diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8652 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.982278481012658,
+  "eval_steps": 500,
+  "global_step": 1230,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004050632911392405,
+      "grad_norm": 6.419734704396513,
+      "learning_rate": 3.2520325203252037e-07,
+      "loss": 1.0356,
+      "step": 1
+    },
+    {
+      "epoch": 0.00810126582278481,
+      "grad_norm": 6.468882313374203,
+      "learning_rate": 6.504065040650407e-07,
+      "loss": 1.0609,
+      "step": 2
+    },
+    {
+      "epoch": 0.012151898734177215,
+      "grad_norm": 6.345731949885464,
+      "learning_rate": 9.75609756097561e-07,
+      "loss": 1.0509,
+      "step": 3
+    },
+    {
+      "epoch": 0.01620253164556962,
+      "grad_norm": 6.20532859333902,
+      "learning_rate": 1.3008130081300815e-06,
+      "loss": 1.0437,
+      "step": 4
+    },
+    {
+      "epoch": 0.020253164556962026,
+      "grad_norm": 6.038525356156134,
+      "learning_rate": 1.6260162601626018e-06,
+      "loss": 1.0321,
+      "step": 5
+    },
+    {
+      "epoch": 0.02430379746835443,
+      "grad_norm": 5.841355470938381,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 1.0285,
+      "step": 6
+    },
+    {
+      "epoch": 0.028354430379746835,
+      "grad_norm": 4.534321544271132,
+      "learning_rate": 2.2764227642276426e-06,
+      "loss": 0.9933,
+      "step": 7
+    },
+    {
+      "epoch": 0.03240506329113924,
+      "grad_norm": 4.195616409642077,
+      "learning_rate": 2.601626016260163e-06,
+      "loss": 0.9906,
+      "step": 8
+    },
+    {
+      "epoch": 0.03645569620253165,
+      "grad_norm": 2.688555365716947,
+      "learning_rate": 2.926829268292683e-06,
+      "loss": 0.952,
+      "step": 9
+    },
+    {
+      "epoch": 0.04050632911392405,
+      "grad_norm": 2.511251204396362,
+      "learning_rate": 3.2520325203252037e-06,
+      "loss": 0.9441,
+      "step": 10
+    },
+    {
+      "epoch": 0.044556962025316456,
+      "grad_norm": 2.5138855352378346,
+      "learning_rate": 3.577235772357724e-06,
+      "loss": 0.9491,
+      "step": 11
+    },
+    {
+      "epoch": 0.04860759493670886,
+      "grad_norm": 3.7669352367471167,
+      "learning_rate": 3.902439024390244e-06,
+      "loss": 0.9285,
+      "step": 12
+    },
+    {
+      "epoch": 0.052658227848101265,
+      "grad_norm": 4.226887117461395,
+      "learning_rate": 4.227642276422765e-06,
+      "loss": 0.9547,
+      "step": 13
+    },
+    {
+      "epoch": 0.05670886075949367,
+      "grad_norm": 3.929759729194547,
+      "learning_rate": 4.552845528455285e-06,
+      "loss": 0.9252,
+      "step": 14
+    },
+    {
+      "epoch": 0.060759493670886074,
+      "grad_norm": 3.719693838660686,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 0.9481,
+      "step": 15
+    },
+    {
+      "epoch": 0.06481012658227848,
+      "grad_norm": 2.814113376669872,
+      "learning_rate": 5.203252032520326e-06,
+      "loss": 0.8965,
+      "step": 16
+    },
+    {
+      "epoch": 0.06886075949367089,
+      "grad_norm": 2.4976644889700954,
+      "learning_rate": 5.528455284552846e-06,
+      "loss": 0.8695,
+      "step": 17
+    },
+    {
+      "epoch": 0.0729113924050633,
+      "grad_norm": 2.0425347265151017,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 0.8541,
+      "step": 18
+    },
+    {
+      "epoch": 0.0769620253164557,
+      "grad_norm": 1.5125624629462948,
+      "learning_rate": 6.178861788617887e-06,
+      "loss": 0.8402,
+      "step": 19
+    },
+    {
+      "epoch": 0.0810126582278481,
+      "grad_norm": 1.8993245583689031,
+      "learning_rate": 6.504065040650407e-06,
+      "loss": 0.8851,
+      "step": 20
+    },
+    {
+      "epoch": 0.08506329113924051,
+      "grad_norm": 1.1555381180973177,
+      "learning_rate": 6.829268292682928e-06,
+      "loss": 0.8079,
+      "step": 21
+    },
+    {
+      "epoch": 0.08911392405063291,
+      "grad_norm": 1.2413543768542576,
+      "learning_rate": 7.154471544715448e-06,
+      "loss": 0.8243,
+      "step": 22
+    },
+    {
+      "epoch": 0.09316455696202532,
+      "grad_norm": 1.0777459907012676,
+      "learning_rate": 7.4796747967479676e-06,
+      "loss": 0.7922,
+      "step": 23
+    },
+    {
+      "epoch": 0.09721518987341772,
+      "grad_norm": 0.8790615743752602,
+      "learning_rate": 7.804878048780489e-06,
+      "loss": 0.7959,
+      "step": 24
+    },
+    {
+      "epoch": 0.10126582278481013,
+      "grad_norm": 0.9575059596101914,
+      "learning_rate": 8.130081300813009e-06,
+      "loss": 0.8012,
+      "step": 25
+    },
+    {
+      "epoch": 0.10531645569620253,
+      "grad_norm": 0.9280608372841942,
+      "learning_rate": 8.45528455284553e-06,
+      "loss": 0.7903,
+      "step": 26
+    },
+    {
+      "epoch": 0.10936708860759493,
+      "grad_norm": 0.827193852176569,
+      "learning_rate": 8.78048780487805e-06,
+      "loss": 0.7723,
+      "step": 27
+    },
+    {
+      "epoch": 0.11341772151898734,
+      "grad_norm": 0.7167019125804834,
+      "learning_rate": 9.10569105691057e-06,
+      "loss": 0.7698,
+      "step": 28
+    },
+    {
+      "epoch": 0.11746835443037974,
+      "grad_norm": 0.8192621136077733,
+      "learning_rate": 9.43089430894309e-06,
+      "loss": 0.7814,
+      "step": 29
+    },
+    {
+      "epoch": 0.12151898734177215,
+      "grad_norm": 0.8117164551163373,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 0.775,
+      "step": 30
+    },
+    {
+      "epoch": 0.12556962025316457,
+      "grad_norm": 0.7975944746496687,
+      "learning_rate": 1.008130081300813e-05,
+      "loss": 0.7595,
+      "step": 31
+    },
+    {
+      "epoch": 0.12962025316455697,
+      "grad_norm": 0.690946209236594,
+      "learning_rate": 1.0406504065040652e-05,
+      "loss": 0.7588,
+      "step": 32
+    },
+    {
+      "epoch": 0.13367088607594937,
+      "grad_norm": 0.5769577653761417,
+      "learning_rate": 1.0731707317073172e-05,
+      "loss": 0.7335,
+      "step": 33
+    },
+    {
+      "epoch": 0.13772151898734178,
+      "grad_norm": 0.7019442068577123,
+      "learning_rate": 1.1056910569105692e-05,
+      "loss": 0.7381,
+      "step": 34
+    },
+    {
+      "epoch": 0.14177215189873418,
+      "grad_norm": 0.6101685211559655,
+      "learning_rate": 1.1382113821138213e-05,
+      "loss": 0.7216,
+      "step": 35
+    },
+    {
+      "epoch": 0.1458227848101266,
+      "grad_norm": 0.5891705789838885,
+      "learning_rate": 1.1707317073170731e-05,
+      "loss": 0.7406,
+      "step": 36
+    },
+    {
+      "epoch": 0.149873417721519,
+      "grad_norm": 0.6305744728664598,
+      "learning_rate": 1.2032520325203254e-05,
+      "loss": 0.7293,
+      "step": 37
+    },
+    {
+      "epoch": 0.1539240506329114,
+      "grad_norm": 0.6256351983244679,
+      "learning_rate": 1.2357723577235774e-05,
+      "loss": 0.7301,
+      "step": 38
+    },
+    {
+      "epoch": 0.1579746835443038,
+      "grad_norm": 0.5733539688789123,
+      "learning_rate": 1.2682926829268294e-05,
+      "loss": 0.7318,
+      "step": 39
+    },
+    {
+      "epoch": 0.1620253164556962,
+      "grad_norm": 0.5988323356918697,
+      "learning_rate": 1.3008130081300815e-05,
+      "loss": 0.726,
+      "step": 40
+    },
+    {
+      "epoch": 0.1660759493670886,
+      "grad_norm": 0.5662677854937769,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.7184,
+      "step": 41
+    },
+    {
+      "epoch": 0.17012658227848101,
+      "grad_norm": 0.7034334266567587,
+      "learning_rate": 1.3658536585365855e-05,
+      "loss": 0.7258,
+      "step": 42
+    },
+    {
+      "epoch": 0.17417721518987342,
+      "grad_norm": 0.47865206779565833,
+      "learning_rate": 1.3983739837398376e-05,
+      "loss": 0.7118,
+      "step": 43
+    },
+    {
+      "epoch": 0.17822784810126582,
+      "grad_norm": 0.7158855059041412,
+      "learning_rate": 1.4308943089430896e-05,
+      "loss": 0.7193,
+      "step": 44
+    },
+    {
+      "epoch": 0.18227848101265823,
+      "grad_norm": 0.613644438037742,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 0.7102,
+      "step": 45
+    },
+    {
+      "epoch": 0.18632911392405063,
+      "grad_norm": 0.5353958828047125,
+      "learning_rate": 1.4959349593495935e-05,
+      "loss": 0.7093,
+      "step": 46
+    },
+    {
+      "epoch": 0.19037974683544304,
+      "grad_norm": 0.6402487934139305,
+      "learning_rate": 1.528455284552846e-05,
+      "loss": 0.7212,
+      "step": 47
+    },
+    {
+      "epoch": 0.19443037974683544,
+      "grad_norm": 0.5196724506450475,
+      "learning_rate": 1.5609756097560978e-05,
+      "loss": 0.7071,
+      "step": 48
+    },
+    {
+      "epoch": 0.19848101265822785,
+      "grad_norm": 0.605594571392198,
+      "learning_rate": 1.5934959349593496e-05,
+      "loss": 0.7308,
+      "step": 49
+    },
+    {
+      "epoch": 0.20253164556962025,
+      "grad_norm": 0.5692759182757724,
+      "learning_rate": 1.6260162601626018e-05,
+      "loss": 0.7087,
+      "step": 50
+    },
+    {
+      "epoch": 0.20658227848101265,
+      "grad_norm": 0.4889127087195634,
+      "learning_rate": 1.6585365853658537e-05,
+      "loss": 0.7014,
+      "step": 51
+    },
+    {
+      "epoch": 0.21063291139240506,
+      "grad_norm": 0.5698234249913086,
+      "learning_rate": 1.691056910569106e-05,
+      "loss": 0.6918,
+      "step": 52
+    },
+    {
+      "epoch": 0.21468354430379746,
+      "grad_norm": 0.5587830913885997,
+      "learning_rate": 1.7235772357723578e-05,
+      "loss": 0.7032,
+      "step": 53
+    },
+    {
+      "epoch": 0.21873417721518987,
+      "grad_norm": 0.42178562394303387,
+      "learning_rate": 1.75609756097561e-05,
+      "loss": 0.6903,
+      "step": 54
+    },
+    {
+      "epoch": 0.22278481012658227,
+      "grad_norm": 0.528168761558852,
+      "learning_rate": 1.788617886178862e-05,
+      "loss": 0.6936,
+      "step": 55
+    },
+    {
+      "epoch": 0.22683544303797468,
+      "grad_norm": 0.45894566295079825,
+      "learning_rate": 1.821138211382114e-05,
+      "loss": 0.6972,
+      "step": 56
+    },
+    {
+      "epoch": 0.23088607594936708,
+      "grad_norm": 0.5205940503732954,
+      "learning_rate": 1.8536585365853663e-05,
+      "loss": 0.6953,
+      "step": 57
+    },
+    {
+      "epoch": 0.23493670886075949,
+      "grad_norm": 0.5328933702396752,
+      "learning_rate": 1.886178861788618e-05,
+      "loss": 0.683,
+      "step": 58
+    },
+    {
+      "epoch": 0.2389873417721519,
+      "grad_norm": 0.5859383835319693,
+      "learning_rate": 1.91869918699187e-05,
+      "loss": 0.6729,
+      "step": 59
+    },
+    {
+      "epoch": 0.2430379746835443,
+      "grad_norm": 0.876219210688,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 0.6923,
+      "step": 60
+    },
+    {
+      "epoch": 0.2470886075949367,
+      "grad_norm": 0.8321445898176479,
+      "learning_rate": 1.983739837398374e-05,
+      "loss": 0.6894,
+      "step": 61
+    },
+    {
+      "epoch": 0.25113924050632913,
+      "grad_norm": 0.4177323232451393,
+      "learning_rate": 2.016260162601626e-05,
+      "loss": 0.6861,
+      "step": 62
+    },
+    {
+      "epoch": 0.25518987341772154,
+      "grad_norm": 0.7163683731693709,
+      "learning_rate": 2.048780487804878e-05,
+      "loss": 0.6743,
+      "step": 63
+    },
+    {
+      "epoch": 0.25924050632911394,
+      "grad_norm": 0.8479006105219709,
+      "learning_rate": 2.0813008130081303e-05,
+      "loss": 0.694,
+      "step": 64
+    },
+    {
+      "epoch": 0.26329113924050634,
+      "grad_norm": 0.9580687668365239,
+      "learning_rate": 2.1138211382113822e-05,
+      "loss": 0.681,
+      "step": 65
+    },
+    {
+      "epoch": 0.26734177215189875,
+      "grad_norm": 0.908688259123547,
+      "learning_rate": 2.1463414634146344e-05,
+      "loss": 0.6903,
+      "step": 66
+    },
+    {
+      "epoch": 0.27139240506329115,
+      "grad_norm": 0.7178148302792761,
+      "learning_rate": 2.1788617886178863e-05,
+      "loss": 0.678,
+      "step": 67
+    },
+    {
+      "epoch": 0.27544303797468356,
+      "grad_norm": 0.5548662712194915,
+      "learning_rate": 2.2113821138211385e-05,
+      "loss": 0.6753,
+      "step": 68
+    },
+    {
+      "epoch": 0.27949367088607596,
+      "grad_norm": 0.6274669885320182,
+      "learning_rate": 2.2439024390243907e-05,
+      "loss": 0.6639,
+      "step": 69
+    },
+    {
+      "epoch": 0.28354430379746837,
+      "grad_norm": 0.7915273323797354,
+      "learning_rate": 2.2764227642276426e-05,
+      "loss": 0.6662,
+      "step": 70
+    },
+    {
+      "epoch": 0.28759493670886077,
+      "grad_norm": 0.9438369694736234,
+      "learning_rate": 2.3089430894308948e-05,
+      "loss": 0.689,
+      "step": 71
+    },
+    {
+      "epoch": 0.2916455696202532,
+      "grad_norm": 0.9885082648829144,
+      "learning_rate": 2.3414634146341463e-05,
+      "loss": 0.6733,
+      "step": 72
+    },
+    {
+      "epoch": 0.2956962025316456,
+      "grad_norm": 1.1184906470332134,
+      "learning_rate": 2.3739837398373985e-05,
+      "loss": 0.6699,
+      "step": 73
+    },
+    {
+      "epoch": 0.299746835443038,
+      "grad_norm": 0.9191161350996355,
+      "learning_rate": 2.4065040650406507e-05,
+      "loss": 0.6855,
+      "step": 74
+    },
+    {
+      "epoch": 0.3037974683544304,
+      "grad_norm": 0.8541969899063963,
+      "learning_rate": 2.4390243902439026e-05,
+      "loss": 0.6685,
+      "step": 75
+    },
+    {
+      "epoch": 0.3078481012658228,
+      "grad_norm": 0.8725581683886978,
+      "learning_rate": 2.4715447154471548e-05,
+      "loss": 0.6667,
+      "step": 76
+    },
+    {
+      "epoch": 0.3118987341772152,
+      "grad_norm": 1.2396251965092913,
+      "learning_rate": 2.5040650406504066e-05,
+      "loss": 0.6855,
+      "step": 77
+    },
+    {
+      "epoch": 0.3159493670886076,
+      "grad_norm": 1.5409086072055316,
+      "learning_rate": 2.536585365853659e-05,
+      "loss": 0.6821,
+      "step": 78
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.8101503418486055,
+      "learning_rate": 2.569105691056911e-05,
+      "loss": 0.6928,
+      "step": 79
+    },
+    {
+      "epoch": 0.3240506329113924,
+      "grad_norm": 0.938440848174476,
+      "learning_rate": 2.601626016260163e-05,
+      "loss": 0.7053,
+      "step": 80
+    },
+    {
+      "epoch": 0.3281012658227848,
+      "grad_norm": 0.8321664016319814,
+      "learning_rate": 2.634146341463415e-05,
+      "loss": 0.646,
+      "step": 81
+    },
+    {
+      "epoch": 0.3321518987341772,
+      "grad_norm": 0.9389177576313694,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.6641,
+      "step": 82
+    },
+    {
+      "epoch": 0.3362025316455696,
+      "grad_norm": 0.9199813652622566,
+      "learning_rate": 2.699186991869919e-05,
+      "loss": 0.6732,
+      "step": 83
+    },
+    {
+      "epoch": 0.34025316455696203,
+      "grad_norm": 0.7370619257005402,
+      "learning_rate": 2.731707317073171e-05,
+      "loss": 0.6722,
+      "step": 84
+    },
+    {
+      "epoch": 0.34430379746835443,
+      "grad_norm": 0.9756846931167642,
+      "learning_rate": 2.764227642276423e-05,
+      "loss": 0.6559,
+      "step": 85
+    },
+    {
+      "epoch": 0.34835443037974684,
+      "grad_norm": 0.8757597303397906,
+      "learning_rate": 2.796747967479675e-05,
+      "loss": 0.663,
+      "step": 86
+    },
+    {
+      "epoch": 0.35240506329113924,
+      "grad_norm": 0.7607712511576015,
+      "learning_rate": 2.829268292682927e-05,
+      "loss": 0.6726,
+      "step": 87
+    },
+    {
+      "epoch": 0.35645569620253165,
+      "grad_norm": 0.8664285334105873,
+      "learning_rate": 2.8617886178861792e-05,
+      "loss": 0.6626,
+      "step": 88
+    },
+    {
+      "epoch": 0.36050632911392405,
+      "grad_norm": 0.8667917358988141,
+      "learning_rate": 2.8943089430894314e-05,
+      "loss": 0.6635,
+      "step": 89
+    },
+    {
+      "epoch": 0.36455696202531646,
+      "grad_norm": 0.8972540126974701,
+      "learning_rate": 2.926829268292683e-05,
+      "loss": 0.6539,
+      "step": 90
+    },
+    {
+      "epoch": 0.36860759493670886,
+      "grad_norm": 0.9235623968422499,
+      "learning_rate": 2.959349593495935e-05,
+      "loss": 0.6705,
+      "step": 91
+    },
+    {
+      "epoch": 0.37265822784810126,
+      "grad_norm": 1.2814894626653324,
+      "learning_rate": 2.991869918699187e-05,
+      "loss": 0.6679,
+      "step": 92
+    },
+    {
+      "epoch": 0.37670886075949367,
+      "grad_norm": 1.023753306669639,
+      "learning_rate": 3.0243902439024392e-05,
+      "loss": 0.6817,
+      "step": 93
+    },
+    {
+      "epoch": 0.3807594936708861,
+      "grad_norm": 1.4606172075724613,
+      "learning_rate": 3.056910569105692e-05,
+      "loss": 0.6747,
+      "step": 94
+    },
+    {
+      "epoch": 0.3848101265822785,
+      "grad_norm": 1.0106531625712747,
+      "learning_rate": 3.089430894308943e-05,
+      "loss": 0.6483,
+      "step": 95
+    },
+    {
+      "epoch": 0.3888607594936709,
+      "grad_norm": 1.4523898882452213,
+      "learning_rate": 3.1219512195121955e-05,
+      "loss": 0.6742,
+      "step": 96
+    },
+    {
+      "epoch": 0.3929113924050633,
+      "grad_norm": 1.3318503634070047,
+      "learning_rate": 3.154471544715447e-05,
+      "loss": 0.6449,
+      "step": 97
+    },
+    {
+      "epoch": 0.3969620253164557,
+      "grad_norm": 1.1257344612786424,
+      "learning_rate": 3.186991869918699e-05,
+      "loss": 0.6692,
+      "step": 98
+    },
+    {
+      "epoch": 0.4010126582278481,
+      "grad_norm": 1.10039926101707,
+      "learning_rate": 3.2195121951219514e-05,
+      "loss": 0.6616,
+      "step": 99
+    },
+    {
+      "epoch": 0.4050632911392405,
+      "grad_norm": 1.323738927373706,
+      "learning_rate": 3.2520325203252037e-05,
+      "loss": 0.6714,
+      "step": 100
+    },
+    {
+      "epoch": 0.4091139240506329,
+      "grad_norm": 0.8269706513103124,
+      "learning_rate": 3.284552845528456e-05,
+      "loss": 0.6559,
+      "step": 101
+    },
+    {
+      "epoch": 0.4131645569620253,
+      "grad_norm": 1.1510047476214877,
+      "learning_rate": 3.3170731707317074e-05,
+      "loss": 0.6603,
+      "step": 102
+    },
+    {
+      "epoch": 0.4172151898734177,
+      "grad_norm": 1.1811998453669896,
+      "learning_rate": 3.3495934959349596e-05,
+      "loss": 0.6578,
+      "step": 103
+    },
+    {
+      "epoch": 0.4212658227848101,
+      "grad_norm": 1.1389960236633714,
+      "learning_rate": 3.382113821138212e-05,
+      "loss": 0.6615,
+      "step": 104
+    },
+    {
+      "epoch": 0.4253164556962025,
+      "grad_norm": 1.0616840489680561,
+      "learning_rate": 3.414634146341463e-05,
+      "loss": 0.6549,
+      "step": 105
+    },
+    {
+      "epoch": 0.4293670886075949,
+      "grad_norm": 1.2517088014762885,
+      "learning_rate": 3.4471544715447155e-05,
+      "loss": 0.6633,
+      "step": 106
+    },
+    {
+      "epoch": 0.43341772151898733,
+      "grad_norm": 0.877420773072822,
+      "learning_rate": 3.479674796747968e-05,
+      "loss": 0.6582,
+      "step": 107
+    },
+    {
+      "epoch": 0.43746835443037974,
+      "grad_norm": 1.576514191601641,
+      "learning_rate": 3.51219512195122e-05,
+      "loss": 0.6656,
+      "step": 108
+    },
+    {
+      "epoch": 0.44151898734177214,
+      "grad_norm": 0.982414660344291,
+      "learning_rate": 3.544715447154472e-05,
+      "loss": 0.6663,
+      "step": 109
+    },
+    {
+      "epoch": 0.44556962025316454,
+      "grad_norm": 1.3042821754290892,
+      "learning_rate": 3.577235772357724e-05,
+      "loss": 0.6619,
+      "step": 110
+    },
+    {
+      "epoch": 0.44962025316455695,
+      "grad_norm": 1.1682104250288972,
+      "learning_rate": 3.609756097560976e-05,
+      "loss": 0.6362,
+      "step": 111
+    },
+    {
+      "epoch": 0.45367088607594935,
+      "grad_norm": 1.1532433172721808,
+      "learning_rate": 3.642276422764228e-05,
+      "loss": 0.6514,
+      "step": 112
+    },
+    {
+      "epoch": 0.45772151898734176,
+      "grad_norm": 0.7698351577794085,
+      "learning_rate": 3.67479674796748e-05,
+      "loss": 0.6531,
+      "step": 113
+    },
+    {
+      "epoch": 0.46177215189873416,
+      "grad_norm": 1.1585563297080284,
+      "learning_rate": 3.7073170731707325e-05,
+      "loss": 0.664,
+      "step": 114
+    },
+    {
+      "epoch": 0.46582278481012657,
+      "grad_norm": 1.0743912406619456,
+      "learning_rate": 3.739837398373984e-05,
+      "loss": 0.6563,
+      "step": 115
+    },
+    {
+      "epoch": 0.46987341772151897,
+      "grad_norm": 1.190644430337651,
+      "learning_rate": 3.772357723577236e-05,
+      "loss": 0.6597,
+      "step": 116
+    },
+    {
+      "epoch": 0.4739240506329114,
+      "grad_norm": 1.2990786848649993,
+      "learning_rate": 3.804878048780488e-05,
+      "loss": 0.6407,
+      "step": 117
+    },
+    {
+      "epoch": 0.4779746835443038,
+      "grad_norm": 1.2116709380884327,
+      "learning_rate": 3.83739837398374e-05,
+      "loss": 0.6586,
+      "step": 118
+    },
+    {
+      "epoch": 0.4820253164556962,
+      "grad_norm": 0.693644711949389,
+      "learning_rate": 3.869918699186992e-05,
+      "loss": 0.6482,
+      "step": 119
+    },
+    {
+      "epoch": 0.4860759493670886,
+      "grad_norm": 0.9459888403658828,
+      "learning_rate": 3.9024390243902444e-05,
+      "loss": 0.6402,
+      "step": 120
+    },
+    {
+      "epoch": 0.490126582278481,
+      "grad_norm": 1.3056210793787857,
+      "learning_rate": 3.9349593495934966e-05,
+      "loss": 0.6536,
+      "step": 121
+    },
+    {
+      "epoch": 0.4941772151898734,
+      "grad_norm": 0.6038024346487839,
+      "learning_rate": 3.967479674796748e-05,
+      "loss": 0.6567,
+      "step": 122
+    },
+    {
+      "epoch": 0.4982278481012658,
+      "grad_norm": 0.9501988570014593,
+      "learning_rate": 4e-05,
+      "loss": 0.6514,
+      "step": 123
+    },
+    {
+      "epoch": 0.5022784810126583,
+      "grad_norm": 0.8253088692651466,
+      "learning_rate": 3.999991946137476e-05,
+      "loss": 0.6417,
+      "step": 124
+    },
+    {
+      "epoch": 0.5063291139240507,
+      "grad_norm": 1.2242635048614376,
+      "learning_rate": 3.999967784614766e-05,
+      "loss": 0.645,
+      "step": 125
+    },
+    {
+      "epoch": 0.5103797468354431,
+      "grad_norm": 0.9278249673401794,
+      "learning_rate": 3.9999275156264656e-05,
+      "loss": 0.6504,
+      "step": 126
+    },
+    {
+      "epoch": 0.5144303797468355,
+      "grad_norm": 1.0782436754172162,
+      "learning_rate": 3.999871139496895e-05,
+      "loss": 0.6503,
+      "step": 127
+    },
+    {
+      "epoch": 0.5184810126582279,
+      "grad_norm": 1.302731202788691,
+      "learning_rate": 3.9997986566800995e-05,
+      "loss": 0.6501,
+      "step": 128
+    },
+    {
+      "epoch": 0.5225316455696203,
+      "grad_norm": 0.8959617239680728,
+      "learning_rate": 3.999710067759846e-05,
+      "loss": 0.6479,
+      "step": 129
+    },
+    {
+      "epoch": 0.5265822784810127,
+      "grad_norm": 1.2611442032789741,
+      "learning_rate": 3.999605373449617e-05,
+      "loss": 0.6531,
+      "step": 130
+    },
+    {
+      "epoch": 0.5306329113924051,
+      "grad_norm": 0.6920194130755605,
+      "learning_rate": 3.9994845745926075e-05,
+      "loss": 0.6418,
+      "step": 131
+    },
+    {
+      "epoch": 0.5346835443037975,
+      "grad_norm": 0.7556754536484948,
+      "learning_rate": 3.999347672161713e-05,
+      "loss": 0.6396,
+      "step": 132
+    },
+    {
+      "epoch": 0.5387341772151899,
+      "grad_norm": 0.8391283000262155,
+      "learning_rate": 3.999194667259528e-05,
+      "loss": 0.6374,
+      "step": 133
+    },
+    {
+      "epoch": 0.5427848101265823,
+      "grad_norm": 0.7637413332943004,
+      "learning_rate": 3.999025561118334e-05,
+      "loss": 0.6427,
+      "step": 134
+    },
+    {
+      "epoch": 0.5468354430379747,
+      "grad_norm": 0.975483346992629,
+      "learning_rate": 3.998840355100086e-05,
+      "loss": 0.6398,
+      "step": 135
+    },
+    {
+      "epoch": 0.5508860759493671,
+      "grad_norm": 1.3487524590388251,
+      "learning_rate": 3.998639050696409e-05,
+      "loss": 0.6409,
+      "step": 136
+    },
+    {
+      "epoch": 0.5549367088607595,
+      "grad_norm": 0.8720392089581317,
+      "learning_rate": 3.998421649528582e-05,
+      "loss": 0.648,
+      "step": 137
+    },
+    {
+      "epoch": 0.5589873417721519,
+      "grad_norm": 1.0434536976384852,
+      "learning_rate": 3.9981881533475234e-05,
+      "loss": 0.6588,
+      "step": 138
+    },
+    {
+      "epoch": 0.5630379746835443,
+      "grad_norm": 0.9399284214467666,
+      "learning_rate": 3.997938564033779e-05,
+      "loss": 0.631,
+      "step": 139
+    },
+    {
+      "epoch": 0.5670886075949367,
+      "grad_norm": 1.0066043807805605,
+      "learning_rate": 3.9976728835975064e-05,
+      "loss": 0.6644,
+      "step": 140
+    },
+    {
+      "epoch": 0.5711392405063291,
+      "grad_norm": 1.0327983326224193,
+      "learning_rate": 3.9973911141784605e-05,
+      "loss": 0.6402,
+      "step": 141
+    },
+    {
+      "epoch": 0.5751898734177215,
+      "grad_norm": 1.0362714759021294,
+      "learning_rate": 3.997093258045973e-05,
+      "loss": 0.6571,
+      "step": 142
+    },
+    {
+      "epoch": 0.579240506329114,
+      "grad_norm": 1.0446934977150661,
+      "learning_rate": 3.996779317598936e-05,
+      "loss": 0.6524,
+      "step": 143
+    },
+    {
+      "epoch": 0.5832911392405064,
+      "grad_norm": 0.7867750021880091,
+      "learning_rate": 3.996449295365782e-05,
+      "loss": 0.6485,
+      "step": 144
+    },
+    {
+      "epoch": 0.5873417721518988,
+      "grad_norm": 0.7547247724967945,
+      "learning_rate": 3.996103194004467e-05,
+      "loss": 0.6387,
+      "step": 145
+    },
+    {
+      "epoch": 0.5913924050632912,
+      "grad_norm": 0.8209631256899239,
+      "learning_rate": 3.995741016302441e-05,
+      "loss": 0.6539,
+      "step": 146
+    },
+    {
+      "epoch": 0.5954430379746836,
+      "grad_norm": 0.6695544350644446,
+      "learning_rate": 3.9953627651766364e-05,
+      "loss": 0.644,
+      "step": 147
+    },
+    {
+      "epoch": 0.599493670886076,
+      "grad_norm": 1.0023469236349212,
+      "learning_rate": 3.9949684436734325e-05,
+      "loss": 0.6465,
+      "step": 148
+    },
+    {
+      "epoch": 0.6035443037974684,
+      "grad_norm": 0.8239375340823033,
+      "learning_rate": 3.994558054968643e-05,
+      "loss": 0.6483,
+      "step": 149
+    },
+    {
+      "epoch": 0.6075949367088608,
+      "grad_norm": 0.6447378923849817,
+      "learning_rate": 3.994131602367481e-05,
+      "loss": 0.6513,
+      "step": 150
+    },
+    {
+      "epoch": 0.6116455696202532,
+      "grad_norm": 0.8867838919436501,
+      "learning_rate": 3.9936890893045376e-05,
+      "loss": 0.6545,
+      "step": 151
+    },
+    {
+      "epoch": 0.6156962025316456,
+      "grad_norm": 0.7629881766077894,
+      "learning_rate": 3.993230519343752e-05,
+      "loss": 0.6351,
+      "step": 152
+    },
+    {
+      "epoch": 0.619746835443038,
+      "grad_norm": 0.724524605100859,
+      "learning_rate": 3.992755896178383e-05,
+      "loss": 0.6352,
+      "step": 153
+    },
+    {
+      "epoch": 0.6237974683544304,
+      "grad_norm": 1.125031681771964,
+      "learning_rate": 3.992265223630981e-05,
+      "loss": 0.6429,
