,
+      "loss": 4.0544,
+      "step": 18
+    },
+    {
+      "epoch": 0.00547550432276657,
+      "grad_norm": 31.959874950729677,
+      "learning_rate": 5.3327172336600444e-06,
+      "loss": 3.8417,
+      "step": 19
+    },
+    {
+      "epoch": 0.005763688760806916,
+      "grad_norm": 28.61342603792043,
+      "learning_rate": 5.425615282112428e-06,
+      "loss": 4.437,
+      "step": 20
+    },
+    {
+      "epoch": 0.006051873198847263,
+      "grad_norm": 47.66734385668608,
+      "learning_rate": 5.513979874192746e-06,
+      "loss": 3.8685,
+      "step": 21
+    },
+    {
+      "epoch": 0.006340057636887608,
+      "grad_norm": 28.43431869602734,
+      "learning_rate": 5.598232966493732e-06,
+      "loss": 3.6345,
+      "step": 22
+    },
+    {
+      "epoch": 0.006628242074927954,
+      "grad_norm": 27.765318083952838,
+      "learning_rate": 5.6787402149051605e-06,
+      "loss": 4.1856,
+      "step": 23
+    },
+    {
+      "epoch": 0.0069164265129683,
+      "grad_norm": 51.311972962556034,
+      "learning_rate": 5.755820565652757e-06,
+      "loss": 3.6776,
+      "step": 24
+    },
+    {
+      "epoch": 0.007204610951008645,
+      "grad_norm": 30.939045377510407,
+      "learning_rate": 5.82975388715503e-06,
+      "loss": 3.4193,
+      "step": 25
+    },
+    {
+      "epoch": 0.007492795389048991,
+      "grad_norm": 25.204320422929406,
+      "learning_rate": 5.900787104154539e-06,
+      "loss": 3.9317,
+      "step": 26
+    },
+    {
+      "epoch": 0.0077809798270893375,
+      "grad_norm": 47.505367505045086,
+      "learning_rate": 5.9691391735511625e-06,
+      "loss": 3.427,
+      "step": 27
+    },
+    {
+      "epoch": 0.008069164265129682,
+      "grad_norm": 43.00244755896901,
+      "learning_rate": 6.03500515487727e-06,
+      "loss": 3.1939,
+      "step": 28
+    },
+    {
+      "epoch": 0.008357348703170028,
+      "grad_norm": 24.738277068462082,
+      "learning_rate": 6.0985595661703904e-06,
+      "loss": 3.6996,
+      "step": 29
+    },
+    {
+      "epoch": 0.008645533141210375,
+      "grad_norm": 27.13279303141791,
+      "learning_rate": 6.159959170695358e-06,
+      "loss": 3.2539,
+      "step": 30
+    },
+    {
+      "epoch": 0.008933717579250721,
+      "grad_norm": 30.518635329934547,
+      "learning_rate": 6.219345306558267e-06,
+      "loss": 3.0071,
+      "step": 31
+    },
+    {
+      "epoch": 0.009221902017291067,
+      "grad_norm": 25.330093895707904,
+      "learning_rate": 6.27684584633728e-06,
+      "loss": 3.4751,
+      "step": 32
+    },
+    {
+      "epoch": 0.009510086455331412,
+      "grad_norm": 20.378547867700338,
+      "learning_rate": 6.332576855076663e-06,
+      "loss": 3.0759,
+      "step": 33
+    },
+    {
+      "epoch": 0.009798270893371758,
+      "grad_norm": 22.71599913260458,
+      "learning_rate": 6.386644000699491e-06,
+      "loss": 2.7783,
+      "step": 34
+    },
+    {
+      "epoch": 0.010086455331412104,
+      "grad_norm": 27.406387142794618,
+      "learning_rate": 6.4391437599198735e-06,
+      "loss": 3.2274,
+      "step": 35
+    },
+    {
+      "epoch": 0.01037463976945245,
+      "grad_norm": 19.64330264999394,
+      "learning_rate": 6.490164454235688e-06,
+      "loss": 2.8711,
+      "step": 36
+    },
+    {
+      "epoch": 0.010662824207492795,
+      "grad_norm": 21.322369767837493,
+      "learning_rate": 6.5397871439471684e-06,
+      "loss": 2.5704,
+      "step": 37
+    },
+    {
+      "epoch": 0.01095100864553314,
+      "grad_norm": 32.31738147342337,
+      "learning_rate": 6.5880864029275e-06,
+      "loss": 3.0204,
+      "step": 38
+    },
+    {
+      "epoch": 0.011239193083573486,
+      "grad_norm": 20.961007392609183,
+      "learning_rate": 6.635130992737469e-06,
+      "loss": 2.6381,
+      "step": 39
+    },
+    {
+      "epoch": 0.011527377521613832,
+      "grad_norm": 17.38274192908384,
+      "learning_rate": 6.680984451379884e-06,
+      "loss": 2.3938,
+      "step": 40
+    },
+    {
+      "epoch": 0.01181556195965418,
+      "grad_norm": 19.507273735865926,
+      "learning_rate": 6.725705609344599e-06,
+      "loss": 2.7986,
+      "step": 41
+    },
+    {
+      "epoch": 0.012103746397694525,
+      "grad_norm": 15.160391921438011,
+      "learning_rate": 6.769349043460203e-06,
+      "loss": 2.515,
+      "step": 42
+    },
+    {
+      "epoch": 0.01239193083573487,
+      "grad_norm": 15.153798413143186,
+      "learning_rate": 6.811965477335649e-06,
+      "loss": 2.2172,
+      "step": 43
+    },
+    {
+      "epoch": 0.012680115273775217,
+      "grad_norm": 15.60502737286468,
+      "learning_rate": 6.853602135761187e-06,
+      "loss": 2.5867,
+      "step": 44
+    },
+    {
+      "epoch": 0.012968299711815562,
+      "grad_norm": 18.906378023267003,
+      "learning_rate": 6.89430305927829e-06,
+      "loss": 2.3077,
+      "step": 45
+    },
+    {
+      "epoch": 0.013256484149855908,
+      "grad_norm": 22.483387730826415,
+      "learning_rate": 6.934109384172617e-06,
+      "loss": 2.0573,
+      "step": 46
+    },
+    {
+      "epoch": 0.013544668587896254,
+      "grad_norm": 16.918339744078224,
+      "learning_rate": 6.973059592352829e-06,
+      "loss": 2.4301,
+      "step": 47
+    },
+    {
+      "epoch": 0.0138328530259366,
+      "grad_norm": 14.359138587395373,
+      "learning_rate": 7.011189734920213e-06,
+      "loss": 2.1018,
+      "step": 48
+    },
+    {
+      "epoch": 0.014121037463976945,
+      "grad_norm": 12.785390213380659,
+      "learning_rate": 7.048533632684716e-06,
+      "loss": 1.9024,
+      "step": 49
+    },
+    {
+      "epoch": 0.01440922190201729,
+      "grad_norm": 13.822682100623124,
+      "learning_rate": 7.085123056422486e-06,
+      "loss": 2.2607,
+      "step": 50
+    },
+    {
+      "epoch": 0.014697406340057636,
+      "grad_norm": 11.248470676887981,
+      "learning_rate": 7.1209878892824225e-06,
+      "loss": 1.9559,
+      "step": 51
+    },
+    {
+      "epoch": 0.014697406340057636,
+      "eval_websight_new_IoU": 0.010778559371829033,
+      "eval_websight_new_MAE_x": 363.0199890136719,
+      "eval_websight_new_MAE_y": 421.5400085449219,
+      "eval_websight_new_coord_probability": 0.0072825150564312935,
+      "eval_websight_new_correct_token_probability": 5.725064966100035e-06,
+      "eval_websight_new_inside_bbox": 0.019999999552965164,
+      "eval_websight_new_loss": 1.8606456518173218,
+      "eval_websight_new_range_100_probability": 0.0008474837522953749,
+      "eval_websight_new_range_20_probability": 0.0001835049915825948,
+      "eval_websight_new_range_5_probability": 4.652893039747141e-05,
+      "eval_websight_new_runtime": 33.4595,
+      "eval_websight_new_samples_per_second": 1.494,
+      "eval_websight_new_steps_per_second": 0.209,
+      "step": 51
+    },
+    {
+      "epoch": 0.014697406340057636,
+      "eval_seeclick_IoU": 0.009962150827050209,
+      "eval_seeclick_MAE_x": 473.05999755859375,
+      "eval_seeclick_MAE_y": 601.760009765625,
+      "eval_seeclick_coord_probability": 0.007472059689462185,
+      "eval_seeclick_correct_token_probability": 5.521962975763017e-06,
+      "eval_seeclick_inside_bbox": 0.05999999865889549,
+      "eval_seeclick_loss": 2.1386313438415527,
+      "eval_seeclick_range_100_probability": 0.0008697480079717934,
+      "eval_seeclick_range_20_probability": 0.00017912087787408382,
+      "eval_seeclick_range_5_probability": 4.512786472332664e-05,
+      "eval_seeclick_runtime": 42.4315,
+      "eval_seeclick_samples_per_second": 1.178,
+      "eval_seeclick_steps_per_second": 0.165,
+      "step": 51
+    },
+    {
+      "epoch": 0.014697406340057636,
+      "eval_icons_IoU": 0.0015337333315983415,
+      "eval_icons_MAE_x": 481.1199951171875,
+      "eval_icons_MAE_y": 428.3800048828125,
+      "eval_icons_coord_probability": 0.007426741532981396,
+      "eval_icons_correct_token_probability": 5.428708846011432e-06,
+      "eval_icons_inside_bbox": 0.0,
+      "eval_icons_loss": 1.865750789642334,
+      "eval_icons_range_100_probability": 0.0008250317187048495,
+      "eval_icons_range_20_probability": 0.0002058930549537763,
+      "eval_icons_range_5_probability": 5.214904376771301e-05,
+      "eval_icons_runtime": 42.6726,
+      "eval_icons_samples_per_second": 1.172,
+      "eval_icons_steps_per_second": 0.164,
+      "step": 51
+    },
+    {
+      "epoch": 0.014985590778097982,
+      "grad_norm": 9.772026011470567,
+      "learning_rate": 7.156156273421995e-06,
+      "loss": 1.7996,
+      "step": 52
+    },
+    {
+      "epoch": 0.01527377521613833,
+      "grad_norm": 12.657681600064546,
+      "learning_rate": 7.1906547426750725e-06,
+      "loss": 2.0842,
+      "step": 53
+    },
+    {
+      "epoch": 0.015561959654178675,
+      "grad_norm": 9.613342066404323,
+      "learning_rate": 7.224508342818619e-06,
+      "loss": 1.8363,
+      "step": 54
+    },
+    {
+      "epoch": 0.01585014409221902,
+      "grad_norm": 11.090633439192665,
+      "learning_rate": 7.25774074080379e-06,
+      "loss": 1.7267,
+      "step": 55
+    },
+    {
+      "epoch": 0.016138328530259365,
+      "grad_norm": 13.126846403161675,
+      "learning_rate": 7.290374324144728e-06,
+      "loss": 1.9524,
+      "step": 56
+    },
+    {
+      "epoch": 0.016426512968299712,
+      "grad_norm": 7.940529281455624,
+      "learning_rate": 7.3224302915104325e-06,
+      "loss": 1.7391,
+      "step": 57
+    },
+    {
+      "epoch": 0.016714697406340056,
+      "grad_norm": 10.327915839031606,
+      "learning_rate": 7.3539287354378455e-06,
+      "loss": 1.6814,
+      "step": 58
+    },
+    {
+      "epoch": 0.017002881844380403,
+      "grad_norm": 12.465134904038113,
+      "learning_rate": 7.38488871797435e-06,
+      "loss": 1.8826,
+      "step": 59
+    },
+    {
+      "epoch": 0.01729106628242075,
+      "grad_norm": 9.518374438515043,
+      "learning_rate": 7.415328339962814e-06,
+      "loss": 1.7086,
+      "step": 60
+    },
+    {
+      "epoch": 0.017579250720461095,
+      "grad_norm": 10.860650734562737,
+      "learning_rate": 7.4452648045998055e-06,
+      "loss": 1.6188,
+      "step": 61
+    },
+    {
+      "epoch": 0.017867435158501442,
+      "grad_norm": 7.824123766994198,
+      "learning_rate": 7.474714475825724e-06,
+      "loss": 1.7929,
+      "step": 62
+    },
+    {
+      "epoch": 0.018155619596541786,
+      "grad_norm": 8.104791112765197,
+      "learning_rate": 7.503692932043134e-06,
+      "loss": 1.6061,
+      "step": 63
+    },
+    {
+      "epoch": 0.018443804034582133,
+      "grad_norm": 8.993205128398746,
+      "learning_rate": 7.532215015604735e-06,
+      "loss": 1.5625,
+      "step": 64
+    },
+    {
+      "epoch": 0.018731988472622477,
+      "grad_norm": 8.176567292713766,
+      "learning_rate": 7.560294878464597e-06,
+      "loss": 1.7126,
+      "step": 65
+    },
+    {
+      "epoch": 0.019020172910662825,
+      "grad_norm": 7.132222732428861,
+      "learning_rate": 7.587946024344118e-06,
+      "loss": 1.5697,
+      "step": 66
+    },
+    {
+      "epoch": 0.01930835734870317,
+      "grad_norm": 8.127988816552792,
+      "learning_rate": 7.615181347727268e-06,
+      "loss": 1.5083,
+      "step": 67
+    },
+    {
+      "epoch": 0.019596541786743516,
+      "grad_norm": 10.280846062275057,
+      "learning_rate": 7.642013169966947e-06,
+      "loss": 1.6577,
+      "step": 68
+    },
+    {
+      "epoch": 0.01988472622478386,
+      "grad_norm": 6.578790507069004,
+      "learning_rate": 7.668453272755549e-06,
+      "loss": 1.5532,
+      "step": 69
+    },
+    {
+      "epoch": 0.020172910662824207,
+      "grad_norm": 10.222158144657401,
+      "learning_rate": 7.69451292918733e-06,
+      "loss": 1.4516,
+      "step": 70
+    },
+    {
+      "epoch": 0.020461095100864555,
+      "grad_norm": 8.033521635824966,
+      "learning_rate": 7.720202932617524e-06,
+      "loss": 1.5968,
+      "step": 71
+    },
+    {
+      "epoch": 0.0207492795389049,
+      "grad_norm": 6.283023849212295,
+      "learning_rate": 7.745533623503144e-06,
+      "loss": 1.4865,
+      "step": 72
+    },
+    {
+      "epoch": 0.021037463976945246,
+      "grad_norm": 11.541429466909177,
+      "learning_rate": 7.770514914392505e-06,
+      "loss": 1.3943,
+      "step": 73
+    },
+    {
+      "epoch": 0.02132564841498559,
+      "grad_norm": 6.810441213975766,
+      "learning_rate": 7.795156313214624e-06,
+      "loss": 1.5454,
+      "step": 74
+    },
+    {
+      "epoch": 0.021613832853025938,
+      "grad_norm": 7.50975549320261,
+      "learning_rate": 7.819466945005417e-06,
+      "loss": 1.4353,
+      "step": 75
+    },
+    {
+      "epoch": 0.02190201729106628,
+      "grad_norm": 7.460716357179955,
+      "learning_rate": 7.843455572194956e-06,
+      "loss": 1.3462,
+      "step": 76
+    },
+    {
+      "epoch": 0.02219020172910663,
+      "grad_norm": 7.5211049677596,
+      "learning_rate": 7.867130613568634e-06,
+      "loss": 1.4812,
+      "step": 77
+    },
+    {
+      "epoch": 0.022478386167146973,
+      "grad_norm": 6.112353977933034,
+      "learning_rate": 7.890500162004926e-06,
+      "loss": 1.3572,
+      "step": 78
+    },
+    {
+      "epoch": 0.02276657060518732,
+      "grad_norm": 10.0445955512976,
+      "learning_rate": 7.913572001083272e-06,
+      "loss": 1.296,
+      "step": 79
+    },
+    {
+      "epoch": 0.023054755043227664,
+      "grad_norm": 6.124066253475085,
+      "learning_rate": 7.93635362064734e-06,
+      "loss": 1.4276,
+      "step": 80
+    },
+    {
+      "epoch": 0.02334293948126801,
+      "grad_norm": 6.316055475181084,
+      "learning_rate": 7.958852231401552e-06,
+      "loss": 1.3069,
+      "step": 81
+    },
+    {
+      "epoch": 0.02363112391930836,
+      "grad_norm": 6.578049211563219,
+      "learning_rate": 7.981074778612054e-06,
+      "loss": 1.2461,
+      "step": 82
+    },
+    {
+      "epoch": 0.023919308357348703,
+      "grad_norm": 7.444772664188824,
+      "learning_rate": 8.003027954977264e-06,
+      "loss": 1.3804,
+      "step": 83
+    },
+    {
+      "epoch": 0.02420749279538905,
+      "grad_norm": 6.5602592933063395,
+      "learning_rate": 8.024718212727658e-06,
+      "loss": 1.27,
+      "step": 84
+    },
+    {
+      "epoch": 0.024495677233429394,
+      "grad_norm": 6.942091189640846,
+      "learning_rate": 8.04615177500955e-06,
+      "loss": 1.1982,
+      "step": 85
+    },
+    {
+      "epoch": 0.02478386167146974,
+      "grad_norm": 6.438612199361165,
+      "learning_rate": 8.067334646603105e-06,
+      "loss": 1.3237,
+      "step": 86
+    },
+    {
+      "epoch": 0.025072046109510086,
+      "grad_norm": 5.547834949064384,
+      "learning_rate": 8.088272624020777e-06,
+      "loss": 1.2506,
+      "step": 87
+    },
+    {
+      "epoch": 0.025360230547550433,
+      "grad_norm": 7.768644495811314,
+      "learning_rate": 8.108971305028645e-06,
+      "loss": 1.1535,
+      "step": 88
+    },
+    {
+      "epoch": 0.025648414985590777,
+      "grad_norm": 6.0374882659496505,
+      "learning_rate": 8.129436097629779e-06,
+      "loss": 1.2679,
+      "step": 89
+    },
+    {
+      "epoch": 0.025936599423631124,
+      "grad_norm": 5.491709410176797,
+      "learning_rate": 8.149672228545746e-06,
+      "loss": 1.1669,
+      "step": 90
+    },
+    {
+      "epoch": 0.02622478386167147,
+      "grad_norm": 6.2627480902334725,
+      "learning_rate": 8.16968475122944e-06,
+      "loss": 1.104,
+      "step": 91
+    },
+    {
+      "epoch": 0.026512968299711816,
+      "grad_norm": 6.052249863888647,
+      "learning_rate": 8.189478553440074e-06,
+      "loss": 1.225,
+      "step": 92
+    },
+    {
+      "epoch": 0.02680115273775216,
+      "grad_norm": 5.386417663387223,
+      "learning_rate": 8.209058364408657e-06,
+      "loss": 1.1572,
+      "step": 93
+    },
+    {
+      "epoch": 0.027089337175792507,
+      "grad_norm": 6.281249602777029,
+      "learning_rate": 8.228428761620285e-06,
+      "loss": 1.0714,
+      "step": 94
+    },
+    {
+      "epoch": 0.027377521613832854,
+      "grad_norm": 5.836760674441374,
+      "learning_rate": 8.247594177237559e-06,
+      "loss": 1.1769,
+      "step": 95
+    },
+    {
+      "epoch": 0.0276657060518732,
+      "grad_norm": 5.266384098127444,
+      "learning_rate": 8.266558904187668e-06,
+      "loss": 1.0825,
+      "step": 96
+    },
+    {
+      "epoch": 0.027953890489913546,
+      "grad_norm": 5.929747177357997,
+      "learning_rate": 8.28532710193407e-06,
+      "loss": 1.0329,
+      "step": 97
+    },
+    {
+      "epoch": 0.02824207492795389,
+      "grad_norm": 5.622834489781087,
+      "learning_rate": 8.303902801952174e-06,
+      "loss": 1.1473,
+      "step": 98
+    },
+    {
+      "epoch": 0.028530259365994237,
+      "grad_norm": 5.150387365532939,
+      "learning_rate": 8.32228991292705e-06,
+      "loss": 1.0387,
+      "step": 99
+    },
+    {
+      "epoch": 0.02881844380403458,
+      "grad_norm": 5.629354297769571,
+      "learning_rate": 8.340492225689944e-06,
+      "loss": 0.9917,
+      "step": 100
+    },
+    {
+      "epoch": 0.02910662824207493,
+      "grad_norm": 5.70491743358872,
+      "learning_rate": 8.358513417909158e-06,
+      "loss": 1.1078,
+      "step": 101
+    },
+    {
+      "epoch": 0.029394812680115272,
+      "grad_norm": 5.053891183636838,
+      "learning_rate": 8.376357058549878e-06,
+      "loss": 1.0231,
+      "step": 102
+    },
+    {
+      "epoch": 0.02968299711815562,
+      "grad_norm": 5.675823683975542,
+      "learning_rate": 8.394026612116405e-06,
+      "loss": 0.9555,
+      "step": 103
+    },
+    {
+      "epoch": 0.029971181556195964,
+      "grad_norm": 5.463972580079715,
+      "learning_rate": 8.41152544268945e-06,
+      "loss": 1.0771,
+      "step": 104
+    },
+    {
+      "epoch": 0.03025936599423631,
+      "grad_norm": 4.999123462776384,
+      "learning_rate": 8.42885681777026e-06,
+      "loss": 0.9859,
+      "step": 105
+    },
+    {
+      "epoch": 0.03054755043227666,
+      "grad_norm": 5.497431583114866,
+      "learning_rate": 8.446023911942528e-06,
+      "loss": 0.9313,
+      "step": 106
+    },
+    {
+      "epoch": 0.030835734870317003,
+      "grad_norm": 5.391678295140468,
+      "learning_rate": 8.463029810362388e-06,
+      "loss": 1.0353,
+      "step": 107
+    },
+    {
+      "epoch": 0.03112391930835735,
+      "grad_norm": 4.891677757189106,
+      "learning_rate": 8.479877512086076e-06,
+      "loss": 0.9647,
+      "step": 108
+    },
+    {
+      "epoch": 0.0314121037463977,
+      "grad_norm": 5.610726694133473,
+      "learning_rate": 8.496569933244228e-06,
+      "loss": 0.8908,
+      "step": 109
+    },
+    {
+      "epoch": 0.03170028818443804,
+      "grad_norm": 5.817735866127734,
+      "learning_rate": 8.513109910071246e-06,
+      "loss": 1.0143,
+      "step": 110
+    },
+    {
+      "epoch": 0.031988472622478385,
+      "grad_norm": 4.871866053471662,
+      "learning_rate": 8.529500201797555e-06,
+      "loss": 0.9479,
+      "step": 111
+    },
+    {
+      "epoch": 0.03227665706051873,
+      "grad_norm": 5.31764218824648,
+      "learning_rate": 8.545743493412182e-06,
+      "loss": 0.8705,
+      "step": 112
+    },
+    {
+      "epoch": 0.03256484149855908,
+      "grad_norm": 5.081131222064519,
+      "learning_rate": 8.561842398302535e-06,
+      "loss": 0.9911,
+      "step": 113
+    },
+    {
+      "epoch": 0.032853025936599424,
+      "grad_norm": 4.610845570193001,
+      "learning_rate": 8.577799460777888e-06,
+      "loss": 0.9058,
+      "step": 114
+    },
+    {
+      "epoch": 0.03314121037463977,
+      "grad_norm": 5.478425064148529,
+      "learning_rate": 8.593617158482677e-06,
+      "loss": 0.8407,
+      "step": 115
+    },
+    {
+      "epoch": 0.03342939481268011,
+      "grad_norm": 4.9308395705928385,
+      "learning_rate": 8.609297904705302e-06,
+      "loss": 0.9558,
+      "step": 116
+    },
+    {
+      "epoch": 0.03371757925072046,
+      "grad_norm": 4.4313658226774,
+      "learning_rate": 8.624844050587858e-06,
+      "loss": 0.9037,
+      "step": 117
+    },
+    {
+      "epoch": 0.03400576368876081,
+      "grad_norm": 5.1957219052400525,
+      "learning_rate": 8.640257887241806e-06,
+      "loss": 0.8201,
+      "step": 118
+    },
+    {
+      "epoch": 0.03429394812680115,
+      "grad_norm": 4.637596099695778,
+      "learning_rate": 8.655541647774393e-06,
+      "loss": 0.9466,
+      "step": 119
+    },
+    {
+      "epoch": 0.0345821325648415,
+      "grad_norm": 4.210947996449233,
+      "learning_rate": 8.67069750923027e-06,
+      "loss": 0.8812,
+      "step": 120
+    },
+    {
+      "epoch": 0.034870317002881845,
+      "grad_norm": 4.656876963041862,
+      "learning_rate": 8.68572759445255e-06,
+      "loss": 0.7964,
+      "step": 121
+    },
+    {
+      "epoch": 0.03515850144092219,
+      "grad_norm": 4.598584830368883,
+      "learning_rate": 8.700633973867262e-06,
+      "loss": 0.9246,
+      "step": 122
+    },
+    {
+      "epoch": 0.03544668587896253,
+      "grad_norm": 4.2321321744684415,
+      "learning_rate": 8.715418667194985e-06,
+      "loss": 0.8862,
+      "step": 123
+    },
+    {
+      "epoch": 0.035734870317002884,
+      "grad_norm": 4.600363613453042,
+      "learning_rate": 8.73008364509318e-06,
+      "loss": 0.7905,
+      "step": 124
+    },
+    {
+      "epoch": 0.03602305475504323,
+      "grad_norm": 4.198077372961204,
+      "learning_rate": 8.744630830732545e-06,
+      "loss": 0.9113,
+      "step": 125
+    },
+    {
+      "epoch": 0.03631123919308357,
+      "grad_norm": 3.7501430166268435,
+      "learning_rate": 8.75906210131059e-06,
+      "loss": 0.8415,
+      "step": 126
+    },
+    {
+      "epoch": 0.036599423631123916,
+      "grad_norm": 4.419910726022508,
+      "learning_rate": 8.773379289505366e-06,
+      "loss": 0.752,
+      "step": 127
+    },
+    {
+      "epoch": 0.03688760806916427,
+      "grad_norm": 3.997242466871886,
+      "learning_rate": 8.787584184872193e-06,
+      "loss": 0.8776,
+      "step": 128
+    },
+    {
+      "epoch": 0.03717579250720461,
+      "grad_norm": 3.534947859211222,
+      "learning_rate": 8.801678535186037e-06,
+      "loss": 0.8254,
+      "step": 129
+    },
+    {
+      "epoch": 0.037463976945244955,
+      "grad_norm": 4.428264660634081,
+      "learning_rate": 8.815664047732054e-06,
+      "loss": 0.7262,
+      "step": 130
+    },
+    {
+      "epoch": 0.037752161383285306,
+      "grad_norm": 3.8981897659675044,
+      "learning_rate": 8.829542390546686e-06,
+      "loss": 0.8572,
+      "step": 131
+    },
+    {
+      "epoch": 0.03804034582132565,
+      "grad_norm": 3.4427721693779323,
+      "learning_rate": 8.843315193611574e-06,
+      "loss": 0.8358,
+      "step": 132
+    },
+    {
+      "epoch": 0.038328530259365994,
+      "grad_norm": 3.9905563522535226,
+      "learning_rate": 8.856984050002403e-06,
+      "loss": 0.7198,
+      "step": 133
+    },
+    {
+      "epoch": 0.03861671469740634,
+      "grad_norm": 3.706844015046595,
+      "learning_rate": 8.870550516994724e-06,
+      "loss": 0.8521,
+      "step": 134
+    },
+    {
+      "epoch": 0.03890489913544669,
+      "grad_norm": 3.186539722022302,
+      "learning_rate": 8.884016117128679e-06,
+      "loss": 0.8066,
+      "step": 135
+    },
+    {
+      "epoch": 0.03919308357348703,
+      "grad_norm": 3.8254743339541295,
+      "learning_rate": 8.897382339234405e-06,
+      "loss": 0.712,
+      "step": 136
+    },
+    {
+      "epoch": 0.039481268011527376,
+      "grad_norm": 3.500633177755673,
+      "learning_rate": 8.910650639419908e-06,
+      "loss": 0.8213,
+      "step": 137
+    },
+    {
+      "epoch": 0.03976945244956772,
+      "grad_norm": 3.079033701156694,
+      "learning_rate": 8.923822442023006e-06,
+      "loss": 0.8025,
+      "step": 138
+    },
+    {
+      "epoch": 0.04005763688760807,
+      "grad_norm": 3.548379275418023,
+      "learning_rate": 8.936899140528881e-06,
+      "loss": 0.6977,
+      "step": 139
+    },
+    {
+      "epoch": 0.040345821325648415,
+      "grad_norm": 3.0662833147413306,
+      "learning_rate": 8.949882098454784e-06,
+      "loss": 0.7955,
+      "step": 140
+    },
+    {
+      "epoch": 0.04063400576368876,
+      "grad_norm": 2.768024362479292,
+      "learning_rate": 8.962772650203216e-06,
+      "loss": 0.7805,
+      "step": 141
+    },
+    {
+      "epoch": 0.04092219020172911,
+      "grad_norm": 3.2237959260312223,
+      "learning_rate": 8.975572101884981e-06,
+      "loss": 0.6739,
+      "step": 142
+    },
+    {
+      "epoch": 0.041210374639769454,
+      "grad_norm": 2.7734152839051283,
+      "learning_rate": 8.988281732113356e-06,
+      "loss": 0.8002,
+      "step": 143
+    },
+    {
+      "epoch": 0.0414985590778098,
+      "grad_norm": 2.250679674647897,
+      "learning_rate": 9.0009027927706e-06,
+      "loss": 0.7839,
+      "step": 144
+    },
+    {
+      "epoch": 0.04178674351585014,
+      "grad_norm": 3.1539427375278724,
+      "learning_rate": 9.013436509747905e-06,
+      "loss": 0.6708,
+      "step": 145
+    },
+    {
+      "epoch": 0.04207492795389049,
+      "grad_norm": 2.4309547666916878,
+      "learning_rate": 9.025884083659961e-06,
+      "loss": 0.7986,
+      "step": 146
+    },
+    {
+      "epoch": 0.042363112391930836,
+      "grad_norm": 2.01309967729301,
+      "learning_rate": 9.038246690535104e-06,
+      "loss": 0.766,
+      "step": 147
+    },
+    {
+      "epoch": 0.04265129682997118,
+      "grad_norm": 3.1015443582771693,
+      "learning_rate": 9.05052548248208e-06,
+      "loss": 0.6383,
+      "step": 148
+    },
+    {
+      "epoch": 0.042939481268011524,
+      "grad_norm": 2.2747882705969196,
+      "learning_rate": 9.062721588334355e-06,
+      "loss": 0.7744,
+      "step": 149
+    },
+    {
+      "epoch": 0.043227665706051875,
+      "grad_norm": 1.874678323320859,
+      "learning_rate": 9.074836114272873e-06,
+      "loss": 0.7578,
+      "step": 150
+    },
+    {
+      "epoch": 0.04351585014409222,
+      "grad_norm": 2.73548781933127,
+      "learning_rate": 9.08687014442814e-06,
+      "loss": 0.6651,
+      "step": 151
+    },
+    {
+      "epoch": 0.04380403458213256,
+      "grad_norm": 2.07172770752639,
+      "learning_rate": 9.098824741462414e-06,
+      "loss": 0.7656,
+      "step": 152
+    },
+    {
+      "epoch": 0.044092219020172914,
+      "grad_norm": 2.426006794767885,
+      "learning_rate": 9.11070094713281e-06,
+      "loss": 0.7627,
+      "step": 153
+    },
+    {
+      "epoch": 0.04438040345821326,
+      "grad_norm": 2.6608504533829533,
+      "learning_rate": 9.12249978283609e-06,
+      "loss": 0.6537,
+      "step": 154
+    },
+    {
+      "epoch": 0.0446685878962536,
+      "grad_norm": 2.1136226301554877,
+      "learning_rate": 9.134222250135783e-06,
+      "loss": 0.7838,
+      "step": 155
+    },
+    {
+      "epoch": 0.044956772334293946,
+      "grad_norm": 1.7542973386472682,
+      "learning_rate": 9.145869331272382e-06,
+      "loss": 0.7444,
+      "step": 156
+    },
+    {
+      "epoch": 0.0452449567723343,
+      "grad_norm": 2.649915281124027,
+      "learning_rate": 9.15744198965723e-06,
+      "loss": 0.6349,
+      "step": 157
+    },
+    {
+      "epoch": 0.04553314121037464,
+      "grad_norm": 1.9522249812385406,
+      "learning_rate": 9.16894117035073e-06,
+      "loss": 0.7693,
+      "step": 158
+    },
+    {
+      "epoch": 0.045821325648414984,
+      "grad_norm": 1.8613859415577985,
+      "learning_rate": 9.18036780052546e-06,
+      "loss": 0.7551,
+      "step": 159
+    },
+    {
+      "epoch": 0.04610951008645533,
+      "grad_norm": 2.627616979894939,
+      "learning_rate": 9.191722789914796e-06,
+      "loss": 0.6259,
+      "step": 160
+    },
+    {
+      "epoch": 0.04639769452449568,
+      "grad_norm": 2.3971206406714805,
+      "learning_rate": 9.203007031247519e-06,
+      "loss": 0.7471,
+      "step": 161
+    },
+    {
+      "epoch": 0.04668587896253602,
+      "grad_norm": 1.8047186312374244,
+      "learning_rate": 9.214221400669006e-06,
+      "loss": 0.7432,
+      "step": 162
+    },
+    {
+      "epoch": 0.04697406340057637,
+      "grad_norm": 2.631275686948647,
+      "learning_rate": 9.225366758149435e-06,
+      "loss": 0.6261,
+      "step": 163
+    },
+    {
+      "epoch": 0.04726224783861672,
+      "grad_norm": 2.345316372051671,
+      "learning_rate": 9.23644394787951e-06,
+      "loss": 0.737,
+      "step": 164
+    },
+    {
+      "epoch": 0.04755043227665706,
+      "grad_norm": 1.8856785473095028,
+      "learning_rate": 9.247453798654177e-06,
+      "loss": 0.7196,
+      "step": 165
+    },
+    {
+      "epoch": 0.047838616714697406,
+      "grad_norm": 2.6540847088644464,
+      "learning_rate": 9.258397124244722e-06,
+      "loss": 0.6193,
+      "step": 166
+    },
+    {
+      "epoch": 0.04812680115273775,
+      "grad_norm": 1.9912082892186653,
+      "learning_rate": 9.2692747237597e-06,
+      "loss": 0.7346,
+      "step": 167
+    },
+    {
+      "epoch": 0.0484149855907781,
+      "grad_norm": 1.9022891213962014,
+      "learning_rate": 9.280087381995114e-06,
+      "loss": 0.7211,
+      "step": 168
+    },
+    {
+      "epoch": 0.048703170028818445,
+      "grad_norm": 2.855680994233883,
+      "learning_rate": 9.290835869774166e-06,
+      "loss": 0.6233,
+      "step": 169
+    },
+    {
+      "epoch": 0.04899135446685879,
+      "grad_norm": 1.8992692898334764,
+      "learning_rate": 9.301520944277006e-06,
+      "loss": 0.7406,
+      "step": 170
+    },
+    {
+      "epoch": 0.04927953890489913,
+      "grad_norm": 1.8061418922496177,
+      "learning_rate": 9.31214334936082e-06,
+      "loss": 0.7206,
+      "step": 171
+    },
+    {
+      "epoch": 0.04956772334293948,
+      "grad_norm": 3.037852726523747,
+      "learning_rate": 9.32270381587056e-06,
+      "loss": 0.605,
+      "step": 172
+    },
+    {
+      "epoch": 0.04985590778097983,
+      "grad_norm": 2.1650228003560645,
+      "learning_rate": 9.333203061940696e-06,
+      "loss": 0.7204,
+      "step": 173
+    },
+    {
+      "epoch": 0.05014409221902017,
+      "grad_norm": 2.1427179809283405,
+      "learning_rate": 9.343641793288234e-06,
+      "loss": 0.7136,
+      "step": 174
+    },
+    {
+      "epoch": 0.05043227665706052,
+      "grad_norm": 2.7496668086096823,
+      "learning_rate": 9.354020703497389e-06,
+      "loss": 0.6172,
+      "step": 175
+    },
+    {
+      "epoch": 0.050720461095100866,
+      "grad_norm": 2.469362015895731,
+      "learning_rate": 9.3643404742961e-06,
+      "loss": 0.7237,
+      "step": 176
+    },
+    {
+      "epoch": 0.05100864553314121,
+      "grad_norm": 2.1037607295004004,
+      "learning_rate": 9.374601775824737e-06,
+      "loss": 0.7092,
+      "step": 177
+    },
+    {
+      "epoch": 0.051296829971181554,
+      "grad_norm": 2.6105303567931823,
+      "learning_rate": 9.384805266897236e-06,
+      "loss": 0.6011,
+      "step": 178
+    },
+    {
+      "epoch": 0.051585014409221905,
+      "grad_norm": 2.4143791803196546,
+      "learning_rate": 9.394951595254911e-06,
+      "loss": 0.7142,
+      "step": 179
+    },
+    {
+      "epoch": 0.05187319884726225,
+      "grad_norm": 1.7090641211386361,
+      "learning_rate": 9.405041397813202e-06,
+      "loss": 0.6932,
+      "step": 180
+    },
+    {
+      "epoch": 0.05216138328530259,
+      "grad_norm": 2.999007860084135,
+      "learning_rate": 9.41507530090159e-06,
+      "loss": 0.6001,
+      "step": 181
+    },
+    {
+      "epoch": 0.05244956772334294,
+      "grad_norm": 2.5026419005785936,
+      "learning_rate": 9.425053920496896e-06,
+      "loss": 0.6877,
+      "step": 182
+    },
+    {
+      "epoch": 0.05273775216138329,
+      "grad_norm": 1.5888472945782244,
+      "learning_rate": 9.434977862450193e-06,
+      "loss": 0.6942,
+      "step": 183
+    },
+    {
+      "epoch": 0.05302593659942363,
+      "grad_norm": 2.621111877977237,
+      "learning_rate": 9.44484772270753e-06,
+      "loss": 0.5951,
+      "step": 184
+    },
+    {
+      "epoch": 0.053314121037463975,
+      "grad_norm": 1.9813131092916778,
+      "learning_rate": 9.454664087524683e-06,
+      "loss": 0.6937,
+      "step": 185
+    },
+    {
+      "epoch": 0.05360230547550432,
+      "grad_norm": 1.870575239040046,
+      "learning_rate": 9.464427533676113e-06,
+      "loss": 0.7063,
+      "step": 186
+    },
+    {
+      "epoch": 0.05389048991354467,
+      "grad_norm": 2.353331569351921,
+      "learning_rate": 9.47413862865831e-06,
+      "loss": 0.5961,
+      "step": 187
+    },
+    {
+      "epoch": 0.054178674351585014,
+      "grad_norm": 1.794698720408445,
+      "learning_rate": 9.483797930887741e-06,
+      "loss": 0.6878,
+      "step": 188
+    },
+    {
+      "epoch": 0.05446685878962536,
+      "grad_norm": 2.0116398279578016,
+      "learning_rate": 9.493405989893522e-06,
+      "loss": 0.6783,
+      "step": 189
+    },
+    {
+      "epoch": 0.05475504322766571,
+      "grad_norm": 2.741855766208623,
+      "learning_rate": 9.502963346505015e-06,
+      "loss": 0.5801,
+      "step": 190
+    },
+    {
+      "epoch": 0.05504322766570605,
+      "grad_norm": 2.38993948854855,
+      "learning_rate": 9.512470533034512e-06,
+      "loss": 0.711,
+      "step": 191
+    },
+    {
+      "epoch": 0.0553314121037464,
+      "grad_norm": 1.9766927221810036,
+      "learning_rate": 9.521928073455125e-06,
+      "loss": 0.6893,
+      "step": 192
+    },
+    {
+      "epoch": 0.05561959654178674,
+      "grad_norm": 2.6954209105434246,
+      "learning_rate": 9.531336483574083e-06,
+      "loss": 0.5935,
+      "step": 193
+    },
+    {
+      "epoch": 0.05590778097982709,
+      "grad_norm": 2.2338669962164928,
+      "learning_rate": 9.540696271201526e-06,
+      "loss": 0.6966,
+      "step": 194
+    },
+    {
+      "epoch": 0.056195965417867436,
+      "grad_norm": 1.5236329673106341,
+      "learning_rate": 9.550007936314986e-06,
+      "loss": 0.672,
+      "step": 195
+    },
+    {
+      "epoch": 0.05648414985590778,
+      "grad_norm": 3.0857241897790577,
+      "learning_rate": 9.559271971219628e-06,
+      "loss": 0.5804,
+      "step": 196
+    },
+    {
+      "epoch": 0.056772334293948123,
+      "grad_norm": 3.08494922150488,
+      "learning_rate": 9.568488860704454e-06,
+      "loss": 0.7248,
+      "step": 197
+    },
+    {
+      "epoch": 0.057060518731988474,
+      "grad_norm": 1.7510197355931456,
+      "learning_rate": 9.577659082194506e-06,
+      "loss": 0.6751,
+      "step": 198
+    },
+    {
+      "epoch": 0.05734870317002882,
+      "grad_norm": 3.8852483936339173,
+      "learning_rate": 9.586783105899282e-06,
+      "loss": 0.5727,
+      "step": 199
+    },
+    {
+      "epoch": 0.05763688760806916,
+      "grad_norm": 3.929409013124462,
+      "learning_rate": 9.595861394957398e-06,
+      "loss": 0.6807,
+      "step": 200
+    },
+    {
+      "epoch": 0.05792507204610951,
+      "grad_norm": 1.5785031461666155,
+      "learning_rate": 9.604894405577658e-06,
+      "loss": 0.6806,
+      "step": 201
+    },
+    {
+      "epoch": 0.05821325648414986,
+      "grad_norm": 3.1878732741345246,
+      "learning_rate": 9.613882587176614e-06,
+      "loss": 0.5734,
+      "step": 202
+    },
+    {
+      "epoch": 0.0585014409221902,
+      "grad_norm": 3.1854201233525523,
+      "learning_rate": 9.622826382512747e-06,
+      "loss": 0.6922,
+      "step": 203
+    },
+    {
+      "epoch": 0.058789625360230545,
+      "grad_norm": 1.8526625111973063,
+      "learning_rate": 9.631726227817333e-06,
+      "loss": 0.6509,
+      "step": 204
+    },
+    {
+      "epoch": 0.059077809798270896,
+      "grad_norm": 2.7294299798853348,
+      "learning_rate": 9.640582552922113e-06,
+      "loss": 0.5793,
+      "step": 205
+    },
+    {
+      "epoch": 0.05936599423631124,
+      "grad_norm": 3.3941744620197984,
+      "learning_rate": 9.64939578138386e-06,
+      "loss": 0.6888,
+      "step": 206
+    },
+    {
+      "epoch": 0.059654178674351584,
+      "grad_norm": 1.9883789922192263,
+      "learning_rate": 9.658166330605937e-06,
+      "loss": 0.6699,
+      "step": 207
+    },
+    {
+      "epoch": 0.05994236311239193,
+      "grad_norm": 2.758267978856053,
+      "learning_rate": 9.666894611956906e-06,
+      "loss": 0.5638,
+      "step": 208
+    },
+    {
+      "epoch": 0.06023054755043228,
+      "grad_norm": 1.8171103966983788,
+      "learning_rate": 9.67558103088632e-06,
+      "loss": 0.6893,
+      "step": 209
+    },
+    {
+      "epoch": 0.06051873198847262,
+      "grad_norm": 2.1888003833329712,
+      "learning_rate": 9.684225987037717e-06,
+      "loss": 0.6822,
+      "step": 210
+    },
+    {
+      "epoch": 0.060806916426512966,
+      "grad_norm": 2.6024308217775913,
+      "learning_rate": 9.69282987435897e-06,
+      "loss": 0.5748,
+      "step": 211
+    },
+    {
+      "epoch": 0.06109510086455332,
+      "grad_norm": 1.8525049491547192,
+      "learning_rate": 9.701393081209986e-06,
+      "loss": 0.666,
+      "step": 212
+    },
+    {
+      "epoch": 0.06138328530259366,
+      "grad_norm": 2.0029479305883666,
+      "learning_rate": 9.709915990467912e-06,
+      "loss": 0.6596,
+      "step": 213
+    },
+    {
+      "epoch": 0.061671469740634005,
+      "grad_norm": 2.76483624262101,
+      "learning_rate": 9.718398979629844e-06,
+      "loss": 0.575,
+      "step": 214
+    },
+    {
+      "epoch": 0.06195965417867435,
+      "grad_norm": 1.6541639511464132,
+      "learning_rate": 9.726842420913163e-06,
+      "loss": 0.6784,
+      "step": 215
+    },
+    {
+      "epoch": 0.0622478386167147,
+      "grad_norm": 2.20119899046267,
+      "learning_rate": 9.735246681353532e-06,
+      "loss": 0.6708,
+      "step": 216
+    },
+    {
+      "epoch": 0.06253602305475504,
+      "grad_norm": 3.0557683367774144,
+      "learning_rate": 9.743612122900627e-06,
+      "loss": 0.5411,
+      "step": 217
+    },
+    {
+      "epoch": 0.0628242074927954,
+      "grad_norm": 1.8749695111686993,
+      "learning_rate": 9.751939102511684e-06,
+      "loss": 0.6743,
+      "step": 218
+    },
+    {
+      "epoch": 0.06311239193083573,
+      "grad_norm": 1.482297281360979,
+      "learning_rate": 9.760227972242893e-06,
+      "loss": 0.664,
+      "step": 219
+    },
+    {
+      "epoch": 0.06340057636887608,
+      "grad_norm": 3.422962977642731,
+      "learning_rate": 9.768479079338704e-06,
+      "loss": 0.5599,
+      "step": 220
+    },
+    {
+      "epoch": 0.06368876080691642,
+      "grad_norm": 1.9338581326114699,
+      "learning_rate": 9.776692766319116e-06,
+      "loss": 0.6797,
+      "step": 221
+    },
+    {
+      "epoch": 0.06397694524495677,
+      "grad_norm": 1.2609633785017869,
+      "learning_rate": 9.78486937106501e-06,
+      "loss": 0.669,
+      "step": 222
+    },
+    {
+      "epoch": 0.06426512968299712,
+      "grad_norm": 3.5571118701177826,
+      "learning_rate": 9.793009226901535e-06,
+      "loss": 0.555,
+      "step": 223
+    },
+    {
+      "epoch": 0.06455331412103746,
+      "grad_norm": 2.9462476487007074,
+      "learning_rate": 9.801112662679638e-06,
+      "loss": 0.6844,
+      "step": 224
+    },
+    {
+      "epoch": 0.06484149855907781,
+      "grad_norm": 1.5889839937538799,
+      "learning_rate": 9.809180002855807e-06,
+      "loss": 0.6593,
+      "step": 225
+    },
+    {
+      "epoch": 0.06512968299711816,
+      "grad_norm": 3.7602466428938217,
+      "learning_rate": 9.817211567569991e-06,
+      "loss": 0.5668,
+      "step": 226
+    },
+    {
+      "epoch": 0.0654178674351585,
+      "grad_norm": 2.030001995574258,
+      "learning_rate": 9.825207672721862e-06,
+      "loss": 0.6841,
+      "step": 227
+    },
+    {
+      "epoch": 0.06570605187319885,
+      "grad_norm": 2.233169312531071,
+      "learning_rate": 9.833168630045344e-06,
+      "loss": 0.6487,
+      "step": 228
+    },
+    {
+      "epoch": 0.0659942363112392,
+      "grad_norm": 2.5736297275905424,
+      "learning_rate": 9.841094747181556e-06,
+      "loss": 0.562,
+      "step": 229
+    },
+    {
+      "epoch": 0.06628242074927954,
+      "grad_norm": 2.291853135098293,
+      "learning_rate": 9.848986327750132e-06,
+      "loss": 0.6562,
+      "step": 230
+    },
+    {
+      "epoch": 0.06657060518731989,
+      "grad_norm": 1.8728261595091804,
+      "learning_rate": 9.85684367141902e-06,
+      "loss": 0.6473,
+      "step": 231
+    },
+    {
+      "epoch": 0.06685878962536022,
+      "grad_norm": 2.583515503245737,
+      "learning_rate": 9.864667073972758e-06,
+      "loss": 0.5391,
+      "step": 232
+    },
+    {
+      "epoch": 0.06714697406340057,
+      "grad_norm": 1.747241423930901,
+      "learning_rate": 9.872456827379283e-06,
+      "loss": 0.6686,
+      "step": 233
+    },
+    {
+      "epoch": 0.06743515850144093,
+      "grad_norm": 1.3701584391406454,
+      "learning_rate": 9.880213219855314e-06,
+      "loss": 0.6519,
+      "step": 234
+    },
+    {
+      "epoch": 0.06772334293948126,
+      "grad_norm": 3.4287773141998237,
+      "learning_rate": 9.887936535930344e-06,
+      "loss": 0.5332,
+      "step": 235
+    },
+    {
+      "epoch": 0.06801152737752161,
+      "grad_norm": 1.7901734286428725,
+      "learning_rate": 9.895627056509262e-06,
+      "loss": 0.6371,
+      "step": 236
+    },
+    {
+      "epoch": 0.06829971181556196,
+      "grad_norm": 1.6397325447306252,
+      "learning_rate": 9.90328505893366e-06,
+      "loss": 0.6518,
+      "step": 237
+    },
+    {
+      "epoch": 0.0685878962536023,
+      "grad_norm": 2.745535047975084,
+      "learning_rate": 9.91091081704185e-06,
+      "loss": 0.5285,
+      "step": 238
+    },
+    {
+      "epoch": 0.06887608069164265,
+      "grad_norm": 1.3871287337041953,
+      "learning_rate": 9.91850460122761e-06,
+      "loss": 0.6584,
+      "step": 239
+    },
+    {
+      "epoch": 0.069164265129683,
+      "grad_norm": 3.6251919092078766,
+      "learning_rate": 9.926066678497726e-06,
+      "loss": 0.6559,
+      "step": 240
+    },
+    {
+      "epoch": 0.06945244956772334,
+      "grad_norm": 1.982539491543836,
+      "learning_rate": 9.93359731252832e-06,
+      "loss": 0.5447,
+      "step": 241
+    },
+    {
+      "epoch": 0.06974063400576369,
+      "grad_norm": 3.5261780386103623,
+      "learning_rate": 9.941096763720006e-06,
+      "loss": 0.6706,
+      "step": 242
+    },
+    {
+      "epoch": 0.07002881844380403,
+      "grad_norm": 1.911397605573411,
+      "learning_rate": 9.948565289251939e-06,
+      "loss": 0.6375,
+      "step": 243
+    },
+    {
+      "epoch": 0.07031700288184438,
+      "grad_norm": 2.285007903188301,
+      "learning_rate": 9.956003143134718e-06,
+      "loss": 0.5516,
+      "step": 244
+    },
+    {
+      "epoch": 0.07060518731988473,
+      "grad_norm": 1.7565615568409982,
+      "learning_rate": 9.963410576262233e-06,
+      "loss": 0.6452,
+      "step": 245
+    },
+    {
+      "epoch": 0.07089337175792507,
+      "grad_norm": 1.9698534688810974,
+      "learning_rate": 9.97078783646244e-06,
+      "loss": 0.6603,
+      "step": 246
+    },
+    {
+      "epoch": 0.07118155619596542,
+      "grad_norm": 2.5102089169613193,
+      "learning_rate": 9.978135168547126e-06,
+      "loss": 0.5465,
+      "step": 247
+    },
+    {
+      "epoch": 0.07146974063400577,
+      "grad_norm": 1.7303136494447997,
+      "learning_rate": 9.985452814360637e-06,
+      "loss": 0.6296,
+      "step": 248
+    },
+    {
+      "epoch": 0.0717579250720461,
+      "grad_norm": 1.11740583936279,
+      "learning_rate": 9.992741012827652e-06,
+      "loss": 0.6626,
+      "step": 249
+    },
+    {
+      "epoch": 0.07204610951008646,
+      "grad_norm": 4.068639871871646,
+      "learning_rate": 1e-05,
+      "loss": 0.524,
+      "step": 250
+    },
+    {
+      "epoch": 0.07204610951008646,
+      "eval_websight_new_IoU": 0.005624114070087671,
+      "eval_websight_new_MAE_x": 305.29998779296875,
+      "eval_websight_new_MAE_y": 421.6199951171875,
+      "eval_websight_new_coord_probability": 0.717555820941925,
+      "eval_websight_new_correct_token_probability": 0.005016825627535582,
+      "eval_websight_new_inside_bbox": 0.019999999552965164,
+      "eval_websight_new_loss": 0.5373218059539795,
+      "eval_websight_new_range_100_probability": 0.11471053212881088,
+      "eval_websight_new_range_20_probability": 0.027784831821918488,
+      "eval_websight_new_range_5_probability": 0.009580916725099087,
+      "eval_websight_new_runtime": 34.5936,
+      "eval_websight_new_samples_per_second": 1.445,
+      "eval_websight_new_steps_per_second": 0.202,
+      "step": 250
+    },
+    {
+      "epoch": 0.07204610951008646,
+      "eval_seeclick_IoU": 0.0044020069763064384,
+      "eval_seeclick_MAE_x": 491.29998779296875,
+      "eval_seeclick_MAE_y": 601.5999755859375,
+      "eval_seeclick_coord_probability": 0.7230032086372375,
+      "eval_seeclick_correct_token_probability": 0.004017556551843882,
+      "eval_seeclick_inside_bbox": 0.05999999865889549,
+      "eval_seeclick_loss": 0.5848657488822937,
+      "eval_seeclick_range_100_probability": 0.12672168016433716,
+      "eval_seeclick_range_20_probability": 0.03310141712427139,
+      "eval_seeclick_range_5_probability": 0.009075821377336979,
+      "eval_seeclick_runtime": 42.2516,
+      "eval_seeclick_samples_per_second": 1.183,
+      "eval_seeclick_steps_per_second": 0.166,
+      "step": 250
+    },
+    {
+      "epoch": 0.07204610951008646,
+      "eval_icons_IoU": 0.0,
+      "eval_icons_MAE_x": 620.3400268554688,
+      "eval_icons_MAE_y": 428.2200012207031,
+      "eval_icons_coord_probability": 0.7095824480056763,
+      "eval_icons_correct_token_probability": 0.002493268111720681,
+      "eval_icons_inside_bbox": 0.0,
+      "eval_icons_loss": 0.6193642616271973,
+      "eval_icons_range_100_probability": 0.22341783344745636,
+      "eval_icons_range_20_probability": 0.06995800137519836,
+      "eval_icons_range_5_probability": 0.018928950652480125,
+      "eval_icons_runtime": 43.3165,
+      "eval_icons_samples_per_second": 1.154,
+      "eval_icons_steps_per_second": 0.162,
+      "step": 250
+    },
+    {
+      "epoch": 0.07233429394812681,
+      "grad_norm": 1.4228089719555008,
+      "learning_rate": 1e-05,
+      "loss": 0.6483,
+      "step": 251
+    },
+    {
+      "epoch": 0.07262247838616714,
+      "grad_norm": 1.4726753507806032,
+      "learning_rate": 1e-05,
+      "loss": 0.6351,
+      "step": 252
+    },
+    {
+      "epoch": 0.0729106628242075,
+      "grad_norm": 2.5730722722107267,
+      "learning_rate": 1e-05,
+      "loss": 0.5223,
+      "step": 253
+    },
+    {
+      "epoch": 0.07319884726224783,
+      "grad_norm": 1.6668812999341671,
+      "learning_rate": 1e-05,
+      "loss": 0.6679,
+      "step": 254
+    },
+    {
+      "epoch": 0.07348703170028818,
+      "grad_norm": 1.2513607195413772,
+      "learning_rate": 1e-05,
+      "loss": 0.6325,
+      "step": 255
+    },
+    {
+      "epoch": 0.07377521613832853,
+      "grad_norm": 4.071539500131123,
+      "learning_rate": 1e-05,
+      "loss": 0.5157,
+      "step": 256
+    },
+    {
+      "epoch": 0.07406340057636887,
+      "grad_norm": 2.7672062492774874,
+      "learning_rate": 1e-05,
+      "loss": 0.6494,
+      "step": 257
+    },
+    {
+      "epoch": 0.07435158501440922,
+      "grad_norm": 1.5093154739351127,
+      "learning_rate": 1e-05,
+      "loss": 0.6371,
+      "step": 258
+    },
+    {
+      "epoch": 0.07463976945244957,
+      "grad_norm": 2.6997085498081073,
+      "learning_rate": 1e-05,
+      "loss": 0.5089,
+      "step": 259
+    },
+    {
+      "epoch": 0.07492795389048991,
+      "grad_norm": 2.533984042380557,
+      "learning_rate": 1e-05,
+      "loss": 0.6263,
+      "step": 260
+    },
+    {
+      "epoch": 0.07521613832853026,
+      "grad_norm": 1.3204406580989467,
+      "learning_rate": 1e-05,
+      "loss": 0.6447,
+      "step": 261
+    },
+    {
+      "epoch": 0.07550432276657061,
+      "grad_norm": 2.0059628945560593,
+      "learning_rate": 1e-05,
+      "loss": 0.5287,
+      "step": 262
+    },
+    {
+      "epoch": 0.07579250720461095,
+      "grad_norm": 2.055654119489871,
+      "learning_rate": 1e-05,
+      "loss": 0.6279,
+      "step": 263
+    },
+    {
+      "epoch": 0.0760806916426513,
+      "grad_norm": 1.6448620459151915,
+      "learning_rate": 1e-05,
+      "loss": 0.6352,
+      "step": 264
+    },
+    {
+      "epoch": 0.07636887608069164,
+      "grad_norm": 2.325043774953538,
+      "learning_rate": 1e-05,
+      "loss": 0.5142,
+      "step": 265
+    },
+    {
+      "epoch": 0.07665706051873199,
+      "grad_norm": 1.5225370910236076,
+      "learning_rate": 1e-05,
+      "loss": 0.6399,
+      "step": 266
+    },
+    {
+      "epoch": 0.07694524495677234,
+      "grad_norm": 1.8189992608714503,
