diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4053 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.994773519163763,
+  "eval_steps": 500,
+  "global_step": 573,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.005226480836236934,
+      "grad_norm": 6.817761533477436,
+      "learning_rate": 1.7241379310344828e-07,
+      "loss": 1.0919,
+      "step": 1
+    },
+    {
+      "epoch": 0.010452961672473868,
+      "grad_norm": 7.115516920325354,
+      "learning_rate": 3.4482758620689656e-07,
+      "loss": 1.1216,
+      "step": 2
+    },
+    {
+      "epoch": 0.0156794425087108,
+      "grad_norm": 6.6059432881920035,
+      "learning_rate": 5.172413793103449e-07,
+      "loss": 1.0823,
+      "step": 3
+    },
+    {
+      "epoch": 0.020905923344947737,
+      "grad_norm": 7.040897553026814,
+      "learning_rate": 6.896551724137931e-07,
+      "loss": 1.1163,
+      "step": 4
+    },
+    {
+      "epoch": 0.02613240418118467,
+      "grad_norm": 6.283884644524741,
+      "learning_rate": 8.620689655172415e-07,
+      "loss": 1.0475,
+      "step": 5
+    },
+    {
+      "epoch": 0.0313588850174216,
+      "grad_norm": 6.519407744623921,
+      "learning_rate": 1.0344827586206898e-06,
+      "loss": 1.0632,
+      "step": 6
+    },
+    {
+      "epoch": 0.036585365853658534,
+      "grad_norm": 6.169268594264987,
+      "learning_rate": 1.2068965517241381e-06,
+      "loss": 1.0867,
+      "step": 7
+    },
+    {
+      "epoch": 0.041811846689895474,
+      "grad_norm": 6.0230600012209905,
+      "learning_rate": 1.3793103448275862e-06,
+      "loss": 1.0679,
+      "step": 8
+    },
+    {
+      "epoch": 0.047038327526132406,
+      "grad_norm": 4.974697996657908,
+      "learning_rate": 1.5517241379310346e-06,
+      "loss": 1.0599,
+      "step": 9
+    },
+    {
+      "epoch": 0.05226480836236934,
+      "grad_norm": 4.610162697639871,
+      "learning_rate": 1.724137931034483e-06,
+      "loss": 1.0377,
+      "step": 10
+    },
+    {
+      "epoch": 0.05749128919860627,
+      "grad_norm": 3.231983464184593,
+      "learning_rate": 1.896551724137931e-06,
+      "loss": 0.9836,
+      "step": 11
+    },
+    {
+      "epoch": 0.0627177700348432,
+      "grad_norm": 2.8094756280140682,
+      "learning_rate": 2.0689655172413796e-06,
+      "loss": 1.0003,
+      "step": 12
+    },
+    {
+      "epoch": 0.06794425087108014,
+      "grad_norm": 2.5493675085491594,
+      "learning_rate": 2.241379310344828e-06,
+      "loss": 0.9805,
+      "step": 13
+    },
+    {
+      "epoch": 0.07317073170731707,
+      "grad_norm": 2.5421388777679423,
+      "learning_rate": 2.4137931034482762e-06,
+      "loss": 0.9785,
+      "step": 14
+    },
+    {
+      "epoch": 0.078397212543554,
+      "grad_norm": 2.4038506323025564,
+      "learning_rate": 2.5862068965517246e-06,
+      "loss": 0.9579,
+      "step": 15
+    },
+    {
+      "epoch": 0.08362369337979095,
+      "grad_norm": 3.118376309649565,
+      "learning_rate": 2.7586206896551725e-06,
+      "loss": 0.9266,
+      "step": 16
+    },
+    {
+      "epoch": 0.08885017421602788,
+      "grad_norm": 3.256072028222514,
+      "learning_rate": 2.931034482758621e-06,
+      "loss": 0.9496,
+      "step": 17
+    },
+    {
+      "epoch": 0.09407665505226481,
+      "grad_norm": 3.0097390475425785,
+      "learning_rate": 3.103448275862069e-06,
+      "loss": 0.9417,
+      "step": 18
+    },
+    {
+      "epoch": 0.09930313588850175,
+      "grad_norm": 2.5804981571566685,
+      "learning_rate": 3.2758620689655175e-06,
+      "loss": 0.9133,
+      "step": 19
+    },
+    {
+      "epoch": 0.10452961672473868,
+      "grad_norm": 2.1253546888275614,
+      "learning_rate": 3.448275862068966e-06,
+      "loss": 0.9181,
+      "step": 20
+    },
+    {
+      "epoch": 0.10975609756097561,
+      "grad_norm": 1.4630576625319751,
+      "learning_rate": 3.620689655172414e-06,
+      "loss": 0.8708,
+      "step": 21
+    },
+    {
+      "epoch": 0.11498257839721254,
+      "grad_norm": 1.8224890188617031,
+      "learning_rate": 3.793103448275862e-06,
+      "loss": 0.8532,
+      "step": 22
+    },
+    {
+      "epoch": 0.12020905923344948,
+      "grad_norm": 1.947817512922658,
+      "learning_rate": 3.96551724137931e-06,
+      "loss": 0.867,
+      "step": 23
+    },
+    {
+      "epoch": 0.1254355400696864,
+      "grad_norm": 1.6741792992922881,
+      "learning_rate": 4.137931034482759e-06,
+      "loss": 0.8596,
+      "step": 24
+    },
+    {
+      "epoch": 0.13066202090592335,
+      "grad_norm": 1.1380185267106577,
+      "learning_rate": 4.310344827586207e-06,
+      "loss": 0.8119,
+      "step": 25
+    },
+    {
+      "epoch": 0.13588850174216027,
+      "grad_norm": 1.2049716071539038,
+      "learning_rate": 4.482758620689656e-06,
+      "loss": 0.8374,
+      "step": 26
+    },
+    {
+      "epoch": 0.14111498257839722,
+      "grad_norm": 1.1289607052380697,
+      "learning_rate": 4.655172413793104e-06,
+      "loss": 0.7917,
+      "step": 27
+    },
+    {
+      "epoch": 0.14634146341463414,
+      "grad_norm": 1.2662899307258315,
+      "learning_rate": 4.8275862068965525e-06,
+      "loss": 0.8187,
+      "step": 28
+    },
+    {
+      "epoch": 0.15156794425087108,
+      "grad_norm": 1.3017674903802239,
+      "learning_rate": 5e-06,
+      "loss": 0.8106,
+      "step": 29
+    },
+    {
+      "epoch": 0.156794425087108,
+      "grad_norm": 0.9437556124885579,
+      "learning_rate": 5.172413793103449e-06,
+      "loss": 0.7807,
+      "step": 30
+    },
+    {
+      "epoch": 0.16202090592334495,
+      "grad_norm": 0.9854643253704647,
+      "learning_rate": 5.344827586206896e-06,
+      "loss": 0.7987,
+      "step": 31
+    },
+    {
+      "epoch": 0.1672473867595819,
+      "grad_norm": 1.0216232146030035,
+      "learning_rate": 5.517241379310345e-06,
+      "loss": 0.7978,
+      "step": 32
+    },
+    {
+      "epoch": 0.17247386759581881,
+      "grad_norm": 0.8155263699639064,
+      "learning_rate": 5.689655172413794e-06,
+      "loss": 0.7872,
+      "step": 33
+    },
+    {
+      "epoch": 0.17770034843205576,
+      "grad_norm": 0.8283995090284338,
+      "learning_rate": 5.862068965517242e-06,
+      "loss": 0.7704,
+      "step": 34
+    },
+    {
+      "epoch": 0.18292682926829268,
+      "grad_norm": 0.7185821947364432,
+      "learning_rate": 6.03448275862069e-06,
+      "loss": 0.7581,
+      "step": 35
+    },
+    {
+      "epoch": 0.18815331010452963,
+      "grad_norm": 0.770352836030317,
+      "learning_rate": 6.206896551724138e-06,
+      "loss": 0.7787,
+      "step": 36
+    },
+    {
+      "epoch": 0.19337979094076654,
+      "grad_norm": 0.7388881984398195,
+      "learning_rate": 6.379310344827587e-06,
+      "loss": 0.745,
+      "step": 37
+    },
+    {
+      "epoch": 0.1986062717770035,
+      "grad_norm": 0.8358555735535642,
+      "learning_rate": 6.551724137931035e-06,
+      "loss": 0.7604,
+      "step": 38
+    },
+    {
+      "epoch": 0.2038327526132404,
+      "grad_norm": 0.793851198978272,
+      "learning_rate": 6.724137931034484e-06,
+      "loss": 0.7544,
+      "step": 39
+    },
+    {
+      "epoch": 0.20905923344947736,
+      "grad_norm": 0.7772514888903087,
+      "learning_rate": 6.896551724137932e-06,
+      "loss": 0.732,
+      "step": 40
+    },
+    {
+      "epoch": 0.21428571428571427,
+      "grad_norm": 0.8148652288090457,
+      "learning_rate": 7.0689655172413796e-06,
+      "loss": 0.7297,
+      "step": 41
+    },
+    {
+      "epoch": 0.21951219512195122,
+      "grad_norm": 0.8289759706310655,
+      "learning_rate": 7.241379310344828e-06,
+      "loss": 0.7383,
+      "step": 42
+    },
+    {
+      "epoch": 0.22473867595818817,
+      "grad_norm": 0.7140088762027519,
+      "learning_rate": 7.413793103448277e-06,
+      "loss": 0.7461,
+      "step": 43
+    },
+    {
+      "epoch": 0.22996515679442509,
+      "grad_norm": 0.6669906555709666,
+      "learning_rate": 7.586206896551724e-06,
+      "loss": 0.7343,
+      "step": 44
+    },
+    {
+      "epoch": 0.23519163763066203,
+      "grad_norm": 0.7425162700101843,
+      "learning_rate": 7.758620689655173e-06,
+      "loss": 0.7211,
+      "step": 45
+    },
+    {
+      "epoch": 0.24041811846689895,
+      "grad_norm": 0.7667317947005017,
+      "learning_rate": 7.93103448275862e-06,
+      "loss": 0.7307,
+      "step": 46
+    },
+    {
+      "epoch": 0.2456445993031359,
+      "grad_norm": 0.6701254662379068,
+      "learning_rate": 8.103448275862069e-06,
+      "loss": 0.7244,
+      "step": 47
+    },
+    {
+      "epoch": 0.2508710801393728,
+      "grad_norm": 0.6974840533976149,
+      "learning_rate": 8.275862068965518e-06,
+      "loss": 0.7222,
+      "step": 48
+    },
+    {
+      "epoch": 0.25609756097560976,
+      "grad_norm": 0.7601606441382056,
+      "learning_rate": 8.448275862068966e-06,
+      "loss": 0.7325,
+      "step": 49
+    },
+    {
+      "epoch": 0.2613240418118467,
+      "grad_norm": 0.6932741056532161,
+      "learning_rate": 8.620689655172414e-06,
+      "loss": 0.7166,
+      "step": 50
+    },
+    {
+      "epoch": 0.2665505226480836,
+      "grad_norm": 0.7185062343660426,
+      "learning_rate": 8.793103448275862e-06,
+      "loss": 0.7367,
+      "step": 51
+    },
+    {
+      "epoch": 0.27177700348432055,
+      "grad_norm": 0.6728422963531511,
+      "learning_rate": 8.965517241379312e-06,
+      "loss": 0.7104,
+      "step": 52
+    },
+    {
+      "epoch": 0.2770034843205575,
+      "grad_norm": 0.661725549134852,
+      "learning_rate": 9.13793103448276e-06,
+      "loss": 0.7183,
+      "step": 53
+    },
+    {
+      "epoch": 0.28222996515679444,
+      "grad_norm": 0.6362203257614645,
+      "learning_rate": 9.310344827586207e-06,
+      "loss": 0.7153,
+      "step": 54
+    },
+    {
+      "epoch": 0.2874564459930314,
+      "grad_norm": 0.5890508129443038,
+      "learning_rate": 9.482758620689655e-06,
+      "loss": 0.7093,
+      "step": 55
+    },
+    {
+      "epoch": 0.2926829268292683,
+      "grad_norm": 0.6116919862768194,
+      "learning_rate": 9.655172413793105e-06,
+      "loss": 0.7176,
+      "step": 56
+    },
+    {
+      "epoch": 0.2979094076655052,
+      "grad_norm": 0.7050152514142426,
+      "learning_rate": 9.827586206896553e-06,
+      "loss": 0.7211,
+      "step": 57
+    },
+    {
+      "epoch": 0.30313588850174217,
+      "grad_norm": 0.6296132679526024,
+      "learning_rate": 1e-05,
+      "loss": 0.6993,
+      "step": 58
+    },
+    {
+      "epoch": 0.3083623693379791,
+      "grad_norm": 0.6453523097142025,
+      "learning_rate": 9.999906969801156e-06,
+      "loss": 0.6963,
+      "step": 59
+    },
+    {
+      "epoch": 0.313588850174216,
+      "grad_norm": 0.6955516834757508,
+      "learning_rate": 9.999627882666474e-06,
+      "loss": 0.7151,
+      "step": 60
+    },
+    {
+      "epoch": 0.31881533101045295,
+      "grad_norm": 0.5580219589427374,
+      "learning_rate": 9.999162748981362e-06,
+      "loss": 0.7049,
+      "step": 61
+    },
+    {
+      "epoch": 0.3240418118466899,
+      "grad_norm": 0.5910429225441691,
+      "learning_rate": 9.998511586054415e-06,
+      "loss": 0.7013,
+      "step": 62
+    },
+    {
+      "epoch": 0.32926829268292684,
+      "grad_norm": 0.8120038822728922,
+      "learning_rate": 9.997674418116759e-06,
+      "loss": 0.717,
+      "step": 63
+    },
+    {
+      "epoch": 0.3344947735191638,
+      "grad_norm": 0.7517280345835676,
+      "learning_rate": 9.996651276321153e-06,
+      "loss": 0.6952,
+      "step": 64
+    },
+    {
+      "epoch": 0.3397212543554007,
+      "grad_norm": 0.6469274840068534,
+      "learning_rate": 9.995442198740832e-06,
+      "loss": 0.701,
+      "step": 65
+    },
+    {
+      "epoch": 0.34494773519163763,
+      "grad_norm": 0.9251247112850067,
+      "learning_rate": 9.994047230368087e-06,
+      "loss": 0.7124,
+      "step": 66
+    },
+    {
+      "epoch": 0.3501742160278746,
+      "grad_norm": 0.662124581217118,
+      "learning_rate": 9.99246642311259e-06,
+      "loss": 0.7041,
+      "step": 67
+    },
+    {
+      "epoch": 0.3554006968641115,
+      "grad_norm": 0.6490157119284249,
+      "learning_rate": 9.99069983579947e-06,
+      "loss": 0.6885,
+      "step": 68
+    },
+    {
+      "epoch": 0.3606271777003484,
+      "grad_norm": 0.8298745583053626,
+      "learning_rate": 9.988747534167112e-06,
+      "loss": 0.7179,
+      "step": 69
+    },
+    {
+      "epoch": 0.36585365853658536,
+      "grad_norm": 0.561322661520838,
+      "learning_rate": 9.98660959086472e-06,
+      "loss": 0.6955,
+      "step": 70
+    },
+    {
+      "epoch": 0.3710801393728223,
+      "grad_norm": 0.7057604500980034,
+      "learning_rate": 9.98428608544961e-06,
+      "loss": 0.6903,
+      "step": 71
+    },
+    {
+      "epoch": 0.37630662020905925,
