diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7342 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.66816,
+  "eval_steps": 500,
+  "global_step": 1044,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00064,
+      "grad_norm": 2.353492879681207,
+      "learning_rate": 0.0,
+      "loss": 0.5764,
+      "step": 1
+    },
+    {
+      "epoch": 0.00128,
+      "grad_norm": 2.567574488545934,
+      "learning_rate": 6.329113924050633e-07,
+      "loss": 0.6154,
+      "step": 2
+    },
+    {
+      "epoch": 0.00192,
+      "grad_norm": 2.452013107124523,
+      "learning_rate": 1.2658227848101265e-06,
+      "loss": 0.5763,
+      "step": 3
+    },
+    {
+      "epoch": 0.00256,
+      "grad_norm": 2.6431662747915237,
+      "learning_rate": 1.8987341772151901e-06,
+      "loss": 0.5897,
+      "step": 4
+    },
+    {
+      "epoch": 0.0032,
+      "grad_norm": 2.3930737468544336,
+      "learning_rate": 2.531645569620253e-06,
+      "loss": 0.5846,
+      "step": 5
+    },
+    {
+      "epoch": 0.00384,
+      "grad_norm": 2.594961068226395,
+      "learning_rate": 3.1645569620253167e-06,
+      "loss": 0.6,
+      "step": 6
+    },
+    {
+      "epoch": 0.00448,
+      "grad_norm": 1.8595975894553154,
+      "learning_rate": 3.7974683544303802e-06,
+      "loss": 0.5579,
+      "step": 7
+    },
+    {
+      "epoch": 0.00512,
+      "grad_norm": 1.9840836213880872,
+      "learning_rate": 4.430379746835443e-06,
+      "loss": 0.5861,
+      "step": 8
+    },
+    {
+      "epoch": 0.00576,
+      "grad_norm": 1.3806056222455356,
+      "learning_rate": 5.063291139240506e-06,
+      "loss": 0.5483,
+      "step": 9
+    },
+    {
+      "epoch": 0.0064,
+      "grad_norm": 1.2378254593133642,
+      "learning_rate": 5.69620253164557e-06,
+      "loss": 0.5306,
+      "step": 10
+    },
+    {
+      "epoch": 0.00704,
+      "grad_norm": 1.1874568518218973,
+      "learning_rate": 6.329113924050633e-06,
+      "loss": 0.5569,
+      "step": 11
+    },
+    {
+      "epoch": 0.00768,
+      "grad_norm": 1.4792517105132126,
+      "learning_rate": 6.9620253164556965e-06,
+      "loss": 0.511,
+      "step": 12
+    },
+    {
+      "epoch": 0.00832,
+      "grad_norm": 1.5972888909878988,
+      "learning_rate": 7.5949367088607605e-06,
+      "loss": 0.4674,
+      "step": 13
+    },
+    {
+      "epoch": 0.00896,
+      "grad_norm": 1.5705765751301897,
+      "learning_rate": 8.227848101265822e-06,
+      "loss": 0.4798,
+      "step": 14
+    },
+    {
+      "epoch": 0.0096,
+      "grad_norm": 1.2132734057876817,
+      "learning_rate": 8.860759493670886e-06,
+      "loss": 0.4411,
+      "step": 15
+    },
+    {
+      "epoch": 0.01024,
+      "grad_norm": 1.0220542665261685,
+      "learning_rate": 9.49367088607595e-06,
+      "loss": 0.4258,
+      "step": 16
+    },
+    {
+      "epoch": 0.01088,
+      "grad_norm": 1.3769301736898738,
+      "learning_rate": 1.0126582278481012e-05,
+      "loss": 0.4543,
+      "step": 17
+    },
+    {
+      "epoch": 0.01152,
+      "grad_norm": 1.2058856706461065,
+      "learning_rate": 1.0759493670886076e-05,
+      "loss": 0.4526,
+      "step": 18
+    },
+    {
+      "epoch": 0.01216,
+      "grad_norm": 0.9279758572161336,
+      "learning_rate": 1.139240506329114e-05,
+      "loss": 0.4146,
+      "step": 19
+    },
+    {
+      "epoch": 0.0128,
+      "grad_norm": 0.7302969117448893,
+      "learning_rate": 1.2025316455696203e-05,
+      "loss": 0.4266,
+      "step": 20
+    },
+    {
+      "epoch": 0.01344,
+      "grad_norm": 0.7558992094228546,
+      "learning_rate": 1.2658227848101267e-05,
+      "loss": 0.4315,
+      "step": 21
+    },
+    {
+      "epoch": 0.01408,
+      "grad_norm": 0.6498887321885588,
+      "learning_rate": 1.3291139240506329e-05,
+      "loss": 0.3667,
+      "step": 22
+    },
+    {
+      "epoch": 0.01472,
+      "grad_norm": 0.7438110419135116,
+      "learning_rate": 1.3924050632911393e-05,
+      "loss": 0.3827,
+      "step": 23
+    },
+    {
+      "epoch": 0.01536,
+      "grad_norm": 0.6626828380488425,
+      "learning_rate": 1.4556962025316457e-05,
+      "loss": 0.3536,
+      "step": 24
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 0.7920276455155202,
+      "learning_rate": 1.5189873417721521e-05,
+      "loss": 0.3942,
+      "step": 25
+    },
+    {
+      "epoch": 0.01664,
+      "grad_norm": 0.6737828955016701,
+      "learning_rate": 1.5822784810126583e-05,
+      "loss": 0.3749,
+      "step": 26
+    },
+    {
+      "epoch": 0.01728,
+      "grad_norm": 0.5503753217412305,
+      "learning_rate": 1.6455696202531644e-05,
+      "loss": 0.3718,
+      "step": 27
+    },
+    {
+      "epoch": 0.01792,
+      "grad_norm": 0.7894180890361427,
+      "learning_rate": 1.7088607594936708e-05,
+      "loss": 0.3703,
+      "step": 28
+    },
+    {
+      "epoch": 0.01856,
+      "grad_norm": 0.52887713912012,
+      "learning_rate": 1.7721518987341772e-05,
+      "loss": 0.3576,
+      "step": 29
+    },
+    {
+      "epoch": 0.0192,
+      "grad_norm": 0.5935309461197494,
+      "learning_rate": 1.8354430379746836e-05,
+      "loss": 0.384,
+      "step": 30
+    },
+    {
+      "epoch": 0.01984,
+      "grad_norm": 0.4947385860590809,
+      "learning_rate": 1.89873417721519e-05,
+      "loss": 0.3778,
+      "step": 31
+    },
+    {
+      "epoch": 0.02048,
+      "grad_norm": 0.5605217572196567,
+      "learning_rate": 1.962025316455696e-05,
+      "loss": 0.3838,
+      "step": 32
+    },
+    {
+      "epoch": 0.02112,
+      "grad_norm": 0.6836657396769625,
+      "learning_rate": 2.0253164556962025e-05,
+      "loss": 0.3482,
+      "step": 33
+    },
+    {
+      "epoch": 0.02176,
+      "grad_norm": 0.43798176485984003,
+      "learning_rate": 2.088607594936709e-05,
+      "loss": 0.3612,
+      "step": 34
+    },
+    {
+      "epoch": 0.0224,
+      "grad_norm": 0.4240945316235695,
+      "learning_rate": 2.1518987341772153e-05,
+      "loss": 0.3599,
+      "step": 35
+    },
+    {
+      "epoch": 0.02304,
+      "grad_norm": 0.4290568610982825,
+      "learning_rate": 2.2151898734177217e-05,
+      "loss": 0.3722,
+      "step": 36
+    },
+    {
+      "epoch": 0.02368,
+      "grad_norm": 0.46533348793486895,
+      "learning_rate": 2.278481012658228e-05,
+      "loss": 0.3627,
+      "step": 37
+    },
+    {
+      "epoch": 0.02432,
+      "grad_norm": 0.48639931562775274,
+      "learning_rate": 2.341772151898734e-05,
+      "loss": 0.3582,
+      "step": 38
+    },
+    {
+      "epoch": 0.02496,
+      "grad_norm": 0.4403975390170576,
+      "learning_rate": 2.4050632911392405e-05,
+      "loss": 0.353,
+      "step": 39
+    },
+    {
+      "epoch": 0.0256,
+      "grad_norm": 0.36233863650138515,
+      "learning_rate": 2.468354430379747e-05,
+      "loss": 0.3451,
+      "step": 40
+    },
+    {
+      "epoch": 0.02624,
+      "grad_norm": 0.4175915746669695,
+      "learning_rate": 2.5316455696202533e-05,
+      "loss": 0.3059,
+      "step": 41
+    },
+    {
+      "epoch": 0.02688,
+      "grad_norm": 0.4694739579410676,
+      "learning_rate": 2.5949367088607597e-05,
+      "loss": 0.3332,
+      "step": 42
+    },
+    {
+      "epoch": 0.02752,
+      "grad_norm": 0.3958470601870478,
+      "learning_rate": 2.6582278481012658e-05,
+      "loss": 0.3231,
+      "step": 43
+    },
+    {
+      "epoch": 0.02816,
+      "grad_norm": 0.43940504584870443,
+      "learning_rate": 2.7215189873417722e-05,
+      "loss": 0.3403,
+      "step": 44
+    },
+    {
+      "epoch": 0.0288,
+      "grad_norm": 0.45387568054253924,
+      "learning_rate": 2.7848101265822786e-05,
+      "loss": 0.3522,
+      "step": 45
+    },
+    {
+      "epoch": 0.02944,
+      "grad_norm": 0.47587749147936115,
+      "learning_rate": 2.848101265822785e-05,
+      "loss": 0.344,
+      "step": 46
+    },
+    {
+      "epoch": 0.03008,
+      "grad_norm": 0.39610819412760284,
+      "learning_rate": 2.9113924050632914e-05,
+      "loss": 0.337,
+      "step": 47
+    },
+    {
+      "epoch": 0.03072,
+      "grad_norm": 0.4363309227849157,
+      "learning_rate": 2.9746835443037974e-05,
+      "loss": 0.3483,
+      "step": 48
+    },
+    {
+      "epoch": 0.03136,
+      "grad_norm": 0.40862381905697387,
+      "learning_rate": 3.0379746835443042e-05,
+      "loss": 0.3305,
+      "step": 49
+    },
+    {
+      "epoch": 0.032,
+      "grad_norm": 0.39589715535465136,
+      "learning_rate": 3.10126582278481e-05,
+      "loss": 0.3398,
+      "step": 50
+    },
+    {
+      "epoch": 0.03264,
+      "grad_norm": 0.3989012389437649,
+      "learning_rate": 3.1645569620253167e-05,
+      "loss": 0.3273,
+      "step": 51
+    },
+    {
+      "epoch": 0.03328,
+      "grad_norm": 0.42021137918245616,
+      "learning_rate": 3.227848101265823e-05,
+      "loss": 0.3267,
+      "step": 52
+    },
+    {
+      "epoch": 0.03392,
+      "grad_norm": 0.4047725101285825,
+      "learning_rate": 3.291139240506329e-05,
+      "loss": 0.2908,
+      "step": 53
+    },
+    {
+      "epoch": 0.03456,
+      "grad_norm": 0.4172826076889594,
+      "learning_rate": 3.354430379746836e-05,
+      "loss": 0.3348,
+      "step": 54
+    },
+    {
+      "epoch": 0.0352,
+      "grad_norm": 0.42253047907814617,
+      "learning_rate": 3.4177215189873416e-05,
+      "loss": 0.3292,
+      "step": 55
+    },
+    {
+      "epoch": 0.03584,
+      "grad_norm": 0.42758322444932984,
+      "learning_rate": 3.4810126582278487e-05,
+      "loss": 0.3289,
+      "step": 56
+    },
+    {
+      "epoch": 0.03648,
+      "grad_norm": 0.4053738271148249,
+      "learning_rate": 3.5443037974683544e-05,
+      "loss": 0.3276,
+      "step": 57
+    },
+    {
+      "epoch": 0.03712,
+      "grad_norm": 0.43996040806002545,
+      "learning_rate": 3.607594936708861e-05,
+      "loss": 0.3467,
+      "step": 58
+    },
+    {
+      "epoch": 0.03776,
+      "grad_norm": 0.4254049766589256,
+      "learning_rate": 3.670886075949367e-05,
+      "loss": 0.3338,
+      "step": 59
+    },
+    {
+      "epoch": 0.0384,
+      "grad_norm": 0.43855409848587795,
+      "learning_rate": 3.7341772151898736e-05,
+      "loss": 0.3274,
+      "step": 60
+    },
+    {
+      "epoch": 0.03904,
+      "grad_norm": 0.4365281497934103,
+      "learning_rate": 3.79746835443038e-05,
+      "loss": 0.3347,
+      "step": 61
+    },
+    {
+      "epoch": 0.03968,
+      "grad_norm": 0.40792773541529964,
+      "learning_rate": 3.8607594936708864e-05,
+      "loss": 0.3153,
+      "step": 62
+    },
+    {
+      "epoch": 0.04032,
+      "grad_norm": 0.4494242243007552,
+      "learning_rate": 3.924050632911392e-05,
+      "loss": 0.3203,
+      "step": 63
+    },
+    {
+      "epoch": 0.04096,
+      "grad_norm": 0.4788676847095667,
+      "learning_rate": 3.987341772151899e-05,
+      "loss": 0.3539,
+      "step": 64
+    },
+    {
+      "epoch": 0.0416,
+      "grad_norm": 0.40546778912791753,
+      "learning_rate": 4.050632911392405e-05,
+      "loss": 0.3119,
+      "step": 65
+    },
+    {
+      "epoch": 0.04224,
+      "grad_norm": 0.4700613066573691,
+      "learning_rate": 4.113924050632912e-05,
+      "loss": 0.3464,
+      "step": 66
+    },
+    {
+      "epoch": 0.04288,
+      "grad_norm": 0.4185455474650584,
+      "learning_rate": 4.177215189873418e-05,
+      "loss": 0.3148,
+      "step": 67
+    },
+    {
+      "epoch": 0.04352,
+      "grad_norm": 0.42213917822060293,
+      "learning_rate": 4.240506329113924e-05,
+      "loss": 0.3264,
+      "step": 68
+    },
+    {
+      "epoch": 0.04416,
+      "grad_norm": 0.4110186066327522,
+      "learning_rate": 4.3037974683544305e-05,
+      "loss": 0.2942,
+      "step": 69
+    },
+    {
+      "epoch": 0.0448,
+      "grad_norm": 0.3969987305256835,
+      "learning_rate": 4.367088607594937e-05,
+      "loss": 0.3296,
+      "step": 70
+    },
+    {
+      "epoch": 0.04544,
+      "grad_norm": 0.458340252055146,
+      "learning_rate": 4.430379746835443e-05,
+      "loss": 0.3153,
+      "step": 71
+    },
+    {
+      "epoch": 0.04608,
+      "grad_norm": 0.4460994424040082,
+      "learning_rate": 4.49367088607595e-05,
+      "loss": 0.3237,
+      "step": 72
+    },
+    {
+      "epoch": 0.04672,
+      "grad_norm": 0.48880157137172375,
+      "learning_rate": 4.556962025316456e-05,
+      "loss": 0.3139,
+      "step": 73
+    },
+    {
+      "epoch": 0.04736,
+      "grad_norm": 0.4187304633592487,
+      "learning_rate": 4.6202531645569625e-05,
+      "loss": 0.308,
+      "step": 74
+    },
+    {
+      "epoch": 0.048,
+      "grad_norm": 0.3977527819047683,
+      "learning_rate": 4.683544303797468e-05,
+      "loss": 0.2984,
+      "step": 75
+    },
+    {
+      "epoch": 0.04864,
+      "grad_norm": 0.44371228034060706,
+      "learning_rate": 4.7468354430379746e-05,
+      "loss": 0.3201,
+      "step": 76
+    },
+    {
+      "epoch": 0.04928,
+      "grad_norm": 0.5076283851103198,
+      "learning_rate": 4.810126582278481e-05,
+      "loss": 0.3267,
+      "step": 77
+    },
+    {
+      "epoch": 0.04992,
+      "grad_norm": 0.3927986005300883,
+      "learning_rate": 4.8734177215189874e-05,
+      "loss": 0.2903,
+      "step": 78
+    },
+    {
+      "epoch": 0.05056,
+      "grad_norm": 0.46176244675521244,
+      "learning_rate": 4.936708860759494e-05,
+      "loss": 0.3235,
+      "step": 79
+    },
+    {
+      "epoch": 0.0512,
+      "grad_norm": 0.421631930666922,
+      "learning_rate": 5e-05,
+      "loss": 0.3148,
+      "step": 80
+    },
+    {
+      "epoch": 0.05184,
+      "grad_norm": 0.4061280972034586,
+      "learning_rate": 4.999994390459423e-05,
+      "loss": 0.3244,
+      "step": 81
+    },
+    {
+      "epoch": 0.05248,
+      "grad_norm": 0.41908336407872915,
+      "learning_rate": 4.999977561862863e-05,
+      "loss": 0.3172,
+      "step": 82
+    },
+    {
+      "epoch": 0.05312,
+      "grad_norm": 0.3871831774034262,
+      "learning_rate": 4.999949514285842e-05,
+      "loss": 0.3176,
+      "step": 83
+    },
+    {
+      "epoch": 0.05376,
+      "grad_norm": 0.4569094416746116,
+      "learning_rate": 4.999910247854227e-05,
+      "loss": 0.3302,
+      "step": 84
+    },
+    {
+      "epoch": 0.0544,
+      "grad_norm": 0.47566240228622386,
+      "learning_rate": 4.99985976274423e-05,
+      "loss": 0.3266,
+      "step": 85
+    },
+    {
+      "epoch": 0.05504,
+      "grad_norm": 0.47728079070137763,
+      "learning_rate": 4.9997980591824126e-05,
+      "loss": 0.3072,
+      "step": 86
+    },
+    {
+      "epoch": 0.05568,
+      "grad_norm": 0.39221570157146257,
+      "learning_rate": 4.999725137445674e-05,
+      "loss": 0.3185,
+      "step": 87
+    },
+    {
+      "epoch": 0.05632,
+      "grad_norm": 0.5220845323376405,
+      "learning_rate": 4.999640997861264e-05,
+      "loss": 0.3206,
+      "step": 88
+    },
+    {
+      "epoch": 0.05696,
+      "grad_norm": 0.45138937219188247,
+      "learning_rate": 4.999545640806766e-05,
+      "loss": 0.322,
+      "step": 89
+    },
+    {
+      "epoch": 0.0576,
+      "grad_norm": 0.43231283770656975,
+      "learning_rate": 4.999439066710111e-05,
+      "loss": 0.3413,
+      "step": 90
+    },
+    {
+      "epoch": 0.05824,
+      "grad_norm": 0.5003943169755616,
+      "learning_rate": 4.999321276049563e-05,
+      "loss": 0.3182,
+      "step": 91
+    },
+    {
+      "epoch": 0.05888,
+      "grad_norm": 0.37424460362649736,
+      "learning_rate": 4.9991922693537226e-05,
+      "loss": 0.3065,
+      "step": 92
+    },
+    {
+      "epoch": 0.05952,
+      "grad_norm": 0.44875045802288094,
+      "learning_rate": 4.9990520472015246e-05,
+      "loss": 0.2941,
+      "step": 93
+    },
+    {
+      "epoch": 0.06016,
+      "grad_norm": 0.4021940194343882,
+      "learning_rate": 4.998900610222236e-05,
+      "loss": 0.3047,
+      "step": 94
+    },
+    {
+      "epoch": 0.0608,
+      "grad_norm": 0.3881453179250032,
+      "learning_rate": 4.998737959095449e-05,
+      "loss": 0.2772,
+      "step": 95
+    },
+    {
+      "epoch": 0.06144,
+      "grad_norm": 0.44699341616410987,
+      "learning_rate": 4.998564094551081e-05,
+      "loss": 0.3147,
+      "step": 96
+    },
+    {
+      "epoch": 0.06208,
+      "grad_norm": 0.43480749835897514,
+      "learning_rate": 4.998379017369375e-05,
+      "loss": 0.3106,
+      "step": 97
+    },
+    {
+      "epoch": 0.06272,
+      "grad_norm": 0.46102101762706826,
+      "learning_rate": 4.9981827283808855e-05,
+      "loss": 0.3075,
+      "step": 98
+    },
+    {
+      "epoch": 0.06336,
+      "grad_norm": 0.43677419770737613,
+      "learning_rate": 4.997975228466489e-05,
+      "loss": 0.3346,
+      "step": 99
+    },
+    {
+      "epoch": 0.064,
+      "grad_norm": 0.3990810545801376,
+      "learning_rate": 4.997756518557368e-05,
+      "loss": 0.2792,
+      "step": 100
+    },
+    {
+      "epoch": 0.06464,
+      "grad_norm": 0.37423119672037614,
+      "learning_rate": 4.997526599635011e-05,
+      "loss": 0.2843,
+      "step": 101
+    },
+    {
+      "epoch": 0.06528,
+      "grad_norm": 0.4832295984241697,
+      "learning_rate": 4.997285472731211e-05,
+      "loss": 0.3335,
+      "step": 102
+    },
+    {
+      "epoch": 0.06592,
+      "grad_norm": 0.42554025051835676,
+      "learning_rate": 4.997033138928056e-05,
+      "loss": 0.3134,
+      "step": 103
+    },
+    {
+      "epoch": 0.06656,
+      "grad_norm": 0.4527604445470986,
+      "learning_rate": 4.996769599357927e-05,
+      "loss": 0.3082,
+      "step": 104
+    },
+    {
+      "epoch": 0.0672,
+      "grad_norm": 0.43328941260689235,
+      "learning_rate": 4.996494855203493e-05,
+      "loss": 0.3233,
+      "step": 105
+    },
+    {
+      "epoch": 0.06784,
+      "grad_norm": 0.5549824145577879,
+      "learning_rate": 4.996208907697706e-05,
+      "loss": 0.3203,
+      "step": 106
+    },
+    {
+      "epoch": 0.06848,
+      "grad_norm": 0.5226904312118057,
+      "learning_rate": 4.995911758123791e-05,
+      "loss": 0.2832,
+      "step": 107
+    },
+    {
+      "epoch": 0.06912,
+      "grad_norm": 0.4619131188070299,
+      "learning_rate": 4.995603407815249e-05,
+      "loss": 0.2958,
+      "step": 108
+    },
+    {
+      "epoch": 0.06976,
+      "grad_norm": 0.4694617073433174,
+      "learning_rate": 4.9952838581558406e-05,
+      "loss": 0.3022,
+      "step": 109
+    },
+    {
+      "epoch": 0.0704,
+      "grad_norm": 0.4111272867245987,
+      "learning_rate": 4.994953110579588e-05,
+      "loss": 0.3123,
+      "step": 110
+    },
+    {
+      "epoch": 0.07104,
+      "grad_norm": 0.49830144949207156,
+      "learning_rate": 4.9946111665707645e-05,
+      "loss": 0.3252,
+      "step": 111
+    },
+    {
+      "epoch": 0.07168,
+      "grad_norm": 0.36298788411716454,
+      "learning_rate": 4.99425802766389e-05,
+      "loss": 0.2886,
+      "step": 112
+    },
+    {
+      "epoch": 0.07232,
+      "grad_norm": 0.4553410362166831,
+      "learning_rate": 4.993893695443721e-05,
+      "loss": 0.3014,
+      "step": 113
+    },
+    {
+      "epoch": 0.07296,
+      "grad_norm": 0.3942254183560885,
+      "learning_rate": 4.993518171545247e-05,
+      "loss": 0.2905,
+      "step": 114
+    },
+    {
+      "epoch": 0.0736,
+      "grad_norm": 0.43886036816836405,
+      "learning_rate": 4.993131457653682e-05,
+      "loss": 0.3016,
+      "step": 115
+    },
+    {
+      "epoch": 0.07424,
+      "grad_norm": 0.4199519149927171,
+      "learning_rate": 4.992733555504454e-05,
+      "loss": 0.3172,
+      "step": 116
+    },
+    {
+      "epoch": 0.07488,
+      "grad_norm": 0.4190328574403923,
+      "learning_rate": 4.9923244668832025e-05,
+      "loss": 0.3061,
+      "step": 117
+    },
+    {
+      "epoch": 0.07552,
+      "grad_norm": 0.42026190311885486,
+      "learning_rate": 4.991904193625768e-05,
+      "loss": 0.311,
+      "step": 118
+    },
+    {
+      "epoch": 0.07616,
+      "grad_norm": 0.3651966339148085,
+      "learning_rate": 4.991472737618181e-05,
+      "loss": 0.2732,
+      "step": 119
+    },
+    {
+      "epoch": 0.0768,
+      "grad_norm": 0.40672045467505213,
+      "learning_rate": 4.991030100796659e-05,
+      "loss": 0.3012,
+      "step": 120
+    },
+    {
+      "epoch": 0.07744,
+      "grad_norm": 0.5292094384621548,
+      "learning_rate": 4.9905762851475915e-05,
+      "loss": 0.3151,
+      "step": 121
+    },
+    {
+      "epoch": 0.07808,
+      "grad_norm": 0.39692105389592514,
+      "learning_rate": 4.990111292707537e-05,
+      "loss": 0.3141,
+      "step": 122
+    },
+    {
+      "epoch": 0.07872,
+      "grad_norm": 0.39038544937881386,
+      "learning_rate": 4.989635125563211e-05,
+      "loss": 0.2889,
+      "step": 123
+    },
+    {
+      "epoch": 0.07936,
+      "grad_norm": 0.38869575358852054,
+      "learning_rate": 4.989147785851477e-05,
+      "loss": 0.3083,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.40766962966440357,
+      "learning_rate": 4.988649275759335e-05,
+      "loss": 0.3186,
+      "step": 125
+    },
+    {
+      "epoch": 0.08064,
+      "grad_norm": 0.3818281612303487,
+      "learning_rate": 4.988139597523916e-05,
+      "loss": 0.2913,
+      "step": 126
+    },
+    {
+      "epoch": 0.08128,
+      "grad_norm": 0.4307784596567231,
+      "learning_rate": 4.98761875343247e-05,
+      "loss": 0.2836,
+      "step": 127
+    },
+    {
+      "epoch": 0.08192,
+      "grad_norm": 0.34729585966363985,
+      "learning_rate": 4.987086745822352e-05,
+      "loss": 0.2844,
+      "step": 128
+    },
+    {
+      "epoch": 0.08256,
+      "grad_norm": 0.3896516068942497,
+      "learning_rate": 4.986543577081017e-05,
+      "loss": 0.3214,
+      "step": 129
+    },
+    {
+      "epoch": 0.0832,
+      "grad_norm": 0.3605049005213821,
+      "learning_rate": 4.985989249646007e-05,
+      "loss": 0.3018,
+      "step": 130
+    },
+    {
+      "epoch": 0.08384,
+      "grad_norm": 0.3728422361203703,
+      "learning_rate": 4.985423766004939e-05,