+      "learning_rate": 1e-05,
+      "loss": 0.6533,
+      "step": 267
+    },
+    {
+      "epoch": 0.07723342939481267,
+      "grad_norm": 2.1680286049966604,
+      "learning_rate": 1e-05,
+      "loss": 0.5007,
+      "step": 268
+    },
+    {
+      "epoch": 0.07752161383285303,
+      "grad_norm": 3.9033303562018555,
+      "learning_rate": 1e-05,
+      "loss": 0.6512,
+      "step": 269
+    },
+    {
+      "epoch": 0.07780979827089338,
+      "grad_norm": 1.0489029684926237,
+      "learning_rate": 1e-05,
+      "loss": 0.6411,
+      "step": 270
+    },
+    {
+      "epoch": 0.07809798270893371,
+      "grad_norm": 3.264163568820453,
+      "learning_rate": 1e-05,
+      "loss": 0.5221,
+      "step": 271
+    },
+    {
+      "epoch": 0.07838616714697406,
+      "grad_norm": 1.8691632613548,
+      "learning_rate": 1e-05,
+      "loss": 0.6288,
+      "step": 272
+    },
+    {
+      "epoch": 0.07867435158501442,
+      "grad_norm": 1.2243714190429924,
+      "learning_rate": 1e-05,
+      "loss": 0.6447,
+      "step": 273
+    },
+    {
+      "epoch": 0.07896253602305475,
+      "grad_norm": 2.578411914899318,
+      "learning_rate": 1e-05,
+      "loss": 0.5224,
+      "step": 274
+    },
+    {
+      "epoch": 0.0792507204610951,
+      "grad_norm": 1.4950396387903995,
+      "learning_rate": 1e-05,
+      "loss": 0.648,
+      "step": 275
+    },
+    {
+      "epoch": 0.07953890489913544,
+      "grad_norm": 1.5020358400986247,
+      "learning_rate": 1e-05,
+      "loss": 0.6502,
+      "step": 276
+    },
+    {
+      "epoch": 0.07982708933717579,
+      "grad_norm": 1.9186735356285112,
+      "learning_rate": 1e-05,
+      "loss": 0.5023,
+      "step": 277
+    },
+    {
+      "epoch": 0.08011527377521614,
+      "grad_norm": 1.5322143468026472,
+      "learning_rate": 1e-05,
+      "loss": 0.6347,
+      "step": 278
+    },
+    {
+      "epoch": 0.08040345821325648,
+      "grad_norm": 0.9374030254644127,
+      "learning_rate": 1e-05,
+      "loss": 0.6476,
+      "step": 279
+    },
+    {
+      "epoch": 0.08069164265129683,
+      "grad_norm": 2.453546930013716,
+      "learning_rate": 1e-05,
+      "loss": 0.5115,
+      "step": 280
+    },
+    {
+      "epoch": 0.08097982708933718,
+      "grad_norm": 1.525975667252706,
+      "learning_rate": 1e-05,
+      "loss": 0.6151,
+      "step": 281
+    },
+    {
+      "epoch": 0.08126801152737752,
+      "grad_norm": 1.0710752976867846,
+      "learning_rate": 1e-05,
+      "loss": 0.6447,
+      "step": 282
+    },
+    {
+      "epoch": 0.08155619596541787,
+      "grad_norm": 2.322551664301356,
+      "learning_rate": 1e-05,
+      "loss": 0.495,
+      "step": 283
+    },
+    {
+      "epoch": 0.08184438040345822,
+      "grad_norm": 1.7580633577114375,
+      "learning_rate": 1e-05,
+      "loss": 0.6064,
+      "step": 284
+    },
+    {
+      "epoch": 0.08213256484149856,
+      "grad_norm": 0.98115460737944,
+      "learning_rate": 1e-05,
+      "loss": 0.6263,
+      "step": 285
+    },
+    {
+      "epoch": 0.08242074927953891,
+      "grad_norm": 1.9173191103169804,
+      "learning_rate": 1e-05,
+      "loss": 0.4884,
+      "step": 286
+    },
+    {
+      "epoch": 0.08270893371757924,
+      "grad_norm": 2.2318016275660106,
+      "learning_rate": 1e-05,
+      "loss": 0.6324,
+      "step": 287
+    },
+    {
+      "epoch": 0.0829971181556196,
+      "grad_norm": 1.1124465074110315,
+      "learning_rate": 1e-05,
+      "loss": 0.6226,
+      "step": 288
+    },
+    {
+      "epoch": 0.08328530259365995,
+      "grad_norm": 2.4864735694052196,
+      "learning_rate": 1e-05,
+      "loss": 0.5043,
+      "step": 289
+    },
+    {
+      "epoch": 0.08357348703170028,
+      "grad_norm": 1.6794470102067223,
+      "learning_rate": 1e-05,
+      "loss": 0.6325,
+      "step": 290
+    },
+    {
+      "epoch": 0.08386167146974063,
+      "grad_norm": 1.4315168581948596,
+      "learning_rate": 1e-05,
+      "loss": 0.6229,
+      "step": 291
+    },
+    {
+      "epoch": 0.08414985590778098,
+      "grad_norm": 3.4610927042017146,
+      "learning_rate": 1e-05,
+      "loss": 0.5054,
+      "step": 292
+    },
+    {
+      "epoch": 0.08443804034582132,
+      "grad_norm": 2.7295557683706115,
+      "learning_rate": 1e-05,
+      "loss": 0.6311,
+      "step": 293
+    },
+    {
+      "epoch": 0.08472622478386167,
+      "grad_norm": 1.0191280989324512,
+      "learning_rate": 1e-05,
+      "loss": 0.6217,
+      "step": 294
+    },
+    {
+      "epoch": 0.08501440922190202,
+      "grad_norm": 1.9113003308196765,
+      "learning_rate": 1e-05,
+      "loss": 0.497,
+      "step": 295
+    },
+    {
+      "epoch": 0.08530259365994236,
+      "grad_norm": 1.5293635470357692,
+      "learning_rate": 1e-05,
+      "loss": 0.5931,
+      "step": 296
+    },
+    {
+      "epoch": 0.08559077809798271,
+      "grad_norm": 1.0296246850281991,
+      "learning_rate": 1e-05,
+      "loss": 0.6092,
+      "step": 297
+    },
+    {
+      "epoch": 0.08587896253602305,
+      "grad_norm": 2.4398890133828863,
+      "learning_rate": 1e-05,
+      "loss": 0.5003,
+      "step": 298
+    },
+    {
+      "epoch": 0.0861671469740634,
+      "grad_norm": 1.36298863155866,
+      "learning_rate": 1e-05,
+      "loss": 0.6619,
+      "step": 299
+    },
+    {
+      "epoch": 0.08645533141210375,
+      "grad_norm": 1.0616564773227006,
+      "learning_rate": 1e-05,
+      "loss": 0.6138,
+      "step": 300
+    },
+    {
+      "epoch": 0.08674351585014409,
+      "grad_norm": 1.8635512737499922,
+      "learning_rate": 1e-05,
+      "loss": 0.5046,
+      "step": 301
+    },
+    {
+      "epoch": 0.08703170028818444,
+      "grad_norm": 2.196386789113054,
+      "learning_rate": 1e-05,
+      "loss": 0.666,
+      "step": 302
+    },
+    {
+      "epoch": 0.08731988472622479,
+      "grad_norm": 1.1206758236967824,
+      "learning_rate": 1e-05,
+      "loss": 0.6356,
+      "step": 303
+    },
+    {
+      "epoch": 0.08760806916426513,
+      "grad_norm": 2.2610129339817098,
+      "learning_rate": 1e-05,
+      "loss": 0.4975,
+      "step": 304
+    },
+    {
+      "epoch": 0.08789625360230548,
+      "grad_norm": 2.9824324925991728,
+      "learning_rate": 1e-05,
+      "loss": 0.6007,
+      "step": 305
+    },
+    {
+      "epoch": 0.08818443804034583,
+      "grad_norm": 1.2488543260255234,
+      "learning_rate": 1e-05,
+      "loss": 0.6167,
+      "step": 306
+    },
+    {
+      "epoch": 0.08847262247838616,
+      "grad_norm": 2.425461246914256,
+      "learning_rate": 1e-05,
+      "loss": 0.5042,
+      "step": 307
+    },
+    {
+      "epoch": 0.08876080691642652,
+      "grad_norm": 2.538485398849825,
+      "learning_rate": 1e-05,
+      "loss": 0.5771,
+      "step": 308
+    },
+    {
+      "epoch": 0.08904899135446685,
+      "grad_norm": 1.4754723688947828,
+      "learning_rate": 1e-05,
+      "loss": 0.6217,
+      "step": 309
+    },
+    {
+      "epoch": 0.0893371757925072,
+      "grad_norm": 1.4215133658462726,
+      "learning_rate": 1e-05,
+      "loss": 0.5053,
+      "step": 310
+    },
+    {
+      "epoch": 0.08962536023054755,
+      "grad_norm": 4.144241341252387,
+      "learning_rate": 1e-05,
+      "loss": 0.602,
+      "step": 311
+    },
+    {
+      "epoch": 0.08991354466858789,
+      "grad_norm": 1.2942442719728213,
+      "learning_rate": 1e-05,
+      "loss": 0.6281,
+      "step": 312
+    },
+    {
+      "epoch": 0.09020172910662824,
+      "grad_norm": 3.009460634432624,
+      "learning_rate": 1e-05,
+      "loss": 0.5039,
+      "step": 313
+    },
+    {
+      "epoch": 0.0904899135446686,
+      "grad_norm": 1.96286308394569,
+      "learning_rate": 1e-05,
+      "loss": 0.6256,
+      "step": 314
+    },
+    {
+      "epoch": 0.09077809798270893,
+      "grad_norm": 1.2472287157708806,
+      "learning_rate": 1e-05,
+      "loss": 0.6124,
+      "step": 315
+    },
+    {
+      "epoch": 0.09106628242074928,
+      "grad_norm": 1.9483588310709774,
+      "learning_rate": 1e-05,
+      "loss": 0.4877,
+      "step": 316
+    },
+    {
+      "epoch": 0.09135446685878963,
+      "grad_norm": 2.539116120554538,
+      "learning_rate": 1e-05,
+      "loss": 0.6175,
+      "step": 317
+    },
+    {
+      "epoch": 0.09164265129682997,
+      "grad_norm": 0.9162670920016374,
+      "learning_rate": 1e-05,
+      "loss": 0.604,
+      "step": 318
+    },
+    {
+      "epoch": 0.09193083573487032,
+      "grad_norm": 3.9023447397861357,
+      "learning_rate": 1e-05,
+      "loss": 0.4915,
+      "step": 319
+    },
+    {
+      "epoch": 0.09221902017291066,
+      "grad_norm": 3.676239398664723,
+      "learning_rate": 1e-05,
+      "loss": 0.6053,
+      "step": 320
+    },
+    {
+      "epoch": 0.09250720461095101,
+      "grad_norm": 1.2454556899715417,
+      "learning_rate": 1e-05,
+      "loss": 0.6014,
+      "step": 321
+    },
+    {
+      "epoch": 0.09279538904899136,
+      "grad_norm": 1.2458713787172075,
+      "learning_rate": 1e-05,
+      "loss": 0.4911,
+      "step": 322
+    },
+    {
+      "epoch": 0.0930835734870317,
+      "grad_norm": 1.866125511189336,
+      "learning_rate": 1e-05,
+      "loss": 0.5944,
+      "step": 323
+    },
+    {
+      "epoch": 0.09337175792507205,
+      "grad_norm": 1.190775702837205,
+      "learning_rate": 1e-05,
+      "loss": 0.6043,
+      "step": 324
+    },
+    {
+      "epoch": 0.0936599423631124,
+      "grad_norm": 2.197604572030403,
+      "learning_rate": 1e-05,
+      "loss": 0.4878,
+      "step": 325
+    },
+    {
+      "epoch": 0.09394812680115273,
+      "grad_norm": 1.6586113238955515,
+      "learning_rate": 1e-05,
+      "loss": 0.6132,
+      "step": 326
+    },
+    {
+      "epoch": 0.09423631123919309,
+      "grad_norm": 1.1165382336552463,
+      "learning_rate": 1e-05,
+      "loss": 0.6293,
+      "step": 327
+    },
+    {
+      "epoch": 0.09452449567723344,
+      "grad_norm": 2.970713124920137,
+      "learning_rate": 1e-05,
+      "loss": 0.4946,
+      "step": 328
+    },
+    {
+      "epoch": 0.09481268011527377,
+      "grad_norm": 4.388834623628037,
+      "learning_rate": 1e-05,
+      "loss": 0.6313,
+      "step": 329
+    },
+    {
+      "epoch": 0.09510086455331412,
+      "grad_norm": 1.1696448667484844,
+      "learning_rate": 1e-05,
+      "loss": 0.6282,
+      "step": 330
+    },
+    {
+      "epoch": 0.09538904899135446,
+      "grad_norm": 1.7819396121412316,
+      "learning_rate": 1e-05,
+      "loss": 0.4802,
+      "step": 331
+    },
+    {
+      "epoch": 0.09567723342939481,
+      "grad_norm": 1.819061336829697,
+      "learning_rate": 1e-05,
+      "loss": 0.5983,
+      "step": 332
+    },
+    {
+      "epoch": 0.09596541786743516,
+      "grad_norm": 0.9982024428129266,
+      "learning_rate": 1e-05,
+      "loss": 0.611,
+      "step": 333
+    },
+    {
+      "epoch": 0.0962536023054755,
+      "grad_norm": 2.0853295943930963,
+      "learning_rate": 1e-05,
+      "loss": 0.4803,
+      "step": 334
+    },
+    {
+      "epoch": 0.09654178674351585,
+      "grad_norm": 2.2050193425368882,
+      "learning_rate": 1e-05,
+      "loss": 0.596,
+      "step": 335
+    },
+    {
+      "epoch": 0.0968299711815562,
+      "grad_norm": 1.106955298911173,
+      "learning_rate": 1e-05,
+      "loss": 0.62,
+      "step": 336
+    },
+    {
+      "epoch": 0.09711815561959654,
+      "grad_norm": 1.5018103554984554,
+      "learning_rate": 1e-05,
+      "loss": 0.493,
+      "step": 337
+    },
+    {
+      "epoch": 0.09740634005763689,
+      "grad_norm": 3.096782380894111,
+      "learning_rate": 1e-05,
+      "loss": 0.6066,
+      "step": 338
+    },
+    {
+      "epoch": 0.09769452449567724,
+      "grad_norm": 1.109033538155625,
+      "learning_rate": 1e-05,
+      "loss": 0.6185,
+      "step": 339
+    },
+    {
+      "epoch": 0.09798270893371758,
+      "grad_norm": 2.342402894228767,
+      "learning_rate": 1e-05,
+      "loss": 0.4814,
+      "step": 340
+    },
+    {
+      "epoch": 0.09827089337175793,
+      "grad_norm": 2.2503418644206037,
+      "learning_rate": 1e-05,
+      "loss": 0.6239,
+      "step": 341
+    },
+    {
+      "epoch": 0.09855907780979827,
+      "grad_norm": 1.46191303710194,
+      "learning_rate": 1e-05,
+      "loss": 0.6265,
+      "step": 342
+    },
+    {
+      "epoch": 0.09884726224783862,
+      "grad_norm": 3.1383944100395422,
+      "learning_rate": 1e-05,
+      "loss": 0.4819,
+      "step": 343
+    },
+    {
+      "epoch": 0.09913544668587897,
+      "grad_norm": 2.7192549497220133,
+      "learning_rate": 1e-05,
+      "loss": 0.6198,
+      "step": 344
+    },
+    {
+      "epoch": 0.0994236311239193,
+      "grad_norm": 1.5945490264235185,
+      "learning_rate": 1e-05,
+      "loss": 0.6076,
+      "step": 345
+    },
+    {
+      "epoch": 0.09971181556195965,
+      "grad_norm": 1.4625706633703939,
+      "learning_rate": 1e-05,
+      "loss": 0.4706,
+      "step": 346
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4528442192556827,
+      "learning_rate": 1e-05,
+      "loss": 0.616,
+      "step": 347
+    },
+    {
+      "epoch": 0.10028818443804034,
+      "grad_norm": 1.0813691773744674,
+      "learning_rate": 1e-05,
+      "loss": 0.5994,
+      "step": 348
+    },
+    {
+      "epoch": 0.1005763688760807,
+      "grad_norm": 3.8456540718069,
+      "learning_rate": 1e-05,
+      "loss": 0.4725,
+      "step": 349
+    },
+    {
+      "epoch": 0.10086455331412104,
+      "grad_norm": 2.4854019549048316,
+      "learning_rate": 1e-05,
+      "loss": 0.5904,
+      "step": 350
+    },
+    {
+      "epoch": 0.10115273775216138,
+      "grad_norm": 1.0716853325250935,
+      "learning_rate": 1e-05,
+      "loss": 0.6085,
+      "step": 351
+    },
+    {
+      "epoch": 0.10144092219020173,
+      "grad_norm": 1.922098976464248,
+      "learning_rate": 1e-05,
+      "loss": 0.4914,
+      "step": 352
+    },
+    {
+      "epoch": 0.10172910662824207,
+      "grad_norm": 3.2765105360887437,
+      "learning_rate": 1e-05,
+      "loss": 0.5768,
+      "step": 353
+    },
+    {
+      "epoch": 0.10201729106628242,
+      "grad_norm": 1.9159363681527248,
+      "learning_rate": 1e-05,
+      "loss": 0.603,
+      "step": 354
+    },
+    {
+      "epoch": 0.10230547550432277,
+      "grad_norm": 1.2291699169535677,
+      "learning_rate": 1e-05,
+      "loss": 0.4772,
+      "step": 355
+    },
+    {
+      "epoch": 0.10259365994236311,
+      "grad_norm": 1.7153489314835757,
+      "learning_rate": 1e-05,
+      "loss": 0.5885,
+      "step": 356
+    },
+    {
+      "epoch": 0.10288184438040346,
+      "grad_norm": 1.038416686137203,
+      "learning_rate": 1e-05,
+      "loss": 0.6023,
+      "step": 357
+    },
+    {
+      "epoch": 0.10317002881844381,
+      "grad_norm": 2.3964872434191378,
+      "learning_rate": 1e-05,
+      "loss": 0.4559,
+      "step": 358
+    },
+    {
+      "epoch": 0.10345821325648415,
+      "grad_norm": 2.2157046001432477,
+      "learning_rate": 1e-05,
+      "loss": 0.6163,
+      "step": 359
+    },
+    {
+      "epoch": 0.1037463976945245,
+      "grad_norm": 1.6549497014322723,
+      "learning_rate": 1e-05,
+      "loss": 0.6095,
+      "step": 360
+    },
+    {
+      "epoch": 0.10403458213256485,
+      "grad_norm": 2.109594268757546,
+      "learning_rate": 1e-05,
+      "loss": 0.4789,
+      "step": 361
+    },
+    {
+      "epoch": 0.10432276657060519,
+      "grad_norm": 2.33695262239474,
+      "learning_rate": 1e-05,
+      "loss": 0.5936,
+      "step": 362
+    },
+    {
+      "epoch": 0.10461095100864554,
+      "grad_norm": 1.0480109993898619,
+      "learning_rate": 1e-05,
+      "loss": 0.6026,
+      "step": 363
+    },
+    {
+      "epoch": 0.10489913544668587,
+      "grad_norm": 1.7960215966295159,
+      "learning_rate": 1e-05,
+      "loss": 0.4674,
+      "step": 364
+    },
+    {
+      "epoch": 0.10518731988472622,
+      "grad_norm": 1.568070753765201,
+      "learning_rate": 1e-05,
+      "loss": 0.5882,
+      "step": 365
+    },
+    {
+      "epoch": 0.10547550432276658,
+      "grad_norm": 0.9702989899863376,
+      "learning_rate": 1e-05,
+      "loss": 0.5989,
+      "step": 366
+    },
+    {
+      "epoch": 0.10576368876080691,
+      "grad_norm": 2.0281939963414506,
+      "learning_rate": 1e-05,
+      "loss": 0.4724,
+      "step": 367
+    },
+    {
+      "epoch": 0.10605187319884726,
+      "grad_norm": 2.838118600581876,
+      "learning_rate": 1e-05,
+      "loss": 0.6056,
+      "step": 368
+    },
+    {
+      "epoch": 0.10634005763688761,
+      "grad_norm": 1.2565327880090889,
+      "learning_rate": 1e-05,
+      "loss": 0.6003,
+      "step": 369
+    },
+    {
+      "epoch": 0.10662824207492795,
+      "grad_norm": 7.053305951324435,
+      "learning_rate": 1e-05,
+      "loss": 0.4619,
+      "step": 370
+    },
+    {
+      "epoch": 0.1069164265129683,
+      "grad_norm": 2.979895739358605,
+      "learning_rate": 1e-05,
+      "loss": 0.6202,
+      "step": 371
+    },
+    {
+      "epoch": 0.10720461095100864,
+      "grad_norm": 1.0289485571866739,
+      "learning_rate": 1e-05,
+      "loss": 0.6135,
+      "step": 372
+    },
+    {
+      "epoch": 0.10749279538904899,
+      "grad_norm": 2.219135041670798,
+      "learning_rate": 1e-05,
+      "loss": 0.4634,
+      "step": 373
+    },
+    {
+      "epoch": 0.10778097982708934,
+      "grad_norm": 3.53549590626335,
+      "learning_rate": 1e-05,
+      "loss": 0.611,
+      "step": 374
+    },
+    {
+      "epoch": 0.10806916426512968,
+      "grad_norm": 1.3712409115660724,
+      "learning_rate": 1e-05,
+      "loss": 0.6088,
+      "step": 375
+    },
+    {
+      "epoch": 0.10835734870317003,
+      "grad_norm": 3.2608083851225844,
+      "learning_rate": 1e-05,
+      "loss": 0.4678,
+      "step": 376
+    },
+    {
+      "epoch": 0.10864553314121038,
+      "grad_norm": 2.533467816629062,
+      "learning_rate": 1e-05,
+      "loss": 0.5746,
+      "step": 377
+    },
+    {
+      "epoch": 0.10893371757925072,
+      "grad_norm": 1.089968662571353,
+      "learning_rate": 1e-05,
+      "loss": 0.6114,
+      "step": 378
+    },
+    {
+      "epoch": 0.10922190201729107,
+      "grad_norm": 2.334185467609639,
+      "learning_rate": 1e-05,
+      "loss": 0.4675,
+      "step": 379
+    },
+    {
+      "epoch": 0.10951008645533142,
+      "grad_norm": 4.320804894689986,
+      "learning_rate": 1e-05,
+      "loss": 0.5976,
+      "step": 380
+    },
+    {
+      "epoch": 0.10979827089337175,
+      "grad_norm": 1.5020827021639047,
+      "learning_rate": 1e-05,
+      "loss": 0.5988,
+      "step": 381
+    },
+    {
+      "epoch": 0.1100864553314121,
+      "grad_norm": 2.9204916922627935,
+      "learning_rate": 1e-05,
+      "loss": 0.4747,
+      "step": 382
+    },
+    {
+      "epoch": 0.11037463976945244,
+      "grad_norm": 1.937969689473588,
+      "learning_rate": 1e-05,
+      "loss": 0.5801,
+      "step": 383
+    },
+    {
+      "epoch": 0.1106628242074928,
+      "grad_norm": 1.146181403848022,
+      "learning_rate": 1e-05,
+      "loss": 0.61,
+      "step": 384
+    },
+    {
+      "epoch": 0.11095100864553314,
+      "grad_norm": 2.5044582375638473,
+      "learning_rate": 1e-05,
+      "loss": 0.4579,
+      "step": 385
+    },
+    {
+      "epoch": 0.11123919308357348,
+      "grad_norm": 2.5779323013366646,
+      "learning_rate": 1e-05,
+      "loss": 0.58,
+      "step": 386
+    },
+    {
+      "epoch": 0.11152737752161383,
+      "grad_norm": 1.5585121429585433,
+      "learning_rate": 1e-05,
+      "loss": 0.5955,
+      "step": 387
+    },
+    {
+      "epoch": 0.11181556195965418,
+      "grad_norm": 3.0977720667830804,
+      "learning_rate": 1e-05,
+      "loss": 0.4567,
+      "step": 388
+    },
+    {
+      "epoch": 0.11210374639769452,
+      "grad_norm": 2.155885125341923,
+      "learning_rate": 1e-05,
+      "loss": 0.5899,
+      "step": 389
+    },
+    {
+      "epoch": 0.11239193083573487,
+      "grad_norm": 1.1642592175270015,
+      "learning_rate": 1e-05,
+      "loss": 0.6091,
+      "step": 390
+    },
+    {
+      "epoch": 0.11268011527377522,
+      "grad_norm": 1.4538116292580978,
+      "learning_rate": 1e-05,
+      "loss": 0.4796,
+      "step": 391
+    },
+    {
+      "epoch": 0.11296829971181556,
+      "grad_norm": 1.4638013421860472,
+      "learning_rate": 1e-05,
+      "loss": 0.6164,
+      "step": 392
+    },
+    {
+      "epoch": 0.11325648414985591,
+      "grad_norm": 1.0857945976367118,
+      "learning_rate": 1e-05,
+      "loss": 0.5894,
+      "step": 393
+    },
+    {
+      "epoch": 0.11354466858789625,
+      "grad_norm": 3.472121988951218,
+      "learning_rate": 1e-05,
+      "loss": 0.4398,
+      "step": 394
+    },
+    {
+      "epoch": 0.1138328530259366,
+      "grad_norm": 2.389754760861759,
+      "learning_rate": 1e-05,
+      "loss": 0.5925,
+      "step": 395
+    },
+    {
+      "epoch": 0.11412103746397695,
+      "grad_norm": 1.5297289713983515,
+      "learning_rate": 1e-05,
+      "loss": 0.6061,
+      "step": 396
+    },
+    {
+      "epoch": 0.11440922190201729,
+      "grad_norm": 2.1975425756434346,
+      "learning_rate": 1e-05,
+      "loss": 0.4559,
+      "step": 397
+    },
+    {
+      "epoch": 0.11469740634005764,
+      "grad_norm": 1.8118169367788128,
+      "learning_rate": 1e-05,
+      "loss": 0.6079,
+      "step": 398
+    },
+    {
+      "epoch": 0.11498559077809799,
+      "grad_norm": 1.2899700312487847,
+      "learning_rate": 1e-05,
+      "loss": 0.6056,
+      "step": 399
+    },
+    {
+      "epoch": 0.11527377521613832,
+      "grad_norm": 2.878458234124749,
+      "learning_rate": 1e-05,
+      "loss": 0.4528,
+      "step": 400
+    },
+    {
+      "epoch": 0.11556195965417868,
+      "grad_norm": 3.2408021931829905,
+      "learning_rate": 1e-05,
+      "loss": 0.5949,
+      "step": 401
+    },
+    {
+      "epoch": 0.11585014409221903,
+      "grad_norm": 1.1833224811328793,
+      "learning_rate": 1e-05,
+      "loss": 0.5915,
+      "step": 402
+    },
+    {
+      "epoch": 0.11613832853025936,
+      "grad_norm": 3.6325309431722776,
+      "learning_rate": 1e-05,
+      "loss": 0.4572,
+      "step": 403
+    },
+    {
+      "epoch": 0.11642651296829971,
+      "grad_norm": 3.063499009766476,
+      "learning_rate": 1e-05,
+      "loss": 0.619,
+      "step": 404
+    },
+    {
+      "epoch": 0.11671469740634005,
+      "grad_norm": 1.2112833482431893,
+      "learning_rate": 1e-05,
+      "loss": 0.5913,
+      "step": 405
+    },
+    {
+      "epoch": 0.1170028818443804,
+      "grad_norm": 1.896771682236592,
+      "learning_rate": 1e-05,
+      "loss": 0.4453,
+      "step": 406
+    },
+    {
+      "epoch": 0.11729106628242075,
+      "grad_norm": 2.4464259596256586,
+      "learning_rate": 1e-05,
+      "loss": 0.6023,
+      "step": 407
+    },
+    {
+      "epoch": 0.11757925072046109,
+      "grad_norm": 1.0719402348113563,
+      "learning_rate": 1e-05,
+      "loss": 0.5975,
+      "step": 408
+    },
+    {
+      "epoch": 0.11786743515850144,
+      "grad_norm": 1.6364604305431,
+      "learning_rate": 1e-05,
+      "loss": 0.4546,
+      "step": 409
+    },
+    {
+      "epoch": 0.11815561959654179,
+      "grad_norm": 1.9226298920457043,
+      "learning_rate": 1e-05,
+      "loss": 0.5961,
+      "step": 410
+    },
+    {
+      "epoch": 0.11844380403458213,
+      "grad_norm": 1.8746477254601188,
+      "learning_rate": 1e-05,
+      "loss": 0.6082,
+      "step": 411
+    },
+    {
+      "epoch": 0.11873198847262248,
+      "grad_norm": 2.9894133572547164,
+      "learning_rate": 1e-05,
+      "loss": 0.4684,
+      "step": 412
+    },
+    {
+      "epoch": 0.11902017291066283,
+      "grad_norm": 2.424736405906688,
+      "learning_rate": 1e-05,
+      "loss": 0.5914,
+      "step": 413
+    },
+    {
+      "epoch": 0.11930835734870317,
+      "grad_norm": 1.0130518532973238,
+      "learning_rate": 1e-05,
+      "loss": 0.597,
+      "step": 414
+    },
+    {
+      "epoch": 0.11959654178674352,
+      "grad_norm": 2.5468271019669584,
+      "learning_rate": 1e-05,
+      "loss": 0.4359,
+      "step": 415
+    },
+    {
+      "epoch": 0.11988472622478386,
+      "grad_norm": 3.017222974339237,
+      "learning_rate": 1e-05,
+      "loss": 0.5772,
+      "step": 416
+    },
+    {
+      "epoch": 0.1201729106628242,
+      "grad_norm": 1.1193530816512283,
+      "learning_rate": 1e-05,
+      "loss": 0.6038,
+      "step": 417
+    },
+    {
+      "epoch": 0.12046109510086456,
+      "grad_norm": 2.762433222336846,
+      "learning_rate": 1e-05,
+      "loss": 0.4539,
+      "step": 418
+    },
+    {
+      "epoch": 0.1207492795389049,
+      "grad_norm": 2.6483256028210294,
+      "learning_rate": 1e-05,
+      "loss": 0.5616,
+      "step": 419
+    },
+    {
+      "epoch": 0.12103746397694524,
+      "grad_norm": 1.2060796510223488,
+      "learning_rate": 1e-05,
+      "loss": 0.5961,
+      "step": 420
+    },
+    {
+      "epoch": 0.1213256484149856,
+      "grad_norm": 2.6985162932794675,
+      "learning_rate": 1e-05,
+      "loss": 0.4467,
+      "step": 421
+    },
+    {
+      "epoch": 0.12161383285302593,
+      "grad_norm": 2.379383849739707,
+      "learning_rate": 1e-05,
+      "loss": 0.5801,
+      "step": 422
+    },
+    {
+      "epoch": 0.12190201729106628,
+      "grad_norm": 1.652316386479186,
+      "learning_rate": 1e-05,
+      "loss": 0.5923,
+      "step": 423
+    },
+    {
+      "epoch": 0.12219020172910663,
+      "grad_norm": 2.679015156162535,
+      "learning_rate": 1e-05,
+      "loss": 0.4607,
+      "step": 424
+    },
+    {
+      "epoch": 0.12247838616714697,
+      "grad_norm": 2.786695020448989,
+      "learning_rate": 1e-05,
+      "loss": 0.5573,
+      "step": 425
+    },
+    {
+      "epoch": 0.12276657060518732,
+      "grad_norm": 1.2997492162885074,
+      "learning_rate": 1e-05,
+      "loss": 0.5965,
+      "step": 426
+    },
+    {
+      "epoch": 0.12305475504322766,
+      "grad_norm": 2.6794316898764845,
+      "learning_rate": 1e-05,
+      "loss": 0.4543,
+      "step": 427
+    },
+    {
+      "epoch": 0.12334293948126801,
+      "grad_norm": 2.243329329192076,
+      "learning_rate": 1e-05,
+      "loss": 0.5881,
+      "step": 428
+    },
+    {
+      "epoch": 0.12363112391930836,
+      "grad_norm": 1.3668366873798132,
+      "learning_rate": 1e-05,
+      "loss": 0.5906,
+      "step": 429
+    },
+    {
+      "epoch": 0.1239193083573487,
+      "grad_norm": 2.8724616352195658,
+      "learning_rate": 1e-05,
+      "loss": 0.4461,
+      "step": 430
+    },
+    {
+      "epoch": 0.12420749279538905,
+      "grad_norm": 3.3669598621394816,
+      "learning_rate": 1e-05,
+      "loss": 0.5512,
+      "step": 431
+    },
+    {
+      "epoch": 0.1244956772334294,
+      "grad_norm": 1.7398174447091903,
+      "learning_rate": 1e-05,
+      "loss": 0.5987,
+      "step": 432
+    },
+    {
+      "epoch": 0.12478386167146974,
+      "grad_norm": 2.125126459710019,
+      "learning_rate": 1e-05,
+      "loss": 0.4399,
+      "step": 433
+    },
+    {
+      "epoch": 0.1250720461095101,
+      "grad_norm": 1.9668441023227823,
+      "learning_rate": 1e-05,
+      "loss": 0.5572,
+      "step": 434
+    },
+    {
+      "epoch": 0.12536023054755044,
+      "grad_norm": 1.4135472721759272,
+      "learning_rate": 1e-05,
+      "loss": 0.5917,
+      "step": 435
+    },
+    {
+      "epoch": 0.1256484149855908,
+      "grad_norm": 2.6557482752537354,
+      "learning_rate": 1e-05,
+      "loss": 0.4216,
+      "step": 436
+    },
+    {
+      "epoch": 0.1259365994236311,
+      "grad_norm": 1.9071029220151672,
+      "learning_rate": 1e-05,
+      "loss": 0.5837,
+      "step": 437
+    },
+    {
+      "epoch": 0.12622478386167146,
+      "grad_norm": 1.5418422732583796,
+      "learning_rate": 1e-05,
+      "loss": 0.6133,
+      "step": 438
+    },
+    {
+      "epoch": 0.12651296829971181,
+      "grad_norm": 3.490914074598071,
+      "learning_rate": 1e-05,
+      "loss": 0.43,
+      "step": 439
+    },
+    {
+      "epoch": 0.12680115273775217,
+      "grad_norm": 2.6635869698716643,
+      "learning_rate": 1e-05,
+      "loss": 0.5967,
+      "step": 440
+    },
+    {
+      "epoch": 0.12708933717579252,
+      "grad_norm": 1.0604435551098819,
+      "learning_rate": 1e-05,
+      "loss": 0.5985,
+      "step": 441
+    },
+    {
+      "epoch": 0.12737752161383284,
+      "grad_norm": 3.469986231723641,
+      "learning_rate": 1e-05,
+      "loss": 0.451,
+      "step": 442
+    },
+    {
+      "epoch": 0.1276657060518732,
+      "grad_norm": 2.833020290341045,
+      "learning_rate": 1e-05,
+      "loss": 0.5725,
+      "step": 443
+    },
+    {
+      "epoch": 0.12795389048991354,
+      "grad_norm": 1.0051541036464473,
+      "learning_rate": 1e-05,
+      "loss": 0.5809,
+      "step": 444
+    },
+    {
+      "epoch": 0.1282420749279539,
+      "grad_norm": 1.8142770422576717,
+      "learning_rate": 1e-05,
+      "loss": 0.4529,
+      "step": 445
+    },
+    {
+      "epoch": 0.12853025936599424,
+      "grad_norm": 3.9811298650754323,
+      "learning_rate": 1e-05,
+      "loss": 0.5488,
+      "step": 446
+    },
+    {
+      "epoch": 0.1288184438040346,
+      "grad_norm": 1.218327361702739,
+      "learning_rate": 1e-05,
+      "loss": 0.5773,
+      "step": 447
+    },
+    {
+      "epoch": 0.12910662824207492,
+      "grad_norm": 2.201286370240042,
+      "learning_rate": 1e-05,
+      "loss": 0.4534,
+      "step": 448
+    },
+    {
+      "epoch": 0.12939481268011527,
+      "grad_norm": 1.7478083257483388,
+      "learning_rate": 1e-05,
+      "loss": 0.6036,
+      "step": 449
+    },
+    {
+      "epoch": 0.12968299711815562,
+      "grad_norm": 1.427855547613876,
+      "learning_rate": 1e-05,
+      "loss": 0.5867,
+      "step": 450
+    },
+    {
+      "epoch": 0.12997118155619597,
+      "grad_norm": 1.4621987146776787,
+      "learning_rate": 1e-05,
+      "loss": 0.4413,
+      "step": 451
+    },
+    {
+      "epoch": 0.13025936599423632,
+      "grad_norm": 1.4666772098953285,
+      "learning_rate": 1e-05,
+      "loss": 0.5748,
+      "step": 452
+    },
+    {
+      "epoch": 0.13054755043227664,
+      "grad_norm": 1.213595026215865,
+      "learning_rate": 1e-05,
+      "loss": 0.5982,
+      "step": 453
+    },
+    {
+      "epoch": 0.130835734870317,
+      "grad_norm": 2.355536720115263,
+      "learning_rate": 1e-05,
+      "loss": 0.445,
+      "step": 454
+    },
+    {
+      "epoch": 0.13112391930835735,
+      "grad_norm": 2.920619249973288,
+      "learning_rate": 1e-05,
+      "loss": 0.5817,
+      "step": 455
+    },
+    {
+      "epoch": 0.1314121037463977,
+      "grad_norm": 1.3428697439989274,
+      "learning_rate": 1e-05,
+      "loss": 0.5694,
+      "step": 456
+    },
+    {
+      "epoch": 0.13170028818443805,
+      "grad_norm": 1.743289386727595,
+      "learning_rate": 1e-05,
+      "loss": 0.4312,
+      "step": 457
+    },
+    {
+      "epoch": 0.1319884726224784,
+      "grad_norm": 2.255421712493541,
+      "learning_rate": 1e-05,
+      "loss": 0.5958,
+      "step": 458
+    },
+    {
+      "epoch": 0.13227665706051872,
+      "grad_norm": 1.1463419678293008,
+      "learning_rate": 1e-05,
+      "loss": 0.5809,
+      "step": 459
+    },
+    {
+      "epoch": 0.13256484149855907,
+      "grad_norm": 2.746233739943775,
+      "learning_rate": 1e-05,
+      "loss": 0.4354,
+      "step": 460
+    },
+    {
+      "epoch": 0.13285302593659942,
+      "grad_norm": 2.4487357894621224,
+      "learning_rate": 1e-05,
+      "loss": 0.5515,
+      "step": 461
+    },
+    {
+      "epoch": 0.13314121037463977,
+      "grad_norm": 1.735262829126143,
+      "learning_rate": 1e-05,
+      "loss": 0.5891,
+      "step": 462
+    },
+    {
+      "epoch": 0.13342939481268012,
+      "grad_norm": 3.6653451701990347,
+      "learning_rate": 1e-05,
+      "loss": 0.4155,
+      "step": 463
+    },
+    {
+      "epoch": 0.13371757925072045,
+      "grad_norm": 3.039932302637225,
+      "learning_rate": 1e-05,
+      "loss": 0.5491,
+      "step": 464
+    },
+    {
+      "epoch": 0.1340057636887608,
+      "grad_norm": 1.1770276022521715,
+      "learning_rate": 1e-05,
+      "loss": 0.581,
+      "step": 465
+    },
+    {
+      "epoch": 0.13429394812680115,
+      "grad_norm": 4.167641032839015,
+      "learning_rate": 1e-05,
+      "loss": 0.442,
+      "step": 466
+    },
+    {
+      "epoch": 0.1345821325648415,
+      "grad_norm": 1.6053065509968596,
+      "learning_rate": 1e-05,
+      "loss": 0.5822,
+      "step": 467
+    },
+    {
+      "epoch": 0.13487031700288185,
+      "grad_norm": 1.5032997619338024,
+      "learning_rate": 1e-05,
+      "loss": 0.5772,
+      "step": 468
+    },
+    {
+      "epoch": 0.1351585014409222,
+      "grad_norm": 3.270618757600374,
+      "learning_rate": 1e-05,
+      "loss": 0.4348,
+      "step": 469
+    },
+    {
+      "epoch": 0.13544668587896252,
+      "grad_norm": 2.791741501570059,
+      "learning_rate": 1e-05,
+      "loss": 0.5844,
+      "step": 470
+    },
+    {
+      "epoch": 0.13573487031700288,
+      "grad_norm": 1.1770255237309266,
+      "learning_rate": 1e-05,
+      "loss": 0.574,
+      "step": 471
+    },
+    {
+      "epoch": 0.13602305475504323,
+      "grad_norm": 4.736775435466847,
+      "learning_rate": 1e-05,
+      "loss": 0.4244,
+      "step": 472
+    },
+    {
+      "epoch": 0.13631123919308358,
+      "grad_norm": 1.7837329171849312,
+      "learning_rate": 1e-05,
+      "loss": 0.593,
+      "step": 473
+    },
+    {
+      "epoch": 0.13659942363112393,
+      "grad_norm": 1.149430666649388,
+      "learning_rate": 1e-05,
+      "loss": 0.5877,
+      "step": 474
+    },
+    {
+      "epoch": 0.13688760806916425,
+      "grad_norm": 2.318623131551547,
+      "learning_rate": 1e-05,
+      "loss": 0.4416,
+      "step": 475
+    },
+    {
+      "epoch": 0.1371757925072046,
+      "grad_norm": 2.8828159601083247,
+      "learning_rate": 1e-05,
+      "loss": 0.5403,
+      "step": 476
+    },
+    {
+      "epoch": 0.13746397694524495,
+      "grad_norm": 1.723279748983139,
+      "learning_rate": 1e-05,
+      "loss": 0.5806,
+      "step": 477
+    },
+    {
+      "epoch": 0.1377521613832853,
+      "grad_norm": 1.7523594088222687,
+      "learning_rate": 1e-05,
+      "loss": 0.4269,
+      "step": 478
+    },
+    {
+      "epoch": 0.13804034582132566,
+      "grad_norm": 4.094525500917702,
+      "learning_rate": 1e-05,
+      "loss": 0.5912,
+      "step": 479
+    },
+    {
+      "epoch": 0.138328530259366,
+      "grad_norm": 1.219868256258515,
+      "learning_rate": 1e-05,
+      "loss": 0.5927,
+      "step": 480
+    },
+    {
+      "epoch": 0.13861671469740633,
+      "grad_norm": 3.8503024267859858,
+      "learning_rate": 1e-05,
+      "loss": 0.4374,
+      "step": 481
+    },
+    {
+      "epoch": 0.13890489913544668,
+      "grad_norm": 3.7215917656004915,
+      "learning_rate": 1e-05,
+      "loss": 0.57,
+      "step": 482
+    },
+    {
+      "epoch": 0.13919308357348703,
+      "grad_norm": 1.326839803262488,
+      "learning_rate": 1e-05,
+      "loss": 0.5719,
+      "step": 483
+    },
+    {
+      "epoch": 0.13948126801152738,
+      "grad_norm": 3.1101448521293342,
+      "learning_rate": 1e-05,
+      "loss": 0.4192,
+      "step": 484
+    },
+    {
+      "epoch": 0.13976945244956773,
+      "grad_norm": 1.5713028450262407,
+      "learning_rate": 1e-05,
+      "loss": 0.5521,
+      "step": 485
+    },
+    {
+      "epoch": 0.14005763688760806,
+      "grad_norm": 1.4812387301104615,
+      "learning_rate": 1e-05,
+      "loss": 0.5961,
+      "step": 486
+    },
+    {
+      "epoch": 0.1403458213256484,
+      "grad_norm": 3.112719692709699,
+      "learning_rate": 1e-05,
+      "loss": 0.425,
+      "step": 487
+    },
+    {
+      "epoch": 0.14063400576368876,
+      "grad_norm": 3.117300102115347,
+      "learning_rate": 1e-05,
+      "loss": 0.5913,
+      "step": 488
+    },
+    {
+      "epoch": 0.1409221902017291,
+      "grad_norm": 1.281642593906318,
+      "learning_rate": 1e-05,
+      "loss": 0.568,
+      "step": 489
+    },
+    {
+      "epoch": 0.14121037463976946,
+      "grad_norm": 3.130491230859037,
+      "learning_rate": 1e-05,
+      "loss": 0.4269,
+      "step": 490
+    },
+    {
+      "epoch": 0.1414985590778098,
+      "grad_norm": 2.4765855622656496,
+      "learning_rate": 1e-05,
+      "loss": 0.5728,
+      "step": 491
+    },
+    {
+      "epoch": 0.14178674351585013,
+      "grad_norm": 1.3995884984100582,
+      "learning_rate": 1e-05,
+      "loss": 0.5658,
+      "step": 492
+    },
+    {
+      "epoch": 0.14207492795389048,
+      "grad_norm": 1.739429378644102,
+      "learning_rate": 1e-05,
+      "loss": 0.4273,
+      "step": 493
+    },
+    {
+      "epoch": 0.14236311239193083,
+      "grad_norm": 2.0295421203874318,
+      "learning_rate": 1e-05,
+      "loss": 0.5422,
+      "step": 494
+    },
+    {
+      "epoch": 0.14265129682997119,
+      "grad_norm": 1.3170660825967557,
+      "learning_rate": 1e-05,
+      "loss": 0.5949,
+      "step": 495
+    },
+    {
+      "epoch": 0.14293948126801154,
+      "grad_norm": 2.6091612579691197,
+      "learning_rate": 1e-05,
+      "loss": 0.4109,
+      "step": 496
+    },
+    {
+      "epoch": 0.14322766570605186,
+      "grad_norm": 3.6534911957875242,
+      "learning_rate": 1e-05,
+      "loss": 0.5765,
+      "step": 497
+    },
+    {
+      "epoch": 0.1435158501440922,
+      "grad_norm": 1.2762671692669125,
+      "learning_rate": 1e-05,
+      "loss": 0.5757,
+      "step": 498
+    },
+    {
+      "epoch": 0.14380403458213256,
+      "grad_norm": 2.0657668399019644,
+      "learning_rate": 1e-05,
+      "loss": 0.4346,
+      "step": 499
+    },
+    {
+      "epoch": 0.1440922190201729,
+      "grad_norm": 1.914852314151166,
+      "learning_rate": 1e-05,
+      "loss": 0.57,
+      "step": 500
+    },
+    {
+      "epoch": 0.1440922190201729,
+      "eval_websight_new_IoU": 0.01463321316987276,
+      "eval_websight_new_MAE_x": 286.7799987792969,
+      "eval_websight_new_MAE_y": 421.5400085449219,
+      "eval_websight_new_coord_probability": 0.73215651512146,
+      "eval_websight_new_correct_token_probability": 0.0021086945198476315,
+      "eval_websight_new_inside_bbox": 0.019999999552965164,
+      "eval_websight_new_loss": 0.4361584484577179,
+      "eval_websight_new_range_100_probability": 0.0867907926440239,
+      "eval_websight_new_range_20_probability": 0.017009830102324486,
+      "eval_websight_new_range_5_probability": 0.0048213982954621315,
+      "eval_websight_new_runtime": 34.1634,
+      "eval_websight_new_samples_per_second": 1.464,
+      "eval_websight_new_steps_per_second": 0.205,
+      "step": 500
+    },
+    {
+      "epoch": 0.1440922190201729,
+      "eval_seeclick_IoU": 0.03221755847334862,
+      "eval_seeclick_MAE_x": 473.1000061035156,
+      "eval_seeclick_MAE_y": 601.5999755859375,
+      "eval_seeclick_coord_probability": 0.7388027310371399,
+      "eval_seeclick_correct_token_probability": 0.002116676652804017,
+      "eval_seeclick_inside_bbox": 0.05999999865889549,
+      "eval_seeclick_loss": 0.4738258123397827,
+      "eval_seeclick_range_100_probability": 0.1094726100564003,
+      "eval_seeclick_range_20_probability": 0.029427895322442055,
+      "eval_seeclick_range_5_probability": 0.005799091421067715,
+      "eval_seeclick_runtime": 42.7801,
+      "eval_seeclick_samples_per_second": 1.169,
+      "eval_seeclick_steps_per_second": 0.164,
+      "step": 500
+    },
+    {
+      "epoch": 0.1440922190201729,
+      "eval_icons_IoU": 0.0019313165685161948,
+      "eval_icons_MAE_x": 473.5400085449219,
+      "eval_icons_MAE_y": 448.8800048828125,
+      "eval_icons_coord_probability": 0.7443975806236267,
+      "eval_icons_correct_token_probability": 0.0036313915625214577,
+      "eval_icons_inside_bbox": 0.0,
+      "eval_icons_loss": 0.5505882501602173,
+      "eval_icons_range_100_probability": 0.2653256356716156,
+      "eval_icons_range_20_probability": 0.09607495367527008,
+      "eval_icons_range_5_probability": 0.02670016698539257,
+      "eval_icons_runtime": 43.0247,
+      "eval_icons_samples_per_second": 1.162,
+      "eval_icons_steps_per_second": 0.163,
+      "step": 500
+    },
+    {
+      "epoch": 0.14438040345821326,
+      "grad_norm": 1.8448071513284736,
+      "learning_rate": 1e-05,
+      "loss": 0.579,
+      "step": 501
+    },
+    {
+      "epoch": 0.14466858789625361,
+      "grad_norm": 3.1326321130794916,
+      "learning_rate": 1e-05,
+      "loss": 0.4292,
+      "step": 502
+    },
+    {
+      "epoch": 0.14495677233429394,
+      "grad_norm": 1.9410665755091052,
+      "learning_rate": 1e-05,
+      "loss": 0.5894,
+      "step": 503
+    },
+    {
+      "epoch": 0.1452449567723343,
+      "grad_norm": 1.8431514766601858,
+      "learning_rate": 1e-05,
+      "loss": 0.5626,
+      "step": 504
+    },
+    {
+      "epoch": 0.14553314121037464,
+      "grad_norm": 3.548647616359119,
+      "learning_rate": 1e-05,
+      "loss": 0.4507,
+      "step": 505
+    },
+    {
+      "epoch": 0.145821325648415,
+      "grad_norm": 2.5418153973369972,
+      "learning_rate": 1e-05,
+      "loss": 0.562,
+      "step": 506
+    },
+    {
+      "epoch": 0.14610951008645534,
+      "grad_norm": 1.3964779815637525,
+      "learning_rate": 1e-05,
+      "loss": 0.5818,
+      "step": 507
+    },
+    {
+      "epoch": 0.14639769452449566,
+      "grad_norm": 2.5166763904911633,
+      "learning_rate": 1e-05,
+      "loss": 0.4376,
+      "step": 508
+    },
+    {
+      "epoch": 0.14668587896253601,
+      "grad_norm": 1.6927625167741949,
+      "learning_rate": 1e-05,
+      "loss": 0.5474,
+      "step": 509
+    },
+    {
+      "epoch": 0.14697406340057637,
+      "grad_norm": 1.7542553767621742,
+      "learning_rate": 1e-05,
+      "loss": 0.5643,
+      "step": 510
+    },
+    {
+      "epoch": 0.14726224783861672,
+      "grad_norm": 2.9349303403137244,
+      "learning_rate": 1e-05,
+      "loss": 0.4413,
+      "step": 511
+    },
+    {
+      "epoch": 0.14755043227665707,
+      "grad_norm": 1.9760710131732777,
+      "learning_rate": 1e-05,
+      "loss": 0.5451,
+      "step": 512
+    },
+    {
+      "epoch": 0.14783861671469742,
+      "grad_norm": 1.449825928543996,
+      "learning_rate": 1e-05,
+      "loss": 0.5691,
+      "step": 513
+    },
+    {
+      "epoch": 0.14812680115273774,
+      "grad_norm": 4.073458689095103,
+      "learning_rate": 1e-05,
+      "loss": 0.4131,
+      "step": 514
+    },
+    {
+      "epoch": 0.1484149855907781,
+      "grad_norm": 2.801815070531123,
+      "learning_rate": 1e-05,
+      "loss": 0.571,
+      "step": 515
+    },
+    {
+      "epoch": 0.14870317002881844,
+      "grad_norm": 2.126190537412442,
+      "learning_rate": 1e-05,
+      "loss": 0.5916,
+      "step": 516
+    },
+    {
+      "epoch": 0.1489913544668588,
+      "grad_norm": 1.4518893561832873,
+      "learning_rate": 1e-05,
+      "loss": 0.4218,
+      "step": 517
+    },
+    {
+      "epoch": 0.14927953890489915,
+      "grad_norm": 3.180599704926899,
+      "learning_rate": 1e-05,
+      "loss": 0.5596,
+      "step": 518
+    },
+    {
+      "epoch": 0.14956772334293947,
+      "grad_norm": 1.4565202457275603,
+      "learning_rate": 1e-05,
+      "loss": 0.5746,
+      "step": 519
+    },
+    {
+      "epoch": 0.14985590778097982,
+      "grad_norm": 2.1482842856757154,
+      "learning_rate": 1e-05,
+      "loss": 0.4091,
+      "step": 520
+    },
+    {
+      "epoch": 0.15014409221902017,
+      "grad_norm": 2.2267048860256082,
+      "learning_rate": 1e-05,
+      "loss": 0.5719,
+      "step": 521
+    },
+    {
+      "epoch": 0.15043227665706052,
+      "grad_norm": 1.1936041027002116,
+      "learning_rate": 1e-05,
+      "loss": 0.5787,
+      "step": 522
+    },
+    {
+      "epoch": 0.15072046109510087,
+      "grad_norm": 4.109477616663216,
+      "learning_rate": 1e-05,
+      "loss": 0.4387,
+      "step": 523
+    },
+    {
+      "epoch": 0.15100864553314122,
+      "grad_norm": 3.145535845039723,
+      "learning_rate": 1e-05,
+      "loss": 0.5522,
+      "step": 524
+    },
+    {
+      "epoch": 0.15129682997118155,
+      "grad_norm": 1.174791602931962,
+      "learning_rate": 1e-05,
+      "loss": 0.5596,
+      "step": 525
+    },
+    {
+      "epoch": 0.1515850144092219,
+      "grad_norm": 3.2884466993425816,
+      "learning_rate": 1e-05,
+      "loss": 0.4216,
+      "step": 526
+    },
+    {
+      "epoch": 0.15187319884726225,
+      "grad_norm": 2.1818494760236895,
+      "learning_rate": 1e-05,
+      "loss": 0.5388,
+      "step": 527
+    },
+    {
+      "epoch": 0.1521613832853026,
+      "grad_norm": 1.2537362607320528,
+      "learning_rate": 1e-05,
+      "loss": 0.583,
+      "step": 528
+    },
+    {
+      "epoch": 0.15244956772334295,
+      "grad_norm": 1.9383363422951394,
+      "learning_rate": 1e-05,
+      "loss": 0.4226,
+      "step": 529
+    },
+    {
+      "epoch": 0.15273775216138327,
+      "grad_norm": 3.4770943698334813,
+      "learning_rate": 1e-05,
+      "loss": 0.5313,
+      "step": 530
+    },
+    {
+      "epoch": 0.15302593659942362,
+      "grad_norm": 1.6673674360106558,
+      "learning_rate": 1e-05,
+      "loss": 0.5822,
+      "step": 531
+    },
+    {
+      "epoch": 0.15331412103746397,
+      "grad_norm": 1.9256390602904025,
+      "learning_rate": 1e-05,
+      "loss": 0.41,
+      "step": 532
+    },
+    {
+      "epoch": 0.15360230547550432,
+      "grad_norm": 3.4372138253608133,
+      "learning_rate": 1e-05,
+      "loss": 0.5564,
+      "step": 533
+    },
+    {
+      "epoch": 0.15389048991354468,
+      "grad_norm": 1.139878259611146,
+      "learning_rate": 1e-05,
+      "loss": 0.5731,
+      "step": 534
+    },
+    {
+      "epoch": 0.15417867435158503,
+      "grad_norm": 2.5258128668121658,
+      "learning_rate": 1e-05,
+      "loss": 0.4287,
+      "step": 535
+    },
+    {
+      "epoch": 0.15446685878962535,
+      "grad_norm": 2.264631946485219,
+      "learning_rate": 1e-05,
+      "loss": 0.5801,
+      "step": 536
+    },
+    {
+      "epoch": 0.1547550432276657,
+      "grad_norm": 1.15717597829775,
+      "learning_rate": 1e-05,
+      "loss": 0.5764,
+      "step": 537
+    },
+    {
+      "epoch": 0.15504322766570605,
+      "grad_norm": 2.19598195468663,
+      "learning_rate": 1e-05,
+      "loss": 0.4152,
+      "step": 538
+    },
+    {
+      "epoch": 0.1553314121037464,
+      "grad_norm": 1.5974769077267827,
+      "learning_rate": 1e-05,
+      "loss": 0.5567,
+      "step": 539
+    },
+    {
+      "epoch": 0.15561959654178675,
+      "grad_norm": 1.6439935505821885,
+      "learning_rate": 1e-05,
+      "loss": 0.5763,
+      "step": 540
+    },
+    {
+      "epoch": 0.15590778097982708,
+      "grad_norm": 1.8035286132055963,
+      "learning_rate": 1e-05,
+      "loss": 0.4244,
+      "step": 541
+    },
+    {
+      "epoch": 0.15619596541786743,
+      "grad_norm": 3.3289981506089608,
+      "learning_rate": 1e-05,
+      "loss": 0.5306,
+      "step": 542
+    },
+    {
+      "epoch": 0.15648414985590778,
+      "grad_norm": 1.3575419479537025,
+      "learning_rate": 1e-05,
+      "loss": 0.5522,
+      "step": 543
+    },
+    {
+      "epoch": 0.15677233429394813,
+      "grad_norm": 4.060042360386747,
+      "learning_rate": 1e-05,
+      "loss": 0.4354,
+      "step": 544
+    },
+    {
+      "epoch": 0.15706051873198848,
+      "grad_norm": 2.719748279737449,
+      "learning_rate": 1e-05,
+      "loss": 0.5546,
+      "step": 545
+    },
+    {
+      "epoch": 0.15734870317002883,
+      "grad_norm": 1.1607951338878562,
+      "learning_rate": 1e-05,
+      "loss": 0.5659,
+      "step": 546
+    },
+    {
+      "epoch": 0.15763688760806915,
+      "grad_norm": 2.2262089217725882,
+      "learning_rate": 1e-05,
+      "loss": 0.4216,
+      "step": 547
+    },
+    {
+      "epoch": 0.1579250720461095,
+      "grad_norm": 2.152306989000815,
+      "learning_rate": 1e-05,
+      "loss": 0.5796,
+      "step": 548
+    },
+    {
+      "epoch": 0.15821325648414986,
+      "grad_norm": 1.9929809674573478,
+      "learning_rate": 1e-05,
+      "loss": 0.5694,
+      "step": 549
+    },
+    {
+      "epoch": 0.1585014409221902,
+      "grad_norm": 1.6817494253176306,
+      "learning_rate": 1e-05,
+      "loss": 0.4244,
+      "step": 550
+    },
+    {
+      "epoch": 0.15878962536023056,
+      "grad_norm": 3.3303932325186083,
+      "learning_rate": 1e-05,
+      "loss": 0.5609,
+      "step": 551
+    },
+    {
+      "epoch": 0.15907780979827088,
+      "grad_norm": 1.4879625617138847,
+      "learning_rate": 1e-05,
+      "loss": 0.551,
+      "step": 552
+    },
+    {
+      "epoch": 0.15936599423631123,
+      "grad_norm": 1.7347403864057702,
+      "learning_rate": 1e-05,
