diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,17193 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9851665267282396,
+  "eval_steps": 10,
+  "global_step": 2200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00044780296669465436,
+      "grad_norm": 2.2432782649993896,
+      "learning_rate": 3.7313432835820895e-07,
+      "loss": 1.8132,
+      "step": 1
+    },
+    {
+      "epoch": 0.0008956059333893087,
+      "grad_norm": 2.291522264480591,
+      "learning_rate": 7.462686567164179e-07,
+      "loss": 1.8956,
+      "step": 2
+    },
+    {
+      "epoch": 0.001343408900083963,
+      "grad_norm": 1.6618109941482544,
+      "learning_rate": 1.119402985074627e-06,
+      "loss": 1.8312,
+      "step": 3
+    },
+    {
+      "epoch": 0.0017912118667786174,
+      "grad_norm": 2.319044589996338,
+      "learning_rate": 1.4925373134328358e-06,
+      "loss": 1.8718,
+      "step": 4
+    },
+    {
+      "epoch": 0.002239014833473272,
+      "grad_norm": 1.671183705329895,
+      "learning_rate": 1.8656716417910446e-06,
+      "loss": 1.8156,
+      "step": 5
+    },
+    {
+      "epoch": 0.002686817800167926,
+      "grad_norm": 1.857389211654663,
+      "learning_rate": 2.238805970149254e-06,
+      "loss": 1.8177,
+      "step": 6
+    },
+    {
+      "epoch": 0.0031346207668625807,
+      "grad_norm": 1.590662956237793,
+      "learning_rate": 2.6119402985074627e-06,
+      "loss": 1.8255,
+      "step": 7
+    },
+    {
+      "epoch": 0.003582423733557235,
+      "grad_norm": 1.9237611293792725,
+      "learning_rate": 2.9850746268656716e-06,
+      "loss": 1.8032,
+      "step": 8
+    },
+    {
+      "epoch": 0.004030226700251889,
+      "grad_norm": 1.6888021230697632,
+      "learning_rate": 3.358208955223881e-06,
+      "loss": 1.7965,
+      "step": 9
+    },
+    {
+      "epoch": 0.004478029666946544,
+      "grad_norm": 2.0715696811676025,
+      "learning_rate": 3.7313432835820893e-06,
+      "loss": 1.9659,
+      "step": 10
+    },
+    {
+      "epoch": 0.004478029666946544,
+      "eval_loss": 1.819678783416748,
+      "eval_runtime": 1739.3924,
+      "eval_samples_per_second": 2.568,
+      "eval_steps_per_second": 2.568,
+      "step": 10
+    },
+    {
+      "epoch": 0.0049258326336411975,
+      "grad_norm": 1.545844316482544,
+      "learning_rate": 4.1044776119402985e-06,
+      "loss": 1.8914,
+      "step": 11
+    },
+    {
+      "epoch": 0.005373635600335852,
+      "grad_norm": 1.37221360206604,
+      "learning_rate": 4.477611940298508e-06,
+      "loss": 1.7632,
+      "step": 12
+    },
+    {
+      "epoch": 0.005821438567030507,
+      "grad_norm": 1.4589499235153198,
+      "learning_rate": 4.850746268656717e-06,
+      "loss": 1.7984,
+      "step": 13
+    },
+    {
+      "epoch": 0.006269241533725161,
+      "grad_norm": 1.5319428443908691,
+      "learning_rate": 5.2238805970149255e-06,
+      "loss": 1.9122,
+      "step": 14
+    },
+    {
+      "epoch": 0.006717044500419815,
+      "grad_norm": 1.2317570447921753,
+      "learning_rate": 5.597014925373135e-06,
+      "loss": 1.7678,
+      "step": 15
+    },
+    {
+      "epoch": 0.00716484746711447,
+      "grad_norm": 1.4415943622589111,
+      "learning_rate": 5.970149253731343e-06,
+      "loss": 1.9129,
+      "step": 16
+    },
+    {
+      "epoch": 0.007612650433809124,
+      "grad_norm": 1.0773462057113647,
+      "learning_rate": 6.343283582089552e-06,
+      "loss": 1.4962,
+      "step": 17
+    },
+    {
+      "epoch": 0.008060453400503778,
+      "grad_norm": 1.1921244859695435,
+      "learning_rate": 6.716417910447762e-06,
+      "loss": 1.7538,
+      "step": 18
+    },
+    {
+      "epoch": 0.008508256367198432,
+      "grad_norm": 1.0547364950180054,
+      "learning_rate": 7.08955223880597e-06,
+      "loss": 1.6643,
+      "step": 19
+    },
+    {
+      "epoch": 0.008956059333893087,
+      "grad_norm": 1.065486192703247,
+      "learning_rate": 7.4626865671641785e-06,
+      "loss": 1.5438,
+      "step": 20
+    },
+    {
+      "epoch": 0.008956059333893087,
+      "eval_loss": 1.666861891746521,
+      "eval_runtime": 1742.6586,
+      "eval_samples_per_second": 2.563,
+      "eval_steps_per_second": 2.563,
+      "step": 20
+    },
+    {
+      "epoch": 0.009403862300587741,
+      "grad_norm": 0.9827540516853333,
+      "learning_rate": 7.835820895522389e-06,
+      "loss": 1.5295,
+      "step": 21
+    },
+    {
+      "epoch": 0.009851665267282395,
+      "grad_norm": 0.9350996613502502,
+      "learning_rate": 8.208955223880597e-06,
+      "loss": 1.6664,
+      "step": 22
+    },
+    {
+      "epoch": 0.01029946823397705,
+      "grad_norm": 1.0762792825698853,
+      "learning_rate": 8.582089552238805e-06,
+      "loss": 1.6782,
+      "step": 23
+    },
+    {
+      "epoch": 0.010747271200671704,
+      "grad_norm": 0.7595136165618896,
+      "learning_rate": 8.955223880597016e-06,
+      "loss": 1.6303,
+      "step": 24
+    },
+    {
+      "epoch": 0.01119507416736636,
+      "grad_norm": 0.6962496638298035,
+      "learning_rate": 9.328358208955226e-06,
+      "loss": 1.5204,
+      "step": 25
+    },
+    {
+      "epoch": 0.011642877134061013,
+      "grad_norm": 0.6340360641479492,
+      "learning_rate": 9.701492537313434e-06,
+      "loss": 1.4292,
+      "step": 26
+    },
+    {
+      "epoch": 0.012090680100755667,
+      "grad_norm": 0.6120555996894836,
+      "learning_rate": 1.0074626865671643e-05,
+      "loss": 1.4588,
+      "step": 27
+    },
+    {
+      "epoch": 0.012538483067450323,
+      "grad_norm": 0.6970665454864502,
+      "learning_rate": 1.0447761194029851e-05,
+      "loss": 1.6072,
+      "step": 28
+    },
+    {
+      "epoch": 0.012986286034144976,
+      "grad_norm": 0.7495852112770081,
+      "learning_rate": 1.082089552238806e-05,
+      "loss": 1.4976,
+      "step": 29
+    },
+    {
+      "epoch": 0.01343408900083963,
+      "grad_norm": 0.5586577653884888,
+      "learning_rate": 1.119402985074627e-05,
+      "loss": 1.5005,
+      "step": 30
+    },
+    {
+      "epoch": 0.01343408900083963,
+      "eval_loss": 1.4981863498687744,
+      "eval_runtime": 1737.6969,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 30
+    },
+    {
+      "epoch": 0.013881891967534286,
+      "grad_norm": 0.5340927839279175,
+      "learning_rate": 1.1567164179104478e-05,
+      "loss": 1.5658,
+      "step": 31
+    },
+    {
+      "epoch": 0.01432969493422894,
+      "grad_norm": 0.5143851637840271,
+      "learning_rate": 1.1940298507462686e-05,
+      "loss": 1.5624,
+      "step": 32
+    },
+    {
+      "epoch": 0.014777497900923593,
+      "grad_norm": 0.5012256503105164,
+      "learning_rate": 1.2313432835820896e-05,
+      "loss": 1.2532,
+      "step": 33
+    },
+    {
+      "epoch": 0.015225300867618249,
+      "grad_norm": 0.5775434970855713,
+      "learning_rate": 1.2686567164179105e-05,
+      "loss": 1.372,
+      "step": 34
+    },
+    {
+      "epoch": 0.0156731038343129,
+      "grad_norm": 0.611817479133606,
+      "learning_rate": 1.3059701492537313e-05,
+      "loss": 1.4631,
+      "step": 35
+    },
+    {
+      "epoch": 0.016120906801007556,
+      "grad_norm": 0.629241406917572,
+      "learning_rate": 1.3432835820895523e-05,
+      "loss": 1.4753,
+      "step": 36
+    },
+    {
+      "epoch": 0.016568709767702212,
+      "grad_norm": 0.5267626643180847,
+      "learning_rate": 1.3805970149253733e-05,
+      "loss": 1.371,
+      "step": 37
+    },
+    {
+      "epoch": 0.017016512734396864,
+      "grad_norm": 0.6829407811164856,
+      "learning_rate": 1.417910447761194e-05,
+      "loss": 1.4017,
+      "step": 38
+    },
+    {
+      "epoch": 0.01746431570109152,
+      "grad_norm": 0.6046289205551147,
+      "learning_rate": 1.455223880597015e-05,
+      "loss": 1.3686,
+      "step": 39
+    },
+    {
+      "epoch": 0.017912118667786175,
+      "grad_norm": 0.610405683517456,
+      "learning_rate": 1.4925373134328357e-05,
+      "loss": 1.269,
+      "step": 40
+    },
+    {
+      "epoch": 0.017912118667786175,
+      "eval_loss": 1.3559142351150513,
+      "eval_runtime": 1741.7432,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 2.565,
+      "step": 40
+    },
+    {
+      "epoch": 0.018359921634480827,
+      "grad_norm": 0.5271261930465698,
+      "learning_rate": 1.529850746268657e-05,
+      "loss": 1.3167,
+      "step": 41
+    },
+    {
+      "epoch": 0.018807724601175482,
+      "grad_norm": 0.549258291721344,
+      "learning_rate": 1.5671641791044777e-05,
+      "loss": 1.4023,
+      "step": 42
+    },
+    {
+      "epoch": 0.019255527567870138,
+      "grad_norm": 0.5591861605644226,
+      "learning_rate": 1.6044776119402986e-05,
+      "loss": 1.2516,
+      "step": 43
+    },
+    {
+      "epoch": 0.01970333053456479,
+      "grad_norm": 0.4712078273296356,
+      "learning_rate": 1.6417910447761194e-05,
+      "loss": 1.4436,
+      "step": 44
+    },
+    {
+      "epoch": 0.020151133501259445,
+      "grad_norm": 0.4768025279045105,
+      "learning_rate": 1.6791044776119406e-05,
+      "loss": 1.3387,
+      "step": 45
+    },
+    {
+      "epoch": 0.0205989364679541,
+      "grad_norm": 0.5674676299095154,
+      "learning_rate": 1.716417910447761e-05,
+      "loss": 1.1163,
+      "step": 46
+    },
+    {
+      "epoch": 0.021046739434648753,
+      "grad_norm": 0.5266850590705872,
+      "learning_rate": 1.7537313432835823e-05,
+      "loss": 1.4061,
+      "step": 47
+    },
+    {
+      "epoch": 0.02149454240134341,
+      "grad_norm": 0.5684650540351868,
+      "learning_rate": 1.791044776119403e-05,
+      "loss": 1.3917,
+      "step": 48
+    },
+    {
+      "epoch": 0.021942345368038064,
+      "grad_norm": 0.4995711147785187,
+      "learning_rate": 1.828358208955224e-05,
+      "loss": 1.2096,
+      "step": 49
+    },
+    {
+      "epoch": 0.02239014833473272,
+      "grad_norm": 0.4433654546737671,
+      "learning_rate": 1.865671641791045e-05,
+      "loss": 1.1679,
+      "step": 50
+    },
+    {
+      "epoch": 0.02239014833473272,
+      "eval_loss": 1.2627769708633423,
+      "eval_runtime": 1741.8103,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 2.565,
+      "step": 50
+    },
+    {
+      "epoch": 0.02283795130142737,
+      "grad_norm": 0.49915778636932373,
+      "learning_rate": 1.9029850746268656e-05,
+      "loss": 1.1626,
+      "step": 51
+    },
+    {
+      "epoch": 0.023285754268122027,
+      "grad_norm": 0.5690376162528992,
+      "learning_rate": 1.9402985074626868e-05,
+      "loss": 1.2634,
+      "step": 52
+    },
+    {
+      "epoch": 0.023733557234816682,
+      "grad_norm": 0.46489912271499634,
+      "learning_rate": 1.9776119402985073e-05,
+      "loss": 1.3055,
+      "step": 53
+    },
+    {
+      "epoch": 0.024181360201511334,
+      "grad_norm": 0.46374455094337463,
+      "learning_rate": 2.0149253731343285e-05,
+      "loss": 1.1938,
+      "step": 54
+    },
+    {
+      "epoch": 0.02462916316820599,
+      "grad_norm": 0.43488311767578125,
+      "learning_rate": 2.0522388059701493e-05,
+      "loss": 1.2376,
+      "step": 55
+    },
+    {
+      "epoch": 0.025076966134900645,
+      "grad_norm": 0.5964159965515137,
+      "learning_rate": 2.0895522388059702e-05,
+      "loss": 1.2211,
+      "step": 56
+    },
+    {
+      "epoch": 0.025524769101595297,
+      "grad_norm": 0.3947959244251251,
+      "learning_rate": 2.126865671641791e-05,
+      "loss": 1.1828,
+      "step": 57
+    },
+    {
+      "epoch": 0.025972572068289953,
+      "grad_norm": 0.42681577801704407,
+      "learning_rate": 2.164179104477612e-05,
+      "loss": 1.3245,
+      "step": 58
+    },
+    {
+      "epoch": 0.02642037503498461,
+      "grad_norm": 0.4505462944507599,
+      "learning_rate": 2.201492537313433e-05,
+      "loss": 1.2595,
+      "step": 59
+    },
+    {
+      "epoch": 0.02686817800167926,
+      "grad_norm": 0.44249483942985535,
+      "learning_rate": 2.238805970149254e-05,
+      "loss": 1.3007,
+      "step": 60
+    },
+    {
+      "epoch": 0.02686817800167926,
+      "eval_loss": 1.2291107177734375,
+      "eval_runtime": 1742.3595,
+      "eval_samples_per_second": 2.564,
+      "eval_steps_per_second": 2.564,
+      "step": 60
+    },
+    {
+      "epoch": 0.027315980968373916,
+      "grad_norm": 0.4635663330554962,
+      "learning_rate": 2.2761194029850747e-05,
+      "loss": 1.2106,
+      "step": 61
+    },
+    {
+      "epoch": 0.02776378393506857,
+      "grad_norm": 0.4279429614543915,
+      "learning_rate": 2.3134328358208956e-05,
+      "loss": 1.3232,
+      "step": 62
+    },
+    {
+      "epoch": 0.028211586901763223,
+      "grad_norm": 0.47193002700805664,
+      "learning_rate": 2.3507462686567168e-05,
+      "loss": 1.1924,
+      "step": 63
+    },
+    {
+      "epoch": 0.02865938986845788,
+      "grad_norm": 0.46152573823928833,
+      "learning_rate": 2.3880597014925373e-05,
+      "loss": 1.1613,
+      "step": 64
+    },
+    {
+      "epoch": 0.029107192835152534,
+      "grad_norm": 0.41957393288612366,
+      "learning_rate": 2.4253731343283584e-05,
+      "loss": 1.2249,
+      "step": 65
+    },
+    {
+      "epoch": 0.029554995801847186,
+      "grad_norm": 0.5162342190742493,
+      "learning_rate": 2.4626865671641793e-05,
+      "loss": 1.2557,
+      "step": 66
+    },
+    {
+      "epoch": 0.030002798768541842,
+      "grad_norm": 0.4564678370952606,
+      "learning_rate": 2.5e-05,
+      "loss": 1.0555,
+      "step": 67
+    },
+    {
+      "epoch": 0.030450601735236497,
+      "grad_norm": 0.441526859998703,
+      "learning_rate": 2.537313432835821e-05,
+      "loss": 1.1295,
+      "step": 68
+    },
+    {
+      "epoch": 0.03089840470193115,
+      "grad_norm": 0.38912713527679443,
+      "learning_rate": 2.574626865671642e-05,
+      "loss": 1.0991,
+      "step": 69
+    },
+    {
+      "epoch": 0.0313462076686258,
+      "grad_norm": 0.411045104265213,
+      "learning_rate": 2.6119402985074626e-05,
+      "loss": 1.2926,
+      "step": 70
+    },
+    {
+      "epoch": 0.0313462076686258,
+      "eval_loss": 1.207950234413147,
+      "eval_runtime": 1742.1349,
+      "eval_samples_per_second": 2.564,
+      "eval_steps_per_second": 2.564,
+      "step": 70
+    },
+    {
+      "epoch": 0.03179401063532046,
+      "grad_norm": 0.6341650485992432,
+      "learning_rate": 2.6492537313432835e-05,
+      "loss": 1.2046,
+      "step": 71
+    },
+    {
+      "epoch": 0.03224181360201511,
+      "grad_norm": 0.4355488717556,
+      "learning_rate": 2.6865671641791047e-05,
+      "loss": 1.2758,
+      "step": 72
+    },
+    {
+      "epoch": 0.032689616568709765,
+      "grad_norm": 0.44128844141960144,
+      "learning_rate": 2.7238805970149255e-05,
+      "loss": 1.35,
+      "step": 73
+    },
+    {
+      "epoch": 0.033137419535404423,
+      "grad_norm": 0.43141600489616394,
+      "learning_rate": 2.7611940298507467e-05,
+      "loss": 1.2572,
+      "step": 74
+    },
+    {
+      "epoch": 0.033585222502099076,
+      "grad_norm": 0.46453097462654114,
+      "learning_rate": 2.7985074626865672e-05,
+      "loss": 1.1533,
+      "step": 75
+    },
+    {
+      "epoch": 0.03403302546879373,
+      "grad_norm": 0.39782240986824036,
+      "learning_rate": 2.835820895522388e-05,
+      "loss": 1.1852,
+      "step": 76
+    },
+    {
+      "epoch": 0.034480828435488386,
+      "grad_norm": 0.45064836740493774,
+      "learning_rate": 2.8731343283582092e-05,
+      "loss": 1.079,
+      "step": 77
+    },
+    {
+      "epoch": 0.03492863140218304,
+      "grad_norm": 0.4778350591659546,
+      "learning_rate": 2.91044776119403e-05,
+      "loss": 1.3059,
+      "step": 78
+    },
+    {
+      "epoch": 0.03537643436887769,
+      "grad_norm": 0.3985428512096405,
+      "learning_rate": 2.9477611940298512e-05,
+      "loss": 1.2544,
+      "step": 79
+    },
+    {
+      "epoch": 0.03582423733557235,
+      "grad_norm": 0.43038174510002136,
+      "learning_rate": 2.9850746268656714e-05,
+      "loss": 1.3496,
+      "step": 80
+    },
+    {
+      "epoch": 0.03582423733557235,
+      "eval_loss": 1.190480351448059,
+      "eval_runtime": 1741.9424,
+      "eval_samples_per_second": 2.564,
+      "eval_steps_per_second": 2.564,
+      "step": 80
+    },
+    {
+      "epoch": 0.036272040302267,
+      "grad_norm": 0.46937325596809387,
+      "learning_rate": 3.0223880597014926e-05,
+      "loss": 1.1186,
+      "step": 81
+    },
+    {
+      "epoch": 0.036719843268961654,
+      "grad_norm": 0.45396608114242554,
+      "learning_rate": 3.059701492537314e-05,
+      "loss": 1.1812,
+      "step": 82
+    },
+    {
+      "epoch": 0.03716764623565631,
+      "grad_norm": 0.4596857726573944,
+      "learning_rate": 3.0970149253731346e-05,
+      "loss": 1.1643,
+      "step": 83
+    },
+    {
+      "epoch": 0.037615449202350965,
+      "grad_norm": 0.4687398076057434,
+      "learning_rate": 3.1343283582089554e-05,
+      "loss": 1.1045,
+      "step": 84
+    },
+    {
+      "epoch": 0.03806325216904562,
+      "grad_norm": 0.40774309635162354,
+      "learning_rate": 3.171641791044776e-05,
+      "loss": 1.1006,
+      "step": 85
+    },
+    {
+      "epoch": 0.038511055135740276,
+      "grad_norm": 0.41461247205734253,
+      "learning_rate": 3.208955223880597e-05,
+      "loss": 1.1628,
+      "step": 86
+    },
+    {
+      "epoch": 0.03895885810243493,
+      "grad_norm": 0.5309916138648987,
+      "learning_rate": 3.246268656716418e-05,
+      "loss": 1.1437,
+      "step": 87
+    },
+    {
+      "epoch": 0.03940666106912958,
+      "grad_norm": 0.4281834661960602,
+      "learning_rate": 3.283582089552239e-05,
+      "loss": 1.0383,
+      "step": 88
+    },
+    {
+      "epoch": 0.03985446403582424,
+      "grad_norm": 0.4634588956832886,
+      "learning_rate": 3.32089552238806e-05,
+      "loss": 1.1905,
+      "step": 89
+    },
+    {
+      "epoch": 0.04030226700251889,
+      "grad_norm": 0.48842886090278625,
+      "learning_rate": 3.358208955223881e-05,
+      "loss": 1.1012,
+      "step": 90
+    },
+    {
+      "epoch": 0.04030226700251889,
+      "eval_loss": 1.1734172105789185,
+      "eval_runtime": 1742.402,
+      "eval_samples_per_second": 2.564,
+      "eval_steps_per_second": 2.564,
+      "step": 90
+    },
+    {
+      "epoch": 0.04075006996921354,
+      "grad_norm": 0.4782605767250061,
+      "learning_rate": 3.395522388059701e-05,
+      "loss": 1.2087,
+      "step": 91
+    },
+    {
+      "epoch": 0.0411978729359082,
+      "grad_norm": 0.6223856210708618,
+      "learning_rate": 3.432835820895522e-05,
+      "loss": 1.1121,
+      "step": 92
+    },
+    {
+      "epoch": 0.041645675902602854,
+      "grad_norm": 0.4409349858760834,
+      "learning_rate": 3.470149253731344e-05,
+      "loss": 1.1133,
+      "step": 93
+    },
+    {
+      "epoch": 0.042093478869297506,
+      "grad_norm": 0.5995984077453613,
+      "learning_rate": 3.5074626865671645e-05,
+      "loss": 1.1062,
+      "step": 94
+    },
+    {
+      "epoch": 0.042541281835992165,
+      "grad_norm": 0.5431959629058838,
+      "learning_rate": 3.5447761194029854e-05,
+      "loss": 1.0471,
+      "step": 95
+    },
+    {
+      "epoch": 0.04298908480268682,
+      "grad_norm": 0.4902937114238739,
+      "learning_rate": 3.582089552238806e-05,
+      "loss": 1.2069,
+      "step": 96
+    },
+    {
+      "epoch": 0.043436887769381476,
+      "grad_norm": 0.4660329818725586,
+      "learning_rate": 3.619402985074627e-05,
+      "loss": 1.1259,
+      "step": 97
+    },
+    {
+      "epoch": 0.04388469073607613,
+      "grad_norm": 0.5536847114562988,
+      "learning_rate": 3.656716417910448e-05,
+      "loss": 1.2128,
+      "step": 98
+    },
+    {
+      "epoch": 0.04433249370277078,
+      "grad_norm": 0.6095202565193176,
+      "learning_rate": 3.694029850746269e-05,
+      "loss": 1.0796,
+      "step": 99
+    },
+    {
+      "epoch": 0.04478029666946544,
+      "grad_norm": 0.5868849754333496,
+      "learning_rate": 3.73134328358209e-05,
+      "loss": 1.2881,
+      "step": 100
+    },
+    {
+      "epoch": 0.04478029666946544,
+      "eval_loss": 1.152649998664856,
+      "eval_runtime": 1741.8326,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 2.565,
+      "step": 100
+    },
+    {
+      "epoch": 0.04522809963616009,
+      "grad_norm": 0.5384364724159241,
+      "learning_rate": 3.7686567164179104e-05,
+      "loss": 1.1073,
+      "step": 101
+    },
+    {
+      "epoch": 0.04567590260285474,
+      "grad_norm": 0.6125591397285461,
+      "learning_rate": 3.805970149253731e-05,
+      "loss": 1.0919,
+      "step": 102
+    },
+    {
+      "epoch": 0.0461237055695494,
+      "grad_norm": 0.4984678030014038,
+      "learning_rate": 3.843283582089552e-05,
+      "loss": 1.2438,
+      "step": 103
+    },
+    {
+      "epoch": 0.046571508536244054,
+      "grad_norm": 0.4974757432937622,
+      "learning_rate": 3.8805970149253736e-05,
+      "loss": 1.1709,
+      "step": 104
+    },
+    {
+      "epoch": 0.047019311502938706,
+      "grad_norm": 0.4727949798107147,
+      "learning_rate": 3.9179104477611945e-05,
+      "loss": 1.2356,
+      "step": 105
+    },
+    {
+      "epoch": 0.047467114469633365,
+      "grad_norm": 0.5323635935783386,
+      "learning_rate": 3.9552238805970146e-05,
+      "loss": 1.1092,
+      "step": 106
+    },
+    {
+      "epoch": 0.04791491743632802,
+      "grad_norm": 0.4718462824821472,
+      "learning_rate": 3.992537313432836e-05,
+      "loss": 1.1643,
+      "step": 107
+    },
+    {
+      "epoch": 0.04836272040302267,
+      "grad_norm": 0.46946635842323303,
+      "learning_rate": 4.029850746268657e-05,
+      "loss": 1.1401,
+      "step": 108
+    },
+    {
+      "epoch": 0.04881052336971733,
+      "grad_norm": 0.5624064207077026,
+      "learning_rate": 4.067164179104478e-05,
+      "loss": 1.0968,
+      "step": 109
+    },
+    {
+      "epoch": 0.04925832633641198,
+      "grad_norm": 0.5305824875831604,
+      "learning_rate": 4.104477611940299e-05,
+      "loss": 1.1556,
+      "step": 110
+    },
+    {
+      "epoch": 0.04925832633641198,
+      "eval_loss": 1.1313787698745728,
+      "eval_runtime": 1741.4825,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 2.565,
+      "step": 110
+    },
+    {
+      "epoch": 0.04970612930310663,
+      "grad_norm": 0.47058171033859253,
+      "learning_rate": 4.1417910447761195e-05,
+      "loss": 1.1151,
+      "step": 111
+    },
+    {
+      "epoch": 0.05015393226980129,
+      "grad_norm": 0.6491269469261169,
+      "learning_rate": 4.1791044776119404e-05,
+      "loss": 1.0862,
+      "step": 112
+    },
+    {
+      "epoch": 0.05060173523649594,
+      "grad_norm": 0.5460953712463379,
+      "learning_rate": 4.216417910447761e-05,
+      "loss": 1.0627,
+      "step": 113
+    },
+    {
+      "epoch": 0.051049538203190595,
+      "grad_norm": 0.5378393530845642,
+      "learning_rate": 4.253731343283582e-05,
+      "loss": 1.2173,
+      "step": 114
+    },
+    {
+      "epoch": 0.051497341169885254,
+      "grad_norm": 0.6210533976554871,
+      "learning_rate": 4.2910447761194036e-05,
+      "loss": 1.0156,
+      "step": 115
+    },
+    {
+      "epoch": 0.051945144136579906,
+      "grad_norm": 0.4818199574947357,
+      "learning_rate": 4.328358208955224e-05,
+      "loss": 1.1848,
+      "step": 116
+    },
+    {
+      "epoch": 0.05239294710327456,
+      "grad_norm": 0.6459422707557678,
+      "learning_rate": 4.3656716417910446e-05,
+      "loss": 1.0064,
+      "step": 117
+    },
+    {
+      "epoch": 0.05284075006996922,
+      "grad_norm": 0.5303052067756653,
+      "learning_rate": 4.402985074626866e-05,
+      "loss": 1.1831,
+      "step": 118
+    },
+    {
+      "epoch": 0.05328855303666387,
+      "grad_norm": 0.6115245819091797,
+      "learning_rate": 4.440298507462687e-05,
+      "loss": 1.2621,
+      "step": 119
+    },
+    {
+      "epoch": 0.05373635600335852,
+      "grad_norm": 0.5204371213912964,
+      "learning_rate": 4.477611940298508e-05,
+      "loss": 1.1908,
+      "step": 120
+    },
+    {
+      "epoch": 0.05373635600335852,
+      "eval_loss": 1.114864468574524,
+      "eval_runtime": 1741.2804,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 2.565,
+      "step": 120
+    },
+    {
+      "epoch": 0.05418415897005318,
+      "grad_norm": 0.5293467044830322,
+      "learning_rate": 4.5149253731343286e-05,
+      "loss": 1.2622,
+      "step": 121
+    },
+    {
+      "epoch": 0.05463196193674783,
+      "grad_norm": 0.5757855176925659,
+      "learning_rate": 4.5522388059701495e-05,
+      "loss": 0.9621,
+      "step": 122
+    },
+    {
+      "epoch": 0.055079764903442484,
+      "grad_norm": 0.558157205581665,
+      "learning_rate": 4.58955223880597e-05,
+      "loss": 1.0987,
+      "step": 123
+    },
+    {
+      "epoch": 0.05552756787013714,
+      "grad_norm": 0.6928860545158386,
+      "learning_rate": 4.626865671641791e-05,
+      "loss": 1.0925,
+      "step": 124
+    },
+    {
+      "epoch": 0.055975370836831795,
+      "grad_norm": 0.6394439935684204,
+      "learning_rate": 4.664179104477612e-05,
+      "loss": 1.302,
+      "step": 125
+    },
+    {
+      "epoch": 0.05642317380352645,
+      "grad_norm": 0.6799196004867554,
+      "learning_rate": 4.7014925373134335e-05,
+      "loss": 0.9764,
+      "step": 126
+    },
+    {
+      "epoch": 0.056870976770221106,
+      "grad_norm": 0.5544326901435852,
+      "learning_rate": 4.738805970149254e-05,
+      "loss": 1.0112,
+      "step": 127
+    },
+    {
+      "epoch": 0.05731877973691576,
+      "grad_norm": 0.5835520625114441,
+      "learning_rate": 4.7761194029850745e-05,
+      "loss": 1.0672,
+      "step": 128
+    },
+    {
+      "epoch": 0.05776658270361041,
+      "grad_norm": 0.5551716089248657,
+      "learning_rate": 4.813432835820896e-05,
+      "loss": 1.3243,
+      "step": 129
+    },
+    {
+      "epoch": 0.05821438567030507,
+      "grad_norm": 0.5320628881454468,
+      "learning_rate": 4.850746268656717e-05,
+      "loss": 1.0686,
+      "step": 130
+    },
+    {
+      "epoch": 0.05821438567030507,
+      "eval_loss": 1.0987714529037476,
+      "eval_runtime": 1741.4607,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 2.565,
+      "step": 130
+    },
+    {
+      "epoch": 0.05866218863699972,
+      "grad_norm": 0.5460705161094666,
+      "learning_rate": 4.888059701492538e-05,
+      "loss": 1.3141,
+      "step": 131
+    },
+    {
+      "epoch": 0.05910999160369437,
+      "grad_norm": 0.5375784039497375,
+      "learning_rate": 4.9253731343283586e-05,
+      "loss": 0.7433,
+      "step": 132
+    },
+    {
+      "epoch": 0.05955779457038903,
+      "grad_norm": 0.6023723483085632,
+      "learning_rate": 4.9626865671641794e-05,
+      "loss": 1.0964,
+      "step": 133
+    },
+    {
+      "epoch": 0.060005597537083684,
+      "grad_norm": 0.7922960519790649,
+      "learning_rate": 5e-05,
+      "loss": 1.0928,
+      "step": 134
+    },
+    {
+      "epoch": 0.060453400503778336,
+      "grad_norm": 0.6330304145812988,
+      "learning_rate": 4.9988457987072945e-05,
+      "loss": 0.9644,
+      "step": 135
+    },
+    {
+      "epoch": 0.060901203470472995,
+      "grad_norm": 0.5186926126480103,
+      "learning_rate": 4.9976915974145894e-05,
+      "loss": 1.3319,
+      "step": 136
+    },
+    {
+      "epoch": 0.06134900643716765,
+      "grad_norm": 0.9184876084327698,
+      "learning_rate": 4.996537396121884e-05,
+      "loss": 0.8197,
+      "step": 137
+    },
+    {
+      "epoch": 0.0617968094038623,
+      "grad_norm": 0.7908456325531006,
+      "learning_rate": 4.9953831948291786e-05,
+      "loss": 1.0536,
+      "step": 138
+    },
+    {
+      "epoch": 0.06224461237055696,
+      "grad_norm": 0.6290323138237,
+      "learning_rate": 4.994228993536473e-05,
+      "loss": 1.0967,
+      "step": 139
+    },
+    {
+      "epoch": 0.0626924153372516,
+      "grad_norm": 0.7094379663467407,
+      "learning_rate": 4.993074792243768e-05,
+      "loss": 1.1115,
+      "step": 140
+    },
+    {
+      "epoch": 0.0626924153372516,
+      "eval_loss": 1.0867949724197388,
+      "eval_runtime": 1740.7431,
+      "eval_samples_per_second": 2.566,
+      "eval_steps_per_second": 2.566,
+      "step": 140
+    },
+    {
+      "epoch": 0.06314021830394627,
+      "grad_norm": 0.6611003279685974,
+      "learning_rate": 4.991920590951062e-05,
+      "loss": 1.0284,
+      "step": 141
+    },
+    {
+      "epoch": 0.06358802127064092,
+      "grad_norm": 0.670655369758606,
+      "learning_rate": 4.990766389658357e-05,
+      "loss": 1.0436,
+      "step": 142
+    },
+    {
+      "epoch": 0.06403582423733557,
+      "grad_norm": 0.700566828250885,
+      "learning_rate": 4.989612188365651e-05,
+      "loss": 0.9554,
+      "step": 143
+    },
+    {
+      "epoch": 0.06448362720403022,
+      "grad_norm": 0.6415517330169678,
+      "learning_rate": 4.9884579870729456e-05,
+      "loss": 1.0908,
+      "step": 144
+    },
+    {
+      "epoch": 0.06493143017072488,
+      "grad_norm": 0.6611580848693848,
+      "learning_rate": 4.98730378578024e-05,
+      "loss": 1.0711,
+      "step": 145
+    },
+    {
+      "epoch": 0.06537923313741953,
+      "grad_norm": 0.6099154949188232,
+      "learning_rate": 4.986149584487535e-05,
+      "loss": 1.0865,
+      "step": 146
+    },
+    {
+      "epoch": 0.0658270361041142,
+      "grad_norm": 0.6479048132896423,
+      "learning_rate": 4.984995383194829e-05,
+      "loss": 1.2223,
+      "step": 147
+    },
+    {
+      "epoch": 0.06627483907080885,
+      "grad_norm": 0.7321466207504272,
+      "learning_rate": 4.983841181902124e-05,
+      "loss": 1.0052,
+      "step": 148
+    },
+    {
+      "epoch": 0.0667226420375035,
+      "grad_norm": 0.6522185206413269,
+      "learning_rate": 4.982686980609418e-05,
+      "loss": 1.0421,
+      "step": 149
+    },
+    {
+      "epoch": 0.06717044500419815,
+      "grad_norm": 0.633614182472229,
+      "learning_rate": 4.981532779316713e-05,
+      "loss": 1.0914,
+      "step": 150
+    },
+    {
+      "epoch": 0.06717044500419815,
+      "eval_loss": 1.0725953578948975,
+      "eval_runtime": 1740.6944,
+      "eval_samples_per_second": 2.566,
+      "eval_steps_per_second": 2.566,
+      "step": 150
+    },
+    {
+      "epoch": 0.0676182479708928,
+      "grad_norm": 0.7060382962226868,
+      "learning_rate": 4.9803785780240074e-05,
+      "loss": 0.891,
+      "step": 151
+    },
+    {
+      "epoch": 0.06806605093758746,
+      "grad_norm": 0.6778922080993652,
+      "learning_rate": 4.9792243767313024e-05,
+      "loss": 0.9576,
+      "step": 152
+    },
+    {
+      "epoch": 0.06851385390428212,
+      "grad_norm": 0.54057776927948,
+      "learning_rate": 4.9780701754385966e-05,
+      "loss": 1.0744,
+      "step": 153
+    },
+    {
+      "epoch": 0.06896165687097677,
+      "grad_norm": 0.6451223492622375,
+      "learning_rate": 4.9769159741458916e-05,
+      "loss": 1.0027,
+      "step": 154
+    },
+    {
+      "epoch": 0.06940945983767143,
+      "grad_norm": 0.6291092038154602,
+      "learning_rate": 4.975761772853186e-05,
+      "loss": 1.2297,
+      "step": 155
+    },
+    {
+      "epoch": 0.06985726280436608,
+      "grad_norm": 0.6851316094398499,
+      "learning_rate": 4.974607571560481e-05,
+      "loss": 1.1893,
+      "step": 156
+    },
+    {
+      "epoch": 0.07030506577106073,
+      "grad_norm": 0.5764050483703613,
+      "learning_rate": 4.973453370267775e-05,
+      "loss": 1.1657,
+      "step": 157
+    },
+    {
+      "epoch": 0.07075286873775538,
+      "grad_norm": 0.5566059350967407,
+      "learning_rate": 4.97229916897507e-05,
+      "loss": 1.0751,
+      "step": 158
+    },
+    {
+      "epoch": 0.07120067170445005,
+      "grad_norm": 0.6990348696708679,
+      "learning_rate": 4.971144967682364e-05,
+      "loss": 1.0401,
+      "step": 159
+    },
+    {
+      "epoch": 0.0716484746711447,
+      "grad_norm": 0.6392049789428711,
+      "learning_rate": 4.9699907663896585e-05,
+      "loss": 1.1104,
+      "step": 160
+    },
+    {
+      "epoch": 0.0716484746711447,
+      "eval_loss": 1.0557206869125366,
+      "eval_runtime": 1740.3422,
+      "eval_samples_per_second": 2.567,
+      "eval_steps_per_second": 2.567,
+      "step": 160
+    },
+    {
+      "epoch": 0.07209627763783935,
+      "grad_norm": 0.5888108015060425,
+      "learning_rate": 4.9688365650969534e-05,
+      "loss": 1.0651,
+      "step": 161
+    },
+    {
+      "epoch": 0.072544080604534,
+      "grad_norm": 0.6952875256538391,
+      "learning_rate": 4.967682363804248e-05,
+      "loss": 1.0902,
+      "step": 162
+    },
+    {
+      "epoch": 0.07299188357122866,
+      "grad_norm": 0.6969994902610779,
+      "learning_rate": 4.9665281625115426e-05,
+      "loss": 0.8659,
+      "step": 163
+    },
+    {
+      "epoch": 0.07343968653792331,
+      "grad_norm": 0.5647038817405701,
+      "learning_rate": 4.965373961218837e-05,
+      "loss": 1.0399,
+      "step": 164
+    },
+    {
+      "epoch": 0.07388748950461797,
+      "grad_norm": 0.5780266523361206,
+      "learning_rate": 4.964219759926131e-05,
+      "loss": 1.2407,
+      "step": 165
+    },
+    {
+      "epoch": 0.07433529247131263,
+      "grad_norm": 0.700043797492981,
+      "learning_rate": 4.9630655586334254e-05,
+      "loss": 1.1373,
+      "step": 166
+    },
+    {
+      "epoch": 0.07478309543800728,
+      "grad_norm": 0.5190924406051636,
+      "learning_rate": 4.96191135734072e-05,
+      "loss": 1.0578,
+      "step": 167
+    },
+    {
+      "epoch": 0.07523089840470193,
+      "grad_norm": 0.5897246599197388,
+      "learning_rate": 4.9607571560480146e-05,
+      "loss": 0.9403,
+      "step": 168
+    },
+    {
+      "epoch": 0.07567870137139658,
+      "grad_norm": 0.7634034156799316,
+      "learning_rate": 4.9596029547553095e-05,
+      "loss": 0.9184,
+      "step": 169
+    },
+    {
+      "epoch": 0.07612650433809123,
+      "grad_norm": 0.6055589914321899,
+      "learning_rate": 4.958448753462604e-05,
+      "loss": 1.0687,
+      "step": 170
+    },
+    {
+      "epoch": 0.07612650433809123,
+      "eval_loss": 1.0446605682373047,
+      "eval_runtime": 1740.2042,
+      "eval_samples_per_second": 2.567,
+      "eval_steps_per_second": 2.567,
+      "step": 170
+    },
+    {
+      "epoch": 0.0765743073047859,
+      "grad_norm": 0.6777039766311646,
+      "learning_rate": 4.957294552169899e-05,
+      "loss": 1.0545,
+      "step": 171
+    },
+    {
+      "epoch": 0.07702211027148055,
+      "grad_norm": 0.602261483669281,
+      "learning_rate": 4.956140350877193e-05,
+      "loss": 0.9887,
+      "step": 172
+    },
+    {
+      "epoch": 0.0774699132381752,
+      "grad_norm": 0.6870952248573303,
+      "learning_rate": 4.954986149584488e-05,
+      "loss": 0.767,
+      "step": 173
+    },
+    {
+      "epoch": 0.07791771620486986,
+      "grad_norm": 0.6584283709526062,
+      "learning_rate": 4.953831948291782e-05,
+      "loss": 1.0888,
+      "step": 174
+    },
+    {
+      "epoch": 0.07836551917156451,
+      "grad_norm": 0.7014456391334534,
+      "learning_rate": 4.952677746999077e-05,
+      "loss": 0.9995,
+      "step": 175
+    },
+    {
+      "epoch": 0.07881332213825916,
+      "grad_norm": 0.7976059913635254,
+      "learning_rate": 4.9515235457063714e-05,
+      "loss": 0.907,
+      "step": 176
+    },
+    {
+      "epoch": 0.07926112510495383,
+      "grad_norm": 0.6465182900428772,
+      "learning_rate": 4.950369344413666e-05,
+      "loss": 1.0508,
+      "step": 177
+    },
+    {
+      "epoch": 0.07970892807164848,
+      "grad_norm": 0.7543056607246399,
+      "learning_rate": 4.9492151431209606e-05,
+      "loss": 1.1151,
+      "step": 178
+    },
+    {
+      "epoch": 0.08015673103834313,
+      "grad_norm": 0.729171633720398,
+      "learning_rate": 4.9480609418282555e-05,
+      "loss": 1.0579,
+      "step": 179
+    },
+    {
+      "epoch": 0.08060453400503778,
+      "grad_norm": 0.6621066927909851,
+      "learning_rate": 4.94690674053555e-05,
+      "loss": 0.915,
+      "step": 180
+    },
+    {
+      "epoch": 0.08060453400503778,
+      "eval_loss": 1.0335291624069214,
+      "eval_runtime": 1740.3768,
+      "eval_samples_per_second": 2.567,
+      "eval_steps_per_second": 2.567,
+      "step": 180
+    },
+    {
+      "epoch": 0.08105233697173243,
+      "grad_norm": 0.7245395183563232,
+      "learning_rate": 4.945752539242845e-05,
+      "loss": 1.0651,
+      "step": 181
+    },
+    {
+      "epoch": 0.08150013993842709,
+      "grad_norm": 0.611437976360321,
+      "learning_rate": 4.944598337950139e-05,
+      "loss": 1.2536,
+      "step": 182
+    },
+    {
+      "epoch": 0.08194794290512175,
+      "grad_norm": 0.6172018051147461,
+      "learning_rate": 4.943444136657433e-05,
+      "loss": 1.197,
+      "step": 183
+    },
+    {
+      "epoch": 0.0823957458718164,
+      "grad_norm": 0.7766463160514832,
+      "learning_rate": 4.9422899353647275e-05,
+      "loss": 1.1375,
+      "step": 184
+    },
+    {
+      "epoch": 0.08284354883851106,
+      "grad_norm": 0.6918166279792786,
+      "learning_rate": 4.9411357340720224e-05,
+      "loss": 0.9037,
+      "step": 185
+    },
+    {
+      "epoch": 0.08329135180520571,
+      "grad_norm": 0.5888485908508301,
+      "learning_rate": 4.939981532779317e-05,
+      "loss": 1.0212,
+      "step": 186
+    },
+    {
+      "epoch": 0.08373915477190036,
+      "grad_norm": 0.6069585680961609,
+      "learning_rate": 4.938827331486611e-05,
+      "loss": 1.0665,
+      "step": 187
+    },
+    {
+      "epoch": 0.08418695773859501,
+      "grad_norm": 0.5719831585884094,
+      "learning_rate": 4.937673130193906e-05,
+      "loss": 1.0863,
+      "step": 188
+    },
+    {
+      "epoch": 0.08463476070528968,
+      "grad_norm": 0.6351702213287354,
+      "learning_rate": 4.9365189289012e-05,
+      "loss": 1.0613,
+      "step": 189
+    },
+    {
+      "epoch": 0.08508256367198433,
+      "grad_norm": 0.7906411290168762,
+      "learning_rate": 4.935364727608495e-05,
+      "loss": 1.0697,
+      "step": 190
+    },
+    {
+      "epoch": 0.08508256367198433,
+      "eval_loss": 1.0170701742172241,
+      "eval_runtime": 1740.2195,
+      "eval_samples_per_second": 2.567,
+      "eval_steps_per_second": 2.567,
+      "step": 190
+    },
+    {
+      "epoch": 0.08553036663867898,
+      "grad_norm": 0.8668298125267029,
+      "learning_rate": 4.9342105263157894e-05,
+      "loss": 0.9181,
+      "step": 191
+    },
+    {
+      "epoch": 0.08597816960537363,
+      "grad_norm": 0.7080042958259583,
+      "learning_rate": 4.933056325023084e-05,
+      "loss": 0.9964,
+      "step": 192
+    },
+    {
+      "epoch": 0.08642597257206829,
+      "grad_norm": 0.5576188564300537,
+      "learning_rate": 4.9319021237303786e-05,
+      "loss": 0.8769,
+      "step": 193
+    },
+    {
+      "epoch": 0.08687377553876295,
+      "grad_norm": 0.685297966003418,
+      "learning_rate": 4.9307479224376735e-05,
+      "loss": 1.1107,
+      "step": 194
+    },
+    {
+      "epoch": 0.0873215785054576,
+      "grad_norm": 0.6759443879127502,
+      "learning_rate": 4.929593721144968e-05,
+      "loss": 1.0284,
+      "step": 195
+    },
+    {
+      "epoch": 0.08776938147215226,
+      "grad_norm": 0.7277442216873169,
+      "learning_rate": 4.928439519852263e-05,
+      "loss": 1.0706,
+      "step": 196
+    },
+    {
+      "epoch": 0.08821718443884691,
+      "grad_norm": 0.7153092622756958,
+      "learning_rate": 4.927285318559557e-05,
+      "loss": 1.046,
+      "step": 197
+    },
+    {
+      "epoch": 0.08866498740554156,
+      "grad_norm": 0.7023780941963196,
+      "learning_rate": 4.926131117266852e-05,
+      "loss": 1.1325,
+      "step": 198
+    },
+    {
+      "epoch": 0.08911279037223621,
+      "grad_norm": 0.6779207587242126,
+      "learning_rate": 4.924976915974146e-05,
+      "loss": 0.8576,
+      "step": 199
+    },
+    {
+      "epoch": 0.08956059333893088,
+      "grad_norm": 0.630558967590332,
+      "learning_rate": 4.923822714681441e-05,
+      "loss": 0.9907,
+      "step": 200
+    },
+    {
+      "epoch": 0.08956059333893088,
+      "eval_loss": 1.006293773651123,
+      "eval_runtime": 1739.5536,
+      "eval_samples_per_second": 2.568,
+      "eval_steps_per_second": 2.568,
+      "step": 200
+    },
+    {
+      "epoch": 0.09000839630562553,
+      "grad_norm": 0.59002685546875,
+      "learning_rate": 4.9226685133887354e-05,
+      "loss": 0.9343,
+      "step": 201
+    },
+    {
+      "epoch": 0.09045619927232018,
+      "grad_norm": 0.728977382183075,
+      "learning_rate": 4.92151431209603e-05,
+      "loss": 0.9973,
+      "step": 202
+    },
+    {
+      "epoch": 0.09090400223901483,
+      "grad_norm": 0.6520524621009827,
+      "learning_rate": 4.9203601108033246e-05,
+      "loss": 0.9441,
+      "step": 203
+    },
+    {
+      "epoch": 0.09135180520570949,
+      "grad_norm": 0.7074622511863708,
+      "learning_rate": 4.919205909510619e-05,
+      "loss": 0.9559,
+      "step": 204
+    },
+    {
+      "epoch": 0.09179960817240414,
+      "grad_norm": 0.7225532531738281,
+      "learning_rate": 4.918051708217913e-05,
+      "loss": 0.9087,
+      "step": 205
+    },
+    {
+      "epoch": 0.0922474111390988,
+      "grad_norm": 0.7118654251098633,
+      "learning_rate": 4.916897506925208e-05,
+      "loss": 0.9611,
+      "step": 206
+    },
+    {
+      "epoch": 0.09269521410579346,
+      "grad_norm": 0.6199821829795837,
+      "learning_rate": 4.915743305632502e-05,
+      "loss": 0.9679,
+      "step": 207
+    },
+    {
+      "epoch": 0.09314301707248811,
+      "grad_norm": 0.6176968812942505,
+      "learning_rate": 4.914589104339797e-05,
+      "loss": 1.0421,
+      "step": 208
+    },
+    {
+      "epoch": 0.09359082003918276,
+      "grad_norm": 0.811738133430481,
+      "learning_rate": 4.9134349030470915e-05,
+      "loss": 0.8281,
+      "step": 209
+    },
+    {
+      "epoch": 0.09403862300587741,
+      "grad_norm": 0.7109337449073792,
+      "learning_rate": 4.912280701754386e-05,
+      "loss": 1.1078,
+      "step": 210
+    },
+    {
+      "epoch": 0.09403862300587741,
+      "eval_loss": 0.996062695980072,
+      "eval_runtime": 1737.4871,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 210
+    },
+    {
+      "epoch": 0.09448642597257206,
+      "grad_norm": 0.6960453987121582,
+      "learning_rate": 4.911126500461681e-05,
+      "loss": 0.9872,
+      "step": 211
+    },
+    {
+      "epoch": 0.09493422893926673,
+      "grad_norm": 0.7350823879241943,
+      "learning_rate": 4.909972299168975e-05,
+      "loss": 0.8907,
+      "step": 212
+    },
+    {
+      "epoch": 0.09538203190596138,
+      "grad_norm": 0.722599983215332,
+      "learning_rate": 4.90881809787627e-05,
+      "loss": 1.0963,
+      "step": 213
+    },
+    {
+      "epoch": 0.09582983487265603,
+      "grad_norm": 0.6781536936759949,
+      "learning_rate": 4.907663896583564e-05,
+      "loss": 0.8545,
+      "step": 214
+    },
+    {
+      "epoch": 0.09627763783935069,
+      "grad_norm": 0.7184162735939026,
+      "learning_rate": 4.906509695290859e-05,
+      "loss": 0.9652,
+      "step": 215
+    },
+    {
+      "epoch": 0.09672544080604534,
+      "grad_norm": 0.8125225305557251,
+      "learning_rate": 4.905355493998153e-05,
+      "loss": 1.0223,
+      "step": 216
+    },
+    {
+      "epoch": 0.09717324377273999,
+      "grad_norm": 0.7977463603019714,
+      "learning_rate": 4.904201292705448e-05,
+      "loss": 1.2184,
+      "step": 217
+    },
+    {
+      "epoch": 0.09762104673943466,
+      "grad_norm": 0.696886420249939,
+      "learning_rate": 4.9030470914127425e-05,
+      "loss": 0.9601,
+      "step": 218
+    },
+    {
+      "epoch": 0.09806884970612931,
+      "grad_norm": 0.6477362513542175,
+      "learning_rate": 4.9018928901200375e-05,
+      "loss": 1.166,
+      "step": 219
+    },
+    {
+      "epoch": 0.09851665267282396,
+      "grad_norm": 0.7791169285774231,
+      "learning_rate": 4.900738688827332e-05,
+      "loss": 1.0317,
+      "step": 220
+    },
+    {
+      "epoch": 0.09851665267282396,
+      "eval_loss": 0.9866155982017517,
+      "eval_runtime": 1737.5401,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 220
+    },
+    {
+      "epoch": 0.09896445563951861,
+      "grad_norm": 0.7042242884635925,
+      "learning_rate": 4.899584487534627e-05,
+      "loss": 1.0636,
+      "step": 221
+    },
+    {
+      "epoch": 0.09941225860621326,
+      "grad_norm": 0.6456022262573242,
+      "learning_rate": 4.898430286241921e-05,
+      "loss": 1.0597,
+      "step": 222
+    },
+    {
+      "epoch": 0.09986006157290792,
+      "grad_norm": 0.796667754650116,
+      "learning_rate": 4.897276084949215e-05,
+      "loss": 1.0915,
+      "step": 223
+    },
+    {
+      "epoch": 0.10030786453960258,
+      "grad_norm": 0.7900235056877136,
+      "learning_rate": 4.89612188365651e-05,
+      "loss": 0.9049,
+      "step": 224
+    },
+    {
+      "epoch": 0.10075566750629723,
+      "grad_norm": 0.689604640007019,
+      "learning_rate": 4.8949676823638044e-05,
+      "loss": 0.9269,
+      "step": 225
+    },
+    {
+      "epoch": 0.10120347047299189,
+      "grad_norm": 0.8840981125831604,
+      "learning_rate": 4.8938134810710987e-05,
+      "loss": 0.9206,
+      "step": 226
+    },
+    {
+      "epoch": 0.10165127343968654,
+      "grad_norm": 0.9159313440322876,
+      "learning_rate": 4.8926592797783936e-05,
+      "loss": 1.0473,
+      "step": 227
+    },
+    {
+      "epoch": 0.10209907640638119,
+      "grad_norm": 0.7433324456214905,
+      "learning_rate": 4.891505078485688e-05,
+      "loss": 0.8495,
+      "step": 228
+    },
+    {
+      "epoch": 0.10254687937307584,
+      "grad_norm": 0.813946008682251,
+      "learning_rate": 4.890350877192983e-05,
+      "loss": 0.9191,
+      "step": 229
+    },
+    {
+      "epoch": 0.10299468233977051,
+      "grad_norm": 0.7069205045700073,
+      "learning_rate": 4.889196675900277e-05,
+      "loss": 0.8153,
+      "step": 230
+    },
+    {
+      "epoch": 0.10299468233977051,
+      "eval_loss": 0.9733306765556335,
+      "eval_runtime": 1737.4261,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 230
+    },
+    {
+      "epoch": 0.10344248530646516,
+      "grad_norm": 0.7403346300125122,
+      "learning_rate": 4.888042474607572e-05,
+      "loss": 0.8176,
+      "step": 231
+    },
+    {
+      "epoch": 0.10389028827315981,
+      "grad_norm": 0.779050886631012,
+      "learning_rate": 4.886888273314866e-05,
+      "loss": 1.1263,
+      "step": 232
+    },
+    {
+      "epoch": 0.10433809123985446,
+      "grad_norm": 0.7971842885017395,
+      "learning_rate": 4.885734072022161e-05,
+      "loss": 0.7987,
+      "step": 233
+    },
+    {
+      "epoch": 0.10478589420654912,
+      "grad_norm": 0.8085238933563232,
+      "learning_rate": 4.8845798707294554e-05,
+      "loss": 0.8065,
+      "step": 234
+    },
+    {
+      "epoch": 0.10523369717324377,
+      "grad_norm": 0.7707525491714478,
+      "learning_rate": 4.88342566943675e-05,
+      "loss": 0.9083,
+      "step": 235
+    },
+    {
+      "epoch": 0.10568150013993843,
+      "grad_norm": 0.7241260409355164,
+      "learning_rate": 4.8822714681440446e-05,
+      "loss": 0.9536,
+      "step": 236
+    },
+    {
+      "epoch": 0.10612930310663309,
+      "grad_norm": 0.760982871055603,
+      "learning_rate": 4.881117266851339e-05,
+      "loss": 0.9606,
+      "step": 237
+    },
+    {
+      "epoch": 0.10657710607332774,
+      "grad_norm": 0.7906852960586548,
+      "learning_rate": 4.879963065558634e-05,
+      "loss": 0.6995,
+      "step": 238
+    },
+    {
+      "epoch": 0.10702490904002239,
+      "grad_norm": 0.8570147752761841,
+      "learning_rate": 4.878808864265928e-05,
+      "loss": 1.0152,
+      "step": 239
+    },
+    {
+      "epoch": 0.10747271200671704,
+      "grad_norm": 0.7524833679199219,
+      "learning_rate": 4.877654662973223e-05,
+      "loss": 0.9486,
+      "step": 240
+    },
+    {
+      "epoch": 0.10747271200671704,
+      "eval_loss": 0.962479829788208,
+      "eval_runtime": 1737.2243,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 240
+    },
+    {
+      "epoch": 0.1079205149734117,
+      "grad_norm": 0.6601128578186035,
+      "learning_rate": 4.876500461680517e-05,
+      "loss": 0.8763,
+      "step": 241
+    },
+    {
+      "epoch": 0.10836831794010636,
+      "grad_norm": 0.7229903936386108,
+      "learning_rate": 4.875346260387812e-05,
+      "loss": 1.0229,
+      "step": 242
+    },
+    {
+      "epoch": 0.10881612090680101,
+      "grad_norm": 0.7775028944015503,
+      "learning_rate": 4.8741920590951065e-05,
+      "loss": 0.9157,
+      "step": 243
+    },
+    {
+      "epoch": 0.10926392387349566,
+      "grad_norm": 0.8382840752601624,
+      "learning_rate": 4.873037857802401e-05,
+      "loss": 0.942,
+      "step": 244
+    },
+    {
+      "epoch": 0.10971172684019032,
+      "grad_norm": 0.6494114995002747,
+      "learning_rate": 4.871883656509695e-05,
+      "loss": 1.0415,
+      "step": 245
+    },
+    {
+      "epoch": 0.11015952980688497,
+      "grad_norm": 0.6916049122810364,
+      "learning_rate": 4.87072945521699e-05,
+      "loss": 0.9285,
+      "step": 246
+    },
+    {
+      "epoch": 0.11060733277357962,
+      "grad_norm": 0.7655143737792969,
+      "learning_rate": 4.869575253924284e-05,
+      "loss": 0.8565,
+      "step": 247
+    },
+    {
+      "epoch": 0.11105513574027429,
+      "grad_norm": 0.8227872848510742,
+      "learning_rate": 4.868421052631579e-05,
+      "loss": 0.933,
+      "step": 248
+    },
+    {
+      "epoch": 0.11150293870696894,
+      "grad_norm": 0.6931448578834534,
+      "learning_rate": 4.8672668513388734e-05,
+      "loss": 0.8383,
+      "step": 249
+    },
+    {
+      "epoch": 0.11195074167366359,
+      "grad_norm": 0.8759434819221497,
+      "learning_rate": 4.8661126500461684e-05,
+      "loss": 0.9163,
+      "step": 250
+    },
+    {
+      "epoch": 0.11195074167366359,
+      "eval_loss": 0.9523711204528809,
+      "eval_runtime": 1737.1441,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 250
+    },
+    {
+      "epoch": 0.11239854464035824,
+      "grad_norm": 0.7037357687950134,
+      "learning_rate": 4.8649584487534626e-05,
+      "loss": 1.0964,
+      "step": 251
+    },
+    {
+      "epoch": 0.1128463476070529,
+      "grad_norm": 0.7379258871078491,
+      "learning_rate": 4.8638042474607576e-05,
+      "loss": 1.0948,
+      "step": 252
+    },
+    {
+      "epoch": 0.11329415057374755,
+      "grad_norm": 0.6891792416572571,
+      "learning_rate": 4.862650046168052e-05,
+      "loss": 1.0872,
+      "step": 253
+    },
+    {
+      "epoch": 0.11374195354044221,
+      "grad_norm": 0.6693180203437805,
+      "learning_rate": 4.861495844875347e-05,
+      "loss": 1.0345,
+      "step": 254
+    },
+    {
+      "epoch": 0.11418975650713686,
+      "grad_norm": 0.7131677269935608,
+      "learning_rate": 4.860341643582641e-05,
+      "loss": 0.8185,
+      "step": 255
+    },
+    {
+      "epoch": 0.11463755947383152,
+      "grad_norm": 0.6727960109710693,
+      "learning_rate": 4.859187442289936e-05,
+      "loss": 1.047,
+      "step": 256
+    },
+    {
+      "epoch": 0.11508536244052617,
+      "grad_norm": 0.7159959673881531,
+      "learning_rate": 4.85803324099723e-05,
+      "loss": 1.0208,
+      "step": 257
+    },
+    {
+      "epoch": 0.11553316540722082,
+      "grad_norm": 0.7980951070785522,
+      "learning_rate": 4.856879039704525e-05,
+      "loss": 0.829,
+      "step": 258
+    },
+    {
+      "epoch": 0.11598096837391547,
+      "grad_norm": 0.6713117957115173,
+      "learning_rate": 4.8557248384118194e-05,
+      "loss": 0.8269,
+      "step": 259
+    },
+    {
+      "epoch": 0.11642877134061014,
+      "grad_norm": 0.8814539313316345,
+      "learning_rate": 4.854570637119114e-05,
+      "loss": 0.8441,
+      "step": 260
+    },
+    {
+      "epoch": 0.11642877134061014,
+      "eval_loss": 0.9442258477210999,
+      "eval_runtime": 1737.8062,
+      "eval_samples_per_second": 2.57,
+      "eval_steps_per_second": 2.57,
+      "step": 260
+    },
+    {
+      "epoch": 0.11687657430730479,
+      "grad_norm": 0.7765588164329529,
+      "learning_rate": 4.8534164358264086e-05,
+      "loss": 0.9471,
+      "step": 261
+    },
+    {
+      "epoch": 0.11732437727399944,
+      "grad_norm": 0.7720988988876343,
+      "learning_rate": 4.852262234533703e-05,
+      "loss": 1.1604,
+      "step": 262
+    },
+    {
+      "epoch": 0.1177721802406941,
+      "grad_norm": 0.6977170705795288,
+      "learning_rate": 4.851108033240998e-05,
+      "loss": 0.9949,
+      "step": 263
+    },
+    {
+      "epoch": 0.11821998320738875,
+      "grad_norm": 0.7317233681678772,
+      "learning_rate": 4.849953831948292e-05,
+      "loss": 1.0061,
+      "step": 264
+    },
+    {
+      "epoch": 0.1186677861740834,
+      "grad_norm": 0.79209303855896,
+      "learning_rate": 4.848799630655586e-05,
+      "loss": 0.8906,
+      "step": 265
+    },
+    {
+      "epoch": 0.11911558914077806,
+      "grad_norm": 0.7325365543365479,
+      "learning_rate": 4.8476454293628806e-05,
+      "loss": 0.8362,
+      "step": 266
+    },
+    {
+      "epoch": 0.11956339210747272,
+      "grad_norm": 0.7075287699699402,
+      "learning_rate": 4.8464912280701755e-05,
+      "loss": 1.0315,
+      "step": 267
+    },
+    {
+      "epoch": 0.12001119507416737,
+      "grad_norm": 0.7248746752738953,
+      "learning_rate": 4.84533702677747e-05,
+      "loss": 1.0958,
+      "step": 268
+    },
+    {
+      "epoch": 0.12045899804086202,
+      "grad_norm": 0.8397259712219238,
+      "learning_rate": 4.844182825484765e-05,
+      "loss": 0.8609,
+      "step": 269
+    },
+    {
+      "epoch": 0.12090680100755667,
+      "grad_norm": 0.6971864700317383,
+      "learning_rate": 4.843028624192059e-05,
+      "loss": 1.0048,
+      "step": 270
+    },
+    {
+      "epoch": 0.12090680100755667,
+      "eval_loss": 0.9325976371765137,
+      "eval_runtime": 1737.3018,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 270
+    },
+    {
+      "epoch": 0.12135460397425132,
+      "grad_norm": 0.7825617790222168,
+      "learning_rate": 4.841874422899354e-05,
+      "loss": 1.0109,
+      "step": 271
+    },
+    {
+      "epoch": 0.12180240694094599,
+      "grad_norm": 0.7868365049362183,
+      "learning_rate": 4.840720221606648e-05,
+      "loss": 0.8101,
+      "step": 272
+    },
+    {
+      "epoch": 0.12225020990764064,
+      "grad_norm": 0.7207172513008118,
+      "learning_rate": 4.839566020313943e-05,
+      "loss": 0.9749,
+      "step": 273
+    },
+    {
+      "epoch": 0.1226980128743353,
+      "grad_norm": 0.6478491425514221,
+      "learning_rate": 4.8384118190212374e-05,
+      "loss": 0.9946,
+      "step": 274
+    },
+    {
+      "epoch": 0.12314581584102995,
+      "grad_norm": 0.8110407590866089,
+      "learning_rate": 4.837257617728532e-05,
+      "loss": 1.0575,
+      "step": 275
+    },
+    {
+      "epoch": 0.1235936188077246,
+      "grad_norm": 0.8174804449081421,
+      "learning_rate": 4.8361034164358266e-05,
+      "loss": 1.1115,
+      "step": 276
