diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5362 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.996517107150174,
+  "eval_steps": 500,
+  "global_step": 760,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.006556033599672198,
+      "grad_norm": 6.308972265166019,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 1.085,
+      "step": 1
+    },
+    {
+      "epoch": 0.013112067199344397,
+      "grad_norm": 6.3669176951390085,
+      "learning_rate": 2.105263157894737e-06,
+      "loss": 1.0943,
+      "step": 2
+    },
+    {
+      "epoch": 0.019668100799016593,
+      "grad_norm": 6.248172461223065,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 1.0945,
+      "step": 3
+    },
+    {
+      "epoch": 0.026224134398688793,
+      "grad_norm": 5.720783796325484,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 1.0822,
+      "step": 4
+    },
+    {
+      "epoch": 0.03278016799836099,
+      "grad_norm": 4.166580116763421,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 1.0376,
+      "step": 5
+    },
+    {
+      "epoch": 0.03933620159803319,
+      "grad_norm": 2.5258613072829443,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 1.0088,
+      "step": 6
+    },
+    {
+      "epoch": 0.04589223519770539,
+      "grad_norm": 3.241153982591279,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 0.9787,
+      "step": 7
+    },
+    {
+      "epoch": 0.05244826879737759,
+      "grad_norm": 4.215997338140802,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 1.0021,
+      "step": 8
+    },
+    {
+      "epoch": 0.05900430239704978,
+      "grad_norm": 3.7917272082574036,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 0.9883,
+      "step": 9
+    },
+    {
+      "epoch": 0.06556033599672198,
+      "grad_norm": 3.1091596358257414,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 0.9323,
+      "step": 10
+    },
+    {
+      "epoch": 0.07211636959639418,
+      "grad_norm": 2.7854345292697746,
+      "learning_rate": 1.1578947368421053e-05,
+      "loss": 0.9234,
+      "step": 11
+    },
+    {
+      "epoch": 0.07867240319606637,
+      "grad_norm": 1.8650709570608603,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 0.9023,
+      "step": 12
+    },
+    {
+      "epoch": 0.08522843679573858,
+      "grad_norm": 1.574226758545731,
+      "learning_rate": 1.3684210526315791e-05,
+      "loss": 0.8693,
+      "step": 13
+    },
+    {
+      "epoch": 0.09178447039541078,
+      "grad_norm": 1.4177768637857613,
+      "learning_rate": 1.4736842105263159e-05,
+      "loss": 0.8558,
+      "step": 14
+    },
+    {
+      "epoch": 0.09834050399508297,
+      "grad_norm": 1.2125668795375901,
+      "learning_rate": 1.578947368421053e-05,
+      "loss": 0.8394,
+      "step": 15
+    },
+    {
+      "epoch": 0.10489653759475517,
+      "grad_norm": 1.1398741474728522,
+      "learning_rate": 1.6842105263157896e-05,
+      "loss": 0.8373,
+      "step": 16
+    },
+    {
+      "epoch": 0.11145257119442738,
+      "grad_norm": 1.0380532605315191,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 0.8138,
+      "step": 17
+    },
+    {
+      "epoch": 0.11800860479409957,
+      "grad_norm": 1.1944105144048436,
+      "learning_rate": 1.894736842105263e-05,
+      "loss": 0.8178,
+      "step": 18
+    },
+    {
+      "epoch": 0.12456463839377177,
+      "grad_norm": 0.9889022037403388,
+      "learning_rate": 2e-05,
+      "loss": 0.803,
+      "step": 19
+    },
+    {
+      "epoch": 0.13112067199344396,
+      "grad_norm": 0.7531044929728199,
+      "learning_rate": 2.105263157894737e-05,
+      "loss": 0.806,
+      "step": 20
+    },
+    {
+      "epoch": 0.13767670559311618,
+      "grad_norm": 0.9451120842150798,
+      "learning_rate": 2.210526315789474e-05,
+      "loss": 0.7904,
+      "step": 21
+    },
+    {
+      "epoch": 0.14423273919278837,
+      "grad_norm": 0.8762905130696029,
+      "learning_rate": 2.3157894736842107e-05,
+      "loss": 0.8001,
+      "step": 22
+    },
+    {
+      "epoch": 0.15078877279246056,
+      "grad_norm": 0.6699562168478991,
+      "learning_rate": 2.4210526315789474e-05,
+      "loss": 0.7822,
+      "step": 23
+    },
+    {
+      "epoch": 0.15734480639213275,
+      "grad_norm": 0.6276115227175743,
+      "learning_rate": 2.526315789473684e-05,
+      "loss": 0.7773,
+      "step": 24
+    },
+    {
+      "epoch": 0.16390083999180496,
+      "grad_norm": 0.6290807862288751,
+      "learning_rate": 2.6315789473684215e-05,
+      "loss": 0.7783,
+      "step": 25
+    },
+    {
+      "epoch": 0.17045687359147715,
+      "grad_norm": 0.5742564332284298,
+      "learning_rate": 2.7368421052631583e-05,
+      "loss": 0.7674,
+      "step": 26
+    },
+    {
+      "epoch": 0.17701290719114934,
+      "grad_norm": 0.7412922803237626,
+      "learning_rate": 2.842105263157895e-05,
+      "loss": 0.7698,
+      "step": 27
+    },
+    {
+      "epoch": 0.18356894079082156,
+      "grad_norm": 1.0547770655616728,
+      "learning_rate": 2.9473684210526317e-05,
+      "loss": 0.7655,
+      "step": 28
+    },
+    {
+      "epoch": 0.19012497439049375,
+      "grad_norm": 1.5527104367024405,
+      "learning_rate": 3.052631578947369e-05,
+      "loss": 0.7707,
+      "step": 29
+    },
+    {
+      "epoch": 0.19668100799016594,
+      "grad_norm": 0.7294937893627019,
+      "learning_rate": 3.157894736842106e-05,
+      "loss": 0.7578,
+      "step": 30
+    },
+    {
+      "epoch": 0.20323704158983816,
+      "grad_norm": 0.9536771340964642,
+      "learning_rate": 3.2631578947368426e-05,
+      "loss": 0.748,
+      "step": 31
+    },
+    {
+      "epoch": 0.20979307518951035,
+      "grad_norm": 1.6534518524100028,
+      "learning_rate": 3.368421052631579e-05,
+      "loss": 0.7588,
+      "step": 32
+    },
+    {
+      "epoch": 0.21634910878918254,
+      "grad_norm": 0.6082911982013689,
+      "learning_rate": 3.473684210526316e-05,
+      "loss": 0.7424,
+      "step": 33
+    },
+    {
+      "epoch": 0.22290514238885475,
+      "grad_norm": 1.4728558033024115,
+      "learning_rate": 3.578947368421053e-05,
+      "loss": 0.7423,
+      "step": 34
+    },
+    {
+      "epoch": 0.22946117598852694,
+      "grad_norm": 0.8950252912362415,
+      "learning_rate": 3.6842105263157895e-05,
+      "loss": 0.7396,
+      "step": 35
+    },
+    {
+      "epoch": 0.23601720958819913,
+      "grad_norm": 1.2202323688159324,
+      "learning_rate": 3.789473684210526e-05,
+      "loss": 0.7394,
+      "step": 36
+    },
+    {
+      "epoch": 0.24257324318787135,
+      "grad_norm": 1.0375163505188818,
+      "learning_rate": 3.8947368421052636e-05,
+      "loss": 0.735,
+      "step": 37
+    },
+    {
+      "epoch": 0.24912927678754354,
+      "grad_norm": 1.1817870027992188,
+      "learning_rate": 4e-05,
+      "loss": 0.7316,
+      "step": 38
+    },
+    {
+      "epoch": 0.25568531038721576,
+      "grad_norm": 1.7472819001700346,
+      "learning_rate": 4.105263157894738e-05,
+      "loss": 0.7371,
+      "step": 39
+    },
+    {
+      "epoch": 0.2622413439868879,
+      "grad_norm": 0.6997273968133122,
+      "learning_rate": 4.210526315789474e-05,
+      "loss": 0.7303,
+      "step": 40
+    },
+    {
+      "epoch": 0.26879737758656014,
+      "grad_norm": 2.321753393397282,
+      "learning_rate": 4.315789473684211e-05,
+      "loss": 0.7458,
+      "step": 41
+    },
+    {
+      "epoch": 0.27535341118623236,
+      "grad_norm": 1.366635958874213,
+      "learning_rate": 4.421052631578948e-05,
+      "loss": 0.7176,
+      "step": 42
+    },
+    {
+      "epoch": 0.2819094447859045,
+      "grad_norm": 2.6212092693642552,
+      "learning_rate": 4.5263157894736846e-05,
+      "loss": 0.7315,
+      "step": 43
+    },
+    {
+      "epoch": 0.28846547838557673,
+      "grad_norm": 2.5130277449596496,
+      "learning_rate": 4.6315789473684214e-05,
+      "loss": 0.753,
+      "step": 44
+    },
+    {
+      "epoch": 0.29502151198524895,
+      "grad_norm": 1.6285728910727462,
+      "learning_rate": 4.736842105263158e-05,
+      "loss": 0.7297,
+      "step": 45
+    },
+    {
+      "epoch": 0.3015775455849211,
+      "grad_norm": 2.0964313396830456,
+      "learning_rate": 4.842105263157895e-05,
+      "loss": 0.7265,
+      "step": 46
+    },
+    {
+      "epoch": 0.30813357918459333,
+      "grad_norm": 2.0887773848236697,
+      "learning_rate": 4.947368421052632e-05,
+      "loss": 0.7239,
+      "step": 47
+    },
+    {
+      "epoch": 0.3146896127842655,
+      "grad_norm": 1.1072655635857458,
+      "learning_rate": 5.052631578947368e-05,
+      "loss": 0.7193,
+      "step": 48
+    },
+    {
+      "epoch": 0.3212456463839377,
+      "grad_norm": 1.4488623000707561,
+      "learning_rate": 5.157894736842106e-05,
+      "loss": 0.7203,
+      "step": 49
+    },
+    {
+      "epoch": 0.32780167998360993,
+      "grad_norm": 1.2322683716693836,
+      "learning_rate": 5.263157894736843e-05,
+      "loss": 0.7238,
+      "step": 50
+    },
+    {
+      "epoch": 0.3343577135832821,
+      "grad_norm": 1.38345500389746,
+      "learning_rate": 5.368421052631579e-05,
+      "loss": 0.7142,
+      "step": 51
+    },
+    {
+      "epoch": 0.3409137471829543,
+      "grad_norm": 0.9094126438399451,
+      "learning_rate": 5.4736842105263165e-05,
+      "loss": 0.72,
+      "step": 52
+    },
+    {
+      "epoch": 0.3474697807826265,
+      "grad_norm": 1.2022041950473747,
+      "learning_rate": 5.5789473684210526e-05,
+      "loss": 0.7075,
+      "step": 53
+    },
+    {
+      "epoch": 0.3540258143822987,
+      "grad_norm": 1.6189179778780058,
+      "learning_rate": 5.68421052631579e-05,
+      "loss": 0.7105,
+      "step": 54
+    },
+    {
+      "epoch": 0.3605818479819709,
+      "grad_norm": 1.2006446280190224,
+      "learning_rate": 5.789473684210527e-05,
+      "loss": 0.7151,
+      "step": 55
+    },
+    {
+      "epoch": 0.3671378815816431,
+      "grad_norm": 1.6321920047861564,
+      "learning_rate": 5.8947368421052634e-05,
+      "loss": 0.7136,
+      "step": 56
+    },
+    {
+      "epoch": 0.3736939151813153,
+      "grad_norm": 1.5287305591592921,
+      "learning_rate": 6.000000000000001e-05,
+      "loss": 0.7107,
+      "step": 57
+    },
+    {
+      "epoch": 0.3802499487809875,
+      "grad_norm": 1.3257983915749183,
+      "learning_rate": 6.105263157894738e-05,
+      "loss": 0.7183,
+      "step": 58
+    },
+    {
+      "epoch": 0.3868059823806597,
+      "grad_norm": 0.8173893638911733,
+      "learning_rate": 6.210526315789474e-05,
+      "loss": 0.7087,
+      "step": 59
+    },
+    {
+      "epoch": 0.3933620159803319,
+      "grad_norm": 1.5027100216095572,
+      "learning_rate": 6.315789473684212e-05,
+      "loss": 0.7099,
+      "step": 60
+    },
+    {
+      "epoch": 0.3999180495800041,
+      "grad_norm": 0.9118886385672123,
+      "learning_rate": 6.421052631578948e-05,
+      "loss": 0.7056,
+      "step": 61
+    },
+    {
+      "epoch": 0.4064740831796763,
+      "grad_norm": 8.129014604966502,
+      "learning_rate": 6.526315789473685e-05,
+      "loss": 0.7359,
+      "step": 62
+    },
+    {
+      "epoch": 0.4130301167793485,
+      "grad_norm": 3.8240665538108343,
+      "learning_rate": 6.631578947368421e-05,
+      "loss": 0.7738,
+      "step": 63
+    },
+    {
+      "epoch": 0.4195861503790207,
+      "grad_norm": 3.710700934507178,
+      "learning_rate": 6.736842105263159e-05,
+      "loss": 0.7309,
+      "step": 64
+    },
+    {
+      "epoch": 0.4261421839786929,
+      "grad_norm": 1.7575755137455045,
+      "learning_rate": 6.842105263157895e-05,
+      "loss": 0.7373,
+      "step": 65
+    },
+    {
+      "epoch": 0.4326982175783651,
+      "grad_norm": 1.29443521888165,
+      "learning_rate": 6.947368421052632e-05,
+      "loss": 0.7149,
+      "step": 66
+    },
+    {
+      "epoch": 0.4392542511780373,
+      "grad_norm": 1.9730328727947009,
+      "learning_rate": 7.052631578947368e-05,
+      "loss": 0.7309,
+      "step": 67
+    },
+    {
+      "epoch": 0.4458102847777095,
+      "grad_norm": 1.5434748912530512,
+      "learning_rate": 7.157894736842105e-05,
+      "loss": 0.7307,
+      "step": 68
+    },
+    {
+      "epoch": 0.45236631837738167,
+      "grad_norm": 1.3098144472083486,
+      "learning_rate": 7.263157894736843e-05,
+      "loss": 0.7226,
+      "step": 69
+    },
+    {
+      "epoch": 0.4589223519770539,
+      "grad_norm": 1.4763764747327572,
+      "learning_rate": 7.368421052631579e-05,
+      "loss": 0.7122,
+      "step": 70
+    },
+    {
+      "epoch": 0.4654783855767261,
+      "grad_norm": 1.782399664190018,
+      "learning_rate": 7.473684210526316e-05,
+      "loss": 0.7105,
+      "step": 71
+    },
+    {
+      "epoch": 0.47203441917639827,
+      "grad_norm": 0.8645965374178922,
+      "learning_rate": 7.578947368421052e-05,
+      "loss": 0.7189,
+      "step": 72
+    },
+    {
+      "epoch": 0.4785904527760705,
+      "grad_norm": 1.1783141500345207,
+      "learning_rate": 7.68421052631579e-05,
+      "loss": 0.7155,
+      "step": 73
+    },
+    {
+      "epoch": 0.4851464863757427,
+      "grad_norm": 1.631405575969008,
+      "learning_rate": 7.789473684210527e-05,
+      "loss": 0.7163,
+      "step": 74
+    },
+    {
+      "epoch": 0.49170251997541486,
+      "grad_norm": 1.799584170211315,
+      "learning_rate": 7.894736842105263e-05,
+      "loss": 0.724,
+      "step": 75
+    },
+    {
+      "epoch": 0.4982585535750871,
+      "grad_norm": 131.28302876683892,
+      "learning_rate": 8e-05,
+      "loss": 0.9347,
+      "step": 76
+    },
+    {
+      "epoch": 0.5048145871747592,
+      "grad_norm": 2.879295120261663,
+      "learning_rate": 7.999957809295807e-05,
+      "loss": 0.7474,
+      "step": 77
+    },
+    {
+      "epoch": 0.5113706207744315,
+      "grad_norm": 0.9410659862896121,
+      "learning_rate": 7.99983123807325e-05,
+      "loss": 0.7124,
+      "step": 78
+    },
+    {
+      "epoch": 0.5179266543741037,
+      "grad_norm": 2.4498983574935487,
+      "learning_rate": 7.999620289002397e-05,
+      "loss": 0.7294,
+      "step": 79
+    },
+    {
+      "epoch": 0.5244826879737758,
+      "grad_norm": 1.3312297085268832,
+      "learning_rate": 7.999324966533291e-05,
+      "loss": 0.7155,
+      "step": 80
+    },
+    {
+      "epoch": 0.5310387215734481,
+      "grad_norm": 1.9333142275612973,
+      "learning_rate": 7.998945276895866e-05,
+      "loss": 0.7249,
+      "step": 81
+    },
+    {
+      "epoch": 0.5375947551731203,
+      "grad_norm": 1.5284340416931426,
+      "learning_rate": 7.998481228099806e-05,
+      "loss": 0.7199,
+      "step": 82
+    },
+    {
+      "epoch": 0.5441507887727924,
+      "grad_norm": 1.5374661844560404,
+      "learning_rate": 7.997932829934386e-05,
+      "loss": 0.7174,
+      "step": 83
+    },
+    {
+      "epoch": 0.5507068223724647,
+      "grad_norm": 1.099517635025565,
+      "learning_rate": 7.997300093968255e-05,
+      "loss": 0.7106,
+      "step": 84
+    },
+    {
+      "epoch": 0.5572628559721369,
+      "grad_norm": 1.4642952447149862,
+      "learning_rate": 7.996583033549204e-05,
+      "loss": 0.7087,
+      "step": 85
+    },
+    {
+      "epoch": 0.563818889571809,
+      "grad_norm": 1.4883363499473765,
+      "learning_rate": 7.995781663803876e-05,
+      "loss": 0.724,
+      "step": 86
+    },
+    {
+      "epoch": 0.5703749231714813,
+      "grad_norm": 0.8588229153701384,
+      "learning_rate": 7.994896001637443e-05,
+      "loss": 0.7052,
+      "step": 87
+    },
+    {
+      "epoch": 0.5769309567711535,
+      "grad_norm": 1.1329548393299773,
+      "learning_rate": 7.993926065733265e-05,
+      "loss": 0.7092,
+      "step": 88
+    },
+    {
+      "epoch": 0.5834869903708256,
+      "grad_norm": 1.9381285339000986,
+      "learning_rate": 7.99287187655248e-05,
+      "loss": 0.7075,
+      "step": 89
+    },
+    {
+      "epoch": 0.5900430239704979,
+      "grad_norm": 1.0298360769847832,
+      "learning_rate": 7.991733456333579e-05,
+      "loss": 0.7088,
+      "step": 90
+    },
+    {
+      "epoch": 0.5965990575701701,
+      "grad_norm": 1.5913971706858665,
+      "learning_rate": 7.990510829091938e-05,
+      "loss": 0.7044,
+      "step": 91
+    },
+    {
+      "epoch": 0.6031550911698422,
+      "grad_norm": 1.0752295835923937,
+      "learning_rate": 7.98920402061931e-05,
+      "loss": 0.698,
+      "step": 92
+    },
+    {
+      "epoch": 0.6097111247695144,
+      "grad_norm": 1.3411827568812442,
+      "learning_rate": 7.987813058483278e-05,
+      "loss": 0.6897,
+      "step": 93
+    },
+    {
+      "epoch": 0.6162671583691867,
+      "grad_norm": 1.0994663107424685,
+      "learning_rate": 7.98633797202668e-05,
+      "loss": 0.7009,
+      "step": 94
+    },
+    {
+      "epoch": 0.6228231919688588,
+      "grad_norm": 1.5572209303699358,
+      "learning_rate": 7.984778792366983e-05,
+      "loss": 0.7001,
+      "step": 95
+    },
+    {
+      "epoch": 0.629379225568531,
+      "grad_norm": 1.2622939681539915,