+      "loss": 0.4,
+      "step": 553
+    },
+    {
+      "epoch": 0.15965417867435158,
+      "grad_norm": 1.7652909618145627,
+      "learning_rate": 1e-05,
+      "loss": 0.5797,
+      "step": 554
+    },
+    {
+      "epoch": 0.15994236311239193,
+      "grad_norm": 1.7303096867055798,
+      "learning_rate": 1e-05,
+      "loss": 0.5722,
+      "step": 555
+    },
+    {
+      "epoch": 0.16023054755043228,
+      "grad_norm": 1.7126273955726241,
+      "learning_rate": 1e-05,
+      "loss": 0.3985,
+      "step": 556
+    },
+    {
+      "epoch": 0.16051873198847263,
+      "grad_norm": 1.4902869117595645,
+      "learning_rate": 1e-05,
+      "loss": 0.5386,
+      "step": 557
+    },
+    {
+      "epoch": 0.16080691642651296,
+      "grad_norm": 1.2679155140149418,
+      "learning_rate": 1e-05,
+      "loss": 0.5573,
+      "step": 558
+    },
+    {
+      "epoch": 0.1610951008645533,
+      "grad_norm": 4.209937191606016,
+      "learning_rate": 1e-05,
+      "loss": 0.386,
+      "step": 559
+    },
+    {
+      "epoch": 0.16138328530259366,
+      "grad_norm": 2.1327055619520903,
+      "learning_rate": 1e-05,
+      "loss": 0.5512,
+      "step": 560
+    },
+    {
+      "epoch": 0.161671469740634,
+      "grad_norm": 1.8745967111214592,
+      "learning_rate": 1e-05,
+      "loss": 0.5706,
+      "step": 561
+    },
+    {
+      "epoch": 0.16195965417867436,
+      "grad_norm": 3.004705979570041,
+      "learning_rate": 1e-05,
+      "loss": 0.3967,
+      "step": 562
+    },
+    {
+      "epoch": 0.16224783861671468,
+      "grad_norm": 3.705567418644865,
+      "learning_rate": 1e-05,
+      "loss": 0.5565,
+      "step": 563
+    },
+    {
+      "epoch": 0.16253602305475504,
+      "grad_norm": 1.879018023645976,
+      "learning_rate": 1e-05,
+      "loss": 0.5613,
+      "step": 564
+    },
+    {
+      "epoch": 0.1628242074927954,
+      "grad_norm": 2.2060710644133685,
+      "learning_rate": 1e-05,
+      "loss": 0.3915,
+      "step": 565
+    },
+    {
+      "epoch": 0.16311239193083574,
+      "grad_norm": 2.07947260611661,
+      "learning_rate": 1e-05,
+      "loss": 0.5639,
+      "step": 566
+    },
+    {
+      "epoch": 0.1634005763688761,
+      "grad_norm": 1.6378908418091689,
+      "learning_rate": 1e-05,
+      "loss": 0.5699,
+      "step": 567
+    },
+    {
+      "epoch": 0.16368876080691644,
+      "grad_norm": 3.000890981624951,
+      "learning_rate": 1e-05,
+      "loss": 0.4058,
+      "step": 568
+    },
+    {
+      "epoch": 0.16397694524495676,
+      "grad_norm": 2.126292063396071,
+      "learning_rate": 1e-05,
+      "loss": 0.5236,
+      "step": 569
+    },
+    {
+      "epoch": 0.1642651296829971,
+      "grad_norm": 1.5309333888574757,
+      "learning_rate": 1e-05,
+      "loss": 0.5497,
+      "step": 570
+    },
+    {
+      "epoch": 0.16455331412103746,
+      "grad_norm": 4.411481950034936,
+      "learning_rate": 1e-05,
+      "loss": 0.4087,
+      "step": 571
+    },
+    {
+      "epoch": 0.16484149855907781,
+      "grad_norm": 2.537810253449623,
+      "learning_rate": 1e-05,
+      "loss": 0.502,
+      "step": 572
+    },
+    {
+      "epoch": 0.16512968299711817,
+      "grad_norm": 1.3858443640190015,
+      "learning_rate": 1e-05,
+      "loss": 0.5688,
+      "step": 573
+    },
+    {
+      "epoch": 0.1654178674351585,
+      "grad_norm": 3.1550677968939924,
+      "learning_rate": 1e-05,
+      "loss": 0.4229,
+      "step": 574
+    },
+    {
+      "epoch": 0.16570605187319884,
+      "grad_norm": 1.833600330517365,
+      "learning_rate": 1e-05,
+      "loss": 0.539,
+      "step": 575
+    },
+    {
+      "epoch": 0.1659942363112392,
+      "grad_norm": 1.8400406919145424,
+      "learning_rate": 1e-05,
+      "loss": 0.5519,
+      "step": 576
+    },
+    {
+      "epoch": 0.16628242074927954,
+      "grad_norm": 2.606139497529331,
+      "learning_rate": 1e-05,
+      "loss": 0.4,
+      "step": 577
+    },
+    {
+      "epoch": 0.1665706051873199,
+      "grad_norm": 2.0181636162880308,
+      "learning_rate": 1e-05,
+      "loss": 0.5852,
+      "step": 578
+    },
+    {
+      "epoch": 0.16685878962536024,
+      "grad_norm": 1.6839491696790014,
+      "learning_rate": 1e-05,
+      "loss": 0.5687,
+      "step": 579
+    },
+    {
+      "epoch": 0.16714697406340057,
+      "grad_norm": 3.854172359890152,
+      "learning_rate": 1e-05,
+      "loss": 0.4043,
+      "step": 580
+    },
+    {
+      "epoch": 0.16743515850144092,
+      "grad_norm": 3.0960591563701434,
+      "learning_rate": 1e-05,
+      "loss": 0.5288,
+      "step": 581
+    },
+    {
+      "epoch": 0.16772334293948127,
+      "grad_norm": 2.1395720918645265,
+      "learning_rate": 1e-05,
+      "loss": 0.5714,
+      "step": 582
+    },
+    {
+      "epoch": 0.16801152737752162,
+      "grad_norm": 3.4674051279756855,
+      "learning_rate": 1e-05,
+      "loss": 0.4016,
+      "step": 583
+    },
+    {
+      "epoch": 0.16829971181556197,
+      "grad_norm": 3.0515951310737193,
+      "learning_rate": 1e-05,
+      "loss": 0.5489,
+      "step": 584
+    },
+    {
+      "epoch": 0.1685878962536023,
+      "grad_norm": 1.3654445978000875,
+      "learning_rate": 1e-05,
+      "loss": 0.5505,
+      "step": 585
+    },
+    {
+      "epoch": 0.16887608069164264,
+      "grad_norm": 2.571044000683727,
+      "learning_rate": 1e-05,
+      "loss": 0.4286,
+      "step": 586
+    },
+    {
+      "epoch": 0.169164265129683,
+      "grad_norm": 2.262059543392501,
+      "learning_rate": 1e-05,
+      "loss": 0.5691,
+      "step": 587
+    },
+    {
+      "epoch": 0.16945244956772335,
+      "grad_norm": 2.3723813138832313,
+      "learning_rate": 1e-05,
+      "loss": 0.5505,
+      "step": 588
+    },
+    {
+      "epoch": 0.1697406340057637,
+      "grad_norm": 2.2783042616446365,
+      "learning_rate": 1e-05,
+      "loss": 0.4146,
+      "step": 589
+    },
+    {
+      "epoch": 0.17002881844380405,
+      "grad_norm": 3.3019889343278206,
+      "learning_rate": 1e-05,
+      "loss": 0.5751,
+      "step": 590
+    },
+    {
+      "epoch": 0.17031700288184437,
+      "grad_norm": 1.4577549924848103,
+      "learning_rate": 1e-05,
+      "loss": 0.5515,
+      "step": 591
+    },
+    {
+      "epoch": 0.17060518731988472,
+      "grad_norm": 4.535404100747485,
+      "learning_rate": 1e-05,
+      "loss": 0.3968,
+      "step": 592
+    },
+    {
+      "epoch": 0.17089337175792507,
+      "grad_norm": 2.007603274675207,
+      "learning_rate": 1e-05,
+      "loss": 0.5427,
+      "step": 593
+    },
+    {
+      "epoch": 0.17118155619596542,
+      "grad_norm": 1.38832578741342,
+      "learning_rate": 1e-05,
+      "loss": 0.558,
+      "step": 594
+    },
+    {
+      "epoch": 0.17146974063400577,
+      "grad_norm": 2.5680251846041497,
+      "learning_rate": 1e-05,
+      "loss": 0.3968,
+      "step": 595
+    },
+    {
+      "epoch": 0.1717579250720461,
+      "grad_norm": 3.6595150087684094,
+      "learning_rate": 1e-05,
+      "loss": 0.5459,
+      "step": 596
+    },
+    {
+      "epoch": 0.17204610951008645,
+      "grad_norm": 2.2315423996352934,
+      "learning_rate": 1e-05,
+      "loss": 0.5505,
+      "step": 597
+    },
+    {
+      "epoch": 0.1723342939481268,
+      "grad_norm": 2.401220339770531,
+      "learning_rate": 1e-05,
+      "loss": 0.3946,
+      "step": 598
+    },
+    {
+      "epoch": 0.17262247838616715,
+      "grad_norm": 2.1875816223005824,
+      "learning_rate": 1e-05,
+      "loss": 0.5647,
+      "step": 599
+    },
+    {
+      "epoch": 0.1729106628242075,
+      "grad_norm": 1.973156691171414,
+      "learning_rate": 1e-05,
+      "loss": 0.5456,
+      "step": 600
+    },
+    {
+      "epoch": 0.17319884726224785,
+      "grad_norm": 3.448549394643256,
+      "learning_rate": 1e-05,
+      "loss": 0.396,
+      "step": 601
+    },
+    {
+      "epoch": 0.17348703170028817,
+      "grad_norm": 2.295210141260302,
+      "learning_rate": 1e-05,
+      "loss": 0.5485,
+      "step": 602
+    },
+    {
+      "epoch": 0.17377521613832853,
+      "grad_norm": 1.9451387743663426,
+      "learning_rate": 1e-05,
+      "loss": 0.5646,
+      "step": 603
+    },
+    {
+      "epoch": 0.17406340057636888,
+      "grad_norm": 3.865943774754975,
+      "learning_rate": 1e-05,
+      "loss": 0.3931,
+      "step": 604
+    },
+    {
+      "epoch": 0.17435158501440923,
+      "grad_norm": 2.878396544005281,
+      "learning_rate": 1e-05,
+      "loss": 0.5514,
+      "step": 605
+    },
+    {
+      "epoch": 0.17463976945244958,
+      "grad_norm": 1.4862887491338082,
+      "learning_rate": 1e-05,
+      "loss": 0.5606,
+      "step": 606
+    },
+    {
+      "epoch": 0.1749279538904899,
+      "grad_norm": 1.886792204755533,
+      "learning_rate": 1e-05,
+      "loss": 0.3835,
+      "step": 607
+    },
+    {
+      "epoch": 0.17521613832853025,
+      "grad_norm": 2.93011855036884,
+      "learning_rate": 1e-05,
+      "loss": 0.548,
+      "step": 608
+    },
+    {
+      "epoch": 0.1755043227665706,
+      "grad_norm": 2.327293328559529,
+      "learning_rate": 1e-05,
+      "loss": 0.5625,
+      "step": 609
+    },
+    {
+      "epoch": 0.17579250720461095,
+      "grad_norm": 3.8373218363052906,
+      "learning_rate": 1e-05,
+      "loss": 0.3928,
+      "step": 610
+    },
+    {
+      "epoch": 0.1760806916426513,
+      "grad_norm": 2.4655793860953956,
+      "learning_rate": 1e-05,
+      "loss": 0.5778,
+      "step": 611
+    },
+    {
+      "epoch": 0.17636887608069166,
+      "grad_norm": 1.849299129303253,
+      "learning_rate": 1e-05,
+      "loss": 0.5402,
+      "step": 612
+    },
+    {
+      "epoch": 0.17665706051873198,
+      "grad_norm": 5.654202874511459,
+      "learning_rate": 1e-05,
+      "loss": 0.3951,
+      "step": 613
+    },
+    {
+      "epoch": 0.17694524495677233,
+      "grad_norm": 4.320339997345097,
+      "learning_rate": 1e-05,
+      "loss": 0.5332,
+      "step": 614
+    },
+    {
+      "epoch": 0.17723342939481268,
+      "grad_norm": 1.9081807363214465,
+      "learning_rate": 1e-05,
+      "loss": 0.5533,
+      "step": 615
+    },
+    {
+      "epoch": 0.17752161383285303,
+      "grad_norm": 1.6998027358164398,
+      "learning_rate": 1e-05,
+      "loss": 0.4038,
+      "step": 616
+    },
+    {
+      "epoch": 0.17780979827089338,
+      "grad_norm": 1.8637243057421535,
+      "learning_rate": 1e-05,
+      "loss": 0.5201,
+      "step": 617
+    },
+    {
+      "epoch": 0.1780979827089337,
+      "grad_norm": 2.3678771540872177,
+      "learning_rate": 1e-05,
+      "loss": 0.5529,
+      "step": 618
+    },
+    {
+      "epoch": 0.17838616714697406,
+      "grad_norm": 5.57593000831114,
+      "learning_rate": 1e-05,
+      "loss": 0.4073,
+      "step": 619
+    },
+    {
+      "epoch": 0.1786743515850144,
+      "grad_norm": 4.218960810252173,
+      "learning_rate": 1e-05,
+      "loss": 0.5473,
+      "step": 620
+    },
+    {
+      "epoch": 0.17896253602305476,
+      "grad_norm": 1.6011666265746873,
+      "learning_rate": 1e-05,
+      "loss": 0.5605,
+      "step": 621
+    },
+    {
+      "epoch": 0.1792507204610951,
+      "grad_norm": 2.7083163727318076,
+      "learning_rate": 1e-05,
+      "loss": 0.3935,
+      "step": 622
+    },
+    {
+      "epoch": 0.17953890489913546,
+      "grad_norm": 1.6812217120008799,
+      "learning_rate": 1e-05,
+      "loss": 0.5633,
+      "step": 623
+    },
+    {
+      "epoch": 0.17982708933717578,
+      "grad_norm": 1.6809096482590729,
+      "learning_rate": 1e-05,
+      "loss": 0.5554,
+      "step": 624
+    },
+    {
+      "epoch": 0.18011527377521613,
+      "grad_norm": 2.679832939952373,
+      "learning_rate": 1e-05,
+      "loss": 0.4127,
+      "step": 625
+    },
+    {
+      "epoch": 0.18040345821325648,
+      "grad_norm": 2.756781505942509,
+      "learning_rate": 1e-05,
+      "loss": 0.564,
+      "step": 626
+    },
+    {
+      "epoch": 0.18069164265129684,
+      "grad_norm": 1.6826096690716983,
+      "learning_rate": 1e-05,
+      "loss": 0.5281,
+      "step": 627
+    },
+    {
+      "epoch": 0.1809798270893372,
+      "grad_norm": 3.861133291619558,
+      "learning_rate": 1e-05,
+      "loss": 0.3834,
+      "step": 628
+    },
+    {
+      "epoch": 0.1812680115273775,
+      "grad_norm": 2.485939817416258,
+      "learning_rate": 1e-05,
+      "loss": 0.5667,
+      "step": 629
+    },
+    {
+      "epoch": 0.18155619596541786,
+      "grad_norm": 1.6639736644632237,
+      "learning_rate": 1e-05,
+      "loss": 0.5406,
+      "step": 630
+    },
+    {
+      "epoch": 0.1818443804034582,
+      "grad_norm": 2.519406384304628,
+      "learning_rate": 1e-05,
+      "loss": 0.4043,
+      "step": 631
+    },
+    {
+      "epoch": 0.18213256484149856,
+      "grad_norm": 4.059356699784603,
+      "learning_rate": 1e-05,
+      "loss": 0.528,
+      "step": 632
+    },
+    {
+      "epoch": 0.1824207492795389,
+      "grad_norm": 2.1055667595745526,
+      "learning_rate": 1e-05,
+      "loss": 0.5426,
+      "step": 633
+    },
+    {
+      "epoch": 0.18270893371757926,
+      "grad_norm": 4.327703295132949,
+      "learning_rate": 1e-05,
+      "loss": 0.3954,
+      "step": 634
+    },
+    {
+      "epoch": 0.1829971181556196,
+      "grad_norm": 1.8548168277556574,
+      "learning_rate": 1e-05,
+      "loss": 0.5559,
+      "step": 635
+    },
+    {
+      "epoch": 0.18328530259365994,
+      "grad_norm": 1.3956688231830576,
+      "learning_rate": 1e-05,
+      "loss": 0.5349,
+      "step": 636
+    },
+    {
+      "epoch": 0.1835734870317003,
+      "grad_norm": 2.9016659878557407,
+      "learning_rate": 1e-05,
+      "loss": 0.403,
+      "step": 637
+    },
+    {
+      "epoch": 0.18386167146974064,
+      "grad_norm": 3.005895623836115,
+      "learning_rate": 1e-05,
+      "loss": 0.5422,
+      "step": 638
+    },
+    {
+      "epoch": 0.184149855907781,
+      "grad_norm": 2.0525646021136716,
+      "learning_rate": 1e-05,
+      "loss": 0.5513,
+      "step": 639
+    },
+    {
+      "epoch": 0.1844380403458213,
+      "grad_norm": 1.6391771026267206,
+      "learning_rate": 1e-05,
+      "loss": 0.3948,
+      "step": 640
+    },
+    {
+      "epoch": 0.18472622478386166,
+      "grad_norm": 1.7490421062019743,
+      "learning_rate": 1e-05,
+      "loss": 0.5699,
+      "step": 641
+    },
+    {
+      "epoch": 0.18501440922190202,
+      "grad_norm": 1.5702471437958372,
+      "learning_rate": 1e-05,
+      "loss": 0.5545,
+      "step": 642
+    },
+    {
+      "epoch": 0.18530259365994237,
+      "grad_norm": 4.152208977372896,
+      "learning_rate": 1e-05,
+      "loss": 0.3948,
+      "step": 643
+    },
+    {
+      "epoch": 0.18559077809798272,
+      "grad_norm": 2.2787741438401916,
+      "learning_rate": 1e-05,
+      "loss": 0.5362,
+      "step": 644
+    },
+    {
+      "epoch": 0.18587896253602307,
+      "grad_norm": 1.1631130314710791,
+      "learning_rate": 1e-05,
+      "loss": 0.5307,
+      "step": 645
+    },
+    {
+      "epoch": 0.1861671469740634,
+      "grad_norm": 2.5882277274941075,
+      "learning_rate": 1e-05,
+      "loss": 0.377,
+      "step": 646
+    },
+    {
+      "epoch": 0.18645533141210374,
+      "grad_norm": 2.2812164584511123,
+      "learning_rate": 1e-05,
+      "loss": 0.5317,
+      "step": 647
+    },
+    {
+      "epoch": 0.1867435158501441,
+      "grad_norm": 1.672678780235995,
+      "learning_rate": 1e-05,
+      "loss": 0.5354,
+      "step": 648
+    },
+    {
+      "epoch": 0.18703170028818444,
+      "grad_norm": 1.818573544529323,
+      "learning_rate": 1e-05,
+      "loss": 0.4098,
+      "step": 649
+    },
+    {
+      "epoch": 0.1873198847262248,
+      "grad_norm": 2.2179155359228626,
+      "learning_rate": 1e-05,
+      "loss": 0.5467,
+      "step": 650
+    },
+    {
+      "epoch": 0.18760806916426512,
+      "grad_norm": 1.8033963460692448,
+      "learning_rate": 1e-05,
+      "loss": 0.5473,
+      "step": 651
+    },
+    {
+      "epoch": 0.18789625360230547,
+      "grad_norm": 3.473690365508948,
+      "learning_rate": 1e-05,
+      "loss": 0.3873,
+      "step": 652
+    },
+    {
+      "epoch": 0.18818443804034582,
+      "grad_norm": 2.389456829919027,
+      "learning_rate": 1e-05,
+      "loss": 0.529,
+      "step": 653
+    },
+    {
+      "epoch": 0.18847262247838617,
+      "grad_norm": 1.6131730909394986,
+      "learning_rate": 1e-05,
+      "loss": 0.5501,
+      "step": 654
+    },
+    {
+      "epoch": 0.18876080691642652,
+      "grad_norm": 1.8623033968265517,
+      "learning_rate": 1e-05,
+      "loss": 0.374,
+      "step": 655
+    },
+    {
+      "epoch": 0.18904899135446687,
+      "grad_norm": 2.3707633059771918,
+      "learning_rate": 1e-05,
+      "loss": 0.5654,
+      "step": 656
+    },
+    {
+      "epoch": 0.1893371757925072,
+      "grad_norm": 1.68411902183965,
+      "learning_rate": 1e-05,
+      "loss": 0.546,
+      "step": 657
+    },
+    {
+      "epoch": 0.18962536023054755,
+      "grad_norm": 6.49456396269767,
+      "learning_rate": 1e-05,
+      "loss": 0.3892,
+      "step": 658
+    },
+    {
+      "epoch": 0.1899135446685879,
+      "grad_norm": 3.5905436001612623,
+      "learning_rate": 1e-05,
+      "loss": 0.5414,
+      "step": 659
+    },
+    {
+      "epoch": 0.19020172910662825,
+      "grad_norm": 1.5502358348297345,
+      "learning_rate": 1e-05,
+      "loss": 0.5596,
+      "step": 660
+    },
+    {
+      "epoch": 0.1904899135446686,
+      "grad_norm": 3.0716995823954765,
+      "learning_rate": 1e-05,
+      "loss": 0.3834,
+      "step": 661
+    },
+    {
+      "epoch": 0.19077809798270892,
+      "grad_norm": 2.932377541022994,
+      "learning_rate": 1e-05,
+      "loss": 0.5263,
+      "step": 662
+    },
+    {
+      "epoch": 0.19106628242074927,
+      "grad_norm": 2.019552377168776,
+      "learning_rate": 1e-05,
+      "loss": 0.538,
+      "step": 663
+    },
+    {
+      "epoch": 0.19135446685878962,
+      "grad_norm": 1.8477246530970768,
+      "learning_rate": 1e-05,
+      "loss": 0.4037,
+      "step": 664
+    },
+    {
+      "epoch": 0.19164265129682997,
+      "grad_norm": 2.4470624112951285,
+      "learning_rate": 1e-05,
+      "loss": 0.5781,
+      "step": 665
+    },
+    {
+      "epoch": 0.19193083573487033,
+      "grad_norm": 1.410434793166915,
+      "learning_rate": 1e-05,
+      "loss": 0.5366,
+      "step": 666
+    },
+    {
+      "epoch": 0.19221902017291068,
+      "grad_norm": 3.4079539947667468,
+      "learning_rate": 1e-05,
+      "loss": 0.3929,
+      "step": 667
+    },
+    {
+      "epoch": 0.192507204610951,
+      "grad_norm": 3.2474549702210043,
+      "learning_rate": 1e-05,
+      "loss": 0.5414,
+      "step": 668
+    },
+    {
+      "epoch": 0.19279538904899135,
+      "grad_norm": 2.062451109518947,
+      "learning_rate": 1e-05,
+      "loss": 0.5455,
+      "step": 669
+    },
+    {
+      "epoch": 0.1930835734870317,
+      "grad_norm": 3.3242299042170345,
+      "learning_rate": 1e-05,
+      "loss": 0.3868,
+      "step": 670
+    },
+    {
+      "epoch": 0.19337175792507205,
+      "grad_norm": 2.7024222046537263,
+      "learning_rate": 1e-05,
+      "loss": 0.5606,
+      "step": 671
+    },
+    {
+      "epoch": 0.1936599423631124,
+      "grad_norm": 1.7649507400502558,
+      "learning_rate": 1e-05,
+      "loss": 0.5599,
+      "step": 672
+    },
+    {
+      "epoch": 0.19394812680115273,
+      "grad_norm": 2.783929652797413,
+      "learning_rate": 1e-05,
+      "loss": 0.3847,
+      "step": 673
+    },
+    {
+      "epoch": 0.19423631123919308,
+      "grad_norm": 3.241842986522019,
+      "learning_rate": 1e-05,
+      "loss": 0.5442,
+      "step": 674
+    },
+    {
+      "epoch": 0.19452449567723343,
+      "grad_norm": 1.3303643604091253,
+      "learning_rate": 1e-05,
+      "loss": 0.5389,
+      "step": 675
+    },
+    {
+      "epoch": 0.19481268011527378,
+      "grad_norm": 1.195358403648017,
+      "learning_rate": 1e-05,
+      "loss": 0.3667,
+      "step": 676
+    },
+    {
+      "epoch": 0.19510086455331413,
+      "grad_norm": 3.1427074417202316,
+      "learning_rate": 1e-05,
+      "loss": 0.5527,
+      "step": 677
+    },
+    {
+      "epoch": 0.19538904899135448,
+      "grad_norm": 1.4903596488788504,
+      "learning_rate": 1e-05,
+      "loss": 0.5401,
+      "step": 678
+    },
+    {
+      "epoch": 0.1956772334293948,
+      "grad_norm": 1.893268281849623,
+      "learning_rate": 1e-05,
+      "loss": 0.3845,
+      "step": 679
+    },
+    {
+      "epoch": 0.19596541786743515,
+      "grad_norm": 3.178108430580706,
+      "learning_rate": 1e-05,
+      "loss": 0.5808,
+      "step": 680
+    },
+    {
+      "epoch": 0.1962536023054755,
+      "grad_norm": 1.314861960044517,
+      "learning_rate": 1e-05,
+      "loss": 0.5211,
+      "step": 681
+    },
+    {
+      "epoch": 0.19654178674351586,
+      "grad_norm": 1.6646688411495776,
+      "learning_rate": 1e-05,
+      "loss": 0.3811,
+      "step": 682
+    },
+    {
+      "epoch": 0.1968299711815562,
+      "grad_norm": 1.731434277680317,
+      "learning_rate": 1e-05,
+      "loss": 0.547,
+      "step": 683
+    },
+    {
+      "epoch": 0.19711815561959653,
+      "grad_norm": 1.7992172856050568,
+      "learning_rate": 1e-05,
+      "loss": 0.5394,
+      "step": 684
+    },
+    {
+      "epoch": 0.19740634005763688,
+      "grad_norm": 2.3807111350441916,
+      "learning_rate": 1e-05,
+      "loss": 0.4015,
+      "step": 685
+    },
+    {
+      "epoch": 0.19769452449567723,
+      "grad_norm": 2.7222627650122377,
+      "learning_rate": 1e-05,
+      "loss": 0.5471,
+      "step": 686
+    },
+    {
+      "epoch": 0.19798270893371758,
+      "grad_norm": 1.6445933398287433,
+      "learning_rate": 1e-05,
+      "loss": 0.534,
+      "step": 687
+    },
+    {
+      "epoch": 0.19827089337175793,
+      "grad_norm": 1.6151151672539203,
+      "learning_rate": 1e-05,
+      "loss": 0.372,
+      "step": 688
+    },
+    {
+      "epoch": 0.19855907780979828,
+      "grad_norm": 1.670692914013598,
+      "learning_rate": 1e-05,
+      "loss": 0.576,
+      "step": 689
+    },
+    {
+      "epoch": 0.1988472622478386,
+      "grad_norm": 1.6555833971116176,
+      "learning_rate": 1e-05,
+      "loss": 0.5351,
+      "step": 690
+    },
+    {
+      "epoch": 0.19913544668587896,
+      "grad_norm": 3.2247248878775276,
+      "learning_rate": 1e-05,
+      "loss": 0.3772,
+      "step": 691
+    },
+    {
+      "epoch": 0.1994236311239193,
+      "grad_norm": 1.928176484072151,
+      "learning_rate": 1e-05,
+      "loss": 0.5423,
+      "step": 692
+    },
+    {
+      "epoch": 0.19971181556195966,
+      "grad_norm": 1.6063135640174004,
+      "learning_rate": 1e-05,
+      "loss": 0.545,
+      "step": 693
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.871140642527986,
+      "learning_rate": 1e-05,
+      "loss": 0.375,
+      "step": 694
+    },
+    {
+      "epoch": 0.20028818443804033,
+      "grad_norm": 1.8679099541746644,
+      "learning_rate": 1e-05,
+      "loss": 0.5682,
+      "step": 695
+    },
+    {
+      "epoch": 0.20057636887608069,
+      "grad_norm": 1.7962709463328976,
+      "learning_rate": 1e-05,
+      "loss": 0.531,
+      "step": 696
+    },
+    {
+      "epoch": 0.20086455331412104,
+      "grad_norm": 2.493526817760628,
+      "learning_rate": 1e-05,
+      "loss": 0.3761,
+      "step": 697
+    },
+    {
+      "epoch": 0.2011527377521614,
+      "grad_norm": 2.435720278720426,
+      "learning_rate": 1e-05,
+      "loss": 0.542,
+      "step": 698
+    },
+    {
+      "epoch": 0.20144092219020174,
+      "grad_norm": 1.2839405537920643,
+      "learning_rate": 1e-05,
+      "loss": 0.5419,
+      "step": 699
+    },
+    {
+      "epoch": 0.2017291066282421,
+      "grad_norm": 3.593143484480332,
+      "learning_rate": 1e-05,
+      "loss": 0.3861,
+      "step": 700
+    },
+    {
+      "epoch": 0.2020172910662824,
+      "grad_norm": 3.4186791670908727,
+      "learning_rate": 1e-05,
+      "loss": 0.5457,
+      "step": 701
+    },
+    {
+      "epoch": 0.20230547550432276,
+      "grad_norm": 1.863503051365215,
+      "learning_rate": 1e-05,
+      "loss": 0.5403,
+      "step": 702
+    },
+    {
+      "epoch": 0.2025936599423631,
+      "grad_norm": 2.34662159725905,
+      "learning_rate": 1e-05,
+      "loss": 0.3919,
+      "step": 703
+    },
+    {
+      "epoch": 0.20288184438040346,
+      "grad_norm": 3.256613608855492,
+      "learning_rate": 1e-05,
+      "loss": 0.5304,
+      "step": 704
+    },
+    {
+      "epoch": 0.20317002881844382,
+      "grad_norm": 1.3295205271030683,
+      "learning_rate": 1e-05,
+      "loss": 0.5202,
+      "step": 705
+    },
+    {
+      "epoch": 0.20345821325648414,
+      "grad_norm": 3.7719510069622966,
+      "learning_rate": 1e-05,
+      "loss": 0.3744,
+      "step": 706
+    },
+    {
+      "epoch": 0.2037463976945245,
+      "grad_norm": 1.9907034061327187,
+      "learning_rate": 1e-05,
+      "loss": 0.5341,
+      "step": 707
+    },
+    {
+      "epoch": 0.20403458213256484,
+      "grad_norm": 2.0807541314953215,
+      "learning_rate": 1e-05,
+      "loss": 0.5247,
+      "step": 708
+    },
+    {
+      "epoch": 0.2043227665706052,
+      "grad_norm": 2.1837956868707886,
+      "learning_rate": 1e-05,
+      "loss": 0.3812,
+      "step": 709
+    },
+    {
+      "epoch": 0.20461095100864554,
+      "grad_norm": 2.347923727511402,
+      "learning_rate": 1e-05,
+      "loss": 0.5407,
+      "step": 710
+    },
+    {
+      "epoch": 0.2048991354466859,
+      "grad_norm": 1.8202318162086493,
+      "learning_rate": 1e-05,
+      "loss": 0.5271,
+      "step": 711
+    },
+    {
+      "epoch": 0.20518731988472622,
+      "grad_norm": 4.588594008908393,
+      "learning_rate": 1e-05,
+      "loss": 0.3766,
+      "step": 712
+    },
+    {
+      "epoch": 0.20547550432276657,
+      "grad_norm": 2.5052719320350194,
+      "learning_rate": 1e-05,
+      "loss": 0.491,
+      "step": 713
+    },
+    {
+      "epoch": 0.20576368876080692,
+      "grad_norm": 2.272802999770661,
+      "learning_rate": 1e-05,
+      "loss": 0.5417,
+      "step": 714
+    },
+    {
+      "epoch": 0.20605187319884727,
+      "grad_norm": 2.5405417145197604,
+      "learning_rate": 1e-05,
+      "loss": 0.3714,
+      "step": 715
+    },
+    {
+      "epoch": 0.20634005763688762,
+      "grad_norm": 2.576715752621875,
+      "learning_rate": 1e-05,
+      "loss": 0.5217,
+      "step": 716
+    },
+    {
+      "epoch": 0.20662824207492794,
+      "grad_norm": 2.7334756717737028,
+      "learning_rate": 1e-05,
+      "loss": 0.5387,
+      "step": 717
+    },
+    {
+      "epoch": 0.2069164265129683,
+      "grad_norm": 2.4819037880568433,
+      "learning_rate": 1e-05,
+      "loss": 0.3714,
+      "step": 718
+    },
+    {
+      "epoch": 0.20720461095100864,
+      "grad_norm": 2.2184880201269226,
+      "learning_rate": 1e-05,
+      "loss": 0.5229,
+      "step": 719
+    },
+    {
+      "epoch": 0.207492795389049,
+      "grad_norm": 1.5608036614364995,
+      "learning_rate": 1e-05,
+      "loss": 0.5182,
+      "step": 720
+    },
+    {
+      "epoch": 0.20778097982708935,
+      "grad_norm": 1.8292187821414867,
+      "learning_rate": 1e-05,
+      "loss": 0.3629,
+      "step": 721
+    },
+    {
+      "epoch": 0.2080691642651297,
+      "grad_norm": 2.266258143468976,
+      "learning_rate": 1e-05,
+      "loss": 0.5391,
+      "step": 722
+    },
+    {
+      "epoch": 0.20835734870317002,
+      "grad_norm": 1.6281311577197308,
+      "learning_rate": 1e-05,
+      "loss": 0.5395,
+      "step": 723
+    },
+    {
+      "epoch": 0.20864553314121037,
+      "grad_norm": 2.8517201813458475,
+      "learning_rate": 1e-05,
+      "loss": 0.3878,
+      "step": 724
+    },
+    {
+      "epoch": 0.20893371757925072,
+      "grad_norm": 2.12071000582212,
+      "learning_rate": 1e-05,
+      "loss": 0.544,
+      "step": 725
+    },
+    {
+      "epoch": 0.20922190201729107,
+      "grad_norm": 1.4786087983239724,
+      "learning_rate": 1e-05,
+      "loss": 0.5314,
+      "step": 726
+    },
+    {
+      "epoch": 0.20951008645533142,
+      "grad_norm": 1.5577610909440798,
+      "learning_rate": 1e-05,
+      "loss": 0.3701,
+      "step": 727
+    },
+    {
+      "epoch": 0.20979827089337175,
+      "grad_norm": 3.1641406948540847,
+      "learning_rate": 1e-05,
+      "loss": 0.5592,
+      "step": 728
+    },
+    {
+      "epoch": 0.2100864553314121,
+      "grad_norm": 1.4810639216076793,
+      "learning_rate": 1e-05,
+      "loss": 0.5247,
+      "step": 729
+    },
+    {
+      "epoch": 0.21037463976945245,
+      "grad_norm": 3.1653545929532205,
+      "learning_rate": 1e-05,
+      "loss": 0.3695,
+      "step": 730
+    },
+    {
+      "epoch": 0.2106628242074928,
+      "grad_norm": 2.1220789892916563,
+      "learning_rate": 1e-05,
+      "loss": 0.5401,
+      "step": 731
+    },
+    {
+      "epoch": 0.21095100864553315,
+      "grad_norm": 2.251796845084247,
+      "learning_rate": 1e-05,
+      "loss": 0.5468,
+      "step": 732
+    },
+    {
+      "epoch": 0.21123919308357347,
+      "grad_norm": 1.8026942421448902,
+      "learning_rate": 1e-05,
+      "loss": 0.3785,
+      "step": 733
+    },
+    {
+      "epoch": 0.21152737752161382,
+      "grad_norm": 2.674342159575944,
+      "learning_rate": 1e-05,
+      "loss": 0.5194,
+      "step": 734
+    },
+    {
+      "epoch": 0.21181556195965417,
+      "grad_norm": 1.6456776112403628,
+      "learning_rate": 1e-05,
+      "loss": 0.5291,
+      "step": 735
+    },
+    {
+      "epoch": 0.21210374639769453,
+      "grad_norm": 12.680453282507559,
+      "learning_rate": 1e-05,
+      "loss": 0.3911,
+      "step": 736
+    },
+    {
+      "epoch": 0.21239193083573488,
+      "grad_norm": 2.691332725684379,
+      "learning_rate": 1e-05,
+      "loss": 0.5383,
+      "step": 737
+    },
+    {
+      "epoch": 0.21268011527377523,
+      "grad_norm": 1.5529169372421723,
+      "learning_rate": 1e-05,
+      "loss": 0.5354,
+      "step": 738
+    },
+    {
+      "epoch": 0.21296829971181555,
+      "grad_norm": 6.518363469400075,
+      "learning_rate": 1e-05,
+      "loss": 0.3985,
+      "step": 739
+    },
+    {
+      "epoch": 0.2132564841498559,
+      "grad_norm": 2.3740170317253706,
+      "learning_rate": 1e-05,
+      "loss": 0.5679,
+      "step": 740
+    },
+    {
+      "epoch": 0.21354466858789625,
+      "grad_norm": 1.7889993945011,
+      "learning_rate": 1e-05,
+      "loss": 0.5279,
+      "step": 741
+    },
+    {
+      "epoch": 0.2138328530259366,
+      "grad_norm": 3.515910998913752,
+      "learning_rate": 1e-05,
+      "loss": 0.3865,
+      "step": 742
+    },
+    {
+      "epoch": 0.21412103746397695,
+      "grad_norm": 2.548307289511562,
+      "learning_rate": 1e-05,
+      "loss": 0.5277,
+      "step": 743
+    },
+    {
+      "epoch": 0.21440922190201728,
+      "grad_norm": 1.3488652772286054,
+      "learning_rate": 1e-05,
+      "loss": 0.5174,
+      "step": 744
+    },
+    {
+      "epoch": 0.21469740634005763,
+      "grad_norm": 4.249962624071357,
+      "learning_rate": 1e-05,
+      "loss": 0.3561,
+      "step": 745
+    },
+    {
+      "epoch": 0.21498559077809798,
+      "grad_norm": 2.4901332890788437,
+      "learning_rate": 1e-05,
+      "loss": 0.5815,
+      "step": 746
+    },
+    {
+      "epoch": 0.21527377521613833,
+      "grad_norm": 1.3670992784434364,
+      "learning_rate": 1e-05,
+      "loss": 0.5222,
+      "step": 747
+    },
+    {
+      "epoch": 0.21556195965417868,
+      "grad_norm": 2.110011740407682,
+      "learning_rate": 1e-05,
+      "loss": 0.378,
+      "step": 748
+    },
+    {
+      "epoch": 0.21585014409221903,
+      "grad_norm": 2.1655089030291013,
+      "learning_rate": 1e-05,
+      "loss": 0.5121,
+      "step": 749
+    },
+    {
+      "epoch": 0.21613832853025935,
+      "grad_norm": 1.4820566524783922,
+      "learning_rate": 1e-05,
+      "loss": 0.5469,
+      "step": 750
+    },
+    {
+      "epoch": 0.21613832853025935,
+      "eval_websight_new_IoU": 0.01727907732129097,
+      "eval_websight_new_MAE_x": 286.7799987792969,
+      "eval_websight_new_MAE_y": 523.760009765625,
+      "eval_websight_new_coord_probability": 0.7429283261299133,
+      "eval_websight_new_correct_token_probability": 0.0011671152897179127,
+      "eval_websight_new_inside_bbox": 0.019999999552965164,
+      "eval_websight_new_loss": 0.3837199807167053,
+      "eval_websight_new_range_100_probability": 0.09132597595453262,
+      "eval_websight_new_range_20_probability": 0.01654745265841484,
+      "eval_websight_new_range_5_probability": 0.003870067186653614,
+      "eval_websight_new_runtime": 33.8429,
+      "eval_websight_new_samples_per_second": 1.477,
+      "eval_websight_new_steps_per_second": 0.207,
+      "step": 750
+    },
+    {
+      "epoch": 0.21613832853025935,
+      "eval_seeclick_IoU": 0.03237491846084595,
+      "eval_seeclick_MAE_x": 473.1000061035156,
+      "eval_seeclick_MAE_y": 515.7999877929688,
+      "eval_seeclick_coord_probability": 0.7453280687332153,
+      "eval_seeclick_correct_token_probability": 0.001050356775522232,
+      "eval_seeclick_inside_bbox": 0.05999999865889549,
+      "eval_seeclick_loss": 0.4347634017467499,
+      "eval_seeclick_range_100_probability": 0.11753538995981216,
+      "eval_seeclick_range_20_probability": 0.03401518613100052,
+      "eval_seeclick_range_5_probability": 0.006139020901173353,
+      "eval_seeclick_runtime": 42.062,
+      "eval_seeclick_samples_per_second": 1.189,
+      "eval_seeclick_steps_per_second": 0.166,
+      "step": 750
+    },
+    {
+      "epoch": 0.21613832853025935,
+      "eval_icons_IoU": 0.003064297139644623,
+      "eval_icons_MAE_x": 473.5400085449219,
+      "eval_icons_MAE_y": 634.4000244140625,
+      "eval_icons_coord_probability": 0.7481094598770142,
+      "eval_icons_correct_token_probability": 0.005065434146672487,
+      "eval_icons_inside_bbox": 0.0,
+      "eval_icons_loss": 0.5070137977600098,
+      "eval_icons_range_100_probability": 0.2729437053203583,
+      "eval_icons_range_20_probability": 0.09965262562036514,
+      "eval_icons_range_5_probability": 0.02936328575015068,
+      "eval_icons_runtime": 43.2884,
+      "eval_icons_samples_per_second": 1.155,
+      "eval_icons_steps_per_second": 0.162,
+      "step": 750
+    },
+    {
+      "epoch": 0.2164265129682997,
+      "grad_norm": 2.1888763443405295,
+      "learning_rate": 1e-05,
+      "loss": 0.3598,
+      "step": 751
+    },
+    {
+      "epoch": 0.21671469740634006,
+      "grad_norm": 2.2774890567448027,
+      "learning_rate": 1e-05,
+      "loss": 0.5035,
+      "step": 752
+    },
+    {
+      "epoch": 0.2170028818443804,
+      "grad_norm": 1.4442456121352483,
+      "learning_rate": 1e-05,
+      "loss": 0.5184,
+      "step": 753
+    },
+    {
+      "epoch": 0.21729106628242076,
+      "grad_norm": 3.785601337006597,
+      "learning_rate": 1e-05,
+      "loss": 0.3839,
+      "step": 754
+    },
+    {
+      "epoch": 0.21757925072046108,
+      "grad_norm": 3.1321856047605956,
+      "learning_rate": 1e-05,
+      "loss": 0.5039,
+      "step": 755
+    },
+    {
+      "epoch": 0.21786743515850143,
+      "grad_norm": 1.9440011815842064,
+      "learning_rate": 1e-05,
+      "loss": 0.5302,
+      "step": 756
+    },
+    {
+      "epoch": 0.21815561959654178,
+      "grad_norm": 2.7511348005259344,
+      "learning_rate": 1e-05,
+      "loss": 0.3573,
+      "step": 757
+    },
+    {
+      "epoch": 0.21844380403458213,
+      "grad_norm": 2.775030827757665,
+      "learning_rate": 1e-05,
+      "loss": 0.5631,
+      "step": 758
+    },
+    {
+      "epoch": 0.21873198847262249,
+      "grad_norm": 1.933287538834539,
+      "learning_rate": 1e-05,
+      "loss": 0.5257,
+      "step": 759
+    },
+    {
+      "epoch": 0.21902017291066284,
+      "grad_norm": 1.6546472853128393,
+      "learning_rate": 1e-05,
+      "loss": 0.3678,
+      "step": 760
+    },
+    {
+      "epoch": 0.21930835734870316,
+      "grad_norm": 2.604311079856137,
+      "learning_rate": 1e-05,
+      "loss": 0.5335,
+      "step": 761
+    },
+    {
+      "epoch": 0.2195965417867435,
+      "grad_norm": 2.2191616680988573,
+      "learning_rate": 1e-05,
+      "loss": 0.5456,
+      "step": 762
+    },
+    {
+      "epoch": 0.21988472622478386,
+      "grad_norm": 2.3295803403283553,
+      "learning_rate": 1e-05,
+      "loss": 0.3866,
+      "step": 763
+    },
+    {
+      "epoch": 0.2201729106628242,
+      "grad_norm": 1.9475593674895146,
+      "learning_rate": 1e-05,
+      "loss": 0.5194,
+      "step": 764
+    },
+    {
+      "epoch": 0.22046109510086456,
+      "grad_norm": 1.6865455352852274,
+      "learning_rate": 1e-05,
+      "loss": 0.5289,
+      "step": 765
+    },
+    {
+      "epoch": 0.22074927953890489,
+      "grad_norm": 1.7278367147952058,
+      "learning_rate": 1e-05,
+      "loss": 0.3569,
+      "step": 766
+    },
+    {
+      "epoch": 0.22103746397694524,
+      "grad_norm": 2.9832195025090593,
+      "learning_rate": 1e-05,
+      "loss": 0.5357,
+      "step": 767
+    },
+    {
+      "epoch": 0.2213256484149856,
+      "grad_norm": 1.3242254653617855,
+      "learning_rate": 1e-05,
+      "loss": 0.5151,
+      "step": 768
+    },
+    {
+      "epoch": 0.22161383285302594,
+      "grad_norm": 1.8377308740842033,
+      "learning_rate": 1e-05,
+      "loss": 0.373,
+      "step": 769
+    },
+    {
+      "epoch": 0.2219020172910663,
+      "grad_norm": 2.2036161876977163,
+      "learning_rate": 1e-05,
+      "loss": 0.5327,
+      "step": 770
+    },
+    {
+      "epoch": 0.22219020172910664,
+      "grad_norm": 1.6665259044580085,
+      "learning_rate": 1e-05,
+      "loss": 0.5187,
+      "step": 771
+    },
+    {
+      "epoch": 0.22247838616714696,
+      "grad_norm": 2.606597787113097,
+      "learning_rate": 1e-05,
+      "loss": 0.3599,
+      "step": 772
+    },
+    {
+      "epoch": 0.22276657060518731,
+      "grad_norm": 3.7691354716746854,
+      "learning_rate": 1e-05,
+      "loss": 0.5282,
+      "step": 773
+    },
+    {
+      "epoch": 0.22305475504322766,
+      "grad_norm": 1.926087044611918,
+      "learning_rate": 1e-05,
+      "loss": 0.5186,
+      "step": 774
+    },
+    {
+      "epoch": 0.22334293948126802,
+      "grad_norm": 2.1872704449275466,
+      "learning_rate": 1e-05,
+      "loss": 0.3543,
+      "step": 775
+    },
+    {
+      "epoch": 0.22363112391930837,
+      "grad_norm": 2.442419185379171,
+      "learning_rate": 1e-05,
+      "loss": 0.5499,
+      "step": 776
+    },
+    {
+      "epoch": 0.2239193083573487,
+      "grad_norm": 1.7615743180506984,
+      "learning_rate": 1e-05,
+      "loss": 0.5129,
+      "step": 777
+    },
+    {
+      "epoch": 0.22420749279538904,
+      "grad_norm": 3.5936095177088605,
+      "learning_rate": 1e-05,
+      "loss": 0.3595,
+      "step": 778
+    },
+    {
+      "epoch": 0.2244956772334294,
+      "grad_norm": 3.9778261145163514,
+      "learning_rate": 1e-05,
+      "loss": 0.5125,
+      "step": 779
+    },
+    {
+      "epoch": 0.22478386167146974,
+      "grad_norm": 1.9736950314580592,
+      "learning_rate": 1e-05,
+      "loss": 0.5193,
+      "step": 780
+    },
+    {
+      "epoch": 0.2250720461095101,
+      "grad_norm": 2.964728492787288,
+      "learning_rate": 1e-05,
+      "loss": 0.3669,
+      "step": 781
+    },
+    {
+      "epoch": 0.22536023054755044,
+      "grad_norm": 2.784188417504383,
+      "learning_rate": 1e-05,
+      "loss": 0.5407,
+      "step": 782
+    },
+    {
+      "epoch": 0.22564841498559077,
+      "grad_norm": 3.202046534108722,
+      "learning_rate": 1e-05,
+      "loss": 0.5128,
+      "step": 783
+    },
+    {
+      "epoch": 0.22593659942363112,
+      "grad_norm": 2.0773279346175006,
+      "learning_rate": 1e-05,
+      "loss": 0.3715,
+      "step": 784
+    },
+    {
+      "epoch": 0.22622478386167147,
+      "grad_norm": 2.7721339436817463,
+      "learning_rate": 1e-05,
+      "loss": 0.5678,
+      "step": 785
+    },
+    {
+      "epoch": 0.22651296829971182,
+      "grad_norm": 2.052925896005973,
+      "learning_rate": 1e-05,
+      "loss": 0.5251,
+      "step": 786
+    },
+    {
+      "epoch": 0.22680115273775217,
+      "grad_norm": 1.7600006840401639,
+      "learning_rate": 1e-05,
+      "loss": 0.3686,
+      "step": 787
+    },
+    {
+      "epoch": 0.2270893371757925,
+      "grad_norm": 3.076061233667846,
+      "learning_rate": 1e-05,
+      "loss": 0.5062,
+      "step": 788
+    },
+    {
+      "epoch": 0.22737752161383284,
+      "grad_norm": 1.8007049947721312,
+      "learning_rate": 1e-05,
+      "loss": 0.5221,
+      "step": 789
+    },
+    {
+      "epoch": 0.2276657060518732,
+      "grad_norm": 3.1757006191847856,
+      "learning_rate": 1e-05,
+      "loss": 0.3777,
+      "step": 790
+    },
+    {
+      "epoch": 0.22795389048991355,
+      "grad_norm": 2.836473562759461,
+      "learning_rate": 1e-05,
+      "loss": 0.5497,
+      "step": 791
+    },
+    {
+      "epoch": 0.2282420749279539,
+      "grad_norm": 1.9915125344875975,
+      "learning_rate": 1e-05,
+      "loss": 0.517,
+      "step": 792
+    },
+    {
+      "epoch": 0.22853025936599425,
+      "grad_norm": 1.9379971324561747,
+      "learning_rate": 1e-05,
+      "loss": 0.3648,
+      "step": 793
+    },
+    {
+      "epoch": 0.22881844380403457,
+      "grad_norm": 2.420860422194887,
+      "learning_rate": 1e-05,
+      "loss": 0.5482,
+      "step": 794
+    },
+    {
+      "epoch": 0.22910662824207492,
+      "grad_norm": 2.1897133743195734,
+      "learning_rate": 1e-05,
+      "loss": 0.5273,
+      "step": 795
+    },
+    {
+      "epoch": 0.22939481268011527,
+      "grad_norm": 1.319229454100792,
+      "learning_rate": 1e-05,
+      "loss": 0.3784,
+      "step": 796
+    },
+    {
+      "epoch": 0.22968299711815562,
+      "grad_norm": 2.118156858440375,
+      "learning_rate": 1e-05,
+      "loss": 0.5125,
+      "step": 797
+    },
+    {
+      "epoch": 0.22997118155619597,
+      "grad_norm": 1.448708704573136,
+      "learning_rate": 1e-05,
+      "loss": 0.5189,
+      "step": 798
+    },
+    {
+      "epoch": 0.2302593659942363,
+      "grad_norm": 2.0463003698498,
+      "learning_rate": 1e-05,
+      "loss": 0.3895,
+      "step": 799
+    },
+    {
+      "epoch": 0.23054755043227665,
+      "grad_norm": 2.006341116386733,
+      "learning_rate": 1e-05,
+      "loss": 0.5205,
+      "step": 800
+    },
+    {
+      "epoch": 0.230835734870317,
+      "grad_norm": 2.0493659703648808,
+      "learning_rate": 1e-05,
+      "loss": 0.5181,
+      "step": 801
+    },
+    {
+      "epoch": 0.23112391930835735,
+      "grad_norm": 1.4723595443050466,
+      "learning_rate": 1e-05,
+      "loss": 0.3702,
+      "step": 802
+    },
+    {
+      "epoch": 0.2314121037463977,
+      "grad_norm": 1.9647835931699873,
+      "learning_rate": 1e-05,
+      "loss": 0.5322,
+      "step": 803
+    },
+    {
+      "epoch": 0.23170028818443805,
+      "grad_norm": 1.5165203723354632,
+      "learning_rate": 1e-05,
+      "loss": 0.524,
+      "step": 804
+    },
+    {
+      "epoch": 0.23198847262247838,
+      "grad_norm": 3.1727305578969958,
+      "learning_rate": 1e-05,
+      "loss": 0.3691,
+      "step": 805
+    },
+    {
+      "epoch": 0.23227665706051873,
+      "grad_norm": 2.455598470170328,
+      "learning_rate": 1e-05,
+      "loss": 0.5294,
+      "step": 806
+    },
+    {
+      "epoch": 0.23256484149855908,
+      "grad_norm": 1.6062968633545127,
+      "learning_rate": 1e-05,
+      "loss": 0.5264,
+      "step": 807
+    },
+    {
+      "epoch": 0.23285302593659943,
+      "grad_norm": 1.6464516538182972,
+      "learning_rate": 1e-05,
+      "loss": 0.3525,
+      "step": 808
+    },
+    {
+      "epoch": 0.23314121037463978,
+      "grad_norm": 2.9692256786690727,
+      "learning_rate": 1e-05,
+      "loss": 0.5277,
+      "step": 809
+    },
+    {
+      "epoch": 0.2334293948126801,
+      "grad_norm": 1.6283622611780912,
+      "learning_rate": 1e-05,
+      "loss": 0.5228,
+      "step": 810
+    },
+    {
+      "epoch": 0.23371757925072045,
+      "grad_norm": 3.452754600099684,
+      "learning_rate": 1e-05,
+      "loss": 0.352,
+      "step": 811
+    },
+    {
+      "epoch": 0.2340057636887608,
+      "grad_norm": 2.7458062624189483,
+      "learning_rate": 1e-05,
+      "loss": 0.5292,
+      "step": 812
+    },
+    {
+      "epoch": 0.23429394812680115,
+      "grad_norm": 1.4757636230943205,
+      "learning_rate": 1e-05,
+      "loss": 0.5178,
+      "step": 813
+    },
+    {
+      "epoch": 0.2345821325648415,
+      "grad_norm": 3.401862068032278,
+      "learning_rate": 1e-05,
+      "loss": 0.3441,
+      "step": 814
+    },
+    {
+      "epoch": 0.23487031700288186,
+      "grad_norm": 2.4224097077513997,
+      "learning_rate": 1e-05,
+      "loss": 0.5594,
+      "step": 815
+    },
+    {
+      "epoch": 0.23515850144092218,
+      "grad_norm": 1.5672899266594948,
+      "learning_rate": 1e-05,
+      "loss": 0.5232,
+      "step": 816
+    },
+    {
+      "epoch": 0.23544668587896253,
+      "grad_norm": 1.956384688894368,
+      "learning_rate": 1e-05,
+      "loss": 0.3676,
+      "step": 817
+    },
+    {
+      "epoch": 0.23573487031700288,
+      "grad_norm": 3.716738993954289,
+      "learning_rate": 1e-05,
+      "loss": 0.5148,
+      "step": 818
+    },
+    {
+      "epoch": 0.23602305475504323,
+      "grad_norm": 1.9951514137318538,
+      "learning_rate": 1e-05,
+      "loss": 0.5198,
+      "step": 819
+    },
+    {
+      "epoch": 0.23631123919308358,
+      "grad_norm": 3.4242391430302153,
+      "learning_rate": 1e-05,
+      "loss": 0.3363,
+      "step": 820
+    },
+    {
+      "epoch": 0.2365994236311239,
+      "grad_norm": 2.772418569518423,
+      "learning_rate": 1e-05,
+      "loss": 0.5342,
+      "step": 821
+    },
+    {
+      "epoch": 0.23688760806916426,
+      "grad_norm": 1.4757115320734255,
+      "learning_rate": 1e-05,
+      "loss": 0.5151,
+      "step": 822
+    },
+    {
+      "epoch": 0.2371757925072046,
+      "grad_norm": 2.1011737845303204,
+      "learning_rate": 1e-05,
+      "loss": 0.3586,
+      "step": 823
+    },
+    {
+      "epoch": 0.23746397694524496,
+      "grad_norm": 2.9805427737176577,
+      "learning_rate": 1e-05,
+      "loss": 0.5272,
+      "step": 824
+    },
+    {
+      "epoch": 0.2377521613832853,
+      "grad_norm": 1.3117274089431927,
+      "learning_rate": 1e-05,
+      "loss": 0.5079,
+      "step": 825
+    },
+    {
+      "epoch": 0.23804034582132566,
+      "grad_norm": 2.3328559839201892,
+      "learning_rate": 1e-05,
+      "loss": 0.3454,
+      "step": 826
+    },
+    {
+      "epoch": 0.23832853025936598,
+      "grad_norm": 3.856422134773823,
+      "learning_rate": 1e-05,
+      "loss": 0.5304,
+      "step": 827
+    },
+    {
+      "epoch": 0.23861671469740633,
+      "grad_norm": 1.4266057337400306,
+      "learning_rate": 1e-05,
+      "loss": 0.5173,
+      "step": 828
+    },
+    {
+      "epoch": 0.23890489913544669,
+      "grad_norm": 1.8049556866780567,
+      "learning_rate": 1e-05,
+      "loss": 0.3554,
+      "step": 829
+    },
+    {
+      "epoch": 0.23919308357348704,
+      "grad_norm": 3.0646174126487233,
+      "learning_rate": 1e-05,
+      "loss": 0.5417,
+      "step": 830
+    },
+    {
+      "epoch": 0.2394812680115274,
+      "grad_norm": 1.5027819303170922,
+      "learning_rate": 1e-05,
+      "loss": 0.5095,
+      "step": 831
+    },
+    {
+      "epoch": 0.2397694524495677,
+      "grad_norm": 2.8189905860249995,
+      "learning_rate": 1e-05,
+      "loss": 0.374,
+      "step": 832
+    },
+    {
+      "epoch": 0.24005763688760806,
+      "grad_norm": 2.7766436345114687,
+      "learning_rate": 1e-05,
+      "loss": 0.5262,
+      "step": 833
+    },
+    {
+      "epoch": 0.2403458213256484,
+      "grad_norm": 1.6962941988775544,
+      "learning_rate": 1e-05,
+      "loss": 0.5094,
+      "step": 834
+    },
+    {
+      "epoch": 0.24063400576368876,
+      "grad_norm": 3.9779864695789833,
+      "learning_rate": 1e-05,
+      "loss": 0.3477,
+      "step": 835
+    },
+    {
+      "epoch": 0.24092219020172911,
+      "grad_norm": 3.946522752119048,
+      "learning_rate": 1e-05,
+      "loss": 0.5166,
+      "step": 836
+    },
+    {
+      "epoch": 0.24121037463976946,
+      "grad_norm": 2.3943887033548426,
+      "learning_rate": 1e-05,
+      "loss": 0.5168,
+      "step": 837
+    },
+    {
+      "epoch": 0.2414985590778098,
+      "grad_norm": 2.329503948048036,
+      "learning_rate": 1e-05,
+      "loss": 0.35,
+      "step": 838
+    },
+    {
+      "epoch": 0.24178674351585014,
+      "grad_norm": 2.4187939347443494,
+      "learning_rate": 1e-05,
+      "loss": 0.4892,