+      "step": 154
+    },
+    {
+      "epoch": 0.6278481012658228,
+      "grad_norm": 0.8127951683768748,
+      "learning_rate": 3.991758505653355e-05,
+      "loss": 0.6392,
+      "step": 155
+    },
+    {
+      "epoch": 0.6318987341772152,
+      "grad_norm": 0.6062905621259084,
+      "learning_rate": 3.991235746326543e-05,
+      "loss": 0.6408,
+      "step": 156
+    },
+    {
+      "epoch": 0.6359493670886076,
+      "grad_norm": 0.6247685044175117,
+      "learning_rate": 3.9906969498607745e-05,
+      "loss": 0.6491,
+      "step": 157
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5977238269690631,
+      "learning_rate": 3.990142120595444e-05,
+      "loss": 0.638,
+      "step": 158
+    },
+    {
+      "epoch": 0.6440506329113924,
+      "grad_norm": 0.7998647152500128,
+      "learning_rate": 3.98957126299907e-05,
+      "loss": 0.6298,
+      "step": 159
+    },
+    {
+      "epoch": 0.6481012658227848,
+      "grad_norm": 0.8063944655162688,
+      "learning_rate": 3.9889843816692596e-05,
+      "loss": 0.6284,
+      "step": 160
+    },
+    {
+      "epoch": 0.6521518987341772,
+      "grad_norm": 0.754505486295503,
+      "learning_rate": 3.9883814813326766e-05,
+      "loss": 0.6379,
+      "step": 161
+    },
+    {
+      "epoch": 0.6562025316455696,
+      "grad_norm": 0.7082608260632913,
+      "learning_rate": 3.9877625668449956e-05,
+      "loss": 0.636,
+      "step": 162
+    },
+    {
+      "epoch": 0.660253164556962,
+      "grad_norm": 0.927054123811773,
+      "learning_rate": 3.98712764319087e-05,
+      "loss": 0.6417,
+      "step": 163
+    },
+    {
+      "epoch": 0.6643037974683544,
+      "grad_norm": 1.1147260609706908,
+      "learning_rate": 3.9864767154838864e-05,
+      "loss": 0.6354,
+      "step": 164
+    },
+    {
+      "epoch": 0.6683544303797468,
+      "grad_norm": 0.8527417634828821,
+      "learning_rate": 3.9858097889665277e-05,
+      "loss": 0.6291,
+      "step": 165
+    },
+    {
+      "epoch": 0.6724050632911392,
+      "grad_norm": 0.5920713441405214,
+      "learning_rate": 3.985126869010129e-05,
+      "loss": 0.6332,
+      "step": 166
+    },
+    {
+      "epoch": 0.6764556962025317,
+      "grad_norm": 0.8325268912168107,
+      "learning_rate": 3.984427961114833e-05,
+      "loss": 0.6361,
+      "step": 167
+    },
+    {
+      "epoch": 0.6805063291139241,
+      "grad_norm": 1.0083404644531042,
+      "learning_rate": 3.9837130709095475e-05,
+      "loss": 0.6275,
+      "step": 168
+    },
+    {
+      "epoch": 0.6845569620253165,
+      "grad_norm": 0.9250641837804169,
+      "learning_rate": 3.982982204151901e-05,
+      "loss": 0.6459,
+      "step": 169
+    },
+    {
+      "epoch": 0.6886075949367089,
+      "grad_norm": 0.7009433574624989,
+      "learning_rate": 3.982235366728193e-05,
+      "loss": 0.6312,
+      "step": 170
+    },
+    {
+      "epoch": 0.6926582278481013,
+      "grad_norm": 0.7092584479485344,
+      "learning_rate": 3.9814725646533505e-05,
+      "loss": 0.635,
+      "step": 171
+    },
+    {
+      "epoch": 0.6967088607594937,
+      "grad_norm": 1.0441333549274805,
+      "learning_rate": 3.9806938040708746e-05,
+      "loss": 0.6248,
+      "step": 172
+    },
+    {
+      "epoch": 0.7007594936708861,
+      "grad_norm": 0.8482517862198069,
+      "learning_rate": 3.9798990912527976e-05,
+      "loss": 0.6257,
+      "step": 173
+    },
+    {
+      "epoch": 0.7048101265822785,
+      "grad_norm": 0.8055830690306399,
+      "learning_rate": 3.979088432599627e-05,
+      "loss": 0.6246,
+      "step": 174
+    },
+    {
+      "epoch": 0.7088607594936709,
+      "grad_norm": 0.9297406465193896,
+      "learning_rate": 3.9782618346402964e-05,
+      "loss": 0.6225,
+      "step": 175
+    },
+    {
+      "epoch": 0.7129113924050633,
+      "grad_norm": 0.7041120657719969,
+      "learning_rate": 3.977419304032111e-05,
+      "loss": 0.6369,
+      "step": 176
+    },
+    {
+      "epoch": 0.7169620253164557,
+      "grad_norm": 0.8104723322922875,
+      "learning_rate": 3.976560847560697e-05,
+      "loss": 0.629,
+      "step": 177
+    },
+    {
+      "epoch": 0.7210126582278481,
+      "grad_norm": 0.8704026305168808,
+      "learning_rate": 3.9756864721399456e-05,
+      "loss": 0.6328,
+      "step": 178
+    },
+    {
+      "epoch": 0.7250632911392405,
+      "grad_norm": 0.6968053758350573,
+      "learning_rate": 3.974796184811956e-05,
+      "loss": 0.6417,
+      "step": 179
+    },
+    {
+      "epoch": 0.7291139240506329,
+      "grad_norm": 0.7866238465554998,
+      "learning_rate": 3.973889992746979e-05,
+      "loss": 0.6343,
+      "step": 180
+    },
+    {
+      "epoch": 0.7331645569620253,
+      "grad_norm": 0.7546643019806617,
+      "learning_rate": 3.972967903243361e-05,
+      "loss": 0.6292,
+      "step": 181
+    },
+    {
+      "epoch": 0.7372151898734177,
+      "grad_norm": 0.6575391302830953,
+      "learning_rate": 3.972029923727486e-05,
+      "loss": 0.6296,
+      "step": 182
+    },
+    {
+      "epoch": 0.7412658227848101,
+      "grad_norm": 0.6552694711659468,
+      "learning_rate": 3.971076061753709e-05,
+      "loss": 0.6272,
+      "step": 183
+    },
+    {
+      "epoch": 0.7453164556962025,
+      "grad_norm": 0.5862296611393707,
+      "learning_rate": 3.9701063250043066e-05,
+      "loss": 0.6449,
+      "step": 184
+    },
+    {
+      "epoch": 0.7493670886075949,
+      "grad_norm": 0.7018844045130789,
+      "learning_rate": 3.969120721289402e-05,
+      "loss": 0.6292,
+      "step": 185
+    },
+    {
+      "epoch": 0.7534177215189873,
+      "grad_norm": 0.770689689786004,
+      "learning_rate": 3.9681192585469146e-05,
+      "loss": 0.6264,
+      "step": 186
+    },
+    {
+      "epoch": 0.7574683544303797,
+      "grad_norm": 0.5954084338872954,
+      "learning_rate": 3.9671019448424865e-05,
+      "loss": 0.6481,
+      "step": 187
+    },
+    {
+      "epoch": 0.7615189873417721,
+      "grad_norm": 0.6199980301172392,
+      "learning_rate": 3.966068788369422e-05,
+      "loss": 0.6148,
+      "step": 188
+    },
+    {
+      "epoch": 0.7655696202531646,
+      "grad_norm": 0.8173013119370307,
+      "learning_rate": 3.965019797448622e-05,
+      "loss": 0.6178,
+      "step": 189
+    },
+    {
+      "epoch": 0.769620253164557,
+      "grad_norm": 0.6639036096412569,
+      "learning_rate": 3.963954980528515e-05,
+      "loss": 0.6187,
+      "step": 190
+    },
+    {
+      "epoch": 0.7736708860759494,
+      "grad_norm": 0.6436499514124145,
+      "learning_rate": 3.9628743461849905e-05,
+      "loss": 0.6237,
+      "step": 191
+    },
+    {
+      "epoch": 0.7777215189873418,
+      "grad_norm": 0.6683554874025449,
+      "learning_rate": 3.961777903121329e-05,
+      "loss": 0.6176,
+      "step": 192
+    },
+    {
+      "epoch": 0.7817721518987342,
+      "grad_norm": 0.5946361394746551,
+      "learning_rate": 3.960665660168131e-05,
+      "loss": 0.6348,
+      "step": 193
+    },
+    {
+      "epoch": 0.7858227848101266,
+      "grad_norm": 0.6987813122869841,
+      "learning_rate": 3.9595376262832485e-05,
+      "loss": 0.6325,
+      "step": 194
+    },
+    {
+      "epoch": 0.789873417721519,
+      "grad_norm": 0.6345625278771495,
+      "learning_rate": 3.9583938105517127e-05,
+      "loss": 0.6343,
+      "step": 195
+    },
+    {
+      "epoch": 0.7939240506329114,
+      "grad_norm": 0.6243560896666179,
+      "learning_rate": 3.957234222185657e-05,
+      "loss": 0.6251,
+      "step": 196
+    },
+    {
+      "epoch": 0.7979746835443038,
+      "grad_norm": 0.42394021137602217,
+      "learning_rate": 3.9560588705242474e-05,
+      "loss": 0.631,
+      "step": 197
+    },
+    {
+      "epoch": 0.8020253164556962,
+      "grad_norm": 0.5133227649035279,
+      "learning_rate": 3.954867765033605e-05,
+      "loss": 0.622,
+      "step": 198
+    },
+    {
+      "epoch": 0.8060759493670886,
+      "grad_norm": 0.5664403169301392,
+      "learning_rate": 3.953660915306728e-05,
+      "loss": 0.6307,
+      "step": 199
+    },
+    {
+      "epoch": 0.810126582278481,
+      "grad_norm": 0.5873814918521808,
+      "learning_rate": 3.952438331063419e-05,
+      "loss": 0.6256,
+      "step": 200
+    },
+    {
+      "epoch": 0.8141772151898734,
+      "grad_norm": 0.5921761560279892,
+      "learning_rate": 3.951200022150205e-05,
+      "loss": 0.617,
+      "step": 201
+    },
+    {
+      "epoch": 0.8182278481012658,
+      "grad_norm": 0.4593284340406492,
+      "learning_rate": 3.949945998540253e-05,
+      "loss": 0.6245,
+      "step": 202
+    },
+    {
+      "epoch": 0.8222784810126582,
+      "grad_norm": 0.4025412904328983,
+      "learning_rate": 3.9486762703332993e-05,
+      "loss": 0.6282,
+      "step": 203
+    },
+    {
+      "epoch": 0.8263291139240506,
+      "grad_norm": 0.514394896832602,
+      "learning_rate": 3.947390847755559e-05,
+      "loss": 0.6297,
+      "step": 204
+    },
+    {
+      "epoch": 0.830379746835443,
+      "grad_norm": 0.5070889778573305,
+      "learning_rate": 3.946089741159648e-05,
+      "loss": 0.6271,
+      "step": 205
+    },
+    {
+      "epoch": 0.8344303797468354,
+      "grad_norm": 0.4001566423476405,
+      "learning_rate": 3.944772961024501e-05,
+      "loss": 0.6076,
+      "step": 206
+    },
+    {
+      "epoch": 0.8384810126582278,
+      "grad_norm": 0.5059910345064784,
+      "learning_rate": 3.943440517955285e-05,
+      "loss": 0.5985,
+      "step": 207
+    },
+    {
+      "epoch": 0.8425316455696202,
+      "grad_norm": 0.4516563432525227,
+      "learning_rate": 3.9420924226833126e-05,
+      "loss": 0.6243,
+      "step": 208
+    },
+    {
+      "epoch": 0.8465822784810126,
+      "grad_norm": 0.4734921831089004,
+      "learning_rate": 3.9407286860659566e-05,
+      "loss": 0.6303,
+      "step": 209
+    },
+    {
+      "epoch": 0.850632911392405,
+      "grad_norm": 0.5430777152806864,
+      "learning_rate": 3.9393493190865657e-05,
+      "loss": 0.6172,
+      "step": 210
+    },
+    {
+      "epoch": 0.8546835443037974,
+      "grad_norm": 0.5758613367441068,
+      "learning_rate": 3.937954332854371e-05,
+      "loss": 0.6206,
+      "step": 211
+    },
+    {
+      "epoch": 0.8587341772151899,
+      "grad_norm": 0.5597300745515237,
+      "learning_rate": 3.9365437386044016e-05,
+      "loss": 0.6119,
+      "step": 212
+    },
+    {
+      "epoch": 0.8627848101265823,
+      "grad_norm": 0.43951264693051123,
+      "learning_rate": 3.935117547697387e-05,
+      "loss": 0.6203,
+      "step": 213
+    },
+    {
+      "epoch": 0.8668354430379747,
+      "grad_norm": 0.5567908436119582,
+      "learning_rate": 3.933675771619675e-05,
+      "loss": 0.618,
+      "step": 214
+    },
+    {
+      "epoch": 0.8708860759493671,
+      "grad_norm": 0.6327523165811465,
+      "learning_rate": 3.932218421983131e-05,
+      "loss": 0.629,
+      "step": 215
+    },
+    {
+      "epoch": 0.8749367088607595,
+      "grad_norm": 0.4694935520148255,
+      "learning_rate": 3.9307455105250484e-05,
+      "loss": 0.6209,
+      "step": 216
+    },
+    {
+      "epoch": 0.8789873417721519,
+      "grad_norm": 0.6597972893140409,
+      "learning_rate": 3.929257049108054e-05,
+      "loss": 0.6131,
+      "step": 217
+    },
+    {
+      "epoch": 0.8830379746835443,
+      "grad_norm": 0.6056798092961302,
+      "learning_rate": 3.927753049720011e-05,
+      "loss": 0.6271,
+      "step": 218
+    },
+    {
+      "epoch": 0.8870886075949367,
+      "grad_norm": 0.5251909161727825,
+      "learning_rate": 3.9262335244739234e-05,
+      "loss": 0.6034,
+      "step": 219
+    },
+    {
+      "epoch": 0.8911392405063291,
+      "grad_norm": 0.5392373280824562,
+      "learning_rate": 3.92469848560784e-05,
+      "loss": 0.6192,
+      "step": 220
+    },
+    {
+      "epoch": 0.8951898734177215,
+      "grad_norm": 0.5956728647242585,
+      "learning_rate": 3.923147945484751e-05,
+      "loss": 0.6259,
+      "step": 221
+    },
+    {
+      "epoch": 0.8992405063291139,
+      "grad_norm": 0.772140946606301,
+      "learning_rate": 3.9215819165924956e-05,
+      "loss": 0.6059,
+      "step": 222
+    },
+    {
+      "epoch": 0.9032911392405063,
+      "grad_norm": 0.5915662728507896,
+      "learning_rate": 3.920000411543654e-05,
+      "loss": 0.616,
+      "step": 223
+    },
+    {
+      "epoch": 0.9073417721518987,
+      "grad_norm": 1.0215867361300608,
+      "learning_rate": 3.9184034430754495e-05,
+      "loss": 0.6567,
+      "step": 224
+    },
+    {
+      "epoch": 0.9113924050632911,
+      "grad_norm": 0.5724789295692316,
+      "learning_rate": 3.916791024049648e-05,
+      "loss": 0.6202,
+      "step": 225
+    },
+    {
+      "epoch": 0.9154430379746835,
+      "grad_norm": 0.4914124777922318,
+      "learning_rate": 3.91516316745245e-05,
+      "loss": 0.6224,
+      "step": 226
+    },
+    {
+      "epoch": 0.9194936708860759,
+      "grad_norm": 0.6910453743353475,
+      "learning_rate": 3.913519886394389e-05,
+      "loss": 0.6093,
+      "step": 227
+    },
+    {
+      "epoch": 0.9235443037974683,
+      "grad_norm": 0.6799953363105539,
+      "learning_rate": 3.911861194110225e-05,
+      "loss": 0.6314,
+      "step": 228
+    },
+    {
+      "epoch": 0.9275949367088607,
+      "grad_norm": 0.45218789514147056,
+      "learning_rate": 3.910187103958837e-05,
+      "loss": 0.6141,
+      "step": 229
+    },
+    {
+      "epoch": 0.9316455696202531,
+      "grad_norm": 0.5538080788761198,
+      "learning_rate": 3.908497629423117e-05,
+      "loss": 0.6241,
+      "step": 230
+    },
+    {
+      "epoch": 0.9356962025316455,
+      "grad_norm": 0.6817103486777154,
+      "learning_rate": 3.9067927841098614e-05,
+      "loss": 0.606,
+      "step": 231
+    },
+    {
+      "epoch": 0.9397468354430379,
+      "grad_norm": 0.5900744785058248,
+      "learning_rate": 3.9050725817496594e-05,
+      "loss": 0.6262,
+      "step": 232
+    },
+    {
+      "epoch": 0.9437974683544303,
+      "grad_norm": 0.5173282908682416,
+      "learning_rate": 3.9033370361967844e-05,
+      "loss": 0.6207,
+      "step": 233
+    },
+    {
+      "epoch": 0.9478481012658228,
+      "grad_norm": 0.5870588325745825,
+      "learning_rate": 3.901586161429081e-05,
+      "loss": 0.6252,
+      "step": 234
+    },
+    {
+      "epoch": 0.9518987341772152,
+      "grad_norm": 0.5079477338056164,
+      "learning_rate": 3.8998199715478545e-05,
+      "loss": 0.6098,
+      "step": 235
+    },
+    {
+      "epoch": 0.9559493670886076,
+      "grad_norm": 0.430226790105489,
+      "learning_rate": 3.8980384807777564e-05,
+      "loss": 0.621,
+      "step": 236
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5615392859050055,
+      "learning_rate": 3.896241703466667e-05,
+      "loss": 0.6071,
+      "step": 237
+    },
+    {
+      "epoch": 0.9640506329113924,
+      "grad_norm": 0.523253852130537,
+      "learning_rate": 3.894429654085585e-05,
+      "loss": 0.6122,
+      "step": 238
+    },
+    {
+      "epoch": 0.9681012658227848,
+      "grad_norm": 0.4420528335765794,
+      "learning_rate": 3.892602347228505e-05,
+      "loss": 0.6157,
+      "step": 239
+    },
+    {
+      "epoch": 0.9721518987341772,
+      "grad_norm": 0.41115212736282225,
+      "learning_rate": 3.890759797612307e-05,
+      "loss": 0.6152,
+      "step": 240
+    },
+    {
+      "epoch": 0.9762025316455696,
+      "grad_norm": 0.44636143335550216,
+      "learning_rate": 3.888902020076632e-05,
+      "loss": 0.6005,
+      "step": 241
+    },
+    {
+      "epoch": 0.980253164556962,
+      "grad_norm": 0.4019075301602835,
+      "learning_rate": 3.887029029583764e-05,
+      "loss": 0.6227,
+      "step": 242
+    },
+    {
+      "epoch": 0.9843037974683544,
+      "grad_norm": 0.4306300872708938,
+      "learning_rate": 3.8851408412185125e-05,
+      "loss": 0.6205,
+      "step": 243
+    },
+    {
+      "epoch": 0.9883544303797468,
+      "grad_norm": 0.4508673735411875,
+      "learning_rate": 3.8832374701880855e-05,
+      "loss": 0.6195,
+      "step": 244
+    },
+    {
+      "epoch": 0.9924050632911392,
+      "grad_norm": 0.4460284974173539,
+      "learning_rate": 3.881318931821972e-05,
+      "loss": 0.6137,
+      "step": 245
+    },
+    {
+      "epoch": 0.9964556962025316,
+      "grad_norm": 0.6620733383461588,
+      "learning_rate": 3.879385241571817e-05,
+      "loss": 0.6271,
+      "step": 246
+    },
+    {
+      "epoch": 1.000506329113924,
+      "grad_norm": 0.6418500118441146,
+      "learning_rate": 3.8774364150112955e-05,
+      "loss": 0.6275,
+      "step": 247
+    },
+    {
+      "epoch": 1.0045569620253165,
+      "grad_norm": 0.8136349045612613,
+      "learning_rate": 3.8754724678359884e-05,
+      "loss": 0.5777,
+      "step": 248
+    },
+    {
+      "epoch": 1.0086075949367088,
+      "grad_norm": 0.6679632495308905,
+      "learning_rate": 3.873493415863256e-05,
+      "loss": 0.5711,
+      "step": 249
+    },
+    {
+      "epoch": 1.0126582278481013,
+      "grad_norm": 0.5959319129414856,
+      "learning_rate": 3.871499275032111e-05,
+      "loss": 0.5787,
+      "step": 250
+    },
+    {
+      "epoch": 1.0167088607594936,
+      "grad_norm": 0.6430696388574605,
+      "learning_rate": 3.869490061403091e-05,
+      "loss": 0.5673,
+      "step": 251
+    },
+    {
+      "epoch": 1.0207594936708861,
+      "grad_norm": 0.544984912850782,
+      "learning_rate": 3.867465791158124e-05,
+      "loss": 0.5683,
+      "step": 252
+    },
+    {
+      "epoch": 1.0248101265822784,
+      "grad_norm": 0.7362762563722631,
+      "learning_rate": 3.865426480600407e-05,
+      "loss": 0.5694,
+      "step": 253
+    },
+    {
+      "epoch": 1.028860759493671,
+      "grad_norm": 0.5794696380523973,
+      "learning_rate": 3.863372146154264e-05,
+      "loss": 0.5729,
+      "step": 254
+    },
+    {
+      "epoch": 1.0329113924050632,
+      "grad_norm": 0.5733014934160112,
+      "learning_rate": 3.861302804365024e-05,
+      "loss": 0.5661,
+      "step": 255
+    },
+    {
+      "epoch": 1.0369620253164558,
+      "grad_norm": 0.4786303469517065,
+      "learning_rate": 3.85921847189888e-05,
+      "loss": 0.5655,
+      "step": 256
+    },
+    {
+      "epoch": 1.041012658227848,
+      "grad_norm": 0.6826478883843755,
+      "learning_rate": 3.85711916554276e-05,
+      "loss": 0.5712,
+      "step": 257
+    },
+    {
+      "epoch": 1.0450632911392406,
+      "grad_norm": 0.9364724026112793,
+      "learning_rate": 3.85500490220419e-05,
+      "loss": 0.5795,
+      "step": 258
+    },
+    {
+      "epoch": 1.0491139240506329,
+      "grad_norm": 1.0475582845392084,
+      "learning_rate": 3.852875698911154e-05,
+      "loss": 0.573,
+      "step": 259
+    },
+    {
+      "epoch": 1.0531645569620254,
+      "grad_norm": 0.6083372017149284,
+      "learning_rate": 3.850731572811963e-05,
+      "loss": 0.5616,
+      "step": 260
+    },
+    {
+      "epoch": 1.0572151898734177,
+      "grad_norm": 0.5604894555151325,
+      "learning_rate": 3.848572541175116e-05,
+      "loss": 0.5785,
+      "step": 261
+    },
+    {
+      "epoch": 1.0612658227848102,
+      "grad_norm": 0.7821848710201438,
+      "learning_rate": 3.846398621389154e-05,
+      "loss": 0.5725,
+      "step": 262
+    },
+    {
+      "epoch": 1.0653164556962025,
+      "grad_norm": 0.9868004882114855,
+      "learning_rate": 3.84420983096253e-05,
+      "loss": 0.5707,
+      "step": 263
+    },
+    {
+      "epoch": 1.069367088607595,
+      "grad_norm": 0.9270090414739185,
+      "learning_rate": 3.8420061875234606e-05,
+      "loss": 0.5639,
+      "step": 264
+    },
+    {
+      "epoch": 1.0734177215189873,
+      "grad_norm": 0.5531781731155561,
+      "learning_rate": 3.839787708819787e-05,
+      "loss": 0.5804,
+      "step": 265
+    },
+    {
+      "epoch": 1.0774683544303798,
+      "grad_norm": 0.4481409615302291,
+      "learning_rate": 3.8375544127188325e-05,
+      "loss": 0.5616,
+      "step": 266
+    },
+    {
+      "epoch": 1.081518987341772,
+      "grad_norm": 0.654105100578237,
+      "learning_rate": 3.8353063172072564e-05,
+      "loss": 0.5877,
+      "step": 267
+    },
+    {
+      "epoch": 1.0855696202531646,
+      "grad_norm": 0.7447750137008016,
+      "learning_rate": 3.8330434403909105e-05,
+      "loss": 0.5651,
+      "step": 268
+    },
+    {
+      "epoch": 1.089620253164557,
+      "grad_norm": 0.7392800029886211,
+      "learning_rate": 3.8307658004946934e-05,
+      "loss": 0.5773,
+      "step": 269
+    },
+    {
+      "epoch": 1.0936708860759494,
+      "grad_norm": 0.44204022611256316,
+      "learning_rate": 3.8284734158624046e-05,
+      "loss": 0.5705,
+      "step": 270
+    },
+    {
+      "epoch": 1.0977215189873417,
+      "grad_norm": 0.6077340454471977,
+      "learning_rate": 3.826166304956594e-05,
+      "loss": 0.5836,
+      "step": 271
+    },
+    {
+      "epoch": 1.1017721518987342,
+      "grad_norm": 0.6151265821529015,
+      "learning_rate": 3.8238444863584164e-05,
+      "loss": 0.5664,
+      "step": 272
+    },
+    {
+      "epoch": 1.1058227848101265,
+      "grad_norm": 0.5465061730099202,
+      "learning_rate": 3.821507978767479e-05,
+      "loss": 0.5698,
+      "step": 273
+    },
+    {
+      "epoch": 1.109873417721519,
+      "grad_norm": 0.5265938501273955,
+      "learning_rate": 3.819156801001693e-05,
+      "loss": 0.5685,
+      "step": 274
+    },
+    {
+      "epoch": 1.1139240506329113,
+      "grad_norm": 0.5577967604452362,
+      "learning_rate": 3.816790971997121e-05,
+      "loss": 0.5548,
+      "step": 275
+    },
+    {
+      "epoch": 1.1179746835443038,
+      "grad_norm": 0.5439938843787003,
+      "learning_rate": 3.8144105108078246e-05,
+      "loss": 0.5675,
+      "step": 276
+    },
+    {
+      "epoch": 1.1220253164556961,
+      "grad_norm": 0.4788564833120996,
+      "learning_rate": 3.81201543660571e-05,
+      "loss": 0.563,
+      "step": 277
+    },
+    {
+      "epoch": 1.1260759493670887,
+      "grad_norm": 0.4632396498105618,
+      "learning_rate": 3.809605768680377e-05,
+      "loss": 0.5781,
+      "step": 278
+    },
+    {
+      "epoch": 1.130126582278481,
+      "grad_norm": 0.3850028625713948,
+      "learning_rate": 3.807181526438958e-05,
+      "loss": 0.5644,
+      "step": 279
+    },
+    {
+      "epoch": 1.1341772151898735,
+      "grad_norm": 0.4671010608108467,
+      "learning_rate": 3.8047427294059697e-05,
+      "loss": 0.573,
+      "step": 280
+    },
+    {
+      "epoch": 1.1382278481012658,
+      "grad_norm": 0.4784008004531782,
+      "learning_rate": 3.802289397223145e-05,
+      "loss": 0.5732,
+      "step": 281
+    },
+    {
+      "epoch": 1.1422784810126583,
+      "grad_norm": 0.5321991694312344,
+      "learning_rate": 3.7998215496492854e-05,
+      "loss": 0.5734,
+      "step": 282
+    },
+    {
+      "epoch": 1.1463291139240506,
+      "grad_norm": 0.5474546588175947,
+      "learning_rate": 3.797339206560096e-05,
+      "loss": 0.5528,
+      "step": 283
+    },
+    {
+      "epoch": 1.150379746835443,
+      "grad_norm": 0.8267056413786941,
+      "learning_rate": 3.794842387948027e-05,
+      "loss": 0.5517,
+      "step": 284
+    },
+    {
+      "epoch": 1.1544303797468354,
+      "grad_norm": 0.5577504862577234,
+      "learning_rate": 3.7923311139221114e-05,
+      "loss": 0.5619,
+      "step": 285
+    },
+    {
+      "epoch": 1.158481012658228,
+      "grad_norm": 0.604140833797346,
+      "learning_rate": 3.7898054047078054e-05,
+      "loss": 0.5596,
+      "step": 286
+    },
+    {
+      "epoch": 1.1625316455696202,
+      "grad_norm": 0.7261600093058989,
+      "learning_rate": 3.787265280646825e-05,
+      "loss": 0.5672,
+      "step": 287
+    },
+    {
+      "epoch": 1.1665822784810127,
+      "grad_norm": 0.6340066996352606,
+      "learning_rate": 3.7847107621969786e-05,
+      "loss": 0.5807,
+      "step": 288
+    },
+    {
+      "epoch": 1.170632911392405,
+      "grad_norm": 0.47832867406444857,
+      "learning_rate": 3.7821418699320064e-05,
+      "loss": 0.5804,
+      "step": 289
+    },
+    {
+      "epoch": 1.1746835443037975,
+      "grad_norm": 0.7214954007778146,
+      "learning_rate": 3.7795586245414145e-05,
+      "loss": 0.5761,
+      "step": 290
+    },
+    {
+      "epoch": 1.1787341772151898,
+      "grad_norm": 0.6149356343973613,
+      "learning_rate": 3.776961046830306e-05,
+      "loss": 0.5618,
+      "step": 291
+    },
+    {
+      "epoch": 1.1827848101265823,
+      "grad_norm": 0.4342127772328317,
+      "learning_rate": 3.774349157719215e-05,
+      "loss": 0.5636,
+      "step": 292
+    },
+    {
+      "epoch": 1.1868354430379746,
+      "grad_norm": 0.4910519801600507,
+      "learning_rate": 3.7717229782439365e-05,
+      "loss": 0.5727,
+      "step": 293
+    },
+    {
+      "epoch": 1.1908860759493671,
+      "grad_norm": 0.5622865206645108,
+      "learning_rate": 3.769082529555359e-05,
+      "loss": 0.5598,
+      "step": 294
+    },
+    {
+      "epoch": 1.1949367088607594,
+      "grad_norm": 0.5042233658795227,
+      "learning_rate": 3.766427832919294e-05,
+      "loss": 0.5717,
+      "step": 295
+    },
+    {
+      "epoch": 1.198987341772152,
+      "grad_norm": 0.4521143189669999,
+      "learning_rate": 3.7637589097163024e-05,
+      "loss": 0.5786,
+      "step": 296
+    },
+    {
+      "epoch": 1.2030379746835442,
+      "grad_norm": 0.4750591611700796,
+      "learning_rate": 3.761075781441526e-05,
+      "loss": 0.5714,
+      "step": 297
+    },
+    {
+      "epoch": 1.2070886075949367,
+      "grad_norm": 0.49772921047605395,
+      "learning_rate": 3.75837846970451e-05,
+      "loss": 0.5665,
+      "step": 298
+    },
+    {
+      "epoch": 1.211139240506329,
+      "grad_norm": 0.5187791996271776,
+      "learning_rate": 3.755666996229032e-05,
+      "loss": 0.5686,
+      "step": 299
+    },
+    {
+      "epoch": 1.2151898734177216,
+      "grad_norm": 0.43418979528967366,
+      "learning_rate": 3.752941382852927e-05,
+      "loss": 0.5635,
+      "step": 300
+    },
+    {
+      "epoch": 1.2192405063291138,
+      "grad_norm": 0.45446475345980974,
+      "learning_rate": 3.7502016515279115e-05,
+      "loss": 0.5791,
+      "step": 301
+    },
+    {
+      "epoch": 1.2232911392405064,
+      "grad_norm": 0.5002420918045158,
+      "learning_rate": 3.7474478243194043e-05,
+      "loss": 0.5781,
+      "step": 302
+    },
+    {
+      "epoch": 1.2273417721518987,
+      "grad_norm": 0.46192092640484517,
+      "learning_rate": 3.744679923406351e-05,
+      "loss": 0.5759,
+      "step": 303
+    },
+    {
+      "epoch": 1.2313924050632912,
+      "grad_norm": 0.49398100213978385,
+      "learning_rate": 3.741897971081043e-05,
+      "loss": 0.5869,
+      "step": 304
+    },
+    {
+      "epoch": 1.2354430379746835,
+      "grad_norm": 0.4633583815593787,
+      "learning_rate": 3.739101989748946e-05,
+      "loss": 0.5776,
+      "step": 305
+    },
+    {
+      "epoch": 1.239493670886076,
+      "grad_norm": 0.4760405598663205,
+      "learning_rate": 3.7362920019285066e-05,
+      "loss": 0.5612,
+      "step": 306
+    },
+    {
+      "epoch": 1.2435443037974683,
+      "grad_norm": 0.5218169703446232,
+      "learning_rate": 3.73346803025098e-05,
+      "loss": 0.5548,
+      "step": 307
+    },
+    {
+      "epoch": 1.2475949367088608,
+      "grad_norm": 0.4124239763044585,
+      "learning_rate": 3.730630097460247e-05,
+      "loss": 0.5573,
+      "step": 308
+    },
+    {
+      "epoch": 1.251645569620253,