+    },
+    {
+      "epoch": 0.12404142177441925,
+      "grad_norm": 0.7011256217956543,
+      "learning_rate": 4.8349492151431215e-05,
+      "loss": 1.0387,
+      "step": 277
+    },
+    {
+      "epoch": 0.12448922474111392,
+      "grad_norm": 0.736200213432312,
+      "learning_rate": 4.833795013850416e-05,
+      "loss": 0.6846,
+      "step": 278
+    },
+    {
+      "epoch": 0.12493702770780857,
+      "grad_norm": 0.7827062606811523,
+      "learning_rate": 4.832640812557711e-05,
+      "loss": 0.8612,
+      "step": 279
+    },
+    {
+      "epoch": 0.1253848306745032,
+      "grad_norm": 0.7452975511550903,
+      "learning_rate": 4.831486611265005e-05,
+      "loss": 0.8897,
+      "step": 280
+    },
+    {
+      "epoch": 0.1253848306745032,
+      "eval_loss": 0.9226668477058411,
+      "eval_runtime": 1737.6416,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 280
+    },
+    {
+      "epoch": 0.12583263364119787,
+      "grad_norm": 0.6109349131584167,
+      "learning_rate": 4.8303324099723e-05,
+      "loss": 0.8418,
+      "step": 281
+    },
+    {
+      "epoch": 0.12628043660789254,
+      "grad_norm": 0.6637207269668579,
+      "learning_rate": 4.829178208679594e-05,
+      "loss": 0.949,
+      "step": 282
+    },
+    {
+      "epoch": 0.12672823957458718,
+      "grad_norm": 0.8242074251174927,
+      "learning_rate": 4.8280240073868885e-05,
+      "loss": 0.8032,
+      "step": 283
+    },
+    {
+      "epoch": 0.12717604254128184,
+      "grad_norm": 0.725492000579834,
+      "learning_rate": 4.826869806094183e-05,
+      "loss": 0.949,
+      "step": 284
+    },
+    {
+      "epoch": 0.12762384550797648,
+      "grad_norm": 0.7447363138198853,
+      "learning_rate": 4.8257156048014777e-05,
+      "loss": 0.8985,
+      "step": 285
+    },
+    {
+      "epoch": 0.12807164847467115,
+      "grad_norm": 0.8081015348434448,
+      "learning_rate": 4.824561403508772e-05,
+      "loss": 0.7457,
+      "step": 286
+    },
+    {
+      "epoch": 0.1285194514413658,
+      "grad_norm": 0.735405683517456,
+      "learning_rate": 4.823407202216066e-05,
+      "loss": 0.9313,
+      "step": 287
+    },
+    {
+      "epoch": 0.12896725440806045,
+      "grad_norm": 0.8218609690666199,
+      "learning_rate": 4.822253000923361e-05,
+      "loss": 0.997,
+      "step": 288
+    },
+    {
+      "epoch": 0.12941505737475512,
+      "grad_norm": 0.6740204095840454,
+      "learning_rate": 4.8210987996306554e-05,
+      "loss": 1.0128,
+      "step": 289
+    },
+    {
+      "epoch": 0.12986286034144975,
+      "grad_norm": 0.9718872904777527,
+      "learning_rate": 4.81994459833795e-05,
+      "loss": 0.9457,
+      "step": 290
+    },
+    {
+      "epoch": 0.12986286034144975,
+      "eval_loss": 0.9150700569152832,
+      "eval_runtime": 1737.2884,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 290
+    },
+    {
+      "epoch": 0.13031066330814442,
+      "grad_norm": 0.8290822505950928,
+      "learning_rate": 4.8187903970452446e-05,
+      "loss": 1.0297,
+      "step": 291
+    },
+    {
+      "epoch": 0.13075846627483906,
+      "grad_norm": 0.9061235189437866,
+      "learning_rate": 4.8176361957525395e-05,
+      "loss": 0.7766,
+      "step": 292
+    },
+    {
+      "epoch": 0.13120626924153372,
+      "grad_norm": 0.6609340310096741,
+      "learning_rate": 4.816481994459834e-05,
+      "loss": 0.9471,
+      "step": 293
+    },
+    {
+      "epoch": 0.1316540722082284,
+      "grad_norm": 0.7579784989356995,
+      "learning_rate": 4.815327793167129e-05,
+      "loss": 0.9807,
+      "step": 294
+    },
+    {
+      "epoch": 0.13210187517492303,
+      "grad_norm": 0.7428559064865112,
+      "learning_rate": 4.814173591874423e-05,
+      "loss": 0.9862,
+      "step": 295
+    },
+    {
+      "epoch": 0.1325496781416177,
+      "grad_norm": 0.8760770559310913,
+      "learning_rate": 4.813019390581718e-05,
+      "loss": 0.7563,
+      "step": 296
+    },
+    {
+      "epoch": 0.13299748110831233,
+      "grad_norm": 0.7334847450256348,
+      "learning_rate": 4.811865189289012e-05,
+      "loss": 0.9554,
+      "step": 297
+    },
+    {
+      "epoch": 0.133445284075007,
+      "grad_norm": 0.8345491886138916,
+      "learning_rate": 4.810710987996307e-05,
+      "loss": 0.9365,
+      "step": 298
+    },
+    {
+      "epoch": 0.13389308704170166,
+      "grad_norm": 0.8603995442390442,
+      "learning_rate": 4.8095567867036014e-05,
+      "loss": 0.8086,
+      "step": 299
+    },
+    {
+      "epoch": 0.1343408900083963,
+      "grad_norm": 0.7264519333839417,
+      "learning_rate": 4.808402585410896e-05,
+      "loss": 0.9419,
+      "step": 300
+    },
+    {
+      "epoch": 0.1343408900083963,
+      "eval_loss": 0.9055705666542053,
+      "eval_runtime": 1737.7291,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 300
+    },
+    {
+      "epoch": 0.13478869297509097,
+      "grad_norm": 0.7371543645858765,
+      "learning_rate": 4.8072483841181906e-05,
+      "loss": 0.7665,
+      "step": 301
+    },
+    {
+      "epoch": 0.1352364959417856,
+      "grad_norm": 0.6498313546180725,
+      "learning_rate": 4.8060941828254855e-05,
+      "loss": 0.9527,
+      "step": 302
+    },
+    {
+      "epoch": 0.13568429890848027,
+      "grad_norm": 0.7187266945838928,
+      "learning_rate": 4.80493998153278e-05,
+      "loss": 0.9354,
+      "step": 303
+    },
+    {
+      "epoch": 0.1361321018751749,
+      "grad_norm": 0.6874094605445862,
+      "learning_rate": 4.803785780240074e-05,
+      "loss": 0.9004,
+      "step": 304
+    },
+    {
+      "epoch": 0.13657990484186958,
+      "grad_norm": 0.7286795973777771,
+      "learning_rate": 4.802631578947368e-05,
+      "loss": 0.7882,
+      "step": 305
+    },
+    {
+      "epoch": 0.13702770780856424,
+      "grad_norm": 0.6951384544372559,
+      "learning_rate": 4.801477377654663e-05,
+      "loss": 0.9683,
+      "step": 306
+    },
+    {
+      "epoch": 0.13747551077525888,
+      "grad_norm": 0.7582501769065857,
+      "learning_rate": 4.8003231763619575e-05,
+      "loss": 0.9498,
+      "step": 307
+    },
+    {
+      "epoch": 0.13792331374195355,
+      "grad_norm": 0.8873469233512878,
+      "learning_rate": 4.7991689750692524e-05,
+      "loss": 1.0083,
+      "step": 308
+    },
+    {
+      "epoch": 0.13837111670864818,
+      "grad_norm": 0.8048168420791626,
+      "learning_rate": 4.798014773776547e-05,
+      "loss": 0.9438,
+      "step": 309
+    },
+    {
+      "epoch": 0.13881891967534285,
+      "grad_norm": 0.6849039793014526,
+      "learning_rate": 4.7968605724838416e-05,
+      "loss": 0.9656,
+      "step": 310
+    },
+    {
+      "epoch": 0.13881891967534285,
+      "eval_loss": 0.8966045379638672,
+      "eval_runtime": 1736.6349,
+      "eval_samples_per_second": 2.572,
+      "eval_steps_per_second": 2.572,
+      "step": 310
+    },
+    {
+      "epoch": 0.13926672264203752,
+      "grad_norm": 0.7566764950752258,
+      "learning_rate": 4.795706371191136e-05,
+      "loss": 0.8435,
+      "step": 311
+    },
+    {
+      "epoch": 0.13971452560873215,
+      "grad_norm": 0.8874958157539368,
+      "learning_rate": 4.79455216989843e-05,
+      "loss": 1.028,
+      "step": 312
+    },
+    {
+      "epoch": 0.14016232857542682,
+      "grad_norm": 0.830262303352356,
+      "learning_rate": 4.793397968605725e-05,
+      "loss": 0.8655,
+      "step": 313
+    },
+    {
+      "epoch": 0.14061013154212146,
+      "grad_norm": 0.8174196481704712,
+      "learning_rate": 4.7922437673130193e-05,
+      "loss": 0.9186,
+      "step": 314
+    },
+    {
+      "epoch": 0.14105793450881612,
+      "grad_norm": 0.7265437841415405,
+      "learning_rate": 4.791089566020314e-05,
+      "loss": 1.0262,
+      "step": 315
+    },
+    {
+      "epoch": 0.14150573747551076,
+      "grad_norm": 0.7852171063423157,
+      "learning_rate": 4.7899353647276085e-05,
+      "loss": 0.8919,
+      "step": 316
+    },
+    {
+      "epoch": 0.14195354044220543,
+      "grad_norm": 0.8389447331428528,
+      "learning_rate": 4.7887811634349035e-05,
+      "loss": 1.0888,
+      "step": 317
+    },
+    {
+      "epoch": 0.1424013434089001,
+      "grad_norm": 0.9585886597633362,
+      "learning_rate": 4.787626962142198e-05,
+      "loss": 0.9379,
+      "step": 318
+    },
+    {
+      "epoch": 0.14284914637559473,
+      "grad_norm": 0.8709464073181152,
+      "learning_rate": 4.786472760849493e-05,
+      "loss": 0.8661,
+      "step": 319
+    },
+    {
+      "epoch": 0.1432969493422894,
+      "grad_norm": 0.7449386715888977,
+      "learning_rate": 4.785318559556787e-05,
+      "loss": 0.9359,
+      "step": 320
+    },
+    {
+      "epoch": 0.1432969493422894,
+      "eval_loss": 0.8892737030982971,
+      "eval_runtime": 1737.4213,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 320
+    },
+    {
+      "epoch": 0.14374475230898404,
+      "grad_norm": 0.9661244750022888,
+      "learning_rate": 4.784164358264082e-05,
+      "loss": 0.7743,
+      "step": 321
+    },
+    {
+      "epoch": 0.1441925552756787,
+      "grad_norm": 0.7624433040618896,
+      "learning_rate": 4.783010156971376e-05,
+      "loss": 1.0915,
+      "step": 322
+    },
+    {
+      "epoch": 0.14464035824237337,
+      "grad_norm": 0.8274304270744324,
+      "learning_rate": 4.7818559556786704e-05,
+      "loss": 0.9038,
+      "step": 323
+    },
+    {
+      "epoch": 0.145088161209068,
+      "grad_norm": 0.746432900428772,
+      "learning_rate": 4.780701754385965e-05,
+      "loss": 1.0964,
+      "step": 324
+    },
+    {
+      "epoch": 0.14553596417576267,
+      "grad_norm": 0.8682849407196045,
+      "learning_rate": 4.7795475530932596e-05,
+      "loss": 0.849,
+      "step": 325
+    },
+    {
+      "epoch": 0.1459837671424573,
+      "grad_norm": 0.761936604976654,
+      "learning_rate": 4.778393351800554e-05,
+      "loss": 0.8772,
+      "step": 326
+    },
+    {
+      "epoch": 0.14643157010915198,
+      "grad_norm": 0.8509253859519958,
+      "learning_rate": 4.777239150507849e-05,
+      "loss": 1.1404,
+      "step": 327
+    },
+    {
+      "epoch": 0.14687937307584661,
+      "grad_norm": 1.0035078525543213,
+      "learning_rate": 4.776084949215143e-05,
+      "loss": 0.8262,
+      "step": 328
+    },
+    {
+      "epoch": 0.14732717604254128,
+      "grad_norm": 0.8431288003921509,
+      "learning_rate": 4.774930747922438e-05,
+      "loss": 0.9464,
+      "step": 329
+    },
+    {
+      "epoch": 0.14777497900923595,
+      "grad_norm": 0.7923704385757446,
+      "learning_rate": 4.773776546629732e-05,
+      "loss": 1.1387,
+      "step": 330
+    },
+    {
+      "epoch": 0.14777497900923595,
+      "eval_loss": 0.8819145560264587,
+      "eval_runtime": 1737.0418,
+      "eval_samples_per_second": 2.572,
+      "eval_steps_per_second": 2.572,
+      "step": 330
+    },
+    {
+      "epoch": 0.14822278197593058,
+      "grad_norm": 0.7270142436027527,
+      "learning_rate": 4.772622345337027e-05,
+      "loss": 1.0828,
+      "step": 331
+    },
+    {
+      "epoch": 0.14867058494262525,
+      "grad_norm": 0.7770641446113586,
+      "learning_rate": 4.7714681440443215e-05,
+      "loss": 0.875,
+      "step": 332
+    },
+    {
+      "epoch": 0.1491183879093199,
+      "grad_norm": 0.7159392833709717,
+      "learning_rate": 4.7703139427516164e-05,
+      "loss": 0.818,
+      "step": 333
+    },
+    {
+      "epoch": 0.14956619087601455,
+      "grad_norm": 0.7923138737678528,
+      "learning_rate": 4.7691597414589107e-05,
+      "loss": 1.0655,
+      "step": 334
+    },
+    {
+      "epoch": 0.15001399384270922,
+      "grad_norm": 0.7565407156944275,
+      "learning_rate": 4.7680055401662056e-05,
+      "loss": 0.7377,
+      "step": 335
+    },
+    {
+      "epoch": 0.15046179680940386,
+      "grad_norm": 0.713832437992096,
+      "learning_rate": 4.7668513388735e-05,
+      "loss": 1.0459,
+      "step": 336
+    },
+    {
+      "epoch": 0.15090959977609852,
+      "grad_norm": 0.774344801902771,
+      "learning_rate": 4.765697137580794e-05,
+      "loss": 1.0397,
+      "step": 337
+    },
+    {
+      "epoch": 0.15135740274279316,
+      "grad_norm": 0.7079201936721802,
+      "learning_rate": 4.764542936288089e-05,
+      "loss": 0.9431,
+      "step": 338
+    },
+    {
+      "epoch": 0.15180520570948783,
+      "grad_norm": 0.8966906070709229,
+      "learning_rate": 4.763388734995383e-05,
+      "loss": 0.8048,
+      "step": 339
+    },
+    {
+      "epoch": 0.15225300867618247,
+      "grad_norm": 0.7640495300292969,
+      "learning_rate": 4.762234533702678e-05,
+      "loss": 1.0161,
+      "step": 340
+    },
+    {
+      "epoch": 0.15225300867618247,
+      "eval_loss": 0.8745113611221313,
+      "eval_runtime": 1737.1711,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 340
+    },
+    {
+      "epoch": 0.15270081164287713,
+      "grad_norm": 0.9535894393920898,
+      "learning_rate": 4.7610803324099725e-05,
+      "loss": 1.1023,
+      "step": 341
+    },
+    {
+      "epoch": 0.1531486146095718,
+      "grad_norm": 0.7694385051727295,
+      "learning_rate": 4.7599261311172675e-05,
+      "loss": 1.0471,
+      "step": 342
+    },
+    {
+      "epoch": 0.15359641757626644,
+      "grad_norm": 0.7791855931282043,
+      "learning_rate": 4.758771929824562e-05,
+      "loss": 1.0974,
+      "step": 343
+    },
+    {
+      "epoch": 0.1540442205429611,
+      "grad_norm": 0.8553569912910461,
+      "learning_rate": 4.757617728531856e-05,
+      "loss": 1.0196,
+      "step": 344
+    },
+    {
+      "epoch": 0.15449202350965574,
+      "grad_norm": 0.8217856287956238,
+      "learning_rate": 4.75646352723915e-05,
+      "loss": 0.6761,
+      "step": 345
+    },
+    {
+      "epoch": 0.1549398264763504,
+      "grad_norm": 0.8384736776351929,
+      "learning_rate": 4.755309325946445e-05,
+      "loss": 1.0592,
+      "step": 346
+    },
+    {
+      "epoch": 0.15538762944304507,
+      "grad_norm": 0.774459958076477,
+      "learning_rate": 4.7541551246537394e-05,
+      "loss": 0.7733,
+      "step": 347
+    },
+    {
+      "epoch": 0.1558354324097397,
+      "grad_norm": 0.8029676675796509,
+      "learning_rate": 4.7530009233610344e-05,
+      "loss": 0.9376,
+      "step": 348
+    },
+    {
+      "epoch": 0.15628323537643438,
+      "grad_norm": 0.7330321073532104,
+      "learning_rate": 4.7518467220683286e-05,
+      "loss": 0.986,
+      "step": 349
+    },
+    {
+      "epoch": 0.15673103834312901,
+      "grad_norm": 0.7578117251396179,
+      "learning_rate": 4.7506925207756236e-05,
+      "loss": 0.7285,
+      "step": 350
+    },
+    {
+      "epoch": 0.15673103834312901,
+      "eval_loss": 0.867455005645752,
+      "eval_runtime": 1737.4591,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 350
+    },
+    {
+      "epoch": 0.15717884130982368,
+      "grad_norm": 0.9032571911811829,
+      "learning_rate": 4.749538319482918e-05,
+      "loss": 0.933,
+      "step": 351
+    },
+    {
+      "epoch": 0.15762664427651832,
+      "grad_norm": 0.9054291844367981,
+      "learning_rate": 4.748384118190213e-05,
+      "loss": 0.7717,
+      "step": 352
+    },
+    {
+      "epoch": 0.15807444724321298,
+      "grad_norm": 0.9551945328712463,
+      "learning_rate": 4.747229916897507e-05,
+      "loss": 0.9073,
+      "step": 353
+    },
+    {
+      "epoch": 0.15852225020990765,
+      "grad_norm": 0.9689154624938965,
+      "learning_rate": 4.746075715604802e-05,
+      "loss": 0.8513,
+      "step": 354
+    },
+    {
+      "epoch": 0.1589700531766023,
+      "grad_norm": 0.7683622241020203,
+      "learning_rate": 4.744921514312096e-05,
+      "loss": 0.9604,
+      "step": 355
+    },
+    {
+      "epoch": 0.15941785614329695,
+      "grad_norm": 0.7971686720848083,
+      "learning_rate": 4.743767313019391e-05,
+      "loss": 0.3917,
+      "step": 356
+    },
+    {
+      "epoch": 0.1598656591099916,
+      "grad_norm": 0.7324389815330505,
+      "learning_rate": 4.7426131117266854e-05,
+      "loss": 0.9423,
+      "step": 357
+    },
+    {
+      "epoch": 0.16031346207668626,
+      "grad_norm": 0.8222924470901489,
+      "learning_rate": 4.7414589104339804e-05,
+      "loss": 0.7726,
+      "step": 358
+    },
+    {
+      "epoch": 0.16076126504338092,
+      "grad_norm": 0.7495877742767334,
+      "learning_rate": 4.7403047091412746e-05,
+      "loss": 0.7838,
+      "step": 359
+    },
+    {
+      "epoch": 0.16120906801007556,
+      "grad_norm": 0.7453298568725586,
+      "learning_rate": 4.7391505078485696e-05,
+      "loss": 1.0874,
+      "step": 360
+    },
+    {
+      "epoch": 0.16120906801007556,
+      "eval_loss": 0.8566550612449646,
+      "eval_runtime": 1737.1905,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 360
+    },
+    {
+      "epoch": 0.16165687097677023,
+      "grad_norm": 0.8878437280654907,
+      "learning_rate": 4.737996306555864e-05,
+      "loss": 0.7624,
+      "step": 361
+    },
+    {
+      "epoch": 0.16210467394346487,
+      "grad_norm": 0.771511971950531,
+      "learning_rate": 4.736842105263158e-05,
+      "loss": 0.7768,
+      "step": 362
+    },
+    {
+      "epoch": 0.16255247691015953,
+      "grad_norm": 0.7700636386871338,
+      "learning_rate": 4.735687903970453e-05,
+      "loss": 0.857,
+      "step": 363
+    },
+    {
+      "epoch": 0.16300027987685417,
+      "grad_norm": 0.8953984975814819,
+      "learning_rate": 4.734533702677747e-05,
+      "loss": 0.8731,
+      "step": 364
+    },
+    {
+      "epoch": 0.16344808284354884,
+      "grad_norm": 1.056182622909546,
+      "learning_rate": 4.7333795013850415e-05,
+      "loss": 0.7347,
+      "step": 365
+    },
+    {
+      "epoch": 0.1638958858102435,
+      "grad_norm": 0.8084350824356079,
+      "learning_rate": 4.732225300092336e-05,
+      "loss": 0.8404,
+      "step": 366
+    },
+    {
+      "epoch": 0.16434368877693814,
+      "grad_norm": 0.8122984766960144,
+      "learning_rate": 4.731071098799631e-05,
+      "loss": 0.7705,
+      "step": 367
+    },
+    {
+      "epoch": 0.1647914917436328,
+      "grad_norm": 0.810514509677887,
+      "learning_rate": 4.729916897506925e-05,
+      "loss": 0.8933,
+      "step": 368
+    },
+    {
+      "epoch": 0.16523929471032744,
+      "grad_norm": 0.8424896597862244,
+      "learning_rate": 4.72876269621422e-05,
+      "loss": 0.6952,
+      "step": 369
+    },
+    {
+      "epoch": 0.1656870976770221,
+      "grad_norm": 0.7785488963127136,
+      "learning_rate": 4.727608494921514e-05,
+      "loss": 0.8959,
+      "step": 370
+    },
+    {
+      "epoch": 0.1656870976770221,
+      "eval_loss": 0.8474584817886353,
+      "eval_runtime": 1737.3041,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 370
+    },
+    {
+      "epoch": 0.16613490064371678,
+      "grad_norm": 0.7459981441497803,
+      "learning_rate": 4.726454293628809e-05,
+      "loss": 0.8296,
+      "step": 371
+    },
+    {
+      "epoch": 0.16658270361041141,
+      "grad_norm": 0.8447751998901367,
+      "learning_rate": 4.7253000923361034e-05,
+      "loss": 0.7229,
+      "step": 372
+    },
+    {
+      "epoch": 0.16703050657710608,
+      "grad_norm": 0.8873215317726135,
+      "learning_rate": 4.7241458910433983e-05,
+      "loss": 0.9665,
+      "step": 373
+    },
+    {
+      "epoch": 0.16747830954380072,
+      "grad_norm": 0.8727468848228455,
+      "learning_rate": 4.7229916897506926e-05,
+      "loss": 0.8588,
+      "step": 374
+    },
+    {
+      "epoch": 0.16792611251049538,
+      "grad_norm": 0.8442516326904297,
+      "learning_rate": 4.7218374884579875e-05,
+      "loss": 0.8735,
+      "step": 375
+    },
+    {
+      "epoch": 0.16837391547719002,
+      "grad_norm": 0.741875410079956,
+      "learning_rate": 4.720683287165282e-05,
+      "loss": 0.7119,
+      "step": 376
+    },
+    {
+      "epoch": 0.1688217184438847,
+      "grad_norm": 0.9967590570449829,
+      "learning_rate": 4.719529085872577e-05,
+      "loss": 0.6724,
+      "step": 377
+    },
+    {
+      "epoch": 0.16926952141057935,
+      "grad_norm": 0.8461303114891052,
+      "learning_rate": 4.718374884579871e-05,
+      "loss": 0.8864,
+      "step": 378
+    },
+    {
+      "epoch": 0.169717324377274,
+      "grad_norm": 0.8811253309249878,
+      "learning_rate": 4.717220683287166e-05,
+      "loss": 0.8461,
+      "step": 379
+    },
+    {
+      "epoch": 0.17016512734396866,
+      "grad_norm": 0.7719868421554565,
+      "learning_rate": 4.71606648199446e-05,
+      "loss": 0.8755,
+      "step": 380
+    },
+    {
+      "epoch": 0.17016512734396866,
+      "eval_loss": 0.8378962874412537,
+      "eval_runtime": 1737.6913,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 380
+    },
+    {
+      "epoch": 0.1706129303106633,
+      "grad_norm": 0.904470682144165,
+      "learning_rate": 4.714912280701755e-05,
+      "loss": 0.869,
+      "step": 381
+    },
+    {
+      "epoch": 0.17106073327735796,
+      "grad_norm": 0.988050639629364,
+      "learning_rate": 4.7137580794090494e-05,
+      "loss": 0.8023,
+      "step": 382
+    },
+    {
+      "epoch": 0.17150853624405263,
+      "grad_norm": 0.7240595817565918,
+      "learning_rate": 4.7126038781163437e-05,
+      "loss": 0.6097,
+      "step": 383
+    },
+    {
+      "epoch": 0.17195633921074727,
+      "grad_norm": 0.811393141746521,
+      "learning_rate": 4.711449676823638e-05,
+      "loss": 0.7306,
+      "step": 384
+    },
+    {
+      "epoch": 0.17240414217744193,
+      "grad_norm": 0.867587149143219,
+      "learning_rate": 4.710295475530933e-05,
+      "loss": 0.7562,
+      "step": 385
+    },
+    {
+      "epoch": 0.17285194514413657,
+      "grad_norm": 0.842566728591919,
+      "learning_rate": 4.709141274238227e-05,
+      "loss": 0.5919,
+      "step": 386
+    },
+    {
+      "epoch": 0.17329974811083124,
+      "grad_norm": 0.9528062343597412,
+      "learning_rate": 4.7079870729455214e-05,
+      "loss": 0.8223,
+      "step": 387
+    },
+    {
+      "epoch": 0.1737475510775259,
+      "grad_norm": 0.7800689935684204,
+      "learning_rate": 4.706832871652816e-05,
+      "loss": 0.7363,
+      "step": 388
+    },
+    {
+      "epoch": 0.17419535404422054,
+      "grad_norm": 1.0290626287460327,
+      "learning_rate": 4.7056786703601106e-05,
+      "loss": 0.7701,
+      "step": 389
+    },
+    {
+      "epoch": 0.1746431570109152,
+      "grad_norm": 0.7647327780723572,
+      "learning_rate": 4.7045244690674055e-05,
+      "loss": 0.877,
+      "step": 390
+    },
+    {
+      "epoch": 0.1746431570109152,
+      "eval_loss": 0.8322415351867676,
+      "eval_runtime": 1737.8875,
+      "eval_samples_per_second": 2.57,
+      "eval_steps_per_second": 2.57,
+      "step": 390
+    },
+    {
+      "epoch": 0.17509095997760984,
+      "grad_norm": 1.024598240852356,
+      "learning_rate": 4.7033702677747e-05,
+      "loss": 0.9069,
+      "step": 391
+    },
+    {
+      "epoch": 0.1755387629443045,
+      "grad_norm": 0.8539007902145386,
+      "learning_rate": 4.702216066481995e-05,
+      "loss": 0.904,
+      "step": 392
+    },
+    {
+      "epoch": 0.17598656591099915,
+      "grad_norm": 0.7349446415901184,
+      "learning_rate": 4.701061865189289e-05,
+      "loss": 0.7279,
+      "step": 393
+    },
+    {
+      "epoch": 0.17643436887769381,
+      "grad_norm": 0.9322994947433472,
+      "learning_rate": 4.699907663896584e-05,
+      "loss": 1.0765,
+      "step": 394
+    },
+    {
+      "epoch": 0.17688217184438848,
+      "grad_norm": 0.7974610924720764,
+      "learning_rate": 4.698753462603878e-05,
+      "loss": 0.7591,
+      "step": 395
+    },
+    {
+      "epoch": 0.17732997481108312,
+      "grad_norm": 0.9433238506317139,
+      "learning_rate": 4.697599261311173e-05,
+      "loss": 0.7526,
+      "step": 396
+    },
+    {
+      "epoch": 0.17777777777777778,
+      "grad_norm": 0.8383938074111938,
+      "learning_rate": 4.6964450600184674e-05,
+      "loss": 0.725,
+      "step": 397
+    },
+    {
+      "epoch": 0.17822558074447242,
+      "grad_norm": 0.7757347226142883,
+      "learning_rate": 4.695290858725762e-05,
+      "loss": 1.063,
+      "step": 398
+    },
+    {
+      "epoch": 0.1786733837111671,
+      "grad_norm": 0.8449235558509827,
+      "learning_rate": 4.6941366574330566e-05,
+      "loss": 0.7043,
+      "step": 399
+    },
+    {
+      "epoch": 0.17912118667786175,
+      "grad_norm": 0.7594418525695801,
+      "learning_rate": 4.6929824561403515e-05,
+      "loss": 0.9083,
+      "step": 400
+    },
+    {
+      "epoch": 0.17912118667786175,
+      "eval_loss": 0.8224214911460876,
+      "eval_runtime": 1738.3524,
+      "eval_samples_per_second": 2.57,
+      "eval_steps_per_second": 2.57,
+      "step": 400
+    },
+    {
+      "epoch": 0.1795689896445564,
+      "grad_norm": 0.7965478301048279,
+      "learning_rate": 4.691828254847646e-05,
+      "loss": 0.9748,
+      "step": 401
+    },
+    {
+      "epoch": 0.18001679261125106,
+      "grad_norm": 0.8964141607284546,
+      "learning_rate": 4.690674053554941e-05,
+      "loss": 1.0344,
+      "step": 402
+    },
+    {
+      "epoch": 0.1804645955779457,
+      "grad_norm": 0.7425351142883301,
+      "learning_rate": 4.689519852262235e-05,
+      "loss": 0.6737,
+      "step": 403
+    },
+    {
+      "epoch": 0.18091239854464036,
+      "grad_norm": 0.7083563208580017,
+      "learning_rate": 4.688365650969529e-05,
+      "loss": 0.8339,
+      "step": 404
+    },
+    {
+      "epoch": 0.181360201511335,
+      "grad_norm": 0.7600324153900146,
+      "learning_rate": 4.6872114496768235e-05,
+      "loss": 0.8032,
+      "step": 405
+    },
+    {
+      "epoch": 0.18180800447802967,
+      "grad_norm": 0.7678210735321045,
+      "learning_rate": 4.6860572483841184e-05,
+      "loss": 0.8703,
+      "step": 406
+    },
+    {
+      "epoch": 0.18225580744472433,
+      "grad_norm": 0.8270692229270935,
+      "learning_rate": 4.684903047091413e-05,
+      "loss": 0.8807,
+      "step": 407
+    },
+    {
+      "epoch": 0.18270361041141897,
+      "grad_norm": 0.8562888503074646,
+      "learning_rate": 4.6837488457987076e-05,
+      "loss": 0.9365,
+      "step": 408
+    },
+    {
+      "epoch": 0.18315141337811364,
+      "grad_norm": 0.9993149042129517,
+      "learning_rate": 4.682594644506002e-05,
+      "loss": 0.8716,
+      "step": 409
+    },
+    {
+      "epoch": 0.18359921634480827,
+      "grad_norm": 0.795988917350769,
+      "learning_rate": 4.681440443213297e-05,
+      "loss": 0.7853,
+      "step": 410
+    },
+    {
+      "epoch": 0.18359921634480827,
+      "eval_loss": 0.8142237663269043,
+      "eval_runtime": 1737.637,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 410
+    },
+    {
+      "epoch": 0.18404701931150294,
+      "grad_norm": 0.7437682151794434,
+      "learning_rate": 4.680286241920591e-05,
+      "loss": 0.7872,
+      "step": 411
+    },
+    {
+      "epoch": 0.1844948222781976,
+      "grad_norm": 0.8209964036941528,
+      "learning_rate": 4.6791320406278854e-05,
+      "loss": 0.7995,
+      "step": 412
+    },
+    {
+      "epoch": 0.18494262524489224,
+      "grad_norm": 0.9489561319351196,
+      "learning_rate": 4.67797783933518e-05,
+      "loss": 0.7761,
+      "step": 413
+    },
+    {
+      "epoch": 0.1853904282115869,
+      "grad_norm": 0.8502386212348938,
+      "learning_rate": 4.6768236380424746e-05,
+      "loss": 0.9846,
+      "step": 414
+    },
+    {
+      "epoch": 0.18583823117828155,
+      "grad_norm": 0.7570764422416687,
+      "learning_rate": 4.6756694367497695e-05,
+      "loss": 1.116,
+      "step": 415
+    },
+    {
+      "epoch": 0.18628603414497621,
+      "grad_norm": 0.8194412589073181,
+      "learning_rate": 4.674515235457064e-05,
+      "loss": 0.8261,
+      "step": 416
+    },
+    {
+      "epoch": 0.18673383711167085,
+      "grad_norm": 1.0201462507247925,
+      "learning_rate": 4.673361034164359e-05,
+      "loss": 1.0355,
+      "step": 417
+    },
+    {
+      "epoch": 0.18718164007836552,
+      "grad_norm": 0.7511212229728699,
+      "learning_rate": 4.672206832871653e-05,
+      "loss": 0.921,
+      "step": 418
+    },
+    {
+      "epoch": 0.18762944304506018,
+      "grad_norm": 0.8299571871757507,
+      "learning_rate": 4.671052631578948e-05,
+      "loss": 1.0033,
+      "step": 419
+    },
+    {
+      "epoch": 0.18807724601175482,
+      "grad_norm": 0.9280295968055725,
+      "learning_rate": 4.669898430286242e-05,
+      "loss": 0.7556,
+      "step": 420
+    },
+    {
+      "epoch": 0.18807724601175482,
+      "eval_loss": 0.8086807727813721,
+      "eval_runtime": 1738.0909,
+      "eval_samples_per_second": 2.57,
+      "eval_steps_per_second": 2.57,
+      "step": 420
+    },
+    {
+      "epoch": 0.1885250489784495,
+      "grad_norm": 0.8704283237457275,
+      "learning_rate": 4.668744228993537e-05,
+      "loss": 0.9123,
+      "step": 421
+    },
+    {
+      "epoch": 0.18897285194514413,
+      "grad_norm": 1.0498746633529663,
+      "learning_rate": 4.6675900277008313e-05,
+      "loss": 0.6306,
+      "step": 422
+    },
+    {
+      "epoch": 0.1894206549118388,
+      "grad_norm": 0.7668930888175964,
+      "learning_rate": 4.6664358264081256e-05,
+      "loss": 1.1142,
+      "step": 423
+    },
+    {
+      "epoch": 0.18986845787853346,
+      "grad_norm": 1.0189166069030762,
+      "learning_rate": 4.6652816251154205e-05,
+      "loss": 0.8831,
+      "step": 424
+    },
+    {
+      "epoch": 0.1903162608452281,
+      "grad_norm": 0.9827046990394592,
+      "learning_rate": 4.664127423822715e-05,
+      "loss": 0.6772,
+      "step": 425
+    },
+    {
+      "epoch": 0.19076406381192276,
+      "grad_norm": 0.8419145941734314,
+      "learning_rate": 4.662973222530009e-05,
+      "loss": 0.8903,
+      "step": 426
+    },
+    {
+      "epoch": 0.1912118667786174,
+      "grad_norm": 1.0089313983917236,
+      "learning_rate": 4.661819021237304e-05,
+      "loss": 0.8986,
+      "step": 427
+    },
+    {
+      "epoch": 0.19165966974531207,
+      "grad_norm": 0.9125351309776306,
+      "learning_rate": 4.660664819944598e-05,
+      "loss": 0.8019,
+      "step": 428
+    },
+    {
+      "epoch": 0.1921074727120067,
+      "grad_norm": 0.9646217823028564,
+      "learning_rate": 4.659510618651893e-05,
+      "loss": 0.6646,
+      "step": 429
+    },
+    {
+      "epoch": 0.19255527567870137,
+      "grad_norm": 1.0330066680908203,
+      "learning_rate": 4.6583564173591875e-05,
+      "loss": 0.9278,
+      "step": 430
+    },
+    {
+      "epoch": 0.19255527567870137,
+      "eval_loss": 0.7997502684593201,
+      "eval_runtime": 1738.9851,
+      "eval_samples_per_second": 2.569,
+      "eval_steps_per_second": 2.569,
+      "step": 430
+    },
+    {
+      "epoch": 0.19300307864539604,
+      "grad_norm": 0.7883182168006897,
+      "learning_rate": 4.6572022160664824e-05,
+      "loss": 0.7367,
+      "step": 431
+    },
+    {
+      "epoch": 0.19345088161209067,
+      "grad_norm": 0.8542357087135315,
+      "learning_rate": 4.656048014773777e-05,
+      "loss": 0.8034,
+      "step": 432
+    },
+    {
+      "epoch": 0.19389868457878534,
+      "grad_norm": 0.8576323390007019,
+      "learning_rate": 4.6548938134810716e-05,
+      "loss": 1.0407,
+      "step": 433
+    },
+    {
+      "epoch": 0.19434648754547998,
+      "grad_norm": 0.8395902514457703,
+      "learning_rate": 4.653739612188366e-05,
+      "loss": 0.8512,
+      "step": 434
+    },
+    {
+      "epoch": 0.19479429051217464,
+      "grad_norm": 0.8623055815696716,
+      "learning_rate": 4.652585410895661e-05,
+      "loss": 0.5575,
+      "step": 435
+    },
+    {
+      "epoch": 0.1952420934788693,
+      "grad_norm": 0.711990237236023,
+      "learning_rate": 4.651431209602955e-05,
+      "loss": 0.9274,
+      "step": 436
+    },
+    {
+      "epoch": 0.19568989644556395,
+      "grad_norm": 0.9932701587677002,
+      "learning_rate": 4.650277008310249e-05,
+      "loss": 0.7234,
+      "step": 437
+    },
+    {
+      "epoch": 0.19613769941225861,
+      "grad_norm": 0.9657753705978394,
+      "learning_rate": 4.649122807017544e-05,
+      "loss": 0.8719,
+      "step": 438
+    },
+    {
+      "epoch": 0.19658550237895325,
+      "grad_norm": 0.8279201984405518,
+      "learning_rate": 4.6479686057248385e-05,
+      "loss": 0.7891,
+      "step": 439
+    },
+    {
+      "epoch": 0.19703330534564792,
+      "grad_norm": 0.9909638166427612,
+      "learning_rate": 4.6468144044321335e-05,
+      "loss": 0.6438,
+      "step": 440
+    },
+    {
+      "epoch": 0.19703330534564792,
+      "eval_loss": 0.7911050915718079,
+      "eval_runtime": 1739.4085,
+      "eval_samples_per_second": 2.568,
+      "eval_steps_per_second": 2.568,
+      "step": 440
+    },
+    {
+      "epoch": 0.19748110831234256,
+      "grad_norm": 0.7535269260406494,
+      "learning_rate": 4.645660203139428e-05,
+      "loss": 0.834,
+      "step": 441
+    },
+    {
+      "epoch": 0.19792891127903722,
+      "grad_norm": 0.8972837328910828,
+      "learning_rate": 4.6445060018467227e-05,
+      "loss": 0.7625,
+      "step": 442
+    },
+    {
+      "epoch": 0.1983767142457319,
+      "grad_norm": 0.8356785774230957,
+      "learning_rate": 4.643351800554017e-05,
+      "loss": 0.8538,
+      "step": 443
+    },
+    {
+      "epoch": 0.19882451721242653,
+      "grad_norm": 0.8921677470207214,
+      "learning_rate": 4.642197599261311e-05,
+      "loss": 0.9137,
+      "step": 444
+    },
+    {
+      "epoch": 0.1992723201791212,
+      "grad_norm": 0.8165355324745178,
+      "learning_rate": 4.6410433979686054e-05,
+      "loss": 0.8285,
+      "step": 445
+    },
+    {
+      "epoch": 0.19972012314581583,
+      "grad_norm": 0.8382999897003174,
+      "learning_rate": 4.6398891966759004e-05,
+      "loss": 1.0426,
+      "step": 446
+    },
+    {
+      "epoch": 0.2001679261125105,
+      "grad_norm": 0.7926031351089478,
+      "learning_rate": 4.6387349953831946e-05,
+      "loss": 0.841,
+      "step": 447
+    },
+    {
+      "epoch": 0.20061572907920516,
+      "grad_norm": 0.8137739300727844,
+      "learning_rate": 4.6375807940904896e-05,
+      "loss": 0.8253,
+      "step": 448
+    },
+    {
+      "epoch": 0.2010635320458998,
+      "grad_norm": 0.8974906802177429,
+      "learning_rate": 4.636426592797784e-05,
+      "loss": 0.7984,
+      "step": 449
+    },
+    {
+      "epoch": 0.20151133501259447,
+      "grad_norm": 0.7735929489135742,
+      "learning_rate": 4.635272391505079e-05,
+      "loss": 0.6582,
+      "step": 450
+    },
+    {
+      "epoch": 0.20151133501259447,
+      "eval_loss": 0.7823182940483093,
+      "eval_runtime": 1739.8443,
+      "eval_samples_per_second": 2.567,
+      "eval_steps_per_second": 2.567,
+      "step": 450
+    },
+    {
+      "epoch": 0.2019591379792891,
+      "grad_norm": 0.7329652905464172,
+      "learning_rate": 4.634118190212373e-05,
+      "loss": 0.8273,
+      "step": 451
+    },
+    {
+      "epoch": 0.20240694094598377,
+      "grad_norm": 0.7918416261672974,
+      "learning_rate": 4.632963988919668e-05,
+      "loss": 0.8176,
+      "step": 452
+    },
+    {
+      "epoch": 0.2028547439126784,
+      "grad_norm": 0.9216796159744263,
+      "learning_rate": 4.631809787626962e-05,
+      "loss": 0.7303,
+      "step": 453
+    },
+    {
+      "epoch": 0.20330254687937308,
+      "grad_norm": 1.078320026397705,
+      "learning_rate": 4.630655586334257e-05,
+      "loss": 0.5538,
+      "step": 454
+    },
+    {
+      "epoch": 0.20375034984606774,
+      "grad_norm": 0.8843169808387756,
+      "learning_rate": 4.6295013850415514e-05,
+      "loss": 0.7284,
+      "step": 455
+    },
+    {
+      "epoch": 0.20419815281276238,
+      "grad_norm": 1.031596302986145,
+      "learning_rate": 4.6283471837488464e-05,
+      "loss": 0.8675,
+      "step": 456
+    },
+    {
+      "epoch": 0.20464595577945704,
+      "grad_norm": 1.0795527696609497,
+      "learning_rate": 4.6271929824561406e-05,
+      "loss": 0.8009,
+      "step": 457
+    },
+    {
+      "epoch": 0.20509375874615168,
+      "grad_norm": 0.9157360196113586,
+      "learning_rate": 4.6260387811634356e-05,
+      "loss": 0.6112,
+      "step": 458
+    },
+    {
+      "epoch": 0.20554156171284635,
+      "grad_norm": 0.8338711857795715,
+      "learning_rate": 4.62488457987073e-05,
+      "loss": 0.8584,
+      "step": 459
+    },
+    {
+      "epoch": 0.20598936467954101,
+      "grad_norm": 0.70074063539505,
+      "learning_rate": 4.623730378578025e-05,
+      "loss": 0.882,
+      "step": 460
+    },
+    {
+      "epoch": 0.20598936467954101,
+      "eval_loss": 0.7754072546958923,
+      "eval_runtime": 1740.1151,
+      "eval_samples_per_second": 2.567,
+      "eval_steps_per_second": 2.567,
+      "step": 460
+    },
+    {
+      "epoch": 0.20643716764623565,
+      "grad_norm": 0.8252310156822205,
+      "learning_rate": 4.622576177285319e-05,
+      "loss": 0.8817,
+      "step": 461
+    },
+    {
+      "epoch": 0.20688497061293032,
+      "grad_norm": 0.9494065046310425,
+      "learning_rate": 4.621421975992613e-05,
+      "loss": 0.8243,
+      "step": 462
+    },
+    {
+      "epoch": 0.20733277357962496,
+      "grad_norm": 0.74778151512146,
+      "learning_rate": 4.620267774699908e-05,
+      "loss": 1.1506,
+      "step": 463
+    },
+    {
+      "epoch": 0.20778057654631962,
+      "grad_norm": 0.8175053596496582,
+      "learning_rate": 4.6191135734072025e-05,
+      "loss": 0.7699,
+      "step": 464
+    },
+    {
+      "epoch": 0.20822837951301426,
+      "grad_norm": 0.7756770849227905,
+      "learning_rate": 4.617959372114497e-05,
+      "loss": 0.8569,
+      "step": 465
+    },
+    {
+      "epoch": 0.20867618247970893,
+      "grad_norm": 0.8187942504882812,
+      "learning_rate": 4.616805170821791e-05,
+      "loss": 0.8084,
+      "step": 466
+    },
+    {
+      "epoch": 0.2091239854464036,
+      "grad_norm": 0.7939863204956055,
+      "learning_rate": 4.615650969529086e-05,
+      "loss": 0.5601,
+      "step": 467
+    },
+    {
+      "epoch": 0.20957178841309823,
+      "grad_norm": 0.8753362894058228,
+      "learning_rate": 4.61449676823638e-05,
+      "loss": 0.799,
+      "step": 468
+    },
+    {
+      "epoch": 0.2100195913797929,
+      "grad_norm": 0.7502310276031494,
+      "learning_rate": 4.613342566943675e-05,
+      "loss": 0.78,
+      "step": 469
+    },
+    {
+      "epoch": 0.21046739434648754,
+      "grad_norm": 0.933039665222168,
+      "learning_rate": 4.6121883656509694e-05,
+      "loss": 0.5448,
+      "step": 470
+    },
+    {
+      "epoch": 0.21046739434648754,
+      "eval_loss": 0.7699535489082336,
+      "eval_runtime": 1739.9004,
+      "eval_samples_per_second": 2.567,
+      "eval_steps_per_second": 2.567,
+      "step": 470
+    },
+    {
+      "epoch": 0.2109151973131822,
+      "grad_norm": 1.137333631515503,
+      "learning_rate": 4.6110341643582644e-05,
+      "loss": 0.7728,
+      "step": 471
+    },
+    {
+      "epoch": 0.21136300027987687,
+      "grad_norm": 0.9676854610443115,
+      "learning_rate": 4.6098799630655586e-05,
+      "loss": 0.8342,
+      "step": 472
+    },
+    {
+      "epoch": 0.2118108032465715,
+      "grad_norm": 0.9194774627685547,
+      "learning_rate": 4.6087257617728535e-05,
+      "loss": 0.8051,
+      "step": 473
+    },
+    {
+      "epoch": 0.21225860621326617,
+      "grad_norm": 0.9370580315589905,
+      "learning_rate": 4.607571560480148e-05,
+      "loss": 0.7046,
+      "step": 474
+    },
+    {
+      "epoch": 0.2127064091799608,
+      "grad_norm": 0.8880735635757446,
+      "learning_rate": 4.606417359187443e-05,
+      "loss": 0.743,
+      "step": 475
+    },
+    {
+      "epoch": 0.21315421214665548,
+      "grad_norm": 0.8558589220046997,
+      "learning_rate": 4.605263157894737e-05,
+      "loss": 0.9362,
+      "step": 476
+    },
+    {
+      "epoch": 0.2136020151133501,
+      "grad_norm": 0.8161864876747131,
+      "learning_rate": 4.604108956602032e-05,
+      "loss": 0.7904,
+      "step": 477
+    },
+    {
+      "epoch": 0.21404981808004478,
+      "grad_norm": 0.9424650073051453,
+      "learning_rate": 4.602954755309326e-05,
+      "loss": 0.7731,
+      "step": 478
+    },
+    {
+      "epoch": 0.21449762104673945,
+      "grad_norm": 0.8773455023765564,
+      "learning_rate": 4.601800554016621e-05,
+      "loss": 0.7889,
+      "step": 479
+    },
+    {
+      "epoch": 0.21494542401343408,
+      "grad_norm": 0.8586567640304565,
+      "learning_rate": 4.6006463527239154e-05,
+      "loss": 0.9641,
+      "step": 480
+    },
+    {
+      "epoch": 0.21494542401343408,
+      "eval_loss": 0.7605552077293396,
+      "eval_runtime": 1741.1131,
+      "eval_samples_per_second": 2.566,
+      "eval_steps_per_second": 2.566,
+      "step": 480
+    },
+    {
+      "epoch": 0.21539322698012875,
+      "grad_norm": 0.768221378326416,
+      "learning_rate": 4.5994921514312103e-05,
+      "loss": 0.9106,
+      "step": 481
+    },
+    {
+      "epoch": 0.2158410299468234,
+      "grad_norm": 1.0304666757583618,
+      "learning_rate": 4.5983379501385046e-05,
+      "loss": 0.6707,
+      "step": 482
+    },
+    {
+      "epoch": 0.21628883291351805,
+      "grad_norm": 0.8881514072418213,
+      "learning_rate": 4.597183748845799e-05,
+      "loss": 0.8598,
+      "step": 483
+    },
+    {
+      "epoch": 0.21673663588021272,
+      "grad_norm": 0.907448410987854,
+      "learning_rate": 4.596029547553093e-05,
+      "loss": 0.7609,
+      "step": 484
+    },
+    {
+      "epoch": 0.21718443884690736,
+      "grad_norm": 0.9400318264961243,
+      "learning_rate": 4.594875346260388e-05,
+      "loss": 0.8799,
+      "step": 485
+    },
+    {
+      "epoch": 0.21763224181360202,
+      "grad_norm": 1.0344361066818237,
+      "learning_rate": 4.593721144967682e-05,
+      "loss": 0.8069,
+      "step": 486
+    },
+    {
+      "epoch": 0.21808004478029666,
+      "grad_norm": 0.8733990788459778,
+      "learning_rate": 4.5925669436749766e-05,
+      "loss": 0.668,
+      "step": 487
+    },
+    {
+      "epoch": 0.21852784774699133,
+      "grad_norm": 0.7444555759429932,
+      "learning_rate": 4.5914127423822715e-05,
+      "loss": 0.6643,
+      "step": 488
+    },
+    {
+      "epoch": 0.21897565071368597,
+      "grad_norm": 0.8918318152427673,
+      "learning_rate": 4.590258541089566e-05,
+      "loss": 0.656,
+      "step": 489
+    },
+    {
+      "epoch": 0.21942345368038063,
+      "grad_norm": 0.9341442584991455,
+      "learning_rate": 4.589104339796861e-05,
+      "loss": 0.7254,
+      "step": 490
+    },
+    {
+      "epoch": 0.21942345368038063,
+      "eval_loss": 0.7527384757995605,
+      "eval_runtime": 1741.6165,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 2.565,
+      "step": 490
+    },
+    {
+      "epoch": 0.2198712566470753,
+      "grad_norm": 0.9269483089447021,
+      "learning_rate": 4.587950138504155e-05,
+      "loss": 0.6856,
+      "step": 491
+    },
+    {
+      "epoch": 0.22031905961376994,
+      "grad_norm": 0.9737778306007385,
+      "learning_rate": 4.58679593721145e-05,
+      "loss": 0.6296,
+      "step": 492
+    },
+    {
+      "epoch": 0.2207668625804646,
+      "grad_norm": 1.1320279836654663,
+      "learning_rate": 4.585641735918744e-05,
+      "loss": 0.9568,
+      "step": 493
+    },
+    {
+      "epoch": 0.22121466554715924,
+      "grad_norm": 0.9716588854789734,
+      "learning_rate": 4.584487534626039e-05,
+      "loss": 0.9283,
+      "step": 494
+    },
+    {
+      "epoch": 0.2216624685138539,
+      "grad_norm": 0.9201183915138245,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.5564,
+      "step": 495
+    },
+    {
+      "epoch": 0.22211027148054857,
+      "grad_norm": 0.956973671913147,
+      "learning_rate": 4.582179132040628e-05,
+      "loss": 0.7652,
+      "step": 496
+    },
+    {
+      "epoch": 0.2225580744472432,
+      "grad_norm": 0.9087381958961487,
+      "learning_rate": 4.5810249307479226e-05,
+      "loss": 0.5507,
+      "step": 497
+    },
+    {
+      "epoch": 0.22300587741393788,
+      "grad_norm": 0.8070569038391113,
+      "learning_rate": 4.5798707294552175e-05,
+      "loss": 0.7761,
+      "step": 498
+    },
+    {
+      "epoch": 0.2234536803806325,
+      "grad_norm": 0.9219637513160706,
+      "learning_rate": 4.578716528162512e-05,
+      "loss": 0.8153,
+      "step": 499
+    },
+    {
+      "epoch": 0.22390148334732718,
+      "grad_norm": 0.921072781085968,
+      "learning_rate": 4.577562326869807e-05,
+      "loss": 0.616,
+      "step": 500
+    },
+    {
+      "epoch": 0.22390148334732718,
+      "eval_loss": 0.7446861863136292,
+      "eval_runtime": 1731.4556,
+      "eval_samples_per_second": 2.58,
+      "eval_steps_per_second": 2.58,
+      "step": 500
+    },
+    {
+      "epoch": 0.22434928631402182,
+      "grad_norm": 0.8725545406341553,
+      "learning_rate": 4.576408125577101e-05,
+      "loss": 0.6828,
+      "step": 501
+    },
+    {
+      "epoch": 0.22479708928071648,
+      "grad_norm": 1.0268959999084473,
+      "learning_rate": 4.575253924284396e-05,
+      "loss": 0.6286,
+      "step": 502
+    },
+    {
+      "epoch": 0.22524489224741115,
+      "grad_norm": 0.8673756718635559,
+      "learning_rate": 4.57409972299169e-05,
+      "loss": 0.7195,
+      "step": 503
+    },
+    {
+      "epoch": 0.2256926952141058,
+      "grad_norm": 0.7618908286094666,
+      "learning_rate": 4.5729455216989844e-05,
+      "loss": 0.8099,
+      "step": 504
+    },
+    {
+      "epoch": 0.22614049818080045,
+      "grad_norm": 0.9637864828109741,
+      "learning_rate": 4.571791320406279e-05,
+      "loss": 0.5663,
+      "step": 505
+    },
+    {
+      "epoch": 0.2265883011474951,
+      "grad_norm": 0.8402149081230164,
+      "learning_rate": 4.5706371191135736e-05,
+      "loss": 0.8883,
+      "step": 506
+    },
+    {
+      "epoch": 0.22703610411418976,
+      "grad_norm": 1.0460970401763916,
+      "learning_rate": 4.569482917820868e-05,
+      "loss": 0.7933,
+      "step": 507
+    },
+    {
+      "epoch": 0.22748390708088442,
+      "grad_norm": 1.0066616535186768,
+      "learning_rate": 4.568328716528163e-05,
+      "loss": 0.6868,
+      "step": 508
+    },
+    {
+      "epoch": 0.22793171004757906,
+      "grad_norm": 0.933358371257782,
+      "learning_rate": 4.567174515235457e-05,
+      "loss": 0.6533,
+      "step": 509
+    },
+    {
+      "epoch": 0.22837951301427373,
+      "grad_norm": 0.9774124622344971,
+      "learning_rate": 4.566020313942752e-05,
+      "loss": 0.5852,
+      "step": 510
+    },
+    {
+      "epoch": 0.22837951301427373,
+      "eval_loss": 0.73744797706604,
+      "eval_runtime": 1735.3738,
+      "eval_samples_per_second": 2.574,
+      "eval_steps_per_second": 2.574,
+      "step": 510
+    },
+    {
+      "epoch": 0.22882731598096837,
+      "grad_norm": 0.8680145740509033,
+      "learning_rate": 4.564866112650046e-05,
+      "loss": 0.8256,
+      "step": 511
+    },
+    {
+      "epoch": 0.22927511894766303,
+      "grad_norm": 0.9629591107368469,
+      "learning_rate": 4.5637119113573406e-05,
+      "loss": 0.786,
+      "step": 512
+    },
+    {
+      "epoch": 0.22972292191435767,
+      "grad_norm": 0.8956183791160583,
+      "learning_rate": 4.5625577100646355e-05,
+      "loss": 0.6659,
+      "step": 513
+    },
+    {
+      "epoch": 0.23017072488105234,
+      "grad_norm": 0.9961169958114624,
+      "learning_rate": 4.56140350877193e-05,
+      "loss": 0.8279,
+      "step": 514
+    },
+    {
+      "epoch": 0.230618527847747,
+      "grad_norm": 0.9245657920837402,
+      "learning_rate": 4.560249307479225e-05,
+      "loss": 0.5056,
+      "step": 515
+    },
+    {
+      "epoch": 0.23106633081444164,
+      "grad_norm": 1.0266315937042236,
+      "learning_rate": 4.559095106186519e-05,
+      "loss": 0.839,
+      "step": 516
+    },
+    {
+      "epoch": 0.2315141337811363,
+      "grad_norm": 1.0880862474441528,
+      "learning_rate": 4.557940904893814e-05,
+      "loss": 0.7539,
+      "step": 517
+    },
+    {
+      "epoch": 0.23196193674783094,
+      "grad_norm": 0.894558846950531,
+      "learning_rate": 4.556786703601108e-05,
+      "loss": 0.7628,
+      "step": 518
+    },
+    {
+      "epoch": 0.2324097397145256,
+      "grad_norm": 0.9117352366447449,
+      "learning_rate": 4.555632502308403e-05,
+      "loss": 0.6321,
+      "step": 519
+    },
+    {
+      "epoch": 0.23285754268122028,
+      "grad_norm": 0.7979990839958191,
+      "learning_rate": 4.5544783010156974e-05,
+      "loss": 0.9759,
+      "step": 520
+    },
+    {
+      "epoch": 0.23285754268122028,
+      "eval_loss": 0.731434166431427,
+      "eval_runtime": 1732.8361,
+      "eval_samples_per_second": 2.578,
+      "eval_steps_per_second": 2.578,
+      "step": 520
+    },
+    {
+      "epoch": 0.2333053456479149,
+      "grad_norm": 1.0542393922805786,
+      "learning_rate": 4.553324099722992e-05,
+      "loss": 0.4714,
+      "step": 521
+    },
+    {
+      "epoch": 0.23375314861460958,
+      "grad_norm": 1.0756099224090576,
+      "learning_rate": 4.5521698984302866e-05,
+      "loss": 0.6873,
+      "step": 522
+    },
+    {
+      "epoch": 0.23420095158130422,
+      "grad_norm": 1.0123642683029175,
+      "learning_rate": 4.551015697137581e-05,
+      "loss": 0.6545,
+      "step": 523
+    },
+    {
+      "epoch": 0.23464875454799888,
+      "grad_norm": 1.0468355417251587,
+      "learning_rate": 4.549861495844876e-05,
+      "loss": 0.7371,
+      "step": 524
+    },
+    {
+      "epoch": 0.23509655751469352,
+      "grad_norm": 0.91986483335495,
+      "learning_rate": 4.54870729455217e-05,
+      "loss": 0.6783,
+      "step": 525
+    },
+    {
+      "epoch": 0.2355443604813882,
+      "grad_norm": 0.8713955283164978,
+      "learning_rate": 4.547553093259464e-05,
+      "loss": 0.7152,
+      "step": 526
+    },
+    {
+      "epoch": 0.23599216344808285,
+      "grad_norm": 1.010161280632019,
+      "learning_rate": 4.546398891966759e-05,
+      "loss": 0.7729,
+      "step": 527
+    },
+    {
+      "epoch": 0.2364399664147775,
+      "grad_norm": 0.9573583602905273,
+      "learning_rate": 4.5452446906740535e-05,
+      "loss": 0.6187,
+      "step": 528
+    },
+    {
+      "epoch": 0.23688776938147216,
+      "grad_norm": 0.9014054536819458,
+      "learning_rate": 4.5440904893813484e-05,
+      "loss": 0.6841,
+      "step": 529
+    },
+    {
+      "epoch": 0.2373355723481668,
+      "grad_norm": 0.7979456782341003,
+      "learning_rate": 4.542936288088643e-05,
+      "loss": 0.6041,
+      "step": 530
+    },
+    {
+      "epoch": 0.2373355723481668,
+      "eval_loss": 0.7225666046142578,
+      "eval_runtime": 1730.4142,
+      "eval_samples_per_second": 2.581,
+      "eval_steps_per_second": 2.581,
+      "step": 530
+    },
+    {
+      "epoch": 0.23778337531486146,
+      "grad_norm": 0.9193875193595886,
+      "learning_rate": 4.5417820867959376e-05,
+      "loss": 0.5635,
+      "step": 531
+    },
+    {
+      "epoch": 0.23823117828155613,
+      "grad_norm": 0.9741266369819641,
+      "learning_rate": 4.540627885503232e-05,
+      "loss": 0.6157,
+      "step": 532
+    },
+    {
+      "epoch": 0.23867898124825077,
+      "grad_norm": 0.9680760502815247,
+      "learning_rate": 4.539473684210527e-05,
+      "loss": 0.6919,
+      "step": 533
+    },
+    {
+      "epoch": 0.23912678421494543,
+      "grad_norm": 0.9001390337944031,
+      "learning_rate": 4.538319482917821e-05,
+      "loss": 0.6982,
+      "step": 534
+    },
+    {
+      "epoch": 0.23957458718164007,
+      "grad_norm": 0.921018660068512,
+      "learning_rate": 4.537165281625116e-05,
+      "loss": 0.5572,
+      "step": 535
+    },
+    {
+      "epoch": 0.24002239014833474,
+      "grad_norm": 0.833634078502655,
+      "learning_rate": 4.53601108033241e-05,
+      "loss": 0.8303,
+      "step": 536
+    },
+    {
+      "epoch": 0.24047019311502937,
+      "grad_norm": 1.0180853605270386,
+      "learning_rate": 4.5348568790397045e-05,
+      "loss": 0.6601,
+      "step": 537
+    },
+    {
+      "epoch": 0.24091799608172404,
+      "grad_norm": 0.8896169066429138,
+      "learning_rate": 4.5337026777469995e-05,
+      "loss": 0.7986,
+      "step": 538
+    },
+    {
+      "epoch": 0.2413657990484187,
+      "grad_norm": 1.0047816038131714,
+      "learning_rate": 4.532548476454294e-05,
+      "loss": 0.71,
+      "step": 539
+    },
+    {
+      "epoch": 0.24181360201511334,
+      "grad_norm": 0.9067107439041138,
+      "learning_rate": 4.531394275161589e-05,
+      "loss": 0.9264,
+      "step": 540
+    },
+    {
+      "epoch": 0.24181360201511334,
+      "eval_loss": 0.7173718214035034,
+      "eval_runtime": 1732.183,
+      "eval_samples_per_second": 2.579,
+      "eval_steps_per_second": 2.579,
+      "step": 540
+    },
+    {
+      "epoch": 0.242261404981808,
+      "grad_norm": 0.9225302338600159,
+      "learning_rate": 4.530240073868883e-05,
+      "loss": 0.7265,
+      "step": 541
+    },
+    {
+      "epoch": 0.24270920794850265,
+      "grad_norm": 1.0222615003585815,
+      "learning_rate": 4.529085872576178e-05,
+      "loss": 0.6444,
+      "step": 542
+    },
+    {
+      "epoch": 0.2431570109151973,
+      "grad_norm": 0.8889837861061096,
+      "learning_rate": 4.527931671283472e-05,
+      "loss": 0.6853,
+      "step": 543
+    },
+    {
+      "epoch": 0.24360481388189198,
+      "grad_norm": 0.9772764444351196,
+      "learning_rate": 4.5267774699907664e-05,
+      "loss": 0.5919,
+      "step": 544
+    },
+    {
+      "epoch": 0.24405261684858662,
+      "grad_norm": 0.794398307800293,
+      "learning_rate": 4.5256232686980606e-05,
+      "loss": 0.9762,
+      "step": 545
+    },
+    {
+      "epoch": 0.24450041981528128,
+      "grad_norm": 0.8529706597328186,
+      "learning_rate": 4.5244690674053556e-05,
+      "loss": 1.0409,
+      "step": 546
+    },
+    {
+      "epoch": 0.24494822278197592,
+      "grad_norm": 0.9057673811912537,
+      "learning_rate": 4.52331486611265e-05,
+      "loss": 0.6886,
+      "step": 547
+    },
+    {
+      "epoch": 0.2453960257486706,
+      "grad_norm": 0.9492160677909851,
+      "learning_rate": 4.522160664819945e-05,
+      "loss": 0.8578,
+      "step": 548
+    },
+    {
+      "epoch": 0.24584382871536523,
+      "grad_norm": 1.1789013147354126,
+      "learning_rate": 4.521006463527239e-05,
+      "loss": 0.7631,
+      "step": 549
+    },
+    {
+      "epoch": 0.2462916316820599,
+      "grad_norm": 1.0071775913238525,
+      "learning_rate": 4.519852262234534e-05,
+      "loss": 0.6618,
+      "step": 550
+    },
+    {
+      "epoch": 0.2462916316820599,
+      "eval_loss": 0.7097620368003845,
+      "eval_runtime": 1735.094,
+      "eval_samples_per_second": 2.575,
+      "eval_steps_per_second": 2.575,
+      "step": 550
+    },
+    {
+      "epoch": 0.24673943464875456,
+      "grad_norm": 0.8603365421295166,
+      "learning_rate": 4.518698060941828e-05,
+      "loss": 0.5254,
+      "step": 551
+    },
+    {
+      "epoch": 0.2471872376154492,
+      "grad_norm": 1.0445563793182373,
+      "learning_rate": 4.517543859649123e-05,
+      "loss": 0.7521,
+      "step": 552
+    },
+    {
+      "epoch": 0.24763504058214386,
