diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,32505 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5765350245027385,
+  "eval_steps": 250,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0002882675122513693,
+      "grad_norm": 6585.90112269416,
+      "learning_rate": 0.0,
+      "loss": 70.0206,
+      "num_input_tokens_seen": 134792,
+      "step": 1
+    },
+    {
+      "epoch": 0.0002882675122513693,
+      "eval_websight_new_IoU": 0.0,
+      "eval_websight_new_MAE_x": 30.116134643554688,
+      "eval_websight_new_MAE_y": 81.22258377075195,
+      "eval_websight_new_NUM_probability": 8.40946690061628e-08,
+      "eval_websight_new_inside_bbox": 0.0,
+      "eval_websight_new_loss": 30257.947265625,
+      "eval_websight_new_loss_ce": 3.8668471574783325,
+      "eval_websight_new_loss_xval": 30184.0,
+      "eval_websight_new_runtime": 36.6864,
+      "eval_websight_new_samples_per_second": 1.363,
+      "eval_websight_new_steps_per_second": 0.055,
+      "num_input_tokens_seen": 134792,
+      "step": 1
+    },
+    {
+      "epoch": 0.0002882675122513693,
+      "eval_seeclick_IoU": 0.0,
+      "eval_seeclick_MAE_x": 35.54628944396973,
+      "eval_seeclick_MAE_y": 98.56820297241211,
+      "eval_seeclick_NUM_probability": 9.84709807028139e-08,
+      "eval_seeclick_inside_bbox": 0.0,
+      "eval_seeclick_loss": 26904.9765625,
+      "eval_seeclick_loss_ce": 4.496192216873169,
+      "eval_seeclick_loss_xval": 26752.0,
+      "eval_seeclick_runtime": 61.4943,
+      "eval_seeclick_samples_per_second": 0.813,
+      "eval_seeclick_steps_per_second": 0.033,
+      "num_input_tokens_seen": 134792,
+      "step": 1
+    },
+    {
+      "epoch": 0.0002882675122513693,
+      "eval_icons_IoU": 0.0,
+      "eval_icons_MAE_x": 40.47983169555664,
+      "eval_icons_MAE_y": 118.87767791748047,
+      "eval_icons_NUM_probability": 6.818678244258081e-08,
+      "eval_icons_inside_bbox": 0.0,
+      "eval_icons_loss": 32003.8046875,
+      "eval_icons_loss_ce": 3.8045945167541504,
+      "eval_icons_loss_xval": 32144.0,
+      "eval_icons_runtime": 64.6968,
+      "eval_icons_samples_per_second": 0.773,
+      "eval_icons_steps_per_second": 0.031,
+      "num_input_tokens_seen": 134792,
+      "step": 1
+    },
+    {
+      "epoch": 0.0002882675122513693,
+      "loss": 32243.8203125,
+      "loss_ce": 3.820674419403076,
+      "loss_xval": 32256.0,
+      "num_input_tokens_seen": 134792,
+      "step": 1
+    },
+    {
+      "epoch": 0.0005765350245027386,
+      "grad_norm": 996706.798337614,
+      "learning_rate": 1.255369169267456e-05,
+      "loss": 31499.8438,
+      "num_input_tokens_seen": 269736,
+      "step": 2
+    },
+    {
+      "epoch": 0.0005765350245027386,
+      "loss": 31795.837890625,
+      "loss_ce": 3.8380894660949707,
+      "loss_xval": 31744.0,
+      "num_input_tokens_seen": 269736,
+      "step": 2
+    },
+    {
+      "epoch": 0.0008648025367541078,
+      "grad_norm": 786586.824957904,
+      "learning_rate": 1.9897130578503877e-05,
+      "loss": 29204.1328,
+      "num_input_tokens_seen": 442296,
+      "step": 3
+    },
+    {
+      "epoch": 0.0008648025367541078,
+      "loss": 32579.794921875,
+      "loss_ce": 3.79455304145813,
+      "loss_xval": 32512.0,
+      "num_input_tokens_seen": 442296,
+      "step": 3
+    },
+    {
+      "epoch": 0.0011530700490054772,
+      "grad_norm": 269341.23696228134,
+      "learning_rate": 2.510738338534912e-05,
+      "loss": 6028.1396,
+      "num_input_tokens_seen": 577064,
+      "step": 4
+    },
+    {
+      "epoch": 0.0011530700490054772,
+      "loss": 5972.423828125,
+      "loss_ce": 4.423999309539795,
+      "loss_xval": 5952.0,
+      "num_input_tokens_seen": 577064,
+      "step": 4
+    },
+    {
+      "epoch": 0.0014413375612568463,
+      "grad_norm": 18537.21188035262,
+      "learning_rate": 2.9148769435775147e-05,
+      "loss": 121.8909,
+      "num_input_tokens_seen": 712088,
+      "step": 5
+    },
+    {
+      "epoch": 0.0014413375612568463,
+      "loss": 133.62535095214844,
+      "loss_ce": 4.250359535217285,
+      "loss_xval": 129.0,
+      "num_input_tokens_seen": 712088,
+      "step": 5
+    },
+    {
+      "epoch": 0.0017296050735082155,
+      "grad_norm": 32400.299293533124,
+      "learning_rate": 3.2450822271178436e-05,
+      "loss": 392.5652,
+      "num_input_tokens_seen": 884592,
+      "step": 6
+    },
+    {
+      "epoch": 0.0017296050735082155,
+      "loss": 344.605712890625,
+      "loss_ce": 4.355719566345215,
+      "loss_xval": 340.0,
+      "num_input_tokens_seen": 884592,
+      "step": 6
+    },
+    {
+      "epoch": 0.002017872585759585,
+      "grad_norm": 42990.192777167664,
+      "learning_rate": 3.524266816342358e-05,
+      "loss": 486.4419,
+      "num_input_tokens_seen": 1019368,
+      "step": 7
+    },
+    {
+      "epoch": 0.002017872585759585,
+      "loss": 469.3589782714844,
+      "loss_ce": 5.35897159576416,
+      "loss_xval": 464.0,
+      "num_input_tokens_seen": 1019368,
+      "step": 7
+    },
+    {
+      "epoch": 0.0023061400980109543,
+      "grad_norm": 22176.080277855566,
+      "learning_rate": 3.766107507802368e-05,
+      "loss": 163.1039,
+      "num_input_tokens_seen": 1154456,
+      "step": 8
+    },
+    {
+      "epoch": 0.0023061400980109543,
+      "loss": 207.9622344970703,
+      "loss_ce": 4.587244033813477,
+      "loss_xval": 203.0,
+      "num_input_tokens_seen": 1154456,
+      "step": 8
+    },
+    {
+      "epoch": 0.0025944076102623233,
+      "grad_norm": 5706.931463243618,
+      "learning_rate": 3.9794261157007754e-05,
+      "loss": 57.5576,
+      "num_input_tokens_seen": 1326888,
+      "step": 9
+    },
+    {
+      "epoch": 0.0025944076102623233,
+      "loss": 39.85100555419922,
+      "loss_ce": 4.835380554199219,
+      "loss_xval": 35.0,
+      "num_input_tokens_seen": 1326888,
+      "step": 9
+    },
+    {
+      "epoch": 0.0028826751225136927,
+      "grad_norm": 2176.5381898893447,
+      "learning_rate": 4.1702461128449717e-05,
+      "loss": 19.5141,
+      "num_input_tokens_seen": 1461696,
+      "step": 10
+    },
+    {
+      "epoch": 0.0028826751225136927,
+      "loss": 24.284374237060547,
+      "loss_ce": 5.823436737060547,
+      "loss_xval": 18.5,
+      "num_input_tokens_seen": 1461696,
+      "step": 10
+    },
+    {
+      "epoch": 0.003170942634765062,
+      "grad_norm": 945.057485234417,
+      "learning_rate": 4.342863797226275e-05,
+      "loss": 7.9961,
+      "num_input_tokens_seen": 1596760,
+      "step": 11
+    },
+    {
+      "epoch": 0.003170942634765062,
+      "loss": 8.878573417663574,
+      "loss_ce": 5.153963088989258,
+      "loss_xval": 3.71875,
+      "num_input_tokens_seen": 1596760,
+      "step": 11
+    },
+    {
+      "epoch": 0.003459210147016431,
+      "grad_norm": 1958.6007838874298,
+      "learning_rate": 4.5004513963852995e-05,
+      "loss": 18.0951,
+      "num_input_tokens_seen": 1769296,
+      "step": 12
+    },
+    {
+      "epoch": 0.003459210147016431,
+      "loss": 13.637873649597168,
+      "loss_ce": 5.114436149597168,
+      "loss_xval": 8.5,
+      "num_input_tokens_seen": 1769296,
+      "step": 12
+    },
+    {
+      "epoch": 0.0037474776592678004,
+      "grad_norm": 3655.5592110276825,
+      "learning_rate": 4.6454179348870823e-05,
+      "loss": 40.8945,
+      "num_input_tokens_seen": 1904080,
+      "step": 13
+    },
+    {
+      "epoch": 0.0037474776592678004,
+      "loss": 47.78251647949219,
+      "loss_ce": 6.032514572143555,
+      "loss_xval": 41.75,
+      "num_input_tokens_seen": 1904080,
+      "step": 13
+    },
+    {
+      "epoch": 0.00403574517151917,
+      "grad_norm": 3601.49256025649,
+      "learning_rate": 4.779635985609814e-05,
+      "loss": 42.2144,
+      "num_input_tokens_seen": 2039112,
+      "step": 14
+    },
+    {
+      "epoch": 0.00403574517151917,
+      "loss": 44.717384338378906,
+      "loss_ce": 5.5298848152160645,
+      "loss_xval": 39.25,
+      "num_input_tokens_seen": 2039112,
+      "step": 14
+    },
+    {
+      "epoch": 0.004324012683770539,
+      "grad_norm": 3510.379349357715,
+      "learning_rate": 4.904590001427903e-05,
+      "loss": 41.0742,
+      "num_input_tokens_seen": 2211640,
+      "step": 15
+    },
+    {
+      "epoch": 0.004324012683770539,
+      "loss": 29.72612762451172,
+      "loss_ce": 5.2730021476745605,
+      "loss_xval": 24.5,
+      "num_input_tokens_seen": 2211640,
+      "step": 15
+    },
+    {
+      "epoch": 0.004612280196021909,
+      "grad_norm": 2298.404288795249,
+      "learning_rate": 5.021476677069824e-05,
+      "loss": 22.6203,
+      "num_input_tokens_seen": 2346448,
+      "step": 16
+    },
+    {
+      "epoch": 0.004612280196021909,
+      "loss": 27.352113723754883,
+      "loss_ce": 6.406801223754883,
+      "loss_xval": 21.0,
+      "num_input_tokens_seen": 2346448,
+      "step": 16
+    },
+    {
+      "epoch": 0.004900547708273278,
+      "grad_norm": 532.1801093126115,
+      "learning_rate": 5.1312748314320346e-05,
+      "loss": 8.118,
+      "num_input_tokens_seen": 2481456,
+      "step": 17
+    },
+    {
+      "epoch": 0.004900547708273278,
+      "loss": 8.194659233093262,
+      "loss_ce": 6.041337966918945,
+      "loss_xval": 2.15625,
+      "num_input_tokens_seen": 2481456,
+      "step": 17
+    },
+    {
+      "epoch": 0.0051888152205246466,
+      "grad_norm": 837.9495849069981,
+      "learning_rate": 5.2347952849682313e-05,
+      "loss": 10.8715,
+      "num_input_tokens_seen": 2654080,
+      "step": 18
+    },
+    {
+      "epoch": 0.0051888152205246466,
+      "loss": 11.446012496948242,
+      "loss_ce": 6.5553879737854,
+      "loss_xval": 4.875,
+      "num_input_tokens_seen": 2654080,
+      "step": 18
+    },
+    {
+      "epoch": 0.005477082732776016,
+      "grad_norm": 1683.4434230931104,
+      "learning_rate": 5.332717233660044e-05,
+      "loss": 17.1665,
+      "num_input_tokens_seen": 2788840,
+      "step": 19
+    },
+    {
+      "epoch": 0.005477082732776016,
+      "loss": 17.23530387878418,
+      "loss_ce": 7.58686637878418,
+      "loss_xval": 9.625,
+      "num_input_tokens_seen": 2788840,
+      "step": 19
+    },
+    {
+      "epoch": 0.005765350245027385,
+      "grad_norm": 2513.5637155552545,
+      "learning_rate": 5.4256152821124276e-05,
+      "loss": 26.6961,
+      "num_input_tokens_seen": 2923840,
+      "step": 20
+    },
+    {
+      "epoch": 0.005765350245027385,
+      "loss": 24.286407470703125,
+      "loss_ce": 6.841094970703125,
+      "loss_xval": 17.5,
+      "num_input_tokens_seen": 2923840,
+      "step": 20
+    },
+    {
+      "epoch": 0.006053617757278754,
+      "grad_norm": 2736.6744740093,
+      "learning_rate": 5.513979874192746e-05,
+      "loss": 30.3839,
+      "num_input_tokens_seen": 3096336,
+      "step": 21
+    },
+    {
+      "epoch": 0.006053617757278754,
+      "loss": 30.83609962463379,
+      "loss_ce": 7.554849624633789,
+      "loss_xval": 23.25,
+      "num_input_tokens_seen": 3096336,
+      "step": 21
+    },
+    {
+      "epoch": 0.006341885269530124,
+      "grad_norm": 1821.5514426743405,
+      "learning_rate": 5.598232966493732e-05,
+      "loss": 18.9078,
+      "num_input_tokens_seen": 3231216,
+      "step": 22
+    },
+    {
+      "epoch": 0.006341885269530124,
+      "loss": 20.68878746032715,
+      "loss_ce": 8.735663414001465,
+      "loss_xval": 11.9375,
+      "num_input_tokens_seen": 3231216,
+      "step": 22
+    },
+    {
+      "epoch": 0.006630152781781493,
+      "grad_norm": 943.9392413077285,
+      "learning_rate": 5.6787402149051605e-05,
+      "loss": 10.9002,
+      "num_input_tokens_seen": 3366376,
+      "step": 23
+    },
+    {
+      "epoch": 0.006630152781781493,
+      "loss": 9.826754570007324,
+      "loss_ce": 7.108004570007324,
+      "loss_xval": 2.71875,
+      "num_input_tokens_seen": 3366376,
+      "step": 23
+    },
+    {
+      "epoch": 0.006918420294032862,
+      "grad_norm": 517.680000027031,
+      "learning_rate": 5.755820565652756e-05,
+      "loss": 10.216,
+      "num_input_tokens_seen": 3538864,
+      "step": 24
+    },
+    {
+      "epoch": 0.006918420294032862,
+      "loss": 8.601203918457031,
+      "loss_ce": 7.817756175994873,
+      "loss_xval": 0.78515625,
+      "num_input_tokens_seen": 3538864,
+      "step": 24
+    },
+    {
+      "epoch": 0.007206687806284232,
+      "grad_norm": 1570.5899753940823,
+      "learning_rate": 5.8297538871550295e-05,
+      "loss": 15.9455,
+      "num_input_tokens_seen": 3673728,
+      "step": 25
+    },
+    {
+      "epoch": 0.007206687806284232,
+      "loss": 18.513473510742188,
+      "loss_ce": 8.568161010742188,
+      "loss_xval": 9.9375,
+      "num_input_tokens_seen": 3673728,
+      "step": 25
+    },
+    {
+      "epoch": 0.007494955318535601,
+      "grad_norm": 1736.7074884499743,
+      "learning_rate": 5.900787104154538e-05,
+      "loss": 17.7182,
+      "num_input_tokens_seen": 3808768,
+      "step": 26
+    },
+    {
+      "epoch": 0.007494955318535601,
+      "loss": 16.724374771118164,
+      "loss_ce": 7.150156497955322,
+      "loss_xval": 9.5625,
+      "num_input_tokens_seen": 3808768,
+      "step": 26
+    },
+    {
+      "epoch": 0.007783222830786971,
+      "grad_norm": 1604.4374265767592,
+      "learning_rate": 5.9691391735511625e-05,
+      "loss": 17.0378,
+      "num_input_tokens_seen": 3981328,
+      "step": 27
+    },
+    {
+      "epoch": 0.007783222830786971,
+      "loss": 17.871387481689453,
+      "loss_ce": 7.668261528015137,
+      "loss_xval": 10.1875,
+      "num_input_tokens_seen": 3981328,
+      "step": 27
+    },
+    {
+      "epoch": 0.00807149034303834,
+      "grad_norm": 1119.1745907415943,
+      "learning_rate": 6.03500515487727e-05,
+      "loss": 12.7019,
+      "num_input_tokens_seen": 4116112,
+      "step": 28
+    },
+    {
+      "epoch": 0.00807149034303834,
+      "loss": 11.71876049041748,
+      "loss_ce": 8.15430736541748,
+      "loss_xval": 3.5625,
+      "num_input_tokens_seen": 4116112,
+      "step": 28
+    },
+    {
+      "epoch": 0.00835975785528971,
+      "grad_norm": 652.7927458908872,
+      "learning_rate": 6.09855956617039e-05,
+      "loss": 9.0132,
+      "num_input_tokens_seen": 4251240,
+      "step": 29
+    },
+    {
+      "epoch": 0.00835975785528971,
+      "loss": 8.91965103149414,
+      "loss_ce": 6.815158843994141,
+      "loss_xval": 2.109375,
+      "num_input_tokens_seen": 4251240,
+      "step": 29
+    },
+    {
+      "epoch": 0.008648025367541078,
+      "grad_norm": 309.08013570955944,
+      "learning_rate": 6.159959170695358e-05,
+      "loss": 8.2797,
+      "num_input_tokens_seen": 4423704,
+      "step": 30
+    },
+    {
+      "epoch": 0.008648025367541078,
+      "loss": 7.2059736251831055,
+      "loss_ce": 6.9381513595581055,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 4423704,
+      "step": 30
+    },
+    {
+      "epoch": 0.008936292879792447,
+      "grad_norm": 821.4961563326211,
+      "learning_rate": 6.219345306558267e-05,
+      "loss": 10.3026,
+      "num_input_tokens_seen": 4558472,
+      "step": 31
+    },
+    {
+      "epoch": 0.008936292879792447,
+      "loss": 12.730304718017578,
+      "loss_ce": 7.624835968017578,
+      "loss_xval": 5.09375,
+      "num_input_tokens_seen": 4558472,
+      "step": 31
+    },
+    {
+      "epoch": 0.009224560392043817,
+      "grad_norm": 810.3899443561476,
+      "learning_rate": 6.276845846337281e-05,
+      "loss": 9.2442,
+      "num_input_tokens_seen": 4693568,
+      "step": 32
+    },
+    {
+      "epoch": 0.009224560392043817,
+      "loss": 9.102476119995117,
+      "loss_ce": 6.311460494995117,
+      "loss_xval": 2.796875,
+      "num_input_tokens_seen": 4693568,
+      "step": 32
+    },
+    {
+      "epoch": 0.009512827904295185,
+      "grad_norm": 1126.8158352879109,
+      "learning_rate": 6.332576855076663e-05,
+      "loss": 12.4498,
+      "num_input_tokens_seen": 4865944,
+      "step": 33
+    },
+    {
+      "epoch": 0.009512827904295185,
+      "loss": 9.520813941955566,
+      "loss_ce": 6.290345191955566,
+      "loss_xval": 3.234375,
+      "num_input_tokens_seen": 4865944,
+      "step": 33
+    },
+    {
+      "epoch": 0.009801095416546555,
+      "grad_norm": 946.1031004612911,
+      "learning_rate": 6.386644000699491e-05,
+      "loss": 10.5841,
+      "num_input_tokens_seen": 5000728,
+      "step": 34
+    },
+    {
+      "epoch": 0.009801095416546555,
+      "loss": 13.091533660888672,
+      "loss_ce": 6.704814434051514,
+      "loss_xval": 6.375,
+      "num_input_tokens_seen": 5000728,
+      "step": 34
+    },
+    {
+      "epoch": 0.010089362928797925,
+      "grad_norm": 254.32293701379814,
+      "learning_rate": 6.439143759919874e-05,
+      "loss": 6.2691,
+      "num_input_tokens_seen": 5135872,
+      "step": 35
+    },
+    {
+      "epoch": 0.010089362928797925,
+      "loss": 6.317028999328613,
+      "loss_ce": 5.744763374328613,
+      "loss_xval": 0.5703125,
+      "num_input_tokens_seen": 5135872,
+      "step": 35
+    },
+    {
+      "epoch": 0.010377630441049293,
+      "grad_norm": 96.83144759509878,
+      "learning_rate": 6.490164454235687e-05,
+      "loss": 6.4683,
+      "num_input_tokens_seen": 5308288,
+      "step": 36
+    },
+    {
+      "epoch": 0.010377630441049293,
+      "loss": 6.021927833557129,
+      "loss_ce": 5.571000099182129,
+      "loss_xval": 0.451171875,
+      "num_input_tokens_seen": 5308288,
+      "step": 36
+    },
+    {
+      "epoch": 0.010665897953300663,
+      "grad_norm": 329.47055076201457,
+      "learning_rate": 6.539787143947167e-05,
+      "loss": 6.44,
+      "num_input_tokens_seen": 5443160,
+      "step": 37
+    },
+    {
+      "epoch": 0.010665897953300663,
+      "loss": 6.1701226234436035,
+      "loss_ce": 5.8368706703186035,
+      "loss_xval": 0.333984375,
+      "num_input_tokens_seen": 5443160,
+      "step": 37
+    },
+    {
+      "epoch": 0.010954165465552033,
+      "grad_norm": 850.6431793469337,
+      "learning_rate": 6.5880864029275e-05,
+      "loss": 8.3984,
+      "num_input_tokens_seen": 5578416,
+      "step": 38
+    },
+    {
+      "epoch": 0.010954165465552033,
+      "loss": 7.8684587478637695,
+      "loss_ce": 5.1809587478637695,
+      "loss_xval": 2.6875,
+      "num_input_tokens_seen": 5578416,
+      "step": 38
+    },
+    {
+      "epoch": 0.011242432977803401,
+      "grad_norm": 692.1496892412431,
+      "learning_rate": 6.63513099273747e-05,
+      "loss": 7.6066,
+      "num_input_tokens_seen": 5750856,
+      "step": 39
+    },
+    {
+      "epoch": 0.011242432977803401,
+      "loss": 8.899791717529297,
+      "loss_ce": 4.835338592529297,
+      "loss_xval": 4.0625,
+      "num_input_tokens_seen": 5750856,
+      "step": 39
+    },
+    {
+      "epoch": 0.01153070049005477,
+      "grad_norm": 468.94665746005865,
+      "learning_rate": 6.680984451379883e-05,
+      "loss": 6.0939,
+      "num_input_tokens_seen": 5885624,
+      "step": 40
+    },
+    {
+      "epoch": 0.01153070049005477,
+      "loss": 5.63425350189209,
+      "loss_ce": 4.90231990814209,
+      "loss_xval": 0.73046875,
+      "num_input_tokens_seen": 5885624,
+      "step": 40
+    },
+    {
+      "epoch": 0.01181896800230614,
+      "grad_norm": 426.5270199999428,
+      "learning_rate": 6.725705609344598e-05,
+      "loss": 5.3084,
+      "num_input_tokens_seen": 6020824,
+      "step": 41
+    },
+    {
+      "epoch": 0.01181896800230614,
+      "loss": 5.0226240158081055,
+      "loss_ce": 4.4718427658081055,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 6020824,
+      "step": 41
+    },
+    {
+      "epoch": 0.012107235514557509,
+      "grad_norm": 75.92312869871212,
+      "learning_rate": 6.769349043460202e-05,
+      "loss": 4.6539,
+      "num_input_tokens_seen": 6193280,
+      "step": 42
+    },
+    {
+      "epoch": 0.012107235514557509,
+      "loss": 4.303558826446533,
+      "loss_ce": 4.052459716796875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 6193280,
+      "step": 42
+    },
+    {
+      "epoch": 0.012395503026808878,
+      "grad_norm": 397.7092047833716,
+      "learning_rate": 6.81196547733565e-05,
+      "loss": 5.0739,
+      "num_input_tokens_seen": 6328136,
+      "step": 43
+    },
+    {
+      "epoch": 0.012395503026808878,
+      "loss": 5.631143093109131,
+      "loss_ce": 4.236611843109131,
+      "loss_xval": 1.390625,
+      "num_input_tokens_seen": 6328136,
+      "step": 43
+    },
+    {
+      "epoch": 0.012683770539060248,
+      "grad_norm": 430.60447024117013,
+      "learning_rate": 6.853602135761187e-05,
+      "loss": 4.6489,
+      "num_input_tokens_seen": 6463240,
+      "step": 44
+    },
+    {
+      "epoch": 0.012683770539060248,
+      "loss": 4.990880966186523,
+      "loss_ce": 3.8263304233551025,
+      "loss_xval": 1.1640625,
+      "num_input_tokens_seen": 6463240,
+      "step": 44
+    },
+    {
+      "epoch": 0.012972038051311616,
+      "grad_norm": 540.73323665623,
+      "learning_rate": 6.89430305927829e-05,
+      "loss": 5.161,
+      "num_input_tokens_seen": 6635752,
+      "step": 45
+    },
+    {
+      "epoch": 0.012972038051311616,
+      "loss": 4.3311381340026855,
+      "loss_ce": 3.3467631340026855,
+      "loss_xval": 0.984375,
+      "num_input_tokens_seen": 6635752,
+      "step": 45
+    },
+    {
+      "epoch": 0.013260305563562986,
+      "grad_norm": 458.0472938948524,
+      "learning_rate": 6.934109384172616e-05,
+      "loss": 4.7264,
+      "num_input_tokens_seen": 6770488,
+      "step": 46
+    },
+    {
+      "epoch": 0.013260305563562986,
+      "loss": 5.214451789855957,
+      "loss_ce": 3.6080069541931152,
+      "loss_xval": 1.609375,
+      "num_input_tokens_seen": 6770488,
+      "step": 46
+    },
+    {
+      "epoch": 0.013548573075814356,
+      "grad_norm": 129.32422015778562,
+      "learning_rate": 6.973059592352828e-05,
+      "loss": 3.311,
+      "num_input_tokens_seen": 6905640,
+      "step": 47
+    },
+    {
+      "epoch": 0.013548573075814356,
+      "loss": 3.4464430809020996,
+      "loss_ce": 3.2276930809020996,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 6905640,
+      "step": 47
+    },
+    {
+      "epoch": 0.013836840588065724,
+      "grad_norm": 65.53517362370098,
+      "learning_rate": 7.011189734920213e-05,
+      "loss": 3.3215,
+      "num_input_tokens_seen": 7078176,
+      "step": 48
+    },
+    {
+      "epoch": 0.013836840588065724,
+      "loss": 3.060483455657959,
+      "loss_ce": 2.844907283782959,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 7078176,
+      "step": 48
+    },
+    {
+      "epoch": 0.014125108100317094,
+      "grad_norm": 215.52012341313568,
+      "learning_rate": 7.048533632684716e-05,
+      "loss": 3.4409,
+      "num_input_tokens_seen": 7213000,
+      "step": 49
+    },
+    {
+      "epoch": 0.014125108100317094,
+      "loss": 3.4835710525512695,
+      "loss_ce": 3.1407976150512695,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 7213000,
+      "step": 49
+    },
+    {
+      "epoch": 0.014413375612568464,
+      "grad_norm": 431.0270260191144,
+      "learning_rate": 7.085123056422486e-05,
+      "loss": 3.7639,
+      "num_input_tokens_seen": 7348040,
+      "step": 50
+    },
+    {
+      "epoch": 0.014413375612568464,
+      "loss": 3.667371988296509,
+      "loss_ce": 2.829481363296509,
+      "loss_xval": 0.8359375,
+      "num_input_tokens_seen": 7348040,
+      "step": 50
+    },
+    {
+      "epoch": 0.014701643124819834,
+      "grad_norm": 378.86249167362797,
+      "learning_rate": 7.120987889282422e-05,
+      "loss": 3.703,
+      "num_input_tokens_seen": 7520472,
+      "step": 51
+    },
+    {
+      "epoch": 0.014701643124819834,
+      "eval_websight_new_IoU": 0.0,
+      "eval_websight_new_MAE_x": 0.5620081424713135,
+      "eval_websight_new_MAE_y": 0.8609068095684052,
+      "eval_websight_new_NUM_probability": 2.2496185010822956e-05,
+      "eval_websight_new_inside_bbox": 0.0,
+      "eval_websight_new_loss": 2.959818124771118,
+      "eval_websight_new_loss_ce": 2.5184309482574463,
+      "eval_websight_new_loss_xval": 0.4293212890625,
+      "eval_websight_new_runtime": 35.5903,
+      "eval_websight_new_samples_per_second": 1.405,
+      "eval_websight_new_steps_per_second": 0.056,
+      "num_input_tokens_seen": 7520472,
+      "step": 51
+    },
+    {
+      "epoch": 0.014701643124819834,
+      "eval_seeclick_IoU": 0.007790456060320139,
+      "eval_seeclick_MAE_x": 0.39892005920410156,
+      "eval_seeclick_MAE_y": 0.5314352512359619,
+      "eval_seeclick_NUM_probability": 3.157063019898487e-05,
+      "eval_seeclick_inside_bbox": 0.015625,
+      "eval_seeclick_loss": 3.310542345046997,
+      "eval_seeclick_loss_ce": 2.78228759765625,
+      "eval_seeclick_loss_xval": 0.504058837890625,
+      "eval_seeclick_runtime": 63.5135,
+      "eval_seeclick_samples_per_second": 0.787,
+      "eval_seeclick_steps_per_second": 0.031,
+      "num_input_tokens_seen": 7520472,
+      "step": 51
+    },
+    {
+      "epoch": 0.014701643124819834,
+      "eval_icons_IoU": 9.042318561114371e-05,
+      "eval_icons_MAE_x": 0.8792010545730591,
+      "eval_icons_MAE_y": 0.6797044575214386,
+      "eval_icons_NUM_probability": 1.8500017176847905e-05,
+      "eval_icons_inside_bbox": 0.0,
+      "eval_icons_loss": 3.088502883911133,
+      "eval_icons_loss_ce": 2.412390947341919,
+      "eval_icons_loss_xval": 0.6337890625,
+      "eval_icons_runtime": 67.6782,
+      "eval_icons_samples_per_second": 0.739,
+      "eval_icons_steps_per_second": 0.03,
+      "num_input_tokens_seen": 7520472,
+      "step": 51
+    },
+    {
+      "epoch": 0.014701643124819834,
+      "loss": 2.99432373046875,
+      "loss_ce": 2.43255615234375,
+      "loss_xval": 0.5625,
+      "num_input_tokens_seen": 7520472,
+      "step": 51
+    },
+    {
+      "epoch": 0.014989910637071202,
+      "grad_norm": 217.68238108622742,
+      "learning_rate": 7.156156273421994e-05,
+      "loss": 3.0923,
+      "num_input_tokens_seen": 7655320,
+      "step": 52
+    },
+    {
+      "epoch": 0.014989910637071202,
+      "loss": 3.246007204055786,
+      "loss_ce": 2.754307985305786,
+      "loss_xval": 0.4921875,
+      "num_input_tokens_seen": 7655320,
+      "step": 52
+    },
+    {
+      "epoch": 0.015278178149322572,
+      "grad_norm": 91.05016392033016,
+      "learning_rate": 7.190654742675073e-05,
+      "loss": 2.5477,
+      "num_input_tokens_seen": 7790448,
+      "step": 53
+    },
+    {
+      "epoch": 0.015278178149322572,
+      "loss": 2.598111391067505,
+      "loss_ce": 2.440823793411255,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 7790448,
+      "step": 53
+    },
+    {
+      "epoch": 0.015566445661573941,
+      "grad_norm": 144.81034266553996,
+      "learning_rate": 7.224508342818619e-05,
+      "loss": 2.71,
+      "num_input_tokens_seen": 7962992,
+      "step": 54
+    },
+    {
+      "epoch": 0.015566445661573941,
+      "loss": 2.4725775718688965,
+      "loss_ce": 2.2418036460876465,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 7962992,
+      "step": 54
+    },
+    {
+      "epoch": 0.01585471317382531,
+      "grad_norm": 232.08300955372366,
+      "learning_rate": 7.25774074080379e-05,
+      "loss": 2.7493,
+      "num_input_tokens_seen": 8097832,
+      "step": 55
+    },
+    {
+      "epoch": 0.01585471317382531,
+      "loss": 2.8998870849609375,
+      "loss_ce": 2.3862152099609375,
+      "loss_xval": 0.515625,
+      "num_input_tokens_seen": 8097832,
+      "step": 55
+    },
+    {
+      "epoch": 0.01614298068607668,
+      "grad_norm": 312.2394215634939,
+      "learning_rate": 7.290374324144727e-05,
+      "loss": 2.719,
+      "num_input_tokens_seen": 8232800,
+      "step": 56
+    },
+    {
+      "epoch": 0.01614298068607668,
+      "loss": 2.6347908973693848,
+      "loss_ce": 2.1626229286193848,
+      "loss_xval": 0.47265625,
+      "num_input_tokens_seen": 8232800,
+      "step": 56
+    },
+    {
+      "epoch": 0.016431248198328047,
+      "grad_norm": 235.52491076478066,
+      "learning_rate": 7.322430291510432e-05,
+      "loss": 2.5759,
+      "num_input_tokens_seen": 8405208,
+      "step": 57
+    },
+    {
+      "epoch": 0.016431248198328047,
+      "loss": 2.5117573738098145,
+      "loss_ce": 2.0200581550598145,
+      "loss_xval": 0.4921875,
+      "num_input_tokens_seen": 8405208,
+      "step": 57
+    },
+    {
+      "epoch": 0.01671951571057942,
+      "grad_norm": 63.56214683438855,
+      "learning_rate": 7.353928735437845e-05,
+      "loss": 2.2731,
+      "num_input_tokens_seen": 8539984,
+      "step": 58
+    },
+    {
+      "epoch": 0.01671951571057942,
+      "loss": 2.37797212600708,
+      "loss_ce": 2.14963960647583,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 8539984,
+      "step": 58
+    },
+    {
+      "epoch": 0.017007783222830787,
+      "grad_norm": 37.67195118941927,
+      "learning_rate": 7.38488871797435e-05,
+      "loss": 2.0322,
+      "num_input_tokens_seen": 8675088,
+      "step": 59
+    },
+    {
+      "epoch": 0.017007783222830787,
+      "loss": 2.0352678298950195,
+      "loss_ce": 1.9146623611450195,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 8675088,
+      "step": 59
+    },
+    {
+      "epoch": 0.017296050735082155,
+      "grad_norm": 167.7360509976244,
+      "learning_rate": 7.415328339962814e-05,
+      "loss": 2.2125,
+      "num_input_tokens_seen": 8847512,
+      "step": 60
+    },
+    {
+      "epoch": 0.017296050735082155,
+      "loss": 2.0584282875061035,
+      "loss_ce": 1.800371766090393,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 8847512,
+      "step": 60
+    },
+    {
+      "epoch": 0.017584318247333527,
+      "grad_norm": 250.3157839443666,
+      "learning_rate": 7.445264804599805e-05,
+      "loss": 2.318,
+      "num_input_tokens_seen": 8982392,
+      "step": 61
+    },
+    {
+      "epoch": 0.017584318247333527,
+      "loss": 2.417539596557617,
+      "loss_ce": 1.9382915496826172,
+      "loss_xval": 0.478515625,
+      "num_input_tokens_seen": 8982392,
+      "step": 61
+    },
+    {
+      "epoch": 0.017872585759584895,
+      "grad_norm": 178.8772377202071,
+      "learning_rate": 7.474714475825724e-05,
+      "loss": 1.9919,
+      "num_input_tokens_seen": 9117408,
+      "step": 62
+    },
+    {
+      "epoch": 0.017872585759584895,
+      "loss": 2.056333303451538,
+      "loss_ce": 1.7208842039108276,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 9117408,
+      "step": 62
+    },
+    {
+      "epoch": 0.018160853271836263,
+      "grad_norm": 92.49710835233134,
+      "learning_rate": 7.503692932043134e-05,
+      "loss": 1.91,
+      "num_input_tokens_seen": 9289880,
+      "step": 63
+    },
+    {
+      "epoch": 0.018160853271836263,
+      "loss": 1.787151575088501,
+      "loss_ce": 1.618694543838501,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 9289880,
+      "step": 63
+    },
+    {
+      "epoch": 0.018449120784087635,
+      "grad_norm": 36.307905662307306,
+      "learning_rate": 7.532215015604735e-05,
+      "loss": 1.814,
+      "num_input_tokens_seen": 9424680,
+      "step": 64
+    },
+    {
+      "epoch": 0.018449120784087635,
+      "loss": 1.890938401222229,
+      "loss_ce": 1.742561936378479,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 9424680,
+      "step": 64
+    },
+    {
+      "epoch": 0.018737388296339003,
+      "grad_norm": 188.18119861000676,
+      "learning_rate": 7.560294878464597e-05,
+      "loss": 1.8272,
+      "num_input_tokens_seen": 9559816,
+      "step": 65
+    },
+    {
+      "epoch": 0.018737388296339003,
+      "loss": 1.7736570835113525,
+      "loss_ce": 1.5479490756988525,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 9559816,
+      "step": 65
+    },
+    {
+      "epoch": 0.01902565580859037,
+      "grad_norm": 159.85223743496317,
+      "learning_rate": 7.587946024344119e-05,
+      "loss": 1.834,
+      "num_input_tokens_seen": 9732376,
+      "step": 66
+    },
+    {
+      "epoch": 0.01902565580859037,
+      "loss": 1.7517627477645874,
+      "loss_ce": 1.4713672399520874,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 9732376,
+      "step": 66
+    },
+    {
+      "epoch": 0.019313923320841742,
+      "grad_norm": 101.15058869594668,
+      "learning_rate": 7.615181347727268e-05,
+      "loss": 1.7053,
+      "num_input_tokens_seen": 9867176,
+      "step": 67
+    },
+    {
+      "epoch": 0.019313923320841742,
+      "loss": 1.7496159076690674,
+      "loss_ce": 1.5835392475128174,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 9867176,
+      "step": 67
+    },
+    {
+      "epoch": 0.01960219083309311,
+      "grad_norm": 58.32331425616856,
+      "learning_rate": 7.642013169966947e-05,
+      "loss": 1.543,
+      "num_input_tokens_seen": 10002352,
+      "step": 68
+    },
+    {
+      "epoch": 0.01960219083309311,
+      "loss": 1.5576841831207275,
+      "loss_ce": 1.4476988315582275,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 10002352,
+      "step": 68
+    },
+    {
+      "epoch": 0.01989045834534448,
+      "grad_norm": 44.59367162259951,
+      "learning_rate": 7.66845327275555e-05,
+      "loss": 1.5735,
+      "num_input_tokens_seen": 10174928,
+      "step": 69
+    },
+    {
+      "epoch": 0.01989045834534448,
+      "loss": 1.4471509456634521,
+      "loss_ce": 1.3455884456634521,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 10174928,
+      "step": 69
+    },
+    {
+      "epoch": 0.02017872585759585,
+      "grad_norm": 140.15921471971498,
+      "learning_rate": 7.69451292918733e-05,
+      "loss": 1.6784,
+      "num_input_tokens_seen": 10309696,
+      "step": 70
+    },
+    {
+      "epoch": 0.02017872585759585,
+      "loss": 1.7239234447479248,
+      "loss_ce": 1.4675757884979248,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 10309696,
+      "step": 70
+    },
+    {
+      "epoch": 0.020466993369847218,
+      "grad_norm": 121.97678463125082,
+      "learning_rate": 7.720202932617523e-05,
+      "loss": 1.4855,
+      "num_input_tokens_seen": 10444840,
+      "step": 71
+    },
+    {
+      "epoch": 0.020466993369847218,
+      "loss": 1.5414938926696777,
+      "loss_ce": 1.3424582481384277,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 10444840,
+      "step": 71
+    },
+    {
+      "epoch": 0.020755260882098586,
+      "grad_norm": 71.6510231372808,
+      "learning_rate": 7.745533623503144e-05,
+      "loss": 1.4846,
+      "num_input_tokens_seen": 10617216,
+      "step": 72
+    },
+    {
+      "epoch": 0.020755260882098586,
+      "loss": 1.3716990947723389,
+      "loss_ce": 1.2538402080535889,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 10617216,
+      "step": 72
+    },
+    {
+      "epoch": 0.021043528394349958,
+      "grad_norm": 17.112653827119473,
+      "learning_rate": 7.770514914392505e-05,
+      "loss": 1.4478,
+      "num_input_tokens_seen": 10752048,
+      "step": 73
+    },
+    {
+      "epoch": 0.021043528394349958,
+      "loss": 1.5130277872085571,
+      "loss_ce": 1.3716398477554321,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 10752048,
+      "step": 73
+    },
+    {
+      "epoch": 0.021331795906601326,
+      "grad_norm": 54.30781148516282,
+      "learning_rate": 7.795156313214625e-05,
+      "loss": 1.3234,
+      "num_input_tokens_seen": 10887048,
+      "step": 74
+    },
+    {
+      "epoch": 0.021331795906601326,
+      "loss": 1.3435890674591064,
+      "loss_ce": 1.2606728076934814,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 10887048,
+      "step": 74
+    },
+    {
+      "epoch": 0.021620063418852694,
+      "grad_norm": 126.25618466514295,
+      "learning_rate": 7.819466945005417e-05,
+      "loss": 1.4311,
+      "num_input_tokens_seen": 11059536,
+      "step": 75
+    },
+    {
+      "epoch": 0.021620063418852694,
+      "loss": 1.3155971765518188,
+      "loss_ce": 1.1788784265518188,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 11059536,
+      "step": 75
+    },
+    {
+      "epoch": 0.021908330931104066,
+      "grad_norm": 71.84571532321051,
+      "learning_rate": 7.843455572194957e-05,
+      "loss": 1.3668,
+      "num_input_tokens_seen": 11194248,
+      "step": 76
+    },
+    {
+      "epoch": 0.021908330931104066,
+      "loss": 1.4305455684661865,
+      "loss_ce": 1.2662389278411865,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 11194248,
+      "step": 76
+    },
+    {
+      "epoch": 0.022196598443355434,
+      "grad_norm": 18.12151091821412,
+      "learning_rate": 7.867130613568635e-05,
+      "loss": 1.2159,
+      "num_input_tokens_seen": 11329488,
+      "step": 77
+    },
+    {
+      "epoch": 0.022196598443355434,
+      "loss": 1.2611539363861084,
+      "loss_ce": 1.1754605770111084,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 11329488,
+      "step": 77
+    },
+    {
+      "epoch": 0.022484865955606802,
+      "grad_norm": 16.94095251304994,
+      "learning_rate": 7.890500162004926e-05,
+      "loss": 1.2511,
+      "num_input_tokens_seen": 11501960,
+      "step": 78
+    },
+    {
+      "epoch": 0.022484865955606802,
+      "loss": 1.1699042320251465,
+      "loss_ce": 1.1077704429626465,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 11501960,
+      "step": 78
+    },
+    {
+      "epoch": 0.022773133467858173,
+      "grad_norm": 49.28829387349209,
+      "learning_rate": 7.913572001083273e-05,
+      "loss": 1.2514,
+      "num_input_tokens_seen": 11636752,
+      "step": 79
+    },
+    {
+      "epoch": 0.022773133467858173,
+      "loss": 1.2829480171203613,
+      "loss_ce": 1.1588635444641113,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 11636752,
+      "step": 79
+    },
+    {
+      "epoch": 0.02306140098010954,
+      "grad_norm": 86.79137128570692,
+      "learning_rate": 7.93635362064734e-05,
+      "loss": 1.1792,
+      "num_input_tokens_seen": 11771712,
+      "step": 80
+    },
+    {
+      "epoch": 0.02306140098010954,
+      "loss": 1.2060190439224243,
+      "loss_ce": 1.0835214853286743,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 11771712,
+      "step": 80
+    },
+    {
+      "epoch": 0.02334966849236091,
+      "grad_norm": 36.119976096556634,
+      "learning_rate": 7.958852231401551e-05,
+      "loss": 1.1689,
+      "num_input_tokens_seen": 11944424,
+      "step": 81
+    },
+    {
+      "epoch": 0.02334966849236091,
+      "loss": 1.1105940341949463,
+      "loss_ce": 1.0218183994293213,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 11944424,
+      "step": 81
+    },
+    {
+      "epoch": 0.02363793600461228,
+      "grad_norm": 12.644434454126534,
+      "learning_rate": 7.981074778612054e-05,
+      "loss": 1.1655,
+      "num_input_tokens_seen": 12079392,
+      "step": 82
+    },
+    {
+      "epoch": 0.02363793600461228,
+      "loss": 1.2264204025268555,
+      "loss_ce": 1.1139326095581055,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 12079392,
+      "step": 82
+    },
+    {
+      "epoch": 0.02392620351686365,
+      "grad_norm": 15.168981426164274,
+      "learning_rate": 8.003027954977264e-05,
+      "loss": 1.0712,
+      "num_input_tokens_seen": 12214464,
+      "step": 83
+    },
+    {
+      "epoch": 0.02392620351686365,
+      "loss": 1.1029493808746338,
+      "loss_ce": 1.0238783359527588,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 12214464,
+      "step": 83
+    },
+    {
+      "epoch": 0.024214471029115017,
+      "grad_norm": 81.32602157915647,
+      "learning_rate": 8.024718212727658e-05,
+      "loss": 1.1252,
+      "num_input_tokens_seen": 12387112,
+      "step": 84
+    },
+    {
+      "epoch": 0.024214471029115017,
+      "loss": 1.02632737159729,
+      "loss_ce": 0.9486294984817505,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 12387112,
+      "step": 84
+    },
+    {
+      "epoch": 0.02450273854136639,
+      "grad_norm": 43.813459977991975,
+      "learning_rate": 8.04615177500955e-05,
+      "loss": 1.1105,
+      "num_input_tokens_seen": 12521856,
+      "step": 85
+    },
+    {
+      "epoch": 0.02450273854136639,
+      "loss": 1.1658377647399902,
+      "loss_ce": 1.0337576866149902,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 12521856,
+      "step": 85
+    },
+    {
+      "epoch": 0.024791006053617757,
+      "grad_norm": 11.656507104084508,
+      "learning_rate": 8.067334646603104e-05,
+      "loss": 0.9894,
+      "num_input_tokens_seen": 12657040,
+      "step": 86
+    },
+    {
+      "epoch": 0.024791006053617757,
+      "loss": 1.0080392360687256,
+      "loss_ce": 0.9314705729484558,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 12657040,
+      "step": 86
+    },
+    {
+      "epoch": 0.025079273565869125,
+      "grad_norm": 12.676570492524768,
+      "learning_rate": 8.088272624020777e-05,
+      "loss": 1.025,
+      "num_input_tokens_seen": 12829568,
+      "step": 87
+    },
+    {
+      "epoch": 0.025079273565869125,
+      "loss": 0.9655556678771973,
+      "loss_ce": 0.9059548377990723,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 12829568,
+      "step": 87
+    },
+    {
+      "epoch": 0.025367541078120497,
+      "grad_norm": 40.00413507298051,
+      "learning_rate": 8.108971305028644e-05,
+      "loss": 1.0396,
+      "num_input_tokens_seen": 12964304,
+      "step": 88
+    },
+    {
+      "epoch": 0.025367541078120497,
+      "loss": 1.0686454772949219,
+      "loss_ce": 0.9660453796386719,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 12964304,
+      "step": 88
+    },
+    {
+      "epoch": 0.025655808590371865,
+      "grad_norm": 57.04743386781687,
+      "learning_rate": 8.129436097629779e-05,
+      "loss": 0.9546,
+      "num_input_tokens_seen": 13099336,
+      "step": 89
+    },
+    {
+      "epoch": 0.025655808590371865,
+      "loss": 0.9784692525863647,
+      "loss_ce": 0.8800195455551147,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 13099336,
+      "step": 89
+    },
+    {
+      "epoch": 0.025944076102623233,
+      "grad_norm": 11.991983730016383,
+      "learning_rate": 8.149672228545746e-05,
+      "loss": 0.9712,
+      "num_input_tokens_seen": 13271816,
+      "step": 90
+    },
+    {
+      "epoch": 0.025944076102623233,
+      "loss": 0.9001408815383911,
+      "loss_ce": 0.8415776491165161,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 13271816,
+      "step": 90
+    },
+    {
+      "epoch": 0.026232343614874604,
+      "grad_norm": 17.076861856400765,
+      "learning_rate": 8.16968475122944e-05,
+      "loss": 0.939,
+      "num_input_tokens_seen": 13406616,
+      "step": 91
+    },
+    {
+      "epoch": 0.026232343614874604,
+      "loss": 0.9755129218101501,
+      "loss_ce": 0.9001650214195251,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 13406616,
+      "step": 91
+    },
+    {
+      "epoch": 0.026520611127125972,
+      "grad_norm": 13.426874489381975,
+      "learning_rate": 8.189478553440074e-05,
+      "loss": 0.8669,
+      "num_input_tokens_seen": 13541608,
+      "step": 92
+    },
+    {
+      "epoch": 0.026520611127125972,
+      "loss": 0.8719437122344971,
+      "loss_ce": 0.8101913928985596,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 13541608,
+      "step": 92
+    },
+    {
+      "epoch": 0.02680887863937734,
+      "grad_norm": 45.10735406441355,
+      "learning_rate": 8.209058364408656e-05,
+      "loss": 0.9199,
+      "num_input_tokens_seen": 13714096,
+      "step": 93
+    },
+    {
+      "epoch": 0.02680887863937734,
+      "loss": 0.8327312469482422,
+      "loss_ce": 0.7779521942138672,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 13714096,
+      "step": 93
+    },
+    {
+      "epoch": 0.027097146151628712,
+      "grad_norm": 22.312072425981693,
+      "learning_rate": 8.228428761620285e-05,
+      "loss": 0.8977,
+      "num_input_tokens_seen": 13848928,
+      "step": 94
+    },
+    {
+      "epoch": 0.027097146151628712,
+      "loss": 0.9612225294113159,
+      "loss_ce": 0.8525189161300659,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 13848928,
+      "step": 94
+    },
+    {
+      "epoch": 0.02738541366388008,
+      "grad_norm": 14.377576713290015,
+      "learning_rate": 8.247594177237559e-05,
+      "loss": 0.8293,
+      "num_input_tokens_seen": 13984080,
+      "step": 95
+    },
+    {
+      "epoch": 0.02738541366388008,
+      "loss": 0.8508036732673645,
+      "loss_ce": 0.7708476185798645,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 13984080,
+      "step": 95
+    },
+    {
+      "epoch": 0.02767368117613145,
+      "grad_norm": 22.570750319320567,
+      "learning_rate": 8.266558904187667e-05,
+      "loss": 0.842,
+      "num_input_tokens_seen": 14156520,
+      "step": 96
+    },
+    {
+      "epoch": 0.02767368117613145,
+      "loss": 0.7710494995117188,
+      "loss_ce": 0.7205734252929688,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 14156520,
+      "step": 96
+    },
+    {
+      "epoch": 0.02796194868838282,
+      "grad_norm": 36.281544865910995,
+      "learning_rate": 8.285327101934069e-05,
+      "loss": 0.8365,
+      "num_input_tokens_seen": 14291312,
+      "step": 97
+    },
+    {
+      "epoch": 0.02796194868838282,
+      "loss": 0.8652549982070923,
+      "loss_ce": 0.7860618829727173,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 14291312,
+      "step": 97
+    },
+    {
+      "epoch": 0.028250216200634188,
+      "grad_norm": 10.406571859201414,
+      "learning_rate": 8.303902801952174e-05,
+      "loss": 0.7746,
+      "num_input_tokens_seen": 14426480,
+      "step": 98
+    },
+    {
+      "epoch": 0.028250216200634188,
+      "loss": 0.8105872869491577,
+      "loss_ce": 0.7177222967147827,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 14426480,
+      "step": 98
+    },
+    {
+      "epoch": 0.02853848371288556,
+      "grad_norm": 11.45020509791116,
+      "learning_rate": 8.322289912927049e-05,
+      "loss": 0.7788,
+      "num_input_tokens_seen": 14598976,
+      "step": 99
+    },
+    {
+      "epoch": 0.02853848371288556,
+      "loss": 0.7215403318405151,
+      "loss_ce": 0.6690806150436401,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 14598976,
+      "step": 99
+    },
+    {
+      "epoch": 0.028826751225136928,
+      "grad_norm": 33.00790033062601,
+      "learning_rate": 8.340492225689943e-05,
+      "loss": 0.7793,
+      "num_input_tokens_seen": 14733720,
+      "step": 100
+    },
+    {
+      "epoch": 0.028826751225136928,
+      "loss": 0.8156184554100037,
+      "loss_ce": 0.7315425276756287,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 14733720,
+      "step": 100
+    },
+    {
+      "epoch": 0.029115018737388296,
+      "grad_norm": 20.940949519344258,
+      "learning_rate": 8.358513417909158e-05,
+      "loss": 0.694,
+      "num_input_tokens_seen": 14868704,
+      "step": 101
+    },
+    {
+      "epoch": 0.029115018737388296,
+      "loss": 0.7094327807426453,
+      "loss_ce": 0.6524564623832703,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 14868704,
+      "step": 101
+    },
+    {
+      "epoch": 0.029403286249639667,
+      "grad_norm": 11.7677303756099,
+      "learning_rate": 8.376357058549878e-05,
+      "loss": 0.7215,
+      "num_input_tokens_seen": 15041320,
+      "step": 102
+    },
+    {
+      "epoch": 0.029403286249639667,
+      "loss": 0.6571950912475586,
+      "loss_ce": 0.6112051010131836,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 15041320,
+      "step": 102
+    },
+    {
+      "epoch": 0.029691553761891035,
+      "grad_norm": 13.477928271939529,
+      "learning_rate": 8.394026612116405e-05,
+      "loss": 0.7139,
+      "num_input_tokens_seen": 15176128,
+      "step": 103
+    },
+    {
+      "epoch": 0.029691553761891035,
+      "loss": 0.7441321611404419,
+      "loss_ce": 0.6728125810623169,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 15176128,
+      "step": 103
+    },
+    {
+      "epoch": 0.029979821274142403,
+      "grad_norm": 28.21428885187996,
+      "learning_rate": 8.41152544268945e-05,
+      "loss": 0.6557,
+      "num_input_tokens_seen": 15311280,
+      "step": 104
+    },
+    {
+      "epoch": 0.029979821274142403,
+      "loss": 0.6761811971664429,
+      "loss_ce": 0.6040834188461304,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 15311280,
+      "step": 104
+    },
+    {
+      "epoch": 0.030268088786393775,
+      "grad_norm": 10.481020095069384,
+      "learning_rate": 8.42885681777026e-05,
+      "loss": 0.6775,
+      "num_input_tokens_seen": 15483784,
+      "step": 105
+    },
+    {
+      "epoch": 0.030268088786393775,
+      "loss": 0.6159076690673828,
+      "loss_ce": 0.5645313262939453,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 15483784,
+      "step": 105
+    },
+    {
+      "epoch": 0.030556356298645143,
+      "grad_norm": 12.736851354361377,
+      "learning_rate": 8.446023911942528e-05,
+      "loss": 0.6781,
+      "num_input_tokens_seen": 15618472,
+      "step": 106
+    },
+    {
+      "epoch": 0.030556356298645143,
+      "loss": 0.6953324675559998,
+      "loss_ce": 0.6300248503684998,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 15618472,
+      "step": 106
+    },
+    {
+      "epoch": 0.03084462381089651,
+      "grad_norm": 15.301730909445215,
+      "learning_rate": 8.463029810362388e-05,
+      "loss": 0.6077,
+      "num_input_tokens_seen": 15753616,
+      "step": 107
+    },
+    {
+      "epoch": 0.03084462381089651,
+      "loss": 0.6367394924163818,
+      "loss_ce": 0.5607202053070068,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 15753616,
+      "step": 107
+    },
+    {
+      "epoch": 0.031132891323147883,
+      "grad_norm": 30.79147422211605,
+      "learning_rate": 8.479877512086075e-05,
+      "loss": 0.617,
+      "num_input_tokens_seen": 15926144,
+      "step": 108
+    },
+    {
+      "epoch": 0.031132891323147883,
+      "loss": 0.5589581727981567,
+      "loss_ce": 0.5138074159622192,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 15926144,
+      "step": 108
+    },
+    {
+      "epoch": 0.03142115883539925,
+      "grad_norm": 8.250685702158895,
+      "learning_rate": 8.496569933244227e-05,
+      "loss": 0.6087,
+      "num_input_tokens_seen": 16061008,
+      "step": 109
+    },
+    {
+      "epoch": 0.03142115883539925,
+      "loss": 0.631111741065979,
+      "loss_ce": 0.564949631690979,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 16061008,
+      "step": 109
+    },
+    {
+      "epoch": 0.03170942634765062,
+      "grad_norm": 23.093026007657294,
+      "learning_rate": 8.513109910071247e-05,
+      "loss": 0.5623,
+      "num_input_tokens_seen": 16196128,
+      "step": 110
+    },
+    {
+      "epoch": 0.03170942634765062,
+      "loss": 0.5758240222930908,
+      "loss_ce": 0.5295593738555908,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 16196128,
+      "step": 110
+    },
+    {
+      "epoch": 0.03199769385990199,
+      "grad_norm": 19.06137292165344,
+      "learning_rate": 8.529500201797555e-05,
+      "loss": 0.5789,
+      "num_input_tokens_seen": 16368512,
+      "step": 111
+    },
+    {
+      "epoch": 0.03199769385990199,
+      "loss": 0.5283578634262085,
+      "loss_ce": 0.4861825108528137,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 16368512,
+      "step": 111
+    },
+    {
+      "epoch": 0.03228596137215336,
+      "grad_norm": 10.639807503403174,
+      "learning_rate": 8.545743493412182e-05,
+      "loss": 0.5752,
+      "num_input_tokens_seen": 16503288,
+      "step": 112
+    },
+    {
+      "epoch": 0.03228596137215336,
+      "loss": 0.5936908721923828,
+      "loss_ce": 0.5266132950782776,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 16503288,
+      "step": 112
+    },
+    {
+      "epoch": 0.03257422888440473,
+      "grad_norm": 16.230556762656512,
+      "learning_rate": 8.561842398302536e-05,
+      "loss": 0.5299,
+      "num_input_tokens_seen": 16638280,
+      "step": 113
+    },
+    {
+      "epoch": 0.03257422888440473,
+      "loss": 0.5637577772140503,
+      "loss_ce": 0.5001896619796753,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 16638280,
+      "step": 113
+    },
+    {
+      "epoch": 0.032862496396656095,
+      "grad_norm": 20.814059821840747,
+      "learning_rate": 8.577799460777888e-05,
+      "loss": 0.5511,
+      "num_input_tokens_seen": 16810784,
+      "step": 114
+    },
+    {
+      "epoch": 0.032862496396656095,
+      "loss": 0.5012564659118652,
+      "loss_ce": 0.4504447281360626,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 16810784,
+      "step": 114
+    },
+    {
+      "epoch": 0.03315076390890746,
+      "grad_norm": 7.387134587767617,
+      "learning_rate": 8.593617158482676e-05,
+      "loss": 0.5444,
+      "num_input_tokens_seen": 16945520,
+      "step": 115
+    },
+    {
+      "epoch": 0.03315076390890746,
+      "loss": 0.560407280921936,
+      "loss_ce": 0.507184624671936,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 16945520,
+      "step": 115
+    },
+    {
+      "epoch": 0.03343903142115884,
+      "grad_norm": 6.797458685600256,
+      "learning_rate": 8.609297904705301e-05,
+      "loss": 0.5066,
+      "num_input_tokens_seen": 17080640,
+      "step": 116
+    },
+    {
+      "epoch": 0.03343903142115884,
+      "loss": 0.5233986377716064,
+      "loss_ce": 0.47078636288642883,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 17080640,
+      "step": 116
+    },
+    {
+      "epoch": 0.033727298933410206,
+      "grad_norm": 18.86772578309578,
+      "learning_rate": 8.624844050587858e-05,
+      "loss": 0.5119,
+      "num_input_tokens_seen": 17253208,
+      "step": 117
+    },
+    {
+      "epoch": 0.033727298933410206,
+      "loss": 0.47286179661750793,
+      "loss_ce": 0.42783311009407043,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 17253208,
+      "step": 117
+    },
+    {
+      "epoch": 0.034015566445661574,
+      "grad_norm": 8.023085850245478,
+      "learning_rate": 8.640257887241806e-05,
+      "loss": 0.5306,
+      "num_input_tokens_seen": 17387920,
+      "step": 118
+    },
+    {
+      "epoch": 0.034015566445661574,
+      "loss": 0.5475594997406006,
+      "loss_ce": 0.4861886501312256,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 17387920,
+      "step": 118
+    },
+    {
+      "epoch": 0.03430383395791294,
+      "grad_norm": 6.890785610877055,
+      "learning_rate": 8.655541647774393e-05,
+      "loss": 0.4852,
+      "num_input_tokens_seen": 17522904,
+      "step": 119
+    },
+    {
+      "epoch": 0.03430383395791294,
+      "loss": 0.5079690217971802,
+      "loss_ce": 0.4557229280471802,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 17522904,
+      "step": 119
+    },
+    {
+      "epoch": 0.03459210147016431,
+      "grad_norm": 11.321006785298598,
+      "learning_rate": 8.67069750923027e-05,
+      "loss": 0.4889,
+      "num_input_tokens_seen": 17695320,
+      "step": 120
+    },
+    {
+      "epoch": 0.03459210147016431,
+      "loss": 0.44401293992996216,
+      "loss_ce": 0.40258532762527466,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 17695320,
+      "step": 120
+    },
+    {
+      "epoch": 0.03488036898241568,
+      "grad_norm": 13.101992884551423,
+      "learning_rate": 8.68572759445255e-05,
+      "loss": 0.5142,
+      "num_input_tokens_seen": 17830080,
+      "step": 121
+    },
+    {
+      "epoch": 0.03488036898241568,
+      "loss": 0.5295823812484741,
+      "loss_ce": 0.4632371664047241,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 17830080,
+      "step": 121
+    },
+    {
+      "epoch": 0.035168636494667053,
+      "grad_norm": 7.865552687957865,
+      "learning_rate": 8.700633973867262e-05,
+      "loss": 0.4699,
+      "num_input_tokens_seen": 17965200,
+      "step": 122
+    },
+    {
+      "epoch": 0.035168636494667053,
+      "loss": 0.49172335863113403,
+      "loss_ce": 0.4352964162826538,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 17965200,
+      "step": 122
+    },
+    {
+      "epoch": 0.03545690400691842,
+      "grad_norm": 8.34823959747774,
+      "learning_rate": 8.715418667194984e-05,
+      "loss": 0.4788,
+      "num_input_tokens_seen": 18137616,
+      "step": 123
+    },
+    {
+      "epoch": 0.03545690400691842,
+      "loss": 0.4347504675388336,
+      "loss_ce": 0.3913239538669586,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 18137616,
+      "step": 123
+    },
+    {
+      "epoch": 0.03574517151916979,
+      "grad_norm": 19.73006652507538,
+      "learning_rate": 8.73008364509318e-05,
+      "loss": 0.5042,
+      "num_input_tokens_seen": 18272416,
+      "step": 124
+    },
+    {
+      "epoch": 0.03574517151916979,
+      "loss": 0.5189793705940247,
+      "loss_ce": 0.44375354051589966,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 18272416,
+      "step": 124
+    },
+    {
+      "epoch": 0.03603343903142116,
+      "grad_norm": 7.744786447742085,
+      "learning_rate": 8.744630830732546e-05,
+      "loss": 0.4524,
+      "num_input_tokens_seen": 18407576,
+      "step": 125
+    },
+    {
+      "epoch": 0.03603343903142116,
+      "loss": 0.47686684131622314,
+      "loss_ce": 0.42333900928497314,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 18407576,
+      "step": 125
+    },
+    {
+      "epoch": 0.036321706543672526,
+      "grad_norm": 13.281162567225987,
+      "learning_rate": 8.75906210131059e-05,
+      "loss": 0.4611,
+      "num_input_tokens_seen": 18580112,
+      "step": 126
+    },
+    {
+      "epoch": 0.036321706543672526,
+      "loss": 0.41387027502059937,
+      "loss_ce": 0.37471622228622437,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 18580112,
+      "step": 126
+    },
+    {
+      "epoch": 0.036609974055923894,
+      "grad_norm": 15.583006378460968,
+      "learning_rate": 8.773379289505366e-05,
+      "loss": 0.4688,
+      "num_input_tokens_seen": 18714824,
+      "step": 127
+    },
+    {
+      "epoch": 0.036609974055923894,
+      "loss": 0.4860740303993225,
+      "loss_ce": 0.4178062081336975,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 18714824,
+      "step": 127
+    },
+    {
+      "epoch": 0.03689824156817527,
+      "grad_norm": 24.714679606403106,
+      "learning_rate": 8.787584184872191e-05,
+      "loss": 0.4365,
+      "num_input_tokens_seen": 18850080,
+      "step": 128
+    },
+    {
+      "epoch": 0.03689824156817527,
+      "loss": 0.45160603523254395,
+      "loss_ce": 0.40163350105285645,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 18850080,
+      "step": 128
+    },
+    {
+      "epoch": 0.03718650908042664,
+      "grad_norm": 8.236941080781937,
+      "learning_rate": 8.801678535186036e-05,
+      "loss": 0.4549,
+      "num_input_tokens_seen": 19022432,
+      "step": 129
+    },
+    {
+      "epoch": 0.03718650908042664,
+      "loss": 0.40648266673088074,
+      "loss_ce": 0.35817334055900574,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 19022432,
+      "step": 129
+    },
+    {
+      "epoch": 0.037474776592678005,
+      "grad_norm": 8.741920747317353,
+      "learning_rate": 8.815664047732053e-05,
+      "loss": 0.447,
+      "num_input_tokens_seen": 19157248,
+      "step": 130
+    },
+    {
+      "epoch": 0.037474776592678005,
+      "loss": 0.4431639313697815,
+      "loss_ce": 0.3914671540260315,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 19157248,
+      "step": 130
+    },
+    {
+      "epoch": 0.03776304410492937,
+      "grad_norm": 11.13044720072461,
+      "learning_rate": 8.829542390546686e-05,
+      "loss": 0.4074,
+      "num_input_tokens_seen": 19292208,
+      "step": 131
+    },
+    {
+      "epoch": 0.03776304410492937,
+      "loss": 0.43208760023117065,
+      "loss_ce": 0.38637226819992065,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 19292208,
+      "step": 131
+    },
+    {
+      "epoch": 0.03805131161718074,
+      "grad_norm": 7.227212476089253,
+      "learning_rate": 8.843315193611575e-05,
+      "loss": 0.4255,
+      "num_input_tokens_seen": 19464680,
+      "step": 132
+    },
+    {
+      "epoch": 0.03805131161718074,
+      "loss": 0.3858657479286194,
+      "loss_ce": 0.3452315926551819,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 19464680,
+      "step": 132
+    },
+    {
+      "epoch": 0.038339579129432116,
+      "grad_norm": 7.444913961645206,
+      "learning_rate": 8.856984050002403e-05,
+      "loss": 0.4414,
+      "num_input_tokens_seen": 19599448,
+      "step": 133
+    },
+    {
+      "epoch": 0.038339579129432116,
+      "loss": 0.4538487195968628,
+      "loss_ce": 0.3836582899093628,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 19599448,
+      "step": 133
+    },
+    {
+      "epoch": 0.038627846641683485,
+      "grad_norm": 8.496712439073113,
+      "learning_rate": 8.870550516994724e-05,
+      "loss": 0.3984,
+      "num_input_tokens_seen": 19734512,
+      "step": 134
+    },
+    {
+      "epoch": 0.038627846641683485,
+      "loss": 0.4241769313812256,
+      "loss_ce": 0.3713052272796631,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 19734512,
+      "step": 134
+    },
+    {
+      "epoch": 0.03891611415393485,
+      "grad_norm": 10.293290447905349,
+      "learning_rate": 8.884016117128679e-05,
+      "loss": 0.3947,
+      "num_input_tokens_seen": 19907056,
+      "step": 135
+    },
+    {
+      "epoch": 0.03891611415393485,
+      "loss": 0.3680115342140198,
+      "loss_ce": 0.3244324326515198,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 19907056,
+      "step": 135
+    },
+    {
+      "epoch": 0.03920438166618622,
+      "grad_norm": 7.220263516510419,
+      "learning_rate": 8.897382339234403e-05,
+      "loss": 0.4308,
+      "num_input_tokens_seen": 20041840,
+      "step": 136
+    },
+    {
+      "epoch": 0.03920438166618622,
+      "loss": 0.44666624069213867,
+      "loss_ce": 0.37360715866088867,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 20041840,
+      "step": 136
+    },
+    {
+      "epoch": 0.03949264917843759,
+      "grad_norm": 8.10053991289463,
+      "learning_rate": 8.910650639419907e-05,
+      "loss": 0.3892,
+      "num_input_tokens_seen": 20177016,
+      "step": 137
+    },
+    {
+      "epoch": 0.03949264917843759,
+      "loss": 0.4107305407524109,
+      "loss_ce": 0.3623601794242859,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 20177016,
+      "step": 137
+    },
+    {
+      "epoch": 0.03978091669068896,
+      "grad_norm": 17.842547960477276,
+      "learning_rate": 8.923822442023005e-05,
+      "loss": 0.3782,
+      "num_input_tokens_seen": 20349472,
+      "step": 138
+    },
+    {
+      "epoch": 0.03978091669068896,
+      "loss": 0.3521159589290619,
+      "loss_ce": 0.3164714276790619,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 20349472,
+      "step": 138
+    },
+    {
+      "epoch": 0.04006918420294033,
+      "grad_norm": 6.227925946167984,
+      "learning_rate": 8.936899140528881e-05,
+      "loss": 0.405,
+      "num_input_tokens_seen": 20484288,
+      "step": 139
+    },
+    {
+      "epoch": 0.04006918420294033,
+      "loss": 0.4243292212486267,
+      "loss_ce": 0.3574804663658142,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 20484288,
+      "step": 139
+    },
+    {
+      "epoch": 0.0403574517151917,
+      "grad_norm": 19.35797505142977,
+      "learning_rate": 8.949882098454784e-05,
+      "loss": 0.3686,
+      "num_input_tokens_seen": 20619376,
+      "step": 140
+    },
+    {
+      "epoch": 0.0403574517151917,
+      "loss": 0.388784259557724,
+      "loss_ce": 0.341237872838974,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 20619376,
+      "step": 140
+    },
+    {
+      "epoch": 0.04064571922744307,
+      "grad_norm": 17.99154294533077,
+      "learning_rate": 8.962772650203216e-05,
+      "loss": 0.378,
+      "num_input_tokens_seen": 20791928,
+      "step": 141
+    },
+    {
+      "epoch": 0.04064571922744307,
+      "loss": 0.33298397064208984,
+      "loss_ce": 0.29975032806396484,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 20791928,
+      "step": 141
+    },
+    {
+      "epoch": 0.040933986739694436,
+      "grad_norm": 16.729983956716715,
+      "learning_rate": 8.97557210188498e-05,
+      "loss": 0.3874,
+      "num_input_tokens_seen": 20926808,
+      "step": 142
+    },
+    {
+      "epoch": 0.040933986739694436,
+      "loss": 0.39781898260116577,
+      "loss_ce": 0.34123939275741577,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 20926808,
+      "step": 142
+    },
+    {
+      "epoch": 0.041222254251945804,
+      "grad_norm": 20.555304511946005,
+      "learning_rate": 8.988281732113356e-05,
+      "loss": 0.3588,
+      "num_input_tokens_seen": 21061904,
+      "step": 143
+    },
+    {
+      "epoch": 0.041222254251945804,
+      "loss": 0.3827919661998749,
+      "loss_ce": 0.3249458968639374,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 21061904,
+      "step": 143
+    },
+    {
+      "epoch": 0.04151052176419717,
+      "grad_norm": 11.077156251039169,
+      "learning_rate": 9.000902792770599e-05,
+      "loss": 0.3618,
+      "num_input_tokens_seen": 21234296,
+      "step": 144
+    },
+    {
+      "epoch": 0.04151052176419717,
+      "loss": 0.3408665657043457,
+      "loss_ce": 0.2942509651184082,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 21234296,
+      "step": 144
+    },
+    {
+      "epoch": 0.04179878927644855,
+      "grad_norm": 7.537437415570367,
+      "learning_rate": 9.013436509747905e-05,
+      "loss": 0.3696,
+      "num_input_tokens_seen": 21369176,
+      "step": 145
+    },
+    {
+      "epoch": 0.04179878927644855,
+      "loss": 0.37368881702423096,
+      "loss_ce": 0.31585800647735596,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 21369176,
+      "step": 145
+    },
+    {
+      "epoch": 0.042087056788699916,
+      "grad_norm": 22.75373071527674,
+      "learning_rate": 9.025884083659961e-05,
+      "loss": 0.3445,
+      "num_input_tokens_seen": 21504168,
+      "step": 146
+    },
+    {
+      "epoch": 0.042087056788699916,
+      "loss": 0.3730278015136719,
+      "loss_ce": 0.3152580261230469,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 21504168,
+      "step": 146
+    },
+    {
+      "epoch": 0.042375324300951284,
+      "grad_norm": 8.750173390173343,
+      "learning_rate": 9.038246690535104e-05,
+      "loss": 0.3446,
+      "num_input_tokens_seen": 21676760,
+      "step": 147
+    },
+    {
+      "epoch": 0.042375324300951284,
+      "loss": 0.319990873336792,
+      "loss_ce": 0.279219388961792,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 21676760,
+      "step": 147
+    },
+    {
+      "epoch": 0.04266359181320265,
+      "grad_norm": 26.65442808332653,
+      "learning_rate": 9.050525482482079e-05,
+      "loss": 0.3634,
+      "num_input_tokens_seen": 21811504,
+      "step": 148
+    },
+    {
+      "epoch": 0.04266359181320265,
+      "loss": 0.3779270052909851,
+      "loss_ce": 0.3187229037284851,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 21811504,
+      "step": 148
+    },
+    {
+      "epoch": 0.04295185932545402,
+      "grad_norm": 9.765345028862136,
+      "learning_rate": 9.062721588334353e-05,
+      "loss": 0.328,
+      "num_input_tokens_seen": 21946624,
+      "step": 149
+    },
+    {
+      "epoch": 0.04295185932545402,
+      "loss": 0.35912925004959106,
+      "loss_ce": 0.31243735551834106,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 21946624,
+      "step": 149
+    },
+    {
+      "epoch": 0.04324012683770539,
+      "grad_norm": 19.40125110129968,
+      "learning_rate": 9.074836114272874e-05,
+      "loss": 0.3332,
+      "num_input_tokens_seen": 22119168,
+      "step": 150
+    },
+    {
+      "epoch": 0.04324012683770539,
+      "loss": 0.30681443214416504,
+      "loss_ce": 0.26843857765197754,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 22119168,
+      "step": 150
+    },
+    {
+      "epoch": 0.04352839434995676,
+      "grad_norm": 28.48434500578458,
+      "learning_rate": 9.086870144428141e-05,
+      "loss": 0.3599,
+      "num_input_tokens_seen": 22254016,
+      "step": 151
+    },
+    {
+      "epoch": 0.04352839434995676,
+      "loss": 0.37152981758117676,
+      "loss_ce": 0.30808377265930176,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 22254016,
+      "step": 151
+    },
+    {
+      "epoch": 0.04381666186220813,
+      "grad_norm": 7.423383361353423,
+      "learning_rate": 9.098824741462412e-05,
+      "loss": 0.318,
+      "num_input_tokens_seen": 22389096,
+      "step": 152
+    },
+    {
+      "epoch": 0.04381666186220813,
+      "loss": 0.35003936290740967,
+      "loss_ce": 0.29791533946990967,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 22389096,
+      "step": 152
+    },
+    {
+      "epoch": 0.0441049293744595,
+      "grad_norm": 27.525794406351512,
+      "learning_rate": 9.110700947132808e-05,
+      "loss": 0.3236,
+      "num_input_tokens_seen": 22561800,
+      "step": 153
+    },
+    {
+      "epoch": 0.0441049293744595,
+      "loss": 0.29091864824295044,
+      "loss_ce": 0.25377875566482544,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 22561800,
+      "step": 153
+    },
+    {
+      "epoch": 0.04439319688671087,
+      "grad_norm": 11.347675941007187,
+      "learning_rate": 9.12249978283609e-05,
+      "loss": 0.3372,
+      "num_input_tokens_seen": 22696600,
+      "step": 154
+    },
+    {
+      "epoch": 0.04439319688671087,
+      "loss": 0.34830331802368164,
+      "loss_ce": 0.28949594497680664,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 22696600,
+      "step": 154
+    },
+    {
+      "epoch": 0.044681464398962235,
+      "grad_norm": 14.974404743639585,
+      "learning_rate": 9.134222250135783e-05,
+      "loss": 0.3118,
+      "num_input_tokens_seen": 22831720,
+      "step": 155
+    },
+    {
+      "epoch": 0.044681464398962235,
+      "loss": 0.33259499073028564,
+      "loss_ce": 0.28230202198028564,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 22831720,
+      "step": 155
+    },
+    {
+      "epoch": 0.044969731911213603,
+      "grad_norm": 14.382221862767326,
+      "learning_rate": 9.145869331272382e-05,
+      "loss": 0.3275,
+      "num_input_tokens_seen": 23004192,
+      "step": 156
+    },
+    {
+      "epoch": 0.044969731911213603,
+      "loss": 0.2943967580795288,
+      "loss_ce": 0.2529538869857788,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 23004192,
+      "step": 156
+    },
+    {
+      "epoch": 0.04525799942346498,
+      "grad_norm": 9.718539618915857,
+      "learning_rate": 9.157441989657229e-05,
+      "loss": 0.326,
+      "num_input_tokens_seen": 23138984,
+      "step": 157
+    },
+    {
+      "epoch": 0.04525799942346498,
+      "loss": 0.324166864156723,
+      "loss_ce": 0.26999813318252563,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 23138984,
+      "step": 157
+    },
+    {
+      "epoch": 0.04554626693571635,
+      "grad_norm": 26.006683493223683,
+      "learning_rate": 9.168941170350729e-05,
+      "loss": 0.3042,
+      "num_input_tokens_seen": 23274144,
+      "step": 158
+    },
+    {
+      "epoch": 0.04554626693571635,
+      "loss": 0.31835854053497314,
+      "loss_ce": 0.27255165576934814,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 23274144,
+      "step": 158
+    },
+    {
+      "epoch": 0.045834534447967715,
+      "grad_norm": 6.575263387107517,
+      "learning_rate": 9.18036780052546e-05,
+      "loss": 0.3047,
+      "num_input_tokens_seen": 23446704,
+      "step": 159
+    },
+    {
+      "epoch": 0.045834534447967715,
+      "loss": 0.2826117277145386,
+      "loss_ce": 0.24026857316493988,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 23446704,
+      "step": 159
+    },
+    {
+      "epoch": 0.04612280196021908,
+      "grad_norm": 22.489834908558066,
+      "learning_rate": 9.191722789914795e-05,
+      "loss": 0.3228,
+      "num_input_tokens_seen": 23581568,
+      "step": 160
+    },
+    {
+      "epoch": 0.04612280196021908,
+      "loss": 0.32048097252845764,
+      "loss_ce": 0.26686158776283264,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 23581568,
+      "step": 160
+    },
+    {
+      "epoch": 0.04641106947247045,
+      "grad_norm": 8.378774904416845,
+      "learning_rate": 9.203007031247519e-05,
+      "loss": 0.2805,
+      "num_input_tokens_seen": 23716744,
+      "step": 161
+    },
+    {
+      "epoch": 0.04641106947247045,
+      "loss": 0.29976916313171387,
+      "loss_ce": 0.25937914848327637,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 23716744,
+      "step": 161
+    },
+    {
+      "epoch": 0.04669933698472182,
+      "grad_norm": 7.8011862892598565,
+      "learning_rate": 9.214221400669005e-05,
+      "loss": 0.2955,
+      "num_input_tokens_seen": 23889208,
+      "step": 162
+    },
+    {
+      "epoch": 0.04669933698472182,
+      "loss": 0.25634777545928955,
+      "loss_ce": 0.22622694075107574,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 23889208,
+      "step": 162
+    },
+    {
+      "epoch": 0.046987604496973194,
+      "grad_norm": 37.03342557004555,
+      "learning_rate": 9.225366758149434e-05,
+      "loss": 0.3199,
+      "num_input_tokens_seen": 24023880,
+      "step": 163
+    },
+    {
+      "epoch": 0.046987604496973194,
+      "loss": 0.3286735713481903,
+      "loss_ce": 0.2644645869731903,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 24023880,
+      "step": 163
+    },
+    {
+      "epoch": 0.04727587200922456,
+      "grad_norm": 10.660900550482202,
+      "learning_rate": 9.236443947879511e-05,
+      "loss": 0.285,
+      "num_input_tokens_seen": 24158960,
+      "step": 164
+    },
+    {
+      "epoch": 0.04727587200922456,
+      "loss": 0.3078290522098541,
+      "loss_ce": 0.2633801996707916,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 24158960,
+      "step": 164
+    },
+    {
+      "epoch": 0.04756413952147593,
+      "grad_norm": 35.74031519686993,
+      "learning_rate": 9.247453798654176e-05,
+      "loss": 0.2879,
+      "num_input_tokens_seen": 24331240,
+      "step": 165
+    },
+    {
+      "epoch": 0.04756413952147593,
+      "loss": 0.2579522728919983,
+      "loss_ce": 0.2230096459388733,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 24331240,
+      "step": 165
+    },
+    {
+      "epoch": 0.0478524070337273,
+      "grad_norm": 10.026673974037537,
+      "learning_rate": 9.25839712424472e-05,
+      "loss": 0.3088,
+      "num_input_tokens_seen": 24466016,
+      "step": 166
+    },
+    {
+      "epoch": 0.0478524070337273,
+      "loss": 0.3210294842720032,
+      "loss_ce": 0.2577055096626282,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 24466016,
+      "step": 166
+    },
+    {
+      "epoch": 0.048140674545978666,
+      "grad_norm": 30.320189176787448,
+      "learning_rate": 9.269274723759701e-05,
+      "loss": 0.2739,
+      "num_input_tokens_seen": 24601288,
+      "step": 167
+    },
+    {
+      "epoch": 0.048140674545978666,
+      "loss": 0.29633641242980957,
+      "loss_ce": 0.24610450863838196,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 24601288,
+      "step": 167
+    },
+    {
+      "epoch": 0.048428942058230035,
+      "grad_norm": 11.424628650689558,
+      "learning_rate": 9.280087381995114e-05,
+      "loss": 0.2855,
+      "num_input_tokens_seen": 24773760,
+      "step": 168
+    },
+    {
+      "epoch": 0.048428942058230035,
+      "loss": 0.2584998905658722,
+      "loss_ce": 0.2163856327533722,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 24773760,
+      "step": 168
+    },
+    {
+      "epoch": 0.04871720957048141,
+      "grad_norm": 5.036083595566636,
+      "learning_rate": 9.290835869774165e-05,
+      "loss": 0.2927,
+      "num_input_tokens_seen": 24908480,
+      "step": 169
+    },
+    {
+      "epoch": 0.04871720957048141,
+      "loss": 0.2933216691017151,
+      "loss_ce": 0.2397633194923401,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 24908480,
+      "step": 169
+    },
+    {
+      "epoch": 0.04900547708273278,
+      "grad_norm": 27.447329626580203,
+      "learning_rate": 9.301520944277006e-05,
+      "loss": 0.2688,
+      "num_input_tokens_seen": 25043544,
+      "step": 170
+    },
+    {
+      "epoch": 0.04900547708273278,
+      "loss": 0.2911906838417053,
+      "loss_ce": 0.23934133350849152,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 25043544,
+      "step": 170
+    },
+    {
+      "epoch": 0.049293744594984146,
+      "grad_norm": 12.281459529846494,
+      "learning_rate": 9.31214334936082e-05,
+      "loss": 0.2634,
+      "num_input_tokens_seen": 25216144,
+      "step": 171
+    },
+    {
+      "epoch": 0.049293744594984146,
+      "loss": 0.23304098844528198,
+      "loss_ce": 0.20543783903121948,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 25216144,
+      "step": 171
+    },
+    {
+      "epoch": 0.049582012107235514,
+      "grad_norm": 31.969966090566412,
+      "learning_rate": 9.32270381587056e-05,
+      "loss": 0.2875,
+      "num_input_tokens_seen": 25350928,
+      "step": 172
+    },
+    {
+      "epoch": 0.049582012107235514,
+      "loss": 0.3008914291858673,
+      "loss_ce": 0.2368045151233673,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 25350928,
+      "step": 172
+    },
+    {
+      "epoch": 0.04987027961948688,
+      "grad_norm": 12.583456227749283,
+      "learning_rate": 9.333203061940695e-05,
+      "loss": 0.254,
+      "num_input_tokens_seen": 25485920,
+      "step": 173
+    },
+    {
+      "epoch": 0.04987027961948688,
+      "loss": 0.27169662714004517,
+      "loss_ce": 0.22727830708026886,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 25485920,
+      "step": 173
+    },
+    {
+      "epoch": 0.05015854713173825,
+      "grad_norm": 24.09416200574407,
+      "learning_rate": 9.343641793288233e-05,
+      "loss": 0.257,
+      "num_input_tokens_seen": 25658352,
+      "step": 174
+    },
+    {
+      "epoch": 0.05015854713173825,
+      "loss": 0.24052289128303528,
+      "loss_ce": 0.20500043034553528,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 25658352,
+      "step": 174
+    },
+    {
+      "epoch": 0.050446814643989625,
+      "grad_norm": 14.657839942966513,
+      "learning_rate": 9.35402070349739e-05,
+      "loss": 0.2797,
+      "num_input_tokens_seen": 25793096,
+      "step": 175
+    },
+    {
+      "epoch": 0.050446814643989625,
+      "loss": 0.27785447239875793,
+      "loss_ce": 0.22736310958862305,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 25793096,
+      "step": 175
+    },
+    {
+      "epoch": 0.05073508215624099,
+      "grad_norm": 19.861551728685,
+      "learning_rate": 9.364340474296099e-05,
+      "loss": 0.2496,
+      "num_input_tokens_seen": 25928248,
+      "step": 176
+    },
+    {
+      "epoch": 0.05073508215624099,
+      "loss": 0.26949241757392883,
+      "loss_ce": 0.22600486874580383,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 25928248,
+      "step": 176
+    },
+    {
+      "epoch": 0.05102334966849236,
+      "grad_norm": 16.02692098193626,
+      "learning_rate": 9.374601775824737e-05,
+      "loss": 0.2526,
+      "num_input_tokens_seen": 26100808,
+      "step": 177
+    },
+    {
+      "epoch": 0.05102334966849236,
+      "loss": 0.22894251346588135,
+      "loss_ce": 0.19715844094753265,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 26100808,
+      "step": 177
+    },
+    {
+      "epoch": 0.05131161718074373,
+      "grad_norm": 12.014856064552857,
+      "learning_rate": 9.384805266897235e-05,
+      "loss": 0.2682,
+      "num_input_tokens_seen": 26235648,
+      "step": 178
+    },
+    {
+      "epoch": 0.05131161718074373,
+      "loss": 0.27641862630844116,
+      "loss_ce": 0.22081559896469116,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 26235648,
+      "step": 178
+    },
+    {
+      "epoch": 0.0515998846929951,
+      "grad_norm": 17.513701257593887,
+      "learning_rate": 9.394951595254911e-05,
+      "loss": 0.2362,
+      "num_input_tokens_seen": 26370816,
+      "step": 179
+    },
+    {
+      "epoch": 0.0515998846929951,
+      "loss": 0.25344526767730713,
+      "loss_ce": 0.21006453037261963,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 26370816,
+      "step": 179
+    },
+    {
+      "epoch": 0.051888152205246466,
+      "grad_norm": 6.160409492259628,
+      "learning_rate": 9.405041397813203e-05,
+      "loss": 0.2412,
+      "num_input_tokens_seen": 26543240,
+      "step": 180
+    },
+    {
+      "epoch": 0.051888152205246466,
+      "loss": 0.21339178085327148,
+      "loss_ce": 0.18534612655639648,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 26543240,
+      "step": 180
+    },
+    {
+      "epoch": 0.05217641971749784,
+      "grad_norm": 20.522481305314027,
+      "learning_rate": 9.415075300901591e-05,
+      "loss": 0.2485,
+      "num_input_tokens_seen": 26678072,
+      "step": 181
+    },
+    {
+      "epoch": 0.05217641971749784,
+      "loss": 0.25265252590179443,
+      "loss_ce": 0.21207940578460693,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 26678072,
+      "step": 181
+    },
+    {
+      "epoch": 0.05246468722974921,
+      "grad_norm": 6.0065164673896705,
+      "learning_rate": 9.425053920496895e-05,
+      "loss": 0.2401,
+      "num_input_tokens_seen": 26813064,
+      "step": 182
+    },
+    {
+      "epoch": 0.05246468722974921,
+      "loss": 0.26624372601509094,
+      "loss_ce": 0.21688155829906464,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 26813064,
+      "step": 182
+    },
+    {
+      "epoch": 0.05275295474200058,
+      "grad_norm": 31.84749340439083,
+      "learning_rate": 9.434977862450192e-05,
+      "loss": 0.2362,
+      "num_input_tokens_seen": 26985560,
+      "step": 183
+    },
+    {
+      "epoch": 0.05275295474200058,
+      "loss": 0.22047150135040283,
+      "loss_ce": 0.18293488025665283,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 26985560,
+      "step": 183
+    },
+    {
+      "epoch": 0.053041222254251945,
+      "grad_norm": 5.518518017772564,
+      "learning_rate": 9.44484772270753e-05,
+      "loss": 0.2407,
+      "num_input_tokens_seen": 27120344,
+      "step": 184
+    },
+    {
+      "epoch": 0.053041222254251945,
+      "loss": 0.24068568646907806,
+      "loss_ce": 0.19535183906555176,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 27120344,
+      "step": 184
+    },
+    {
+      "epoch": 0.05332948976650331,
+      "grad_norm": 42.021018230561694,
+      "learning_rate": 9.454664087524682e-05,
+      "loss": 0.2365,
+      "num_input_tokens_seen": 27255400,
+      "step": 185
+    },
+    {
+      "epoch": 0.05332948976650331,
+      "loss": 0.257222980260849,
+      "loss_ce": 0.207692950963974,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 27255400,
+      "step": 185
+    },
+    {
+      "epoch": 0.05361775727875468,
+      "grad_norm": 6.447048391106758,
+      "learning_rate": 9.464427533676112e-05,
+      "loss": 0.2219,
+      "num_input_tokens_seen": 27428048,
+      "step": 186
+    },
+    {
+      "epoch": 0.05361775727875468,
+      "loss": 0.20040246844291687,
+      "loss_ce": 0.17156335711479187,
+      "loss_xval": 0.02880859375,
+      "num_input_tokens_seen": 27428048,
+      "step": 186
+    },
+    {
+      "epoch": 0.053906024791006056,
+      "grad_norm": 27.203776139916368,
+      "learning_rate": 9.474138628658309e-05,
+      "loss": 0.2491,
+      "num_input_tokens_seen": 27562848,
+      "step": 187
+    },
+    {
+      "epoch": 0.053906024791006056,
+      "loss": 0.2541777789592743,
+      "loss_ce": 0.2019316852092743,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 27562848,
+      "step": 187
+    },
+    {
+      "epoch": 0.054194292303257424,
+      "grad_norm": 15.749244789077474,
+      "learning_rate": 9.48379793088774e-05,
+      "loss": 0.2185,
+      "num_input_tokens_seen": 27697960,
+      "step": 188
+    },
+    {
+      "epoch": 0.054194292303257424,
+      "loss": 0.23451972007751465,
+      "loss_ce": 0.19426703453063965,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 27697960,
+      "step": 188
+    },
+    {
+      "epoch": 0.05448255981550879,
+      "grad_norm": 35.73821338153578,
+      "learning_rate": 9.493405989893521e-05,
+      "loss": 0.2304,
+      "num_input_tokens_seen": 27870424,
+      "step": 189
+    },
+    {
+      "epoch": 0.05448255981550879,
+      "loss": 0.19829484820365906,
+      "loss_ce": 0.16846391558647156,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 27870424,
+      "step": 189
+    },
+    {
+      "epoch": 0.05477082732776016,
+      "grad_norm": 9.214313193062441,
+      "learning_rate": 9.502963346505015e-05,
+      "loss": 0.2347,
+      "num_input_tokens_seen": 28005296,
+      "step": 190
+    },
+    {
+      "epoch": 0.05477082732776016,
+      "loss": 0.241441547870636,
+      "loss_ce": 0.1921709179878235,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 28005296,
+      "step": 190
+    },
+    {
+      "epoch": 0.05505909484001153,
+      "grad_norm": 31.476319932486625,
+      "learning_rate": 9.512470533034511e-05,
+      "loss": 0.2196,
+      "num_input_tokens_seen": 28140448,
+      "step": 191
+    },
+    {
+      "epoch": 0.05505909484001153,
+      "loss": 0.23726461827754974,
+      "loss_ce": 0.19161030650138855,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 28140448,
+      "step": 191
+    },
+    {
+      "epoch": 0.0553473623522629,
+      "grad_norm": 5.069492722780138,
+      "learning_rate": 9.521928073455124e-05,
+      "loss": 0.2243,
+      "num_input_tokens_seen": 28312832,
+      "step": 192
+    },
+    {
+      "epoch": 0.0553473623522629,
+      "loss": 0.1921926885843277,
+      "loss_ce": 0.16705383360385895,
+      "loss_xval": 0.025146484375,
+      "num_input_tokens_seen": 28312832,
+      "step": 192
+    },
+    {
+      "epoch": 0.05563562986451427,
+      "grad_norm": 14.335192287118785,
+      "learning_rate": 9.531336483574082e-05,
+      "loss": 0.23,
+      "num_input_tokens_seen": 28447600,
+      "step": 193
+    },
+    {
+      "epoch": 0.05563562986451427,
+      "loss": 0.2333214282989502,
+      "loss_ce": 0.1864311695098877,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 28447600,
+      "step": 193
+    },
+    {
+      "epoch": 0.05592389737676564,
+      "grad_norm": 20.486809993451025,
+      "learning_rate": 9.540696271201525e-05,
+      "loss": 0.2041,
+      "num_input_tokens_seen": 28582696,
+      "step": 194
+    },
+    {
+      "epoch": 0.05592389737676564,
+      "loss": 0.22749033570289612,
+      "loss_ce": 0.18444529175758362,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 28582696,
+      "step": 194
+    },
+    {
+      "epoch": 0.05621216488901701,
+      "grad_norm": 34.51668792130806,
+      "learning_rate": 9.550007936314986e-05,
+      "loss": 0.212,
+      "num_input_tokens_seen": 28755208,
+      "step": 195
+    },
+    {
+      "epoch": 0.05621216488901701,
+      "loss": 0.18501949310302734,
+      "loss_ce": 0.15514278411865234,
+      "loss_xval": 0.0299072265625,
+      "num_input_tokens_seen": 28755208,
+      "step": 195
+    },
+    {
+      "epoch": 0.056500432401268376,
+      "grad_norm": 17.745854925138033,
+      "learning_rate": 9.559271971219628e-05,
+      "loss": 0.2279,
+      "num_input_tokens_seen": 28890120,
+      "step": 196
+    },
+    {
+      "epoch": 0.056500432401268376,
+      "loss": 0.241400346159935,
+      "loss_ce": 0.184027299284935,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 28890120,
+      "step": 196
+    },
+    {
+      "epoch": 0.056788699913519744,
+      "grad_norm": 54.68747170055578,
+      "learning_rate": 9.568488860704453e-05,
+      "loss": 0.2097,
+      "num_input_tokens_seen": 29025136,
+      "step": 197
+    },
+    {
+      "epoch": 0.056788699913519744,
+      "loss": 0.21792283654212952,
+      "loss_ce": 0.17619004845619202,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 29025136,
+      "step": 197
+    },
+    {
+      "epoch": 0.05707696742577112,
+      "grad_norm": 15.10090377209552,
+      "learning_rate": 9.577659082194505e-05,
+      "loss": 0.1959,
+      "num_input_tokens_seen": 29197896,
+      "step": 198
+    },
+    {
+      "epoch": 0.05707696742577112,
+      "loss": 0.17415854334831238,
+      "loss_ce": 0.15132376551628113,
+      "loss_xval": 0.0228271484375,
+      "num_input_tokens_seen": 29197896,
+      "step": 198
+    },
+    {
+      "epoch": 0.05736523493802249,
+      "grad_norm": 37.689905888594836,
+      "learning_rate": 9.586783105899282e-05,
+      "loss": 0.2207,
+      "num_input_tokens_seen": 29332752,
+      "step": 199
+    },
+    {
+      "epoch": 0.05736523493802249,
+      "loss": 0.22800716757774353,
+      "loss_ce": 0.17243465781211853,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 29332752,
+      "step": 199
+    },
+    {
+      "epoch": 0.057653502450273855,
+      "grad_norm": 13.807671874727676,
+      "learning_rate": 9.595861394957398e-05,
+      "loss": 0.1868,
+      "num_input_tokens_seen": 29467792,
+      "step": 200
+    },
+    {
+      "epoch": 0.057653502450273855,
+      "loss": 0.20190562307834625,
+      "loss_ce": 0.16923655569553375,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 29467792,
+      "step": 200
+    },
+    {
+      "epoch": 0.05794176996252522,
+      "grad_norm": 44.84280562592291,
+      "learning_rate": 9.604894405577657e-05,
+      "loss": 0.2107,
+      "num_input_tokens_seen": 29640320,
+      "step": 201
+    },
+    {
+      "epoch": 0.05794176996252522,
+      "loss": 0.1847914755344391,
+      "loss_ce": 0.1469191610813141,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 29640320,
+      "step": 201
+    },
+    {
+      "epoch": 0.05823003747477659,
+      "grad_norm": 28.05197649963492,
+      "learning_rate": 9.613882587176614e-05,
+      "loss": 0.2106,
+      "num_input_tokens_seen": 29775112,
+      "step": 202
+    },
+    {
+      "epoch": 0.05823003747477659,
+      "loss": 0.21230345964431763,
+      "loss_ce": 0.16901427507400513,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 29775112,
+      "step": 202
+    },
+    {
+      "epoch": 0.05851830498702796,
+      "grad_norm": 44.536626160823914,
+      "learning_rate": 9.622826382512748e-05,
+      "loss": 0.1915,
+      "num_input_tokens_seen": 29910208,
+      "step": 203
+    },
+    {
+      "epoch": 0.05851830498702796,
+      "loss": 0.20585274696350098,
+      "loss_ce": 0.16460824012756348,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 29910208,
+      "step": 203
+    },
+    {
+      "epoch": 0.058806572499279335,
+      "grad_norm": 61.770898696786716,
+      "learning_rate": 9.631726227817333e-05,
+      "loss": 0.204,
+      "num_input_tokens_seen": 30082816,
+      "step": 204
+    },
+    {
+      "epoch": 0.058806572499279335,
+      "loss": 0.18048658967018127,
+      "loss_ce": 0.14203444123268127,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 30082816,
+      "step": 204
+    },
+    {
+      "epoch": 0.0590948400115307,
+      "grad_norm": 5.133503721665133,
+      "learning_rate": 9.640582552922112e-05,
+      "loss": 0.2018,
+      "num_input_tokens_seen": 30217600,
+      "step": 205
+    },
+    {
+      "epoch": 0.0590948400115307,
+      "loss": 0.21230003237724304,
+      "loss_ce": 0.16855308413505554,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 30217600,
+      "step": 205
+    },
+    {
+      "epoch": 0.05938310752378207,
+      "grad_norm": 41.13106645832814,
+      "learning_rate": 9.64939578138386e-05,
+      "loss": 0.1953,
+      "num_input_tokens_seen": 30352520,
+      "step": 206
+    },
+    {
+      "epoch": 0.05938310752378207,
+      "loss": 0.20415394008159637,
+      "loss_ce": 0.16251270473003387,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 30352520,
+      "step": 206
+    },
+    {
+      "epoch": 0.05967137503603344,
+      "grad_norm": 7.532245333884863,
+      "learning_rate": 9.658166330605936e-05,
+      "loss": 0.1872,
+      "num_input_tokens_seen": 30525000,
+      "step": 207
+    },
+    {
+      "epoch": 0.05967137503603344,
+      "loss": 0.16164681315422058,
+      "loss_ce": 0.13517281413078308,
+      "loss_xval": 0.0264892578125,
+      "num_input_tokens_seen": 30525000,
+      "step": 207
+    },
+    {
+      "epoch": 0.05995964254828481,
+      "grad_norm": 68.5759865275484,
+      "learning_rate": 9.666894611956906e-05,
+      "loss": 0.2288,
+      "num_input_tokens_seen": 30659808,
+      "step": 208
+    },
+    {
+      "epoch": 0.05995964254828481,
+      "loss": 0.23044301569461823,
+      "loss_ce": 0.16342641413211823,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 30659808,
+      "step": 208
+    },
+    {
+      "epoch": 0.060247910060536175,
+      "grad_norm": 41.614219942429486,
+      "learning_rate": 9.67558103088632e-05,
+      "loss": 0.1869,
+      "num_input_tokens_seen": 30794960,
+      "step": 209
+    },
+    {
+      "epoch": 0.060247910060536175,
+      "loss": 0.1920596957206726,
+      "loss_ce": 0.1534244418144226,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 30794960,
+      "step": 209
+    },
+    {
+      "epoch": 0.06053617757278755,
+      "grad_norm": 54.20840230591495,
+      "learning_rate": 9.684225987037716e-05,
+      "loss": 0.1939,
+      "num_input_tokens_seen": 30967408,
+      "step": 210
+    },
+    {
+      "epoch": 0.06053617757278755,
+      "loss": 0.1660287082195282,
+      "loss_ce": 0.1323220431804657,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 30967408,
+      "step": 210
+    },
+    {
+      "epoch": 0.06082444508503892,
+      "grad_norm": 78.17547740458902,
+      "learning_rate": 9.692829874358969e-05,
+      "loss": 0.2254,
+      "num_input_tokens_seen": 31102128,
+      "step": 211
+    },
+    {
+      "epoch": 0.06082444508503892,
+      "loss": 0.21866479516029358,
+      "loss_ce": 0.15173974633216858,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 31102128,
+      "step": 211
+    },
+    {
+      "epoch": 0.061112712597290286,
+      "grad_norm": 31.53452634225132,
+      "learning_rate": 9.701393081209986e-05,
+      "loss": 0.1722,
+      "num_input_tokens_seen": 31237240,
+      "step": 212
+    },
+    {
+      "epoch": 0.061112712597290286,
+      "loss": 0.1789591759443283,
+      "loss_ce": 0.1464884728193283,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 31237240,
+      "step": 212
+    },
+    {
+      "epoch": 0.061400980109541654,
+      "grad_norm": 113.61471944977563,
+      "learning_rate": 9.709915990467911e-05,
+      "loss": 0.2605,
+      "num_input_tokens_seen": 31409720,
+      "step": 213
+    },
+    {
+      "epoch": 0.061400980109541654,
+      "loss": 0.20028024911880493,
+      "loss_ce": 0.13012033700942993,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 31409720,
+      "step": 213
+    },
+    {
+      "epoch": 0.06168924762179302,
+      "grad_norm": 13.533712490533789,
+      "learning_rate": 9.718398979629844e-05,
+      "loss": 0.1859,
+      "num_input_tokens_seen": 31544496,
+      "step": 214
+    },
+    {
+      "epoch": 0.06168924762179302,
+      "loss": 0.18483048677444458,
+      "loss_ce": 0.14607316255569458,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 31544496,
+      "step": 214
+    },
+    {
+      "epoch": 0.06197751513404439,
+      "grad_norm": 153.1001234370429,
+      "learning_rate": 9.726842420913164e-05,
+      "loss": 0.3144,
+      "num_input_tokens_seen": 31679576,
+      "step": 215
+    },
+    {
+      "epoch": 0.06197751513404439,
+      "loss": 0.24575066566467285,
+      "loss_ce": 0.13930535316467285,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 31679576,
+      "step": 215
+    },
+    {
+      "epoch": 0.062265782646295766,
+      "grad_norm": 114.7913479047939,
+      "learning_rate": 9.735246681353531e-05,
+      "loss": 0.2533,
+      "num_input_tokens_seen": 31852216,
+      "step": 216
+    },
+    {
+      "epoch": 0.062265782646295766,
+      "loss": 0.1870504915714264,
+      "loss_ce": 0.12320771813392639,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 31852216,
+      "step": 216
+    },
+    {
+      "epoch": 0.06255405015854713,
+      "grad_norm": 127.62253929979651,
+      "learning_rate": 9.743612122900626e-05,
+      "loss": 0.2881,
+      "num_input_tokens_seen": 31987000,
+      "step": 217
+    },
+    {
+      "epoch": 0.06255405015854713,
+      "loss": 0.2577114701271057,
+      "loss_ce": 0.14381985366344452,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 31987000,
+      "step": 217
+    },
+    {
+      "epoch": 0.0628423176707985,
+      "grad_norm": 235.00377257321716,
+      "learning_rate": 9.751939102511684e-05,
+      "loss": 0.4947,
+      "num_input_tokens_seen": 32122032,
+      "step": 218
+    },
+    {
+      "epoch": 0.0628423176707985,
+      "loss": 0.3868434727191925,
+      "loss_ce": 0.14050555229187012,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 32122032,
+      "step": 218
+    },
+    {
+      "epoch": 0.06313058518304987,
+      "grad_norm": 12.837730790336758,
+      "learning_rate": 9.760227972242893e-05,
+      "loss": 0.1585,
+      "num_input_tokens_seen": 32294488,
+      "step": 219
+    },
+    {
+      "epoch": 0.06313058518304987,
+      "loss": 0.1404755860567093,
+      "loss_ce": 0.11587841808795929,
+      "loss_xval": 0.024658203125,
+      "num_input_tokens_seen": 32294488,
+      "step": 219
+    },
+    {
+      "epoch": 0.06341885269530124,
+      "grad_norm": 265.78254131812344,
+      "learning_rate": 9.768479079338703e-05,
+      "loss": 0.6116,
+      "num_input_tokens_seen": 32429264,
+      "step": 220
+    },
+    {
+      "epoch": 0.06341885269530124,
+      "loss": 0.46776074171066284,
+      "loss_ce": 0.14061228930950165,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 32429264,
+      "step": 220
+    },
+    {
+      "epoch": 0.0637071202075526,
+      "grad_norm": 161.92075515006673,
+      "learning_rate": 9.776692766319115e-05,
+      "loss": 0.3099,
+      "num_input_tokens_seen": 32564304,
+      "step": 221
+    },
+    {
+      "epoch": 0.0637071202075526,
+      "loss": 0.30831146240234375,
+      "loss_ce": 0.12667082250118256,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 32564304,
+      "step": 221
+    },
+    {
+      "epoch": 0.06399538771980398,
+      "grad_norm": 213.97148511685998,
+      "learning_rate": 9.784869371065011e-05,
+      "loss": 0.4637,
+      "num_input_tokens_seen": 32736864,
+      "step": 222
+    },
+    {
+      "epoch": 0.06399538771980398,
+      "loss": 0.28785890340805054,
+      "loss_ce": 0.11268798261880875,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 32736864,
+      "step": 222
+    },
+    {
+      "epoch": 0.06428365523205534,
+      "grad_norm": 279.0437182260339,
+      "learning_rate": 9.793009226901534e-05,
+      "loss": 0.6737,
+      "num_input_tokens_seen": 32871640,
+      "step": 223
+    },
+    {
+      "epoch": 0.06428365523205534,
+      "loss": 0.5465670824050903,
+      "loss_ce": 0.14129364490509033,
+      "loss_xval": 0.40625,
+      "num_input_tokens_seen": 32871640,
+      "step": 223
+    },
+    {
+      "epoch": 0.06457192274430672,
+      "grad_norm": 148.45369429539775,
+      "learning_rate": 9.801112662679638e-05,
+      "loss": 0.3168,
+      "num_input_tokens_seen": 33006704,
+      "step": 224
+    },
+    {
+      "epoch": 0.06457192274430672,
+      "loss": 0.21143169701099396,
+      "loss_ce": 0.12302227318286896,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 33006704,
+      "step": 224
+    },
+    {
+      "epoch": 0.06486019025655809,
+      "grad_norm": 444.77590098553196,
+      "learning_rate": 9.809180002855806e-05,
+      "loss": 1.4688,
+      "num_input_tokens_seen": 33179216,
+      "step": 225
+    },
+    {
+      "epoch": 0.06486019025655809,
+      "loss": 0.8774189949035645,
+      "loss_ce": 0.10886429250240326,
+      "loss_xval": 0.76953125,
+      "num_input_tokens_seen": 33179216,
+      "step": 225
+    },
+    {
+      "epoch": 0.06514845776880945,
+      "grad_norm": 31.205263410938144,
+      "learning_rate": 9.817211567569992e-05,
+      "loss": 0.1857,
+      "num_input_tokens_seen": 33313952,
+      "step": 226
+    },
+    {
+      "epoch": 0.06514845776880945,
+      "loss": 0.18774542212486267,
+      "loss_ce": 0.13681158423423767,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 33313952,
+      "step": 226
+    },
+    {
+      "epoch": 0.06543672528106083,
+      "grad_norm": 598.3519060203606,
+      "learning_rate": 9.825207672721861e-05,
+      "loss": 2.4385,
+      "num_input_tokens_seen": 33449056,
+      "step": 227
+    },
+    {
+      "epoch": 0.06543672528106083,
+      "loss": 1.5101361274719238,
+      "loss_ce": 0.12537051737308502,
+      "loss_xval": 1.3828125,
+      "num_input_tokens_seen": 33449056,
+      "step": 227
+    },
+    {
+      "epoch": 0.06572499279331219,
+      "grad_norm": 384.2944927117941,
+      "learning_rate": 9.833168630045344e-05,
+      "loss": 1.1605,
+      "num_input_tokens_seen": 33621544,
+      "step": 228
+    },
+    {
+      "epoch": 0.06572499279331219,
+      "loss": 0.7385019063949585,
+      "loss_ce": 0.11203701794147491,
+      "loss_xval": 0.625,
+      "num_input_tokens_seen": 33621544,
+      "step": 228
+    },
+    {
+      "epoch": 0.06601326030556356,
+      "grad_norm": 505.26874849553155,
+      "learning_rate": 9.841094747181556e-05,
+      "loss": 1.8865,
+      "num_input_tokens_seen": 33756344,
+      "step": 229
+    },
+    {
+      "epoch": 0.06601326030556356,
+      "loss": 1.398691177368164,
+      "loss_ce": 0.13599595427513123,
+      "loss_xval": 1.265625,
+      "num_input_tokens_seen": 33756344,
+      "step": 229
+    },
+    {
+      "epoch": 0.06630152781781493,
+      "grad_norm": 634.3239419924421,
+      "learning_rate": 9.848986327750132e-05,
+      "loss": 2.831,
+      "num_input_tokens_seen": 33891360,
+      "step": 230
+    },
+    {
+      "epoch": 0.06630152781781493,
+      "loss": 1.8530738353729248,
+      "loss_ce": 0.1235816702246666,
+      "loss_xval": 1.7265625,
+      "num_input_tokens_seen": 33891360,
+      "step": 230
+    },
+    {
+      "epoch": 0.0665897953300663,
+      "grad_norm": 320.3658674116094,
+      "learning_rate": 9.856843671419021e-05,
+      "loss": 0.8727,
+      "num_input_tokens_seen": 34063912,
+      "step": 231
+    },
+    {
+      "epoch": 0.0665897953300663,
+      "loss": 0.7934414744377136,
+      "loss_ce": 0.11863675713539124,
+      "loss_xval": 0.67578125,
+      "num_input_tokens_seen": 34063912,
+      "step": 231
+    },
+    {
+      "epoch": 0.06687806284231768,
+      "grad_norm": 714.0511131829795,
+      "learning_rate": 9.864667073972757e-05,
+      "loss": 3.6725,
+      "num_input_tokens_seen": 34198672,
+      "step": 232
+    },
+    {
+      "epoch": 0.06687806284231768,
+      "loss": 2.818943500518799,
+      "loss_ce": 0.13925601541996002,
+      "loss_xval": 2.6875,
+      "num_input_tokens_seen": 34198672,
+      "step": 232
+    },
+    {
+      "epoch": 0.06716633035456904,
+      "grad_norm": 193.6951410568273,
+      "learning_rate": 9.872456827379282e-05,
+      "loss": 0.4295,
+      "num_input_tokens_seen": 34333616,
+      "step": 233
+    },
+    {
+      "epoch": 0.06716633035456904,
+      "loss": 0.4650017321109772,
+      "loss_ce": 0.13553395867347717,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 34333616,
+      "step": 233
+    },
+    {
+      "epoch": 0.06745459786682041,
+      "grad_norm": 719.8870289760067,
+      "learning_rate": 9.880213219855314e-05,
+      "loss": 3.872,
+      "num_input_tokens_seen": 34506024,
+      "step": 234
+    },
+    {
+      "epoch": 0.06745459786682041,
+      "loss": 2.788909435272217,
+      "loss_ce": 0.12094061076641083,
+      "loss_xval": 2.671875,
+      "num_input_tokens_seen": 34506024,
+      "step": 234
+    },
+    {
+      "epoch": 0.06774286537907177,
+      "grad_norm": 181.98869361563385,
+      "learning_rate": 9.887936535930344e-05,
+      "loss": 0.4589,
+      "num_input_tokens_seen": 34640728,
+      "step": 235
+    },
+    {
+      "epoch": 0.06774286537907177,
+      "loss": 0.5512518882751465,
+      "loss_ce": 0.14207220077514648,
+      "loss_xval": 0.41015625,
+      "num_input_tokens_seen": 34640728,
+      "step": 235
+    },
+    {
+      "epoch": 0.06803113289132315,
+      "grad_norm": 703.1804675651726,
+      "learning_rate": 9.895627056509262e-05,
+      "loss": 4.0544,
+      "num_input_tokens_seen": 34775784,
+      "step": 236
+    },
+    {
+      "epoch": 0.06803113289132315,
+      "loss": 3.0673837661743164,
+      "loss_ce": 0.14746198058128357,
+      "loss_xval": 2.921875,
+      "num_input_tokens_seen": 34775784,
+      "step": 236
+    },
+    {
+      "epoch": 0.06831940040357452,
+      "grad_norm": 192.37560557603808,
+      "learning_rate": 9.90328505893366e-05,
+      "loss": 0.5742,
+      "num_input_tokens_seen": 34948304,
+      "step": 237
+    },
+    {
+      "epoch": 0.06831940040357452,
+      "loss": 0.7511662244796753,
+      "loss_ce": 0.14130297303199768,
+      "loss_xval": 0.609375,
+      "num_input_tokens_seen": 34948304,
+      "step": 237
+    },
+    {
+      "epoch": 0.06860766791582588,
+      "grad_norm": 670.4064003150474,
+      "learning_rate": 9.91091081704185e-05,
+      "loss": 3.8927,
+      "num_input_tokens_seen": 35083072,
+      "step": 238
+    },
+    {
+      "epoch": 0.06860766791582588,
+      "loss": 3.181490898132324,
+      "loss_ce": 0.16391250491142273,
+      "loss_xval": 3.015625,
+      "num_input_tokens_seen": 35083072,
+      "step": 238
+    },
+    {
+      "epoch": 0.06889593542807726,
+      "grad_norm": 184.37943098250278,
+      "learning_rate": 9.918504601227611e-05,
+      "loss": 0.5801,
+      "num_input_tokens_seen": 35218160,
+      "step": 239
+    },
+    {
+      "epoch": 0.06889593542807726,
+      "loss": 0.808303713798523,
+      "loss_ce": 0.15058887004852295,
+      "loss_xval": 0.65625,
+      "num_input_tokens_seen": 35218160,
+      "step": 239
+    },
+    {
+      "epoch": 0.06918420294032862,
+      "grad_norm": 486.97199962741814,
+      "learning_rate": 9.926066678497726e-05,
+      "loss": 2.5227,
+      "num_input_tokens_seen": 35390600,
+      "step": 240
+    },
+    {
+      "epoch": 0.06918420294032862,
+      "loss": 2.4577555656433105,
+      "loss_ce": 0.15697449445724487,
+      "loss_xval": 2.296875,
+      "num_input_tokens_seen": 35390600,
+      "step": 240
+    },
+    {
+      "epoch": 0.06947247045258,
+      "grad_norm": 291.8847677589011,
+      "learning_rate": 9.933597312528319e-05,
+      "loss": 1.0913,
+      "num_input_tokens_seen": 35525384,
+      "step": 241
+    },
+    {
+      "epoch": 0.06947247045258,
+      "loss": 1.3332394361495972,
+      "loss_ce": 0.2033565640449524,
+      "loss_xval": 1.1328125,
+      "num_input_tokens_seen": 35525384,
+      "step": 241
+    },
+    {
+      "epoch": 0.06976073796483136,
+      "grad_norm": 248.55809130995584,
+      "learning_rate": 9.941096763720006e-05,
+      "loss": 0.8804,
+      "num_input_tokens_seen": 35660456,
+      "step": 242
+    },
+    {
+      "epoch": 0.06976073796483136,
+      "loss": 0.9395829439163208,
+      "loss_ce": 0.1744462549686432,
+      "loss_xval": 0.765625,
+      "num_input_tokens_seen": 35660456,
+      "step": 242
+    },
+    {
+      "epoch": 0.07004900547708273,
+      "grad_norm": 344.70726428259974,
+      "learning_rate": 9.948565289251937e-05,
+      "loss": 1.5722,
+      "num_input_tokens_seen": 35832752,
+      "step": 243
+    },
+    {
+      "epoch": 0.07004900547708273,
+      "loss": 1.5424445867538452,
+      "loss_ce": 0.19088204205036163,
+      "loss_xval": 1.3515625,
+      "num_input_tokens_seen": 35832752,
+      "step": 243
+    },
+    {
+      "epoch": 0.07033727298933411,
+      "grad_norm": 35.03505347015722,
+      "learning_rate": 9.956003143134718e-05,
+      "loss": 0.2743,
+      "num_input_tokens_seen": 35967456,
+      "step": 244
+    },
+    {
+      "epoch": 0.07033727298933411,
+      "loss": 0.3062777519226074,
+      "loss_ce": 0.2516207695007324,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 35967456,
+      "step": 244
+    },
+    {
+      "epoch": 0.07062554050158547,
+      "grad_norm": 324.27535084781664,
+      "learning_rate": 9.963410576262232e-05,
+      "loss": 1.4836,
+      "num_input_tokens_seen": 36102544,
+      "step": 245
+    },
+    {
+      "epoch": 0.07062554050158547,