+      "grad_norm": 0.3813194489460432,
+      "learning_rate": 3.727778226412628e-05,
+      "loss": 0.5813,
+      "step": 309
+    },
+    {
+      "epoch": 1.2556962025316456,
+      "grad_norm": 0.3956505750304584,
+      "learning_rate": 3.7249124400767006e-05,
+      "loss": 0.5553,
+      "step": 310
+    },
+    {
+      "epoch": 1.259746835443038,
+      "grad_norm": 0.4489438400980702,
+      "learning_rate": 3.722032761533114e-05,
+      "loss": 0.5758,
+      "step": 311
+    },
+    {
+      "epoch": 1.2637974683544304,
+      "grad_norm": 0.4261762223514116,
+      "learning_rate": 3.719139213974403e-05,
+      "loss": 0.5745,
+      "step": 312
+    },
+    {
+      "epoch": 1.2678481012658227,
+      "grad_norm": 0.5208839999933685,
+      "learning_rate": 3.7162318207048006e-05,
+      "loss": 0.5695,
+      "step": 313
+    },
+    {
+      "epoch": 1.2718987341772152,
+      "grad_norm": 0.4560594848976192,
+      "learning_rate": 3.713310605140055e-05,
+      "loss": 0.57,
+      "step": 314
+    },
+    {
+      "epoch": 1.2759493670886077,
+      "grad_norm": 0.4098600569326759,
+      "learning_rate": 3.710375590807233e-05,
+      "loss": 0.5661,
+      "step": 315
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5284606714857439,
+      "learning_rate": 3.7074268013445365e-05,
+      "loss": 0.5712,
+      "step": 316
+    },
+    {
+      "epoch": 1.2840506329113923,
+      "grad_norm": 0.5665583719157021,
+      "learning_rate": 3.7044642605011114e-05,
+      "loss": 0.5539,
+      "step": 317
+    },
+    {
+      "epoch": 1.2881012658227848,
+      "grad_norm": 0.3734766948490748,
+      "learning_rate": 3.701487992136854e-05,
+      "loss": 0.558,
+      "step": 318
+    },
+    {
+      "epoch": 1.2921518987341774,
+      "grad_norm": 0.45982261326228385,
+      "learning_rate": 3.69849802022222e-05,
+      "loss": 0.5739,
+      "step": 319
+    },
+    {
+      "epoch": 1.2962025316455696,
+      "grad_norm": 0.5568446846581506,
+      "learning_rate": 3.6954943688380334e-05,
+      "loss": 0.554,
+      "step": 320
+    },
+    {
+      "epoch": 1.300253164556962,
+      "grad_norm": 0.48720325598199,
+      "learning_rate": 3.692477062175289e-05,
+      "loss": 0.5744,
+      "step": 321
+    },
+    {
+      "epoch": 1.3043037974683545,
+      "grad_norm": 0.47503573126401366,
+      "learning_rate": 3.689446124534958e-05,
+      "loss": 0.5618,
+      "step": 322
+    },
+    {
+      "epoch": 1.308354430379747,
+      "grad_norm": 0.4955521238904994,
+      "learning_rate": 3.686401580327799e-05,
+      "loss": 0.5704,
+      "step": 323
+    },
+    {
+      "epoch": 1.3124050632911393,
+      "grad_norm": 0.5701111210791449,
+      "learning_rate": 3.683343454074149e-05,
+      "loss": 0.5672,
+      "step": 324
+    },
+    {
+      "epoch": 1.3164556962025316,
+      "grad_norm": 0.5991840620159615,
+      "learning_rate": 3.6802717704037386e-05,
+      "loss": 0.5653,
+      "step": 325
+    },
+    {
+      "epoch": 1.320506329113924,
+      "grad_norm": 0.4326079889304455,
+      "learning_rate": 3.6771865540554855e-05,
+      "loss": 0.561,
+      "step": 326
+    },
+    {
+      "epoch": 1.3245569620253166,
+      "grad_norm": 0.6060498110477124,
+      "learning_rate": 3.674087829877297e-05,
+      "loss": 0.5641,
+      "step": 327
+    },
+    {
+      "epoch": 1.3286075949367089,
+      "grad_norm": 0.46841216648579825,
+      "learning_rate": 3.6709756228258735e-05,
+      "loss": 0.5674,
+      "step": 328
+    },
+    {
+      "epoch": 1.3326582278481012,
+      "grad_norm": 0.48762204452299246,
+      "learning_rate": 3.667849957966501e-05,
+      "loss": 0.555,
+      "step": 329
+    },
+    {
+      "epoch": 1.3367088607594937,
+      "grad_norm": 0.6354441274008239,
+      "learning_rate": 3.6647108604728546e-05,
+      "loss": 0.5642,
+      "step": 330
+    },
+    {
+      "epoch": 1.3407594936708862,
+      "grad_norm": 0.4331916494331695,
+      "learning_rate": 3.661558355626795e-05,
+      "loss": 0.5755,
+      "step": 331
+    },
+    {
+      "epoch": 1.3448101265822785,
+      "grad_norm": 0.5679493126451431,
+      "learning_rate": 3.658392468818163e-05,
+      "loss": 0.5804,
+      "step": 332
+    },
+    {
+      "epoch": 1.3488607594936708,
+      "grad_norm": 0.40107613533144937,
+      "learning_rate": 3.655213225544574e-05,
+      "loss": 0.5653,
+      "step": 333
+    },
+    {
+      "epoch": 1.3529113924050633,
+      "grad_norm": 0.5278411014832515,
+      "learning_rate": 3.652020651411218e-05,
+      "loss": 0.5662,
+      "step": 334
+    },
+    {
+      "epoch": 1.3569620253164558,
+      "grad_norm": 0.8087628657193692,
+      "learning_rate": 3.6488147721306474e-05,
+      "loss": 0.5792,
+      "step": 335
+    },
+    {
+      "epoch": 1.3610126582278481,
+      "grad_norm": 0.565821913963553,
+      "learning_rate": 3.645595613522574e-05,
+      "loss": 0.5654,
+      "step": 336
+    },
+    {
+      "epoch": 1.3650632911392404,
+      "grad_norm": 0.4144549187979962,
+      "learning_rate": 3.642363201513657e-05,
+      "loss": 0.5732,
+      "step": 337
+    },
+    {
+      "epoch": 1.369113924050633,
+      "grad_norm": 0.44148708280180815,
+      "learning_rate": 3.6391175621373006e-05,
+      "loss": 0.5849,
+      "step": 338
+    },
+    {
+      "epoch": 1.3731645569620254,
+      "grad_norm": 0.459234796309001,
+      "learning_rate": 3.6358587215334355e-05,
+      "loss": 0.5661,
+      "step": 339
+    },
+    {
+      "epoch": 1.3772151898734177,
+      "grad_norm": 0.49716284367234853,
+      "learning_rate": 3.632586705948318e-05,
+      "loss": 0.5594,
+      "step": 340
+    },
+    {
+      "epoch": 1.38126582278481,
+      "grad_norm": 0.4460806845987647,
+      "learning_rate": 3.629301541734311e-05,
+      "loss": 0.571,
+      "step": 341
+    },
+    {
+      "epoch": 1.3853164556962025,
+      "grad_norm": 0.4228637449047461,
+      "learning_rate": 3.626003255349676e-05,
+      "loss": 0.556,
+      "step": 342
+    },
+    {
+      "epoch": 1.389367088607595,
+      "grad_norm": 0.5402862633113682,
+      "learning_rate": 3.622691873358357e-05,
+      "loss": 0.5609,
+      "step": 343
+    },
+    {
+      "epoch": 1.3934177215189874,
+      "grad_norm": 0.4294271921972551,
+      "learning_rate": 3.61936742242977e-05,
+      "loss": 0.5661,
+      "step": 344
+    },
+    {
+      "epoch": 1.3974683544303796,
+      "grad_norm": 0.44944697172949083,
+      "learning_rate": 3.6160299293385864e-05,
+      "loss": 0.5652,
+      "step": 345
+    },
+    {
+      "epoch": 1.4015189873417722,
+      "grad_norm": 0.5691192956470723,
+      "learning_rate": 3.612679420964516e-05,
+      "loss": 0.5711,
+      "step": 346
+    },
+    {
+      "epoch": 1.4055696202531647,
+      "grad_norm": 0.5500694656266102,
+      "learning_rate": 3.609315924292092e-05,
+      "loss": 0.5637,
+      "step": 347
+    },
+    {
+      "epoch": 1.409620253164557,
+      "grad_norm": 0.46131748533801237,
+      "learning_rate": 3.6059394664104554e-05,
+      "loss": 0.5681,
+      "step": 348
+    },
+    {
+      "epoch": 1.4136708860759493,
+      "grad_norm": 0.4152374368192273,
+      "learning_rate": 3.602550074513133e-05,
+      "loss": 0.5586,
+      "step": 349
+    },
+    {
+      "epoch": 1.4177215189873418,
+      "grad_norm": 0.4341469396917868,
+      "learning_rate": 3.599147775897822e-05,
+      "loss": 0.5771,
+      "step": 350
+    },
+    {
+      "epoch": 1.4217721518987343,
+      "grad_norm": 0.5410925171648765,
+      "learning_rate": 3.595732597966167e-05,
+      "loss": 0.5621,
+      "step": 351
+    },
+    {
+      "epoch": 1.4258227848101266,
+      "grad_norm": 0.4796295241268698,
+      "learning_rate": 3.592304568223542e-05,
+      "loss": 0.5717,
+      "step": 352
+    },
+    {
+      "epoch": 1.4298734177215189,
+      "grad_norm": 0.4834425445129118,
+      "learning_rate": 3.588863714278826e-05,
+      "loss": 0.5649,
+      "step": 353
+    },
+    {
+      "epoch": 1.4339240506329114,
+      "grad_norm": 0.42641702387885366,
+      "learning_rate": 3.585410063844186e-05,
+      "loss": 0.567,
+      "step": 354
+    },
+    {
+      "epoch": 1.437974683544304,
+      "grad_norm": 0.46702983446490315,
+      "learning_rate": 3.581943644734846e-05,
+      "loss": 0.5623,
+      "step": 355
+    },
+    {
+      "epoch": 1.4420253164556962,
+      "grad_norm": 0.4326465828298296,
+      "learning_rate": 3.578464484868869e-05,
+      "loss": 0.562,
+      "step": 356
+    },
+    {
+      "epoch": 1.4460759493670885,
+      "grad_norm": 0.3817051379162761,
+      "learning_rate": 3.5749726122669316e-05,
+      "loss": 0.561,
+      "step": 357
+    },
+    {
+      "epoch": 1.450126582278481,
+      "grad_norm": 0.5183492358266165,
+      "learning_rate": 3.5714680550520943e-05,
+      "loss": 0.5713,
+      "step": 358
+    },
+    {
+      "epoch": 1.4541772151898735,
+      "grad_norm": 0.40302051392272603,
+      "learning_rate": 3.5679508414495794e-05,
+      "loss": 0.5557,
+      "step": 359
+    },
+    {
+      "epoch": 1.4582278481012658,
+      "grad_norm": 0.45659039739758844,
+      "learning_rate": 3.564420999786543e-05,
+      "loss": 0.5641,
+      "step": 360
+    },
+    {
+      "epoch": 1.4622784810126581,
+      "grad_norm": 0.5052864103375094,
+      "learning_rate": 3.560878558491842e-05,
+      "loss": 0.562,
+      "step": 361
+    },
+    {
+      "epoch": 1.4663291139240506,
+      "grad_norm": 0.3591904510237819,
+      "learning_rate": 3.5573235460958145e-05,
+      "loss": 0.5713,
+      "step": 362
+    },
+    {
+      "epoch": 1.4703797468354431,
+      "grad_norm": 0.484705643054314,
+      "learning_rate": 3.553755991230039e-05,
+      "loss": 0.572,
+      "step": 363
+    },
+    {
+      "epoch": 1.4744303797468354,
+      "grad_norm": 0.5847081519338315,
+      "learning_rate": 3.5501759226271144e-05,
+      "loss": 0.5634,
+      "step": 364
+    },
+    {
+      "epoch": 1.4784810126582277,
+      "grad_norm": 0.36075565446748264,
+      "learning_rate": 3.546583369120419e-05,
+      "loss": 0.5524,
+      "step": 365
+    },
+    {
+      "epoch": 1.4825316455696202,
+      "grad_norm": 0.6210643434420019,
+      "learning_rate": 3.5429783596438864e-05,
+      "loss": 0.5756,
+      "step": 366
+    },
+    {
+      "epoch": 1.4865822784810128,
+      "grad_norm": 0.3720838570619001,
+      "learning_rate": 3.539360923231766e-05,
+      "loss": 0.5407,
+      "step": 367
+    },
+    {
+      "epoch": 1.490632911392405,
+      "grad_norm": 0.6287444935546549,
+      "learning_rate": 3.535731089018394e-05,
+      "loss": 0.5546,
+      "step": 368
+    },
+    {
+      "epoch": 1.4946835443037974,
+      "grad_norm": 0.4119614114449107,
+      "learning_rate": 3.532088886237956e-05,
+      "loss": 0.5678,
+      "step": 369
+    },
+    {
+      "epoch": 1.4987341772151899,
+      "grad_norm": 0.37700690106090856,
+      "learning_rate": 3.528434344224253e-05,
+      "loss": 0.5609,
+      "step": 370
+    },
+    {
+      "epoch": 1.5027848101265824,
+      "grad_norm": 0.38895852858783275,
+      "learning_rate": 3.524767492410464e-05,
+      "loss": 0.5708,
+      "step": 371
+    },
+    {
+      "epoch": 1.5068354430379747,
+      "grad_norm": 0.32893314811964247,
+      "learning_rate": 3.521088360328908e-05,
+      "loss": 0.5729,
+      "step": 372
+    },
+    {
+      "epoch": 1.510886075949367,
+      "grad_norm": 0.36577077425748256,
+      "learning_rate": 3.517396977610811e-05,
+      "loss": 0.5722,
+      "step": 373
+    },
+    {
+      "epoch": 1.5149367088607595,
+      "grad_norm": 0.390019931651527,
+      "learning_rate": 3.5136933739860595e-05,
+      "loss": 0.5659,
+      "step": 374
+    },
+    {
+      "epoch": 1.518987341772152,
+      "grad_norm": 0.3971568099709169,
+      "learning_rate": 3.509977579282971e-05,
+      "loss": 0.5566,
+      "step": 375
+    },
+    {
+      "epoch": 1.5230379746835443,
+      "grad_norm": 0.3544722332260646,
+      "learning_rate": 3.5062496234280424e-05,
+      "loss": 0.5614,
+      "step": 376
+    },
+    {
+      "epoch": 1.5270886075949366,
+      "grad_norm": 0.4458496144858772,
+      "learning_rate": 3.502509536445719e-05,
+      "loss": 0.5695,
+      "step": 377
+    },
+    {
+      "epoch": 1.531139240506329,
+      "grad_norm": 0.3979654963668021,
+      "learning_rate": 3.498757348458147e-05,
+      "loss": 0.5642,
+      "step": 378
+    },
+    {
+      "epoch": 1.5351898734177216,
+      "grad_norm": 0.4058374632122085,
+      "learning_rate": 3.4949930896849324e-05,
+      "loss": 0.5508,
+      "step": 379
+    },
+    {
+      "epoch": 1.539240506329114,
+      "grad_norm": 0.4201794747659719,
+      "learning_rate": 3.491216790442899e-05,
+      "loss": 0.5598,
+      "step": 380
+    },
+    {
+      "epoch": 1.5432911392405062,
+      "grad_norm": 0.4823909516903219,
+      "learning_rate": 3.487428481145839e-05,
+      "loss": 0.5635,
+      "step": 381
+    },
+    {
+      "epoch": 1.5473417721518987,
+      "grad_norm": 0.4001764569996043,
+      "learning_rate": 3.483628192304278e-05,
+      "loss": 0.5758,
+      "step": 382
+    },
+    {
+      "epoch": 1.5513924050632912,
+      "grad_norm": 0.4430158042646442,
+      "learning_rate": 3.479815954525219e-05,
+      "loss": 0.5741,
+      "step": 383
+    },
+    {
+      "epoch": 1.5554430379746835,
+      "grad_norm": 0.5015113531009421,
+      "learning_rate": 3.475991798511899e-05,
+      "loss": 0.5667,
+      "step": 384
+    },
+    {
+      "epoch": 1.5594936708860758,
+      "grad_norm": 0.44574776756511597,
+      "learning_rate": 3.4721557550635464e-05,
+      "loss": 0.5429,
+      "step": 385
+    },
+    {
+      "epoch": 1.5635443037974683,
+      "grad_norm": 0.48628074820711237,
+      "learning_rate": 3.468307855075128e-05,
+      "loss": 0.5573,
+      "step": 386
+    },
+    {
+      "epoch": 1.5675949367088609,
+      "grad_norm": 0.4171158331348657,
+      "learning_rate": 3.4644481295371005e-05,
+      "loss": 0.5597,
+      "step": 387
+    },
+    {
+      "epoch": 1.5716455696202531,
+      "grad_norm": 0.4980757453679588,
+      "learning_rate": 3.460576609535163e-05,
+      "loss": 0.5704,
+      "step": 388
+    },
+    {
+      "epoch": 1.5756962025316454,
+      "grad_norm": 0.4991580186867127,
+      "learning_rate": 3.456693326250006e-05,
+      "loss": 0.5655,
+      "step": 389
+    },
+    {
+      "epoch": 1.579746835443038,
+      "grad_norm": 0.44632101714158634,
+      "learning_rate": 3.452798310957058e-05,
+      "loss": 0.5734,
+      "step": 390
+    },
+    {
+      "epoch": 1.5837974683544305,
+      "grad_norm": 0.5361407563259633,
+      "learning_rate": 3.4488915950262386e-05,
+      "loss": 0.5599,
+      "step": 391
+    },
+    {
+      "epoch": 1.5878481012658228,
+      "grad_norm": 0.37411128435943447,
+      "learning_rate": 3.4449732099216985e-05,
+      "loss": 0.5529,
+      "step": 392
+    },
+    {
+      "epoch": 1.591898734177215,
+      "grad_norm": 0.4274363852876439,
+      "learning_rate": 3.441043187201574e-05,
+      "loss": 0.5608,
+      "step": 393
+    },
+    {
+      "epoch": 1.5959493670886076,
+      "grad_norm": 0.4530128065273998,
+      "learning_rate": 3.437101558517728e-05,
+      "loss": 0.5607,
+      "step": 394
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.38859768369743847,
+      "learning_rate": 3.433148355615496e-05,
+      "loss": 0.5534,
+      "step": 395
+    },
+    {
+      "epoch": 1.6040506329113924,
+      "grad_norm": 0.3907763281294501,
+      "learning_rate": 3.4291836103334294e-05,
+      "loss": 0.5665,
+      "step": 396
+    },
+    {
+      "epoch": 1.6081012658227847,
+      "grad_norm": 0.4296485522923601,
+      "learning_rate": 3.425207354603043e-05,
+      "loss": 0.5706,
+      "step": 397
+    },
+    {
+      "epoch": 1.6121518987341772,
+      "grad_norm": 0.9738686688112259,
+      "learning_rate": 3.421219620448553e-05,
+      "loss": 0.5886,
+      "step": 398
+    },
+    {
+      "epoch": 1.6162025316455697,
+      "grad_norm": 0.37202346378857615,
+      "learning_rate": 3.417220439986623e-05,
+      "loss": 0.5592,
+      "step": 399
+    },
+    {
+      "epoch": 1.620253164556962,
+      "grad_norm": 0.3517455356638131,
+      "learning_rate": 3.4132098454261024e-05,
+      "loss": 0.5503,
+      "step": 400
+    },
+    {
+      "epoch": 1.6243037974683543,
+      "grad_norm": 0.3714583975084134,
+      "learning_rate": 3.4091878690677676e-05,
+      "loss": 0.5637,
+      "step": 401
+    },
+    {
+      "epoch": 1.6283544303797468,
+      "grad_norm": 0.4315757763590048,
+      "learning_rate": 3.405154543304065e-05,
+      "loss": 0.5639,
+      "step": 402
+    },
+    {
+      "epoch": 1.6324050632911393,
+      "grad_norm": 0.3606273302213119,
+      "learning_rate": 3.401109900618843e-05,
+      "loss": 0.5593,
+      "step": 403
+    },
+    {
+      "epoch": 1.6364556962025316,
+      "grad_norm": 0.37433422651982035,
+      "learning_rate": 3.3970539735870996e-05,
+      "loss": 0.5586,
+      "step": 404
+    },
+    {
+      "epoch": 1.640506329113924,
+      "grad_norm": 0.38994927726081396,
+      "learning_rate": 3.392986794874714e-05,
+      "loss": 0.5857,
+      "step": 405
+    },
+    {
+      "epoch": 1.6445569620253164,
+      "grad_norm": 0.35384427917061045,
+      "learning_rate": 3.388908397238184e-05,
+      "loss": 0.5565,
+      "step": 406
+    },
+    {
+      "epoch": 1.648607594936709,
+      "grad_norm": 0.3749177926523868,
+      "learning_rate": 3.384818813524362e-05,
+      "loss": 0.5644,
+      "step": 407
+    },
+    {
+      "epoch": 1.6526582278481012,
+      "grad_norm": 0.40744537467391445,
+      "learning_rate": 3.380718076670195e-05,
+      "loss": 0.5682,
+      "step": 408
+    },
+    {
+      "epoch": 1.6567088607594935,
+      "grad_norm": 0.3374378542026639,
+      "learning_rate": 3.376606219702454e-05,
+      "loss": 0.5485,
+      "step": 409
+    },
+    {
+      "epoch": 1.660759493670886,
+      "grad_norm": 0.38997606864547935,
+      "learning_rate": 3.372483275737468e-05,
+      "loss": 0.5523,
+      "step": 410
+    },
+    {
+      "epoch": 1.6648101265822786,
+      "grad_norm": 0.39377639677763737,
+      "learning_rate": 3.368349277980861e-05,
+      "loss": 0.5702,
+      "step": 411
+    },
+    {
+      "epoch": 1.6688607594936709,
+      "grad_norm": 0.3399824097454189,
+      "learning_rate": 3.3642042597272844e-05,
+      "loss": 0.5551,
+      "step": 412
+    },
+    {
+      "epoch": 1.6729113924050631,
+      "grad_norm": 0.48036500985737474,
+      "learning_rate": 3.360048254360144e-05,
+      "loss": 0.5424,
+      "step": 413
+    },
+    {
+      "epoch": 1.6769620253164557,
+      "grad_norm": 0.41399368049320473,
+      "learning_rate": 3.355881295351336e-05,
+      "loss": 0.5554,
+      "step": 414
+    },
+    {
+      "epoch": 1.6810126582278482,
+      "grad_norm": 0.33102567097569896,
+      "learning_rate": 3.351703416260975e-05,
+      "loss": 0.5723,
+      "step": 415
+    },
+    {
+      "epoch": 1.6850632911392405,
+      "grad_norm": 0.4338522702080373,
+      "learning_rate": 3.347514650737126e-05,
+      "loss": 0.5527,
+      "step": 416
+    },
+    {
+      "epoch": 1.6891139240506328,
+      "grad_norm": 0.5082517618045325,
+      "learning_rate": 3.3433150325155295e-05,
+      "loss": 0.5608,
+      "step": 417
+    },
+    {
+      "epoch": 1.6931645569620253,
+      "grad_norm": 0.6569446513598711,
+      "learning_rate": 3.339104595419334e-05,
+      "loss": 0.5742,
+      "step": 418
+    },
+    {
+      "epoch": 1.6972151898734178,
+      "grad_norm": 0.4797359533509065,
+      "learning_rate": 3.3348833733588204e-05,
+      "loss": 0.5627,
+      "step": 419
+    },
+    {
+      "epoch": 1.70126582278481,
+      "grad_norm": 0.46410871729456027,
+      "learning_rate": 3.3306514003311305e-05,
+      "loss": 0.5476,
+      "step": 420
+    },
+    {
+      "epoch": 1.7053164556962024,
+      "grad_norm": 0.43730157510071466,
+      "learning_rate": 3.326408710419996e-05,
+      "loss": 0.5525,
+      "step": 421
+    },
+    {
+      "epoch": 1.709367088607595,
+      "grad_norm": 1.3847396850244527,
+      "learning_rate": 3.322155337795454e-05,
+      "loss": 0.6289,
+      "step": 422
+    },
+    {
+      "epoch": 1.7134177215189874,
+      "grad_norm": 0.4759994029565324,
+      "learning_rate": 3.317891316713587e-05,
+      "loss": 0.5515,
+      "step": 423
+    },
+    {
+      "epoch": 1.7174683544303797,
+      "grad_norm": 0.48818694054454465,
+      "learning_rate": 3.313616681516231e-05,
+      "loss": 0.5617,
+      "step": 424
+    },
+    {
+      "epoch": 1.721518987341772,
+      "grad_norm": 0.37052441370988803,
+      "learning_rate": 3.309331466630713e-05,
+      "loss": 0.5541,
+      "step": 425
+    },
+    {
+      "epoch": 1.7255696202531645,
+      "grad_norm": 0.48728887317194824,
+      "learning_rate": 3.305035706569563e-05,
+      "loss": 0.565,
+      "step": 426
+    },
+    {
+      "epoch": 1.729620253164557,
+      "grad_norm": 0.3597893912190749,
+      "learning_rate": 3.3007294359302433e-05,
+      "loss": 0.5604,
+      "step": 427
+    },
+    {
+      "epoch": 1.7336708860759493,
+      "grad_norm": 1.948440789957399,
+      "learning_rate": 3.296412689394864e-05,
+      "loss": 0.6033,
+      "step": 428
+    },
+    {
+      "epoch": 1.7377215189873416,
+      "grad_norm": 0.45047860459174066,
+      "learning_rate": 3.292085501729909e-05,
+      "loss": 0.5616,
+      "step": 429
+    },
+    {
+      "epoch": 1.7417721518987341,
+      "grad_norm": 0.49539757552531155,
+      "learning_rate": 3.2877479077859534e-05,
+      "loss": 0.563,
+      "step": 430
+    },
+    {
+      "epoch": 1.7458227848101266,
+      "grad_norm": 1.5366198532745963,
+      "learning_rate": 3.283399942497381e-05,
+      "loss": 0.5803,
+      "step": 431
+    },
+    {
+      "epoch": 1.749873417721519,
+      "grad_norm": 0.48515112648520387,
+      "learning_rate": 3.279041640882108e-05,
+      "loss": 0.5446,
+      "step": 432
+    },
+    {
+      "epoch": 1.7539240506329112,
+      "grad_norm": 0.6776618873669393,
+      "learning_rate": 3.2746730380412964e-05,
+      "loss": 0.551,
+      "step": 433
+    },
+    {
+      "epoch": 1.7579746835443038,
+      "grad_norm": 0.6091075678160074,
+      "learning_rate": 3.2702941691590726e-05,
+      "loss": 0.5557,
+      "step": 434
+    },
+    {
+      "epoch": 1.7620253164556963,
+      "grad_norm": 0.3605769201712125,
+      "learning_rate": 3.265905069502244e-05,
+      "loss": 0.5705,
+      "step": 435
+    },
+    {
+      "epoch": 1.7660759493670886,
+      "grad_norm": 0.6147153597335849,
+      "learning_rate": 3.261505774420016e-05,
+      "loss": 0.5677,
+      "step": 436
+    },
+    {
+      "epoch": 1.7701265822784809,
+      "grad_norm": 0.463706053990247,
+      "learning_rate": 3.257096319343707e-05,
+      "loss": 0.5464,
+      "step": 437
+    },
+    {
+      "epoch": 1.7741772151898734,
+      "grad_norm": 0.40563931003837356,
+      "learning_rate": 3.2526767397864614e-05,
+      "loss": 0.5739,
+      "step": 438
+    },
+    {
+      "epoch": 1.7782278481012659,
+      "grad_norm": 0.5380648207543253,
+      "learning_rate": 3.248247071342966e-05,
+      "loss": 0.5525,
+      "step": 439
+    },
+    {
+      "epoch": 1.7822784810126582,
+      "grad_norm": 0.44567268177178354,
+      "learning_rate": 3.243807349689161e-05,
+      "loss": 0.5618,
+      "step": 440
+    },
+    {
+      "epoch": 1.7863291139240505,
+      "grad_norm": 0.31617209032968163,
+      "learning_rate": 3.2393576105819544e-05,
+      "loss": 0.5522,
+      "step": 441
+    },
+    {
+      "epoch": 1.790379746835443,
+      "grad_norm": 0.4681969902820236,
+      "learning_rate": 3.2348978898589333e-05,
+      "loss": 0.5601,
+      "step": 442
+    },
+    {
+      "epoch": 1.7944303797468355,
+      "grad_norm": 0.4267087271577437,
+      "learning_rate": 3.230428223438075e-05,
+      "loss": 0.5621,
+      "step": 443
+    },
+    {
+      "epoch": 1.7984810126582278,
+      "grad_norm": 0.36322900088690185,
+      "learning_rate": 3.225948647317459e-05,
+      "loss": 0.5564,
+      "step": 444
+    },
+    {
+      "epoch": 1.80253164556962,
+      "grad_norm": 0.4827186927692806,
+      "learning_rate": 3.2214591975749745e-05,
+      "loss": 0.5653,
+      "step": 445
+    },
+    {
+      "epoch": 1.8065822784810126,
+      "grad_norm": 0.5243280384305498,
+      "learning_rate": 3.216959910368034e-05,
+      "loss": 0.5596,
+      "step": 446
+    },
+    {
+      "epoch": 1.8106329113924051,
+      "grad_norm": 0.43288911624323223,
+      "learning_rate": 3.212450821933277e-05,
+      "loss": 0.5667,
+      "step": 447
+    },
+    {
+      "epoch": 1.8146835443037974,
+      "grad_norm": 0.5458570337085531,
+      "learning_rate": 3.207931968586281e-05,
+      "loss": 0.5711,
+      "step": 448
+    },
+    {
+      "epoch": 1.81873417721519,
+      "grad_norm": 0.4518765392996167,
+      "learning_rate": 3.203403386721272e-05,
+      "loss": 0.5576,
+      "step": 449
+    },
+    {
+      "epoch": 1.8227848101265822,
+      "grad_norm": 0.36828223449644804,
+      "learning_rate": 3.1988651128108245e-05,
+      "loss": 0.5584,
+      "step": 450
+    },
+    {
+      "epoch": 1.8268354430379747,
+      "grad_norm": 0.4780858049999546,
+      "learning_rate": 3.194317183405573e-05,
+      "loss": 0.5547,
+      "step": 451
+    },
+    {
+      "epoch": 1.8308860759493673,
+      "grad_norm": 0.38170681631362263,
+      "learning_rate": 3.189759635133914e-05,
+      "loss": 0.5629,
+      "step": 452
+    },
+    {
+      "epoch": 1.8349367088607595,
+      "grad_norm": 0.38638104531552125,
+      "learning_rate": 3.185192504701718e-05,
+      "loss": 0.5549,
+      "step": 453
+    },
+    {
+      "epoch": 1.8389873417721518,
+      "grad_norm": 0.5152744170682028,
+      "learning_rate": 3.1806158288920234e-05,
+      "loss": 0.5661,
+      "step": 454
+    },
+    {
+      "epoch": 1.8430379746835444,
+      "grad_norm": 0.36362911125642816,
+      "learning_rate": 3.1760296445647477e-05,
+      "loss": 0.5613,
+      "step": 455
+    },
+    {
+      "epoch": 1.8470886075949369,
+      "grad_norm": 0.5314218296868294,
+      "learning_rate": 3.1714339886563896e-05,
+      "loss": 0.557,
+      "step": 456
+    },
+    {
+      "epoch": 1.8511392405063292,
+      "grad_norm": 0.5330374162530912,
+      "learning_rate": 3.166828898179731e-05,
+      "loss": 0.5668,
+      "step": 457
+    },
+    {
+      "epoch": 1.8551898734177215,
+      "grad_norm": 0.38083526010136615,
+      "learning_rate": 3.162214410223536e-05,
+      "loss": 0.5717,
+      "step": 458
+    },
+    {
+      "epoch": 1.859240506329114,
+      "grad_norm": 0.4549635085910908,
+      "learning_rate": 3.157590561952257e-05,
+      "loss": 0.5418,
+      "step": 459
+    },
+    {
+      "epoch": 1.8632911392405065,
+      "grad_norm": 0.3353554170207971,
+      "learning_rate": 3.152957390605732e-05,
+      "loss": 0.5634,
+      "step": 460
+    },
+    {
+      "epoch": 1.8673417721518988,
+      "grad_norm": 0.40993223932982603,
+      "learning_rate": 3.148314933498886e-05,
+      "loss": 0.5584,
+      "step": 461
+    },
+    {
+      "epoch": 1.871392405063291,
+      "grad_norm": 0.39536476993258535,
+      "learning_rate": 3.143663228021431e-05,
+      "loss": 0.5607,
+      "step": 462
+    },
+    {
+      "epoch": 1.8754430379746836,
+      "grad_norm": 0.38751214929238653,