+      "loss": 0.3097,
+      "step": 131
+    },
+    {
+      "epoch": 0.08448,
+      "grad_norm": 0.4536755532298521,
+      "learning_rate": 4.9848471286954965e-05,
+      "loss": 0.3266,
+      "step": 132
+    },
+    {
+      "epoch": 0.08512,
+      "grad_norm": 0.3266252900478582,
+      "learning_rate": 4.984259340305416e-05,
+      "loss": 0.2886,
+      "step": 133
+    },
+    {
+      "epoch": 0.08576,
+      "grad_norm": 0.3884896844083972,
+      "learning_rate": 4.9836604034724753e-05,
+      "loss": 0.2917,
+      "step": 134
+    },
+    {
+      "epoch": 0.0864,
+      "grad_norm": 0.4147698757428519,
+      "learning_rate": 4.9830503208844835e-05,
+      "loss": 0.2927,
+      "step": 135
+    },
+    {
+      "epoch": 0.08704,
+      "grad_norm": 0.35299011989641577,
+      "learning_rate": 4.982429095279266e-05,
+      "loss": 0.2941,
+      "step": 136
+    },
+    {
+      "epoch": 0.08768,
+      "grad_norm": 0.43335162639635905,
+      "learning_rate": 4.981796729444655e-05,
+      "loss": 0.3239,
+      "step": 137
+    },
+    {
+      "epoch": 0.08832,
+      "grad_norm": 0.3940276865887743,
+      "learning_rate": 4.981153226218477e-05,
+      "loss": 0.3076,
+      "step": 138
+    },
+    {
+      "epoch": 0.08896,
+      "grad_norm": 0.3265746011811658,
+      "learning_rate": 4.9804985884885366e-05,
+      "loss": 0.2868,
+      "step": 139
+    },
+    {
+      "epoch": 0.0896,
+      "grad_norm": 0.38117660073360304,
+      "learning_rate": 4.979832819192608e-05,
+      "loss": 0.3121,
+      "step": 140
+    },
+    {
+      "epoch": 0.09024,
+      "grad_norm": 0.33607121469181583,
+      "learning_rate": 4.97915592131842e-05,
+      "loss": 0.2878,
+      "step": 141
+    },
+    {
+      "epoch": 0.09088,
+      "grad_norm": 0.3692306728222237,
+      "learning_rate": 4.97846789790364e-05,
+      "loss": 0.3072,
+      "step": 142
+    },
+    {
+      "epoch": 0.09152,
+      "grad_norm": 0.35801789239892756,
+      "learning_rate": 4.977768752035865e-05,
+      "loss": 0.3115,
+      "step": 143
+    },
+    {
+      "epoch": 0.09216,
+      "grad_norm": 0.3930347756960167,
+      "learning_rate": 4.9770584868526046e-05,
+      "loss": 0.3116,
+      "step": 144
+    },
+    {
+      "epoch": 0.0928,
+      "grad_norm": 0.4285263500774063,
+      "learning_rate": 4.9763371055412685e-05,
+      "loss": 0.3079,
+      "step": 145
+    },
+    {
+      "epoch": 0.09344,
+      "grad_norm": 0.4008078130919119,
+      "learning_rate": 4.9756046113391496e-05,
+      "loss": 0.3289,
+      "step": 146
+    },
+    {
+      "epoch": 0.09408,
+      "grad_norm": 0.38210060332023665,
+      "learning_rate": 4.974861007533413e-05,
+      "loss": 0.2861,
+      "step": 147
+    },
+    {
+      "epoch": 0.09472,
+      "grad_norm": 0.4259001953879997,
+      "learning_rate": 4.97410629746108e-05,
+      "loss": 0.2918,
+      "step": 148
+    },
+    {
+      "epoch": 0.09536,
+      "grad_norm": 0.4076596626408926,
+      "learning_rate": 4.9733404845090116e-05,
+      "loss": 0.2841,
+      "step": 149
+    },
+    {
+      "epoch": 0.096,
+      "grad_norm": 0.40095369405114606,
+      "learning_rate": 4.972563572113894e-05,
+      "loss": 0.2902,
+      "step": 150
+    },
+    {
+      "epoch": 0.09664,
+      "grad_norm": 0.44874206133172484,
+      "learning_rate": 4.971775563762226e-05,
+      "loss": 0.2912,
+      "step": 151
+    },
+    {
+      "epoch": 0.09728,
+      "grad_norm": 0.41878107643356327,
+      "learning_rate": 4.970976462990299e-05,
+      "loss": 0.3106,
+      "step": 152
+    },
+    {
+      "epoch": 0.09792,
+      "grad_norm": 0.3886961745648121,
+      "learning_rate": 4.970166273384183e-05,
+      "loss": 0.2866,
+      "step": 153
+    },
+    {
+      "epoch": 0.09856,
+      "grad_norm": 0.34534258751325136,
+      "learning_rate": 4.969344998579712e-05,
+      "loss": 0.2657,
+      "step": 154
+    },
+    {
+      "epoch": 0.0992,
+      "grad_norm": 0.3755269967559221,
+      "learning_rate": 4.9685126422624644e-05,
+      "loss": 0.3146,
+      "step": 155
+    },
+    {
+      "epoch": 0.09984,
+      "grad_norm": 0.40024019989003823,
+      "learning_rate": 4.96766920816775e-05,
+      "loss": 0.3144,
+      "step": 156
+    },
+    {
+      "epoch": 0.10048,
+      "grad_norm": 0.3528612408949507,
+      "learning_rate": 4.966814700080592e-05,
+      "loss": 0.3175,
+      "step": 157
+    },
+    {
+      "epoch": 0.10112,
+      "grad_norm": 0.37989058888383176,
+      "learning_rate": 4.965949121835707e-05,
+      "loss": 0.3336,
+      "step": 158
+    },
+    {
+      "epoch": 0.10176,
+      "grad_norm": 0.3935853059250065,
+      "learning_rate": 4.965072477317493e-05,
+      "loss": 0.3093,
+      "step": 159
+    },
+    {
+      "epoch": 0.1024,
+      "grad_norm": 0.3481727131458977,
+      "learning_rate": 4.964184770460007e-05,
+      "loss": 0.2928,
+      "step": 160
+    },
+    {
+      "epoch": 0.10304,
+      "grad_norm": 0.407095370695997,
+      "learning_rate": 4.963286005246953e-05,
+      "loss": 0.2819,
+      "step": 161
+    },
+    {
+      "epoch": 0.10368,
+      "grad_norm": 0.36659617800740035,
+      "learning_rate": 4.9623761857116594e-05,
+      "loss": 0.3195,
+      "step": 162
+    },
+    {
+      "epoch": 0.10432,
+      "grad_norm": 0.3734389648125262,
+      "learning_rate": 4.961455315937059e-05,
+      "loss": 0.2812,
+      "step": 163
+    },
+    {
+      "epoch": 0.10496,
+      "grad_norm": 0.3413505309534122,
+      "learning_rate": 4.960523400055679e-05,
+      "loss": 0.2934,
+      "step": 164
+    },
+    {
+      "epoch": 0.1056,
+      "grad_norm": 0.3301887986531996,
+      "learning_rate": 4.959580442249615e-05,
+      "loss": 0.3012,
+      "step": 165
+    },
+    {
+      "epoch": 0.10624,
+      "grad_norm": 0.3123585093078612,
+      "learning_rate": 4.958626446750515e-05,
+      "loss": 0.3202,
+      "step": 166
+    },
+    {
+      "epoch": 0.10688,
+      "grad_norm": 0.35966543460940326,
+      "learning_rate": 4.9576614178395595e-05,
+      "loss": 0.3006,
+      "step": 167
+    },
+    {
+      "epoch": 0.10752,
+      "grad_norm": 0.332623020117398,
+      "learning_rate": 4.956685359847445e-05,
+      "loss": 0.297,
+      "step": 168
+    },
+    {
+      "epoch": 0.10816,
+      "grad_norm": 0.3142621220673422,
+      "learning_rate": 4.95569827715436e-05,
+      "loss": 0.2812,
+      "step": 169
+    },
+    {
+      "epoch": 0.1088,
+      "grad_norm": 0.32468349021820947,
+      "learning_rate": 4.9547001741899693e-05,
+      "loss": 0.2968,
+      "step": 170
+    },
+    {
+      "epoch": 0.10944,
+      "grad_norm": 0.3408361426728115,
+      "learning_rate": 4.953691055433391e-05,
+      "loss": 0.2903,
+      "step": 171
+    },
+    {
+      "epoch": 0.11008,
+      "grad_norm": 0.36227080591435323,
+      "learning_rate": 4.952670925413182e-05,
+      "loss": 0.3035,
+      "step": 172
+    },
+    {
+      "epoch": 0.11072,
+      "grad_norm": 0.33583803873672646,
+      "learning_rate": 4.951639788707307e-05,
+      "loss": 0.2845,
+      "step": 173
+    },
+    {
+      "epoch": 0.11136,
+      "grad_norm": 0.34619921286834143,
+      "learning_rate": 4.950597649943132e-05,
+      "loss": 0.278,
+      "step": 174
+    },
+    {
+      "epoch": 0.112,
+      "grad_norm": 0.3404081978672373,
+      "learning_rate": 4.9495445137973905e-05,
+      "loss": 0.2929,
+      "step": 175
+    },
+    {
+      "epoch": 0.11264,
+      "grad_norm": 0.3310437000635691,
+      "learning_rate": 4.948480384996172e-05,
+      "loss": 0.2872,
+      "step": 176
+    },
+    {
+      "epoch": 0.11328,
+      "grad_norm": 0.35248099294249946,
+      "learning_rate": 4.947405268314894e-05,
+      "loss": 0.3003,
+      "step": 177
+    },
+    {
+      "epoch": 0.11392,
+      "grad_norm": 0.4133726433387471,
+      "learning_rate": 4.9463191685782865e-05,
+      "loss": 0.307,
+      "step": 178
+    },
+    {
+      "epoch": 0.11456,
+      "grad_norm": 0.31342845143199094,
+      "learning_rate": 4.945222090660364e-05,
+      "loss": 0.2846,
+      "step": 179
+    },
+    {
+      "epoch": 0.1152,
+      "grad_norm": 0.3518931188946702,
+      "learning_rate": 4.944114039484411e-05,
+      "loss": 0.3046,
+      "step": 180
+    },
+    {
+      "epoch": 0.11584,
+      "grad_norm": 0.3421314438441075,
+      "learning_rate": 4.942995020022954e-05,
+      "loss": 0.3244,
+      "step": 181
+    },
+    {
+      "epoch": 0.11648,
+      "grad_norm": 0.3267692222072987,
+      "learning_rate": 4.9418650372977384e-05,
+      "loss": 0.2887,
+      "step": 182
+    },
+    {
+      "epoch": 0.11712,
+      "grad_norm": 0.3150731393402504,
+      "learning_rate": 4.940724096379714e-05,
+      "loss": 0.2826,
+      "step": 183
+    },
+    {
+      "epoch": 0.11776,
+      "grad_norm": 0.28931360046418647,
+      "learning_rate": 4.9395722023890035e-05,
+      "loss": 0.2806,
+      "step": 184
+    },
+    {
+      "epoch": 0.1184,
+      "grad_norm": 0.3456286699266885,
+      "learning_rate": 4.938409360494883e-05,
+      "loss": 0.2835,
+      "step": 185
+    },
+    {
+      "epoch": 0.11904,
+      "grad_norm": 0.30464221501484345,
+      "learning_rate": 4.937235575915761e-05,
+      "loss": 0.2917,
+      "step": 186
+    },
+    {
+      "epoch": 0.11968,
+      "grad_norm": 0.3227096120306476,
+      "learning_rate": 4.93605085391915e-05,
+      "loss": 0.2923,
+      "step": 187
+    },
+    {
+      "epoch": 0.12032,
+      "grad_norm": 0.366171085820676,
+      "learning_rate": 4.934855199821647e-05,
+      "loss": 0.2835,
+      "step": 188
+    },
+    {
+      "epoch": 0.12096,
+      "grad_norm": 0.3304740847644211,
+      "learning_rate": 4.93364861898891e-05,
+      "loss": 0.2763,
+      "step": 189
+    },
+    {
+      "epoch": 0.1216,
+      "grad_norm": 0.35804683150654454,
+      "learning_rate": 4.932431116835628e-05,
+      "loss": 0.297,
+      "step": 190
+    },
+    {
+      "epoch": 0.12224,
+      "grad_norm": 0.3280992910010731,
+      "learning_rate": 4.931202698825503e-05,
+      "loss": 0.2814,
+      "step": 191
+    },
+    {
+      "epoch": 0.12288,
+      "grad_norm": 0.29841743502194873,
+      "learning_rate": 4.929963370471226e-05,
+      "loss": 0.2878,
+      "step": 192
+    },
+    {
+      "epoch": 0.12352,
+      "grad_norm": 0.2854325525422512,
+      "learning_rate": 4.928713137334445e-05,
+      "loss": 0.2816,
+      "step": 193
+    },
+    {
+      "epoch": 0.12416,
+      "grad_norm": 0.3265192973580523,
+      "learning_rate": 4.927452005025748e-05,
+      "loss": 0.3019,
+      "step": 194
+    },
+    {
+      "epoch": 0.1248,
+      "grad_norm": 0.31439513957912246,
+      "learning_rate": 4.926179979204633e-05,
+      "loss": 0.3023,
+      "step": 195
+    },
+    {
+      "epoch": 0.12544,
+      "grad_norm": 0.3043558334720137,
+      "learning_rate": 4.9248970655794836e-05,
+      "loss": 0.2921,
+      "step": 196
+    },
+    {
+      "epoch": 0.12608,
+      "grad_norm": 0.3083301719907164,
+      "learning_rate": 4.923603269907545e-05,
+      "loss": 0.2746,
+      "step": 197
+    },
+    {
+      "epoch": 0.12672,
+      "grad_norm": 0.3091270018423747,
+      "learning_rate": 4.922298597994897e-05,
+      "loss": 0.2842,
+      "step": 198
+    },
+    {
+      "epoch": 0.12736,
+      "grad_norm": 0.34382329346733104,
+      "learning_rate": 4.920983055696428e-05,
+      "loss": 0.2937,
+      "step": 199
+    },
+    {
+      "epoch": 0.128,
+      "grad_norm": 0.35936428325700825,
+      "learning_rate": 4.919656648915807e-05,
+      "loss": 0.2847,
+      "step": 200
+    },
+    {
+      "epoch": 0.12864,
+      "grad_norm": 0.3044935865997733,
+      "learning_rate": 4.918319383605462e-05,
+      "loss": 0.2575,
+      "step": 201
+    },
+    {
+      "epoch": 0.12928,
+      "grad_norm": 0.3547660540984471,
+      "learning_rate": 4.916971265766547e-05,
+      "loss": 0.2949,
+      "step": 202
+    },
+    {
+      "epoch": 0.12992,
+      "grad_norm": 0.36659141032868914,
+      "learning_rate": 4.915612301448919e-05,
+      "loss": 0.2949,
+      "step": 203
+    },
+    {
+      "epoch": 0.13056,
+      "grad_norm": 0.3603069106129497,
+      "learning_rate": 4.914242496751111e-05,
+      "loss": 0.2882,
+      "step": 204
+    },
+    {
+      "epoch": 0.1312,
+      "grad_norm": 0.3963859179748674,
+      "learning_rate": 4.912861857820303e-05,
+      "loss": 0.3027,
+      "step": 205
+    },
+    {
+      "epoch": 0.13184,
+      "grad_norm": 0.37181091318644754,
+      "learning_rate": 4.911470390852295e-05,
+      "loss": 0.3017,
+      "step": 206
+    },
+    {
+      "epoch": 0.13248,
+      "grad_norm": 0.31409311742594026,
+      "learning_rate": 4.91006810209148e-05,
+      "loss": 0.2884,
+      "step": 207
+    },
+    {
+      "epoch": 0.13312,
+      "grad_norm": 0.3722479719556711,
+      "learning_rate": 4.9086549978308125e-05,
+      "loss": 0.2991,
+      "step": 208
+    },
+    {
+      "epoch": 0.13376,
+      "grad_norm": 0.3004507702000884,
+      "learning_rate": 4.9072310844117875e-05,
+      "loss": 0.272,
+      "step": 209
+    },
+    {
+      "epoch": 0.1344,
+      "grad_norm": 0.33911903822311734,
+      "learning_rate": 4.9057963682244045e-05,
+      "loss": 0.2908,
+      "step": 210
+    },
+    {
+      "epoch": 0.13504,
+      "grad_norm": 0.34057084273821486,
+      "learning_rate": 4.904350855707142e-05,
+      "loss": 0.2869,
+      "step": 211
+    },
+    {
+      "epoch": 0.13568,
+      "grad_norm": 0.31234744073094967,
+      "learning_rate": 4.9028945533469276e-05,
+      "loss": 0.2653,
+      "step": 212
+    },
+    {
+      "epoch": 0.13632,
+      "grad_norm": 0.3755970970195266,
+      "learning_rate": 4.9014274676791134e-05,
+      "loss": 0.3031,
+      "step": 213
+    },
+    {
+      "epoch": 0.13696,
+      "grad_norm": 0.32331976083334474,
+      "learning_rate": 4.899949605287439e-05,
+      "loss": 0.2801,
+      "step": 214
+    },
+    {
+      "epoch": 0.1376,
+      "grad_norm": 0.3755925580675204,
+      "learning_rate": 4.898460972804009e-05,
+      "loss": 0.3092,
+      "step": 215
+    },
+    {
+      "epoch": 0.13824,
+      "grad_norm": 0.36036624380998206,
+      "learning_rate": 4.896961576909258e-05,
+      "loss": 0.2817,
+      "step": 216
+    },
+    {
+      "epoch": 0.13888,
+      "grad_norm": 0.36082215281448893,
+      "learning_rate": 4.895451424331924e-05,
+      "loss": 0.2721,
+      "step": 217
+    },
+    {
+      "epoch": 0.13952,
+      "grad_norm": 0.34106303407373795,
+      "learning_rate": 4.893930521849016e-05,
+      "loss": 0.2675,
+      "step": 218
+    },
+    {
+      "epoch": 0.14016,
+      "grad_norm": 0.392247984251282,
+      "learning_rate": 4.892398876285785e-05,
+      "loss": 0.2904,
+      "step": 219
+    },
+    {
+      "epoch": 0.1408,
+      "grad_norm": 0.3855496076265292,
+      "learning_rate": 4.890856494515696e-05,
+      "loss": 0.2887,
+      "step": 220
+    },
+    {
+      "epoch": 0.14144,
+      "grad_norm": 0.35740557532357425,
+      "learning_rate": 4.889303383460389e-05,
+      "loss": 0.2814,
+      "step": 221
+    },
+    {
+      "epoch": 0.14208,
+      "grad_norm": 0.36474251567342575,
+      "learning_rate": 4.887739550089656e-05,
+      "loss": 0.2788,
+      "step": 222
+    },
+    {
+      "epoch": 0.14272,
+      "grad_norm": 0.36975719759429415,
+      "learning_rate": 4.886165001421408e-05,
+      "loss": 0.3103,
+      "step": 223
+    },
+    {
+      "epoch": 0.14336,
+      "grad_norm": 0.32122517879682544,
+      "learning_rate": 4.884579744521638e-05,
+      "loss": 0.2846,
+      "step": 224
+    },
+    {
+      "epoch": 0.144,
+      "grad_norm": 0.34843218274712867,
+      "learning_rate": 4.8829837865043984e-05,
+      "loss": 0.2916,
+      "step": 225
+    },
+    {
+      "epoch": 0.14464,
+      "grad_norm": 0.31168966766131073,
+      "learning_rate": 4.881377134531762e-05,
+      "loss": 0.2678,
+      "step": 226
+    },
+    {
+      "epoch": 0.14528,
+      "grad_norm": 0.32795585476788547,
+      "learning_rate": 4.8797597958137914e-05,
+      "loss": 0.2804,
+      "step": 227
+    },
+    {
+      "epoch": 0.14592,
+      "grad_norm": 0.366243634878421,
+      "learning_rate": 4.878131777608509e-05,
+      "loss": 0.2848,
+      "step": 228
+    },
+    {
+      "epoch": 0.14656,
+      "grad_norm": 0.32937437081275867,
+      "learning_rate": 4.876493087221862e-05,
+      "loss": 0.2971,
+      "step": 229
+    },
+    {
+      "epoch": 0.1472,
+      "grad_norm": 0.5263731374338324,
+      "learning_rate": 4.87484373200769e-05,
+      "loss": 0.2902,
+      "step": 230
+    },
+    {
+      "epoch": 0.14784,
+      "grad_norm": 0.3289646826837418,
+      "learning_rate": 4.8731837193676944e-05,
+      "loss": 0.2792,
+      "step": 231
+    },
+    {
+      "epoch": 0.14848,
+      "grad_norm": 0.3388926800791491,
+      "learning_rate": 4.871513056751401e-05,
+      "loss": 0.2995,
+      "step": 232
+    },
+    {
+      "epoch": 0.14912,
+      "grad_norm": 0.3530964158377889,
+      "learning_rate": 4.8698317516561295e-05,
+      "loss": 0.3012,
+      "step": 233
+    },
+    {
+      "epoch": 0.14976,
+      "grad_norm": 0.3661383465989981,
+      "learning_rate": 4.8681398116269596e-05,
+      "loss": 0.2867,
+      "step": 234
+    },
+    {
+      "epoch": 0.1504,
+      "grad_norm": 0.33175219745607853,
+      "learning_rate": 4.8664372442566954e-05,
+      "loss": 0.2935,
+      "step": 235
+    },
+    {
+      "epoch": 0.15104,
+      "grad_norm": 0.33548701697224265,
+      "learning_rate": 4.864724057185834e-05,
+      "loss": 0.2689,
+      "step": 236
+    },
+    {
+      "epoch": 0.15168,
+      "grad_norm": 0.32268615702639075,
+      "learning_rate": 4.86300025810253e-05,
+      "loss": 0.2809,
+      "step": 237
+    },
+    {
+      "epoch": 0.15232,
+      "grad_norm": 0.30024229483347803,
+      "learning_rate": 4.861265854742559e-05,
+      "loss": 0.2744,
+      "step": 238
+    },
+    {
+      "epoch": 0.15296,
+      "grad_norm": 0.34761571238874817,
+      "learning_rate": 4.859520854889287e-05,
+      "loss": 0.2937,
+      "step": 239
+    },
+    {
+      "epoch": 0.1536,
+      "grad_norm": 0.29306794724474977,
+      "learning_rate": 4.857765266373631e-05,
+      "loss": 0.2962,
+      "step": 240
+    },
+    {
+      "epoch": 0.15424,
+      "grad_norm": 0.3293199526625673,
+      "learning_rate": 4.855999097074027e-05,
+      "loss": 0.2938,
+      "step": 241
+    },
+    {
+      "epoch": 0.15488,
+      "grad_norm": 0.30974364555406575,
+      "learning_rate": 4.8542223549163946e-05,
+      "loss": 0.2725,
+      "step": 242
+    },
+    {
+      "epoch": 0.15552,
+      "grad_norm": 0.328770395916147,
+      "learning_rate": 4.8524350478740985e-05,
+      "loss": 0.267,
+      "step": 243
+    },
+    {
+      "epoch": 0.15616,
+      "grad_norm": 0.36960435497430866,
+      "learning_rate": 4.850637183967917e-05,
+      "loss": 0.2852,
+      "step": 244
+    },
+    {
+      "epoch": 0.1568,
+      "grad_norm": 0.3202372900146965,
+      "learning_rate": 4.848828771266002e-05,
+      "loss": 0.2885,
+      "step": 245
+    },
+    {
+      "epoch": 0.15744,
+      "grad_norm": 0.38632291902597493,
+      "learning_rate": 4.847009817883844e-05,
+      "loss": 0.2833,
+      "step": 246
+    },
+    {
+      "epoch": 0.15808,
+      "grad_norm": 0.3039843721273191,
+      "learning_rate": 4.845180331984239e-05,
+      "loss": 0.2724,
+      "step": 247
+    },
+    {
+      "epoch": 0.15872,
+      "grad_norm": 0.37777270335325575,
+      "learning_rate": 4.843340321777246e-05,
+      "loss": 0.2654,
+      "step": 248
+    },
+    {
+      "epoch": 0.15936,
+      "grad_norm": 0.32765169609391825,
+      "learning_rate": 4.8414897955201556e-05,
+      "loss": 0.2624,
+      "step": 249
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.30937134246453424,
+      "learning_rate": 4.8396287615174484e-05,
+      "loss": 0.2588,
+      "step": 250
+    },
+    {
+      "epoch": 0.16064,
+      "grad_norm": 0.3283849158663661,
+      "learning_rate": 4.837757228120762e-05,
+      "loss": 0.2895,
+      "step": 251
+    },
+    {
+      "epoch": 0.16128,
+      "grad_norm": 0.31535959823298476,
+      "learning_rate": 4.835875203728849e-05,
+      "loss": 0.2552,
+      "step": 252
+    },
+    {
+      "epoch": 0.16192,
+      "grad_norm": 0.473186549819399,
+      "learning_rate": 4.8339826967875443e-05,
+      "loss": 0.288,
+      "step": 253
+    },
+    {
+      "epoch": 0.16256,
+      "grad_norm": 0.36640618590983065,
+      "learning_rate": 4.832079715789724e-05,
+      "loss": 0.2991,
+      "step": 254
+    },
+    {
+      "epoch": 0.1632,
+      "grad_norm": 0.40519416561462235,
+      "learning_rate": 4.830166269275267e-05,
+      "loss": 0.2916,
+      "step": 255
+    },
+    {
+      "epoch": 0.16384,
+      "grad_norm": 0.3537388526585215,
+      "learning_rate": 4.828242365831016e-05,
+      "loss": 0.2757,
+      "step": 256
+    },
+    {
+      "epoch": 0.16448,
+      "grad_norm": 0.3817071905765977,
+      "learning_rate": 4.8263080140907455e-05,
+      "loss": 0.2916,
+      "step": 257
+    },
+    {
+      "epoch": 0.16512,
+      "grad_norm": 0.361265701608818,
+      "learning_rate": 4.824363222735114e-05,
+      "loss": 0.2755,
+      "step": 258
+    },
+    {
+      "epoch": 0.16576,
+      "grad_norm": 0.35887429797588116,
+      "learning_rate": 4.82240800049163e-05,
+      "loss": 0.2801,
+      "step": 259
+    },
+    {
+      "epoch": 0.1664,
+      "grad_norm": 0.4048340907079254,
+      "learning_rate": 4.820442356134612e-05,
+      "loss": 0.2716,
+      "step": 260
+    },
+    {
+      "epoch": 0.16704,
+      "grad_norm": 0.3554137106703619,
+      "learning_rate": 4.8184662984851505e-05,
+      "loss": 0.3015,
+      "step": 261
+    },
+    {
+      "epoch": 0.16768,
+      "grad_norm": 0.3540300438998987,