+      "grad_norm": 0.6658460439147413,
+      "learning_rate": 9.981777104384251e-06,
+      "loss": 0.7082,
+      "step": 72
+    },
+    {
+      "epoch": 0.38153310104529614,
+      "grad_norm": 0.5943504077793226,
+      "learning_rate": 9.979082741033047e-06,
+      "loss": 0.6768,
+      "step": 73
+    },
+    {
+      "epoch": 0.3867595818815331,
+      "grad_norm": 0.6864915702650408,
+      "learning_rate": 9.976203095658859e-06,
+      "loss": 0.6761,
+      "step": 74
+    },
+    {
+      "epoch": 0.39198606271777003,
+      "grad_norm": 0.6390276002240977,
+      "learning_rate": 9.97313827541928e-06,
+      "loss": 0.6934,
+      "step": 75
+    },
+    {
+      "epoch": 0.397212543554007,
+      "grad_norm": 0.6038585831748531,
+      "learning_rate": 9.969888394362648e-06,
+      "loss": 0.7041,
+      "step": 76
+    },
+    {
+      "epoch": 0.4024390243902439,
+      "grad_norm": 0.6334231152676686,
+      "learning_rate": 9.96645357342379e-06,
+      "loss": 0.6945,
+      "step": 77
+    },
+    {
+      "epoch": 0.4076655052264808,
+      "grad_norm": 0.5462598060569209,
+      "learning_rate": 9.96283394041954e-06,
+      "loss": 0.696,
+      "step": 78
+    },
+    {
+      "epoch": 0.41289198606271776,
+      "grad_norm": 0.6338524439755173,
+      "learning_rate": 9.959029630043969e-06,
+      "loss": 0.691,
+      "step": 79
+    },
+    {
+      "epoch": 0.4181184668989547,
+      "grad_norm": 0.6411928123304964,
+      "learning_rate": 9.955040783863373e-06,
+      "loss": 0.692,
+      "step": 80
+    },
+    {
+      "epoch": 0.42334494773519166,
+      "grad_norm": 0.8030581042650948,
+      "learning_rate": 9.950867550311019e-06,
+      "loss": 0.7196,
+      "step": 81
+    },
+    {
+      "epoch": 0.42857142857142855,
+      "grad_norm": 0.6024953623866571,
+      "learning_rate": 9.946510084681602e-06,
+      "loss": 0.6726,
+      "step": 82
+    },
+    {
+      "epoch": 0.4337979094076655,
+      "grad_norm": 0.6074570740511613,
+      "learning_rate": 9.941968549125481e-06,
+      "loss": 0.6897,
+      "step": 83
+    },
+    {
+      "epoch": 0.43902439024390244,
+      "grad_norm": 0.7048846718482608,
+      "learning_rate": 9.937243112642639e-06,
+      "loss": 0.6918,
+      "step": 84
+    },
+    {
+      "epoch": 0.4442508710801394,
+      "grad_norm": 0.5918640923441442,
+      "learning_rate": 9.932333951076395e-06,
+      "loss": 0.6766,
+      "step": 85
+    },
+    {
+      "epoch": 0.44947735191637633,
+      "grad_norm": 0.5926297060359819,
+      "learning_rate": 9.927241247106856e-06,
+      "loss": 0.695,
+      "step": 86
+    },
+    {
+      "epoch": 0.4547038327526132,
+      "grad_norm": 0.553208274597187,
+      "learning_rate": 9.921965190244129e-06,
+      "loss": 0.6534,
+      "step": 87
+    },
+    {
+      "epoch": 0.45993031358885017,
+      "grad_norm": 0.7161403077582934,
+      "learning_rate": 9.916505976821262e-06,
+      "loss": 0.6964,
+      "step": 88
+    },
+    {
+      "epoch": 0.4651567944250871,
+      "grad_norm": 0.6949109145902641,
+      "learning_rate": 9.910863809986942e-06,
+      "loss": 0.6948,
+      "step": 89
+    },
+    {
+      "epoch": 0.47038327526132406,
+      "grad_norm": 0.669998554662786,
+      "learning_rate": 9.905038899697924e-06,
+      "loss": 0.6748,
+      "step": 90
+    },
+    {
+      "epoch": 0.47560975609756095,
+      "grad_norm": 0.812799215586504,
+      "learning_rate": 9.899031462711237e-06,
+      "loss": 0.698,
+      "step": 91
+    },
+    {
+      "epoch": 0.4808362369337979,
+      "grad_norm": 0.7765125728252844,
+      "learning_rate": 9.892841722576103e-06,
+      "loss": 0.6839,
+      "step": 92
+    },
+    {
+      "epoch": 0.48606271777003485,
+      "grad_norm": 0.7279630881961654,
+      "learning_rate": 9.886469909625624e-06,
+      "loss": 0.6865,
+      "step": 93
+    },
+    {
+      "epoch": 0.4912891986062718,
+      "grad_norm": 0.8884193887396585,
+      "learning_rate": 9.879916260968212e-06,
+      "loss": 0.6693,
+      "step": 94
+    },
+    {
+      "epoch": 0.4965156794425087,
+      "grad_norm": 0.607572845135145,
+      "learning_rate": 9.87318102047876e-06,
+      "loss": 0.6812,
+      "step": 95
+    },
+    {
+      "epoch": 0.5017421602787456,
+      "grad_norm": 0.7203962118826617,
+      "learning_rate": 9.866264438789573e-06,
+      "loss": 0.686,
+      "step": 96
+    },
+    {
+      "epoch": 0.5069686411149826,
+      "grad_norm": 0.6785442509634554,
+      "learning_rate": 9.85916677328104e-06,
+      "loss": 0.7037,
+      "step": 97
+    },
+    {
+      "epoch": 0.5121951219512195,
+      "grad_norm": 0.5917891054628419,
+      "learning_rate": 9.851888288072053e-06,
+      "loss": 0.7037,
+      "step": 98
+    },
+    {
+      "epoch": 0.5174216027874564,
+      "grad_norm": 0.7601539373800206,
+      "learning_rate": 9.844429254010184e-06,
+      "loss": 0.6969,
+      "step": 99
+    },
+    {
+      "epoch": 0.5226480836236934,
+      "grad_norm": 0.6020652320578571,
+      "learning_rate": 9.836789948661602e-06,
+      "loss": 0.6879,
+      "step": 100
+    },
+    {
+      "epoch": 0.5278745644599303,
+      "grad_norm": 0.6764576509816881,
+      "learning_rate": 9.828970656300743e-06,
+      "loss": 0.6898,
+      "step": 101
+    },
+    {
+      "epoch": 0.5331010452961672,
+      "grad_norm": 0.6666874805819255,
+      "learning_rate": 9.82097166789974e-06,
+      "loss": 0.6778,
+      "step": 102
+    },
+    {
+      "epoch": 0.5383275261324042,
+      "grad_norm": 0.7360464789092225,
+      "learning_rate": 9.81279328111758e-06,
+      "loss": 0.6882,
+      "step": 103
+    },
+    {
+      "epoch": 0.5435540069686411,
+      "grad_norm": 0.701626438763124,
+      "learning_rate": 9.804435800289047e-06,
+      "loss": 0.6893,
+      "step": 104
+    },
+    {
+      "epoch": 0.5487804878048781,
+      "grad_norm": 0.7646586589346112,
+      "learning_rate": 9.795899536413383e-06,
+      "loss": 0.663,
+      "step": 105
+    },
+    {
+      "epoch": 0.554006968641115,
+      "grad_norm": 0.7827135637222683,
+      "learning_rate": 9.787184807142713e-06,
+      "loss": 0.6782,
+      "step": 106
+    },
+    {
+      "epoch": 0.5592334494773519,
+      "grad_norm": 0.7963355955603488,
+      "learning_rate": 9.778291936770241e-06,
+      "loss": 0.6948,
+      "step": 107
+    },
+    {
+      "epoch": 0.5644599303135889,
+      "grad_norm": 0.9185384976633354,
+      "learning_rate": 9.769221256218165e-06,
+      "loss": 0.6855,
+      "step": 108
+    },
+    {
+      "epoch": 0.5696864111498258,
+      "grad_norm": 0.6715876776695767,
+      "learning_rate": 9.759973103025369e-06,
+      "loss": 0.6692,
+      "step": 109
+    },
+    {
+      "epoch": 0.5749128919860628,
+      "grad_norm": 1.2389220927650733,
+      "learning_rate": 9.750547821334868e-06,
+      "loss": 0.6881,
+      "step": 110
+    },
+    {
+      "epoch": 0.5801393728222997,
+      "grad_norm": 0.7818470562308032,
+      "learning_rate": 9.740945761880993e-06,
+      "loss": 0.6483,
+      "step": 111
+    },
+    {
+      "epoch": 0.5853658536585366,
+      "grad_norm": 0.8194633115862917,
+      "learning_rate": 9.731167281976346e-06,
+      "loss": 0.6573,
+      "step": 112
+    },
+    {
+      "epoch": 0.5905923344947736,
+      "grad_norm": 0.8554081467672775,
+      "learning_rate": 9.721212745498493e-06,
+      "loss": 0.6831,
+      "step": 113
+    },
+    {
+      "epoch": 0.5958188153310104,
+      "grad_norm": 0.6301357489652583,
+      "learning_rate": 9.711082522876445e-06,
+      "loss": 0.6883,
+      "step": 114
+    },
+    {
+      "epoch": 0.6010452961672473,
+      "grad_norm": 0.7320145359964577,
+      "learning_rate": 9.700776991076846e-06,
+      "loss": 0.6826,
+      "step": 115
+    },
+    {
+      "epoch": 0.6062717770034843,
+      "grad_norm": 0.7164618151118382,
+      "learning_rate": 9.690296533589967e-06,
+      "loss": 0.6669,
+      "step": 116
+    },
+    {
+      "epoch": 0.6114982578397212,
+      "grad_norm": 0.6788237892173536,
+      "learning_rate": 9.679641540415428e-06,
+      "loss": 0.6659,
+      "step": 117
+    },
+    {
+      "epoch": 0.6167247386759582,
+      "grad_norm": 0.6158473714288532,
+      "learning_rate": 9.66881240804768e-06,
+      "loss": 0.6597,
+      "step": 118
+    },
+    {
+      "epoch": 0.6219512195121951,
+      "grad_norm": 0.6555526405519565,
+      "learning_rate": 9.657809539461256e-06,
+      "loss": 0.6831,
+      "step": 119
+    },
+    {
+      "epoch": 0.627177700348432,
+      "grad_norm": 0.6887980438356484,
+      "learning_rate": 9.64663334409578e-06,
+      "loss": 0.666,
+      "step": 120
+    },
+    {
+      "epoch": 0.632404181184669,
+      "grad_norm": 0.5979509900819844,
+      "learning_rate": 9.635284237840721e-06,
+      "loss": 0.6553,
+      "step": 121
+    },
+    {
+      "epoch": 0.6376306620209059,
+      "grad_norm": 0.7294362846964269,
+      "learning_rate": 9.623762643019927e-06,
+      "loss": 0.6771,
+      "step": 122
+    },
+    {
+      "epoch": 0.6428571428571429,
+      "grad_norm": 0.7545085300162823,
+      "learning_rate": 9.612068988375898e-06,
+      "loss": 0.6715,
+      "step": 123
+    },
+    {
+      "epoch": 0.6480836236933798,
+      "grad_norm": 0.6749790083440244,
+      "learning_rate": 9.60020370905384e-06,
+      "loss": 0.6514,
+      "step": 124
+    },
+    {
+      "epoch": 0.6533101045296167,
+      "grad_norm": 0.804530126401735,
+      "learning_rate": 9.588167246585474e-06,
+      "loss": 0.6459,
+      "step": 125
+    },
+    {
+      "epoch": 0.6585365853658537,
+      "grad_norm": 0.6831220443498838,
+      "learning_rate": 9.575960048872595e-06,
+      "loss": 0.6896,
+      "step": 126
+    },
+    {
+      "epoch": 0.6637630662020906,
+      "grad_norm": 0.5810171537948514,
+      "learning_rate": 9.563582570170418e-06,
+      "loss": 0.6613,
+      "step": 127
+    },
+    {
+      "epoch": 0.6689895470383276,
+      "grad_norm": 0.7304494902535995,
+      "learning_rate": 9.551035271070665e-06,
+      "loss": 0.6516,
+      "step": 128
+    },
+    {
+      "epoch": 0.6742160278745645,
+      "grad_norm": 0.6119862390653155,
+      "learning_rate": 9.538318618484426e-06,
+      "loss": 0.6618,
+      "step": 129
+    },
+    {
+      "epoch": 0.6794425087108014,
+      "grad_norm": 0.6430165539813878,
+      "learning_rate": 9.52543308562479e-06,
+      "loss": 0.6864,
+      "step": 130
+    },
+    {
+      "epoch": 0.6846689895470384,
+      "grad_norm": 0.6194884070250605,
+      "learning_rate": 9.512379151989229e-06,
+      "loss": 0.6784,
+      "step": 131
+    },
+    {
+      "epoch": 0.6898954703832753,
+      "grad_norm": 0.5516355539801995,
+      "learning_rate": 9.499157303341761e-06,
+      "loss": 0.6653,
+      "step": 132
+    },
+    {
+      "epoch": 0.6951219512195121,
+      "grad_norm": 0.622640994125776,
+      "learning_rate": 9.485768031694872e-06,
+      "loss": 0.6731,
+      "step": 133
+    },
+    {
+      "epoch": 0.7003484320557491,
+      "grad_norm": 0.547286279646913,
+      "learning_rate": 9.4722118352912e-06,
+      "loss": 0.6663,
+      "step": 134
+    },
+    {
+      "epoch": 0.705574912891986,
+      "grad_norm": 0.6363064057591127,
+      "learning_rate": 9.458489218585003e-06,
+      "loss": 0.655,
+      "step": 135
+    },
+    {
+      "epoch": 0.710801393728223,
+      "grad_norm": 0.6134952050030124,
+      "learning_rate": 9.44460069222339e-06,
+      "loss": 0.6749,
+      "step": 136
+    },
+    {
+      "epoch": 0.7160278745644599,
+      "grad_norm": 0.5937836538107625,
+      "learning_rate": 9.430546773027302e-06,
+      "loss": 0.6686,
+      "step": 137
+    },
+    {
+      "epoch": 0.7212543554006968,
+      "grad_norm": 0.6644117521702412,
+      "learning_rate": 9.416327983972304e-06,
+      "loss": 0.6594,
+      "step": 138
+    },
+    {
+      "epoch": 0.7264808362369338,
+      "grad_norm": 0.6220644455779103,
+      "learning_rate": 9.401944854169103e-06,
+      "loss": 0.6749,
+      "step": 139
+    },
+    {
+      "epoch": 0.7317073170731707,
+      "grad_norm": 0.6547771063435059,
+      "learning_rate": 9.38739791884387e-06,
+      "loss": 0.6579,
+      "step": 140
+    },
+    {
+      "epoch": 0.7369337979094077,
+      "grad_norm": 0.561284464609663,
+      "learning_rate": 9.372687719318316e-06,
+      "loss": 0.681,
+      "step": 141
+    },
+    {
+      "epoch": 0.7421602787456446,
+      "grad_norm": 0.6038846576238117,
+      "learning_rate": 9.35781480298956e-06,
+      "loss": 0.6666,
+      "step": 142
+    },
+    {
+      "epoch": 0.7473867595818815,
+      "grad_norm": 0.572240770754845,
+      "learning_rate": 9.342779723309746e-06,
+      "loss": 0.6238,
+      "step": 143
+    },
+    {
+      "epoch": 0.7526132404181185,
+      "grad_norm": 0.5890147611510671,