+      "step": 839
+    },
+    {
+      "epoch": 0.2420749279538905,
+      "grad_norm": 1.47487900374655,
+      "learning_rate": 1e-05,
+      "loss": 0.5119,
+      "step": 840
+    },
+    {
+      "epoch": 0.24236311239193084,
+      "grad_norm": 3.800528926250578,
+      "learning_rate": 1e-05,
+      "loss": 0.3457,
+      "step": 841
+    },
+    {
+      "epoch": 0.2426512968299712,
+      "grad_norm": 2.5069874848836817,
+      "learning_rate": 1e-05,
+      "loss": 0.4978,
+      "step": 842
+    },
+    {
+      "epoch": 0.24293948126801151,
+      "grad_norm": 1.8611740241971595,
+      "learning_rate": 1e-05,
+      "loss": 0.517,
+      "step": 843
+    },
+    {
+      "epoch": 0.24322766570605187,
+      "grad_norm": 1.8766061195321277,
+      "learning_rate": 1e-05,
+      "loss": 0.3647,
+      "step": 844
+    },
+    {
+      "epoch": 0.24351585014409222,
+      "grad_norm": 4.214162592822582,
+      "learning_rate": 1e-05,
+      "loss": 0.5036,
+      "step": 845
+    },
+    {
+      "epoch": 0.24380403458213257,
+      "grad_norm": 2.2303791341440635,
+      "learning_rate": 1e-05,
+      "loss": 0.5178,
+      "step": 846
+    },
+    {
+      "epoch": 0.24409221902017292,
+      "grad_norm": 2.0840787909368825,
+      "learning_rate": 1e-05,
+      "loss": 0.3632,
+      "step": 847
+    },
+    {
+      "epoch": 0.24438040345821327,
+      "grad_norm": 2.0604828244399993,
+      "learning_rate": 1e-05,
+      "loss": 0.5498,
+      "step": 848
+    },
+    {
+      "epoch": 0.2446685878962536,
+      "grad_norm": 1.784944626914137,
+      "learning_rate": 1e-05,
+      "loss": 0.5183,
+      "step": 849
+    },
+    {
+      "epoch": 0.24495677233429394,
+      "grad_norm": 2.0313924719542897,
+      "learning_rate": 1e-05,
+      "loss": 0.3522,
+      "step": 850
+    },
+    {
+      "epoch": 0.2452449567723343,
+      "grad_norm": 2.52825075621921,
+      "learning_rate": 1e-05,
+      "loss": 0.5492,
+      "step": 851
+    },
+    {
+      "epoch": 0.24553314121037464,
+      "grad_norm": 1.3334641465216632,
+      "learning_rate": 1e-05,
+      "loss": 0.5258,
+      "step": 852
+    },
+    {
+      "epoch": 0.245821325648415,
+      "grad_norm": 2.2192424001626936,
+      "learning_rate": 1e-05,
+      "loss": 0.364,
+      "step": 853
+    },
+    {
+      "epoch": 0.24610951008645532,
+      "grad_norm": 2.460333177583637,
+      "learning_rate": 1e-05,
+      "loss": 0.5014,
+      "step": 854
+    },
+    {
+      "epoch": 0.24639769452449567,
+      "grad_norm": 2.606972913728769,
+      "learning_rate": 1e-05,
+      "loss": 0.5121,
+      "step": 855
+    },
+    {
+      "epoch": 0.24668587896253602,
+      "grad_norm": 1.4715149103233114,
+      "learning_rate": 1e-05,
+      "loss": 0.3742,
+      "step": 856
+    },
+    {
+      "epoch": 0.24697406340057637,
+      "grad_norm": 1.8883963445055671,
+      "learning_rate": 1e-05,
+      "loss": 0.5134,
+      "step": 857
+    },
+    {
+      "epoch": 0.24726224783861672,
+      "grad_norm": 1.5747373487704703,
+      "learning_rate": 1e-05,
+      "loss": 0.5187,
+      "step": 858
+    },
+    {
+      "epoch": 0.24755043227665707,
+      "grad_norm": 1.6956743231061768,
+      "learning_rate": 1e-05,
+      "loss": 0.3711,
+      "step": 859
+    },
+    {
+      "epoch": 0.2478386167146974,
+      "grad_norm": 2.0715775798740377,
+      "learning_rate": 1e-05,
+      "loss": 0.5167,
+      "step": 860
+    },
+    {
+      "epoch": 0.24812680115273775,
+      "grad_norm": 1.4829408931933024,
+      "learning_rate": 1e-05,
+      "loss": 0.5145,
+      "step": 861
+    },
+    {
+      "epoch": 0.2484149855907781,
+      "grad_norm": 3.0141285576727066,
+      "learning_rate": 1e-05,
+      "loss": 0.3511,
+      "step": 862
+    },
+    {
+      "epoch": 0.24870317002881845,
+      "grad_norm": 2.4860482609325785,
+      "learning_rate": 1e-05,
+      "loss": 0.534,
+      "step": 863
+    },
+    {
+      "epoch": 0.2489913544668588,
+      "grad_norm": 1.2180242319881303,
+      "learning_rate": 1e-05,
+      "loss": 0.5122,
+      "step": 864
+    },
+    {
+      "epoch": 0.24927953890489912,
+      "grad_norm": 1.5460133411288588,
+      "learning_rate": 1e-05,
+      "loss": 0.3442,
+      "step": 865
+    },
+    {
+      "epoch": 0.24956772334293947,
+      "grad_norm": 2.270173704845945,
+      "learning_rate": 1e-05,
+      "loss": 0.5334,
+      "step": 866
+    },
+    {
+      "epoch": 0.24985590778097982,
+      "grad_norm": 1.704269971401325,
+      "learning_rate": 1e-05,
+      "loss": 0.5026,
+      "step": 867
+    },
+    {
+      "epoch": 0.2501440922190202,
+      "grad_norm": 1.4900878720973898,
+      "learning_rate": 1e-05,
+      "loss": 0.3544,
+      "step": 868
+    },
+    {
+      "epoch": 0.2504322766570605,
+      "grad_norm": 2.5843336862710884,
+      "learning_rate": 1e-05,
+      "loss": 0.5405,
+      "step": 869
+    },
+    {
+      "epoch": 0.2507204610951009,
+      "grad_norm": 1.3856792728436378,
+      "learning_rate": 1e-05,
+      "loss": 0.5126,
+      "step": 870
+    },
+    {
+      "epoch": 0.25100864553314123,
+      "grad_norm": 1.772133582598731,
+      "learning_rate": 1e-05,
+      "loss": 0.3522,
+      "step": 871
+    },
+    {
+      "epoch": 0.2512968299711816,
+      "grad_norm": 2.5978131685660655,
+      "learning_rate": 1e-05,
+      "loss": 0.5394,
+      "step": 872
+    },
+    {
+      "epoch": 0.2515850144092219,
+      "grad_norm": 1.4779713020696061,
+      "learning_rate": 1e-05,
+      "loss": 0.5172,
+      "step": 873
+    },
+    {
+      "epoch": 0.2518731988472622,
+      "grad_norm": 2.9554944616355,
+      "learning_rate": 1e-05,
+      "loss": 0.3455,
+      "step": 874
+    },
+    {
+      "epoch": 0.2521613832853026,
+      "grad_norm": 3.6095839901225832,
+      "learning_rate": 1e-05,
+      "loss": 0.5261,
+      "step": 875
+    },
+    {
+      "epoch": 0.2524495677233429,
+      "grad_norm": 1.863377495387552,
+      "learning_rate": 1e-05,
+      "loss": 0.5096,
+      "step": 876
+    },
+    {
+      "epoch": 0.2527377521613833,
+      "grad_norm": 2.6368724875045957,
+      "learning_rate": 1e-05,
+      "loss": 0.3461,
+      "step": 877
+    },
+    {
+      "epoch": 0.25302593659942363,
+      "grad_norm": 2.23478728416295,
+      "learning_rate": 1e-05,
+      "loss": 0.5454,
+      "step": 878
+    },
+    {
+      "epoch": 0.253314121037464,
+      "grad_norm": 2.1140083394089895,
+      "learning_rate": 1e-05,
+      "loss": 0.5199,
+      "step": 879
+    },
+    {
+      "epoch": 0.25360230547550433,
+      "grad_norm": 1.6564815866196856,
+      "learning_rate": 1e-05,
+      "loss": 0.3376,
+      "step": 880
+    },
+    {
+      "epoch": 0.2538904899135447,
+      "grad_norm": 2.2729324647369564,
+      "learning_rate": 1e-05,
+      "loss": 0.5389,
+      "step": 881
+    },
+    {
+      "epoch": 0.25417867435158503,
+      "grad_norm": 1.917905469875635,
+      "learning_rate": 1e-05,
+      "loss": 0.5209,
+      "step": 882
+    },
+    {
+      "epoch": 0.2544668587896254,
+      "grad_norm": 2.495268518614722,
+      "learning_rate": 1e-05,
+      "loss": 0.3325,
+      "step": 883
+    },
+    {
+      "epoch": 0.2547550432276657,
+      "grad_norm": 2.20632661090717,
+      "learning_rate": 1e-05,
+      "loss": 0.5398,
+      "step": 884
+    },
+    {
+      "epoch": 0.25504322766570603,
+      "grad_norm": 2.978182106779898,
+      "learning_rate": 1e-05,
+      "loss": 0.5031,
+      "step": 885
+    },
+    {
+      "epoch": 0.2553314121037464,
+      "grad_norm": 2.023394065311876,
+      "learning_rate": 1e-05,
+      "loss": 0.3531,
+      "step": 886
+    },
+    {
+      "epoch": 0.25561959654178673,
+      "grad_norm": 2.9384319821079714,
+      "learning_rate": 1e-05,
+      "loss": 0.5277,
+      "step": 887
+    },
+    {
+      "epoch": 0.2559077809798271,
+      "grad_norm": 2.305370048844002,
+      "learning_rate": 1e-05,
+      "loss": 0.5057,
+      "step": 888
+    },
+    {
+      "epoch": 0.25619596541786743,
+      "grad_norm": 2.6623452134444583,
+      "learning_rate": 1e-05,
+      "loss": 0.3657,
+      "step": 889
+    },
+    {
+      "epoch": 0.2564841498559078,
+      "grad_norm": 2.5367479962613975,
+      "learning_rate": 1e-05,
+      "loss": 0.5066,
+      "step": 890
+    },
+    {
+      "epoch": 0.25677233429394813,
+      "grad_norm": 1.419286418468924,
+      "learning_rate": 1e-05,
+      "loss": 0.5091,
+      "step": 891
+    },
+    {
+      "epoch": 0.2570605187319885,
+      "grad_norm": 2.1384303010117605,
+      "learning_rate": 1e-05,
+      "loss": 0.3489,
+      "step": 892
+    },
+    {
+      "epoch": 0.25734870317002884,
+      "grad_norm": 3.2342022984795125,
+      "learning_rate": 1e-05,
+      "loss": 0.527,
+      "step": 893
+    },
+    {
+      "epoch": 0.2576368876080692,
+      "grad_norm": 2.0204922629256608,
+      "learning_rate": 1e-05,
+      "loss": 0.5069,
+      "step": 894
+    },
+    {
+      "epoch": 0.2579250720461095,
+      "grad_norm": 4.76560770896601,
+      "learning_rate": 1e-05,
+      "loss": 0.3782,
+      "step": 895
+    },
+    {
+      "epoch": 0.25821325648414983,
+      "grad_norm": 3.2405572098774535,
+      "learning_rate": 1e-05,
+      "loss": 0.512,
+      "step": 896
+    },
+    {
+      "epoch": 0.2585014409221902,
+      "grad_norm": 1.799772119453473,
+      "learning_rate": 1e-05,
+      "loss": 0.5146,
+      "step": 897
+    },
+    {
+      "epoch": 0.25878962536023054,
+      "grad_norm": 1.8846168408199326,
+      "learning_rate": 1e-05,
+      "loss": 0.3511,
+      "step": 898
+    },
+    {
+      "epoch": 0.2590778097982709,
+      "grad_norm": 3.2201950166329483,
+      "learning_rate": 1e-05,
+      "loss": 0.5116,
+      "step": 899
+    },
+    {
+      "epoch": 0.25936599423631124,
+      "grad_norm": 2.538171058228147,
+      "learning_rate": 1e-05,
+      "loss": 0.5338,
+      "step": 900
+    },
+    {
+      "epoch": 0.2596541786743516,
+      "grad_norm": 3.35038891535301,
+      "learning_rate": 1e-05,
+      "loss": 0.3602,
+      "step": 901
+    },
+    {
+      "epoch": 0.25994236311239194,
+      "grad_norm": 2.3003666047019924,
+      "learning_rate": 1e-05,
+      "loss": 0.5488,
+      "step": 902
+    },
+    {
+      "epoch": 0.2602305475504323,
+      "grad_norm": 1.6378904104692014,
+      "learning_rate": 1e-05,
+      "loss": 0.5089,
+      "step": 903
+    },
+    {
+      "epoch": 0.26051873198847264,
+      "grad_norm": 1.9298438110931269,
+      "learning_rate": 1e-05,
+      "loss": 0.3465,
+      "step": 904
+    },
+    {
+      "epoch": 0.260806916426513,
+      "grad_norm": 2.4619128302226434,
+      "learning_rate": 1e-05,
+      "loss": 0.528,
+      "step": 905
+    },
+    {
+      "epoch": 0.2610951008645533,
+      "grad_norm": 1.3476313349250084,
+      "learning_rate": 1e-05,
+      "loss": 0.5095,
+      "step": 906
+    },
+    {
+      "epoch": 0.26138328530259364,
+      "grad_norm": 2.7887628111368166,
+      "learning_rate": 1e-05,
+      "loss": 0.3477,
+      "step": 907
+    },
+    {
+      "epoch": 0.261671469740634,
+      "grad_norm": 2.365357342524135,
+      "learning_rate": 1e-05,
+      "loss": 0.4832,
+      "step": 908
+    },
+    {
+      "epoch": 0.26195965417867434,
+      "grad_norm": 2.3627181380185625,
+      "learning_rate": 1e-05,
+      "loss": 0.5089,
+      "step": 909
+    },
+    {
+      "epoch": 0.2622478386167147,
+      "grad_norm": 2.876074731526754,
+      "learning_rate": 1e-05,
+      "loss": 0.3697,
+      "step": 910
+    },
+    {
+      "epoch": 0.26253602305475504,
+      "grad_norm": 2.6280887251899174,
+      "learning_rate": 1e-05,
+      "loss": 0.5383,
+      "step": 911
+    },
+    {
+      "epoch": 0.2628242074927954,
+      "grad_norm": 1.6832899182461254,
+      "learning_rate": 1e-05,
+      "loss": 0.5016,
+      "step": 912
+    },
+    {
+      "epoch": 0.26311239193083574,
+      "grad_norm": 1.1162671633980008,
+      "learning_rate": 1e-05,
+      "loss": 0.3379,
+      "step": 913
+    },
+    {
+      "epoch": 0.2634005763688761,
+      "grad_norm": 3.0120401514277373,
+      "learning_rate": 1e-05,
+      "loss": 0.528,
+      "step": 914
+    },
+    {
+      "epoch": 0.26368876080691644,
+      "grad_norm": 1.416544071898569,
+      "learning_rate": 1e-05,
+      "loss": 0.5118,
+      "step": 915
+    },
+    {
+      "epoch": 0.2639769452449568,
+      "grad_norm": 1.7925956476565545,
+      "learning_rate": 1e-05,
+      "loss": 0.3528,
+      "step": 916
+    },
+    {
+      "epoch": 0.2642651296829971,
+      "grad_norm": 2.6421410122111126,
+      "learning_rate": 1e-05,
+      "loss": 0.5276,
+      "step": 917
+    },
+    {
+      "epoch": 0.26455331412103744,
+      "grad_norm": 1.4449283784010145,
+      "learning_rate": 1e-05,
+      "loss": 0.5097,
+      "step": 918
+    },
+    {
+      "epoch": 0.2648414985590778,
+      "grad_norm": 1.62279020218201,
+      "learning_rate": 1e-05,
+      "loss": 0.3468,
+      "step": 919
+    },
+    {
+      "epoch": 0.26512968299711814,
+      "grad_norm": 2.412267330061665,
+      "learning_rate": 1e-05,
+      "loss": 0.5339,
+      "step": 920
+    },
+    {
+      "epoch": 0.2654178674351585,
+      "grad_norm": 2.012484193733642,
+      "learning_rate": 1e-05,
+      "loss": 0.501,
+      "step": 921
+    },
+    {
+      "epoch": 0.26570605187319885,
+      "grad_norm": 1.3534784316568522,
+      "learning_rate": 1e-05,
+      "loss": 0.339,
+      "step": 922
+    },
+    {
+      "epoch": 0.2659942363112392,
+      "grad_norm": 2.7367124146915134,
+      "learning_rate": 1e-05,
+      "loss": 0.5254,
+      "step": 923
+    },
+    {
+      "epoch": 0.26628242074927955,
+      "grad_norm": 1.9157433438308276,
+      "learning_rate": 1e-05,
+      "loss": 0.5042,
+      "step": 924
+    },
+    {
+      "epoch": 0.2665706051873199,
+      "grad_norm": 2.141332264753019,
+      "learning_rate": 1e-05,
+      "loss": 0.3566,
+      "step": 925
+    },
+    {
+      "epoch": 0.26685878962536025,
+      "grad_norm": 2.1711057484674297,
+      "learning_rate": 1e-05,
+      "loss": 0.5131,
+      "step": 926
+    },
+    {
+      "epoch": 0.2671469740634006,
+      "grad_norm": 1.4690749120533202,
+      "learning_rate": 1e-05,
+      "loss": 0.5162,
+      "step": 927
+    },
+    {
+      "epoch": 0.2674351585014409,
+      "grad_norm": 2.6789176677358673,
+      "learning_rate": 1e-05,
+      "loss": 0.3531,
+      "step": 928
+    },
+    {
+      "epoch": 0.26772334293948125,
+      "grad_norm": 2.264548478937063,
+      "learning_rate": 1e-05,
+      "loss": 0.5097,
+      "step": 929
+    },
+    {
+      "epoch": 0.2680115273775216,
+      "grad_norm": 1.5144619403127813,
+      "learning_rate": 1e-05,
+      "loss": 0.5055,
+      "step": 930
+    },
+    {
+      "epoch": 0.26829971181556195,
+      "grad_norm": 2.552370856032433,
+      "learning_rate": 1e-05,
+      "loss": 0.3574,
+      "step": 931
+    },
+    {
+      "epoch": 0.2685878962536023,
+      "grad_norm": 3.0566838022283416,
+      "learning_rate": 1e-05,
+      "loss": 0.5114,
+      "step": 932
+    },
+    {
+      "epoch": 0.26887608069164265,
+      "grad_norm": 1.9664727540315903,
+      "learning_rate": 1e-05,
+      "loss": 0.5135,
+      "step": 933
+    },
+    {
+      "epoch": 0.269164265129683,
+      "grad_norm": 1.9116619780341022,
+      "learning_rate": 1e-05,
+      "loss": 0.3532,
+      "step": 934
+    },
+    {
+      "epoch": 0.26945244956772335,
+      "grad_norm": 2.749209246235332,
+      "learning_rate": 1e-05,
+      "loss": 0.5303,
+      "step": 935
+    },
+    {
+      "epoch": 0.2697406340057637,
+      "grad_norm": 1.9429900975170704,
+      "learning_rate": 1e-05,
+      "loss": 0.5016,
+      "step": 936
+    },
+    {
+      "epoch": 0.27002881844380405,
+      "grad_norm": 2.1919804364456263,
+      "learning_rate": 1e-05,
+      "loss": 0.3363,
+      "step": 937
+    },
+    {
+      "epoch": 0.2703170028818444,
+      "grad_norm": 2.9447218226456044,
+      "learning_rate": 1e-05,
+      "loss": 0.542,
+      "step": 938
+    },
+    {
+      "epoch": 0.2706051873198847,
+      "grad_norm": 1.5941809809269842,
+      "learning_rate": 1e-05,
+      "loss": 0.5145,
+      "step": 939
+    },
+    {
+      "epoch": 0.27089337175792505,
+      "grad_norm": 1.7321234967773047,
+      "learning_rate": 1e-05,
+      "loss": 0.357,
+      "step": 940
+    },
+    {
+      "epoch": 0.2711815561959654,
+      "grad_norm": 3.3328482893388967,
+      "learning_rate": 1e-05,
+      "loss": 0.4793,
+      "step": 941
+    },
+    {
+      "epoch": 0.27146974063400575,
+      "grad_norm": 1.5301144291370694,
+      "learning_rate": 1e-05,
+      "loss": 0.512,
+      "step": 942
+    },
+    {
+      "epoch": 0.2717579250720461,
+      "grad_norm": 2.162147900076396,
+      "learning_rate": 1e-05,
+      "loss": 0.3378,
+      "step": 943
+    },
+    {
+      "epoch": 0.27204610951008645,
+      "grad_norm": 2.82530953683193,
+      "learning_rate": 1e-05,
+      "loss": 0.5237,
+      "step": 944
+    },
+    {
+      "epoch": 0.2723342939481268,
+      "grad_norm": 1.8044675645639783,
+      "learning_rate": 1e-05,
+      "loss": 0.5173,
+      "step": 945
+    },
+    {
+      "epoch": 0.27262247838616716,
+      "grad_norm": 5.141915343195708,
+      "learning_rate": 1e-05,
+      "loss": 0.3404,
+      "step": 946
+    },
+    {
+      "epoch": 0.2729106628242075,
+      "grad_norm": 2.1772464536272325,
+      "learning_rate": 1e-05,
+      "loss": 0.502,
+      "step": 947
+    },
+    {
+      "epoch": 0.27319884726224786,
+      "grad_norm": 1.4401664870261714,
+      "learning_rate": 1e-05,
+      "loss": 0.5154,
+      "step": 948
+    },
+    {
+      "epoch": 0.2734870317002882,
+      "grad_norm": 3.0831382894278683,
+      "learning_rate": 1e-05,
+      "loss": 0.3639,
+      "step": 949
+    },
+    {
+      "epoch": 0.2737752161383285,
+      "grad_norm": 2.6477007846939737,
+      "learning_rate": 1e-05,
+      "loss": 0.4888,
+      "step": 950
+    },
+    {
+      "epoch": 0.27406340057636885,
+      "grad_norm": 2.0297943635857134,
+      "learning_rate": 1e-05,
+      "loss": 0.5256,
+      "step": 951
+    },
+    {
+      "epoch": 0.2743515850144092,
+      "grad_norm": 1.8996304186524549,
+      "learning_rate": 1e-05,
+      "loss": 0.3546,
+      "step": 952
+    },
+    {
+      "epoch": 0.27463976945244956,
+      "grad_norm": 2.7855096498536134,
+      "learning_rate": 1e-05,
+      "loss": 0.5055,
+      "step": 953
+    },
+    {
+      "epoch": 0.2749279538904899,
+      "grad_norm": 1.5800546792173475,
+      "learning_rate": 1e-05,
+      "loss": 0.5026,
+      "step": 954
+    },
+    {
+      "epoch": 0.27521613832853026,
+      "grad_norm": 5.7576587032094455,
+      "learning_rate": 1e-05,
+      "loss": 0.3315,
+      "step": 955
+    },
+    {
+      "epoch": 0.2755043227665706,
+      "grad_norm": 4.201893465596969,
+      "learning_rate": 1e-05,
+      "loss": 0.5388,
+      "step": 956
+    },
+    {
+      "epoch": 0.27579250720461096,
+      "grad_norm": 1.4216292133657598,
+      "learning_rate": 1e-05,
+      "loss": 0.5087,
+      "step": 957
+    },
+    {
+      "epoch": 0.2760806916426513,
+      "grad_norm": 2.226321484319974,
+      "learning_rate": 1e-05,
+      "loss": 0.3545,
+      "step": 958
+    },
+    {
+      "epoch": 0.27636887608069166,
+      "grad_norm": 3.313341371873014,
+      "learning_rate": 1e-05,
+      "loss": 0.5253,
+      "step": 959
+    },
+    {
+      "epoch": 0.276657060518732,
+      "grad_norm": 1.6604053305299187,
+      "learning_rate": 1e-05,
+      "loss": 0.5076,
+      "step": 960
+    },
+    {
+      "epoch": 0.2769452449567723,
+      "grad_norm": 3.371233896844695,
+      "learning_rate": 1e-05,
+      "loss": 0.3361,
+      "step": 961
+    },
+    {
+      "epoch": 0.27723342939481266,
+      "grad_norm": 3.2121544076481245,
+      "learning_rate": 1e-05,
+      "loss": 0.5218,
+      "step": 962
+    },
+    {
+      "epoch": 0.277521613832853,
+      "grad_norm": 1.5130474234128468,
+      "learning_rate": 1e-05,
+      "loss": 0.5112,
+      "step": 963
+    },
+    {
+      "epoch": 0.27780979827089336,
+      "grad_norm": 2.5496314917510325,
+      "learning_rate": 1e-05,
+      "loss": 0.3398,
+      "step": 964
+    },
+    {
+      "epoch": 0.2780979827089337,
+      "grad_norm": 3.987921686601577,
+      "learning_rate": 1e-05,
+      "loss": 0.5268,
+      "step": 965
+    },
+    {
+      "epoch": 0.27838616714697406,
+      "grad_norm": 2.2533295714840587,
+      "learning_rate": 1e-05,
+      "loss": 0.5015,
+      "step": 966
+    },
+    {
+      "epoch": 0.2786743515850144,
+      "grad_norm": 1.5513596680091195,
+      "learning_rate": 1e-05,
+      "loss": 0.3512,
+      "step": 967
+    },
+    {
+      "epoch": 0.27896253602305476,
+      "grad_norm": 2.013799560979713,
+      "learning_rate": 1e-05,
+      "loss": 0.5339,
+      "step": 968
+    },
+    {
+      "epoch": 0.2792507204610951,
+      "grad_norm": 1.7163970279679512,
+      "learning_rate": 1e-05,
+      "loss": 0.516,
+      "step": 969
+    },
+    {
+      "epoch": 0.27953890489913547,
+      "grad_norm": 1.740456061909977,
+      "learning_rate": 1e-05,
+      "loss": 0.3458,
+      "step": 970
+    },
+    {
+      "epoch": 0.2798270893371758,
+      "grad_norm": 2.600900511052735,
+      "learning_rate": 1e-05,
+      "loss": 0.5783,
+      "step": 971
+    },
+    {
+      "epoch": 0.2801152737752161,
+      "grad_norm": 1.6952006028519035,
+      "learning_rate": 1e-05,
+      "loss": 0.5113,
+      "step": 972
+    },
+    {
+      "epoch": 0.28040345821325646,
+      "grad_norm": 1.6348837324966576,
+      "learning_rate": 1e-05,
+      "loss": 0.3416,
+      "step": 973
+    },
+    {
+      "epoch": 0.2806916426512968,
+      "grad_norm": 3.203856767932054,
+      "learning_rate": 1e-05,
+      "loss": 0.4982,
+      "step": 974
+    },
+    {
+      "epoch": 0.28097982708933716,
+      "grad_norm": 1.4188826070149956,
+      "learning_rate": 1e-05,
+      "loss": 0.4952,
+      "step": 975
+    },
+    {
+      "epoch": 0.2812680115273775,
+      "grad_norm": 1.5009416983479824,
+      "learning_rate": 1e-05,
+      "loss": 0.3347,
+      "step": 976
+    },
+    {
+      "epoch": 0.28155619596541787,
+      "grad_norm": 2.277957231446711,
+      "learning_rate": 1e-05,
+      "loss": 0.4995,
+      "step": 977
+    },
+    {
+      "epoch": 0.2818443804034582,
+      "grad_norm": 1.3294390252675674,
+      "learning_rate": 1e-05,
+      "loss": 0.5116,
+      "step": 978
+    },
+    {
+      "epoch": 0.28213256484149857,
+      "grad_norm": 1.4958690683527438,
+      "learning_rate": 1e-05,
+      "loss": 0.3253,
+      "step": 979
+    },
+    {
+      "epoch": 0.2824207492795389,
+      "grad_norm": 2.4378365852311936,
+      "learning_rate": 1e-05,
+      "loss": 0.4753,
+      "step": 980
+    },
+    {
+      "epoch": 0.28270893371757927,
+      "grad_norm": 1.6091312154798967,
+      "learning_rate": 1e-05,
+      "loss": 0.5081,
+      "step": 981
+    },
+    {
+      "epoch": 0.2829971181556196,
+      "grad_norm": 2.1393277030619946,
+      "learning_rate": 1e-05,
+      "loss": 0.3604,
+      "step": 982
+    },
+    {
+      "epoch": 0.2832853025936599,
+      "grad_norm": 2.1222333160060027,
+      "learning_rate": 1e-05,
+      "loss": 0.4901,
+      "step": 983
+    },
+    {
+      "epoch": 0.28357348703170027,
+      "grad_norm": 1.4829404798184205,
+      "learning_rate": 1e-05,
+      "loss": 0.5092,
+      "step": 984
+    },
+    {
+      "epoch": 0.2838616714697406,
+      "grad_norm": 1.757665907427271,
+      "learning_rate": 1e-05,
+      "loss": 0.3501,
+      "step": 985
+    },
+    {
+      "epoch": 0.28414985590778097,
+      "grad_norm": 2.106202963340238,
+      "learning_rate": 1e-05,
+      "loss": 0.5245,
+      "step": 986
+    },
+    {
+      "epoch": 0.2844380403458213,
+      "grad_norm": 1.4356411481647915,
+      "learning_rate": 1e-05,
+      "loss": 0.5126,
+      "step": 987
+    },
+    {
+      "epoch": 0.28472622478386167,
+      "grad_norm": 1.8367029706409999,
+      "learning_rate": 1e-05,
+      "loss": 0.3274,
+      "step": 988
+    },
+    {
+      "epoch": 0.285014409221902,
+      "grad_norm": 2.2418570564991716,
+      "learning_rate": 1e-05,
+      "loss": 0.5185,
+      "step": 989
+    },
+    {
+      "epoch": 0.28530259365994237,
+      "grad_norm": 2.498112838150171,
+      "learning_rate": 1e-05,
+      "loss": 0.5071,
+      "step": 990
+    },
+    {
+      "epoch": 0.2855907780979827,
+      "grad_norm": 2.4362164461254325,
+      "learning_rate": 1e-05,
+      "loss": 0.368,
+      "step": 991
+    },
+    {
+      "epoch": 0.2858789625360231,
+      "grad_norm": 2.3850112864480444,
+      "learning_rate": 1e-05,
+      "loss": 0.5489,
+      "step": 992
+    },
+    {
+      "epoch": 0.2861671469740634,
+      "grad_norm": 2.1723865818093855,
+      "learning_rate": 1e-05,
+      "loss": 0.4988,
+      "step": 993
+    },
+    {
+      "epoch": 0.2864553314121037,
+      "grad_norm": 1.5193362742613035,
+      "learning_rate": 1e-05,
+      "loss": 0.349,
+      "step": 994
+    },
+    {
+      "epoch": 0.28674351585014407,
+      "grad_norm": 2.3047313648767527,
+      "learning_rate": 1e-05,
+      "loss": 0.4986,
+      "step": 995
+    },
+    {
+      "epoch": 0.2870317002881844,
+      "grad_norm": 1.849037996593631,
+      "learning_rate": 1e-05,
+      "loss": 0.5007,
+      "step": 996
+    },
+    {
+      "epoch": 0.2873198847262248,
+      "grad_norm": 1.5160434534484937,
+      "learning_rate": 1e-05,
+      "loss": 0.3342,
+      "step": 997
+    },
+    {
+      "epoch": 0.2876080691642651,
+      "grad_norm": 3.656074368240668,
+      "learning_rate": 1e-05,
+      "loss": 0.5135,
+      "step": 998
+    },
+    {
+      "epoch": 0.2878962536023055,
+      "grad_norm": 1.7931777660285688,
+      "learning_rate": 1e-05,
+      "loss": 0.5173,
+      "step": 999
+    },
+    {
+      "epoch": 0.2881844380403458,
+      "grad_norm": 1.7995325426446684,
+      "learning_rate": 1e-05,
+      "loss": 0.3335,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2881844380403458,
+      "eval_websight_new_IoU": 0.017799582332372665,
+      "eval_websight_new_MAE_x": 286.7799987792969,
+      "eval_websight_new_MAE_y": 626.0999755859375,
+      "eval_websight_new_coord_probability": 0.7444506883621216,
+      "eval_websight_new_correct_token_probability": 0.00132176058832556,
+      "eval_websight_new_inside_bbox": 0.019999999552965164,
+      "eval_websight_new_loss": 0.34697380661964417,
+      "eval_websight_new_range_100_probability": 0.08497080951929092,
+      "eval_websight_new_range_20_probability": 0.016971057280898094,
+      "eval_websight_new_range_5_probability": 0.0037731751799583435,
+      "eval_websight_new_runtime": 34.1093,
+      "eval_websight_new_samples_per_second": 1.466,
+      "eval_websight_new_steps_per_second": 0.205,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2881844380403458,
+      "eval_seeclick_IoU": 0.03177700936794281,
+      "eval_seeclick_MAE_x": 473.1000061035156,
+      "eval_seeclick_MAE_y": 420.239990234375,
+      "eval_seeclick_coord_probability": 0.7443803548812866,
+      "eval_seeclick_correct_token_probability": 0.0007717523258179426,
+      "eval_seeclick_inside_bbox": 0.07999999821186066,
+      "eval_seeclick_loss": 0.41178855299949646,
+      "eval_seeclick_range_100_probability": 0.13013127446174622,
+      "eval_seeclick_range_20_probability": 0.050893329083919525,
+      "eval_seeclick_range_5_probability": 0.0074063134379684925,
+      "eval_seeclick_runtime": 42.0879,
+      "eval_seeclick_samples_per_second": 1.188,
+      "eval_seeclick_steps_per_second": 0.166,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2881844380403458,
+      "eval_icons_IoU": 0.0025627599097788334,
+      "eval_icons_MAE_x": 473.5400085449219,
+      "eval_icons_MAE_y": 447.05999755859375,
+      "eval_icons_coord_probability": 0.7488621473312378,
+      "eval_icons_correct_token_probability": 0.0066015468910336494,
+      "eval_icons_inside_bbox": 0.0,
+      "eval_icons_loss": 0.4837815463542938,
+      "eval_icons_range_100_probability": 0.2755989730358124,
+      "eval_icons_range_20_probability": 0.10048653930425644,
+      "eval_icons_range_5_probability": 0.0324726328253746,
+      "eval_icons_runtime": 43.1872,
+      "eval_icons_samples_per_second": 1.158,
+      "eval_icons_steps_per_second": 0.162,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2884726224783862,
+      "grad_norm": 2.338705947652742,
+      "learning_rate": 1e-05,
+      "loss": 0.5499,
+      "step": 1001
+    },
+    {
+      "epoch": 0.2887608069164265,
+      "grad_norm": 3.19564411545996,
+      "learning_rate": 1e-05,
+      "loss": 0.4956,
+      "step": 1002
+    },
+    {
+      "epoch": 0.2890489913544669,
+      "grad_norm": 1.791379287140292,
+      "learning_rate": 1e-05,
+      "loss": 0.3211,
+      "step": 1003
+    },
+    {
+      "epoch": 0.28933717579250723,
+      "grad_norm": 2.542255608805061,
+      "learning_rate": 1e-05,
+      "loss": 0.5143,
+      "step": 1004
+    },
+    {
+      "epoch": 0.2896253602305475,
+      "grad_norm": 1.5656152479121874,
+      "learning_rate": 1e-05,
+      "loss": 0.503,
+      "step": 1005
+    },
+    {
+      "epoch": 0.2899135446685879,
+      "grad_norm": 1.875361162771763,
+      "learning_rate": 1e-05,
+      "loss": 0.3259,
+      "step": 1006
+    },
+    {
+      "epoch": 0.2902017291066282,
+      "grad_norm": 2.5654391299631047,
+      "learning_rate": 1e-05,
+      "loss": 0.4782,
+      "step": 1007
+    },
+    {
+      "epoch": 0.2904899135446686,
+      "grad_norm": 1.687344227505042,
+      "learning_rate": 1e-05,
+      "loss": 0.5132,
+      "step": 1008
+    },
+    {
+      "epoch": 0.2907780979827089,
+      "grad_norm": 1.4118840040804839,
+      "learning_rate": 1e-05,
+      "loss": 0.3425,
+      "step": 1009
+    },
+    {
+      "epoch": 0.2910662824207493,
+      "grad_norm": 3.0684612023857323,
+      "learning_rate": 1e-05,
+      "loss": 0.4977,
+      "step": 1010
+    },
+    {
+      "epoch": 0.29135446685878963,
+      "grad_norm": 2.3222928133076612,
+      "learning_rate": 1e-05,
+      "loss": 0.5029,
+      "step": 1011
+    },
+    {
+      "epoch": 0.29164265129683,
+      "grad_norm": 1.4926787652225055,
+      "learning_rate": 1e-05,
+      "loss": 0.3512,
+      "step": 1012
+    },
+    {
+      "epoch": 0.29193083573487033,
+      "grad_norm": 2.225007072873174,
+      "learning_rate": 1e-05,
+      "loss": 0.4875,
+      "step": 1013
+    },
+    {
+      "epoch": 0.2922190201729107,
+      "grad_norm": 1.3432831467936621,
+      "learning_rate": 1e-05,
+      "loss": 0.5122,
+      "step": 1014
+    },
+    {
+      "epoch": 0.29250720461095103,
+      "grad_norm": 1.4827838645606013,
+      "learning_rate": 1e-05,
+      "loss": 0.3436,
+      "step": 1015
+    },
+    {
+      "epoch": 0.29279538904899133,
+      "grad_norm": 2.3026347865131322,
+      "learning_rate": 1e-05,
+      "loss": 0.553,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2930835734870317,
+      "grad_norm": 2.0832470548307267,
+      "learning_rate": 1e-05,
+      "loss": 0.5061,
+      "step": 1017
+    },
+    {
+      "epoch": 0.29337175792507203,
+      "grad_norm": 1.5728771509894899,
+      "learning_rate": 1e-05,
+      "loss": 0.3495,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2936599423631124,
+      "grad_norm": 2.7257448830204827,
+      "learning_rate": 1e-05,
+      "loss": 0.5238,
+      "step": 1019
+    },
+    {
+      "epoch": 0.29394812680115273,
+      "grad_norm": 2.2189118689950584,
+      "learning_rate": 1e-05,
+      "loss": 0.5084,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2942363112391931,
+      "grad_norm": 2.915582284166355,
+      "learning_rate": 1e-05,
+      "loss": 0.3676,
+      "step": 1021
+    },
+    {
+      "epoch": 0.29452449567723343,
+      "grad_norm": 2.502896874980711,
+      "learning_rate": 1e-05,
+      "loss": 0.4961,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2948126801152738,
+      "grad_norm": 2.1008151251217253,
+      "learning_rate": 1e-05,
+      "loss": 0.5092,
+      "step": 1023
+    },
+    {
+      "epoch": 0.29510086455331414,
+      "grad_norm": 1.710263389551021,
+      "learning_rate": 1e-05,
+      "loss": 0.3457,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2953890489913545,
+      "grad_norm": 3.556721941846693,
+      "learning_rate": 1e-05,
+      "loss": 0.4959,
+      "step": 1025
+    },
+    {
+      "epoch": 0.29567723342939484,
+      "grad_norm": 1.4887559265779642,
+      "learning_rate": 1e-05,
+      "loss": 0.5115,
+      "step": 1026
+    },
+    {
+      "epoch": 0.29596541786743513,
+      "grad_norm": 1.8539460109675998,
+      "learning_rate": 1e-05,
+      "loss": 0.3516,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2962536023054755,
+      "grad_norm": 2.3425154740643555,
+      "learning_rate": 1e-05,
+      "loss": 0.4887,
+      "step": 1028
+    },
+    {
+      "epoch": 0.29654178674351583,
+      "grad_norm": 1.1950489518835659,
+      "learning_rate": 1e-05,
+      "loss": 0.4877,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2968299711815562,
+      "grad_norm": 1.325524273952331,
+      "learning_rate": 1e-05,
+      "loss": 0.3462,
+      "step": 1030
+    },
+    {
+      "epoch": 0.29711815561959654,
+      "grad_norm": 3.6218677604895317,
+      "learning_rate": 1e-05,
+      "loss": 0.5156,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2974063400576369,
+      "grad_norm": 1.3493498512154727,
+      "learning_rate": 1e-05,
+      "loss": 0.495,
+      "step": 1032
+    },
+    {
+      "epoch": 0.29769452449567724,
+      "grad_norm": 2.143130694787277,
+      "learning_rate": 1e-05,
+      "loss": 0.3186,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2979827089337176,
+      "grad_norm": 2.1363154839009724,
+      "learning_rate": 1e-05,
+      "loss": 0.5064,
+      "step": 1034
+    },
+    {
+      "epoch": 0.29827089337175794,
+      "grad_norm": 1.699553612367511,
+      "learning_rate": 1e-05,
+      "loss": 0.4945,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2985590778097983,
+      "grad_norm": 2.0868848043456336,
+      "learning_rate": 1e-05,
+      "loss": 0.3304,
+      "step": 1036
+    },
+    {
+      "epoch": 0.29884726224783864,
+      "grad_norm": 2.2575186930196534,
+      "learning_rate": 1e-05,
+      "loss": 0.5199,
+      "step": 1037
+    },
+    {
+      "epoch": 0.29913544668587894,
+      "grad_norm": 1.7017161362946382,
+      "learning_rate": 1e-05,
+      "loss": 0.5047,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2994236311239193,
+      "grad_norm": 2.37203629858315,
+      "learning_rate": 1e-05,
+      "loss": 0.3492,
+      "step": 1039
+    },
+    {
+      "epoch": 0.29971181556195964,
+      "grad_norm": 2.665105073295838,
+      "learning_rate": 1e-05,
+      "loss": 0.534,
+      "step": 1040
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.048127299268337,
+      "learning_rate": 1e-05,
+      "loss": 0.4985,
+      "step": 1041
+    },
+    {
+      "epoch": 0.30028818443804034,
+      "grad_norm": 2.546255498986328,
+      "learning_rate": 1e-05,
+      "loss": 0.3369,
+      "step": 1042
+    },
+    {
+      "epoch": 0.3005763688760807,
+      "grad_norm": 3.327265663756822,
+      "learning_rate": 1e-05,
+      "loss": 0.4978,
+      "step": 1043
+    },
+    {
+      "epoch": 0.30086455331412104,
+      "grad_norm": 1.5376186167854946,
+      "learning_rate": 1e-05,
+      "loss": 0.5036,
+      "step": 1044
+    },
+    {
+      "epoch": 0.3011527377521614,
+      "grad_norm": 2.326408967350416,
+      "learning_rate": 1e-05,
+      "loss": 0.3184,
+      "step": 1045
+    },
+    {
+      "epoch": 0.30144092219020174,
+      "grad_norm": 2.327117884034042,
+      "learning_rate": 1e-05,
+      "loss": 0.4875,
+      "step": 1046
+    },
+    {
+      "epoch": 0.3017291066282421,
+      "grad_norm": 1.8993417820337035,
+      "learning_rate": 1e-05,
+      "loss": 0.5083,
+      "step": 1047
+    },
+    {
+      "epoch": 0.30201729106628245,
+      "grad_norm": 1.986867371144744,
+      "learning_rate": 1e-05,
+      "loss": 0.3337,
+      "step": 1048
+    },
+    {
+      "epoch": 0.30230547550432274,
+      "grad_norm": 2.468355040160504,
+      "learning_rate": 1e-05,
+      "loss": 0.5177,
+      "step": 1049
+    },
+    {
+      "epoch": 0.3025936599423631,
+      "grad_norm": 1.3565315890460994,
+      "learning_rate": 1e-05,
+      "loss": 0.4926,
+      "step": 1050
+    },
+    {
+      "epoch": 0.30288184438040344,
+      "grad_norm": 3.4523937722407285,
+      "learning_rate": 1e-05,
+      "loss": 0.3439,
+      "step": 1051
+    },
+    {
+      "epoch": 0.3031700288184438,
+      "grad_norm": 2.843193735309629,
+      "learning_rate": 1e-05,
+      "loss": 0.4768,
+      "step": 1052
+    },
+    {
+      "epoch": 0.30345821325648414,
+      "grad_norm": 1.2883347670704832,
+      "learning_rate": 1e-05,
+      "loss": 0.5009,
+      "step": 1053
+    },
+    {
+      "epoch": 0.3037463976945245,
+      "grad_norm": 1.5757669538464156,
+      "learning_rate": 1e-05,
+      "loss": 0.3412,
+      "step": 1054
+    },
+    {
+      "epoch": 0.30403458213256485,
+      "grad_norm": 2.2483605576330694,
+      "learning_rate": 1e-05,
+      "loss": 0.4609,
+      "step": 1055
+    },
+    {
+      "epoch": 0.3043227665706052,
+      "grad_norm": 1.737514661382086,
+      "learning_rate": 1e-05,
+      "loss": 0.4975,
+      "step": 1056
+    },
+    {
+      "epoch": 0.30461095100864555,
+      "grad_norm": 2.941418614338563,
+      "learning_rate": 1e-05,
+      "loss": 0.3121,
+      "step": 1057
+    },
+    {
+      "epoch": 0.3048991354466859,
+      "grad_norm": 3.4503887886225684,
+      "learning_rate": 1e-05,
+      "loss": 0.5021,
+      "step": 1058
+    },
+    {
+      "epoch": 0.30518731988472625,
+      "grad_norm": 2.3020992021384186,
+      "learning_rate": 1e-05,
+      "loss": 0.4961,
+      "step": 1059
+    },
+    {
+      "epoch": 0.30547550432276654,
+      "grad_norm": 1.606250972485056,
+      "learning_rate": 1e-05,
+      "loss": 0.3234,
+      "step": 1060
+    },
+    {
+      "epoch": 0.3057636887608069,
+      "grad_norm": 3.204938244559325,
+      "learning_rate": 1e-05,
+      "loss": 0.5312,
+      "step": 1061
+    },
+    {
+      "epoch": 0.30605187319884725,
+      "grad_norm": 1.7501247432174412,
+      "learning_rate": 1e-05,
+      "loss": 0.4954,
+      "step": 1062
+    },
+    {
+      "epoch": 0.3063400576368876,
+      "grad_norm": 2.462337105028575,
+      "learning_rate": 1e-05,
+      "loss": 0.3362,
+      "step": 1063
+    },
+    {
+      "epoch": 0.30662824207492795,
+      "grad_norm": 2.9218099597484413,
+      "learning_rate": 1e-05,
+      "loss": 0.4813,
+      "step": 1064
+    },
+    {
+      "epoch": 0.3069164265129683,
+      "grad_norm": 1.3206159124427526,
+      "learning_rate": 1e-05,
+      "loss": 0.4995,
+      "step": 1065
+    },
+    {
+      "epoch": 0.30720461095100865,
+      "grad_norm": 2.3711038543775724,
+      "learning_rate": 1e-05,
+      "loss": 0.3198,
+      "step": 1066
+    },
+    {
+      "epoch": 0.307492795389049,
+      "grad_norm": 2.869326574113226,
+      "learning_rate": 1e-05,
+      "loss": 0.5007,
+      "step": 1067
+    },
+    {
+      "epoch": 0.30778097982708935,
+      "grad_norm": 1.6101459790709531,
+      "learning_rate": 1e-05,
+      "loss": 0.4893,
+      "step": 1068
+    },
+    {
+      "epoch": 0.3080691642651297,
+      "grad_norm": 2.0601913564631342,
+      "learning_rate": 1e-05,
+      "loss": 0.3208,
+      "step": 1069
+    },
+    {
+      "epoch": 0.30835734870317005,
+      "grad_norm": 2.0557299549071004,
+      "learning_rate": 1e-05,
+      "loss": 0.4669,
+      "step": 1070
+    },
+    {
+      "epoch": 0.30864553314121035,
+      "grad_norm": 1.6732954854682147,
+      "learning_rate": 1e-05,
+      "loss": 0.486,
+      "step": 1071
+    },
+    {
+      "epoch": 0.3089337175792507,
+      "grad_norm": 1.6402686890449796,
+      "learning_rate": 1e-05,
+      "loss": 0.3418,
+      "step": 1072
+    },
+    {
+      "epoch": 0.30922190201729105,
+      "grad_norm": 2.569262783943642,
+      "learning_rate": 1e-05,
+      "loss": 0.5016,
+      "step": 1073
+    },
+    {
+      "epoch": 0.3095100864553314,
+      "grad_norm": 1.3435841034096272,
+      "learning_rate": 1e-05,
+      "loss": 0.4929,
+      "step": 1074
+    },
+    {
+      "epoch": 0.30979827089337175,
+      "grad_norm": 1.9873277200581418,
+      "learning_rate": 1e-05,
+      "loss": 0.335,
+      "step": 1075
+    },
+    {
+      "epoch": 0.3100864553314121,
+      "grad_norm": 3.0261184720776995,
+      "learning_rate": 1e-05,
+      "loss": 0.4929,
+      "step": 1076
+    },
+    {
+      "epoch": 0.31037463976945245,
+      "grad_norm": 1.4833336101948864,
+      "learning_rate": 1e-05,
+      "loss": 0.4887,
+      "step": 1077
+    },
+    {
+      "epoch": 0.3106628242074928,
+      "grad_norm": 1.6388746026047194,
+      "learning_rate": 1e-05,
+      "loss": 0.3443,
+      "step": 1078
+    },
+    {
+      "epoch": 0.31095100864553316,
+      "grad_norm": 2.17100085821424,
+      "learning_rate": 1e-05,
+      "loss": 0.5004,
+      "step": 1079
+    },
+    {
+      "epoch": 0.3112391930835735,
+      "grad_norm": 1.5009698236618427,
+      "learning_rate": 1e-05,
+      "loss": 0.5029,
+      "step": 1080
+    },
+    {
+      "epoch": 0.31152737752161386,
+      "grad_norm": 1.3224191389643551,
+      "learning_rate": 1e-05,
+      "loss": 0.3305,
+      "step": 1081
+    },
+    {
+      "epoch": 0.31181556195965415,
+      "grad_norm": 2.363374841564539,
+      "learning_rate": 1e-05,
+      "loss": 0.4831,
+      "step": 1082
+    },
+    {
+      "epoch": 0.3121037463976945,
+      "grad_norm": 1.835170709153892,
+      "learning_rate": 1e-05,
+      "loss": 0.5032,
+      "step": 1083
+    },
+    {
+      "epoch": 0.31239193083573485,
+      "grad_norm": 1.7927680679006808,
+      "learning_rate": 1e-05,
+      "loss": 0.3288,
+      "step": 1084
+    },
+    {
+      "epoch": 0.3126801152737752,
+      "grad_norm": 2.442094038750008,
+      "learning_rate": 1e-05,
+      "loss": 0.4659,
+      "step": 1085
+    },
+    {
+      "epoch": 0.31296829971181556,
+      "grad_norm": 1.2804687823000351,
+      "learning_rate": 1e-05,
+      "loss": 0.4878,
+      "step": 1086
+    },
+    {
+      "epoch": 0.3132564841498559,
+      "grad_norm": 1.6619241894603738,
+      "learning_rate": 1e-05,
+      "loss": 0.3325,
+      "step": 1087
+    },
+    {
+      "epoch": 0.31354466858789626,
+      "grad_norm": 2.6481437855016523,
+      "learning_rate": 1e-05,
+      "loss": 0.5162,
+      "step": 1088
+    },
+    {
+      "epoch": 0.3138328530259366,
+      "grad_norm": 2.0292716630329117,
+      "learning_rate": 1e-05,
+      "loss": 0.5062,
+      "step": 1089
+    },
+    {
+      "epoch": 0.31412103746397696,
+      "grad_norm": 1.6257203678204113,
+      "learning_rate": 1e-05,
+      "loss": 0.3118,
+      "step": 1090
+    },
+    {
+      "epoch": 0.3144092219020173,
+      "grad_norm": 2.616220849824178,
+      "learning_rate": 1e-05,
+      "loss": 0.508,
+      "step": 1091
+    },
+    {
+      "epoch": 0.31469740634005766,
+      "grad_norm": 2.431223788935733,
+      "learning_rate": 1e-05,
+      "loss": 0.5045,
+      "step": 1092
+    },
+    {
+      "epoch": 0.31498559077809796,
+      "grad_norm": 1.6089488492295851,
+      "learning_rate": 1e-05,
+      "loss": 0.3272,
+      "step": 1093
+    },
+    {
+      "epoch": 0.3152737752161383,
+      "grad_norm": 2.5294353517048087,
+      "learning_rate": 1e-05,
+      "loss": 0.4893,
+      "step": 1094
+    },
+    {
+      "epoch": 0.31556195965417866,
+      "grad_norm": 1.8357982310621475,
+      "learning_rate": 1e-05,
+      "loss": 0.5081,
+      "step": 1095
+    },
+    {
+      "epoch": 0.315850144092219,
+      "grad_norm": 1.4577791992451414,
+      "learning_rate": 1e-05,
+      "loss": 0.3463,
+      "step": 1096
+    },
+    {
+      "epoch": 0.31613832853025936,
+      "grad_norm": 2.384433196506135,
+      "learning_rate": 1e-05,
+      "loss": 0.5092,
+      "step": 1097
+    },
+    {
+      "epoch": 0.3164265129682997,
+      "grad_norm": 2.123313500223575,
+      "learning_rate": 1e-05,
+      "loss": 0.4831,
+      "step": 1098
+    },
+    {
+      "epoch": 0.31671469740634006,
+      "grad_norm": 1.6293099403350886,
+      "learning_rate": 1e-05,
+      "loss": 0.351,
+      "step": 1099
+    },
+    {
+      "epoch": 0.3170028818443804,
+      "grad_norm": 2.472562235724969,
+      "learning_rate": 1e-05,
+      "loss": 0.5227,
+      "step": 1100
+    },
+    {
+      "epoch": 0.31729106628242076,
+      "grad_norm": 1.6770690157243782,
+      "learning_rate": 1e-05,
+      "loss": 0.4898,
+      "step": 1101
+    },
+    {
+      "epoch": 0.3175792507204611,
+      "grad_norm": 2.3972959029136014,
+      "learning_rate": 1e-05,
+      "loss": 0.317,
+      "step": 1102
+    },
+    {
+      "epoch": 0.31786743515850147,
+      "grad_norm": 2.664672422405968,
+      "learning_rate": 1e-05,
+      "loss": 0.4909,
+      "step": 1103
+    },
+    {
+      "epoch": 0.31815561959654176,
+      "grad_norm": 1.5753150829159668,
+      "learning_rate": 1e-05,
+      "loss": 0.4941,
+      "step": 1104
+    },
+    {
+      "epoch": 0.3184438040345821,
+      "grad_norm": 1.933326489331015,
+      "learning_rate": 1e-05,
+      "loss": 0.3533,
+      "step": 1105
+    },
+    {
+      "epoch": 0.31873198847262246,
+      "grad_norm": 2.420775407156173,
+      "learning_rate": 1e-05,
+      "loss": 0.4497,
+      "step": 1106
+    },
+    {
+      "epoch": 0.3190201729106628,
+      "grad_norm": 1.7918076644983714,
+      "learning_rate": 1e-05,
+      "loss": 0.4898,
+      "step": 1107
+    },
+    {
+      "epoch": 0.31930835734870316,
+      "grad_norm": 1.2780275917344892,
+      "learning_rate": 1e-05,
+      "loss": 0.3087,
+      "step": 1108
+    },
+    {
+      "epoch": 0.3195965417867435,
+      "grad_norm": 3.4097096363935027,
+      "learning_rate": 1e-05,
+      "loss": 0.4985,
+      "step": 1109
+    },
+    {
+      "epoch": 0.31988472622478387,
+      "grad_norm": 1.5942865708242027,
+      "learning_rate": 1e-05,
+      "loss": 0.4861,
+      "step": 1110
+    },
+    {
+      "epoch": 0.3201729106628242,
+      "grad_norm": 2.932396606270518,
+      "learning_rate": 1e-05,
+      "loss": 0.3253,
+      "step": 1111
+    },
+    {
+      "epoch": 0.32046109510086457,
+      "grad_norm": 1.976735231620706,
+      "learning_rate": 1e-05,
+      "loss": 0.4609,
+      "step": 1112
+    },
+    {
+      "epoch": 0.3207492795389049,
+      "grad_norm": 2.756056048048722,
+      "learning_rate": 1e-05,
+      "loss": 0.4999,
+      "step": 1113
+    },
+    {
+      "epoch": 0.32103746397694527,
+      "grad_norm": 1.7328647439434661,
+      "learning_rate": 1e-05,
+      "loss": 0.3196,
+      "step": 1114
+    },
+    {
+      "epoch": 0.32132564841498557,
+      "grad_norm": 2.262601047104278,
+      "learning_rate": 1e-05,
+      "loss": 0.5254,
+      "step": 1115
+    },
+    {
+      "epoch": 0.3216138328530259,
+      "grad_norm": 1.6506093692491353,
+      "learning_rate": 1e-05,
+      "loss": 0.4981,
+      "step": 1116
+    },
+    {
+      "epoch": 0.32190201729106627,
+      "grad_norm": 3.2703496210059018,
+      "learning_rate": 1e-05,
+      "loss": 0.3375,
+      "step": 1117
+    },
+    {
+      "epoch": 0.3221902017291066,
+      "grad_norm": 4.868071164616914,
+      "learning_rate": 1e-05,
+      "loss": 0.5018,
+      "step": 1118
+    },
+    {
+      "epoch": 0.32247838616714697,
+      "grad_norm": 1.6085911689626369,
+      "learning_rate": 1e-05,
+      "loss": 0.4998,
+      "step": 1119
+    },
+    {
+      "epoch": 0.3227665706051873,
+      "grad_norm": 3.4486472045757277,
+      "learning_rate": 1e-05,
+      "loss": 0.3183,
+      "step": 1120
+    },
+    {
+      "epoch": 0.32305475504322767,
+      "grad_norm": 2.6922668896039514,
+      "learning_rate": 1e-05,
+      "loss": 0.498,
+      "step": 1121
+    },
+    {
+      "epoch": 0.323342939481268,
+      "grad_norm": 1.6744870417845208,
+      "learning_rate": 1e-05,
+      "loss": 0.4932,
+      "step": 1122
+    },
+    {
+      "epoch": 0.32363112391930837,
+      "grad_norm": 2.323000645504488,
+      "learning_rate": 1e-05,
+      "loss": 0.3225,
+      "step": 1123
+    },
+    {
+      "epoch": 0.3239193083573487,
+      "grad_norm": 3.6972537266210423,
+      "learning_rate": 1e-05,
+      "loss": 0.4985,
+      "step": 1124
+    },
+    {
+      "epoch": 0.3242074927953891,
+      "grad_norm": 3.1444825583795826,
+      "learning_rate": 1e-05,