+      "learning_rate": 4.816479836411065e-05,
+      "loss": 0.2833,
+      "step": 262
+    },
+    {
+      "epoch": 0.16832,
+      "grad_norm": 0.4227691893247339,
+      "learning_rate": 4.8144829788268674e-05,
+      "loss": 0.2958,
+      "step": 263
+    },
+    {
+      "epoch": 0.16896,
+      "grad_norm": 0.35085282951596225,
+      "learning_rate": 4.8124757346937214e-05,
+      "loss": 0.2633,
+      "step": 264
+    },
+    {
+      "epoch": 0.1696,
+      "grad_norm": 0.3304398208876773,
+      "learning_rate": 4.810458113019401e-05,
+      "loss": 0.275,
+      "step": 265
+    },
+    {
+      "epoch": 0.17024,
+      "grad_norm": 0.32563241609865123,
+      "learning_rate": 4.8084301228582485e-05,
+      "loss": 0.2576,
+      "step": 266
+    },
+    {
+      "epoch": 0.17088,
+      "grad_norm": 0.38836797469119294,
+      "learning_rate": 4.8063917733111406e-05,
+      "loss": 0.3099,
+      "step": 267
+    },
+    {
+      "epoch": 0.17152,
+      "grad_norm": 0.3217274402475735,
+      "learning_rate": 4.8043430735254404e-05,
+      "loss": 0.2813,
+      "step": 268
+    },
+    {
+      "epoch": 0.17216,
+      "grad_norm": 0.36813790588683976,
+      "learning_rate": 4.8022840326949595e-05,
+      "loss": 0.2804,
+      "step": 269
+    },
+    {
+      "epoch": 0.1728,
+      "grad_norm": 0.37938400264260286,
+      "learning_rate": 4.800214660059916e-05,
+      "loss": 0.2805,
+      "step": 270
+    },
+    {
+      "epoch": 0.17344,
+      "grad_norm": 0.3415094791341939,
+      "learning_rate": 4.798134964906894e-05,
+      "loss": 0.2875,
+      "step": 271
+    },
+    {
+      "epoch": 0.17408,
+      "grad_norm": 0.37458115660567415,
+      "learning_rate": 4.7960449565688e-05,
+      "loss": 0.2635,
+      "step": 272
+    },
+    {
+      "epoch": 0.17472,
+      "grad_norm": 0.3250425989019864,
+      "learning_rate": 4.793944644424826e-05,
+      "loss": 0.2855,
+      "step": 273
+    },
+    {
+      "epoch": 0.17536,
+      "grad_norm": 0.3766228590098436,
+      "learning_rate": 4.7918340379003985e-05,
+      "loss": 0.2875,
+      "step": 274
+    },
+    {
+      "epoch": 0.176,
+      "grad_norm": 0.32502593608424385,
+      "learning_rate": 4.7897131464671444e-05,
+      "loss": 0.3004,
+      "step": 275
+    },
+    {
+      "epoch": 0.17664,
+      "grad_norm": 0.36374996921029645,
+      "learning_rate": 4.787581979642845e-05,
+      "loss": 0.2554,
+      "step": 276
+    },
+    {
+      "epoch": 0.17728,
+      "grad_norm": 0.3271666456259006,
+      "learning_rate": 4.7854405469913944e-05,
+      "loss": 0.274,
+      "step": 277
+    },
+    {
+      "epoch": 0.17792,
+      "grad_norm": 0.3208313062202937,
+      "learning_rate": 4.7832888581227545e-05,
+      "loss": 0.2765,
+      "step": 278
+    },
+    {
+      "epoch": 0.17856,
+      "grad_norm": 0.33957962251078677,
+      "learning_rate": 4.7811269226929135e-05,
+      "loss": 0.2801,
+      "step": 279
+    },
+    {
+      "epoch": 0.1792,
+      "grad_norm": 0.3491584815817293,
+      "learning_rate": 4.7789547504038443e-05,
+      "loss": 0.2692,
+      "step": 280
+    },
+    {
+      "epoch": 0.17984,
+      "grad_norm": 0.34068986400037704,
+      "learning_rate": 4.776772351003458e-05,
+      "loss": 0.281,
+      "step": 281
+    },
+    {
+      "epoch": 0.18048,
+      "grad_norm": 0.32860891352594357,
+      "learning_rate": 4.774579734285559e-05,
+      "loss": 0.2877,
+      "step": 282
+    },
+    {
+      "epoch": 0.18112,
+      "grad_norm": 0.32161948162561343,
+      "learning_rate": 4.772376910089808e-05,
+      "loss": 0.2791,
+      "step": 283
+    },
+    {
+      "epoch": 0.18176,
+      "grad_norm": 0.29944358983259645,
+      "learning_rate": 4.770163888301668e-05,
+      "loss": 0.2706,
+      "step": 284
+    },
+    {
+      "epoch": 0.1824,
+      "grad_norm": 0.33259975876524245,
+      "learning_rate": 4.7679406788523685e-05,
+      "loss": 0.2779,
+      "step": 285
+    },
+    {
+      "epoch": 0.18304,
+      "grad_norm": 0.3192766834044811,
+      "learning_rate": 4.765707291718856e-05,
+      "loss": 0.2754,
+      "step": 286
+    },
+    {
+      "epoch": 0.18368,
+      "grad_norm": 0.3333201828315899,
+      "learning_rate": 4.7634637369237526e-05,
+      "loss": 0.2776,
+      "step": 287
+    },
+    {
+      "epoch": 0.18432,
+      "grad_norm": 0.3504320913830548,
+      "learning_rate": 4.7612100245353045e-05,
+      "loss": 0.2874,
+      "step": 288
+    },
+    {
+      "epoch": 0.18496,
+      "grad_norm": 0.34297542367165534,
+      "learning_rate": 4.758946164667347e-05,
+      "loss": 0.2915,
+      "step": 289
+    },
+    {
+      "epoch": 0.1856,
+      "grad_norm": 0.37229747511831845,
+      "learning_rate": 4.7566721674792515e-05,
+      "loss": 0.2782,
+      "step": 290
+    },
+    {
+      "epoch": 0.18624,
+      "grad_norm": 0.3317199247294434,
+      "learning_rate": 4.7543880431758805e-05,
+      "loss": 0.2782,
+      "step": 291
+    },
+    {
+      "epoch": 0.18688,
+      "grad_norm": 0.3375135597374568,
+      "learning_rate": 4.752093802007543e-05,
+      "loss": 0.3017,
+      "step": 292
+    },
+    {
+      "epoch": 0.18752,
+      "grad_norm": 0.3006592553400984,
+      "learning_rate": 4.7497894542699534e-05,
+      "loss": 0.2621,
+      "step": 293
+    },
+    {
+      "epoch": 0.18816,
+      "grad_norm": 0.3457799673936055,
+      "learning_rate": 4.747475010304175e-05,
+      "loss": 0.2814,
+      "step": 294
+    },
+    {
+      "epoch": 0.1888,
+      "grad_norm": 0.30139639958651177,
+      "learning_rate": 4.745150480496583e-05,
+      "loss": 0.2963,
+      "step": 295
+    },
+    {
+      "epoch": 0.18944,
+      "grad_norm": 0.3165095553552411,
+      "learning_rate": 4.742815875278811e-05,
+      "loss": 0.295,
+      "step": 296
+    },
+    {
+      "epoch": 0.19008,
+      "grad_norm": 0.49950391816814954,
+      "learning_rate": 4.7404712051277115e-05,
+      "loss": 0.2564,
+      "step": 297
+    },
+    {
+      "epoch": 0.19072,
+      "grad_norm": 0.4675033045283917,
+      "learning_rate": 4.738116480565301e-05,
+      "loss": 0.2955,
+      "step": 298
+    },
+    {
+      "epoch": 0.19136,
+      "grad_norm": 0.320627703534577,
+      "learning_rate": 4.7357517121587183e-05,
+      "loss": 0.2793,
+      "step": 299
+    },
+    {
+      "epoch": 0.192,
+      "grad_norm": 0.30638839622164654,
+      "learning_rate": 4.7333769105201735e-05,
+      "loss": 0.2433,
+      "step": 300
+    },
+    {
+      "epoch": 0.19264,
+      "grad_norm": 0.33993107017240015,
+      "learning_rate": 4.730992086306906e-05,
+      "loss": 0.2839,
+      "step": 301
+    },
+    {
+      "epoch": 0.19328,
+      "grad_norm": 0.31266754895882487,
+      "learning_rate": 4.728597250221129e-05,
+      "loss": 0.2838,
+      "step": 302
+    },
+    {
+      "epoch": 0.19392,
+      "grad_norm": 0.3333824096931821,
+      "learning_rate": 4.726192413009986e-05,
+      "loss": 0.2956,
+      "step": 303
+    },
+    {
+      "epoch": 0.19456,
+      "grad_norm": 0.31668134967973227,
+      "learning_rate": 4.723777585465503e-05,
+      "loss": 0.2943,
+      "step": 304
+    },
+    {
+      "epoch": 0.1952,
+      "grad_norm": 0.3516833949768645,
+      "learning_rate": 4.7213527784245395e-05,
+      "loss": 0.2778,
+      "step": 305
+    },
+    {
+      "epoch": 0.19584,
+      "grad_norm": 0.3221013371360776,
+      "learning_rate": 4.718918002768738e-05,
+      "loss": 0.2775,
+      "step": 306
+    },
+    {
+      "epoch": 0.19648,
+      "grad_norm": 0.2912869296710327,
+      "learning_rate": 4.716473269424476e-05,
+      "loss": 0.2838,
+      "step": 307
+    },
+    {
+      "epoch": 0.19712,
+      "grad_norm": 0.3063238267835587,
+      "learning_rate": 4.714018589362818e-05,
+      "loss": 0.3116,
+      "step": 308
+    },
+    {
+      "epoch": 0.19776,
+      "grad_norm": 0.2678309458069726,
+      "learning_rate": 4.7115539735994666e-05,
+      "loss": 0.2741,
+      "step": 309
+    },
+    {
+      "epoch": 0.1984,
+      "grad_norm": 0.3203497385070914,
+      "learning_rate": 4.709079433194712e-05,
+      "loss": 0.278,
+      "step": 310
+    },
+    {
+      "epoch": 0.19904,
+      "grad_norm": 0.2886560865479684,
+      "learning_rate": 4.7065949792533804e-05,
+      "loss": 0.2579,
+      "step": 311
+    },
+    {
+      "epoch": 0.19968,
+      "grad_norm": 0.34979254223314304,
+      "learning_rate": 4.70410062292479e-05,
+      "loss": 0.2943,
+      "step": 312
+    },
+    {
+      "epoch": 0.20032,
+      "grad_norm": 0.30129370691362123,
+      "learning_rate": 4.701596375402694e-05,
+      "loss": 0.2935,
+      "step": 313
+    },
+    {
+      "epoch": 0.20096,
+      "grad_norm": 0.3028194799255164,
+      "learning_rate": 4.699082247925234e-05,
+      "loss": 0.3014,
+      "step": 314
+    },
+    {
+      "epoch": 0.2016,
+      "grad_norm": 0.3245596740386509,
+      "learning_rate": 4.6965582517748923e-05,
+      "loss": 0.2938,
+      "step": 315
+    },
+    {
+      "epoch": 0.20224,
+      "grad_norm": 0.35740876844794395,
+      "learning_rate": 4.694024398278434e-05,
+      "loss": 0.2888,
+      "step": 316
+    },
+    {
+      "epoch": 0.20288,
+      "grad_norm": 0.3378708759886966,
+      "learning_rate": 4.691480698806864e-05,
+      "loss": 0.3004,
+      "step": 317
+    },
+    {
+      "epoch": 0.20352,
+      "grad_norm": 0.2824072245359557,
+      "learning_rate": 4.6889271647753694e-05,
+      "loss": 0.2641,
+      "step": 318
+    },
+    {
+      "epoch": 0.20416,
+      "grad_norm": 0.3174107262885623,
+      "learning_rate": 4.686363807643272e-05,
+      "loss": 0.2558,
+      "step": 319
+    },
+    {
+      "epoch": 0.2048,
+      "grad_norm": 0.3176431256834232,
+      "learning_rate": 4.683790638913979e-05,
+      "loss": 0.301,
+      "step": 320
+    },
+    {
+      "epoch": 0.20544,
+      "grad_norm": 0.3001231365631309,
+      "learning_rate": 4.681207670134923e-05,
+      "loss": 0.285,
+      "step": 321
+    },
+    {
+      "epoch": 0.20608,
+      "grad_norm": 0.2901884788933691,
+      "learning_rate": 4.6786149128975196e-05,
+      "loss": 0.2787,
+      "step": 322
+    },
+    {
+      "epoch": 0.20672,
+      "grad_norm": 0.2954537995473784,
+      "learning_rate": 4.6760123788371114e-05,
+      "loss": 0.2827,
+      "step": 323
+    },
+    {
+      "epoch": 0.20736,
+      "grad_norm": 0.3101653694159506,
+      "learning_rate": 4.6734000796329125e-05,
+      "loss": 0.2848,
+      "step": 324
+    },
+    {
+      "epoch": 0.208,
+      "grad_norm": 0.2874105687683901,
+      "learning_rate": 4.670778027007964e-05,
+      "loss": 0.2973,
+      "step": 325
+    },
+    {
+      "epoch": 0.20864,
+      "grad_norm": 0.3620198319608371,
+      "learning_rate": 4.6681462327290725e-05,
+      "loss": 0.2675,
+      "step": 326
+    },
+    {
+      "epoch": 0.20928,
+      "grad_norm": 0.3162114156351334,
+      "learning_rate": 4.6655047086067636e-05,
+      "loss": 0.2788,
+      "step": 327
+    },
+    {
+      "epoch": 0.20992,
+      "grad_norm": 0.30972198988187555,
+      "learning_rate": 4.662853466495229e-05,
+      "loss": 0.2714,
+      "step": 328
+    },
+    {
+      "epoch": 0.21056,
+      "grad_norm": 0.2972071351365883,
+      "learning_rate": 4.660192518292266e-05,
+      "loss": 0.2704,
+      "step": 329
+    },
+    {
+      "epoch": 0.2112,
+      "grad_norm": 0.4583399375330476,
+      "learning_rate": 4.6575218759392336e-05,
+      "loss": 0.2706,
+      "step": 330
+    },
+    {
+      "epoch": 0.21184,
+      "grad_norm": 0.29570973103502374,
+      "learning_rate": 4.654841551420993e-05,
+      "loss": 0.2591,
+      "step": 331
+    },
+    {
+      "epoch": 0.21248,
+      "grad_norm": 0.30593116338066634,
+      "learning_rate": 4.6521515567658557e-05,
+      "loss": 0.268,
+      "step": 332
+    },
+    {
+      "epoch": 0.21312,
+      "grad_norm": 0.3260999005002705,
+      "learning_rate": 4.6494519040455275e-05,
+      "loss": 0.2799,
+      "step": 333
+    },
+    {
+      "epoch": 0.21376,
+      "grad_norm": 0.2968856649158912,
+      "learning_rate": 4.6467426053750596e-05,
+      "loss": 0.2642,
+      "step": 334
+    },
+    {
+      "epoch": 0.2144,
+      "grad_norm": 0.29642619891446786,
+      "learning_rate": 4.644023672912788e-05,
+      "loss": 0.2777,
+      "step": 335
+    },
+    {
+      "epoch": 0.21504,
+      "grad_norm": 0.30350566844855414,
+      "learning_rate": 4.641295118860281e-05,
+      "loss": 0.2879,
+      "step": 336
+    },
+    {
+      "epoch": 0.21568,
+      "grad_norm": 0.26464623569777385,
+      "learning_rate": 4.638556955462289e-05,
+      "loss": 0.2608,
+      "step": 337
+    },
+    {
+      "epoch": 0.21632,
+      "grad_norm": 0.331526482181387,
+      "learning_rate": 4.635809195006681e-05,
+      "loss": 0.2915,
+      "step": 338
+    },
+    {
+      "epoch": 0.21696,
+      "grad_norm": 0.31249619783308275,
+      "learning_rate": 4.633051849824396e-05,
+      "loss": 0.2587,
+      "step": 339
+    },
+    {
+      "epoch": 0.2176,
+      "grad_norm": 0.3342213823905452,
+      "learning_rate": 4.630284932289386e-05,
+      "loss": 0.2861,
+      "step": 340
+    },
+    {
+      "epoch": 0.21824,
+      "grad_norm": 0.3443174294305122,
+      "learning_rate": 4.6275084548185605e-05,
+      "loss": 0.2951,
+      "step": 341
+    },
+    {
+      "epoch": 0.21888,
+      "grad_norm": 0.32706429001657567,
+      "learning_rate": 4.6247224298717294e-05,
+      "loss": 0.2555,
+      "step": 342
+    },
+    {
+      "epoch": 0.21952,
+      "grad_norm": 0.3059927880992944,
+      "learning_rate": 4.621926869951549e-05,
+      "loss": 0.246,
+      "step": 343
+    },
+    {
+      "epoch": 0.22016,
+      "grad_norm": 0.3825724352839679,
+      "learning_rate": 4.619121787603464e-05,
+      "loss": 0.2768,
+      "step": 344
+    },
+    {
+      "epoch": 0.2208,
+      "grad_norm": 0.31461054559886714,
+      "learning_rate": 4.6163071954156546e-05,
+      "loss": 0.2765,
+      "step": 345
+    },
+    {
+      "epoch": 0.22144,
+      "grad_norm": 0.30279471004384817,
+      "learning_rate": 4.6134831060189743e-05,
+      "loss": 0.283,
+      "step": 346
+    },
+    {
+      "epoch": 0.22208,
+      "grad_norm": 0.3131415453093515,
+      "learning_rate": 4.610649532086899e-05,
+      "loss": 0.2738,
+      "step": 347
+    },
+    {
+      "epoch": 0.22272,
+      "grad_norm": 0.34626214360910446,
+      "learning_rate": 4.607806486335468e-05,
+      "loss": 0.2899,
+      "step": 348
+    },
+    {
+      "epoch": 0.22336,
+      "grad_norm": 0.3000462942424186,
+      "learning_rate": 4.604953981523225e-05,
+      "loss": 0.288,
+      "step": 349
+    },
+    {
+      "epoch": 0.224,
+      "grad_norm": 0.3076604165950532,
+      "learning_rate": 4.602092030451162e-05,
+      "loss": 0.2707,
+      "step": 350
+    },
+    {
+      "epoch": 0.22464,
+      "grad_norm": 0.3224147646178712,
+      "learning_rate": 4.599220645962665e-05,
+      "loss": 0.2867,
+      "step": 351
+    },
+    {
+      "epoch": 0.22528,
+      "grad_norm": 0.29505276608299713,
+      "learning_rate": 4.596339840943452e-05,
+      "loss": 0.2688,
+      "step": 352
+    },
+    {
+      "epoch": 0.22592,
+      "grad_norm": 0.2820029386723236,
+      "learning_rate": 4.5934496283215164e-05,
+      "loss": 0.2675,
+      "step": 353
+    },
+    {
+      "epoch": 0.22656,
+      "grad_norm": 0.2937811078323571,
+      "learning_rate": 4.590550021067072e-05,
+      "loss": 0.2586,
+      "step": 354
+    },
+    {
+      "epoch": 0.2272,
+      "grad_norm": 0.3089230579758364,
+      "learning_rate": 4.587641032192488e-05,
+      "loss": 0.2678,
+      "step": 355
+    },
+    {
+      "epoch": 0.22784,
+      "grad_norm": 0.3141525256276034,
+      "learning_rate": 4.58472267475224e-05,
+      "loss": 0.2828,
+      "step": 356
+    },
+    {
+      "epoch": 0.22848,
+      "grad_norm": 0.3221425300292192,
+      "learning_rate": 4.5817949618428405e-05,
+      "loss": 0.2684,
+      "step": 357
+    },
+    {
+      "epoch": 0.22912,
+      "grad_norm": 0.3081385245871579,
+      "learning_rate": 4.578857906602791e-05,
+      "loss": 0.2686,
+      "step": 358
+    },
+    {
+      "epoch": 0.22976,
+      "grad_norm": 0.29474067735994697,
+      "learning_rate": 4.575911522212516e-05,
+      "loss": 0.2853,
+      "step": 359
+    },
+    {
+      "epoch": 0.2304,
+      "grad_norm": 0.26872460388574704,
+      "learning_rate": 4.572955821894305e-05,
+      "loss": 0.271,
+      "step": 360
+    },
+    {
+      "epoch": 0.23104,
+      "grad_norm": 0.2795986689368969,
+      "learning_rate": 4.5699908189122556e-05,
+      "loss": 0.237,
+      "step": 361
+    },
+    {
+      "epoch": 0.23168,
+      "grad_norm": 0.3164595478830514,
+      "learning_rate": 4.5670165265722096e-05,
+      "loss": 0.2575,
+      "step": 362
+    },
+    {
+      "epoch": 0.23232,
+      "grad_norm": 0.3195893610809849,
+      "learning_rate": 4.5640329582217004e-05,
+      "loss": 0.2857,
+      "step": 363
+    },
+    {
+      "epoch": 0.23296,
+      "grad_norm": 0.31680454787494927,
+      "learning_rate": 4.5610401272498836e-05,
+      "loss": 0.2554,
+      "step": 364
+    },
+    {
+      "epoch": 0.2336,
+      "grad_norm": 0.28146209486053303,
+      "learning_rate": 4.558038047087486e-05,
+      "loss": 0.2643,
+      "step": 365
+    },
+    {
+      "epoch": 0.23424,
+      "grad_norm": 0.30175920777877385,
+      "learning_rate": 4.555026731206741e-05,
+      "loss": 0.2492,
+      "step": 366
+    },
+    {
+      "epoch": 0.23488,
+      "grad_norm": 0.29436533992649466,
+      "learning_rate": 4.5520061931213255e-05,
+      "loss": 0.2793,
+      "step": 367
+    },
+    {
+      "epoch": 0.23552,
+      "grad_norm": 0.2829560308292762,
+      "learning_rate": 4.548976446386305e-05,
+      "loss": 0.274,
+      "step": 368
+    },
+    {
+      "epoch": 0.23616,
+      "grad_norm": 0.29408627346856103,
+      "learning_rate": 4.545937504598069e-05,
+      "loss": 0.2728,
+      "step": 369
+    },
+    {
+      "epoch": 0.2368,
+      "grad_norm": 0.2889978012340055,
+      "learning_rate": 4.5428893813942727e-05,
+      "loss": 0.2692,
+      "step": 370
+    },
+    {
+      "epoch": 0.23744,
+      "grad_norm": 0.27148243923702725,
+      "learning_rate": 4.539832090453771e-05,
+      "loss": 0.2655,
+      "step": 371
+    },
+    {
+      "epoch": 0.23808,
+      "grad_norm": 0.31050739584394577,
+      "learning_rate": 4.536765645496564e-05,
+      "loss": 0.262,
+      "step": 372
+    },
+    {
+      "epoch": 0.23872,
+      "grad_norm": 0.2996079504786343,
+      "learning_rate": 4.533690060283727e-05,
+      "loss": 0.2662,
+      "step": 373
+    },
+    {
+      "epoch": 0.23936,
+      "grad_norm": 0.2800563400226751,
+      "learning_rate": 4.530605348617357e-05,
+      "loss": 0.275,
+      "step": 374
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.31562058411792526,
+      "learning_rate": 4.527511524340508e-05,
+      "loss": 0.2656,
+      "step": 375
+    },
+    {
+      "epoch": 0.24064,
+      "grad_norm": 0.3090111413770133,
+      "learning_rate": 4.5244086013371246e-05,
+      "loss": 0.2646,
+      "step": 376
+    },
+    {
+      "epoch": 0.24128,
+      "grad_norm": 0.30489397336811647,
+      "learning_rate": 4.521296593531985e-05,
+      "loss": 0.2624,
+      "step": 377
+    },
+    {
+      "epoch": 0.24192,
+      "grad_norm": 0.29221796413694495,
+      "learning_rate": 4.518175514890636e-05,
+      "loss": 0.2656,
+      "step": 378
+    },
+    {
+      "epoch": 0.24256,
+      "grad_norm": 0.33233722630021206,
+      "learning_rate": 4.515045379419332e-05,
+      "loss": 0.2803,
+      "step": 379
+    },
+    {
+      "epoch": 0.2432,
+      "grad_norm": 0.2999774916809297,
+      "learning_rate": 4.5119062011649714e-05,
+      "loss": 0.275,
+      "step": 380
+    },
+    {
+      "epoch": 0.24384,
+      "grad_norm": 0.3355386902700075,
+      "learning_rate": 4.508757994215031e-05,
+      "loss": 0.2807,
+      "step": 381
+    },
+    {
+      "epoch": 0.24448,
+      "grad_norm": 0.35909608718324937,
+      "learning_rate": 4.505600772697507e-05,
+      "loss": 0.2789,
+      "step": 382
+    },
+    {
+      "epoch": 0.24512,
+      "grad_norm": 0.3388616226576712,
+      "learning_rate": 4.5024345507808495e-05,
+      "loss": 0.2665,
+      "step": 383
+    },
+    {
+      "epoch": 0.24576,
+      "grad_norm": 0.3402302385955381,
+      "learning_rate": 4.499259342673898e-05,
+      "loss": 0.2766,
+      "step": 384
+    },
+    {
+      "epoch": 0.2464,
+      "grad_norm": 0.3134053278907894,
+      "learning_rate": 4.4960751626258205e-05,
+      "loss": 0.2637,
+      "step": 385
+    },
+    {
+      "epoch": 0.24704,
+      "grad_norm": 0.34108714934521905,
+      "learning_rate": 4.492882024926046e-05,
+      "loss": 0.2732,
+      "step": 386
+    },
+    {
+      "epoch": 0.24768,
+      "grad_norm": 0.322628980519524,
+      "learning_rate": 4.489679943904203e-05,
+      "loss": 0.2942,
+      "step": 387
+    },
+    {
+      "epoch": 0.24832,
+      "grad_norm": 0.2872634294452658,
+      "learning_rate": 4.486468933930055e-05,
+      "loss": 0.2809,
+      "step": 388
+    },
+    {
+      "epoch": 0.24896,
+      "grad_norm": 0.3112032329339079,
+      "learning_rate": 4.483249009413433e-05,
+      "loss": 0.2596,
+      "step": 389
+    },
+    {
+      "epoch": 0.2496,
+      "grad_norm": 0.3091772617757821,
+      "learning_rate": 4.4800201848041765e-05,
+      "loss": 0.2895,
+      "step": 390
+    },
+    {
+      "epoch": 0.25024,
+      "grad_norm": 0.3162533728810715,
+      "learning_rate": 4.476782474592062e-05,
+      "loss": 0.2629,
+      "step": 391
+    },
+    {
+      "epoch": 0.25088,
+      "grad_norm": 0.29776738506438605,
+      "learning_rate": 4.473535893306744e-05,
+      "loss": 0.269,
+      "step": 392
+    },
+    {
+      "epoch": 0.25152,