+      "grad_norm": 0.8923339247703552,
+      "learning_rate": 4.5163896583564174e-05,
+      "loss": 0.5268,
+      "step": 553
+    },
+    {
+      "epoch": 0.2480828435488385,
+      "grad_norm": 0.9316530227661133,
+      "learning_rate": 4.5152354570637124e-05,
+      "loss": 0.5405,
+      "step": 554
+    },
+    {
+      "epoch": 0.24853064651553317,
+      "grad_norm": 0.9488258361816406,
+      "learning_rate": 4.5140812557710066e-05,
+      "loss": 0.6399,
+      "step": 555
+    },
+    {
+      "epoch": 0.24897844948222783,
+      "grad_norm": 0.9134349226951599,
+      "learning_rate": 4.5129270544783016e-05,
+      "loss": 0.7562,
+      "step": 556
+    },
+    {
+      "epoch": 0.24942625244892247,
+      "grad_norm": 0.8964117169380188,
+      "learning_rate": 4.511772853185596e-05,
+      "loss": 0.6248,
+      "step": 557
+    },
+    {
+      "epoch": 0.24987405541561714,
+      "grad_norm": 0.8586876392364502,
+      "learning_rate": 4.510618651892891e-05,
+      "loss": 0.7153,
+      "step": 558
+    },
+    {
+      "epoch": 0.2503218583823118,
+      "grad_norm": 0.9428457021713257,
+      "learning_rate": 4.509464450600185e-05,
+      "loss": 0.8549,
+      "step": 559
+    },
+    {
+      "epoch": 0.2507696613490064,
+      "grad_norm": 0.8758842945098877,
+      "learning_rate": 4.50831024930748e-05,
+      "loss": 0.7171,
+      "step": 560
+    },
+    {
+      "epoch": 0.2507696613490064,
+      "eval_loss": 0.703180193901062,
+      "eval_runtime": 1745.1552,
+      "eval_samples_per_second": 2.56,
+      "eval_steps_per_second": 2.56,
+      "step": 560
+    },
+    {
+      "epoch": 0.2512174643157011,
+      "grad_norm": 0.9924363493919373,
+      "learning_rate": 4.507156048014774e-05,
+      "loss": 0.3454,
+      "step": 561
+    },
+    {
+      "epoch": 0.25166526728239574,
+      "grad_norm": 0.8703386187553406,
+      "learning_rate": 4.5060018467220685e-05,
+      "loss": 0.7339,
+      "step": 562
+    },
+    {
+      "epoch": 0.2521130702490904,
+      "grad_norm": 0.9063810110092163,
+      "learning_rate": 4.5048476454293634e-05,
+      "loss": 0.5657,
+      "step": 563
+    },
+    {
+      "epoch": 0.2525608732157851,
+      "grad_norm": 0.9477468132972717,
+      "learning_rate": 4.503693444136658e-05,
+      "loss": 0.8622,
+      "step": 564
+    },
+    {
+      "epoch": 0.2530086761824797,
+      "grad_norm": 0.8709380030632019,
+      "learning_rate": 4.502539242843952e-05,
+      "loss": 0.7344,
+      "step": 565
+    },
+    {
+      "epoch": 0.25345647914917435,
+      "grad_norm": 1.0082261562347412,
+      "learning_rate": 4.501385041551246e-05,
+      "loss": 0.8182,
+      "step": 566
+    },
+    {
+      "epoch": 0.253904282115869,
+      "grad_norm": 0.891644299030304,
+      "learning_rate": 4.500230840258541e-05,
+      "loss": 0.6383,
+      "step": 567
+    },
+    {
+      "epoch": 0.2543520850825637,
+      "grad_norm": 0.9304959774017334,
+      "learning_rate": 4.4990766389658354e-05,
+      "loss": 0.7813,
+      "step": 568
+    },
+    {
+      "epoch": 0.25479988804925835,
+      "grad_norm": 0.9278199672698975,
+      "learning_rate": 4.4979224376731304e-05,
+      "loss": 0.5809,
+      "step": 569
+    },
+    {
+      "epoch": 0.25524769101595296,
+      "grad_norm": 0.9517792463302612,
+      "learning_rate": 4.4967682363804246e-05,
+      "loss": 0.5944,
+      "step": 570
+    },
+    {
+      "epoch": 0.25524769101595296,
+      "eval_loss": 0.6951916813850403,
+      "eval_runtime": 1748.0579,
+      "eval_samples_per_second": 2.555,
+      "eval_steps_per_second": 2.555,
+      "step": 570
+    },
+    {
+      "epoch": 0.2556954939826476,
+      "grad_norm": 0.9592674374580383,
+      "learning_rate": 4.4956140350877196e-05,
+      "loss": 0.5948,
+      "step": 571
+    },
+    {
+      "epoch": 0.2561432969493423,
+      "grad_norm": 0.7364266514778137,
+      "learning_rate": 4.494459833795014e-05,
+      "loss": 0.6486,
+      "step": 572
+    },
+    {
+      "epoch": 0.25659109991603696,
+      "grad_norm": 0.8261834383010864,
+      "learning_rate": 4.493305632502309e-05,
+      "loss": 0.55,
+      "step": 573
+    },
+    {
+      "epoch": 0.2570389028827316,
+      "grad_norm": 0.8784179091453552,
+      "learning_rate": 4.492151431209603e-05,
+      "loss": 0.7123,
+      "step": 574
+    },
+    {
+      "epoch": 0.25748670584942623,
+      "grad_norm": 0.930644154548645,
+      "learning_rate": 4.490997229916898e-05,
+      "loss": 0.7046,
+      "step": 575
+    },
+    {
+      "epoch": 0.2579345088161209,
+      "grad_norm": 1.0528159141540527,
+      "learning_rate": 4.489843028624192e-05,
+      "loss": 0.6414,
+      "step": 576
+    },
+    {
+      "epoch": 0.25838231178281557,
+      "grad_norm": 0.9306286573410034,
+      "learning_rate": 4.488688827331487e-05,
+      "loss": 0.6506,
+      "step": 577
+    },
+    {
+      "epoch": 0.25883011474951023,
+      "grad_norm": 1.1190401315689087,
+      "learning_rate": 4.4875346260387814e-05,
+      "loss": 0.505,
+      "step": 578
+    },
+    {
+      "epoch": 0.2592779177162049,
+      "grad_norm": 0.9627339839935303,
+      "learning_rate": 4.4863804247460764e-05,
+      "loss": 0.4562,
+      "step": 579
+    },
+    {
+      "epoch": 0.2597257206828995,
+      "grad_norm": 1.0335285663604736,
+      "learning_rate": 4.4852262234533706e-05,
+      "loss": 0.6464,
+      "step": 580
+    },
+    {
+      "epoch": 0.2597257206828995,
+      "eval_loss": 0.689763605594635,
+      "eval_runtime": 1738.6553,
+      "eval_samples_per_second": 2.569,
+      "eval_steps_per_second": 2.569,
+      "step": 580
+    },
+    {
+      "epoch": 0.2601735236495942,
+      "grad_norm": 1.1265842914581299,
+      "learning_rate": 4.4840720221606656e-05,
+      "loss": 0.5609,
+      "step": 581
+    },
+    {
+      "epoch": 0.26062132661628884,
+      "grad_norm": 1.0893819332122803,
+      "learning_rate": 4.48291782086796e-05,
+      "loss": 0.8671,
+      "step": 582
+    },
+    {
+      "epoch": 0.2610691295829835,
+      "grad_norm": 0.9546157717704773,
+      "learning_rate": 4.481763619575254e-05,
+      "loss": 0.7458,
+      "step": 583
+    },
+    {
+      "epoch": 0.2615169325496781,
+      "grad_norm": 0.9704911112785339,
+      "learning_rate": 4.480609418282548e-05,
+      "loss": 0.8723,
+      "step": 584
+    },
+    {
+      "epoch": 0.2619647355163728,
+      "grad_norm": 1.0225448608398438,
+      "learning_rate": 4.479455216989843e-05,
+      "loss": 0.8101,
+      "step": 585
+    },
+    {
+      "epoch": 0.26241253848306745,
+      "grad_norm": 0.9812183976173401,
+      "learning_rate": 4.4783010156971375e-05,
+      "loss": 0.7722,
+      "step": 586
+    },
+    {
+      "epoch": 0.2628603414497621,
+      "grad_norm": 0.8933651447296143,
+      "learning_rate": 4.4771468144044325e-05,
+      "loss": 0.6636,
+      "step": 587
+    },
+    {
+      "epoch": 0.2633081444164568,
+      "grad_norm": 0.8690710067749023,
+      "learning_rate": 4.475992613111727e-05,
+      "loss": 0.6494,
+      "step": 588
+    },
+    {
+      "epoch": 0.2637559473831514,
+      "grad_norm": 0.9920389652252197,
+      "learning_rate": 4.474838411819021e-05,
+      "loss": 0.3813,
+      "step": 589
+    },
+    {
+      "epoch": 0.26420375034984606,
+      "grad_norm": 0.9435338973999023,
+      "learning_rate": 4.473684210526316e-05,
+      "loss": 0.5487,
+      "step": 590
+    },
+    {
+      "epoch": 0.26420375034984606,
+      "eval_loss": 0.6839437484741211,
+      "eval_runtime": 1734.3464,
+      "eval_samples_per_second": 2.576,
+      "eval_steps_per_second": 2.576,
+      "step": 590
+    },
+    {
+      "epoch": 0.2646515533165407,
+      "grad_norm": 0.7974052429199219,
+      "learning_rate": 4.47253000923361e-05,
+      "loss": 0.6446,
+      "step": 591
+    },
+    {
+      "epoch": 0.2650993562832354,
+      "grad_norm": 0.9143849015235901,
+      "learning_rate": 4.471375807940905e-05,
+      "loss": 0.5984,
+      "step": 592
+    },
+    {
+      "epoch": 0.26554715924993005,
+      "grad_norm": 0.9549746513366699,
+      "learning_rate": 4.4702216066481994e-05,
+      "loss": 0.6503,
+      "step": 593
+    },
+    {
+      "epoch": 0.26599496221662466,
+      "grad_norm": 1.10163414478302,
+      "learning_rate": 4.469067405355494e-05,
+      "loss": 0.5042,
+      "step": 594
+    },
+    {
+      "epoch": 0.26644276518331933,
+      "grad_norm": 1.0820388793945312,
+      "learning_rate": 4.4679132040627886e-05,
+      "loss": 0.5365,
+      "step": 595
+    },
+    {
+      "epoch": 0.266890568150014,
+      "grad_norm": 1.3410756587982178,
+      "learning_rate": 4.4667590027700835e-05,
+      "loss": 0.5666,
+      "step": 596
+    },
+    {
+      "epoch": 0.26733837111670866,
+      "grad_norm": 1.1864416599273682,
+      "learning_rate": 4.465604801477378e-05,
+      "loss": 0.6475,
+      "step": 597
+    },
+    {
+      "epoch": 0.2677861740834033,
+      "grad_norm": 1.0692484378814697,
+      "learning_rate": 4.464450600184673e-05,
+      "loss": 0.6315,
+      "step": 598
+    },
+    {
+      "epoch": 0.26823397705009794,
+      "grad_norm": 1.0647058486938477,
+      "learning_rate": 4.463296398891967e-05,
+      "loss": 0.7084,
+      "step": 599
+    },
+    {
+      "epoch": 0.2686817800167926,
+      "grad_norm": 0.9759196043014526,
+      "learning_rate": 4.462142197599262e-05,
+      "loss": 0.8468,
+      "step": 600
+    },
+    {
+      "epoch": 0.2686817800167926,
+      "eval_loss": 0.6730876564979553,
+      "eval_runtime": 1733.1552,
+      "eval_samples_per_second": 2.577,
+      "eval_steps_per_second": 2.577,
+      "step": 600
+    },
+    {
+      "epoch": 0.26912958298348727,
+      "grad_norm": 0.8030557632446289,
+      "learning_rate": 4.460987996306556e-05,
+      "loss": 0.4763,
+      "step": 601
+    },
+    {
+      "epoch": 0.26957738595018194,
+      "grad_norm": 1.0184237957000732,
+      "learning_rate": 4.459833795013851e-05,
+      "loss": 0.5457,
+      "step": 602
+    },
+    {
+      "epoch": 0.2700251889168766,
+      "grad_norm": 1.0043987035751343,
+      "learning_rate": 4.4586795937211454e-05,
+      "loss": 0.7558,
+      "step": 603
+    },
+    {
+      "epoch": 0.2704729918835712,
+      "grad_norm": 1.0344120264053345,
+      "learning_rate": 4.4575253924284396e-05,
+      "loss": 0.6735,
+      "step": 604
+    },
+    {
+      "epoch": 0.2709207948502659,
+      "grad_norm": 0.8846050500869751,
+      "learning_rate": 4.456371191135734e-05,
+      "loss": 0.5962,
+      "step": 605
+    },
+    {
+      "epoch": 0.27136859781696054,
+      "grad_norm": 0.9319285750389099,
+      "learning_rate": 4.455216989843029e-05,
+      "loss": 0.6983,
+      "step": 606
+    },
+    {
+      "epoch": 0.2718164007836552,
+      "grad_norm": 0.878506600856781,
+      "learning_rate": 4.454062788550323e-05,
+      "loss": 0.7427,
+      "step": 607
+    },
+    {
+      "epoch": 0.2722642037503498,
+      "grad_norm": 1.0764738321304321,
+      "learning_rate": 4.452908587257618e-05,
+      "loss": 0.5658,
+      "step": 608
+    },
+    {
+      "epoch": 0.2727120067170445,
+      "grad_norm": 0.9462092518806458,
+      "learning_rate": 4.451754385964912e-05,
+      "loss": 0.6126,
+      "step": 609
+    },
+    {
+      "epoch": 0.27315980968373915,
+      "grad_norm": 0.9421838521957397,
+      "learning_rate": 4.450600184672207e-05,
+      "loss": 0.8937,
+      "step": 610
+    },
+    {
+      "epoch": 0.27315980968373915,
+      "eval_loss": 0.6644588112831116,
+      "eval_runtime": 1730.9171,
+      "eval_samples_per_second": 2.581,
+      "eval_steps_per_second": 2.581,
+      "step": 610
+    },
+    {
+      "epoch": 0.2736076126504338,
+      "grad_norm": 1.1748048067092896,
+      "learning_rate": 4.4494459833795015e-05,
+      "loss": 0.5851,
+      "step": 611
+    },
+    {
+      "epoch": 0.2740554156171285,
+      "grad_norm": 0.882234513759613,
+      "learning_rate": 4.4482917820867964e-05,
+      "loss": 0.7101,
+      "step": 612
+    },
+    {
+      "epoch": 0.2745032185838231,
+      "grad_norm": 0.9383102059364319,
+      "learning_rate": 4.447137580794091e-05,
+      "loss": 0.6291,
+      "step": 613
+    },
+    {
+      "epoch": 0.27495102155051776,
+      "grad_norm": 1.1847293376922607,
+      "learning_rate": 4.445983379501385e-05,
+      "loss": 0.6358,
+      "step": 614
+    },
+    {
+      "epoch": 0.2753988245172124,
+      "grad_norm": 1.3309704065322876,
+      "learning_rate": 4.44482917820868e-05,
+      "loss": 0.5924,
+      "step": 615
+    },
+    {
+      "epoch": 0.2758466274839071,
+      "grad_norm": 0.8071926236152649,
+      "learning_rate": 4.443674976915974e-05,
+      "loss": 0.6001,
+      "step": 616
+    },
+    {
+      "epoch": 0.27629443045060176,
+      "grad_norm": 0.928219199180603,
+      "learning_rate": 4.442520775623269e-05,
+      "loss": 0.6982,
+      "step": 617
+    },
+    {
+      "epoch": 0.27674223341729637,
+      "grad_norm": 0.9898912310600281,
+      "learning_rate": 4.4413665743305634e-05,
+      "loss": 0.6513,
+      "step": 618
+    },
+    {
+      "epoch": 0.27719003638399103,
+      "grad_norm": 1.0699115991592407,
+      "learning_rate": 4.440212373037858e-05,
+      "loss": 0.5662,
+      "step": 619
+    },
+    {
+      "epoch": 0.2776378393506857,
+      "grad_norm": 0.9030242562294006,
+      "learning_rate": 4.4390581717451526e-05,
+      "loss": 0.6892,
+      "step": 620
+    },
+    {
+      "epoch": 0.2776378393506857,
+      "eval_loss": 0.6576302647590637,
+      "eval_runtime": 1732.4607,
+      "eval_samples_per_second": 2.578,
+      "eval_steps_per_second": 2.578,
+      "step": 620
+    },
+    {
+      "epoch": 0.27808564231738037,
+      "grad_norm": 1.031383752822876,
+      "learning_rate": 4.4379039704524475e-05,
+      "loss": 0.4274,
+      "step": 621
+    },
+    {
+      "epoch": 0.27853344528407503,
+      "grad_norm": 0.8698911070823669,
+      "learning_rate": 4.436749769159742e-05,
+      "loss": 0.6138,
+      "step": 622
+    },
+    {
+      "epoch": 0.27898124825076964,
+      "grad_norm": 1.0288652181625366,
+      "learning_rate": 4.435595567867036e-05,
+      "loss": 0.6111,
+      "step": 623
+    },
+    {
+      "epoch": 0.2794290512174643,
+      "grad_norm": 1.21888267993927,
+      "learning_rate": 4.434441366574331e-05,
+      "loss": 0.5836,
+      "step": 624
+    },
+    {
+      "epoch": 0.279876854184159,
+      "grad_norm": 0.9868746399879456,
+      "learning_rate": 4.433287165281625e-05,
+      "loss": 0.665,
+      "step": 625
+    },
+    {
+      "epoch": 0.28032465715085364,
+      "grad_norm": 1.1043651103973389,
+      "learning_rate": 4.4321329639889195e-05,
+      "loss": 0.6688,
+      "step": 626
+    },
+    {
+      "epoch": 0.2807724601175483,
+      "grad_norm": 0.9997096657752991,
+      "learning_rate": 4.4309787626962144e-05,
+      "loss": 0.6676,
+      "step": 627
+    },
+    {
+      "epoch": 0.2812202630842429,
+      "grad_norm": 0.8945668339729309,
+      "learning_rate": 4.429824561403509e-05,
+      "loss": 0.5091,
+      "step": 628
+    },
+    {
+      "epoch": 0.2816680660509376,
+      "grad_norm": 0.9280533790588379,
+      "learning_rate": 4.4286703601108036e-05,
+      "loss": 0.3896,
+      "step": 629
+    },
+    {
+      "epoch": 0.28211586901763225,
+      "grad_norm": 0.9969098567962646,
+      "learning_rate": 4.427516158818098e-05,
+      "loss": 0.674,
+      "step": 630
+    },
+    {
+      "epoch": 0.28211586901763225,
+      "eval_loss": 0.6506349444389343,
+      "eval_runtime": 1732.7002,
+      "eval_samples_per_second": 2.578,
+      "eval_steps_per_second": 2.578,
+      "step": 630
+    },
+    {
+      "epoch": 0.2825636719843269,
+      "grad_norm": 0.9867515563964844,
+      "learning_rate": 4.426361957525393e-05,
+      "loss": 0.5049,
+      "step": 631
+    },
+    {
+      "epoch": 0.2830114749510215,
+      "grad_norm": 0.9162868857383728,
+      "learning_rate": 4.425207756232687e-05,
+      "loss": 0.8855,
+      "step": 632
+    },
+    {
+      "epoch": 0.2834592779177162,
+      "grad_norm": 1.0479141473770142,
+      "learning_rate": 4.424053554939982e-05,
+      "loss": 0.686,
+      "step": 633
+    },
+    {
+      "epoch": 0.28390708088441086,
+      "grad_norm": 0.9793229699134827,
+      "learning_rate": 4.422899353647276e-05,
+      "loss": 0.8075,
+      "step": 634
+    },
+    {
+      "epoch": 0.2843548838511055,
+      "grad_norm": 1.1976258754730225,
+      "learning_rate": 4.421745152354571e-05,
+      "loss": 0.7487,
+      "step": 635
+    },
+    {
+      "epoch": 0.2848026868178002,
+      "grad_norm": 1.13443922996521,
+      "learning_rate": 4.4205909510618655e-05,
+      "loss": 0.7534,
+      "step": 636
+    },
+    {
+      "epoch": 0.2852504897844948,
+      "grad_norm": 1.0198792219161987,
+      "learning_rate": 4.4194367497691604e-05,
+      "loss": 0.6477,
+      "step": 637
+    },
+    {
+      "epoch": 0.28569829275118946,
+      "grad_norm": 1.0110708475112915,
+      "learning_rate": 4.418282548476455e-05,
+      "loss": 0.8262,
+      "step": 638
+    },
+    {
+      "epoch": 0.28614609571788413,
+      "grad_norm": 0.9380364418029785,
+      "learning_rate": 4.417128347183749e-05,
+      "loss": 0.9128,
+      "step": 639
+    },
+    {
+      "epoch": 0.2865938986845788,
+      "grad_norm": 0.9336382746696472,
+      "learning_rate": 4.415974145891044e-05,
+      "loss": 0.8021,
+      "step": 640
+    },
+    {
+      "epoch": 0.2865938986845788,
+      "eval_loss": 0.6476196646690369,
+      "eval_runtime": 1733.8978,
+      "eval_samples_per_second": 2.576,
+      "eval_steps_per_second": 2.576,
+      "step": 640
+    },
+    {
+      "epoch": 0.28704170165127346,
+      "grad_norm": 1.0600725412368774,
+      "learning_rate": 4.414819944598338e-05,
+      "loss": 0.4586,
+      "step": 641
+    },
+    {
+      "epoch": 0.2874895046179681,
+      "grad_norm": 1.0352274179458618,
+      "learning_rate": 4.413665743305633e-05,
+      "loss": 0.5248,
+      "step": 642
+    },
+    {
+      "epoch": 0.28793730758466274,
+      "grad_norm": 0.9235619306564331,
+      "learning_rate": 4.412511542012927e-05,
+      "loss": 0.6809,
+      "step": 643
+    },
+    {
+      "epoch": 0.2883851105513574,
+      "grad_norm": 1.0097838640213013,
+      "learning_rate": 4.4113573407202216e-05,
+      "loss": 0.6336,
+      "step": 644
+    },
+    {
+      "epoch": 0.28883291351805207,
+      "grad_norm": 0.9870620965957642,
+      "learning_rate": 4.410203139427516e-05,
+      "loss": 0.6283,
+      "step": 645
+    },
+    {
+      "epoch": 0.28928071648474674,
+      "grad_norm": 0.9528816938400269,
+      "learning_rate": 4.409048938134811e-05,
+      "loss": 0.5438,
+      "step": 646
+    },
+    {
+      "epoch": 0.28972851945144135,
+      "grad_norm": 1.0778117179870605,
+      "learning_rate": 4.407894736842105e-05,
+      "loss": 0.6447,
+      "step": 647
+    },
+    {
+      "epoch": 0.290176322418136,
+      "grad_norm": 1.1421031951904297,
+      "learning_rate": 4.4067405355494e-05,
+      "loss": 0.8675,
+      "step": 648
+    },
+    {
+      "epoch": 0.2906241253848307,
+      "grad_norm": 1.1636042594909668,
+      "learning_rate": 4.405586334256694e-05,
+      "loss": 0.6389,
+      "step": 649
+    },
+    {
+      "epoch": 0.29107192835152534,
+      "grad_norm": 0.926356315612793,
+      "learning_rate": 4.404432132963989e-05,
+      "loss": 0.4951,
+      "step": 650
+    },
+    {
+      "epoch": 0.29107192835152534,
+      "eval_loss": 0.6369218230247498,
+      "eval_runtime": 1733.9976,
+      "eval_samples_per_second": 2.576,
+      "eval_steps_per_second": 2.576,
+      "step": 650
+    },
+    {
+      "epoch": 0.29151973131822,
+      "grad_norm": 0.835151731967926,
+      "learning_rate": 4.4032779316712835e-05,
+      "loss": 0.6196,
+      "step": 651
+    },
+    {
+      "epoch": 0.2919675342849146,
+      "grad_norm": 0.8978967666625977,
+      "learning_rate": 4.4021237303785784e-05,
+      "loss": 0.8643,
+      "step": 652
+    },
+    {
+      "epoch": 0.2924153372516093,
+      "grad_norm": 0.956881582736969,
+      "learning_rate": 4.4009695290858727e-05,
+      "loss": 0.5167,
+      "step": 653
+    },
+    {
+      "epoch": 0.29286314021830395,
+      "grad_norm": 0.8161943554878235,
+      "learning_rate": 4.3998153277931676e-05,
+      "loss": 0.3244,
+      "step": 654
+    },
+    {
+      "epoch": 0.2933109431849986,
+      "grad_norm": 1.175564169883728,
+      "learning_rate": 4.398661126500462e-05,
+      "loss": 0.6833,
+      "step": 655
+    },
+    {
+      "epoch": 0.29375874615169323,
+      "grad_norm": 0.8921207189559937,
+      "learning_rate": 4.397506925207757e-05,
+      "loss": 0.5833,
+      "step": 656
+    },
+    {
+      "epoch": 0.2942065491183879,
+      "grad_norm": 1.0308955907821655,
+      "learning_rate": 4.396352723915051e-05,
+      "loss": 0.6401,
+      "step": 657
+    },
+    {
+      "epoch": 0.29465435208508256,
+      "grad_norm": 0.8860639333724976,
+      "learning_rate": 4.395198522622346e-05,
+      "loss": 0.6095,
+      "step": 658
+    },
+    {
+      "epoch": 0.2951021550517772,
+      "grad_norm": 1.0290175676345825,
+      "learning_rate": 4.39404432132964e-05,
+      "loss": 0.8198,
+      "step": 659
+    },
+    {
+      "epoch": 0.2955499580184719,
+      "grad_norm": 1.1758856773376465,
+      "learning_rate": 4.392890120036935e-05,
+      "loss": 0.5188,
+      "step": 660
+    },
+    {
+      "epoch": 0.2955499580184719,
+      "eval_loss": 0.6303642988204956,
+      "eval_runtime": 1735.5608,
+      "eval_samples_per_second": 2.574,
+      "eval_steps_per_second": 2.574,
+      "step": 660
+    },
+    {
+      "epoch": 0.2959977609851665,
+      "grad_norm": 0.9396849274635315,
+      "learning_rate": 4.3917359187442294e-05,
+      "loss": 0.5156,
+      "step": 661
+    },
+    {
+      "epoch": 0.29644556395186117,
+      "grad_norm": 1.129699468612671,
+      "learning_rate": 4.390581717451524e-05,
+      "loss": 0.6017,
+      "step": 662
+    },
+    {
+      "epoch": 0.29689336691855583,
+      "grad_norm": 1.2421824932098389,
+      "learning_rate": 4.3894275161588186e-05,
+      "loss": 0.6726,
+      "step": 663
+    },
+    {
+      "epoch": 0.2973411698852505,
+      "grad_norm": 0.9329867959022522,
+      "learning_rate": 4.388273314866113e-05,
+      "loss": 0.5167,
+      "step": 664
+    },
+    {
+      "epoch": 0.29778897285194517,
+      "grad_norm": 1.0112805366516113,
+      "learning_rate": 4.387119113573407e-05,
+      "loss": 0.704,
+      "step": 665
+    },
+    {
+      "epoch": 0.2982367758186398,
+      "grad_norm": 1.0268175601959229,
+      "learning_rate": 4.3859649122807014e-05,
+      "loss": 0.6813,
+      "step": 666
+    },
+    {
+      "epoch": 0.29868457878533444,
+      "grad_norm": 0.8663952350616455,
+      "learning_rate": 4.3848107109879964e-05,
+      "loss": 0.7125,
+      "step": 667
+    },
+    {
+      "epoch": 0.2991323817520291,
+      "grad_norm": 1.1380077600479126,
+      "learning_rate": 4.3836565096952906e-05,
+      "loss": 0.5561,
+      "step": 668
+    },
+    {
+      "epoch": 0.2995801847187238,
+      "grad_norm": 1.1226226091384888,
+      "learning_rate": 4.3825023084025856e-05,
+      "loss": 0.6075,
+      "step": 669
+    },
+    {
+      "epoch": 0.30002798768541844,
+      "grad_norm": 1.0129913091659546,
+      "learning_rate": 4.38134810710988e-05,
+      "loss": 0.6707,
+      "step": 670
+    },
+    {
+      "epoch": 0.30002798768541844,
+      "eval_loss": 0.6253139972686768,
+      "eval_runtime": 1735.9274,
+      "eval_samples_per_second": 2.573,
+      "eval_steps_per_second": 2.573,
+      "step": 670
+    },
+    {
+      "epoch": 0.30047579065211305,
+      "grad_norm": 0.9254921674728394,
+      "learning_rate": 4.380193905817175e-05,
+      "loss": 0.5443,
+      "step": 671
+    },
+    {
+      "epoch": 0.3009235936188077,
+      "grad_norm": 0.9654009938240051,
+      "learning_rate": 4.379039704524469e-05,
+      "loss": 0.6952,
+      "step": 672
+    },
+    {
+      "epoch": 0.3013713965855024,
+      "grad_norm": 1.0288201570510864,
+      "learning_rate": 4.377885503231764e-05,
+      "loss": 0.5304,
+      "step": 673
+    },
+    {
+      "epoch": 0.30181919955219705,
+      "grad_norm": 1.2195161581039429,
+      "learning_rate": 4.376731301939058e-05,
+      "loss": 0.7748,
+      "step": 674
+    },
+    {
+      "epoch": 0.3022670025188917,
+      "grad_norm": 1.2199338674545288,
+      "learning_rate": 4.375577100646353e-05,
+      "loss": 0.3766,
+      "step": 675
+    },
+    {
+      "epoch": 0.3027148054855863,
+      "grad_norm": 1.010966181755066,
+      "learning_rate": 4.3744228993536474e-05,
+      "loss": 0.8081,
+      "step": 676
+    },
+    {
+      "epoch": 0.303162608452281,
+      "grad_norm": 1.1240795850753784,
+      "learning_rate": 4.3732686980609424e-05,
+      "loss": 0.8757,
+      "step": 677
+    },
+    {
+      "epoch": 0.30361041141897566,
+      "grad_norm": 1.2343254089355469,
+      "learning_rate": 4.3721144967682366e-05,
+      "loss": 0.7133,
+      "step": 678
+    },
+    {
+      "epoch": 0.3040582143856703,
+      "grad_norm": 0.9523788690567017,
+      "learning_rate": 4.3709602954755316e-05,
+      "loss": 0.6402,
+      "step": 679
+    },
+    {
+      "epoch": 0.30450601735236493,
+      "grad_norm": 0.9624511003494263,
+      "learning_rate": 4.369806094182826e-05,
+      "loss": 0.7798,
+      "step": 680
+    },
+    {
+      "epoch": 0.30450601735236493,
+      "eval_loss": 0.620413601398468,
+      "eval_runtime": 1735.0028,
+      "eval_samples_per_second": 2.575,
+      "eval_steps_per_second": 2.575,
+      "step": 680
+    },
+    {
+      "epoch": 0.3049538203190596,
+      "grad_norm": 1.0228697061538696,
+      "learning_rate": 4.368651892890121e-05,
+      "loss": 0.47,
+      "step": 681
+    },
+    {
+      "epoch": 0.30540162328575426,
+      "grad_norm": 0.9654502272605896,
+      "learning_rate": 4.367497691597415e-05,
+      "loss": 0.597,
+      "step": 682
+    },
+    {
+      "epoch": 0.30584942625244893,
+      "grad_norm": 1.1050530672073364,
+      "learning_rate": 4.366343490304709e-05,
+      "loss": 0.6511,
+      "step": 683
+    },
+    {
+      "epoch": 0.3062972292191436,
+      "grad_norm": 0.8982037901878357,
+      "learning_rate": 4.3651892890120035e-05,
+      "loss": 0.4284,
+      "step": 684
+    },
+    {
+      "epoch": 0.3067450321858382,
+      "grad_norm": 1.0873932838439941,
+      "learning_rate": 4.3640350877192985e-05,
+      "loss": 0.7394,
+      "step": 685
+    },
+    {
+      "epoch": 0.3071928351525329,
+      "grad_norm": 1.0383609533309937,
+      "learning_rate": 4.362880886426593e-05,
+      "loss": 0.5046,
+      "step": 686
+    },
+    {
+      "epoch": 0.30764063811922754,
+      "grad_norm": 1.1227695941925049,
+      "learning_rate": 4.361726685133888e-05,
+      "loss": 0.34,
+      "step": 687
+    },
+    {
+      "epoch": 0.3080884410859222,
+      "grad_norm": 0.9949764013290405,
+      "learning_rate": 4.360572483841182e-05,
+      "loss": 0.4576,
+      "step": 688
+    },
+    {
+      "epoch": 0.30853624405261687,
+      "grad_norm": 1.1606419086456299,
+      "learning_rate": 4.359418282548476e-05,
+      "loss": 0.5196,
+      "step": 689
+    },
+    {
+      "epoch": 0.3089840470193115,
+      "grad_norm": 1.3154551982879639,
+      "learning_rate": 4.358264081255771e-05,
+      "loss": 0.5968,
+      "step": 690
+    },
+    {
+      "epoch": 0.3089840470193115,
+      "eval_loss": 0.6166278123855591,
+      "eval_runtime": 1736.0293,
+      "eval_samples_per_second": 2.573,
+      "eval_steps_per_second": 2.573,
+      "step": 690
+    },
+    {
+      "epoch": 0.30943184998600615,
+      "grad_norm": 1.2704520225524902,
+      "learning_rate": 4.3571098799630654e-05,
+      "loss": 0.668,
+      "step": 691
+    },
+    {
+      "epoch": 0.3098796529527008,
+      "grad_norm": 1.2325021028518677,
+      "learning_rate": 4.3559556786703603e-05,
+      "loss": 0.4863,
+      "step": 692
+    },
+    {
+      "epoch": 0.3103274559193955,
+      "grad_norm": 1.1729905605316162,
+      "learning_rate": 4.3548014773776546e-05,
+      "loss": 0.7218,
+      "step": 693
+    },
+    {
+      "epoch": 0.31077525888609014,
+      "grad_norm": 1.1204662322998047,
+      "learning_rate": 4.3536472760849495e-05,
+      "loss": 0.9501,
+      "step": 694
+    },
+    {
+      "epoch": 0.31122306185278475,
+      "grad_norm": 1.0278434753417969,
+      "learning_rate": 4.352493074792244e-05,
+      "loss": 0.5807,
+      "step": 695
+    },
+    {
+      "epoch": 0.3116708648194794,
+      "grad_norm": 0.9404197335243225,
+      "learning_rate": 4.351338873499539e-05,
+      "loss": 0.4889,
+      "step": 696
+    },
+    {
+      "epoch": 0.3121186677861741,
+      "grad_norm": 1.0532269477844238,
+      "learning_rate": 4.350184672206833e-05,
+      "loss": 0.5541,
+      "step": 697
+    },
+    {
+      "epoch": 0.31256647075286875,
+      "grad_norm": 1.015628695487976,
+      "learning_rate": 4.349030470914128e-05,
+      "loss": 0.5046,
+      "step": 698
+    },
+    {
+      "epoch": 0.3130142737195634,
+      "grad_norm": 0.9070887565612793,
+      "learning_rate": 4.347876269621422e-05,
+      "loss": 0.6346,
+      "step": 699
+    },
+    {
+      "epoch": 0.31346207668625803,
+      "grad_norm": 1.0546003580093384,
+      "learning_rate": 4.346722068328717e-05,
+      "loss": 0.4904,
+      "step": 700
+    },
+    {
+      "epoch": 0.31346207668625803,
+      "eval_loss": 0.6045358777046204,
+      "eval_runtime": 1734.5753,
+      "eval_samples_per_second": 2.575,
+      "eval_steps_per_second": 2.575,
+      "step": 700
+    },
+    {
+      "epoch": 0.3139098796529527,
+      "grad_norm": 1.2544265985488892,
+      "learning_rate": 4.3455678670360114e-05,
+      "loss": 0.6624,
+      "step": 701
+    },
+    {
+      "epoch": 0.31435768261964736,
+      "grad_norm": 0.9051249027252197,
+      "learning_rate": 4.3444136657433057e-05,
+      "loss": 0.5293,
+      "step": 702
+    },
+    {
+      "epoch": 0.314805485586342,
+      "grad_norm": 1.036263108253479,
+      "learning_rate": 4.3432594644506006e-05,
+      "loss": 0.7179,
+      "step": 703
+    },
+    {
+      "epoch": 0.31525328855303664,
+      "grad_norm": 1.0076425075531006,
+      "learning_rate": 4.342105263157895e-05,
+      "loss": 0.643,
+      "step": 704
+    },
+    {
+      "epoch": 0.3157010915197313,
+      "grad_norm": 1.236649751663208,
+      "learning_rate": 4.340951061865189e-05,
+      "loss": 0.8407,
+      "step": 705
+    },
+    {
+      "epoch": 0.31614889448642597,
+      "grad_norm": 1.3177474737167358,
+      "learning_rate": 4.339796860572484e-05,
+      "loss": 0.6383,
+      "step": 706
+    },
+    {
+      "epoch": 0.31659669745312063,
+      "grad_norm": 1.1313387155532837,
+      "learning_rate": 4.338642659279778e-05,
+      "loss": 0.7257,
+      "step": 707
+    },
+    {
+      "epoch": 0.3170445004198153,
+      "grad_norm": 1.0173510313034058,
+      "learning_rate": 4.337488457987073e-05,
+      "loss": 0.4731,
+      "step": 708
+    },
+    {
+      "epoch": 0.3174923033865099,
+      "grad_norm": 1.1304107904434204,
+      "learning_rate": 4.3363342566943675e-05,
+      "loss": 0.5356,
+      "step": 709
+    },
+    {
+      "epoch": 0.3179401063532046,
+      "grad_norm": 0.9915953278541565,
+      "learning_rate": 4.3351800554016625e-05,
+      "loss": 0.5619,
+      "step": 710
+    },
+    {
+      "epoch": 0.3179401063532046,
+      "eval_loss": 0.598295271396637,
+      "eval_runtime": 1735.247,
+      "eval_samples_per_second": 2.574,
+      "eval_steps_per_second": 2.574,
+      "step": 710
+    },
+    {
+      "epoch": 0.31838790931989924,
+      "grad_norm": 1.1005866527557373,
+      "learning_rate": 4.334025854108957e-05,
+      "loss": 0.839,
+      "step": 711
+    },
+    {
+      "epoch": 0.3188357122865939,
+      "grad_norm": 0.9614087343215942,
+      "learning_rate": 4.3328716528162517e-05,
+      "loss": 0.6166,
+      "step": 712
+    },
+    {
+      "epoch": 0.3192835152532886,
+      "grad_norm": 1.0046594142913818,
+      "learning_rate": 4.331717451523546e-05,
+      "loss": 0.4292,
+      "step": 713
+    },
+    {
+      "epoch": 0.3197313182199832,
+      "grad_norm": 1.0307554006576538,
+      "learning_rate": 4.33056325023084e-05,
+      "loss": 0.7486,
+      "step": 714
+    },
+    {
+      "epoch": 0.32017912118667785,
+      "grad_norm": 1.0028913021087646,
+      "learning_rate": 4.329409048938135e-05,
+      "loss": 0.5175,
+      "step": 715
+    },
+    {
+      "epoch": 0.3206269241533725,
+      "grad_norm": 1.2011581659317017,
+      "learning_rate": 4.3282548476454294e-05,
+      "loss": 0.5116,
+      "step": 716
+    },
+    {
+      "epoch": 0.3210747271200672,
+      "grad_norm": 0.9902753829956055,
+      "learning_rate": 4.327100646352724e-05,
+      "loss": 0.3947,
+      "step": 717
+    },
+    {
+      "epoch": 0.32152253008676185,
+      "grad_norm": 0.945852518081665,
+      "learning_rate": 4.3259464450600186e-05,
+      "loss": 0.6836,
+      "step": 718
+    },
+    {
+      "epoch": 0.32197033305345646,
+      "grad_norm": 1.0151649713516235,
+      "learning_rate": 4.3247922437673135e-05,
+      "loss": 0.6962,
+      "step": 719
+    },
+    {
+      "epoch": 0.3224181360201511,
+      "grad_norm": 1.1140705347061157,
+      "learning_rate": 4.323638042474608e-05,
+      "loss": 0.3008,
+      "step": 720
+    },
+    {
+      "epoch": 0.3224181360201511,
+      "eval_loss": 0.5958486199378967,
+      "eval_runtime": 1737.2477,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 720
+    },
+    {
+      "epoch": 0.3228659389868458,
+      "grad_norm": 1.09420907497406,
+      "learning_rate": 4.322483841181903e-05,
+      "loss": 0.6773,
+      "step": 721
+    },
+    {
+      "epoch": 0.32331374195354046,
+      "grad_norm": 1.2701865434646606,
+      "learning_rate": 4.321329639889197e-05,
+      "loss": 0.6063,
+      "step": 722
+    },
+    {
+      "epoch": 0.3237615449202351,
+      "grad_norm": 1.0487083196640015,
+      "learning_rate": 4.320175438596491e-05,
+      "loss": 0.6196,
+      "step": 723
+    },
+    {
+      "epoch": 0.32420934788692973,
+      "grad_norm": 1.1093167066574097,
+      "learning_rate": 4.3190212373037855e-05,
+      "loss": 0.5419,
+      "step": 724
+    },
+    {
+      "epoch": 0.3246571508536244,
+      "grad_norm": 1.1824114322662354,
+      "learning_rate": 4.3178670360110804e-05,
+      "loss": 0.6709,
+      "step": 725
+    },
+    {
+      "epoch": 0.32510495382031906,
+      "grad_norm": 1.077880859375,
+      "learning_rate": 4.316712834718375e-05,
+      "loss": 0.6136,
+      "step": 726
+    },
+    {
+      "epoch": 0.32555275678701373,
+      "grad_norm": 0.8415115475654602,
+      "learning_rate": 4.3155586334256696e-05,
+      "loss": 0.5034,
+      "step": 727
+    },
+    {
+      "epoch": 0.32600055975370834,
+      "grad_norm": 1.0477930307388306,
+      "learning_rate": 4.314404432132964e-05,
+      "loss": 0.5437,
+      "step": 728
+    },
+    {
+      "epoch": 0.326448362720403,
+      "grad_norm": 1.1078599691390991,
+      "learning_rate": 4.313250230840259e-05,
+      "loss": 0.648,
+      "step": 729
+    },
+    {
+      "epoch": 0.3268961656870977,
+      "grad_norm": 0.946194589138031,
+      "learning_rate": 4.312096029547553e-05,
+      "loss": 0.5502,
+      "step": 730
+    },
+    {
+      "epoch": 0.3268961656870977,
+      "eval_loss": 0.5906120538711548,
+      "eval_runtime": 1737.2839,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 730
+    },
+    {
+      "epoch": 0.32734396865379234,
+      "grad_norm": 0.9358011484146118,
+      "learning_rate": 4.310941828254848e-05,
+      "loss": 0.5899,
+      "step": 731
+    },
+    {
+      "epoch": 0.327791771620487,
+      "grad_norm": 1.104750633239746,
+      "learning_rate": 4.309787626962142e-05,
+      "loss": 0.6237,
+      "step": 732
+    },
+    {
+      "epoch": 0.3282395745871816,
+      "grad_norm": 0.8731098771095276,
+      "learning_rate": 4.308633425669437e-05,
+      "loss": 0.5086,
+      "step": 733
+    },
+    {
+      "epoch": 0.3286873775538763,
+      "grad_norm": 1.0071966648101807,
+      "learning_rate": 4.3074792243767315e-05,
+      "loss": 0.6828,
+      "step": 734
+    },
+    {
+      "epoch": 0.32913518052057095,
+      "grad_norm": 1.203137993812561,
+      "learning_rate": 4.3063250230840264e-05,
+      "loss": 0.4755,
+      "step": 735
+    },
+    {
+      "epoch": 0.3295829834872656,
+      "grad_norm": 0.9339362382888794,
+      "learning_rate": 4.305170821791321e-05,
+      "loss": 0.5706,
+      "step": 736
+    },
+    {
+      "epoch": 0.3300307864539603,
+      "grad_norm": 1.1789636611938477,
+      "learning_rate": 4.3040166204986156e-05,
+      "loss": 0.8102,
+      "step": 737
+    },
+    {
+      "epoch": 0.3304785894206549,
+      "grad_norm": 1.0512614250183105,
+      "learning_rate": 4.30286241920591e-05,
+      "loss": 0.3716,
+      "step": 738
+    },
+    {
+      "epoch": 0.33092639238734955,
+      "grad_norm": 1.079956293106079,
+      "learning_rate": 4.301708217913204e-05,
+      "loss": 0.4777,
+      "step": 739
+    },
+    {
+      "epoch": 0.3313741953540442,
+      "grad_norm": 0.9875457882881165,
+      "learning_rate": 4.300554016620499e-05,
+      "loss": 0.5379,
+      "step": 740
+    },
+    {
+      "epoch": 0.3313741953540442,
+      "eval_loss": 0.5849510431289673,
+      "eval_runtime": 1736.9612,
+      "eval_samples_per_second": 2.572,
+      "eval_steps_per_second": 2.572,
+      "step": 740
+    },
+    {
+      "epoch": 0.3318219983207389,
+      "grad_norm": 1.170372486114502,
+      "learning_rate": 4.2993998153277933e-05,
+      "loss": 0.6948,
+      "step": 741
+    },
+    {
+      "epoch": 0.33226980128743355,
+      "grad_norm": 1.150221347808838,
+      "learning_rate": 4.298245614035088e-05,
+      "loss": 0.6632,
+      "step": 742
+    },
+    {
+      "epoch": 0.33271760425412816,
+      "grad_norm": 0.9060766696929932,
+      "learning_rate": 4.2970914127423825e-05,
+      "loss": 0.4876,
+      "step": 743
+    },
+    {
+      "epoch": 0.33316540722082283,
+      "grad_norm": 1.0560237169265747,
+      "learning_rate": 4.295937211449677e-05,
+      "loss": 0.4003,
+      "step": 744
+    },
+    {
+      "epoch": 0.3336132101875175,
+      "grad_norm": 1.204740047454834,
+      "learning_rate": 4.294783010156971e-05,
+      "loss": 0.5309,
+      "step": 745
+    },
+    {
+      "epoch": 0.33406101315421216,
+      "grad_norm": 1.0319269895553589,
+      "learning_rate": 4.293628808864266e-05,
+      "loss": 0.4674,
+      "step": 746
+    },
+    {
+      "epoch": 0.3345088161209068,
+      "grad_norm": 1.0303031206130981,
+      "learning_rate": 4.29247460757156e-05,
+      "loss": 0.7802,
+      "step": 747
+    },
+    {
+      "epoch": 0.33495661908760144,
+      "grad_norm": 1.2019189596176147,
+      "learning_rate": 4.291320406278855e-05,
+      "loss": 0.6186,
+      "step": 748
+    },
+    {
+      "epoch": 0.3354044220542961,
+      "grad_norm": 1.0149527788162231,
+      "learning_rate": 4.2901662049861495e-05,
+      "loss": 0.3242,
+      "step": 749
+    },
+    {
+      "epoch": 0.33585222502099077,
+      "grad_norm": 1.458787441253662,
+      "learning_rate": 4.2890120036934444e-05,
+      "loss": 0.8334,
+      "step": 750
+    },
+    {
+      "epoch": 0.33585222502099077,
+      "eval_loss": 0.5777731537818909,
+      "eval_runtime": 1737.7233,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 750
+    },
+    {
+      "epoch": 0.33630002798768543,
+      "grad_norm": 0.9327331781387329,
+      "learning_rate": 4.2878578024007387e-05,
+      "loss": 0.494,
+      "step": 751
+    },
+    {
+      "epoch": 0.33674783095438005,
+      "grad_norm": 0.9456080794334412,
+      "learning_rate": 4.2867036011080336e-05,
+      "loss": 0.3454,
+      "step": 752
+    },
+    {
+      "epoch": 0.3371956339210747,
+      "grad_norm": 1.035635232925415,
+      "learning_rate": 4.285549399815328e-05,
+      "loss": 0.6425,
+      "step": 753
+    },
+    {
+      "epoch": 0.3376434368877694,
+      "grad_norm": 1.1074715852737427,
+      "learning_rate": 4.284395198522623e-05,
+      "loss": 0.653,
+      "step": 754
+    },
+    {
+      "epoch": 0.33809123985446404,
+      "grad_norm": 1.190762996673584,
+      "learning_rate": 4.283240997229917e-05,
+      "loss": 0.7805,
+      "step": 755
+    },
+    {
+      "epoch": 0.3385390428211587,
+      "grad_norm": 0.8755838871002197,
+      "learning_rate": 4.282086795937212e-05,
+      "loss": 0.5877,
+      "step": 756
+    },
+    {
+      "epoch": 0.3389868457878533,
+      "grad_norm": 0.9096702337265015,
+      "learning_rate": 4.280932594644506e-05,
+      "loss": 0.6816,
+      "step": 757
+    },
+    {
+      "epoch": 0.339434648754548,
+      "grad_norm": 0.9515770673751831,
+      "learning_rate": 4.279778393351801e-05,
+      "loss": 0.6265,
+      "step": 758
+    },
+    {
+      "epoch": 0.33988245172124265,
+      "grad_norm": 1.0240706205368042,
+      "learning_rate": 4.2786241920590955e-05,
+      "loss": 0.4192,
+      "step": 759
+    },
+    {
+      "epoch": 0.3403302546879373,
+      "grad_norm": 1.1301640272140503,
+      "learning_rate": 4.2774699907663904e-05,
+      "loss": 0.4906,
+      "step": 760
+    },
+    {
+      "epoch": 0.3403302546879373,
+      "eval_loss": 0.5691836476325989,
+      "eval_runtime": 1736.0523,
+      "eval_samples_per_second": 2.573,
+      "eval_steps_per_second": 2.573,
+      "step": 760
+    },
+    {
+      "epoch": 0.340778057654632,
+      "grad_norm": 1.1020630598068237,
+      "learning_rate": 4.2763157894736847e-05,
+      "loss": 0.6314,
+      "step": 761
+    },
+    {
+      "epoch": 0.3412258606213266,
+      "grad_norm": 0.9853556752204895,
+      "learning_rate": 4.275161588180979e-05,
+      "loss": 0.288,
+      "step": 762
+    },
+    {
+      "epoch": 0.34167366358802126,
+      "grad_norm": 0.965360164642334,
+      "learning_rate": 4.274007386888273e-05,
+      "loss": 0.5128,
+      "step": 763
+    },
+    {
+      "epoch": 0.3421214665547159,
+      "grad_norm": 1.1848323345184326,
+      "learning_rate": 4.272853185595568e-05,
+      "loss": 0.6169,
+      "step": 764
+    },
+    {
+      "epoch": 0.3425692695214106,
+      "grad_norm": 1.1289207935333252,
+      "learning_rate": 4.2716989843028624e-05,
+      "loss": 0.687,
+      "step": 765
+    },
+    {
+      "epoch": 0.34301707248810526,
+      "grad_norm": 1.0918805599212646,
+      "learning_rate": 4.2705447830101566e-05,
+      "loss": 0.5477,
+      "step": 766
+    },
+    {
+      "epoch": 0.34346487545479987,
+      "grad_norm": 0.9774726033210754,
+      "learning_rate": 4.2693905817174516e-05,
+      "loss": 0.6455,
+      "step": 767
+    },
+    {
+      "epoch": 0.34391267842149453,
+      "grad_norm": 1.1208592653274536,
+      "learning_rate": 4.268236380424746e-05,
+      "loss": 0.492,
+      "step": 768
+    },
+    {
+      "epoch": 0.3443604813881892,
+      "grad_norm": 1.0818626880645752,
+      "learning_rate": 4.267082179132041e-05,
+      "loss": 0.7567,
+      "step": 769
+    },
+    {
+      "epoch": 0.34480828435488386,
+      "grad_norm": 1.057212471961975,
+      "learning_rate": 4.265927977839335e-05,
+      "loss": 0.6216,
+      "step": 770
+    },
+    {
+      "epoch": 0.34480828435488386,
+      "eval_loss": 0.5637683868408203,
+      "eval_runtime": 1737.2303,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 770
+    },
+    {
+      "epoch": 0.34525608732157853,
+      "grad_norm": 1.2107055187225342,
+      "learning_rate": 4.26477377654663e-05,
+      "loss": 0.4313,
+      "step": 771
+    },
+    {
+      "epoch": 0.34570389028827314,
+      "grad_norm": 1.0803090333938599,
+      "learning_rate": 4.263619575253924e-05,
+      "loss": 0.7932,
+      "step": 772
+    },
+    {
+      "epoch": 0.3461516932549678,
+      "grad_norm": 0.9676966667175293,
+      "learning_rate": 4.262465373961219e-05,
+      "loss": 0.57,
+      "step": 773
+    },
+    {
+      "epoch": 0.3465994962216625,
+      "grad_norm": 1.0771076679229736,
+      "learning_rate": 4.2613111726685134e-05,
+      "loss": 0.591,
+      "step": 774
+    },
+    {
+      "epoch": 0.34704729918835714,
+      "grad_norm": 1.1740727424621582,
+      "learning_rate": 4.2601569713758084e-05,
+      "loss": 0.8161,
+      "step": 775
+    },
+    {
+      "epoch": 0.3474951021550518,
+      "grad_norm": 1.026063323020935,
+      "learning_rate": 4.2590027700831026e-05,
+      "loss": 0.6341,
+      "step": 776
+    },
+    {
+      "epoch": 0.3479429051217464,
+      "grad_norm": 1.220108985900879,
+      "learning_rate": 4.2578485687903976e-05,
+      "loss": 0.5722,
+      "step": 777
+    },
+    {
+      "epoch": 0.3483907080884411,
+      "grad_norm": 0.8669452667236328,
+      "learning_rate": 4.256694367497692e-05,
+      "loss": 0.6794,
+      "step": 778
+    },
+    {
+      "epoch": 0.34883851105513575,
+      "grad_norm": 1.0848174095153809,
+      "learning_rate": 4.255540166204987e-05,
+      "loss": 0.4824,
+      "step": 779
+    },
+    {
+      "epoch": 0.3492863140218304,
+      "grad_norm": 0.985701858997345,
+      "learning_rate": 4.254385964912281e-05,
+      "loss": 0.6193,
+      "step": 780
+    },
+    {
+      "epoch": 0.3492863140218304,
+      "eval_loss": 0.5568503737449646,
+      "eval_runtime": 1733.3312,
+      "eval_samples_per_second": 2.577,
+      "eval_steps_per_second": 2.577,
+      "step": 780
+    },
+    {
+      "epoch": 0.349734116988525,
+      "grad_norm": 1.01187264919281,
+      "learning_rate": 4.253231763619576e-05,
+      "loss": 0.6011,
+      "step": 781
+    },
+    {
+      "epoch": 0.3501819199552197,
+      "grad_norm": 1.1729329824447632,
+      "learning_rate": 4.25207756232687e-05,
+      "loss": 0.4948,
+      "step": 782
+    },
+    {
+      "epoch": 0.35062972292191436,
+      "grad_norm": 1.123641014099121,
+      "learning_rate": 4.2509233610341645e-05,
+      "loss": 0.6441,
+      "step": 783
+    },
+    {
+      "epoch": 0.351077525888609,
+      "grad_norm": 1.071698784828186,
+      "learning_rate": 4.249769159741459e-05,
+      "loss": 0.3689,
+      "step": 784
+    },
+    {
+      "epoch": 0.3515253288553037,
+      "grad_norm": 1.2566313743591309,
+      "learning_rate": 4.248614958448754e-05,
+      "loss": 0.574,
+      "step": 785
+    },
+    {
+      "epoch": 0.3519731318219983,
+      "grad_norm": 0.8795626163482666,
+      "learning_rate": 4.247460757156048e-05,
+      "loss": 0.574,
+      "step": 786
+    },
+    {
+      "epoch": 0.35242093478869296,
+      "grad_norm": 1.1304922103881836,
+      "learning_rate": 4.246306555863343e-05,
+      "loss": 0.574,
+      "step": 787
+    },
+    {
+      "epoch": 0.35286873775538763,
+      "grad_norm": 0.9006721377372742,
+      "learning_rate": 4.245152354570637e-05,
+      "loss": 0.6294,
+      "step": 788
+    },
+    {
+      "epoch": 0.3533165407220823,
+      "grad_norm": 1.0647133588790894,
+      "learning_rate": 4.2439981532779314e-05,
+      "loss": 0.3783,
+      "step": 789
+    },
+    {
+      "epoch": 0.35376434368877696,
+      "grad_norm": 1.190906286239624,
+      "learning_rate": 4.2428439519852263e-05,
+      "loss": 0.5647,
+      "step": 790
+    },
+    {
+      "epoch": 0.35376434368877696,
+      "eval_loss": 0.5501430630683899,
+      "eval_runtime": 1734.9154,
+      "eval_samples_per_second": 2.575,
+      "eval_steps_per_second": 2.575,
+      "step": 790
+    },
+    {
+      "epoch": 0.35421214665547157,
+      "grad_norm": 1.3009206056594849,
+      "learning_rate": 4.2416897506925206e-05,
+      "loss": 0.5486,
+      "step": 791
+    },
+    {
+      "epoch": 0.35465994962216624,
+      "grad_norm": 0.9215652346611023,
+      "learning_rate": 4.2405355493998155e-05,
+      "loss": 0.3632,
+      "step": 792
+    },
+    {
+      "epoch": 0.3551077525888609,
+      "grad_norm": 0.9980651140213013,
+      "learning_rate": 4.23938134810711e-05,
+      "loss": 0.4124,
+      "step": 793
+    },
+    {
+      "epoch": 0.35555555555555557,
+      "grad_norm": 1.074180006980896,
+      "learning_rate": 4.238227146814405e-05,
+      "loss": 0.6014,
+      "step": 794
+    },
+    {
+      "epoch": 0.35600335852225024,
+      "grad_norm": 1.0638560056686401,
+      "learning_rate": 4.237072945521699e-05,
+      "loss": 0.454,
+      "step": 795
+    },
+    {
+      "epoch": 0.35645116148894485,
+      "grad_norm": 1.1408402919769287,
+      "learning_rate": 4.235918744228994e-05,
+      "loss": 0.5157,
+      "step": 796
+    },
+    {
+      "epoch": 0.3568989644556395,
+      "grad_norm": 1.1119613647460938,
+      "learning_rate": 4.234764542936288e-05,
+      "loss": 0.6754,
+      "step": 797
+    },
+    {
+      "epoch": 0.3573467674223342,
+      "grad_norm": 1.1197654008865356,
+      "learning_rate": 4.233610341643583e-05,
+      "loss": 0.4981,
+      "step": 798
+    },
+    {
+      "epoch": 0.35779457038902884,
+      "grad_norm": 1.090043306350708,
+      "learning_rate": 4.2324561403508774e-05,
+      "loss": 0.716,
+      "step": 799
+    },
+    {
+      "epoch": 0.3582423733557235,
+      "grad_norm": 1.050110936164856,
+      "learning_rate": 4.2313019390581723e-05,
+      "loss": 0.4693,
+      "step": 800
+    },
+    {
+      "epoch": 0.3582423733557235,
+      "eval_loss": 0.5443281531333923,
+      "eval_runtime": 1730.4364,
+      "eval_samples_per_second": 2.581,
+      "eval_steps_per_second": 2.581,
+      "step": 800
+    },
+    {
+      "epoch": 0.3586901763224181,
+      "grad_norm": 1.0029529333114624,
+      "learning_rate": 4.2301477377654666e-05,
+      "loss": 0.6834,
+      "step": 801
+    },
+    {
+      "epoch": 0.3591379792891128,
+      "grad_norm": 0.8656796813011169,
+      "learning_rate": 4.228993536472761e-05,
+      "loss": 0.4182,
+      "step": 802
+    },
+    {
+      "epoch": 0.35958578225580745,
+      "grad_norm": 0.9622914791107178,
+      "learning_rate": 4.227839335180056e-05,
+      "loss": 0.5761,
+      "step": 803
+    },
+    {
+      "epoch": 0.3600335852225021,
+      "grad_norm": 1.291436791419983,
+      "learning_rate": 4.22668513388735e-05,
+      "loss": 0.7108,
+      "step": 804
+    },
+    {
+      "epoch": 0.3604813881891967,
+      "grad_norm": 1.2705053091049194,
+      "learning_rate": 4.225530932594644e-05,
+      "loss": 0.5405,
+      "step": 805
+    },
+    {
+      "epoch": 0.3609291911558914,
+      "grad_norm": 1.041524052619934,
+      "learning_rate": 4.224376731301939e-05,
+      "loss": 0.3357,
+      "step": 806
+    },
+    {
+      "epoch": 0.36137699412258606,
+      "grad_norm": 1.0177375078201294,
+      "learning_rate": 4.2232225300092335e-05,
+      "loss": 0.4982,
+      "step": 807
+    },
+    {
+      "epoch": 0.3618247970892807,
+      "grad_norm": 1.0229461193084717,
+      "learning_rate": 4.2220683287165285e-05,
+      "loss": 0.572,
+      "step": 808
+    },
+    {
+      "epoch": 0.3622726000559754,
+      "grad_norm": 1.037686824798584,
+      "learning_rate": 4.220914127423823e-05,
+      "loss": 0.7705,
+      "step": 809
+    },
+    {
+      "epoch": 0.36272040302267,
+      "grad_norm": 1.0016790628433228,
+      "learning_rate": 4.2197599261311177e-05,
+      "loss": 0.5448,
+      "step": 810
+    },
+    {
+      "epoch": 0.36272040302267,
+      "eval_loss": 0.537259042263031,
+      "eval_runtime": 1761.6249,
+      "eval_samples_per_second": 2.536,
+      "eval_steps_per_second": 2.536,
+      "step": 810
+    },
+    {
+      "epoch": 0.36316820598936467,
+      "grad_norm": 1.0614540576934814,
+      "learning_rate": 4.218605724838412e-05,
+      "loss": 0.5356,
+      "step": 811
+    },
+    {
+      "epoch": 0.36361600895605933,
+      "grad_norm": 0.9634736180305481,
+      "learning_rate": 4.217451523545707e-05,
+      "loss": 0.5571,
+      "step": 812
+    },
+    {
+      "epoch": 0.364063811922754,
+      "grad_norm": 1.0522750616073608,
+      "learning_rate": 4.216297322253001e-05,
+      "loss": 0.5121,
+      "step": 813
+    },
+    {
+      "epoch": 0.36451161488944867,
+      "grad_norm": 1.1789124011993408,
+      "learning_rate": 4.215143120960296e-05,
+      "loss": 0.5571,
+      "step": 814
+    },
+    {
+      "epoch": 0.3649594178561433,
+      "grad_norm": 1.1370149850845337,
+      "learning_rate": 4.21398891966759e-05,
+      "loss": 0.3999,
+      "step": 815
+    },
+    {
+      "epoch": 0.36540722082283794,
+      "grad_norm": 1.006212830543518,
+      "learning_rate": 4.2128347183748846e-05,
+      "loss": 0.5291,
+      "step": 816
+    },
+    {
+      "epoch": 0.3658550237895326,
+      "grad_norm": 1.1692012548446655,
+      "learning_rate": 4.2116805170821795e-05,
+      "loss": 0.53,
+      "step": 817
+    },
+    {
+      "epoch": 0.3663028267562273,
+      "grad_norm": 1.0599125623703003,
+      "learning_rate": 4.210526315789474e-05,
+      "loss": 0.8076,
+      "step": 818
+    },
+    {
+      "epoch": 0.36675062972292194,
+      "grad_norm": 1.2722653150558472,
+      "learning_rate": 4.209372114496769e-05,
+      "loss": 0.4796,
+      "step": 819
+    },
+    {
+      "epoch": 0.36719843268961655,
+      "grad_norm": 1.155556321144104,
+      "learning_rate": 4.208217913204063e-05,
+      "loss": 0.5004,
+      "step": 820
+    },
+    {
+      "epoch": 0.36719843268961655,
+      "eval_loss": 0.5317677855491638,
+      "eval_runtime": 1759.931,
+      "eval_samples_per_second": 2.538,
+      "eval_steps_per_second": 2.538,
+      "step": 820
+    },
+    {
+      "epoch": 0.3676462356563112,
+      "grad_norm": 1.2540476322174072,
+      "learning_rate": 4.207063711911358e-05,
+      "loss": 0.6489,
+      "step": 821
+    },
+    {
+      "epoch": 0.3680940386230059,
+      "grad_norm": 1.5235843658447266,
+      "learning_rate": 4.205909510618652e-05,
+      "loss": 0.457,
+      "step": 822
+    },
+    {
+      "epoch": 0.36854184158970055,
+      "grad_norm": 1.1602377891540527,
+      "learning_rate": 4.2047553093259464e-05,
+      "loss": 0.5357,
+      "step": 823
+    },
+    {
+      "epoch": 0.3689896445563952,
+      "grad_norm": 1.1831027269363403,
+      "learning_rate": 4.203601108033241e-05,
+      "loss": 0.5766,
+      "step": 824
+    },
+    {
+      "epoch": 0.3694374475230898,
+      "grad_norm": 1.0105196237564087,
+      "learning_rate": 4.2024469067405356e-05,
+      "loss": 0.4265,
+      "step": 825
+    },
+    {
+      "epoch": 0.3698852504897845,
+      "grad_norm": 1.036521315574646,
+      "learning_rate": 4.20129270544783e-05,
+      "loss": 0.5833,
+      "step": 826
+    },
+    {
+      "epoch": 0.37033305345647916,
+      "grad_norm": 1.0582512617111206,
+      "learning_rate": 4.200138504155125e-05,
+      "loss": 0.279,
+      "step": 827
+    },
+    {
+      "epoch": 0.3707808564231738,
+      "grad_norm": 1.1272261142730713,
+      "learning_rate": 4.198984302862419e-05,
+      "loss": 0.9891,
+      "step": 828
+    },
+    {
+      "epoch": 0.37122865938986843,
+      "grad_norm": 0.911460816860199,
+      "learning_rate": 4.197830101569714e-05,
+      "loss": 0.4668,