+      "loss": 0.4948,
+      "step": 1125
+    },
+    {
+      "epoch": 0.32449567723342937,
+      "grad_norm": 2.5453979516419074,
+      "learning_rate": 1e-05,
+      "loss": 0.3321,
+      "step": 1126
+    },
+    {
+      "epoch": 0.3247838616714697,
+      "grad_norm": 2.2607852092163196,
+      "learning_rate": 1e-05,
+      "loss": 0.4742,
+      "step": 1127
+    },
+    {
+      "epoch": 0.32507204610951007,
+      "grad_norm": 2.1275514433154257,
+      "learning_rate": 1e-05,
+      "loss": 0.4905,
+      "step": 1128
+    },
+    {
+      "epoch": 0.3253602305475504,
+      "grad_norm": 1.637432413303326,
+      "learning_rate": 1e-05,
+      "loss": 0.3423,
+      "step": 1129
+    },
+    {
+      "epoch": 0.3256484149855908,
+      "grad_norm": 3.034997933911647,
+      "learning_rate": 1e-05,
+      "loss": 0.5107,
+      "step": 1130
+    },
+    {
+      "epoch": 0.3259365994236311,
+      "grad_norm": 1.4552193149878974,
+      "learning_rate": 1e-05,
+      "loss": 0.4914,
+      "step": 1131
+    },
+    {
+      "epoch": 0.3262247838616715,
+      "grad_norm": 4.1907660132530005,
+      "learning_rate": 1e-05,
+      "loss": 0.3091,
+      "step": 1132
+    },
+    {
+      "epoch": 0.3265129682997118,
+      "grad_norm": 2.008765562856227,
+      "learning_rate": 1e-05,
+      "loss": 0.5149,
+      "step": 1133
+    },
+    {
+      "epoch": 0.3268011527377522,
+      "grad_norm": 2.1363305914665207,
+      "learning_rate": 1e-05,
+      "loss": 0.4963,
+      "step": 1134
+    },
+    {
+      "epoch": 0.3270893371757925,
+      "grad_norm": 1.9298203593679761,
+      "learning_rate": 1e-05,
+      "loss": 0.3166,
+      "step": 1135
+    },
+    {
+      "epoch": 0.3273775216138329,
+      "grad_norm": 2.430956307266632,
+      "learning_rate": 1e-05,
+      "loss": 0.464,
+      "step": 1136
+    },
+    {
+      "epoch": 0.3276657060518732,
+      "grad_norm": 2.2524622025954537,
+      "learning_rate": 1e-05,
+      "loss": 0.4901,
+      "step": 1137
+    },
+    {
+      "epoch": 0.3279538904899135,
+      "grad_norm": 2.688017518403025,
+      "learning_rate": 1e-05,
+      "loss": 0.3251,
+      "step": 1138
+    },
+    {
+      "epoch": 0.3282420749279539,
+      "grad_norm": 2.0325657283588567,
+      "learning_rate": 1e-05,
+      "loss": 0.5062,
+      "step": 1139
+    },
+    {
+      "epoch": 0.3285302593659942,
+      "grad_norm": 1.7806061781593912,
+      "learning_rate": 1e-05,
+      "loss": 0.4768,
+      "step": 1140
+    },
+    {
+      "epoch": 0.3288184438040346,
+      "grad_norm": 2.156722268754788,
+      "learning_rate": 1e-05,
+      "loss": 0.3264,
+      "step": 1141
+    },
+    {
+      "epoch": 0.32910662824207493,
+      "grad_norm": 2.7039124237656016,
+      "learning_rate": 1e-05,
+      "loss": 0.5116,
+      "step": 1142
+    },
+    {
+      "epoch": 0.3293948126801153,
+      "grad_norm": 1.846803783065928,
+      "learning_rate": 1e-05,
+      "loss": 0.4927,
+      "step": 1143
+    },
+    {
+      "epoch": 0.32968299711815563,
+      "grad_norm": 1.9812623456518241,
+      "learning_rate": 1e-05,
+      "loss": 0.3019,
+      "step": 1144
+    },
+    {
+      "epoch": 0.329971181556196,
+      "grad_norm": 2.6877800273170434,
+      "learning_rate": 1e-05,
+      "loss": 0.5018,
+      "step": 1145
+    },
+    {
+      "epoch": 0.33025936599423633,
+      "grad_norm": 1.8582044690750283,
+      "learning_rate": 1e-05,
+      "loss": 0.503,
+      "step": 1146
+    },
+    {
+      "epoch": 0.3305475504322767,
+      "grad_norm": 2.5289453697355757,
+      "learning_rate": 1e-05,
+      "loss": 0.3375,
+      "step": 1147
+    },
+    {
+      "epoch": 0.330835734870317,
+      "grad_norm": 3.94007933818287,
+      "learning_rate": 1e-05,
+      "loss": 0.4854,
+      "step": 1148
+    },
+    {
+      "epoch": 0.33112391930835733,
+      "grad_norm": 2.3006899651143424,
+      "learning_rate": 1e-05,
+      "loss": 0.4887,
+      "step": 1149
+    },
+    {
+      "epoch": 0.3314121037463977,
+      "grad_norm": 1.9633104945925637,
+      "learning_rate": 1e-05,
+      "loss": 0.334,
+      "step": 1150
+    },
+    {
+      "epoch": 0.33170028818443803,
+      "grad_norm": 2.523047665002316,
+      "learning_rate": 1e-05,
+      "loss": 0.5205,
+      "step": 1151
+    },
+    {
+      "epoch": 0.3319884726224784,
+      "grad_norm": 1.73966685174154,
+      "learning_rate": 1e-05,
+      "loss": 0.4991,
+      "step": 1152
+    },
+    {
+      "epoch": 0.33227665706051873,
+      "grad_norm": 1.8066746398505633,
+      "learning_rate": 1e-05,
+      "loss": 0.3437,
+      "step": 1153
+    },
+    {
+      "epoch": 0.3325648414985591,
+      "grad_norm": 2.283090155831804,
+      "learning_rate": 1e-05,
+      "loss": 0.5052,
+      "step": 1154
+    },
+    {
+      "epoch": 0.33285302593659943,
+      "grad_norm": 1.5571410132639631,
+      "learning_rate": 1e-05,
+      "loss": 0.4896,
+      "step": 1155
+    },
+    {
+      "epoch": 0.3331412103746398,
+      "grad_norm": 1.3363582344794764,
+      "learning_rate": 1e-05,
+      "loss": 0.3343,
+      "step": 1156
+    },
+    {
+      "epoch": 0.33342939481268014,
+      "grad_norm": 2.992721080077841,
+      "learning_rate": 1e-05,
+      "loss": 0.4711,
+      "step": 1157
+    },
+    {
+      "epoch": 0.3337175792507205,
+      "grad_norm": 1.816968349529476,
+      "learning_rate": 1e-05,
+      "loss": 0.4921,
+      "step": 1158
+    },
+    {
+      "epoch": 0.3340057636887608,
+      "grad_norm": 1.6857815219001941,
+      "learning_rate": 1e-05,
+      "loss": 0.3426,
+      "step": 1159
+    },
+    {
+      "epoch": 0.33429394812680113,
+      "grad_norm": 3.0959603364657204,
+      "learning_rate": 1e-05,
+      "loss": 0.513,
+      "step": 1160
+    },
+    {
+      "epoch": 0.3345821325648415,
+      "grad_norm": 1.4472170523508918,
+      "learning_rate": 1e-05,
+      "loss": 0.4973,
+      "step": 1161
+    },
+    {
+      "epoch": 0.33487031700288183,
+      "grad_norm": 2.283479435772585,
+      "learning_rate": 1e-05,
+      "loss": 0.3309,
+      "step": 1162
+    },
+    {
+      "epoch": 0.3351585014409222,
+      "grad_norm": 2.50287615214733,
+      "learning_rate": 1e-05,
+      "loss": 0.493,
+      "step": 1163
+    },
+    {
+      "epoch": 0.33544668587896254,
+      "grad_norm": 2.1825900702502294,
+      "learning_rate": 1e-05,
+      "loss": 0.4941,
+      "step": 1164
+    },
+    {
+      "epoch": 0.3357348703170029,
+      "grad_norm": 1.977525837420698,
+      "learning_rate": 1e-05,
+      "loss": 0.3375,
+      "step": 1165
+    },
+    {
+      "epoch": 0.33602305475504324,
+      "grad_norm": 2.6235561088202357,
+      "learning_rate": 1e-05,
+      "loss": 0.4653,
+      "step": 1166
+    },
+    {
+      "epoch": 0.3363112391930836,
+      "grad_norm": 1.8806137088648773,
+      "learning_rate": 1e-05,
+      "loss": 0.4834,
+      "step": 1167
+    },
+    {
+      "epoch": 0.33659942363112394,
+      "grad_norm": 2.076944516350051,
+      "learning_rate": 1e-05,
+      "loss": 0.3133,
+      "step": 1168
+    },
+    {
+      "epoch": 0.3368876080691643,
+      "grad_norm": 2.830054702700364,
+      "learning_rate": 1e-05,
+      "loss": 0.5036,
+      "step": 1169
+    },
+    {
+      "epoch": 0.3371757925072046,
+      "grad_norm": 1.8246097676492032,
+      "learning_rate": 1e-05,
+      "loss": 0.4886,
+      "step": 1170
+    },
+    {
+      "epoch": 0.33746397694524494,
+      "grad_norm": 1.9362712545579375,
+      "learning_rate": 1e-05,
+      "loss": 0.3026,
+      "step": 1171
+    },
+    {
+      "epoch": 0.3377521613832853,
+      "grad_norm": 2.4855666695575493,
+      "learning_rate": 1e-05,
+      "loss": 0.4988,
+      "step": 1172
+    },
+    {
+      "epoch": 0.33804034582132564,
+      "grad_norm": 1.6627595920456808,
+      "learning_rate": 1e-05,
+      "loss": 0.492,
+      "step": 1173
+    },
+    {
+      "epoch": 0.338328530259366,
+      "grad_norm": 1.9276841704652814,
+      "learning_rate": 1e-05,
+      "loss": 0.3227,
+      "step": 1174
+    },
+    {
+      "epoch": 0.33861671469740634,
+      "grad_norm": 2.576002040137353,
+      "learning_rate": 1e-05,
+      "loss": 0.5259,
+      "step": 1175
+    },
+    {
+      "epoch": 0.3389048991354467,
+      "grad_norm": 1.9066436871077268,
+      "learning_rate": 1e-05,
+      "loss": 0.4944,
+      "step": 1176
+    },
+    {
+      "epoch": 0.33919308357348704,
+      "grad_norm": 1.5441205950825012,
+      "learning_rate": 1e-05,
+      "loss": 0.337,
+      "step": 1177
+    },
+    {
+      "epoch": 0.3394812680115274,
+      "grad_norm": 3.2639640612206944,
+      "learning_rate": 1e-05,
+      "loss": 0.4634,
+      "step": 1178
+    },
+    {
+      "epoch": 0.33976945244956774,
+      "grad_norm": 2.258471482981334,
+      "learning_rate": 1e-05,
+      "loss": 0.5032,
+      "step": 1179
+    },
+    {
+      "epoch": 0.3400576368876081,
+      "grad_norm": 2.008940650731972,
+      "learning_rate": 1e-05,
+      "loss": 0.3102,
+      "step": 1180
+    },
+    {
+      "epoch": 0.3403458213256484,
+      "grad_norm": 3.0963746293113514,
+      "learning_rate": 1e-05,
+      "loss": 0.5367,
+      "step": 1181
+    },
+    {
+      "epoch": 0.34063400576368874,
+      "grad_norm": 1.915204271376301,
+      "learning_rate": 1e-05,
+      "loss": 0.4929,
+      "step": 1182
+    },
+    {
+      "epoch": 0.3409221902017291,
+      "grad_norm": 2.286096240621281,
+      "learning_rate": 1e-05,
+      "loss": 0.3084,
+      "step": 1183
+    },
+    {
+      "epoch": 0.34121037463976944,
+      "grad_norm": 2.6863781439408223,
+      "learning_rate": 1e-05,
+      "loss": 0.4711,
+      "step": 1184
+    },
+    {
+      "epoch": 0.3414985590778098,
+      "grad_norm": 1.8449861890807961,
+      "learning_rate": 1e-05,
+      "loss": 0.4948,
+      "step": 1185
+    },
+    {
+      "epoch": 0.34178674351585014,
+      "grad_norm": 2.7888935869922826,
+      "learning_rate": 1e-05,
+      "loss": 0.3196,
+      "step": 1186
+    },
+    {
+      "epoch": 0.3420749279538905,
+      "grad_norm": 2.709673365782931,
+      "learning_rate": 1e-05,
+      "loss": 0.5203,
+      "step": 1187
+    },
+    {
+      "epoch": 0.34236311239193085,
+      "grad_norm": 1.7317494751780376,
+      "learning_rate": 1e-05,
+      "loss": 0.4959,
+      "step": 1188
+    },
+    {
+      "epoch": 0.3426512968299712,
+      "grad_norm": 2.1063969485892895,
+      "learning_rate": 1e-05,
+      "loss": 0.3123,
+      "step": 1189
+    },
+    {
+      "epoch": 0.34293948126801155,
+      "grad_norm": 2.2464240545060257,
+      "learning_rate": 1e-05,
+      "loss": 0.5109,
+      "step": 1190
+    },
+    {
+      "epoch": 0.3432276657060519,
+      "grad_norm": 2.4253077947170296,
+      "learning_rate": 1e-05,
+      "loss": 0.4969,
+      "step": 1191
+    },
+    {
+      "epoch": 0.3435158501440922,
+      "grad_norm": 2.4332739078321004,
+      "learning_rate": 1e-05,
+      "loss": 0.334,
+      "step": 1192
+    },
+    {
+      "epoch": 0.34380403458213254,
+      "grad_norm": 3.482343282441452,
+      "learning_rate": 1e-05,
+      "loss": 0.4952,
+      "step": 1193
+    },
+    {
+      "epoch": 0.3440922190201729,
+      "grad_norm": 2.38694438541596,
+      "learning_rate": 1e-05,
+      "loss": 0.485,
+      "step": 1194
+    },
+    {
+      "epoch": 0.34438040345821325,
+      "grad_norm": 2.410215883752191,
+      "learning_rate": 1e-05,
+      "loss": 0.3394,
+      "step": 1195
+    },
+    {
+      "epoch": 0.3446685878962536,
+      "grad_norm": 2.3249140368948633,
+      "learning_rate": 1e-05,
+      "loss": 0.5061,
+      "step": 1196
+    },
+    {
+      "epoch": 0.34495677233429395,
+      "grad_norm": 1.8467063337590413,
+      "learning_rate": 1e-05,
+      "loss": 0.4908,
+      "step": 1197
+    },
+    {
+      "epoch": 0.3452449567723343,
+      "grad_norm": 1.762787342750634,
+      "learning_rate": 1e-05,
+      "loss": 0.3195,
+      "step": 1198
+    },
+    {
+      "epoch": 0.34553314121037465,
+      "grad_norm": 2.3874762172329436,
+      "learning_rate": 1e-05,
+      "loss": 0.4993,
+      "step": 1199
+    },
+    {
+      "epoch": 0.345821325648415,
+      "grad_norm": 2.8834715847733507,
+      "learning_rate": 1e-05,
+      "loss": 0.4904,
+      "step": 1200
+    },
+    {
+      "epoch": 0.34610951008645535,
+      "grad_norm": 2.3666481840738345,
+      "learning_rate": 1e-05,
+      "loss": 0.3251,
+      "step": 1201
+    },
+    {
+      "epoch": 0.3463976945244957,
+      "grad_norm": 3.6847967110324222,
+      "learning_rate": 1e-05,
+      "loss": 0.5112,
+      "step": 1202
+    },
+    {
+      "epoch": 0.346685878962536,
+      "grad_norm": 1.5959837004659907,
+      "learning_rate": 1e-05,
+      "loss": 0.4939,
+      "step": 1203
+    },
+    {
+      "epoch": 0.34697406340057635,
+      "grad_norm": 2.1624166756889753,
+      "learning_rate": 1e-05,
+      "loss": 0.3097,
+      "step": 1204
+    },
+    {
+      "epoch": 0.3472622478386167,
+      "grad_norm": 2.3874538395666125,
+      "learning_rate": 1e-05,
+      "loss": 0.4892,
+      "step": 1205
+    },
+    {
+      "epoch": 0.34755043227665705,
+      "grad_norm": 2.2499510745749514,
+      "learning_rate": 1e-05,
+      "loss": 0.4915,
+      "step": 1206
+    },
+    {
+      "epoch": 0.3478386167146974,
+      "grad_norm": 1.404668578845065,
+      "learning_rate": 1e-05,
+      "loss": 0.3006,
+      "step": 1207
+    },
+    {
+      "epoch": 0.34812680115273775,
+      "grad_norm": 2.7956025858497378,
+      "learning_rate": 1e-05,
+      "loss": 0.4777,
+      "step": 1208
+    },
+    {
+      "epoch": 0.3484149855907781,
+      "grad_norm": 2.379588347293559,
+      "learning_rate": 1e-05,
+      "loss": 0.4961,
+      "step": 1209
+    },
+    {
+      "epoch": 0.34870317002881845,
+      "grad_norm": 2.0538953562342837,
+      "learning_rate": 1e-05,
+      "loss": 0.3216,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3489913544668588,
+      "grad_norm": 2.355086879363187,
+      "learning_rate": 1e-05,
+      "loss": 0.4709,
+      "step": 1211
+    },
+    {
+      "epoch": 0.34927953890489916,
+      "grad_norm": 2.073100433132766,
+      "learning_rate": 1e-05,
+      "loss": 0.4815,
+      "step": 1212
+    },
+    {
+      "epoch": 0.3495677233429395,
+      "grad_norm": 1.673457990794969,
+      "learning_rate": 1e-05,
+      "loss": 0.3215,
+      "step": 1213
+    },
+    {
+      "epoch": 0.3498559077809798,
+      "grad_norm": 2.6904057157051153,
+      "learning_rate": 1e-05,
+      "loss": 0.491,
+      "step": 1214
+    },
+    {
+      "epoch": 0.35014409221902015,
+      "grad_norm": 1.6598497495637436,
+      "learning_rate": 1e-05,
+      "loss": 0.4948,
+      "step": 1215
+    },
+    {
+      "epoch": 0.3504322766570605,
+      "grad_norm": 1.6461641354690129,
+      "learning_rate": 1e-05,
+      "loss": 0.3062,
+      "step": 1216
+    },
+    {
+      "epoch": 0.35072046109510085,
+      "grad_norm": 2.972128860288283,
+      "learning_rate": 1e-05,
+      "loss": 0.4939,
+      "step": 1217
+    },
+    {
+      "epoch": 0.3510086455331412,
+      "grad_norm": 1.9455660845083287,
+      "learning_rate": 1e-05,
+      "loss": 0.4873,
+      "step": 1218
+    },
+    {
+      "epoch": 0.35129682997118156,
+      "grad_norm": 2.7234024735995908,
+      "learning_rate": 1e-05,
+      "loss": 0.337,
+      "step": 1219
+    },
+    {
+      "epoch": 0.3515850144092219,
+      "grad_norm": 2.8288095617482347,
+      "learning_rate": 1e-05,
+      "loss": 0.4854,
+      "step": 1220
+    },
+    {
+      "epoch": 0.35187319884726226,
+      "grad_norm": 1.4670768600406208,
+      "learning_rate": 1e-05,
+      "loss": 0.4928,
+      "step": 1221
+    },
+    {
+      "epoch": 0.3521613832853026,
+      "grad_norm": 2.138084040796824,
+      "learning_rate": 1e-05,
+      "loss": 0.3109,
+      "step": 1222
+    },
+    {
+      "epoch": 0.35244956772334296,
+      "grad_norm": 3.625345475281414,
+      "learning_rate": 1e-05,
+      "loss": 0.5104,
+      "step": 1223
+    },
+    {
+      "epoch": 0.3527377521613833,
+      "grad_norm": 1.5645525491115844,
+      "learning_rate": 1e-05,
+      "loss": 0.4739,
+      "step": 1224
+    },
+    {
+      "epoch": 0.3530259365994236,
+      "grad_norm": 1.388550786833075,
+      "learning_rate": 1e-05,
+      "loss": 0.3304,
+      "step": 1225
+    },
+    {
+      "epoch": 0.35331412103746396,
+      "grad_norm": 2.4455139912586077,
+      "learning_rate": 1e-05,
+      "loss": 0.4635,
+      "step": 1226
+    },
+    {
+      "epoch": 0.3536023054755043,
+      "grad_norm": 1.865557352159103,
+      "learning_rate": 1e-05,
+      "loss": 0.488,
+      "step": 1227
+    },
+    {
+      "epoch": 0.35389048991354466,
+      "grad_norm": 2.5138800299778428,
+      "learning_rate": 1e-05,
+      "loss": 0.3237,
+      "step": 1228
+    },
+    {
+      "epoch": 0.354178674351585,
+      "grad_norm": 2.7204347633921686,
+      "learning_rate": 1e-05,
+      "loss": 0.4777,
+      "step": 1229
+    },
+    {
+      "epoch": 0.35446685878962536,
+      "grad_norm": 1.8944054861645894,
+      "learning_rate": 1e-05,
+      "loss": 0.4904,
+      "step": 1230
+    },
+    {
+      "epoch": 0.3547550432276657,
+      "grad_norm": 1.7954201368104827,
+      "learning_rate": 1e-05,
+      "loss": 0.328,
+      "step": 1231
+    },
+    {
+      "epoch": 0.35504322766570606,
+      "grad_norm": 2.920565212870612,
+      "learning_rate": 1e-05,
+      "loss": 0.464,
+      "step": 1232
+    },
+    {
+      "epoch": 0.3553314121037464,
+      "grad_norm": 2.724892083275897,
+      "learning_rate": 1e-05,
+      "loss": 0.5022,
+      "step": 1233
+    },
+    {
+      "epoch": 0.35561959654178676,
+      "grad_norm": 1.3347016429534064,
+      "learning_rate": 1e-05,
+      "loss": 0.2885,
+      "step": 1234
+    },
+    {
+      "epoch": 0.3559077809798271,
+      "grad_norm": 3.2167881848599342,
+      "learning_rate": 1e-05,
+      "loss": 0.4999,
+      "step": 1235
+    },
+    {
+      "epoch": 0.3561959654178674,
+      "grad_norm": 1.87428169441781,
+      "learning_rate": 1e-05,
+      "loss": 0.4836,
+      "step": 1236
+    },
+    {
+      "epoch": 0.35648414985590776,
+      "grad_norm": 2.8143313945360267,
+      "learning_rate": 1e-05,
+      "loss": 0.3185,
+      "step": 1237
+    },
+    {
+      "epoch": 0.3567723342939481,
+      "grad_norm": 2.3846807789170965,
+      "learning_rate": 1e-05,
+      "loss": 0.4839,
+      "step": 1238
+    },
+    {
+      "epoch": 0.35706051873198846,
+      "grad_norm": 2.035876155842584,
+      "learning_rate": 1e-05,
+      "loss": 0.4875,
+      "step": 1239
+    },
+    {
+      "epoch": 0.3573487031700288,
+      "grad_norm": 1.9676162943433577,
+      "learning_rate": 1e-05,
+      "loss": 0.3067,
+      "step": 1240
+    },
+    {
+      "epoch": 0.35763688760806917,
+      "grad_norm": 3.1917087666871975,
+      "learning_rate": 1e-05,
+      "loss": 0.4951,
+      "step": 1241
+    },
+    {
+      "epoch": 0.3579250720461095,
+      "grad_norm": 1.6257615101345382,
+      "learning_rate": 1e-05,
+      "loss": 0.4707,
+      "step": 1242
+    },
+    {
+      "epoch": 0.35821325648414987,
+      "grad_norm": 2.759397431881197,
+      "learning_rate": 1e-05,
+      "loss": 0.3148,
+      "step": 1243
+    },
+    {
+      "epoch": 0.3585014409221902,
+      "grad_norm": 2.6349765334068067,
+      "learning_rate": 1e-05,
+      "loss": 0.4737,
+      "step": 1244
+    },
+    {
+      "epoch": 0.35878962536023057,
+      "grad_norm": 1.5678831740735477,
+      "learning_rate": 1e-05,
+      "loss": 0.4899,
+      "step": 1245
+    },
+    {
+      "epoch": 0.3590778097982709,
+      "grad_norm": 1.6145035667601524,
+      "learning_rate": 1e-05,
+      "loss": 0.302,
+      "step": 1246
+    },
+    {
+      "epoch": 0.3593659942363112,
+      "grad_norm": 2.9853048113082368,
+      "learning_rate": 1e-05,
+      "loss": 0.5149,
+      "step": 1247
+    },
+    {
+      "epoch": 0.35965417867435157,
+      "grad_norm": 1.357609329275561,
+      "learning_rate": 1e-05,
+      "loss": 0.4912,
+      "step": 1248
+    },
+    {
+      "epoch": 0.3599423631123919,
+      "grad_norm": 2.121872333122013,
+      "learning_rate": 1e-05,
+      "loss": 0.322,
+      "step": 1249
+    },
+    {
+      "epoch": 0.36023054755043227,
+      "grad_norm": 2.145641816572467,
+      "learning_rate": 1e-05,
+      "loss": 0.4628,
+      "step": 1250
+    },
+    {
+      "epoch": 0.36023054755043227,
+      "eval_websight_new_IoU": 0.017749059945344925,
+      "eval_websight_new_MAE_x": 286.7799987792969,
+      "eval_websight_new_MAE_y": 621.280029296875,
+      "eval_websight_new_coord_probability": 0.7440074682235718,
+      "eval_websight_new_correct_token_probability": 0.0020548468455672264,
+      "eval_websight_new_inside_bbox": 0.03999999910593033,
+      "eval_websight_new_loss": 0.3225766122341156,
+      "eval_websight_new_range_100_probability": 0.09060951322317123,
+      "eval_websight_new_range_20_probability": 0.016936045140028,
+      "eval_websight_new_range_5_probability": 0.004216435365378857,
+      "eval_websight_new_runtime": 34.1266,
+      "eval_websight_new_samples_per_second": 1.465,
+      "eval_websight_new_steps_per_second": 0.205,
+      "step": 1250
+    },
+    {
+      "epoch": 0.36023054755043227,
+      "eval_seeclick_IoU": 0.03234250470995903,
+      "eval_seeclick_MAE_x": 473.1000061035156,
+      "eval_seeclick_MAE_y": 587.97998046875,
+      "eval_seeclick_coord_probability": 0.7468330264091492,
+      "eval_seeclick_correct_token_probability": 0.0008442556136287749,
+      "eval_seeclick_inside_bbox": 0.07999999821186066,
+      "eval_seeclick_loss": 0.39571070671081543,
+      "eval_seeclick_range_100_probability": 0.12909740209579468,
+      "eval_seeclick_range_20_probability": 0.0456046499311924,
+      "eval_seeclick_range_5_probability": 0.00795473251491785,
+      "eval_seeclick_runtime": 43.2477,
+      "eval_seeclick_samples_per_second": 1.156,
+      "eval_seeclick_steps_per_second": 0.162,
+      "step": 1250
+    },
+    {
+      "epoch": 0.36023054755043227,
+      "eval_icons_IoU": 0.003231856506317854,
+      "eval_icons_MAE_x": 473.5400085449219,
+      "eval_icons_MAE_y": 375.55999755859375,
+      "eval_icons_coord_probability": 0.7471686601638794,
+      "eval_icons_correct_token_probability": 0.0060055917128920555,
+      "eval_icons_inside_bbox": 0.0,
+      "eval_icons_loss": 0.4698539972305298,
+      "eval_icons_range_100_probability": 0.27581825852394104,
+      "eval_icons_range_20_probability": 0.10117445886135101,
+      "eval_icons_range_5_probability": 0.03151170164346695,
+      "eval_icons_runtime": 42.8179,
+      "eval_icons_samples_per_second": 1.168,
+      "eval_icons_steps_per_second": 0.163,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3605187319884726,
+      "grad_norm": 2.9267574931825977,
+      "learning_rate": 1e-05,
+      "loss": 0.4939,
+      "step": 1251
+    },
+    {
+      "epoch": 0.36080691642651297,
+      "grad_norm": 1.4161651526852423,
+      "learning_rate": 1e-05,
+      "loss": 0.3099,
+      "step": 1252
+    },
+    {
+      "epoch": 0.3610951008645533,
+      "grad_norm": 2.9329296516871315,
+      "learning_rate": 1e-05,
+      "loss": 0.4942,
+      "step": 1253
+    },
+    {
+      "epoch": 0.36138328530259367,
+      "grad_norm": 1.7443290807635572,
+      "learning_rate": 1e-05,
+      "loss": 0.4993,
+      "step": 1254
+    },
+    {
+      "epoch": 0.361671469740634,
+      "grad_norm": 1.5849987171284718,
+      "learning_rate": 1e-05,
+      "loss": 0.3201,
+      "step": 1255
+    },
+    {
+      "epoch": 0.3619596541786744,
+      "grad_norm": 3.388739534462795,
+      "learning_rate": 1e-05,
+      "loss": 0.4998,
+      "step": 1256
+    },
+    {
+      "epoch": 0.3622478386167147,
+      "grad_norm": 2.7104139892986048,
+      "learning_rate": 1e-05,
+      "loss": 0.4875,
+      "step": 1257
+    },
+    {
+      "epoch": 0.362536023054755,
+      "grad_norm": 2.159282660602506,
+      "learning_rate": 1e-05,
+      "loss": 0.3115,
+      "step": 1258
+    },
+    {
+      "epoch": 0.36282420749279537,
+      "grad_norm": 2.5475589208738225,
+      "learning_rate": 1e-05,
+      "loss": 0.471,
+      "step": 1259
+    },
+    {
+      "epoch": 0.3631123919308357,
+      "grad_norm": 1.7243894630942112,
+      "learning_rate": 1e-05,
+      "loss": 0.4764,
+      "step": 1260
+    },
+    {
+      "epoch": 0.36340057636887607,
+      "grad_norm": 1.777580820937449,
+      "learning_rate": 1e-05,
+      "loss": 0.3162,
+      "step": 1261
+    },
+    {
+      "epoch": 0.3636887608069164,
+      "grad_norm": 3.2518330570615053,
+      "learning_rate": 1e-05,
+      "loss": 0.4846,
+      "step": 1262
+    },
+    {
+      "epoch": 0.3639769452449568,
+      "grad_norm": 2.302001017306,
+      "learning_rate": 1e-05,
+      "loss": 0.5045,
+      "step": 1263
+    },
+    {
+      "epoch": 0.3642651296829971,
+      "grad_norm": 2.528883284010198,
+      "learning_rate": 1e-05,
+      "loss": 0.3067,
+      "step": 1264
+    },
+    {
+      "epoch": 0.3645533141210375,
+      "grad_norm": 3.2737716372913637,
+      "learning_rate": 1e-05,
+      "loss": 0.5047,
+      "step": 1265
+    },
+    {
+      "epoch": 0.3648414985590778,
+      "grad_norm": 1.8572245666074538,
+      "learning_rate": 1e-05,
+      "loss": 0.4921,
+      "step": 1266
+    },
+    {
+      "epoch": 0.3651296829971182,
+      "grad_norm": 2.419883123172451,
+      "learning_rate": 1e-05,
+      "loss": 0.3014,
+      "step": 1267
+    },
+    {
+      "epoch": 0.3654178674351585,
+      "grad_norm": 2.380087061074094,
+      "learning_rate": 1e-05,
+      "loss": 0.4939,
+      "step": 1268
+    },
+    {
+      "epoch": 0.3657060518731988,
+      "grad_norm": 1.455830805212864,
+      "learning_rate": 1e-05,
+      "loss": 0.4874,
+      "step": 1269
+    },
+    {
+      "epoch": 0.3659942363112392,
+      "grad_norm": 1.7106818638178742,
+      "learning_rate": 1e-05,
+      "loss": 0.3375,
+      "step": 1270
+    },
+    {
+      "epoch": 0.3662824207492795,
+      "grad_norm": 2.938412142230153,
+      "learning_rate": 1e-05,
+      "loss": 0.5121,
+      "step": 1271
+    },
+    {
+      "epoch": 0.3665706051873199,
+      "grad_norm": 1.944071545920773,
+      "learning_rate": 1e-05,
+      "loss": 0.4717,
+      "step": 1272
+    },
+    {
+      "epoch": 0.3668587896253602,
+      "grad_norm": 2.127957365004338,
+      "learning_rate": 1e-05,
+      "loss": 0.3184,
+      "step": 1273
+    },
+    {
+      "epoch": 0.3671469740634006,
+      "grad_norm": 2.570902478592638,
+      "learning_rate": 1e-05,
+      "loss": 0.4889,
+      "step": 1274
+    },
+    {
+      "epoch": 0.36743515850144093,
+      "grad_norm": 1.657296767571961,
+      "learning_rate": 1e-05,
+      "loss": 0.4818,
+      "step": 1275
+    },
+    {
+      "epoch": 0.3677233429394813,
+      "grad_norm": 1.830962402830421,
+      "learning_rate": 1e-05,
+      "loss": 0.307,
+      "step": 1276
+    },
+    {
+      "epoch": 0.36801152737752163,
+      "grad_norm": 3.2433727313856098,
+      "learning_rate": 1e-05,
+      "loss": 0.5087,
+      "step": 1277
+    },
+    {
+      "epoch": 0.368299711815562,
+      "grad_norm": 3.2623484670700353,
+      "learning_rate": 1e-05,
+      "loss": 0.4897,
+      "step": 1278
+    },
+    {
+      "epoch": 0.36858789625360233,
+      "grad_norm": 1.8457997670091872,
+      "learning_rate": 1e-05,
+      "loss": 0.3077,
+      "step": 1279
+    },
+    {
+      "epoch": 0.3688760806916426,
+      "grad_norm": 3.4046358478665413,
+      "learning_rate": 1e-05,
+      "loss": 0.5126,
+      "step": 1280
+    },
+    {
+      "epoch": 0.369164265129683,
+      "grad_norm": 2.1078191717348997,
+      "learning_rate": 1e-05,
+      "loss": 0.4961,
+      "step": 1281
+    },
+    {
+      "epoch": 0.36945244956772333,
+      "grad_norm": 2.6760556258353914,
+      "learning_rate": 1e-05,
+      "loss": 0.2994,
+      "step": 1282
+    },
+    {
+      "epoch": 0.3697406340057637,
+      "grad_norm": 2.7779267593627672,
+      "learning_rate": 1e-05,
+      "loss": 0.4815,
+      "step": 1283
+    },
+    {
+      "epoch": 0.37002881844380403,
+      "grad_norm": 1.9673132430282592,
+      "learning_rate": 1e-05,
+      "loss": 0.4755,
+      "step": 1284
+    },
+    {
+      "epoch": 0.3703170028818444,
+      "grad_norm": 1.6916980677277118,
+      "learning_rate": 1e-05,
+      "loss": 0.3235,
+      "step": 1285
+    },
+    {
+      "epoch": 0.37060518731988473,
+      "grad_norm": 2.2892499234506247,
+      "learning_rate": 1e-05,
+      "loss": 0.4651,
+      "step": 1286
+    },
+    {
+      "epoch": 0.3708933717579251,
+      "grad_norm": 1.9612395551381552,
+      "learning_rate": 1e-05,
+      "loss": 0.4843,
+      "step": 1287
+    },
+    {
+      "epoch": 0.37118155619596543,
+      "grad_norm": 2.3321421597678538,
+      "learning_rate": 1e-05,
+      "loss": 0.3176,
+      "step": 1288
+    },
+    {
+      "epoch": 0.3714697406340058,
+      "grad_norm": 2.775891285608071,
+      "learning_rate": 1e-05,
+      "loss": 0.4586,
+      "step": 1289
+    },
+    {
+      "epoch": 0.37175792507204614,
+      "grad_norm": 1.5916999326429562,
+      "learning_rate": 1e-05,
+      "loss": 0.4813,
+      "step": 1290
+    },
+    {
+      "epoch": 0.37204610951008643,
+      "grad_norm": 1.3546781839457918,
+      "learning_rate": 1e-05,
+      "loss": 0.3079,
+      "step": 1291
+    },
+    {
+      "epoch": 0.3723342939481268,
+      "grad_norm": 2.034301394316897,
+      "learning_rate": 1e-05,
+      "loss": 0.4496,
+      "step": 1292
+    },
+    {
+      "epoch": 0.37262247838616713,
+      "grad_norm": 2.5830915613085708,
+      "learning_rate": 1e-05,
+      "loss": 0.5008,
+      "step": 1293
+    },
+    {
+      "epoch": 0.3729106628242075,
+      "grad_norm": 1.4480034078466113,
+      "learning_rate": 1e-05,
+      "loss": 0.3008,
+      "step": 1294
+    },
+    {
+      "epoch": 0.37319884726224783,
+      "grad_norm": 2.6955945052568198,
+      "learning_rate": 1e-05,
+      "loss": 0.4906,
+      "step": 1295
+    },
+    {
+      "epoch": 0.3734870317002882,
+      "grad_norm": 1.3768118074381839,
+      "learning_rate": 1e-05,
+      "loss": 0.4803,
+      "step": 1296
+    },
+    {
+      "epoch": 0.37377521613832854,
+      "grad_norm": 2.2906236875197687,
+      "learning_rate": 1e-05,
+      "loss": 0.3053,
+      "step": 1297
+    },
+    {
+      "epoch": 0.3740634005763689,
+      "grad_norm": 3.1020107631734337,
+      "learning_rate": 1e-05,
+      "loss": 0.5018,
+      "step": 1298
+    },
+    {
+      "epoch": 0.37435158501440924,
+      "grad_norm": 3.2403181026251033,
+      "learning_rate": 1e-05,
+      "loss": 0.4919,
+      "step": 1299
+    },
+    {
+      "epoch": 0.3746397694524496,
+      "grad_norm": 1.8184864154877236,
+      "learning_rate": 1e-05,
+      "loss": 0.3025,
+      "step": 1300
+    },
+    {
+      "epoch": 0.37492795389048994,
+      "grad_norm": 2.117689221882779,
+      "learning_rate": 1e-05,
+      "loss": 0.4782,
+      "step": 1301
+    },
+    {
+      "epoch": 0.37521613832853024,
+      "grad_norm": 2.611897522826078,
+      "learning_rate": 1e-05,
+      "loss": 0.4858,
+      "step": 1302
+    },
+    {
+      "epoch": 0.3755043227665706,
+      "grad_norm": 1.6906029962405396,
+      "learning_rate": 1e-05,
+      "loss": 0.3183,
+      "step": 1303
+    },
+    {
+      "epoch": 0.37579250720461094,
+      "grad_norm": 2.740084728664951,
+      "learning_rate": 1e-05,
+      "loss": 0.5032,
+      "step": 1304
+    },
+    {
+      "epoch": 0.3760806916426513,
+      "grad_norm": 1.9411042343379972,
+      "learning_rate": 1e-05,
+      "loss": 0.4928,
+      "step": 1305
+    },
+    {
+      "epoch": 0.37636887608069164,
+      "grad_norm": 2.422741433303496,
+      "learning_rate": 1e-05,
+      "loss": 0.3027,
+      "step": 1306
+    },
+    {
+      "epoch": 0.376657060518732,
+      "grad_norm": 2.2839314934791886,
+      "learning_rate": 1e-05,
+      "loss": 0.4648,
+      "step": 1307
+    },
+    {
+      "epoch": 0.37694524495677234,
+      "grad_norm": 2.115259153820208,
+      "learning_rate": 1e-05,
+      "loss": 0.4783,
+      "step": 1308
+    },
+    {
+      "epoch": 0.3772334293948127,
+      "grad_norm": 2.1455241632791027,
+      "learning_rate": 1e-05,
+      "loss": 0.3167,
+      "step": 1309
+    },
+    {
+      "epoch": 0.37752161383285304,
+      "grad_norm": 3.1279634221134094,
+      "learning_rate": 1e-05,
+      "loss": 0.4851,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3778097982708934,
+      "grad_norm": 1.9221366164413791,
+      "learning_rate": 1e-05,
+      "loss": 0.4933,
+      "step": 1311
+    },
+    {
+      "epoch": 0.37809798270893374,
+      "grad_norm": 1.4491711126374,
+      "learning_rate": 1e-05,
+      "loss": 0.3365,
+      "step": 1312
+    },
+    {
+      "epoch": 0.37838616714697404,
+      "grad_norm": 2.748317882630484,
+      "learning_rate": 1e-05,
+      "loss": 0.4923,
+      "step": 1313
+    },
+    {
+      "epoch": 0.3786743515850144,
+      "grad_norm": 1.6488192337586374,
+      "learning_rate": 1e-05,
+      "loss": 0.4873,
+      "step": 1314
+    },
+    {
+      "epoch": 0.37896253602305474,
+      "grad_norm": 1.8069141908922857,
+      "learning_rate": 1e-05,
+      "loss": 0.2988,
+      "step": 1315
+    },
+    {
+      "epoch": 0.3792507204610951,
+      "grad_norm": 2.497952537050359,
+      "learning_rate": 1e-05,
+      "loss": 0.4943,
+      "step": 1316
+    },
+    {
+      "epoch": 0.37953890489913544,
+      "grad_norm": 1.668255912182165,
+      "learning_rate": 1e-05,
+      "loss": 0.4814,
+      "step": 1317
+    },
+    {
+      "epoch": 0.3798270893371758,
+      "grad_norm": 2.6220758634279675,
+      "learning_rate": 1e-05,
+      "loss": 0.2971,
+      "step": 1318
+    },
+    {
+      "epoch": 0.38011527377521614,
+      "grad_norm": 3.064543845303531,
+      "learning_rate": 1e-05,
+      "loss": 0.4628,
+      "step": 1319
+    },
+    {
+      "epoch": 0.3804034582132565,
+      "grad_norm": 2.1629034925763686,
+      "learning_rate": 1e-05,
+      "loss": 0.4831,
+      "step": 1320
+    },
+    {
+      "epoch": 0.38069164265129685,
+      "grad_norm": 3.4293882979735906,
+      "learning_rate": 1e-05,
+      "loss": 0.3098,
+      "step": 1321
+    },
+    {
+      "epoch": 0.3809798270893372,
+      "grad_norm": 3.153726299390492,
+      "learning_rate": 1e-05,
+      "loss": 0.498,
+      "step": 1322
+    },
+    {
+      "epoch": 0.38126801152737755,
+      "grad_norm": 1.8266526696789749,
+      "learning_rate": 1e-05,
+      "loss": 0.4866,
+      "step": 1323
+    },
+    {
+      "epoch": 0.38155619596541784,
+      "grad_norm": 1.8106958910859294,
+      "learning_rate": 1e-05,
+      "loss": 0.2919,
+      "step": 1324
+    },
+    {
+      "epoch": 0.3818443804034582,
+      "grad_norm": 2.4230832410767764,
+      "learning_rate": 1e-05,
+      "loss": 0.5213,
+      "step": 1325
+    },
+    {
+      "epoch": 0.38213256484149855,
+      "grad_norm": 1.7093502651446373,
+      "learning_rate": 1e-05,
+      "loss": 0.4743,
+      "step": 1326
+    },
+    {
+      "epoch": 0.3824207492795389,
+      "grad_norm": 1.6023156819322093,
+      "learning_rate": 1e-05,
+      "loss": 0.2964,
+      "step": 1327
+    },
+    {
+      "epoch": 0.38270893371757925,
+      "grad_norm": 2.7148544933679832,
+      "learning_rate": 1e-05,
+      "loss": 0.4697,
+      "step": 1328
+    },
+    {
+      "epoch": 0.3829971181556196,
+      "grad_norm": 1.592202006561088,
+      "learning_rate": 1e-05,
+      "loss": 0.4831,
+      "step": 1329
+    },
+    {
+      "epoch": 0.38328530259365995,
+      "grad_norm": 1.5708819141233699,
+      "learning_rate": 1e-05,
+      "loss": 0.3069,
+      "step": 1330
+    },
+    {
+      "epoch": 0.3835734870317003,
+      "grad_norm": 2.193907768911475,
+      "learning_rate": 1e-05,
+      "loss": 0.4949,
+      "step": 1331
+    },
+    {
+      "epoch": 0.38386167146974065,
+      "grad_norm": 1.82617787309074,
+      "learning_rate": 1e-05,
+      "loss": 0.4841,
+      "step": 1332
+    },
+    {
+      "epoch": 0.384149855907781,
+      "grad_norm": 1.649677103567239,
+      "learning_rate": 1e-05,
+      "loss": 0.3103,
+      "step": 1333
+    },
+    {
+      "epoch": 0.38443804034582135,
+      "grad_norm": 3.814639114475869,
+      "learning_rate": 1e-05,
+      "loss": 0.4801,
+      "step": 1334
+    },
+    {
+      "epoch": 0.38472622478386165,
+      "grad_norm": 2.798330556491224,
+      "learning_rate": 1e-05,
+      "loss": 0.4931,
+      "step": 1335
+    },
+    {
+      "epoch": 0.385014409221902,
+      "grad_norm": 2.3047698188749006,
+      "learning_rate": 1e-05,
+      "loss": 0.3029,
+      "step": 1336
+    },
+    {
+      "epoch": 0.38530259365994235,
+      "grad_norm": 2.4107253949315135,
+      "learning_rate": 1e-05,
+      "loss": 0.4941,
+      "step": 1337
+    },
+    {
+      "epoch": 0.3855907780979827,
+      "grad_norm": 2.340576269044811,
+      "learning_rate": 1e-05,
+      "loss": 0.4791,
+      "step": 1338
+    },
+    {
+      "epoch": 0.38587896253602305,
+      "grad_norm": 2.3449681373393587,
+      "learning_rate": 1e-05,
+      "loss": 0.2941,
+      "step": 1339
+    },
+    {
+      "epoch": 0.3861671469740634,
+      "grad_norm": 2.396207023795265,
+      "learning_rate": 1e-05,
+      "loss": 0.4957,
+      "step": 1340
+    },
+    {
+      "epoch": 0.38645533141210375,
+      "grad_norm": 2.2731384422383605,
+      "learning_rate": 1e-05,
+      "loss": 0.4772,
+      "step": 1341
+    },
+    {
+      "epoch": 0.3867435158501441,
+      "grad_norm": 2.1533263214602636,
+      "learning_rate": 1e-05,
+      "loss": 0.3438,
+      "step": 1342
+    },
+    {
+      "epoch": 0.38703170028818445,
+      "grad_norm": 2.2200595691980127,
+      "learning_rate": 1e-05,
+      "loss": 0.4787,
+      "step": 1343
+    },
+    {
+      "epoch": 0.3873198847262248,
+      "grad_norm": 1.7112106300037293,
+      "learning_rate": 1e-05,
+      "loss": 0.4826,
+      "step": 1344
+    },
+    {
+      "epoch": 0.38760806916426516,
+      "grad_norm": 1.9886221802591015,
+      "learning_rate": 1e-05,
+      "loss": 0.2897,
+      "step": 1345
+    },
+    {
+      "epoch": 0.38789625360230545,
+      "grad_norm": 4.093146799119056,
+      "learning_rate": 1e-05,
+      "loss": 0.5198,
+      "step": 1346
+    },
+    {
+      "epoch": 0.3881844380403458,
+      "grad_norm": 1.893178744352946,
+      "learning_rate": 1e-05,
+      "loss": 0.4722,
+      "step": 1347
+    },
+    {
+      "epoch": 0.38847262247838615,
+      "grad_norm": 2.3039484364608693,
+      "learning_rate": 1e-05,
+      "loss": 0.3054,
+      "step": 1348
+    },
+    {
+      "epoch": 0.3887608069164265,
+      "grad_norm": 2.6947301278259728,
+      "learning_rate": 1e-05,
+      "loss": 0.497,
+      "step": 1349
+    },
+    {
+      "epoch": 0.38904899135446686,
+      "grad_norm": 2.1200112126908897,
+      "learning_rate": 1e-05,
+      "loss": 0.4719,
+      "step": 1350
+    },
+    {
+      "epoch": 0.3893371757925072,
+      "grad_norm": 1.4961252933768772,
+      "learning_rate": 1e-05,
+      "loss": 0.3102,
+      "step": 1351
+    },
+    {
+      "epoch": 0.38962536023054756,
+      "grad_norm": 2.141053986186834,
+      "learning_rate": 1e-05,
+      "loss": 0.4916,
+      "step": 1352
+    },
+    {
+      "epoch": 0.3899135446685879,
+      "grad_norm": 3.1611691572610483,
+      "learning_rate": 1e-05,
+      "loss": 0.4837,
+      "step": 1353
+    },
+    {
+      "epoch": 0.39020172910662826,
+      "grad_norm": 1.6877173749956633,
+      "learning_rate": 1e-05,
+      "loss": 0.3019,
+      "step": 1354
+    },
+    {
+      "epoch": 0.3904899135446686,
+      "grad_norm": 2.357194862463654,
+      "learning_rate": 1e-05,
+      "loss": 0.4183,
+      "step": 1355
+    },
+    {
+      "epoch": 0.39077809798270896,
+      "grad_norm": 1.7644790755685829,
+      "learning_rate": 1e-05,
+      "loss": 0.491,
+      "step": 1356
+    },
+    {
+      "epoch": 0.39106628242074926,
+      "grad_norm": 1.748201040487368,
+      "learning_rate": 1e-05,
+      "loss": 0.3265,
+      "step": 1357
+    },
+    {
+      "epoch": 0.3913544668587896,
+      "grad_norm": 2.7881708335472766,
+      "learning_rate": 1e-05,
+      "loss": 0.4923,
+      "step": 1358
+    },
+    {
+      "epoch": 0.39164265129682996,
+      "grad_norm": 1.9879518805417644,
+      "learning_rate": 1e-05,
+      "loss": 0.4709,
+      "step": 1359
+    },
+    {
+      "epoch": 0.3919308357348703,
+      "grad_norm": 1.5462275623714095,
+      "learning_rate": 1e-05,
+      "loss": 0.3173,
+      "step": 1360
+    },
+    {
+      "epoch": 0.39221902017291066,
+      "grad_norm": 2.245245037347222,
+      "learning_rate": 1e-05,
+      "loss": 0.5091,
+      "step": 1361
+    },
+    {
+      "epoch": 0.392507204610951,
+      "grad_norm": 2.0100895309071953,
+      "learning_rate": 1e-05,
+      "loss": 0.4904,
+      "step": 1362
+    },
+    {
+      "epoch": 0.39279538904899136,
+      "grad_norm": 2.0630055451528286,
+      "learning_rate": 1e-05,
+      "loss": 0.3033,
+      "step": 1363
+    },
+    {
+      "epoch": 0.3930835734870317,
+      "grad_norm": 2.5843217211195606,
+      "learning_rate": 1e-05,
+      "loss": 0.4676,
+      "step": 1364
+    },
+    {
+      "epoch": 0.39337175792507206,
+      "grad_norm": 1.790497685082556,
+      "learning_rate": 1e-05,
+      "loss": 0.4872,
+      "step": 1365
+    },
+    {
+      "epoch": 0.3936599423631124,
+      "grad_norm": 3.0926875606313944,
+      "learning_rate": 1e-05,
+      "loss": 0.3062,
+      "step": 1366
+    },
+    {
+      "epoch": 0.39394812680115276,
+      "grad_norm": 3.684659310424777,
+      "learning_rate": 1e-05,
+      "loss": 0.4969,
+      "step": 1367
+    },
+    {
+      "epoch": 0.39423631123919306,
+      "grad_norm": 2.6374079921845732,
+      "learning_rate": 1e-05,
+      "loss": 0.481,
+      "step": 1368
+    },
+    {
+      "epoch": 0.3945244956772334,
+      "grad_norm": 2.529947938300062,
+      "learning_rate": 1e-05,
+      "loss": 0.2729,
+      "step": 1369
+    },
+    {
+      "epoch": 0.39481268011527376,
+      "grad_norm": 2.6403102299203063,
+      "learning_rate": 1e-05,
+      "loss": 0.5024,
+      "step": 1370
+    },
+    {
+      "epoch": 0.3951008645533141,
+      "grad_norm": 2.4139843294062597,
+      "learning_rate": 1e-05,
+      "loss": 0.4853,
+      "step": 1371
+    },
+    {
+      "epoch": 0.39538904899135446,
+      "grad_norm": 1.919988233786583,
+      "learning_rate": 1e-05,
+      "loss": 0.2881,
+      "step": 1372
+    },
+    {
+      "epoch": 0.3956772334293948,
+      "grad_norm": 2.8129664017254736,
+      "learning_rate": 1e-05,
+      "loss": 0.4699,
+      "step": 1373
+    },
+    {
+      "epoch": 0.39596541786743517,
+      "grad_norm": 1.634030689856549,
+      "learning_rate": 1e-05,
+      "loss": 0.4757,
+      "step": 1374
+    },
+    {
+      "epoch": 0.3962536023054755,
+      "grad_norm": 3.424058629062118,
+      "learning_rate": 1e-05,
+      "loss": 0.3152,
+      "step": 1375
+    },
+    {
+      "epoch": 0.39654178674351587,
+      "grad_norm": 3.112838442542796,
+      "learning_rate": 1e-05,
+      "loss": 0.5075,
+      "step": 1376
+    },
+    {
+      "epoch": 0.3968299711815562,
+      "grad_norm": 2.1388924889196996,
+      "learning_rate": 1e-05,
+      "loss": 0.4765,
+      "step": 1377
+    },
+    {
+      "epoch": 0.39711815561959657,
+      "grad_norm": 2.3876696427532207,
+      "learning_rate": 1e-05,
+      "loss": 0.3046,
+      "step": 1378
+    },
+    {
+      "epoch": 0.39740634005763686,
+      "grad_norm": 2.7980681050344045,
+      "learning_rate": 1e-05,
+      "loss": 0.5026,
+      "step": 1379
+    },
+    {
+      "epoch": 0.3976945244956772,
+      "grad_norm": 1.6533707032971585,
+      "learning_rate": 1e-05,
+      "loss": 0.4778,
+      "step": 1380
+    },
+    {
+      "epoch": 0.39798270893371757,
+      "grad_norm": 3.6450302461981057,
+      "learning_rate": 1e-05,
+      "loss": 0.3155,
+      "step": 1381
+    },
+    {
+      "epoch": 0.3982708933717579,
+      "grad_norm": 2.9253563119175645,
+      "learning_rate": 1e-05,
+      "loss": 0.5041,
+      "step": 1382
+    },
+    {
+      "epoch": 0.39855907780979827,
+      "grad_norm": 1.9576576125421563,
+      "learning_rate": 1e-05,
+      "loss": 0.4854,
+      "step": 1383
+    },
+    {
+      "epoch": 0.3988472622478386,
+      "grad_norm": 1.992720781724827,
+      "learning_rate": 1e-05,
+      "loss": 0.2959,
+      "step": 1384
+    },
+    {
+      "epoch": 0.39913544668587897,
+      "grad_norm": 3.110274141952733,
+      "learning_rate": 1e-05,
+      "loss": 0.5109,
+      "step": 1385
+    },
+    {
+      "epoch": 0.3994236311239193,
+      "grad_norm": 1.4837612858925824,
+      "learning_rate": 1e-05,
+      "loss": 0.4757,
+      "step": 1386
+    },
+    {
+      "epoch": 0.39971181556195967,
+      "grad_norm": 1.5299937302980349,
+      "learning_rate": 1e-05,
+      "loss": 0.2849,
+      "step": 1387
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3616359226141648,
+      "learning_rate": 1e-05,
+      "loss": 0.5097,
+      "step": 1388
+    },
+    {
+      "epoch": 0.4002881844380404,
+      "grad_norm": 2.2733226729071183,
+      "learning_rate": 1e-05,
+      "loss": 0.4905,
+      "step": 1389
+    },
+    {
+      "epoch": 0.40057636887608067,
+      "grad_norm": 1.9042849334434573,
+      "learning_rate": 1e-05,
+      "loss": 0.3066,
+      "step": 1390
+    },
+    {
+      "epoch": 0.400864553314121,
+      "grad_norm": 3.129236639675653,
+      "learning_rate": 1e-05,
+      "loss": 0.4525,
+      "step": 1391
+    },
+    {
+      "epoch": 0.40115273775216137,
+      "grad_norm": 1.60501688777708,
+      "learning_rate": 1e-05,
+      "loss": 0.4824,
+      "step": 1392
+    },
+    {
+      "epoch": 0.4014409221902017,
+      "grad_norm": 2.4101207174232164,
+      "learning_rate": 1e-05,
+      "loss": 0.3235,
+      "step": 1393
+    },
+    {
+      "epoch": 0.40172910662824207,
+      "grad_norm": 2.5438807558281376,
+      "learning_rate": 1e-05,
+      "loss": 0.4584,
+      "step": 1394
+    },
+    {
+      "epoch": 0.4020172910662824,
+      "grad_norm": 2.0730066361358466,
+      "learning_rate": 1e-05,
+      "loss": 0.4803,
+      "step": 1395
+    },
+    {
+      "epoch": 0.4023054755043228,
+      "grad_norm": 1.6283441998766814,
+      "learning_rate": 1e-05,
+      "loss": 0.2952,
+      "step": 1396
+    },
+    {
+      "epoch": 0.4025936599423631,
+      "grad_norm": 3.131426981929381,
+      "learning_rate": 1e-05,
+      "loss": 0.4915,
+      "step": 1397
+    },
+    {
+      "epoch": 0.4028818443804035,
+      "grad_norm": 2.184896822021097,
+      "learning_rate": 1e-05,
+      "loss": 0.4905,
+      "step": 1398
+    },
+    {
+      "epoch": 0.4031700288184438,
+      "grad_norm": 1.9986154852514275,
+      "learning_rate": 1e-05,
+      "loss": 0.292,
+      "step": 1399
+    },
+    {
+      "epoch": 0.4034582132564842,
+      "grad_norm": 2.522326409736831,
+      "learning_rate": 1e-05,
+      "loss": 0.4719,
+      "step": 1400
+    },
+    {
+      "epoch": 0.4037463976945245,
+      "grad_norm": 1.7452401996021782,
+      "learning_rate": 1e-05,
+      "loss": 0.4777,
+      "step": 1401
+    },
+    {
+      "epoch": 0.4040345821325648,
+      "grad_norm": 1.6123031076560754,
+      "learning_rate": 1e-05,
+      "loss": 0.32,
+      "step": 1402
+    },
+    {
+      "epoch": 0.4043227665706052,
+      "grad_norm": 2.912808244320673,
+      "learning_rate": 1e-05,
+      "loss": 0.5106,
+      "step": 1403
+    },
+    {
+      "epoch": 0.4046109510086455,
+      "grad_norm": 1.5889190205139632,
+      "learning_rate": 1e-05,
+      "loss": 0.4701,
+      "step": 1404
+    },
+    {
+      "epoch": 0.4048991354466859,
+      "grad_norm": 2.6005923423253963,
+      "learning_rate": 1e-05,
+      "loss": 0.3034,
+      "step": 1405
+    },
+    {
+      "epoch": 0.4051873198847262,
+      "grad_norm": 2.370651320021173,
+      "learning_rate": 1e-05,
+      "loss": 0.5104,
+      "step": 1406
+    },
+    {
+      "epoch": 0.4054755043227666,
+      "grad_norm": 1.8283032148070109,
+      "learning_rate": 1e-05,
+      "loss": 0.4757,
+      "step": 1407
+    },
+    {
+      "epoch": 0.40576368876080693,
+      "grad_norm": 1.7580438693746454,
+      "learning_rate": 1e-05,
+      "loss": 0.294,
+      "step": 1408
+    },
+    {
+      "epoch": 0.4060518731988473,
+      "grad_norm": 2.563673768843064,
+      "learning_rate": 1e-05,
+      "loss": 0.4781,
+      "step": 1409
+    },
+    {
+      "epoch": 0.40634005763688763,
+      "grad_norm": 2.047885702235358,
+      "learning_rate": 1e-05,