+      "loss": 1.3415637016296387,
+      "loss_ce": 0.18629023432731628,
+      "loss_xval": 1.15625,
+      "num_input_tokens_seen": 36102544,
+      "step": 245
+    },
+    {
+      "epoch": 0.07091380801383684,
+      "grad_norm": 192.40093602923258,
+      "learning_rate": 9.97078783646244e-05,
+      "loss": 0.7443,
+      "num_input_tokens_seen": 36275024,
+      "step": 246
+    },
+    {
+      "epoch": 0.07091380801383684,
+      "loss": 0.865687370300293,
+      "loss_ce": 0.19967174530029297,
+      "loss_xval": 0.6640625,
+      "num_input_tokens_seen": 36275024,
+      "step": 246
+    },
+    {
+      "epoch": 0.0712020755260882,
+      "grad_norm": 169.35354206434812,
+      "learning_rate": 9.978135168547127e-05,
+      "loss": 0.6059,
+      "num_input_tokens_seen": 36409848,
+      "step": 247
+    },
+    {
+      "epoch": 0.0712020755260882,
+      "loss": 0.6389449834823608,
+      "loss_ce": 0.2558884024620056,
+      "loss_xval": 0.3828125,
+      "num_input_tokens_seen": 36409848,
+      "step": 247
+    },
+    {
+      "epoch": 0.07149034303833958,
+      "grad_norm": 236.26593297962225,
+      "learning_rate": 9.985452814360636e-05,
+      "loss": 0.9012,
+      "num_input_tokens_seen": 36544976,
+      "step": 248
+    },
+    {
+      "epoch": 0.07149034303833958,
+      "loss": 1.0095927715301514,
+      "loss_ce": 0.1878153681755066,
+      "loss_xval": 0.8203125,
+      "num_input_tokens_seen": 36544976,
+      "step": 248
+    },
+    {
+      "epoch": 0.07177861055059095,
+      "grad_norm": 48.60897485178449,
+      "learning_rate": 9.992741012827652e-05,
+      "loss": 0.2498,
+      "num_input_tokens_seen": 36717640,
+      "step": 249
+    },
+    {
+      "epoch": 0.07177861055059095,
+      "loss": 0.20433716475963593,
+      "loss_ce": 0.17581848800182343,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 36717640,
+      "step": 249
+    },
+    {
+      "epoch": 0.07206687806284232,
+      "grad_norm": 179.656878403396,
+      "learning_rate": 0.0001,
+      "loss": 0.6386,
+      "num_input_tokens_seen": 36852424,
+      "step": 250
+    },
+    {
+      "epoch": 0.07206687806284232,
+      "eval_websight_new_IoU": 0.0,
+      "eval_websight_new_MAE_x": 0.6111758947372437,
+      "eval_websight_new_MAE_y": 0.5629599392414093,
+      "eval_websight_new_NUM_probability": 0.6707842648029327,
+      "eval_websight_new_inside_bbox": 0.0,
+      "eval_websight_new_loss": 0.5729646682739258,
+      "eval_websight_new_loss_ce": 0.1514492705464363,
+      "eval_websight_new_loss_xval": 0.4156494140625,
+      "eval_websight_new_runtime": 35.618,
+      "eval_websight_new_samples_per_second": 1.404,
+      "eval_websight_new_steps_per_second": 0.056,
+      "num_input_tokens_seen": 36852424,
+      "step": 250
+    },
+    {
+      "epoch": 0.07206687806284232,
+      "eval_seeclick_IoU": 0.0,
+      "eval_seeclick_MAE_x": 0.5661886930465698,
+      "eval_seeclick_MAE_y": 0.5124521255493164,
+      "eval_seeclick_NUM_probability": 0.6034610271453857,
+      "eval_seeclick_inside_bbox": 0.0,
+      "eval_seeclick_loss": 0.5685375332832336,
+      "eval_seeclick_loss_ce": 0.22021447867155075,
+      "eval_seeclick_loss_xval": 0.3525390625,
+      "eval_seeclick_runtime": 64.6475,
+      "eval_seeclick_samples_per_second": 0.773,
+      "eval_seeclick_steps_per_second": 0.031,
+      "num_input_tokens_seen": 36852424,
+      "step": 250
+    },
+    {
+      "epoch": 0.07206687806284232,
+      "eval_icons_IoU": 0.0,
+      "eval_icons_MAE_x": 0.6235586404800415,
+      "eval_icons_MAE_y": 0.6090098023414612,
+      "eval_icons_NUM_probability": 0.7112331092357635,
+      "eval_icons_inside_bbox": 0.0,
+      "eval_icons_loss": 0.5959538817405701,
+      "eval_icons_loss_ce": 0.18405180424451828,
+      "eval_icons_loss_xval": 0.41455078125,
+      "eval_icons_runtime": 64.7835,
+      "eval_icons_samples_per_second": 0.772,
+      "eval_icons_steps_per_second": 0.031,
+      "num_input_tokens_seen": 36852424,
+      "step": 250
+    },
+    {
+      "epoch": 0.07206687806284232,
+      "loss": 0.5958322882652283,
+      "loss_ce": 0.18811748921871185,
+      "loss_xval": 0.408203125,
+      "num_input_tokens_seen": 36852424,
+      "step": 250
+    },
+    {
+      "epoch": 0.07235514557509369,
+      "grad_norm": 178.65768699117723,
+      "learning_rate": 0.0001,
+      "loss": 0.5964,
+      "num_input_tokens_seen": 36987608,
+      "step": 251
+    },
+    {
+      "epoch": 0.07235514557509369,
+      "loss": 0.589705228805542,
+      "loss_ce": 0.15757633745670319,
+      "loss_xval": 0.431640625,
+      "num_input_tokens_seen": 36987608,
+      "step": 251
+    },
+    {
+      "epoch": 0.07264341308734505,
+      "grad_norm": 38.66699265104924,
+      "learning_rate": 0.0001,
+      "loss": 0.2277,
+      "num_input_tokens_seen": 37160176,
+      "step": 252
+    },
+    {
+      "epoch": 0.07264341308734505,
+      "loss": 0.20238327980041504,
+      "loss_ce": 0.17584824562072754,
+      "loss_xval": 0.0264892578125,
+      "num_input_tokens_seen": 37160176,
+      "step": 252
+    },
+    {
+      "epoch": 0.07293168059959643,
+      "grad_norm": 185.87773020429194,
+      "learning_rate": 0.0001,
+      "loss": 0.636,
+      "num_input_tokens_seen": 37295120,
+      "step": 253
+    },
+    {
+      "epoch": 0.07293168059959643,
+      "loss": 0.6577854752540588,
+      "loss_ce": 0.20099833607673645,
+      "loss_xval": 0.45703125,
+      "num_input_tokens_seen": 37295120,
+      "step": 253
+    },
+    {
+      "epoch": 0.07321994811184779,
+      "grad_norm": 102.75700880750043,
+      "learning_rate": 0.0001,
+      "loss": 0.3128,
+      "num_input_tokens_seen": 37430168,
+      "step": 254
+    },
+    {
+      "epoch": 0.07321994811184779,
+      "loss": 0.2869405746459961,
+      "loss_ce": 0.1482686698436737,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 37430168,
+      "step": 254
+    },
+    {
+      "epoch": 0.07350821562409916,
+      "grad_norm": 87.97296152813514,
+      "learning_rate": 0.0001,
+      "loss": 0.3084,
+      "num_input_tokens_seen": 37602528,
+      "step": 255
+    },
+    {
+      "epoch": 0.07350821562409916,
+      "loss": 0.28620052337646484,
+      "loss_ce": 0.17829035222530365,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 37602528,
+      "step": 255
+    },
+    {
+      "epoch": 0.07379648313635054,
+      "grad_norm": 164.32807477823272,
+      "learning_rate": 0.0001,
+      "loss": 0.539,
+      "num_input_tokens_seen": 37737296,
+      "step": 256
+    },
+    {
+      "epoch": 0.07379648313635054,
+      "loss": 0.5527489185333252,
+      "loss_ce": 0.1921532154083252,
+      "loss_xval": 0.361328125,
+      "num_input_tokens_seen": 37737296,
+      "step": 256
+    },
+    {
+      "epoch": 0.0740847506486019,
+      "grad_norm": 29.183798965456198,
+      "learning_rate": 0.0001,
+      "loss": 0.1747,
+      "num_input_tokens_seen": 37872304,
+      "step": 257
+    },
+    {
+      "epoch": 0.0740847506486019,
+      "loss": 0.1750638484954834,
+      "loss_ce": 0.1369626522064209,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 37872304,
+      "step": 257
+    },
+    {
+      "epoch": 0.07437301816085327,
+      "grad_norm": 113.89670618085759,
+      "learning_rate": 0.0001,
+      "loss": 0.3648,
+      "num_input_tokens_seen": 38044744,
+      "step": 258
+    },
+    {
+      "epoch": 0.07437301816085327,
+      "loss": 0.37308743596076965,
+      "loss_ce": 0.15116359293460846,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 38044744,
+      "step": 258
+    },
+    {
+      "epoch": 0.07466128567310464,
+      "grad_norm": 114.85316241618703,
+      "learning_rate": 0.0001,
+      "loss": 0.3363,
+      "num_input_tokens_seen": 38179512,
+      "step": 259
+    },
+    {
+      "epoch": 0.07466128567310464,
+      "loss": 0.34536540508270264,
+      "loss_ce": 0.16409097611904144,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 38179512,
+      "step": 259
+    },
+    {
+      "epoch": 0.07494955318535601,
+      "grad_norm": 14.752637398271474,
+      "learning_rate": 0.0001,
+      "loss": 0.1474,
+      "num_input_tokens_seen": 38314592,
+      "step": 260
+    },
+    {
+      "epoch": 0.07494955318535601,
+      "loss": 0.13977420330047607,
+      "loss_ce": 0.12003696709871292,
+      "loss_xval": 0.019775390625,
+      "num_input_tokens_seen": 38314592,
+      "step": 260
+    },
+    {
+      "epoch": 0.07523782069760739,
+      "grad_norm": 120.3078403264174,
+      "learning_rate": 0.0001,
+      "loss": 0.3626,
+      "num_input_tokens_seen": 38487080,
+      "step": 261
+    },
+    {
+      "epoch": 0.07523782069760739,
+      "loss": 0.31943660974502563,
+      "loss_ce": 0.12986139953136444,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 38487080,
+      "step": 261
+    },
+    {
+      "epoch": 0.07552608820985875,
+      "grad_norm": 79.9536443457315,
+      "learning_rate": 0.0001,
+      "loss": 0.2404,
+      "num_input_tokens_seen": 38621784,
+      "step": 262
+    },
+    {
+      "epoch": 0.07552608820985875,
+      "loss": 0.2604348063468933,
+      "loss_ce": 0.1555764079093933,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 38621784,
+      "step": 262
+    },
+    {
+      "epoch": 0.07581435572211012,
+      "grad_norm": 67.24270419291493,
+      "learning_rate": 0.0001,
+      "loss": 0.1914,
+      "num_input_tokens_seen": 38756824,
+      "step": 263
+    },
+    {
+      "epoch": 0.07581435572211012,
+      "loss": 0.17697831988334656,
+      "loss_ce": 0.11179277300834656,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 38756824,
+      "step": 263
+    },
+    {
+      "epoch": 0.07610262323436148,
+      "grad_norm": 109.17180729202337,
+      "learning_rate": 0.0001,
+      "loss": 0.3252,
+      "num_input_tokens_seen": 38929360,
+      "step": 264
+    },
+    {
+      "epoch": 0.07610262323436148,
+      "loss": 0.3131757378578186,
+      "loss_ce": 0.120792917907238,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 38929360,
+      "step": 264
+    },
+    {
+      "epoch": 0.07639089074661286,
+      "grad_norm": 17.105499866069646,
+      "learning_rate": 0.0001,
+      "loss": 0.1472,
+      "num_input_tokens_seen": 39064128,
+      "step": 265
+    },
+    {
+      "epoch": 0.07639089074661286,
+      "loss": 0.17089056968688965,
+      "loss_ce": 0.14578986167907715,
+      "loss_xval": 0.025146484375,
+      "num_input_tokens_seen": 39064128,
+      "step": 265
+    },
+    {
+      "epoch": 0.07667915825886423,
+      "grad_norm": 82.84505450129119,
+      "learning_rate": 0.0001,
+      "loss": 0.2117,
+      "num_input_tokens_seen": 39199264,
+      "step": 266
+    },
+    {
+      "epoch": 0.07667915825886423,
+      "loss": 0.20540225505828857,
+      "loss_ce": 0.09560001641511917,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 39199264,
+      "step": 266
+    },
+    {
+      "epoch": 0.0769674257711156,
+      "grad_norm": 70.99727079552451,
+      "learning_rate": 0.0001,
+      "loss": 0.2063,
+      "num_input_tokens_seen": 39371720,
+      "step": 267
+    },
+    {
+      "epoch": 0.0769674257711156,
+      "loss": 0.1930188685655594,
+      "loss_ce": 0.10299200564622879,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 39371720,
+      "step": 267
+    },
+    {
+      "epoch": 0.07725569328336697,
+      "grad_norm": 24.615430995589435,
+      "learning_rate": 0.0001,
+      "loss": 0.1345,
+      "num_input_tokens_seen": 39506472,
+      "step": 268
+    },
+    {
+      "epoch": 0.07725569328336697,
+      "loss": 0.15381276607513428,
+      "loss_ce": 0.13003194332122803,
+      "loss_xval": 0.0238037109375,
+      "num_input_tokens_seen": 39506472,
+      "step": 268
+    },
+    {
+      "epoch": 0.07754396079561833,
+      "grad_norm": 81.15937577572338,
+      "learning_rate": 0.0001,
+      "loss": 0.1952,
+      "num_input_tokens_seen": 39641416,
+      "step": 269
+    },
+    {
+      "epoch": 0.07754396079561833,
+      "loss": 0.19636116921901703,
+      "loss_ce": 0.08729134500026703,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 39641416,
+      "step": 269
+    },
+    {
+      "epoch": 0.0778322283078697,
+      "grad_norm": 33.693167336460625,
+      "learning_rate": 0.0001,
+      "loss": 0.1362,
+      "num_input_tokens_seen": 39814000,
+      "step": 270
+    },
+    {
+      "epoch": 0.0778322283078697,
+      "loss": 0.1107083410024643,
+      "loss_ce": 0.0908261388540268,
+      "loss_xval": 0.0198974609375,
+      "num_input_tokens_seen": 39814000,
+      "step": 270
+    },
+    {
+      "epoch": 0.07812049582012107,
+      "grad_norm": 49.838891424712216,
+      "learning_rate": 0.0001,
+      "loss": 0.1504,
+      "num_input_tokens_seen": 39948960,
+      "step": 271
+    },
+    {
+      "epoch": 0.07812049582012107,
+      "loss": 0.16035011410713196,
+      "loss_ce": 0.11991433054208755,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 39948960,
+      "step": 271
+    },
+    {
+      "epoch": 0.07840876333237244,
+      "grad_norm": 71.84009907388065,
+      "learning_rate": 0.0001,
+      "loss": 0.1673,
+      "num_input_tokens_seen": 40084048,
+      "step": 272
+    },
+    {
+      "epoch": 0.07840876333237244,
+      "loss": 0.158762127161026,
+      "loss_ce": 0.079172283411026,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 40084048,
+      "step": 272
+    },
+    {
+      "epoch": 0.07869703084462382,
+      "grad_norm": 14.134734889641313,
+      "learning_rate": 0.0001,
+      "loss": 0.1095,
+      "num_input_tokens_seen": 40256624,
+      "step": 273
+    },
+    {
+      "epoch": 0.07869703084462382,
+      "loss": 0.09496979415416718,
+      "loss_ce": 0.08466248214244843,
+      "loss_xval": 0.01031494140625,
+      "num_input_tokens_seen": 40256624,
+      "step": 273
+    },
+    {
+      "epoch": 0.07898529835687518,
+      "grad_norm": 65.02772231537142,
+      "learning_rate": 0.0001,
+      "loss": 0.1654,
+      "num_input_tokens_seen": 40391416,
+      "step": 274
+    },
+    {
+      "epoch": 0.07898529835687518,
+      "loss": 0.1933080554008484,
+      "loss_ce": 0.10480707883834839,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 40391416,
+      "step": 274
+    },
+    {
+      "epoch": 0.07927356586912655,
+      "grad_norm": 22.172397413986033,
+      "learning_rate": 0.0001,
+      "loss": 0.0932,
+      "num_input_tokens_seen": 40526544,
+      "step": 275
+    },
+    {
+      "epoch": 0.07927356586912655,
+      "loss": 0.09251715242862701,
+      "loss_ce": 0.07400824129581451,
+      "loss_xval": 0.0185546875,
+      "num_input_tokens_seen": 40526544,
+      "step": 275
+    },
+    {
+      "epoch": 0.07956183338137791,
+      "grad_norm": 30.37442186919003,
+      "learning_rate": 0.0001,
+      "loss": 0.1126,
+      "num_input_tokens_seen": 40699040,
+      "step": 276
+    },
+    {
+      "epoch": 0.07956183338137791,
+      "loss": 0.10694645345211029,
+      "loss_ce": 0.07613895833492279,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 40699040,
+      "step": 276
+    },
+    {
+      "epoch": 0.07985010089362929,
+      "grad_norm": 47.63916778777146,
+      "learning_rate": 0.0001,
+      "loss": 0.1285,
+      "num_input_tokens_seen": 40833864,
+      "step": 277
+    },
+    {
+      "epoch": 0.07985010089362929,
+      "loss": 0.12656483054161072,
+      "loss_ce": 0.09027943015098572,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 40833864,
+      "step": 277
+    },
+    {
+      "epoch": 0.08013836840588066,
+      "grad_norm": 5.830462727623499,
+      "learning_rate": 0.0001,
+      "loss": 0.0816,
+      "num_input_tokens_seen": 40968944,
+      "step": 278
+    },
+    {
+      "epoch": 0.08013836840588066,
+      "loss": 0.0807839035987854,
+      "loss_ce": 0.06702810525894165,
+      "loss_xval": 0.01373291015625,
+      "num_input_tokens_seen": 40968944,
+      "step": 278
+    },
+    {
+      "epoch": 0.08042663591813203,
+      "grad_norm": 50.914697191412415,
+      "learning_rate": 0.0001,
+      "loss": 0.131,
+      "num_input_tokens_seen": 41141552,
+      "step": 279
+    },
+    {
+      "epoch": 0.08042663591813203,
+      "loss": 0.10976387560367584,
+      "loss_ce": 0.07019783556461334,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 41141552,
+      "step": 279
+    },
+    {
+      "epoch": 0.0807149034303834,
+      "grad_norm": 27.192581330339348,
+      "learning_rate": 0.0001,
+      "loss": 0.096,
+      "num_input_tokens_seen": 41276352,
+      "step": 280
+    },
+    {
+      "epoch": 0.0807149034303834,
+      "loss": 0.11412344127893448,
+      "loss_ce": 0.08633718639612198,
+      "loss_xval": 0.02783203125,
+      "num_input_tokens_seen": 41276352,
+      "step": 280
+    },
+    {
+      "epoch": 0.08100317094263476,
+      "grad_norm": 35.97693953893691,
+      "learning_rate": 0.0001,
+      "loss": 0.093,
+      "num_input_tokens_seen": 41411352,
+      "step": 281
+    },
+    {
+      "epoch": 0.08100317094263476,
+      "loss": 0.08904524892568588,
+      "loss_ce": 0.06298323720693588,
+      "loss_xval": 0.026123046875,
+      "num_input_tokens_seen": 41411352,
+      "step": 281
+    },
+    {
+      "epoch": 0.08129143845488614,
+      "grad_norm": 39.86205075250757,
+      "learning_rate": 0.0001,
+      "loss": 0.115,
+      "num_input_tokens_seen": 41583784,
+      "step": 282
+    },
+    {
+      "epoch": 0.08129143845488614,
+      "loss": 0.1079903095960617,
+      "loss_ce": 0.0660133808851242,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 41583784,
+      "step": 282
+    },
+    {
+      "epoch": 0.0815797059671375,
+      "grad_norm": 16.584565294970684,
+      "learning_rate": 0.0001,
+      "loss": 0.0897,
+      "num_input_tokens_seen": 41718576,
+      "step": 283
+    },
+    {
+      "epoch": 0.0815797059671375,
+      "loss": 0.1088862419128418,
+      "loss_ce": 0.08702802658081055,
+      "loss_xval": 0.0218505859375,
+      "num_input_tokens_seen": 41718576,
+      "step": 283
+    },
+    {
+      "epoch": 0.08186797347938887,
+      "grad_norm": 39.839454046307246,
+      "learning_rate": 0.0001,
+      "loss": 0.093,
+      "num_input_tokens_seen": 41853736,
+      "step": 284
+    },
+    {
+      "epoch": 0.08186797347938887,
+      "loss": 0.09195973724126816,
+      "loss_ce": 0.05681874603033066,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 41853736,
+      "step": 284
+    },
+    {
+      "epoch": 0.08215624099164025,
+      "grad_norm": 4.847042953173711,
+      "learning_rate": 0.0001,
+      "loss": 0.079,
+      "num_input_tokens_seen": 42026296,
+      "step": 285
+    },
+    {
+      "epoch": 0.08215624099164025,
+      "loss": 0.06571406126022339,
+      "loss_ce": 0.056989844888448715,
+      "loss_xval": 0.00872802734375,
+      "num_input_tokens_seen": 42026296,
+      "step": 285
+    },
+    {
+      "epoch": 0.08244450850389161,
+      "grad_norm": 34.08495666742016,
+      "learning_rate": 0.0001,
+      "loss": 0.0979,
+      "num_input_tokens_seen": 42161136,
+      "step": 286
+    },
+    {
+      "epoch": 0.08244450850389161,
+      "loss": 0.11285368353128433,
+      "loss_ce": 0.08119169622659683,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 42161136,
+      "step": 286
+    },
+    {
+      "epoch": 0.08273277601614298,
+      "grad_norm": 17.053837094150378,
+      "learning_rate": 0.0001,
+      "loss": 0.0704,
+      "num_input_tokens_seen": 42296312,
+      "step": 287
+    },
+    {
+      "epoch": 0.08273277601614298,
+      "loss": 0.07522790133953094,
+      "loss_ce": 0.056322261691093445,
+      "loss_xval": 0.0189208984375,
+      "num_input_tokens_seen": 42296312,
+      "step": 287
+    },
+    {
+      "epoch": 0.08302104352839434,
+      "grad_norm": 26.03619151270456,
+      "learning_rate": 0.0001,
+      "loss": 0.088,
+      "num_input_tokens_seen": 42468952,
+      "step": 288
+    },
+    {
+      "epoch": 0.08302104352839434,
+      "loss": 0.0716395452618599,
+      "loss_ce": 0.051970966160297394,
+      "loss_xval": 0.0196533203125,
+      "num_input_tokens_seen": 42468952,
+      "step": 288
+    },
+    {
+      "epoch": 0.08330931104064572,
+      "grad_norm": 19.104599746913536,
+      "learning_rate": 0.0001,
+      "loss": 0.0795,
+      "num_input_tokens_seen": 42603720,
+      "step": 289
+    },
+    {
+      "epoch": 0.08330931104064572,
+      "loss": 0.09239511936903,
+      "loss_ce": 0.07012491673231125,
+      "loss_xval": 0.022216796875,
+      "num_input_tokens_seen": 42603720,
+      "step": 289
+    },
+    {
+      "epoch": 0.0835975785528971,
+      "grad_norm": 12.898381782846887,
+      "learning_rate": 0.0001,
+      "loss": 0.0611,
+      "num_input_tokens_seen": 42738816,
+      "step": 290
+    },
+    {
+      "epoch": 0.0835975785528971,
+      "loss": 0.062205493450164795,
+      "loss_ce": 0.046939074993133545,
+      "loss_xval": 0.0152587890625,
+      "num_input_tokens_seen": 42738816,
+      "step": 290
+    },
+    {
+      "epoch": 0.08388584606514846,
+      "grad_norm": 20.1050497393893,
+      "learning_rate": 0.0001,
+      "loss": 0.0789,
+      "num_input_tokens_seen": 42911264,
+      "step": 291
+    },
+    {
+      "epoch": 0.08388584606514846,
+      "loss": 0.0620073564350605,
+      "loss_ce": 0.0473284013569355,
+      "loss_xval": 0.0146484375,
+      "num_input_tokens_seen": 42911264,
+      "step": 291
+    },
+    {
+      "epoch": 0.08417411357739983,
+      "grad_norm": 2.8626189497123953,
+      "learning_rate": 0.0001,
+      "loss": 0.0681,
+      "num_input_tokens_seen": 43046240,
+      "step": 292
+    },
+    {
+      "epoch": 0.08417411357739983,
+      "loss": 0.08566500246524811,
+      "loss_ce": 0.07007052004337311,
+      "loss_xval": 0.015625,
+      "num_input_tokens_seen": 43046240,
+      "step": 292
+    },
+    {
+      "epoch": 0.08446238108965119,
+      "grad_norm": 26.39635415241842,
+      "learning_rate": 0.0001,
+      "loss": 0.0655,
+      "num_input_tokens_seen": 43181360,
+      "step": 293
+    },
+    {
+      "epoch": 0.08446238108965119,
+      "loss": 0.06426546722650528,
+      "loss_ce": 0.04400942474603653,
+      "loss_xval": 0.020263671875,
+      "num_input_tokens_seen": 43181360,
+      "step": 293
+    },
+    {
+      "epoch": 0.08475064860190257,
+      "grad_norm": 7.208164119900281,
+      "learning_rate": 0.0001,
+      "loss": 0.0653,
+      "num_input_tokens_seen": 43353920,
+      "step": 294
+    },
+    {
+      "epoch": 0.08475064860190257,
+      "loss": 0.053133852779865265,
+      "loss_ce": 0.042292483150959015,
+      "loss_xval": 0.0108642578125,
+      "num_input_tokens_seen": 43353920,
+      "step": 294
+    },
+    {
+      "epoch": 0.08503891611415393,
+      "grad_norm": 26.98415268638157,
+      "learning_rate": 0.0001,
+      "loss": 0.073,
+      "num_input_tokens_seen": 43488696,
+      "step": 295
+    },
+    {
+      "epoch": 0.08503891611415393,
+      "loss": 0.08344084024429321,
+      "loss_ce": 0.06291013956069946,
+      "loss_xval": 0.0205078125,
+      "num_input_tokens_seen": 43488696,
+      "step": 295
+    },
+    {
+      "epoch": 0.0853271836264053,
+      "grad_norm": 10.145044575777948,
+      "learning_rate": 0.0001,
+      "loss": 0.053,
+      "num_input_tokens_seen": 43623704,
+      "step": 296
+    },
+    {
+      "epoch": 0.0853271836264053,
+      "loss": 0.05334699898958206,
+      "loss_ce": 0.04183042794466019,
+      "loss_xval": 0.01153564453125,
+      "num_input_tokens_seen": 43623704,
+      "step": 296
+    },
+    {
+      "epoch": 0.08561545113865668,
+      "grad_norm": 22.607482953404165,
+      "learning_rate": 0.0001,
+      "loss": 0.0696,
+      "num_input_tokens_seen": 43796312,
+      "step": 297
+    },
+    {
+      "epoch": 0.08561545113865668,
+      "loss": 0.056711770594120026,
+      "loss_ce": 0.039812661707401276,
+      "loss_xval": 0.016845703125,
+      "num_input_tokens_seen": 43796312,
+      "step": 297
+    },
+    {
+      "epoch": 0.08590371865090804,
+      "grad_norm": 13.95908121236066,
+      "learning_rate": 0.0001,
+      "loss": 0.0629,
+      "num_input_tokens_seen": 43931104,
+      "step": 298
+    },
+    {
+      "epoch": 0.08590371865090804,
+      "loss": 0.07877244055271149,
+      "loss_ce": 0.057333845645189285,
+      "loss_xval": 0.021484375,
+      "num_input_tokens_seen": 43931104,
+      "step": 298
+    },
+    {
+      "epoch": 0.08619198616315941,
+      "grad_norm": 21.14603428598075,
+      "learning_rate": 0.0001,
+      "loss": 0.0536,
+      "num_input_tokens_seen": 44066328,
+      "step": 299
+    },
+    {
+      "epoch": 0.08619198616315941,
+      "loss": 0.05007364600896835,
+      "loss_ce": 0.03374292701482773,
+      "loss_xval": 0.016357421875,
+      "num_input_tokens_seen": 44066328,
+      "step": 299
+    },
+    {
+      "epoch": 0.08648025367541078,
+      "grad_norm": 16.200678478671318,
+      "learning_rate": 0.0001,
+      "loss": 0.0663,
+      "num_input_tokens_seen": 44238784,
+      "step": 300
+    },
+    {
+      "epoch": 0.08648025367541078,
+      "loss": 0.04962464049458504,
+      "loss_ce": 0.03678818419575691,
+      "loss_xval": 0.0128173828125,
+      "num_input_tokens_seen": 44238784,
+      "step": 300
+    },
+    {
+      "epoch": 0.08676852118766215,
+      "grad_norm": 22.293967998334683,
+      "learning_rate": 0.0001,
+      "loss": 0.0643,
+      "num_input_tokens_seen": 44373552,
+      "step": 301
+    },
+    {
+      "epoch": 0.08676852118766215,
+      "loss": 0.07732580602169037,
+      "loss_ce": 0.05923651158809662,
+      "loss_xval": 0.01806640625,
+      "num_input_tokens_seen": 44373552,
+      "step": 301
+    },
+    {
+      "epoch": 0.08705678869991353,
+      "grad_norm": 15.846001468722628,
+      "learning_rate": 0.0001,
+      "loss": 0.0481,
+      "num_input_tokens_seen": 44508624,
+      "step": 302
+    },
+    {
+      "epoch": 0.08705678869991353,
+      "loss": 0.0516769215464592,
+      "loss_ce": 0.0360976979136467,
+      "loss_xval": 0.01556396484375,
+      "num_input_tokens_seen": 44508624,
+      "step": 302
+    },
+    {
+      "epoch": 0.08734505621216489,
+      "grad_norm": 20.491333852475268,
+      "learning_rate": 0.0001,
+      "loss": 0.0634,
+      "num_input_tokens_seen": 44681000,
+      "step": 303
+    },
+    {
+      "epoch": 0.08734505621216489,
+      "loss": 0.053738269954919815,
+      "loss_ce": 0.034756336361169815,
+      "loss_xval": 0.01904296875,
+      "num_input_tokens_seen": 44681000,
+      "step": 303
+    },
+    {
+      "epoch": 0.08763332372441626,
+      "grad_norm": 13.708979707336596,
+      "learning_rate": 0.0001,
+      "loss": 0.0586,
+      "num_input_tokens_seen": 44815696,
+      "step": 304
+    },
+    {
+      "epoch": 0.08763332372441626,
+      "loss": 0.0743015930056572,
+      "loss_ce": 0.05784880369901657,
+      "loss_xval": 0.0164794921875,
+      "num_input_tokens_seen": 44815696,
+      "step": 304
+    },
+    {
+      "epoch": 0.08792159123666762,
+      "grad_norm": 15.17868299260897,
+      "learning_rate": 0.0001,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 44950768,
+      "step": 305
+    },
+    {
+      "epoch": 0.08792159123666762,
+      "loss": 0.05023224651813507,
+      "loss_ce": 0.03353913128376007,
+      "loss_xval": 0.0167236328125,
+      "num_input_tokens_seen": 44950768,
+      "step": 305
+    },
+    {
+      "epoch": 0.088209858748919,
+      "grad_norm": 14.183441934506382,
+      "learning_rate": 0.0001,
+      "loss": 0.0524,
+      "num_input_tokens_seen": 45123376,
+      "step": 306
+    },
+    {
+      "epoch": 0.088209858748919,
+      "loss": 0.039744533598423004,
+      "loss_ce": 0.029437221586704254,
+      "loss_xval": 0.01031494140625,
+      "num_input_tokens_seen": 45123376,
+      "step": 306
+    },
+    {
+      "epoch": 0.08849812626117036,
+      "grad_norm": 15.764900216413375,
+      "learning_rate": 0.0001,
+      "loss": 0.0557,
+      "num_input_tokens_seen": 45258176,
+      "step": 307
+    },
+    {
+      "epoch": 0.08849812626117036,
+      "loss": 0.06498374789953232,
+      "loss_ce": 0.05014457553625107,
+      "loss_xval": 0.01483154296875,
+      "num_input_tokens_seen": 45258176,
+      "step": 307
+    },
+    {
+      "epoch": 0.08878639377342173,
+      "grad_norm": 18.067172708638466,
+      "learning_rate": 0.0001,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 45393264,
+      "step": 308
+    },
+    {
+      "epoch": 0.08878639377342173,
+      "loss": 0.043388769030570984,
+      "loss_ce": 0.028351232409477234,
+      "loss_xval": 0.0150146484375,
+      "num_input_tokens_seen": 45393264,
+      "step": 308
+    },
+    {
+      "epoch": 0.08907466128567311,
+      "grad_norm": 26.07091528719645,
+      "learning_rate": 0.0001,
+      "loss": 0.0596,
+      "num_input_tokens_seen": 45565840,
+      "step": 309
+    },
+    {
+      "epoch": 0.08907466128567311,
+      "loss": 0.043060220777988434,
+      "loss_ce": 0.027099527418613434,
+      "loss_xval": 0.0159912109375,
+      "num_input_tokens_seen": 45565840,
+      "step": 309
+    },
+    {
+      "epoch": 0.08936292879792447,
+      "grad_norm": 11.447550254988329,
+      "learning_rate": 0.0001,
+      "loss": 0.0516,
+      "num_input_tokens_seen": 45700504,
+      "step": 310
+    },
+    {
+      "epoch": 0.08936292879792447,
+      "loss": 0.06326834857463837,
+      "loss_ce": 0.04712073877453804,
+      "loss_xval": 0.01611328125,
+      "num_input_tokens_seen": 45700504,
+      "step": 310
+    },
+    {
+      "epoch": 0.08965119631017585,
+      "grad_norm": 33.1345531958039,
+      "learning_rate": 0.0001,
+      "loss": 0.0523,
+      "num_input_tokens_seen": 45835696,
+      "step": 311
+    },
+    {
+      "epoch": 0.08965119631017585,
+      "loss": 0.04893258213996887,
+      "loss_ce": 0.02565530315041542,
+      "loss_xval": 0.0233154296875,
+      "num_input_tokens_seen": 45835696,
+      "step": 311
+    },
+    {
+      "epoch": 0.08993946382242721,
+      "grad_norm": 5.875168912525042,
+      "learning_rate": 0.0001,
+      "loss": 0.047,
+      "num_input_tokens_seen": 46008144,
+      "step": 312
+    },
+    {
+      "epoch": 0.08993946382242721,
+      "loss": 0.0328424833714962,
+      "loss_ce": 0.02461036667227745,
+      "loss_xval": 0.00823974609375,
+      "num_input_tokens_seen": 46008144,
+      "step": 312
+    },
+    {
+      "epoch": 0.09022773133467858,
+      "grad_norm": 29.067938055683037,
+      "learning_rate": 0.0001,
+      "loss": 0.0599,
+      "num_input_tokens_seen": 46142920,
+      "step": 313
+    },
+    {
+      "epoch": 0.09022773133467858,
+      "loss": 0.07382102310657501,
+      "loss_ce": 0.046645116060972214,
+      "loss_xval": 0.0272216796875,
+      "num_input_tokens_seen": 46142920,
+      "step": 313
+    },
+    {
+      "epoch": 0.09051599884692996,
+      "grad_norm": 14.145042558138448,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "num_input_tokens_seen": 46278128,
+      "step": 314
+    },
+    {
+      "epoch": 0.09051599884692996,
+      "loss": 0.03773703798651695,
+      "loss_ce": 0.023843908682465553,
+      "loss_xval": 0.013916015625,
+      "num_input_tokens_seen": 46278128,
+      "step": 314
+    },
+    {
+      "epoch": 0.09080426635918132,
+      "grad_norm": 26.968528320820493,
+      "learning_rate": 0.0001,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 46450640,
+      "step": 315
+    },
+    {
+      "epoch": 0.09080426635918132,
+      "loss": 0.039110615849494934,
+      "loss_ce": 0.023127034306526184,
+      "loss_xval": 0.0159912109375,
+      "num_input_tokens_seen": 46450640,
+      "step": 315
+    },
+    {
+      "epoch": 0.0910925338714327,
+      "grad_norm": 21.67591186653818,
+      "learning_rate": 0.0001,
+      "loss": 0.0508,
+      "num_input_tokens_seen": 46585496,
+      "step": 316
+    },
+    {
+      "epoch": 0.0910925338714327,
+      "loss": 0.056679822504520416,
+      "loss_ce": 0.04201994091272354,
+      "loss_xval": 0.0146484375,
+      "num_input_tokens_seen": 46585496,
+      "step": 316
+    },
+    {
+      "epoch": 0.09138080138368405,
+      "grad_norm": 26.61148582926146,
+      "learning_rate": 0.0001,
+      "loss": 0.0416,
+      "num_input_tokens_seen": 46720720,
+      "step": 317
+    },
+    {
+      "epoch": 0.09138080138368405,
+      "loss": 0.04069584235548973,
+      "loss_ce": 0.02201908454298973,
+      "loss_xval": 0.0186767578125,
+      "num_input_tokens_seen": 46720720,
+      "step": 317
+    },
+    {
+      "epoch": 0.09166906889593543,
+      "grad_norm": 41.72680946967378,
+      "learning_rate": 0.0001,
+      "loss": 0.0668,
+      "num_input_tokens_seen": 46893176,
+      "step": 318
+    },
+    {
+      "epoch": 0.09166906889593543,
+      "loss": 0.04476112499833107,
+      "loss_ce": 0.01906532794237137,
+      "loss_xval": 0.025634765625,
+      "num_input_tokens_seen": 46893176,
+      "step": 318
+    },
+    {
+      "epoch": 0.09195733640818679,
+      "grad_norm": 2.8774329155971197,
+      "learning_rate": 0.0001,
+      "loss": 0.0397,
+      "num_input_tokens_seen": 47027912,
+      "step": 319
+    },
+    {
+      "epoch": 0.09195733640818679,
+      "loss": 0.05090821534395218,
+      "loss_ce": 0.03768647834658623,
+      "loss_xval": 0.01324462890625,
+      "num_input_tokens_seen": 47027912,
+      "step": 319
+    },
+    {
+      "epoch": 0.09224560392043817,
+      "grad_norm": 45.18862966542774,
+      "learning_rate": 0.0001,
+      "loss": 0.0591,
+      "num_input_tokens_seen": 47163008,
+      "step": 320
+    },
+    {
+      "epoch": 0.09224560392043817,
+      "loss": 0.05182437598705292,
+      "loss_ce": 0.018575472757220268,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 47163008,
+      "step": 320
+    },
+    {
+      "epoch": 0.09253387143268954,
+      "grad_norm": 21.166922384966146,
+      "learning_rate": 0.0001,
+      "loss": 0.0423,
+      "num_input_tokens_seen": 47335528,
+      "step": 321
+    },
+    {
+      "epoch": 0.09253387143268954,
+      "loss": 0.028793595731258392,
+      "loss_ce": 0.017944596707820892,
+      "loss_xval": 0.0108642578125,
+      "num_input_tokens_seen": 47335528,
+      "step": 321
+    },
+    {
+      "epoch": 0.0928221389449409,
+      "grad_norm": 38.85691929951026,
+      "learning_rate": 0.0001,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 47470296,
+      "step": 322
+    },
+    {
+      "epoch": 0.0928221389449409,
+      "loss": 0.07056182622909546,
+      "loss_ce": 0.03659576177597046,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 47470296,
+      "step": 322
+    },
+    {
+      "epoch": 0.09311040645719228,
+      "grad_norm": 45.36783852672683,
+      "learning_rate": 0.0001,
+      "loss": 0.0599,
+      "num_input_tokens_seen": 47605336,
+      "step": 323
+    },
+    {
+      "epoch": 0.09311040645719228,
+      "loss": 0.04938621446490288,
+      "loss_ce": 0.017876815050840378,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 47605336,
+      "step": 323
+    },
+    {
+      "epoch": 0.09339867396944364,
+      "grad_norm": 17.114573694016176,
+      "learning_rate": 0.0001,
+      "loss": 0.0453,
+      "num_input_tokens_seen": 47777800,
+      "step": 324
+    },
+    {
+      "epoch": 0.09339867396944364,
+      "loss": 0.026957228779792786,
+      "loss_ce": 0.01701231300830841,
+      "loss_xval": 0.00994873046875,
+      "num_input_tokens_seen": 47777800,
+      "step": 324
+    },
+    {
+      "epoch": 0.09368694148169501,
+      "grad_norm": 64.38369471444744,
+      "learning_rate": 0.0001,
+      "loss": 0.0997,
+      "num_input_tokens_seen": 47912616,
+      "step": 325
+    },
+    {
+      "epoch": 0.09368694148169501,
+      "loss": 0.08857642859220505,
+      "loss_ce": 0.03495704382658005,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 47912616,
+      "step": 325
+    },
+    {
+      "epoch": 0.09397520899394639,
+      "grad_norm": 8.95093976787684,
+      "learning_rate": 0.0001,
+      "loss": 0.0257,
+      "num_input_tokens_seen": 48047752,
+      "step": 326
+    },
+    {
+      "epoch": 0.09397520899394639,
+      "loss": 0.027160916477441788,
+      "loss_ce": 0.016243252903223038,
+      "loss_xval": 0.01092529296875,
+      "num_input_tokens_seen": 48047752,
+      "step": 326
+    },
+    {
+      "epoch": 0.09426347650619775,
+      "grad_norm": 73.2897144597942,
+      "learning_rate": 0.0001,
+      "loss": 0.1164,
+      "num_input_tokens_seen": 48220264,
+      "step": 327
+    },
+    {
+      "epoch": 0.09426347650619775,
+      "loss": 0.07901627570390701,
+      "loss_ce": 0.014685220085084438,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 48220264,
+      "step": 327
+    },
+    {
+      "epoch": 0.09455174401844912,
+      "grad_norm": 48.71043719154727,
+      "learning_rate": 0.0001,
+      "loss": 0.0726,
+      "num_input_tokens_seen": 48355104,
+      "step": 328
+    },
+    {
+      "epoch": 0.09455174401844912,
+      "loss": 0.0679316371679306,
+      "loss_ce": 0.0327753871679306,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 48355104,
+      "step": 328
+    },
+    {
+      "epoch": 0.09484001153070049,
+      "grad_norm": 59.90845726737974,
+      "learning_rate": 0.0001,
+      "loss": 0.0818,
+      "num_input_tokens_seen": 48490152,
+      "step": 329
+    },
+    {
+      "epoch": 0.09484001153070049,
+      "loss": 0.06156587600708008,
+      "loss_ce": 0.015392780303955078,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 48490152,
+      "step": 329
+    },
+    {
+      "epoch": 0.09512827904295186,
+      "grad_norm": 95.59853445230955,
+      "learning_rate": 0.0001,
+      "loss": 0.1703,
+      "num_input_tokens_seen": 48662608,
+      "step": 330
+    },
+    {
+      "epoch": 0.09512827904295186,
+      "loss": 0.11032046377658844,
+      "loss_ce": 0.01553286612033844,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 48662608,
+      "step": 330
+    },
+    {
+      "epoch": 0.09541654655520324,
+      "grad_norm": 16.26928208251906,
+      "learning_rate": 0.0001,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 48797416,
+      "step": 331
+    },
+    {
+      "epoch": 0.09541654655520324,
+      "loss": 0.048342738300561905,
+      "loss_ce": 0.03313354030251503,
+      "loss_xval": 0.01519775390625,
+      "num_input_tokens_seen": 48797416,
+      "step": 331
+    },
+    {
+      "epoch": 0.0957048140674546,
+      "grad_norm": 122.17471882082314,
+      "learning_rate": 0.0001,
+      "loss": 0.2516,
+      "num_input_tokens_seen": 48932560,
+      "step": 332
+    },
+    {
+      "epoch": 0.0957048140674546,
+      "loss": 0.17840638756752014,
+      "loss_ce": 0.01605287566781044,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 48932560,
+      "step": 332
+    },
+    {
+      "epoch": 0.09599308157970597,
+      "grad_norm": 52.88764880613315,
+      "learning_rate": 0.0001,
+      "loss": 0.0842,
+      "num_input_tokens_seen": 49105232,
+      "step": 333
+    },
+    {
+      "epoch": 0.09599308157970597,
+      "loss": 0.05301094055175781,
+      "loss_ce": 0.017061231657862663,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 49105232,
+      "step": 333
+    },
+    {
+      "epoch": 0.09628134909195733,
+      "grad_norm": 115.37258236272477,
+      "learning_rate": 0.0001,
+      "loss": 0.2361,
+      "num_input_tokens_seen": 49240080,
+      "step": 334
+    },
+    {
+      "epoch": 0.09628134909195733,
+      "loss": 0.20233498513698578,
+      "loss_ce": 0.03271828591823578,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 49240080,
+      "step": 334
+    },
+    {
+      "epoch": 0.09656961660420871,
+      "grad_norm": 135.4813468564314,
+      "learning_rate": 0.0001,
+      "loss": 0.2934,
+      "num_input_tokens_seen": 49375304,
+      "step": 335
+    },
+    {
+      "epoch": 0.09656961660420871,
+      "loss": 0.21803660690784454,
+      "loss_ce": 0.015644025057554245,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 49375304,
+      "step": 335
+    },
+    {
+      "epoch": 0.09685788411646007,
+      "grad_norm": 45.24827134034747,
+      "learning_rate": 0.0001,
+      "loss": 0.0666,
+      "num_input_tokens_seen": 49547824,
+      "step": 336
+    },
+    {
+      "epoch": 0.09685788411646007,
+      "loss": 0.04550432041287422,
+      "loss_ce": 0.014895188622176647,
+      "loss_xval": 0.0306396484375,
+      "num_input_tokens_seen": 49547824,
+      "step": 336
+    },
+    {
+      "epoch": 0.09714615162871144,
+      "grad_norm": 174.6835376599181,
+      "learning_rate": 0.0001,
+      "loss": 0.4999,
+      "num_input_tokens_seen": 49682744,
+      "step": 337
+    },
+    {
+      "epoch": 0.09714615162871144,
+      "loss": 0.34568923711776733,
+      "loss_ce": 0.028062283992767334,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 49682744,
+      "step": 337
+    },
+    {
+      "epoch": 0.09743441914096282,
+      "grad_norm": 34.17773463839456,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "num_input_tokens_seen": 49817848,
+      "step": 338
+    },
+    {
+      "epoch": 0.09743441914096282,
+      "loss": 0.03003338724374771,
+      "loss_ce": 0.013782776892185211,
+      "loss_xval": 0.0162353515625,
+      "num_input_tokens_seen": 49817848,
+      "step": 338
+    },
+    {
+      "epoch": 0.09772268665321418,
+      "grad_norm": 185.13851267540525,
+      "learning_rate": 0.0001,
+      "loss": 0.5886,
+      "num_input_tokens_seen": 49990280,
+      "step": 339
+    },
+    {
+      "epoch": 0.09772268665321418,
+      "loss": 0.4133293032646179,
+      "loss_ce": 0.014159374870359898,
+      "loss_xval": 0.3984375,
+      "num_input_tokens_seen": 49990280,
+      "step": 339
+    },
+    {
+      "epoch": 0.09801095416546556,
+      "grad_norm": 150.00087701324276,
+      "learning_rate": 0.0001,
+      "loss": 0.3805,
+      "num_input_tokens_seen": 50125128,
+      "step": 340
+    },
+    {
+      "epoch": 0.09801095416546556,
+      "loss": 0.26180437207221985,
+      "loss_ce": 0.03206804767251015,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 50125128,
+      "step": 340
+    },
+    {
+      "epoch": 0.09829922167771692,
+      "grad_norm": 129.76070714652064,
+      "learning_rate": 0.0001,
+      "loss": 0.2883,
+      "num_input_tokens_seen": 50260248,
+      "step": 341
+    },
+    {
+      "epoch": 0.09829922167771692,
+      "loss": 0.21383962035179138,
+      "loss_ce": 0.015231214463710785,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 50260248,
+      "step": 341
+    },
+    {
+      "epoch": 0.09858748918996829,
+      "grad_norm": 234.05736326944017,
+      "learning_rate": 0.0001,
+      "loss": 0.9074,
+      "num_input_tokens_seen": 50432816,
+      "step": 342
+    },
+    {
+      "epoch": 0.09858748918996829,
+      "loss": 0.6195493936538696,
+      "loss_ce": 0.017986856400966644,
+      "loss_xval": 0.6015625,
+      "num_input_tokens_seen": 50432816,
+      "step": 342
+    },
+    {
+      "epoch": 0.09887575670221967,
+      "grad_norm": 43.17761008074917,
+      "learning_rate": 0.0001,
+      "loss": 0.0801,
+      "num_input_tokens_seen": 50567648,
+      "step": 343
+    },
+    {
+      "epoch": 0.09887575670221967,
+      "loss": 0.1131780743598938,
+      "loss_ce": 0.0380437970161438,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 50567648,
+      "step": 343
+    },
+    {
+      "epoch": 0.09916402421447103,
+      "grad_norm": 293.5641993425453,
+      "learning_rate": 0.0001,
+      "loss": 1.3699,
+      "num_input_tokens_seen": 50702776,
+      "step": 344
+    },
+    {
+      "epoch": 0.09916402421447103,
+      "loss": 1.069295048713684,
+      "loss_ce": 0.013631051406264305,
+      "loss_xval": 1.0546875,
+      "num_input_tokens_seen": 50702776,
+      "step": 344
+    },
+    {
+      "epoch": 0.0994522917267224,
+      "grad_norm": 131.71555163423727,
+      "learning_rate": 0.0001,
+      "loss": 0.3338,
+      "num_input_tokens_seen": 50875256,
+      "step": 345
+    },
+    {
+      "epoch": 0.0994522917267224,
+      "loss": 0.2189842015504837,
+      "loss_ce": 0.0239158496260643,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 50875256,
+      "step": 345
+    },
+    {
+      "epoch": 0.09974055923897376,
+      "grad_norm": 233.3297271775485,
+      "learning_rate": 0.0001,
+      "loss": 0.8797,
+      "num_input_tokens_seen": 51010072,
+      "step": 346
+    },
+    {
+      "epoch": 0.09974055923897376,
+      "loss": 0.7266819477081299,
+      "loss_ce": 0.038693614304065704,
+      "loss_xval": 0.6875,
+      "num_input_tokens_seen": 51010072,
+      "step": 346
+    },
+    {
+      "epoch": 0.10002882675122514,
+      "grad_norm": 225.78961503277478,
+      "learning_rate": 0.0001,
+      "loss": 0.8029,
+      "num_input_tokens_seen": 51145272,
+      "step": 347
+    },
+    {
+      "epoch": 0.10002882675122514,
+      "loss": 0.6846581697463989,
+      "loss_ce": 0.019130853936076164,
+      "loss_xval": 0.6640625,
+      "num_input_tokens_seen": 51145272,
+      "step": 347
+    },
+    {
+      "epoch": 0.1003170942634765,
+      "grad_norm": 147.2061227303931,
+      "learning_rate": 0.0001,
+      "loss": 0.4121,
+      "num_input_tokens_seen": 51317720,
+      "step": 348
+    },
+    {
+      "epoch": 0.1003170942634765,
+      "loss": 0.32414305210113525,
+      "loss_ce": 0.03166259080171585,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 51317720,
+      "step": 348
+    },
+    {
+      "epoch": 0.10060536177572788,
+      "grad_norm": 263.8637916525708,
+      "learning_rate": 0.0001,
+      "loss": 1.165,
+      "num_input_tokens_seen": 51452536,
+      "step": 349
+    },
+    {
+      "epoch": 0.10060536177572788,
+      "loss": 0.889431357383728,
+      "loss_ce": 0.04763442277908325,
+      "loss_xval": 0.84375,
+      "num_input_tokens_seen": 51452536,
+      "step": 349
+    },
+    {
+      "epoch": 0.10089362928797925,
+      "grad_norm": 73.43080285637714,
+      "learning_rate": 0.0001,
+      "loss": 0.1414,
+      "num_input_tokens_seen": 51587552,
+      "step": 350
+    },
+    {
+      "epoch": 0.10089362928797925,
+      "loss": 0.11348080635070801,
+      "loss_ce": 0.02668880857527256,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 51587552,
+      "step": 350
+    },
+    {
+      "epoch": 0.10118189680023061,
+      "grad_norm": 307.91044587597645,
+      "learning_rate": 0.0001,
+      "loss": 1.6928,
+      "num_input_tokens_seen": 51760016,
+      "step": 351
+    },
+    {
+      "epoch": 0.10118189680023061,
+      "loss": 1.2498633861541748,
+      "loss_ce": 0.04380872845649719,
+      "loss_xval": 1.203125,
+      "num_input_tokens_seen": 51760016,
+      "step": 351
+    },
+    {
+      "epoch": 0.10147016431248199,
+      "grad_norm": 22.52344498921592,
+      "learning_rate": 0.0001,
+      "loss": 0.0778,
+      "num_input_tokens_seen": 51894824,
+      "step": 352
+    },
+    {
+      "epoch": 0.10147016431248199,
+      "loss": 0.08678604662418365,
+      "loss_ce": 0.051645055413246155,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 51894824,
+      "step": 352
+    },
+    {
+      "epoch": 0.10175843182473335,
+      "grad_norm": 314.1793084218993,
+      "learning_rate": 0.0001,
+      "loss": 1.7642,
+      "num_input_tokens_seen": 52029744,
+      "step": 353
+    },
+    {
+      "epoch": 0.10175843182473335,
+      "loss": 1.4714159965515137,
+      "loss_ce": 0.02805667743086815,
+      "loss_xval": 1.4453125,
+      "num_input_tokens_seen": 52029744,
+      "step": 353
+    },
+    {
+      "epoch": 0.10204669933698472,
+      "grad_norm": 85.04274380544564,
+      "learning_rate": 0.0001,
+      "loss": 0.1873,
+      "num_input_tokens_seen": 52202232,
+      "step": 354
+    },
+    {
+      "epoch": 0.10204669933698472,
+      "loss": 0.15907925367355347,
+      "loss_ce": 0.03090541996061802,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 52202232,
+      "step": 354
+    },
+    {
+      "epoch": 0.1023349668492361,
+      "grad_norm": 282.4662109698405,
+      "learning_rate": 0.0001,
+      "loss": 1.365,
+      "num_input_tokens_seen": 52337056,
+      "step": 355
+    },
+    {
+      "epoch": 0.1023349668492361,
+      "loss": 1.2567670345306396,
+      "loss_ce": 0.05559512972831726,
+      "loss_xval": 1.203125,
+      "num_input_tokens_seen": 52337056,
+      "step": 355
+    },
+    {
+      "epoch": 0.10262323436148746,
+      "grad_norm": 102.65515390331076,
+      "learning_rate": 0.0001,
+      "loss": 0.2295,
+      "num_input_tokens_seen": 52472048,
+      "step": 356
+    },
+    {
+      "epoch": 0.10262323436148746,
+      "loss": 0.2835312783718109,
+      "loss_ce": 0.025718793272972107,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 52472048,
+      "step": 356
+    },
+    {
+      "epoch": 0.10291150187373883,
+      "grad_norm": 252.48243156019015,
+      "learning_rate": 0.0001,
+      "loss": 1.344,
+      "num_input_tokens_seen": 52644424,
+      "step": 357
+    },
+    {
+      "epoch": 0.10291150187373883,
+      "loss": 0.7913997173309326,
+      "loss_ce": 0.029192738234996796,
+      "loss_xval": 0.76171875,
+      "num_input_tokens_seen": 52644424,
+      "step": 357
+    },
+    {
+      "epoch": 0.1031997693859902,
+      "grad_norm": 90.84763123217391,
+      "learning_rate": 0.0001,
+      "loss": 0.2428,
+      "num_input_tokens_seen": 52779144,
+      "step": 358
+    },
+    {
+      "epoch": 0.1031997693859902,
+      "loss": 0.2381390482187271,
+      "loss_ce": 0.05472839996218681,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 52779144,
+      "step": 358
+    },
+    {
+      "epoch": 0.10348803689824157,
+      "grad_norm": 328.37470612355446,
+      "learning_rate": 0.0001,
+      "loss": 2.1571,
+      "num_input_tokens_seen": 52914216,
+      "step": 359
+    },
+    {
+      "epoch": 0.10348803689824157,
+      "loss": 1.3795331716537476,
+      "loss_ce": 0.027970610186457634,
+      "loss_xval": 1.3515625,
+      "num_input_tokens_seen": 52914216,
+      "step": 359
+    },
+    {
+      "epoch": 0.10377630441049293,
+      "grad_norm": 188.38399390828792,
+      "learning_rate": 0.0001,
+      "loss": 0.7471,
+      "num_input_tokens_seen": 53086680,
+      "step": 360
+    },
+    {
+      "epoch": 0.10377630441049293,
+      "loss": 0.49086835980415344,
+      "loss_ce": 0.03334883227944374,
+      "loss_xval": 0.45703125,
+      "num_input_tokens_seen": 53086680,
+      "step": 360
+    },
+    {
+      "epoch": 0.1040645719227443,
+      "grad_norm": 305.99410572222166,
+      "learning_rate": 0.0001,
+      "loss": 1.6008,
+      "num_input_tokens_seen": 53221528,
+      "step": 361
+    },
+    {
+      "epoch": 0.1040645719227443,
+      "loss": 1.3899487257003784,
+      "loss_ce": 0.08233153820037842,
+      "loss_xval": 1.3046875,
+      "num_input_tokens_seen": 53221528,
+      "step": 361
+    },
+    {
+      "epoch": 0.10435283943499568,
+      "grad_norm": 259.50968601472374,
+      "learning_rate": 0.0001,
+      "loss": 1.1558,
+      "num_input_tokens_seen": 53356560,
+      "step": 362
+    },
+    {
+      "epoch": 0.10435283943499568,
+      "loss": 1.188092589378357,
+      "loss_ce": 0.04844410717487335,
+      "loss_xval": 1.140625,
+      "num_input_tokens_seen": 53356560,
+      "step": 362
+    },
+    {
+      "epoch": 0.10464110694724704,
+      "grad_norm": 194.27041230322703,
+      "learning_rate": 0.0001,
+      "loss": 0.8075,
+      "num_input_tokens_seen": 53529184,
+      "step": 363
+    },
+    {
+      "epoch": 0.10464110694724704,
+      "loss": 0.5688841342926025,
+      "loss_ce": 0.03690172731876373,
+      "loss_xval": 0.53125,
+      "num_input_tokens_seen": 53529184,
+      "step": 363
+    },
+    {
+      "epoch": 0.10492937445949842,
+      "grad_norm": 167.4660911295136,
+      "learning_rate": 0.0001,
+      "loss": 0.6972,
+      "num_input_tokens_seen": 53664024,
+      "step": 364
+    },
+    {
+      "epoch": 0.10492937445949842,
+      "loss": 0.770238995552063,
+      "loss_ce": 0.0729733482003212,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 53664024,
+      "step": 364
+    },
+    {
+      "epoch": 0.10521764197174978,
+      "grad_norm": 236.66961574420125,
+      "learning_rate": 0.0001,
+      "loss": 1.3009,
+      "num_input_tokens_seen": 53799152,
+      "step": 365
+    },
+    {
+      "epoch": 0.10521764197174978,
+      "loss": 0.8208091259002686,
+      "loss_ce": 0.03858250379562378,
+      "loss_xval": 0.78125,
+      "num_input_tokens_seen": 53799152,
+      "step": 365
+    },
+    {
+      "epoch": 0.10550590948400115,
+      "grad_norm": 186.46485981451747,
+      "learning_rate": 0.0001,
+      "loss": 1.0544,
+      "num_input_tokens_seen": 53971760,
+      "step": 366
+    },
+    {
+      "epoch": 0.10550590948400115,
+      "loss": 0.4651481509208679,
+      "loss_ce": 0.03814617544412613,
+      "loss_xval": 0.427734375,
+      "num_input_tokens_seen": 53971760,
+      "step": 366
+    },
+    {
+      "epoch": 0.10579417699625253,
+      "grad_norm": 198.19441855401575,
+      "learning_rate": 0.0001,
+      "loss": 0.9633,
+      "num_input_tokens_seen": 54106544,
+      "step": 367
+    },
+    {
+      "epoch": 0.10579417699625253,
+      "loss": 0.6854445934295654,
+      "loss_ce": 0.06679220497608185,
+      "loss_xval": 0.6171875,
+      "num_input_tokens_seen": 54106544,
+      "step": 367
+    },
+    {
+      "epoch": 0.10608244450850389,
+      "grad_norm": 173.81988841584263,
+      "learning_rate": 0.0001,
+      "loss": 0.7825,
+      "num_input_tokens_seen": 54241616,
+      "step": 368
+    },
+    {
+      "epoch": 0.10608244450850389,
+      "loss": 0.26582324504852295,
+      "loss_ce": 0.03242482244968414,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 54241616,
+      "step": 368
+    },
+    {
+      "epoch": 0.10637071202075526,
+      "grad_norm": 203.49345931635668,
+      "learning_rate": 0.0001,
+      "loss": 1.0004,
+      "num_input_tokens_seen": 54414272,
+      "step": 369
+    },
+    {
+      "epoch": 0.10637071202075526,
+      "loss": 0.8784339427947998,
+      "loss_ce": 0.04493783041834831,
+      "loss_xval": 0.83203125,
+      "num_input_tokens_seen": 54414272,
+      "step": 369
+    },
+    {
+      "epoch": 0.10665897953300663,
+      "grad_norm": 349.5785229713507,
+      "learning_rate": 0.0001,
+      "loss": 1.8979,
+      "num_input_tokens_seen": 54548968,
+      "step": 370
+    },
+    {
+      "epoch": 0.10665897953300663,
+      "loss": 1.4802519083023071,
+      "loss_ce": 0.07986137270927429,
+      "loss_xval": 1.3984375,
+      "num_input_tokens_seen": 54548968,
+      "step": 370
+    },
+    {
+      "epoch": 0.106947247045258,
+      "grad_norm": 74.2585336466779,
+      "learning_rate": 0.0001,
+      "loss": 0.1741,
+      "num_input_tokens_seen": 54683992,
+      "step": 371
+    },
+    {
+      "epoch": 0.106947247045258,
+      "loss": 0.19732186198234558,
+      "loss_ce": 0.06493661552667618,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 54683992,
+      "step": 371
+    },
+    {
+      "epoch": 0.10723551455750936,
+      "grad_norm": 349.3312247600097,
+      "learning_rate": 0.0001,
+      "loss": 1.9217,
+      "num_input_tokens_seen": 54856472,
+      "step": 372
+    },
+    {
+      "epoch": 0.10723551455750936,
+      "loss": 1.735950231552124,
+      "loss_ce": 0.09630171954631805,
+      "loss_xval": 1.640625,
+      "num_input_tokens_seen": 54856472,
+      "step": 372
+    },
+    {
+      "epoch": 0.10752378206976074,
+      "grad_norm": 11.985561447706766,
+      "learning_rate": 0.0001,
+      "loss": 0.1713,
+      "num_input_tokens_seen": 54991288,
+      "step": 373
+    },
+    {
+      "epoch": 0.10752378206976074,
+      "loss": 0.19064179062843323,
+      "loss_ce": 0.15305939316749573,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 54991288,
+      "step": 373
+    },
+    {
+      "epoch": 0.10781204958201211,
+      "grad_norm": 330.89079551948697,
+      "learning_rate": 0.0001,
+      "loss": 1.8265,
+      "num_input_tokens_seen": 55126536,
+      "step": 374
+    },
+    {
+      "epoch": 0.10781204958201211,
+      "loss": 1.5767598152160645,
+      "loss_ce": 0.08359581232070923,
+      "loss_xval": 1.4921875,
+      "num_input_tokens_seen": 55126536,
+      "step": 374
+    },
+    {
+      "epoch": 0.10810031709426347,
+      "grad_norm": 24.35537299172033,
+      "learning_rate": 0.0001,
+      "loss": 0.1601,
+      "num_input_tokens_seen": 55299168,
+      "step": 375
+    },
+    {
+      "epoch": 0.10810031709426347,
+      "loss": 0.126111701130867,
+      "loss_ce": 0.0854165107011795,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 55299168,
+      "step": 375
+    },
+    {
+      "epoch": 0.10838858460651485,
+      "grad_norm": 285.65327944843983,
+      "learning_rate": 0.0001,
+      "loss": 1.5406,
+      "num_input_tokens_seen": 55433896,
+      "step": 376
+    },
+    {
+      "epoch": 0.10838858460651485,
+      "loss": 1.349381446838379,
+      "loss_ce": 0.1408853530883789,
+      "loss_xval": 1.2109375,
+      "num_input_tokens_seen": 55433896,
+      "step": 376
+    },
+    {
+      "epoch": 0.10867685211876621,
+      "grad_norm": 24.788341756279642,
+      "learning_rate": 0.0001,
+      "loss": 0.1178,
+      "num_input_tokens_seen": 55568936,
+      "step": 377
+    },
+    {
+      "epoch": 0.10867685211876621,
+      "loss": 0.12482555210590363,
+      "loss_ce": 0.08234508335590363,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 55568936,
+      "step": 377
+    },
+    {
+      "epoch": 0.10896511963101758,
+      "grad_norm": 249.08100726158384,
+      "learning_rate": 0.0001,
+      "loss": 1.3048,
+      "num_input_tokens_seen": 55741328,
+      "step": 378
+    },
+    {
+      "epoch": 0.10896511963101758,
+      "loss": 1.1989914178848267,
+      "loss_ce": 0.09742889553308487,
+      "loss_xval": 1.1015625,
+      "num_input_tokens_seen": 55741328,
+      "step": 378
+    },
+    {
+      "epoch": 0.10925338714326896,
+      "grad_norm": 41.92556961556827,
+      "learning_rate": 0.0001,
+      "loss": 0.1731,
+      "num_input_tokens_seen": 55876224,
+      "step": 379
+    },
+    {
+      "epoch": 0.10925338714326896,
+      "loss": 0.23099157214164734,
+      "loss_ce": 0.14755651354789734,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 55876224,
+      "step": 379
+    },
+    {
+      "epoch": 0.10954165465552032,
+      "grad_norm": 200.56550278580963,
+      "learning_rate": 0.0001,
+      "loss": 0.9848,
+      "num_input_tokens_seen": 56011368,
+      "step": 380
+    },
+    {
+      "epoch": 0.10954165465552032,
+      "loss": 1.0198489427566528,
+      "loss_ce": 0.09260280430316925,
+      "loss_xval": 0.92578125,
+      "num_input_tokens_seen": 56011368,
+      "step": 380
+    },
+    {
+      "epoch": 0.1098299221677717,
+      "grad_norm": 85.41435392852556,
+      "learning_rate": 0.0001,
+      "loss": 0.3012,
+      "num_input_tokens_seen": 56183928,
+      "step": 381
+    },
+    {
+      "epoch": 0.1098299221677717,
+      "loss": 0.28134530782699585,
+      "loss_ce": 0.09213632345199585,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 56183928,
+      "step": 381
+    },
+    {
+      "epoch": 0.11011818968002306,
+      "grad_norm": 138.8160607265775,
+      "learning_rate": 0.0001,
+      "loss": 0.5655,
+      "num_input_tokens_seen": 56318752,
+      "step": 382
+    },
+    {
+      "epoch": 0.11011818968002306,
+      "loss": 0.6360880136489868,
+      "loss_ce": 0.131449356675148,
+      "loss_xval": 0.50390625,
+      "num_input_tokens_seen": 56318752,
+      "step": 382
+    },
+    {
+      "epoch": 0.11040645719227443,
+      "grad_norm": 125.28655562326834,
+      "learning_rate": 0.0001,
+      "loss": 0.4428,
+      "num_input_tokens_seen": 56453792,
+      "step": 383
+    },
+    {
+      "epoch": 0.11040645719227443,
+      "loss": 0.40254080295562744,
+      "loss_ce": 0.07270681858062744,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 56453792,
+      "step": 383
+    },
+    {
+      "epoch": 0.1106947247045258,
+      "grad_norm": 81.8855747466578,
+      "learning_rate": 0.0001,
+      "loss": 0.2815,
+      "num_input_tokens_seen": 56626256,
+      "step": 384
+    },
+    {
+      "epoch": 0.1106947247045258,
+      "loss": 0.1878707855939865,
+      "loss_ce": 0.06659393757581711,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 56626256,
+      "step": 384
+    },
+    {
+      "epoch": 0.11098299221677717,
+      "grad_norm": 134.69538035742877,
+      "learning_rate": 0.0001,
+      "loss": 0.5708,
+      "num_input_tokens_seen": 56761000,
+      "step": 385
+    },
+    {
+      "epoch": 0.11098299221677717,
+      "loss": 0.6164034605026245,
+      "loss_ce": 0.1088351458311081,
+      "loss_xval": 0.5078125,
+      "num_input_tokens_seen": 56761000,
+      "step": 385
+    },
+    {
+      "epoch": 0.11127125972902854,
+      "grad_norm": 51.9899502391935,
+      "learning_rate": 0.0001,
+      "loss": 0.1245,
+      "num_input_tokens_seen": 56896072,
+      "step": 386
+    },
+    {
+      "epoch": 0.11127125972902854,
+      "loss": 0.11218064278364182,
+      "loss_ce": 0.050565652549266815,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 56896072,
+      "step": 386
+    },
+    {
+      "epoch": 0.1115595272412799,
+      "grad_norm": 134.96651808205962,
+      "learning_rate": 0.0001,
+      "loss": 0.5115,
+      "num_input_tokens_seen": 57068584,
+      "step": 387
+    },
+    {
+      "epoch": 0.1115595272412799,
+      "loss": 0.45479512214660645,
+      "loss_ce": 0.04708027094602585,
+      "loss_xval": 0.408203125,
+      "num_input_tokens_seen": 57068584,
+      "step": 387
+    },
+    {
+      "epoch": 0.11184779475353128,
+      "grad_norm": 3.9921421710123757,
+      "learning_rate": 0.0001,
+      "loss": 0.0814,
+      "num_input_tokens_seen": 57203448,
+      "step": 388
+    },
+    {
+      "epoch": 0.11184779475353128,
+      "loss": 0.11016294360160828,
+      "loss_ce": 0.08075925707817078,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 57203448,
+      "step": 388
+    },
+    {
+      "epoch": 0.11213606226578264,
+      "grad_norm": 154.34349074142602,
+      "learning_rate": 0.0001,
+      "loss": 0.5716,
+      "num_input_tokens_seen": 57338608,
+      "step": 389
+    },
+    {
+      "epoch": 0.11213606226578264,
+      "loss": 0.5179411172866821,
+      "loss_ce": 0.03551921993494034,
+      "loss_xval": 0.482421875,
+      "num_input_tokens_seen": 57338608,
+      "step": 389
+    },
+    {
+      "epoch": 0.11242432977803402,
+      "grad_norm": 28.336565736169785,
+      "learning_rate": 0.0001,
+      "loss": 0.0975,
+      "num_input_tokens_seen": 57511192,
+      "step": 390
+    },
+    {
+      "epoch": 0.11242432977803402,
+      "loss": 0.10524661839008331,
+      "loss_ce": 0.036337923258543015,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 57511192,
+      "step": 390
+    },
+    {
+      "epoch": 0.11271259729028539,
+      "grad_norm": 111.97474819267906,
+      "learning_rate": 0.0001,
+      "loss": 0.3659,
+      "num_input_tokens_seen": 57646056,
+      "step": 391
+    },
+    {
+      "epoch": 0.11271259729028539,
+      "loss": 0.434956431388855,
+      "loss_ce": 0.05531773343682289,
+      "loss_xval": 0.37890625,
+      "num_input_tokens_seen": 57646056,
+      "step": 391
+    },
+    {
+      "epoch": 0.11300086480253675,
+      "grad_norm": 54.641254485706455,
+      "learning_rate": 0.0001,
+      "loss": 0.1127,
+      "num_input_tokens_seen": 57781256,
+      "step": 392
+    },
+    {
+      "epoch": 0.11300086480253675,
+      "loss": 0.11866104602813721,
+      "loss_ce": 0.02973281964659691,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 57781256,
+      "step": 392
+    },
+    {
+      "epoch": 0.11328913231478813,
+      "grad_norm": 65.2115128081605,
+      "learning_rate": 0.0001,
+      "loss": 0.1522,
+      "num_input_tokens_seen": 57953776,
+      "step": 393
+    },
+    {
+      "epoch": 0.11328913231478813,
+      "loss": 0.16455772519111633,
+      "loss_ce": 0.02844933047890663,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 57953776,
+      "step": 393
+    },
+    {
+      "epoch": 0.11357739982703949,
+      "grad_norm": 72.45753153535337,
+      "learning_rate": 0.0001,
+      "loss": 0.1703,
+      "num_input_tokens_seen": 58088568,
+      "step": 394
+    },
+    {
+      "epoch": 0.11357739982703949,
+      "loss": 0.16719621419906616,
+      "loss_ce": 0.045064859092235565,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 58088568,
+      "step": 394
+    },
+    {
+      "epoch": 0.11386566733929086,
+      "grad_norm": 16.350567249204943,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "num_input_tokens_seen": 58223688,
+      "step": 395
+    },
+    {
+      "epoch": 0.11386566733929086,
+      "loss": 0.03724553436040878,
+      "loss_ce": 0.022444508969783783,
+      "loss_xval": 0.0147705078125,
+      "num_input_tokens_seen": 58223688,
+      "step": 395
+    },
+    {
+      "epoch": 0.11415393485154224,
+      "grad_norm": 80.20883394198827,
+      "learning_rate": 0.0001,
+      "loss": 0.1971,
+      "num_input_tokens_seen": 58396152,
+      "step": 396
+    },
+    {
+      "epoch": 0.11415393485154224,
+      "loss": 0.1776159405708313,
+      "loss_ce": 0.0247839093208313,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 58396152,
+      "step": 396
+    },
+    {
+      "epoch": 0.1144422023637936,
+      "grad_norm": 13.45347570730963,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 58531008,
+      "step": 397
+    },
+    {
+      "epoch": 0.1144422023637936,
+      "loss": 0.05418435111641884,
+      "loss_ce": 0.03697243705391884,
+      "loss_xval": 0.0172119140625,
+      "num_input_tokens_seen": 58531008,
+      "step": 397
+    },
+    {
+      "epoch": 0.11473046987604497,
+      "grad_norm": 66.82835410580104,
+      "learning_rate": 0.0001,
+      "loss": 0.1362,
+      "num_input_tokens_seen": 58666032,
+      "step": 398
+    },
+    {
+      "epoch": 0.11473046987604497,
+      "loss": 0.1477038562297821,
+      "loss_ce": 0.017088618129491806,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 58666032,
+      "step": 398
+    },
+    {
+      "epoch": 0.11501873738829634,
+      "grad_norm": 33.58390379792348,
+      "learning_rate": 0.0001,
+      "loss": 0.0668,
+      "num_input_tokens_seen": 58838640,
+      "step": 399
+    },
+    {
+      "epoch": 0.11501873738829634,
+      "loss": 0.04878139868378639,
+      "loss_ce": 0.019392970949411392,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 58838640,
+      "step": 399
+    },
+    {
+      "epoch": 0.11530700490054771,
+      "grad_norm": 37.13352731577908,
+      "learning_rate": 0.0001,
+      "loss": 0.0724,
+      "num_input_tokens_seen": 58973440,
+      "step": 400
+    },
+    {
+      "epoch": 0.11530700490054771,
+      "loss": 0.08155344426631927,
+      "loss_ce": 0.037638649344444275,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 58973440,
+      "step": 400
+    },
+    {
+      "epoch": 0.11559527241279907,
+      "grad_norm": 58.5907348005972,
+      "learning_rate": 0.0001,
+      "loss": 0.1073,
+      "num_input_tokens_seen": 59108488,
+      "step": 401
+    },
+    {
+      "epoch": 0.11559527241279907,
+      "loss": 0.0918554961681366,
+      "loss_ce": 0.011746861040592194,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 59108488,
+      "step": 401
+    },
+    {
+      "epoch": 0.11588353992505045,
+      "grad_norm": 21.688355763409646,
+      "learning_rate": 0.0001,
+      "loss": 0.0501,
+      "num_input_tokens_seen": 59281024,
+      "step": 402
+    },
+    {
+      "epoch": 0.11588353992505045,
+      "loss": 0.028341641649603844,
+      "loss_ce": 0.016180386766791344,
+      "loss_xval": 0.01214599609375,
+      "num_input_tokens_seen": 59281024,
+      "step": 402
+    },
+    {
+      "epoch": 0.11617180743730182,
+      "grad_norm": 62.076919731958114,
+      "learning_rate": 0.0001,
+      "loss": 0.1351,
+      "num_input_tokens_seen": 59415824,
+      "step": 403
+    },
+    {
+      "epoch": 0.11617180743730182,
+      "loss": 0.15151304006576538,
+      "loss_ce": 0.032372407615184784,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 59415824,
+      "step": 403
+    },
+    {
+      "epoch": 0.11646007494955318,
+      "grad_norm": 10.21746319986935,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "num_input_tokens_seen": 59550856,
+      "step": 404
+    },
+    {
+      "epoch": 0.11646007494955318,
+      "loss": 0.019688857719302177,
+      "loss_ce": 0.010304702445864677,
+      "loss_xval": 0.0093994140625,
+      "num_input_tokens_seen": 59550856,
+      "step": 404
+    },
+    {
+      "epoch": 0.11674834246180456,
+      "grad_norm": 51.93656716076114,
+      "learning_rate": 0.0001,
+      "loss": 0.0985,
+      "num_input_tokens_seen": 59723384,
+      "step": 405
+    },
+    {
+      "epoch": 0.11674834246180456,
+      "loss": 0.11036922037601471,
+      "loss_ce": 0.013872633688151836,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 59723384,
+      "step": 405
+    },
+    {
+      "epoch": 0.11703660997405592,
+      "grad_norm": 7.218077949279075,
+      "learning_rate": 0.0001,
+      "loss": 0.0292,
+      "num_input_tokens_seen": 59858224,
+      "step": 406
+    },
+    {
+      "epoch": 0.11703660997405592,
+      "loss": 0.03941994905471802,
+      "loss_ce": 0.03063851408660412,
+      "loss_xval": 0.0087890625,
+      "num_input_tokens_seen": 59858224,
+      "step": 406
+    },
+    {
+      "epoch": 0.1173248774863073,
+      "grad_norm": 38.771615811302745,
+      "learning_rate": 0.0001,
+      "loss": 0.0552,
+      "num_input_tokens_seen": 59993184,
+      "step": 407
+    },
+    {
+      "epoch": 0.1173248774863073,
+      "loss": 0.0620470829308033,
+      "loss_ce": 0.009465297684073448,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 59993184,
+      "step": 407
+    },
+    {
+      "epoch": 0.11761314499855867,
+      "grad_norm": 21.05248979073788,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "num_input_tokens_seen": 60165568,
+      "step": 408
+    },
+    {
+      "epoch": 0.11761314499855867,
+      "loss": 0.027896784245967865,
+      "loss_ce": 0.010372065007686615,
+      "loss_xval": 0.017578125,
+      "num_input_tokens_seen": 60165568,
+      "step": 408
+    },
+    {
+      "epoch": 0.11790141251081003,
+      "grad_norm": 24.493599892108804,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 60300328,
+      "step": 409
+    },
+    {
+      "epoch": 0.11790141251081003,
+      "loss": 0.05128861218690872,
+      "loss_ce": 0.027080543339252472,
+      "loss_xval": 0.024169921875,
+      "num_input_tokens_seen": 60300328,
+      "step": 409
+    },
+    {
+      "epoch": 0.1181896800230614,
+      "grad_norm": 31.759586149499057,
+      "learning_rate": 0.0001,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 60435400,
+      "step": 410
+    },
+    {
+      "epoch": 0.1181896800230614,
+      "loss": 0.03685798496007919,
+      "loss_ce": 0.008232494816184044,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 60435400,
+      "step": 410
+    },
+    {
+      "epoch": 0.11847794753531277,
+      "grad_norm": 16.79595672478788,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "num_input_tokens_seen": 60607824,
+      "step": 411
+    },
+    {
+      "epoch": 0.11847794753531277,
+      "loss": 0.021884309127926826,
+      "loss_ce": 0.009623872116208076,
+      "loss_xval": 0.01226806640625,
+      "num_input_tokens_seen": 60607824,
+      "step": 411
+    },
+    {
+      "epoch": 0.11876621504756414,
+      "grad_norm": 33.97780050017176,
+      "learning_rate": 0.0001,
+      "loss": 0.0523,
+      "num_input_tokens_seen": 60742616,
+      "step": 412
+    },
+    {
+      "epoch": 0.11876621504756414,
+      "loss": 0.05837390199303627,
+      "loss_ce": 0.02347705140709877,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 60742616,
+      "step": 412
+    },
+    {
+      "epoch": 0.1190544825598155,
+      "grad_norm": 8.233141537208109,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "num_input_tokens_seen": 60877904,
+      "step": 413
+    },
+    {
+      "epoch": 0.1190544825598155,
+      "loss": 0.015995457768440247,
+      "loss_ce": 0.007229284383356571,
+      "loss_xval": 0.0087890625,
+      "num_input_tokens_seen": 60877904,
+      "step": 413
+    },
+    {
+      "epoch": 0.11934275007206688,
+      "grad_norm": 34.01052905899437,
+      "learning_rate": 0.0001,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 61050352,
+      "step": 414
+    },
+    {
+      "epoch": 0.11934275007206688,
+      "loss": 0.042255695909261703,
+      "loss_ce": 0.007847124710679054,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 61050352,
+      "step": 414
+    },
+    {
+      "epoch": 0.11963101758431825,
+      "grad_norm": 3.9841818309777546,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "num_input_tokens_seen": 61185088,
+      "step": 415
+    },
+    {
+      "epoch": 0.11963101758431825,
+      "loss": 0.029205160215497017,
+      "loss_ce": 0.02410300448536873,
+      "loss_xval": 0.005096435546875,
+      "num_input_tokens_seen": 61185088,
+      "step": 415
+    },
+    {
+      "epoch": 0.11991928509656961,
+      "grad_norm": 33.199360451976254,
+      "learning_rate": 0.0001,
+      "loss": 0.0404,
+      "num_input_tokens_seen": 61320072,
+      "step": 416
+    },
+    {
+      "epoch": 0.11991928509656961,
+      "loss": 0.04171193018555641,
+      "loss_ce": 0.006433609873056412,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 61320072,
+      "step": 416
+    },
+    {
+      "epoch": 0.12020755260882099,
+      "grad_norm": 5.067381489080464,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "num_input_tokens_seen": 61492440,
+      "step": 417
+    },
+    {
+      "epoch": 0.12020755260882099,
+      "loss": 0.011221684515476227,
+      "loss_ce": 0.007321156561374664,
+      "loss_xval": 0.00390625,
+      "num_input_tokens_seen": 61492440,
+      "step": 417
+    },
+    {
+      "epoch": 0.12049582012107235,
+      "grad_norm": 26.45059905845838,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "num_input_tokens_seen": 61627200,
+      "step": 418
+    },
+    {
+      "epoch": 0.12049582012107235,
+      "loss": 0.04936272278428078,
+      "loss_ce": 0.022110525518655777,
+      "loss_xval": 0.0272216796875,
+      "num_input_tokens_seen": 61627200,
+      "step": 418
+    },
+    {
+      "epoch": 0.12078408763332373,
+      "grad_norm": 1.786488274094738,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "num_input_tokens_seen": 61762344,
+      "step": 419
+    },
+    {
+      "epoch": 0.12078408763332373,
+      "loss": 0.010375022888183594,
+      "loss_ce": 0.005908012855798006,
+      "loss_xval": 0.00445556640625,
+      "num_input_tokens_seen": 61762344,
+      "step": 419
+    },
+    {
+      "epoch": 0.1210723551455751,
+      "grad_norm": 21.591453869968998,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "num_input_tokens_seen": 61934872,
+      "step": 420
+    },
+    {
+      "epoch": 0.1210723551455751,
+      "loss": 0.026244094595313072,
+      "loss_ce": 0.006491592153906822,
+      "loss_xval": 0.019775390625,
+      "num_input_tokens_seen": 61934872,
+      "step": 420
+    },
+    {
+      "epoch": 0.12136062265782646,
+      "grad_norm": 1.327132608773384,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "num_input_tokens_seen": 62069656,
+      "step": 421
+    },
+    {
+      "epoch": 0.12136062265782646,
+      "loss": 0.029985252767801285,
+      "loss_ce": 0.02338201180100441,
+      "loss_xval": 0.006591796875,
+      "num_input_tokens_seen": 62069656,
+      "step": 421
+    },
+    {
+      "epoch": 0.12164889017007784,
+      "grad_norm": 15.53135837910297,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "num_input_tokens_seen": 62204656,
+      "step": 422
+    },
+    {
+      "epoch": 0.12164889017007784,
+      "loss": 0.017050063237547874,
+      "loss_ce": 0.0061705466359853745,
+      "loss_xval": 0.0108642578125,
+      "num_input_tokens_seen": 62204656,
+      "step": 422
+    },
+    {
+      "epoch": 0.1219371576823292,
+      "grad_norm": 3.8104541358240582,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "num_input_tokens_seen": 62377208,
+      "step": 423
+    },
+    {
+      "epoch": 0.1219371576823292,
+      "loss": 0.008296707645058632,
+      "loss_ce": 0.005021790042519569,
+      "loss_xval": 0.0032806396484375,
+      "num_input_tokens_seen": 62377208,
+      "step": 423
+    },
+    {
+      "epoch": 0.12222542519458057,
+      "grad_norm": 13.700976362376961,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "num_input_tokens_seen": 62512016,
+      "step": 424
+    },
+    {
+      "epoch": 0.12222542519458057,
+      "loss": 0.02982976660132408,
+      "loss_ce": 0.019648339599370956,
+      "loss_xval": 0.01019287109375,
+      "num_input_tokens_seen": 62512016,
+      "step": 424
+    },
+    {
+      "epoch": 0.12251369270683193,
+      "grad_norm": 5.461601173582089,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "num_input_tokens_seen": 62645520,
+      "step": 425
+    },
+    {
+      "epoch": 0.12251369270683193,
+      "loss": 0.009898142889142036,
+      "loss_ce": 0.0057897139340639114,
+      "loss_xval": 0.004119873046875,
+      "num_input_tokens_seen": 62645520,
+      "step": 425
+    },
+    {
+      "epoch": 0.12280196021908331,
+      "grad_norm": 12.868724266222875,
+      "learning_rate": 0.0001,
+      "loss": 0.0219,
+      "num_input_tokens_seen": 62818008,
+      "step": 426
+    },
+    {
+      "epoch": 0.12280196021908331,
+      "loss": 0.011952612549066544,
+      "loss_ce": 0.004845831543207169,
+      "loss_xval": 0.007110595703125,
+      "num_input_tokens_seen": 62818008,
+      "step": 426
+    },
+    {
+      "epoch": 0.12309022773133468,
+      "grad_norm": 5.962625423686631,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "num_input_tokens_seen": 62952776,
+      "step": 427
+    },
+    {
+      "epoch": 0.12309022773133468,
+      "loss": 0.027931518852710724,
+      "loss_ce": 0.02284843474626541,
+      "loss_xval": 0.005096435546875,
+      "num_input_tokens_seen": 62952776,
+      "step": 427
+    },
+    {
+      "epoch": 0.12337849524358604,
+      "grad_norm": 13.40354193419642,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "num_input_tokens_seen": 63087864,
+      "step": 428
+    },
+    {
+      "epoch": 0.12337849524358604,
+      "loss": 0.012452416121959686,
+      "loss_ce": 0.004254631698131561,
+      "loss_xval": 0.0081787109375,
+      "num_input_tokens_seen": 63087864,
+      "step": 428
+    },
+    {
+      "epoch": 0.12366676275583742,
+      "grad_norm": 3.119445339769395,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "num_input_tokens_seen": 63260320,
+      "step": 429
+    },
+    {
+      "epoch": 0.12366676275583742,
+      "loss": 0.007456950377672911,
+      "loss_ce": 0.004637889098376036,
+      "loss_xval": 0.0028228759765625,
+      "num_input_tokens_seen": 63260320,
+      "step": 429
+    },
+    {
+      "epoch": 0.12395503026808878,
+      "grad_norm": 11.337685211261718,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "num_input_tokens_seen": 63395120,
+      "step": 430
+    },
+    {
+      "epoch": 0.12395503026808878,
+      "loss": 0.023988714441657066,
+      "loss_ce": 0.01637076400220394,
+      "loss_xval": 0.00762939453125,
+      "num_input_tokens_seen": 63395120,
+      "step": 430
+    },
+    {
+      "epoch": 0.12424329778034016,
+      "grad_norm": 0.620352290348227,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "num_input_tokens_seen": 63530144,
+      "step": 431
+    },
+    {
+      "epoch": 0.12424329778034016,
+      "loss": 0.007503397762775421,
+      "loss_ce": 0.0034550500568002462,
+      "loss_xval": 0.004058837890625,
+      "num_input_tokens_seen": 63530144,
+      "step": 431
+    },
+    {
+      "epoch": 0.12453156529259153,
+      "grad_norm": 8.636674289000593,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "num_input_tokens_seen": 63702656,
+      "step": 432
+    },
+    {
+      "epoch": 0.12453156529259153,
+      "loss": 0.009306371212005615,
+      "loss_ce": 0.004097401164472103,
+      "loss_xval": 0.005218505859375,
+      "num_input_tokens_seen": 63702656,
+      "step": 432
+    },
+    {
+      "epoch": 0.12481983280484289,
+      "grad_norm": 1.1493436039330713,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "num_input_tokens_seen": 63837392,
+      "step": 433
+    },
+    {
+      "epoch": 0.12481983280484289,
+      "loss": 0.019667595624923706,
+      "loss_ce": 0.01628205180168152,
+      "loss_xval": 0.003387451171875,
+      "num_input_tokens_seen": 63837392,
+      "step": 433
+    },
+    {
+      "epoch": 0.12510810031709427,
+      "grad_norm": 6.650570588966674,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "num_input_tokens_seen": 63972432,
+      "step": 434
+    },
+    {
+      "epoch": 0.12510810031709427,
+      "loss": 0.006774544715881348,
+      "loss_ce": 0.002935052150860429,
+      "loss_xval": 0.00384521484375,
+      "num_input_tokens_seen": 63972432,
+      "step": 434
+    },
+    {
+      "epoch": 0.12539636782934563,
+      "grad_norm": 1.5652798425645655,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "num_input_tokens_seen": 64144888,
+      "step": 435
+    },
+    {
+      "epoch": 0.12539636782934563,
+      "loss": 0.008809241466224194,
+      "loss_ce": 0.005465658847242594,
+      "loss_xval": 0.0033416748046875,
+      "num_input_tokens_seen": 64144888,
+      "step": 435
+    },
+    {
+      "epoch": 0.125684635341597,
+      "grad_norm": 7.078695658340988,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "num_input_tokens_seen": 64279712,
+      "step": 436
+    },
+    {
+      "epoch": 0.125684635341597,
+      "loss": 0.02221812680363655,
+      "loss_ce": 0.0176176019012928,
+      "loss_xval": 0.004608154296875,
+      "num_input_tokens_seen": 64279712,
+      "step": 436
+    },
+    {
+      "epoch": 0.12597290285384838,
+      "grad_norm": 4.718172704478871,
+      "learning_rate": 0.0001,
+      "loss": 0.0067,
+      "num_input_tokens_seen": 64414904,
+      "step": 437
+    },
+    {
+      "epoch": 0.12597290285384838,
+      "loss": 0.006065657362341881,
+      "loss_ce": 0.0030663516372442245,
+      "loss_xval": 0.0030059814453125,
+      "num_input_tokens_seen": 64414904,
+      "step": 437
+    },
+    {
+      "epoch": 0.12626117036609974,
+      "grad_norm": 6.2712386779629785,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "num_input_tokens_seen": 64587336,
+      "step": 438
+    },
+    {
+      "epoch": 0.12626117036609974,
+      "loss": 0.0074829827062785625,
+      "loss_ce": 0.0033840904943645,
+      "loss_xval": 0.00408935546875,
+      "num_input_tokens_seen": 64587336,
+      "step": 438
+    },
+    {
+      "epoch": 0.1265494378783511,
+      "grad_norm": 10.047732594806881,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "num_input_tokens_seen": 64722136,
+      "step": 439
+    },
+    {
+      "epoch": 0.1265494378783511,
+      "loss": 0.02131899818778038,
+      "loss_ce": 0.01652010902762413,
+      "loss_xval": 0.004791259765625,
+      "num_input_tokens_seen": 64722136,
+      "step": 439
+    },
+    {
+      "epoch": 0.1268377053906025,
+      "grad_norm": 0.806131296148929,
+      "learning_rate": 0.0001,
+      "loss": 0.0055,
+      "num_input_tokens_seen": 64857232,
+      "step": 440
+    },
+    {
+      "epoch": 0.1268377053906025,
+      "loss": 0.005497447215020657,
+      "loss_ce": 0.002595416735857725,
+      "loss_xval": 0.002899169921875,
+      "num_input_tokens_seen": 64857232,
+      "step": 440
+    },
+    {
+      "epoch": 0.12712597290285385,
+      "grad_norm": 9.806963011163788,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "num_input_tokens_seen": 65029712,
+      "step": 441
+    },
+    {
+      "epoch": 0.12712597290285385,
+      "loss": 0.00795245636254549,
+      "loss_ce": 0.0030295890755951405,
+      "loss_xval": 0.004913330078125,
+      "num_input_tokens_seen": 65029712,
+      "step": 441
+    },
+    {
+      "epoch": 0.1274142404151052,
+      "grad_norm": 6.870364954781495,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "num_input_tokens_seen": 65164528,
+      "step": 442
+    },
+    {
+      "epoch": 0.1274142404151052,
+      "loss": 0.02025909721851349,
+      "loss_ce": 0.01654548943042755,
+      "loss_xval": 0.0037078857421875,
+      "num_input_tokens_seen": 65164528,
+      "step": 442
+    },
+    {
+      "epoch": 0.12770250792735657,
+      "grad_norm": 5.060108314966909,
+      "learning_rate": 0.0001,
+      "loss": 0.0061,
+      "num_input_tokens_seen": 65299784,
+      "step": 443
+    },
+    {
+      "epoch": 0.12770250792735657,
+      "loss": 0.005887602921575308,
+      "loss_ce": 0.0022655476350337267,
+      "loss_xval": 0.0036163330078125,
+      "num_input_tokens_seen": 65299784,
+      "step": 443
+    },
+    {
+      "epoch": 0.12799077543960796,
+      "grad_norm": 9.71208773417023,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "num_input_tokens_seen": 65472424,
+      "step": 444
+    },
+    {
+      "epoch": 0.12799077543960796,
+      "loss": 0.007333572953939438,
+      "loss_ce": 0.002883728127926588,
+      "loss_xval": 0.00445556640625,
+      "num_input_tokens_seen": 65472424,
+      "step": 444
+    },
+    {
+      "epoch": 0.12827904295185932,
+      "grad_norm": 0.5369327335786107,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "num_input_tokens_seen": 65607192,
+      "step": 445
+    },
+    {
+      "epoch": 0.12827904295185932,
+      "loss": 0.01974431797862053,
+      "loss_ce": 0.016584794968366623,
+      "loss_xval": 0.0031585693359375,
+      "num_input_tokens_seen": 65607192,
+      "step": 445
+    },
+    {
+      "epoch": 0.12856731046411068,
+      "grad_norm": 11.725794535136131,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "num_input_tokens_seen": 65742360,
+      "step": 446
+    },
+    {
+      "epoch": 0.12856731046411068,
+      "loss": 0.008027734234929085,
+      "loss_ce": 0.0024067778140306473,
+      "loss_xval": 0.005615234375,
+      "num_input_tokens_seen": 65742360,
+      "step": 446
+    },
+    {
+      "epoch": 0.12885557797636207,
+      "grad_norm": 8.498973502599162,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "num_input_tokens_seen": 65914968,
+      "step": 447
+    },
+    {
+      "epoch": 0.12885557797636207,
+      "loss": 0.006696837022900581,
+      "loss_ce": 0.003360884264111519,
+      "loss_xval": 0.0033416748046875,
+      "num_input_tokens_seen": 65914968,
+      "step": 447
+    },
+    {
+      "epoch": 0.12914384548861343,
+      "grad_norm": 7.918505058803784,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "num_input_tokens_seen": 66049832,
+      "step": 448
+    },
+    {
+      "epoch": 0.12914384548861343,
+      "loss": 0.023593148216605186,
+      "loss_ce": 0.01844712160527706,
+      "loss_xval": 0.005157470703125,
+      "num_input_tokens_seen": 66049832,
+      "step": 448
+    },
+    {
+      "epoch": 0.1294321130008648,
+      "grad_norm": 15.69806868410208,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "num_input_tokens_seen": 66184864,
+      "step": 449
+    },
+    {
+      "epoch": 0.1294321130008648,
+      "loss": 0.009837846271693707,
+      "loss_ce": 0.0018842025892809033,
+      "loss_xval": 0.0079345703125,
+      "num_input_tokens_seen": 66184864,
+      "step": 449
+    },
+    {
+      "epoch": 0.12972038051311618,
+      "grad_norm": 3.43071398992182,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "num_input_tokens_seen": 66357288,
+      "step": 450
+    },
+    {
+      "epoch": 0.12972038051311618,
+      "loss": 0.0050545441918075085,
+      "loss_ce": 0.0021372544579207897,
+      "loss_xval": 0.0029144287109375,
+      "num_input_tokens_seen": 66357288,
+      "step": 450
+    },
+    {
+      "epoch": 0.13000864802536755,
+      "grad_norm": 13.550465489619242,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "num_input_tokens_seen": 66492088,
+      "step": 451
+    },
+    {
+      "epoch": 0.13000864802536755,
+      "loss": 0.02118588052690029,
+      "loss_ce": 0.015063291415572166,
+      "loss_xval": 0.006134033203125,
+      "num_input_tokens_seen": 66492088,
+      "step": 451
+    },
+    {
+      "epoch": 0.1302969155376189,
+      "grad_norm": 13.399452499456793,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "num_input_tokens_seen": 66627072,
+      "step": 452
+    },
+    {
+      "epoch": 0.1302969155376189,
+      "loss": 0.009072705172002316,
+      "loss_ce": 0.002290173899382353,
+      "loss_xval": 0.00677490234375,
+      "num_input_tokens_seen": 66627072,
+      "step": 452
+    },
+    {
+      "epoch": 0.13058518304987027,
+      "grad_norm": 4.240862155234506,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "num_input_tokens_seen": 66799488,
+      "step": 453
+    },
+    {
+      "epoch": 0.13058518304987027,
+      "loss": 0.004820866510272026,
+      "loss_ce": 0.0020142027642577887,
+      "loss_xval": 0.0028076171875,
+      "num_input_tokens_seen": 66799488,
+      "step": 453
+    },
+    {
+      "epoch": 0.13087345056212166,
+      "grad_norm": 16.168784353206163,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "num_input_tokens_seen": 66934336,
+      "step": 454
+    },
+    {
+      "epoch": 0.13087345056212166,
+      "loss": 0.024435311555862427,
+      "loss_ce": 0.016409188508987427,
+      "loss_xval": 0.008056640625,
+      "num_input_tokens_seen": 66934336,
+      "step": 454
+    },
+    {
+      "epoch": 0.13116171807437302,
+      "grad_norm": 3.851014867434399,
+      "learning_rate": 0.0001,
+      "loss": 0.0044,
+      "num_input_tokens_seen": 67069320,
+      "step": 455
+    },
+    {
+      "epoch": 0.13116171807437302,
+      "loss": 0.004337950609624386,
+      "loss_ce": 0.0016276079695671797,
+      "loss_xval": 0.002716064453125,
+      "num_input_tokens_seen": 67069320,
+      "step": 455
+    },
+    {
+      "epoch": 0.13144998558662438,
+      "grad_norm": 18.412053523066213,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "num_input_tokens_seen": 67241768,
+      "step": 456
+    },
+    {
+      "epoch": 0.13144998558662438,
+      "loss": 0.01089160144329071,
+      "loss_ce": 0.0026671146042644978,
+      "loss_xval": 0.00823974609375,
+      "num_input_tokens_seen": 67241768,
+      "step": 456
+    },
+    {
+      "epoch": 0.13173825309887577,
+      "grad_norm": 19.223558233035803,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "num_input_tokens_seen": 67376624,
+      "step": 457
+    },
+    {
+      "epoch": 0.13173825309887577,
+      "loss": 0.02506151795387268,
+      "loss_ce": 0.015362650156021118,
+      "loss_xval": 0.00970458984375,
+      "num_input_tokens_seen": 67376624,
+      "step": 457
+    },
+    {
+      "epoch": 0.13202652061112713,
+      "grad_norm": 9.699651714957488,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "num_input_tokens_seen": 67511632,
+      "step": 458
+    },
+    {
+      "epoch": 0.13202652061112713,
+      "loss": 0.005822490900754929,
+      "loss_ce": 0.001746486988849938,
+      "loss_xval": 0.00408935546875,
+      "num_input_tokens_seen": 67511632,
+      "step": 458
+    },
+    {
+      "epoch": 0.1323147881233785,
+      "grad_norm": 34.073273051818276,
+      "learning_rate": 0.0001,
+      "loss": 0.0445,
+      "num_input_tokens_seen": 67684128,
+      "step": 459
+    },
+    {
+      "epoch": 0.1323147881233785,
+      "loss": 0.026470568031072617,
+      "loss_ce": 0.002163316821679473,
+      "loss_xval": 0.0242919921875,
+      "num_input_tokens_seen": 67684128,
+      "step": 459
+    },
+    {
+      "epoch": 0.13260305563562985,
+      "grad_norm": 15.624662552836503,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "num_input_tokens_seen": 67818912,
+      "step": 460
+    },
+    {
+      "epoch": 0.13260305563562985,
+      "loss": 0.02330195903778076,
+      "loss_ce": 0.015149950981140137,
+      "loss_xval": 0.0081787109375,
+      "num_input_tokens_seen": 67818912,
+      "step": 460
+    },
+    {
+      "epoch": 0.13289132314788124,
+      "grad_norm": 31.190077918049116,
+      "learning_rate": 0.0001,
+      "loss": 0.0311,
+      "num_input_tokens_seen": 67954008,
+      "step": 461
+    },
+    {
+      "epoch": 0.13289132314788124,
+      "loss": 0.0190433319658041,
+      "loss_ce": 0.0015491305384784937,
+      "loss_xval": 0.0174560546875,
+      "num_input_tokens_seen": 67954008,
+      "step": 461
+    },
+    {
+      "epoch": 0.1331795906601326,
+      "grad_norm": 47.20861973030621,
+      "learning_rate": 0.0001,
+      "loss": 0.0742,
+      "num_input_tokens_seen": 68126440,
+      "step": 462
+    },
+    {
+      "epoch": 0.1331795906601326,
+      "loss": 0.04881232604384422,
+      "loss_ce": 0.0018457742407917976,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 68126440,
+      "step": 462
+    },
+    {
+      "epoch": 0.13346785817238396,
+      "grad_norm": 2.6336244187033753,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "num_input_tokens_seen": 68261192,
+      "step": 463
+    },
+    {
+      "epoch": 0.13346785817238396,
+      "loss": 0.018522823229432106,
+      "loss_ce": 0.014969432726502419,
+      "loss_xval": 0.0035552978515625,
+      "num_input_tokens_seen": 68261192,
+      "step": 463
+    },
+    {
+      "epoch": 0.13375612568463535,
+      "grad_norm": 57.78556207273457,
+      "learning_rate": 0.0001,
+      "loss": 0.0957,
+      "num_input_tokens_seen": 68396264,
+      "step": 464
+    },
+    {
+      "epoch": 0.13375612568463535,
+      "loss": 0.06289799511432648,
+      "loss_ce": 0.001466113142669201,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 68396264,
+      "step": 464
+    },
+    {
+      "epoch": 0.1340443931968867,
+      "grad_norm": 58.46829235355628,
+      "learning_rate": 0.0001,
+      "loss": 0.1064,
+      "num_input_tokens_seen": 68568856,
+      "step": 465
+    },
+    {
+      "epoch": 0.1340443931968867,
+      "loss": 0.08666277676820755,
+      "loss_ce": 0.002129080705344677,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 68568856,
+      "step": 465
+    },
+    {
+      "epoch": 0.13433266070913807,
+      "grad_norm": 12.440288342385827,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "num_input_tokens_seen": 68703656,
+      "step": 466
+    },
+    {
+      "epoch": 0.13433266070913807,
+      "loss": 0.020905818790197372,
+      "loss_ce": 0.015027369372546673,
+      "loss_xval": 0.005889892578125,
+      "num_input_tokens_seen": 68703656,
+      "step": 466
+    },
+    {
+      "epoch": 0.13462092822138946,
+      "grad_norm": 77.7980371215157,
+      "learning_rate": 0.0001,
+      "loss": 0.1704,
+      "num_input_tokens_seen": 68838648,
+      "step": 467
+    },
+    {
+      "epoch": 0.13462092822138946,
+      "loss": 0.11293500661849976,
+      "loss_ce": 0.001667913980782032,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 68838648,
+      "step": 467
+    },
+    {
+      "epoch": 0.13490919573364082,
+      "grad_norm": 55.77808662197712,
+      "learning_rate": 0.0001,
+      "loss": 0.0969,
+      "num_input_tokens_seen": 69011080,
+      "step": 468
+    },
+    {
+      "epoch": 0.13490919573364082,
+      "loss": 0.0929902121424675,
+      "loss_ce": 0.0019257587846368551,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 69011080,
+      "step": 468
+    },
+    {
+      "epoch": 0.13519746324589219,
+      "grad_norm": 36.90146649165042,
+      "learning_rate": 0.0001,
+      "loss": 0.0613,
+      "num_input_tokens_seen": 69145880,
+      "step": 469
+    },
+    {
+      "epoch": 0.13519746324589219,
+      "loss": 0.03158336132764816,
+      "loss_ce": 0.016259726136922836,
+      "loss_xval": 0.01531982421875,
+      "num_input_tokens_seen": 69145880,
+      "step": 469
+    },
+    {
+      "epoch": 0.13548573075814355,
+      "grad_norm": 92.9330163112296,
+      "learning_rate": 0.0001,
+      "loss": 0.2396,
+      "num_input_tokens_seen": 69280872,
+      "step": 470
+    },
+    {
+      "epoch": 0.13548573075814355,
+      "loss": 0.17209458351135254,
+      "loss_ce": 0.0018064901232719421,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 69280872,
+      "step": 470
+    },
+    {
+      "epoch": 0.13577399827039494,
+      "grad_norm": 35.17731963697524,
+      "learning_rate": 0.0001,
+      "loss": 0.052,
+      "num_input_tokens_seen": 69453408,
+      "step": 471
+    },
+    {
+      "epoch": 0.13577399827039494,
+      "loss": 0.06354454159736633,
+      "loss_ce": 0.0019295515958219767,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 69453408,
+      "step": 471
+    },
+    {
+      "epoch": 0.1360622657826463,
+      "grad_norm": 78.21099175332745,
+      "learning_rate": 0.0001,
+      "loss": 0.1981,
+      "num_input_tokens_seen": 69588224,
+      "step": 472
+    },
+    {
+      "epoch": 0.1360622657826463,
+      "loss": 0.10047487169504166,
+      "loss_ce": 0.017772231251001358,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 69588224,
+      "step": 472
+    },
+    {
+      "epoch": 0.13635053329489766,
+      "grad_norm": 109.31947715057484,
+      "learning_rate": 0.0001,
+      "loss": 0.3297,
+      "num_input_tokens_seen": 69723480,
+      "step": 473
+    },
+    {
+      "epoch": 0.13635053329489766,
+      "loss": 0.27944549918174744,
+      "loss_ce": 0.0019796781707555056,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 69723480,
+      "step": 473
+    },
+    {
+      "epoch": 0.13663880080714905,
+      "grad_norm": 4.425859631788829,
+      "learning_rate": 0.0001,
+      "loss": 0.0315,
+      "num_input_tokens_seen": 69895896,
+      "step": 474
+    },
+    {
+      "epoch": 0.13663880080714905,
+      "loss": 0.024713829159736633,
+      "loss_ce": 0.0023749619722366333,
+      "loss_xval": 0.0223388671875,
+      "num_input_tokens_seen": 69895896,
+      "step": 474
+    },
+    {
+      "epoch": 0.1369270683194004,
+      "grad_norm": 121.91800725272553,
+      "learning_rate": 0.0001,
+      "loss": 0.4467,
+      "num_input_tokens_seen": 70030768,
+      "step": 475
+    },
+    {
+      "epoch": 0.1369270683194004,
+      "loss": 0.25983697175979614,
+      "loss_ce": 0.01789361983537674,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 70030768,
+      "step": 475
+    },
+    {
+      "epoch": 0.13721533583165177,
+      "grad_norm": 102.73057111430288,
+      "learning_rate": 0.0001,
+      "loss": 0.2963,
+      "num_input_tokens_seen": 70165784,
+      "step": 476
+    },
+    {
+      "epoch": 0.13721533583165177,
+      "loss": 0.3070387840270996,
+      "loss_ce": 0.002839577617123723,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 70165784,
+      "step": 476
+    },
+    {
+      "epoch": 0.13750360334390313,
+      "grad_norm": 54.2162435395872,
+      "learning_rate": 0.0001,
+      "loss": 0.126,
+      "num_input_tokens_seen": 70338232,
+      "step": 477
+    },
+    {
+      "epoch": 0.13750360334390313,
+      "loss": 0.022026684135198593,
+      "loss_ce": 0.004997876472771168,
+      "loss_xval": 0.01708984375,
+      "num_input_tokens_seen": 70338232,
+      "step": 477
+    },
+    {
+      "epoch": 0.13779187085615452,
+      "grad_norm": 154.58136133365142,
+      "learning_rate": 0.0001,
+      "loss": 0.6935,
+      "num_input_tokens_seen": 70472952,
+      "step": 478
+    },
+    {
+      "epoch": 0.13779187085615452,
+      "loss": 0.506955623626709,
+      "loss_ce": 0.01940680667757988,
+      "loss_xval": 0.48828125,
+      "num_input_tokens_seen": 70472952,
+      "step": 478
+    },
+    {
+      "epoch": 0.13808013836840588,
+      "grad_norm": 48.70800586926015,
+      "learning_rate": 0.0001,
+      "loss": 0.0818,
+      "num_input_tokens_seen": 70607984,
+      "step": 479
+    },
+    {
+      "epoch": 0.13808013836840588,
+      "loss": 0.12578168511390686,
+      "loss_ce": 0.003101026639342308,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 70607984,
+      "step": 479
+    },
+    {
+      "epoch": 0.13836840588065724,
+      "grad_norm": 143.0906218557431,
+      "learning_rate": 0.0001,
+      "loss": 0.643,
+      "num_input_tokens_seen": 70780496,
+      "step": 480
+    },
+    {
+      "epoch": 0.13836840588065724,
+      "loss": 0.3119392395019531,
+      "loss_ce": 0.004077905789017677,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 70780496,
+      "step": 480
+    },
+    {
+      "epoch": 0.13865667339290863,
+      "grad_norm": 157.58702639731285,
+      "learning_rate": 0.0001,
+      "loss": 0.7503,
+      "num_input_tokens_seen": 70915288,
+      "step": 481
+    },
+    {
+      "epoch": 0.13865667339290863,
+      "loss": 0.6724344491958618,
+      "loss_ce": 0.0200907364487648,
+      "loss_xval": 0.65234375,
+      "num_input_tokens_seen": 70915288,
+      "step": 481
+    },
+    {
+      "epoch": 0.13894494090516,
+      "grad_norm": 49.559912277556116,
+      "learning_rate": 0.0001,
+      "loss": 0.1068,
+      "num_input_tokens_seen": 71050288,
+      "step": 482
+    },
+    {
+      "epoch": 0.13894494090516,
+      "loss": 0.024329371750354767,
+      "loss_ce": 0.0039360010996460915,
+      "loss_xval": 0.0203857421875,
+      "num_input_tokens_seen": 71050288,
+      "step": 482
+    },
+    {
+      "epoch": 0.13923320841741135,
+      "grad_norm": 207.90175767039054,
+      "learning_rate": 0.0001,
+      "loss": 1.3603,
+      "num_input_tokens_seen": 71222816,
+      "step": 483
+    },
+    {
+      "epoch": 0.13923320841741135,
+      "loss": 0.9580897092819214,
+      "loss_ce": 0.004964680410921574,
+      "loss_xval": 0.953125,
+      "num_input_tokens_seen": 71222816,
+      "step": 483
+    },
+    {
+      "epoch": 0.1395214759296627,
+      "grad_norm": 68.2165654788575,
+      "learning_rate": 0.0001,
+      "loss": 0.1706,
+      "num_input_tokens_seen": 71357624,
+      "step": 484
+    },
+    {
+      "epoch": 0.1395214759296627,
+      "loss": 0.18781864643096924,
+      "loss_ce": 0.01801883429288864,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 71357624,
+      "step": 484
+    },
+    {
+      "epoch": 0.1398097434419141,
+      "grad_norm": 194.29247828283172,
+      "learning_rate": 0.0001,
+      "loss": 1.242,
+      "num_input_tokens_seen": 71492848,
+      "step": 485
+    },
+    {
+      "epoch": 0.1398097434419141,
+      "loss": 0.8029749989509583,
+      "loss_ce": 0.0046351575292646885,
+      "loss_xval": 0.796875,
+      "num_input_tokens_seen": 71492848,
+      "step": 485
+    },
+    {
+      "epoch": 0.14009801095416546,
+      "grad_norm": 186.8894329562773,
+      "learning_rate": 0.0001,
+      "loss": 1.195,
+      "num_input_tokens_seen": 71665440,
+      "step": 486
+    },
+    {
+      "epoch": 0.14009801095416546,
+      "loss": 1.040757179260254,
+      "loss_ce": 0.009507151320576668,
+      "loss_xval": 1.03125,
+      "num_input_tokens_seen": 71665440,
+      "step": 486
+    },
+    {
+      "epoch": 0.14038627846641683,
+      "grad_norm": 97.3981341751347,
+      "learning_rate": 0.0001,
+      "loss": 0.365,
+      "num_input_tokens_seen": 71800264,
+      "step": 487
+    },
+    {
+      "epoch": 0.14038627846641683,
+      "loss": 0.25503528118133545,
+      "loss_ce": 0.02139267325401306,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 71800264,
+      "step": 487
+    },
+    {
+      "epoch": 0.14067454597866821,
+      "grad_norm": 240.16409642375302,
+      "learning_rate": 0.0001,
+      "loss": 1.9637,
+      "num_input_tokens_seen": 71935360,
+      "step": 488
+    },
+    {
+      "epoch": 0.14067454597866821,
+      "loss": 1.6441001892089844,
+      "loss_ce": 0.007381412200629711,
+      "loss_xval": 1.640625,
+      "num_input_tokens_seen": 71935360,
+      "step": 488
+    },
+    {
+      "epoch": 0.14096281349091958,
+      "grad_norm": 3.504384559163347,
+      "learning_rate": 0.0001,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 72107824,
+      "step": 489
+    },
+    {
+      "epoch": 0.14096281349091958,
+      "loss": 0.027433183044195175,
+      "loss_ce": 0.005040910094976425,
+      "loss_xval": 0.0223388671875,
+      "num_input_tokens_seen": 72107824,
+      "step": 489
+    },
+    {
+      "epoch": 0.14125108100317094,
+      "grad_norm": 251.4256804889384,
+      "learning_rate": 0.0001,
+      "loss": 1.9899,
+      "num_input_tokens_seen": 72242616,
+      "step": 490
+    },
+    {
+      "epoch": 0.14125108100317094,
+      "loss": 1.516852617263794,
+      "loss_ce": 0.016852546483278275,
+      "loss_xval": 1.5,
+      "num_input_tokens_seen": 72242616,
+      "step": 490
+    },
+    {
+      "epoch": 0.14153934851542233,
+      "grad_norm": 120.34478539836553,
+      "learning_rate": 0.0001,
+      "loss": 0.4263,
+      "num_input_tokens_seen": 72377696,
+      "step": 491
+    },
+    {
+      "epoch": 0.14153934851542233,
+      "loss": 0.3436959981918335,
+      "loss_ce": 0.008979195728898048,
+      "loss_xval": 0.333984375,
+      "num_input_tokens_seen": 72377696,
+      "step": 491
+    },
+    {
+      "epoch": 0.1418276160276737,
+      "grad_norm": 191.44123174477102,
+      "learning_rate": 0.0001,
+      "loss": 1.1365,
+      "num_input_tokens_seen": 72550192,
+      "step": 492
+    },
+    {
+      "epoch": 0.1418276160276737,
+      "loss": 0.9527350068092346,
+      "loss_ce": 0.006445930339396,
+      "loss_xval": 0.9453125,
+      "num_input_tokens_seen": 72550192,
+      "step": 492
+    },
+    {
+      "epoch": 0.14211588353992505,
+      "grad_norm": 142.707840044168,
+      "learning_rate": 0.0001,
+      "loss": 0.8139,
+      "num_input_tokens_seen": 72685120,
+      "step": 493
+    },
+    {
+      "epoch": 0.14211588353992505,
+      "loss": 0.5713205933570862,
+      "loss_ce": 0.031037403270602226,
+      "loss_xval": 0.5390625,
+      "num_input_tokens_seen": 72685120,
+      "step": 493
+    },
+    {
+      "epoch": 0.1424041510521764,
+      "grad_norm": 83.07267900422377,
+      "learning_rate": 0.0001,
+      "loss": 0.3097,
+      "num_input_tokens_seen": 72820296,
+      "step": 494
+    },
+    {
+      "epoch": 0.1424041510521764,
+      "loss": 0.1652831882238388,
+      "loss_ce": 0.009643547236919403,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 72820296,
+      "step": 494
+    },
+    {
+      "epoch": 0.1426924185644278,
+      "grad_norm": 1183.2500602177688,
+      "learning_rate": 0.0001,
+      "loss": 16.1702,
+      "num_input_tokens_seen": 72992760,
+      "step": 495
+    },
+    {
+      "epoch": 0.1426924185644278,
+      "loss": 7.769603729248047,
+      "loss_ce": 0.03132236748933792,
+      "loss_xval": 7.75,
+      "num_input_tokens_seen": 72992760,
+      "step": 495
+    },
+    {
+      "epoch": 0.14298068607667916,
+      "grad_norm": 468.23328751112655,
+      "learning_rate": 0.0001,
+      "loss": 2.9613,
+      "num_input_tokens_seen": 73127528,
+      "step": 496
+    },
+    {
+      "epoch": 0.14298068607667916,
+      "loss": 2.834989547729492,
+      "loss_ce": 0.2217084765434265,
+      "loss_xval": 2.609375,
+      "num_input_tokens_seen": 73127528,
+      "step": 496
+    },
+    {
+      "epoch": 0.14326895358893052,
+      "grad_norm": 3547.408541295576,
+      "learning_rate": 0.0001,
+      "loss": 63.6958,
+      "num_input_tokens_seen": 73262672,
+      "step": 497
+    },
+    {
+      "epoch": 0.14326895358893052,
+      "loss": 68.32894134521484,
+      "loss_ce": 0.5789394378662109,
+      "loss_xval": 68.0,
+      "num_input_tokens_seen": 73262672,
+      "step": 497
+    },
+    {
+      "epoch": 0.1435572211011819,
+      "grad_norm": 2957.9835631037604,
+      "learning_rate": 0.0001,
+      "loss": 88.0121,
+      "num_input_tokens_seen": 73435112,
+      "step": 498
+    },
+    {
+      "epoch": 0.1435572211011819,
+      "loss": 90.78068542480469,
+      "loss_ce": 1.7181835174560547,
+      "loss_xval": 89.0,
+      "num_input_tokens_seen": 73435112,
+      "step": 498
+    },
+    {
+      "epoch": 0.14384548861343327,
+      "grad_norm": 144.12515736515584,
+      "learning_rate": 0.0001,
+      "loss": 3.3137,
+      "num_input_tokens_seen": 73569888,
+      "step": 499
+    },
+    {
+      "epoch": 0.14384548861343327,
+      "loss": 3.01613450050354,
+      "loss_ce": 2.08986496925354,
+      "loss_xval": 0.92578125,