+      "learning_rate": 9.327583039765453e-06,
+      "loss": 0.6637,
+      "step": 144
+    },
+    {
+      "epoch": 0.7578397212543554,
+      "grad_norm": 0.6636633864116531,
+      "learning_rate": 9.31222531785688e-06,
+      "loss": 0.6523,
+      "step": 145
+    },
+    {
+      "epoch": 0.7630662020905923,
+      "grad_norm": 0.6188433137259545,
+      "learning_rate": 9.296707129076794e-06,
+      "loss": 0.6626,
+      "step": 146
+    },
+    {
+      "epoch": 0.7682926829268293,
+      "grad_norm": 0.6163886375278169,
+      "learning_rate": 9.281029050889274e-06,
+      "loss": 0.6786,
+      "step": 147
+    },
+    {
+      "epoch": 0.7735191637630662,
+      "grad_norm": 0.7346550992330867,
+      "learning_rate": 9.26519166670821e-06,
+      "loss": 0.6861,
+      "step": 148
+    },
+    {
+      "epoch": 0.7787456445993032,
+      "grad_norm": 0.5852860888717738,
+      "learning_rate": 9.2491955658756e-06,
+      "loss": 0.6591,
+      "step": 149
+    },
+    {
+      "epoch": 0.7839721254355401,
+      "grad_norm": 0.7028963677577197,
+      "learning_rate": 9.233041343639623e-06,
+      "loss": 0.6881,
+      "step": 150
+    },
+    {
+      "epoch": 0.789198606271777,
+      "grad_norm": 0.8189320800786853,
+      "learning_rate": 9.216729601132481e-06,
+      "loss": 0.6702,
+      "step": 151
+    },
+    {
+      "epoch": 0.794425087108014,
+      "grad_norm": 0.5963261479645,
+      "learning_rate": 9.200260945348034e-06,
+      "loss": 0.6652,
+      "step": 152
+    },
+    {
+      "epoch": 0.7996515679442509,
+      "grad_norm": 0.6745799083758508,
+      "learning_rate": 9.183635989119211e-06,
+      "loss": 0.6569,
+      "step": 153
+    },
+    {
+      "epoch": 0.8048780487804879,
+      "grad_norm": 0.7674431433443214,
+      "learning_rate": 9.166855351095205e-06,
+      "loss": 0.6629,
+      "step": 154
+    },
+    {
+      "epoch": 0.8101045296167247,
+      "grad_norm": 0.5864935078587289,
+      "learning_rate": 9.149919655718453e-06,
+      "loss": 0.6615,
+      "step": 155
+    },
+    {
+      "epoch": 0.8153310104529616,
+      "grad_norm": 0.6569771197412858,
+      "learning_rate": 9.132829533201397e-06,
+      "loss": 0.6321,
+      "step": 156
+    },
+    {
+      "epoch": 0.8205574912891986,
+      "grad_norm": 0.6855458963431387,
+      "learning_rate": 9.115585619503039e-06,
+      "loss": 0.6471,
+      "step": 157
+    },
+    {
+      "epoch": 0.8257839721254355,
+      "grad_norm": 0.5765451182091219,
+      "learning_rate": 9.098188556305262e-06,
+      "loss": 0.6748,
+      "step": 158
+    },
+    {
+      "epoch": 0.8310104529616724,
+      "grad_norm": 0.6567732753183342,
+      "learning_rate": 9.080638990988971e-06,
+      "loss": 0.6564,
+      "step": 159
+    },
+    {
+      "epoch": 0.8362369337979094,
+      "grad_norm": 0.636552981471604,
+      "learning_rate": 9.062937576609983e-06,
+      "loss": 0.654,
+      "step": 160
+    },
+    {
+      "epoch": 0.8414634146341463,
+      "grad_norm": 0.5529034964515249,
+      "learning_rate": 9.045084971874738e-06,
+      "loss": 0.6558,
+      "step": 161
+    },
+    {
+      "epoch": 0.8466898954703833,
+      "grad_norm": 0.5893696671833625,
+      "learning_rate": 9.027081841115784e-06,
+      "loss": 0.6663,
+      "step": 162
+    },
+    {
+      "epoch": 0.8519163763066202,
+      "grad_norm": 0.5667170304864082,
+      "learning_rate": 9.008928854267054e-06,
+      "loss": 0.6443,
+      "step": 163
+    },
+    {
+      "epoch": 0.8571428571428571,
+      "grad_norm": 0.5725785658325226,
+      "learning_rate": 8.99062668683894e-06,
+      "loss": 0.6536,
+      "step": 164
+    },
+    {
+      "epoch": 0.8623693379790941,
+      "grad_norm": 0.6060232996236128,
+      "learning_rate": 8.97217601989315e-06,
+      "loss": 0.6579,
+      "step": 165
+    },
+    {
+      "epoch": 0.867595818815331,
+      "grad_norm": 0.5509869618020639,
+      "learning_rate": 8.95357754001737e-06,
+      "loss": 0.6713,
+      "step": 166
+    },
+    {
+      "epoch": 0.872822299651568,
+      "grad_norm": 0.6324223806192756,
+      "learning_rate": 8.934831939299715e-06,
+      "loss": 0.6696,
+      "step": 167
+    },
+    {
+      "epoch": 0.8780487804878049,
+      "grad_norm": 0.6405745054270299,
+      "learning_rate": 8.91593991530297e-06,
+      "loss": 0.6584,
+      "step": 168
+    },
+    {
+      "epoch": 0.8832752613240418,
+      "grad_norm": 0.6307439054892403,
+      "learning_rate": 8.896902171038629e-06,
+      "loss": 0.6529,
+      "step": 169
+    },
+    {
+      "epoch": 0.8885017421602788,
+      "grad_norm": 0.6611063399387049,
+      "learning_rate": 8.877719414940751e-06,
+      "loss": 0.6816,
+      "step": 170
+    },
+    {
+      "epoch": 0.8937282229965157,
+      "grad_norm": 0.5327557983597507,
+      "learning_rate": 8.85839236083958e-06,
+      "loss": 0.6591,
+      "step": 171
+    },
+    {
+      "epoch": 0.8989547038327527,
+      "grad_norm": 0.6306768954606544,
+      "learning_rate": 8.838921727934992e-06,
+      "loss": 0.6725,
+      "step": 172
+    },
+    {
+      "epoch": 0.9041811846689896,
+      "grad_norm": 0.5906681591023982,
+      "learning_rate": 8.819308240769726e-06,
+      "loss": 0.663,
+      "step": 173
+    },
+    {
+      "epoch": 0.9094076655052264,
+      "grad_norm": 0.6242722153083872,
+      "learning_rate": 8.799552629202424e-06,
+      "loss": 0.6593,
+      "step": 174
+    },
+    {
+      "epoch": 0.9146341463414634,
+      "grad_norm": 0.6206172349107745,
+      "learning_rate": 8.779655628380479e-06,
+      "loss": 0.6366,
+      "step": 175
+    },
+    {
+      "epoch": 0.9198606271777003,
+      "grad_norm": 0.6446754833762357,
+      "learning_rate": 8.759617978712667e-06,
+      "loss": 0.6588,
+      "step": 176
+    },
+    {
+      "epoch": 0.9250871080139372,
+      "grad_norm": 0.6524808843585924,
+      "learning_rate": 8.7394404258416e-06,
+      "loss": 0.6708,
+      "step": 177
+    },
+    {
+      "epoch": 0.9303135888501742,
+      "grad_norm": 0.5695275564016344,
+      "learning_rate": 8.71912372061598e-06,
+      "loss": 0.6581,
+      "step": 178
+    },
+    {
+      "epoch": 0.9355400696864111,
+      "grad_norm": 0.5524125516832253,
+      "learning_rate": 8.69866861906266e-06,
+      "loss": 0.675,
+      "step": 179
+    },
+    {
+      "epoch": 0.9407665505226481,
+      "grad_norm": 0.6301898047784469,
+      "learning_rate": 8.678075882358506e-06,
+      "loss": 0.6315,
+      "step": 180
+    },
+    {
+      "epoch": 0.945993031358885,
+      "grad_norm": 0.5726932230656162,
+      "learning_rate": 8.657346276802071e-06,
+      "loss": 0.6407,
+      "step": 181
+    },
+    {
+      "epoch": 0.9512195121951219,
+      "grad_norm": 0.6267676856772942,
+      "learning_rate": 8.636480573785089e-06,
+      "loss": 0.6665,
+      "step": 182
+    },
+    {
+      "epoch": 0.9564459930313589,
+      "grad_norm": 0.6244142776173546,
+      "learning_rate": 8.615479549763756e-06,
+      "loss": 0.6618,
+      "step": 183
+    },
+    {
+      "epoch": 0.9616724738675958,
+      "grad_norm": 0.6177405231356075,
+      "learning_rate": 8.594343986229854e-06,
+      "loss": 0.6773,
+      "step": 184
+    },
+    {
+      "epoch": 0.9668989547038328,
+      "grad_norm": 0.5988900835082392,
+      "learning_rate": 8.57307466968165e-06,
+      "loss": 0.6429,
+      "step": 185
+    },
+    {
+      "epoch": 0.9721254355400697,
+      "grad_norm": 0.6447854583228403,
+      "learning_rate": 8.551672391594646e-06,
+      "loss": 0.6735,
+      "step": 186
+    },
+    {
+      "epoch": 0.9773519163763066,
+      "grad_norm": 0.5761133455065623,
+      "learning_rate": 8.530137948392113e-06,
+      "loss": 0.6614,
+      "step": 187
+    },
+    {
+      "epoch": 0.9825783972125436,
+      "grad_norm": 0.5943974371608253,
+      "learning_rate": 8.508472141415468e-06,
+      "loss": 0.6539,
+      "step": 188
+    },
+    {
+      "epoch": 0.9878048780487805,
+      "grad_norm": 0.6718313091561632,
+      "learning_rate": 8.48667577689444e-06,
+      "loss": 0.6346,
+      "step": 189
+    },
+    {
+      "epoch": 0.9930313588850174,
+      "grad_norm": 0.5856014973725608,
+      "learning_rate": 8.46474966591708e-06,
+      "loss": 0.6626,
+      "step": 190
+    },
+    {
+      "epoch": 0.9982578397212544,
+      "grad_norm": 0.6113432602736558,
+      "learning_rate": 8.442694624399576e-06,
+      "loss": 0.6604,
+      "step": 191
+    },
+    {
+      "epoch": 1.0034843205574913,
+      "grad_norm": 1.1692152399954596,
+      "learning_rate": 8.420511473055887e-06,
+      "loss": 1.0489,
+      "step": 192
+    },
+    {
+      "epoch": 1.0087108013937283,
+      "grad_norm": 0.6145614135481637,
+      "learning_rate": 8.398201037367202e-06,
+      "loss": 0.6419,
+      "step": 193
+    },
+    {
+      "epoch": 1.0139372822299653,
+      "grad_norm": 0.6268987577515908,
+      "learning_rate": 8.37576414755123e-06,
+      "loss": 0.6137,
+      "step": 194
+    },
+    {
+      "epoch": 1.019163763066202,
+      "grad_norm": 0.6416953015407134,
+      "learning_rate": 8.3532016385313e-06,
+      "loss": 0.6243,
+      "step": 195
+    },
+    {
+      "epoch": 1.024390243902439,
+      "grad_norm": 0.5704666555598586,
+      "learning_rate": 8.330514349905295e-06,
+      "loss": 0.6368,
+      "step": 196
+    },
+    {
+      "epoch": 1.029616724738676,
+      "grad_norm": 0.503382880864932,
+      "learning_rate": 8.307703125914397e-06,
+      "loss": 0.5435,
+      "step": 197
+    },
+    {
+      "epoch": 1.0348432055749128,
+      "grad_norm": 0.5773976279224134,
+      "learning_rate": 8.284768815411693e-06,
+      "loss": 0.6031,
+      "step": 198
+    },
+    {
+      "epoch": 1.0400696864111498,
+      "grad_norm": 0.6463491533346394,
+      "learning_rate": 8.261712271830564e-06,
+      "loss": 0.5943,
+      "step": 199
+    },
+    {
+      "epoch": 1.0452961672473868,
+      "grad_norm": 0.6023840946052278,
+      "learning_rate": 8.23853435315295e-06,
+      "loss": 0.5491,
+      "step": 200
+    },
+    {
+      "epoch": 1.0505226480836236,
+      "grad_norm": 0.5738792581365361,
+      "learning_rate": 8.215235921877403e-06,
+      "loss": 0.5604,
+      "step": 201
+    },
+    {
+      "epoch": 1.0557491289198606,
+      "grad_norm": 0.6156147642415088,
+      "learning_rate": 8.191817844986998e-06,
+      "loss": 0.6434,
+      "step": 202
+    },
+    {
+      "epoch": 1.0609756097560976,
+      "grad_norm": 0.5569699792461741,
+      "learning_rate": 8.168280993917078e-06,
+      "loss": 0.5594,
+      "step": 203
+    },
+    {
+      "epoch": 1.0662020905923344,
+      "grad_norm": 0.5657310358177506,
+      "learning_rate": 8.144626244522812e-06,
+      "loss": 0.5686,
+      "step": 204
+    },
+    {
+      "epoch": 1.0714285714285714,
+      "grad_norm": 0.6672649446625757,
+      "learning_rate": 8.120854477046621e-06,
+      "loss": 0.6803,
+      "step": 205
+    },
+    {
+      "epoch": 1.0766550522648084,
+      "grad_norm": 0.5612168930959153,
+      "learning_rate": 8.096966576085406e-06,
+      "loss": 0.5824,
+      "step": 206
+    },
+    {
+      "epoch": 1.0818815331010452,
+      "grad_norm": 0.6579205469061564,
+      "learning_rate": 8.072963430557636e-06,
+      "loss": 0.6028,
+      "step": 207
+    },
+    {
+      "epoch": 1.0871080139372822,
+      "grad_norm": 0.5598751666256221,
+      "learning_rate": 8.048845933670274e-06,
+      "loss": 0.5777,
+      "step": 208
+    },
+    {
+      "epoch": 1.0923344947735192,
+      "grad_norm": 0.5843330235385993,
+      "learning_rate": 8.024614982885527e-06,
+      "loss": 0.5577,
+      "step": 209
+    },
+    {
+      "epoch": 1.0975609756097562,
+      "grad_norm": 0.601858208353747,
+      "learning_rate": 8.00027147988747e-06,
+      "loss": 0.6143,
+      "step": 210
+    },
+    {
+      "epoch": 1.102787456445993,
+      "grad_norm": 0.627628607597956,
+      "learning_rate": 7.975816330548466e-06,
+      "loss": 0.596,
+      "step": 211
+    },
+    {
+      "epoch": 1.10801393728223,
+      "grad_norm": 0.6241333668783621,
+      "learning_rate": 7.951250444895485e-06,
+      "loss": 0.6494,
+      "step": 212
+    },
+    {
+      "epoch": 1.113240418118467,
+      "grad_norm": 0.5382934099983673,
+      "learning_rate": 7.92657473707621e-06,
+      "loss": 0.5217,
+      "step": 213
+    },
+    {
+      "epoch": 1.1184668989547037,
+      "grad_norm": 0.6953324804344473,
+      "learning_rate": 7.901790125325049e-06,
+      "loss": 0.6941,
+      "step": 214
+    },
+    {
+      "epoch": 1.1236933797909407,
+      "grad_norm": 0.5453365037414081,
+      "learning_rate": 7.876897531928943e-06,
+      "loss": 0.6076,
+      "step": 215
+    },
+    {
+      "epoch": 1.1289198606271778,
+      "grad_norm": 0.5736100460893016,
+      "learning_rate": 7.851897883193057e-06,
