diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8652 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.99290780141844,
+  "eval_steps": 500,
+  "global_step": 1230,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004052684903748734,
+      "grad_norm": 6.599023045587286,
+      "learning_rate": 3.2520325203252037e-07,
+      "loss": 1.0514,
+      "step": 1
+    },
+    {
+      "epoch": 0.008105369807497468,
+      "grad_norm": 6.47445504180245,
+      "learning_rate": 6.504065040650407e-07,
+      "loss": 1.0362,
+      "step": 2
+    },
+    {
+      "epoch": 0.0121580547112462,
+      "grad_norm": 6.686247018328671,
+      "learning_rate": 9.75609756097561e-07,
+      "loss": 1.0544,
+      "step": 3
+    },
+    {
+      "epoch": 0.016210739614994935,
+      "grad_norm": 6.421449402635541,
+      "learning_rate": 1.3008130081300815e-06,
+      "loss": 1.0401,
+      "step": 4
+    },
+    {
+      "epoch": 0.020263424518743668,
+      "grad_norm": 6.225536077141368,
+      "learning_rate": 1.6260162601626018e-06,
+      "loss": 1.042,
+      "step": 5
+    },
+    {
+      "epoch": 0.0243161094224924,
+      "grad_norm": 6.079425532546916,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 1.039,
+      "step": 6
+    },
+    {
+      "epoch": 0.028368794326241134,
+      "grad_norm": 4.747162327497833,
+      "learning_rate": 2.2764227642276426e-06,
+      "loss": 0.9873,
+      "step": 7
+    },
+    {
+      "epoch": 0.03242147922998987,
+      "grad_norm": 4.591298946716306,
+      "learning_rate": 2.601626016260163e-06,
+      "loss": 1.0035,
+      "step": 8
+    },
+    {
+      "epoch": 0.0364741641337386,
+      "grad_norm": 2.8712908458609747,
+      "learning_rate": 2.926829268292683e-06,
+      "loss": 0.9471,
+      "step": 9
+    },
+    {
+      "epoch": 0.040526849037487336,
+      "grad_norm": 2.562130612426219,
+      "learning_rate": 3.2520325203252037e-06,
+      "loss": 0.9319,
+      "step": 10
+    },
+    {
+      "epoch": 0.044579533941236066,
+      "grad_norm": 2.575689379913838,
+      "learning_rate": 3.577235772357724e-06,
+      "loss": 0.9467,
+      "step": 11
+    },
+    {
+      "epoch": 0.0486322188449848,
+      "grad_norm": 3.7930907311939817,
+      "learning_rate": 3.902439024390244e-06,
+      "loss": 0.9029,
+      "step": 12
+    },
+    {
+      "epoch": 0.05268490374873354,
+      "grad_norm": 4.170067449993447,
+      "learning_rate": 4.227642276422765e-06,
+      "loss": 0.924,
+      "step": 13
+    },
+    {
+      "epoch": 0.05673758865248227,
+      "grad_norm": 3.965497393388768,
+      "learning_rate": 4.552845528455285e-06,
+      "loss": 0.9197,
+      "step": 14
+    },
+    {
+      "epoch": 0.060790273556231005,
+      "grad_norm": 3.738532012764317,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 0.8887,
+      "step": 15
+    },
+    {
+      "epoch": 0.06484295845997974,
+      "grad_norm": 2.792925295998095,
+      "learning_rate": 5.203252032520326e-06,
+      "loss": 0.8642,
+      "step": 16
+    },
+    {
+      "epoch": 0.06889564336372847,
+      "grad_norm": 2.5039001786604347,
+      "learning_rate": 5.528455284552846e-06,
+      "loss": 0.839,
+      "step": 17
+    },
+    {
+      "epoch": 0.0729483282674772,
+      "grad_norm": 2.0553313011702494,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 0.8238,
+      "step": 18
+    },
+    {
+      "epoch": 0.07700101317122594,
+      "grad_norm": 1.5773393617325848,
+      "learning_rate": 6.178861788617887e-06,
+      "loss": 0.8125,
+      "step": 19
+    },
+    {
+      "epoch": 0.08105369807497467,
+      "grad_norm": 1.2593251921114146,
+      "learning_rate": 6.504065040650407e-06,
+      "loss": 0.7996,
+      "step": 20
+    },
+    {
+      "epoch": 0.0851063829787234,
+      "grad_norm": 1.211398822943914,
+      "learning_rate": 6.829268292682928e-06,
+      "loss": 0.791,
+      "step": 21
+    },
+    {
+      "epoch": 0.08915906788247213,
+      "grad_norm": 1.3023956820532772,
+      "learning_rate": 7.154471544715448e-06,
+      "loss": 0.7845,
+      "step": 22
+    },
+    {
+      "epoch": 0.09321175278622088,
+      "grad_norm": 1.1161869174225816,
+      "learning_rate": 7.4796747967479676e-06,
+      "loss": 0.7643,
+      "step": 23
+    },
+    {
+      "epoch": 0.0972644376899696,
+      "grad_norm": 1.006673189543252,
+      "learning_rate": 7.804878048780489e-06,
+      "loss": 0.7651,
+      "step": 24
+    },
+    {
+      "epoch": 0.10131712259371833,
+      "grad_norm": 0.893820815392647,
+      "learning_rate": 8.130081300813009e-06,
+      "loss": 0.7303,
+      "step": 25
+    },
+    {
+      "epoch": 0.10536980749746708,
+      "grad_norm": 0.9560473442751164,
+      "learning_rate": 8.45528455284553e-06,
+      "loss": 0.7368,
+      "step": 26
+    },
+    {
+      "epoch": 0.1094224924012158,
+      "grad_norm": 0.8802110889639827,
+      "learning_rate": 8.78048780487805e-06,
+      "loss": 0.7253,
+      "step": 27
+    },
+    {
+      "epoch": 0.11347517730496454,
+      "grad_norm": 0.748527512955968,
+      "learning_rate": 9.10569105691057e-06,
+      "loss": 0.7189,
+      "step": 28
+    },
+    {
+      "epoch": 0.11752786220871327,
+      "grad_norm": 0.7125428348195815,
+      "learning_rate": 9.43089430894309e-06,
+      "loss": 0.7207,
+      "step": 29
+    },
+    {
+      "epoch": 0.12158054711246201,
+      "grad_norm": 0.8467727891606993,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 0.7194,
+      "step": 30
+    },
+    {
+      "epoch": 0.12563323201621074,
+      "grad_norm": 0.8073050367246051,
+      "learning_rate": 1.008130081300813e-05,
+      "loss": 0.7013,
+      "step": 31
+    },
+    {
+      "epoch": 0.12968591691995948,
+      "grad_norm": 0.6270370447029,
+      "learning_rate": 1.0406504065040652e-05,
+      "loss": 0.7167,
+      "step": 32
+    },
+    {
+      "epoch": 0.1337386018237082,
+      "grad_norm": 0.7265413634779462,
+      "learning_rate": 1.0731707317073172e-05,
+      "loss": 0.7171,
+      "step": 33
+    },
+    {
+      "epoch": 0.13779128672745694,
+      "grad_norm": 0.6576559175700657,
+      "learning_rate": 1.1056910569105692e-05,
+      "loss": 0.6808,
+      "step": 34
+    },
+    {
+      "epoch": 0.14184397163120568,
+      "grad_norm": 0.8234127148418743,
+      "learning_rate": 1.1382113821138213e-05,
+      "loss": 0.6799,
+      "step": 35
+    },
+    {
+      "epoch": 0.1458966565349544,
+      "grad_norm": 0.6058143860968938,
+      "learning_rate": 1.1707317073170731e-05,
+      "loss": 0.6916,
+      "step": 36
+    },
+    {
+      "epoch": 0.14994934143870314,
+      "grad_norm": 0.6467038659776014,
+      "learning_rate": 1.2032520325203254e-05,
+      "loss": 0.6841,
+      "step": 37
+    },
+    {
+      "epoch": 0.1540020263424519,
+      "grad_norm": 0.7316541503441805,
+      "learning_rate": 1.2357723577235774e-05,
+      "loss": 0.678,
+      "step": 38
+    },
+    {
+      "epoch": 0.1580547112462006,
+      "grad_norm": 0.7472149382046376,
+      "learning_rate": 1.2682926829268294e-05,
+      "loss": 0.6889,
+      "step": 39
+    },
+    {
+      "epoch": 0.16210739614994935,
+      "grad_norm": 0.5793146399762638,
+      "learning_rate": 1.3008130081300815e-05,
+      "loss": 0.6842,
+      "step": 40
+    },
+    {
+      "epoch": 0.1661600810536981,
+      "grad_norm": 0.6875199051363161,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.6668,
+      "step": 41
+    },
+    {
+      "epoch": 0.1702127659574468,
+      "grad_norm": 0.5703715776394133,
+      "learning_rate": 1.3658536585365855e-05,
+      "loss": 0.652,
+      "step": 42
+    },
+    {
+      "epoch": 0.17426545086119555,
+      "grad_norm": 0.5352526787481642,
+      "learning_rate": 1.3983739837398376e-05,
+      "loss": 0.6664,
+      "step": 43
+    },
+    {
+      "epoch": 0.17831813576494426,
+      "grad_norm": 0.6169194929882444,
+      "learning_rate": 1.4308943089430896e-05,
+      "loss": 0.6611,
+      "step": 44
+    },
+    {
+      "epoch": 0.182370820668693,
+      "grad_norm": 0.5580859953378566,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 0.6501,
+      "step": 45
+    },
+    {
+      "epoch": 0.18642350557244175,
+      "grad_norm": 0.6224025899791772,
+      "learning_rate": 1.4959349593495935e-05,
+      "loss": 0.6579,
+      "step": 46
+    },
+    {
+      "epoch": 0.19047619047619047,
+      "grad_norm": 0.5381919334902069,
+      "learning_rate": 1.528455284552846e-05,
+      "loss": 0.6634,
+      "step": 47
+    },
+    {
+      "epoch": 0.1945288753799392,
+      "grad_norm": 0.6010567456784308,
+      "learning_rate": 1.5609756097560978e-05,
+      "loss": 0.6487,
+      "step": 48
+    },
+    {
+      "epoch": 0.19858156028368795,
+      "grad_norm": 0.6488990433414341,
+      "learning_rate": 1.5934959349593496e-05,
+      "loss": 0.6578,
+      "step": 49
+    },
+    {
+      "epoch": 0.20263424518743667,
+      "grad_norm": 0.4477246718607412,
+      "learning_rate": 1.6260162601626018e-05,
+      "loss": 0.6503,
+      "step": 50
+    },
+    {
+      "epoch": 0.2066869300911854,
+      "grad_norm": 0.6426477379434364,
+      "learning_rate": 1.6585365853658537e-05,
+      "loss": 0.6696,
+      "step": 51
+    },
+    {
+      "epoch": 0.21073961499493415,
+      "grad_norm": 0.5044801809994581,
+      "learning_rate": 1.691056910569106e-05,
+      "loss": 0.6517,
+      "step": 52
+    },
+    {
+      "epoch": 0.21479229989868287,
+      "grad_norm": 0.5119480455859519,
+      "learning_rate": 1.7235772357723578e-05,
+      "loss": 0.6511,
+      "step": 53
+    },
+    {
+      "epoch": 0.2188449848024316,
+      "grad_norm": 0.5584164405076512,
+      "learning_rate": 1.75609756097561e-05,
+      "loss": 0.6513,
+      "step": 54
+    },
+    {
+      "epoch": 0.22289766970618036,
+      "grad_norm": 0.4851555351275332,
+      "learning_rate": 1.788617886178862e-05,
+      "loss": 0.6578,
+      "step": 55
+    },
+    {
+      "epoch": 0.22695035460992907,
+      "grad_norm": 0.4791477452289817,
+      "learning_rate": 1.821138211382114e-05,
+      "loss": 0.6411,
+      "step": 56
+    },
+    {
+      "epoch": 0.23100303951367782,
+      "grad_norm": 0.6818115167524869,
+      "learning_rate": 1.8536585365853663e-05,
+      "loss": 0.6364,
+      "step": 57
+    },
+    {
+      "epoch": 0.23505572441742653,
+      "grad_norm": 0.674996775789408,
+      "learning_rate": 1.886178861788618e-05,
+      "loss": 0.6445,
+      "step": 58
+    },
+    {
+      "epoch": 0.23910840932117527,
+      "grad_norm": 0.5049699624546058,
+      "learning_rate": 1.91869918699187e-05,
+      "loss": 0.6467,
+      "step": 59
+    },
+    {
+      "epoch": 0.24316109422492402,
+      "grad_norm": 0.6910583208230778,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 0.6444,
+      "step": 60
+    },
+    {
+      "epoch": 0.24721377912867273,
+      "grad_norm": 0.7937751674365281,
+      "learning_rate": 1.983739837398374e-05,
+      "loss": 0.6325,
+      "step": 61
+    },
+    {
+      "epoch": 0.2512664640324215,
+      "grad_norm": 0.6497295123578288,
+      "learning_rate": 2.016260162601626e-05,
+      "loss": 0.6362,
+      "step": 62
+    },
+    {
+      "epoch": 0.2553191489361702,
+      "grad_norm": 0.83011660343835,
+      "learning_rate": 2.048780487804878e-05,
+      "loss": 0.6275,
+      "step": 63
+    },
+    {
+      "epoch": 0.25937183383991896,
+      "grad_norm": 0.7993008910455739,
+      "learning_rate": 2.0813008130081303e-05,
+      "loss": 0.6435,
+      "step": 64
+    },
+    {
+      "epoch": 0.2634245187436677,
+      "grad_norm": 1.08783893201976,
+      "learning_rate": 2.1138211382113822e-05,
+      "loss": 0.6449,
+      "step": 65
+    },
+    {
+      "epoch": 0.2674772036474164,
+      "grad_norm": 0.6897874653005953,
+      "learning_rate": 2.1463414634146344e-05,
+      "loss": 0.6305,
+      "step": 66
+    },
+    {
+      "epoch": 0.27152988855116517,
+      "grad_norm": 0.8919102060332686,
+      "learning_rate": 2.1788617886178863e-05,
+      "loss": 0.6267,
+      "step": 67
+    },
+    {
+      "epoch": 0.2755825734549139,
+      "grad_norm": 0.9844431258155775,
+      "learning_rate": 2.2113821138211385e-05,
+      "loss": 0.6296,
+      "step": 68
+    },
+    {
+      "epoch": 0.2796352583586626,
+      "grad_norm": 0.9194076598934081,
+      "learning_rate": 2.2439024390243907e-05,
+      "loss": 0.6344,
+      "step": 69
+    },
+    {
+      "epoch": 0.28368794326241137,
+      "grad_norm": 1.1065466031734117,
+      "learning_rate": 2.2764227642276426e-05,
+      "loss": 0.6243,
+      "step": 70
+    },
+    {
+      "epoch": 0.2877406281661601,
+      "grad_norm": 0.7208673076659101,
+      "learning_rate": 2.3089430894308948e-05,
+      "loss": 0.6159,
+      "step": 71
+    },
+    {
+      "epoch": 0.2917933130699088,
+      "grad_norm": 0.73452231554592,
+      "learning_rate": 2.3414634146341463e-05,
+      "loss": 0.6083,
+      "step": 72
+    },
+    {
+      "epoch": 0.29584599797365757,
+      "grad_norm": 0.8253144357432102,
+      "learning_rate": 2.3739837398373985e-05,
+      "loss": 0.6276,
+      "step": 73
+    },
+    {
+      "epoch": 0.2998986828774063,
+      "grad_norm": 0.8734929655139548,
+      "learning_rate": 2.4065040650406507e-05,
+      "loss": 0.6129,
+      "step": 74
+    },
+    {
+      "epoch": 0.303951367781155,
+      "grad_norm": 0.9112601084025713,
+      "learning_rate": 2.4390243902439026e-05,
+      "loss": 0.6236,
+      "step": 75
+    },
+    {
+      "epoch": 0.3080040526849038,
+      "grad_norm": 0.8415231090703136,
+      "learning_rate": 2.4715447154471548e-05,
+      "loss": 0.6263,
+      "step": 76
+    },
+    {
+      "epoch": 0.3120567375886525,
+      "grad_norm": 0.8462930373145203,
+      "learning_rate": 2.5040650406504066e-05,
+      "loss": 0.6229,
+      "step": 77
+    },
+    {
+      "epoch": 0.3161094224924012,
+      "grad_norm": 0.8727171669136281,
+      "learning_rate": 2.536585365853659e-05,
+      "loss": 0.6155,
+      "step": 78
+    },
+    {
+      "epoch": 0.32016210739615,
+      "grad_norm": 0.9599251286576289,
+      "learning_rate": 2.569105691056911e-05,
+      "loss": 0.6137,
+      "step": 79
+    },
+    {
+      "epoch": 0.3242147922998987,
+      "grad_norm": 1.0662735894475077,
+      "learning_rate": 2.601626016260163e-05,
+      "loss": 0.6111,
+      "step": 80
+    },
+    {
+      "epoch": 0.3282674772036474,
+      "grad_norm": 0.7853059023523702,
+      "learning_rate": 2.634146341463415e-05,
+      "loss": 0.6187,
+      "step": 81
+    },
+    {
+      "epoch": 0.3323201621073962,
+      "grad_norm": 0.9795949167273307,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.6168,
+      "step": 82
+    },
+    {
+      "epoch": 0.3363728470111449,
+      "grad_norm": 1.0445722702179447,
+      "learning_rate": 2.699186991869919e-05,
+      "loss": 0.6119,
+      "step": 83
+    },
+    {
+      "epoch": 0.3404255319148936,
+      "grad_norm": 0.8674808814604583,
+      "learning_rate": 2.731707317073171e-05,
+      "loss": 0.6123,
+      "step": 84
+    },
+    {
+      "epoch": 0.3444782168186423,
+      "grad_norm": 0.7114489673791807,
+      "learning_rate": 2.764227642276423e-05,
+      "loss": 0.6202,
+      "step": 85
+    },
+    {
+      "epoch": 0.3485309017223911,
+      "grad_norm": 0.7731889768192329,
+      "learning_rate": 2.796747967479675e-05,
+      "loss": 0.6287,
+      "step": 86
+    },
+    {
+      "epoch": 0.3525835866261398,
+      "grad_norm": 0.956648109933667,
+      "learning_rate": 2.829268292682927e-05,
+      "loss": 0.604,
+      "step": 87
+    },
+    {
+      "epoch": 0.3566362715298885,
+      "grad_norm": 1.0370738380771238,
+      "learning_rate": 2.8617886178861792e-05,
+      "loss": 0.609,
+      "step": 88
+    },
+    {
+      "epoch": 0.3606889564336373,
+      "grad_norm": 1.1777960561988736,
+      "learning_rate": 2.8943089430894314e-05,
+      "loss": 0.6166,
+      "step": 89
+    },
+    {
+      "epoch": 0.364741641337386,
+      "grad_norm": 1.086791362669415,
+      "learning_rate": 2.926829268292683e-05,
+      "loss": 0.6205,
+      "step": 90
+    },
+    {
+      "epoch": 0.36879432624113473,
+      "grad_norm": 0.9770790180691501,