+      "num_input_tokens_seen": 73569888,
+      "step": 499
+    },
+    {
+      "epoch": 0.14413375612568463,
+      "grad_norm": 2182.1687365844186,
+      "learning_rate": 0.0001,
+      "loss": 42.8611,
+      "num_input_tokens_seen": 73704904,
+      "step": 500
+    },
+    {
+      "epoch": 0.14413375612568463,
+      "eval_websight_new_IoU": 0.0,
+      "eval_websight_new_MAE_x": 16.08986473083496,
+      "eval_websight_new_MAE_y": 14.622143268585205,
+      "eval_websight_new_NUM_probability": 0.00971380015835166,
+      "eval_websight_new_inside_bbox": 0.0,
+      "eval_websight_new_loss": 261.8992614746094,
+      "eval_websight_new_loss_ce": 1.582156479358673,
+      "eval_websight_new_loss_xval": 260.5,
+      "eval_websight_new_runtime": 36.8782,
+      "eval_websight_new_samples_per_second": 1.356,
+      "eval_websight_new_steps_per_second": 0.054,
+      "num_input_tokens_seen": 73704904,
+      "step": 500
+    },
+    {
+      "epoch": 0.14413375612568463,
+      "eval_seeclick_IoU": 0.0,
+      "eval_seeclick_MAE_x": 15.54664421081543,
+      "eval_seeclick_MAE_y": 14.642383098602295,
+      "eval_seeclick_NUM_probability": 0.02366485446691513,
+      "eval_seeclick_inside_bbox": 0.0,
+      "eval_seeclick_loss": 251.6716766357422,
+      "eval_seeclick_loss_ce": 1.5070286989212036,
+      "eval_seeclick_loss_xval": 248.625,
+      "eval_seeclick_runtime": 66.5403,
+      "eval_seeclick_samples_per_second": 0.751,
+      "eval_seeclick_steps_per_second": 0.03,
+      "num_input_tokens_seen": 73704904,
+      "step": 500
+    },
+    {
+      "epoch": 0.14413375612568463,
+      "eval_icons_IoU": 0.0,
+      "eval_icons_MAE_x": 16.21195411682129,
+      "eval_icons_MAE_y": 14.966073036193848,
+      "eval_icons_NUM_probability": 0.017178870271891356,
+      "eval_icons_inside_bbox": 0.0,
+      "eval_icons_loss": 252.5450439453125,
+      "eval_icons_loss_ce": 1.3136942386627197,
+      "eval_icons_loss_xval": 250.5625,
+      "eval_icons_runtime": 63.7918,
+      "eval_icons_samples_per_second": 0.784,
+      "eval_icons_steps_per_second": 0.031,
+      "num_input_tokens_seen": 73704904,
+      "step": 500
+    },
+    {
+      "epoch": 0.14413375612568463,
+      "loss": 254.4305419921875,
+      "loss_ce": 1.3055286407470703,
+      "loss_xval": 253.0,
+      "num_input_tokens_seen": 73704904,
+      "step": 500
+    },
+    {
+      "epoch": 0.144422023637936,
+      "grad_norm": 5415.594835990045,
+      "learning_rate": 0.0001,
+      "loss": 249.5239,
+      "num_input_tokens_seen": 73877312,
+      "step": 501
+    },
+    {
+      "epoch": 0.144422023637936,
+      "loss": 248.81724548339844,
+      "loss_ce": 1.3172447681427002,
+      "loss_xval": 248.0,
+      "num_input_tokens_seen": 73877312,
+      "step": 501
+    },
+    {
+      "epoch": 0.14471029115018738,
+      "grad_norm": 437.05624804314454,
+      "learning_rate": 0.0001,
+      "loss": 5.4789,
+      "num_input_tokens_seen": 74012064,
+      "step": 502
+    },
+    {
+      "epoch": 0.14471029115018738,
+      "loss": 6.111766815185547,
+      "loss_ce": 0.8031734228134155,
+      "loss_xval": 5.3125,
+      "num_input_tokens_seen": 74012064,
+      "step": 502
+    },
+    {
+      "epoch": 0.14499855866243874,
+      "grad_norm": 6017.803408310277,
+      "learning_rate": 0.0001,
+      "loss": 35.1959,
+      "num_input_tokens_seen": 74147200,
+      "step": 503
+    },
+    {
+      "epoch": 0.14499855866243874,
+      "loss": 24.930065155029297,
+      "loss_ce": 0.7581892609596252,
+      "loss_xval": 24.125,
+      "num_input_tokens_seen": 74147200,
+      "step": 503
+    },
+    {
+      "epoch": 0.1452868261746901,
+      "grad_norm": 4925.158357501441,
+      "learning_rate": 0.0001,
+      "loss": 231.4661,
+      "num_input_tokens_seen": 74319696,
+      "step": 504
+    },
+    {
+      "epoch": 0.1452868261746901,
+      "loss": 234.67117309570312,
+      "loss_ce": 2.17117977142334,
+      "loss_xval": 232.0,
+      "num_input_tokens_seen": 74319696,
+      "step": 504
+    },
+    {
+      "epoch": 0.1455750936869415,
+      "grad_norm": 6083.288980314157,
+      "learning_rate": 0.0001,
+      "loss": 33.7177,
+      "num_input_tokens_seen": 74454544,
+      "step": 505
+    },
+    {
+      "epoch": 0.1455750936869415,
+      "loss": 40.13539505004883,
+      "loss_ce": 1.9478960037231445,
+      "loss_xval": 38.25,
+      "num_input_tokens_seen": 74454544,
+      "step": 505
+    },
+    {
+      "epoch": 0.14586336119919285,
+      "grad_norm": 1590.7858721635075,
+      "learning_rate": 0.0001,
+      "loss": 35.5616,
+      "num_input_tokens_seen": 74589784,
+      "step": 506
+    },
+    {
+      "epoch": 0.14586336119919285,
+      "loss": 31.166776657104492,
+      "loss_ce": 2.5730271339416504,
+      "loss_xval": 28.625,
+      "num_input_tokens_seen": 74589784,
+      "step": 506
+    },
+    {
+      "epoch": 0.14615162871144421,
+      "grad_norm": 386.09725857908813,
+      "learning_rate": 0.0001,
+      "loss": 5.0356,
+      "num_input_tokens_seen": 74762248,
+      "step": 507
+    },
+    {
+      "epoch": 0.14615162871144421,
+      "loss": 4.796041488647461,
+      "loss_ce": 3.13100266456604,
+      "loss_xval": 1.6640625,
+      "num_input_tokens_seen": 74762248,
+      "step": 507
+    },
+    {
+      "epoch": 0.14643989622369558,
+      "grad_norm": 948.5710892886063,
+      "learning_rate": 0.0001,
+      "loss": 19.8862,
+      "num_input_tokens_seen": 74897016,
+      "step": 508
+    },
+    {
+      "epoch": 0.14643989622369558,
+      "loss": 19.3146915435791,
+      "loss_ce": 2.8303165435791016,
+      "loss_xval": 16.5,
+      "num_input_tokens_seen": 74897016,
+      "step": 508
+    },
+    {
+      "epoch": 0.14672816373594696,
+      "grad_norm": 748.4834585091971,
+      "learning_rate": 0.0001,
+      "loss": 5.8444,
+      "num_input_tokens_seen": 75032128,
+      "step": 509
+    },
+    {
+      "epoch": 0.14672816373594696,
+      "loss": 6.053933143615723,
+      "loss_ce": 2.7746365070343018,
+      "loss_xval": 3.28125,
+      "num_input_tokens_seen": 75032128,
+      "step": 509
+    },
+    {
+      "epoch": 0.14701643124819833,
+      "grad_norm": 676.2903234596315,
+      "learning_rate": 0.0001,
+      "loss": 5.9856,
+      "num_input_tokens_seen": 75204696,
+      "step": 510
+    },
+    {
+      "epoch": 0.14701643124819833,
+      "loss": 5.366189956665039,
+      "loss_ce": 3.422830104827881,
+      "loss_xval": 1.9453125,
+      "num_input_tokens_seen": 75204696,
+      "step": 510
+    },
+    {
+      "epoch": 0.1473046987604497,
+      "grad_norm": 894.8470585288454,
+      "learning_rate": 0.0001,
+      "loss": 9.989,
+      "num_input_tokens_seen": 75339472,
+      "step": 511
+    },
+    {
+      "epoch": 0.1473046987604497,
+      "loss": 10.071243286132812,
+      "loss_ce": 2.645461320877075,
+      "loss_xval": 7.4375,
+      "num_input_tokens_seen": 75339472,
+      "step": 511
+    },
+    {
+      "epoch": 0.14759296627270108,
+      "grad_norm": 1193.6287681118333,
+      "learning_rate": 0.0001,
+      "loss": 2.2828,
+      "num_input_tokens_seen": 75474672,
+      "step": 512
+    },
+    {
+      "epoch": 0.14759296627270108,
+      "loss": 2.0058786869049072,
+      "loss_ce": 1.4199411869049072,
+      "loss_xval": 0.5859375,
+      "num_input_tokens_seen": 75474672,
+      "step": 512
+    },
+    {
+      "epoch": 0.14788123378495244,
+      "grad_norm": 220.89561497284404,
+      "learning_rate": 0.0001,
+      "loss": 2.2091,
+      "num_input_tokens_seen": 75647152,
+      "step": 513
+    },
+    {
+      "epoch": 0.14788123378495244,
+      "loss": 2.175510883331299,
+      "loss_ce": 1.0495343208312988,
+      "loss_xval": 1.125,
+      "num_input_tokens_seen": 75647152,
+      "step": 513
+    },
+    {
+      "epoch": 0.1481695012972038,
+      "grad_norm": 245.22955417098655,
+      "learning_rate": 0.0001,
+      "loss": 2.1856,
+      "num_input_tokens_seen": 75781872,
+      "step": 514
+    },
+    {
+      "epoch": 0.1481695012972038,
+      "loss": 2.1912238597869873,
+      "loss_ce": 0.5808722972869873,
+      "loss_xval": 1.609375,
+      "num_input_tokens_seen": 75781872,
+      "step": 514
+    },
+    {
+      "epoch": 0.1484577688094552,
+      "grad_norm": 59.04888675936134,
+      "learning_rate": 0.0001,
+      "loss": 1.0063,
+      "num_input_tokens_seen": 75916936,
+      "step": 515
+    },
+    {
+      "epoch": 0.1484577688094552,
+      "loss": 0.8350945711135864,
+      "loss_ce": 0.5239373445510864,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 75916936,
+      "step": 515
+    },
+    {
+      "epoch": 0.14874603632170655,
+      "grad_norm": 240.02413750733743,
+      "learning_rate": 0.0001,
+      "loss": 1.9709,
+      "num_input_tokens_seen": 76089512,
+      "step": 516
+    },
+    {
+      "epoch": 0.14874603632170655,
+      "loss": 1.302797794342041,
+      "loss_ce": 0.599184513092041,
+      "loss_xval": 0.703125,
+      "num_input_tokens_seen": 76089512,
+      "step": 516
+    },
+    {
+      "epoch": 0.1490343038339579,
+      "grad_norm": 98.03365050550292,
+      "learning_rate": 0.0001,
+      "loss": 0.8544,
+      "num_input_tokens_seen": 76224288,
+      "step": 517
+    },
+    {
+      "epoch": 0.1490343038339579,
+      "loss": 0.7350907325744629,
+      "loss_ce": 0.4651932716369629,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 76224288,
+      "step": 517
+    },
+    {
+      "epoch": 0.14932257134620927,
+      "grad_norm": 90.15703552305258,
+      "learning_rate": 0.0001,
+      "loss": 0.7748,
+      "num_input_tokens_seen": 76359384,
+      "step": 518
+    },
+    {
+      "epoch": 0.14932257134620927,
+      "loss": 0.6386263370513916,
+      "loss_ce": 0.4304353594779968,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 76359384,
+      "step": 518
+    },
+    {
+      "epoch": 0.14961083885846066,
+      "grad_norm": 205.27603193926652,
+      "learning_rate": 0.0001,
+      "loss": 1.8418,
+      "num_input_tokens_seen": 76531944,
+      "step": 519
+    },
+    {
+      "epoch": 0.14961083885846066,
+      "loss": 1.29929518699646,
+      "loss_ce": 0.5361114740371704,
+      "loss_xval": 0.76171875,
+      "num_input_tokens_seen": 76531944,
+      "step": 519
+    },
+    {
+      "epoch": 0.14989910637071202,
+      "grad_norm": 78.52290789401435,
+      "learning_rate": 0.0001,
+      "loss": 0.7662,
+      "num_input_tokens_seen": 76666672,
+      "step": 520
+    },
+    {
+      "epoch": 0.14989910637071202,
+      "loss": 0.9968472123146057,
+      "loss_ce": 0.4050503075122833,
+      "loss_xval": 0.59375,
+      "num_input_tokens_seen": 76666672,
+      "step": 520
+    },
+    {
+      "epoch": 0.15018737388296338,
+      "grad_norm": 207.7463656688498,
+      "learning_rate": 0.0001,
+      "loss": 1.3291,
+      "num_input_tokens_seen": 76801760,
+      "step": 521
+    },
+    {
+      "epoch": 0.15018737388296338,
+      "loss": 1.097721815109253,
+      "loss_ce": 0.37360072135925293,
+      "loss_xval": 0.72265625,
+      "num_input_tokens_seen": 76801760,
+      "step": 521
+    },
+    {
+      "epoch": 0.15047564139521477,
+      "grad_norm": 135.28340638854004,
+      "learning_rate": 0.0001,
+      "loss": 1.0985,
+      "num_input_tokens_seen": 76974288,
+      "step": 522
+    },
+    {
+      "epoch": 0.15047564139521477,
+      "loss": 1.4575904607772827,
+      "loss_ce": 0.49762946367263794,
+      "loss_xval": 0.9609375,
+      "num_input_tokens_seen": 76974288,
+      "step": 522
+    },
+    {
+      "epoch": 0.15076390890746613,
+      "grad_norm": 24.17987630181005,
+      "learning_rate": 0.0001,
+      "loss": 0.53,
+      "num_input_tokens_seen": 77109096,
+      "step": 523
+    },
+    {
+      "epoch": 0.15076390890746613,
+      "loss": 0.5469415187835693,
+      "loss_ce": 0.40826964378356934,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 77109096,
+      "step": 523
+    },
+    {
+      "epoch": 0.1510521764197175,
+      "grad_norm": 100.11571223596611,
+      "learning_rate": 0.0001,
+      "loss": 0.8684,
+      "num_input_tokens_seen": 77244008,
+      "step": 524
+    },
+    {
+      "epoch": 0.1510521764197175,
+      "loss": 0.9112217426300049,
+      "loss_ce": 0.3496982455253601,
+      "loss_xval": 0.5625,
+      "num_input_tokens_seen": 77244008,
+      "step": 524
+    },
+    {
+      "epoch": 0.15134044393196885,
+      "grad_norm": 49.317152729974175,
+      "learning_rate": 0.0001,
+      "loss": 0.6577,
+      "num_input_tokens_seen": 77416440,
+      "step": 525
+    },
+    {
+      "epoch": 0.15134044393196885,
+      "loss": 0.6915754079818726,
+      "loss_ce": 0.48088207840919495,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 77416440,
+      "step": 525
+    },
+    {
+      "epoch": 0.15162871144422024,
+      "grad_norm": 37.88388740046188,
+      "learning_rate": 0.0001,
+      "loss": 0.5633,
+      "num_input_tokens_seen": 77551096,
+      "step": 526
+    },
+    {
+      "epoch": 0.15162871144422024,
+      "loss": 0.6765055656433105,
+      "loss_ce": 0.40758466720581055,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 77551096,
+      "step": 526
+    },
+    {
+      "epoch": 0.1519169789564716,
+      "grad_norm": 96.29838625038265,
+      "learning_rate": 0.0001,
+      "loss": 0.539,
+      "num_input_tokens_seen": 77686080,
+      "step": 527
+    },
+    {
+      "epoch": 0.1519169789564716,
+      "loss": 0.3767795264720917,
+      "loss_ce": 0.2926425635814667,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 77686080,
+      "step": 527
+    },
+    {
+      "epoch": 0.15220524646872297,
+      "grad_norm": 33.551167364865414,
+      "learning_rate": 0.0001,
+      "loss": 0.4591,
+      "num_input_tokens_seen": 77858584,
+      "step": 528
+    },
+    {
+      "epoch": 0.15220524646872297,
+      "loss": 0.4692811965942383,
+      "loss_ce": 0.3921327590942383,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 77858584,
+      "step": 528
+    },
+    {
+      "epoch": 0.15249351398097435,
+      "grad_norm": 25.602899058495524,
+      "learning_rate": 0.0001,
+      "loss": 0.3878,
+      "num_input_tokens_seen": 77993448,
+      "step": 529
+    },
+    {
+      "epoch": 0.15249351398097435,
+      "loss": 0.4130549132823944,
+      "loss_ce": 0.3271784484386444,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 77993448,
+      "step": 529
+    },
+    {
+      "epoch": 0.15278178149322572,
+      "grad_norm": 25.35178128801067,
+      "learning_rate": 0.0001,
+      "loss": 0.3704,
+      "num_input_tokens_seen": 78128592,
+      "step": 530
+    },
+    {
+      "epoch": 0.15278178149322572,
+      "loss": 0.3342554569244385,
+      "loss_ce": 0.2731592655181885,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 78128592,
+      "step": 530
+    },
+    {
+      "epoch": 0.15307004900547708,
+      "grad_norm": 14.81325168312329,
+      "learning_rate": 0.0001,
+      "loss": 0.4114,
+      "num_input_tokens_seen": 78301192,
+      "step": 531
+    },
+    {
+      "epoch": 0.15307004900547708,
+      "loss": 0.42171674966812134,
+      "loss_ce": 0.37812238931655884,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 78301192,
+      "step": 531
+    },
+    {
+      "epoch": 0.15335831651772847,
+      "grad_norm": 24.1013008056333,
+      "learning_rate": 0.0001,
+      "loss": 0.3783,
+      "num_input_tokens_seen": 78436040,
+      "step": 532
+    },
+    {
+      "epoch": 0.15335831651772847,
+      "loss": 0.38421040773391724,
+      "loss_ce": 0.31337910890579224,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 78436040,
+      "step": 532
+    },
+    {
+      "epoch": 0.15364658402997983,
+      "grad_norm": 37.4499157075249,
+      "learning_rate": 0.0001,
+      "loss": 0.392,
+      "num_input_tokens_seen": 78571040,
+      "step": 533
+    },
+    {
+      "epoch": 0.15364658402997983,
+      "loss": 0.3791619539260864,
+      "loss_ce": 0.2706414461135864,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 78571040,
+      "step": 533
+    },
+    {
+      "epoch": 0.1539348515422312,
+      "grad_norm": 12.129510955905058,
+      "learning_rate": 0.0001,
+      "loss": 0.3663,
+      "num_input_tokens_seen": 78743488,
+      "step": 534
+    },
+    {
+      "epoch": 0.1539348515422312,
+      "loss": 0.3491227328777313,
+      "loss_ce": 0.3058335483074188,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 78743488,
+      "step": 534
+    },
+    {
+      "epoch": 0.15422311905448255,
+      "grad_norm": 23.735009533377152,
+      "learning_rate": 0.0001,
+      "loss": 0.3452,
+      "num_input_tokens_seen": 78878208,
+      "step": 535
+    },
+    {
+      "epoch": 0.15422311905448255,
+      "loss": 0.39459940791130066,
+      "loss_ce": 0.27881571650505066,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 78878208,
+      "step": 535
+    },
+    {
+      "epoch": 0.15451138656673394,
+      "grad_norm": 18.28117522583193,
+      "learning_rate": 0.0001,
+      "loss": 0.3246,
+      "num_input_tokens_seen": 79013256,
+      "step": 536
+    },
+    {
+      "epoch": 0.15451138656673394,
+      "loss": 0.2895457148551941,
+      "loss_ce": 0.2394358515739441,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 79013256,
+      "step": 536
+    },
+    {
+      "epoch": 0.1547996540789853,
+      "grad_norm": 6.24255692241842,
+      "learning_rate": 0.0001,
+      "loss": 0.3245,
+      "num_input_tokens_seen": 79185768,
+      "step": 537
+    },
+    {
+      "epoch": 0.1547996540789853,
+      "loss": 0.3146858811378479,
+      "loss_ce": 0.2829018235206604,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 79185768,
+      "step": 537
+    },
+    {
+      "epoch": 0.15508792159123666,
+      "grad_norm": 18.842244440998723,
+      "learning_rate": 0.0001,
+      "loss": 0.302,
+      "num_input_tokens_seen": 79320536,
+      "step": 538
+    },
+    {
+      "epoch": 0.15508792159123666,
+      "loss": 0.32134073972702026,
+      "loss_ce": 0.24397867918014526,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 79320536,
+      "step": 538
+    },
+    {
+      "epoch": 0.15537618910348805,
+      "grad_norm": 13.401811339057847,
+      "learning_rate": 0.0001,
+      "loss": 0.2766,
+      "num_input_tokens_seen": 79455824,
+      "step": 539
+    },
+    {
+      "epoch": 0.15537618910348805,
+      "loss": 0.255174458026886,
+      "loss_ce": 0.2099168747663498,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 79455824,
+      "step": 539
+    },
+    {
+      "epoch": 0.1556644566157394,
+      "grad_norm": 8.87835003264218,
+      "learning_rate": 0.0001,
+      "loss": 0.2796,
+      "num_input_tokens_seen": 79628288,
+      "step": 540
+    },
+    {
+      "epoch": 0.1556644566157394,
+      "loss": 0.28175222873687744,
+      "loss_ce": 0.24963247776031494,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 79628288,
+      "step": 540
+    },
+    {
+      "epoch": 0.15595272412799077,
+      "grad_norm": 10.764691594571945,
+      "learning_rate": 0.0001,
+      "loss": 0.2552,
+      "num_input_tokens_seen": 79763104,
+      "step": 541
+    },
+    {
+      "epoch": 0.15595272412799077,
+      "loss": 0.2863326072692871,
+      "loss_ce": 0.2172713279724121,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 79763104,
+      "step": 541
+    },
+    {
+      "epoch": 0.15624099164024213,
+      "grad_norm": 14.007259191137436,
+      "learning_rate": 0.0001,
+      "loss": 0.2418,
+      "num_input_tokens_seen": 79898184,
+      "step": 542
+    },
+    {
+      "epoch": 0.15624099164024213,
+      "loss": 0.21299250423908234,
+      "loss_ce": 0.18197138607501984,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 79898184,
+      "step": 542
+    },
+    {
+      "epoch": 0.15652925915249352,
+      "grad_norm": 7.639771826560467,
+      "learning_rate": 0.0001,
+      "loss": 0.2555,
+      "num_input_tokens_seen": 80070632,
+      "step": 543
+    },
+    {
+      "epoch": 0.15652925915249352,
+      "loss": 0.24649299681186676,
+      "loss_ce": 0.22470344603061676,
+      "loss_xval": 0.021728515625,
+      "num_input_tokens_seen": 80070632,
+      "step": 543
+    },
+    {
+      "epoch": 0.15681752666474488,
+      "grad_norm": 16.12636329610739,
+      "learning_rate": 0.0001,
+      "loss": 0.2312,
+      "num_input_tokens_seen": 80205424,
+      "step": 544
+    },
+    {
+      "epoch": 0.15681752666474488,
+      "loss": 0.26397019624710083,
+      "loss_ce": 0.19713670015335083,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 80205424,
+      "step": 544
+    },
+    {
+      "epoch": 0.15710579417699624,
+      "grad_norm": 7.993418479045426,
+      "learning_rate": 0.0001,
+      "loss": 0.2087,
+      "num_input_tokens_seen": 80340528,
+      "step": 545
+    },
+    {
+      "epoch": 0.15710579417699624,
+      "loss": 0.18511179089546204,
+      "loss_ce": 0.16072824597358704,
+      "loss_xval": 0.0244140625,
+      "num_input_tokens_seen": 80340528,
+      "step": 545
+    },
+    {
+      "epoch": 0.15739406168924763,
+      "grad_norm": 10.595619750311025,
+      "learning_rate": 0.0001,
+      "loss": 0.2364,
+      "num_input_tokens_seen": 80513160,
+      "step": 546
+    },
+    {
+      "epoch": 0.15739406168924763,
+      "loss": 0.24273675680160522,
+      "loss_ce": 0.21314996480941772,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 80513160,
+      "step": 546
+    },
+    {
+      "epoch": 0.157682329201499,
+      "grad_norm": 6.930279341659058,
+      "learning_rate": 0.0001,
+      "loss": 0.1913,
+      "num_input_tokens_seen": 80648040,
+      "step": 547
+    },
+    {
+      "epoch": 0.157682329201499,
+      "loss": 0.22323384881019592,
+      "loss_ce": 0.18223348259925842,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 80648040,
+      "step": 547
+    },
+    {
+      "epoch": 0.15797059671375036,
+      "grad_norm": 8.714721085520644,
+      "learning_rate": 0.0001,
+      "loss": 0.1862,
+      "num_input_tokens_seen": 80783152,
+      "step": 548
+    },
+    {
+      "epoch": 0.15797059671375036,
+      "loss": 0.15961964428424835,
+      "loss_ce": 0.13665516674518585,
+      "loss_xval": 0.02294921875,
+      "num_input_tokens_seen": 80783152,
+      "step": 548
+    },
+    {
+      "epoch": 0.15825886422600172,
+      "grad_norm": 6.696238976718923,
+      "learning_rate": 0.0001,
+      "loss": 0.2109,
+      "num_input_tokens_seen": 80955488,
+      "step": 549
+    },
+    {
+      "epoch": 0.15825886422600172,
+      "loss": 0.20110364258289337,
+      "loss_ce": 0.17869611084461212,
+      "loss_xval": 0.0224609375,
+      "num_input_tokens_seen": 80955488,
+      "step": 549
+    },
+    {
+      "epoch": 0.1585471317382531,
+      "grad_norm": 4.674473836001116,
+      "learning_rate": 0.0001,
+      "loss": 0.1755,
+      "num_input_tokens_seen": 81090160,
+      "step": 550
+    },
+    {
+      "epoch": 0.1585471317382531,
+      "loss": 0.20668087899684906,
+      "loss_ce": 0.16780148446559906,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 81090160,
+      "step": 550
+    },
+    {
+      "epoch": 0.15883539925050447,
+      "grad_norm": 5.464329287182501,
+      "learning_rate": 0.0001,
+      "loss": 0.1676,
+      "num_input_tokens_seen": 81225456,
+      "step": 551
+    },
+    {
+      "epoch": 0.15883539925050447,
+      "loss": 0.14501583576202393,
+      "loss_ce": 0.12630856037139893,
+      "loss_xval": 0.0186767578125,
+      "num_input_tokens_seen": 81225456,
+      "step": 551
+    },
+    {
+      "epoch": 0.15912366676275583,
+      "grad_norm": 15.54738056116537,
+      "learning_rate": 0.0001,
+      "loss": 0.1926,
+      "num_input_tokens_seen": 81397864,
+      "step": 552
+    },
+    {
+      "epoch": 0.15912366676275583,
+      "loss": 0.19035065174102783,
+      "loss_ce": 0.16454041004180908,
+      "loss_xval": 0.0257568359375,
+      "num_input_tokens_seen": 81397864,
+      "step": 552
+    },
+    {
+      "epoch": 0.15941193427500722,
+      "grad_norm": 9.823333175994385,
+      "learning_rate": 0.0001,
+      "loss": 0.1606,
+      "num_input_tokens_seen": 81532616,
+      "step": 553
+    },
+    {
+      "epoch": 0.15941193427500722,
+      "loss": 0.18590956926345825,
+      "loss_ce": 0.155170738697052,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 81532616,
+      "step": 553
+    },
+    {
+      "epoch": 0.15970020178725858,
+      "grad_norm": 11.081119125897253,
+      "learning_rate": 0.0001,
+      "loss": 0.1459,
+      "num_input_tokens_seen": 81667744,
+      "step": 554
+    },
+    {
+      "epoch": 0.15970020178725858,
+      "loss": 0.1296243667602539,
+      "loss_ce": 0.11304950714111328,
+      "loss_xval": 0.0166015625,
+      "num_input_tokens_seen": 81667744,
+      "step": 554
+    },
+    {
+      "epoch": 0.15998846929950994,
+      "grad_norm": 10.365269177618455,
+      "learning_rate": 0.0001,
+      "loss": 0.1766,
+      "num_input_tokens_seen": 81840440,
+      "step": 555
+    },
+    {
+      "epoch": 0.15998846929950994,
+      "loss": 0.17812153697013855,
+      "loss_ce": 0.15227314829826355,
+      "loss_xval": 0.02587890625,
+      "num_input_tokens_seen": 81840440,
+      "step": 555
+    },
+    {
+      "epoch": 0.16027673681176133,
+      "grad_norm": 16.301766057599238,
+      "learning_rate": 0.0001,
+      "loss": 0.1505,
+      "num_input_tokens_seen": 81975280,
+      "step": 556
+    },
+    {
+      "epoch": 0.16027673681176133,
+      "loss": 0.1683354675769806,
+      "loss_ce": 0.1410680115222931,
+      "loss_xval": 0.0272216796875,
+      "num_input_tokens_seen": 81975280,
+      "step": 556
+    },
+    {
+      "epoch": 0.1605650043240127,
+      "grad_norm": 3.957116361728929,
+      "learning_rate": 0.0001,
+      "loss": 0.1385,
+      "num_input_tokens_seen": 82110312,
+      "step": 557
+    },
+    {
+      "epoch": 0.1605650043240127,
+      "loss": 0.12488369643688202,
+      "loss_ce": 0.10235408693552017,
+      "loss_xval": 0.0225830078125,
+      "num_input_tokens_seen": 82110312,
+      "step": 557
+    },
+    {
+      "epoch": 0.16085327183626405,
+      "grad_norm": 16.430070798635164,
+      "learning_rate": 0.0001,
+      "loss": 0.1404,
+      "num_input_tokens_seen": 82282880,
+      "step": 558
+    },
+    {
+      "epoch": 0.16085327183626405,
+      "loss": 0.13027092814445496,
+      "loss_ce": 0.11491294205188751,
+      "loss_xval": 0.015380859375,
+      "num_input_tokens_seen": 82282880,
+      "step": 558
+    },
+    {
+      "epoch": 0.1611415393485154,
+      "grad_norm": 5.8377270996113575,
+      "learning_rate": 0.0001,
+      "loss": 0.1266,
+      "num_input_tokens_seen": 82417696,
+      "step": 559
+    },
+    {
+      "epoch": 0.1611415393485154,
+      "loss": 0.14256012439727783,
+      "loss_ce": 0.12391388416290283,
+      "loss_xval": 0.0186767578125,
+      "num_input_tokens_seen": 82417696,
+      "step": 559
+    },
+    {
+      "epoch": 0.1614298068607668,
+      "grad_norm": 8.064025990014587,
+      "learning_rate": 0.0001,
+      "loss": 0.1194,
+      "num_input_tokens_seen": 82552728,
+      "step": 560
+    },
+    {
+      "epoch": 0.1614298068607668,
+      "loss": 0.10484252870082855,
+      "loss_ce": 0.08726440370082855,
+      "loss_xval": 0.017578125,
+      "num_input_tokens_seen": 82552728,
+      "step": 560
+    },
+    {
+      "epoch": 0.16171807437301816,
+      "grad_norm": 20.098214952476035,
+      "learning_rate": 0.0001,
+      "loss": 0.124,
+      "num_input_tokens_seen": 82725352,
+      "step": 561
+    },
+    {
+      "epoch": 0.16171807437301816,
+      "loss": 0.11904975771903992,
+      "loss_ce": 0.09915992617607117,
+      "loss_xval": 0.0198974609375,
+      "num_input_tokens_seen": 82725352,
+      "step": 561
+    },
+    {
+      "epoch": 0.16200634188526952,
+      "grad_norm": 7.901668286399249,
+      "learning_rate": 0.0001,
+      "loss": 0.1153,
+      "num_input_tokens_seen": 82860184,
+      "step": 562
+    },
+    {
+      "epoch": 0.16200634188526952,
+      "loss": 0.12794066965579987,
+      "loss_ce": 0.10678437352180481,
+      "loss_xval": 0.0211181640625,
+      "num_input_tokens_seen": 82860184,
+      "step": 562
+    },
+    {
+      "epoch": 0.1622946093975209,
+      "grad_norm": 6.294120755456603,
+      "learning_rate": 0.0001,
+      "loss": 0.0987,
+      "num_input_tokens_seen": 82995280,
+      "step": 563
+    },
+    {
+      "epoch": 0.1622946093975209,
+      "loss": 0.09468768537044525,
+      "loss_ce": 0.079879030585289,
+      "loss_xval": 0.01483154296875,
+      "num_input_tokens_seen": 82995280,
+      "step": 563
+    },
+    {
+      "epoch": 0.16258287690977227,
+      "grad_norm": 11.065605042496632,
+      "learning_rate": 0.0001,
+      "loss": 0.114,
+      "num_input_tokens_seen": 83167720,
+      "step": 564
+    },
+    {
+      "epoch": 0.16258287690977227,
+      "loss": 0.10833622515201569,
+      "loss_ce": 0.08724857866764069,
+      "loss_xval": 0.0211181640625,
+      "num_input_tokens_seen": 83167720,
+      "step": 564
+    },
+    {
+      "epoch": 0.16287114442202363,
+      "grad_norm": 9.562876080765191,
+      "learning_rate": 0.0001,
+      "loss": 0.0975,
+      "num_input_tokens_seen": 83302488,
+      "step": 565
+    },
+    {
+      "epoch": 0.16287114442202363,
+      "loss": 0.11148280650377274,
+      "loss_ce": 0.09303493797779083,
+      "loss_xval": 0.0184326171875,
+      "num_input_tokens_seen": 83302488,
+      "step": 565
+    },
+    {
+      "epoch": 0.163159411934275,
+      "grad_norm": 6.714478129953445,
+      "learning_rate": 0.0001,
+      "loss": 0.0921,
+      "num_input_tokens_seen": 83437424,
+      "step": 566
+    },
+    {
+      "epoch": 0.163159411934275,
+      "loss": 0.0829567164182663,
+      "loss_ce": 0.06752245128154755,
+      "loss_xval": 0.01544189453125,
+      "num_input_tokens_seen": 83437424,
+      "step": 566
+    },
+    {
+      "epoch": 0.16344767944652638,
+      "grad_norm": 4.429600475589518,
+      "learning_rate": 0.0001,
+      "loss": 0.1009,
+      "num_input_tokens_seen": 83610000,
+      "step": 567
+    },
+    {
+      "epoch": 0.16344767944652638,
+      "loss": 0.09480486810207367,
+      "loss_ce": 0.07885180413722992,
+      "loss_xval": 0.0159912109375,
+      "num_input_tokens_seen": 83610000,
+      "step": 567
+    },
+    {
+      "epoch": 0.16373594695877774,
+      "grad_norm": 8.38652460181145,
+      "learning_rate": 0.0001,
+      "loss": 0.0913,
+      "num_input_tokens_seen": 83744816,
+      "step": 568
+    },
+    {
+      "epoch": 0.16373594695877774,
+      "loss": 0.10490494966506958,
+      "loss_ce": 0.08726578950881958,
+      "loss_xval": 0.017578125,
+      "num_input_tokens_seen": 83744816,
+      "step": 568
+    },
+    {
+      "epoch": 0.1640242144710291,
+      "grad_norm": 4.1872332233055545,
+      "learning_rate": 0.0001,
+      "loss": 0.0846,
+      "num_input_tokens_seen": 83879928,
+      "step": 569
+    },
+    {
+      "epoch": 0.1640242144710291,
+      "loss": 0.07664161920547485,
+      "loss_ce": 0.060459673404693604,
+      "loss_xval": 0.0162353515625,
+      "num_input_tokens_seen": 83879928,
+      "step": 569
+    },
+    {
+      "epoch": 0.1643124819832805,
+      "grad_norm": 6.256329066436033,
+      "learning_rate": 0.0001,
+      "loss": 0.0895,
+      "num_input_tokens_seen": 84052392,
+      "step": 570
+    },
+    {
+      "epoch": 0.1643124819832805,
+      "loss": 0.08122982084751129,
+      "loss_ce": 0.06586040556430817,
+      "loss_xval": 0.015380859375,
+      "num_input_tokens_seen": 84052392,
+      "step": 570
+    },
+    {
+      "epoch": 0.16460074949553186,
+      "grad_norm": 5.398141020968809,
+      "learning_rate": 0.0001,
+      "loss": 0.0777,
+      "num_input_tokens_seen": 84187232,
+      "step": 571
+    },
+    {
+      "epoch": 0.16460074949553186,
+      "loss": 0.0879916399717331,
+      "loss_ce": 0.0741061419248581,
+      "loss_xval": 0.013916015625,
+      "num_input_tokens_seen": 84187232,
+      "step": 571
+    },
+    {
+      "epoch": 0.16488901700778322,
+      "grad_norm": 4.902929545953649,
+      "learning_rate": 0.0001,
+      "loss": 0.0698,
+      "num_input_tokens_seen": 84322272,
+      "step": 572
+    },
+    {
+      "epoch": 0.16488901700778322,
+      "loss": 0.06459660828113556,
+      "loss_ce": 0.05026097595691681,
+      "loss_xval": 0.01434326171875,
+      "num_input_tokens_seen": 84322272,
+      "step": 572
+    },
+    {
+      "epoch": 0.16517728452003458,
+      "grad_norm": 9.525756844667548,
+      "learning_rate": 0.0001,
+      "loss": 0.08,
+      "num_input_tokens_seen": 84494768,
+      "step": 573
+    },
+    {
+      "epoch": 0.16517728452003458,
+      "loss": 0.07009953260421753,
+      "loss_ce": 0.05596226826310158,
+      "loss_xval": 0.01416015625,
+      "num_input_tokens_seen": 84494768,
+      "step": 573
+    },
+    {
+      "epoch": 0.16546555203228597,
+      "grad_norm": 5.313776134745736,
+      "learning_rate": 0.0001,
+      "loss": 0.0721,
+      "num_input_tokens_seen": 84629568,
+      "step": 574
+    },
+    {
+      "epoch": 0.16546555203228597,
+      "loss": 0.08415266871452332,
+      "loss_ce": 0.07041975855827332,
+      "loss_xval": 0.01373291015625,
+      "num_input_tokens_seen": 84629568,
+      "step": 574
+    },
+    {
+      "epoch": 0.16575381954453733,
+      "grad_norm": 9.375819271631455,
+      "learning_rate": 0.0001,
+      "loss": 0.0646,
+      "num_input_tokens_seen": 84764472,
+      "step": 575
+    },
+    {
+      "epoch": 0.16575381954453733,
+      "loss": 0.05780894309282303,
+      "loss_ce": 0.04357249289751053,
+      "loss_xval": 0.01422119140625,
+      "num_input_tokens_seen": 84764472,
+      "step": 575
+    },
+    {
+      "epoch": 0.1660420870567887,
+      "grad_norm": 5.726067399275158,
+      "learning_rate": 0.0001,
+      "loss": 0.0764,
+      "num_input_tokens_seen": 84937008,
+      "step": 576
+    },
+    {
+      "epoch": 0.1660420870567887,
+      "loss": 0.07145722210407257,
+      "loss_ce": 0.05638917163014412,
+      "loss_xval": 0.01507568359375,
+      "num_input_tokens_seen": 84937008,
+      "step": 576
+    },
+    {
+      "epoch": 0.16633035456904008,
+      "grad_norm": 9.691222941127505,
+      "learning_rate": 0.0001,
+      "loss": 0.0636,
+      "num_input_tokens_seen": 85071872,
+      "step": 577
+    },
+    {
+      "epoch": 0.16633035456904008,
+      "loss": 0.07909021526575089,
+      "loss_ce": 0.06402216106653214,
+      "loss_xval": 0.01507568359375,
+      "num_input_tokens_seen": 85071872,
+      "step": 577
+    },
+    {
+      "epoch": 0.16661862208129144,
+      "grad_norm": 9.189882675536726,
+      "learning_rate": 0.0001,
+      "loss": 0.0571,
+      "num_input_tokens_seen": 85207072,
+      "step": 578
+    },
+    {
+      "epoch": 0.16661862208129144,
+      "loss": 0.04689406231045723,
+      "loss_ce": 0.03543471172451973,
+      "loss_xval": 0.011474609375,
+      "num_input_tokens_seen": 85207072,
+      "step": 578
+    },
+    {
+      "epoch": 0.1669068895935428,
+      "grad_norm": 7.183323348640227,
+      "learning_rate": 0.0001,
+      "loss": 0.0647,
+      "num_input_tokens_seen": 85379712,
+      "step": 579
+    },
+    {
+      "epoch": 0.1669068895935428,
+      "loss": 0.0562661737203598,
+      "loss_ce": 0.0435708612203598,
+      "loss_xval": 0.0126953125,
+      "num_input_tokens_seen": 85379712,
+      "step": 579
+    },
+    {
+      "epoch": 0.1671951571057942,
+      "grad_norm": 7.625349683880436,
+      "learning_rate": 0.0001,
+      "loss": 0.0572,
+      "num_input_tokens_seen": 85514480,
+      "step": 580
+    },
+    {
+      "epoch": 0.1671951571057942,
+      "loss": 0.07185040414333344,
+      "loss_ce": 0.058354005217552185,
+      "loss_xval": 0.01348876953125,
+      "num_input_tokens_seen": 85514480,
+      "step": 580
+    },
+    {
+      "epoch": 0.16748342461804555,
+      "grad_norm": 7.025990709824806,
+      "learning_rate": 0.0001,
+      "loss": 0.048,
+      "num_input_tokens_seen": 85649536,
+      "step": 581
+    },
+    {
+      "epoch": 0.16748342461804555,
+      "loss": 0.0399930477142334,
+      "loss_ce": 0.02949499897658825,
+      "loss_xval": 0.010498046875,
+      "num_input_tokens_seen": 85649536,
+      "step": 581
+    },
+    {
+      "epoch": 0.1677716921302969,
+      "grad_norm": 9.258264080755461,
+      "learning_rate": 0.0001,
+      "loss": 0.0616,
+      "num_input_tokens_seen": 85822208,
+      "step": 582
+    },
+    {
+      "epoch": 0.1677716921302969,
+      "loss": 0.06172851473093033,
+      "loss_ce": 0.05022338777780533,
+      "loss_xval": 0.011474609375,
+      "num_input_tokens_seen": 85822208,
+      "step": 582
+    },
+    {
+      "epoch": 0.16805995964254827,
+      "grad_norm": 5.89116810275108,
+      "learning_rate": 0.0001,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 85957008,
+      "step": 583
+    },
+    {
+      "epoch": 0.16805995964254827,
+      "loss": 0.06277771294116974,
+      "loss_ce": 0.04876251146197319,
+      "loss_xval": 0.0140380859375,
+      "num_input_tokens_seen": 85957008,
+      "step": 583
+    },
+    {
+      "epoch": 0.16834822715479966,
+      "grad_norm": 10.182576802733074,
+      "learning_rate": 0.0001,
+      "loss": 0.0435,
+      "num_input_tokens_seen": 86092000,
+      "step": 584
+    },
+    {
+      "epoch": 0.16834822715479966,
+      "loss": 0.03622671216726303,
+      "loss_ce": 0.02325674146413803,
+      "loss_xval": 0.012939453125,
+      "num_input_tokens_seen": 86092000,
+      "step": 584
+    },
+    {
+      "epoch": 0.16863649466705102,
+      "grad_norm": 15.131169214119659,
+      "learning_rate": 0.0001,
+      "loss": 0.052,
+      "num_input_tokens_seen": 86264520,
+      "step": 585
+    },
+    {
+      "epoch": 0.16863649466705102,
+      "loss": 0.04348108172416687,
+      "loss_ce": 0.029843537136912346,
+      "loss_xval": 0.01361083984375,
+      "num_input_tokens_seen": 86264520,
+      "step": 585
+    },
+    {
+      "epoch": 0.16892476217930238,
+      "grad_norm": 6.965049780112206,
+      "learning_rate": 0.0001,
+      "loss": 0.0414,
+      "num_input_tokens_seen": 86399272,
+      "step": 586
+    },
+    {
+      "epoch": 0.16892476217930238,
+      "loss": 0.04911273717880249,
+      "loss_ce": 0.04123920202255249,
+      "loss_xval": 0.00787353515625,
+      "num_input_tokens_seen": 86399272,
+      "step": 586
+    },
+    {
+      "epoch": 0.16921302969155377,
+      "grad_norm": 4.913891385632238,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "num_input_tokens_seen": 86534408,
+      "step": 587
+    },
+    {
+      "epoch": 0.16921302969155377,
+      "loss": 0.027581913396716118,
+      "loss_ce": 0.020135624334216118,
+      "loss_xval": 0.0074462890625,
+      "num_input_tokens_seen": 86534408,
+      "step": 587
+    },
+    {
+      "epoch": 0.16950129720380513,
+      "grad_norm": 2.482330696638268,
+      "learning_rate": 0.0001,
+      "loss": 0.0473,
+      "num_input_tokens_seen": 86706880,
+      "step": 588
+    },
+    {
+      "epoch": 0.16950129720380513,
+      "loss": 0.045104727149009705,
+      "loss_ce": 0.037139639258384705,
+      "loss_xval": 0.0079345703125,
+      "num_input_tokens_seen": 86706880,
+      "step": 588
+    },
+    {
+      "epoch": 0.1697895647160565,
+      "grad_norm": 7.563645548775605,
+      "learning_rate": 0.0001,
+      "loss": 0.037,
+      "num_input_tokens_seen": 86841632,
+      "step": 589
+    },
+    {
+      "epoch": 0.1697895647160565,
+      "loss": 0.04473777860403061,
+      "loss_ce": 0.03411003202199936,
+      "loss_xval": 0.0106201171875,
+      "num_input_tokens_seen": 86841632,
+      "step": 589
+    },
+    {
+      "epoch": 0.17007783222830786,
+      "grad_norm": 6.679156719751855,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "num_input_tokens_seen": 86976736,
+      "step": 590
+    },
+    {
+      "epoch": 0.17007783222830786,
+      "loss": 0.024894479662179947,
+      "loss_ce": 0.017337564378976822,
+      "loss_xval": 0.007568359375,
+      "num_input_tokens_seen": 86976736,
+      "step": 590
+    },
+    {
+      "epoch": 0.17036609974055925,
+      "grad_norm": 4.787149514617314,
+      "learning_rate": 0.0001,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 87149400,
+      "step": 591
+    },
+    {
+      "epoch": 0.17036609974055925,
+      "loss": 0.04200609773397446,
+      "loss_ce": 0.028608879074454308,
+      "loss_xval": 0.013427734375,
+      "num_input_tokens_seen": 87149400,
+      "step": 591
+    },
+    {
+      "epoch": 0.1706543672528106,
+      "grad_norm": 11.614698173696578,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "num_input_tokens_seen": 87284192,
+      "step": 592
+    },
+    {
+      "epoch": 0.1706543672528106,
+      "loss": 0.04323431849479675,
+      "loss_ce": 0.03306815028190613,
+      "loss_xval": 0.01019287109375,
+      "num_input_tokens_seen": 87284192,
+      "step": 592
+    },
+    {
+      "epoch": 0.17094263476506197,
+      "grad_norm": 2.397756579315926,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "num_input_tokens_seen": 87419176,
+      "step": 593
+    },
+    {
+      "epoch": 0.17094263476506197,
+      "loss": 0.020117074251174927,
+      "loss_ce": 0.013338357210159302,
+      "loss_xval": 0.00677490234375,
+      "num_input_tokens_seen": 87419176,
+      "step": 593
+    },
+    {
+      "epoch": 0.17123090227731336,
+      "grad_norm": 9.654477039282371,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "num_input_tokens_seen": 87591752,
+      "step": 594
+    },
+    {
+      "epoch": 0.17123090227731336,
+      "loss": 0.028524693101644516,
+      "loss_ce": 0.019094761461019516,
+      "loss_xval": 0.0093994140625,
+      "num_input_tokens_seen": 87591752,
+      "step": 594
+    },
+    {
+      "epoch": 0.17151916978956472,
+      "grad_norm": 5.367170509753708,
+      "learning_rate": 0.0001,
+      "loss": 0.0299,
+      "num_input_tokens_seen": 87726488,
+      "step": 595
+    },
+    {
+      "epoch": 0.17151916978956472,
+      "loss": 0.038070403039455414,
+      "loss_ce": 0.027030669152736664,
+      "loss_xval": 0.01104736328125,
+      "num_input_tokens_seen": 87726488,
+      "step": 595
+    },
+    {
+      "epoch": 0.17180743730181608,
+      "grad_norm": 5.695877234150102,
+      "learning_rate": 0.0001,
+      "loss": 0.0245,
+      "num_input_tokens_seen": 87861664,
+      "step": 596
+    },
+    {
+      "epoch": 0.17180743730181608,
+      "loss": 0.02125311642885208,
+      "loss_ce": 0.010629183612763882,
+      "loss_xval": 0.0106201171875,
+      "num_input_tokens_seen": 87861664,
+      "step": 596
+    },
+    {
+      "epoch": 0.17209570481406747,
+      "grad_norm": 5.263429908122342,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "num_input_tokens_seen": 88034152,
+      "step": 597
+    },
+    {
+      "epoch": 0.17209570481406747,
+      "loss": 0.03428371623158455,
+      "loss_ce": 0.025364950299263,
+      "loss_xval": 0.0089111328125,
+      "num_input_tokens_seen": 88034152,
+      "step": 597
+    },
+    {
+      "epoch": 0.17238397232631883,
+      "grad_norm": 8.564515731366683,
+      "learning_rate": 0.0001,
+      "loss": 0.0273,
+      "num_input_tokens_seen": 88168984,
+      "step": 598
+    },
+    {
+      "epoch": 0.17238397232631883,
+      "loss": 0.03684600442647934,
+      "loss_ce": 0.025626981630921364,
+      "loss_xval": 0.01123046875,
+      "num_input_tokens_seen": 88168984,
+      "step": 598
+    },
+    {
+      "epoch": 0.1726722398385702,
+      "grad_norm": 9.058382947276888,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "num_input_tokens_seen": 88303992,
+      "step": 599
+    },
+    {
+      "epoch": 0.1726722398385702,
+      "loss": 0.016890352591872215,
+      "loss_ce": 0.00854760967195034,
+      "loss_xval": 0.00836181640625,
+      "num_input_tokens_seen": 88303992,
+      "step": 599
+    },
+    {
+      "epoch": 0.17296050735082155,
+      "grad_norm": 8.279263806480797,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "num_input_tokens_seen": 88476368,
+      "step": 600
+    },
+    {
+      "epoch": 0.17296050735082155,
+      "loss": 0.026282694190740585,
+      "loss_ce": 0.01596393808722496,
+      "loss_xval": 0.01031494140625,
+      "num_input_tokens_seen": 88476368,
+      "step": 600
+    },
+    {
+      "epoch": 0.17324877486307294,
+      "grad_norm": 15.192892311476902,
+      "learning_rate": 0.0001,
+      "loss": 0.031,
+      "num_input_tokens_seen": 88611096,
+      "step": 601
+    },
+    {
+      "epoch": 0.17324877486307294,
+      "loss": 0.036679480224847794,
+      "loss_ce": 0.023183079436421394,
+      "loss_xval": 0.01348876953125,
+      "num_input_tokens_seen": 88611096,
+      "step": 601
+    },
+    {
+      "epoch": 0.1735370423753243,
+      "grad_norm": 4.358761804382471,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "num_input_tokens_seen": 88746200,
+      "step": 602
+    },
+    {
+      "epoch": 0.1735370423753243,
+      "loss": 0.014774775132536888,
+      "loss_ce": 0.008205866441130638,
+      "loss_xval": 0.006561279296875,
+      "num_input_tokens_seen": 88746200,
+      "step": 602
+    },
+    {
+      "epoch": 0.17382530988757566,
+      "grad_norm": 20.32798700944008,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "num_input_tokens_seen": 88918640,
+      "step": 603
+    },
+    {
+      "epoch": 0.17382530988757566,
+      "loss": 0.03068140149116516,
+      "loss_ce": 0.014491826295852661,
+      "loss_xval": 0.0162353515625,
+      "num_input_tokens_seen": 88918640,
+      "step": 603
+    },
+    {
+      "epoch": 0.17411357739982705,
+      "grad_norm": 1.549470192136012,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "num_input_tokens_seen": 89053512,
+      "step": 604
+    },
+    {
+      "epoch": 0.17411357739982705,
+      "loss": 0.03184882178902626,
+      "loss_ce": 0.023847492411732674,
+      "loss_xval": 0.00799560546875,
+      "num_input_tokens_seen": 89053512,
+      "step": 604
+    },
+    {
+      "epoch": 0.1744018449120784,
+      "grad_norm": 22.991785780271417,
+      "learning_rate": 0.0001,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 89188560,
+      "step": 605
+    },
+    {
+      "epoch": 0.1744018449120784,
+      "loss": 0.025367528200149536,
+      "loss_ce": 0.006500035524368286,
+      "loss_xval": 0.0189208984375,
+      "num_input_tokens_seen": 89188560,
+      "step": 605
+    },
+    {
+      "epoch": 0.17469011242432977,
+      "grad_norm": 10.869462051705277,
+      "learning_rate": 0.0001,
+      "loss": 0.0263,
+      "num_input_tokens_seen": 89361072,
+      "step": 606
+    },
+    {
+      "epoch": 0.17469011242432977,
+      "loss": 0.023846114054322243,
+      "loss_ce": 0.013390026986598969,
+      "loss_xval": 0.01043701171875,
+      "num_input_tokens_seen": 89361072,
+      "step": 606
+    },
+    {
+      "epoch": 0.17497837993658114,
+      "grad_norm": 23.37286302176032,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "num_input_tokens_seen": 89495952,
+      "step": 607
+    },
+    {
+      "epoch": 0.17497837993658114,
+      "loss": 0.03683967888355255,
+      "loss_ce": 0.0210620928555727,
+      "loss_xval": 0.0157470703125,
+      "num_input_tokens_seen": 89495952,
+      "step": 607
+    },
+    {
+      "epoch": 0.17526664744883252,
+      "grad_norm": 14.326681445640265,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "num_input_tokens_seen": 89630984,
+      "step": 608
+    },
+    {
+      "epoch": 0.17526664744883252,
+      "loss": 0.012337716296315193,
+      "loss_ce": 0.004445107653737068,
+      "loss_xval": 0.00787353515625,
+      "num_input_tokens_seen": 89630984,
+      "step": 608
+    },
+    {
+      "epoch": 0.17555491496108389,
+      "grad_norm": 26.295209187947552,
+      "learning_rate": 0.0001,
+      "loss": 0.041,
+      "num_input_tokens_seen": 89803488,
+      "step": 609
+    },
+    {
+      "epoch": 0.17555491496108389,
+      "loss": 0.03529471904039383,
+      "loss_ce": 0.013344951905310154,
+      "loss_xval": 0.02197265625,
+      "num_input_tokens_seen": 89803488,
+      "step": 609
+    },
+    {
+      "epoch": 0.17584318247333525,
+      "grad_norm": 30.32216800331154,
+      "learning_rate": 0.0001,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 89938248,
+      "step": 610
+    },
+    {
+      "epoch": 0.17584318247333525,
+      "loss": 0.0387270450592041,
+      "loss_ce": 0.020073173567652702,
+      "loss_xval": 0.0186767578125,
+      "num_input_tokens_seen": 89938248,
+      "step": 610
+    },
+    {
+      "epoch": 0.17613144998558664,
+      "grad_norm": 16.927809740603625,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "num_input_tokens_seen": 90073272,
+      "step": 611
+    },
+    {
+      "epoch": 0.17613144998558664,
+      "loss": 0.024793002754449844,
+      "loss_ce": 0.00980124156922102,
+      "loss_xval": 0.0150146484375,
+      "num_input_tokens_seen": 90073272,
+      "step": 611
+    },
+    {
+      "epoch": 0.176419717497838,
+      "grad_norm": 43.083889005233594,
+      "learning_rate": 0.0001,
+      "loss": 0.0792,
+      "num_input_tokens_seen": 90245760,
+      "step": 612
+    },
+    {
+      "epoch": 0.176419717497838,
+      "loss": 0.05786873772740364,
+      "loss_ce": 0.01784493774175644,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 90245760,
+      "step": 612
+    },
+    {
+      "epoch": 0.17670798501008936,
+      "grad_norm": 4.502134202248031,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "num_input_tokens_seen": 90380496,
+      "step": 613
+    },
+    {
+      "epoch": 0.17670798501008936,
+      "loss": 0.03407268226146698,
+      "loss_ce": 0.026391787454485893,
+      "loss_xval": 0.0076904296875,
+      "num_input_tokens_seen": 90380496,
+      "step": 613
+    },
+    {
+      "epoch": 0.17699625252234072,
+      "grad_norm": 46.63692973207371,
+      "learning_rate": 0.0001,
+      "loss": 0.0823,
+      "num_input_tokens_seen": 90515528,
+      "step": 614
+    },
+    {
+      "epoch": 0.17699625252234072,
+      "loss": 0.05757968872785568,
+      "loss_ce": 0.006340674124658108,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 90515528,
+      "step": 614
+    },
+    {
+      "epoch": 0.1772845200345921,
+      "grad_norm": 14.615420262559676,
+      "learning_rate": 0.0001,
+      "loss": 0.0273,
+      "num_input_tokens_seen": 90688016,
+      "step": 615
+    },
+    {
+      "epoch": 0.1772845200345921,
+      "loss": 0.02321012318134308,
+      "loss_ce": 0.010972573421895504,
+      "loss_xval": 0.01220703125,
+      "num_input_tokens_seen": 90688016,
+      "step": 615
+    },
+    {
+      "epoch": 0.17757278754684347,
+      "grad_norm": 55.01904078945244,
+      "learning_rate": 0.0001,
+      "loss": 0.1086,
+      "num_input_tokens_seen": 90822800,
+      "step": 616
+    },
+    {
+      "epoch": 0.17757278754684347,
+      "loss": 0.06659901142120361,
+      "loss_ce": 0.020364880561828613,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 90822800,
+      "step": 616
+    },
+    {
+      "epoch": 0.17786105505909483,
+      "grad_norm": 34.16231678059451,
+      "learning_rate": 0.0001,
+      "loss": 0.056,
+      "num_input_tokens_seen": 90956392,
+      "step": 617
+    },
+    {
+      "epoch": 0.17786105505909483,
+      "loss": 0.02499392069876194,
+      "loss_ce": 0.004318260587751865,
+      "loss_xval": 0.0206298828125,
+      "num_input_tokens_seen": 90956392,
+      "step": 617
+    },
+    {
+      "epoch": 0.17814932257134622,
+      "grad_norm": 59.454902458580015,
+      "learning_rate": 0.0001,
+      "loss": 0.1291,
+      "num_input_tokens_seen": 91128864,
+      "step": 618
+    },
+    {
+      "epoch": 0.17814932257134622,
+      "loss": 0.08348599821329117,
+      "loss_ce": 0.0071920487098395824,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 91128864,
+      "step": 618
+    },
+    {
+      "epoch": 0.17843759008359758,
+      "grad_norm": 76.31438217203026,
+      "learning_rate": 0.0001,
+      "loss": 0.2037,
+      "num_input_tokens_seen": 91263672,
+      "step": 619
+    },
+    {
+      "epoch": 0.17843759008359758,
+      "loss": 0.10991793125867844,
+      "loss_ce": 0.018182091414928436,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 91263672,
+      "step": 619
+    },
+    {
+      "epoch": 0.17872585759584894,
+      "grad_norm": 33.35197047290332,
+      "learning_rate": 0.0001,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 91398624,
+      "step": 620
+    },
+    {
+      "epoch": 0.17872585759584894,
+      "loss": 0.04731812700629234,
+      "loss_ce": 0.003342297160997987,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 91398624,
+      "step": 620
+    },
+    {
+      "epoch": 0.17901412510810033,
+      "grad_norm": 104.89909603419869,
+      "learning_rate": 0.0001,
+      "loss": 0.3838,
+      "num_input_tokens_seen": 91571032,
+      "step": 621
+    },
+    {
+      "epoch": 0.17901412510810033,
+      "loss": 0.25730788707733154,
+      "loss_ce": 0.008772751316428185,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 91571032,
+      "step": 621
+    },
+    {
+      "epoch": 0.1793023926203517,
+      "grad_norm": 2.5475132522315587,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "num_input_tokens_seen": 91705872,
+      "step": 622
+    },
+    {
+      "epoch": 0.1793023926203517,
+      "loss": 0.024016141891479492,
+      "loss_ce": 0.016756772994995117,
+      "loss_xval": 0.00726318359375,
+      "num_input_tokens_seen": 91705872,
+      "step": 622
+    },
+    {
+      "epoch": 0.17959066013260305,
+      "grad_norm": 112.72775966928485,
+      "learning_rate": 0.0001,
+      "loss": 0.4688,
+      "num_input_tokens_seen": 91841016,
+      "step": 623
+    },
+    {
+      "epoch": 0.17959066013260305,
+      "loss": 0.2900291681289673,
+      "loss_ce": 0.0035301523748785257,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 91841016,
+      "step": 623
+    },
+    {
+      "epoch": 0.17987892764485441,
+      "grad_norm": 29.55511489330518,
+      "learning_rate": 0.0001,
+      "loss": 0.0575,
+      "num_input_tokens_seen": 92013560,
+      "step": 624
+    },
+    {
+      "epoch": 0.17987892764485441,
+      "loss": 0.030054394155740738,
+      "loss_ce": 0.009149854071438313,
+      "loss_xval": 0.0208740234375,
+      "num_input_tokens_seen": 92013560,
+      "step": 624
+    },
+    {
+      "epoch": 0.1801671951571058,
+      "grad_norm": 137.37470881232548,
+      "learning_rate": 0.0001,
+      "loss": 0.7034,
+      "num_input_tokens_seen": 92148376,
+      "step": 625
+    },
+    {
+      "epoch": 0.1801671951571058,
+      "loss": 0.40648776292800903,
+      "loss_ce": 0.01659516617655754,
+      "loss_xval": 0.390625,
+      "num_input_tokens_seen": 92148376,
+      "step": 625
+    },
+    {
+      "epoch": 0.18045546266935716,
+      "grad_norm": 79.15124205085066,
+      "learning_rate": 0.0001,
+      "loss": 0.2883,
+      "num_input_tokens_seen": 92283352,
+      "step": 626
+    },
+    {
+      "epoch": 0.18045546266935716,
+      "loss": 0.09200652688741684,
+      "loss_ce": 0.003688656259328127,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 92283352,
+      "step": 626
+    },
+    {
+      "epoch": 0.18074373018160853,
+      "grad_norm": 135.12030726633364,
+      "learning_rate": 0.0001,
+      "loss": 0.6984,
+      "num_input_tokens_seen": 92455888,
+      "step": 627
+    },
+    {
+      "epoch": 0.18074373018160853,
+      "loss": 0.5769015550613403,
+      "loss_ce": 0.008542162366211414,
+      "loss_xval": 0.5703125,
+      "num_input_tokens_seen": 92455888,
+      "step": 627
+    },
+    {
+      "epoch": 0.18103199769385991,
+      "grad_norm": 129.47937214572832,
+      "learning_rate": 0.0001,
+      "loss": 0.6816,
+      "num_input_tokens_seen": 92590720,
+      "step": 628
+    },
+    {
+      "epoch": 0.18103199769385991,
+      "loss": 0.3277011215686798,
+      "loss_ce": 0.021304648369550705,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 92590720,
+      "step": 628
+    },
+    {
+      "epoch": 0.18132026520611128,
+      "grad_norm": 103.8174512889404,
+      "learning_rate": 0.0001,
+      "loss": 0.425,
+      "num_input_tokens_seen": 92725696,
+      "step": 629
+    },
+    {
+      "epoch": 0.18132026520611128,
+      "loss": 0.5176563262939453,
+      "loss_ce": 0.004228599369525909,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 92725696,
+      "step": 629
+    },
+    {
+      "epoch": 0.18160853271836264,
+      "grad_norm": 148.7103706126506,
+      "learning_rate": 0.0001,
+      "loss": 0.8761,
+      "num_input_tokens_seen": 92898232,
+      "step": 630
+    },
+    {
+      "epoch": 0.18160853271836264,
+      "loss": 0.683671236038208,
+      "loss_ce": 0.011307948268949986,
+      "loss_xval": 0.671875,
+      "num_input_tokens_seen": 92898232,
+      "step": 630
+    },
+    {
+      "epoch": 0.181896800230614,
+      "grad_norm": 81.09085148470248,
+      "learning_rate": 0.0001,
+      "loss": 0.2746,
+      "num_input_tokens_seen": 93032992,
+      "step": 631
+    },
+    {
+      "epoch": 0.181896800230614,
+      "loss": 0.3315165042877197,
+      "loss_ce": 0.021213779225945473,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 93032992,
+      "step": 631
+    },
+    {
+      "epoch": 0.1821850677428654,
+      "grad_norm": 134.49367353766172,
+      "learning_rate": 0.0001,
+      "loss": 0.7242,
+      "num_input_tokens_seen": 93167984,
+      "step": 632
+    },
+    {
+      "epoch": 0.1821850677428654,
+      "loss": 0.6176624298095703,
+      "loss_ce": 0.005846067331731319,
+      "loss_xval": 0.61328125,
+      "num_input_tokens_seen": 93167984,
+      "step": 632
+    },
+    {
+      "epoch": 0.18247333525511675,
+      "grad_norm": 82.28557695859297,
+      "learning_rate": 0.0001,
+      "loss": 0.295,
+      "num_input_tokens_seen": 93340568,
+      "step": 633
+    },
+    {
+      "epoch": 0.18247333525511675,
+      "loss": 0.32379481196403503,
+      "loss_ce": 0.011294808238744736,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 93340568,
+      "step": 633
+    },
+    {
+      "epoch": 0.1827616027673681,
+      "grad_norm": 115.07263488502592,
+      "learning_rate": 0.0001,
+      "loss": 0.5633,
+      "num_input_tokens_seen": 93475448,
+      "step": 634
+    },
+    {
+      "epoch": 0.1827616027673681,
+      "loss": 0.43479472398757935,
+      "loss_ce": 0.021464616060256958,
+      "loss_xval": 0.4140625,
+      "num_input_tokens_seen": 93475448,
+      "step": 634
+    },
+    {
+      "epoch": 0.1830498702796195,
+      "grad_norm": 86.6727154697341,
+      "learning_rate": 0.0001,
+      "loss": 0.3386,
+      "num_input_tokens_seen": 93610488,
+      "step": 635
+    },
+    {
+      "epoch": 0.1830498702796195,
+      "loss": 0.44108590483665466,
+      "loss_ce": 0.006027324125170708,
+      "loss_xval": 0.435546875,
+      "num_input_tokens_seen": 93610488,
+      "step": 635
+    },
+    {
+      "epoch": 0.18333813779187086,
+      "grad_norm": 96.04215122807649,
+      "learning_rate": 0.0001,
+      "loss": 0.3946,
+      "num_input_tokens_seen": 93782984,
+      "step": 636
+    },
+    {
+      "epoch": 0.18333813779187086,
+      "loss": 0.3574924170970917,
+      "loss_ce": 0.01032444927841425,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 93782984,
+      "step": 636
+    },
+    {
+      "epoch": 0.18362640530412222,
+      "grad_norm": 81.44290209799817,
+      "learning_rate": 0.0001,
+      "loss": 0.3073,
+      "num_input_tokens_seen": 93917776,
+      "step": 637
+    },
+    {
+      "epoch": 0.18362640530412222,
+      "loss": 0.3946782946586609,
+      "loss_ce": 0.01772518828511238,
+      "loss_xval": 0.376953125,
+      "num_input_tokens_seen": 93917776,
+      "step": 637
+    },
+    {
+      "epoch": 0.18391467281637358,
+      "grad_norm": 50.44860046713553,
+      "learning_rate": 0.0001,
+      "loss": 0.1365,
+      "num_input_tokens_seen": 94052704,
+      "step": 638
+    },
+    {
+      "epoch": 0.18391467281637358,
+      "loss": 0.17554911971092224,
+      "loss_ce": 0.007824521511793137,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 94052704,
+      "step": 638
+    },
+    {
+      "epoch": 0.18420294032862497,
+      "grad_norm": 76.33064807793744,
+      "learning_rate": 0.0001,
+      "loss": 0.2706,
+      "num_input_tokens_seen": 94225128,
+      "step": 639
+    },
+    {
+      "epoch": 0.18420294032862497,
+      "loss": 0.27219584584236145,
+      "loss_ce": 0.010232958011329174,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 94225128,
+      "step": 639
+    },
+    {
+      "epoch": 0.18449120784087633,
+      "grad_norm": 2.9693015176271773,
+      "learning_rate": 0.0001,
+      "loss": 0.028,
+      "num_input_tokens_seen": 94359936,
+      "step": 640
+    },
+    {
+      "epoch": 0.18449120784087633,
+      "loss": 0.029297899454832077,
+      "loss_ce": 0.018612932413816452,
+      "loss_xval": 0.01068115234375,
+      "num_input_tokens_seen": 94359936,
+      "step": 640
+    },
+    {
+      "epoch": 0.1847794753531277,
+      "grad_norm": 78.18709435660247,
+      "learning_rate": 0.0001,
+      "loss": 0.294,
+      "num_input_tokens_seen": 94495104,
+      "step": 641
+    },
+    {
+      "epoch": 0.1847794753531277,
+      "loss": 0.23611602187156677,
+      "loss_ce": 0.006379688158631325,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 94495104,
+      "step": 641
+    },
+    {
+      "epoch": 0.18506774286537908,
+      "grad_norm": 36.01208626464418,
+      "learning_rate": 0.0001,
+      "loss": 0.0792,
+      "num_input_tokens_seen": 94667656,
+      "step": 642
+    },
+    {
+      "epoch": 0.18506774286537908,
+      "loss": 0.084808848798275,
+      "loss_ce": 0.008514909073710442,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 94667656,
+      "step": 642
+    },
+    {
+      "epoch": 0.18535601037763044,
+      "grad_norm": 71.56368152016078,
+      "learning_rate": 0.0001,
+      "loss": 0.2428,
+      "num_input_tokens_seen": 94802504,
+      "step": 643
+    },
+    {
+      "epoch": 0.18535601037763044,
+      "loss": 0.23120847344398499,
+      "loss_ce": 0.01905028149485588,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 94802504,
+      "step": 643
+    },
+    {
+      "epoch": 0.1856442778898818,
+      "grad_norm": 48.96488819717949,
+      "learning_rate": 0.0001,
+      "loss": 0.1255,
+      "num_input_tokens_seen": 94937664,
+      "step": 644
+    },
+    {
+      "epoch": 0.1856442778898818,
+      "loss": 0.14206674695014954,
+      "loss_ce": 0.005836280062794685,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 94937664,
+      "step": 644
+    },
+    {
+      "epoch": 0.1859325454021332,
+      "grad_norm": 41.79536487740748,
+      "learning_rate": 0.0001,
+      "loss": 0.1032,
+      "num_input_tokens_seen": 95110192,
+      "step": 645
+    },
+    {
+      "epoch": 0.1859325454021332,
+      "loss": 0.10236231237649918,
+      "loss_ce": 0.007940929383039474,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 95110192,
+      "step": 645
+    },
+    {
+      "epoch": 0.18622081291438455,
+      "grad_norm": 55.58375506803601,
+      "learning_rate": 0.0001,
+      "loss": 0.1666,
+      "num_input_tokens_seen": 95245008,
+      "step": 646
+    },
+    {
+      "epoch": 0.18622081291438455,
+      "loss": 0.18501755595207214,
+      "loss_ce": 0.022175753489136696,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 95245008,
+      "step": 646
+    },
+    {
+      "epoch": 0.18650908042663591,
+      "grad_norm": 10.179949024257645,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "num_input_tokens_seen": 95380144,
+      "step": 647
+    },
+    {
+      "epoch": 0.18650908042663591,
+      "loss": 0.020446889102458954,
+      "loss_ce": 0.0061799222603440285,
+      "loss_xval": 0.0142822265625,
+      "num_input_tokens_seen": 95380144,
+      "step": 647
+    },
+    {
+      "epoch": 0.18679734793888728,
+      "grad_norm": 57.440270290522434,
+      "learning_rate": 0.0001,
+      "loss": 0.1697,
+      "num_input_tokens_seen": 95552656,
+      "step": 648
+    },
+    {
+      "epoch": 0.18679734793888728,
+      "loss": 0.1528579443693161,
+      "loss_ce": 0.011012241244316101,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 95552656,
+      "step": 648
+    },
+    {
+      "epoch": 0.18708561545113866,
+      "grad_norm": 15.56417725902896,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 95687496,
+      "step": 649
+    },
+    {
+      "epoch": 0.18708561545113866,
+      "loss": 0.04545750468969345,
+      "loss_ce": 0.01956333965063095,
+      "loss_xval": 0.02587890625,
+      "num_input_tokens_seen": 95687496,
+      "step": 649
+    },
+    {
+      "epoch": 0.18737388296339003,
+      "grad_norm": 47.34908410562357,
+      "learning_rate": 0.0001,
+      "loss": 0.1196,
+      "num_input_tokens_seen": 95822704,
+      "step": 650
+    },
+    {
+      "epoch": 0.18737388296339003,
+      "loss": 0.12299247086048126,
+      "loss_ce": 0.008124311454594135,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 95822704,
+      "step": 650
+    },
+    {
+      "epoch": 0.1876621504756414,
+      "grad_norm": 30.048830082668516,
+      "learning_rate": 0.0001,
+      "loss": 0.0648,
+      "num_input_tokens_seen": 95995240,
+      "step": 651
+    },
+    {
+      "epoch": 0.1876621504756414,
+      "loss": 0.05971190333366394,
+      "loss_ce": 0.01140257716178894,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 95995240,
+      "step": 651
+    },
+    {
+      "epoch": 0.18795041798789278,
+      "grad_norm": 31.16166204897759,
+      "learning_rate": 0.0001,
+      "loss": 0.0639,
+      "num_input_tokens_seen": 96130024,
+      "step": 652
+    },
+    {
+      "epoch": 0.18795041798789278,
+      "loss": 0.07345931231975555,
+      "loss_ce": 0.017398525029420853,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 96130024,
+      "step": 652
+    },
+    {
+      "epoch": 0.18823868550014414,
+      "grad_norm": 36.79897939166327,
+      "learning_rate": 0.0001,
+      "loss": 0.0791,
+      "num_input_tokens_seen": 96265224,
+      "step": 653
+    },
+    {
+      "epoch": 0.18823868550014414,
+      "loss": 0.07308580726385117,
+      "loss_ce": 0.0050010960549116135,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 96265224,
+      "step": 653
+    },
+    {
+      "epoch": 0.1885269530123955,
+      "grad_norm": 10.089448574264999,
+      "learning_rate": 0.0001,
+      "loss": 0.0271,
+      "num_input_tokens_seen": 96437712,
+      "step": 654
+    },
+    {
+      "epoch": 0.1885269530123955,
+      "loss": 0.018405906856060028,
+      "loss_ce": 0.008525840938091278,
+      "loss_xval": 0.0098876953125,
+      "num_input_tokens_seen": 96437712,
+      "step": 654
+    },
+    {
+      "epoch": 0.18881522052464686,
+      "grad_norm": 40.4976776327298,
+      "learning_rate": 0.0001,
+      "loss": 0.0954,
+      "num_input_tokens_seen": 96572520,
+      "step": 655
+    },
+    {
+      "epoch": 0.18881522052464686,
+      "loss": 0.09976939857006073,
+      "loss_ce": 0.01621226966381073,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 96572520,
+      "step": 655
+    },
+    {
+      "epoch": 0.18910348803689825,
+      "grad_norm": 6.340013313622405,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "num_input_tokens_seen": 96707664,
+      "step": 656
+    },
+    {
+      "epoch": 0.18910348803689825,
+      "loss": 0.016466163098812103,
+      "loss_ce": 0.005746862851083279,
+      "loss_xval": 0.0107421875,
+      "num_input_tokens_seen": 96707664,
+      "step": 656
+    },
+    {
+      "epoch": 0.1893917555491496,
+      "grad_norm": 35.91003124190502,
+      "learning_rate": 0.0001,
+      "loss": 0.0812,
+      "num_input_tokens_seen": 96880240,
+      "step": 657
+    },
+    {
+      "epoch": 0.1893917555491496,
+      "loss": 0.07167474180459976,
+      "loss_ce": 0.008564389310777187,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 96880240,
+      "step": 657
+    },
+    {
+      "epoch": 0.18968002306140097,
+      "grad_norm": 15.718505266645101,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "num_input_tokens_seen": 97014944,
+      "step": 658
+    },
+    {
+      "epoch": 0.18968002306140097,
+      "loss": 0.04226473718881607,
+      "loss_ce": 0.01769808493554592,
+      "loss_xval": 0.0245361328125,
+      "num_input_tokens_seen": 97014944,
+      "step": 658
+    },
+    {
+      "epoch": 0.18996829057365236,
+      "grad_norm": 24.304796089580112,
+      "learning_rate": 0.0001,
+      "loss": 0.0398,
+      "num_input_tokens_seen": 97150216,
+      "step": 659
+    },
+    {
+      "epoch": 0.18996829057365236,
+      "loss": 0.04164959490299225,
+      "loss_ce": 0.003929869271814823,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 97150216,
+      "step": 659
+    },
+    {
+      "epoch": 0.19025655808590372,
+      "grad_norm": 22.686082472452636,
+      "learning_rate": 0.0001,
+      "loss": 0.0456,
+      "num_input_tokens_seen": 97322824,
+      "step": 660
+    },
+    {
+      "epoch": 0.19025655808590372,
+      "loss": 0.041275255382061005,
+      "loss_ce": 0.013794178143143654,
+      "loss_xval": 0.0274658203125,
+      "num_input_tokens_seen": 97322824,
+      "step": 660
+    },
+    {
+      "epoch": 0.19054482559815508,
+      "grad_norm": 10.923850233615978,
+      "learning_rate": 0.0001,
+      "loss": 0.0263,
+      "num_input_tokens_seen": 97457624,
+      "step": 661
+    },
+    {
+      "epoch": 0.19054482559815508,
+      "loss": 0.03429355472326279,
+      "loss_ce": 0.020759012550115585,
+      "loss_xval": 0.0135498046875,
+      "num_input_tokens_seen": 97457624,
+      "step": 661
+    },
+    {
+      "epoch": 0.19083309311040647,
+      "grad_norm": 24.784504672493853,
+      "learning_rate": 0.0001,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 97592712,
+      "step": 662
+    },
+    {
+      "epoch": 0.19083309311040647,
+      "loss": 0.038780368864536285,
+      "loss_ce": 0.00423447135835886,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 97592712,
+      "step": 662
+    },
+    {
+      "epoch": 0.19112136062265783,
+      "grad_norm": 1.9143655089211267,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "num_input_tokens_seen": 97765232,
+      "step": 663
+    },
+    {
+      "epoch": 0.19112136062265783,
+      "loss": 0.013931534253060818,
+      "loss_ce": 0.009016296826303005,
+      "loss_xval": 0.004913330078125,
+      "num_input_tokens_seen": 97765232,
+      "step": 663
+    },
+    {
+      "epoch": 0.1914096281349092,
+      "grad_norm": 25.36963085600519,
+      "learning_rate": 0.0001,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 97900040,
+      "step": 664
+    },
+    {
+      "epoch": 0.1914096281349092,
+      "loss": 0.05295349657535553,
+      "loss_ce": 0.01700379140675068,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 97900040,
+      "step": 664
+    },
+    {
+      "epoch": 0.19169789564716055,
+      "grad_norm": 7.826629884765334,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "num_input_tokens_seen": 98035160,
+      "step": 665
+    },
+    {
+      "epoch": 0.19169789564716055,
+      "loss": 0.012243002653121948,
+      "loss_ce": 0.003486365545541048,
+      "loss_xval": 0.00872802734375,
+      "num_input_tokens_seen": 98035160,
+      "step": 665
+    },
+    {
+      "epoch": 0.19198616315941194,
+      "grad_norm": 22.006637229215777,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "num_input_tokens_seen": 98207640,
+      "step": 666
+    },
+    {
+      "epoch": 0.19198616315941194,
+      "loss": 0.03680358827114105,
+      "loss_ce": 0.007033688947558403,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 98207640,
+      "step": 666
+    },
+    {
+      "epoch": 0.1922744306716633,
+      "grad_norm": 11.690800860739543,
+      "learning_rate": 0.0001,
+      "loss": 0.0231,
+      "num_input_tokens_seen": 98342408,
+      "step": 667
+    },
+    {
+      "epoch": 0.1922744306716633,
+      "loss": 0.02821066603064537,
+      "loss_ce": 0.01575949415564537,
+      "loss_xval": 0.012451171875,
+      "num_input_tokens_seen": 98342408,
+      "step": 667
+    },
+    {
+      "epoch": 0.19256269818391467,
+      "grad_norm": 16.620870830369185,
+      "learning_rate": 0.0001,
+      "loss": 0.0225,
+      "num_input_tokens_seen": 98477512,
+      "step": 668
+    },
+    {
+      "epoch": 0.19256269818391467,
+      "loss": 0.02170976996421814,
+      "loss_ce": 0.0031550817657262087,
+      "loss_xval": 0.0185546875,
+      "num_input_tokens_seen": 98477512,
+      "step": 668
+    },
+    {
+      "epoch": 0.19285096569616605,
+      "grad_norm": 14.19915075241969,
+      "learning_rate": 0.0001,
+      "loss": 0.0252,
+      "num_input_tokens_seen": 98649920,
+      "step": 669
+    },
+    {
+      "epoch": 0.19285096569616605,
+      "loss": 0.016638725996017456,
+      "loss_ce": 0.005190819036215544,
+      "loss_xval": 0.011474609375,
+      "num_input_tokens_seen": 98649920,
+      "step": 669
+    },
+    {
+      "epoch": 0.19313923320841742,
+      "grad_norm": 11.114309643706601,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "num_input_tokens_seen": 98784720,
+      "step": 670
+    },
+    {
+      "epoch": 0.19313923320841742,
+      "loss": 0.028857605531811714,
+      "loss_ce": 0.016703978180885315,
+      "loss_xval": 0.01214599609375,
+      "num_input_tokens_seen": 98784720,
+      "step": 670
+    },
+    {
+      "epoch": 0.19342750072066878,
+      "grad_norm": 15.49157272490426,
+      "learning_rate": 0.0001,
+      "loss": 0.0214,
+      "num_input_tokens_seen": 98919832,
+      "step": 671
+    },
+    {
+      "epoch": 0.19342750072066878,
+      "loss": 0.020468585193157196,
+      "loss_ce": 0.003065934870392084,
+      "loss_xval": 0.0174560546875,
+      "num_input_tokens_seen": 98919832,
+      "step": 671
+    },
+    {
+      "epoch": 0.19371576823292014,
+      "grad_norm": 6.639652599123291,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "num_input_tokens_seen": 99092272,
+      "step": 672
+    },
+    {
+      "epoch": 0.19371576823292014,
+      "loss": 0.021392960101366043,
+      "loss_ce": 0.014747758395969868,
+      "loss_xval": 0.00665283203125,
+      "num_input_tokens_seen": 99092272,
+      "step": 672
+    },
+    {
+      "epoch": 0.19400403574517153,
+      "grad_norm": 15.668890241160856,
+      "learning_rate": 0.0001,
+      "loss": 0.0259,
+      "num_input_tokens_seen": 99227176,
+      "step": 673
+    },
+    {
+      "epoch": 0.19400403574517153,
+      "loss": 0.03154763579368591,
+      "loss_ce": 0.013893214985728264,
+      "loss_xval": 0.0177001953125,
+      "num_input_tokens_seen": 99227176,
+      "step": 673
+    },
+    {
+      "epoch": 0.1942923032574229,
+      "grad_norm": 4.130485453269547,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "num_input_tokens_seen": 99362336,
+      "step": 674
+    },
+    {
+      "epoch": 0.1942923032574229,
+      "loss": 0.009054271504282951,
+      "loss_ce": 0.0033589282538741827,
+      "loss_xval": 0.005706787109375,
+      "num_input_tokens_seen": 99362336,
+      "step": 674
+    },
+    {
+      "epoch": 0.19458057076967425,
+      "grad_norm": 12.946791305625748,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "num_input_tokens_seen": 99534984,
+      "step": 675
+    },
+    {
+      "epoch": 0.19458057076967425,
+      "loss": 0.01664827950298786,
+      "loss_ce": 0.004082666710019112,
+      "loss_xval": 0.0125732421875,
+      "num_input_tokens_seen": 99534984,
+      "step": 675
+    },
+    {
+      "epoch": 0.19486883828192564,
+      "grad_norm": 2.015939391342304,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "num_input_tokens_seen": 99669744,
+      "step": 676
+    },
+    {
+      "epoch": 0.19486883828192564,
+      "loss": 0.01962101459503174,
+      "loss_ce": 0.015081525780260563,
+      "loss_xval": 0.004547119140625,
+      "num_input_tokens_seen": 99669744,
+      "step": 676
+    },
+    {
+      "epoch": 0.195157105794177,
+      "grad_norm": 10.133951813447016,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "num_input_tokens_seen": 99804800,
+      "step": 677
+    },
+    {
+      "epoch": 0.195157105794177,
+      "loss": 0.012844789773225784,
+      "loss_ce": 0.0024726272094994783,
+      "loss_xval": 0.0103759765625,
+      "num_input_tokens_seen": 99804800,
+      "step": 677
+    },
+    {
+      "epoch": 0.19544537330642836,
+      "grad_norm": 0.7346199890040997,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "num_input_tokens_seen": 99977304,
+      "step": 678
+    },
+    {
+      "epoch": 0.19544537330642836,
+      "loss": 0.011482784524559975,
+      "loss_ce": 0.008564542047679424,
+      "loss_xval": 0.0029144287109375,
+      "num_input_tokens_seen": 99977304,
+      "step": 678
+    },
+    {
+      "epoch": 0.19573364081867972,
+      "grad_norm": 8.687525798170709,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "num_input_tokens_seen": 100112096,
+      "step": 679
+    },
+    {
+      "epoch": 0.19573364081867972,
+      "loss": 0.022103361785411835,
+      "loss_ce": 0.01443963497877121,
+      "loss_xval": 0.007659912109375,
+      "num_input_tokens_seen": 100112096,
+      "step": 679
+    },
+    {
+      "epoch": 0.1960219083309311,
+      "grad_norm": 0.622549468221611,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "num_input_tokens_seen": 100247088,
+      "step": 680
+    },
+    {
+      "epoch": 0.1960219083309311,
+      "loss": 0.006348044611513615,
+      "loss_ce": 0.0022729947231709957,
+      "loss_xval": 0.00408935546875,
+      "num_input_tokens_seen": 100247088,
+      "step": 680
+    },
+    {
+      "epoch": 0.19631017584318247,
+      "grad_norm": 7.192674512777773,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "num_input_tokens_seen": 100419584,
+      "step": 681
+    },
+    {
+      "epoch": 0.19631017584318247,
+      "loss": 0.009196193888783455,
+      "loss_ce": 0.00399676151573658,
+      "loss_xval": 0.00518798828125,
+      "num_input_tokens_seen": 100419584,
+      "step": 681
+    },
+    {
+      "epoch": 0.19659844335543383,
+      "grad_norm": 0.7655775148366553,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "num_input_tokens_seen": 100554328,
+      "step": 682
+    },
+    {
+      "epoch": 0.19659844335543383,
+      "loss": 0.01690344139933586,
+      "loss_ce": 0.012867492623627186,
+      "loss_xval": 0.0040283203125,
+      "num_input_tokens_seen": 100554328,
+      "step": 682
+    },
+    {
+      "epoch": 0.19688671086768522,
+      "grad_norm": 7.242367892671911,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "num_input_tokens_seen": 100689416,
+      "step": 683
+    },
+    {
+      "epoch": 0.19688671086768522,
+      "loss": 0.009902188554406166,
+      "loss_ce": 0.002890774980187416,
+      "loss_xval": 0.00701904296875,
+      "num_input_tokens_seen": 100689416,
+      "step": 683
+    },
+    {
+      "epoch": 0.19717497837993658,
+      "grad_norm": 1.2326886099093166,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "num_input_tokens_seen": 100861864,
+      "step": 684
+    },
+    {
+      "epoch": 0.19717497837993658,
+      "loss": 0.006663970649242401,
+      "loss_ce": 0.0037590786814689636,
+      "loss_xval": 0.002899169921875,
+      "num_input_tokens_seen": 100861864,
+      "step": 684
+    },
+    {
+      "epoch": 0.19746324589218794,
+      "grad_norm": 6.2741974114302375,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "num_input_tokens_seen": 100996592,
+      "step": 685
+    },
+    {
+      "epoch": 0.19746324589218794,
+      "loss": 0.01968025602400303,
+      "loss_ce": 0.013534778729081154,
+      "loss_xval": 0.006134033203125,
+      "num_input_tokens_seen": 100996592,
+      "step": 685
+    },
+    {
+      "epoch": 0.19775151340443933,
+      "grad_norm": 2.38411098441469,
+      "learning_rate": 0.0001,
+      "loss": 0.0059,
+      "num_input_tokens_seen": 101131656,
+      "step": 686
+    },
+    {
+      "epoch": 0.19775151340443933,
+      "loss": 0.005984088871628046,
+      "loss_ce": 0.0022304265294224024,
+      "loss_xval": 0.003753662109375,
+      "num_input_tokens_seen": 101131656,
+      "step": 686
+    },
+    {
+      "epoch": 0.1980397809166907,
+      "grad_norm": 4.6953093680133104,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "num_input_tokens_seen": 101304152,
+      "step": 687
+    },
+    {
+      "epoch": 0.1980397809166907,
+      "loss": 0.012748443521559238,
+      "loss_ce": 0.009401046670973301,
+      "loss_xval": 0.0033416748046875,
+      "num_input_tokens_seen": 101304152,
+      "step": 687
+    },
+    {
+      "epoch": 0.19832804842894206,
+      "grad_norm": 2.421385194885947,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "num_input_tokens_seen": 101439032,
+      "step": 688
+    },
+    {
+      "epoch": 0.19832804842894206,
+      "loss": 0.0180523619055748,
+      "loss_ce": 0.014428399503231049,
+      "loss_xval": 0.003631591796875,
+      "num_input_tokens_seen": 101439032,
+      "step": 688
+    },
+    {
+      "epoch": 0.19861631594119342,
+      "grad_norm": 4.6660438919960985,
+      "learning_rate": 0.0001,
+      "loss": 0.0062,
+      "num_input_tokens_seen": 101574136,
+      "step": 689
+    },
+    {
+      "epoch": 0.19861631594119342,
+      "loss": 0.0060625383630394936,
+      "loss_ce": 0.0021047904156148434,
+      "loss_xval": 0.00396728515625,
+      "num_input_tokens_seen": 101574136,
+      "step": 689
+    },
+    {
+      "epoch": 0.1989045834534448,
+      "grad_norm": 4.1145697211269425,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "num_input_tokens_seen": 101746584,
+      "step": 690
+    },
+    {
+      "epoch": 0.1989045834534448,
+      "loss": 0.007049012929201126,
+      "loss_ce": 0.003686357755213976,
+      "loss_xval": 0.00335693359375,
+      "num_input_tokens_seen": 101746584,
+      "step": 690
+    },
+    {
+      "epoch": 0.19919285096569617,
+      "grad_norm": 3.2637992516048917,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "num_input_tokens_seen": 101881448,
+      "step": 691
+    },
+    {
+      "epoch": 0.19919285096569617,
+      "loss": 0.019784413278102875,
+      "loss_ce": 0.01585432142019272,
+      "loss_xval": 0.003936767578125,
+      "num_input_tokens_seen": 101881448,
+      "step": 691
+    },
+    {
+      "epoch": 0.19948111847794753,
+      "grad_norm": 4.695312015427915,
+      "learning_rate": 0.0001,
+      "loss": 0.0054,
+      "num_input_tokens_seen": 102016520,
+      "step": 692
+    },
+    {
+      "epoch": 0.19948111847794753,
+      "loss": 0.005383410956710577,
+      "loss_ce": 0.0017766145756468177,
+      "loss_xval": 0.00360107421875,
+      "num_input_tokens_seen": 102016520,
+      "step": 692
+    },
+    {
+      "epoch": 0.19976938599019892,
+      "grad_norm": 1.9018232659248924,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "num_input_tokens_seen": 102189000,
+      "step": 693
+    },
+    {
+      "epoch": 0.19976938599019892,
+      "loss": 0.006288346368819475,
+      "loss_ce": 0.0036495295353233814,
+      "loss_xval": 0.0026397705078125,
+      "num_input_tokens_seen": 102189000,
+      "step": 693
+    },
+    {
+      "epoch": 0.20005765350245028,
+      "grad_norm": 5.098530938515272,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "num_input_tokens_seen": 102323736,
+      "step": 694
+    },
+    {
+      "epoch": 0.20005765350245028,
+      "loss": 0.016563966870307922,
+      "loss_ce": 0.012800769880414009,
+      "loss_xval": 0.0037689208984375,
+      "num_input_tokens_seen": 102323736,
+      "step": 694
+    },
+    {
+      "epoch": 0.20034592101470164,
+      "grad_norm": 0.43345396702123434,
+      "learning_rate": 0.0001,
+      "loss": 0.0052,
+      "num_input_tokens_seen": 102458768,
+      "step": 695
+    },
+    {
+      "epoch": 0.20034592101470164,
+      "loss": 0.004343533888459206,
+      "loss_ce": 0.0017161609139293432,
+      "loss_xval": 0.00262451171875,
+      "num_input_tokens_seen": 102458768,
+      "step": 695
+    },
+    {
+      "epoch": 0.200634188526953,
+      "grad_norm": 5.545324349512925,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "num_input_tokens_seen": 102631224,
+      "step": 696
+    },
+    {
+      "epoch": 0.200634188526953,
+      "loss": 0.007838984951376915,
+      "loss_ce": 0.004875918850302696,
+      "loss_xval": 0.002960205078125,
+      "num_input_tokens_seen": 102631224,
+      "step": 696
+    },
+    {
+      "epoch": 0.2009224560392044,
+      "grad_norm": 2.378620571444512,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "num_input_tokens_seen": 102766000,
+      "step": 697
+    },
+    {
+      "epoch": 0.2009224560392044,
+      "loss": 0.016783371567726135,
+      "loss_ce": 0.014165535569190979,
+      "loss_xval": 0.00262451171875,
+      "num_input_tokens_seen": 102766000,
+      "step": 697
+    },
+    {
+      "epoch": 0.20121072355145575,
+      "grad_norm": 4.147280176248775,
+      "learning_rate": 0.0001,
+      "loss": 0.0054,
+      "num_input_tokens_seen": 102901240,
+      "step": 698
+    },
+    {
+      "epoch": 0.20121072355145575,
+      "loss": 0.0046518342569470406,
+      "loss_ce": 0.0016077059553936124,
+      "loss_xval": 0.0030517578125,
+      "num_input_tokens_seen": 102901240,
+      "step": 698
+    },
+    {
+      "epoch": 0.2014989910637071,
+      "grad_norm": 4.334942145827397,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "num_input_tokens_seen": 103073640,
+      "step": 699
+    },
+    {
+      "epoch": 0.2014989910637071,
+      "loss": 0.008360013365745544,
+      "loss_ce": 0.005029782652854919,
+      "loss_xval": 0.003326416015625,
+      "num_input_tokens_seen": 103073640,
+      "step": 699
+    },
+    {
+      "epoch": 0.2017872585759585,
+      "grad_norm": 2.053810403349566,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "num_input_tokens_seen": 103208504,
+      "step": 700
+    },
+    {
+      "epoch": 0.2017872585759585,
+      "loss": 0.018476512283086777,
+      "loss_ce": 0.015882518142461777,
+      "loss_xval": 0.002593994140625,
+      "num_input_tokens_seen": 103208504,
+      "step": 700
+    },
+    {
+      "epoch": 0.20207552608820986,
+      "grad_norm": 6.464624251825052,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "num_input_tokens_seen": 103343472,
+      "step": 701
+    },
+    {
+      "epoch": 0.20207552608820986,
+      "loss": 0.006439526565372944,
+      "loss_ce": 0.0027354557532817125,
+      "loss_xval": 0.0037078857421875,
+      "num_input_tokens_seen": 103343472,
+      "step": 701
+    },
+    {
+      "epoch": 0.20236379360046122,
+      "grad_norm": 2.7451701528896693,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "num_input_tokens_seen": 103515840,
+      "step": 702
+    },
+    {
+      "epoch": 0.20236379360046122,
+      "loss": 0.007059911731630564,
+      "loss_ce": 0.004708150401711464,
+      "loss_xval": 0.002349853515625,
+      "num_input_tokens_seen": 103515840,
+      "step": 702
+    },
+    {
+      "epoch": 0.20265206111271258,
+      "grad_norm": 3.7432213865383077,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "num_input_tokens_seen": 103650608,
+      "step": 703
+    },
+    {
+      "epoch": 0.20265206111271258,
+      "loss": 0.016106905415654182,
+      "loss_ce": 0.01323634572327137,
+      "loss_xval": 0.00286865234375,
+      "num_input_tokens_seen": 103650608,
+      "step": 703
+    },
+    {
+      "epoch": 0.20294032862496397,
+      "grad_norm": 5.118536747197136,
+      "learning_rate": 0.0001,
+      "loss": 0.0066,
+      "num_input_tokens_seen": 103785736,
+      "step": 704
+    },
+    {
+      "epoch": 0.20294032862496397,
+      "loss": 0.0067821950651705265,
+      "loss_ce": 0.00405754754319787,
+      "loss_xval": 0.0027313232421875,
+      "num_input_tokens_seen": 103785736,
+      "step": 704
+    },
+    {
+      "epoch": 0.20322859613721533,
+      "grad_norm": 0.7402285579455906,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "num_input_tokens_seen": 103958312,
+      "step": 705
+    },
+    {
+      "epoch": 0.20322859613721533,
+      "loss": 0.004566242918372154,
+      "loss_ce": 0.002834369894117117,
+      "loss_xval": 0.00173187255859375,
+      "num_input_tokens_seen": 103958312,
+      "step": 705
+    },
+    {
+      "epoch": 0.2035168636494667,
+      "grad_norm": 3.5297354964922336,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "num_input_tokens_seen": 104093096,
+      "step": 706
+    },
+    {
+      "epoch": 0.2035168636494667,
+      "loss": 0.018068354576826096,
+      "loss_ce": 0.015509648248553276,
+      "loss_xval": 0.0025634765625,
+      "num_input_tokens_seen": 104093096,
+      "step": 706
+    },
+    {
+      "epoch": 0.20380513116171808,
+      "grad_norm": 2.0238108681732356,
+      "learning_rate": 0.0001,
+      "loss": 0.0037,
+      "num_input_tokens_seen": 104228072,
+      "step": 707
+    },
+    {
+      "epoch": 0.20380513116171808,
+      "loss": 0.0034702238626778126,
+      "loss_ce": 0.0015657362528145313,
+      "loss_xval": 0.0019073486328125,
+      "num_input_tokens_seen": 104228072,
+      "step": 707
+    },
+    {
+      "epoch": 0.20409339867396945,
+      "grad_norm": 2.8823502193143353,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "num_input_tokens_seen": 104400696,
+      "step": 708
+    },
+    {
+      "epoch": 0.20409339867396945,
+      "loss": 0.00361913931556046,
+      "loss_ce": 0.001668875222094357,
+      "loss_xval": 0.001953125,
+      "num_input_tokens_seen": 104400696,
+      "step": 708
+    },
+    {
+      "epoch": 0.2043816661862208,
+      "grad_norm": 3.995092893791365,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "num_input_tokens_seen": 104535552,
+      "step": 709
+    },
+    {
+      "epoch": 0.2043816661862208,
+      "loss": 0.014175212942063808,
+      "loss_ce": 0.011864460073411465,
+      "loss_xval": 0.0023040771484375,
+      "num_input_tokens_seen": 104535552,
+      "step": 709
+    },
+    {
+      "epoch": 0.2046699336984722,
+      "grad_norm": 0.9230410321679317,
+      "learning_rate": 0.0001,
+      "loss": 0.0063,
+      "num_input_tokens_seen": 104670800,
+      "step": 710
+    },
+    {
+      "epoch": 0.2046699336984722,
+      "loss": 0.0030780972447246313,
+      "loss_ce": 0.001169794937595725,
+      "loss_xval": 0.0019073486328125,
+      "num_input_tokens_seen": 104670800,
+      "step": 710
+    },
+    {
+      "epoch": 0.20495820121072356,
+      "grad_norm": 6.386372898528216,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "num_input_tokens_seen": 104843208,
+      "step": 711
+    },
+    {
+      "epoch": 0.20495820121072356,
+      "loss": 0.004922128282487392,
+      "loss_ce": 0.0021431210916489363,
+      "loss_xval": 0.002777099609375,
+      "num_input_tokens_seen": 104843208,
+      "step": 711
+    },
+    {
+      "epoch": 0.20524646872297492,
+      "grad_norm": 4.493493805742633,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "num_input_tokens_seen": 104977992,
+      "step": 712
+    },
+    {
+      "epoch": 0.20524646872297492,
+      "loss": 0.01235833391547203,
+      "loss_ce": 0.010562565177679062,
+      "loss_xval": 0.00179290771484375,
+      "num_input_tokens_seen": 104977992,
+      "step": 712
+    },
+    {
+      "epoch": 0.20553473623522628,
+      "grad_norm": 4.1652319135334634,
+      "learning_rate": 0.0001,
+      "loss": 0.006,
+      "num_input_tokens_seen": 105113024,
+      "step": 713
+    },
+    {
+      "epoch": 0.20553473623522628,
+      "loss": 0.007076140493154526,
+      "loss_ce": 0.004552717786282301,
+      "loss_xval": 0.0025177001953125,
+      "num_input_tokens_seen": 105113024,
+      "step": 713
+    },
+    {
+      "epoch": 0.20582300374747767,
+      "grad_norm": 9.277139468460556,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "num_input_tokens_seen": 105285544,
+      "step": 714
+    },
+    {
+      "epoch": 0.20582300374747767,
+      "loss": 0.012906506657600403,
+      "loss_ce": 0.007316067814826965,
+      "loss_xval": 0.005584716796875,
+      "num_input_tokens_seen": 105285544,
+      "step": 714
+    },
+    {
+      "epoch": 0.20611127125972903,
+      "grad_norm": 3.598033343833172,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "num_input_tokens_seen": 105420280,
+      "step": 715
+    },
+    {
+      "epoch": 0.20611127125972903,
+      "loss": 0.014619720168411732,
+      "loss_ce": 0.01264942903071642,
+      "loss_xval": 0.0019683837890625,
+      "num_input_tokens_seen": 105420280,
+      "step": 715
+    },
+    {
+      "epoch": 0.2063995387719804,
+      "grad_norm": 7.377418814876543,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "num_input_tokens_seen": 105555360,
+      "step": 716
+    },
+    {
+      "epoch": 0.2063995387719804,
+      "loss": 0.006381037645041943,
+      "loss_ce": 0.0024461778812110424,
+      "loss_xval": 0.003936767578125,
+      "num_input_tokens_seen": 105555360,
+      "step": 716
+    },
+    {
+      "epoch": 0.20668780628423178,
+      "grad_norm": 11.760060582895239,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "num_input_tokens_seen": 105728048,
+      "step": 717
+    },
+    {
+      "epoch": 0.20668780628423178,
+      "loss": 0.008494282141327858,
+      "loss_ce": 0.0015858651604503393,
+      "loss_xval": 0.00689697265625,
+      "num_input_tokens_seen": 105728048,
+      "step": 717
+    },
+    {
+      "epoch": 0.20697607379648314,
+      "grad_norm": 3.2649079419303106,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "num_input_tokens_seen": 105862864,
+      "step": 718
+    },
+    {
+      "epoch": 0.20697607379648314,
+      "loss": 0.015600248239934444,
+      "loss_ce": 0.013268514536321163,
+      "loss_xval": 0.0023345947265625,
+      "num_input_tokens_seen": 105862864,
+      "step": 718
+    },
+    {
+      "epoch": 0.2072643413087345,
+      "grad_norm": 8.78753059548886,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "num_input_tokens_seen": 105998024,
+      "step": 719
+    },
+    {
+      "epoch": 0.2072643413087345,
+      "loss": 0.007219268009066582,
+      "loss_ce": 0.0020961295813322067,
+      "loss_xval": 0.005126953125,
+      "num_input_tokens_seen": 105998024,
+      "step": 719
+    },
+    {
+      "epoch": 0.20755260882098586,
+      "grad_norm": 9.84300808250944,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "num_input_tokens_seen": 106170576,
+      "step": 720
+    },
+    {
+      "epoch": 0.20755260882098586,
+      "loss": 0.007366933859884739,
+      "loss_ce": 0.0020110984332859516,
+      "loss_xval": 0.00537109375,
+      "num_input_tokens_seen": 106170576,
+      "step": 720
+    },
+    {
+      "epoch": 0.20784087633323725,
+      "grad_norm": 1.824367661797582,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "num_input_tokens_seen": 106305344,
+      "step": 721
+    },
+    {
+      "epoch": 0.20784087633323725,
+      "loss": 0.014551606960594654,
+      "loss_ce": 0.013377157039940357,
+      "loss_xval": 0.0011749267578125,
+      "num_input_tokens_seen": 106305344,
+      "step": 721
+    },
+    {
+      "epoch": 0.2081291438454886,
+      "grad_norm": 11.832146455499752,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "num_input_tokens_seen": 106440480,
+      "step": 722
+    },
+    {
+      "epoch": 0.2081291438454886,
+      "loss": 0.006810145918279886,
+      "loss_ce": 0.001046138582751155,
+      "loss_xval": 0.005767822265625,
+      "num_input_tokens_seen": 106440480,
+      "step": 722
+    },
+    {
+      "epoch": 0.20841741135773997,
+      "grad_norm": 7.026477339332676,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "num_input_tokens_seen": 106612912,
+      "step": 723
+    },
+    {
+      "epoch": 0.20841741135773997,
+      "loss": 0.006631569936871529,
+      "loss_ce": 0.0018555691931396723,
+      "loss_xval": 0.0047607421875,
+      "num_input_tokens_seen": 106612912,
+      "step": 723
+    },
+    {
+      "epoch": 0.20870567886999136,
+      "grad_norm": 8.645951427184539,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "num_input_tokens_seen": 106747672,
+      "step": 724
+    },
+    {
+      "epoch": 0.20870567886999136,
+      "loss": 0.016665926203131676,
+      "loss_ce": 0.0138621237128973,
+      "loss_xval": 0.0028076171875,
+      "num_input_tokens_seen": 106747672,
+      "step": 724
+    },
+    {
+      "epoch": 0.20899394638224272,
+      "grad_norm": 16.810786634717683,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "num_input_tokens_seen": 106882776,
+      "step": 725
+    },
+    {
+      "epoch": 0.20899394638224272,
+      "loss": 0.011873760260641575,
+      "loss_ce": 0.0009560962789691985,
+      "loss_xval": 0.01092529296875,
+      "num_input_tokens_seen": 106882776,
+      "step": 725
+    },
+    {
+      "epoch": 0.20928221389449408,
+      "grad_norm": 6.357963151247145,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "num_input_tokens_seen": 107055424,
+      "step": 726
+    },
+    {
+      "epoch": 0.20928221389449408,
+      "loss": 0.008962000720202923,
+      "loss_ce": 0.0044415839947760105,
+      "loss_xval": 0.0045166015625,
+      "num_input_tokens_seen": 107055424,
+      "step": 726
+    },
+    {
+      "epoch": 0.20957048140674547,
+      "grad_norm": 12.00840382089868,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "num_input_tokens_seen": 107190240,
+      "step": 727
+    },
+    {
+      "epoch": 0.20957048140674547,
+      "loss": 0.018986305221915245,
+      "loss_ce": 0.014607032760977745,
+      "loss_xval": 0.00439453125,
+      "num_input_tokens_seen": 107190240,
+      "step": 727
+    },
+    {
+      "epoch": 0.20985874891899683,
+      "grad_norm": 16.77374168533911,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "num_input_tokens_seen": 107325304,
+      "step": 728
+    },
+    {
+      "epoch": 0.20985874891899683,
+      "loss": 0.013724341988563538,
+      "loss_ce": 0.0011892463080585003,
+      "loss_xval": 0.01251220703125,
+      "num_input_tokens_seen": 107325304,
+      "step": 728
+    },
+    {
+      "epoch": 0.2101470164312482,
+      "grad_norm": 0.7581546666650136,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "num_input_tokens_seen": 107497904,
+      "step": 729
+    },
+    {
+      "epoch": 0.2101470164312482,
+      "loss": 0.005003034602850676,
+      "loss_ce": 0.002600728999823332,
+      "loss_xval": 0.0023956298828125,
+      "num_input_tokens_seen": 107497904,
+      "step": 729
+    },
+    {
+      "epoch": 0.21043528394349956,
+      "grad_norm": 22.638176333350234,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "num_input_tokens_seen": 107632768,
+      "step": 730
+    },
+    {
+      "epoch": 0.21043528394349956,
+      "loss": 0.02648492529988289,
+      "loss_ce": 0.01167626865208149,
+      "loss_xval": 0.01483154296875,
+      "num_input_tokens_seen": 107632768,
+      "step": 730
+    },
+    {
+      "epoch": 0.21072355145575095,
+      "grad_norm": 24.982746535895913,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "num_input_tokens_seen": 107767784,
+      "step": 731
+    },
+    {
+      "epoch": 0.21072355145575095,
+      "loss": 0.031112845987081528,
+      "loss_ce": 0.001465020701289177,
+      "loss_xval": 0.0296630859375,
+      "num_input_tokens_seen": 107767784,
+      "step": 731
+    },
+    {
+      "epoch": 0.2110118189680023,
+      "grad_norm": 1.4820439364157452,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "num_input_tokens_seen": 107940288,
+      "step": 732
+    },
+    {
+      "epoch": 0.2110118189680023,
+      "loss": 0.004679238889366388,
+      "loss_ce": 0.0020709396339952946,
+      "loss_xval": 0.0026092529296875,
+      "num_input_tokens_seen": 107940288,
+      "step": 732
+    },
+    {
+      "epoch": 0.21130008648025367,
+      "grad_norm": 30.58091880038745,
+      "learning_rate": 0.0001,
+      "loss": 0.0558,
+      "num_input_tokens_seen": 108075104,
+      "step": 733
+    },
+    {
+      "epoch": 0.21130008648025367,
+      "loss": 0.041614316403865814,
+      "loss_ce": 0.014255308546125889,
+      "loss_xval": 0.02734375,
+      "num_input_tokens_seen": 108075104,
+      "step": 733
+    },
+    {
+      "epoch": 0.21158835399250506,
+      "grad_norm": 29.47286835944433,
+      "learning_rate": 0.0001,
+      "loss": 0.0444,
+      "num_input_tokens_seen": 108210048,
+      "step": 734
+    },
+    {
+      "epoch": 0.21158835399250506,
+      "loss": 0.0443190336227417,
+      "loss_ce": 0.0008620026055723429,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 108210048,
+      "step": 734
+    },
+    {
+      "epoch": 0.21187662150475642,
+      "grad_norm": 4.104779558726963,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "num_input_tokens_seen": 108382560,
+      "step": 735
+    },
+    {
+      "epoch": 0.21187662150475642,
+      "loss": 0.005712998099625111,
+      "loss_ce": 0.002589714713394642,
+      "loss_xval": 0.0031280517578125,
+      "num_input_tokens_seen": 108382560,
+      "step": 735
+    },
+    {
+      "epoch": 0.21216488901700778,
+      "grad_norm": 37.49739334499608,
+      "learning_rate": 0.0001,
+      "loss": 0.0792,
+      "num_input_tokens_seen": 108517344,
+      "step": 736
+    },
+    {
+      "epoch": 0.21216488901700778,
+      "loss": 0.05222557112574577,
+      "loss_ce": 0.012400129809975624,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 108517344,
+      "step": 736
+    },
+    {
+      "epoch": 0.21245315652925914,
+      "grad_norm": 34.181805068474205,
+      "learning_rate": 0.0001,
+      "loss": 0.0587,
+      "num_input_tokens_seen": 108652376,
+      "step": 737
+    },
+    {
+      "epoch": 0.21245315652925914,
+      "loss": 0.06710252165794373,
+      "loss_ce": 0.0008488551247864962,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 108652376,
+      "step": 737
+    },
+    {
+      "epoch": 0.21274142404151053,
+      "grad_norm": 7.407982752291219,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "num_input_tokens_seen": 108824816,
+      "step": 738
+    },
+    {
+      "epoch": 0.21274142404151053,
+      "loss": 0.004085796885192394,
+      "loss_ce": 0.001343029784038663,
+      "loss_xval": 0.00274658203125,
+      "num_input_tokens_seen": 108824816,
+      "step": 738
+    },
+    {
+      "epoch": 0.2130296915537619,
+      "grad_norm": 45.64979852027761,
+      "learning_rate": 0.0001,
+      "loss": 0.115,
+      "num_input_tokens_seen": 108959632,
+      "step": 739
+    },
+    {
+      "epoch": 0.2130296915537619,
+      "loss": 0.07691901922225952,
+      "loss_ce": 0.017562326043844223,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 108959632,
+      "step": 739
+    },
+    {
+      "epoch": 0.21331795906601325,
+      "grad_norm": 36.967704163197965,
+      "learning_rate": 0.0001,
+      "loss": 0.0698,
+      "num_input_tokens_seen": 109094712,
+      "step": 740
+    },
+    {
+      "epoch": 0.21331795906601325,
+      "loss": 0.08530056476593018,
+      "loss_ce": 0.0009499795269221067,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 109094712,
+      "step": 740
+    },
+    {
+      "epoch": 0.21360622657826464,
+      "grad_norm": 14.27081537855223,
+      "learning_rate": 0.0001,
+      "loss": 0.0316,
+      "num_input_tokens_seen": 109267256,
+      "step": 741
+    },
+    {
+      "epoch": 0.21360622657826464,
+      "loss": 0.0032704181503504515,
+      "loss_ce": 0.001323015196248889,
+      "loss_xval": 0.00194549560546875,
+      "num_input_tokens_seen": 109267256,
+      "step": 741
+    },
+    {
+      "epoch": 0.213894494090516,
+      "grad_norm": 57.42413597809985,
+      "learning_rate": 0.0001,
+      "loss": 0.1725,
+      "num_input_tokens_seen": 109401992,
+      "step": 742
+    },
+    {
+      "epoch": 0.213894494090516,
+      "loss": 0.1181582510471344,
+      "loss_ce": 0.011102594435214996,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 109401992,
+      "step": 742
+    },
+    {
+      "epoch": 0.21418276160276736,
+      "grad_norm": 43.418170426670386,
+      "learning_rate": 0.0001,
+      "loss": 0.098,
+      "num_input_tokens_seen": 109536952,
+      "step": 743
+    },
+    {
+      "epoch": 0.21418276160276736,
+      "loss": 0.12911713123321533,
+      "loss_ce": 0.0012484844774007797,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 109536952,
+      "step": 743
+    },
+    {
+      "epoch": 0.21447102911501872,
+      "grad_norm": 19.583171126203737,
+      "learning_rate": 0.0001,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 109709448,
+      "step": 744
+    },
+    {
+      "epoch": 0.21447102911501872,
+      "loss": 0.0038100657984614372,
+      "loss_ce": 0.0019217908848077059,
+      "loss_xval": 0.00189208984375,
+      "num_input_tokens_seen": 109709448,
+      "step": 744
+    },
+    {
+      "epoch": 0.2147592966272701,
+      "grad_norm": 67.0891843399185,
+      "learning_rate": 0.0001,
+      "loss": 0.2363,
+      "num_input_tokens_seen": 109844328,
+      "step": 745
+    },
+    {
+      "epoch": 0.2147592966272701,
+      "loss": 0.16546039283275604,
+      "loss_ce": 0.01458149217069149,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 109844328,
+      "step": 745
+    },
+    {
+      "epoch": 0.21504756413952147,
+      "grad_norm": 41.61436187130198,
+      "learning_rate": 0.0001,
+      "loss": 0.0962,
+      "num_input_tokens_seen": 109979464,
+      "step": 746
+    },
+    {
+      "epoch": 0.21504756413952147,
+      "loss": 0.1428573727607727,
+      "loss_ce": 0.002293400000780821,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 109979464,
+      "step": 746
+    },
+    {
+      "epoch": 0.21533583165177284,
+      "grad_norm": 35.76768484393859,
+      "learning_rate": 0.0001,
+      "loss": 0.1025,
+      "num_input_tokens_seen": 110152024,
+      "step": 747
+    },
+    {
+      "epoch": 0.21533583165177284,
+      "loss": 0.010296022519469261,
+      "loss_ce": 0.00212112651206553,
+      "loss_xval": 0.0081787109375,
+      "num_input_tokens_seen": 110152024,
+      "step": 747
+    },
+    {
+      "epoch": 0.21562409916402422,
+      "grad_norm": 81.41444916351666,
+      "learning_rate": 0.0001,
+      "loss": 0.3448,
+      "num_input_tokens_seen": 110286800,
+      "step": 748
+    },
+    {
+      "epoch": 0.21562409916402422,
+      "loss": 0.2511296272277832,
+      "loss_ce": 0.011261474341154099,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 110286800,
+      "step": 748
+    },
+    {
+      "epoch": 0.21591236667627559,
+      "grad_norm": 33.949280913156656,
+      "learning_rate": 0.0001,
+      "loss": 0.0738,
+      "num_input_tokens_seen": 110421864,
+      "step": 749
+    },
+    {
+      "epoch": 0.21591236667627559,
+      "loss": 0.12540201842784882,
+      "loss_ce": 0.0011344377417117357,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 110421864,
+      "step": 749
+    },
+    {
+      "epoch": 0.21620063418852695,
+      "grad_norm": 64.0371404773295,
+      "learning_rate": 0.0001,
+      "loss": 0.2511,
+      "num_input_tokens_seen": 110594344,
+      "step": 750
+    },
+    {
+      "epoch": 0.21620063418852695,
+      "eval_websight_new_IoU": 0.0,
+      "eval_websight_new_MAE_x": 0.8172395527362823,
+      "eval_websight_new_MAE_y": 0.8070051372051239,
+      "eval_websight_new_NUM_probability": 0.9898514747619629,
+      "eval_websight_new_inside_bbox": 0.0,
+      "eval_websight_new_loss": 0.6402856707572937,
+      "eval_websight_new_loss_ce": 0.0014999214326962829,
+      "eval_websight_new_loss_xval": 0.638671875,
+      "eval_websight_new_runtime": 35.9819,
+      "eval_websight_new_samples_per_second": 1.39,
+      "eval_websight_new_steps_per_second": 0.056,
+      "num_input_tokens_seen": 110594344,
+      "step": 750
+    },
+    {
+      "epoch": 0.21620063418852695,
+      "eval_seeclick_IoU": 0.0,
+      "eval_seeclick_MAE_x": 0.6540522873401642,
+      "eval_seeclick_MAE_y": 0.6515002846717834,
+      "eval_seeclick_NUM_probability": 0.9914124011993408,
+      "eval_seeclick_inside_bbox": 0.0,
+      "eval_seeclick_loss": 0.4462566375732422,
+      "eval_seeclick_loss_ce": 0.012402037624269724,
+      "eval_seeclick_loss_xval": 0.4342041015625,
+      "eval_seeclick_runtime": 64.6128,
+      "eval_seeclick_samples_per_second": 0.774,
+      "eval_seeclick_steps_per_second": 0.031,
+      "num_input_tokens_seen": 110594344,