+      "step": 829
+    },
+    {
+      "epoch": 0.3716764623565631,
+      "grad_norm": 1.0097686052322388,
+      "learning_rate": 4.196675900277008e-05,
+      "loss": 0.4885,
+      "step": 830
+    },
+    {
+      "epoch": 0.3716764623565631,
+      "eval_loss": 0.5257967114448547,
+      "eval_runtime": 1764.0371,
+      "eval_samples_per_second": 2.532,
+      "eval_steps_per_second": 2.532,
+      "step": 830
+    },
+    {
+      "epoch": 0.37212426532325776,
+      "grad_norm": 1.0220969915390015,
+      "learning_rate": 4.195521698984303e-05,
+      "loss": 0.3719,
+      "step": 831
+    },
+    {
+      "epoch": 0.37257206828995243,
+      "grad_norm": 1.0850633382797241,
+      "learning_rate": 4.1943674976915975e-05,
+      "loss": 0.6425,
+      "step": 832
+    },
+    {
+      "epoch": 0.3730198712566471,
+      "grad_norm": 1.0111445188522339,
+      "learning_rate": 4.1932132963988924e-05,
+      "loss": 0.5395,
+      "step": 833
+    },
+    {
+      "epoch": 0.3734676742233417,
+      "grad_norm": 0.9761947989463806,
+      "learning_rate": 4.192059095106187e-05,
+      "loss": 0.427,
+      "step": 834
+    },
+    {
+      "epoch": 0.37391547719003637,
+      "grad_norm": 1.1134892702102661,
+      "learning_rate": 4.1909048938134816e-05,
+      "loss": 0.4008,
+      "step": 835
+    },
+    {
+      "epoch": 0.37436328015673104,
+      "grad_norm": 1.1790378093719482,
+      "learning_rate": 4.189750692520776e-05,
+      "loss": 0.709,
+      "step": 836
+    },
+    {
+      "epoch": 0.3748110831234257,
+      "grad_norm": 1.3236595392227173,
+      "learning_rate": 4.188596491228071e-05,
+      "loss": 0.5734,
+      "step": 837
+    },
+    {
+      "epoch": 0.37525888609012037,
+      "grad_norm": 1.2026864290237427,
+      "learning_rate": 4.187442289935365e-05,
+      "loss": 0.7704,
+      "step": 838
+    },
+    {
+      "epoch": 0.375706689056815,
+      "grad_norm": 1.0784732103347778,
+      "learning_rate": 4.18628808864266e-05,
+      "loss": 0.6095,
+      "step": 839
+    },
+    {
+      "epoch": 0.37615449202350965,
+      "grad_norm": 1.0638351440429688,
+      "learning_rate": 4.185133887349954e-05,
+      "loss": 0.8335,
+      "step": 840
+    },
+    {
+      "epoch": 0.37615449202350965,
+      "eval_loss": 0.5203561782836914,
+      "eval_runtime": 1759.4932,
+      "eval_samples_per_second": 2.539,
+      "eval_steps_per_second": 2.539,
+      "step": 840
+    },
+    {
+      "epoch": 0.3766022949902043,
+      "grad_norm": 1.3564544916152954,
+      "learning_rate": 4.1839796860572486e-05,
+      "loss": 0.5154,
+      "step": 841
+    },
+    {
+      "epoch": 0.377050097956899,
+      "grad_norm": 0.9023253917694092,
+      "learning_rate": 4.1828254847645435e-05,
+      "loss": 0.2059,
+      "step": 842
+    },
+    {
+      "epoch": 0.37749790092359364,
+      "grad_norm": 0.8635544180870056,
+      "learning_rate": 4.181671283471838e-05,
+      "loss": 0.3199,
+      "step": 843
+    },
+    {
+      "epoch": 0.37794570389028825,
+      "grad_norm": 1.0583134889602661,
+      "learning_rate": 4.180517082179132e-05,
+      "loss": 0.7493,
+      "step": 844
+    },
+    {
+      "epoch": 0.3783935068569829,
+      "grad_norm": 1.0581653118133545,
+      "learning_rate": 4.179362880886426e-05,
+      "loss": 0.8315,
+      "step": 845
+    },
+    {
+      "epoch": 0.3788413098236776,
+      "grad_norm": 1.022495150566101,
+      "learning_rate": 4.178208679593721e-05,
+      "loss": 0.5801,
+      "step": 846
+    },
+    {
+      "epoch": 0.37928911279037225,
+      "grad_norm": 1.0368661880493164,
+      "learning_rate": 4.1770544783010155e-05,
+      "loss": 0.4629,
+      "step": 847
+    },
+    {
+      "epoch": 0.3797369157570669,
+      "grad_norm": 1.0277600288391113,
+      "learning_rate": 4.1759002770083104e-05,
+      "loss": 0.4293,
+      "step": 848
+    },
+    {
+      "epoch": 0.38018471872376153,
+      "grad_norm": 0.9926353096961975,
+      "learning_rate": 4.174746075715605e-05,
+      "loss": 0.545,
+      "step": 849
+    },
+    {
+      "epoch": 0.3806325216904562,
+      "grad_norm": 1.0656540393829346,
+      "learning_rate": 4.1735918744228996e-05,
+      "loss": 0.4953,
+      "step": 850
+    },
+    {
+      "epoch": 0.3806325216904562,
+      "eval_loss": 0.5157405138015747,
+      "eval_runtime": 1765.5766,
+      "eval_samples_per_second": 2.53,
+      "eval_steps_per_second": 2.53,
+      "step": 850
+    },
+    {
+      "epoch": 0.38108032465715086,
+      "grad_norm": 1.0566133260726929,
+      "learning_rate": 4.172437673130194e-05,
+      "loss": 0.3186,
+      "step": 851
+    },
+    {
+      "epoch": 0.3815281276238455,
+      "grad_norm": 1.0668479204177856,
+      "learning_rate": 4.171283471837489e-05,
+      "loss": 0.8912,
+      "step": 852
+    },
+    {
+      "epoch": 0.38197593059054014,
+      "grad_norm": 0.9448925852775574,
+      "learning_rate": 4.170129270544783e-05,
+      "loss": 0.5759,
+      "step": 853
+    },
+    {
+      "epoch": 0.3824237335572348,
+      "grad_norm": 0.9699609279632568,
+      "learning_rate": 4.168975069252078e-05,
+      "loss": 0.7082,
+      "step": 854
+    },
+    {
+      "epoch": 0.38287153652392947,
+      "grad_norm": 1.1473195552825928,
+      "learning_rate": 4.167820867959372e-05,
+      "loss": 0.3057,
+      "step": 855
+    },
+    {
+      "epoch": 0.38331933949062413,
+      "grad_norm": 1.154795527458191,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.4783,
+      "step": 856
+    },
+    {
+      "epoch": 0.3837671424573188,
+      "grad_norm": 1.1641485691070557,
+      "learning_rate": 4.1655124653739615e-05,
+      "loss": 0.428,
+      "step": 857
+    },
+    {
+      "epoch": 0.3842149454240134,
+      "grad_norm": 1.0281574726104736,
+      "learning_rate": 4.1643582640812564e-05,
+      "loss": 0.4742,
+      "step": 858
+    },
+    {
+      "epoch": 0.3846627483907081,
+      "grad_norm": 1.2137088775634766,
+      "learning_rate": 4.163204062788551e-05,
+      "loss": 0.6097,
+      "step": 859
+    },
+    {
+      "epoch": 0.38511055135740274,
+      "grad_norm": 1.0084288120269775,
+      "learning_rate": 4.1620498614958456e-05,
+      "loss": 0.386,
+      "step": 860
+    },
+    {
+      "epoch": 0.38511055135740274,
+      "eval_loss": 0.5101390480995178,
+      "eval_runtime": 1766.2692,
+      "eval_samples_per_second": 2.529,
+      "eval_steps_per_second": 2.529,
+      "step": 860
+    },
+    {
+      "epoch": 0.3855583543240974,
+      "grad_norm": 1.200697898864746,
+      "learning_rate": 4.16089566020314e-05,
+      "loss": 0.5233,
+      "step": 861
+    },
+    {
+      "epoch": 0.3860061572907921,
+      "grad_norm": 1.0038084983825684,
+      "learning_rate": 4.159741458910434e-05,
+      "loss": 0.4321,
+      "step": 862
+    },
+    {
+      "epoch": 0.3864539602574867,
+      "grad_norm": 1.0555037260055542,
+      "learning_rate": 4.1585872576177284e-05,
+      "loss": 0.4842,
+      "step": 863
+    },
+    {
+      "epoch": 0.38690176322418135,
+      "grad_norm": 1.0753816366195679,
+      "learning_rate": 4.157433056325023e-05,
+      "loss": 0.5594,
+      "step": 864
+    },
+    {
+      "epoch": 0.387349566190876,
+      "grad_norm": 0.92066490650177,
+      "learning_rate": 4.1562788550323176e-05,
+      "loss": 0.4383,
+      "step": 865
+    },
+    {
+      "epoch": 0.3877973691575707,
+      "grad_norm": 1.1402944326400757,
+      "learning_rate": 4.155124653739612e-05,
+      "loss": 0.4339,
+      "step": 866
+    },
+    {
+      "epoch": 0.38824517212426535,
+      "grad_norm": 1.1352843046188354,
+      "learning_rate": 4.153970452446907e-05,
+      "loss": 0.6557,
+      "step": 867
+    },
+    {
+      "epoch": 0.38869297509095996,
+      "grad_norm": 1.0460140705108643,
+      "learning_rate": 4.152816251154201e-05,
+      "loss": 0.4145,
+      "step": 868
+    },
+    {
+      "epoch": 0.3891407780576546,
+      "grad_norm": 1.0888867378234863,
+      "learning_rate": 4.151662049861496e-05,
+      "loss": 0.6837,
+      "step": 869
+    },
+    {
+      "epoch": 0.3895885810243493,
+      "grad_norm": 1.1084412336349487,
+      "learning_rate": 4.15050784856879e-05,
+      "loss": 0.5287,
+      "step": 870
+    },
+    {
+      "epoch": 0.3895885810243493,
+      "eval_loss": 0.5020670294761658,
+      "eval_runtime": 1766.2603,
+      "eval_samples_per_second": 2.529,
+      "eval_steps_per_second": 2.529,
+      "step": 870
+    },
+    {
+      "epoch": 0.39003638399104396,
+      "grad_norm": 1.0915954113006592,
+      "learning_rate": 4.149353647276085e-05,
+      "loss": 0.48,
+      "step": 871
+    },
+    {
+      "epoch": 0.3904841869577386,
+      "grad_norm": 1.0187283754348755,
+      "learning_rate": 4.1481994459833794e-05,
+      "loss": 0.6489,
+      "step": 872
+    },
+    {
+      "epoch": 0.39093198992443323,
+      "grad_norm": 1.15608549118042,
+      "learning_rate": 4.1470452446906744e-05,
+      "loss": 0.6752,
+      "step": 873
+    },
+    {
+      "epoch": 0.3913797928911279,
+      "grad_norm": 1.2111482620239258,
+      "learning_rate": 4.1458910433979686e-05,
+      "loss": 0.5118,
+      "step": 874
+    },
+    {
+      "epoch": 0.39182759585782256,
+      "grad_norm": 0.9627257585525513,
+      "learning_rate": 4.1447368421052636e-05,
+      "loss": 0.4093,
+      "step": 875
+    },
+    {
+      "epoch": 0.39227539882451723,
+      "grad_norm": 0.8828468918800354,
+      "learning_rate": 4.143582640812558e-05,
+      "loss": 0.5358,
+      "step": 876
+    },
+    {
+      "epoch": 0.39272320179121184,
+      "grad_norm": 0.963966429233551,
+      "learning_rate": 4.142428439519853e-05,
+      "loss": 0.8033,
+      "step": 877
+    },
+    {
+      "epoch": 0.3931710047579065,
+      "grad_norm": 0.9207846522331238,
+      "learning_rate": 4.141274238227147e-05,
+      "loss": 0.2906,
+      "step": 878
+    },
+    {
+      "epoch": 0.39361880772460117,
+      "grad_norm": 1.082098126411438,
+      "learning_rate": 4.140120036934442e-05,
+      "loss": 0.2977,
+      "step": 879
+    },
+    {
+      "epoch": 0.39406661069129584,
+      "grad_norm": 1.2937519550323486,
+      "learning_rate": 4.138965835641736e-05,
+      "loss": 0.407,
+      "step": 880
+    },
+    {
+      "epoch": 0.39406661069129584,
+      "eval_loss": 0.49985942244529724,
+      "eval_runtime": 1767.4919,
+      "eval_samples_per_second": 2.527,
+      "eval_steps_per_second": 2.527,
+      "step": 880
+    },
+    {
+      "epoch": 0.3945144136579905,
+      "grad_norm": 1.326373815536499,
+      "learning_rate": 4.137811634349031e-05,
+      "loss": 0.4247,
+      "step": 881
+    },
+    {
+      "epoch": 0.3949622166246851,
+      "grad_norm": 1.042016625404358,
+      "learning_rate": 4.1366574330563254e-05,
+      "loss": 0.5957,
+      "step": 882
+    },
+    {
+      "epoch": 0.3954100195913798,
+      "grad_norm": 1.083022952079773,
+      "learning_rate": 4.13550323176362e-05,
+      "loss": 0.5244,
+      "step": 883
+    },
+    {
+      "epoch": 0.39585782255807445,
+      "grad_norm": 1.174361228942871,
+      "learning_rate": 4.134349030470914e-05,
+      "loss": 0.8584,
+      "step": 884
+    },
+    {
+      "epoch": 0.3963056255247691,
+      "grad_norm": 1.1042940616607666,
+      "learning_rate": 4.133194829178209e-05,
+      "loss": 0.2712,
+      "step": 885
+    },
+    {
+      "epoch": 0.3967534284914638,
+      "grad_norm": 1.1315444707870483,
+      "learning_rate": 4.132040627885503e-05,
+      "loss": 0.5922,
+      "step": 886
+    },
+    {
+      "epoch": 0.3972012314581584,
+      "grad_norm": 1.0596696138381958,
+      "learning_rate": 4.130886426592798e-05,
+      "loss": 0.3525,
+      "step": 887
+    },
+    {
+      "epoch": 0.39764903442485305,
+      "grad_norm": 1.0386688709259033,
+      "learning_rate": 4.1297322253000924e-05,
+      "loss": 0.7037,
+      "step": 888
+    },
+    {
+      "epoch": 0.3980968373915477,
+      "grad_norm": 1.3697748184204102,
+      "learning_rate": 4.128578024007387e-05,
+      "loss": 0.3933,
+      "step": 889
+    },
+    {
+      "epoch": 0.3985446403582424,
+      "grad_norm": 1.1484674215316772,
+      "learning_rate": 4.1274238227146816e-05,
+      "loss": 0.4183,
+      "step": 890
+    },
+    {
+      "epoch": 0.3985446403582424,
+      "eval_loss": 0.4953237771987915,
+      "eval_runtime": 1762.0817,
+      "eval_samples_per_second": 2.535,
+      "eval_steps_per_second": 2.535,
+      "step": 890
+    },
+    {
+      "epoch": 0.39899244332493705,
+      "grad_norm": 1.1704154014587402,
+      "learning_rate": 4.126269621421976e-05,
+      "loss": 0.565,
+      "step": 891
+    },
+    {
+      "epoch": 0.39944024629163166,
+      "grad_norm": 1.0532429218292236,
+      "learning_rate": 4.125115420129271e-05,
+      "loss": 0.442,
+      "step": 892
+    },
+    {
+      "epoch": 0.39988804925832633,
+      "grad_norm": 1.23839271068573,
+      "learning_rate": 4.123961218836565e-05,
+      "loss": 0.6296,
+      "step": 893
+    },
+    {
+      "epoch": 0.400335852225021,
+      "grad_norm": 1.184489369392395,
+      "learning_rate": 4.12280701754386e-05,
+      "loss": 0.4153,
+      "step": 894
+    },
+    {
+      "epoch": 0.40078365519171566,
+      "grad_norm": 0.9506192803382874,
+      "learning_rate": 4.121652816251154e-05,
+      "loss": 0.4286,
+      "step": 895
+    },
+    {
+      "epoch": 0.4012314581584103,
+      "grad_norm": 1.1313607692718506,
+      "learning_rate": 4.120498614958449e-05,
+      "loss": 0.5289,
+      "step": 896
+    },
+    {
+      "epoch": 0.40167926112510494,
+      "grad_norm": 1.0801643133163452,
+      "learning_rate": 4.1193444136657434e-05,
+      "loss": 0.6045,
+      "step": 897
+    },
+    {
+      "epoch": 0.4021270640917996,
+      "grad_norm": 1.1855746507644653,
+      "learning_rate": 4.1181902123730384e-05,
+      "loss": 0.5275,
+      "step": 898
+    },
+    {
+      "epoch": 0.40257486705849427,
+      "grad_norm": 0.948414146900177,
+      "learning_rate": 4.1170360110803326e-05,
+      "loss": 0.5052,
+      "step": 899
+    },
+    {
+      "epoch": 0.40302267002518893,
+      "grad_norm": 0.9570877552032471,
+      "learning_rate": 4.1158818097876276e-05,
+      "loss": 0.3068,
+      "step": 900
+    },
+    {
+      "epoch": 0.40302267002518893,
+      "eval_loss": 0.48936161398887634,
+      "eval_runtime": 1738.8904,
+      "eval_samples_per_second": 2.569,
+      "eval_steps_per_second": 2.569,
+      "step": 900
+    },
+    {
+      "epoch": 0.40347047299188354,
+      "grad_norm": 1.0901323556900024,
+      "learning_rate": 4.114727608494922e-05,
+      "loss": 0.5199,
+      "step": 901
+    },
+    {
+      "epoch": 0.4039182759585782,
+      "grad_norm": 1.0945810079574585,
+      "learning_rate": 4.113573407202216e-05,
+      "loss": 0.4492,
+      "step": 902
+    },
+    {
+      "epoch": 0.4043660789252729,
+      "grad_norm": 1.1015689373016357,
+      "learning_rate": 4.112419205909511e-05,
+      "loss": 0.39,
+      "step": 903
+    },
+    {
+      "epoch": 0.40481388189196754,
+      "grad_norm": 1.30820631980896,
+      "learning_rate": 4.111265004616805e-05,
+      "loss": 0.5595,
+      "step": 904
+    },
+    {
+      "epoch": 0.4052616848586622,
+      "grad_norm": 1.3023115396499634,
+      "learning_rate": 4.1101108033240995e-05,
+      "loss": 0.7433,
+      "step": 905
+    },
+    {
+      "epoch": 0.4057094878253568,
+      "grad_norm": 1.2857636213302612,
+      "learning_rate": 4.1089566020313945e-05,
+      "loss": 0.7139,
+      "step": 906
+    },
+    {
+      "epoch": 0.4061572907920515,
+      "grad_norm": 1.0232025384902954,
+      "learning_rate": 4.107802400738689e-05,
+      "loss": 0.345,
+      "step": 907
+    },
+    {
+      "epoch": 0.40660509375874615,
+      "grad_norm": 1.0496271848678589,
+      "learning_rate": 4.106648199445984e-05,
+      "loss": 0.2903,
+      "step": 908
+    },
+    {
+      "epoch": 0.4070528967254408,
+      "grad_norm": 1.170128345489502,
+      "learning_rate": 4.105493998153278e-05,
+      "loss": 0.5301,
+      "step": 909
+    },
+    {
+      "epoch": 0.4075006996921355,
+      "grad_norm": 0.9690027832984924,
+      "learning_rate": 4.104339796860573e-05,
+      "loss": 0.5454,
+      "step": 910
+    },
+    {
+      "epoch": 0.4075006996921355,
+      "eval_loss": 0.48335975408554077,
+      "eval_runtime": 1729.8426,
+      "eval_samples_per_second": 2.582,
+      "eval_steps_per_second": 2.582,
+      "step": 910
+    },
+    {
+      "epoch": 0.4079485026588301,
+      "grad_norm": 1.0750856399536133,
+      "learning_rate": 4.103185595567867e-05,
+      "loss": 0.396,
+      "step": 911
+    },
+    {
+      "epoch": 0.40839630562552476,
+      "grad_norm": 0.9750925302505493,
+      "learning_rate": 4.102031394275162e-05,
+      "loss": 0.3567,
+      "step": 912
+    },
+    {
+      "epoch": 0.4088441085922194,
+      "grad_norm": 1.2686316967010498,
+      "learning_rate": 4.100877192982456e-05,
+      "loss": 0.4793,
+      "step": 913
+    },
+    {
+      "epoch": 0.4092919115589141,
+      "grad_norm": 1.0663954019546509,
+      "learning_rate": 4.099722991689751e-05,
+      "loss": 0.7291,
+      "step": 914
+    },
+    {
+      "epoch": 0.40973971452560876,
+      "grad_norm": 1.1266555786132812,
+      "learning_rate": 4.0985687903970455e-05,
+      "loss": 0.729,
+      "step": 915
+    },
+    {
+      "epoch": 0.41018751749230337,
+      "grad_norm": 1.218075156211853,
+      "learning_rate": 4.09741458910434e-05,
+      "loss": 0.6157,
+      "step": 916
+    },
+    {
+      "epoch": 0.41063532045899803,
+      "grad_norm": 1.142625093460083,
+      "learning_rate": 4.096260387811635e-05,
+      "loss": 0.5192,
+      "step": 917
+    },
+    {
+      "epoch": 0.4110831234256927,
+      "grad_norm": 1.0120737552642822,
+      "learning_rate": 4.095106186518929e-05,
+      "loss": 0.5094,
+      "step": 918
+    },
+    {
+      "epoch": 0.41153092639238736,
+      "grad_norm": 1.1946959495544434,
+      "learning_rate": 4.093951985226224e-05,
+      "loss": 0.4118,
+      "step": 919
+    },
+    {
+      "epoch": 0.41197872935908203,
+      "grad_norm": 0.9003228545188904,
+      "learning_rate": 4.092797783933518e-05,
+      "loss": 0.3265,
+      "step": 920
+    },
+    {
+      "epoch": 0.41197872935908203,
+      "eval_loss": 0.477049857378006,
+      "eval_runtime": 1730.2627,
+      "eval_samples_per_second": 2.582,
+      "eval_steps_per_second": 2.582,
+      "step": 920
+    },
+    {
+      "epoch": 0.41242653232577664,
+      "grad_norm": 1.0392440557479858,
+      "learning_rate": 4.091643582640813e-05,
+      "loss": 0.3326,
+      "step": 921
+    },
+    {
+      "epoch": 0.4128743352924713,
+      "grad_norm": 1.0120160579681396,
+      "learning_rate": 4.0904893813481074e-05,
+      "loss": 0.4699,
+      "step": 922
+    },
+    {
+      "epoch": 0.41332213825916597,
+      "grad_norm": 1.2628374099731445,
+      "learning_rate": 4.0893351800554016e-05,
+      "loss": 0.455,
+      "step": 923
+    },
+    {
+      "epoch": 0.41376994122586064,
+      "grad_norm": 1.1514098644256592,
+      "learning_rate": 4.088180978762696e-05,
+      "loss": 0.4138,
+      "step": 924
+    },
+    {
+      "epoch": 0.41421774419255525,
+      "grad_norm": 1.059490442276001,
+      "learning_rate": 4.087026777469991e-05,
+      "loss": 0.3612,
+      "step": 925
+    },
+    {
+      "epoch": 0.4146655471592499,
+      "grad_norm": 0.92525714635849,
+      "learning_rate": 4.085872576177285e-05,
+      "loss": 0.4365,
+      "step": 926
+    },
+    {
+      "epoch": 0.4151133501259446,
+      "grad_norm": 1.0232129096984863,
+      "learning_rate": 4.08471837488458e-05,
+      "loss": 0.3875,
+      "step": 927
+    },
+    {
+      "epoch": 0.41556115309263925,
+      "grad_norm": 1.272973656654358,
+      "learning_rate": 4.083564173591874e-05,
+      "loss": 0.3881,
+      "step": 928
+    },
+    {
+      "epoch": 0.4160089560593339,
+      "grad_norm": 1.49363374710083,
+      "learning_rate": 4.082409972299169e-05,
+      "loss": 0.6014,
+      "step": 929
+    },
+    {
+      "epoch": 0.4164567590260285,
+      "grad_norm": 1.0534272193908691,
+      "learning_rate": 4.0812557710064635e-05,
+      "loss": 0.339,
+      "step": 930
+    },
+    {
+      "epoch": 0.4164567590260285,
+      "eval_loss": 0.4716756343841553,
+      "eval_runtime": 1730.6307,
+      "eval_samples_per_second": 2.581,
+      "eval_steps_per_second": 2.581,
+      "step": 930
+    },
+    {
+      "epoch": 0.4169045619927232,
+      "grad_norm": 1.0245121717453003,
+      "learning_rate": 4.0801015697137584e-05,
+      "loss": 0.7959,
+      "step": 931
+    },
+    {
+      "epoch": 0.41735236495941785,
+      "grad_norm": 1.0729509592056274,
+      "learning_rate": 4.078947368421053e-05,
+      "loss": 0.3776,
+      "step": 932
+    },
+    {
+      "epoch": 0.4178001679261125,
+      "grad_norm": 1.1630797386169434,
+      "learning_rate": 4.0777931671283476e-05,
+      "loss": 0.5052,
+      "step": 933
+    },
+    {
+      "epoch": 0.4182479708928072,
+      "grad_norm": 0.955295205116272,
+      "learning_rate": 4.076638965835642e-05,
+      "loss": 0.3525,
+      "step": 934
+    },
+    {
+      "epoch": 0.4186957738595018,
+      "grad_norm": 1.0843642950057983,
+      "learning_rate": 4.075484764542937e-05,
+      "loss": 0.4731,
+      "step": 935
+    },
+    {
+      "epoch": 0.41914357682619646,
+      "grad_norm": 0.9524621367454529,
+      "learning_rate": 4.074330563250231e-05,
+      "loss": 0.3274,
+      "step": 936
+    },
+    {
+      "epoch": 0.41959137979289113,
+      "grad_norm": 1.1125507354736328,
+      "learning_rate": 4.073176361957526e-05,
+      "loss": 0.4461,
+      "step": 937
+    },
+    {
+      "epoch": 0.4200391827595858,
+      "grad_norm": 1.0988569259643555,
+      "learning_rate": 4.07202216066482e-05,
+      "loss": 0.3417,
+      "step": 938
+    },
+    {
+      "epoch": 0.42048698572628046,
+      "grad_norm": 1.270066261291504,
+      "learning_rate": 4.070867959372115e-05,
+      "loss": 0.5205,
+      "step": 939
+    },
+    {
+      "epoch": 0.42093478869297507,
+      "grad_norm": 1.0238381624221802,
+      "learning_rate": 4.0697137580794095e-05,
+      "loss": 0.4402,
+      "step": 940
+    },
+    {
+      "epoch": 0.42093478869297507,
+      "eval_loss": 0.4671795666217804,
+      "eval_runtime": 1732.843,
+      "eval_samples_per_second": 2.578,
+      "eval_steps_per_second": 2.578,
+      "step": 940
+    },
+    {
+      "epoch": 0.42138259165966974,
+      "grad_norm": 0.8779534697532654,
+      "learning_rate": 4.068559556786704e-05,
+      "loss": 0.3124,
+      "step": 941
+    },
+    {
+      "epoch": 0.4218303946263644,
+      "grad_norm": 1.201365351676941,
+      "learning_rate": 4.067405355493999e-05,
+      "loss": 0.2708,
+      "step": 942
+    },
+    {
+      "epoch": 0.42227819759305907,
+      "grad_norm": 1.1746056079864502,
+      "learning_rate": 4.066251154201293e-05,
+      "loss": 0.3772,
+      "step": 943
+    },
+    {
+      "epoch": 0.42272600055975373,
+      "grad_norm": 1.3690954446792603,
+      "learning_rate": 4.065096952908587e-05,
+      "loss": 0.569,
+      "step": 944
+    },
+    {
+      "epoch": 0.42317380352644834,
+      "grad_norm": 1.2674678564071655,
+      "learning_rate": 4.0639427516158815e-05,
+      "loss": 0.3943,
+      "step": 945
+    },
+    {
+      "epoch": 0.423621606493143,
+      "grad_norm": 1.16683030128479,
+      "learning_rate": 4.0627885503231764e-05,
+      "loss": 0.5455,
+      "step": 946
+    },
+    {
+      "epoch": 0.4240694094598377,
+      "grad_norm": 1.226865291595459,
+      "learning_rate": 4.061634349030471e-05,
+      "loss": 0.4399,
+      "step": 947
+    },
+    {
+      "epoch": 0.42451721242653234,
+      "grad_norm": 0.9444255232810974,
+      "learning_rate": 4.0604801477377656e-05,
+      "loss": 0.5685,
+      "step": 948
+    },
+    {
+      "epoch": 0.42496501539322695,
+      "grad_norm": 1.2744015455245972,
+      "learning_rate": 4.05932594644506e-05,
+      "loss": 0.7167,
+      "step": 949
+    },
+    {
+      "epoch": 0.4254128183599216,
+      "grad_norm": 0.9263292551040649,
+      "learning_rate": 4.058171745152355e-05,
+      "loss": 0.2135,
+      "step": 950
+    },
+    {
+      "epoch": 0.4254128183599216,
+      "eval_loss": 0.46275782585144043,
+      "eval_runtime": 1737.1967,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 950
+    },
+    {
+      "epoch": 0.4258606213266163,
+      "grad_norm": 1.1970070600509644,
+      "learning_rate": 4.057017543859649e-05,
+      "loss": 0.4788,
+      "step": 951
+    },
+    {
+      "epoch": 0.42630842429331095,
+      "grad_norm": 1.0885076522827148,
+      "learning_rate": 4.055863342566944e-05,
+      "loss": 0.1885,
+      "step": 952
+    },
+    {
+      "epoch": 0.4267562272600056,
+      "grad_norm": 1.155806064605713,
+      "learning_rate": 4.054709141274238e-05,
+      "loss": 0.3639,
+      "step": 953
+    },
+    {
+      "epoch": 0.4272040302267002,
+      "grad_norm": 0.9880549311637878,
+      "learning_rate": 4.053554939981533e-05,
+      "loss": 0.4937,
+      "step": 954
+    },
+    {
+      "epoch": 0.4276518331933949,
+      "grad_norm": 1.2488168478012085,
+      "learning_rate": 4.0524007386888275e-05,
+      "loss": 0.3599,
+      "step": 955
+    },
+    {
+      "epoch": 0.42809963616008956,
+      "grad_norm": 1.1837854385375977,
+      "learning_rate": 4.0512465373961224e-05,
+      "loss": 0.4442,
+      "step": 956
+    },
+    {
+      "epoch": 0.4285474391267842,
+      "grad_norm": 1.3064805269241333,
+      "learning_rate": 4.050092336103417e-05,
+      "loss": 0.6201,
+      "step": 957
+    },
+    {
+      "epoch": 0.4289952420934789,
+      "grad_norm": 1.2136608362197876,
+      "learning_rate": 4.0489381348107116e-05,
+      "loss": 0.6388,
+      "step": 958
+    },
+    {
+      "epoch": 0.4294430450601735,
+      "grad_norm": 1.073799729347229,
+      "learning_rate": 4.047783933518006e-05,
+      "loss": 0.5717,
+      "step": 959
+    },
+    {
+      "epoch": 0.42989084802686817,
+      "grad_norm": 1.0983150005340576,
+      "learning_rate": 4.046629732225301e-05,
+      "loss": 0.6641,
+      "step": 960
+    },
+    {
+      "epoch": 0.42989084802686817,
+      "eval_loss": 0.457937628030777,
+      "eval_runtime": 1737.7661,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 960
+    },
+    {
+      "epoch": 0.43033865099356283,
+      "grad_norm": 1.405277132987976,
+      "learning_rate": 4.045475530932595e-05,
+      "loss": 0.5287,
+      "step": 961
+    },
+    {
+      "epoch": 0.4307864539602575,
+      "grad_norm": 0.9269359111785889,
+      "learning_rate": 4.044321329639889e-05,
+      "loss": 0.2857,
+      "step": 962
+    },
+    {
+      "epoch": 0.43123425692695216,
+      "grad_norm": 1.2387222051620483,
+      "learning_rate": 4.0431671283471836e-05,
+      "loss": 0.4877,
+      "step": 963
+    },
+    {
+      "epoch": 0.4316820598936468,
+      "grad_norm": 0.8296558856964111,
+      "learning_rate": 4.0420129270544785e-05,
+      "loss": 0.302,
+      "step": 964
+    },
+    {
+      "epoch": 0.43212986286034144,
+      "grad_norm": 1.1528252363204956,
+      "learning_rate": 4.040858725761773e-05,
+      "loss": 0.3738,
+      "step": 965
+    },
+    {
+      "epoch": 0.4325776658270361,
+      "grad_norm": 1.3704917430877686,
+      "learning_rate": 4.039704524469067e-05,
+      "loss": 0.5414,
+      "step": 966
+    },
+    {
+      "epoch": 0.4330254687937308,
+      "grad_norm": 1.276453971862793,
+      "learning_rate": 4.038550323176362e-05,
+      "loss": 0.6736,
+      "step": 967
+    },
+    {
+      "epoch": 0.43347327176042544,
+      "grad_norm": 0.9381370544433594,
+      "learning_rate": 4.037396121883656e-05,
+      "loss": 0.4799,
+      "step": 968
+    },
+    {
+      "epoch": 0.43392107472712005,
+      "grad_norm": 1.047900915145874,
+      "learning_rate": 4.036241920590951e-05,
+      "loss": 0.4303,
+      "step": 969
+    },
+    {
+      "epoch": 0.4343688776938147,
+      "grad_norm": 1.3831329345703125,
+      "learning_rate": 4.0350877192982455e-05,
+      "loss": 0.4318,
+      "step": 970
+    },
+    {
+      "epoch": 0.4343688776938147,
+      "eval_loss": 0.4529132843017578,
+      "eval_runtime": 1743.0645,
+      "eval_samples_per_second": 2.563,
+      "eval_steps_per_second": 2.563,
+      "step": 970
+    },
+    {
+      "epoch": 0.4348166806605094,
+      "grad_norm": 1.0554351806640625,
+      "learning_rate": 4.0339335180055404e-05,
+      "loss": 0.3476,
+      "step": 971
+    },
+    {
+      "epoch": 0.43526448362720405,
+      "grad_norm": 1.0319788455963135,
+      "learning_rate": 4.0327793167128346e-05,
+      "loss": 0.4205,
+      "step": 972
+    },
+    {
+      "epoch": 0.4357122865938987,
+      "grad_norm": 1.0676701068878174,
+      "learning_rate": 4.0316251154201296e-05,
+      "loss": 0.4594,
+      "step": 973
+    },
+    {
+      "epoch": 0.4361600895605933,
+      "grad_norm": 1.3130353689193726,
+      "learning_rate": 4.030470914127424e-05,
+      "loss": 0.5385,
+      "step": 974
+    },
+    {
+      "epoch": 0.436607892527288,
+      "grad_norm": 1.171388030052185,
+      "learning_rate": 4.029316712834719e-05,
+      "loss": 0.3008,
+      "step": 975
+    },
+    {
+      "epoch": 0.43705569549398265,
+      "grad_norm": 1.1871421337127686,
+      "learning_rate": 4.028162511542013e-05,
+      "loss": 0.518,
+      "step": 976
+    },
+    {
+      "epoch": 0.4375034984606773,
+      "grad_norm": 1.2678828239440918,
+      "learning_rate": 4.027008310249308e-05,
+      "loss": 0.4653,
+      "step": 977
+    },
+    {
+      "epoch": 0.43795130142737193,
+      "grad_norm": 1.2946516275405884,
+      "learning_rate": 4.025854108956602e-05,
+      "loss": 0.4609,
+      "step": 978
+    },
+    {
+      "epoch": 0.4383991043940666,
+      "grad_norm": 1.4077446460723877,
+      "learning_rate": 4.024699907663897e-05,
+      "loss": 0.7787,
+      "step": 979
+    },
+    {
+      "epoch": 0.43884690736076126,
+      "grad_norm": 1.2191566228866577,
+      "learning_rate": 4.0235457063711914e-05,
+      "loss": 0.4084,
+      "step": 980
+    },
+    {
+      "epoch": 0.43884690736076126,
+      "eval_loss": 0.4486415684223175,
+      "eval_runtime": 1746.9057,
+      "eval_samples_per_second": 2.557,
+      "eval_steps_per_second": 2.557,
+      "step": 980
+    },
+    {
+      "epoch": 0.43929471032745593,
+      "grad_norm": 1.2254949808120728,
+      "learning_rate": 4.0223915050784864e-05,
+      "loss": 0.3569,
+      "step": 981
+    },
+    {
+      "epoch": 0.4397425132941506,
+      "grad_norm": 0.9401348233222961,
+      "learning_rate": 4.0212373037857806e-05,
+      "loss": 0.3883,
+      "step": 982
+    },
+    {
+      "epoch": 0.4401903162608452,
+      "grad_norm": 1.2795180082321167,
+      "learning_rate": 4.020083102493075e-05,
+      "loss": 0.4951,
+      "step": 983
+    },
+    {
+      "epoch": 0.44063811922753987,
+      "grad_norm": 1.2436785697937012,
+      "learning_rate": 4.018928901200369e-05,
+      "loss": 0.4338,
+      "step": 984
+    },
+    {
+      "epoch": 0.44108592219423454,
+      "grad_norm": 1.1067421436309814,
+      "learning_rate": 4.017774699907664e-05,
+      "loss": 0.4877,
+      "step": 985
+    },
+    {
+      "epoch": 0.4415337251609292,
+      "grad_norm": 1.0319819450378418,
+      "learning_rate": 4.0166204986149584e-05,
+      "loss": 0.3879,
+      "step": 986
+    },
+    {
+      "epoch": 0.44198152812762387,
+      "grad_norm": 1.2362407445907593,
+      "learning_rate": 4.015466297322253e-05,
+      "loss": 0.6812,
+      "step": 987
+    },
+    {
+      "epoch": 0.4424293310943185,
+      "grad_norm": 1.3658016920089722,
+      "learning_rate": 4.0143120960295476e-05,
+      "loss": 0.5854,
+      "step": 988
+    },
+    {
+      "epoch": 0.44287713406101314,
+      "grad_norm": 1.0967875719070435,
+      "learning_rate": 4.0131578947368425e-05,
+      "loss": 0.4034,
+      "step": 989
+    },
+    {
+      "epoch": 0.4433249370277078,
+      "grad_norm": 1.025646686553955,
+      "learning_rate": 4.012003693444137e-05,
+      "loss": 0.2335,
+      "step": 990
+    },
+    {
+      "epoch": 0.4433249370277078,
+      "eval_loss": 0.4429576098918915,
+      "eval_runtime": 1743.9799,
+      "eval_samples_per_second": 2.561,
+      "eval_steps_per_second": 2.561,
+      "step": 990
+    },
+    {
+      "epoch": 0.4437727399944025,
+      "grad_norm": 1.1534349918365479,
+      "learning_rate": 4.010849492151431e-05,
+      "loss": 0.4779,
+      "step": 991
+    },
+    {
+      "epoch": 0.44422054296109714,
+      "grad_norm": 1.3908884525299072,
+      "learning_rate": 4.009695290858726e-05,
+      "loss": 0.4558,
+      "step": 992
+    },
+    {
+      "epoch": 0.44466834592779175,
+      "grad_norm": 1.4888511896133423,
+      "learning_rate": 4.00854108956602e-05,
+      "loss": 0.5484,
+      "step": 993
+    },
+    {
+      "epoch": 0.4451161488944864,
+      "grad_norm": 1.1760286092758179,
+      "learning_rate": 4.007386888273315e-05,
+      "loss": 0.3029,
+      "step": 994
+    },
+    {
+      "epoch": 0.4455639518611811,
+      "grad_norm": 1.3324183225631714,
+      "learning_rate": 4.0062326869806094e-05,
+      "loss": 0.2873,
+      "step": 995
+    },
+    {
+      "epoch": 0.44601175482787575,
+      "grad_norm": 1.3383257389068604,
+      "learning_rate": 4.0050784856879044e-05,
+      "loss": 0.4015,
+      "step": 996
+    },
+    {
+      "epoch": 0.4464595577945704,
+      "grad_norm": 1.0451068878173828,
+      "learning_rate": 4.0039242843951986e-05,
+      "loss": 0.3162,
+      "step": 997
+    },
+    {
+      "epoch": 0.446907360761265,
+      "grad_norm": 1.1706875562667847,
+      "learning_rate": 4.0027700831024936e-05,
+      "loss": 0.5261,
+      "step": 998
+    },
+    {
+      "epoch": 0.4473551637279597,
+      "grad_norm": 1.3044228553771973,
+      "learning_rate": 4.001615881809788e-05,
+      "loss": 0.4846,
+      "step": 999
+    },
+    {
+      "epoch": 0.44780296669465436,
+      "grad_norm": 1.2698777914047241,
+      "learning_rate": 4.000461680517083e-05,
+      "loss": 0.3915,
+      "step": 1000
+    },
+    {
+      "epoch": 0.44780296669465436,
+      "eval_loss": 0.43732336163520813,
+      "eval_runtime": 1742.5158,
+      "eval_samples_per_second": 2.564,
+      "eval_steps_per_second": 2.564,
+      "step": 1000
+    },
+    {
+      "epoch": 0.448250769661349,
+      "grad_norm": 1.1517488956451416,
+      "learning_rate": 3.999307479224377e-05,
+      "loss": 0.4172,
+      "step": 1001
+    },
+    {
+      "epoch": 0.44869857262804363,
+      "grad_norm": 0.9378765821456909,
+      "learning_rate": 3.998153277931671e-05,
+      "loss": 0.3321,
+      "step": 1002
+    },
+    {
+      "epoch": 0.4491463755947383,
+      "grad_norm": 1.2423738241195679,
+      "learning_rate": 3.996999076638966e-05,
+      "loss": 0.3526,
+      "step": 1003
+    },
+    {
+      "epoch": 0.44959417856143297,
+      "grad_norm": 1.476121187210083,
+      "learning_rate": 3.9958448753462605e-05,
+      "loss": 0.4987,
+      "step": 1004
+    },
+    {
+      "epoch": 0.45004198152812763,
+      "grad_norm": 1.0502073764801025,
+      "learning_rate": 3.994690674053555e-05,
+      "loss": 0.351,
+      "step": 1005
+    },
+    {
+      "epoch": 0.4504897844948223,
+      "grad_norm": 1.0767852067947388,
+      "learning_rate": 3.99353647276085e-05,
+      "loss": 0.4662,
+      "step": 1006
+    },
+    {
+      "epoch": 0.4509375874615169,
+      "grad_norm": 1.2034271955490112,
+      "learning_rate": 3.992382271468144e-05,
+      "loss": 0.5037,
+      "step": 1007
+    },
+    {
+      "epoch": 0.4513853904282116,
+      "grad_norm": 0.9691647887229919,
+      "learning_rate": 3.991228070175439e-05,
+      "loss": 0.3718,
+      "step": 1008
+    },
+    {
+      "epoch": 0.45183319339490624,
+      "grad_norm": 1.139522910118103,
+      "learning_rate": 3.990073868882733e-05,
+      "loss": 0.5095,
+      "step": 1009
+    },
+    {
+      "epoch": 0.4522809963616009,
+      "grad_norm": 1.2414535284042358,
+      "learning_rate": 3.988919667590028e-05,
+      "loss": 0.4806,
+      "step": 1010
+    },
+    {
+      "epoch": 0.4522809963616009,
+      "eval_loss": 0.43430060148239136,
+      "eval_runtime": 1742.0984,
+      "eval_samples_per_second": 2.564,
+      "eval_steps_per_second": 2.564,
+      "step": 1010
+    },
+    {
+      "epoch": 0.4527287993282956,
+      "grad_norm": 1.1482878923416138,
+      "learning_rate": 3.987765466297322e-05,
+      "loss": 0.2714,
+      "step": 1011
+    },
+    {
+      "epoch": 0.4531766022949902,
+      "grad_norm": 1.3171560764312744,
+      "learning_rate": 3.986611265004617e-05,
+      "loss": 0.4045,
+      "step": 1012
+    },
+    {
+      "epoch": 0.45362440526168485,
+      "grad_norm": 1.0673166513442993,
+      "learning_rate": 3.9854570637119115e-05,
+      "loss": 0.5288,
+      "step": 1013
+    },
+    {
+      "epoch": 0.4540722082283795,
+      "grad_norm": 1.1590580940246582,
+      "learning_rate": 3.9843028624192065e-05,
+      "loss": 0.4681,
+      "step": 1014
+    },
+    {
+      "epoch": 0.4545200111950742,
+      "grad_norm": 1.1945033073425293,
+      "learning_rate": 3.983148661126501e-05,
+      "loss": 0.3764,
+      "step": 1015
+    },
+    {
+      "epoch": 0.45496781416176885,
+      "grad_norm": 1.2974134683609009,
+      "learning_rate": 3.981994459833795e-05,
+      "loss": 0.3517,
+      "step": 1016
+    },
+    {
+      "epoch": 0.45541561712846346,
+      "grad_norm": 1.2538046836853027,
+      "learning_rate": 3.98084025854109e-05,
+      "loss": 0.5615,
+      "step": 1017
+    },
+    {
+      "epoch": 0.4558634200951581,
+      "grad_norm": 1.0046873092651367,
+      "learning_rate": 3.979686057248384e-05,
+      "loss": 0.2909,
+      "step": 1018
+    },
+    {
+      "epoch": 0.4563112230618528,
+      "grad_norm": 1.049226999282837,
+      "learning_rate": 3.978531855955679e-05,
+      "loss": 0.4252,
+      "step": 1019
+    },
+    {
+      "epoch": 0.45675902602854745,
+      "grad_norm": 1.1835418939590454,
+      "learning_rate": 3.9773776546629734e-05,
+      "loss": 0.3724,
+      "step": 1020
+    },
+    {
+      "epoch": 0.45675902602854745,
+      "eval_loss": 0.4290771484375,
+      "eval_runtime": 1743.8313,
+      "eval_samples_per_second": 2.562,
+      "eval_steps_per_second": 2.562,
+      "step": 1020
+    },
+    {
+      "epoch": 0.4572068289952421,
+      "grad_norm": 0.9210574626922607,
+      "learning_rate": 3.976223453370268e-05,
+      "loss": 0.3855,
+      "step": 1021
+    },
+    {
+      "epoch": 0.45765463196193673,
+      "grad_norm": 1.167533278465271,
+      "learning_rate": 3.9750692520775626e-05,
+      "loss": 0.5742,
+      "step": 1022
+    },
+    {
+      "epoch": 0.4581024349286314,
+      "grad_norm": 0.9851381778717041,
+      "learning_rate": 3.973915050784857e-05,
+      "loss": 0.3645,
+      "step": 1023
+    },
+    {
+      "epoch": 0.45855023789532606,
+      "grad_norm": 1.2704929113388062,
+      "learning_rate": 3.972760849492151e-05,
+      "loss": 0.4467,
+      "step": 1024
+    },
+    {
+      "epoch": 0.45899804086202073,
+      "grad_norm": 1.1447492837905884,
+      "learning_rate": 3.971606648199446e-05,
+      "loss": 0.3171,
+      "step": 1025
+    },
+    {
+      "epoch": 0.45944584382871534,
+      "grad_norm": 1.1135897636413574,
+      "learning_rate": 3.97045244690674e-05,
+      "loss": 0.5033,
+      "step": 1026
+    },
+    {
+      "epoch": 0.45989364679541,
+      "grad_norm": 1.3386653661727905,
+      "learning_rate": 3.969298245614035e-05,
+      "loss": 0.6805,
+      "step": 1027
+    },
+    {
+      "epoch": 0.46034144976210467,
+      "grad_norm": 0.9690244197845459,
+      "learning_rate": 3.9681440443213295e-05,
+      "loss": 0.2986,
+      "step": 1028
+    },
+    {
+      "epoch": 0.46078925272879934,
+      "grad_norm": 1.201811671257019,
+      "learning_rate": 3.9669898430286245e-05,
+      "loss": 0.4149,
+      "step": 1029
+    },
+    {
+      "epoch": 0.461237055695494,
+      "grad_norm": 0.9881176948547363,
+      "learning_rate": 3.965835641735919e-05,
+      "loss": 0.4466,
+      "step": 1030
+    },
+    {
+      "epoch": 0.461237055695494,
+      "eval_loss": 0.423381507396698,
+      "eval_runtime": 1742.3334,
+      "eval_samples_per_second": 2.564,
+      "eval_steps_per_second": 2.564,
+      "step": 1030
+    },
+    {
+      "epoch": 0.4616848586621886,
+      "grad_norm": 1.0963369607925415,
+      "learning_rate": 3.9646814404432136e-05,
+      "loss": 0.5915,
+      "step": 1031
+    },
+    {
+      "epoch": 0.4621326616288833,
+      "grad_norm": 1.0670597553253174,
+      "learning_rate": 3.963527239150508e-05,
+      "loss": 0.4366,
+      "step": 1032
+    },
+    {
+      "epoch": 0.46258046459557794,
+      "grad_norm": 1.0526273250579834,
+      "learning_rate": 3.962373037857803e-05,
+      "loss": 0.3719,
+      "step": 1033
+    },
+    {
+      "epoch": 0.4630282675622726,
+      "grad_norm": 1.1232683658599854,
+      "learning_rate": 3.961218836565097e-05,
+      "loss": 0.5496,
+      "step": 1034
+    },
+    {
+      "epoch": 0.4634760705289673,
+      "grad_norm": 1.2455625534057617,
+      "learning_rate": 3.960064635272392e-05,
+      "loss": 0.3716,
+      "step": 1035
+    },
+    {
+      "epoch": 0.4639238734956619,
+      "grad_norm": 1.116101622581482,
+      "learning_rate": 3.958910433979686e-05,
+      "loss": 0.2784,
+      "step": 1036
+    },
+    {
+      "epoch": 0.46437167646235655,
+      "grad_norm": 1.1567442417144775,
+      "learning_rate": 3.957756232686981e-05,
+      "loss": 0.4029,
+      "step": 1037
+    },
+    {
+      "epoch": 0.4648194794290512,
+      "grad_norm": 1.0800672769546509,
+      "learning_rate": 3.9566020313942755e-05,
+      "loss": 0.2335,
+      "step": 1038
+    },
+    {
+      "epoch": 0.4652672823957459,
+      "grad_norm": 0.8808770775794983,
+      "learning_rate": 3.9554478301015704e-05,
+      "loss": 0.2624,
+      "step": 1039
+    },
+    {
+      "epoch": 0.46571508536244055,
+      "grad_norm": 1.176003098487854,
+      "learning_rate": 3.954293628808865e-05,
+      "loss": 0.3314,
+      "step": 1040
+    },
+    {
+      "epoch": 0.46571508536244055,
+      "eval_loss": 0.41852641105651855,
+      "eval_runtime": 1741.3418,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 2.565,
+      "step": 1040
+    },
+    {
+      "epoch": 0.46616288832913516,
+      "grad_norm": 1.1637083292007446,
+      "learning_rate": 3.953139427516159e-05,
+      "loss": 0.2736,
+      "step": 1041
+    },
+    {
+      "epoch": 0.4666106912958298,
+      "grad_norm": 1.36524498462677,
+      "learning_rate": 3.951985226223454e-05,
+      "loss": 0.3522,
+      "step": 1042
+    },
+    {
+      "epoch": 0.4670584942625245,
+      "grad_norm": 1.0998141765594482,
+      "learning_rate": 3.950831024930748e-05,
+      "loss": 0.4015,
+      "step": 1043
+    },
+    {
+      "epoch": 0.46750629722921916,
+      "grad_norm": 1.8939387798309326,
+      "learning_rate": 3.9496768236380424e-05,
+      "loss": 0.4661,
+      "step": 1044
+    },
+    {
+      "epoch": 0.4679541001959138,
+      "grad_norm": 1.365972876548767,
+      "learning_rate": 3.948522622345337e-05,
+      "loss": 0.3268,
+      "step": 1045
+    },
+    {
+      "epoch": 0.46840190316260844,
+      "grad_norm": 1.078254222869873,
+      "learning_rate": 3.9473684210526316e-05,
+      "loss": 0.37,
+      "step": 1046
+    },
+    {
+      "epoch": 0.4688497061293031,
+      "grad_norm": 1.2513341903686523,
+      "learning_rate": 3.946214219759926e-05,
+      "loss": 0.7105,
+      "step": 1047
+    },
+    {
+      "epoch": 0.46929750909599777,
+      "grad_norm": 1.2173830270767212,
+      "learning_rate": 3.945060018467221e-05,
+      "loss": 0.4167,
+      "step": 1048
+    },
+    {
+      "epoch": 0.46974531206269243,
+      "grad_norm": 1.5618706941604614,
+      "learning_rate": 3.943905817174515e-05,
+      "loss": 0.3482,
+      "step": 1049
+    },
+    {
+      "epoch": 0.47019311502938704,
+      "grad_norm": 1.234503984451294,
+      "learning_rate": 3.94275161588181e-05,
+      "loss": 0.5212,
+      "step": 1050
+    },
+    {
+      "epoch": 0.47019311502938704,
+      "eval_loss": 0.41508200764656067,
+      "eval_runtime": 1741.1289,
+      "eval_samples_per_second": 2.566,
+      "eval_steps_per_second": 2.566,
+      "step": 1050
+    },
+    {
+      "epoch": 0.4706409179960817,
+      "grad_norm": 1.360834002494812,
+      "learning_rate": 3.941597414589104e-05,
+      "loss": 0.2982,
+      "step": 1051
+    },
+    {
+      "epoch": 0.4710887209627764,
+      "grad_norm": 1.036285638809204,
+      "learning_rate": 3.940443213296399e-05,
+      "loss": 0.2856,
+      "step": 1052
+    },
+    {
+      "epoch": 0.47153652392947104,
+      "grad_norm": 1.3228135108947754,
+      "learning_rate": 3.9392890120036935e-05,
+      "loss": 0.4763,
+      "step": 1053
+    },
+    {
+      "epoch": 0.4719843268961657,
+      "grad_norm": 1.2010780572891235,
+      "learning_rate": 3.9381348107109884e-05,
+      "loss": 0.3307,
+      "step": 1054
+    },
+    {
+      "epoch": 0.4724321298628603,
+      "grad_norm": 1.1086570024490356,
+      "learning_rate": 3.936980609418283e-05,
+      "loss": 0.4546,
+      "step": 1055
+    },
+    {
+      "epoch": 0.472879932829555,
+      "grad_norm": 1.1960604190826416,
+      "learning_rate": 3.9358264081255776e-05,
+      "loss": 0.405,
+      "step": 1056
+    },
+    {
+      "epoch": 0.47332773579624965,
+      "grad_norm": 1.186313509941101,
+      "learning_rate": 3.934672206832872e-05,
+      "loss": 0.4105,
+      "step": 1057
+    },
+    {
+      "epoch": 0.4737755387629443,
+      "grad_norm": 0.9901049137115479,
+      "learning_rate": 3.933518005540167e-05,
+      "loss": 0.4515,
+      "step": 1058
+    },
+    {
+      "epoch": 0.474223341729639,
+      "grad_norm": 1.000322699546814,
+      "learning_rate": 3.932363804247461e-05,
+      "loss": 0.2171,
+      "step": 1059
+    },
+    {
+      "epoch": 0.4746711446963336,
+      "grad_norm": 1.177179217338562,
+      "learning_rate": 3.931209602954756e-05,
+      "loss": 0.3213,
+      "step": 1060
+    },
+    {
+      "epoch": 0.4746711446963336,
+      "eval_loss": 0.4125591516494751,
+      "eval_runtime": 1740.7949,
+      "eval_samples_per_second": 2.566,
+      "eval_steps_per_second": 2.566,
+      "step": 1060
+    },
+    {
+      "epoch": 0.47511894766302826,
+      "grad_norm": 0.945931613445282,
+      "learning_rate": 3.93005540166205e-05,
+      "loss": 0.4175,
+      "step": 1061
+    },
+    {
+      "epoch": 0.4755667506297229,
+      "grad_norm": 1.0835816860198975,
+      "learning_rate": 3.9289012003693445e-05,
+      "loss": 0.2497,
+      "step": 1062
+    },
+    {
+      "epoch": 0.4760145535964176,
+      "grad_norm": 1.0238778591156006,
+      "learning_rate": 3.927746999076639e-05,
+      "loss": 0.2807,
+      "step": 1063
+    },
+    {
+      "epoch": 0.47646235656311225,
+      "grad_norm": 1.2828023433685303,
+      "learning_rate": 3.926592797783934e-05,
+      "loss": 0.3368,
+      "step": 1064
+    },
+    {
+      "epoch": 0.47691015952980687,
+      "grad_norm": 1.323936104774475,
+      "learning_rate": 3.925438596491228e-05,
+      "loss": 0.4173,
+      "step": 1065
+    },
+    {
+      "epoch": 0.47735796249650153,
+      "grad_norm": 1.270038366317749,
+      "learning_rate": 3.924284395198522e-05,
+      "loss": 0.4249,
+      "step": 1066
+    },
+    {
+      "epoch": 0.4778057654631962,
+      "grad_norm": 1.2153494358062744,
+      "learning_rate": 3.923130193905817e-05,
+      "loss": 0.2474,
+      "step": 1067
+    },
+    {
+      "epoch": 0.47825356842989086,
+      "grad_norm": 1.2241121530532837,
+      "learning_rate": 3.9219759926131115e-05,
+      "loss": 0.3937,
+      "step": 1068
+    },
+    {
+      "epoch": 0.47870137139658553,
+      "grad_norm": 1.294264793395996,
+      "learning_rate": 3.9208217913204064e-05,
+      "loss": 0.7127,
+      "step": 1069
+    },
+    {
+      "epoch": 0.47914917436328014,
+      "grad_norm": 1.1305440664291382,
+      "learning_rate": 3.9196675900277007e-05,
+      "loss": 0.552,
+      "step": 1070
+    },
+    {
+      "epoch": 0.47914917436328014,
+      "eval_loss": 0.4095277786254883,
+      "eval_runtime": 1740.4302,
+      "eval_samples_per_second": 2.567,
+      "eval_steps_per_second": 2.567,
+      "step": 1070
+    },
+    {
+      "epoch": 0.4795969773299748,
+      "grad_norm": 1.3406230211257935,
+      "learning_rate": 3.9185133887349956e-05,
+      "loss": 0.4512,
+      "step": 1071
+    },
+    {
+      "epoch": 0.48004478029666947,
+      "grad_norm": 1.2547788619995117,
+      "learning_rate": 3.91735918744229e-05,
+      "loss": 0.4066,
+      "step": 1072
+    },
+    {
+      "epoch": 0.48049258326336414,
+      "grad_norm": 1.0894948244094849,
+      "learning_rate": 3.916204986149585e-05,
+      "loss": 0.5634,
+      "step": 1073
+    },
+    {
+      "epoch": 0.48094038623005875,
+      "grad_norm": 1.1141157150268555,
+      "learning_rate": 3.915050784856879e-05,
+      "loss": 0.5119,
+      "step": 1074
+    },
+    {
+      "epoch": 0.4813881891967534,
+      "grad_norm": 1.0408393144607544,
+      "learning_rate": 3.913896583564174e-05,
+      "loss": 0.3573,
+      "step": 1075
+    },
+    {
+      "epoch": 0.4818359921634481,
+      "grad_norm": 1.20759117603302,
+      "learning_rate": 3.912742382271468e-05,
+      "loss": 0.3264,
+      "step": 1076
+    },
+    {
+      "epoch": 0.48228379513014275,
+      "grad_norm": 1.1531039476394653,
+      "learning_rate": 3.911588180978763e-05,
+      "loss": 0.4826,
+      "step": 1077
+    },
+    {
+      "epoch": 0.4827315980968374,
+      "grad_norm": 1.6651794910430908,
+      "learning_rate": 3.9104339796860575e-05,
+      "loss": 0.4274,
+      "step": 1078
+    },
+    {
+      "epoch": 0.483179401063532,
+      "grad_norm": 0.916927695274353,
+      "learning_rate": 3.9092797783933524e-05,
+      "loss": 0.3246,
+      "step": 1079
+    },
+    {
+      "epoch": 0.4836272040302267,
+      "grad_norm": 1.0637052059173584,
+      "learning_rate": 3.9081255771006467e-05,
+      "loss": 0.3588,
+      "step": 1080
+    },
+    {
+      "epoch": 0.4836272040302267,
+      "eval_loss": 0.40346789360046387,
+      "eval_runtime": 1739.6386,
+      "eval_samples_per_second": 2.568,
+      "eval_steps_per_second": 2.568,
+      "step": 1080
+    },
+    {
+      "epoch": 0.48407500699692135,
+      "grad_norm": 1.419047236442566,
+      "learning_rate": 3.9069713758079416e-05,
+      "loss": 0.2933,
+      "step": 1081
+    },
+    {
+      "epoch": 0.484522809963616,
+      "grad_norm": 1.3041396141052246,
+      "learning_rate": 3.905817174515236e-05,
+      "loss": 0.6344,
+      "step": 1082
+    },
+    {
+      "epoch": 0.4849706129303107,
+      "grad_norm": 1.1709884405136108,
+      "learning_rate": 3.90466297322253e-05,
+      "loss": 0.7341,
+      "step": 1083
+    },
+    {
+      "epoch": 0.4854184158970053,
+      "grad_norm": 1.2271945476531982,
+      "learning_rate": 3.9035087719298244e-05,
+      "loss": 0.6506,
+      "step": 1084
+    },
+    {
+      "epoch": 0.48586621886369996,
+      "grad_norm": 1.1007139682769775,
+      "learning_rate": 3.902354570637119e-05,
+      "loss": 0.4708,
+      "step": 1085
+    },
+    {
+      "epoch": 0.4863140218303946,
+      "grad_norm": 1.242135763168335,
+      "learning_rate": 3.9012003693444136e-05,
+      "loss": 0.6122,
+      "step": 1086
+    },
+    {
+      "epoch": 0.4867618247970893,
+      "grad_norm": 1.1739150285720825,
+      "learning_rate": 3.9000461680517085e-05,
+      "loss": 0.23,
+      "step": 1087
+    },
+    {
+      "epoch": 0.48720962776378396,
+      "grad_norm": 1.2357375621795654,
+      "learning_rate": 3.898891966759003e-05,
+      "loss": 0.3669,
+      "step": 1088
+    },
+    {
+      "epoch": 0.48765743073047857,
+      "grad_norm": 0.9539027214050293,
+      "learning_rate": 3.897737765466298e-05,
+      "loss": 0.3564,
+      "step": 1089
+    },
+    {
+      "epoch": 0.48810523369717324,
+      "grad_norm": 1.1826626062393188,
+      "learning_rate": 3.896583564173592e-05,
+      "loss": 0.5192,
+      "step": 1090
+    },
+    {
+      "epoch": 0.48810523369717324,
+      "eval_loss": 0.39695626497268677,
+      "eval_runtime": 1739.8638,
+      "eval_samples_per_second": 2.567,
+      "eval_steps_per_second": 2.567,
+      "step": 1090
+    },
+    {
+      "epoch": 0.4885530366638679,
+      "grad_norm": 1.36544668674469,
+      "learning_rate": 3.895429362880887e-05,
+      "loss": 0.4346,
+      "step": 1091
+    },
+    {
+      "epoch": 0.48900083963056257,
+      "grad_norm": 1.2532953023910522,
+      "learning_rate": 3.894275161588181e-05,
+      "loss": 0.3558,
+      "step": 1092
+    },
+    {
+      "epoch": 0.48944864259725723,
+      "grad_norm": 1.2060133218765259,
+      "learning_rate": 3.8931209602954754e-05,
+      "loss": 0.354,
+      "step": 1093
+    },
+    {
+      "epoch": 0.48989644556395184,
+      "grad_norm": 1.3531482219696045,
+      "learning_rate": 3.8919667590027704e-05,
+      "loss": 0.7354,
+      "step": 1094
+    },
+    {
+      "epoch": 0.4903442485306465,
+      "grad_norm": 1.2521357536315918,
+      "learning_rate": 3.8908125577100646e-05,
+      "loss": 0.4558,
+      "step": 1095
+    },
+    {
+      "epoch": 0.4907920514973412,
+      "grad_norm": 1.319419264793396,
+      "learning_rate": 3.8896583564173596e-05,
+      "loss": 0.4897,
+      "step": 1096
+    },
+    {
+      "epoch": 0.49123985446403584,
+      "grad_norm": 1.3026233911514282,
+      "learning_rate": 3.888504155124654e-05,
+      "loss": 0.5194,
+      "step": 1097
+    },
+    {
+      "epoch": 0.49168765743073045,
+      "grad_norm": 1.3030741214752197,
+      "learning_rate": 3.887349953831949e-05,
+      "loss": 0.5338,
+      "step": 1098
+    },
+    {
+      "epoch": 0.4921354603974251,
+      "grad_norm": 1.2530677318572998,
+      "learning_rate": 3.886195752539243e-05,
+      "loss": 0.3714,
+      "step": 1099
+    },
+    {
+      "epoch": 0.4925832633641198,
+      "grad_norm": 1.1098151206970215,
+      "learning_rate": 3.885041551246538e-05,
+      "loss": 0.3104,
+      "step": 1100
+    },
+    {
+      "epoch": 0.4925832633641198,
+      "eval_loss": 0.39361706376075745,
+      "eval_runtime": 1739.4256,
+      "eval_samples_per_second": 2.568,
+      "eval_steps_per_second": 2.568,
+      "step": 1100
+    },
+    {
+      "epoch": 0.49303106633081445,
+      "grad_norm": 1.1043074131011963,
+      "learning_rate": 3.883887349953832e-05,
+      "loss": 0.4308,
+      "step": 1101
+    },
+    {
+      "epoch": 0.4934788692975091,
+      "grad_norm": 1.0311301946640015,
+      "learning_rate": 3.8827331486611265e-05,
+      "loss": 0.2908,
+      "step": 1102
+    },
+    {
+      "epoch": 0.4939266722642037,
+      "grad_norm": 1.250179409980774,
+      "learning_rate": 3.8815789473684214e-05,
+      "loss": 0.4199,
+      "step": 1103
+    },
+    {
+      "epoch": 0.4943744752308984,