+      "loss": 0.5852,
+      "step": 216
+    },
+    {
+      "epoch": 1.1341463414634148,
+      "grad_norm": 0.5648285470025594,
+      "learning_rate": 7.82679210940631e-06,
+      "loss": 0.6033,
+      "step": 217
+    },
+    {
+      "epoch": 1.1393728222996515,
+      "grad_norm": 0.5867845846947239,
+      "learning_rate": 7.801581144806752e-06,
+      "loss": 0.5587,
+      "step": 218
+    },
+    {
+      "epoch": 1.1445993031358885,
+      "grad_norm": 0.5513023482330085,
+      "learning_rate": 7.7762659275468e-06,
+      "loss": 0.618,
+      "step": 219
+    },
+    {
+      "epoch": 1.1498257839721253,
+      "grad_norm": 0.6153474010859067,
+      "learning_rate": 7.750847399658336e-06,
+      "loss": 0.6282,
+      "step": 220
+    },
+    {
+      "epoch": 1.1550522648083623,
+      "grad_norm": 0.4877966585426567,
+      "learning_rate": 7.725326507017644e-06,
+      "loss": 0.5451,
+      "step": 221
+    },
+    {
+      "epoch": 1.1602787456445993,
+      "grad_norm": 0.5463512600607151,
+      "learning_rate": 7.699704199310204e-06,
+      "loss": 0.5972,
+      "step": 222
+    },
+    {
+      "epoch": 1.1655052264808363,
+      "grad_norm": 0.6352304968816307,
+      "learning_rate": 7.673981429995372e-06,
+      "loss": 0.6114,
+      "step": 223
+    },
+    {
+      "epoch": 1.170731707317073,
+      "grad_norm": 0.510832677269982,
+      "learning_rate": 7.648159156270884e-06,
+      "loss": 0.6268,
+      "step": 224
+    },
+    {
+      "epoch": 1.17595818815331,
+      "grad_norm": 0.5814622275101878,
+      "learning_rate": 7.622238339037248e-06,
+      "loss": 0.6141,
+      "step": 225
+    },
+    {
+      "epoch": 1.181184668989547,
+      "grad_norm": 0.5818228914475463,
+      "learning_rate": 7.596219942861971e-06,
+      "loss": 0.5655,
+      "step": 226
+    },
+    {
+      "epoch": 1.1864111498257839,
+      "grad_norm": 0.5842183364885295,
+      "learning_rate": 7.570104935943685e-06,
+      "loss": 0.5983,
+      "step": 227
+    },
+    {
+      "epoch": 1.1916376306620209,
+      "grad_norm": 0.5752307733655048,
+      "learning_rate": 7.5438942900761035e-06,
+      "loss": 0.6072,
+      "step": 228
+    },
+    {
+      "epoch": 1.1968641114982579,
+      "grad_norm": 0.585743260967966,
+      "learning_rate": 7.517588980611864e-06,
+      "loss": 0.6306,
+      "step": 229
+    },
+    {
+      "epoch": 1.202090592334495,
+      "grad_norm": 0.48400699770777905,
+      "learning_rate": 7.491189986426236e-06,
+      "loss": 0.523,
+      "step": 230
+    },
+    {
+      "epoch": 1.2073170731707317,
+      "grad_norm": 0.6000420590770226,
+      "learning_rate": 7.464698289880689e-06,
+      "loss": 0.6248,
+      "step": 231
+    },
+    {
+      "epoch": 1.2125435540069687,
+      "grad_norm": 0.566825613388456,
+      "learning_rate": 7.438114876786344e-06,
+      "loss": 0.5892,
+      "step": 232
+    },
+    {
+      "epoch": 1.2177700348432055,
+      "grad_norm": 0.5942135634459118,
+      "learning_rate": 7.411440736367281e-06,
+      "loss": 0.604,
+      "step": 233
+    },
+    {
+      "epoch": 1.2229965156794425,
+      "grad_norm": 0.5308639194617243,
+      "learning_rate": 7.384676861223738e-06,
+      "loss": 0.5938,
+      "step": 234
+    },
+    {
+      "epoch": 1.2282229965156795,
+      "grad_norm": 0.5233297325602637,
+      "learning_rate": 7.3578242472951635e-06,
+      "loss": 0.5606,
+      "step": 235
+    },
+    {
+      "epoch": 1.2334494773519165,
+      "grad_norm": 0.5506498491313724,
+      "learning_rate": 7.330883893823164e-06,
+      "loss": 0.6321,
+      "step": 236
+    },
+    {
+      "epoch": 1.2386759581881532,
+      "grad_norm": 0.5032393653634389,
+      "learning_rate": 7.303856803314313e-06,
+      "loss": 0.5772,
+      "step": 237
+    },
+    {
+      "epoch": 1.2439024390243902,
+      "grad_norm": 0.5516215897044785,
+      "learning_rate": 7.276743981502856e-06,
+      "loss": 0.6249,
+      "step": 238
+    },
+    {
+      "epoch": 1.2491289198606272,
+      "grad_norm": 0.5922013934962067,
+      "learning_rate": 7.249546437313273e-06,
+      "loss": 0.5742,
+      "step": 239
+    },
+    {
+      "epoch": 1.254355400696864,
+      "grad_norm": 0.563594800466747,
+      "learning_rate": 7.22226518282274e-06,
+      "loss": 0.5931,
+      "step": 240
+    },
+    {
+      "epoch": 1.259581881533101,
+      "grad_norm": 0.5865049138488002,
+      "learning_rate": 7.194901233223471e-06,
+      "loss": 0.6026,
+      "step": 241
+    },
+    {
+      "epoch": 1.264808362369338,
+      "grad_norm": 0.6049741584214127,
+      "learning_rate": 7.167455606784936e-06,
+      "loss": 0.5973,
+      "step": 242
+    },
+    {
+      "epoch": 1.270034843205575,
+      "grad_norm": 0.600516789714151,
+      "learning_rate": 7.139929324815965e-06,
+      "loss": 0.639,
+      "step": 243
+    },
+    {
+      "epoch": 1.2752613240418118,
+      "grad_norm": 0.5799862799629144,
+      "learning_rate": 7.112323411626756e-06,
+      "loss": 0.5975,
+      "step": 244
+    },
+    {
+      "epoch": 1.2804878048780488,
+      "grad_norm": 0.608405822867294,
+      "learning_rate": 7.084638894490744e-06,
+      "loss": 0.5932,
+      "step": 245
+    },
+    {
+      "epoch": 1.2857142857142856,
+      "grad_norm": 0.6086512791994181,
+      "learning_rate": 7.056876803606384e-06,
+      "loss": 0.6211,
+      "step": 246
+    },
+    {
+      "epoch": 1.2909407665505226,
+      "grad_norm": 0.6114934942027832,
+      "learning_rate": 7.029038172058809e-06,
+      "loss": 0.5816,
+      "step": 247
+    },
+    {
+      "epoch": 1.2961672473867596,
+      "grad_norm": 0.5531245874896416,
+      "learning_rate": 7.00112403578139e-06,
+      "loss": 0.5793,
+      "step": 248
+    },
+    {
+      "epoch": 1.3013937282229966,
+      "grad_norm": 0.590859726592878,
+      "learning_rate": 6.9731354335171885e-06,
+      "loss": 0.5969,
+      "step": 249
+    },
+    {
+      "epoch": 1.3066202090592334,
+      "grad_norm": 0.6199140175680848,
+      "learning_rate": 6.945073406780296e-06,
+      "loss": 0.6263,
+      "step": 250
+    },
+    {
+      "epoch": 1.3118466898954704,
+      "grad_norm": 0.49378151455087976,
+      "learning_rate": 6.916938999817085e-06,
+      "loss": 0.6258,
+      "step": 251
+    },
+    {
+      "epoch": 1.3170731707317074,
+      "grad_norm": 0.5481220680272908,
+      "learning_rate": 6.888733259567343e-06,
+      "loss": 0.5697,
+      "step": 252
+    },
+    {
+      "epoch": 1.3222996515679442,
+      "grad_norm": 0.5601237104950061,
+      "learning_rate": 6.860457235625322e-06,
+      "loss": 0.6192,
+      "step": 253
+    },
+    {
+      "epoch": 1.3275261324041812,
+      "grad_norm": 0.5562695792344781,
+      "learning_rate": 6.832111980200672e-06,
+      "loss": 0.5595,
+      "step": 254
+    },
+    {
+      "epoch": 1.3327526132404182,
+      "grad_norm": 0.6375068604321416,
+      "learning_rate": 6.803698548079294e-06,
+      "loss": 0.6001,
+      "step": 255
+    },
+    {
+      "epoch": 1.3379790940766552,
+      "grad_norm": 0.5443785529911263,
+      "learning_rate": 6.775217996584082e-06,
+      "loss": 0.5915,
+      "step": 256
+    },
+    {
+      "epoch": 1.343205574912892,
+      "grad_norm": 0.5408122608427677,
+      "learning_rate": 6.746671385535586e-06,
+      "loss": 0.5953,
+      "step": 257
+    },
+    {
+      "epoch": 1.348432055749129,
+      "grad_norm": 0.5539163863396951,
+      "learning_rate": 6.7180597772125665e-06,
+      "loss": 0.6388,
+      "step": 258
+    },
+    {
+      "epoch": 1.3536585365853657,
+      "grad_norm": 0.5362462583433603,
+      "learning_rate": 6.689384236312465e-06,
+      "loss": 0.5838,
+      "step": 259
+    },
+    {
+      "epoch": 1.3588850174216027,
+      "grad_norm": 0.541689780691623,
+      "learning_rate": 6.660645829911794e-06,
+      "loss": 0.6318,
+      "step": 260
+    },
+    {
+      "epoch": 1.3641114982578397,
+      "grad_norm": 0.5106593953968394,
+      "learning_rate": 6.631845627426418e-06,
+      "loss": 0.5425,
+      "step": 261
+    },
+    {
+      "epoch": 1.3693379790940767,
+      "grad_norm": 0.49403851854670217,
+      "learning_rate": 6.602984700571758e-06,
+      "loss": 0.5305,
+      "step": 262
+    },
+    {
+      "epoch": 1.3745644599303135,
+      "grad_norm": 0.5967203227127262,
+      "learning_rate": 6.574064123322925e-06,
+      "loss": 0.6405,
+      "step": 263
+    },
+    {
+      "epoch": 1.3797909407665505,
+      "grad_norm": 0.5601246715123179,
+      "learning_rate": 6.545084971874738e-06,
+      "loss": 0.5802,
+      "step": 264
+    },
+    {
+      "epoch": 1.3850174216027875,
+      "grad_norm": 0.5342203081001657,
+      "learning_rate": 6.516048324601685e-06,
+      "loss": 0.6166,
+      "step": 265
+    },
+    {
+      "epoch": 1.3902439024390243,
+      "grad_norm": 0.5175415102386944,
+      "learning_rate": 6.486955262017794e-06,
+      "loss": 0.6049,
+      "step": 266
+    },
+    {
+      "epoch": 1.3954703832752613,
+      "grad_norm": 0.603808078724887,
+      "learning_rate": 6.457806866736424e-06,
+      "loss": 0.615,
+      "step": 267
+    },
+    {
+      "epoch": 1.4006968641114983,
+      "grad_norm": 0.5271092125681837,
+      "learning_rate": 6.42860422342998e-06,
+      "loss": 0.522,
+      "step": 268
+    },
+    {
+      "epoch": 1.4059233449477353,
+      "grad_norm": 0.5871002546738227,
+      "learning_rate": 6.399348418789545e-06,
+      "loss": 0.6314,
+      "step": 269
+    },
+    {
+      "epoch": 1.411149825783972,
+      "grad_norm": 0.5188920938550411,
+      "learning_rate": 6.37004054148445e-06,
+      "loss": 0.5822,
+      "step": 270
+    },
+    {
+      "epoch": 1.416376306620209,
+      "grad_norm": 0.5152138837098115,
+      "learning_rate": 6.3406816821217554e-06,
+      "loss": 0.6093,
+      "step": 271
+    },
+    {
+      "epoch": 1.4216027874564459,
+      "grad_norm": 0.513754045434498,
+      "learning_rate": 6.311272933205672e-06,
+      "loss": 0.5832,
+      "step": 272
+    },
+    {
+      "epoch": 1.4268292682926829,
+      "grad_norm": 0.4912944860658243,
+      "learning_rate": 6.281815389096903e-06,
+      "loss": 0.529,
+      "step": 273
+    },
+    {
+      "epoch": 1.4320557491289199,
+      "grad_norm": 0.5427353767757749,
+      "learning_rate": 6.2523101459719204e-06,
+      "loss": 0.6526,
+      "step": 274
+    },
+    {
+      "epoch": 1.4372822299651569,
+      "grad_norm": 0.5535697011306765,
+      "learning_rate": 6.222758301782183e-06,
+      "loss": 0.6282,
+      "step": 275
+    },
+    {
+      "epoch": 1.4425087108013936,
+      "grad_norm": 0.5162479310938117,
+      "learning_rate": 6.193160956213262e-06,
+      "loss": 0.5903,
+      "step": 276
+    },
+    {
+      "epoch": 1.4477351916376306,
+      "grad_norm": 0.5604814223022786,
+      "learning_rate": 6.163519210643939e-06,
+      "loss": 0.6083,
+      "step": 277
+    },
+    {
+      "epoch": 1.4529616724738676,
+      "grad_norm": 0.5139952218594641,
+      "learning_rate": 6.133834168105206e-06,
+      "loss": 0.5271,
+      "step": 278
+    },
+    {
+      "epoch": 1.4581881533101044,
+      "grad_norm": 0.6229450021349342,
+      "learning_rate": 6.104106933239227e-06,
+      "loss": 0.5636,
+      "step": 279
+    },
+    {
+      "epoch": 1.4634146341463414,
+      "grad_norm": 0.5408488715069601,
+      "learning_rate": 6.07433861225823e-06,
+      "loss": 0.6205,
+      "step": 280
+    },
+    {
+      "epoch": 1.4686411149825784,
+      "grad_norm": 0.49674309270840106,
+      "learning_rate": 6.044530312903343e-06,
+      "loss": 0.5794,
+      "step": 281
+    },
+    {
+      "epoch": 1.4738675958188154,
+      "grad_norm": 0.582431202829567,
+      "learning_rate": 6.014683144403375e-06,
+      "loss": 0.6167,
+      "step": 282
+    },
+    {
+      "epoch": 1.4790940766550522,
+      "grad_norm": 0.573524016791654,
+      "learning_rate": 5.9847982174335314e-06,
+      "loss": 0.6008,
+      "step": 283
+    },
+    {
+      "epoch": 1.4843205574912892,
+      "grad_norm": 0.5130183259716176,
+      "learning_rate": 5.954876644074092e-06,
+      "loss": 0.6057,
+      "step": 284
+    },
+    {
+      "epoch": 1.489547038327526,
+      "grad_norm": 0.547570250155865,
+      "learning_rate": 5.924919537769025e-06,
+      "loss": 0.5872,
+      "step": 285
+    },
+    {
+      "epoch": 1.494773519163763,
+      "grad_norm": 0.5786698298674792,
+      "learning_rate": 5.894928013284551e-06,
+      "loss": 0.549,
+      "step": 286
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.5765527388977582,
+      "learning_rate": 5.8649031866676685e-06,
+      "loss": 0.6325,
+      "step": 287
+    },
+    {
+      "epoch": 1.505226480836237,
+      "grad_norm": 0.4800435309675829,
+      "learning_rate": 5.834846175204612e-06,
+      "loss": 0.5278,