+      "step": 750
+    },
+    {
+      "epoch": 0.21620063418852695,
+      "eval_icons_IoU": 0.0,
+      "eval_icons_MAE_x": 1.0583945512771606,
+      "eval_icons_MAE_y": 1.0435316562652588,
+      "eval_icons_NUM_probability": 0.9888060688972473,
+      "eval_icons_inside_bbox": 0.0,
+      "eval_icons_loss": 1.1147381067276,
+      "eval_icons_loss_ce": 0.0066304560750722885,
+      "eval_icons_loss_xval": 1.10546875,
+      "eval_icons_runtime": 65.9871,
+      "eval_icons_samples_per_second": 0.758,
+      "eval_icons_steps_per_second": 0.03,
+      "num_input_tokens_seen": 110594344,
+      "step": 750
+    },
+    {
+      "epoch": 0.21620063418852695,
+      "loss": 1.12251615524292,
+      "loss_ce": 0.008258317597210407,
+      "loss_xval": 1.1171875,
+      "num_input_tokens_seen": 110594344,
+      "step": 750
+    },
+    {
+      "epoch": 0.21648890170077834,
+      "grad_norm": 100.39901558775324,
+      "learning_rate": 0.0001,
+      "loss": 0.5285,
+      "num_input_tokens_seen": 110729120,
+      "step": 751
+    },
+    {
+      "epoch": 0.21648890170077834,
+      "loss": 0.433902382850647,
+      "loss_ce": 0.012515652924776077,
+      "loss_xval": 0.421875,
+      "num_input_tokens_seen": 110729120,
+      "step": 751
+    },
+    {
+      "epoch": 0.2167771692130297,
+      "grad_norm": 14.068410073467993,
+      "learning_rate": 0.0001,
+      "loss": 0.0389,
+      "num_input_tokens_seen": 110864456,
+      "step": 752
+    },
+    {
+      "epoch": 0.2167771692130297,
+      "loss": 0.07106943428516388,
+      "loss_ce": 0.0024048807099461555,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 110864456,
+      "step": 752
+    },
+    {
+      "epoch": 0.21706543672528106,
+      "grad_norm": 108.40390401257352,
+      "learning_rate": 0.0001,
+      "loss": 0.6725,
+      "num_input_tokens_seen": 111037032,
+      "step": 753
+    },
+    {
+      "epoch": 0.21706543672528106,
+      "loss": 0.3352723717689514,
+      "loss_ce": 0.0029969951137900352,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 111037032,
+      "step": 753
+    },
+    {
+      "epoch": 0.21735370423753242,
+      "grad_norm": 109.21261440688775,
+      "learning_rate": 0.0001,
+      "loss": 0.6504,
+      "num_input_tokens_seen": 111171856,
+      "step": 754
+    },
+    {
+      "epoch": 0.21735370423753242,
+      "loss": 0.5691009759902954,
+      "loss_ce": 0.011483820155262947,
+      "loss_xval": 0.55859375,
+      "num_input_tokens_seen": 111171856,
+      "step": 754
+    },
+    {
+      "epoch": 0.2176419717497838,
+      "grad_norm": 41.5565355210608,
+      "learning_rate": 0.0001,
+      "loss": 0.145,
+      "num_input_tokens_seen": 111306816,
+      "step": 755
+    },
+    {
+      "epoch": 0.2176419717497838,
+      "loss": 0.01940779760479927,
+      "loss_ce": 0.0028062332421541214,
+      "loss_xval": 0.0166015625,
+      "num_input_tokens_seen": 111306816,
+      "step": 755
+    },
+    {
+      "epoch": 0.21793023926203517,
+      "grad_norm": 162.7730201135358,
+      "learning_rate": 0.0001,
+      "loss": 1.5395,
+      "num_input_tokens_seen": 111479384,
+      "step": 756
+    },
+    {
+      "epoch": 0.21793023926203517,
+      "loss": 1.0829627513885498,
+      "loss_ce": 0.002884726971387863,
+      "loss_xval": 1.078125,
+      "num_input_tokens_seen": 111479384,
+      "step": 756
+    },
+    {
+      "epoch": 0.21821850677428653,
+      "grad_norm": 65.1372279307664,
+      "learning_rate": 0.0001,
+      "loss": 0.2507,
+      "num_input_tokens_seen": 111614168,
+      "step": 757
+    },
+    {
+      "epoch": 0.21821850677428653,
+      "loss": 0.2695873975753784,
+      "loss_ce": 0.01116455439478159,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 111614168,
+      "step": 757
+    },
+    {
+      "epoch": 0.21850677428653792,
+      "grad_norm": 145.76586500867106,
+      "learning_rate": 0.0001,
+      "loss": 1.1751,
+      "num_input_tokens_seen": 111749232,
+      "step": 758
+    },
+    {
+      "epoch": 0.21850677428653792,
+      "loss": 0.6525642275810242,
+      "loss_ce": 0.0026618903502821922,
+      "loss_xval": 0.6484375,
+      "num_input_tokens_seen": 111749232,
+      "step": 758
+    },
+    {
+      "epoch": 0.21879504179878928,
+      "grad_norm": 409.62868599068196,
+      "learning_rate": 0.0001,
+      "loss": 4.1763,
+      "num_input_tokens_seen": 111921744,
+      "step": 759
+    },
+    {
+      "epoch": 0.21879504179878928,
+      "loss": 3.8221874237060547,
+      "loss_ce": 0.011640791781246662,
+      "loss_xval": 3.8125,
+      "num_input_tokens_seen": 111921744,
+      "step": 759
+    },
+    {
+      "epoch": 0.21908330931104064,
+      "grad_norm": 135.94398972900095,
+      "learning_rate": 0.0001,
+      "loss": 0.9475,
+      "num_input_tokens_seen": 112056560,
+      "step": 760
+    },
+    {
+      "epoch": 0.21908330931104064,
+      "loss": 0.940850019454956,
+      "loss_ce": 0.3182913661003113,
+      "loss_xval": 0.62109375,
+      "num_input_tokens_seen": 112056560,
+      "step": 760
+    },
+    {
+      "epoch": 0.219371576823292,
+      "grad_norm": 738.893441811946,
+      "learning_rate": 0.0001,
+      "loss": 9.1575,
+      "num_input_tokens_seen": 112191688,
+      "step": 761
+    },
+    {
+      "epoch": 0.219371576823292,
+      "loss": 10.175888061523438,
+      "loss_ce": 0.6133874654769897,
+      "loss_xval": 9.5625,
+      "num_input_tokens_seen": 112191688,
+      "step": 761
+    },
+    {
+      "epoch": 0.2196598443355434,
+      "grad_norm": 832.3972251198659,
+      "learning_rate": 0.0001,
+      "loss": 33.4835,
+      "num_input_tokens_seen": 112364192,
+      "step": 762
+    },
+    {
+      "epoch": 0.2196598443355434,
+      "loss": 32.80980682373047,
+      "loss_ce": 0.3723085820674896,
+      "loss_xval": 32.5,
+      "num_input_tokens_seen": 112364192,
+      "step": 762
+    },
+    {
+      "epoch": 0.21994811184779475,
+      "grad_norm": 755.1576609542743,
+      "learning_rate": 0.0001,
+      "loss": 12.6196,
+      "num_input_tokens_seen": 112498976,
+      "step": 763
+    },
+    {
+      "epoch": 0.21994811184779475,
+      "loss": 11.760293960571289,
+      "loss_ce": 0.40091896057128906,
+      "loss_xval": 11.375,
+      "num_input_tokens_seen": 112498976,
+      "step": 763
+    },
+    {
+      "epoch": 0.22023637936004611,
+      "grad_norm": 138.2397155003218,
+      "learning_rate": 0.0001,
+      "loss": 0.9111,
+      "num_input_tokens_seen": 112634040,
+      "step": 764
+    },
+    {
+      "epoch": 0.22023637936004611,
+      "loss": 0.974545419216156,
+      "loss_ce": 0.2006196528673172,
+      "loss_xval": 0.7734375,
+      "num_input_tokens_seen": 112634040,
+      "step": 764
+    },
+    {
+      "epoch": 0.2205246468722975,
+      "grad_norm": 173.75011876445583,
+      "learning_rate": 0.0001,
+      "loss": 1.5548,
+      "num_input_tokens_seen": 112806632,
+      "step": 765
+    },
+    {
+      "epoch": 0.2205246468722975,
+      "loss": 1.4252033233642578,
+      "loss_ce": 0.2347736358642578,
+      "loss_xval": 1.1875,
+      "num_input_tokens_seen": 112806632,
+      "step": 765
+    },
+    {
+      "epoch": 0.22081291438454886,
+      "grad_norm": 80.24372426526006,
+      "learning_rate": 0.0001,
+      "loss": 0.4294,
+      "num_input_tokens_seen": 112941480,
+      "step": 766
+    },
+    {
+      "epoch": 0.22081291438454886,
+      "loss": 0.3662590980529785,
+      "loss_ce": 0.21873709559440613,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 112941480,
+      "step": 766
+    },
+    {
+      "epoch": 0.22110118189680023,
+      "grad_norm": 241.67489394610638,
+      "learning_rate": 0.0001,
+      "loss": 1.8493,
+      "num_input_tokens_seen": 113076600,
+      "step": 767
+    },
+    {
+      "epoch": 0.22110118189680023,
+      "loss": 1.8362443447113037,
+      "loss_ce": 0.2532365918159485,
+      "loss_xval": 1.5859375,
+      "num_input_tokens_seen": 113076600,
+      "step": 767
+    },
+    {
+      "epoch": 0.2213894494090516,
+      "grad_norm": 309.9244098220946,
+      "learning_rate": 0.0001,
+      "loss": 0.6891,
+      "num_input_tokens_seen": 113249144,
+      "step": 768
+    },
+    {
+      "epoch": 0.2213894494090516,
+      "loss": 0.7872165441513062,
+      "loss_ce": 0.7117770910263062,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 113249144,
+      "step": 768
+    },
+    {
+      "epoch": 0.22167771692130298,
+      "grad_norm": 159.45569658811266,
+      "learning_rate": 0.0001,
+      "loss": 0.7886,
+      "num_input_tokens_seen": 113383944,
+      "step": 769
+    },
+    {
+      "epoch": 0.22167771692130298,
+      "loss": 0.9091415405273438,
+      "loss_ce": 0.17281340062618256,
+      "loss_xval": 0.734375,
+      "num_input_tokens_seen": 113383944,
+      "step": 769
+    },
+    {
+      "epoch": 0.22196598443355434,
+      "grad_norm": 76.54296676716736,
+      "learning_rate": 0.0001,
+      "loss": 0.5255,
+      "num_input_tokens_seen": 113519064,
+      "step": 770
+    },
+    {
+      "epoch": 0.22196598443355434,
+      "loss": 0.4586948752403259,
+      "loss_ce": 0.14302101731300354,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 113519064,
+      "step": 770
+    },
+    {
+      "epoch": 0.2222542519458057,
+      "grad_norm": 33.441254496122355,
+      "learning_rate": 0.0001,
+      "loss": 0.2996,
+      "num_input_tokens_seen": 113691528,
+      "step": 771
+    },
+    {
+      "epoch": 0.2222542519458057,
+      "loss": 0.3153448700904846,
+      "loss_ce": 0.20883852243423462,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 113691528,
+      "step": 771
+    },
+    {
+      "epoch": 0.2225425194580571,
+      "grad_norm": 94.5394479291071,
+      "learning_rate": 0.0001,
+      "loss": 0.7111,
+      "num_input_tokens_seen": 113826248,
+      "step": 772
+    },
+    {
+      "epoch": 0.2225425194580571,
+      "loss": 0.6232205033302307,
+      "loss_ce": 0.1376248300075531,
+      "loss_xval": 0.486328125,
+      "num_input_tokens_seen": 113826248,
+      "step": 772
+    },
+    {
+      "epoch": 0.22283078697030845,
+      "grad_norm": 31.618367864103973,
+      "learning_rate": 0.0001,
+      "loss": 0.2197,
+      "num_input_tokens_seen": 113961288,
+      "step": 773
+    },
+    {
+      "epoch": 0.22283078697030845,
+      "loss": 0.20101110637187958,
+      "loss_ce": 0.10189002007246017,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 113961288,
+      "step": 773
+    },
+    {
+      "epoch": 0.2231190544825598,
+      "grad_norm": 79.57061276729071,
+      "learning_rate": 0.0001,
+      "loss": 0.5602,
+      "num_input_tokens_seen": 114133776,
+      "step": 774
+    },
+    {
+      "epoch": 0.2231190544825598,
+      "loss": 0.5369690656661987,
+      "loss_ce": 0.18931280076503754,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 114133776,
+      "step": 774
+    },
+    {
+      "epoch": 0.2234073219948112,
+      "grad_norm": 63.51037426847027,
+      "learning_rate": 0.0001,
+      "loss": 0.3826,
+      "num_input_tokens_seen": 114268600,
+      "step": 775
+    },
+    {
+      "epoch": 0.2234073219948112,
+      "loss": 0.4234406352043152,
+      "loss_ce": 0.11557929962873459,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 114268600,
+      "step": 775
+    },
+    {
+      "epoch": 0.22369558950706256,
+      "grad_norm": 33.74071638554486,
+      "learning_rate": 0.0001,
+      "loss": 0.2024,
+      "num_input_tokens_seen": 114403672,
+      "step": 776
+    },
+    {
+      "epoch": 0.22369558950706256,
+      "loss": 0.18327657878398895,
+      "loss_ce": 0.08977071940898895,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 114403672,
+      "step": 776
+    },
+    {
+      "epoch": 0.22398385701931392,
+      "grad_norm": 73.48859012338815,
+      "learning_rate": 0.0001,
+      "loss": 0.4483,
+      "num_input_tokens_seen": 114576152,
+      "step": 777
+    },
+    {
+      "epoch": 0.22398385701931392,
+      "loss": 0.4633466303348541,
+      "loss_ce": 0.10030952095985413,
+      "loss_xval": 0.36328125,
+      "num_input_tokens_seen": 114576152,
+      "step": 777
+    },
+    {
+      "epoch": 0.22427212453156528,
+      "grad_norm": 14.942164124982677,
+      "learning_rate": 0.0001,
+      "loss": 0.1079,
+      "num_input_tokens_seen": 114710920,
+      "step": 778
+    },
+    {
+      "epoch": 0.22427212453156528,
+      "loss": 0.12119618058204651,
+      "loss_ce": 0.09102955460548401,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 114710920,
+      "step": 778
+    },
+    {
+      "epoch": 0.22456039204381667,
+      "grad_norm": 57.038676561237565,
+      "learning_rate": 0.0001,
+      "loss": 0.2876,
+      "num_input_tokens_seen": 114845984,
+      "step": 779
+    },
+    {
+      "epoch": 0.22456039204381667,
+      "loss": 0.27952879667282104,
+      "loss_ce": 0.06114499643445015,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 114845984,
+      "step": 779
+    },
+    {
+      "epoch": 0.22484865955606803,
+      "grad_norm": 43.27715155294268,
+      "learning_rate": 0.0001,
+      "loss": 0.2181,
+      "num_input_tokens_seen": 115018592,
+      "step": 780
+    },
+    {
+      "epoch": 0.22484865955606803,
+      "loss": 0.2002837359905243,
+      "loss_ce": 0.06808160245418549,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 115018592,
+      "step": 780
+    },
+    {
+      "epoch": 0.2251369270683194,
+      "grad_norm": 29.573825175062915,
+      "learning_rate": 0.0001,
+      "loss": 0.141,
+      "num_input_tokens_seen": 115153416,
+      "step": 781
+    },
+    {
+      "epoch": 0.2251369270683194,
+      "loss": 0.16036775708198547,
+      "loss_ce": 0.09045198559761047,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 115153416,
+      "step": 781
+    },
+    {
+      "epoch": 0.22542519458057078,
+      "grad_norm": 55.78457083485312,
+      "learning_rate": 0.0001,
+      "loss": 0.2542,
+      "num_input_tokens_seen": 115288464,
+      "step": 782
+    },
+    {
+      "epoch": 0.22542519458057078,
+      "loss": 0.2591482400894165,
+      "loss_ce": 0.0452810674905777,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 115288464,
+      "step": 782
+    },
+    {
+      "epoch": 0.22571346209282214,
+      "grad_norm": 2.6055577715854104,
+      "learning_rate": 0.0001,
+      "loss": 0.0719,
+      "num_input_tokens_seen": 115461056,
+      "step": 783
+    },
+    {
+      "epoch": 0.22571346209282214,
+      "loss": 0.05857803672552109,
+      "loss_ce": 0.04785110801458359,
+      "loss_xval": 0.0107421875,
+      "num_input_tokens_seen": 115461056,
+      "step": 783
+    },
+    {
+      "epoch": 0.2260017296050735,
+      "grad_norm": 47.20006023553052,
+      "learning_rate": 0.0001,
+      "loss": 0.2008,
+      "num_input_tokens_seen": 115595912,
+      "step": 784
+    },
+    {
+      "epoch": 0.2260017296050735,
+      "loss": 0.2157151997089386,
+      "loss_ce": 0.0576341450214386,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 115595912,
+      "step": 784
+    },
+    {
+      "epoch": 0.22628999711732486,
+      "grad_norm": 28.185111544321373,
+      "learning_rate": 0.0001,
+      "loss": 0.0922,
+      "num_input_tokens_seen": 115730992,
+      "step": 785
+    },
+    {
+      "epoch": 0.22628999711732486,
+      "loss": 0.08684153854846954,
+      "loss_ce": 0.029437974095344543,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 115730992,
+      "step": 785
+    },
+    {
+      "epoch": 0.22657826462957625,
+      "grad_norm": 29.89508500227094,
+      "learning_rate": 0.0001,
+      "loss": 0.1067,
+      "num_input_tokens_seen": 115903480,
+      "step": 786
+    },
+    {
+      "epoch": 0.22657826462957625,
+      "loss": 0.09849099814891815,
+      "loss_ce": 0.04423074051737785,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 115903480,
+      "step": 786
+    },
+    {
+      "epoch": 0.22686653214182761,
+      "grad_norm": 42.04327426869395,
+      "learning_rate": 0.0001,
+      "loss": 0.1484,
+      "num_input_tokens_seen": 116038288,
+      "step": 787
+    },
+    {
+      "epoch": 0.22686653214182761,
+      "loss": 0.160066157579422,
+      "loss_ce": 0.0313430055975914,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 116038288,
+      "step": 787
+    },
+    {
+      "epoch": 0.22715479965407898,
+      "grad_norm": 9.903206689698122,
+      "learning_rate": 0.0001,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 116173248,
+      "step": 788
+    },
+    {
+      "epoch": 0.22715479965407898,
+      "loss": 0.032735370099544525,
+      "loss_ce": 0.018895648419857025,
+      "loss_xval": 0.01385498046875,
+      "num_input_tokens_seen": 116173248,
+      "step": 788
+    },
+    {
+      "epoch": 0.22744306716633036,
+      "grad_norm": 39.38878364407851,
+      "learning_rate": 0.0001,
+      "loss": 0.1426,
+      "num_input_tokens_seen": 116345712,
+      "step": 789
+    },
+    {
+      "epoch": 0.22744306716633036,
+      "loss": 0.14974749088287354,
+      "loss_ce": 0.03658831864595413,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 116345712,
+      "step": 789
+    },
+    {
+      "epoch": 0.22773133467858173,
+      "grad_norm": 10.849920705998569,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "num_input_tokens_seen": 116480424,
+      "step": 790
+    },
+    {
+      "epoch": 0.22773133467858173,
+      "loss": 0.042529068887233734,
+      "loss_ce": 0.024920426309108734,
+      "loss_xval": 0.017578125,
+      "num_input_tokens_seen": 116480424,
+      "step": 790
+    },
+    {
+      "epoch": 0.2280196021908331,
+      "grad_norm": 29.27412447713324,
+      "learning_rate": 0.0001,
+      "loss": 0.0822,
+      "num_input_tokens_seen": 116615552,
+      "step": 791
+    },
+    {
+      "epoch": 0.2280196021908331,
+      "loss": 0.07675212621688843,
+      "loss_ce": 0.012237967923283577,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 116615552,
+      "step": 791
+    },
+    {
+      "epoch": 0.22830786970308448,
+      "grad_norm": 27.650517838759587,
+      "learning_rate": 0.0001,
+      "loss": 0.0588,
+      "num_input_tokens_seen": 116788096,
+      "step": 792
+    },
+    {
+      "epoch": 0.22830786970308448,
+      "loss": 0.05113198608160019,
+      "loss_ce": 0.02311684750020504,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 116788096,
+      "step": 792
+    },
+    {
+      "epoch": 0.22859613721533584,
+      "grad_norm": 16.993351779114146,
+      "learning_rate": 0.0001,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 116922872,
+      "step": 793
+    },
+    {
+      "epoch": 0.22859613721533584,
+      "loss": 0.04983597993850708,
+      "loss_ce": 0.02609330229461193,
+      "loss_xval": 0.023681640625,
+      "num_input_tokens_seen": 116922872,
+      "step": 793
+    },
+    {
+      "epoch": 0.2288844047275872,
+      "grad_norm": 24.271465227011593,
+      "learning_rate": 0.0001,
+      "loss": 0.0618,
+      "num_input_tokens_seen": 117057920,
+      "step": 794
+    },
+    {
+      "epoch": 0.2288844047275872,
+      "loss": 0.054998017847537994,
+      "loss_ce": 0.008885959163308144,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 117057920,
+      "step": 794
+    },
+    {
+      "epoch": 0.22917267223983856,
+      "grad_norm": 6.142005548753007,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "num_input_tokens_seen": 117230352,
+      "step": 795
+    },
+    {
+      "epoch": 0.22917267223983856,
+      "loss": 0.024952255189418793,
+      "loss_ce": 0.017422042787075043,
+      "loss_xval": 0.007537841796875,
+      "num_input_tokens_seen": 117230352,
+      "step": 795
+    },
+    {
+      "epoch": 0.22946093975208995,
+      "grad_norm": 25.269181779510625,
+      "learning_rate": 0.0001,
+      "loss": 0.0654,
+      "num_input_tokens_seen": 117365136,
+      "step": 796
+    },
+    {
+      "epoch": 0.22946093975208995,
+      "loss": 0.07435447722673416,
+      "loss_ce": 0.02610618993639946,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 117365136,
+      "step": 796
+    },
+    {
+      "epoch": 0.2297492072643413,
+      "grad_norm": 4.149447856918701,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "num_input_tokens_seen": 117500200,
+      "step": 797
+    },
+    {
+      "epoch": 0.2297492072643413,
+      "loss": 0.014349638484418392,
+      "loss_ce": 0.0065791006200015545,
+      "loss_xval": 0.007781982421875,
+      "num_input_tokens_seen": 117500200,
+      "step": 797
+    },
+    {
+      "epoch": 0.23003747477659267,
+      "grad_norm": 20.90550876360466,
+      "learning_rate": 0.0001,
+      "loss": 0.0545,
+      "num_input_tokens_seen": 117672592,
+      "step": 798
+    },
+    {
+      "epoch": 0.23003747477659267,
+      "loss": 0.053098104894161224,
+      "loss_ce": 0.015897177159786224,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 117672592,
+      "step": 798
+    },
+    {
+      "epoch": 0.23032574228884406,
+      "grad_norm": 12.299732683231854,
+      "learning_rate": 0.0001,
+      "loss": 0.0297,
+      "num_input_tokens_seen": 117807336,
+      "step": 799
+    },
+    {
+      "epoch": 0.23032574228884406,
+      "loss": 0.040189050137996674,
+      "loss_ce": 0.024312280118465424,
+      "loss_xval": 0.015869140625,
+      "num_input_tokens_seen": 117807336,
+      "step": 799
+    },
+    {
+      "epoch": 0.23061400980109542,
+      "grad_norm": 12.779776228914203,
+      "learning_rate": 0.0001,
+      "loss": 0.0245,
+      "num_input_tokens_seen": 117942352,
+      "step": 800
+    },
+    {
+      "epoch": 0.23061400980109542,
+      "loss": 0.021533731371164322,
+      "loss_ce": 0.005611184053122997,
+      "loss_xval": 0.015869140625,
+      "num_input_tokens_seen": 117942352,
+      "step": 800
+    },
+    {
+      "epoch": 0.23090227731334678,
+      "grad_norm": 18.23997338878261,
+      "learning_rate": 0.0001,
+      "loss": 0.0437,
+      "num_input_tokens_seen": 118114760,
+      "step": 801
+    },
+    {
+      "epoch": 0.23090227731334678,
+      "loss": 0.03563398867845535,
+      "loss_ce": 0.012455889955163002,
+      "loss_xval": 0.023193359375,
+      "num_input_tokens_seen": 118114760,
+      "step": 801
+    },
+    {
+      "epoch": 0.23119054482559814,
+      "grad_norm": 5.126952756766402,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "num_input_tokens_seen": 118249608,
+      "step": 802
+    },
+    {
+      "epoch": 0.23119054482559814,
+      "loss": 0.027275238186120987,
+      "loss_ce": 0.020496521145105362,
+      "loss_xval": 0.00677490234375,
+      "num_input_tokens_seen": 118249608,
+      "step": 802
+    },
+    {
+      "epoch": 0.23147881233784953,
+      "grad_norm": 20.333284577996896,
+      "learning_rate": 0.0001,
+      "loss": 0.0414,
+      "num_input_tokens_seen": 118384776,
+      "step": 803
+    },
+    {
+      "epoch": 0.23147881233784953,
+      "loss": 0.03544127941131592,
+      "loss_ce": 0.004832149483263493,
+      "loss_xval": 0.0306396484375,
+      "num_input_tokens_seen": 118384776,
+      "step": 803
+    },
+    {
+      "epoch": 0.2317670798501009,
+      "grad_norm": 1.0000957750891695,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "num_input_tokens_seen": 118557304,
+      "step": 804
+    },
+    {
+      "epoch": 0.2317670798501009,
+      "loss": 0.014575895853340626,
+      "loss_ce": 0.011049207299947739,
+      "loss_xval": 0.0035247802734375,
+      "num_input_tokens_seen": 118557304,
+      "step": 804
+    },
+    {
+      "epoch": 0.23205534736235225,
+      "grad_norm": 18.193515679118878,
+      "learning_rate": 0.0001,
+      "loss": 0.0369,
+      "num_input_tokens_seen": 118692096,
+      "step": 805
+    },
+    {
+      "epoch": 0.23205534736235225,
+      "loss": 0.04329147934913635,
+      "loss_ce": 0.017046362161636353,
+      "loss_xval": 0.0262451171875,
+      "num_input_tokens_seen": 118692096,
+      "step": 805
+    },
+    {
+      "epoch": 0.23234361487460364,
+      "grad_norm": 5.562258804450563,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "num_input_tokens_seen": 118827128,
+      "step": 806
+    },
+    {
+      "epoch": 0.23234361487460364,
+      "loss": 0.010109086520969868,
+      "loss_ce": 0.004446167498826981,
+      "loss_xval": 0.00567626953125,
+      "num_input_tokens_seen": 118827128,
+      "step": 806
+    },
+    {
+      "epoch": 0.232631882386855,
+      "grad_norm": 13.555981820540076,
+      "learning_rate": 0.0001,
+      "loss": 0.0296,
+      "num_input_tokens_seen": 118999824,
+      "step": 807
+    },
+    {
+      "epoch": 0.232631882386855,
+      "loss": 0.026791900396347046,
+      "loss_ce": 0.009984343312680721,
+      "loss_xval": 0.016845703125,
+      "num_input_tokens_seen": 118999824,
+      "step": 807
+    },
+    {
+      "epoch": 0.23292014989910637,
+      "grad_norm": 9.654390742337505,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "num_input_tokens_seen": 119134800,
+      "step": 808
+    },
+    {
+      "epoch": 0.23292014989910637,
+      "loss": 0.028388747945427895,
+      "loss_ce": 0.01809288002550602,
+      "loss_xval": 0.01031494140625,
+      "num_input_tokens_seen": 119134800,
+      "step": 808
+    },
+    {
+      "epoch": 0.23320841741135773,
+      "grad_norm": 9.473802505614938,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "num_input_tokens_seen": 119270000,
+      "step": 809
+    },
+    {
+      "epoch": 0.23320841741135773,
+      "loss": 0.014459874480962753,
+      "loss_ce": 0.003813053946942091,
+      "loss_xval": 0.0106201171875,
+      "num_input_tokens_seen": 119270000,
+      "step": 809
+    },
+    {
+      "epoch": 0.23349668492360912,
+      "grad_norm": 12.29129249764753,
+      "learning_rate": 0.0001,
+      "loss": 0.0254,
+      "num_input_tokens_seen": 119442552,
+      "step": 810
+    },
+    {
+      "epoch": 0.23349668492360912,
+      "loss": 0.0196915864944458,
+      "loss_ce": 0.007125974632799625,
+      "loss_xval": 0.0125732421875,
+      "num_input_tokens_seen": 119442552,
+      "step": 810
+    },
+    {
+      "epoch": 0.23378495243586048,
+      "grad_norm": 6.4491102706539865,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "num_input_tokens_seen": 119577288,
+      "step": 811
+    },
+    {
+      "epoch": 0.23378495243586048,
+      "loss": 0.02235257439315319,
+      "loss_ce": 0.016197558492422104,
+      "loss_xval": 0.00616455078125,
+      "num_input_tokens_seen": 119577288,
+      "step": 811
+    },
+    {
+      "epoch": 0.23407321994811184,
+      "grad_norm": 11.972644652281343,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "num_input_tokens_seen": 119712456,
+      "step": 812
+    },
+    {
+      "epoch": 0.23407321994811184,
+      "loss": 0.017476797103881836,
+      "loss_ce": 0.0034692296758294106,
+      "loss_xval": 0.0140380859375,
+      "num_input_tokens_seen": 119712456,
+      "step": 812
+    },
+    {
+      "epoch": 0.23436148746036323,
+      "grad_norm": 2.3991287299948647,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "num_input_tokens_seen": 119884928,
+      "step": 813
+    },
+    {
+      "epoch": 0.23436148746036323,
+      "loss": 0.0121365487575531,
+      "loss_ce": 0.009936422109603882,
+      "loss_xval": 0.002197265625,
+      "num_input_tokens_seen": 119884928,
+      "step": 813
+    },
+    {
+      "epoch": 0.2346497549726146,
+      "grad_norm": 11.432710545599601,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "num_input_tokens_seen": 120019720,
+      "step": 814
+    },
+    {
+      "epoch": 0.2346497549726146,
+      "loss": 0.02629391849040985,
+      "loss_ce": 0.014300510287284851,
+      "loss_xval": 0.011962890625,
+      "num_input_tokens_seen": 120019720,
+      "step": 814
+    },
+    {
+      "epoch": 0.23493802248486595,
+      "grad_norm": 0.4407423912378353,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "num_input_tokens_seen": 120154800,
+      "step": 815
+    },
+    {
+      "epoch": 0.23493802248486595,
+      "loss": 0.004980717785656452,
+      "loss_ce": 0.002568875439465046,
+      "loss_xval": 0.002410888671875,
+      "num_input_tokens_seen": 120154800,
+      "step": 815
+    },
+    {
+      "epoch": 0.23522628999711734,
+      "grad_norm": 10.567351106248262,
+      "learning_rate": 0.0001,
+      "loss": 0.0256,
+      "num_input_tokens_seen": 120327264,
+      "step": 816
+    },
+    {
+      "epoch": 0.23522628999711734,
+      "loss": 0.023881588131189346,
+      "loss_ce": 0.012002620846033096,
+      "loss_xval": 0.01190185546875,
+      "num_input_tokens_seen": 120327264,
+      "step": 816
+    },
+    {
+      "epoch": 0.2355145575093687,
+      "grad_norm": 1.9443929350512452,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "num_input_tokens_seen": 120462024,
+      "step": 817
+    },
+    {
+      "epoch": 0.2355145575093687,
+      "loss": 0.019425533711910248,
+      "loss_ce": 0.016307972371578217,
+      "loss_xval": 0.00311279296875,
+      "num_input_tokens_seen": 120462024,
+      "step": 817
+    },
+    {
+      "epoch": 0.23580282502162006,
+      "grad_norm": 8.887875354058915,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "num_input_tokens_seen": 120596992,
+      "step": 818
+    },
+    {
+      "epoch": 0.23580282502162006,
+      "loss": 0.011268608272075653,
+      "loss_ce": 0.002510063350200653,
+      "loss_xval": 0.0087890625,
+      "num_input_tokens_seen": 120596992,
+      "step": 818
+    },
+    {
+      "epoch": 0.23609109253387142,
+      "grad_norm": 3.4867306523058232,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "num_input_tokens_seen": 120769480,
+      "step": 819
+    },
+    {
+      "epoch": 0.23609109253387142,
+      "loss": 0.011347447521984577,
+      "loss_ce": 0.008326207287609577,
+      "loss_xval": 0.003021240234375,
+      "num_input_tokens_seen": 120769480,
+      "step": 819
+    },
+    {
+      "epoch": 0.2363793600461228,
+      "grad_norm": 7.558834106455437,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "num_input_tokens_seen": 120904232,
+      "step": 820
+    },
+    {
+      "epoch": 0.2363793600461228,
+      "loss": 0.023228928446769714,
+      "loss_ce": 0.01735810935497284,
+      "loss_xval": 0.005859375,
+      "num_input_tokens_seen": 120904232,
+      "step": 820
+    },
+    {
+      "epoch": 0.23666762755837417,
+      "grad_norm": 3.8589553586698733,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "num_input_tokens_seen": 121039600,
+      "step": 821
+    },
+    {
+      "epoch": 0.23666762755837417,
+      "loss": 0.007957377471029758,
+      "loss_ce": 0.005234637297689915,
+      "loss_xval": 0.002716064453125,
+      "num_input_tokens_seen": 121039600,
+      "step": 821
+    },
+    {
+      "epoch": 0.23695589507062553,
+      "grad_norm": 6.547441002040102,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "num_input_tokens_seen": 121212112,
+      "step": 822
+    },
+    {
+      "epoch": 0.23695589507062553,
+      "loss": 0.009982486255466938,
+      "loss_ce": 0.0054372744634747505,
+      "loss_xval": 0.004547119140625,
+      "num_input_tokens_seen": 121212112,
+      "step": 822
+    },
+    {
+      "epoch": 0.23724416258287692,
+      "grad_norm": 3.574675045046867,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "num_input_tokens_seen": 121346864,
+      "step": 823
+    },
+    {
+      "epoch": 0.23724416258287692,
+      "loss": 0.018714873120188713,
+      "loss_ce": 0.0160426776856184,
+      "loss_xval": 0.0026702880859375,
+      "num_input_tokens_seen": 121346864,
+      "step": 823
+    },
+    {
+      "epoch": 0.23753243009512828,
+      "grad_norm": 5.120405630275477,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "num_input_tokens_seen": 121482008,
+      "step": 824
+    },
+    {
+      "epoch": 0.23753243009512828,
+      "loss": 0.006162785924971104,
+      "loss_ce": 0.002063893945887685,
+      "loss_xval": 0.00408935546875,
+      "num_input_tokens_seen": 121482008,
+      "step": 824
+    },
+    {
+      "epoch": 0.23782069760737964,
+      "grad_norm": 3.847733022136177,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "num_input_tokens_seen": 121654536,
+      "step": 825
+    },
+    {
+      "epoch": 0.23782069760737964,
+      "loss": 0.006931713782250881,
+      "loss_ce": 0.004517009947448969,
+      "loss_xval": 0.002410888671875,
+      "num_input_tokens_seen": 121654536,
+      "step": 825
+    },
+    {
+      "epoch": 0.238108965119631,
+      "grad_norm": 4.460862200587386,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "num_input_tokens_seen": 121789288,
+      "step": 826
+    },
+    {
+      "epoch": 0.238108965119631,
+      "loss": 0.017785396426916122,
+      "loss_ce": 0.015078868716955185,
+      "loss_xval": 0.0027008056640625,
+      "num_input_tokens_seen": 121789288,
+      "step": 826
+    },
+    {
+      "epoch": 0.2383972326318824,
+      "grad_norm": 3.2989980361041567,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "num_input_tokens_seen": 121924200,
+      "step": 827
+    },
+    {
+      "epoch": 0.2383972326318824,
+      "loss": 0.0073768566362559795,
+      "loss_ce": 0.0046989391557872295,
+      "loss_xval": 0.002685546875,
+      "num_input_tokens_seen": 121924200,
+      "step": 827
+    },
+    {
+      "epoch": 0.23868550014413376,
+      "grad_norm": 3.652485050248189,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "num_input_tokens_seen": 122096688,
+      "step": 828
+    },
+    {
+      "epoch": 0.23868550014413376,
+      "loss": 0.011565220542252064,
+      "loss_ce": 0.009024632163345814,
+      "loss_xval": 0.002532958984375,
+      "num_input_tokens_seen": 122096688,
+      "step": 828
+    },
+    {
+      "epoch": 0.23897376765638512,
+      "grad_norm": 2.7805410201893728,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "num_input_tokens_seen": 122231512,
+      "step": 829
+    },
+    {
+      "epoch": 0.23897376765638512,
+      "loss": 0.019701950252056122,
+      "loss_ce": 0.017360679805278778,
+      "loss_xval": 0.0023345947265625,
+      "num_input_tokens_seen": 122231512,
+      "step": 829
+    },
+    {
+      "epoch": 0.2392620351686365,
+      "grad_norm": 3.7491616956615665,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "num_input_tokens_seen": 122366616,
+      "step": 830
+    },
+    {
+      "epoch": 0.2392620351686365,
+      "loss": 0.004523873329162598,
+      "loss_ce": 0.001920342561788857,
+      "loss_xval": 0.0026092529296875,
+      "num_input_tokens_seen": 122366616,
+      "step": 830
+    },
+    {
+      "epoch": 0.23955030268088787,
+      "grad_norm": 1.7379208880039498,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "num_input_tokens_seen": 122539112,
+      "step": 831
+    },
+    {
+      "epoch": 0.23955030268088787,
+      "loss": 0.008797908201813698,
+      "loss_ce": 0.006952548865228891,
+      "loss_xval": 0.0018463134765625,
+      "num_input_tokens_seen": 122539112,
+      "step": 831
+    },
+    {
+      "epoch": 0.23983857019313923,
+      "grad_norm": 2.942166352575048,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "num_input_tokens_seen": 122673944,
+      "step": 832
+    },
+    {
+      "epoch": 0.23983857019313923,
+      "loss": 0.017966344952583313,
+      "loss_ce": 0.01571376621723175,
+      "loss_xval": 0.00225830078125,
+      "num_input_tokens_seen": 122673944,
+      "step": 832
+    },
+    {
+      "epoch": 0.2401268377053906,
+      "grad_norm": 1.8229508953203557,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "num_input_tokens_seen": 122809096,
+      "step": 833
+    },
+    {
+      "epoch": 0.2401268377053906,
+      "loss": 0.003673751372843981,
+      "loss_ce": 0.0015937874559313059,
+      "loss_xval": 0.0020751953125,
+      "num_input_tokens_seen": 122809096,
+      "step": 833
+    },
+    {
+      "epoch": 0.24041510521764198,
+      "grad_norm": 2.8603712211321946,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "num_input_tokens_seen": 122981720,
+      "step": 834
+    },
+    {
+      "epoch": 0.24041510521764198,
+      "loss": 0.01071496307849884,
+      "loss_ce": 0.008584454655647278,
+      "loss_xval": 0.00213623046875,
+      "num_input_tokens_seen": 122981720,
+      "step": 834
+    },
+    {
+      "epoch": 0.24070337272989334,
+      "grad_norm": 1.672332473164387,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "num_input_tokens_seen": 123116504,
+      "step": 835
+    },
+    {
+      "epoch": 0.24070337272989334,
+      "loss": 0.015596328303217888,
+      "loss_ce": 0.013956962153315544,
+      "loss_xval": 0.00164031982421875,
+      "num_input_tokens_seen": 123116504,
+      "step": 835
+    },
+    {
+      "epoch": 0.2409916402421447,
+      "grad_norm": 2.8517776188793613,
+      "learning_rate": 0.0001,
+      "loss": 0.0055,
+      "num_input_tokens_seen": 123251632,
+      "step": 836
+    },
+    {
+      "epoch": 0.2409916402421447,
+      "loss": 0.004265207797288895,
+      "loss_ce": 0.0019525473471730947,
+      "loss_xval": 0.0023193359375,
+      "num_input_tokens_seen": 123251632,
+      "step": 836
+    },
+    {
+      "epoch": 0.2412799077543961,
+      "grad_norm": 1.3730800143412834,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "num_input_tokens_seen": 123424136,
+      "step": 837
+    },
+    {
+      "epoch": 0.2412799077543961,
+      "loss": 0.007244983687996864,
+      "loss_ce": 0.005767742171883583,
+      "loss_xval": 0.0014801025390625,
+      "num_input_tokens_seen": 123424136,
+      "step": 837
+    },
+    {
+      "epoch": 0.24156817526664745,
+      "grad_norm": 3.164522197042215,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "num_input_tokens_seen": 123558920,
+      "step": 838
+    },
+    {
+      "epoch": 0.24156817526664745,
+      "loss": 0.018805041909217834,
+      "loss_ce": 0.016788974404335022,
+      "loss_xval": 0.00201416015625,
+      "num_input_tokens_seen": 123558920,
+      "step": 838
+    },
+    {
+      "epoch": 0.2418564427788988,
+      "grad_norm": 0.6940947497193604,
+      "learning_rate": 0.0001,
+      "loss": 0.0054,
+      "num_input_tokens_seen": 123694040,
+      "step": 839
+    },
+    {
+      "epoch": 0.2418564427788988,
+      "loss": 0.0029890744481235743,
+      "loss_ce": 0.0013792722020298243,
+      "loss_xval": 0.00160980224609375,
+      "num_input_tokens_seen": 123694040,
+      "step": 839
+    },
+    {
+      "epoch": 0.2421447102911502,
+      "grad_norm": 3.414009592985551,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "num_input_tokens_seen": 123866680,
+      "step": 840
+    },
+    {
+      "epoch": 0.2421447102911502,
+      "loss": 0.008075650781393051,
+      "loss_ce": 0.005059178918600082,
+      "loss_xval": 0.003021240234375,
+      "num_input_tokens_seen": 123866680,
+      "step": 840
+    },
+    {
+      "epoch": 0.24243297780340156,
+      "grad_norm": 1.4163402518449046,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "num_input_tokens_seen": 124001520,
+      "step": 841
+    },
+    {
+      "epoch": 0.24243297780340156,
+      "loss": 0.015565956011414528,
+      "loss_ce": 0.014158331789076328,
+      "loss_xval": 0.00140380859375,
+      "num_input_tokens_seen": 124001520,
+      "step": 841
+    },
+    {
+      "epoch": 0.24272124531565292,
+      "grad_norm": 1.403378922157825,
+      "learning_rate": 0.0001,
+      "loss": 0.0047,
+      "num_input_tokens_seen": 124136760,
+      "step": 842
+    },
+    {
+      "epoch": 0.24272124531565292,
+      "loss": 0.003177802776917815,
+      "loss_ce": 0.001676719170063734,
+      "loss_xval": 0.00150299072265625,
+      "num_input_tokens_seen": 124136760,
+      "step": 842
+    },
+    {
+      "epoch": 0.24300951282790428,
+      "grad_norm": 0.9423805294699967,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "num_input_tokens_seen": 124309304,
+      "step": 843
+    },
+    {
+      "epoch": 0.24300951282790428,
+      "loss": 0.008656498044729233,
+      "loss_ce": 0.007049556355923414,
+      "loss_xval": 0.00160980224609375,
+      "num_input_tokens_seen": 124309304,
+      "step": 843
+    },
+    {
+      "epoch": 0.24329778034015567,
+      "grad_norm": 0.9049718050772102,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "num_input_tokens_seen": 124444096,
+      "step": 844
+    },
+    {
+      "epoch": 0.24329778034015567,
+      "loss": 0.012917540036141872,
+      "loss_ce": 0.01146986149251461,
+      "loss_xval": 0.0014495849609375,
+      "num_input_tokens_seen": 124444096,
+      "step": 844
+    },
+    {
+      "epoch": 0.24358604785240703,
+      "grad_norm": 0.8273414421621751,
+      "learning_rate": 0.0001,
+      "loss": 0.005,
+      "num_input_tokens_seen": 124579144,
+      "step": 845
+    },
+    {
+      "epoch": 0.24358604785240703,
+      "loss": 0.002799539128318429,
+      "loss_ce": 0.0014577193651348352,
+      "loss_xval": 0.0013427734375,
+      "num_input_tokens_seen": 124579144,
+      "step": 845
+    },
+    {
+      "epoch": 0.2438743153646584,
+      "grad_norm": 0.7750075248220467,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "num_input_tokens_seen": 124751800,
+      "step": 846
+    },
+    {
+      "epoch": 0.2438743153646584,
+      "loss": 0.010742152109742165,
+      "loss_ce": 0.008784258738160133,
+      "loss_xval": 0.001953125,
+      "num_input_tokens_seen": 124751800,
+      "step": 846
+    },
+    {
+      "epoch": 0.24416258287690978,
+      "grad_norm": 0.5817994642876541,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "num_input_tokens_seen": 124886816,
+      "step": 847
+    },
+    {
+      "epoch": 0.24416258287690978,
+      "loss": 0.016359955072402954,
+      "loss_ce": 0.01504865288734436,
+      "loss_xval": 0.001312255859375,
+      "num_input_tokens_seen": 124886816,
+      "step": 847
+    },
+    {
+      "epoch": 0.24445085038916115,
+      "grad_norm": 1.7709905019760608,
+      "learning_rate": 0.0001,
+      "loss": 0.0044,
+      "num_input_tokens_seen": 125021848,
+      "step": 848
+    },
+    {
+      "epoch": 0.24445085038916115,
+      "loss": 0.0030480134300887585,
+      "loss_ce": 0.0013533341698348522,
+      "loss_xval": 0.0016937255859375,
+      "num_input_tokens_seen": 125021848,
+      "step": 848
+    },
+    {
+      "epoch": 0.2447391179014125,
+      "grad_norm": 2.165383770527757,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "num_input_tokens_seen": 125194344,
+      "step": 849
+    },
+    {
+      "epoch": 0.2447391179014125,
+      "loss": 0.005808062851428986,
+      "loss_ce": 0.003933139145374298,
+      "loss_xval": 0.0018768310546875,
+      "num_input_tokens_seen": 125194344,
+      "step": 849
+    },
+    {
+      "epoch": 0.24502738541366387,
+      "grad_norm": 0.2620781135391585,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "num_input_tokens_seen": 125329208,
+      "step": 850
+    },
+    {
+      "epoch": 0.24502738541366387,
+      "loss": 0.014183461666107178,
+      "loss_ce": 0.012954652309417725,
+      "loss_xval": 0.00122833251953125,
+      "num_input_tokens_seen": 125329208,
+      "step": 850
+    },
+    {
+      "epoch": 0.24531565292591526,
+      "grad_norm": 1.9324423430851303,
+      "learning_rate": 0.0001,
+      "loss": 0.0055,
+      "num_input_tokens_seen": 125464352,
+      "step": 851
+    },
+    {
+      "epoch": 0.24531565292591526,
+      "loss": 0.0028029857203364372,
+      "loss_ce": 0.0011912761256098747,
+      "loss_xval": 0.00160980224609375,
+      "num_input_tokens_seen": 125464352,
+      "step": 851
+    },
+    {
+      "epoch": 0.24560392043816662,
+      "grad_norm": 1.994122040931301,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "num_input_tokens_seen": 125636976,
+      "step": 852
+    },
+    {
+      "epoch": 0.24560392043816662,
+      "loss": 0.006850851699709892,
+      "loss_ce": 0.005015028640627861,
+      "loss_xval": 0.00183868408203125,
+      "num_input_tokens_seen": 125636976,
+      "step": 852
+    },
+    {
+      "epoch": 0.24589218795041798,
+      "grad_norm": 0.6645628358683997,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "num_input_tokens_seen": 125771696,
+      "step": 853
+    },
+    {
+      "epoch": 0.24589218795041798,
+      "loss": 0.01390957273542881,
+      "loss_ce": 0.0125958863645792,
+      "loss_xval": 0.001312255859375,
+      "num_input_tokens_seen": 125771696,
+      "step": 853
+    },
+    {
+      "epoch": 0.24618045546266937,
+      "grad_norm": 0.9193485395039878,
+      "learning_rate": 0.0001,
+      "loss": 0.0067,
+      "num_input_tokens_seen": 125907032,
+      "step": 854
+    },
+    {
+      "epoch": 0.24618045546266937,
+      "loss": 0.008167913183569908,
+      "loss_ce": 0.006742647383362055,
+      "loss_xval": 0.00142669677734375,
+      "num_input_tokens_seen": 125907032,
+      "step": 854
+    },
+    {
+      "epoch": 0.24646872297492073,
+      "grad_norm": 1.0377375477638926,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "num_input_tokens_seen": 126079496,
+      "step": 855
+    },
+    {
+      "epoch": 0.24646872297492073,
+      "loss": 0.005995092913508415,
+      "loss_ce": 0.004644213244318962,
+      "loss_xval": 0.00135040283203125,
+      "num_input_tokens_seen": 126079496,
+      "step": 855
+    },
+    {
+      "epoch": 0.2467569904871721,
+      "grad_norm": 0.4591843203940489,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "num_input_tokens_seen": 126214408,
+      "step": 856
+    },
+    {
+      "epoch": 0.2467569904871721,
+      "loss": 0.013096168637275696,
+      "loss_ce": 0.01175482664257288,
+      "loss_xval": 0.0013427734375,
+      "num_input_tokens_seen": 126214408,
+      "step": 856
+    },
+    {
+      "epoch": 0.24704525799942348,
+      "grad_norm": 0.3583351807212624,
+      "learning_rate": 0.0001,
+      "loss": 0.005,
+      "num_input_tokens_seen": 126349512,
+      "step": 857
+    },
+    {
+      "epoch": 0.24704525799942348,
+      "loss": 0.0021648311521857977,
+      "loss_ce": 0.0009698771755211055,
+      "loss_xval": 0.00119781494140625,
+      "num_input_tokens_seen": 126349512,
+      "step": 857
+    },
+    {
+      "epoch": 0.24733352551167484,
+      "grad_norm": 0.4255231077383957,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "num_input_tokens_seen": 126521928,
+      "step": 858
+    },
+    {
+      "epoch": 0.24733352551167484,
+      "loss": 0.007635718677192926,
+      "loss_ce": 0.0065132444724440575,
+      "loss_xval": 0.00112152099609375,
+      "num_input_tokens_seen": 126521928,
+      "step": 858
+    },
+    {
+      "epoch": 0.2476217930239262,
+      "grad_norm": 1.9034822694333349,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "num_input_tokens_seen": 126656648,
+      "step": 859
+    },
+    {
+      "epoch": 0.2476217930239262,
+      "loss": 0.011844690889120102,
+      "loss_ce": 0.010460909456014633,
+      "loss_xval": 0.00138092041015625,
+      "num_input_tokens_seen": 126656648,
+      "step": 859
+    },
+    {
+      "epoch": 0.24791006053617756,
+      "grad_norm": 2.160556800745774,
+      "learning_rate": 0.0001,
+      "loss": 0.0038,
+      "num_input_tokens_seen": 126791656,
+      "step": 860
+    },
+    {
+      "epoch": 0.24791006053617756,
+      "loss": 0.0028993161395192146,
+      "loss_ce": 0.0011569531634449959,
+      "loss_xval": 0.001739501953125,
+      "num_input_tokens_seen": 126791656,
+      "step": 860
+    },
+    {
+      "epoch": 0.24819832804842895,
+      "grad_norm": 0.8898372077055581,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "num_input_tokens_seen": 126964216,
+      "step": 861
+    },
+    {
+      "epoch": 0.24819832804842895,
+      "loss": 0.0031397445127367973,
+      "loss_ce": 0.002113114111125469,
+      "loss_xval": 0.00102996826171875,
+      "num_input_tokens_seen": 126964216,
+      "step": 861
+    },
+    {
+      "epoch": 0.2484865955606803,
+      "grad_norm": 3.312099635274805,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "num_input_tokens_seen": 127099080,
+      "step": 862
+    },
+    {
+      "epoch": 0.2484865955606803,
+      "loss": 0.01764540746808052,
+      "loss_ce": 0.015682268887758255,
+      "loss_xval": 0.0019683837890625,
+      "num_input_tokens_seen": 127099080,
+      "step": 862
+    },
+    {
+      "epoch": 0.24877486307293167,
+      "grad_norm": 1.5009503860441187,
+      "learning_rate": 0.0001,
+      "loss": 0.0045,
+      "num_input_tokens_seen": 127234144,
+      "step": 863
+    },
+    {
+      "epoch": 0.24877486307293167,
+      "loss": 0.0024071503430604935,
+      "loss_ce": 0.0009451676160097122,
+      "loss_xval": 0.00146484375,
+      "num_input_tokens_seen": 127234144,
+      "step": 863
+    },
+    {
+      "epoch": 0.24906313058518306,
+      "grad_norm": 3.065705952173178,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "num_input_tokens_seen": 127406696,
+      "step": 864
+    },
+    {
+      "epoch": 0.24906313058518306,
+      "loss": 0.005417156033217907,
+      "loss_ce": 0.003864574246108532,
+      "loss_xval": 0.001556396484375,
+      "num_input_tokens_seen": 127406696,
+      "step": 864
+    },
+    {
+      "epoch": 0.24935139809743442,
+      "grad_norm": 5.191152496926581,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "num_input_tokens_seen": 127541608,
+      "step": 865
+    },
+    {
+      "epoch": 0.24935139809743442,
+      "loss": 0.015241998247802258,
+      "loss_ce": 0.012771028093993664,
+      "loss_xval": 0.002471923828125,
+      "num_input_tokens_seen": 127541608,
+      "step": 865
+    },
+    {
+      "epoch": 0.24963966560968578,
+      "grad_norm": 1.9883630746272334,
+      "learning_rate": 0.0001,
+      "loss": 0.0043,
+      "num_input_tokens_seen": 127676656,
+      "step": 866
+    },
+    {
+      "epoch": 0.24963966560968578,
+      "loss": 0.0024233288131654263,
+      "loss_ce": 0.0009551472612656653,
+      "loss_xval": 0.00146484375,
+      "num_input_tokens_seen": 127676656,
+      "step": 866
+    },
+    {
+      "epoch": 0.24992793312193715,
+      "grad_norm": 4.045749456376484,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "num_input_tokens_seen": 127849168,
+      "step": 867
+    },
+    {
+      "epoch": 0.24992793312193715,
+      "loss": 0.005908184219151735,
+      "loss_ce": 0.004185848403722048,
+      "loss_xval": 0.0017242431640625,
+      "num_input_tokens_seen": 127849168,
+      "step": 867
+    },
+    {
+      "epoch": 0.25021620063418853,
+      "grad_norm": 6.916765603415226,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "num_input_tokens_seen": 127983992,
+      "step": 868
+    },
+    {
+      "epoch": 0.25021620063418853,
+      "loss": 0.01603618450462818,
+      "loss_ce": 0.012471349909901619,
+      "loss_xval": 0.003570556640625,
+      "num_input_tokens_seen": 127983992,
+      "step": 868
+    },
+    {
+      "epoch": 0.2505044681464399,
+      "grad_norm": 2.7933517156089267,
+      "learning_rate": 0.0001,
+      "loss": 0.0035,
+      "num_input_tokens_seen": 128118960,
+      "step": 869
+    },
+    {
+      "epoch": 0.2505044681464399,
+      "loss": 0.002570713870227337,
+      "loss_ce": 0.000868404982611537,
+      "loss_xval": 0.00170135498046875,
+      "num_input_tokens_seen": 128118960,
+      "step": 869
+    },
+    {
+      "epoch": 0.25079273565869126,
+      "grad_norm": 5.686330272706534,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "num_input_tokens_seen": 128291560,
+      "step": 870
+    },
+    {
+      "epoch": 0.25079273565869126,
+      "loss": 0.005967923440039158,
+      "loss_ce": 0.0036791053134948015,
+      "loss_xval": 0.002288818359375,
+      "num_input_tokens_seen": 128291560,
+      "step": 870
+    },
+    {
+      "epoch": 0.2510810031709426,
+      "grad_norm": 10.381696412661938,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "num_input_tokens_seen": 128426336,
+      "step": 871
+    },
+    {
+      "epoch": 0.2510810031709426,
+      "loss": 0.020823419094085693,
+      "loss_ce": 0.013220726512372494,
+      "loss_xval": 0.007598876953125,
+      "num_input_tokens_seen": 128426336,
+      "step": 871
+    },
+    {
+      "epoch": 0.251369270683194,
+      "grad_norm": 5.533838864070437,
+      "learning_rate": 0.0001,
+      "loss": 0.0045,
+      "num_input_tokens_seen": 128561360,
+      "step": 872
+    },
+    {
+      "epoch": 0.251369270683194,
+      "loss": 0.004516711458563805,
+      "loss_ce": 0.0008450658060610294,
+      "loss_xval": 0.0036773681640625,
+      "num_input_tokens_seen": 128561360,
+      "step": 872
+    },
+    {
+      "epoch": 0.2516575381954454,
+      "grad_norm": 5.185474513960417,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "num_input_tokens_seen": 128733848,
+      "step": 873
+    },
+    {
+      "epoch": 0.2516575381954454,
+      "loss": 0.003517691045999527,
+      "loss_ce": 0.0014730134280398488,
+      "loss_xval": 0.002044677734375,
+      "num_input_tokens_seen": 128733848,
+      "step": 873
+    },
+    {
+      "epoch": 0.25194580570769676,
+      "grad_norm": 11.881789355652334,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "num_input_tokens_seen": 128868688,
+      "step": 874
+    },
+    {
+      "epoch": 0.25194580570769676,
+      "loss": 0.020403128117322922,
+      "loss_ce": 0.011839132755994797,
+      "loss_xval": 0.008544921875,
+      "num_input_tokens_seen": 128868688,
+      "step": 874
+    },
+    {
+      "epoch": 0.2522340732199481,
+      "grad_norm": 8.080500825807434,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "num_input_tokens_seen": 129003864,
+      "step": 875
+    },
+    {
+      "epoch": 0.2522340732199481,
+      "loss": 0.0072290534153580666,
+      "loss_ce": 0.0008051031036302447,
+      "loss_xval": 0.00640869140625,
+      "num_input_tokens_seen": 129003864,
+      "step": 875
+    },
+    {
+      "epoch": 0.2525223407321995,
+      "grad_norm": 3.481422564697328,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "num_input_tokens_seen": 129176360,
+      "step": 876
+    },
+    {
+      "epoch": 0.2525223407321995,
+      "loss": 0.0051642959006130695,
+      "loss_ce": 0.003819615114480257,
+      "loss_xval": 0.0013427734375,
+      "num_input_tokens_seen": 129176360,
+      "step": 876
+    },
+    {
+      "epoch": 0.25281060824445084,
+      "grad_norm": 11.855942235668593,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "num_input_tokens_seen": 129311152,
+      "step": 877
+    },
+    {
+      "epoch": 0.25281060824445084,
+      "loss": 0.022560758516192436,
+      "loss_ce": 0.014446896500885487,
+      "loss_xval": 0.00811767578125,
+      "num_input_tokens_seen": 129311152,
+      "step": 877
+    },
+    {
+      "epoch": 0.2530988757567022,
+      "grad_norm": 8.119722893152582,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "num_input_tokens_seen": 129446176,
+      "step": 878
+    },
+    {
+      "epoch": 0.2530988757567022,
+      "loss": 0.007987035438418388,
+      "loss_ce": 0.0008916989318095148,
+      "loss_xval": 0.007080078125,
+      "num_input_tokens_seen": 129446176,
+      "step": 878
+    },
+    {
+      "epoch": 0.25338714326895356,
+      "grad_norm": 4.820818363547437,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "num_input_tokens_seen": 129618720,
+      "step": 879
+    },
+    {
+      "epoch": 0.25338714326895356,
+      "loss": 0.004318573512136936,
+      "loss_ce": 0.0032614259980618954,
+      "loss_xval": 0.00106048583984375,
+      "num_input_tokens_seen": 129618720,
+      "step": 879
+    },
+    {
+      "epoch": 0.253675410781205,
+      "grad_norm": 14.62637741629135,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "num_input_tokens_seen": 129753536,
+      "step": 880
+    },
+    {
+      "epoch": 0.253675410781205,
+      "loss": 0.024506494402885437,
+      "loss_ce": 0.012139246799051762,
+      "loss_xval": 0.01239013671875,
+      "num_input_tokens_seen": 129753536,
+      "step": 880
+    },
+    {
+      "epoch": 0.25396367829345634,
+      "grad_norm": 11.717901677148392,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "num_input_tokens_seen": 129888712,
+      "step": 881
+    },
+    {
+      "epoch": 0.25396367829345634,
+      "loss": 0.015599137172102928,
+      "loss_ce": 0.000851518358103931,
+      "loss_xval": 0.0147705078125,
+      "num_input_tokens_seen": 129888712,
+      "step": 881
+    },
+    {
+      "epoch": 0.2542519458057077,
+      "grad_norm": 1.4620165698355299,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "num_input_tokens_seen": 130061136,
+      "step": 882
+    },
+    {
+      "epoch": 0.2542519458057077,
+      "loss": 0.003730251919478178,
+      "loss_ce": 0.0018252875888720155,
+      "loss_xval": 0.0019073486328125,
+      "num_input_tokens_seen": 130061136,
+      "step": 882
+    },
+    {
+      "epoch": 0.25454021331795906,
+      "grad_norm": 11.906553202096525,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "num_input_tokens_seen": 130195896,
+      "step": 883
+    },
+    {
+      "epoch": 0.25454021331795906,
+      "loss": 0.02044621855020523,
+      "loss_ce": 0.013198292814195156,
+      "loss_xval": 0.00726318359375,
+      "num_input_tokens_seen": 130195896,
+      "step": 883
+    },
+    {
+      "epoch": 0.2548284808302104,
+      "grad_norm": 9.009958862531548,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "num_input_tokens_seen": 130330992,
+      "step": 884
+    },
+    {
+      "epoch": 0.2548284808302104,
+      "loss": 0.010740547440946102,
+      "loss_ce": 0.0007994465995579958,
+      "loss_xval": 0.00994873046875,
+      "num_input_tokens_seen": 130330992,
+      "step": 884
+    },
+    {
+      "epoch": 0.2551167483424618,
+      "grad_norm": 4.994362602119953,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "num_input_tokens_seen": 130503464,
+      "step": 885
+    },
+    {
+      "epoch": 0.2551167483424618,
+      "loss": 0.00263015553355217,
+      "loss_ce": 0.0014218504074960947,
+      "loss_xval": 0.0012054443359375,
+      "num_input_tokens_seen": 130503464,
+      "step": 885
+    },
+    {
+      "epoch": 0.25540501585471315,
+      "grad_norm": 16.541549750545187,
+      "learning_rate": 0.0001,
+      "loss": 0.0254,
+      "num_input_tokens_seen": 130638304,
+      "step": 886
+    },
+    {
+      "epoch": 0.25540501585471315,
+      "loss": 0.0236981064081192,
+      "loss_ce": 0.009553208947181702,
+      "loss_xval": 0.01416015625,
+      "num_input_tokens_seen": 130638304,
+      "step": 886
+    },
+    {
+      "epoch": 0.25569328336696456,
+      "grad_norm": 14.441262893561307,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "num_input_tokens_seen": 130773464,
+      "step": 887
+    },
+    {
+      "epoch": 0.25569328336696456,
+      "loss": 0.021735940128564835,
+      "loss_ce": 0.0007093290332704782,
+      "loss_xval": 0.02099609375,
+      "num_input_tokens_seen": 130773464,
+      "step": 887
+    },
+    {
+      "epoch": 0.2559815508792159,
+      "grad_norm": 1.030551625521142,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "num_input_tokens_seen": 130945968,
+      "step": 888
+    },
+    {
+      "epoch": 0.2559815508792159,
+      "loss": 0.007245267741382122,
+      "loss_ce": 0.0021755348425358534,
+      "loss_xval": 0.00506591796875,
+      "num_input_tokens_seen": 130945968,
+      "step": 888
+    },
+    {
+      "epoch": 0.2562698183914673,
+      "grad_norm": 10.336845186935472,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "num_input_tokens_seen": 131080728,
+      "step": 889
+    },
+    {
+      "epoch": 0.2562698183914673,
+      "loss": 0.017896853387355804,
+      "loss_ce": 0.01371975801885128,
+      "loss_xval": 0.004180908203125,
+      "num_input_tokens_seen": 131080728,
+      "step": 889
+    },
+    {
+      "epoch": 0.25655808590371865,
+      "grad_norm": 7.421496056016462,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "num_input_tokens_seen": 131215752,
+      "step": 890
+    },
+    {
+      "epoch": 0.25655808590371865,
+      "loss": 0.010436332784593105,
+      "loss_ce": 0.0010254747467115521,
+      "loss_xval": 0.0093994140625,
+      "num_input_tokens_seen": 131215752,
+      "step": 890
+    },
+    {
+      "epoch": 0.25684635341597,
+      "grad_norm": 7.379148752876616,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "num_input_tokens_seen": 131388184,
+      "step": 891
+    },
+    {
+      "epoch": 0.25684635341597,
+      "loss": 0.0033525533508509398,
+      "loss_ce": 0.001976401312276721,
+      "loss_xval": 0.001373291015625,
+      "num_input_tokens_seen": 131388184,
+      "step": 891
+    },
+    {
+      "epoch": 0.25713462092822137,
+      "grad_norm": 19.251412926550937,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "num_input_tokens_seen": 131522984,
+      "step": 892
+    },
+    {
+      "epoch": 0.25713462092822137,
+      "loss": 0.028828779235482216,
+      "loss_ce": 0.01163975428789854,
+      "loss_xval": 0.0172119140625,
+      "num_input_tokens_seen": 131522984,
+      "step": 892
+    },
+    {
+      "epoch": 0.2574228884404728,
+      "grad_norm": 15.799551940261379,
+      "learning_rate": 0.0001,
+      "loss": 0.0231,
+      "num_input_tokens_seen": 131658112,
+      "step": 893
+    },
+    {
+      "epoch": 0.2574228884404728,
+      "loss": 0.028818532824516296,
+      "loss_ce": 0.0006965833017602563,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 131658112,
+      "step": 893
+    },
+    {
+      "epoch": 0.25771115595272415,
+      "grad_norm": 1.0845714678383525,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "num_input_tokens_seen": 131830608,
+      "step": 894
+    },
+    {
+      "epoch": 0.25771115595272415,
+      "loss": 0.006088267080485821,
+      "loss_ce": 0.0015697581693530083,
+      "loss_xval": 0.0045166015625,
+      "num_input_tokens_seen": 131830608,
+      "step": 894
+    },
+    {
+      "epoch": 0.2579994234649755,
+      "grad_norm": 16.644940036843188,
+      "learning_rate": 0.0001,
+      "loss": 0.0289,
+      "num_input_tokens_seen": 131965448,
+      "step": 895
+    },
+    {
+      "epoch": 0.2579994234649755,
+      "loss": 0.026386450976133347,
+      "loss_ce": 0.014148902148008347,
+      "loss_xval": 0.01220703125,
+      "num_input_tokens_seen": 131965448,
+      "step": 895
+    },
+    {
+      "epoch": 0.25828769097722687,
+      "grad_norm": 17.960460147753857,
+      "learning_rate": 0.0001,
+      "loss": 0.026,
+      "num_input_tokens_seen": 132100560,
+      "step": 896
+    },
+    {
+      "epoch": 0.25828769097722687,
+      "loss": 0.03656185418367386,
+      "loss_ce": 0.0006731800967827439,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 132100560,
+      "step": 896
+    },
+    {
+      "epoch": 0.25857595848947823,
+      "grad_norm": 5.864950391364441,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "num_input_tokens_seen": 132273072,
+      "step": 897
+    },
+    {
+      "epoch": 0.25857595848947823,
+      "loss": 0.017272870987653732,
+      "loss_ce": 0.0020751184783875942,
+      "loss_xval": 0.01519775390625,
+      "num_input_tokens_seen": 132273072,
+      "step": 897
+    },
+    {
+      "epoch": 0.2588642260017296,
+      "grad_norm": 5.535883983296717,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "num_input_tokens_seen": 132407760,
+      "step": 898
+    },
+    {
+      "epoch": 0.2588642260017296,
+      "loss": 0.01192381139844656,
+      "loss_ce": 0.010679743252694607,
+      "loss_xval": 0.00124359130859375,
+      "num_input_tokens_seen": 132407760,
+      "step": 898
+    },
+    {
+      "epoch": 0.25915249351398095,
+      "grad_norm": 2.705241713165992,
+      "learning_rate": 0.0001,
+      "loss": 0.0039,
+      "num_input_tokens_seen": 132542840,
+      "step": 899
+    },
+    {
+      "epoch": 0.25915249351398095,
+      "loss": 0.004840262699872255,
+      "loss_ce": 0.000699408701620996,
+      "loss_xval": 0.004150390625,
+      "num_input_tokens_seen": 132542840,
+      "step": 899
+    },
+    {
+      "epoch": 0.25944076102623237,
+      "grad_norm": 12.68596683547975,
+      "learning_rate": 0.0001,
+      "loss": 0.0286,
+      "num_input_tokens_seen": 132715368,
+      "step": 900
+    },
+    {
+      "epoch": 0.25944076102623237,
+      "loss": 0.010654434561729431,
+      "loss_ce": 0.008203492499887943,
+      "loss_xval": 0.0024566650390625,
+      "num_input_tokens_seen": 132715368,
+      "step": 900
+    },
+    {
+      "epoch": 0.25972902853848373,
+      "grad_norm": 26.31430231024215,
+      "learning_rate": 0.0001,
+      "loss": 0.0572,
+      "num_input_tokens_seen": 132850240,
+      "step": 901
+    },
+    {
+      "epoch": 0.25972902853848373,
+      "loss": 0.046879298985004425,
+      "loss_ce": 0.011906154453754425,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 132850240,
+      "step": 901
+    },
+    {
+      "epoch": 0.2600172960507351,
+      "grad_norm": 26.183178589900663,
+      "learning_rate": 0.0001,
+      "loss": 0.0528,
+      "num_input_tokens_seen": 132985336,
+      "step": 902
+    },
+    {
+      "epoch": 0.2600172960507351,
+      "loss": 0.06659163534641266,
+      "loss_ce": 0.0006736628711223602,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 132985336,
+      "step": 902
+    },
+    {
+      "epoch": 0.26030556356298645,
+      "grad_norm": 14.531282187694886,
+      "learning_rate": 0.0001,
+      "loss": 0.0282,
+      "num_input_tokens_seen": 133157976,
+      "step": 903
+    },
+    {
+      "epoch": 0.26030556356298645,
+      "loss": 0.04092877358198166,
+      "loss_ce": 0.0007676435052417219,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 133157976,
+      "step": 903
+    },
+    {
+      "epoch": 0.2605938310752378,
+      "grad_norm": 4.771268098697512,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "num_input_tokens_seen": 133292824,
+      "step": 904
+    },
+    {
+      "epoch": 0.2605938310752378,
+      "loss": 0.017029020935297012,
+      "loss_ce": 0.009620878845453262,
+      "loss_xval": 0.007415771484375,
+      "num_input_tokens_seen": 133292824,
+      "step": 904
+    },
+    {
+      "epoch": 0.2608820985874892,
+      "grad_norm": 7.83416493572119,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "num_input_tokens_seen": 133428056,
+      "step": 905
+    },
+    {
+      "epoch": 0.2608820985874892,
+      "loss": 0.00283773522824049,
+      "loss_ce": 0.0006976902368478477,
+      "loss_xval": 0.00213623046875,
+      "num_input_tokens_seen": 133428056,
+      "step": 905
+    },
+    {
+      "epoch": 0.26117036609974054,
+      "grad_norm": 21.668543085444146,
+      "learning_rate": 0.0001,
+      "loss": 0.0442,
+      "num_input_tokens_seen": 133600712,
+      "step": 906
+    },
+    {
+      "epoch": 0.26117036609974054,
+      "loss": 0.01621638983488083,
+      "loss_ce": 0.000980487558990717,
+      "loss_xval": 0.0152587890625,
+      "num_input_tokens_seen": 133600712,
+      "step": 906
+    },
+    {
+      "epoch": 0.26145863361199195,
+      "grad_norm": 33.911068651601695,
+      "learning_rate": 0.0001,
+      "loss": 0.0903,
+      "num_input_tokens_seen": 133735480,
+      "step": 907
+    },
+    {
+      "epoch": 0.26145863361199195,
+      "loss": 0.08236135542392731,
+      "loss_ce": 0.011316439136862755,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 133735480,
+      "step": 907
+    },
+    {
+      "epoch": 0.2617469011242433,
+      "grad_norm": 33.76515245614444,
+      "learning_rate": 0.0001,
+      "loss": 0.0865,
+      "num_input_tokens_seen": 133870456,
+      "step": 908
+    },
+    {
+      "epoch": 0.2617469011242433,
+      "loss": 0.10329385101795197,
+      "loss_ce": 0.0006937556900084019,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 133870456,
+      "step": 908
+    },
+    {
+      "epoch": 0.2620351686364947,
+      "grad_norm": 22.571864793058605,
+      "learning_rate": 0.0001,
+      "loss": 0.0465,
+      "num_input_tokens_seen": 134043096,
+      "step": 909
+    },
+    {
+      "epoch": 0.2620351686364947,
+      "loss": 0.06501968204975128,
+      "loss_ce": 0.0009938071016222239,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 134043096,
+      "step": 909
+    },
+    {
+      "epoch": 0.26232343614874604,
+      "grad_norm": 11.579752563407906,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "num_input_tokens_seen": 134177904,
+      "step": 910
+    },
+    {
+      "epoch": 0.26232343614874604,
+      "loss": 0.03199625760316849,
+      "loss_ce": 0.013319501653313637,
+      "loss_xval": 0.0186767578125,
+      "num_input_tokens_seen": 134177904,
+      "step": 910
+    },
+    {
+      "epoch": 0.2626117036609974,
+      "grad_norm": 10.671083819097934,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "num_input_tokens_seen": 134312864,
+      "step": 911
+    },
+    {
+      "epoch": 0.2626117036609974,
+      "loss": 0.006020676344633102,
+      "loss_ce": 0.0007373203989118338,
+      "loss_xval": 0.005279541015625,
+      "num_input_tokens_seen": 134312864,
+      "step": 911
+    },
+    {
+      "epoch": 0.26289997117324876,
+      "grad_norm": 17.29284770364731,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "num_input_tokens_seen": 134485392,
+      "step": 912
+    },
+    {
+      "epoch": 0.26289997117324876,
+      "loss": 0.011055695824325085,
+      "loss_ce": 0.0016448370879516006,
+      "loss_xval": 0.0093994140625,
+      "num_input_tokens_seen": 134485392,
+      "step": 912
+    },
+    {
+      "epoch": 0.2631882386855001,
+      "grad_norm": 20.517777314388695,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "num_input_tokens_seen": 134620280,
+      "step": 913
+    },
+    {
+      "epoch": 0.2631882386855001,
+      "loss": 0.03527616709470749,
+      "loss_ce": 0.009626143611967564,
+      "loss_xval": 0.025634765625,
+      "num_input_tokens_seen": 134620280,
+      "step": 913
+    },
+    {
+      "epoch": 0.26347650619775154,
+      "grad_norm": 13.194845052998755,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "num_input_tokens_seen": 134755240,
+      "step": 914
+    },
+    {
+      "epoch": 0.26347650619775154,
+      "loss": 0.02211439609527588,
+      "loss_ce": 0.0007368340156972408,
+      "loss_xval": 0.0213623046875,
+      "num_input_tokens_seen": 134755240,
+      "step": 914
+    },
+    {
+      "epoch": 0.2637647737100029,
+      "grad_norm": 1.4668499046176524,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "num_input_tokens_seen": 134927608,
+      "step": 915
+    },
+    {
+      "epoch": 0.2637647737100029,
+      "loss": 0.00330064888112247,
+      "loss_ce": 0.0014142810832709074,
+      "loss_xval": 0.00188446044921875,
+      "num_input_tokens_seen": 134927608,
+      "step": 915
+    },
+    {
+      "epoch": 0.26405304122225426,
+      "grad_norm": 13.229827627513,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "num_input_tokens_seen": 135062368,
+      "step": 916
+    },
+    {
+      "epoch": 0.26405304122225426,
+      "loss": 0.021753720939159393,
+      "loss_ce": 0.010431700386106968,
+      "loss_xval": 0.0113525390625,
+      "num_input_tokens_seen": 135062368,
+      "step": 916
+    },
+    {
+      "epoch": 0.2643413087345056,
+      "grad_norm": 15.10165445453856,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "num_input_tokens_seen": 135197440,
+      "step": 917
+    },
+    {
+      "epoch": 0.2643413087345056,
+      "loss": 0.027548931539058685,
+      "loss_ce": 0.0007239800179377198,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 135197440,
+      "step": 917
+    },
+    {
+      "epoch": 0.264629576246757,
+      "grad_norm": 10.113371750330549,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "num_input_tokens_seen": 135369888,
+      "step": 918
+    },
+    {
+      "epoch": 0.264629576246757,
+      "loss": 0.019287362694740295,
+      "loss_ce": 0.0010988865979015827,
+      "loss_xval": 0.0181884765625,
+      "num_input_tokens_seen": 135369888,
+      "step": 918
+    },
+    {
+      "epoch": 0.26491784375900834,
+      "grad_norm": 7.503561052273172,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "num_input_tokens_seen": 135504720,
+      "step": 919
+    },
+    {
+      "epoch": 0.26491784375900834,
+      "loss": 0.014727404341101646,
+      "loss_ce": 0.007189563009887934,
+      "loss_xval": 0.007537841796875,
+      "num_input_tokens_seen": 135504720,
+      "step": 919
+    },
+    {
+      "epoch": 0.2652061112712597,
+      "grad_norm": 12.942512436123494,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "num_input_tokens_seen": 135639816,
+      "step": 920
+    },
+    {
+      "epoch": 0.2652061112712597,
+      "loss": 0.009108353406190872,
+      "loss_ce": 0.0009105693316087127,
+      "loss_xval": 0.0081787109375,
+      "num_input_tokens_seen": 135639816,
+      "step": 920
+    },
+    {
+      "epoch": 0.2654943787835111,
+      "grad_norm": 23.22476813528809,
+      "learning_rate": 0.0001,
+      "loss": 0.049,
+      "num_input_tokens_seen": 135812408,
+      "step": 921
+    },
+    {
+      "epoch": 0.2654943787835111,
+      "loss": 0.03185856714844704,
+      "loss_ce": 0.0026837619952857494,
+      "loss_xval": 0.0291748046875,
+      "num_input_tokens_seen": 135812408,
+      "step": 921
+    },
+    {
+      "epoch": 0.2657826462957625,
+      "grad_norm": 30.64846542366728,
+      "learning_rate": 0.0001,
+      "loss": 0.0798,
+      "num_input_tokens_seen": 135947240,
+      "step": 922
+    },
+    {
+      "epoch": 0.2657826462957625,
+      "loss": 0.08230604231357574,
+      "loss_ce": 0.013214251026511192,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 135947240,
+      "step": 922
+    },
+    {
+      "epoch": 0.26607091380801384,
+      "grad_norm": 30.414646214325735,
+      "learning_rate": 0.0001,
+      "loss": 0.0749,
+      "num_input_tokens_seen": 136082312,
+      "step": 923
+    },
+    {
+      "epoch": 0.26607091380801384,
+      "loss": 0.08990393579006195,
+      "loss_ce": 0.0007315789698623121,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 136082312,
+      "step": 923
+    },
+    {
+      "epoch": 0.2663591813202652,
+      "grad_norm": 24.71366898539121,
+      "learning_rate": 0.0001,
+      "loss": 0.0544,
+      "num_input_tokens_seen": 136254824,
+      "step": 924
+    },
+    {
+      "epoch": 0.2663591813202652,
+      "loss": 0.06479089707136154,
+      "loss_ce": 0.0011007109424099326,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 136254824,
+      "step": 924
+    },
+    {
+      "epoch": 0.26664744883251656,
+      "grad_norm": 20.403882779981654,
+      "learning_rate": 0.0001,
+      "loss": 0.0387,
+      "num_input_tokens_seen": 136389768,
+      "step": 925
+    },
+    {
+      "epoch": 0.26664744883251656,
+      "loss": 0.04584236443042755,
+      "loss_ce": 0.008092120289802551,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 136389768,
+      "step": 925
+    },
+    {
+      "epoch": 0.2669357163447679,
+      "grad_norm": 21.71097006562949,
+      "learning_rate": 0.0001,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 136524832,
+      "step": 926
+    },
+    {
+      "epoch": 0.2669357163447679,
+      "loss": 0.03015214577317238,
+      "loss_ce": 0.0007484586676582694,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 136524832,
+      "step": 926
+    },
+    {
+      "epoch": 0.2672239838570193,
+      "grad_norm": 25.048860969375237,
+      "learning_rate": 0.0001,
+      "loss": 0.0588,
+      "num_input_tokens_seen": 136697280,
+      "step": 927
+    },
+    {
+      "epoch": 0.2672239838570193,
+      "loss": 0.039126113057136536,
+      "loss_ce": 0.0023066524881869555,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 136697280,
+      "step": 927
+    },
+    {
+      "epoch": 0.2675122513692707,
+      "grad_norm": 23.604295185328763,
+      "learning_rate": 0.0001,
+      "loss": 0.051,
+      "num_input_tokens_seen": 136832016,
+      "step": 928
+    },
+    {
+      "epoch": 0.2675122513692707,
+      "loss": 0.05441943556070328,
+      "loss_ce": 0.009802735410630703,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 136832016,
+      "step": 928
+    },
+    {
+      "epoch": 0.26780051888152206,
+      "grad_norm": 14.373659940022728,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "num_input_tokens_seen": 136967048,
+      "step": 929
+    },
+    {
+      "epoch": 0.26780051888152206,
+      "loss": 0.026713449507951736,
+      "loss_ce": 0.0007887678220868111,
+      "loss_xval": 0.02587890625,
+      "num_input_tokens_seen": 136967048,
+      "step": 929
+    },
+    {
+      "epoch": 0.2680887863937734,
+      "grad_norm": 1.625112853934714,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "num_input_tokens_seen": 137139536,
+      "step": 930
+    },
+    {
+      "epoch": 0.2680887863937734,
+      "loss": 0.004050274379551411,
+      "loss_ce": 0.0012865260941907763,
+      "loss_xval": 0.0027618408203125,
+      "num_input_tokens_seen": 137139536,
+      "step": 930
+    },
+    {
+      "epoch": 0.2683770539060248,
+      "grad_norm": 7.604170035840317,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "num_input_tokens_seen": 137274200,
+      "step": 931
+    },
+    {
+      "epoch": 0.2683770539060248,
+      "loss": 0.017092669382691383,
+      "loss_ce": 0.010817492380738258,
+      "loss_xval": 0.00628662109375,
+      "num_input_tokens_seen": 137274200,
+      "step": 931
+    },
+    {
+      "epoch": 0.26866532141827615,
+      "grad_norm": 9.838193297363524,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "num_input_tokens_seen": 137409296,
+      "step": 932
+    },
+    {
+      "epoch": 0.26866532141827615,
+      "loss": 0.01571030355989933,
+      "loss_ce": 0.0009321661200374365,
+      "loss_xval": 0.0147705078125,
+      "num_input_tokens_seen": 137409296,
+      "step": 932
+    },
+    {
+      "epoch": 0.2689535889305275,
+      "grad_norm": 8.518101589222372,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "num_input_tokens_seen": 137581768,
+      "step": 933
+    },
+    {
+      "epoch": 0.2689535889305275,
+      "loss": 0.012986189685761929,
+      "loss_ce": 0.001626020995900035,
+      "loss_xval": 0.0113525390625,
+      "num_input_tokens_seen": 137581768,
+      "step": 933
+    },
+    {
+      "epoch": 0.2692418564427789,
+      "grad_norm": 9.815786119364088,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "num_input_tokens_seen": 137716504,
+      "step": 934
+    },
+    {
+      "epoch": 0.2692418564427789,
+      "loss": 0.021232970058918,
+      "loss_ce": 0.011879331432282925,
+      "loss_xval": 0.00933837890625,
+      "num_input_tokens_seen": 137716504,
+      "step": 934
+    },
+    {
+      "epoch": 0.2695301239550303,
+      "grad_norm": 15.504062175280504,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "num_input_tokens_seen": 137851616,
+      "step": 935
+    },
+    {
+      "epoch": 0.2695301239550303,
+      "loss": 0.015094293281435966,
+      "loss_ce": 0.0007739191642031074,
+      "loss_xval": 0.01434326171875,
+      "num_input_tokens_seen": 137851616,
+      "step": 935
+    },
+    {
+      "epoch": 0.26981839146728165,
+      "grad_norm": 21.816170072043693,
+      "learning_rate": 0.0001,
+      "loss": 0.0501,
+      "num_input_tokens_seen": 138024032,
+      "step": 936
+    },
+    {
+      "epoch": 0.26981839146728165,
+      "loss": 0.03761507570743561,
+      "loss_ce": 0.006273522041738033,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 138024032,
+      "step": 936
+    },
+    {
+      "epoch": 0.270106658979533,
+      "grad_norm": 24.016219232791897,
+      "learning_rate": 0.0001,
+      "loss": 0.054,
+      "num_input_tokens_seen": 138158760,
+      "step": 937
+    },
+    {
+      "epoch": 0.270106658979533,
+      "loss": 0.057084821164608,
+      "loss_ce": 0.008043071255087852,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 138158760,
+      "step": 937
+    },
+    {
+      "epoch": 0.27039492649178437,
+      "grad_norm": 20.950077621507994,
+      "learning_rate": 0.0001,
+      "loss": 0.0387,
+      "num_input_tokens_seen": 138293760,
+      "step": 938
+    },
+    {
+      "epoch": 0.27039492649178437,
+      "loss": 0.046418577432632446,
+      "loss_ce": 0.0007947962731122971,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 138293760,
+      "step": 938
+    },
+    {
+      "epoch": 0.27068319400403573,
+      "grad_norm": 16.409492997422056,
+      "learning_rate": 0.0001,
+      "loss": 0.0306,
+      "num_input_tokens_seen": 138466136,
+      "step": 939
+    },
+    {
+      "epoch": 0.27068319400403573,
+      "loss": 0.030278921127319336,
+      "loss_ce": 0.0010888581164181232,
+      "loss_xval": 0.0291748046875,
+      "num_input_tokens_seen": 138466136,
+      "step": 939
+    },
+    {
+      "epoch": 0.2709714615162871,
+      "grad_norm": 15.24922810529818,
+      "learning_rate": 0.0001,
+      "loss": 0.0268,
+      "num_input_tokens_seen": 138600880,
+      "step": 940
+    },
+    {
+      "epoch": 0.2709714615162871,
+      "loss": 0.03236062824726105,
+      "loss_ce": 0.012325836345553398,
+      "loss_xval": 0.02001953125,
+      "num_input_tokens_seen": 138600880,
+      "step": 940
+    },
+    {
+      "epoch": 0.2712597290285385,
+      "grad_norm": 19.295290220045818,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "num_input_tokens_seen": 138735968,
+      "step": 941
+    },
+    {
+      "epoch": 0.2712597290285385,
+      "loss": 0.025427913293242455,
+      "loss_ce": 0.0008002271642908454,
+      "loss_xval": 0.024658203125,
+      "num_input_tokens_seen": 138735968,
+      "step": 941
+    },
+    {
+      "epoch": 0.27154799654078987,
+      "grad_norm": 25.769955372399583,
+      "learning_rate": 0.0001,
+      "loss": 0.0632,
+      "num_input_tokens_seen": 138908344,
+      "step": 942
+    },
+    {
+      "epoch": 0.27154799654078987,
+      "loss": 0.04995029419660568,
+      "loss_ce": 0.0012747579021379352,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 138908344,
+      "step": 942
+    },
+    {
+      "epoch": 0.27183626405304123,
+      "grad_norm": 29.89708338319636,
+      "learning_rate": 0.0001,
+      "loss": 0.0834,
+      "num_input_tokens_seen": 139043064,
+      "step": 943
+    },
+    {
+      "epoch": 0.27183626405304123,
+      "loss": 0.09044340252876282,
+      "loss_ce": 0.012196330353617668,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 139043064,
+      "step": 943
+    },
+    {
+      "epoch": 0.2721245315652926,
+      "grad_norm": 29.221058418100938,
+      "learning_rate": 0.0001,
+      "loss": 0.0752,
+      "num_input_tokens_seen": 139178080,
+      "step": 944
+    },
+    {
+      "epoch": 0.2721245315652926,
+      "loss": 0.0861777663230896,
+      "loss_ce": 0.0012168283574283123,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 139178080,
+      "step": 944
+    },
+    {
+      "epoch": 0.27241279907754395,
+      "grad_norm": 26.289080802161784,
+      "learning_rate": 0.0001,
+      "loss": 0.0653,
+      "num_input_tokens_seen": 139350632,
+      "step": 945
+    },
+    {
+      "epoch": 0.27241279907754395,
+      "loss": 0.07178245484828949,
+      "loss_ce": 0.0016530563589185476,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 139350632,
+      "step": 945
+    },
+    {
+      "epoch": 0.2727010665897953,
+      "grad_norm": 24.600247355148554,
+      "learning_rate": 0.0001,
+      "loss": 0.0575,
+      "num_input_tokens_seen": 139485448,
+      "step": 946
+    },
+    {
+      "epoch": 0.2727010665897953,
+      "loss": 0.060692962259054184,
+      "loss_ce": 0.007531340233981609,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 139485448,
+      "step": 946
+    },
+    {
+      "epoch": 0.2729893341020467,
+      "grad_norm": 25.384371769491704,
+      "learning_rate": 0.0001,
+      "loss": 0.0584,
+      "num_input_tokens_seen": 139620448,
+      "step": 947
+    },
+    {
+      "epoch": 0.2729893341020467,
+      "loss": 0.04900098592042923,
+      "loss_ce": 0.0009663160308264196,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 139620448,
+      "step": 947
+    },
+    {
+      "epoch": 0.2732776016142981,
+      "grad_norm": 25.47315877753157,
+      "learning_rate": 0.0001,
+      "loss": 0.0646,
+      "num_input_tokens_seen": 139792856,
+      "step": 948
+    },
+    {
+      "epoch": 0.2732776016142981,
+      "loss": 0.049740906804800034,
+      "loss_ce": 0.0014315814478322864,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 139792856,
+      "step": 948
+    },
+    {
+      "epoch": 0.27356586912654945,
+      "grad_norm": 20.431395255995,
+      "learning_rate": 0.0001,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 139927664,
+      "step": 949
+    },
+    {
+      "epoch": 0.27356586912654945,
+      "loss": 0.05533936247229576,
+      "loss_ce": 0.01728394255042076,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 139927664,
+      "step": 949
+    },
+    {
+      "epoch": 0.2738541366388008,
+      "grad_norm": 9.615385475945006,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "num_input_tokens_seen": 140062672,
+      "step": 950
+    },
+    {
+      "epoch": 0.2738541366388008,
+      "loss": 0.013754522427916527,
+      "loss_ce": 0.0008455874631181359,
+      "loss_xval": 0.012939453125,
+      "num_input_tokens_seen": 140062672,
+      "step": 950
+    },
+    {
+      "epoch": 0.2741424041510522,
+      "grad_norm": 2.7294140895326406,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "num_input_tokens_seen": 140235104,
+      "step": 951
+    },
+    {
+      "epoch": 0.2741424041510522,
+      "loss": 0.003939633257687092,
+      "loss_ce": 0.002550129545852542,
+      "loss_xval": 0.0013885498046875,
+      "num_input_tokens_seen": 140235104,
+      "step": 951
+    },
+    {
+      "epoch": 0.27443067166330354,
+      "grad_norm": 10.861198971518537,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "num_input_tokens_seen": 140369880,
+      "step": 952
+    },
+    {
+      "epoch": 0.27443067166330354,
+      "loss": 0.027800941839814186,
+      "loss_ce": 0.01453342568129301,
+      "loss_xval": 0.01324462890625,
+      "num_input_tokens_seen": 140369880,
+      "step": 952
+    },
+    {
+      "epoch": 0.2747189391755549,
+      "grad_norm": 13.40011589641844,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "num_input_tokens_seen": 140504904,
+      "step": 953
+    },
+    {
+      "epoch": 0.2747189391755549,
+      "loss": 0.02331135794520378,
+      "loss_ce": 0.0008504212601110339,
+      "loss_xval": 0.0224609375,
+      "num_input_tokens_seen": 140504904,
+      "step": 953
+    },
+    {
+      "epoch": 0.27500720668780626,
+      "grad_norm": 12.303345775184155,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "num_input_tokens_seen": 140677608,
+      "step": 954
+    },
+    {
+      "epoch": 0.27500720668780626,
+      "loss": 0.02076645940542221,
+      "loss_ce": 0.0022804364562034607,
+      "loss_xval": 0.0184326171875,
+      "num_input_tokens_seen": 140677608,
+      "step": 954
+    },
+    {
+      "epoch": 0.2752954742000577,
+      "grad_norm": 10.961128742719117,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "num_input_tokens_seen": 140812384,
+      "step": 955
+    },
+    {
+      "epoch": 0.2752954742000577,
+      "loss": 0.02713668718934059,
+      "loss_ce": 0.014540554955601692,
+      "loss_xval": 0.0125732421875,
+      "num_input_tokens_seen": 140812384,
+      "step": 955
+    },
+    {
+      "epoch": 0.27558374171230904,
+      "grad_norm": 11.45692148425062,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "num_input_tokens_seen": 140947408,
+      "step": 956
+    },
+    {
+      "epoch": 0.27558374171230904,
+      "loss": 0.011002715677022934,
+      "loss_ce": 0.0008022149559110403,
+      "loss_xval": 0.01019287109375,
+      "num_input_tokens_seen": 140947408,
+      "step": 956
+    },
+    {
+      "epoch": 0.2758720092245604,
+      "grad_norm": 12.276193654056682,
+      "learning_rate": 0.0001,
+      "loss": 0.0219,
+      "num_input_tokens_seen": 141119968,
+      "step": 957
+    },
+    {
+      "epoch": 0.2758720092245604,
+      "loss": 0.012618216685950756,
+      "loss_ce": 0.0017005529953166842,
+      "loss_xval": 0.01092529296875,
+      "num_input_tokens_seen": 141119968,
+      "step": 957
+    },
+    {
+      "epoch": 0.27616027673681176,
+      "grad_norm": 10.322703497206954,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "num_input_tokens_seen": 141254712,
+      "step": 958
+    },
+    {
+      "epoch": 0.27616027673681176,
+      "loss": 0.022379161790013313,
+      "loss_ce": 0.011713268235325813,
+      "loss_xval": 0.01068115234375,
+      "num_input_tokens_seen": 141254712,
+      "step": 958
+    },
+    {
+      "epoch": 0.2764485442490631,
+      "grad_norm": 5.142704767137008,
+      "learning_rate": 0.0001,
+      "loss": 0.0041,
+      "num_input_tokens_seen": 141389776,
+      "step": 959
+    },
+    {
+      "epoch": 0.2764485442490631,
+      "loss": 0.005325574427843094,
+      "loss_ce": 0.0007574743358418345,
+      "loss_xval": 0.00457763671875,
+      "num_input_tokens_seen": 141389776,
+      "step": 959
+    },
+    {
+      "epoch": 0.2767368117613145,
+      "grad_norm": 0.8162921258077296,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "num_input_tokens_seen": 141562256,
+      "step": 960
+    },
+    {
+      "epoch": 0.2767368117613145,
+      "loss": 0.0015741356182843447,
+      "loss_ce": 0.0009709365549497306,
+      "loss_xval": 0.00060272216796875,
+      "num_input_tokens_seen": 141562256,
+      "step": 960
+    },
+    {
+      "epoch": 0.27702507927356584,
+      "grad_norm": 4.287498292630918,
+      "learning_rate": 0.0001,
+      "loss": 0.006,
+      "num_input_tokens_seen": 141697080,
+      "step": 961
+    },
+    {
+      "epoch": 0.27702507927356584,
+      "loss": 0.008994442410767078,
+      "loss_ce": 0.006450039334595203,
+      "loss_xval": 0.0025482177734375,
+      "num_input_tokens_seen": 141697080,
+      "step": 961
+    },
+    {
+      "epoch": 0.27731334678581726,
+      "grad_norm": 4.414203081683492,
+      "learning_rate": 0.0001,
+      "loss": 0.0038,
+      "num_input_tokens_seen": 141832144,
+      "step": 962
+    },
+    {
+      "epoch": 0.27731334678581726,
+      "loss": 0.004648888483643532,
+      "loss_ce": 0.000725472578778863,
+      "loss_xval": 0.003936767578125,
+      "num_input_tokens_seen": 141832144,
+      "step": 962
+    },
+    {
+      "epoch": 0.2776016142980686,
+      "grad_norm": 3.131954367557028,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "num_input_tokens_seen": 142004616,
+      "step": 963
+    },
+    {
+      "epoch": 0.2776016142980686,
+      "loss": 0.0035226130858063698,
+      "loss_ce": 0.0009610439883545041,
+      "loss_xval": 0.0025634765625,
+      "num_input_tokens_seen": 142004616,
+      "step": 963
+    },
+    {
+      "epoch": 0.27788988181032,
+      "grad_norm": 3.7388693449607406,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "num_input_tokens_seen": 142139456,
+      "step": 964
+    },
+    {
+      "epoch": 0.27788988181032,
+      "loss": 0.013125386089086533,
+      "loss_ce": 0.010863270610570908,
+      "loss_xval": 0.00225830078125,
+      "num_input_tokens_seen": 142139456,
+      "step": 964
+    },
+    {
+      "epoch": 0.27817814932257134,
+      "grad_norm": 7.853043705392883,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "num_input_tokens_seen": 142274552,
+      "step": 965
+    },
+    {
+      "epoch": 0.27817814932257134,
+      "loss": 0.0050224545411765575,
+      "loss_ce": 0.0006698851939290762,
+      "loss_xval": 0.004364013671875,
+      "num_input_tokens_seen": 142274552,
+      "step": 965
+    },
+    {
+      "epoch": 0.2784664168348227,
+      "grad_norm": 13.798301364906187,
+      "learning_rate": 0.0001,
+      "loss": 0.0246,
+      "num_input_tokens_seen": 142446992,
+      "step": 966
+    },
+    {
+      "epoch": 0.2784664168348227,
+      "loss": 0.01616474986076355,
+      "loss_ce": 0.0009364773286506534,
+      "loss_xval": 0.0152587890625,
+      "num_input_tokens_seen": 142446992,
+      "step": 966
+    },
+    {
+      "epoch": 0.27875468434707407,
+      "grad_norm": 21.11692774117553,
+      "learning_rate": 0.0001,
+      "loss": 0.048,
+      "num_input_tokens_seen": 142581824,
+      "step": 967
+    },
+    {
+      "epoch": 0.27875468434707407,
+      "loss": 0.05509456992149353,
+      "loss_ce": 0.011607019230723381,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 142581824,
+      "step": 967
+    },
+    {
+      "epoch": 0.2790429518593254,
+      "grad_norm": 30.422620769917025,
+      "learning_rate": 0.0001,
+      "loss": 0.0882,
+      "num_input_tokens_seen": 142716784,
+      "step": 968
+    },
+    {
+      "epoch": 0.2790429518593254,
+      "loss": 0.09613828361034393,
+      "loss_ce": 0.0006793015636503696,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 142716784,
+      "step": 968
+    },
+    {
+      "epoch": 0.27933121937157684,
+      "grad_norm": 44.02363731095848,
+      "learning_rate": 0.0001,
+      "loss": 0.1874,
+      "num_input_tokens_seen": 142889200,
+      "step": 969
+    },
+    {
+      "epoch": 0.27933121937157684,
+      "loss": 0.19272679090499878,
+      "loss_ce": 0.0013205313589423895,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 142889200,
+      "step": 969
+    },
+    {
+      "epoch": 0.2796194868838282,
+      "grad_norm": 62.57380169348872,
+      "learning_rate": 0.0001,
+      "loss": 0.379,
+      "num_input_tokens_seen": 143023960,
+      "step": 970
+    },
+    {
+      "epoch": 0.2796194868838282,
+      "loss": 0.39476674795150757,
+      "loss_ce": 0.010245269164443016,
+      "loss_xval": 0.384765625,
+      "num_input_tokens_seen": 143023960,
+      "step": 970
+    },
+    {
+      "epoch": 0.27990775439607957,
+      "grad_norm": 80.40325261016795,
+      "learning_rate": 0.0001,
+      "loss": 0.6164,
+      "num_input_tokens_seen": 143158960,
+      "step": 971
+    },
+    {
+      "epoch": 0.27990775439607957,
+      "loss": 0.5973809957504272,
+      "loss_ce": 0.000701274024322629,
+      "loss_xval": 0.59765625,
+      "num_input_tokens_seen": 143158960,
+      "step": 971
+    },
+    {
+      "epoch": 0.28019602190833093,
+      "grad_norm": 76.35630842860024,
+      "learning_rate": 0.0001,
+      "loss": 0.5774,
+      "num_input_tokens_seen": 143331528,
+      "step": 972
+    },
+    {
+      "epoch": 0.28019602190833093,
+      "loss": 0.5543111562728882,
+      "loss_ce": 0.0010884751100093126,
+      "loss_xval": 0.5546875,
+      "num_input_tokens_seen": 143331528,
+      "step": 972
+    },
+    {
+      "epoch": 0.2804842894205823,
+      "grad_norm": 33.64460824242602,
+      "learning_rate": 0.0001,
+      "loss": 0.1209,
+      "num_input_tokens_seen": 143466336,
+      "step": 973
+    },
+    {
+      "epoch": 0.2804842894205823,
+      "loss": 0.128205806016922,
+      "loss_ce": 0.010774169117212296,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 143466336,
+      "step": 973
+    },
+    {
+      "epoch": 0.28077255693283365,
+      "grad_norm": 26.28356998850835,
+      "learning_rate": 0.0001,
+      "loss": 0.0743,
+      "num_input_tokens_seen": 143601336,
+      "step": 974
+    },
+    {
+      "epoch": 0.28077255693283365,
+      "loss": 0.0688517838716507,
+      "loss_ce": 0.0009196547325700521,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 143601336,
+      "step": 974
+    },
+    {
+      "epoch": 0.281060824445085,
+      "grad_norm": 57.38440801081276,
+      "learning_rate": 0.0001,
+      "loss": 0.3491,
+      "num_input_tokens_seen": 143773952,
+      "step": 975
+    },
+    {
+      "epoch": 0.281060824445085,
+      "loss": 0.35013845562934875,
+      "loss_ce": 0.004191188141703606,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 143773952,
+      "step": 975
+    },
+    {
+      "epoch": 0.28134909195733643,
+      "grad_norm": 33.886311086551565,
+      "learning_rate": 0.0001,
+      "loss": 0.1299,
+      "num_input_tokens_seen": 143908696,
+      "step": 976
+    },
+    {
+      "epoch": 0.28134909195733643,
+      "loss": 0.15857642889022827,
+      "loss_ce": 0.013068612664937973,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 143908696,
+      "step": 976
+    },
+    {
+      "epoch": 0.2816373594695878,
+      "grad_norm": 17.302379582721162,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "num_input_tokens_seen": 144043712,
+      "step": 977
+    },
+    {
+      "epoch": 0.2816373594695878,
+      "loss": 0.027677983045578003,
+      "loss_ce": 0.001127689378336072,
+      "loss_xval": 0.026611328125,
+      "num_input_tokens_seen": 144043712,
+      "step": 977
+    },
+    {
+      "epoch": 0.28192562698183915,
+      "grad_norm": 41.012904645947934,
+      "learning_rate": 0.0001,
+      "loss": 0.1911,
+      "num_input_tokens_seen": 144216272,
+      "step": 978
+    },
+    {
+      "epoch": 0.28192562698183915,
+      "loss": 0.2025768756866455,
+      "loss_ce": 0.003480201121419668,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 144216272,
+      "step": 978
+    },
+    {
+      "epoch": 0.2822138944940905,
+      "grad_norm": 15.54293449596175,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "num_input_tokens_seen": 144351040,
+      "step": 979
+    },
+    {
+      "epoch": 0.2822138944940905,
+      "loss": 0.05929769203066826,
+      "loss_ce": 0.009187827818095684,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 144351040,
+      "step": 979
+    },
+    {
+      "epoch": 0.2825021620063419,
+      "grad_norm": 22.751053245751585,
+      "learning_rate": 0.0001,
+      "loss": 0.0628,
+      "num_input_tokens_seen": 144486072,
+      "step": 980
+    },
+    {
+      "epoch": 0.2825021620063419,
+      "loss": 0.04631127417087555,
+      "loss_ce": 0.0014504313003271818,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 144486072,
+      "step": 980
+    },
+    {
+      "epoch": 0.28279042951859323,
+      "grad_norm": 24.906885248483842,
+      "learning_rate": 0.0001,
+      "loss": 0.0798,
+      "num_input_tokens_seen": 144658504,
+      "step": 981
+    },
+    {
+      "epoch": 0.28279042951859323,
+      "loss": 0.09189799427986145,
+      "loss_ce": 0.0021763185504823923,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 144658504,
+      "step": 981
+    },
+    {
+      "epoch": 0.28307869703084465,
+      "grad_norm": 9.041544339456737,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "num_input_tokens_seen": 144793368,
+      "step": 982
+    },
+    {
+      "epoch": 0.28307869703084465,
+      "loss": 0.015240021049976349,
+      "loss_ce": 0.011438675224781036,
+      "loss_xval": 0.0037994384765625,
+      "num_input_tokens_seen": 144793368,
+      "step": 982
+    },
+    {
+      "epoch": 0.283366964543096,
+      "grad_norm": 29.264667801560147,
+      "learning_rate": 0.0001,
+      "loss": 0.1046,
+      "num_input_tokens_seen": 144928696,
+      "step": 983
+    },
+    {
+      "epoch": 0.283366964543096,
+      "loss": 0.07928408682346344,
+      "loss_ce": 0.0017084057908505201,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 144928696,
+      "step": 983
+    },
+    {
+      "epoch": 0.2836552320553474,
+      "grad_norm": 3.9211013715650176,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "num_input_tokens_seen": 145101192,
+      "step": 984
+    },
+    {
+      "epoch": 0.2836552320553474,
+      "loss": 0.011925269849598408,
+      "loss_ce": 0.002686073537915945,
+      "loss_xval": 0.00921630859375,
+      "num_input_tokens_seen": 145101192,
+      "step": 984
+    },
+    {
+      "epoch": 0.28394349956759873,
+      "grad_norm": 33.30285734842819,
+      "learning_rate": 0.0001,
+      "loss": 0.1483,
+      "num_input_tokens_seen": 145235976,
+      "step": 985
+    },
+    {
+      "epoch": 0.28394349956759873,
+      "loss": 0.0838775783777237,
+      "loss_ce": 0.008682267740368843,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 145235976,
+      "step": 985
+    },
+    {
+      "epoch": 0.2842317670798501,
+      "grad_norm": 25.700668378333628,
+      "learning_rate": 0.0001,
+      "loss": 0.0873,
+      "num_input_tokens_seen": 145371144,
+      "step": 986
+    },
+    {
+      "epoch": 0.2842317670798501,
+      "loss": 0.05714728683233261,
+      "loss_ce": 0.0016968436539173126,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 145371144,
+      "step": 986
+    },
+    {
+      "epoch": 0.28452003459210146,
+      "grad_norm": 26.210793499594022,
+      "learning_rate": 0.0001,
+      "loss": 0.0977,
+      "num_input_tokens_seen": 145543648,
+      "step": 987
+    },
+    {
+      "epoch": 0.28452003459210146,
+      "loss": 0.04943012818694115,
+      "loss_ce": 0.0018532233079895377,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 145543648,
+      "step": 987
+    },
+    {
+      "epoch": 0.2848083021043528,
+      "grad_norm": 53.09264170885138,
+      "learning_rate": 0.0001,
+      "loss": 0.3692,
+      "num_input_tokens_seen": 145678456,
+      "step": 988
+    },
+    {
+      "epoch": 0.2848083021043528,
+      "loss": 0.23942574858665466,
+      "loss_ce": 0.009933562949299812,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 145678456,
+      "step": 988
+    },
+    {
+      "epoch": 0.28509656961660423,
+      "grad_norm": 7.6051148878985035,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "num_input_tokens_seen": 145813592,
+      "step": 989
+    },
+    {
+      "epoch": 0.28509656961660423,
+      "loss": 0.007788852788507938,
+      "loss_ce": 0.0017158547416329384,
+      "loss_xval": 0.006072998046875,
+      "num_input_tokens_seen": 145813592,
+      "step": 989
+    },
+    {
+      "epoch": 0.2853848371288556,
+      "grad_norm": 58.91672638655118,
+      "learning_rate": 0.0001,
+      "loss": 0.4474,
+      "num_input_tokens_seen": 145986328,
+      "step": 990
+    },
+    {
+      "epoch": 0.2853848371288556,
+      "loss": 0.3273075222969055,
+      "loss_ce": 0.002112208865582943,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 145986328,
+      "step": 990
+    },
+    {
+      "epoch": 0.28567310464110696,
+      "grad_norm": 49.15384164290371,
+      "learning_rate": 0.0001,
+      "loss": 0.3252,
+      "num_input_tokens_seen": 146121160,
+      "step": 991
+    },
+    {
+      "epoch": 0.28567310464110696,
+      "loss": 0.23587465286254883,
+      "loss_ce": 0.008579719811677933,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 146121160,
+      "step": 991
+    },
+    {
+      "epoch": 0.2859613721533583,
+      "grad_norm": 31.272820081609783,
+      "learning_rate": 0.0001,
+      "loss": 0.1358,
+      "num_input_tokens_seen": 146256144,
+      "step": 992
+    },
+    {
+      "epoch": 0.2859613721533583,
+      "loss": 0.13801869750022888,
+      "loss_ce": 0.00172718265093863,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 146256144,
+      "step": 992
+    },
+    {
+      "epoch": 0.2862496396656097,
+      "grad_norm": 65.15782599733767,
+      "learning_rate": 0.0001,
+      "loss": 0.5779,
+      "num_input_tokens_seen": 146428640,
+      "step": 993
+    },
+    {
+      "epoch": 0.2862496396656097,
+      "loss": 0.49440979957580566,
+      "loss_ce": 0.0027106208726763725,
+      "loss_xval": 0.4921875,
+      "num_input_tokens_seen": 146428640,
+      "step": 993
+    },
+    {
+      "epoch": 0.28653790717786104,
+      "grad_norm": 0.9831012404283345,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "num_input_tokens_seen": 146563456,
+      "step": 994
+    },
+    {
+      "epoch": 0.28653790717786104,
+      "loss": 0.025697484612464905,
+      "loss_ce": 0.014360204339027405,
+      "loss_xval": 0.0113525390625,
+      "num_input_tokens_seen": 146563456,
+      "step": 994
+    },
+    {
+      "epoch": 0.2868261746901124,
+      "grad_norm": 58.8648520424778,
+      "learning_rate": 0.0001,
+      "loss": 0.4717,
+      "num_input_tokens_seen": 146698480,
+      "step": 995
+    },
+    {
+      "epoch": 0.2868261746901124,
+      "loss": 0.46646779775619507,
+      "loss_ce": 0.0016240678960457444,
+      "loss_xval": 0.46484375,
+      "num_input_tokens_seen": 146698480,
+      "step": 995
+    },
+    {
+      "epoch": 0.2871144422023638,
+      "grad_norm": 17.60872940599237,
+      "learning_rate": 0.0001,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 146871072,
+      "step": 996
+    },
+    {
+      "epoch": 0.2871144422023638,
+      "loss": 0.04825851321220398,
+      "loss_ce": 0.003336637280881405,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 146871072,
+      "step": 996
+    },
+    {
+      "epoch": 0.2874027097146152,
+      "grad_norm": 48.696681592737896,
+      "learning_rate": 0.0001,
+      "loss": 0.3301,
+      "num_input_tokens_seen": 147005872,
+      "step": 997
+    },
+    {
+      "epoch": 0.2874027097146152,
+      "loss": 0.3169158101081848,
+      "loss_ce": 0.010275169275701046,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 147005872,
+      "step": 997
+    },
+    {
+      "epoch": 0.28769097722686654,
+      "grad_norm": 24.725614729868106,
+      "learning_rate": 0.0001,
+      "loss": 0.0927,
+      "num_input_tokens_seen": 147140992,
+      "step": 998
+    },
+    {
+      "epoch": 0.28769097722686654,
+      "loss": 0.09264456480741501,
+      "loss_ce": 0.0017632140079513192,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 147140992,
+      "step": 998
+    },
+    {
+      "epoch": 0.2879792447391179,
+      "grad_norm": 42.92469792044564,
+      "learning_rate": 0.0001,
+      "loss": 0.2791,
+      "num_input_tokens_seen": 147313528,
+      "step": 999
+    },
+    {
+      "epoch": 0.2879792447391179,
+      "loss": 0.22426754236221313,
+      "loss_ce": 0.0030761342495679855,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 147313528,
+      "step": 999
+    },
+    {
+      "epoch": 0.28826751225136926,
+      "grad_norm": 29.84366516038228,
+      "learning_rate": 0.0001,
+      "loss": 0.145,
+      "num_input_tokens_seen": 147448352,
+      "step": 1000
+    },
+    {
+      "epoch": 0.28826751225136926,
+      "eval_websight_new_IoU": 0.0006760411197319627,
+      "eval_websight_new_MAE_x": 0.48339928686618805,
+      "eval_websight_new_MAE_y": 0.4158986359834671,
+      "eval_websight_new_NUM_probability": 0.9786257445812225,
+      "eval_websight_new_inside_bbox": 0.0,
+      "eval_websight_new_loss": 0.21313229203224182,
+      "eval_websight_new_loss_ce": 0.002203845651820302,
+      "eval_websight_new_loss_xval": 0.2103271484375,
+      "eval_websight_new_runtime": 35.3244,
+      "eval_websight_new_samples_per_second": 1.415,
+      "eval_websight_new_steps_per_second": 0.057,
+      "num_input_tokens_seen": 147448352,
+      "step": 1000
+    },
+    {
+      "epoch": 0.28826751225136926,
+      "eval_seeclick_IoU": 0.010920957662165165,
+      "eval_seeclick_MAE_x": 0.3465300649404526,
+      "eval_seeclick_MAE_y": 0.3193105012178421,
+      "eval_seeclick_NUM_probability": 0.980810135602951,
+      "eval_seeclick_inside_bbox": 0.0,
+      "eval_seeclick_loss": 0.1418568640947342,
+      "eval_seeclick_loss_ce": 0.012138516176491976,
+      "eval_seeclick_loss_xval": 0.12896728515625,
+      "eval_seeclick_runtime": 66.5626,
+      "eval_seeclick_samples_per_second": 0.751,
+      "eval_seeclick_steps_per_second": 0.03,
+      "num_input_tokens_seen": 147448352,
+      "step": 1000
+    },
+    {
+      "epoch": 0.28826751225136926,
+      "eval_icons_IoU": 0.0,
+      "eval_icons_MAE_x": 0.41779157519340515,
+      "eval_icons_MAE_y": 0.4029218852519989,
+      "eval_icons_NUM_probability": 0.9756582975387573,
+      "eval_icons_inside_bbox": 0.0,
+      "eval_icons_loss": 0.2196388989686966,
+      "eval_icons_loss_ce": 0.005240184138529003,
+      "eval_icons_loss_xval": 0.21490478515625,
+      "eval_icons_runtime": 63.1802,
+      "eval_icons_samples_per_second": 0.791,
+      "eval_icons_steps_per_second": 0.032,
+      "num_input_tokens_seen": 147448352,
+      "step": 1000
+    },
+    {
+      "epoch": 0.28826751225136926,
+      "loss": 0.22050818800926208,
+      "loss_ce": 0.006885141599923372,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 147448352,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2885557797636206,
+      "grad_norm": 38.16026161198317,
+      "learning_rate": 0.0001,
+      "loss": 0.2163,
+      "num_input_tokens_seen": 147583504,
+      "step": 1001
+    },
+    {
+      "epoch": 0.2885557797636206,
+      "loss": 0.21293914318084717,
+      "loss_ce": 0.0027340645901858807,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 147583504,