+      "learning_rate": 3.1390023116375624e-05,
+      "loss": 0.5596,
+      "step": 463
+    },
+    {
+      "epoch": 1.879493670886076,
+      "grad_norm": 0.4264220016952534,
+      "learning_rate": 3.134332221885661e-05,
+      "loss": 0.5487,
+      "step": 464
+    },
+    {
+      "epoch": 1.8835443037974684,
+      "grad_norm": 0.40442386064997504,
+      "learning_rate": 3.129652996377987e-05,
+      "loss": 0.5688,
+      "step": 465
+    },
+    {
+      "epoch": 1.8875949367088607,
+      "grad_norm": 0.4381700586484839,
+      "learning_rate": 3.12496467280038e-05,
+      "loss": 0.5744,
+      "step": 466
+    },
+    {
+      "epoch": 1.8916455696202532,
+      "grad_norm": 0.34892290560351114,
+      "learning_rate": 3.120267288911952e-05,
+      "loss": 0.567,
+      "step": 467
+    },
+    {
+      "epoch": 1.8956962025316457,
+      "grad_norm": 0.4025278304918358,
+      "learning_rate": 3.11556088254479e-05,
+      "loss": 0.5587,
+      "step": 468
+    },
+    {
+      "epoch": 1.899746835443038,
+      "grad_norm": 0.33747469618336473,
+      "learning_rate": 3.11084549160364e-05,
+      "loss": 0.5559,
+      "step": 469
+    },
+    {
+      "epoch": 1.9037974683544303,
+      "grad_norm": 0.4042881396006844,
+      "learning_rate": 3.106121154065615e-05,
+      "loss": 0.5518,
+      "step": 470
+    },
+    {
+      "epoch": 1.9078481012658228,
+      "grad_norm": 0.3553782294080473,
+      "learning_rate": 3.1013879079798805e-05,
+      "loss": 0.5541,
+      "step": 471
+    },
+    {
+      "epoch": 1.9118987341772153,
+      "grad_norm": 0.3260934393470081,
+      "learning_rate": 3.096645791467348e-05,
+      "loss": 0.5619,
+      "step": 472
+    },
+    {
+      "epoch": 1.9159493670886076,
+      "grad_norm": 0.36734724274348723,
+      "learning_rate": 3.091894842720373e-05,
+      "loss": 0.5575,
+      "step": 473
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.33400752248633975,
+      "learning_rate": 3.0871351000024425e-05,
+      "loss": 0.5639,
+      "step": 474
+    },
+    {
+      "epoch": 1.9240506329113924,
+      "grad_norm": 0.3602230182667905,
+      "learning_rate": 3.0823666016478716e-05,
+      "loss": 0.5463,
+      "step": 475
+    },
+    {
+      "epoch": 1.928101265822785,
+      "grad_norm": 0.30753659958519136,
+      "learning_rate": 3.0775893860614896e-05,
+      "loss": 0.5549,
+      "step": 476
+    },
+    {
+      "epoch": 1.9321518987341773,
+      "grad_norm": 0.34712241259259075,
+      "learning_rate": 3.0728034917183336e-05,
+      "loss": 0.556,
+      "step": 477
+    },
+    {
+      "epoch": 1.9362025316455695,
+      "grad_norm": 0.33837068135005066,
+      "learning_rate": 3.06800895716334e-05,
+      "loss": 0.564,
+      "step": 478
+    },
+    {
+      "epoch": 1.940253164556962,
+      "grad_norm": 0.34462562646045874,
+      "learning_rate": 3.063205821011029e-05,
+      "loss": 0.5714,
+      "step": 479
+    },
+    {
+      "epoch": 1.9443037974683546,
+      "grad_norm": 0.3243356969257693,
+      "learning_rate": 3.0583941219452016e-05,
+      "loss": 0.5546,
+      "step": 480
+    },
+    {
+      "epoch": 1.9483544303797469,
+      "grad_norm": 0.3839057091660223,
+      "learning_rate": 3.053573898718618e-05,
+      "loss": 0.5567,
+      "step": 481
+    },
+    {
+      "epoch": 1.9524050632911392,
+      "grad_norm": 0.3210173029213212,
+      "learning_rate": 3.0487451901526956e-05,
+      "loss": 0.5588,
+      "step": 482
+    },
+    {
+      "epoch": 1.9564556962025317,
+      "grad_norm": 0.3388518082793255,
+      "learning_rate": 3.0439080351371875e-05,
+      "loss": 0.5512,
+      "step": 483
+    },
+    {
+      "epoch": 1.9605063291139242,
+      "grad_norm": 0.3579697634420425,
+      "learning_rate": 3.0390624726298764e-05,
+      "loss": 0.568,
+      "step": 484
+    },
+    {
+      "epoch": 1.9645569620253165,
+      "grad_norm": 0.36775126817924286,
+      "learning_rate": 3.034208541656255e-05,
+      "loss": 0.5624,
+      "step": 485
+    },
+    {
+      "epoch": 1.9686075949367088,
+      "grad_norm": 0.3276033736654695,
+      "learning_rate": 3.029346281309218e-05,
+      "loss": 0.5537,
+      "step": 486
+    },
+    {
+      "epoch": 1.9726582278481013,
+      "grad_norm": 0.3525853651304472,
+      "learning_rate": 3.0244757307487415e-05,
+      "loss": 0.5659,
+      "step": 487
+    },
+    {
+      "epoch": 1.9767088607594938,
+      "grad_norm": 0.39428836637636505,
+      "learning_rate": 3.019596929201569e-05,
+      "loss": 0.5604,
+      "step": 488
+    },
+    {
+      "epoch": 1.980759493670886,
+      "grad_norm": 0.3838737553785362,
+      "learning_rate": 3.0147099159608985e-05,
+      "loss": 0.5446,
+      "step": 489
+    },
+    {
+      "epoch": 1.9848101265822784,
+      "grad_norm": 0.29548402710173977,
+      "learning_rate": 3.0098147303860616e-05,
+      "loss": 0.5652,
+      "step": 490
+    },
+    {
+      "epoch": 1.988860759493671,
+      "grad_norm": 0.378651665225724,
+      "learning_rate": 3.0049114119022117e-05,
+      "loss": 0.5419,
+      "step": 491
+    },
+    {
+      "epoch": 1.9929113924050634,
+      "grad_norm": 0.3181527564513318,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.559,
+      "step": 492
+    },
+    {
+      "epoch": 1.9969620253164557,
+      "grad_norm": 0.3450865449338005,
+      "learning_rate": 2.995080534235264e-05,
+      "loss": 0.5666,
+      "step": 493
+    },
+    {
+      "epoch": 2.001012658227848,
+      "grad_norm": 0.3367965015363785,
+      "learning_rate": 2.9901530542287044e-05,
+      "loss": 0.5462,
+      "step": 494
+    },
+    {
+      "epoch": 2.0050632911392405,
+      "grad_norm": 0.44988528948768663,
+      "learning_rate": 2.9852175996655676e-05,
+      "loss": 0.5098,
+      "step": 495
+    },
+    {
+      "epoch": 2.009113924050633,
+      "grad_norm": 0.4163816816585076,
+      "learning_rate": 2.980274210295326e-05,
+      "loss": 0.4898,
+      "step": 496
+    },
+    {
+      "epoch": 2.013164556962025,
+      "grad_norm": 0.45007815230415055,
+      "learning_rate": 2.9753229259313578e-05,
+      "loss": 0.5057,
+      "step": 497
+    },
+    {
+      "epoch": 2.0172151898734176,
+      "grad_norm": 0.4430664254846143,
+      "learning_rate": 2.9703637864506274e-05,
+      "loss": 0.4967,
+      "step": 498
+    },
+    {
+      "epoch": 2.02126582278481,
+      "grad_norm": 0.49552508735744655,
+      "learning_rate": 2.965396831793362e-05,
+      "loss": 0.4924,
+      "step": 499
+    },
+    {
+      "epoch": 2.0253164556962027,
+      "grad_norm": 0.3693434422379149,
+      "learning_rate": 2.9604221019627316e-05,
+      "loss": 0.491,
+      "step": 500
+    },
+    {
+      "epoch": 2.0293670886075947,
+      "grad_norm": 0.4383463968530797,
+      "learning_rate": 2.955439637024526e-05,
+      "loss": 0.4948,
+      "step": 501
+    },
+    {
+      "epoch": 2.0334177215189873,
+      "grad_norm": 0.39851016707101605,
+      "learning_rate": 2.9504494771068334e-05,
+      "loss": 0.4983,
+      "step": 502
+    },
+    {
+      "epoch": 2.0374683544303798,
+      "grad_norm": 0.38611729196818895,
+      "learning_rate": 2.9454516623997156e-05,
+      "loss": 0.4964,
+      "step": 503
+    },
+    {
+      "epoch": 2.0415189873417723,
+      "grad_norm": 0.36647590694430154,
+      "learning_rate": 2.9404462331548847e-05,
+      "loss": 0.5056,
+      "step": 504
+    },
+    {
+      "epoch": 2.0455696202531644,
+      "grad_norm": 0.3544351204338852,
+      "learning_rate": 2.93543322968538e-05,
+      "loss": 0.4949,
+      "step": 505
+    },
+    {
+      "epoch": 2.049620253164557,
+      "grad_norm": 0.4256301379028352,
+      "learning_rate": 2.9304126923652428e-05,
+      "loss": 0.4961,
+      "step": 506
+    },
+    {
+      "epoch": 2.0536708860759494,
+      "grad_norm": 0.436787289484188,
+      "learning_rate": 2.9253846616291896e-05,
+      "loss": 0.4876,
+      "step": 507
+    },
+    {
+      "epoch": 2.057721518987342,
+      "grad_norm": 0.3258718937206479,
+      "learning_rate": 2.9203491779722896e-05,
+      "loss": 0.473,
+      "step": 508
+    },
+    {
+      "epoch": 2.061772151898734,
+      "grad_norm": 0.3970558236581134,
+      "learning_rate": 2.9153062819496357e-05,
+      "loss": 0.4934,
+      "step": 509
+    },
+    {
+      "epoch": 2.0658227848101265,
+      "grad_norm": 0.30924039589523367,
+      "learning_rate": 2.9102560141760178e-05,
+      "loss": 0.4901,
+      "step": 510
+    },
+    {
+      "epoch": 2.069873417721519,
+      "grad_norm": 0.8870978136757431,
+      "learning_rate": 2.9051984153256004e-05,
+      "loss": 0.4917,
+      "step": 511
+    },
+    {
+      "epoch": 2.0739240506329115,
+      "grad_norm": 0.406140222445099,
+      "learning_rate": 2.900133526131588e-05,
+      "loss": 0.4999,
+      "step": 512
+    },
+    {
+      "epoch": 2.0779746835443036,
+      "grad_norm": 0.3414142747292381,
+      "learning_rate": 2.8950613873859025e-05,
+      "loss": 0.49,
+      "step": 513
+    },
+    {
+      "epoch": 2.082025316455696,
+      "grad_norm": 0.4143813795559816,
+      "learning_rate": 2.8899820399388515e-05,
+      "loss": 0.4961,
+      "step": 514
+    },
+    {
+      "epoch": 2.0860759493670886,
+      "grad_norm": 0.36510159210924936,
+      "learning_rate": 2.8848955246988012e-05,
+      "loss": 0.4962,
+      "step": 515
+    },
+    {
+      "epoch": 2.090126582278481,
+      "grad_norm": 0.3978716246714353,
+      "learning_rate": 2.879801882631847e-05,
+      "loss": 0.499,
+      "step": 516
+    },
+    {
+      "epoch": 2.094177215189873,
+      "grad_norm": 0.38239010519150707,
+      "learning_rate": 2.8747011547614808e-05,
+      "loss": 0.5062,
+      "step": 517
+    },
+    {
+      "epoch": 2.0982278481012657,
+      "grad_norm": 0.38036441021028927,
+      "learning_rate": 2.8695933821682635e-05,
+      "loss": 0.5053,
+      "step": 518
+    },
+    {
+      "epoch": 2.1022784810126582,
+      "grad_norm": 0.42094687490080146,
+      "learning_rate": 2.864478605989494e-05,
+      "loss": 0.4908,
+      "step": 519
+    },
+    {
+      "epoch": 2.1063291139240508,
+      "grad_norm": 0.3144308977108092,
+      "learning_rate": 2.8593568674188765e-05,
+      "loss": 0.5017,
+      "step": 520
+    },
+    {
+      "epoch": 2.110379746835443,
+      "grad_norm": 0.3575528727730122,
+      "learning_rate": 2.8542282077061892e-05,
+      "loss": 0.4895,
+      "step": 521
+    },
+    {
+      "epoch": 2.1144303797468353,
+      "grad_norm": 0.34714553602682713,
+      "learning_rate": 2.8490926681569523e-05,
+      "loss": 0.4895,
+      "step": 522
+    },
+    {
+      "epoch": 2.118481012658228,
+      "grad_norm": 0.5776022474918991,
+      "learning_rate": 2.8439502901320956e-05,
+      "loss": 0.4921,
+      "step": 523
+    },
+    {
+      "epoch": 2.1225316455696204,
+      "grad_norm": 0.33779218410703893,
+      "learning_rate": 2.8388011150476237e-05,
+      "loss": 0.497,
+      "step": 524
+    },
+    {
+      "epoch": 2.1265822784810124,
+      "grad_norm": 0.3429631641185881,
+      "learning_rate": 2.8336451843742866e-05,
+      "loss": 0.5015,
+      "step": 525
+    },
+    {
+      "epoch": 2.130632911392405,
+      "grad_norm": 0.39392498686170413,
+      "learning_rate": 2.8284825396372387e-05,
+      "loss": 0.4883,
+      "step": 526
+    },
+    {
+      "epoch": 2.1346835443037975,
+      "grad_norm": 0.3248496422527101,
+      "learning_rate": 2.8233132224157132e-05,
+      "loss": 0.4881,
+      "step": 527
+    },
+    {
+      "epoch": 2.13873417721519,
+      "grad_norm": 1.3386612941210276,
+      "learning_rate": 2.8181372743426805e-05,
+      "loss": 0.5011,
+      "step": 528
+    },
+    {
+      "epoch": 2.1427848101265825,
+      "grad_norm": 0.3267699677243998,
+      "learning_rate": 2.8129547371045128e-05,
+      "loss": 0.4995,
+      "step": 529
+    },
+    {
+      "epoch": 2.1468354430379746,
+      "grad_norm": 0.3474880727133314,
+      "learning_rate": 2.8077656524406534e-05,
+      "loss": 0.4965,
+      "step": 530
+    },
+    {
+      "epoch": 2.150886075949367,
+      "grad_norm": 0.3243167848082411,
+      "learning_rate": 2.802570062143278e-05,
+      "loss": 0.4904,
+      "step": 531
+    },
+    {
+      "epoch": 2.1549367088607596,
+      "grad_norm": 0.3296107344366493,
+      "learning_rate": 2.7973680080569555e-05,
+      "loss": 0.503,
+      "step": 532
+    },
+    {
+      "epoch": 2.1589873417721517,
+      "grad_norm": 1.006168288368498,
+      "learning_rate": 2.792159532078314e-05,
+      "loss": 0.519,
+      "step": 533
+    },
+    {
+      "epoch": 2.163037974683544,
+      "grad_norm": 0.801665591880324,
+      "learning_rate": 2.7869446761557033e-05,
+      "loss": 0.5068,
+      "step": 534
+    },
+    {
+      "epoch": 2.1670886075949367,
+      "grad_norm": 0.40524833235559626,
+      "learning_rate": 2.781723482288857e-05,
+      "loss": 0.4991,
+      "step": 535
+    },
+    {
+      "epoch": 2.1711392405063292,
+      "grad_norm": 0.3330661478033663,
+      "learning_rate": 2.7764959925285517e-05,
+      "loss": 0.4923,
+      "step": 536
+    },
+    {
+      "epoch": 2.1751898734177217,
+      "grad_norm": 0.3877804650175095,
+      "learning_rate": 2.771262248976272e-05,
+      "loss": 0.4992,
+      "step": 537
+    },
+    {
+      "epoch": 2.179240506329114,
+      "grad_norm": 0.3268811770701431,
+      "learning_rate": 2.7660222937838677e-05,
+      "loss": 0.4947,
+      "step": 538
+    },
+    {
+      "epoch": 2.1832911392405063,
+      "grad_norm": 0.3899736934682659,
+      "learning_rate": 2.7607761691532186e-05,
+      "loss": 0.4946,
+      "step": 539
+    },
+    {
+      "epoch": 2.187341772151899,
+      "grad_norm": 0.34479446045397066,
+      "learning_rate": 2.7555239173358916e-05,
+      "loss": 0.5063,
+      "step": 540
+    },
+    {
+      "epoch": 2.191392405063291,
+      "grad_norm": 0.34419845180911535,
+      "learning_rate": 2.7502655806328e-05,
+      "loss": 0.4889,
+      "step": 541
+    },
+    {
+      "epoch": 2.1954430379746834,
+      "grad_norm": 0.3675038222650802,
+      "learning_rate": 2.7450012013938648e-05,
+      "loss": 0.5061,
+      "step": 542
+    },
+    {
+      "epoch": 2.199493670886076,
+      "grad_norm": 0.3284495858669357,
+      "learning_rate": 2.739730822017673e-05,
+      "loss": 0.4966,
+      "step": 543
+    },
+    {
+      "epoch": 2.2035443037974685,
+      "grad_norm": 0.371280213456399,
+      "learning_rate": 2.7344544849511355e-05,
+      "loss": 0.4909,
+      "step": 544
+    },
+    {
+      "epoch": 2.207594936708861,
+      "grad_norm": 0.3744442283764528,
+      "learning_rate": 2.7291722326891456e-05,
+      "loss": 0.4824,
+      "step": 545
+    },
+    {
+      "epoch": 2.211645569620253,
+      "grad_norm": 0.3580616021676279,
+      "learning_rate": 2.723884107774236e-05,
+      "loss": 0.5104,
+      "step": 546
+    },
+    {
+      "epoch": 2.2156962025316456,
+      "grad_norm": 0.3716278660164839,
+      "learning_rate": 2.718590152796239e-05,
+      "loss": 0.4919,
+      "step": 547
+    },
+    {
+      "epoch": 2.219746835443038,
+      "grad_norm": 0.3398801374717807,
+      "learning_rate": 2.71329041039194e-05,
+      "loss": 0.5064,
+      "step": 548
+    },
+    {
+      "epoch": 2.22379746835443,
+      "grad_norm": 0.2816392909659232,
+      "learning_rate": 2.7079849232447357e-05,
+      "loss": 0.4995,
+      "step": 549
+    },
+    {
+      "epoch": 2.2278481012658227,
+      "grad_norm": 0.327348367271328,
+      "learning_rate": 2.7026737340842895e-05,
+      "loss": 0.4914,
+      "step": 550
+    },
+    {
+      "epoch": 2.231898734177215,
+      "grad_norm": 0.3288314516045633,
+      "learning_rate": 2.697356885686189e-05,
+      "loss": 0.513,
+      "step": 551
+    },
+    {
+      "epoch": 2.2359493670886077,
+      "grad_norm": 0.2944476557138394,
+      "learning_rate": 2.6920344208716014e-05,
+      "loss": 0.4986,
+      "step": 552
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.3243688654228835,
+      "learning_rate": 2.6867063825069252e-05,
+      "loss": 0.4935,
+      "step": 553
+    },
+    {
+      "epoch": 2.2440506329113923,
+      "grad_norm": 0.2983842764803659,
+      "learning_rate": 2.6813728135034494e-05,
+      "loss": 0.4908,
+      "step": 554
+    },
+    {
+      "epoch": 2.248101265822785,
+      "grad_norm": 0.3343943542203276,
+      "learning_rate": 2.6760337568170056e-05,
+      "loss": 0.5064,
+      "step": 555
+    },
+    {
+      "epoch": 2.2521518987341773,
+      "grad_norm": 0.3343718906470914,
+      "learning_rate": 2.6706892554476226e-05,
+      "loss": 0.484,
+      "step": 556
+    },
+    {
+      "epoch": 2.2562025316455694,
+      "grad_norm": 0.36640992756672114,
+      "learning_rate": 2.6653393524391795e-05,
+      "loss": 0.5017,
+      "step": 557
+    },
+    {
+      "epoch": 2.260253164556962,
+      "grad_norm": 0.3947573272507659,
+      "learning_rate": 2.6599840908790592e-05,
+      "loss": 0.5067,
+      "step": 558
+    },
+    {
+      "epoch": 2.2643037974683544,
+      "grad_norm": 0.6105793515708365,
+      "learning_rate": 2.6546235138978028e-05,
+      "loss": 0.4942,
+      "step": 559
+    },
+    {
+      "epoch": 2.268354430379747,
+      "grad_norm": 0.4046205137637671,
+      "learning_rate": 2.6492576646687597e-05,
+      "loss": 0.4982,
+      "step": 560
+    },
+    {
+      "epoch": 2.2724050632911394,
+      "grad_norm": 0.3949293135617647,
+      "learning_rate": 2.6438865864077425e-05,
+      "loss": 0.4968,
+      "step": 561
+    },
+    {
+      "epoch": 2.2764556962025315,
+      "grad_norm": 0.321730042845409,
+      "learning_rate": 2.6385103223726766e-05,
+      "loss": 0.4988,
+      "step": 562
+    },
+    {
+      "epoch": 2.280506329113924,
+      "grad_norm": 0.41207033435408724,
+      "learning_rate": 2.6331289158632537e-05,
+      "loss": 0.5032,
+      "step": 563
+    },
+    {
+      "epoch": 2.2845569620253166,
+      "grad_norm": 0.38332352188363794,
+      "learning_rate": 2.6277424102205817e-05,
+      "loss": 0.509,
+      "step": 564
+    },
+    {
+      "epoch": 2.2886075949367086,
+      "grad_norm": 0.39278764645368525,
+      "learning_rate": 2.6223508488268374e-05,
+      "loss": 0.4888,
+      "step": 565
+    },
+    {
+      "epoch": 2.292658227848101,
+      "grad_norm": 0.3556146431272337,
+      "learning_rate": 2.6169542751049148e-05,
+      "loss": 0.5015,
+      "step": 566
+    },
+    {
+      "epoch": 2.2967088607594937,
+      "grad_norm": 0.4675412328106165,
+      "learning_rate": 2.6115527325180754e-05,
+      "loss": 0.4999,
+      "step": 567
+    },
+    {
+      "epoch": 2.300759493670886,
+      "grad_norm": 0.38763539498360156,
+      "learning_rate": 2.606146264569603e-05,
+      "loss": 0.501,
+      "step": 568
+    },
+    {
+      "epoch": 2.3048101265822787,
+      "grad_norm": 0.2934849632625072,
+      "learning_rate": 2.6007349148024447e-05,
+      "loss": 0.4863,
+      "step": 569
+    },
+    {
+      "epoch": 2.3088607594936708,
+      "grad_norm": 0.41939508896823435,
+      "learning_rate": 2.5953187267988694e-05,
+      "loss": 0.4965,
+      "step": 570
+    },
+    {
+      "epoch": 2.3129113924050633,
+      "grad_norm": 0.30858227669925525,
+      "learning_rate": 2.5898977441801097e-05,
+      "loss": 0.5097,
+      "step": 571
+    },
+    {
+      "epoch": 2.316962025316456,
+      "grad_norm": 0.3392066055425425,
+      "learning_rate": 2.584472010606015e-05,
+      "loss": 0.4925,
+      "step": 572
+    },
+    {
+      "epoch": 2.321012658227848,
+      "grad_norm": 0.3157340315012667,
+      "learning_rate": 2.5790415697746976e-05,
+      "loss": 0.507,
+      "step": 573
+    },
+    {
+      "epoch": 2.3250632911392404,
+      "grad_norm": 0.305608540789959,
+      "learning_rate": 2.5736064654221808e-05,
+      "loss": 0.5007,
+      "step": 574
+    },
+    {
+      "epoch": 2.329113924050633,
+      "grad_norm": 0.288554811447723,
+      "learning_rate": 2.568166741322048e-05,
+      "loss": 0.4969,
+      "step": 575
+    },
+    {
+      "epoch": 2.3331645569620254,
+      "grad_norm": 0.3426994078267748,
+      "learning_rate": 2.56272244128509e-05,
+      "loss": 0.5026,
+      "step": 576
+    },
+    {
+      "epoch": 2.337215189873418,
+      "grad_norm": 0.28677192948845487,
+      "learning_rate": 2.55727360915895e-05,
+      "loss": 0.4818,
+      "step": 577
+    },
+    {
+      "epoch": 2.34126582278481,
+      "grad_norm": 1.8214467222757453,
+      "learning_rate": 2.5518202888277734e-05,
+      "loss": 0.5487,
+      "step": 578
+    },
+    {
+      "epoch": 2.3453164556962025,
+      "grad_norm": 0.354157226469596,
+      "learning_rate": 2.5463625242118523e-05,
+      "loss": 0.4937,
+      "step": 579
+    },
+    {
+      "epoch": 2.349367088607595,
+      "grad_norm": 0.3471145420689805,
+      "learning_rate": 2.5409003592672723e-05,
+      "loss": 0.5068,
+      "step": 580
+    },
+    {
+      "epoch": 2.353417721518987,
+      "grad_norm": 0.3799652685794428,
+      "learning_rate": 2.535433837985559e-05,
+      "loss": 0.5072,
+      "step": 581
+    },
+    {
+      "epoch": 2.3574683544303796,
+      "grad_norm": 0.3363089430378653,
+      "learning_rate": 2.529963004393324e-05,
+      "loss": 0.4983,
+      "step": 582
+    },
+    {
+      "epoch": 2.361518987341772,
+      "grad_norm": 0.34624657519578783,
+      "learning_rate": 2.524487902551908e-05,
+      "loss": 0.504,
+      "step": 583
+    },
+    {
+      "epoch": 2.3655696202531646,
+      "grad_norm": 0.30675165525677733,
+      "learning_rate": 2.519008576557029e-05,
+      "loss": 0.5049,
+      "step": 584
+    },
+    {
+      "epoch": 2.369620253164557,
+      "grad_norm": 0.29925984341431244,
+      "learning_rate": 2.5135250705384254e-05,
+      "loss": 0.4986,
+      "step": 585
+    },
+    {
+      "epoch": 2.3736708860759492,
+      "grad_norm": 0.37437139458382257,
+      "learning_rate": 2.5080374286595007e-05,
+      "loss": 0.4862,
+      "step": 586
+    },
+    {
+      "epoch": 2.3777215189873417,
+      "grad_norm": 0.3000063425566951,
+      "learning_rate": 2.5025456951169677e-05,
+      "loss": 0.5057,
+      "step": 587
+    },
+    {
+      "epoch": 2.3817721518987343,
+      "grad_norm": 0.3351887530481943,
+      "learning_rate": 2.4970499141404942e-05,
+      "loss": 0.4853,
+      "step": 588
+    },
+    {
+      "epoch": 2.3858227848101268,
+      "grad_norm": 0.32490235357972813,
+      "learning_rate": 2.491550129992345e-05,
+      "loss": 0.4994,
+      "step": 589
+    },
+    {
+      "epoch": 2.389873417721519,
+      "grad_norm": 0.34505189543624304,
+      "learning_rate": 2.486046386967024e-05,
+      "loss": 0.4947,
+      "step": 590
+    },
+    {
+      "epoch": 2.3939240506329114,
+      "grad_norm": 0.35489616578197736,
+      "learning_rate": 2.4805387293909214e-05,
+      "loss": 0.5045,
+      "step": 591
+    },
+    {
+      "epoch": 2.397974683544304,
+      "grad_norm": 0.3595753175497515,
+      "learning_rate": 2.4750272016219552e-05,
+      "loss": 0.4893,
+      "step": 592
+    },
+    {
+      "epoch": 2.4020253164556964,
+      "grad_norm": 0.34651348159007905,
+      "learning_rate": 2.4695118480492114e-05,
+      "loss": 0.5008,
+      "step": 593
+    },
+    {
+      "epoch": 2.4060759493670885,
+      "grad_norm": 0.3879192012111034,
+      "learning_rate": 2.4639927130925898e-05,
+      "loss": 0.5019,
+      "step": 594
+    },
+    {
+      "epoch": 2.410126582278481,
+      "grad_norm": 0.3357089903686379,
+      "learning_rate": 2.458469841202444e-05,
+      "loss": 0.4997,
+      "step": 595
+    },
+    {
+      "epoch": 2.4141772151898735,
+      "grad_norm": 0.3841039825296223,
+      "learning_rate": 2.452943276859226e-05,
+      "loss": 0.4961,
+      "step": 596
+    },
+    {
+      "epoch": 2.418227848101266,
+      "grad_norm": 5.100502157148463,
+      "learning_rate": 2.447413064573125e-05,
+      "loss": 0.5742,
+      "step": 597
+    },
+    {
+      "epoch": 2.422278481012658,
+      "grad_norm": 0.6270842232320574,
+      "learning_rate": 2.4418792488837095e-05,
+      "loss": 0.5065,
+      "step": 598
+    },
+    {
+      "epoch": 2.4263291139240506,
+      "grad_norm": 0.36950054901179313,
+      "learning_rate": 2.4363418743595713e-05,
+      "loss": 0.5103,
+      "step": 599
+    },
+    {
+      "epoch": 2.430379746835443,
+      "grad_norm": 0.4861356324021188,
+      "learning_rate": 2.430800985597963e-05,
+      "loss": 0.4929,
+      "step": 600
+    },
+    {
+      "epoch": 2.4344303797468356,
+      "grad_norm": 13.517266214794379,
+      "learning_rate": 2.4252566272244415e-05,
+      "loss": 0.5341,
+      "step": 601
+    },
+    {
+      "epoch": 2.4384810126582277,
+      "grad_norm": 0.7168481055285383,
+      "learning_rate": 2.4197088438925063e-05,
+      "loss": 0.4965,
+      "step": 602
+    },
+    {
+      "epoch": 2.44253164556962,
+      "grad_norm": 0.41709722719658915,
+      "learning_rate": 2.4141576802832417e-05,
+      "loss": 0.4811,
+      "step": 603
+    },
+    {
+      "epoch": 2.4465822784810127,
+      "grad_norm": 0.7226209863223575,
+      "learning_rate": 2.408603181104957e-05,
+      "loss": 0.4954,
+      "step": 604
+    },
+    {
+      "epoch": 2.4506329113924052,
+      "grad_norm": 0.45524409867146154,
+      "learning_rate": 2.4030453910928245e-05,
+      "loss": 0.5009,
+      "step": 605
+    },
+    {
+      "epoch": 2.4546835443037973,
+      "grad_norm": 0.499584457013481,
+      "learning_rate": 2.397484355008521e-05,
+      "loss": 0.4952,
+      "step": 606
+    },
+    {
+      "epoch": 2.45873417721519,
+      "grad_norm": 1.5181283032228206,
+      "learning_rate": 2.3919201176398662e-05,
+      "loss": 0.5084,
+      "step": 607
+    },
+    {
+      "epoch": 2.4627848101265823,
+      "grad_norm": 2.1462470695404288,
+      "learning_rate": 2.3863527238004633e-05,
+      "loss": 0.5137,
+      "step": 608
+    },
+    {
+      "epoch": 2.466835443037975,
+      "grad_norm": 2.1631014156800936,
+      "learning_rate": 2.380782218329337e-05,
+      "loss": 0.523,
+      "step": 609
+    },
+    {
+      "epoch": 2.470886075949367,
+      "grad_norm": 0.4431397182034638,
+      "learning_rate": 2.3752086460905725e-05,
+      "loss": 0.5047,
+      "step": 610
+    },
+    {
+      "epoch": 2.4749367088607594,
+      "grad_norm": 0.34823615310793665,
+      "learning_rate": 2.3696320519729544e-05,
+      "loss": 0.502,
+      "step": 611
+    },
+    {
+      "epoch": 2.478987341772152,
+      "grad_norm": 0.4157555702834231,
+      "learning_rate": 2.3640524808896045e-05,
+      "loss": 0.506,
+      "step": 612
+    },
+    {
+      "epoch": 2.4830379746835445,
+      "grad_norm": 22.81463098885398,
+      "learning_rate": 2.3584699777776222e-05,
+      "loss": 0.5384,
+      "step": 613
+    },
+    {
+      "epoch": 2.4870886075949366,
+      "grad_norm": 0.6612212851418253,
+      "learning_rate": 2.3528845875977195e-05,
+      "loss": 0.5065,
+      "step": 614
+    },
+    {
+      "epoch": 2.491139240506329,
+      "grad_norm": 0.5285452139800472,
+      "learning_rate": 2.3472963553338614e-05,
+      "loss": 0.501,
+      "step": 615
+    },
+    {
+      "epoch": 2.4951898734177216,
+      "grad_norm": 0.5212626242043013,
+      "learning_rate": 2.341705325992901e-05,
+      "loss": 0.4925,
+      "step": 616
+    },
+    {
+      "epoch": 2.499240506329114,
+      "grad_norm": 0.47885138040065867,
+      "learning_rate": 2.336111544604222e-05,
+      "loss": 0.5009,