+      "learning_rate": 7.98313555239563e-05,
+      "loss": 0.7027,
+      "step": 96
+    },
+    {
+      "epoch": 0.6359352591682033,
+      "grad_norm": 1.5287178198020728,
+      "learning_rate": 7.98140828677735e-05,
+      "loss": 0.6977,
+      "step": 97
+    },
+    {
+      "epoch": 0.6424912927678754,
+      "grad_norm": 1.1853944350439,
+      "learning_rate": 7.979597031949415e-05,
+      "loss": 0.6943,
+      "step": 98
+    },
+    {
+      "epoch": 0.6490473263675476,
+      "grad_norm": 1.1219402341610256,
+      "learning_rate": 7.977701826120888e-05,
+      "loss": 0.6984,
+      "step": 99
+    },
+    {
+      "epoch": 0.6556033599672199,
+      "grad_norm": 1.8026221843535832,
+      "learning_rate": 7.975722709271799e-05,
+      "loss": 0.6955,
+      "step": 100
+    },
+    {
+      "epoch": 0.662159393566892,
+      "grad_norm": 1.0073031848270164,
+      "learning_rate": 7.973659723152317e-05,
+      "loss": 0.6942,
+      "step": 101
+    },
+    {
+      "epoch": 0.6687154271665642,
+      "grad_norm": 1.4709635304665,
+      "learning_rate": 7.97151291128186e-05,
+      "loss": 0.7024,
+      "step": 102
+    },
+    {
+      "epoch": 0.6752714607662365,
+      "grad_norm": 1.5710635409318867,
+      "learning_rate": 7.96928231894818e-05,
+      "loss": 0.6917,
+      "step": 103
+    },
+    {
+      "epoch": 0.6818274943659086,
+      "grad_norm": 0.8520728596618132,
+      "learning_rate": 7.96696799320641e-05,
+      "loss": 0.6821,
+      "step": 104
+    },
+    {
+      "epoch": 0.6883835279655808,
+      "grad_norm": 1.3459049216152963,
+      "learning_rate": 7.964569982878063e-05,
+      "loss": 0.6916,
+      "step": 105
+    },
+    {
+      "epoch": 0.694939561565253,
+      "grad_norm": 0.9826433336489447,
+      "learning_rate": 7.962088338550013e-05,
+      "loss": 0.6894,
+      "step": 106
+    },
+    {
+      "epoch": 0.7014955951649252,
+      "grad_norm": 1.2896938337982984,
+      "learning_rate": 7.959523112573422e-05,
+      "loss": 0.6933,
+      "step": 107
+    },
+    {
+      "epoch": 0.7080516287645974,
+      "grad_norm": 0.9867952374397386,
+      "learning_rate": 7.956874359062632e-05,
+      "loss": 0.697,
+      "step": 108
+    },
+    {
+      "epoch": 0.7146076623642696,
+      "grad_norm": 1.2639647577953248,
+      "learning_rate": 7.954142133894033e-05,
+      "loss": 0.6894,
+      "step": 109
+    },
+    {
+      "epoch": 0.7211636959639418,
+      "grad_norm": 1.357158506896498,
+      "learning_rate": 7.951326494704878e-05,
+      "loss": 0.691,
+      "step": 110
+    },
+    {
+      "epoch": 0.727719729563614,
+      "grad_norm": 0.8936367138602801,
+      "learning_rate": 7.948427500892065e-05,
+      "loss": 0.6887,
+      "step": 111
+    },
+    {
+      "epoch": 0.7342757631632862,
+      "grad_norm": 1.0046258611386971,
+      "learning_rate": 7.94544521361089e-05,
+      "loss": 0.6866,
+      "step": 112
+    },
+    {
+      "epoch": 0.7408317967629584,
+      "grad_norm": 1.0742250060740215,
+      "learning_rate": 7.942379695773753e-05,
+      "loss": 0.6888,
+      "step": 113
+    },
+    {
+      "epoch": 0.7473878303626306,
+      "grad_norm": 0.7284144457213086,
+      "learning_rate": 7.939231012048833e-05,
+      "loss": 0.6811,
+      "step": 114
+    },
+    {
+      "epoch": 0.7539438639623028,
+      "grad_norm": 0.7052427321018496,
+      "learning_rate": 7.93599922885872e-05,
+      "loss": 0.6763,
+      "step": 115
+    },
+    {
+      "epoch": 0.760499897561975,
+      "grad_norm": 0.798707901395817,
+      "learning_rate": 7.932684414379021e-05,
+      "loss": 0.692,
+      "step": 116
+    },
+    {
+      "epoch": 0.7670559311616472,
+      "grad_norm": 1.5879889624779187,
+      "learning_rate": 7.929286638536913e-05,
+      "loss": 0.6908,
+      "step": 117
+    },
+    {
+      "epoch": 0.7736119647613194,
+      "grad_norm": 0.8333519158322553,
+      "learning_rate": 7.925805973009672e-05,
+      "loss": 0.6734,
+      "step": 118
+    },
+    {
+      "epoch": 0.7801679983609916,
+      "grad_norm": 1.0218779201091506,
+      "learning_rate": 7.922242491223167e-05,
+      "loss": 0.684,
+      "step": 119
+    },
+    {
+      "epoch": 0.7867240319606638,
+      "grad_norm": 1.2082599250889219,
+      "learning_rate": 7.918596268350296e-05,
+      "loss": 0.6765,
+      "step": 120
+    },
+    {
+      "epoch": 0.793280065560336,
+      "grad_norm": 1.179921751372486,
+      "learning_rate": 7.914867381309418e-05,
+      "loss": 0.6868,
+      "step": 121
+    },
+    {
+      "epoch": 0.7998360991600082,
+      "grad_norm": 0.9486398114671882,
+      "learning_rate": 7.911055908762718e-05,
+      "loss": 0.6749,
+      "step": 122
+    },
+    {
+      "epoch": 0.8063921327596804,
+      "grad_norm": 0.8000482601006661,
+      "learning_rate": 7.90716193111455e-05,
+      "loss": 0.6747,
+      "step": 123
+    },
+    {
+      "epoch": 0.8129481663593526,
+      "grad_norm": 0.9517101245313929,
+      "learning_rate": 7.903185530509743e-05,
+      "loss": 0.6794,
+      "step": 124
+    },
+    {
+      "epoch": 0.8195041999590248,
+      "grad_norm": 0.9986534633574224,
+      "learning_rate": 7.899126790831869e-05,
+      "loss": 0.6774,
+      "step": 125
+    },
+    {
+      "epoch": 0.826060233558697,
+      "grad_norm": 1.1257859217497668,
+      "learning_rate": 7.894985797701472e-05,
+      "loss": 0.6793,
+      "step": 126
+    },
+    {
+      "epoch": 0.8326162671583692,
+      "grad_norm": 1.0292173321571256,
+      "learning_rate": 7.890762638474256e-05,
+      "loss": 0.6826,
+      "step": 127
+    },
+    {
+      "epoch": 0.8391723007580414,
+      "grad_norm": 0.7557871567857994,
+      "learning_rate": 7.886457402239256e-05,
+      "loss": 0.6792,
+      "step": 128
+    },
+    {
+      "epoch": 0.8457283343577136,
+      "grad_norm": 1.10326991797423,
+      "learning_rate": 7.882070179816944e-05,
+      "loss": 0.6786,
+      "step": 129
+    },
+    {
+      "epoch": 0.8522843679573858,
+      "grad_norm": 0.6776535159011668,
+      "learning_rate": 7.877601063757323e-05,
+      "loss": 0.6769,
+      "step": 130
+    },
+    {
+      "epoch": 0.858840401557058,
+      "grad_norm": 0.8010200105501427,
+      "learning_rate": 7.873050148337967e-05,
+      "loss": 0.6748,
+      "step": 131
+    },
+    {
+      "epoch": 0.8653964351567301,
+      "grad_norm": 0.9126529329815608,
+      "learning_rate": 7.868417529562043e-05,
+      "loss": 0.6632,
+      "step": 132
+    },
+    {
+      "epoch": 0.8719524687564024,
+      "grad_norm": 1.145222649075524,
+      "learning_rate": 7.863703305156273e-05,
+      "loss": 0.6756,
+      "step": 133
+    },
+    {
+      "epoch": 0.8785085023560746,
+      "grad_norm": 1.082669097422374,
+      "learning_rate": 7.858907574568882e-05,
+      "loss": 0.6765,
+      "step": 134
+    },
+    {
+      "epoch": 0.8850645359557467,
+      "grad_norm": 1.0310987601522028,
+      "learning_rate": 7.854030438967494e-05,
+      "loss": 0.6738,
+      "step": 135
+    },
+    {
+      "epoch": 0.891620569555419,
+      "grad_norm": 1.0160706638367318,
+      "learning_rate": 7.849072001237001e-05,
+      "loss": 0.6778,
+      "step": 136
+    },
+    {
+      "epoch": 0.8981766031550912,
+      "grad_norm": 0.9374366244899177,
+      "learning_rate": 7.844032365977396e-05,
+      "loss": 0.6736,
+      "step": 137
+    },
+    {
+      "epoch": 0.9047326367547633,
+      "grad_norm": 0.7008738978429749,
+      "learning_rate": 7.838911639501557e-05,
+      "loss": 0.6781,
+      "step": 138
+    },
+    {
+      "epoch": 0.9112886703544356,
+      "grad_norm": 0.4532137724870558,
+      "learning_rate": 7.833709929833012e-05,
+      "loss": 0.6686,
+      "step": 139
+    },
+    {
+      "epoch": 0.9178447039541078,
+      "grad_norm": 0.5139137753219752,
+      "learning_rate": 7.828427346703657e-05,
+      "loss": 0.6672,
+      "step": 140
+    },
+    {
+      "epoch": 0.9244007375537799,
+      "grad_norm": 0.47106346244337904,
+      "learning_rate": 7.823064001551445e-05,
+      "loss": 0.6621,
+      "step": 141
+    },
+    {
+      "epoch": 0.9309567711534522,
+      "grad_norm": 0.3844200151143341,
+      "learning_rate": 7.81762000751803e-05,
+      "loss": 0.6669,
+      "step": 142
+    },
+    {
+      "epoch": 0.9375128047531244,
+      "grad_norm": 0.4844251889310036,
+      "learning_rate": 7.812095479446383e-05,
+      "loss": 0.6606,
+      "step": 143
+    },
+    {
+      "epoch": 0.9440688383527965,
+      "grad_norm": 0.5614156709455322,
+      "learning_rate": 7.806490533878368e-05,
+      "loss": 0.6627,
+      "step": 144
+    },
+    {
+      "epoch": 0.9506248719524688,
+      "grad_norm": 0.6295196907491479,
+      "learning_rate": 7.800805289052286e-05,
+      "loss": 0.6652,
+      "step": 145
+    },
+    {
+      "epoch": 0.957180905552141,
+      "grad_norm": 0.7172605017331375,
+      "learning_rate": 7.795039864900378e-05,
+      "loss": 0.6632,
+      "step": 146
+    },
+    {
+      "epoch": 0.9637369391518131,
+      "grad_norm": 0.9441139561703984,
+      "learning_rate": 7.789194383046295e-05,
+      "loss": 0.6681,
+      "step": 147
+    },
+    {
+      "epoch": 0.9702929727514854,
+      "grad_norm": 1.3151350146737344,
+      "learning_rate": 7.783268966802539e-05,
+      "loss": 0.6703,
+      "step": 148
+    },
+    {
+      "epoch": 0.9768490063511576,
+      "grad_norm": 0.5115168470062825,
+      "learning_rate": 7.777263741167849e-05,
+      "loss": 0.6647,
+      "step": 149
+    },
+    {
+      "epoch": 0.9834050399508297,
+      "grad_norm": 1.0163356112728301,
+      "learning_rate": 7.771178832824573e-05,
+      "loss": 0.666,
+      "step": 150
+    },
+    {
+      "epoch": 0.9899610735505019,
+      "grad_norm": 1.1732972717592203,
+      "learning_rate": 7.765014370135999e-05,
+      "loss": 0.6592,
+      "step": 151
+    },
+    {
+      "epoch": 0.9965171071501742,
+      "grad_norm": 0.7851604401937097,
+      "learning_rate": 7.758770483143634e-05,
+      "loss": 0.6592,
+      "step": 152
+    },
+    {
+      "epoch": 1.0055316533497234,
+      "grad_norm": 1.6420611019668498,
+      "learning_rate": 7.752447303564475e-05,
+      "loss": 1.1949,
+      "step": 153
+    },
+    {
+      "epoch": 1.0120876869493955,
+      "grad_norm": 0.9808444668187579,
+      "learning_rate": 7.74604496478822e-05,
+      "loss": 0.6428,
+      "step": 154
+    },
+    {
+      "epoch": 1.0186437205490677,
+      "grad_norm": 1.1426869091876994,
+      "learning_rate": 7.73956360187446e-05,
+      "loss": 0.6346,
+      "step": 155
+    },
+    {
+      "epoch": 1.02519975414874,
+      "grad_norm": 1.147145143696543,
+      "learning_rate": 7.733003351549829e-05,
+      "loss": 0.6388,
+      "step": 156
+    },
+    {
+      "epoch": 1.0317557877484123,
+      "grad_norm": 0.7817012939106086,
+      "learning_rate": 7.726364352205117e-05,
+      "loss": 0.6408,
+      "step": 157
+    },
+    {
+      "epoch": 1.0383118213480844,
+      "grad_norm": 0.694581928412185,
+      "learning_rate": 7.719646743892352e-05,
+      "loss": 0.6284,
+      "step": 158
+    },
+    {
+      "epoch": 1.0448678549477566,
+      "grad_norm": 0.6070626955744831,
+      "learning_rate": 7.712850668321846e-05,
+      "loss": 0.6405,
+      "step": 159
+    },
+    {
+      "epoch": 1.0514238885474287,
+      "grad_norm": 0.6514623348225942,
+      "learning_rate": 7.705976268859207e-05,
+      "loss": 0.6395,
+      "step": 160
+    },
+    {
+      "epoch": 1.057979922147101,
+      "grad_norm": 0.5619592930813068,
+      "learning_rate": 7.699023690522315e-05,
+      "loss": 0.629,
+      "step": 161
+    },
+    {
+      "epoch": 1.0645359557467733,
+      "grad_norm": 0.5769129425178826,
+      "learning_rate": 7.691993079978252e-05,
+      "loss": 0.6381,
+      "step": 162
+    },
+    {
+      "epoch": 1.0710919893464454,
+      "grad_norm": 0.8198780194642444,
+      "learning_rate": 7.684884585540227e-05,
+      "loss": 0.6363,
+      "step": 163
+    },
+    {
+      "epoch": 1.0776480229461176,
+      "grad_norm": 0.7916477488592911,
+      "learning_rate": 7.677698357164431e-05,
+      "loss": 0.635,
+      "step": 164
+    },
+    {
+      "epoch": 1.0842040565457898,
+      "grad_norm": 0.5931606825910369,
+      "learning_rate": 7.670434546446886e-05,
+      "loss": 0.6333,
+      "step": 165
+    },
+    {
+      "epoch": 1.090760090145462,
+      "grad_norm": 0.5877139627697113,
+      "learning_rate": 7.663093306620231e-05,
+      "loss": 0.629,
+      "step": 166
+    },
+    {
+      "epoch": 1.097316123745134,
+      "grad_norm": 0.852370483441162,
+      "learning_rate": 7.655674792550507e-05,
+      "loss": 0.6395,
+      "step": 167
+    },
+    {
+      "epoch": 1.1038721573448065,
+      "grad_norm": 1.00327271618352,
+      "learning_rate": 7.648179160733883e-05,
+      "loss": 0.6478,
+      "step": 168
+    },
+    {
+      "epoch": 1.1104281909444786,
+      "grad_norm": 1.0959509941262375,
+      "learning_rate": 7.640606569293347e-05,
+      "loss": 0.6415,
+      "step": 169
+    },
+    {
+      "epoch": 1.1169842245441508,
+      "grad_norm": 0.7671153051839551,
+      "learning_rate": 7.632957177975387e-05,
+      "loss": 0.6401,
+      "step": 170
+    },
+    {
+      "epoch": 1.123540258143823,
+      "grad_norm": 0.7544679819781906,
+      "learning_rate": 7.625231148146601e-05,
+      "loss": 0.6365,
+      "step": 171
+    },
+    {
+      "epoch": 1.1300962917434951,
+      "grad_norm": 0.7948382138503196,
+      "learning_rate": 7.61742864279031e-05,
+      "loss": 0.6363,
+      "step": 172
+    },
+    {
+      "epoch": 1.1366523253431673,
+      "grad_norm": 0.7047533773597212,
+      "learning_rate": 7.609549826503115e-05,
+      "loss": 0.6421,
+      "step": 173
+    },
+    {
+      "epoch": 1.1432083589428397,
+      "grad_norm": 0.7374204351284568,
+      "learning_rate": 7.601594865491414e-05,
+      "loss": 0.6407,
+      "step": 174
+    },
+    {
+      "epoch": 1.1497643925425118,
+      "grad_norm": 0.6911928835033512,
+      "learning_rate": 7.593563927567916e-05,
+      "loss": 0.625,
+      "step": 175
+    },
+    {
+      "epoch": 1.156320426142184,
+      "grad_norm": 0.4803947112250499,
+      "learning_rate": 7.585457182148081e-05,
+      "loss": 0.628,
+      "step": 176
+    },
+    {
+      "epoch": 1.1628764597418562,
+      "grad_norm": 0.3950061404414093,
+      "learning_rate": 7.577274800246558e-05,
+      "loss": 0.6357,
+      "step": 177
+    },
+    {
+      "epoch": 1.1694324933415283,
+      "grad_norm": 0.4280389547638801,
+      "learning_rate": 7.569016954473577e-05,
+      "loss": 0.6434,
+      "step": 178
+    },
+    {
+      "epoch": 1.1759885269412005,
+      "grad_norm": 0.5103954479058104,
+      "learning_rate": 7.560683819031298e-05,
+      "loss": 0.6325,
+      "step": 179
+    },
+    {
+      "epoch": 1.1825445605408729,
+      "grad_norm": 0.48405163068225376,
+      "learning_rate": 7.552275569710152e-05,
+      "loss": 0.621,
+      "step": 180
+    },
+    {
+      "epoch": 1.189100594140545,
+      "grad_norm": 0.49505546120752403,
+      "learning_rate": 7.543792383885113e-05,
+      "loss": 0.6335,
+      "step": 181
+    },
+    {
+      "epoch": 1.1956566277402172,
+      "grad_norm": 0.5549014381017361,
+      "learning_rate": 7.535234440511979e-05,
+      "loss": 0.6401,
+      "step": 182
+    },
+    {
+      "epoch": 1.2022126613398894,
+      "grad_norm": 0.5679718866243005,
+      "learning_rate": 7.526601920123574e-05,
+      "loss": 0.6339,
+      "step": 183
+    },
+    {
+      "epoch": 1.2087686949395615,
+      "grad_norm": 0.5837881333403254,
+      "learning_rate": 7.517895004825956e-05,
+      "loss": 0.6201,
+      "step": 184
+    },
+    {
+      "epoch": 1.2153247285392337,
+      "grad_norm": 0.5275232093842687,
+      "learning_rate": 7.509113878294572e-05,
+      "loss": 0.6313,
+      "step": 185
+    },
+    {
+      "epoch": 1.2218807621389058,
+      "grad_norm": 0.4187731695243854,
+      "learning_rate": 7.500258725770375e-05,
+      "loss": 0.6297,
+      "step": 186
+    },
+    {
+      "epoch": 1.2284367957385782,
+      "grad_norm": 0.3496822319709094,
+      "learning_rate": 7.491329734055926e-05,
+      "loss": 0.6341,
+      "step": 187
+    },
+    {
+      "epoch": 1.2349928293382504,
+      "grad_norm": 0.2892298691529975,
+      "learning_rate": 7.48232709151145e-05,
+      "loss": 0.6289,
+      "step": 188
+    },
+    {
+      "epoch": 1.2415488629379225,
+      "grad_norm": 0.3970392684873568,
+      "learning_rate": 7.473250988050861e-05,
+      "loss": 0.6327,
+      "step": 189
+    },
+    {
+      "epoch": 1.2481048965375947,
+      "grad_norm": 0.5366046878238813,
+      "learning_rate": 7.464101615137756e-05,
+      "loss": 0.6234,
+      "step": 190
+    },
+    {
+      "epoch": 1.2546609301372669,
+      "grad_norm": 0.6365895325361283,
+      "learning_rate": 7.454879165781379e-05,
+      "loss": 0.6322,
+      "step": 191
+    },