+      "step": 288
+    },
+    {
+      "epoch": 1.510452961672474,
+      "grad_norm": 0.6218236849239587,
+      "learning_rate": 5.804758097379281e-06,
+      "loss": 0.6634,
+      "step": 289
+    },
+    {
+      "epoch": 1.5156794425087108,
+      "grad_norm": 0.5073235422648865,
+      "learning_rate": 5.774640072831622e-06,
+      "loss": 0.5773,
+      "step": 290
+    },
+    {
+      "epoch": 1.5209059233449478,
+      "grad_norm": 0.5687256206289278,
+      "learning_rate": 5.74449322231596e-06,
+      "loss": 0.6216,
+      "step": 291
+    },
+    {
+      "epoch": 1.5261324041811846,
+      "grad_norm": 0.495289281164082,
+      "learning_rate": 5.714318667659295e-06,
+      "loss": 0.5697,
+      "step": 292
+    },
+    {
+      "epoch": 1.5313588850174216,
+      "grad_norm": 0.5071599473157437,
+      "learning_rate": 5.684117531719552e-06,
+      "loss": 0.5902,
+      "step": 293
+    },
+    {
+      "epoch": 1.5365853658536586,
+      "grad_norm": 0.5465492422582399,
+      "learning_rate": 5.653890938343806e-06,
+      "loss": 0.6084,
+      "step": 294
+    },
+    {
+      "epoch": 1.5418118466898956,
+      "grad_norm": 0.5794183886322944,
+      "learning_rate": 5.623640012326455e-06,
+      "loss": 0.6074,
+      "step": 295
+    },
+    {
+      "epoch": 1.5470383275261324,
+      "grad_norm": 0.5077288815232027,
+      "learning_rate": 5.593365879367361e-06,
+      "loss": 0.559,
+      "step": 296
+    },
+    {
+      "epoch": 1.5522648083623694,
+      "grad_norm": 0.535785650486145,
+      "learning_rate": 5.56306966602997e-06,
+      "loss": 0.6259,
+      "step": 297
+    },
+    {
+      "epoch": 1.5574912891986061,
+      "grad_norm": 0.5574339114144872,
+      "learning_rate": 5.532752499699381e-06,
+      "loss": 0.6274,
+      "step": 298
+    },
+    {
+      "epoch": 1.5627177700348431,
+      "grad_norm": 0.4864171169075875,
+      "learning_rate": 5.502415508540401e-06,
+      "loss": 0.5689,
+      "step": 299
+    },
+    {
+      "epoch": 1.5679442508710801,
+      "grad_norm": 0.5395355580960868,
+      "learning_rate": 5.472059821455554e-06,
+      "loss": 0.615,
+      "step": 300
+    },
+    {
+      "epoch": 1.5731707317073171,
+      "grad_norm": 0.5367333644494985,
+      "learning_rate": 5.441686568043086e-06,
+      "loss": 0.6209,
+      "step": 301
+    },
+    {
+      "epoch": 1.5783972125435541,
+      "grad_norm": 0.5629949580706008,
+      "learning_rate": 5.411296878554918e-06,
+      "loss": 0.5778,
+      "step": 302
+    },
+    {
+      "epoch": 1.583623693379791,
+      "grad_norm": 0.5279025042856722,
+      "learning_rate": 5.380891883854591e-06,
+      "loss": 0.5891,
+      "step": 303
+    },
+    {
+      "epoch": 1.588850174216028,
+      "grad_norm": 0.529056626502548,
+      "learning_rate": 5.3504727153751865e-06,
+      "loss": 0.5781,
+      "step": 304
+    },
+    {
+      "epoch": 1.5940766550522647,
+      "grad_norm": 0.5143855575664987,
+      "learning_rate": 5.320040505077222e-06,
+      "loss": 0.6108,
+      "step": 305
+    },
+    {
+      "epoch": 1.5993031358885017,
+      "grad_norm": 0.5496043322415672,
+      "learning_rate": 5.289596385406527e-06,
+      "loss": 0.5925,
+      "step": 306
+    },
+    {
+      "epoch": 1.6045296167247387,
+      "grad_norm": 0.45788921944124966,
+      "learning_rate": 5.259141489252104e-06,
+      "loss": 0.5211,
+      "step": 307
+    },
+    {
+      "epoch": 1.6097560975609757,
+      "grad_norm": 0.5610785336406,
+      "learning_rate": 5.228676949903974e-06,
+      "loss": 0.6835,
+      "step": 308
+    },
+    {
+      "epoch": 1.6149825783972127,
+      "grad_norm": 0.5041137193985311,
+      "learning_rate": 5.198203901010993e-06,
+      "loss": 0.5614,
+      "step": 309
+    },
+    {
+      "epoch": 1.6202090592334495,
+      "grad_norm": 0.6140523018892134,
+      "learning_rate": 5.167723476538683e-06,
+      "loss": 0.6063,
+      "step": 310
+    },
+    {
+      "epoch": 1.6254355400696863,
+      "grad_norm": 0.47853290908855395,
+      "learning_rate": 5.137236810727025e-06,
+      "loss": 0.5358,
+      "step": 311
+    },
+    {
+      "epoch": 1.6306620209059233,
+      "grad_norm": 0.5819289571367962,
+      "learning_rate": 5.106745038048251e-06,
+      "loss": 0.6031,
+      "step": 312
+    },
+    {
+      "epoch": 1.6358885017421603,
+      "grad_norm": 0.5268059390545785,
+      "learning_rate": 5.07624929316463e-06,
+      "loss": 0.5926,
+      "step": 313
+    },
+    {
+      "epoch": 1.6411149825783973,
+      "grad_norm": 0.5158701213233892,
+      "learning_rate": 5.045750710886248e-06,
+      "loss": 0.5909,
+      "step": 314
+    },
+    {
+      "epoch": 1.6463414634146343,
+      "grad_norm": 0.5151166538253583,
+      "learning_rate": 5.015250426128772e-06,
+      "loss": 0.5394,
+      "step": 315
+    },
+    {
+      "epoch": 1.651567944250871,
+      "grad_norm": 0.5538710708429964,
+      "learning_rate": 4.984749573871228e-06,
+      "loss": 0.6349,
+      "step": 316
+    },
+    {
+      "epoch": 1.656794425087108,
+      "grad_norm": 0.5631924824167399,
+      "learning_rate": 4.954249289113753e-06,
+      "loss": 0.6271,
+      "step": 317
+    },
+    {
+      "epoch": 1.6620209059233448,
+      "grad_norm": 0.5557831986406585,
+      "learning_rate": 4.923750706835371e-06,
+      "loss": 0.6067,
+      "step": 318
+    },
+    {
+      "epoch": 1.6672473867595818,
+      "grad_norm": 0.5129556814943586,
+      "learning_rate": 4.8932549619517514e-06,
+      "loss": 0.5688,
+      "step": 319
+    },
+    {
+      "epoch": 1.6724738675958188,
+      "grad_norm": 0.510154514617586,
+      "learning_rate": 4.862763189272976e-06,
+      "loss": 0.5993,
+      "step": 320
+    },
+    {
+      "epoch": 1.6777003484320558,
+      "grad_norm": 0.5144230379042817,
+      "learning_rate": 4.832276523461317e-06,
+      "loss": 0.5251,
+      "step": 321
+    },
+    {
+      "epoch": 1.6829268292682928,
+      "grad_norm": 0.6787456643186844,
+      "learning_rate": 4.801796098989009e-06,
+      "loss": 0.6171,
+      "step": 322
+    },
+    {
+      "epoch": 1.6881533101045296,
+      "grad_norm": 0.5288392604393932,
+      "learning_rate": 4.771323050096028e-06,
+      "loss": 0.5868,
+      "step": 323
+    },
+    {
+      "epoch": 1.6933797909407664,
+      "grad_norm": 0.5337562030536104,
+      "learning_rate": 4.740858510747897e-06,
+      "loss": 0.6094,
+      "step": 324
+    },
+    {
+      "epoch": 1.6986062717770034,
+      "grad_norm": 0.5526357082902859,
+      "learning_rate": 4.710403614593475e-06,
+      "loss": 0.551,
+      "step": 325
+    },
+    {
+      "epoch": 1.7038327526132404,
+      "grad_norm": 0.5667609879832234,
+      "learning_rate": 4.679959494922779e-06,
+      "loss": 0.6207,
+      "step": 326
+    },
+    {
+      "epoch": 1.7090592334494774,
+      "grad_norm": 0.5052465715351889,
+      "learning_rate": 4.649527284624814e-06,
+      "loss": 0.5813,
+      "step": 327
+    },
+    {
+      "epoch": 1.7142857142857144,
+      "grad_norm": 0.6012385103563326,
+      "learning_rate": 4.619108116145411e-06,
+      "loss": 0.5997,
+      "step": 328
+    },
+    {
+      "epoch": 1.7195121951219512,
+      "grad_norm": 0.49875874383318625,
+      "learning_rate": 4.588703121445084e-06,
+      "loss": 0.6162,
+      "step": 329
+    },
+    {
+      "epoch": 1.7247386759581882,
+      "grad_norm": 0.5203036540095274,
+      "learning_rate": 4.558313431956914e-06,
+      "loss": 0.5616,
+      "step": 330
+    },
+    {
+      "epoch": 1.729965156794425,
+      "grad_norm": 0.6015714400165652,
+      "learning_rate": 4.527940178544446e-06,
+      "loss": 0.6088,
+      "step": 331
+    },
+    {
+      "epoch": 1.735191637630662,
+      "grad_norm": 0.4998018729629211,
+      "learning_rate": 4.4975844914596015e-06,
+      "loss": 0.5808,
+      "step": 332
+    },
+    {
+      "epoch": 1.740418118466899,
+      "grad_norm": 0.5755597332027752,
+      "learning_rate": 4.467247500300621e-06,
+      "loss": 0.6187,
+      "step": 333
+    },
+    {
+      "epoch": 1.745644599303136,
+      "grad_norm": 0.4930485861382567,
+      "learning_rate": 4.436930333970033e-06,
+      "loss": 0.5532,
+      "step": 334
+    },
+    {
+      "epoch": 1.750871080139373,
+      "grad_norm": 0.5496423220513306,
+      "learning_rate": 4.40663412063264e-06,
+      "loss": 0.6331,
+      "step": 335
+    },
+    {
+      "epoch": 1.7560975609756098,
+      "grad_norm": 0.5442491839702545,
+      "learning_rate": 4.376359987673547e-06,
+      "loss": 0.5759,
+      "step": 336
+    },
+    {
+      "epoch": 1.7613240418118465,
+      "grad_norm": 0.5248586938914367,
+      "learning_rate": 4.346109061656196e-06,
+      "loss": 0.5859,
+      "step": 337
+    },
+    {
+      "epoch": 1.7665505226480835,
+      "grad_norm": 0.5323054479309529,
+      "learning_rate": 4.31588246828045e-06,
+      "loss": 0.6084,
+      "step": 338
+    },
+    {
+      "epoch": 1.7717770034843205,
+      "grad_norm": 0.47627566350608475,
+      "learning_rate": 4.285681332340708e-06,
+      "loss": 0.5706,
+      "step": 339
+    },
+    {
+      "epoch": 1.7770034843205575,
+      "grad_norm": 0.5358815728520783,
+      "learning_rate": 4.255506777684041e-06,
+      "loss": 0.5904,
+      "step": 340
+    },
+    {
+      "epoch": 1.7822299651567945,
+      "grad_norm": 0.48425524407967563,
+      "learning_rate": 4.225359927168379e-06,
+      "loss": 0.5667,
+      "step": 341
+    },
+    {
+      "epoch": 1.7874564459930313,
+      "grad_norm": 0.5596432588171234,
+      "learning_rate": 4.195241902620721e-06,
+      "loss": 0.6049,
+      "step": 342
+    },
+    {
+      "epoch": 1.7926829268292683,
+      "grad_norm": 0.5440361119871323,
+      "learning_rate": 4.165153824795391e-06,
+      "loss": 0.6078,
+      "step": 343
+    },
+    {
+      "epoch": 1.797909407665505,
+      "grad_norm": 0.4826979180979176,
+      "learning_rate": 4.135096813332333e-06,
+      "loss": 0.588,
+      "step": 344
+    },
+    {
+      "epoch": 1.8031358885017421,
+      "grad_norm": 0.5397256499018573,
+      "learning_rate": 4.105071986715449e-06,
+      "loss": 0.6015,
+      "step": 345
+    },
+    {
+      "epoch": 1.8083623693379791,
+      "grad_norm": 0.4510163396115861,
+      "learning_rate": 4.075080462230977e-06,
+      "loss": 0.5697,
+      "step": 346
+    },
+    {
+      "epoch": 1.8135888501742161,
+      "grad_norm": 0.5218309859466103,
+      "learning_rate": 4.04512335592591e-06,
+      "loss": 0.5698,
+      "step": 347
+    },
+    {
+      "epoch": 1.8188153310104531,
+      "grad_norm": 0.5100313135625644,
+      "learning_rate": 4.015201782566471e-06,
+      "loss": 0.5787,
+      "step": 348
+    },
+    {
+      "epoch": 1.82404181184669,
+      "grad_norm": 0.498790862195705,
+      "learning_rate": 3.985316855596627e-06,
+      "loss": 0.6,
+      "step": 349
+    },
+    {
+      "epoch": 1.8292682926829267,
+      "grad_norm": 0.4754481818294577,
+      "learning_rate": 3.955469687096657e-06,
+      "loss": 0.5759,
+      "step": 350
+    },
+    {
+      "epoch": 1.8344947735191637,
+      "grad_norm": 0.5366460340127677,
+      "learning_rate": 3.9256613877417715e-06,
+      "loss": 0.6505,
+      "step": 351
+    },
+    {
+      "epoch": 1.8397212543554007,
+      "grad_norm": 0.48550128940070797,
+      "learning_rate": 3.895893066760774e-06,
+      "loss": 0.5581,
+      "step": 352
+    },
+    {
+      "epoch": 1.8449477351916377,
+      "grad_norm": 0.4645102288396099,
+      "learning_rate": 3.866165831894796e-06,
+      "loss": 0.6126,
+      "step": 353
+    },
+    {
+      "epoch": 1.8501742160278747,
+      "grad_norm": 0.5483359900678575,
+      "learning_rate": 3.836480789356063e-06,
+      "loss": 0.6353,
+      "step": 354
+    },
+    {
+      "epoch": 1.8554006968641115,
+      "grad_norm": 0.4726139182789402,
+      "learning_rate": 3.806839043786738e-06,
+      "loss": 0.5625,
+      "step": 355
+    },
+    {
+      "epoch": 1.8606271777003485,
+      "grad_norm": 0.46997478052113384,
+      "learning_rate": 3.777241698217819e-06,
+      "loss": 0.5839,
+      "step": 356
+    },
+    {
+      "epoch": 1.8658536585365852,
+      "grad_norm": 0.487480595451344,
+      "learning_rate": 3.747689854028081e-06,
+      "loss": 0.596,
+      "step": 357
+    },
+    {
+      "epoch": 1.8710801393728222,
+      "grad_norm": 0.4691428215776054,
+      "learning_rate": 3.7181846109031007e-06,
+      "loss": 0.5786,
+      "step": 358
+    },
+    {
+      "epoch": 1.8763066202090593,
+      "grad_norm": 0.49771800498221574,
+      "learning_rate": 3.6887270667943285e-06,
+      "loss": 0.5763,
+      "step": 359
+    },
+    {
+      "epoch": 1.8815331010452963,
+      "grad_norm": 0.48008730260324617,
+      "learning_rate": 3.6593183178782454e-06,
+      "loss": 0.6022,