+      "step": 617
+    },
+    {
+      "epoch": 2.503291139240506,
+      "grad_norm": 0.434087250267316,
+      "learning_rate": 2.33051505621937e-05,
+      "loss": 0.487,
+      "step": 618
+    },
+    {
+      "epoch": 2.5073417721518987,
+      "grad_norm": 0.4694677079487178,
+      "learning_rate": 2.324915905911693e-05,
+      "loss": 0.4811,
+      "step": 619
+    },
+    {
+      "epoch": 2.511392405063291,
+      "grad_norm": 0.4017250968200373,
+      "learning_rate": 2.319314138775977e-05,
+      "loss": 0.4951,
+      "step": 620
+    },
+    {
+      "epoch": 2.5154430379746833,
+      "grad_norm": 0.4175267791224966,
+      "learning_rate": 2.3137097999280856e-05,
+      "loss": 0.511,
+      "step": 621
+    },
+    {
+      "epoch": 2.519493670886076,
+      "grad_norm": 0.37764701320901795,
+      "learning_rate": 2.308102934504593e-05,
+      "loss": 0.4941,
+      "step": 622
+    },
+    {
+      "epoch": 2.5235443037974683,
+      "grad_norm": 0.4289471038911866,
+      "learning_rate": 2.3024935876624222e-05,
+      "loss": 0.4904,
+      "step": 623
+    },
+    {
+      "epoch": 2.527594936708861,
+      "grad_norm": 0.33926027209075776,
+      "learning_rate": 2.2968818045784813e-05,
+      "loss": 0.4952,
+      "step": 624
+    },
+    {
+      "epoch": 2.5316455696202533,
+      "grad_norm": 0.4494694014412692,
+      "learning_rate": 2.2912676304493006e-05,
+      "loss": 0.4893,
+      "step": 625
+    },
+    {
+      "epoch": 2.5356962025316454,
+      "grad_norm": 2.0816314178130724,
+      "learning_rate": 2.2856511104906668e-05,
+      "loss": 0.5544,
+      "step": 626
+    },
+    {
+      "epoch": 2.539746835443038,
+      "grad_norm": 1.1709759915972844,
+      "learning_rate": 2.2800322899372586e-05,
+      "loss": 0.5463,
+      "step": 627
+    },
+    {
+      "epoch": 2.5437974683544304,
+      "grad_norm": 0.3491224445104732,
+      "learning_rate": 2.2744112140422844e-05,
+      "loss": 0.5017,
+      "step": 628
+    },
+    {
+      "epoch": 2.547848101265823,
+      "grad_norm": 0.3467037892850443,
+      "learning_rate": 2.2687879280771177e-05,
+      "loss": 0.5209,
+      "step": 629
+    },
+    {
+      "epoch": 2.5518987341772155,
+      "grad_norm": 0.3363908272471808,
+      "learning_rate": 2.26316247733093e-05,
+      "loss": 0.4923,
+      "step": 630
+    },
+    {
+      "epoch": 2.5559493670886075,
+      "grad_norm": 0.3259864945841708,
+      "learning_rate": 2.257534907110328e-05,
+      "loss": 0.4949,
+      "step": 631
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.2967840235212126,
+      "learning_rate": 2.2519052627389882e-05,
+      "loss": 0.5118,
+      "step": 632
+    },
+    {
+      "epoch": 2.5640506329113926,
+      "grad_norm": 0.30541026129564197,
+      "learning_rate": 2.246273589557294e-05,
+      "loss": 0.4947,
+      "step": 633
+    },
+    {
+      "epoch": 2.5681012658227846,
+      "grad_norm": 0.2863024881646755,
+      "learning_rate": 2.240639932921966e-05,
+      "loss": 0.5008,
+      "step": 634
+    },
+    {
+      "epoch": 2.572151898734177,
+      "grad_norm": 0.32955552048457437,
+      "learning_rate": 2.2350043382056995e-05,
+      "loss": 0.4959,
+      "step": 635
+    },
+    {
+      "epoch": 2.5762025316455697,
+      "grad_norm": 0.28325095098356,
+      "learning_rate": 2.2293668507968015e-05,
+      "loss": 0.4989,
+      "step": 636
+    },
+    {
+      "epoch": 2.580253164556962,
+      "grad_norm": 0.2900468374800593,
+      "learning_rate": 2.2237275160988186e-05,
+      "loss": 0.5126,
+      "step": 637
+    },
+    {
+      "epoch": 2.5843037974683547,
+      "grad_norm": 0.2650348985004949,
+      "learning_rate": 2.2180863795301787e-05,
+      "loss": 0.4976,
+      "step": 638
+    },
+    {
+      "epoch": 2.5883544303797468,
+      "grad_norm": 0.3214181012259551,
+      "learning_rate": 2.212443486523819e-05,
+      "loss": 0.5017,
+      "step": 639
+    },
+    {
+      "epoch": 2.5924050632911393,
+      "grad_norm": 0.3205053220722113,
+      "learning_rate": 2.2067988825268243e-05,
+      "loss": 0.4964,
+      "step": 640
+    },
+    {
+      "epoch": 2.596455696202532,
+      "grad_norm": 0.2827789033102027,
+      "learning_rate": 2.2011526130000596e-05,
+      "loss": 0.5019,
+      "step": 641
+    },
+    {
+      "epoch": 2.600506329113924,
+      "grad_norm": 0.2984055614361101,
+      "learning_rate": 2.1955047234178038e-05,
+      "loss": 0.5009,
+      "step": 642
+    },
+    {
+      "epoch": 2.6045569620253164,
+      "grad_norm": 0.29590050808572643,
+      "learning_rate": 2.1898552592673825e-05,
+      "loss": 0.4958,
+      "step": 643
+    },
+    {
+      "epoch": 2.608607594936709,
+      "grad_norm": 0.3097371796364634,
+      "learning_rate": 2.184204266048803e-05,
+      "loss": 0.4866,
+      "step": 644
+    },
+    {
+      "epoch": 2.6126582278481014,
+      "grad_norm": 0.33386876924884296,
+      "learning_rate": 2.1785517892743887e-05,
+      "loss": 0.5013,
+      "step": 645
+    },
+    {
+      "epoch": 2.616708860759494,
+      "grad_norm": 0.31945937739316527,
+      "learning_rate": 2.17289787446841e-05,
+      "loss": 0.5219,
+      "step": 646
+    },
+    {
+      "epoch": 2.620759493670886,
+      "grad_norm": 0.33585872327975774,
+      "learning_rate": 2.1672425671667198e-05,
+      "loss": 0.4989,
+      "step": 647
+    },
+    {
+      "epoch": 2.6248101265822785,
+      "grad_norm": 0.2965485980858943,
+      "learning_rate": 2.161585912916385e-05,
+      "loss": 0.4952,
+      "step": 648
+    },
+    {
+      "epoch": 2.628860759493671,
+      "grad_norm": 0.31486258935118056,
+      "learning_rate": 2.1559279572753214e-05,
+      "loss": 0.5052,
+      "step": 649
+    },
+    {
+      "epoch": 2.632911392405063,
+      "grad_norm": 0.30265625055706813,
+      "learning_rate": 2.1502687458119268e-05,
+      "loss": 0.4841,
+      "step": 650
+    },
+    {
+      "epoch": 2.6369620253164556,
+      "grad_norm": 0.3277379376547064,
+      "learning_rate": 2.1446083241047116e-05,
+      "loss": 0.4981,
+      "step": 651
+    },
+    {
+      "epoch": 2.641012658227848,
+      "grad_norm": 0.24904160281417645,
+      "learning_rate": 2.1389467377419333e-05,
+      "loss": 0.5072,
+      "step": 652
+    },
+    {
+      "epoch": 2.6450632911392407,
+      "grad_norm": 0.34593113037345763,
+      "learning_rate": 2.133284032321232e-05,
+      "loss": 0.4986,
+      "step": 653
+    },
+    {
+      "epoch": 2.649113924050633,
+      "grad_norm": 0.6218527662417912,
+      "learning_rate": 2.1276202534492566e-05,
+      "loss": 0.4989,
+      "step": 654
+    },
+    {
+      "epoch": 2.6531645569620252,
+      "grad_norm": 0.2945311075603498,
+      "learning_rate": 2.121955446741306e-05,
+      "loss": 0.501,
+      "step": 655
+    },
+    {
+      "epoch": 2.6572151898734178,
+      "grad_norm": 0.2663290393326927,
+      "learning_rate": 2.1162896578209517e-05,
+      "loss": 0.4967,
+      "step": 656
+    },
+    {
+      "epoch": 2.6612658227848103,
+      "grad_norm": 0.3119302436442297,
+      "learning_rate": 2.1106229323196813e-05,
+      "loss": 0.5088,
+      "step": 657
+    },
+    {
+      "epoch": 2.6653164556962023,
+      "grad_norm": 0.41476819647727714,
+      "learning_rate": 2.1049553158765214e-05,
+      "loss": 0.4887,
+      "step": 658
+    },
+    {
+      "epoch": 2.669367088607595,
+      "grad_norm": 11.78231553636138,
+      "learning_rate": 2.0992868541376764e-05,
+      "loss": 0.5388,
+      "step": 659
+    },
+    {
+      "epoch": 2.6734177215189874,
+      "grad_norm": 0.4365125210712558,
+      "learning_rate": 2.093617592756158e-05,
+      "loss": 0.4948,
+      "step": 660
+    },
+    {
+      "epoch": 2.67746835443038,
+      "grad_norm": 0.31114978667191523,
+      "learning_rate": 2.0879475773914167e-05,
+      "loss": 0.5015,
+      "step": 661
+    },
+    {
+      "epoch": 2.6815189873417724,
+      "grad_norm": 0.4146782829942934,
+      "learning_rate": 2.082276853708978e-05,
+      "loss": 0.4967,
+      "step": 662
+    },
+    {
+      "epoch": 2.6855696202531645,
+      "grad_norm": 0.3812195393048291,
+      "learning_rate": 2.076605467380071e-05,
+      "loss": 0.5166,
+      "step": 663
+    },
+    {
+      "epoch": 2.689620253164557,
+      "grad_norm": 0.37903767170833474,
+      "learning_rate": 2.0709334640812613e-05,
+      "loss": 0.4961,
+      "step": 664
+    },
+    {
+      "epoch": 2.6936708860759495,
+      "grad_norm": 0.52580989972627,
+      "learning_rate": 2.0652608894940824e-05,
+      "loss": 0.51,
+      "step": 665
+    },
+    {
+      "epoch": 2.6977215189873416,
+      "grad_norm": 0.36712593345400835,
+      "learning_rate": 2.0595877893046722e-05,
+      "loss": 0.488,
+      "step": 666
+    },
+    {
+      "epoch": 2.701772151898734,
+      "grad_norm": 0.3328913647440034,
+      "learning_rate": 2.0539142092033985e-05,
+      "loss": 0.508,
+      "step": 667
+    },
+    {
+      "epoch": 2.7058227848101266,
+      "grad_norm": 0.32003901618592007,
+      "learning_rate": 2.048240194884496e-05,
+      "loss": 0.4946,
+      "step": 668
+    },
+    {
+      "epoch": 2.709873417721519,
+      "grad_norm": 0.3301993959500028,
+      "learning_rate": 2.042565792045695e-05,
+      "loss": 0.4909,
+      "step": 669
+    },
+    {
+      "epoch": 2.7139240506329116,
+      "grad_norm": 0.2744475444482885,
+      "learning_rate": 2.036891046387857e-05,
+      "loss": 0.5,
+      "step": 670
+    },
+    {
+      "epoch": 2.7179746835443037,
+      "grad_norm": 0.32809749308393504,
+      "learning_rate": 2.0312160036146036e-05,
+      "loss": 0.491,
+      "step": 671
+    },
+    {
+      "epoch": 2.7220253164556962,
+      "grad_norm": 0.30750716083880114,
+      "learning_rate": 2.025540709431948e-05,
+      "loss": 0.4989,
+      "step": 672
+    },
+    {
+      "epoch": 2.7260759493670887,
+      "grad_norm": 0.28943815630682496,
+      "learning_rate": 2.0198652095479298e-05,
+      "loss": 0.4984,
+      "step": 673
+    },
+    {
+      "epoch": 2.730126582278481,
+      "grad_norm": 0.2881157695162525,
+      "learning_rate": 2.014189549672245e-05,
+      "loss": 0.5014,
+      "step": 674
+    },
+    {
+      "epoch": 2.7341772151898733,
+      "grad_norm": 0.2806392834371474,
+      "learning_rate": 2.0085137755158776e-05,
+      "loss": 0.4932,
+      "step": 675
+    },
+    {
+      "epoch": 2.738227848101266,
+      "grad_norm": 0.2773335971042127,
+      "learning_rate": 2.0028379327907327e-05,
+      "loss": 0.5046,
+      "step": 676
+    },
+    {
+      "epoch": 2.7422784810126584,
+      "grad_norm": 0.2826284142718375,
+      "learning_rate": 1.9971620672092676e-05,
+      "loss": 0.4976,
+      "step": 677
+    },
+    {
+      "epoch": 2.746329113924051,
+      "grad_norm": 0.2986767910072417,
+      "learning_rate": 1.991486224484123e-05,
+      "loss": 0.4968,
+      "step": 678
+    },
+    {
+      "epoch": 2.750379746835443,
+      "grad_norm": 0.30054234718350253,
+      "learning_rate": 1.985810450327756e-05,
+      "loss": 0.5025,
+      "step": 679
+    },
+    {
+      "epoch": 2.7544303797468355,
+      "grad_norm": 0.2950486683021978,
+      "learning_rate": 1.9801347904520706e-05,
+      "loss": 0.4956,
+      "step": 680
+    },
+    {
+      "epoch": 2.758481012658228,
+      "grad_norm": 0.324177443397006,
+      "learning_rate": 1.974459290568053e-05,
+      "loss": 0.5066,
+      "step": 681
+    },
+    {
+      "epoch": 2.76253164556962,
+      "grad_norm": 0.27152359190252556,
+      "learning_rate": 1.968783996385397e-05,
+      "loss": 0.4999,
+      "step": 682
+    },
+    {
+      "epoch": 2.7665822784810126,
+      "grad_norm": 0.2944919748598089,
+      "learning_rate": 1.963108953612143e-05,
+      "loss": 0.496,
+      "step": 683
+    },
+    {
+      "epoch": 2.770632911392405,
+      "grad_norm": 0.29673519781485264,
+      "learning_rate": 1.9574342079543056e-05,
+      "loss": 0.4946,
+      "step": 684
+    },
+    {
+      "epoch": 2.7746835443037976,
+      "grad_norm": 0.33930894720010774,
+      "learning_rate": 1.9517598051155046e-05,
+      "loss": 0.5071,
+      "step": 685
+    },
+    {
+      "epoch": 2.77873417721519,
+      "grad_norm": 0.26736060180781274,
+      "learning_rate": 1.9460857907966025e-05,
+      "loss": 0.4995,
+      "step": 686
+    },
+    {
+      "epoch": 2.782784810126582,
+      "grad_norm": 0.3439353920383225,
+      "learning_rate": 1.9404122106953285e-05,
+      "loss": 0.5019,
+      "step": 687
+    },
+    {
+      "epoch": 2.7868354430379747,
+      "grad_norm": 0.9970820846861362,
+      "learning_rate": 1.9347391105059176e-05,
+      "loss": 0.5047,
+      "step": 688
+    },
+    {
+      "epoch": 2.790886075949367,
+      "grad_norm": 0.3010018324565462,
+      "learning_rate": 1.92906653591874e-05,
+      "loss": 0.4875,
+      "step": 689
+    },
+    {
+      "epoch": 2.7949367088607593,
+      "grad_norm": 0.28481746430786187,
+      "learning_rate": 1.9233945326199295e-05,
+      "loss": 0.494,
+      "step": 690
+    },
+    {
+      "epoch": 2.798987341772152,
+      "grad_norm": 0.49321964293840137,
+      "learning_rate": 1.917723146291022e-05,
+      "loss": 0.5164,
+      "step": 691
+    },
+    {
+      "epoch": 2.8030379746835443,
+      "grad_norm": 0.2820155083140123,
+      "learning_rate": 1.912052422608584e-05,
+      "loss": 0.5006,
+      "step": 692
+    },
+    {
+      "epoch": 2.807088607594937,
+      "grad_norm": 0.27116943667193055,
+      "learning_rate": 1.9063824072438428e-05,
+      "loss": 0.4787,
+      "step": 693
+    },
+    {
+      "epoch": 2.8111392405063294,
+      "grad_norm": 0.29034572565112515,
+      "learning_rate": 1.9007131458623246e-05,
+      "loss": 0.4959,
+      "step": 694
+    },
+    {
+      "epoch": 2.8151898734177214,
+      "grad_norm": 0.26392237362060966,
+      "learning_rate": 1.895044684123479e-05,
+      "loss": 0.4931,
+      "step": 695
+    },
+    {
+      "epoch": 2.819240506329114,
+      "grad_norm": 0.5128884499474436,
+      "learning_rate": 1.8893770676803194e-05,
+      "loss": 0.497,
+      "step": 696
+    },
+    {
+      "epoch": 2.8232911392405065,
+      "grad_norm": 0.2868335613383996,
+      "learning_rate": 1.8837103421790486e-05,
+      "loss": 0.5029,
+      "step": 697
+    },
+    {
+      "epoch": 2.8273417721518985,
+      "grad_norm": 0.26636637122357504,
+      "learning_rate": 1.8780445532586952e-05,
+      "loss": 0.4981,
+      "step": 698
+    },
+    {
+      "epoch": 2.831392405063291,
+      "grad_norm": 0.29680323008494824,
+      "learning_rate": 1.872379746550743e-05,
+      "loss": 0.5051,
+      "step": 699
+    },
+    {
+      "epoch": 2.8354430379746836,
+      "grad_norm": 0.27385946685455087,
+      "learning_rate": 1.866715967678769e-05,
+      "loss": 0.5032,
+      "step": 700
+    },
+    {
+      "epoch": 2.839493670886076,
+      "grad_norm": 0.26001201267666185,
+      "learning_rate": 1.861053262258067e-05,
+      "loss": 0.5074,
+      "step": 701
+    },
+    {
+      "epoch": 2.8435443037974686,
+      "grad_norm": 0.2805641918603186,
+      "learning_rate": 1.8553916758952897e-05,
+      "loss": 0.4958,
+      "step": 702
+    },
+    {
+      "epoch": 2.8475949367088607,
+      "grad_norm": 0.28986825152106743,
+      "learning_rate": 1.8497312541880735e-05,
+      "loss": 0.4937,
+      "step": 703
+    },
+    {
+      "epoch": 2.851645569620253,
+      "grad_norm": 0.28369099033078077,
+      "learning_rate": 1.8440720427246786e-05,
+      "loss": 0.502,
+      "step": 704
+    },
+    {
+      "epoch": 2.8556962025316457,
+      "grad_norm": 0.2919656443835669,
+      "learning_rate": 1.8384140870836157e-05,
+      "loss": 0.5082,
+      "step": 705
+    },
+    {
+      "epoch": 2.8597468354430378,
+      "grad_norm": 0.2664921679216166,
+      "learning_rate": 1.8327574328332806e-05,
+      "loss": 0.505,
+      "step": 706
+    },
+    {
+      "epoch": 2.8637974683544303,
+      "grad_norm": 0.29428695397325555,
+      "learning_rate": 1.8271021255315906e-05,
+      "loss": 0.503,
+      "step": 707
+    },
+    {
+      "epoch": 2.867848101265823,
+      "grad_norm": 0.2853908259271052,
+      "learning_rate": 1.8214482107256117e-05,
+      "loss": 0.4918,
+      "step": 708
+    },
+    {
+      "epoch": 2.8718987341772153,
+      "grad_norm": 0.30099944623918073,
+      "learning_rate": 1.8157957339511968e-05,
+      "loss": 0.4954,
+      "step": 709
+    },
+    {
+      "epoch": 2.875949367088608,
+      "grad_norm": 0.2873404478888209,
+      "learning_rate": 1.8101447407326182e-05,
+      "loss": 0.4945,
+      "step": 710
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.6169675428426853,
+      "learning_rate": 1.8044952765821966e-05,
+      "loss": 0.5087,
+      "step": 711
+    },
+    {
+      "epoch": 2.8840506329113924,
+      "grad_norm": 0.3664128060446159,
+      "learning_rate": 1.7988473869999407e-05,
+      "loss": 0.4988,
+      "step": 712
+    },
+    {
+      "epoch": 2.888101265822785,
+      "grad_norm": 0.2666137193564976,
+      "learning_rate": 1.7932011174731764e-05,
+      "loss": 0.4984,
+      "step": 713
+    },
+    {
+      "epoch": 2.892151898734177,
+      "grad_norm": 0.30450867302842793,
+      "learning_rate": 1.7875565134761817e-05,
+      "loss": 0.5044,
+      "step": 714
+    },
+    {
+      "epoch": 2.8962025316455695,
+      "grad_norm": 0.2777054612750143,
+      "learning_rate": 1.7819136204698226e-05,
+      "loss": 0.5024,
+      "step": 715
+    },
+    {
+      "epoch": 2.900253164556962,
+      "grad_norm": 0.31595583070675237,
+      "learning_rate": 1.776272483901182e-05,
+      "loss": 0.5023,
+      "step": 716
+    },
+    {
+      "epoch": 2.9043037974683545,
+      "grad_norm": 0.27652306775648533,
+      "learning_rate": 1.7706331492031995e-05,
+      "loss": 0.4921,
+      "step": 717
+    },
+    {
+      "epoch": 2.908354430379747,
+      "grad_norm": 0.3356938852819067,
+      "learning_rate": 1.764995661794301e-05,
+      "loss": 0.4943,
+      "step": 718
+    },
+    {
+      "epoch": 2.912405063291139,
+      "grad_norm": 0.25477192357607475,
+      "learning_rate": 1.759360067078035e-05,
+      "loss": 0.4932,
+      "step": 719
+    },
+    {
+      "epoch": 2.9164556962025316,
+      "grad_norm": 0.32170296144179,
+      "learning_rate": 1.7537264104427064e-05,
+      "loss": 0.4866,
+      "step": 720
+    },
+    {
+      "epoch": 2.920506329113924,
+      "grad_norm": 0.26129804898505915,
+      "learning_rate": 1.748094737261012e-05,
+      "loss": 0.4863,
+      "step": 721
+    },
+    {
+      "epoch": 2.9245569620253162,
+      "grad_norm": 0.2819649467600713,
+      "learning_rate": 1.7424650928896726e-05,
+      "loss": 0.5093,
+      "step": 722
+    },
+    {
+      "epoch": 2.9286075949367087,
+      "grad_norm": 0.6358912233315348,
+      "learning_rate": 1.7368375226690712e-05,
+      "loss": 0.5033,
+      "step": 723
+    },
+    {
+      "epoch": 2.9326582278481013,
+      "grad_norm": 0.24305442202337005,
+      "learning_rate": 1.731212071922883e-05,
+      "loss": 0.5011,
+      "step": 724
+    },
+    {
+      "epoch": 2.9367088607594938,
+      "grad_norm": 0.2882121070901065,
+      "learning_rate": 1.7255887859577156e-05,
+      "loss": 0.4821,
+      "step": 725
+    },
+    {
+      "epoch": 2.9407594936708863,
+      "grad_norm": 0.25812893750630145,
+      "learning_rate": 1.7199677100627427e-05,
+      "loss": 0.4982,
+      "step": 726
+    },
+    {
+      "epoch": 2.9448101265822784,
+      "grad_norm": 0.2683752673734316,
+      "learning_rate": 1.7143488895093343e-05,
+      "loss": 0.5044,
+      "step": 727
+    },
+    {
+      "epoch": 2.948860759493671,
+      "grad_norm": 0.25830681877617395,
+      "learning_rate": 1.7087323695506994e-05,
+      "loss": 0.4913,
+      "step": 728
+    },
+    {
+      "epoch": 2.9529113924050634,
+      "grad_norm": 0.24530027541523106,
+      "learning_rate": 1.7031181954215194e-05,
+      "loss": 0.485,
+      "step": 729
+    },
+    {
+      "epoch": 2.9569620253164555,
+      "grad_norm": 0.2786883472819882,
+      "learning_rate": 1.6975064123375788e-05,
+      "loss": 0.4994,
+      "step": 730
+    },
+    {
+      "epoch": 2.961012658227848,
+      "grad_norm": 0.3989758377044572,
+      "learning_rate": 1.6918970654954084e-05,
+      "loss": 0.4905,
+      "step": 731
+    },
+    {
+      "epoch": 2.9650632911392405,
+      "grad_norm": 0.2682660089614656,
+      "learning_rate": 1.686290200071915e-05,
+      "loss": 0.4965,
+      "step": 732
+    },
+    {
+      "epoch": 2.969113924050633,
+      "grad_norm": 0.2795115517282762,
+      "learning_rate": 1.6806858612240234e-05,
+      "loss": 0.4912,
+      "step": 733
+    },
+    {
+      "epoch": 2.9731645569620255,
+      "grad_norm": 0.27423619448060793,
+      "learning_rate": 1.6750840940883078e-05,
+      "loss": 0.4932,
+      "step": 734
+    },
+    {
+      "epoch": 2.9772151898734176,
+      "grad_norm": 0.33106055283977964,
+      "learning_rate": 1.6694849437806305e-05,
+      "loss": 0.4873,
+      "step": 735
+    },
+    {
+      "epoch": 2.98126582278481,
+      "grad_norm": 0.28900594551432296,
+      "learning_rate": 1.663888455395778e-05,
+      "loss": 0.5005,
+      "step": 736
+    },
+    {
+      "epoch": 2.9853164556962026,
+      "grad_norm": 0.3252364720795426,
+      "learning_rate": 1.6582946740070995e-05,
+      "loss": 0.4981,
+      "step": 737
+    },
+    {
+      "epoch": 2.9893670886075947,
+      "grad_norm": 0.2726682860625895,
+      "learning_rate": 1.6527036446661396e-05,
+      "loss": 0.495,
+      "step": 738
+    },
+    {
+      "epoch": 2.993417721518987,
+      "grad_norm": 0.3263157863509486,
+      "learning_rate": 1.6471154124022818e-05,
+      "loss": 0.5013,
+      "step": 739
+    },
+    {
+      "epoch": 2.9974683544303797,
+      "grad_norm": 0.283838252361991,
+      "learning_rate": 1.6415300222223788e-05,
+      "loss": 0.4927,
+      "step": 740
+    },
+    {
+      "epoch": 3.0015189873417722,
+      "grad_norm": 0.3270964454340484,
+      "learning_rate": 1.6359475191103958e-05,
+      "loss": 0.4914,
+      "step": 741
+    },
+    {
+      "epoch": 3.0055696202531648,
+      "grad_norm": 0.38759301653586653,
+      "learning_rate": 1.6303679480270466e-05,
+      "loss": 0.4387,
+      "step": 742
+    },
+    {
+      "epoch": 3.009620253164557,
+      "grad_norm": 0.5808439270565191,
+      "learning_rate": 1.624791353909428e-05,
+      "loss": 0.4327,
+      "step": 743
+    },
+    {
+      "epoch": 3.0136708860759494,
+      "grad_norm": 0.52694161781581,
+      "learning_rate": 1.619217781670663e-05,
+      "loss": 0.4518,
+      "step": 744
+    },
+    {
+      "epoch": 3.017721518987342,
+      "grad_norm": 0.43784334636937866,
+      "learning_rate": 1.6136472761995373e-05,
+      "loss": 0.4438,
+      "step": 745
+    },
+    {
+      "epoch": 3.0217721518987344,
+      "grad_norm": 0.40724174686734,
+      "learning_rate": 1.608079882360134e-05,
+      "loss": 0.4433,
+      "step": 746
+    },
+    {
+      "epoch": 3.0258227848101265,
+      "grad_norm": 0.3779834328111871,
+      "learning_rate": 1.60251564499148e-05,
+      "loss": 0.4309,
+      "step": 747
+    },
+    {
+      "epoch": 3.029873417721519,
+      "grad_norm": 0.41582853242743434,
+      "learning_rate": 1.596954608907176e-05,
+      "loss": 0.4347,
+      "step": 748
+    },
+    {
+      "epoch": 3.0339240506329115,
+      "grad_norm": 0.39814711868196895,
+      "learning_rate": 1.591396818895043e-05,
+      "loss": 0.4471,
+      "step": 749
+    },
+    {
+      "epoch": 3.037974683544304,
+      "grad_norm": 0.4477986518707895,
+      "learning_rate": 1.585842319716759e-05,
+      "loss": 0.4388,
+      "step": 750
+    },
+    {
+      "epoch": 3.042025316455696,
+      "grad_norm": 0.3778400781870757,
+      "learning_rate": 1.5802911561074944e-05,
+      "loss": 0.445,
+      "step": 751
+    },
+    {
+      "epoch": 3.0460759493670886,
+      "grad_norm": 0.40367664547205206,
+      "learning_rate": 1.5747433727755595e-05,
+      "loss": 0.4324,
+      "step": 752
+    },
+    {
+      "epoch": 3.050126582278481,
+      "grad_norm": 0.8332995503141809,
+      "learning_rate": 1.5691990144020376e-05,
+      "loss": 0.4645,
+      "step": 753
+    },
+    {
+      "epoch": 3.0541772151898736,
+      "grad_norm": 0.36785024256386895,
+      "learning_rate": 1.5636581256404297e-05,
+      "loss": 0.4222,
+      "step": 754
+    },
+    {
+      "epoch": 3.0582278481012657,
+      "grad_norm": 0.3746729813885356,
+      "learning_rate": 1.558120751116291e-05,
+      "loss": 0.4461,
+      "step": 755
+    },
+    {
+      "epoch": 3.062278481012658,
+      "grad_norm": 0.30391457255974724,
+      "learning_rate": 1.552586935426876e-05,
+      "loss": 0.4323,
+      "step": 756
+    },
+    {
+      "epoch": 3.0663291139240507,
+      "grad_norm": 0.3322405150327349,
+      "learning_rate": 1.547056723140774e-05,
+      "loss": 0.4415,
+      "step": 757
+    },
+    {
+      "epoch": 3.0703797468354432,
+      "grad_norm": 0.3268416873407739,
+      "learning_rate": 1.5415301587975565e-05,
+      "loss": 0.4521,
+      "step": 758
+    },
+    {
+      "epoch": 3.0744303797468353,
+      "grad_norm": 0.3178497352232047,
+      "learning_rate": 1.536007286907411e-05,
+      "loss": 0.4375,
+      "step": 759
+    },
+    {
+      "epoch": 3.078481012658228,
+      "grad_norm": 0.32390671059350573,
+      "learning_rate": 1.5304881519507896e-05,
+      "loss": 0.4448,
+      "step": 760
+    },
+    {
+      "epoch": 3.0825316455696203,
+      "grad_norm": 0.4533019441743116,
+      "learning_rate": 1.5249727983780453e-05,
+      "loss": 0.4532,
+      "step": 761
+    },
+    {
+      "epoch": 3.086582278481013,
+      "grad_norm": 0.3055035146310139,
+      "learning_rate": 1.5194612706090786e-05,
+      "loss": 0.4545,
+      "step": 762
+    },
+    {
+      "epoch": 3.090632911392405,
+      "grad_norm": 0.2892608312058464,
+      "learning_rate": 1.5139536130329771e-05,
+      "loss": 0.443,
+      "step": 763
+    },
+    {
+      "epoch": 3.0946835443037974,
+      "grad_norm": 0.3254127257137983,
+      "learning_rate": 1.508449870007656e-05,
+      "loss": 0.4591,
+      "step": 764
+    },
+    {
+      "epoch": 3.09873417721519,
+      "grad_norm": 0.2729479256061733,
+      "learning_rate": 1.5029500858595056e-05,
+      "loss": 0.431,
+      "step": 765
+    },
+    {
+      "epoch": 3.1027848101265825,
+      "grad_norm": 0.33230989250196347,
+      "learning_rate": 1.4974543048830328e-05,
+      "loss": 0.4426,
+      "step": 766
+    },
+    {
+      "epoch": 3.1068354430379745,
+      "grad_norm": 0.26671998761726246,
+      "learning_rate": 1.4919625713405e-05,
+      "loss": 0.4369,
+      "step": 767
+    },
+    {
+      "epoch": 3.110886075949367,
+      "grad_norm": 0.3041465456969174,
+      "learning_rate": 1.4864749294615756e-05,
+      "loss": 0.4276,
+      "step": 768
+    },
+    {
+      "epoch": 3.1149367088607596,
+      "grad_norm": 0.27736374328453695,
+      "learning_rate": 1.4809914234429716e-05,
+      "loss": 0.4454,
+      "step": 769
+    },
+    {
+      "epoch": 3.118987341772152,
+      "grad_norm": 0.3206616408467089,
+      "learning_rate": 1.4755120974480923e-05,
+      "loss": 0.4531,
+      "step": 770
+    },
+    {
+      "epoch": 3.123037974683544,
+      "grad_norm": 0.2505120191227989,
+      "learning_rate": 1.4700369956066771e-05,
+      "loss": 0.441,
+      "step": 771
+    },
+    {