+    {
+      "epoch": 1.2612169637369393,
+      "grad_norm": 0.6390242865748291,
+      "learning_rate": 7.445583834532546e-05,
+      "loss": 0.6252,
+      "step": 192
+    },
+    {
+      "epoch": 1.2677729973366114,
+      "grad_norm": 0.7130289120451936,
+      "learning_rate": 7.436215817479541e-05,
+      "loss": 0.6369,
+      "step": 193
+    },
+    {
+      "epoch": 1.2743290309362836,
+      "grad_norm": 0.8001337986119181,
+      "learning_rate": 7.426775312243986e-05,
+      "loss": 0.6341,
+      "step": 194
+    },
+    {
+      "epoch": 1.2808850645359557,
+      "grad_norm": 0.9421645194421999,
+      "learning_rate": 7.41726251797666e-05,
+      "loss": 0.629,
+      "step": 195
+    },
+    {
+      "epoch": 1.287441098135628,
+      "grad_norm": 1.0651209082990578,
+      "learning_rate": 7.407677635353308e-05,
+      "loss": 0.6274,
+      "step": 196
+    },
+    {
+      "epoch": 1.2939971317353,
+      "grad_norm": 0.8478545298742448,
+      "learning_rate": 7.398020866570407e-05,
+      "loss": 0.6373,
+      "step": 197
+    },
+    {
+      "epoch": 1.3005531653349722,
+      "grad_norm": 0.5323333743534195,
+      "learning_rate": 7.388292415340888e-05,
+      "loss": 0.631,
+      "step": 198
+    },
+    {
+      "epoch": 1.3071091989346446,
+      "grad_norm": 0.37990592076356283,
+      "learning_rate": 7.37849248688986e-05,
+      "loss": 0.631,
+      "step": 199
+    },
+    {
+      "epoch": 1.3136652325343168,
+      "grad_norm": 0.4732271711472085,
+      "learning_rate": 7.368621287950264e-05,
+      "loss": 0.6328,
+      "step": 200
+    },
+    {
+      "epoch": 1.320221266133989,
+      "grad_norm": 0.5259491499827896,
+      "learning_rate": 7.358679026758515e-05,
+      "loss": 0.6283,
+      "step": 201
+    },
+    {
+      "epoch": 1.326777299733661,
+      "grad_norm": 0.5000279931190099,
+      "learning_rate": 7.348665913050115e-05,
+      "loss": 0.6208,
+      "step": 202
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.3912490196632365,
+      "learning_rate": 7.338582158055224e-05,
+      "loss": 0.6251,
+      "step": 203
+    },
+    {
+      "epoch": 1.3398893669330056,
+      "grad_norm": 0.38905604815288286,
+      "learning_rate": 7.328427974494201e-05,
+      "loss": 0.6179,
+      "step": 204
+    },
+    {
+      "epoch": 1.3464454005326778,
+      "grad_norm": 0.4906878706390755,
+      "learning_rate": 7.318203576573126e-05,
+      "loss": 0.6271,
+      "step": 205
+    },
+    {
+      "epoch": 1.35300143413235,
+      "grad_norm": 0.4146373419913017,
+      "learning_rate": 7.307909179979274e-05,
+      "loss": 0.6256,
+      "step": 206
+    },
+    {
+      "epoch": 1.3595574677320221,
+      "grad_norm": 0.29246282870192875,
+      "learning_rate": 7.297545001876563e-05,
+      "loss": 0.6219,
+      "step": 207
+    },
+    {
+      "epoch": 1.3661135013316943,
+      "grad_norm": 0.27459922699544653,
+      "learning_rate": 7.28711126090098e-05,
+      "loss": 0.6346,
+      "step": 208
+    },
+    {
+      "epoch": 1.3726695349313665,
+      "grad_norm": 0.3594793278431249,
+      "learning_rate": 7.276608177155968e-05,
+      "loss": 0.6234,
+      "step": 209
+    },
+    {
+      "epoch": 1.3792255685310386,
+      "grad_norm": 0.46199653506152194,
+      "learning_rate": 7.266035972207773e-05,
+      "loss": 0.6328,
+      "step": 210
+    },
+    {
+      "epoch": 1.385781602130711,
+      "grad_norm": 0.5084391547285126,
+      "learning_rate": 7.25539486908078e-05,
+      "loss": 0.6287,
+      "step": 211
+    },
+    {
+      "epoch": 1.3923376357303832,
+      "grad_norm": 0.4970090588981773,
+      "learning_rate": 7.24468509225281e-05,
+      "loss": 0.6338,
+      "step": 212
+    },
+    {
+      "epoch": 1.3988936693300553,
+      "grad_norm": 0.4297079110960391,
+      "learning_rate": 7.233906867650373e-05,
+      "loss": 0.6246,
+      "step": 213
+    },
+    {
+      "epoch": 1.4054497029297275,
+      "grad_norm": 0.4277415770961674,
+      "learning_rate": 7.223060422643914e-05,
+      "loss": 0.6235,
+      "step": 214
+    },
+    {
+      "epoch": 1.4120057365293996,
+      "grad_norm": 0.5211718034258063,
+      "learning_rate": 7.212145986043007e-05,
+      "loss": 0.626,
+      "step": 215
+    },
+    {
+      "epoch": 1.418561770129072,
+      "grad_norm": 0.7113602181910502,
+      "learning_rate": 7.201163788091536e-05,
+      "loss": 0.626,
+      "step": 216
+    },
+    {
+      "epoch": 1.4251178037287442,
+      "grad_norm": 0.8702986528721038,
+      "learning_rate": 7.190114060462837e-05,
+      "loss": 0.6285,
+      "step": 217
+    },
+    {
+      "epoch": 1.4316738373284164,
+      "grad_norm": 0.88781618603766,
+      "learning_rate": 7.178997036254799e-05,
+      "loss": 0.625,
+      "step": 218
+    },
+    {
+      "epoch": 1.4382298709280885,
+      "grad_norm": 0.8543987595137604,
+      "learning_rate": 7.167812949984966e-05,
+      "loss": 0.6369,
+      "step": 219
+    },
+    {
+      "epoch": 1.4447859045277607,
+      "grad_norm": 0.6591694598278867,
+      "learning_rate": 7.156562037585576e-05,
+      "loss": 0.6309,
+      "step": 220
+    },
+    {
+      "epoch": 1.4513419381274328,
+      "grad_norm": 0.435460183285974,
+      "learning_rate": 7.145244536398584e-05,
+      "loss": 0.6337,
+      "step": 221
+    },
+    {
+      "epoch": 1.457897971727105,
+      "grad_norm": 0.5394064377655016,
+      "learning_rate": 7.133860685170665e-05,
+      "loss": 0.6272,
+      "step": 222
+    },
+    {
+      "epoch": 1.4644540053267772,
+      "grad_norm": 0.6560512889140389,
+      "learning_rate": 7.12241072404817e-05,
+      "loss": 0.6317,
+      "step": 223
+    },
+    {
+      "epoch": 1.4710100389264495,
+      "grad_norm": 0.5108872994150431,
+      "learning_rate": 7.110894894572056e-05,
+      "loss": 0.6266,
+      "step": 224
+    },
+    {
+      "epoch": 1.4775660725261217,
+      "grad_norm": 0.4508746314491801,
+      "learning_rate": 7.099313439672806e-05,
+      "loss": 0.6222,
+      "step": 225
+    },
+    {
+      "epoch": 1.4841221061257939,
+      "grad_norm": 0.5932762714408137,
+      "learning_rate": 7.087666603665284e-05,
+      "loss": 0.6174,
+      "step": 226
+    },
+    {
+      "epoch": 1.490678139725466,
+      "grad_norm": 0.5703671444864514,
+      "learning_rate": 7.0759546322436e-05,
+      "loss": 0.6286,
+      "step": 227
+    },
+    {
+      "epoch": 1.4972341733251384,
+      "grad_norm": 0.4219759927229514,
+      "learning_rate": 7.064177772475912e-05,
+      "loss": 0.621,
+      "step": 228
+    },
+    {
+      "epoch": 1.5037902069248106,
+      "grad_norm": 0.3899488374409842,
+      "learning_rate": 7.052336272799227e-05,
+      "loss": 0.6368,
+      "step": 229
+    },
+    {
+      "epoch": 1.5103462405244827,
+      "grad_norm": 0.3352814743100072,
+      "learning_rate": 7.040430383014146e-05,
+      "loss": 0.6273,
+      "step": 230
+    },
+    {
+      "epoch": 1.516902274124155,
+      "grad_norm": 0.49330992250119887,
+      "learning_rate": 7.02846035427961e-05,
+      "loss": 0.6268,
+      "step": 231
+    },
+    {
+      "epoch": 1.523458307723827,
+      "grad_norm": 0.7275390777504656,
+      "learning_rate": 7.016426439107586e-05,
+      "loss": 0.6198,
+      "step": 232
+    },
+    {
+      "epoch": 1.5300143413234992,
+      "grad_norm": 0.5944986948743242,
+      "learning_rate": 7.004328891357753e-05,
+      "loss": 0.6321,
+      "step": 233
+    },
+    {
+      "epoch": 1.5365703749231714,
+      "grad_norm": 0.40136071416301794,
+      "learning_rate": 6.992167966232143e-05,
+      "loss": 0.6205,
+      "step": 234
+    },
+    {
+      "epoch": 1.5431264085228436,
+      "grad_norm": 0.42683320761735183,
+      "learning_rate": 6.979943920269749e-05,
+      "loss": 0.6282,
+      "step": 235
+    },
+    {
+      "epoch": 1.5496824421225157,
+      "grad_norm": 0.37233786072575653,
+      "learning_rate": 6.967657011341126e-05,
+      "loss": 0.6216,
+      "step": 236
+    },
+    {
+      "epoch": 1.556238475722188,
+      "grad_norm": 0.367274159532209,
+      "learning_rate": 6.955307498642948e-05,
+      "loss": 0.6224,
+      "step": 237
+    },
+    {
+      "epoch": 1.5627945093218603,
+      "grad_norm": 0.3660447730807658,
+      "learning_rate": 6.942895642692527e-05,
+      "loss": 0.6202,
+      "step": 238
+    },
+    {
+      "epoch": 1.5693505429215324,
+      "grad_norm": 0.38882305238122433,
+      "learning_rate": 6.930421705322339e-05,
+      "loss": 0.6195,
+      "step": 239
+    },
+    {
+      "epoch": 1.5759065765212048,
+      "grad_norm": 0.5684648344288802,
+      "learning_rate": 6.917885949674483e-05,
+      "loss": 0.6228,
+      "step": 240
+    },
+    {
+      "epoch": 1.582462610120877,
+      "grad_norm": 0.7431328174808526,
+      "learning_rate": 6.905288640195141e-05,
+      "loss": 0.626,
+      "step": 241
+    },
+    {
+      "epoch": 1.5890186437205491,
+      "grad_norm": 0.7410518651663662,
+      "learning_rate": 6.892630042628988e-05,
+      "loss": 0.6248,
+      "step": 242
+    },
+    {
+      "epoch": 1.5955746773202213,
+      "grad_norm": 0.6219837555078648,
+      "learning_rate": 6.879910424013599e-05,
+      "loss": 0.6295,
+      "step": 243
+    },
+    {
+      "epoch": 1.6021307109198935,
+      "grad_norm": 0.5396699034196659,
+      "learning_rate": 6.867130052673806e-05,
+      "loss": 0.6231,
+      "step": 244
+    },
+    {
+      "epoch": 1.6086867445195656,
+      "grad_norm": 0.4531223244262461,
+      "learning_rate": 6.854289198216042e-05,
+      "loss": 0.6346,
+      "step": 245
+    },
+    {
+      "epoch": 1.6152427781192378,
+      "grad_norm": 0.3242170520225258,
+      "learning_rate": 6.841388131522656e-05,
+      "loss": 0.6292,
+      "step": 246
+    },
+    {
+      "epoch": 1.62179881171891,
+      "grad_norm": 0.34710026048248854,
+      "learning_rate": 6.828427124746191e-05,
+      "loss": 0.6153,
+      "step": 247
+    },
+    {
+      "epoch": 1.628354845318582,
+      "grad_norm": 0.44267957645174644,
+      "learning_rate": 6.815406451303647e-05,
+      "loss": 0.6205,
+      "step": 248
+    },
+    {
+      "epoch": 1.6349108789182545,
+      "grad_norm": 0.5545408679102592,
+      "learning_rate": 6.802326385870715e-05,
+      "loss": 0.621,
+      "step": 249
+    },
+    {
+      "epoch": 1.6414669125179266,
+      "grad_norm": 0.5443247431153629,
+      "learning_rate": 6.789187204375981e-05,
+      "loss": 0.622,
+      "step": 250
+    },
+    {
+      "epoch": 1.6480229461175988,
+      "grad_norm": 0.4301197066284,
+      "learning_rate": 6.775989183995108e-05,
+      "loss": 0.6135,
+      "step": 251
+    },
+    {
+      "epoch": 1.6545789797172712,
+      "grad_norm": 0.396698152583572,
+      "learning_rate": 6.762732603144978e-05,
+      "loss": 0.6216,
+      "step": 252
+    },
+    {
+      "epoch": 1.6611350133169434,
+      "grad_norm": 0.3515261284616528,
+      "learning_rate": 6.749417741477836e-05,
+      "loss": 0.6187,
+      "step": 253
+    },
+    {
+      "epoch": 1.6676910469166155,
+      "grad_norm": 0.35858581397311556,
+      "learning_rate": 6.736044879875373e-05,
+      "loss": 0.6185,
+      "step": 254
+    },
+    {
+      "epoch": 1.6742470805162877,
+      "grad_norm": 0.34392278311565416,
+      "learning_rate": 6.722614300442815e-05,
+      "loss": 0.6154,
+      "step": 255
+    },
+    {
+      "epoch": 1.6808031141159598,
+      "grad_norm": 0.36707892916862706,
+      "learning_rate": 6.709126286502965e-05,
+      "loss": 0.6179,
+      "step": 256
+    },
+    {
+      "epoch": 1.687359147715632,
+      "grad_norm": 0.441219103876971,
+      "learning_rate": 6.695581122590225e-05,
+      "loss": 0.6227,
+      "step": 257
+    },
+    {
+      "epoch": 1.6939151813153042,
+      "grad_norm": 0.3994835806954666,
+      "learning_rate": 6.681979094444596e-05,
+      "loss": 0.6192,
+      "step": 258
+    },
+    {
+      "epoch": 1.7004712149149763,
+      "grad_norm": 0.433716188444932,
+      "learning_rate": 6.668320489005654e-05,
+      "loss": 0.622,
+      "step": 259
+    },
+    {
+      "epoch": 1.7070272485146485,
+      "grad_norm": 0.47283003533467305,
+      "learning_rate": 6.654605594406486e-05,
+      "loss": 0.6253,
+      "step": 260
+    },
+    {
+      "epoch": 1.7135832821143209,
+      "grad_norm": 0.49013239605830206,
+      "learning_rate": 6.640834699967626e-05,
+      "loss": 0.6214,
+      "step": 261
+    },
+    {
+      "epoch": 1.720139315713993,
+      "grad_norm": 0.49324672918814405,
+      "learning_rate": 6.627008096190938e-05,
+      "loss": 0.6182,
+      "step": 262
+    },
+    {
+      "epoch": 1.7266953493136652,
+      "grad_norm": 0.5169572714827575,
+      "learning_rate": 6.6131260747535e-05,
+      "loss": 0.6259,
+      "step": 263
+    },
+    {
+      "epoch": 1.7332513829133376,
+      "grad_norm": 0.5873107078500235,
+      "learning_rate": 6.59918892850144e-05,
+      "loss": 0.6235,
+      "step": 264
+    },
+    {
+      "epoch": 1.7398074165130097,
+      "grad_norm": 0.5020487886809576,
+      "learning_rate": 6.585196951443763e-05,
+      "loss": 0.6221,
+      "step": 265
+    },
+    {
+      "epoch": 1.746363450112682,
+      "grad_norm": 0.329791614699069,
+      "learning_rate": 6.571150438746157e-05,
+      "loss": 0.6265,
+      "step": 266
+    },
+    {
+      "epoch": 1.752919483712354,
+      "grad_norm": 0.2861139606661766,
+      "learning_rate": 6.557049686724751e-05,
+      "loss": 0.624,
+      "step": 267
+    },
+    {
+      "epoch": 1.7594755173120262,
+      "grad_norm": 0.46750896287166216,
+      "learning_rate": 6.542894992839873e-05,
+      "loss": 0.6241,
+      "step": 268
+    },
+    {
+      "epoch": 1.7660315509116984,
+      "grad_norm": 0.5684934196993104,
+      "learning_rate": 6.528686655689774e-05,
+      "loss": 0.6152,
+      "step": 269
+    },
+    {
+      "epoch": 1.7725875845113706,
+      "grad_norm": 0.48490900499130857,
+      "learning_rate": 6.514424975004329e-05,
+      "loss": 0.6111,
+      "step": 270
+    },
+    {
+      "epoch": 1.7791436181110427,
+      "grad_norm": 0.5506105973703118,
+      "learning_rate": 6.500110251638715e-05,
+      "loss": 0.6152,
+      "step": 271
+    },
+    {
+      "epoch": 1.7856996517107149,
+      "grad_norm": 0.4539294761333573,
+      "learning_rate": 6.48574278756706e-05,
+      "loss": 0.6273,
+      "step": 272
+    },
+    {
+      "epoch": 1.7922556853103873,
+      "grad_norm": 0.34705648209401146,
+      "learning_rate": 6.471322885876077e-05,
+      "loss": 0.6204,
+      "step": 273
+    },
+    {
+      "epoch": 1.7988117189100594,
+      "grad_norm": 0.2427469168923526,
+      "learning_rate": 6.456850850758673e-05,
+      "loss": 0.6214,
+      "step": 274
+    },
+    {
+      "epoch": 1.8053677525097316,
+      "grad_norm": 0.3004384124304078,
+      "learning_rate": 6.44232698750752e-05,
+      "loss": 0.6135,
+      "step": 275
+    },
+    {
+      "epoch": 1.811923786109404,
+      "grad_norm": 0.29782974812920376,
+      "learning_rate": 6.427751602508628e-05,
+      "loss": 0.6104,
+      "step": 276
+    },
+    {
+      "epoch": 1.8184798197090761,
+      "grad_norm": 0.29267694689244445,
+      "learning_rate": 6.413125003234876e-05,
+      "loss": 0.624,
+      "step": 277
+    },
+    {
+      "epoch": 1.8250358533087483,
+      "grad_norm": 0.3462332305273243,
+      "learning_rate": 6.398447498239527e-05,
+      "loss": 0.612,
+      "step": 278
+    },
+    {
+      "epoch": 1.8315918869084205,
+      "grad_norm": 0.3949978565357491,
+      "learning_rate": 6.383719397149715e-05,
+      "loss": 0.622,
+      "step": 279
+    },
+    {
+      "epoch": 1.8381479205080926,
+      "grad_norm": 0.40536973102729335,
+      "learning_rate": 6.368941010659921e-05,
+      "loss": 0.6118,
+      "step": 280
+    },
+    {
+      "epoch": 1.8447039541077648,
+      "grad_norm": 0.3621578404050215,
+      "learning_rate": 6.354112650525407e-05,
+      "loss": 0.6149,
+      "step": 281
+    },
+    {
+      "epoch": 1.851259987707437,
+      "grad_norm": 0.3445107311397209,
+      "learning_rate": 6.339234629555655e-05,
+      "loss": 0.6196,
+      "step": 282
+    },
+    {
+      "epoch": 1.857816021307109,
+      "grad_norm": 0.30605392165712425,
+      "learning_rate": 6.324307261607754e-05,
+      "loss": 0.6239,
+      "step": 283
+    },
+    {
+      "epoch": 1.8643720549067813,
+      "grad_norm": 0.3080701040643038,
+      "learning_rate": 6.309330861579786e-05,
+      "loss": 0.629,
+      "step": 284
+    },
+    {
+      "epoch": 1.8709280885064536,
+      "grad_norm": 0.3349937492345958,
+      "learning_rate": 6.294305745404185e-05,
+      "loss": 0.6224,
+      "step": 285
+    },
+    {
+      "epoch": 1.8774841221061258,
+      "grad_norm": 0.3396848816828842,
+      "learning_rate": 6.279232230041065e-05,
+      "loss": 0.6182,
+      "step": 286
+    },
+    {
+      "epoch": 1.884040155705798,
+      "grad_norm": 0.3410093342010557,
+      "learning_rate": 6.26411063347154e-05,