+      "step": 1001
+    },
+    {
+      "epoch": 0.288844047275872,
+      "grad_norm": 34.97874588867303,
+      "learning_rate": 0.0001,
+      "loss": 0.2006,
+      "num_input_tokens_seen": 147756064,
+      "step": 1002
+    },
+    {
+      "epoch": 0.288844047275872,
+      "loss": 0.15403415262699127,
+      "loss_ce": 0.0029111080802977085,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 147756064,
+      "step": 1002
+    },
+    {
+      "epoch": 0.2891323147881234,
+      "grad_norm": 29.870015158590114,
+      "learning_rate": 0.0001,
+      "loss": 0.1551,
+      "num_input_tokens_seen": 147890952,
+      "step": 1003
+    },
+    {
+      "epoch": 0.2891323147881234,
+      "loss": 0.1824108362197876,
+      "loss_ce": 0.013831727206707,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 147890952,
+      "step": 1003
+    },
+    {
+      "epoch": 0.28942058230037476,
+      "grad_norm": 52.54062919848932,
+      "learning_rate": 0.0001,
+      "loss": 0.1827,
+      "num_input_tokens_seen": 148026080,
+      "step": 1004
+    },
+    {
+      "epoch": 0.28942058230037476,
+      "loss": 0.1736568808555603,
+      "loss_ce": 0.0031246549915522337,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 148026080,
+      "step": 1004
+    },
+    {
+      "epoch": 0.2897088498126261,
+      "grad_norm": 23.43538994398458,
+      "learning_rate": 0.0001,
+      "loss": 0.0984,
+      "num_input_tokens_seen": 148198608,
+      "step": 1005
+    },
+    {
+      "epoch": 0.2897088498126261,
+      "loss": 0.10520273447036743,
+      "loss_ce": 0.002785742050036788,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 148198608,
+      "step": 1005
+    },
+    {
+      "epoch": 0.2899971173248775,
+      "grad_norm": 24.570694864205628,
+      "learning_rate": 0.0001,
+      "loss": 0.1066,
+      "num_input_tokens_seen": 148333400,
+      "step": 1006
+    },
+    {
+      "epoch": 0.2899971173248775,
+      "loss": 0.1271902620792389,
+      "loss_ce": 0.011772781610488892,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 148333400,
+      "step": 1006
+    },
+    {
+      "epoch": 0.29028538483712885,
+      "grad_norm": 21.721609732412762,
+      "learning_rate": 0.0001,
+      "loss": 0.0815,
+      "num_input_tokens_seen": 148468376,
+      "step": 1007
+    },
+    {
+      "epoch": 0.29028538483712885,
+      "loss": 0.07593949139118195,
+      "loss_ce": 0.0031245506834238768,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 148468376,
+      "step": 1007
+    },
+    {
+      "epoch": 0.2905736523493802,
+      "grad_norm": 19.166959945295666,
+      "learning_rate": 0.0001,
+      "loss": 0.0741,
+      "num_input_tokens_seen": 148640952,
+      "step": 1008
+    },
+    {
+      "epoch": 0.2905736523493802,
+      "loss": 0.07962779700756073,
+      "loss_ce": 0.014259150251746178,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 148640952,
+      "step": 1008
+    },
+    {
+      "epoch": 0.29086191986163157,
+      "grad_norm": 21.311366697639038,
+      "learning_rate": 0.0001,
+      "loss": 0.0842,
+      "num_input_tokens_seen": 148775840,
+      "step": 1009
+    },
+    {
+      "epoch": 0.29086191986163157,
+      "loss": 0.07820656895637512,
+      "loss_ce": 0.013600854203104973,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 148775840,
+      "step": 1009
+    },
+    {
+      "epoch": 0.291150187373883,
+      "grad_norm": 19.015665857193742,
+      "learning_rate": 0.0001,
+      "loss": 0.0639,
+      "num_input_tokens_seen": 148910984,
+      "step": 1010
+    },
+    {
+      "epoch": 0.291150187373883,
+      "loss": 0.05878002196550369,
+      "loss_ce": 0.00336010055616498,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 148910984,
+      "step": 1010
+    },
+    {
+      "epoch": 0.29143845488613435,
+      "grad_norm": 19.15943729010692,
+      "learning_rate": 0.0001,
+      "loss": 0.0723,
+      "num_input_tokens_seen": 149083560,
+      "step": 1011
+    },
+    {
+      "epoch": 0.29143845488613435,
+      "loss": 0.06524774432182312,
+      "loss_ce": 0.005860530771315098,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 149083560,
+      "step": 1011
+    },
+    {
+      "epoch": 0.2917267223983857,
+      "grad_norm": 18.53212069603917,
+      "learning_rate": 0.0001,
+      "loss": 0.0679,
+      "num_input_tokens_seen": 149218400,
+      "step": 1012
+    },
+    {
+      "epoch": 0.2917267223983857,
+      "loss": 0.06433580815792084,
+      "loss_ce": 0.014164908789098263,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 149218400,
+      "step": 1012
+    },
+    {
+      "epoch": 0.29201498991063707,
+      "grad_norm": 16.539777806069416,
+      "learning_rate": 0.0001,
+      "loss": 0.0612,
+      "num_input_tokens_seen": 149353464,
+      "step": 1013
+    },
+    {
+      "epoch": 0.29201498991063707,
+      "loss": 0.0547209233045578,
+      "loss_ce": 0.0029631140641868114,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 149353464,
+      "step": 1013
+    },
+    {
+      "epoch": 0.29230325742288843,
+      "grad_norm": 18.267120678013114,
+      "learning_rate": 0.0001,
+      "loss": 0.0727,
+      "num_input_tokens_seen": 149525984,
+      "step": 1014
+    },
+    {
+      "epoch": 0.29230325742288843,
+      "loss": 0.05868646502494812,
+      "loss_ce": 0.0076000383123755455,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 149525984,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2925915249351398,
+      "grad_norm": 12.986732799219686,
+      "learning_rate": 0.0001,
+      "loss": 0.0425,
+      "num_input_tokens_seen": 149660760,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2925915249351398,
+      "loss": 0.056522078812122345,
+      "loss_ce": 0.01465196069329977,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 149660760,
+      "step": 1015
+    },
+    {
+      "epoch": 0.29287979244739115,
+      "grad_norm": 16.2358668446718,
+      "learning_rate": 0.0001,
+      "loss": 0.052,
+      "num_input_tokens_seen": 149795824,
+      "step": 1016
+    },
+    {
+      "epoch": 0.29287979244739115,
+      "loss": 0.04977504909038544,
+      "loss_ce": 0.002167628612369299,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 149795824,
+      "step": 1016
+    },
+    {
+      "epoch": 0.29316805995964257,
+      "grad_norm": 10.364906246580064,
+      "learning_rate": 0.0001,
+      "loss": 0.0348,
+      "num_input_tokens_seen": 149968296,
+      "step": 1017
+    },
+    {
+      "epoch": 0.29316805995964257,
+      "loss": 0.039077188819646835,
+      "loss_ce": 0.016387369483709335,
+      "loss_xval": 0.022705078125,
+      "num_input_tokens_seen": 149968296,
+      "step": 1017
+    },
+    {
+      "epoch": 0.29345632747189393,
+      "grad_norm": 13.811124874886177,
+      "learning_rate": 0.0001,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 150103088,
+      "step": 1018
+    },
+    {
+      "epoch": 0.29345632747189393,
+      "loss": 0.054243750870227814,
+      "loss_ce": 0.013930030167102814,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 150103088,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2937445949841453,
+      "grad_norm": 8.931755673306386,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "num_input_tokens_seen": 150238168,
+      "step": 1019
+    },
+    {
+      "epoch": 0.2937445949841453,
+      "loss": 0.020072992891073227,
+      "loss_ce": 0.0018311117310076952,
+      "loss_xval": 0.0181884765625,
+      "num_input_tokens_seen": 150238168,
+      "step": 1019
+    },
+    {
+      "epoch": 0.29403286249639665,
+      "grad_norm": 13.364194602767919,
+      "learning_rate": 0.0001,
+      "loss": 0.0416,
+      "num_input_tokens_seen": 150410816,
+      "step": 1020
+    },
+    {
+      "epoch": 0.29403286249639665,
+      "loss": 0.04038884490728378,
+      "loss_ce": 0.009215136058628559,
+      "loss_xval": 0.0311279296875,
+      "num_input_tokens_seen": 150410816,
+      "step": 1020
+    },
+    {
+      "epoch": 0.294321130008648,
+      "grad_norm": 8.488080883654083,
+      "learning_rate": 0.0001,
+      "loss": 0.0252,
+      "num_input_tokens_seen": 150545608,
+      "step": 1021
+    },
+    {
+      "epoch": 0.294321130008648,
+      "loss": 0.026289835572242737,
+      "loss_ce": 0.014265909790992737,
+      "loss_xval": 0.01202392578125,
+      "num_input_tokens_seen": 150545608,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2946093975208994,
+      "grad_norm": 13.343837433926826,
+      "learning_rate": 0.0001,
+      "loss": 0.0378,
+      "num_input_tokens_seen": 150680792,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2946093975208994,
+      "loss": 0.0328974649310112,
+      "loss_ce": 0.0027918717823922634,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 150680792,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2948976650331508,
+      "grad_norm": 7.013524864159005,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "num_input_tokens_seen": 150853376,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2948976650331508,
+      "loss": 0.013356918469071388,
+      "loss_ce": 0.004079573787748814,
+      "loss_xval": 0.00927734375,
+      "num_input_tokens_seen": 150853376,
+      "step": 1023
+    },
+    {
+      "epoch": 0.29518593254540215,
+      "grad_norm": 13.120764246368939,
+      "learning_rate": 0.0001,
+      "loss": 0.0397,
+      "num_input_tokens_seen": 150988160,
+      "step": 1024
+    },
+    {
+      "epoch": 0.29518593254540215,
+      "loss": 0.043137479573488235,
+      "loss_ce": 0.012879302725195885,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 150988160,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2954742000576535,
+      "grad_norm": 5.383806173983863,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "num_input_tokens_seen": 151123136,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2954742000576535,
+      "loss": 0.010275648906826973,
+      "loss_ce": 0.0012500756420195103,
+      "loss_xval": 0.009033203125,
+      "num_input_tokens_seen": 151123136,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2957624675699049,
+      "grad_norm": 13.615844972319005,
+      "learning_rate": 0.0001,
+      "loss": 0.0424,
+      "num_input_tokens_seen": 151295736,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2957624675699049,
+      "loss": 0.035312261432409286,
+      "loss_ce": 0.005817023105919361,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 151295736,
+      "step": 1026
+    },
+    {
+      "epoch": 0.29605073508215624,
+      "grad_norm": 3.120567186484494,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "num_input_tokens_seen": 151430624,
+      "step": 1027
+    },
+    {
+      "epoch": 0.29605073508215624,
+      "loss": 0.019799476489424706,
+      "loss_ce": 0.01299024187028408,
+      "loss_xval": 0.006805419921875,
+      "num_input_tokens_seen": 151430624,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2963390025944076,
+      "grad_norm": 12.13040539309611,
+      "learning_rate": 0.0001,
+      "loss": 0.0291,
+      "num_input_tokens_seen": 151565696,
+      "step": 1028
+    },
+    {
+      "epoch": 0.2963390025944076,
+      "loss": 0.029170066118240356,
+      "loss_ce": 0.001063377596437931,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 151565696,
+      "step": 1028
+    },
+    {
+      "epoch": 0.29662727010665896,
+      "grad_norm": 1.6662810240924648,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "num_input_tokens_seen": 151738232,
+      "step": 1029
+    },
+    {
+      "epoch": 0.29662727010665896,
+      "loss": 0.005300058051943779,
+      "loss_ce": 0.0031552445143461227,
+      "loss_xval": 0.0021514892578125,
+      "num_input_tokens_seen": 151738232,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2969155376189104,
+      "grad_norm": 11.118915433420447,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "num_input_tokens_seen": 151873024,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2969155376189104,
+      "loss": 0.03775542229413986,
+      "loss_ce": 0.012639455497264862,
+      "loss_xval": 0.025146484375,
+      "num_input_tokens_seen": 151873024,
+      "step": 1030
+    },
+    {
+      "epoch": 0.29720380513116174,
+      "grad_norm": 0.7622954876728699,
+      "learning_rate": 0.0001,
+      "loss": 0.0041,
+      "num_input_tokens_seen": 152008160,
+      "step": 1031
+    },
+    {
+      "epoch": 0.29720380513116174,
+      "loss": 0.004262763075530529,
+      "loss_ce": 0.0008714973228052258,
+      "loss_xval": 0.003387451171875,
+      "num_input_tokens_seen": 152008160,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2974920726434131,
+      "grad_norm": 11.233492609765298,
+      "learning_rate": 0.0001,
+      "loss": 0.0283,
+      "num_input_tokens_seen": 152180696,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2974920726434131,
+      "loss": 0.022938229143619537,
+      "loss_ce": 0.0020642043091356754,
+      "loss_xval": 0.0208740234375,
+      "num_input_tokens_seen": 152180696,
+      "step": 1032
+    },
+    {
+      "epoch": 0.29778034015566446,
+      "grad_norm": 1.8711093823146188,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "num_input_tokens_seen": 152315464,
+      "step": 1033
+    },
+    {
+      "epoch": 0.29778034015566446,
+      "loss": 0.011964991688728333,
+      "loss_ce": 0.009271815419197083,
+      "loss_xval": 0.002685546875,
+      "num_input_tokens_seen": 152315464,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2980686076679158,
+      "grad_norm": 9.643338816839094,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "num_input_tokens_seen": 152450504,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2980686076679158,
+      "loss": 0.017920546233654022,
+      "loss_ce": 0.0008764795493334532,
+      "loss_xval": 0.01708984375,
+      "num_input_tokens_seen": 152450504,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2983568751801672,
+      "grad_norm": 3.7999975124459855,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "num_input_tokens_seen": 152622880,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2983568751801672,
+      "loss": 0.0050286888144910336,
+      "loss_ce": 0.0016564966645091772,
+      "loss_xval": 0.0033721923828125,
+      "num_input_tokens_seen": 152622880,
+      "step": 1035
+    },
+    {
+      "epoch": 0.29864514269241854,
+      "grad_norm": 7.570198961949153,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "num_input_tokens_seen": 152757704,
+      "step": 1036
+    },
+    {
+      "epoch": 0.29864514269241854,
+      "loss": 0.019874121993780136,
+      "loss_ce": 0.008231667801737785,
+      "loss_xval": 0.01165771484375,
+      "num_input_tokens_seen": 152757704,
+      "step": 1036
+    },
+    {
+      "epoch": 0.29893341020466996,
+      "grad_norm": 5.035619336789141,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "num_input_tokens_seen": 152892736,
+      "step": 1037
+    },
+    {
+      "epoch": 0.29893341020466996,
+      "loss": 0.006385215558111668,
+      "loss_ce": 0.0007489998824894428,
+      "loss_xval": 0.005645751953125,
+      "num_input_tokens_seen": 152892736,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2992216777169213,
+      "grad_norm": 6.429868675290196,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "num_input_tokens_seen": 153065136,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2992216777169213,
+      "loss": 0.009618373587727547,
+      "loss_ce": 0.0013137776404619217,
+      "loss_xval": 0.00830078125,
+      "num_input_tokens_seen": 153065136,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2995099452291727,
+      "grad_norm": 6.432639398582564,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "num_input_tokens_seen": 153199952,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2995099452291727,
+      "loss": 0.01743425987660885,
+      "loss_ce": 0.009648462757468224,
+      "loss_xval": 0.007781982421875,
+      "num_input_tokens_seen": 153199952,
+      "step": 1039
+    },
+    {
+      "epoch": 0.29979821274142404,
+      "grad_norm": 3.7247511285128807,
+      "learning_rate": 0.0001,
+      "loss": 0.0049,
+      "num_input_tokens_seen": 153335216,
+      "step": 1040
+    },
+    {
+      "epoch": 0.29979821274142404,
+      "loss": 0.003518037497997284,
+      "loss_ce": 0.0005702301859855652,
+      "loss_xval": 0.0029449462890625,
+      "num_input_tokens_seen": 153335216,
+      "step": 1040
+    },
+    {
+      "epoch": 0.3000864802536754,
+      "grad_norm": 7.104556921570073,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "num_input_tokens_seen": 153507568,
+      "step": 1041
+    },
+    {
+      "epoch": 0.3000864802536754,
+      "loss": 0.009859236888587475,
+      "loss_ce": 0.0010854334104806185,
+      "loss_xval": 0.0087890625,
+      "num_input_tokens_seen": 153507568,
+      "step": 1041
+    },
+    {
+      "epoch": 0.30037474776592676,
+      "grad_norm": 1.2721608266605768,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "num_input_tokens_seen": 153642344,
+      "step": 1042
+    },
+    {
+      "epoch": 0.30037474776592676,
+      "loss": 0.010791032575070858,
+      "loss_ce": 0.00926181674003601,
+      "loss_xval": 0.00152587890625,
+      "num_input_tokens_seen": 153642344,
+      "step": 1042
+    },
+    {
+      "epoch": 0.3006630152781781,
+      "grad_norm": 7.144400955958297,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "num_input_tokens_seen": 153777424,
+      "step": 1043
+    },
+    {
+      "epoch": 0.3006630152781781,
+      "loss": 0.00973152182996273,
+      "loss_ce": 0.00044654967496171594,
+      "loss_xval": 0.00927734375,
+      "num_input_tokens_seen": 153777424,
+      "step": 1043
+    },
+    {
+      "epoch": 0.30095128279042954,
+      "grad_norm": 1.5694669073126917,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "num_input_tokens_seen": 153949848,
+      "step": 1044
+    },
+    {
+      "epoch": 0.30095128279042954,
+      "loss": 0.0023211517836898565,
+      "loss_ce": 0.0010503807570785284,
+      "loss_xval": 0.00127410888671875,
+      "num_input_tokens_seen": 153949848,
+      "step": 1044
+    },
+    {
+      "epoch": 0.3012395503026809,
+      "grad_norm": 6.43614642477153,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "num_input_tokens_seen": 154084664,
+      "step": 1045
+    },
+    {
+      "epoch": 0.3012395503026809,
+      "loss": 0.017825186252593994,
+      "loss_ce": 0.01016527321189642,
+      "loss_xval": 0.007659912109375,
+      "num_input_tokens_seen": 154084664,
+      "step": 1045
+    },
+    {
+      "epoch": 0.30152781781493226,
+      "grad_norm": 4.690194926699786,
+      "learning_rate": 0.0001,
+      "loss": 0.0049,
+      "num_input_tokens_seen": 154219688,
+      "step": 1046
+    },
+    {
+      "epoch": 0.30152781781493226,
+      "loss": 0.004704832099378109,
+      "loss_ce": 0.00046861107693985105,
+      "loss_xval": 0.004241943359375,
+      "num_input_tokens_seen": 154219688,
+      "step": 1046
+    },
+    {
+      "epoch": 0.3018160853271836,
+      "grad_norm": 3.929857549293112,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "num_input_tokens_seen": 154392072,
+      "step": 1047
+    },
+    {
+      "epoch": 0.3018160853271836,
+      "loss": 0.0042261965572834015,
+      "loss_ce": 0.0011534581426531076,
+      "loss_xval": 0.0030670166015625,
+      "num_input_tokens_seen": 154392072,
+      "step": 1047
+    },
+    {
+      "epoch": 0.302104352839435,
+      "grad_norm": 6.606956071895859,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "num_input_tokens_seen": 154526880,
+      "step": 1048
+    },
+    {
+      "epoch": 0.302104352839435,
+      "loss": 0.01597825065255165,
+      "loss_ce": 0.00860062800347805,
+      "loss_xval": 0.00738525390625,
+      "num_input_tokens_seen": 154526880,
+      "step": 1048
+    },
+    {
+      "epoch": 0.30239262035168635,
+      "grad_norm": 0.3879556290843241,
+      "learning_rate": 0.0001,
+      "loss": 0.0018,
+      "num_input_tokens_seen": 154661976,
+      "step": 1049
+    },
+    {
+      "epoch": 0.30239262035168635,
+      "loss": 0.000958413234911859,
+      "loss_ce": 0.00039574544643983245,
+      "loss_xval": 0.0005645751953125,
+      "num_input_tokens_seen": 154661976,
+      "step": 1049
+    },
+    {
+      "epoch": 0.3026808878639377,
+      "grad_norm": 5.967744172554036,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "num_input_tokens_seen": 154834640,
+      "step": 1050
+    },
+    {
+      "epoch": 0.3026808878639377,
+      "loss": 0.007467512972652912,
+      "loss_ce": 0.000837569241411984,
+      "loss_xval": 0.006622314453125,
+      "num_input_tokens_seen": 154834640,
+      "step": 1050
+    },
+    {
+      "epoch": 0.3029691553761891,
+      "grad_norm": 3.674411800824582,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "num_input_tokens_seen": 154969320,
+      "step": 1051
+    },
+    {
+      "epoch": 0.3029691553761891,
+      "loss": 0.010981084778904915,
+      "loss_ce": 0.008468152955174446,
+      "loss_xval": 0.0025177001953125,
+      "num_input_tokens_seen": 154969320,
+      "step": 1051
+    },
+    {
+      "epoch": 0.3032574228884405,
+      "grad_norm": 3.6628554067126227,
+      "learning_rate": 0.0001,
+      "loss": 0.0033,
+      "num_input_tokens_seen": 155104376,
+      "step": 1052
+    },
+    {
+      "epoch": 0.3032574228884405,
+      "loss": 0.0028882953338325024,
+      "loss_ce": 0.0003457994607742876,
+      "loss_xval": 0.0025482177734375,
+      "num_input_tokens_seen": 155104376,
+      "step": 1052
+    },
+    {
+      "epoch": 0.30354569040069185,
+      "grad_norm": 6.352602041437757,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "num_input_tokens_seen": 155276944,
+      "step": 1053
+    },
+    {
+      "epoch": 0.30354569040069185,
+      "loss": 0.006636136211454868,
+      "loss_ce": 0.0006050996598787606,
+      "loss_xval": 0.00604248046875,
+      "num_input_tokens_seen": 155276944,
+      "step": 1053
+    },
+    {
+      "epoch": 0.3038339579129432,
+      "grad_norm": 1.16276976605343,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "num_input_tokens_seen": 155411768,
+      "step": 1054
+    },
+    {
+      "epoch": 0.3038339579129432,
+      "loss": 0.013181800954043865,
+      "loss_ce": 0.012291069142520428,
+      "loss_xval": 0.00089263916015625,
+      "num_input_tokens_seen": 155411768,
+      "step": 1054
+    },
+    {
+      "epoch": 0.30412222542519457,
+      "grad_norm": 5.495397592642692,
+      "learning_rate": 0.0001,
+      "loss": 0.0059,
+      "num_input_tokens_seen": 155546872,
+      "step": 1055
+    },
+    {
+      "epoch": 0.30412222542519457,
+      "loss": 0.005437557119876146,
+      "loss_ce": 0.0003220479120500386,
+      "loss_xval": 0.005126953125,
+      "num_input_tokens_seen": 155546872,
+      "step": 1055
+    },
+    {
+      "epoch": 0.30441049293744593,
+      "grad_norm": 5.63390137909816,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "num_input_tokens_seen": 155719568,
+      "step": 1056
+    },
+    {
+      "epoch": 0.30441049293744593,
+      "loss": 0.005506637506186962,
+      "loss_ce": 0.00046742259291931987,
+      "loss_xval": 0.005035400390625,
+      "num_input_tokens_seen": 155719568,
+      "step": 1056
+    },
+    {
+      "epoch": 0.3046987604496973,
+      "grad_norm": 1.0159675221600082,
+      "learning_rate": 0.0001,
+      "loss": 0.0058,
+      "num_input_tokens_seen": 155854352,
+      "step": 1057
+    },
+    {
+      "epoch": 0.3046987604496973,
+      "loss": 0.010783128440380096,
+      "loss_ce": 0.010038308799266815,
+      "loss_xval": 0.000743865966796875,
+      "num_input_tokens_seen": 155854352,
+      "step": 1057
+    },
+    {
+      "epoch": 0.3049870279619487,
+      "grad_norm": 6.196608356578114,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "num_input_tokens_seen": 155989472,
+      "step": 1058
+    },
+    {
+      "epoch": 0.3049870279619487,
+      "loss": 0.006915869191288948,
+      "loss_ce": 0.0006788394530303776,
+      "loss_xval": 0.0062255859375,
+      "num_input_tokens_seen": 155989472,
+      "step": 1058
+    },
+    {
+      "epoch": 0.30527529547420007,
+      "grad_norm": 3.793207536774932,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "num_input_tokens_seen": 156161928,
+      "step": 1059
+    },
+    {
+      "epoch": 0.30527529547420007,
+      "loss": 0.0034458579029887915,
+      "loss_ce": 0.0010359229054301977,
+      "loss_xval": 0.002410888671875,
+      "num_input_tokens_seen": 156161928,
+      "step": 1059
+    },
+    {
+      "epoch": 0.30556356298645143,
+      "grad_norm": 2.9765471186640684,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "num_input_tokens_seen": 156296776,
+      "step": 1060
+    },
+    {
+      "epoch": 0.30556356298645143,
+      "loss": 0.012287916615605354,
+      "loss_ce": 0.010464491322636604,
+      "loss_xval": 0.00182342529296875,
+      "num_input_tokens_seen": 156296776,
+      "step": 1060
+    },
+    {
+      "epoch": 0.3058518304987028,
+      "grad_norm": 6.3222162275498555,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "num_input_tokens_seen": 156431832,
+      "step": 1061
+    },
+    {
+      "epoch": 0.3058518304987028,
+      "loss": 0.00641957763582468,
+      "loss_ce": 0.000300803454592824,
+      "loss_xval": 0.006103515625,
+      "num_input_tokens_seen": 156431832,
+      "step": 1061
+    },
+    {
+      "epoch": 0.30614009801095415,
+      "grad_norm": 2.708296585407212,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "num_input_tokens_seen": 156604280,
+      "step": 1062
+    },
+    {
+      "epoch": 0.30614009801095415,
+      "loss": 0.0020502936094999313,
+      "loss_ce": 0.0005210767267271876,
+      "loss_xval": 0.00152587890625,
+      "num_input_tokens_seen": 156604280,
+      "step": 1062
+    },
+    {
+      "epoch": 0.3064283655232055,
+      "grad_norm": 3.646756298160841,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "num_input_tokens_seen": 156739048,
+      "step": 1063
+    },
+    {
+      "epoch": 0.3064283655232055,
+      "loss": 0.012669045478105545,
+      "loss_ce": 0.010387856513261795,
+      "loss_xval": 0.002288818359375,
+      "num_input_tokens_seen": 156739048,
+      "step": 1063
+    },
+    {
+      "epoch": 0.30671663303545693,
+      "grad_norm": 5.864763562119953,
+      "learning_rate": 0.0001,
+      "loss": 0.0064,
+      "num_input_tokens_seen": 156874232,
+      "step": 1064
+    },
+    {
+      "epoch": 0.30671663303545693,
+      "loss": 0.00592215359210968,
+      "loss_ce": 0.0003298069932498038,
+      "loss_xval": 0.005584716796875,
+      "num_input_tokens_seen": 156874232,
+      "step": 1064
+    },
+    {
+      "epoch": 0.3070049005477083,
+      "grad_norm": 1.621707410019325,
+      "learning_rate": 0.0001,
+      "loss": 0.0065,
+      "num_input_tokens_seen": 157046744,
+      "step": 1065
+    },
+    {
+      "epoch": 0.3070049005477083,
+      "loss": 0.0015599572798237205,
+      "loss_ce": 0.0005266511579975486,
+      "loss_xval": 0.00102996826171875,
+      "num_input_tokens_seen": 157046744,
+      "step": 1065
+    },
+    {
+      "epoch": 0.30729316805995965,
+      "grad_norm": 4.474351846510715,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "num_input_tokens_seen": 157181640,
+      "step": 1066
+    },
+    {
+      "epoch": 0.30729316805995965,
+      "loss": 0.008225830271840096,
+      "loss_ce": 0.005221756175160408,
+      "loss_xval": 0.0030059814453125,
+      "num_input_tokens_seen": 157181640,
+      "step": 1066
+    },
+    {
+      "epoch": 0.307581435572211,
+      "grad_norm": 5.802771539488246,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "num_input_tokens_seen": 157316736,
+      "step": 1067
+    },
+    {
+      "epoch": 0.307581435572211,
+      "loss": 0.00557685736566782,
+      "loss_ce": 0.00023628136841580272,
+      "loss_xval": 0.005340576171875,
+      "num_input_tokens_seen": 157316736,
+      "step": 1067
+    },
+    {
+      "epoch": 0.3078697030844624,
+      "grad_norm": 0.6427301743774646,
+      "learning_rate": 0.0001,
+      "loss": 0.0047,
+      "num_input_tokens_seen": 157489248,
+      "step": 1068
+    },
+    {
+      "epoch": 0.3078697030844624,
+      "loss": 0.0010960090439766645,
+      "loss_ce": 0.0004949558642692864,
+      "loss_xval": 0.00060272216796875,
+      "num_input_tokens_seen": 157489248,
+      "step": 1068
+    },
+    {
+      "epoch": 0.30815797059671374,
+      "grad_norm": 5.649654271338128,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "num_input_tokens_seen": 157624120,
+      "step": 1069
+    },
+    {
+      "epoch": 0.30815797059671374,
+      "loss": 0.0163407139480114,
+      "loss_ce": 0.012455443851649761,
+      "loss_xval": 0.0038909912109375,
+      "num_input_tokens_seen": 157624120,
+      "step": 1069
+    },
+    {
+      "epoch": 0.3084462381089651,
+      "grad_norm": 5.799699774111246,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "num_input_tokens_seen": 157759240,
+      "step": 1070
+    },
+    {
+      "epoch": 0.3084462381089651,
+      "loss": 0.010901054367423058,
+      "loss_ce": 0.005217155907303095,
+      "loss_xval": 0.00567626953125,
+      "num_input_tokens_seen": 157759240,
+      "step": 1070
+    },
+    {
+      "epoch": 0.3087345056212165,
+      "grad_norm": 1.1247419801395206,
+      "learning_rate": 0.0001,
+      "loss": 0.0054,
+      "num_input_tokens_seen": 157931680,
+      "step": 1071
+    },
+    {
+      "epoch": 0.3087345056212165,
+      "loss": 0.0013144547119736671,
+      "loss_ce": 0.0008340412168763578,
+      "loss_xval": 0.00048065185546875,
+      "num_input_tokens_seen": 157931680,
+      "step": 1071
+    },
+    {
+      "epoch": 0.3090227731334679,
+      "grad_norm": 7.909503645706065,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "num_input_tokens_seen": 158066368,
+      "step": 1072
+    },
+    {
+      "epoch": 0.3090227731334679,
+      "loss": 0.01710927113890648,
+      "loss_ce": 0.008610125631093979,
+      "loss_xval": 0.00848388671875,
+      "num_input_tokens_seen": 158066368,
+      "step": 1072
+    },
+    {
+      "epoch": 0.30931104064571924,
+      "grad_norm": 7.299799078860211,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "num_input_tokens_seen": 158201368,
+      "step": 1073
+    },
+    {
+      "epoch": 0.30931104064571924,
+      "loss": 0.012196368537843227,
+      "loss_ce": 0.0033806031569838524,
+      "loss_xval": 0.0087890625,
+      "num_input_tokens_seen": 158201368,
+      "step": 1073
+    },
+    {
+      "epoch": 0.3095993081579706,
+      "grad_norm": 1.2005303324186114,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "num_input_tokens_seen": 158373776,
+      "step": 1074
+    },
+    {
+      "epoch": 0.3095993081579706,
+      "loss": 0.0010331561788916588,
+      "loss_ce": 0.0005875517963431776,
+      "loss_xval": 0.000446319580078125,
+      "num_input_tokens_seen": 158373776,
+      "step": 1074
+    },
+    {
+      "epoch": 0.30988757567022196,
+      "grad_norm": 9.154735585892116,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "num_input_tokens_seen": 158508480,
+      "step": 1075
+    },
+    {
+      "epoch": 0.30988757567022196,
+      "loss": 0.019368529319763184,
+      "loss_ce": 0.008641598746180534,
+      "loss_xval": 0.0107421875,
+      "num_input_tokens_seen": 158508480,
+      "step": 1075
+    },
+    {
+      "epoch": 0.3101758431824733,
+      "grad_norm": 8.412756668445324,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "num_input_tokens_seen": 158643488,
+      "step": 1076
+    },
+    {
+      "epoch": 0.3101758431824733,
+      "loss": 0.012810402549803257,
+      "loss_ce": 0.000229530967772007,
+      "loss_xval": 0.0125732421875,
+      "num_input_tokens_seen": 158643488,
+      "step": 1076
+    },
+    {
+      "epoch": 0.3104641106947247,
+      "grad_norm": 0.9466170936867265,
+      "learning_rate": 0.0001,
+      "loss": 0.0062,
+      "num_input_tokens_seen": 158816016,
+      "step": 1077
+    },
+    {
+      "epoch": 0.3104641106947247,
+      "loss": 0.0014774627052247524,
+      "loss_ce": 0.0010189838940277696,
+      "loss_xval": 0.000457763671875,
+      "num_input_tokens_seen": 158816016,
+      "step": 1077
+    },
+    {
+      "epoch": 0.3107523782069761,
+      "grad_norm": 10.21379456640987,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "num_input_tokens_seen": 158950880,
+      "step": 1078
+    },
+    {
+      "epoch": 0.3107523782069761,
+      "loss": 0.02332136780023575,
+      "loss_ce": 0.009924151003360748,
+      "loss_xval": 0.013427734375,
+      "num_input_tokens_seen": 158950880,
+      "step": 1078
+    },
+    {
+      "epoch": 0.31104064571922746,
+      "grad_norm": 9.987037711665085,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "num_input_tokens_seen": 159085888,
+      "step": 1079
+    },
+    {
+      "epoch": 0.31104064571922746,
+      "loss": 0.018559180200099945,
+      "loss_ce": 0.00021811538317706436,
+      "loss_xval": 0.018310546875,
+      "num_input_tokens_seen": 159085888,
+      "step": 1079
+    },
+    {
+      "epoch": 0.3113289132314788,
+      "grad_norm": 0.20392553303226543,
+      "learning_rate": 0.0001,
+      "loss": 0.0068,
+      "num_input_tokens_seen": 159258384,
+      "step": 1080
+    },
+    {
+      "epoch": 0.3113289132314788,
+      "loss": 0.0019423088524490595,
+      "loss_ce": 0.0007568916189484298,
+      "loss_xval": 0.00118255615234375,
+      "num_input_tokens_seen": 159258384,
+      "step": 1080
+    },
+    {
+      "epoch": 0.3116171807437302,
+      "grad_norm": 9.791986473139499,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "num_input_tokens_seen": 159393072,
+      "step": 1081
+    },
+    {
+      "epoch": 0.3116171807437302,
+      "loss": 0.021431514993309975,
+      "loss_ce": 0.009758541360497475,
+      "loss_xval": 0.01165771484375,
+      "num_input_tokens_seen": 159393072,
+      "step": 1081
+    },
+    {
+      "epoch": 0.31190544825598154,
+      "grad_norm": 9.607565793359004,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "num_input_tokens_seen": 159528224,
+      "step": 1082
+    },
+    {
+      "epoch": 0.31190544825598154,
+      "loss": 0.01773345097899437,
+      "loss_ce": 0.000231619254918769,
+      "loss_xval": 0.0174560546875,
+      "num_input_tokens_seen": 159528224,
+      "step": 1082
+    },
+    {
+      "epoch": 0.3121937157682329,
+      "grad_norm": 0.8754551559488629,
+      "learning_rate": 0.0001,
+      "loss": 0.0064,
+      "num_input_tokens_seen": 159700648,
+      "step": 1083
+    },
+    {
+      "epoch": 0.3121937157682329,
+      "loss": 0.002320789499208331,
+      "loss_ce": 0.0010037652682512999,
+      "loss_xval": 0.00131988525390625,
+      "num_input_tokens_seen": 159700648,
+      "step": 1083
+    },
+    {
+      "epoch": 0.31248198328048427,
+      "grad_norm": 10.787525641949458,
+      "learning_rate": 0.0001,
+      "loss": 0.0255,
+      "num_input_tokens_seen": 159835472,
+      "step": 1084
+    },
+    {
+      "epoch": 0.31248198328048427,
+      "loss": 0.022629978135228157,
+      "loss_ce": 0.008164646103978157,
+      "loss_xval": 0.01446533203125,
+      "num_input_tokens_seen": 159835472,
+      "step": 1084
+    },
+    {
+      "epoch": 0.3127702507927357,
+      "grad_norm": 10.646109982293996,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "num_input_tokens_seen": 159970448,
+      "step": 1085
+    },
+    {
+      "epoch": 0.3127702507927357,
+      "loss": 0.02264305204153061,
+      "loss_ce": 0.00024314915935974568,
+      "loss_xval": 0.0224609375,
+      "num_input_tokens_seen": 159970448,
+      "step": 1085
+    },
+    {
+      "epoch": 0.31305851830498704,
+      "grad_norm": 0.27325773479874504,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "num_input_tokens_seen": 160142976,
+      "step": 1086
+    },
+    {
+      "epoch": 0.31305851830498704,
+      "loss": 0.0023754406720399857,
+      "loss_ce": 0.0006254484760574996,
+      "loss_xval": 0.00174713134765625,
+      "num_input_tokens_seen": 160142976,
+      "step": 1086
+    },
+    {
+      "epoch": 0.3133467858172384,
+      "grad_norm": 11.219846459935127,
+      "learning_rate": 0.0001,
+      "loss": 0.0282,
+      "num_input_tokens_seen": 160277728,
+      "step": 1087
+    },
+    {
+      "epoch": 0.3133467858172384,
+      "loss": 0.025487396866083145,
+      "loss_ce": 0.011311981827020645,
+      "loss_xval": 0.01416015625,
+      "num_input_tokens_seen": 160277728,
+      "step": 1087
+    },
+    {
+      "epoch": 0.31363505332948977,
+      "grad_norm": 11.484214546937917,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "num_input_tokens_seen": 160412976,
+      "step": 1088
+    },
+    {
+      "epoch": 0.31363505332948977,
+      "loss": 0.027778614312410355,
+      "loss_ce": 0.0002670185058377683,
+      "loss_xval": 0.0274658203125,
+      "num_input_tokens_seen": 160412976,
+      "step": 1088
+    },
+    {
+      "epoch": 0.3139233208417411,
+      "grad_norm": 0.5123509703035186,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "num_input_tokens_seen": 160585480,
+      "step": 1089
+    },
+    {
+      "epoch": 0.3139233208417411,
+      "loss": 0.0037535782903432846,
+      "loss_ce": 0.0007361529860645533,
+      "loss_xval": 0.003021240234375,
+      "num_input_tokens_seen": 160585480,
+      "step": 1089
+    },
+    {
+      "epoch": 0.3142115883539925,
+      "grad_norm": 10.85298039866953,
+      "learning_rate": 0.0001,
+      "loss": 0.0258,
+      "num_input_tokens_seen": 160720264,
+      "step": 1090
+    },
+    {
+      "epoch": 0.3142115883539925,
+      "loss": 0.020125051960349083,
+      "loss_ce": 0.00652947137132287,
+      "loss_xval": 0.01361083984375,
+      "num_input_tokens_seen": 160720264,
+      "step": 1090
+    },
+    {
+      "epoch": 0.31449985586624385,
+      "grad_norm": 10.865952405702181,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "num_input_tokens_seen": 160855376,
+      "step": 1091
+    },
+    {
+      "epoch": 0.31449985586624385,
+      "loss": 0.025808095932006836,
+      "loss_ce": 0.00029540024115704,
+      "loss_xval": 0.0255126953125,
+      "num_input_tokens_seen": 160855376,
+      "step": 1091
+    },
+    {
+      "epoch": 0.31478812337849527,
+      "grad_norm": 0.6638405627439249,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "num_input_tokens_seen": 161027808,
+      "step": 1092
+    },
+    {
+      "epoch": 0.31478812337849527,
+      "loss": 0.0030655530281364918,
+      "loss_ce": 0.0007996229687705636,
+      "loss_xval": 0.00225830078125,
+      "num_input_tokens_seen": 161027808,
+      "step": 1092
+    },
+    {
+      "epoch": 0.3150763908907466,
+      "grad_norm": 11.225511097099034,
+      "learning_rate": 0.0001,
+      "loss": 0.0288,
+      "num_input_tokens_seen": 161162600,
+      "step": 1093
+    },
+    {
+      "epoch": 0.3150763908907466,
+      "loss": 0.022820211946964264,
+      "loss_ce": 0.009468772448599339,
+      "loss_xval": 0.01336669921875,
+      "num_input_tokens_seen": 161162600,
+      "step": 1093
+    },
+    {
+      "epoch": 0.315364658402998,
+      "grad_norm": 9.31150135276372,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "num_input_tokens_seen": 161297704,
+      "step": 1094
+    },
+    {
+      "epoch": 0.315364658402998,
+      "loss": 0.021114954724907875,
+      "loss_ce": 0.0003630008432082832,
+      "loss_xval": 0.020751953125,
+      "num_input_tokens_seen": 161297704,
+      "step": 1094
+    },
+    {
+      "epoch": 0.31565292591524935,
+      "grad_norm": 3.7042286765060615,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "num_input_tokens_seen": 161470192,
+      "step": 1095
+    },
+    {
+      "epoch": 0.31565292591524935,
+      "loss": 0.001126689836382866,
+      "loss_ce": 0.00049559585750103,
+      "loss_xval": 0.000629425048828125,
+      "num_input_tokens_seen": 161470192,
+      "step": 1095
+    },
+    {
+      "epoch": 0.3159411934275007,
+      "grad_norm": 14.440635946622175,
+      "learning_rate": 0.0001,
+      "loss": 0.0437,
+      "num_input_tokens_seen": 161604920,
+      "step": 1096
+    },
+    {
+      "epoch": 0.3159411934275007,
+      "loss": 0.03371185064315796,
+      "loss_ce": 0.009496153332293034,
+      "loss_xval": 0.024169921875,
+      "num_input_tokens_seen": 161604920,
+      "step": 1096
+    },
+    {
+      "epoch": 0.31622946093975207,
+      "grad_norm": 10.8120636880313,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "num_input_tokens_seen": 161739992,
+      "step": 1097
+    },
+    {
+      "epoch": 0.31622946093975207,
+      "loss": 0.028470251709222794,
+      "loss_ce": 0.00036356275086291134,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 161739992,
+      "step": 1097
+    },
+    {
+      "epoch": 0.31651772845200343,
+      "grad_norm": 4.845232306751907,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "num_input_tokens_seen": 161912496,
+      "step": 1098
+    },
+    {
+      "epoch": 0.31651772845200343,
+      "loss": 0.0014338723849505186,
+      "loss_ce": 0.0007434121216647327,
+      "loss_xval": 0.000690460205078125,
+      "num_input_tokens_seen": 161912496,
+      "step": 1098
+    },
+    {
+      "epoch": 0.31680599596425485,
+      "grad_norm": 16.83238346653988,
+      "learning_rate": 0.0001,
+      "loss": 0.0573,
+      "num_input_tokens_seen": 162047344,
+      "step": 1099
+    },
+    {
+      "epoch": 0.31680599596425485,
+      "loss": 0.042024776339530945,
+      "loss_ce": 0.007814568467438221,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 162047344,
+      "step": 1099
+    },
+    {
+      "epoch": 0.3170942634765062,
+      "grad_norm": 11.875846180074877,
+      "learning_rate": 0.0001,
+      "loss": 0.0276,
+      "num_input_tokens_seen": 162182456,
+      "step": 1100
+    },
+    {
+      "epoch": 0.3170942634765062,
+      "loss": 0.03627585619688034,
+      "loss_ce": 0.0006008032942190766,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 162182456,
+      "step": 1100
+    },
+    {
+      "epoch": 0.31738253098875757,
+      "grad_norm": 6.8405890813471935,
+      "learning_rate": 0.0001,
+      "loss": 0.0213,
+      "num_input_tokens_seen": 162354808,
+      "step": 1101
+    },
+    {
+      "epoch": 0.31738253098875757,
+      "loss": 0.005464246030896902,
+      "loss_ce": 0.004270245786756277,
+      "loss_xval": 0.001190185546875,
+      "num_input_tokens_seen": 162354808,
+      "step": 1101
+    },
+    {
+      "epoch": 0.31767079850100893,
+      "grad_norm": 20.75383958252492,
+      "learning_rate": 0.0001,
+      "loss": 0.0855,
+      "num_input_tokens_seen": 162489592,
+      "step": 1102
+    },
+    {
+      "epoch": 0.31767079850100893,
+      "loss": 0.06375418603420258,
+      "loss_ce": 0.009616006165742874,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 162489592,
+      "step": 1102
+    },
+    {
+      "epoch": 0.3179590660132603,
+      "grad_norm": 14.355246236937758,
+      "learning_rate": 0.0001,
+      "loss": 0.0399,
+      "num_input_tokens_seen": 162624624,
+      "step": 1103
+    },
+    {
+      "epoch": 0.3179590660132603,
+      "loss": 0.05262349545955658,
+      "loss_ce": 0.0004384391359053552,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 162624624,
+      "step": 1103
+    },
+    {
+      "epoch": 0.31824733352551166,
+      "grad_norm": 7.739226969232351,
+      "learning_rate": 0.0001,
+      "loss": 0.0272,
+      "num_input_tokens_seen": 162797120,
+      "step": 1104
+    },
+    {
+      "epoch": 0.31824733352551166,
+      "loss": 0.00172033766284585,
+      "loss_ce": 0.0007585571147501469,
+      "loss_xval": 0.0009613037109375,
+      "num_input_tokens_seen": 162797120,
+      "step": 1104
+    },
+    {
+      "epoch": 0.318535601037763,
+      "grad_norm": 22.913991416793884,
+      "learning_rate": 0.0001,
+      "loss": 0.1058,
+      "num_input_tokens_seen": 162931936,
+      "step": 1105
+    },
+    {
+      "epoch": 0.318535601037763,
+      "loss": 0.08138446509838104,
+      "loss_ce": 0.010888861492276192,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 162931936,
+      "step": 1105
+    },
+    {
+      "epoch": 0.31882386855001443,
+      "grad_norm": 13.233226662544276,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "num_input_tokens_seen": 163067024,
+      "step": 1106
+    },
+    {
+      "epoch": 0.31882386855001443,
+      "loss": 0.05254264920949936,
+      "loss_ce": 0.000540698878467083,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 163067024,
+      "step": 1106
+    },
+    {
+      "epoch": 0.3191121360622658,
+      "grad_norm": 13.283548229086884,
+      "learning_rate": 0.0001,
+      "loss": 0.0508,
+      "num_input_tokens_seen": 163239560,
+      "step": 1107
+    },
+    {
+      "epoch": 0.3191121360622658,
+      "loss": 0.008430368266999722,
+      "loss_ce": 0.0013655491638928652,
+      "loss_xval": 0.007080078125,
+      "num_input_tokens_seen": 163239560,
+      "step": 1107
+    },
+    {
+      "epoch": 0.31940040357451716,
+      "grad_norm": 26.958071158147696,
+      "learning_rate": 0.0001,
+      "loss": 0.1463,
+      "num_input_tokens_seen": 163374336,
+      "step": 1108
+    },
+    {
+      "epoch": 0.31940040357451716,
+      "loss": 0.11349773406982422,
+      "loss_ce": 0.008700372651219368,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 163374336,
+      "step": 1108
+    },
+    {
+      "epoch": 0.3196886710867685,
+      "grad_norm": 8.341302719853255,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "num_input_tokens_seen": 163509360,
+      "step": 1109
+    },
+    {
+      "epoch": 0.3196886710867685,
+      "loss": 0.032069962471723557,
+      "loss_ce": 0.0006063411710783839,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 163509360,
+      "step": 1109
+    },
+    {
+      "epoch": 0.3199769385990199,
+      "grad_norm": 25.057511538159552,
+      "learning_rate": 0.0001,
+      "loss": 0.1433,
+      "num_input_tokens_seen": 163681776,
+      "step": 1110
+    },
+    {
+      "epoch": 0.3199769385990199,
+      "loss": 0.05327451229095459,
+      "loss_ce": 0.0007842776831239462,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 163681776,
+      "step": 1110
+    },
+    {
+      "epoch": 0.32026520611127124,
+      "grad_norm": 31.97887247106539,
+      "learning_rate": 0.0001,
+      "loss": 0.2089,
+      "num_input_tokens_seen": 163816592,
+      "step": 1111
+    },
+    {
+      "epoch": 0.32026520611127124,
+      "loss": 0.1736195832490921,
+      "loss_ce": 0.014439894817769527,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 163816592,
+      "step": 1111
+    },
+    {
+      "epoch": 0.32055347362352266,
+      "grad_norm": 3.7657790177530446,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "num_input_tokens_seen": 163951688,
+      "step": 1112
+    },
+    {
+      "epoch": 0.32055347362352266,
+      "loss": 0.0032639564014971256,
+      "loss_ce": 0.0006890358054079115,
+      "loss_xval": 0.0025787353515625,
+      "num_input_tokens_seen": 163951688,
+      "step": 1112
+    },
+    {
+      "epoch": 0.320841741135774,
+      "grad_norm": 41.76094920123107,
+      "learning_rate": 0.0001,
+      "loss": 0.3675,
+      "num_input_tokens_seen": 164124288,
+      "step": 1113
+    },
+    {
+      "epoch": 0.320841741135774,
+      "loss": 0.22158637642860413,
+      "loss_ce": 0.0008832494495436549,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 164124288,
+      "step": 1113
+    },
+    {
+      "epoch": 0.3211300086480254,
+      "grad_norm": 23.535877158829695,
+      "learning_rate": 0.0001,
+      "loss": 0.1256,
+      "num_input_tokens_seen": 164259104,
+      "step": 1114
+    },
+    {
+      "epoch": 0.3211300086480254,
+      "loss": 0.11367707699537277,
+      "loss_ce": 0.008574537932872772,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 164259104,
+      "step": 1114
+    },
+    {
+      "epoch": 0.32141827616027674,
+      "grad_norm": 36.60361659058581,
+      "learning_rate": 0.0001,
+      "loss": 0.281,
+      "num_input_tokens_seen": 164394192,
+      "step": 1115
+    },
+    {
+      "epoch": 0.32141827616027674,
+      "loss": 0.1536085605621338,
+      "loss_ce": 0.0010206589940935373,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 164394192,
+      "step": 1115
+    },
+    {
+      "epoch": 0.3217065436725281,
+      "grad_norm": 49.90161265894237,
+      "learning_rate": 0.0001,
+      "loss": 0.5425,
+      "num_input_tokens_seen": 164566656,
+      "step": 1116
+    },
+    {
+      "epoch": 0.3217065436725281,
+      "loss": 0.412850946187973,
+      "loss_ce": 0.003915389999747276,
+      "loss_xval": 0.408203125,
+      "num_input_tokens_seen": 164566656,
+      "step": 1116
+    },
+    {
+      "epoch": 0.32199481118477946,
+      "grad_norm": 9.574391739728737,
+      "learning_rate": 0.0001,
+      "loss": 0.046,
+      "num_input_tokens_seen": 164701456,
+      "step": 1117
+    },
+    {
+      "epoch": 0.32199481118477946,
+      "loss": 0.05413966625928879,
+      "loss_ce": 0.012193256989121437,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 164701456,
+      "step": 1117
+    },
+    {
+      "epoch": 0.3222830786970308,
+      "grad_norm": 54.59205629061163,
+      "learning_rate": 0.0001,
+      "loss": 0.6666,
+      "num_input_tokens_seen": 164836512,
+      "step": 1118
+    },
+    {
+      "epoch": 0.3222830786970308,
+      "loss": 0.6489585638046265,
+      "loss_ce": 0.0024742158129811287,
+      "loss_xval": 0.6484375,
+      "num_input_tokens_seen": 164836512,
+      "step": 1118
+    },
+    {
+      "epoch": 0.32257134620928224,
+      "grad_norm": 11.693643308618372,
+      "learning_rate": 0.0001,
+      "loss": 0.046,
+      "num_input_tokens_seen": 165008952,
+      "step": 1119
+    },
+    {
+      "epoch": 0.32257134620928224,
+      "loss": 0.01354275457561016,
+      "loss_ce": 0.00258694333024323,
+      "loss_xval": 0.010986328125,
+      "num_input_tokens_seen": 165008952,
+      "step": 1119
+    },
+    {
+      "epoch": 0.3228596137215336,
+      "grad_norm": 31.216037525281035,
+      "learning_rate": 0.0001,
+      "loss": 0.204,
+      "num_input_tokens_seen": 165143728,
+      "step": 1120
+    },
+    {
+      "epoch": 0.3228596137215336,
+      "loss": 0.19186297059059143,
+      "loss_ce": 0.009123703464865685,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 165143728,
+      "step": 1120
+    },
+    {
+      "epoch": 0.32314788123378496,
+      "grad_norm": 146.70057585480131,
+      "learning_rate": 0.0001,
+      "loss": 2.2466,
+      "num_input_tokens_seen": 165278776,
+      "step": 1121
+    },
+    {
+      "epoch": 0.32314788123378496,
+      "loss": 2.0149452686309814,
+      "loss_ce": 0.0032265656627714634,
+      "loss_xval": 2.015625,
+      "num_input_tokens_seen": 165278776,
+      "step": 1121
+    },
+    {
+      "epoch": 0.3234361487460363,
+      "grad_norm": 128.1209425151165,
+      "learning_rate": 0.0001,
+      "loss": 3.5429,
+      "num_input_tokens_seen": 165451272,
+      "step": 1122
+    },
+    {
+      "epoch": 0.3234361487460363,
+      "loss": 3.4982962608337402,
+      "loss_ce": 0.006108833011239767,
+      "loss_xval": 3.5,
+      "num_input_tokens_seen": 165451272,
+      "step": 1122
+    },
+    {
+      "epoch": 0.3237244162582877,
+      "grad_norm": 23.896542536561327,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "num_input_tokens_seen": 165586072,
+      "step": 1123
+    },
+    {
+      "epoch": 0.3237244162582877,
+      "loss": 0.17959138751029968,
+      "loss_ce": 0.03127596154808998,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 165586072,
+      "step": 1123
+    },
+    {
+      "epoch": 0.32401268377053905,
+      "grad_norm": 225.41097447165146,
+      "learning_rate": 0.0001,
+      "loss": 7.3387,
+      "num_input_tokens_seen": 165721048,
+      "step": 1124
+    },
+    {
+      "epoch": 0.32401268377053905,
+      "loss": 7.041110038757324,
+      "loss_ce": 0.08407886326313019,
+      "loss_xval": 6.96875,
+      "num_input_tokens_seen": 165721048,
+      "step": 1124
+    },
+    {
+      "epoch": 0.3243009512827904,
+      "grad_norm": 27.324685193473073,
+      "learning_rate": 0.0001,
+      "loss": 0.3048,
+      "num_input_tokens_seen": 165893552,
+      "step": 1125
+    },
+    {
+      "epoch": 0.3243009512827904,
+      "loss": 0.26803261041641235,
+      "loss_ce": 0.10726602375507355,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 165893552,
+      "step": 1125
+    },
+    {
+      "epoch": 0.3245892187950418,
+      "grad_norm": 83.20874980582771,
+      "learning_rate": 0.0001,
+      "loss": 1.505,
+      "num_input_tokens_seen": 166028344,
+      "step": 1126
+    },
+    {
+      "epoch": 0.3245892187950418,
+      "loss": 1.5246477127075195,
+      "loss_ce": 0.06371021270751953,
+      "loss_xval": 1.4609375,
+      "num_input_tokens_seen": 166028344,
+      "step": 1126
+    },
+    {
+      "epoch": 0.3248774863072932,
+      "grad_norm": 117.91214459350986,
+      "learning_rate": 0.0001,
+      "loss": 1.802,
+      "num_input_tokens_seen": 166163440,
+      "step": 1127
+    },
+    {
+      "epoch": 0.3248774863072932,
+      "loss": 1.7308454513549805,
+      "loss_ce": 0.021861031651496887,
+      "loss_xval": 1.7109375,
+      "num_input_tokens_seen": 166163440,
+      "step": 1127
+    },
+    {
+      "epoch": 0.32516575381954455,
+      "grad_norm": 21.969901497136338,
+      "learning_rate": 0.0001,
+      "loss": 0.1225,
+      "num_input_tokens_seen": 166335888,
+      "step": 1128
+    },
+    {
+      "epoch": 0.32516575381954455,
+      "loss": 0.14037910103797913,
+      "loss_ce": 0.017393263056874275,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 166335888,
+      "step": 1128
+    },
+    {
+      "epoch": 0.3254540213317959,
+      "grad_norm": 70.85543800871753,
+      "learning_rate": 0.0001,
+      "loss": 0.8779,
+      "num_input_tokens_seen": 166470640,
+      "step": 1129
+    },
+    {
+      "epoch": 0.3254540213317959,
+      "loss": 0.817480206489563,
+      "loss_ce": 0.014257535338401794,
+      "loss_xval": 0.8046875,
+      "num_input_tokens_seen": 166470640,
+      "step": 1129
+    },
+    {
+      "epoch": 0.32574228884404727,
+      "grad_norm": 10.650929566560954,
+      "learning_rate": 0.0001,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 166605624,
+      "step": 1130
+    },
+    {
+      "epoch": 0.32574228884404727,
+      "loss": 0.036035455763339996,
+      "loss_ce": 0.0022219824604690075,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 166605624,
+      "step": 1130
+    },
+    {
+      "epoch": 0.32603055635629863,
+      "grad_norm": 88.29133006200985,
+      "learning_rate": 0.0001,
+      "loss": 1.0776,
+      "num_input_tokens_seen": 166778096,
+      "step": 1131
+    },
+    {
+      "epoch": 0.32603055635629863,
+      "loss": 0.9238502979278564,
+      "loss_ce": 0.021994847804307938,
+      "loss_xval": 0.90234375,
+      "num_input_tokens_seen": 166778096,
+      "step": 1131
+    },
+    {
+      "epoch": 0.32631882386855,
+      "grad_norm": 36956.58996931595,
+      "learning_rate": 0.0001,
+      "loss": 148.2867,
+      "num_input_tokens_seen": 166912928,
+      "step": 1132
+    },
+    {
+      "epoch": 0.32631882386855,
+      "loss": 155.36557006835938,
+      "loss_ce": 20.115570068359375,
+      "loss_xval": 135.0,
+      "num_input_tokens_seen": 166912928,
+      "step": 1132
+    },
+    {
+      "epoch": 0.3266070913808014,
+      "grad_norm": 37411.494503156595,
+      "learning_rate": 0.0001,
+      "loss": 961.649,
+      "num_input_tokens_seen": 167047896,
+      "step": 1133
+    },
+    {
+      "epoch": 0.3266070913808014,
+      "loss": 1922.889404296875,
+      "loss_ce": 9.014402389526367,
+      "loss_xval": 1912.0,
+      "num_input_tokens_seen": 167047896,
+      "step": 1133
+    },
+    {
+      "epoch": 0.32689535889305277,
+      "grad_norm": 35611.01625548505,
+      "learning_rate": 0.0001,
+      "loss": 1535.11,
+      "num_input_tokens_seen": 167220624,
+      "step": 1134
+    },
+    {
+      "epoch": 0.32689535889305277,
+      "loss": 0.5700771808624268,
+      "loss_ce": 0.013924892991781235,
+      "loss_xval": 0.5546875,
+      "num_input_tokens_seen": 167220624,
+      "step": 1134
+    },
+    {
+      "epoch": 0.32718362640530413,
+      "grad_norm": 3903.1403432917123,
+      "learning_rate": 0.0001,
+      "loss": 64.9901,
+      "num_input_tokens_seen": 167355400,
+      "step": 1135
+    },
+    {
+      "epoch": 0.32718362640530413,
+      "loss": 102.77651977539062,
+      "loss_ce": 5.159328937530518,
+      "loss_xval": 97.5,
+      "num_input_tokens_seen": 167355400,
+      "step": 1135
+    },
+    {
+      "epoch": 0.3274718939175555,
+      "grad_norm": 14367.867932467441,
+      "learning_rate": 0.0001,
+      "loss": 52.0497,
+      "num_input_tokens_seen": 167490400,
+      "step": 1136
+    },
+    {
+      "epoch": 0.3274718939175555,
+      "loss": 103.61935424804688,
+      "loss_ce": 1.8678855895996094,
+      "loss_xval": 102.0,
+      "num_input_tokens_seen": 167490400,
+      "step": 1136
+    },
+    {
+      "epoch": 0.32776016142980685,
+      "grad_norm": 128.34438076454498,
+      "learning_rate": 0.0001,
+      "loss": 0.487,
+      "num_input_tokens_seen": 167662864,
+      "step": 1137
+    },
+    {
+      "epoch": 0.32776016142980685,
+      "loss": 0.5655533075332642,
+      "loss_ce": 0.23571929335594177,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 167662864,
+      "step": 1137
+    },
+    {
+      "epoch": 0.3280484289420582,
+      "grad_norm": 186.8314494397255,
+      "learning_rate": 0.0001,
+      "loss": 1.0293,
+      "num_input_tokens_seen": 167797576,
+      "step": 1138
+    },
+    {
+      "epoch": 0.3280484289420582,
+      "loss": 1.259447693824768,
+      "loss_ce": 0.9324213266372681,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 167797576,
+      "step": 1138
+    },
+    {
+      "epoch": 0.3283366964543096,
+      "grad_norm": 124.40677251025926,
+      "learning_rate": 0.0001,
+      "loss": 2.4235,
+      "num_input_tokens_seen": 167932648,
+      "step": 1139
+    },
+    {
+      "epoch": 0.3283366964543096,
+      "loss": 1.974762201309204,
+      "loss_ce": 1.721283197402954,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 167932648,
+      "step": 1139
+    },
+    {
+      "epoch": 0.328624963966561,
+      "grad_norm": 111.30133258078932,
+      "learning_rate": 0.0001,
+      "loss": 3.6092,
+      "num_input_tokens_seen": 168105168,
+      "step": 1140
+    },
+    {
+      "epoch": 0.328624963966561,
+      "loss": 3.867354393005371,
+      "loss_ce": 3.805678367614746,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 168105168,
+      "step": 1140
+    },
+    {
+      "epoch": 0.32891323147881235,
+      "grad_norm": 121.31372025179807,
+      "learning_rate": 0.0001,
+      "loss": 3.274,
+      "num_input_tokens_seen": 168239888,
+      "step": 1141
+    },
+    {
+      "epoch": 0.32891323147881235,
+      "loss": 3.6110503673553467,
+      "loss_ce": 3.0675933361053467,
+      "loss_xval": 0.54296875,
+      "num_input_tokens_seen": 168239888,
+      "step": 1141
+    },
+    {
+      "epoch": 0.3292014989910637,
+      "grad_norm": 397.8345944363598,
+      "learning_rate": 0.0001,
+      "loss": 5.6374,
+      "num_input_tokens_seen": 168374952,
+      "step": 1142
+    },
+    {
+      "epoch": 0.3292014989910637,
+      "loss": 4.382977485656738,
+      "loss_ce": 2.9259462356567383,
+      "loss_xval": 1.453125,
+      "num_input_tokens_seen": 168374952,
+      "step": 1142
+    },
+    {
+      "epoch": 0.3294897665033151,
+      "grad_norm": 12260.97294175282,
+      "learning_rate": 0.0001,
+      "loss": 120.2437,
+      "num_input_tokens_seen": 168547464,
+      "step": 1143
+    },
+    {
+      "epoch": 0.3294897665033151,
+      "loss": 110.66117858886719,
+      "loss_ce": 4.442431926727295,
+      "loss_xval": 106.0,
+      "num_input_tokens_seen": 168547464,
+      "step": 1143
+    },
+    {
+      "epoch": 0.32977803401556643,
+      "grad_norm": 16729.315621184192,
+      "learning_rate": 0.0001,
+      "loss": 437.4442,
+      "num_input_tokens_seen": 168682168,
+      "step": 1144
+    },
+    {
+      "epoch": 0.32977803401556643,
+      "loss": 455.2748107910156,
+      "loss_ce": 8.024798393249512,
+      "loss_xval": 448.0,
+      "num_input_tokens_seen": 168682168,
+      "step": 1144
+    },
+    {
+      "epoch": 0.3300663015278178,
+      "grad_norm": 38370.55638467301,
+      "learning_rate": 0.0001,
+      "loss": 1914.62,
+      "num_input_tokens_seen": 168817304,
+      "step": 1145
+    },
+    {
+      "epoch": 0.3300663015278178,
+      "loss": 1790.501220703125,
+      "loss_ce": 7.501255035400391,
+      "loss_xval": 1784.0,
+      "num_input_tokens_seen": 168817304,
+      "step": 1145
+    },
+    {
+      "epoch": 0.33035456904006916,
+      "grad_norm": 575.5757979659641,
+      "learning_rate": 0.0001,
+      "loss": 19.334,
+      "num_input_tokens_seen": 168989928,
+      "step": 1146
+    },
+    {
+      "epoch": 0.33035456904006916,
+      "loss": 19.11969757080078,
+      "loss_ce": 5.244698524475098,
+      "loss_xval": 13.875,
+      "num_input_tokens_seen": 168989928,
+      "step": 1146
+    },
+    {
+      "epoch": 0.3306428365523206,
+      "grad_norm": 363.34565051839263,
+      "learning_rate": 0.0001,
+      "loss": 17.0378,
+      "num_input_tokens_seen": 169124768,
+      "step": 1147
+    },
+    {
+      "epoch": 0.3306428365523206,
+      "loss": 17.8129825592041,
+      "loss_ce": 5.852045059204102,
+      "loss_xval": 11.9375,
+      "num_input_tokens_seen": 169124768,
+      "step": 1147
+    },
+    {
+      "epoch": 0.33093110406457193,
+      "grad_norm": 260.9191068247743,
+      "learning_rate": 0.0001,
+      "loss": 13.149,
+      "num_input_tokens_seen": 169259872,
+      "step": 1148
+    },
+    {
+      "epoch": 0.33093110406457193,
+      "loss": 12.404302597045898,
+      "loss_ce": 6.541021347045898,
+      "loss_xval": 5.875,
+      "num_input_tokens_seen": 169259872,
+      "step": 1148
+    },
+    {
+      "epoch": 0.3312193715768233,
+      "grad_norm": 209.5617449763566,
+      "learning_rate": 0.0001,
+      "loss": 10.899,
+      "num_input_tokens_seen": 169432424,
+      "step": 1149
+    },
+    {
+      "epoch": 0.3312193715768233,
+      "loss": 11.033249855041504,
+      "loss_ce": 6.748093605041504,
+      "loss_xval": 4.28125,
+      "num_input_tokens_seen": 169432424,
+      "step": 1149
+    },
+    {
+      "epoch": 0.33150763908907466,
+      "grad_norm": 123.85290589512873,
+      "learning_rate": 0.0001,
+      "loss": 8.0386,
+      "num_input_tokens_seen": 169567184,
+      "step": 1150
+    },
+    {
+      "epoch": 0.33150763908907466,
+      "loss": 8.336545944213867,
+      "loss_ce": 6.424436569213867,
+      "loss_xval": 1.9140625,
+      "num_input_tokens_seen": 169567184,
+      "step": 1150
+    },
+    {
+      "epoch": 0.331795906601326,
+      "grad_norm": 57.909997902104216,
+      "learning_rate": 0.0001,
+      "loss": 6.8854,
+      "num_input_tokens_seen": 169700752,
+      "step": 1151
+    },
+    {
+      "epoch": 0.331795906601326,
+      "loss": 6.737767219543457,
+      "loss_ce": 6.636631965637207,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 169700752,
+      "step": 1151
+    },
+    {
+      "epoch": 0.3320841741135774,
+      "grad_norm": 58.06386206153396,
+      "learning_rate": 0.0001,
+      "loss": 6.633,
+      "num_input_tokens_seen": 169873256,
+      "step": 1152
+    },
+    {
+      "epoch": 0.3320841741135774,
+      "loss": 6.838903427124023,
+      "loss_ce": 6.601720809936523,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 169873256,
+      "step": 1152
+    },
+    {
+      "epoch": 0.3323724416258288,
+      "grad_norm": 101.93742036056541,
+      "learning_rate": 0.0001,
+      "loss": 7.0618,
+      "num_input_tokens_seen": 170008032,
+      "step": 1153
+    },
+    {
+      "epoch": 0.3323724416258288,
+      "loss": 6.780313491821289,
+      "loss_ce": 6.040079116821289,
+      "loss_xval": 0.7421875,
+      "num_input_tokens_seen": 170008032,
+      "step": 1153
+    },
+    {
+      "epoch": 0.33266070913808016,
+      "grad_norm": 139.2700863542851,
+      "learning_rate": 0.0001,
+      "loss": 7.8894,
+      "num_input_tokens_seen": 170143120,
+      "step": 1154
+    },
+    {
+      "epoch": 0.33266070913808016,
+      "loss": 7.91797399520874,
+      "loss_ce": 6.046880722045898,
+      "loss_xval": 1.875,
+      "num_input_tokens_seen": 170143120,
+      "step": 1154
+    },
+    {
+      "epoch": 0.3329489766503315,
+      "grad_norm": 149.41658480780492,
+      "learning_rate": 0.0001,
+      "loss": 7.7876,
+      "num_input_tokens_seen": 170315632,
+      "step": 1155
+    },
+    {
+      "epoch": 0.3329489766503315,
+      "loss": 8.109418869018555,
+      "loss_ce": 5.894575119018555,
+      "loss_xval": 2.21875,
+      "num_input_tokens_seen": 170315632,
+      "step": 1155
+    },
+    {
+      "epoch": 0.3332372441625829,
+      "grad_norm": 156.32637537503285,
+      "learning_rate": 0.0001,
+      "loss": 7.672,
+      "num_input_tokens_seen": 170450384,
+      "step": 1156
+    },
+    {
+      "epoch": 0.3332372441625829,
+      "loss": 7.428304672241211,
+      "loss_ce": 5.239828109741211,
+      "loss_xval": 2.1875,
+      "num_input_tokens_seen": 170450384,
+      "step": 1156
+    },
+    {
+      "epoch": 0.33352551167483424,
+      "grad_norm": 146.9594323082445,
+      "learning_rate": 0.0001,
+      "loss": 7.1932,
+      "num_input_tokens_seen": 170585472,
+      "step": 1157
+    },
+    {
+      "epoch": 0.33352551167483424,
+      "loss": 7.230058670043945,
+      "loss_ce": 5.124589920043945,
+      "loss_xval": 2.109375,
+      "num_input_tokens_seen": 170585472,
+      "step": 1157
+    },
+    {
+      "epoch": 0.3338137791870856,
+      "grad_norm": 107.99167328343886,
+      "learning_rate": 0.0001,
+      "loss": 5.884,
+      "num_input_tokens_seen": 170758080,
+      "step": 1158
+    },
+    {
+      "epoch": 0.3338137791870856,
+      "loss": 6.086777687072754,
+      "loss_ce": 4.977890968322754,
+      "loss_xval": 1.109375,
+      "num_input_tokens_seen": 170758080,
+      "step": 1158
+    },
+    {
+      "epoch": 0.33410204669933696,
+      "grad_norm": 67.67594647471446,
+      "learning_rate": 0.0001,
+      "loss": 4.9364,
+      "num_input_tokens_seen": 170892832,
+      "step": 1159
+    },
+    {
+      "epoch": 0.33410204669933696,
+      "loss": 4.73914098739624,
+      "loss_ce": 4.37781286239624,
+      "loss_xval": 0.361328125,
+      "num_input_tokens_seen": 170892832,
+      "step": 1159
+    },
+    {
+      "epoch": 0.3343903142115884,
+      "grad_norm": 27.192686899599124,
+      "learning_rate": 0.0001,
+      "loss": 4.5496,
+      "num_input_tokens_seen": 171027808,
+      "step": 1160
+    },
+    {
+      "epoch": 0.3343903142115884,
+      "loss": 4.496192932128906,
+      "loss_ce": 4.365577697753906,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 171027808,
+      "step": 1160
+    },
+    {
+      "epoch": 0.33467858172383974,
+      "grad_norm": 38.851402590144424,
+      "learning_rate": 0.0001,
+      "loss": 4.4712,
+      "num_input_tokens_seen": 171200368,
+      "step": 1161
+    },
+    {
+      "epoch": 0.33467858172383974,
+      "loss": 4.607762336730957,
+      "loss_ce": 4.399876594543457,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 171200368,
+      "step": 1161
+    },
+    {
+      "epoch": 0.3349668492360911,
+      "grad_norm": 76.11601509063742,
+      "learning_rate": 0.0001,
+      "loss": 4.634,
+      "num_input_tokens_seen": 171335200,
+      "step": 1162
+    },
+    {
+      "epoch": 0.3349668492360911,
+      "loss": 4.723811149597168,
+      "loss_ce": 3.919123888015747,
+      "loss_xval": 0.8046875,
+      "num_input_tokens_seen": 171335200,
+      "step": 1162
+    },
+    {
+      "epoch": 0.33525511674834246,
+      "grad_norm": 99.2778880784972,
+      "learning_rate": 0.0001,
+      "loss": 5.0097,
+      "num_input_tokens_seen": 171470312,
+      "step": 1163
+    },
+    {
+      "epoch": 0.33525511674834246,
+      "loss": 4.773837089538574,
+      "loss_ce": 4.016513347625732,
+      "loss_xval": 0.7578125,
+      "num_input_tokens_seen": 171470312,
+      "step": 1163
+    },
+    {
+      "epoch": 0.3355433842605938,
+      "grad_norm": 123.6317341440415,
+      "learning_rate": 0.0001,
+      "loss": 5.297,
+      "num_input_tokens_seen": 171642888,
+      "step": 1164
+    },
+    {
+      "epoch": 0.3355433842605938,
+      "loss": 5.352497577667236,
+      "loss_ce": 4.102497577667236,
+      "loss_xval": 1.25,
+      "num_input_tokens_seen": 171642888,
+      "step": 1164
+    },
+    {
+      "epoch": 0.3358316517728452,
+      "grad_norm": 104.57501260677738,
+      "learning_rate": 0.0001,
+      "loss": 4.8253,
+      "num_input_tokens_seen": 171777616,
+      "step": 1165
+    },
+    {
+      "epoch": 0.3358316517728452,
+      "loss": 4.950669765472412,
+      "loss_ce": 3.735337734222412,
+      "loss_xval": 1.21875,
+      "num_input_tokens_seen": 171777616,
+      "step": 1165
+    },
+    {
+      "epoch": 0.33611991928509655,
+      "grad_norm": 81.33891217763653,
+      "learning_rate": 0.0001,
+      "loss": 4.5942,
+      "num_input_tokens_seen": 171912712,
+      "step": 1166
+    },
+    {
+      "epoch": 0.33611991928509655,
+      "loss": 4.381400108337402,
+      "loss_ce": 3.8672401905059814,
+      "loss_xval": 0.515625,
+      "num_input_tokens_seen": 171912712,
+      "step": 1166
+    },
+    {
+      "epoch": 0.33640818679734796,
+      "grad_norm": 55.303487211926544,
+      "learning_rate": 0.0001,
+      "loss": 4.1956,
+      "num_input_tokens_seen": 172085144,
+      "step": 1167
+    },
+    {
+      "epoch": 0.33640818679734796,
+      "loss": 4.305488109588623,
+      "loss_ce": 4.013373851776123,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 172085144,
+      "step": 1167
+    },
+    {
+      "epoch": 0.3366964543095993,
+      "grad_norm": 16.54202519736511,
+      "learning_rate": 0.0001,
+      "loss": 3.8247,
+      "num_input_tokens_seen": 172219936,
+      "step": 1168
+    },
+    {
+      "epoch": 0.3366964543095993,
+      "loss": 3.7913873195648193,
+      "loss_ce": 3.6247613430023193,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 172219936,
+      "step": 1168
+    },
+    {
+      "epoch": 0.3369847218218507,
+      "grad_norm": 26.200108410757043,
+      "learning_rate": 0.0001,
+      "loss": 3.9935,
+      "num_input_tokens_seen": 172354984,
+      "step": 1169
+    },
+    {
+      "epoch": 0.3369847218218507,
+      "loss": 3.902951717376709,
+      "loss_ce": 3.763547420501709,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 172354984,
+      "step": 1169
+    },
+    {
+      "epoch": 0.33727298933410205,
+      "grad_norm": 47.584733452244784,
+      "learning_rate": 0.0001,
+      "loss": 4.0044,
+      "num_input_tokens_seen": 172527480,
+      "step": 1170
+    },
+    {
+      "epoch": 0.33727298933410205,
+      "loss": 4.209903717041016,
+      "loss_ce": 3.900089740753174,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 172527480,
+      "step": 1170
+    },
+    {
+      "epoch": 0.3375612568463534,
+      "grad_norm": 67.07075172810855,
+      "learning_rate": 0.0001,
+      "loss": 4.0503,
+      "num_input_tokens_seen": 172662256,
+      "step": 1171
+    },
+    {
+      "epoch": 0.3375612568463534,
+      "loss": 3.90846586227417,
+      "loss_ce": 3.51808500289917,
+      "loss_xval": 0.390625,
+      "num_input_tokens_seen": 172662256,
+      "step": 1171
+    },
+    {
+      "epoch": 0.33784952435860477,
+      "grad_norm": 78.6025130293129,
+      "learning_rate": 0.0001,
+      "loss": 4.3344,
+      "num_input_tokens_seen": 172797288,
+      "step": 1172
+    },
+    {
+      "epoch": 0.33784952435860477,
+      "loss": 4.278440475463867,
+      "loss_ce": 3.666623592376709,
+      "loss_xval": 0.61328125,
+      "num_input_tokens_seen": 172797288,
+      "step": 1172
+    },
+    {
+      "epoch": 0.33813779187085613,
+      "grad_norm": 65.58126499126392,
+      "learning_rate": 0.0001,
+      "loss": 4.0663,
+      "num_input_tokens_seen": 172969768,
+      "step": 1173
+    },
+    {
+      "epoch": 0.33813779187085613,
+      "loss": 4.310020446777344,
+      "loss_ce": 3.809532642364502,
+      "loss_xval": 0.5,
+      "num_input_tokens_seen": 172969768,
+      "step": 1173
+    },
+    {
+      "epoch": 0.33842605938310755,
+      "grad_norm": 53.08952868260317,
+      "learning_rate": 0.0001,
+      "loss": 3.8061,
+      "num_input_tokens_seen": 173104648,
+      "step": 1174
+    },
+    {
+      "epoch": 0.33842605938310755,
+      "loss": 3.6648547649383545,
+      "loss_ce": 3.4049670696258545,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 173104648,
+      "step": 1174
+    },
+    {
+      "epoch": 0.3387143268953589,
+      "grad_norm": 40.470371532877415,
+      "learning_rate": 0.0001,
+      "loss": 3.8175,
+      "num_input_tokens_seen": 173239736,
+      "step": 1175
+    },
+    {
+      "epoch": 0.3387143268953589,
+      "loss": 3.7364258766174316,
+      "loss_ce": 3.5298829078674316,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 173239736,
+      "step": 1175
+    },
+    {
+      "epoch": 0.33900259440761027,
+      "grad_norm": 9.107618648325966,
+      "learning_rate": 0.0001,
+      "loss": 3.5961,
+      "num_input_tokens_seen": 173412152,
+      "step": 1176
+    },
+    {
+      "epoch": 0.33900259440761027,
+      "loss": 3.7630152702331543,
+      "loss_ce": 3.6769556999206543,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 173412152,
+      "step": 1176
+    },
+    {
+      "epoch": 0.33929086191986163,
+      "grad_norm": 16.62523082092502,
+      "learning_rate": 0.0001,
+      "loss": 3.4821,
+      "num_input_tokens_seen": 173546992,
+      "step": 1177
+    },
+    {
+      "epoch": 0.33929086191986163,
+      "loss": 3.445162296295166,
+      "loss_ce": 3.302217960357666,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 173546992,
+      "step": 1177
+    },
+    {
+      "epoch": 0.339579129432113,
+      "grad_norm": 32.240317086567416,
+      "learning_rate": 0.0001,
+      "loss": 3.693,
+      "num_input_tokens_seen": 173681984,
+      "step": 1178
+    },
+    {
+      "epoch": 0.339579129432113,
+      "loss": 3.5491960048675537,
+      "loss_ce": 3.4296891689300537,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 173681984,
+      "step": 1178
+    },
+    {
+      "epoch": 0.33986739694436435,
+      "grad_norm": 52.75172198229805,
+      "learning_rate": 0.0001,
+      "loss": 3.7589,
+      "num_input_tokens_seen": 173854440,
+      "step": 1179
+    },
+    {
+      "epoch": 0.33986739694436435,
+      "loss": 3.916414976119995,
+      "loss_ce": 3.636019468307495,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 173854440,
+      "step": 1179
+    },
+    {
+      "epoch": 0.3401556644566157,
+      "grad_norm": 50.607619311064305,
+      "learning_rate": 0.0001,
+      "loss": 3.5949,
+      "num_input_tokens_seen": 173989424,
+      "step": 1180
+    },
+    {
+      "epoch": 0.3401556644566157,
+      "loss": 3.5826573371887207,
+      "loss_ce": 3.2337803840637207,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 173989424,
+      "step": 1180
+    },
+    {
+      "epoch": 0.34044393196886713,
+      "grad_norm": 40.62064593668211,
+      "learning_rate": 0.0001,
+      "loss": 3.6989,
+      "num_input_tokens_seen": 174124608,
+      "step": 1181
+    },
+    {
+      "epoch": 0.34044393196886713,
+      "loss": 3.5342178344726562,
+      "loss_ce": 3.3720474243164062,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 174124608,
+      "step": 1181
+    },
+    {
+      "epoch": 0.3407321994811185,
+      "grad_norm": 35.836911748651296,
+      "learning_rate": 0.0001,
+      "loss": 3.5721,
+      "num_input_tokens_seen": 174297080,
+      "step": 1182
+    },
+    {
+      "epoch": 0.3407321994811185,
+      "loss": 3.7330305576324463,
+      "loss_ce": 3.5919783115386963,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 174297080,
+      "step": 1182
+    },
+    {
+      "epoch": 0.34102046699336985,
+      "grad_norm": 16.492342672168352,
+      "learning_rate": 0.0001,
+      "loss": 3.3894,
+      "num_input_tokens_seen": 174432032,
+      "step": 1183
+    },
+    {
+      "epoch": 0.34102046699336985,
+      "loss": 3.316561222076416,
+      "loss_ce": 3.188875675201416,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 174432032,
+      "step": 1183
+    },
+    {
+      "epoch": 0.3413087345056212,
+      "grad_norm": 5.432967850334228,
+      "learning_rate": 0.0001,
+      "loss": 3.5067,
+      "num_input_tokens_seen": 174567080,
+      "step": 1184
+    },
+    {
+      "epoch": 0.3413087345056212,
+      "loss": 3.3840579986572266,
+      "loss_ce": 3.3247928619384766,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 174567080,
+      "step": 1184
+    },
+    {
+      "epoch": 0.3415970020178726,
+      "grad_norm": 11.636727885745076,
+      "learning_rate": 0.0001,
+      "loss": 3.456,
+      "num_input_tokens_seen": 174739592,
+      "step": 1185
+    },
+    {
+      "epoch": 0.3415970020178726,
+      "loss": 3.6579012870788574,
+      "loss_ce": 3.5756258964538574,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 174739592,
+      "step": 1185
+    },
+    {
+      "epoch": 0.34188526953012394,
+      "grad_norm": 28.073243632677382,
+      "learning_rate": 0.0001,
+      "loss": 3.3726,
+      "num_input_tokens_seen": 174874352,
+      "step": 1186
+    },
+    {
+      "epoch": 0.34188526953012394,
+      "loss": 3.2818756103515625,
+      "loss_ce": 3.1490631103515625,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 174874352,
+      "step": 1186
+    },
+    {
+      "epoch": 0.3421735370423753,
+      "grad_norm": 37.38558090166038,
+      "learning_rate": 0.0001,
+      "loss": 3.6097,
+      "num_input_tokens_seen": 175009344,
+      "step": 1187
+    },
+    {
+      "epoch": 0.3421735370423753,
+      "loss": 3.4909286499023438,
+      "loss_ce": 3.3041610717773438,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 175009344,
+      "step": 1187
+    },
+    {
+      "epoch": 0.3424618045546267,
+      "grad_norm": 32.136012294484296,
+      "learning_rate": 0.0001,
+      "loss": 3.5214,
+      "num_input_tokens_seen": 175181992,
+      "step": 1188
+    },
+    {
+      "epoch": 0.3424618045546267,
+      "loss": 3.7463483810424805,
+      "loss_ce": 3.5837507247924805,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 175181992,
+      "step": 1188
+    },
+    {
+      "epoch": 0.3427500720668781,
+      "grad_norm": 30.42733750158407,
+      "learning_rate": 0.0001,
+      "loss": 3.3589,
+      "num_input_tokens_seen": 175316728,
+      "step": 1189
+    },
+    {
+      "epoch": 0.3427500720668781,
+      "loss": 3.259042739868164,
+      "loss_ce": 3.126535415649414,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 175316728,
+      "step": 1189
+    },
+    {
+      "epoch": 0.34303833957912944,
+      "grad_norm": 21.403150656984053,
+      "learning_rate": 0.0001,
+      "loss": 3.5281,
+      "num_input_tokens_seen": 175451896,
+      "step": 1190
+    },
+    {
+      "epoch": 0.34303833957912944,
+      "loss": 3.385063409805298,
+      "loss_ce": 3.290947198867798,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 175451896,
+      "step": 1190
+    },
+    {
+      "epoch": 0.3433266070913808,
+      "grad_norm": 3.7029830708447267,
+      "learning_rate": 0.0001,
+      "loss": 3.406,
+      "num_input_tokens_seen": 175624320,
+      "step": 1191
+    },
+    {
+      "epoch": 0.3433266070913808,
+      "loss": 3.61179780960083,
+      "loss_ce": 3.54966402053833,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 175624320,
+      "step": 1191
+    },
+    {
+      "epoch": 0.34361487460363216,
+      "grad_norm": 9.347038645733567,
+      "learning_rate": 0.0001,
+      "loss": 3.2743,
+      "num_input_tokens_seen": 175759136,
+      "step": 1192
+    },
+    {
+      "epoch": 0.34361487460363216,
+      "loss": 3.2146201133728027,
+      "loss_ce": 3.1162314414978027,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 175759136,
+      "step": 1192
+    },
+    {
+      "epoch": 0.3439031421158835,
+      "grad_norm": 13.15268747761076,
+      "learning_rate": 0.0001,
+      "loss": 3.4675,
+      "num_input_tokens_seen": 175894248,
+      "step": 1193
+    },
+    {
+      "epoch": 0.3439031421158835,
+      "loss": 3.3171091079711914,
+      "loss_ce": 3.268967628479004,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 175894248,
+      "step": 1193
+    },
+    {
+      "epoch": 0.34419140962813494,
+      "grad_norm": 24.383472160045933,
+      "learning_rate": 0.0001,
+      "loss": 3.45,
+      "num_input_tokens_seen": 176066792,
+      "step": 1194
+    },
+    {
+      "epoch": 0.34419140962813494,
+      "loss": 3.663198947906494,
+      "loss_ce": 3.574148654937744,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 176066792,
+      "step": 1194
+    },
+    {
+      "epoch": 0.3444796771403863,
+      "grad_norm": 21.76778370017577,
+      "learning_rate": 0.0001,
+      "loss": 3.281,
+      "num_input_tokens_seen": 176201608,
+      "step": 1195
+    },
+    {
+      "epoch": 0.3444796771403863,
+      "loss": 3.234495162963867,
+      "loss_ce": 3.100400924682617,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 176201608,
+      "step": 1195
+    },
+    {
+      "epoch": 0.34476794465263766,
+      "grad_norm": 16.369405491156957,
+      "learning_rate": 0.0001,
+      "loss": 3.4813,
+      "num_input_tokens_seen": 176336784,
+      "step": 1196
+    },
+    {
+      "epoch": 0.34476794465263766,
+      "loss": 3.335388660430908,
+      "loss_ce": 3.2846226692199707,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 176336784,
+      "step": 1196
+    },
+    {
+      "epoch": 0.345056212164889,
+      "grad_norm": 16.78593163365355,
+      "learning_rate": 0.0001,
+      "loss": 3.3888,
+      "num_input_tokens_seen": 176509320,
+      "step": 1197
+    },
+    {
+      "epoch": 0.345056212164889,
+      "loss": 3.5923142433166504,
+      "loss_ce": 3.5209641456604004,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 176509320,
+      "step": 1197
+    },
+    {
+      "epoch": 0.3453444796771404,
+      "grad_norm": 6.970010570859095,
+      "learning_rate": 0.0001,
+      "loss": 3.2365,
+      "num_input_tokens_seen": 176644096,
+      "step": 1198
+    },
+    {
+      "epoch": 0.3453444796771404,
+      "loss": 3.164623975753784,
+      "loss_ce": 3.082195997238159,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 176644096,
+      "step": 1198
+    },
+    {
+      "epoch": 0.34563274718939174,
+      "grad_norm": 5.43104937026321,
+      "learning_rate": 0.0001,
+      "loss": 3.4491,
+      "num_input_tokens_seen": 176779144,
+      "step": 1199
+    },
+    {
+      "epoch": 0.34563274718939174,
+      "loss": 3.300656318664551,
+      "loss_ce": 3.257901191711426,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 176779144,
+      "step": 1199
+    },
+    {
+      "epoch": 0.3459210147016431,
+      "grad_norm": 6.256442475593534,
+      "learning_rate": 0.0001,
+      "loss": 3.3549,
+      "num_input_tokens_seen": 176951656,
+      "step": 1200
+    },
+    {
+      "epoch": 0.3459210147016431,
+      "loss": 3.5743367671966553,
+      "loss_ce": 3.5145528316497803,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 176951656,
+      "step": 1200
+    },
+    {
+      "epoch": 0.3462092822138945,
+      "grad_norm": 12.289159214842655,
+      "learning_rate": 0.0001,
+      "loss": 3.2364,
+      "num_input_tokens_seen": 177086424,
+      "step": 1201
+    },
+    {
+      "epoch": 0.3462092822138945,
+      "loss": 3.1313276290893555,
+      "loss_ce": 3.0714216232299805,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 177086424,
+      "step": 1201
+    },
+    {
+      "epoch": 0.3464975497261459,
+      "grad_norm": 16.61564645292739,
+      "learning_rate": 0.0001,
+      "loss": 3.4868,
+      "num_input_tokens_seen": 177221504,
+      "step": 1202
+    },
+    {
+      "epoch": 0.3464975497261459,
+      "loss": 3.34700083732605,
+      "loss_ce": 3.2785804271698,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 177221504,
+      "step": 1202
+    },
+    {
+      "epoch": 0.34678581723839724,
+      "grad_norm": 12.253516525379556,
+      "learning_rate": 0.0001,
+      "loss": 3.3667,
+      "num_input_tokens_seen": 177394048,
+      "step": 1203
+    },
+    {
+      "epoch": 0.34678581723839724,
+      "loss": 3.6046361923217773,
+      "loss_ce": 3.5355443954467773,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 177394048,
+      "step": 1203
+    },
+    {
+      "epoch": 0.3470740847506486,
+      "grad_norm": 12.680372469910816,
+      "learning_rate": 0.0001,
+      "loss": 3.2299,
+      "num_input_tokens_seen": 177528816,
+      "step": 1204
+    },
+    {
+      "epoch": 0.3470740847506486,
+      "loss": 3.1183619499206543,
+      "loss_ce": 3.0578150749206543,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 177528816,
+      "step": 1204
+    },
+    {
+      "epoch": 0.34736235226289996,
+      "grad_norm": 9.269347803679004,
+      "learning_rate": 0.0001,
+      "loss": 3.4058,
+      "num_input_tokens_seen": 177663768,
+      "step": 1205
+    },
+    {
+      "epoch": 0.34736235226289996,
+      "loss": 3.276123523712158,
+      "loss_ce": 3.232819080352783,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 177663768,
+      "step": 1205
+    },
+    {
+      "epoch": 0.3476506197751513,
+      "grad_norm": 4.73661481131897,
+      "learning_rate": 0.0001,
+      "loss": 3.3351,
+      "num_input_tokens_seen": 177836392,
+      "step": 1206
+    },
+    {
+      "epoch": 0.3476506197751513,
+      "loss": 3.5520756244659424,
+      "loss_ce": 3.5076420307159424,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 177836392,
+      "step": 1206
+    },
+    {
+      "epoch": 0.3479388872874027,
+      "grad_norm": 6.329114181074099,
+      "learning_rate": 0.0001,
+      "loss": 3.2034,
+      "num_input_tokens_seen": 177971168,
+      "step": 1207
+    },
+    {
+      "epoch": 0.3479388872874027,
+      "loss": 3.1209933757781982,
+      "loss_ce": 3.0564181804656982,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 177971168,
+      "step": 1207
+    },
+    {
+      "epoch": 0.3482271547996541,
+      "grad_norm": 6.65566027578529,
+      "learning_rate": 0.0001,
+      "loss": 3.41,
+      "num_input_tokens_seen": 178106168,
+      "step": 1208
+    },
+    {
+      "epoch": 0.3482271547996541,
+      "loss": 3.28511381149292,
+      "loss_ce": 3.2503085136413574,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 178106168,
+      "step": 1208
+    },
+    {
+      "epoch": 0.34851542231190547,
+      "grad_norm": 13.210336902177081,
+      "learning_rate": 0.0001,
+      "loss": 3.3444,
+      "num_input_tokens_seen": 178278728,
+      "step": 1209
+    },
+    {
+      "epoch": 0.34851542231190547,
+      "loss": 3.5691750049591064,
+      "loss_ce": 3.5233376026153564,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 178278728,
+      "step": 1209
+    },
+    {
+      "epoch": 0.3488036898241568,
+      "grad_norm": 11.57224749405127,
+      "learning_rate": 0.0001,
+      "loss": 3.2052,
+      "num_input_tokens_seen": 178413576,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3488036898241568,
+      "loss": 3.1245784759521484,
+      "loss_ce": 3.0383968353271484,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 178413576,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3490919573364082,
+      "grad_norm": 6.020088437621785,
+      "learning_rate": 0.0001,
+      "loss": 3.4461,
+      "num_input_tokens_seen": 178548736,
+      "step": 1211
+    },
+    {
+      "epoch": 0.3490919573364082,
+      "loss": 3.2839789390563965,
+      "loss_ce": 3.255948543548584,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 178548736,
+      "step": 1211
+    },
+    {
+      "epoch": 0.34938022484865955,
+      "grad_norm": 6.669576843869337,
+      "learning_rate": 0.0001,
+      "loss": 3.3146,
+      "num_input_tokens_seen": 178721200,
+      "step": 1212
+    },
+    {
+      "epoch": 0.34938022484865955,
+      "loss": 3.528001308441162,
+      "loss_ce": 3.485276699066162,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 178721200,
+      "step": 1212
+    },
+    {
+      "epoch": 0.3496684923609109,
+      "grad_norm": 1.913119357143956,
+      "learning_rate": 0.0001,
+      "loss": 3.1879,
+      "num_input_tokens_seen": 178856032,
+      "step": 1213
+    },
+    {
+      "epoch": 0.3496684923609109,
+      "loss": 3.0951290130615234,
+      "loss_ce": 3.0421199798583984,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 178856032,
+      "step": 1213
+    },
+    {
+      "epoch": 0.34995675987316227,
+      "grad_norm": 7.057834925247301,
+      "learning_rate": 0.0001,
+      "loss": 3.4293,
+      "num_input_tokens_seen": 178991144,
+      "step": 1214
+    },
+    {
+      "epoch": 0.34995675987316227,
+      "loss": 3.283670663833618,
+      "loss_ce": 3.247324228286743,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 178991144,
+      "step": 1214
+    },
+    {
+      "epoch": 0.3502450273854137,
+      "grad_norm": 6.700408570182067,
+      "learning_rate": 0.0001,
+      "loss": 3.3265,
+      "num_input_tokens_seen": 179163648,
+      "step": 1215
+    },
+    {
+      "epoch": 0.3502450273854137,
+      "loss": 3.5651307106018066,
+      "loss_ce": 3.5232300758361816,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 179163648,
+      "step": 1215
+    },
+    {
+      "epoch": 0.35053329489766505,
+      "grad_norm": 8.093485597628083,
+      "learning_rate": 0.0001,
+      "loss": 3.1862,
+      "num_input_tokens_seen": 179298472,
+      "step": 1216
+    },
+    {
+      "epoch": 0.35053329489766505,
+      "loss": 3.0838124752044678,
+      "loss_ce": 3.0416371822357178,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 179298472,
+      "step": 1216
+    },
+    {
+      "epoch": 0.3508215624099164,
+      "grad_norm": 10.871592735786908,
+      "learning_rate": 0.0001,
+      "loss": 3.3771,
+      "num_input_tokens_seen": 179433416,
+      "step": 1217
+    },
+    {
+      "epoch": 0.3508215624099164,
+      "loss": 3.2459707260131836,
+      "loss_ce": 3.212386131286621,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 179433416,
+      "step": 1217
+    },
+    {
+      "epoch": 0.35110982992216777,
+      "grad_norm": 2.6312062789379613,
+      "learning_rate": 0.0001,
+      "loss": 3.3024,
+      "num_input_tokens_seen": 179605904,
+      "step": 1218
+    },
+    {
+      "epoch": 0.35110982992216777,
+      "loss": 3.5236358642578125,
+      "loss_ce": 3.48828125,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 179605904,
+      "step": 1218
+    },
+    {
+      "epoch": 0.35139809743441913,
+      "grad_norm": 1.5401386473978997,
+      "learning_rate": 0.0001,
+      "loss": 3.1598,
+      "num_input_tokens_seen": 179740720,
+      "step": 1219
+    },
+    {
+      "epoch": 0.35139809743441913,
+      "loss": 3.079009532928467,
+      "loss_ce": 3.033355236053467,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 179740720,
+      "step": 1219
+    },
+    {
+      "epoch": 0.3516863649466705,
+      "grad_norm": 1.6760672752235724,
+      "learning_rate": 0.0001,
+      "loss": 3.3668,
+      "num_input_tokens_seen": 179875752,
+      "step": 1220
+    },
+    {
+      "epoch": 0.3516863649466705,
+      "loss": 3.232072353363037,
+      "loss_ce": 3.211595058441162,
+      "loss_xval": 0.0205078125,
+      "num_input_tokens_seen": 179875752,
+      "step": 1220
+    },
+    {
+      "epoch": 0.35197463245892185,
+      "grad_norm": 8.250299885867918,
+      "learning_rate": 0.0001,
+      "loss": 3.3206,
+      "num_input_tokens_seen": 180048296,
+      "step": 1221
+    },
+    {
+      "epoch": 0.35197463245892185,
+      "loss": 3.5465445518493652,
+      "loss_ce": 3.519040584564209,
+      "loss_xval": 0.0274658203125,
+      "num_input_tokens_seen": 180048296,
+      "step": 1221
+    },
+    {
+      "epoch": 0.35226289997117327,
+      "grad_norm": 7.823759219188541,
+      "learning_rate": 0.0001,
+      "loss": 3.1555,
+      "num_input_tokens_seen": 180183080,
+      "step": 1222
+    },
+    {
+      "epoch": 0.35226289997117327,
+      "loss": 3.0791444778442383,
+      "loss_ce": 3.0272035598754883,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 180183080,
+      "step": 1222
+    },
+    {
+      "epoch": 0.35255116748342463,
+      "grad_norm": 3.5751367712534434,
+      "learning_rate": 0.0001,
+      "loss": 3.3744,
+      "num_input_tokens_seen": 180318208,
+      "step": 1223
+    },
+    {
+      "epoch": 0.35255116748342463,
+      "loss": 3.250316619873047,
+      "loss_ce": 3.2265663146972656,
+      "loss_xval": 0.0238037109375,
+      "num_input_tokens_seen": 180318208,
+      "step": 1223
+    },
+    {
+      "epoch": 0.352839434995676,
+      "grad_norm": 8.228880590132833,
+      "learning_rate": 0.0001,
+      "loss": 3.3038,
+      "num_input_tokens_seen": 180490680,
+      "step": 1224
+    },
+    {
+      "epoch": 0.352839434995676,
+      "loss": 3.5290775299072266,
+      "loss_ce": 3.498727798461914,
+      "loss_xval": 0.0303955078125,
+      "num_input_tokens_seen": 180490680,
+      "step": 1224
+    },
+    {
+      "epoch": 0.35312770250792735,
+      "grad_norm": 2.959232776381592,
+      "learning_rate": 0.0001,
+      "loss": 3.1494,
+      "num_input_tokens_seen": 180625504,
+      "step": 1225
+    },
+    {
+      "epoch": 0.35312770250792735,
+      "loss": 3.0728955268859863,
+      "loss_ce": 3.0282177925109863,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 180625504,
+      "step": 1225
+    },
+    {
+      "epoch": 0.3534159700201787,
+      "grad_norm": 4.857446287211869,
+      "learning_rate": 0.0001,
+      "loss": 3.3712,
+      "num_input_tokens_seen": 180760592,
+      "step": 1226
+    },
+    {
+      "epoch": 0.3534159700201787,
+      "loss": 3.2360877990722656,
+      "loss_ce": 3.214008331298828,
+      "loss_xval": 0.0220947265625,
+      "num_input_tokens_seen": 180760592,
+      "step": 1226
+    },
+    {
+      "epoch": 0.3537042375324301,
+      "grad_norm": 1.6530887063049307,
+      "learning_rate": 0.0001,
+      "loss": 3.3088,
+      "num_input_tokens_seen": 180933232,
+      "step": 1227
+    },
+    {
+      "epoch": 0.3537042375324301,
+      "loss": 3.561347484588623,
+      "loss_ce": 3.5332255363464355,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 180933232,
+      "step": 1227
+    },
+    {
+      "epoch": 0.35399250504468144,
+      "grad_norm": 5.015458929984309,
+      "learning_rate": 0.0001,
+      "loss": 3.1558,
+      "num_input_tokens_seen": 181068040,
+      "step": 1228
+    },
+    {
+      "epoch": 0.35399250504468144,
+      "loss": 3.0620033740997314,
+      "loss_ce": 3.0263283252716064,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 181068040,
+      "step": 1228
+    },
+    {
+      "epoch": 0.35428077255693285,
+      "grad_norm": 8.938959384444903,
+      "learning_rate": 0.0001,
+      "loss": 3.3659,
+      "num_input_tokens_seen": 181203056,
+      "step": 1229
+    },
+    {
+      "epoch": 0.35428077255693285,
+      "loss": 3.2330641746520996,
+      "loss_ce": 3.206925868988037,
+      "loss_xval": 0.026123046875,
+      "num_input_tokens_seen": 181203056,
+      "step": 1229
+    },
+    {
+      "epoch": 0.3545690400691842,
+      "grad_norm": 2.5622534586346046,
+      "learning_rate": 0.0001,
+      "loss": 3.2811,
+      "num_input_tokens_seen": 181375560,
+      "step": 1230
+    },
+    {
+      "epoch": 0.3545690400691842,
+      "loss": 3.504213333129883,
+      "loss_ce": 3.477785110473633,
+      "loss_xval": 0.0263671875,
+      "num_input_tokens_seen": 181375560,
+      "step": 1230
+    },
+    {
+      "epoch": 0.3548573075814356,
+      "grad_norm": 2.349014538352901,
+      "learning_rate": 0.0001,
+      "loss": 3.1492,
+      "num_input_tokens_seen": 181510392,
+      "step": 1231
+    },
+    {
+      "epoch": 0.3548573075814356,
+      "loss": 3.057544231414795,
+      "loss_ce": 3.024768352508545,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 181510392,
+      "step": 1231
+    },
+    {
+      "epoch": 0.35514557509368694,
+      "grad_norm": 2.760313775015949,
+      "learning_rate": 0.0001,
+      "loss": 3.3364,
+      "num_input_tokens_seen": 181645400,
+      "step": 1232
+    },
+    {
+      "epoch": 0.35514557509368694,
+      "loss": 3.192826271057129,
+      "loss_ce": 3.1752023696899414,
+      "loss_xval": 0.017578125,
+      "num_input_tokens_seen": 181645400,
+      "step": 1232
+    },
+    {
+      "epoch": 0.3554338426059383,
+      "grad_norm": 3.7620834491475583,
+      "learning_rate": 0.0001,
+      "loss": 3.2669,
+      "num_input_tokens_seen": 181817928,
+      "step": 1233
+    },
+    {
+      "epoch": 0.3554338426059383,
+      "loss": 3.4794349670410156,
+      "loss_ce": 3.4616966247558594,
+      "loss_xval": 0.0177001953125,
+      "num_input_tokens_seen": 181817928,
+      "step": 1233
+    },
+    {
+      "epoch": 0.35572211011818966,
+      "grad_norm": 6.472062853720295,
+      "learning_rate": 0.0001,
+      "loss": 3.1369,
+      "num_input_tokens_seen": 181952752,
+      "step": 1234
+    },
+    {
+      "epoch": 0.35572211011818966,
+      "loss": 3.054111957550049,
+      "loss_ce": 3.013737201690674,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 181952752,
+      "step": 1234
+    },
+    {
+      "epoch": 0.356010377630441,
+      "grad_norm": 2.8472546072170744,
+      "learning_rate": 0.0001,
+      "loss": 3.3564,
+      "num_input_tokens_seen": 182087864,
+      "step": 1235
+    },
+    {
+      "epoch": 0.356010377630441,
+      "loss": 3.2165331840515137,
+      "loss_ce": 3.2055697441101074,
+      "loss_xval": 0.010986328125,
+      "num_input_tokens_seen": 182087864,
+      "step": 1235
+    },
+    {
+      "epoch": 0.35629864514269244,
+      "grad_norm": 6.785687454321519,
+      "learning_rate": 0.0001,
+      "loss": 3.2471,
+      "num_input_tokens_seen": 182260296,
+      "step": 1236
+    },
+    {
+      "epoch": 0.35629864514269244,
+      "loss": 3.4405596256256104,
+      "loss_ce": 3.4217989444732666,
+      "loss_xval": 0.018798828125,
+      "num_input_tokens_seen": 182260296,
+      "step": 1236
+    },
+    {
+      "epoch": 0.3565869126549438,
+      "grad_norm": 2.146040129543714,
+      "learning_rate": 0.0001,
+      "loss": 3.1209,
+      "num_input_tokens_seen": 182395104,
+      "step": 1237
+    },
+    {
+      "epoch": 0.3565869126549438,
+      "loss": 3.0413930416107178,
+      "loss_ce": 3.0080831050872803,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 182395104,
+      "step": 1237
+    },
+    {
+      "epoch": 0.35687518016719516,
+      "grad_norm": 2.974408010039056,
+      "learning_rate": 0.0001,
+      "loss": 3.3152,
+      "num_input_tokens_seen": 182530136,
+      "step": 1238
+    },
+    {
+      "epoch": 0.35687518016719516,
+      "loss": 3.1985116004943848,
+      "loss_ce": 3.1820931434631348,
+      "loss_xval": 0.016357421875,
+      "num_input_tokens_seen": 182530136,
+      "step": 1238
+    },
+    {
+      "epoch": 0.3571634476794465,
+      "grad_norm": 1.4409787530889788,
+      "learning_rate": 0.0001,
+      "loss": 3.243,
+      "num_input_tokens_seen": 182702624,
+      "step": 1239
+    },
+    {
+      "epoch": 0.3571634476794465,
+      "loss": 3.4570600986480713,
+      "loss_ce": 3.43831467628479,
+      "loss_xval": 0.018798828125,
+      "num_input_tokens_seen": 182702624,
+      "step": 1239
+    },
+    {
+      "epoch": 0.3574517151916979,
+      "grad_norm": 3.507385210087722,
+      "learning_rate": 0.0001,
+      "loss": 3.1193,
+      "num_input_tokens_seen": 182837392,
+      "step": 1240
+    },
+    {
+      "epoch": 0.3574517151916979,
+      "loss": 3.025404453277588,
+      "loss_ce": 3.001112461090088,
+      "loss_xval": 0.0242919921875,
+      "num_input_tokens_seen": 182837392,
+      "step": 1240
+    },
+    {
+      "epoch": 0.35773998270394924,
+      "grad_norm": 7.001759358957166,
+      "learning_rate": 0.0001,
+      "loss": 3.3462,
+      "num_input_tokens_seen": 182972408,
+      "step": 1241
+    },
+    {
+      "epoch": 0.35773998270394924,
+      "loss": 3.2213611602783203,
+      "loss_ce": 3.2040576934814453,
+      "loss_xval": 0.017333984375,
+      "num_input_tokens_seen": 182972408,
+      "step": 1241
+    },
+    {
+      "epoch": 0.35802825021620066,
+      "grad_norm": 2.538978744474648,
+      "learning_rate": 0.0001,
+      "loss": 3.2317,
+      "num_input_tokens_seen": 183144792,
+      "step": 1242
+    },
+    {
+      "epoch": 0.35802825021620066,
+      "loss": 3.4401183128356934,
+      "loss_ce": 3.4241347312927246,
+      "loss_xval": 0.0159912109375,
+      "num_input_tokens_seen": 183144792,
+      "step": 1242
+    },
+    {
+      "epoch": 0.358316517728452,
+      "grad_norm": 1.4479057315326487,
+      "learning_rate": 0.0001,
+      "loss": 3.1175,
+      "num_input_tokens_seen": 183279616,
+      "step": 1243
+    },
+    {
+      "epoch": 0.358316517728452,
+      "loss": 3.026390314102173,
+      "loss_ce": 2.9969866275787354,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 183279616,
+      "step": 1243
+    },
+    {
+      "epoch": 0.3586047852407034,
+      "grad_norm": 2.4414483373827145,
+      "learning_rate": 0.0001,
+      "loss": 3.3614,
+      "num_input_tokens_seen": 183414816,
+      "step": 1244
+    },
+    {
+      "epoch": 0.3586047852407034,
+      "loss": 3.198207378387451,
+      "loss_ce": 3.186450481414795,
+      "loss_xval": 0.01177978515625,
+      "num_input_tokens_seen": 183414816,
+      "step": 1244
+    },
+    {
+      "epoch": 0.35889305275295474,
+      "grad_norm": 5.1675116274784205,
+      "learning_rate": 0.0001,
+      "loss": 3.2394,
+      "num_input_tokens_seen": 183587312,
+      "step": 1245
+    },
+    {
+      "epoch": 0.35889305275295474,
+      "loss": 3.4583864212036133,
+      "loss_ce": 3.444554328918457,
+      "loss_xval": 0.01385498046875,
+      "num_input_tokens_seen": 183587312,
+      "step": 1245
+    },
+    {
+      "epoch": 0.3591813202652061,
+      "grad_norm": 3.0400243211955966,
+      "learning_rate": 0.0001,
+      "loss": 3.1146,
+      "num_input_tokens_seen": 183722120,
+      "step": 1246
+    },
+    {
+      "epoch": 0.3591813202652061,
+      "loss": 3.0098531246185303,
+      "loss_ce": 2.9842259883880615,
+      "loss_xval": 0.025634765625,
+      "num_input_tokens_seen": 183722120,
+      "step": 1246
+    },
+    {
+      "epoch": 0.35946958777745747,
+      "grad_norm": 1.5009990234436026,
+      "learning_rate": 0.0001,
+      "loss": 3.2998,
+      "num_input_tokens_seen": 183857160,
+      "step": 1247
+    },
+    {
+      "epoch": 0.35946958777745747,
+      "loss": 3.1848437786102295,
+      "loss_ce": 3.1744449138641357,
+      "loss_xval": 0.0103759765625,
+      "num_input_tokens_seen": 183857160,
+      "step": 1247
+    },
+    {
+      "epoch": 0.35975785528970883,
+      "grad_norm": 4.303761457456079,
+      "learning_rate": 0.0001,
+      "loss": 3.2222,
+      "num_input_tokens_seen": 184029680,
+      "step": 1248
+    },
+    {
+      "epoch": 0.35975785528970883,
+      "loss": 3.4391422271728516,
+      "loss_ce": 3.4264888763427734,
+      "loss_xval": 0.01263427734375,
+      "num_input_tokens_seen": 184029680,
+      "step": 1248
+    },
+    {
+      "epoch": 0.36004612280196024,
+      "grad_norm": 1.402284473926413,
+      "learning_rate": 0.0001,
+      "loss": 3.0968,
+      "num_input_tokens_seen": 184164528,
+      "step": 1249
+    },
+    {
+      "epoch": 0.36004612280196024,
+      "loss": 3.003244638442993,
+      "loss_ce": 2.981996774673462,
+      "loss_xval": 0.021240234375,
+      "num_input_tokens_seen": 184164528,
+      "step": 1249
+    },
+    {
+      "epoch": 0.3603343903142116,
+      "grad_norm": 4.291548001512167,
+      "learning_rate": 0.0001,
+      "loss": 3.3356,
+      "num_input_tokens_seen": 184299688,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3603343903142116,
+      "eval_websight_new_IoU": 0.06377699971199036,
+      "eval_websight_new_MAE_x": 0.11152733862400055,
+      "eval_websight_new_MAE_y": 0.070404052734375,
+      "eval_websight_new_NUM_probability": 0.10237714275717735,
+      "eval_websight_new_inside_bbox": 0.07465277798473835,
+      "eval_websight_new_loss": 3.1641979217529297,
+      "eval_websight_new_loss_ce": 3.14769971370697,
+      "eval_websight_new_loss_xval": 0.011007308959960938,
+      "eval_websight_new_runtime": 35.8519,
+      "eval_websight_new_samples_per_second": 1.395,
+      "eval_websight_new_steps_per_second": 0.056,
+      "num_input_tokens_seen": 184299688,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3603343903142116,
+      "eval_seeclick_IoU": 0.0957968607544899,
+      "eval_seeclick_MAE_x": 0.12743081152439117,
+      "eval_seeclick_MAE_y": 0.1018507219851017,
+      "eval_seeclick_NUM_probability": 0.09817633777856827,
+      "eval_seeclick_inside_bbox": 0.2725694477558136,
+      "eval_seeclick_loss": 2.995253324508667,
+      "eval_seeclick_loss_ce": 2.976414203643799,
+      "eval_seeclick_loss_xval": 0.021217823028564453,
+      "eval_seeclick_runtime": 62.8692,
+      "eval_seeclick_samples_per_second": 0.795,
+      "eval_seeclick_steps_per_second": 0.032,
+      "num_input_tokens_seen": 184299688,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3603343903142116,
+      "eval_icons_IoU": 0.017014755867421627,
+      "eval_icons_MAE_x": 0.09833865612745285,
+      "eval_icons_MAE_y": 0.08148108422756195,
+      "eval_icons_NUM_probability": 0.09484974667429924,
+      "eval_icons_inside_bbox": 0.02777777798473835,
+      "eval_icons_loss": 3.481070041656494,
+      "eval_icons_loss_ce": 3.4514541625976562,
+      "eval_icons_loss_xval": 0.010906219482421875,
+      "eval_icons_runtime": 65.653,
+      "eval_icons_samples_per_second": 0.762,
+      "eval_icons_steps_per_second": 0.03,
+      "num_input_tokens_seen": 184299688,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3603343903142116,
+      "loss": 3.4562795162200928,
+      "loss_ce": 3.4461476802825928,
+      "loss_xval": 0.0101318359375,
+      "num_input_tokens_seen": 184299688,
+      "step": 1250
+    },
+    {
+      "epoch": 0.36062265782646297,
+      "grad_norm": 1.935079332583963,
+      "learning_rate": 0.0001,
+      "loss": 3.2156,
+      "num_input_tokens_seen": 184472272,
+      "step": 1251
+    },
+    {
+      "epoch": 0.36062265782646297,
+      "loss": 3.437589406967163,
+      "loss_ce": 3.425779104232788,
+      "loss_xval": 0.0118408203125,
+      "num_input_tokens_seen": 184472272,
+      "step": 1251
+    },
+    {
+      "epoch": 0.36091092533871433,
+      "grad_norm": 3.5075152116035633,
+      "learning_rate": 0.0001,
+      "loss": 3.0755,
+      "num_input_tokens_seen": 184606984,
+      "step": 1252
+    },
+    {
+      "epoch": 0.36091092533871433,
+      "loss": 2.988539695739746,
+      "loss_ce": 2.9702444076538086,
+      "loss_xval": 0.018310546875,
+      "num_input_tokens_seen": 184606984,
+      "step": 1252
+    },
+    {
+      "epoch": 0.3611991928509657,
+      "grad_norm": 4.050070963936819,
+      "learning_rate": 0.0001,
+      "loss": 3.2879,
+      "num_input_tokens_seen": 184741984,
+      "step": 1253
+    },
+    {
+      "epoch": 0.3611991928509657,
+      "loss": 3.1794817447662354,
+      "loss_ce": 3.167839288711548,
+      "loss_xval": 0.01165771484375,
+      "num_input_tokens_seen": 184741984,
+      "step": 1253
+    },
+    {
+      "epoch": 0.36148746036321705,
+      "grad_norm": 1.6716885708857125,
+      "learning_rate": 0.0001,
+      "loss": 3.1998,
+      "num_input_tokens_seen": 184914544,
+      "step": 1254
+    },
+    {
+      "epoch": 0.36148746036321705,
+      "loss": 3.411571979522705,
+      "loss_ce": 3.401855945587158,
+      "loss_xval": 0.00970458984375,
+      "num_input_tokens_seen": 184914544,
+      "step": 1254
+    },
+    {
+      "epoch": 0.3617757278754684,
+      "grad_norm": 2.3067478578370966,