+      "grad_norm": 0.29061925034244496,
+      "learning_rate": 4.470280455517687e-05,
+      "loss": 0.2621,
+      "step": 393
+    },
+    {
+      "epoch": 0.25216,
+      "grad_norm": 0.32125404165323423,
+      "learning_rate": 4.4670161758340966e-05,
+      "loss": 0.2739,
+      "step": 394
+    },
+    {
+      "epoch": 0.2528,
+      "grad_norm": 0.32765191926742254,
+      "learning_rate": 4.4637430689048634e-05,
+      "loss": 0.2603,
+      "step": 395
+    },
+    {
+      "epoch": 0.25344,
+      "grad_norm": 0.3248426091429505,
+      "learning_rate": 4.460461149418486e-05,
+      "loss": 0.2562,
+      "step": 396
+    },
+    {
+      "epoch": 0.25408,
+      "grad_norm": 0.3293450138223353,
+      "learning_rate": 4.4571704321030125e-05,
+      "loss": 0.2909,
+      "step": 397
+    },
+    {
+      "epoch": 0.25472,
+      "grad_norm": 0.3268103816628839,
+      "learning_rate": 4.4538709317259754e-05,
+      "loss": 0.2476,
+      "step": 398
+    },
+    {
+      "epoch": 0.25536,
+      "grad_norm": 0.28789034131409996,
+      "learning_rate": 4.4505626630943174e-05,
+      "loss": 0.2615,
+      "step": 399
+    },
+    {
+      "epoch": 0.256,
+      "grad_norm": 0.3326450624021029,
+      "learning_rate": 4.447245641054334e-05,
+      "loss": 0.2741,
+      "step": 400
+    },
+    {
+      "epoch": 0.25664,
+      "grad_norm": 0.2849399630023341,
+      "learning_rate": 4.443919880491598e-05,
+      "loss": 0.2526,
+      "step": 401
+    },
+    {
+      "epoch": 0.25728,
+      "grad_norm": 0.28894563785283384,
+      "learning_rate": 4.440585396330904e-05,
+      "loss": 0.2688,
+      "step": 402
+    },
+    {
+      "epoch": 0.25792,
+      "grad_norm": 0.3197497488847442,
+      "learning_rate": 4.437242203536189e-05,
+      "loss": 0.2728,
+      "step": 403
+    },
+    {
+      "epoch": 0.25856,
+      "grad_norm": 0.3037447602492967,
+      "learning_rate": 4.433890317110475e-05,
+      "loss": 0.2839,
+      "step": 404
+    },
+    {
+      "epoch": 0.2592,
+      "grad_norm": 0.29041308751589046,
+      "learning_rate": 4.430529752095795e-05,
+      "loss": 0.2712,
+      "step": 405
+    },
+    {
+      "epoch": 0.25984,
+      "grad_norm": 0.29621099818039576,
+      "learning_rate": 4.427160523573131e-05,
+      "loss": 0.2801,
+      "step": 406
+    },
+    {
+      "epoch": 0.26048,
+      "grad_norm": 0.29524960348391543,
+      "learning_rate": 4.4237826466623406e-05,
+      "loss": 0.2871,
+      "step": 407
+    },
+    {
+      "epoch": 0.26112,
+      "grad_norm": 0.2966251537463695,
+      "learning_rate": 4.420396136522095e-05,
+      "loss": 0.2728,
+      "step": 408
+    },
+    {
+      "epoch": 0.26176,
+      "grad_norm": 0.2875766541626561,
+      "learning_rate": 4.417001008349807e-05,
+      "loss": 0.2585,
+      "step": 409
+    },
+    {
+      "epoch": 0.2624,
+      "grad_norm": 0.3113316229409783,
+      "learning_rate": 4.413597277381564e-05,
+      "loss": 0.2707,
+      "step": 410
+    },
+    {
+      "epoch": 0.26304,
+      "grad_norm": 0.2811939335957826,
+      "learning_rate": 4.410184958892059e-05,
+      "loss": 0.2655,
+      "step": 411
+    },
+    {
+      "epoch": 0.26368,
+      "grad_norm": 0.29793389103823087,
+      "learning_rate": 4.4067640681945244e-05,
+      "loss": 0.2637,
+      "step": 412
+    },
+    {
+      "epoch": 0.26432,
+      "grad_norm": 0.3448593520224143,
+      "learning_rate": 4.403334620640659e-05,
+      "loss": 0.2603,
+      "step": 413
+    },
+    {
+      "epoch": 0.26496,
+      "grad_norm": 0.3031638665356794,
+      "learning_rate": 4.3998966316205636e-05,
+      "loss": 0.2538,
+      "step": 414
+    },
+    {
+      "epoch": 0.2656,
+      "grad_norm": 0.2532330310355455,
+      "learning_rate": 4.396450116562669e-05,
+      "loss": 0.2453,
+      "step": 415
+    },
+    {
+      "epoch": 0.26624,
+      "grad_norm": 0.27984325783985575,
+      "learning_rate": 4.392995090933669e-05,
+      "loss": 0.2527,
+      "step": 416
+    },
+    {
+      "epoch": 0.26688,
+      "grad_norm": 0.2853666077902048,
+      "learning_rate": 4.389531570238448e-05,
+      "loss": 0.2419,
+      "step": 417
+    },
+    {
+      "epoch": 0.26752,
+      "grad_norm": 0.34087513199677344,
+      "learning_rate": 4.386059570020015e-05,
+      "loss": 0.2782,
+      "step": 418
+    },
+    {
+      "epoch": 0.26816,
+      "grad_norm": 0.27330226400607055,
+      "learning_rate": 4.382579105859428e-05,
+      "loss": 0.2739,
+      "step": 419
+    },
+    {
+      "epoch": 0.2688,
+      "grad_norm": 0.31913954123535876,
+      "learning_rate": 4.379090193375735e-05,
+      "loss": 0.2685,
+      "step": 420
+    },
+    {
+      "epoch": 0.26944,
+      "grad_norm": 0.28177893210943633,
+      "learning_rate": 4.375592848225889e-05,
+      "loss": 0.2494,
+      "step": 421
+    },
+    {
+      "epoch": 0.27008,
+      "grad_norm": 0.28222647654040717,
+      "learning_rate": 4.372087086104692e-05,
+      "loss": 0.2769,
+      "step": 422
+    },
+    {
+      "epoch": 0.27072,
+      "grad_norm": 0.30927952594934816,
+      "learning_rate": 4.368572922744715e-05,
+      "loss": 0.2708,
+      "step": 423
+    },
+    {
+      "epoch": 0.27136,
+      "grad_norm": 0.2949859599241957,
+      "learning_rate": 4.365050373916233e-05,
+      "loss": 0.2753,
+      "step": 424
+    },
+    {
+      "epoch": 0.272,
+      "grad_norm": 0.29610748060111175,
+      "learning_rate": 4.3615194554271485e-05,
+      "loss": 0.2335,
+      "step": 425
+    },
+    {
+      "epoch": 0.27264,
+      "grad_norm": 0.3034357877644478,
+      "learning_rate": 4.357980183122927e-05,
+      "loss": 0.2726,
+      "step": 426
+    },
+    {
+      "epoch": 0.27328,
+      "grad_norm": 0.3059352582010788,
+      "learning_rate": 4.354432572886522e-05,
+      "loss": 0.271,
+      "step": 427
+    },
+    {
+      "epoch": 0.27392,
+      "grad_norm": 0.2771240245917802,
+      "learning_rate": 4.3508766406383036e-05,
+      "loss": 0.2784,
+      "step": 428
+    },
+    {
+      "epoch": 0.27456,
+      "grad_norm": 0.27908141628363836,
+      "learning_rate": 4.347312402335989e-05,
+      "loss": 0.2685,
+      "step": 429
+    },
+    {
+      "epoch": 0.2752,
+      "grad_norm": 0.28707815382315055,
+      "learning_rate": 4.34373987397457e-05,
+      "loss": 0.278,
+      "step": 430
+    },
+    {
+      "epoch": 0.27584,
+      "grad_norm": 0.2752693747267436,
+      "learning_rate": 4.340159071586241e-05,
+      "loss": 0.2447,
+      "step": 431
+    },
+    {
+      "epoch": 0.27648,
+      "grad_norm": 0.28362003524505025,
+      "learning_rate": 4.336570011240326e-05,
+      "loss": 0.2457,
+      "step": 432
+    },
+    {
+      "epoch": 0.27712,
+      "grad_norm": 0.6135783826974505,
+      "learning_rate": 4.33297270904321e-05,
+      "loss": 0.268,
+      "step": 433
+    },
+    {
+      "epoch": 0.27776,
+      "grad_norm": 0.27762568110334607,
+      "learning_rate": 4.329367181138262e-05,
+      "loss": 0.2765,
+      "step": 434
+    },
+    {
+      "epoch": 0.2784,
+      "grad_norm": 0.2858205844663221,
+      "learning_rate": 4.3257534437057674e-05,
+      "loss": 0.2569,
+      "step": 435
+    },
+    {
+      "epoch": 0.27904,
+      "grad_norm": 0.30296957565612004,
+      "learning_rate": 4.3221315129628506e-05,
+      "loss": 0.249,
+      "step": 436
+    },
+    {
+      "epoch": 0.27968,
+      "grad_norm": 0.27957554300787335,
+      "learning_rate": 4.318501405163406e-05,
+      "loss": 0.2628,
+      "step": 437
+    },
+    {
+      "epoch": 0.28032,
+      "grad_norm": 0.26204789686898283,
+      "learning_rate": 4.314863136598022e-05,
+      "loss": 0.2475,
+      "step": 438
+    },
+    {
+      "epoch": 0.28096,
+      "grad_norm": 0.3058255372708976,
+      "learning_rate": 4.311216723593913e-05,
+      "loss": 0.2866,
+      "step": 439
+    },
+    {
+      "epoch": 0.2816,
+      "grad_norm": 0.3251040999392898,
+      "learning_rate": 4.307562182514838e-05,
+      "loss": 0.2622,
+      "step": 440
+    },
+    {
+      "epoch": 0.28224,
+      "grad_norm": 0.30006956753828234,
+      "learning_rate": 4.3038995297610364e-05,
+      "loss": 0.2488,
+      "step": 441
+    },
+    {
+      "epoch": 0.28288,
+      "grad_norm": 0.30274868902673485,
+      "learning_rate": 4.3002287817691466e-05,
+      "loss": 0.2867,
+      "step": 442
+    },
+    {
+      "epoch": 0.28352,
+      "grad_norm": 0.38145313435180433,
+      "learning_rate": 4.2965499550121354e-05,
+      "loss": 0.2687,
+      "step": 443
+    },
+    {
+      "epoch": 0.28416,
+      "grad_norm": 0.3094348741385053,
+      "learning_rate": 4.2928630659992267e-05,
+      "loss": 0.2712,
+      "step": 444
+    },
+    {
+      "epoch": 0.2848,
+      "grad_norm": 0.34366344981433733,
+      "learning_rate": 4.2891681312758225e-05,
+      "loss": 0.2605,
+      "step": 445
+    },
+    {
+      "epoch": 0.28544,
+      "grad_norm": 0.2807715782879043,
+      "learning_rate": 4.285465167423432e-05,
+      "loss": 0.272,
+      "step": 446
+    },
+    {
+      "epoch": 0.28608,
+      "grad_norm": 0.3038994998195099,
+      "learning_rate": 4.281754191059596e-05,
+      "loss": 0.262,
+      "step": 447
+    },
+    {
+      "epoch": 0.28672,
+      "grad_norm": 0.30453615561443753,
+      "learning_rate": 4.278035218837813e-05,
+      "loss": 0.2889,
+      "step": 448
+    },
+    {
+      "epoch": 0.28736,
+      "grad_norm": 0.291161967151285,
+      "learning_rate": 4.274308267447463e-05,
+      "loss": 0.2519,
+      "step": 449
+    },
+    {
+      "epoch": 0.288,
+      "grad_norm": 0.2956628354845414,
+      "learning_rate": 4.270573353613734e-05,
+      "loss": 0.2709,
+      "step": 450
+    },
+    {
+      "epoch": 0.28864,
+      "grad_norm": 0.2796475447880478,
+      "learning_rate": 4.266830494097546e-05,
+      "loss": 0.2543,
+      "step": 451
+    },
+    {
+      "epoch": 0.28928,
+      "grad_norm": 0.3103279388881702,
+      "learning_rate": 4.263079705695478e-05,
+      "loss": 0.2638,
+      "step": 452
+    },
+    {
+      "epoch": 0.28992,
+      "grad_norm": 0.3546905046069225,
+      "learning_rate": 4.2593210052396894e-05,
+      "loss": 0.2545,
+      "step": 453
+    },
+    {
+      "epoch": 0.29056,
+      "grad_norm": 0.2790744270430645,
+      "learning_rate": 4.255554409597846e-05,
+      "loss": 0.2405,
+      "step": 454
+    },
+    {
+      "epoch": 0.2912,
+      "grad_norm": 0.3313654210870624,
+      "learning_rate": 4.2517799356730445e-05,
+      "loss": 0.2692,
+      "step": 455
+    },
+    {
+      "epoch": 0.29184,
+      "grad_norm": 0.3016384994176408,
+      "learning_rate": 4.247997600403737e-05,
+      "loss": 0.2601,
+      "step": 456
+    },
+    {
+      "epoch": 0.29248,
+      "grad_norm": 0.2993078346559434,
+      "learning_rate": 4.244207420763655e-05,
+      "loss": 0.2513,
+      "step": 457
+    },
+    {
+      "epoch": 0.29312,
+      "grad_norm": 0.31622969353803504,
+      "learning_rate": 4.240409413761731e-05,
+      "loss": 0.2551,
+      "step": 458
+    },
+    {
+      "epoch": 0.29376,
+      "grad_norm": 0.30524388925934887,
+      "learning_rate": 4.236603596442023e-05,
+      "loss": 0.2631,
+      "step": 459
+    },
+    {
+      "epoch": 0.2944,
+      "grad_norm": 0.30637282729194065,
+      "learning_rate": 4.232789985883642e-05,
+      "loss": 0.2696,
+      "step": 460
+    },
+    {
+      "epoch": 0.29504,
+      "grad_norm": 0.28616384975047443,
+      "learning_rate": 4.2289685992006704e-05,
+      "loss": 0.277,
+      "step": 461
+    },
+    {
+      "epoch": 0.29568,
+      "grad_norm": 0.3103128382558485,
+      "learning_rate": 4.225139453542087e-05,
+      "loss": 0.2619,
+      "step": 462
+    },
+    {
+      "epoch": 0.29632,
+      "grad_norm": 0.29703792526543216,
+      "learning_rate": 4.2213025660916904e-05,
+      "loss": 0.2676,
+      "step": 463
+    },
+    {
+      "epoch": 0.29696,
+      "grad_norm": 0.3168951655026096,
+      "learning_rate": 4.21745795406802e-05,
+      "loss": 0.2473,
+      "step": 464
+    },
+    {
+      "epoch": 0.2976,
+      "grad_norm": 0.2671842042378145,
+      "learning_rate": 4.2136056347242835e-05,
+      "loss": 0.2505,
+      "step": 465
+    },
+    {
+      "epoch": 0.29824,
+      "grad_norm": 0.29683839331480205,
+      "learning_rate": 4.209745625348273e-05,
+      "loss": 0.284,
+      "step": 466
+    },
+    {
+      "epoch": 0.29888,
+      "grad_norm": 0.32464847892381105,
+      "learning_rate": 4.205877943262291e-05,
+      "loss": 0.2462,
+      "step": 467
+    },
+    {
+      "epoch": 0.29952,
+      "grad_norm": 0.25754594330485725,
+      "learning_rate": 4.2020026058230756e-05,
+      "loss": 0.2688,
+      "step": 468
+    },
+    {
+      "epoch": 0.30016,
+      "grad_norm": 0.31150290164451966,
+      "learning_rate": 4.1981196304217154e-05,
+      "loss": 0.2641,
+      "step": 469
+    },
+    {
+      "epoch": 0.3008,
+      "grad_norm": 0.29116448915632004,
+      "learning_rate": 4.1942290344835765e-05,
+      "loss": 0.2643,
+      "step": 470
+    },
+    {
+      "epoch": 0.30144,
+      "grad_norm": 0.2990863020226855,
+      "learning_rate": 4.190330835468224e-05,
+      "loss": 0.267,
+      "step": 471
+    },
+    {
+      "epoch": 0.30208,
+      "grad_norm": 0.3072242093075254,
+      "learning_rate": 4.1864250508693424e-05,
+      "loss": 0.2791,
+      "step": 472
+    },
+    {
+      "epoch": 0.30272,
+      "grad_norm": 0.31058783224981473,
+      "learning_rate": 4.182511698214657e-05,
+      "loss": 0.2712,
+      "step": 473
+    },
+    {
+      "epoch": 0.30336,
+      "grad_norm": 0.2798304078793174,
+      "learning_rate": 4.1785907950658573e-05,
+      "loss": 0.2593,
+      "step": 474
+    },
+    {
+      "epoch": 0.304,
+      "grad_norm": 0.31535291727654785,
+      "learning_rate": 4.1746623590185155e-05,
+      "loss": 0.2536,
+      "step": 475
+    },
+    {
+      "epoch": 0.30464,
+      "grad_norm": 0.2990140489891215,
+      "learning_rate": 4.170726407702008e-05,
+      "loss": 0.247,
+      "step": 476
+    },
+    {
+      "epoch": 0.30528,
+      "grad_norm": 0.33391702160348635,
+      "learning_rate": 4.1667829587794374e-05,
+      "loss": 0.2835,
+      "step": 477
+    },
+    {
+      "epoch": 0.30592,
+      "grad_norm": 0.32213023722063483,
+      "learning_rate": 4.162832029947553e-05,
+      "loss": 0.2745,
+      "step": 478
+    },
+    {
+      "epoch": 0.30656,
+      "grad_norm": 0.27871899544394474,
+      "learning_rate": 4.158873638936672e-05,
+      "loss": 0.2476,
+      "step": 479
+    },
+    {
+      "epoch": 0.3072,
+      "grad_norm": 0.32481480905586535,
+      "learning_rate": 4.154907803510599e-05,
+      "loss": 0.2528,
+      "step": 480
+    },
+    {
+      "epoch": 0.30784,
+      "grad_norm": 0.29044250985733133,
+      "learning_rate": 4.150934541466545e-05,
+      "loss": 0.2626,
+      "step": 481
+    },
+    {
+      "epoch": 0.30848,
+      "grad_norm": 0.3001465924251364,
+      "learning_rate": 4.146953870635049e-05,
+      "loss": 0.2733,
+      "step": 482
+    },
+    {
+      "epoch": 0.30912,
+      "grad_norm": 0.2924668088023568,
+      "learning_rate": 4.1429658088798996e-05,
+      "loss": 0.2639,
+      "step": 483
+    },
+    {
+      "epoch": 0.30976,
+      "grad_norm": 0.2978329946137925,
+      "learning_rate": 4.138970374098052e-05,
+      "loss": 0.2493,
+      "step": 484
+    },
+    {
+      "epoch": 0.3104,
+      "grad_norm": 0.32039490633737067,
+      "learning_rate": 4.1349675842195494e-05,
+      "loss": 0.2679,
+      "step": 485
+    },
+    {
+      "epoch": 0.31104,
+      "grad_norm": 0.2784834783735675,
+      "learning_rate": 4.130957457207441e-05,
+      "loss": 0.2511,
+      "step": 486
+    },
+    {
+      "epoch": 0.31168,
+      "grad_norm": 0.3231096974927409,
+      "learning_rate": 4.126940011057703e-05,
+      "loss": 0.2417,
+      "step": 487
+    },
+    {
+      "epoch": 0.31232,
+      "grad_norm": 0.29574649661084784,
+      "learning_rate": 4.1229152637991576e-05,
+      "loss": 0.2683,
+      "step": 488
+    },
+    {
+      "epoch": 0.31296,
+      "grad_norm": 0.3198759705633704,
+      "learning_rate": 4.1188832334933906e-05,
+      "loss": 0.2485,
+      "step": 489
+    },
+    {
+      "epoch": 0.3136,
+      "grad_norm": 0.2743856200052568,
+      "learning_rate": 4.114843938234673e-05,
+      "loss": 0.2753,
+      "step": 490
+    },
+    {
+      "epoch": 0.31424,
+      "grad_norm": 0.31005901539666614,
+      "learning_rate": 4.1107973961498763e-05,
+      "loss": 0.2591,
+      "step": 491
+    },
+    {
+      "epoch": 0.31488,
+      "grad_norm": 0.3300495689975318,
+      "learning_rate": 4.106743625398395e-05,
+      "loss": 0.2675,
+      "step": 492
+    },
+    {
+      "epoch": 0.31552,
+      "grad_norm": 0.3068927581792596,
+      "learning_rate": 4.102682644172061e-05,
+      "loss": 0.2871,
+      "step": 493
+    },
+    {
+      "epoch": 0.31616,
+      "grad_norm": 0.3264341456984584,
+      "learning_rate": 4.0986144706950677e-05,
+      "loss": 0.2755,
+      "step": 494
+    },
+    {
+      "epoch": 0.3168,
+      "grad_norm": 0.31711346944454233,
+      "learning_rate": 4.094539123223881e-05,
+      "loss": 0.2442,
+      "step": 495
+    },
+    {
+      "epoch": 0.31744,
+      "grad_norm": 0.2862941880474931,
+      "learning_rate": 4.090456620047162e-05,
+      "loss": 0.2447,
+      "step": 496
+    },
+    {
+      "epoch": 0.31808,
+      "grad_norm": 0.3455300856074319,
+      "learning_rate": 4.086366979485685e-05,
+      "loss": 0.2806,
+      "step": 497
+    },
+    {
+      "epoch": 0.31872,
+      "grad_norm": 0.3038196785649234,
+      "learning_rate": 4.082270219892254e-05,
+      "loss": 0.2643,
+      "step": 498
+    },
+    {
+      "epoch": 0.31936,
+      "grad_norm": 0.27271340321003223,
+      "learning_rate": 4.0781663596516206e-05,
+      "loss": 0.2433,
+      "step": 499
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.29835333522383656,
+      "learning_rate": 4.074055417180401e-05,
+      "loss": 0.2704,
+      "step": 500
+    },
+    {
+      "epoch": 0.32064,
+      "grad_norm": 0.3196694498557124,
+      "learning_rate": 4.069937410926994e-05,
+      "loss": 0.2549,
+      "step": 501
+    },
+    {
+      "epoch": 0.32128,
+      "grad_norm": 0.28825300365435247,
+      "learning_rate": 4.0658123593714984e-05,
+      "loss": 0.2647,
+      "step": 502
+    },
+    {
+      "epoch": 0.32192,
+      "grad_norm": 0.2786545499138097,
+      "learning_rate": 4.061680281025628e-05,
+      "loss": 0.2477,
+      "step": 503
+    },
+    {
+      "epoch": 0.32256,
+      "grad_norm": 0.29978230649243376,
+      "learning_rate": 4.057541194432634e-05,
+      "loss": 0.2421,
+      "step": 504
+    },
+    {
+      "epoch": 0.3232,
+      "grad_norm": 0.3012784250324835,
+      "learning_rate": 4.053395118167214e-05,
+      "loss": 0.2581,
+      "step": 505
+    },
+    {
+      "epoch": 0.32384,
+      "grad_norm": 0.29499478568673526,
+      "learning_rate": 4.049242070835435e-05,
+      "loss": 0.2874,
+      "step": 506
+    },
+    {
+      "epoch": 0.32448,
+      "grad_norm": 0.31674834634025634,
+      "learning_rate": 4.0450820710746475e-05,
+      "loss": 0.2608,
+      "step": 507
+    },
+    {
+      "epoch": 0.32512,
+      "grad_norm": 0.2865283632314673,
+      "learning_rate": 4.0409151375534014e-05,
+      "loss": 0.2637,
+      "step": 508
+    },
+    {
+      "epoch": 0.32576,
+      "grad_norm": 0.30827781105540003,
+      "learning_rate": 4.036741288971362e-05,
+      "loss": 0.2558,
+      "step": 509
+    },
+    {
+      "epoch": 0.3264,
+      "grad_norm": 0.31853194371996735,
+      "learning_rate": 4.032560544059228e-05,
+      "loss": 0.255,
+      "step": 510
+    },
+    {
+      "epoch": 0.32704,
+      "grad_norm": 0.38294307444540887,
+      "learning_rate": 4.028372921578647e-05,
+      "loss": 0.2641,
+      "step": 511
+    },
+    {
+      "epoch": 0.32768,
+      "grad_norm": 0.2864115669013345,
+      "learning_rate": 4.024178440322128e-05,
+      "loss": 0.2507,
+      "step": 512
+    },
+    {
+      "epoch": 0.32832,
+      "grad_norm": 0.2773279212490367,
+      "learning_rate": 4.019977119112962e-05,
+      "loss": 0.2604,
+      "step": 513
+    },
+    {
+      "epoch": 0.32896,
+      "grad_norm": 0.30350170146466277,
+      "learning_rate": 4.015768976805135e-05,
+      "loss": 0.303,
+      "step": 514
+    },
+    {
+      "epoch": 0.3296,
+      "grad_norm": 0.3019836637698821,
+      "learning_rate": 4.011554032283242e-05,
+      "loss": 0.2342,
+      "step": 515
+    },
+    {
+      "epoch": 0.33024,
+      "grad_norm": 0.33693079309207474,
+      "learning_rate": 4.007332304462406e-05,
+      "loss": 0.2503,
+      "step": 516
+    },
+    {
+      "epoch": 0.33088,
+      "grad_norm": 0.301807035388646,
+      "learning_rate": 4.003103812288189e-05,
+      "loss": 0.2681,
+      "step": 517
+    },
+    {
+      "epoch": 0.33152,
+      "grad_norm": 0.27363500419504855,
+      "learning_rate": 3.99886857473651e-05,
+      "loss": 0.2523,
+      "step": 518
+    },
+    {
+      "epoch": 0.33216,
+      "grad_norm": 0.3288948783067086,
+      "learning_rate": 3.9946266108135575e-05,
+      "loss": 0.2771,
+      "step": 519
+    },
+    {
+      "epoch": 0.3328,
+      "grad_norm": 0.3116926537482921,
+      "learning_rate": 3.9903779395557087e-05,
+      "loss": 0.261,
+      "step": 520
+    },
+    {
+      "epoch": 0.33344,
+      "grad_norm": 0.2620289802478765,
+      "learning_rate": 3.9861225800294366e-05,
+      "loss": 0.2295,
+      "step": 521
+    },
+    {
+      "epoch": 0.33408,
+      "grad_norm": 0.32247525286268586,
+      "learning_rate": 3.981860551331232e-05,
+      "loss": 0.2671,
+      "step": 522
+    },
+    {
+      "epoch": 0.33472,
+      "grad_norm": 0.29485974202493387,
+      "learning_rate": 3.977591872587512e-05,
+      "loss": 0.2559,
+      "step": 523
+    },