+      "grad_norm": 1.0847578048706055,
+      "learning_rate": 3.880424746075716e-05,
+      "loss": 0.474,
+      "step": 1104
+    },
+    {
+      "epoch": 0.49482227819759306,
+      "grad_norm": 1.0442312955856323,
+      "learning_rate": 3.87927054478301e-05,
+      "loss": 0.2112,
+      "step": 1105
+    },
+    {
+      "epoch": 0.4952700811642877,
+      "grad_norm": 1.09678316116333,
+      "learning_rate": 3.878116343490305e-05,
+      "loss": 0.3507,
+      "step": 1106
+    },
+    {
+      "epoch": 0.4957178841309824,
+      "grad_norm": 0.8428501486778259,
+      "learning_rate": 3.876962142197599e-05,
+      "loss": 0.1935,
+      "step": 1107
+    },
+    {
+      "epoch": 0.496165687097677,
+      "grad_norm": 1.255764365196228,
+      "learning_rate": 3.875807940904894e-05,
+      "loss": 0.4886,
+      "step": 1108
+    },
+    {
+      "epoch": 0.49661349006437167,
+      "grad_norm": 1.2545994520187378,
+      "learning_rate": 3.8746537396121883e-05,
+      "loss": 0.4899,
+      "step": 1109
+    },
+    {
+      "epoch": 0.49706129303106633,
+      "grad_norm": 0.9944257140159607,
+      "learning_rate": 3.873499538319483e-05,
+      "loss": 0.3231,
+      "step": 1110
+    },
+    {
+      "epoch": 0.49706129303106633,
+      "eval_loss": 0.388060063123703,
+      "eval_runtime": 1739.6113,
+      "eval_samples_per_second": 2.568,
+      "eval_steps_per_second": 2.568,
+      "step": 1110
+    },
+    {
+      "epoch": 0.497509095997761,
+      "grad_norm": 1.3695605993270874,
+      "learning_rate": 3.8723453370267775e-05,
+      "loss": 0.3714,
+      "step": 1111
+    },
+    {
+      "epoch": 0.49795689896445566,
+      "grad_norm": 1.5230897665023804,
+      "learning_rate": 3.8711911357340725e-05,
+      "loss": 0.8132,
+      "step": 1112
+    },
+    {
+      "epoch": 0.4984047019311503,
+      "grad_norm": 1.4187443256378174,
+      "learning_rate": 3.870036934441367e-05,
+      "loss": 0.4506,
+      "step": 1113
+    },
+    {
+      "epoch": 0.49885250489784494,
+      "grad_norm": 1.4074289798736572,
+      "learning_rate": 3.868882733148662e-05,
+      "loss": 0.4261,
+      "step": 1114
+    },
+    {
+      "epoch": 0.4993003078645396,
+      "grad_norm": 1.161434292793274,
+      "learning_rate": 3.867728531855956e-05,
+      "loss": 0.4857,
+      "step": 1115
+    },
+    {
+      "epoch": 0.49974811083123427,
+      "grad_norm": 1.053652048110962,
+      "learning_rate": 3.866574330563251e-05,
+      "loss": 0.2793,
+      "step": 1116
+    },
+    {
+      "epoch": 0.5001959137979289,
+      "grad_norm": 1.2816743850708008,
+      "learning_rate": 3.865420129270545e-05,
+      "loss": 0.4757,
+      "step": 1117
+    },
+    {
+      "epoch": 0.5006437167646236,
+      "grad_norm": 1.2309850454330444,
+      "learning_rate": 3.8642659279778394e-05,
+      "loss": 0.4121,
+      "step": 1118
+    },
+    {
+      "epoch": 0.5010915197313183,
+      "grad_norm": 1.3217289447784424,
+      "learning_rate": 3.8631117266851343e-05,
+      "loss": 0.4224,
+      "step": 1119
+    },
+    {
+      "epoch": 0.5015393226980128,
+      "grad_norm": 1.1132780313491821,
+      "learning_rate": 3.8619575253924286e-05,
+      "loss": 0.4038,
+      "step": 1120
+    },
+    {
+      "epoch": 0.5015393226980128,
+      "eval_loss": 0.38312047719955444,
+      "eval_runtime": 1739.3273,
+      "eval_samples_per_second": 2.568,
+      "eval_steps_per_second": 2.568,
+      "step": 1120
+    },
+    {
+      "epoch": 0.5019871256647075,
+      "grad_norm": 1.0189847946166992,
+      "learning_rate": 3.8608033240997235e-05,
+      "loss": 0.3739,
+      "step": 1121
+    },
+    {
+      "epoch": 0.5024349286314022,
+      "grad_norm": 0.9733580350875854,
+      "learning_rate": 3.859649122807018e-05,
+      "loss": 0.3766,
+      "step": 1122
+    },
+    {
+      "epoch": 0.5028827315980968,
+      "grad_norm": 1.2229063510894775,
+      "learning_rate": 3.858494921514312e-05,
+      "loss": 0.4201,
+      "step": 1123
+    },
+    {
+      "epoch": 0.5033305345647915,
+      "grad_norm": 1.1701637506484985,
+      "learning_rate": 3.857340720221606e-05,
+      "loss": 0.4795,
+      "step": 1124
+    },
+    {
+      "epoch": 0.5037783375314862,
+      "grad_norm": 1.2924901247024536,
+      "learning_rate": 3.856186518928901e-05,
+      "loss": 0.4878,
+      "step": 1125
+    },
+    {
+      "epoch": 0.5042261404981808,
+      "grad_norm": 1.1813278198242188,
+      "learning_rate": 3.8550323176361955e-05,
+      "loss": 0.402,
+      "step": 1126
+    },
+    {
+      "epoch": 0.5046739434648755,
+      "grad_norm": 1.4923017024993896,
+      "learning_rate": 3.8538781163434905e-05,
+      "loss": 0.464,
+      "step": 1127
+    },
+    {
+      "epoch": 0.5051217464315702,
+      "grad_norm": 1.1882212162017822,
+      "learning_rate": 3.852723915050785e-05,
+      "loss": 0.2678,
+      "step": 1128
+    },
+    {
+      "epoch": 0.5055695493982648,
+      "grad_norm": 0.9495273232460022,
+      "learning_rate": 3.8515697137580797e-05,
+      "loss": 0.2749,
+      "step": 1129
+    },
+    {
+      "epoch": 0.5060173523649594,
+      "grad_norm": 1.1850472688674927,
+      "learning_rate": 3.850415512465374e-05,
+      "loss": 0.2712,
+      "step": 1130
+    },
+    {
+      "epoch": 0.5060173523649594,
+      "eval_loss": 0.3788580894470215,
+      "eval_runtime": 1739.553,
+      "eval_samples_per_second": 2.568,
+      "eval_steps_per_second": 2.568,
+      "step": 1130
+    },
+    {
+      "epoch": 0.506465155331654,
+      "grad_norm": 1.2247016429901123,
+      "learning_rate": 3.849261311172669e-05,
+      "loss": 0.5771,
+      "step": 1131
+    },
+    {
+      "epoch": 0.5069129582983487,
+      "grad_norm": 1.2585265636444092,
+      "learning_rate": 3.848107109879963e-05,
+      "loss": 0.4273,
+      "step": 1132
+    },
+    {
+      "epoch": 0.5073607612650434,
+      "grad_norm": 1.1470332145690918,
+      "learning_rate": 3.846952908587258e-05,
+      "loss": 0.4164,
+      "step": 1133
+    },
+    {
+      "epoch": 0.507808564231738,
+      "grad_norm": 1.2961275577545166,
+      "learning_rate": 3.845798707294552e-05,
+      "loss": 0.4638,
+      "step": 1134
+    },
+    {
+      "epoch": 0.5082563671984327,
+      "grad_norm": 1.2521647214889526,
+      "learning_rate": 3.844644506001847e-05,
+      "loss": 0.2843,
+      "step": 1135
+    },
+    {
+      "epoch": 0.5087041701651274,
+      "grad_norm": 1.1778457164764404,
+      "learning_rate": 3.8434903047091415e-05,
+      "loss": 0.4627,
+      "step": 1136
+    },
+    {
+      "epoch": 0.509151973131822,
+      "grad_norm": 1.1717246770858765,
+      "learning_rate": 3.8423361034164365e-05,
+      "loss": 0.386,
+      "step": 1137
+    },
+    {
+      "epoch": 0.5095997760985167,
+      "grad_norm": 1.1839237213134766,
+      "learning_rate": 3.841181902123731e-05,
+      "loss": 0.35,
+      "step": 1138
+    },
+    {
+      "epoch": 0.5100475790652113,
+      "grad_norm": 1.2348573207855225,
+      "learning_rate": 3.8400277008310257e-05,
+      "loss": 0.5258,
+      "step": 1139
+    },
+    {
+      "epoch": 0.5104953820319059,
+      "grad_norm": 1.4483014345169067,
+      "learning_rate": 3.83887349953832e-05,
+      "loss": 0.3698,
+      "step": 1140
+    },
+    {
+      "epoch": 0.5104953820319059,
+      "eval_loss": 0.3735828101634979,
+      "eval_runtime": 1738.8301,
+      "eval_samples_per_second": 2.569,
+      "eval_steps_per_second": 2.569,
+      "step": 1140
+    },
+    {
+      "epoch": 0.5109431849986006,
+      "grad_norm": 1.017661452293396,
+      "learning_rate": 3.837719298245614e-05,
+      "loss": 0.2918,
+      "step": 1141
+    },
+    {
+      "epoch": 0.5113909879652953,
+      "grad_norm": 1.1523457765579224,
+      "learning_rate": 3.836565096952909e-05,
+      "loss": 0.4667,
+      "step": 1142
+    },
+    {
+      "epoch": 0.5118387909319899,
+      "grad_norm": 1.0640584230422974,
+      "learning_rate": 3.8354108956602034e-05,
+      "loss": 0.3532,
+      "step": 1143
+    },
+    {
+      "epoch": 0.5122865938986846,
+      "grad_norm": 1.2194321155548096,
+      "learning_rate": 3.8342566943674976e-05,
+      "loss": 0.4472,
+      "step": 1144
+    },
+    {
+      "epoch": 0.5127343968653792,
+      "grad_norm": 1.1272408962249756,
+      "learning_rate": 3.833102493074792e-05,
+      "loss": 0.3863,
+      "step": 1145
+    },
+    {
+      "epoch": 0.5131821998320739,
+      "grad_norm": 1.1771918535232544,
+      "learning_rate": 3.831948291782087e-05,
+      "loss": 0.3669,
+      "step": 1146
+    },
+    {
+      "epoch": 0.5136300027987686,
+      "grad_norm": 1.154484748840332,
+      "learning_rate": 3.830794090489381e-05,
+      "loss": 0.4898,
+      "step": 1147
+    },
+    {
+      "epoch": 0.5140778057654632,
+      "grad_norm": 1.1640018224716187,
+      "learning_rate": 3.829639889196676e-05,
+      "loss": 0.2343,
+      "step": 1148
+    },
+    {
+      "epoch": 0.5145256087321578,
+      "grad_norm": 1.1481584310531616,
+      "learning_rate": 3.82848568790397e-05,
+      "loss": 0.2811,
+      "step": 1149
+    },
+    {
+      "epoch": 0.5149734116988525,
+      "grad_norm": 1.322683334350586,
+      "learning_rate": 3.827331486611265e-05,
+      "loss": 0.279,
+      "step": 1150
+    },
+    {
+      "epoch": 0.5149734116988525,
+      "eval_loss": 0.36834877729415894,
+      "eval_runtime": 1738.7448,
+      "eval_samples_per_second": 2.569,
+      "eval_steps_per_second": 2.569,
+      "step": 1150
+    },
+    {
+      "epoch": 0.5154212146655471,
+      "grad_norm": 1.4827194213867188,
+      "learning_rate": 3.8261772853185595e-05,
+      "loss": 0.4846,
+      "step": 1151
+    },
+    {
+      "epoch": 0.5158690176322418,
+      "grad_norm": 1.0381073951721191,
+      "learning_rate": 3.8250230840258544e-05,
+      "loss": 0.2628,
+      "step": 1152
+    },
+    {
+      "epoch": 0.5163168205989365,
+      "grad_norm": 1.0377421379089355,
+      "learning_rate": 3.823868882733149e-05,
+      "loss": 0.2416,
+      "step": 1153
+    },
+    {
+      "epoch": 0.5167646235656311,
+      "grad_norm": 1.114513635635376,
+      "learning_rate": 3.8227146814404436e-05,
+      "loss": 0.3389,
+      "step": 1154
+    },
+    {
+      "epoch": 0.5172124265323258,
+      "grad_norm": 1.1325706243515015,
+      "learning_rate": 3.821560480147738e-05,
+      "loss": 0.5003,
+      "step": 1155
+    },
+    {
+      "epoch": 0.5176602294990205,
+      "grad_norm": 0.9513109922409058,
+      "learning_rate": 3.820406278855033e-05,
+      "loss": 0.5039,
+      "step": 1156
+    },
+    {
+      "epoch": 0.5181080324657151,
+      "grad_norm": 1.3812943696975708,
+      "learning_rate": 3.819252077562327e-05,
+      "loss": 0.3025,
+      "step": 1157
+    },
+    {
+      "epoch": 0.5185558354324098,
+      "grad_norm": 1.07074773311615,
+      "learning_rate": 3.818097876269622e-05,
+      "loss": 0.4571,
+      "step": 1158
+    },
+    {
+      "epoch": 0.5190036383991044,
+      "grad_norm": 1.3915842771530151,
+      "learning_rate": 3.816943674976916e-05,
+      "loss": 0.3686,
+      "step": 1159
+    },
+    {
+      "epoch": 0.519451441365799,
+      "grad_norm": 1.3118082284927368,
+      "learning_rate": 3.815789473684211e-05,
+      "loss": 0.268,
+      "step": 1160
+    },
+    {
+      "epoch": 0.519451441365799,
+      "eval_loss": 0.36535876989364624,
+      "eval_runtime": 1738.7404,
+      "eval_samples_per_second": 2.569,
+      "eval_steps_per_second": 2.569,
+      "step": 1160
+    },
+    {
+      "epoch": 0.5198992443324937,
+      "grad_norm": 1.2658312320709229,
+      "learning_rate": 3.8146352723915055e-05,
+      "loss": 0.4729,
+      "step": 1161
+    },
+    {
+      "epoch": 0.5203470472991883,
+      "grad_norm": 1.1261812448501587,
+      "learning_rate": 3.8134810710988e-05,
+      "loss": 0.3151,
+      "step": 1162
+    },
+    {
+      "epoch": 0.520794850265883,
+      "grad_norm": 1.354509711265564,
+      "learning_rate": 3.812326869806094e-05,
+      "loss": 0.4913,
+      "step": 1163
+    },
+    {
+      "epoch": 0.5212426532325777,
+      "grad_norm": 1.498897910118103,
+      "learning_rate": 3.811172668513389e-05,
+      "loss": 0.4962,
+      "step": 1164
+    },
+    {
+      "epoch": 0.5216904561992723,
+      "grad_norm": 1.4946974515914917,
+      "learning_rate": 3.810018467220683e-05,
+      "loss": 0.6484,
+      "step": 1165
+    },
+    {
+      "epoch": 0.522138259165967,
+      "grad_norm": 1.1827718019485474,
+      "learning_rate": 3.808864265927978e-05,
+      "loss": 0.3115,
+      "step": 1166
+    },
+    {
+      "epoch": 0.5225860621326617,
+      "grad_norm": 1.1302978992462158,
+      "learning_rate": 3.8077100646352724e-05,
+      "loss": 0.404,
+      "step": 1167
+    },
+    {
+      "epoch": 0.5230338650993562,
+      "grad_norm": 1.1893664598464966,
+      "learning_rate": 3.806555863342567e-05,
+      "loss": 0.4,
+      "step": 1168
+    },
+    {
+      "epoch": 0.5234816680660509,
+      "grad_norm": 1.1274092197418213,
+      "learning_rate": 3.8054016620498616e-05,
+      "loss": 0.4143,
+      "step": 1169
+    },
+    {
+      "epoch": 0.5239294710327456,
+      "grad_norm": 1.305298924446106,
+      "learning_rate": 3.804247460757156e-05,
+      "loss": 0.1854,
+      "step": 1170
+    },
+    {
+      "epoch": 0.5239294710327456,
+      "eval_loss": 0.3615010380744934,
+      "eval_runtime": 1738.5553,
+      "eval_samples_per_second": 2.569,
+      "eval_steps_per_second": 2.569,
+      "step": 1170
+    },
+    {
+      "epoch": 0.5243772739994402,
+      "grad_norm": 1.3721306324005127,
+      "learning_rate": 3.803093259464451e-05,
+      "loss": 0.5466,
+      "step": 1171
+    },
+    {
+      "epoch": 0.5248250769661349,
+      "grad_norm": 1.2742910385131836,
+      "learning_rate": 3.801939058171745e-05,
+      "loss": 0.4656,
+      "step": 1172
+    },
+    {
+      "epoch": 0.5252728799328296,
+      "grad_norm": 1.2864550352096558,
+      "learning_rate": 3.80078485687904e-05,
+      "loss": 0.5515,
+      "step": 1173
+    },
+    {
+      "epoch": 0.5257206828995242,
+      "grad_norm": 1.3748183250427246,
+      "learning_rate": 3.799630655586334e-05,
+      "loss": 0.4864,
+      "step": 1174
+    },
+    {
+      "epoch": 0.5261684858662189,
+      "grad_norm": 1.0401967763900757,
+      "learning_rate": 3.798476454293629e-05,
+      "loss": 0.3758,
+      "step": 1175
+    },
+    {
+      "epoch": 0.5266162888329136,
+      "grad_norm": 1.117813229560852,
+      "learning_rate": 3.7973222530009235e-05,
+      "loss": 0.3174,
+      "step": 1176
+    },
+    {
+      "epoch": 0.5270640917996082,
+      "grad_norm": 1.2500524520874023,
+      "learning_rate": 3.7961680517082184e-05,
+      "loss": 0.6011,
+      "step": 1177
+    },
+    {
+      "epoch": 0.5275118947663028,
+      "grad_norm": 1.2969615459442139,
+      "learning_rate": 3.795013850415513e-05,
+      "loss": 0.3895,
+      "step": 1178
+    },
+    {
+      "epoch": 0.5279596977329974,
+      "grad_norm": 1.301548719406128,
+      "learning_rate": 3.7938596491228076e-05,
+      "loss": 0.5013,
+      "step": 1179
+    },
+    {
+      "epoch": 0.5284075006996921,
+      "grad_norm": 1.1384559869766235,
+      "learning_rate": 3.792705447830102e-05,
+      "loss": 0.4033,
+      "step": 1180
+    },
+    {
+      "epoch": 0.5284075006996921,
+      "eval_loss": 0.35599184036254883,
+      "eval_runtime": 1738.2824,
+      "eval_samples_per_second": 2.57,
+      "eval_steps_per_second": 2.57,
+      "step": 1180
+    },
+    {
+      "epoch": 0.5288553036663868,
+      "grad_norm": 1.06693434715271,
+      "learning_rate": 3.791551246537397e-05,
+      "loss": 0.2809,
+      "step": 1181
+    },
+    {
+      "epoch": 0.5293031066330814,
+      "grad_norm": 1.5773426294326782,
+      "learning_rate": 3.790397045244691e-05,
+      "loss": 0.3743,
+      "step": 1182
+    },
+    {
+      "epoch": 0.5297509095997761,
+      "grad_norm": 1.1038140058517456,
+      "learning_rate": 3.789242843951985e-05,
+      "loss": 0.15,
+      "step": 1183
+    },
+    {
+      "epoch": 0.5301987125664708,
+      "grad_norm": 1.9519281387329102,
+      "learning_rate": 3.7880886426592796e-05,
+      "loss": 0.5218,
+      "step": 1184
+    },
+    {
+      "epoch": 0.5306465155331654,
+      "grad_norm": 1.0286353826522827,
+      "learning_rate": 3.7869344413665745e-05,
+      "loss": 0.3094,
+      "step": 1185
+    },
+    {
+      "epoch": 0.5310943184998601,
+      "grad_norm": 1.140738606452942,
+      "learning_rate": 3.785780240073869e-05,
+      "loss": 0.3357,
+      "step": 1186
+    },
+    {
+      "epoch": 0.5315421214665547,
+      "grad_norm": 1.3645200729370117,
+      "learning_rate": 3.784626038781164e-05,
+      "loss": 0.4584,
+      "step": 1187
+    },
+    {
+      "epoch": 0.5319899244332493,
+      "grad_norm": 1.1500276327133179,
+      "learning_rate": 3.783471837488458e-05,
+      "loss": 0.4634,
+      "step": 1188
+    },
+    {
+      "epoch": 0.532437727399944,
+      "grad_norm": 1.3467127084732056,
+      "learning_rate": 3.782317636195753e-05,
+      "loss": 0.3407,
+      "step": 1189
+    },
+    {
+      "epoch": 0.5328855303666387,
+      "grad_norm": 1.0754728317260742,
+      "learning_rate": 3.781163434903047e-05,
+      "loss": 0.4932,
+      "step": 1190
+    },
+    {
+      "epoch": 0.5328855303666387,
+      "eval_loss": 0.35155370831489563,
+      "eval_runtime": 1738.1857,
+      "eval_samples_per_second": 2.57,
+      "eval_steps_per_second": 2.57,
+      "step": 1190
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 1.2793216705322266,
+      "learning_rate": 3.780009233610342e-05,
+      "loss": 0.3066,
+      "step": 1191
+    },
+    {
+      "epoch": 0.533781136300028,
+      "grad_norm": 1.366106390953064,
+      "learning_rate": 3.7788550323176364e-05,
+      "loss": 0.2695,
+      "step": 1192
+    },
+    {
+      "epoch": 0.5342289392667227,
+      "grad_norm": 1.2858949899673462,
+      "learning_rate": 3.7777008310249306e-05,
+      "loss": 0.4392,
+      "step": 1193
+    },
+    {
+      "epoch": 0.5346767422334173,
+      "grad_norm": 1.2085633277893066,
+      "learning_rate": 3.7765466297322256e-05,
+      "loss": 0.4836,
+      "step": 1194
+    },
+    {
+      "epoch": 0.535124545200112,
+      "grad_norm": 1.352075457572937,
+      "learning_rate": 3.77539242843952e-05,
+      "loss": 0.2062,
+      "step": 1195
+    },
+    {
+      "epoch": 0.5355723481668067,
+      "grad_norm": 0.9742631912231445,
+      "learning_rate": 3.774238227146815e-05,
+      "loss": 0.2527,
+      "step": 1196
+    },
+    {
+      "epoch": 0.5360201511335012,
+      "grad_norm": 1.268314242362976,
+      "learning_rate": 3.773084025854109e-05,
+      "loss": 0.387,
+      "step": 1197
+    },
+    {
+      "epoch": 0.5364679541001959,
+      "grad_norm": 1.1327756643295288,
+      "learning_rate": 3.771929824561404e-05,
+      "loss": 0.2698,
+      "step": 1198
+    },
+    {
+      "epoch": 0.5369157570668905,
+      "grad_norm": 1.1978434324264526,
+      "learning_rate": 3.770775623268698e-05,
+      "loss": 0.2696,
+      "step": 1199
+    },
+    {
+      "epoch": 0.5373635600335852,
+      "grad_norm": 1.2355518341064453,
+      "learning_rate": 3.769621421975993e-05,
+      "loss": 0.3093,
+      "step": 1200
+    },
+    {
+      "epoch": 0.5373635600335852,
+      "eval_loss": 0.34780725836753845,
+      "eval_runtime": 1738.201,
+      "eval_samples_per_second": 2.57,
+      "eval_steps_per_second": 2.57,
+      "step": 1200
+    },
+    {
+      "epoch": 0.5378113630002799,
+      "grad_norm": 0.8600719571113586,
+      "learning_rate": 3.7684672206832874e-05,
+      "loss": 0.1402,
+      "step": 1201
+    },
+    {
+      "epoch": 0.5382591659669745,
+      "grad_norm": 1.3365446329116821,
+      "learning_rate": 3.767313019390582e-05,
+      "loss": 0.251,
+      "step": 1202
+    },
+    {
+      "epoch": 0.5387069689336692,
+      "grad_norm": 1.1617084741592407,
+      "learning_rate": 3.7661588180978766e-05,
+      "loss": 0.3694,
+      "step": 1203
+    },
+    {
+      "epoch": 0.5391547719003639,
+      "grad_norm": 1.145098328590393,
+      "learning_rate": 3.765004616805171e-05,
+      "loss": 0.4229,
+      "step": 1204
+    },
+    {
+      "epoch": 0.5396025748670585,
+      "grad_norm": 1.3333810567855835,
+      "learning_rate": 3.763850415512465e-05,
+      "loss": 0.2195,
+      "step": 1205
+    },
+    {
+      "epoch": 0.5400503778337532,
+      "grad_norm": 1.2409133911132812,
+      "learning_rate": 3.76269621421976e-05,
+      "loss": 0.4328,
+      "step": 1206
+    },
+    {
+      "epoch": 0.5404981808004478,
+      "grad_norm": 1.194001317024231,
+      "learning_rate": 3.7615420129270544e-05,
+      "loss": 0.2599,
+      "step": 1207
+    },
+    {
+      "epoch": 0.5409459837671424,
+      "grad_norm": 1.3317934274673462,
+      "learning_rate": 3.760387811634349e-05,
+      "loss": 0.3035,
+      "step": 1208
+    },
+    {
+      "epoch": 0.5413937867338371,
+      "grad_norm": 1.0461061000823975,
+      "learning_rate": 3.7592336103416436e-05,
+      "loss": 0.1203,
+      "step": 1209
+    },
+    {
+      "epoch": 0.5418415897005318,
+      "grad_norm": 1.2445271015167236,
+      "learning_rate": 3.7580794090489385e-05,
+      "loss": 0.2857,
+      "step": 1210
+    },
+    {
+      "epoch": 0.5418415897005318,
+      "eval_loss": 0.3450431227684021,
+      "eval_runtime": 1738.2005,
+      "eval_samples_per_second": 2.57,
+      "eval_steps_per_second": 2.57,
+      "step": 1210
+    },
+    {
+      "epoch": 0.5422893926672264,
+      "grad_norm": 1.0992363691329956,
+      "learning_rate": 3.756925207756233e-05,
+      "loss": 0.2729,
+      "step": 1211
+    },
+    {
+      "epoch": 0.5427371956339211,
+      "grad_norm": 1.0799282789230347,
+      "learning_rate": 3.755771006463528e-05,
+      "loss": 0.3273,
+      "step": 1212
+    },
+    {
+      "epoch": 0.5431849986006158,
+      "grad_norm": 1.0961830615997314,
+      "learning_rate": 3.754616805170822e-05,
+      "loss": 0.3408,
+      "step": 1213
+    },
+    {
+      "epoch": 0.5436328015673104,
+      "grad_norm": 1.1589432954788208,
+      "learning_rate": 3.753462603878117e-05,
+      "loss": 0.2953,
+      "step": 1214
+    },
+    {
+      "epoch": 0.5440806045340051,
+      "grad_norm": 1.4813915491104126,
+      "learning_rate": 3.752308402585411e-05,
+      "loss": 0.5627,
+      "step": 1215
+    },
+    {
+      "epoch": 0.5445284075006996,
+      "grad_norm": 1.0269968509674072,
+      "learning_rate": 3.751154201292706e-05,
+      "loss": 0.3505,
+      "step": 1216
+    },
+    {
+      "epoch": 0.5449762104673943,
+      "grad_norm": 1.140181541442871,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.2568,
+      "step": 1217
+    },
+    {
+      "epoch": 0.545424013434089,
+      "grad_norm": 1.115614414215088,
+      "learning_rate": 3.7488457987072946e-05,
+      "loss": 0.2442,
+      "step": 1218
+    },
+    {
+      "epoch": 0.5458718164007836,
+      "grad_norm": 1.0412161350250244,
+      "learning_rate": 3.7476915974145895e-05,
+      "loss": 0.2746,
+      "step": 1219
+    },
+    {
+      "epoch": 0.5463196193674783,
+      "grad_norm": 1.283544659614563,
+      "learning_rate": 3.746537396121884e-05,
+      "loss": 0.2452,
+      "step": 1220
+    },
+    {
+      "epoch": 0.5463196193674783,
+      "eval_loss": 0.34167274832725525,
+      "eval_runtime": 1738.5376,
+      "eval_samples_per_second": 2.569,
+      "eval_steps_per_second": 2.569,
+      "step": 1220
+    },
+    {
+      "epoch": 0.546767422334173,
+      "grad_norm": 1.2286393642425537,
+      "learning_rate": 3.745383194829179e-05,
+      "loss": 0.3005,
+      "step": 1221
+    },
+    {
+      "epoch": 0.5472152253008676,
+      "grad_norm": 1.4023370742797852,
+      "learning_rate": 3.744228993536473e-05,
+      "loss": 0.3721,
+      "step": 1222
+    },
+    {
+      "epoch": 0.5476630282675623,
+      "grad_norm": 1.4825446605682373,
+      "learning_rate": 3.743074792243767e-05,
+      "loss": 0.3667,
+      "step": 1223
+    },
+    {
+      "epoch": 0.548110831234257,
+      "grad_norm": 1.1614080667495728,
+      "learning_rate": 3.7419205909510615e-05,
+      "loss": 0.249,
+      "step": 1224
+    },
+    {
+      "epoch": 0.5485586342009516,
+      "grad_norm": 1.1903398036956787,
+      "learning_rate": 3.7407663896583565e-05,
+      "loss": 0.3123,
+      "step": 1225
+    },
+    {
+      "epoch": 0.5490064371676462,
+      "grad_norm": 1.2279646396636963,
+      "learning_rate": 3.739612188365651e-05,
+      "loss": 0.3751,
+      "step": 1226
+    },
+    {
+      "epoch": 0.5494542401343409,
+      "grad_norm": 1.272114634513855,
+      "learning_rate": 3.738457987072946e-05,
+      "loss": 0.2775,
+      "step": 1227
+    },
+    {
+      "epoch": 0.5499020431010355,
+      "grad_norm": 1.0231918096542358,
+      "learning_rate": 3.73730378578024e-05,
+      "loss": 0.2722,
+      "step": 1228
+    },
+    {
+      "epoch": 0.5503498460677302,
+      "grad_norm": 1.0256974697113037,
+      "learning_rate": 3.736149584487535e-05,
+      "loss": 0.3589,
+      "step": 1229
+    },
+    {
+      "epoch": 0.5507976490344249,
+      "grad_norm": 1.0507721900939941,
+      "learning_rate": 3.734995383194829e-05,
+      "loss": 0.2636,
+      "step": 1230
+    },
+    {
+      "epoch": 0.5507976490344249,
+      "eval_loss": 0.33938801288604736,
+      "eval_runtime": 1737.4881,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 1230
+    },
+    {
+      "epoch": 0.5512454520011195,
+      "grad_norm": 1.433797836303711,
+      "learning_rate": 3.733841181902124e-05,
+      "loss": 0.2577,
+      "step": 1231
+    },
+    {
+      "epoch": 0.5516932549678142,
+      "grad_norm": 1.160000205039978,
+      "learning_rate": 3.732686980609418e-05,
+      "loss": 0.3427,
+      "step": 1232
+    },
+    {
+      "epoch": 0.5521410579345088,
+      "grad_norm": 1.1534769535064697,
+      "learning_rate": 3.731532779316713e-05,
+      "loss": 0.2084,
+      "step": 1233
+    },
+    {
+      "epoch": 0.5525888609012035,
+      "grad_norm": 1.4578096866607666,
+      "learning_rate": 3.7303785780240075e-05,
+      "loss": 0.398,
+      "step": 1234
+    },
+    {
+      "epoch": 0.5530366638678981,
+      "grad_norm": 1.6965374946594238,
+      "learning_rate": 3.7292243767313025e-05,
+      "loss": 0.5219,
+      "step": 1235
+    },
+    {
+      "epoch": 0.5534844668345927,
+      "grad_norm": 1.1840635538101196,
+      "learning_rate": 3.728070175438597e-05,
+      "loss": 0.3324,
+      "step": 1236
+    },
+    {
+      "epoch": 0.5539322698012874,
+      "grad_norm": 1.2262353897094727,
+      "learning_rate": 3.726915974145892e-05,
+      "loss": 0.3844,
+      "step": 1237
+    },
+    {
+      "epoch": 0.5543800727679821,
+      "grad_norm": 1.120253086090088,
+      "learning_rate": 3.725761772853186e-05,
+      "loss": 0.2836,
+      "step": 1238
+    },
+    {
+      "epoch": 0.5548278757346767,
+      "grad_norm": 1.3321200609207153,
+      "learning_rate": 3.724607571560481e-05,
+      "loss": 0.4458,
+      "step": 1239
+    },
+    {
+      "epoch": 0.5552756787013714,
+      "grad_norm": 1.2431046962738037,
+      "learning_rate": 3.723453370267775e-05,
+      "loss": 0.3338,
+      "step": 1240
+    },
+    {
+      "epoch": 0.5552756787013714,
+      "eval_loss": 0.33455634117126465,
+      "eval_runtime": 1735.9972,
+      "eval_samples_per_second": 2.573,
+      "eval_steps_per_second": 2.573,
+      "step": 1240
+    },
+    {
+      "epoch": 0.5557234816680661,
+      "grad_norm": 1.113531231880188,
+      "learning_rate": 3.7222991689750694e-05,
+      "loss": 0.1868,
+      "step": 1241
+    },
+    {
+      "epoch": 0.5561712846347607,
+      "grad_norm": 1.1995195150375366,
+      "learning_rate": 3.7211449676823636e-05,
+      "loss": 0.5076,
+      "step": 1242
+    },
+    {
+      "epoch": 0.5566190876014554,
+      "grad_norm": 1.3281294107437134,
+      "learning_rate": 3.7199907663896586e-05,
+      "loss": 0.3138,
+      "step": 1243
+    },
+    {
+      "epoch": 0.5570668905681501,
+      "grad_norm": 1.101189136505127,
+      "learning_rate": 3.718836565096953e-05,
+      "loss": 0.3515,
+      "step": 1244
+    },
+    {
+      "epoch": 0.5575146935348446,
+      "grad_norm": 1.8178327083587646,
+      "learning_rate": 3.717682363804247e-05,
+      "loss": 0.2748,
+      "step": 1245
+    },
+    {
+      "epoch": 0.5579624965015393,
+      "grad_norm": 1.3038519620895386,
+      "learning_rate": 3.716528162511542e-05,
+      "loss": 0.2843,
+      "step": 1246
+    },
+    {
+      "epoch": 0.558410299468234,
+      "grad_norm": 1.1656179428100586,
+      "learning_rate": 3.715373961218836e-05,
+      "loss": 0.3318,
+      "step": 1247
+    },
+    {
+      "epoch": 0.5588581024349286,
+      "grad_norm": 1.280800223350525,
+      "learning_rate": 3.714219759926131e-05,
+      "loss": 0.4303,
+      "step": 1248
+    },
+    {
+      "epoch": 0.5593059054016233,
+      "grad_norm": 1.352533221244812,
+      "learning_rate": 3.7130655586334255e-05,
+      "loss": 0.3739,
+      "step": 1249
+    },
+    {
+      "epoch": 0.559753708368318,
+      "grad_norm": 1.1881428956985474,
+      "learning_rate": 3.7119113573407204e-05,
+      "loss": 0.1833,
+      "step": 1250
+    },
+    {
+      "epoch": 0.559753708368318,
+      "eval_loss": 0.33061400055885315,
+      "eval_runtime": 1730.969,
+      "eval_samples_per_second": 2.581,
+      "eval_steps_per_second": 2.581,
+      "step": 1250
+    },
+    {
+      "epoch": 0.5602015113350126,
+      "grad_norm": 0.903508722782135,
+      "learning_rate": 3.710757156048015e-05,
+      "loss": 0.1424,
+      "step": 1251
+    },
+    {
+      "epoch": 0.5606493143017073,
+      "grad_norm": 1.3883750438690186,
+      "learning_rate": 3.7096029547553096e-05,
+      "loss": 0.5074,
+      "step": 1252
+    },
+    {
+      "epoch": 0.561097117268402,
+      "grad_norm": 1.451582431793213,
+      "learning_rate": 3.708448753462604e-05,
+      "loss": 0.4956,
+      "step": 1253
+    },
+    {
+      "epoch": 0.5615449202350966,
+      "grad_norm": 1.3098926544189453,
+      "learning_rate": 3.707294552169899e-05,
+      "loss": 0.3404,
+      "step": 1254
+    },
+    {
+      "epoch": 0.5619927232017912,
+      "grad_norm": 1.2385728359222412,
+      "learning_rate": 3.706140350877193e-05,
+      "loss": 0.1917,
+      "step": 1255
+    },
+    {
+      "epoch": 0.5624405261684858,
+      "grad_norm": 1.0623195171356201,
+      "learning_rate": 3.704986149584488e-05,
+      "loss": 0.2205,
+      "step": 1256
+    },
+    {
+      "epoch": 0.5628883291351805,
+      "grad_norm": 1.0155680179595947,
+      "learning_rate": 3.703831948291782e-05,
+      "loss": 0.3373,
+      "step": 1257
+    },
+    {
+      "epoch": 0.5633361321018752,
+      "grad_norm": 1.3060396909713745,
+      "learning_rate": 3.702677746999077e-05,
+      "loss": 0.4576,
+      "step": 1258
+    },
+    {
+      "epoch": 0.5637839350685698,
+      "grad_norm": 0.9929831027984619,
+      "learning_rate": 3.7015235457063715e-05,
+      "loss": 0.1533,
+      "step": 1259
+    },
+    {
+      "epoch": 0.5642317380352645,
+      "grad_norm": 1.223260760307312,
+      "learning_rate": 3.7003693444136664e-05,
+      "loss": 0.3359,
+      "step": 1260
+    },
+    {
+      "epoch": 0.5642317380352645,
+      "eval_loss": 0.32611918449401855,
+      "eval_runtime": 1729.8713,
+      "eval_samples_per_second": 2.582,
+      "eval_steps_per_second": 2.582,
+      "step": 1260
+    },
+    {
+      "epoch": 0.5646795410019592,
+      "grad_norm": 1.2260444164276123,
+      "learning_rate": 3.699215143120961e-05,
+      "loss": 0.3395,
+      "step": 1261
+    },
+    {
+      "epoch": 0.5651273439686538,
+      "grad_norm": 1.2700910568237305,
+      "learning_rate": 3.698060941828255e-05,
+      "loss": 0.2942,
+      "step": 1262
+    },
+    {
+      "epoch": 0.5655751469353485,
+      "grad_norm": 1.1794884204864502,
+      "learning_rate": 3.696906740535549e-05,
+      "loss": 0.3885,
+      "step": 1263
+    },
+    {
+      "epoch": 0.566022949902043,
+      "grad_norm": 1.5495160818099976,
+      "learning_rate": 3.695752539242844e-05,
+      "loss": 0.442,
+      "step": 1264
+    },
+    {
+      "epoch": 0.5664707528687377,
+      "grad_norm": 1.0414493083953857,
+      "learning_rate": 3.6945983379501384e-05,
+      "loss": 0.227,
+      "step": 1265
+    },
+    {
+      "epoch": 0.5669185558354324,
+      "grad_norm": 1.0589414834976196,
+      "learning_rate": 3.6934441366574334e-05,
+      "loss": 0.211,
+      "step": 1266
+    },
+    {
+      "epoch": 0.567366358802127,
+      "grad_norm": 1.1238470077514648,
+      "learning_rate": 3.6922899353647276e-05,
+      "loss": 0.4103,
+      "step": 1267
+    },
+    {
+      "epoch": 0.5678141617688217,
+      "grad_norm": 1.4189252853393555,
+      "learning_rate": 3.691135734072022e-05,
+      "loss": 0.487,
+      "step": 1268
+    },
+    {
+      "epoch": 0.5682619647355164,
+      "grad_norm": 1.078750491142273,
+      "learning_rate": 3.689981532779317e-05,
+      "loss": 0.2973,
+      "step": 1269
+    },
+    {
+      "epoch": 0.568709767702211,
+      "grad_norm": 1.2650834321975708,
+      "learning_rate": 3.688827331486611e-05,
+      "loss": 0.2079,
+      "step": 1270
+    },
+    {
+      "epoch": 0.568709767702211,
+      "eval_loss": 0.32332947850227356,
+      "eval_runtime": 1738.8495,
+      "eval_samples_per_second": 2.569,
+      "eval_steps_per_second": 2.569,
+      "step": 1270
+    },
+    {
+      "epoch": 0.5691575706689057,
+      "grad_norm": 1.3321959972381592,
+      "learning_rate": 3.687673130193906e-05,
+      "loss": 0.3667,
+      "step": 1271
+    },
+    {
+      "epoch": 0.5696053736356004,
+      "grad_norm": 1.6029940843582153,
+      "learning_rate": 3.6865189289012e-05,
+      "loss": 0.4847,
+      "step": 1272
+    },
+    {
+      "epoch": 0.570053176602295,
+      "grad_norm": 1.2480456829071045,
+      "learning_rate": 3.685364727608495e-05,
+      "loss": 0.3311,
+      "step": 1273
+    },
+    {
+      "epoch": 0.5705009795689896,
+      "grad_norm": 1.4163548946380615,
+      "learning_rate": 3.6842105263157895e-05,
+      "loss": 0.2867,
+      "step": 1274
+    },
+    {
+      "epoch": 0.5709487825356843,
+      "grad_norm": 1.233312964439392,
+      "learning_rate": 3.6830563250230844e-05,
+      "loss": 0.2895,
+      "step": 1275
+    },
+    {
+      "epoch": 0.5713965855023789,
+      "grad_norm": 1.4446786642074585,
+      "learning_rate": 3.681902123730379e-05,
+      "loss": 0.3115,
+      "step": 1276
+    },
+    {
+      "epoch": 0.5718443884690736,
+      "grad_norm": 1.1519650220870972,
+      "learning_rate": 3.6807479224376736e-05,
+      "loss": 0.2677,
+      "step": 1277
+    },
+    {
+      "epoch": 0.5722921914357683,
+      "grad_norm": 1.3036447763442993,
+      "learning_rate": 3.679593721144968e-05,
+      "loss": 0.497,
+      "step": 1278
+    },
+    {
+      "epoch": 0.5727399944024629,
+      "grad_norm": 1.2030128240585327,
+      "learning_rate": 3.678439519852263e-05,
+      "loss": 0.2617,
+      "step": 1279
+    },
+    {
+      "epoch": 0.5731877973691576,
+      "grad_norm": 1.4113270044326782,
+      "learning_rate": 3.677285318559557e-05,
+      "loss": 0.3102,
+      "step": 1280
+    },
+    {
+      "epoch": 0.5731877973691576,
+      "eval_loss": 0.3202439546585083,
+      "eval_runtime": 1741.4969,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 2.565,
+      "step": 1280
+    },
+    {
+      "epoch": 0.5736356003358523,
+      "grad_norm": 1.2645702362060547,
+      "learning_rate": 3.676131117266851e-05,
+      "loss": 0.5035,
+      "step": 1281
+    },
+    {
+      "epoch": 0.5740834033025469,
+      "grad_norm": 0.8573437929153442,
+      "learning_rate": 3.674976915974146e-05,
+      "loss": 0.2647,
+      "step": 1282
+    },
+    {
+      "epoch": 0.5745312062692415,
+      "grad_norm": 1.2316564321517944,
+      "learning_rate": 3.6738227146814405e-05,
+      "loss": 0.3333,
+      "step": 1283
+    },
+    {
+      "epoch": 0.5749790092359361,
+      "grad_norm": 1.1025688648223877,
+      "learning_rate": 3.672668513388735e-05,
+      "loss": 0.4113,
+      "step": 1284
+    },
+    {
+      "epoch": 0.5754268122026308,
+      "grad_norm": 1.2769172191619873,
+      "learning_rate": 3.67151431209603e-05,
+      "loss": 0.3261,
+      "step": 1285
+    },
+    {
+      "epoch": 0.5758746151693255,
+      "grad_norm": 1.4074493646621704,
+      "learning_rate": 3.670360110803324e-05,
+      "loss": 0.5384,
+      "step": 1286
+    },
+    {
+      "epoch": 0.5763224181360201,
+      "grad_norm": 1.248610258102417,
+      "learning_rate": 3.669205909510619e-05,
+      "loss": 0.3164,
+      "step": 1287
+    },
+    {
+      "epoch": 0.5767702211027148,
+      "grad_norm": 1.2112908363342285,
+      "learning_rate": 3.668051708217913e-05,
+      "loss": 0.4445,
+      "step": 1288
+    },
+    {
+      "epoch": 0.5772180240694095,
+      "grad_norm": 1.2379544973373413,
+      "learning_rate": 3.666897506925208e-05,
+      "loss": 0.38,
+      "step": 1289
+    },
+    {
+      "epoch": 0.5776658270361041,
+      "grad_norm": 1.0747911930084229,
+      "learning_rate": 3.6657433056325024e-05,
+      "loss": 0.3254,
+      "step": 1290
+    },
+    {
+      "epoch": 0.5776658270361041,
+      "eval_loss": 0.3167678415775299,
+      "eval_runtime": 1742.3404,
+      "eval_samples_per_second": 2.564,
+      "eval_steps_per_second": 2.564,
+      "step": 1290
+    },
+    {
+      "epoch": 0.5781136300027988,
+      "grad_norm": 0.9601860046386719,
+      "learning_rate": 3.664589104339797e-05,
+      "loss": 0.183,
+      "step": 1291
+    },
+    {
+      "epoch": 0.5785614329694935,
+      "grad_norm": 0.8449709415435791,
+      "learning_rate": 3.6634349030470916e-05,
+      "loss": 0.1727,
+      "step": 1292
+    },
+    {
+      "epoch": 0.579009235936188,
+      "grad_norm": 1.3243083953857422,
+      "learning_rate": 3.662280701754386e-05,
+      "loss": 0.2925,
+      "step": 1293
+    },
+    {
+      "epoch": 0.5794570389028827,
+      "grad_norm": 1.3475184440612793,
+      "learning_rate": 3.661126500461681e-05,
+      "loss": 0.4296,
+      "step": 1294
+    },
+    {
+      "epoch": 0.5799048418695774,
+      "grad_norm": 1.3963221311569214,
+      "learning_rate": 3.659972299168975e-05,
+      "loss": 0.5055,
+      "step": 1295
+    },
+    {
+      "epoch": 0.580352644836272,
+      "grad_norm": 1.177854061126709,
+      "learning_rate": 3.65881809787627e-05,
+      "loss": 0.2422,
+      "step": 1296
+    },
+    {
+      "epoch": 0.5808004478029667,
+      "grad_norm": 1.038023591041565,
+      "learning_rate": 3.657663896583564e-05,
+      "loss": 0.1763,
+      "step": 1297
+    },
+    {
+      "epoch": 0.5812482507696614,
+      "grad_norm": 1.0623871088027954,
+      "learning_rate": 3.656509695290859e-05,
+      "loss": 0.2948,
+      "step": 1298
+    },
+    {
+      "epoch": 0.581696053736356,
+      "grad_norm": 1.0913971662521362,
+      "learning_rate": 3.6553554939981534e-05,
+      "loss": 0.2271,
+      "step": 1299
+    },
+    {
+      "epoch": 0.5821438567030507,
+      "grad_norm": 1.0985262393951416,
+      "learning_rate": 3.6542012927054484e-05,
+      "loss": 0.3481,
+      "step": 1300
+    },
+    {
+      "epoch": 0.5821438567030507,
+      "eval_loss": 0.3124310374259949,
+      "eval_runtime": 1741.5934,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 2.565,
+      "step": 1300
+    },
+    {
+      "epoch": 0.5825916596697454,
+      "grad_norm": 1.1117430925369263,
+      "learning_rate": 3.6530470914127426e-05,
+      "loss": 0.4351,
+      "step": 1301
+    },
+    {
+      "epoch": 0.58303946263644,
+      "grad_norm": 1.3771589994430542,
+      "learning_rate": 3.651892890120037e-05,
+      "loss": 0.2323,
+      "step": 1302
+    },
+    {
+      "epoch": 0.5834872656031346,
+      "grad_norm": 1.3689393997192383,
+      "learning_rate": 3.650738688827331e-05,
+      "loss": 0.2608,
+      "step": 1303
+    },
+    {
+      "epoch": 0.5839350685698292,
+      "grad_norm": 1.3155872821807861,
+      "learning_rate": 3.649584487534626e-05,
+      "loss": 0.3222,
+      "step": 1304
+    },
+    {
+      "epoch": 0.5843828715365239,
+      "grad_norm": 1.2287445068359375,
+      "learning_rate": 3.6484302862419204e-05,
+      "loss": 0.2483,
+      "step": 1305
+    },
+    {
+      "epoch": 0.5848306745032186,
+      "grad_norm": 1.4728074073791504,
+      "learning_rate": 3.647276084949215e-05,
+      "loss": 0.4875,
+      "step": 1306
+    },
+    {
+      "epoch": 0.5852784774699132,
+      "grad_norm": 0.9676113128662109,
+      "learning_rate": 3.6461218836565096e-05,
+      "loss": 0.2232,
+      "step": 1307
+    },
+    {
+      "epoch": 0.5857262804366079,
+      "grad_norm": 1.1757640838623047,
+      "learning_rate": 3.6449676823638045e-05,
+      "loss": 0.2085,
+      "step": 1308
+    },
+    {
+      "epoch": 0.5861740834033026,
+      "grad_norm": 1.0415418148040771,
+      "learning_rate": 3.643813481071099e-05,
+      "loss": 0.259,
+      "step": 1309
+    },
+    {
+      "epoch": 0.5866218863699972,
+      "grad_norm": 1.1401948928833008,
+      "learning_rate": 3.642659279778394e-05,
+      "loss": 0.2322,
+      "step": 1310
+    },
+    {
+      "epoch": 0.5866218863699972,
+      "eval_loss": 0.30859696865081787,
+      "eval_runtime": 1742.4133,
+      "eval_samples_per_second": 2.564,
+      "eval_steps_per_second": 2.564,
+      "step": 1310
+    },
+    {
+      "epoch": 0.5870696893366919,
+      "grad_norm": 1.0934462547302246,
+      "learning_rate": 3.641505078485688e-05,
+      "loss": 0.1186,
+      "step": 1311
+    },
+    {
+      "epoch": 0.5875174923033865,
+      "grad_norm": 1.0055665969848633,
+      "learning_rate": 3.640350877192983e-05,
+      "loss": 0.3582,
+      "step": 1312
+    },
+    {
+      "epoch": 0.5879652952700811,
+      "grad_norm": 1.4080592393875122,
+      "learning_rate": 3.639196675900277e-05,
+      "loss": 0.3942,
+      "step": 1313
+    },
+    {
+      "epoch": 0.5884130982367758,
+      "grad_norm": 1.1707267761230469,
+      "learning_rate": 3.638042474607572e-05,
+      "loss": 0.246,
+      "step": 1314
+    },
+    {
+      "epoch": 0.5888609012034705,
+      "grad_norm": 1.1650632619857788,
+      "learning_rate": 3.6368882733148664e-05,
+      "loss": 0.177,
+      "step": 1315
+    },
+    {
+      "epoch": 0.5893087041701651,
+      "grad_norm": 1.2004889249801636,
+      "learning_rate": 3.635734072022161e-05,
+      "loss": 0.2384,
+      "step": 1316
+    },
+    {
+      "epoch": 0.5897565071368598,
+      "grad_norm": 1.0767138004302979,
+      "learning_rate": 3.6345798707294556e-05,
+      "loss": 0.2802,
+      "step": 1317
+    },
+    {
+      "epoch": 0.5902043101035545,
+      "grad_norm": 1.1107332706451416,
+      "learning_rate": 3.63342566943675e-05,
+      "loss": 0.1619,
+      "step": 1318
+    },
+    {
+      "epoch": 0.5906521130702491,
+      "grad_norm": 1.2549272775650024,
+      "learning_rate": 3.632271468144045e-05,
+      "loss": 0.2781,
+      "step": 1319
+    },
+    {
+      "epoch": 0.5910999160369438,
+      "grad_norm": 1.3902312517166138,
+      "learning_rate": 3.631117266851339e-05,
+      "loss": 0.3365,
+      "step": 1320
+    },
+    {
+      "epoch": 0.5910999160369438,
+      "eval_loss": 0.3084462285041809,
+      "eval_runtime": 1741.5592,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 2.565,
+      "step": 1320
+    },
+    {
+      "epoch": 0.5915477190036385,
+      "grad_norm": 1.1804687976837158,
+      "learning_rate": 3.629963065558634e-05,
+      "loss": 0.2902,
+      "step": 1321
+    },
+    {
+      "epoch": 0.591995521970333,
+      "grad_norm": 0.9639450907707214,
+      "learning_rate": 3.628808864265928e-05,
+      "loss": 0.1469,
+      "step": 1322
+    },
+    {
+      "epoch": 0.5924433249370277,
+      "grad_norm": 1.1308844089508057,
+      "learning_rate": 3.6276546629732225e-05,
+      "loss": 0.2995,
+      "step": 1323
+    },
+    {
+      "epoch": 0.5928911279037223,
+      "grad_norm": 1.3623363971710205,
+      "learning_rate": 3.626500461680517e-05,
+      "loss": 0.4717,
+      "step": 1324
+    },
+    {
+      "epoch": 0.593338930870417,
+      "grad_norm": 1.0320011377334595,
+      "learning_rate": 3.625346260387812e-05,
+      "loss": 0.2517,
+      "step": 1325
+    },
+    {
+      "epoch": 0.5937867338371117,
+      "grad_norm": 1.0291073322296143,
+      "learning_rate": 3.624192059095106e-05,
+      "loss": 0.368,
+      "step": 1326
+    },
+    {
+      "epoch": 0.5942345368038063,
+      "grad_norm": 0.9665423631668091,
+      "learning_rate": 3.623037857802401e-05,
+      "loss": 0.2019,
+      "step": 1327
+    },
+    {
+      "epoch": 0.594682339770501,
+      "grad_norm": 1.0355550050735474,
+      "learning_rate": 3.621883656509695e-05,
+      "loss": 0.1839,
+      "step": 1328
+    },
+    {
+      "epoch": 0.5951301427371957,
+      "grad_norm": 1.0803695917129517,
+      "learning_rate": 3.62072945521699e-05,
+      "loss": 0.2234,
+      "step": 1329
+    },
+    {
+      "epoch": 0.5955779457038903,
+      "grad_norm": 1.0858778953552246,
+      "learning_rate": 3.619575253924284e-05,
+      "loss": 0.3195,
+      "step": 1330
+    },
+    {
+      "epoch": 0.5955779457038903,
+      "eval_loss": 0.3059205412864685,
+      "eval_runtime": 1742.4212,
+      "eval_samples_per_second": 2.564,
+      "eval_steps_per_second": 2.564,
+      "step": 1330
+    },
+    {
+      "epoch": 0.5960257486705849,
+      "grad_norm": 1.0849155187606812,
+      "learning_rate": 3.618421052631579e-05,
+      "loss": 0.2967,
+      "step": 1331
+    },
+    {
+      "epoch": 0.5964735516372796,
+      "grad_norm": 1.3284932374954224,
+      "learning_rate": 3.6172668513388735e-05,
+      "loss": 0.4775,
+      "step": 1332
+    },
+    {
+      "epoch": 0.5969213546039742,
+      "grad_norm": 1.0310250520706177,
+      "learning_rate": 3.6161126500461685e-05,
+      "loss": 0.2639,
+      "step": 1333
+    },
+    {
+      "epoch": 0.5973691575706689,
+      "grad_norm": 1.4501495361328125,
+      "learning_rate": 3.614958448753463e-05,
+      "loss": 0.4214,
+      "step": 1334
+    },
+    {
+      "epoch": 0.5978169605373636,
+      "grad_norm": 0.9147229194641113,
+      "learning_rate": 3.613804247460758e-05,
+      "loss": 0.1917,
+      "step": 1335
+    },
+    {
+      "epoch": 0.5982647635040582,
+      "grad_norm": 1.262824296951294,
+      "learning_rate": 3.612650046168052e-05,
+      "loss": 0.2901,
+      "step": 1336
+    },
+    {
+      "epoch": 0.5987125664707529,
+      "grad_norm": 1.1430469751358032,
+      "learning_rate": 3.611495844875347e-05,
+      "loss": 0.3782,
+      "step": 1337
+    },
+    {
+      "epoch": 0.5991603694374475,
+      "grad_norm": 1.228498101234436,
+      "learning_rate": 3.610341643582641e-05,
+      "loss": 0.3737,
+      "step": 1338
+    },
+    {
+      "epoch": 0.5996081724041422,
+      "grad_norm": 1.1623265743255615,
+      "learning_rate": 3.609187442289936e-05,
+      "loss": 0.1474,
+      "step": 1339
+    },
+    {
+      "epoch": 0.6000559753708369,
+      "grad_norm": 1.1078261137008667,
+      "learning_rate": 3.60803324099723e-05,
+      "loss": 0.2201,
+      "step": 1340
+    },
+    {
+      "epoch": 0.6000559753708369,
+      "eval_loss": 0.3014572560787201,
+      "eval_runtime": 1741.5342,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 2.565,
+      "step": 1340
+    },
+    {
+      "epoch": 0.6005037783375314,
+      "grad_norm": 1.2032644748687744,
+      "learning_rate": 3.6068790397045246e-05,
+      "loss": 0.1858,
+      "step": 1341
+    },
+    {
+      "epoch": 0.6009515813042261,
+      "grad_norm": 0.9365187883377075,
+      "learning_rate": 3.605724838411819e-05,
+      "loss": 0.2956,
+      "step": 1342
+    },
+    {
+      "epoch": 0.6013993842709208,
+      "grad_norm": 1.3125098943710327,
+      "learning_rate": 3.604570637119114e-05,
+      "loss": 0.3214,
+      "step": 1343
+    },
+    {
+      "epoch": 0.6018471872376154,
+      "grad_norm": 1.2828110456466675,
+      "learning_rate": 3.603416435826408e-05,
+      "loss": 0.3273,
+      "step": 1344
+    },
+    {
+      "epoch": 0.6022949902043101,
+      "grad_norm": 1.0955266952514648,
+      "learning_rate": 3.602262234533702e-05,
+      "loss": 0.1819,
+      "step": 1345
+    },
+    {
+      "epoch": 0.6027427931710048,
+      "grad_norm": 1.1152938604354858,
+      "learning_rate": 3.601108033240997e-05,
+      "loss": 0.2204,
+      "step": 1346
+    },
+    {
+      "epoch": 0.6031905961376994,
+      "grad_norm": 1.3604631423950195,
+      "learning_rate": 3.5999538319482915e-05,
+      "loss": 0.2325,
+      "step": 1347
+    },
+    {
+      "epoch": 0.6036383991043941,
+      "grad_norm": 1.0130641460418701,
+      "learning_rate": 3.5987996306555864e-05,
+      "loss": 0.2148,
+      "step": 1348
+    },
+    {
+      "epoch": 0.6040862020710888,
+      "grad_norm": 0.9263889193534851,
+      "learning_rate": 3.597645429362881e-05,
+      "loss": 0.1883,
+      "step": 1349
+    },
+    {
+      "epoch": 0.6045340050377834,
+      "grad_norm": 1.4137290716171265,
+      "learning_rate": 3.5964912280701756e-05,
+      "loss": 0.3185,
+      "step": 1350
+    },
+    {
+      "epoch": 0.6045340050377834,
+      "eval_loss": 0.2987724840641022,
+      "eval_runtime": 1740.5422,
+      "eval_samples_per_second": 2.566,
+      "eval_steps_per_second": 2.566,
+      "step": 1350
+    },
+    {
+      "epoch": 0.604981808004478,
+      "grad_norm": 1.3324114084243774,
+      "learning_rate": 3.59533702677747e-05,
+      "loss": 0.2055,
+      "step": 1351
+    },
+    {
+      "epoch": 0.6054296109711726,
+      "grad_norm": 1.4274815320968628,
+      "learning_rate": 3.594182825484765e-05,
+      "loss": 0.434,
+      "step": 1352
+    },
+    {
+      "epoch": 0.6058774139378673,
+      "grad_norm": 1.538460612297058,
+      "learning_rate": 3.593028624192059e-05,
+      "loss": 0.4736,
+      "step": 1353
+    },
+    {
+      "epoch": 0.606325216904562,
+      "grad_norm": 1.4670627117156982,
+      "learning_rate": 3.591874422899354e-05,
+      "loss": 0.5298,
+      "step": 1354
+    },
+    {
+      "epoch": 0.6067730198712566,
+      "grad_norm": 1.2798479795455933,
+      "learning_rate": 3.590720221606648e-05,
+      "loss": 0.3156,
+      "step": 1355
+    },
+    {
+      "epoch": 0.6072208228379513,
+      "grad_norm": 1.0595784187316895,
+      "learning_rate": 3.589566020313943e-05,
+      "loss": 0.2297,
+      "step": 1356
+    },
+    {
+      "epoch": 0.607668625804646,
+      "grad_norm": 1.2012929916381836,
+      "learning_rate": 3.5884118190212375e-05,
+      "loss": 0.3495,
+      "step": 1357
+    },
+    {
+      "epoch": 0.6081164287713406,
+      "grad_norm": 1.1378802061080933,
+      "learning_rate": 3.5872576177285324e-05,
+      "loss": 0.186,
+      "step": 1358
+    },
+    {
+      "epoch": 0.6085642317380353,
+      "grad_norm": 1.5004686117172241,
+      "learning_rate": 3.586103416435827e-05,
+      "loss": 0.245,
+      "step": 1359
+    },
+    {
+      "epoch": 0.6090120347047299,
+      "grad_norm": 1.2461063861846924,
+      "learning_rate": 3.5849492151431216e-05,
+      "loss": 0.3392,
+      "step": 1360
+    },
+    {
+      "epoch": 0.6090120347047299,
+      "eval_loss": 0.29492369294166565,
+      "eval_runtime": 1741.3422,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 2.565,
+      "step": 1360
+    },
+    {
+      "epoch": 0.6094598376714245,
+      "grad_norm": 1.1116904020309448,
+      "learning_rate": 3.583795013850416e-05,
+      "loss": 0.496,
+      "step": 1361
+    },
+    {
+      "epoch": 0.6099076406381192,
+      "grad_norm": 1.0401750802993774,
+      "learning_rate": 3.58264081255771e-05,
+      "loss": 0.2223,
+      "step": 1362
+    },
+    {
+      "epoch": 0.6103554436048139,
+      "grad_norm": 1.5595612525939941,
+      "learning_rate": 3.5814866112650044e-05,
+      "loss": 0.2784,
+      "step": 1363
+    },
+    {
+      "epoch": 0.6108032465715085,
+      "grad_norm": 1.3437434434890747,
+      "learning_rate": 3.5803324099722994e-05,
+      "loss": 0.3927,
+      "step": 1364
+    },
+    {
+      "epoch": 0.6112510495382032,
+      "grad_norm": 1.167486548423767,
+      "learning_rate": 3.5791782086795936e-05,
+      "loss": 0.3245,
+      "step": 1365
+    },
+    {
+      "epoch": 0.6116988525048979,
+      "grad_norm": 1.1190348863601685,
+      "learning_rate": 3.5780240073868886e-05,
+      "loss": 0.2274,
+      "step": 1366
+    },
+    {
+      "epoch": 0.6121466554715925,
+      "grad_norm": 1.1168073415756226,
+      "learning_rate": 3.576869806094183e-05,
+      "loss": 0.2366,
+      "step": 1367
+    },
+    {
+      "epoch": 0.6125944584382872,
+      "grad_norm": 1.1127575635910034,
+      "learning_rate": 3.575715604801478e-05,
+      "loss": 0.2578,
+      "step": 1368
+    },
+    {
+      "epoch": 0.6130422614049819,
+      "grad_norm": 1.1270995140075684,
+      "learning_rate": 3.574561403508772e-05,
+      "loss": 0.1832,
+      "step": 1369
+    },
+    {
+      "epoch": 0.6134900643716764,
+      "grad_norm": 1.4525073766708374,
+      "learning_rate": 3.573407202216066e-05,
+      "loss": 0.1798,
+      "step": 1370
+    },
+    {
+      "epoch": 0.6134900643716764,
+      "eval_loss": 0.2917921543121338,
+      "eval_runtime": 1741.2161,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 2.565,
+      "step": 1370
+    },
+    {
+      "epoch": 0.6139378673383711,
+      "grad_norm": 1.565889596939087,
+      "learning_rate": 3.572253000923361e-05,
+      "loss": 0.3886,
+      "step": 1371
+    },
+    {
+      "epoch": 0.6143856703050657,
+      "grad_norm": 1.1590737104415894,
+      "learning_rate": 3.5710987996306555e-05,
+      "loss": 0.262,
+      "step": 1372
+    },
+    {
+      "epoch": 0.6148334732717604,
+      "grad_norm": 1.444347620010376,
+      "learning_rate": 3.5699445983379504e-05,
+      "loss": 0.2706,
+      "step": 1373
+    },
+    {
+      "epoch": 0.6152812762384551,
+      "grad_norm": 1.4933663606643677,
+      "learning_rate": 3.568790397045245e-05,
+      "loss": 0.3262,
+      "step": 1374
+    },
+    {
+      "epoch": 0.6157290792051497,
+      "grad_norm": 0.8957858681678772,
+      "learning_rate": 3.5676361957525396e-05,
+      "loss": 0.1987,
+      "step": 1375
+    },
+    {
+      "epoch": 0.6161768821718444,
+      "grad_norm": 1.3272017240524292,
+      "learning_rate": 3.566481994459834e-05,
+      "loss": 0.2494,
+      "step": 1376
+    },
+    {
+      "epoch": 0.6166246851385391,
+      "grad_norm": 1.256388545036316,
+      "learning_rate": 3.565327793167129e-05,
+      "loss": 0.4452,
+      "step": 1377
+    },
+    {
+      "epoch": 0.6170724881052337,
+      "grad_norm": 1.1355270147323608,
+      "learning_rate": 3.564173591874423e-05,
+      "loss": 0.2524,
+      "step": 1378
+    },
+    {
+      "epoch": 0.6175202910719284,
+      "grad_norm": 1.2793563604354858,
+      "learning_rate": 3.563019390581718e-05,
+      "loss": 0.3375,
+      "step": 1379