+      "learning_rate": 2.959349593495935e-05,
+      "loss": 0.6038,
+      "step": 91
+    },
+    {
+      "epoch": 0.3728470111448835,
+      "grad_norm": 1.0327060681903977,
+      "learning_rate": 2.991869918699187e-05,
+      "loss": 0.5924,
+      "step": 92
+    },
+    {
+      "epoch": 0.3768996960486322,
+      "grad_norm": 1.1577643786737397,
+      "learning_rate": 3.0243902439024392e-05,
+      "loss": 0.5966,
+      "step": 93
+    },
+    {
+      "epoch": 0.38095238095238093,
+      "grad_norm": 0.9851717301392412,
+      "learning_rate": 3.056910569105692e-05,
+      "loss": 0.6025,
+      "step": 94
+    },
+    {
+      "epoch": 0.3850050658561297,
+      "grad_norm": 0.9305559134381358,
+      "learning_rate": 3.089430894308943e-05,
+      "loss": 0.6155,
+      "step": 95
+    },
+    {
+      "epoch": 0.3890577507598784,
+      "grad_norm": 1.0531383999376154,
+      "learning_rate": 3.1219512195121955e-05,
+      "loss": 0.6034,
+      "step": 96
+    },
+    {
+      "epoch": 0.39311043566362713,
+      "grad_norm": 1.4739439835432528,
+      "learning_rate": 3.154471544715447e-05,
+      "loss": 0.608,
+      "step": 97
+    },
+    {
+      "epoch": 0.3971631205673759,
+      "grad_norm": 0.8903811047308882,
+      "learning_rate": 3.186991869918699e-05,
+      "loss": 0.6009,
+      "step": 98
+    },
+    {
+      "epoch": 0.4012158054711246,
+      "grad_norm": 1.2890094312999683,
+      "learning_rate": 3.2195121951219514e-05,
+      "loss": 0.6145,
+      "step": 99
+    },
+    {
+      "epoch": 0.40526849037487334,
+      "grad_norm": 1.5377901588674554,
+      "learning_rate": 3.2520325203252037e-05,
+      "loss": 0.6165,
+      "step": 100
+    },
+    {
+      "epoch": 0.4093211752786221,
+      "grad_norm": 0.9798279967084077,
+      "learning_rate": 3.284552845528456e-05,
+      "loss": 0.6121,
+      "step": 101
+    },
+    {
+      "epoch": 0.4133738601823708,
+      "grad_norm": 2.374826379079598,
+      "learning_rate": 3.3170731707317074e-05,
+      "loss": 0.608,
+      "step": 102
+    },
+    {
+      "epoch": 0.41742654508611954,
+      "grad_norm": 1.8195762020549644,
+      "learning_rate": 3.3495934959349596e-05,
+      "loss": 0.607,
+      "step": 103
+    },
+    {
+      "epoch": 0.4214792299898683,
+      "grad_norm": 1.8943041468337254,
+      "learning_rate": 3.382113821138212e-05,
+      "loss": 0.611,
+      "step": 104
+    },
+    {
+      "epoch": 0.425531914893617,
+      "grad_norm": 1.9196628831168747,
+      "learning_rate": 3.414634146341463e-05,
+      "loss": 0.6119,
+      "step": 105
+    },
+    {
+      "epoch": 0.42958459979736574,
+      "grad_norm": 1.3202498996535552,
+      "learning_rate": 3.4471544715447155e-05,
+      "loss": 0.5882,
+      "step": 106
+    },
+    {
+      "epoch": 0.4336372847011145,
+      "grad_norm": 1.3681923413251689,
+      "learning_rate": 3.479674796747968e-05,
+      "loss": 0.6032,
+      "step": 107
+    },
+    {
+      "epoch": 0.4376899696048632,
+      "grad_norm": 1.4144616570974533,
+      "learning_rate": 3.51219512195122e-05,
+      "loss": 0.5977,
+      "step": 108
+    },
+    {
+      "epoch": 0.44174265450861194,
+      "grad_norm": 1.0429952162358156,
+      "learning_rate": 3.544715447154472e-05,
+      "loss": 0.6193,
+      "step": 109
+    },
+    {
+      "epoch": 0.4457953394123607,
+      "grad_norm": 1.5534020292030635,
+      "learning_rate": 3.577235772357724e-05,
+      "loss": 0.6133,
+      "step": 110
+    },
+    {
+      "epoch": 0.44984802431610943,
+      "grad_norm": 0.9920400286293563,
+      "learning_rate": 3.609756097560976e-05,
+      "loss": 0.6137,
+      "step": 111
+    },
+    {
+      "epoch": 0.45390070921985815,
+      "grad_norm": 1.7497537633064735,
+      "learning_rate": 3.642276422764228e-05,
+      "loss": 0.6036,
+      "step": 112
+    },
+    {
+      "epoch": 0.4579533941236069,
+      "grad_norm": 1.4500100577955872,
+      "learning_rate": 3.67479674796748e-05,
+      "loss": 0.6024,
+      "step": 113
+    },
+    {
+      "epoch": 0.46200607902735563,
+      "grad_norm": 1.436311058181195,
+      "learning_rate": 3.7073170731707325e-05,
+      "loss": 0.6028,
+      "step": 114
+    },
+    {
+      "epoch": 0.46605876393110435,
+      "grad_norm": 1.2660602143533284,
+      "learning_rate": 3.739837398373984e-05,
+      "loss": 0.5965,
+      "step": 115
+    },
+    {
+      "epoch": 0.47011144883485306,
+      "grad_norm": 1.183335814860972,
+      "learning_rate": 3.772357723577236e-05,
+      "loss": 0.6009,
+      "step": 116
+    },
+    {
+      "epoch": 0.47416413373860183,
+      "grad_norm": 1.186882558761872,
+      "learning_rate": 3.804878048780488e-05,
+      "loss": 0.6078,
+      "step": 117
+    },
+    {
+      "epoch": 0.47821681864235055,
+      "grad_norm": 1.2935580575767696,
+      "learning_rate": 3.83739837398374e-05,
+      "loss": 0.5862,
+      "step": 118
+    },
+    {
+      "epoch": 0.48226950354609927,
+      "grad_norm": 1.0182066029183403,
+      "learning_rate": 3.869918699186992e-05,
+      "loss": 0.5932,
+      "step": 119
+    },
+    {
+      "epoch": 0.48632218844984804,
+      "grad_norm": 1.233971548054037,
+      "learning_rate": 3.9024390243902444e-05,
+      "loss": 0.6064,
+      "step": 120
+    },
+    {
+      "epoch": 0.49037487335359675,
+      "grad_norm": 1.0587585632805925,
+      "learning_rate": 3.9349593495934966e-05,
+      "loss": 0.614,
+      "step": 121
+    },
+    {
+      "epoch": 0.49442755825734547,
+      "grad_norm": 1.026550579721245,
+      "learning_rate": 3.967479674796748e-05,
+      "loss": 0.5964,
+      "step": 122
+    },
+    {
+      "epoch": 0.49848024316109424,
+      "grad_norm": 1.0202621301409174,
+      "learning_rate": 4e-05,
+      "loss": 0.5996,
+      "step": 123
+    },
+    {
+      "epoch": 0.502532928064843,
+      "grad_norm": 0.9645823620574748,
+      "learning_rate": 3.999991946137476e-05,
+      "loss": 0.591,
+      "step": 124
+    },
+    {
+      "epoch": 0.5065856129685917,
+      "grad_norm": 0.9727069245390664,
+      "learning_rate": 3.999967784614766e-05,
+      "loss": 0.5927,
+      "step": 125
+    },
+    {
+      "epoch": 0.5106382978723404,
+      "grad_norm": 0.9575106494245289,
+      "learning_rate": 3.9999275156264656e-05,
+      "loss": 0.5841,
+      "step": 126
+    },
+    {
+      "epoch": 0.5146909827760892,
+      "grad_norm": 0.9244278735699253,
+      "learning_rate": 3.999871139496895e-05,
+      "loss": 0.5865,
+      "step": 127
+    },
+    {
+      "epoch": 0.5187436676798379,
+      "grad_norm": 0.8645848502796644,
+      "learning_rate": 3.9997986566800995e-05,
+      "loss": 0.5965,
+      "step": 128
+    },
+    {
+      "epoch": 0.5227963525835866,
+      "grad_norm": 0.7987115788482113,
+      "learning_rate": 3.999710067759846e-05,
+      "loss": 0.6002,
+      "step": 129
+    },
+    {
+      "epoch": 0.5268490374873354,
+      "grad_norm": 0.8484821578185601,
+      "learning_rate": 3.999605373449617e-05,
+      "loss": 0.5954,
+      "step": 130
+    },
+    {
+      "epoch": 0.5309017223910841,
+      "grad_norm": 0.971580608401574,
+      "learning_rate": 3.9994845745926075e-05,
+      "loss": 0.5927,
+      "step": 131
+    },
+    {
+      "epoch": 0.5349544072948328,
+      "grad_norm": 0.9747622535790835,
+      "learning_rate": 3.999347672161713e-05,
+      "loss": 0.5905,
+      "step": 132
+    },
+    {
+      "epoch": 0.5390070921985816,
+      "grad_norm": 1.1166476115686765,
+      "learning_rate": 3.999194667259528e-05,
+      "loss": 0.6016,
+      "step": 133
+    },
+    {
+      "epoch": 0.5430597771023303,
+      "grad_norm": 0.6761046565653213,
+      "learning_rate": 3.999025561118334e-05,
+      "loss": 0.5908,
+      "step": 134
+    },
+    {
+      "epoch": 0.547112462006079,
+      "grad_norm": 0.6839236271714765,
+      "learning_rate": 3.998840355100086e-05,
+      "loss": 0.5944,
+      "step": 135
+    },
+    {
+      "epoch": 0.5511651469098278,
+      "grad_norm": 0.7029152980988005,
+      "learning_rate": 3.998639050696409e-05,
+      "loss": 0.6005,
+      "step": 136
+    },
+    {
+      "epoch": 0.5552178318135765,
+      "grad_norm": 0.6258283013178816,
+      "learning_rate": 3.998421649528582e-05,
+      "loss": 0.5913,
+      "step": 137
+    },
+    {
+      "epoch": 0.5592705167173252,
+      "grad_norm": 0.9719197743275402,
+      "learning_rate": 3.9981881533475234e-05,
+      "loss": 0.5843,
+      "step": 138
+    },
+    {
+      "epoch": 0.563323201621074,
+      "grad_norm": 0.7244697798564573,
+      "learning_rate": 3.997938564033779e-05,
+      "loss": 0.5927,
+      "step": 139
+    },
+    {
+      "epoch": 0.5673758865248227,
+      "grad_norm": 0.6116455492566624,
+      "learning_rate": 3.9976728835975064e-05,
+      "loss": 0.5961,
+      "step": 140
+    },
+    {
+      "epoch": 0.5714285714285714,
+      "grad_norm": 0.5597555162335052,
+      "learning_rate": 3.9973911141784605e-05,
+      "loss": 0.5854,
+      "step": 141
+    },
+    {
+      "epoch": 0.5754812563323202,
+      "grad_norm": 0.6355188391656353,
+      "learning_rate": 3.997093258045973e-05,
+      "loss": 0.6044,
+      "step": 142
+    },
+    {
+      "epoch": 0.5795339412360689,
+      "grad_norm": 0.8091179535502104,
+      "learning_rate": 3.996779317598936e-05,
+      "loss": 0.578,
+      "step": 143
+    },
+    {
+      "epoch": 0.5835866261398176,
+      "grad_norm": 1.3524064626747305,
+      "learning_rate": 3.996449295365782e-05,
+      "loss": 0.6639,
+      "step": 144
+    },
+    {
+      "epoch": 0.5876393110435664,
+      "grad_norm": 0.6144276860924925,
+      "learning_rate": 3.996103194004467e-05,
+      "loss": 0.5946,
+      "step": 145
+    },
+    {
+      "epoch": 0.5916919959473151,
+      "grad_norm": 0.8900594482542501,
+      "learning_rate": 3.995741016302441e-05,
+      "loss": 0.5927,
+      "step": 146
+    },
+    {
+      "epoch": 0.5957446808510638,
+      "grad_norm": 1.1307082657573655,
+      "learning_rate": 3.9953627651766364e-05,
+      "loss": 0.6019,
+      "step": 147
+    },
+    {
+      "epoch": 0.5997973657548126,
+      "grad_norm": 0.6093630761114724,
+      "learning_rate": 3.9949684436734325e-05,
+      "loss": 0.5832,
+      "step": 148
+    },
+    {
+      "epoch": 0.6038500506585613,
+      "grad_norm": 0.8927507499625856,
+      "learning_rate": 3.994558054968643e-05,
+      "loss": 0.5923,
+      "step": 149
+    },
+    {
+      "epoch": 0.60790273556231,
+      "grad_norm": 0.9044347991443313,
+      "learning_rate": 3.994131602367481e-05,
+      "loss": 0.6008,
+      "step": 150
+    },
+    {
+      "epoch": 0.6119554204660588,
+      "grad_norm": 0.6889495233948225,
+      "learning_rate": 3.9936890893045376e-05,
+      "loss": 0.5769,
+      "step": 151
+    },
+    {
+      "epoch": 0.6160081053698075,
+      "grad_norm": 0.5636111290380746,
+      "learning_rate": 3.993230519343752e-05,
+      "loss": 0.5864,
+      "step": 152
+    },
+    {
+      "epoch": 0.6200607902735562,
+      "grad_norm": 0.6063529038504148,
+      "learning_rate": 3.992755896178383e-05,
+      "loss": 0.5976,
+      "step": 153
+    },
+    {
+      "epoch": 0.624113475177305,
+      "grad_norm": 0.5962853484434704,
+      "learning_rate": 3.992265223630981e-05,
+      "loss": 0.5981,
+      "step": 154
+    },
+    {
+      "epoch": 0.6281661600810537,
+      "grad_norm": 0.594526735068891,
+      "learning_rate": 3.991758505653355e-05,
+      "loss": 0.5989,
+      "step": 155
+    },
+    {
+      "epoch": 0.6322188449848024,
+      "grad_norm": 0.6732444351353275,
+      "learning_rate": 3.991235746326543e-05,
+      "loss": 0.5866,
+      "step": 156
+    },
+    {
+      "epoch": 0.6362715298885512,
+      "grad_norm": 0.692304303556478,
+      "learning_rate": 3.9906969498607745e-05,
+      "loss": 0.5918,
+      "step": 157
+    },
+    {
+      "epoch": 0.6403242147923,
+      "grad_norm": 0.6637571415045488,
+      "learning_rate": 3.990142120595444e-05,
+      "loss": 0.5884,
+      "step": 158
+    },
+    {
+      "epoch": 0.6443768996960486,
+      "grad_norm": 0.7688867999656185,
+      "learning_rate": 3.98957126299907e-05,
+      "loss": 0.5869,
+      "step": 159
+    },
+    {
+      "epoch": 0.6484295845997974,
+      "grad_norm": 0.7389578037108177,
+      "learning_rate": 3.9889843816692596e-05,
+      "loss": 0.5933,
+      "step": 160
+    },
+    {
+      "epoch": 0.6524822695035462,
+      "grad_norm": 0.6443966658312461,
+      "learning_rate": 3.9883814813326766e-05,
+      "loss": 0.5819,
+      "step": 161
+    },
+    {
+      "epoch": 0.6565349544072948,
+      "grad_norm": 0.7022403859302037,
+      "learning_rate": 3.9877625668449956e-05,
+      "loss": 0.5956,
+      "step": 162
+    },
+    {
+      "epoch": 0.6605876393110436,
+      "grad_norm": 0.6605782438262765,
+      "learning_rate": 3.98712764319087e-05,
+      "loss": 0.5794,
+      "step": 163
+    },
+    {
+      "epoch": 0.6646403242147924,
+      "grad_norm": 0.5956978402705035,
+      "learning_rate": 3.9864767154838864e-05,
+      "loss": 0.5855,
+      "step": 164
+    },
+    {
+      "epoch": 0.668693009118541,
+      "grad_norm": 0.5830874156149116,
+      "learning_rate": 3.9858097889665277e-05,
+      "loss": 0.593,
+      "step": 165
+    },
+    {
+      "epoch": 0.6727456940222898,
+      "grad_norm": 0.629907908872092,
+      "learning_rate": 3.985126869010129e-05,
+      "loss": 0.5861,
+      "step": 166
+    },
+    {
+      "epoch": 0.6767983789260384,
+      "grad_norm": 0.6851671018144985,
+      "learning_rate": 3.984427961114833e-05,
+      "loss": 0.5746,
+      "step": 167
+    },
+    {
+      "epoch": 0.6808510638297872,
+      "grad_norm": 0.5650956231842051,
+      "learning_rate": 3.9837130709095475e-05,
+      "loss": 0.6105,
+      "step": 168
+    },
+    {
+      "epoch": 0.684903748733536,
+      "grad_norm": 0.51559674706075,
+      "learning_rate": 3.982982204151901e-05,
+      "loss": 0.5921,
+      "step": 169
+    },
+    {
+      "epoch": 0.6889564336372846,
+      "grad_norm": 0.5385953173582195,
+      "learning_rate": 3.982235366728193e-05,
+      "loss": 0.5903,
+      "step": 170
+    },
+    {
+      "epoch": 0.6930091185410334,
+      "grad_norm": 0.595519385692879,
+      "learning_rate": 3.9814725646533505e-05,
+      "loss": 0.5723,
+      "step": 171
+    },
+    {
+      "epoch": 0.6970618034447822,
+      "grad_norm": 0.6876881404617907,
+      "learning_rate": 3.9806938040708746e-05,
+      "loss": 0.5759,
+      "step": 172
+    },
+    {
+      "epoch": 0.7011144883485309,
+      "grad_norm": 0.8264617094791407,
+      "learning_rate": 3.9798990912527976e-05,
+      "loss": 0.575,
+      "step": 173
+    },
+    {
+      "epoch": 0.7051671732522796,
+      "grad_norm": 0.5893607996668139,
+      "learning_rate": 3.979088432599627e-05,
+      "loss": 0.5775,
+      "step": 174
+    },
+    {
+      "epoch": 0.7092198581560284,
+      "grad_norm": 0.4919074691226593,
+      "learning_rate": 3.9782618346402964e-05,
+      "loss": 0.5583,
+      "step": 175
+    },
+    {
+      "epoch": 0.713272543059777,
+      "grad_norm": 0.6457413301136418,
+      "learning_rate": 3.977419304032111e-05,
+      "loss": 0.5829,
+      "step": 176
+    },
+    {
+      "epoch": 0.7173252279635258,
+      "grad_norm": 0.7887202287284762,
+      "learning_rate": 3.976560847560697e-05,
+      "loss": 0.588,
+      "step": 177
+    },
+    {
+      "epoch": 0.7213779128672746,
+      "grad_norm": 0.6236837362263904,
+      "learning_rate": 3.9756864721399456e-05,
+      "loss": 0.5762,
+      "step": 178
+    },
+    {
+      "epoch": 0.7254305977710233,
+      "grad_norm": 0.7917934414327535,
+      "learning_rate": 3.974796184811956e-05,
+      "loss": 0.578,
+      "step": 179
+    },
+    {
+      "epoch": 0.729483282674772,
+      "grad_norm": 0.7864273354959346,
+      "learning_rate": 3.973889992746979e-05,
+      "loss": 0.576,
+      "step": 180
+    },
+    {
+      "epoch": 0.7335359675785208,
+      "grad_norm": 0.47702577115895367,
+      "learning_rate": 3.972967903243361e-05,
+      "loss": 0.574,
+      "step": 181
+    },
+    {
+      "epoch": 0.7375886524822695,
+      "grad_norm": 0.6016512243297031,
+      "learning_rate": 3.972029923727486e-05,