+      "step": 360
+    },
+    {
+      "epoch": 1.8867595818815333,
+      "grad_norm": 0.49688177136494294,
+      "learning_rate": 3.6299594585155513e-06,
+      "loss": 0.5949,
+      "step": 361
+    },
+    {
+      "epoch": 1.89198606271777,
+      "grad_norm": 0.48272168297238965,
+      "learning_rate": 3.600651581210457e-06,
+      "loss": 0.5868,
+      "step": 362
+    },
+    {
+      "epoch": 1.8972125435540068,
+      "grad_norm": 0.4282471284231044,
+      "learning_rate": 3.5713957765700224e-06,
+      "loss": 0.5522,
+      "step": 363
+    },
+    {
+      "epoch": 1.9024390243902438,
+      "grad_norm": 0.49857423331197087,
+      "learning_rate": 3.542193133263576e-06,
+      "loss": 0.6013,
+      "step": 364
+    },
+    {
+      "epoch": 1.9076655052264808,
+      "grad_norm": 0.5051671065894443,
+      "learning_rate": 3.5130447379822076e-06,
+      "loss": 0.5448,
+      "step": 365
+    },
+    {
+      "epoch": 1.9128919860627178,
+      "grad_norm": 0.49126598359477613,
+      "learning_rate": 3.483951675398316e-06,
+      "loss": 0.6007,
+      "step": 366
+    },
+    {
+      "epoch": 1.9181184668989548,
+      "grad_norm": 0.5076571957591817,
+      "learning_rate": 3.4549150281252635e-06,
+      "loss": 0.5685,
+      "step": 367
+    },
+    {
+      "epoch": 1.9233449477351916,
+      "grad_norm": 0.512416353580941,
+      "learning_rate": 3.425935876677077e-06,
+      "loss": 0.5857,
+      "step": 368
+    },
+    {
+      "epoch": 1.9285714285714286,
+      "grad_norm": 0.5368818226681046,
+      "learning_rate": 3.397015299428242e-06,
+      "loss": 0.5809,
+      "step": 369
+    },
+    {
+      "epoch": 1.9337979094076654,
+      "grad_norm": 0.5120830184486495,
+      "learning_rate": 3.3681543725735843e-06,
+      "loss": 0.6067,
+      "step": 370
+    },
+    {
+      "epoch": 1.9390243902439024,
+      "grad_norm": 0.4785710907485341,
+      "learning_rate": 3.339354170088207e-06,
+      "loss": 0.5398,
+      "step": 371
+    },
+    {
+      "epoch": 1.9442508710801394,
+      "grad_norm": 0.45732087968552365,
+      "learning_rate": 3.3106157636875356e-06,
+      "loss": 0.5702,
+      "step": 372
+    },
+    {
+      "epoch": 1.9494773519163764,
+      "grad_norm": 0.49177652612113,
+      "learning_rate": 3.2819402227874364e-06,
+      "loss": 0.5925,
+      "step": 373
+    },
+    {
+      "epoch": 1.9547038327526134,
+      "grad_norm": 0.522898733840705,
+      "learning_rate": 3.253328614464414e-06,
+      "loss": 0.6521,
+      "step": 374
+    },
+    {
+      "epoch": 1.9599303135888502,
+      "grad_norm": 0.4962984913773303,
+      "learning_rate": 3.2247820034159182e-06,
+      "loss": 0.5637,
+      "step": 375
+    },
+    {
+      "epoch": 1.965156794425087,
+      "grad_norm": 0.48961258731018903,
+      "learning_rate": 3.196301451920708e-06,
+      "loss": 0.6078,
+      "step": 376
+    },
+    {
+      "epoch": 1.970383275261324,
+      "grad_norm": 0.516946862548738,
+      "learning_rate": 3.16788801979933e-06,
+      "loss": 0.614,
+      "step": 377
+    },
+    {
+      "epoch": 1.975609756097561,
+      "grad_norm": 0.49422567410473595,
+      "learning_rate": 3.1395427643746802e-06,
+      "loss": 0.5407,
+      "step": 378
+    },
+    {
+      "epoch": 1.980836236933798,
+      "grad_norm": 0.5096634642555786,
+      "learning_rate": 3.111266740432658e-06,
+      "loss": 0.5792,
+      "step": 379
+    },
+    {
+      "epoch": 1.986062717770035,
+      "grad_norm": 0.5002493838245908,
+      "learning_rate": 3.0830610001829173e-06,
+      "loss": 0.5552,
+      "step": 380
+    },
+    {
+      "epoch": 1.9912891986062717,
+      "grad_norm": 0.4634141103131062,
+      "learning_rate": 3.0549265932197055e-06,
+      "loss": 0.5474,
+      "step": 381
+    },
+    {
+      "epoch": 1.9965156794425087,
+      "grad_norm": 0.49968954452588915,
+      "learning_rate": 3.026864566482813e-06,
+      "loss": 0.5926,
+      "step": 382
+    },
+    {
+      "epoch": 2.0017421602787455,
+      "grad_norm": 0.9566301326202498,
+      "learning_rate": 2.99887596421861e-06,
+      "loss": 0.9583,
+      "step": 383
+    },
+    {
+      "epoch": 2.0069686411149825,
+      "grad_norm": 0.4819191529898956,
+      "learning_rate": 2.9709618279411922e-06,
+      "loss": 0.4765,
+      "step": 384
+    },
+    {
+      "epoch": 2.0121951219512195,
+      "grad_norm": 0.5177350268902499,
+      "learning_rate": 2.9431231963936176e-06,
+      "loss": 0.5513,
+      "step": 385
+    },
+    {
+      "epoch": 2.0174216027874565,
+      "grad_norm": 0.49571084166916257,
+      "learning_rate": 2.9153611055092586e-06,
+      "loss": 0.5568,
+      "step": 386
+    },
+    {
+      "epoch": 2.0226480836236935,
+      "grad_norm": 0.46758191028533336,
+      "learning_rate": 2.8876765883732447e-06,
+      "loss": 0.5378,
+      "step": 387
+    },
+    {
+      "epoch": 2.0278745644599305,
+      "grad_norm": 0.5062906288655116,
+      "learning_rate": 2.860070675184036e-06,
+      "loss": 0.5742,
+      "step": 388
+    },
+    {
+      "epoch": 2.033101045296167,
+      "grad_norm": 0.48782738802655556,
+      "learning_rate": 2.8325443932150646e-06,
+      "loss": 0.5343,
+      "step": 389
+    },
+    {
+      "epoch": 2.038327526132404,
+      "grad_norm": 0.5414387332438313,
+      "learning_rate": 2.805098766776529e-06,
+      "loss": 0.5849,
+      "step": 390
+    },
+    {
+      "epoch": 2.043554006968641,
+      "grad_norm": 0.4642400730103952,
+      "learning_rate": 2.7777348171772604e-06,
+      "loss": 0.5105,
+      "step": 391
+    },
+    {
+      "epoch": 2.048780487804878,
+      "grad_norm": 0.5304251807046266,
+      "learning_rate": 2.750453562686729e-06,
+      "loss": 0.5143,
+      "step": 392
+    },
+    {
+      "epoch": 2.054006968641115,
+      "grad_norm": 0.5576744024363436,
+      "learning_rate": 2.7232560184971437e-06,
+      "loss": 0.5839,
+      "step": 393
+    },
+    {
+      "epoch": 2.059233449477352,
+      "grad_norm": 0.45314772448223345,
+      "learning_rate": 2.6961431966856866e-06,
+      "loss": 0.5233,
+      "step": 394
+    },
+    {
+      "epoch": 2.0644599303135887,
+      "grad_norm": 0.4577070909297848,
+      "learning_rate": 2.669116106176838e-06,
+      "loss": 0.5494,
+      "step": 395
+    },
+    {
+      "epoch": 2.0696864111498257,
+      "grad_norm": 0.4395553331677893,
+      "learning_rate": 2.6421757527048373e-06,
+      "loss": 0.5097,
+      "step": 396
+    },
+    {
+      "epoch": 2.0749128919860627,
+      "grad_norm": 0.5287787777736306,
+      "learning_rate": 2.615323138776264e-06,
+      "loss": 0.5719,
+      "step": 397
+    },
+    {
+      "epoch": 2.0801393728222997,
+      "grad_norm": 0.5093454684767568,
+      "learning_rate": 2.588559263632719e-06,
+      "loss": 0.5265,
+      "step": 398
+    },
+    {
+      "epoch": 2.0853658536585367,
+      "grad_norm": 0.5080566318736497,
+      "learning_rate": 2.5618851232136576e-06,
+      "loss": 0.5772,
+      "step": 399
+    },
+    {
+      "epoch": 2.0905923344947737,
+      "grad_norm": 0.4499734335969841,
+      "learning_rate": 2.5353017101193124e-06,
+      "loss": 0.4555,
+      "step": 400
+    },
+    {
+      "epoch": 2.0958188153310107,
+      "grad_norm": 0.5291566057629548,
+      "learning_rate": 2.508810013573767e-06,
+      "loss": 0.5643,
+      "step": 401
+    },
+    {
+      "epoch": 2.1010452961672472,
+      "grad_norm": 0.46824151574429873,
+      "learning_rate": 2.4824110193881384e-06,
+      "loss": 0.5065,
+      "step": 402
+    },
+    {
+      "epoch": 2.1062717770034842,
+      "grad_norm": 0.4617359511491741,
+      "learning_rate": 2.4561057099238973e-06,
+      "loss": 0.563,
+      "step": 403
+    },
+    {
+      "epoch": 2.1114982578397212,
+      "grad_norm": 0.4697636992836946,
+      "learning_rate": 2.4298950640563155e-06,
+      "loss": 0.5056,
+      "step": 404
+    },
+    {
+      "epoch": 2.1167247386759582,
+      "grad_norm": 0.48460609918415953,
+      "learning_rate": 2.4037800571380297e-06,
+      "loss": 0.5517,
+      "step": 405
+    },
+    {
+      "epoch": 2.1219512195121952,
+      "grad_norm": 0.465322807158819,
+      "learning_rate": 2.377761660962754e-06,
+      "loss": 0.5764,
+      "step": 406
+    },
+    {
+      "epoch": 2.1271777003484322,
+      "grad_norm": 0.49593267777356703,
+      "learning_rate": 2.3518408437291155e-06,
+      "loss": 0.5357,
+      "step": 407
+    },
+    {
+      "epoch": 2.132404181184669,
+      "grad_norm": 0.48746378777099386,
+      "learning_rate": 2.3260185700046295e-06,
+      "loss": 0.5321,
+      "step": 408
+    },
+    {
+      "epoch": 2.137630662020906,
+      "grad_norm": 0.46071947103740635,
+      "learning_rate": 2.3002958006897985e-06,
+      "loss": 0.5468,
+      "step": 409
+    },
+    {
+      "epoch": 2.142857142857143,
+      "grad_norm": 0.440111968230906,
+      "learning_rate": 2.2746734929823596e-06,
+      "loss": 0.531,
+      "step": 410
+    },
+    {
+      "epoch": 2.14808362369338,
+      "grad_norm": 0.4575672027883791,
+      "learning_rate": 2.249152600341665e-06,
+      "loss": 0.52,
+      "step": 411
+    },
+    {
+      "epoch": 2.153310104529617,
+      "grad_norm": 0.5283338120770418,
+      "learning_rate": 2.2237340724532007e-06,
+      "loss": 0.6015,
+      "step": 412
+    },
+    {
+      "epoch": 2.158536585365854,
+      "grad_norm": 0.4532501199512347,
+      "learning_rate": 2.1984188551932513e-06,
+      "loss": 0.4911,
+      "step": 413
+    },
+    {
+      "epoch": 2.1637630662020904,
+      "grad_norm": 0.5151201704194432,
+      "learning_rate": 2.173207890593693e-06,
+      "loss": 0.5519,
+      "step": 414
+    },
+    {
+      "epoch": 2.1689895470383274,
+      "grad_norm": 0.4813092159395286,
+      "learning_rate": 2.148102116806944e-06,
+      "loss": 0.531,
+      "step": 415
+    },
+    {
+      "epoch": 2.1742160278745644,
+      "grad_norm": 0.47141365400566,
+      "learning_rate": 2.123102468071058e-06,
+      "loss": 0.5249,
+      "step": 416
+    },
+    {
+      "epoch": 2.1794425087108014,
+      "grad_norm": 0.4935502426085609,
+      "learning_rate": 2.0982098746749524e-06,
+      "loss": 0.5261,
+      "step": 417
+    },
+    {
+      "epoch": 2.1846689895470384,
+      "grad_norm": 0.5305754761741215,
+      "learning_rate": 2.0734252629237892e-06,
+      "loss": 0.609,
+      "step": 418
+    },
+    {
+      "epoch": 2.1898954703832754,
+      "grad_norm": 0.482899473961335,
+      "learning_rate": 2.048749555104516e-06,
+      "loss": 0.5314,
+      "step": 419
+    },
+    {
+      "epoch": 2.1951219512195124,
+      "grad_norm": 0.44109993576183376,
+      "learning_rate": 2.0241836694515338e-06,
+      "loss": 0.54,
+      "step": 420
+    },
+    {
+      "epoch": 2.200348432055749,
+      "grad_norm": 0.4541137377252633,
+      "learning_rate": 1.9997285201125328e-06,
+      "loss": 0.5182,
+      "step": 421
+    },
+    {
+      "epoch": 2.205574912891986,
+      "grad_norm": 0.47179876294255096,
+      "learning_rate": 1.975385017114473e-06,
+      "loss": 0.5786,
+      "step": 422
+    },
+    {
+      "epoch": 2.210801393728223,
+      "grad_norm": 0.49973970600486517,
+      "learning_rate": 1.9511540663297284e-06,
+      "loss": 0.5183,
+      "step": 423
+    },
+    {
+      "epoch": 2.21602787456446,
+      "grad_norm": 0.46356554425956825,
+      "learning_rate": 1.9270365694423654e-06,
+      "loss": 0.5159,
+      "step": 424
+    },
+    {
+      "epoch": 2.221254355400697,
+      "grad_norm": 0.5241753937590147,
+      "learning_rate": 1.903033423914596e-06,
+      "loss": 0.5957,
+      "step": 425
+    },
+    {
+      "epoch": 2.226480836236934,
+      "grad_norm": 0.44148518090830996,
+      "learning_rate": 1.8791455229533806e-06,
+      "loss": 0.4488,
+      "step": 426
+    },
+    {
+      "epoch": 2.231707317073171,
+      "grad_norm": 0.5038639568195054,
+      "learning_rate": 1.8553737554771883e-06,
+      "loss": 0.6349,
+      "step": 427
+    },
+    {
+      "epoch": 2.2369337979094075,
+      "grad_norm": 0.49606752296755446,
+      "learning_rate": 1.8317190060829242e-06,
+      "loss": 0.5089,
+      "step": 428
+    },
+    {
+      "epoch": 2.2421602787456445,
+      "grad_norm": 0.45579655008006387,
+      "learning_rate": 1.808182155013003e-06,
+      "loss": 0.4899,
+      "step": 429
+    },
+    {
+      "epoch": 2.2473867595818815,
+      "grad_norm": 0.464962788957996,
+      "learning_rate": 1.7847640781225982e-06,
+      "loss": 0.5486,
+      "step": 430
+    },
+    {
+      "epoch": 2.2526132404181185,
+      "grad_norm": 0.4984600881906674,
+      "learning_rate": 1.7614656468470508e-06,
+      "loss": 0.6217,
+      "step": 431
+    },
+    {
+      "epoch": 2.2578397212543555,
+      "grad_norm": 0.4386510268154617,
+      "learning_rate": 1.7382877281694355e-06,