+      "loss": 0.6188,
+      "step": 287
+    },
+    {
+      "epoch": 1.8905961893054704,
+      "grad_norm": 0.314163527323835,
+      "learning_rate": 6.248941274691017e-05,
+      "loss": 0.6169,
+      "step": 288
+    },
+    {
+      "epoch": 1.8971522229051425,
+      "grad_norm": 0.3976946162007582,
+      "learning_rate": 6.233724473702457e-05,
+      "loss": 0.6195,
+      "step": 289
+    },
+    {
+      "epoch": 1.9037082565048147,
+      "grad_norm": 0.5601353024746372,
+      "learning_rate": 6.218460551509636e-05,
+      "loss": 0.6206,
+      "step": 290
+    },
+    {
+      "epoch": 1.9102642901044868,
+      "grad_norm": 0.6132983016524085,
+      "learning_rate": 6.203149830110367e-05,
+      "loss": 0.6138,
+      "step": 291
+    },
+    {
+      "epoch": 1.916820323704159,
+      "grad_norm": 0.5021407478552538,
+      "learning_rate": 6.18779263248971e-05,
+      "loss": 0.6219,
+      "step": 292
+    },
+    {
+      "epoch": 1.9233763573038312,
+      "grad_norm": 0.3890448472238098,
+      "learning_rate": 6.172389282613151e-05,
+      "loss": 0.6187,
+      "step": 293
+    },
+    {
+      "epoch": 1.9299323909035033,
+      "grad_norm": 0.3361740572901459,
+      "learning_rate": 6.156940105419785e-05,
+      "loss": 0.6218,
+      "step": 294
+    },
+    {
+      "epoch": 1.9364884245031755,
+      "grad_norm": 0.2907876046977178,
+      "learning_rate": 6.141445426815443e-05,
+      "loss": 0.6166,
+      "step": 295
+    },
+    {
+      "epoch": 1.9430444581028476,
+      "grad_norm": 0.2632987500846716,
+      "learning_rate": 6.125905573665824e-05,
+      "loss": 0.6232,
+      "step": 296
+    },
+    {
+      "epoch": 1.9496004917025198,
+      "grad_norm": 0.32120831457165155,
+      "learning_rate": 6.110320873789604e-05,
+      "loss": 0.6242,
+      "step": 297
+    },
+    {
+      "epoch": 1.9561565253021922,
+      "grad_norm": 0.32170026271883145,
+      "learning_rate": 6.094691655951512e-05,
+      "loss": 0.6094,
+      "step": 298
+    },
+    {
+      "epoch": 1.9627125589018644,
+      "grad_norm": 0.27126547953075625,
+      "learning_rate": 6.079018249855402e-05,
+      "loss": 0.622,
+      "step": 299
+    },
+    {
+      "epoch": 1.9692685925015365,
+      "grad_norm": 0.25823866111956445,
+      "learning_rate": 6.063300986137297e-05,
+      "loss": 0.6195,
+      "step": 300
+    },
+    {
+      "epoch": 1.975824626101209,
+      "grad_norm": 0.2777978831118673,
+      "learning_rate": 6.047540196358405e-05,
+      "loss": 0.6188,
+      "step": 301
+    },
+    {
+      "epoch": 1.982380659700881,
+      "grad_norm": 0.3097245776135163,
+      "learning_rate": 6.0317362129981375e-05,
+      "loss": 0.6161,
+      "step": 302
+    },
+    {
+      "epoch": 1.9889366933005532,
+      "grad_norm": 0.3108082557836966,
+      "learning_rate": 6.015889369447088e-05,
+      "loss": 0.6166,
+      "step": 303
+    },
+    {
+      "epoch": 1.9954927269002254,
+      "grad_norm": 0.2659644339645489,
+      "learning_rate": 6.000000000000001e-05,
+      "loss": 0.6143,
+      "step": 304
+    },
+    {
+      "epoch": 2.006556033599672,
+      "grad_norm": 0.3001511379848704,
+      "learning_rate": 5.9840684398487186e-05,
+      "loss": 0.5867,
+      "step": 305
+    },
+    {
+      "epoch": 2.0131120671993443,
+      "grad_norm": 0.3404835526439102,
+      "learning_rate": 5.968095025075114e-05,
+      "loss": 0.5745,
+      "step": 306
+    },
+    {
+      "epoch": 2.0196681007990165,
+      "grad_norm": 0.5738437895797627,
+      "learning_rate": 5.952080092643993e-05,
+      "loss": 0.5828,
+      "step": 307
+    },
+    {
+      "epoch": 2.0262241343986886,
+      "grad_norm": 0.7841436327218428,
+      "learning_rate": 5.936023980395997e-05,
+      "loss": 0.583,
+      "step": 308
+    },
+    {
+      "epoch": 2.032780167998361,
+      "grad_norm": 0.7382363346356913,
+      "learning_rate": 5.919927027040463e-05,
+      "loss": 0.5795,
+      "step": 309
+    },
+    {
+      "epoch": 2.039336201598033,
+      "grad_norm": 0.6545848867535631,
+      "learning_rate": 5.903789572148295e-05,
+      "loss": 0.5835,
+      "step": 310
+    },
+    {
+      "epoch": 2.0458922351977056,
+      "grad_norm": 0.516762214742333,
+      "learning_rate": 5.887611956144782e-05,
+      "loss": 0.5787,
+      "step": 311
+    },
+    {
+      "epoch": 2.0524482687973777,
+      "grad_norm": 0.5327353778668117,
+      "learning_rate": 5.871394520302432e-05,
+      "loss": 0.5742,
+      "step": 312
+    },
+    {
+      "epoch": 2.05900430239705,
+      "grad_norm": 0.5080742962254841,
+      "learning_rate": 5.8551376067337626e-05,
+      "loss": 0.5737,
+      "step": 313
+    },
+    {
+      "epoch": 2.065560335996722,
+      "grad_norm": 0.5649012551531187,
+      "learning_rate": 5.838841558384091e-05,
+      "loss": 0.5764,
+      "step": 314
+    },
+    {
+      "epoch": 2.0721163695963942,
+      "grad_norm": 0.5820350754778492,
+      "learning_rate": 5.8225067190242925e-05,
+      "loss": 0.5716,
+      "step": 315
+    },
+    {
+      "epoch": 2.0786724031960664,
+      "grad_norm": 0.46076842055118317,
+      "learning_rate": 5.806133433243558e-05,
+      "loss": 0.5753,
+      "step": 316
+    },
+    {
+      "epoch": 2.0852284367957385,
+      "grad_norm": 0.47933858405340773,
+      "learning_rate": 5.789722046442114e-05,
+      "loss": 0.575,
+      "step": 317
+    },
+    {
+      "epoch": 2.0917844703954107,
+      "grad_norm": 0.5168151561479127,
+      "learning_rate": 5.7732729048239444e-05,
+      "loss": 0.5749,
+      "step": 318
+    },
+    {
+      "epoch": 2.098340503995083,
+      "grad_norm": 0.40840899989073715,
+      "learning_rate": 5.756786355389482e-05,
+      "loss": 0.5802,
+      "step": 319
+    },
+    {
+      "epoch": 2.104896537594755,
+      "grad_norm": 0.44997063706673934,
+      "learning_rate": 5.740262745928293e-05,
+      "loss": 0.5717,
+      "step": 320
+    },
+    {
+      "epoch": 2.111452571194427,
+      "grad_norm": 0.5369629713906064,
+      "learning_rate": 5.723702425011738e-05,
+      "loss": 0.5751,
+      "step": 321
+    },
+    {
+      "epoch": 2.1180086047940994,
+      "grad_norm": 0.4475818760429054,
+      "learning_rate": 5.707105741985615e-05,
+      "loss": 0.5765,
+      "step": 322
+    },
+    {
+      "epoch": 2.124564638393772,
+      "grad_norm": 0.30932774384801376,
+      "learning_rate": 5.6904730469627985e-05,
+      "loss": 0.5785,
+      "step": 323
+    },
+    {
+      "epoch": 2.131120671993444,
+      "grad_norm": 0.33040348186144713,
+      "learning_rate": 5.673804690815845e-05,
+      "loss": 0.5768,
+      "step": 324
+    },
+    {
+      "epoch": 2.1376767055931163,
+      "grad_norm": 0.3576099667078179,
+      "learning_rate": 5.6571010251695954e-05,
+      "loss": 0.5815,
+      "step": 325
+    },
+    {
+      "epoch": 2.1442327391927885,
+      "grad_norm": 0.4237003547581017,
+      "learning_rate": 5.6403624023937614e-05,
+      "loss": 0.5747,
+      "step": 326
+    },
+    {
+      "epoch": 2.1507887727924606,
+      "grad_norm": 0.4700100373051473,
+      "learning_rate": 5.62358917559548e-05,
+      "loss": 0.5751,
+      "step": 327
+    },
+    {
+      "epoch": 2.1573448063921328,
+      "grad_norm": 0.39617955284785017,
+      "learning_rate": 5.606781698611879e-05,
+      "loss": 0.5798,
+      "step": 328
+    },
+    {
+      "epoch": 2.163900839991805,
+      "grad_norm": 0.28548306735776896,
+      "learning_rate": 5.5899403260026006e-05,
+      "loss": 0.5724,
+      "step": 329
+    },
+    {
+      "epoch": 2.170456873591477,
+      "grad_norm": 0.3315286342160672,
+      "learning_rate": 5.573065413042333e-05,
+      "loss": 0.5721,
+      "step": 330
+    },
+    {
+      "epoch": 2.1770129071911493,
+      "grad_norm": 0.3343441243180511,
+      "learning_rate": 5.556157315713305e-05,
+      "loss": 0.5783,
+      "step": 331
+    },
+    {
+      "epoch": 2.1835689407908214,
+      "grad_norm": 0.29129418213987884,
+      "learning_rate": 5.5392163906977835e-05,
+      "loss": 0.577,
+      "step": 332
+    },
+    {
+      "epoch": 2.1901249743904936,
+      "grad_norm": 0.25286013518278194,
+      "learning_rate": 5.522242995370545e-05,
+      "loss": 0.5698,
+      "step": 333
+    },
+    {
+      "epoch": 2.1966810079901657,
+      "grad_norm": 0.5001313540973934,
+      "learning_rate": 5.505237487791343e-05,
+      "loss": 0.5915,
+      "step": 334
+    },
+    {
+      "epoch": 2.2032370415898384,
+      "grad_norm": 0.2322461169124981,
+      "learning_rate": 5.488200226697345e-05,
+      "loss": 0.5707,
+      "step": 335
+    },
+    {
+      "epoch": 2.2097930751895105,
+      "grad_norm": 0.24729941417632514,
+      "learning_rate": 5.471131571495574e-05,
+      "loss": 0.5688,
+      "step": 336
+    },
+    {
+      "epoch": 2.2163491087891827,
+      "grad_norm": 0.25095902176236934,
+      "learning_rate": 5.454031882255319e-05,
+      "loss": 0.5804,
+      "step": 337
+    },
+    {
+      "epoch": 2.222905142388855,
+      "grad_norm": 0.26393993849887953,
+      "learning_rate": 5.4369015197005506e-05,
+      "loss": 0.5741,
+      "step": 338
+    },
+    {
+      "epoch": 2.229461175988527,
+      "grad_norm": 0.32581537912637687,
+      "learning_rate": 5.419740845202292e-05,
+      "loss": 0.5841,
+      "step": 339
+    },
+    {
+      "epoch": 2.236017209588199,
+      "grad_norm": 0.2928430349319568,
+      "learning_rate": 5.4025502207710184e-05,
+      "loss": 0.5763,
+      "step": 340
+    },
+    {
+      "epoch": 2.2425732431878713,
+      "grad_norm": 0.28344192622339104,
+      "learning_rate": 5.385330009049003e-05,
+      "loss": 0.5748,
+      "step": 341
+    },
+    {
+      "epoch": 2.2491292767875435,
+      "grad_norm": 0.2108350778995995,
+      "learning_rate": 5.368080573302676e-05,
+      "loss": 0.5677,
+      "step": 342
+    },
+    {
+      "epoch": 2.2556853103872156,
+      "grad_norm": 0.2654969873255356,
+      "learning_rate": 5.3508022774149574e-05,
+      "loss": 0.5759,
+      "step": 343
+    },
+    {
+      "epoch": 2.262241343986888,
+      "grad_norm": 0.26893594961047856,
+      "learning_rate": 5.333495485877583e-05,
+      "loss": 0.5713,
+      "step": 344
+    },
+    {
+      "epoch": 2.26879737758656,
+      "grad_norm": 0.2359706290249212,
+      "learning_rate": 5.3161605637834135e-05,
+      "loss": 0.5826,
+      "step": 345
+    },
+    {
+      "epoch": 2.2753534111862326,
+      "grad_norm": 0.2001826448668912,
+      "learning_rate": 5.298797876818735e-05,
+      "loss": 0.5828,
+      "step": 346
+    },
+    {
+      "epoch": 2.2819094447859047,
+      "grad_norm": 0.18687800048623612,
+      "learning_rate": 5.2814077912555415e-05,
+      "loss": 0.5674,
+      "step": 347
+    },
+    {
+      "epoch": 2.288465478385577,
+      "grad_norm": 0.18869821797720956,
+      "learning_rate": 5.263990673943811e-05,
+      "loss": 0.5795,
+      "step": 348
+    },
+    {
+      "epoch": 2.295021511985249,
+      "grad_norm": 0.18002946746844742,
+      "learning_rate": 5.246546892303766e-05,
+      "loss": 0.5766,
+      "step": 349
+    },
+    {
+      "epoch": 2.3015775455849212,
+      "grad_norm": 0.2532930596607861,
+      "learning_rate": 5.229076814318122e-05,
+      "loss": 0.5742,
+      "step": 350
+    },
+    {
+      "epoch": 2.3081335791845934,
+      "grad_norm": 0.3187321065178083,
+      "learning_rate": 5.211580808524325e-05,
+      "loss": 0.5739,
+      "step": 351
+    },
+    {
+      "epoch": 2.3146896127842655,
+      "grad_norm": 0.3465322974483502,
+      "learning_rate": 5.194059244006779e-05,
+      "loss": 0.5699,
+      "step": 352
+    },
+    {
+      "epoch": 2.3212456463839377,
+      "grad_norm": 0.48722607810944585,
+      "learning_rate": 5.176512490389055e-05,
+      "loss": 0.5756,
+      "step": 353
+    },
+    {
+      "epoch": 2.32780167998361,
+      "grad_norm": 1.1074199903854496,
+      "learning_rate": 5.158940917826099e-05,
+      "loss": 0.5832,
+      "step": 354
+    },
+    {
+      "epoch": 2.334357713583282,
+      "grad_norm": 0.3988094901700725,
+      "learning_rate": 5.141344896996422e-05,
+      "loss": 0.5871,
+      "step": 355
+    },
+    {
+      "epoch": 2.340913747182954,
+      "grad_norm": 0.3381422893134184,
+      "learning_rate": 5.123724799094279e-05,
+      "loss": 0.5908,
+      "step": 356
+    },
+    {
+      "epoch": 2.3474697807826264,
+      "grad_norm": 0.30952489856697063,
+      "learning_rate": 5.106080995821836e-05,
+      "loss": 0.5763,
+      "step": 357
+    },
+    {
+      "epoch": 2.3540258143822985,
+      "grad_norm": 0.3302426550471252,
+      "learning_rate": 5.088413859381341e-05,
+      "loss": 0.5796,
+      "step": 358
+    },
+    {
+      "epoch": 2.3605818479819707,
+      "grad_norm": 0.2513297320397786,
+      "learning_rate": 5.070723762467254e-05,
+      "loss": 0.5749,
+      "step": 359
+    },
+    {
+      "epoch": 2.3671378815816433,
+      "grad_norm": 0.26942873256331556,
+      "learning_rate": 5.053011078258397e-05,
+      "loss": 0.5782,
+      "step": 360
+    },
+    {
+      "epoch": 2.3736939151813155,
+      "grad_norm": 0.26811935982451485,
+      "learning_rate": 5.0352761804100835e-05,
+      "loss": 0.5893,
+      "step": 361
+    },
+    {
+      "epoch": 2.3802499487809876,
+      "grad_norm": 0.24961381636090246,
+      "learning_rate": 5.017519443046226e-05,
+      "loss": 0.5752,
+      "step": 362
+    },
+    {
+      "epoch": 2.3868059823806598,
+      "grad_norm": 0.2789479751757785,
+      "learning_rate": 4.999741240751451e-05,
+      "loss": 0.5819,
+      "step": 363
+    },
+    {
+      "epoch": 2.393362015980332,
+      "grad_norm": 0.23968714486879794,
+      "learning_rate": 4.981941948563197e-05,
+      "loss": 0.5864,
+      "step": 364
+    },
+    {
+      "epoch": 2.399918049580004,
+      "grad_norm": 0.3892339244230853,
+      "learning_rate": 4.9641219419637985e-05,
+      "loss": 0.589,
+      "step": 365
+    },
+    {
+      "epoch": 2.4064740831796763,
+      "grad_norm": 0.19471837418765098,
+      "learning_rate": 4.94628159687257e-05,
+      "loss": 0.5841,
+      "step": 366
+    },
+    {
+      "epoch": 2.4130301167793484,
+      "grad_norm": 0.19750205892208292,
+      "learning_rate": 4.928421289637871e-05,
+      "loss": 0.5735,
+      "step": 367
+    },
+    {
+      "epoch": 2.4195861503790206,
+      "grad_norm": 0.2849687278419682,
+      "learning_rate": 4.9105413970291747e-05,
+      "loss": 0.5806,
+      "step": 368
+    },
+    {
+      "epoch": 2.4261421839786927,
+      "grad_norm": 0.4039957657566193,
+      "learning_rate": 4.892642296229107e-05,
+      "loss": 0.5802,
+      "step": 369
+    },
+    {
+      "epoch": 2.432698217578365,
+      "grad_norm": 0.339476402567376,
+      "learning_rate": 4.874724364825504e-05,
+      "loss": 0.583,
+      "step": 370
+    },
+    {
+      "epoch": 2.4392542511780375,
+      "grad_norm": 0.2352273125693621,
+      "learning_rate": 4.856787980803437e-05,
+      "loss": 0.5822,
+      "step": 371
+    },
+    {
+      "epoch": 2.4458102847777097,
+      "grad_norm": 0.20048904331921943,
+      "learning_rate": 4.8388335225372416e-05,
+      "loss": 0.5865,
+      "step": 372
+    },
+    {
+      "epoch": 2.452366318377382,
+      "grad_norm": 0.20167869688700724,
+      "learning_rate": 4.820861368782537e-05,
+      "loss": 0.5753,
+      "step": 373
+    },
+    {
+      "epoch": 2.458922351977054,
+      "grad_norm": 0.30771705944153094,
+      "learning_rate": 4.802871898668237e-05,
+      "loss": 0.5888,
+      "step": 374
+    },
+    {
+      "epoch": 2.465478385576726,
+      "grad_norm": 0.38232619534443235,
+      "learning_rate": 4.7848654916885446e-05,
+      "loss": 0.5729,
+      "step": 375
+    },
+    {
+      "epoch": 2.4720344191763983,
+      "grad_norm": 0.2297372798599205,
+      "learning_rate": 4.7668425276949546e-05,
+      "loss": 0.5806,
+      "step": 376
+    },
+    {
+      "epoch": 2.4785904527760705,
+      "grad_norm": 0.22126295840938795,
+      "learning_rate": 4.74880338688824e-05,
+      "loss": 0.5761,
+      "step": 377
+    },
+    {
+      "epoch": 2.4851464863757426,
+      "grad_norm": 0.2882635600513661,
+      "learning_rate": 4.730748449810429e-05,
+      "loss": 0.577,
+      "step": 378
+    },
+    {
+      "epoch": 2.491702519975415,
+      "grad_norm": 0.306926273041758,
+      "learning_rate": 4.712678097336773e-05,
+      "loss": 0.5822,
+      "step": 379
+    },
+    {
+      "epoch": 2.498258553575087,
+      "grad_norm": 0.22726474030427726,
+      "learning_rate": 4.694592710667723e-05,
+      "loss": 0.5795,
+      "step": 380
+    },
+    {
+      "epoch": 2.504814587174759,
+      "grad_norm": 0.25322867792835857,
+      "learning_rate": 4.6764926713208756e-05,
+      "loss": 0.5762,
+      "step": 381
+    },
+    {
+      "epoch": 2.5113706207744313,
+      "grad_norm": 0.22298267407895275,
+      "learning_rate": 4.658378361122936e-05,
+      "loss": 0.5737,