+      "loss": 0.4847,
+      "step": 1410
+    },
+    {
+      "epoch": 0.406628242074928,
+      "grad_norm": 1.603045143656897,
+      "learning_rate": 1e-05,
+      "loss": 0.2888,
+      "step": 1411
+    },
+    {
+      "epoch": 0.4069164265129683,
+      "grad_norm": 2.5886615969240387,
+      "learning_rate": 1e-05,
+      "loss": 0.4748,
+      "step": 1412
+    },
+    {
+      "epoch": 0.4072046109510086,
+      "grad_norm": 1.6802958158922825,
+      "learning_rate": 1e-05,
+      "loss": 0.479,
+      "step": 1413
+    },
+    {
+      "epoch": 0.407492795389049,
+      "grad_norm": 1.619522368911882,
+      "learning_rate": 1e-05,
+      "loss": 0.2841,
+      "step": 1414
+    },
+    {
+      "epoch": 0.40778097982708933,
+      "grad_norm": 2.434723005486242,
+      "learning_rate": 1e-05,
+      "loss": 0.4428,
+      "step": 1415
+    },
+    {
+      "epoch": 0.4080691642651297,
+      "grad_norm": 2.3743119348863813,
+      "learning_rate": 1e-05,
+      "loss": 0.4708,
+      "step": 1416
+    },
+    {
+      "epoch": 0.40835734870317003,
+      "grad_norm": 1.5103269192763225,
+      "learning_rate": 1e-05,
+      "loss": 0.2837,
+      "step": 1417
+    },
+    {
+      "epoch": 0.4086455331412104,
+      "grad_norm": 2.6875133844687644,
+      "learning_rate": 1e-05,
+      "loss": 0.4426,
+      "step": 1418
+    },
+    {
+      "epoch": 0.40893371757925073,
+      "grad_norm": 1.742967920823814,
+      "learning_rate": 1e-05,
+      "loss": 0.4762,
+      "step": 1419
+    },
+    {
+      "epoch": 0.4092219020172911,
+      "grad_norm": 1.5959741836972012,
+      "learning_rate": 1e-05,
+      "loss": 0.3049,
+      "step": 1420
+    },
+    {
+      "epoch": 0.40951008645533143,
+      "grad_norm": 2.534248957154458,
+      "learning_rate": 1e-05,
+      "loss": 0.4601,
+      "step": 1421
+    },
+    {
+      "epoch": 0.4097982708933718,
+      "grad_norm": 1.9664757858549118,
+      "learning_rate": 1e-05,
+      "loss": 0.482,
+      "step": 1422
+    },
+    {
+      "epoch": 0.4100864553314121,
+      "grad_norm": 1.5027571897809393,
+      "learning_rate": 1e-05,
+      "loss": 0.2945,
+      "step": 1423
+    },
+    {
+      "epoch": 0.41037463976945243,
+      "grad_norm": 2.952620010745475,
+      "learning_rate": 1e-05,
+      "loss": 0.5281,
+      "step": 1424
+    },
+    {
+      "epoch": 0.4106628242074928,
+      "grad_norm": 1.9273716488939223,
+      "learning_rate": 1e-05,
+      "loss": 0.477,
+      "step": 1425
+    },
+    {
+      "epoch": 0.41095100864553313,
+      "grad_norm": 1.9634936397917053,
+      "learning_rate": 1e-05,
+      "loss": 0.3252,
+      "step": 1426
+    },
+    {
+      "epoch": 0.4112391930835735,
+      "grad_norm": 2.693940835759522,
+      "learning_rate": 1e-05,
+      "loss": 0.4795,
+      "step": 1427
+    },
+    {
+      "epoch": 0.41152737752161384,
+      "grad_norm": 2.4222487408861277,
+      "learning_rate": 1e-05,
+      "loss": 0.4759,
+      "step": 1428
+    },
+    {
+      "epoch": 0.4118155619596542,
+      "grad_norm": 2.1830351365182246,
+      "learning_rate": 1e-05,
+      "loss": 0.306,
+      "step": 1429
+    },
+    {
+      "epoch": 0.41210374639769454,
+      "grad_norm": 2.4769551938384864,
+      "learning_rate": 1e-05,
+      "loss": 0.4586,
+      "step": 1430
+    },
+    {
+      "epoch": 0.4123919308357349,
+      "grad_norm": 2.161859985283154,
+      "learning_rate": 1e-05,
+      "loss": 0.4992,
+      "step": 1431
+    },
+    {
+      "epoch": 0.41268011527377524,
+      "grad_norm": 1.9311564431354833,
+      "learning_rate": 1e-05,
+      "loss": 0.3019,
+      "step": 1432
+    },
+    {
+      "epoch": 0.4129682997118156,
+      "grad_norm": 2.545808915398012,
+      "learning_rate": 1e-05,
+      "loss": 0.488,
+      "step": 1433
+    },
+    {
+      "epoch": 0.4132564841498559,
+      "grad_norm": 2.0614808574437897,
+      "learning_rate": 1e-05,
+      "loss": 0.4795,
+      "step": 1434
+    },
+    {
+      "epoch": 0.41354466858789624,
+      "grad_norm": 1.56057299395756,
+      "learning_rate": 1e-05,
+      "loss": 0.3019,
+      "step": 1435
+    },
+    {
+      "epoch": 0.4138328530259366,
+      "grad_norm": 2.2399026840728262,
+      "learning_rate": 1e-05,
+      "loss": 0.4724,
+      "step": 1436
+    },
+    {
+      "epoch": 0.41412103746397694,
+      "grad_norm": 2.105709306541403,
+      "learning_rate": 1e-05,
+      "loss": 0.4693,
+      "step": 1437
+    },
+    {
+      "epoch": 0.4144092219020173,
+      "grad_norm": 2.162936762973091,
+      "learning_rate": 1e-05,
+      "loss": 0.3169,
+      "step": 1438
+    },
+    {
+      "epoch": 0.41469740634005764,
+      "grad_norm": 2.5542633457686272,
+      "learning_rate": 1e-05,
+      "loss": 0.4678,
+      "step": 1439
+    },
+    {
+      "epoch": 0.414985590778098,
+      "grad_norm": 1.4167261304479106,
+      "learning_rate": 1e-05,
+      "loss": 0.4745,
+      "step": 1440
+    },
+    {
+      "epoch": 0.41527377521613834,
+      "grad_norm": 1.61449310937752,
+      "learning_rate": 1e-05,
+      "loss": 0.291,
+      "step": 1441
+    },
+    {
+      "epoch": 0.4155619596541787,
+      "grad_norm": 2.460741009092124,
+      "learning_rate": 1e-05,
+      "loss": 0.4947,
+      "step": 1442
+    },
+    {
+      "epoch": 0.41585014409221904,
+      "grad_norm": 1.2816142509425972,
+      "learning_rate": 1e-05,
+      "loss": 0.4696,
+      "step": 1443
+    },
+    {
+      "epoch": 0.4161383285302594,
+      "grad_norm": 1.6593680240842172,
+      "learning_rate": 1e-05,
+      "loss": 0.2982,
+      "step": 1444
+    },
+    {
+      "epoch": 0.4164265129682997,
+      "grad_norm": 2.304564977019813,
+      "learning_rate": 1e-05,
+      "loss": 0.428,
+      "step": 1445
+    },
+    {
+      "epoch": 0.41671469740634004,
+      "grad_norm": 1.5971229954726822,
+      "learning_rate": 1e-05,
+      "loss": 0.4756,
+      "step": 1446
+    },
+    {
+      "epoch": 0.4170028818443804,
+      "grad_norm": 1.752756272258912,
+      "learning_rate": 1e-05,
+      "loss": 0.2723,
+      "step": 1447
+    },
+    {
+      "epoch": 0.41729106628242074,
+      "grad_norm": 2.2574850631520462,
+      "learning_rate": 1e-05,
+      "loss": 0.494,
+      "step": 1448
+    },
+    {
+      "epoch": 0.4175792507204611,
+      "grad_norm": 1.5487360772919614,
+      "learning_rate": 1e-05,
+      "loss": 0.4746,
+      "step": 1449
+    },
+    {
+      "epoch": 0.41786743515850144,
+      "grad_norm": 1.6966590813465043,
+      "learning_rate": 1e-05,
+      "loss": 0.3112,
+      "step": 1450
+    },
+    {
+      "epoch": 0.4181556195965418,
+      "grad_norm": 3.2315311759123118,
+      "learning_rate": 1e-05,
+      "loss": 0.4753,
+      "step": 1451
+    },
+    {
+      "epoch": 0.41844380403458215,
+      "grad_norm": 2.2574661317758937,
+      "learning_rate": 1e-05,
+      "loss": 0.4802,
+      "step": 1452
+    },
+    {
+      "epoch": 0.4187319884726225,
+      "grad_norm": 3.9803998971006664,
+      "learning_rate": 1e-05,
+      "loss": 0.2902,
+      "step": 1453
+    },
+    {
+      "epoch": 0.41902017291066285,
+      "grad_norm": 3.209780933951834,
+      "learning_rate": 1e-05,
+      "loss": 0.4602,
+      "step": 1454
+    },
+    {
+      "epoch": 0.41930835734870314,
+      "grad_norm": 2.230305806397594,
+      "learning_rate": 1e-05,
+      "loss": 0.4744,
+      "step": 1455
+    },
+    {
+      "epoch": 0.4195965417867435,
+      "grad_norm": 2.0377336323745716,
+      "learning_rate": 1e-05,
+      "loss": 0.314,
+      "step": 1456
+    },
+    {
+      "epoch": 0.41988472622478384,
+      "grad_norm": 2.0423812560080714,
+      "learning_rate": 1e-05,
+      "loss": 0.4396,
+      "step": 1457
+    },
+    {
+      "epoch": 0.4201729106628242,
+      "grad_norm": 2.058425649358438,
+      "learning_rate": 1e-05,
+      "loss": 0.4805,
+      "step": 1458
+    },
+    {
+      "epoch": 0.42046109510086455,
+      "grad_norm": 2.107814332893319,
+      "learning_rate": 1e-05,
+      "loss": 0.282,
+      "step": 1459
+    },
+    {
+      "epoch": 0.4207492795389049,
+      "grad_norm": 4.456406890207841,
+      "learning_rate": 1e-05,
+      "loss": 0.4971,
+      "step": 1460
+    },
+    {
+      "epoch": 0.42103746397694525,
+      "grad_norm": 1.8378400632612824,
+      "learning_rate": 1e-05,
+      "loss": 0.4748,
+      "step": 1461
+    },
+    {
+      "epoch": 0.4213256484149856,
+      "grad_norm": 2.11974453191249,
+      "learning_rate": 1e-05,
+      "loss": 0.3012,
+      "step": 1462
+    },
+    {
+      "epoch": 0.42161383285302595,
+      "grad_norm": 2.686223251549633,
+      "learning_rate": 1e-05,
+      "loss": 0.503,
+      "step": 1463
+    },
+    {
+      "epoch": 0.4219020172910663,
+      "grad_norm": 1.8509864605201591,
+      "learning_rate": 1e-05,
+      "loss": 0.4759,
+      "step": 1464
+    },
+    {
+      "epoch": 0.42219020172910665,
+      "grad_norm": 2.6826435072967882,
+      "learning_rate": 1e-05,
+      "loss": 0.2867,
+      "step": 1465
+    },
+    {
+      "epoch": 0.42247838616714695,
+      "grad_norm": 4.054905726717292,
+      "learning_rate": 1e-05,
+      "loss": 0.5015,
+      "step": 1466
+    },
+    {
+      "epoch": 0.4227665706051873,
+      "grad_norm": 1.7056820602826313,
+      "learning_rate": 1e-05,
+      "loss": 0.4786,
+      "step": 1467
+    },
+    {
+      "epoch": 0.42305475504322765,
+      "grad_norm": 2.373737807042705,
+      "learning_rate": 1e-05,
+      "loss": 0.2919,
+      "step": 1468
+    },
+    {
+      "epoch": 0.423342939481268,
+      "grad_norm": 3.1455534455823435,
+      "learning_rate": 1e-05,
+      "loss": 0.4915,
+      "step": 1469
+    },
+    {
+      "epoch": 0.42363112391930835,
+      "grad_norm": 1.5609259233210926,
+      "learning_rate": 1e-05,
+      "loss": 0.4792,
+      "step": 1470
+    },
+    {
+      "epoch": 0.4239193083573487,
+      "grad_norm": 2.2781872873900832,
+      "learning_rate": 1e-05,
+      "loss": 0.2805,
+      "step": 1471
+    },
+    {
+      "epoch": 0.42420749279538905,
+      "grad_norm": 2.5535708657805336,
+      "learning_rate": 1e-05,
+      "loss": 0.4633,
+      "step": 1472
+    },
+    {
+      "epoch": 0.4244956772334294,
+      "grad_norm": 2.576172876923238,
+      "learning_rate": 1e-05,
+      "loss": 0.4713,
+      "step": 1473
+    },
+    {
+      "epoch": 0.42478386167146975,
+      "grad_norm": 2.413693568839609,
+      "learning_rate": 1e-05,
+      "loss": 0.2942,
+      "step": 1474
+    },
+    {
+      "epoch": 0.4250720461095101,
+      "grad_norm": 3.7683768285466663,
+      "learning_rate": 1e-05,
+      "loss": 0.5096,
+      "step": 1475
+    },
+    {
+      "epoch": 0.42536023054755046,
+      "grad_norm": 1.4097587762945767,
+      "learning_rate": 1e-05,
+      "loss": 0.4767,
+      "step": 1476
+    },
+    {
+      "epoch": 0.42564841498559075,
+      "grad_norm": 1.9611481825788895,
+      "learning_rate": 1e-05,
+      "loss": 0.3,
+      "step": 1477
+    },
+    {
+      "epoch": 0.4259365994236311,
+      "grad_norm": 2.7697927302614125,
+      "learning_rate": 1e-05,
+      "loss": 0.4952,
+      "step": 1478
+    },
+    {
+      "epoch": 0.42622478386167145,
+      "grad_norm": 2.0120709047065715,
+      "learning_rate": 1e-05,
+      "loss": 0.4861,
+      "step": 1479
+    },
+    {
+      "epoch": 0.4265129682997118,
+      "grad_norm": 2.239333012962598,
+      "learning_rate": 1e-05,
+      "loss": 0.3059,
+      "step": 1480
+    },
+    {
+      "epoch": 0.42680115273775215,
+      "grad_norm": 3.9713986125353244,
+      "learning_rate": 1e-05,
+      "loss": 0.5035,
+      "step": 1481
+    },
+    {
+      "epoch": 0.4270893371757925,
+      "grad_norm": 2.135974392250055,
+      "learning_rate": 1e-05,
+      "loss": 0.4727,
+      "step": 1482
+    },
+    {
+      "epoch": 0.42737752161383286,
+      "grad_norm": 1.7694741634549267,
+      "learning_rate": 1e-05,
+      "loss": 0.3134,
+      "step": 1483
+    },
+    {
+      "epoch": 0.4276657060518732,
+      "grad_norm": 2.201081975407542,
+      "learning_rate": 1e-05,
+      "loss": 0.4837,
+      "step": 1484
+    },
+    {
+      "epoch": 0.42795389048991356,
+      "grad_norm": 2.1049177714570804,
+      "learning_rate": 1e-05,
+      "loss": 0.4713,
+      "step": 1485
+    },
+    {
+      "epoch": 0.4282420749279539,
+      "grad_norm": 1.828062091862777,
+      "learning_rate": 1e-05,
+      "loss": 0.3028,
+      "step": 1486
+    },
+    {
+      "epoch": 0.42853025936599426,
+      "grad_norm": 3.388980148519981,
+      "learning_rate": 1e-05,
+      "loss": 0.4695,
+      "step": 1487
+    },
+    {
+      "epoch": 0.42881844380403455,
+      "grad_norm": 2.071387382431709,
+      "learning_rate": 1e-05,
+      "loss": 0.4655,
+      "step": 1488
+    },
+    {
+      "epoch": 0.4291066282420749,
+      "grad_norm": 2.473212913479042,
+      "learning_rate": 1e-05,
+      "loss": 0.2895,
+      "step": 1489
+    },
+    {
+      "epoch": 0.42939481268011526,
+      "grad_norm": 3.5488756794281264,
+      "learning_rate": 1e-05,
+      "loss": 0.4806,
+      "step": 1490
+    },
+    {
+      "epoch": 0.4296829971181556,
+      "grad_norm": 2.070394697000306,
+      "learning_rate": 1e-05,
+      "loss": 0.4697,
+      "step": 1491
+    },
+    {
+      "epoch": 0.42997118155619596,
+      "grad_norm": 1.7693617947762856,
+      "learning_rate": 1e-05,
+      "loss": 0.2963,
+      "step": 1492
+    },
+    {
+      "epoch": 0.4302593659942363,
+      "grad_norm": 2.2658425376271634,
+      "learning_rate": 1e-05,
+      "loss": 0.4988,
+      "step": 1493
+    },
+    {
+      "epoch": 0.43054755043227666,
+      "grad_norm": 1.635647565764361,
+      "learning_rate": 1e-05,
+      "loss": 0.4618,
+      "step": 1494
+    },
+    {
+      "epoch": 0.430835734870317,
+      "grad_norm": 2.844256003261886,
+      "learning_rate": 1e-05,
+      "loss": 0.2956,
+      "step": 1495
+    },
+    {
+      "epoch": 0.43112391930835736,
+      "grad_norm": 1.8942081514772837,
+      "learning_rate": 1e-05,
+      "loss": 0.4516,
+      "step": 1496
+    },
+    {
+      "epoch": 0.4314121037463977,
+      "grad_norm": 1.8172410700735635,
+      "learning_rate": 1e-05,
+      "loss": 0.4769,
+      "step": 1497
+    },
+    {
+      "epoch": 0.43170028818443806,
+      "grad_norm": 2.007527346024837,
+      "learning_rate": 1e-05,
+      "loss": 0.2824,
+      "step": 1498
+    },
+    {
+      "epoch": 0.43198847262247836,
+      "grad_norm": 3.226794105209606,
+      "learning_rate": 1e-05,
+      "loss": 0.5084,
+      "step": 1499
+    },
+    {
+      "epoch": 0.4322766570605187,
+      "grad_norm": 1.906344602352076,
+      "learning_rate": 1e-05,
+      "loss": 0.4764,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4322766570605187,
+      "eval_websight_new_IoU": 0.018606476485729218,
+      "eval_websight_new_MAE_x": 286.7799987792969,
+      "eval_websight_new_MAE_y": 592.5800170898438,
+      "eval_websight_new_coord_probability": 0.7464273571968079,
+      "eval_websight_new_correct_token_probability": 0.0013086475664749742,
+      "eval_websight_new_inside_bbox": 0.019999999552965164,
+      "eval_websight_new_loss": 0.3021695017814636,
+      "eval_websight_new_range_100_probability": 0.09368956834077835,
+      "eval_websight_new_range_20_probability": 0.017098380252718925,
+      "eval_websight_new_range_5_probability": 0.003493149532005191,
+      "eval_websight_new_runtime": 34.573,
+      "eval_websight_new_samples_per_second": 1.446,
+      "eval_websight_new_steps_per_second": 0.202,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4322766570605187,
+      "eval_seeclick_IoU": 0.0315781868994236,
+      "eval_seeclick_MAE_x": 473.1000061035156,
+      "eval_seeclick_MAE_y": 555.760009765625,
+      "eval_seeclick_coord_probability": 0.7461659908294678,
+      "eval_seeclick_correct_token_probability": 0.0007614689529873431,
+      "eval_seeclick_inside_bbox": 0.07999999821186066,
+      "eval_seeclick_loss": 0.3881937861442566,
+      "eval_seeclick_range_100_probability": 0.1348019391298294,
+      "eval_seeclick_range_20_probability": 0.053533121943473816,
+      "eval_seeclick_range_5_probability": 0.008099403232336044,
+      "eval_seeclick_runtime": 41.6794,
+      "eval_seeclick_samples_per_second": 1.2,
+      "eval_seeclick_steps_per_second": 0.168,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4322766570605187,
+      "eval_icons_IoU": 0.002965953666716814,
+      "eval_icons_MAE_x": 473.5400085449219,
+      "eval_icons_MAE_y": 583.760009765625,
+      "eval_icons_coord_probability": 0.747755229473114,
+      "eval_icons_correct_token_probability": 0.00685780169442296,
+      "eval_icons_inside_bbox": 0.0,
+      "eval_icons_loss": 0.45823726058006287,
+      "eval_icons_range_100_probability": 0.27524334192276,
+      "eval_icons_range_20_probability": 0.09912868589162827,
+      "eval_icons_range_5_probability": 0.030710894614458084,
+      "eval_icons_runtime": 43.7339,
+      "eval_icons_samples_per_second": 1.143,
+      "eval_icons_steps_per_second": 0.16,
+      "step": 1500
+    },
+    {
+      "epoch": 0.43256484149855906,
+      "grad_norm": 1.5872574925183616,
+      "learning_rate": 1e-05,
+      "loss": 0.2881,
+      "step": 1501
+    },
+    {
+      "epoch": 0.4328530259365994,
+      "grad_norm": 2.513250812854745,
+      "learning_rate": 1e-05,
+      "loss": 0.4873,
+      "step": 1502
+    },
+    {
+      "epoch": 0.43314121037463976,
+      "grad_norm": 1.4467163565839434,
+      "learning_rate": 1e-05,
+      "loss": 0.4754,
+      "step": 1503
+    },
+    {
+      "epoch": 0.4334293948126801,
+      "grad_norm": 2.8143907431672854,
+      "learning_rate": 1e-05,
+      "loss": 0.2882,
+      "step": 1504
+    },
+    {
+      "epoch": 0.43371757925072046,
+      "grad_norm": 3.1965223788772543,
+      "learning_rate": 1e-05,
+      "loss": 0.464,
+      "step": 1505
+    },
+    {
+      "epoch": 0.4340057636887608,
+      "grad_norm": 1.9414117477439994,
+      "learning_rate": 1e-05,
+      "loss": 0.4703,
+      "step": 1506
+    },
+    {
+      "epoch": 0.43429394812680117,
+      "grad_norm": 1.476265941565406,
+      "learning_rate": 1e-05,
+      "loss": 0.2882,
+      "step": 1507
+    },
+    {
+      "epoch": 0.4345821325648415,
+      "grad_norm": 3.288435398577001,
+      "learning_rate": 1e-05,
+      "loss": 0.4863,
+      "step": 1508
+    },
+    {
+      "epoch": 0.43487031700288187,
+      "grad_norm": 1.7781756587565274,
+      "learning_rate": 1e-05,
+      "loss": 0.4782,
+      "step": 1509
+    },
+    {
+      "epoch": 0.43515850144092216,
+      "grad_norm": 1.5639576209977695,
+      "learning_rate": 1e-05,
+      "loss": 0.2804,
+      "step": 1510
+    },
+    {
+      "epoch": 0.4354466858789625,
+      "grad_norm": 2.5849426529224067,
+      "learning_rate": 1e-05,
+      "loss": 0.4566,
+      "step": 1511
+    },
+    {
+      "epoch": 0.43573487031700286,
+      "grad_norm": 2.224779390377121,
+      "learning_rate": 1e-05,
+      "loss": 0.4695,
+      "step": 1512
+    },
+    {
+      "epoch": 0.4360230547550432,
+      "grad_norm": 2.518276835035036,
+      "learning_rate": 1e-05,
+      "loss": 0.2947,
+      "step": 1513
+    },
+    {
+      "epoch": 0.43631123919308357,
+      "grad_norm": 2.16698257127677,
+      "learning_rate": 1e-05,
+      "loss": 0.4633,
+      "step": 1514
+    },
+    {
+      "epoch": 0.4365994236311239,
+      "grad_norm": 2.0026538905015023,
+      "learning_rate": 1e-05,
+      "loss": 0.4654,
+      "step": 1515
+    },
+    {
+      "epoch": 0.43688760806916427,
+      "grad_norm": 1.5740477738801,
+      "learning_rate": 1e-05,
+      "loss": 0.3039,
+      "step": 1516
+    },
+    {
+      "epoch": 0.4371757925072046,
+      "grad_norm": 2.237092615867006,
+      "learning_rate": 1e-05,
+      "loss": 0.4779,
+      "step": 1517
+    },
+    {
+      "epoch": 0.43746397694524497,
+      "grad_norm": 2.8136904641085607,
+      "learning_rate": 1e-05,
+      "loss": 0.4734,
+      "step": 1518
+    },
+    {
+      "epoch": 0.4377521613832853,
+      "grad_norm": 1.4916193545442005,
+      "learning_rate": 1e-05,
+      "loss": 0.2616,
+      "step": 1519
+    },
+    {
+      "epoch": 0.43804034582132567,
+      "grad_norm": 2.319559728956529,
+      "learning_rate": 1e-05,
+      "loss": 0.479,
+      "step": 1520
+    },
+    {
+      "epoch": 0.43832853025936597,
+      "grad_norm": 1.9413915090878375,
+      "learning_rate": 1e-05,
+      "loss": 0.4712,
+      "step": 1521
+    },
+    {
+      "epoch": 0.4386167146974063,
+      "grad_norm": 1.5467348645246979,
+      "learning_rate": 1e-05,
+      "loss": 0.2958,
+      "step": 1522
+    },
+    {
+      "epoch": 0.43890489913544667,
+      "grad_norm": 3.514060979099786,
+      "learning_rate": 1e-05,
+      "loss": 0.4728,
+      "step": 1523
+    },
+    {
+      "epoch": 0.439193083573487,
+      "grad_norm": 1.816045519131449,
+      "learning_rate": 1e-05,
+      "loss": 0.4772,
+      "step": 1524
+    },
+    {
+      "epoch": 0.43948126801152737,
+      "grad_norm": 2.0350489231668996,
+      "learning_rate": 1e-05,
+      "loss": 0.3024,
+      "step": 1525
+    },
+    {
+      "epoch": 0.4397694524495677,
+      "grad_norm": 3.2219810135777163,
+      "learning_rate": 1e-05,
+      "loss": 0.4523,
+      "step": 1526
+    },
+    {
+      "epoch": 0.4400576368876081,
+      "grad_norm": 2.0807276821580425,
+      "learning_rate": 1e-05,
+      "loss": 0.4596,
+      "step": 1527
+    },
+    {
+      "epoch": 0.4403458213256484,
+      "grad_norm": 2.09696549398811,
+      "learning_rate": 1e-05,
+      "loss": 0.3015,
+      "step": 1528
+    },
+    {
+      "epoch": 0.4406340057636888,
+      "grad_norm": 3.4772008092345343,
+      "learning_rate": 1e-05,
+      "loss": 0.4883,
+      "step": 1529
+    },
+    {
+      "epoch": 0.4409221902017291,
+      "grad_norm": 1.82544342216196,
+      "learning_rate": 1e-05,
+      "loss": 0.4658,
+      "step": 1530
+    },
+    {
+      "epoch": 0.4412103746397695,
+      "grad_norm": 1.723235618489549,
+      "learning_rate": 1e-05,
+      "loss": 0.2821,
+      "step": 1531
+    },
+    {
+      "epoch": 0.44149855907780977,
+      "grad_norm": 2.3441861237895467,
+      "learning_rate": 1e-05,
+      "loss": 0.4624,
+      "step": 1532
+    },
+    {
+      "epoch": 0.4417867435158501,
+      "grad_norm": 1.6247000768593272,
+      "learning_rate": 1e-05,
+      "loss": 0.4691,
+      "step": 1533
+    },
+    {
+      "epoch": 0.4420749279538905,
+      "grad_norm": 1.9745718388471114,
+      "learning_rate": 1e-05,
+      "loss": 0.2873,
+      "step": 1534
+    },
+    {
+      "epoch": 0.4423631123919308,
+      "grad_norm": 3.324646745868719,
+      "learning_rate": 1e-05,
+      "loss": 0.4745,
+      "step": 1535
+    },
+    {
+      "epoch": 0.4426512968299712,
+      "grad_norm": 1.8867451406839792,
+      "learning_rate": 1e-05,
+      "loss": 0.4717,
+      "step": 1536
+    },
+    {
+      "epoch": 0.4429394812680115,
+      "grad_norm": 2.4174374781034964,
+      "learning_rate": 1e-05,
+      "loss": 0.2877,
+      "step": 1537
+    },
+    {
+      "epoch": 0.4432276657060519,
+      "grad_norm": 2.5017887054244565,
+      "learning_rate": 1e-05,
+      "loss": 0.4598,
+      "step": 1538
+    },
+    {
+      "epoch": 0.4435158501440922,
+      "grad_norm": 1.7850722546794648,
+      "learning_rate": 1e-05,
+      "loss": 0.4674,
+      "step": 1539
+    },
+    {
+      "epoch": 0.4438040345821326,
+      "grad_norm": 2.148436440003259,
+      "learning_rate": 1e-05,
+      "loss": 0.2897,
+      "step": 1540
+    },
+    {
+      "epoch": 0.44409221902017293,
+      "grad_norm": 3.2171685780383497,
+      "learning_rate": 1e-05,
+      "loss": 0.4662,
+      "step": 1541
+    },
+    {
+      "epoch": 0.4443804034582133,
+      "grad_norm": 1.5096168271723365,
+      "learning_rate": 1e-05,
+      "loss": 0.4606,
+      "step": 1542
+    },
+    {
+      "epoch": 0.4446685878962536,
+      "grad_norm": 1.8582772475076226,
+      "learning_rate": 1e-05,
+      "loss": 0.2873,
+      "step": 1543
+    },
+    {
+      "epoch": 0.4449567723342939,
+      "grad_norm": 2.5279416610391325,
+      "learning_rate": 1e-05,
+      "loss": 0.4558,
+      "step": 1544
+    },
+    {
+      "epoch": 0.4452449567723343,
+      "grad_norm": 1.9269150992686295,
+      "learning_rate": 1e-05,
+      "loss": 0.4652,
+      "step": 1545
+    },
+    {
+      "epoch": 0.44553314121037463,
+      "grad_norm": 2.1094191576045493,
+      "learning_rate": 1e-05,
+      "loss": 0.2997,
+      "step": 1546
+    },
+    {
+      "epoch": 0.445821325648415,
+      "grad_norm": 2.7335428758417453,
+      "learning_rate": 1e-05,
+      "loss": 0.4816,
+      "step": 1547
+    },
+    {
+      "epoch": 0.44610951008645533,
+      "grad_norm": 1.4570127507970505,
+      "learning_rate": 1e-05,
+      "loss": 0.4741,
+      "step": 1548
+    },
+    {
+      "epoch": 0.4463976945244957,
+      "grad_norm": 2.0189080812348483,
+      "learning_rate": 1e-05,
+      "loss": 0.2861,
+      "step": 1549
+    },
+    {
+      "epoch": 0.44668587896253603,
+      "grad_norm": 3.269122993577927,
+      "learning_rate": 1e-05,
+      "loss": 0.4595,
+      "step": 1550
+    },
+    {
+      "epoch": 0.4469740634005764,
+      "grad_norm": 1.5717104712633572,
+      "learning_rate": 1e-05,
+      "loss": 0.4653,
+      "step": 1551
+    },
+    {
+      "epoch": 0.44726224783861673,
+      "grad_norm": 2.0530804084811005,
+      "learning_rate": 1e-05,
+      "loss": 0.2712,
+      "step": 1552
+    },
+    {
+      "epoch": 0.4475504322766571,
+      "grad_norm": 2.2505052949895505,
+      "learning_rate": 1e-05,
+      "loss": 0.4901,
+      "step": 1553
+    },
+    {
+      "epoch": 0.4478386167146974,
+      "grad_norm": 2.041908678963047,
+      "learning_rate": 1e-05,
+      "loss": 0.4785,
+      "step": 1554
+    },
+    {
+      "epoch": 0.44812680115273773,
+      "grad_norm": 1.6502146780344642,
+      "learning_rate": 1e-05,
+      "loss": 0.2822,
+      "step": 1555
+    },
+    {
+      "epoch": 0.4484149855907781,
+      "grad_norm": 2.09226285052814,
+      "learning_rate": 1e-05,
+      "loss": 0.4674,
+      "step": 1556
+    },
+    {
+      "epoch": 0.44870317002881843,
+      "grad_norm": 1.682934145653075,
+      "learning_rate": 1e-05,
+      "loss": 0.4692,
+      "step": 1557
+    },
+    {
+      "epoch": 0.4489913544668588,
+      "grad_norm": 1.7936155554218134,
+      "learning_rate": 1e-05,
+      "loss": 0.2736,
+      "step": 1558
+    },
+    {
+      "epoch": 0.44927953890489913,
+      "grad_norm": 2.9374696279877224,
+      "learning_rate": 1e-05,
+      "loss": 0.4824,
+      "step": 1559
+    },
+    {
+      "epoch": 0.4495677233429395,
+      "grad_norm": 2.018776048115124,
+      "learning_rate": 1e-05,
+      "loss": 0.4673,
+      "step": 1560
+    },
+    {
+      "epoch": 0.44985590778097984,
+      "grad_norm": 1.8145829483936216,
+      "learning_rate": 1e-05,
+      "loss": 0.2616,
+      "step": 1561
+    },
+    {
+      "epoch": 0.4501440922190202,
+      "grad_norm": 2.311170758105593,
+      "learning_rate": 1e-05,
+      "loss": 0.4661,
+      "step": 1562
+    },
+    {
+      "epoch": 0.45043227665706054,
+      "grad_norm": 2.1575239049221206,
+      "learning_rate": 1e-05,
+      "loss": 0.4726,
+      "step": 1563
+    },
+    {
+      "epoch": 0.4507204610951009,
+      "grad_norm": 1.398521466330883,
+      "learning_rate": 1e-05,
+      "loss": 0.2966,
+      "step": 1564
+    },
+    {
+      "epoch": 0.4510086455331412,
+      "grad_norm": 3.0139786653151392,
+      "learning_rate": 1e-05,
+      "loss": 0.4256,
+      "step": 1565
+    },
+    {
+      "epoch": 0.45129682997118153,
+      "grad_norm": 1.5206472061961696,
+      "learning_rate": 1e-05,
+      "loss": 0.4687,
+      "step": 1566
+    },
+    {
+      "epoch": 0.4515850144092219,
+      "grad_norm": 2.160624694251746,
+      "learning_rate": 1e-05,
+      "loss": 0.3051,
+      "step": 1567
+    },
+    {
+      "epoch": 0.45187319884726224,
+      "grad_norm": 2.5482743201834444,
+      "learning_rate": 1e-05,
+      "loss": 0.443,
+      "step": 1568
+    },
+    {
+      "epoch": 0.4521613832853026,
+      "grad_norm": 2.0224049925494327,
+      "learning_rate": 1e-05,
+      "loss": 0.4585,
+      "step": 1569
+    },
+    {
+      "epoch": 0.45244956772334294,
+      "grad_norm": 2.460064527237816,
+      "learning_rate": 1e-05,
+      "loss": 0.2961,
+      "step": 1570
+    },
+    {
+      "epoch": 0.4527377521613833,
+      "grad_norm": 3.2057774393838154,
+      "learning_rate": 1e-05,
+      "loss": 0.4577,
+      "step": 1571
+    },
+    {
+      "epoch": 0.45302593659942364,
+      "grad_norm": 1.7868727145251238,
+      "learning_rate": 1e-05,
+      "loss": 0.4715,
+      "step": 1572
+    },
+    {
+      "epoch": 0.453314121037464,
+      "grad_norm": 2.29698423854333,
+      "learning_rate": 1e-05,
+      "loss": 0.2844,
+      "step": 1573
+    },
+    {
+      "epoch": 0.45360230547550434,
+      "grad_norm": 2.917727972513402,
+      "learning_rate": 1e-05,
+      "loss": 0.4916,
+      "step": 1574
+    },
+    {
+      "epoch": 0.4538904899135447,
+      "grad_norm": 1.6247585355469811,
+      "learning_rate": 1e-05,
+      "loss": 0.4669,
+      "step": 1575
+    },
+    {
+      "epoch": 0.454178674351585,
+      "grad_norm": 1.5614244317128618,
+      "learning_rate": 1e-05,
+      "loss": 0.2599,
+      "step": 1576
+    },
+    {
+      "epoch": 0.45446685878962534,
+      "grad_norm": 2.1206996263612243,
+      "learning_rate": 1e-05,
+      "loss": 0.4601,
+      "step": 1577
+    },
+    {
+      "epoch": 0.4547550432276657,
+      "grad_norm": 2.047320752803363,
+      "learning_rate": 1e-05,
+      "loss": 0.4831,
+      "step": 1578
+    },
+    {
+      "epoch": 0.45504322766570604,
+      "grad_norm": 2.1405983343150456,
+      "learning_rate": 1e-05,
+      "loss": 0.2845,
+      "step": 1579
+    },
+    {
+      "epoch": 0.4553314121037464,
+      "grad_norm": 2.458083877361452,
+      "learning_rate": 1e-05,
+      "loss": 0.4693,
+      "step": 1580
+    },
+    {
+      "epoch": 0.45561959654178674,
+      "grad_norm": 1.6355539726377584,
+      "learning_rate": 1e-05,
+      "loss": 0.4725,
+      "step": 1581
+    },
+    {
+      "epoch": 0.4559077809798271,
+      "grad_norm": 1.878882098430483,
+      "learning_rate": 1e-05,
+      "loss": 0.2986,
+      "step": 1582
+    },
+    {
+      "epoch": 0.45619596541786744,
+      "grad_norm": 3.4414263782465637,
+      "learning_rate": 1e-05,
+      "loss": 0.4829,
+      "step": 1583
+    },
+    {
+      "epoch": 0.4564841498559078,
+      "grad_norm": 1.7070299972485468,
+      "learning_rate": 1e-05,
+      "loss": 0.4626,
+      "step": 1584
+    },
+    {
+      "epoch": 0.45677233429394815,
+      "grad_norm": 1.4912568027028905,
+      "learning_rate": 1e-05,
+      "loss": 0.2611,
+      "step": 1585
+    },
+    {
+      "epoch": 0.4570605187319885,
+      "grad_norm": 2.5139741315017567,
+      "learning_rate": 1e-05,
+      "loss": 0.4912,
+      "step": 1586
+    },
+    {
+      "epoch": 0.4573487031700288,
+      "grad_norm": 1.7131976761887835,
+      "learning_rate": 1e-05,
+      "loss": 0.478,
+      "step": 1587
+    },
+    {
+      "epoch": 0.45763688760806914,
+      "grad_norm": 1.5223364212938868,
+      "learning_rate": 1e-05,
+      "loss": 0.2757,
+      "step": 1588
+    },
+    {
+      "epoch": 0.4579250720461095,
+      "grad_norm": 2.090334564468152,
+      "learning_rate": 1e-05,
+      "loss": 0.4376,
+      "step": 1589
+    },
+    {
+      "epoch": 0.45821325648414984,
+      "grad_norm": 1.384471336032835,
+      "learning_rate": 1e-05,
+      "loss": 0.4659,
+      "step": 1590
+    },
+    {
+      "epoch": 0.4585014409221902,
+      "grad_norm": 1.6695985549873653,
+      "learning_rate": 1e-05,
+      "loss": 0.275,
+      "step": 1591
+    },
+    {
+      "epoch": 0.45878962536023055,
+      "grad_norm": 2.2964318697011254,
+      "learning_rate": 1e-05,
+      "loss": 0.4346,
+      "step": 1592
+    },
+    {
+      "epoch": 0.4590778097982709,
+      "grad_norm": 1.8497059623298822,
+      "learning_rate": 1e-05,
+      "loss": 0.4704,
+      "step": 1593
+    },
+    {
+      "epoch": 0.45936599423631125,
+      "grad_norm": 1.6835772981604133,
+      "learning_rate": 1e-05,
+      "loss": 0.2774,
+      "step": 1594
+    },
+    {
+      "epoch": 0.4596541786743516,
+      "grad_norm": 2.7826512914461503,
+      "learning_rate": 1e-05,
+      "loss": 0.4566,
+      "step": 1595
+    },
+    {
+      "epoch": 0.45994236311239195,
+      "grad_norm": 1.9990675916202083,
+      "learning_rate": 1e-05,
+      "loss": 0.4626,
+      "step": 1596
+    },
+    {
+      "epoch": 0.4602305475504323,
+      "grad_norm": 1.6885590526293213,
+      "learning_rate": 1e-05,
+      "loss": 0.273,
+      "step": 1597
+    },
+    {
+      "epoch": 0.4605187319884726,
+      "grad_norm": 3.0511730495487233,
+      "learning_rate": 1e-05,
+      "loss": 0.4417,
+      "step": 1598
+    },
+    {
+      "epoch": 0.46080691642651295,
+      "grad_norm": 1.9622648574010686,
+      "learning_rate": 1e-05,
+      "loss": 0.467,
+      "step": 1599
+    },
+    {
+      "epoch": 0.4610951008645533,
+      "grad_norm": 1.5492189056892085,
+      "learning_rate": 1e-05,
+      "loss": 0.2778,
+      "step": 1600
+    },
+    {
+      "epoch": 0.46138328530259365,
+      "grad_norm": 2.616357665417584,
+      "learning_rate": 1e-05,
+      "loss": 0.4732,
+      "step": 1601
+    },
+    {
+      "epoch": 0.461671469740634,
+      "grad_norm": 2.220624824895359,
+      "learning_rate": 1e-05,
+      "loss": 0.4628,
+      "step": 1602
+    },
+    {
+      "epoch": 0.46195965417867435,
+      "grad_norm": 2.5573975084738882,
+      "learning_rate": 1e-05,
+      "loss": 0.2894,
+      "step": 1603
+    },
+    {
+      "epoch": 0.4622478386167147,
+      "grad_norm": 2.4115610321105403,
+      "learning_rate": 1e-05,
+      "loss": 0.4605,
+      "step": 1604
+    },
+    {
+      "epoch": 0.46253602305475505,
+      "grad_norm": 1.9242603208135884,
+      "learning_rate": 1e-05,
+      "loss": 0.4705,
+      "step": 1605
+    },
+    {
+      "epoch": 0.4628242074927954,
+      "grad_norm": 2.0542088440136035,
+      "learning_rate": 1e-05,
+      "loss": 0.2821,
+      "step": 1606
+    },
+    {
+      "epoch": 0.46311239193083575,
+      "grad_norm": 2.234799390002291,
+      "learning_rate": 1e-05,
+      "loss": 0.5011,
+      "step": 1607
+    },
+    {
+      "epoch": 0.4634005763688761,
+      "grad_norm": 2.3923851771137374,
+      "learning_rate": 1e-05,
+      "loss": 0.4651,
+      "step": 1608
+    },
+    {
+      "epoch": 0.4636887608069164,
+      "grad_norm": 1.4616520651039109,
+      "learning_rate": 1e-05,
+      "loss": 0.2639,
+      "step": 1609
+    },
+    {
+      "epoch": 0.46397694524495675,
+      "grad_norm": 2.9292817123777364,
+      "learning_rate": 1e-05,
+      "loss": 0.4638,
+      "step": 1610
+    },
+    {
+      "epoch": 0.4642651296829971,
+      "grad_norm": 1.7415947974614805,
+      "learning_rate": 1e-05,
+      "loss": 0.4762,
+      "step": 1611
+    },
+    {
+      "epoch": 0.46455331412103745,
+      "grad_norm": 1.670760159121049,
+      "learning_rate": 1e-05,
+      "loss": 0.2658,
+      "step": 1612
+    },
+    {
+      "epoch": 0.4648414985590778,
+      "grad_norm": 2.5213464652038535,
+      "learning_rate": 1e-05,
+      "loss": 0.4756,
+      "step": 1613
+    },
+    {
+      "epoch": 0.46512968299711815,
+      "grad_norm": 1.628758078685135,
+      "learning_rate": 1e-05,
+      "loss": 0.465,
+      "step": 1614
+    },
+    {
+      "epoch": 0.4654178674351585,
+      "grad_norm": 1.5732845415610626,
+      "learning_rate": 1e-05,
+      "loss": 0.2923,
+      "step": 1615
+    },
+    {
+      "epoch": 0.46570605187319886,
+      "grad_norm": 3.030593908464025,
+      "learning_rate": 1e-05,
+      "loss": 0.4769,
+      "step": 1616
+    },
+    {
+      "epoch": 0.4659942363112392,
+      "grad_norm": 1.6352362506907436,
+      "learning_rate": 1e-05,
+      "loss": 0.4661,
+      "step": 1617
+    },
+    {
+      "epoch": 0.46628242074927956,
+      "grad_norm": 1.6462901234522942,
+      "learning_rate": 1e-05,
+      "loss": 0.2857,
+      "step": 1618
+    },
+    {
+      "epoch": 0.4665706051873199,
+      "grad_norm": 2.305484901094868,
+      "learning_rate": 1e-05,
+      "loss": 0.4336,
+      "step": 1619
+    },
+    {
+      "epoch": 0.4668587896253602,
+      "grad_norm": 2.557398249605915,
+      "learning_rate": 1e-05,
+      "loss": 0.4593,
+      "step": 1620
+    },
+    {
+      "epoch": 0.46714697406340056,
+      "grad_norm": 2.1407733470243153,
+      "learning_rate": 1e-05,
+      "loss": 0.2638,
+      "step": 1621
+    },
+    {
+      "epoch": 0.4674351585014409,
+      "grad_norm": 3.3332545163100904,
+      "learning_rate": 1e-05,
+      "loss": 0.4365,
+      "step": 1622
+    },
+    {
+      "epoch": 0.46772334293948126,
+      "grad_norm": 1.9781546237826027,
+      "learning_rate": 1e-05,
+      "loss": 0.4655,
+      "step": 1623
+    },
+    {
+      "epoch": 0.4680115273775216,
+      "grad_norm": 2.521960806890818,
+      "learning_rate": 1e-05,
+      "loss": 0.2799,
+      "step": 1624
+    },
+    {
+      "epoch": 0.46829971181556196,
+      "grad_norm": 2.510635807760795,
+      "learning_rate": 1e-05,
+      "loss": 0.4609,
+      "step": 1625
+    },
+    {
+      "epoch": 0.4685878962536023,
+      "grad_norm": 1.9715299569973497,
+      "learning_rate": 1e-05,
+      "loss": 0.4747,
+      "step": 1626
+    },
+    {
+      "epoch": 0.46887608069164266,
+      "grad_norm": 1.5624894931516837,
+      "learning_rate": 1e-05,
+      "loss": 0.2825,
+      "step": 1627
+    },
+    {
+      "epoch": 0.469164265129683,
+      "grad_norm": 2.340010305197747,
+      "learning_rate": 1e-05,
+      "loss": 0.4476,
+      "step": 1628
+    },
+    {
+      "epoch": 0.46945244956772336,
+      "grad_norm": 2.181830008102016,
+      "learning_rate": 1e-05,
+      "loss": 0.4608,
+      "step": 1629
+    },
+    {
+      "epoch": 0.4697406340057637,
+      "grad_norm": 2.3829939547019072,
+      "learning_rate": 1e-05,
+      "loss": 0.2766,
+      "step": 1630
+    },
+    {
+      "epoch": 0.470028818443804,
+      "grad_norm": 2.1314038791907612,
+      "learning_rate": 1e-05,
+      "loss": 0.4594,
+      "step": 1631
+    },
+    {
+      "epoch": 0.47031700288184436,
+      "grad_norm": 1.8878699816203297,
+      "learning_rate": 1e-05,
+      "loss": 0.468,
+      "step": 1632
+    },
+    {
+      "epoch": 0.4706051873198847,
+      "grad_norm": 1.4414170428945374,
+      "learning_rate": 1e-05,
+      "loss": 0.2663,
+      "step": 1633
+    },
+    {
+      "epoch": 0.47089337175792506,
+      "grad_norm": 3.1604308673524635,
+      "learning_rate": 1e-05,
+      "loss": 0.4356,
+      "step": 1634
+    },
+    {
+      "epoch": 0.4711815561959654,
+      "grad_norm": 2.393902564518669,
+      "learning_rate": 1e-05,
+      "loss": 0.4609,
+      "step": 1635
+    },
+    {
+      "epoch": 0.47146974063400576,
+      "grad_norm": 2.334488135606776,
+      "learning_rate": 1e-05,
+      "loss": 0.2752,
+      "step": 1636
+    },
+    {
+      "epoch": 0.4717579250720461,
+      "grad_norm": 2.615335064480252,
+      "learning_rate": 1e-05,
+      "loss": 0.4731,
+      "step": 1637
+    },
+    {
+      "epoch": 0.47204610951008646,
+      "grad_norm": 1.9090025764184533,
+      "learning_rate": 1e-05,
+      "loss": 0.4607,
+      "step": 1638
+    },
+    {
+      "epoch": 0.4723342939481268,
+      "grad_norm": 1.5997617443798198,
+      "learning_rate": 1e-05,
+      "loss": 0.2916,
+      "step": 1639
+    },
+    {
+      "epoch": 0.47262247838616717,
+      "grad_norm": 2.8454983572286365,
+      "learning_rate": 1e-05,
+      "loss": 0.4584,
+      "step": 1640
+    },
+    {
+      "epoch": 0.4729106628242075,
+      "grad_norm": 2.415401984530564,
+      "learning_rate": 1e-05,
+      "loss": 0.4604,
+      "step": 1641
+    },
+    {
+      "epoch": 0.4731988472622478,
+      "grad_norm": 2.0629566279765394,
+      "learning_rate": 1e-05,
+      "loss": 0.2785,
+      "step": 1642
+    },
+    {
+      "epoch": 0.47348703170028816,
+      "grad_norm": 2.6809143836691964,
+      "learning_rate": 1e-05,
+      "loss": 0.4633,
+      "step": 1643
+    },
+    {
+      "epoch": 0.4737752161383285,
+      "grad_norm": 1.6133613428462672,
+      "learning_rate": 1e-05,
+      "loss": 0.4627,
+      "step": 1644
+    },
+    {
+      "epoch": 0.47406340057636887,
+      "grad_norm": 1.9295680584512247,
+      "learning_rate": 1e-05,
+      "loss": 0.2943,
+      "step": 1645
+    },
+    {
+      "epoch": 0.4743515850144092,
+      "grad_norm": 2.5630263242961457,
+      "learning_rate": 1e-05,
+      "loss": 0.4583,
+      "step": 1646
+    },
+    {
+      "epoch": 0.47463976945244957,
+      "grad_norm": 2.871829494714822,
+      "learning_rate": 1e-05,
+      "loss": 0.4664,
+      "step": 1647
+    },
+    {
+      "epoch": 0.4749279538904899,
+      "grad_norm": 1.5701051155975592,
+      "learning_rate": 1e-05,
+      "loss": 0.2993,
+      "step": 1648
+    },
+    {
+      "epoch": 0.47521613832853027,
+      "grad_norm": 2.1813422991508493,
+      "learning_rate": 1e-05,
+      "loss": 0.4104,
+      "step": 1649
+    },
+    {
+      "epoch": 0.4755043227665706,
+      "grad_norm": 1.9274624400881828,
+      "learning_rate": 1e-05,
+      "loss": 0.465,
+      "step": 1650
+    },
+    {
+      "epoch": 0.47579250720461097,
+      "grad_norm": 2.188434136825765,
+      "learning_rate": 1e-05,
+      "loss": 0.2711,
+      "step": 1651
+    },
+    {
+      "epoch": 0.4760806916426513,
+      "grad_norm": 2.9606405106504594,
+      "learning_rate": 1e-05,
+      "loss": 0.4851,
+      "step": 1652
+    },
+    {
+      "epoch": 0.4763688760806916,
+      "grad_norm": 1.8623058087282203,
+      "learning_rate": 1e-05,
+      "loss": 0.4622,
+      "step": 1653
+    },
+    {
+      "epoch": 0.47665706051873197,
+      "grad_norm": 1.7217779197591614,
+      "learning_rate": 1e-05,
+      "loss": 0.2973,
+      "step": 1654
+    },
+    {
+      "epoch": 0.4769452449567723,
+      "grad_norm": 4.474901292845818,
+      "learning_rate": 1e-05,
+      "loss": 0.4646,
+      "step": 1655
+    },
+    {
+      "epoch": 0.47723342939481267,
+      "grad_norm": 1.77208449825262,
+      "learning_rate": 1e-05,
+      "loss": 0.4651,
+      "step": 1656
+    },
+    {
+      "epoch": 0.477521613832853,
+      "grad_norm": 1.46297419544384,
+      "learning_rate": 1e-05,
+      "loss": 0.2826,
+      "step": 1657
+    },
+    {
+      "epoch": 0.47780979827089337,
+      "grad_norm": 2.4772530667091313,
+      "learning_rate": 1e-05,
+      "loss": 0.4465,
+      "step": 1658
+    },
+    {
+      "epoch": 0.4780979827089337,
+      "grad_norm": 2.231787991203313,
+      "learning_rate": 1e-05,
+      "loss": 0.4674,
+      "step": 1659
+    },
+    {
+      "epoch": 0.4783861671469741,
+      "grad_norm": 2.0476087840540496,
+      "learning_rate": 1e-05,
+      "loss": 0.2902,
+      "step": 1660
+    },
+    {
+      "epoch": 0.4786743515850144,
+      "grad_norm": 3.351809836897069,
+      "learning_rate": 1e-05,
+      "loss": 0.4738,
+      "step": 1661
+    },
+    {
+      "epoch": 0.4789625360230548,
+      "grad_norm": 1.4230338456345113,
+      "learning_rate": 1e-05,
+      "loss": 0.4641,
+      "step": 1662
+    },
+    {
+      "epoch": 0.4792507204610951,
+      "grad_norm": 2.4454093220425532,
+      "learning_rate": 1e-05,
+      "loss": 0.2929,
+      "step": 1663
+    },
+    {
+      "epoch": 0.4795389048991354,
+      "grad_norm": 2.3791521392022865,
+      "learning_rate": 1e-05,
+      "loss": 0.4637,
+      "step": 1664
+    },
+    {
+      "epoch": 0.47982708933717577,
+      "grad_norm": 1.746024248598935,
+      "learning_rate": 1e-05,
+      "loss": 0.4694,
+      "step": 1665
+    },
+    {
+      "epoch": 0.4801152737752161,
+      "grad_norm": 2.4297254692858217,
+      "learning_rate": 1e-05,
+      "loss": 0.2814,
+      "step": 1666
+    },
+    {
+      "epoch": 0.4804034582132565,
+      "grad_norm": 3.6694270809864418,
+      "learning_rate": 1e-05,
+      "loss": 0.4592,
+      "step": 1667
+    },
+    {
+      "epoch": 0.4806916426512968,
+      "grad_norm": 1.4818977252200454,
+      "learning_rate": 1e-05,
+      "loss": 0.464,
+      "step": 1668
+    },
+    {
+      "epoch": 0.4809798270893372,
+      "grad_norm": 2.401649865474904,
+      "learning_rate": 1e-05,
+      "loss": 0.2726,
+      "step": 1669
+    },
+    {
+      "epoch": 0.4812680115273775,
+      "grad_norm": 2.8556667154383093,
+      "learning_rate": 1e-05,
+      "loss": 0.4213,
+      "step": 1670
+    },
+    {
+      "epoch": 0.4815561959654179,
+      "grad_norm": 1.705405878147625,
+      "learning_rate": 1e-05,
+      "loss": 0.4577,
+      "step": 1671
+    },
+    {
+      "epoch": 0.48184438040345823,
+      "grad_norm": 1.5484368382923146,
+      "learning_rate": 1e-05,
+      "loss": 0.286,
+      "step": 1672
+    },
+    {
+      "epoch": 0.4821325648414986,
+      "grad_norm": 2.325960044809156,
+      "learning_rate": 1e-05,
+      "loss": 0.4655,
+      "step": 1673
+    },
+    {
+      "epoch": 0.48242074927953893,
+      "grad_norm": 2.377847976610141,
+      "learning_rate": 1e-05,
+      "loss": 0.4642,
+      "step": 1674
+    },
+    {
+      "epoch": 0.4827089337175792,
+      "grad_norm": 2.880226609097058,
+      "learning_rate": 1e-05,
+      "loss": 0.2817,
+      "step": 1675
+    },
+    {
+      "epoch": 0.4829971181556196,
+      "grad_norm": 2.3133838506031754,
+      "learning_rate": 1e-05,
+      "loss": 0.4739,
+      "step": 1676
+    },
+    {
+      "epoch": 0.4832853025936599,
+      "grad_norm": 1.5933475041603575,
+      "learning_rate": 1e-05,
+      "loss": 0.4686,
+      "step": 1677
+    },
+    {
+      "epoch": 0.4835734870317003,
+      "grad_norm": 2.3560075097273594,
+      "learning_rate": 1e-05,
+      "loss": 0.2734,
+      "step": 1678
+    },
+    {
+      "epoch": 0.48386167146974063,
+      "grad_norm": 3.2853984031850767,
+      "learning_rate": 1e-05,
+      "loss": 0.4232,
+      "step": 1679
+    },
+    {
+      "epoch": 0.484149855907781,
+      "grad_norm": 1.7081589209357984,
+      "learning_rate": 1e-05,
+      "loss": 0.4635,
+      "step": 1680
+    },
+    {
+      "epoch": 0.48443804034582133,
+      "grad_norm": 2.393075804247037,
+      "learning_rate": 1e-05,
+      "loss": 0.2859,
+      "step": 1681
+    },
+    {
+      "epoch": 0.4847262247838617,
+      "grad_norm": 3.529738877729882,
+      "learning_rate": 1e-05,
+      "loss": 0.4492,
+      "step": 1682
+    },
+    {
+      "epoch": 0.48501440922190203,
+      "grad_norm": 1.9093370333183477,
+      "learning_rate": 1e-05,
+      "loss": 0.45,
+      "step": 1683
+    },
+    {
+      "epoch": 0.4853025936599424,
+      "grad_norm": 1.7693072755843469,
+      "learning_rate": 1e-05,
+      "loss": 0.263,
+      "step": 1684
+    },
+    {
+      "epoch": 0.48559077809798273,
+      "grad_norm": 4.568922047885823,
+      "learning_rate": 1e-05,
+      "loss": 0.4148,
+      "step": 1685
+    },
+    {
+      "epoch": 0.48587896253602303,
+      "grad_norm": 2.134908794525408,
+      "learning_rate": 1e-05,
+      "loss": 0.4702,
+      "step": 1686
+    },
+    {
+      "epoch": 0.4861671469740634,
+      "grad_norm": 1.439007644118069,
+      "learning_rate": 1e-05,
+      "loss": 0.2985,
+      "step": 1687
+    },
+    {
+      "epoch": 0.48645533141210373,
+      "grad_norm": 3.3774076129624273,
+      "learning_rate": 1e-05,
+      "loss": 0.4384,
+      "step": 1688
+    },
+    {
+      "epoch": 0.4867435158501441,
+      "grad_norm": 1.915779414442965,
+      "learning_rate": 1e-05,
+      "loss": 0.4735,
+      "step": 1689
+    },
+    {
+      "epoch": 0.48703170028818443,
+      "grad_norm": 1.4489762424037487,
+      "learning_rate": 1e-05,
+      "loss": 0.2735,
+      "step": 1690
+    },
+    {
+      "epoch": 0.4873198847262248,
+      "grad_norm": 2.950035274296331,
+      "learning_rate": 1e-05,
+      "loss": 0.437,
+      "step": 1691
+    },
+    {
+      "epoch": 0.48760806916426513,
+      "grad_norm": 1.8952923358414893,
+      "learning_rate": 1e-05,
+      "loss": 0.4547,
+      "step": 1692
+    },
+    {
+      "epoch": 0.4878962536023055,
+      "grad_norm": 1.4305584083910485,
+      "learning_rate": 1e-05,
+      "loss": 0.2796,
+      "step": 1693
+    },
+    {
+      "epoch": 0.48818443804034584,
+      "grad_norm": 2.4632837605887734,
+      "learning_rate": 1e-05,
+      "loss": 0.4292,
+      "step": 1694
+    },
+    {
+      "epoch": 0.4884726224783862,
+      "grad_norm": 1.3955015462464664,