+      "loss": 0.5065,
+      "step": 432
+    },
+    {
+      "epoch": 2.2630662020905925,
+      "grad_norm": 0.4628957678602141,
+      "learning_rate": 1.7152311845883096e-06,
+      "loss": 0.5339,
+      "step": 433
+    },
+    {
+      "epoch": 2.2682926829268295,
+      "grad_norm": 0.4511882273309374,
+      "learning_rate": 1.692296874085605e-06,
+      "loss": 0.5833,
+      "step": 434
+    },
+    {
+      "epoch": 2.273519163763066,
+      "grad_norm": 0.4423489129104317,
+      "learning_rate": 1.6694856500947081e-06,
+      "loss": 0.5414,
+      "step": 435
+    },
+    {
+      "epoch": 2.278745644599303,
+      "grad_norm": 0.4278102250494057,
+      "learning_rate": 1.6467983614686995e-06,
+      "loss": 0.5291,
+      "step": 436
+    },
+    {
+      "epoch": 2.28397212543554,
+      "grad_norm": 0.4899503913317031,
+      "learning_rate": 1.62423585244877e-06,
+      "loss": 0.5207,
+      "step": 437
+    },
+    {
+      "epoch": 2.289198606271777,
+      "grad_norm": 0.45314192403359066,
+      "learning_rate": 1.601798962632799e-06,
+      "loss": 0.5233,
+      "step": 438
+    },
+    {
+      "epoch": 2.294425087108014,
+      "grad_norm": 0.4675516971595869,
+      "learning_rate": 1.5794885269441152e-06,
+      "loss": 0.5789,
+      "step": 439
+    },
+    {
+      "epoch": 2.2996515679442506,
+      "grad_norm": 0.4684838825648316,
+      "learning_rate": 1.5573053756004253e-06,
+      "loss": 0.5447,
+      "step": 440
+    },
+    {
+      "epoch": 2.3048780487804876,
+      "grad_norm": 0.4742515090348089,
+      "learning_rate": 1.53525033408292e-06,
+      "loss": 0.5522,
+      "step": 441
+    },
+    {
+      "epoch": 2.3101045296167246,
+      "grad_norm": 0.4255877001891176,
+      "learning_rate": 1.5133242231055622e-06,
+      "loss": 0.52,
+      "step": 442
+    },
+    {
+      "epoch": 2.3153310104529616,
+      "grad_norm": 0.4703877181502192,
+      "learning_rate": 1.491527858584535e-06,
+      "loss": 0.5629,
+      "step": 443
+    },
+    {
+      "epoch": 2.3205574912891986,
+      "grad_norm": 0.4564707024075428,
+      "learning_rate": 1.4698620516078882e-06,
+      "loss": 0.5899,
+      "step": 444
+    },
+    {
+      "epoch": 2.3257839721254356,
+      "grad_norm": 0.4393071035962851,
+      "learning_rate": 1.4483276084053567e-06,
+      "loss": 0.514,
+      "step": 445
+    },
+    {
+      "epoch": 2.3310104529616726,
+      "grad_norm": 0.4382243891353936,
+      "learning_rate": 1.4269253303183516e-06,
+      "loss": 0.5766,
+      "step": 446
+    },
+    {
+      "epoch": 2.3362369337979096,
+      "grad_norm": 0.46207515860487863,
+      "learning_rate": 1.4056560137701469e-06,
+      "loss": 0.492,
+      "step": 447
+    },
+    {
+      "epoch": 2.341463414634146,
+      "grad_norm": 0.4779708014243619,
+      "learning_rate": 1.3845204502362442e-06,
+      "loss": 0.5503,
+      "step": 448
+    },
+    {
+      "epoch": 2.346689895470383,
+      "grad_norm": 0.4446567240946519,
+      "learning_rate": 1.3635194262149131e-06,
+      "loss": 0.5173,
+      "step": 449
+    },
+    {
+      "epoch": 2.35191637630662,
+      "grad_norm": 0.5194569385386936,
+      "learning_rate": 1.3426537231979309e-06,
+      "loss": 0.5464,
+      "step": 450
+    },
+    {
+      "epoch": 2.357142857142857,
+      "grad_norm": 0.47302377318250843,
+      "learning_rate": 1.3219241176414948e-06,
+      "loss": 0.5341,
+      "step": 451
+    },
+    {
+      "epoch": 2.362369337979094,
+      "grad_norm": 0.4316213228653094,
+      "learning_rate": 1.3013313809373396e-06,
+      "loss": 0.5283,
+      "step": 452
+    },
+    {
+      "epoch": 2.3675958188153308,
+      "grad_norm": 0.4347857005792825,
+      "learning_rate": 1.28087627938402e-06,
+      "loss": 0.5103,
+      "step": 453
+    },
+    {
+      "epoch": 2.3728222996515678,
+      "grad_norm": 0.4599959414145429,
+      "learning_rate": 1.2605595741584015e-06,
+      "loss": 0.5989,
+      "step": 454
+    },
+    {
+      "epoch": 2.3780487804878048,
+      "grad_norm": 0.4477421618895101,
+      "learning_rate": 1.2403820212873347e-06,
+      "loss": 0.5247,
+      "step": 455
+    },
+    {
+      "epoch": 2.3832752613240418,
+      "grad_norm": 0.46194912531798293,
+      "learning_rate": 1.2203443716195213e-06,
+      "loss": 0.547,
+      "step": 456
+    },
+    {
+      "epoch": 2.3885017421602788,
+      "grad_norm": 0.48246695872510925,
+      "learning_rate": 1.200447370797576e-06,
+      "loss": 0.5319,
+      "step": 457
+    },
+    {
+      "epoch": 2.3937282229965158,
+      "grad_norm": 0.44364446337995106,
+      "learning_rate": 1.1806917592302763e-06,
+      "loss": 0.5079,
+      "step": 458
+    },
+    {
+      "epoch": 2.3989547038327528,
+      "grad_norm": 0.45366308339292766,
+      "learning_rate": 1.16107827206501e-06,
+      "loss": 0.528,
+      "step": 459
+    },
+    {
+      "epoch": 2.40418118466899,
+      "grad_norm": 0.44739930129966016,
+      "learning_rate": 1.1416076391604197e-06,
+      "loss": 0.5351,
+      "step": 460
+    },
+    {
+      "epoch": 2.4094076655052263,
+      "grad_norm": 0.4842777825049656,
+      "learning_rate": 1.12228058505925e-06,
+      "loss": 0.5459,
+      "step": 461
+    },
+    {
+      "epoch": 2.4146341463414633,
+      "grad_norm": 0.46656539985547907,
+      "learning_rate": 1.1030978289613725e-06,
+      "loss": 0.5377,
+      "step": 462
+    },
+    {
+      "epoch": 2.4198606271777003,
+      "grad_norm": 0.4434636263867975,
+      "learning_rate": 1.0840600846970333e-06,
+      "loss": 0.5494,
+      "step": 463
+    },
+    {
+      "epoch": 2.4250871080139373,
+      "grad_norm": 0.42991245492599656,
+      "learning_rate": 1.0651680607002861e-06,
+      "loss": 0.5372,
+      "step": 464
+    },
+    {
+      "epoch": 2.4303135888501743,
+      "grad_norm": 0.47167305733035525,
+      "learning_rate": 1.0464224599826301e-06,
+      "loss": 0.5127,
+      "step": 465
+    },
+    {
+      "epoch": 2.435540069686411,
+      "grad_norm": 0.460998662886707,
+      "learning_rate": 1.0278239801068518e-06,
+      "loss": 0.5137,
+      "step": 466
+    },
+    {
+      "epoch": 2.440766550522648,
+      "grad_norm": 0.5048182621907691,
+      "learning_rate": 1.0093733131610621e-06,
+      "loss": 0.6034,
+      "step": 467
+    },
+    {
+      "epoch": 2.445993031358885,
+      "grad_norm": 0.46718193125046437,
+      "learning_rate": 9.91071145732948e-07,
+      "loss": 0.5201,
+      "step": 468
+    },
+    {
+      "epoch": 2.451219512195122,
+      "grad_norm": 0.40697553295642236,
+      "learning_rate": 9.729181588842184e-07,
+      "loss": 0.4763,
+      "step": 469
+    },
+    {
+      "epoch": 2.456445993031359,
+      "grad_norm": 0.4282850894981929,
+      "learning_rate": 9.549150281252633e-07,
+      "loss": 0.5551,
+      "step": 470
+    },
+    {
+      "epoch": 2.461672473867596,
+      "grad_norm": 0.4425949078660112,
+      "learning_rate": 9.370624233900183e-07,
+      "loss": 0.509,
+      "step": 471
+    },
+    {
+      "epoch": 2.466898954703833,
+      "grad_norm": 0.43823474167181586,
+      "learning_rate": 9.193610090110305e-07,
+      "loss": 0.5605,
+      "step": 472
+    },
+    {
+      "epoch": 2.47212543554007,
+      "grad_norm": 0.45300520608343586,
+      "learning_rate": 9.018114436947373e-07,
+      "loss": 0.5394,
+      "step": 473
+    },
+    {
+      "epoch": 2.4773519163763065,
+      "grad_norm": 0.4830008293506701,
+      "learning_rate": 8.844143804969624e-07,
+      "loss": 0.5614,
+      "step": 474
+    },
+    {
+      "epoch": 2.4825783972125435,
+      "grad_norm": 0.4313889868995408,
+      "learning_rate": 8.671704667986036e-07,
+      "loss": 0.495,
+      "step": 475
+    },
+    {
+      "epoch": 2.4878048780487805,
+      "grad_norm": 0.4652585907754659,
+      "learning_rate": 8.500803442815475e-07,
+      "loss": 0.5337,
+      "step": 476
+    },
+    {
+      "epoch": 2.4930313588850175,
+      "grad_norm": 0.4434579349259814,
+      "learning_rate": 8.331446489047956e-07,
+      "loss": 0.5432,
+      "step": 477
+    },
+    {
+      "epoch": 2.4982578397212545,
+      "grad_norm": 0.4260838162732741,
+      "learning_rate": 8.163640108807897e-07,
+      "loss": 0.5526,
+      "step": 478
+    },
+    {
+      "epoch": 2.503484320557491,
+      "grad_norm": 0.41851893716167266,
+      "learning_rate": 7.997390546519668e-07,
+      "loss": 0.5274,
+      "step": 479
+    },
+    {
+      "epoch": 2.508710801393728,
+      "grad_norm": 0.43352778515121865,
+      "learning_rate": 7.832703988675195e-07,
+      "loss": 0.5498,
+      "step": 480
+    },
+    {
+      "epoch": 2.513937282229965,
+      "grad_norm": 0.45074130711852517,
+      "learning_rate": 7.669586563603782e-07,
+      "loss": 0.5531,
+      "step": 481
+    },
+    {
+      "epoch": 2.519163763066202,
+      "grad_norm": 0.45864866148701783,
+      "learning_rate": 7.508044341244014e-07,
+      "loss": 0.5914,
+      "step": 482
+    },
+    {
+      "epoch": 2.524390243902439,
+      "grad_norm": 0.43380289787943704,
+      "learning_rate": 7.348083332917927e-07,
+      "loss": 0.5302,
+      "step": 483
+    },
+    {
+      "epoch": 2.529616724738676,
+      "grad_norm": 0.44541179316397,
+      "learning_rate": 7.189709491107272e-07,
+      "loss": 0.5103,
+      "step": 484
+    },
+    {
+      "epoch": 2.534843205574913,
+      "grad_norm": 0.42812608402735486,
+      "learning_rate": 7.032928709232062e-07,
+      "loss": 0.5131,
+      "step": 485
+    },
+    {
+      "epoch": 2.54006968641115,
+      "grad_norm": 0.4641631244885751,
+      "learning_rate": 6.87774682143122e-07,
+      "loss": 0.5406,
+      "step": 486
+    },
+    {
+      "epoch": 2.5452961672473866,
+      "grad_norm": 0.46466095362861476,
+      "learning_rate": 6.724169602345487e-07,
+      "loss": 0.5524,
+      "step": 487
+    },
+    {
+      "epoch": 2.5505226480836236,
+      "grad_norm": 0.44159109445526523,
+      "learning_rate": 6.572202766902569e-07,
+      "loss": 0.5107,
+      "step": 488
+    },
+    {
+      "epoch": 2.5557491289198606,
+      "grad_norm": 0.43095228829120374,
+      "learning_rate": 6.421851970104409e-07,
+      "loss": 0.5465,
+      "step": 489
+    },
+    {
+      "epoch": 2.5609756097560976,
+      "grad_norm": 0.4398781503440029,
+      "learning_rate": 6.273122806816845e-07,
+      "loss": 0.5306,
+      "step": 490
+    },
+    {
+      "epoch": 2.5662020905923346,
+      "grad_norm": 0.455532997496077,
+      "learning_rate": 6.126020811561323e-07,
+      "loss": 0.5881,
+      "step": 491
+    },
+    {
+      "epoch": 2.571428571428571,
+      "grad_norm": 0.3948516059085693,
+      "learning_rate": 5.980551458308981e-07,
+      "loss": 0.4535,
+      "step": 492
+    },
+    {
+      "epoch": 2.576655052264808,
+      "grad_norm": 0.4009473999258072,
+      "learning_rate": 5.836720160276971e-07,
+      "loss": 0.5696,
+      "step": 493
+    },
+    {
+      "epoch": 2.581881533101045,
+      "grad_norm": 0.4065638987549273,
+      "learning_rate": 5.694532269726977e-07,
+      "loss": 0.5412,
+      "step": 494
+    },
+    {
+      "epoch": 2.587108013937282,
+      "grad_norm": 0.4539932581334291,
+      "learning_rate": 5.553993077766124e-07,
+      "loss": 0.5886,
+      "step": 495
+    },
+    {
+      "epoch": 2.592334494773519,
+      "grad_norm": 0.42936026661045573,
+      "learning_rate": 5.415107814149978e-07,
+      "loss": 0.519,
+      "step": 496
+    },
+    {
+      "epoch": 2.597560975609756,
+      "grad_norm": 0.4335795734646765,
+      "learning_rate": 5.277881647088024e-07,
+      "loss": 0.5866,
+      "step": 497
+    },
+    {
+      "epoch": 2.602787456445993,
+      "grad_norm": 0.44871188719474847,
+      "learning_rate": 5.1423196830513e-07,
+      "loss": 0.5824,
+      "step": 498
+    },
+    {
+      "epoch": 2.60801393728223,
+      "grad_norm": 0.4649294419801843,
+      "learning_rate": 5.008426966582386e-07,
+      "loss": 0.5086,
+      "step": 499
+    },
+    {
+      "epoch": 2.6132404181184667,
+      "grad_norm": 0.47884736570244324,
+      "learning_rate": 4.87620848010772e-07,
+      "loss": 0.5853,
+      "step": 500
+    },
+    {
+      "epoch": 2.6184668989547037,
+      "grad_norm": 0.4070247906117974,
+      "learning_rate": 4.7456691437521243e-07,
+      "loss": 0.4705,
+      "step": 501
+    },
+    {
+      "epoch": 2.6236933797909407,
+      "grad_norm": 0.44228619733522523,
+      "learning_rate": 4.6168138151557516e-07,
+      "loss": 0.6052,
+      "step": 502
+    },
+    {
+      "epoch": 2.6289198606271778,
+      "grad_norm": 0.4186718748315985,
+      "learning_rate": 4.4896472892933693e-07,
+      "loss": 0.5156,
+      "step": 503
+    },
+    {
+      "epoch": 2.6341463414634148,
+      "grad_norm": 0.41927468655653893,