+      "epoch": 3.1270886075949367,
+      "grad_norm": 0.4071492462756337,
+      "learning_rate": 1.4645661620144413e-05,
+      "loss": 0.4389,
+      "step": 772
+    },
+    {
+      "epoch": 3.131139240506329,
+      "grad_norm": 0.26008944633134895,
+      "learning_rate": 1.4590996407327284e-05,
+      "loss": 0.4445,
+      "step": 773
+    },
+    {
+      "epoch": 3.1351898734177217,
+      "grad_norm": 0.2605585631072323,
+      "learning_rate": 1.4536374757881487e-05,
+      "loss": 0.4257,
+      "step": 774
+    },
+    {
+      "epoch": 3.1392405063291138,
+      "grad_norm": 1.1848476437459738,
+      "learning_rate": 1.4481797111722271e-05,
+      "loss": 0.4667,
+      "step": 775
+    },
+    {
+      "epoch": 3.1432911392405063,
+      "grad_norm": 0.27846014718987344,
+      "learning_rate": 1.4427263908410507e-05,
+      "loss": 0.4353,
+      "step": 776
+    },
+    {
+      "epoch": 3.147341772151899,
+      "grad_norm": 1.8372612716616885,
+      "learning_rate": 1.4372775587149108e-05,
+      "loss": 0.4892,
+      "step": 777
+    },
+    {
+      "epoch": 3.1513924050632913,
+      "grad_norm": 0.353845500950243,
+      "learning_rate": 1.4318332586779522e-05,
+      "loss": 0.443,
+      "step": 778
+    },
+    {
+      "epoch": 3.1554430379746834,
+      "grad_norm": 0.23943161634189533,
+      "learning_rate": 1.4263935345778202e-05,
+      "loss": 0.4365,
+      "step": 779
+    },
+    {
+      "epoch": 3.159493670886076,
+      "grad_norm": 0.32891935371745334,
+      "learning_rate": 1.420958430225303e-05,
+      "loss": 0.4377,
+      "step": 780
+    },
+    {
+      "epoch": 3.1635443037974684,
+      "grad_norm": 0.24710851277997384,
+      "learning_rate": 1.415527989393985e-05,
+      "loss": 0.4426,
+      "step": 781
+    },
+    {
+      "epoch": 3.167594936708861,
+      "grad_norm": 0.32069858682542973,
+      "learning_rate": 1.410102255819891e-05,
+      "loss": 0.4446,
+      "step": 782
+    },
+    {
+      "epoch": 3.171645569620253,
+      "grad_norm": 1.0399232934130094,
+      "learning_rate": 1.404681273201131e-05,
+      "loss": 0.4402,
+      "step": 783
+    },
+    {
+      "epoch": 3.1756962025316455,
+      "grad_norm": 0.33206128828816883,
+      "learning_rate": 1.399265085197556e-05,
+      "loss": 0.4311,
+      "step": 784
+    },
+    {
+      "epoch": 3.179746835443038,
+      "grad_norm": 0.2778521843235964,
+      "learning_rate": 1.393853735430398e-05,
+      "loss": 0.4411,
+      "step": 785
+    },
+    {
+      "epoch": 3.1837974683544306,
+      "grad_norm": 0.3082150117819678,
+      "learning_rate": 1.3884472674819246e-05,
+      "loss": 0.4373,
+      "step": 786
+    },
+    {
+      "epoch": 3.1878481012658226,
+      "grad_norm": 0.28117179390263936,
+      "learning_rate": 1.3830457248950864e-05,
+      "loss": 0.4352,
+      "step": 787
+    },
+    {
+      "epoch": 3.191898734177215,
+      "grad_norm": 0.26818073915612445,
+      "learning_rate": 1.377649151173163e-05,
+      "loss": 0.4322,
+      "step": 788
+    },
+    {
+      "epoch": 3.1959493670886077,
+      "grad_norm": 0.2948532134553097,
+      "learning_rate": 1.3722575897794181e-05,
+      "loss": 0.4487,
+      "step": 789
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.2569664686324319,
+      "learning_rate": 1.3668710841367472e-05,
+      "loss": 0.4516,
+      "step": 790
+    },
+    {
+      "epoch": 3.2040506329113922,
+      "grad_norm": 0.28595779584664105,
+      "learning_rate": 1.361489677627324e-05,
+      "loss": 0.4468,
+      "step": 791
+    },
+    {
+      "epoch": 3.2081012658227848,
+      "grad_norm": 0.23132279295729594,
+      "learning_rate": 1.3561134135922585e-05,
+      "loss": 0.443,
+      "step": 792
+    },
+    {
+      "epoch": 3.2121518987341773,
+      "grad_norm": 0.25231804778270683,
+      "learning_rate": 1.350742335331241e-05,
+      "loss": 0.4538,
+      "step": 793
+    },
+    {
+      "epoch": 3.21620253164557,
+      "grad_norm": 0.27290228025614766,
+      "learning_rate": 1.345376486102198e-05,
+      "loss": 0.4363,
+      "step": 794
+    },
+    {
+      "epoch": 3.220253164556962,
+      "grad_norm": 0.25261071723620154,
+      "learning_rate": 1.3400159091209414e-05,
+      "loss": 0.4428,
+      "step": 795
+    },
+    {
+      "epoch": 3.2243037974683544,
+      "grad_norm": 5.916686969278828,
+      "learning_rate": 1.3346606475608216e-05,
+      "loss": 0.4608,
+      "step": 796
+    },
+    {
+      "epoch": 3.228354430379747,
+      "grad_norm": 0.32844613705153125,
+      "learning_rate": 1.3293107445523781e-05,
+      "loss": 0.428,
+      "step": 797
+    },
+    {
+      "epoch": 3.2324050632911394,
+      "grad_norm": 0.24586457415846405,
+      "learning_rate": 1.3239662431829949e-05,
+      "loss": 0.4437,
+      "step": 798
+    },
+    {
+      "epoch": 3.2364556962025315,
+      "grad_norm": 0.2707700719122805,
+      "learning_rate": 1.3186271864965509e-05,
+      "loss": 0.445,
+      "step": 799
+    },
+    {
+      "epoch": 3.240506329113924,
+      "grad_norm": 0.27051901965070674,
+      "learning_rate": 1.3132936174930756e-05,
+      "loss": 0.4445,
+      "step": 800
+    },
+    {
+      "epoch": 3.2445569620253165,
+      "grad_norm": 0.25362887047078425,
+      "learning_rate": 1.3079655791283995e-05,
+      "loss": 0.4509,
+      "step": 801
+    },
+    {
+      "epoch": 3.248607594936709,
+      "grad_norm": 0.2654854942433358,
+      "learning_rate": 1.3026431143138108e-05,
+      "loss": 0.4521,
+      "step": 802
+    },
+    {
+      "epoch": 3.252658227848101,
+      "grad_norm": 0.24762429762748903,
+      "learning_rate": 1.2973262659157114e-05,
+      "loss": 0.4409,
+      "step": 803
+    },
+    {
+      "epoch": 3.2567088607594936,
+      "grad_norm": 0.26824854796026826,
+      "learning_rate": 1.2920150767552651e-05,
+      "loss": 0.4373,
+      "step": 804
+    },
+    {
+      "epoch": 3.260759493670886,
+      "grad_norm": 0.24347171631369263,
+      "learning_rate": 1.2867095896080607e-05,
+      "loss": 0.4472,
+      "step": 805
+    },
+    {
+      "epoch": 3.2648101265822786,
+      "grad_norm": 0.3019461952558301,
+      "learning_rate": 1.2814098472037612e-05,
+      "loss": 0.4386,
+      "step": 806
+    },
+    {
+      "epoch": 3.2688607594936707,
+      "grad_norm": 0.6187908750332304,
+      "learning_rate": 1.276115892225764e-05,
+      "loss": 0.4644,
+      "step": 807
+    },
+    {
+      "epoch": 3.2729113924050632,
+      "grad_norm": 0.2587725156085,
+      "learning_rate": 1.2708277673108555e-05,
+      "loss": 0.4459,
+      "step": 808
+    },
+    {
+      "epoch": 3.2769620253164558,
+      "grad_norm": 0.27515503284543796,
+      "learning_rate": 1.2655455150488649e-05,
+      "loss": 0.4551,
+      "step": 809
+    },
+    {
+      "epoch": 3.2810126582278483,
+      "grad_norm": 0.24615575914208188,
+      "learning_rate": 1.2602691779823272e-05,
+      "loss": 0.4453,
+      "step": 810
+    },
+    {
+      "epoch": 3.2850632911392403,
+      "grad_norm": 0.2630009606115076,
+      "learning_rate": 1.2549987986061355e-05,
+      "loss": 0.4431,
+      "step": 811
+    },
+    {
+      "epoch": 3.289113924050633,
+      "grad_norm": 0.2580754561372567,
+      "learning_rate": 1.2497344193672005e-05,
+      "loss": 0.4367,
+      "step": 812
+    },
+    {
+      "epoch": 3.2931645569620254,
+      "grad_norm": 0.2837072373992648,
+      "learning_rate": 1.2444760826641092e-05,
+      "loss": 0.4398,
+      "step": 813
+    },
+    {
+      "epoch": 3.297215189873418,
+      "grad_norm": 0.2531958775622318,
+      "learning_rate": 1.2392238308467817e-05,
+      "loss": 0.4466,
+      "step": 814
+    },
+    {
+      "epoch": 3.30126582278481,
+      "grad_norm": 0.2556813376228238,
+      "learning_rate": 1.2339777062161326e-05,
+      "loss": 0.4474,
+      "step": 815
+    },
+    {
+      "epoch": 3.3053164556962025,
+      "grad_norm": 0.3001568637940637,
+      "learning_rate": 1.2287377510237293e-05,
+      "loss": 0.4378,
+      "step": 816
+    },
+    {
+      "epoch": 3.309367088607595,
+      "grad_norm": 0.22938697537636352,
+      "learning_rate": 1.2235040074714488e-05,
+      "loss": 0.4409,
+      "step": 817
+    },
+    {
+      "epoch": 3.3134177215189875,
+      "grad_norm": 0.3473165925559644,
+      "learning_rate": 1.2182765177111434e-05,
+      "loss": 0.4503,
+      "step": 818
+    },
+    {
+      "epoch": 3.3174683544303796,
+      "grad_norm": 0.24603962274560726,
+      "learning_rate": 1.213055323844297e-05,
+      "loss": 0.4404,
+      "step": 819
+    },
+    {
+      "epoch": 3.321518987341772,
+      "grad_norm": 0.3206179927387132,
+      "learning_rate": 1.2078404679216864e-05,
+      "loss": 0.453,
+      "step": 820
+    },
+    {
+      "epoch": 3.3255696202531646,
+      "grad_norm": 0.25363198868469705,
+      "learning_rate": 1.2026319919430458e-05,
+      "loss": 0.4413,
+      "step": 821
+    },
+    {
+      "epoch": 3.329620253164557,
+      "grad_norm": 0.25217302275559556,
+      "learning_rate": 1.1974299378567227e-05,
+      "loss": 0.4342,
+      "step": 822
+    },
+    {
+      "epoch": 3.333670886075949,
+      "grad_norm": 0.24752443535726784,
+      "learning_rate": 1.1922343475593462e-05,
+      "loss": 0.4383,
+      "step": 823
+    },
+    {
+      "epoch": 3.3377215189873417,
+      "grad_norm": 0.2580091578059851,
+      "learning_rate": 1.187045262895488e-05,
+      "loss": 0.4465,
+      "step": 824
+    },
+    {
+      "epoch": 3.3417721518987342,
+      "grad_norm": 0.26375798566822783,
+      "learning_rate": 1.1818627256573203e-05,
+      "loss": 0.4395,
+      "step": 825
+    },
+    {
+      "epoch": 3.3458227848101267,
+      "grad_norm": 0.27329998971765795,
+      "learning_rate": 1.1766867775842864e-05,
+      "loss": 0.45,
+      "step": 826
+    },
+    {
+      "epoch": 3.349873417721519,
+      "grad_norm": 0.2575072428729799,
+      "learning_rate": 1.1715174603627615e-05,
+      "loss": 0.4375,
+      "step": 827
+    },
+    {
+      "epoch": 3.3539240506329113,
+      "grad_norm": 0.2509289673700171,
+      "learning_rate": 1.1663548156257147e-05,
+      "loss": 0.4421,
+      "step": 828
+    },
+    {
+      "epoch": 3.357974683544304,
+      "grad_norm": 0.2487519635606791,
+      "learning_rate": 1.161198884952377e-05,
+      "loss": 0.4358,
+      "step": 829
+    },
+    {
+      "epoch": 3.3620253164556964,
+      "grad_norm": 0.23940160916873635,
+      "learning_rate": 1.1560497098679056e-05,
+      "loss": 0.4464,
+      "step": 830
+    },
+    {
+      "epoch": 3.3660759493670884,
+      "grad_norm": 0.23210503219543444,
+      "learning_rate": 1.1509073318430479e-05,
+      "loss": 0.4501,
+      "step": 831
+    },
+    {
+      "epoch": 3.370126582278481,
+      "grad_norm": 0.2635345417306887,
+      "learning_rate": 1.1457717922938116e-05,
+      "loss": 0.4486,
+      "step": 832
+    },
+    {
+      "epoch": 3.3741772151898735,
+      "grad_norm": 0.2304174151072277,
+      "learning_rate": 1.1406431325811233e-05,
+      "loss": 0.4485,
+      "step": 833
+    },
+    {
+      "epoch": 3.378227848101266,
+      "grad_norm": 0.24903879881163132,
+      "learning_rate": 1.135521394010506e-05,
+      "loss": 0.4436,
+      "step": 834
+    },
+    {
+      "epoch": 3.382278481012658,
+      "grad_norm": 0.2569176337971067,
+      "learning_rate": 1.1304066178317367e-05,
+      "loss": 0.4446,
+      "step": 835
+    },
+    {
+      "epoch": 3.3863291139240506,
+      "grad_norm": 0.2536720181847794,
+      "learning_rate": 1.1252988452385199e-05,
+      "loss": 0.4425,
+      "step": 836
+    },
+    {
+      "epoch": 3.390379746835443,
+      "grad_norm": 0.2582046791045345,
+      "learning_rate": 1.1201981173681536e-05,
+      "loss": 0.4474,
+      "step": 837
+    },
+    {
+      "epoch": 3.3944303797468356,
+      "grad_norm": 0.2426425915432118,
+      "learning_rate": 1.1151044753011991e-05,
+      "loss": 0.4505,
+      "step": 838
+    },
+    {
+      "epoch": 3.3984810126582277,
+      "grad_norm": 0.2648981375806747,
+      "learning_rate": 1.1100179600611491e-05,
+      "loss": 0.4378,
+      "step": 839
+    },
+    {
+      "epoch": 3.40253164556962,
+      "grad_norm": 0.2614202479027756,
+      "learning_rate": 1.1049386126140985e-05,
+      "loss": 0.4384,
+      "step": 840
+    },
+    {
+      "epoch": 3.4065822784810127,
+      "grad_norm": 0.23499090887559482,
+      "learning_rate": 1.0998664738684128e-05,
+      "loss": 0.445,
+      "step": 841
+    },
+    {
+      "epoch": 3.410632911392405,
+      "grad_norm": 0.2426309033190602,
+      "learning_rate": 1.0948015846744e-05,
+      "loss": 0.4475,
+      "step": 842
+    },
+    {
+      "epoch": 3.4146835443037973,
+      "grad_norm": 0.24740998900198563,
+      "learning_rate": 1.0897439858239832e-05,
+      "loss": 0.437,
+      "step": 843
+    },
+    {
+      "epoch": 3.41873417721519,
+      "grad_norm": 0.23352294479093633,
+      "learning_rate": 1.0846937180503652e-05,
+      "loss": 0.4382,
+      "step": 844
+    },
+    {
+      "epoch": 3.4227848101265823,
+      "grad_norm": 0.26732210547960766,
+      "learning_rate": 1.0796508220277117e-05,
+      "loss": 0.4413,
+      "step": 845
+    },
+    {
+      "epoch": 3.426835443037975,
+      "grad_norm": 0.25414614149694775,
+      "learning_rate": 1.0746153383708107e-05,
+      "loss": 0.4287,
+      "step": 846
+    },
+    {
+      "epoch": 3.430886075949367,
+      "grad_norm": 0.2520278564724915,
+      "learning_rate": 1.0695873076347579e-05,
+      "loss": 0.4357,
+      "step": 847
+    },
+    {
+      "epoch": 3.4349367088607594,
+      "grad_norm": 0.2709469412549575,
+      "learning_rate": 1.0645667703146205e-05,
+      "loss": 0.438,
+      "step": 848
+    },
+    {
+      "epoch": 3.438987341772152,
+      "grad_norm": 0.2458100841930303,
+      "learning_rate": 1.0595537668451161e-05,
+      "loss": 0.4403,
+      "step": 849
+    },
+    {
+      "epoch": 3.4430379746835444,
+      "grad_norm": 0.2616112653509615,
+      "learning_rate": 1.0545483376002854e-05,
+      "loss": 0.4466,
+      "step": 850
+    },
+    {
+      "epoch": 3.4470886075949365,
+      "grad_norm": 0.25509069400948275,
+      "learning_rate": 1.0495505228931676e-05,
+      "loss": 0.4335,
+      "step": 851
+    },
+    {
+      "epoch": 3.451139240506329,
+      "grad_norm": 0.4978461516314591,
+      "learning_rate": 1.044560362975474e-05,
+      "loss": 0.4566,
+      "step": 852
+    },
+    {
+      "epoch": 3.4551898734177215,
+      "grad_norm": 0.2449098882270892,
+      "learning_rate": 1.0395778980372695e-05,
+      "loss": 0.4495,
+      "step": 853
+    },
+    {
+      "epoch": 3.459240506329114,
+      "grad_norm": 0.23645177316989824,
+      "learning_rate": 1.0346031682066381e-05,
+      "loss": 0.4491,
+      "step": 854
+    },
+    {
+      "epoch": 3.463291139240506,
+      "grad_norm": 0.23573758860566152,
+      "learning_rate": 1.0296362135493724e-05,
+      "loss": 0.4444,
+      "step": 855
+    },
+    {
+      "epoch": 3.4673417721518986,
+      "grad_norm": 0.2416770136819717,
+      "learning_rate": 1.0246770740686422e-05,
+      "loss": 0.4347,
+      "step": 856
+    },
+    {
+      "epoch": 3.471392405063291,
+      "grad_norm": 0.23579655171116387,
+      "learning_rate": 1.0197257897046743e-05,
+      "loss": 0.4365,
+      "step": 857
+    },
+    {
+      "epoch": 3.4754430379746837,
+      "grad_norm": 0.24720326358750438,
+      "learning_rate": 1.014782400334433e-05,
+      "loss": 0.449,
+      "step": 858
+    },
+    {
+      "epoch": 3.479493670886076,
+      "grad_norm": 0.25349611096662095,
+      "learning_rate": 1.009846945771296e-05,
+      "loss": 0.4293,
+      "step": 859
+    },
+    {
+      "epoch": 3.4835443037974683,
+      "grad_norm": 0.233288655961289,
+      "learning_rate": 1.0049194657647363e-05,
+      "loss": 0.4366,
+      "step": 860
+    },
+    {
+      "epoch": 3.487594936708861,
+      "grad_norm": 0.24184513295984525,
+      "learning_rate": 1.0000000000000006e-05,
+      "loss": 0.4479,
+      "step": 861
+    },
+    {
+      "epoch": 3.4916455696202533,
+      "grad_norm": 0.23531216591207993,
+      "learning_rate": 9.950885880977891e-06,
+      "loss": 0.4378,
+      "step": 862
+    },
+    {
+      "epoch": 3.4956962025316454,
+      "grad_norm": 0.23255796422564154,
+      "learning_rate": 9.901852696139382e-06,
+      "loss": 0.4529,
+      "step": 863
+    },
+    {
+      "epoch": 3.499746835443038,
+      "grad_norm": 0.2558969631214962,
+      "learning_rate": 9.852900840391027e-06,
+      "loss": 0.4468,
+      "step": 864
+    },
+    {
+      "epoch": 3.5037974683544304,
+      "grad_norm": 0.2181824538312049,
+      "learning_rate": 9.804030707984313e-06,
+      "loss": 0.4269,
+      "step": 865
+    },
+    {
+      "epoch": 3.507848101265823,
+      "grad_norm": 0.22147505775541257,
+      "learning_rate": 9.755242692512599e-06,
+      "loss": 0.443,
+      "step": 866
+    },
+    {
+      "epoch": 3.5118987341772154,
+      "grad_norm": 0.25472155362629706,
+      "learning_rate": 9.70653718690782e-06,
+      "loss": 0.4481,
+      "step": 867
+    },
+    {
+      "epoch": 3.5159493670886075,
+      "grad_norm": 0.23340467869965525,
+      "learning_rate": 9.657914583437454e-06,
+      "loss": 0.4441,
+      "step": 868
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.23140697465090843,
+      "learning_rate": 9.609375273701246e-06,
+      "loss": 0.4393,
+      "step": 869
+    },
+    {
+      "epoch": 3.5240506329113925,
+      "grad_norm": 0.2516771943344089,
+      "learning_rate": 9.560919648628133e-06,
+      "loss": 0.4559,
+      "step": 870
+    },
+    {
+      "epoch": 3.5281012658227846,
+      "grad_norm": 0.22504282815872784,
+      "learning_rate": 9.512548098473047e-06,
+      "loss": 0.4496,
+      "step": 871
+    },
+    {
+      "epoch": 3.532151898734177,
+      "grad_norm": 0.22010115079133077,
+      "learning_rate": 9.464261012813825e-06,
+      "loss": 0.4378,
+      "step": 872
+    },
+    {
+      "epoch": 3.5362025316455696,
+      "grad_norm": 0.2147265283607473,
+      "learning_rate": 9.416058780547987e-06,
+      "loss": 0.4377,
+      "step": 873
+    },
+    {
+      "epoch": 3.540253164556962,
+      "grad_norm": 1.0127530649010235,
+      "learning_rate": 9.367941789889714e-06,
+      "loss": 0.466,
+      "step": 874
+    },
+    {
+      "epoch": 3.5443037974683547,
+      "grad_norm": 0.23239104802796964,
+      "learning_rate": 9.319910428366607e-06,
+      "loss": 0.4412,
+      "step": 875
+    },
+    {
+      "epoch": 3.5483544303797467,
+      "grad_norm": 0.22424120579505216,
+      "learning_rate": 9.271965082816667e-06,
+      "loss": 0.4378,
+      "step": 876
+    },
+    {
+      "epoch": 3.5524050632911393,
+      "grad_norm": 0.2110723659475602,
+      "learning_rate": 9.224106139385111e-06,
+      "loss": 0.446,
+      "step": 877
+    },
+    {
+      "epoch": 3.5564556962025318,
+      "grad_norm": 0.23666491962274638,
+      "learning_rate": 9.176333983521291e-06,
+      "loss": 0.4327,
+      "step": 878
+    },
+    {
+      "epoch": 3.560506329113924,
+      "grad_norm": 0.22549984079832297,
+      "learning_rate": 9.12864899997558e-06,
+      "loss": 0.4388,
+      "step": 879
+    },
+    {
+      "epoch": 3.5645569620253164,
+      "grad_norm": 0.22037989492611926,
+      "learning_rate": 9.08105157279628e-06,
+      "loss": 0.4403,
+      "step": 880
+    },
+    {
+      "epoch": 3.568607594936709,
+      "grad_norm": 0.260911430948045,
+      "learning_rate": 9.03354208532653e-06,
+      "loss": 0.4392,
+      "step": 881
+    },
+    {
+      "epoch": 3.5726582278481014,
+      "grad_norm": 0.22861615873195762,
+      "learning_rate": 8.986120920201205e-06,
+      "loss": 0.4395,
+      "step": 882
+    },
+    {
+      "epoch": 3.576708860759494,
+      "grad_norm": 0.21503063900563363,
+      "learning_rate": 8.938788459343852e-06,
+      "loss": 0.4554,
+      "step": 883
+    },
+    {
+      "epoch": 3.580759493670886,
+      "grad_norm": 0.22811753017775552,
+      "learning_rate": 8.8915450839636e-06,
+      "loss": 0.4356,
+      "step": 884
+    },
+    {
+      "epoch": 3.5848101265822785,
+      "grad_norm": 0.2219550436894594,
+      "learning_rate": 8.844391174552116e-06,
+      "loss": 0.4459,
+      "step": 885
+    },
+    {
+      "epoch": 3.588860759493671,
+      "grad_norm": 0.20667453788165044,
+      "learning_rate": 8.797327110880479e-06,
+      "loss": 0.4334,
+      "step": 886
+    },
+    {
+      "epoch": 3.592911392405063,
+      "grad_norm": 0.2183378817974902,
+      "learning_rate": 8.750353271996206e-06,
+      "loss": 0.4504,
+      "step": 887
+    },
+    {
+      "epoch": 3.5969620253164556,
+      "grad_norm": 0.22172522211995077,
+      "learning_rate": 8.703470036220132e-06,
+      "loss": 0.4518,
+      "step": 888
+    },
+    {
+      "epoch": 3.601012658227848,
+      "grad_norm": 0.3256351728411888,
+      "learning_rate": 8.656677781143394e-06,
+      "loss": 0.4666,
+      "step": 889
+    },
+    {
+      "epoch": 3.6050632911392406,
+      "grad_norm": 0.22914143482988372,
+      "learning_rate": 8.609976883624377e-06,
+      "loss": 0.4297,
+      "step": 890
+    },
+    {
+      "epoch": 3.609113924050633,
+      "grad_norm": 0.23934581081694853,
+      "learning_rate": 8.563367719785698e-06,
+      "loss": 0.4525,
+      "step": 891
+    },
+    {
+      "epoch": 3.613164556962025,
+      "grad_norm": 0.24887475022214542,
+      "learning_rate": 8.516850665011138e-06,
+      "loss": 0.4283,
+      "step": 892
+    },
+    {
+      "epoch": 3.6172151898734177,
+      "grad_norm": 0.223733865178949,
+      "learning_rate": 8.47042609394269e-06,
+      "loss": 0.4405,
+      "step": 893
+    },
+    {
+      "epoch": 3.6212658227848102,
+      "grad_norm": 0.24219168495999985,
+      "learning_rate": 8.424094380477432e-06,
+      "loss": 0.4391,
+      "step": 894
+    },
+    {
+      "epoch": 3.6253164556962023,
+      "grad_norm": 0.22437222601475848,
+      "learning_rate": 8.37785589776465e-06,
+      "loss": 0.4383,
+      "step": 895
+    },
+    {
+      "epoch": 3.629367088607595,
+      "grad_norm": 0.21667191092057198,
+      "learning_rate": 8.331711018202694e-06,
+      "loss": 0.4335,
+      "step": 896
+    },
+    {
+      "epoch": 3.6334177215189873,
+      "grad_norm": 1.0292359051286124,
+      "learning_rate": 8.285660113436104e-06,
+      "loss": 0.4517,
+      "step": 897
+    },
+    {
+      "epoch": 3.63746835443038,
+      "grad_norm": 0.23596360976939565,
+      "learning_rate": 8.239703554352527e-06,
+      "loss": 0.4531,
+      "step": 898
+    },
+    {
+      "epoch": 3.6415189873417724,
+      "grad_norm": 0.23488837756560854,
+      "learning_rate": 8.193841711079775e-06,
+      "loss": 0.4447,
+      "step": 899
+    },
+    {
+      "epoch": 3.6455696202531644,
+      "grad_norm": 0.23662587616526212,
+      "learning_rate": 8.148074952982828e-06,
+      "loss": 0.441,
+      "step": 900
+    },
+    {
+      "epoch": 3.649620253164557,
+      "grad_norm": 0.25098852143019745,
+      "learning_rate": 8.102403648660859e-06,
+      "loss": 0.4357,
+      "step": 901
+    },
+    {
+      "epoch": 3.6536708860759495,
+      "grad_norm": 0.24050208757878186,
+      "learning_rate": 8.056828165944282e-06,
+      "loss": 0.4322,
+      "step": 902
+    },
+    {
+      "epoch": 3.6577215189873415,
+      "grad_norm": 0.22231160238354156,
+      "learning_rate": 8.011348871891762e-06,
+      "loss": 0.4612,
+      "step": 903
+    },
+    {
+      "epoch": 3.661772151898734,
+      "grad_norm": 0.23483918551758565,
+      "learning_rate": 7.965966132787287e-06,
+      "loss": 0.4508,
+      "step": 904
+    },
+    {
+      "epoch": 3.6658227848101266,
+      "grad_norm": 0.23181136087519447,
+      "learning_rate": 7.920680314137189e-06,
+      "loss": 0.4265,
+      "step": 905
+    },
+    {
+      "epoch": 3.669873417721519,
+      "grad_norm": 0.23318303198475615,
+      "learning_rate": 7.875491780667246e-06,
+      "loss": 0.446,
+      "step": 906
+    },
+    {
+      "epoch": 3.6739240506329116,
+      "grad_norm": 0.21565725088887816,
+      "learning_rate": 7.830400896319667e-06,
+      "loss": 0.4509,
+      "step": 907
+    },
+    {
+      "epoch": 3.6779746835443037,
+      "grad_norm": 0.23559133723678866,
+      "learning_rate": 7.785408024250259e-06,
+      "loss": 0.4418,
+      "step": 908
+    },
+    {
+      "epoch": 3.682025316455696,
+      "grad_norm": 0.24688364662270415,
+      "learning_rate": 7.74051352682542e-06,
+      "loss": 0.4275,
+      "step": 909
+    },
+    {
+      "epoch": 3.6860759493670887,
+      "grad_norm": 0.2328063735576249,
+      "learning_rate": 7.695717765619257e-06,
+      "loss": 0.4479,
+      "step": 910
+    },
+    {
+      "epoch": 3.690126582278481,
+      "grad_norm": 0.22994292265321883,
+      "learning_rate": 7.651021101410673e-06,
+      "loss": 0.4396,
+      "step": 911
+    },
+    {
+      "epoch": 3.6941772151898733,
+      "grad_norm": 0.24928409580514282,
+      "learning_rate": 7.606423894180464e-06,
+      "loss": 0.4424,
+      "step": 912
+    },
+    {
+      "epoch": 3.698227848101266,
+      "grad_norm": 0.22246677879465013,
+      "learning_rate": 7.56192650310839e-06,
+      "loss": 0.4383,
+      "step": 913
+    },
+    {
+      "epoch": 3.7022784810126583,
+      "grad_norm": 0.21648378227813259,
+      "learning_rate": 7.517529286570349e-06,
+      "loss": 0.4387,
+      "step": 914
+    },
+    {
+      "epoch": 3.706329113924051,
+      "grad_norm": 0.21703656712285405,
+      "learning_rate": 7.473232602135387e-06,
+      "loss": 0.4428,
+      "step": 915
+    },
+    {
+      "epoch": 3.710379746835443,
+      "grad_norm": 0.2374776186820649,
+      "learning_rate": 7.429036806562935e-06,
+      "loss": 0.4431,
+      "step": 916
+    },
+    {
+      "epoch": 3.7144303797468354,
+      "grad_norm": 0.2214680978131251,
+      "learning_rate": 7.3849422557998455e-06,
+      "loss": 0.4313,
+      "step": 917
+    },
+    {
+      "epoch": 3.718481012658228,
+      "grad_norm": 0.23132845994454135,
+      "learning_rate": 7.340949304977567e-06,
+      "loss": 0.4333,
+      "step": 918
+    },
+    {
+      "epoch": 3.72253164556962,
+      "grad_norm": 0.23280527487305142,
+      "learning_rate": 7.297058308409282e-06,
+      "loss": 0.438,
+      "step": 919
+    },
+    {
+      "epoch": 3.7265822784810125,
+      "grad_norm": 0.23101296478911068,
+      "learning_rate": 7.25326961958704e-06,
+      "loss": 0.446,
+      "step": 920
+    },
+    {
+      "epoch": 3.730632911392405,
+      "grad_norm": 0.22102249658316175,
+      "learning_rate": 7.209583591178921e-06,
+      "loss": 0.4408,
+      "step": 921
+    },
+    {
+      "epoch": 3.7346835443037976,
+      "grad_norm": 0.23433873051361206,
+      "learning_rate": 7.1660005750261925e-06,
+      "loss": 0.4328,
+      "step": 922
+    },
+    {
+      "epoch": 3.73873417721519,
+      "grad_norm": 0.2167461196423833,
+      "learning_rate": 7.1225209221404765e-06,
+      "loss": 0.4405,
+      "step": 923
+    },
+    {
+      "epoch": 3.742784810126582,
+      "grad_norm": 0.23286054385494478,
+      "learning_rate": 7.079144982700909e-06,
+      "loss": 0.4408,
+      "step": 924
+    },
+    {
+      "epoch": 3.7468354430379747,
+      "grad_norm": 0.209545671956717,
+      "learning_rate": 7.0358731060513695e-06,
+      "loss": 0.4442,
+      "step": 925
+    },
+    {
+      "epoch": 3.750886075949367,