+    {
+      "epoch": 0.33536,
+      "grad_norm": 0.2615445610289309,
+      "learning_rate": 3.973316562954539e-05,
+      "loss": 0.2514,
+      "step": 524
+    },
+    {
+      "epoch": 0.336,
+      "grad_norm": 0.28540916266337385,
+      "learning_rate": 3.969034641618332e-05,
+      "loss": 0.2845,
+      "step": 525
+    },
+    {
+      "epoch": 0.33664,
+      "grad_norm": 0.2973099909096069,
+      "learning_rate": 3.964746127794579e-05,
+      "loss": 0.2627,
+      "step": 526
+    },
+    {
+      "epoch": 0.33728,
+      "grad_norm": 0.2829998813648647,
+      "learning_rate": 3.9604510407285524e-05,
+      "loss": 0.2502,
+      "step": 527
+    },
+    {
+      "epoch": 0.33792,
+      "grad_norm": 0.2923882761100174,
+      "learning_rate": 3.956149399695027e-05,
+      "loss": 0.2524,
+      "step": 528
+    },
+    {
+      "epoch": 0.33856,
+      "grad_norm": 0.31889981262923656,
+      "learning_rate": 3.951841223998186e-05,
+      "loss": 0.2802,
+      "step": 529
+    },
+    {
+      "epoch": 0.3392,
+      "grad_norm": 0.27084760232555066,
+      "learning_rate": 3.947526532971538e-05,
+      "loss": 0.2447,
+      "step": 530
+    },
+    {
+      "epoch": 0.33984,
+      "grad_norm": 0.277167819276775,
+      "learning_rate": 3.943205345977831e-05,
+      "loss": 0.2599,
+      "step": 531
+    },
+    {
+      "epoch": 0.34048,
+      "grad_norm": 0.45847772602327513,
+      "learning_rate": 3.9388776824089636e-05,
+      "loss": 0.2568,
+      "step": 532
+    },
+    {
+      "epoch": 0.34112,
+      "grad_norm": 0.2646005627784305,
+      "learning_rate": 3.9345435616858995e-05,
+      "loss": 0.2674,
+      "step": 533
+    },
+    {
+      "epoch": 0.34176,
+      "grad_norm": 0.2704833186174627,
+      "learning_rate": 3.930203003258579e-05,
+      "loss": 0.2566,
+      "step": 534
+    },
+    {
+      "epoch": 0.3424,
+      "grad_norm": 0.27320797496261323,
+      "learning_rate": 3.9258560266058335e-05,
+      "loss": 0.2689,
+      "step": 535
+    },
+    {
+      "epoch": 0.34304,
+      "grad_norm": 0.26245552780653786,
+      "learning_rate": 3.921502651235297e-05,
+      "loss": 0.2533,
+      "step": 536
+    },
+    {
+      "epoch": 0.34368,
+      "grad_norm": 0.2686721573275851,
+      "learning_rate": 3.917142896683318e-05,
+      "loss": 0.2713,
+      "step": 537
+    },
+    {
+      "epoch": 0.34432,
+      "grad_norm": 0.27677482339556025,
+      "learning_rate": 3.912776782514872e-05,
+      "loss": 0.2541,
+      "step": 538
+    },
+    {
+      "epoch": 0.34496,
+      "grad_norm": 0.2752780592634784,
+      "learning_rate": 3.908404328323475e-05,
+      "loss": 0.2724,
+      "step": 539
+    },
+    {
+      "epoch": 0.3456,
+      "grad_norm": 0.30938115723324516,
+      "learning_rate": 3.904025553731094e-05,
+      "loss": 0.2413,
+      "step": 540
+    },
+    {
+      "epoch": 0.34624,
+      "grad_norm": 0.26305059884444465,
+      "learning_rate": 3.8996404783880606e-05,
+      "loss": 0.2429,
+      "step": 541
+    },
+    {
+      "epoch": 0.34688,
+      "grad_norm": 0.27017911566688885,
+      "learning_rate": 3.895249121972981e-05,
+      "loss": 0.2425,
+      "step": 542
+    },
+    {
+      "epoch": 0.34752,
+      "grad_norm": 0.2807572440929959,
+      "learning_rate": 3.890851504192649e-05,
+      "loss": 0.2692,
+      "step": 543
+    },
+    {
+      "epoch": 0.34816,
+      "grad_norm": 0.2937755464668312,
+      "learning_rate": 3.8864476447819565e-05,
+      "loss": 0.2471,
+      "step": 544
+    },
+    {
+      "epoch": 0.3488,
+      "grad_norm": 0.26722639745381527,
+      "learning_rate": 3.8820375635038067e-05,
+      "loss": 0.257,
+      "step": 545
+    },
+    {
+      "epoch": 0.34944,
+      "grad_norm": 0.32942353225222054,
+      "learning_rate": 3.877621280149022e-05,
+      "loss": 0.2673,
+      "step": 546
+    },
+    {
+      "epoch": 0.35008,
+      "grad_norm": 0.26071685079047663,
+      "learning_rate": 3.87319881453626e-05,
+      "loss": 0.235,
+      "step": 547
+    },
+    {
+      "epoch": 0.35072,
+      "grad_norm": 0.2703795884584656,
+      "learning_rate": 3.868770186511921e-05,
+      "loss": 0.2618,
+      "step": 548
+    },
+    {
+      "epoch": 0.35136,
+      "grad_norm": 0.3063325675005532,
+      "learning_rate": 3.864335415950059e-05,
+      "loss": 0.2437,
+      "step": 549
+    },
+    {
+      "epoch": 0.352,
+      "grad_norm": 0.25931696906001084,
+      "learning_rate": 3.859894522752297e-05,
+      "loss": 0.2498,
+      "step": 550
+    },
+    {
+      "epoch": 0.35264,
+      "grad_norm": 0.317347010295714,
+      "learning_rate": 3.8554475268477274e-05,
+      "loss": 0.2573,
+      "step": 551
+    },
+    {
+      "epoch": 0.35328,
+      "grad_norm": 0.2879966431448922,
+      "learning_rate": 3.850994448192836e-05,
+      "loss": 0.2601,
+      "step": 552
+    },
+    {
+      "epoch": 0.35392,
+      "grad_norm": 0.2704993494133405,
+      "learning_rate": 3.846535306771404e-05,
+      "loss": 0.2495,
+      "step": 553
+    },
+    {
+      "epoch": 0.35456,
+      "grad_norm": 0.32753739181266783,
+      "learning_rate": 3.842070122594417e-05,
+      "loss": 0.2647,
+      "step": 554
+    },
+    {
+      "epoch": 0.3552,
+      "grad_norm": 0.27693874665061385,
+      "learning_rate": 3.837598915699981e-05,
+      "loss": 0.2459,
+      "step": 555
+    },
+    {
+      "epoch": 0.35584,
+      "grad_norm": 0.3033680637986832,
+      "learning_rate": 3.83312170615323e-05,
+      "loss": 0.2384,
+      "step": 556
+    },
+    {
+      "epoch": 0.35648,
+      "grad_norm": 0.3039697232443411,
+      "learning_rate": 3.8286385140462334e-05,
+      "loss": 0.2473,
+      "step": 557
+    },
+    {
+      "epoch": 0.35712,
+      "grad_norm": 0.3047824145474792,
+      "learning_rate": 3.8241493594979114e-05,
+      "loss": 0.2635,
+      "step": 558
+    },
+    {
+      "epoch": 0.35776,
+      "grad_norm": 0.2991402315294844,
+      "learning_rate": 3.8196542626539396e-05,
+      "loss": 0.2637,
+      "step": 559
+    },
+    {
+      "epoch": 0.3584,
+      "grad_norm": 0.3092917498688086,
+      "learning_rate": 3.815153243686658e-05,
+      "loss": 0.2648,
+      "step": 560
+    },
+    {
+      "epoch": 0.35904,
+      "grad_norm": 0.30146699010235556,
+      "learning_rate": 3.8106463227949885e-05,
+      "loss": 0.265,
+      "step": 561
+    },
+    {
+      "epoch": 0.35968,
+      "grad_norm": 0.2769544706427565,
+      "learning_rate": 3.806133520204335e-05,
+      "loss": 0.2352,
+      "step": 562
+    },
+    {
+      "epoch": 0.36032,
+      "grad_norm": 0.35477609684955097,
+      "learning_rate": 3.801614856166495e-05,
+      "loss": 0.2441,
+      "step": 563
+    },
+    {
+      "epoch": 0.36096,
+      "grad_norm": 0.29044978422937073,
+      "learning_rate": 3.797090350959574e-05,
+      "loss": 0.2478,
+      "step": 564
+    },
+    {
+      "epoch": 0.3616,
+      "grad_norm": 0.2847992844131417,
+      "learning_rate": 3.792560024887887e-05,
+      "loss": 0.253,
+      "step": 565
+    },
+    {
+      "epoch": 0.36224,
+      "grad_norm": 0.3082206555429372,
+      "learning_rate": 3.788023898281873e-05,
+      "loss": 0.252,
+      "step": 566
+    },
+    {
+      "epoch": 0.36288,
+      "grad_norm": 0.29684039873581247,
+      "learning_rate": 3.783481991498001e-05,
+      "loss": 0.2573,
+      "step": 567
+    },
+    {
+      "epoch": 0.36352,
+      "grad_norm": 0.2799704203242341,
+      "learning_rate": 3.7789343249186805e-05,
+      "loss": 0.2654,
+      "step": 568
+    },
+    {
+      "epoch": 0.36416,
+      "grad_norm": 0.3072567979064305,
+      "learning_rate": 3.774380918952165e-05,
+      "loss": 0.2749,
+      "step": 569
+    },
+    {
+      "epoch": 0.3648,
+      "grad_norm": 0.2903629157466471,
+      "learning_rate": 3.76982179403247e-05,
+      "loss": 0.2379,
+      "step": 570
+    },
+    {
+      "epoch": 0.36544,
+      "grad_norm": 0.27576649206877873,
+      "learning_rate": 3.7652569706192694e-05,
+      "loss": 0.2682,
+      "step": 571
+    },
+    {
+      "epoch": 0.36608,
+      "grad_norm": 0.27013938988147623,
+      "learning_rate": 3.760686469197815e-05,
+      "loss": 0.2492,
+      "step": 572
+    },
+    {
+      "epoch": 0.36672,
+      "grad_norm": 0.36320284367672434,
+      "learning_rate": 3.756110310278835e-05,
+      "loss": 0.249,
+      "step": 573
+    },
+    {
+      "epoch": 0.36736,
+      "grad_norm": 0.2455885509270789,
+      "learning_rate": 3.7515285143984524e-05,
+      "loss": 0.2429,
+      "step": 574
+    },
+    {
+      "epoch": 0.368,
+      "grad_norm": 0.2968349995967185,
+      "learning_rate": 3.746941102118082e-05,
+      "loss": 0.2503,
+      "step": 575
+    },
+    {
+      "epoch": 0.36864,
+      "grad_norm": 0.25701314511324636,
+      "learning_rate": 3.742348094024342e-05,
+      "loss": 0.2721,
+      "step": 576
+    },
+    {
+      "epoch": 0.36928,
+      "grad_norm": 0.3043146256349847,
+      "learning_rate": 3.7377495107289664e-05,
+      "loss": 0.2285,
+      "step": 577
+    },
+    {
+      "epoch": 0.36992,
+      "grad_norm": 0.2977740197991099,
+      "learning_rate": 3.7331453728687054e-05,
+      "loss": 0.2529,
+      "step": 578
+    },
+    {
+      "epoch": 0.37056,
+      "grad_norm": 0.331830692934904,
+      "learning_rate": 3.7285357011052394e-05,
+      "loss": 0.2391,
+      "step": 579
+    },
+    {
+      "epoch": 0.3712,
+      "grad_norm": 0.29610508157500476,
+      "learning_rate": 3.72392051612508e-05,
+      "loss": 0.2692,
+      "step": 580
+    },
+    {
+      "epoch": 0.37184,
+      "grad_norm": 0.2695864248403297,
+      "learning_rate": 3.719299838639481e-05,
+      "loss": 0.232,
+      "step": 581
+    },
+    {
+      "epoch": 0.37248,
+      "grad_norm": 0.3011787829747353,
+      "learning_rate": 3.714673689384346e-05,
+      "loss": 0.2734,
+      "step": 582
+    },
+    {
+      "epoch": 0.37312,
+      "grad_norm": 0.29463703959958826,
+      "learning_rate": 3.710042089120131e-05,
+      "loss": 0.2482,
+      "step": 583
+    },
+    {
+      "epoch": 0.37376,
+      "grad_norm": 0.2891496575716541,
+      "learning_rate": 3.705405058631756e-05,
+      "loss": 0.263,
+      "step": 584
+    },
+    {
+      "epoch": 0.3744,
+      "grad_norm": 0.28223121574762366,
+      "learning_rate": 3.700762618728508e-05,
+      "loss": 0.2566,
+      "step": 585
+    },
+    {
+      "epoch": 0.37504,
+      "grad_norm": 0.30203997125282284,
+      "learning_rate": 3.696114790243955e-05,
+      "loss": 0.2641,
+      "step": 586
+    },
+    {
+      "epoch": 0.37568,
+      "grad_norm": 0.30418487005364714,
+      "learning_rate": 3.691461594035841e-05,
+      "loss": 0.2788,
+      "step": 587
+    },
+    {
+      "epoch": 0.37632,
+      "grad_norm": 0.27057259769384845,
+      "learning_rate": 3.686803050986e-05,
+      "loss": 0.2475,
+      "step": 588
+    },
+    {
+      "epoch": 0.37696,
+      "grad_norm": 0.32392640185038807,
+      "learning_rate": 3.6821391820002596e-05,
+      "loss": 0.2435,
+      "step": 589
+    },
+    {
+      "epoch": 0.3776,
+      "grad_norm": 0.2594129682407732,
+      "learning_rate": 3.677470008008352e-05,
+      "loss": 0.2465,
+      "step": 590
+    },
+    {
+      "epoch": 0.37824,
+      "grad_norm": 0.28387705453798534,
+      "learning_rate": 3.672795549963814e-05,
+      "loss": 0.2456,
+      "step": 591
+    },
+    {
+      "epoch": 0.37888,
+      "grad_norm": 0.2950074433710385,
+      "learning_rate": 3.668115828843894e-05,
+      "loss": 0.2501,
+      "step": 592
+    },
+    {
+      "epoch": 0.37952,
+      "grad_norm": 0.281455028183716,
+      "learning_rate": 3.6634308656494605e-05,
+      "loss": 0.2507,
+      "step": 593
+    },
+    {
+      "epoch": 0.38016,
+      "grad_norm": 0.29396350647979413,
+      "learning_rate": 3.6587406814049075e-05,
+      "loss": 0.2408,
+      "step": 594
+    },
+    {
+      "epoch": 0.3808,
+      "grad_norm": 0.27711962772675214,
+      "learning_rate": 3.6540452971580566e-05,
+      "loss": 0.2586,
+      "step": 595
+    },
+    {
+      "epoch": 0.38144,
+      "grad_norm": 0.294567573462565,
+      "learning_rate": 3.649344733980068e-05,
+      "loss": 0.2552,
+      "step": 596
+    },
+    {
+      "epoch": 0.38208,
+      "grad_norm": 0.27673346556360123,
+      "learning_rate": 3.6446390129653405e-05,
+      "loss": 0.2498,
+      "step": 597
+    },
+    {
+      "epoch": 0.38272,
+      "grad_norm": 0.2818818939618678,
+      "learning_rate": 3.63992815523142e-05,
+      "loss": 0.2411,
+      "step": 598
+    },
+    {
+      "epoch": 0.38336,
+      "grad_norm": 0.29234462981277415,
+      "learning_rate": 3.6352121819189075e-05,
+      "loss": 0.2425,
+      "step": 599
+    },
+    {
+      "epoch": 0.384,
+      "grad_norm": 0.3269998641272275,
+      "learning_rate": 3.630491114191354e-05,
+      "loss": 0.2405,
+      "step": 600
+    },
+    {
+      "epoch": 0.38464,
+      "grad_norm": 0.3001209139639063,
+      "learning_rate": 3.62576497323518e-05,
+      "loss": 0.2793,
+      "step": 601
+    },
+    {
+      "epoch": 0.38528,
+      "grad_norm": 0.28134693156999974,
+      "learning_rate": 3.621033780259566e-05,
+      "loss": 0.2388,
+      "step": 602
+    },
+    {
+      "epoch": 0.38592,
+      "grad_norm": 0.3244703748522845,
+      "learning_rate": 3.616297556496369e-05,
+      "loss": 0.238,
+      "step": 603
+    },
+    {
+      "epoch": 0.38656,
+      "grad_norm": 0.2637566816735972,
+      "learning_rate": 3.6115563232000206e-05,
+      "loss": 0.2354,
+      "step": 604
+    },
+    {
+      "epoch": 0.3872,
+      "grad_norm": 0.28762941282849863,
+      "learning_rate": 3.6068101016474314e-05,
+      "loss": 0.237,
+      "step": 605
+    },
+    {
+      "epoch": 0.38784,
+      "grad_norm": 0.29578283278832357,
+      "learning_rate": 3.602058913137902e-05,
+      "loss": 0.2536,
+      "step": 606
+    },
+    {
+      "epoch": 0.38848,
+      "grad_norm": 0.25838120579268425,
+      "learning_rate": 3.597302778993018e-05,
+      "loss": 0.2536,
+      "step": 607
+    },
+    {
+      "epoch": 0.38912,
+      "grad_norm": 0.29178349766440004,
+      "learning_rate": 3.592541720556561e-05,
+      "loss": 0.2468,
+      "step": 608
+    },
+    {
+      "epoch": 0.38976,
+      "grad_norm": 0.28538733473554107,
+      "learning_rate": 3.5877757591944137e-05,
+      "loss": 0.2543,
+      "step": 609
+    },
+    {
+      "epoch": 0.3904,
+      "grad_norm": 0.28910068675727285,
+      "learning_rate": 3.583004916294458e-05,
+      "loss": 0.2587,
+      "step": 610
+    },
+    {
+      "epoch": 0.39104,
+      "grad_norm": 0.27742239985535233,
+      "learning_rate": 3.578229213266483e-05,
+      "loss": 0.2526,
+      "step": 611
+    },
+    {
+      "epoch": 0.39168,
+      "grad_norm": 0.294115969883614,
+      "learning_rate": 3.573448671542088e-05,
+      "loss": 0.2502,
+      "step": 612
+    },
+    {
+      "epoch": 0.39232,
+      "grad_norm": 0.29258891729454395,
+      "learning_rate": 3.568663312574589e-05,
+      "loss": 0.2351,
+      "step": 613
+    },
+    {
+      "epoch": 0.39296,
+      "grad_norm": 0.2851564904354055,
+      "learning_rate": 3.563873157838917e-05,
+      "loss": 0.2585,
+      "step": 614
+    },
+    {
+      "epoch": 0.3936,
+      "grad_norm": 0.2810424321596881,
+      "learning_rate": 3.5590782288315266e-05,
+      "loss": 0.2703,
+      "step": 615
+    },
+    {
+      "epoch": 0.39424,
+      "grad_norm": 0.2963400750722823,
+      "learning_rate": 3.5542785470702965e-05,
+      "loss": 0.2583,
+      "step": 616
+    },
+    {
+      "epoch": 0.39488,
+      "grad_norm": 0.28554352127230415,
+      "learning_rate": 3.549474134094434e-05,
+      "loss": 0.2607,
+      "step": 617
+    },
+    {
+      "epoch": 0.39552,
+      "grad_norm": 0.2751785354484056,
+      "learning_rate": 3.54466501146438e-05,
+      "loss": 0.2423,
+      "step": 618
+    },
+    {
+      "epoch": 0.39616,
+      "grad_norm": 0.3109111333105493,
+      "learning_rate": 3.539851200761707e-05,
+      "loss": 0.2826,
+      "step": 619
+    },
+    {
+      "epoch": 0.3968,
+      "grad_norm": 0.28380724524869966,
+      "learning_rate": 3.535032723589031e-05,
+      "loss": 0.2382,
+      "step": 620
+    },
+    {
+      "epoch": 0.39744,
+      "grad_norm": 0.29031803555316743,
+      "learning_rate": 3.5302096015699035e-05,
+      "loss": 0.2529,
+      "step": 621
+    },
+    {
+      "epoch": 0.39808,
+      "grad_norm": 0.2862319408791201,
+      "learning_rate": 3.5253818563487266e-05,
+      "loss": 0.2497,
+      "step": 622
+    },
+    {
+      "epoch": 0.39872,
+      "grad_norm": 0.2597359081273056,
+      "learning_rate": 3.5205495095906444e-05,
+      "loss": 0.2502,
+      "step": 623
+    },
+    {
+      "epoch": 0.39936,
+      "grad_norm": 0.26645746598257863,
+      "learning_rate": 3.5157125829814534e-05,
+      "loss": 0.2427,
+      "step": 624
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.2589738737012416,
+      "learning_rate": 3.510871098227503e-05,
+      "loss": 0.2314,
+      "step": 625
+    },
+    {
+      "epoch": 0.40064,
+      "grad_norm": 0.2808606689504793,
+      "learning_rate": 3.506025077055597e-05,
+      "loss": 0.2906,
+      "step": 626
+    },
+    {
+      "epoch": 0.40128,
+      "grad_norm": 0.27935188272224154,
+      "learning_rate": 3.501174541212898e-05,
+      "loss": 0.2419,
+      "step": 627
+    },
+    {
+      "epoch": 0.40192,
+      "grad_norm": 0.2907671332799639,
+      "learning_rate": 3.496319512466827e-05,
+      "loss": 0.2596,
+      "step": 628
+    },
+    {
+      "epoch": 0.40256,
+      "grad_norm": 0.2844727991856023,
+      "learning_rate": 3.4914600126049696e-05,
+      "loss": 0.2459,
+      "step": 629
+    },
+    {
+      "epoch": 0.4032,
+      "grad_norm": 0.2719868034683634,
+      "learning_rate": 3.486596063434975e-05,
+      "loss": 0.2431,
+      "step": 630
+    },
+    {
+      "epoch": 0.40384,
+      "grad_norm": 0.2823824998273708,
+      "learning_rate": 3.481727686784458e-05,
+      "loss": 0.2445,
+      "step": 631
+    },
+    {
+      "epoch": 0.40448,
+      "grad_norm": 0.2756425905858028,
+      "learning_rate": 3.476854904500905e-05,
+      "loss": 0.2463,
+      "step": 632
+    },
+    {
+      "epoch": 0.40512,
+      "grad_norm": 0.2779852039855442,
+      "learning_rate": 3.471977738451572e-05,
+      "loss": 0.2534,
+      "step": 633
+    },
+    {
+      "epoch": 0.40576,
+      "grad_norm": 0.2950432225935468,
+      "learning_rate": 3.467096210523387e-05,
+      "loss": 0.2532,
+      "step": 634
+    },
+    {
+      "epoch": 0.4064,
+      "grad_norm": 0.24821255764789482,
+      "learning_rate": 3.4622103426228534e-05,
+      "loss": 0.2616,
+      "step": 635
+    },
+    {
+      "epoch": 0.40704,
+      "grad_norm": 0.2665965175038227,
+      "learning_rate": 3.4573201566759515e-05,
+      "loss": 0.2451,
+      "step": 636
+    },
+    {
+      "epoch": 0.40768,
+      "grad_norm": 0.2944539832776922,
+      "learning_rate": 3.4524256746280374e-05,
+      "loss": 0.2546,
+      "step": 637
+    },
+    {
+      "epoch": 0.40832,
+      "grad_norm": 0.26681868196558595,
+      "learning_rate": 3.447526918443748e-05,
+      "loss": 0.2489,
+      "step": 638
+    },
+    {
+      "epoch": 0.40896,
+      "grad_norm": 0.2700230642277879,
+      "learning_rate": 3.442623910106901e-05,
+      "loss": 0.2297,
+      "step": 639
+    },
+    {
+      "epoch": 0.4096,
+      "grad_norm": 0.25839557803313135,
+      "learning_rate": 3.4377166716203945e-05,
+      "loss": 0.2397,
+      "step": 640
+    },
+    {
+      "epoch": 0.41024,
+      "grad_norm": 0.3021408976030671,
+      "learning_rate": 3.4328052250061124e-05,
+      "loss": 0.2455,
+      "step": 641
+    },
+    {
+      "epoch": 0.41088,
+      "grad_norm": 0.2859479339462905,
+      "learning_rate": 3.4278895923048226e-05,
+      "loss": 0.2547,
+      "step": 642
+    },
+    {
+      "epoch": 0.41152,
+      "grad_norm": 0.2745261932057034,
+      "learning_rate": 3.4229697955760767e-05,
+      "loss": 0.2293,
+      "step": 643
+    },
+    {
+      "epoch": 0.41216,
+      "grad_norm": 0.2909265259735638,
+      "learning_rate": 3.4180458568981136e-05,
+      "loss": 0.2788,
+      "step": 644
+    },
+    {
+      "epoch": 0.4128,
+      "grad_norm": 0.25620787622393093,
+      "learning_rate": 3.4131177983677624e-05,
+      "loss": 0.2258,
+      "step": 645
+    },
+    {
+      "epoch": 0.41344,
+      "grad_norm": 0.2838977519379636,
+      "learning_rate": 3.4081856421003364e-05,
+      "loss": 0.2364,
+      "step": 646
+    },
+    {
+      "epoch": 0.41408,
+      "grad_norm": 0.2779231236669113,
+      "learning_rate": 3.403249410229542e-05,
+      "loss": 0.2465,
+      "step": 647
+    },
+    {
+      "epoch": 0.41472,
+      "grad_norm": 0.33020342043739015,
+      "learning_rate": 3.398309124907372e-05,
+      "loss": 0.2562,
+      "step": 648
+    },
+    {
+      "epoch": 0.41536,
+      "grad_norm": 0.24666782803318293,
+      "learning_rate": 3.393364808304012e-05,
+      "loss": 0.2712,
+      "step": 649
+    },
+    {
+      "epoch": 0.416,
+      "grad_norm": 0.30815484394945014,
+      "learning_rate": 3.3884164826077384e-05,
+      "loss": 0.2456,
+      "step": 650
+    },
+    {
+      "epoch": 0.41664,
+      "grad_norm": 0.2812813367796177,
+      "learning_rate": 3.383464170024817e-05,
+      "loss": 0.2441,
+      "step": 651
+    },
+    {
+      "epoch": 0.41728,
+      "grad_norm": 0.28607485450453224,
+      "learning_rate": 3.378507892779408e-05,
+      "loss": 0.2701,
+      "step": 652
+    },
+    {
+      "epoch": 0.41792,
+      "grad_norm": 0.2960472952173197,
+      "learning_rate": 3.37354767311346e-05,
+      "loss": 0.2532,
+      "step": 653
+    },
+    {
+      "epoch": 0.41856,
+      "grad_norm": 0.2832878309358752,
+      "learning_rate": 3.368583533286617e-05,
+      "loss": 0.235,