+    },
+    {
+      "epoch": 0.617968094038623,
+      "grad_norm": 1.4125409126281738,
+      "learning_rate": 3.561865189289012e-05,
+      "loss": 0.4409,
+      "step": 1380
+    },
+    {
+      "epoch": 0.617968094038623,
+      "eval_loss": 0.2876901924610138,
+      "eval_runtime": 1737.7665,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 1380
+    },
+    {
+      "epoch": 0.6184158970053176,
+      "grad_norm": 0.9987519979476929,
+      "learning_rate": 3.5607109879963065e-05,
+      "loss": 0.2368,
+      "step": 1381
+    },
+    {
+      "epoch": 0.6188636999720123,
+      "grad_norm": 1.0614521503448486,
+      "learning_rate": 3.5595567867036015e-05,
+      "loss": 0.2734,
+      "step": 1382
+    },
+    {
+      "epoch": 0.619311502938707,
+      "grad_norm": 0.9792033433914185,
+      "learning_rate": 3.558402585410896e-05,
+      "loss": 0.1667,
+      "step": 1383
+    },
+    {
+      "epoch": 0.6197593059054016,
+      "grad_norm": 1.4053237438201904,
+      "learning_rate": 3.55724838411819e-05,
+      "loss": 0.3517,
+      "step": 1384
+    },
+    {
+      "epoch": 0.6202071088720963,
+      "grad_norm": 1.050905466079712,
+      "learning_rate": 3.556094182825485e-05,
+      "loss": 0.304,
+      "step": 1385
+    },
+    {
+      "epoch": 0.620654911838791,
+      "grad_norm": 1.0837103128433228,
+      "learning_rate": 3.554939981532779e-05,
+      "loss": 0.1562,
+      "step": 1386
+    },
+    {
+      "epoch": 0.6211027148054856,
+      "grad_norm": 1.3736618757247925,
+      "learning_rate": 3.553785780240074e-05,
+      "loss": 0.2442,
+      "step": 1387
+    },
+    {
+      "epoch": 0.6215505177721803,
+      "grad_norm": 1.0219933986663818,
+      "learning_rate": 3.5526315789473684e-05,
+      "loss": 0.1763,
+      "step": 1388
+    },
+    {
+      "epoch": 0.6219983207388748,
+      "grad_norm": 1.1732109785079956,
+      "learning_rate": 3.551477377654663e-05,
+      "loss": 0.2701,
+      "step": 1389
+    },
+    {
+      "epoch": 0.6224461237055695,
+      "grad_norm": 1.294912576675415,
+      "learning_rate": 3.5503231763619576e-05,
+      "loss": 0.2805,
+      "step": 1390
+    },
+    {
+      "epoch": 0.6224461237055695,
+      "eval_loss": 0.28569599986076355,
+      "eval_runtime": 1737.3228,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 1390
+    },
+    {
+      "epoch": 0.6228939266722642,
+      "grad_norm": 0.9582709074020386,
+      "learning_rate": 3.5491689750692525e-05,
+      "loss": 0.2353,
+      "step": 1391
+    },
+    {
+      "epoch": 0.6233417296389588,
+      "grad_norm": 1.2713261842727661,
+      "learning_rate": 3.548014773776547e-05,
+      "loss": 0.3263,
+      "step": 1392
+    },
+    {
+      "epoch": 0.6237895326056535,
+      "grad_norm": 0.983315110206604,
+      "learning_rate": 3.546860572483842e-05,
+      "loss": 0.4056,
+      "step": 1393
+    },
+    {
+      "epoch": 0.6242373355723482,
+      "grad_norm": 1.2925885915756226,
+      "learning_rate": 3.545706371191136e-05,
+      "loss": 0.3493,
+      "step": 1394
+    },
+    {
+      "epoch": 0.6246851385390428,
+      "grad_norm": 1.1899527311325073,
+      "learning_rate": 3.54455216989843e-05,
+      "loss": 0.2079,
+      "step": 1395
+    },
+    {
+      "epoch": 0.6251329415057375,
+      "grad_norm": 1.4036329984664917,
+      "learning_rate": 3.543397968605725e-05,
+      "loss": 0.3941,
+      "step": 1396
+    },
+    {
+      "epoch": 0.6255807444724322,
+      "grad_norm": 1.3126832246780396,
+      "learning_rate": 3.5422437673130195e-05,
+      "loss": 0.2855,
+      "step": 1397
+    },
+    {
+      "epoch": 0.6260285474391268,
+      "grad_norm": 0.9315616488456726,
+      "learning_rate": 3.5410895660203144e-05,
+      "loss": 0.1829,
+      "step": 1398
+    },
+    {
+      "epoch": 0.6264763504058214,
+      "grad_norm": 1.2371937036514282,
+      "learning_rate": 3.5399353647276087e-05,
+      "loss": 0.3544,
+      "step": 1399
+    },
+    {
+      "epoch": 0.6269241533725161,
+      "grad_norm": 1.414616346359253,
+      "learning_rate": 3.5387811634349036e-05,
+      "loss": 0.2595,
+      "step": 1400
+    },
+    {
+      "epoch": 0.6269241533725161,
+      "eval_loss": 0.28282201290130615,
+      "eval_runtime": 1742.7887,
+      "eval_samples_per_second": 2.563,
+      "eval_steps_per_second": 2.563,
+      "step": 1400
+    },
+    {
+      "epoch": 0.6273719563392107,
+      "grad_norm": 1.3357312679290771,
+      "learning_rate": 3.537626962142198e-05,
+      "loss": 0.3193,
+      "step": 1401
+    },
+    {
+      "epoch": 0.6278197593059054,
+      "grad_norm": 1.2589373588562012,
+      "learning_rate": 3.536472760849492e-05,
+      "loss": 0.2254,
+      "step": 1402
+    },
+    {
+      "epoch": 0.6282675622726,
+      "grad_norm": 1.0175939798355103,
+      "learning_rate": 3.5353185595567864e-05,
+      "loss": 0.1809,
+      "step": 1403
+    },
+    {
+      "epoch": 0.6287153652392947,
+      "grad_norm": 1.132478952407837,
+      "learning_rate": 3.534164358264081e-05,
+      "loss": 0.2762,
+      "step": 1404
+    },
+    {
+      "epoch": 0.6291631682059894,
+      "grad_norm": 1.0150865316390991,
+      "learning_rate": 3.5330101569713756e-05,
+      "loss": 0.1599,
+      "step": 1405
+    },
+    {
+      "epoch": 0.629610971172684,
+      "grad_norm": 1.0766525268554688,
+      "learning_rate": 3.5318559556786705e-05,
+      "loss": 0.4027,
+      "step": 1406
+    },
+    {
+      "epoch": 0.6300587741393787,
+      "grad_norm": 1.0079785585403442,
+      "learning_rate": 3.530701754385965e-05,
+      "loss": 0.1424,
+      "step": 1407
+    },
+    {
+      "epoch": 0.6305065771060733,
+      "grad_norm": 1.1688034534454346,
+      "learning_rate": 3.52954755309326e-05,
+      "loss": 0.2678,
+      "step": 1408
+    },
+    {
+      "epoch": 0.6309543800727679,
+      "grad_norm": 1.4623137712478638,
+      "learning_rate": 3.528393351800554e-05,
+      "loss": 0.3958,
+      "step": 1409
+    },
+    {
+      "epoch": 0.6314021830394626,
+      "grad_norm": 1.6712130308151245,
+      "learning_rate": 3.527239150507849e-05,
+      "loss": 0.4481,
+      "step": 1410
+    },
+    {
+      "epoch": 0.6314021830394626,
+      "eval_loss": 0.28085920214653015,
+      "eval_runtime": 1744.3896,
+      "eval_samples_per_second": 2.561,
+      "eval_steps_per_second": 2.561,
+      "step": 1410
+    },
+    {
+      "epoch": 0.6318499860061573,
+      "grad_norm": 1.0096598863601685,
+      "learning_rate": 3.526084949215143e-05,
+      "loss": 0.1529,
+      "step": 1411
+    },
+    {
+      "epoch": 0.6322977889728519,
+      "grad_norm": 1.2336546182632446,
+      "learning_rate": 3.524930747922438e-05,
+      "loss": 0.3192,
+      "step": 1412
+    },
+    {
+      "epoch": 0.6327455919395466,
+      "grad_norm": 1.1255513429641724,
+      "learning_rate": 3.5237765466297324e-05,
+      "loss": 0.1696,
+      "step": 1413
+    },
+    {
+      "epoch": 0.6331933949062413,
+      "grad_norm": 1.4015638828277588,
+      "learning_rate": 3.522622345337027e-05,
+      "loss": 0.3031,
+      "step": 1414
+    },
+    {
+      "epoch": 0.6336411978729359,
+      "grad_norm": 1.450169324874878,
+      "learning_rate": 3.5214681440443216e-05,
+      "loss": 0.3528,
+      "step": 1415
+    },
+    {
+      "epoch": 0.6340890008396306,
+      "grad_norm": 1.0433216094970703,
+      "learning_rate": 3.5203139427516165e-05,
+      "loss": 0.1812,
+      "step": 1416
+    },
+    {
+      "epoch": 0.6345368038063253,
+      "grad_norm": 1.7097972631454468,
+      "learning_rate": 3.519159741458911e-05,
+      "loss": 0.4827,
+      "step": 1417
+    },
+    {
+      "epoch": 0.6349846067730198,
+      "grad_norm": 1.2833137512207031,
+      "learning_rate": 3.518005540166206e-05,
+      "loss": 0.3412,
+      "step": 1418
+    },
+    {
+      "epoch": 0.6354324097397145,
+      "grad_norm": 1.1409881114959717,
+      "learning_rate": 3.5168513388735e-05,
+      "loss": 0.2976,
+      "step": 1419
+    },
+    {
+      "epoch": 0.6358802127064092,
+      "grad_norm": 1.182773232460022,
+      "learning_rate": 3.515697137580794e-05,
+      "loss": 0.2242,
+      "step": 1420
+    },
+    {
+      "epoch": 0.6358802127064092,
+      "eval_loss": 0.2783552408218384,
+      "eval_runtime": 1744.4292,
+      "eval_samples_per_second": 2.561,
+      "eval_steps_per_second": 2.561,
+      "step": 1420
+    },
+    {
+      "epoch": 0.6363280156731038,
+      "grad_norm": 1.0328185558319092,
+      "learning_rate": 3.514542936288089e-05,
+      "loss": 0.2419,
+      "step": 1421
+    },
+    {
+      "epoch": 0.6367758186397985,
+      "grad_norm": 1.4960310459136963,
+      "learning_rate": 3.5133887349953834e-05,
+      "loss": 0.3231,
+      "step": 1422
+    },
+    {
+      "epoch": 0.6372236216064932,
+      "grad_norm": 0.8000389337539673,
+      "learning_rate": 3.512234533702678e-05,
+      "loss": 0.1323,
+      "step": 1423
+    },
+    {
+      "epoch": 0.6376714245731878,
+      "grad_norm": 0.9059187173843384,
+      "learning_rate": 3.511080332409972e-05,
+      "loss": 0.1372,
+      "step": 1424
+    },
+    {
+      "epoch": 0.6381192275398825,
+      "grad_norm": 1.1321302652359009,
+      "learning_rate": 3.509926131117267e-05,
+      "loss": 0.2905,
+      "step": 1425
+    },
+    {
+      "epoch": 0.6385670305065771,
+      "grad_norm": 1.164314866065979,
+      "learning_rate": 3.508771929824561e-05,
+      "loss": 0.3124,
+      "step": 1426
+    },
+    {
+      "epoch": 0.6390148334732718,
+      "grad_norm": 1.1748840808868408,
+      "learning_rate": 3.507617728531856e-05,
+      "loss": 0.3135,
+      "step": 1427
+    },
+    {
+      "epoch": 0.6394626364399664,
+      "grad_norm": 1.3437206745147705,
+      "learning_rate": 3.5064635272391503e-05,
+      "loss": 0.3196,
+      "step": 1428
+    },
+    {
+      "epoch": 0.639910439406661,
+      "grad_norm": 1.161334753036499,
+      "learning_rate": 3.505309325946445e-05,
+      "loss": 0.2674,
+      "step": 1429
+    },
+    {
+      "epoch": 0.6403582423733557,
+      "grad_norm": 1.2514257431030273,
+      "learning_rate": 3.5041551246537395e-05,
+      "loss": 0.2372,
+      "step": 1430
+    },
+    {
+      "epoch": 0.6403582423733557,
+      "eval_loss": 0.27350693941116333,
+      "eval_runtime": 1743.9261,
+      "eval_samples_per_second": 2.561,
+      "eval_steps_per_second": 2.561,
+      "step": 1430
+    },
+    {
+      "epoch": 0.6408060453400504,
+      "grad_norm": 1.1305255889892578,
+      "learning_rate": 3.5030009233610345e-05,
+      "loss": 0.2812,
+      "step": 1431
+    },
+    {
+      "epoch": 0.641253848306745,
+      "grad_norm": 1.2445604801177979,
+      "learning_rate": 3.501846722068329e-05,
+      "loss": 0.1822,
+      "step": 1432
+    },
+    {
+      "epoch": 0.6417016512734397,
+      "grad_norm": 1.1391922235488892,
+      "learning_rate": 3.500692520775624e-05,
+      "loss": 0.257,
+      "step": 1433
+    },
+    {
+      "epoch": 0.6421494542401344,
+      "grad_norm": 1.5449109077453613,
+      "learning_rate": 3.499538319482918e-05,
+      "loss": 0.1946,
+      "step": 1434
+    },
+    {
+      "epoch": 0.642597257206829,
+      "grad_norm": 1.9147025346755981,
+      "learning_rate": 3.498384118190213e-05,
+      "loss": 0.3978,
+      "step": 1435
+    },
+    {
+      "epoch": 0.6430450601735237,
+      "grad_norm": 1.0643846988677979,
+      "learning_rate": 3.497229916897507e-05,
+      "loss": 0.2178,
+      "step": 1436
+    },
+    {
+      "epoch": 0.6434928631402183,
+      "grad_norm": 1.2681896686553955,
+      "learning_rate": 3.496075715604802e-05,
+      "loss": 0.1863,
+      "step": 1437
+    },
+    {
+      "epoch": 0.6439406661069129,
+      "grad_norm": 1.3740264177322388,
+      "learning_rate": 3.494921514312096e-05,
+      "loss": 0.1325,
+      "step": 1438
+    },
+    {
+      "epoch": 0.6443884690736076,
+      "grad_norm": 1.1765505075454712,
+      "learning_rate": 3.493767313019391e-05,
+      "loss": 0.3691,
+      "step": 1439
+    },
+    {
+      "epoch": 0.6448362720403022,
+      "grad_norm": 1.2055754661560059,
+      "learning_rate": 3.4926131117266855e-05,
+      "loss": 0.2446,
+      "step": 1440
+    },
+    {
+      "epoch": 0.6448362720403022,
+      "eval_loss": 0.27113863825798035,
+      "eval_runtime": 1744.5256,
+      "eval_samples_per_second": 2.561,
+      "eval_steps_per_second": 2.561,
+      "step": 1440
+    },
+    {
+      "epoch": 0.6452840750069969,
+      "grad_norm": 0.9709833860397339,
+      "learning_rate": 3.49145891043398e-05,
+      "loss": 0.2387,
+      "step": 1441
+    },
+    {
+      "epoch": 0.6457318779736916,
+      "grad_norm": 1.0405006408691406,
+      "learning_rate": 3.490304709141274e-05,
+      "loss": 0.1372,
+      "step": 1442
+    },
+    {
+      "epoch": 0.6461796809403862,
+      "grad_norm": 0.8977200984954834,
+      "learning_rate": 3.489150507848569e-05,
+      "loss": 0.1927,
+      "step": 1443
+    },
+    {
+      "epoch": 0.6466274839070809,
+      "grad_norm": 1.368783712387085,
+      "learning_rate": 3.487996306555863e-05,
+      "loss": 0.3845,
+      "step": 1444
+    },
+    {
+      "epoch": 0.6470752868737756,
+      "grad_norm": 0.966835081577301,
+      "learning_rate": 3.4868421052631575e-05,
+      "loss": 0.2843,
+      "step": 1445
+    },
+    {
+      "epoch": 0.6475230898404702,
+      "grad_norm": 1.244189977645874,
+      "learning_rate": 3.4856879039704525e-05,
+      "loss": 0.2754,
+      "step": 1446
+    },
+    {
+      "epoch": 0.6479708928071648,
+      "grad_norm": 1.1181278228759766,
+      "learning_rate": 3.484533702677747e-05,
+      "loss": 0.2702,
+      "step": 1447
+    },
+    {
+      "epoch": 0.6484186957738595,
+      "grad_norm": 1.1835464239120483,
+      "learning_rate": 3.4833795013850417e-05,
+      "loss": 0.3099,
+      "step": 1448
+    },
+    {
+      "epoch": 0.6488664987405541,
+      "grad_norm": 1.0337269306182861,
+      "learning_rate": 3.482225300092336e-05,
+      "loss": 0.1738,
+      "step": 1449
+    },
+    {
+      "epoch": 0.6493143017072488,
+      "grad_norm": 1.0362787246704102,
+      "learning_rate": 3.481071098799631e-05,
+      "loss": 0.3784,
+      "step": 1450
+    },
+    {
+      "epoch": 0.6493143017072488,
+      "eval_loss": 0.26652002334594727,
+      "eval_runtime": 1745.891,
+      "eval_samples_per_second": 2.559,
+      "eval_steps_per_second": 2.559,
+      "step": 1450
+    },
+    {
+      "epoch": 0.6497621046739435,
+      "grad_norm": 1.4951701164245605,
+      "learning_rate": 3.479916897506925e-05,
+      "loss": 0.3286,
+      "step": 1451
+    },
+    {
+      "epoch": 0.6502099076406381,
+      "grad_norm": 1.4321891069412231,
+      "learning_rate": 3.47876269621422e-05,
+      "loss": 0.418,
+      "step": 1452
+    },
+    {
+      "epoch": 0.6506577106073328,
+      "grad_norm": 1.2720353603363037,
+      "learning_rate": 3.477608494921514e-05,
+      "loss": 0.2375,
+      "step": 1453
+    },
+    {
+      "epoch": 0.6511055135740275,
+      "grad_norm": 0.9470394253730774,
+      "learning_rate": 3.476454293628809e-05,
+      "loss": 0.1657,
+      "step": 1454
+    },
+    {
+      "epoch": 0.6515533165407221,
+      "grad_norm": 1.3401026725769043,
+      "learning_rate": 3.4753000923361035e-05,
+      "loss": 0.2846,
+      "step": 1455
+    },
+    {
+      "epoch": 0.6520011195074167,
+      "grad_norm": 0.9690277576446533,
+      "learning_rate": 3.4741458910433985e-05,
+      "loss": 0.1218,
+      "step": 1456
+    },
+    {
+      "epoch": 0.6524489224741113,
+      "grad_norm": 1.0513359308242798,
+      "learning_rate": 3.472991689750693e-05,
+      "loss": 0.193,
+      "step": 1457
+    },
+    {
+      "epoch": 0.652896725440806,
+      "grad_norm": 1.2486155033111572,
+      "learning_rate": 3.4718374884579877e-05,
+      "loss": 0.4864,
+      "step": 1458
+    },
+    {
+      "epoch": 0.6533445284075007,
+      "grad_norm": 1.1349366903305054,
+      "learning_rate": 3.470683287165282e-05,
+      "loss": 0.1879,
+      "step": 1459
+    },
+    {
+      "epoch": 0.6537923313741953,
+      "grad_norm": 1.3820996284484863,
+      "learning_rate": 3.469529085872577e-05,
+      "loss": 0.4345,
+      "step": 1460
+    },
+    {
+      "epoch": 0.6537923313741953,
+      "eval_loss": 0.2638843357563019,
+      "eval_runtime": 1741.2113,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 2.565,
+      "step": 1460
+    },
+    {
+      "epoch": 0.65424013434089,
+      "grad_norm": 1.2622342109680176,
+      "learning_rate": 3.468374884579871e-05,
+      "loss": 0.2013,
+      "step": 1461
+    },
+    {
+      "epoch": 0.6546879373075847,
+      "grad_norm": 0.8820511698722839,
+      "learning_rate": 3.4672206832871654e-05,
+      "loss": 0.1695,
+      "step": 1462
+    },
+    {
+      "epoch": 0.6551357402742793,
+      "grad_norm": 1.4605138301849365,
+      "learning_rate": 3.4660664819944596e-05,
+      "loss": 0.2984,
+      "step": 1463
+    },
+    {
+      "epoch": 0.655583543240974,
+      "grad_norm": 0.8579814434051514,
+      "learning_rate": 3.4649122807017546e-05,
+      "loss": 0.1279,
+      "step": 1464
+    },
+    {
+      "epoch": 0.6560313462076687,
+      "grad_norm": 1.398829460144043,
+      "learning_rate": 3.463758079409049e-05,
+      "loss": 0.3021,
+      "step": 1465
+    },
+    {
+      "epoch": 0.6564791491743632,
+      "grad_norm": 1.234954833984375,
+      "learning_rate": 3.462603878116344e-05,
+      "loss": 0.1898,
+      "step": 1466
+    },
+    {
+      "epoch": 0.6569269521410579,
+      "grad_norm": 1.1851727962493896,
+      "learning_rate": 3.461449676823638e-05,
+      "loss": 0.2184,
+      "step": 1467
+    },
+    {
+      "epoch": 0.6573747551077526,
+      "grad_norm": 1.3027613162994385,
+      "learning_rate": 3.460295475530933e-05,
+      "loss": 0.218,
+      "step": 1468
+    },
+    {
+      "epoch": 0.6578225580744472,
+      "grad_norm": 1.4629859924316406,
+      "learning_rate": 3.459141274238227e-05,
+      "loss": 0.3775,
+      "step": 1469
+    },
+    {
+      "epoch": 0.6582703610411419,
+      "grad_norm": 1.2793594598770142,
+      "learning_rate": 3.4579870729455215e-05,
+      "loss": 0.4109,
+      "step": 1470
+    },
+    {
+      "epoch": 0.6582703610411419,
+      "eval_loss": 0.2607268989086151,
+      "eval_runtime": 1741.3184,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 2.565,
+      "step": 1470
+    },
+    {
+      "epoch": 0.6587181640078366,
+      "grad_norm": 1.1653704643249512,
+      "learning_rate": 3.4568328716528164e-05,
+      "loss": 0.1692,
+      "step": 1471
+    },
+    {
+      "epoch": 0.6591659669745312,
+      "grad_norm": 1.1862878799438477,
+      "learning_rate": 3.455678670360111e-05,
+      "loss": 0.2082,
+      "step": 1472
+    },
+    {
+      "epoch": 0.6596137699412259,
+      "grad_norm": 1.1138640642166138,
+      "learning_rate": 3.4545244690674056e-05,
+      "loss": 0.1064,
+      "step": 1473
+    },
+    {
+      "epoch": 0.6600615729079206,
+      "grad_norm": 1.390017032623291,
+      "learning_rate": 3.4533702677747e-05,
+      "loss": 0.3819,
+      "step": 1474
+    },
+    {
+      "epoch": 0.6605093758746152,
+      "grad_norm": 1.2229011058807373,
+      "learning_rate": 3.452216066481995e-05,
+      "loss": 0.2179,
+      "step": 1475
+    },
+    {
+      "epoch": 0.6609571788413098,
+      "grad_norm": 1.1169127225875854,
+      "learning_rate": 3.451061865189289e-05,
+      "loss": 0.212,
+      "step": 1476
+    },
+    {
+      "epoch": 0.6614049818080044,
+      "grad_norm": 1.3647346496582031,
+      "learning_rate": 3.449907663896584e-05,
+      "loss": 0.2052,
+      "step": 1477
+    },
+    {
+      "epoch": 0.6618527847746991,
+      "grad_norm": 1.423122525215149,
+      "learning_rate": 3.448753462603878e-05,
+      "loss": 0.1691,
+      "step": 1478
+    },
+    {
+      "epoch": 0.6623005877413938,
+      "grad_norm": 1.2200878858566284,
+      "learning_rate": 3.447599261311173e-05,
+      "loss": 0.3275,
+      "step": 1479
+    },
+    {
+      "epoch": 0.6627483907080884,
+      "grad_norm": 1.3014010190963745,
+      "learning_rate": 3.4464450600184675e-05,
+      "loss": 0.3449,
+      "step": 1480
+    },
+    {
+      "epoch": 0.6627483907080884,
+      "eval_loss": 0.25847139954566956,
+      "eval_runtime": 1741.5279,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 2.565,
+      "step": 1480
+    },
+    {
+      "epoch": 0.6631961936747831,
+      "grad_norm": 1.3815656900405884,
+      "learning_rate": 3.445290858725762e-05,
+      "loss": 0.1357,
+      "step": 1481
+    },
+    {
+      "epoch": 0.6636439966414778,
+      "grad_norm": 1.3851501941680908,
+      "learning_rate": 3.444136657433057e-05,
+      "loss": 0.388,
+      "step": 1482
+    },
+    {
+      "epoch": 0.6640917996081724,
+      "grad_norm": 0.9760904908180237,
+      "learning_rate": 3.442982456140351e-05,
+      "loss": 0.2618,
+      "step": 1483
+    },
+    {
+      "epoch": 0.6645396025748671,
+      "grad_norm": 1.1593406200408936,
+      "learning_rate": 3.441828254847645e-05,
+      "loss": 0.1924,
+      "step": 1484
+    },
+    {
+      "epoch": 0.6649874055415617,
+      "grad_norm": 1.4297518730163574,
+      "learning_rate": 3.44067405355494e-05,
+      "loss": 0.2887,
+      "step": 1485
+    },
+    {
+      "epoch": 0.6654352085082563,
+      "grad_norm": 1.0096676349639893,
+      "learning_rate": 3.4395198522622344e-05,
+      "loss": 0.1669,
+      "step": 1486
+    },
+    {
+      "epoch": 0.665883011474951,
+      "grad_norm": 1.343379020690918,
+      "learning_rate": 3.4383656509695293e-05,
+      "loss": 0.3973,
+      "step": 1487
+    },
+    {
+      "epoch": 0.6663308144416457,
+      "grad_norm": 1.2132956981658936,
+      "learning_rate": 3.4372114496768236e-05,
+      "loss": 0.2312,
+      "step": 1488
+    },
+    {
+      "epoch": 0.6667786174083403,
+      "grad_norm": 1.2106480598449707,
+      "learning_rate": 3.4360572483841185e-05,
+      "loss": 0.4319,
+      "step": 1489
+    },
+    {
+      "epoch": 0.667226420375035,
+      "grad_norm": 1.4173866510391235,
+      "learning_rate": 3.434903047091413e-05,
+      "loss": 0.3643,
+      "step": 1490
+    },
+    {
+      "epoch": 0.667226420375035,
+      "eval_loss": 0.25605136156082153,
+      "eval_runtime": 1742.0461,
+      "eval_samples_per_second": 2.564,
+      "eval_steps_per_second": 2.564,
+      "step": 1490
+    },
+    {
+      "epoch": 0.6676742233417297,
+      "grad_norm": 1.229823112487793,
+      "learning_rate": 3.433748845798708e-05,
+      "loss": 0.2643,
+      "step": 1491
+    },
+    {
+      "epoch": 0.6681220263084243,
+      "grad_norm": 1.2856589555740356,
+      "learning_rate": 3.432594644506002e-05,
+      "loss": 0.2516,
+      "step": 1492
+    },
+    {
+      "epoch": 0.668569829275119,
+      "grad_norm": 1.4023637771606445,
+      "learning_rate": 3.431440443213297e-05,
+      "loss": 0.2539,
+      "step": 1493
+    },
+    {
+      "epoch": 0.6690176322418137,
+      "grad_norm": 1.1964390277862549,
+      "learning_rate": 3.430286241920591e-05,
+      "loss": 0.4095,
+      "step": 1494
+    },
+    {
+      "epoch": 0.6694654352085082,
+      "grad_norm": 1.2144992351531982,
+      "learning_rate": 3.4291320406278855e-05,
+      "loss": 0.3216,
+      "step": 1495
+    },
+    {
+      "epoch": 0.6699132381752029,
+      "grad_norm": 1.6141177415847778,
+      "learning_rate": 3.4279778393351804e-05,
+      "loss": 0.3221,
+      "step": 1496
+    },
+    {
+      "epoch": 0.6703610411418975,
+      "grad_norm": 1.077372431755066,
+      "learning_rate": 3.4268236380424747e-05,
+      "loss": 0.3164,
+      "step": 1497
+    },
+    {
+      "epoch": 0.6708088441085922,
+      "grad_norm": 1.9878309965133667,
+      "learning_rate": 3.4256694367497696e-05,
+      "loss": 0.2299,
+      "step": 1498
+    },
+    {
+      "epoch": 0.6712566470752869,
+      "grad_norm": 1.0988008975982666,
+      "learning_rate": 3.424515235457064e-05,
+      "loss": 0.1888,
+      "step": 1499
+    },
+    {
+      "epoch": 0.6717044500419815,
+      "grad_norm": 1.2697983980178833,
+      "learning_rate": 3.423361034164359e-05,
+      "loss": 0.2446,
+      "step": 1500
+    },
+    {
+      "epoch": 0.6717044500419815,
+      "eval_loss": 0.25102800130844116,
+      "eval_runtime": 1741.064,
+      "eval_samples_per_second": 2.566,
+      "eval_steps_per_second": 2.566,
+      "step": 1500
+    },
+    {
+      "epoch": 0.6721522530086762,
+      "grad_norm": 1.1101131439208984,
+      "learning_rate": 3.422206832871653e-05,
+      "loss": 0.1175,
+      "step": 1501
+    },
+    {
+      "epoch": 0.6726000559753709,
+      "grad_norm": 1.8745020627975464,
+      "learning_rate": 3.421052631578947e-05,
+      "loss": 0.5006,
+      "step": 1502
+    },
+    {
+      "epoch": 0.6730478589420655,
+      "grad_norm": 1.5899895429611206,
+      "learning_rate": 3.4198984302862416e-05,
+      "loss": 0.4807,
+      "step": 1503
+    },
+    {
+      "epoch": 0.6734956619087601,
+      "grad_norm": 1.6755222082138062,
+      "learning_rate": 3.4187442289935365e-05,
+      "loss": 0.4125,
+      "step": 1504
+    },
+    {
+      "epoch": 0.6739434648754548,
+      "grad_norm": 1.5390435457229614,
+      "learning_rate": 3.417590027700831e-05,
+      "loss": 0.524,
+      "step": 1505
+    },
+    {
+      "epoch": 0.6743912678421494,
+      "grad_norm": 1.2704511880874634,
+      "learning_rate": 3.416435826408126e-05,
+      "loss": 0.2664,
+      "step": 1506
+    },
+    {
+      "epoch": 0.6748390708088441,
+      "grad_norm": 0.9589602947235107,
+      "learning_rate": 3.41528162511542e-05,
+      "loss": 0.2286,
+      "step": 1507
+    },
+    {
+      "epoch": 0.6752868737755388,
+      "grad_norm": 1.1151490211486816,
+      "learning_rate": 3.414127423822715e-05,
+      "loss": 0.3238,
+      "step": 1508
+    },
+    {
+      "epoch": 0.6757346767422334,
+      "grad_norm": 1.0681778192520142,
+      "learning_rate": 3.412973222530009e-05,
+      "loss": 0.2079,
+      "step": 1509
+    },
+    {
+      "epoch": 0.6761824797089281,
+      "grad_norm": 1.1055569648742676,
+      "learning_rate": 3.411819021237304e-05,
+      "loss": 0.2807,
+      "step": 1510
+    },
+    {
+      "epoch": 0.6761824797089281,
+      "eval_loss": 0.24902376532554626,
+      "eval_runtime": 1742.0489,
+      "eval_samples_per_second": 2.564,
+      "eval_steps_per_second": 2.564,
+      "step": 1510
+    },
+    {
+      "epoch": 0.6766302826756228,
+      "grad_norm": 1.061822533607483,
+      "learning_rate": 3.4106648199445984e-05,
+      "loss": 0.2028,
+      "step": 1511
+    },
+    {
+      "epoch": 0.6770780856423174,
+      "grad_norm": 1.3082396984100342,
+      "learning_rate": 3.409510618651893e-05,
+      "loss": 0.4129,
+      "step": 1512
+    },
+    {
+      "epoch": 0.6775258886090121,
+      "grad_norm": 1.0236505270004272,
+      "learning_rate": 3.4083564173591876e-05,
+      "loss": 0.1439,
+      "step": 1513
+    },
+    {
+      "epoch": 0.6779736915757066,
+      "grad_norm": 1.150169849395752,
+      "learning_rate": 3.4072022160664825e-05,
+      "loss": 0.1417,
+      "step": 1514
+    },
+    {
+      "epoch": 0.6784214945424013,
+      "grad_norm": 0.9641677141189575,
+      "learning_rate": 3.406048014773777e-05,
+      "loss": 0.1417,
+      "step": 1515
+    },
+    {
+      "epoch": 0.678869297509096,
+      "grad_norm": 1.1167751550674438,
+      "learning_rate": 3.404893813481072e-05,
+      "loss": 0.1216,
+      "step": 1516
+    },
+    {
+      "epoch": 0.6793171004757906,
+      "grad_norm": 1.2652047872543335,
+      "learning_rate": 3.403739612188366e-05,
+      "loss": 0.253,
+      "step": 1517
+    },
+    {
+      "epoch": 0.6797649034424853,
+      "grad_norm": 1.4064199924468994,
+      "learning_rate": 3.402585410895661e-05,
+      "loss": 0.3362,
+      "step": 1518
+    },
+    {
+      "epoch": 0.68021270640918,
+      "grad_norm": 1.244307041168213,
+      "learning_rate": 3.401431209602955e-05,
+      "loss": 0.2374,
+      "step": 1519
+    },
+    {
+      "epoch": 0.6806605093758746,
+      "grad_norm": 1.3351123332977295,
+      "learning_rate": 3.4002770083102494e-05,
+      "loss": 0.3342,
+      "step": 1520
+    },
+    {
+      "epoch": 0.6806605093758746,
+      "eval_loss": 0.24461551010608673,
+      "eval_runtime": 1740.9353,
+      "eval_samples_per_second": 2.566,
+      "eval_steps_per_second": 2.566,
+      "step": 1520
+    },
+    {
+      "epoch": 0.6811083123425693,
+      "grad_norm": 1.5558222532272339,
+      "learning_rate": 3.3991228070175444e-05,
+      "loss": 0.1321,
+      "step": 1521
+    },
+    {
+      "epoch": 0.681556115309264,
+      "grad_norm": 1.5526158809661865,
+      "learning_rate": 3.3979686057248386e-05,
+      "loss": 0.3917,
+      "step": 1522
+    },
+    {
+      "epoch": 0.6820039182759586,
+      "grad_norm": 1.3235844373703003,
+      "learning_rate": 3.396814404432133e-05,
+      "loss": 0.2082,
+      "step": 1523
+    },
+    {
+      "epoch": 0.6824517212426532,
+      "grad_norm": 1.226735234260559,
+      "learning_rate": 3.395660203139427e-05,
+      "loss": 0.1909,
+      "step": 1524
+    },
+    {
+      "epoch": 0.6828995242093479,
+      "grad_norm": 1.3214516639709473,
+      "learning_rate": 3.394506001846722e-05,
+      "loss": 0.2406,
+      "step": 1525
+    },
+    {
+      "epoch": 0.6833473271760425,
+      "grad_norm": 1.1992199420928955,
+      "learning_rate": 3.3933518005540164e-05,
+      "loss": 0.216,
+      "step": 1526
+    },
+    {
+      "epoch": 0.6837951301427372,
+      "grad_norm": 1.197937250137329,
+      "learning_rate": 3.392197599261311e-05,
+      "loss": 0.193,
+      "step": 1527
+    },
+    {
+      "epoch": 0.6842429331094319,
+      "grad_norm": 1.0902233123779297,
+      "learning_rate": 3.3910433979686056e-05,
+      "loss": 0.1702,
+      "step": 1528
+    },
+    {
+      "epoch": 0.6846907360761265,
+      "grad_norm": 1.1615121364593506,
+      "learning_rate": 3.3898891966759005e-05,
+      "loss": 0.2149,
+      "step": 1529
+    },
+    {
+      "epoch": 0.6851385390428212,
+      "grad_norm": 1.0904998779296875,
+      "learning_rate": 3.388734995383195e-05,
+      "loss": 0.1792,
+      "step": 1530
+    },
+    {
+      "epoch": 0.6851385390428212,
+      "eval_loss": 0.24200741946697235,
+      "eval_runtime": 1740.0794,
+      "eval_samples_per_second": 2.567,
+      "eval_steps_per_second": 2.567,
+      "step": 1530
+    },
+    {
+      "epoch": 0.6855863420095158,
+      "grad_norm": 1.5489822626113892,
+      "learning_rate": 3.38758079409049e-05,
+      "loss": 0.3297,
+      "step": 1531
+    },
+    {
+      "epoch": 0.6860341449762105,
+      "grad_norm": 1.0661543607711792,
+      "learning_rate": 3.386426592797784e-05,
+      "loss": 0.1182,
+      "step": 1532
+    },
+    {
+      "epoch": 0.6864819479429051,
+      "grad_norm": 1.2064769268035889,
+      "learning_rate": 3.385272391505079e-05,
+      "loss": 0.153,
+      "step": 1533
+    },
+    {
+      "epoch": 0.6869297509095997,
+      "grad_norm": 1.2851231098175049,
+      "learning_rate": 3.384118190212373e-05,
+      "loss": 0.4473,
+      "step": 1534
+    },
+    {
+      "epoch": 0.6873775538762944,
+      "grad_norm": 1.2064697742462158,
+      "learning_rate": 3.382963988919668e-05,
+      "loss": 0.2223,
+      "step": 1535
+    },
+    {
+      "epoch": 0.6878253568429891,
+      "grad_norm": 1.3766595125198364,
+      "learning_rate": 3.3818097876269623e-05,
+      "loss": 0.3017,
+      "step": 1536
+    },
+    {
+      "epoch": 0.6882731598096837,
+      "grad_norm": 1.2292721271514893,
+      "learning_rate": 3.380655586334257e-05,
+      "loss": 0.146,
+      "step": 1537
+    },
+    {
+      "epoch": 0.6887209627763784,
+      "grad_norm": 1.3682496547698975,
+      "learning_rate": 3.3795013850415515e-05,
+      "loss": 0.2525,
+      "step": 1538
+    },
+    {
+      "epoch": 0.6891687657430731,
+      "grad_norm": 1.5135877132415771,
+      "learning_rate": 3.3783471837488465e-05,
+      "loss": 0.2383,
+      "step": 1539
+    },
+    {
+      "epoch": 0.6896165687097677,
+      "grad_norm": 1.3297990560531616,
+      "learning_rate": 3.377192982456141e-05,
+      "loss": 0.2684,
+      "step": 1540
+    },
+    {
+      "epoch": 0.6896165687097677,
+      "eval_loss": 0.23915019631385803,
+      "eval_runtime": 1739.054,
+      "eval_samples_per_second": 2.569,
+      "eval_steps_per_second": 2.569,
+      "step": 1540
+    },
+    {
+      "epoch": 0.6900643716764624,
+      "grad_norm": 1.10771644115448,
+      "learning_rate": 3.376038781163435e-05,
+      "loss": 0.1979,
+      "step": 1541
+    },
+    {
+      "epoch": 0.6905121746431571,
+      "grad_norm": 0.916994571685791,
+      "learning_rate": 3.374884579870729e-05,
+      "loss": 0.0774,
+      "step": 1542
+    },
+    {
+      "epoch": 0.6909599776098516,
+      "grad_norm": 1.2376232147216797,
+      "learning_rate": 3.373730378578024e-05,
+      "loss": 0.1507,
+      "step": 1543
+    },
+    {
+      "epoch": 0.6914077805765463,
+      "grad_norm": 1.1735001802444458,
+      "learning_rate": 3.3725761772853185e-05,
+      "loss": 0.3846,
+      "step": 1544
+    },
+    {
+      "epoch": 0.691855583543241,
+      "grad_norm": 1.585693359375,
+      "learning_rate": 3.371421975992613e-05,
+      "loss": 0.4625,
+      "step": 1545
+    },
+    {
+      "epoch": 0.6923033865099356,
+      "grad_norm": 1.234339952468872,
+      "learning_rate": 3.370267774699908e-05,
+      "loss": 0.1853,
+      "step": 1546
+    },
+    {
+      "epoch": 0.6927511894766303,
+      "grad_norm": 0.9094129204750061,
+      "learning_rate": 3.369113573407202e-05,
+      "loss": 0.1299,
+      "step": 1547
+    },
+    {
+      "epoch": 0.693198992443325,
+      "grad_norm": 1.1160544157028198,
+      "learning_rate": 3.367959372114497e-05,
+      "loss": 0.1527,
+      "step": 1548
+    },
+    {
+      "epoch": 0.6936467954100196,
+      "grad_norm": 1.3147051334381104,
+      "learning_rate": 3.366805170821791e-05,
+      "loss": 0.2995,
+      "step": 1549
+    },
+    {
+      "epoch": 0.6940945983767143,
+      "grad_norm": 1.2512911558151245,
+      "learning_rate": 3.365650969529086e-05,
+      "loss": 0.2386,
+      "step": 1550
+    },
+    {
+      "epoch": 0.6940945983767143,
+      "eval_loss": 0.23694820702075958,
+      "eval_runtime": 1738.7269,
+      "eval_samples_per_second": 2.569,
+      "eval_steps_per_second": 2.569,
+      "step": 1550
+    },
+    {
+      "epoch": 0.6945424013434089,
+      "grad_norm": 1.0559110641479492,
+      "learning_rate": 3.36449676823638e-05,
+      "loss": 0.2887,
+      "step": 1551
+    },
+    {
+      "epoch": 0.6949902043101036,
+      "grad_norm": 1.3872283697128296,
+      "learning_rate": 3.363342566943675e-05,
+      "loss": 0.1972,
+      "step": 1552
+    },
+    {
+      "epoch": 0.6954380072767982,
+      "grad_norm": 1.2875914573669434,
+      "learning_rate": 3.3621883656509695e-05,
+      "loss": 0.3023,
+      "step": 1553
+    },
+    {
+      "epoch": 0.6958858102434928,
+      "grad_norm": 1.1694672107696533,
+      "learning_rate": 3.3610341643582645e-05,
+      "loss": 0.2716,
+      "step": 1554
+    },
+    {
+      "epoch": 0.6963336132101875,
+      "grad_norm": 1.1989940404891968,
+      "learning_rate": 3.359879963065559e-05,
+      "loss": 0.2673,
+      "step": 1555
+    },
+    {
+      "epoch": 0.6967814161768822,
+      "grad_norm": 1.1650522947311401,
+      "learning_rate": 3.3587257617728537e-05,
+      "loss": 0.2633,
+      "step": 1556
+    },
+    {
+      "epoch": 0.6972292191435768,
+      "grad_norm": 1.125189185142517,
+      "learning_rate": 3.357571560480148e-05,
+      "loss": 0.1559,
+      "step": 1557
+    },
+    {
+      "epoch": 0.6976770221102715,
+      "grad_norm": 0.9679086804389954,
+      "learning_rate": 3.356417359187443e-05,
+      "loss": 0.1219,
+      "step": 1558
+    },
+    {
+      "epoch": 0.6981248250769662,
+      "grad_norm": 1.3153406381607056,
+      "learning_rate": 3.355263157894737e-05,
+      "loss": 0.2164,
+      "step": 1559
+    },
+    {
+      "epoch": 0.6985726280436608,
+      "grad_norm": 1.0555087327957153,
+      "learning_rate": 3.354108956602032e-05,
+      "loss": 0.1167,
+      "step": 1560
+    },
+    {
+      "epoch": 0.6985726280436608,
+      "eval_loss": 0.23435254395008087,
+      "eval_runtime": 1735.1927,
+      "eval_samples_per_second": 2.574,
+      "eval_steps_per_second": 2.574,
+      "step": 1560
+    },
+    {
+      "epoch": 0.6990204310103555,
+      "grad_norm": 1.350873589515686,
+      "learning_rate": 3.352954755309326e-05,
+      "loss": 0.2516,
+      "step": 1561
+    },
+    {
+      "epoch": 0.69946823397705,
+      "grad_norm": 1.7134159803390503,
+      "learning_rate": 3.3518005540166206e-05,
+      "loss": 0.2061,
+      "step": 1562
+    },
+    {
+      "epoch": 0.6999160369437447,
+      "grad_norm": 1.3558813333511353,
+      "learning_rate": 3.350646352723915e-05,
+      "loss": 0.2233,
+      "step": 1563
+    },
+    {
+      "epoch": 0.7003638399104394,
+      "grad_norm": 1.2703967094421387,
+      "learning_rate": 3.34949215143121e-05,
+      "loss": 0.2354,
+      "step": 1564
+    },
+    {
+      "epoch": 0.700811642877134,
+      "grad_norm": 1.2839316129684448,
+      "learning_rate": 3.348337950138504e-05,
+      "loss": 0.2084,
+      "step": 1565
+    },
+    {
+      "epoch": 0.7012594458438287,
+      "grad_norm": 1.9036198854446411,
+      "learning_rate": 3.347183748845799e-05,
+      "loss": 0.2555,
+      "step": 1566
+    },
+    {
+      "epoch": 0.7017072488105234,
+      "grad_norm": 1.2471376657485962,
+      "learning_rate": 3.346029547553093e-05,
+      "loss": 0.2167,
+      "step": 1567
+    },
+    {
+      "epoch": 0.702155051777218,
+      "grad_norm": 1.5329457521438599,
+      "learning_rate": 3.344875346260388e-05,
+      "loss": 0.3522,
+      "step": 1568
+    },
+    {
+      "epoch": 0.7026028547439127,
+      "grad_norm": 1.1027286052703857,
+      "learning_rate": 3.3437211449676824e-05,
+      "loss": 0.3117,
+      "step": 1569
+    },
+    {
+      "epoch": 0.7030506577106074,
+      "grad_norm": 1.2840759754180908,
+      "learning_rate": 3.342566943674977e-05,
+      "loss": 0.2366,
+      "step": 1570
+    },
+    {
+      "epoch": 0.7030506577106074,
+      "eval_loss": 0.23120713233947754,
+      "eval_runtime": 1731.9849,
+      "eval_samples_per_second": 2.579,
+      "eval_steps_per_second": 2.579,
+      "step": 1570
+    },
+    {
+      "epoch": 0.703498460677302,
+      "grad_norm": 1.1187115907669067,
+      "learning_rate": 3.3414127423822716e-05,
+      "loss": 0.3006,
+      "step": 1571
+    },
+    {
+      "epoch": 0.7039462636439966,
+      "grad_norm": 1.0415239334106445,
+      "learning_rate": 3.340258541089566e-05,
+      "loss": 0.224,
+      "step": 1572
+    },
+    {
+      "epoch": 0.7043940666106913,
+      "grad_norm": 1.0761853456497192,
+      "learning_rate": 3.339104339796861e-05,
+      "loss": 0.3569,
+      "step": 1573
+    },
+    {
+      "epoch": 0.7048418695773859,
+      "grad_norm": 1.53033447265625,
+      "learning_rate": 3.337950138504155e-05,
+      "loss": 0.3501,
+      "step": 1574
+    },
+    {
+      "epoch": 0.7052896725440806,
+      "grad_norm": 1.0321897268295288,
+      "learning_rate": 3.33679593721145e-05,
+      "loss": 0.194,
+      "step": 1575
+    },
+    {
+      "epoch": 0.7057374755107753,
+      "grad_norm": 1.5243761539459229,
+      "learning_rate": 3.335641735918744e-05,
+      "loss": 0.2557,
+      "step": 1576
+    },
+    {
+      "epoch": 0.7061852784774699,
+      "grad_norm": 1.264231562614441,
+      "learning_rate": 3.334487534626039e-05,
+      "loss": 0.179,
+      "step": 1577
+    },
+    {
+      "epoch": 0.7066330814441646,
+      "grad_norm": 1.316745400428772,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.434,
+      "step": 1578
+    },
+    {
+      "epoch": 0.7070808844108593,
+      "grad_norm": 1.1383960247039795,
+      "learning_rate": 3.3321791320406284e-05,
+      "loss": 0.1775,
+      "step": 1579
+    },
+    {
+      "epoch": 0.7075286873775539,
+      "grad_norm": 1.0016801357269287,
+      "learning_rate": 3.331024930747923e-05,
+      "loss": 0.1749,
+      "step": 1580
+    },
+    {
+      "epoch": 0.7075286873775539,
+      "eval_loss": 0.22776223719120026,
+      "eval_runtime": 1740.0928,
+      "eval_samples_per_second": 2.567,
+      "eval_steps_per_second": 2.567,
+      "step": 1580
+    },
+    {
+      "epoch": 0.7079764903442485,
+      "grad_norm": 1.7598121166229248,
+      "learning_rate": 3.329870729455217e-05,
+      "loss": 0.2445,
+      "step": 1581
+    },
+    {
+      "epoch": 0.7084242933109431,
+      "grad_norm": 1.9481260776519775,
+      "learning_rate": 3.328716528162512e-05,
+      "loss": 0.2726,
+      "step": 1582
+    },
+    {
+      "epoch": 0.7088720962776378,
+      "grad_norm": 1.1749162673950195,
+      "learning_rate": 3.327562326869806e-05,
+      "loss": 0.103,
+      "step": 1583
+    },
+    {
+      "epoch": 0.7093198992443325,
+      "grad_norm": 1.4559996128082275,
+      "learning_rate": 3.3264081255771004e-05,
+      "loss": 0.2297,
+      "step": 1584
+    },
+    {
+      "epoch": 0.7097677022110271,
+      "grad_norm": 1.4041540622711182,
+      "learning_rate": 3.3252539242843954e-05,
+      "loss": 0.2676,
+      "step": 1585
+    },
+    {
+      "epoch": 0.7102155051777218,
+      "grad_norm": 1.2508280277252197,
+      "learning_rate": 3.3240997229916896e-05,
+      "loss": 0.254,
+      "step": 1586
+    },
+    {
+      "epoch": 0.7106633081444165,
+      "grad_norm": 0.9396544694900513,
+      "learning_rate": 3.3229455216989845e-05,
+      "loss": 0.1049,
+      "step": 1587
+    },
+    {
+      "epoch": 0.7111111111111111,
+      "grad_norm": 1.3375426530838013,
+      "learning_rate": 3.321791320406279e-05,
+      "loss": 0.2525,
+      "step": 1588
+    },
+    {
+      "epoch": 0.7115589140778058,
+      "grad_norm": 1.631959080696106,
+      "learning_rate": 3.320637119113574e-05,
+      "loss": 0.2584,
+      "step": 1589
+    },
+    {
+      "epoch": 0.7120067170445005,
+      "grad_norm": 1.31275475025177,
+      "learning_rate": 3.319482917820868e-05,
+      "loss": 0.1589,
+      "step": 1590
+    },
+    {
+      "epoch": 0.7120067170445005,
+      "eval_loss": 0.2232540100812912,
+      "eval_runtime": 1739.5515,
+      "eval_samples_per_second": 2.568,
+      "eval_steps_per_second": 2.568,
+      "step": 1590
+    },
+    {
+      "epoch": 0.712454520011195,
+      "grad_norm": 1.3619765043258667,
+      "learning_rate": 3.318328716528163e-05,
+      "loss": 0.1927,
+      "step": 1591
+    },
+    {
+      "epoch": 0.7129023229778897,
+      "grad_norm": 1.2817407846450806,
+      "learning_rate": 3.317174515235457e-05,
+      "loss": 0.2036,
+      "step": 1592
+    },
+    {
+      "epoch": 0.7133501259445844,
+      "grad_norm": 1.3498672246932983,
+      "learning_rate": 3.316020313942752e-05,
+      "loss": 0.1583,
+      "step": 1593
+    },
+    {
+      "epoch": 0.713797928911279,
+      "grad_norm": 1.0232353210449219,
+      "learning_rate": 3.3148661126500464e-05,
+      "loss": 0.1753,
+      "step": 1594
+    },
+    {
+      "epoch": 0.7142457318779737,
+      "grad_norm": 1.239030361175537,
+      "learning_rate": 3.313711911357341e-05,
+      "loss": 0.2525,
+      "step": 1595
+    },
+    {
+      "epoch": 0.7146935348446684,
+      "grad_norm": 0.9768178462982178,
+      "learning_rate": 3.3125577100646356e-05,
+      "loss": 0.1509,
+      "step": 1596
+    },
+    {
+      "epoch": 0.715141337811363,
+      "grad_norm": 1.2771555185317993,
+      "learning_rate": 3.31140350877193e-05,
+      "loss": 0.22,
+      "step": 1597
+    },
+    {
+      "epoch": 0.7155891407780577,
+      "grad_norm": 0.8508195877075195,
+      "learning_rate": 3.310249307479225e-05,
+      "loss": 0.1854,
+      "step": 1598
+    },
+    {
+      "epoch": 0.7160369437447524,
+      "grad_norm": 1.0700000524520874,
+      "learning_rate": 3.309095106186519e-05,
+      "loss": 0.2283,
+      "step": 1599
+    },
+    {
+      "epoch": 0.716484746711447,
+      "grad_norm": 1.2275710105895996,
+      "learning_rate": 3.307940904893814e-05,
+      "loss": 0.1604,
+      "step": 1600
+    },
+    {
+      "epoch": 0.716484746711447,
+      "eval_loss": 0.22277671098709106,
+      "eval_runtime": 1739.7877,
+      "eval_samples_per_second": 2.568,
+      "eval_steps_per_second": 2.568,
+      "step": 1600
+    },
+    {
+      "epoch": 0.7169325496781416,
+      "grad_norm": 0.8512494564056396,
+      "learning_rate": 3.306786703601108e-05,
+      "loss": 0.0913,
+      "step": 1601
+    },
+    {
+      "epoch": 0.7173803526448362,
+      "grad_norm": 1.4188917875289917,
+      "learning_rate": 3.3056325023084025e-05,
+      "loss": 0.3004,
+      "step": 1602
+    },
+    {
+      "epoch": 0.7178281556115309,
+      "grad_norm": 1.1424413919448853,
+      "learning_rate": 3.304478301015697e-05,
+      "loss": 0.2261,
+      "step": 1603
+    },
+    {
+      "epoch": 0.7182759585782256,
+      "grad_norm": 1.630253553390503,
+      "learning_rate": 3.303324099722992e-05,
+      "loss": 0.3728,
+      "step": 1604
+    },
+    {
+      "epoch": 0.7187237615449202,
+      "grad_norm": 1.3553986549377441,
+      "learning_rate": 3.302169898430286e-05,
+      "loss": 0.2097,
+      "step": 1605
+    },
+    {
+      "epoch": 0.7191715645116149,
+      "grad_norm": 1.4127997159957886,
+      "learning_rate": 3.301015697137581e-05,
+      "loss": 0.2059,
+      "step": 1606
+    },
+    {
+      "epoch": 0.7196193674783096,
+      "grad_norm": 1.2591644525527954,
+      "learning_rate": 3.299861495844875e-05,
+      "loss": 0.2187,
+      "step": 1607
+    },
+    {
+      "epoch": 0.7200671704450042,
+      "grad_norm": 1.056017279624939,
+      "learning_rate": 3.29870729455217e-05,
+      "loss": 0.1817,
+      "step": 1608
+    },
+    {
+      "epoch": 0.7205149734116989,
+      "grad_norm": 1.230894923210144,
+      "learning_rate": 3.2975530932594644e-05,
+      "loss": 0.3076,
+      "step": 1609
+    },
+    {
+      "epoch": 0.7209627763783935,
+      "grad_norm": 1.300376534461975,
+      "learning_rate": 3.296398891966759e-05,
+      "loss": 0.2628,
+      "step": 1610
+    },
+    {
+      "epoch": 0.7209627763783935,
+      "eval_loss": 0.21946477890014648,
+      "eval_runtime": 1739.4143,
+      "eval_samples_per_second": 2.568,
+      "eval_steps_per_second": 2.568,
+      "step": 1610
+    },
+    {
+      "epoch": 0.7214105793450881,
+      "grad_norm": 1.413095235824585,
+      "learning_rate": 3.2952446906740536e-05,
+      "loss": 0.2967,
+      "step": 1611
+    },
+    {
+      "epoch": 0.7218583823117828,
+      "grad_norm": 1.1679149866104126,
+      "learning_rate": 3.2940904893813485e-05,
+      "loss": 0.3076,
+      "step": 1612
+    },
+    {
+      "epoch": 0.7223061852784775,
+      "grad_norm": 1.018036961555481,
+      "learning_rate": 3.292936288088643e-05,
+      "loss": 0.1127,
+      "step": 1613
+    },
+    {
+      "epoch": 0.7227539882451721,
+      "grad_norm": 0.766292929649353,
+      "learning_rate": 3.291782086795938e-05,
+      "loss": 0.0642,
+      "step": 1614
+    },
+    {
+      "epoch": 0.7232017912118668,
+      "grad_norm": 1.2774684429168701,
+      "learning_rate": 3.290627885503232e-05,
+      "loss": 0.1783,
+      "step": 1615
+    },
+    {
+      "epoch": 0.7236495941785615,
+      "grad_norm": 1.4727896451950073,
+      "learning_rate": 3.289473684210527e-05,
+      "loss": 0.2811,
+      "step": 1616
+    },
+    {
+      "epoch": 0.7240973971452561,
+      "grad_norm": 1.0615532398223877,
+      "learning_rate": 3.288319482917821e-05,
+      "loss": 0.2028,
+      "step": 1617
+    },
+    {
+      "epoch": 0.7245452001119508,
+      "grad_norm": 1.279786229133606,
+      "learning_rate": 3.287165281625116e-05,
+      "loss": 0.2692,
+      "step": 1618
+    },
+    {
+      "epoch": 0.7249930030786454,
+      "grad_norm": 1.2355388402938843,
+      "learning_rate": 3.2860110803324104e-05,
+      "loss": 0.2413,
+      "step": 1619
+    },
+    {
+      "epoch": 0.72544080604534,
+      "grad_norm": 1.3490946292877197,
+      "learning_rate": 3.2848568790397046e-05,
+      "loss": 0.2309,
+      "step": 1620
+    },
+    {
+      "epoch": 0.72544080604534,
+      "eval_loss": 0.21591566503047943,
+      "eval_runtime": 1739.1002,
+      "eval_samples_per_second": 2.569,
+      "eval_steps_per_second": 2.569,
+      "step": 1620
+    },
+    {
+      "epoch": 0.7258886090120347,
+      "grad_norm": 1.1078510284423828,
+      "learning_rate": 3.2837026777469996e-05,
+      "loss": 0.1262,
+      "step": 1621
+    },
+    {
+      "epoch": 0.7263364119787293,
+      "grad_norm": 1.130131483078003,
+      "learning_rate": 3.282548476454294e-05,
+      "loss": 0.146,
+      "step": 1622
+    },
+    {
+      "epoch": 0.726784214945424,
+      "grad_norm": 1.545876383781433,
+      "learning_rate": 3.281394275161588e-05,
+      "loss": 0.3484,
+      "step": 1623
+    },
+    {
+      "epoch": 0.7272320179121187,
+      "grad_norm": 1.292954444885254,
+      "learning_rate": 3.2802400738688824e-05,
+      "loss": 0.2202,
+      "step": 1624
+    },
+    {
+      "epoch": 0.7276798208788133,
+      "grad_norm": 1.3415828943252563,
+      "learning_rate": 3.279085872576177e-05,
+      "loss": 0.2863,
+      "step": 1625
+    },
+    {
+      "epoch": 0.728127623845508,
+      "grad_norm": 1.084219217300415,
+      "learning_rate": 3.2779316712834716e-05,
+      "loss": 0.2006,
+      "step": 1626
+    },
+    {
+      "epoch": 0.7285754268122027,
+      "grad_norm": 1.0170210599899292,
+      "learning_rate": 3.2767774699907665e-05,
+      "loss": 0.1657,
+      "step": 1627
+    },
+    {
+      "epoch": 0.7290232297788973,
+      "grad_norm": 1.1863627433776855,
+      "learning_rate": 3.275623268698061e-05,
+      "loss": 0.1701,
+      "step": 1628
+    },
+    {
+      "epoch": 0.7294710327455919,
+      "grad_norm": 1.0183345079421997,
+      "learning_rate": 3.274469067405356e-05,
+      "loss": 0.13,
+      "step": 1629
+    },
+    {
+      "epoch": 0.7299188357122866,
+      "grad_norm": 1.2636232376098633,
+      "learning_rate": 3.27331486611265e-05,
+      "loss": 0.3096,
+      "step": 1630
+    },
+    {
+      "epoch": 0.7299188357122866,
+      "eval_loss": 0.21282623708248138,
+      "eval_runtime": 1738.9211,
+      "eval_samples_per_second": 2.569,
+      "eval_steps_per_second": 2.569,
+      "step": 1630
+    },
+    {
+      "epoch": 0.7303666386789812,
+      "grad_norm": 1.625278115272522,
+      "learning_rate": 3.272160664819945e-05,
+      "loss": 0.1581,
+      "step": 1631
+    },
+    {
+      "epoch": 0.7308144416456759,
+      "grad_norm": 1.0692945718765259,
+      "learning_rate": 3.271006463527239e-05,
+      "loss": 0.1466,
+      "step": 1632
+    },
+    {
+      "epoch": 0.7312622446123705,
+      "grad_norm": 1.545180320739746,
+      "learning_rate": 3.269852262234534e-05,
+      "loss": 0.2468,
+      "step": 1633
+    },
+    {
+      "epoch": 0.7317100475790652,
+      "grad_norm": 1.2902580499649048,
+      "learning_rate": 3.2686980609418284e-05,
+      "loss": 0.1939,
+      "step": 1634
+    },
+    {
+      "epoch": 0.7321578505457599,
+      "grad_norm": 1.4147045612335205,
+      "learning_rate": 3.267543859649123e-05,
+      "loss": 0.3051,
+      "step": 1635
+    },
+    {
+      "epoch": 0.7326056535124545,
+      "grad_norm": 1.821250557899475,
+      "learning_rate": 3.2663896583564176e-05,
+      "loss": 0.5277,
+      "step": 1636
+    },
+    {
+      "epoch": 0.7330534564791492,
+      "grad_norm": 1.1706265211105347,
+      "learning_rate": 3.2652354570637125e-05,
+      "loss": 0.2537,
+      "step": 1637
+    },
+    {
+      "epoch": 0.7335012594458439,
+      "grad_norm": 1.3979747295379639,
+      "learning_rate": 3.264081255771007e-05,
+      "loss": 0.1647,
+      "step": 1638
+    },
+    {
+      "epoch": 0.7339490624125384,
+      "grad_norm": 1.1551800966262817,
+      "learning_rate": 3.262927054478302e-05,
+      "loss": 0.2547,
+      "step": 1639
+    },
+    {
+      "epoch": 0.7343968653792331,
+      "grad_norm": 0.986116349697113,
+      "learning_rate": 3.261772853185596e-05,
+      "loss": 0.1303,
+      "step": 1640
+    },
+    {
+      "epoch": 0.7343968653792331,
+      "eval_loss": 0.21062257885932922,
+      "eval_runtime": 1739.0011,
+      "eval_samples_per_second": 2.569,
+      "eval_steps_per_second": 2.569,
+      "step": 1640
+    },
+    {
+      "epoch": 0.7348446683459278,
+      "grad_norm": 0.9228246808052063,
+      "learning_rate": 3.26061865189289e-05,
+      "loss": 0.1241,
+      "step": 1641
+    },
+    {
+      "epoch": 0.7352924713126224,
+      "grad_norm": 1.1444522142410278,
+      "learning_rate": 3.2594644506001845e-05,
+      "loss": 0.2111,
+      "step": 1642
+    },
+    {
+      "epoch": 0.7357402742793171,
+      "grad_norm": 1.3221601247787476,
+      "learning_rate": 3.2583102493074794e-05,
+      "loss": 0.5413,
+      "step": 1643
+    },
+    {
+      "epoch": 0.7361880772460118,
+      "grad_norm": 1.3881961107254028,
+      "learning_rate": 3.257156048014774e-05,
+      "loss": 0.2686,
+      "step": 1644
+    },
+    {
+      "epoch": 0.7366358802127064,
+      "grad_norm": 1.500131368637085,
+      "learning_rate": 3.2560018467220686e-05,
+      "loss": 0.3149,
+      "step": 1645
+    },
+    {
+      "epoch": 0.7370836831794011,
+      "grad_norm": 1.0553194284439087,
+      "learning_rate": 3.254847645429363e-05,
+      "loss": 0.1628,
+      "step": 1646
+    },
+    {
+      "epoch": 0.7375314861460958,
+      "grad_norm": 0.9254956841468811,
+      "learning_rate": 3.253693444136657e-05,
+      "loss": 0.1078,
+      "step": 1647
+    },
+    {
+      "epoch": 0.7379792891127904,
+      "grad_norm": 1.1450732946395874,
+      "learning_rate": 3.252539242843952e-05,
+      "loss": 0.2339,
+      "step": 1648
+    },
+    {
+      "epoch": 0.738427092079485,
+      "grad_norm": 0.9290254712104797,
+      "learning_rate": 3.251385041551246e-05,
+      "loss": 0.1774,
+      "step": 1649
+    },
+    {
+      "epoch": 0.7388748950461796,
+      "grad_norm": 0.8868831396102905,
+      "learning_rate": 3.250230840258541e-05,
+      "loss": 0.1535,
+      "step": 1650
+    },
+    {
+      "epoch": 0.7388748950461796,
+      "eval_loss": 0.2098311334848404,
+      "eval_runtime": 1738.0385,
+      "eval_samples_per_second": 2.57,
+      "eval_steps_per_second": 2.57,
+      "step": 1650
+    },
+    {
+      "epoch": 0.7393226980128743,
+      "grad_norm": 1.1324498653411865,
+      "learning_rate": 3.2490766389658355e-05,
+      "loss": 0.1097,
+      "step": 1651
+    },
+    {
+      "epoch": 0.739770500979569,
+      "grad_norm": 1.3903108835220337,
+      "learning_rate": 3.2479224376731305e-05,
+      "loss": 0.24,
+      "step": 1652
+    },
+    {
+      "epoch": 0.7402183039462636,
+      "grad_norm": 1.320954442024231,
+      "learning_rate": 3.246768236380425e-05,
+      "loss": 0.2528,
+      "step": 1653
+    },
+    {
+      "epoch": 0.7406661069129583,