+      "grad_norm": 0.22012408194787375,
+      "learning_rate": 6.99270564069757e-06,
+      "loss": 0.4415,
+      "step": 926
+    },
+    {
+      "epoch": 3.7549367088607593,
+      "grad_norm": 0.2136659498209704,
+      "learning_rate": 6.949642934304375e-06,
+      "loss": 0.4437,
+      "step": 927
+    },
+    {
+      "epoch": 3.7589873417721518,
+      "grad_norm": 0.22189054757925264,
+      "learning_rate": 6.906685333692871e-06,
+      "loss": 0.4196,
+      "step": 928
+    },
+    {
+      "epoch": 3.7630379746835443,
+      "grad_norm": 0.24046849553582597,
+      "learning_rate": 6.86383318483769e-06,
+      "loss": 0.4388,
+      "step": 929
+    },
+    {
+      "epoch": 3.767088607594937,
+      "grad_norm": 0.2344221233851987,
+      "learning_rate": 6.821086832864139e-06,
+      "loss": 0.451,
+      "step": 930
+    },
+    {
+      "epoch": 3.7711392405063293,
+      "grad_norm": 0.2096767247506317,
+      "learning_rate": 6.77844662204546e-06,
+      "loss": 0.4432,
+      "step": 931
+    },
+    {
+      "epoch": 3.7751898734177214,
+      "grad_norm": 0.21596748340369631,
+      "learning_rate": 6.7359128958000455e-06,
+      "loss": 0.4422,
+      "step": 932
+    },
+    {
+      "epoch": 3.779240506329114,
+      "grad_norm": 0.2332541034540522,
+      "learning_rate": 6.693485996688695e-06,
+      "loss": 0.448,
+      "step": 933
+    },
+    {
+      "epoch": 3.7832911392405064,
+      "grad_norm": 0.22853097835310243,
+      "learning_rate": 6.651166266411801e-06,
+      "loss": 0.4402,
+      "step": 934
+    },
+    {
+      "epoch": 3.7873417721518985,
+      "grad_norm": 0.24531101757838236,
+      "learning_rate": 6.6089540458066725e-06,
+      "loss": 0.4349,
+      "step": 935
+    },
+    {
+      "epoch": 3.791392405063291,
+      "grad_norm": 0.24060692523071392,
+      "learning_rate": 6.566849674844711e-06,
+      "loss": 0.4374,
+      "step": 936
+    },
+    {
+      "epoch": 3.7954430379746835,
+      "grad_norm": 0.24175422603584337,
+      "learning_rate": 6.524853492628747e-06,
+      "loss": 0.4378,
+      "step": 937
+    },
+    {
+      "epoch": 3.799493670886076,
+      "grad_norm": 0.23622364718652,
+      "learning_rate": 6.4829658373902536e-06,
+      "loss": 0.4501,
+      "step": 938
+    },
+    {
+      "epoch": 3.8035443037974686,
+      "grad_norm": 0.22719806172023863,
+      "learning_rate": 6.441187046486648e-06,
+      "loss": 0.4445,
+      "step": 939
+    },
+    {
+      "epoch": 3.8075949367088606,
+      "grad_norm": 0.21836537786557286,
+      "learning_rate": 6.399517456398567e-06,
+      "loss": 0.4467,
+      "step": 940
+    },
+    {
+      "epoch": 3.811645569620253,
+      "grad_norm": 0.21014124915972784,
+      "learning_rate": 6.357957402727164e-06,
+      "loss": 0.4304,
+      "step": 941
+    },
+    {
+      "epoch": 3.8156962025316457,
+      "grad_norm": 0.23971490859669997,
+      "learning_rate": 6.316507220191395e-06,
+      "loss": 0.4421,
+      "step": 942
+    },
+    {
+      "epoch": 3.8197468354430377,
+      "grad_norm": 0.2207813544349984,
+      "learning_rate": 6.275167242625331e-06,
+      "loss": 0.4548,
+      "step": 943
+    },
+    {
+      "epoch": 3.8237974683544302,
+      "grad_norm": 0.2355979156066042,
+      "learning_rate": 6.233937802975471e-06,
+      "loss": 0.4338,
+      "step": 944
+    },
+    {
+      "epoch": 3.8278481012658228,
+      "grad_norm": 0.21286654559267287,
+      "learning_rate": 6.192819233298046e-06,
+      "loss": 0.4456,
+      "step": 945
+    },
+    {
+      "epoch": 3.8318987341772153,
+      "grad_norm": 0.21156902989815146,
+      "learning_rate": 6.151811864756383e-06,
+      "loss": 0.4332,
+      "step": 946
+    },
+    {
+      "epoch": 3.835949367088608,
+      "grad_norm": 0.21714643661657546,
+      "learning_rate": 6.1109160276181655e-06,
+      "loss": 0.4382,
+      "step": 947
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.21869278714234278,
+      "learning_rate": 6.070132051252868e-06,
+      "loss": 0.4454,
+      "step": 948
+    },
+    {
+      "epoch": 3.8440506329113924,
+      "grad_norm": 0.21969358760254182,
+      "learning_rate": 6.0294602641290034e-06,
+      "loss": 0.4461,
+      "step": 949
+    },
+    {
+      "epoch": 3.848101265822785,
+      "grad_norm": 0.20610155412554018,
+      "learning_rate": 5.988900993811575e-06,
+      "loss": 0.4388,
+      "step": 950
+    },
+    {
+      "epoch": 3.852151898734177,
+      "grad_norm": 0.20813274392332634,
+      "learning_rate": 5.948454566959363e-06,
+      "loss": 0.435,
+      "step": 951
+    },
+    {
+      "epoch": 3.8562025316455695,
+      "grad_norm": 0.20806717511788417,
+      "learning_rate": 5.908121309322328e-06,
+      "loss": 0.4324,
+      "step": 952
+    },
+    {
+      "epoch": 3.860253164556962,
+      "grad_norm": 0.24204646024694815,
+      "learning_rate": 5.867901545738976e-06,
+      "loss": 0.4435,
+      "step": 953
+    },
+    {
+      "epoch": 3.8643037974683545,
+      "grad_norm": 0.23939801049302076,
+      "learning_rate": 5.827795600133774e-06,
+      "loss": 0.4553,
+      "step": 954
+    },
+    {
+      "epoch": 3.868354430379747,
+      "grad_norm": 0.19585761085965928,
+      "learning_rate": 5.787803795514466e-06,
+      "loss": 0.4397,
+      "step": 955
+    },
+    {
+      "epoch": 3.872405063291139,
+      "grad_norm": 0.23415082868521614,
+      "learning_rate": 5.747926453969576e-06,
+      "loss": 0.4542,
+      "step": 956
+    },
+    {
+      "epoch": 3.8764556962025316,
+      "grad_norm": 0.24200812798837856,
+      "learning_rate": 5.708163896665708e-06,
+      "loss": 0.4349,
+      "step": 957
+    },
+    {
+      "epoch": 3.880506329113924,
+      "grad_norm": 0.22524091319381087,
+      "learning_rate": 5.668516443845047e-06,
+      "loss": 0.4328,
+      "step": 958
+    },
+    {
+      "epoch": 3.884556962025316,
+      "grad_norm": 0.21404111164415368,
+      "learning_rate": 5.6289844148227225e-06,
+      "loss": 0.4447,
+      "step": 959
+    },
+    {
+      "epoch": 3.8886075949367087,
+      "grad_norm": 0.23214746464453134,
+      "learning_rate": 5.5895681279842615e-06,
+      "loss": 0.4362,
+      "step": 960
+    },
+    {
+      "epoch": 3.8926582278481012,
+      "grad_norm": 0.2367094898709184,
+      "learning_rate": 5.550267900783019e-06,
+      "loss": 0.4358,
+      "step": 961
+    },
+    {
+      "epoch": 3.8967088607594937,
+      "grad_norm": 0.2165280778375853,
+      "learning_rate": 5.511084049737623e-06,
+      "loss": 0.4446,
+      "step": 962
+    },
+    {
+      "epoch": 3.9007594936708863,
+      "grad_norm": 0.21678810012843822,
+      "learning_rate": 5.4720168904294215e-06,
+      "loss": 0.4305,
+      "step": 963
+    },
+    {
+      "epoch": 3.9048101265822783,
+      "grad_norm": 0.21576517544995039,
+      "learning_rate": 5.433066737499948e-06,
+      "loss": 0.443,
+      "step": 964
+    },
+    {
+      "epoch": 3.908860759493671,
+      "grad_norm": 0.20835707316124955,
+      "learning_rate": 5.394233904648376e-06,
+      "loss": 0.4343,
+      "step": 965
+    },
+    {
+      "epoch": 3.9129113924050634,
+      "grad_norm": 0.19874550408903927,
+      "learning_rate": 5.355518704628997e-06,
+      "loss": 0.4478,
+      "step": 966
+    },
+    {
+      "epoch": 3.9169620253164554,
+      "grad_norm": 0.19977138393975258,
+      "learning_rate": 5.316921449248731e-06,
+      "loss": 0.4483,
+      "step": 967
+    },
+    {
+      "epoch": 3.921012658227848,
+      "grad_norm": 0.20818864901334272,
+      "learning_rate": 5.278442449364538e-06,
+      "loss": 0.4375,
+      "step": 968
+    },
+    {
+      "epoch": 3.9250632911392405,
+      "grad_norm": 0.20490103477878543,
+      "learning_rate": 5.240082014881016e-06,
+      "loss": 0.4386,
+      "step": 969
+    },
+    {
+      "epoch": 3.929113924050633,
+      "grad_norm": 0.2274593202358075,
+      "learning_rate": 5.201840454747822e-06,
+      "loss": 0.4408,
+      "step": 970
+    },
+    {
+      "epoch": 3.9331645569620255,
+      "grad_norm": 0.22254488494265065,
+      "learning_rate": 5.163718076957223e-06,
+      "loss": 0.4327,
+      "step": 971
+    },
+    {
+      "epoch": 3.9372151898734176,
+      "grad_norm": 0.21448881382747717,
+      "learning_rate": 5.125715188541609e-06,
+      "loss": 0.4421,
+      "step": 972
+    },
+    {
+      "epoch": 3.94126582278481,
+      "grad_norm": 0.2400002803577591,
+      "learning_rate": 5.087832095571021e-06,
+      "loss": 0.4402,
+      "step": 973
+    },
+    {
+      "epoch": 3.9453164556962026,
+      "grad_norm": 0.2124674740062061,
+      "learning_rate": 5.0500691031506766e-06,
+      "loss": 0.4274,
+      "step": 974
+    },
+    {
+      "epoch": 3.9493670886075947,
+      "grad_norm": 0.2163467457542528,
+      "learning_rate": 5.01242651541854e-06,
+      "loss": 0.4299,
+      "step": 975
+    },
+    {
+      "epoch": 3.953417721518987,
+      "grad_norm": 0.21141854454343234,
+      "learning_rate": 4.974904635542815e-06,
+      "loss": 0.4561,
+      "step": 976
+    },
+    {
+      "epoch": 3.9574683544303797,
+      "grad_norm": 0.21543716305781313,
+      "learning_rate": 4.937503765719582e-06,
+      "loss": 0.4342,
+      "step": 977
+    },
+    {
+      "epoch": 3.961518987341772,
+      "grad_norm": 0.21361693709797636,
+      "learning_rate": 4.900224207170299e-06,
+      "loss": 0.4419,
+      "step": 978
+    },
+    {
+      "epoch": 3.9655696202531647,
+      "grad_norm": 0.21181716900021674,
+      "learning_rate": 4.8630662601394065e-06,
+      "loss": 0.448,
+      "step": 979
+    },
+    {
+      "epoch": 3.969620253164557,
+      "grad_norm": 0.21650411759259824,
+      "learning_rate": 4.8260302238918995e-06,
+      "loss": 0.4345,
+      "step": 980
+    },
+    {
+      "epoch": 3.9736708860759493,
+      "grad_norm": 0.20579492785944545,
+      "learning_rate": 4.789116396710924e-06,
+      "loss": 0.4306,
+      "step": 981
+    },
+    {
+      "epoch": 3.977721518987342,
+      "grad_norm": 0.21291814375626159,
+      "learning_rate": 4.752325075895368e-06,
+      "loss": 0.4442,
+      "step": 982
+    },
+    {
+      "epoch": 3.981772151898734,
+      "grad_norm": 0.20456310505757413,
+      "learning_rate": 4.715656557757473e-06,
+      "loss": 0.4501,
+      "step": 983
+    },
+    {
+      "epoch": 3.9858227848101264,
+      "grad_norm": 0.20769756546559634,
+      "learning_rate": 4.679111137620442e-06,
+      "loss": 0.4458,
+      "step": 984
+    },
+    {
+      "epoch": 3.989873417721519,
+      "grad_norm": 0.2091411251619509,
+      "learning_rate": 4.6426891098160585e-06,
+      "loss": 0.4604,
+      "step": 985
+    },
+    {
+      "epoch": 3.9939240506329114,
+      "grad_norm": 0.20344280414543497,
+      "learning_rate": 4.6063907676823474e-06,
+      "loss": 0.4487,
+      "step": 986
+    },
+    {
+      "epoch": 3.997974683544304,
+      "grad_norm": 0.21624311365447793,
+      "learning_rate": 4.570216403561141e-06,
+      "loss": 0.4349,
+      "step": 987
+    },
+    {
+      "epoch": 4.002025316455696,
+      "grad_norm": 0.23386497939862205,
+      "learning_rate": 4.534166308795815e-06,
+      "loss": 0.4336,
+      "step": 988
+    },
+    {
+      "epoch": 4.006075949367089,
+      "grad_norm": 0.2961009682037505,
+      "learning_rate": 4.498240773728859e-06,
+      "loss": 0.4119,
+      "step": 989
+    },
+    {
+      "epoch": 4.010126582278481,
+      "grad_norm": 0.25523030672318026,
+      "learning_rate": 4.462440087699609e-06,
+      "loss": 0.4047,
+      "step": 990
+    },
+    {
+      "epoch": 4.014177215189873,
+      "grad_norm": 0.277229188464042,
+      "learning_rate": 4.426764539041861e-06,
+      "loss": 0.4101,
+      "step": 991
+    },
+    {
+      "epoch": 4.018227848101266,
+      "grad_norm": 0.3669576700860659,
+      "learning_rate": 4.391214415081582e-06,
+      "loss": 0.4111,
+      "step": 992
+    },
+    {
+      "epoch": 4.022278481012658,
+      "grad_norm": 0.2874258650621852,
+      "learning_rate": 4.355790002134579e-06,
+      "loss": 0.4133,
+      "step": 993
+    },
+    {
+      "epoch": 4.02632911392405,
+      "grad_norm": 0.2450446160107371,
+      "learning_rate": 4.320491585504207e-06,
+      "loss": 0.3933,
+      "step": 994
+    },
+    {
+      "epoch": 4.030379746835443,
+      "grad_norm": 0.34265764764315193,
+      "learning_rate": 4.2853194494790615e-06,
+      "loss": 0.3937,
+      "step": 995
+    },
+    {
+      "epoch": 4.034430379746835,
+      "grad_norm": 0.3411108232683835,
+      "learning_rate": 4.250273877330691e-06,
+      "loss": 0.4078,
+      "step": 996
+    },
+    {
+      "epoch": 4.038481012658228,
+      "grad_norm": 0.2264091125330083,
+      "learning_rate": 4.215355151311313e-06,
+      "loss": 0.4076,
+      "step": 997
+    },
+    {
+      "epoch": 4.04253164556962,
+      "grad_norm": 0.25132834085086786,
+      "learning_rate": 4.180563552651542e-06,
+      "loss": 0.3841,
+      "step": 998
+    },
+    {
+      "epoch": 4.046582278481012,
+      "grad_norm": 0.31882996468335423,
+      "learning_rate": 4.145899361558147e-06,
+      "loss": 0.4044,
+      "step": 999
+    },
+    {
+      "epoch": 4.050632911392405,
+      "grad_norm": 0.2656495916919769,
+      "learning_rate": 4.111362857211738e-06,
+      "loss": 0.4025,
+      "step": 1000
+    },
+    {
+      "epoch": 4.054683544303797,
+      "grad_norm": 0.22243482954720437,
+      "learning_rate": 4.076954317764592e-06,
+      "loss": 0.4031,
+      "step": 1001
+    },
+    {
+      "epoch": 4.0587341772151895,
+      "grad_norm": 0.2213578632685723,
+      "learning_rate": 4.042674020338335e-06,
+      "loss": 0.4183,
+      "step": 1002
+    },
+    {
+      "epoch": 4.062784810126582,
+      "grad_norm": 0.25299178685332185,
+      "learning_rate": 4.0085222410217835e-06,
+      "loss": 0.3812,
+      "step": 1003
+    },
+    {
+      "epoch": 4.0668354430379745,
+      "grad_norm": 0.2390299570120058,
+      "learning_rate": 3.974499254868674e-06,
+      "loss": 0.4041,
+      "step": 1004
+    },
+    {
+      "epoch": 4.0708860759493675,
+      "grad_norm": 0.22198125787024656,
+      "learning_rate": 3.940605335895451e-06,
+      "loss": 0.4102,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0749367088607595,
+      "grad_norm": 0.2222390374935532,
+      "learning_rate": 3.90684075707908e-06,
+      "loss": 0.4084,
+      "step": 1006
+    },
+    {
+      "epoch": 4.078987341772152,
+      "grad_norm": 0.23471817371695292,
+      "learning_rate": 3.8732057903548505e-06,
+      "loss": 0.4194,
+      "step": 1007
+    },
+    {
+      "epoch": 4.083037974683545,
+      "grad_norm": 0.21773695571160714,
+      "learning_rate": 3.8397007066141375e-06,
+      "loss": 0.408,
+      "step": 1008
+    },
+    {
+      "epoch": 4.087088607594937,
+      "grad_norm": 0.2171524188914636,
+      "learning_rate": 3.806325775702304e-06,
+      "loss": 0.4144,
+      "step": 1009
+    },
+    {
+      "epoch": 4.091139240506329,
+      "grad_norm": 0.21763731058892402,
+      "learning_rate": 3.773081266416434e-06,
+      "loss": 0.4042,
+      "step": 1010
+    },
+    {
+      "epoch": 4.095189873417722,
+      "grad_norm": 0.21960424221696428,
+      "learning_rate": 3.739967446503245e-06,
+      "loss": 0.4089,
+      "step": 1011
+    },
+    {
+      "epoch": 4.099240506329114,
+      "grad_norm": 0.2280508133887294,
+      "learning_rate": 3.706984582656894e-06,
+      "loss": 0.4038,
+      "step": 1012
+    },
+    {
+      "epoch": 4.103291139240507,
+      "grad_norm": 0.2189203124180907,
+      "learning_rate": 3.6741329405168237e-06,
+      "loss": 0.4097,
+      "step": 1013
+    },
+    {
+      "epoch": 4.107341772151899,
+      "grad_norm": 0.21276383349272174,
+      "learning_rate": 3.641412784665648e-06,
+      "loss": 0.3935,
+      "step": 1014
+    },
+    {
+      "epoch": 4.111392405063291,
+      "grad_norm": 0.19944173239941004,
+      "learning_rate": 3.608824378627005e-06,
+      "loss": 0.4134,
+      "step": 1015
+    },
+    {
+      "epoch": 4.115443037974684,
+      "grad_norm": 0.2145645431327895,
+      "learning_rate": 3.5763679848634337e-06,
+      "loss": 0.3922,
+      "step": 1016
+    },
+    {
+      "epoch": 4.119493670886076,
+      "grad_norm": 0.22178503443831962,
+      "learning_rate": 3.544043864774269e-06,
+      "loss": 0.3992,
+      "step": 1017
+    },
+    {
+      "epoch": 4.123544303797468,
+      "grad_norm": 0.2159787054536486,
+      "learning_rate": 3.5118522786935282e-06,
+      "loss": 0.4036,
+      "step": 1018
+    },
+    {
+      "epoch": 4.127594936708861,
+      "grad_norm": 0.2138178247140783,
+      "learning_rate": 3.479793485887819e-06,
+      "loss": 0.3994,
+      "step": 1019
+    },
+    {
+      "epoch": 4.131645569620253,
+      "grad_norm": 0.2132222880334988,
+      "learning_rate": 3.4478677445542653e-06,
+      "loss": 0.3947,
+      "step": 1020
+    },
+    {
+      "epoch": 4.135696202531646,
+      "grad_norm": 0.21516232257354634,
+      "learning_rate": 3.4160753118183767e-06,
+      "loss": 0.401,
+      "step": 1021
+    },
+    {
+      "epoch": 4.139746835443038,
+      "grad_norm": 0.23446227425405888,
+      "learning_rate": 3.3844164437320527e-06,
+      "loss": 0.4181,
+      "step": 1022
+    },
+    {
+      "epoch": 4.14379746835443,
+      "grad_norm": 0.20980901052221326,
+      "learning_rate": 3.3528913952714558e-06,
+      "loss": 0.4046,
+      "step": 1023
+    },
+    {
+      "epoch": 4.147848101265823,
+      "grad_norm": 0.20396663465471457,
+      "learning_rate": 3.321500420335e-06,
+      "loss": 0.4174,
+      "step": 1024
+    },
+    {
+      "epoch": 4.151898734177215,
+      "grad_norm": 0.20356713219163547,
+      "learning_rate": 3.290243771741275e-06,
+      "loss": 0.3897,
+      "step": 1025
+    },
+    {
+      "epoch": 4.155949367088607,
+      "grad_norm": 0.23497542466368426,
+      "learning_rate": 3.2591217012270325e-06,
+      "loss": 0.4164,
+      "step": 1026
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.20336386451488214,
+      "learning_rate": 3.228134459445149e-06,
+      "loss": 0.409,
+      "step": 1027
+    },
+    {
+      "epoch": 4.164050632911392,
+      "grad_norm": 0.20810864105504334,
+      "learning_rate": 3.1972822959626205e-06,
+      "loss": 0.4082,
+      "step": 1028
+    },
+    {
+      "epoch": 4.168101265822785,
+      "grad_norm": 0.2022311718814409,
+      "learning_rate": 3.166565459258513e-06,
+      "loss": 0.4042,
+      "step": 1029
+    },
+    {
+      "epoch": 4.172151898734177,
+      "grad_norm": 0.2050703177579398,
+      "learning_rate": 3.1359841967220193e-06,
+      "loss": 0.3995,
+      "step": 1030
+    },
+    {
+      "epoch": 4.176202531645569,
+      "grad_norm": 0.20237114097728034,
+      "learning_rate": 3.105538754650419e-06,
+      "loss": 0.4188,
+      "step": 1031
+    },
+    {
+      "epoch": 4.180253164556962,
+      "grad_norm": 0.20401758370995274,
+      "learning_rate": 3.07522937824712e-06,
+      "loss": 0.4043,
+      "step": 1032
+    },
+    {
+      "epoch": 4.184303797468354,
+      "grad_norm": 0.20295891879257102,
+      "learning_rate": 3.0450563116196697e-06,
+      "loss": 0.3965,
+      "step": 1033
+    },
+    {
+      "epoch": 4.188354430379746,
+      "grad_norm": 0.21547753299903669,
+      "learning_rate": 3.0150197977778008e-06,
+      "loss": 0.4029,
+      "step": 1034
+    },
+    {
+      "epoch": 4.192405063291139,
+      "grad_norm": 0.2102516986755663,
+      "learning_rate": 2.985120078631465e-06,
+      "loss": 0.4008,
+      "step": 1035
+    },
+    {
+      "epoch": 4.1964556962025314,
+      "grad_norm": 0.19589054050390972,
+      "learning_rate": 2.9553573949888893e-06,
+      "loss": 0.4088,
+      "step": 1036
+    },
+    {
+      "epoch": 4.200506329113924,
+      "grad_norm": 0.21211776310964586,
+      "learning_rate": 2.9257319865546384e-06,
+      "loss": 0.3864,
+      "step": 1037
+    },
+    {
+      "epoch": 4.2045569620253165,
+      "grad_norm": 0.21529356511078765,
+      "learning_rate": 2.896244091927678e-06,
+      "loss": 0.4023,
+      "step": 1038
+    },
+    {
+      "epoch": 4.2086075949367086,
+      "grad_norm": 0.20441572160069463,
+      "learning_rate": 2.8668939485994584e-06,
+      "loss": 0.3803,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2126582278481015,
+      "grad_norm": 0.20378253751249942,
+      "learning_rate": 2.837681792951994e-06,
+      "loss": 0.4248,
+      "step": 1040
+    },
+    {
+      "epoch": 4.216708860759494,
+      "grad_norm": 0.19756292840401754,
+      "learning_rate": 2.808607860255981e-06,
+      "loss": 0.4013,
+      "step": 1041
+    },
+    {
+      "epoch": 4.220759493670886,
+      "grad_norm": 0.20407006532033903,
+      "learning_rate": 2.7796723846688634e-06,
+      "loss": 0.4094,
+      "step": 1042
+    },
+    {
+      "epoch": 4.224810126582279,
+      "grad_norm": 0.19395812014286515,
+      "learning_rate": 2.7508755992329937e-06,
+      "loss": 0.4054,
+      "step": 1043
+    },
+    {
+      "epoch": 4.228860759493671,
+      "grad_norm": 0.2216152075510637,
+      "learning_rate": 2.722217735873718e-06,
+      "loss": 0.4131,
+      "step": 1044
+    },
+    {
+      "epoch": 4.232911392405064,
+      "grad_norm": 0.20948229559715498,
+      "learning_rate": 2.6936990253975315e-06,
+      "loss": 0.4048,
+      "step": 1045
+    },
+    {
+      "epoch": 4.236962025316456,
+      "grad_norm": 0.19928132574657414,
+      "learning_rate": 2.665319697490205e-06,
+      "loss": 0.4141,
+      "step": 1046
+    },
+    {
+      "epoch": 4.241012658227848,
+      "grad_norm": 0.20046472432352644,
+      "learning_rate": 2.637079980714945e-06,
+      "loss": 0.4144,
+      "step": 1047
+    },
+    {
+      "epoch": 4.245063291139241,
+      "grad_norm": 0.20468379118206548,
+      "learning_rate": 2.6089801025105453e-06,
+      "loss": 0.4073,
+      "step": 1048
+    },
+    {
+      "epoch": 4.249113924050633,
+      "grad_norm": 0.2040922112674285,
+      "learning_rate": 2.581020289189571e-06,
+      "loss": 0.4069,
+      "step": 1049
+    },
+    {
+      "epoch": 4.253164556962025,
+      "grad_norm": 0.21290162273609006,
+      "learning_rate": 2.553200765936501e-06,
+      "loss": 0.411,
+      "step": 1050
+    },
+    {
+      "epoch": 4.257215189873418,
+      "grad_norm": 0.20292550754338695,
+      "learning_rate": 2.525521756805962e-06,
+      "loss": 0.4076,
+      "step": 1051
+    },
+    {
+      "epoch": 4.26126582278481,
+      "grad_norm": 0.20847566628703418,
+      "learning_rate": 2.497983484720885e-06,
+      "loss": 0.4126,
+      "step": 1052
+    },
+    {
+      "epoch": 4.265316455696203,
+      "grad_norm": 0.20581052480965326,
+      "learning_rate": 2.470586171470728e-06,
+      "loss": 0.3937,
+      "step": 1053
+    },
+    {
+      "epoch": 4.269367088607595,
+      "grad_norm": 0.19588586724743176,
+      "learning_rate": 2.4433300377096836e-06,
+      "loss": 0.4023,
+      "step": 1054
+    },
+    {
+      "epoch": 4.273417721518987,
+      "grad_norm": 0.20706907451023684,
+      "learning_rate": 2.4162153029549073e-06,
+      "loss": 0.4085,
+      "step": 1055
+    },
+    {
+      "epoch": 4.27746835443038,
+      "grad_norm": 0.20783791600821222,
+      "learning_rate": 2.3892421855847458e-06,
+      "loss": 0.3946,
+      "step": 1056
+    },
+    {
+      "epoch": 4.281518987341772,
+      "grad_norm": 0.19950059230083925,
+      "learning_rate": 2.362410902836978e-06,
+      "loss": 0.3997,
+      "step": 1057
+    },
+    {
+      "epoch": 4.285569620253165,
+      "grad_norm": 0.20415807308706932,
+      "learning_rate": 2.3357216708070653e-06,
+      "loss": 0.411,
+      "step": 1058
+    },
+    {
+      "epoch": 4.289620253164557,
+      "grad_norm": 0.19845882619613556,
+      "learning_rate": 2.309174704446411e-06,
+      "loss": 0.4094,
+      "step": 1059
+    },
+    {
+      "epoch": 4.293670886075949,
+      "grad_norm": 0.2034097771431483,
+      "learning_rate": 2.2827702175606437e-06,
+      "loss": 0.3943,
+      "step": 1060
+    },
+    {
+      "epoch": 4.297721518987342,
+      "grad_norm": 0.1953207669876288,
+      "learning_rate": 2.256508422807855e-06,
+      "loss": 0.389,
+      "step": 1061
+    },
+    {
+      "epoch": 4.301772151898734,
+      "grad_norm": 0.20179511202013173,
+      "learning_rate": 2.230389531696946e-06,
+      "loss": 0.3966,
+      "step": 1062
+    },
+    {
+      "epoch": 4.305822784810126,
+      "grad_norm": 0.2017846361204614,
+      "learning_rate": 2.204413754585857e-06,
+      "loss": 0.4067,
+      "step": 1063
+    },
+    {
+      "epoch": 4.309873417721519,
+      "grad_norm": 0.19938093903420234,
+      "learning_rate": 2.1785813006799406e-06,
+      "loss": 0.4103,
+      "step": 1064
+    },
+    {
+      "epoch": 4.313924050632911,
+      "grad_norm": 0.19689925863281701,
+      "learning_rate": 2.1528923780302224e-06,
+      "loss": 0.3969,
+      "step": 1065
+    },
+    {
+      "epoch": 4.317974683544303,
+      "grad_norm": 0.19896049787313308,
+      "learning_rate": 2.127347193531757e-06,
+      "loss": 0.3859,
+      "step": 1066
+    },
+    {
+      "epoch": 4.322025316455696,
+      "grad_norm": 0.20221898225520293,
+      "learning_rate": 2.101945952921942e-06,
+      "loss": 0.4031,
+      "step": 1067
+    },
+    {
+      "epoch": 4.326075949367088,
+      "grad_norm": 0.20950619915082347,
+      "learning_rate": 2.0766888607788906e-06,
+      "loss": 0.4103,
+      "step": 1068
+    },
+    {
+      "epoch": 4.330126582278481,
+      "grad_norm": 0.20604056080498478,
+      "learning_rate": 2.0515761205197337e-06,
+      "loss": 0.4071,
+      "step": 1069
+    },
+    {
+      "epoch": 4.334177215189873,
+      "grad_norm": 0.201562681828564,
+      "learning_rate": 2.0266079343990453e-06,
+      "loss": 0.407,
+      "step": 1070
+    },
+    {
+      "epoch": 4.3382278481012655,
+      "grad_norm": 0.21470607942576178,
+      "learning_rate": 2.0017845035071494e-06,
+      "loss": 0.4065,
+      "step": 1071
+    },
+    {
+      "epoch": 4.3422784810126585,
+      "grad_norm": 0.19869285322437136,
+      "learning_rate": 1.9771060277685537e-06,
+      "loss": 0.4002,
+      "step": 1072
+    },
+    {
+      "epoch": 4.3463291139240505,
+      "grad_norm": 0.20610587626084576,
+      "learning_rate": 1.95257270594031e-06,
+      "loss": 0.4065,
+      "step": 1073
+    },
+    {
+      "epoch": 4.3503797468354435,
+      "grad_norm": 0.20934056110562457,
+      "learning_rate": 1.9281847356104188e-06,
+      "loss": 0.4073,
+      "step": 1074
+    },
+    {
+      "epoch": 4.3544303797468356,
+      "grad_norm": 0.20223445265657802,
+      "learning_rate": 1.9039423131962365e-06,
+      "loss": 0.3926,
+      "step": 1075
+    },
+    {
+      "epoch": 4.358481012658228,
+      "grad_norm": 0.19970345662293046,
+      "learning_rate": 1.8798456339429027e-06,
+      "loss": 0.4052,
+      "step": 1076
+    },
+    {
+      "epoch": 4.362531645569621,
+      "grad_norm": 0.21062052903991385,
+      "learning_rate": 1.8558948919217612e-06,
+      "loss": 0.4069,
+      "step": 1077
+    },
+    {
+      "epoch": 4.366582278481013,
+      "grad_norm": 0.19942746973330516,
+      "learning_rate": 1.8320902800287954e-06,
+      "loss": 0.4032,
+      "step": 1078
+    },
+    {
+      "epoch": 4.370632911392405,
+      "grad_norm": 0.20815824984232562,
+      "learning_rate": 1.8084319899830726e-06,
+      "loss": 0.4111,
+      "step": 1079
+    },
+    {