+      "loss": 0.5781,
+      "step": 182
+    },
+    {
+      "epoch": 0.7416413373860182,
+      "grad_norm": 3.613647139307396,
+      "learning_rate": 3.971076061753709e-05,
+      "loss": 0.6272,
+      "step": 183
+    },
+    {
+      "epoch": 0.745694022289767,
+      "grad_norm": 0.8424059579908663,
+      "learning_rate": 3.9701063250043066e-05,
+      "loss": 0.5804,
+      "step": 184
+    },
+    {
+      "epoch": 0.7497467071935157,
+      "grad_norm": 0.7171898071813968,
+      "learning_rate": 3.969120721289402e-05,
+      "loss": 0.5734,
+      "step": 185
+    },
+    {
+      "epoch": 0.7537993920972644,
+      "grad_norm": 0.6430828612704044,
+      "learning_rate": 3.9681192585469146e-05,
+      "loss": 0.5759,
+      "step": 186
+    },
+    {
+      "epoch": 0.7578520770010132,
+      "grad_norm": 0.5411949543263597,
+      "learning_rate": 3.9671019448424865e-05,
+      "loss": 0.5772,
+      "step": 187
+    },
+    {
+      "epoch": 0.7619047619047619,
+      "grad_norm": 0.6629915907028012,
+      "learning_rate": 3.966068788369422e-05,
+      "loss": 0.5696,
+      "step": 188
+    },
+    {
+      "epoch": 0.7659574468085106,
+      "grad_norm": 0.6197495345472117,
+      "learning_rate": 3.965019797448622e-05,
+      "loss": 0.5804,
+      "step": 189
+    },
+    {
+      "epoch": 0.7700101317122594,
+      "grad_norm": 0.43877020272001177,
+      "learning_rate": 3.963954980528515e-05,
+      "loss": 0.5816,
+      "step": 190
+    },
+    {
+      "epoch": 0.7740628166160081,
+      "grad_norm": 0.6372327284226272,
+      "learning_rate": 3.9628743461849905e-05,
+      "loss": 0.5725,
+      "step": 191
+    },
+    {
+      "epoch": 0.7781155015197568,
+      "grad_norm": 0.5616348893638621,
+      "learning_rate": 3.961777903121329e-05,
+      "loss": 0.5799,
+      "step": 192
+    },
+    {
+      "epoch": 0.7821681864235056,
+      "grad_norm": 0.538090864506061,
+      "learning_rate": 3.960665660168131e-05,
+      "loss": 0.5825,
+      "step": 193
+    },
+    {
+      "epoch": 0.7862208713272543,
+      "grad_norm": 0.6543667433759188,
+      "learning_rate": 3.9595376262832485e-05,
+      "loss": 0.5902,
+      "step": 194
+    },
+    {
+      "epoch": 0.790273556231003,
+      "grad_norm": 0.5569867081731052,
+      "learning_rate": 3.9583938105517127e-05,
+      "loss": 0.5873,
+      "step": 195
+    },
+    {
+      "epoch": 0.7943262411347518,
+      "grad_norm": 0.5146474726912704,
+      "learning_rate": 3.957234222185657e-05,
+      "loss": 0.5687,
+      "step": 196
+    },
+    {
+      "epoch": 0.7983789260385005,
+      "grad_norm": 0.6075094664699849,
+      "learning_rate": 3.9560588705242474e-05,
+      "loss": 0.5751,
+      "step": 197
+    },
+    {
+      "epoch": 0.8024316109422492,
+      "grad_norm": 0.5592164823460768,
+      "learning_rate": 3.954867765033605e-05,
+      "loss": 0.5628,
+      "step": 198
+    },
+    {
+      "epoch": 0.806484295845998,
+      "grad_norm": 0.6269616882516431,
+      "learning_rate": 3.953660915306728e-05,
+      "loss": 0.5618,
+      "step": 199
+    },
+    {
+      "epoch": 0.8105369807497467,
+      "grad_norm": 0.49159093632367734,
+      "learning_rate": 3.952438331063419e-05,
+      "loss": 0.5672,
+      "step": 200
+    },
+    {
+      "epoch": 0.8145896656534954,
+      "grad_norm": 0.5412141385890064,
+      "learning_rate": 3.951200022150205e-05,
+      "loss": 0.5526,
+      "step": 201
+    },
+    {
+      "epoch": 0.8186423505572442,
+      "grad_norm": 0.5523680647526468,
+      "learning_rate": 3.949945998540253e-05,
+      "loss": 0.5738,
+      "step": 202
+    },
+    {
+      "epoch": 0.8226950354609929,
+      "grad_norm": 0.5209321699786994,
+      "learning_rate": 3.9486762703332993e-05,
+      "loss": 0.5768,
+      "step": 203
+    },
+    {
+      "epoch": 0.8267477203647416,
+      "grad_norm": 0.4951370311494425,
+      "learning_rate": 3.947390847755559e-05,
+      "loss": 0.57,
+      "step": 204
+    },
+    {
+      "epoch": 0.8308004052684904,
+      "grad_norm": 0.625356227016392,
+      "learning_rate": 3.946089741159648e-05,
+      "loss": 0.5533,
+      "step": 205
+    },
+    {
+      "epoch": 0.8348530901722391,
+      "grad_norm": 0.5991522110754549,
+      "learning_rate": 3.944772961024501e-05,
+      "loss": 0.5625,
+      "step": 206
+    },
+    {
+      "epoch": 0.8389057750759878,
+      "grad_norm": 0.470587268822535,
+      "learning_rate": 3.943440517955285e-05,
+      "loss": 0.5822,
+      "step": 207
+    },
+    {
+      "epoch": 0.8429584599797366,
+      "grad_norm": 0.6911532138154093,
+      "learning_rate": 3.9420924226833126e-05,
+      "loss": 0.5888,
+      "step": 208
+    },
+    {
+      "epoch": 0.8470111448834853,
+      "grad_norm": 0.630396527076442,
+      "learning_rate": 3.9407286860659566e-05,
+      "loss": 0.5753,
+      "step": 209
+    },
+    {
+      "epoch": 0.851063829787234,
+      "grad_norm": 0.5508056236682327,
+      "learning_rate": 3.9393493190865657e-05,
+      "loss": 0.5563,
+      "step": 210
+    },
+    {
+      "epoch": 0.8551165146909828,
+      "grad_norm": 0.678269742132698,
+      "learning_rate": 3.937954332854371e-05,
+      "loss": 0.5547,
+      "step": 211
+    },
+    {
+      "epoch": 0.8591691995947315,
+      "grad_norm": 0.6585364970834816,
+      "learning_rate": 3.9365437386044016e-05,
+      "loss": 0.5739,
+      "step": 212
+    },
+    {
+      "epoch": 0.8632218844984803,
+      "grad_norm": 0.4660216266936489,
+      "learning_rate": 3.935117547697387e-05,
+      "loss": 0.5817,
+      "step": 213
+    },
+    {
+      "epoch": 0.867274569402229,
+      "grad_norm": 0.5923321720157314,
+      "learning_rate": 3.933675771619675e-05,
+      "loss": 0.569,
+      "step": 214
+    },
+    {
+      "epoch": 0.8713272543059777,
+      "grad_norm": 0.5694789946119627,
+      "learning_rate": 3.932218421983131e-05,
+      "loss": 0.5746,
+      "step": 215
+    },
+    {
+      "epoch": 0.8753799392097265,
+      "grad_norm": 0.5569276407724351,
+      "learning_rate": 3.9307455105250484e-05,
+      "loss": 0.5436,
+      "step": 216
+    },
+    {
+      "epoch": 0.8794326241134752,
+      "grad_norm": 0.4820247724431898,
+      "learning_rate": 3.929257049108054e-05,
+      "loss": 0.581,
+      "step": 217
+    },
+    {
+      "epoch": 0.8834853090172239,
+      "grad_norm": 0.47948368383095985,
+      "learning_rate": 3.927753049720011e-05,
+      "loss": 0.5687,
+      "step": 218
+    },
+    {
+      "epoch": 0.8875379939209727,
+      "grad_norm": 0.6066256958719515,
+      "learning_rate": 3.9262335244739234e-05,
+      "loss": 0.573,
+      "step": 219
+    },
+    {
+      "epoch": 0.8915906788247214,
+      "grad_norm": 0.6211350059735337,
+      "learning_rate": 3.92469848560784e-05,
+      "loss": 0.5804,
+      "step": 220
+    },
+    {
+      "epoch": 0.8956433637284701,
+      "grad_norm": 0.5206410858855138,
+      "learning_rate": 3.923147945484751e-05,
+      "loss": 0.5648,
+      "step": 221
+    },
+    {
+      "epoch": 0.8996960486322189,
+      "grad_norm": 0.42830071517698765,
+      "learning_rate": 3.9215819165924956e-05,
+      "loss": 0.5603,
+      "step": 222
+    },
+    {
+      "epoch": 0.9037487335359676,
+      "grad_norm": 0.41200542849521593,
+      "learning_rate": 3.920000411543654e-05,
+      "loss": 0.5651,
+      "step": 223
+    },
+    {
+      "epoch": 0.9078014184397163,
+      "grad_norm": 0.4290409268364959,
+      "learning_rate": 3.9184034430754495e-05,
+      "loss": 0.565,
+      "step": 224
+    },
+    {
+      "epoch": 0.9118541033434651,
+      "grad_norm": 0.4405104643499867,
+      "learning_rate": 3.916791024049648e-05,
+      "loss": 0.5632,
+      "step": 225
+    },
+    {
+      "epoch": 0.9159067882472138,
+      "grad_norm": 0.42779429421192805,
+      "learning_rate": 3.91516316745245e-05,
+      "loss": 0.5613,
+      "step": 226
+    },
+    {
+      "epoch": 0.9199594731509625,
+      "grad_norm": 0.4206017277162419,
+      "learning_rate": 3.913519886394389e-05,
+      "loss": 0.5753,
+      "step": 227
+    },
+    {
+      "epoch": 0.9240121580547113,
+      "grad_norm": 0.4582584226329624,
+      "learning_rate": 3.911861194110225e-05,
+      "loss": 0.5742,
+      "step": 228
+    },
+    {
+      "epoch": 0.92806484295846,
+      "grad_norm": 0.4800620788616759,
+      "learning_rate": 3.910187103958837e-05,
+      "loss": 0.5665,
+      "step": 229
+    },
+    {
+      "epoch": 0.9321175278622087,
+      "grad_norm": 0.5245292915572244,
+      "learning_rate": 3.908497629423117e-05,
+      "loss": 0.5757,
+      "step": 230
+    },
+    {
+      "epoch": 0.9361702127659575,
+      "grad_norm": 0.47694404335999496,
+      "learning_rate": 3.9067927841098614e-05,
+      "loss": 0.5649,
+      "step": 231
+    },
+    {
+      "epoch": 0.9402228976697061,
+      "grad_norm": 0.5225235646206117,
+      "learning_rate": 3.9050725817496594e-05,
+      "loss": 0.5697,
+      "step": 232
+    },
+    {
+      "epoch": 0.9442755825734549,
+      "grad_norm": 0.5304360342225947,
+      "learning_rate": 3.9033370361967844e-05,
+      "loss": 0.5702,
+      "step": 233
+    },
+    {
+      "epoch": 0.9483282674772037,
+      "grad_norm": 0.5091531550627542,
+      "learning_rate": 3.901586161429081e-05,
+      "loss": 0.5696,
+      "step": 234
+    },
+    {
+      "epoch": 0.9523809523809523,
+      "grad_norm": 0.5719550465702286,
+      "learning_rate": 3.8998199715478545e-05,
+      "loss": 0.5663,
+      "step": 235
+    },
+    {
+      "epoch": 0.9564336372847011,
+      "grad_norm": 0.70978299490918,
+      "learning_rate": 3.8980384807777564e-05,
+      "loss": 0.5736,
+      "step": 236
+    },
+    {
+      "epoch": 0.9604863221884499,
+      "grad_norm": 0.4549419224132296,
+      "learning_rate": 3.896241703466667e-05,
+      "loss": 0.5723,
+      "step": 237
+    },
+    {
+      "epoch": 0.9645390070921985,
+      "grad_norm": 0.5712759760236225,
+      "learning_rate": 3.894429654085585e-05,
+      "loss": 0.567,
+      "step": 238
+    },
+    {
+      "epoch": 0.9685916919959473,
+      "grad_norm": 0.525829280693854,
+      "learning_rate": 3.892602347228505e-05,
+      "loss": 0.555,
+      "step": 239
+    },
+    {
+      "epoch": 0.9726443768996961,
+      "grad_norm": 0.4739037818682846,
+      "learning_rate": 3.890759797612307e-05,
+      "loss": 0.556,
+      "step": 240
+    },
+    {
+      "epoch": 0.9766970618034447,
+      "grad_norm": 0.617108641706374,
+      "learning_rate": 3.888902020076632e-05,
+      "loss": 0.5656,
+      "step": 241
+    },
+    {
+      "epoch": 0.9807497467071935,
+      "grad_norm": 0.4781533838474172,
+      "learning_rate": 3.887029029583764e-05,
+      "loss": 0.5753,
+      "step": 242
+    },
+    {
+      "epoch": 0.9848024316109423,
+      "grad_norm": 0.42492821789126733,
+      "learning_rate": 3.8851408412185125e-05,
+      "loss": 0.5644,
+      "step": 243
+    },
+    {
+      "epoch": 0.9888551165146909,
+      "grad_norm": 0.48867856629364304,
+      "learning_rate": 3.8832374701880855e-05,
+      "loss": 0.5503,
+      "step": 244
+    },
+    {
+      "epoch": 0.9929078014184397,
+      "grad_norm": 0.4492864320443739,
+      "learning_rate": 3.881318931821972e-05,
+      "loss": 0.5727,
+      "step": 245
+    },
+    {
+      "epoch": 0.9969604863221885,
+      "grad_norm": 0.4713016518803041,
+      "learning_rate": 3.879385241571817e-05,
+      "loss": 0.5711,
+      "step": 246
+    },
+    {
+      "epoch": 1.0030395136778116,
+      "grad_norm": 0.557748540238745,
+      "learning_rate": 3.8774364150112955e-05,
+      "loss": 0.5424,
+      "step": 247
+    },
+    {
+      "epoch": 1.0070921985815602,
+      "grad_norm": 0.4621148175796167,
+      "learning_rate": 3.8754724678359884e-05,
+      "loss": 0.5231,
+      "step": 248
+    },
+    {
+      "epoch": 1.011144883485309,
+      "grad_norm": 0.6000954945612695,
+      "learning_rate": 3.873493415863256e-05,
+      "loss": 0.5134,
+      "step": 249
+    },
+    {
+      "epoch": 1.0151975683890577,
+      "grad_norm": 0.5535341535106301,
+      "learning_rate": 3.871499275032111e-05,
+      "loss": 0.5215,
+      "step": 250
+    },
+    {
+      "epoch": 1.0192502532928065,
+      "grad_norm": 0.5322754355179768,
+      "learning_rate": 3.869490061403091e-05,
+      "loss": 0.5264,
+      "step": 251
+    },
+    {
+      "epoch": 1.0233029381965553,
+      "grad_norm": 0.5582598034606526,
+      "learning_rate": 3.867465791158124e-05,
+      "loss": 0.5041,
+      "step": 252
+    },
+    {
+      "epoch": 1.027355623100304,
+      "grad_norm": 0.4871602974793124,
+      "learning_rate": 3.865426480600407e-05,
+      "loss": 0.5165,
+      "step": 253
+    },
+    {
+      "epoch": 1.0314083080040526,
+      "grad_norm": 0.45220811140253986,
+      "learning_rate": 3.863372146154264e-05,
+      "loss": 0.494,
+      "step": 254
+    },
+    {
+      "epoch": 1.0354609929078014,
+      "grad_norm": 0.5781802420159897,
+      "learning_rate": 3.861302804365024e-05,
+      "loss": 0.5204,
+      "step": 255
+    },
+    {
+      "epoch": 1.0395136778115501,
+      "grad_norm": 0.49766362648363055,
+      "learning_rate": 3.85921847189888e-05,
+      "loss": 0.5194,
+      "step": 256
+    },
+    {
+      "epoch": 1.043566362715299,
+      "grad_norm": 0.7462236846176246,
+      "learning_rate": 3.85711916554276e-05,
+      "loss": 0.5143,
+      "step": 257
+    },
+    {
+      "epoch": 1.0476190476190477,
+      "grad_norm": 0.5299219145348918,
+      "learning_rate": 3.85500490220419e-05,
+      "loss": 0.5255,
+      "step": 258
+    },
+    {
+      "epoch": 1.0516717325227964,
+      "grad_norm": 0.6539443857614234,
+      "learning_rate": 3.852875698911154e-05,
+      "loss": 0.5098,
+      "step": 259
+    },
+    {
+      "epoch": 1.055724417426545,
+      "grad_norm": 0.6274335444857269,
+      "learning_rate": 3.850731572811963e-05,
+      "loss": 0.5156,
+      "step": 260
+    },
+    {
+      "epoch": 1.0597771023302938,
+      "grad_norm": 0.5194345553113586,
+      "learning_rate": 3.848572541175116e-05,
+      "loss": 0.527,
+      "step": 261
+    },
+    {
+      "epoch": 1.0638297872340425,
+      "grad_norm": 0.46848556483468673,
+      "learning_rate": 3.846398621389154e-05,
+      "loss": 0.5111,
+      "step": 262
+    },
+    {
+      "epoch": 1.0678824721377913,
+      "grad_norm": 0.49878842850120786,
+      "learning_rate": 3.84420983096253e-05,
+      "loss": 0.536,
+      "step": 263
+    },
+    {
+      "epoch": 1.07193515704154,
+      "grad_norm": 0.6059383530690796,
+      "learning_rate": 3.8420061875234606e-05,
+      "loss": 0.5249,
+      "step": 264
+    },
+    {
+      "epoch": 1.0759878419452888,
+      "grad_norm": 0.5790965591868338,
+      "learning_rate": 3.839787708819787e-05,
+      "loss": 0.5153,
+      "step": 265
+    },
+    {
+      "epoch": 1.0800405268490374,
+      "grad_norm": 0.5563861248066765,
+      "learning_rate": 3.8375544127188325e-05,
+      "loss": 0.5205,
+      "step": 266
+    },
+    {
+      "epoch": 1.0840932117527862,
+      "grad_norm": 0.46485524046621957,
+      "learning_rate": 3.8353063172072564e-05,
+      "loss": 0.5222,
+      "step": 267
+    },
+    {
+      "epoch": 1.088145896656535,
+      "grad_norm": 0.6112401445203366,
+      "learning_rate": 3.8330434403909105e-05,
+      "loss": 0.5027,
+      "step": 268
+    },
+    {
+      "epoch": 1.0921985815602837,
+      "grad_norm": 0.6391308635690197,
+      "learning_rate": 3.8307658004946934e-05,
+      "loss": 0.5111,
+      "step": 269
+    },
+    {
+      "epoch": 1.0962512664640325,
+      "grad_norm": 0.47946903579815237,
+      "learning_rate": 3.8284734158624046e-05,
+      "loss": 0.5165,
+      "step": 270
+    },
+    {
+      "epoch": 1.1003039513677813,
+      "grad_norm": 0.5023507479940311,
+      "learning_rate": 3.826166304956594e-05,
+      "loss": 0.5179,
+      "step": 271
+    },
+    {
+      "epoch": 1.1043566362715298,
+      "grad_norm": 0.5185841466978587,
+      "learning_rate": 3.8238444863584164e-05,
+      "loss": 0.522,
+      "step": 272
+    },
+    {
+      "epoch": 1.1084093211752786,
+      "grad_norm": 0.4517335732324085,
+      "learning_rate": 3.821507978767479e-05,