+      "grad_norm": 1.5540986061096191,
+      "learning_rate": 3.24561403508772e-05,
+      "loss": 0.2835,
+      "step": 1654
+    },
+    {
+      "epoch": 0.741113909879653,
+      "grad_norm": 1.2737802267074585,
+      "learning_rate": 3.244459833795014e-05,
+      "loss": 0.2013,
+      "step": 1655
+    },
+    {
+      "epoch": 0.7415617128463476,
+      "grad_norm": 1.185079574584961,
+      "learning_rate": 3.243305632502309e-05,
+      "loss": 0.1936,
+      "step": 1656
+    },
+    {
+      "epoch": 0.7420095158130423,
+      "grad_norm": 1.8175549507141113,
+      "learning_rate": 3.242151431209603e-05,
+      "loss": 0.3241,
+      "step": 1657
+    },
+    {
+      "epoch": 0.7424573187797369,
+      "grad_norm": 0.9585305452346802,
+      "learning_rate": 3.240997229916898e-05,
+      "loss": 0.1021,
+      "step": 1658
+    },
+    {
+      "epoch": 0.7429051217464315,
+      "grad_norm": 1.0971755981445312,
+      "learning_rate": 3.239843028624192e-05,
+      "loss": 0.1455,
+      "step": 1659
+    },
+    {
+      "epoch": 0.7433529247131262,
+      "grad_norm": 1.2596510648727417,
+      "learning_rate": 3.238688827331487e-05,
+      "loss": 0.1561,
+      "step": 1660
+    },
+    {
+      "epoch": 0.7433529247131262,
+      "eval_loss": 0.2056712806224823,
+      "eval_runtime": 1736.8376,
+      "eval_samples_per_second": 2.572,
+      "eval_steps_per_second": 2.572,
+      "step": 1660
+    },
+    {
+      "epoch": 0.7438007276798209,
+      "grad_norm": 1.399214506149292,
+      "learning_rate": 3.2375346260387815e-05,
+      "loss": 0.1979,
+      "step": 1661
+    },
+    {
+      "epoch": 0.7442485306465155,
+      "grad_norm": 1.5399458408355713,
+      "learning_rate": 3.236380424746076e-05,
+      "loss": 0.2839,
+      "step": 1662
+    },
+    {
+      "epoch": 0.7446963336132102,
+      "grad_norm": 1.3398548364639282,
+      "learning_rate": 3.23522622345337e-05,
+      "loss": 0.2559,
+      "step": 1663
+    },
+    {
+      "epoch": 0.7451441365799049,
+      "grad_norm": 1.3190560340881348,
+      "learning_rate": 3.234072022160665e-05,
+      "loss": 0.123,
+      "step": 1664
+    },
+    {
+      "epoch": 0.7455919395465995,
+      "grad_norm": 0.886492908000946,
+      "learning_rate": 3.232917820867959e-05,
+      "loss": 0.1186,
+      "step": 1665
+    },
+    {
+      "epoch": 0.7460397425132942,
+      "grad_norm": 1.5858815908432007,
+      "learning_rate": 3.231763619575254e-05,
+      "loss": 0.2775,
+      "step": 1666
+    },
+    {
+      "epoch": 0.7464875454799889,
+      "grad_norm": 1.419754147529602,
+      "learning_rate": 3.2306094182825484e-05,
+      "loss": 0.2343,
+      "step": 1667
+    },
+    {
+      "epoch": 0.7469353484466834,
+      "grad_norm": 1.1208510398864746,
+      "learning_rate": 3.2294552169898434e-05,
+      "loss": 0.1272,
+      "step": 1668
+    },
+    {
+      "epoch": 0.7473831514133781,
+      "grad_norm": 0.9657753705978394,
+      "learning_rate": 3.2283010156971376e-05,
+      "loss": 0.1537,
+      "step": 1669
+    },
+    {
+      "epoch": 0.7478309543800727,
+      "grad_norm": 1.117485761642456,
+      "learning_rate": 3.2271468144044326e-05,
+      "loss": 0.2412,
+      "step": 1670
+    },
+    {
+      "epoch": 0.7478309543800727,
+      "eval_loss": 0.2037663459777832,
+      "eval_runtime": 1736.9404,
+      "eval_samples_per_second": 2.572,
+      "eval_steps_per_second": 2.572,
+      "step": 1670
+    },
+    {
+      "epoch": 0.7482787573467674,
+      "grad_norm": 1.138968825340271,
+      "learning_rate": 3.225992613111727e-05,
+      "loss": 0.221,
+      "step": 1671
+    },
+    {
+      "epoch": 0.7487265603134621,
+      "grad_norm": 1.2097197771072388,
+      "learning_rate": 3.224838411819021e-05,
+      "loss": 0.1445,
+      "step": 1672
+    },
+    {
+      "epoch": 0.7491743632801567,
+      "grad_norm": 1.4289298057556152,
+      "learning_rate": 3.223684210526316e-05,
+      "loss": 0.2555,
+      "step": 1673
+    },
+    {
+      "epoch": 0.7496221662468514,
+      "grad_norm": 1.1956260204315186,
+      "learning_rate": 3.22253000923361e-05,
+      "loss": 0.2039,
+      "step": 1674
+    },
+    {
+      "epoch": 0.7500699692135461,
+      "grad_norm": 1.2446436882019043,
+      "learning_rate": 3.221375807940905e-05,
+      "loss": 0.2794,
+      "step": 1675
+    },
+    {
+      "epoch": 0.7505177721802407,
+      "grad_norm": 1.1709624528884888,
+      "learning_rate": 3.2202216066481995e-05,
+      "loss": 0.1036,
+      "step": 1676
+    },
+    {
+      "epoch": 0.7509655751469353,
+      "grad_norm": 1.0143916606903076,
+      "learning_rate": 3.2190674053554944e-05,
+      "loss": 0.1106,
+      "step": 1677
+    },
+    {
+      "epoch": 0.75141337811363,
+      "grad_norm": 1.3979700803756714,
+      "learning_rate": 3.217913204062789e-05,
+      "loss": 0.2883,
+      "step": 1678
+    },
+    {
+      "epoch": 0.7518611810803246,
+      "grad_norm": 1.49893319606781,
+      "learning_rate": 3.2167590027700836e-05,
+      "loss": 0.3638,
+      "step": 1679
+    },
+    {
+      "epoch": 0.7523089840470193,
+      "grad_norm": 1.2170891761779785,
+      "learning_rate": 3.215604801477378e-05,
+      "loss": 0.2359,
+      "step": 1680
+    },
+    {
+      "epoch": 0.7523089840470193,
+      "eval_loss": 0.20157921314239502,
+      "eval_runtime": 1735.3712,
+      "eval_samples_per_second": 2.574,
+      "eval_steps_per_second": 2.574,
+      "step": 1680
+    },
+    {
+      "epoch": 0.752756787013714,
+      "grad_norm": 1.215842366218567,
+      "learning_rate": 3.214450600184672e-05,
+      "loss": 0.3012,
+      "step": 1681
+    },
+    {
+      "epoch": 0.7532045899804086,
+      "grad_norm": 1.4815665483474731,
+      "learning_rate": 3.213296398891967e-05,
+      "loss": 0.1521,
+      "step": 1682
+    },
+    {
+      "epoch": 0.7536523929471033,
+      "grad_norm": 1.3670876026153564,
+      "learning_rate": 3.2121421975992614e-05,
+      "loss": 0.1288,
+      "step": 1683
+    },
+    {
+      "epoch": 0.754100195913798,
+      "grad_norm": 1.0420467853546143,
+      "learning_rate": 3.2109879963065556e-05,
+      "loss": 0.1539,
+      "step": 1684
+    },
+    {
+      "epoch": 0.7545479988804926,
+      "grad_norm": 1.1879098415374756,
+      "learning_rate": 3.2098337950138506e-05,
+      "loss": 0.1517,
+      "step": 1685
+    },
+    {
+      "epoch": 0.7549958018471873,
+      "grad_norm": 1.3479251861572266,
+      "learning_rate": 3.208679593721145e-05,
+      "loss": 0.26,
+      "step": 1686
+    },
+    {
+      "epoch": 0.7554436048138818,
+      "grad_norm": 2.0593390464782715,
+      "learning_rate": 3.20752539242844e-05,
+      "loss": 0.5097,
+      "step": 1687
+    },
+    {
+      "epoch": 0.7558914077805765,
+      "grad_norm": 1.2824532985687256,
+      "learning_rate": 3.206371191135734e-05,
+      "loss": 0.2245,
+      "step": 1688
+    },
+    {
+      "epoch": 0.7563392107472712,
+      "grad_norm": 1.1562902927398682,
+      "learning_rate": 3.205216989843029e-05,
+      "loss": 0.1307,
+      "step": 1689
+    },
+    {
+      "epoch": 0.7567870137139658,
+      "grad_norm": 0.6965445280075073,
+      "learning_rate": 3.204062788550323e-05,
+      "loss": 0.0786,
+      "step": 1690
+    },
+    {
+      "epoch": 0.7567870137139658,
+      "eval_loss": 0.1975894570350647,
+      "eval_runtime": 1736.2667,
+      "eval_samples_per_second": 2.573,
+      "eval_steps_per_second": 2.573,
+      "step": 1690
+    },
+    {
+      "epoch": 0.7572348166806605,
+      "grad_norm": 1.2004364728927612,
+      "learning_rate": 3.202908587257618e-05,
+      "loss": 0.2382,
+      "step": 1691
+    },
+    {
+      "epoch": 0.7576826196473552,
+      "grad_norm": 0.8067834377288818,
+      "learning_rate": 3.2017543859649124e-05,
+      "loss": 0.081,
+      "step": 1692
+    },
+    {
+      "epoch": 0.7581304226140498,
+      "grad_norm": 1.2369734048843384,
+      "learning_rate": 3.2006001846722074e-05,
+      "loss": 0.1239,
+      "step": 1693
+    },
+    {
+      "epoch": 0.7585782255807445,
+      "grad_norm": 1.0439966917037964,
+      "learning_rate": 3.1994459833795016e-05,
+      "loss": 0.1213,
+      "step": 1694
+    },
+    {
+      "epoch": 0.7590260285474392,
+      "grad_norm": 0.9752621650695801,
+      "learning_rate": 3.1982917820867966e-05,
+      "loss": 0.1779,
+      "step": 1695
+    },
+    {
+      "epoch": 0.7594738315141338,
+      "grad_norm": 1.333712100982666,
+      "learning_rate": 3.197137580794091e-05,
+      "loss": 0.1714,
+      "step": 1696
+    },
+    {
+      "epoch": 0.7599216344808284,
+      "grad_norm": 1.0253794193267822,
+      "learning_rate": 3.195983379501385e-05,
+      "loss": 0.1056,
+      "step": 1697
+    },
+    {
+      "epoch": 0.7603694374475231,
+      "grad_norm": 1.007927656173706,
+      "learning_rate": 3.19482917820868e-05,
+      "loss": 0.1837,
+      "step": 1698
+    },
+    {
+      "epoch": 0.7608172404142177,
+      "grad_norm": 1.5209026336669922,
+      "learning_rate": 3.193674976915974e-05,
+      "loss": 0.1403,
+      "step": 1699
+    },
+    {
+      "epoch": 0.7612650433809124,
+      "grad_norm": 1.6739745140075684,
+      "learning_rate": 3.192520775623269e-05,
+      "loss": 0.2188,
+      "step": 1700
+    },
+    {
+      "epoch": 0.7612650433809124,
+      "eval_loss": 0.19731631875038147,
+      "eval_runtime": 1732.9245,
+      "eval_samples_per_second": 2.578,
+      "eval_steps_per_second": 2.578,
+      "step": 1700
+    },
+    {
+      "epoch": 0.761712846347607,
+      "grad_norm": 1.293473482131958,
+      "learning_rate": 3.1913665743305635e-05,
+      "loss": 0.1678,
+      "step": 1701
+    },
+    {
+      "epoch": 0.7621606493143017,
+      "grad_norm": 1.1437653303146362,
+      "learning_rate": 3.190212373037858e-05,
+      "loss": 0.2575,
+      "step": 1702
+    },
+    {
+      "epoch": 0.7626084522809964,
+      "grad_norm": 1.0263285636901855,
+      "learning_rate": 3.189058171745152e-05,
+      "loss": 0.2893,
+      "step": 1703
+    },
+    {
+      "epoch": 0.763056255247691,
+      "grad_norm": 1.3024674654006958,
+      "learning_rate": 3.187903970452447e-05,
+      "loss": 0.1215,
+      "step": 1704
+    },
+    {
+      "epoch": 0.7635040582143857,
+      "grad_norm": 1.4158358573913574,
+      "learning_rate": 3.186749769159741e-05,
+      "loss": 0.2106,
+      "step": 1705
+    },
+    {
+      "epoch": 0.7639518611810803,
+      "grad_norm": 0.9089943170547485,
+      "learning_rate": 3.185595567867036e-05,
+      "loss": 0.1623,
+      "step": 1706
+    },
+    {
+      "epoch": 0.7643996641477749,
+      "grad_norm": 1.3617576360702515,
+      "learning_rate": 3.1844413665743304e-05,
+      "loss": 0.2851,
+      "step": 1707
+    },
+    {
+      "epoch": 0.7648474671144696,
+      "grad_norm": 1.1946085691452026,
+      "learning_rate": 3.183287165281625e-05,
+      "loss": 0.17,
+      "step": 1708
+    },
+    {
+      "epoch": 0.7652952700811643,
+      "grad_norm": 1.0318683385849,
+      "learning_rate": 3.1821329639889196e-05,
+      "loss": 0.212,
+      "step": 1709
+    },
+    {
+      "epoch": 0.7657430730478589,
+      "grad_norm": 2.042214870452881,
+      "learning_rate": 3.1809787626962145e-05,
+      "loss": 0.2701,
+      "step": 1710
+    },
+    {
+      "epoch": 0.7657430730478589,
+      "eval_loss": 0.19515901803970337,
+      "eval_runtime": 1728.648,
+      "eval_samples_per_second": 2.584,
+      "eval_steps_per_second": 2.584,
+      "step": 1710
+    },
+    {
+      "epoch": 0.7661908760145536,
+      "grad_norm": 0.8775007724761963,
+      "learning_rate": 3.179824561403509e-05,
+      "loss": 0.2296,
+      "step": 1711
+    },
+    {
+      "epoch": 0.7666386789812483,
+      "grad_norm": 0.797558605670929,
+      "learning_rate": 3.178670360110804e-05,
+      "loss": 0.1135,
+      "step": 1712
+    },
+    {
+      "epoch": 0.7670864819479429,
+      "grad_norm": 1.1293439865112305,
+      "learning_rate": 3.177516158818098e-05,
+      "loss": 0.1872,
+      "step": 1713
+    },
+    {
+      "epoch": 0.7675342849146376,
+      "grad_norm": 1.4189733266830444,
+      "learning_rate": 3.176361957525393e-05,
+      "loss": 0.2343,
+      "step": 1714
+    },
+    {
+      "epoch": 0.7679820878813323,
+      "grad_norm": 1.0621099472045898,
+      "learning_rate": 3.175207756232687e-05,
+      "loss": 0.2167,
+      "step": 1715
+    },
+    {
+      "epoch": 0.7684298908480268,
+      "grad_norm": 1.3043614625930786,
+      "learning_rate": 3.174053554939982e-05,
+      "loss": 0.3389,
+      "step": 1716
+    },
+    {
+      "epoch": 0.7688776938147215,
+      "grad_norm": 0.9801878333091736,
+      "learning_rate": 3.1728993536472764e-05,
+      "loss": 0.102,
+      "step": 1717
+    },
+    {
+      "epoch": 0.7693254967814162,
+      "grad_norm": 1.019263744354248,
+      "learning_rate": 3.171745152354571e-05,
+      "loss": 0.1738,
+      "step": 1718
+    },
+    {
+      "epoch": 0.7697732997481108,
+      "grad_norm": 1.5536470413208008,
+      "learning_rate": 3.1705909510618656e-05,
+      "loss": 0.2715,
+      "step": 1719
+    },
+    {
+      "epoch": 0.7702211027148055,
+      "grad_norm": 1.2832285165786743,
+      "learning_rate": 3.16943674976916e-05,
+      "loss": 0.1412,
+      "step": 1720
+    },
+    {
+      "epoch": 0.7702211027148055,
+      "eval_loss": 0.19498112797737122,
+      "eval_runtime": 1726.0687,
+      "eval_samples_per_second": 2.588,
+      "eval_steps_per_second": 2.588,
+      "step": 1720
+    },
+    {
+      "epoch": 0.7706689056815001,
+      "grad_norm": 1.2903026342391968,
+      "learning_rate": 3.168282548476455e-05,
+      "loss": 0.175,
+      "step": 1721
+    },
+    {
+      "epoch": 0.7711167086481948,
+      "grad_norm": 1.4443283081054688,
+      "learning_rate": 3.167128347183749e-05,
+      "loss": 0.245,
+      "step": 1722
+    },
+    {
+      "epoch": 0.7715645116148895,
+      "grad_norm": 1.4279271364212036,
+      "learning_rate": 3.165974145891043e-05,
+      "loss": 0.2023,
+      "step": 1723
+    },
+    {
+      "epoch": 0.7720123145815841,
+      "grad_norm": 0.9861289858818054,
+      "learning_rate": 3.1648199445983376e-05,
+      "loss": 0.2625,
+      "step": 1724
+    },
+    {
+      "epoch": 0.7724601175482787,
+      "grad_norm": 1.038965106010437,
+      "learning_rate": 3.1636657433056325e-05,
+      "loss": 0.0921,
+      "step": 1725
+    },
+    {
+      "epoch": 0.7729079205149734,
+      "grad_norm": 1.2597821950912476,
+      "learning_rate": 3.162511542012927e-05,
+      "loss": 0.1168,
+      "step": 1726
+    },
+    {
+      "epoch": 0.773355723481668,
+      "grad_norm": 1.182401418685913,
+      "learning_rate": 3.161357340720222e-05,
+      "loss": 0.1085,
+      "step": 1727
+    },
+    {
+      "epoch": 0.7738035264483627,
+      "grad_norm": 1.0120970010757446,
+      "learning_rate": 3.160203139427516e-05,
+      "loss": 0.0935,
+      "step": 1728
+    },
+    {
+      "epoch": 0.7742513294150574,
+      "grad_norm": 1.0572340488433838,
+      "learning_rate": 3.159048938134811e-05,
+      "loss": 0.1821,
+      "step": 1729
+    },
+    {
+      "epoch": 0.774699132381752,
+      "grad_norm": 1.1466655731201172,
+      "learning_rate": 3.157894736842105e-05,
+      "loss": 0.1782,
+      "step": 1730
+    },
+    {
+      "epoch": 0.774699132381752,
+      "eval_loss": 0.19265195727348328,
+      "eval_runtime": 1728.1783,
+      "eval_samples_per_second": 2.585,
+      "eval_steps_per_second": 2.585,
+      "step": 1730
+    },
+    {
+      "epoch": 0.7751469353484467,
+      "grad_norm": 0.9281951189041138,
+      "learning_rate": 3.1567405355494e-05,
+      "loss": 0.0994,
+      "step": 1731
+    },
+    {
+      "epoch": 0.7755947383151414,
+      "grad_norm": 1.089085578918457,
+      "learning_rate": 3.1555863342566944e-05,
+      "loss": 0.1445,
+      "step": 1732
+    },
+    {
+      "epoch": 0.776042541281836,
+      "grad_norm": 1.286909580230713,
+      "learning_rate": 3.154432132963989e-05,
+      "loss": 0.2776,
+      "step": 1733
+    },
+    {
+      "epoch": 0.7764903442485307,
+      "grad_norm": 1.348159909248352,
+      "learning_rate": 3.1532779316712836e-05,
+      "loss": 0.3719,
+      "step": 1734
+    },
+    {
+      "epoch": 0.7769381472152252,
+      "grad_norm": 1.3471921682357788,
+      "learning_rate": 3.1521237303785785e-05,
+      "loss": 0.1505,
+      "step": 1735
+    },
+    {
+      "epoch": 0.7773859501819199,
+      "grad_norm": 0.8467832207679749,
+      "learning_rate": 3.150969529085873e-05,
+      "loss": 0.0806,
+      "step": 1736
+    },
+    {
+      "epoch": 0.7778337531486146,
+      "grad_norm": 1.291153907775879,
+      "learning_rate": 3.149815327793168e-05,
+      "loss": 0.2666,
+      "step": 1737
+    },
+    {
+      "epoch": 0.7782815561153092,
+      "grad_norm": 1.6741520166397095,
+      "learning_rate": 3.148661126500462e-05,
+      "loss": 0.2447,
+      "step": 1738
+    },
+    {
+      "epoch": 0.7787293590820039,
+      "grad_norm": 1.1718069314956665,
+      "learning_rate": 3.147506925207757e-05,
+      "loss": 0.2378,
+      "step": 1739
+    },
+    {
+      "epoch": 0.7791771620486986,
+      "grad_norm": 1.080367088317871,
+      "learning_rate": 3.146352723915051e-05,
+      "loss": 0.2847,
+      "step": 1740
+    },
+    {
+      "epoch": 0.7791771620486986,
+      "eval_loss": 0.19086259603500366,
+      "eval_runtime": 1723.5914,
+      "eval_samples_per_second": 2.592,
+      "eval_steps_per_second": 2.592,
+      "step": 1740
+    },
+    {
+      "epoch": 0.7796249650153932,
+      "grad_norm": 0.9889196753501892,
+      "learning_rate": 3.1451985226223454e-05,
+      "loss": 0.0902,
+      "step": 1741
+    },
+    {
+      "epoch": 0.7800727679820879,
+      "grad_norm": 1.2149578332901,
+      "learning_rate": 3.14404432132964e-05,
+      "loss": 0.3468,
+      "step": 1742
+    },
+    {
+      "epoch": 0.7805205709487826,
+      "grad_norm": 1.186657190322876,
+      "learning_rate": 3.1428901200369346e-05,
+      "loss": 0.1753,
+      "step": 1743
+    },
+    {
+      "epoch": 0.7809683739154772,
+      "grad_norm": 0.8789498805999756,
+      "learning_rate": 3.141735918744229e-05,
+      "loss": 0.1154,
+      "step": 1744
+    },
+    {
+      "epoch": 0.7814161768821718,
+      "grad_norm": 0.8804444074630737,
+      "learning_rate": 3.140581717451524e-05,
+      "loss": 0.1065,
+      "step": 1745
+    },
+    {
+      "epoch": 0.7818639798488665,
+      "grad_norm": 1.1631603240966797,
+      "learning_rate": 3.139427516158818e-05,
+      "loss": 0.2699,
+      "step": 1746
+    },
+    {
+      "epoch": 0.7823117828155611,
+      "grad_norm": 1.3118648529052734,
+      "learning_rate": 3.1382733148661123e-05,
+      "loss": 0.2221,
+      "step": 1747
+    },
+    {
+      "epoch": 0.7827595857822558,
+      "grad_norm": 1.4373090267181396,
+      "learning_rate": 3.137119113573407e-05,
+      "loss": 0.1979,
+      "step": 1748
+    },
+    {
+      "epoch": 0.7832073887489505,
+      "grad_norm": 1.4228724241256714,
+      "learning_rate": 3.1359649122807015e-05,
+      "loss": 0.2494,
+      "step": 1749
+    },
+    {
+      "epoch": 0.7836551917156451,
+      "grad_norm": 1.2517611980438232,
+      "learning_rate": 3.1348107109879965e-05,
+      "loss": 0.1731,
+      "step": 1750
+    },
+    {
+      "epoch": 0.7836551917156451,
+      "eval_loss": 0.18768282234668732,
+      "eval_runtime": 1728.9333,
+      "eval_samples_per_second": 2.584,
+      "eval_steps_per_second": 2.584,
+      "step": 1750
+    },
+    {
+      "epoch": 0.7841029946823398,
+      "grad_norm": 1.355423927307129,
+      "learning_rate": 3.133656509695291e-05,
+      "loss": 0.2274,
+      "step": 1751
+    },
+    {
+      "epoch": 0.7845507976490345,
+      "grad_norm": 1.472356915473938,
+      "learning_rate": 3.132502308402586e-05,
+      "loss": 0.2816,
+      "step": 1752
+    },
+    {
+      "epoch": 0.7849986006157291,
+      "grad_norm": 1.4461036920547485,
+      "learning_rate": 3.13134810710988e-05,
+      "loss": 0.2883,
+      "step": 1753
+    },
+    {
+      "epoch": 0.7854464035824237,
+      "grad_norm": 1.3502665758132935,
+      "learning_rate": 3.130193905817175e-05,
+      "loss": 0.1779,
+      "step": 1754
+    },
+    {
+      "epoch": 0.7858942065491183,
+      "grad_norm": 0.954616904258728,
+      "learning_rate": 3.129039704524469e-05,
+      "loss": 0.1198,
+      "step": 1755
+    },
+    {
+      "epoch": 0.786342009515813,
+      "grad_norm": 0.916275143623352,
+      "learning_rate": 3.127885503231764e-05,
+      "loss": 0.1592,
+      "step": 1756
+    },
+    {
+      "epoch": 0.7867898124825077,
+      "grad_norm": 1.4052035808563232,
+      "learning_rate": 3.126731301939058e-05,
+      "loss": 0.4243,
+      "step": 1757
+    },
+    {
+      "epoch": 0.7872376154492023,
+      "grad_norm": 1.2046221494674683,
+      "learning_rate": 3.125577100646353e-05,
+      "loss": 0.2051,
+      "step": 1758
+    },
+    {
+      "epoch": 0.787685418415897,
+      "grad_norm": 0.7101848125457764,
+      "learning_rate": 3.1244228993536475e-05,
+      "loss": 0.086,
+      "step": 1759
+    },
+    {
+      "epoch": 0.7881332213825917,
+      "grad_norm": 1.0928889513015747,
+      "learning_rate": 3.123268698060942e-05,
+      "loss": 0.2118,
+      "step": 1760
+    },
+    {
+      "epoch": 0.7881332213825917,
+      "eval_loss": 0.18407580256462097,
+      "eval_runtime": 1725.0622,
+      "eval_samples_per_second": 2.589,
+      "eval_steps_per_second": 2.589,
+      "step": 1760
+    },
+    {
+      "epoch": 0.7885810243492863,
+      "grad_norm": 1.032731294631958,
+      "learning_rate": 3.122114496768237e-05,
+      "loss": 0.1622,
+      "step": 1761
+    },
+    {
+      "epoch": 0.789028827315981,
+      "grad_norm": 1.4532630443572998,
+      "learning_rate": 3.120960295475531e-05,
+      "loss": 0.1498,
+      "step": 1762
+    },
+    {
+      "epoch": 0.7894766302826757,
+      "grad_norm": 0.9604041576385498,
+      "learning_rate": 3.119806094182825e-05,
+      "loss": 0.0749,
+      "step": 1763
+    },
+    {
+      "epoch": 0.7899244332493702,
+      "grad_norm": 1.3322830200195312,
+      "learning_rate": 3.11865189289012e-05,
+      "loss": 0.2035,
+      "step": 1764
+    },
+    {
+      "epoch": 0.7903722362160649,
+      "grad_norm": 0.9661667943000793,
+      "learning_rate": 3.1174976915974145e-05,
+      "loss": 0.1066,
+      "step": 1765
+    },
+    {
+      "epoch": 0.7908200391827596,
+      "grad_norm": 1.23445463180542,
+      "learning_rate": 3.1163434903047094e-05,
+      "loss": 0.2315,
+      "step": 1766
+    },
+    {
+      "epoch": 0.7912678421494542,
+      "grad_norm": 0.9666409492492676,
+      "learning_rate": 3.1151892890120037e-05,
+      "loss": 0.0874,
+      "step": 1767
+    },
+    {
+      "epoch": 0.7917156451161489,
+      "grad_norm": 1.6275707483291626,
+      "learning_rate": 3.1140350877192986e-05,
+      "loss": 0.3015,
+      "step": 1768
+    },
+    {
+      "epoch": 0.7921634480828436,
+      "grad_norm": 1.0952938795089722,
+      "learning_rate": 3.112880886426593e-05,
+      "loss": 0.1406,
+      "step": 1769
+    },
+    {
+      "epoch": 0.7926112510495382,
+      "grad_norm": 1.5493882894515991,
+      "learning_rate": 3.111726685133888e-05,
+      "loss": 0.2373,
+      "step": 1770
+    },
+    {
+      "epoch": 0.7926112510495382,
+      "eval_loss": 0.18200978636741638,
+      "eval_runtime": 1742.6654,
+      "eval_samples_per_second": 2.563,
+      "eval_steps_per_second": 2.563,
+      "step": 1770
+    },
+    {
+      "epoch": 0.7930590540162329,
+      "grad_norm": 1.2741529941558838,
+      "learning_rate": 3.110572483841182e-05,
+      "loss": 0.2427,
+      "step": 1771
+    },
+    {
+      "epoch": 0.7935068569829276,
+      "grad_norm": 0.6875128149986267,
+      "learning_rate": 3.109418282548476e-05,
+      "loss": 0.0656,
+      "step": 1772
+    },
+    {
+      "epoch": 0.7939546599496222,
+      "grad_norm": 1.2775335311889648,
+      "learning_rate": 3.108264081255771e-05,
+      "loss": 0.2231,
+      "step": 1773
+    },
+    {
+      "epoch": 0.7944024629163168,
+      "grad_norm": 1.0050452947616577,
+      "learning_rate": 3.1071098799630655e-05,
+      "loss": 0.125,
+      "step": 1774
+    },
+    {
+      "epoch": 0.7948502658830114,
+      "grad_norm": 1.2733467817306519,
+      "learning_rate": 3.1059556786703604e-05,
+      "loss": 0.1813,
+      "step": 1775
+    },
+    {
+      "epoch": 0.7952980688497061,
+      "grad_norm": 1.0973080396652222,
+      "learning_rate": 3.104801477377655e-05,
+      "loss": 0.1617,
+      "step": 1776
+    },
+    {
+      "epoch": 0.7957458718164008,
+      "grad_norm": 1.3167555332183838,
+      "learning_rate": 3.1036472760849496e-05,
+      "loss": 0.1869,
+      "step": 1777
+    },
+    {
+      "epoch": 0.7961936747830954,
+      "grad_norm": 1.404090404510498,
+      "learning_rate": 3.102493074792244e-05,
+      "loss": 0.1659,
+      "step": 1778
+    },
+    {
+      "epoch": 0.7966414777497901,
+      "grad_norm": 1.5618826150894165,
+      "learning_rate": 3.101338873499539e-05,
+      "loss": 0.2905,
+      "step": 1779
+    },
+    {
+      "epoch": 0.7970892807164848,
+      "grad_norm": 1.1656345129013062,
+      "learning_rate": 3.100184672206833e-05,
+      "loss": 0.1532,
+      "step": 1780
+    },
+    {
+      "epoch": 0.7970892807164848,
+      "eval_loss": 0.18052628636360168,
+      "eval_runtime": 1741.697,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 2.565,
+      "step": 1780
+    },
+    {
+      "epoch": 0.7975370836831794,
+      "grad_norm": 0.970252275466919,
+      "learning_rate": 3.0990304709141274e-05,
+      "loss": 0.1948,
+      "step": 1781
+    },
+    {
+      "epoch": 0.7979848866498741,
+      "grad_norm": 1.1388535499572754,
+      "learning_rate": 3.0978762696214216e-05,
+      "loss": 0.1178,
+      "step": 1782
+    },
+    {
+      "epoch": 0.7984326896165687,
+      "grad_norm": 1.0984439849853516,
+      "learning_rate": 3.0967220683287166e-05,
+      "loss": 0.1222,
+      "step": 1783
+    },
+    {
+      "epoch": 0.7988804925832633,
+      "grad_norm": 1.2920303344726562,
+      "learning_rate": 3.095567867036011e-05,
+      "loss": 0.2673,
+      "step": 1784
+    },
+    {
+      "epoch": 0.799328295549958,
+      "grad_norm": 0.9241845011711121,
+      "learning_rate": 3.094413665743306e-05,
+      "loss": 0.1297,
+      "step": 1785
+    },
+    {
+      "epoch": 0.7997760985166527,
+      "grad_norm": 1.1384260654449463,
+      "learning_rate": 3.0932594644506e-05,
+      "loss": 0.1618,
+      "step": 1786
+    },
+    {
+      "epoch": 0.8002239014833473,
+      "grad_norm": 1.1058518886566162,
+      "learning_rate": 3.092105263157895e-05,
+      "loss": 0.1894,
+      "step": 1787
+    },
+    {
+      "epoch": 0.800671704450042,
+      "grad_norm": 1.4290608167648315,
+      "learning_rate": 3.090951061865189e-05,
+      "loss": 0.2922,
+      "step": 1788
+    },
+    {
+      "epoch": 0.8011195074167367,
+      "grad_norm": 1.589939832687378,
+      "learning_rate": 3.089796860572484e-05,
+      "loss": 0.2531,
+      "step": 1789
+    },
+    {
+      "epoch": 0.8015673103834313,
+      "grad_norm": 1.2515995502471924,
+      "learning_rate": 3.0886426592797784e-05,
+      "loss": 0.1397,
+      "step": 1790
+    },
+    {
+      "epoch": 0.8015673103834313,
+      "eval_loss": 0.17911021411418915,
+      "eval_runtime": 1743.1592,
+      "eval_samples_per_second": 2.563,
+      "eval_steps_per_second": 2.563,
+      "step": 1790
+    },
+    {
+      "epoch": 0.802015113350126,
+      "grad_norm": 1.4639934301376343,
+      "learning_rate": 3.0874884579870734e-05,
+      "loss": 0.155,
+      "step": 1791
+    },
+    {
+      "epoch": 0.8024629163168207,
+      "grad_norm": 1.5946593284606934,
+      "learning_rate": 3.0863342566943676e-05,
+      "loss": 0.335,
+      "step": 1792
+    },
+    {
+      "epoch": 0.8029107192835152,
+      "grad_norm": 0.9911496639251709,
+      "learning_rate": 3.0851800554016626e-05,
+      "loss": 0.1262,
+      "step": 1793
+    },
+    {
+      "epoch": 0.8033585222502099,
+      "grad_norm": 0.9423514604568481,
+      "learning_rate": 3.084025854108957e-05,
+      "loss": 0.1645,
+      "step": 1794
+    },
+    {
+      "epoch": 0.8038063252169045,
+      "grad_norm": 1.6770713329315186,
+      "learning_rate": 3.082871652816252e-05,
+      "loss": 0.2112,
+      "step": 1795
+    },
+    {
+      "epoch": 0.8042541281835992,
+      "grad_norm": 1.091598629951477,
+      "learning_rate": 3.081717451523546e-05,
+      "loss": 0.1308,
+      "step": 1796
+    },
+    {
+      "epoch": 0.8047019311502939,
+      "grad_norm": 1.0683177709579468,
+      "learning_rate": 3.08056325023084e-05,
+      "loss": 0.2161,
+      "step": 1797
+    },
+    {
+      "epoch": 0.8051497341169885,
+      "grad_norm": 1.2820827960968018,
+      "learning_rate": 3.079409048938135e-05,
+      "loss": 0.1894,
+      "step": 1798
+    },
+    {
+      "epoch": 0.8055975370836832,
+      "grad_norm": 1.0889778137207031,
+      "learning_rate": 3.0782548476454295e-05,
+      "loss": 0.1841,
+      "step": 1799
+    },
+    {
+      "epoch": 0.8060453400503779,
+      "grad_norm": 0.9384577870368958,
+      "learning_rate": 3.0771006463527244e-05,
+      "loss": 0.1522,
+      "step": 1800
+    },
+    {
+      "epoch": 0.8060453400503779,
+      "eval_loss": 0.1766439974308014,
+      "eval_runtime": 1744.9642,
+      "eval_samples_per_second": 2.56,
+      "eval_steps_per_second": 2.56,
+      "step": 1800
+    },
+    {
+      "epoch": 0.8064931430170725,
+      "grad_norm": 1.1987580060958862,
+      "learning_rate": 3.075946445060019e-05,
+      "loss": 0.1167,
+      "step": 1801
+    },
+    {
+      "epoch": 0.8069409459837671,
+      "grad_norm": 1.2142102718353271,
+      "learning_rate": 3.074792243767313e-05,
+      "loss": 0.1917,
+      "step": 1802
+    },
+    {
+      "epoch": 0.8073887489504618,
+      "grad_norm": 1.063592553138733,
+      "learning_rate": 3.073638042474607e-05,
+      "loss": 0.196,
+      "step": 1803
+    },
+    {
+      "epoch": 0.8078365519171564,
+      "grad_norm": 0.9051306843757629,
+      "learning_rate": 3.072483841181902e-05,
+      "loss": 0.104,
+      "step": 1804
+    },
+    {
+      "epoch": 0.8082843548838511,
+      "grad_norm": 1.501448392868042,
+      "learning_rate": 3.0713296398891964e-05,
+      "loss": 0.1321,
+      "step": 1805
+    },
+    {
+      "epoch": 0.8087321578505458,
+      "grad_norm": 1.3337589502334595,
+      "learning_rate": 3.0701754385964913e-05,
+      "loss": 0.1714,
+      "step": 1806
+    },
+    {
+      "epoch": 0.8091799608172404,
+      "grad_norm": 1.1337471008300781,
+      "learning_rate": 3.0690212373037856e-05,
+      "loss": 0.1375,
+      "step": 1807
+    },
+    {
+      "epoch": 0.8096277637839351,
+      "grad_norm": 0.8848599195480347,
+      "learning_rate": 3.0678670360110805e-05,
+      "loss": 0.0759,
+      "step": 1808
+    },
+    {
+      "epoch": 0.8100755667506297,
+      "grad_norm": 1.6946122646331787,
+      "learning_rate": 3.066712834718375e-05,
+      "loss": 0.1689,
+      "step": 1809
+    },
+    {
+      "epoch": 0.8105233697173244,
+      "grad_norm": 1.4061602354049683,
+      "learning_rate": 3.06555863342567e-05,
+      "loss": 0.2621,
+      "step": 1810
+    },
+    {
+      "epoch": 0.8105233697173244,
+      "eval_loss": 0.17439012229442596,
+      "eval_runtime": 1743.6105,
+      "eval_samples_per_second": 2.562,
+      "eval_steps_per_second": 2.562,
+      "step": 1810
+    },
+    {
+      "epoch": 0.8109711726840191,
+      "grad_norm": 1.1560529470443726,
+      "learning_rate": 3.064404432132964e-05,
+      "loss": 0.1473,
+      "step": 1811
+    },
+    {
+      "epoch": 0.8114189756507136,
+      "grad_norm": 1.1303647756576538,
+      "learning_rate": 3.063250230840259e-05,
+      "loss": 0.1474,
+      "step": 1812
+    },
+    {
+      "epoch": 0.8118667786174083,
+      "grad_norm": 1.3138867616653442,
+      "learning_rate": 3.062096029547553e-05,
+      "loss": 0.1713,
+      "step": 1813
+    },
+    {
+      "epoch": 0.812314581584103,
+      "grad_norm": 1.5874935388565063,
+      "learning_rate": 3.060941828254848e-05,
+      "loss": 0.2144,
+      "step": 1814
+    },
+    {
+      "epoch": 0.8127623845507976,
+      "grad_norm": 1.0073288679122925,
+      "learning_rate": 3.0597876269621424e-05,
+      "loss": 0.1259,
+      "step": 1815
+    },
+    {
+      "epoch": 0.8132101875174923,
+      "grad_norm": 1.4054961204528809,
+      "learning_rate": 3.058633425669437e-05,
+      "loss": 0.1752,
+      "step": 1816
+    },
+    {
+      "epoch": 0.813657990484187,
+      "grad_norm": 1.0050817728042603,
+      "learning_rate": 3.0574792243767316e-05,
+      "loss": 0.1,
+      "step": 1817
+    },
+    {
+      "epoch": 0.8141057934508816,
+      "grad_norm": 0.8871022462844849,
+      "learning_rate": 3.0563250230840265e-05,
+      "loss": 0.0854,
+      "step": 1818
+    },
+    {
+      "epoch": 0.8145535964175763,
+      "grad_norm": 0.8566880822181702,
+      "learning_rate": 3.055170821791321e-05,
+      "loss": 0.0894,
+      "step": 1819
+    },
+    {
+      "epoch": 0.815001399384271,
+      "grad_norm": 1.0436166524887085,
+      "learning_rate": 3.054016620498615e-05,
+      "loss": 0.1256,
+      "step": 1820
+    },
+    {
+      "epoch": 0.815001399384271,
+      "eval_loss": 0.17154671251773834,
+      "eval_runtime": 1740.5344,
+      "eval_samples_per_second": 2.566,
+      "eval_steps_per_second": 2.566,
+      "step": 1820
+    },
+    {
+      "epoch": 0.8154492023509656,
+      "grad_norm": 1.0768319368362427,
+      "learning_rate": 3.052862419205909e-05,
+      "loss": 0.1093,
+      "step": 1821
+    },
+    {
+      "epoch": 0.8158970053176602,
+      "grad_norm": 0.9550846219062805,
+      "learning_rate": 3.051708217913204e-05,
+      "loss": 0.1221,
+      "step": 1822
+    },
+    {
+      "epoch": 0.8163448082843549,
+      "grad_norm": 1.3339128494262695,
+      "learning_rate": 3.050554016620499e-05,
+      "loss": 0.3042,
+      "step": 1823
+    },
+    {
+      "epoch": 0.8167926112510495,
+      "grad_norm": 1.2556220293045044,
+      "learning_rate": 3.049399815327793e-05,
+      "loss": 0.1797,
+      "step": 1824
+    },
+    {
+      "epoch": 0.8172404142177442,
+      "grad_norm": 1.1306003332138062,
+      "learning_rate": 3.048245614035088e-05,
+      "loss": 0.1557,
+      "step": 1825
+    },
+    {
+      "epoch": 0.8176882171844388,
+      "grad_norm": 1.104524850845337,
+      "learning_rate": 3.0470914127423823e-05,
+      "loss": 0.1996,
+      "step": 1826
+    },
+    {
+      "epoch": 0.8181360201511335,
+      "grad_norm": 1.4180043935775757,
+      "learning_rate": 3.045937211449677e-05,
+      "loss": 0.1997,
+      "step": 1827
+    },
+    {
+      "epoch": 0.8185838231178282,
+      "grad_norm": 1.6681851148605347,
+      "learning_rate": 3.0447830101569712e-05,
+      "loss": 0.1433,
+      "step": 1828
+    },
+    {
+      "epoch": 0.8190316260845228,
+      "grad_norm": 1.2225019931793213,
+      "learning_rate": 3.043628808864266e-05,
+      "loss": 0.1352,
+      "step": 1829
+    },
+    {
+      "epoch": 0.8194794290512175,
+      "grad_norm": 1.310770034790039,
+      "learning_rate": 3.0424746075715604e-05,
+      "loss": 0.135,
+      "step": 1830
+    },
+    {
+      "epoch": 0.8194794290512175,
+      "eval_loss": 0.16900593042373657,
+      "eval_runtime": 1740.5996,
+      "eval_samples_per_second": 2.566,
+      "eval_steps_per_second": 2.566,
+      "step": 1830
+    },
+    {
+      "epoch": 0.8199272320179121,
+      "grad_norm": 1.2285445928573608,
+      "learning_rate": 3.0413204062788553e-05,
+      "loss": 0.1948,
+      "step": 1831
+    },
+    {
+      "epoch": 0.8203750349846067,
+      "grad_norm": 1.2398518323898315,
+      "learning_rate": 3.0401662049861496e-05,
+      "loss": 0.2515,
+      "step": 1832
+    },
+    {
+      "epoch": 0.8208228379513014,
+      "grad_norm": 1.177215814590454,
+      "learning_rate": 3.0390120036934445e-05,
+      "loss": 0.1068,
+      "step": 1833
+    },
+    {
+      "epoch": 0.8212706409179961,
+      "grad_norm": 1.2884360551834106,
+      "learning_rate": 3.0378578024007388e-05,
+      "loss": 0.2032,
+      "step": 1834
+    },
+    {
+      "epoch": 0.8217184438846907,
+      "grad_norm": 1.3158836364746094,
+      "learning_rate": 3.0367036011080334e-05,
+      "loss": 0.1355,
+      "step": 1835
+    },
+    {
+      "epoch": 0.8221662468513854,
+      "grad_norm": 1.5116773843765259,
+      "learning_rate": 3.035549399815328e-05,
+      "loss": 0.2563,
+      "step": 1836
+    },
+    {
+      "epoch": 0.8226140498180801,
+      "grad_norm": 0.8042264580726624,
+      "learning_rate": 3.0343951985226226e-05,
+      "loss": 0.0907,
+      "step": 1837
+    },
+    {
+      "epoch": 0.8230618527847747,
+      "grad_norm": 0.9426969289779663,
+      "learning_rate": 3.0332409972299168e-05,
+      "loss": 0.1182,
+      "step": 1838
+    },
+    {
+      "epoch": 0.8235096557514694,
+      "grad_norm": 1.0019307136535645,
+      "learning_rate": 3.0320867959372118e-05,
+      "loss": 0.1143,
+      "step": 1839
+    },
+    {
+      "epoch": 0.8239574587181641,
+      "grad_norm": 1.2967493534088135,
+      "learning_rate": 3.030932594644506e-05,
+      "loss": 0.1422,
+      "step": 1840
+    },
+    {
+      "epoch": 0.8239574587181641,
+      "eval_loss": 0.16731417179107666,
+      "eval_runtime": 1740.7881,
+      "eval_samples_per_second": 2.566,
+      "eval_steps_per_second": 2.566,
+      "step": 1840
+    },
+    {
+      "epoch": 0.8244052616848586,
+      "grad_norm": 1.4186187982559204,
+      "learning_rate": 3.029778393351801e-05,
+      "loss": 0.214,
+      "step": 1841
+    },
+    {
+      "epoch": 0.8248530646515533,
+      "grad_norm": 0.8762224316596985,
+      "learning_rate": 3.0286241920590952e-05,
+      "loss": 0.1384,
+      "step": 1842
+    },
+    {
+      "epoch": 0.825300867618248,
+      "grad_norm": 1.3150262832641602,
+      "learning_rate": 3.02746999076639e-05,
+      "loss": 0.1528,
+      "step": 1843
+    },
+    {
+      "epoch": 0.8257486705849426,
+      "grad_norm": 1.600975513458252,
+      "learning_rate": 3.0263157894736844e-05,
+      "loss": 0.1479,
+      "step": 1844
+    },
+    {
+      "epoch": 0.8261964735516373,
+      "grad_norm": 1.3228427171707153,
+      "learning_rate": 3.025161588180979e-05,
+      "loss": 0.128,
+      "step": 1845
+    },
+    {
+      "epoch": 0.8266442765183319,
+      "grad_norm": 1.5124461650848389,
+      "learning_rate": 3.0240073868882733e-05,
+      "loss": 0.1793,
+      "step": 1846
+    },
+    {
+      "epoch": 0.8270920794850266,
+      "grad_norm": 1.3643213510513306,
+      "learning_rate": 3.022853185595568e-05,
+      "loss": 0.2281,
+      "step": 1847
+    },
+    {
+      "epoch": 0.8275398824517213,
+      "grad_norm": 1.2589600086212158,
+      "learning_rate": 3.0216989843028625e-05,
+      "loss": 0.2255,
+      "step": 1848
+    },
+    {
+      "epoch": 0.8279876854184159,
+      "grad_norm": 0.9553655385971069,
+      "learning_rate": 3.0205447830101567e-05,
+      "loss": 0.1416,
+      "step": 1849
+    },
+    {
+      "epoch": 0.8284354883851105,
+      "grad_norm": 1.096134901046753,
+      "learning_rate": 3.0193905817174517e-05,
+      "loss": 0.1515,
+      "step": 1850
+    },
+    {
+      "epoch": 0.8284354883851105,
+      "eval_loss": 0.1658756136894226,
+      "eval_runtime": 1741.1344,
+      "eval_samples_per_second": 2.566,
+      "eval_steps_per_second": 2.566,
+      "step": 1850
+    },
+    {
+      "epoch": 0.8288832913518052,
+      "grad_norm": 1.5498076677322388,
+      "learning_rate": 3.018236380424746e-05,
+      "loss": 0.1799,
+      "step": 1851
+    },
+    {
+      "epoch": 0.8293310943184998,
+      "grad_norm": 1.2013968229293823,
+      "learning_rate": 3.017082179132041e-05,
+      "loss": 0.1325,
+      "step": 1852
+    },
+    {
+      "epoch": 0.8297788972851945,
+      "grad_norm": 1.2413219213485718,
+      "learning_rate": 3.015927977839335e-05,
+      "loss": 0.1424,
+      "step": 1853
+    },
+    {
+      "epoch": 0.8302267002518892,
+      "grad_norm": 1.4597523212432861,
+      "learning_rate": 3.01477377654663e-05,
+      "loss": 0.1868,
+      "step": 1854
+    },
+    {
+      "epoch": 0.8306745032185838,
+      "grad_norm": 0.7172173261642456,
+      "learning_rate": 3.0136195752539243e-05,
+      "loss": 0.0705,
+      "step": 1855
+    },
+    {
+      "epoch": 0.8311223061852785,
+      "grad_norm": 1.2041913270950317,
+      "learning_rate": 3.012465373961219e-05,
+      "loss": 0.2077,
+      "step": 1856
+    },
+    {
+      "epoch": 0.8315701091519732,
+      "grad_norm": 1.3840017318725586,
+      "learning_rate": 3.0113111726685132e-05,
+      "loss": 0.2763,
+      "step": 1857
+    },
+    {
+      "epoch": 0.8320179121186678,
+      "grad_norm": 1.2809327840805054,
+      "learning_rate": 3.010156971375808e-05,
+      "loss": 0.2566,
+      "step": 1858
+    },
+    {
+      "epoch": 0.8324657150853625,
+      "grad_norm": 0.8672228455543518,
+      "learning_rate": 3.0090027700831024e-05,
+      "loss": 0.1191,
+      "step": 1859
+    },
+    {
+      "epoch": 0.832913518052057,
+      "grad_norm": 1.0470631122589111,
+      "learning_rate": 3.0078485687903973e-05,
+      "loss": 0.0841,
+      "step": 1860
+    },
+    {
+      "epoch": 0.832913518052057,
+      "eval_loss": 0.16294632852077484,
+      "eval_runtime": 1740.0658,
+      "eval_samples_per_second": 2.567,
+      "eval_steps_per_second": 2.567,
+      "step": 1860
+    },
+    {
+      "epoch": 0.8333613210187517,
+      "grad_norm": 1.225232720375061,
+      "learning_rate": 3.0066943674976916e-05,
+      "loss": 0.1085,
+      "step": 1861
+    },
+    {
+      "epoch": 0.8338091239854464,
+      "grad_norm": 1.1180237531661987,
+      "learning_rate": 3.0055401662049865e-05,
+      "loss": 0.3731,
+      "step": 1862
+    },
+    {
+      "epoch": 0.834256926952141,
+      "grad_norm": 1.4320118427276611,
+      "learning_rate": 3.0043859649122808e-05,
+      "loss": 0.2216,
+      "step": 1863
+    },
+    {
+      "epoch": 0.8347047299188357,
+      "grad_norm": 0.8964130282402039,
+      "learning_rate": 3.0032317636195757e-05,
+      "loss": 0.1404,
+      "step": 1864
+    },
+    {
+      "epoch": 0.8351525328855304,
+      "grad_norm": 1.3593807220458984,
+      "learning_rate": 3.00207756232687e-05,
+      "loss": 0.2552,
+      "step": 1865
+    },
+    {
+      "epoch": 0.835600335852225,
+      "grad_norm": 1.2691015005111694,
+      "learning_rate": 3.0009233610341646e-05,
+      "loss": 0.151,
+      "step": 1866
+    },
+    {
+      "epoch": 0.8360481388189197,
+      "grad_norm": 1.2501640319824219,
+      "learning_rate": 2.999769159741459e-05,
+      "loss": 0.2293,
+      "step": 1867
+    },
+    {
+      "epoch": 0.8364959417856144,
+      "grad_norm": 0.9922981262207031,
+      "learning_rate": 2.9986149584487538e-05,
+      "loss": 0.0955,
+      "step": 1868
+    },
+    {
+      "epoch": 0.836943744752309,
+      "grad_norm": 1.6564064025878906,
+      "learning_rate": 2.997460757156048e-05,
+      "loss": 0.2517,
+      "step": 1869
+    },
+    {
+      "epoch": 0.8373915477190036,
+      "grad_norm": 1.8210774660110474,
+      "learning_rate": 2.996306555863343e-05,
+      "loss": 0.2969,
+      "step": 1870
+    },
+    {
+      "epoch": 0.8373915477190036,
+      "eval_loss": 0.1613447070121765,
+      "eval_runtime": 1743.648,
+      "eval_samples_per_second": 2.562,
+      "eval_steps_per_second": 2.562,
+      "step": 1870
+    },
+    {
+      "epoch": 0.8378393506856983,
+      "grad_norm": 1.4837088584899902,
+      "learning_rate": 2.9951523545706373e-05,
+      "loss": 0.1553,
+      "step": 1871
+    },
+    {
+      "epoch": 0.8382871536523929,
+      "grad_norm": 0.8095635175704956,
+      "learning_rate": 2.9939981532779315e-05,
+      "loss": 0.0953,
+      "step": 1872
+    },
+    {
+      "epoch": 0.8387349566190876,
+      "grad_norm": 0.9664444327354431,
+      "learning_rate": 2.9928439519852265e-05,
+      "loss": 0.1132,
+      "step": 1873
+    },
+    {
+      "epoch": 0.8391827595857823,
+      "grad_norm": 1.040560245513916,
+      "learning_rate": 2.9916897506925207e-05,
+      "loss": 0.1133,
+      "step": 1874
+    },
+    {
+      "epoch": 0.8396305625524769,
+      "grad_norm": 1.034613013267517,
+      "learning_rate": 2.9905355493998157e-05,
+      "loss": 0.0934,
+      "step": 1875
+    },
+    {
+      "epoch": 0.8400783655191716,
+      "grad_norm": 1.4855910539627075,
+      "learning_rate": 2.98938134810711e-05,
+      "loss": 0.1679,
+      "step": 1876
+    },
+    {
+      "epoch": 0.8405261684858663,
+      "grad_norm": 1.272095799446106,
+      "learning_rate": 2.9882271468144045e-05,
+      "loss": 0.2133,
+      "step": 1877
+    },
+    {
+      "epoch": 0.8409739714525609,
+      "grad_norm": 0.9249767661094666,
+      "learning_rate": 2.9870729455216988e-05,
+      "loss": 0.1258,
+      "step": 1878
+    },
+    {
+      "epoch": 0.8414217744192555,
+      "grad_norm": 1.2314205169677734,
+      "learning_rate": 2.9859187442289937e-05,
+      "loss": 0.1198,
+      "step": 1879
+    },
+    {
+      "epoch": 0.8418695773859501,
+      "grad_norm": 1.135740876197815,
+      "learning_rate": 2.984764542936288e-05,
+      "loss": 0.1507,
+      "step": 1880
+    },
+    {
+      "epoch": 0.8418695773859501,
+      "eval_loss": 0.1596778929233551,
+      "eval_runtime": 1744.6636,
+      "eval_samples_per_second": 2.56,
+      "eval_steps_per_second": 2.56,
+      "step": 1880
+    },
+    {
+      "epoch": 0.8423173803526448,
+      "grad_norm": 0.9603174328804016,
+      "learning_rate": 2.983610341643583e-05,
+      "loss": 0.1832,
+      "step": 1881
+    },
+    {
+      "epoch": 0.8427651833193395,
+      "grad_norm": 0.9922119975090027,
+      "learning_rate": 2.9824561403508772e-05,
+      "loss": 0.1203,
+      "step": 1882
+    },
+    {
+      "epoch": 0.8432129862860341,
+      "grad_norm": 1.2410778999328613,
+      "learning_rate": 2.981301939058172e-05,
+      "loss": 0.2341,
+      "step": 1883
+    },
+    {
+      "epoch": 0.8436607892527288,
+      "grad_norm": 1.2529115676879883,
+      "learning_rate": 2.9801477377654664e-05,
+      "loss": 0.1256,
+      "step": 1884
+    },
+    {
+      "epoch": 0.8441085922194235,
+      "grad_norm": 1.0369658470153809,
+      "learning_rate": 2.978993536472761e-05,
+      "loss": 0.2077,
+      "step": 1885
+    },
+    {
+      "epoch": 0.8445563951861181,
+      "grad_norm": 0.8883606791496277,
+      "learning_rate": 2.9778393351800556e-05,
+      "loss": 0.0912,
+      "step": 1886
+    },
+    {
+      "epoch": 0.8450041981528128,
+      "grad_norm": 1.1145747900009155,
+      "learning_rate": 2.9766851338873502e-05,
+      "loss": 0.169,
+      "step": 1887
+    },
+    {
+      "epoch": 0.8454520011195075,
+      "grad_norm": 1.1187107563018799,
+      "learning_rate": 2.9755309325946444e-05,
+      "loss": 0.1202,
+      "step": 1888
+    },
+    {
+      "epoch": 0.845899804086202,
+      "grad_norm": 1.2027275562286377,
+      "learning_rate": 2.9743767313019394e-05,
+      "loss": 0.1931,
+      "step": 1889
+    },
+    {
+      "epoch": 0.8463476070528967,
+      "grad_norm": 1.1199167966842651,
+      "learning_rate": 2.9732225300092336e-05,
+      "loss": 0.1389,
+      "step": 1890
+    },
+    {
+      "epoch": 0.8463476070528967,
+      "eval_loss": 0.157403826713562,
+      "eval_runtime": 1740.7291,
+      "eval_samples_per_second": 2.566,
+      "eval_steps_per_second": 2.566,
+      "step": 1890
+    },
+    {
+      "epoch": 0.8467954100195914,
+      "grad_norm": 1.1820210218429565,
+      "learning_rate": 2.9720683287165286e-05,
+      "loss": 0.1506,
+      "step": 1891
+    },
+    {
+      "epoch": 0.847243212986286,
+      "grad_norm": 1.314258337020874,
+      "learning_rate": 2.970914127423823e-05,
+      "loss": 0.18,
+      "step": 1892
+    },
+    {
+      "epoch": 0.8476910159529807,
+      "grad_norm": 1.2523696422576904,
+      "learning_rate": 2.9697599261311178e-05,
+      "loss": 0.1401,
+      "step": 1893
+    },
+    {
+      "epoch": 0.8481388189196754,
+      "grad_norm": 1.6710721254348755,
+      "learning_rate": 2.968605724838412e-05,
+      "loss": 0.1575,
+      "step": 1894
+    },
+    {
+      "epoch": 0.84858662188637,
+      "grad_norm": 1.1848293542861938,
+      "learning_rate": 2.9674515235457066e-05,
+      "loss": 0.1888,
+      "step": 1895
+    },
+    {
+      "epoch": 0.8490344248530647,
+      "grad_norm": 1.042680263519287,
+      "learning_rate": 2.966297322253001e-05,
+      "loss": 0.1032,
+      "step": 1896
+    },
+    {
+      "epoch": 0.8494822278197594,
+      "grad_norm": 1.2438651323318481,
+      "learning_rate": 2.9651431209602958e-05,
+      "loss": 0.1433,
+      "step": 1897
+    },
+    {
+      "epoch": 0.8499300307864539,
+      "grad_norm": 1.0681265592575073,
+      "learning_rate": 2.96398891966759e-05,
+      "loss": 0.1787,
+      "step": 1898
+    },
+    {
+      "epoch": 0.8503778337531486,
+      "grad_norm": 1.2646775245666504,
+      "learning_rate": 2.9628347183748844e-05,
+      "loss": 0.1186,
+      "step": 1899
+    },
+    {
+      "epoch": 0.8508256367198432,
+      "grad_norm": 1.3411589860916138,
+      "learning_rate": 2.9616805170821793e-05,
+      "loss": 0.25,
+      "step": 1900
+    },
+    {
+      "epoch": 0.8508256367198432,
+      "eval_loss": 0.1552475243806839,
+      "eval_runtime": 1740.6806,
+      "eval_samples_per_second": 2.566,
+      "eval_steps_per_second": 2.566,
+      "step": 1900
+    },
+    {
+      "epoch": 0.8512734396865379,
+      "grad_norm": 1.544445276260376,
+      "learning_rate": 2.9605263157894735e-05,
+      "loss": 0.2102,
+      "step": 1901
+    },
+    {
+      "epoch": 0.8517212426532326,
+      "grad_norm": 1.0015203952789307,
+      "learning_rate": 2.9593721144967685e-05,
+      "loss": 0.0845,
+      "step": 1902
+    },
+    {
+      "epoch": 0.8521690456199272,
+      "grad_norm": 0.8234665393829346,
+      "learning_rate": 2.9582179132040627e-05,
+      "loss": 0.0787,
+      "step": 1903
+    },
+    {
+      "epoch": 0.8526168485866219,
+      "grad_norm": 1.0346554517745972,
+      "learning_rate": 2.9570637119113577e-05,
+      "loss": 0.0953,
+      "step": 1904
+    },
+    {
+      "epoch": 0.8530646515533166,
+      "grad_norm": 1.5251343250274658,
+      "learning_rate": 2.955909510618652e-05,
+      "loss": 0.1945,
+      "step": 1905
+    },
+    {
+      "epoch": 0.8535124545200112,
+      "grad_norm": 1.2917299270629883,
+      "learning_rate": 2.9547553093259465e-05,
+      "loss": 0.191,
+      "step": 1906
+    },
+    {
+      "epoch": 0.8539602574867059,
+      "grad_norm": 1.1073006391525269,
+      "learning_rate": 2.9536011080332408e-05,
+      "loss": 0.1684,
+      "step": 1907
+    },
+    {
+      "epoch": 0.8544080604534005,
+      "grad_norm": 1.1586658954620361,
+      "learning_rate": 2.9524469067405357e-05,
+      "loss": 0.1412,
+      "step": 1908
+    },
+    {
+      "epoch": 0.8548558634200951,
+      "grad_norm": 0.9181356430053711,
+      "learning_rate": 2.95129270544783e-05,
+      "loss": 0.1547,
+      "step": 1909
+    },
+    {
+      "epoch": 0.8553036663867898,
+      "grad_norm": 1.1670259237289429,
+      "learning_rate": 2.950138504155125e-05,
+      "loss": 0.2027,
+      "step": 1910
+    },
+    {
+      "epoch": 0.8553036663867898,
+      "eval_loss": 0.15608276426792145,
+      "eval_runtime": 1740.3593,
+      "eval_samples_per_second": 2.567,
+      "eval_steps_per_second": 2.567,
+      "step": 1910
+    },
+    {
+      "epoch": 0.8557514693534845,
+      "grad_norm": 0.8703057765960693,
+      "learning_rate": 2.9489843028624192e-05,
+      "loss": 0.1132,
+      "step": 1911
+    },
+    {
+      "epoch": 0.8561992723201791,
+      "grad_norm": 1.286847472190857,
+      "learning_rate": 2.947830101569714e-05,
+      "loss": 0.2887,
+      "step": 1912
+    },
+    {
+      "epoch": 0.8566470752868738,
+      "grad_norm": 1.367441177368164,
+      "learning_rate": 2.9466759002770084e-05,
+      "loss": 0.1678,
+      "step": 1913
+    },
+    {
+      "epoch": 0.8570948782535684,
+      "grad_norm": 1.5393425226211548,
+      "learning_rate": 2.9455216989843033e-05,
+      "loss": 0.3088,
+      "step": 1914
+    },
+    {
+      "epoch": 0.8575426812202631,
+      "grad_norm": 1.5887715816497803,
+      "learning_rate": 2.9443674976915976e-05,
+      "loss": 0.1537,
+      "step": 1915
+    },
+    {
+      "epoch": 0.8579904841869578,
+      "grad_norm": 1.3844943046569824,
+      "learning_rate": 2.9432132963988922e-05,
+      "loss": 0.2481,
+      "step": 1916
+    },
+    {
+      "epoch": 0.8584382871536524,
+      "grad_norm": 1.306236743927002,
+      "learning_rate": 2.9420590951061865e-05,
+      "loss": 0.1246,
+      "step": 1917
+    },
+    {
+      "epoch": 0.858886090120347,
+      "grad_norm": 1.9353609085083008,
+      "learning_rate": 2.9409048938134814e-05,
+      "loss": 0.2403,
+      "step": 1918
+    },
+    {
+      "epoch": 0.8593338930870417,
+      "grad_norm": 1.0708073377609253,
+      "learning_rate": 2.9397506925207757e-05,
+      "loss": 0.1256,
+      "step": 1919
+    },
+    {
+      "epoch": 0.8597816960537363,
+      "grad_norm": 1.6582993268966675,
+      "learning_rate": 2.9385964912280706e-05,
+      "loss": 0.1708,
+      "step": 1920
+    },
+    {
+      "epoch": 0.8597816960537363,
+      "eval_loss": 0.15275830030441284,
+      "eval_runtime": 1741.7262,
+      "eval_samples_per_second": 2.565,
+      "eval_steps_per_second": 2.565,
+      "step": 1920
+    },
+    {
+      "epoch": 0.860229499020431,
+      "grad_norm": 1.1031070947647095,
+      "learning_rate": 2.937442289935365e-05,
+      "loss": 0.1313,
+      "step": 1921
+    },
+    {
+      "epoch": 0.8606773019871257,
+      "grad_norm": 1.6089885234832764,
+      "learning_rate": 2.9362880886426598e-05,
+      "loss": 0.1299,
+      "step": 1922
+    },
+    {
+      "epoch": 0.8611251049538203,
+      "grad_norm": 1.2939296960830688,
+      "learning_rate": 2.935133887349954e-05,
+      "loss": 0.1808,
+      "step": 1923
+    },
+    {
+      "epoch": 0.861572907920515,
+      "grad_norm": 0.9729406237602234,
+      "learning_rate": 2.9339796860572483e-05,
+      "loss": 0.1002,
+      "step": 1924
+    },
+    {
+      "epoch": 0.8620207108872097,
+      "grad_norm": 1.3479326963424683,
+      "learning_rate": 2.9328254847645433e-05,
+      "loss": 0.1815,
+      "step": 1925
+    },
+    {
+      "epoch": 0.8624685138539043,
+      "grad_norm": 1.0860453844070435,
+      "learning_rate": 2.9316712834718375e-05,
+      "loss": 0.2335,
+      "step": 1926
+    },
+    {
+      "epoch": 0.8629163168205989,
+      "grad_norm": 1.1711890697479248,
+      "learning_rate": 2.930517082179132e-05,
+      "loss": 0.1488,
+      "step": 1927
+    },
+    {
+      "epoch": 0.8633641197872935,
+      "grad_norm": 0.9251280426979065,
+      "learning_rate": 2.9293628808864264e-05,
+      "loss": 0.119,
+      "step": 1928