+      "learning_rate": 0.0001,
+      "loss": 3.0845,
+      "num_input_tokens_seen": 185049448,
+      "step": 1255
+    },
+    {
+      "epoch": 0.3617757278754684,
+      "loss": 2.986309289932251,
+      "loss_ce": 2.9643900394439697,
+      "loss_xval": 0.02197265625,
+      "num_input_tokens_seen": 185049448,
+      "step": 1255
+    },
+    {
+      "epoch": 0.36206399538771983,
+      "grad_norm": 1.6005172720100542,
+      "learning_rate": 0.0001,
+      "loss": 3.2921,
+      "num_input_tokens_seen": 185184592,
+      "step": 1256
+    },
+    {
+      "epoch": 0.36206399538771983,
+      "loss": 3.1446337699890137,
+      "loss_ce": 3.137214183807373,
+      "loss_xval": 0.007415771484375,
+      "num_input_tokens_seen": 185184592,
+      "step": 1256
+    },
+    {
+      "epoch": 0.3623522628999712,
+      "grad_norm": 5.222734644437509,
+      "learning_rate": 0.0001,
+      "loss": 3.2062,
+      "num_input_tokens_seen": 185357176,
+      "step": 1257
+    },
+    {
+      "epoch": 0.3623522628999712,
+      "loss": 3.431488513946533,
+      "loss_ce": 3.4212613105773926,
+      "loss_xval": 0.01025390625,
+      "num_input_tokens_seen": 185357176,
+      "step": 1257
+    },
+    {
+      "epoch": 0.36264053041222255,
+      "grad_norm": 1.6688855379400882,
+      "learning_rate": 0.0001,
+      "loss": 3.0652,
+      "num_input_tokens_seen": 185492032,
+      "step": 1258
+    },
+    {
+      "epoch": 0.36264053041222255,
+      "loss": 2.9728851318359375,
+      "loss_ce": 2.950164794921875,
+      "loss_xval": 0.022705078125,
+      "num_input_tokens_seen": 185492032,
+      "step": 1258
+    },
+    {
+      "epoch": 0.3629287979244739,
+      "grad_norm": 2.497505358631419,
+      "learning_rate": 0.0001,
+      "loss": 3.2925,
+      "num_input_tokens_seen": 185627128,
+      "step": 1259
+    },
+    {
+      "epoch": 0.3629287979244739,
+      "loss": 3.153623580932617,
+      "loss_ce": 3.142393112182617,
+      "loss_xval": 0.01123046875,
+      "num_input_tokens_seen": 185627128,
+      "step": 1259
+    },
+    {
+      "epoch": 0.3632170654367253,
+      "grad_norm": 1.3269461584050222,
+      "learning_rate": 0.0001,
+      "loss": 3.2288,
+      "num_input_tokens_seen": 185799800,
+      "step": 1260
+    },
+    {
+      "epoch": 0.3632170654367253,
+      "loss": 3.494309186935425,
+      "loss_ce": 3.483990430831909,
+      "loss_xval": 0.01031494140625,
+      "num_input_tokens_seen": 185799800,
+      "step": 1260
+    },
+    {
+      "epoch": 0.36350533294897663,
+      "grad_norm": 2.999073864405773,
+      "learning_rate": 0.0001,
+      "loss": 3.0764,
+      "num_input_tokens_seen": 185934760,
+      "step": 1261
+    },
+    {
+      "epoch": 0.36350533294897663,
+      "loss": 2.960078477859497,
+      "loss_ce": 2.9436981678009033,
+      "loss_xval": 0.016357421875,
+      "num_input_tokens_seen": 185934760,
+      "step": 1261
+    },
+    {
+      "epoch": 0.363793600461228,
+      "grad_norm": 4.427508635851775,
+      "learning_rate": 0.0001,
+      "loss": 3.2545,
+      "num_input_tokens_seen": 186069728,
+      "step": 1262
+    },
+    {
+      "epoch": 0.363793600461228,
+      "loss": 3.137115240097046,
+      "loss_ce": 3.1245877742767334,
+      "loss_xval": 0.01251220703125,
+      "num_input_tokens_seen": 186069728,
+      "step": 1262
+    },
+    {
+      "epoch": 0.3640818679734794,
+      "grad_norm": 1.3243023539960694,
+      "learning_rate": 0.0001,
+      "loss": 3.2047,
+      "num_input_tokens_seen": 186242096,
+      "step": 1263
+    },
+    {
+      "epoch": 0.3640818679734794,
+      "loss": 3.4566519260406494,
+      "loss_ce": 3.4479429721832275,
+      "loss_xval": 0.00872802734375,
+      "num_input_tokens_seen": 186242096,
+      "step": 1263
+    },
+    {
+      "epoch": 0.3643701354857308,
+      "grad_norm": 1.492897020777446,
+      "learning_rate": 0.0001,
+      "loss": 3.0452,
+      "num_input_tokens_seen": 186376880,
+      "step": 1264
+    },
+    {
+      "epoch": 0.3643701354857308,
+      "loss": 2.953927993774414,
+      "loss_ce": 2.9349613189697266,
+      "loss_xval": 0.0189208984375,
+      "num_input_tokens_seen": 186376880,
+      "step": 1264
+    },
+    {
+      "epoch": 0.36465840299798213,
+      "grad_norm": 1.450417138161176,
+      "learning_rate": 0.0001,
+      "loss": 3.2671,
+      "num_input_tokens_seen": 186511944,
+      "step": 1265
+    },
+    {
+      "epoch": 0.36465840299798213,
+      "loss": 3.1252617835998535,
+      "loss_ce": 3.1166253089904785,
+      "loss_xval": 0.0086669921875,
+      "num_input_tokens_seen": 186511944,
+      "step": 1265
+    },
+    {
+      "epoch": 0.3649466705102335,
+      "grad_norm": 3.524096074421607,
+      "learning_rate": 0.0001,
+      "loss": 3.1648,
+      "num_input_tokens_seen": 186684480,
+      "step": 1266
+    },
+    {
+      "epoch": 0.3649466705102335,
+      "loss": 3.383352041244507,
+      "loss_ce": 3.3738229274749756,
+      "loss_xval": 0.009521484375,
+      "num_input_tokens_seen": 186684480,
+      "step": 1266
+    },
+    {
+      "epoch": 0.36523493802248486,
+      "grad_norm": 1.8573647817225514,
+      "learning_rate": 0.0001,
+      "loss": 3.0348,
+      "num_input_tokens_seen": 186819320,
+      "step": 1267
+    },
+    {
+      "epoch": 0.36523493802248486,
+      "loss": 2.93626070022583,
+      "loss_ce": 2.91801118850708,
+      "loss_xval": 0.018310546875,
+      "num_input_tokens_seen": 186819320,
+      "step": 1267
+    },
+    {
+      "epoch": 0.3655232055347362,
+      "grad_norm": 1.6620050972238471,
+      "learning_rate": 0.0001,
+      "loss": 3.2388,
+      "num_input_tokens_seen": 186954360,
+      "step": 1268
+    },
+    {
+      "epoch": 0.3655232055347362,
+      "loss": 3.094966173171997,
+      "loss_ce": 3.0861923694610596,
+      "loss_xval": 0.0087890625,
+      "num_input_tokens_seen": 186954360,
+      "step": 1268
+    },
+    {
+      "epoch": 0.3658114730469876,
+      "grad_norm": 1.6554413422711625,
+      "learning_rate": 0.0001,
+      "loss": 3.1737,
+      "num_input_tokens_seen": 187126824,
+      "step": 1269
+    },
+    {
+      "epoch": 0.3658114730469876,
+      "loss": 3.4225668907165527,
+      "loss_ce": 3.4151244163513184,
+      "loss_xval": 0.0074462890625,
+      "num_input_tokens_seen": 187126824,
+      "step": 1269
+    },
+    {
+      "epoch": 0.366099740559239,
+      "grad_norm": 1.7051155693016407,
+      "learning_rate": 0.0001,
+      "loss": 3.0186,
+      "num_input_tokens_seen": 187261608,
+      "step": 1270
+    },
+    {
+      "epoch": 0.366099740559239,
+      "loss": 2.921408176422119,
+      "loss_ce": 2.9070725440979004,
+      "loss_xval": 0.01434326171875,
+      "num_input_tokens_seen": 187261608,
+      "step": 1270
+    },
+    {
+      "epoch": 0.36638800807149036,
+      "grad_norm": 4.265203473661172,
+      "learning_rate": 0.0001,
+      "loss": 3.236,
+      "num_input_tokens_seen": 187396552,
+      "step": 1271
+    },
+    {
+      "epoch": 0.36638800807149036,
+      "loss": 3.108557939529419,
+      "loss_ce": 3.0978996753692627,
+      "loss_xval": 0.01068115234375,
+      "num_input_tokens_seen": 187396552,
+      "step": 1271
+    },
+    {
+      "epoch": 0.3666762755837417,
+      "grad_norm": 1.495532032734321,
+      "learning_rate": 0.0001,
+      "loss": 3.1714,
+      "num_input_tokens_seen": 187569184,
+      "step": 1272
+    },
+    {
+      "epoch": 0.3666762755837417,
+      "loss": 3.429234504699707,
+      "loss_ce": 3.420933723449707,
+      "loss_xval": 0.00830078125,
+      "num_input_tokens_seen": 187569184,
+      "step": 1272
+    },
+    {
+      "epoch": 0.3669645430959931,
+      "grad_norm": 1.5397754658188914,
+      "learning_rate": 0.0001,
+      "loss": 2.9948,
+      "num_input_tokens_seen": 187703888,
+      "step": 1273
+    },
+    {
+      "epoch": 0.3669645430959931,
+      "loss": 2.907548427581787,
+      "loss_ce": 2.8901076316833496,
+      "loss_xval": 0.0174560546875,
+      "num_input_tokens_seen": 187703888,
+      "step": 1273
+    },
+    {
+      "epoch": 0.36725281060824444,
+      "grad_norm": 2.2236850531763146,
+      "learning_rate": 0.0001,
+      "loss": 3.2323,
+      "num_input_tokens_seen": 187838936,
+      "step": 1274
+    },
+    {
+      "epoch": 0.36725281060824444,
+      "loss": 3.095729351043701,
+      "loss_ce": 3.0877223014831543,
+      "loss_xval": 0.00799560546875,
+      "num_input_tokens_seen": 187838936,
+      "step": 1274
+    },
+    {
+      "epoch": 0.3675410781204958,
+      "grad_norm": 2.909090515749406,
+      "learning_rate": 0.0001,
+      "loss": 3.1036,
+      "num_input_tokens_seen": 188011592,
+      "step": 1275
+    },
+    {
+      "epoch": 0.3675410781204958,
+      "loss": 3.313185214996338,
+      "loss_ce": 3.3051209449768066,
+      "loss_xval": 0.008056640625,
+      "num_input_tokens_seen": 188011592,
+      "step": 1275
+    },
+    {
+      "epoch": 0.36782934563274716,
+      "grad_norm": 2.5094342235680953,
+      "learning_rate": 0.0001,
+      "loss": 2.9943,
+      "num_input_tokens_seen": 188146424,
+      "step": 1276
+    },
+    {
+      "epoch": 0.36782934563274716,
+      "loss": 2.892660617828369,
+      "loss_ce": 2.876692295074463,
+      "loss_xval": 0.0159912109375,
+      "num_input_tokens_seen": 188146424,
+      "step": 1276
+    },
+    {
+      "epoch": 0.3681176131449986,
+      "grad_norm": 2.8488290702724988,
+      "learning_rate": 0.0001,
+      "loss": 3.2013,
+      "num_input_tokens_seen": 188281472,
+      "step": 1277
+    },
+    {
+      "epoch": 0.3681176131449986,
+      "loss": 3.077254295349121,
+      "loss_ce": 3.068915367126465,
+      "loss_xval": 0.00836181640625,
+      "num_input_tokens_seen": 188281472,
+      "step": 1277
+    },
+    {
+      "epoch": 0.36840588065724994,
+      "grad_norm": 1.8587305330114212,
+      "learning_rate": 0.0001,
+      "loss": 3.1322,
+      "num_input_tokens_seen": 188454168,
+      "step": 1278
+    },
+    {
+      "epoch": 0.36840588065724994,
+      "loss": 3.390192985534668,
+      "loss_ce": 3.3831357955932617,
+      "loss_xval": 0.007049560546875,
+      "num_input_tokens_seen": 188454168,
+      "step": 1278
+    },
+    {
+      "epoch": 0.3686941481695013,
+      "grad_norm": 1.6701222144616605,
+      "learning_rate": 0.0001,
+      "loss": 2.961,
+      "num_input_tokens_seen": 188588904,
+      "step": 1279
+    },
+    {
+      "epoch": 0.3686941481695013,
+      "loss": 2.8679046630859375,
+      "loss_ce": 2.8522682189941406,
+      "loss_xval": 0.015625,
+      "num_input_tokens_seen": 188588904,
+      "step": 1279
+    },
+    {
+      "epoch": 0.36898241568175266,
+      "grad_norm": 3.397635660389669,
+      "learning_rate": 0.0001,
+      "loss": 3.1997,
+      "num_input_tokens_seen": 188724008,
+      "step": 1280
+    },
+    {
+      "epoch": 0.36898241568175266,
+      "loss": 3.0433454513549805,
+      "loss_ce": 3.0348386764526367,
+      "loss_xval": 0.00848388671875,
+      "num_input_tokens_seen": 188724008,
+      "step": 1280
+    },
+    {
+      "epoch": 0.369270683194004,
+      "grad_norm": 1.7129980805898721,
+      "learning_rate": 0.0001,
+      "loss": 3.0605,
+      "num_input_tokens_seen": 188896536,
+      "step": 1281
+    },
+    {
+      "epoch": 0.369270683194004,
+      "loss": 3.2636823654174805,
+      "loss_ce": 3.25540828704834,
+      "loss_xval": 0.00830078125,
+      "num_input_tokens_seen": 188896536,
+      "step": 1281
+    },
+    {
+      "epoch": 0.3695589507062554,
+      "grad_norm": 1.674071501036216,
+      "learning_rate": 0.0001,
+      "loss": 2.9341,
+      "num_input_tokens_seen": 189031424,
+      "step": 1282
+    },
+    {
+      "epoch": 0.3695589507062554,
+      "loss": 2.8433218002319336,
+      "loss_ce": 2.828688621520996,
+      "loss_xval": 0.0146484375,
+      "num_input_tokens_seen": 189031424,
+      "step": 1282
+    },
+    {
+      "epoch": 0.3698472182185068,
+      "grad_norm": 2.632451663005747,
+      "learning_rate": 0.0001,
+      "loss": 3.1643,
+      "num_input_tokens_seen": 189166496,
+      "step": 1283
+    },
+    {
+      "epoch": 0.3698472182185068,
+      "loss": 3.038170337677002,
+      "loss_ce": 3.0290989875793457,
+      "loss_xval": 0.00909423828125,
+      "num_input_tokens_seen": 189166496,
+      "step": 1283
+    },
+    {
+      "epoch": 0.37013548573075816,
+      "grad_norm": 2.0903033276496874,
+      "learning_rate": 0.0001,
+      "loss": 3.0753,
+      "num_input_tokens_seen": 189338952,
+      "step": 1284
+    },
+    {
+      "epoch": 0.37013548573075816,
+      "loss": 3.3279080390930176,
+      "loss_ce": 3.3198323249816895,
+      "loss_xval": 0.008056640625,
+      "num_input_tokens_seen": 189338952,
+      "step": 1284
+    },
+    {
+      "epoch": 0.3704237532430095,
+      "grad_norm": 1.8716433509508976,
+      "learning_rate": 0.0001,
+      "loss": 2.9242,
+      "num_input_tokens_seen": 189473744,
+      "step": 1285
+    },
+    {
+      "epoch": 0.3704237532430095,
+      "loss": 2.816052198410034,
+      "loss_ce": 2.800999402999878,
+      "loss_xval": 0.01507568359375,
+      "num_input_tokens_seen": 189473744,
+      "step": 1285
+    },
+    {
+      "epoch": 0.3707120207552609,
+      "grad_norm": 2.9900748452227637,
+      "learning_rate": 0.0001,
+      "loss": 3.1373,
+      "num_input_tokens_seen": 189608856,
+      "step": 1286
+    },
+    {
+      "epoch": 0.3707120207552609,
+      "loss": 3.0003795623779297,
+      "loss_ce": 2.991331100463867,
+      "loss_xval": 0.009033203125,
+      "num_input_tokens_seen": 189608856,
+      "step": 1286
+    },
+    {
+      "epoch": 0.37100028826751225,
+      "grad_norm": 2.0213465984800454,
+      "learning_rate": 0.0001,
+      "loss": 3.0273,
+      "num_input_tokens_seen": 189781312,
+      "step": 1287
+    },
+    {
+      "epoch": 0.37100028826751225,
+      "loss": 3.2566895484924316,
+      "loss_ce": 3.2470154762268066,
+      "loss_xval": 0.0096435546875,
+      "num_input_tokens_seen": 189781312,
+      "step": 1287
+    },
+    {
+      "epoch": 0.3712885557797636,
+      "grad_norm": 2.0433003449427587,
+      "learning_rate": 0.0001,
+      "loss": 2.874,
+      "num_input_tokens_seen": 189916048,
+      "step": 1288
+    },
+    {
+      "epoch": 0.3712885557797636,
+      "loss": 2.785282611846924,
+      "loss_ce": 2.7689785957336426,
+      "loss_xval": 0.016357421875,
+      "num_input_tokens_seen": 189916048,
+      "step": 1288
+    },
+    {
+      "epoch": 0.37157682329201497,
+      "grad_norm": 3.2920199908713124,
+      "learning_rate": 0.0001,
+      "loss": 3.0956,
+      "num_input_tokens_seen": 190051064,
+      "step": 1289
+    },
+    {
+      "epoch": 0.37157682329201497,
+      "loss": 2.984703302383423,
+      "loss_ce": 2.971916437149048,
+      "loss_xval": 0.0128173828125,
+      "num_input_tokens_seen": 190051064,
+      "step": 1289
+    },
+    {
+      "epoch": 0.3718650908042664,
+      "grad_norm": 2.2241646450849935,
+      "learning_rate": 0.0001,
+      "loss": 3.0093,
+      "num_input_tokens_seen": 190223664,
+      "step": 1290
+    },
+    {
+      "epoch": 0.3718650908042664,
+      "loss": 3.2635810375213623,
+      "loss_ce": 3.2557456493377686,
+      "loss_xval": 0.0078125,
+      "num_input_tokens_seen": 190223664,
+      "step": 1290
+    },
+    {
+      "epoch": 0.37215335831651775,
+      "grad_norm": 2.3558950213229597,
+      "learning_rate": 0.0001,
+      "loss": 2.8469,
+      "num_input_tokens_seen": 190358544,
+      "step": 1291
+    },
+    {
+      "epoch": 0.37215335831651775,
+      "loss": 2.741013526916504,
+      "loss_ce": 2.72432804107666,
+      "loss_xval": 0.0167236328125,
+      "num_input_tokens_seen": 190358544,
+      "step": 1291
+    },
+    {
+      "epoch": 0.3724416258287691,
+      "grad_norm": 3.155953361472079,
+      "learning_rate": 0.0001,
+      "loss": 3.0072,
+      "num_input_tokens_seen": 190493376,
+      "step": 1292
+    },
+    {
+      "epoch": 0.3724416258287691,
+      "loss": 2.8824682235717773,
+      "loss_ce": 2.8714895248413086,
+      "loss_xval": 0.010986328125,
+      "num_input_tokens_seen": 190493376,
+      "step": 1292
+    },
+    {
+      "epoch": 0.37272989334102047,
+      "grad_norm": 2.677295979464333,
+      "learning_rate": 0.0001,
+      "loss": 2.9604,
+      "num_input_tokens_seen": 190665872,
+      "step": 1293
+    },
+    {
+      "epoch": 0.37272989334102047,
+      "loss": 3.2144699096679688,
+      "loss_ce": 3.202442169189453,
+      "loss_xval": 0.01202392578125,
+      "num_input_tokens_seen": 190665872,
+      "step": 1293
+    },
+    {
+      "epoch": 0.37301816085327183,
+      "grad_norm": 2.6246963289673144,
+      "learning_rate": 0.0001,
+      "loss": 2.7918,
+      "num_input_tokens_seen": 190800688,
+      "step": 1294
+    },
+    {
+      "epoch": 0.37301816085327183,
+      "loss": 2.69120717048645,
+      "loss_ce": 2.6758530139923096,
+      "loss_xval": 0.015380859375,
+      "num_input_tokens_seen": 190800688,
+      "step": 1294
+    },
+    {
+      "epoch": 0.3733064283655232,
+      "grad_norm": 3.4790256525746,
+      "learning_rate": 0.0001,
+      "loss": 2.9933,
+      "num_input_tokens_seen": 190935744,
+      "step": 1295
+    },
+    {
+      "epoch": 0.3733064283655232,
+      "loss": 2.8375144004821777,
+      "loss_ce": 2.8264365196228027,
+      "loss_xval": 0.0111083984375,
+      "num_input_tokens_seen": 190935744,
+      "step": 1295
+    },
+    {
+      "epoch": 0.37359469587777455,
+      "grad_norm": 2.8185221481052762,
+      "learning_rate": 0.0001,
+      "loss": 2.8728,
+      "num_input_tokens_seen": 191108200,
+      "step": 1296
+    },
+    {
+      "epoch": 0.37359469587777455,
+      "loss": 3.109882354736328,
+      "loss_ce": 3.100902557373047,
+      "loss_xval": 0.00897216796875,
+      "num_input_tokens_seen": 191108200,
+      "step": 1296
+    },
+    {
+      "epoch": 0.37388296339002597,
+      "grad_norm": 3.0080030797377493,
+      "learning_rate": 0.0001,
+      "loss": 2.7236,
+      "num_input_tokens_seen": 191242960,
+      "step": 1297
+    },
+    {
+      "epoch": 0.37388296339002597,
+      "loss": 2.617522954940796,
+      "loss_ce": 2.5978238582611084,
+      "loss_xval": 0.0196533203125,
+      "num_input_tokens_seen": 191242960,
+      "step": 1297
+    },
+    {
+      "epoch": 0.37417123090227733,
+      "grad_norm": 3.4015052487127733,
+      "learning_rate": 0.0001,
+      "loss": 2.9454,
+      "num_input_tokens_seen": 191378048,
+      "step": 1298
+    },
+    {
+      "epoch": 0.37417123090227733,
+      "loss": 2.8113441467285156,
+      "loss_ce": 2.799163818359375,
+      "loss_xval": 0.01220703125,
+      "num_input_tokens_seen": 191378048,
+      "step": 1298
+    },
+    {
+      "epoch": 0.3744594984145287,
+      "grad_norm": 3.1268663171814013,
+      "learning_rate": 0.0001,
+      "loss": 2.7806,
+      "num_input_tokens_seen": 191550472,
+      "step": 1299
+    },
+    {
+      "epoch": 0.3744594984145287,
+      "loss": 2.998398780822754,
+      "loss_ce": 2.9863977432250977,
+      "loss_xval": 0.01202392578125,
+      "num_input_tokens_seen": 191550472,
+      "step": 1299
+    },
+    {
+      "epoch": 0.37474776592678005,
+      "grad_norm": 3.4915024725359314,
+      "learning_rate": 0.0001,
+      "loss": 2.6396,
+      "num_input_tokens_seen": 191685248,
+      "step": 1300
+    },
+    {
+      "epoch": 0.37474776592678005,
+      "loss": 2.528428554534912,
+      "loss_ce": 2.5081725120544434,
+      "loss_xval": 0.020263671875,
+      "num_input_tokens_seen": 191685248,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3750360334390314,
+      "grad_norm": 4.385977856328783,
+      "learning_rate": 0.0001,
+      "loss": 2.8341,
+      "num_input_tokens_seen": 191820368,
+      "step": 1301
+    },
+    {
+      "epoch": 0.3750360334390314,
+      "loss": 2.688603401184082,
+      "loss_ce": 2.677090644836426,
+      "loss_xval": 0.01153564453125,
+      "num_input_tokens_seen": 191820368,
+      "step": 1301
+    },
+    {
+      "epoch": 0.3753243009512828,
+      "grad_norm": 3.9234660899501947,
+      "learning_rate": 0.0001,
+      "loss": 2.6939,
+      "num_input_tokens_seen": 191992848,
+      "step": 1302
+    },
+    {
+      "epoch": 0.3753243009512828,
+      "loss": 2.939253330230713,
+      "loss_ce": 2.9202942848205566,
+      "loss_xval": 0.0189208984375,
+      "num_input_tokens_seen": 191992848,
+      "step": 1302
+    },
+    {
+      "epoch": 0.37561256846353414,
+      "grad_norm": 4.531139506615138,
+      "learning_rate": 0.0001,
+      "loss": 2.5161,
+      "num_input_tokens_seen": 192127608,
+      "step": 1303
+    },
+    {
+      "epoch": 0.37561256846353414,
+      "loss": 2.4107353687286377,
+      "loss_ce": 2.385352373123169,
+      "loss_xval": 0.025390625,
+      "num_input_tokens_seen": 192127608,
+      "step": 1303
+    },
+    {
+      "epoch": 0.37590083597578555,
+      "grad_norm": 4.966347398089012,
+      "learning_rate": 0.0001,
+      "loss": 2.7311,
+      "num_input_tokens_seen": 192262696,
+      "step": 1304
+    },
+    {
+      "epoch": 0.37590083597578555,
+      "loss": 2.579951763153076,
+      "loss_ce": 2.568812847137451,
+      "loss_xval": 0.0111083984375,
+      "num_input_tokens_seen": 192262696,
+      "step": 1304
+    },
+    {
+      "epoch": 0.3761891034880369,
+      "grad_norm": 4.887690213966296,
+      "learning_rate": 0.0001,
+      "loss": 2.5636,
+      "num_input_tokens_seen": 192435160,
+      "step": 1305
+    },
+    {
+      "epoch": 0.3761891034880369,
+      "loss": 2.807976007461548,
+      "loss_ce": 2.7880098819732666,
+      "loss_xval": 0.02001953125,
+      "num_input_tokens_seen": 192435160,
+      "step": 1305
+    },
+    {
+      "epoch": 0.3764773710002883,
+      "grad_norm": 5.596325586836034,
+      "learning_rate": 0.0001,
+      "loss": 2.3777,
+      "num_input_tokens_seen": 192569928,
+      "step": 1306
+    },
+    {
+      "epoch": 0.3764773710002883,
+      "loss": 2.276872158050537,
+      "loss_ce": 2.245103359222412,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 192569928,
+      "step": 1306
+    },
+    {
+      "epoch": 0.37676563851253964,
+      "grad_norm": 5.342816876559327,
+      "learning_rate": 0.0001,
+      "loss": 2.5818,
+      "num_input_tokens_seen": 192705032,
+      "step": 1307
+    },
+    {
+      "epoch": 0.37676563851253964,
+      "loss": 2.4396731853485107,
+      "loss_ce": 2.425520658493042,
+      "loss_xval": 0.01416015625,
+      "num_input_tokens_seen": 192705032,
+      "step": 1307
+    },
+    {
+      "epoch": 0.377053906024791,
+      "grad_norm": 5.187116600816184,
+      "learning_rate": 0.0001,
+      "loss": 2.4064,
+      "num_input_tokens_seen": 192877544,
+      "step": 1308
+    },
+    {
+      "epoch": 0.377053906024791,
+      "loss": 2.6360042095184326,
+      "loss_ce": 2.61273455619812,
+      "loss_xval": 0.0233154296875,
+      "num_input_tokens_seen": 192877544,
+      "step": 1308
+    },
+    {
+      "epoch": 0.37734217353704236,
+      "grad_norm": 5.392912569319807,
+      "learning_rate": 0.0001,
+      "loss": 2.2167,
+      "num_input_tokens_seen": 193012368,
+      "step": 1309
+    },
+    {
+      "epoch": 0.37734217353704236,
+      "loss": 2.115727663040161,
+      "loss_ce": 2.085911989212036,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 193012368,
+      "step": 1309
+    },
+    {
+      "epoch": 0.3776304410492937,
+      "grad_norm": 5.981915610875332,
+      "learning_rate": 0.0001,
+      "loss": 2.4108,
+      "num_input_tokens_seen": 193147368,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3776304410492937,
+      "loss": 2.273207664489746,
+      "loss_ce": 2.25192928314209,
+      "loss_xval": 0.021240234375,
+      "num_input_tokens_seen": 193147368,
+      "step": 1310
+    },
+    {
+      "epoch": 0.37791870856154514,
+      "grad_norm": 5.340813411172702,
+      "learning_rate": 0.0001,
+      "loss": 2.2597,
+      "num_input_tokens_seen": 193319824,
+      "step": 1311
+    },
+    {
+      "epoch": 0.37791870856154514,
+      "loss": 2.519282579421997,
+      "loss_ce": 2.4933578968048096,
+      "loss_xval": 0.02587890625,
+      "num_input_tokens_seen": 193319824,
+      "step": 1311
+    },
+    {
+      "epoch": 0.3782069760737965,
+      "grad_norm": 5.5534880340477635,
+      "learning_rate": 0.0001,
+      "loss": 2.0486,
+      "num_input_tokens_seen": 193454552,
+      "step": 1312
+    },
+    {
+      "epoch": 0.3782069760737965,
+      "loss": 1.9361722469329834,
+      "loss_ce": 1.902007818222046,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 193454552,
+      "step": 1312
+    },
+    {
+      "epoch": 0.37849524358604786,
+      "grad_norm": 5.382293657982696,
+      "learning_rate": 0.0001,
+      "loss": 2.2029,
+      "num_input_tokens_seen": 193589712,
+      "step": 1313
+    },
+    {
+      "epoch": 0.37849524358604786,
+      "loss": 2.060214042663574,
+      "loss_ce": 2.0438222885131836,
+      "loss_xval": 0.016357421875,
+      "num_input_tokens_seen": 193589712,
+      "step": 1313
+    },
+    {
+      "epoch": 0.3787835110982992,
+      "grad_norm": 5.398565780604443,
+      "learning_rate": 0.0001,
+      "loss": 2.0614,
+      "num_input_tokens_seen": 193762352,
+      "step": 1314
+    },
+    {
+      "epoch": 0.3787835110982992,
+      "loss": 2.323631763458252,
+      "loss_ce": 2.3039631843566895,
+      "loss_xval": 0.0196533203125,
+      "num_input_tokens_seen": 193762352,
+      "step": 1314
+    },
+    {
+      "epoch": 0.3790717786105506,
+      "grad_norm": 6.089319320163971,
+      "learning_rate": 0.0001,
+      "loss": 1.8249,
+      "num_input_tokens_seen": 193897064,
+      "step": 1315
+    },
+    {
+      "epoch": 0.3790717786105506,
+      "loss": 1.7183966636657715,
+      "loss_ce": 1.693814754486084,
+      "loss_xval": 0.0245361328125,
+      "num_input_tokens_seen": 193897064,
+      "step": 1315
+    },
+    {
+      "epoch": 0.37936004612280194,
+      "grad_norm": 6.00686223988536,
+      "learning_rate": 0.0001,
+      "loss": 1.9911,
+      "num_input_tokens_seen": 194032176,
+      "step": 1316
+    },
+    {
+      "epoch": 0.37936004612280194,
+      "loss": 1.8674993515014648,
+      "loss_ce": 1.8527288436889648,
+      "loss_xval": 0.0147705078125,
+      "num_input_tokens_seen": 194032176,
+      "step": 1316
+    },
+    {
+      "epoch": 0.3796483136350533,
+      "grad_norm": 6.384847108031611,
+      "learning_rate": 0.0001,
+      "loss": 1.7857,
+      "num_input_tokens_seen": 194204600,
+      "step": 1317
+    },
+    {
+      "epoch": 0.3796483136350533,
+      "loss": 2.01257586479187,
+      "loss_ce": 1.9936702251434326,
+      "loss_xval": 0.0189208984375,
+      "num_input_tokens_seen": 194204600,
+      "step": 1317
+    },
+    {
+      "epoch": 0.3799365811473047,
+      "grad_norm": 7.057178007836694,
+      "learning_rate": 0.0001,
+      "loss": 1.5645,
+      "num_input_tokens_seen": 194339368,
+      "step": 1318
+    },
+    {
+      "epoch": 0.3799365811473047,
+      "loss": 1.46513032913208,
+      "loss_ce": 1.43845796585083,
+      "loss_xval": 0.026611328125,
+      "num_input_tokens_seen": 194339368,
+      "step": 1318
+    },
+    {
+      "epoch": 0.3802248486595561,
+      "grad_norm": 7.503319455739542,
+      "learning_rate": 0.0001,
+      "loss": 1.6781,
+      "num_input_tokens_seen": 194474432,
+      "step": 1319
+    },
+    {
+      "epoch": 0.3802248486595561,
+      "loss": 1.5578892230987549,
+      "loss_ce": 1.5376484394073486,
+      "loss_xval": 0.020263671875,
+      "num_input_tokens_seen": 194474432,
+      "step": 1319
+    },
+    {
+      "epoch": 0.38051311617180744,
+      "grad_norm": 7.2656151061317304,
+      "learning_rate": 0.0001,
+      "loss": 1.5075,
+      "num_input_tokens_seen": 194647032,
+      "step": 1320
+    },
+    {
+      "epoch": 0.38051311617180744,
+      "loss": 1.741824984550476,
+      "loss_ce": 1.7225226163864136,
+      "loss_xval": 0.019287109375,
+      "num_input_tokens_seen": 194647032,
+      "step": 1320
+    },
+    {
+      "epoch": 0.3808013836840588,
+      "grad_norm": 7.579715480943731,
+      "learning_rate": 0.0001,
+      "loss": 1.2582,
+      "num_input_tokens_seen": 194781840,
+      "step": 1321
+    },
+    {
+      "epoch": 0.3808013836840588,
+      "loss": 1.1731231212615967,
+      "loss_ce": 1.1493346691131592,
+      "loss_xval": 0.0238037109375,
+      "num_input_tokens_seen": 194781840,
+      "step": 1321
+    },
+    {
+      "epoch": 0.38108965119631016,
+      "grad_norm": 7.840261357937563,
+      "learning_rate": 0.0001,
+      "loss": 1.3399,
+      "num_input_tokens_seen": 194916848,
+      "step": 1322
+    },
+    {
+      "epoch": 0.38108965119631016,
+      "loss": 1.1873873472213745,
+      "loss_ce": 1.1677416563034058,
+      "loss_xval": 0.0196533203125,
+      "num_input_tokens_seen": 194916848,
+      "step": 1322
+    },
+    {
+      "epoch": 0.3813779187085615,
+      "grad_norm": 8.101935906238767,
+      "learning_rate": 0.0001,
+      "loss": 1.1675,
+      "num_input_tokens_seen": 195089424,
+      "step": 1323
+    },
+    {
+      "epoch": 0.3813779187085615,
+      "loss": 1.3871322870254517,
+      "loss_ce": 1.367089867591858,
+      "loss_xval": 0.02001953125,
+      "num_input_tokens_seen": 195089424,
+      "step": 1323
+    },
+    {
+      "epoch": 0.38166618622081294,
+      "grad_norm": 10.133729748812186,
+      "learning_rate": 0.0001,
+      "loss": 0.9085,
+      "num_input_tokens_seen": 195224168,
+      "step": 1324
+    },
+    {
+      "epoch": 0.38166618622081294,
+      "loss": 0.8485641479492188,
+      "loss_ce": 0.8170852661132812,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 195224168,
+      "step": 1324
+    },
+    {
+      "epoch": 0.3819544537330643,
+      "grad_norm": 9.184673489968239,
+      "learning_rate": 0.0001,
+      "loss": 0.9749,
+      "num_input_tokens_seen": 195359168,
+      "step": 1325
+    },
+    {
+      "epoch": 0.3819544537330643,
+      "loss": 0.8387224674224854,
+      "loss_ce": 0.8197863101959229,
+      "loss_xval": 0.0189208984375,
+      "num_input_tokens_seen": 195359168,
+      "step": 1325
+    },
+    {
+      "epoch": 0.38224272124531566,
+      "grad_norm": 8.462154836119543,
+      "learning_rate": 0.0001,
+      "loss": 0.8089,
+      "num_input_tokens_seen": 195531720,
+      "step": 1326
+    },
+    {
+      "epoch": 0.38224272124531566,
+      "loss": 1.000867247581482,
+      "loss_ce": 0.9825414419174194,
+      "loss_xval": 0.018310546875,
+      "num_input_tokens_seen": 195531720,
+      "step": 1326
+    },
+    {
+      "epoch": 0.382530988757567,
+      "grad_norm": 8.262590257922607,
+      "learning_rate": 0.0001,
+      "loss": 0.5586,
+      "num_input_tokens_seen": 195666576,
+      "step": 1327
+    },
+    {
+      "epoch": 0.382530988757567,
+      "loss": 0.5303316116333008,
+      "loss_ce": 0.507390022277832,
+      "loss_xval": 0.02294921875,
+      "num_input_tokens_seen": 195666576,
+      "step": 1327
+    },
+    {
+      "epoch": 0.3828192562698184,
+      "grad_norm": 8.508081307007473,
+      "learning_rate": 0.0001,
+      "loss": 0.6148,
+      "num_input_tokens_seen": 195801680,
+      "step": 1328
+    },
+    {
+      "epoch": 0.3828192562698184,
+      "loss": 0.4951069951057434,
+      "loss_ce": 0.4788067936897278,
+      "loss_xval": 0.016357421875,
+      "num_input_tokens_seen": 195801680,
+      "step": 1328
+    },
+    {
+      "epoch": 0.38310752378206975,
+      "grad_norm": 7.359277842901975,
+      "learning_rate": 0.0001,
+      "loss": 0.4927,
+      "num_input_tokens_seen": 195974144,
+      "step": 1329
+    },
+    {
+      "epoch": 0.38310752378206975,
+      "loss": 0.6365176439285278,
+      "loss_ce": 0.6132174730300903,
+      "loss_xval": 0.0233154296875,
+      "num_input_tokens_seen": 195974144,
+      "step": 1329
+    },
+    {
+      "epoch": 0.3833957912943211,
+      "grad_norm": 5.939757783528577,
+      "learning_rate": 0.0001,
+      "loss": 0.2844,
+      "num_input_tokens_seen": 196108816,
+      "step": 1330
+    },
+    {
+      "epoch": 0.3833957912943211,
+      "loss": 0.29377344250679016,
+      "loss_ce": 0.26371362805366516,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 196108816,
+      "step": 1330
+    },
+    {
+      "epoch": 0.3836840588065725,
+      "grad_norm": 6.159986104423088,
+      "learning_rate": 0.0001,
+      "loss": 0.3409,
+      "num_input_tokens_seen": 196244112,
+      "step": 1331
+    },
+    {
+      "epoch": 0.3836840588065725,
+      "loss": 0.23033326864242554,
+      "loss_ce": 0.20579713582992554,
+      "loss_xval": 0.0245361328125,
+      "num_input_tokens_seen": 196244112,
+      "step": 1331
+    },
+    {
+      "epoch": 0.3839723263188239,
+      "grad_norm": 4.77282512062471,
+      "learning_rate": 0.0001,
+      "loss": 0.2807,
+      "num_input_tokens_seen": 196416592,
+      "step": 1332
+    },
+    {
+      "epoch": 0.3839723263188239,
+      "loss": 0.35939767956733704,
+      "loss_ce": 0.3379667103290558,
+      "loss_xval": 0.021484375,
+      "num_input_tokens_seen": 196416592,
+      "step": 1332
+    },
+    {
+      "epoch": 0.38426059383107525,
+      "grad_norm": 4.051100825473909,
+      "learning_rate": 0.0001,
+      "loss": 0.1492,
+      "num_input_tokens_seen": 196551408,
+      "step": 1333
+    },
+    {
+      "epoch": 0.38426059383107525,
+      "loss": 0.1672614961862564,
+      "loss_ce": 0.14396895468235016,
+      "loss_xval": 0.0233154296875,
+      "num_input_tokens_seen": 196551408,
+      "step": 1333
+    },
+    {
+      "epoch": 0.3845488613433266,
+      "grad_norm": 4.06293877344607,
+      "learning_rate": 0.0001,
+      "loss": 0.2038,
+      "num_input_tokens_seen": 196686448,
+      "step": 1334
+    },
+    {
+      "epoch": 0.3845488613433266,
+      "loss": 0.1264897584915161,
+      "loss_ce": 0.10885824263095856,
+      "loss_xval": 0.017578125,
+      "num_input_tokens_seen": 196686448,
+      "step": 1334
+    },
+    {
+      "epoch": 0.38483712885557797,
+      "grad_norm": 3.037738848199535,
+      "learning_rate": 0.0001,
+      "loss": 0.1679,
+      "num_input_tokens_seen": 196858936,
+      "step": 1335
+    },
+    {
+      "epoch": 0.38483712885557797,
+      "loss": 0.2194724977016449,
+      "loss_ce": 0.19611892104148865,
+      "loss_xval": 0.0233154296875,
+      "num_input_tokens_seen": 196858936,
+      "step": 1335
+    },
+    {
+      "epoch": 0.38512539636782933,
+      "grad_norm": 3.9874713827981862,
+      "learning_rate": 0.0001,
+      "loss": 0.0949,
+      "num_input_tokens_seen": 196993680,
+      "step": 1336
+    },
+    {
+      "epoch": 0.38512539636782933,
+      "loss": 0.09953860938549042,
+      "loss_ce": 0.07335452735424042,
+      "loss_xval": 0.026123046875,
+      "num_input_tokens_seen": 196993680,
+      "step": 1336
+    },
+    {
+      "epoch": 0.3854136638800807,
+      "grad_norm": 3.842858227911105,
+      "learning_rate": 0.0001,
+      "loss": 0.1263,
+      "num_input_tokens_seen": 197128720,
+      "step": 1337
+    },
+    {
+      "epoch": 0.3854136638800807,
+      "loss": 0.07819326221942902,
+      "loss_ce": 0.061484888195991516,
+      "loss_xval": 0.0167236328125,
+      "num_input_tokens_seen": 197128720,
+      "step": 1337
+    },
+    {
+      "epoch": 0.3857019313923321,
+      "grad_norm": 3.7661320103755913,
+      "learning_rate": 0.0001,
+      "loss": 0.1149,
+      "num_input_tokens_seen": 197301192,
+      "step": 1338
+    },
+    {
+      "epoch": 0.3857019313923321,
+      "loss": 0.1548667848110199,
+      "loss_ce": 0.1375785768032074,
+      "loss_xval": 0.017333984375,
+      "num_input_tokens_seen": 197301192,
+      "step": 1338
+    },
+    {
+      "epoch": 0.38599019890458347,
+      "grad_norm": 2.0935235814016324,
+      "learning_rate": 0.0001,
+      "loss": 0.063,
+      "num_input_tokens_seen": 197436008,
+      "step": 1339
+    },
+    {
+      "epoch": 0.38599019890458347,
+      "loss": 0.06575188040733337,
+      "loss_ce": 0.04433616250753403,
+      "loss_xval": 0.0213623046875,
+      "num_input_tokens_seen": 197436008,
+      "step": 1339
+    },
+    {
+      "epoch": 0.38627846641683483,
+      "grad_norm": 9.864120567956032,
+      "learning_rate": 0.0001,
+      "loss": 0.0988,
+      "num_input_tokens_seen": 197571040,
+      "step": 1340
+    },
+    {
+      "epoch": 0.38627846641683483,
+      "loss": 0.04486007243394852,
+      "loss_ce": 0.030917353928089142,
+      "loss_xval": 0.013916015625,
+      "num_input_tokens_seen": 197571040,
+      "step": 1340
+    },
+    {
+      "epoch": 0.3865667339290862,
+      "grad_norm": 2.030815217945069,
+      "learning_rate": 0.0001,
+      "loss": 0.0922,
+      "num_input_tokens_seen": 197743672,
+      "step": 1341
+    },
+    {
+      "epoch": 0.3865667339290862,
+      "loss": 0.1213720515370369,
+      "loss_ce": 0.1082647442817688,
+      "loss_xval": 0.01312255859375,
+      "num_input_tokens_seen": 197743672,
+      "step": 1341
+    },
+    {
+      "epoch": 0.38685500144133755,
+      "grad_norm": 1.9671558855419173,
+      "learning_rate": 0.0001,
+      "loss": 0.052,
+      "num_input_tokens_seen": 197878456,
+      "step": 1342
+    },
+    {
+      "epoch": 0.38685500144133755,
+      "loss": 0.05662340298295021,
+      "loss_ce": 0.03942674770951271,
+      "loss_xval": 0.0172119140625,
+      "num_input_tokens_seen": 197878456,
+      "step": 1342
+    },
+    {
+      "epoch": 0.3871432689535889,
+      "grad_norm": 2.4283101718006312,
+      "learning_rate": 0.0001,
+      "loss": 0.0894,
+      "num_input_tokens_seen": 198013456,
+      "step": 1343
+    },
+    {
+      "epoch": 0.3871432689535889,
+      "loss": 0.04897967353463173,
+      "loss_ce": 0.03391161933541298,
+      "loss_xval": 0.01507568359375,
+      "num_input_tokens_seen": 198013456,
+      "step": 1343
+    },
+    {
+      "epoch": 0.3874315364658403,
+      "grad_norm": 2.7356786477473936,
+      "learning_rate": 0.0001,
+      "loss": 0.0838,
+      "num_input_tokens_seen": 198186000,
+      "step": 1344
+    },
+    {
+      "epoch": 0.3874315364658403,
+      "loss": 0.10481792688369751,
+      "loss_ce": 0.08823162317276001,
+      "loss_xval": 0.0166015625,
+      "num_input_tokens_seen": 198186000,
+      "step": 1344
+    },
+    {
+      "epoch": 0.3877198039780917,
+      "grad_norm": 1.3907189763308183,
+      "learning_rate": 0.0001,
+      "loss": 0.0507,
+      "num_input_tokens_seen": 198320848,
+      "step": 1345
+    },
+    {
+      "epoch": 0.3877198039780917,
+      "loss": 0.05430437996983528,
+      "loss_ce": 0.03217913582921028,
+      "loss_xval": 0.0220947265625,
+      "num_input_tokens_seen": 198320848,
+      "step": 1345
+    },
+    {
+      "epoch": 0.38800807149034305,
+      "grad_norm": 4.385028674819962,
+      "learning_rate": 0.0001,
+      "loss": 0.0654,
+      "num_input_tokens_seen": 198456080,
+      "step": 1346
+    },
+    {
+      "epoch": 0.38800807149034305,
+      "loss": 0.037943094968795776,
+      "loss_ce": 0.020853251218795776,
+      "loss_xval": 0.01708984375,
+      "num_input_tokens_seen": 198456080,
+      "step": 1346
+    },
+    {
+      "epoch": 0.3882963390025944,
+      "grad_norm": 2.0849586519173657,
+      "learning_rate": 0.0001,
+      "loss": 0.0545,
+      "num_input_tokens_seen": 198628672,
+      "step": 1347
+    },
+    {
+      "epoch": 0.3882963390025944,
+      "loss": 0.06738258898258209,
+      "loss_ce": 0.051727067679166794,
+      "loss_xval": 0.015625,
+      "num_input_tokens_seen": 198628672,
+      "step": 1347
+    },
+    {
+      "epoch": 0.3885846065148458,
+      "grad_norm": 1.3551925654534938,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "num_input_tokens_seen": 198763384,
+      "step": 1348
+    },
+    {
+      "epoch": 0.3885846065148458,
+      "loss": 0.045364461839199066,
+      "loss_ce": 0.029098594561219215,
+      "loss_xval": 0.0162353515625,
+      "num_input_tokens_seen": 198763384,
+      "step": 1348
+    },
+    {
+      "epoch": 0.38887287402709714,
+      "grad_norm": 1.2264368422957106,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "num_input_tokens_seen": 198898480,
+      "step": 1349
+    },
+    {
+      "epoch": 0.38887287402709714,
+      "loss": 0.025431422516703606,
+      "loss_ce": 0.009157923981547356,
+      "loss_xval": 0.0162353515625,
+      "num_input_tokens_seen": 198898480,
+      "step": 1349
+    },
+    {
+      "epoch": 0.3891611415393485,
+      "grad_norm": 3.0852120648983212,
+      "learning_rate": 0.0001,
+      "loss": 0.049,
+      "num_input_tokens_seen": 199070928,
+      "step": 1350
+    },
+    {
+      "epoch": 0.3891611415393485,
+      "loss": 0.04758167266845703,
+      "loss_ce": 0.03284931182861328,
+      "loss_xval": 0.01470947265625,
+      "num_input_tokens_seen": 199070928,
+      "step": 1350
+    },
+    {
+      "epoch": 0.38944940905159986,
+      "grad_norm": 4.074322664880206,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "num_input_tokens_seen": 199205680,
+      "step": 1351
+    },
+    {
+      "epoch": 0.38944940905159986,
+      "loss": 0.043620556592941284,
+      "loss_ce": 0.027080029249191284,
+      "loss_xval": 0.0166015625,
+      "num_input_tokens_seen": 199205680,
+      "step": 1351
+    },
+    {
+      "epoch": 0.3897376765638513,
+      "grad_norm": 1.149605580995087,
+      "learning_rate": 0.0001,
+      "loss": 0.0348,
+      "num_input_tokens_seen": 199340672,
+      "step": 1352
+    },
+    {
+      "epoch": 0.3897376765638513,
+      "loss": 0.021780086681246758,
+      "loss_ce": 0.008981777355074883,
+      "loss_xval": 0.0128173828125,
+      "num_input_tokens_seen": 199340672,
+      "step": 1352
+    },
+    {
+      "epoch": 0.39002594407610264,
+      "grad_norm": 1.139692322477526,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "num_input_tokens_seen": 199513160,
+      "step": 1353
+    },
+    {
+      "epoch": 0.39002594407610264,
+      "loss": 0.041254542768001556,
+      "loss_ce": 0.03088238276541233,
+      "loss_xval": 0.0103759765625,
+      "num_input_tokens_seen": 199513160,
+      "step": 1353
+    },
+    {
+      "epoch": 0.390314211588354,
+      "grad_norm": 0.806504861954892,
+      "learning_rate": 0.0001,
+      "loss": 0.0278,
+      "num_input_tokens_seen": 199647992,
+      "step": 1354
+    },
+    {
+      "epoch": 0.390314211588354,
+      "loss": 0.034722212702035904,
+      "loss_ce": 0.023095015436410904,
+      "loss_xval": 0.0115966796875,
+      "num_input_tokens_seen": 199647992,
+      "step": 1354
+    },
+    {
+      "epoch": 0.39060247910060536,
+      "grad_norm": 2.061867339897939,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "num_input_tokens_seen": 199782968,
+      "step": 1355
+    },
+    {
+      "epoch": 0.39060247910060536,
+      "loss": 0.01884021796286106,
+      "loss_ce": 0.009284400381147861,
+      "loss_xval": 0.00958251953125,
+      "num_input_tokens_seen": 199782968,
+      "step": 1355
+    },
+    {
+      "epoch": 0.3908907466128567,
+      "grad_norm": 1.4719134776992004,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "num_input_tokens_seen": 199955464,
+      "step": 1356
+    },
+    {
+      "epoch": 0.3908907466128567,
+      "loss": 0.044432446360588074,
+      "loss_ce": 0.0333278626203537,
+      "loss_xval": 0.0111083984375,
+      "num_input_tokens_seen": 199955464,
+      "step": 1356
+    },
+    {
+      "epoch": 0.3911790141251081,
+      "grad_norm": 1.8078027247986106,
+      "learning_rate": 0.0001,
+      "loss": 0.0257,
+      "num_input_tokens_seen": 200090192,
+      "step": 1357
+    },
+    {
+      "epoch": 0.3911790141251081,
+      "loss": 0.031802982091903687,
+      "loss_ce": 0.019832465797662735,
+      "loss_xval": 0.011962890625,
+      "num_input_tokens_seen": 200090192,
+      "step": 1357
+    },
+    {
+      "epoch": 0.39146728163735944,
+      "grad_norm": 1.1044953565112376,
+      "learning_rate": 0.0001,
+      "loss": 0.0228,
+      "num_input_tokens_seen": 200225288,
+      "step": 1358
+    },
+    {
+      "epoch": 0.39146728163735944,
+      "loss": 0.016087032854557037,
+      "loss_ce": 0.009964443743228912,
+      "loss_xval": 0.006134033203125,
+      "num_input_tokens_seen": 200225288,
+      "step": 1358
+    },
+    {
+      "epoch": 0.39175554914961086,
+      "grad_norm": 0.9945927909285693,
+      "learning_rate": 0.0001,
+      "loss": 0.0254,
+      "num_input_tokens_seen": 200397888,
+      "step": 1359
+    },
+    {
+      "epoch": 0.39175554914961086,
+      "loss": 0.02666654624044895,
+      "loss_ce": 0.018461132422089577,
+      "loss_xval": 0.0081787109375,
+      "num_input_tokens_seen": 200397888,
+      "step": 1359
+    },
+    {
+      "epoch": 0.3920438166618622,
+      "grad_norm": 0.494338140152747,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "num_input_tokens_seen": 200532680,
+      "step": 1360
+    },
+    {
+      "epoch": 0.3920438166618622,
+      "loss": 0.02553568407893181,
+      "loss_ce": 0.013740640133619308,
+      "loss_xval": 0.01177978515625,
+      "num_input_tokens_seen": 200532680,
+      "step": 1360
+    },
+    {
+      "epoch": 0.3923320841741136,
+      "grad_norm": 1.3856157113105607,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "num_input_tokens_seen": 200667592,
+      "step": 1361
+    },
+    {
+      "epoch": 0.3923320841741136,
+      "loss": 0.011884558014571667,
+      "loss_ce": 0.0035341857001185417,
+      "loss_xval": 0.00836181640625,
+      "num_input_tokens_seen": 200667592,
+      "step": 1361
+    },
+    {
+      "epoch": 0.39262035168636494,
+      "grad_norm": 0.7369679296640576,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "num_input_tokens_seen": 200839960,
+      "step": 1362
+    },
+    {
+      "epoch": 0.39262035168636494,
+      "loss": 0.02030220441520214,
+      "loss_ce": 0.01264992170035839,
+      "loss_xval": 0.007659912109375,
+      "num_input_tokens_seen": 200839960,
+      "step": 1362
+    },
+    {
+      "epoch": 0.3929086191986163,
+      "grad_norm": 1.2784878807936497,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "num_input_tokens_seen": 200974784,
+      "step": 1363
+    },
+    {
+      "epoch": 0.3929086191986163,
+      "loss": 0.027085553854703903,
+      "loss_ce": 0.019082318991422653,
+      "loss_xval": 0.00799560546875,
+      "num_input_tokens_seen": 200974784,
+      "step": 1363
+    },
+    {
+      "epoch": 0.39319688671086767,
+      "grad_norm": 0.973016613201439,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "num_input_tokens_seen": 201109784,
+      "step": 1364
+    },
+    {
+      "epoch": 0.39319688671086767,
+      "loss": 0.008179396390914917,
+      "loss_ce": 0.0029017634224146605,
+      "loss_xval": 0.005279541015625,
+      "num_input_tokens_seen": 201109784,
+      "step": 1364
+    },
+    {
+      "epoch": 0.393485154223119,
+      "grad_norm": 0.7713254196549865,
+      "learning_rate": 0.0001,
+      "loss": 0.0226,
+      "num_input_tokens_seen": 201282384,
+      "step": 1365
+    },
+    {
+      "epoch": 0.393485154223119,
+      "loss": 0.015684355050325394,
+      "loss_ce": 0.010984647087752819,
+      "loss_xval": 0.00469970703125,
+      "num_input_tokens_seen": 201282384,
+      "step": 1365
+    },
+    {
+      "epoch": 0.39377342173537044,
+      "grad_norm": 0.9308851890759041,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "num_input_tokens_seen": 201417136,
+      "step": 1366
+    },
+    {
+      "epoch": 0.39377342173537044,
+      "loss": 0.023319490253925323,
+      "loss_ce": 0.015459306538105011,
+      "loss_xval": 0.00787353515625,
+      "num_input_tokens_seen": 201417136,
+      "step": 1366
+    },
+    {
+      "epoch": 0.3940616892476218,
+      "grad_norm": 0.6404491910207054,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "num_input_tokens_seen": 201552240,
+      "step": 1367
+    },
+    {
+      "epoch": 0.3940616892476218,
+      "loss": 0.008579766377806664,
+      "loss_ce": 0.0033879633992910385,
+      "loss_xval": 0.00518798828125,
+      "num_input_tokens_seen": 201552240,
+      "step": 1367
+    },
+    {
+      "epoch": 0.39434995675987317,
+      "grad_norm": 0.6986686066295519,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "num_input_tokens_seen": 201724872,
+      "step": 1368
+    },
+    {
+      "epoch": 0.39434995675987317,
+      "loss": 0.02358327805995941,
+      "loss_ce": 0.018536433577537537,
+      "loss_xval": 0.005035400390625,
+      "num_input_tokens_seen": 201724872,
+      "step": 1368
+    },
+    {
+      "epoch": 0.3946382242721245,
+      "grad_norm": 0.8428508345107775,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "num_input_tokens_seen": 201859616,
+      "step": 1369
+    },
+    {
+      "epoch": 0.3946382242721245,
+      "loss": 0.017313163727521896,
+      "loss_ce": 0.01113144587725401,
+      "loss_xval": 0.006195068359375,
+      "num_input_tokens_seen": 201859616,
+      "step": 1369
+    },
+    {
+      "epoch": 0.3949264917843759,
+      "grad_norm": 0.8350447749155997,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "num_input_tokens_seen": 201994760,
+      "step": 1370
+    },
+    {
+      "epoch": 0.3949264917843759,
+      "loss": 0.007084501441568136,
+      "loss_ce": 0.0034090406261384487,
+      "loss_xval": 0.0036773681640625,
+      "num_input_tokens_seen": 201994760,
+      "step": 1370
+    },
+    {
+      "epoch": 0.39521475929662725,
+      "grad_norm": 0.9544725638554615,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "num_input_tokens_seen": 202167240,
+      "step": 1371
+    },
+    {
+      "epoch": 0.39521475929662725,
+      "loss": 0.014551658183336258,
+      "loss_ce": 0.011322516947984695,
+      "loss_xval": 0.00323486328125,
+      "num_input_tokens_seen": 202167240,
+      "step": 1371
+    },
+    {
+      "epoch": 0.39550302680887867,
+      "grad_norm": 0.6628523590881815,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "num_input_tokens_seen": 202302224,
+      "step": 1372
+    },
+    {
+      "epoch": 0.39550302680887867,
+      "loss": 0.01799912378191948,
+      "loss_ce": 0.011701060459017754,
+      "loss_xval": 0.00628662109375,
+      "num_input_tokens_seen": 202302224,
+      "step": 1372
+    },
+    {
+      "epoch": 0.39579129432113,
+      "grad_norm": 0.46876947007397934,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "num_input_tokens_seen": 202437312,
+      "step": 1373
+    },
+    {
+      "epoch": 0.39579129432113,
+      "loss": 0.009276005439460278,
+      "loss_ce": 0.00552234286442399,
+      "loss_xval": 0.003753662109375,
+      "num_input_tokens_seen": 202437312,
+      "step": 1373
+    },
+    {
+      "epoch": 0.3960795618333814,
+      "grad_norm": 1.8683003497946353,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "num_input_tokens_seen": 202609792,
+      "step": 1374
+    },
+    {
+      "epoch": 0.3960795618333814,
+      "loss": 0.01582440733909607,
+      "loss_ce": 0.011563390493392944,
+      "loss_xval": 0.0042724609375,
+      "num_input_tokens_seen": 202609792,
+      "step": 1374
+    },
+    {
+      "epoch": 0.39636782934563275,
+      "grad_norm": 0.2811121747222105,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "num_input_tokens_seen": 202744600,
+      "step": 1375
+    },
+    {
+      "epoch": 0.39636782934563275,
+      "loss": 0.01710539683699608,
+      "loss_ce": 0.012304600328207016,
+      "loss_xval": 0.004791259765625,
+      "num_input_tokens_seen": 202744600,
+      "step": 1375
+    },
+    {
+      "epoch": 0.3966560968578841,
+      "grad_norm": 1.500880723672373,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "num_input_tokens_seen": 202879656,
+      "step": 1376
+    },
+    {
+      "epoch": 0.3966560968578841,
+      "loss": 0.004621902015060186,
+      "loss_ce": 0.002170005114749074,
+      "loss_xval": 0.0024566650390625,
+      "num_input_tokens_seen": 202879656,
+      "step": 1376
+    },
+    {
+      "epoch": 0.39694436437013547,
+      "grad_norm": 1.098144830738785,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "num_input_tokens_seen": 203052192,
+      "step": 1377
+    },
+    {
+      "epoch": 0.39694436437013547,
+      "loss": 0.01708078756928444,
+      "loss_ce": 0.012887481600046158,
+      "loss_xval": 0.004180908203125,
+      "num_input_tokens_seen": 203052192,
+      "step": 1377
+    },
+    {
+      "epoch": 0.39723263188238683,
+      "grad_norm": 0.46400715903117085,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "num_input_tokens_seen": 203186880,
+      "step": 1378
+    },
+    {
+      "epoch": 0.39723263188238683,
+      "loss": 0.017731178551912308,
+      "loss_ce": 0.013500679284334183,
+      "loss_xval": 0.004241943359375,
+      "num_input_tokens_seen": 203186880,
+      "step": 1378
+    },
+    {
+      "epoch": 0.39752089939463825,
+      "grad_norm": 1.944436649986528,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "num_input_tokens_seen": 203321928,
+      "step": 1379
+    },
+    {
+      "epoch": 0.39752089939463825,
+      "loss": 0.004524157382547855,
+      "loss_ce": 0.002213404979556799,
+      "loss_xval": 0.0023040771484375,
+      "num_input_tokens_seen": 203321928,
+      "step": 1379
+    },
+    {
+      "epoch": 0.3978091669068896,
+      "grad_norm": 1.8472827415052435,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "num_input_tokens_seen": 203494472,
+      "step": 1380
+    },
+    {
+      "epoch": 0.3978091669068896,
+      "loss": 0.00947756040841341,
+      "loss_ce": 0.006767218001186848,
+      "loss_xval": 0.002716064453125,
+      "num_input_tokens_seen": 203494472,
+      "step": 1380
+    },
+    {
+      "epoch": 0.39809743441914097,
+      "grad_norm": 1.0137558806292657,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "num_input_tokens_seen": 203629256,
+      "step": 1381
+    },
+    {
+      "epoch": 0.39809743441914097,
+      "loss": 0.014530917629599571,
+      "loss_ce": 0.010992785915732384,
+      "loss_xval": 0.0035400390625,
+      "num_input_tokens_seen": 203629256,
+      "step": 1381
+    },
+    {
+      "epoch": 0.39838570193139233,
+      "grad_norm": 1.4902002985671599,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "num_input_tokens_seen": 203764232,
+      "step": 1382
+    },
+    {
+      "epoch": 0.39838570193139233,
+      "loss": 0.005032642744481564,
+      "loss_ce": 0.002303226850926876,
+      "loss_xval": 0.0027313232421875,
+      "num_input_tokens_seen": 203764232,
+      "step": 1382
+    },
+    {
+      "epoch": 0.3986739694436437,
+      "grad_norm": 1.9965416544832317,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "num_input_tokens_seen": 203936776,
+      "step": 1383
+    },
+    {
+      "epoch": 0.3986739694436437,
+      "loss": 0.010603894479572773,
+      "loss_ce": 0.00803088117390871,
+      "loss_xval": 0.0025787353515625,
+      "num_input_tokens_seen": 203936776,
+      "step": 1383
+    },
+    {
+      "epoch": 0.39896223695589506,
+      "grad_norm": 0.7761101964481655,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "num_input_tokens_seen": 204071632,
+      "step": 1384
+    },
+    {
+      "epoch": 0.39896223695589506,
+      "loss": 0.016848795115947723,
+      "loss_ce": 0.013341179117560387,
+      "loss_xval": 0.003509521484375,
+      "num_input_tokens_seen": 204071632,
+      "step": 1384
+    },
+    {
+      "epoch": 0.3992505044681464,
+      "grad_norm": 1.9679208914351352,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "num_input_tokens_seen": 204206592,
+      "step": 1385
+    },
+    {
+      "epoch": 0.3992505044681464,
+      "loss": 0.004215900786221027,
+      "loss_ce": 0.001496975775808096,
+      "loss_xval": 0.002716064453125,
+      "num_input_tokens_seen": 204206592,
+      "step": 1385
+    },
+    {
+      "epoch": 0.39953877198039783,
+      "grad_norm": 1.9443220159955805,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "num_input_tokens_seen": 204378992,
+      "step": 1386
+    },
+    {
+      "epoch": 0.39953877198039783,
+      "loss": 0.007329082116484642,
+      "loss_ce": 0.0047522541135549545,
+      "loss_xval": 0.0025787353515625,
+      "num_input_tokens_seen": 204378992,
+      "step": 1386
+    },
+    {
+      "epoch": 0.3998270394926492,
+      "grad_norm": 0.9898315702902115,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "num_input_tokens_seen": 204513824,
+      "step": 1387
+    },
+    {
+      "epoch": 0.3998270394926492,
+      "loss": 0.016465310007333755,
+      "loss_ce": 0.013494614511728287,
+      "loss_xval": 0.0029754638671875,
+      "num_input_tokens_seen": 204513824,
+      "step": 1387
+    },
+    {
+      "epoch": 0.40011530700490056,
+      "grad_norm": 1.7837581013050972,
+      "learning_rate": 0.0001,
+      "loss": 0.0045,
+      "num_input_tokens_seen": 204648864,
+      "step": 1388
+    },
+    {
+      "epoch": 0.40011530700490056,
+      "loss": 0.0035922147799283266,
+      "loss_ce": 0.001431188895367086,
+      "loss_xval": 0.002166748046875,
+      "num_input_tokens_seen": 204648864,
+      "step": 1388
+    },
+    {
+      "epoch": 0.4004035745171519,
+      "grad_norm": 1.231230547855382,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "num_input_tokens_seen": 204821408,
+      "step": 1389
+    },
+    {
+      "epoch": 0.4004035745171519,
+      "loss": 0.006246610544621944,
+      "loss_ce": 0.004241033457219601,
+      "loss_xval": 0.0019989013671875,
+      "num_input_tokens_seen": 204821408,
+      "step": 1389
+    },
+    {
+      "epoch": 0.4006918420294033,
+      "grad_norm": 0.45237187857054606,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "num_input_tokens_seen": 204956256,
+      "step": 1390
+    },
+    {
+      "epoch": 0.4006918420294033,
+      "loss": 0.014893912710249424,
+      "loss_ce": 0.011824988760054111,
+      "loss_xval": 0.0030670166015625,
+      "num_input_tokens_seen": 204956256,
+      "step": 1390
+    },
+    {
+      "epoch": 0.40098010954165464,
+      "grad_norm": 2.2686371794323406,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "num_input_tokens_seen": 205091240,
+      "step": 1391
+    },
+    {
+      "epoch": 0.40098010954165464,
+      "loss": 0.008735474199056625,
+      "loss_ce": 0.006970223039388657,
+      "loss_xval": 0.00176239013671875,
+      "num_input_tokens_seen": 205091240,
+      "step": 1391
+    },
+    {
+      "epoch": 0.401268377053906,
+      "grad_norm": 1.7761112950576003,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "num_input_tokens_seen": 205263688,
+      "step": 1392
+    },
+    {
+      "epoch": 0.401268377053906,
+      "loss": 0.00662598293274641,
+      "loss_ce": 0.004417273215949535,
+      "loss_xval": 0.0022125244140625,
+      "num_input_tokens_seen": 205263688,
+      "step": 1392
+    },
+    {
+      "epoch": 0.4015566445661574,
+      "grad_norm": 0.8047908874304497,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "num_input_tokens_seen": 205398456,
+      "step": 1393
+    },
+    {
+      "epoch": 0.4015566445661574,
+      "loss": 0.015318496152758598,
+      "loss_ce": 0.01345787849277258,
+      "loss_xval": 0.001861572265625,
+      "num_input_tokens_seen": 205398456,
+      "step": 1393
+    },
+    {
+      "epoch": 0.4018449120784088,
+      "grad_norm": 2.1236601899090837,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "num_input_tokens_seen": 205533424,
+      "step": 1394
+    },
+    {
+      "epoch": 0.4018449120784088,
+      "loss": 0.0035114334896206856,
+      "loss_ce": 0.0014324233634397388,
+      "loss_xval": 0.0020751953125,
+      "num_input_tokens_seen": 205533424,
+      "step": 1394
+    },
+    {
+      "epoch": 0.40213317959066014,
+      "grad_norm": 1.465889493088798,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "num_input_tokens_seen": 205705968,
+      "step": 1395
+    },
+    {
+      "epoch": 0.40213317959066014,
+      "loss": 0.007632310502231121,
+      "loss_ce": 0.006197507493197918,
+      "loss_xval": 0.001434326171875,
+      "num_input_tokens_seen": 205705968,
+      "step": 1395
+    },
+    {
+      "epoch": 0.4024214471029115,
+      "grad_norm": 1.1457608913972734,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "num_input_tokens_seen": 205840784,
+      "step": 1396
+    },
+    {
+      "epoch": 0.4024214471029115,
+      "loss": 0.013797684572637081,
+      "loss_ce": 0.011010094545781612,
+      "loss_xval": 0.0027923583984375,
+      "num_input_tokens_seen": 205840784,
+      "step": 1396
+    },
+    {
+      "epoch": 0.40270971461516286,
+      "grad_norm": 2.24633480417486,
+      "learning_rate": 0.0001,
+      "loss": 0.0047,
+      "num_input_tokens_seen": 205975904,
+      "step": 1397
+    },
+    {
+      "epoch": 0.40270971461516286,
+      "loss": 0.0036817528307437897,
+      "loss_ce": 0.0019260382978245616,
+      "loss_xval": 0.0017547607421875,
+      "num_input_tokens_seen": 205975904,
+      "step": 1397
+    },
+    {
+      "epoch": 0.4029979821274142,
+      "grad_norm": 0.4192766702455911,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "num_input_tokens_seen": 206148408,
+      "step": 1398
+    },
+    {
+      "epoch": 0.4029979821274142,
+      "loss": 0.005562697537243366,
+      "loss_ce": 0.004066859371960163,
+      "loss_xval": 0.001495361328125,
+      "num_input_tokens_seen": 206148408,
+      "step": 1398
+    },
+    {
+      "epoch": 0.4032862496396656,
+      "grad_norm": 2.0131614169700516,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "num_input_tokens_seen": 206283240,
+      "step": 1399
+    },
+    {
+      "epoch": 0.4032862496396656,
+      "loss": 0.018222136422991753,
+      "loss_ce": 0.015307707712054253,
+      "loss_xval": 0.0029144287109375,
+      "num_input_tokens_seen": 206283240,
+      "step": 1399
+    },
+    {
+      "epoch": 0.403574517151917,
+      "grad_norm": 0.8860071239442986,
+      "learning_rate": 0.0001,
+      "loss": 0.0041,
+      "num_input_tokens_seen": 206418328,
+      "step": 1400
+    },
+    {
+      "epoch": 0.403574517151917,
+      "loss": 0.00250605377368629,
+      "loss_ce": 0.0011155966203659773,
+      "loss_xval": 0.0013885498046875,
+      "num_input_tokens_seen": 206418328,
+      "step": 1400
+    },
+    {
+      "epoch": 0.40386278466416836,
+      "grad_norm": 0.5472273941624404,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "num_input_tokens_seen": 206590920,
+      "step": 1401
+    },
+    {
+      "epoch": 0.40386278466416836,
+      "loss": 0.005478480365127325,
+      "loss_ce": 0.0040613203309476376,
+      "loss_xval": 0.0014190673828125,
+      "num_input_tokens_seen": 206590920,
+      "step": 1401
+    },
+    {
+      "epoch": 0.4041510521764197,
+      "grad_norm": 0.9079608553207296,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "num_input_tokens_seen": 206725768,
+      "step": 1402
+    },
+    {
+      "epoch": 0.4041510521764197,
+      "loss": 0.015894869342446327,
+      "loss_ce": 0.014095762744545937,
+      "loss_xval": 0.001800537109375,
+      "num_input_tokens_seen": 206725768,
+      "step": 1402
+    },
+    {
+      "epoch": 0.4044393196886711,
+      "grad_norm": 1.020068142575505,
+      "learning_rate": 0.0001,
+      "loss": 0.0039,
+      "num_input_tokens_seen": 206860864,
+      "step": 1403
+    },
+    {
+      "epoch": 0.4044393196886711,
+      "loss": 0.0025417383294552565,
+      "loss_ce": 0.0014025743585079908,
+      "loss_xval": 0.00113677978515625,
+      "num_input_tokens_seen": 206860864,
+      "step": 1403
+    },
+    {
+      "epoch": 0.40472758720092245,
+      "grad_norm": 0.7036622702056449,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "num_input_tokens_seen": 207033376,
+      "step": 1404
+    },
+    {
+      "epoch": 0.40472758720092245,
+      "loss": 0.003258179873228073,
+      "loss_ce": 0.002056073397397995,
+      "loss_xval": 0.0012054443359375,
+      "num_input_tokens_seen": 207033376,
+      "step": 1404
+    },
+    {
+      "epoch": 0.4050158547131738,
+      "grad_norm": 1.2717500529898216,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "num_input_tokens_seen": 207168192,
+      "step": 1405
+    },
+    {
+      "epoch": 0.4050158547131738,
+      "loss": 0.013880310580134392,
+      "loss_ce": 0.011514244601130486,
+      "loss_xval": 0.0023651123046875,
+      "num_input_tokens_seen": 207168192,
+      "step": 1405
+    },
+    {
+      "epoch": 0.40530412222542517,
+      "grad_norm": 0.3145447559978116,
+      "learning_rate": 0.0001,
+      "loss": 0.0038,
+      "num_input_tokens_seen": 207303128,
+      "step": 1406
+    },
+    {
+      "epoch": 0.40530412222542517,
+      "loss": 0.0033487228211015463,
+      "loss_ce": 0.002387657528743148,
+      "loss_xval": 0.0009613037109375,
+      "num_input_tokens_seen": 207303128,
+      "step": 1406
+    },
+    {
+      "epoch": 0.4055923897376766,
+      "grad_norm": 0.596831291599684,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "num_input_tokens_seen": 207475568,
+      "step": 1407
+    },
+    {
+      "epoch": 0.4055923897376766,
+      "loss": 0.004142426885664463,
+      "loss_ce": 0.0028172959573566914,
+      "loss_xval": 0.0013275146484375,
+      "num_input_tokens_seen": 207475568,
+      "step": 1407
+    },
+    {
+      "epoch": 0.40588065724992795,
+      "grad_norm": 0.20424467166333724,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "num_input_tokens_seen": 207610408,
+      "step": 1408
+    },
+    {
+      "epoch": 0.40588065724992795,
+      "loss": 0.012081284075975418,
+      "loss_ce": 0.010298389941453934,
+      "loss_xval": 0.0017852783203125,
+      "num_input_tokens_seen": 207610408,
+      "step": 1408
+    },
+    {
+      "epoch": 0.4061689247621793,
+      "grad_norm": 0.38475049889262486,
+      "learning_rate": 0.0001,
+      "loss": 0.0034,
+      "num_input_tokens_seen": 207745656,
+      "step": 1409
+    },
+    {
+      "epoch": 0.4061689247621793,
+      "loss": 0.0022318486589938402,
+      "loss_ce": 0.001303923549130559,
+      "loss_xval": 0.000926971435546875,
+      "num_input_tokens_seen": 207745656,
+      "step": 1409
+    },
+    {
+      "epoch": 0.40645719227443067,
+      "grad_norm": 0.44183879593684616,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "num_input_tokens_seen": 207918072,
+      "step": 1410
+    },
+    {
+      "epoch": 0.40645719227443067,
+      "loss": 0.003880149219185114,
+      "loss_ce": 0.0025850594975054264,
+      "loss_xval": 0.0012969970703125,
+      "num_input_tokens_seen": 207918072,
+      "step": 1410
+    },
+    {
+      "epoch": 0.40674545978668203,
+      "grad_norm": 0.5407026287377048,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "num_input_tokens_seen": 208052832,
+      "step": 1411
+    },
+    {
+      "epoch": 0.40674545978668203,
+      "loss": 0.013677089475095272,
+      "loss_ce": 0.011812656186521053,
+      "loss_xval": 0.001861572265625,
+      "num_input_tokens_seen": 208052832,
+      "step": 1411
+    },
+    {
+      "epoch": 0.4070337272989334,
+      "grad_norm": 0.32728890027750474,
+      "learning_rate": 0.0001,
+      "loss": 0.0029,
+      "num_input_tokens_seen": 208187952,
+      "step": 1412
+    },
+    {
+      "epoch": 0.4070337272989334,
+      "loss": 0.0026520793326199055,
+      "loss_ce": 0.0017924610292539,
+      "loss_xval": 0.000858306884765625,
+      "num_input_tokens_seen": 208187952,
+      "step": 1412
+    },
+    {
+      "epoch": 0.4073219948111848,
+      "grad_norm": 0.42571153146403357,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "num_input_tokens_seen": 208360488,
+      "step": 1413
+    },
+    {
+      "epoch": 0.4073219948111848,
+      "loss": 0.005467297974973917,
+      "loss_ce": 0.004301669541746378,
+      "loss_xval": 0.00116729736328125,
+      "num_input_tokens_seen": 208360488,
+      "step": 1413
+    },
+    {
+      "epoch": 0.40761026232343617,
+      "grad_norm": 0.17038287039479713,
+      "learning_rate": 0.0001,
+      "loss": 0.0062,
+      "num_input_tokens_seen": 208495232,
+      "step": 1414
+    },
+    {
+      "epoch": 0.40761026232343617,
+      "loss": 0.010554078966379166,
+      "loss_ce": 0.009341483004391193,
+      "loss_xval": 0.00121307373046875,
+      "num_input_tokens_seen": 208495232,
+      "step": 1414
+    },
+    {
+      "epoch": 0.40789852983568753,
+      "grad_norm": 0.19734809280784743,
+      "learning_rate": 0.0001,
+      "loss": 0.0029,
+      "num_input_tokens_seen": 208630288,
+      "step": 1415
+    },
+    {
+      "epoch": 0.40789852983568753,
+      "loss": 0.002154544461518526,
+      "loss_ce": 0.0012824092991650105,
+      "loss_xval": 0.000873565673828125,
+      "num_input_tokens_seen": 208630288,
+      "step": 1415
+    },
+    {
+      "epoch": 0.4081867973479389,
+      "grad_norm": 0.31871405754082566,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "num_input_tokens_seen": 208802840,
+      "step": 1416
+    },
+    {
+      "epoch": 0.4081867973479389,
+      "loss": 0.00468908017501235,
+      "loss_ce": 0.0037377900443971157,
+      "loss_xval": 0.000949859619140625,
+      "num_input_tokens_seen": 208802840,
+      "step": 1416
+    },
+    {
+      "epoch": 0.40847506486019025,
+      "grad_norm": 0.5962579211692774,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "num_input_tokens_seen": 208937632,
+      "step": 1417
+    },
+    {
+      "epoch": 0.40847506486019025,
+      "loss": 0.017369378358125687,
+      "loss_ce": 0.015037644654512405,
+      "loss_xval": 0.0023345947265625,
+      "num_input_tokens_seen": 208937632,
+      "step": 1417
+    },
+    {
+      "epoch": 0.4087633323724416,
+      "grad_norm": 0.9337263657247876,
+      "learning_rate": 0.0001,
+      "loss": 0.0054,
+      "num_input_tokens_seen": 209072952,
+      "step": 1418
+    },
+    {
+      "epoch": 0.4087633323724416,
+      "loss": 0.0018941645976155996,
+      "loss_ce": 0.0009700541850179434,
+      "loss_xval": 0.00092315673828125,
+      "num_input_tokens_seen": 209072952,
+      "step": 1418
+    },
+    {
+      "epoch": 0.409051599884693,
+      "grad_norm": 0.4026671847572958,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "num_input_tokens_seen": 209245536,
+      "step": 1419
+    },
+    {
+      "epoch": 0.409051599884693,
+      "loss": 0.006634651683270931,
+      "loss_ce": 0.005855022929608822,
+      "loss_xval": 0.0007781982421875,
+      "num_input_tokens_seen": 209245536,
+      "step": 1419
+    },
+    {
+      "epoch": 0.4093398673969444,
+      "grad_norm": 0.5314251752133781,
+      "learning_rate": 0.0001,
+      "loss": 0.0056,
+      "num_input_tokens_seen": 209380256,
+      "step": 1420
+    },
+    {
+      "epoch": 0.4093398673969444,
+      "loss": 0.009849036112427711,
+      "loss_ce": 0.008433545008301735,
+      "loss_xval": 0.0014190673828125,
+      "num_input_tokens_seen": 209380256,
+      "step": 1420
+    },
+    {
+      "epoch": 0.40962813490919575,
+      "grad_norm": 0.20906559044152748,
+      "learning_rate": 0.0001,
+      "loss": 0.0049,
+      "num_input_tokens_seen": 209515304,
+      "step": 1421
+    },
+    {
+      "epoch": 0.40962813490919575,
+      "loss": 0.002209738129749894,
+      "loss_ce": 0.0014544283039867878,
+      "loss_xval": 0.00075531005859375,
+      "num_input_tokens_seen": 209515304,
+      "step": 1421
+    },
+    {
+      "epoch": 0.4099164024214471,
+      "grad_norm": 0.26415178813757945,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "num_input_tokens_seen": 209687688,
+      "step": 1422
+    },
+    {
+      "epoch": 0.4099164024214471,
+      "loss": 0.002843210706487298,
+      "loss_ce": 0.0018246864201501012,
+      "loss_xval": 0.0010223388671875,
+      "num_input_tokens_seen": 209687688,
+      "step": 1422
+    },
+    {
+      "epoch": 0.4102046699336985,
+      "grad_norm": 0.5415690781467778,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "num_input_tokens_seen": 209822464,
+      "step": 1423
+    },
+    {
+      "epoch": 0.4102046699336985,
+      "loss": 0.012525908648967743,
+      "loss_ce": 0.011467330157756805,
+      "loss_xval": 0.00106048583984375,
+      "num_input_tokens_seen": 209822464,
+      "step": 1423
+    },
+    {
+      "epoch": 0.41049293744594983,
+      "grad_norm": 0.8888751647930228,
+      "learning_rate": 0.0001,
+      "loss": 0.0034,
+      "num_input_tokens_seen": 209957616,
+      "step": 1424
+    },
+    {
+      "epoch": 0.41049293744594983,
+      "loss": 0.0023013115860521793,
+      "loss_ce": 0.0016084673115983605,
+      "loss_xval": 0.00069427490234375,
+      "num_input_tokens_seen": 209957616,
+      "step": 1424
+    },
+    {
+      "epoch": 0.4107812049582012,
+      "grad_norm": 0.3638501786792088,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "num_input_tokens_seen": 210130112,
+      "step": 1425
+    },
+    {
+      "epoch": 0.4107812049582012,
+      "loss": 0.0025470005348324776,
+      "loss_ce": 0.0015520796878263354,
+      "loss_xval": 0.0009918212890625,
+      "num_input_tokens_seen": 210130112,
+      "step": 1425
+    },
+    {
+      "epoch": 0.41106947247045256,
+      "grad_norm": 0.39490131021782965,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "num_input_tokens_seen": 210264872,
+      "step": 1426
+    },
+    {
+      "epoch": 0.41106947247045256,
+      "loss": 0.011294020339846611,
+      "loss_ce": 0.010136259719729424,
+      "loss_xval": 0.00115966796875,
+      "num_input_tokens_seen": 210264872,
+      "step": 1426
+    },
+    {
+      "epoch": 0.411357739982704,
+      "grad_norm": 0.5864661523256884,
+      "learning_rate": 0.0001,
+      "loss": 0.0033,
+      "num_input_tokens_seen": 210399880,
+      "step": 1427
+    },
+    {
+      "epoch": 0.411357739982704,
+      "loss": 0.0016113725723698735,
+      "loss_ce": 0.0009144750656560063,
+      "loss_xval": 0.000698089599609375,
+      "num_input_tokens_seen": 210399880,
+      "step": 1427
+    },
+    {
+      "epoch": 0.41164600749495533,
+      "grad_norm": 0.34916345557654377,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "num_input_tokens_seen": 210572416,
+      "step": 1428
+    },
+    {
+      "epoch": 0.41164600749495533,
+      "loss": 0.0048339469358325005,
+      "loss_ce": 0.0039680106565356255,
+      "loss_xval": 0.000865936279296875,
+      "num_input_tokens_seen": 210572416,
+      "step": 1428
+    },
+    {
+      "epoch": 0.4119342750072067,
+      "grad_norm": 0.29858260038323814,
+      "learning_rate": 0.0001,
+      "loss": 0.0064,
+      "num_input_tokens_seen": 210707192,
+      "step": 1429
+    },
+    {
+      "epoch": 0.4119342750072067,
+      "loss": 0.011549213901162148,
+      "loss_ce": 0.010494926944375038,
+      "loss_xval": 0.0010528564453125,
+      "num_input_tokens_seen": 210707192,
+      "step": 1429
+    },
+    {
+      "epoch": 0.41222254251945806,
+      "grad_norm": 0.33206058421520657,
+      "learning_rate": 0.0001,
+      "loss": 0.0019,
+      "num_input_tokens_seen": 210842240,
+      "step": 1430
+    },
+    {
+      "epoch": 0.41222254251945806,
+      "loss": 0.001508349203504622,
+      "loss_ce": 0.0008450687164440751,
+      "loss_xval": 0.00066375732421875,
+      "num_input_tokens_seen": 210842240,
+      "step": 1430
+    },
+    {
+      "epoch": 0.4125108100317094,
+      "grad_norm": 0.40896687569605294,
+      "learning_rate": 0.0001,
+      "loss": 0.0063,
+      "num_input_tokens_seen": 211014728,
+      "step": 1431
+    },
+    {
+      "epoch": 0.4125108100317094,
+      "loss": 0.003282485995441675,
+      "loss_ce": 0.002321182284504175,
+      "loss_xval": 0.0009613037109375,
+      "num_input_tokens_seen": 211014728,
+      "step": 1431
+    },
+    {
+      "epoch": 0.4127990775439608,
+      "grad_norm": 0.3321492134803992,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "num_input_tokens_seen": 211149584,
+      "step": 1432
+    },
+    {
+      "epoch": 0.4127990775439608,
+      "loss": 0.013707758858799934,
+      "loss_ce": 0.012455584481358528,
+      "loss_xval": 0.001251220703125,
+      "num_input_tokens_seen": 211149584,
+      "step": 1432
+    },
+    {
+      "epoch": 0.41308734505621214,
+      "grad_norm": 0.3826896427237596,
+      "learning_rate": 0.0001,
+      "loss": 0.003,
+      "num_input_tokens_seen": 211284688,
+      "step": 1433
+    },
+    {
+      "epoch": 0.41308734505621214,
+      "loss": 0.0016002252232283354,
+      "loss_ce": 0.0010268285404890776,
+      "loss_xval": 0.00057220458984375,
+      "num_input_tokens_seen": 211284688,
+      "step": 1433
+    },
+    {
+      "epoch": 0.41337561256846356,
+      "grad_norm": 0.3051905722859315,
+      "learning_rate": 0.0001,
+      "loss": 0.0061,
+      "num_input_tokens_seen": 211457040,
+      "step": 1434
+    },
+    {
+      "epoch": 0.41337561256846356,
+      "loss": 0.002180540468543768,
+      "loss_ce": 0.0013804077170789242,
+      "loss_xval": 0.00080108642578125,
+      "num_input_tokens_seen": 211457040,
+      "step": 1434
+    },
+    {
+      "epoch": 0.4136638800807149,
+      "grad_norm": 0.6753060604545037,
+      "learning_rate": 0.0001,
+      "loss": 0.0062,
+      "num_input_tokens_seen": 211591744,
+      "step": 1435
+    },
+    {
+      "epoch": 0.4136638800807149,
+      "loss": 0.011045737192034721,
+      "loss_ce": 0.00985841266810894,
+      "loss_xval": 0.001190185546875,
+      "num_input_tokens_seen": 211591744,
+      "step": 1435
+    },
+    {
+      "epoch": 0.4139521475929663,
+      "grad_norm": 0.47751356907696696,
+      "learning_rate": 0.0001,
+      "loss": 0.004,
+      "num_input_tokens_seen": 211726792,
+      "step": 1436
+    },
+    {
+      "epoch": 0.4139521475929663,
+      "loss": 0.0013247502502053976,
+      "loss_ce": 0.0006855500396341085,
+      "loss_xval": 0.000640869140625,
+      "num_input_tokens_seen": 211726792,
+      "step": 1436
+    },
+    {
+      "epoch": 0.41424041510521764,
+      "grad_norm": 0.48960108155618903,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "num_input_tokens_seen": 211899272,
+      "step": 1437
+    },
+    {
+      "epoch": 0.41424041510521764,
+      "loss": 0.008046119473874569,
+      "loss_ce": 0.0073160817846655846,
+      "loss_xval": 0.000728607177734375,
+      "num_input_tokens_seen": 211899272,
+      "step": 1437
+    },
+    {
+      "epoch": 0.414528682617469,
+      "grad_norm": 0.712602502926373,
+      "learning_rate": 0.0001,
+      "loss": 0.0066,
+      "num_input_tokens_seen": 212034000,
+      "step": 1438
+    },
+    {
+      "epoch": 0.414528682617469,
+      "loss": 0.01144714467227459,
+      "loss_ce": 0.010413838550448418,
+      "loss_xval": 0.00102996826171875,
+      "num_input_tokens_seen": 212034000,
+      "step": 1438
+    },
+    {
+      "epoch": 0.41481695012972036,
+      "grad_norm": 0.4055228692809266,
+      "learning_rate": 0.0001,
+      "loss": 0.0039,
+      "num_input_tokens_seen": 212169088,
+      "step": 1439
+    },
+    {
+      "epoch": 0.41481695012972036,
+      "loss": 0.003060694085434079,
+      "loss_ce": 0.0025781351141631603,
+      "loss_xval": 0.0004825592041015625,
+      "num_input_tokens_seen": 212169088,
+      "step": 1439
+    },
+    {
+      "epoch": 0.4151052176419717,
+      "grad_norm": 0.4669629111386831,
+      "learning_rate": 0.0001,
+      "loss": 0.0062,
+      "num_input_tokens_seen": 212341600,
+      "step": 1440
+    },
+    {
+      "epoch": 0.4151052176419717,
+      "loss": 0.0033066237810999155,
+      "loss_ce": 0.0024750197771936655,
+      "loss_xval": 0.00083160400390625,
+      "num_input_tokens_seen": 212341600,
+      "step": 1440
+    },
+    {
+      "epoch": 0.41539348515422314,
+      "grad_norm": 0.272947734067498,
+      "learning_rate": 0.0001,
+      "loss": 0.0048,
+      "num_input_tokens_seen": 212476392,
+      "step": 1441
+    },
+    {
+      "epoch": 0.41539348515422314,
+      "loss": 0.008491845801472664,
+      "loss_ce": 0.007533402647823095,
+      "loss_xval": 0.000957489013671875,
+      "num_input_tokens_seen": 212476392,
+      "step": 1441
+    },
+    {
+      "epoch": 0.4156817526664745,
+      "grad_norm": 0.6574337302620793,
+      "learning_rate": 0.0001,
+      "loss": 0.0023,
+      "num_input_tokens_seen": 212611504,
+      "step": 1442
+    },
+    {
+      "epoch": 0.4156817526664745,
+      "loss": 0.001446253852918744,
+      "loss_ce": 0.0008158752461895347,
+      "loss_xval": 0.000629425048828125,
+      "num_input_tokens_seen": 212611504,
+      "step": 1442
+    },
+    {
+      "epoch": 0.41597002017872586,
+      "grad_norm": 0.5084837390614846,
+      "learning_rate": 0.0001,
+      "loss": 0.0058,
+      "num_input_tokens_seen": 212784000,
+      "step": 1443
+    },
+    {
+      "epoch": 0.41597002017872586,
+      "loss": 0.001965146278962493,
+      "loss_ce": 0.001111130928620696,
+      "loss_xval": 0.0008544921875,
+      "num_input_tokens_seen": 212784000,
+      "step": 1443
+    },
+    {
+      "epoch": 0.4162582876909772,
+      "grad_norm": 0.10623955238908554,
+      "learning_rate": 0.0001,
+      "loss": 0.0057,
+      "num_input_tokens_seen": 212918800,
+      "step": 1444
+    },
+    {
+      "epoch": 0.4162582876909772,
+      "loss": 0.009658853523433208,
+      "loss_ce": 0.00887636374682188,
+      "loss_xval": 0.000782012939453125,
+      "num_input_tokens_seen": 212918800,
+      "step": 1444
+    },
+    {
+      "epoch": 0.4165465552032286,
+      "grad_norm": 0.26211817991715775,
+      "learning_rate": 0.0001,
+      "loss": 0.0018,
+      "num_input_tokens_seen": 213053888,
+      "step": 1445
+    },
+    {
+      "epoch": 0.4165465552032286,
+      "loss": 0.001269388129003346,
+      "loss_ce": 0.0007265089661814272,
+      "loss_xval": 0.00054168701171875,
+      "num_input_tokens_seen": 213053888,
+      "step": 1445
+    },
+    {
+      "epoch": 0.41683482271547995,
+      "grad_norm": 0.3816847147583729,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "num_input_tokens_seen": 213226456,
+      "step": 1446
+    },
+    {
+      "epoch": 0.41683482271547995,
+      "loss": 0.0040206024423241615,
+      "loss_ce": 0.003081233473494649,
+      "loss_xval": 0.00093841552734375,
+      "num_input_tokens_seen": 213226456,
+      "step": 1446
+    },
+    {
+      "epoch": 0.4171230902277313,
+      "grad_norm": 0.826796537032878,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "num_input_tokens_seen": 213361392,
+      "step": 1447
+    },
+    {
+      "epoch": 0.4171230902277313,
+      "loss": 0.01285298727452755,
+      "loss_ce": 0.011970838531851768,
+      "loss_xval": 0.000881195068359375,
+      "num_input_tokens_seen": 213361392,
+      "step": 1447
+    },
+    {
+      "epoch": 0.4174113577399827,
+      "grad_norm": 0.7368095171873357,
+      "learning_rate": 0.0001,
+      "loss": 0.0042,
+      "num_input_tokens_seen": 213496472,
+      "step": 1448
+    },
+    {
+      "epoch": 0.4174113577399827,
+      "loss": 0.001192914554849267,
+      "loss_ce": 0.0007071368163451552,
+      "loss_xval": 0.0004863739013671875,
+      "num_input_tokens_seen": 213496472,
+      "step": 1448
+    },
+    {
+      "epoch": 0.4176996252522341,
+      "grad_norm": 0.7264266905359824,
+      "learning_rate": 0.0001,
+      "loss": 0.0064,
+      "num_input_tokens_seen": 213669032,
+      "step": 1449
+    },
+    {
+      "epoch": 0.4176996252522341,
+      "loss": 0.00256285909563303,
+      "loss_ce": 0.00180325738620013,
+      "loss_xval": 0.000759124755859375,
+      "num_input_tokens_seen": 213669032,
+      "step": 1449
+    },
+    {
+      "epoch": 0.41798789276448545,
+      "grad_norm": 0.8756888162528366,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "num_input_tokens_seen": 213803896,
+      "step": 1450
+    },
+    {
+      "epoch": 0.41798789276448545,
+      "loss": 0.011397944763302803,
+      "loss_ce": 0.010598527267575264,
+      "loss_xval": 0.00080108642578125,
+      "num_input_tokens_seen": 213803896,
+      "step": 1450
+    },
+    {
+      "epoch": 0.4182761602767368,
+      "grad_norm": 0.6703661093710185,
+      "learning_rate": 0.0001,
+      "loss": 0.0019,
+      "num_input_tokens_seen": 213938976,
+      "step": 1451
+    },
+    {
+      "epoch": 0.4182761602767368,
+      "loss": 0.0011391295120120049,
+      "loss_ce": 0.0006146086379885674,
+      "loss_xval": 0.00052642822265625,
+      "num_input_tokens_seen": 213938976,
+      "step": 1451
+    },
+    {
+      "epoch": 0.41856442778898817,
+      "grad_norm": 1.1802014879345701,
+      "learning_rate": 0.0001,
+      "loss": 0.0064,
+      "num_input_tokens_seen": 214111544,
+      "step": 1452
+    },
+    {
+      "epoch": 0.41856442778898817,
+      "loss": 0.002023919951170683,
+      "loss_ce": 0.0014066543662920594,
+      "loss_xval": 0.00061798095703125,
+      "num_input_tokens_seen": 214111544,
+      "step": 1452
+    },
+    {
+      "epoch": 0.41885269530123953,
+      "grad_norm": 0.40651186118730853,
+      "learning_rate": 0.0001,
+      "loss": 0.0049,
+      "num_input_tokens_seen": 214246264,
+      "step": 1453
+    },
+    {
+      "epoch": 0.41885269530123953,
+      "loss": 0.008859092369675636,
+      "loss_ce": 0.008055144920945168,
+      "loss_xval": 0.000804901123046875,
+      "num_input_tokens_seen": 214246264,
+      "step": 1453
+    },
+    {
+      "epoch": 0.41914096281349095,
+      "grad_norm": 1.2407763603561572,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "num_input_tokens_seen": 214381288,
+      "step": 1454
+    },
+    {
+      "epoch": 0.41914096281349095,
+      "loss": 0.0013691552449017763,
+      "loss_ce": 0.0006729729357175529,
+      "loss_xval": 0.00069427490234375,
+      "num_input_tokens_seen": 214381288,
+      "step": 1454
+    },
+    {
+      "epoch": 0.4194292303257423,
+      "grad_norm": 0.2690818929859582,
+      "learning_rate": 0.0001,
+      "loss": 0.0063,
+      "num_input_tokens_seen": 214553744,
+      "step": 1455
+    },
+    {
+      "epoch": 0.4194292303257423,
+      "loss": 0.002206414006650448,
+      "loss_ce": 0.0016423157649114728,
+      "loss_xval": 0.0005645751953125,
+      "num_input_tokens_seen": 214553744,
+      "step": 1455
+    },
+    {
+      "epoch": 0.41971749783799367,
+      "grad_norm": 1.0529441379573192,
+      "learning_rate": 0.0001,
+      "loss": 0.0052,
+      "num_input_tokens_seen": 214688496,
+      "step": 1456
+    },
+    {
+      "epoch": 0.41971749783799367,
+      "loss": 0.008993783965706825,
+      "loss_ce": 0.008180061355233192,
+      "loss_xval": 0.000812530517578125,
+      "num_input_tokens_seen": 214688496,
+      "step": 1456
+    },
+    {
+      "epoch": 0.42000576535024503,
+      "grad_norm": 0.8358229366317473,
+      "learning_rate": 0.0001,
+      "loss": 0.0014,
+      "num_input_tokens_seen": 214823544,
+      "step": 1457
+    },
+    {
+      "epoch": 0.42000576535024503,
+      "loss": 0.0011128491023555398,
+      "loss_ce": 0.000553042278625071,
+      "loss_xval": 0.000560760498046875,
+      "num_input_tokens_seen": 214823544,
+      "step": 1457
+    },
+    {
+      "epoch": 0.4202940328624964,
+      "grad_norm": 0.665178161614472,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "num_input_tokens_seen": 214996008,
+      "step": 1458
+    },
+    {
+      "epoch": 0.4202940328624964,
+      "loss": 0.0021955184638500214,
+      "loss_ce": 0.001614253968000412,
+      "loss_xval": 0.000579833984375,
+      "num_input_tokens_seen": 214996008,
+      "step": 1458
+    },
+    {
+      "epoch": 0.42058230037474775,
+      "grad_norm": 0.9507247443250465,
+      "learning_rate": 0.0001,
+      "loss": 0.0044,
+      "num_input_tokens_seen": 215130744,
+      "step": 1459
+    },
+    {
+      "epoch": 0.42058230037474775,
+      "loss": 0.00750365573912859,
+      "loss_ce": 0.006683019455522299,
+      "loss_xval": 0.000820159912109375,
+      "num_input_tokens_seen": 215130744,
+      "step": 1459
+    },
+    {
+      "epoch": 0.4208705678869991,
+      "grad_norm": 0.35005294419163846,
+      "learning_rate": 0.0001,
+      "loss": 0.0019,
+      "num_input_tokens_seen": 215265888,
+      "step": 1460
+    },
+    {
+      "epoch": 0.4208705678869991,
+      "loss": 0.002211757702752948,
+      "loss_ce": 0.001746245427057147,
+      "loss_xval": 0.00046539306640625,
+      "num_input_tokens_seen": 215265888,
+      "step": 1460
+    },
+    {
+      "epoch": 0.42115883539925053,
+      "grad_norm": 0.8883463319709012,
+      "learning_rate": 0.0001,
+      "loss": 0.0061,
+      "num_input_tokens_seen": 215438312,
+      "step": 1461
+    },
+    {
+      "epoch": 0.42115883539925053,
+      "loss": 0.0020642499439418316,
+      "loss_ce": 0.0014379244530573487,
+      "loss_xval": 0.0006256103515625,
+      "num_input_tokens_seen": 215438312,
+      "step": 1461
+    },
+    {
+      "epoch": 0.4214471029115019,
+      "grad_norm": 0.38416254140929673,
+      "learning_rate": 0.0001,
+      "loss": 0.0063,
+      "num_input_tokens_seen": 215573144,
+      "step": 1462
+    },
+    {
+      "epoch": 0.4214471029115019,
+      "loss": 0.009878046810626984,
+      "loss_ce": 0.009044773876667023,
+      "loss_xval": 0.00083160400390625,
+      "num_input_tokens_seen": 215573144,
+      "step": 1462
+    },
+    {
+      "epoch": 0.42173537042375325,
+      "grad_norm": 1.170583224481744,
+      "learning_rate": 0.0001,
+      "loss": 0.0018,
+      "num_input_tokens_seen": 215708112,
+      "step": 1463
+    },
+    {
+      "epoch": 0.42173537042375325,
+      "loss": 0.0011670843232423067,
+      "loss_ce": 0.0005476729129441082,
+      "loss_xval": 0.00061798095703125,
+      "num_input_tokens_seen": 215708112,
+      "step": 1463
+    },
+    {
+      "epoch": 0.4220236379360046,
+      "grad_norm": 0.5614263522034938,
+      "learning_rate": 0.0001,
+      "loss": 0.0062,
+      "num_input_tokens_seen": 215880648,
+      "step": 1464
+    },
+    {
+      "epoch": 0.4220236379360046,
+      "loss": 0.0018251438159495592,
+      "loss_ce": 0.0012097854632884264,
+      "loss_xval": 0.000614166259765625,
+      "num_input_tokens_seen": 215880648,
+      "step": 1464
+    },
+    {
+      "epoch": 0.422311905448256,
+      "grad_norm": 0.4011228468575239,
+      "learning_rate": 0.0001,
+      "loss": 0.004,
+      "num_input_tokens_seen": 216015600,
+      "step": 1465
+    },
+    {
+      "epoch": 0.422311905448256,
+      "loss": 0.006949270609766245,
+      "loss_ce": 0.006290758494287729,
+      "loss_xval": 0.000659942626953125,
+      "num_input_tokens_seen": 216015600,
+      "step": 1465
+    },
+    {
+      "epoch": 0.42260017296050734,
+      "grad_norm": 0.15231738215848292,
+      "learning_rate": 0.0001,
+      "loss": 0.0018,
+      "num_input_tokens_seen": 216150536,
+      "step": 1466
+    },
+    {
+      "epoch": 0.42260017296050734,
+      "loss": 0.0014463100815191865,
+      "loss_ce": 0.000987592851743102,
+      "loss_xval": 0.000457763671875,
+      "num_input_tokens_seen": 216150536,
+      "step": 1466
+    },
+    {
+      "epoch": 0.4228884404727587,
+      "grad_norm": 0.9513563965505365,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "num_input_tokens_seen": 216323024,
+      "step": 1467
+    },
+    {
+      "epoch": 0.4228884404727587,
+      "loss": 0.0019695989321917295,
+      "loss_ce": 0.0012667409610003233,
+      "loss_xval": 0.000701904296875,
+      "num_input_tokens_seen": 216323024,
+      "step": 1467
+    },
+    {
+      "epoch": 0.4231767079850101,
+      "grad_norm": 0.5957654342033346,
+      "learning_rate": 0.0001,
+      "loss": 0.0058,
+      "num_input_tokens_seen": 216457800,
+      "step": 1468
+    },
+    {
+      "epoch": 0.4231767079850101,
+      "loss": 0.009507661685347557,
+      "loss_ce": 0.008915429934859276,
+      "loss_xval": 0.000591278076171875,
+      "num_input_tokens_seen": 216457800,
+      "step": 1468
+    },
+    {
+      "epoch": 0.4234649754972615,
+      "grad_norm": 0.7614670433113088,
+      "learning_rate": 0.0001,
+      "loss": 0.0018,
+      "num_input_tokens_seen": 216592960,
+      "step": 1469
+    },
+    {
+      "epoch": 0.4234649754972615,
+      "loss": 0.0016541981603950262,
+      "loss_ce": 0.0011575722601264715,
+      "loss_xval": 0.00049591064453125,
+      "num_input_tokens_seen": 216592960,
+      "step": 1469
+    },
+    {
+      "epoch": 0.42375324300951284,
+      "grad_norm": 1.270700724520404,
+      "learning_rate": 0.0001,
+      "loss": 0.005,
+      "num_input_tokens_seen": 216765584,
+      "step": 1470
+    },
+    {
+      "epoch": 0.42375324300951284,
+      "loss": 0.0014716139994561672,
+      "loss_ce": 0.0008128635818138719,
+      "loss_xval": 0.000659942626953125,
+      "num_input_tokens_seen": 216765584,
+      "step": 1470
+    },
+    {
+      "epoch": 0.4240415105217642,
+      "grad_norm": 0.5875145747840825,
+      "learning_rate": 0.0001,
+      "loss": 0.0055,
+      "num_input_tokens_seen": 216900376,
+      "step": 1471
+    },
+    {
+      "epoch": 0.4240415105217642,
+      "loss": 0.009993959218263626,
+      "loss_ce": 0.009399700909852982,
+      "loss_xval": 0.0005950927734375,
+      "num_input_tokens_seen": 216900376,
+      "step": 1471
+    },
+    {
+      "epoch": 0.42432977803401556,
+      "grad_norm": 0.3045933975449817,
+      "learning_rate": 0.0001,
+      "loss": 0.0036,
+      "num_input_tokens_seen": 217035464,
+      "step": 1472
+    },
+    {
+      "epoch": 0.42432977803401556,
+      "loss": 0.000996191636659205,
+      "loss_ce": 0.00048502214485779405,
+      "loss_xval": 0.00051116943359375,
+      "num_input_tokens_seen": 217035464,
+      "step": 1472
+    },
+    {
+      "epoch": 0.4246180455462669,
+      "grad_norm": 0.5158276363401353,
+      "learning_rate": 0.0001,
+      "loss": 0.0046,
+      "num_input_tokens_seen": 217207904,
+      "step": 1473
+    },
+    {
+      "epoch": 0.4246180455462669,
+      "loss": 0.001819569617509842,
+      "loss_ce": 0.0012697763741016388,
+      "loss_xval": 0.00054931640625,
+      "num_input_tokens_seen": 217207904,
+      "step": 1473
+    },
+    {
+      "epoch": 0.4249063130585183,
+      "grad_norm": 0.28704228820610606,
+      "learning_rate": 0.0001,
+      "loss": 0.0058,
+      "num_input_tokens_seen": 217342744,
+      "step": 1474
+    },
+    {
+      "epoch": 0.4249063130585183,
+      "loss": 0.010570655576884747,
+      "loss_ce": 0.009992490522563457,
+      "loss_xval": 0.000579833984375,
+      "num_input_tokens_seen": 217342744,
+      "step": 1474
+    },
+    {
+      "epoch": 0.4251945805707697,
+      "grad_norm": 0.10951631417043917,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "num_input_tokens_seen": 217477784,
+      "step": 1475
+    },
+    {
+      "epoch": 0.4251945805707697,
+      "loss": 0.0009299661032855511,
+      "loss_ce": 0.0004757787100970745,
+      "loss_xval": 0.000453948974609375,
+      "num_input_tokens_seen": 217477784,
+      "step": 1475
+    },
+    {
+      "epoch": 0.42548284808302106,
+      "grad_norm": 0.1979496940686989,
+      "learning_rate": 0.0001,
+      "loss": 0.0054,
+      "num_input_tokens_seen": 217650280,
+      "step": 1476
+    },
+    {
+      "epoch": 0.42548284808302106,
+      "loss": 0.0023711256217211485,
+      "loss_ce": 0.0018344454001635313,
+      "loss_xval": 0.000537872314453125,
+      "num_input_tokens_seen": 217650280,
+      "step": 1476
+    },
+    {
+      "epoch": 0.4257711155952724,
+      "grad_norm": 0.22928083269061086,
+      "learning_rate": 0.0001,
+      "loss": 0.0058,
+      "num_input_tokens_seen": 217785024,
+      "step": 1477
+    },
+    {
+      "epoch": 0.4257711155952724,
+      "loss": 0.010517632588744164,
+      "loss_ce": 0.009880578145384789,
+      "loss_xval": 0.000637054443359375,
+      "num_input_tokens_seen": 217785024,
+      "step": 1477
+    },
+    {
+      "epoch": 0.4260593831075238,
+      "grad_norm": 0.29270217498181766,
+      "learning_rate": 0.0001,
+      "loss": 0.0032,
+      "num_input_tokens_seen": 217920096,
+      "step": 1478
+    },
+    {
+      "epoch": 0.4260593831075238,
+      "loss": 0.0040700966492295265,
+      "loss_ce": 0.0037148527335375547,
+      "loss_xval": 0.000354766845703125,
+      "num_input_tokens_seen": 217920096,
+      "step": 1478
+    },
+    {
+      "epoch": 0.42634765061977514,
+      "grad_norm": 0.2624178326681669,
+      "learning_rate": 0.0001,
+      "loss": 0.0054,
+      "num_input_tokens_seen": 218092560,
+      "step": 1479
+    },
+    {
+      "epoch": 0.42634765061977514,
+      "loss": 0.0016003092750906944,
+      "loss_ce": 0.0011647185310721397,
+      "loss_xval": 0.00043487548828125,
+      "num_input_tokens_seen": 218092560,
+      "step": 1479
+    },
+    {
+      "epoch": 0.4266359181320265,
+      "grad_norm": 0.6343513122065932,
+      "learning_rate": 0.0001,
+      "loss": 0.0057,
+      "num_input_tokens_seen": 218227344,
+      "step": 1480
+    },
+    {
+      "epoch": 0.4266359181320265,
+      "loss": 0.010287400335073471,
+      "loss_ce": 0.009708281606435776,
+      "loss_xval": 0.000579833984375,
+      "num_input_tokens_seen": 218227344,
+      "step": 1480
+    },
+    {
+      "epoch": 0.42692418564427786,
+      "grad_norm": 0.39304353113000834,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "num_input_tokens_seen": 218362440,
+      "step": 1481
+    },
+    {
+      "epoch": 0.42692418564427786,
+      "loss": 0.0008061963599175215,
+      "loss_ce": 0.00044070067815482616,
+      "loss_xval": 0.0003662109375,
+      "num_input_tokens_seen": 218362440,
+      "step": 1481
+    },
+    {
+      "epoch": 0.4272124531565293,
+      "grad_norm": 12.747643337776775,
+      "learning_rate": 0.0001,
+      "loss": 0.0642,
+      "num_input_tokens_seen": 218535016,
+      "step": 1482
+    },
+    {
+      "epoch": 0.4272124531565293,
+      "loss": 0.11798842996358871,
+      "loss_ce": 0.0037561291828751564,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 218535016,
+      "step": 1482
+    },
+    {
+      "epoch": 0.42750072066878064,
+      "grad_norm": 4.612439740601446,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "num_input_tokens_seen": 218669808,
+      "step": 1483
+    },
+    {
+      "epoch": 0.42750072066878064,
+      "loss": 0.01200609840452671,
+      "loss_ce": 0.009215647354722023,
+      "loss_xval": 0.0027923583984375,
+      "num_input_tokens_seen": 218669808,
+      "step": 1483
+    },
+    {
+      "epoch": 0.427788988181032,
+      "grad_norm": 3.9401909327583904,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "num_input_tokens_seen": 218804864,
+      "step": 1484
+    },
+    {
+      "epoch": 0.427788988181032,
+      "loss": 0.003662645351141691,
+      "loss_ce": 0.0008264179341495037,
+      "loss_xval": 0.002838134765625,
+      "num_input_tokens_seen": 218804864,
+      "step": 1484
+    },
+    {
+      "epoch": 0.42807725569328337,
+      "grad_norm": 2.515274796217096,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "num_input_tokens_seen": 218977456,
+      "step": 1485
+    },
+    {
+      "epoch": 0.42807725569328337,
+      "loss": 0.01962754875421524,
+      "loss_ce": 0.01210115011781454,
+      "loss_xval": 0.007537841796875,
+      "num_input_tokens_seen": 218977456,
+      "step": 1485
+    },
+    {
+      "epoch": 0.4283655232055347,
+      "grad_norm": 5.907577211244805,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "num_input_tokens_seen": 219112264,
+      "step": 1486
+    },
+    {
+      "epoch": 0.4283655232055347,
+      "loss": 0.021406885236501694,
+      "loss_ce": 0.011324640363454819,
+      "loss_xval": 0.01007080078125,
+      "num_input_tokens_seen": 219112264,
+      "step": 1486
+    },
+    {
+      "epoch": 0.4286537907177861,
+      "grad_norm": 1.8264295246632904,
+      "learning_rate": 0.0001,
+      "loss": 0.0064,
+      "num_input_tokens_seen": 219247248,
+      "step": 1487
+    },
+    {
+      "epoch": 0.4286537907177861,
+      "loss": 0.0038554510101675987,
+      "loss_ce": 0.0008065543370321393,
+      "loss_xval": 0.0030517578125,
+      "num_input_tokens_seen": 219247248,
+      "step": 1487
+    },
+    {
+      "epoch": 0.42894205823003745,
+      "grad_norm": 5.810858391504198,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "num_input_tokens_seen": 219419848,
+      "step": 1488
+    },
+    {
+      "epoch": 0.42894205823003745,
+      "loss": 0.013821925967931747,
+      "loss_ce": 0.0071347616612911224,
+      "loss_xval": 0.006683349609375,
+      "num_input_tokens_seen": 219419848,
+      "step": 1488
+    },
+    {
+      "epoch": 0.42923032574228887,
+      "grad_norm": 6.5915972029136105,
+      "learning_rate": 0.0001,
+      "loss": 0.0214,
+      "num_input_tokens_seen": 219554760,
+      "step": 1489
+    },
+    {
+      "epoch": 0.42923032574228887,
+      "loss": 0.02457248792052269,
+      "loss_ce": 0.01850711926817894,
+      "loss_xval": 0.006072998046875,
+      "num_input_tokens_seen": 219554760,
+      "step": 1489
+    },
+    {
+      "epoch": 0.4295185932545402,
+      "grad_norm": 3.073420111985128,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "num_input_tokens_seen": 219689760,
+      "step": 1490
+    },
+    {
+      "epoch": 0.4295185932545402,
+      "loss": 0.012378985062241554,
+      "loss_ce": 0.005218798760324717,
+      "loss_xval": 0.007171630859375,
+      "num_input_tokens_seen": 219689760,
+      "step": 1490
+    },
+    {
+      "epoch": 0.4298068607667916,
+      "grad_norm": 7.785497592641905,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "num_input_tokens_seen": 219862240,
+      "step": 1491
+    },
+    {
+      "epoch": 0.4298068607667916,
+      "loss": 0.00931491144001484,
+      "loss_ce": 0.0023263858165591955,
+      "loss_xval": 0.006988525390625,
+      "num_input_tokens_seen": 219862240,
+      "step": 1491
+    },
+    {
+      "epoch": 0.43009512827904295,
+      "grad_norm": 2.140618993640047,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "num_input_tokens_seen": 219997032,
+      "step": 1492
+    },
+    {
+      "epoch": 0.43009512827904295,
+      "loss": 0.013707047328352928,
+      "loss_ce": 0.011228448711335659,
+      "loss_xval": 0.002471923828125,
+      "num_input_tokens_seen": 219997032,
+      "step": 1492
+    },
+    {
+      "epoch": 0.4303833957912943,
+      "grad_norm": 7.906417162325639,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "num_input_tokens_seen": 220132048,
+      "step": 1493
+    },
+    {
+      "epoch": 0.4303833957912943,
+      "loss": 0.009202736429870129,
+      "loss_ce": 0.0014970483025535941,
+      "loss_xval": 0.0076904296875,
+      "num_input_tokens_seen": 220132048,
+      "step": 1493
+    },
+    {
+      "epoch": 0.43067166330354567,
+      "grad_norm": 8.165985635154161,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "num_input_tokens_seen": 220304544,
+      "step": 1494
+    },
+    {
+      "epoch": 0.43067166330354567,
+      "loss": 0.010727154091000557,
+      "loss_ce": 0.0025369995273649693,
+      "loss_xval": 0.0081787109375,
+      "num_input_tokens_seen": 220304544,
+      "step": 1494
+    },
+    {
+      "epoch": 0.43095993081579703,
+      "grad_norm": 2.5659378266589496,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "num_input_tokens_seen": 220439288,
+      "step": 1495
+    },
+    {
+      "epoch": 0.43095993081579703,
+      "loss": 0.012238608673214912,
+      "loss_ce": 0.008729087188839912,
+      "loss_xval": 0.003509521484375,
+      "num_input_tokens_seen": 220439288,
+      "step": 1495
+    },
+    {
+      "epoch": 0.43124819832804845,
+      "grad_norm": 9.235219599695903,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "num_input_tokens_seen": 220574336,
+      "step": 1496
+    },
+    {
+      "epoch": 0.43124819832804845,
+      "loss": 0.009838768281042576,
+      "loss_ce": 0.000660606543533504,
+      "loss_xval": 0.0091552734375,
+      "num_input_tokens_seen": 220574336,
+      "step": 1496
+    },
+    {
+      "epoch": 0.4315364658402998,
+      "grad_norm": 2.91361378442123,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "num_input_tokens_seen": 220746832,
+      "step": 1497
+    },
+    {
+      "epoch": 0.4315364658402998,
+      "loss": 0.007760161068290472,
+      "loss_ce": 0.004582518711686134,
+      "loss_xval": 0.003173828125,
+      "num_input_tokens_seen": 220746832,
+      "step": 1497
+    },
+    {
+      "epoch": 0.43182473335255117,
+      "grad_norm": 8.514799900684112,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "num_input_tokens_seen": 220881584,
+      "step": 1498
+    },
+    {
+      "epoch": 0.43182473335255117,
+      "loss": 0.017550978809595108,
+      "loss_ce": 0.010741745121777058,
+      "loss_xval": 0.006805419921875,
+      "num_input_tokens_seen": 220881584,
+      "step": 1498
+    },
+    {
+      "epoch": 0.43211300086480253,
+      "grad_norm": 8.262073445905802,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "num_input_tokens_seen": 221016736,
+      "step": 1499
+    },
+    {
+      "epoch": 0.43211300086480253,
+      "loss": 0.008337888866662979,
+      "loss_ce": 0.0006856058025732636,
+      "loss_xval": 0.007659912109375,
+      "num_input_tokens_seen": 221016736,
+      "step": 1499
+    },
+    {
+      "epoch": 0.4324012683770539,
+      "grad_norm": 4.097932252545542,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "num_input_tokens_seen": 221189264,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4324012683770539,
+      "eval_websight_new_IoU": 0.0,
+      "eval_websight_new_MAE_x": 0.11273882165551186,
+      "eval_websight_new_MAE_y": 0.13310284167528152,
+      "eval_websight_new_NUM_probability": 0.9946940243244171,
+      "eval_websight_new_inside_bbox": 0.0,
+      "eval_websight_new_loss": 0.024503586813807487,
+      "eval_websight_new_loss_ce": 0.0006861111614853144,
+      "eval_websight_new_loss_xval": 0.02381134033203125,
+      "eval_websight_new_runtime": 35.3106,
+      "eval_websight_new_samples_per_second": 1.416,
+      "eval_websight_new_steps_per_second": 0.057,
+      "num_input_tokens_seen": 221189264,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4324012683770539,
+      "eval_seeclick_IoU": 0.002761509735137224,
+      "eval_seeclick_MAE_x": 0.08998997882008553,
+      "eval_seeclick_MAE_y": 0.09225813671946526,
+      "eval_seeclick_NUM_probability": 0.9951084852218628,
+      "eval_seeclick_inside_bbox": 0.1493055559694767,
+      "eval_seeclick_loss": 0.024002188816666603,
+      "eval_seeclick_loss_ce": 0.010312775615602732,
+      "eval_seeclick_loss_xval": 0.014301300048828125,
+      "eval_seeclick_runtime": 65.2763,
+      "eval_seeclick_samples_per_second": 0.766,
+      "eval_seeclick_steps_per_second": 0.031,
+      "num_input_tokens_seen": 221189264,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4324012683770539,
+      "eval_icons_IoU": 0.0,
+      "eval_icons_MAE_x": 0.11441490426659584,
+      "eval_icons_MAE_y": 0.14393816888332367,
+      "eval_icons_NUM_probability": 0.9947239756584167,
+      "eval_icons_inside_bbox": 0.0,
+      "eval_icons_loss": 0.03313892334699631,
+      "eval_icons_loss_ce": 0.013403720688074827,
+      "eval_icons_loss_xval": 0.019145965576171875,
+      "eval_icons_runtime": 66.7805,
+      "eval_icons_samples_per_second": 0.749,
+      "eval_icons_steps_per_second": 0.03,
+      "num_input_tokens_seen": 221189264,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4324012683770539,
+      "loss": 0.03301580622792244,
+      "loss_ce": 0.013606625609099865,
+      "loss_xval": 0.0194091796875,
+      "num_input_tokens_seen": 221189264,
+      "step": 1500
+    },
+    {
+      "epoch": 0.43268953588930525,
+      "grad_norm": 11.926736727618014,
+      "learning_rate": 0.0001,
+      "loss": 0.0258,
+      "num_input_tokens_seen": 221324040,
+      "step": 1501
+    },
+    {
+      "epoch": 0.43268953588930525,
+      "loss": 0.0268569216132164,
+      "loss_ce": 0.0121627077460289,
+      "loss_xval": 0.01470947265625,
+      "num_input_tokens_seen": 221324040,
+      "step": 1501
+    },
+    {
+      "epoch": 0.43297780340155667,
+      "grad_norm": 3.8871982189529573,
+      "learning_rate": 0.0001,
+      "loss": 0.0043,
+      "num_input_tokens_seen": 221459064,
+      "step": 1502
+    },
+    {
+      "epoch": 0.43297780340155667,
+      "loss": 0.002975265495479107,
+      "loss_ce": 0.0007198259700089693,
+      "loss_xval": 0.00225830078125,
+      "num_input_tokens_seen": 221459064,
+      "step": 1502
+    },
+    {
+      "epoch": 0.43326607091380803,
+      "grad_norm": 10.091104014332721,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "num_input_tokens_seen": 221631456,