+      "loss": 0.5142,
+      "step": 273
+    },
+    {
+      "epoch": 1.1124620060790273,
+      "grad_norm": 0.46604263922948747,
+      "learning_rate": 3.819156801001693e-05,
+      "loss": 0.5001,
+      "step": 274
+    },
+    {
+      "epoch": 1.1165146909827761,
+      "grad_norm": 0.5147897257218349,
+      "learning_rate": 3.816790971997121e-05,
+      "loss": 0.5155,
+      "step": 275
+    },
+    {
+      "epoch": 1.1205673758865249,
+      "grad_norm": 0.4254210258903041,
+      "learning_rate": 3.8144105108078246e-05,
+      "loss": 0.5156,
+      "step": 276
+    },
+    {
+      "epoch": 1.1246200607902737,
+      "grad_norm": 0.5500677061121733,
+      "learning_rate": 3.81201543660571e-05,
+      "loss": 0.5038,
+      "step": 277
+    },
+    {
+      "epoch": 1.1286727456940222,
+      "grad_norm": 0.5780053780820302,
+      "learning_rate": 3.809605768680377e-05,
+      "loss": 0.5303,
+      "step": 278
+    },
+    {
+      "epoch": 1.132725430597771,
+      "grad_norm": 0.48038898918683315,
+      "learning_rate": 3.807181526438958e-05,
+      "loss": 0.4946,
+      "step": 279
+    },
+    {
+      "epoch": 1.1367781155015197,
+      "grad_norm": 0.4737413153652073,
+      "learning_rate": 3.8047427294059697e-05,
+      "loss": 0.5096,
+      "step": 280
+    },
+    {
+      "epoch": 1.1408308004052685,
+      "grad_norm": 0.6956302709338007,
+      "learning_rate": 3.802289397223145e-05,
+      "loss": 0.5313,
+      "step": 281
+    },
+    {
+      "epoch": 1.1448834853090173,
+      "grad_norm": 0.5529285979675868,
+      "learning_rate": 3.7998215496492854e-05,
+      "loss": 0.5103,
+      "step": 282
+    },
+    {
+      "epoch": 1.148936170212766,
+      "grad_norm": 0.4784796828354979,
+      "learning_rate": 3.797339206560096e-05,
+      "loss": 0.4977,
+      "step": 283
+    },
+    {
+      "epoch": 1.1529888551165146,
+      "grad_norm": 0.6014494635914271,
+      "learning_rate": 3.794842387948027e-05,
+      "loss": 0.5375,
+      "step": 284
+    },
+    {
+      "epoch": 1.1570415400202634,
+      "grad_norm": 0.5726868670007542,
+      "learning_rate": 3.7923311139221114e-05,
+      "loss": 0.5235,
+      "step": 285
+    },
+    {
+      "epoch": 1.1610942249240122,
+      "grad_norm": 0.526775308165516,
+      "learning_rate": 3.7898054047078054e-05,
+      "loss": 0.5084,
+      "step": 286
+    },
+    {
+      "epoch": 1.165146909827761,
+      "grad_norm": 0.5666402167707913,
+      "learning_rate": 3.787265280646825e-05,
+      "loss": 0.5303,
+      "step": 287
+    },
+    {
+      "epoch": 1.1691995947315097,
+      "grad_norm": 0.41073723547254526,
+      "learning_rate": 3.7847107621969786e-05,
+      "loss": 0.52,
+      "step": 288
+    },
+    {
+      "epoch": 1.1732522796352582,
+      "grad_norm": 0.47841656699954654,
+      "learning_rate": 3.7821418699320064e-05,
+      "loss": 0.5107,
+      "step": 289
+    },
+    {
+      "epoch": 1.177304964539007,
+      "grad_norm": 0.4954360249327025,
+      "learning_rate": 3.7795586245414145e-05,
+      "loss": 0.4967,
+      "step": 290
+    },
+    {
+      "epoch": 1.1813576494427558,
+      "grad_norm": 0.4943642634271794,
+      "learning_rate": 3.776961046830306e-05,
+      "loss": 0.5206,
+      "step": 291
+    },
+    {
+      "epoch": 1.1854103343465046,
+      "grad_norm": 0.4902327515935834,
+      "learning_rate": 3.774349157719215e-05,
+      "loss": 0.5156,
+      "step": 292
+    },
+    {
+      "epoch": 1.1894630192502533,
+      "grad_norm": 0.4757190487796499,
+      "learning_rate": 3.7717229782439365e-05,
+      "loss": 0.52,
+      "step": 293
+    },
+    {
+      "epoch": 1.193515704154002,
+      "grad_norm": 0.44536843193264514,
+      "learning_rate": 3.769082529555359e-05,
+      "loss": 0.5373,
+      "step": 294
+    },
+    {
+      "epoch": 1.1975683890577509,
+      "grad_norm": 0.4467646110528696,
+      "learning_rate": 3.766427832919294e-05,
+      "loss": 0.4917,
+      "step": 295
+    },
+    {
+      "epoch": 1.2016210739614994,
+      "grad_norm": 0.47799260112851605,
+      "learning_rate": 3.7637589097163024e-05,
+      "loss": 0.5166,
+      "step": 296
+    },
+    {
+      "epoch": 1.2056737588652482,
+      "grad_norm": 0.4735206304259062,
+      "learning_rate": 3.761075781441526e-05,
+      "loss": 0.5146,
+      "step": 297
+    },
+    {
+      "epoch": 1.209726443768997,
+      "grad_norm": 0.45005909508358527,
+      "learning_rate": 3.75837846970451e-05,
+      "loss": 0.5383,
+      "step": 298
+    },
+    {
+      "epoch": 1.2137791286727457,
+      "grad_norm": 0.5473442276927805,
+      "learning_rate": 3.755666996229032e-05,
+      "loss": 0.5077,
+      "step": 299
+    },
+    {
+      "epoch": 1.2178318135764945,
+      "grad_norm": 0.4609985977708474,
+      "learning_rate": 3.752941382852927e-05,
+      "loss": 0.541,
+      "step": 300
+    },
+    {
+      "epoch": 1.221884498480243,
+      "grad_norm": 0.4898295955606378,
+      "learning_rate": 3.7502016515279115e-05,
+      "loss": 0.5231,
+      "step": 301
+    },
+    {
+      "epoch": 1.2259371833839918,
+      "grad_norm": 0.48905209617225276,
+      "learning_rate": 3.7474478243194043e-05,
+      "loss": 0.497,
+      "step": 302
+    },
+    {
+      "epoch": 1.2299898682877406,
+      "grad_norm": 0.5177014980600935,
+      "learning_rate": 3.744679923406351e-05,
+      "loss": 0.4971,
+      "step": 303
+    },
+    {
+      "epoch": 1.2340425531914894,
+      "grad_norm": 0.462703001863924,
+      "learning_rate": 3.741897971081043e-05,
+      "loss": 0.504,
+      "step": 304
+    },
+    {
+      "epoch": 1.2380952380952381,
+      "grad_norm": 0.49446811749889646,
+      "learning_rate": 3.739101989748946e-05,
+      "loss": 0.5001,
+      "step": 305
+    },
+    {
+      "epoch": 1.242147922998987,
+      "grad_norm": 0.46788869282102064,
+      "learning_rate": 3.7362920019285066e-05,
+      "loss": 0.5289,
+      "step": 306
+    },
+    {
+      "epoch": 1.2462006079027357,
+      "grad_norm": 0.5216593834012441,
+      "learning_rate": 3.73346803025098e-05,
+      "loss": 0.5269,
+      "step": 307
+    },
+    {
+      "epoch": 1.2502532928064842,
+      "grad_norm": 0.3844022698212344,
+      "learning_rate": 3.730630097460247e-05,
+      "loss": 0.5085,
+      "step": 308
+    },
+    {
+      "epoch": 1.254305977710233,
+      "grad_norm": 0.5163509112687578,
+      "learning_rate": 3.727778226412628e-05,
+      "loss": 0.5086,
+      "step": 309
+    },
+    {
+      "epoch": 1.2583586626139818,
+      "grad_norm": 0.5400855629158948,
+      "learning_rate": 3.7249124400767006e-05,
+      "loss": 0.5215,
+      "step": 310
+    },
+    {
+      "epoch": 1.2624113475177305,
+      "grad_norm": 0.36928789401375145,
+      "learning_rate": 3.722032761533114e-05,
+      "loss": 0.5078,
+      "step": 311
+    },
+    {
+      "epoch": 1.2664640324214793,
+      "grad_norm": 0.49916489726792845,
+      "learning_rate": 3.719139213974403e-05,
+      "loss": 0.5268,
+      "step": 312
+    },
+    {
+      "epoch": 1.2705167173252279,
+      "grad_norm": 0.43250673810166007,
+      "learning_rate": 3.7162318207048006e-05,
+      "loss": 0.5081,
+      "step": 313
+    },
+    {
+      "epoch": 1.2745694022289766,
+      "grad_norm": 0.3641961668596623,
+      "learning_rate": 3.713310605140055e-05,
+      "loss": 0.5157,
+      "step": 314
+    },
+    {
+      "epoch": 1.2786220871327254,
+      "grad_norm": 0.4708017584884893,
+      "learning_rate": 3.710375590807233e-05,
+      "loss": 0.5239,
+      "step": 315
+    },
+    {
+      "epoch": 1.2826747720364742,
+      "grad_norm": 0.4469251197502781,
+      "learning_rate": 3.7074268013445365e-05,
+      "loss": 0.5191,
+      "step": 316
+    },
+    {
+      "epoch": 1.286727456940223,
+      "grad_norm": 0.4170792540276034,
+      "learning_rate": 3.7044642605011114e-05,
+      "loss": 0.5271,
+      "step": 317
+    },
+    {
+      "epoch": 1.2907801418439715,
+      "grad_norm": 0.8326916771854185,
+      "learning_rate": 3.701487992136854e-05,
+      "loss": 0.5223,
+      "step": 318
+    },
+    {
+      "epoch": 1.2948328267477205,
+      "grad_norm": 0.38862422748363334,
+      "learning_rate": 3.69849802022222e-05,
+      "loss": 0.5254,
+      "step": 319
+    },
+    {
+      "epoch": 1.298885511651469,
+      "grad_norm": 1.263726522086109,
+      "learning_rate": 3.6954943688380334e-05,
+      "loss": 0.5264,
+      "step": 320
+    },
+    {
+      "epoch": 1.3029381965552178,
+      "grad_norm": 0.39284217743353667,
+      "learning_rate": 3.692477062175289e-05,
+      "loss": 0.5009,
+      "step": 321
+    },
+    {
+      "epoch": 1.3069908814589666,
+      "grad_norm": 0.5596339534907246,
+      "learning_rate": 3.689446124534958e-05,
+      "loss": 0.5079,
+      "step": 322
+    },
+    {
+      "epoch": 1.3110435663627154,
+      "grad_norm": 0.47488138701711996,
+      "learning_rate": 3.686401580327799e-05,
+      "loss": 0.5179,
+      "step": 323
+    },
+    {
+      "epoch": 1.3150962512664641,
+      "grad_norm": 0.42193105336906295,
+      "learning_rate": 3.683343454074149e-05,
+      "loss": 0.5112,
+      "step": 324
+    },
+    {
+      "epoch": 1.3191489361702127,
+      "grad_norm": 0.41714536262216384,
+      "learning_rate": 3.6802717704037386e-05,
+      "loss": 0.5359,
+      "step": 325
+    },
+    {
+      "epoch": 1.3232016210739614,
+      "grad_norm": 0.4523953322512892,
+      "learning_rate": 3.6771865540554855e-05,
+      "loss": 0.5164,
+      "step": 326
+    },
+    {
+      "epoch": 1.3272543059777102,
+      "grad_norm": 0.5124556975023729,
+      "learning_rate": 3.674087829877297e-05,
+      "loss": 0.5295,
+      "step": 327
+    },
+    {
+      "epoch": 1.331306990881459,
+      "grad_norm": 0.5163116635384236,
+      "learning_rate": 3.6709756228258735e-05,
+      "loss": 0.5229,
+      "step": 328
+    },
+    {
+      "epoch": 1.3353596757852078,
+      "grad_norm": 0.4757158040810718,
+      "learning_rate": 3.667849957966501e-05,
+      "loss": 0.5158,
+      "step": 329
+    },
+    {
+      "epoch": 1.3394123606889563,
+      "grad_norm": 0.4247555418435866,
+      "learning_rate": 3.6647108604728546e-05,
+      "loss": 0.5168,
+      "step": 330
+    },
+    {
+      "epoch": 1.3434650455927053,
+      "grad_norm": 0.5571299450683076,
+      "learning_rate": 3.661558355626795e-05,
+      "loss": 0.5192,
+      "step": 331
+    },
+    {
+      "epoch": 1.3475177304964538,
+      "grad_norm": 0.5501084967380048,
+      "learning_rate": 3.658392468818163e-05,
+      "loss": 0.5299,
+      "step": 332
+    },
+    {
+      "epoch": 1.3515704154002026,
+      "grad_norm": 0.4472445692430267,
+      "learning_rate": 3.655213225544574e-05,
+      "loss": 0.5086,
+      "step": 333
+    },
+    {
+      "epoch": 1.3556231003039514,
+      "grad_norm": 0.41696880674648146,
+      "learning_rate": 3.652020651411218e-05,
+      "loss": 0.5273,
+      "step": 334
+    },
+    {
+      "epoch": 1.3596757852077002,
+      "grad_norm": 0.4019100709648588,
+      "learning_rate": 3.6488147721306474e-05,
+      "loss": 0.5153,
+      "step": 335
+    },
+    {
+      "epoch": 1.363728470111449,
+      "grad_norm": 0.5436194254852269,
+      "learning_rate": 3.645595613522574e-05,
+      "loss": 0.506,
+      "step": 336
+    },
+    {
+      "epoch": 1.3677811550151975,
+      "grad_norm": 0.46236889571111445,
+      "learning_rate": 3.642363201513657e-05,
+      "loss": 0.5052,
+      "step": 337
+    },
+    {
+      "epoch": 1.3718338399189463,
+      "grad_norm": 0.4903359934637521,
+      "learning_rate": 3.6391175621373006e-05,
+      "loss": 0.515,
+      "step": 338
+    },
+    {
+      "epoch": 1.375886524822695,
+      "grad_norm": 0.4995139536483641,
+      "learning_rate": 3.6358587215334355e-05,
+      "loss": 0.4996,
+      "step": 339
+    },
+    {
+      "epoch": 1.3799392097264438,
+      "grad_norm": 0.50836386237976,
+      "learning_rate": 3.632586705948318e-05,
+      "loss": 0.5174,
+      "step": 340
+    },
+    {
+      "epoch": 1.3839918946301926,
+      "grad_norm": 0.5838425279381627,
+      "learning_rate": 3.629301541734311e-05,
+      "loss": 0.5068,
+      "step": 341
+    },
+    {
+      "epoch": 1.3880445795339411,
+      "grad_norm": 0.4779767665462759,
+      "learning_rate": 3.626003255349676e-05,
+      "loss": 0.5146,
+      "step": 342
+    },
+    {
+      "epoch": 1.39209726443769,
+      "grad_norm": 0.6312449403964024,
+      "learning_rate": 3.622691873358357e-05,
+      "loss": 0.5017,
+      "step": 343
+    },
+    {
+      "epoch": 1.3961499493414387,
+      "grad_norm": 0.6777622678473971,
+      "learning_rate": 3.61936742242977e-05,
+      "loss": 0.526,
+      "step": 344
+    },
+    {
+      "epoch": 1.4002026342451874,
+      "grad_norm": 0.5438415883023792,
+      "learning_rate": 3.6160299293385864e-05,
+      "loss": 0.5111,
+      "step": 345
+    },
+    {
+      "epoch": 1.4042553191489362,
+      "grad_norm": 0.6023331337036119,
+      "learning_rate": 3.612679420964516e-05,
+      "loss": 0.5092,
+      "step": 346
+    },
+    {
+      "epoch": 1.408308004052685,
+      "grad_norm": 0.6440575479298628,
+      "learning_rate": 3.609315924292092e-05,
+      "loss": 0.5015,
+      "step": 347
+    },
+    {
+      "epoch": 1.4123606889564337,
+      "grad_norm": 0.36807012829611774,
+      "learning_rate": 3.6059394664104554e-05,
+      "loss": 0.5132,
+      "step": 348
+    },
+    {
+      "epoch": 1.4164133738601823,
+      "grad_norm": 0.5321152648650316,
+      "learning_rate": 3.602550074513133e-05,
+      "loss": 0.5046,
+      "step": 349
+    },
+    {
+      "epoch": 1.420466058763931,
+      "grad_norm": 0.3879443485912782,
+      "learning_rate": 3.599147775897822e-05,
+      "loss": 0.5147,
+      "step": 350
+    },
+    {
+      "epoch": 1.4245187436676798,
+      "grad_norm": 0.4533658510888897,
+      "learning_rate": 3.595732597966167e-05,
+      "loss": 0.5221,
+      "step": 351
+    },
+    {
+      "epoch": 1.4285714285714286,
+      "grad_norm": 0.3514284045242658,
+      "learning_rate": 3.592304568223542e-05,
+      "loss": 0.5261,
+      "step": 352
+    },
+    {
+      "epoch": 1.4326241134751774,
+      "grad_norm": 0.42960114927603743,
+      "learning_rate": 3.588863714278826e-05,
+      "loss": 0.5239,
+      "step": 353
+    },
+    {
+      "epoch": 1.436676798378926,
+      "grad_norm": 0.4157234266114839,
+      "learning_rate": 3.585410063844186e-05,
+      "loss": 0.5136,
+      "step": 354
+    },
+    {
+      "epoch": 1.4407294832826747,
+      "grad_norm": 0.3868995120542772,
+      "learning_rate": 3.581943644734846e-05,
+      "loss": 0.5134,
+      "step": 355
+    },
+    {
+      "epoch": 1.4447821681864235,
+      "grad_norm": 0.3725545268594919,
+      "learning_rate": 3.578464484868869e-05,
+      "loss": 0.5028,
+      "step": 356
+    },
+    {
+      "epoch": 1.4488348530901722,
+      "grad_norm": 0.4222244466611466,
+      "learning_rate": 3.5749726122669316e-05,
+      "loss": 0.4973,
+      "step": 357
+    },
+    {
+      "epoch": 1.452887537993921,
+      "grad_norm": 0.3404146357414848,
+      "learning_rate": 3.5714680550520943e-05,
+      "loss": 0.5097,
+      "step": 358
+    },
+    {
+      "epoch": 1.4569402228976698,
+      "grad_norm": 0.48650870685084163,
+      "learning_rate": 3.5679508414495794e-05,
+      "loss": 0.5137,
+      "step": 359
+    },
+    {
+      "epoch": 1.4609929078014185,
+      "grad_norm": 0.47532553407670547,
+      "learning_rate": 3.564420999786543e-05,
+      "loss": 0.511,
+      "step": 360
+    },
+    {
+      "epoch": 1.465045592705167,
+      "grad_norm": 0.40615098973085634,
+      "learning_rate": 3.560878558491842e-05,
+      "loss": 0.4951,
+      "step": 361
+    },
+    {
+      "epoch": 1.4690982776089159,
+      "grad_norm": 0.3843917841242262,
+      "learning_rate": 3.5573235460958145e-05,
+      "loss": 0.5204,
+      "step": 362
+    },
+    {
+      "epoch": 1.4731509625126646,
+      "grad_norm": 0.4302570838829243,
+      "learning_rate": 3.553755991230039e-05,
+      "loss": 0.5055,
+      "step": 363
+    },
+    {
+      "epoch": 1.4772036474164134,
+      "grad_norm": 0.5110468740171087,