+      "learning_rate": 1e-05,
+      "loss": 0.4585,
+      "step": 1695
+    },
+    {
+      "epoch": 0.48876080691642654,
+      "grad_norm": 2.2759160277281127,
+      "learning_rate": 1e-05,
+      "loss": 0.2604,
+      "step": 1696
+    },
+    {
+      "epoch": 0.48904899135446683,
+      "grad_norm": 3.0743248862314223,
+      "learning_rate": 1e-05,
+      "loss": 0.4332,
+      "step": 1697
+    },
+    {
+      "epoch": 0.4893371757925072,
+      "grad_norm": 1.4239389086894434,
+      "learning_rate": 1e-05,
+      "loss": 0.4519,
+      "step": 1698
+    },
+    {
+      "epoch": 0.48962536023054753,
+      "grad_norm": 1.3716393538929104,
+      "learning_rate": 1e-05,
+      "loss": 0.2514,
+      "step": 1699
+    },
+    {
+      "epoch": 0.4899135446685879,
+      "grad_norm": 2.363439394993871,
+      "learning_rate": 1e-05,
+      "loss": 0.4481,
+      "step": 1700
+    },
+    {
+      "epoch": 0.49020172910662824,
+      "grad_norm": 1.9900927483142574,
+      "learning_rate": 1e-05,
+      "loss": 0.4683,
+      "step": 1701
+    },
+    {
+      "epoch": 0.4904899135446686,
+      "grad_norm": 1.8185424866723199,
+      "learning_rate": 1e-05,
+      "loss": 0.2972,
+      "step": 1702
+    },
+    {
+      "epoch": 0.49077809798270894,
+      "grad_norm": 3.3545570355044085,
+      "learning_rate": 1e-05,
+      "loss": 0.4683,
+      "step": 1703
+    },
+    {
+      "epoch": 0.4910662824207493,
+      "grad_norm": 1.6559295493133095,
+      "learning_rate": 1e-05,
+      "loss": 0.4598,
+      "step": 1704
+    },
+    {
+      "epoch": 0.49135446685878964,
+      "grad_norm": 1.4775431660851086,
+      "learning_rate": 1e-05,
+      "loss": 0.2662,
+      "step": 1705
+    },
+    {
+      "epoch": 0.49164265129683,
+      "grad_norm": 3.486095927001384,
+      "learning_rate": 1e-05,
+      "loss": 0.4906,
+      "step": 1706
+    },
+    {
+      "epoch": 0.49193083573487034,
+      "grad_norm": 2.219624772953442,
+      "learning_rate": 1e-05,
+      "loss": 0.4599,
+      "step": 1707
+    },
+    {
+      "epoch": 0.49221902017291064,
+      "grad_norm": 1.9844505305072102,
+      "learning_rate": 1e-05,
+      "loss": 0.2641,
+      "step": 1708
+    },
+    {
+      "epoch": 0.492507204610951,
+      "grad_norm": 2.820281651509339,
+      "learning_rate": 1e-05,
+      "loss": 0.4751,
+      "step": 1709
+    },
+    {
+      "epoch": 0.49279538904899134,
+      "grad_norm": 1.8429376323004725,
+      "learning_rate": 1e-05,
+      "loss": 0.4615,
+      "step": 1710
+    },
+    {
+      "epoch": 0.4930835734870317,
+      "grad_norm": 1.876894295065116,
+      "learning_rate": 1e-05,
+      "loss": 0.2637,
+      "step": 1711
+    },
+    {
+      "epoch": 0.49337175792507204,
+      "grad_norm": 3.6828539449790103,
+      "learning_rate": 1e-05,
+      "loss": 0.4508,
+      "step": 1712
+    },
+    {
+      "epoch": 0.4936599423631124,
+      "grad_norm": 1.49070007978551,
+      "learning_rate": 1e-05,
+      "loss": 0.4665,
+      "step": 1713
+    },
+    {
+      "epoch": 0.49394812680115274,
+      "grad_norm": 1.8332686252713897,
+      "learning_rate": 1e-05,
+      "loss": 0.2685,
+      "step": 1714
+    },
+    {
+      "epoch": 0.4942363112391931,
+      "grad_norm": 3.653039605835981,
+      "learning_rate": 1e-05,
+      "loss": 0.4509,
+      "step": 1715
+    },
+    {
+      "epoch": 0.49452449567723344,
+      "grad_norm": 2.2361933476641727,
+      "learning_rate": 1e-05,
+      "loss": 0.4678,
+      "step": 1716
+    },
+    {
+      "epoch": 0.4948126801152738,
+      "grad_norm": 2.2198315477674417,
+      "learning_rate": 1e-05,
+      "loss": 0.2864,
+      "step": 1717
+    },
+    {
+      "epoch": 0.49510086455331415,
+      "grad_norm": 2.9662040776422613,
+      "learning_rate": 1e-05,
+      "loss": 0.4596,
+      "step": 1718
+    },
+    {
+      "epoch": 0.49538904899135444,
+      "grad_norm": 1.8533439742640203,
+      "learning_rate": 1e-05,
+      "loss": 0.4601,
+      "step": 1719
+    },
+    {
+      "epoch": 0.4956772334293948,
+      "grad_norm": 2.05752974142649,
+      "learning_rate": 1e-05,
+      "loss": 0.2892,
+      "step": 1720
+    },
+    {
+      "epoch": 0.49596541786743514,
+      "grad_norm": 2.5147899712900736,
+      "learning_rate": 1e-05,
+      "loss": 0.4467,
+      "step": 1721
+    },
+    {
+      "epoch": 0.4962536023054755,
+      "grad_norm": 1.4329665661640703,
+      "learning_rate": 1e-05,
+      "loss": 0.4569,
+      "step": 1722
+    },
+    {
+      "epoch": 0.49654178674351584,
+      "grad_norm": 1.9863338498032903,
+      "learning_rate": 1e-05,
+      "loss": 0.2768,
+      "step": 1723
+    },
+    {
+      "epoch": 0.4968299711815562,
+      "grad_norm": 2.7689961736244513,
+      "learning_rate": 1e-05,
+      "loss": 0.4393,
+      "step": 1724
+    },
+    {
+      "epoch": 0.49711815561959655,
+      "grad_norm": 1.863142892863334,
+      "learning_rate": 1e-05,
+      "loss": 0.4576,
+      "step": 1725
+    },
+    {
+      "epoch": 0.4974063400576369,
+      "grad_norm": 1.749129294111609,
+      "learning_rate": 1e-05,
+      "loss": 0.2481,
+      "step": 1726
+    },
+    {
+      "epoch": 0.49769452449567725,
+      "grad_norm": 2.4098307488706117,
+      "learning_rate": 1e-05,
+      "loss": 0.4352,
+      "step": 1727
+    },
+    {
+      "epoch": 0.4979827089337176,
+      "grad_norm": 1.7004384919576656,
+      "learning_rate": 1e-05,
+      "loss": 0.4631,
+      "step": 1728
+    },
+    {
+      "epoch": 0.49827089337175795,
+      "grad_norm": 1.6741308399940007,
+      "learning_rate": 1e-05,
+      "loss": 0.2644,
+      "step": 1729
+    },
+    {
+      "epoch": 0.49855907780979825,
+      "grad_norm": 2.2932672534152436,
+      "learning_rate": 1e-05,
+      "loss": 0.4367,
+      "step": 1730
+    },
+    {
+      "epoch": 0.4988472622478386,
+      "grad_norm": 2.6349075799650565,
+      "learning_rate": 1e-05,
+      "loss": 0.4599,
+      "step": 1731
+    },
+    {
+      "epoch": 0.49913544668587895,
+      "grad_norm": 1.8889506861333187,
+      "learning_rate": 1e-05,
+      "loss": 0.2671,
+      "step": 1732
+    },
+    {
+      "epoch": 0.4994236311239193,
+      "grad_norm": 2.6226440853671735,
+      "learning_rate": 1e-05,
+      "loss": 0.465,
+      "step": 1733
+    },
+    {
+      "epoch": 0.49971181556195965,
+      "grad_norm": 1.7201450504469826,
+      "learning_rate": 1e-05,
+      "loss": 0.4531,
+      "step": 1734
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2914202968030986,
+      "learning_rate": 1e-05,
+      "loss": 0.2772,
+      "step": 1735
+    },
+    {
+      "epoch": 0.5002881844380404,
+      "grad_norm": 2.50152200887206,
+      "learning_rate": 1e-05,
+      "loss": 0.4495,
+      "step": 1736
+    },
+    {
+      "epoch": 0.5005763688760807,
+      "grad_norm": 1.693806283768107,
+      "learning_rate": 1e-05,
+      "loss": 0.4638,
+      "step": 1737
+    },
+    {
+      "epoch": 0.500864553314121,
+      "grad_norm": 1.9028704339010702,
+      "learning_rate": 1e-05,
+      "loss": 0.3013,
+      "step": 1738
+    },
+    {
+      "epoch": 0.5011527377521614,
+      "grad_norm": 2.5933183846887995,
+      "learning_rate": 1e-05,
+      "loss": 0.4828,
+      "step": 1739
+    },
+    {
+      "epoch": 0.5014409221902018,
+      "grad_norm": 2.0412542395277455,
+      "learning_rate": 1e-05,
+      "loss": 0.4587,
+      "step": 1740
+    },
+    {
+      "epoch": 0.5017291066282421,
+      "grad_norm": 1.9856149329061437,
+      "learning_rate": 1e-05,
+      "loss": 0.278,
+      "step": 1741
+    },
+    {
+      "epoch": 0.5020172910662825,
+      "grad_norm": 2.4016386267429217,
+      "learning_rate": 1e-05,
+      "loss": 0.4721,
+      "step": 1742
+    },
+    {
+      "epoch": 0.5023054755043228,
+      "grad_norm": 2.3613869701064427,
+      "learning_rate": 1e-05,
+      "loss": 0.4709,
+      "step": 1743
+    },
+    {
+      "epoch": 0.5025936599423632,
+      "grad_norm": 1.8895381195885375,
+      "learning_rate": 1e-05,
+      "loss": 0.2667,
+      "step": 1744
+    },
+    {
+      "epoch": 0.5028818443804035,
+      "grad_norm": 2.519056644274537,
+      "learning_rate": 1e-05,
+      "loss": 0.4173,
+      "step": 1745
+    },
+    {
+      "epoch": 0.5031700288184437,
+      "grad_norm": 1.597784148194564,
+      "learning_rate": 1e-05,
+      "loss": 0.4478,
+      "step": 1746
+    },
+    {
+      "epoch": 0.5034582132564841,
+      "grad_norm": 1.8673531027230266,
+      "learning_rate": 1e-05,
+      "loss": 0.264,
+      "step": 1747
+    },
+    {
+      "epoch": 0.5037463976945245,
+      "grad_norm": 2.9027343527283103,
+      "learning_rate": 1e-05,
+      "loss": 0.4356,
+      "step": 1748
+    },
+    {
+      "epoch": 0.5040345821325648,
+      "grad_norm": 3.034671484976285,
+      "learning_rate": 1e-05,
+      "loss": 0.4664,
+      "step": 1749
+    },
+    {
+      "epoch": 0.5043227665706052,
+      "grad_norm": 2.7764727211021434,
+      "learning_rate": 1e-05,
+      "loss": 0.2722,
+      "step": 1750
+    },
+    {
+      "epoch": 0.5043227665706052,
+      "eval_websight_new_IoU": 0.019636476412415504,
+      "eval_websight_new_MAE_x": 286.760009765625,
+      "eval_websight_new_MAE_y": 594.97998046875,
+      "eval_websight_new_coord_probability": 0.7459741830825806,
+      "eval_websight_new_correct_token_probability": 0.0011280778562650084,
+      "eval_websight_new_inside_bbox": 0.019999999552965164,
+      "eval_websight_new_loss": 0.2900213897228241,
+      "eval_websight_new_range_100_probability": 0.09146662056446075,
+      "eval_websight_new_range_20_probability": 0.015594979748129845,
+      "eval_websight_new_range_5_probability": 0.0030590335372835398,
+      "eval_websight_new_runtime": 33.7773,
+      "eval_websight_new_samples_per_second": 1.48,
+      "eval_websight_new_steps_per_second": 0.207,
+      "step": 1750
+    },
+    {
+      "epoch": 0.5043227665706052,
+      "eval_seeclick_IoU": 0.034775398671627045,
+      "eval_seeclick_MAE_x": 473.1000061035156,
+      "eval_seeclick_MAE_y": 505.5199890136719,
+      "eval_seeclick_coord_probability": 0.7473167181015015,
+      "eval_seeclick_correct_token_probability": 0.0006066637579351664,
+      "eval_seeclick_inside_bbox": 0.05999999865889549,
+      "eval_seeclick_loss": 0.38515812158584595,
+      "eval_seeclick_range_100_probability": 0.13240179419517517,
+      "eval_seeclick_range_20_probability": 0.05072963237762451,
+      "eval_seeclick_range_5_probability": 0.007617821916937828,
+      "eval_seeclick_runtime": 42.4492,
+      "eval_seeclick_samples_per_second": 1.178,
+      "eval_seeclick_steps_per_second": 0.165,
+      "step": 1750
+    },
+    {
+      "epoch": 0.5043227665706052,
+      "eval_icons_IoU": 0.0025072256103157997,
+      "eval_icons_MAE_x": 473.5400085449219,
+      "eval_icons_MAE_y": 546.8800048828125,
+      "eval_icons_coord_probability": 0.7476227283477783,
+      "eval_icons_correct_token_probability": 0.00905266497284174,
+      "eval_icons_inside_bbox": 0.0,
+      "eval_icons_loss": 0.4429156482219696,
+      "eval_icons_range_100_probability": 0.276961088180542,
+      "eval_icons_range_20_probability": 0.09574487060308456,
+      "eval_icons_range_5_probability": 0.03115815669298172,
+      "eval_icons_runtime": 43.2189,
+      "eval_icons_samples_per_second": 1.157,
+      "eval_icons_steps_per_second": 0.162,
+      "step": 1750
+    },
+    {
+      "epoch": 0.5046109510086455,
+      "grad_norm": 3.9831774379628264,
+      "learning_rate": 1e-05,
+      "loss": 0.4618,
+      "step": 1751
+    },
+    {
+      "epoch": 0.5048991354466859,
+      "grad_norm": 1.4374301573776382,
+      "learning_rate": 1e-05,
+      "loss": 0.4568,
+      "step": 1752
+    },
+    {
+      "epoch": 0.5051873198847262,
+      "grad_norm": 1.8921007186490801,
+      "learning_rate": 1e-05,
+      "loss": 0.2785,
+      "step": 1753
+    },
+    {
+      "epoch": 0.5054755043227666,
+      "grad_norm": 2.615752425693824,
+      "learning_rate": 1e-05,
+      "loss": 0.477,
+      "step": 1754
+    },
+    {
+      "epoch": 0.5057636887608069,
+      "grad_norm": 1.533452814309213,
+      "learning_rate": 1e-05,
+      "loss": 0.4545,
+      "step": 1755
+    },
+    {
+      "epoch": 0.5060518731988473,
+      "grad_norm": 1.9957275008919058,
+      "learning_rate": 1e-05,
+      "loss": 0.2784,
+      "step": 1756
+    },
+    {
+      "epoch": 0.5063400576368876,
+      "grad_norm": 3.1653940246653987,
+      "learning_rate": 1e-05,
+      "loss": 0.4532,
+      "step": 1757
+    },
+    {
+      "epoch": 0.506628242074928,
+      "grad_norm": 1.407081682029088,
+      "learning_rate": 1e-05,
+      "loss": 0.4559,
+      "step": 1758
+    },
+    {
+      "epoch": 0.5069164265129683,
+      "grad_norm": 1.5261723664185132,
+      "learning_rate": 1e-05,
+      "loss": 0.2775,
+      "step": 1759
+    },
+    {
+      "epoch": 0.5072046109510087,
+      "grad_norm": 2.8459742883006585,
+      "learning_rate": 1e-05,
+      "loss": 0.4497,
+      "step": 1760
+    },
+    {
+      "epoch": 0.507492795389049,
+      "grad_norm": 1.52626873226206,
+      "learning_rate": 1e-05,
+      "loss": 0.4485,
+      "step": 1761
+    },
+    {
+      "epoch": 0.5077809798270894,
+      "grad_norm": 2.120607391919743,
+      "learning_rate": 1e-05,
+      "loss": 0.2781,
+      "step": 1762
+    },
+    {
+      "epoch": 0.5080691642651297,
+      "grad_norm": 2.4977431110147483,
+      "learning_rate": 1e-05,
+      "loss": 0.4563,
+      "step": 1763
+    },
+    {
+      "epoch": 0.5083573487031701,
+      "grad_norm": 2.0923075056563833,
+      "learning_rate": 1e-05,
+      "loss": 0.4692,
+      "step": 1764
+    },
+    {
+      "epoch": 0.5086455331412104,
+      "grad_norm": 1.6577734390908576,
+      "learning_rate": 1e-05,
+      "loss": 0.2685,
+      "step": 1765
+    },
+    {
+      "epoch": 0.5089337175792508,
+      "grad_norm": 3.4894389660396223,
+      "learning_rate": 1e-05,
+      "loss": 0.4617,
+      "step": 1766
+    },
+    {
+      "epoch": 0.5092219020172911,
+      "grad_norm": 1.7791014605602602,
+      "learning_rate": 1e-05,
+      "loss": 0.4646,
+      "step": 1767
+    },
+    {
+      "epoch": 0.5095100864553314,
+      "grad_norm": 1.7600775080832534,
+      "learning_rate": 1e-05,
+      "loss": 0.2607,
+      "step": 1768
+    },
+    {
+      "epoch": 0.5097982708933717,
+      "grad_norm": 3.635526592272463,
+      "learning_rate": 1e-05,
+      "loss": 0.4646,
+      "step": 1769
+    },
+    {
+      "epoch": 0.5100864553314121,
+      "grad_norm": 1.4535745053954856,
+      "learning_rate": 1e-05,
+      "loss": 0.4504,
+      "step": 1770
+    },
+    {
+      "epoch": 0.5103746397694524,
+      "grad_norm": 1.9015723045727277,
+      "learning_rate": 1e-05,
+      "loss": 0.2665,
+      "step": 1771
+    },
+    {
+      "epoch": 0.5106628242074928,
+      "grad_norm": 2.7503315131178994,
+      "learning_rate": 1e-05,
+      "loss": 0.445,
+      "step": 1772
+    },
+    {
+      "epoch": 0.5109510086455331,
+      "grad_norm": 1.7947410749054482,
+      "learning_rate": 1e-05,
+      "loss": 0.4552,
+      "step": 1773
+    },
+    {
+      "epoch": 0.5112391930835735,
+      "grad_norm": 2.0285048593527675,
+      "learning_rate": 1e-05,
+      "loss": 0.2817,
+      "step": 1774
+    },
+    {
+      "epoch": 0.5115273775216138,
+      "grad_norm": 4.085655591407628,
+      "learning_rate": 1e-05,
+      "loss": 0.4605,
+      "step": 1775
+    },
+    {
+      "epoch": 0.5118155619596542,
+      "grad_norm": 1.409100641755321,
+      "learning_rate": 1e-05,
+      "loss": 0.4568,
+      "step": 1776
+    },
+    {
+      "epoch": 0.5121037463976945,
+      "grad_norm": 2.450601632694455,
+      "learning_rate": 1e-05,
+      "loss": 0.2464,
+      "step": 1777
+    },
+    {
+      "epoch": 0.5123919308357349,
+      "grad_norm": 2.6694118572635075,
+      "learning_rate": 1e-05,
+      "loss": 0.4843,
+      "step": 1778
+    },
+    {
+      "epoch": 0.5126801152737752,
+      "grad_norm": 2.68944159707934,
+      "learning_rate": 1e-05,
+      "loss": 0.4542,
+      "step": 1779
+    },
+    {
+      "epoch": 0.5129682997118156,
+      "grad_norm": 1.4155636999897319,
+      "learning_rate": 1e-05,
+      "loss": 0.2684,
+      "step": 1780
+    },
+    {
+      "epoch": 0.5132564841498559,
+      "grad_norm": 3.014353805737626,
+      "learning_rate": 1e-05,
+      "loss": 0.4892,
+      "step": 1781
+    },
+    {
+      "epoch": 0.5135446685878963,
+      "grad_norm": 1.4594450185730636,
+      "learning_rate": 1e-05,
+      "loss": 0.4457,
+      "step": 1782
+    },
+    {
+      "epoch": 0.5138328530259366,
+      "grad_norm": 2.008615649097747,
+      "learning_rate": 1e-05,
+      "loss": 0.2421,
+      "step": 1783
+    },
+    {
+      "epoch": 0.514121037463977,
+      "grad_norm": 3.3669893461047993,
+      "learning_rate": 1e-05,
+      "loss": 0.4537,
+      "step": 1784
+    },
+    {
+      "epoch": 0.5144092219020173,
+      "grad_norm": 2.8345861052522427,
+      "learning_rate": 1e-05,
+      "loss": 0.4673,
+      "step": 1785
+    },
+    {
+      "epoch": 0.5146974063400577,
+      "grad_norm": 1.323601890023609,
+      "learning_rate": 1e-05,
+      "loss": 0.2613,
+      "step": 1786
+    },
+    {
+      "epoch": 0.514985590778098,
+      "grad_norm": 2.74391568395978,
+      "learning_rate": 1e-05,
+      "loss": 0.464,
+      "step": 1787
+    },
+    {
+      "epoch": 0.5152737752161384,
+      "grad_norm": 1.5384663730957742,
+      "learning_rate": 1e-05,
+      "loss": 0.4502,
+      "step": 1788
+    },
+    {
+      "epoch": 0.5155619596541787,
+      "grad_norm": 2.276530745494601,
+      "learning_rate": 1e-05,
+      "loss": 0.2797,
+      "step": 1789
+    },
+    {
+      "epoch": 0.515850144092219,
+      "grad_norm": 3.2809096257420007,
+      "learning_rate": 1e-05,
+      "loss": 0.4668,
+      "step": 1790
+    },
+    {
+      "epoch": 0.5161383285302593,
+      "grad_norm": 2.230621188973937,
+      "learning_rate": 1e-05,
+      "loss": 0.4692,
+      "step": 1791
+    },
+    {
+      "epoch": 0.5164265129682997,
+      "grad_norm": 1.3677062826287025,
+      "learning_rate": 1e-05,
+      "loss": 0.267,
+      "step": 1792
+    },
+    {
+      "epoch": 0.51671469740634,
+      "grad_norm": 2.6172921544831285,
+      "learning_rate": 1e-05,
+      "loss": 0.4505,
+      "step": 1793
+    },
+    {
+      "epoch": 0.5170028818443804,
+      "grad_norm": 1.4649113503607931,
+      "learning_rate": 1e-05,
+      "loss": 0.4589,
+      "step": 1794
+    },
+    {
+      "epoch": 0.5172910662824207,
+      "grad_norm": 2.683917428362708,
+      "learning_rate": 1e-05,
+      "loss": 0.2774,
+      "step": 1795
+    },
+    {
+      "epoch": 0.5175792507204611,
+      "grad_norm": 3.110624395145058,
+      "learning_rate": 1e-05,
+      "loss": 0.4531,
+      "step": 1796
+    },
+    {
+      "epoch": 0.5178674351585014,
+      "grad_norm": 1.8149679816302997,
+      "learning_rate": 1e-05,
+      "loss": 0.4653,
+      "step": 1797
+    },
+    {
+      "epoch": 0.5181556195965418,
+      "grad_norm": 1.5170431570832181,
+      "learning_rate": 1e-05,
+      "loss": 0.258,
+      "step": 1798
+    },
+    {
+      "epoch": 0.5184438040345821,
+      "grad_norm": 3.2083316155370487,
+      "learning_rate": 1e-05,
+      "loss": 0.4573,
+      "step": 1799
+    },
+    {
+      "epoch": 0.5187319884726225,
+      "grad_norm": 2.014435392735327,
+      "learning_rate": 1e-05,
+      "loss": 0.4509,
+      "step": 1800
+    },
+    {
+      "epoch": 0.5190201729106628,
+      "grad_norm": 2.700097817195559,
+      "learning_rate": 1e-05,
+      "loss": 0.2678,
+      "step": 1801
+    },
+    {
+      "epoch": 0.5193083573487032,
+      "grad_norm": 3.5033374580324566,
+      "learning_rate": 1e-05,
+      "loss": 0.4434,
+      "step": 1802
+    },
+    {
+      "epoch": 0.5195965417867435,
+      "grad_norm": 1.4833570072957702,
+      "learning_rate": 1e-05,
+      "loss": 0.4629,
+      "step": 1803
+    },
+    {
+      "epoch": 0.5198847262247839,
+      "grad_norm": 2.2591788072257843,
+      "learning_rate": 1e-05,
+      "loss": 0.2993,
+      "step": 1804
+    },
+    {
+      "epoch": 0.5201729106628242,
+      "grad_norm": 3.3331176425439524,
+      "learning_rate": 1e-05,
+      "loss": 0.4366,
+      "step": 1805
+    },
+    {
+      "epoch": 0.5204610951008646,
+      "grad_norm": 1.8628877406825821,
+      "learning_rate": 1e-05,
+      "loss": 0.4607,
+      "step": 1806
+    },
+    {
+      "epoch": 0.5207492795389049,
+      "grad_norm": 1.8380156721674201,
+      "learning_rate": 1e-05,
+      "loss": 0.2554,
+      "step": 1807
+    },
+    {
+      "epoch": 0.5210374639769453,
+      "grad_norm": 3.0873032711346275,
+      "learning_rate": 1e-05,
+      "loss": 0.4712,
+      "step": 1808
+    },
+    {
+      "epoch": 0.5213256484149856,
+      "grad_norm": 1.6313365237093287,
+      "learning_rate": 1e-05,
+      "loss": 0.4621,
+      "step": 1809
+    },
+    {
+      "epoch": 0.521613832853026,
+      "grad_norm": 2.4406031488855664,
+      "learning_rate": 1e-05,
+      "loss": 0.2919,
+      "step": 1810
+    },
+    {
+      "epoch": 0.5219020172910663,
+      "grad_norm": 2.741731552872433,
+      "learning_rate": 1e-05,
+      "loss": 0.4344,
+      "step": 1811
+    },
+    {
+      "epoch": 0.5221902017291066,
+      "grad_norm": 1.5711327264546315,
+      "learning_rate": 1e-05,
+      "loss": 0.4532,
+      "step": 1812
+    },
+    {
+      "epoch": 0.5224783861671469,
+      "grad_norm": 1.6069010910310144,
+      "learning_rate": 1e-05,
+      "loss": 0.2954,
+      "step": 1813
+    },
+    {
+      "epoch": 0.5227665706051873,
+      "grad_norm": 2.448133217933075,
+      "learning_rate": 1e-05,
+      "loss": 0.4851,
+      "step": 1814
+    },
+    {
+      "epoch": 0.5230547550432276,
+      "grad_norm": 1.9430375921496603,
+      "learning_rate": 1e-05,
+      "loss": 0.459,
+      "step": 1815
+    },
+    {
+      "epoch": 0.523342939481268,
+      "grad_norm": 2.8756318716673293,
+      "learning_rate": 1e-05,
+      "loss": 0.2794,
+      "step": 1816
+    },
+    {
+      "epoch": 0.5236311239193083,
+      "grad_norm": 2.933961020242428,
+      "learning_rate": 1e-05,
+      "loss": 0.4325,
+      "step": 1817
+    },
+    {
+      "epoch": 0.5239193083573487,
+      "grad_norm": 1.5475470865668899,
+      "learning_rate": 1e-05,
+      "loss": 0.4658,
+      "step": 1818
+    },
+    {
+      "epoch": 0.524207492795389,
+      "grad_norm": 2.0619442989106505,
+      "learning_rate": 1e-05,
+      "loss": 0.2823,
+      "step": 1819
+    },
+    {
+      "epoch": 0.5244956772334294,
+      "grad_norm": 2.7536751220796605,
+      "learning_rate": 1e-05,
+      "loss": 0.4881,
+      "step": 1820
+    },
+    {
+      "epoch": 0.5247838616714697,
+      "grad_norm": 1.8208384751595372,
+      "learning_rate": 1e-05,
+      "loss": 0.4595,
+      "step": 1821
+    },
+    {
+      "epoch": 0.5250720461095101,
+      "grad_norm": 2.217366030697359,
+      "learning_rate": 1e-05,
+      "loss": 0.27,
+      "step": 1822
+    },
+    {
+      "epoch": 0.5253602305475504,
+      "grad_norm": 3.216314333500646,
+      "learning_rate": 1e-05,
+      "loss": 0.4161,
+      "step": 1823
+    },
+    {
+      "epoch": 0.5256484149855908,
+      "grad_norm": 2.680423152848142,
+      "learning_rate": 1e-05,
+      "loss": 0.4627,
+      "step": 1824
+    },
+    {
+      "epoch": 0.5259365994236311,
+      "grad_norm": 1.8212022253223754,
+      "learning_rate": 1e-05,
+      "loss": 0.2813,
+      "step": 1825
+    },
+    {
+      "epoch": 0.5262247838616715,
+      "grad_norm": 3.6904326994867915,
+      "learning_rate": 1e-05,
+      "loss": 0.46,
+      "step": 1826
+    },
+    {
+      "epoch": 0.5265129682997118,
+      "grad_norm": 1.874899690129343,
+      "learning_rate": 1e-05,
+      "loss": 0.4453,
+      "step": 1827
+    },
+    {
+      "epoch": 0.5268011527377522,
+      "grad_norm": 1.5384880091568887,
+      "learning_rate": 1e-05,
+      "loss": 0.258,
+      "step": 1828
+    },
+    {
+      "epoch": 0.5270893371757925,
+      "grad_norm": 2.8060985419907145,
+      "learning_rate": 1e-05,
+      "loss": 0.483,
+      "step": 1829
+    },
+    {
+      "epoch": 0.5273775216138329,
+      "grad_norm": 1.9478776801876367,
+      "learning_rate": 1e-05,
+      "loss": 0.4598,
+      "step": 1830
+    },
+    {
+      "epoch": 0.5276657060518732,
+      "grad_norm": 1.780793099901807,
+      "learning_rate": 1e-05,
+      "loss": 0.2834,
+      "step": 1831
+    },
+    {
+      "epoch": 0.5279538904899136,
+      "grad_norm": 3.0492293138847093,
+      "learning_rate": 1e-05,
+      "loss": 0.4545,
+      "step": 1832
+    },
+    {
+      "epoch": 0.5282420749279539,
+      "grad_norm": 1.4573449451227571,
+      "learning_rate": 1e-05,
+      "loss": 0.4484,
+      "step": 1833
+    },
+    {
+      "epoch": 0.5285302593659942,
+      "grad_norm": 2.213340764443074,
+      "learning_rate": 1e-05,
+      "loss": 0.2808,
+      "step": 1834
+    },
+    {
+      "epoch": 0.5288184438040345,
+      "grad_norm": 2.877377727558519,
+      "learning_rate": 1e-05,
+      "loss": 0.4732,
+      "step": 1835
+    },
+    {
+      "epoch": 0.5291066282420749,
+      "grad_norm": 1.6714668937969217,
+      "learning_rate": 1e-05,
+      "loss": 0.4585,
+      "step": 1836
+    },
+    {
+      "epoch": 0.5293948126801152,
+      "grad_norm": 1.9132058027926757,
+      "learning_rate": 1e-05,
+      "loss": 0.2797,
+      "step": 1837
+    },
+    {
+      "epoch": 0.5296829971181556,
+      "grad_norm": 3.1423187466760276,
+      "learning_rate": 1e-05,
+      "loss": 0.4523,
+      "step": 1838
+    },
+    {
+      "epoch": 0.5299711815561959,
+      "grad_norm": 1.6421478176952211,
+      "learning_rate": 1e-05,
+      "loss": 0.4609,
+      "step": 1839
+    },
+    {
+      "epoch": 0.5302593659942363,
+      "grad_norm": 2.285409906363161,
+      "learning_rate": 1e-05,
+      "loss": 0.2752,
+      "step": 1840
+    },
+    {
+      "epoch": 0.5305475504322766,
+      "grad_norm": 3.089328294282285,
+      "learning_rate": 1e-05,
+      "loss": 0.4611,
+      "step": 1841
+    },
+    {
+      "epoch": 0.530835734870317,
+      "grad_norm": 2.53730982001956,
+      "learning_rate": 1e-05,
+      "loss": 0.4734,
+      "step": 1842
+    },
+    {
+      "epoch": 0.5311239193083573,
+      "grad_norm": 1.893790387928652,
+      "learning_rate": 1e-05,
+      "loss": 0.2447,
+      "step": 1843
+    },
+    {
+      "epoch": 0.5314121037463977,
+      "grad_norm": 3.015291369506835,
+      "learning_rate": 1e-05,
+      "loss": 0.4632,
+      "step": 1844
+    },
+    {
+      "epoch": 0.531700288184438,
+      "grad_norm": 1.8123259306752908,
+      "learning_rate": 1e-05,
+      "loss": 0.4598,
+      "step": 1845
+    },
+    {
+      "epoch": 0.5319884726224784,
+      "grad_norm": 1.4412994338870306,
+      "learning_rate": 1e-05,
+      "loss": 0.2632,
+      "step": 1846
+    },
+    {
+      "epoch": 0.5322766570605187,
+      "grad_norm": 3.1533731896340558,
+      "learning_rate": 1e-05,
+      "loss": 0.4111,
+      "step": 1847
+    },
+    {
+      "epoch": 0.5325648414985591,
+      "grad_norm": 1.7342875541722724,
+      "learning_rate": 1e-05,
+      "loss": 0.4592,
+      "step": 1848
+    },
+    {
+      "epoch": 0.5328530259365994,
+      "grad_norm": 1.8922708143846898,
+      "learning_rate": 1e-05,
+      "loss": 0.2664,
+      "step": 1849
+    },
+    {
+      "epoch": 0.5331412103746398,
+      "grad_norm": 2.4403878437728093,
+      "learning_rate": 1e-05,
+      "loss": 0.431,
+      "step": 1850
+    },
+    {
+      "epoch": 0.5334293948126801,
+      "grad_norm": 1.5409715979370118,
+      "learning_rate": 1e-05,
+      "loss": 0.4519,
+      "step": 1851
+    },
+    {
+      "epoch": 0.5337175792507205,
+      "grad_norm": 1.819497807365805,
+      "learning_rate": 1e-05,
+      "loss": 0.2755,
+      "step": 1852
+    },
+    {
+      "epoch": 0.5340057636887608,
+      "grad_norm": 3.2243638315771785,
+      "learning_rate": 1e-05,
+      "loss": 0.4747,
+      "step": 1853
+    },
+    {
+      "epoch": 0.5342939481268012,
+      "grad_norm": 1.6823526981268717,
+      "learning_rate": 1e-05,
+      "loss": 0.4483,
+      "step": 1854
+    },
+    {
+      "epoch": 0.5345821325648416,
+      "grad_norm": 1.4946564867384682,
+      "learning_rate": 1e-05,
+      "loss": 0.2604,
+      "step": 1855
+    },
+    {
+      "epoch": 0.5348703170028818,
+      "grad_norm": 2.9848690934637467,
+      "learning_rate": 1e-05,
+      "loss": 0.4461,
+      "step": 1856
+    },
+    {
+      "epoch": 0.5351585014409221,
+      "grad_norm": 2.1098009369876443,
+      "learning_rate": 1e-05,
+      "loss": 0.4643,
+      "step": 1857
+    },
+    {
+      "epoch": 0.5354466858789625,
+      "grad_norm": 1.2031374446979346,
+      "learning_rate": 1e-05,
+      "loss": 0.2603,
+      "step": 1858
+    },
+    {
+      "epoch": 0.5357348703170028,
+      "grad_norm": 2.6112357037271328,
+      "learning_rate": 1e-05,
+      "loss": 0.4528,
+      "step": 1859
+    },
+    {
+      "epoch": 0.5360230547550432,
+      "grad_norm": 1.584410884861249,
+      "learning_rate": 1e-05,
+      "loss": 0.4434,
+      "step": 1860
+    },
+    {
+      "epoch": 0.5363112391930835,
+      "grad_norm": 1.9740859563312552,
+      "learning_rate": 1e-05,
+      "loss": 0.2508,
+      "step": 1861
+    },
+    {
+      "epoch": 0.5365994236311239,
+      "grad_norm": 2.9887653882763687,
+      "learning_rate": 1e-05,
+      "loss": 0.4589,
+      "step": 1862
+    },
+    {
+      "epoch": 0.5368876080691642,
+      "grad_norm": 1.8961156323859274,
+      "learning_rate": 1e-05,
+      "loss": 0.4487,
+      "step": 1863
+    },
+    {
+      "epoch": 0.5371757925072046,
+      "grad_norm": 1.459533941523443,
+      "learning_rate": 1e-05,
+      "loss": 0.2757,
+      "step": 1864
+    },
+    {
+      "epoch": 0.537463976945245,
+      "grad_norm": 2.908063744858589,
+      "learning_rate": 1e-05,
+      "loss": 0.5206,
+      "step": 1865
+    },
+    {
+      "epoch": 0.5377521613832853,
+      "grad_norm": 1.9124436660439734,
+      "learning_rate": 1e-05,
+      "loss": 0.4649,
+      "step": 1866
+    },
+    {
+      "epoch": 0.5380403458213256,
+      "grad_norm": 2.0064850758455224,
+      "learning_rate": 1e-05,
+      "loss": 0.2519,
+      "step": 1867
+    },
+    {
+      "epoch": 0.538328530259366,
+      "grad_norm": 3.333105967232054,
+      "learning_rate": 1e-05,
+      "loss": 0.4499,
+      "step": 1868
+    },
+    {
+      "epoch": 0.5386167146974064,
+      "grad_norm": 2.009065187974275,
+      "learning_rate": 1e-05,
+      "loss": 0.4622,
+      "step": 1869
+    },
+    {
+      "epoch": 0.5389048991354467,
+      "grad_norm": 1.8532428290322112,
+      "learning_rate": 1e-05,
+      "loss": 0.2713,
+      "step": 1870
+    },
+    {
+      "epoch": 0.539193083573487,
+      "grad_norm": 3.02397352823615,
+      "learning_rate": 1e-05,
+      "loss": 0.4841,
+      "step": 1871
+    },
+    {
+      "epoch": 0.5394812680115274,
+      "grad_norm": 1.5324680567007694,
+      "learning_rate": 1e-05,
+      "loss": 0.4499,
+      "step": 1872
+    },
+    {
+      "epoch": 0.5397694524495678,
+      "grad_norm": 2.0931095142787752,
+      "learning_rate": 1e-05,
+      "loss": 0.2645,
+      "step": 1873
+    },
+    {
+      "epoch": 0.5400576368876081,
+      "grad_norm": 3.376271359266738,
+      "learning_rate": 1e-05,
+      "loss": 0.46,
+      "step": 1874
+    },
+    {
+      "epoch": 0.5403458213256485,
+      "grad_norm": 1.8229770106171024,
+      "learning_rate": 1e-05,
+      "loss": 0.4527,
+      "step": 1875
+    },
+    {
+      "epoch": 0.5406340057636888,
+      "grad_norm": 1.8368714849623498,
+      "learning_rate": 1e-05,
+      "loss": 0.2622,
+      "step": 1876
+    },
+    {
+      "epoch": 0.5409221902017292,
+      "grad_norm": 2.933593558121684,
+      "learning_rate": 1e-05,
+      "loss": 0.444,
+      "step": 1877
+    },
+    {
+      "epoch": 0.5412103746397694,
+      "grad_norm": 1.5578430127564136,
+      "learning_rate": 1e-05,
+      "loss": 0.4569,
+      "step": 1878
+    },
+    {
+      "epoch": 0.5414985590778097,
+      "grad_norm": 2.089309179173947,
+      "learning_rate": 1e-05,
+      "loss": 0.2708,
+      "step": 1879
+    },
+    {
+      "epoch": 0.5417867435158501,
+      "grad_norm": 2.512466219808063,
+      "learning_rate": 1e-05,
+      "loss": 0.4623,
+      "step": 1880
+    },
+    {
+      "epoch": 0.5420749279538905,
+      "grad_norm": 2.0389629009086105,
+      "learning_rate": 1e-05,
+      "loss": 0.4448,
+      "step": 1881
+    },
+    {
+      "epoch": 0.5423631123919308,
+      "grad_norm": 1.9030719199770536,
+      "learning_rate": 1e-05,
+      "loss": 0.2624,
+      "step": 1882
+    },
+    {
+      "epoch": 0.5426512968299712,
+      "grad_norm": 2.307530192752208,
+      "learning_rate": 1e-05,
+      "loss": 0.4589,
+      "step": 1883
+    },
+    {
+      "epoch": 0.5429394812680115,
+      "grad_norm": 1.629736150698527,
+      "learning_rate": 1e-05,
+      "loss": 0.4514,
+      "step": 1884
+    },
+    {
+      "epoch": 0.5432276657060519,
+      "grad_norm": 2.4088248450727927,
+      "learning_rate": 1e-05,
+      "loss": 0.2347,
+      "step": 1885
+    },
+    {
+      "epoch": 0.5435158501440922,
+      "grad_norm": 3.1431509643435582,
+      "learning_rate": 1e-05,
+      "loss": 0.4812,
+      "step": 1886
+    },
+    {
+      "epoch": 0.5438040345821326,
+      "grad_norm": 1.8826733706424852,
+      "learning_rate": 1e-05,
+      "loss": 0.4583,
+      "step": 1887
+    },
+    {
+      "epoch": 0.5440922190201729,
+      "grad_norm": 2.5953362787037273,
+      "learning_rate": 1e-05,
+      "loss": 0.278,
+      "step": 1888
+    },
+    {
+      "epoch": 0.5443804034582133,
+      "grad_norm": 2.687749652434169,
+      "learning_rate": 1e-05,
+      "loss": 0.4513,
+      "step": 1889
+    },
+    {
+      "epoch": 0.5446685878962536,
+      "grad_norm": 1.8653911290409602,
+      "learning_rate": 1e-05,
+      "loss": 0.4465,
+      "step": 1890
+    },
+    {
+      "epoch": 0.544956772334294,
+      "grad_norm": 2.2593739986321064,
+      "learning_rate": 1e-05,
+      "loss": 0.2572,
+      "step": 1891
+    },
+    {
+      "epoch": 0.5452449567723343,
+      "grad_norm": 2.5782587798696697,
+      "learning_rate": 1e-05,
+      "loss": 0.453,
+      "step": 1892
+    },
+    {
+      "epoch": 0.5455331412103747,
+      "grad_norm": 2.8581536633741833,
+      "learning_rate": 1e-05,
+      "loss": 0.4571,
+      "step": 1893
+    },
+    {
+      "epoch": 0.545821325648415,
+      "grad_norm": 3.2311140930376188,
+      "learning_rate": 1e-05,
+      "loss": 0.2625,
+      "step": 1894
+    },
+    {
+      "epoch": 0.5461095100864554,
+      "grad_norm": 3.776262194209603,
+      "learning_rate": 1e-05,
+      "loss": 0.4864,
+      "step": 1895
+    },
+    {
+      "epoch": 0.5463976945244957,
+      "grad_norm": 1.913305405328957,
+      "learning_rate": 1e-05,
+      "loss": 0.4489,
+      "step": 1896
+    },
+    {
+      "epoch": 0.5466858789625361,
+      "grad_norm": 2.293688537709464,
+      "learning_rate": 1e-05,
+      "loss": 0.2809,
+      "step": 1897
+    },
+    {
+      "epoch": 0.5469740634005764,
+      "grad_norm": 2.850231559640113,
+      "learning_rate": 1e-05,
+      "loss": 0.4485,
+      "step": 1898
+    },
+    {
+      "epoch": 0.5472622478386168,
+      "grad_norm": 1.7905016688720539,
+      "learning_rate": 1e-05,
+      "loss": 0.4526,
+      "step": 1899
+    },
+    {
+      "epoch": 0.547550432276657,
+      "grad_norm": 2.081835389558327,
+      "learning_rate": 1e-05,
+      "loss": 0.2573,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5478386167146974,
+      "grad_norm": 3.4872844874424436,
+      "learning_rate": 1e-05,
+      "loss": 0.451,
+      "step": 1901
+    },
+    {
+      "epoch": 0.5481268011527377,
+      "grad_norm": 2.1065416879642966,
+      "learning_rate": 1e-05,
+      "loss": 0.456,
+      "step": 1902
+    },
+    {
+      "epoch": 0.5484149855907781,
+      "grad_norm": 2.0926646314436694,
+      "learning_rate": 1e-05,
+      "loss": 0.2693,
+      "step": 1903
+    },
+    {
+      "epoch": 0.5487031700288184,
+      "grad_norm": 2.245442164994045,
+      "learning_rate": 1e-05,
+      "loss": 0.4263,
+      "step": 1904
+    },
+    {
+      "epoch": 0.5489913544668588,
+      "grad_norm": 2.2731655613094084,
+      "learning_rate": 1e-05,
+      "loss": 0.4583,
+      "step": 1905
+    },
+    {
+      "epoch": 0.5492795389048991,
+      "grad_norm": 2.285044456313453,
+      "learning_rate": 1e-05,
+      "loss": 0.2473,
+      "step": 1906
+    },
+    {
+      "epoch": 0.5495677233429395,
+      "grad_norm": 3.9068601259022016,
+      "learning_rate": 1e-05,
+      "loss": 0.429,
+      "step": 1907
+    },
+    {
+      "epoch": 0.5498559077809798,
+      "grad_norm": 1.82427426832822,
+      "learning_rate": 1e-05,
+      "loss": 0.4535,
+      "step": 1908
+    },
+    {
+      "epoch": 0.5501440922190202,
+      "grad_norm": 1.7584421286956202,
+      "learning_rate": 1e-05,
+      "loss": 0.2741,
+      "step": 1909
+    },
+    {
+      "epoch": 0.5504322766570605,
+      "grad_norm": 2.6504193638075457,
+      "learning_rate": 1e-05,
+      "loss": 0.441,
+      "step": 1910
+    },
+    {
+      "epoch": 0.5507204610951009,
+      "grad_norm": 1.9922239506499178,
+      "learning_rate": 1e-05,
+      "loss": 0.4509,
+      "step": 1911
+    },
+    {
+      "epoch": 0.5510086455331412,
+      "grad_norm": 2.6730502047285776,
+      "learning_rate": 1e-05,
+      "loss": 0.2758,
+      "step": 1912
+    },
+    {
+      "epoch": 0.5512968299711816,
+      "grad_norm": 3.401251289453041,
+      "learning_rate": 1e-05,
+      "loss": 0.475,
+      "step": 1913
+    },
+    {
+      "epoch": 0.5515850144092219,
+      "grad_norm": 2.1860798214734856,
+      "learning_rate": 1e-05,
+      "loss": 0.4443,
+      "step": 1914
+    },
+    {
+      "epoch": 0.5518731988472623,
+      "grad_norm": 1.4306611831603044,
+      "learning_rate": 1e-05,
+      "loss": 0.2711,
+      "step": 1915
+    },
+    {
+      "epoch": 0.5521613832853026,
+      "grad_norm": 2.882775587691361,
+      "learning_rate": 1e-05,
+      "loss": 0.4758,
+      "step": 1916
+    },
+    {
+      "epoch": 0.552449567723343,
+      "grad_norm": 2.1634947836778182,
+      "learning_rate": 1e-05,
+      "loss": 0.4501,
+      "step": 1917
+    },
+    {
+      "epoch": 0.5527377521613833,
+      "grad_norm": 1.8622904481100178,
+      "learning_rate": 1e-05,
+      "loss": 0.2574,
+      "step": 1918
+    },
+    {
+      "epoch": 0.5530259365994237,
+      "grad_norm": 2.9059816826291085,
+      "learning_rate": 1e-05,
+      "loss": 0.4656,
+      "step": 1919
+    },
+    {
+      "epoch": 0.553314121037464,
+      "grad_norm": 1.6339918829837667,
+      "learning_rate": 1e-05,
+      "loss": 0.4526,
+      "step": 1920
+    },
+    {
+      "epoch": 0.5536023054755044,
+      "grad_norm": 1.8584706483869857,
+      "learning_rate": 1e-05,
+      "loss": 0.2711,
+      "step": 1921
+    },
+    {
+      "epoch": 0.5538904899135446,
+      "grad_norm": 3.372951866244728,
+      "learning_rate": 1e-05,
+      "loss": 0.4183,
+      "step": 1922
+    },
+    {
+      "epoch": 0.554178674351585,
+      "grad_norm": 1.8625918927995937,
+      "learning_rate": 1e-05,
+      "loss": 0.4487,
+      "step": 1923
+    },
+    {
+      "epoch": 0.5544668587896253,
+      "grad_norm": 2.0076964461445876,
+      "learning_rate": 1e-05,
+      "loss": 0.2611,
+      "step": 1924
+    },
+    {
+      "epoch": 0.5547550432276657,
+      "grad_norm": 3.659663255000959,
+      "learning_rate": 1e-05,
+      "loss": 0.4282,
+      "step": 1925
+    },
+    {
+      "epoch": 0.555043227665706,
+      "grad_norm": 3.1758154607924904,
+      "learning_rate": 1e-05,
+      "loss": 0.4535,
+      "step": 1926
+    },
+    {
+      "epoch": 0.5553314121037464,
+      "grad_norm": 1.8110667318905647,
+      "learning_rate": 1e-05,
+      "loss": 0.2486,
+      "step": 1927
+    },
+    {
+      "epoch": 0.5556195965417867,
+      "grad_norm": 3.191348287919045,
+      "learning_rate": 1e-05,
+      "loss": 0.4748,
+      "step": 1928
+    },
+    {
+      "epoch": 0.5559077809798271,
+      "grad_norm": 2.223725809958116,
+      "learning_rate": 1e-05,
+      "loss": 0.4491,
+      "step": 1929
+    },
+    {
+      "epoch": 0.5561959654178674,
+      "grad_norm": 1.4495361720657725,
+      "learning_rate": 1e-05,
+      "loss": 0.2756,
+      "step": 1930
+    },
+    {
+      "epoch": 0.5564841498559078,
+      "grad_norm": 2.5107807968663756,
+      "learning_rate": 1e-05,
+      "loss": 0.4407,
+      "step": 1931
+    },
+    {
+      "epoch": 0.5567723342939481,
+      "grad_norm": 2.3669842152948815,
+      "learning_rate": 1e-05,
+      "loss": 0.4479,
+      "step": 1932
+    },
+    {
+      "epoch": 0.5570605187319885,
+      "grad_norm": 1.3904215328448146,
+      "learning_rate": 1e-05,
+      "loss": 0.2666,
+      "step": 1933
+    },
+    {
+      "epoch": 0.5573487031700288,
+      "grad_norm": 3.4674459239613507,
+      "learning_rate": 1e-05,
+      "loss": 0.4352,
+      "step": 1934
+    },
+    {
+      "epoch": 0.5576368876080692,
+      "grad_norm": 2.7767132905402256,
+      "learning_rate": 1e-05,
+      "loss": 0.463,
+      "step": 1935
+    },
+    {
+      "epoch": 0.5579250720461095,
+      "grad_norm": 1.9492353474204853,
+      "learning_rate": 1e-05,
+      "loss": 0.2803,
+      "step": 1936
+    },
+    {
+      "epoch": 0.5582132564841499,
+      "grad_norm": 2.691510021933372,
+      "learning_rate": 1e-05,
+      "loss": 0.4226,
+      "step": 1937
+    },
+    {
+      "epoch": 0.5585014409221902,
+      "grad_norm": 1.5621947677382966,
+      "learning_rate": 1e-05,
+      "loss": 0.4541,
+      "step": 1938
+    },
+    {
+      "epoch": 0.5587896253602306,
+      "grad_norm": 1.5869357621000888,
+      "learning_rate": 1e-05,
+      "loss": 0.2691,
+      "step": 1939
+    },
+    {
+      "epoch": 0.5590778097982709,
+      "grad_norm": 3.205006071970074,
+      "learning_rate": 1e-05,
+      "loss": 0.4332,
+      "step": 1940
+    },
+    {
+      "epoch": 0.5593659942363113,
+      "grad_norm": 2.3376904211592966,
+      "learning_rate": 1e-05,
+      "loss": 0.4633,
+      "step": 1941
+    },
+    {
+      "epoch": 0.5596541786743516,
+      "grad_norm": 1.9904667136804608,
+      "learning_rate": 1e-05,
+      "loss": 0.2459,
+      "step": 1942
+    },
+    {
+      "epoch": 0.559942363112392,
+      "grad_norm": 3.4591905583054086,
+      "learning_rate": 1e-05,
+      "loss": 0.4183,
+      "step": 1943
+    },
+    {
+      "epoch": 0.5602305475504322,
+      "grad_norm": 1.753562242041457,
+      "learning_rate": 1e-05,
+      "loss": 0.4409,
+      "step": 1944
+    },
+    {
+      "epoch": 0.5605187319884726,
+      "grad_norm": 1.902508296796601,
+      "learning_rate": 1e-05,
+      "loss": 0.2568,
+      "step": 1945
+    },
+    {
+      "epoch": 0.5608069164265129,
+      "grad_norm": 2.423376203861388,
+      "learning_rate": 1e-05,
+      "loss": 0.4601,
+      "step": 1946
+    },
+    {
+      "epoch": 0.5610951008645533,
+      "grad_norm": 1.754324898109713,
+      "learning_rate": 1e-05,
+      "loss": 0.4634,
+      "step": 1947
+    },
+    {
+      "epoch": 0.5613832853025936,
+      "grad_norm": 1.8735754639317097,
+      "learning_rate": 1e-05,
+      "loss": 0.2709,
+      "step": 1948
+    },
+    {
+      "epoch": 0.561671469740634,
+      "grad_norm": 3.452931451866348,
+      "learning_rate": 1e-05,
+      "loss": 0.4295,
+      "step": 1949
+    },
+    {
+      "epoch": 0.5619596541786743,
+      "grad_norm": 1.5226875454891986,
+      "learning_rate": 1e-05,
+      "loss": 0.4518,
+      "step": 1950
+    },
+    {
+      "epoch": 0.5622478386167147,
+      "grad_norm": 2.211168511151061,
+      "learning_rate": 1e-05,
+      "loss": 0.2622,
+      "step": 1951
+    },
+    {
+      "epoch": 0.562536023054755,
+      "grad_norm": 2.1656830793762434,
+      "learning_rate": 1e-05,
+      "loss": 0.4609,
+      "step": 1952
+    },
+    {
+      "epoch": 0.5628242074927954,
+      "grad_norm": 1.6655781931884497,
+      "learning_rate": 1e-05,
+      "loss": 0.4571,
+      "step": 1953
+    },
+    {
+      "epoch": 0.5631123919308357,
+      "grad_norm": 2.141959730144056,
+      "learning_rate": 1e-05,
+      "loss": 0.2444,
+      "step": 1954
+    },
+    {
+      "epoch": 0.5634005763688761,
+      "grad_norm": 2.9061920590169312,
+      "learning_rate": 1e-05,
+      "loss": 0.4446,
+      "step": 1955
+    },
+    {
+      "epoch": 0.5636887608069164,
+      "grad_norm": 1.790019799247271,
+      "learning_rate": 1e-05,
+      "loss": 0.4592,
+      "step": 1956
+    },
+    {
+      "epoch": 0.5639769452449568,
+      "grad_norm": 1.6906293802811507,
+      "learning_rate": 1e-05,
+      "loss": 0.2459,
+      "step": 1957
+    },
+    {
+      "epoch": 0.5642651296829971,
+      "grad_norm": 2.65820233033142,
+      "learning_rate": 1e-05,
+      "loss": 0.4584,
+      "step": 1958
+    },
+    {
+      "epoch": 0.5645533141210375,
+      "grad_norm": 1.6689396683044961,
+      "learning_rate": 1e-05,
+      "loss": 0.443,
+      "step": 1959
+    },
+    {
+      "epoch": 0.5648414985590778,
+      "grad_norm": 1.7996097107784526,
+      "learning_rate": 1e-05,
+      "loss": 0.276,
+      "step": 1960
+    },
+    {
+      "epoch": 0.5651296829971182,
+      "grad_norm": 2.7420215876567333,
+      "learning_rate": 1e-05,
+      "loss": 0.4471,
+      "step": 1961
+    },
+    {
+      "epoch": 0.5654178674351585,
+      "grad_norm": 2.0964142455103643,
+      "learning_rate": 1e-05,
+      "loss": 0.4463,
+      "step": 1962
+    },
+    {
+      "epoch": 0.5657060518731989,
+      "grad_norm": 2.3527628641881058,
+      "learning_rate": 1e-05,
+      "loss": 0.2624,
+      "step": 1963
+    },
+    {
+      "epoch": 0.5659942363112392,
+      "grad_norm": 3.6000461194743734,
+      "learning_rate": 1e-05,
+      "loss": 0.4486,
+      "step": 1964
+    },
+    {
+      "epoch": 0.5662824207492796,
+      "grad_norm": 1.734018448165514,
+      "learning_rate": 1e-05,
+      "loss": 0.4459,
+      "step": 1965
+    },
+    {
+      "epoch": 0.5665706051873198,
+      "grad_norm": 1.736113764684417,
+      "learning_rate": 1e-05,
+      "loss": 0.2535,
+      "step": 1966
+    },
+    {
+      "epoch": 0.5668587896253602,
+      "grad_norm": 2.820831180274689,
+      "learning_rate": 1e-05,
+      "loss": 0.4533,
+      "step": 1967
+    },
+    {
+      "epoch": 0.5671469740634005,
+      "grad_norm": 1.7774150858091302,
+      "learning_rate": 1e-05,
+      "loss": 0.4648,
+      "step": 1968
+    },
+    {
+      "epoch": 0.5674351585014409,
+      "grad_norm": 2.2558021269642152,
+      "learning_rate": 1e-05,
+      "loss": 0.2666,
+      "step": 1969
+    },
+    {
+      "epoch": 0.5677233429394812,
+      "grad_norm": 2.877564985648358,
+      "learning_rate": 1e-05,
+      "loss": 0.4544,
+      "step": 1970
+    },
+    {
+      "epoch": 0.5680115273775216,
+      "grad_norm": 3.172130672807336,
+      "learning_rate": 1e-05,
+      "loss": 0.4458,
+      "step": 1971
+    },
+    {
+      "epoch": 0.5682997118155619,
+      "grad_norm": 1.7251175232041882,
+      "learning_rate": 1e-05,
+      "loss": 0.2623,
+      "step": 1972
+    },
+    {
+      "epoch": 0.5685878962536023,
+      "grad_norm": 3.306136752859454,
+      "learning_rate": 1e-05,
+      "loss": 0.4066,
+      "step": 1973
+    },
+    {
+      "epoch": 0.5688760806916426,
+      "grad_norm": 1.9264481567426284,
+      "learning_rate": 1e-05,
+      "loss": 0.4513,
+      "step": 1974
+    },
+    {
+      "epoch": 0.569164265129683,
+      "grad_norm": 2.4078500074421965,
+      "learning_rate": 1e-05,
+      "loss": 0.25,
+      "step": 1975
+    },
+    {
+      "epoch": 0.5694524495677233,
+      "grad_norm": 2.897428927921229,
+      "learning_rate": 1e-05,
+      "loss": 0.424,
+      "step": 1976
+    },
+    {
+      "epoch": 0.5697406340057637,
+      "grad_norm": 2.3378278523124263,
+      "learning_rate": 1e-05,
+      "loss": 0.459,
+      "step": 1977
+    },
+    {
+      "epoch": 0.570028818443804,
+      "grad_norm": 2.210653772895215,
+      "learning_rate": 1e-05,
+      "loss": 0.2728,
+      "step": 1978
+    },
+    {
+      "epoch": 0.5703170028818444,
+      "grad_norm": 3.3872854527593295,
+      "learning_rate": 1e-05,
+      "loss": 0.4635,
+      "step": 1979
+    },
+    {
+      "epoch": 0.5706051873198847,
+      "grad_norm": 1.8882172250271712,
+      "learning_rate": 1e-05,
+      "loss": 0.4606,
+      "step": 1980
+    },
+    {