+      "step": 1503
+    },
+    {
+      "epoch": 0.43326607091380803,
+      "loss": 0.01266709715127945,
+      "loss_ce": 0.0017875813646242023,
+      "loss_xval": 0.0108642578125,
+      "num_input_tokens_seen": 221631456,
+      "step": 1503
+    },
+    {
+      "epoch": 0.4335543384260594,
+      "grad_norm": 12.153293017387893,
+      "learning_rate": 0.0001,
+      "loss": 0.0266,
+      "num_input_tokens_seen": 221766224,
+      "step": 1504
+    },
+    {
+      "epoch": 0.4335543384260594,
+      "loss": 0.02504376508295536,
+      "loss_ce": 0.00867108441889286,
+      "loss_xval": 0.016357421875,
+      "num_input_tokens_seen": 221766224,
+      "step": 1504
+    },
+    {
+      "epoch": 0.43384260593831075,
+      "grad_norm": 0.4609686671533385,
+      "learning_rate": 0.0001,
+      "loss": 0.0024,
+      "num_input_tokens_seen": 221901320,
+      "step": 1505
+    },
+    {
+      "epoch": 0.43384260593831075,
+      "loss": 0.0018655576277524233,
+      "loss_ce": 0.0009276189375668764,
+      "loss_xval": 0.00093841552734375,
+      "num_input_tokens_seen": 221901320,
+      "step": 1505
+    },
+    {
+      "epoch": 0.4341308734505621,
+      "grad_norm": 11.897519898174188,
+      "learning_rate": 0.0001,
+      "loss": 0.0252,
+      "num_input_tokens_seen": 222073768,
+      "step": 1506
+    },
+    {
+      "epoch": 0.4341308734505621,
+      "loss": 0.017838824540376663,
+      "loss_ce": 0.0018323541153222322,
+      "loss_xval": 0.0159912109375,
+      "num_input_tokens_seen": 222073768,
+      "step": 1506
+    },
+    {
+      "epoch": 0.4344191409628135,
+      "grad_norm": 6.977404050056601,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "num_input_tokens_seen": 222208576,
+      "step": 1507
+    },
+    {
+      "epoch": 0.4344191409628135,
+      "loss": 0.016943227499723434,
+      "loss_ce": 0.010710012167692184,
+      "loss_xval": 0.0062255859375,
+      "num_input_tokens_seen": 222208576,
+      "step": 1507
+    },
+    {
+      "epoch": 0.43470740847506484,
+      "grad_norm": 8.024467376062539,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "num_input_tokens_seen": 222343576,
+      "step": 1508
+    },
+    {
+      "epoch": 0.43470740847506484,
+      "loss": 0.011738698929548264,
+      "loss_ce": 0.0030793354380875826,
+      "loss_xval": 0.0086669921875,
+      "num_input_tokens_seen": 222343576,
+      "step": 1508
+    },
+    {
+      "epoch": 0.43499567598731625,
+      "grad_norm": 12.988579617987343,
+      "learning_rate": 0.0001,
+      "loss": 0.0281,
+      "num_input_tokens_seen": 222516032,
+      "step": 1509
+    },
+    {
+      "epoch": 0.43499567598731625,
+      "loss": 0.01961071789264679,
+      "loss_ce": 0.0027116083074361086,
+      "loss_xval": 0.016845703125,
+      "num_input_tokens_seen": 222516032,
+      "step": 1509
+    },
+    {
+      "epoch": 0.4352839434995676,
+      "grad_norm": 0.8972896129622868,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "num_input_tokens_seen": 222650880,
+      "step": 1510
+    },
+    {
+      "epoch": 0.4352839434995676,
+      "loss": 0.014146210625767708,
+      "loss_ce": 0.012663723900914192,
+      "loss_xval": 0.0014801025390625,
+      "num_input_tokens_seen": 222650880,
+      "step": 1510
+    },
+    {
+      "epoch": 0.435572211011819,
+      "grad_norm": 12.577305655701169,
+      "learning_rate": 0.0001,
+      "loss": 0.0239,
+      "num_input_tokens_seen": 222785984,
+      "step": 1511
+    },
+    {
+      "epoch": 0.435572211011819,
+      "loss": 0.019875552505254745,
+      "loss_ce": 0.0024881642311811447,
+      "loss_xval": 0.017333984375,
+      "num_input_tokens_seen": 222785984,
+      "step": 1511
+    },
+    {
+      "epoch": 0.43586047852407034,
+      "grad_norm": 8.79078606333596,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "num_input_tokens_seen": 222958576,
+      "step": 1512
+    },
+    {
+      "epoch": 0.43586047852407034,
+      "loss": 0.009240476414561272,
+      "loss_ce": 0.0009969149250537157,
+      "loss_xval": 0.00823974609375,
+      "num_input_tokens_seen": 222958576,
+      "step": 1512
+    },
+    {
+      "epoch": 0.4361487460363217,
+      "grad_norm": 6.959399811417409,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "num_input_tokens_seen": 223093344,
+      "step": 1513
+    },
+    {
+      "epoch": 0.4361487460363217,
+      "loss": 0.017352435737848282,
+      "loss_ce": 0.010268543846905231,
+      "loss_xval": 0.007080078125,
+      "num_input_tokens_seen": 223093344,
+      "step": 1513
+    },
+    {
+      "epoch": 0.43643701354857306,
+      "grad_norm": 13.354758671346826,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "num_input_tokens_seen": 223228440,
+      "step": 1514
+    },
+    {
+      "epoch": 0.43643701354857306,
+      "loss": 0.019258903339505196,
+      "loss_ce": 0.0004600748070515692,
+      "loss_xval": 0.018798828125,
+      "num_input_tokens_seen": 223228440,
+      "step": 1514
+    },
+    {
+      "epoch": 0.4367252810608244,
+      "grad_norm": 1.494140077747622,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "num_input_tokens_seen": 223400944,
+      "step": 1515
+    },
+    {
+      "epoch": 0.4367252810608244,
+      "loss": 0.0038938431534916162,
+      "loss_ce": 0.0023775009904056787,
+      "loss_xval": 0.00151824951171875,
+      "num_input_tokens_seen": 223400944,
+      "step": 1515
+    },
+    {
+      "epoch": 0.43701354857307584,
+      "grad_norm": 13.336987482759294,
+      "learning_rate": 0.0001,
+      "loss": 0.0262,
+      "num_input_tokens_seen": 223535744,
+      "step": 1516
+    },
+    {
+      "epoch": 0.43701354857307584,
+      "loss": 0.026453927159309387,
+      "loss_ce": 0.0064954305998981,
+      "loss_xval": 0.02001953125,
+      "num_input_tokens_seen": 223535744,
+      "step": 1516
+    },
+    {
+      "epoch": 0.4373018160853272,
+      "grad_norm": 10.647586085016922,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "num_input_tokens_seen": 223670856,
+      "step": 1517
+    },
+    {
+      "epoch": 0.4373018160853272,
+      "loss": 0.012804286554455757,
+      "loss_ce": 0.0009253188036382198,
+      "loss_xval": 0.01190185546875,
+      "num_input_tokens_seen": 223670856,
+      "step": 1517
+    },
+    {
+      "epoch": 0.43759008359757856,
+      "grad_norm": 6.0494325716405015,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "num_input_tokens_seen": 223843392,
+      "step": 1518
+    },
+    {
+      "epoch": 0.43759008359757856,
+      "loss": 0.00721039017662406,
+      "loss_ce": 0.0020567341707646847,
+      "loss_xval": 0.005157470703125,
+      "num_input_tokens_seen": 223843392,
+      "step": 1518
+    },
+    {
+      "epoch": 0.4378783511098299,
+      "grad_norm": 13.82175508494315,
+      "learning_rate": 0.0001,
+      "loss": 0.0286,
+      "num_input_tokens_seen": 223978120,
+      "step": 1519
+    },
+    {
+      "epoch": 0.4378783511098299,
+      "loss": 0.027493847534060478,
+      "loss_ce": 0.0073064700700342655,
+      "loss_xval": 0.0201416015625,
+      "num_input_tokens_seen": 223978120,
+      "step": 1519
+    },
+    {
+      "epoch": 0.4381666186220813,
+      "grad_norm": 1.9713786580914048,
+      "learning_rate": 0.0001,
+      "loss": 0.0022,
+      "num_input_tokens_seen": 224113256,
+      "step": 1520
+    },
+    {
+      "epoch": 0.4381666186220813,
+      "loss": 0.0016766273183748126,
+      "loss_ce": 0.000655718962661922,
+      "loss_xval": 0.0010223388671875,
+      "num_input_tokens_seen": 224113256,
+      "step": 1520
+    },
+    {
+      "epoch": 0.43845488613433264,
+      "grad_norm": 12.724023038845006,
+      "learning_rate": 0.0001,
+      "loss": 0.0266,
+      "num_input_tokens_seen": 224285712,
+      "step": 1521
+    },
+    {
+      "epoch": 0.43845488613433264,
+      "loss": 0.01906008832156658,
+      "loss_ce": 0.0006045823683962226,
+      "loss_xval": 0.0184326171875,
+      "num_input_tokens_seen": 224285712,
+      "step": 1521
+    },
+    {
+      "epoch": 0.438743153646584,
+      "grad_norm": 9.66225346832843,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "num_input_tokens_seen": 224420576,
+      "step": 1522
+    },
+    {
+      "epoch": 0.438743153646584,
+      "loss": 0.017540834844112396,
+      "loss_ce": 0.007927797734737396,
+      "loss_xval": 0.0096435546875,
+      "num_input_tokens_seen": 224420576,
+      "step": 1522
+    },
+    {
+      "epoch": 0.4390314211588354,
+      "grad_norm": 7.291506408736569,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "num_input_tokens_seen": 224555656,
+      "step": 1523
+    },
+    {
+      "epoch": 0.4390314211588354,
+      "loss": 0.009626722894608974,
+      "loss_ce": 0.0011428362922742963,
+      "loss_xval": 0.00848388671875,
+      "num_input_tokens_seen": 224555656,
+      "step": 1523
+    },
+    {
+      "epoch": 0.4393196886710868,
+      "grad_norm": 15.069879681871626,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "num_input_tokens_seen": 224728192,
+      "step": 1524
+    },
+    {
+      "epoch": 0.4393196886710868,
+      "loss": 0.027095580473542213,
+      "loss_ce": 0.001460814499296248,
+      "loss_xval": 0.025634765625,
+      "num_input_tokens_seen": 224728192,
+      "step": 1524
+    },
+    {
+      "epoch": 0.43960795618333814,
+      "grad_norm": 2.966582613154629,
+      "learning_rate": 0.0001,
+      "loss": 0.0056,
+      "num_input_tokens_seen": 224862904,
+      "step": 1525
+    },
+    {
+      "epoch": 0.43960795618333814,
+      "loss": 0.007811859715729952,
+      "loss_ce": 0.006425216794013977,
+      "loss_xval": 0.0013885498046875,
+      "num_input_tokens_seen": 224862904,
+      "step": 1525
+    },
+    {
+      "epoch": 0.4398962236955895,
+      "grad_norm": 12.893278036449145,
+      "learning_rate": 0.0001,
+      "loss": 0.0228,
+      "num_input_tokens_seen": 224997968,
+      "step": 1526
+    },
+    {
+      "epoch": 0.4398962236955895,
+      "loss": 0.02297937497496605,
+      "loss_ce": 0.0010067173279821873,
+      "loss_xval": 0.02197265625,
+      "num_input_tokens_seen": 224997968,
+      "step": 1526
+    },
+    {
+      "epoch": 0.44018449120784087,
+      "grad_norm": 11.262248783259581,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "num_input_tokens_seen": 225170504,
+      "step": 1527
+    },
+    {
+      "epoch": 0.44018449120784087,
+      "loss": 0.015866370871663094,
+      "loss_ce": 0.0014544444857165217,
+      "loss_xval": 0.014404296875,
+      "num_input_tokens_seen": 225170504,
+      "step": 1527
+    },
+    {
+      "epoch": 0.44047275872009223,
+      "grad_norm": 4.933947175820202,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "num_input_tokens_seen": 225305240,
+      "step": 1528
+    },
+    {
+      "epoch": 0.44047275872009223,
+      "loss": 0.014280106872320175,
+      "loss_ce": 0.0090959332883358,
+      "loss_xval": 0.00518798828125,
+      "num_input_tokens_seen": 225305240,
+      "step": 1528
+    },
+    {
+      "epoch": 0.4407610262323436,
+      "grad_norm": 13.410309682135633,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "num_input_tokens_seen": 225440432,
+      "step": 1529
+    },
+    {
+      "epoch": 0.4407610262323436,
+      "loss": 0.023734629154205322,
+      "loss_ce": 0.0011058447416871786,
+      "loss_xval": 0.0225830078125,
+      "num_input_tokens_seen": 225440432,
+      "step": 1529
+    },
+    {
+      "epoch": 0.441049293744595,
+      "grad_norm": 3.972873618108575,
+      "learning_rate": 0.0001,
+      "loss": 0.0068,
+      "num_input_tokens_seen": 225612808,
+      "step": 1530
+    },
+    {
+      "epoch": 0.441049293744595,
+      "loss": 0.0027857262175530195,
+      "loss_ce": 0.0006599860498681664,
+      "loss_xval": 0.0021209716796875,
+      "num_input_tokens_seen": 225612808,
+      "step": 1530
+    },
+    {
+      "epoch": 0.44133756125684637,
+      "grad_norm": 9.76112441466362,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "num_input_tokens_seen": 225747616,
+      "step": 1531
+    },
+    {
+      "epoch": 0.44133756125684637,
+      "loss": 0.02597951330244541,
+      "loss_ce": 0.012353414669632912,
+      "loss_xval": 0.01361083984375,
+      "num_input_tokens_seen": 225747616,
+      "step": 1531
+    },
+    {
+      "epoch": 0.44162582876909773,
+      "grad_norm": 9.468773076230084,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "num_input_tokens_seen": 225882616,
+      "step": 1532
+    },
+    {
+      "epoch": 0.44162582876909773,
+      "loss": 0.011748703196644783,
+      "loss_ce": 0.0003885347687173635,
+      "loss_xval": 0.0113525390625,
+      "num_input_tokens_seen": 225882616,
+      "step": 1532
+    },
+    {
+      "epoch": 0.4419140962813491,
+      "grad_norm": 3.5672797055456202,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "num_input_tokens_seen": 226055320,
+      "step": 1533
+    },
+    {
+      "epoch": 0.4419140962813491,
+      "loss": 0.008645853027701378,
+      "loss_ce": 0.006130059715360403,
+      "loss_xval": 0.0025177001953125,
+      "num_input_tokens_seen": 226055320,
+      "step": 1533
+    },
+    {
+      "epoch": 0.44220236379360045,
+      "grad_norm": 10.752583877190121,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "num_input_tokens_seen": 226190064,
+      "step": 1534
+    },
+    {
+      "epoch": 0.44220236379360045,
+      "loss": 0.028693024069070816,
+      "loss_ce": 0.012366121634840965,
+      "loss_xval": 0.016357421875,
+      "num_input_tokens_seen": 226190064,
+      "step": 1534
+    },
+    {
+      "epoch": 0.4424906313058518,
+      "grad_norm": 3.465595947976591,
+      "learning_rate": 0.0001,
+      "loss": 0.0033,
+      "num_input_tokens_seen": 226325024,
+      "step": 1535
+    },
+    {
+      "epoch": 0.4424906313058518,
+      "loss": 0.0033873319625854492,
+      "loss_ce": 0.0011738539906218648,
+      "loss_xval": 0.0022125244140625,
+      "num_input_tokens_seen": 226325024,
+      "step": 1535
+    },
+    {
+      "epoch": 0.4427788988181032,
+      "grad_norm": 7.28989255271461,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "num_input_tokens_seen": 226497560,
+      "step": 1536
+    },
+    {
+      "epoch": 0.4427788988181032,
+      "loss": 0.010725769214332104,
+      "loss_ce": 0.0028865656349807978,
+      "loss_xval": 0.0078125,
+      "num_input_tokens_seen": 226497560,
+      "step": 1536
+    },
+    {
+      "epoch": 0.4430671663303546,
+      "grad_norm": 6.376705732830883,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "num_input_tokens_seen": 226632384,
+      "step": 1537
+    },
+    {
+      "epoch": 0.4430671663303546,
+      "loss": 0.014991188421845436,
+      "loss_ce": 0.009280585683882236,
+      "loss_xval": 0.005706787109375,
+      "num_input_tokens_seen": 226632384,
+      "step": 1537
+    },
+    {
+      "epoch": 0.44335543384260595,
+      "grad_norm": 3.6645045269522893,
+      "learning_rate": 0.0001,
+      "loss": 0.0037,
+      "num_input_tokens_seen": 226767464,
+      "step": 1538
+    },
+    {
+      "epoch": 0.44335543384260595,
+      "loss": 0.00352754769846797,
+      "loss_ce": 0.0010041255736723542,
+      "loss_xval": 0.0025177001953125,
+      "num_input_tokens_seen": 226767464,
+      "step": 1538
+    },
+    {
+      "epoch": 0.4436437013548573,
+      "grad_norm": 8.060341189612492,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "num_input_tokens_seen": 226939952,
+      "step": 1539
+    },
+    {
+      "epoch": 0.4436437013548573,
+      "loss": 0.010525913909077644,
+      "loss_ce": 0.0016758155543357134,
+      "loss_xval": 0.00885009765625,
+      "num_input_tokens_seen": 226939952,
+      "step": 1539
+    },
+    {
+      "epoch": 0.4439319688671087,
+      "grad_norm": 1.1441313267565794,
+      "learning_rate": 0.0001,
+      "loss": 0.0064,
+      "num_input_tokens_seen": 227074760,
+      "step": 1540
+    },
+    {
+      "epoch": 0.4439319688671087,
+      "loss": 0.011146601289510727,
+      "loss_ce": 0.01025300845503807,
+      "loss_xval": 0.00089263916015625,
+      "num_input_tokens_seen": 227074760,
+      "step": 1540
+    },
+    {
+      "epoch": 0.44422023637936003,
+      "grad_norm": 7.097086335065152,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "num_input_tokens_seen": 227209952,
+      "step": 1541
+    },
+    {
+      "epoch": 0.44422023637936003,
+      "loss": 0.010442393831908703,
+      "loss_ce": 0.0031830251682549715,
+      "loss_xval": 0.00726318359375,
+      "num_input_tokens_seen": 227209952,
+      "step": 1541
+    },
+    {
+      "epoch": 0.4445085038916114,
+      "grad_norm": 5.103558104826032,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "num_input_tokens_seen": 227382392,
+      "step": 1542
+    },
+    {
+      "epoch": 0.4445085038916114,
+      "loss": 0.004657773766666651,
+      "loss_ce": 0.0005798626225441694,
+      "loss_xval": 0.00408935546875,
+      "num_input_tokens_seen": 227382392,
+      "step": 1542
+    },
+    {
+      "epoch": 0.4447967714038628,
+      "grad_norm": 4.064031752568589,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "num_input_tokens_seen": 227517256,
+      "step": 1543
+    },
+    {
+      "epoch": 0.4447967714038628,
+      "loss": 0.012953763827681541,
+      "loss_ce": 0.010111814364790916,
+      "loss_xval": 0.002838134765625,
+      "num_input_tokens_seen": 227517256,
+      "step": 1543
+    },
+    {
+      "epoch": 0.4450850389161142,
+      "grad_norm": 7.784847830840979,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "num_input_tokens_seen": 227652296,
+      "step": 1544
+    },
+    {
+      "epoch": 0.4450850389161142,
+      "loss": 0.00782286748290062,
+      "loss_ce": 0.0004109111032448709,
+      "loss_xval": 0.007415771484375,
+      "num_input_tokens_seen": 227652296,
+      "step": 1544
+    },
+    {
+      "epoch": 0.44537330642836553,
+      "grad_norm": 1.442024922252899,
+      "learning_rate": 0.0001,
+      "loss": 0.0064,
+      "num_input_tokens_seen": 227824768,
+      "step": 1545
+    },
+    {
+      "epoch": 0.44537330642836553,
+      "loss": 0.0016156777273863554,
+      "loss_ce": 0.0008408173453062773,
+      "loss_xval": 0.000774383544921875,
+      "num_input_tokens_seen": 227824768,
+      "step": 1545
+    },
+    {
+      "epoch": 0.4456615739406169,
+      "grad_norm": 6.247328802780809,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "num_input_tokens_seen": 227959496,
+      "step": 1546
+    },
+    {
+      "epoch": 0.4456615739406169,
+      "loss": 0.01608903706073761,
+      "loss_ce": 0.01050813589245081,
+      "loss_xval": 0.005584716796875,
+      "num_input_tokens_seen": 227959496,
+      "step": 1546
+    },
+    {
+      "epoch": 0.44594984145286826,
+      "grad_norm": 5.746839659032812,
+      "learning_rate": 0.0001,
+      "loss": 0.0055,
+      "num_input_tokens_seen": 228094560,
+      "step": 1547
+    },
+    {
+      "epoch": 0.44594984145286826,
+      "loss": 0.004420662298798561,
+      "loss_ce": 0.0004896164173260331,
+      "loss_xval": 0.003936767578125,
+      "num_input_tokens_seen": 228094560,
+      "step": 1547
+    },
+    {
+      "epoch": 0.4462381089651196,
+      "grad_norm": 1.546779408913841,
+      "learning_rate": 0.0001,
+      "loss": 0.0063,
+      "num_input_tokens_seen": 228266960,
+      "step": 1548
+    },
+    {
+      "epoch": 0.4462381089651196,
+      "loss": 0.003417445346713066,
+      "loss_ce": 0.0025643836706876755,
+      "loss_xval": 0.0008544921875,
+      "num_input_tokens_seen": 228266960,
+      "step": 1548
+    },
+    {
+      "epoch": 0.446526376477371,
+      "grad_norm": 6.600943517960566,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "num_input_tokens_seen": 228401736,
+      "step": 1549
+    },
+    {
+      "epoch": 0.446526376477371,
+      "loss": 0.014280472882091999,
+      "loss_ce": 0.008085404522716999,
+      "loss_xval": 0.006195068359375,
+      "num_input_tokens_seen": 228401736,
+      "step": 1549
+    },
+    {
+      "epoch": 0.4468146439896224,
+      "grad_norm": 3.7195384151417747,
+      "learning_rate": 0.0001,
+      "loss": 0.0027,
+      "num_input_tokens_seen": 228536792,
+      "step": 1550
+    },
+    {
+      "epoch": 0.4468146439896224,
+      "loss": 0.002413647249341011,
+      "loss_ce": 0.0004652906209230423,
+      "loss_xval": 0.00194549560546875,
+      "num_input_tokens_seen": 228536792,
+      "step": 1550
+    },
+    {
+      "epoch": 0.44710291150187376,
+      "grad_norm": 3.112689711889062,
+      "learning_rate": 0.0001,
+      "loss": 0.0061,
+      "num_input_tokens_seen": 228709232,
+      "step": 1551
+    },
+    {
+      "epoch": 0.44710291150187376,
+      "loss": 0.0021983864717185497,
+      "loss_ce": 0.00048463381244800985,
+      "loss_xval": 0.00171661376953125,
+      "num_input_tokens_seen": 228709232,
+      "step": 1551
+    },
+    {
+      "epoch": 0.4473911790141251,
+      "grad_norm": 5.5390335008117,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "num_input_tokens_seen": 228843976,
+      "step": 1552
+    },
+    {
+      "epoch": 0.4473911790141251,
+      "loss": 0.012496663257479668,
+      "loss_ce": 0.008344365283846855,
+      "loss_xval": 0.004150390625,
+      "num_input_tokens_seen": 228843976,
+      "step": 1552
+    },
+    {
+      "epoch": 0.4476794465263765,
+      "grad_norm": 1.0931289083866806,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "num_input_tokens_seen": 228979048,
+      "step": 1553
+    },
+    {
+      "epoch": 0.4476794465263765,
+      "loss": 0.0007365739438682795,
+      "loss_ce": 0.0003422295849304646,
+      "loss_xval": 0.0003948211669921875,
+      "num_input_tokens_seen": 228979048,
+      "step": 1553
+    },
+    {
+      "epoch": 0.44796771403862784,
+      "grad_norm": 4.597918279702102,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "num_input_tokens_seen": 229151512,
+      "step": 1554
+    },
+    {
+      "epoch": 0.44796771403862784,
+      "loss": 0.00347797479480505,
+      "loss_ce": 0.00044147565495222807,
+      "loss_xval": 0.0030364990234375,
+      "num_input_tokens_seen": 229151512,
+      "step": 1554
+    },
+    {
+      "epoch": 0.4482559815508792,
+      "grad_norm": 4.97960024521966,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "num_input_tokens_seen": 229286432,
+      "step": 1555
+    },
+    {
+      "epoch": 0.4482559815508792,
+      "loss": 0.011004405096173286,
+      "loss_ce": 0.007620769552886486,
+      "loss_xval": 0.003387451171875,
+      "num_input_tokens_seen": 229286432,
+      "step": 1555
+    },
+    {
+      "epoch": 0.44854424906313056,
+      "grad_norm": 0.2455556335516732,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "num_input_tokens_seen": 229421552,
+      "step": 1556
+    },
+    {
+      "epoch": 0.44854424906313056,
+      "loss": 0.0007927416008897126,
+      "loss_ce": 0.00031506994855590165,
+      "loss_xval": 0.000476837158203125,
+      "num_input_tokens_seen": 229421552,
+      "step": 1556
+    },
+    {
+      "epoch": 0.448832516575382,
+      "grad_norm": 5.320435855878716,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "num_input_tokens_seen": 229594152,
+      "step": 1557
+    },
+    {
+      "epoch": 0.448832516575382,
+      "loss": 0.004450153559446335,
+      "loss_ce": 0.0004847753734793514,
+      "loss_xval": 0.00396728515625,
+      "num_input_tokens_seen": 229594152,
+      "step": 1557
+    },
+    {
+      "epoch": 0.44912078408763334,
+      "grad_norm": 4.645618704016488,
+      "learning_rate": 0.0001,
+      "loss": 0.0067,
+      "num_input_tokens_seen": 229728976,
+      "step": 1558
+    },
+    {
+      "epoch": 0.44912078408763334,
+      "loss": 0.00939270481467247,
+      "loss_ce": 0.006438221782445908,
+      "loss_xval": 0.002960205078125,
+      "num_input_tokens_seen": 229728976,
+      "step": 1558
+    },
+    {
+      "epoch": 0.4494090515998847,
+      "grad_norm": 1.4425698536358162,
+      "learning_rate": 0.0001,
+      "loss": 0.0018,
+      "num_input_tokens_seen": 229864032,
+      "step": 1559
+    },
+    {
+      "epoch": 0.4494090515998847,
+      "loss": 0.0012257457710802555,
+      "loss_ce": 0.0005226494395174086,
+      "loss_xval": 0.000701904296875,
+      "num_input_tokens_seen": 229864032,
+      "step": 1559
+    },
+    {
+      "epoch": 0.44969731911213606,
+      "grad_norm": 6.097482104716617,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "num_input_tokens_seen": 230036728,
+      "step": 1560
+    },
+    {
+      "epoch": 0.44969731911213606,
+      "loss": 0.006726293824613094,
+      "loss_ce": 0.0020990660414099693,
+      "loss_xval": 0.004638671875,
+      "num_input_tokens_seen": 230036728,
+      "step": 1560
+    },
+    {
+      "epoch": 0.4499855866243874,
+      "grad_norm": 4.215032668485646,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "num_input_tokens_seen": 230171520,
+      "step": 1561
+    },
+    {
+      "epoch": 0.4499855866243874,
+      "loss": 0.012901779264211655,
+      "loss_ce": 0.009932037442922592,
+      "loss_xval": 0.0029754638671875,
+      "num_input_tokens_seen": 230171520,
+      "step": 1561
+    },
+    {
+      "epoch": 0.4502738541366388,
+      "grad_norm": 2.535015434197968,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "num_input_tokens_seen": 230306656,
+      "step": 1562
+    },
+    {
+      "epoch": 0.4502738541366388,
+      "loss": 0.0014059317763894796,
+      "loss_ce": 0.0003254187176935375,
+      "loss_xval": 0.0010833740234375,
+      "num_input_tokens_seen": 230306656,
+      "step": 1562
+    },
+    {
+      "epoch": 0.45056212164889015,
+      "grad_norm": 6.983310804340718,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "num_input_tokens_seen": 230479136,
+      "step": 1563
+    },
+    {
+      "epoch": 0.45056212164889015,
+      "loss": 0.007696349173784256,
+      "loss_ce": 0.0015699451323598623,
+      "loss_xval": 0.006134033203125,
+      "num_input_tokens_seen": 230479136,
+      "step": 1563
+    },
+    {
+      "epoch": 0.45085038916114156,
+      "grad_norm": 3.982794410049864,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "num_input_tokens_seen": 230613944,
+      "step": 1564
+    },
+    {
+      "epoch": 0.45085038916114156,
+      "loss": 0.011464044451713562,
+      "loss_ce": 0.008942529559135437,
+      "loss_xval": 0.0025177001953125,
+      "num_input_tokens_seen": 230613944,
+      "step": 1564
+    },
+    {
+      "epoch": 0.4511386566733929,
+      "grad_norm": 3.6182417808098513,
+      "learning_rate": 0.0001,
+      "loss": 0.0026,
+      "num_input_tokens_seen": 230748992,
+      "step": 1565
+    },
+    {
+      "epoch": 0.4511386566733929,
+      "loss": 0.0019969851709902287,
+      "loss_ce": 0.00036429459578357637,
+      "loss_xval": 0.0016326904296875,
+      "num_input_tokens_seen": 230748992,
+      "step": 1565
+    },
+    {
+      "epoch": 0.4514269241856443,
+      "grad_norm": 7.749495228699737,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "num_input_tokens_seen": 230921360,
+      "step": 1566
+    },
+    {
+      "epoch": 0.4514269241856443,
+      "loss": 0.00788209494203329,
+      "loss_ce": 0.0005044708959758282,
+      "loss_xval": 0.00738525390625,
+      "num_input_tokens_seen": 230921360,
+      "step": 1566
+    },
+    {
+      "epoch": 0.45171519169789565,
+      "grad_norm": 3.8659449472484058,
+      "learning_rate": 0.0001,
+      "loss": 0.0062,
+      "num_input_tokens_seen": 231056128,
+      "step": 1567
+    },
+    {
+      "epoch": 0.45171519169789565,
+      "loss": 0.0096367122605443,
+      "loss_ce": 0.00724871177226305,
+      "loss_xval": 0.00238037109375,
+      "num_input_tokens_seen": 231056128,
+      "step": 1567
+    },
+    {
+      "epoch": 0.452003459210147,
+      "grad_norm": 4.199783172254735,
+      "learning_rate": 0.0001,
+      "loss": 0.0058,
+      "num_input_tokens_seen": 231191184,
+      "step": 1568
+    },
+    {
+      "epoch": 0.452003459210147,
+      "loss": 0.007501209620386362,
+      "loss_ce": 0.005417431704699993,
+      "loss_xval": 0.0020904541015625,
+      "num_input_tokens_seen": 231191184,
+      "step": 1568
+    },
+    {
+      "epoch": 0.45229172672239837,
+      "grad_norm": 8.37342705145472,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "num_input_tokens_seen": 231363672,
+      "step": 1569
+    },
+    {
+      "epoch": 0.45229172672239837,
+      "loss": 0.010011442005634308,
+      "loss_ce": 0.0010507181286811829,
+      "loss_xval": 0.00897216796875,
+      "num_input_tokens_seen": 231363672,
+      "step": 1569
+    },
+    {
+      "epoch": 0.45257999423464973,
+      "grad_norm": 4.643924955363012,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "num_input_tokens_seen": 231498432,
+      "step": 1570
+    },
+    {
+      "epoch": 0.45257999423464973,
+      "loss": 0.012639187276363373,
+      "loss_ce": 0.008965633809566498,
+      "loss_xval": 0.0036773681640625,
+      "num_input_tokens_seen": 231498432,
+      "step": 1570
+    },
+    {
+      "epoch": 0.45286826174690115,
+      "grad_norm": 3.286366249617848,
+      "learning_rate": 0.0001,
+      "loss": 0.0025,
+      "num_input_tokens_seen": 231633632,
+      "step": 1571
+    },
+    {
+      "epoch": 0.45286826174690115,
+      "loss": 0.0012818737886846066,
+      "loss_ce": 0.0002752706059254706,
+      "loss_xval": 0.001007080078125,
+      "num_input_tokens_seen": 231633632,
+      "step": 1571
+    },
+    {
+      "epoch": 0.4531565292591525,
+      "grad_norm": 7.813638068618443,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "num_input_tokens_seen": 231806216,
+      "step": 1572
+    },
+    {
+      "epoch": 0.4531565292591525,
+      "loss": 0.00804380513727665,
+      "loss_ce": 0.0005097782704979181,
+      "loss_xval": 0.007537841796875,
+      "num_input_tokens_seen": 231806216,
+      "step": 1572
+    },
+    {
+      "epoch": 0.45344479677140387,
+      "grad_norm": 4.395192620509415,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "num_input_tokens_seen": 231940920,
+      "step": 1573
+    },
+    {
+      "epoch": 0.45344479677140387,
+      "loss": 0.01320941187441349,
+      "loss_ce": 0.009152481332421303,
+      "loss_xval": 0.004058837890625,
+      "num_input_tokens_seen": 231940920,
+      "step": 1573
+    },
+    {
+      "epoch": 0.45373306428365523,
+      "grad_norm": 3.5827965481902977,
+      "learning_rate": 0.0001,
+      "loss": 0.0031,
+      "num_input_tokens_seen": 232075888,
+      "step": 1574
+    },
+    {
+      "epoch": 0.45373306428365523,
+      "loss": 0.0021838760003447533,
+      "loss_ce": 0.0009717558859847486,
+      "loss_xval": 0.00121307373046875,
+      "num_input_tokens_seen": 232075888,
+      "step": 1574
+    },
+    {
+      "epoch": 0.4540213317959066,
+      "grad_norm": 7.978160524452476,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "num_input_tokens_seen": 232248472,
+      "step": 1575
+    },
+    {
+      "epoch": 0.4540213317959066,
+      "loss": 0.00975605845451355,
+      "loss_ce": 0.0014400180662050843,
+      "loss_xval": 0.00830078125,
+      "num_input_tokens_seen": 232248472,
+      "step": 1575
+    },
+    {
+      "epoch": 0.45430959930815795,
+      "grad_norm": 4.626883789668973,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "num_input_tokens_seen": 232383320,
+      "step": 1576
+    },
+    {
+      "epoch": 0.45430959930815795,
+      "loss": 0.014307888224720955,
+      "loss_ce": 0.009922893717885017,
+      "loss_xval": 0.00439453125,
+      "num_input_tokens_seen": 232383320,
+      "step": 1576
+    },
+    {
+      "epoch": 0.4545978668204093,
+      "grad_norm": 2.864667933219195,
+      "learning_rate": 0.0001,
+      "loss": 0.0034,
+      "num_input_tokens_seen": 232518432,
+      "step": 1577
+    },
+    {
+      "epoch": 0.4545978668204093,
+      "loss": 0.0009719881927594543,
+      "loss_ce": 0.00024481158470734954,
+      "loss_xval": 0.000728607177734375,
+      "num_input_tokens_seen": 232518432,
+      "step": 1577
+    },
+    {
+      "epoch": 0.45488613433266073,
+      "grad_norm": 7.476888287506308,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "num_input_tokens_seen": 232690968,
+      "step": 1578
+    },
+    {
+      "epoch": 0.45488613433266073,
+      "loss": 0.008270667865872383,
+      "loss_ce": 0.0009350047912448645,
+      "loss_xval": 0.00732421875,
+      "num_input_tokens_seen": 232690968,
+      "step": 1578
+    },
+    {
+      "epoch": 0.4551744018449121,
+      "grad_norm": 5.18413115965893,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "num_input_tokens_seen": 232825776,
+      "step": 1579
+    },
+    {
+      "epoch": 0.4551744018449121,
+      "loss": 0.014096993952989578,
+      "loss_ce": 0.008470315486192703,
+      "loss_xval": 0.005615234375,
+      "num_input_tokens_seen": 232825776,
+      "step": 1579
+    },
+    {
+      "epoch": 0.45546266935716345,
+      "grad_norm": 1.2851043654472376,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "num_input_tokens_seen": 232960856,
+      "step": 1580
+    },
+    {
+      "epoch": 0.45546266935716345,
+      "loss": 0.0006986982771195471,
+      "loss_ce": 0.00029136016382835805,
+      "loss_xval": 0.000408172607421875,
+      "num_input_tokens_seen": 232960856,
+      "step": 1580
+    },
+    {
+      "epoch": 0.4557509368694148,
+      "grad_norm": 5.719385694562323,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "num_input_tokens_seen": 233133512,
+      "step": 1581
+    },
+    {
+      "epoch": 0.4557509368694148,
+      "loss": 0.0053933728486299515,
+      "loss_ce": 0.000987397157587111,
+      "loss_xval": 0.00439453125,
+      "num_input_tokens_seen": 233133512,
+      "step": 1581
+    },
+    {
+      "epoch": 0.4560392043816662,
+      "grad_norm": 4.430066093964202,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "num_input_tokens_seen": 233268232,
+      "step": 1582
+    },
+    {
+      "epoch": 0.4560392043816662,
+      "loss": 0.013365356251597404,
+      "loss_ce": 0.008581725880503654,
+      "loss_xval": 0.004791259765625,
+      "num_input_tokens_seen": 233268232,
+      "step": 1582
+    },
+    {
+      "epoch": 0.45632747189391754,
+      "grad_norm": 0.44684700881540795,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "num_input_tokens_seen": 233403200,
+      "step": 1583
+    },
+    {
+      "epoch": 0.45632747189391754,
+      "loss": 0.0007035763701424003,
+      "loss_ce": 0.00021958664001431316,
+      "loss_xval": 0.000484466552734375,
+      "num_input_tokens_seen": 233403200,
+      "step": 1583
+    },
+    {
+      "epoch": 0.45661573940616895,
+      "grad_norm": 3.695879823804059,
+      "learning_rate": 0.0001,
+      "loss": 0.0041,
+      "num_input_tokens_seen": 233575744,
+      "step": 1584
+    },
+    {
+      "epoch": 0.45661573940616895,
+      "loss": 0.0021955205593258142,
+      "loss_ce": 0.0003329944738652557,
+      "loss_xval": 0.001861572265625,
+      "num_input_tokens_seen": 233575744,
+      "step": 1584
+    },
+    {
+      "epoch": 0.4569040069184203,
+      "grad_norm": 2.263990778477124,
+      "learning_rate": 0.0001,
+      "loss": 0.0069,
+      "num_input_tokens_seen": 233710616,
+      "step": 1585
+    },
+    {
+      "epoch": 0.4569040069184203,
+      "loss": 0.01318388245999813,
+      "loss_ce": 0.011165907606482506,
+      "loss_xval": 0.00201416015625,
+      "num_input_tokens_seen": 233710616,
+      "step": 1585
+    },
+    {
+      "epoch": 0.4571922744306717,
+      "grad_norm": 2.0982288678549565,
+      "learning_rate": 0.0001,
+      "loss": 0.0021,
+      "num_input_tokens_seen": 233845624,
+      "step": 1586
+    },
+    {
+      "epoch": 0.4571922744306717,
+      "loss": 0.0013415648136287928,
+      "loss_ce": 0.000979049364104867,
+      "loss_xval": 0.000362396240234375,
+      "num_input_tokens_seen": 233845624,
+      "step": 1586
+    },
+    {
+      "epoch": 0.45748054194292304,
+      "grad_norm": 4.895005433909848,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "num_input_tokens_seen": 234018112,
+      "step": 1587
+    },
+    {
+      "epoch": 0.45748054194292304,
+      "loss": 0.004058973863720894,
+      "loss_ce": 0.0008851455640979111,
+      "loss_xval": 0.003173828125,
+      "num_input_tokens_seen": 234018112,
+      "step": 1587
+    },
+    {
+      "epoch": 0.4577688094551744,
+      "grad_norm": 3.559117590151945,
+      "learning_rate": 0.0001,
+      "loss": 0.0059,
+      "num_input_tokens_seen": 234152976,
+      "step": 1588
+    },
+    {
+      "epoch": 0.4577688094551744,
+      "loss": 0.009978879243135452,
+      "loss_ce": 0.006557096727192402,
+      "loss_xval": 0.00341796875,
+      "num_input_tokens_seen": 234152976,
+      "step": 1588
+    },
+    {
+      "epoch": 0.45805707696742576,
+      "grad_norm": 0.49625808451973086,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "num_input_tokens_seen": 234288176,
+      "step": 1589
+    },
+    {
+      "epoch": 0.45805707696742576,
+      "loss": 0.0014246056089177728,
+      "loss_ce": 0.0008972237701527774,
+      "loss_xval": 0.00052642822265625,
+      "num_input_tokens_seen": 234288176,
+      "step": 1589
+    },
+    {
+      "epoch": 0.4583453444796771,
+      "grad_norm": 3.467829765242924,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "num_input_tokens_seen": 234460856,
+      "step": 1590
+    },
+    {
+      "epoch": 0.4583453444796771,
+      "loss": 0.002010901691392064,
+      "loss_ce": 0.00042875594226643443,
+      "loss_xval": 0.00157928466796875,
+      "num_input_tokens_seen": 234460856,
+      "step": 1590
+    },
+    {
+      "epoch": 0.45863361199192854,
+      "grad_norm": 3.129702725598525,
+      "learning_rate": 0.0001,
+      "loss": 0.0064,
+      "num_input_tokens_seen": 234595656,
+      "step": 1591
+    },
+    {
+      "epoch": 0.45863361199192854,
+      "loss": 0.011616665869951248,
+      "loss_ce": 0.00884147360920906,
+      "loss_xval": 0.002777099609375,
+      "num_input_tokens_seen": 234595656,
+      "step": 1591
+    },
+    {
+      "epoch": 0.4589218795041799,
+      "grad_norm": 0.33646227501719334,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "num_input_tokens_seen": 234730832,
+      "step": 1592
+    },
+    {
+      "epoch": 0.4589218795041799,
+      "loss": 0.0010132191237062216,
+      "loss_ce": 0.0002991555375047028,
+      "loss_xval": 0.000713348388671875,
+      "num_input_tokens_seen": 234730832,
+      "step": 1592
+    },
+    {
+      "epoch": 0.45921014701643126,
+      "grad_norm": 1.961299073545222,
+      "learning_rate": 0.0001,
+      "loss": 0.0045,
+      "num_input_tokens_seen": 234903184,
+      "step": 1593
+    },
+    {
+      "epoch": 0.45921014701643126,
+      "loss": 0.001006225822493434,
+      "loss_ce": 0.000321726081892848,
+      "loss_xval": 0.000682830810546875,
+      "num_input_tokens_seen": 234903184,
+      "step": 1593
+    },
+    {
+      "epoch": 0.4594984145286826,
+      "grad_norm": 1.721840997569964,
+      "learning_rate": 0.0001,
+      "loss": 0.0051,
+      "num_input_tokens_seen": 235038056,
+      "step": 1594
+    },
+    {
+      "epoch": 0.4594984145286826,
+      "loss": 0.009549458511173725,
+      "loss_ce": 0.008188565261662006,
+      "loss_xval": 0.0013580322265625,
+      "num_input_tokens_seen": 235038056,
+      "step": 1594
+    },
+    {
+      "epoch": 0.459786682040934,
+      "grad_norm": 0.17062183891564095,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "num_input_tokens_seen": 235173176,
+      "step": 1595
+    },
+    {
+      "epoch": 0.459786682040934,
+      "loss": 0.0006244688993319869,
+      "loss_ce": 0.00021963415201753378,
+      "loss_xval": 0.00040435791015625,
+      "num_input_tokens_seen": 235173176,
+      "step": 1595
+    },
+    {
+      "epoch": 0.46007494955318534,
+      "grad_norm": 1.483910756864113,
+      "learning_rate": 0.0001,
+      "loss": 0.0063,
+      "num_input_tokens_seen": 235345632,
+      "step": 1596
+    },
+    {
+      "epoch": 0.46007494955318534,
+      "loss": 0.0008567938930355012,
+      "loss_ce": 0.00030771593446843326,
+      "loss_xval": 0.00054931640625,
+      "num_input_tokens_seen": 235345632,
+      "step": 1596
+    },
+    {
+      "epoch": 0.4603632170654367,
+      "grad_norm": 0.7808873756743552,
+      "learning_rate": 0.0001,
+      "loss": 0.0061,
+      "num_input_tokens_seen": 235480456,
+      "step": 1597
+    },
+    {
+      "epoch": 0.4603632170654367,
+      "loss": 0.011714190244674683,
+      "loss_ce": 0.011104553937911987,
+      "loss_xval": 0.0006103515625,
+      "num_input_tokens_seen": 235480456,
+      "step": 1597
+    },
+    {
+      "epoch": 0.4606514845776881,
+      "grad_norm": 1.2412921623326705,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "num_input_tokens_seen": 235615480,
+      "step": 1598
+    },
+    {
+      "epoch": 0.4606514845776881,
+      "loss": 0.0004532962338998914,
+      "loss_ce": 0.00021571211982518435,
+      "loss_xval": 0.00023746490478515625,
+      "num_input_tokens_seen": 235615480,
+      "step": 1598
+    },
+    {
+      "epoch": 0.4609397520899395,
+      "grad_norm": 2.2899244520261948,
+      "learning_rate": 0.0001,
+      "loss": 0.0046,
+      "num_input_tokens_seen": 235788088,
+      "step": 1599
+    },
+    {
+      "epoch": 0.4609397520899395,
+      "loss": 0.001188235473819077,
+      "loss_ce": 0.0003819038684014231,
+      "loss_xval": 0.000804901123046875,
+      "num_input_tokens_seen": 235788088,
+      "step": 1599
+    },
+    {
+      "epoch": 0.46122801960219084,
+      "grad_norm": 1.2176522588627314,
+      "learning_rate": 0.0001,
+      "loss": 0.0049,
+      "num_input_tokens_seen": 235922920,
+      "step": 1600
+    },
+    {
+      "epoch": 0.46122801960219084,
+      "loss": 0.00936118047684431,
+      "loss_ce": 0.008414182811975479,
+      "loss_xval": 0.000946044921875,
+      "num_input_tokens_seen": 235922920,
+      "step": 1600
+    },
+    {
+      "epoch": 0.4615162871144422,
+      "grad_norm": 0.9907528345995467,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "num_input_tokens_seen": 236058200,
+      "step": 1601
+    },
+    {
+      "epoch": 0.4615162871144422,
+      "loss": 0.0004800662863999605,
+      "loss_ce": 0.00026364182122051716,
+      "loss_xval": 0.00021648406982421875,
+      "num_input_tokens_seen": 236058200,
+      "step": 1601
+    },
+    {
+      "epoch": 0.46180455462669356,
+      "grad_norm": 2.525567127516242,
+      "learning_rate": 0.0001,
+      "loss": 0.0058,
+      "num_input_tokens_seen": 236230672,
+      "step": 1602
+    },
+    {
+      "epoch": 0.46180455462669356,
+      "loss": 0.0014512970810756087,
+      "loss_ce": 0.000483317649923265,
+      "loss_xval": 0.00096893310546875,
+      "num_input_tokens_seen": 236230672,
+      "step": 1602
+    },
+    {
+      "epoch": 0.4620928221389449,
+      "grad_norm": 2.373056990094241,
+      "learning_rate": 0.0001,
+      "loss": 0.0044,
+      "num_input_tokens_seen": 236365440,
+      "step": 1603
+    },
+    {
+      "epoch": 0.4620928221389449,
+      "loss": 0.007948949001729488,
+      "loss_ce": 0.006201818119734526,
+      "loss_xval": 0.00174713134765625,
+      "num_input_tokens_seen": 236365440,
+      "step": 1603
+    },
+    {
+      "epoch": 0.4623810896511963,
+      "grad_norm": 1.3101942198197403,
+      "learning_rate": 0.0001,
+      "loss": 0.0018,
+      "num_input_tokens_seen": 236500472,
+      "step": 1604
+    },
+    {
+      "epoch": 0.4623810896511963,
+      "loss": 0.001278056064620614,
+      "loss_ce": 0.00020278830197639763,
+      "loss_xval": 0.00107574462890625,
+      "num_input_tokens_seen": 236500472,
+      "step": 1604
+    },
+    {
+      "epoch": 0.4626693571634477,
+      "grad_norm": 1.1630350641608882,
+      "learning_rate": 0.0001,
+      "loss": 0.0039,
+      "num_input_tokens_seen": 236672968,
+      "step": 1605
+    },
+    {
+      "epoch": 0.4626693571634477,
+      "loss": 0.0008471074397675693,
+      "loss_ce": 0.00029254582477733493,
+      "loss_xval": 0.000553131103515625,
+      "num_input_tokens_seen": 236672968,
+      "step": 1605
+    },
+    {
+      "epoch": 0.46295762467569906,
+      "grad_norm": 2.9590444115680916,
+      "learning_rate": 0.0001,
+      "loss": 0.0066,
+      "num_input_tokens_seen": 236807832,
+      "step": 1606
+    },
+    {
+      "epoch": 0.46295762467569906,
+      "loss": 0.010720960795879364,
+      "loss_ce": 0.009732477366924286,
+      "loss_xval": 0.0009918212890625,
+      "num_input_tokens_seen": 236807832,
+      "step": 1606
+    },
+    {
+      "epoch": 0.4632458921879504,
+      "grad_norm": 6.278245676438151,
+      "learning_rate": 0.0001,
+      "loss": 0.0063,
+      "num_input_tokens_seen": 236942912,
+      "step": 1607
+    },
+    {
+      "epoch": 0.4632458921879504,
+      "loss": 0.004414130933582783,
+      "loss_ce": 0.00018935362459160388,
+      "loss_xval": 0.00421142578125,
+      "num_input_tokens_seen": 236942912,
+      "step": 1607
+    },
+    {
+      "epoch": 0.4635341597002018,
+      "grad_norm": 10.105785581159694,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "num_input_tokens_seen": 237115424,
+      "step": 1608
+    },
+    {
+      "epoch": 0.4635341597002018,
+      "loss": 0.014623086899518967,
+      "loss_ce": 0.0003790073096752167,
+      "loss_xval": 0.01422119140625,
+      "num_input_tokens_seen": 237115424,
+      "step": 1608
+    },
+    {
+      "epoch": 0.46382242721245315,
+      "grad_norm": 14.24392109551276,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "num_input_tokens_seen": 237250240,
+      "step": 1609
+    },
+    {
+      "epoch": 0.46382242721245315,
+      "loss": 0.04370329529047012,
+      "loss_ce": 0.011598804034292698,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 237250240,
+      "step": 1609
+    },
+    {
+      "epoch": 0.4641106947247045,
+      "grad_norm": 19.783437624937637,
+      "learning_rate": 0.0001,
+      "loss": 0.056,
+      "num_input_tokens_seen": 237385408,
+      "step": 1610
+    },
+    {
+      "epoch": 0.4641106947247045,
+      "loss": 0.062287457287311554,
+      "loss_ce": 0.00021470512729138136,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 237385408,
+      "step": 1610
+    },
+    {
+      "epoch": 0.46439896223695587,
+      "grad_norm": 27.140839690327642,
+      "learning_rate": 0.0001,
+      "loss": 0.1082,
+      "num_input_tokens_seen": 237557888,
+      "step": 1611
+    },
+    {
+      "epoch": 0.46439896223695587,
+      "loss": 0.10900463163852692,
+      "loss_ce": 0.0005451533943414688,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 237557888,
+      "step": 1611
+    },
+    {
+      "epoch": 0.4646872297492073,
+      "grad_norm": 31.234951537714885,
+      "learning_rate": 0.0001,
+      "loss": 0.1447,
+      "num_input_tokens_seen": 237692728,
+      "step": 1612
+    },
+    {
+      "epoch": 0.4646872297492073,
+      "loss": 0.14542171359062195,
+      "loss_ce": 0.008214689791202545,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 237692728,
+      "step": 1612
+    },
+    {
+      "epoch": 0.46497549726145865,
+      "grad_norm": 23.55866477887437,
+      "learning_rate": 0.0001,
+      "loss": 0.083,
+      "num_input_tokens_seen": 237827928,
+      "step": 1613
+    },
+    {
+      "epoch": 0.46497549726145865,
+      "loss": 0.07829268276691437,
+      "loss_ce": 0.0005338959745131433,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 237827928,
+      "step": 1613
+    },
+    {
+      "epoch": 0.46526376477371,
+      "grad_norm": 3.971405211022664,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "num_input_tokens_seen": 238000440,
+      "step": 1614
+    },
+    {
+      "epoch": 0.46526376477371,
+      "loss": 0.0032971161417663097,
+      "loss_ce": 0.0005486266454681754,
+      "loss_xval": 0.00274658203125,
+      "num_input_tokens_seen": 238000440,
+      "step": 1614
+    },
+    {
+      "epoch": 0.46555203228596137,
+      "grad_norm": 16.527875696120834,
+      "learning_rate": 0.0001,
+      "loss": 0.0465,
+      "num_input_tokens_seen": 238135312,
+      "step": 1615
+    },
+    {
+      "epoch": 0.46555203228596137,
+      "loss": 0.051238447427749634,
+      "loss_ce": 0.007415205705910921,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 238135312,
+      "step": 1615
+    },
+    {
+      "epoch": 0.46584029979821273,
+      "grad_norm": 22.9706387506555,
+      "learning_rate": 0.0001,
+      "loss": 0.0801,
+      "num_input_tokens_seen": 238270408,
+      "step": 1616
+    },
+    {
+      "epoch": 0.46584029979821273,
+      "loss": 0.07985322177410126,
+      "loss_ce": 0.0002633807889651507,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 238270408,
+      "step": 1616
+    },
+    {
+      "epoch": 0.4661285673104641,
+      "grad_norm": 10.81122800922098,
+      "learning_rate": 0.0001,
+      "loss": 0.0243,
+      "num_input_tokens_seen": 238442976,
+      "step": 1617
+    },
+    {
+      "epoch": 0.4661285673104641,
+      "loss": 0.021433737128973007,
+      "loss_ce": 0.002375510986894369,
+      "loss_xval": 0.01904296875,
+      "num_input_tokens_seen": 238442976,
+      "step": 1617
+    },
+    {
+      "epoch": 0.46641683482271545,
+      "grad_norm": 9.549291981727233,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "num_input_tokens_seen": 238577672,
+      "step": 1618
+    },
+    {
+      "epoch": 0.46641683482271545,
+      "loss": 0.023982558399438858,
+      "loss_ce": 0.010119949467480183,
+      "loss_xval": 0.01385498046875,
+      "num_input_tokens_seen": 238577672,
+      "step": 1618
+    },
+    {
+      "epoch": 0.46670510233496687,
+      "grad_norm": 18.65749974230963,
+      "learning_rate": 0.0001,
+      "loss": 0.055,
+      "num_input_tokens_seen": 238712600,
+      "step": 1619
+    },
+    {
+      "epoch": 0.46670510233496687,
+      "loss": 0.05401700735092163,
+      "loss_ce": 0.0002755502355284989,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 238712600,
+      "step": 1619
+    },
+    {
+      "epoch": 0.46699336984721823,
+      "grad_norm": 7.856487019792692,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "num_input_tokens_seen": 238885080,
+      "step": 1620
+    },
+    {
+      "epoch": 0.46699336984721823,
+      "loss": 0.012627718970179558,
+      "loss_ce": 0.000374911876861006,
+      "loss_xval": 0.01226806640625,
+      "num_input_tokens_seen": 238885080,
+      "step": 1620
+    },
+    {
+      "epoch": 0.4672816373594696,
+      "grad_norm": 11.188000000786419,
+      "learning_rate": 0.0001,
+      "loss": 0.0247,
+      "num_input_tokens_seen": 239019832,
+      "step": 1621
+    },
+    {
+      "epoch": 0.4672816373594696,
+      "loss": 0.025063853710889816,
+      "loss_ce": 0.007562023587524891,
+      "loss_xval": 0.0174560546875,
+      "num_input_tokens_seen": 239019832,
+      "step": 1621
+    },
+    {
+      "epoch": 0.46756990487172095,
+      "grad_norm": 16.834899974492362,
+      "learning_rate": 0.0001,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 239154848,
+      "step": 1622
+    },
+    {
+      "epoch": 0.46756990487172095,
+      "loss": 0.045431505888700485,
+      "loss_ce": 0.0006011828663758934,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 239154848,
+      "step": 1622
+    },
+    {
+      "epoch": 0.4678581723839723,
+      "grad_norm": 2.7871025358915906,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "num_input_tokens_seen": 239327320,
+      "step": 1623
+    },
+    {
+      "epoch": 0.4678581723839723,
+      "loss": 0.008208648301661015,
+      "loss_ce": 0.00468577491119504,
+      "loss_xval": 0.0035247802734375,
+      "num_input_tokens_seen": 239327320,
+      "step": 1623
+    },
+    {
+      "epoch": 0.4681464398962237,
+      "grad_norm": 13.562232424207437,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 239462112,
+      "step": 1624
+    },
+    {
+      "epoch": 0.4681464398962237,
+      "loss": 0.033323727548122406,
+      "loss_ce": 0.010069331154227257,
+      "loss_xval": 0.023193359375,
+      "num_input_tokens_seen": 239462112,
+      "step": 1624
+    },
+    {
+      "epoch": 0.46843470740847504,
+      "grad_norm": 11.587959949993882,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "num_input_tokens_seen": 239597232,
+      "step": 1625
+    },
+    {
+      "epoch": 0.46843470740847504,
+      "loss": 0.02408481389284134,
+      "loss_ce": 0.00032687990460544825,
+      "loss_xval": 0.0238037109375,
+      "num_input_tokens_seen": 239597232,
+      "step": 1625
+    },
+    {
+      "epoch": 0.46872297492072645,
+      "grad_norm": 5.998804309422722,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "num_input_tokens_seen": 239769608,
+      "step": 1626
+    },
+    {
+      "epoch": 0.46872297492072645,
+      "loss": 0.003973469138145447,
+      "loss_ce": 0.0004830213147215545,
+      "loss_xval": 0.0034942626953125,
+      "num_input_tokens_seen": 239769608,
+      "step": 1626
+    },
+    {
+      "epoch": 0.4690112424329778,
+      "grad_norm": 16.26213066494406,
+      "learning_rate": 0.0001,
+      "loss": 0.05,
+      "num_input_tokens_seen": 239904400,
+      "step": 1627
+    },
+    {
+      "epoch": 0.4690112424329778,
+      "loss": 0.044366300106048584,
+      "loss_ce": 0.00798934418708086,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 239904400,
+      "step": 1627
+    },
+    {
+      "epoch": 0.4692995099452292,
+      "grad_norm": 4.206662652807161,
+      "learning_rate": 0.0001,
+      "loss": 0.0064,
+      "num_input_tokens_seen": 240039488,
+      "step": 1628
+    },
+    {
+      "epoch": 0.4692995099452292,
+      "loss": 0.005551893264055252,
+      "loss_ce": 0.00031812855741009116,
+      "loss_xval": 0.0052490234375,
+      "num_input_tokens_seen": 240039488,
+      "step": 1628
+    },
+    {
+      "epoch": 0.46958777745748054,
+      "grad_norm": 16.1538880969112,
+      "learning_rate": 0.0001,
+      "loss": 0.0509,
+      "num_input_tokens_seen": 240211976,
+      "step": 1629
+    },
+    {
+      "epoch": 0.46958777745748054,
+      "loss": 0.029739946126937866,
+      "loss_ce": 0.0005498826503753662,
+      "loss_xval": 0.0291748046875,
+      "num_input_tokens_seen": 240211976,
+      "step": 1629
+    },
+    {
+      "epoch": 0.4698760449697319,
+      "grad_norm": 17.30655354392214,
+      "learning_rate": 0.0001,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 240346792,
+      "step": 1630
+    },
+    {
+      "epoch": 0.4698760449697319,
+      "loss": 0.051143795251846313,
+      "loss_ce": 0.007106929086148739,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 240346792,
+      "step": 1630
+    },
+    {
+      "epoch": 0.47016431248198326,
+      "grad_norm": 4.871710309339203,
+      "learning_rate": 0.0001,
+      "loss": 0.0058,
+      "num_input_tokens_seen": 240481968,
+      "step": 1631
+    },
+    {
+      "epoch": 0.47016431248198326,
+      "loss": 0.0038508400321006775,
+      "loss_ce": 0.0004004462971352041,
+      "loss_xval": 0.003448486328125,
+      "num_input_tokens_seen": 240481968,
+      "step": 1631
+    },
+    {
+      "epoch": 0.4704525799942347,
+      "grad_norm": 22.22578398457069,
+      "learning_rate": 0.0001,
+      "loss": 0.0886,
+      "num_input_tokens_seen": 240654584,
+      "step": 1632
+    },
+    {
+      "epoch": 0.4704525799942347,
+      "loss": 0.06573119014501572,
+      "loss_ce": 0.00039305369136855006,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 240654584,
+      "step": 1632
+    },
+    {
+      "epoch": 0.47074084750648604,
+      "grad_norm": 10.295121901232179,
+      "learning_rate": 0.0001,
+      "loss": 0.0256,
+      "num_input_tokens_seen": 240789440,
+      "step": 1633
+    },
+    {
+      "epoch": 0.47074084750648604,
+      "loss": 0.02956472337245941,
+      "loss_ce": 0.01014028675854206,
+      "loss_xval": 0.0194091796875,
+      "num_input_tokens_seen": 240789440,
+      "step": 1633
+    },
+    {
+      "epoch": 0.4710291150187374,
+      "grad_norm": 14.965752387695096,
+      "learning_rate": 0.0001,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 240924440,
+      "step": 1634
+    },
+    {
+      "epoch": 0.4710291150187374,
+      "loss": 0.03500206023454666,
+      "loss_ce": 0.00042564034811221063,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 240924440,
+      "step": 1634
+    },
+    {
+      "epoch": 0.47131738253098876,
+      "grad_norm": 17.858622605015228,
+      "learning_rate": 0.0001,
+      "loss": 0.0651,
+      "num_input_tokens_seen": 241096856,
+      "step": 1635
+    },
+    {
+      "epoch": 0.47131738253098876,
+      "loss": 0.05295751243829727,
+      "loss_ce": 0.00037572276778519154,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 241096856,
+      "step": 1635
+    },
+    {
+      "epoch": 0.4716056500432401,
+      "grad_norm": 4.614423817877169,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "num_input_tokens_seen": 241231656,
+      "step": 1636
+    },
+    {
+      "epoch": 0.4716056500432401,
+      "loss": 0.014617221429944038,
+      "loss_ce": 0.008631961420178413,
+      "loss_xval": 0.0059814453125,
+      "num_input_tokens_seen": 241231656,
+      "step": 1636
+    },
+    {
+      "epoch": 0.4718939175554915,
+      "grad_norm": 19.190589867557332,
+      "learning_rate": 0.0001,
+      "loss": 0.0701,
+      "num_input_tokens_seen": 241366888,
+      "step": 1637
+    },
+    {
+      "epoch": 0.4718939175554915,
+      "loss": 0.06205695495009422,
+      "loss_ce": 0.0008692118572071195,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 241366888,
+      "step": 1637
+    },
+    {
+      "epoch": 0.47218218506774284,
+      "grad_norm": 4.293632420630124,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "num_input_tokens_seen": 241539280,
+      "step": 1638
+    },
+    {
+      "epoch": 0.47218218506774284,
+      "loss": 0.005224054679274559,
+      "loss_ce": 0.0013445076765492558,
+      "loss_xval": 0.003875732421875,
+      "num_input_tokens_seen": 241539280,
+      "step": 1638
+    },
+    {
+      "epoch": 0.47247045257999426,
+      "grad_norm": 17.120904008588155,
+      "learning_rate": 0.0001,
+      "loss": 0.0587,
+      "num_input_tokens_seen": 241674192,
+      "step": 1639
+    },
+    {
+      "epoch": 0.47247045257999426,
+      "loss": 0.058812811970710754,
+      "loss_ce": 0.007817939855158329,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 241674192,
+      "step": 1639
+    },
+    {
+      "epoch": 0.4727587200922456,
+      "grad_norm": 12.678106685232084,
+      "learning_rate": 0.0001,
+      "loss": 0.0325,
+      "num_input_tokens_seen": 241809352,
+      "step": 1640
+    },
+    {
+      "epoch": 0.4727587200922456,
+      "loss": 0.02766924723982811,
+      "loss_ce": 0.00029497960349544883,
+      "loss_xval": 0.02734375,
+      "num_input_tokens_seen": 241809352,
+      "step": 1640
+    },
+    {
+      "epoch": 0.473046987604497,
+      "grad_norm": 10.072616136148826,
+      "learning_rate": 0.0001,
+      "loss": 0.0232,
+      "num_input_tokens_seen": 241981848,
+      "step": 1641
+    },
+    {
+      "epoch": 0.473046987604497,
+      "loss": 0.01985846646130085,
+      "loss_ce": 0.0006323920097202063,
+      "loss_xval": 0.019287109375,
+      "num_input_tokens_seen": 241981848,
+      "step": 1641
+    },
+    {
+      "epoch": 0.47333525511674834,
+      "grad_norm": 16.424813937686693,
+      "learning_rate": 0.0001,
+      "loss": 0.056,
+      "num_input_tokens_seen": 242116624,
+      "step": 1642
+    },
+    {
+      "epoch": 0.47333525511674834,
+      "loss": 0.05718439444899559,
+      "loss_ce": 0.00887506827712059,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 242116624,
+      "step": 1642
+    },
+    {
+      "epoch": 0.4736235226289997,
+      "grad_norm": 1.9623067562963932,
+      "learning_rate": 0.0001,
+      "loss": 0.002,
+      "num_input_tokens_seen": 242251720,
+      "step": 1643
+    },
+    {
+      "epoch": 0.4736235226289997,
+      "loss": 0.002029548864811659,
+      "loss_ce": 0.0002747882972471416,
+      "loss_xval": 0.0017547607421875,
+      "num_input_tokens_seen": 242251720,
+      "step": 1643
+    },
+    {
+      "epoch": 0.47391179014125107,
+      "grad_norm": 15.945841150302297,
+      "learning_rate": 0.0001,
+      "loss": 0.0542,
+      "num_input_tokens_seen": 242424312,
+      "step": 1644
+    },
+    {
+      "epoch": 0.47391179014125107,
+      "loss": 0.04716937243938446,
+      "loss_ce": 0.0005079961847513914,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 242424312,
+      "step": 1644
+    },
+    {
+      "epoch": 0.4742000576535024,
+      "grad_norm": 4.53648202392857,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "num_input_tokens_seen": 242559168,
+      "step": 1645
+    },
+    {
+      "epoch": 0.4742000576535024,
+      "loss": 0.01110774278640747,
+      "loss_ce": 0.006743729114532471,
+      "loss_xval": 0.004364013671875,
+      "num_input_tokens_seen": 242559168,
+      "step": 1645
+    },
+    {
+      "epoch": 0.47448832516575384,
+      "grad_norm": 13.462705008051305,
+      "learning_rate": 0.0001,
+      "loss": 0.0368,
+      "num_input_tokens_seen": 242694168,
+      "step": 1646
+    },
+    {
+      "epoch": 0.47448832516575384,
+      "loss": 0.03658619150519371,
+      "loss_ce": 0.00027027411852031946,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 242694168,
+      "step": 1646
+    },
+    {
+      "epoch": 0.4747765926780052,
+      "grad_norm": 10.000124083236205,
+      "learning_rate": 0.0001,
+      "loss": 0.0252,
+      "num_input_tokens_seen": 242866584,
+      "step": 1647
+    },
+    {
+      "epoch": 0.4747765926780052,
+      "loss": 0.018924858421087265,
+      "loss_ce": 0.0005990548525005579,
+      "loss_xval": 0.018310546875,
+      "num_input_tokens_seen": 242866584,
+      "step": 1647
+    },
+    {
+      "epoch": 0.47506486019025657,
+      "grad_norm": 8.050537914125309,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "num_input_tokens_seen": 243001280,
+      "step": 1648
+    },
+    {
+      "epoch": 0.47506486019025657,
+      "loss": 0.020638510584831238,
+      "loss_ce": 0.005127951968461275,
+      "loss_xval": 0.0155029296875,
+      "num_input_tokens_seen": 243001280,
+      "step": 1648
+    },
+    {
+      "epoch": 0.4753531277025079,
+      "grad_norm": 12.371986608475902,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "num_input_tokens_seen": 243136352,
+      "step": 1649
+    },
+    {
+      "epoch": 0.4753531277025079,
+      "loss": 0.03136796876788139,
+      "loss_ce": 0.0009114270796999335,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 243136352,
+      "step": 1649
+    },
+    {
+      "epoch": 0.4756413952147593,
+      "grad_norm": 1.410721305972707,
+      "learning_rate": 0.0001,
+      "loss": 0.0055,
+      "num_input_tokens_seen": 243308768,
+      "step": 1650
+    },
+    {
+      "epoch": 0.4756413952147593,
+      "loss": 0.0017137866234406829,
+      "loss_ce": 0.0004358630394563079,
+      "loss_xval": 0.00128173828125,
+      "num_input_tokens_seen": 243308768,
+      "step": 1650
+    },
+    {
+      "epoch": 0.47592966272701065,
+      "grad_norm": 10.94040115444206,
+      "learning_rate": 0.0001,
+      "loss": 0.0276,
+      "num_input_tokens_seen": 243443504,
+      "step": 1651
+    },
+    {
+      "epoch": 0.47592966272701065,
+      "loss": 0.030689680948853493,
+      "loss_ce": 0.005024399608373642,
+      "loss_xval": 0.025634765625,
+      "num_input_tokens_seen": 243443504,
+      "step": 1651
+    },
+    {
+      "epoch": 0.476217930239262,
+      "grad_norm": 2.230035570569663,
+      "learning_rate": 0.0001,
+      "loss": 0.0018,
+      "num_input_tokens_seen": 243578616,
+      "step": 1652
+    },
+    {
+      "epoch": 0.476217930239262,
+      "loss": 0.001764374552294612,
+      "loss_ce": 0.00029666972113773227,
+      "loss_xval": 0.00146484375,
+      "num_input_tokens_seen": 243578616,
+      "step": 1652
+    },
+    {
+      "epoch": 0.4765061977515134,
+      "grad_norm": 9.022289083119624,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "num_input_tokens_seen": 243751184,
+      "step": 1653
+    },
+    {
+      "epoch": 0.4765061977515134,
+      "loss": 0.018094517290592194,
+      "loss_ce": 0.0011114849476143718,
+      "loss_xval": 0.0169677734375,
+      "num_input_tokens_seen": 243751184,
+      "step": 1653
+    },
+    {
+      "epoch": 0.4767944652637648,
+      "grad_norm": 4.739831753216289,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "num_input_tokens_seen": 243886016,
+      "step": 1654
+    },
+    {
+      "epoch": 0.4767944652637648,
+      "loss": 0.012720847502350807,
+      "loss_ce": 0.007132316008210182,
+      "loss_xval": 0.005584716796875,
+      "num_input_tokens_seen": 243886016,
+      "step": 1654
+    },
+    {
+      "epoch": 0.47708273277601615,
+      "grad_norm": 6.834353804413808,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "num_input_tokens_seen": 244021200,
+      "step": 1655
+    },
+    {
+      "epoch": 0.47708273277601615,
+      "loss": 0.009534979239106178,
+      "loss_ce": 0.0003110412508249283,
+      "loss_xval": 0.00921630859375,
+      "num_input_tokens_seen": 244021200,
+      "step": 1655
+    },
+    {
+      "epoch": 0.4773710002882675,
+      "grad_norm": 6.508606064768901,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "num_input_tokens_seen": 244193648,
+      "step": 1656
+    },
+    {
+      "epoch": 0.4773710002882675,
+      "loss": 0.008853845298290253,
+      "loss_ce": 0.0007209099130704999,
+      "loss_xval": 0.00811767578125,
+      "num_input_tokens_seen": 244193648,
+      "step": 1656
+    },
+    {
+      "epoch": 0.47765926780051887,
+      "grad_norm": 4.791951822258596,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "num_input_tokens_seen": 244328336,
+      "step": 1657
+    },
+    {
+      "epoch": 0.47765926780051887,
+      "loss": 0.015914246439933777,
+      "loss_ce": 0.010794922709465027,
+      "loss_xval": 0.005126953125,
+      "num_input_tokens_seen": 244328336,
+      "step": 1657
+    },
+    {
+      "epoch": 0.47794753531277023,
+      "grad_norm": 8.518485511752413,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "num_input_tokens_seen": 244463576,
+      "step": 1658
+    },
+    {
+      "epoch": 0.47794753531277023,
+      "loss": 0.013599589467048645,
+      "loss_ce": 0.00042362435488030314,
+      "loss_xval": 0.01318359375,
+      "num_input_tokens_seen": 244463576,
+      "step": 1658
+    },
+    {
+      "epoch": 0.4782358028250216,
+      "grad_norm": 1.213327411200081,
+      "learning_rate": 0.0001,
+      "loss": 0.0059,
+      "num_input_tokens_seen": 244636112,
+      "step": 1659
+    },
+    {
+      "epoch": 0.4782358028250216,
+      "loss": 0.002218852750957012,
+      "loss_ce": 0.000748763675801456,
+      "loss_xval": 0.00147247314453125,
+      "num_input_tokens_seen": 244636112,
+      "step": 1659
+    },
+    {
+      "epoch": 0.478524070337273,
+      "grad_norm": 9.181071206460544,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "num_input_tokens_seen": 244770920,
+      "step": 1660
+    },
+    {
+      "epoch": 0.478524070337273,
+      "loss": 0.02432107925415039,
+      "loss_ce": 0.007727145683020353,
+      "loss_xval": 0.0166015625,
+      "num_input_tokens_seen": 244770920,
+      "step": 1660
+    },
+    {
+      "epoch": 0.47881233784952437,
+      "grad_norm": 3.4061375786237496,
+      "learning_rate": 0.0001,
+      "loss": 0.0037,
+      "num_input_tokens_seen": 244906072,
+      "step": 1661
+    },
+    {
+      "epoch": 0.47881233784952437,
+      "loss": 0.002656823955476284,
+      "loss_ce": 0.000368005596101284,
+      "loss_xval": 0.002288818359375,
+      "num_input_tokens_seen": 244906072,
+      "step": 1661
+    },
+    {
+      "epoch": 0.47910060536177573,
+      "grad_norm": 6.917202408403764,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "num_input_tokens_seen": 245078544,
+      "step": 1662
+    },
+    {
+      "epoch": 0.47910060536177573,
+      "loss": 0.012157600373029709,
+      "loss_ce": 0.000423592166043818,
+      "loss_xval": 0.01171875,
+      "num_input_tokens_seen": 245078544,
+      "step": 1662
+    },
+    {
+      "epoch": 0.4793888728740271,
+      "grad_norm": 6.442072131706248,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "num_input_tokens_seen": 245213312,
+      "step": 1663
+    },
+    {
+      "epoch": 0.4793888728740271,
+      "loss": 0.01970660500228405,
+      "loss_ce": 0.01142108254134655,
+      "loss_xval": 0.00830078125,
+      "num_input_tokens_seen": 245213312,
+      "step": 1663
+    },
+    {
+      "epoch": 0.47967714038627846,
+      "grad_norm": 3.332300474485623,
+      "learning_rate": 0.0001,
+      "loss": 0.0032,
+      "num_input_tokens_seen": 245348296,
+      "step": 1664
+    },
+    {
+      "epoch": 0.47967714038627846,
+      "loss": 0.0035281481686979532,
+      "loss_ce": 0.0002932848874479532,
+      "loss_xval": 0.00323486328125,
+      "num_input_tokens_seen": 245348296,
+      "step": 1664
+    },
+    {
+      "epoch": 0.4799654078985298,
+      "grad_norm": 7.695169704124381,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "num_input_tokens_seen": 245520760,
+      "step": 1665
+    },
+    {
+      "epoch": 0.4799654078985298,
+      "loss": 0.013785162940621376,
+      "loss_ce": 0.0004032054857816547,
+      "loss_xval": 0.01336669921875,
+      "num_input_tokens_seen": 245520760,
+      "step": 1665
+    },
+    {
+      "epoch": 0.4802536754107812,
+      "grad_norm": 0.8855467303934494,
+      "learning_rate": 0.0001,
+      "loss": 0.0051,
+      "num_input_tokens_seen": 245655544,
+      "step": 1666
+    },
+    {
+      "epoch": 0.4802536754107812,
+      "loss": 0.009379335679113865,
+      "loss_ce": 0.008762546814978123,
+      "loss_xval": 0.00061798095703125,
+      "num_input_tokens_seen": 245655544,
+      "step": 1666
+    },
+    {
+      "epoch": 0.4805419429230326,
+      "grad_norm": 6.393784109798928,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "num_input_tokens_seen": 245790752,
+      "step": 1667
+    },
+    {
+      "epoch": 0.4805419429230326,
+      "loss": 0.009521054103970528,
+      "loss_ce": 0.0002742270880844444,
+      "loss_xval": 0.00927734375,
+      "num_input_tokens_seen": 245790752,
+      "step": 1667
+    },
+    {
+      "epoch": 0.48083021043528396,
+      "grad_norm": 3.936996774178376,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "num_input_tokens_seen": 245963240,
+      "step": 1668
+    },
+    {
+      "epoch": 0.48083021043528396,
+      "loss": 0.004050452262163162,
+      "loss_ce": 0.00040741637349128723,
+      "loss_xval": 0.0036468505859375,
+      "num_input_tokens_seen": 245963240,
+      "step": 1668
+    },
+    {
+      "epoch": 0.4811184779475353,
+      "grad_norm": 3.686948208785727,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "num_input_tokens_seen": 246098016,
+      "step": 1669
+    },
+    {
+      "epoch": 0.4811184779475353,
+      "loss": 0.011370341293513775,
+      "loss_ce": 0.007866541855037212,
+      "loss_xval": 0.003509521484375,
+      "num_input_tokens_seen": 246098016,
+      "step": 1669
+    },
+    {
+      "epoch": 0.4814067454597867,
+      "grad_norm": 5.263022503025478,
+      "learning_rate": 0.0001,
+      "loss": 0.0066,
+      "num_input_tokens_seen": 246232992,
+      "step": 1670
+    },
+    {
+      "epoch": 0.4814067454597867,
+      "loss": 0.006171601824462414,
+      "loss_ce": 0.00024737673811614513,
+      "loss_xval": 0.00592041015625,
+      "num_input_tokens_seen": 246232992,
+      "step": 1670
+    },
+    {
+      "epoch": 0.48169501297203804,
+      "grad_norm": 0.8220510945207464,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "num_input_tokens_seen": 246405480,
+      "step": 1671
+    },
+    {
+      "epoch": 0.48169501297203804,
+      "loss": 0.0007558593060821295,
+      "loss_ce": 0.00026328652165830135,
+      "loss_xval": 0.000492095947265625,
+      "num_input_tokens_seen": 246405480,
+      "step": 1671
+    },
+    {
+      "epoch": 0.4819832804842894,
+      "grad_norm": 5.489817625132127,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "num_input_tokens_seen": 246540264,
+      "step": 1672
+    },
+    {
+      "epoch": 0.4819832804842894,
+      "loss": 0.014926630072295666,
+      "loss_ce": 0.007903773337602615,
+      "loss_xval": 0.00701904296875,
+      "num_input_tokens_seen": 246540264,
+      "step": 1672
+    },
+    {
+      "epoch": 0.4822715479965408,
+      "grad_norm": 2.518344153341866,
+      "learning_rate": 0.0001,
+      "loss": 0.0043,
+      "num_input_tokens_seen": 246675400,
+      "step": 1673
+    },
+    {
+      "epoch": 0.4822715479965408,
+      "loss": 0.005512160249054432,
+      "loss_ce": 0.004024427849799395,
+      "loss_xval": 0.00148773193359375,
+      "num_input_tokens_seen": 246675400,
+      "step": 1673
+    },
+    {
+      "epoch": 0.4825598155087922,
+      "grad_norm": 3.545674469895499,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "num_input_tokens_seen": 246847800,
+      "step": 1674
+    },
+    {
+      "epoch": 0.4825598155087922,
+      "loss": 0.003581702709197998,
+      "loss_ce": 0.00027626747032627463,
+      "loss_xval": 0.0033111572265625,
+      "num_input_tokens_seen": 246847800,
+      "step": 1674
+    },
+    {
+      "epoch": 0.48284808302104354,
+      "grad_norm": 4.677097197848902,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "num_input_tokens_seen": 246982656,
+      "step": 1675
+    },
+    {
+      "epoch": 0.48284808302104354,
+      "loss": 0.014951646327972412,
+      "loss_ce": 0.009744584560394287,
+      "loss_xval": 0.005218505859375,
+      "num_input_tokens_seen": 246982656,
+      "step": 1675
+    },
+    {
+      "epoch": 0.4831363505332949,
+      "grad_norm": 0.06732599670808967,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "num_input_tokens_seen": 247117624,
+      "step": 1676
+    },
+    {
+      "epoch": 0.4831363505332949,
+      "loss": 0.0005512872594408691,
+      "loss_ce": 0.00023967419110704213,
+      "loss_xval": 0.0003108978271484375,
+      "num_input_tokens_seen": 247117624,
+      "step": 1676
+    },
+    {
+      "epoch": 0.48342461804554626,
+      "grad_norm": 4.186459031367488,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "num_input_tokens_seen": 247290104,
+      "step": 1677
+    },
+    {
+      "epoch": 0.48342461804554626,
+      "loss": 0.004762859083712101,
+      "loss_ce": 0.0006620598142035306,
+      "loss_xval": 0.00408935546875,
+      "num_input_tokens_seen": 247290104,
+      "step": 1677
+    },
+    {
+      "epoch": 0.4837128855577976,
+      "grad_norm": 2.814617904179985,
+      "learning_rate": 0.0001,
+      "loss": 0.0069,
+      "num_input_tokens_seen": 247424832,
+      "step": 1678
+    },
+    {
+      "epoch": 0.4837128855577976,
+      "loss": 0.011508513242006302,
+      "loss_ce": 0.00955061987042427,
+      "loss_xval": 0.001953125,
+      "num_input_tokens_seen": 247424832,
+      "step": 1678
+    },
+    {
+      "epoch": 0.484001153070049,
+      "grad_norm": 1.9229211822494456,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "num_input_tokens_seen": 247559832,
+      "step": 1679
+    },
+    {
+      "epoch": 0.484001153070049,
+      "loss": 0.0012820811243727803,
+      "loss_ce": 0.00022779422579333186,
+      "loss_xval": 0.0010528564453125,
+      "num_input_tokens_seen": 247559832,
+      "step": 1679
+    },
+    {
+      "epoch": 0.4842894205823004,
+      "grad_norm": 3.973230759058265,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "num_input_tokens_seen": 247732560,
+      "step": 1680
+    },
+    {
+      "epoch": 0.4842894205823004,
+      "loss": 0.004148183390498161,
+      "loss_ce": 0.00030106125632300973,
+      "loss_xval": 0.00384521484375,
+      "num_input_tokens_seen": 247732560,
+      "step": 1680
+    },
+    {
+      "epoch": 0.48457768809455176,
+      "grad_norm": 1.1343054790113372,
+      "learning_rate": 0.0001,
+      "loss": 0.0054,
+      "num_input_tokens_seen": 247867400,
+      "step": 1681
+    },
+    {
+      "epoch": 0.48457768809455176,
+      "loss": 0.0073390984907746315,
+      "loss_ce": 0.00671324972063303,
+      "loss_xval": 0.0006256103515625,
+      "num_input_tokens_seen": 247867400,
+      "step": 1681
+    },
+    {
+      "epoch": 0.4848659556068031,
+      "grad_norm": 2.8233753979799636,
+      "learning_rate": 0.0001,
+      "loss": 0.0032,
+      "num_input_tokens_seen": 248002544,
+      "step": 1682
+    },
+    {
+      "epoch": 0.4848659556068031,
+      "loss": 0.0029423628002405167,
+      "loss_ce": 0.0011876022908836603,
+      "loss_xval": 0.0017547607421875,
+      "num_input_tokens_seen": 248002544,
+      "step": 1682
+    },
+    {
+      "epoch": 0.4851542231190545,
+      "grad_norm": 3.1356041359079727,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "num_input_tokens_seen": 248174976,
+      "step": 1683
+    },
+    {
+      "epoch": 0.4851542231190545,
+      "loss": 0.0028833728283643723,
+      "loss_ce": 0.0003389697812963277,
+      "loss_xval": 0.0025482177734375,
+      "num_input_tokens_seen": 248174976,
+      "step": 1683
+    },
+    {
+      "epoch": 0.48544249063130585,
+      "grad_norm": 0.16825660364549808,
+      "learning_rate": 0.0001,
+      "loss": 0.0031,
+      "num_input_tokens_seen": 248309784,
+      "step": 1684
+    },
+    {
+      "epoch": 0.48544249063130585,
+      "loss": 0.005761802662163973,
+      "loss_ce": 0.005357802379876375,
+      "loss_xval": 0.00040435791015625,
+      "num_input_tokens_seen": 248309784,
+      "step": 1684
+    },
+    {
+      "epoch": 0.4857307581435572,
+      "grad_norm": 2.9574485021450614,
+      "learning_rate": 0.0001,
+      "loss": 0.0025,
+      "num_input_tokens_seen": 248444784,
+      "step": 1685
+    },
+    {
+      "epoch": 0.4857307581435572,
+      "loss": 0.0023077609948813915,
+      "loss_ce": 0.00037084840005263686,
+      "loss_xval": 0.0019378662109375,
+      "num_input_tokens_seen": 248444784,
+      "step": 1685
+    },
+    {
+      "epoch": 0.48601902565580857,
+      "grad_norm": 1.9710236987670244,
+      "learning_rate": 0.0001,
+      "loss": 0.0041,
+      "num_input_tokens_seen": 248617416,
+      "step": 1686
+    },
+    {
+      "epoch": 0.48601902565580857,
+      "loss": 0.001532458234578371,
+      "loss_ce": 0.00034704094287008047,
+      "loss_xval": 0.00118255615234375,
+      "num_input_tokens_seen": 248617416,
+      "step": 1686
+    },
+    {
+      "epoch": 0.48630729316806,
+      "grad_norm": 1.6574027342722102,
+      "learning_rate": 0.0001,
+      "loss": 0.0044,
+      "num_input_tokens_seen": 248752320,
+      "step": 1687
+    },
+    {
+      "epoch": 0.48630729316806,
+      "loss": 0.007482764311134815,
+      "loss_ce": 0.006698128767311573,
+      "loss_xval": 0.00078582763671875,
+      "num_input_tokens_seen": 248752320,
+      "step": 1687
+    },
+    {
+      "epoch": 0.48659556068031135,
+      "grad_norm": 3.402714718010406,
+      "learning_rate": 0.0001,
+      "loss": 0.003,
+      "num_input_tokens_seen": 248887352,
+      "step": 1688
+    },
+    {
+      "epoch": 0.48659556068031135,
+      "loss": 0.002607089001685381,
+      "loss_ce": 0.00019810753292404115,
+      "loss_xval": 0.002410888671875,
+      "num_input_tokens_seen": 248887352,
+      "step": 1688
+    },
+    {
+      "epoch": 0.4868838281925627,
+      "grad_norm": 1.0612269469325708,
+      "learning_rate": 0.0001,
+      "loss": 0.0037,
+      "num_input_tokens_seen": 249060080,
+      "step": 1689
+    },
+    {
+      "epoch": 0.4868838281925627,
+      "loss": 0.000750762177631259,
+      "loss_ce": 0.00023482434335164726,
+      "loss_xval": 0.000514984130859375,
+      "num_input_tokens_seen": 249060080,
+      "step": 1689
+    },
+    {
+      "epoch": 0.48717209570481407,
+      "grad_norm": 2.6202481583055857,
+      "learning_rate": 0.0001,
+      "loss": 0.0065,
+      "num_input_tokens_seen": 249194848,
+      "step": 1690
+    },
+    {
+      "epoch": 0.48717209570481407,
+      "loss": 0.010731605812907219,
+      "loss_ce": 0.009190468117594719,
+      "loss_xval": 0.0015411376953125,
+      "num_input_tokens_seen": 249194848,
+      "step": 1690
+    },
+    {
+      "epoch": 0.48746036321706543,
+      "grad_norm": 3.391612477312743,
+      "learning_rate": 0.0001,
+      "loss": 0.003,
+      "num_input_tokens_seen": 249329816,
+      "step": 1691
+    },
+    {
+      "epoch": 0.48746036321706543,
+      "loss": 0.002606350462883711,
+      "loss_ce": 0.00022121102665551007,
+      "loss_xval": 0.00238037109375,
+      "num_input_tokens_seen": 249329816,
+      "step": 1691
+    },
+    {
+      "epoch": 0.4877486307293168,
+      "grad_norm": 0.29936391697277404,
+      "learning_rate": 0.0001,
+      "loss": 0.0049,
+      "num_input_tokens_seen": 249502304,
+      "step": 1692
+    },
+    {
+      "epoch": 0.4877486307293168,
+      "loss": 0.001809053821489215,
+      "loss_ce": 0.0015478662680834532,
+      "loss_xval": 0.0002613067626953125,
+      "num_input_tokens_seen": 249502304,
+      "step": 1692
+    },
+    {
+      "epoch": 0.48803689824156815,
+      "grad_norm": 3.5497603899955794,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "num_input_tokens_seen": 249637096,
+      "step": 1693
+    },
+    {
+      "epoch": 0.48803689824156815,
+      "loss": 0.01052839308977127,
+      "loss_ce": 0.007900066673755646,
+      "loss_xval": 0.00262451171875,
+      "num_input_tokens_seen": 249637096,
+      "step": 1693
+    },
+    {
+      "epoch": 0.48832516575381957,
+      "grad_norm": 4.189153427151052,
+      "learning_rate": 0.0001,
+      "loss": 0.0048,
+      "num_input_tokens_seen": 249772064,
+      "step": 1694
+    },
+    {
+      "epoch": 0.48832516575381957,
+      "loss": 0.0034670508466660976,
+      "loss_ce": 0.00020929938182234764,
+      "loss_xval": 0.003265380859375,
+      "num_input_tokens_seen": 249772064,
+      "step": 1694
+    },
+    {
+      "epoch": 0.48861343326607093,
+      "grad_norm": 0.6277208093933458,
+      "learning_rate": 0.0001,
+      "loss": 0.0045,
+      "num_input_tokens_seen": 249944584,
+      "step": 1695
+    },
+    {
+      "epoch": 0.48861343326607093,
+      "loss": 0.0007549943984486163,
+      "loss_ce": 0.00036637208540923893,
+      "loss_xval": 0.00038909912109375,
+      "num_input_tokens_seen": 249944584,
+      "step": 1695
+    },
+    {
+      "epoch": 0.4889017007783223,
+      "grad_norm": 3.3679481066575474,
+      "learning_rate": 0.0001,
+      "loss": 0.006,
+      "num_input_tokens_seen": 250079368,
+      "step": 1696
+    },
+    {
+      "epoch": 0.4889017007783223,
+      "loss": 0.008834033273160458,
+      "loss_ce": 0.006375460885465145,
+      "loss_xval": 0.0024566650390625,
+      "num_input_tokens_seen": 250079368,
+      "step": 1696
+    },
+    {
+      "epoch": 0.48918996829057365,
+      "grad_norm": 4.179445419185496,
+      "learning_rate": 0.0001,
+      "loss": 0.0043,
+      "num_input_tokens_seen": 250214560,
+      "step": 1697
+    },
+    {
+      "epoch": 0.48918996829057365,
+      "loss": 0.004044243134558201,
+      "loss_ce": 0.00017232507525477558,
+      "loss_xval": 0.003875732421875,
+      "num_input_tokens_seen": 250214560,
+      "step": 1697
+    },
+    {
+      "epoch": 0.489478235802825,
+      "grad_norm": 1.138036604988876,
+      "learning_rate": 0.0001,
+      "loss": 0.0051,
+      "num_input_tokens_seen": 250387136,
+      "step": 1698
+    },
+    {
+      "epoch": 0.489478235802825,
+      "loss": 0.0008948410395532846,
+      "loss_ce": 0.00018316156638320535,
+      "loss_xval": 0.000713348388671875,
+      "num_input_tokens_seen": 250387136,
+      "step": 1698
+    },
+    {
+      "epoch": 0.4897665033150764,
+      "grad_norm": 2.9359114211323214,
+      "learning_rate": 0.0001,
+      "loss": 0.0066,
+      "num_input_tokens_seen": 250521960,
+      "step": 1699
+    },
+    {
+      "epoch": 0.4897665033150764,
+      "loss": 0.008544711396098137,
+      "loss_ce": 0.006434229668229818,
+      "loss_xval": 0.002105712890625,
+      "num_input_tokens_seen": 250521960,
+      "step": 1699
+    },
+    {
+      "epoch": 0.49005477082732773,
+      "grad_norm": 4.246204134764685,
+      "learning_rate": 0.0001,
+      "loss": 0.0044,
+      "num_input_tokens_seen": 250657000,
+      "step": 1700
+    },
+    {
+      "epoch": 0.49005477082732773,
+      "loss": 0.004051597788929939,
+      "loss_ce": 0.000160606752615422,
+      "loss_xval": 0.0038909912109375,
+      "num_input_tokens_seen": 250657000,
+      "step": 1700
+    },
+    {
+      "epoch": 0.49034303833957915,
+      "grad_norm": 1.6708110961924532,
+      "learning_rate": 0.0001,
+      "loss": 0.0059,
+      "num_input_tokens_seen": 250829520,
+      "step": 1701
+    },
+    {
+      "epoch": 0.49034303833957915,
+      "loss": 0.0034429319202899933,
+      "loss_ce": 0.0023800618946552277,
+      "loss_xval": 0.00106048583984375,
+      "num_input_tokens_seen": 250829520,
+      "step": 1701
+    },
+    {
+      "epoch": 0.4906313058518305,
+      "grad_norm": 2.29495962909665,
+      "learning_rate": 0.0001,
+      "loss": 0.006,
+      "num_input_tokens_seen": 250964360,
+      "step": 1702
+    },
+    {
+      "epoch": 0.4906313058518305,
+      "loss": 0.009584905579686165,
+      "loss_ce": 0.008198263123631477,
+      "loss_xval": 0.0013885498046875,
+      "num_input_tokens_seen": 250964360,
+      "step": 1702
+    },
+    {
+      "epoch": 0.4909195733640819,
+      "grad_norm": 4.051342750901919,
+      "learning_rate": 0.0001,
+      "loss": 0.004,
+      "num_input_tokens_seen": 251099480,
+      "step": 1703
+    },
+    {
+      "epoch": 0.4909195733640819,
+      "loss": 0.0038174032233655453,
+      "loss_ce": 0.0001457572216168046,
+      "loss_xval": 0.0036773681640625,
+      "num_input_tokens_seen": 251099480,
+      "step": 1703
+    },
+    {
+      "epoch": 0.49120784087633323,
+      "grad_norm": 2.4280188175113104,
+      "learning_rate": 0.0001,
+      "loss": 0.0067,
+      "num_input_tokens_seen": 251271904,
+      "step": 1704
+    },
+    {
+      "epoch": 0.49120784087633323,
+      "loss": 0.002196403918787837,
+      "loss_ce": 0.0004912342992611229,
+      "loss_xval": 0.001708984375,
+      "num_input_tokens_seen": 251271904,
+      "step": 1704
+    },
+    {
+      "epoch": 0.4914961083885846,
+      "grad_norm": 0.9184818784150366,
+      "learning_rate": 0.0001,
+      "loss": 0.0049,
+      "num_input_tokens_seen": 251406736,
+      "step": 1705
+    },
+    {
+      "epoch": 0.4914961083885846,
+      "loss": 0.009084072895348072,
+      "loss_ce": 0.008687344379723072,
+      "loss_xval": 0.000396728515625,
+      "num_input_tokens_seen": 251406736,
+      "step": 1705
+    },
+    {
+      "epoch": 0.49178437590083596,
+      "grad_norm": 3.2573763940603477,
+      "learning_rate": 0.0001,
+      "loss": 0.0028,
+      "num_input_tokens_seen": 251541728,
+      "step": 1706
+    },
+    {
+      "epoch": 0.49178437590083596,
+      "loss": 0.00256602605804801,
+      "loss_ce": 0.00013606389984488487,
+      "loss_xval": 0.0024261474609375,
+      "num_input_tokens_seen": 251541728,
+      "step": 1706
+    },
+    {
+      "epoch": 0.4920726434130873,
+      "grad_norm": 2.9257386010126085,
+      "learning_rate": 0.0001,
+      "loss": 0.0066,
+      "num_input_tokens_seen": 251714192,
+      "step": 1707
+    },
+    {
+      "epoch": 0.4920726434130873,
+      "loss": 0.0038506370037794113,
+      "loss_ce": 0.001319585251621902,
+      "loss_xval": 0.002532958984375,
+      "num_input_tokens_seen": 251714192,
+      "step": 1707
+    },
+    {
+      "epoch": 0.49236091092533874,
+      "grad_norm": 0.546968323447548,
+      "learning_rate": 0.0001,
+      "loss": 0.0041,
+      "num_input_tokens_seen": 251848992,
+      "step": 1708
+    },
+    {
+      "epoch": 0.49236091092533874,
+      "loss": 0.007951841689646244,
+      "loss_ce": 0.007509575225412846,
+      "loss_xval": 0.0004425048828125,
+      "num_input_tokens_seen": 251848992,
+      "step": 1708
+    },
+    {
+      "epoch": 0.4926491784375901,
+      "grad_norm": 1.8227164441424433,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "num_input_tokens_seen": 251984240,
+      "step": 1709
+    },
+    {
+      "epoch": 0.4926491784375901,
+      "loss": 0.0008962653810158372,
+      "loss_ce": 0.00012998809688724577,
+      "loss_xval": 0.000766754150390625,
+      "num_input_tokens_seen": 251984240,
+      "step": 1709
+    },
+    {
+      "epoch": 0.49293744594984146,
+      "grad_norm": 2.5433669083958867,
+      "learning_rate": 0.0001,
+      "loss": 0.0068,
+      "num_input_tokens_seen": 252156808,
+      "step": 1710
+    },
+    {
+      "epoch": 0.49293744594984146,
+      "loss": 0.002464288379997015,
+      "loss_ce": 0.0006427703192457557,
+      "loss_xval": 0.00182342529296875,
+      "num_input_tokens_seen": 252156808,
+      "step": 1710
+    },
+    {
+      "epoch": 0.4932257134620928,
+      "grad_norm": 1.8610536126480004,
+      "learning_rate": 0.0001,
+      "loss": 0.0043,
+      "num_input_tokens_seen": 252291600,
+      "step": 1711
+    },
+    {
+      "epoch": 0.4932257134620928,
+      "loss": 0.007783045060932636,
+      "loss_ce": 0.0064679281786084175,
+      "loss_xval": 0.001312255859375,
+      "num_input_tokens_seen": 252291600,
+      "step": 1711
+    },
+    {
+      "epoch": 0.4935139809743442,
+      "grad_norm": 0.8857956722653375,
+      "learning_rate": 0.0001,
+      "loss": 0.0016,
+      "num_input_tokens_seen": 252426768,
+      "step": 1712
+    },
+    {
+      "epoch": 0.4935139809743442,
+      "loss": 0.0005321372882463038,
+      "loss_ce": 0.00012110365059925243,
+      "loss_xval": 0.0004119873046875,
+      "num_input_tokens_seen": 252426768,
+      "step": 1712
+    },
+    {
+      "epoch": 0.49380224848659554,
+      "grad_norm": 0.5796544010149521,
+      "learning_rate": 0.0001,
+      "loss": 0.0064,
+      "num_input_tokens_seen": 252599320,
+      "step": 1713
+    },
+    {
+      "epoch": 0.49380224848659554,
+      "loss": 0.0036989161744713783,
+      "loss_ce": 0.0034384438768029213,
+      "loss_xval": 0.0002613067626953125,
+      "num_input_tokens_seen": 252599320,
+      "step": 1713
+    },
+    {
+      "epoch": 0.49409051599884696,
+      "grad_norm": 0.7655430660339378,
+      "learning_rate": 0.0001,
+      "loss": 0.0038,
+      "num_input_tokens_seen": 252734120,
+      "step": 1714
+    },
+    {
+      "epoch": 0.49409051599884696,
+      "loss": 0.006788511760532856,
+      "loss_ce": 0.006498952396214008,
+      "loss_xval": 0.0002899169921875,
+      "num_input_tokens_seen": 252734120,
+      "step": 1714
+    },
+    {
+      "epoch": 0.4943787835110983,
+      "grad_norm": 0.7202521182045087,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "num_input_tokens_seen": 252869224,
+      "step": 1715
+    },
+    {
+      "epoch": 0.4943787835110983,
+      "loss": 0.00029843367519788444,
+      "loss_ce": 0.00011461296526249498,
+      "loss_xval": 0.00018405914306640625,
+      "num_input_tokens_seen": 252869224,
+      "step": 1715
+    },
+    {
+      "epoch": 0.4946670510233497,
+      "grad_norm": 0.20232642834434011,
+      "learning_rate": 0.0001,
+      "loss": 0.0053,
+      "num_input_tokens_seen": 253041680,
+      "step": 1716
+    },
+    {
+      "epoch": 0.4946670510233497,
+      "loss": 0.0009578837198205292,
+      "loss_ce": 0.0007630957406945527,
+      "loss_xval": 0.000194549560546875,
+      "num_input_tokens_seen": 253041680,
+      "step": 1716
+    },
+    {
+      "epoch": 0.49495531853560104,
+      "grad_norm": 1.1670590573363684,
+      "learning_rate": 0.0001,
+      "loss": 0.0048,
+      "num_input_tokens_seen": 253176552,
+      "step": 1717
+    },
+    {
+      "epoch": 0.49495531853560104,
+      "loss": 0.008777554146945477,
+      "loss_ce": 0.008417423814535141,
+      "loss_xval": 0.0003604888916015625,
+      "num_input_tokens_seen": 253176552,
+      "step": 1717
+    },
+    {
+      "epoch": 0.4952435860478524,
+      "grad_norm": 1.4988461928806927,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "num_input_tokens_seen": 253311680,
+      "step": 1718
+    },
+    {
+      "epoch": 0.4952435860478524,
+      "loss": 0.0006728050066158175,
+      "loss_ce": 0.00010918345651589334,
+      "loss_xval": 0.0005645751953125,
+      "num_input_tokens_seen": 253311680,
+      "step": 1718
+    },
+    {
+      "epoch": 0.49553185356010376,
+      "grad_norm": 0.8959117241888527,
+      "learning_rate": 0.0001,
+      "loss": 0.0043,
+      "num_input_tokens_seen": 253484192,
+      "step": 1719
+    },
+    {
+      "epoch": 0.49553185356010376,
+      "loss": 0.0006553968414664268,
+      "loss_ce": 0.00014923422713764012,
+      "loss_xval": 0.000507354736328125,
+      "num_input_tokens_seen": 253484192,
+      "step": 1719
+    },
+    {
+      "epoch": 0.4958201210723551,
+      "grad_norm": 0.19446805773449719,
+      "learning_rate": 0.0001,
+      "loss": 0.0034,
+      "num_input_tokens_seen": 253618976,
+      "step": 1720
+    },
+    {
+      "epoch": 0.4958201210723551,
+      "loss": 0.0064426204189658165,
+      "loss_ce": 0.006185128353536129,
+      "loss_xval": 0.0002574920654296875,
+      "num_input_tokens_seen": 253618976,
+      "step": 1720
+    },
+    {
+      "epoch": 0.49610838858460654,
+      "grad_norm": 0.5419645813642442,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "num_input_tokens_seen": 253754168,
+      "step": 1721
+    },
+    {
+      "epoch": 0.49610838858460654,
+      "loss": 0.00029128711321391165,
+      "loss_ce": 0.0001080028378055431,
+      "loss_xval": 0.00018310546875,
+      "num_input_tokens_seen": 253754168,
+      "step": 1721
+    },
+    {
+      "epoch": 0.4963966560968579,
+      "grad_norm": 0.16764965953239006,
+      "learning_rate": 0.0001,
+      "loss": 0.0031,
+      "num_input_tokens_seen": 253926776,
+      "step": 1722
+    },
+    {
+      "epoch": 0.4963966560968579,
+      "loss": 0.00037144176894798875,
+      "loss_ce": 0.00013326163752935827,
+      "loss_xval": 0.0002384185791015625,
+      "num_input_tokens_seen": 253926776,
+      "step": 1722
+    },
+    {
+      "epoch": 0.49668492360910926,
+      "grad_norm": 0.5143012980626315,
+      "learning_rate": 0.0001,
+      "loss": 0.006,
+      "num_input_tokens_seen": 254061584,
+      "step": 1723
+    },
+    {
+      "epoch": 0.49668492360910926,
+      "loss": 0.011035285890102386,
+      "loss_ce": 0.010848961770534515,
+      "loss_xval": 0.00018596649169921875,
+      "num_input_tokens_seen": 254061584,
+      "step": 1723
+    },
+    {
+      "epoch": 0.4969731911213606,
+      "grad_norm": 0.6497143775508163,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "num_input_tokens_seen": 254196720,
+      "step": 1724
+    },
+    {
+      "epoch": 0.4969731911213606,
+      "loss": 0.0003499590093269944,
+      "loss_ce": 0.00016345609037671238,
+      "loss_xval": 0.000186920166015625,
+      "num_input_tokens_seen": 254196720,
+      "step": 1724
+    },
+    {
+      "epoch": 0.497261458633612,
+      "grad_norm": 0.3602351152941275,
+      "learning_rate": 0.0001,
+      "loss": 0.0066,
+      "num_input_tokens_seen": 254369112,
+      "step": 1725
+    },
+    {
+      "epoch": 0.497261458633612,
+      "loss": 0.0003184653469361365,
+      "loss_ce": 0.00011902822006959468,
+      "loss_xval": 0.00019931793212890625,
+      "num_input_tokens_seen": 254369112,
+      "step": 1725
+    },
+    {
+      "epoch": 0.49754972614586335,
+      "grad_norm": 0.9732779173841355,
+      "learning_rate": 0.0001,
+      "loss": 0.0049,
+      "num_input_tokens_seen": 254503904,
+      "step": 1726
+    },
+    {
+      "epoch": 0.49754972614586335,
+      "loss": 0.009241588413715363,
+      "loss_ce": 0.00895286351442337,
+      "loss_xval": 0.0002880096435546875,
+      "num_input_tokens_seen": 254503904,
+      "step": 1726
+    },
+    {
+      "epoch": 0.4978379936581147,
+      "grad_norm": 1.8909439477260013,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "num_input_tokens_seen": 254638912,
+      "step": 1727
+    },
+    {
+      "epoch": 0.4978379936581147,
+      "loss": 0.0012038424611091614,
+      "loss_ce": 0.00030452755163423717,
+      "loss_xval": 0.0009002685546875,
+      "num_input_tokens_seen": 254638912,
+      "step": 1727
+    },
+    {
+      "epoch": 0.4981262611703661,
+      "grad_norm": 2.5298356679902243,
+      "learning_rate": 0.0001,
+      "loss": 0.0055,
+      "num_input_tokens_seen": 254811496,
+      "step": 1728
+    },
+    {
+      "epoch": 0.4981262611703661,
+      "loss": 0.0020236079581081867,
+      "loss_ce": 0.0001296107075177133,
+      "loss_xval": 0.00189208984375,
+      "num_input_tokens_seen": 254811496,
+      "step": 1728
+    },
+    {
+      "epoch": 0.4984145286826175,
+      "grad_norm": 3.4742859942049025,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "num_input_tokens_seen": 254946296,
+      "step": 1729
+    },
+    {
+      "epoch": 0.4984145286826175,
+      "loss": 0.012024440802633762,
+      "loss_ce": 0.008614100515842438,
+      "loss_xval": 0.00341796875,
+      "num_input_tokens_seen": 254946296,
+      "step": 1729
+    },
+    {
+      "epoch": 0.49870279619486885,
+      "grad_norm": 5.673154601826852,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "num_input_tokens_seen": 255081392,
+      "step": 1730
+    },
+    {
+      "epoch": 0.49870279619486885,
+      "loss": 0.007622344419360161,
+      "loss_ce": 0.00011502046254463494,
+      "loss_xval": 0.00750732421875,
+      "num_input_tokens_seen": 255081392,
+      "step": 1730
+    },
+    {
+      "epoch": 0.4989910637071202,
+      "grad_norm": 10.11347194036202,
+      "learning_rate": 0.0001,
+      "loss": 0.027,
+      "num_input_tokens_seen": 255253968,
+      "step": 1731
+    },
+    {
+      "epoch": 0.4989910637071202,
+      "loss": 0.021371876820921898,
+      "loss_ce": 0.00011638353316811845,
+      "loss_xval": 0.021240234375,
+      "num_input_tokens_seen": 255253968,
+      "step": 1731
+    },
+    {
+      "epoch": 0.49927933121937157,
+      "grad_norm": 17.873211376225438,
+      "learning_rate": 0.0001,
+      "loss": 0.074,
+      "num_input_tokens_seen": 255388760,
+      "step": 1732
+    },
+    {
+      "epoch": 0.49927933121937157,
+      "loss": 0.07567343860864639,
+      "loss_ce": 0.007741309702396393,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 255388760,
+      "step": 1732
+    },
+    {
+      "epoch": 0.49956759873162293,
+      "grad_norm": 28.20506840621217,
+      "learning_rate": 0.0001,
+      "loss": 0.1821,
+      "num_input_tokens_seen": 255523872,
+      "step": 1733
+    },
+    {
+      "epoch": 0.49956759873162293,
+      "loss": 0.18528181314468384,
+      "loss_ce": 0.00010114896576851606,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 255523872,
+      "step": 1733
+    },
+    {
+      "epoch": 0.4998558662438743,
+      "grad_norm": 32.107443205850224,
+      "learning_rate": 0.0001,
+      "loss": 0.2388,
+      "num_input_tokens_seen": 255696352,
+      "step": 1734
+    },
+    {
+      "epoch": 0.4998558662438743,
+      "loss": 0.24280518293380737,
+      "loss_ce": 0.00012939998123329133,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 255696352,
+      "step": 1734
+    },
+    {
+      "epoch": 0.5001441337561257,
+      "grad_norm": 20.192170135400502,
+      "learning_rate": 0.0001,
+      "loss": 0.1051,
+      "num_input_tokens_seen": 255831144,
+      "step": 1735
+    },
+    {
+      "epoch": 0.5001441337561257,
+      "loss": 0.11830128729343414,
+      "loss_ce": 0.012649435549974442,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 255831144,
+      "step": 1735
+    },
+    {
+      "epoch": 0.5004324012683771,
+      "grad_norm": 4.138208841567372,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "num_input_tokens_seen": 255966200,
+      "step": 1736
+    },
+    {
+      "epoch": 0.5004324012683771,
+      "loss": 0.009793087840080261,
+      "loss_ce": 0.00044708000496029854,
+      "loss_xval": 0.00933837890625,
+      "num_input_tokens_seen": 255966200,
+      "step": 1736
+    },
+    {
+      "epoch": 0.5007206687806284,
+      "grad_norm": 20.913269450739836,
+      "learning_rate": 0.0001,
+      "loss": 0.1158,
+      "num_input_tokens_seen": 256138808,
+      "step": 1737
+    },
+    {
+      "epoch": 0.5007206687806284,
+      "loss": 0.11857728660106659,
+      "loss_ce": 0.00023012000019662082,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 256138808,
+      "step": 1737
+    },
+    {
+      "epoch": 0.5010089362928798,
+      "grad_norm": 16.722405346702054,
+      "learning_rate": 0.0001,
+      "loss": 0.0794,
+      "num_input_tokens_seen": 256273608,
+      "step": 1738
+    },
+    {
+      "epoch": 0.5010089362928798,
+      "loss": 0.09203828871250153,
+      "loss_ce": 0.006344921886920929,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 256273608,
+      "step": 1738
+    },
+    {
+      "epoch": 0.5012972038051312,
+      "grad_norm": 2.1893892288404886,
+      "learning_rate": 0.0001,
+      "loss": 0.0049,
+      "num_input_tokens_seen": 256408632,
+      "step": 1739
+    },
+    {
+      "epoch": 0.5012972038051312,
+      "loss": 0.0053595867939293385,
+      "loss_ce": 0.0009078349103219807,
+      "loss_xval": 0.00445556640625,
+      "num_input_tokens_seen": 256408632,
+      "step": 1739
+    },
+    {
+      "epoch": 0.5015854713173825,
+      "grad_norm": 14.686392840406226,
+      "learning_rate": 0.0001,
+      "loss": 0.0608,
+      "num_input_tokens_seen": 256581120,
+      "step": 1740
+    },
+    {
+      "epoch": 0.5015854713173825,
+      "loss": 0.058756496757268906,
+      "loss_ce": 0.0004679245757870376,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 256581120,
+      "step": 1740
+    },
+    {
+      "epoch": 0.5018737388296339,
+      "grad_norm": 11.310540214461815,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "num_input_tokens_seen": 256715976,
+      "step": 1741
+    },
+    {
+      "epoch": 0.5018737388296339,
+      "loss": 0.04806603491306305,
+      "loss_ce": 0.008728873915970325,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 256715976,
+      "step": 1741
+    },
+    {
+      "epoch": 0.5021620063418852,
+      "grad_norm": 1.33807983469355,
+      "learning_rate": 0.0001,
+      "loss": 0.0034,
+      "num_input_tokens_seen": 256851024,
+      "step": 1742
+    },
+    {
+      "epoch": 0.5021620063418852,
+      "loss": 0.003706869902089238,
+      "loss_ce": 0.0002393101021880284,
+      "loss_xval": 0.0034637451171875,
+      "num_input_tokens_seen": 256851024,
+      "step": 1742
+    },
+    {
+      "epoch": 0.5024502738541367,
+      "grad_norm": 10.614122739332574,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "num_input_tokens_seen": 257023632,
+      "step": 1743
+    },
+    {
+      "epoch": 0.5024502738541367,
+      "loss": 0.027926770970225334,
+      "loss_ce": 0.00026258552679792047,
+      "loss_xval": 0.0277099609375,
+      "num_input_tokens_seen": 257023632,
+      "step": 1743
+    },
+    {
+      "epoch": 0.502738541366388,
+      "grad_norm": 7.657015223009477,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "num_input_tokens_seen": 257158440,
+      "step": 1744
+    },
+    {
+      "epoch": 0.502738541366388,
+      "loss": 0.02629696950316429,
+      "loss_ce": 0.007711765356361866,
+      "loss_xval": 0.0185546875,
+      "num_input_tokens_seen": 257158440,
+      "step": 1744
+    },
+    {
+      "epoch": 0.5030268088786394,
+      "grad_norm": 2.2808369659423535,
+      "learning_rate": 0.0001,
+      "loss": 0.0023,
+      "num_input_tokens_seen": 257293408,
+      "step": 1745
+    },
+    {
+      "epoch": 0.5030268088786394,
+      "loss": 0.0011452080216258764,
+      "loss_ce": 0.0002578140702098608,
+      "loss_xval": 0.000888824462890625,
+      "num_input_tokens_seen": 257293408,
+      "step": 1745
+    },
+    {
+      "epoch": 0.5033150763908908,
+      "grad_norm": 8.013615070959606,
+      "learning_rate": 0.0001,
+      "loss": 0.0228,
+      "num_input_tokens_seen": 257465864,
+      "step": 1746
+    },
+    {
+      "epoch": 0.5033150763908908,
+      "loss": 0.01814066618680954,
+      "loss_ce": 0.00030314087052829564,
+      "loss_xval": 0.017822265625,
+      "num_input_tokens_seen": 257465864,
+      "step": 1746
+    },
+    {
+      "epoch": 0.5036033439031421,
+      "grad_norm": 3.457585227141799,
+      "learning_rate": 0.0001,
+      "loss": 0.0069,
+      "num_input_tokens_seen": 257600728,
+      "step": 1747
+    },
+    {
+      "epoch": 0.5036033439031421,
+      "loss": 0.011352559551596642,
+      "loss_ce": 0.005506536923348904,
+      "loss_xval": 0.005859375,
+      "num_input_tokens_seen": 257600728,
+      "step": 1747
+    },
+    {
+      "epoch": 0.5038916114153935,
+      "grad_norm": 5.38481600143461,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "num_input_tokens_seen": 257735848,
+      "step": 1748
+    },
+    {
+      "epoch": 0.5038916114153935,
+      "loss": 0.005389847327023745,
+      "loss_ce": 0.00027815281646326184,
+      "loss_xval": 0.005126953125,
+      "num_input_tokens_seen": 257735848,
+      "step": 1748
+    },
+    {
+      "epoch": 0.5041798789276448,
+      "grad_norm": 6.642525855981442,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "num_input_tokens_seen": 257908416,
+      "step": 1749
+    },
+    {
+      "epoch": 0.5041798789276448,
+      "loss": 0.01205461099743843,
+      "loss_ce": 0.00032060168450698256,
+      "loss_xval": 0.01171875,
+      "num_input_tokens_seen": 257908416,
+      "step": 1749
+    },
+    {
+      "epoch": 0.5044681464398962,
+      "grad_norm": 1.1926943126740923,
+      "learning_rate": 0.0001,
+      "loss": 0.0057,
+      "num_input_tokens_seen": 258043136,
+      "step": 1750
+    },
+    {
+      "epoch": 0.5044681464398962,
+      "eval_websight_new_IoU": 0.0,
+      "eval_websight_new_MAE_x": 0.08808813989162445,
+      "eval_websight_new_MAE_y": 0.08523305132985115,
+      "eval_websight_new_NUM_probability": 0.99669548869133,
+      "eval_websight_new_inside_bbox": 0.0,
+      "eval_websight_new_loss": 0.013562072068452835,
+      "eval_websight_new_loss_ce": 0.00028378186107147485,
+      "eval_websight_new_loss_xval": 0.01325225830078125,
+      "eval_websight_new_runtime": 35.8548,
+      "eval_websight_new_samples_per_second": 1.395,
+      "eval_websight_new_steps_per_second": 0.056,
+      "num_input_tokens_seen": 258043136,
+      "step": 1750
+    },
+    {
+      "epoch": 0.5044681464398962,
+      "eval_seeclick_IoU": 0.006532402941957116,
+      "eval_seeclick_MAE_x": 0.13058312609791756,
+      "eval_seeclick_MAE_y": 0.1318240687251091,
+      "eval_seeclick_NUM_probability": 0.9967434406280518,
+      "eval_seeclick_inside_bbox": 0.0711805559694767,
+      "eval_seeclick_loss": 0.03064095973968506,
+      "eval_seeclick_loss_ce": 0.00829253252595663,
+      "eval_seeclick_loss_xval": 0.02297210693359375,
+      "eval_seeclick_runtime": 63.4085,
+      "eval_seeclick_samples_per_second": 0.789,
+      "eval_seeclick_steps_per_second": 0.032,
+      "num_input_tokens_seen": 258043136,
+      "step": 1750
+    },
+    {
+      "epoch": 0.5044681464398962,
+      "eval_icons_IoU": 0.0,
+      "eval_icons_MAE_x": 0.11214055120944977,
+      "eval_icons_MAE_y": 0.1266588643193245,
+      "eval_icons_NUM_probability": 0.9963801205158234,
+      "eval_icons_inside_bbox": 0.0,
+      "eval_icons_loss": 0.028072558343410492,
+      "eval_icons_loss_ce": 0.004345909343101084,
+      "eval_icons_loss_xval": 0.02297210693359375,
+      "eval_icons_runtime": 66.7092,
+      "eval_icons_samples_per_second": 0.75,
+      "eval_icons_steps_per_second": 0.03,
+      "num_input_tokens_seen": 258043136,
+      "step": 1750
+    },
+    {
+      "epoch": 0.5044681464398962,
+      "loss": 0.02850811555981636,
+      "loss_ce": 0.005391049198806286,
+      "loss_xval": 0.0230712890625,
+      "num_input_tokens_seen": 258043136,
+      "step": 1750
+    },
+    {
+      "epoch": 0.5047564139521475,
+      "grad_norm": 7.831924408909281,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "num_input_tokens_seen": 258178240,
+      "step": 1751
+    },
+    {
+      "epoch": 0.5047564139521475,
+      "loss": 0.013527311384677887,
+      "loss_ce": 0.0002903116401284933,
+      "loss_xval": 0.01324462890625,