+      "step": 654
+    },
+    {
+      "epoch": 0.4192,
+      "grad_norm": 0.2906843275840211,
+      "learning_rate": 3.363615495576115e-05,
+      "loss": 0.2564,
+      "step": 655
+    },
+    {
+      "epoch": 0.41984,
+      "grad_norm": 0.2790824461789213,
+      "learning_rate": 3.3586435822766785e-05,
+      "loss": 0.2568,
+      "step": 656
+    },
+    {
+      "epoch": 0.42048,
+      "grad_norm": 0.28446065705330625,
+      "learning_rate": 3.35366781570043e-05,
+      "loss": 0.2363,
+      "step": 657
+    },
+    {
+      "epoch": 0.42112,
+      "grad_norm": 0.3157219415215847,
+      "learning_rate": 3.3486882181767785e-05,
+      "loss": 0.2606,
+      "step": 658
+    },
+    {
+      "epoch": 0.42176,
+      "grad_norm": 0.3056369034770089,
+      "learning_rate": 3.34370481205233e-05,
+      "loss": 0.2617,
+      "step": 659
+    },
+    {
+      "epoch": 0.4224,
+      "grad_norm": 0.2766010888311108,
+      "learning_rate": 3.3387176196907776e-05,
+      "loss": 0.2609,
+      "step": 660
+    },
+    {
+      "epoch": 0.42304,
+      "grad_norm": 0.2545434388799889,
+      "learning_rate": 3.333726663472809e-05,
+      "loss": 0.2414,
+      "step": 661
+    },
+    {
+      "epoch": 0.42368,
+      "grad_norm": 0.27376260910857747,
+      "learning_rate": 3.328731965796e-05,
+      "loss": 0.244,
+      "step": 662
+    },
+    {
+      "epoch": 0.42432,
+      "grad_norm": 0.2840566587736982,
+      "learning_rate": 3.32373354907472e-05,
+      "loss": 0.2336,
+      "step": 663
+    },
+    {
+      "epoch": 0.42496,
+      "grad_norm": 0.27188917531513446,
+      "learning_rate": 3.3187314357400244e-05,
+      "loss": 0.2344,
+      "step": 664
+    },
+    {
+      "epoch": 0.4256,
+      "grad_norm": 0.265226445811272,
+      "learning_rate": 3.31372564823956e-05,
+      "loss": 0.2649,
+      "step": 665
+    },
+    {
+      "epoch": 0.42624,
+      "grad_norm": 0.28854870363229795,
+      "learning_rate": 3.308716209037462e-05,
+      "loss": 0.2538,
+      "step": 666
+    },
+    {
+      "epoch": 0.42688,
+      "grad_norm": 0.2618928575977395,
+      "learning_rate": 3.303703140614251e-05,
+      "loss": 0.2532,
+      "step": 667
+    },
+    {
+      "epoch": 0.42752,
+      "grad_norm": 0.26410396546652154,
+      "learning_rate": 3.2986864654667356e-05,
+      "loss": 0.2402,
+      "step": 668
+    },
+    {
+      "epoch": 0.42816,
+      "grad_norm": 0.24094546777729578,
+      "learning_rate": 3.293666206107911e-05,
+      "loss": 0.2516,
+      "step": 669
+    },
+    {
+      "epoch": 0.4288,
+      "grad_norm": 0.28378779340674526,
+      "learning_rate": 3.2886423850668555e-05,
+      "loss": 0.2487,
+      "step": 670
+    },
+    {
+      "epoch": 0.42944,
+      "grad_norm": 0.2800453476199491,
+      "learning_rate": 3.283615024888632e-05,
+      "loss": 0.2404,
+      "step": 671
+    },
+    {
+      "epoch": 0.43008,
+      "grad_norm": 0.28987843294031523,
+      "learning_rate": 3.278584148134185e-05,
+      "loss": 0.2677,
+      "step": 672
+    },
+    {
+      "epoch": 0.43072,
+      "grad_norm": 0.29824090133191716,
+      "learning_rate": 3.2735497773802406e-05,
+      "loss": 0.2682,
+      "step": 673
+    },
+    {
+      "epoch": 0.43136,
+      "grad_norm": 0.2718813071517211,
+      "learning_rate": 3.268511935219204e-05,
+      "loss": 0.241,
+      "step": 674
+    },
+    {
+      "epoch": 0.432,
+      "grad_norm": 0.3320650785911746,
+      "learning_rate": 3.263470644259059e-05,
+      "loss": 0.2341,
+      "step": 675
+    },
+    {
+      "epoch": 0.43264,
+      "grad_norm": 0.31100418326629736,
+      "learning_rate": 3.258425927123266e-05,
+      "loss": 0.2663,
+      "step": 676
+    },
+    {
+      "epoch": 0.43328,
+      "grad_norm": 0.2749981217880558,
+      "learning_rate": 3.253377806450664e-05,
+      "loss": 0.2485,
+      "step": 677
+    },
+    {
+      "epoch": 0.43392,
+      "grad_norm": 0.2491062102162234,
+      "learning_rate": 3.248326304895361e-05,
+      "loss": 0.2258,
+      "step": 678
+    },
+    {
+      "epoch": 0.43456,
+      "grad_norm": 0.27061105202173463,
+      "learning_rate": 3.2432714451266395e-05,
+      "loss": 0.2609,
+      "step": 679
+    },
+    {
+      "epoch": 0.4352,
+      "grad_norm": 0.2855069550319216,
+      "learning_rate": 3.238213249828853e-05,
+      "loss": 0.2582,
+      "step": 680
+    },
+    {
+      "epoch": 0.43584,
+      "grad_norm": 0.2705470532603105,
+      "learning_rate": 3.233151741701322e-05,
+      "loss": 0.2501,
+      "step": 681
+    },
+    {
+      "epoch": 0.43648,
+      "grad_norm": 0.25333939918230564,
+      "learning_rate": 3.228086943458235e-05,
+      "loss": 0.2563,
+      "step": 682
+    },
+    {
+      "epoch": 0.43712,
+      "grad_norm": 0.24493528767421877,
+      "learning_rate": 3.223018877828546e-05,
+      "loss": 0.2207,
+      "step": 683
+    },
+    {
+      "epoch": 0.43776,
+      "grad_norm": 0.27279546295316043,
+      "learning_rate": 3.2179475675558694e-05,
+      "loss": 0.2449,
+      "step": 684
+    },
+    {
+      "epoch": 0.4384,
+      "grad_norm": 0.2479820800381958,
+      "learning_rate": 3.212873035398383e-05,
+      "loss": 0.249,
+      "step": 685
+    },
+    {
+      "epoch": 0.43904,
+      "grad_norm": 0.2646928915495465,
+      "learning_rate": 3.207795304128721e-05,
+      "loss": 0.2287,
+      "step": 686
+    },
+    {
+      "epoch": 0.43968,
+      "grad_norm": 0.26738698042119885,
+      "learning_rate": 3.2027143965338765e-05,
+      "loss": 0.2483,
+      "step": 687
+    },
+    {
+      "epoch": 0.44032,
+      "grad_norm": 0.2741701894109012,
+      "learning_rate": 3.1976303354150935e-05,
+      "loss": 0.2244,
+      "step": 688
+    },
+    {
+      "epoch": 0.44096,
+      "grad_norm": 0.23871128665206856,
+      "learning_rate": 3.192543143587771e-05,
+      "loss": 0.2331,
+      "step": 689
+    },
+    {
+      "epoch": 0.4416,
+      "grad_norm": 0.24189532225791055,
+      "learning_rate": 3.187452843881357e-05,
+      "loss": 0.2476,
+      "step": 690
+    },
+    {
+      "epoch": 0.44224,
+      "grad_norm": 0.25539624756502316,
+      "learning_rate": 3.182359459139243e-05,
+      "loss": 0.2266,
+      "step": 691
+    },
+    {
+      "epoch": 0.44288,
+      "grad_norm": 0.2756480505672449,
+      "learning_rate": 3.1772630122186695e-05,
+      "loss": 0.2487,
+      "step": 692
+    },
+    {
+      "epoch": 0.44352,
+      "grad_norm": 0.2668909881136519,
+      "learning_rate": 3.172163525990617e-05,
+      "loss": 0.2646,
+      "step": 693
+    },
+    {
+      "epoch": 0.44416,
+      "grad_norm": 0.2371652011452226,
+      "learning_rate": 3.167061023339706e-05,
+      "loss": 0.2497,
+      "step": 694
+    },
+    {
+      "epoch": 0.4448,
+      "grad_norm": 0.25390001109163235,
+      "learning_rate": 3.161955527164092e-05,
+      "loss": 0.2508,
+      "step": 695
+    },
+    {
+      "epoch": 0.44544,
+      "grad_norm": 0.2402366710016075,
+      "learning_rate": 3.1568470603753666e-05,
+      "loss": 0.236,
+      "step": 696
+    },
+    {
+      "epoch": 0.44608,
+      "grad_norm": 0.2784039809880136,
+      "learning_rate": 3.1517356458984485e-05,
+      "loss": 0.2478,
+      "step": 697
+    },
+    {
+      "epoch": 0.44672,
+      "grad_norm": 0.24727119521104232,
+      "learning_rate": 3.14662130667149e-05,
+      "loss": 0.2217,
+      "step": 698
+    },
+    {
+      "epoch": 0.44736,
+      "grad_norm": 0.27458593279014226,
+      "learning_rate": 3.141504065645765e-05,
+      "loss": 0.2589,
+      "step": 699
+    },
+    {
+      "epoch": 0.448,
+      "grad_norm": 0.24446850331658782,
+      "learning_rate": 3.1363839457855704e-05,
+      "loss": 0.2475,
+      "step": 700
+    },
+    {
+      "epoch": 0.44864,
+      "grad_norm": 0.2559445330730872,
+      "learning_rate": 3.131260970068122e-05,
+      "loss": 0.2469,
+      "step": 701
+    },
+    {
+      "epoch": 0.44928,
+      "grad_norm": 0.26623620144423177,
+      "learning_rate": 3.1261351614834536e-05,
+      "loss": 0.2382,
+      "step": 702
+    },
+    {
+      "epoch": 0.44992,
+      "grad_norm": 0.22387150133617711,
+      "learning_rate": 3.121006543034307e-05,
+      "loss": 0.2601,
+      "step": 703
+    },
+    {
+      "epoch": 0.45056,
+      "grad_norm": 0.23724739980041212,
+      "learning_rate": 3.11587513773604e-05,
+      "loss": 0.222,
+      "step": 704
+    },
+    {
+      "epoch": 0.4512,
+      "grad_norm": 0.25564336027468965,
+      "learning_rate": 3.1107409686165105e-05,
+      "loss": 0.2499,
+      "step": 705
+    },
+    {
+      "epoch": 0.45184,
+      "grad_norm": 0.2530073332441141,
+      "learning_rate": 3.1056040587159856e-05,
+      "loss": 0.2266,
+      "step": 706
+    },
+    {
+      "epoch": 0.45248,
+      "grad_norm": 0.2426770783855302,
+      "learning_rate": 3.100464431087027e-05,
+      "loss": 0.235,
+      "step": 707
+    },
+    {
+      "epoch": 0.45312,
+      "grad_norm": 0.24939202296967658,
+      "learning_rate": 3.095322108794396e-05,
+      "loss": 0.2346,
+      "step": 708
+    },
+    {
+      "epoch": 0.45376,
+      "grad_norm": 0.2670752936110549,
+      "learning_rate": 3.0901771149149434e-05,
+      "loss": 0.2269,
+      "step": 709
+    },
+    {
+      "epoch": 0.4544,
+      "grad_norm": 0.24835502202214765,
+      "learning_rate": 3.0850294725375105e-05,
+      "loss": 0.245,
+      "step": 710
+    },
+    {
+      "epoch": 0.45504,
+      "grad_norm": 0.2398459646425662,
+      "learning_rate": 3.079879204762827e-05,
+      "loss": 0.2329,
+      "step": 711
+    },
+    {
+      "epoch": 0.45568,
+      "grad_norm": 0.2579451505404363,
+      "learning_rate": 3.074726334703399e-05,
+      "loss": 0.2375,
+      "step": 712
+    },
+    {
+      "epoch": 0.45632,
+      "grad_norm": 0.23992084217792944,
+      "learning_rate": 3.069570885483414e-05,
+      "loss": 0.2493,
+      "step": 713
+    },
+    {
+      "epoch": 0.45696,
+      "grad_norm": 0.26286173696168813,
+      "learning_rate": 3.064412880238636e-05,
+      "loss": 0.249,
+      "step": 714
+    },
+    {
+      "epoch": 0.4576,
+      "grad_norm": 0.26001195813349676,
+      "learning_rate": 3.059252342116293e-05,
+      "loss": 0.248,
+      "step": 715
+    },
+    {
+      "epoch": 0.45824,
+      "grad_norm": 0.2454394896408352,
+      "learning_rate": 3.054089294274985e-05,
+      "loss": 0.2461,
+      "step": 716
+    },
+    {
+      "epoch": 0.45888,
+      "grad_norm": 0.2740552510767714,
+      "learning_rate": 3.048923759884573e-05,
+      "loss": 0.2363,
+      "step": 717
+    },
+    {
+      "epoch": 0.45952,
+      "grad_norm": 0.28565496441503085,
+      "learning_rate": 3.043755762126077e-05,
+      "loss": 0.2621,
+      "step": 718
+    },
+    {
+      "epoch": 0.46016,
+      "grad_norm": 0.2558300796052746,
+      "learning_rate": 3.038585324191571e-05,
+      "loss": 0.2526,
+      "step": 719
+    },
+    {
+      "epoch": 0.4608,
+      "grad_norm": 0.2560629430046937,
+      "learning_rate": 3.0334124692840815e-05,
+      "loss": 0.2558,
+      "step": 720
+    },
+    {
+      "epoch": 0.46144,
+      "grad_norm": 0.2722189672194111,
+      "learning_rate": 3.0282372206174785e-05,
+      "loss": 0.2496,
+      "step": 721
+    },
+    {
+      "epoch": 0.46208,
+      "grad_norm": 0.3113674681390855,
+      "learning_rate": 3.023059601416376e-05,
+      "loss": 0.2493,
+      "step": 722
+    },
+    {
+      "epoch": 0.46272,
+      "grad_norm": 0.23312450422876216,
+      "learning_rate": 3.017879634916027e-05,
+      "loss": 0.2394,
+      "step": 723
+    },
+    {
+      "epoch": 0.46336,
+      "grad_norm": 0.2473881664409505,
+      "learning_rate": 3.012697344362217e-05,
+      "loss": 0.2349,
+      "step": 724
+    },
+    {
+      "epoch": 0.464,
+      "grad_norm": 0.2661149365212982,
+      "learning_rate": 3.0075127530111606e-05,
+      "loss": 0.2242,
+      "step": 725
+    },
+    {
+      "epoch": 0.46464,
+      "grad_norm": 0.2747099671427922,
+      "learning_rate": 3.0023258841294e-05,
+      "loss": 0.242,
+      "step": 726
+    },
+    {
+      "epoch": 0.46528,
+      "grad_norm": 0.257152761339917,
+      "learning_rate": 2.9971367609936945e-05,
+      "loss": 0.2522,
+      "step": 727
+    },
+    {
+      "epoch": 0.46592,
+      "grad_norm": 0.27242290443778083,
+      "learning_rate": 2.9919454068909218e-05,
+      "loss": 0.2245,
+      "step": 728
+    },
+    {
+      "epoch": 0.46656,
+      "grad_norm": 0.31224799275083753,
+      "learning_rate": 2.986751845117972e-05,
+      "loss": 0.2562,
+      "step": 729
+    },
+    {
+      "epoch": 0.4672,
+      "grad_norm": 0.2578554546888466,
+      "learning_rate": 2.981556098981641e-05,
+      "loss": 0.2805,
+      "step": 730
+    },
+    {
+      "epoch": 0.46784,
+      "grad_norm": 0.3024846857246908,
+      "learning_rate": 2.9763581917985274e-05,
+      "loss": 0.2632,
+      "step": 731
+    },
+    {
+      "epoch": 0.46848,
+      "grad_norm": 0.2824166641987713,
+      "learning_rate": 2.9711581468949295e-05,
+      "loss": 0.2579,
+      "step": 732
+    },
+    {
+      "epoch": 0.46912,
+      "grad_norm": 0.2944097917138229,
+      "learning_rate": 2.965955987606736e-05,
+      "loss": 0.2716,
+      "step": 733
+    },
+    {
+      "epoch": 0.46976,
+      "grad_norm": 0.2841807990688873,
+      "learning_rate": 2.9607517372793268e-05,
+      "loss": 0.2498,
+      "step": 734
+    },
+    {
+      "epoch": 0.4704,
+      "grad_norm": 0.3201603618025787,
+      "learning_rate": 2.9555454192674638e-05,
+      "loss": 0.2498,
+      "step": 735
+    },
+    {
+      "epoch": 0.47104,
+      "grad_norm": 0.29583253286832584,
+      "learning_rate": 2.9503370569351895e-05,
+      "loss": 0.2611,
+      "step": 736
+    },
+    {
+      "epoch": 0.47168,
+      "grad_norm": 0.28051105666181864,
+      "learning_rate": 2.9451266736557202e-05,
+      "loss": 0.2575,
+      "step": 737
+    },
+    {
+      "epoch": 0.47232,
+      "grad_norm": 0.24888601598003757,
+      "learning_rate": 2.9399142928113403e-05,
+      "loss": 0.2189,
+      "step": 738
+    },
+    {
+      "epoch": 0.47296,
+      "grad_norm": 0.27307930235300004,
+      "learning_rate": 2.9346999377932992e-05,
+      "loss": 0.2625,
+      "step": 739
+    },
+    {
+      "epoch": 0.4736,
+      "grad_norm": 0.25703656850177853,
+      "learning_rate": 2.929483632001706e-05,
+      "loss": 0.2418,
+      "step": 740
+    },
+    {
+      "epoch": 0.47424,
+      "grad_norm": 0.2746230772665275,
+      "learning_rate": 2.9242653988454237e-05,
+      "loss": 0.2436,
+      "step": 741
+    },
+    {
+      "epoch": 0.47488,
+      "grad_norm": 0.27201723917797194,
+      "learning_rate": 2.919045261741965e-05,
+      "loss": 0.2379,
+      "step": 742
+    },
+    {
+      "epoch": 0.47552,
+      "grad_norm": 0.2907921867384871,
+      "learning_rate": 2.9138232441173867e-05,
+      "loss": 0.2387,
+      "step": 743
+    },
+    {
+      "epoch": 0.47616,
+      "grad_norm": 0.2981223972195515,
+      "learning_rate": 2.9085993694061853e-05,
+      "loss": 0.2427,
+      "step": 744
+    },
+    {
+      "epoch": 0.4768,
+      "grad_norm": 0.28193295528586837,
+      "learning_rate": 2.9033736610511887e-05,
+      "loss": 0.2613,
+      "step": 745
+    },
+    {
+      "epoch": 0.47744,
+      "grad_norm": 0.2757657774306609,
+      "learning_rate": 2.898146142503456e-05,
+      "loss": 0.2627,
+      "step": 746
+    },
+    {
+      "epoch": 0.47808,
+      "grad_norm": 0.2653548783721028,
+      "learning_rate": 2.8929168372221693e-05,
+      "loss": 0.2496,
+      "step": 747
+    },
+    {
+      "epoch": 0.47872,
+      "grad_norm": 0.29977110233394766,
+      "learning_rate": 2.88768576867453e-05,
+      "loss": 0.2561,
+      "step": 748
+    },
+    {
+      "epoch": 0.47936,
+      "grad_norm": 0.24436988805469104,
+      "learning_rate": 2.8824529603356497e-05,
+      "loss": 0.2231,
+      "step": 749
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.29080555610569137,
+      "learning_rate": 2.8772184356884496e-05,
+      "loss": 0.2602,
+      "step": 750
+    },
+    {
+      "epoch": 0.48064,
+      "grad_norm": 0.2357009376696896,
+      "learning_rate": 2.8719822182235516e-05,
+      "loss": 0.2552,
+      "step": 751
+    },
+    {
+      "epoch": 0.48128,
+      "grad_norm": 0.2612000455514733,
+      "learning_rate": 2.8667443314391762e-05,
+      "loss": 0.229,
+      "step": 752
+    },
+    {
+      "epoch": 0.48192,
+      "grad_norm": 0.27847227931686186,
+      "learning_rate": 2.8615047988410337e-05,
+      "loss": 0.2431,
+      "step": 753
+    },
+    {
+      "epoch": 0.48256,
+      "grad_norm": 0.25403155179332854,
+      "learning_rate": 2.8562636439422202e-05,
+      "loss": 0.2432,
+      "step": 754
+    },
+    {
+      "epoch": 0.4832,
+      "grad_norm": 0.2667691204165141,
+      "learning_rate": 2.8510208902631135e-05,
+      "loss": 0.2509,
+      "step": 755
+    },
+    {
+      "epoch": 0.48384,
+      "grad_norm": 0.2777759815907825,
+      "learning_rate": 2.845776561331265e-05,
+      "loss": 0.253,
+      "step": 756
+    },
+    {
+      "epoch": 0.48448,
+      "grad_norm": 0.2740727680932021,
+      "learning_rate": 2.8405306806812947e-05,
+      "loss": 0.224,
+      "step": 757
+    },
+    {
+      "epoch": 0.48512,
+      "grad_norm": 0.2817937944167678,
+      "learning_rate": 2.835283271854788e-05,
+      "loss": 0.2628,
+      "step": 758
+    },
+    {
+      "epoch": 0.48576,
+      "grad_norm": 0.24766172176538703,
+      "learning_rate": 2.8300343584001853e-05,
+      "loss": 0.2456,
+      "step": 759
+    },
+    {
+      "epoch": 0.4864,
+      "grad_norm": 0.24618641032586008,
+      "learning_rate": 2.8247839638726825e-05,
+      "loss": 0.2587,
+      "step": 760
+    },
+    {
+      "epoch": 0.48704,
+      "grad_norm": 0.2601684998720227,
+      "learning_rate": 2.8195321118341204e-05,
+      "loss": 0.2397,
+      "step": 761
+    },
+    {
+      "epoch": 0.48768,
+      "grad_norm": 0.25920726062087374,
+      "learning_rate": 2.8142788258528813e-05,
+      "loss": 0.2572,
+      "step": 762
+    },
+    {
+      "epoch": 0.48832,
+      "grad_norm": 0.2577373926665209,
+      "learning_rate": 2.8090241295037804e-05,
+      "loss": 0.253,
+      "step": 763
+    },
+    {
+      "epoch": 0.48896,
+      "grad_norm": 0.2769909768648415,
+      "learning_rate": 2.8037680463679648e-05,
+      "loss": 0.2477,
+      "step": 764
+    },
+    {
+      "epoch": 0.4896,
+      "grad_norm": 0.25021183963273425,
+      "learning_rate": 2.7985106000328033e-05,
+      "loss": 0.2399,
+      "step": 765
+    },
+    {
+      "epoch": 0.49024,
+      "grad_norm": 0.2751049696728052,
+      "learning_rate": 2.793251814091782e-05,
+      "loss": 0.2365,
+      "step": 766
+    },
+    {
+      "epoch": 0.49088,
+      "grad_norm": 0.2785539736743758,
+      "learning_rate": 2.787991712144401e-05,
+      "loss": 0.2599,
+      "step": 767
+    },
+    {
+      "epoch": 0.49152,
+      "grad_norm": 0.25754493785695,
+      "learning_rate": 2.7827303177960638e-05,
+      "loss": 0.2654,
+      "step": 768
+    },
+    {
+      "epoch": 0.49216,
+      "grad_norm": 0.2977071861983136,
+      "learning_rate": 2.7774676546579737e-05,
+      "loss": 0.2513,
+      "step": 769
+    },
+    {
+      "epoch": 0.4928,
+      "grad_norm": 0.2627055142139323,
+      "learning_rate": 2.7722037463470297e-05,
+      "loss": 0.2314,
+      "step": 770
+    },
+    {
+      "epoch": 0.49344,
+      "grad_norm": 0.24348337022176492,
+      "learning_rate": 2.7669386164857174e-05,
+      "loss": 0.2159,
+      "step": 771
+    },
+    {
+      "epoch": 0.49408,
+      "grad_norm": 0.2567881766174943,
+      "learning_rate": 2.7616722887020034e-05,
+      "loss": 0.2309,
+      "step": 772
+    },
+    {
+      "epoch": 0.49472,
+      "grad_norm": 0.2960178723617407,
+      "learning_rate": 2.7564047866292336e-05,
+      "loss": 0.2348,
+      "step": 773
+    },
+    {
+      "epoch": 0.49536,
+      "grad_norm": 0.2573058308269504,
+      "learning_rate": 2.7511361339060193e-05,
+      "loss": 0.2578,
+      "step": 774
+    },
+    {
+      "epoch": 0.496,
+      "grad_norm": 0.23934502760321336,
+      "learning_rate": 2.7458663541761376e-05,
+      "loss": 0.2539,
+      "step": 775
+    },
+    {
+      "epoch": 0.49664,
+      "grad_norm": 0.2559326444177946,
+      "learning_rate": 2.740595471088424e-05,
+      "loss": 0.233,
+      "step": 776
+    },
+    {
+      "epoch": 0.49728,
+      "grad_norm": 0.29032921794905275,
+      "learning_rate": 2.735323508296664e-05,
+      "loss": 0.2542,
+      "step": 777
+    },
+    {
+      "epoch": 0.49792,
+      "grad_norm": 0.2575784084305343,
+      "learning_rate": 2.7300504894594892e-05,
+      "loss": 0.242,
+      "step": 778
+    },
+    {
+      "epoch": 0.49856,
+      "grad_norm": 0.2448048925081302,
+      "learning_rate": 2.7247764382402696e-05,
+      "loss": 0.2691,
+      "step": 779
+    },
+    {
+      "epoch": 0.4992,
+      "grad_norm": 0.25496846358680036,
+      "learning_rate": 2.7195013783070093e-05,
+      "loss": 0.2622,
+      "step": 780
+    },
+    {
+      "epoch": 0.49984,
+      "grad_norm": 0.2645276293483241,
+      "learning_rate": 2.7142253333322387e-05,
+      "loss": 0.2483,
+      "step": 781
+    },
+    {
+      "epoch": 0.50048,
+      "grad_norm": 0.24891173303191308,
+      "learning_rate": 2.7089483269929074e-05,
+      "loss": 0.2511,
+      "step": 782
+    },
+    {
+      "epoch": 0.50112,
+      "grad_norm": 0.24013450456492896,
+      "learning_rate": 2.703670382970281e-05,
+      "loss": 0.2351,
+      "step": 783
+    },
+    {
+      "epoch": 0.50176,
+      "grad_norm": 0.25116583720506236,
+      "learning_rate": 2.698391524949832e-05,
+      "loss": 0.2396,
+      "step": 784
+    },
+    {
+      "epoch": 0.5024,
+      "grad_norm": 0.2613555823733014,