+      "step": 382
+    },
+    {
+      "epoch": 2.5179266543741035,
+      "grad_norm": 0.20258291898015557,
+      "learning_rate": 4.640250162201656e-05,
+      "loss": 0.5787,
+      "step": 383
+    },
+    {
+      "epoch": 2.5244826879737756,
+      "grad_norm": 0.22752302168713118,
+      "learning_rate": 4.622108456977773e-05,
+      "loss": 0.58,
+      "step": 384
+    },
+    {
+      "epoch": 2.5310387215734482,
+      "grad_norm": 0.7757782276758474,
+      "learning_rate": 4.6039536281569476e-05,
+      "loss": 0.5823,
+      "step": 385
+    },
+    {
+      "epoch": 2.5375947551731204,
+      "grad_norm": 0.2825695671491106,
+      "learning_rate": 4.585786058721687e-05,
+      "loss": 0.5617,
+      "step": 386
+    },
+    {
+      "epoch": 2.5441507887727925,
+      "grad_norm": 0.18948706553634664,
+      "learning_rate": 4.567606131923263e-05,
+      "loss": 0.5747,
+      "step": 387
+    },
+    {
+      "epoch": 2.5507068223724647,
+      "grad_norm": 0.21721693917710894,
+      "learning_rate": 4.549414231273633e-05,
+      "loss": 0.5829,
+      "step": 388
+    },
+    {
+      "epoch": 2.557262855972137,
+      "grad_norm": 0.2383552245325273,
+      "learning_rate": 4.531210740537347e-05,
+      "loss": 0.5845,
+      "step": 389
+    },
+    {
+      "epoch": 2.563818889571809,
+      "grad_norm": 0.30583160784152247,
+      "learning_rate": 4.512996043723453e-05,
+      "loss": 0.5865,
+      "step": 390
+    },
+    {
+      "epoch": 2.570374923171481,
+      "grad_norm": 0.2812991313238803,
+      "learning_rate": 4.494770525077392e-05,
+      "loss": 0.5886,
+      "step": 391
+    },
+    {
+      "epoch": 2.5769309567711534,
+      "grad_norm": 0.2573334610716088,
+      "learning_rate": 4.476534569072895e-05,
+      "loss": 0.5812,
+      "step": 392
+    },
+    {
+      "epoch": 2.5834869903708255,
+      "grad_norm": 0.18065438415446586,
+      "learning_rate": 4.458288560403878e-05,
+      "loss": 0.5745,
+      "step": 393
+    },
+    {
+      "epoch": 2.590043023970498,
+      "grad_norm": 0.2944192743209924,
+      "learning_rate": 4.440032883976318e-05,
+      "loss": 0.5771,
+      "step": 394
+    },
+    {
+      "epoch": 2.5965990575701703,
+      "grad_norm": 0.23346564334137798,
+      "learning_rate": 4.421767924900136e-05,
+      "loss": 0.5773,
+      "step": 395
+    },
+    {
+      "epoch": 2.6031550911698424,
+      "grad_norm": 0.2274103844648281,
+      "learning_rate": 4.403494068481074e-05,
+      "loss": 0.5792,
+      "step": 396
+    },
+    {
+      "epoch": 2.6097111247695146,
+      "grad_norm": 0.18276656304603,
+      "learning_rate": 4.385211700212567e-05,
+      "loss": 0.5858,
+      "step": 397
+    },
+    {
+      "epoch": 2.6162671583691868,
+      "grad_norm": 0.18285901538118296,
+      "learning_rate": 4.3669212057676145e-05,
+      "loss": 0.5799,
+      "step": 398
+    },
+    {
+      "epoch": 2.622823191968859,
+      "grad_norm": 0.18097190512177241,
+      "learning_rate": 4.348622970990634e-05,
+      "loss": 0.5739,
+      "step": 399
+    },
+    {
+      "epoch": 2.629379225568531,
+      "grad_norm": 0.1930953632606967,
+      "learning_rate": 4.33031738188933e-05,
+      "loss": 0.5753,
+      "step": 400
+    },
+    {
+      "epoch": 2.6359352591682033,
+      "grad_norm": 0.17356931537284068,
+      "learning_rate": 4.312004824626551e-05,
+      "loss": 0.5777,
+      "step": 401
+    },
+    {
+      "epoch": 2.6424912927678754,
+      "grad_norm": 0.1734706404199846,
+      "learning_rate": 4.293685685512142e-05,
+      "loss": 0.5752,
+      "step": 402
+    },
+    {
+      "epoch": 2.6490473263675476,
+      "grad_norm": 0.19379812038615699,
+      "learning_rate": 4.275360350994791e-05,
+      "loss": 0.5795,
+      "step": 403
+    },
+    {
+      "epoch": 2.6556033599672197,
+      "grad_norm": 0.1660813082194996,
+      "learning_rate": 4.257029207653881e-05,
+      "loss": 0.5826,
+      "step": 404
+    },
+    {
+      "epoch": 2.662159393566892,
+      "grad_norm": 0.19954108364376924,
+      "learning_rate": 4.238692642191336e-05,
+      "loss": 0.5825,
+      "step": 405
+    },
+    {
+      "epoch": 2.668715427166564,
+      "grad_norm": 0.19322396699850317,
+      "learning_rate": 4.220351041423462e-05,
+      "loss": 0.5841,
+      "step": 406
+    },
+    {
+      "epoch": 2.6752714607662362,
+      "grad_norm": 0.17735120440062302,
+      "learning_rate": 4.202004792272785e-05,
+      "loss": 0.5798,
+      "step": 407
+    },
+    {
+      "epoch": 2.6818274943659084,
+      "grad_norm": 0.17855572539839387,
+      "learning_rate": 4.183654281759888e-05,
+      "loss": 0.5785,
+      "step": 408
+    },
+    {
+      "epoch": 2.6883835279655806,
+      "grad_norm": 0.1998138629729567,
+      "learning_rate": 4.165299896995253e-05,
+      "loss": 0.5782,
+      "step": 409
+    },
+    {
+      "epoch": 2.694939561565253,
+      "grad_norm": 0.19733364770879444,
+      "learning_rate": 4.1469420251710905e-05,
+      "loss": 0.5738,
+      "step": 410
+    },
+    {
+      "epoch": 2.7014955951649253,
+      "grad_norm": 0.17455960566932077,
+      "learning_rate": 4.128581053553169e-05,
+      "loss": 0.5804,
+      "step": 411
+    },
+    {
+      "epoch": 2.7080516287645975,
+      "grad_norm": 0.2277904402914904,
+      "learning_rate": 4.110217369472649e-05,
+      "loss": 0.5823,
+      "step": 412
+    },
+    {
+      "epoch": 2.7146076623642696,
+      "grad_norm": 0.17727829284042818,
+      "learning_rate": 4.091851360317912e-05,
+      "loss": 0.5806,
+      "step": 413
+    },
+    {
+      "epoch": 2.721163695963942,
+      "grad_norm": 0.20033803661764574,
+      "learning_rate": 4.07348341352639e-05,
+      "loss": 0.5737,
+      "step": 414
+    },
+    {
+      "epoch": 2.727719729563614,
+      "grad_norm": 0.28585140789050134,
+      "learning_rate": 4.055113916576386e-05,
+      "loss": 0.5922,
+      "step": 415
+    },
+    {
+      "epoch": 2.734275763163286,
+      "grad_norm": 0.1770226397602663,
+      "learning_rate": 4.0367432569789065e-05,
+      "loss": 0.5708,
+      "step": 416
+    },
+    {
+      "epoch": 2.7408317967629583,
+      "grad_norm": 0.19810748724107435,
+      "learning_rate": 4.0183718222694823e-05,
+      "loss": 0.5822,
+      "step": 417
+    },
+    {
+      "epoch": 2.7473878303626305,
+      "grad_norm": 0.16916185476319684,
+      "learning_rate": 4e-05,
+      "loss": 0.5781,
+      "step": 418
+    },
+    {
+      "epoch": 2.753943863962303,
+      "grad_norm": 0.1832943105404772,
+      "learning_rate": 3.9816281777305176e-05,
+      "loss": 0.5812,
+      "step": 419
+    },
+    {
+      "epoch": 2.7604998975619752,
+      "grad_norm": 0.1502272563296327,
+      "learning_rate": 3.963256743021095e-05,
+      "loss": 0.5716,
+      "step": 420
+    },
+    {
+      "epoch": 2.7670559311616474,
+      "grad_norm": 0.21339511828926633,
+      "learning_rate": 3.944886083423615e-05,
+      "loss": 0.5801,
+      "step": 421
+    },
+    {
+      "epoch": 2.7736119647613195,
+      "grad_norm": 0.177457878421534,
+      "learning_rate": 3.92651658647361e-05,
+      "loss": 0.5747,
+      "step": 422
+    },
+    {
+      "epoch": 2.7801679983609917,
+      "grad_norm": 0.1839395596929418,
+      "learning_rate": 3.908148639682089e-05,
+      "loss": 0.5843,
+      "step": 423
+    },
+    {
+      "epoch": 2.786724031960664,
+      "grad_norm": 0.19753047943291085,
+      "learning_rate": 3.889782630527353e-05,
+      "loss": 0.574,
+      "step": 424
+    },
+    {
+      "epoch": 2.793280065560336,
+      "grad_norm": 0.14640408330908058,
+      "learning_rate": 3.8714189464468334e-05,
+      "loss": 0.5786,
+      "step": 425
+    },
+    {
+      "epoch": 2.799836099160008,
+      "grad_norm": 0.19103262838391485,
+      "learning_rate": 3.853057974828911e-05,
+      "loss": 0.5798,
+      "step": 426
+    },
+    {
+      "epoch": 2.8063921327596804,
+      "grad_norm": 0.1673182640044429,
+      "learning_rate": 3.834700103004747e-05,
+      "loss": 0.5729,
+      "step": 427
+    },
+    {
+      "epoch": 2.8129481663593525,
+      "grad_norm": 0.20105657869010776,
+      "learning_rate": 3.816345718240113e-05,
+      "loss": 0.5744,
+      "step": 428
+    },
+    {
+      "epoch": 2.8195041999590247,
+      "grad_norm": 0.13814925979553985,
+      "learning_rate": 3.797995207727217e-05,
+      "loss": 0.5694,
+      "step": 429
+    },
+    {
+      "epoch": 2.826060233558697,
+      "grad_norm": 0.16253212459987165,
+      "learning_rate": 3.779648958576538e-05,
+      "loss": 0.5732,
+      "step": 430
+    },
+    {
+      "epoch": 2.832616267158369,
+      "grad_norm": 0.14306544183315967,
+      "learning_rate": 3.7613073578086644e-05,
+      "loss": 0.5737,
+      "step": 431
+    },
+    {
+      "epoch": 2.839172300758041,
+      "grad_norm": 0.16148301199590984,
+      "learning_rate": 3.74297079234612e-05,
+      "loss": 0.5696,
+      "step": 432
+    },
+    {
+      "epoch": 2.8457283343577133,
+      "grad_norm": 0.1591738101867343,
+      "learning_rate": 3.7246396490052117e-05,
+      "loss": 0.5752,
+      "step": 433
+    },
+    {
+      "epoch": 2.852284367957386,
+      "grad_norm": 0.14229281054940401,
+      "learning_rate": 3.706314314487859e-05,
+      "loss": 0.5778,
+      "step": 434
+    },
+    {
+      "epoch": 2.858840401557058,
+      "grad_norm": 0.18401979626969228,
+      "learning_rate": 3.687995175373449e-05,
+      "loss": 0.5785,
+      "step": 435
+    },
+    {
+      "epoch": 2.8653964351567303,
+      "grad_norm": 0.14532042872787476,
+      "learning_rate": 3.669682618110671e-05,
+      "loss": 0.5717,
+      "step": 436
+    },
+    {
+      "epoch": 2.8719524687564024,
+      "grad_norm": 0.14280350348324158,
+      "learning_rate": 3.6513770290093674e-05,
+      "loss": 0.5737,
+      "step": 437
+    },
+    {
+      "epoch": 2.8785085023560746,
+      "grad_norm": 0.17828968160150296,
+      "learning_rate": 3.6330787942323855e-05,
+      "loss": 0.5739,
+      "step": 438
+    },
+    {
+      "epoch": 2.8850645359557467,
+      "grad_norm": 0.1680340039645173,
+      "learning_rate": 3.614788299787434e-05,
+      "loss": 0.5719,
+      "step": 439
+    },
+    {
+      "epoch": 2.891620569555419,
+      "grad_norm": 0.16773501838978377,
+      "learning_rate": 3.5965059315189274e-05,
+      "loss": 0.5861,
+      "step": 440
+    },
+    {
+      "epoch": 2.898176603155091,
+      "grad_norm": 0.1466119984148444,
+      "learning_rate": 3.578232075099866e-05,
+      "loss": 0.5719,
+      "step": 441
+    },
+    {
+      "epoch": 2.9047326367547632,
+      "grad_norm": 0.18860006392370388,
+      "learning_rate": 3.559967116023683e-05,
+      "loss": 0.573,
+      "step": 442
+    },
+    {
+      "epoch": 2.911288670354436,
+      "grad_norm": 0.17180023728534166,
+      "learning_rate": 3.541711439596122e-05,
+      "loss": 0.5715,
+      "step": 443
+    },
+    {
+      "epoch": 2.917844703954108,
+      "grad_norm": 0.16863095150334853,
+      "learning_rate": 3.523465430927106e-05,
+      "loss": 0.587,
+      "step": 444
+    },
+    {
+      "epoch": 2.92440073755378,
+      "grad_norm": 0.18865259295148487,
+      "learning_rate": 3.5052294749226094e-05,
+      "loss": 0.5705,
+      "step": 445
+    },
+    {
+      "epoch": 2.9309567711534523,
+      "grad_norm": 0.3900404738786457,
+      "learning_rate": 3.4870039562765475e-05,
+      "loss": 0.5808,
+      "step": 446
+    },
+    {
+      "epoch": 2.9375128047531245,
+      "grad_norm": 0.1577690647524753,
+      "learning_rate": 3.4687892594626536e-05,
+      "loss": 0.5767,
+      "step": 447
+    },
+    {
+      "epoch": 2.9440688383527966,
+      "grad_norm": 0.17930301136018978,
+      "learning_rate": 3.4505857687263675e-05,
+      "loss": 0.5794,
+      "step": 448
+    },
+    {
+      "epoch": 2.950624871952469,
+      "grad_norm": 0.21442963322771297,
+      "learning_rate": 3.432393868076739e-05,
+      "loss": 0.5813,
+      "step": 449
+    },
+    {
+      "epoch": 2.957180905552141,
+      "grad_norm": 0.5263365227065998,
+      "learning_rate": 3.414213941278314e-05,
+      "loss": 0.5712,
+      "step": 450
+    },
+    {
+      "epoch": 2.963736939151813,
+      "grad_norm": 0.24486881857008144,
+      "learning_rate": 3.396046371843052e-05,
+      "loss": 0.5862,
+      "step": 451
+    },
+    {
+      "epoch": 2.9702929727514853,
+      "grad_norm": 0.14980108475089587,
+      "learning_rate": 3.377891543022229e-05,
+      "loss": 0.5801,
+      "step": 452
+    },
+    {
+      "epoch": 2.9768490063511575,
+      "grad_norm": 0.20536856095424458,
+      "learning_rate": 3.3597498377983444e-05,
+      "loss": 0.5802,
+      "step": 453
+    },
+    {
+      "epoch": 2.9834050399508296,
+      "grad_norm": 0.18782303046647886,
+      "learning_rate": 3.341621638877064e-05,
+      "loss": 0.5784,
+      "step": 454
+    },
+    {
+      "epoch": 2.989961073550502,
+      "grad_norm": 0.14515446375387317,
+      "learning_rate": 3.3235073286791264e-05,
+      "loss": 0.5754,
+      "step": 455
+    },
+    {
+      "epoch": 2.996517107150174,
+      "grad_norm": 0.16161434398106808,
+      "learning_rate": 3.305407289332279e-05,
+      "loss": 0.5756,
+      "step": 456
+    },
+    {
+      "epoch": 3.0055316533497236,
+      "grad_norm": 0.3533323244538683,
+      "learning_rate": 3.287321902663229e-05,
+      "loss": 1.0491,
+      "step": 457
+    },
+    {
+      "epoch": 3.0120876869493958,
+      "grad_norm": 0.3045514737546258,
+      "learning_rate": 3.269251550189573e-05,
+      "loss": 0.5446,
+      "step": 458
+    },
+    {
+      "epoch": 3.018643720549068,
+      "grad_norm": 0.25472049022532234,
+      "learning_rate": 3.251196613111761e-05,
+      "loss": 0.5407,
+      "step": 459
+    },
+    {
+      "epoch": 3.02519975414874,
+      "grad_norm": 0.25510858573604617,
+      "learning_rate": 3.2331574723050474e-05,
+      "loss": 0.5411,
+      "step": 460
+    },
+    {
+      "epoch": 3.0317557877484123,
+      "grad_norm": 0.38483951060765287,
+      "learning_rate": 3.2151345083114574e-05,
+      "loss": 0.5396,
+      "step": 461
+    },
+    {
+      "epoch": 3.0383118213480844,
+      "grad_norm": 0.2558618267692145,
+      "learning_rate": 3.197128101331764e-05,
+      "loss": 0.5349,
+      "step": 462
+    },
+    {
+      "epoch": 3.0448678549477566,
+      "grad_norm": 0.23801295747592094,
+      "learning_rate": 3.179138631217463e-05,
+      "loss": 0.5439,
+      "step": 463
+    },
+    {
+      "epoch": 3.0514238885474287,
+      "grad_norm": 0.2746658371274729,
+      "learning_rate": 3.161166477462759e-05,
+      "loss": 0.5463,
+      "step": 464
+    },
+    {
+      "epoch": 3.057979922147101,
+      "grad_norm": 0.23307565195751595,
+      "learning_rate": 3.1432120191965647e-05,
+      "loss": 0.538,
+      "step": 465
+    },
+    {
+      "epoch": 3.064535955746773,
+      "grad_norm": 0.23849808366960823,
+      "learning_rate": 3.125275635174497e-05,
+      "loss": 0.5431,
+      "step": 466
+    },
+    {
+      "epoch": 3.0710919893464452,
+      "grad_norm": 0.22477045443908192,
+      "learning_rate": 3.1073577037708935e-05,
+      "loss": 0.5422,
+      "step": 467
+    },
+    {
+      "epoch": 3.0776480229461174,
+      "grad_norm": 0.2060382177048335,
+      "learning_rate": 3.089458602970828e-05,
+      "loss": 0.5446,
+      "step": 468
+    },
+    {
+      "epoch": 3.08420405654579,
+      "grad_norm": 0.22200054079896087,
+      "learning_rate": 3.0715787103621294e-05,
+      "loss": 0.5479,
+      "step": 469
+    },
+    {
+      "epoch": 3.090760090145462,
+      "grad_norm": 0.17039781849738478,
+      "learning_rate": 3.0537184031274306e-05,
+      "loss": 0.5305,
+      "step": 470
+    },
+    {
+      "epoch": 3.0973161237451343,
+      "grad_norm": 0.16941181980633563,
+      "learning_rate": 3.0358780580362025e-05,
+      "loss": 0.5261,
+      "step": 471
+    },
+    {
+      "epoch": 3.1038721573448065,
+      "grad_norm": 0.17660138294537184,
+      "learning_rate": 3.0180580514368037e-05,
+      "loss": 0.5312,
+      "step": 472
+    },
+    {
+      "epoch": 3.1104281909444786,
+      "grad_norm": 0.18073996525501446,
+      "learning_rate": 3.0002587592485497e-05,
+      "loss": 0.5371,
+      "step": 473
+    },
+    {
+      "epoch": 3.116984224544151,
+      "grad_norm": 0.3988440151798082,
+      "learning_rate": 2.9824805569537747e-05,
+      "loss": 0.5513,
+      "step": 474
+    },
+    {
+      "epoch": 3.123540258143823,
+      "grad_norm": 0.20423237269747027,
+      "learning_rate": 2.9647238195899168e-05,
+      "loss": 0.5375,
+      "step": 475
+    },
+    {
+      "epoch": 3.130096291743495,
+      "grad_norm": 0.17213193598614746,
+      "learning_rate": 2.9469889217416045e-05,
+      "loss": 0.5437,
+      "step": 476
+    },
+    {
+      "epoch": 3.1366523253431673,
+      "grad_norm": 0.21207611299854595,
+      "learning_rate": 2.9292762375327483e-05,
+      "loss": 0.5339,
+      "step": 477
+    },
+    {