+      "num_input_tokens_seen": 258178240,
+      "step": 1751
+    },
+    {
+      "epoch": 0.505044681464399,
+      "grad_norm": 4.176178068633609,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "num_input_tokens_seen": 258350680,
+      "step": 1752
+    },
+    {
+      "epoch": 0.505044681464399,
+      "loss": 0.006143480539321899,
+      "loss_ce": 0.0004366936336737126,
+      "loss_xval": 0.005706787109375,
+      "num_input_tokens_seen": 258350680,
+      "step": 1752
+    },
+    {
+      "epoch": 0.5053329489766504,
+      "grad_norm": 5.643783467526188,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "num_input_tokens_seen": 258485472,
+      "step": 1753
+    },
+    {
+      "epoch": 0.5053329489766504,
+      "loss": 0.014654072932898998,
+      "loss_ce": 0.008191976696252823,
+      "loss_xval": 0.0064697265625,
+      "num_input_tokens_seen": 258485472,
+      "step": 1753
+    },
+    {
+      "epoch": 0.5056212164889017,
+      "grad_norm": 8.477634537026859,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "num_input_tokens_seen": 258620400,
+      "step": 1754
+    },
+    {
+      "epoch": 0.5056212164889017,
+      "loss": 0.01704748347401619,
+      "loss_ce": 0.00027044457965530455,
+      "loss_xval": 0.0167236328125,
+      "num_input_tokens_seen": 258620400,
+      "step": 1754
+    },
+    {
+      "epoch": 0.5059094840011531,
+      "grad_norm": 0.35876812800575075,
+      "learning_rate": 0.0001,
+      "loss": 0.0048,
+      "num_input_tokens_seen": 258792904,
+      "step": 1755
+    },
+    {
+      "epoch": 0.5059094840011531,
+      "loss": 0.0015499168075621128,
+      "loss_ce": 0.00030537188285961747,
+      "loss_xval": 0.00124359130859375,
+      "num_input_tokens_seen": 258792904,
+      "step": 1755
+    },
+    {
+      "epoch": 0.5061977515134044,
+      "grad_norm": 9.14907245606204,
+      "learning_rate": 0.0001,
+      "loss": 0.0295,
+      "num_input_tokens_seen": 258927752,
+      "step": 1756
+    },
+    {
+      "epoch": 0.5061977515134044,
+      "loss": 0.02945183776319027,
+      "loss_ce": 0.01081322692334652,
+      "loss_xval": 0.0186767578125,
+      "num_input_tokens_seen": 258927752,
+      "step": 1756
+    },
+    {
+      "epoch": 0.5064860190256558,
+      "grad_norm": 6.475913983913407,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "num_input_tokens_seen": 259063000,
+      "step": 1757
+    },
+    {
+      "epoch": 0.5064860190256558,
+      "loss": 0.010992877185344696,
+      "loss_ce": 0.00026594821247272193,
+      "loss_xval": 0.0107421875,
+      "num_input_tokens_seen": 259063000,
+      "step": 1757
+    },
+    {
+      "epoch": 0.5067742865379071,
+      "grad_norm": 5.060122557936152,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "num_input_tokens_seen": 259235464,
+      "step": 1758
+    },
+    {
+      "epoch": 0.5067742865379071,
+      "loss": 0.0065870825201272964,
+      "loss_ce": 0.00031190545996651053,
+      "loss_xval": 0.00628662109375,
+      "num_input_tokens_seen": 259235464,
+      "step": 1758
+    },
+    {
+      "epoch": 0.5070625540501585,
+      "grad_norm": 10.671722543873086,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "num_input_tokens_seen": 259370384,
+      "step": 1759
+    },
+    {
+      "epoch": 0.5070625540501585,
+      "loss": 0.033719681203365326,
+      "loss_ce": 0.0073677548207342625,
+      "loss_xval": 0.0263671875,
+      "num_input_tokens_seen": 259370384,
+      "step": 1759
+    },
+    {
+      "epoch": 0.50735082156241,
+      "grad_norm": 3.14861007606756,
+      "learning_rate": 0.0001,
+      "loss": 0.0038,
+      "num_input_tokens_seen": 259505544,
+      "step": 1760
+    },
+    {
+      "epoch": 0.50735082156241,
+      "loss": 0.003241258906200528,
+      "loss_ce": 0.0002581655571702868,
+      "loss_xval": 0.00299072265625,
+      "num_input_tokens_seen": 259505544,
+      "step": 1760
+    },
+    {
+      "epoch": 0.5076390890746613,
+      "grad_norm": 8.30325635822721,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "num_input_tokens_seen": 259677936,
+      "step": 1761
+    },
+    {
+      "epoch": 0.5076390890746613,
+      "loss": 0.016355641186237335,
+      "loss_ce": 0.0003110232937615365,
+      "loss_xval": 0.0159912109375,
+      "num_input_tokens_seen": 259677936,
+      "step": 1761
+    },
+    {
+      "epoch": 0.5079273565869127,
+      "grad_norm": 9.637160112658565,
+      "learning_rate": 0.0001,
+      "loss": 0.03,
+      "num_input_tokens_seen": 259812696,
+      "step": 1762
+    },
+    {
+      "epoch": 0.5079273565869127,
+      "loss": 0.028667747974395752,
+      "loss_ce": 0.006618797779083252,
+      "loss_xval": 0.0220947265625,
+      "num_input_tokens_seen": 259812696,
+      "step": 1762
+    },
+    {
+      "epoch": 0.508215624099164,
+      "grad_norm": 0.7828280927300808,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "num_input_tokens_seen": 259947800,
+      "step": 1763
+    },
+    {
+      "epoch": 0.508215624099164,
+      "loss": 0.0010191877372562885,
+      "loss_ce": 0.0002243001654278487,
+      "loss_xval": 0.00079345703125,
+      "num_input_tokens_seen": 259947800,
+      "step": 1763
+    },
+    {
+      "epoch": 0.5085038916114154,
+      "grad_norm": 10.565618258218029,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "num_input_tokens_seen": 260120256,
+      "step": 1764
+    },
+    {
+      "epoch": 0.5085038916114154,
+      "loss": 0.027929434552788734,
+      "loss_ce": 0.000417837843997404,
+      "loss_xval": 0.0274658203125,
+      "num_input_tokens_seen": 260120256,
+      "step": 1764
+    },
+    {
+      "epoch": 0.5087921591236667,
+      "grad_norm": 7.7345371447851194,
+      "learning_rate": 0.0001,
+      "loss": 0.0213,
+      "num_input_tokens_seen": 260255000,
+      "step": 1765
+    },
+    {
+      "epoch": 0.5087921591236667,
+      "loss": 0.0220201276242733,
+      "loss_ce": 0.0070436252281069756,
+      "loss_xval": 0.01495361328125,
+      "num_input_tokens_seen": 260255000,
+      "step": 1765
+    },
+    {
+      "epoch": 0.5090804266359181,
+      "grad_norm": 4.785958281539146,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "num_input_tokens_seen": 260389984,
+      "step": 1766
+    },
+    {
+      "epoch": 0.5090804266359181,
+      "loss": 0.006561644375324249,
+      "loss_ce": 0.00021780251699965447,
+      "loss_xval": 0.00634765625,
+      "num_input_tokens_seen": 260389984,
+      "step": 1766
+    },
+    {
+      "epoch": 0.5093686941481695,
+      "grad_norm": 12.14625787907636,
+      "learning_rate": 0.0001,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 260562448,
+      "step": 1767
+    },
+    {
+      "epoch": 0.5093686941481695,
+      "loss": 0.03634582459926605,
+      "loss_ce": 0.0002740452764555812,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 260562448,
+      "step": 1767
+    },
+    {
+      "epoch": 0.5096569616604208,
+      "grad_norm": 4.580556944591601,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "num_input_tokens_seen": 260697216,
+      "step": 1768
+    },
+    {
+      "epoch": 0.5096569616604208,
+      "loss": 0.013978520408272743,
+      "loss_ce": 0.008794346824288368,
+      "loss_xval": 0.00518798828125,
+      "num_input_tokens_seen": 260697216,
+      "step": 1768
+    },
+    {
+      "epoch": 0.5099452291726723,
+      "grad_norm": 9.088690506902278,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "num_input_tokens_seen": 260832144,
+      "step": 1769
+    },
+    {
+      "epoch": 0.5099452291726723,
+      "loss": 0.022012032568454742,
+      "loss_ce": 0.00028351714718155563,
+      "loss_xval": 0.021728515625,
+      "num_input_tokens_seen": 260832144,
+      "step": 1769
+    },
+    {
+      "epoch": 0.5102334966849236,
+      "grad_norm": 11.86962561681491,
+      "learning_rate": 0.0001,
+      "loss": 0.0466,
+      "num_input_tokens_seen": 261004736,
+      "step": 1770
+    },
+    {
+      "epoch": 0.5102334966849236,
+      "loss": 0.036182355135679245,
+      "loss_ce": 0.000751446234062314,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 261004736,
+      "step": 1770
+    },
+    {
+      "epoch": 0.510521764197175,
+      "grad_norm": 0.4202323598180629,
+      "learning_rate": 0.0001,
+      "loss": 0.0042,
+      "num_input_tokens_seen": 261139512,
+      "step": 1771
+    },
+    {
+      "epoch": 0.510521764197175,
+      "loss": 0.00725182332098484,
+      "loss_ce": 0.006140792742371559,
+      "loss_xval": 0.0011138916015625,
+      "num_input_tokens_seen": 261139512,
+      "step": 1771
+    },
+    {
+      "epoch": 0.5108100317094263,
+      "grad_norm": 12.340846024432985,
+      "learning_rate": 0.0001,
+      "loss": 0.0462,
+      "num_input_tokens_seen": 261274520,
+      "step": 1772
+    },
+    {
+      "epoch": 0.5108100317094263,
+      "loss": 0.04091305285692215,
+      "loss_ce": 0.00023312387929763645,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 261274520,
+      "step": 1772
+    },
+    {
+      "epoch": 0.5110982992216777,
+      "grad_norm": 9.083396449002233,
+      "learning_rate": 0.0001,
+      "loss": 0.0313,
+      "num_input_tokens_seen": 261446968,
+      "step": 1773
+    },
+    {
+      "epoch": 0.5110982992216777,
+      "loss": 0.020383324474096298,
+      "loss_ce": 0.00030275824246928096,
+      "loss_xval": 0.02001953125,
+      "num_input_tokens_seen": 261446968,
+      "step": 1773
+    },
+    {
+      "epoch": 0.5113865667339291,
+      "grad_norm": 5.6260253773587685,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "num_input_tokens_seen": 261581808,
+      "step": 1774
+    },
+    {
+      "epoch": 0.5113865667339291,
+      "loss": 0.01707657054066658,
+      "loss_ce": 0.005991059355437756,
+      "loss_xval": 0.0111083984375,
+      "num_input_tokens_seen": 261581808,
+      "step": 1774
+    },
+    {
+      "epoch": 0.5116748342461804,
+      "grad_norm": 13.279502917108802,
+      "learning_rate": 0.0001,
+      "loss": 0.0533,
+      "num_input_tokens_seen": 261716872,
+      "step": 1775
+    },
+    {
+      "epoch": 0.5116748342461804,
+      "loss": 0.04808903858065605,
+      "loss_ce": 0.0002374750911258161,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 261716872,
+      "step": 1775
+    },
+    {
+      "epoch": 0.5119631017584318,
+      "grad_norm": 4.9284076730976185,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "num_input_tokens_seen": 261889328,
+      "step": 1776
+    },
+    {
+      "epoch": 0.5119631017584318,
+      "loss": 0.0062470389530062675,
+      "loss_ce": 0.00025796430418267846,
+      "loss_xval": 0.0059814453125,
+      "num_input_tokens_seen": 261889328,
+      "step": 1776
+    },
+    {
+      "epoch": 0.5122513692706832,
+      "grad_norm": 8.774623567615105,
+      "learning_rate": 0.0001,
+      "loss": 0.0281,
+      "num_input_tokens_seen": 262024152,
+      "step": 1777
+    },
+    {
+      "epoch": 0.5122513692706832,
+      "loss": 0.03278300166130066,
+      "loss_ce": 0.00882670283317566,
+      "loss_xval": 0.02392578125,
+      "num_input_tokens_seen": 262024152,
+      "step": 1777
+    },
+    {
+      "epoch": 0.5125396367829346,
+      "grad_norm": 10.381916712519102,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "num_input_tokens_seen": 262159096,
+      "step": 1778
+    },
+    {
+      "epoch": 0.5125396367829346,
+      "loss": 0.030212176963686943,
+      "loss_ce": 0.0003202082007192075,
+      "loss_xval": 0.0299072265625,
+      "num_input_tokens_seen": 262159096,
+      "step": 1778
+    },
+    {
+      "epoch": 0.512827904295186,
+      "grad_norm": 1.1520748376796106,
+      "learning_rate": 0.0001,
+      "loss": 0.0058,
+      "num_input_tokens_seen": 262331560,
+      "step": 1779
+    },
+    {
+      "epoch": 0.512827904295186,
+      "loss": 0.001818220131099224,
+      "loss_ce": 0.00024370371829718351,
+      "loss_xval": 0.0015716552734375,
+      "num_input_tokens_seen": 262331560,
+      "step": 1779
+    },
+    {
+      "epoch": 0.5131161718074373,
+      "grad_norm": 10.169366738375263,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "num_input_tokens_seen": 262466256,
+      "step": 1780
+    },
+    {
+      "epoch": 0.5131161718074373,
+      "loss": 0.0423487164080143,
+      "loss_ce": 0.010671471245586872,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 262466256,
+      "step": 1780
+    },
+    {
+      "epoch": 0.5134044393196887,
+      "grad_norm": 5.603413067960133,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "num_input_tokens_seen": 262601296,
+      "step": 1781
+    },
+    {
+      "epoch": 0.5134044393196887,
+      "loss": 0.009353495202958584,
+      "loss_ce": 0.0002134803362423554,
+      "loss_xval": 0.0091552734375,
+      "num_input_tokens_seen": 262601296,
+      "step": 1781
+    },
+    {
+      "epoch": 0.51369270683194,
+      "grad_norm": 5.697274476706217,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "num_input_tokens_seen": 262773776,
+      "step": 1782
+    },
+    {
+      "epoch": 0.51369270683194,
+      "loss": 0.01040747668594122,
+      "loss_ce": 0.0002374936593696475,
+      "loss_xval": 0.01019287109375,
+      "num_input_tokens_seen": 262773776,
+      "step": 1782
+    },
+    {
+      "epoch": 0.5139809743441914,
+      "grad_norm": 9.185995857737648,
+      "learning_rate": 0.0001,
+      "loss": 0.0307,
+      "num_input_tokens_seen": 262908520,
+      "step": 1783
+    },
+    {
+      "epoch": 0.5139809743441914,
+      "loss": 0.03518703579902649,
+      "loss_ce": 0.008499410934746265,
+      "loss_xval": 0.0267333984375,
+      "num_input_tokens_seen": 262908520,
+      "step": 1783
+    },
+    {
+      "epoch": 0.5142692418564427,
+      "grad_norm": 0.980429739430197,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "num_input_tokens_seen": 263043456,
+      "step": 1784
+    },
+    {
+      "epoch": 0.5142692418564427,
+      "loss": 0.0013816391583532095,
+      "loss_ce": 0.00020671238598879427,
+      "loss_xval": 0.0011749267578125,
+      "num_input_tokens_seen": 263043456,
+      "step": 1784
+    },
+    {
+      "epoch": 0.5145575093686942,
+      "grad_norm": 7.476174623428016,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "num_input_tokens_seen": 263216024,
+      "step": 1785
+    },
+    {
+      "epoch": 0.5145575093686942,
+      "loss": 0.01764541119337082,
+      "loss_ce": 0.00025039329193532467,
+      "loss_xval": 0.017333984375,
+      "num_input_tokens_seen": 263216024,
+      "step": 1785
+    },
+    {
+      "epoch": 0.5148457768809456,
+      "grad_norm": 5.115796542754154,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "num_input_tokens_seen": 263350784,
+      "step": 1786
+    },
+    {
+      "epoch": 0.5148457768809456,
+      "loss": 0.018444210290908813,
+      "loss_ce": 0.008831174112856388,
+      "loss_xval": 0.0096435546875,
+      "num_input_tokens_seen": 263350784,
+      "step": 1786
+    },
+    {
+      "epoch": 0.5151340443931969,
+      "grad_norm": 3.797885601558983,
+      "learning_rate": 0.0001,
+      "loss": 0.0056,
+      "num_input_tokens_seen": 263485776,
+      "step": 1787
+    },
+    {
+      "epoch": 0.5151340443931969,
+      "loss": 0.004911388270556927,
+      "loss_ce": 0.00021740313968621194,
+      "loss_xval": 0.00469970703125,
+      "num_input_tokens_seen": 263485776,
+      "step": 1787
+    },
+    {
+      "epoch": 0.5154223119054483,
+      "grad_norm": 6.4852572950830485,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "num_input_tokens_seen": 263658240,
+      "step": 1788
+    },
+    {
+      "epoch": 0.5154223119054483,
+      "loss": 0.01347002387046814,
+      "loss_ce": 0.0003093187406193465,
+      "loss_xval": 0.01318359375,
+      "num_input_tokens_seen": 263658240,
+      "step": 1788
+    },
+    {
+      "epoch": 0.5157105794176996,
+      "grad_norm": 0.36006105128606547,
+      "learning_rate": 0.0001,
+      "loss": 0.0046,
+      "num_input_tokens_seen": 263793016,
+      "step": 1789
+    },
+    {
+      "epoch": 0.5157105794176996,
+      "loss": 0.008257750421762466,
+      "loss_ce": 0.007303599268198013,
+      "loss_xval": 0.00095367431640625,
+      "num_input_tokens_seen": 263793016,
+      "step": 1789
+    },
+    {
+      "epoch": 0.515998846929951,
+      "grad_norm": 6.51905227217711,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "num_input_tokens_seen": 263928216,
+      "step": 1790
+    },
+    {
+      "epoch": 0.515998846929951,
+      "loss": 0.012620228342711926,
+      "loss_ce": 0.00019957451149821281,
+      "loss_xval": 0.012451171875,
+      "num_input_tokens_seen": 263928216,
+      "step": 1790
+    },
+    {
+      "epoch": 0.5162871144422023,
+      "grad_norm": 3.3272530057976772,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "num_input_tokens_seen": 264100696,
+      "step": 1791
+    },
+    {
+      "epoch": 0.5162871144422023,
+      "loss": 0.005464356858283281,
+      "loss_ce": 0.0015428480692207813,
+      "loss_xval": 0.00390625,
+      "num_input_tokens_seen": 264100696,
+      "step": 1791
+    },
+    {
+      "epoch": 0.5165753819544537,
+      "grad_norm": 4.536962600288751,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "num_input_tokens_seen": 264235528,
+      "step": 1792
+    },
+    {
+      "epoch": 0.5165753819544537,
+      "loss": 0.01935257576406002,
+      "loss_ce": 0.01350846141576767,
+      "loss_xval": 0.005859375,
+      "num_input_tokens_seen": 264235528,
+      "step": 1792
+    },
+    {
+      "epoch": 0.5168636494667052,
+      "grad_norm": 6.180124664142906,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "num_input_tokens_seen": 264370616,
+      "step": 1793
+    },
+    {
+      "epoch": 0.5168636494667052,
+      "loss": 0.01103239506483078,
+      "loss_ce": 0.0001910249120555818,
+      "loss_xval": 0.0108642578125,
+      "num_input_tokens_seen": 264370616,
+      "step": 1793
+    },
+    {
+      "epoch": 0.5171519169789565,
+      "grad_norm": 0.7206910189063704,
+      "learning_rate": 0.0001,
+      "loss": 0.0053,
+      "num_input_tokens_seen": 264543200,
+      "step": 1794
+    },
+    {
+      "epoch": 0.5171519169789565,
+      "loss": 0.001006332109682262,
+      "loss_ce": 0.00018807948799803853,
+      "loss_xval": 0.00081634521484375,
+      "num_input_tokens_seen": 264543200,
+      "step": 1794
+    },
+    {
+      "epoch": 0.5174401844912079,
+      "grad_norm": 7.206321887425217,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "num_input_tokens_seen": 264678144,
+      "step": 1795
+    },
+    {
+      "epoch": 0.5174401844912079,
+      "loss": 0.02243190072476864,
+      "loss_ce": 0.007913162931799889,
+      "loss_xval": 0.0145263671875,
+      "num_input_tokens_seen": 264678144,
+      "step": 1795
+    },
+    {
+      "epoch": 0.5177284520034592,
+      "grad_norm": 4.848419346001407,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "num_input_tokens_seen": 264813280,
+      "step": 1796
+    },
+    {
+      "epoch": 0.5177284520034592,
+      "loss": 0.007033906411379576,
+      "loss_ce": 0.0008197647985070944,
+      "loss_xval": 0.0062255859375,
+      "num_input_tokens_seen": 264813280,
+      "step": 1796
+    },
+    {
+      "epoch": 0.5180167195157106,
+      "grad_norm": 3.7155723030374883,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "num_input_tokens_seen": 264985632,
+      "step": 1797
+    },
+    {
+      "epoch": 0.5180167195157106,
+      "loss": 0.004432442598044872,
+      "loss_ce": 0.00022864610946271569,
+      "loss_xval": 0.00421142578125,
+      "num_input_tokens_seen": 264985632,
+      "step": 1797
+    },
+    {
+      "epoch": 0.5183049870279619,
+      "grad_norm": 7.999785529658378,
+      "learning_rate": 0.0001,
+      "loss": 0.0261,
+      "num_input_tokens_seen": 265120480,
+      "step": 1798
+    },
+    {
+      "epoch": 0.5183049870279619,
+      "loss": 0.029360122978687286,
+      "loss_ce": 0.011392900720238686,
+      "loss_xval": 0.0179443359375,
+      "num_input_tokens_seen": 265120480,
+      "step": 1798
+    },
+    {
+      "epoch": 0.5185932545402133,
+      "grad_norm": 2.7260965333996285,
+      "learning_rate": 0.0001,
+      "loss": 0.0039,
+      "num_input_tokens_seen": 265255616,
+      "step": 1799
+    },
+    {
+      "epoch": 0.5185932545402133,
+      "loss": 0.0042513152584433556,
+      "loss_ce": 0.0022514602169394493,
+      "loss_xval": 0.0019989013671875,
+      "num_input_tokens_seen": 265255616,
+      "step": 1799
+    },
+    {
+      "epoch": 0.5188815220524647,
+      "grad_norm": 5.770778443890454,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "num_input_tokens_seen": 265428080,
+      "step": 1800
+    },
+    {
+      "epoch": 0.5188815220524647,
+      "loss": 0.009834205731749535,
+      "loss_ce": 0.00018302083481103182,
+      "loss_xval": 0.0096435546875,
+      "num_input_tokens_seen": 265428080,
+      "step": 1800
+    },
+    {
+      "epoch": 0.519169789564716,
+      "grad_norm": 7.326870513573077,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "num_input_tokens_seen": 265562848,
+      "step": 1801
+    },
+    {
+      "epoch": 0.519169789564716,
+      "loss": 0.021131791174411774,
+      "loss_ce": 0.006170548032969236,
+      "loss_xval": 0.01495361328125,
+      "num_input_tokens_seen": 265562848,
+      "step": 1801
+    },
+    {
+      "epoch": 0.5194580570769675,
+      "grad_norm": 0.2707278759222347,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "num_input_tokens_seen": 265697912,
+      "step": 1802
+    },
+    {
+      "epoch": 0.5194580570769675,
+      "loss": 0.0003889158251695335,
+      "loss_ce": 0.00016402750043198466,
+      "loss_xval": 0.000225067138671875,
+      "num_input_tokens_seen": 265697912,
+      "step": 1802
+    },
+    {
+      "epoch": 0.5197463245892188,
+      "grad_norm": 7.100925150836463,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "num_input_tokens_seen": 265870368,
+      "step": 1803
+    },
+    {
+      "epoch": 0.5197463245892188,
+      "loss": 0.014973396435379982,
+      "loss_ce": 0.00018763082334771752,
+      "loss_xval": 0.0147705078125,
+      "num_input_tokens_seen": 265870368,
+      "step": 1803
+    },
+    {
+      "epoch": 0.5200345921014702,
+      "grad_norm": 6.053470067030953,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "num_input_tokens_seen": 266005056,
+      "step": 1804
+    },
+    {
+      "epoch": 0.5200345921014702,
+      "loss": 0.019502371549606323,
+      "loss_ce": 0.008950917981564999,
+      "loss_xval": 0.01055908203125,
+      "num_input_tokens_seen": 266005056,
+      "step": 1804
+    },
+    {
+      "epoch": 0.5203228596137215,
+      "grad_norm": 2.057397577661355,
+      "learning_rate": 0.0001,
+      "loss": 0.0019,
+      "num_input_tokens_seen": 266140120,
+      "step": 1805
+    },
+    {
+      "epoch": 0.5203228596137215,
+      "loss": 0.001837803632952273,
+      "loss_ce": 0.00018031768559012562,
+      "loss_xval": 0.00165557861328125,
+      "num_input_tokens_seen": 266140120,
+      "step": 1805
+    },
+    {
+      "epoch": 0.5206111271259729,
+      "grad_norm": 7.649635337597759,
+      "learning_rate": 0.0001,
+      "loss": 0.0226,
+      "num_input_tokens_seen": 266312648,
+      "step": 1806
+    },
+    {
+      "epoch": 0.5206111271259729,
+      "loss": 0.017032187432050705,
+      "loss_ce": 0.00018648413242772222,
+      "loss_xval": 0.016845703125,
+      "num_input_tokens_seen": 266312648,
+      "step": 1806
+    },
+    {
+      "epoch": 0.5208993946382243,
+      "grad_norm": 4.2351822034209405,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "num_input_tokens_seen": 266447576,
+      "step": 1807
+    },
+    {
+      "epoch": 0.5208993946382243,
+      "loss": 0.013426562771201134,
+      "loss_ce": 0.007937213405966759,
+      "loss_xval": 0.0054931640625,
+      "num_input_tokens_seen": 266447576,
+      "step": 1807
+    },
+    {
+      "epoch": 0.5211876621504756,
+      "grad_norm": 4.046343404175255,
+      "learning_rate": 0.0001,
+      "loss": 0.0057,
+      "num_input_tokens_seen": 266582712,
+      "step": 1808
+    },
+    {
+      "epoch": 0.5211876621504756,
+      "loss": 0.005520365200936794,
+      "loss_ce": 0.00015308592992369086,
+      "loss_xval": 0.00537109375,
+      "num_input_tokens_seen": 266582712,
+      "step": 1808
+    },
+    {
+      "epoch": 0.521475929662727,
+      "grad_norm": 7.701796112055166,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "num_input_tokens_seen": 266755160,
+      "step": 1809
+    },
+    {
+      "epoch": 0.521475929662727,
+      "loss": 0.018067223951220512,
+      "loss_ce": 0.00024495949037373066,
+      "loss_xval": 0.017822265625,
+      "num_input_tokens_seen": 266755160,
+      "step": 1809
+    },
+    {
+      "epoch": 0.5217641971749784,
+      "grad_norm": 2.614532008892686,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "num_input_tokens_seen": 266889984,
+      "step": 1810
+    },
+    {
+      "epoch": 0.5217641971749784,
+      "loss": 0.011933187022805214,
+      "loss_ce": 0.009657721035182476,
+      "loss_xval": 0.0022735595703125,
+      "num_input_tokens_seen": 266889984,
+      "step": 1810
+    },
+    {
+      "epoch": 0.5220524646872298,
+      "grad_norm": 5.425181300265865,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "num_input_tokens_seen": 267025000,
+      "step": 1811
+    },
+    {
+      "epoch": 0.5220524646872298,
+      "loss": 0.009021684527397156,
+      "loss_ce": 0.00024025217862799764,
+      "loss_xval": 0.0087890625,
+      "num_input_tokens_seen": 267025000,
+      "step": 1811
+    },
+    {
+      "epoch": 0.5223407321994811,
+      "grad_norm": 7.369734476588075,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "num_input_tokens_seen": 267197600,
+      "step": 1812
+    },
+    {
+      "epoch": 0.5223407321994811,
+      "loss": 0.01665113866329193,
+      "loss_ce": 0.00018690709839574993,
+      "loss_xval": 0.0164794921875,
+      "num_input_tokens_seen": 267197600,
+      "step": 1812
+    },
+    {
+      "epoch": 0.5226289997117325,
+      "grad_norm": 1.0926084372171434,
+      "learning_rate": 0.0001,
+      "loss": 0.0045,
+      "num_input_tokens_seen": 267332288,
+      "step": 1813
+    },
+    {
+      "epoch": 0.5226289997117325,
+      "loss": 0.008218837901949883,
+      "loss_ce": 0.007448745891451836,
+      "loss_xval": 0.00077056884765625,
+      "num_input_tokens_seen": 267332288,
+      "step": 1813
+    },
+    {
+      "epoch": 0.5229172672239839,
+      "grad_norm": 6.322536422783003,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "num_input_tokens_seen": 267467416,
+      "step": 1814
+    },
+    {
+      "epoch": 0.5229172672239839,
+      "loss": 0.011986615136265755,
+      "loss_ce": 0.0001534248294774443,
+      "loss_xval": 0.0118408203125,
+      "num_input_tokens_seen": 267467416,
+      "step": 1814
+    },
+    {
+      "epoch": 0.5232055347362352,
+      "grad_norm": 6.580074951937422,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "num_input_tokens_seen": 267640160,
+      "step": 1815
+    },
+    {
+      "epoch": 0.5232055347362352,
+      "loss": 0.013704424723982811,
+      "loss_ce": 0.00020039669470861554,
+      "loss_xval": 0.01348876953125,
+      "num_input_tokens_seen": 267640160,
+      "step": 1815
+    },
+    {
+      "epoch": 0.5234938022484866,
+      "grad_norm": 0.5802635231365518,
+      "learning_rate": 0.0001,
+      "loss": 0.0048,
+      "num_input_tokens_seen": 267774944,
+      "step": 1816
+    },
+    {
+      "epoch": 0.5234938022484866,
+      "loss": 0.008905167691409588,
+      "loss_ce": 0.007684941403567791,
+      "loss_xval": 0.001220703125,
+      "num_input_tokens_seen": 267774944,
+      "step": 1816
+    },
+    {
+      "epoch": 0.5237820697607379,
+      "grad_norm": 6.992165248213569,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "num_input_tokens_seen": 267909920,
+      "step": 1817
+    },
+    {
+      "epoch": 0.5237820697607379,
+      "loss": 0.014009732753038406,
+      "loss_ce": 0.00015475264808628708,
+      "loss_xval": 0.01385498046875,
+      "num_input_tokens_seen": 267909920,
+      "step": 1817
+    },
+    {
+      "epoch": 0.5240703372729894,
+      "grad_norm": 5.715831527574991,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "num_input_tokens_seen": 268082416,
+      "step": 1818
+    },
+    {
+      "epoch": 0.5240703372729894,
+      "loss": 0.010570320300757885,
+      "loss_ce": 0.0001714556710794568,
+      "loss_xval": 0.0103759765625,
+      "num_input_tokens_seen": 268082416,
+      "step": 1818
+    },
+    {
+      "epoch": 0.5243586047852407,
+      "grad_norm": 2.103081581176938,
+      "learning_rate": 0.0001,
+      "loss": 0.0061,
+      "num_input_tokens_seen": 268217104,
+      "step": 1819
+    },
+    {
+      "epoch": 0.5243586047852407,
+      "loss": 0.009669620543718338,
+      "loss_ce": 0.008071262389421463,
+      "loss_xval": 0.0016021728515625,
+      "num_input_tokens_seen": 268217104,
+      "step": 1819
+    },
+    {
+      "epoch": 0.5246468722974921,
+      "grad_norm": 7.97819003535755,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "num_input_tokens_seen": 268352232,
+      "step": 1820
+    },
+    {
+      "epoch": 0.5246468722974921,
+      "loss": 0.019121186807751656,
+      "loss_ce": 0.00018502911552786827,
+      "loss_xval": 0.0189208984375,
+      "num_input_tokens_seen": 268352232,
+      "step": 1820
+    },
+    {
+      "epoch": 0.5249351398097435,
+      "grad_norm": 5.277971127917007,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "num_input_tokens_seen": 268524752,
+      "step": 1821
+    },
+    {
+      "epoch": 0.5249351398097435,
+      "loss": 0.009244978427886963,
+      "loss_ce": 0.00015836962847970426,
+      "loss_xval": 0.00909423828125,
+      "num_input_tokens_seen": 268524752,
+      "step": 1821
+    },
+    {
+      "epoch": 0.5252234073219948,
+      "grad_norm": 3.166177126377048,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "num_input_tokens_seen": 268659552,
+      "step": 1822
+    },
+    {
+      "epoch": 0.5252234073219948,
+      "loss": 0.0116716418415308,
+      "loss_ce": 0.008358577266335487,
+      "loss_xval": 0.0033111572265625,
+      "num_input_tokens_seen": 268659552,
+      "step": 1822
+    },
+    {
+      "epoch": 0.5255116748342462,
+      "grad_norm": 8.588310727501373,
+      "learning_rate": 0.0001,
+      "loss": 0.0261,
+      "num_input_tokens_seen": 268794736,
+      "step": 1823
+    },
+    {
+      "epoch": 0.5255116748342462,
+      "loss": 0.0217589084059,
+      "loss_ce": 0.00013720503193326294,
+      "loss_xval": 0.0216064453125,
+      "num_input_tokens_seen": 268794736,
+      "step": 1823
+    },
+    {
+      "epoch": 0.5257999423464975,
+      "grad_norm": 5.4901507624366195,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "num_input_tokens_seen": 268967424,
+      "step": 1824
+    },
+    {
+      "epoch": 0.5257999423464975,
+      "loss": 0.010655020363628864,
+      "loss_ce": 0.0001798617740860209,
+      "loss_xval": 0.010498046875,
+      "num_input_tokens_seen": 268967424,
+      "step": 1824
+    },
+    {
+      "epoch": 0.5260882098587489,
+      "grad_norm": 3.5760439539186324,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "num_input_tokens_seen": 269102152,
+      "step": 1825
+    },
+    {
+      "epoch": 0.5260882098587489,
+      "loss": 0.011267166584730148,
+      "loss_ce": 0.007673721294850111,
+      "loss_xval": 0.00360107421875,
+      "num_input_tokens_seen": 269102152,
+      "step": 1825
+    },
+    {
+      "epoch": 0.5263764773710002,
+      "grad_norm": 8.878070935563086,
+      "learning_rate": 0.0001,
+      "loss": 0.0276,
+      "num_input_tokens_seen": 269237248,
+      "step": 1826
+    },
+    {
+      "epoch": 0.5263764773710002,
+      "loss": 0.022971712052822113,
+      "loss_ce": 0.000144564313814044,
+      "loss_xval": 0.0228271484375,
+      "num_input_tokens_seen": 269237248,
+      "step": 1826
+    },
+    {
+      "epoch": 0.5266647448832517,
+      "grad_norm": 4.414978131465267,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "num_input_tokens_seen": 269409816,
+      "step": 1827
+    },
+    {
+      "epoch": 0.5266647448832517,
+      "loss": 0.007491076365113258,
+      "loss_ce": 0.0002088190958602354,
+      "loss_xval": 0.007293701171875,
+      "num_input_tokens_seen": 269409816,
+      "step": 1827
+    },
+    {
+      "epoch": 0.5269530123955031,
+      "grad_norm": 5.027340329287891,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "num_input_tokens_seen": 269544552,
+      "step": 1828
+    },
+    {
+      "epoch": 0.5269530123955031,
+      "loss": 0.012437459081411362,
+      "loss_ce": 0.005471821874380112,
+      "loss_xval": 0.0069580078125,
+      "num_input_tokens_seen": 269544552,
+      "step": 1828
+    },
+    {
+      "epoch": 0.5272412799077544,
+      "grad_norm": 9.30005855051124,
+      "learning_rate": 0.0001,
+      "loss": 0.0303,
+      "num_input_tokens_seen": 269679608,
+      "step": 1829
+    },
+    {
+      "epoch": 0.5272412799077544,
+      "loss": 0.026024669408798218,
+      "loss_ce": 0.0001457621401641518,
+      "loss_xval": 0.02587890625,
+      "num_input_tokens_seen": 269679608,
+      "step": 1829
+    },
+    {
+      "epoch": 0.5275295474200058,
+      "grad_norm": 3.385395007624453,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "num_input_tokens_seen": 269852144,
+      "step": 1830
+    },
+    {
+      "epoch": 0.5275295474200058,
+      "loss": 0.005543527193367481,
+      "loss_ce": 0.00035553891211748123,
+      "loss_xval": 0.00518798828125,
+      "num_input_tokens_seen": 269852144,
+      "step": 1830
+    },
+    {
+      "epoch": 0.5278178149322571,
+      "grad_norm": 6.672355098963897,
+      "learning_rate": 0.0001,
+      "loss": 0.0213,
+      "num_input_tokens_seen": 269987008,
+      "step": 1831
+    },
+    {
+      "epoch": 0.5278178149322571,
+      "loss": 0.02160380780696869,
+      "loss_ce": 0.009183155372738838,
+      "loss_xval": 0.012451171875,
+      "num_input_tokens_seen": 269987008,
+      "step": 1831
+    },
+    {
+      "epoch": 0.5281060824445085,
+      "grad_norm": 9.189129921555727,
+      "learning_rate": 0.0001,
+      "loss": 0.0302,
+      "num_input_tokens_seen": 270122144,
+      "step": 1832
+    },
+    {
+      "epoch": 0.5281060824445085,
+      "loss": 0.027190502732992172,
+      "loss_ce": 0.00018244492821395397,
+      "loss_xval": 0.0269775390625,
+      "num_input_tokens_seen": 270122144,
+      "step": 1832
+    },
+    {
+      "epoch": 0.5283943499567598,
+      "grad_norm": 0.8676142019276292,
+      "learning_rate": 0.0001,
+      "loss": 0.0057,
+      "num_input_tokens_seen": 270294576,
+      "step": 1833
+    },
+    {
+      "epoch": 0.5283943499567598,
+      "loss": 0.0018882546573877335,
+      "loss_ce": 0.00017068708257284015,
+      "loss_xval": 0.00171661376953125,
+      "num_input_tokens_seen": 270294576,
+      "step": 1833
+    },
+    {
+      "epoch": 0.5286826174690112,
+      "grad_norm": 8.93930668699493,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "num_input_tokens_seen": 270429344,
+      "step": 1834
+    },
+    {
+      "epoch": 0.5286826174690112,
+      "loss": 0.02938574180006981,
+      "loss_ce": 0.006665404886007309,
+      "loss_xval": 0.022705078125,
+      "num_input_tokens_seen": 270429344,
+      "step": 1834
+    },
+    {
+      "epoch": 0.5289708849812627,
+      "grad_norm": 8.682313899729719,
+      "learning_rate": 0.0001,
+      "loss": 0.0271,
+      "num_input_tokens_seen": 270564392,
+      "step": 1835
+    },
+    {
+      "epoch": 0.5289708849812627,
+      "loss": 0.025439847260713577,
+      "loss_ce": 0.00014077400555834174,
+      "loss_xval": 0.0252685546875,
+      "num_input_tokens_seen": 270564392,
+      "step": 1835
+    },
+    {
+      "epoch": 0.529259152493514,
+      "grad_norm": 1.9778568650894202,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "num_input_tokens_seen": 270736976,
+      "step": 1836
+    },
+    {
+      "epoch": 0.529259152493514,
+      "loss": 0.0010150633752346039,
+      "loss_ce": 0.00016915427113417536,
+      "loss_xval": 0.00084686279296875,
+      "num_input_tokens_seen": 270736976,
+      "step": 1836
+    },
+    {
+      "epoch": 0.5295474200057654,
+      "grad_norm": 11.147218175352355,
+      "learning_rate": 0.0001,
+      "loss": 0.0491,
+      "num_input_tokens_seen": 270871816,
+      "step": 1837
+    },
+    {
+      "epoch": 0.5295474200057654,
+      "loss": 0.04592829942703247,
+      "loss_ce": 0.008117021061480045,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 270871816,
+      "step": 1837
+    },
+    {
+      "epoch": 0.5298356875180167,
+      "grad_norm": 7.234562197358699,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "num_input_tokens_seen": 271006896,
+      "step": 1838
+    },
+    {
+      "epoch": 0.5298356875180167,
+      "loss": 0.019972706213593483,
+      "loss_ce": 0.001356984837912023,
+      "loss_xval": 0.0185546875,
+      "num_input_tokens_seen": 271006896,
+      "step": 1838
+    },
+    {
+      "epoch": 0.5301239550302681,
+      "grad_norm": 6.134931735856935,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "num_input_tokens_seen": 271179408,
+      "step": 1839
+    },
+    {
+      "epoch": 0.5301239550302681,
+      "loss": 0.008071623742580414,
+      "loss_ce": 0.00021334676421247423,
+      "loss_xval": 0.00787353515625,
+      "num_input_tokens_seen": 271179408,
+      "step": 1839
+    },
+    {
+      "epoch": 0.5304122225425194,
+      "grad_norm": 13.488095571657878,
+      "learning_rate": 0.0001,
+      "loss": 0.0714,
+      "num_input_tokens_seen": 271314296,
+      "step": 1840
+    },
+    {
+      "epoch": 0.5304122225425194,
+      "loss": 0.0717129334807396,
+      "loss_ce": 0.008877238258719444,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 271314296,
+      "step": 1840
+    },
+    {
+      "epoch": 0.5307004900547708,
+      "grad_norm": 4.802872837119547,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "num_input_tokens_seen": 271449488,
+      "step": 1841
+    },
+    {
+      "epoch": 0.5307004900547708,
+      "loss": 0.012417205609381199,
+      "loss_ce": 0.00016439828323200345,
+      "loss_xval": 0.01226806640625,
+      "num_input_tokens_seen": 271449488,
+      "step": 1841
+    },
+    {
+      "epoch": 0.5309887575670222,
+      "grad_norm": 9.425082060032546,
+      "learning_rate": 0.0001,
+      "loss": 0.0396,
+      "num_input_tokens_seen": 271621952,
+      "step": 1842
+    },
+    {
+      "epoch": 0.5309887575670222,
+      "loss": 0.03309786319732666,
+      "loss_ce": 0.00019991426961496472,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 271621952,
+      "step": 1842
+    },
+    {
+      "epoch": 0.5312770250792735,
+      "grad_norm": 10.929547089579462,
+      "learning_rate": 0.0001,
+      "loss": 0.0485,
+      "num_input_tokens_seen": 271756720,
+      "step": 1843
+    },
+    {
+      "epoch": 0.5312770250792735,
+      "loss": 0.052716903388500214,
+      "loss_ce": 0.008374866098165512,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 271756720,
+      "step": 1843
+    },
+    {
+      "epoch": 0.531565292591525,
+      "grad_norm": 1.3531818049696558,
+      "learning_rate": 0.0001,
+      "loss": 0.002,
+      "num_input_tokens_seen": 271891840,
+      "step": 1844
+    },
+    {
+      "epoch": 0.531565292591525,
+      "loss": 0.0014669632073491812,
+      "loss_ce": 0.00018427114991936833,
+      "loss_xval": 0.00128173828125,
+      "num_input_tokens_seen": 271891840,
+      "step": 1844
+    },
+    {
+      "epoch": 0.5318535601037763,
+      "grad_norm": 12.025348100936883,
+      "learning_rate": 0.0001,
+      "loss": 0.0575,
+      "num_input_tokens_seen": 272064320,
+      "step": 1845
+    },
+    {
+      "epoch": 0.5318535601037763,
+      "loss": 0.04556536301970482,
+      "loss_ce": 0.00021624250803142786,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 272064320,
+      "step": 1845
+    },
+    {
+      "epoch": 0.5321418276160277,
+      "grad_norm": 8.611132853096239,
+      "learning_rate": 0.0001,
+      "loss": 0.0297,
+      "num_input_tokens_seen": 272199200,
+      "step": 1846
+    },
+    {
+      "epoch": 0.5321418276160277,
+      "loss": 0.032210852950811386,
+      "loss_ce": 0.004851843696087599,
+      "loss_xval": 0.02734375,
+      "num_input_tokens_seen": 272199200,
+      "step": 1846
+    },
+    {
+      "epoch": 0.532430095128279,
+      "grad_norm": 5.0086818591762965,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "num_input_tokens_seen": 272334240,
+      "step": 1847
+    },
+    {
+      "epoch": 0.532430095128279,
+      "loss": 0.006865452043712139,
+      "loss_ce": 0.00019736125250346959,
+      "loss_xval": 0.00665283203125,
+      "num_input_tokens_seen": 272334240,
+      "step": 1847
+    },
+    {
+      "epoch": 0.5327183626405304,
+      "grad_norm": 11.640411204028348,
+      "learning_rate": 0.0001,
+      "loss": 0.0558,
+      "num_input_tokens_seen": 272506904,
+      "step": 1848
+    },
+    {
+      "epoch": 0.5327183626405304,
+      "loss": 0.04578360170125961,
+      "loss_ce": 0.0002208554942626506,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 272506904,
+      "step": 1848
+    },
+    {
+      "epoch": 0.5330066301527818,
+      "grad_norm": 2.439540588150794,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "num_input_tokens_seen": 272641704,
+      "step": 1849
+    },
+    {
+      "epoch": 0.5330066301527818,
+      "loss": 0.016928596422076225,
+      "loss_ce": 0.011700552888214588,
+      "loss_xval": 0.005218505859375,
+      "num_input_tokens_seen": 272641704,
+      "step": 1849
+    },
+    {
+      "epoch": 0.5332948976650331,
+      "grad_norm": 10.359025160538259,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "num_input_tokens_seen": 272776880,
+      "step": 1850
+    },
+    {
+      "epoch": 0.5332948976650331,
+      "loss": 0.030838388949632645,
+      "loss_ce": 0.00021399807883426547,
+      "loss_xval": 0.0306396484375,
+      "num_input_tokens_seen": 272776880,
+      "step": 1850
+    },
+    {
+      "epoch": 0.5335831651772845,
+      "grad_norm": 9.252019092841493,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "num_input_tokens_seen": 272949280,
+      "step": 1851
+    },
+    {
+      "epoch": 0.5335831651772845,
+      "loss": 0.031272031366825104,
+      "loss_ce": 0.0002966886095236987,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 272949280,
+      "step": 1851
+    },
+    {
+      "epoch": 0.5338714326895359,
+      "grad_norm": 4.503404550676964,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "num_input_tokens_seen": 273084072,
+      "step": 1852
+    },
+    {
+      "epoch": 0.5338714326895359,
+      "loss": 0.014995909295976162,
+      "loss_ce": 0.007736540865153074,
+      "loss_xval": 0.00726318359375,
+      "num_input_tokens_seen": 273084072,
+      "step": 1852
+    },
+    {
+      "epoch": 0.5341597002017873,
+      "grad_norm": 13.679304448516996,
+      "learning_rate": 0.0001,
+      "loss": 0.069,
+      "num_input_tokens_seen": 273219208,
+      "step": 1853
+    },
+    {
+      "epoch": 0.5341597002017873,
+      "loss": 0.05733451992273331,
+      "loss_ce": 0.000205614764126949,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 273219208,
+      "step": 1853
+    },
+    {
+      "epoch": 0.5344479677140386,
+      "grad_norm": 6.295374009468746,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "num_input_tokens_seen": 273391976,
+      "step": 1854
+    },
+    {
+      "epoch": 0.5344479677140386,
+      "loss": 0.017546821385622025,
+      "loss_ce": 0.002890755422413349,
+      "loss_xval": 0.0146484375,
+      "num_input_tokens_seen": 273391976,
+      "step": 1854
+    },
+    {
+      "epoch": 0.53473623522629,
+      "grad_norm": 8.300922492017126,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "num_input_tokens_seen": 273526776,
+      "step": 1855
+    },
+    {
+      "epoch": 0.53473623522629,
+      "loss": 0.03134995698928833,
+      "loss_ce": 0.01007920689880848,
+      "loss_xval": 0.021240234375,
+      "num_input_tokens_seen": 273526776,
+      "step": 1855
+    },
+    {
+      "epoch": 0.5350245027385414,
+      "grad_norm": 14.090258374182888,
+      "learning_rate": 0.0001,
+      "loss": 0.0721,
+      "num_input_tokens_seen": 273661712,
+      "step": 1856
+    },
+    {
+      "epoch": 0.5350245027385414,
+      "loss": 0.06754553318023682,
+      "loss_ce": 0.0002542752190493047,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 273661712,
+      "step": 1856
+    },
+    {
+      "epoch": 0.5353127702507927,
+      "grad_norm": 4.217604913128783,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "num_input_tokens_seen": 273834160,
+      "step": 1857
+    },
+    {
+      "epoch": 0.5353127702507927,
+      "loss": 0.011369526386260986,
+      "loss_ce": 0.0009859215933829546,
+      "loss_xval": 0.0103759765625,
+      "num_input_tokens_seen": 273834160,
+      "step": 1857
+    },
+    {
+      "epoch": 0.5356010377630441,
+      "grad_norm": 10.066890717856893,
+      "learning_rate": 0.0001,
+      "loss": 0.0434,
+      "num_input_tokens_seen": 273969016,
+      "step": 1858
+    },
+    {
+      "epoch": 0.5356010377630441,
+      "loss": 0.034813039004802704,
+      "loss_ce": 0.006157036870718002,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 273969016,
+      "step": 1858
+    },
+    {
+      "epoch": 0.5358893052752954,
+      "grad_norm": 9.636156631624512,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "num_input_tokens_seen": 274104152,
+      "step": 1859
+    },
+    {
+      "epoch": 0.5358893052752954,
+      "loss": 0.029770299792289734,
+      "loss_ce": 0.00016824921476654708,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 274104152,
+      "step": 1859
+    },
+    {
+      "epoch": 0.5361775727875469,
+      "grad_norm": 6.198292477325853,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "num_input_tokens_seen": 274276760,
+      "step": 1860
+    },
+    {
+      "epoch": 0.5361775727875469,
+      "loss": 0.012494480237364769,
+      "loss_ce": 0.0002264140930492431,
+      "loss_xval": 0.01226806640625,
+      "num_input_tokens_seen": 274276760,
+      "step": 1860
+    },
+    {
+      "epoch": 0.5364658402997982,
+      "grad_norm": 13.371852634114951,
+      "learning_rate": 0.0001,
+      "loss": 0.0723,
+      "num_input_tokens_seen": 274411536,
+      "step": 1861
+    },
+    {
+      "epoch": 0.5364658402997982,
+      "loss": 0.06802873313426971,
+      "loss_ce": 0.008672050200402737,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 274411536,
+      "step": 1861
+    },
+    {
+      "epoch": 0.5367541078120496,
+      "grad_norm": 0.6687132978930205,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "num_input_tokens_seen": 274546536,
+      "step": 1862
+    },
+    {
+      "epoch": 0.5367541078120496,
+      "loss": 0.0009921127930283546,
+      "loss_ce": 0.00022941174393054098,
+      "loss_xval": 0.000762939453125,
+      "num_input_tokens_seen": 274546536,
+      "step": 1862
+    },
+    {
+      "epoch": 0.537042375324301,
+      "grad_norm": 12.9414839972833,
+      "learning_rate": 0.0001,
+      "loss": 0.0686,
+      "num_input_tokens_seen": 274719008,
+      "step": 1863
+    },
+    {
+      "epoch": 0.537042375324301,
+      "loss": 0.06119033694267273,
+      "loss_ce": 0.0002772501902654767,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 274719008,
+      "step": 1863
+    },
+    {
+      "epoch": 0.5373306428365523,
+      "grad_norm": 6.5873752554268545,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "num_input_tokens_seen": 274853752,
+      "step": 1864
+    },
+    {
+      "epoch": 0.5373306428365523,
+      "loss": 0.023443717509508133,
+      "loss_ce": 0.0076966481283307076,
+      "loss_xval": 0.0157470703125,
+      "num_input_tokens_seen": 274853752,
+      "step": 1864
+    },
+    {
+      "epoch": 0.5376189103488037,
+      "grad_norm": 9.18362001601729,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "num_input_tokens_seen": 274988920,
+      "step": 1865
+    },
+    {
+      "epoch": 0.5376189103488037,
+      "loss": 0.03302977979183197,
+      "loss_ce": 0.0003149337135255337,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 274988920,
+      "step": 1865
+    },
+    {
+      "epoch": 0.537907177861055,
+      "grad_norm": 9.78562809125422,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 275161448,
+      "step": 1866
+    },
+    {
+      "epoch": 0.537907177861055,
+      "loss": 0.03606872633099556,
+      "loss_ce": 0.0004241948190610856,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 275161448,
+      "step": 1866
+    },
+    {
+      "epoch": 0.5381954453733064,
+      "grad_norm": 4.664040810762385,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "num_input_tokens_seen": 275296272,
+      "step": 1867
+    },
+    {
+      "epoch": 0.5381954453733064,
+      "loss": 0.019615836441516876,
+      "loss_ce": 0.010201163589954376,
+      "loss_xval": 0.0093994140625,
+      "num_input_tokens_seen": 275296272,
+      "step": 1867
+    },
+    {
+      "epoch": 0.5384837128855579,
+      "grad_norm": 10.882200468816723,
+      "learning_rate": 0.0001,
+      "loss": 0.0486,
+      "num_input_tokens_seen": 275431280,
+      "step": 1868
+    },
+    {
+      "epoch": 0.5384837128855579,
+      "loss": 0.046006783843040466,
+      "loss_ce": 0.0003524859785102308,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 275431280,
+      "step": 1868
+    },
+    {
+      "epoch": 0.5387719803978092,
+      "grad_norm": 0.21921989623535026,
+      "learning_rate": 0.0001,
+      "loss": 0.005,
+      "num_input_tokens_seen": 275603792,
+      "step": 1869
+    },
+    {
+      "epoch": 0.5387719803978092,
+      "loss": 0.0015932258684188128,
+      "loss_ce": 0.0003429587814025581,
+      "loss_xval": 0.001251220703125,
+      "num_input_tokens_seen": 275603792,
+      "step": 1869
+    },
+    {
+      "epoch": 0.5390602479100606,
+      "grad_norm": 9.76831164206979,
+      "learning_rate": 0.0001,
+      "loss": 0.0434,
+      "num_input_tokens_seen": 275738648,
+      "step": 1870
+    },
+    {
+      "epoch": 0.5390602479100606,
+      "loss": 0.04737325385212898,
+      "loss_ce": 0.0074562616646289825,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 275738648,
+      "step": 1870
+    },
+    {
+      "epoch": 0.5393485154223119,
+      "grad_norm": 3.2196833292638347,
+      "learning_rate": 0.0001,
+      "loss": 0.0053,
+      "num_input_tokens_seen": 275873688,
+      "step": 1871
+    },
+    {
+      "epoch": 0.5393485154223119,
+      "loss": 0.00434330478310585,
+      "loss_ce": 0.00030735484324395657,
+      "loss_xval": 0.0040283203125,
+      "num_input_tokens_seen": 275873688,
+      "step": 1871
+    },
+    {
+      "epoch": 0.5396367829345633,
+      "grad_norm": 7.713986595540021,
+      "learning_rate": 0.0001,
+      "loss": 0.0286,
+      "num_input_tokens_seen": 276046200,
+      "step": 1872
+    },
+    {
+      "epoch": 0.5396367829345633,
+      "loss": 0.026184234768152237,
+      "loss_ce": 0.00029007039847783744,
+      "loss_xval": 0.02587890625,
+      "num_input_tokens_seen": 276046200,
+      "step": 1872
+    },
+    {
+      "epoch": 0.5399250504468146,
+      "grad_norm": 5.44441982546994,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "num_input_tokens_seen": 276181072,
+      "step": 1873
+    },
+    {
+      "epoch": 0.5399250504468146,
+      "loss": 0.022309284657239914,
+      "loss_ce": 0.010231953114271164,
+      "loss_xval": 0.0120849609375,
+      "num_input_tokens_seen": 276181072,
+      "step": 1873
+    },
+    {
+      "epoch": 0.540213317959066,
+      "grad_norm": 4.975650159557211,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "num_input_tokens_seen": 276316168,
+      "step": 1874
+    },
+    {
+      "epoch": 0.540213317959066,
+      "loss": 0.011838541366159916,
+      "loss_ce": 0.0002571204677224159,
+      "loss_xval": 0.0115966796875,
+      "num_input_tokens_seen": 276316168,
+      "step": 1874
+    },
+    {
+      "epoch": 0.5405015854713174,
+      "grad_norm": 6.232380580900199,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "num_input_tokens_seen": 276488528,
+      "step": 1875
+    },
+    {
+      "epoch": 0.5405015854713174,
+      "loss": 0.016596008092164993,
+      "loss_ce": 0.0003225088585168123,
+      "loss_xval": 0.0162353515625,
+      "num_input_tokens_seen": 276488528,
+      "step": 1875
+    },
+    {
+      "epoch": 0.5407898529835687,
+      "grad_norm": 2.164403279556159,
+      "learning_rate": 0.0001,
+      "loss": 0.006,
+      "num_input_tokens_seen": 276623368,
+      "step": 1876
+    },
+    {
+      "epoch": 0.5407898529835687,
+      "loss": 0.009121393784880638,
+      "loss_ce": 0.006442523095756769,
+      "loss_xval": 0.002685546875,
+      "num_input_tokens_seen": 276623368,
+      "step": 1876
+    },
+    {
+      "epoch": 0.5410781204958202,
+      "grad_norm": 5.647682673041729,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "num_input_tokens_seen": 276758480,
+      "step": 1877
+    },
+    {
+      "epoch": 0.5410781204958202,
+      "loss": 0.014722021296620369,
+      "loss_ce": 0.0005389765137806535,
+      "loss_xval": 0.01416015625,
+      "num_input_tokens_seen": 276758480,
+      "step": 1877
+    },
+    {
+      "epoch": 0.5413663880080715,
+      "grad_norm": 0.9526118525776436,
+      "learning_rate": 0.0001,
+      "loss": 0.0067,
+      "num_input_tokens_seen": 276930976,
+      "step": 1878
+    },
+    {
+      "epoch": 0.5413663880080715,
+      "loss": 0.0019027346279472113,
+      "loss_ce": 0.00040117441676557064,
+      "loss_xval": 0.00150299072265625,
+      "num_input_tokens_seen": 276930976,
+      "step": 1878
+    },
+    {
+      "epoch": 0.5416546555203229,
+      "grad_norm": 5.5350916551675615,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "num_input_tokens_seen": 277065800,
+      "step": 1879
+    },
+    {
+      "epoch": 0.5416546555203229,
+      "loss": 0.019967705011367798,
+      "loss_ce": 0.006677298806607723,
+      "loss_xval": 0.0133056640625,
+      "num_input_tokens_seen": 277065800,
+      "step": 1879
+    },
+    {
+      "epoch": 0.5419429230325742,
+      "grad_norm": 0.4270329198572373,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "num_input_tokens_seen": 277200872,
+      "step": 1880
+    },
+    {
+      "epoch": 0.5419429230325742,
+      "loss": 0.0007261130958795547,
+      "loss_ce": 0.00027693252195604146,
+      "loss_xval": 0.00045013427734375,
+      "num_input_tokens_seen": 277200872,
+      "step": 1880
+    },
+    {
+      "epoch": 0.5422311905448256,
+      "grad_norm": 5.050655608523017,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "num_input_tokens_seen": 277373448,
+      "step": 1881
+    },
+    {
+      "epoch": 0.5422311905448256,
+      "loss": 0.010783690959215164,
+      "loss_ce": 0.00045349146239459515,
+      "loss_xval": 0.01031494140625,
+      "num_input_tokens_seen": 277373448,
+      "step": 1881
+    },
+    {
+      "epoch": 0.542519458057077,
+      "grad_norm": 1.734342283968104,
+      "learning_rate": 0.0001,
+      "loss": 0.0062,
+      "num_input_tokens_seen": 277508216,
+      "step": 1882
+    },
+    {
+      "epoch": 0.542519458057077,
+      "loss": 0.010175340808928013,
+      "loss_ce": 0.007627123035490513,
+      "loss_xval": 0.0025482177734375,
+      "num_input_tokens_seen": 277508216,
+      "step": 1882
+    },
+    {
+      "epoch": 0.5428077255693283,
+      "grad_norm": 4.093064236279448,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "num_input_tokens_seen": 277643312,
+      "step": 1883
+    },
+    {
+      "epoch": 0.5428077255693283,
+      "loss": 0.007740652654320002,
+      "loss_ce": 0.0003019930445589125,
+      "loss_xval": 0.0074462890625,
+      "num_input_tokens_seen": 277643312,
+      "step": 1883
+    },
+    {
+      "epoch": 0.5430959930815797,
+      "grad_norm": 2.7869194069931886,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "num_input_tokens_seen": 277815768,
+      "step": 1884
+    },
+    {
+      "epoch": 0.5430959930815797,
+      "loss": 0.003793244482949376,
+      "loss_ce": 0.00024176128499675542,
+      "loss_xval": 0.0035552978515625,
+      "num_input_tokens_seen": 277815768,
+      "step": 1884
+    },
+    {
+      "epoch": 0.543384260593831,
+      "grad_norm": 3.0191744495092117,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "num_input_tokens_seen": 277950600,
+      "step": 1885
+    },
+    {
+      "epoch": 0.543384260593831,
+      "loss": 0.014148483984172344,
+      "loss_ce": 0.009771118871867657,
+      "loss_xval": 0.004364013671875,
+      "num_input_tokens_seen": 277950600,
+      "step": 1885
+    },
+    {
+      "epoch": 0.5436725281060825,
+      "grad_norm": 3.7978812660289627,
+      "learning_rate": 0.0001,
+      "loss": 0.0067,
+      "num_input_tokens_seen": 278085648,
+      "step": 1886
+    },
+    {
+      "epoch": 0.5436725281060825,
+      "loss": 0.006118366494774818,
+      "loss_ce": 0.00019795639673247933,
+      "loss_xval": 0.00592041015625,
+      "num_input_tokens_seen": 278085648,
+      "step": 1886
+    },
+    {
+      "epoch": 0.5439607956183338,
+      "grad_norm": 1.212810871733268,
+      "learning_rate": 0.0001,
+      "loss": 0.0054,
+      "num_input_tokens_seen": 278258016,
+      "step": 1887
+    },
+    {
+      "epoch": 0.5439607956183338,
+      "loss": 0.001508941175416112,
+      "loss_ce": 0.00021194410510361195,
+      "loss_xval": 0.0012969970703125,
+      "num_input_tokens_seen": 278258016,
+      "step": 1887
+    },
+    {
+      "epoch": 0.5442490631305852,
+      "grad_norm": 3.660909833810732,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "num_input_tokens_seen": 278392800,
+      "step": 1888
+    },
+    {
+      "epoch": 0.5442490631305852,
+      "loss": 0.012451540678739548,
+      "loss_ce": 0.006084810942411423,
+      "loss_xval": 0.006378173828125,
+      "num_input_tokens_seen": 278392800,
+      "step": 1888
+    },
+    {
+      "epoch": 0.5445373306428366,
+      "grad_norm": 0.2878029698109918,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "num_input_tokens_seen": 278527768,
+      "step": 1889
+    },
+    {
+      "epoch": 0.5445373306428366,
+      "loss": 0.0005817347555421293,
+      "loss_ce": 0.00019430456450209022,
+      "loss_xval": 0.0003871917724609375,
+      "num_input_tokens_seen": 278527768,
+      "step": 1889
+    },
+    {
+      "epoch": 0.5448255981550879,
+      "grad_norm": 3.167419414539796,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "num_input_tokens_seen": 278700264,
+      "step": 1890
+    },
+    {
+      "epoch": 0.5448255981550879,
+      "loss": 0.0045308684930205345,
+      "loss_ce": 0.0001973719772649929,
+      "loss_xval": 0.00433349609375,
+      "num_input_tokens_seen": 278700264,
+      "step": 1890
+    },
+    {
+      "epoch": 0.5451138656673393,
+      "grad_norm": 1.49057272645955,
+      "learning_rate": 0.0001,
+      "loss": 0.0053,
+      "num_input_tokens_seen": 278835072,
+      "step": 1891
+    },
+    {
+      "epoch": 0.5451138656673393,
+      "loss": 0.009009573608636856,
+      "loss_ce": 0.007309172302484512,
+      "loss_xval": 0.00170135498046875,
+      "num_input_tokens_seen": 278835072,
+      "step": 1891
+    },
+    {
+      "epoch": 0.5454021331795906,
+      "grad_norm": 2.4543003292367245,
+      "learning_rate": 0.0001,
+      "loss": 0.0031,
+      "num_input_tokens_seen": 278970104,
+      "step": 1892
+    },
+    {
+      "epoch": 0.5454021331795906,
+      "loss": 0.0029379362240433693,
+      "loss_ce": 0.00036873770295642316,
+      "loss_xval": 0.0025634765625,
+      "num_input_tokens_seen": 278970104,
+      "step": 1892
+    },
+    {
+      "epoch": 0.545690400691842,
+      "grad_norm": 2.4981723443169925,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "num_input_tokens_seen": 279142720,
+      "step": 1893
+    },
+    {
+      "epoch": 0.545690400691842,
+      "loss": 0.0029300451278686523,
+      "loss_ce": 0.00019872195844072849,
+      "loss_xval": 0.0027313232421875,
+      "num_input_tokens_seen": 279142720,
+      "step": 1893
+    },
+    {
+      "epoch": 0.5459786682040934,
+      "grad_norm": 1.2333966495433941,
+      "learning_rate": 0.0001,
+      "loss": 0.005,
+      "num_input_tokens_seen": 279277584,
+      "step": 1894
+    },
+    {
+      "epoch": 0.5459786682040934,
+      "loss": 0.008969184011220932,
+      "loss_ce": 0.008108016103506088,
+      "loss_xval": 0.00086212158203125,
+      "num_input_tokens_seen": 279277584,
+      "step": 1894
+    },
+    {
+      "epoch": 0.5462669357163448,
+      "grad_norm": 3.0088690684316033,
+      "learning_rate": 0.0001,
+      "loss": 0.0045,
+      "num_input_tokens_seen": 279412664,
+      "step": 1895
+    },
+    {
+      "epoch": 0.5462669357163448,
+      "loss": 0.004141474142670631,
+      "loss_ce": 0.0001284123572986573,
+      "loss_xval": 0.0040283203125,
+      "num_input_tokens_seen": 279412664,
+      "step": 1895
+    },
+    {
+      "epoch": 0.5465552032285962,
+      "grad_norm": 0.3048082625937777,
+      "learning_rate": 0.0001,
+      "loss": 0.0047,
+      "num_input_tokens_seen": 279585048,
+      "step": 1896
+    },
+    {
+      "epoch": 0.5465552032285962,
+      "loss": 0.000647731008939445,
+      "loss_ce": 0.00013370058150030673,
+      "loss_xval": 0.000514984130859375,
+      "num_input_tokens_seen": 279585048,
+      "step": 1896
+    },
+    {
+      "epoch": 0.5468434707408475,
+      "grad_norm": 2.8539549965482647,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "num_input_tokens_seen": 279719824,
+      "step": 1897
+    },
+    {
+      "epoch": 0.5468434707408475,
+      "loss": 0.010153815150260925,
+      "loss_ce": 0.006661459803581238,
+      "loss_xval": 0.0034942626953125,
+      "num_input_tokens_seen": 279719824,
+      "step": 1897
+    },
+    {
+      "epoch": 0.5471317382530989,
+      "grad_norm": 1.8076048875343496,
+      "learning_rate": 0.0001,
+      "loss": 0.0026,
+      "num_input_tokens_seen": 279854936,
+      "step": 1898
+    },
+    {
+      "epoch": 0.5471317382530989,
+      "loss": 0.0016184869455173612,
+      "loss_ce": 0.0001383843191433698,
+      "loss_xval": 0.0014801025390625,
+      "num_input_tokens_seen": 279854936,
+      "step": 1898
+    },
+    {
+      "epoch": 0.5474200057653502,
+      "grad_norm": 1.9186558700507859,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "num_input_tokens_seen": 280027392,
+      "step": 1899
+    },
+    {
+      "epoch": 0.5474200057653502,
+      "loss": 0.001938400324434042,
+      "loss_ce": 0.0002294160076417029,
+      "loss_xval": 0.001708984375,
+      "num_input_tokens_seen": 280027392,
+      "step": 1899
+    },
+    {
+      "epoch": 0.5477082732776016,
+      "grad_norm": 2.81423870825424,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "num_input_tokens_seen": 280162184,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5477082732776016,
+      "loss": 0.013608219102025032,
+      "loss_ce": 0.010356190614402294,
+      "loss_xval": 0.0032501220703125,
+      "num_input_tokens_seen": 280162184,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5479965407898529,
+      "grad_norm": 0.24242660438385522,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "num_input_tokens_seen": 280297216,
+      "step": 1901
+    },
+    {
+      "epoch": 0.5479965407898529,
+      "loss": 0.0003102511982433498,
+      "loss_ce": 9.877391858026385e-05,
+      "loss_xval": 0.0002117156982421875,
+      "num_input_tokens_seen": 280297216,
+      "step": 1901
+    },
+    {
+      "epoch": 0.5482848083021044,
+      "grad_norm": 3.045715546334397,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "num_input_tokens_seen": 280469696,
+      "step": 1902
+    },
+    {
+      "epoch": 0.5482848083021044,
+      "loss": 0.004016020335257053,
+      "loss_ce": 0.00010786311759147793,
+      "loss_xval": 0.00390625,
+      "num_input_tokens_seen": 280469696,
+      "step": 1902
+    },
+    {
+      "epoch": 0.5485730758143558,
+      "grad_norm": 1.9843275332507238,
+      "learning_rate": 0.0001,
+      "loss": 0.0062,
+      "num_input_tokens_seen": 280604424,
+      "step": 1903
+    },
+    {
+      "epoch": 0.5485730758143558,
+      "loss": 0.01020785328000784,
+      "loss_ce": 0.008550367318093777,
+      "loss_xval": 0.00165557861328125,
+      "num_input_tokens_seen": 280604424,
+      "step": 1903
+    },
+    {
+      "epoch": 0.5488613433266071,
+      "grad_norm": 1.5118493970879978,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "num_input_tokens_seen": 280739472,
+      "step": 1904
+    },
+    {
+      "epoch": 0.5488613433266071,
+      "loss": 0.0010831006802618504,
+      "loss_ce": 7.172907498897985e-05,
+      "loss_xval": 0.00101470947265625,
+      "num_input_tokens_seen": 280739472,
+      "step": 1904
+    },
+    {
+      "epoch": 0.5491496108388585,
+      "grad_norm": 3.1270755163143464,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "num_input_tokens_seen": 280912072,
+      "step": 1905
+    },
+    {
+      "epoch": 0.5491496108388585,
+      "loss": 0.004354151897132397,
+      "loss_ce": 0.00036397893563844264,
+      "loss_xval": 0.003997802734375,
+      "num_input_tokens_seen": 280912072,
+      "step": 1905
+    },
+    {
+      "epoch": 0.5494378783511098,
+      "grad_norm": 1.1936946363165124,
+      "learning_rate": 0.0001,
+      "loss": 0.0055,
+      "num_input_tokens_seen": 281046904,
+      "step": 1906
+    },
+    {
+      "epoch": 0.5494378783511098,
+      "loss": 0.010138033889234066,
+      "loss_ce": 0.009292601607739925,
+      "loss_xval": 0.00084686279296875,
+      "num_input_tokens_seen": 281046904,
+      "step": 1906
+    },
+    {
+      "epoch": 0.5497261458633612,
+      "grad_norm": 1.7984578180285498,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "num_input_tokens_seen": 281181984,
+      "step": 1907
+    },
+    {
+      "epoch": 0.5497261458633612,
+      "loss": 0.0018197790486738086,
+      "loss_ce": 0.00028245607973076403,
+      "loss_xval": 0.0015411376953125,
+      "num_input_tokens_seen": 281181984,
+      "step": 1907
+    },
+    {
+      "epoch": 0.5500144133756125,
+      "grad_norm": 2.5448443442932898,
+      "learning_rate": 0.0001,
+      "loss": 0.0057,
+      "num_input_tokens_seen": 281354480,
+      "step": 1908
+    },
+    {
+      "epoch": 0.5500144133756125,
+      "loss": 0.002872986253350973,
+      "loss_ce": 7.490596908610314e-05,
+      "loss_xval": 0.0027923583984375,
+      "num_input_tokens_seen": 281354480,
+      "step": 1908
+    },
+    {
+      "epoch": 0.5503026808878639,
+      "grad_norm": 0.48281903020689454,
+      "learning_rate": 0.0001,
+      "loss": 0.0062,
+      "num_input_tokens_seen": 281489320,
+      "step": 1909
+    },
+    {
+      "epoch": 0.5503026808878639,
+      "loss": 0.012108450755476952,
+      "loss_ce": 0.011801367625594139,
+      "loss_xval": 0.0003070831298828125,
+      "num_input_tokens_seen": 281489320,
+      "step": 1909
+    },
+    {
+      "epoch": 0.5505909484001154,
+      "grad_norm": 2.0299099430728376,
+      "learning_rate": 0.0001,
+      "loss": 0.002,
+      "num_input_tokens_seen": 281624280,
+      "step": 1910
+    },
+    {
+      "epoch": 0.5505909484001154,
+      "loss": 0.001749855582602322,
+      "loss_ce": 6.185208621900529e-05,
+      "loss_xval": 0.00168609619140625,
+      "num_input_tokens_seen": 281624280,
+      "step": 1910
+    },
+    {
+      "epoch": 0.5508792159123667,
+      "grad_norm": 2.240271411263848,
+      "learning_rate": 0.0001,
+      "loss": 0.0056,
+      "num_input_tokens_seen": 281796704,
+      "step": 1911
+    },
+    {
+      "epoch": 0.5508792159123667,
+      "loss": 0.0022659413516521454,
+      "loss_ce": 7.153685146477073e-05,
+      "loss_xval": 0.002197265625,
+      "num_input_tokens_seen": 281796704,
+      "step": 1911
+    },
+    {
+      "epoch": 0.5511674834246181,
+      "grad_norm": 0.1330004944553925,
+      "learning_rate": 0.0001,
+      "loss": 0.0029,
+      "num_input_tokens_seen": 281931464,
+      "step": 1912
+    },
+    {
+      "epoch": 0.5511674834246181,
+      "loss": 0.005604051053524017,
+      "loss_ce": 0.005324862897396088,
+      "loss_xval": 0.000278472900390625,
+      "num_input_tokens_seen": 281931464,
+      "step": 1912
+    },
+    {
+      "epoch": 0.5514557509368694,
+      "grad_norm": 2.229550340309766,
+      "learning_rate": 0.0001,
+      "loss": 0.0024,
+      "num_input_tokens_seen": 282066520,
+      "step": 1913
+    },
+    {
+      "epoch": 0.5514557509368694,
+      "loss": 0.0020462670363485813,
+      "loss_ce": 6.262438546400517e-05,
+      "loss_xval": 0.001983642578125,
+      "num_input_tokens_seen": 282066520,
+      "step": 1913
+    },
+    {
+      "epoch": 0.5517440184491208,
+      "grad_norm": 2.3761471501605627,
+      "learning_rate": 0.0001,
+      "loss": 0.0065,
+      "num_input_tokens_seen": 282239008,
+      "step": 1914
+    },
+    {
+      "epoch": 0.5517440184491208,
+      "loss": 0.002356612589210272,
+      "loss_ce": 0.0001564860576763749,
+      "loss_xval": 0.002197265625,
+      "num_input_tokens_seen": 282239008,
+      "step": 1914
+    },
+    {
+      "epoch": 0.5520322859613721,
+      "grad_norm": 0.21665882302988462,
+      "learning_rate": 0.0001,
+      "loss": 0.0054,
+      "num_input_tokens_seen": 282373776,
+      "step": 1915
+    },
+    {
+      "epoch": 0.5520322859613721,
+      "loss": 0.010722009465098381,
+      "loss_ce": 0.010517684742808342,
+      "loss_xval": 0.0002040863037109375,
+      "num_input_tokens_seen": 282373776,
+      "step": 1915
+    },
+    {
+      "epoch": 0.5523205534736235,
+      "grad_norm": 2.1117128245187184,
+      "learning_rate": 0.0001,
+      "loss": 0.0022,
+      "num_input_tokens_seen": 282508896,
+      "step": 1916
+    },
+    {
+      "epoch": 0.5523205534736235,
+      "loss": 0.0018395634833723307,
+      "loss_ce": 6.668292917311192e-05,
+      "loss_xval": 0.00177001953125,
+      "num_input_tokens_seen": 282508896,
+      "step": 1916
+    },
+    {
+      "epoch": 0.5526088209858749,
+      "grad_norm": 2.2123876638762874,
+      "learning_rate": 0.0001,
+      "loss": 0.0048,
+      "num_input_tokens_seen": 282681552,
+      "step": 1917
+    },
+    {
+      "epoch": 0.5526088209858749,
+      "loss": 0.0022970284335315228,
+      "loss_ce": 0.00014172433293424547,
+      "loss_xval": 0.0021514892578125,
+      "num_input_tokens_seen": 282681552,
+      "step": 1917
+    },
+    {
+      "epoch": 0.5528970884981262,
+      "grad_norm": 0.1690305145590322,
+      "learning_rate": 0.0001,
+      "loss": 0.0039,
+      "num_input_tokens_seen": 282816320,
+      "step": 1918
+    },
+    {
+      "epoch": 0.5528970884981262,
+      "loss": 0.0076433876529335976,
+      "loss_ce": 0.007427559234201908,
+      "loss_xval": 0.0002155303955078125,
+      "num_input_tokens_seen": 282816320,
+      "step": 1918
+    },
+    {
+      "epoch": 0.5531853560103777,
+      "grad_norm": 1.9800549126606803,
+      "learning_rate": 0.0001,
+      "loss": 0.0019,
+      "num_input_tokens_seen": 282951264,
+      "step": 1919
+    },
+    {
+      "epoch": 0.5531853560103777,
+      "loss": 0.001511584734544158,
+      "loss_ce": 5.5323973356280476e-05,
+      "loss_xval": 0.00145721435546875,
+      "num_input_tokens_seen": 282951264,
+      "step": 1919
+    },
+    {
+      "epoch": 0.553473623522629,
+      "grad_norm": 2.107008911020967,
+      "learning_rate": 0.0001,
+      "loss": 0.0044,
+      "num_input_tokens_seen": 283123816,
+      "step": 1920
+    },
+    {
+      "epoch": 0.553473623522629,
+      "loss": 0.002012682380154729,
+      "loss_ce": 6.337207014439628e-05,
+      "loss_xval": 0.001953125,
+      "num_input_tokens_seen": 283123816,
+      "step": 1920
+    },
+    {
+      "epoch": 0.5537618910348804,
+      "grad_norm": 0.16338842906769652,
+      "learning_rate": 0.0001,
+      "loss": 0.0048,
+      "num_input_tokens_seen": 283258736,
+      "step": 1921
+    },
+    {
+      "epoch": 0.5537618910348804,
+      "loss": 0.009544363245368004,
+      "loss_ce": 0.009310713037848473,
+      "loss_xval": 0.00023365020751953125,
+      "num_input_tokens_seen": 283258736,
+      "step": 1921
+    },
+    {
+      "epoch": 0.5540501585471317,
+      "grad_norm": 1.9086123569129076,
+      "learning_rate": 0.0001,
+      "loss": 0.002,
+      "num_input_tokens_seen": 283394088,
+      "step": 1922
+    },
+    {
+      "epoch": 0.5540501585471317,
+      "loss": 0.001516035059466958,
+      "loss_ce": 0.00015323443221859634,
+      "loss_xval": 0.00136566162109375,
+      "num_input_tokens_seen": 283394088,
+      "step": 1922
+    },
+    {
+      "epoch": 0.5543384260593831,
+      "grad_norm": 1.9160847816709878,
+      "learning_rate": 0.0001,
+      "loss": 0.0068,
+      "num_input_tokens_seen": 283566624,
+      "step": 1923
+    },
+    {
+      "epoch": 0.5543384260593831,
+      "loss": 0.0018914049724116921,
+      "loss_ce": 8.896048530004919e-05,
+      "loss_xval": 0.001800537109375,
+      "num_input_tokens_seen": 283566624,
+      "step": 1923
+    },
+    {
+      "epoch": 0.5546266935716345,
+      "grad_norm": 0.051652544586469715,
+      "learning_rate": 0.0001,
+      "loss": 0.0036,
+      "num_input_tokens_seen": 283701376,
+      "step": 1924
+    },
+    {
+      "epoch": 0.5546266935716345,
+      "loss": 0.007118854206055403,
+      "loss_ce": 0.006899747531861067,
+      "loss_xval": 0.0002193450927734375,
+      "num_input_tokens_seen": 283701376,
+      "step": 1924
+    },
+    {
+      "epoch": 0.5549149610838858,
+      "grad_norm": 1.980739323892179,
+      "learning_rate": 0.0001,
+      "loss": 0.0019,
+      "num_input_tokens_seen": 283836400,
+      "step": 1925
+    },
+    {
+      "epoch": 0.5549149610838858,
+      "loss": 0.0013998394133523107,
+      "loss_ce": 5.5158656323328614e-05,
+      "loss_xval": 0.0013427734375,
+      "num_input_tokens_seen": 283836400,
+      "step": 1925
+    },
+    {
+      "epoch": 0.5552032285961372,
+      "grad_norm": 2.1895260282147717,
+      "learning_rate": 0.0001,
+      "loss": 0.0065,
+      "num_input_tokens_seen": 284008896,
+      "step": 1926
+    },
+    {
+      "epoch": 0.5552032285961372,
+      "loss": 0.0025523242074996233,
+      "loss_ce": 5.751205389969982e-05,
+      "loss_xval": 0.00250244140625,
+      "num_input_tokens_seen": 284008896,
+      "step": 1926
+    },
+    {
+      "epoch": 0.5554914961083885,
+      "grad_norm": 0.5431506925705571,
+      "learning_rate": 0.0001,
+      "loss": 0.0059,
+      "num_input_tokens_seen": 284143592,
+      "step": 1927
+    },
+    {
+      "epoch": 0.5554914961083885,
+      "loss": 0.011616826057434082,
+      "loss_ce": 0.01074516773223877,
+      "loss_xval": 0.0008697509765625,
+      "num_input_tokens_seen": 284143592,
+      "step": 1927
+    },
+    {
+      "epoch": 0.55577976362064,
+      "grad_norm": 1.1463366590129798,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "num_input_tokens_seen": 284278800,
+      "step": 1928
+    },
+    {
+      "epoch": 0.55577976362064,
+      "loss": 0.0005050962208770216,
+      "loss_ce": 6.75981500535272e-05,
+      "loss_xval": 0.0004367828369140625,
+      "num_input_tokens_seen": 284278800,
+      "step": 1928
+    },
+    {
+      "epoch": 0.5560680311328913,
+      "grad_norm": 1.4869739800687145,
+      "learning_rate": 0.0001,
+      "loss": 0.0041,
+      "num_input_tokens_seen": 284451328,
+      "step": 1929
+    },
+    {
+      "epoch": 0.5560680311328913,
+      "loss": 0.0015234043821692467,
+      "loss_ce": 5.665321441483684e-05,
+      "loss_xval": 0.00146484375,
+      "num_input_tokens_seen": 284451328,
+      "step": 1929
+    },
+    {
+      "epoch": 0.5563562986451427,
+      "grad_norm": 0.4843964706871106,
+      "learning_rate": 0.0001,
+      "loss": 0.0038,
+      "num_input_tokens_seen": 284586216,
+      "step": 1930
+    },
+    {
+      "epoch": 0.5563562986451427,
+      "loss": 0.007321392651647329,
+      "loss_ce": 0.006811653729528189,
+      "loss_xval": 0.00051116943359375,
+      "num_input_tokens_seen": 284586216,
+      "step": 1930
+    },
+    {
+      "epoch": 0.5566445661573941,
+      "grad_norm": 0.958553380827203,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "num_input_tokens_seen": 284721176,
+      "step": 1931
+    },
+    {
+      "epoch": 0.5566445661573941,
+      "loss": 0.00031205732375383377,
+      "loss_ce": 4.8127942136488855e-05,
+      "loss_xval": 0.000263214111328125,
+      "num_input_tokens_seen": 284721176,
+      "step": 1931
+    },
+    {
+      "epoch": 0.5569328336696454,
+      "grad_norm": 1.4978071655368779,
+      "learning_rate": 0.0001,
+      "loss": 0.0044,
+      "num_input_tokens_seen": 284893792,
+      "step": 1932
+    },
+    {
+      "epoch": 0.5569328336696454,
+      "loss": 0.0014730643015354872,
+      "loss_ce": 5.208961374592036e-05,
+      "loss_xval": 0.0014190673828125,
+      "num_input_tokens_seen": 284893792,
+      "step": 1932
+    },
+    {
+      "epoch": 0.5572211011818968,
+      "grad_norm": 0.8566778501180569,
+      "learning_rate": 0.0001,
+      "loss": 0.0065,
+      "num_input_tokens_seen": 285028528,
+      "step": 1933
+    },
+    {
+      "epoch": 0.5572211011818968,
+      "loss": 0.0127099072560668,
+      "loss_ce": 0.01148157473653555,
+      "loss_xval": 0.00122833251953125,
+      "num_input_tokens_seen": 285028528,
+      "step": 1933
+    },
+    {
+      "epoch": 0.5575093686941481,
+      "grad_norm": 0.044511992402982,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "num_input_tokens_seen": 285163648,
+      "step": 1934
+    },
+    {
+      "epoch": 0.5575093686941481,
+      "loss": 0.0001727144990582019,
+      "loss_ce": 4.5816203055437654e-05,
+      "loss_xval": 0.00012683868408203125,
+      "num_input_tokens_seen": 285163648,
+      "step": 1934
+    },
+    {
+      "epoch": 0.5577976362063995,
+      "grad_norm": 0.2350786549418237,
+      "learning_rate": 0.0001,
+      "loss": 0.004,
+      "num_input_tokens_seen": 285336112,
+      "step": 1935
+    },
+    {
+      "epoch": 0.5577976362063995,
+      "loss": 0.0003692085447255522,
+      "loss_ce": 6.03372827754356e-05,
+      "loss_xval": 0.000308990478515625,
+      "num_input_tokens_seen": 285336112,
+      "step": 1935
+    },
+    {
+      "epoch": 0.5580859037186509,
+      "grad_norm": 0.19000274994071045,
+      "learning_rate": 0.0001,
+      "loss": 0.0041,
+      "num_input_tokens_seen": 285470816,
+      "step": 1936
+    },
+    {
+      "epoch": 0.5580859037186509,
+      "loss": 0.007709057070314884,
+      "loss_ce": 0.007516414858400822,
+      "loss_xval": 0.0001926422119140625,
+      "num_input_tokens_seen": 285470816,
+      "step": 1936
+    },
+    {
+      "epoch": 0.5583741712309023,
+      "grad_norm": 0.7432285967026308,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "num_input_tokens_seen": 285605896,
+      "step": 1937
+    },
+    {
+      "epoch": 0.5583741712309023,
+      "loss": 0.00025071113486774266,
+      "loss_ce": 7.595031638629735e-05,
+      "loss_xval": 0.00017452239990234375,
+      "num_input_tokens_seen": 285605896,
+      "step": 1937
+    },
+    {
+      "epoch": 0.5586624387431537,
+      "grad_norm": 0.6835868718300878,
+      "learning_rate": 0.0001,
+      "loss": 0.0051,
+      "num_input_tokens_seen": 285778400,
+      "step": 1938
+    },
+    {
+      "epoch": 0.5586624387431537,
+      "loss": 0.000513557402882725,
+      "loss_ce": 8.464240818284452e-05,
+      "loss_xval": 0.0004291534423828125,
+      "num_input_tokens_seen": 285778400,
+      "step": 1938
+    },
+    {
+      "epoch": 0.558950706255405,
+      "grad_norm": 0.21160262356402706,
+      "learning_rate": 0.0001,
+      "loss": 0.0039,
+      "num_input_tokens_seen": 285913136,
+      "step": 1939
+    },
+    {
+      "epoch": 0.558950706255405,
+      "loss": 0.007476561237126589,
+      "loss_ce": 0.007180326152592897,
+      "loss_xval": 0.0002956390380859375,
+      "num_input_tokens_seen": 285913136,
+      "step": 1939
+    },
+    {
+      "epoch": 0.5592389737676564,
+      "grad_norm": 0.9580294627468874,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "num_input_tokens_seen": 286048264,
+      "step": 1940
+    },
+    {
+      "epoch": 0.5592389737676564,
+      "loss": 0.0003514891432132572,
+      "loss_ce": 7.075125904520974e-05,
+      "loss_xval": 0.0002803802490234375,
+      "num_input_tokens_seen": 286048264,
+      "step": 1940
+    },
+    {
+      "epoch": 0.5595272412799077,
+      "grad_norm": 1.1122987198740457,
+      "learning_rate": 0.0001,
+      "loss": 0.005,
+      "num_input_tokens_seen": 286220784,
+      "step": 1941
+    },
+    {
+      "epoch": 0.5595272412799077,
+      "loss": 0.000870700110681355,
+      "loss_ce": 8.201141463359818e-05,
+      "loss_xval": 0.000789642333984375,
+      "num_input_tokens_seen": 286220784,
+      "step": 1941
+    },
+    {
+      "epoch": 0.5598155087921591,
+      "grad_norm": 0.654693458550546,
+      "learning_rate": 0.0001,
+      "loss": 0.0042,
+      "num_input_tokens_seen": 286355584,
+      "step": 1942
+    },
+    {
+      "epoch": 0.5598155087921591,
+      "loss": 0.008090196177363396,
+      "loss_ce": 0.0076016769744455814,
+      "loss_xval": 0.00048828125,
+      "num_input_tokens_seen": 286355584,
+      "step": 1942
+    },
+    {
+      "epoch": 0.5601037763044104,
+      "grad_norm": 0.06036064124455231,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "num_input_tokens_seen": 286490616,
+      "step": 1943
+    },
+    {
+      "epoch": 0.5601037763044104,
+      "loss": 0.00018729933071881533,
+      "loss_ce": 9.282596147386357e-05,
+      "loss_xval": 9.441375732421875e-05,
+      "num_input_tokens_seen": 286490616,
+      "step": 1943
+    },
+    {
+      "epoch": 0.5603920438166619,
+      "grad_norm": 0.14486727422936166,
+      "learning_rate": 0.0001,
+      "loss": 0.0034,
+      "num_input_tokens_seen": 286663224,
+      "step": 1944
+    },
+    {
+      "epoch": 0.5603920438166619,
+      "loss": 0.00021684798412024975,
+      "loss_ce": 5.150470678927377e-05,
+      "loss_xval": 0.00016498565673828125,
+      "num_input_tokens_seen": 286663224,
+      "step": 1944
+    },
+    {
+      "epoch": 0.5606803113289133,
+      "grad_norm": 0.4009978276180538,
+      "learning_rate": 0.0001,
+      "loss": 0.0049,
+      "num_input_tokens_seen": 286797968,
+      "step": 1945
+    },
+    {
+      "epoch": 0.5606803113289133,
+      "loss": 0.009587855078279972,
+      "loss_ce": 0.009425134398043156,
+      "loss_xval": 0.00016307830810546875,
+      "num_input_tokens_seen": 286797968,
+      "step": 1945
+    },
+    {
+      "epoch": 0.5609685788411646,
+      "grad_norm": 1.2402087854563595,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "num_input_tokens_seen": 286933040,
+      "step": 1946
+    },
+    {
+      "epoch": 0.5609685788411646,
+      "loss": 0.0003389227786101401,
+      "loss_ce": 3.267412830609828e-05,
+      "loss_xval": 0.0003070831298828125,
+      "num_input_tokens_seen": 286933040,
+      "step": 1946
+    },
+    {
+      "epoch": 0.561256846353416,
+      "grad_norm": 1.2725344281821624,
+      "learning_rate": 0.0001,
+      "loss": 0.0064,
+      "num_input_tokens_seen": 287105616,
+      "step": 1947
+    },
+    {
+      "epoch": 0.561256846353416,
+      "loss": 0.0011517228558659554,
+      "loss_ce": 0.00024334801128134131,
+      "loss_xval": 0.00090789794921875,
+      "num_input_tokens_seen": 287105616,
+      "step": 1947
+    },
+    {
+      "epoch": 0.5615451138656673,
+      "grad_norm": 1.1853218638615954,
+      "learning_rate": 0.0001,
+      "loss": 0.0049,
+      "num_input_tokens_seen": 287240328,
+      "step": 1948
+    },
+    {
+      "epoch": 0.5615451138656673,
+      "loss": 0.009418094530701637,
+      "loss_ce": 0.008385742083191872,
+      "loss_xval": 0.00102996826171875,
+      "num_input_tokens_seen": 287240328,
+      "step": 1948
+    },
+    {
+      "epoch": 0.5618333813779187,
+      "grad_norm": 1.1085830242231645,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "num_input_tokens_seen": 287375576,
+      "step": 1949
+    },
+    {
+      "epoch": 0.5618333813779187,
+      "loss": 0.0008955195662565529,
+      "loss_ce": 3.3397995139239356e-05,
+      "loss_xval": 0.00086212158203125,
+      "num_input_tokens_seen": 287375576,
+      "step": 1949
+    },
+    {
+      "epoch": 0.56212164889017,
+      "grad_norm": 1.5125347817040997,
+      "learning_rate": 0.0001,
+      "loss": 0.0056,
+      "num_input_tokens_seen": 287548080,
+      "step": 1950
+    },
+    {
+      "epoch": 0.56212164889017,
+      "loss": 0.0017731888219714165,
+      "loss_ce": 0.0006511909887194633,
+      "loss_xval": 0.00112152099609375,
+      "num_input_tokens_seen": 287548080,
+      "step": 1950
+    },
+    {
+      "epoch": 0.5624099164024214,
+      "grad_norm": 2.704419859032451,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "num_input_tokens_seen": 287682832,
+      "step": 1951
+    },
+    {
+      "epoch": 0.5624099164024214,
+      "loss": 0.011266487650573254,
+      "loss_ce": 0.008250969462096691,
+      "loss_xval": 0.003021240234375,
+      "num_input_tokens_seen": 287682832,
+      "step": 1951
+    },
+    {
+      "epoch": 0.5626981839146729,
+      "grad_norm": 4.881457725475292,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "num_input_tokens_seen": 287817840,
+      "step": 1952
+    },
+    {
+      "epoch": 0.5626981839146729,
+      "loss": 0.009428413584828377,
+      "loss_ce": 6.714653864037246e-05,
+      "loss_xval": 0.00933837890625,
+      "num_input_tokens_seen": 287817840,
+      "step": 1952
+    },
+    {
+      "epoch": 0.5629864514269242,
+      "grad_norm": 8.142683543107829,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "num_input_tokens_seen": 287990176,
+      "step": 1953
+    },
+    {
+      "epoch": 0.5629864514269242,
+      "loss": 0.028103932738304138,
+      "loss_ce": 5.8278888900531456e-05,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 287990176,
+      "step": 1953
+    },
+    {
+      "epoch": 0.5632747189391756,
+      "grad_norm": 12.683316002509716,
+      "learning_rate": 0.0001,
+      "loss": 0.0747,
+      "num_input_tokens_seen": 288124992,
+      "step": 1954
+    },
+    {
+      "epoch": 0.5632747189391756,
+      "loss": 0.0827786922454834,
+      "loss_ce": 0.00831579975783825,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 288124992,
+      "step": 1954
+    },
+    {
+      "epoch": 0.5635629864514269,
+      "grad_norm": 16.530990540062938,
+      "learning_rate": 0.0001,
+      "loss": 0.1233,
+      "num_input_tokens_seen": 288260168,
+      "step": 1955
+    },
+    {
+      "epoch": 0.5635629864514269,
+      "loss": 0.12602970004081726,
+      "loss_ce": 5.31386467628181e-05,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 288260168,
+      "step": 1955
+    },
+    {
+      "epoch": 0.5638512539636783,
+      "grad_norm": 13.143860245392554,
+      "learning_rate": 0.0001,
+      "loss": 0.0874,
+      "num_input_tokens_seen": 288432592,
+      "step": 1956
+    },
+    {
+      "epoch": 0.5638512539636783,
+      "loss": 0.09267762303352356,
+      "loss_ce": 8.729415276320651e-05,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 288432592,
+      "step": 1956
+    },
+    {
+      "epoch": 0.5641395214759297,
+      "grad_norm": 2.6561824504781044,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "num_input_tokens_seen": 288567392,
+      "step": 1957
+    },
+    {
+      "epoch": 0.5641395214759297,
+      "loss": 0.024685639888048172,
+      "loss_ce": 0.008061189204454422,
+      "loss_xval": 0.0166015625,
+      "num_input_tokens_seen": 288567392,
+      "step": 1957
+    },
+    {
+      "epoch": 0.564427788988181,
+      "grad_norm": 6.84049611185504,
+      "learning_rate": 0.0001,
+      "loss": 0.0231,
+      "num_input_tokens_seen": 288702376,
+      "step": 1958
+    },
+    {
+      "epoch": 0.564427788988181,
+      "loss": 0.024861274287104607,
+      "loss_ce": 0.00012677747872658074,
+      "loss_xval": 0.0247802734375,
+      "num_input_tokens_seen": 288702376,
+      "step": 1958
+    },
+    {
+      "epoch": 0.5647160565004324,
+      "grad_norm": 10.650617694885725,
+      "learning_rate": 0.0001,
+      "loss": 0.055,
+      "num_input_tokens_seen": 288875056,
+      "step": 1959
+    },
+    {
+      "epoch": 0.5647160565004324,
+      "loss": 0.0596018023788929,
+      "loss_ce": 0.00021459662821143866,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 288875056,
+      "step": 1959
+    },
+    {
+      "epoch": 0.5650043240126837,
+      "grad_norm": 8.25795596922048,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "num_input_tokens_seen": 289009872,
+      "step": 1960
+    },
+    {
+      "epoch": 0.5650043240126837,
+      "loss": 0.048701632767915726,
+      "loss_ce": 0.008052217774093151,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 289009872,
+      "step": 1960
+    },
+    {
+      "epoch": 0.5652925915249352,
+      "grad_norm": 2.4798138856675305,
+      "learning_rate": 0.0001,
+      "loss": 0.0046,
+      "num_input_tokens_seen": 289144976,
+      "step": 1961
+    },
+    {
+      "epoch": 0.5652925915249352,
+      "loss": 0.005279832519590855,
+      "loss_ce": 0.00028448639204725623,
+      "loss_xval": 0.0050048828125,
+      "num_input_tokens_seen": 289144976,
+      "step": 1961
+    },
+    {
+      "epoch": 0.5655808590371865,
+      "grad_norm": 1.561935056225871,
+      "learning_rate": 0.0001,
+      "loss": 0.0064,
+      "num_input_tokens_seen": 289317568,
+      "step": 1962
+    },
+    {
+      "epoch": 0.5655808590371865,
+      "loss": 0.0034121754579246044,
+      "loss_ce": 0.0003909351071342826,
+      "loss_xval": 0.003021240234375,
+      "num_input_tokens_seen": 289317568,
+      "step": 1962
+    },
+    {
+      "epoch": 0.5658691265494379,
+      "grad_norm": 2.9034700037078753,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "num_input_tokens_seen": 289452392,
+      "step": 1963
+    },
+    {
+      "epoch": 0.5658691265494379,
+      "loss": 0.018562277778983116,
+      "loss_ce": 0.008354147896170616,
+      "loss_xval": 0.01019287109375,
+      "num_input_tokens_seen": 289452392,
+      "step": 1963
+    },
+    {
+      "epoch": 0.5661573940616893,
+      "grad_norm": 3.2432257799309285,
+      "learning_rate": 0.0001,
+      "loss": 0.0059,
+      "num_input_tokens_seen": 289587400,
+      "step": 1964
+    },
+    {
+      "epoch": 0.5661573940616893,
+      "loss": 0.007287283428013325,
+      "loss_ce": 0.0004742340533994138,
+      "loss_xval": 0.006805419921875,
+      "num_input_tokens_seen": 289587400,
+      "step": 1964
+    },
+    {
+      "epoch": 0.5664456615739406,
+      "grad_norm": 4.038297008837272,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "num_input_tokens_seen": 289759840,
+      "step": 1965
+    },
+    {
+      "epoch": 0.5664456615739406,
+      "loss": 0.0066073378548026085,
+      "loss_ce": 0.0006297073559835553,
+      "loss_xval": 0.0059814453125,
+      "num_input_tokens_seen": 289759840,
+      "step": 1965
+    },
+    {
+      "epoch": 0.566733929086192,
+      "grad_norm": 5.565205838171376,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "num_input_tokens_seen": 289894704,
+      "step": 1966
+    },
+    {
+      "epoch": 0.566733929086192,
+      "loss": 0.018734395503997803,
+      "loss_ce": 0.007687033619731665,
+      "loss_xval": 0.01104736328125,
+      "num_input_tokens_seen": 289894704,
+      "step": 1966
+    },
+    {
+      "epoch": 0.5670221965984433,
+      "grad_norm": 5.467379562241336,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "num_input_tokens_seen": 290029936,
+      "step": 1967
+    },
+    {
+      "epoch": 0.5670221965984433,
+      "loss": 0.012764690443873405,
+      "loss_ce": 0.0004889954579994082,
+      "loss_xval": 0.01226806640625,
+      "num_input_tokens_seen": 290029936,
+      "step": 1967
+    },
+    {
+      "epoch": 0.5673104641106947,
+      "grad_norm": 2.99187780606278,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "num_input_tokens_seen": 290202328,
+      "step": 1968
+    },
+    {
+      "epoch": 0.5673104641106947,
+      "loss": 0.00564954150468111,
+      "loss_ce": 0.00032040924998000264,
+      "loss_xval": 0.005340576171875,
+      "num_input_tokens_seen": 290202328,
+      "step": 1968
+    },
+    {
+      "epoch": 0.567598731622946,
+      "grad_norm": 0.48645630930362904,
+      "learning_rate": 0.0001,
+      "loss": 0.0053,
+      "num_input_tokens_seen": 290337224,
+      "step": 1969
+    },
+    {
+      "epoch": 0.567598731622946,
+      "loss": 0.009277354925870895,
+      "loss_ce": 0.008250724524259567,
+      "loss_xval": 0.00102996826171875,
+      "num_input_tokens_seen": 290337224,
+      "step": 1969
+    },
+    {
+      "epoch": 0.5678869991351975,
+      "grad_norm": 2.6207641878735055,
+      "learning_rate": 0.0001,
+      "loss": 0.004,
+      "num_input_tokens_seen": 290472392,
+      "step": 1970
+    },
+    {
+      "epoch": 0.5678869991351975,
+      "loss": 0.003790928516536951,
+      "loss_ce": 0.0001993910118471831,
+      "loss_xval": 0.0035858154296875,
+      "num_input_tokens_seen": 290472392,
+      "step": 1970
+    },
+    {
+      "epoch": 0.5681752666474489,
+      "grad_norm": 1.7294043568848192,
+      "learning_rate": 0.0001,
+      "loss": 0.0064,
+      "num_input_tokens_seen": 290644992,
+      "step": 1971
+    },
+    {
+      "epoch": 0.5681752666474489,
+      "loss": 0.0025520434137433767,
+      "loss_ce": 0.00023270744713954628,
+      "loss_xval": 0.0023193359375,
+      "num_input_tokens_seen": 290644992,
+      "step": 1971
+    },
+    {
+      "epoch": 0.5684635341597002,
+      "grad_norm": 0.5996015104252971,
+      "learning_rate": 0.0001,
+      "loss": 0.0045,
+      "num_input_tokens_seen": 290779760,
+      "step": 1972
+    },
+    {
+      "epoch": 0.5684635341597002,
+      "loss": 0.007963595911860466,
+      "loss_ce": 0.007266698870807886,
+      "loss_xval": 0.000698089599609375,
+      "num_input_tokens_seen": 290779760,
+      "step": 1972
+    },
+    {
+      "epoch": 0.5687518016719516,
+      "grad_norm": 2.5431713335506845,
+      "learning_rate": 0.0001,
+      "loss": 0.0035,
+      "num_input_tokens_seen": 290914776,
+      "step": 1973
+    },
+    {
+      "epoch": 0.5687518016719516,
+      "loss": 0.0034647989086806774,
+      "loss_ce": 8.306984091177583e-05,
+      "loss_xval": 0.003387451171875,
+      "num_input_tokens_seen": 290914776,
+      "step": 1973
+    },
+    {
+      "epoch": 0.5690400691842029,
+      "grad_norm": 3.0238650907286813,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "num_input_tokens_seen": 291087272,
+      "step": 1974
+    },
+    {
+      "epoch": 0.5690400691842029,
+      "loss": 0.005696265026926994,
+      "loss_ce": 0.00045487057650461793,
+      "loss_xval": 0.0052490234375,
+      "num_input_tokens_seen": 291087272,
+      "step": 1974
+    },
+    {
+      "epoch": 0.5693283366964543,
+      "grad_norm": 2.6482166457440774,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "num_input_tokens_seen": 291222040,
+      "step": 1975
+    },
+    {
+      "epoch": 0.5693283366964543,
+      "loss": 0.01379525288939476,
+      "loss_ce": 0.009589549154043198,
+      "loss_xval": 0.00421142578125,
+      "num_input_tokens_seen": 291222040,
+      "step": 1975
+    },
+    {
+      "epoch": 0.5696166042087056,
+      "grad_norm": 2.197137165884688,
+      "learning_rate": 0.0001,
+      "loss": 0.0026,
+      "num_input_tokens_seen": 291356984,
+      "step": 1976
+    },
+    {
+      "epoch": 0.5696166042087056,
+      "loss": 0.0022935783490538597,
+      "loss_ce": 5.339750714483671e-05,
+      "loss_xval": 0.0022430419921875,
+      "num_input_tokens_seen": 291356984,
+      "step": 1976
+    },
+    {
+      "epoch": 0.569904871720957,
+      "grad_norm": 1.9490519769257666,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "num_input_tokens_seen": 291529568,
+      "step": 1977
+    },
+    {
+      "epoch": 0.569904871720957,
+      "loss": 0.0018777311779558659,
+      "loss_ce": 7.337937859119847e-05,
+      "loss_xval": 0.001800537109375,
+      "num_input_tokens_seen": 291529568,
+      "step": 1977
+    },
+    {
+      "epoch": 0.5701931392332085,
+      "grad_norm": 1.9049005772357561,
+      "learning_rate": 0.0001,
+      "loss": 0.0069,
+      "num_input_tokens_seen": 291664280,
+      "step": 1978
+    },
+    {
+      "epoch": 0.5701931392332085,
+      "loss": 0.011196194216609001,
+      "loss_ce": 0.009730396792292595,
+      "loss_xval": 0.00146484375,
+      "num_input_tokens_seen": 291664280,
+      "step": 1978
+    },
+    {
+      "epoch": 0.5704814067454598,
+      "grad_norm": 1.4667761838514597,
+      "learning_rate": 0.0001,
+      "loss": 0.0014,
+      "num_input_tokens_seen": 291799408,
+      "step": 1979
+    },
+    {
+      "epoch": 0.5704814067454598,
+      "loss": 0.0012744119158014655,
+      "loss_ce": 3.8450009014923126e-05,
+      "loss_xval": 0.0012359619140625,
+      "num_input_tokens_seen": 291799408,
+      "step": 1979
+    },
+    {
+      "epoch": 0.5707696742577112,
+      "grad_norm": 0.5428364480716036,
+      "learning_rate": 0.0001,
+      "loss": 0.0058,
+      "num_input_tokens_seen": 291972168,
+      "step": 1980
+    },
+    {
+      "epoch": 0.5707696742577112,
+      "loss": 0.0022662151604890823,
+      "loss_ce": 0.0017080771503970027,
+      "loss_xval": 0.00055694580078125,
+      "num_input_tokens_seen": 291972168,
+      "step": 1980
+    },
+    {
+      "epoch": 0.5710579417699625,
+      "grad_norm": 1.0430200816446416,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "num_input_tokens_seen": 292106936,
+      "step": 1981
+    },
+    {
+      "epoch": 0.5710579417699625,
+      "loss": 0.02298683673143387,
+      "loss_ce": 0.02252192050218582,
+      "loss_xval": 0.00046539306640625,
+      "num_input_tokens_seen": 292106936,
+      "step": 1981
+    },
+    {
+      "epoch": 0.5713462092822139,
+      "grad_norm": 1.6942707055207802,
+      "learning_rate": 0.0001,
+      "loss": 0.0015,
+      "num_input_tokens_seen": 292242048,
+      "step": 1982
+    },
+    {
+      "epoch": 0.5713462092822139,
+      "loss": 0.0013274959055706859,
+      "loss_ce": 7.341417949646711e-05,
+      "loss_xval": 0.001251220703125,
+      "num_input_tokens_seen": 292242048,
+      "step": 1982
+    },
+    {
+      "epoch": 0.5716344767944652,
+      "grad_norm": 1.6282262101978988,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "num_input_tokens_seen": 292414584,
+      "step": 1983
+    },
+    {
+      "epoch": 0.5716344767944652,
+      "loss": 0.005388371646404266,
+      "loss_ce": 0.0033703967928886414,
+      "loss_xval": 0.00201416015625,
+      "num_input_tokens_seen": 292414584,
+      "step": 1983
+    },
+    {
+      "epoch": 0.5719227443067166,
+      "grad_norm": 0.6386831560604371,
+      "learning_rate": 0.0001,
+      "loss": 0.0049,
+      "num_input_tokens_seen": 292549432,
+      "step": 1984
+    },
+    {
+      "epoch": 0.5719227443067166,
+      "loss": 0.00935543142259121,
+      "loss_ce": 0.008496647700667381,
+      "loss_xval": 0.000858306884765625,
+      "num_input_tokens_seen": 292549432,
+      "step": 1984
+    },
+    {
+      "epoch": 0.572211011818968,
+      "grad_norm": 0.552995258421988,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "num_input_tokens_seen": 292684424,
+      "step": 1985
+    },
+    {
+      "epoch": 0.572211011818968,
+      "loss": 0.0003244928375352174,
+      "loss_ce": 9.286919521400705e-05,
+      "loss_xval": 0.00023174285888671875,
+      "num_input_tokens_seen": 292684424,
+      "step": 1985
+    },
+    {
+      "epoch": 0.5724992793312194,
+      "grad_norm": 0.3768425457267613,
+      "learning_rate": 0.0001,
+      "loss": 0.0042,
+      "num_input_tokens_seen": 292856968,
+      "step": 1986
+    },
+    {
+      "epoch": 0.5724992793312194,
+      "loss": 0.0005440630484372377,
+      "loss_ce": 9.464404865866527e-05,
+      "loss_xval": 0.00045013427734375,
+      "num_input_tokens_seen": 292856968,
+      "step": 1986
+    },
+    {
+      "epoch": 0.5727875468434708,
+      "grad_norm": 0.7890485953633781,
+      "learning_rate": 0.0001,
+      "loss": 0.0069,
+      "num_input_tokens_seen": 292991680,
+      "step": 1987
+    },
+    {
+      "epoch": 0.5727875468434708,
+      "loss": 0.012787751853466034,
+      "loss_ce": 0.012313298881053925,
+      "loss_xval": 0.0004749298095703125,
+      "num_input_tokens_seen": 292991680,
+      "step": 1987
+    },
+    {
+      "epoch": 0.5730758143557221,
+      "grad_norm": 1.4873338901823483,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "num_input_tokens_seen": 293126728,
+      "step": 1988
+    },
+    {
+      "epoch": 0.5730758143557221,
+      "loss": 0.000927687855437398,
+      "loss_ce": 3.838654811261222e-05,
+      "loss_xval": 0.000888824462890625,
+      "num_input_tokens_seen": 293126728,
+      "step": 1988
+    },
+    {
+      "epoch": 0.5733640818679735,
+      "grad_norm": 1.6546880055852196,
+      "learning_rate": 0.0001,
+      "loss": 0.005,
+      "num_input_tokens_seen": 293299200,
+      "step": 1989
+    },
+    {
+      "epoch": 0.5733640818679735,
+      "loss": 0.0014969916082918644,
+      "loss_ce": 5.4082367569208145e-05,
+      "loss_xval": 0.00144195556640625,
+      "num_input_tokens_seen": 293299200,
+      "step": 1989
+    },
+    {
+      "epoch": 0.5736523493802248,
+      "grad_norm": 1.2679658205061728,
+      "learning_rate": 0.0001,
+      "loss": 0.0062,
+      "num_input_tokens_seen": 293434000,
+      "step": 1990
+    },
+    {
+      "epoch": 0.5736523493802248,
+      "loss": 0.012005031108856201,
+      "loss_ce": 0.009815394878387451,
+      "loss_xval": 0.002197265625,
+      "num_input_tokens_seen": 293434000,
+      "step": 1990
+    },
+    {
+      "epoch": 0.5739406168924762,
+      "grad_norm": 1.1819654147990233,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "num_input_tokens_seen": 293569008,
+      "step": 1991
+    },
+    {
+      "epoch": 0.5739406168924762,
+      "loss": 0.0011650105006992817,
+      "loss_ce": 5.397993663791567e-05,
+      "loss_xval": 0.0011138916015625,
+      "num_input_tokens_seen": 293569008,
+      "step": 1991
+    },
+    {
+      "epoch": 0.5742288844047276,
+      "grad_norm": 1.6972687442414842,
+      "learning_rate": 0.0001,
+      "loss": 0.0065,
+      "num_input_tokens_seen": 293741424,
+      "step": 1992
+    },
+    {
+      "epoch": 0.5742288844047276,
+      "loss": 0.0014634766848757863,
+      "loss_ce": 5.966809840174392e-05,
+      "loss_xval": 0.00140380859375,
+      "num_input_tokens_seen": 293741424,
+      "step": 1992
+    },
+    {
+      "epoch": 0.5745171519169789,
+      "grad_norm": 2.852577005265175,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "num_input_tokens_seen": 293876288,
+      "step": 1993
+    },
+    {
+      "epoch": 0.5745171519169789,
+      "loss": 0.012815849855542183,
+      "loss_ce": 0.009848015382885933,
+      "loss_xval": 0.002960205078125,
+      "num_input_tokens_seen": 293876288,
+      "step": 1993
+    },
+    {
+      "epoch": 0.5748054194292304,
+      "grad_norm": 4.215657705576143,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "num_input_tokens_seen": 294011360,
+      "step": 1994
+    },
+    {
+      "epoch": 0.5748054194292304,
+      "loss": 0.007397228851914406,
+      "loss_ce": 4.24925638071727e-05,
+      "loss_xval": 0.007354736328125,
+      "num_input_tokens_seen": 294011360,
+      "step": 1994
+    },
+    {
+      "epoch": 0.5750936869414817,
+      "grad_norm": 5.016030693031347,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "num_input_tokens_seen": 294183848,
+      "step": 1995
+    },
+    {
+      "epoch": 0.5750936869414817,
+      "loss": 0.01234062947332859,
+      "loss_ce": 4.204545257380232e-05,
+      "loss_xval": 0.0123291015625,
+      "num_input_tokens_seen": 294183848,
+      "step": 1995
+    },
+    {
+      "epoch": 0.5753819544537331,
+      "grad_norm": 4.917456119960526,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "num_input_tokens_seen": 294318704,
+      "step": 1996
+    },
+    {
+      "epoch": 0.5753819544537331,
+      "loss": 0.020639140158891678,
+      "loss_ce": 0.007631022948771715,
+      "loss_xval": 0.01300048828125,
+      "num_input_tokens_seen": 294318704,
+      "step": 1996
+    },
+    {
+      "epoch": 0.5756702219659844,
+      "grad_norm": 4.2676973987215705,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "num_input_tokens_seen": 294453744,
+      "step": 1997
+    },
+    {
+      "epoch": 0.5756702219659844,
+      "loss": 0.009153537452220917,
+      "loss_ce": 0.00017374010349158198,
+      "loss_xval": 0.00897216796875,
+      "num_input_tokens_seen": 294453744,
+      "step": 1997
+    },
+    {
+      "epoch": 0.5759584894782358,
+      "grad_norm": 3.510931597546058,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "num_input_tokens_seen": 294626248,
+      "step": 1998
+    },
+    {
+      "epoch": 0.5759584894782358,
+      "loss": 0.0059237708337605,
+      "loss_ce": 0.00030472176149487495,
+      "loss_xval": 0.005615234375,
+      "num_input_tokens_seen": 294626248,
+      "step": 1998
+    },
+    {
+      "epoch": 0.5762467569904872,
+      "grad_norm": 2.5517557847208376,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "num_input_tokens_seen": 294761032,
+      "step": 1999
+    },
+    {
+      "epoch": 0.5762467569904872,
+      "loss": 0.015102425590157509,
+      "loss_ce": 0.012073555961251259,
+      "loss_xval": 0.003021240234375,
+      "num_input_tokens_seen": 294761032,
+      "step": 1999
+    },
+    {
+      "epoch": 0.5765350245027385,
+      "grad_norm": 1.4836132785368261,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "num_input_tokens_seen": 294896120,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5765350245027385,
+      "eval_websight_new_IoU": 0.4954599142074585,
+      "eval_websight_new_MAE_x": 0.01227654330432415,
+      "eval_websight_new_MAE_y": 0.012751261238008738,
+      "eval_websight_new_NUM_probability": 0.9992389678955078,
+      "eval_websight_new_inside_bbox": 0.7829861044883728,
+      "eval_websight_new_loss": 0.0003192399744875729,
+      "eval_websight_new_loss_ce": 0.00010440748155815527,
+      "eval_websight_new_loss_xval": 0.00020268559455871582,
+      "eval_websight_new_runtime": 35.7316,
+      "eval_websight_new_samples_per_second": 1.399,
+      "eval_websight_new_steps_per_second": 0.056,
+      "num_input_tokens_seen": 294896120,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5765350245027385,
+      "eval_seeclick_IoU": 0.5345141887664795,
+      "eval_seeclick_MAE_x": 0.015516493003815413,
+      "eval_seeclick_MAE_y": 0.013822767417877913,
+      "eval_seeclick_NUM_probability": 0.9992666840553284,
+      "eval_seeclick_inside_bbox": 0.7829861044883728,
+      "eval_seeclick_loss": 0.007619759999215603,
+      "eval_seeclick_loss_ce": 0.008350821677595377,
+      "eval_seeclick_loss_xval": 0.0003256499767303467,
+      "eval_seeclick_runtime": 64.0935,
+      "eval_seeclick_samples_per_second": 0.78,
+      "eval_seeclick_steps_per_second": 0.031,
+      "num_input_tokens_seen": 294896120,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5765350245027385,
+      "eval_icons_IoU": 0.18187464028596878,
+      "eval_icons_MAE_x": 0.0158988106995821,
+      "eval_icons_MAE_y": 0.017659504897892475,
+      "eval_icons_NUM_probability": 0.9992129504680634,
+      "eval_icons_inside_bbox": 0.2361111119389534,
+      "eval_icons_loss": 0.007123068440705538,
+      "eval_icons_loss_ce": 0.005604170728474855,
+      "eval_icons_loss_xval": 0.00037682056427001953,
+      "eval_icons_runtime": 69.6929,
+      "eval_icons_samples_per_second": 0.717,
+      "eval_icons_steps_per_second": 0.029,
+      "num_input_tokens_seen": 294896120,
+      "step": 2000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 294896120,
+  "num_train_epochs": 3,
+  "save_steps": 250,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2411691198119936.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}