+      "learning_rate": 2.693111776621136e-05,
+      "loss": 0.2401,
+      "step": 785
+    },
+    {
+      "epoch": 0.50304,
+      "grad_norm": 0.23467219955674096,
+      "learning_rate": 2.6878311616777618e-05,
+      "loss": 0.246,
+      "step": 786
+    },
+    {
+      "epoch": 0.50368,
+      "grad_norm": 0.2725254268140766,
+      "learning_rate": 2.68254970381717e-05,
+      "loss": 0.2439,
+      "step": 787
+    },
+    {
+      "epoch": 0.50432,
+      "grad_norm": 0.24774235023130026,
+      "learning_rate": 2.6772674267406005e-05,
+      "loss": 0.263,
+      "step": 788
+    },
+    {
+      "epoch": 0.50496,
+      "grad_norm": 0.23891005756843905,
+      "learning_rate": 2.6719843541529727e-05,
+      "loss": 0.2536,
+      "step": 789
+    },
+    {
+      "epoch": 0.5056,
+      "grad_norm": 0.27768499138453195,
+      "learning_rate": 2.666700509762774e-05,
+      "loss": 0.234,
+      "step": 790
+    },
+    {
+      "epoch": 0.50624,
+      "grad_norm": 0.29215190403296315,
+      "learning_rate": 2.6614159172819564e-05,
+      "loss": 0.2513,
+      "step": 791
+    },
+    {
+      "epoch": 0.50688,
+      "grad_norm": 0.2614573878416528,
+      "learning_rate": 2.6561306004258284e-05,
+      "loss": 0.2529,
+      "step": 792
+    },
+    {
+      "epoch": 0.50752,
+      "grad_norm": 0.25761339968627844,
+      "learning_rate": 2.6508445829129498e-05,
+      "loss": 0.256,
+      "step": 793
+    },
+    {
+      "epoch": 0.50816,
+      "grad_norm": 0.2668111421697042,
+      "learning_rate": 2.645557888465024e-05,
+      "loss": 0.2274,
+      "step": 794
+    },
+    {
+      "epoch": 0.5088,
+      "grad_norm": 0.27451929815166287,
+      "learning_rate": 2.640270540806793e-05,
+      "loss": 0.231,
+      "step": 795
+    },
+    {
+      "epoch": 0.50944,
+      "grad_norm": 0.28276713584045393,
+      "learning_rate": 2.6349825636659292e-05,
+      "loss": 0.2737,
+      "step": 796
+    },
+    {
+      "epoch": 0.51008,
+      "grad_norm": 0.25996717470137165,
+      "learning_rate": 2.629693980772931e-05,
+      "loss": 0.2386,
+      "step": 797
+    },
+    {
+      "epoch": 0.51072,
+      "grad_norm": 0.24933734326173174,
+      "learning_rate": 2.6244048158610148e-05,
+      "loss": 0.2514,
+      "step": 798
+    },
+    {
+      "epoch": 0.51136,
+      "grad_norm": 0.2713739147643552,
+      "learning_rate": 2.619115092666008e-05,
+      "loss": 0.2312,
+      "step": 799
+    },
+    {
+      "epoch": 0.512,
+      "grad_norm": 0.2853248485041968,
+      "learning_rate": 2.6138248349262446e-05,
+      "loss": 0.2464,
+      "step": 800
+    },
+    {
+      "epoch": 0.51264,
+      "grad_norm": 0.2769176496798977,
+      "learning_rate": 2.608534066382457e-05,
+      "loss": 0.2575,
+      "step": 801
+    },
+    {
+      "epoch": 0.51328,
+      "grad_norm": 0.25550021777275495,
+      "learning_rate": 2.60324281077767e-05,
+      "loss": 0.2296,
+      "step": 802
+    },
+    {
+      "epoch": 0.51392,
+      "grad_norm": 0.2717710452679654,
+      "learning_rate": 2.5979510918570943e-05,
+      "loss": 0.228,
+      "step": 803
+    },
+    {
+      "epoch": 0.51456,
+      "grad_norm": 0.2855809708839765,
+      "learning_rate": 2.592658933368019e-05,
+      "loss": 0.2397,
+      "step": 804
+    },
+    {
+      "epoch": 0.5152,
+      "grad_norm": 0.25730499536702794,
+      "learning_rate": 2.5873663590597065e-05,
+      "loss": 0.2218,
+      "step": 805
+    },
+    {
+      "epoch": 0.51584,
+      "grad_norm": 0.24816491362061666,
+      "learning_rate": 2.582073392683284e-05,
+      "loss": 0.2349,
+      "step": 806
+    },
+    {
+      "epoch": 0.51648,
+      "grad_norm": 0.2811332685753554,
+      "learning_rate": 2.5767800579916412e-05,
+      "loss": 0.2319,
+      "step": 807
+    },
+    {
+      "epoch": 0.51712,
+      "grad_norm": 0.2826697719356305,
+      "learning_rate": 2.5714863787393173e-05,
+      "loss": 0.2409,
+      "step": 808
+    },
+    {
+      "epoch": 0.51776,
+      "grad_norm": 0.27440667075281816,
+      "learning_rate": 2.5661923786823994e-05,
+      "loss": 0.2169,
+      "step": 809
+    },
+    {
+      "epoch": 0.5184,
+      "grad_norm": 0.3464183569952111,
+      "learning_rate": 2.5608980815784144e-05,
+      "loss": 0.2471,
+      "step": 810
+    },
+    {
+      "epoch": 0.51904,
+      "grad_norm": 0.23556464983790443,
+      "learning_rate": 2.5556035111862213e-05,
+      "loss": 0.2442,
+      "step": 811
+    },
+    {
+      "epoch": 0.51968,
+      "grad_norm": 0.27365035067332283,
+      "learning_rate": 2.550308691265907e-05,
+      "loss": 0.2549,
+      "step": 812
+    },
+    {
+      "epoch": 0.52032,
+      "grad_norm": 0.2870134081115856,
+      "learning_rate": 2.545013645578676e-05,
+      "loss": 0.258,
+      "step": 813
+    },
+    {
+      "epoch": 0.52096,
+      "grad_norm": 0.2556368997216433,
+      "learning_rate": 2.5397183978867484e-05,
+      "loss": 0.2387,
+      "step": 814
+    },
+    {
+      "epoch": 0.5216,
+      "grad_norm": 0.25964279902581394,
+      "learning_rate": 2.5344229719532486e-05,
+      "loss": 0.2399,
+      "step": 815
+    },
+    {
+      "epoch": 0.52224,
+      "grad_norm": 0.26998265791053805,
+      "learning_rate": 2.529127391542103e-05,
+      "loss": 0.2245,
+      "step": 816
+    },
+    {
+      "epoch": 0.52288,
+      "grad_norm": 0.2843610345936838,
+      "learning_rate": 2.5238316804179296e-05,
+      "loss": 0.2449,
+      "step": 817
+    },
+    {
+      "epoch": 0.52352,
+      "grad_norm": 0.31442295392273417,
+      "learning_rate": 2.5185358623459332e-05,
+      "loss": 0.2356,
+      "step": 818
+    },
+    {
+      "epoch": 0.52416,
+      "grad_norm": 0.2516186869896401,
+      "learning_rate": 2.513239961091799e-05,
+      "loss": 0.2163,
+      "step": 819
+    },
+    {
+      "epoch": 0.5248,
+      "grad_norm": 0.25299650171502486,
+      "learning_rate": 2.507944000421586e-05,
+      "loss": 0.2255,
+      "step": 820
+    },
+    {
+      "epoch": 0.52544,
+      "grad_norm": 0.2497383419095203,
+      "learning_rate": 2.502648004101619e-05,
+      "loss": 0.2346,
+      "step": 821
+    },
+    {
+      "epoch": 0.52608,
+      "grad_norm": 0.2899381319524661,
+      "learning_rate": 2.497351995898382e-05,
+      "loss": 0.2387,
+      "step": 822
+    },
+    {
+      "epoch": 0.52672,
+      "grad_norm": 0.25184447023613826,
+      "learning_rate": 2.492055999578415e-05,
+      "loss": 0.2458,
+      "step": 823
+    },
+    {
+      "epoch": 0.52736,
+      "grad_norm": 0.265296832898281,
+      "learning_rate": 2.486760038908201e-05,
+      "loss": 0.2356,
+      "step": 824
+    },
+    {
+      "epoch": 0.528,
+      "grad_norm": 0.25652770276343023,
+      "learning_rate": 2.481464137654068e-05,
+      "loss": 0.257,
+      "step": 825
+    },
+    {
+      "epoch": 0.52864,
+      "grad_norm": 0.24628391207970465,
+      "learning_rate": 2.476168319582071e-05,
+      "loss": 0.2269,
+      "step": 826
+    },
+    {
+      "epoch": 0.52928,
+      "grad_norm": 0.272941022628505,
+      "learning_rate": 2.4708726084578978e-05,
+      "loss": 0.2361,
+      "step": 827
+    },
+    {
+      "epoch": 0.52992,
+      "grad_norm": 0.25824476732724294,
+      "learning_rate": 2.465577028046752e-05,
+      "loss": 0.2424,
+      "step": 828
+    },
+    {
+      "epoch": 0.53056,
+      "grad_norm": 0.2560477063420002,
+      "learning_rate": 2.4602816021132515e-05,
+      "loss": 0.229,
+      "step": 829
+    },
+    {
+      "epoch": 0.5312,
+      "grad_norm": 0.24996724858142452,
+      "learning_rate": 2.4549863544213243e-05,
+      "loss": 0.2409,
+      "step": 830
+    },
+    {
+      "epoch": 0.53184,
+      "grad_norm": 0.26937230634068454,
+      "learning_rate": 2.449691308734093e-05,
+      "loss": 0.2717,
+      "step": 831
+    },
+    {
+      "epoch": 0.53248,
+      "grad_norm": 0.25399722766750077,
+      "learning_rate": 2.444396488813779e-05,
+      "loss": 0.2458,
+      "step": 832
+    },
+    {
+      "epoch": 0.53312,
+      "grad_norm": 0.25073073586853056,
+      "learning_rate": 2.4391019184215862e-05,
+      "loss": 0.287,
+      "step": 833
+    },
+    {
+      "epoch": 0.53376,
+      "grad_norm": 0.2668798130737852,
+      "learning_rate": 2.4338076213176012e-05,
+      "loss": 0.2379,
+      "step": 834
+    },
+    {
+      "epoch": 0.5344,
+      "grad_norm": 0.2694419392170376,
+      "learning_rate": 2.4285136212606836e-05,
+      "loss": 0.2242,
+      "step": 835
+    },
+    {
+      "epoch": 0.53504,
+      "grad_norm": 0.2514952521033137,
+      "learning_rate": 2.42321994200836e-05,
+      "loss": 0.2309,
+      "step": 836
+    },
+    {
+      "epoch": 0.53568,
+      "grad_norm": 0.26883353919312936,
+      "learning_rate": 2.417926607316716e-05,
+      "loss": 0.2444,
+      "step": 837
+    },
+    {
+      "epoch": 0.53632,
+      "grad_norm": 0.25407436256439836,
+      "learning_rate": 2.412633640940295e-05,
+      "loss": 0.2249,
+      "step": 838
+    },
+    {
+      "epoch": 0.53696,
+      "grad_norm": 0.2471028479062988,
+      "learning_rate": 2.4073410666319816e-05,
+      "loss": 0.2478,
+      "step": 839
+    },
+    {
+      "epoch": 0.5376,
+      "grad_norm": 0.24904978864126115,
+      "learning_rate": 2.4020489081429063e-05,
+      "loss": 0.2556,
+      "step": 840
+    },
+    {
+      "epoch": 0.53824,
+      "grad_norm": 0.24958463538150766,
+      "learning_rate": 2.3967571892223305e-05,
+      "loss": 0.2236,
+      "step": 841
+    },
+    {
+      "epoch": 0.53888,
+      "grad_norm": 0.22568255136010584,
+      "learning_rate": 2.3914659336175428e-05,
+      "loss": 0.2284,
+      "step": 842
+    },
+    {
+      "epoch": 0.53952,
+      "grad_norm": 0.2680013212335924,
+      "learning_rate": 2.386175165073756e-05,
+      "loss": 0.2299,
+      "step": 843
+    },
+    {
+      "epoch": 0.54016,
+      "grad_norm": 0.2513863790200001,
+      "learning_rate": 2.3808849073339923e-05,
+      "loss": 0.2371,
+      "step": 844
+    },
+    {
+      "epoch": 0.5408,
+      "grad_norm": 0.282899676416669,
+      "learning_rate": 2.375595184138986e-05,
+      "loss": 0.241,
+      "step": 845
+    },
+    {
+      "epoch": 0.54144,
+      "grad_norm": 0.2244467766555148,
+      "learning_rate": 2.3703060192270696e-05,
+      "loss": 0.2317,
+      "step": 846
+    },
+    {
+      "epoch": 0.54208,
+      "grad_norm": 0.3412894508050135,
+      "learning_rate": 2.365017436334072e-05,
+      "loss": 0.2308,
+      "step": 847
+    },
+    {
+      "epoch": 0.54272,
+      "grad_norm": 0.24859724986121393,
+      "learning_rate": 2.3597294591932077e-05,
+      "loss": 0.2319,
+      "step": 848
+    },
+    {
+      "epoch": 0.54336,
+      "grad_norm": 0.26058887718049173,
+      "learning_rate": 2.3544421115349772e-05,
+      "loss": 0.2304,
+      "step": 849
+    },
+    {
+      "epoch": 0.544,
+      "grad_norm": 0.24490556556432685,
+      "learning_rate": 2.3491554170870504e-05,
+      "loss": 0.2131,
+      "step": 850
+    },
+    {
+      "epoch": 0.54464,
+      "grad_norm": 0.2742245469914794,
+      "learning_rate": 2.343869399574172e-05,
+      "loss": 0.2335,
+      "step": 851
+    },
+    {
+      "epoch": 0.54528,
+      "grad_norm": 0.26604566820064934,
+      "learning_rate": 2.3385840827180445e-05,
+      "loss": 0.2399,
+      "step": 852
+    },
+    {
+      "epoch": 0.54592,
+      "grad_norm": 0.2935391865349877,
+      "learning_rate": 2.333299490237226e-05,
+      "loss": 0.2509,
+      "step": 853
+    },
+    {
+      "epoch": 0.54656,
+      "grad_norm": 0.2658590860493031,
+      "learning_rate": 2.3280156458470282e-05,
+      "loss": 0.2398,
+      "step": 854
+    },
+    {
+      "epoch": 0.5472,
+      "grad_norm": 0.2519142285543896,
+      "learning_rate": 2.3227325732593997e-05,
+      "loss": 0.2343,
+      "step": 855
+    },
+    {
+      "epoch": 0.54784,
+      "grad_norm": 0.24935913440924726,
+      "learning_rate": 2.317450296182831e-05,
+      "loss": 0.2521,
+      "step": 856
+    },
+    {
+      "epoch": 0.54848,
+      "grad_norm": 0.27273844122511376,
+      "learning_rate": 2.3121688383222384e-05,
+      "loss": 0.2264,
+      "step": 857
+    },
+    {
+      "epoch": 0.54912,
+      "grad_norm": 0.2637618231471123,
+      "learning_rate": 2.306888223378865e-05,
+      "loss": 0.2459,
+      "step": 858
+    },
+    {
+      "epoch": 0.54976,
+      "grad_norm": 0.24626052284155567,
+      "learning_rate": 2.3016084750501685e-05,
+      "loss": 0.2498,
+      "step": 859
+    },
+    {
+      "epoch": 0.5504,
+      "grad_norm": 0.2577670221498842,
+      "learning_rate": 2.2963296170297204e-05,
+      "loss": 0.2426,
+      "step": 860
+    },
+    {
+      "epoch": 0.55104,
+      "grad_norm": 0.2531452855233776,
+      "learning_rate": 2.2910516730070932e-05,
+      "loss": 0.2507,
+      "step": 861
+    },
+    {
+      "epoch": 0.55168,
+      "grad_norm": 0.2504925026124407,
+      "learning_rate": 2.2857746666677626e-05,
+      "loss": 0.2122,
+      "step": 862
+    },
+    {
+      "epoch": 0.55232,
+      "grad_norm": 0.24041491357324582,
+      "learning_rate": 2.280498621692991e-05,
+      "loss": 0.2337,
+      "step": 863
+    },
+    {
+      "epoch": 0.55296,
+      "grad_norm": 0.2546756714997248,
+      "learning_rate": 2.275223561759731e-05,
+      "loss": 0.2467,
+      "step": 864
+    },
+    {
+      "epoch": 0.5536,
+      "grad_norm": 0.2653173276499359,
+      "learning_rate": 2.2699495105405117e-05,
+      "loss": 0.241,
+      "step": 865
+    },
+    {
+      "epoch": 0.55424,
+      "grad_norm": 0.2659857425961046,
+      "learning_rate": 2.264676491703336e-05,
+      "loss": 0.2281,
+      "step": 866
+    },
+    {
+      "epoch": 0.55488,
+      "grad_norm": 0.2769155487501475,
+      "learning_rate": 2.2594045289115764e-05,
+      "loss": 0.226,
+      "step": 867
+    },
+    {
+      "epoch": 0.55552,
+      "grad_norm": 0.24517901275586215,
+      "learning_rate": 2.254133645823862e-05,
+      "loss": 0.2323,
+      "step": 868
+    },
+    {
+      "epoch": 0.55616,
+      "grad_norm": 0.24520820932686727,
+      "learning_rate": 2.2488638660939816e-05,
+      "loss": 0.2502,
+      "step": 869
+    },
+    {
+      "epoch": 0.5568,
+      "grad_norm": 0.2630195872229611,
+      "learning_rate": 2.243595213370767e-05,
+      "loss": 0.2351,
+      "step": 870
+    },
+    {
+      "epoch": 0.55744,
+      "grad_norm": 0.25845227070967053,
+      "learning_rate": 2.238327711297997e-05,
+      "loss": 0.2462,
+      "step": 871
+    },
+    {
+      "epoch": 0.55808,
+      "grad_norm": 0.2407775804369694,
+      "learning_rate": 2.2330613835142832e-05,
+      "loss": 0.2401,
+      "step": 872
+    },
+    {
+      "epoch": 0.55872,
+      "grad_norm": 0.25636378428837236,
+      "learning_rate": 2.2277962536529712e-05,
+      "loss": 0.2146,
+      "step": 873
+    },
+    {
+      "epoch": 0.55936,
+      "grad_norm": 0.2599169204293122,
+      "learning_rate": 2.2225323453420265e-05,
+      "loss": 0.221,
+      "step": 874
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.262003731673188,
+      "learning_rate": 2.2172696822039374e-05,
+      "loss": 0.2396,
+      "step": 875
+    },
+    {
+      "epoch": 0.56064,
+      "grad_norm": 0.22044573742479154,
+      "learning_rate": 2.2120082878556e-05,
+      "loss": 0.2201,
+      "step": 876
+    },
+    {
+      "epoch": 0.56128,
+      "grad_norm": 0.28212886056666625,
+      "learning_rate": 2.206748185908219e-05,
+      "loss": 0.2179,
+      "step": 877
+    },
+    {
+      "epoch": 0.56192,
+      "grad_norm": 0.2543660609179219,
+      "learning_rate": 2.2014893999671976e-05,
+      "loss": 0.2529,
+      "step": 878
+    },
+    {
+      "epoch": 0.56256,
+      "grad_norm": 0.28114291350073906,
+      "learning_rate": 2.1962319536320354e-05,
+      "loss": 0.2444,
+      "step": 879
+    },
+    {
+      "epoch": 0.5632,
+      "grad_norm": 0.24835237823983541,
+      "learning_rate": 2.1909758704962198e-05,
+      "loss": 0.2156,
+      "step": 880
+    },
+    {
+      "epoch": 0.56384,
+      "grad_norm": 0.25599405623703164,
+      "learning_rate": 2.185721174147119e-05,
+      "loss": 0.228,
+      "step": 881
+    },
+    {
+      "epoch": 0.56448,
+      "grad_norm": 0.26709324375032384,
+      "learning_rate": 2.1804678881658798e-05,
+      "loss": 0.231,
+      "step": 882
+    },
+    {
+      "epoch": 0.56512,
+      "grad_norm": 0.26194312329216146,
+      "learning_rate": 2.1752160361273174e-05,
+      "loss": 0.2294,
+      "step": 883
+    },
+    {
+      "epoch": 0.56576,
+      "grad_norm": 0.4306300503220103,
+      "learning_rate": 2.1699656415998153e-05,
+      "loss": 0.2264,
+      "step": 884
+    },
+    {
+      "epoch": 0.5664,
+      "grad_norm": 0.29613259757712557,
+      "learning_rate": 2.164716728145213e-05,
+      "loss": 0.2566,
+      "step": 885
+    },
+    {
+      "epoch": 0.56704,
+      "grad_norm": 0.24662149693794833,
+      "learning_rate": 2.1594693193187062e-05,
+      "loss": 0.2343,
+      "step": 886
+    },
+    {
+      "epoch": 0.56768,
+      "grad_norm": 0.25552195497626806,
+      "learning_rate": 2.1542234386687354e-05,
+      "loss": 0.2432,
+      "step": 887
+    },
+    {
+      "epoch": 0.56832,
+      "grad_norm": 0.251813550789662,
+      "learning_rate": 2.148979109736887e-05,
+      "loss": 0.2556,
+      "step": 888
+    },
+    {
+      "epoch": 0.56896,
+      "grad_norm": 0.2540482580395456,
+      "learning_rate": 2.1437363560577803e-05,
+      "loss": 0.2381,
+      "step": 889
+    },
+    {
+      "epoch": 0.5696,
+      "grad_norm": 0.24111845750526342,
+      "learning_rate": 2.1384952011589675e-05,
+      "loss": 0.2524,
+      "step": 890
+    },
+    {
+      "epoch": 0.57024,
+      "grad_norm": 0.24398046866887468,
+      "learning_rate": 2.1332556685608243e-05,
+      "loss": 0.2517,
+      "step": 891
+    },
+    {
+      "epoch": 0.57088,
+      "grad_norm": 0.22752053201164787,
+      "learning_rate": 2.128017781776448e-05,
+      "loss": 0.2309,
+      "step": 892
+    },
+    {
+      "epoch": 0.57152,
+      "grad_norm": 0.2355688865970114,
+      "learning_rate": 2.122781564311551e-05,
+      "loss": 0.2286,
+      "step": 893
+    },
+    {
+      "epoch": 0.57216,
+      "grad_norm": 0.30704592466341735,
+      "learning_rate": 2.1175470396643505e-05,
+      "loss": 0.235,
+      "step": 894
+    },
+    {
+      "epoch": 0.5728,
+      "grad_norm": 0.2521695242515918,
+      "learning_rate": 2.1123142313254708e-05,
+      "loss": 0.2399,
+      "step": 895
+    },
+    {
+      "epoch": 0.57344,
+      "grad_norm": 0.2517901942950135,
+      "learning_rate": 2.1070831627778303e-05,
+      "loss": 0.251,
+      "step": 896
+    },
+    {
+      "epoch": 0.57408,
+      "grad_norm": 0.2686542226900506,
+      "learning_rate": 2.101853857496545e-05,
+      "loss": 0.2424,
+      "step": 897
+    },
+    {
+      "epoch": 0.57472,
+      "grad_norm": 0.23956702115744746,
+      "learning_rate": 2.0966263389488123e-05,
+      "loss": 0.2337,
+      "step": 898
+    },
+    {
+      "epoch": 0.57536,
+      "grad_norm": 0.2698696392827644,
+      "learning_rate": 2.091400630593816e-05,
+      "loss": 0.211,
+      "step": 899
+    },
+    {
+      "epoch": 0.576,
+      "grad_norm": 0.27164963285710964,
+      "learning_rate": 2.086176755882614e-05,
+      "loss": 0.2049,
+      "step": 900
+    },
+    {
+      "epoch": 0.57664,
+      "grad_norm": 0.3010728945004748,
+      "learning_rate": 2.0809547382580357e-05,
+      "loss": 0.249,
+      "step": 901
+    },
+    {
+      "epoch": 0.57728,
+      "grad_norm": 0.24248362744993088,
+      "learning_rate": 2.075734601154577e-05,
+      "loss": 0.2292,
+      "step": 902
+    },
+    {
+      "epoch": 0.57792,
+      "grad_norm": 0.26337775614858755,
+      "learning_rate": 2.070516367998294e-05,
+      "loss": 0.221,
+      "step": 903
+    },
+    {
+      "epoch": 0.57856,
+      "grad_norm": 0.2762119569843419,
+      "learning_rate": 2.065300062206701e-05,
+      "loss": 0.2399,
+      "step": 904
+    },
+    {
+      "epoch": 0.5792,
+      "grad_norm": 0.2502944790586434,
+      "learning_rate": 2.06008570718866e-05,
+      "loss": 0.2261,
+      "step": 905
+    },
+    {
+      "epoch": 0.57984,
+      "grad_norm": 0.24994600132296746,
+      "learning_rate": 2.0548733263442804e-05,
+      "loss": 0.2318,
+      "step": 906
+    },
+    {
+      "epoch": 0.58048,
+      "grad_norm": 0.2646229850813451,
+      "learning_rate": 2.04966294306481e-05,
+      "loss": 0.242,
+      "step": 907
+    },
+    {
+      "epoch": 0.58112,
+      "grad_norm": 0.23630446548881304,
+      "learning_rate": 2.0444545807325364e-05,
+      "loss": 0.2382,
+      "step": 908
+    },
+    {
+      "epoch": 0.58176,
+      "grad_norm": 0.2763693183478354,
+      "learning_rate": 2.0392482627206734e-05,
+      "loss": 0.2456,
+      "step": 909
+    },
+    {
+      "epoch": 0.5824,
+      "grad_norm": 0.23271232904672204,
+      "learning_rate": 2.034044012393265e-05,
+      "loss": 0.228,
+      "step": 910
+    },
+    {
+      "epoch": 0.58304,
+      "grad_norm": 0.2619212734189822,
+      "learning_rate": 2.028841853105071e-05,
+      "loss": 0.2173,
+      "step": 911
+    },
+    {
+      "epoch": 0.58368,
+      "grad_norm": 0.2343186878321288,
+      "learning_rate": 2.0236418082014728e-05,
+      "loss": 0.2312,
+      "step": 912
+    },
+    {
+      "epoch": 0.58432,
+      "grad_norm": 0.2621287675096788,
+      "learning_rate": 2.01844390101836e-05,
+      "loss": 0.243,
+      "step": 913
+    },
+    {
+      "epoch": 0.58496,
+      "grad_norm": 0.24394920166442188,
+      "learning_rate": 2.013248154882029e-05,
+      "loss": 0.2376,
+      "step": 914
+    },
+    {
+      "epoch": 0.5856,
+      "grad_norm": 0.2630255512164301,
+      "learning_rate": 2.0080545931090784e-05,
+      "loss": 0.2507,
+      "step": 915
+    },
+    {
+      "epoch": 0.58624,