+      "epoch": 3.1432083589428395,
+      "grad_norm": 0.15310274061749113,
+      "learning_rate": 2.9115861406186593e-05,
+      "loss": 0.5303,
+      "step": 478
+    },
+    {
+      "epoch": 3.1497643925425116,
+      "grad_norm": 0.1877335953144904,
+      "learning_rate": 2.8939190041781647e-05,
+      "loss": 0.5319,
+      "step": 479
+    },
+    {
+      "epoch": 3.1563204261421838,
+      "grad_norm": 0.1458292108542802,
+      "learning_rate": 2.8762752009057232e-05,
+      "loss": 0.54,
+      "step": 480
+    },
+    {
+      "epoch": 3.1628764597418564,
+      "grad_norm": 0.17442833234197713,
+      "learning_rate": 2.85865510300358e-05,
+      "loss": 0.5377,
+      "step": 481
+    },
+    {
+      "epoch": 3.1694324933415285,
+      "grad_norm": 0.15457635774978962,
+      "learning_rate": 2.841059082173902e-05,
+      "loss": 0.5389,
+      "step": 482
+    },
+    {
+      "epoch": 3.1759885269412007,
+      "grad_norm": 0.16823476302413515,
+      "learning_rate": 2.823487509610946e-05,
+      "loss": 0.5435,
+      "step": 483
+    },
+    {
+      "epoch": 3.182544560540873,
+      "grad_norm": 0.14604985181804828,
+      "learning_rate": 2.805940755993223e-05,
+      "loss": 0.5377,
+      "step": 484
+    },
+    {
+      "epoch": 3.189100594140545,
+      "grad_norm": 0.14560052298572645,
+      "learning_rate": 2.7884191914756757e-05,
+      "loss": 0.5409,
+      "step": 485
+    },
+    {
+      "epoch": 3.195656627740217,
+      "grad_norm": 0.1472508058071573,
+      "learning_rate": 2.770923185681878e-05,
+      "loss": 0.5455,
+      "step": 486
+    },
+    {
+      "epoch": 3.2022126613398894,
+      "grad_norm": 0.13624234976069768,
+      "learning_rate": 2.7534531076962356e-05,
+      "loss": 0.5433,
+      "step": 487
+    },
+    {
+      "epoch": 3.2087686949395615,
+      "grad_norm": 0.15264871662264248,
+      "learning_rate": 2.7360093260561904e-05,
+      "loss": 0.5372,
+      "step": 488
+    },
+    {
+      "epoch": 3.2153247285392337,
+      "grad_norm": 0.1462659967641793,
+      "learning_rate": 2.7185922087444602e-05,
+      "loss": 0.538,
+      "step": 489
+    },
+    {
+      "epoch": 3.221880762138906,
+      "grad_norm": 0.16303592219574492,
+      "learning_rate": 2.7012021231812666e-05,
+      "loss": 0.542,
+      "step": 490
+    },
+    {
+      "epoch": 3.228436795738578,
+      "grad_norm": 0.14173407966037618,
+      "learning_rate": 2.6838394362165875e-05,
+      "loss": 0.5387,
+      "step": 491
+    },
+    {
+      "epoch": 3.23499282933825,
+      "grad_norm": 0.15432095686939976,
+      "learning_rate": 2.6665045141224193e-05,
+      "loss": 0.5377,
+      "step": 492
+    },
+    {
+      "epoch": 3.2415488629379228,
+      "grad_norm": 0.1482561893469378,
+      "learning_rate": 2.6491977225850446e-05,
+      "loss": 0.5371,
+      "step": 493
+    },
+    {
+      "epoch": 3.248104896537595,
+      "grad_norm": 0.16624560752182815,
+      "learning_rate": 2.6319194266973256e-05,
+      "loss": 0.5359,
+      "step": 494
+    },
+    {
+      "epoch": 3.254660930137267,
+      "grad_norm": 0.15016468812997047,
+      "learning_rate": 2.6146699909509984e-05,
+      "loss": 0.5411,
+      "step": 495
+    },
+    {
+      "epoch": 3.2612169637369393,
+      "grad_norm": 0.2450944209511445,
+      "learning_rate": 2.597449779228983e-05,
+      "loss": 0.5429,
+      "step": 496
+    },
+    {
+      "epoch": 3.2677729973366114,
+      "grad_norm": 0.2412941457078939,
+      "learning_rate": 2.580259154797709e-05,
+      "loss": 0.5343,
+      "step": 497
+    },
+    {
+      "epoch": 3.2743290309362836,
+      "grad_norm": 0.14493021044588159,
+      "learning_rate": 2.563098480299451e-05,
+      "loss": 0.5409,
+      "step": 498
+    },
+    {
+      "epoch": 3.2808850645359557,
+      "grad_norm": 0.16960807003233563,
+      "learning_rate": 2.5459681177446803e-05,
+      "loss": 0.5389,
+      "step": 499
+    },
+    {
+      "epoch": 3.287441098135628,
+      "grad_norm": 0.14869909968414688,
+      "learning_rate": 2.5288684285044283e-05,
+      "loss": 0.5353,
+      "step": 500
+    },
+    {
+      "epoch": 3.2939971317353,
+      "grad_norm": 0.14777225005600433,
+      "learning_rate": 2.5117997733026566e-05,
+      "loss": 0.5393,
+      "step": 501
+    },
+    {
+      "epoch": 3.3005531653349722,
+      "grad_norm": 0.15139629935458157,
+      "learning_rate": 2.4947625122086585e-05,
+      "loss": 0.5435,
+      "step": 502
+    },
+    {
+      "epoch": 3.3071091989346444,
+      "grad_norm": 0.1392849654786053,
+      "learning_rate": 2.477757004629456e-05,
+      "loss": 0.5375,
+      "step": 503
+    },
+    {
+      "epoch": 3.3136652325343166,
+      "grad_norm": 0.1367733939357107,
+      "learning_rate": 2.460783609302218e-05,
+      "loss": 0.5426,
+      "step": 504
+    },
+    {
+      "epoch": 3.3202212661339887,
+      "grad_norm": 0.12699654005182315,
+      "learning_rate": 2.4438426842866966e-05,
+      "loss": 0.5353,
+      "step": 505
+    },
+    {
+      "epoch": 3.3267772997336613,
+      "grad_norm": 0.1390969751788505,
+      "learning_rate": 2.4269345869576676e-05,
+      "loss": 0.539,
+      "step": 506
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 0.12248688460845146,
+      "learning_rate": 2.4100596739973993e-05,
+      "loss": 0.5392,
+      "step": 507
+    },
+    {
+      "epoch": 3.3398893669330056,
+      "grad_norm": 0.13805750150926843,
+      "learning_rate": 2.393218301388123e-05,
+      "loss": 0.5401,
+      "step": 508
+    },
+    {
+      "epoch": 3.346445400532678,
+      "grad_norm": 0.12180259855826225,
+      "learning_rate": 2.3764108244045212e-05,
+      "loss": 0.5465,
+      "step": 509
+    },
+    {
+      "epoch": 3.35300143413235,
+      "grad_norm": 0.12245696463081471,
+      "learning_rate": 2.35963759760624e-05,
+      "loss": 0.5311,
+      "step": 510
+    },
+    {
+      "epoch": 3.359557467732022,
+      "grad_norm": 0.12896584095171493,
+      "learning_rate": 2.342898974830405e-05,
+      "loss": 0.5391,
+      "step": 511
+    },
+    {
+      "epoch": 3.3661135013316943,
+      "grad_norm": 0.12281667944621683,
+      "learning_rate": 2.3261953091841553e-05,
+      "loss": 0.5335,
+      "step": 512
+    },
+    {
+      "epoch": 3.3726695349313665,
+      "grad_norm": 0.11829751242108535,
+      "learning_rate": 2.3095269530372032e-05,
+      "loss": 0.5447,
+      "step": 513
+    },
+    {
+      "epoch": 3.3792255685310386,
+      "grad_norm": 0.13317495284187636,
+      "learning_rate": 2.2928942580143855e-05,
+      "loss": 0.5438,
+      "step": 514
+    },
+    {
+      "epoch": 3.3857816021307108,
+      "grad_norm": 0.12369993983516651,
+      "learning_rate": 2.276297574988263e-05,
+      "loss": 0.5433,
+      "step": 515
+    },
+    {
+      "epoch": 3.392337635730383,
+      "grad_norm": 0.1246149968311202,
+      "learning_rate": 2.2597372540717083e-05,
+      "loss": 0.5412,
+      "step": 516
+    },
+    {
+      "epoch": 3.3988936693300555,
+      "grad_norm": 0.12919720585139893,
+      "learning_rate": 2.2432136446105192e-05,
+      "loss": 0.5442,
+      "step": 517
+    },
+    {
+      "epoch": 3.4054497029297277,
+      "grad_norm": 0.12089802144087428,
+      "learning_rate": 2.226727095176057e-05,
+      "loss": 0.539,
+      "step": 518
+    },
+    {
+      "epoch": 3.4120057365294,
+      "grad_norm": 0.13804336756907107,
+      "learning_rate": 2.210277953557888e-05,
+      "loss": 0.5462,
+      "step": 519
+    },
+    {
+      "epoch": 3.418561770129072,
+      "grad_norm": 0.12982437086238163,
+      "learning_rate": 2.1938665667564435e-05,
+      "loss": 0.5344,
+      "step": 520
+    },
+    {
+      "epoch": 3.425117803728744,
+      "grad_norm": 0.12345004982954612,
+      "learning_rate": 2.177493280975708e-05,
+      "loss": 0.5465,
+      "step": 521
+    },
+    {
+      "epoch": 3.4316738373284164,
+      "grad_norm": 0.14573537922186017,
+      "learning_rate": 2.1611584416159106e-05,
+      "loss": 0.5323,
+      "step": 522
+    },
+    {
+      "epoch": 3.4382298709280885,
+      "grad_norm": 0.1097589192258888,
+      "learning_rate": 2.1448623932662377e-05,
+      "loss": 0.5437,
+      "step": 523
+    },
+    {
+      "epoch": 3.4447859045277607,
+      "grad_norm": 0.1317797633987106,
+      "learning_rate": 2.1286054796975696e-05,
+      "loss": 0.5377,
+      "step": 524
+    },
+    {
+      "epoch": 3.451341938127433,
+      "grad_norm": 0.13321471541724095,
+      "learning_rate": 2.1123880438552187e-05,
+      "loss": 0.5518,
+      "step": 525
+    },
+    {
+      "epoch": 3.457897971727105,
+      "grad_norm": 0.1352447131480959,
+      "learning_rate": 2.096210427851706e-05,
+      "loss": 0.532,
+      "step": 526
+    },
+    {
+      "epoch": 3.464454005326777,
+      "grad_norm": 0.13508733978142928,
+      "learning_rate": 2.0800729729595385e-05,
+      "loss": 0.5314,
+      "step": 527
+    },
+    {
+      "epoch": 3.4710100389264493,
+      "grad_norm": 0.1380525297606027,
+      "learning_rate": 2.063976019604006e-05,
+      "loss": 0.5438,
+      "step": 528
+    },
+    {
+      "epoch": 3.4775660725261215,
+      "grad_norm": 0.13529984461100034,
+      "learning_rate": 2.0479199073560084e-05,
+      "loss": 0.5451,
+      "step": 529
+    },
+    {
+      "epoch": 3.4841221061257936,
+      "grad_norm": 0.13383345341162217,
+      "learning_rate": 2.0319049749248876e-05,
+      "loss": 0.5335,
+      "step": 530
+    },
+    {
+      "epoch": 3.4906781397254663,
+      "grad_norm": 0.12966276679499278,
+      "learning_rate": 2.0159315601512817e-05,
+      "loss": 0.5404,
+      "step": 531
+    },
+    {
+      "epoch": 3.4972341733251384,
+      "grad_norm": 0.12661938479816495,
+      "learning_rate": 2.0000000000000012e-05,
+      "loss": 0.5361,
+      "step": 532
+    },
+    {
+      "epoch": 3.5037902069248106,
+      "grad_norm": 0.25950697427544217,
+      "learning_rate": 1.9841106305529133e-05,
+      "loss": 0.5475,
+      "step": 533
+    },
+    {
+      "epoch": 3.5103462405244827,
+      "grad_norm": 0.12388736409274281,
+      "learning_rate": 1.9682637870018638e-05,
+      "loss": 0.5433,
+      "step": 534
+    },
+    {
+      "epoch": 3.516902274124155,
+      "grad_norm": 0.13692640136699888,
+      "learning_rate": 1.9524598036415973e-05,
+      "loss": 0.5379,
+      "step": 535
+    },
+    {
+      "epoch": 3.523458307723827,
+      "grad_norm": 0.22971792027818427,
+      "learning_rate": 1.9366990138627054e-05,
+      "loss": 0.5354,
+      "step": 536
+    },
+    {
+      "epoch": 3.5300143413234992,
+      "grad_norm": 0.13477440715558586,
+      "learning_rate": 1.9209817501445978e-05,
+      "loss": 0.5408,
+      "step": 537
+    },
+    {
+      "epoch": 3.5365703749231714,
+      "grad_norm": 0.12000818819921334,
+      "learning_rate": 1.9053083440484887e-05,
+      "loss": 0.539,
+      "step": 538
+    },
+    {
+      "epoch": 3.5431264085228436,
+      "grad_norm": 0.18018036064763043,
+      "learning_rate": 1.889679126210397e-05,
+      "loss": 0.54,
+      "step": 539
+    },
+    {
+      "epoch": 3.5496824421225157,
+      "grad_norm": 0.13187429164193115,
+      "learning_rate": 1.8740944263341773e-05,
+      "loss": 0.5336,
+      "step": 540
+    },
+    {
+      "epoch": 3.5562384757221883,
+      "grad_norm": 0.1115344100782877,
+      "learning_rate": 1.8585545731845584e-05,
+      "loss": 0.5457,
+      "step": 541
+    },
+    {
+      "epoch": 3.5627945093218605,
+      "grad_norm": 0.4821099792222361,
+      "learning_rate": 1.8430598945802156e-05,
+      "loss": 0.5429,
+      "step": 542
+    },
+    {
+      "epoch": 3.5693505429215326,
+      "grad_norm": 0.13726412619312758,
+      "learning_rate": 1.8276107173868503e-05,
+      "loss": 0.551,
+      "step": 543
+    },
+    {
+      "epoch": 3.575906576521205,
+      "grad_norm": 0.11727031408692318,
+      "learning_rate": 1.8122073675102935e-05,
+      "loss": 0.5315,
+      "step": 544
+    },
+    {
+      "epoch": 3.582462610120877,
+      "grad_norm": 0.12442269217723008,
+      "learning_rate": 1.7968501698896346e-05,
+      "loss": 0.5338,
+      "step": 545
+    },
+    {
+      "epoch": 3.589018643720549,
+      "grad_norm": 0.11894782465239562,
+      "learning_rate": 1.781539448490365e-05,
+      "loss": 0.5365,
+      "step": 546
+    },
+    {
+      "epoch": 3.5955746773202213,
+      "grad_norm": 0.11002626026600501,
+      "learning_rate": 1.7662755262975432e-05,
+      "loss": 0.5404,
+      "step": 547
+    },
+    {
+      "epoch": 3.6021307109198935,
+      "grad_norm": 0.13251223380465485,
+      "learning_rate": 1.7510587253089842e-05,
+      "loss": 0.543,
+      "step": 548
+    },
+    {
+      "epoch": 3.6086867445195656,
+      "grad_norm": 0.11425745596834401,
+      "learning_rate": 1.7358893665284595e-05,
+      "loss": 0.5345,
+      "step": 549
+    },
+    {
+      "epoch": 3.6152427781192378,
+      "grad_norm": 0.11826825954302657,
+      "learning_rate": 1.7207677699589355e-05,
+      "loss": 0.5463,
+      "step": 550
+    },
+    {
+      "epoch": 3.62179881171891,
+      "grad_norm": 0.10903975348972718,
+      "learning_rate": 1.7056942545958167e-05,
+      "loss": 0.5396,
+      "step": 551
+    },
+    {
+      "epoch": 3.628354845318582,
+      "grad_norm": 0.11693940603496059,
+      "learning_rate": 1.690669138420215e-05,
+      "loss": 0.5305,
+      "step": 552
+    },
+    {
+      "epoch": 3.6349108789182543,
+      "grad_norm": 0.1142560549805311,
+      "learning_rate": 1.6756927383922473e-05,
+      "loss": 0.5391,
+      "step": 553
+    },
+    {
+      "epoch": 3.6414669125179264,
+      "grad_norm": 0.10839349884507322,
+      "learning_rate": 1.6607653704443457e-05,
+      "loss": 0.542,
+      "step": 554
+    },
+    {
+      "epoch": 3.6480229461175986,
+      "grad_norm": 0.11225360540051456,
+      "learning_rate": 1.6458873494745926e-05,
+      "loss": 0.5388,
+      "step": 555
+    },
+    {
+      "epoch": 3.654578979717271,
+      "grad_norm": 0.11188297317863019,
+      "learning_rate": 1.6310589893400804e-05,
+      "loss": 0.5381,
+      "step": 556
+    },
+    {
+      "epoch": 3.6611350133169434,
+      "grad_norm": 0.7105498519897432,
+      "learning_rate": 1.6162806028502852e-05,
+      "loss": 0.5411,
+      "step": 557
+    },
+    {
+      "epoch": 3.6676910469166155,
+      "grad_norm": 0.11475316554461859,
+      "learning_rate": 1.601552501760473e-05,
+      "loss": 0.5416,
+      "step": 558
+    },
+    {
+      "epoch": 3.6742470805162877,
+      "grad_norm": 0.12194606250202103,
+      "learning_rate": 1.5868749967651252e-05,
+      "loss": 0.5396,
+      "step": 559
+    },
+    {
+      "epoch": 3.68080311411596,
+      "grad_norm": 0.12661866626370497,
+      "learning_rate": 1.5722483974913737e-05,
+      "loss": 0.531,
+      "step": 560
+    },
+    {
+      "epoch": 3.687359147715632,
+      "grad_norm": 0.11256332358317793,
+      "learning_rate": 1.5576730124924822e-05,
+      "loss": 0.5418,
+      "step": 561
+    },
+    {
+      "epoch": 3.693915181315304,
+      "grad_norm": 0.11015606275564016,
+      "learning_rate": 1.5431491492413288e-05,
+      "loss": 0.5411,
+      "step": 562
+    },
+    {
+      "epoch": 3.7004712149149763,
+      "grad_norm": 0.12074411265164606,
+      "learning_rate": 1.528677114123923e-05,
+      "loss": 0.53,
+      "step": 563
+    },
+    {
+      "epoch": 3.7070272485146485,
+      "grad_norm": 0.10893671793883805,
+      "learning_rate": 1.5142572124329418e-05,
+      "loss": 0.5397,
+      "step": 564
+    },
+    {
+      "epoch": 3.713583282114321,
+      "grad_norm": 0.11746065306540288,
+      "learning_rate": 1.4998897483612865e-05,
+      "loss": 0.5423,
+      "step": 565
+    },
+    {
+      "epoch": 3.7201393157139933,
+      "grad_norm": 0.1126528745871064,
+      "learning_rate": 1.4855750249956718e-05,
+      "loss": 0.5426,
+      "step": 566
+    },
+    {
+      "epoch": 3.7266953493136654,
+      "grad_norm": 0.3165608272055915,
+      "learning_rate": 1.4713133443102283e-05,
+      "loss": 0.5428,
+      "step": 567
+    },
+    {
+      "epoch": 3.7332513829133376,
+      "grad_norm": 0.11800782137431694,
+      "learning_rate": 1.457105007160129e-05,
+      "loss": 0.5443,
+      "step": 568
+    },
+    {
+      "epoch": 3.7398074165130097,
+      "grad_norm": 0.11122501505766147,
+      "learning_rate": 1.44295031327525e-05,
+      "loss": 0.5397,
+      "step": 569
+    },
+    {
+      "epoch": 3.746363450112682,
+      "grad_norm": 0.12798481597310407,
+      "learning_rate": 1.4288495612538427e-05,
+      "loss": 0.5457,
+      "step": 570
+    },
+    {
+      "epoch": 3.752919483712354,
+      "grad_norm": 0.10441587675286511,
+      "learning_rate": 1.4148030485562362e-05,
+      "loss": 0.5371,
+      "step": 571
+    },
+    {
+      "epoch": 3.7594755173120262,
+      "grad_norm": 0.12743357362001942,
+      "learning_rate": 1.4008110714985623e-05,
+      "loss": 0.5392,
+      "step": 572
+    },