+      "learning_rate": 3.5501759226271144e-05,
+      "loss": 0.5188,
+      "step": 364
+    },
+    {
+      "epoch": 1.4812563323201622,
+      "grad_norm": 0.4404063573793456,
+      "learning_rate": 3.546583369120419e-05,
+      "loss": 0.5254,
+      "step": 365
+    },
+    {
+      "epoch": 1.4853090172239107,
+      "grad_norm": 0.49985110591406146,
+      "learning_rate": 3.5429783596438864e-05,
+      "loss": 0.5182,
+      "step": 366
+    },
+    {
+      "epoch": 1.4893617021276595,
+      "grad_norm": 0.39012242982584083,
+      "learning_rate": 3.539360923231766e-05,
+      "loss": 0.5062,
+      "step": 367
+    },
+    {
+      "epoch": 1.4934143870314083,
+      "grad_norm": 0.5959979678203442,
+      "learning_rate": 3.535731089018394e-05,
+      "loss": 0.5112,
+      "step": 368
+    },
+    {
+      "epoch": 1.497467071935157,
+      "grad_norm": 0.42843565077267814,
+      "learning_rate": 3.532088886237956e-05,
+      "loss": 0.5257,
+      "step": 369
+    },
+    {
+      "epoch": 1.5015197568389058,
+      "grad_norm": 0.550645301428324,
+      "learning_rate": 3.528434344224253e-05,
+      "loss": 0.5306,
+      "step": 370
+    },
+    {
+      "epoch": 1.5055724417426544,
+      "grad_norm": 0.5091182750265051,
+      "learning_rate": 3.524767492410464e-05,
+      "loss": 0.5125,
+      "step": 371
+    },
+    {
+      "epoch": 1.5096251266464034,
+      "grad_norm": 0.374279794609297,
+      "learning_rate": 3.521088360328908e-05,
+      "loss": 0.5122,
+      "step": 372
+    },
+    {
+      "epoch": 1.513677811550152,
+      "grad_norm": 0.5275170206560594,
+      "learning_rate": 3.517396977610811e-05,
+      "loss": 0.5007,
+      "step": 373
+    },
+    {
+      "epoch": 1.5177304964539007,
+      "grad_norm": 0.46321340058218863,
+      "learning_rate": 3.5136933739860595e-05,
+      "loss": 0.5172,
+      "step": 374
+    },
+    {
+      "epoch": 1.5217831813576495,
+      "grad_norm": 0.47124638484942266,
+      "learning_rate": 3.509977579282971e-05,
+      "loss": 0.5178,
+      "step": 375
+    },
+    {
+      "epoch": 1.5258358662613982,
+      "grad_norm": 0.4400460060669203,
+      "learning_rate": 3.5062496234280424e-05,
+      "loss": 0.5095,
+      "step": 376
+    },
+    {
+      "epoch": 1.529888551165147,
+      "grad_norm": 0.44787783607543513,
+      "learning_rate": 3.502509536445719e-05,
+      "loss": 0.4985,
+      "step": 377
+    },
+    {
+      "epoch": 1.5339412360688955,
+      "grad_norm": 0.45270159262455223,
+      "learning_rate": 3.498757348458147e-05,
+      "loss": 0.5065,
+      "step": 378
+    },
+    {
+      "epoch": 1.5379939209726445,
+      "grad_norm": 0.47544400455654506,
+      "learning_rate": 3.4949930896849324e-05,
+      "loss": 0.5083,
+      "step": 379
+    },
+    {
+      "epoch": 1.542046605876393,
+      "grad_norm": 0.4502265060682666,
+      "learning_rate": 3.491216790442899e-05,
+      "loss": 0.5189,
+      "step": 380
+    },
+    {
+      "epoch": 1.5460992907801419,
+      "grad_norm": 0.42386990203135555,
+      "learning_rate": 3.487428481145839e-05,
+      "loss": 0.5122,
+      "step": 381
+    },
+    {
+      "epoch": 1.5501519756838906,
+      "grad_norm": 0.3778941917367464,
+      "learning_rate": 3.483628192304278e-05,
+      "loss": 0.5087,
+      "step": 382
+    },
+    {
+      "epoch": 1.5542046605876392,
+      "grad_norm": 0.41910394900737297,
+      "learning_rate": 3.479815954525219e-05,
+      "loss": 0.5112,
+      "step": 383
+    },
+    {
+      "epoch": 1.5582573454913882,
+      "grad_norm": 0.35683601630209655,
+      "learning_rate": 3.475991798511899e-05,
+      "loss": 0.4946,
+      "step": 384
+    },
+    {
+      "epoch": 1.5623100303951367,
+      "grad_norm": 0.3754391276161083,
+      "learning_rate": 3.4721557550635464e-05,
+      "loss": 0.5099,
+      "step": 385
+    },
+    {
+      "epoch": 1.5663627152988855,
+      "grad_norm": 0.3630853363738009,
+      "learning_rate": 3.468307855075128e-05,
+      "loss": 0.5073,
+      "step": 386
+    },
+    {
+      "epoch": 1.5704154002026343,
+      "grad_norm": 0.39163727676025634,
+      "learning_rate": 3.4644481295371005e-05,
+      "loss": 0.4896,
+      "step": 387
+    },
+    {
+      "epoch": 1.574468085106383,
+      "grad_norm": 0.39962278173165977,
+      "learning_rate": 3.460576609535163e-05,
+      "loss": 0.5208,
+      "step": 388
+    },
+    {
+      "epoch": 1.5785207700101318,
+      "grad_norm": 0.3640182525208087,
+      "learning_rate": 3.456693326250006e-05,
+      "loss": 0.5193,
+      "step": 389
+    },
+    {
+      "epoch": 1.5825734549138804,
+      "grad_norm": 0.3871034964125718,
+      "learning_rate": 3.452798310957058e-05,
+      "loss": 0.5163,
+      "step": 390
+    },
+    {
+      "epoch": 1.5866261398176293,
+      "grad_norm": 0.4708321345242008,
+      "learning_rate": 3.4488915950262386e-05,
+      "loss": 0.5136,
+      "step": 391
+    },
+    {
+      "epoch": 1.590678824721378,
+      "grad_norm": 0.4262443509825757,
+      "learning_rate": 3.4449732099216985e-05,
+      "loss": 0.5019,
+      "step": 392
+    },
+    {
+      "epoch": 1.5947315096251267,
+      "grad_norm": 0.4960883493154585,
+      "learning_rate": 3.441043187201574e-05,
+      "loss": 0.4986,
+      "step": 393
+    },
+    {
+      "epoch": 1.5987841945288754,
+      "grad_norm": 0.46348185074641,
+      "learning_rate": 3.437101558517728e-05,
+      "loss": 0.515,
+      "step": 394
+    },
+    {
+      "epoch": 1.602836879432624,
+      "grad_norm": 0.45562489518813776,
+      "learning_rate": 3.433148355615496e-05,
+      "loss": 0.5181,
+      "step": 395
+    },
+    {
+      "epoch": 1.606889564336373,
+      "grad_norm": 0.517070227140482,
+      "learning_rate": 3.4291836103334294e-05,
+      "loss": 0.5144,
+      "step": 396
+    },
+    {
+      "epoch": 1.6109422492401215,
+      "grad_norm": 0.3736507878674384,
+      "learning_rate": 3.425207354603043e-05,
+      "loss": 0.5117,
+      "step": 397
+    },
+    {
+      "epoch": 1.6149949341438703,
+      "grad_norm": 0.48526320851915467,
+      "learning_rate": 3.421219620448553e-05,
+      "loss": 0.4929,
+      "step": 398
+    },
+    {
+      "epoch": 1.619047619047619,
+      "grad_norm": 0.5179302101710983,
+      "learning_rate": 3.417220439986623e-05,
+      "loss": 0.5075,
+      "step": 399
+    },
+    {
+      "epoch": 1.6231003039513676,
+      "grad_norm": 0.349021646962347,
+      "learning_rate": 3.4132098454261024e-05,
+      "loss": 0.5243,
+      "step": 400
+    },
+    {
+      "epoch": 1.6271529888551166,
+      "grad_norm": 0.44217883346559617,
+      "learning_rate": 3.4091878690677676e-05,
+      "loss": 0.5013,
+      "step": 401
+    },
+    {
+      "epoch": 1.6312056737588652,
+      "grad_norm": 0.4527982424641868,
+      "learning_rate": 3.405154543304065e-05,
+      "loss": 0.5321,
+      "step": 402
+    },
+    {
+      "epoch": 1.635258358662614,
+      "grad_norm": 0.4129390805088078,
+      "learning_rate": 3.401109900618843e-05,
+      "loss": 0.5207,
+      "step": 403
+    },
+    {
+      "epoch": 1.6393110435663627,
+      "grad_norm": 0.3980799190471002,
+      "learning_rate": 3.3970539735870996e-05,
+      "loss": 0.5049,
+      "step": 404
+    },
+    {
+      "epoch": 1.6433637284701115,
+      "grad_norm": 0.4908845598595389,
+      "learning_rate": 3.392986794874714e-05,
+      "loss": 0.5162,
+      "step": 405
+    },
+    {
+      "epoch": 1.6474164133738602,
+      "grad_norm": 0.43055821311817066,
+      "learning_rate": 3.388908397238184e-05,
+      "loss": 0.516,
+      "step": 406
+    },
+    {
+      "epoch": 1.6514690982776088,
+      "grad_norm": 0.37208979604796427,
+      "learning_rate": 3.384818813524362e-05,
+      "loss": 0.5052,
+      "step": 407
+    },
+    {
+      "epoch": 1.6555217831813578,
+      "grad_norm": 0.4802973786274369,
+      "learning_rate": 3.380718076670195e-05,
+      "loss": 0.5155,
+      "step": 408
+    },
+    {
+      "epoch": 1.6595744680851063,
+      "grad_norm": 0.41314622362844466,
+      "learning_rate": 3.376606219702454e-05,
+      "loss": 0.5154,
+      "step": 409
+    },
+    {
+      "epoch": 1.663627152988855,
+      "grad_norm": 0.3771455696682645,
+      "learning_rate": 3.372483275737468e-05,
+      "loss": 0.5211,
+      "step": 410
+    },
+    {
+      "epoch": 1.6676798378926039,
+      "grad_norm": 0.477617651507032,
+      "learning_rate": 3.368349277980861e-05,
+      "loss": 0.511,
+      "step": 411
+    },
+    {
+      "epoch": 1.6717325227963524,
+      "grad_norm": 0.45993886520684657,
+      "learning_rate": 3.3642042597272844e-05,
+      "loss": 0.5165,
+      "step": 412
+    },
+    {
+      "epoch": 1.6757852077001014,
+      "grad_norm": 0.376511524884432,
+      "learning_rate": 3.360048254360144e-05,
+      "loss": 0.5094,
+      "step": 413
+    },
+    {
+      "epoch": 1.67983789260385,
+      "grad_norm": 0.48774517907104364,
+      "learning_rate": 3.355881295351336e-05,
+      "loss": 0.5231,
+      "step": 414
+    },
+    {
+      "epoch": 1.6838905775075987,
+      "grad_norm": 0.5108002410323854,
+      "learning_rate": 3.351703416260975e-05,
+      "loss": 0.4992,
+      "step": 415
+    },
+    {
+      "epoch": 1.6879432624113475,
+      "grad_norm": 0.3770227765274486,
+      "learning_rate": 3.347514650737126e-05,
+      "loss": 0.507,
+      "step": 416
+    },
+    {
+      "epoch": 1.6919959473150963,
+      "grad_norm": 0.581922092754776,
+      "learning_rate": 3.3433150325155295e-05,
+      "loss": 0.4966,
+      "step": 417
+    },
+    {
+      "epoch": 1.696048632218845,
+      "grad_norm": 0.40256271476525934,
+      "learning_rate": 3.339104595419334e-05,
+      "loss": 0.5087,
+      "step": 418
+    },
+    {
+      "epoch": 1.7001013171225936,
+      "grad_norm": 0.40834122655581984,
+      "learning_rate": 3.3348833733588204e-05,
+      "loss": 0.5246,
+      "step": 419
+    },
+    {
+      "epoch": 1.7041540020263426,
+      "grad_norm": 0.47183102353395956,
+      "learning_rate": 3.3306514003311305e-05,
+      "loss": 0.5262,
+      "step": 420
+    },
+    {
+      "epoch": 1.7082066869300911,
+      "grad_norm": 0.3836566756031701,
+      "learning_rate": 3.326408710419996e-05,
+      "loss": 0.5106,
+      "step": 421
+    },
+    {
+      "epoch": 1.71225937183384,
+      "grad_norm": 0.4335688059504242,
+      "learning_rate": 3.322155337795454e-05,
+      "loss": 0.5092,
+      "step": 422
+    },
+    {
+      "epoch": 1.7163120567375887,
+      "grad_norm": 0.5614115497390675,
+      "learning_rate": 3.317891316713587e-05,
+      "loss": 0.5093,
+      "step": 423
+    },
+    {
+      "epoch": 1.7203647416413372,
+      "grad_norm": 0.4133951074959847,
+      "learning_rate": 3.313616681516231e-05,
+      "loss": 0.5011,
+      "step": 424
+    },
+    {
+      "epoch": 1.7244174265450862,
+      "grad_norm": 0.3637814659771627,
+      "learning_rate": 3.309331466630713e-05,
+      "loss": 0.5018,
+      "step": 425
+    },
+    {
+      "epoch": 1.7284701114488348,
+      "grad_norm": 0.46197070614050006,
+      "learning_rate": 3.305035706569563e-05,
+      "loss": 0.5135,
+      "step": 426
+    },
+    {
+      "epoch": 1.7325227963525835,
+      "grad_norm": 0.3976842330233229,
+      "learning_rate": 3.3007294359302433e-05,
+      "loss": 0.5204,
+      "step": 427
+    },
+    {
+      "epoch": 1.7365754812563323,
+      "grad_norm": 0.3778469305830394,
+      "learning_rate": 3.296412689394864e-05,
+      "loss": 0.5056,
+      "step": 428
+    },
+    {
+      "epoch": 1.740628166160081,
+      "grad_norm": 0.5189439001834191,
+      "learning_rate": 3.292085501729909e-05,
+      "loss": 0.505,
+      "step": 429
+    },
+    {
+      "epoch": 1.7446808510638299,
+      "grad_norm": 0.3599803567954272,
+      "learning_rate": 3.2877479077859534e-05,
+      "loss": 0.51,
+      "step": 430
+    },
+    {
+      "epoch": 1.7487335359675784,
+      "grad_norm": 0.5131805580404335,
+      "learning_rate": 3.283399942497381e-05,
+      "loss": 0.5228,
+      "step": 431
+    },
+    {
+      "epoch": 1.7527862208713274,
+      "grad_norm": 0.42081440290177197,
+      "learning_rate": 3.279041640882108e-05,
+      "loss": 0.5172,
+      "step": 432
+    },
+    {
+      "epoch": 1.756838905775076,
+      "grad_norm": 0.360106585548781,
+      "learning_rate": 3.2746730380412964e-05,
+      "loss": 0.5098,
+      "step": 433
+    },
+    {
+      "epoch": 1.7608915906788247,
+      "grad_norm": 0.4092394195612383,
+      "learning_rate": 3.2702941691590726e-05,
+      "loss": 0.5204,
+      "step": 434
+    },
+    {
+      "epoch": 1.7649442755825735,
+      "grad_norm": 0.42537013613511365,
+      "learning_rate": 3.265905069502244e-05,
+      "loss": 0.5088,
+      "step": 435
+    },
+    {
+      "epoch": 1.768996960486322,
+      "grad_norm": 0.4074066455580255,
+      "learning_rate": 3.261505774420016e-05,
+      "loss": 0.5049,
+      "step": 436
+    },
+    {
+      "epoch": 1.773049645390071,
+      "grad_norm": 0.43617451342870184,
+      "learning_rate": 3.257096319343707e-05,
+      "loss": 0.5085,
+      "step": 437
+    },
+    {
+      "epoch": 1.7771023302938196,
+      "grad_norm": 0.4001265046086003,
+      "learning_rate": 3.2526767397864614e-05,
+      "loss": 0.5138,
+      "step": 438
+    },
+    {
+      "epoch": 1.7811550151975684,
+      "grad_norm": 0.36978425541297655,
+      "learning_rate": 3.248247071342966e-05,
+      "loss": 0.5028,
+      "step": 439
+    },
+    {
+      "epoch": 1.7852077001013171,
+      "grad_norm": 0.44176015129427154,
+      "learning_rate": 3.243807349689161e-05,
+      "loss": 0.4952,
+      "step": 440
+    },
+    {
+      "epoch": 1.789260385005066,
+      "grad_norm": 0.30661975612240344,
+      "learning_rate": 3.2393576105819544e-05,
+      "loss": 0.5034,
+      "step": 441
+    },
+    {
+      "epoch": 1.7933130699088147,
+      "grad_norm": 0.4584117550853115,
+      "learning_rate": 3.2348978898589333e-05,
+      "loss": 0.5005,
+      "step": 442
+    },
+    {
+      "epoch": 1.7973657548125632,
+      "grad_norm": 0.35478248501504045,
+      "learning_rate": 3.230428223438075e-05,
+      "loss": 0.507,
+      "step": 443
+    },
+    {
+      "epoch": 1.8014184397163122,
+      "grad_norm": 0.41396256080427335,
+      "learning_rate": 3.225948647317459e-05,
+      "loss": 0.5125,
+      "step": 444
+    },
+    {
+      "epoch": 1.8054711246200608,
+      "grad_norm": 0.47181320797981524,
+      "learning_rate": 3.2214591975749745e-05,
+      "loss": 0.5101,
+      "step": 445
+    },
+    {
+      "epoch": 1.8095238095238095,
+      "grad_norm": 0.4122322160330429,
+      "learning_rate": 3.216959910368034e-05,
+      "loss": 0.4982,
+      "step": 446
+    },
+    {
+      "epoch": 1.8135764944275583,
+      "grad_norm": 0.5691128851101503,
+      "learning_rate": 3.212450821933277e-05,
+      "loss": 0.5004,
+      "step": 447
+    },
+    {
+      "epoch": 1.8176291793313069,
+      "grad_norm": 0.4564192474589319,
+      "learning_rate": 3.207931968586281e-05,
+      "loss": 0.5007,
+      "step": 448
+    },
+    {
+      "epoch": 1.8216818642350558,
+      "grad_norm": 0.5985012738884663,
+      "learning_rate": 3.203403386721272e-05,
+      "loss": 0.5056,
+      "step": 449
+    },
+    {
+      "epoch": 1.8257345491388044,
+      "grad_norm": 0.5291397435563098,
+      "learning_rate": 3.1988651128108245e-05,
+      "loss": 0.5112,
+      "step": 450
+    },
+    {
+      "epoch": 1.8297872340425532,
+      "grad_norm": 0.5046191799902497,
+      "learning_rate": 3.194317183405573e-05,
+      "loss": 0.5317,
+      "step": 451
+    },
+    {
+      "epoch": 1.833839918946302,
+      "grad_norm": 0.5715407202210351,
+      "learning_rate": 3.189759635133914e-05,
+      "loss": 0.5083,
+      "step": 452
+    },
+    {
+      "epoch": 1.8378926038500507,
+      "grad_norm": 0.4680357631599358,
+      "learning_rate": 3.185192504701718e-05,
+      "loss": 0.5033,
+      "step": 453
+    },
+    {
+      "epoch": 1.8419452887537995,
+      "grad_norm": 0.4448820935643253,
+      "learning_rate": 3.1806158288920234e-05,
+      "loss": 0.4861,
+      "step": 454
+    },
+    {
+      "epoch": 1.845997973657548,
+      "grad_norm": 0.45074109091554493,