+      "epoch": 4.374683544303798,
+      "grad_norm": 0.2048783484895622,
+      "learning_rate": 1.7849202123252097e-06,
+      "loss": 0.4102,
+      "step": 1080
+    },
+    {
+      "epoch": 4.37873417721519,
+      "grad_norm": 0.20079406816503792,
+      "learning_rate": 1.7615551364158401e-06,
+      "loss": 0.4115,
+      "step": 1081
+    },
+    {
+      "epoch": 4.382784810126582,
+      "grad_norm": 0.43904162892401566,
+      "learning_rate": 1.738336950434061e-06,
+      "loss": 0.4175,
+      "step": 1082
+    },
+    {
+      "epoch": 4.386835443037975,
+      "grad_norm": 0.20595794171552603,
+      "learning_rate": 1.715265841375957e-06,
+      "loss": 0.3875,
+      "step": 1083
+    },
+    {
+      "epoch": 4.390886075949367,
+      "grad_norm": 0.2007741697510327,
+      "learning_rate": 1.6923419950530684e-06,
+      "loss": 0.3856,
+      "step": 1084
+    },
+    {
+      "epoch": 4.39493670886076,
+      "grad_norm": 0.19623535964805797,
+      "learning_rate": 1.6695655960909008e-06,
+      "loss": 0.4009,
+      "step": 1085
+    },
+    {
+      "epoch": 4.398987341772152,
+      "grad_norm": 0.19647094245458468,
+      "learning_rate": 1.646936827927441e-06,
+      "loss": 0.4082,
+      "step": 1086
+    },
+    {
+      "epoch": 4.403037974683544,
+      "grad_norm": 0.1976567941161319,
+      "learning_rate": 1.6244558728116766e-06,
+      "loss": 0.4187,
+      "step": 1087
+    },
+    {
+      "epoch": 4.407088607594937,
+      "grad_norm": 0.21058434723858532,
+      "learning_rate": 1.6021229118021265e-06,
+      "loss": 0.4047,
+      "step": 1088
+    },
+    {
+      "epoch": 4.411139240506329,
+      "grad_norm": 0.1959494373305206,
+      "learning_rate": 1.5799381247653967e-06,
+      "loss": 0.3932,
+      "step": 1089
+    },
+    {
+      "epoch": 4.415189873417722,
+      "grad_norm": 0.2102575072220191,
+      "learning_rate": 1.5579016903747013e-06,
+      "loss": 0.398,
+      "step": 1090
+    },
+    {
+      "epoch": 4.419240506329114,
+      "grad_norm": 0.1990049619944398,
+      "learning_rate": 1.5360137861084656e-06,
+      "loss": 0.3973,
+      "step": 1091
+    },
+    {
+      "epoch": 4.423291139240506,
+      "grad_norm": 0.20727174827508407,
+      "learning_rate": 1.5142745882488475e-06,
+      "loss": 0.4073,
+      "step": 1092
+    },
+    {
+      "epoch": 4.427341772151899,
+      "grad_norm": 0.20493681012543563,
+      "learning_rate": 1.4926842718803691e-06,
+      "loss": 0.3962,
+      "step": 1093
+    },
+    {
+      "epoch": 4.431392405063291,
+      "grad_norm": 0.20049757362944062,
+      "learning_rate": 1.4712430108884657e-06,
+      "loss": 0.3938,
+      "step": 1094
+    },
+    {
+      "epoch": 4.435443037974683,
+      "grad_norm": 0.19494883447280537,
+      "learning_rate": 1.4499509779581078e-06,
+      "loss": 0.4054,
+      "step": 1095
+    },
+    {
+      "epoch": 4.439493670886076,
+      "grad_norm": 0.19960080973314634,
+      "learning_rate": 1.4288083445723988e-06,
+      "loss": 0.3961,
+      "step": 1096
+    },
+    {
+      "epoch": 4.443544303797468,
+      "grad_norm": 0.19470994165964567,
+      "learning_rate": 1.4078152810112045e-06,
+      "loss": 0.4059,
+      "step": 1097
+    },
+    {
+      "epoch": 4.44759493670886,
+      "grad_norm": 0.21860691733484067,
+      "learning_rate": 1.3869719563497697e-06,
+      "loss": 0.3881,
+      "step": 1098
+    },
+    {
+      "epoch": 4.451645569620253,
+      "grad_norm": 0.19918789428533984,
+      "learning_rate": 1.3662785384573663e-06,
+      "loss": 0.407,
+      "step": 1099
+    },
+    {
+      "epoch": 4.455696202531645,
+      "grad_norm": 0.18821515472865952,
+      "learning_rate": 1.3457351939959383e-06,
+      "loss": 0.4113,
+      "step": 1100
+    },
+    {
+      "epoch": 4.459746835443038,
+      "grad_norm": 0.2003865045995853,
+      "learning_rate": 1.3253420884187551e-06,
+      "loss": 0.4016,
+      "step": 1101
+    },
+    {
+      "epoch": 4.46379746835443,
+      "grad_norm": 0.1942091226315613,
+      "learning_rate": 1.3050993859690953e-06,
+      "loss": 0.4205,
+      "step": 1102
+    },
+    {
+      "epoch": 4.467848101265822,
+      "grad_norm": 0.19768147088415053,
+      "learning_rate": 1.2850072496788869e-06,
+      "loss": 0.4045,
+      "step": 1103
+    },
+    {
+      "epoch": 4.471898734177215,
+      "grad_norm": 0.19741170619660295,
+      "learning_rate": 1.2650658413674434e-06,
+      "loss": 0.3945,
+      "step": 1104
+    },
+    {
+      "epoch": 4.4759493670886075,
+      "grad_norm": 0.19951999654640395,
+      "learning_rate": 1.2452753216401226e-06,
+      "loss": 0.4098,
+      "step": 1105
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.18768433218747588,
+      "learning_rate": 1.2256358498870503e-06,
+      "loss": 0.4135,
+      "step": 1106
+    },
+    {
+      "epoch": 4.4840506329113925,
+      "grad_norm": 0.2010962279125226,
+      "learning_rate": 1.2061475842818337e-06,
+      "loss": 0.3942,
+      "step": 1107
+    },
+    {
+      "epoch": 4.488101265822785,
+      "grad_norm": 0.1927093597025398,
+      "learning_rate": 1.1868106817802816e-06,
+      "loss": 0.4043,
+      "step": 1108
+    },
+    {
+      "epoch": 4.4921518987341775,
+      "grad_norm": 0.20126253781276499,
+      "learning_rate": 1.1676252981191482e-06,
+      "loss": 0.3896,
+      "step": 1109
+    },
+    {
+      "epoch": 4.49620253164557,
+      "grad_norm": 0.1898901558133873,
+      "learning_rate": 1.1485915878148823e-06,
+      "loss": 0.4187,
+      "step": 1110
+    },
+    {
+      "epoch": 4.500253164556962,
+      "grad_norm": 0.19570915731444272,
+      "learning_rate": 1.1297097041623584e-06,
+      "loss": 0.4064,
+      "step": 1111
+    },
+    {
+      "epoch": 4.504303797468355,
+      "grad_norm": 0.18842033367889097,
+      "learning_rate": 1.1109797992336847e-06,
+      "loss": 0.3992,
+      "step": 1112
+    },
+    {
+      "epoch": 4.508354430379747,
+      "grad_norm": 0.20571328230328254,
+      "learning_rate": 1.092402023876933e-06,
+      "loss": 0.3978,
+      "step": 1113
+    },
+    {
+      "epoch": 4.512405063291139,
+      "grad_norm": 0.20780615969361496,
+      "learning_rate": 1.0739765277149527e-06,
+      "loss": 0.3965,
+      "step": 1114
+    },
+    {
+      "epoch": 4.516455696202532,
+      "grad_norm": 0.4536278584408678,
+      "learning_rate": 1.0557034591441596e-06,
+      "loss": 0.4164,
+      "step": 1115
+    },
+    {
+      "epoch": 4.520506329113924,
+      "grad_norm": 0.202549843583585,
+      "learning_rate": 1.0375829653333324e-06,
+      "loss": 0.3967,
+      "step": 1116
+    },
+    {
+      "epoch": 4.524556962025317,
+      "grad_norm": 0.1956109191642723,
+      "learning_rate": 1.0196151922224385e-06,
+      "loss": 0.4038,
+      "step": 1117
+    },
+    {
+      "epoch": 4.528607594936709,
+      "grad_norm": 0.20355577885863113,
+      "learning_rate": 1.0018002845214526e-06,
+      "loss": 0.3989,
+      "step": 1118
+    },
+    {
+      "epoch": 4.532658227848101,
+      "grad_norm": 0.21046401044500926,
+      "learning_rate": 9.841383857091947e-07,
+      "loss": 0.4051,
+      "step": 1119
+    },
+    {
+      "epoch": 4.536708860759494,
+      "grad_norm": 0.19374922752437587,
+      "learning_rate": 9.666296380321616e-07,
+      "loss": 0.4034,
+      "step": 1120
+    },
+    {
+      "epoch": 4.540759493670886,
+      "grad_norm": 0.19554113137394583,
+      "learning_rate": 9.492741825034124e-07,
+      "loss": 0.4092,
+      "step": 1121
+    },
+    {
+      "epoch": 4.544810126582279,
+      "grad_norm": 0.1967188072525825,
+      "learning_rate": 9.320721589013892e-07,
+      "loss": 0.4176,
+      "step": 1122
+    },
+    {
+      "epoch": 4.548860759493671,
+      "grad_norm": 0.1954056544114555,
+      "learning_rate": 9.150237057688339e-07,
+      "loss": 0.399,
+      "step": 1123
+    },
+    {
+      "epoch": 4.552911392405063,
+      "grad_norm": 0.1963846652467093,
+      "learning_rate": 8.981289604116328e-07,
+      "loss": 0.3927,
+      "step": 1124
+    },
+    {
+      "epoch": 4.556962025316456,
+      "grad_norm": 0.19082610043039436,
+      "learning_rate": 8.813880588977542e-07,
+      "loss": 0.4048,
+      "step": 1125
+    },
+    {
+      "epoch": 4.561012658227848,
+      "grad_norm": 0.19310174850064066,
+      "learning_rate": 8.648011360561126e-07,
+      "loss": 0.3977,
+      "step": 1126
+    },
+    {
+      "epoch": 4.56506329113924,
+      "grad_norm": 0.19792236858270332,
+      "learning_rate": 8.483683254755037e-07,
+      "loss": 0.4045,
+      "step": 1127
+    },
+    {
+      "epoch": 4.569113924050633,
+      "grad_norm": 0.19019303695262843,
+      "learning_rate": 8.320897595035227e-07,
+      "loss": 0.401,
+      "step": 1128
+    },
+    {
+      "epoch": 4.573164556962025,
+      "grad_norm": 0.19733672550393266,
+      "learning_rate": 8.159655692455093e-07,
+      "loss": 0.3984,
+      "step": 1129
+    },
+    {
+      "epoch": 4.577215189873417,
+      "grad_norm": 0.1987994664994913,
+      "learning_rate": 7.999958845634648e-07,
+      "loss": 0.3966,
+      "step": 1130
+    },
+    {
+      "epoch": 4.58126582278481,
+      "grad_norm": 0.19197647673390686,
+      "learning_rate": 7.841808340750478e-07,
+      "loss": 0.4001,
+      "step": 1131
+    },
+    {
+      "epoch": 4.585316455696202,
+      "grad_norm": 0.197015395941159,
+      "learning_rate": 7.685205451524869e-07,
+      "loss": 0.4138,
+      "step": 1132
+    },
+    {
+      "epoch": 4.589367088607595,
+      "grad_norm": 0.4023688006004889,
+      "learning_rate": 7.530151439216027e-07,
+      "loss": 0.4185,
+      "step": 1133
+    },
+    {
+      "epoch": 4.593417721518987,
+      "grad_norm": 0.2000328305274102,
+      "learning_rate": 7.376647552607675e-07,
+      "loss": 0.4144,
+      "step": 1134
+    },
+    {
+      "epoch": 4.597468354430379,
+      "grad_norm": 0.20782100437690215,
+      "learning_rate": 7.224695027998963e-07,
+      "loss": 0.392,
+      "step": 1135
+    },
+    {
+      "epoch": 4.601518987341772,
+      "grad_norm": 0.18628749095926794,
+      "learning_rate": 7.07429508919466e-07,
+      "loss": 0.4072,
+      "step": 1136
+    },
+    {
+      "epoch": 4.605569620253164,
+      "grad_norm": 0.19830687026418478,
+      "learning_rate": 6.925448947495206e-07,
+      "loss": 0.404,
+      "step": 1137
+    },
+    {
+      "epoch": 4.609620253164557,
+      "grad_norm": 0.20063579859309066,
+      "learning_rate": 6.778157801686936e-07,
+      "loss": 0.397,
+      "step": 1138
+    },
+    {
+      "epoch": 4.613670886075949,
+      "grad_norm": 0.18882679222440826,
+      "learning_rate": 6.632422838032515e-07,
+      "loss": 0.406,
+      "step": 1139
+    },
+    {
+      "epoch": 4.6177215189873415,
+      "grad_norm": 0.19803965446942645,
+      "learning_rate": 6.488245230261281e-07,
+      "loss": 0.408,
+      "step": 1140
+    },
+    {
+      "epoch": 4.6217721518987345,
+      "grad_norm": 0.19764538472816623,
+      "learning_rate": 6.345626139559868e-07,
+      "loss": 0.4115,
+      "step": 1141
+    },
+    {
+      "epoch": 4.6258227848101265,
+      "grad_norm": 0.19782821696776084,
+      "learning_rate": 6.204566714562866e-07,
+      "loss": 0.4123,
+      "step": 1142
+    },
+    {
+      "epoch": 4.629873417721519,
+      "grad_norm": 0.20317104219915383,
+      "learning_rate": 6.06506809134344e-07,
+      "loss": 0.3965,
+      "step": 1143
+    },
+    {
+      "epoch": 4.633924050632912,
+      "grad_norm": 0.1974641730460825,
+      "learning_rate": 5.927131393404373e-07,
+      "loss": 0.403,
+      "step": 1144
+    },
+    {
+      "epoch": 4.637974683544304,
+      "grad_norm": 0.196235522532761,
+      "learning_rate": 5.790757731668817e-07,
+      "loss": 0.4157,
+      "step": 1145
+    },
+    {
+      "epoch": 4.642025316455696,
+      "grad_norm": 0.19413129537018745,
+      "learning_rate": 5.655948204471507e-07,
+      "loss": 0.4033,
+      "step": 1146
+    },
+    {
+      "epoch": 4.646075949367089,
+      "grad_norm": 0.1955291966673253,
+      "learning_rate": 5.522703897549875e-07,
+      "loss": 0.4028,
+      "step": 1147
+    },
+    {
+      "epoch": 4.650126582278481,
+      "grad_norm": 0.1889072301095045,
+      "learning_rate": 5.391025884035239e-07,
+      "loss": 0.4011,
+      "step": 1148
+    },
+    {
+      "epoch": 4.654177215189874,
+      "grad_norm": 0.19256689788451534,
+      "learning_rate": 5.260915224444207e-07,
+      "loss": 0.3973,
+      "step": 1149
+    },
+    {
+      "epoch": 4.658227848101266,
+      "grad_norm": 0.19483126027789915,
+      "learning_rate": 5.132372966670129e-07,
+      "loss": 0.4079,
+      "step": 1150
+    },
+    {
+      "epoch": 4.662278481012658,
+      "grad_norm": 0.19473978548672682,
+      "learning_rate": 5.005400145974704e-07,
+      "loss": 0.4096,
+      "step": 1151
+    },
+    {
+      "epoch": 4.666329113924051,
+      "grad_norm": 0.19182894839417897,
+      "learning_rate": 4.879997784979562e-07,
+      "loss": 0.3998,
+      "step": 1152
+    },
+    {
+      "epoch": 4.670379746835443,
+      "grad_norm": 0.19525511738435028,
+      "learning_rate": 4.7561668936580984e-07,
+      "loss": 0.4058,
+      "step": 1153
+    },
+    {
+      "epoch": 4.674430379746836,
+      "grad_norm": 0.19076996804772717,
+      "learning_rate": 4.6339084693272306e-07,
+      "loss": 0.4071,
+      "step": 1154
+    },
+    {
+      "epoch": 4.678481012658228,
+      "grad_norm": 0.1986900111163795,
+      "learning_rate": 4.5132234966395847e-07,
+      "loss": 0.389,
+      "step": 1155
+    },
+    {
+      "epoch": 4.68253164556962,
+      "grad_norm": 0.19571096518967623,
+      "learning_rate": 4.3941129475752795e-07,
+      "loss": 0.3974,
+      "step": 1156
+    },
+    {
+      "epoch": 4.686582278481013,
+      "grad_norm": 0.19465632523541707,
+      "learning_rate": 4.27657778143431e-07,
+      "loss": 0.4069,
+      "step": 1157
+    },
+    {
+      "epoch": 4.690632911392405,
+      "grad_norm": 0.19112278799815677,
+      "learning_rate": 4.1606189448287757e-07,
+      "loss": 0.4028,
+      "step": 1158
+    },
+    {
+      "epoch": 4.694683544303797,
+      "grad_norm": 0.19725131045679534,
+      "learning_rate": 4.046237371675177e-07,
+      "loss": 0.4003,
+      "step": 1159
+    },
+    {
+      "epoch": 4.69873417721519,
+      "grad_norm": 0.19569481148204834,
+      "learning_rate": 3.9334339831869963e-07,
+      "loss": 0.4052,
+      "step": 1160
+    },
+    {
+      "epoch": 4.702784810126582,
+      "grad_norm": 0.19816881679960127,
+      "learning_rate": 3.8222096878671955e-07,
+      "loss": 0.407,
+      "step": 1161
+    },
+    {
+      "epoch": 4.706835443037974,
+      "grad_norm": 0.1893180680611313,
+      "learning_rate": 3.7125653815009545e-07,
+      "loss": 0.4036,
+      "step": 1162
+    },
+    {
+      "epoch": 4.710886075949367,
+      "grad_norm": 0.20150245817323417,
+      "learning_rate": 3.6045019471484974e-07,
+      "loss": 0.3902,
+      "step": 1163
+    },
+    {
+      "epoch": 4.714936708860759,
+      "grad_norm": 0.19283556058066573,
+      "learning_rate": 3.498020255137813e-07,
+      "loss": 0.4001,
+      "step": 1164
+    },
+    {
+      "epoch": 4.718987341772152,
+      "grad_norm": 0.20296956301139377,
+      "learning_rate": 3.393121163057811e-07,
+      "loss": 0.3874,
+      "step": 1165
+    },
+    {
+      "epoch": 4.723037974683544,
+      "grad_norm": 0.19432993580525873,
+      "learning_rate": 3.289805515751399e-07,
+      "loss": 0.398,
+      "step": 1166
+    },
+    {
+      "epoch": 4.727088607594936,
+      "grad_norm": 0.18641553766245078,
+      "learning_rate": 3.188074145308573e-07,
+      "loss": 0.4,
+      "step": 1167
+    },
+    {
+      "epoch": 4.731139240506329,
+      "grad_norm": 0.1905324599647721,
+      "learning_rate": 3.087927871059804e-07,
+      "loss": 0.4113,
+      "step": 1168
+    },
+    {
+      "epoch": 4.735189873417721,
+      "grad_norm": 0.1980720922831672,
+      "learning_rate": 2.989367499569418e-07,
+      "loss": 0.4004,
+      "step": 1169
+    },
+    {
+      "epoch": 4.739240506329114,
+      "grad_norm": 0.19398058041623226,
+      "learning_rate": 2.8923938246290917e-07,
+      "loss": 0.4062,
+      "step": 1170
+    },
+    {
+      "epoch": 4.743291139240506,
+      "grad_norm": 0.19796190741952188,
+      "learning_rate": 2.7970076272514804e-07,
+      "loss": 0.3942,
+      "step": 1171
+    },
+    {
+      "epoch": 4.7473417721518985,
+      "grad_norm": 0.19385568395694583,
+      "learning_rate": 2.703209675663887e-07,
+      "loss": 0.4019,
+      "step": 1172
+    },
+    {
+      "epoch": 4.751392405063291,
+      "grad_norm": 0.1932504842098403,
+      "learning_rate": 2.6110007253021374e-07,
+      "loss": 0.4068,
+      "step": 1173
+    },
+    {
+      "epoch": 4.7554430379746835,
+      "grad_norm": 0.1896336965428746,
+      "learning_rate": 2.520381518804471e-07,
+      "loss": 0.4059,
+      "step": 1174
+    },
+    {
+      "epoch": 4.759493670886076,
+      "grad_norm": 0.19034372803244262,
+      "learning_rate": 2.4313527860054585e-07,
+      "loss": 0.4046,
+      "step": 1175
+    },
+    {
+      "epoch": 4.7635443037974685,
+      "grad_norm": 0.1929314326210692,
+      "learning_rate": 2.343915243930317e-07,
+      "loss": 0.399,
+      "step": 1176
+    },
+    {
+      "epoch": 4.767594936708861,
+      "grad_norm": 0.17840224600482255,
+      "learning_rate": 2.2580695967889367e-07,
+      "loss": 0.428,
+      "step": 1177
+    },
+    {
+      "epoch": 4.7716455696202535,
+      "grad_norm": 0.19186801443449564,
+      "learning_rate": 2.1738165359704189e-07,
+      "loss": 0.3875,
+      "step": 1178
+    },
+    {
+      "epoch": 4.775696202531646,
+      "grad_norm": 0.18736326338059783,
+      "learning_rate": 2.0911567400373257e-07,
+      "loss": 0.4007,
+      "step": 1179
+    },
+    {
+      "epoch": 4.779746835443038,
+      "grad_norm": 0.19057590789197418,
+      "learning_rate": 2.0100908747202607e-07,
+      "loss": 0.4043,
+      "step": 1180
+    },
+    {
+      "epoch": 4.783797468354431,
+      "grad_norm": 0.19136983678672795,
+      "learning_rate": 1.9306195929125638e-07,
+      "loss": 0.3984,
+      "step": 1181
+    },
+    {
+      "epoch": 4.787848101265823,
+      "grad_norm": 0.1930856720368308,
+      "learning_rate": 1.8527435346650247e-07,
+      "loss": 0.3885,
+      "step": 1182
+    },
+    {
+      "epoch": 4.791898734177215,
+      "grad_norm": 0.1983822013319305,
+      "learning_rate": 1.7764633271807108e-07,
+      "loss": 0.3971,
+      "step": 1183
+    },
+    {
+      "epoch": 4.795949367088608,
+      "grad_norm": 0.19412501807168506,
+      "learning_rate": 1.7017795848099262e-07,
+      "loss": 0.3953,
+      "step": 1184
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.1929029604138729,
+      "learning_rate": 1.6286929090452596e-07,
+      "loss": 0.4063,
+      "step": 1185
+    },
+    {
+      "epoch": 4.804050632911393,
+      "grad_norm": 0.19213576585067302,
+      "learning_rate": 1.557203888516745e-07,
+      "loss": 0.3987,
+      "step": 1186
+    },
+    {
+      "epoch": 4.808101265822785,
+      "grad_norm": 0.1929065617170961,
+      "learning_rate": 1.487313098987131e-07,
+      "loss": 0.415,
+      "step": 1187
+    },
+    {
+      "epoch": 4.812151898734177,
+      "grad_norm": 0.19205243866715127,
+      "learning_rate": 1.4190211033472402e-07,
+      "loss": 0.4027,
+      "step": 1188
+    },
+    {
+      "epoch": 4.81620253164557,
+      "grad_norm": 0.18835553195847815,
+      "learning_rate": 1.3523284516113955e-07,
+      "loss": 0.4142,
+      "step": 1189
+    },
+    {
+      "epoch": 4.820253164556962,
+      "grad_norm": 0.1930996286856906,
+      "learning_rate": 1.2872356809130682e-07,
+      "loss": 0.4109,
+      "step": 1190
+    },
+    {
+      "epoch": 4.824303797468355,
+      "grad_norm": 0.18858078214858134,
+      "learning_rate": 1.2237433155004807e-07,
+      "loss": 0.4135,
+      "step": 1191
+    },
+    {
+      "epoch": 4.828354430379747,
+      "grad_norm": 0.2023320073362713,
+      "learning_rate": 1.1618518667323886e-07,
+      "loss": 0.3825,
+      "step": 1192
+    },
+    {
+      "epoch": 4.832405063291139,
+      "grad_norm": 0.20700240755976784,
+      "learning_rate": 1.1015618330740385e-07,
+      "loss": 0.4058,
+      "step": 1193
+    },
+    {
+      "epoch": 4.836455696202532,
+      "grad_norm": 0.189961235045587,
+      "learning_rate": 1.042873700093061e-07,
+      "loss": 0.4026,
+      "step": 1194
+    },
+    {
+      "epoch": 4.840506329113924,
+      "grad_norm": 0.20038435391626436,
+      "learning_rate": 9.857879404556291e-08,
+      "loss": 0.3989,
+      "step": 1195
+    },
+    {
+      "epoch": 4.844556962025316,
+      "grad_norm": 0.19711162434149523,
+      "learning_rate": 9.303050139225722e-08,
+      "loss": 0.4034,
+      "step": 1196
+    },
+    {
+      "epoch": 4.848607594936709,
+      "grad_norm": 0.1887296673121165,
+      "learning_rate": 8.76425367345779e-08,
+      "loss": 0.4074,
+      "step": 1197
+    },
+    {
+      "epoch": 4.852658227848101,
+      "grad_norm": 0.1899523032126636,
+      "learning_rate": 8.241494346644897e-08,
+      "loss": 0.3969,
+      "step": 1198
+    },
+    {
+      "epoch": 4.856708860759493,
+      "grad_norm": 0.1876677592060378,
+      "learning_rate": 7.734776369019204e-08,
+      "loss": 0.4052,
+      "step": 1199
+    },
+    {
+      "epoch": 4.860759493670886,
+      "grad_norm": 0.19579619375115953,
+      "learning_rate": 7.244103821617332e-08,
+      "loss": 0.4063,
+      "step": 1200
+    },
+    {
+      "epoch": 4.864810126582278,
+      "grad_norm": 0.19796255998511975,
+      "learning_rate": 6.769480656248606e-08,
+      "loss": 0.4143,
+      "step": 1201
+    },
+    {
+      "epoch": 4.868860759493671,
+      "grad_norm": 0.18649219899482125,
+      "learning_rate": 6.310910695462635e-08,
+      "loss": 0.4081,
+      "step": 1202
+    },
+    {
+      "epoch": 4.872911392405063,
+      "grad_norm": 0.19216362905606238,
+      "learning_rate": 5.8683976325191185e-08,
+      "loss": 0.3932,
+      "step": 1203
+    },
+    {
+      "epoch": 4.876962025316455,
+      "grad_norm": 0.19204416174597172,
+      "learning_rate": 5.4419450313571984e-08,
+      "loss": 0.4082,
+      "step": 1204
+    },
+    {
+      "epoch": 4.881012658227848,
+      "grad_norm": 0.19090059942733026,
+      "learning_rate": 5.031556326567488e-08,
+      "loss": 0.4012,
+      "step": 1205
+    },
+    {
+      "epoch": 4.88506329113924,
+      "grad_norm": 0.19175617617543675,
+      "learning_rate": 4.637234823364312e-08,
+      "loss": 0.4118,
+      "step": 1206
+    },
+    {
+      "epoch": 4.889113924050633,
+      "grad_norm": 0.20269396373965806,
+      "learning_rate": 4.258983697558838e-08,
+      "loss": 0.3904,
+      "step": 1207
+    },
+    {
+      "epoch": 4.8931645569620255,
+      "grad_norm": 0.1922203757691003,
+      "learning_rate": 3.896805995533548e-08,
+      "loss": 0.3959,
+      "step": 1208
+    },
+    {
+      "epoch": 4.8972151898734175,
+      "grad_norm": 0.19729157488607657,
+      "learning_rate": 3.550704634218028e-08,
+      "loss": 0.3957,
+      "step": 1209
+    },
+    {
+      "epoch": 4.9012658227848105,
+      "grad_norm": 0.18308927334554267,
+      "learning_rate": 3.2206824010647676e-08,
+      "loss": 0.4059,
+      "step": 1210
+    },
+    {
+      "epoch": 4.905316455696203,
+      "grad_norm": 0.18990370006971424,
+      "learning_rate": 2.9067419540278476e-08,
+      "loss": 0.3946,
+      "step": 1211
+    },
+    {
+      "epoch": 4.909367088607595,
+      "grad_norm": 0.20342055050399474,
+      "learning_rate": 2.6088858215400638e-08,
+      "loss": 0.4066,
+      "step": 1212
+    },
+    {
+      "epoch": 4.913417721518988,
+      "grad_norm": 0.1965818692184798,
+      "learning_rate": 2.3271164024940564e-08,
+      "loss": 0.4283,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91746835443038,
+      "grad_norm": 0.1962769735048317,
+      "learning_rate": 2.061435966221881e-08,
+      "loss": 0.3979,
+      "step": 1214
+    },
+    {
+      "epoch": 4.921518987341772,
+      "grad_norm": 0.18770453645108612,
+      "learning_rate": 1.811846652477245e-08,
+      "loss": 0.4086,
+      "step": 1215
+    },
+    {
+      "epoch": 4.925569620253165,
+      "grad_norm": 0.18584357510121988,
+      "learning_rate": 1.5783504714184106e-08,
+      "loss": 0.4125,
+      "step": 1216
+    },
+    {
+      "epoch": 4.929620253164557,
+      "grad_norm": 0.18922550664720877,
+      "learning_rate": 1.360949303591097e-08,
+      "loss": 0.414,
+      "step": 1217
+    },
+    {
+      "epoch": 4.93367088607595,
+      "grad_norm": 0.18870723557883531,
+      "learning_rate": 1.1596448999144916e-08,
+      "loss": 0.3995,
+      "step": 1218
+    },
+    {
+      "epoch": 4.937721518987342,
+      "grad_norm": 0.20111618149649702,
+      "learning_rate": 9.744388816668172e-09,
+      "loss": 0.4066,
+      "step": 1219
+    },
+    {
+      "epoch": 4.941772151898734,
+      "grad_norm": 0.18879233992261057,
+      "learning_rate": 8.05332740472009e-09,
+      "loss": 0.3937,
+      "step": 1220
+    },
+    {
+      "epoch": 4.945822784810127,
+      "grad_norm": 0.19540212242153251,
+      "learning_rate": 6.523278382872811e-09,
+      "loss": 0.4006,
+      "step": 1221
+    },
+    {
+      "epoch": 4.949873417721519,
+      "grad_norm": 0.2026158335350584,
+      "learning_rate": 5.15425407393133e-09,
+      "loss": 0.3999,
+      "step": 1222
+    },
+    {
+      "epoch": 4.953924050632912,
+      "grad_norm": 0.18589335446163627,
+      "learning_rate": 3.94626550383137e-09,
+      "loss": 0.4057,
+      "step": 1223
+    },
+    {
+      "epoch": 4.957974683544304,
+      "grad_norm": 0.1854788538185702,
+      "learning_rate": 2.899322401546112e-09,
+      "loss": 0.4046,
+      "step": 1224
+    },
+    {
+      "epoch": 4.962025316455696,
+      "grad_norm": 0.19713817614959497,
+      "learning_rate": 2.013433199010706e-09,
+      "loss": 0.3869,
+      "step": 1225
+    },
+    {
+      "epoch": 4.966075949367089,
+      "grad_norm": 0.19843395839436742,
+      "learning_rate": 1.2886050310556563e-09,
+      "loss": 0.3937,
+      "step": 1226
+    },
+    {
+      "epoch": 4.970126582278481,
+      "grad_norm": 0.20952583436367642,
+      "learning_rate": 7.248437353468695e-10,
+      "loss": 0.3825,
+      "step": 1227
+    },
+    {
+      "epoch": 4.974177215189873,
+      "grad_norm": 0.18785899009853088,
+      "learning_rate": 3.221538523412449e-10,
+      "loss": 0.416,
+      "step": 1228
+    },
+    {
+      "epoch": 4.978227848101266,
+      "grad_norm": 0.1990633300075058,
+      "learning_rate": 8.053862524670663e-11,
+      "loss": 0.3924,
+      "step": 1229
+    },
+    {
+      "epoch": 4.982278481012658,
+      "grad_norm": 0.1959312353918406,
+      "learning_rate": 0.0,
+      "loss": 0.396,
+      "step": 1230
+    },
+    {
+      "epoch": 4.982278481012658,
+      "step": 1230,
+      "total_flos": 6.089346265168577e+18,
+      "train_loss": 0.519107237773213,
+      "train_runtime": 56157.9676,
+      "train_samples_per_second": 2.813,
+      "train_steps_per_second": 0.022
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1230,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "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": 6.089346265168577e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}