+      "learning_rate": 4.3641742982958203e-07,
+      "loss": 0.5552,
+      "step": 504
+    },
+    {
+      "epoch": 2.6393728222996513,
+      "grad_norm": 0.3807275422714154,
+      "learning_rate": 4.240399511274057e-07,
+      "loss": 0.4874,
+      "step": 505
+    },
+    {
+      "epoch": 2.6445993031358883,
+      "grad_norm": 0.43043158611193255,
+      "learning_rate": 4.118327534145278e-07,
+      "loss": 0.5473,
+      "step": 506
+    },
+    {
+      "epoch": 2.6498257839721253,
+      "grad_norm": 0.4427511902610817,
+      "learning_rate": 3.997962909461611e-07,
+      "loss": 0.5476,
+      "step": 507
+    },
+    {
+      "epoch": 2.6550522648083623,
+      "grad_norm": 0.4287387407398484,
+      "learning_rate": 3.8793101162410417e-07,
+      "loss": 0.515,
+      "step": 508
+    },
+    {
+      "epoch": 2.6602787456445993,
+      "grad_norm": 0.4581450049667044,
+      "learning_rate": 3.762373569800737e-07,
+      "loss": 0.5736,
+      "step": 509
+    },
+    {
+      "epoch": 2.6655052264808363,
+      "grad_norm": 0.45437023578463565,
+      "learning_rate": 3.6471576215927897e-07,
+      "loss": 0.5301,
+      "step": 510
+    },
+    {
+      "epoch": 2.6707317073170733,
+      "grad_norm": 0.4205454157334176,
+      "learning_rate": 3.5336665590422147e-07,
+      "loss": 0.5274,
+      "step": 511
+    },
+    {
+      "epoch": 2.6759581881533103,
+      "grad_norm": 0.42921462391644427,
+      "learning_rate": 3.4219046053874604e-07,
+      "loss": 0.5453,
+      "step": 512
+    },
+    {
+      "epoch": 2.681184668989547,
+      "grad_norm": 0.4233533929984576,
+      "learning_rate": 3.3118759195232273e-07,
+      "loss": 0.5573,
+      "step": 513
+    },
+    {
+      "epoch": 2.686411149825784,
+      "grad_norm": 0.3919672353563574,
+      "learning_rate": 3.2035845958457324e-07,
+      "loss": 0.5181,
+      "step": 514
+    },
+    {
+      "epoch": 2.691637630662021,
+      "grad_norm": 0.47372027451981813,
+      "learning_rate": 3.0970346641003346e-07,
+      "loss": 0.5436,
+      "step": 515
+    },
+    {
+      "epoch": 2.696864111498258,
+      "grad_norm": 0.4396594930081907,
+      "learning_rate": 2.9922300892315517e-07,
+      "loss": 0.5475,
+      "step": 516
+    },
+    {
+      "epoch": 2.702090592334495,
+      "grad_norm": 0.4473565138284573,
+      "learning_rate": 2.88917477123557e-07,
+      "loss": 0.5559,
+      "step": 517
+    },
+    {
+      "epoch": 2.7073170731707314,
+      "grad_norm": 0.4024822510249766,
+      "learning_rate": 2.787872545015069e-07,
+      "loss": 0.5059,
+      "step": 518
+    },
+    {
+      "epoch": 2.7125435540069684,
+      "grad_norm": 0.4665623873342051,
+      "learning_rate": 2.6883271802365606e-07,
+      "loss": 0.5585,
+      "step": 519
+    },
+    {
+      "epoch": 2.7177700348432055,
+      "grad_norm": 0.42854221934756087,
+      "learning_rate": 2.5905423811900755e-07,
+      "loss": 0.5104,
+      "step": 520
+    },
+    {
+      "epoch": 2.7229965156794425,
+      "grad_norm": 0.4286908922250654,
+      "learning_rate": 2.4945217866513274e-07,
+      "loss": 0.5518,
+      "step": 521
+    },
+    {
+      "epoch": 2.7282229965156795,
+      "grad_norm": 0.44386460142096057,
+      "learning_rate": 2.4002689697463224e-07,
+      "loss": 0.5305,
+      "step": 522
+    },
+    {
+      "epoch": 2.7334494773519165,
+      "grad_norm": 0.43745356189843465,
+      "learning_rate": 2.307787437818365e-07,
+      "loss": 0.5579,
+      "step": 523
+    },
+    {
+      "epoch": 2.7386759581881535,
+      "grad_norm": 0.4192061612455892,
+      "learning_rate": 2.2170806322976023e-07,
+      "loss": 0.5133,
+      "step": 524
+    },
+    {
+      "epoch": 2.7439024390243905,
+      "grad_norm": 0.44288601777891073,
+      "learning_rate": 2.1281519285728803e-07,
+      "loss": 0.5837,
+      "step": 525
+    },
+    {
+      "epoch": 2.749128919860627,
+      "grad_norm": 0.4161039648633803,
+      "learning_rate": 2.041004635866195e-07,
+      "loss": 0.5105,
+      "step": 526
+    },
+    {
+      "epoch": 2.754355400696864,
+      "grad_norm": 0.4294055964727348,
+      "learning_rate": 1.9556419971095363e-07,
+      "loss": 0.5319,
+      "step": 527
+    },
+    {
+      "epoch": 2.759581881533101,
+      "grad_norm": 0.4358089354550434,
+      "learning_rate": 1.8720671888242058e-07,
+      "loss": 0.4988,
+      "step": 528
+    },
+    {
+      "epoch": 2.764808362369338,
+      "grad_norm": 0.43788412237975727,
+      "learning_rate": 1.7902833210026228e-07,
+      "loss": 0.546,
+      "step": 529
+    },
+    {
+      "epoch": 2.770034843205575,
+      "grad_norm": 0.45097033105165757,
+      "learning_rate": 1.710293436992566e-07,
+      "loss": 0.5696,
+      "step": 530
+    },
+    {
+      "epoch": 2.7752613240418116,
+      "grad_norm": 0.4138845338967673,
+      "learning_rate": 1.6321005133839884e-07,
+      "loss": 0.5316,
+      "step": 531
+    },
+    {
+      "epoch": 2.7804878048780486,
+      "grad_norm": 0.4202899606511126,
+      "learning_rate": 1.5557074598981593e-07,
+      "loss": 0.5301,
+      "step": 532
+    },
+    {
+      "epoch": 2.7857142857142856,
+      "grad_norm": 0.42666653234169716,
+      "learning_rate": 1.4811171192794628e-07,
+      "loss": 0.5697,
+      "step": 533
+    },
+    {
+      "epoch": 2.7909407665505226,
+      "grad_norm": 0.37877132072819447,
+      "learning_rate": 1.4083322671896048e-07,
+      "loss": 0.4895,
+      "step": 534
+    },
+    {
+      "epoch": 2.7961672473867596,
+      "grad_norm": 0.4422175332427632,
+      "learning_rate": 1.337355612104274e-07,
+      "loss": 0.5793,
+      "step": 535
+    },
+    {
+      "epoch": 2.8013937282229966,
+      "grad_norm": 0.4141363458191714,
+      "learning_rate": 1.2681897952124046e-07,
+      "loss": 0.5057,
+      "step": 536
+    },
+    {
+      "epoch": 2.8066202090592336,
+      "grad_norm": 0.4469656727146587,
+      "learning_rate": 1.2008373903178828e-07,
+      "loss": 0.5524,
+      "step": 537
+    },
+    {
+      "epoch": 2.8118466898954706,
+      "grad_norm": 0.42707047626049666,
+      "learning_rate": 1.1353009037437523e-07,
+      "loss": 0.4922,
+      "step": 538
+    },
+    {
+      "epoch": 2.817073170731707,
+      "grad_norm": 0.45141362991328465,
+      "learning_rate": 1.0715827742389717e-07,
+      "loss": 0.5724,
+      "step": 539
+    },
+    {
+      "epoch": 2.822299651567944,
+      "grad_norm": 0.45003668834699156,
+      "learning_rate": 1.0096853728876365e-07,
+      "loss": 0.4959,
+      "step": 540
+    },
+    {
+      "epoch": 2.827526132404181,
+      "grad_norm": 0.40703793301322555,
+      "learning_rate": 9.496110030207673e-08,
+      "loss": 0.5187,
+      "step": 541
+    },
+    {
+      "epoch": 2.832752613240418,
+      "grad_norm": 0.4349893646514051,
+      "learning_rate": 8.913619001305995e-08,
+      "loss": 0.5413,
+      "step": 542
+    },
+    {
+      "epoch": 2.837979094076655,
+      "grad_norm": 0.4413261937466953,
+      "learning_rate": 8.34940231787379e-08,
+      "loss": 0.5597,
+      "step": 543
+    },
+    {
+      "epoch": 2.8432055749128917,
+      "grad_norm": 0.4219769036035547,
+      "learning_rate": 7.803480975587197e-08,
+      "loss": 0.542,
+      "step": 544
+    },
+    {
+      "epoch": 2.8484320557491287,
+      "grad_norm": 0.4673046785235365,
+      "learning_rate": 7.275875289314616e-08,
+      "loss": 0.5495,
+      "step": 545
+    },
+    {
+      "epoch": 2.8536585365853657,
+      "grad_norm": 0.45105426206164123,
+      "learning_rate": 6.766604892360751e-08,
+      "loss": 0.5724,
+      "step": 546
+    },
+    {
+      "epoch": 2.8588850174216027,
+      "grad_norm": 0.43511636060314596,
+      "learning_rate": 6.275688735736141e-08,
+      "loss": 0.5257,
+      "step": 547
+    },
+    {
+      "epoch": 2.8641114982578397,
+      "grad_norm": 0.43349086279126203,
+      "learning_rate": 5.803145087451945e-08,
+      "loss": 0.5131,
+      "step": 548
+    },
+    {
+      "epoch": 2.8693379790940767,
+      "grad_norm": 0.4537611808145007,
+      "learning_rate": 5.348991531839875e-08,
+      "loss": 0.5607,
+      "step": 549
+    },
+    {
+      "epoch": 2.8745644599303137,
+      "grad_norm": 0.4439279551647372,
+      "learning_rate": 4.913244968898279e-08,
+      "loss": 0.5056,
+      "step": 550
+    },
+    {
+      "epoch": 2.8797909407665507,
+      "grad_norm": 0.4411930978992959,
+      "learning_rate": 4.495921613662746e-08,
+      "loss": 0.5415,
+      "step": 551
+    },
+    {
+      "epoch": 2.8850174216027873,
+      "grad_norm": 0.4436199780022762,
+      "learning_rate": 4.097036995603321e-08,
+      "loss": 0.5205,
+      "step": 552
+    },
+    {
+      "epoch": 2.8902439024390243,
+      "grad_norm": 0.42509245773578536,
+      "learning_rate": 3.716605958046071e-08,
+      "loss": 0.5259,
+      "step": 553
+    },
+    {
+      "epoch": 2.8954703832752613,
+      "grad_norm": 0.43131172360952724,
+      "learning_rate": 3.354642657621032e-08,
+      "loss": 0.604,
+      "step": 554
+    },
+    {
+      "epoch": 2.9006968641114983,
+      "grad_norm": 0.4133105200801449,
+      "learning_rate": 3.011160563735349e-08,
+      "loss": 0.4883,
+      "step": 555
+    },
+    {
+      "epoch": 2.9059233449477353,
+      "grad_norm": 0.4391744276562499,
+      "learning_rate": 2.6861724580719562e-08,
+      "loss": 0.5621,
+      "step": 556
+    },
+    {
+      "epoch": 2.911149825783972,
+      "grad_norm": 0.4391177361243526,
+      "learning_rate": 2.3796904341141236e-08,
+      "loss": 0.5334,
+      "step": 557
+    },
+    {
+      "epoch": 2.916376306620209,
+      "grad_norm": 0.4214891944016137,
+      "learning_rate": 2.0917258966953735e-08,
+      "loss": 0.553,
+      "step": 558
+    },
+    {
+      "epoch": 2.921602787456446,
+      "grad_norm": 0.4014499901153732,
+      "learning_rate": 1.8222895615748748e-08,
+      "loss": 0.5358,
+      "step": 559
+    },
+    {
+      "epoch": 2.926829268292683,
+      "grad_norm": 0.4454567928346748,
+      "learning_rate": 1.5713914550389843e-08,
+      "loss": 0.5405,
+      "step": 560
+    },
+    {
+      "epoch": 2.93205574912892,
+      "grad_norm": 0.4178833466790727,
+      "learning_rate": 1.3390409135281001e-08,
+      "loss": 0.5513,
+      "step": 561
+    },
+    {
+      "epoch": 2.937282229965157,
+      "grad_norm": 0.4457705249971793,
+      "learning_rate": 1.125246583288886e-08,
+      "loss": 0.5245,
+      "step": 562
+    },
+    {
+      "epoch": 2.942508710801394,
+      "grad_norm": 0.4645928048601542,
+      "learning_rate": 9.300164200530815e-09,
+      "loss": 0.5701,
+      "step": 563
+    },
+    {
+      "epoch": 2.947735191637631,
+      "grad_norm": 0.4563271003769357,
+      "learning_rate": 7.533576887410188e-09,
+      "loss": 0.4999,
+      "step": 564
+    },
+    {
+      "epoch": 2.952961672473868,
+      "grad_norm": 0.39820608811758146,
+      "learning_rate": 5.95276963191449e-09,
+      "loss": 0.5704,
+      "step": 565
+    },
+    {
+      "epoch": 2.9581881533101044,
+      "grad_norm": 0.4590404415892991,
+      "learning_rate": 4.557801259169048e-09,
+      "loss": 0.4995,
+      "step": 566
+    },
+    {
+      "epoch": 2.9634146341463414,
+      "grad_norm": 0.455741641646435,
+      "learning_rate": 3.348723678847643e-09,
+      "loss": 0.5569,
+      "step": 567
+    },
+    {
+      "epoch": 2.9686411149825784,
+      "grad_norm": 0.4914791262393272,
+      "learning_rate": 2.3255818832423894e-09,
+      "loss": 0.5232,
+      "step": 568
+    },
+    {
+      "epoch": 2.9738675958188154,
+      "grad_norm": 0.4310628707346215,
+      "learning_rate": 1.4884139455861868e-09,
+      "loss": 0.5572,
+      "step": 569
+    },
+    {
+      "epoch": 2.979094076655052,
+      "grad_norm": 0.43105949566502877,
+      "learning_rate": 8.372510186388516e-10,
+      "loss": 0.5603,
+      "step": 570
+    },
+    {
+      "epoch": 2.984320557491289,
+      "grad_norm": 0.3988129414746118,
+      "learning_rate": 3.7211733352748856e-10,
+      "loss": 0.4893,
+      "step": 571
+    },
+    {
+      "epoch": 2.989547038327526,
+      "grad_norm": 0.4326905501521538,
+      "learning_rate": 9.303019884387976e-11,
+      "loss": 0.5622,
+      "step": 572
+    },
+    {
+      "epoch": 2.994773519163763,
+      "grad_norm": 0.4266940308017901,
+      "learning_rate": 0.0,
+      "loss": 0.4927,
+      "step": 573
+    },
+    {
+      "epoch": 2.994773519163763,
+      "step": 573,
+      "total_flos": 520358168002560.0,
+      "train_loss": 0.6214590458553707,
+      "train_runtime": 8980.1589,
+      "train_samples_per_second": 6.133,
+      "train_steps_per_second": 0.064
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 573,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 520358168002560.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}