+      "learning_rate": 3.1760296445647477e-05,
+      "loss": 0.505,
+      "step": 455
+    },
+    {
+      "epoch": 1.850050658561297,
+      "grad_norm": 0.4379497213059703,
+      "learning_rate": 3.1714339886563896e-05,
+      "loss": 0.4982,
+      "step": 456
+    },
+    {
+      "epoch": 1.8541033434650456,
+      "grad_norm": 0.4678026548600231,
+      "learning_rate": 3.166828898179731e-05,
+      "loss": 0.4903,
+      "step": 457
+    },
+    {
+      "epoch": 1.8581560283687943,
+      "grad_norm": 0.359024494586788,
+      "learning_rate": 3.162214410223536e-05,
+      "loss": 0.5138,
+      "step": 458
+    },
+    {
+      "epoch": 1.8622087132725431,
+      "grad_norm": 0.4542815410644416,
+      "learning_rate": 3.157590561952257e-05,
+      "loss": 0.5207,
+      "step": 459
+    },
+    {
+      "epoch": 1.8662613981762917,
+      "grad_norm": 0.3808222897699044,
+      "learning_rate": 3.152957390605732e-05,
+      "loss": 0.5077,
+      "step": 460
+    },
+    {
+      "epoch": 1.8703140830800407,
+      "grad_norm": 0.3966850415278545,
+      "learning_rate": 3.148314933498886e-05,
+      "loss": 0.5025,
+      "step": 461
+    },
+    {
+      "epoch": 1.8743667679837892,
+      "grad_norm": 0.4227418314994302,
+      "learning_rate": 3.143663228021431e-05,
+      "loss": 0.52,
+      "step": 462
+    },
+    {
+      "epoch": 1.878419452887538,
+      "grad_norm": 0.36865592472772585,
+      "learning_rate": 3.1390023116375624e-05,
+      "loss": 0.5094,
+      "step": 463
+    },
+    {
+      "epoch": 1.8824721377912867,
+      "grad_norm": 0.4230086571571748,
+      "learning_rate": 3.134332221885661e-05,
+      "loss": 0.5069,
+      "step": 464
+    },
+    {
+      "epoch": 1.8865248226950353,
+      "grad_norm": 0.4865376888736305,
+      "learning_rate": 3.129652996377987e-05,
+      "loss": 0.506,
+      "step": 465
+    },
+    {
+      "epoch": 1.8905775075987843,
+      "grad_norm": 0.3586565691382082,
+      "learning_rate": 3.12496467280038e-05,
+      "loss": 0.503,
+      "step": 466
+    },
+    {
+      "epoch": 1.8946301925025328,
+      "grad_norm": 0.5173100489627686,
+      "learning_rate": 3.120267288911952e-05,
+      "loss": 0.4963,
+      "step": 467
+    },
+    {
+      "epoch": 1.8986828774062816,
+      "grad_norm": 0.3894794386148582,
+      "learning_rate": 3.11556088254479e-05,
+      "loss": 0.5081,
+      "step": 468
+    },
+    {
+      "epoch": 1.9027355623100304,
+      "grad_norm": 0.434087189230862,
+      "learning_rate": 3.11084549160364e-05,
+      "loss": 0.4978,
+      "step": 469
+    },
+    {
+      "epoch": 1.9067882472137792,
+      "grad_norm": 0.43540656327153193,
+      "learning_rate": 3.106121154065615e-05,
+      "loss": 0.5135,
+      "step": 470
+    },
+    {
+      "epoch": 1.910840932117528,
+      "grad_norm": 0.38835530867502965,
+      "learning_rate": 3.1013879079798805e-05,
+      "loss": 0.5064,
+      "step": 471
+    },
+    {
+      "epoch": 1.9148936170212765,
+      "grad_norm": 0.43364286096721644,
+      "learning_rate": 3.096645791467348e-05,
+      "loss": 0.5017,
+      "step": 472
+    },
+    {
+      "epoch": 1.9189463019250255,
+      "grad_norm": 0.46412885633388007,
+      "learning_rate": 3.091894842720373e-05,
+      "loss": 0.5105,
+      "step": 473
+    },
+    {
+      "epoch": 1.922998986828774,
+      "grad_norm": 0.327685361404935,
+      "learning_rate": 3.0871351000024425e-05,
+      "loss": 0.4937,
+      "step": 474
+    },
+    {
+      "epoch": 1.9270516717325228,
+      "grad_norm": 0.4003732793389209,
+      "learning_rate": 3.0823666016478716e-05,
+      "loss": 0.5026,
+      "step": 475
+    },
+    {
+      "epoch": 1.9311043566362716,
+      "grad_norm": 0.31030464775365824,
+      "learning_rate": 3.0775893860614896e-05,
+      "loss": 0.487,
+      "step": 476
+    },
+    {
+      "epoch": 1.93515704154002,
+      "grad_norm": 0.37844837216061555,
+      "learning_rate": 3.0728034917183336e-05,
+      "loss": 0.4943,
+      "step": 477
+    },
+    {
+      "epoch": 1.939209726443769,
+      "grad_norm": 0.3761041097368198,
+      "learning_rate": 3.06800895716334e-05,
+      "loss": 0.515,
+      "step": 478
+    },
+    {
+      "epoch": 1.9432624113475176,
+      "grad_norm": 0.34105499333301525,
+      "learning_rate": 3.063205821011029e-05,
+      "loss": 0.5022,
+      "step": 479
+    },
+    {
+      "epoch": 1.9473150962512664,
+      "grad_norm": 0.3434204163494274,
+      "learning_rate": 3.0583941219452016e-05,
+      "loss": 0.5059,
+      "step": 480
+    },
+    {
+      "epoch": 1.9513677811550152,
+      "grad_norm": 0.342798220622963,
+      "learning_rate": 3.053573898718618e-05,
+      "loss": 0.5161,
+      "step": 481
+    },
+    {
+      "epoch": 1.955420466058764,
+      "grad_norm": 0.38622806206243165,
+      "learning_rate": 3.0487451901526956e-05,
+      "loss": 0.5219,
+      "step": 482
+    },
+    {
+      "epoch": 1.9594731509625127,
+      "grad_norm": 0.4246240951222412,
+      "learning_rate": 3.0439080351371875e-05,
+      "loss": 0.5032,
+      "step": 483
+    },
+    {
+      "epoch": 1.9635258358662613,
+      "grad_norm": 0.3556257540680832,
+      "learning_rate": 3.0390624726298764e-05,
+      "loss": 0.4864,
+      "step": 484
+    },
+    {
+      "epoch": 1.9675785207700103,
+      "grad_norm": 0.36868876075038537,
+      "learning_rate": 3.034208541656255e-05,
+      "loss": 0.5208,
+      "step": 485
+    },
+    {
+      "epoch": 1.9716312056737588,
+      "grad_norm": 0.3610588748401692,
+      "learning_rate": 3.029346281309218e-05,
+      "loss": 0.5041,
+      "step": 486
+    },
+    {
+      "epoch": 1.9756838905775076,
+      "grad_norm": 0.34937827685673983,
+      "learning_rate": 3.0244757307487415e-05,
+      "loss": 0.4913,
+      "step": 487
+    },
+    {
+      "epoch": 1.9797365754812564,
+      "grad_norm": 0.3477030210550178,
+      "learning_rate": 3.019596929201569e-05,
+      "loss": 0.5064,
+      "step": 488
+    },
+    {
+      "epoch": 1.983789260385005,
+      "grad_norm": 0.4227474125135131,
+      "learning_rate": 3.0147099159608985e-05,
+      "loss": 0.522,
+      "step": 489
+    },
+    {
+      "epoch": 1.987841945288754,
+      "grad_norm": 0.3094207511546184,
+      "learning_rate": 3.0098147303860616e-05,
+      "loss": 0.505,
+      "step": 490
+    },
+    {
+      "epoch": 1.9918946301925025,
+      "grad_norm": 0.4008816235367959,
+      "learning_rate": 3.0049114119022117e-05,
+      "loss": 0.4977,
+      "step": 491
+    },
+    {
+      "epoch": 1.9959473150962512,
+      "grad_norm": 0.3553228837321045,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.503,
+      "step": 492
+    },
+    {
+      "epoch": 2.0020263424518743,
+      "grad_norm": 0.3957562871432883,
+      "learning_rate": 2.995080534235264e-05,
+      "loss": 0.47,
+      "step": 493
+    },
+    {
+      "epoch": 2.0060790273556233,
+      "grad_norm": 0.43386265612843367,
+      "learning_rate": 2.9901530542287044e-05,
+      "loss": 0.4379,
+      "step": 494
+    },
+    {
+      "epoch": 2.010131712259372,
+      "grad_norm": 0.6085637293603964,
+      "learning_rate": 2.9852175996655676e-05,
+      "loss": 0.4398,
+      "step": 495
+    },
+    {
+      "epoch": 2.0141843971631204,
+      "grad_norm": 0.4898735295242637,
+      "learning_rate": 2.980274210295326e-05,
+      "loss": 0.4434,
+      "step": 496
+    },
+    {
+      "epoch": 2.0182370820668694,
+      "grad_norm": 0.5061533783363372,
+      "learning_rate": 2.9753229259313578e-05,
+      "loss": 0.4414,
+      "step": 497
+    },
+    {
+      "epoch": 2.022289766970618,
+      "grad_norm": 0.4269512340349817,
+      "learning_rate": 2.9703637864506274e-05,
+      "loss": 0.4308,
+      "step": 498
+    },
+    {
+      "epoch": 2.026342451874367,
+      "grad_norm": 0.41223315654197756,
+      "learning_rate": 2.965396831793362e-05,
+      "loss": 0.4355,
+      "step": 499
+    },
+    {
+      "epoch": 2.0303951367781155,
+      "grad_norm": 0.4540618269403147,
+      "learning_rate": 2.9604221019627316e-05,
+      "loss": 0.4387,
+      "step": 500
+    },
+    {
+      "epoch": 2.0344478216818644,
+      "grad_norm": 0.43871473472012845,
+      "learning_rate": 2.955439637024526e-05,
+      "loss": 0.4248,
+      "step": 501
+    },
+    {
+      "epoch": 2.038500506585613,
+      "grad_norm": 0.4505008021598321,
+      "learning_rate": 2.9504494771068334e-05,
+      "loss": 0.4446,
+      "step": 502
+    },
+    {
+      "epoch": 2.0425531914893615,
+      "grad_norm": 0.4172192649300292,
+      "learning_rate": 2.9454516623997156e-05,
+      "loss": 0.4302,
+      "step": 503
+    },
+    {
+      "epoch": 2.0466058763931105,
+      "grad_norm": 0.40146772547336085,
+      "learning_rate": 2.9404462331548847e-05,
+      "loss": 0.4439,
+      "step": 504
+    },
+    {
+      "epoch": 2.050658561296859,
+      "grad_norm": 0.39888178150287157,
+      "learning_rate": 2.93543322968538e-05,
+      "loss": 0.4343,
+      "step": 505
+    },
+    {
+      "epoch": 2.054711246200608,
+      "grad_norm": 0.4010324290539672,
+      "learning_rate": 2.9304126923652428e-05,
+      "loss": 0.4338,
+      "step": 506
+    },
+    {
+      "epoch": 2.0587639311043566,
+      "grad_norm": 0.3506868515718636,
+      "learning_rate": 2.9253846616291896e-05,
+      "loss": 0.436,
+      "step": 507
+    },
+    {
+      "epoch": 2.062816616008105,
+      "grad_norm": 0.4129525003336646,
+      "learning_rate": 2.9203491779722896e-05,
+      "loss": 0.4218,
+      "step": 508
+    },
+    {
+      "epoch": 2.066869300911854,
+      "grad_norm": 0.353920518082642,
+      "learning_rate": 2.9153062819496357e-05,
+      "loss": 0.438,
+      "step": 509
+    },
+    {
+      "epoch": 2.0709219858156027,
+      "grad_norm": 0.41031765540668913,
+      "learning_rate": 2.9102560141760178e-05,
+      "loss": 0.4474,
+      "step": 510
+    },
+    {
+      "epoch": 2.0749746707193517,
+      "grad_norm": 0.3560449349411888,
+      "learning_rate": 2.9051984153256004e-05,
+      "loss": 0.4474,
+      "step": 511
+    },
+    {
+      "epoch": 2.0790273556231003,
+      "grad_norm": 0.3737734389564037,
+      "learning_rate": 2.900133526131588e-05,
+      "loss": 0.4438,
+      "step": 512
+    },
+    {
+      "epoch": 2.0830800405268493,
+      "grad_norm": 0.3905759365363995,
+      "learning_rate": 2.8950613873859025e-05,
+      "loss": 0.4455,
+      "step": 513
+    },
+    {
+      "epoch": 2.087132725430598,
+      "grad_norm": 0.3586068139248692,
+      "learning_rate": 2.8899820399388515e-05,
+      "loss": 0.4246,
+      "step": 514
+    },
+    {
+      "epoch": 2.0911854103343464,
+      "grad_norm": 0.3939403685352425,
+      "learning_rate": 2.8848955246988012e-05,
+      "loss": 0.4467,
+      "step": 515
+    },
+    {
+      "epoch": 2.0952380952380953,
+      "grad_norm": 0.38268006746299166,
+      "learning_rate": 2.879801882631847e-05,
+      "loss": 0.4326,
+      "step": 516
+    },
+    {
+      "epoch": 2.099290780141844,
+      "grad_norm": 0.3976079893542882,
+      "learning_rate": 2.8747011547614808e-05,
+      "loss": 0.4519,
+      "step": 517
+    },
+    {
+      "epoch": 2.103343465045593,
+      "grad_norm": 0.4129467361960651,
+      "learning_rate": 2.8695933821682635e-05,
+      "loss": 0.4466,
+      "step": 518
+    },
+    {
+      "epoch": 2.1073961499493414,
+      "grad_norm": 0.4394261124473519,
+      "learning_rate": 2.864478605989494e-05,
+      "loss": 0.44,
+      "step": 519
+    },
+    {
+      "epoch": 2.11144883485309,
+      "grad_norm": 0.47510373003059053,
+      "learning_rate": 2.8593568674188765e-05,
+      "loss": 0.4388,
+      "step": 520
+    },
+    {
+      "epoch": 2.115501519756839,
+      "grad_norm": 0.422149260139249,
+      "learning_rate": 2.8542282077061892e-05,
+      "loss": 0.4411,
+      "step": 521
+    },
+    {
+      "epoch": 2.1195542046605875,
+      "grad_norm": 0.4490399135688369,
+      "learning_rate": 2.8490926681569523e-05,
+      "loss": 0.4471,
+      "step": 522
+    },
+    {
+      "epoch": 2.1236068895643365,
+      "grad_norm": 0.48237993127512185,
+      "learning_rate": 2.8439502901320956e-05,
+      "loss": 0.4446,
+      "step": 523
+    },
+    {
+      "epoch": 2.127659574468085,
+      "grad_norm": 0.3726192184767776,
+      "learning_rate": 2.8388011150476237e-05,
+      "loss": 0.4444,
+      "step": 524
+    },
+    {
+      "epoch": 2.1317122593718336,
+      "grad_norm": 0.44225977855558796,
+      "learning_rate": 2.8336451843742866e-05,
+      "loss": 0.4516,
+      "step": 525
+    },
+    {
+      "epoch": 2.1357649442755826,
+      "grad_norm": 0.3318267644956898,
+      "learning_rate": 2.8284825396372387e-05,
+      "loss": 0.448,
+      "step": 526
+    },
+    {
+      "epoch": 2.139817629179331,
+      "grad_norm": 0.4167815557160867,
+      "learning_rate": 2.8233132224157132e-05,
+      "loss": 0.4404,
+      "step": 527
+    },
+    {
+      "epoch": 2.14387031408308,
+      "grad_norm": 0.399117046215451,
+      "learning_rate": 2.8181372743426805e-05,
+      "loss": 0.4412,
+      "step": 528
+    },
+    {
+      "epoch": 2.1479229989868287,
+      "grad_norm": 0.42996619069159747,
+      "learning_rate": 2.8129547371045128e-05,
+      "loss": 0.4457,
+      "step": 529
+    },
+    {
+      "epoch": 2.1519756838905777,
+      "grad_norm": 0.40658830316335864,
+      "learning_rate": 2.8077656524406534e-05,
+      "loss": 0.4475,
+      "step": 530
+    },
+    {
+      "epoch": 2.1560283687943262,
+      "grad_norm": 0.3900252079960181,
+      "learning_rate": 2.802570062143278e-05,
+      "loss": 0.446,
+      "step": 531
+    },
+    {
+      "epoch": 2.160081053698075,
+      "grad_norm": 0.3457040130531706,
+      "learning_rate": 2.7973680080569555e-05,
+      "loss": 0.4524,
+      "step": 532
+    },
+    {
+      "epoch": 2.164133738601824,
+      "grad_norm": 0.31523937089503834,
+      "learning_rate": 2.792159532078314e-05,
+      "loss": 0.436,
+      "step": 533
+    },
+    {
+      "epoch": 2.1681864235055723,
+      "grad_norm": 0.349969402368419,
+      "learning_rate": 2.7869446761557033e-05,
+      "loss": 0.439,
+      "step": 534
+    },
+    {
+      "epoch": 2.1722391084093213,
+      "grad_norm": 0.35386101101734324,
+      "learning_rate": 2.781723482288857e-05,
+      "loss": 0.4391,
+      "step": 535
+    },
+    {
+      "epoch": 2.17629179331307,
+      "grad_norm": 0.355684435443336,
+      "learning_rate": 2.7764959925285517e-05,
+      "loss": 0.4276,
+      "step": 536
+    },
+    {
+      "epoch": 2.180344478216819,
+      "grad_norm": 0.3370386697794104,
+      "learning_rate": 2.771262248976272e-05,
+      "loss": 0.4387,
+      "step": 537
+    },
+    {
+      "epoch": 2.1843971631205674,
+      "grad_norm": 0.3227019948931399,
+      "learning_rate": 2.7660222937838677e-05,
+      "loss": 0.4414,
+      "step": 538
+    },
+    {
+      "epoch": 2.188449848024316,
+      "grad_norm": 0.37100772248145486,
+      "learning_rate": 2.7607761691532186e-05,
+      "loss": 0.464,
+      "step": 539
+    },
+    {
+      "epoch": 2.192502532928065,
+      "grad_norm": 0.34920108223892243,
+      "learning_rate": 2.7555239173358916e-05,
+      "loss": 0.445,
+      "step": 540
+    },
+    {
+      "epoch": 2.1965552178318135,
+      "grad_norm": 0.33438942817637946,
+      "learning_rate": 2.7502655806328e-05,
+      "loss": 0.4346,
+      "step": 541
+    },
+    {
+      "epoch": 2.2006079027355625,
+      "grad_norm": 0.338233034355839,
+      "learning_rate": 2.7450012013938648e-05,
+      "loss": 0.4546,
+      "step": 542
+    },
+    {
+      "epoch": 2.204660587639311,
+      "grad_norm": 0.3149971370650695,
+      "learning_rate": 2.739730822017673e-05,
+      "loss": 0.4513,
+      "step": 543
+    },
+    {
+      "epoch": 2.2087132725430596,
+      "grad_norm": 0.30565439280200163,
+      "learning_rate": 2.7344544849511355e-05,
+      "loss": 0.4537,
+      "step": 544
+    },
+    {
+      "epoch": 2.2127659574468086,
+      "grad_norm": 0.36768349326667193,
+      "learning_rate": 2.7291722326891456e-05,
+      "loss": 0.4545,
+      "step": 545
+    },
+    {
+      "epoch": 2.216818642350557,
+      "grad_norm": 0.3701403437624965,