+      "grad_norm": 0.4204887892501008,
+      "learning_rate": 2.0028632390063057e-05,
+      "loss": 0.257,
+      "step": 916
+    },
+    {
+      "epoch": 0.58688,
+      "grad_norm": 0.2367459404403816,
+      "learning_rate": 1.9976741158706002e-05,
+      "loss": 0.2233,
+      "step": 917
+    },
+    {
+      "epoch": 0.58752,
+      "grad_norm": 0.2503120076484051,
+      "learning_rate": 1.992487246988839e-05,
+      "loss": 0.2424,
+      "step": 918
+    },
+    {
+      "epoch": 0.58816,
+      "grad_norm": 0.24367871271764782,
+      "learning_rate": 1.9873026556377835e-05,
+      "loss": 0.2545,
+      "step": 919
+    },
+    {
+      "epoch": 0.5888,
+      "grad_norm": 0.27402164758506486,
+      "learning_rate": 1.9821203650839728e-05,
+      "loss": 0.2507,
+      "step": 920
+    },
+    {
+      "epoch": 0.58944,
+      "grad_norm": 0.25175154095139074,
+      "learning_rate": 1.9769403985836242e-05,
+      "loss": 0.2193,
+      "step": 921
+    },
+    {
+      "epoch": 0.59008,
+      "grad_norm": 0.21651552672731597,
+      "learning_rate": 1.971762779382522e-05,
+      "loss": 0.2435,
+      "step": 922
+    },
+    {
+      "epoch": 0.59072,
+      "grad_norm": 0.23741702650241553,
+      "learning_rate": 1.9665875307159194e-05,
+      "loss": 0.2376,
+      "step": 923
+    },
+    {
+      "epoch": 0.59136,
+      "grad_norm": 0.2799239026177196,
+      "learning_rate": 1.961414675808429e-05,
+      "loss": 0.2168,
+      "step": 924
+    },
+    {
+      "epoch": 0.592,
+      "grad_norm": 0.24757148011984073,
+      "learning_rate": 1.956244237873924e-05,
+      "loss": 0.2325,
+      "step": 925
+    },
+    {
+      "epoch": 0.59264,
+      "grad_norm": 0.23945061682371735,
+      "learning_rate": 1.9510762401154274e-05,
+      "loss": 0.2262,
+      "step": 926
+    },
+    {
+      "epoch": 0.59328,
+      "grad_norm": 0.2557150226996925,
+      "learning_rate": 1.945910705725016e-05,
+      "loss": 0.2088,
+      "step": 927
+    },
+    {
+      "epoch": 0.59392,
+      "grad_norm": 0.23522549083096198,
+      "learning_rate": 1.940747657883708e-05,
+      "loss": 0.2083,
+      "step": 928
+    },
+    {
+      "epoch": 0.59456,
+      "grad_norm": 0.22758182784177505,
+      "learning_rate": 1.9355871197613647e-05,
+      "loss": 0.226,
+      "step": 929
+    },
+    {
+      "epoch": 0.5952,
+      "grad_norm": 0.2511880619456736,
+      "learning_rate": 1.9304291145165858e-05,
+      "loss": 0.2414,
+      "step": 930
+    },
+    {
+      "epoch": 0.59584,
+      "grad_norm": 0.23320426421854681,
+      "learning_rate": 1.9252736652966015e-05,
+      "loss": 0.2099,
+      "step": 931
+    },
+    {
+      "epoch": 0.59648,
+      "grad_norm": 0.2595547756610221,
+      "learning_rate": 1.920120795237174e-05,
+      "loss": 0.2442,
+      "step": 932
+    },
+    {
+      "epoch": 0.59712,
+      "grad_norm": 0.26463995276945596,
+      "learning_rate": 1.914970527462489e-05,
+      "loss": 0.2181,
+      "step": 933
+    },
+    {
+      "epoch": 0.59776,
+      "grad_norm": 0.238788378499657,
+      "learning_rate": 1.909822885085058e-05,
+      "loss": 0.2397,
+      "step": 934
+    },
+    {
+      "epoch": 0.5984,
+      "grad_norm": 0.24848561435523608,
+      "learning_rate": 1.9046778912056047e-05,
+      "loss": 0.2162,
+      "step": 935
+    },
+    {
+      "epoch": 0.59904,
+      "grad_norm": 0.2723669206273018,
+      "learning_rate": 1.8995355689129734e-05,
+      "loss": 0.2403,
+      "step": 936
+    },
+    {
+      "epoch": 0.59968,
+      "grad_norm": 0.2480462005432638,
+      "learning_rate": 1.894395941284015e-05,
+      "loss": 0.2394,
+      "step": 937
+    },
+    {
+      "epoch": 0.60032,
+      "grad_norm": 0.24880407030958504,
+      "learning_rate": 1.88925903138349e-05,
+      "loss": 0.226,
+      "step": 938
+    },
+    {
+      "epoch": 0.60096,
+      "grad_norm": 0.31625732486542973,
+      "learning_rate": 1.884124862263961e-05,
+      "loss": 0.2442,
+      "step": 939
+    },
+    {
+      "epoch": 0.6016,
+      "grad_norm": 0.2873656922578689,
+      "learning_rate": 1.8789934569656932e-05,
+      "loss": 0.2354,
+      "step": 940
+    },
+    {
+      "epoch": 0.60224,
+      "grad_norm": 0.2767633515165357,
+      "learning_rate": 1.8738648385165476e-05,
+      "loss": 0.2511,
+      "step": 941
+    },
+    {
+      "epoch": 0.60288,
+      "grad_norm": 0.250122222077665,
+      "learning_rate": 1.868739029931878e-05,
+      "loss": 0.2269,
+      "step": 942
+    },
+    {
+      "epoch": 0.60352,
+      "grad_norm": 0.2548079571334406,
+      "learning_rate": 1.8636160542144305e-05,
+      "loss": 0.2458,
+      "step": 943
+    },
+    {
+      "epoch": 0.60416,
+      "grad_norm": 0.2580498901867541,
+      "learning_rate": 1.858495934354235e-05,
+      "loss": 0.2515,
+      "step": 944
+    },
+    {
+      "epoch": 0.6048,
+      "grad_norm": 0.25461083763349485,
+      "learning_rate": 1.8533786933285106e-05,
+      "loss": 0.2233,
+      "step": 945
+    },
+    {
+      "epoch": 0.60544,
+      "grad_norm": 0.2712220276055523,
+      "learning_rate": 1.8482643541015514e-05,
+      "loss": 0.2379,
+      "step": 946
+    },
+    {
+      "epoch": 0.60608,
+      "grad_norm": 0.2663737415272671,
+      "learning_rate": 1.843152939624635e-05,
+      "loss": 0.2314,
+      "step": 947
+    },
+    {
+      "epoch": 0.60672,
+      "grad_norm": 0.2557966156563651,
+      "learning_rate": 1.8380444728359086e-05,
+      "loss": 0.228,
+      "step": 948
+    },
+    {
+      "epoch": 0.60736,
+      "grad_norm": 0.2561612472416165,
+      "learning_rate": 1.832938976660295e-05,
+      "loss": 0.2266,
+      "step": 949
+    },
+    {
+      "epoch": 0.608,
+      "grad_norm": 0.2501289124941747,
+      "learning_rate": 1.827836474009383e-05,
+      "loss": 0.2238,
+      "step": 950
+    },
+    {
+      "epoch": 0.60864,
+      "grad_norm": 0.24012895290858152,
+      "learning_rate": 1.8227369877813314e-05,
+      "loss": 0.238,
+      "step": 951
+    },
+    {
+      "epoch": 0.60928,
+      "grad_norm": 0.23688285995128988,
+      "learning_rate": 1.8176405408607574e-05,
+      "loss": 0.2258,
+      "step": 952
+    },
+    {
+      "epoch": 0.60992,
+      "grad_norm": 0.25254935913818866,
+      "learning_rate": 1.812547156118644e-05,
+      "loss": 0.24,
+      "step": 953
+    },
+    {
+      "epoch": 0.61056,
+      "grad_norm": 0.24541424446957505,
+      "learning_rate": 1.8074568564122293e-05,
+      "loss": 0.2284,
+      "step": 954
+    },
+    {
+      "epoch": 0.6112,
+      "grad_norm": 0.2269250208128893,
+      "learning_rate": 1.8023696645849064e-05,
+      "loss": 0.2072,
+      "step": 955
+    },
+    {
+      "epoch": 0.61184,
+      "grad_norm": 0.2500455959868291,
+      "learning_rate": 1.7972856034661244e-05,
+      "loss": 0.2375,
+      "step": 956
+    },
+    {
+      "epoch": 0.61248,
+      "grad_norm": 0.25364622496095157,
+      "learning_rate": 1.7922046958712786e-05,
+      "loss": 0.2298,
+      "step": 957
+    },
+    {
+      "epoch": 0.61312,
+      "grad_norm": 0.22853680752648398,
+      "learning_rate": 1.7871269646016177e-05,
+      "loss": 0.2328,
+      "step": 958
+    },
+    {
+      "epoch": 0.61376,
+      "grad_norm": 0.24406267302696807,
+      "learning_rate": 1.782052432444131e-05,
+      "loss": 0.23,
+      "step": 959
+    },
+    {
+      "epoch": 0.6144,
+      "grad_norm": 0.2515871331087937,
+      "learning_rate": 1.7769811221714544e-05,
+      "loss": 0.2198,
+      "step": 960
+    },
+    {
+      "epoch": 0.61504,
+      "grad_norm": 0.22764605431139925,
+      "learning_rate": 1.7719130565417653e-05,
+      "loss": 0.248,
+      "step": 961
+    },
+    {
+      "epoch": 0.61568,
+      "grad_norm": 0.237687399057445,
+      "learning_rate": 1.7668482582986793e-05,
+      "loss": 0.221,
+      "step": 962
+    },
+    {
+      "epoch": 0.61632,
+      "grad_norm": 0.23787989141058516,
+      "learning_rate": 1.7617867501711478e-05,
+      "loss": 0.2122,
+      "step": 963
+    },
+    {
+      "epoch": 0.61696,
+      "grad_norm": 0.23374245040758787,
+      "learning_rate": 1.7567285548733614e-05,
+      "loss": 0.2369,
+      "step": 964
+    },
+    {
+      "epoch": 0.6176,
+      "grad_norm": 0.22015034136047543,
+      "learning_rate": 1.7516736951046394e-05,
+      "loss": 0.2119,
+      "step": 965
+    },
+    {
+      "epoch": 0.61824,
+      "grad_norm": 0.25795835024829716,
+      "learning_rate": 1.746622193549336e-05,
+      "loss": 0.2345,
+      "step": 966
+    },
+    {
+      "epoch": 0.61888,
+      "grad_norm": 0.24333314596409658,
+      "learning_rate": 1.7415740728767337e-05,
+      "loss": 0.219,
+      "step": 967
+    },
+    {
+      "epoch": 0.61952,
+      "grad_norm": 0.25522623213752005,
+      "learning_rate": 1.7365293557409413e-05,
+      "loss": 0.2454,
+      "step": 968
+    },
+    {
+      "epoch": 0.62016,
+      "grad_norm": 0.2281527795598103,
+      "learning_rate": 1.7314880647807967e-05,
+      "loss": 0.2268,
+      "step": 969
+    },
+    {
+      "epoch": 0.6208,
+      "grad_norm": 0.2629018922135576,
+      "learning_rate": 1.7264502226197593e-05,
+      "loss": 0.2131,
+      "step": 970
+    },
+    {
+      "epoch": 0.62144,
+      "grad_norm": 0.2491411547360006,
+      "learning_rate": 1.7214158518658152e-05,
+      "loss": 0.2165,
+      "step": 971
+    },
+    {
+      "epoch": 0.62208,
+      "grad_norm": 0.24072551480063997,
+      "learning_rate": 1.716384975111368e-05,
+      "loss": 0.2248,
+      "step": 972
+    },
+    {
+      "epoch": 0.62272,
+      "grad_norm": 0.2530719564842945,
+      "learning_rate": 1.711357614933145e-05,
+      "loss": 0.2009,
+      "step": 973
+    },
+    {
+      "epoch": 0.62336,
+      "grad_norm": 0.2650100823610753,
+      "learning_rate": 1.7063337938920895e-05,
+      "loss": 0.2266,
+      "step": 974
+    },
+    {
+      "epoch": 0.624,
+      "grad_norm": 0.24212508790865536,
+      "learning_rate": 1.7013135345332653e-05,
+      "loss": 0.2241,
+      "step": 975
+    },
+    {
+      "epoch": 0.62464,
+      "grad_norm": 0.23563595735476697,
+      "learning_rate": 1.6962968593857502e-05,
+      "loss": 0.2339,
+      "step": 976
+    },
+    {
+      "epoch": 0.62528,
+      "grad_norm": 0.25236749445400286,
+      "learning_rate": 1.6912837909625395e-05,
+      "loss": 0.2523,
+      "step": 977
+    },
+    {
+      "epoch": 0.62592,
+      "grad_norm": 0.24023819460817605,
+      "learning_rate": 1.6862743517604407e-05,
+      "loss": 0.2461,
+      "step": 978
+    },
+    {
+      "epoch": 0.62656,
+      "grad_norm": 0.2632992135948861,
+      "learning_rate": 1.6812685642599758e-05,
+      "loss": 0.2356,
+      "step": 979
+    },
+    {
+      "epoch": 0.6272,
+      "grad_norm": 0.24821791130944226,
+      "learning_rate": 1.6762664509252806e-05,
+      "loss": 0.244,
+      "step": 980
+    },
+    {
+      "epoch": 0.62784,
+      "grad_norm": 0.2306992285424631,
+      "learning_rate": 1.6712680342039994e-05,
+      "loss": 0.2115,
+      "step": 981
+    },
+    {
+      "epoch": 0.62848,
+      "grad_norm": 0.2540881408962946,
+      "learning_rate": 1.6662733365271916e-05,
+      "loss": 0.2319,
+      "step": 982
+    },
+    {
+      "epoch": 0.62912,
+      "grad_norm": 0.24509281530559013,
+      "learning_rate": 1.661282380309223e-05,
+      "loss": 0.2402,
+      "step": 983
+    },
+    {
+      "epoch": 0.62976,
+      "grad_norm": 0.2297046642955313,
+      "learning_rate": 1.6562951879476708e-05,
+      "loss": 0.25,
+      "step": 984
+    },
+    {
+      "epoch": 0.6304,
+      "grad_norm": 0.24831460343070508,
+      "learning_rate": 1.6513117818232217e-05,
+      "loss": 0.2216,
+      "step": 985
+    },
+    {
+      "epoch": 0.63104,
+      "grad_norm": 0.272621749507111,
+      "learning_rate": 1.6463321842995717e-05,
+      "loss": 0.2046,
+      "step": 986
+    },
+    {
+      "epoch": 0.63168,
+      "grad_norm": 0.23775794084413449,
+      "learning_rate": 1.641356417723322e-05,
+      "loss": 0.2195,
+      "step": 987
+    },
+    {
+      "epoch": 0.63232,
+      "grad_norm": 0.24935420480465814,
+      "learning_rate": 1.6363845044238868e-05,
+      "loss": 0.2288,
+      "step": 988
+    },
+    {
+      "epoch": 0.63296,
+      "grad_norm": 0.2431399311707703,
+      "learning_rate": 1.631416466713383e-05,
+      "loss": 0.2339,
+      "step": 989
+    },
+    {
+      "epoch": 0.6336,
+      "grad_norm": 0.2629341579809072,
+      "learning_rate": 1.62645232688654e-05,
+      "loss": 0.2122,
+      "step": 990
+    },
+    {
+      "epoch": 0.63424,
+      "grad_norm": 0.2358691945192489,
+      "learning_rate": 1.621492107220593e-05,
+      "loss": 0.2243,
+      "step": 991
+    },
+    {
+      "epoch": 0.63488,
+      "grad_norm": 0.24802048283867612,
+      "learning_rate": 1.6165358299751825e-05,
+      "loss": 0.2414,
+      "step": 992
+    },
+    {
+      "epoch": 0.63552,
+      "grad_norm": 0.26302321903716336,
+      "learning_rate": 1.6115835173922618e-05,
+      "loss": 0.2287,
+      "step": 993
+    },
+    {
+      "epoch": 0.63616,
+      "grad_norm": 0.2522729710309751,
+      "learning_rate": 1.6066351916959876e-05,
+      "loss": 0.2384,
+      "step": 994
+    },
+    {
+      "epoch": 0.6368,
+      "grad_norm": 0.24691458951344603,
+      "learning_rate": 1.6016908750926286e-05,
+      "loss": 0.2275,
+      "step": 995
+    },
+    {
+      "epoch": 0.63744,
+      "grad_norm": 0.24134289572608625,
+      "learning_rate": 1.596750589770459e-05,
+      "loss": 0.2511,
+      "step": 996
+    },
+    {
+      "epoch": 0.63808,
+      "grad_norm": 0.24770605715280714,
+      "learning_rate": 1.5918143578996646e-05,
+      "loss": 0.2288,
+      "step": 997
+    },
+    {
+      "epoch": 0.63872,
+      "grad_norm": 0.24556908211864034,
+      "learning_rate": 1.5868822016322386e-05,
+      "loss": 0.2291,
+      "step": 998
+    },
+    {
+      "epoch": 0.63936,
+      "grad_norm": 0.27576361598675675,
+      "learning_rate": 1.5819541431018874e-05,
+      "loss": 0.247,
+      "step": 999
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.2570737510294073,
+      "learning_rate": 1.577030204423924e-05,
+      "loss": 0.2319,
+      "step": 1000
+    },
+    {
+      "epoch": 0.64064,
+      "grad_norm": 0.2534564438404505,
+      "learning_rate": 1.5721104076951783e-05,
+      "loss": 0.2185,
+      "step": 1001
+    },
+    {
+      "epoch": 0.64128,
+      "grad_norm": 0.2788588872429722,
+      "learning_rate": 1.5671947749938875e-05,
+      "loss": 0.2408,
+      "step": 1002
+    },
+    {
+      "epoch": 0.64192,
+      "grad_norm": 0.25321642004297756,
+      "learning_rate": 1.562283328379605e-05,
+      "loss": 0.234,
+      "step": 1003
+    },
+    {
+      "epoch": 0.64256,
+      "grad_norm": 0.2614491407627203,
+      "learning_rate": 1.5573760898931e-05,
+      "loss": 0.2173,
+      "step": 1004
+    },
+    {
+      "epoch": 0.6432,
+      "grad_norm": 0.2688539511176729,
+      "learning_rate": 1.5524730815562518e-05,
+      "loss": 0.2311,
+      "step": 1005
+    },
+    {
+      "epoch": 0.64384,
+      "grad_norm": 0.2660611603035458,
+      "learning_rate": 1.5475743253719628e-05,
+      "loss": 0.2169,
+      "step": 1006
+    },
+    {
+      "epoch": 0.64448,
+      "grad_norm": 0.2640835654318191,
+      "learning_rate": 1.542679843324049e-05,
+      "loss": 0.2341,
+      "step": 1007
+    },
+    {
+      "epoch": 0.64512,
+      "grad_norm": 0.2407195749259193,
+      "learning_rate": 1.5377896573771465e-05,
+      "loss": 0.2397,
+      "step": 1008
+    },
+    {
+      "epoch": 0.64576,
+      "grad_norm": 0.243922181554778,
+      "learning_rate": 1.5329037894766133e-05,
+      "loss": 0.2504,
+      "step": 1009
+    },
+    {
+      "epoch": 0.6464,
+      "grad_norm": 0.2648799412774992,
+      "learning_rate": 1.528022261548429e-05,
+      "loss": 0.2357,
+      "step": 1010
+    },
+    {
+      "epoch": 0.64704,
+      "grad_norm": 0.2417283438561751,
+      "learning_rate": 1.5231450954990955e-05,
+      "loss": 0.2422,
+      "step": 1011
+    },
+    {
+      "epoch": 0.64768,
+      "grad_norm": 0.24310411258712905,
+      "learning_rate": 1.518272313215543e-05,
+      "loss": 0.2327,
+      "step": 1012
+    },
+    {
+      "epoch": 0.64832,
+      "grad_norm": 0.24347727311974987,
+      "learning_rate": 1.513403936565026e-05,
+      "loss": 0.2224,
+      "step": 1013
+    },
+    {
+      "epoch": 0.64896,
+      "grad_norm": 0.2152097717576801,
+      "learning_rate": 1.5085399873950313e-05,
+      "loss": 0.2131,
+      "step": 1014
+    },
+    {
+      "epoch": 0.6496,
+      "grad_norm": 0.25368340123068406,
+      "learning_rate": 1.5036804875331734e-05,
+      "loss": 0.2666,
+      "step": 1015
+    },
+    {
+      "epoch": 0.65024,
+      "grad_norm": 0.21958417853607198,
+      "learning_rate": 1.498825458787102e-05,
+      "loss": 0.2361,
+      "step": 1016
+    },
+    {
+      "epoch": 0.65088,
+      "grad_norm": 0.2466692548385663,
+      "learning_rate": 1.4939749229444028e-05,
+      "loss": 0.2274,
+      "step": 1017
+    },
+    {
+      "epoch": 0.65152,
+      "grad_norm": 0.25529899685788066,
+      "learning_rate": 1.4891289017724967e-05,
+      "loss": 0.2262,
+      "step": 1018
+    },
+    {
+      "epoch": 0.65216,
+      "grad_norm": 0.23621391564889319,
+      "learning_rate": 1.4842874170185467e-05,
+      "loss": 0.2369,
+      "step": 1019
+    },
+    {
+      "epoch": 0.6528,
+      "grad_norm": 0.23532265918464682,
+      "learning_rate": 1.4794504904093559e-05,
+      "loss": 0.2211,
+      "step": 1020
+    },
+    {
+      "epoch": 0.65344,
+      "grad_norm": 0.24481175887672052,
+      "learning_rate": 1.4746181436512747e-05,
+      "loss": 0.2496,
+      "step": 1021
+    },
+    {
+      "epoch": 0.65408,
+      "grad_norm": 0.2593468164261547,
+      "learning_rate": 1.4697903984300964e-05,
+      "loss": 0.2091,
+      "step": 1022
+    },
+    {
+      "epoch": 0.65472,
+      "grad_norm": 0.24988430452745558,
+      "learning_rate": 1.4649672764109704e-05,
+      "loss": 0.2347,
+      "step": 1023
+    },
+    {
+      "epoch": 0.65536,
+      "grad_norm": 0.28606520719702316,
+      "learning_rate": 1.460148799238293e-05,
+      "loss": 0.2268,
+      "step": 1024
+    },
+    {
+      "epoch": 0.656,
+      "grad_norm": 0.24354085938785522,
+      "learning_rate": 1.4553349885356213e-05,
+      "loss": 0.2351,
+      "step": 1025
+    },
+    {
+      "epoch": 0.65664,
+      "grad_norm": 0.24024216627401732,
+      "learning_rate": 1.4505258659055665e-05,
+      "loss": 0.2123,
+      "step": 1026
+    },
+    {
+      "epoch": 0.65728,
+      "grad_norm": 0.2787995702238703,
+      "learning_rate": 1.4457214529297043e-05,
+      "loss": 0.2518,
+      "step": 1027
+    },
+    {
+      "epoch": 0.65792,
+      "grad_norm": 0.2504422318802916,
+      "learning_rate": 1.4409217711684737e-05,
+      "loss": 0.2271,
+      "step": 1028
+    },
+    {
+      "epoch": 0.65856,
+      "grad_norm": 0.2402553559005155,
+      "learning_rate": 1.4361268421610829e-05,
+      "loss": 0.208,
+      "step": 1029
+    },
+    {
+      "epoch": 0.6592,
+      "grad_norm": 0.24214818854883866,
+      "learning_rate": 1.431336687425411e-05,
+      "loss": 0.2417,
+      "step": 1030
+    },
+    {
+      "epoch": 0.65984,
+      "grad_norm": 0.2269309859275578,
+      "learning_rate": 1.4265513284579118e-05,
+      "loss": 0.2198,
+      "step": 1031
+    },
+    {
+      "epoch": 0.66048,
+      "grad_norm": 0.24028490163911678,
+      "learning_rate": 1.4217707867335179e-05,
+      "loss": 0.2137,
+      "step": 1032
+    },
+    {
+      "epoch": 0.66112,
+      "grad_norm": 0.264876802958815,
+      "learning_rate": 1.4169950837055424e-05,
+      "loss": 0.2333,
+      "step": 1033
+    },
+    {
+      "epoch": 0.66176,
+      "grad_norm": 0.25224688050666716,
+      "learning_rate": 1.4122242408055864e-05,
+      "loss": 0.2301,
+      "step": 1034
+    },
+    {
+      "epoch": 0.6624,
+      "grad_norm": 0.2507646158580805,
+      "learning_rate": 1.4074582794434388e-05,
+      "loss": 0.2404,
+      "step": 1035
+    },
+    {
+      "epoch": 0.66304,
+      "grad_norm": 0.24905885018937385,
+      "learning_rate": 1.4026972210069839e-05,
+      "loss": 0.2227,
+      "step": 1036
+    },
+    {
+      "epoch": 0.66368,
+      "grad_norm": 0.21687785520075453,
+      "learning_rate": 1.3979410868620987e-05,
+      "loss": 0.2332,
+      "step": 1037
+    },
+    {
+      "epoch": 0.66432,
+      "grad_norm": 0.23384411948873113,
+      "learning_rate": 1.3931898983525694e-05,
+      "loss": 0.246,
+      "step": 1038
+    },
+    {
+      "epoch": 0.66496,
+      "grad_norm": 0.2346938183755602,
+      "learning_rate": 1.3884436767999798e-05,
+      "loss": 0.2438,
+      "step": 1039
+    },
+    {
+      "epoch": 0.6656,
+      "grad_norm": 0.22146292146840602,
+      "learning_rate": 1.3837024435036309e-05,
+      "loss": 0.2097,
+      "step": 1040
+    },
+    {
+      "epoch": 0.66624,
+      "grad_norm": 0.24505548039402017,
+      "learning_rate": 1.3789662197404346e-05,
+      "loss": 0.2147,
+      "step": 1041
+    },
+    {
+      "epoch": 0.66688,
+      "grad_norm": 0.27664431296730363,
+      "learning_rate": 1.3742350267648202e-05,
+      "loss": 0.2288,
+      "step": 1042
+    },
+    {
+      "epoch": 0.66752,
+      "grad_norm": 0.22818871336132518,
+      "learning_rate": 1.3695088858086464e-05,
+      "loss": 0.2359,
+      "step": 1043
+    },
+    {
+      "epoch": 0.66816,
+      "grad_norm": 0.23045475526006876,
+      "learning_rate": 1.3647878180810924e-05,
+      "loss": 0.2157,
+      "step": 1044
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1562,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 261,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 478374861078528.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}