+      "epoch": 0.5708933717579251,
+      "grad_norm": 2.9454060428828086,
+      "learning_rate": 1e-05,
+      "loss": 0.2662,
+      "step": 1981
+    },
+    {
+      "epoch": 0.5711815561959654,
+      "grad_norm": 3.275216721117935,
+      "learning_rate": 1e-05,
+      "loss": 0.4199,
+      "step": 1982
+    },
+    {
+      "epoch": 0.5714697406340058,
+      "grad_norm": 1.4034515029833858,
+      "learning_rate": 1e-05,
+      "loss": 0.4517,
+      "step": 1983
+    },
+    {
+      "epoch": 0.5717579250720461,
+      "grad_norm": 1.459181585993267,
+      "learning_rate": 1e-05,
+      "loss": 0.2429,
+      "step": 1984
+    },
+    {
+      "epoch": 0.5720461095100865,
+      "grad_norm": 3.191726955890059,
+      "learning_rate": 1e-05,
+      "loss": 0.4206,
+      "step": 1985
+    },
+    {
+      "epoch": 0.5723342939481268,
+      "grad_norm": 1.5915879246958469,
+      "learning_rate": 1e-05,
+      "loss": 0.454,
+      "step": 1986
+    },
+    {
+      "epoch": 0.5726224783861672,
+      "grad_norm": 2.2028803814240443,
+      "learning_rate": 1e-05,
+      "loss": 0.2551,
+      "step": 1987
+    },
+    {
+      "epoch": 0.5729106628242074,
+      "grad_norm": 2.789248722855279,
+      "learning_rate": 1e-05,
+      "loss": 0.4277,
+      "step": 1988
+    },
+    {
+      "epoch": 0.5731988472622478,
+      "grad_norm": 1.9533881732362772,
+      "learning_rate": 1e-05,
+      "loss": 0.4409,
+      "step": 1989
+    },
+    {
+      "epoch": 0.5734870317002881,
+      "grad_norm": 1.911761756965588,
+      "learning_rate": 1e-05,
+      "loss": 0.2404,
+      "step": 1990
+    },
+    {
+      "epoch": 0.5737752161383285,
+      "grad_norm": 2.5606081431461725,
+      "learning_rate": 1e-05,
+      "loss": 0.4301,
+      "step": 1991
+    },
+    {
+      "epoch": 0.5740634005763688,
+      "grad_norm": 1.88672893655734,
+      "learning_rate": 1e-05,
+      "loss": 0.4517,
+      "step": 1992
+    },
+    {
+      "epoch": 0.5743515850144092,
+      "grad_norm": 1.8707287091071405,
+      "learning_rate": 1e-05,
+      "loss": 0.2709,
+      "step": 1993
+    },
+    {
+      "epoch": 0.5746397694524495,
+      "grad_norm": 2.67820686651302,
+      "learning_rate": 1e-05,
+      "loss": 0.4452,
+      "step": 1994
+    },
+    {
+      "epoch": 0.5749279538904899,
+      "grad_norm": 1.5825442843867061,
+      "learning_rate": 1e-05,
+      "loss": 0.4494,
+      "step": 1995
+    },
+    {
+      "epoch": 0.5752161383285302,
+      "grad_norm": 1.721512530766781,
+      "learning_rate": 1e-05,
+      "loss": 0.2467,
+      "step": 1996
+    },
+    {
+      "epoch": 0.5755043227665706,
+      "grad_norm": 2.2895279745774886,
+      "learning_rate": 1e-05,
+      "loss": 0.4192,
+      "step": 1997
+    },
+    {
+      "epoch": 0.575792507204611,
+      "grad_norm": 2.2188312061584012,
+      "learning_rate": 1e-05,
+      "loss": 0.4421,
+      "step": 1998
+    },
+    {
+      "epoch": 0.5760806916426513,
+      "grad_norm": 1.9452511989807846,
+      "learning_rate": 1e-05,
+      "loss": 0.2404,
+      "step": 1999
+    },
+    {
+      "epoch": 0.5763688760806917,
+      "grad_norm": 3.3970071542160714,
+      "learning_rate": 1e-05,
+      "loss": 0.4299,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5763688760806917,
+      "eval_websight_new_IoU": 0.019274206832051277,
+      "eval_websight_new_MAE_x": 286.760009765625,
+      "eval_websight_new_MAE_y": 587.2000122070312,
+      "eval_websight_new_coord_probability": 0.7459678649902344,
+      "eval_websight_new_correct_token_probability": 0.0011421897215768695,
+      "eval_websight_new_inside_bbox": 0.019999999552965164,
+      "eval_websight_new_loss": 0.2701660394668579,
+      "eval_websight_new_range_100_probability": 0.08858620375394821,
+      "eval_websight_new_range_20_probability": 0.015967553481459618,
+      "eval_websight_new_range_5_probability": 0.002914814744144678,
+      "eval_websight_new_runtime": 34.5561,
+      "eval_websight_new_samples_per_second": 1.447,
+      "eval_websight_new_steps_per_second": 0.203,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5763688760806917,
+      "eval_seeclick_IoU": 0.03321577608585358,
+      "eval_seeclick_MAE_x": 473.1000061035156,
+      "eval_seeclick_MAE_y": 516.47998046875,
+      "eval_seeclick_coord_probability": 0.7474353313446045,
+      "eval_seeclick_correct_token_probability": 0.0005565558094531298,
+      "eval_seeclick_inside_bbox": 0.05999999865889549,
+      "eval_seeclick_loss": 0.36988624930381775,
+      "eval_seeclick_range_100_probability": 0.12603196501731873,
+      "eval_seeclick_range_20_probability": 0.04917369410395622,
+      "eval_seeclick_range_5_probability": 0.007311419118195772,
+      "eval_seeclick_runtime": 42.4256,
+      "eval_seeclick_samples_per_second": 1.179,
+      "eval_seeclick_steps_per_second": 0.165,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5763688760806917,
+      "eval_icons_IoU": 0.0019011390395462513,
+      "eval_icons_MAE_x": 473.5400085449219,
+      "eval_icons_MAE_y": 567.260009765625,
+      "eval_icons_coord_probability": 0.7475740313529968,
+      "eval_icons_correct_token_probability": 0.00864581111818552,
+      "eval_icons_inside_bbox": 0.0,
+      "eval_icons_loss": 0.42921721935272217,
+      "eval_icons_range_100_probability": 0.27580708265304565,
+      "eval_icons_range_20_probability": 0.09803719818592072,
+      "eval_icons_range_5_probability": 0.03172297775745392,
+      "eval_icons_runtime": 43.3526,
+      "eval_icons_samples_per_second": 1.153,
+      "eval_icons_steps_per_second": 0.161,
+      "step": 2000
+    },
+    {
+      "epoch": 0.576657060518732,
+      "grad_norm": 1.9550860468198583,
+      "learning_rate": 1e-05,
+      "loss": 0.4442,
+      "step": 2001
+    },
+    {
+      "epoch": 0.5769452449567724,
+      "grad_norm": 1.577494146650957,
+      "learning_rate": 1e-05,
+      "loss": 0.2566,
+      "step": 2002
+    },
+    {
+      "epoch": 0.5772334293948127,
+      "grad_norm": 2.767049720277851,
+      "learning_rate": 1e-05,
+      "loss": 0.4566,
+      "step": 2003
+    },
+    {
+      "epoch": 0.577521613832853,
+      "grad_norm": 2.5465025491254725,
+      "learning_rate": 1e-05,
+      "loss": 0.4424,
+      "step": 2004
+    },
+    {
+      "epoch": 0.5778097982708934,
+      "grad_norm": 1.845007603334073,
+      "learning_rate": 1e-05,
+      "loss": 0.2723,
+      "step": 2005
+    },
+    {
+      "epoch": 0.5780979827089338,
+      "grad_norm": 3.64718050719754,
+      "learning_rate": 1e-05,
+      "loss": 0.4472,
+      "step": 2006
+    },
+    {
+      "epoch": 0.5783861671469741,
+      "grad_norm": 1.8344972295935609,
+      "learning_rate": 1e-05,
+      "loss": 0.4499,
+      "step": 2007
+    },
+    {
+      "epoch": 0.5786743515850145,
+      "grad_norm": 1.6739145034281844,
+      "learning_rate": 1e-05,
+      "loss": 0.2392,
+      "step": 2008
+    },
+    {
+      "epoch": 0.5789625360230548,
+      "grad_norm": 3.358983656918239,
+      "learning_rate": 1e-05,
+      "loss": 0.4054,
+      "step": 2009
+    },
+    {
+      "epoch": 0.579250720461095,
+      "grad_norm": 1.5959066660117969,
+      "learning_rate": 1e-05,
+      "loss": 0.4551,
+      "step": 2010
+    },
+    {
+      "epoch": 0.5795389048991354,
+      "grad_norm": 1.5841033665933912,
+      "learning_rate": 1e-05,
+      "loss": 0.2738,
+      "step": 2011
+    },
+    {
+      "epoch": 0.5798270893371757,
+      "grad_norm": 3.4534268362892773,
+      "learning_rate": 1e-05,
+      "loss": 0.4443,
+      "step": 2012
+    },
+    {
+      "epoch": 0.5801152737752161,
+      "grad_norm": 2.382349696943559,
+      "learning_rate": 1e-05,
+      "loss": 0.4492,
+      "step": 2013
+    },
+    {
+      "epoch": 0.5804034582132565,
+      "grad_norm": 2.0286916261947745,
+      "learning_rate": 1e-05,
+      "loss": 0.2592,
+      "step": 2014
+    },
+    {
+      "epoch": 0.5806916426512968,
+      "grad_norm": 2.6258538343031015,
+      "learning_rate": 1e-05,
+      "loss": 0.4514,
+      "step": 2015
+    },
+    {
+      "epoch": 0.5809798270893372,
+      "grad_norm": 1.754907531010363,
+      "learning_rate": 1e-05,
+      "loss": 0.4545,
+      "step": 2016
+    },
+    {
+      "epoch": 0.5812680115273775,
+      "grad_norm": 1.6571951399680935,
+      "learning_rate": 1e-05,
+      "loss": 0.2638,
+      "step": 2017
+    },
+    {
+      "epoch": 0.5815561959654179,
+      "grad_norm": 3.0299171142987222,
+      "learning_rate": 1e-05,
+      "loss": 0.441,
+      "step": 2018
+    },
+    {
+      "epoch": 0.5818443804034582,
+      "grad_norm": 2.144761348268048,
+      "learning_rate": 1e-05,
+      "loss": 0.4493,
+      "step": 2019
+    },
+    {
+      "epoch": 0.5821325648414986,
+      "grad_norm": 2.070044806614756,
+      "learning_rate": 1e-05,
+      "loss": 0.2497,
+      "step": 2020
+    },
+    {
+      "epoch": 0.5824207492795389,
+      "grad_norm": 3.099683898606559,
+      "learning_rate": 1e-05,
+      "loss": 0.4386,
+      "step": 2021
+    },
+    {
+      "epoch": 0.5827089337175793,
+      "grad_norm": 2.2102813923737847,
+      "learning_rate": 1e-05,
+      "loss": 0.453,
+      "step": 2022
+    },
+    {
+      "epoch": 0.5829971181556196,
+      "grad_norm": 1.5718659771633983,
+      "learning_rate": 1e-05,
+      "loss": 0.264,
+      "step": 2023
+    },
+    {
+      "epoch": 0.58328530259366,
+      "grad_norm": 2.8147473350664454,
+      "learning_rate": 1e-05,
+      "loss": 0.4552,
+      "step": 2024
+    },
+    {
+      "epoch": 0.5835734870317003,
+      "grad_norm": 1.8180551058536898,
+      "learning_rate": 1e-05,
+      "loss": 0.4588,
+      "step": 2025
+    },
+    {
+      "epoch": 0.5838616714697407,
+      "grad_norm": 1.474364620837879,
+      "learning_rate": 1e-05,
+      "loss": 0.2634,
+      "step": 2026
+    },
+    {
+      "epoch": 0.584149855907781,
+      "grad_norm": 3.029159607505013,
+      "learning_rate": 1e-05,
+      "loss": 0.4647,
+      "step": 2027
+    },
+    {
+      "epoch": 0.5844380403458214,
+      "grad_norm": 2.759542058969036,
+      "learning_rate": 1e-05,
+      "loss": 0.4514,
+      "step": 2028
+    },
+    {
+      "epoch": 0.5847262247838617,
+      "grad_norm": 1.8850118790012194,
+      "learning_rate": 1e-05,
+      "loss": 0.286,
+      "step": 2029
+    },
+    {
+      "epoch": 0.5850144092219021,
+      "grad_norm": 3.26299391417422,
+      "learning_rate": 1e-05,
+      "loss": 0.4338,
+      "step": 2030
+    },
+    {
+      "epoch": 0.5853025936599424,
+      "grad_norm": 2.6609988806535823,
+      "learning_rate": 1e-05,
+      "loss": 0.4535,
+      "step": 2031
+    },
+    {
+      "epoch": 0.5855907780979827,
+      "grad_norm": 1.5752606261152426,
+      "learning_rate": 1e-05,
+      "loss": 0.2454,
+      "step": 2032
+    },
+    {
+      "epoch": 0.585878962536023,
+      "grad_norm": 2.824232903081195,
+      "learning_rate": 1e-05,
+      "loss": 0.4713,
+      "step": 2033
+    },
+    {
+      "epoch": 0.5861671469740634,
+      "grad_norm": 1.417072975639568,
+      "learning_rate": 1e-05,
+      "loss": 0.4496,
+      "step": 2034
+    },
+    {
+      "epoch": 0.5864553314121037,
+      "grad_norm": 2.1161175785011914,
+      "learning_rate": 1e-05,
+      "loss": 0.2347,
+      "step": 2035
+    },
+    {
+      "epoch": 0.5867435158501441,
+      "grad_norm": 2.6492547309497536,
+      "learning_rate": 1e-05,
+      "loss": 0.4195,
+      "step": 2036
+    },
+    {
+      "epoch": 0.5870317002881844,
+      "grad_norm": 2.258183658195679,
+      "learning_rate": 1e-05,
+      "loss": 0.4461,
+      "step": 2037
+    },
+    {
+      "epoch": 0.5873198847262248,
+      "grad_norm": 1.6103495381978243,
+      "learning_rate": 1e-05,
+      "loss": 0.2462,
+      "step": 2038
+    },
+    {
+      "epoch": 0.5876080691642651,
+      "grad_norm": 3.044882260748048,
+      "learning_rate": 1e-05,
+      "loss": 0.4571,
+      "step": 2039
+    },
+    {
+      "epoch": 0.5878962536023055,
+      "grad_norm": 1.6161419156818677,
+      "learning_rate": 1e-05,
+      "loss": 0.4451,
+      "step": 2040
+    },
+    {
+      "epoch": 0.5881844380403458,
+      "grad_norm": 2.041516391278625,
+      "learning_rate": 1e-05,
+      "loss": 0.2591,
+      "step": 2041
+    },
+    {
+      "epoch": 0.5884726224783862,
+      "grad_norm": 3.297007260653783,
+      "learning_rate": 1e-05,
+      "loss": 0.4452,
+      "step": 2042
+    },
+    {
+      "epoch": 0.5887608069164265,
+      "grad_norm": 2.8069238351408665,
+      "learning_rate": 1e-05,
+      "loss": 0.4651,
+      "step": 2043
+    },
+    {
+      "epoch": 0.5890489913544669,
+      "grad_norm": 2.5115994883017785,
+      "learning_rate": 1e-05,
+      "loss": 0.24,
+      "step": 2044
+    },
+    {
+      "epoch": 0.5893371757925072,
+      "grad_norm": 2.9379329035629276,
+      "learning_rate": 1e-05,
+      "loss": 0.4284,
+      "step": 2045
+    },
+    {
+      "epoch": 0.5896253602305476,
+      "grad_norm": 1.441645710217809,
+      "learning_rate": 1e-05,
+      "loss": 0.4577,
+      "step": 2046
+    },
+    {
+      "epoch": 0.5899135446685879,
+      "grad_norm": 2.1597409075201117,
+      "learning_rate": 1e-05,
+      "loss": 0.257,
+      "step": 2047
+    },
+    {
+      "epoch": 0.5902017291066283,
+      "grad_norm": 2.5934115332175978,
+      "learning_rate": 1e-05,
+      "loss": 0.4384,
+      "step": 2048
+    },
+    {
+      "epoch": 0.5904899135446686,
+      "grad_norm": 1.5548112423461338,
+      "learning_rate": 1e-05,
+      "loss": 0.4497,
+      "step": 2049
+    },
+    {
+      "epoch": 0.590778097982709,
+      "grad_norm": 1.6076087527300826,
+      "learning_rate": 1e-05,
+      "loss": 0.27,
+      "step": 2050
+    },
+    {
+      "epoch": 0.5910662824207493,
+      "grad_norm": 2.666318432601144,
+      "learning_rate": 1e-05,
+      "loss": 0.4444,
+      "step": 2051
+    },
+    {
+      "epoch": 0.5913544668587897,
+      "grad_norm": 1.7010832910359102,
+      "learning_rate": 1e-05,
+      "loss": 0.4525,
+      "step": 2052
+    },
+    {
+      "epoch": 0.59164265129683,
+      "grad_norm": 2.0574392333952356,
+      "learning_rate": 1e-05,
+      "loss": 0.2574,
+      "step": 2053
+    },
+    {
+      "epoch": 0.5919308357348703,
+      "grad_norm": 2.6422015683283426,
+      "learning_rate": 1e-05,
+      "loss": 0.4758,
+      "step": 2054
+    },
+    {
+      "epoch": 0.5922190201729106,
+      "grad_norm": 1.7135159638115012,
+      "learning_rate": 1e-05,
+      "loss": 0.4483,
+      "step": 2055
+    },
+    {
+      "epoch": 0.592507204610951,
+      "grad_norm": 1.9732068399032463,
+      "learning_rate": 1e-05,
+      "loss": 0.2607,
+      "step": 2056
+    },
+    {
+      "epoch": 0.5927953890489913,
+      "grad_norm": 2.5308717917953647,
+      "learning_rate": 1e-05,
+      "loss": 0.4478,
+      "step": 2057
+    },
+    {
+      "epoch": 0.5930835734870317,
+      "grad_norm": 1.5856993531847656,
+      "learning_rate": 1e-05,
+      "loss": 0.4423,
+      "step": 2058
+    },
+    {
+      "epoch": 0.593371757925072,
+      "grad_norm": 2.085203981257147,
+      "learning_rate": 1e-05,
+      "loss": 0.2627,
+      "step": 2059
+    },
+    {
+      "epoch": 0.5936599423631124,
+      "grad_norm": 3.137261404606973,
+      "learning_rate": 1e-05,
+      "loss": 0.4271,
+      "step": 2060
+    },
+    {
+      "epoch": 0.5939481268011527,
+      "grad_norm": 1.8869257815420204,
+      "learning_rate": 1e-05,
+      "loss": 0.4401,
+      "step": 2061
+    },
+    {
+      "epoch": 0.5942363112391931,
+      "grad_norm": 1.7277056983931665,
+      "learning_rate": 1e-05,
+      "loss": 0.253,
+      "step": 2062
+    },
+    {
+      "epoch": 0.5945244956772334,
+      "grad_norm": 2.396605176150871,
+      "learning_rate": 1e-05,
+      "loss": 0.4594,
+      "step": 2063
+    },
+    {
+      "epoch": 0.5948126801152738,
+      "grad_norm": 1.9784710057565758,
+      "learning_rate": 1e-05,
+      "loss": 0.4399,
+      "step": 2064
+    },
+    {
+      "epoch": 0.5951008645533141,
+      "grad_norm": 1.581618264593078,
+      "learning_rate": 1e-05,
+      "loss": 0.251,
+      "step": 2065
+    },
+    {
+      "epoch": 0.5953890489913545,
+      "grad_norm": 3.2845507258138884,
+      "learning_rate": 1e-05,
+      "loss": 0.4369,
+      "step": 2066
+    },
+    {
+      "epoch": 0.5956772334293948,
+      "grad_norm": 1.8470353326205402,
+      "learning_rate": 1e-05,
+      "loss": 0.4397,
+      "step": 2067
+    },
+    {
+      "epoch": 0.5959654178674352,
+      "grad_norm": 1.7224709385549548,
+      "learning_rate": 1e-05,
+      "loss": 0.2424,
+      "step": 2068
+    },
+    {
+      "epoch": 0.5962536023054755,
+      "grad_norm": 4.22654189852175,
+      "learning_rate": 1e-05,
+      "loss": 0.4908,
+      "step": 2069
+    },
+    {
+      "epoch": 0.5965417867435159,
+      "grad_norm": 2.028066531551479,
+      "learning_rate": 1e-05,
+      "loss": 0.4393,
+      "step": 2070
+    },
+    {
+      "epoch": 0.5968299711815562,
+      "grad_norm": 2.061785829491306,
+      "learning_rate": 1e-05,
+      "loss": 0.2436,
+      "step": 2071
+    },
+    {
+      "epoch": 0.5971181556195966,
+      "grad_norm": 2.751133923717285,
+      "learning_rate": 1e-05,
+      "loss": 0.4098,
+      "step": 2072
+    },
+    {
+      "epoch": 0.5974063400576369,
+      "grad_norm": 2.53651075542974,
+      "learning_rate": 1e-05,
+      "loss": 0.4598,
+      "step": 2073
+    },
+    {
+      "epoch": 0.5976945244956773,
+      "grad_norm": 2.1147054121136093,
+      "learning_rate": 1e-05,
+      "loss": 0.2832,
+      "step": 2074
+    },
+    {
+      "epoch": 0.5979827089337176,
+      "grad_norm": 2.8304387907203155,
+      "learning_rate": 1e-05,
+      "loss": 0.4446,
+      "step": 2075
+    },
+    {
+      "epoch": 0.5982708933717579,
+      "grad_norm": 1.7704325905974798,
+      "learning_rate": 1e-05,
+      "loss": 0.4484,
+      "step": 2076
+    },
+    {
+      "epoch": 0.5985590778097982,
+      "grad_norm": 1.8008024416461477,
+      "learning_rate": 1e-05,
+      "loss": 0.2708,
+      "step": 2077
+    },
+    {
+      "epoch": 0.5988472622478386,
+      "grad_norm": 3.6878907328456374,
+      "learning_rate": 1e-05,
+      "loss": 0.415,
+      "step": 2078
+    },
+    {
+      "epoch": 0.5991354466858789,
+      "grad_norm": 1.8713840610503134,
+      "learning_rate": 1e-05,
+      "loss": 0.456,
+      "step": 2079
+    },
+    {
+      "epoch": 0.5994236311239193,
+      "grad_norm": 1.8219286489434328,
+      "learning_rate": 1e-05,
+      "loss": 0.2338,
+      "step": 2080
+    },
+    {
+      "epoch": 0.5997118155619596,
+      "grad_norm": 2.3137385073111174,
+      "learning_rate": 1e-05,
+      "loss": 0.4041,
+      "step": 2081
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.544285393211998,
+      "learning_rate": 1e-05,
+      "loss": 0.4518,
+      "step": 2082
+    },
+    {
+      "epoch": 0.6002881844380403,
+      "grad_norm": 2.411733248292235,
+      "learning_rate": 1e-05,
+      "loss": 0.23,
+      "step": 2083
+    },
+    {
+      "epoch": 0.6005763688760807,
+      "grad_norm": 3.112627136276708,
+      "learning_rate": 1e-05,
+      "loss": 0.4607,
+      "step": 2084
+    },
+    {
+      "epoch": 0.600864553314121,
+      "grad_norm": 1.9719538611650116,
+      "learning_rate": 1e-05,
+      "loss": 0.45,
+      "step": 2085
+    },
+    {
+      "epoch": 0.6011527377521614,
+      "grad_norm": 1.7376767223745377,
+      "learning_rate": 1e-05,
+      "loss": 0.2479,
+      "step": 2086
+    },
+    {
+      "epoch": 0.6014409221902017,
+      "grad_norm": 2.8881007892073938,
+      "learning_rate": 1e-05,
+      "loss": 0.4135,
+      "step": 2087
+    },
+    {
+      "epoch": 0.6017291066282421,
+      "grad_norm": 1.6446162108491584,
+      "learning_rate": 1e-05,
+      "loss": 0.4399,
+      "step": 2088
+    },
+    {
+      "epoch": 0.6020172910662824,
+      "grad_norm": 2.7271189379748937,
+      "learning_rate": 1e-05,
+      "loss": 0.2484,
+      "step": 2089
+    },
+    {
+      "epoch": 0.6023054755043228,
+      "grad_norm": 2.4137280688630565,
+      "learning_rate": 1e-05,
+      "loss": 0.4634,
+      "step": 2090
+    },
+    {
+      "epoch": 0.6025936599423631,
+      "grad_norm": 1.332354211507635,
+      "learning_rate": 1e-05,
+      "loss": 0.443,
+      "step": 2091
+    },
+    {
+      "epoch": 0.6028818443804035,
+      "grad_norm": 1.8085084227820358,
+      "learning_rate": 1e-05,
+      "loss": 0.2551,
+      "step": 2092
+    },
+    {
+      "epoch": 0.6031700288184438,
+      "grad_norm": 3.4780717621047175,
+      "learning_rate": 1e-05,
+      "loss": 0.4513,
+      "step": 2093
+    },
+    {
+      "epoch": 0.6034582132564842,
+      "grad_norm": 1.8600354139642854,
+      "learning_rate": 1e-05,
+      "loss": 0.452,
+      "step": 2094
+    },
+    {
+      "epoch": 0.6037463976945245,
+      "grad_norm": 1.3794793343273797,
+      "learning_rate": 1e-05,
+      "loss": 0.2327,
+      "step": 2095
+    },
+    {
+      "epoch": 0.6040345821325649,
+      "grad_norm": 2.9968348958854776,
+      "learning_rate": 1e-05,
+      "loss": 0.4055,
+      "step": 2096
+    },
+    {
+      "epoch": 0.6043227665706052,
+      "grad_norm": 1.8585527895953826,
+      "learning_rate": 1e-05,
+      "loss": 0.4457,
+      "step": 2097
+    },
+    {
+      "epoch": 0.6046109510086455,
+      "grad_norm": 1.6809499829679715,
+      "learning_rate": 1e-05,
+      "loss": 0.2397,
+      "step": 2098
+    },
+    {
+      "epoch": 0.6048991354466858,
+      "grad_norm": 3.1547568437589546,
+      "learning_rate": 1e-05,
+      "loss": 0.466,
+      "step": 2099
+    },
+    {
+      "epoch": 0.6051873198847262,
+      "grad_norm": 1.5434664194751035,
+      "learning_rate": 1e-05,
+      "loss": 0.4355,
+      "step": 2100
+    },
+    {
+      "epoch": 0.6054755043227665,
+      "grad_norm": 1.332929757256648,
+      "learning_rate": 1e-05,
+      "loss": 0.2457,
+      "step": 2101
+    },
+    {
+      "epoch": 0.6057636887608069,
+      "grad_norm": 3.2479715797322,
+      "learning_rate": 1e-05,
+      "loss": 0.4392,
+      "step": 2102
+    },
+    {
+      "epoch": 0.6060518731988472,
+      "grad_norm": 1.564221211658833,
+      "learning_rate": 1e-05,
+      "loss": 0.4557,
+      "step": 2103
+    },
+    {
+      "epoch": 0.6063400576368876,
+      "grad_norm": 1.5854904076813274,
+      "learning_rate": 1e-05,
+      "loss": 0.2338,
+      "step": 2104
+    },
+    {
+      "epoch": 0.6066282420749279,
+      "grad_norm": 2.546217553642099,
+      "learning_rate": 1e-05,
+      "loss": 0.4393,
+      "step": 2105
+    },
+    {
+      "epoch": 0.6069164265129683,
+      "grad_norm": 2.4010235327100222,
+      "learning_rate": 1e-05,
+      "loss": 0.4532,
+      "step": 2106
+    },
+    {
+      "epoch": 0.6072046109510086,
+      "grad_norm": 2.564789725880665,
+      "learning_rate": 1e-05,
+      "loss": 0.2498,
+      "step": 2107
+    },
+    {
+      "epoch": 0.607492795389049,
+      "grad_norm": 2.919474498363441,
+      "learning_rate": 1e-05,
+      "loss": 0.4448,
+      "step": 2108
+    },
+    {
+      "epoch": 0.6077809798270893,
+      "grad_norm": 1.687511716719379,
+      "learning_rate": 1e-05,
+      "loss": 0.4555,
+      "step": 2109
+    },
+    {
+      "epoch": 0.6080691642651297,
+      "grad_norm": 1.8155151009626451,
+      "learning_rate": 1e-05,
+      "loss": 0.2766,
+      "step": 2110
+    },
+    {
+      "epoch": 0.60835734870317,
+      "grad_norm": 3.3289963509643115,
+      "learning_rate": 1e-05,
+      "loss": 0.4555,
+      "step": 2111
+    },
+    {
+      "epoch": 0.6086455331412104,
+      "grad_norm": 1.7171666192458641,
+      "learning_rate": 1e-05,
+      "loss": 0.4466,
+      "step": 2112
+    },
+    {
+      "epoch": 0.6089337175792507,
+      "grad_norm": 1.4719082693173857,
+      "learning_rate": 1e-05,
+      "loss": 0.2462,
+      "step": 2113
+    },
+    {
+      "epoch": 0.6092219020172911,
+      "grad_norm": 2.549437257988022,
+      "learning_rate": 1e-05,
+      "loss": 0.4185,
+      "step": 2114
+    },
+    {
+      "epoch": 0.6095100864553314,
+      "grad_norm": 1.5133571866347024,
+      "learning_rate": 1e-05,
+      "loss": 0.4434,
+      "step": 2115
+    },
+    {
+      "epoch": 0.6097982708933718,
+      "grad_norm": 1.6460959768234367,
+      "learning_rate": 1e-05,
+      "loss": 0.2327,
+      "step": 2116
+    },
+    {
+      "epoch": 0.6100864553314121,
+      "grad_norm": 3.265837251000058,
+      "learning_rate": 1e-05,
+      "loss": 0.4489,
+      "step": 2117
+    },
+    {
+      "epoch": 0.6103746397694525,
+      "grad_norm": 1.9242162867891692,
+      "learning_rate": 1e-05,
+      "loss": 0.4426,
+      "step": 2118
+    },
+    {
+      "epoch": 0.6106628242074928,
+      "grad_norm": 1.6707231066794073,
+      "learning_rate": 1e-05,
+      "loss": 0.2424,
+      "step": 2119
+    },
+    {
+      "epoch": 0.6109510086455331,
+      "grad_norm": 3.040514531953241,
+      "learning_rate": 1e-05,
+      "loss": 0.4243,
+      "step": 2120
+    },
+    {
+      "epoch": 0.6112391930835734,
+      "grad_norm": 2.0241337559111474,
+      "learning_rate": 1e-05,
+      "loss": 0.4487,
+      "step": 2121
+    },
+    {
+      "epoch": 0.6115273775216138,
+      "grad_norm": 2.0137596032860605,
+      "learning_rate": 1e-05,
+      "loss": 0.2659,
+      "step": 2122
+    },
+    {
+      "epoch": 0.6118155619596541,
+      "grad_norm": 2.748840221276641,
+      "learning_rate": 1e-05,
+      "loss": 0.4226,
+      "step": 2123
+    },
+    {
+      "epoch": 0.6121037463976945,
+      "grad_norm": 1.7530949492316532,
+      "learning_rate": 1e-05,
+      "loss": 0.4391,
+      "step": 2124
+    },
+    {
+      "epoch": 0.6123919308357348,
+      "grad_norm": 1.9134842833787362,
+      "learning_rate": 1e-05,
+      "loss": 0.2124,
+      "step": 2125
+    },
+    {
+      "epoch": 0.6126801152737752,
+      "grad_norm": 2.6493052156796493,
+      "learning_rate": 1e-05,
+      "loss": 0.4462,
+      "step": 2126
+    },
+    {
+      "epoch": 0.6129682997118155,
+      "grad_norm": 1.5546265100802588,
+      "learning_rate": 1e-05,
+      "loss": 0.4402,
+      "step": 2127
+    },
+    {
+      "epoch": 0.6132564841498559,
+      "grad_norm": 1.5070519370111153,
+      "learning_rate": 1e-05,
+      "loss": 0.2406,
+      "step": 2128
+    },
+    {
+      "epoch": 0.6135446685878962,
+      "grad_norm": 3.4153867999695797,
+      "learning_rate": 1e-05,
+      "loss": 0.4442,
+      "step": 2129
+    },
+    {
+      "epoch": 0.6138328530259366,
+      "grad_norm": 1.8782783252038626,
+      "learning_rate": 1e-05,
+      "loss": 0.4412,
+      "step": 2130
+    },
+    {
+      "epoch": 0.614121037463977,
+      "grad_norm": 1.4955906768963096,
+      "learning_rate": 1e-05,
+      "loss": 0.2502,
+      "step": 2131
+    },
+    {
+      "epoch": 0.6144092219020173,
+      "grad_norm": 2.656539831948737,
+      "learning_rate": 1e-05,
+      "loss": 0.4265,
+      "step": 2132
+    },
+    {
+      "epoch": 0.6146974063400577,
+      "grad_norm": 1.4498378146223199,
+      "learning_rate": 1e-05,
+      "loss": 0.4532,
+      "step": 2133
+    },
+    {
+      "epoch": 0.614985590778098,
+      "grad_norm": 2.139351593612628,
+      "learning_rate": 1e-05,
+      "loss": 0.2513,
+      "step": 2134
+    },
+    {
+      "epoch": 0.6152737752161384,
+      "grad_norm": 3.958012147042486,
+      "learning_rate": 1e-05,
+      "loss": 0.4616,
+      "step": 2135
+    },
+    {
+      "epoch": 0.6155619596541787,
+      "grad_norm": 1.7572465145136853,
+      "learning_rate": 1e-05,
+      "loss": 0.4482,
+      "step": 2136
+    },
+    {
+      "epoch": 0.615850144092219,
+      "grad_norm": 1.710388138782609,
+      "learning_rate": 1e-05,
+      "loss": 0.2638,
+      "step": 2137
+    },
+    {
+      "epoch": 0.6161383285302594,
+      "grad_norm": 2.8416428396616182,
+      "learning_rate": 1e-05,
+      "loss": 0.443,
+      "step": 2138
+    },
+    {
+      "epoch": 0.6164265129682998,
+      "grad_norm": 1.9421895457064535,
+      "learning_rate": 1e-05,
+      "loss": 0.4569,
+      "step": 2139
+    },
+    {
+      "epoch": 0.6167146974063401,
+      "grad_norm": 1.5315925362235654,
+      "learning_rate": 1e-05,
+      "loss": 0.2386,
+      "step": 2140
+    },
+    {
+      "epoch": 0.6170028818443805,
+      "grad_norm": 2.5412597660103557,
+      "learning_rate": 1e-05,
+      "loss": 0.4376,
+      "step": 2141
+    },
+    {
+      "epoch": 0.6172910662824207,
+      "grad_norm": 1.5915989198552307,
+      "learning_rate": 1e-05,
+      "loss": 0.4401,
+      "step": 2142
+    },
+    {
+      "epoch": 0.617579250720461,
+      "grad_norm": 2.1522199351416056,
+      "learning_rate": 1e-05,
+      "loss": 0.2544,
+      "step": 2143
+    },
+    {
+      "epoch": 0.6178674351585014,
+      "grad_norm": 2.5485317240260077,
+      "learning_rate": 1e-05,
+      "loss": 0.4407,
+      "step": 2144
+    },
+    {
+      "epoch": 0.6181556195965417,
+      "grad_norm": 1.5313382595483551,
+      "learning_rate": 1e-05,
+      "loss": 0.4557,
+      "step": 2145
+    },
+    {
+      "epoch": 0.6184438040345821,
+      "grad_norm": 2.054486267832255,
+      "learning_rate": 1e-05,
+      "loss": 0.2723,
+      "step": 2146
+    },
+    {
+      "epoch": 0.6187319884726225,
+      "grad_norm": 3.4090998875109837,
+      "learning_rate": 1e-05,
+      "loss": 0.4408,
+      "step": 2147
+    },
+    {
+      "epoch": 0.6190201729106628,
+      "grad_norm": 1.9044220054973855,
+      "learning_rate": 1e-05,
+      "loss": 0.4521,
+      "step": 2148
+    },
+    {
+      "epoch": 0.6193083573487032,
+      "grad_norm": 2.7121438865991148,
+      "learning_rate": 1e-05,
+      "loss": 0.2245,
+      "step": 2149
+    },
+    {
+      "epoch": 0.6195965417867435,
+      "grad_norm": 2.5732597325520836,
+      "learning_rate": 1e-05,
+      "loss": 0.445,
+      "step": 2150
+    },
+    {
+      "epoch": 0.6198847262247839,
+      "grad_norm": 2.1823612337363514,
+      "learning_rate": 1e-05,
+      "loss": 0.4467,
+      "step": 2151
+    },
+    {
+      "epoch": 0.6201729106628242,
+      "grad_norm": 1.6829309697016506,
+      "learning_rate": 1e-05,
+      "loss": 0.2499,
+      "step": 2152
+    },
+    {
+      "epoch": 0.6204610951008646,
+      "grad_norm": 2.637279853976246,
+      "learning_rate": 1e-05,
+      "loss": 0.4658,
+      "step": 2153
+    },
+    {
+      "epoch": 0.6207492795389049,
+      "grad_norm": 1.5904686030479704,
+      "learning_rate": 1e-05,
+      "loss": 0.4341,
+      "step": 2154
+    },
+    {
+      "epoch": 0.6210374639769453,
+      "grad_norm": 1.7998781923573408,
+      "learning_rate": 1e-05,
+      "loss": 0.2492,
+      "step": 2155
+    },
+    {
+      "epoch": 0.6213256484149856,
+      "grad_norm": 2.8768727807697743,
+      "learning_rate": 1e-05,
+      "loss": 0.4269,
+      "step": 2156
+    },
+    {
+      "epoch": 0.621613832853026,
+      "grad_norm": 1.6298772350844697,
+      "learning_rate": 1e-05,
+      "loss": 0.4438,
+      "step": 2157
+    },
+    {
+      "epoch": 0.6219020172910663,
+      "grad_norm": 1.9792746786563291,
+      "learning_rate": 1e-05,
+      "loss": 0.2437,
+      "step": 2158
+    },
+    {
+      "epoch": 0.6221902017291067,
+      "grad_norm": 2.6610043866659434,
+      "learning_rate": 1e-05,
+      "loss": 0.4365,
+      "step": 2159
+    },
+    {
+      "epoch": 0.622478386167147,
+      "grad_norm": 2.1428413522782495,
+      "learning_rate": 1e-05,
+      "loss": 0.4459,
+      "step": 2160
+    },
+    {
+      "epoch": 0.6227665706051874,
+      "grad_norm": 2.668678412115329,
+      "learning_rate": 1e-05,
+      "loss": 0.2337,
+      "step": 2161
+    },
+    {
+      "epoch": 0.6230547550432277,
+      "grad_norm": 2.8447343605724194,
+      "learning_rate": 1e-05,
+      "loss": 0.4771,
+      "step": 2162
+    },
+    {
+      "epoch": 0.6233429394812681,
+      "grad_norm": 1.8125656931211491,
+      "learning_rate": 1e-05,
+      "loss": 0.4355,
+      "step": 2163
+    },
+    {
+      "epoch": 0.6236311239193083,
+      "grad_norm": 1.947844304822238,
+      "learning_rate": 1e-05,
+      "loss": 0.266,
+      "step": 2164
+    },
+    {
+      "epoch": 0.6239193083573487,
+      "grad_norm": 3.6323691529340185,
+      "learning_rate": 1e-05,
+      "loss": 0.4631,
+      "step": 2165
+    },
+    {
+      "epoch": 0.624207492795389,
+      "grad_norm": 2.2842934584255774,
+      "learning_rate": 1e-05,
+      "loss": 0.4512,
+      "step": 2166
+    },
+    {
+      "epoch": 0.6244956772334294,
+      "grad_norm": 1.930752931023711,
+      "learning_rate": 1e-05,
+      "loss": 0.2246,
+      "step": 2167
+    },
+    {
+      "epoch": 0.6247838616714697,
+      "grad_norm": 2.560422305241705,
+      "learning_rate": 1e-05,
+      "loss": 0.4496,
+      "step": 2168
+    },
+    {
+      "epoch": 0.6250720461095101,
+      "grad_norm": 1.4329863627493236,
+      "learning_rate": 1e-05,
+      "loss": 0.4413,
+      "step": 2169
+    },
+    {
+      "epoch": 0.6253602305475504,
+      "grad_norm": 1.945434478132978,
+      "learning_rate": 1e-05,
+      "loss": 0.2304,
+      "step": 2170
+    },
+    {
+      "epoch": 0.6256484149855908,
+      "grad_norm": 2.9717639049275038,
+      "learning_rate": 1e-05,
+      "loss": 0.3918,
+      "step": 2171
+    },
+    {
+      "epoch": 0.6259365994236311,
+      "grad_norm": 1.9612900231556427,
+      "learning_rate": 1e-05,
+      "loss": 0.4452,
+      "step": 2172
+    },
+    {
+      "epoch": 0.6262247838616715,
+      "grad_norm": 2.065594340795662,
+      "learning_rate": 1e-05,
+      "loss": 0.2569,
+      "step": 2173
+    },
+    {
+      "epoch": 0.6265129682997118,
+      "grad_norm": 2.4738696719914053,
+      "learning_rate": 1e-05,
+      "loss": 0.4151,
+      "step": 2174
+    },
+    {
+      "epoch": 0.6268011527377522,
+      "grad_norm": 1.8902591724331854,
+      "learning_rate": 1e-05,
+      "loss": 0.4523,
+      "step": 2175
+    },
+    {
+      "epoch": 0.6270893371757925,
+      "grad_norm": 1.6745493299984548,
+      "learning_rate": 1e-05,
+      "loss": 0.2395,
+      "step": 2176
+    },
+    {
+      "epoch": 0.6273775216138329,
+      "grad_norm": 3.2443745361221126,
+      "learning_rate": 1e-05,
+      "loss": 0.4415,
+      "step": 2177
+    },
+    {
+      "epoch": 0.6276657060518732,
+      "grad_norm": 2.276271708958747,
+      "learning_rate": 1e-05,
+      "loss": 0.4486,
+      "step": 2178
+    },
+    {
+      "epoch": 0.6279538904899136,
+      "grad_norm": 1.7325942030950319,
+      "learning_rate": 1e-05,
+      "loss": 0.2308,
+      "step": 2179
+    },
+    {
+      "epoch": 0.6282420749279539,
+      "grad_norm": 2.7488994586572955,
+      "learning_rate": 1e-05,
+      "loss": 0.4428,
+      "step": 2180
+    },
+    {
+      "epoch": 0.6285302593659943,
+      "grad_norm": 1.6960998545372619,
+      "learning_rate": 1e-05,
+      "loss": 0.4439,
+      "step": 2181
+    },
+    {
+      "epoch": 0.6288184438040346,
+      "grad_norm": 1.5366062651961152,
+      "learning_rate": 1e-05,
+      "loss": 0.2323,
+      "step": 2182
+    },
+    {
+      "epoch": 0.629106628242075,
+      "grad_norm": 3.2775864590246266,
+      "learning_rate": 1e-05,
+      "loss": 0.4763,
+      "step": 2183
+    },
+    {
+      "epoch": 0.6293948126801153,
+      "grad_norm": 1.398966470366019,
+      "learning_rate": 1e-05,
+      "loss": 0.4406,
+      "step": 2184
+    },
+    {
+      "epoch": 0.6296829971181557,
+      "grad_norm": 1.6627867337782436,
+      "learning_rate": 1e-05,
+      "loss": 0.2534,
+      "step": 2185
+    },
+    {
+      "epoch": 0.6299711815561959,
+      "grad_norm": 2.3357956265932067,
+      "learning_rate": 1e-05,
+      "loss": 0.4364,
+      "step": 2186
+    },
+    {
+      "epoch": 0.6302593659942363,
+      "grad_norm": 2.4272619214150675,
+      "learning_rate": 1e-05,
+      "loss": 0.4433,
+      "step": 2187
+    },
+    {
+      "epoch": 0.6305475504322766,
+      "grad_norm": 1.5897311251059187,
+      "learning_rate": 1e-05,
+      "loss": 0.2323,
+      "step": 2188
+    },
+    {
+      "epoch": 0.630835734870317,
+      "grad_norm": 3.6255015074985177,
+      "learning_rate": 1e-05,
+      "loss": 0.47,
+      "step": 2189
+    },
+    {
+      "epoch": 0.6311239193083573,
+      "grad_norm": 1.710826581051854,
+      "learning_rate": 1e-05,
+      "loss": 0.4348,
+      "step": 2190
+    },
+    {
+      "epoch": 0.6314121037463977,
+      "grad_norm": 1.3214079312521063,
+      "learning_rate": 1e-05,
+      "loss": 0.2396,
+      "step": 2191
+    },
+    {
+      "epoch": 0.631700288184438,
+      "grad_norm": 3.6581278309385974,
+      "learning_rate": 1e-05,
+      "loss": 0.4557,
+      "step": 2192
+    },
+    {
+      "epoch": 0.6319884726224784,
+      "grad_norm": 1.9873933956261685,
+      "learning_rate": 1e-05,
+      "loss": 0.4396,
+      "step": 2193
+    },
+    {
+      "epoch": 0.6322766570605187,
+      "grad_norm": 1.945389894464319,
+      "learning_rate": 1e-05,
+      "loss": 0.2469,
+      "step": 2194
+    },
+    {
+      "epoch": 0.6325648414985591,
+      "grad_norm": 2.4274401087222532,
+      "learning_rate": 1e-05,
+      "loss": 0.4449,
+      "step": 2195
+    },
+    {
+      "epoch": 0.6328530259365994,
+      "grad_norm": 1.7005894252433857,
+      "learning_rate": 1e-05,
+      "loss": 0.4442,
+      "step": 2196
+    },
+    {
+      "epoch": 0.6331412103746398,
+      "grad_norm": 1.706655592221686,
+      "learning_rate": 1e-05,
+      "loss": 0.2731,
+      "step": 2197
+    },
+    {
+      "epoch": 0.6334293948126801,
+      "grad_norm": 3.1349413931657444,
+      "learning_rate": 1e-05,
+      "loss": 0.433,
+      "step": 2198
+    },
+    {
+      "epoch": 0.6337175792507205,
+      "grad_norm": 1.6117010170738182,
+      "learning_rate": 1e-05,
+      "loss": 0.4403,
+      "step": 2199
+    },
+    {
+      "epoch": 0.6340057636887608,
+      "grad_norm": 2.029578739632731,
+      "learning_rate": 1e-05,
+      "loss": 0.2425,
+      "step": 2200
+    },
+    {
+      "epoch": 0.6342939481268012,
+      "grad_norm": 2.5226996042723933,
+      "learning_rate": 1e-05,
+      "loss": 0.4478,
+      "step": 2201
+    },
+    {
+      "epoch": 0.6345821325648415,
+      "grad_norm": 1.4749346665703282,
+      "learning_rate": 1e-05,
+      "loss": 0.4342,
+      "step": 2202
+    },
+    {
+      "epoch": 0.6348703170028819,
+      "grad_norm": 1.7929226949785588,
+      "learning_rate": 1e-05,
+      "loss": 0.2469,
+      "step": 2203
+    },
+    {
+      "epoch": 0.6351585014409222,
+      "grad_norm": 3.1932894460576,
+      "learning_rate": 1e-05,
+      "loss": 0.4478,
+      "step": 2204
+    },
+    {
+      "epoch": 0.6354466858789626,
+      "grad_norm": 1.6411041485005728,
+      "learning_rate": 1e-05,
+      "loss": 0.4527,
+      "step": 2205
+    },
+    {
+      "epoch": 0.6357348703170029,
+      "grad_norm": 1.782848833184517,
+      "learning_rate": 1e-05,
+      "loss": 0.2319,
+      "step": 2206
+    },
+    {
+      "epoch": 0.6360230547550433,
+      "grad_norm": 2.813445973391478,
+      "learning_rate": 1e-05,
+      "loss": 0.4259,
+      "step": 2207
+    },
+    {
+      "epoch": 0.6363112391930835,
+      "grad_norm": 1.570466548333388,
+      "learning_rate": 1e-05,
+      "loss": 0.4536,
+      "step": 2208
+    },
+    {
+      "epoch": 0.6365994236311239,
+      "grad_norm": 1.9323020194051894,
+      "learning_rate": 1e-05,
+      "loss": 0.2475,
+      "step": 2209
+    },
+    {
+      "epoch": 0.6368876080691642,
+      "grad_norm": 2.848547237007071,
+      "learning_rate": 1e-05,
+      "loss": 0.4506,
+      "step": 2210
+    },
+    {
+      "epoch": 0.6371757925072046,
+      "grad_norm": 1.6034499628142453,
+      "learning_rate": 1e-05,
+      "loss": 0.4457,
+      "step": 2211
+    },
+    {
+      "epoch": 0.6374639769452449,
+      "grad_norm": 1.5031872979236383,
+      "learning_rate": 1e-05,
+      "loss": 0.2531,
+      "step": 2212
+    },
+    {
+      "epoch": 0.6377521613832853,
+      "grad_norm": 4.04524836779013,
+      "learning_rate": 1e-05,
+      "loss": 0.4232,
+      "step": 2213
+    },
+    {
+      "epoch": 0.6380403458213256,
+      "grad_norm": 2.6051989171617915,
+      "learning_rate": 1e-05,
+      "loss": 0.4447,
+      "step": 2214
+    },
+    {
+      "epoch": 0.638328530259366,
+      "grad_norm": 2.4253756509703863,
+      "learning_rate": 1e-05,
+      "loss": 0.2495,
+      "step": 2215
+    },
+    {
+      "epoch": 0.6386167146974063,
+      "grad_norm": 3.992543219607941,
+      "learning_rate": 1e-05,
+      "loss": 0.4477,
+      "step": 2216
+    },
+    {
+      "epoch": 0.6389048991354467,
+      "grad_norm": 2.746369999913768,
+      "learning_rate": 1e-05,
+      "loss": 0.4516,
+      "step": 2217
+    },
+    {
+      "epoch": 0.639193083573487,
+      "grad_norm": 2.161832034593531,
+      "learning_rate": 1e-05,
+      "loss": 0.2387,
+      "step": 2218
+    },
+    {
+      "epoch": 0.6394812680115274,
+      "grad_norm": 2.8117143050302555,
+      "learning_rate": 1e-05,
+      "loss": 0.4242,
+      "step": 2219
+    },
+    {
+      "epoch": 0.6397694524495677,
+      "grad_norm": 2.153834976919712,
+      "learning_rate": 1e-05,
+      "loss": 0.4366,
+      "step": 2220
+    },
+    {
+      "epoch": 0.6400576368876081,
+      "grad_norm": 2.005668586029055,
+      "learning_rate": 1e-05,
+      "loss": 0.2281,
+      "step": 2221
+    },
+    {
+      "epoch": 0.6403458213256484,
+      "grad_norm": 3.7929156336115066,
+      "learning_rate": 1e-05,
+      "loss": 0.4786,
+      "step": 2222
+    },
+    {
+      "epoch": 0.6406340057636888,
+      "grad_norm": 2.4655314378548514,
+      "learning_rate": 1e-05,
+      "loss": 0.4458,
+      "step": 2223
+    },
+    {
+      "epoch": 0.6409221902017291,
+      "grad_norm": 1.9935463871528907,
+      "learning_rate": 1e-05,
+      "loss": 0.243,
+      "step": 2224
+    },
+    {
+      "epoch": 0.6412103746397695,
+      "grad_norm": 2.970243036083692,
+      "learning_rate": 1e-05,
+      "loss": 0.4445,
+      "step": 2225
+    },
+    {
+      "epoch": 0.6414985590778098,
+      "grad_norm": 1.4983249879619065,
+      "learning_rate": 1e-05,
+      "loss": 0.4363,
+      "step": 2226
+    },
+    {
+      "epoch": 0.6417867435158502,
+      "grad_norm": 1.7566778247567378,
+      "learning_rate": 1e-05,
+      "loss": 0.2574,
+      "step": 2227
+    },
+    {
+      "epoch": 0.6420749279538905,
+      "grad_norm": 2.755327867084357,
+      "learning_rate": 1e-05,
+      "loss": 0.4414,
+      "step": 2228
+    },
+    {
+      "epoch": 0.6423631123919309,
+      "grad_norm": 1.5241609644119447,
+      "learning_rate": 1e-05,
+      "loss": 0.4541,
+      "step": 2229
+    },
+    {
+      "epoch": 0.6426512968299711,
+      "grad_norm": 1.6525976260023942,
+      "learning_rate": 1e-05,
+      "loss": 0.2416,
+      "step": 2230
+    },
+    {
+      "epoch": 0.6429394812680115,
+      "grad_norm": 2.6696724083824166,
+      "learning_rate": 1e-05,
+      "loss": 0.4304,
+      "step": 2231
+    },
+    {
+      "epoch": 0.6432276657060518,
+      "grad_norm": 1.6959295340722362,
+      "learning_rate": 1e-05,
+      "loss": 0.4355,
+      "step": 2232
+    },
+    {
+      "epoch": 0.6435158501440922,
+      "grad_norm": 1.5511738255136893,
+      "learning_rate": 1e-05,
+      "loss": 0.2395,
+      "step": 2233
+    },
+    {
+      "epoch": 0.6438040345821325,
+      "grad_norm": 2.8396846352629117,
+      "learning_rate": 1e-05,
+      "loss": 0.3925,
+      "step": 2234
+    },
+    {
+      "epoch": 0.6440922190201729,
+      "grad_norm": 1.6811320261958476,
+      "learning_rate": 1e-05,
+      "loss": 0.4409,
+      "step": 2235
+    },
+    {
+      "epoch": 0.6443804034582132,
+      "grad_norm": 1.5161129898204209,
+      "learning_rate": 1e-05,
+      "loss": 0.2473,
+      "step": 2236
+    },
+    {
+      "epoch": 0.6446685878962536,
+      "grad_norm": 2.276729482868985,
+      "learning_rate": 1e-05,
+      "loss": 0.4029,
+      "step": 2237
+    },
+    {
+      "epoch": 0.6449567723342939,
+      "grad_norm": 2.0327197148905523,
+      "learning_rate": 1e-05,
+      "loss": 0.4414,
+      "step": 2238
+    },
+    {
+      "epoch": 0.6452449567723343,
+      "grad_norm": 1.5297950175355017,
+      "learning_rate": 1e-05,
+      "loss": 0.2331,
+      "step": 2239
+    },
+    {
+      "epoch": 0.6455331412103746,
+      "grad_norm": 2.494499277973723,
+      "learning_rate": 1e-05,
+      "loss": 0.4435,
+      "step": 2240
+    },
+    {
+      "epoch": 0.645821325648415,
+      "grad_norm": 1.6988591974997773,
+      "learning_rate": 1e-05,
+      "loss": 0.4421,
+      "step": 2241
+    },
+    {
+      "epoch": 0.6461095100864553,
+      "grad_norm": 1.6490346279860597,
+      "learning_rate": 1e-05,
+      "loss": 0.2377,
+      "step": 2242
+    },
+    {
+      "epoch": 0.6463976945244957,
+      "grad_norm": 3.2965946421832437,
+      "learning_rate": 1e-05,
+      "loss": 0.4428,
+      "step": 2243
+    },
+    {
+      "epoch": 0.646685878962536,
+      "grad_norm": 1.6432481617393475,
+      "learning_rate": 1e-05,
+      "loss": 0.4413,
+      "step": 2244
+    },
+    {
+      "epoch": 0.6469740634005764,
+      "grad_norm": 2.005173590063975,
+      "learning_rate": 1e-05,
+      "loss": 0.2428,
+      "step": 2245
+    },
+    {
+      "epoch": 0.6472622478386167,
+      "grad_norm": 3.170974833067571,
+      "learning_rate": 1e-05,
+      "loss": 0.4177,
+      "step": 2246
+    },
+    {
+      "epoch": 0.6475504322766571,
+      "grad_norm": 2.075420646325355,
+      "learning_rate": 1e-05,
+      "loss": 0.4267,
+      "step": 2247
+    },
+    {
+      "epoch": 0.6478386167146974,
+      "grad_norm": 1.6139530020843824,
+      "learning_rate": 1e-05,
+      "loss": 0.2549,
+      "step": 2248
+    },
+    {
+      "epoch": 0.6481268011527378,
+      "grad_norm": 2.451013669431529,
+      "learning_rate": 1e-05,
+      "loss": 0.4126,
+      "step": 2249
+    },
+    {
+      "epoch": 0.6484149855907781,
+      "grad_norm": 1.9988837079216608,
+      "learning_rate": 1e-05,
+      "loss": 0.4418,
+      "step": 2250
+    },
+    {
+      "epoch": 0.6484149855907781,
+      "eval_websight_new_IoU": 0.01669546402990818,
+      "eval_websight_new_MAE_x": 286.760009765625,
+      "eval_websight_new_MAE_y": 588.2999877929688,
+      "eval_websight_new_coord_probability": 0.7470889091491699,
+      "eval_websight_new_correct_token_probability": 0.0008188483188860118,
+      "eval_websight_new_inside_bbox": 0.019999999552965164,
+      "eval_websight_new_loss": 0.2525531053543091,
+      "eval_websight_new_range_100_probability": 0.0876244381070137,
+      "eval_websight_new_range_20_probability": 0.014237750321626663,
+      "eval_websight_new_range_5_probability": 0.0023667821660637856,
+      "eval_websight_new_runtime": 33.5452,
+      "eval_websight_new_samples_per_second": 1.491,
+      "eval_websight_new_steps_per_second": 0.209,
+      "step": 2250
+    },
+    {
+      "epoch": 0.6484149855907781,
+      "eval_seeclick_IoU": 0.032595857977867126,
+      "eval_seeclick_MAE_x": 473.1000061035156,
+      "eval_seeclick_MAE_y": 579.4199829101562,
+      "eval_seeclick_coord_probability": 0.748126745223999,
+      "eval_seeclick_correct_token_probability": 0.0005362248048186302,
+      "eval_seeclick_inside_bbox": 0.05999999865889549,
+      "eval_seeclick_loss": 0.36191099882125854,
+      "eval_seeclick_range_100_probability": 0.1297806203365326,
+      "eval_seeclick_range_20_probability": 0.0489816851913929,
+      "eval_seeclick_range_5_probability": 0.0075534796342253685,
+      "eval_seeclick_runtime": 43.4368,
+      "eval_seeclick_samples_per_second": 1.151,
+      "eval_seeclick_steps_per_second": 0.161,
+      "step": 2250
+    },
+    {
+      "epoch": 0.6484149855907781,
+      "eval_icons_IoU": 0.0030106292106211185,
+      "eval_icons_MAE_x": 473.5199890136719,
+      "eval_icons_MAE_y": 569.219970703125,
+      "eval_icons_coord_probability": 0.7479195594787598,
+      "eval_icons_correct_token_probability": 0.009233924560248852,
+      "eval_icons_inside_bbox": 0.0,
+      "eval_icons_loss": 0.42071011662483215,
+      "eval_icons_range_100_probability": 0.2812122106552124,
+      "eval_icons_range_20_probability": 0.10173630714416504,
+      "eval_icons_range_5_probability": 0.035193752497434616,
+      "eval_icons_runtime": 43.6615,
+      "eval_icons_samples_per_second": 1.145,
+      "eval_icons_steps_per_second": 0.16,
+      "step": 2250
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 20000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 6,
+  "save_steps": 250,
+  "total_flos": 1070009904922624.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}