+      "learning_rate": 2.723884107774236e-05,
+      "loss": 0.4406,
+      "step": 546
+    },
+    {
+      "epoch": 2.220871327254306,
+      "grad_norm": 0.36215148140741227,
+      "learning_rate": 2.718590152796239e-05,
+      "loss": 0.4533,
+      "step": 547
+    },
+    {
+      "epoch": 2.2249240121580547,
+      "grad_norm": 0.4130935454727938,
+      "learning_rate": 2.71329041039194e-05,
+      "loss": 0.4377,
+      "step": 548
+    },
+    {
+      "epoch": 2.2289766970618032,
+      "grad_norm": 0.44096410498843186,
+      "learning_rate": 2.7079849232447357e-05,
+      "loss": 0.4503,
+      "step": 549
+    },
+    {
+      "epoch": 2.2330293819655522,
+      "grad_norm": 0.29901955869727204,
+      "learning_rate": 2.7026737340842895e-05,
+      "loss": 0.4478,
+      "step": 550
+    },
+    {
+      "epoch": 2.237082066869301,
+      "grad_norm": 0.43729885851108696,
+      "learning_rate": 2.697356885686189e-05,
+      "loss": 0.4472,
+      "step": 551
+    },
+    {
+      "epoch": 2.2411347517730498,
+      "grad_norm": 0.3935974532733391,
+      "learning_rate": 2.6920344208716014e-05,
+      "loss": 0.4488,
+      "step": 552
+    },
+    {
+      "epoch": 2.2451874366767983,
+      "grad_norm": 0.3613505801042168,
+      "learning_rate": 2.6867063825069252e-05,
+      "loss": 0.4411,
+      "step": 553
+    },
+    {
+      "epoch": 2.2492401215805473,
+      "grad_norm": 0.35253110813727523,
+      "learning_rate": 2.6813728135034494e-05,
+      "loss": 0.4471,
+      "step": 554
+    },
+    {
+      "epoch": 2.253292806484296,
+      "grad_norm": 0.3726277670303869,
+      "learning_rate": 2.6760337568170056e-05,
+      "loss": 0.4376,
+      "step": 555
+    },
+    {
+      "epoch": 2.2573454913880444,
+      "grad_norm": 0.3075152692363883,
+      "learning_rate": 2.6706892554476226e-05,
+      "loss": 0.4418,
+      "step": 556
+    },
+    {
+      "epoch": 2.2613981762917934,
+      "grad_norm": 0.46525698386528336,
+      "learning_rate": 2.6653393524391795e-05,
+      "loss": 0.4431,
+      "step": 557
+    },
+    {
+      "epoch": 2.265450861195542,
+      "grad_norm": 0.30736152987971066,
+      "learning_rate": 2.6599840908790592e-05,
+      "loss": 0.449,
+      "step": 558
+    },
+    {
+      "epoch": 2.269503546099291,
+      "grad_norm": 0.44616953922452857,
+      "learning_rate": 2.6546235138978028e-05,
+      "loss": 0.4438,
+      "step": 559
+    },
+    {
+      "epoch": 2.2735562310030395,
+      "grad_norm": 0.31993291555988845,
+      "learning_rate": 2.6492576646687597e-05,
+      "loss": 0.4369,
+      "step": 560
+    },
+    {
+      "epoch": 2.2776089159067885,
+      "grad_norm": 0.43328981149113904,
+      "learning_rate": 2.6438865864077425e-05,
+      "loss": 0.4358,
+      "step": 561
+    },
+    {
+      "epoch": 2.281661600810537,
+      "grad_norm": 0.3703336479610273,
+      "learning_rate": 2.6385103223726766e-05,
+      "loss": 0.4587,
+      "step": 562
+    },
+    {
+      "epoch": 2.2857142857142856,
+      "grad_norm": 0.34035332176800803,
+      "learning_rate": 2.6331289158632537e-05,
+      "loss": 0.4543,
+      "step": 563
+    },
+    {
+      "epoch": 2.2897669706180346,
+      "grad_norm": 0.7337075215353476,
+      "learning_rate": 2.6277424102205817e-05,
+      "loss": 0.463,
+      "step": 564
+    },
+    {
+      "epoch": 2.293819655521783,
+      "grad_norm": 0.3725300488193857,
+      "learning_rate": 2.6223508488268374e-05,
+      "loss": 0.4289,
+      "step": 565
+    },
+    {
+      "epoch": 2.297872340425532,
+      "grad_norm": 0.3976758875332692,
+      "learning_rate": 2.6169542751049148e-05,
+      "loss": 0.4628,
+      "step": 566
+    },
+    {
+      "epoch": 2.3019250253292807,
+      "grad_norm": 0.5249985049781397,
+      "learning_rate": 2.6115527325180754e-05,
+      "loss": 0.4446,
+      "step": 567
+    },
+    {
+      "epoch": 2.305977710233029,
+      "grad_norm": 0.38854462570778775,
+      "learning_rate": 2.606146264569603e-05,
+      "loss": 0.4585,
+      "step": 568
+    },
+    {
+      "epoch": 2.310030395136778,
+      "grad_norm": 0.3637403502615504,
+      "learning_rate": 2.6007349148024447e-05,
+      "loss": 0.4336,
+      "step": 569
+    },
+    {
+      "epoch": 2.3140830800405268,
+      "grad_norm": 0.2952197897272316,
+      "learning_rate": 2.5953187267988694e-05,
+      "loss": 0.4409,
+      "step": 570
+    },
+    {
+      "epoch": 2.3181357649442758,
+      "grad_norm": 0.30950569598802063,
+      "learning_rate": 2.5898977441801097e-05,
+      "loss": 0.449,
+      "step": 571
+    },
+    {
+      "epoch": 2.3221884498480243,
+      "grad_norm": 0.280585872126244,
+      "learning_rate": 2.584472010606015e-05,
+      "loss": 0.4604,
+      "step": 572
+    },
+    {
+      "epoch": 2.326241134751773,
+      "grad_norm": 0.2882796447003011,
+      "learning_rate": 2.5790415697746976e-05,
+      "loss": 0.451,
+      "step": 573
+    },
+    {
+      "epoch": 2.330293819655522,
+      "grad_norm": 0.31414600750350985,
+      "learning_rate": 2.5736064654221808e-05,
+      "loss": 0.4374,
+      "step": 574
+    },
+    {
+      "epoch": 2.3343465045592704,
+      "grad_norm": 0.32454409259304257,
+      "learning_rate": 2.568166741322048e-05,
+      "loss": 0.4492,
+      "step": 575
+    },
+    {
+      "epoch": 2.3383991894630194,
+      "grad_norm": 0.37787661911337705,
+      "learning_rate": 2.56272244128509e-05,
+      "loss": 0.4149,
+      "step": 576
+    },
+    {
+      "epoch": 2.342451874366768,
+      "grad_norm": 0.29554876401561386,
+      "learning_rate": 2.55727360915895e-05,
+      "loss": 0.44,
+      "step": 577
+    },
+    {
+      "epoch": 2.3465045592705165,
+      "grad_norm": 0.3968092765570855,
+      "learning_rate": 2.5518202888277734e-05,
+      "loss": 0.4449,
+      "step": 578
+    },
+    {
+      "epoch": 2.3505572441742655,
+      "grad_norm": 0.32434456495073655,
+      "learning_rate": 2.5463625242118523e-05,
+      "loss": 0.4326,
+      "step": 579
+    },
+    {
+      "epoch": 2.354609929078014,
+      "grad_norm": 0.3659404647028591,
+      "learning_rate": 2.5409003592672723e-05,
+      "loss": 0.4337,
+      "step": 580
+    },
+    {
+      "epoch": 2.358662613981763,
+      "grad_norm": 0.32923180796301754,
+      "learning_rate": 2.535433837985559e-05,
+      "loss": 0.4541,
+      "step": 581
+    },
+    {
+      "epoch": 2.3627152988855116,
+      "grad_norm": 0.35806497270649884,
+      "learning_rate": 2.529963004393324e-05,
+      "loss": 0.4662,
+      "step": 582
+    },
+    {
+      "epoch": 2.3667679837892606,
+      "grad_norm": 0.33209025811679616,
+      "learning_rate": 2.524487902551908e-05,
+      "loss": 0.4544,
+      "step": 583
+    },
+    {
+      "epoch": 2.370820668693009,
+      "grad_norm": 0.3132716577974625,
+      "learning_rate": 2.519008576557029e-05,
+      "loss": 0.441,
+      "step": 584
+    },
+    {
+      "epoch": 2.3748733535967577,
+      "grad_norm": 0.3189992924541878,
+      "learning_rate": 2.5135250705384254e-05,
+      "loss": 0.444,
+      "step": 585
+    },
+    {
+      "epoch": 2.3789260385005067,
+      "grad_norm": 0.3010755766283583,
+      "learning_rate": 2.5080374286595007e-05,
+      "loss": 0.4431,
+      "step": 586
+    },
+    {
+      "epoch": 2.382978723404255,
+      "grad_norm": 0.29915708172104893,
+      "learning_rate": 2.5025456951169677e-05,
+      "loss": 0.4556,
+      "step": 587
+    },
+    {
+      "epoch": 2.387031408308004,
+      "grad_norm": 0.31512823751757624,
+      "learning_rate": 2.4970499141404942e-05,
+      "loss": 0.4452,
+      "step": 588
+    },
+    {
+      "epoch": 2.3910840932117527,
+      "grad_norm": 0.3008110070225229,
+      "learning_rate": 2.491550129992345e-05,
+      "loss": 0.4551,
+      "step": 589
+    },
+    {
+      "epoch": 2.3951367781155017,
+      "grad_norm": 0.30066378537557736,
+      "learning_rate": 2.486046386967024e-05,
+      "loss": 0.4527,
+      "step": 590
+    },
+    {
+      "epoch": 2.3991894630192503,
+      "grad_norm": 0.31900805950846545,
+      "learning_rate": 2.4805387293909214e-05,
+      "loss": 0.4453,
+      "step": 591
+    },
+    {
+      "epoch": 2.403242147922999,
+      "grad_norm": 0.32272666830661023,
+      "learning_rate": 2.4750272016219552e-05,
+      "loss": 0.4463,
+      "step": 592
+    },
+    {
+      "epoch": 2.407294832826748,
+      "grad_norm": 0.29742438696928303,
+      "learning_rate": 2.4695118480492114e-05,
+      "loss": 0.4394,
+      "step": 593
+    },
+    {
+      "epoch": 2.4113475177304964,
+      "grad_norm": 0.34476301038890383,
+      "learning_rate": 2.4639927130925898e-05,
+      "loss": 0.4355,
+      "step": 594
+    },
+    {
+      "epoch": 2.4154002026342454,
+      "grad_norm": 0.2871740574842531,
+      "learning_rate": 2.458469841202444e-05,
+      "loss": 0.4494,
+      "step": 595
+    },
+    {
+      "epoch": 2.419452887537994,
+      "grad_norm": 0.3655792431410782,
+      "learning_rate": 2.452943276859226e-05,
+      "loss": 0.4321,
+      "step": 596
+    },
+    {
+      "epoch": 2.4235055724417425,
+      "grad_norm": 0.3207807851871291,
+      "learning_rate": 2.447413064573125e-05,
+      "loss": 0.4602,
+      "step": 597
+    },
+    {
+      "epoch": 2.4275582573454915,
+      "grad_norm": 0.3694406663567429,
+      "learning_rate": 2.4418792488837095e-05,
+      "loss": 0.4499,
+      "step": 598
+    },
+    {
+      "epoch": 2.43161094224924,
+      "grad_norm": 0.36992645886717623,
+      "learning_rate": 2.4363418743595713e-05,
+      "loss": 0.4364,
+      "step": 599
+    },
+    {
+      "epoch": 2.435663627152989,
+      "grad_norm": 0.33592862389406797,
+      "learning_rate": 2.430800985597963e-05,
+      "loss": 0.4466,
+      "step": 600
+    },
+    {
+      "epoch": 2.4397163120567376,
+      "grad_norm": 0.3647960976857997,
+      "learning_rate": 2.4252566272244415e-05,
+      "loss": 0.444,
+      "step": 601
+    },
+    {
+      "epoch": 2.443768996960486,
+      "grad_norm": 0.2979463577093382,
+      "learning_rate": 2.4197088438925063e-05,
+      "loss": 0.4498,
+      "step": 602
+    },
+    {
+      "epoch": 2.447821681864235,
+      "grad_norm": 1.4707392002836646,
+      "learning_rate": 2.4141576802832417e-05,
+      "loss": 0.5017,
+      "step": 603
+    },
+    {
+      "epoch": 2.4518743667679836,
+      "grad_norm": 0.36738779001057814,
+      "learning_rate": 2.408603181104957e-05,
+      "loss": 0.4488,
+      "step": 604
+    },
+    {
+      "epoch": 2.4559270516717326,
+      "grad_norm": 0.35488189599425357,
+      "learning_rate": 2.4030453910928245e-05,
+      "loss": 0.4467,
+      "step": 605
+    },
+    {
+      "epoch": 2.459979736575481,
+      "grad_norm": 0.34225882506672584,
+      "learning_rate": 2.397484355008521e-05,
+      "loss": 0.4391,
+      "step": 606
+    },
+    {
+      "epoch": 2.46403242147923,
+      "grad_norm": 0.3708517603180246,
+      "learning_rate": 2.3919201176398662e-05,
+      "loss": 0.4524,
+      "step": 607
+    },
+    {
+      "epoch": 2.4680851063829787,
+      "grad_norm": 0.3573057332459475,
+      "learning_rate": 2.3863527238004633e-05,
+      "loss": 0.4487,
+      "step": 608
+    },
+    {
+      "epoch": 2.4721377912867273,
+      "grad_norm": 0.39226646172926816,
+      "learning_rate": 2.380782218329337e-05,
+      "loss": 0.4475,
+      "step": 609
+    },
+    {
+      "epoch": 2.4761904761904763,
+      "grad_norm": 0.4056394488334609,
+      "learning_rate": 2.3752086460905725e-05,
+      "loss": 0.4367,
+      "step": 610
+    },
+    {
+      "epoch": 2.480243161094225,
+      "grad_norm": 0.3824361039707399,
+      "learning_rate": 2.3696320519729544e-05,
+      "loss": 0.4487,
+      "step": 611
+    },
+    {
+      "epoch": 2.484295845997974,
+      "grad_norm": 0.43321690947275676,
+      "learning_rate": 2.3640524808896045e-05,
+      "loss": 0.4316,
+      "step": 612
+    },
+    {
+      "epoch": 2.4883485309017224,
+      "grad_norm": 0.32906114150191906,
+      "learning_rate": 2.3584699777776222e-05,
+      "loss": 0.4311,
+      "step": 613
+    },
+    {
+      "epoch": 2.4924012158054714,
+      "grad_norm": 0.4135064805641933,
+      "learning_rate": 2.3528845875977195e-05,
+      "loss": 0.4367,
+      "step": 614
+    },
+    {
+      "epoch": 2.49645390070922,
+      "grad_norm": 0.3075867414052995,
+      "learning_rate": 2.3472963553338614e-05,
+      "loss": 0.4509,
+      "step": 615
+    },
+    {
+      "epoch": 2.5005065856129685,
+      "grad_norm": 0.36943563398524304,
+      "learning_rate": 2.341705325992901e-05,
+      "loss": 0.4467,
+      "step": 616
+    },
+    {
+      "epoch": 2.5045592705167175,
+      "grad_norm": 0.32497838346417,
+      "learning_rate": 2.336111544604222e-05,
+      "loss": 0.4531,
+      "step": 617
+    },
+    {
+      "epoch": 2.508611955420466,
+      "grad_norm": 0.34607269225396814,
+      "learning_rate": 2.33051505621937e-05,
+      "loss": 0.4518,
+      "step": 618
+    },
+    {
+      "epoch": 2.512664640324215,
+      "grad_norm": 0.3532800136132749,
+      "learning_rate": 2.324915905911693e-05,
+      "loss": 0.4499,
+      "step": 619
+    },
+    {
+      "epoch": 2.5167173252279635,
+      "grad_norm": 0.3469895859362161,
+      "learning_rate": 2.319314138775977e-05,
+      "loss": 0.4469,
+      "step": 620
+    },
+    {
+      "epoch": 2.520770010131712,
+      "grad_norm": 0.35303597346463045,
+      "learning_rate": 2.3137097999280856e-05,
+      "loss": 0.4631,
+      "step": 621
+    },
+    {
+      "epoch": 2.524822695035461,
+      "grad_norm": 0.33306931131371176,
+      "learning_rate": 2.308102934504593e-05,
+      "loss": 0.4483,
+      "step": 622
+    },
+    {
+      "epoch": 2.5288753799392096,
+      "grad_norm": 0.39904517279416585,
+      "learning_rate": 2.3024935876624222e-05,
+      "loss": 0.4576,
+      "step": 623
+    },
+    {
+      "epoch": 2.5329280648429586,
+      "grad_norm": 0.3153181827968658,
+      "learning_rate": 2.2968818045784813e-05,
+      "loss": 0.4367,
+      "step": 624
+    },
+    {
+      "epoch": 2.536980749746707,
+      "grad_norm": 0.3781010605339802,
+      "learning_rate": 2.2912676304493006e-05,
+      "loss": 0.4457,
+      "step": 625
+    },
+    {
+      "epoch": 2.5410334346504557,
+      "grad_norm": 0.29391441835821075,
+      "learning_rate": 2.2856511104906668e-05,
+      "loss": 0.452,
+      "step": 626
+    },
+    {
+      "epoch": 2.5450861195542047,
+      "grad_norm": 0.36242613095922027,
+      "learning_rate": 2.2800322899372586e-05,
+      "loss": 0.442,
+      "step": 627
+    },
+    {
+      "epoch": 2.5491388044579533,
+      "grad_norm": 0.2911216540986173,
+      "learning_rate": 2.2744112140422844e-05,
+      "loss": 0.4544,
+      "step": 628
+    },
+    {
+      "epoch": 2.5531914893617023,
+      "grad_norm": 0.33314197778100546,
+      "learning_rate": 2.2687879280771177e-05,
+      "loss": 0.4429,
+      "step": 629
+    },
+    {
+      "epoch": 2.557244174265451,
+      "grad_norm": 0.3373905077786125,
+      "learning_rate": 2.26316247733093e-05,
+      "loss": 0.4407,
+      "step": 630
+    },
+    {
+      "epoch": 2.5612968591691994,
+      "grad_norm": 0.29360894161075535,
+      "learning_rate": 2.257534907110328e-05,
+      "loss": 0.452,
+      "step": 631
+    },
+    {
+      "epoch": 2.5653495440729484,
+      "grad_norm": 0.3435933567924771,
+      "learning_rate": 2.2519052627389882e-05,
+      "loss": 0.4229,
+      "step": 632
+    },
+    {
+      "epoch": 2.569402228976697,
+      "grad_norm": 0.31099989928450195,
+      "learning_rate": 2.246273589557294e-05,
+      "loss": 0.4467,
+      "step": 633
+    },
+    {
+      "epoch": 2.573454913880446,
+      "grad_norm": 0.33590756174056147,
+      "learning_rate": 2.240639932921966e-05,
+      "loss": 0.4491,
+      "step": 634
+    },
+    {
+      "epoch": 2.5775075987841944,
+      "grad_norm": 0.3174088406342149,
+      "learning_rate": 2.2350043382056995e-05,
+      "loss": 0.434,
+      "step": 635
+    },
+    {
+      "epoch": 2.581560283687943,
+      "grad_norm": 0.3396256810400546,
+      "learning_rate": 2.2293668507968015e-05,
+      "loss": 0.4386,
+      "step": 636
+    },
+    {
+      "epoch": 2.585612968591692,