+    },
+    {
+      "epoch": 0.8638119227539882,
+      "grad_norm": 1.112289309501648,
+      "learning_rate": 2.9282086795937213e-05,
+      "loss": 0.137,
+      "step": 1929
+    },
+    {
+      "epoch": 0.8642597257206829,
+      "grad_norm": 1.7165166139602661,
+      "learning_rate": 2.9270544783010156e-05,
+      "loss": 0.1479,
+      "step": 1930
+    },
+    {
+      "epoch": 0.8642597257206829,
+      "eval_loss": 0.14976796507835388,
+      "eval_runtime": 1741.861,
+      "eval_samples_per_second": 2.564,
+      "eval_steps_per_second": 2.564,
+      "step": 1930
+    },
+    {
+      "epoch": 0.8647075286873775,
+      "grad_norm": 0.929421067237854,
+      "learning_rate": 2.9259002770083105e-05,
+      "loss": 0.1143,
+      "step": 1931
+    },
+    {
+      "epoch": 0.8651553316540722,
+      "grad_norm": 0.8255317211151123,
+      "learning_rate": 2.9247460757156048e-05,
+      "loss": 0.0891,
+      "step": 1932
+    },
+    {
+      "epoch": 0.8656031346207669,
+      "grad_norm": 0.8592564463615417,
+      "learning_rate": 2.9235918744228997e-05,
+      "loss": 0.0861,
+      "step": 1933
+    },
+    {
+      "epoch": 0.8660509375874615,
+      "grad_norm": 1.107157588005066,
+      "learning_rate": 2.922437673130194e-05,
+      "loss": 0.135,
+      "step": 1934
+    },
+    {
+      "epoch": 0.8664987405541562,
+      "grad_norm": 1.6733407974243164,
+      "learning_rate": 2.9212834718374886e-05,
+      "loss": 0.1944,
+      "step": 1935
+    },
+    {
+      "epoch": 0.8669465435208509,
+      "grad_norm": 1.3344576358795166,
+      "learning_rate": 2.9201292705447832e-05,
+      "loss": 0.1332,
+      "step": 1936
+    },
+    {
+      "epoch": 0.8673943464875454,
+      "grad_norm": 1.0710251331329346,
+      "learning_rate": 2.9189750692520778e-05,
+      "loss": 0.2112,
+      "step": 1937
+    },
+    {
+      "epoch": 0.8678421494542401,
+      "grad_norm": 1.2376161813735962,
+      "learning_rate": 2.917820867959372e-05,
+      "loss": 0.1467,
+      "step": 1938
+    },
+    {
+      "epoch": 0.8682899524209348,
+      "grad_norm": 1.2241551876068115,
+      "learning_rate": 2.916666666666667e-05,
+      "loss": 0.1139,
+      "step": 1939
+    },
+    {
+      "epoch": 0.8687377553876294,
+      "grad_norm": 0.8482950329780579,
+      "learning_rate": 2.9155124653739612e-05,
+      "loss": 0.0913,
+      "step": 1940
+    },
+    {
+      "epoch": 0.8687377553876294,
+      "eval_loss": 0.1484833061695099,
+      "eval_runtime": 1744.4985,
+      "eval_samples_per_second": 2.561,
+      "eval_steps_per_second": 2.561,
+      "step": 1940
+    },
+    {
+      "epoch": 0.8691855583543241,
+      "grad_norm": 0.715686559677124,
+      "learning_rate": 2.9143582640812562e-05,
+      "loss": 0.0635,
+      "step": 1941
+    },
+    {
+      "epoch": 0.8696333613210188,
+      "grad_norm": 1.4325870275497437,
+      "learning_rate": 2.9132040627885504e-05,
+      "loss": 0.2724,
+      "step": 1942
+    },
+    {
+      "epoch": 0.8700811642877134,
+      "grad_norm": 1.6713802814483643,
+      "learning_rate": 2.9120498614958454e-05,
+      "loss": 0.2227,
+      "step": 1943
+    },
+    {
+      "epoch": 0.8705289672544081,
+      "grad_norm": 1.6410531997680664,
+      "learning_rate": 2.9108956602031396e-05,
+      "loss": 0.3713,
+      "step": 1944
+    },
+    {
+      "epoch": 0.8709767702211028,
+      "grad_norm": 1.0280667543411255,
+      "learning_rate": 2.9097414589104342e-05,
+      "loss": 0.1245,
+      "step": 1945
+    },
+    {
+      "epoch": 0.8714245731877974,
+      "grad_norm": 1.0153021812438965,
+      "learning_rate": 2.9085872576177285e-05,
+      "loss": 0.1004,
+      "step": 1946
+    },
+    {
+      "epoch": 0.871872376154492,
+      "grad_norm": 0.8973650336265564,
+      "learning_rate": 2.9074330563250234e-05,
+      "loss": 0.1089,
+      "step": 1947
+    },
+    {
+      "epoch": 0.8723201791211866,
+      "grad_norm": 0.9956257939338684,
+      "learning_rate": 2.9062788550323177e-05,
+      "loss": 0.0945,
+      "step": 1948
+    },
+    {
+      "epoch": 0.8727679820878813,
+      "grad_norm": 0.9029696583747864,
+      "learning_rate": 2.905124653739612e-05,
+      "loss": 0.1923,
+      "step": 1949
+    },
+    {
+      "epoch": 0.873215785054576,
+      "grad_norm": 1.4042998552322388,
+      "learning_rate": 2.903970452446907e-05,
+      "loss": 0.0725,
+      "step": 1950
+    },
+    {
+      "epoch": 0.873215785054576,
+      "eval_loss": 0.14708450436592102,
+      "eval_runtime": 1740.0099,
+      "eval_samples_per_second": 2.567,
+      "eval_steps_per_second": 2.567,
+      "step": 1950
+    },
+    {
+      "epoch": 0.8736635880212706,
+      "grad_norm": 1.163150668144226,
+      "learning_rate": 2.902816251154201e-05,
+      "loss": 0.0904,
+      "step": 1951
+    },
+    {
+      "epoch": 0.8741113909879653,
+      "grad_norm": 0.999988317489624,
+      "learning_rate": 2.901662049861496e-05,
+      "loss": 0.1097,
+      "step": 1952
+    },
+    {
+      "epoch": 0.87455919395466,
+      "grad_norm": 1.109708547592163,
+      "learning_rate": 2.9005078485687904e-05,
+      "loss": 0.0968,
+      "step": 1953
+    },
+    {
+      "epoch": 0.8750069969213546,
+      "grad_norm": 1.0820552110671997,
+      "learning_rate": 2.8993536472760853e-05,
+      "loss": 0.1765,
+      "step": 1954
+    },
+    {
+      "epoch": 0.8754547998880493,
+      "grad_norm": 1.5995961427688599,
+      "learning_rate": 2.8981994459833796e-05,
+      "loss": 0.1388,
+      "step": 1955
+    },
+    {
+      "epoch": 0.8759026028547439,
+      "grad_norm": 1.3009573221206665,
+      "learning_rate": 2.897045244690674e-05,
+      "loss": 0.132,
+      "step": 1956
+    },
+    {
+      "epoch": 0.8763504058214385,
+      "grad_norm": 1.079586386680603,
+      "learning_rate": 2.8958910433979684e-05,
+      "loss": 0.1795,
+      "step": 1957
+    },
+    {
+      "epoch": 0.8767982087881332,
+      "grad_norm": 1.421583890914917,
+      "learning_rate": 2.8947368421052634e-05,
+      "loss": 0.1738,
+      "step": 1958
+    },
+    {
+      "epoch": 0.8772460117548279,
+      "grad_norm": 1.9353923797607422,
+      "learning_rate": 2.8935826408125576e-05,
+      "loss": 0.5605,
+      "step": 1959
+    },
+    {
+      "epoch": 0.8776938147215225,
+      "grad_norm": 1.5885474681854248,
+      "learning_rate": 2.8924284395198525e-05,
+      "loss": 0.2418,
+      "step": 1960
+    },
+    {
+      "epoch": 0.8776938147215225,
+      "eval_loss": 0.14678245782852173,
+      "eval_runtime": 1739.877,
+      "eval_samples_per_second": 2.567,
+      "eval_steps_per_second": 2.567,
+      "step": 1960
+    },
+    {
+      "epoch": 0.8781416176882172,
+      "grad_norm": 1.3782999515533447,
+      "learning_rate": 2.8912742382271468e-05,
+      "loss": 0.1732,
+      "step": 1961
+    },
+    {
+      "epoch": 0.8785894206549119,
+      "grad_norm": 1.1999086141586304,
+      "learning_rate": 2.8901200369344417e-05,
+      "loss": 0.1317,
+      "step": 1962
+    },
+    {
+      "epoch": 0.8790372236216065,
+      "grad_norm": 1.050298810005188,
+      "learning_rate": 2.888965835641736e-05,
+      "loss": 0.1355,
+      "step": 1963
+    },
+    {
+      "epoch": 0.8794850265883012,
+      "grad_norm": 1.6739578247070312,
+      "learning_rate": 2.887811634349031e-05,
+      "loss": 0.2034,
+      "step": 1964
+    },
+    {
+      "epoch": 0.8799328295549959,
+      "grad_norm": 1.3336613178253174,
+      "learning_rate": 2.8866574330563252e-05,
+      "loss": 0.1467,
+      "step": 1965
+    },
+    {
+      "epoch": 0.8803806325216904,
+      "grad_norm": 1.1149473190307617,
+      "learning_rate": 2.8855032317636198e-05,
+      "loss": 0.1364,
+      "step": 1966
+    },
+    {
+      "epoch": 0.8808284354883851,
+      "grad_norm": 0.8522685766220093,
+      "learning_rate": 2.884349030470914e-05,
+      "loss": 0.095,
+      "step": 1967
+    },
+    {
+      "epoch": 0.8812762384550797,
+      "grad_norm": 1.3482296466827393,
+      "learning_rate": 2.883194829178209e-05,
+      "loss": 0.2734,
+      "step": 1968
+    },
+    {
+      "epoch": 0.8817240414217744,
+      "grad_norm": 0.894426703453064,
+      "learning_rate": 2.8820406278855033e-05,
+      "loss": 0.0841,
+      "step": 1969
+    },
+    {
+      "epoch": 0.8821718443884691,
+      "grad_norm": 1.0606603622436523,
+      "learning_rate": 2.8808864265927982e-05,
+      "loss": 0.1532,
+      "step": 1970
+    },
+    {
+      "epoch": 0.8821718443884691,
+      "eval_loss": 0.14509204030036926,
+      "eval_runtime": 1739.6153,
+      "eval_samples_per_second": 2.568,
+      "eval_steps_per_second": 2.568,
+      "step": 1970
+    },
+    {
+      "epoch": 0.8826196473551637,
+      "grad_norm": 1.30607271194458,
+      "learning_rate": 2.8797322253000925e-05,
+      "loss": 0.212,
+      "step": 1971
+    },
+    {
+      "epoch": 0.8830674503218584,
+      "grad_norm": 0.9961697459220886,
+      "learning_rate": 2.8785780240073874e-05,
+      "loss": 0.143,
+      "step": 1972
+    },
+    {
+      "epoch": 0.8835152532885531,
+      "grad_norm": 0.9268302321434021,
+      "learning_rate": 2.8774238227146817e-05,
+      "loss": 0.1197,
+      "step": 1973
+    },
+    {
+      "epoch": 0.8839630562552477,
+      "grad_norm": 1.2642148733139038,
+      "learning_rate": 2.876269621421976e-05,
+      "loss": 0.1186,
+      "step": 1974
+    },
+    {
+      "epoch": 0.8844108592219423,
+      "grad_norm": 1.2498232126235962,
+      "learning_rate": 2.875115420129271e-05,
+      "loss": 0.095,
+      "step": 1975
+    },
+    {
+      "epoch": 0.884858662188637,
+      "grad_norm": 1.8330223560333252,
+      "learning_rate": 2.873961218836565e-05,
+      "loss": 0.2395,
+      "step": 1976
+    },
+    {
+      "epoch": 0.8853064651553316,
+      "grad_norm": 1.2042216062545776,
+      "learning_rate": 2.8728070175438597e-05,
+      "loss": 0.1489,
+      "step": 1977
+    },
+    {
+      "epoch": 0.8857542681220263,
+      "grad_norm": 0.9576921463012695,
+      "learning_rate": 2.871652816251154e-05,
+      "loss": 0.0874,
+      "step": 1978
+    },
+    {
+      "epoch": 0.886202071088721,
+      "grad_norm": 1.0180977582931519,
+      "learning_rate": 2.870498614958449e-05,
+      "loss": 0.1121,
+      "step": 1979
+    },
+    {
+      "epoch": 0.8866498740554156,
+      "grad_norm": 1.520078182220459,
+      "learning_rate": 2.8693444136657432e-05,
+      "loss": 0.2487,
+      "step": 1980
+    },
+    {
+      "epoch": 0.8866498740554156,
+      "eval_loss": 0.14374347031116486,
+      "eval_runtime": 1739.5902,
+      "eval_samples_per_second": 2.568,
+      "eval_steps_per_second": 2.568,
+      "step": 1980
+    },
+    {
+      "epoch": 0.8870976770221103,
+      "grad_norm": 0.7219629883766174,
+      "learning_rate": 2.868190212373038e-05,
+      "loss": 0.0582,
+      "step": 1981
+    },
+    {
+      "epoch": 0.887545479988805,
+      "grad_norm": 0.9952398538589478,
+      "learning_rate": 2.8670360110803324e-05,
+      "loss": 0.1541,
+      "step": 1982
+    },
+    {
+      "epoch": 0.8879932829554996,
+      "grad_norm": 1.2556064128875732,
+      "learning_rate": 2.8658818097876273e-05,
+      "loss": 0.1759,
+      "step": 1983
+    },
+    {
+      "epoch": 0.8884410859221943,
+      "grad_norm": 1.1847312450408936,
+      "learning_rate": 2.8647276084949216e-05,
+      "loss": 0.1542,
+      "step": 1984
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.9823407530784607,
+      "learning_rate": 2.8635734072022162e-05,
+      "loss": 0.107,
+      "step": 1985
+    },
+    {
+      "epoch": 0.8893366918555835,
+      "grad_norm": 1.1333894729614258,
+      "learning_rate": 2.8624192059095108e-05,
+      "loss": 0.2441,
+      "step": 1986
+    },
+    {
+      "epoch": 0.8897844948222782,
+      "grad_norm": 1.157876968383789,
+      "learning_rate": 2.8612650046168054e-05,
+      "loss": 0.163,
+      "step": 1987
+    },
+    {
+      "epoch": 0.8902322977889728,
+      "grad_norm": 0.8168591260910034,
+      "learning_rate": 2.8601108033240996e-05,
+      "loss": 0.07,
+      "step": 1988
+    },
+    {
+      "epoch": 0.8906801007556675,
+      "grad_norm": 0.9307788014411926,
+      "learning_rate": 2.8589566020313946e-05,
+      "loss": 0.1237,
+      "step": 1989
+    },
+    {
+      "epoch": 0.8911279037223622,
+      "grad_norm": 1.2166438102722168,
+      "learning_rate": 2.857802400738689e-05,
+      "loss": 0.1151,
+      "step": 1990
+    },
+    {
+      "epoch": 0.8911279037223622,
+      "eval_loss": 0.14140208065509796,
+      "eval_runtime": 1739.3049,
+      "eval_samples_per_second": 2.568,
+      "eval_steps_per_second": 2.568,
+      "step": 1990
+    },
+    {
+      "epoch": 0.8915757066890568,
+      "grad_norm": 1.0516357421875,
+      "learning_rate": 2.8566481994459838e-05,
+      "loss": 0.1531,
+      "step": 1991
+    },
+    {
+      "epoch": 0.8920235096557515,
+      "grad_norm": 1.1389539241790771,
+      "learning_rate": 2.855493998153278e-05,
+      "loss": 0.0984,
+      "step": 1992
+    },
+    {
+      "epoch": 0.8924713126224462,
+      "grad_norm": 1.6578068733215332,
+      "learning_rate": 2.854339796860573e-05,
+      "loss": 0.1922,
+      "step": 1993
+    },
+    {
+      "epoch": 0.8929191155891408,
+      "grad_norm": 0.9576694965362549,
+      "learning_rate": 2.8531855955678672e-05,
+      "loss": 0.0826,
+      "step": 1994
+    },
+    {
+      "epoch": 0.8933669185558354,
+      "grad_norm": 1.0604113340377808,
+      "learning_rate": 2.852031394275162e-05,
+      "loss": 0.0933,
+      "step": 1995
+    },
+    {
+      "epoch": 0.89381472152253,
+      "grad_norm": 0.9682891368865967,
+      "learning_rate": 2.850877192982456e-05,
+      "loss": 0.1203,
+      "step": 1996
+    },
+    {
+      "epoch": 0.8942625244892247,
+      "grad_norm": 1.1396582126617432,
+      "learning_rate": 2.849722991689751e-05,
+      "loss": 0.0986,
+      "step": 1997
+    },
+    {
+      "epoch": 0.8947103274559194,
+      "grad_norm": 0.9337264895439148,
+      "learning_rate": 2.8485687903970453e-05,
+      "loss": 0.1658,
+      "step": 1998
+    },
+    {
+      "epoch": 0.895158130422614,
+      "grad_norm": 1.4096896648406982,
+      "learning_rate": 2.8474145891043396e-05,
+      "loss": 0.1466,
+      "step": 1999
+    },
+    {
+      "epoch": 0.8956059333893087,
+      "grad_norm": 1.0767900943756104,
+      "learning_rate": 2.8462603878116345e-05,
+      "loss": 0.1029,
+      "step": 2000
+    },
+    {
+      "epoch": 0.8956059333893087,
+      "eval_loss": 0.14072825014591217,
+      "eval_runtime": 1738.5687,
+      "eval_samples_per_second": 2.569,
+      "eval_steps_per_second": 2.569,
+      "step": 2000
+    },
+    {
+      "epoch": 0.8960537363560034,
+      "grad_norm": 1.3448169231414795,
+      "learning_rate": 2.8451061865189288e-05,
+      "loss": 0.1451,
+      "step": 2001
+    },
+    {
+      "epoch": 0.896501539322698,
+      "grad_norm": 1.0144966840744019,
+      "learning_rate": 2.8439519852262237e-05,
+      "loss": 0.0834,
+      "step": 2002
+    },
+    {
+      "epoch": 0.8969493422893927,
+      "grad_norm": 1.3243262767791748,
+      "learning_rate": 2.842797783933518e-05,
+      "loss": 0.2478,
+      "step": 2003
+    },
+    {
+      "epoch": 0.8973971452560873,
+      "grad_norm": 1.26291024684906,
+      "learning_rate": 2.841643582640813e-05,
+      "loss": 0.1247,
+      "step": 2004
+    },
+    {
+      "epoch": 0.8978449482227819,
+      "grad_norm": 1.1598271131515503,
+      "learning_rate": 2.840489381348107e-05,
+      "loss": 0.1975,
+      "step": 2005
+    },
+    {
+      "epoch": 0.8982927511894766,
+      "grad_norm": 0.98848557472229,
+      "learning_rate": 2.8393351800554018e-05,
+      "loss": 0.0948,
+      "step": 2006
+    },
+    {
+      "epoch": 0.8987405541561713,
+      "grad_norm": 1.0264403820037842,
+      "learning_rate": 2.838180978762696e-05,
+      "loss": 0.0812,
+      "step": 2007
+    },
+    {
+      "epoch": 0.8991883571228659,
+      "grad_norm": 0.7366992831230164,
+      "learning_rate": 2.837026777469991e-05,
+      "loss": 0.0588,
+      "step": 2008
+    },
+    {
+      "epoch": 0.8996361600895606,
+      "grad_norm": 0.8434129953384399,
+      "learning_rate": 2.8358725761772852e-05,
+      "loss": 0.0947,
+      "step": 2009
+    },
+    {
+      "epoch": 0.9000839630562553,
+      "grad_norm": 0.9951444268226624,
+      "learning_rate": 2.83471837488458e-05,
+      "loss": 0.09,
+      "step": 2010
+    },
+    {
+      "epoch": 0.9000839630562553,
+      "eval_loss": 0.13967619836330414,
+      "eval_runtime": 1736.9399,
+      "eval_samples_per_second": 2.572,
+      "eval_steps_per_second": 2.572,
+      "step": 2010
+    },
+    {
+      "epoch": 0.9005317660229499,
+      "grad_norm": 1.09131920337677,
+      "learning_rate": 2.8335641735918744e-05,
+      "loss": 0.102,
+      "step": 2011
+    },
+    {
+      "epoch": 0.9009795689896446,
+      "grad_norm": 1.2195484638214111,
+      "learning_rate": 2.8324099722991694e-05,
+      "loss": 0.1977,
+      "step": 2012
+    },
+    {
+      "epoch": 0.9014273719563393,
+      "grad_norm": 1.3848285675048828,
+      "learning_rate": 2.8312557710064636e-05,
+      "loss": 0.2635,
+      "step": 2013
+    },
+    {
+      "epoch": 0.9018751749230338,
+      "grad_norm": 0.9678056240081787,
+      "learning_rate": 2.8301015697137586e-05,
+      "loss": 0.0769,
+      "step": 2014
+    },
+    {
+      "epoch": 0.9023229778897285,
+      "grad_norm": 1.0768543481826782,
+      "learning_rate": 2.8289473684210528e-05,
+      "loss": 0.0976,
+      "step": 2015
+    },
+    {
+      "epoch": 0.9027707808564231,
+      "grad_norm": 0.7608491778373718,
+      "learning_rate": 2.8277931671283474e-05,
+      "loss": 0.0695,
+      "step": 2016
+    },
+    {
+      "epoch": 0.9032185838231178,
+      "grad_norm": 1.0845251083374023,
+      "learning_rate": 2.8266389658356417e-05,
+      "loss": 0.084,
+      "step": 2017
+    },
+    {
+      "epoch": 0.9036663867898125,
+      "grad_norm": 1.199765920639038,
+      "learning_rate": 2.8254847645429366e-05,
+      "loss": 0.1161,
+      "step": 2018
+    },
+    {
+      "epoch": 0.9041141897565071,
+      "grad_norm": 1.084950566291809,
+      "learning_rate": 2.824330563250231e-05,
+      "loss": 0.1145,
+      "step": 2019
+    },
+    {
+      "epoch": 0.9045619927232018,
+      "grad_norm": 0.9587868452072144,
+      "learning_rate": 2.8231763619575258e-05,
+      "loss": 0.1391,
+      "step": 2020
+    },
+    {
+      "epoch": 0.9045619927232018,
+      "eval_loss": 0.13916510343551636,
+      "eval_runtime": 1737.3861,
+      "eval_samples_per_second": 2.571,
+      "eval_steps_per_second": 2.571,
+      "step": 2020
+    },
+    {
+      "epoch": 0.9050097956898965,
+      "grad_norm": 1.1295528411865234,
+      "learning_rate": 2.82202216066482e-05,
+      "loss": 0.1125,
+      "step": 2021
+    },
+    {
+      "epoch": 0.9054575986565911,
+      "grad_norm": 1.32404625415802,
+      "learning_rate": 2.820867959372115e-05,
+      "loss": 0.1223,
+      "step": 2022
+    },
+    {
+      "epoch": 0.9059054016232857,
+      "grad_norm": 1.3158620595932007,
+      "learning_rate": 2.8197137580794093e-05,
+      "loss": 0.2674,
+      "step": 2023
+    },
+    {
+      "epoch": 0.9063532045899804,
+      "grad_norm": 1.20033860206604,
+      "learning_rate": 2.8185595567867035e-05,
+      "loss": 0.1966,
+      "step": 2024
+    },
+    {
+      "epoch": 0.906801007556675,
+      "grad_norm": 0.9327771663665771,
+      "learning_rate": 2.8174053554939985e-05,
+      "loss": 0.117,
+      "step": 2025
+    },
+    {
+      "epoch": 0.9072488105233697,
+      "grad_norm": 1.43967866897583,
+      "learning_rate": 2.8162511542012927e-05,
+      "loss": 0.1195,
+      "step": 2026
+    },
+    {
+      "epoch": 0.9076966134900644,
+      "grad_norm": 1.1177648305892944,
+      "learning_rate": 2.8150969529085873e-05,
+      "loss": 0.1073,
+      "step": 2027
+    },
+    {
+      "epoch": 0.908144416456759,
+      "grad_norm": 0.8828498125076294,
+      "learning_rate": 2.8139427516158816e-05,
+      "loss": 0.1223,
+      "step": 2028
+    },
+    {
+      "epoch": 0.9085922194234537,
+      "grad_norm": 1.082963466644287,
+      "learning_rate": 2.8127885503231765e-05,
+      "loss": 0.166,
+      "step": 2029
+    },
+    {
+      "epoch": 0.9090400223901484,
+      "grad_norm": 0.9051773548126221,
+      "learning_rate": 2.8116343490304708e-05,
+      "loss": 0.0679,
+      "step": 2030
+    },
+    {
+      "epoch": 0.9090400223901484,
+      "eval_loss": 0.13642385601997375,
+      "eval_runtime": 1736.3259,
+      "eval_samples_per_second": 2.573,
+      "eval_steps_per_second": 2.573,
+      "step": 2030
+    },
+    {
+      "epoch": 0.909487825356843,
+      "grad_norm": 1.1784230470657349,
+      "learning_rate": 2.8104801477377657e-05,
+      "loss": 0.1645,
+      "step": 2031
+    },
+    {
+      "epoch": 0.9099356283235377,
+      "grad_norm": 1.8553470373153687,
+      "learning_rate": 2.80932594644506e-05,
+      "loss": 0.1982,
+      "step": 2032
+    },
+    {
+      "epoch": 0.9103834312902322,
+      "grad_norm": 0.9508320093154907,
+      "learning_rate": 2.808171745152355e-05,
+      "loss": 0.0892,
+      "step": 2033
+    },
+    {
+      "epoch": 0.9108312342569269,
+      "grad_norm": 1.0444815158843994,
+      "learning_rate": 2.8070175438596492e-05,
+      "loss": 0.1011,
+      "step": 2034
+    },
+    {
+      "epoch": 0.9112790372236216,
+      "grad_norm": 1.3620988130569458,
+      "learning_rate": 2.8058633425669438e-05,
+      "loss": 0.2077,
+      "step": 2035
+    },
+    {
+      "epoch": 0.9117268401903162,
+      "grad_norm": 1.7878204584121704,
+      "learning_rate": 2.8047091412742384e-05,
+      "loss": 0.1701,
+      "step": 2036
+    },
+    {
+      "epoch": 0.9121746431570109,
+      "grad_norm": 0.9353978037834167,
+      "learning_rate": 2.803554939981533e-05,
+      "loss": 0.0899,
+      "step": 2037
+    },
+    {
+      "epoch": 0.9126224461237056,
+      "grad_norm": 1.4356045722961426,
+      "learning_rate": 2.8024007386888272e-05,
+      "loss": 0.239,
+      "step": 2038
+    },
+    {
+      "epoch": 0.9130702490904002,
+      "grad_norm": 1.3796677589416504,
+      "learning_rate": 2.8012465373961222e-05,
+      "loss": 0.1811,
+      "step": 2039
+    },
+    {
+      "epoch": 0.9135180520570949,
+      "grad_norm": 1.1362677812576294,
+      "learning_rate": 2.8000923361034164e-05,
+      "loss": 0.1263,
+      "step": 2040
+    },
+    {
+      "epoch": 0.9135180520570949,
+      "eval_loss": 0.13533499836921692,
+      "eval_runtime": 1732.9341,
+      "eval_samples_per_second": 2.578,
+      "eval_steps_per_second": 2.578,
+      "step": 2040
+    },
+    {
+      "epoch": 0.9139658550237896,
+      "grad_norm": 1.3013898134231567,
+      "learning_rate": 2.7989381348107114e-05,
+      "loss": 0.1018,
+      "step": 2041
+    },
+    {
+      "epoch": 0.9144136579904842,
+      "grad_norm": 1.7660280466079712,
+      "learning_rate": 2.7977839335180056e-05,
+      "loss": 0.2326,
+      "step": 2042
+    },
+    {
+      "epoch": 0.9148614609571788,
+      "grad_norm": 1.3348053693771362,
+      "learning_rate": 2.7966297322253006e-05,
+      "loss": 0.2302,
+      "step": 2043
+    },
+    {
+      "epoch": 0.9153092639238735,
+      "grad_norm": 1.8742432594299316,
+      "learning_rate": 2.795475530932595e-05,
+      "loss": 0.1871,
+      "step": 2044
+    },
+    {
+      "epoch": 0.9157570668905681,
+      "grad_norm": 0.6203439831733704,
+      "learning_rate": 2.7943213296398894e-05,
+      "loss": 0.0832,
+      "step": 2045
+    },
+    {
+      "epoch": 0.9162048698572628,
+      "grad_norm": 1.1034528017044067,
+      "learning_rate": 2.7931671283471837e-05,
+      "loss": 0.1393,
+      "step": 2046
+    },
+    {
+      "epoch": 0.9166526728239575,
+      "grad_norm": 1.2916886806488037,
+      "learning_rate": 2.7920129270544786e-05,
+      "loss": 0.1032,
+      "step": 2047
+    },
+    {
+      "epoch": 0.9171004757906521,
+      "grad_norm": 1.0839390754699707,
+      "learning_rate": 2.790858725761773e-05,
+      "loss": 0.1538,
+      "step": 2048
+    },
+    {
+      "epoch": 0.9175482787573468,
+      "grad_norm": 1.3720133304595947,
+      "learning_rate": 2.789704524469067e-05,
+      "loss": 0.206,
+      "step": 2049
+    },
+    {
+      "epoch": 0.9179960817240415,
+      "grad_norm": 1.0583562850952148,
+      "learning_rate": 2.788550323176362e-05,
+      "loss": 0.0734,
+      "step": 2050
+    },
+    {
+      "epoch": 0.9179960817240415,
+      "eval_loss": 0.13328154385089874,
+      "eval_runtime": 1735.7016,
+      "eval_samples_per_second": 2.574,
+      "eval_steps_per_second": 2.574,
+      "step": 2050
+    },
+    {
+      "epoch": 0.9184438846907361,
+      "grad_norm": 1.2125797271728516,
+      "learning_rate": 2.7873961218836564e-05,
+      "loss": 0.0837,
+      "step": 2051
+    },
+    {
+      "epoch": 0.9188916876574307,
+      "grad_norm": 1.289035439491272,
+      "learning_rate": 2.7862419205909513e-05,
+      "loss": 0.1494,
+      "step": 2052
+    },
+    {
+      "epoch": 0.9193394906241253,
+      "grad_norm": 1.5854977369308472,
+      "learning_rate": 2.7850877192982456e-05,
+      "loss": 0.2638,
+      "step": 2053
+    },
+    {
+      "epoch": 0.91978729359082,
+      "grad_norm": 1.4358898401260376,
+      "learning_rate": 2.7839335180055405e-05,
+      "loss": 0.2973,
+      "step": 2054
+    },
+    {
+      "epoch": 0.9202350965575147,
+      "grad_norm": 0.9232112169265747,
+      "learning_rate": 2.7827793167128348e-05,
+      "loss": 0.0715,
+      "step": 2055
+    },
+    {
+      "epoch": 0.9206828995242093,
+      "grad_norm": 1.1108778715133667,
+      "learning_rate": 2.7816251154201294e-05,
+      "loss": 0.1234,
+      "step": 2056
+    },
+    {
+      "epoch": 0.921130702490904,
+      "grad_norm": 1.1772187948226929,
+      "learning_rate": 2.7804709141274236e-05,
+      "loss": 0.1565,
+      "step": 2057
+    },
+    {
+      "epoch": 0.9215785054575987,
+      "grad_norm": 1.3795710802078247,
+      "learning_rate": 2.7793167128347186e-05,
+      "loss": 0.161,
+      "step": 2058
+    },
+    {
+      "epoch": 0.9220263084242933,
+      "grad_norm": 1.0228571891784668,
+      "learning_rate": 2.7781625115420128e-05,
+      "loss": 0.1555,
+      "step": 2059
+    },
+    {
+      "epoch": 0.922474111390988,
+      "grad_norm": 0.939876139163971,
+      "learning_rate": 2.7770083102493078e-05,
+      "loss": 0.1374,
+      "step": 2060
+    },
+    {
+      "epoch": 0.922474111390988,
+      "eval_loss": 0.13193786144256592,
+      "eval_runtime": 1733.2853,
+      "eval_samples_per_second": 2.577,
+      "eval_steps_per_second": 2.577,
+      "step": 2060
+    },
+    {
+      "epoch": 0.9229219143576827,
+      "grad_norm": 1.0626474618911743,
+      "learning_rate": 2.775854108956602e-05,
+      "loss": 0.0856,
+      "step": 2061
+    },
+    {
+      "epoch": 0.9233697173243772,
+      "grad_norm": 1.2270374298095703,
+      "learning_rate": 2.774699907663897e-05,
+      "loss": 0.1045,
+      "step": 2062
+    },
+    {
+      "epoch": 0.9238175202910719,
+      "grad_norm": 1.1409982442855835,
+      "learning_rate": 2.7735457063711912e-05,
+      "loss": 0.161,
+      "step": 2063
+    },
+    {
+      "epoch": 0.9242653232577666,
+      "grad_norm": 1.3213545083999634,
+      "learning_rate": 2.772391505078486e-05,
+      "loss": 0.2275,
+      "step": 2064
+    },
+    {
+      "epoch": 0.9247131262244612,
+      "grad_norm": 0.8315839767456055,
+      "learning_rate": 2.7712373037857804e-05,
+      "loss": 0.1097,
+      "step": 2065
+    },
+    {
+      "epoch": 0.9251609291911559,
+      "grad_norm": 1.038310170173645,
+      "learning_rate": 2.770083102493075e-05,
+      "loss": 0.1428,
+      "step": 2066
+    },
+    {
+      "epoch": 0.9256087321578506,
+      "grad_norm": 1.3389686346054077,
+      "learning_rate": 2.7689289012003693e-05,
+      "loss": 0.1982,
+      "step": 2067
+    },
+    {
+      "epoch": 0.9260565351245452,
+      "grad_norm": 1.168298363685608,
+      "learning_rate": 2.7677746999076642e-05,
+      "loss": 0.1804,
+      "step": 2068
+    },
+    {
+      "epoch": 0.9265043380912399,
+      "grad_norm": 0.9898531436920166,
+      "learning_rate": 2.7666204986149585e-05,
+      "loss": 0.0915,
+      "step": 2069
+    },
+    {
+      "epoch": 0.9269521410579346,
+      "grad_norm": 0.9529014825820923,
+      "learning_rate": 2.7654662973222534e-05,
+      "loss": 0.0745,
+      "step": 2070
+    },
+    {
+      "epoch": 0.9269521410579346,
+      "eval_loss": 0.12939994037151337,
+      "eval_runtime": 1728.0102,
+      "eval_samples_per_second": 2.585,
+      "eval_steps_per_second": 2.585,
+      "step": 2070
+    },
+    {
+      "epoch": 0.9273999440246291,
+      "grad_norm": 1.0990973711013794,
+      "learning_rate": 2.7643120960295477e-05,
+      "loss": 0.1563,
+      "step": 2071
+    },
+    {
+      "epoch": 0.9278477469913238,
+      "grad_norm": 1.4120893478393555,
+      "learning_rate": 2.7631578947368426e-05,
+      "loss": 0.1955,
+      "step": 2072
+    },
+    {
+      "epoch": 0.9282955499580184,
+      "grad_norm": 1.8996490240097046,
+      "learning_rate": 2.762003693444137e-05,
+      "loss": 0.1389,
+      "step": 2073
+    },
+    {
+      "epoch": 0.9287433529247131,
+      "grad_norm": 1.2464399337768555,
+      "learning_rate": 2.760849492151431e-05,
+      "loss": 0.12,
+      "step": 2074
+    },
+    {
+      "epoch": 0.9291911558914078,
+      "grad_norm": 1.1876400709152222,
+      "learning_rate": 2.759695290858726e-05,
+      "loss": 0.0893,
+      "step": 2075
+    },
+    {
+      "epoch": 0.9296389588581024,
+      "grad_norm": 1.5894993543624878,
+      "learning_rate": 2.7585410895660203e-05,
+      "loss": 0.1151,
+      "step": 2076
+    },
+    {
+      "epoch": 0.9300867618247971,
+      "grad_norm": 1.2473312616348267,
+      "learning_rate": 2.757386888273315e-05,
+      "loss": 0.0991,
+      "step": 2077
+    },
+    {
+      "epoch": 0.9305345647914918,
+      "grad_norm": 0.9199130535125732,
+      "learning_rate": 2.7562326869806092e-05,
+      "loss": 0.0805,
+      "step": 2078
+    },
+    {
+      "epoch": 0.9309823677581864,
+      "grad_norm": 1.4169944524765015,
+      "learning_rate": 2.755078485687904e-05,
+      "loss": 0.1683,
+      "step": 2079
+    },
+    {
+      "epoch": 0.9314301707248811,
+      "grad_norm": 1.1549464464187622,
+      "learning_rate": 2.7539242843951984e-05,
+      "loss": 0.1183,
+      "step": 2080
+    },
+    {
+      "epoch": 0.9314301707248811,
+      "eval_loss": 0.12832778692245483,
+      "eval_runtime": 1728.5516,
+      "eval_samples_per_second": 2.584,
+      "eval_steps_per_second": 2.584,
+      "step": 2080
+    },
+    {
+      "epoch": 0.9318779736915757,
+      "grad_norm": 1.8567004203796387,
+      "learning_rate": 2.7527700831024933e-05,
+      "loss": 0.2219,
+      "step": 2081
+    },
+    {
+      "epoch": 0.9323257766582703,
+      "grad_norm": 0.8522270321846008,
+      "learning_rate": 2.7516158818097876e-05,
+      "loss": 0.0656,
+      "step": 2082
+    },
+    {
+      "epoch": 0.932773579624965,
+      "grad_norm": 0.920242190361023,
+      "learning_rate": 2.7504616805170825e-05,
+      "loss": 0.0794,
+      "step": 2083
+    },
+    {
+      "epoch": 0.9332213825916597,
+      "grad_norm": 1.5621031522750854,
+      "learning_rate": 2.7493074792243768e-05,
+      "loss": 0.1228,
+      "step": 2084
+    },
+    {
+      "epoch": 0.9336691855583543,
+      "grad_norm": 1.3902122974395752,
+      "learning_rate": 2.7481532779316714e-05,
+      "loss": 0.1225,
+      "step": 2085
+    },
+    {
+      "epoch": 0.934116988525049,
+      "grad_norm": 0.8981521725654602,
+      "learning_rate": 2.746999076638966e-05,
+      "loss": 0.1141,
+      "step": 2086
+    },
+    {
+      "epoch": 0.9345647914917437,
+      "grad_norm": 1.0861537456512451,
+      "learning_rate": 2.7458448753462606e-05,
+      "loss": 0.0923,
+      "step": 2087
+    },
+    {
+      "epoch": 0.9350125944584383,
+      "grad_norm": 0.641179084777832,
+      "learning_rate": 2.744690674053555e-05,
+      "loss": 0.0678,
+      "step": 2088
+    },
+    {
+      "epoch": 0.935460397425133,
+      "grad_norm": 1.2074898481369019,
+      "learning_rate": 2.7435364727608498e-05,
+      "loss": 0.1375,
+      "step": 2089
+    },
+    {
+      "epoch": 0.9359082003918276,
+      "grad_norm": 0.8622816801071167,
+      "learning_rate": 2.742382271468144e-05,
+      "loss": 0.1091,
+      "step": 2090
+    },
+    {
+      "epoch": 0.9359082003918276,
+      "eval_loss": 0.12745097279548645,
+      "eval_runtime": 1733.2884,
+      "eval_samples_per_second": 2.577,
+      "eval_steps_per_second": 2.577,
+      "step": 2090
+    },
+    {
+      "epoch": 0.9363560033585222,
+      "grad_norm": 1.2994965314865112,
+      "learning_rate": 2.741228070175439e-05,
+      "loss": 0.1355,
+      "step": 2091
+    },
+    {
+      "epoch": 0.9368038063252169,
+      "grad_norm": 0.9833651185035706,
+      "learning_rate": 2.7400738688827332e-05,
+      "loss": 0.0922,
+      "step": 2092
+    },
+    {
+      "epoch": 0.9372516092919115,
+      "grad_norm": 0.9704849123954773,
+      "learning_rate": 2.7389196675900282e-05,
+      "loss": 0.0962,
+      "step": 2093
+    },
+    {
+      "epoch": 0.9376994122586062,
+      "grad_norm": 0.9272742867469788,
+      "learning_rate": 2.7377654662973224e-05,
+      "loss": 0.0572,
+      "step": 2094
+    },
+    {
+      "epoch": 0.9381472152253009,
+      "grad_norm": 0.8162627816200256,
+      "learning_rate": 2.736611265004617e-05,
+      "loss": 0.0705,
+      "step": 2095
+    },
+    {
+      "epoch": 0.9385950181919955,
+      "grad_norm": 1.1083530187606812,
+      "learning_rate": 2.7354570637119113e-05,
+      "loss": 0.1832,
+      "step": 2096
+    },
+    {
+      "epoch": 0.9390428211586902,
+      "grad_norm": 0.935386061668396,
+      "learning_rate": 2.7343028624192062e-05,
+      "loss": 0.1217,
+      "step": 2097
+    },
+    {
+      "epoch": 0.9394906241253849,
+      "grad_norm": 1.1456583738327026,
+      "learning_rate": 2.7331486611265005e-05,
+      "loss": 0.0749,
+      "step": 2098
+    },
+    {
+      "epoch": 0.9399384270920795,
+      "grad_norm": 1.2562615871429443,
+      "learning_rate": 2.7319944598337948e-05,
+      "loss": 0.1562,
+      "step": 2099
+    },
+    {
+      "epoch": 0.9403862300587741,
+      "grad_norm": 1.1288965940475464,
+      "learning_rate": 2.7308402585410897e-05,
+      "loss": 0.0863,
+      "step": 2100
+    },
+    {
+      "epoch": 0.9403862300587741,
+      "eval_loss": 0.1265988051891327,
+      "eval_runtime": 1736.0228,
+      "eval_samples_per_second": 2.573,
+      "eval_steps_per_second": 2.573,
+      "step": 2100
+    },
+    {
+      "epoch": 0.9408340330254688,
+      "grad_norm": 1.0694856643676758,
+      "learning_rate": 2.729686057248384e-05,
+      "loss": 0.1046,
+      "step": 2101
+    },
+    {
+      "epoch": 0.9412818359921634,
+      "grad_norm": 0.8078300356864929,
+      "learning_rate": 2.728531855955679e-05,
+      "loss": 0.067,
+      "step": 2102
+    },
+    {
+      "epoch": 0.9417296389588581,
+      "grad_norm": 1.4082826375961304,
+      "learning_rate": 2.727377654662973e-05,
+      "loss": 0.2408,
+      "step": 2103
+    },
+    {
+      "epoch": 0.9421774419255527,
+      "grad_norm": 1.0101279020309448,
+      "learning_rate": 2.726223453370268e-05,
+      "loss": 0.1091,
+      "step": 2104
+    },
+    {
+      "epoch": 0.9426252448922474,
+      "grad_norm": 0.7255677580833435,
+      "learning_rate": 2.7250692520775624e-05,
+      "loss": 0.0572,
+      "step": 2105
+    },
+    {
+      "epoch": 0.9430730478589421,
+      "grad_norm": 1.0651483535766602,
+      "learning_rate": 2.723915050784857e-05,
+      "loss": 0.0956,
+      "step": 2106
+    },
+    {
+      "epoch": 0.9435208508256367,
+      "grad_norm": 1.2711114883422852,
+      "learning_rate": 2.7227608494921512e-05,
+      "loss": 0.1856,
+      "step": 2107
+    },
+    {
+      "epoch": 0.9439686537923314,
+      "grad_norm": 1.3673771619796753,
+      "learning_rate": 2.721606648199446e-05,
+      "loss": 0.1094,
+      "step": 2108
+    },
+    {
+      "epoch": 0.9444164567590261,
+      "grad_norm": 0.7726939916610718,
+      "learning_rate": 2.7204524469067404e-05,
+      "loss": 0.0995,
+      "step": 2109
+    },
+    {
+      "epoch": 0.9448642597257206,
+      "grad_norm": 0.930659294128418,
+      "learning_rate": 2.7192982456140354e-05,
+      "loss": 0.1202,
+      "step": 2110
+    },
+    {
+      "epoch": 0.9448642597257206,
+      "eval_loss": 0.1258155256509781,
+      "eval_runtime": 1731.0481,
+      "eval_samples_per_second": 2.581,
+      "eval_steps_per_second": 2.581,
+      "step": 2110
+    },
+    {
+      "epoch": 0.9453120626924153,
+      "grad_norm": 0.9811476469039917,
+      "learning_rate": 2.7181440443213296e-05,
+      "loss": 0.124,
+      "step": 2111
+    },
+    {
+      "epoch": 0.94575986565911,
+      "grad_norm": 1.239196538925171,
+      "learning_rate": 2.7169898430286246e-05,
+      "loss": 0.1605,
+      "step": 2112
+    },
+    {
+      "epoch": 0.9462076686258046,
+      "grad_norm": 1.453579306602478,
+      "learning_rate": 2.7158356417359188e-05,
+      "loss": 0.1452,
+      "step": 2113
+    },
+    {
+      "epoch": 0.9466554715924993,
+      "grad_norm": 0.9727533459663391,
+      "learning_rate": 2.7146814404432138e-05,
+      "loss": 0.0827,
+      "step": 2114
+    },
+    {
+      "epoch": 0.947103274559194,
+      "grad_norm": 0.8124465346336365,
+      "learning_rate": 2.713527239150508e-05,
+      "loss": 0.0863,
+      "step": 2115
+    },
+    {
+      "epoch": 0.9475510775258886,
+      "grad_norm": 0.8164709806442261,
+      "learning_rate": 2.7123730378578026e-05,
+      "loss": 0.0735,
+      "step": 2116
+    },
+    {
+      "epoch": 0.9479988804925833,
+      "grad_norm": 1.3132597208023071,
+      "learning_rate": 2.711218836565097e-05,
+      "loss": 0.1077,
+      "step": 2117
+    },
+    {
+      "epoch": 0.948446683459278,
+      "grad_norm": 1.138953685760498,
+      "learning_rate": 2.7100646352723918e-05,
+      "loss": 0.1137,
+      "step": 2118
+    },
+    {
+      "epoch": 0.9488944864259726,
+      "grad_norm": 1.3828215599060059,
+      "learning_rate": 2.708910433979686e-05,
+      "loss": 0.0948,
+      "step": 2119
+    },
+    {
+      "epoch": 0.9493422893926672,
+      "grad_norm": 0.6844764947891235,
+      "learning_rate": 2.707756232686981e-05,
+      "loss": 0.0961,
+      "step": 2120
+    },
+    {
+      "epoch": 0.9493422893926672,
+      "eval_loss": 0.1241694986820221,
+      "eval_runtime": 1734.9872,
+      "eval_samples_per_second": 2.575,
+      "eval_steps_per_second": 2.575,
+      "step": 2120
+    },
+    {
+      "epoch": 0.9497900923593618,
+      "grad_norm": 0.9134883284568787,
+      "learning_rate": 2.7066020313942753e-05,
+      "loss": 0.0868,
+      "step": 2121
+    },
+    {
+      "epoch": 0.9502378953260565,
+      "grad_norm": 1.5742448568344116,
+      "learning_rate": 2.7054478301015702e-05,
+      "loss": 0.2265,
+      "step": 2122
+    },
+    {
+      "epoch": 0.9506856982927512,
+      "grad_norm": 1.5208790302276611,
+      "learning_rate": 2.7042936288088645e-05,
+      "loss": 0.2737,
+      "step": 2123
+    },
+    {
+      "epoch": 0.9511335012594458,
+      "grad_norm": 1.644930124282837,
+      "learning_rate": 2.7031394275161587e-05,
+      "loss": 0.1309,
+      "step": 2124
+    },
+    {
+      "epoch": 0.9515813042261405,
+      "grad_norm": 0.7968838810920715,
+      "learning_rate": 2.7019852262234537e-05,
+      "loss": 0.0825,
+      "step": 2125
+    },
+    {
+      "epoch": 0.9520291071928352,
+      "grad_norm": 1.451269507408142,
+      "learning_rate": 2.700831024930748e-05,
+      "loss": 0.1362,
+      "step": 2126
+    },
+    {
+      "epoch": 0.9524769101595298,
+      "grad_norm": 1.1176960468292236,
+      "learning_rate": 2.6996768236380425e-05,
+      "loss": 0.1107,
+      "step": 2127
+    },
+    {
+      "epoch": 0.9529247131262245,
+      "grad_norm": 0.9091848731040955,
+      "learning_rate": 2.6985226223453368e-05,
+      "loss": 0.1113,
+      "step": 2128
+    },
+    {
+      "epoch": 0.9533725160929191,
+      "grad_norm": 1.2914046049118042,
+      "learning_rate": 2.6973684210526317e-05,
+      "loss": 0.1142,
+      "step": 2129
+    },
+    {
+      "epoch": 0.9538203190596137,
+      "grad_norm": 0.965366005897522,
+      "learning_rate": 2.696214219759926e-05,
+      "loss": 0.1009,
+      "step": 2130
+    },
+    {
+      "epoch": 0.9538203190596137,
+      "eval_loss": 0.1229698657989502,
+      "eval_runtime": 1728.6065,
+      "eval_samples_per_second": 2.584,
+      "eval_steps_per_second": 2.584,
+      "step": 2130
+    },
+    {
+      "epoch": 0.9542681220263084,
+      "grad_norm": 1.3068053722381592,
+      "learning_rate": 2.695060018467221e-05,
+      "loss": 0.1627,
+      "step": 2131
+    },
+    {
+      "epoch": 0.9547159249930031,
+      "grad_norm": 1.0245745182037354,
+      "learning_rate": 2.6939058171745152e-05,
+      "loss": 0.0736,
+      "step": 2132
+    },
+    {
+      "epoch": 0.9551637279596977,
+      "grad_norm": 0.8566816449165344,
+      "learning_rate": 2.69275161588181e-05,
+      "loss": 0.1103,
+      "step": 2133
+    },
+    {
+      "epoch": 0.9556115309263924,
+      "grad_norm": 0.952111005783081,
+      "learning_rate": 2.6915974145891044e-05,
+      "loss": 0.09,
+      "step": 2134
+    },
+    {
+      "epoch": 0.9560593338930871,
+      "grad_norm": 1.0342718362808228,
+      "learning_rate": 2.690443213296399e-05,
+      "loss": 0.0958,
+      "step": 2135
+    },
+    {
+      "epoch": 0.9565071368597817,
+      "grad_norm": 1.3856199979782104,
+      "learning_rate": 2.6892890120036936e-05,
+      "loss": 0.1149,
+      "step": 2136
+    },
+    {
+      "epoch": 0.9569549398264764,
+      "grad_norm": 1.2831979990005493,
+      "learning_rate": 2.6881348107109882e-05,
+      "loss": 0.1712,
+      "step": 2137
+    },
+    {
+      "epoch": 0.9574027427931711,
+      "grad_norm": 1.151147484779358,
+      "learning_rate": 2.6869806094182825e-05,
+      "loss": 0.106,
+      "step": 2138
+    },
+    {
+      "epoch": 0.9578505457598656,
+      "grad_norm": 1.2741520404815674,
+      "learning_rate": 2.6858264081255774e-05,
+      "loss": 0.1151,
+      "step": 2139
+    },
+    {
+      "epoch": 0.9582983487265603,
+      "grad_norm": 0.9963933229446411,
+      "learning_rate": 2.6846722068328717e-05,
+      "loss": 0.1024,
+      "step": 2140
+    },
+    {
+      "epoch": 0.9582983487265603,
+      "eval_loss": 0.12240637838840485,
+      "eval_runtime": 1732.9445,
+      "eval_samples_per_second": 2.578,
+      "eval_steps_per_second": 2.578,
+      "step": 2140
+    },
+    {
+      "epoch": 0.9587461516932549,
+      "grad_norm": 1.1386899948120117,
+      "learning_rate": 2.6835180055401666e-05,
+      "loss": 0.0772,
+      "step": 2141
+    },
+    {
+      "epoch": 0.9591939546599496,
+      "grad_norm": 0.8593908548355103,
+      "learning_rate": 2.682363804247461e-05,
+      "loss": 0.0883,
+      "step": 2142
+    },
+    {
+      "epoch": 0.9596417576266443,
+      "grad_norm": 1.1071313619613647,
+      "learning_rate": 2.6812096029547558e-05,
+      "loss": 0.125,
+      "step": 2143
+    },
+    {
+      "epoch": 0.9600895605933389,
+      "grad_norm": 0.6254713535308838,
+      "learning_rate": 2.68005540166205e-05,
+      "loss": 0.0546,
+      "step": 2144
+    },
+    {
+      "epoch": 0.9605373635600336,
+      "grad_norm": 0.7606832981109619,
+      "learning_rate": 2.6789012003693446e-05,
+      "loss": 0.0801,
+      "step": 2145
+    },
+    {
+      "epoch": 0.9609851665267283,
+      "grad_norm": 1.3213703632354736,
+      "learning_rate": 2.677746999076639e-05,
+      "loss": 0.1557,
+      "step": 2146
+    },
+    {
+      "epoch": 0.9614329694934229,
+      "grad_norm": 1.2782256603240967,
+      "learning_rate": 2.676592797783934e-05,
+      "loss": 0.1,
+      "step": 2147
+    },
+    {
+      "epoch": 0.9618807724601175,
+      "grad_norm": 0.9655989408493042,
+      "learning_rate": 2.675438596491228e-05,
+      "loss": 0.1161,
+      "step": 2148
+    },
+    {
+      "epoch": 0.9623285754268122,
+      "grad_norm": 0.8622291088104248,
+      "learning_rate": 2.6742843951985224e-05,
+      "loss": 0.0622,
+      "step": 2149
+    },
+    {
+      "epoch": 0.9627763783935068,
+      "grad_norm": 1.0358092784881592,
+      "learning_rate": 2.6731301939058173e-05,
+      "loss": 0.1415,
+      "step": 2150
+    },
+    {
+      "epoch": 0.9627763783935068,
+      "eval_loss": 0.12110047042369843,
+      "eval_runtime": 1729.86,
+      "eval_samples_per_second": 2.582,
+      "eval_steps_per_second": 2.582,
+      "step": 2150
+    },
+    {
+      "epoch": 0.9632241813602015,
+      "grad_norm": 1.3379656076431274,
+      "learning_rate": 2.6719759926131116e-05,
+      "loss": 0.1732,
+      "step": 2151
+    },
+    {
+      "epoch": 0.9636719843268962,
+      "grad_norm": 1.2671988010406494,
+      "learning_rate": 2.6708217913204065e-05,
+      "loss": 0.1829,
+      "step": 2152
+    },
+    {
+      "epoch": 0.9641197872935908,
+      "grad_norm": 1.749251365661621,
+      "learning_rate": 2.6696675900277008e-05,
+      "loss": 0.2139,
+      "step": 2153
+    },
+    {
+      "epoch": 0.9645675902602855,
+      "grad_norm": 0.9710353016853333,
+      "learning_rate": 2.6685133887349957e-05,
+      "loss": 0.0672,
+      "step": 2154
+    },
+    {
+      "epoch": 0.9650153932269802,
+      "grad_norm": 1.379391074180603,
+      "learning_rate": 2.66735918744229e-05,
+      "loss": 0.2084,
+      "step": 2155
+    },
+    {
+      "epoch": 0.9654631961936748,
+      "grad_norm": 0.9271833896636963,
+      "learning_rate": 2.6662049861495846e-05,
+      "loss": 0.0758,
+      "step": 2156
+    },
+    {
+      "epoch": 0.9659109991603695,
+      "grad_norm": 1.5926554203033447,
+      "learning_rate": 2.6650507848568788e-05,
+      "loss": 0.347,
+      "step": 2157
+    },
+    {
+      "epoch": 0.966358802127064,
+      "grad_norm": 0.7312039732933044,
+      "learning_rate": 2.6638965835641738e-05,
+      "loss": 0.0647,
+      "step": 2158
+    },
+    {
+      "epoch": 0.9668066050937587,
+      "grad_norm": 0.9427241683006287,
+      "learning_rate": 2.662742382271468e-05,
+      "loss": 0.0663,
+      "step": 2159
+    },
+    {
+      "epoch": 0.9672544080604534,
+      "grad_norm": 1.3281080722808838,
+      "learning_rate": 2.661588180978763e-05,
+      "loss": 0.161,
+      "step": 2160
+    },
+    {
+      "epoch": 0.9672544080604534,
+      "eval_loss": 0.11987407505512238,
+      "eval_runtime": 1729.868,
+      "eval_samples_per_second": 2.582,
+      "eval_steps_per_second": 2.582,
+      "step": 2160
+    },
+    {
+      "epoch": 0.967702211027148,
+      "grad_norm": 1.2127536535263062,
+      "learning_rate": 2.6604339796860572e-05,
+      "loss": 0.1544,
+      "step": 2161
+    },
+    {
+      "epoch": 0.9681500139938427,
+      "grad_norm": 1.1174790859222412,
+      "learning_rate": 2.659279778393352e-05,
+      "loss": 0.0893,
+      "step": 2162
+    },
+    {
+      "epoch": 0.9685978169605374,
+      "grad_norm": 1.026856780052185,
+      "learning_rate": 2.6581255771006464e-05,
+      "loss": 0.096,
+      "step": 2163
+    },
+    {
+      "epoch": 0.969045619927232,
+      "grad_norm": 1.323331356048584,
+      "learning_rate": 2.6569713758079414e-05,
+      "loss": 0.2058,
+      "step": 2164
+    },
+    {
+      "epoch": 0.9694934228939267,
+      "grad_norm": 1.1892203092575073,
+      "learning_rate": 2.6558171745152356e-05,
+      "loss": 0.0926,
+      "step": 2165
+    },
+    {
+      "epoch": 0.9699412258606214,
+      "grad_norm": 0.855096161365509,
+      "learning_rate": 2.6546629732225302e-05,
+      "loss": 0.0613,
+      "step": 2166
+    },
+    {
+      "epoch": 0.970389028827316,
+      "grad_norm": 1.0742472410202026,
+      "learning_rate": 2.6535087719298245e-05,
+      "loss": 0.084,
+      "step": 2167
+    },
+    {
+      "epoch": 0.9708368317940106,
+      "grad_norm": 0.9125453233718872,
+      "learning_rate": 2.6523545706371194e-05,
+      "loss": 0.0604,
+      "step": 2168
+    },
+    {
+      "epoch": 0.9712846347607053,
+      "grad_norm": 1.4418150186538696,
+      "learning_rate": 2.6512003693444137e-05,
+      "loss": 0.1148,
+      "step": 2169
+    },
+    {
+      "epoch": 0.9717324377273999,
+      "grad_norm": 0.7091181874275208,
+      "learning_rate": 2.6500461680517086e-05,
+      "loss": 0.0828,
+      "step": 2170
+    },
+    {
+      "epoch": 0.9717324377273999,
+      "eval_loss": 0.11846964806318283,
+      "eval_runtime": 1731.7718,
+      "eval_samples_per_second": 2.579,
+      "eval_steps_per_second": 2.579,
+      "step": 2170
+    },
+    {
+      "epoch": 0.9721802406940946,
+      "grad_norm": 0.848183274269104,
+      "learning_rate": 2.648891966759003e-05,
+      "loss": 0.0814,
+      "step": 2171
+    },
+    {
+      "epoch": 0.9726280436607893,
+      "grad_norm": 1.014132022857666,
+      "learning_rate": 2.6477377654662978e-05,
+      "loss": 0.0882,
+      "step": 2172
+    },
+    {
+      "epoch": 0.9730758466274839,
+      "grad_norm": 0.807933509349823,
+      "learning_rate": 2.646583564173592e-05,
+      "loss": 0.0792,
+      "step": 2173
+    },
+    {
+      "epoch": 0.9735236495941786,
+      "grad_norm": 1.0491405725479126,
+      "learning_rate": 2.6454293628808867e-05,
+      "loss": 0.0683,
+      "step": 2174
+    },
+    {
+      "epoch": 0.9739714525608733,
+      "grad_norm": 1.0016448497772217,
+      "learning_rate": 2.6442751615881813e-05,
+      "loss": 0.0677,
+      "step": 2175
+    },
+    {
+      "epoch": 0.9744192555275679,
+      "grad_norm": 1.008288025856018,
+      "learning_rate": 2.6431209602954755e-05,
+      "loss": 0.0913,
+      "step": 2176
+    },
+    {
+      "epoch": 0.9748670584942625,
+      "grad_norm": 0.911280632019043,
+      "learning_rate": 2.64196675900277e-05,
+      "loss": 0.0986,
+      "step": 2177
+    },
+    {
+      "epoch": 0.9753148614609571,
+      "grad_norm": 1.6649196147918701,
+      "learning_rate": 2.6408125577100644e-05,
+      "loss": 0.1501,
+      "step": 2178
+    },
+    {
+      "epoch": 0.9757626644276518,
+      "grad_norm": 1.039286494255066,
+      "learning_rate": 2.6396583564173593e-05,
+      "loss": 0.0721,
+      "step": 2179
+    },
+    {
+      "epoch": 0.9762104673943465,
+      "grad_norm": 1.164796233177185,
+      "learning_rate": 2.6385041551246536e-05,
+      "loss": 0.1483,
+      "step": 2180
+    },
+    {
+      "epoch": 0.9762104673943465,
+      "eval_loss": 0.11873579770326614,
+      "eval_runtime": 1728.0248,
+      "eval_samples_per_second": 2.585,
+      "eval_steps_per_second": 2.585,
+      "step": 2180
+    },
+    {
+      "epoch": 0.9766582703610411,
+      "grad_norm": 1.7558592557907104,
+      "learning_rate": 2.6373499538319485e-05,
+      "loss": 0.2055,
+      "step": 2181
+    },
+    {
+      "epoch": 0.9771060733277358,
+      "grad_norm": 1.5141433477401733,
+      "learning_rate": 2.6361957525392428e-05,
+      "loss": 0.1019,
+      "step": 2182
+    },
+    {
+      "epoch": 0.9775538762944305,
+      "grad_norm": 1.5504546165466309,
+      "learning_rate": 2.6350415512465377e-05,
+      "loss": 0.1048,
+      "step": 2183
+    },
+    {
+      "epoch": 0.9780016792611251,
+      "grad_norm": 0.9713227152824402,
+      "learning_rate": 2.633887349953832e-05,
+      "loss": 0.0762,
+      "step": 2184
+    },
+    {
+      "epoch": 0.9784494822278198,
+      "grad_norm": 1.3416686058044434,
+      "learning_rate": 2.6327331486611266e-05,
+      "loss": 0.1602,
+      "step": 2185
+    },
+    {
+      "epoch": 0.9788972851945145,
+      "grad_norm": 0.9782314896583557,
+      "learning_rate": 2.6315789473684212e-05,
+      "loss": 0.1174,
+      "step": 2186
+    },
+    {
+      "epoch": 0.979345088161209,
+      "grad_norm": 0.8174190521240234,
+      "learning_rate": 2.6304247460757158e-05,
+      "loss": 0.0682,
+      "step": 2187
+    },
+    {
+      "epoch": 0.9797928911279037,
+      "grad_norm": 1.2250410318374634,
+      "learning_rate": 2.62927054478301e-05,
+      "loss": 0.0933,
+      "step": 2188
+    },
+    {
+      "epoch": 0.9802406940945984,
+      "grad_norm": 1.0255568027496338,
+      "learning_rate": 2.628116343490305e-05,
+      "loss": 0.0666,
+      "step": 2189
+    },
+    {
+      "epoch": 0.980688497061293,
+      "grad_norm": 1.3774869441986084,
+      "learning_rate": 2.6269621421975993e-05,
+      "loss": 0.1049,
+      "step": 2190
+    },
+    {
+      "epoch": 0.980688497061293,
+      "eval_loss": 0.11641395092010498,
+      "eval_runtime": 1733.033,
+      "eval_samples_per_second": 2.578,
+      "eval_steps_per_second": 2.578,
+      "step": 2190
+    },
+    {
+      "epoch": 0.9811363000279877,
+      "grad_norm": 0.9319576025009155,
+      "learning_rate": 2.6258079409048942e-05,
+      "loss": 0.1294,
+      "step": 2191
+    },
+    {
+      "epoch": 0.9815841029946824,
+      "grad_norm": 0.9967508316040039,
+      "learning_rate": 2.6246537396121885e-05,
+      "loss": 0.1042,
+      "step": 2192
+    },
+    {
+      "epoch": 0.982031905961377,
+      "grad_norm": 1.1014225482940674,
+      "learning_rate": 2.6234995383194834e-05,
+      "loss": 0.0992,
+      "step": 2193
+    },
+    {
+      "epoch": 0.9824797089280717,
+      "grad_norm": 1.0176576375961304,
+      "learning_rate": 2.6223453370267777e-05,
+      "loss": 0.0895,
+      "step": 2194
+    },
+    {
+      "epoch": 0.9829275118947663,
+      "grad_norm": 1.006609559059143,
+      "learning_rate": 2.6211911357340723e-05,
+      "loss": 0.1612,
+      "step": 2195
+    },
+    {
+      "epoch": 0.9833753148614609,
+      "grad_norm": 0.9863540530204773,
+      "learning_rate": 2.6200369344413665e-05,
+      "loss": 0.0949,
+      "step": 2196
+    },
+    {
+      "epoch": 0.9838231178281556,
+      "grad_norm": 0.6519593596458435,
+      "learning_rate": 2.6188827331486615e-05,
+      "loss": 0.2872,
+      "step": 2197
+    },
+    {
+      "epoch": 0.9842709207948502,
+      "grad_norm": 0.6476958394050598,
+      "learning_rate": 2.6177285318559557e-05,
+      "loss": 0.0532,
+      "step": 2198
+    },
+    {
+      "epoch": 0.9847187237615449,
+      "grad_norm": 1.404766321182251,
+      "learning_rate": 2.6165743305632507e-05,
+      "loss": 0.1264,
+      "step": 2199
+    },
+    {
+      "epoch": 0.9851665267282396,
+      "grad_norm": 0.9817789196968079,
+      "learning_rate": 2.615420129270545e-05,
+      "loss": 0.1291,
+      "step": 2200
+    },
+    {
+      "epoch": 0.9851665267282396,
+      "eval_loss": 0.1149076521396637,
+      "eval_runtime": 1727.9785,
+      "eval_samples_per_second": 2.585,
+      "eval_steps_per_second": 2.585,
+      "step": 2200
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 4466,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.476867796544799e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}