+    {
+      "epoch": 3.7660315509116984,
+      "grad_norm": 0.1114224145991442,
+      "learning_rate": 1.3868739252465017e-05,
+      "loss": 0.5368,
+      "step": 573
+    },
+    {
+      "epoch": 3.7725875845113706,
+      "grad_norm": 0.10668882720776016,
+      "learning_rate": 1.3729919038090627e-05,
+      "loss": 0.5386,
+      "step": 574
+    },
+    {
+      "epoch": 3.7791436181110427,
+      "grad_norm": 0.11554119750738572,
+      "learning_rate": 1.3591653000323764e-05,
+      "loss": 0.5336,
+      "step": 575
+    },
+    {
+      "epoch": 3.785699651710715,
+      "grad_norm": 0.11456186328555522,
+      "learning_rate": 1.3453944055935151e-05,
+      "loss": 0.5447,
+      "step": 576
+    },
+    {
+      "epoch": 3.792255685310387,
+      "grad_norm": 0.12119615336966778,
+      "learning_rate": 1.3316795109943476e-05,
+      "loss": 0.5333,
+      "step": 577
+    },
+    {
+      "epoch": 3.798811718910059,
+      "grad_norm": 0.1097823442621975,
+      "learning_rate": 1.3180209055554043e-05,
+      "loss": 0.5353,
+      "step": 578
+    },
+    {
+      "epoch": 3.8053677525097314,
+      "grad_norm": 0.12113771390183198,
+      "learning_rate": 1.3044188774097757e-05,
+      "loss": 0.5406,
+      "step": 579
+    },
+    {
+      "epoch": 3.811923786109404,
+      "grad_norm": 0.1076103282360743,
+      "learning_rate": 1.2908737134970367e-05,
+      "loss": 0.5428,
+      "step": 580
+    },
+    {
+      "epoch": 3.818479819709076,
+      "grad_norm": 0.10778481828407349,
+      "learning_rate": 1.2773856995571858e-05,
+      "loss": 0.5346,
+      "step": 581
+    },
+    {
+      "epoch": 3.8250358533087483,
+      "grad_norm": 0.10605992762588787,
+      "learning_rate": 1.2639551201246278e-05,
+      "loss": 0.5394,
+      "step": 582
+    },
+    {
+      "epoch": 3.8315918869084205,
+      "grad_norm": 0.10492552888703595,
+      "learning_rate": 1.2505822585221665e-05,
+      "loss": 0.5402,
+      "step": 583
+    },
+    {
+      "epoch": 3.8381479205080926,
+      "grad_norm": 0.10835031468844004,
+      "learning_rate": 1.2372673968550229e-05,
+      "loss": 0.5449,
+      "step": 584
+    },
+    {
+      "epoch": 3.8447039541077648,
+      "grad_norm": 0.10610757084756944,
+      "learning_rate": 1.2240108160048934e-05,
+      "loss": 0.5366,
+      "step": 585
+    },
+    {
+      "epoch": 3.851259987707437,
+      "grad_norm": 0.10287232050693294,
+      "learning_rate": 1.2108127956240186e-05,
+      "loss": 0.5367,
+      "step": 586
+    },
+    {
+      "epoch": 3.857816021307109,
+      "grad_norm": 0.10929614703844563,
+      "learning_rate": 1.1976736141292853e-05,
+      "loss": 0.5335,
+      "step": 587
+    },
+    {
+      "epoch": 3.8643720549067813,
+      "grad_norm": 0.1204731707453792,
+      "learning_rate": 1.1845935486963546e-05,
+      "loss": 0.5404,
+      "step": 588
+    },
+    {
+      "epoch": 3.870928088506454,
+      "grad_norm": 0.10037277898866681,
+      "learning_rate": 1.1715728752538103e-05,
+      "loss": 0.5386,
+      "step": 589
+    },
+    {
+      "epoch": 3.877484122106126,
+      "grad_norm": 0.10755482937489974,
+      "learning_rate": 1.158611868477344e-05,
+      "loss": 0.5396,
+      "step": 590
+    },
+    {
+      "epoch": 3.884040155705798,
+      "grad_norm": 0.1125655323891997,
+      "learning_rate": 1.1457108017839587e-05,
+      "loss": 0.5522,
+      "step": 591
+    },
+    {
+      "epoch": 3.8905961893054704,
+      "grad_norm": 0.09908512763395612,
+      "learning_rate": 1.1328699473261957e-05,
+      "loss": 0.5389,
+      "step": 592
+    },
+    {
+      "epoch": 3.8971522229051425,
+      "grad_norm": 0.10011512841325966,
+      "learning_rate": 1.1200895759864027e-05,
+      "loss": 0.5402,
+      "step": 593
+    },
+    {
+      "epoch": 3.9037082565048147,
+      "grad_norm": 0.11190228493508152,
+      "learning_rate": 1.107369957371013e-05,
+      "loss": 0.5402,
+      "step": 594
+    },
+    {
+      "epoch": 3.910264290104487,
+      "grad_norm": 0.0997294109524961,
+      "learning_rate": 1.09471135980486e-05,
+      "loss": 0.5356,
+      "step": 595
+    },
+    {
+      "epoch": 3.916820323704159,
+      "grad_norm": 0.10926000465284759,
+      "learning_rate": 1.0821140503255174e-05,
+      "loss": 0.5459,
+      "step": 596
+    },
+    {
+      "epoch": 3.923376357303831,
+      "grad_norm": 0.10485038006440338,
+      "learning_rate": 1.0695782946776619e-05,
+      "loss": 0.5407,
+      "step": 597
+    },
+    {
+      "epoch": 3.9299323909035033,
+      "grad_norm": 0.10291841117205842,
+      "learning_rate": 1.0571043573074737e-05,
+      "loss": 0.5373,
+      "step": 598
+    },
+    {
+      "epoch": 3.9364884245031755,
+      "grad_norm": 0.10153144309813497,
+      "learning_rate": 1.0446925013570545e-05,
+      "loss": 0.5408,
+      "step": 599
+    },
+    {
+      "epoch": 3.9430444581028476,
+      "grad_norm": 0.10605125481156527,
+      "learning_rate": 1.0323429886588743e-05,
+      "loss": 0.5411,
+      "step": 600
+    },
+    {
+      "epoch": 3.94960049170252,
+      "grad_norm": 0.10215451404049354,
+      "learning_rate": 1.020056079730252e-05,
+      "loss": 0.5428,
+      "step": 601
+    },
+    {
+      "epoch": 3.956156525302192,
+      "grad_norm": 0.10818454985255951,
+      "learning_rate": 1.0078320337678584e-05,
+      "loss": 0.5396,
+      "step": 602
+    },
+    {
+      "epoch": 3.962712558901864,
+      "grad_norm": 0.10188342720575627,
+      "learning_rate": 9.956711086422471e-06,
+      "loss": 0.5486,
+      "step": 603
+    },
+    {
+      "epoch": 3.9692685925015363,
+      "grad_norm": 0.10148315246310981,
+      "learning_rate": 9.835735608924155e-06,
+      "loss": 0.5341,
+      "step": 604
+    },
+    {
+      "epoch": 3.975824626101209,
+      "grad_norm": 0.09241997326882184,
+      "learning_rate": 9.715396457203918e-06,
+      "loss": 0.5389,
+      "step": 605
+    },
+    {
+      "epoch": 3.982380659700881,
+      "grad_norm": 0.102047541160635,
+      "learning_rate": 9.595696169858542e-06,
+      "loss": 0.5343,
+      "step": 606
+    },
+    {
+      "epoch": 3.9889366933005532,
+      "grad_norm": 0.16329932937909056,
+      "learning_rate": 9.476637272007748e-06,
+      "loss": 0.5414,
+      "step": 607
+    },
+    {
+      "epoch": 3.9954927269002254,
+      "grad_norm": 0.10040695165913859,
+      "learning_rate": 9.358222275240884e-06,
+      "loss": 0.5365,
+      "step": 608
+    },
+    {
+      "epoch": 4.006556033599672,
+      "grad_norm": 0.16186764492729486,
+      "learning_rate": 9.24045367756401e-06,
+      "loss": 0.5171,
+      "step": 609
+    },
+    {
+      "epoch": 4.013112067199344,
+      "grad_norm": 0.1322033843585489,
+      "learning_rate": 9.123333963347166e-06,
+      "loss": 0.5209,
+      "step": 610
+    },
+    {
+      "epoch": 4.0196681007990165,
+      "grad_norm": 0.12245478327966862,
+      "learning_rate": 9.006865603271952e-06,
+      "loss": 0.5112,
+      "step": 611
+    },
+    {
+      "epoch": 4.026224134398689,
+      "grad_norm": 0.117065406177187,
+      "learning_rate": 8.89105105427945e-06,
+      "loss": 0.5089,
+      "step": 612
+    },
+    {
+      "epoch": 4.032780167998361,
+      "grad_norm": 0.1358673718015058,
+      "learning_rate": 8.775892759518321e-06,
+      "loss": 0.5245,
+      "step": 613
+    },
+    {
+      "epoch": 4.039336201598033,
+      "grad_norm": 0.1436375868810231,
+      "learning_rate": 8.661393148293355e-06,
+      "loss": 0.523,
+      "step": 614
+    },
+    {
+      "epoch": 4.045892235197705,
+      "grad_norm": 0.11924450152050672,
+      "learning_rate": 8.547554636014177e-06,
+      "loss": 0.506,
+      "step": 615
+    },
+    {
+      "epoch": 4.052448268797377,
+      "grad_norm": 0.11988650232489285,
+      "learning_rate": 8.434379624144261e-06,
+      "loss": 0.515,
+      "step": 616
+    },
+    {
+      "epoch": 4.0590043023970495,
+      "grad_norm": 0.13738230782732197,
+      "learning_rate": 8.321870500150347e-06,
+      "loss": 0.511,
+      "step": 617
+    },
+    {
+      "epoch": 4.065560335996722,
+      "grad_norm": 0.1371441628587204,
+      "learning_rate": 8.210029637452016e-06,
+      "loss": 0.5139,
+      "step": 618
+    },
+    {
+      "epoch": 4.072116369596394,
+      "grad_norm": 0.3628915461879963,
+      "learning_rate": 8.098859395371641e-06,
+      "loss": 0.5165,
+      "step": 619
+    },
+    {
+      "epoch": 4.078672403196066,
+      "grad_norm": 0.11429063516858927,
+      "learning_rate": 7.988362119084642e-06,
+      "loss": 0.5136,
+      "step": 620
+    },
+    {
+      "epoch": 4.085228436795739,
+      "grad_norm": 0.13045931327904303,
+      "learning_rate": 7.87854013956994e-06,
+      "loss": 0.5097,
+      "step": 621
+    },
+    {
+      "epoch": 4.091784470395411,
+      "grad_norm": 0.11629419650039642,
+      "learning_rate": 7.769395773560874e-06,
+      "loss": 0.5157,
+      "step": 622
+    },
+    {
+      "epoch": 4.098340503995083,
+      "grad_norm": 0.11362363257154101,
+      "learning_rate": 7.660931323496283e-06,
+      "loss": 0.5188,
+      "step": 623
+    },
+    {
+      "epoch": 4.1048965375947555,
+      "grad_norm": 0.10992983384699095,
+      "learning_rate": 7.553149077471915e-06,
+      "loss": 0.5124,
+      "step": 624
+    },
+    {
+      "epoch": 4.111452571194428,
+      "grad_norm": 0.10426021171852758,
+      "learning_rate": 7.446051309192204e-06,
+      "loss": 0.5142,
+      "step": 625
+    },
+    {
+      "epoch": 4.1180086047941,
+      "grad_norm": 0.10863479923168164,
+      "learning_rate": 7.3396402779222845e-06,
+      "loss": 0.5127,
+      "step": 626
+    },
+    {
+      "epoch": 4.124564638393772,
+      "grad_norm": 0.10635132221465905,
+      "learning_rate": 7.233918228440324e-06,
+      "loss": 0.5162,
+      "step": 627
+    },
+    {
+      "epoch": 4.131120671993444,
+      "grad_norm": 0.09912102469646186,
+      "learning_rate": 7.128887390990198e-06,
+      "loss": 0.5224,
+      "step": 628
+    },
+    {
+      "epoch": 4.137676705593116,
+      "grad_norm": 0.10290328673729038,
+      "learning_rate": 7.024549981234377e-06,
+      "loss": 0.5217,
+      "step": 629
+    },
+    {
+      "epoch": 4.1442327391927885,
+      "grad_norm": 0.1038941435526061,
+      "learning_rate": 6.9209082002072725e-06,
+      "loss": 0.5133,
+      "step": 630
+    },
+    {
+      "epoch": 4.150788772792461,
+      "grad_norm": 0.09603936143418888,
+      "learning_rate": 6.817964234268748e-06,
+      "loss": 0.5176,
+      "step": 631
+    },
+    {
+      "epoch": 4.157344806392133,
+      "grad_norm": 0.11813529879188972,
+      "learning_rate": 6.715720255058e-06,
+      "loss": 0.5152,
+      "step": 632
+    },
+    {
+      "epoch": 4.163900839991805,
+      "grad_norm": 0.09790007822885208,
+      "learning_rate": 6.614178419447781e-06,
+      "loss": 0.5158,
+      "step": 633
+    },
+    {
+      "epoch": 4.170456873591477,
+      "grad_norm": 0.10792344691656378,
+      "learning_rate": 6.513340869498859e-06,
+      "loss": 0.5122,
+      "step": 634
+    },
+    {
+      "epoch": 4.177012907191149,
+      "grad_norm": 0.24107917853631503,
+      "learning_rate": 6.4132097324148556e-06,
+      "loss": 0.5161,
+      "step": 635
+    },
+    {
+      "epoch": 4.183568940790821,
+      "grad_norm": 0.09669004690900093,
+      "learning_rate": 6.313787120497376e-06,
+      "loss": 0.507,
+      "step": 636
+    },
+    {
+      "epoch": 4.190124974390494,
+      "grad_norm": 0.10731913975620888,
+      "learning_rate": 6.215075131101405e-06,
+      "loss": 0.5087,
+      "step": 637
+    },
+    {
+      "epoch": 4.196681007990166,
+      "grad_norm": 0.0997103006367049,
+      "learning_rate": 6.117075846591123e-06,
+      "loss": 0.5201,
+      "step": 638
+    },
+    {
+      "epoch": 4.203237041589838,
+      "grad_norm": 0.09955705599544924,
+      "learning_rate": 6.019791334295955e-06,
+      "loss": 0.5076,
+      "step": 639
+    },
+    {
+      "epoch": 4.20979307518951,
+      "grad_norm": 0.09540357195970695,
+      "learning_rate": 5.923223646466923e-06,
+      "loss": 0.5181,
+      "step": 640
+    },
+    {
+      "epoch": 4.216349108789182,
+      "grad_norm": 0.0952939694361504,
+      "learning_rate": 5.827374820233407e-06,
+      "loss": 0.5195,
+      "step": 641
+    },
+    {
+      "epoch": 4.222905142388854,
+      "grad_norm": 0.09942874484427339,
+      "learning_rate": 5.732246877560146e-06,
+      "loss": 0.5171,
+      "step": 642
+    },
+    {
+      "epoch": 4.2294611759885266,
+      "grad_norm": 0.09564484595006637,
+      "learning_rate": 5.637841825204588e-06,
+      "loss": 0.5131,
+      "step": 643
+    },
+    {
+      "epoch": 4.236017209588199,
+      "grad_norm": 0.1056325087529548,
+      "learning_rate": 5.5441616546745646e-06,
+      "loss": 0.5095,
+      "step": 644
+    },
+    {
+      "epoch": 4.242573243187872,
+      "grad_norm": 0.09442156372590707,
+      "learning_rate": 5.451208342186229e-06,
+      "loss": 0.5139,
+      "step": 645
+    },
+    {
+      "epoch": 4.249129276787544,
+      "grad_norm": 0.08963880451358089,
+      "learning_rate": 5.358983848622452e-06,
+      "loss": 0.5192,
+      "step": 646
+    },
+    {
+      "epoch": 4.255685310387216,
+      "grad_norm": 0.0917773531851403,
+      "learning_rate": 5.26749011949141e-06,
+      "loss": 0.5112,
+      "step": 647
+    },
+    {
+      "epoch": 4.262241343986888,
+      "grad_norm": 0.09650192405895454,
+      "learning_rate": 5.176729084885508e-06,
+      "loss": 0.5156,
+      "step": 648
+    },
+    {
+      "epoch": 4.26879737758656,
+      "grad_norm": 0.09638202151344397,
+      "learning_rate": 5.086702659440743e-06,
+      "loss": 0.5132,
+      "step": 649
+    },
+    {
+      "epoch": 4.275353411186233,
+      "grad_norm": 0.09515733205388084,
+      "learning_rate": 4.99741274229625e-06,
+      "loss": 0.5107,
+      "step": 650
+    },
+    {
+      "epoch": 4.281909444785905,
+      "grad_norm": 0.08694276833101254,
+      "learning_rate": 4.908861217054281e-06,
+      "loss": 0.511,
+      "step": 651
+    },
+    {
+      "epoch": 4.288465478385577,
+      "grad_norm": 0.09365787216266765,
+      "learning_rate": 4.821049951740442e-06,
+      "loss": 0.5196,
+      "step": 652
+    },
+    {
+      "epoch": 4.295021511985249,
+      "grad_norm": 0.09875335551778978,
+      "learning_rate": 4.733980798764273e-06,
+      "loss": 0.5139,
+      "step": 653
+    },
+    {
+      "epoch": 4.301577545584921,
+      "grad_norm": 0.0984165654822695,
+      "learning_rate": 4.647655594880225e-06,
+      "loss": 0.5194,
+      "step": 654
+    },
+    {
+      "epoch": 4.308133579184593,
+      "grad_norm": 0.08981515555040814,
+      "learning_rate": 4.562076161148881e-06,
+      "loss": 0.5159,
+      "step": 655
+    },
+    {
+      "epoch": 4.3146896127842655,
+      "grad_norm": 0.08699993576971486,
+      "learning_rate": 4.4772443028985004e-06,
+      "loss": 0.5107,
+      "step": 656
+    },
+    {
+      "epoch": 4.321245646383938,
+      "grad_norm": 0.09531375447276737,
+      "learning_rate": 4.393161809687021e-06,
+      "loss": 0.5211,
+      "step": 657
+    },
+    {
+      "epoch": 4.32780167998361,
+      "grad_norm": 0.16067767992274057,
+      "learning_rate": 4.3098304552642385e-06,
+      "loss": 0.5163,
+      "step": 658
+    },
+    {
+      "epoch": 4.334357713583282,
+      "grad_norm": 0.09069517004519909,
+      "learning_rate": 4.227251997534416e-06,
+      "loss": 0.5107,
+      "step": 659
+    },
+    {
+      "epoch": 4.340913747182954,
+      "grad_norm": 0.08688482436943581,
+      "learning_rate": 4.1454281785191995e-06,
+      "loss": 0.5118,
+      "step": 660
+    },
+    {
+      "epoch": 4.347469780782626,
+      "grad_norm": 0.09109521686027529,
+      "learning_rate": 4.064360724320846e-06,
+      "loss": 0.5114,
+      "step": 661
+    },
+    {
+      "epoch": 4.3540258143822985,
+      "grad_norm": 0.0885789818679224,
+      "learning_rate": 3.984051345085855e-06,
+      "loss": 0.5126,
+      "step": 662
+    },
+    {
+      "epoch": 4.360581847981971,
+      "grad_norm": 0.09379559333937212,
+      "learning_rate": 3.90450173496887e-06,
+      "loss": 0.5207,
+      "step": 663
+    },
+    {
+      "epoch": 4.367137881581643,
+      "grad_norm": 0.08937679173054215,
+      "learning_rate": 3.825713572096903e-06,
+      "loss": 0.5175,
+      "step": 664
+    },
+    {
+      "epoch": 4.373693915181315,
+      "grad_norm": 0.08924834510849128,
+      "learning_rate": 3.747688518534003e-06,
+      "loss": 0.5186,
+      "step": 665
+    },
+    {
+      "epoch": 4.380249948780987,
+      "grad_norm": 0.09109376121051856,
+      "learning_rate": 3.6704282202461515e-06,
+      "loss": 0.5161,
+      "step": 666
+    },
+    {
+      "epoch": 4.386805982380659,
+      "grad_norm": 0.10119588536048828,
+      "learning_rate": 3.5939343070665243e-06,
+      "loss": 0.5032,
+      "step": 667
+    },
+    {
+      "epoch": 4.3933620159803315,
+      "grad_norm": 0.09644206288351137,
+      "learning_rate": 3.518208392661184e-06,
+      "loss": 0.5188,
+      "step": 668
+    },
+    {
+      "epoch": 4.399918049580004,
+      "grad_norm": 0.08778658376932175,
+      "learning_rate": 3.4432520744949317e-06,
+      "loss": 0.5133,
+      "step": 669
+    },
+    {
+      "epoch": 4.406474083179677,
+      "grad_norm": 0.0941601011641409,
+      "learning_rate": 3.3690669337977e-06,
+      "loss": 0.5061,
+      "step": 670
+    },
+    {
+      "epoch": 4.413030116779349,
+      "grad_norm": 0.09791658312405495,
+      "learning_rate": 3.295654535531161e-06,
+      "loss": 0.5193,
+      "step": 671
+    },
+    {
+      "epoch": 4.419586150379021,
+      "grad_norm": 0.09354050139551012,
+      "learning_rate": 3.2230164283556918e-06,
+      "loss": 0.5172,
+      "step": 672
+    },
+    {
+      "epoch": 4.426142183978693,
+      "grad_norm": 0.08893737381738466,
+      "learning_rate": 3.151154144597741e-06,
+      "loss": 0.5111,
+      "step": 673
+    },
+    {
+      "epoch": 4.432698217578365,
+      "grad_norm": 0.08445332878405215,
+      "learning_rate": 3.080069200217497e-06,
+      "loss": 0.5097,
+      "step": 674
+    },
+    {
+      "epoch": 4.4392542511780375,
+      "grad_norm": 0.08801285298434917,
+      "learning_rate": 3.0097630947768695e-06,
+      "loss": 0.5135,
+      "step": 675
+    },
+    {
+      "epoch": 4.44581028477771,
+      "grad_norm": 0.09402581699959618,
+      "learning_rate": 2.9402373114079295e-06,
+      "loss": 0.5086,
+      "step": 676
+    },
+    {
+      "epoch": 4.452366318377382,
+      "grad_norm": 0.0905312100821144,
+      "learning_rate": 2.871493316781546e-06,
+      "loss": 0.515,
+      "step": 677
+    },
+    {
+      "epoch": 4.458922351977054,
+      "grad_norm": 0.09633255689013458,
+      "learning_rate": 2.803532561076492e-06,
+      "loss": 0.5123,
+      "step": 678
+    },
+    {
+      "epoch": 4.465478385576726,
+      "grad_norm": 0.08803341472121784,
+      "learning_rate": 2.7363564779488448e-06,
+      "loss": 0.5135,
+      "step": 679
+    },
+    {
+      "epoch": 4.472034419176398,
+      "grad_norm": 0.08797758072438489,
+      "learning_rate": 2.669966484501716e-06,
+      "loss": 0.5199,
+      "step": 680
+    },
+    {
+      "epoch": 4.4785904527760705,
+      "grad_norm": 0.09112213281433264,
+      "learning_rate": 2.6043639812554043e-06,
+      "loss": 0.5205,
+      "step": 681
+    },
+    {
+      "epoch": 4.485146486375743,
+      "grad_norm": 0.09544750033107532,
+      "learning_rate": 2.5395503521178143e-06,
+      "loss": 0.5253,
+      "step": 682
+    },
+    {
+      "epoch": 4.491702519975415,
+      "grad_norm": 0.09376536821854904,
+      "learning_rate": 2.4755269643552594e-06,
+      "loss": 0.5164,
+      "step": 683
+    },
+    {
+      "epoch": 4.498258553575087,
+      "grad_norm": 0.09454989320113857,
+      "learning_rate": 2.4122951685636674e-06,
+      "loss": 0.5187,
+      "step": 684
+    },
+    {
+      "epoch": 4.504814587174759,
+      "grad_norm": 0.08421901449177807,
+      "learning_rate": 2.3498562986400187e-06,
+      "loss": 0.5154,
+      "step": 685
+    },
+    {
+      "epoch": 4.511370620774431,
+      "grad_norm": 0.08753626044860159,
+      "learning_rate": 2.2882116717542634e-06,
+      "loss": 0.5176,
+      "step": 686
+    },
+    {
+      "epoch": 4.5179266543741035,
+      "grad_norm": 0.0907417635028892,
+      "learning_rate": 2.22736258832152e-06,
+      "loss": 0.5244,
+      "step": 687
+    },
+    {
+      "epoch": 4.524482687973776,
+      "grad_norm": 0.08866763487206057,
+      "learning_rate": 2.1673103319746146e-06,
+      "loss": 0.5188,
+      "step": 688
+    },
+    {
+      "epoch": 4.531038721573448,
+      "grad_norm": 0.08406194293605605,
+      "learning_rate": 2.1080561695370425e-06,
+      "loss": 0.5117,
+      "step": 689
+    },
+    {
+      "epoch": 4.53759475517312,
+      "grad_norm": 0.08781306185528281,
+      "learning_rate": 2.049601350996233e-06,
+      "loss": 0.5105,
+      "step": 690
+    },
+    {
+      "epoch": 4.544150788772792,
+      "grad_norm": 0.0862714841925732,
+      "learning_rate": 1.9919471094771523e-06,
+      "loss": 0.513,
+      "step": 691
+    },
+    {
+      "epoch": 4.550706822372465,
+      "grad_norm": 0.08316320143150852,
+      "learning_rate": 1.93509466121633e-06,
+      "loss": 0.5143,
+      "step": 692
+    },
+    {
+      "epoch": 4.557262855972137,
+      "grad_norm": 0.08459303029799788,
+      "learning_rate": 1.8790452055361764e-06,
+      "loss": 0.5117,
+      "step": 693
+    },
+    {
+      "epoch": 4.5638188895718095,
+      "grad_norm": 0.08646962847262735,
+      "learning_rate": 1.8237999248197002e-06,
+      "loss": 0.5122,
+      "step": 694
+    },
+    {
+      "epoch": 4.570374923171482,
+      "grad_norm": 0.18896054046080968,
+      "learning_rate": 1.7693599844855568e-06,
+      "loss": 0.5187,
+      "step": 695
+    },
+    {
+      "epoch": 4.576930956771154,
+      "grad_norm": 0.08583158851506041,
+      "learning_rate": 1.7157265329634354e-06,
+      "loss": 0.5206,
+      "step": 696
+    },
+    {
+      "epoch": 4.583486990370826,
+      "grad_norm": 0.0823819160891982,
+      "learning_rate": 1.6629007016698918e-06,
+      "loss": 0.5119,
+      "step": 697
+    },
+    {
+      "epoch": 4.590043023970498,
+      "grad_norm": 0.0856122092163256,
+      "learning_rate": 1.6108836049844434e-06,
+      "loss": 0.5192,
+      "step": 698
+    },
+    {
+      "epoch": 4.59659905757017,
+      "grad_norm": 0.08366071868253996,
+      "learning_rate": 1.5596763402260462e-06,
+      "loss": 0.5169,
+      "step": 699
+    },
+    {
+      "epoch": 4.6031550911698424,
+      "grad_norm": 0.08756847365451516,
+      "learning_rate": 1.5092799876299835e-06,
+      "loss": 0.5142,
+      "step": 700
+    },
+    {
+      "epoch": 4.609711124769515,
+      "grad_norm": 0.08472854571793725,
+      "learning_rate": 1.459695610325067e-06,
+      "loss": 0.5191,
+      "step": 701
+    },
+    {
+      "epoch": 4.616267158369187,
+      "grad_norm": 0.08619093020176126,
+      "learning_rate": 1.4109242543111834e-06,
+      "loss": 0.5166,
+      "step": 702
+    },
+    {
+      "epoch": 4.622823191968859,
+      "grad_norm": 0.08301363626424127,
+      "learning_rate": 1.3629669484372722e-06,
+      "loss": 0.517,
+      "step": 703
+    },
+    {
+      "epoch": 4.629379225568531,
+      "grad_norm": 0.08230801802672034,
+      "learning_rate": 1.3158247043795735e-06,
+      "loss": 0.5033,
+      "step": 704
+    },
+    {
+      "epoch": 4.635935259168203,
+      "grad_norm": 0.08247077823518248,
+      "learning_rate": 1.2694985166203311e-06,
+      "loss": 0.523,
+      "step": 705
+    },
+    {
+      "epoch": 4.642491292767875,
+      "grad_norm": 0.08077290324213217,
+      "learning_rate": 1.2239893624267852e-06,
+      "loss": 0.5218,
+      "step": 706
+    },
+    {
+      "epoch": 4.649047326367548,
+      "grad_norm": 0.08354463750730508,
+      "learning_rate": 1.1792982018305677e-06,
+      "loss": 0.5143,
+      "step": 707
+    },
+    {
+      "epoch": 4.65560335996722,
+      "grad_norm": 0.08393485609493388,
+      "learning_rate": 1.1354259776074472e-06,
+      "loss": 0.5151,
+      "step": 708
+    },
+    {
+      "epoch": 4.662159393566892,
+      "grad_norm": 0.0832402976978181,
+      "learning_rate": 1.0923736152574428e-06,
+      "loss": 0.5233,
+      "step": 709
+    },
+    {
+      "epoch": 4.668715427166564,
+      "grad_norm": 0.0835585475483202,
+      "learning_rate": 1.050142022985292e-06,
+      "loss": 0.5093,
+      "step": 710
+    },
+    {
+      "epoch": 4.675271460766236,
+      "grad_norm": 0.08176206607867041,
+      "learning_rate": 1.0087320916813127e-06,
+      "loss": 0.5225,
+      "step": 711
+    },
+    {
+      "epoch": 4.681827494365908,
+      "grad_norm": 0.08752232597844366,
+      "learning_rate": 9.681446949025752e-07,
+      "loss": 0.5156,
+      "step": 712
+    },
+    {
+      "epoch": 4.6883835279655806,
+      "grad_norm": 0.08108785189147111,
+      "learning_rate": 9.283806888545111e-07,
+      "loss": 0.5138,
+      "step": 713
+    },
+    {
+      "epoch": 4.694939561565253,
+      "grad_norm": 0.08223618174467849,
+      "learning_rate": 8.89440912372832e-07,
+      "loss": 0.5172,
+      "step": 714
+    },
+    {
+      "epoch": 4.701495595164925,
+      "grad_norm": 0.08351263590200621,
+      "learning_rate": 8.513261869058209e-07,
+      "loss": 0.5282,
+      "step": 715
+    },
+    {
+      "epoch": 4.708051628764597,
+      "grad_norm": 0.08286053605490928,
+      "learning_rate": 8.140373164970428e-07,
+      "loss": 0.5244,
+      "step": 716
+    },
+    {
+      "epoch": 4.714607662364269,
+      "grad_norm": 0.08518895529322469,
+      "learning_rate": 7.775750877683452e-07,
+      "loss": 0.5171,
+      "step": 717
+    },
+    {
+      "epoch": 4.721163695963941,
+      "grad_norm": 0.0813459247213106,
+      "learning_rate": 7.419402699032852e-07,
+      "loss": 0.5186,
+      "step": 718
+    },
+    {
+      "epoch": 4.7277197295636135,
+      "grad_norm": 0.0803887295459458,
+      "learning_rate": 7.071336146308883e-07,
+      "loss": 0.5059,
+      "step": 719
+    },
+    {
+      "epoch": 4.734275763163287,
+      "grad_norm": 0.08172953576855356,
+      "learning_rate": 6.731558562097995e-07,
+      "loss": 0.5224,
+      "step": 720
+    },
+    {
+      "epoch": 4.740831796762959,
+      "grad_norm": 0.08549688805268194,
+      "learning_rate": 6.400077114128023e-07,
+      "loss": 0.5165,
+      "step": 721
+    },
+    {
+      "epoch": 4.747387830362631,
+      "grad_norm": 0.08286626015188583,
+      "learning_rate": 6.076898795116792e-07,
+      "loss": 0.51,
+      "step": 722
+    },
+    {
+      "epoch": 4.753943863962303,
+      "grad_norm": 0.0823372680890407,
+      "learning_rate": 5.762030422624732e-07,
+      "loss": 0.5212,
+      "step": 723
+    },
+    {
+      "epoch": 4.760499897561975,
+      "grad_norm": 0.08161043456784334,
+      "learning_rate": 5.455478638911071e-07,
+      "loss": 0.5116,
+      "step": 724
+    },
+    {
+      "epoch": 4.767055931161647,
+      "grad_norm": 0.08090694406998772,
+      "learning_rate": 5.15724991079356e-07,
+      "loss": 0.5223,
+      "step": 725
+    },
+    {
+      "epoch": 4.7736119647613195,
+      "grad_norm": 0.08130626566721383,
+      "learning_rate": 4.867350529512261e-07,
+      "loss": 0.5104,
+      "step": 726
+    },
+    {
+      "epoch": 4.780167998360992,
+      "grad_norm": 0.08487388342418309,
+      "learning_rate": 4.5857866105966763e-07,
+      "loss": 0.5129,
+      "step": 727
+    },
+    {
+      "epoch": 4.786724031960664,
+      "grad_norm": 0.0816069016289385,
+      "learning_rate": 4.3125640937368373e-07,
+      "loss": 0.5123,
+      "step": 728
+    },
+    {
+      "epoch": 4.793280065560336,
+      "grad_norm": 0.08049465996969124,
+      "learning_rate": 4.047688742657885e-07,
+      "loss": 0.5063,
+      "step": 729
+    },
+    {
+      "epoch": 4.799836099160008,
+      "grad_norm": 0.0830575657951303,
+      "learning_rate": 3.791166144998704e-07,
+      "loss": 0.5121,
+      "step": 730
+    },
+    {
+      "epoch": 4.80639213275968,
+      "grad_norm": 0.08387321061054255,
+      "learning_rate": 3.54300171219375e-07,
+      "loss": 0.5146,
+      "step": 731
+    },
+    {
+      "epoch": 4.8129481663593525,
+      "grad_norm": 0.08224120804212026,
+      "learning_rate": 3.3032006793590977e-07,
+      "loss": 0.5206,
+      "step": 732
+    },
+    {
+      "epoch": 4.819504199959025,
+      "grad_norm": 0.08072502199031102,
+      "learning_rate": 3.0717681051819935e-07,
+      "loss": 0.522,
+      "step": 733
+    },
+    {
+      "epoch": 4.826060233558697,
+      "grad_norm": 0.07860877555923813,
+      "learning_rate": 2.848708871814054e-07,
+      "loss": 0.5188,
+      "step": 734
+    },
+    {
+      "epoch": 4.832616267158369,
+      "grad_norm": 0.08001404199090854,
+      "learning_rate": 2.634027684768414e-07,
+      "loss": 0.5124,
+      "step": 735
+    },
+    {
+      "epoch": 4.839172300758041,
+      "grad_norm": 0.08310618061839851,
+      "learning_rate": 2.4277290728202063e-07,
+      "loss": 0.5212,
+      "step": 736
+    },
+    {
+      "epoch": 4.845728334357713,
+      "grad_norm": 0.0797549571951669,
+      "learning_rate": 2.2298173879113481e-07,
+      "loss": 0.5168,
+      "step": 737
+    },
+    {
+      "epoch": 4.8522843679573855,
+      "grad_norm": 0.0824361700062477,
+      "learning_rate": 2.040296805058528e-07,
+      "loss": 0.5158,
+      "step": 738
+    },
+    {
+      "epoch": 4.858840401557058,
+      "grad_norm": 0.08076093089733337,
+      "learning_rate": 1.859171322265141e-07,
+      "loss": 0.5253,
+      "step": 739
+    },
+    {
+      "epoch": 4.86539643515673,
+      "grad_norm": 0.08043719703067892,
+      "learning_rate": 1.6864447604370004e-07,
+      "loss": 0.5244,
+      "step": 740
+    },
+    {
+      "epoch": 4.871952468756403,
+      "grad_norm": 0.08277960997697664,
+      "learning_rate": 1.522120763301782e-07,
+      "loss": 0.5085,
+      "step": 741
+    },
+    {
+      "epoch": 4.878508502356075,
+      "grad_norm": 0.08180092674268333,
+      "learning_rate": 1.3662027973320614e-07,
+      "loss": 0.5202,
+      "step": 742
+    },
+    {
+      "epoch": 4.885064535955747,
+      "grad_norm": 0.08183966123917269,
+      "learning_rate": 1.2186941516722173e-07,
+      "loss": 0.5231,
+      "step": 743
+    },
+    {
+      "epoch": 4.891620569555419,
+      "grad_norm": 0.07963446358425141,
+      "learning_rate": 1.0795979380690657e-07,
+      "loss": 0.5136,
+      "step": 744
+    },
+    {
+      "epoch": 4.8981766031550915,
+      "grad_norm": 0.07852420651092652,
+      "learning_rate": 9.489170908062228e-08,
+      "loss": 0.5204,
+      "step": 745
+    },
+    {
+      "epoch": 4.904732636754764,
+      "grad_norm": 0.12877804510272148,
+      "learning_rate": 8.266543666421544e-08,
+      "loss": 0.5185,
+      "step": 746
+    },
+    {
+      "epoch": 4.911288670354436,
+      "grad_norm": 0.0869160748356661,
+      "learning_rate": 7.128123447520452e-08,
+      "loss": 0.5234,
+      "step": 747
+    },
+    {
+      "epoch": 4.917844703954108,
+      "grad_norm": 0.08086821677688971,
+      "learning_rate": 6.073934266735303e-08,
+      "loss": 0.5129,
+      "step": 748
+    },
+    {
+      "epoch": 4.92440073755378,
+      "grad_norm": 0.07880134225937953,
+      "learning_rate": 5.10399836255715e-08,
+      "loss": 0.5064,
+      "step": 749
+    },
+    {
+      "epoch": 4.930956771153452,
+      "grad_norm": 0.07797418195960645,
+      "learning_rate": 4.218336196125439e-08,
+      "loss": 0.5106,
+      "step": 750
+    },
+    {
+      "epoch": 4.9375128047531245,
+      "grad_norm": 0.08006201262285005,
+      "learning_rate": 3.416966450795922e-08,
+      "loss": 0.5099,
+      "step": 751
+    },
+    {
+      "epoch": 4.944068838352797,
+      "grad_norm": 0.08059149745693742,
+      "learning_rate": 2.699906031745414e-08,
+      "loss": 0.5176,
+      "step": 752
+    },
+    {
+      "epoch": 4.950624871952469,
+      "grad_norm": 0.08251393526468398,
+      "learning_rate": 2.067170065615187e-08,
+      "loss": 0.5235,
+      "step": 753
+    },
+    {
+      "epoch": 4.957180905552141,
+      "grad_norm": 0.08234699138692923,
+      "learning_rate": 1.5187719001943378e-08,
+      "loss": 0.5122,
+      "step": 754
+    },
+    {
+      "epoch": 4.963736939151813,
+      "grad_norm": 0.08191620156821956,
+      "learning_rate": 1.0547231041346806e-08,
+      "loss": 0.5237,
+      "step": 755
+    },
+    {
+      "epoch": 4.970292972751485,
+      "grad_norm": 0.08084676979927179,
+      "learning_rate": 6.750334667091629e-09,
+      "loss": 0.5178,
+      "step": 756
+    },
+    {
+      "epoch": 4.9768490063511575,
+      "grad_norm": 0.08117469896416148,
+      "learning_rate": 3.797109976035884e-09,
+      "loss": 0.5192,
+      "step": 757
+    },
+    {
+      "epoch": 4.98340503995083,
+      "grad_norm": 0.0821326486604627,
+      "learning_rate": 1.6876192675052695e-09,
+      "loss": 0.519,
+      "step": 758
+    },
+    {
+      "epoch": 4.989961073550502,
+      "grad_norm": 0.08465924625593949,
+      "learning_rate": 4.219070419475557e-10,
+      "loss": 0.5078,
+      "step": 759
+    },
+    {
+      "epoch": 4.996517107150174,
+      "grad_norm": 0.08042205253823026,
+      "learning_rate": 0.0,
+      "loss": 0.5193,
+      "step": 760
+    },
+    {
+      "epoch": 4.996517107150174,
+      "step": 760,
+      "total_flos": 2.0225694010775175e+19,
+      "train_loss": 0.10313788385767686,
+      "train_runtime": 35636.9099,
+      "train_samples_per_second": 10.956,
+      "train_steps_per_second": 0.021
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 760,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.0225694010775175e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}