+      "grad_norm": 0.34941651772778226,
+      "learning_rate": 2.2237275160988186e-05,
+      "loss": 0.4453,
+      "step": 637
+    },
+    {
+      "epoch": 2.589665653495441,
+      "grad_norm": 0.3139772349811781,
+      "learning_rate": 2.2180863795301787e-05,
+      "loss": 0.4537,
+      "step": 638
+    },
+    {
+      "epoch": 2.5937183383991895,
+      "grad_norm": 0.33967533783999143,
+      "learning_rate": 2.212443486523819e-05,
+      "loss": 0.4251,
+      "step": 639
+    },
+    {
+      "epoch": 2.597771023302938,
+      "grad_norm": 0.2934155719201838,
+      "learning_rate": 2.2067988825268243e-05,
+      "loss": 0.4341,
+      "step": 640
+    },
+    {
+      "epoch": 2.601823708206687,
+      "grad_norm": 0.3261770300733525,
+      "learning_rate": 2.2011526130000596e-05,
+      "loss": 0.4258,
+      "step": 641
+    },
+    {
+      "epoch": 2.6058763931104356,
+      "grad_norm": 0.32268330268693696,
+      "learning_rate": 2.1955047234178038e-05,
+      "loss": 0.461,
+      "step": 642
+    },
+    {
+      "epoch": 2.6099290780141846,
+      "grad_norm": 0.33757751078098575,
+      "learning_rate": 2.1898552592673825e-05,
+      "loss": 0.4452,
+      "step": 643
+    },
+    {
+      "epoch": 2.613981762917933,
+      "grad_norm": 0.2876189952836773,
+      "learning_rate": 2.184204266048803e-05,
+      "loss": 0.4466,
+      "step": 644
+    },
+    {
+      "epoch": 2.6180344478216817,
+      "grad_norm": 0.3073735223765011,
+      "learning_rate": 2.1785517892743887e-05,
+      "loss": 0.4615,
+      "step": 645
+    },
+    {
+      "epoch": 2.6220871327254307,
+      "grad_norm": 0.3214758388394934,
+      "learning_rate": 2.17289787446841e-05,
+      "loss": 0.444,
+      "step": 646
+    },
+    {
+      "epoch": 2.6261398176291793,
+      "grad_norm": 0.27524340088033705,
+      "learning_rate": 2.1672425671667198e-05,
+      "loss": 0.4447,
+      "step": 647
+    },
+    {
+      "epoch": 2.6301925025329282,
+      "grad_norm": 0.29290924979770494,
+      "learning_rate": 2.161585912916385e-05,
+      "loss": 0.4572,
+      "step": 648
+    },
+    {
+      "epoch": 2.634245187436677,
+      "grad_norm": 0.33879773637830063,
+      "learning_rate": 2.1559279572753214e-05,
+      "loss": 0.4385,
+      "step": 649
+    },
+    {
+      "epoch": 2.6382978723404253,
+      "grad_norm": 0.3068252899768846,
+      "learning_rate": 2.1502687458119268e-05,
+      "loss": 0.4497,
+      "step": 650
+    },
+    {
+      "epoch": 2.6423505572441743,
+      "grad_norm": 0.35991325970743543,
+      "learning_rate": 2.1446083241047116e-05,
+      "loss": 0.4592,
+      "step": 651
+    },
+    {
+      "epoch": 2.646403242147923,
+      "grad_norm": 0.3058493977206401,
+      "learning_rate": 2.1389467377419333e-05,
+      "loss": 0.4611,
+      "step": 652
+    },
+    {
+      "epoch": 2.650455927051672,
+      "grad_norm": 0.32082623202886773,
+      "learning_rate": 2.133284032321232e-05,
+      "loss": 0.4478,
+      "step": 653
+    },
+    {
+      "epoch": 2.6545086119554204,
+      "grad_norm": 0.31452109716194143,
+      "learning_rate": 2.1276202534492566e-05,
+      "loss": 0.4377,
+      "step": 654
+    },
+    {
+      "epoch": 2.658561296859169,
+      "grad_norm": 0.3094283373019122,
+      "learning_rate": 2.121955446741306e-05,
+      "loss": 0.4488,
+      "step": 655
+    },
+    {
+      "epoch": 2.662613981762918,
+      "grad_norm": 0.30272795675876885,
+      "learning_rate": 2.1162896578209517e-05,
+      "loss": 0.4382,
+      "step": 656
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.3273500290740797,
+      "learning_rate": 2.1106229323196813e-05,
+      "loss": 0.4391,
+      "step": 657
+    },
+    {
+      "epoch": 2.6707193515704155,
+      "grad_norm": 0.31595633356171093,
+      "learning_rate": 2.1049553158765214e-05,
+      "loss": 0.4444,
+      "step": 658
+    },
+    {
+      "epoch": 2.674772036474164,
+      "grad_norm": 0.3439033188967827,
+      "learning_rate": 2.0992868541376764e-05,
+      "loss": 0.4346,
+      "step": 659
+    },
+    {
+      "epoch": 2.6788247213779126,
+      "grad_norm": 0.30719713157150863,
+      "learning_rate": 2.093617592756158e-05,
+      "loss": 0.4463,
+      "step": 660
+    },
+    {
+      "epoch": 2.6828774062816616,
+      "grad_norm": 0.35128633744959337,
+      "learning_rate": 2.0879475773914167e-05,
+      "loss": 0.4491,
+      "step": 661
+    },
+    {
+      "epoch": 2.6869300911854106,
+      "grad_norm": 0.30330176869223924,
+      "learning_rate": 2.082276853708978e-05,
+      "loss": 0.4349,
+      "step": 662
+    },
+    {
+      "epoch": 2.690982776089159,
+      "grad_norm": 0.3495879458067684,
+      "learning_rate": 2.076605467380071e-05,
+      "loss": 0.4423,
+      "step": 663
+    },
+    {
+      "epoch": 2.6950354609929077,
+      "grad_norm": 0.3512911574405841,
+      "learning_rate": 2.0709334640812613e-05,
+      "loss": 0.464,
+      "step": 664
+    },
+    {
+      "epoch": 2.6990881458966567,
+      "grad_norm": 0.34062616869990525,
+      "learning_rate": 2.0652608894940824e-05,
+      "loss": 0.4541,
+      "step": 665
+    },
+    {
+      "epoch": 2.7031408308004052,
+      "grad_norm": 0.37044441134231904,
+      "learning_rate": 2.0595877893046722e-05,
+      "loss": 0.436,
+      "step": 666
+    },
+    {
+      "epoch": 2.7071935157041542,
+      "grad_norm": 0.35347530839149116,
+      "learning_rate": 2.0539142092033985e-05,
+      "loss": 0.4403,
+      "step": 667
+    },
+    {
+      "epoch": 2.711246200607903,
+      "grad_norm": 0.33036433030140894,
+      "learning_rate": 2.048240194884496e-05,
+      "loss": 0.4525,
+      "step": 668
+    },
+    {
+      "epoch": 2.7152988855116513,
+      "grad_norm": 0.3328539221896754,
+      "learning_rate": 2.042565792045695e-05,
+      "loss": 0.4503,
+      "step": 669
+    },
+    {
+      "epoch": 2.7193515704154003,
+      "grad_norm": 0.3164094912656424,
+      "learning_rate": 2.036891046387857e-05,
+      "loss": 0.4342,
+      "step": 670
+    },
+    {
+      "epoch": 2.723404255319149,
+      "grad_norm": 0.37646322469939697,
+      "learning_rate": 2.0312160036146036e-05,
+      "loss": 0.4407,
+      "step": 671
+    },
+    {
+      "epoch": 2.727456940222898,
+      "grad_norm": 0.3513642160258712,
+      "learning_rate": 2.025540709431948e-05,
+      "loss": 0.4348,
+      "step": 672
+    },
+    {
+      "epoch": 2.7315096251266464,
+      "grad_norm": 0.36501632566818276,
+      "learning_rate": 2.0198652095479298e-05,
+      "loss": 0.4465,
+      "step": 673
+    },
+    {
+      "epoch": 2.735562310030395,
+      "grad_norm": 0.3416589340750122,
+      "learning_rate": 2.014189549672245e-05,
+      "loss": 0.4482,
+      "step": 674
+    },
+    {
+      "epoch": 2.739614994934144,
+      "grad_norm": 0.3007728333520319,
+      "learning_rate": 2.0085137755158776e-05,
+      "loss": 0.4565,
+      "step": 675
+    },
+    {
+      "epoch": 2.7436676798378925,
+      "grad_norm": 0.30824599172832745,
+      "learning_rate": 2.0028379327907327e-05,
+      "loss": 0.4447,
+      "step": 676
+    },
+    {
+      "epoch": 2.7477203647416415,
+      "grad_norm": 0.3451431554573569,
+      "learning_rate": 1.9971620672092676e-05,
+      "loss": 0.4434,
+      "step": 677
+    },
+    {
+      "epoch": 2.75177304964539,
+      "grad_norm": 0.33097863538793304,
+      "learning_rate": 1.991486224484123e-05,
+      "loss": 0.4476,
+      "step": 678
+    },
+    {
+      "epoch": 2.7558257345491386,
+      "grad_norm": 0.29961467654858276,
+      "learning_rate": 1.985810450327756e-05,
+      "loss": 0.4478,
+      "step": 679
+    },
+    {
+      "epoch": 2.7598784194528876,
+      "grad_norm": 0.3403203420809795,
+      "learning_rate": 1.9801347904520706e-05,
+      "loss": 0.4557,
+      "step": 680
+    },
+    {
+      "epoch": 2.763931104356636,
+      "grad_norm": 0.3156827715850812,
+      "learning_rate": 1.974459290568053e-05,
+      "loss": 0.4569,
+      "step": 681
+    },
+    {
+      "epoch": 2.767983789260385,
+      "grad_norm": 0.29769558939758445,
+      "learning_rate": 1.968783996385397e-05,
+      "loss": 0.4409,
+      "step": 682
+    },
+    {
+      "epoch": 2.7720364741641337,
+      "grad_norm": 0.29791217609480064,
+      "learning_rate": 1.963108953612143e-05,
+      "loss": 0.4539,
+      "step": 683
+    },
+    {
+      "epoch": 2.7760891590678822,
+      "grad_norm": 0.3229550052508969,
+      "learning_rate": 1.9574342079543056e-05,
+      "loss": 0.4562,
+      "step": 684
+    },
+    {
+      "epoch": 2.780141843971631,
+      "grad_norm": 0.28863112206548475,
+      "learning_rate": 1.9517598051155046e-05,
+      "loss": 0.4273,
+      "step": 685
+    },
+    {
+      "epoch": 2.78419452887538,
+      "grad_norm": 0.29913085611783935,
+      "learning_rate": 1.9460857907966025e-05,
+      "loss": 0.4468,
+      "step": 686
+    },
+    {
+      "epoch": 2.7882472137791288,
+      "grad_norm": 0.30504870769965775,
+      "learning_rate": 1.9404122106953285e-05,
+      "loss": 0.4366,
+      "step": 687
+    },
+    {
+      "epoch": 2.7922998986828773,
+      "grad_norm": 0.3183548687307993,
+      "learning_rate": 1.9347391105059176e-05,
+      "loss": 0.4522,
+      "step": 688
+    },
+    {
+      "epoch": 2.7963525835866263,
+      "grad_norm": 0.3019815071253538,
+      "learning_rate": 1.92906653591874e-05,
+      "loss": 0.4474,
+      "step": 689
+    },
+    {
+      "epoch": 2.800405268490375,
+      "grad_norm": 0.30336358210702535,
+      "learning_rate": 1.9233945326199295e-05,
+      "loss": 0.4522,
+      "step": 690
+    },
+    {
+      "epoch": 2.804457953394124,
+      "grad_norm": 0.30822194354763655,
+      "learning_rate": 1.917723146291022e-05,
+      "loss": 0.4348,
+      "step": 691
+    },
+    {
+      "epoch": 2.8085106382978724,
+      "grad_norm": 0.3303242014782378,
+      "learning_rate": 1.912052422608584e-05,
+      "loss": 0.4307,
+      "step": 692
+    },
+    {
+      "epoch": 2.812563323201621,
+      "grad_norm": 0.29794367899266655,
+      "learning_rate": 1.9063824072438428e-05,
+      "loss": 0.4341,
+      "step": 693
+    },
+    {
+      "epoch": 2.81661600810537,
+      "grad_norm": 0.3174664037592215,
+      "learning_rate": 1.9007131458623246e-05,
+      "loss": 0.4393,
+      "step": 694
+    },
+    {
+      "epoch": 2.8206686930091185,
+      "grad_norm": 0.30017042749364053,
+      "learning_rate": 1.895044684123479e-05,
+      "loss": 0.4427,
+      "step": 695
+    },
+    {
+      "epoch": 2.8247213779128675,
+      "grad_norm": 0.3288068825208793,
+      "learning_rate": 1.8893770676803194e-05,
+      "loss": 0.4549,
+      "step": 696
+    },
+    {
+      "epoch": 2.828774062816616,
+      "grad_norm": 0.2759460699432527,
+      "learning_rate": 1.8837103421790486e-05,
+      "loss": 0.4463,
+      "step": 697
+    },
+    {
+      "epoch": 2.8328267477203646,
+      "grad_norm": 0.3079493824829927,
+      "learning_rate": 1.8780445532586952e-05,
+      "loss": 0.4336,
+      "step": 698
+    },
+    {
+      "epoch": 2.8368794326241136,
+      "grad_norm": 0.2988597070602149,
+      "learning_rate": 1.872379746550743e-05,
+      "loss": 0.4521,
+      "step": 699
+    },
+    {
+      "epoch": 2.840932117527862,
+      "grad_norm": 0.2990122406735027,
+      "learning_rate": 1.866715967678769e-05,
+      "loss": 0.4495,
+      "step": 700
+    },
+    {
+      "epoch": 2.844984802431611,
+      "grad_norm": 0.3206623590956083,
+      "learning_rate": 1.861053262258067e-05,
+      "loss": 0.4541,
+      "step": 701
+    },
+    {
+      "epoch": 2.8490374873353597,
+      "grad_norm": 0.3017499744305013,
+      "learning_rate": 1.8553916758952897e-05,
+      "loss": 0.4299,
+      "step": 702
+    },
+    {
+      "epoch": 2.853090172239108,
+      "grad_norm": 0.34254757571502004,
+      "learning_rate": 1.8497312541880735e-05,
+      "loss": 0.4282,
+      "step": 703
+    },
+    {
+      "epoch": 2.857142857142857,
+      "grad_norm": 0.3095873526921389,
+      "learning_rate": 1.8440720427246786e-05,
+      "loss": 0.4473,
+      "step": 704
+    },
+    {
+      "epoch": 2.8611955420466058,
+      "grad_norm": 0.3295921806370062,
+      "learning_rate": 1.8384140870836157e-05,
+      "loss": 0.4427,
+      "step": 705
+    },
+    {
+      "epoch": 2.8652482269503547,
+      "grad_norm": 0.3366039141201994,
+      "learning_rate": 1.8327574328332806e-05,
+      "loss": 0.4477,
+      "step": 706
+    },
+    {
+      "epoch": 2.8693009118541033,
+      "grad_norm": 0.32572680530478754,
+      "learning_rate": 1.8271021255315906e-05,
+      "loss": 0.4455,
+      "step": 707
+    },
+    {
+      "epoch": 2.873353596757852,
+      "grad_norm": 0.339333313576772,
+      "learning_rate": 1.8214482107256117e-05,
+      "loss": 0.4467,
+      "step": 708
+    },
+    {
+      "epoch": 2.877406281661601,
+      "grad_norm": 0.3395683983420118,
+      "learning_rate": 1.8157957339511968e-05,
+      "loss": 0.4385,
+      "step": 709
+    },
+    {
+      "epoch": 2.8814589665653494,
+      "grad_norm": 0.3036800446427919,
+      "learning_rate": 1.8101447407326182e-05,
+      "loss": 0.4377,
+      "step": 710
+    },
+    {
+      "epoch": 2.8855116514690984,
+      "grad_norm": 0.30107971129121613,
+      "learning_rate": 1.8044952765821966e-05,
+      "loss": 0.4422,
+      "step": 711
+    },
+    {
+      "epoch": 2.889564336372847,
+      "grad_norm": 0.3148485900685422,
+      "learning_rate": 1.7988473869999407e-05,
+      "loss": 0.4439,
+      "step": 712
+    },
+    {
+      "epoch": 2.8936170212765955,
+      "grad_norm": 0.2950791393623461,
+      "learning_rate": 1.7932011174731764e-05,
+      "loss": 0.442,
+      "step": 713
+    },
+    {
+      "epoch": 2.8976697061803445,
+      "grad_norm": 0.28569760056665355,
+      "learning_rate": 1.7875565134761817e-05,
+      "loss": 0.4617,
+      "step": 714
+    },
+    {
+      "epoch": 2.9017223910840935,
+      "grad_norm": 0.3015151902480177,
+      "learning_rate": 1.7819136204698226e-05,
+      "loss": 0.4417,
+      "step": 715
+    },
+    {
+      "epoch": 2.905775075987842,
+      "grad_norm": 0.29907971014512114,
+      "learning_rate": 1.776272483901182e-05,
+      "loss": 0.4414,
+      "step": 716
+    },
+    {
+      "epoch": 2.9098277608915906,
+      "grad_norm": 0.29303278600781346,
+      "learning_rate": 1.7706331492031995e-05,
+      "loss": 0.4458,
+      "step": 717
+    },
+    {
+      "epoch": 2.9138804457953396,
+      "grad_norm": 0.3352112405041721,
+      "learning_rate": 1.764995661794301e-05,
+      "loss": 0.4448,
+      "step": 718
+    },
+    {
+      "epoch": 2.917933130699088,
+      "grad_norm": 0.29192361821003954,
+      "learning_rate": 1.759360067078035e-05,
+      "loss": 0.4402,
+      "step": 719
+    },
+    {
+      "epoch": 2.921985815602837,
+      "grad_norm": 0.3295767993291945,
+      "learning_rate": 1.7537264104427064e-05,
+      "loss": 0.4411,
+      "step": 720
+    },
+    {
+      "epoch": 2.9260385005065856,
+      "grad_norm": 0.27014651494193037,
+      "learning_rate": 1.748094737261012e-05,
+      "loss": 0.4511,
+      "step": 721
+    },
+    {
+      "epoch": 2.930091185410334,
+      "grad_norm": 0.29532333503549385,
+      "learning_rate": 1.7424650928896726e-05,
+      "loss": 0.4392,
+      "step": 722
+    },
+    {
+      "epoch": 2.934143870314083,
+      "grad_norm": 0.29101374290441184,
+      "learning_rate": 1.7368375226690712e-05,
+      "loss": 0.4373,
+      "step": 723
+    },
+    {
+      "epoch": 2.9381965552178317,
+      "grad_norm": 0.2812497261151677,
+      "learning_rate": 1.731212071922883e-05,
+      "loss": 0.446,
+      "step": 724
+    },
+    {
+      "epoch": 2.9422492401215807,
+      "grad_norm": 0.34742649925238944,
+      "learning_rate": 1.7255887859577156e-05,
+      "loss": 0.4411,
+      "step": 725
+    },
+    {
+      "epoch": 2.9463019250253293,
+      "grad_norm": 0.2660535524593765,
+      "learning_rate": 1.7199677100627427e-05,
+      "loss": 0.4602,
+      "step": 726
+    },
+    {
+      "epoch": 2.950354609929078,
+      "grad_norm": 0.3596071028043233,
+      "learning_rate": 1.7143488895093343e-05,
+      "loss": 0.4479,
+      "step": 727
+    },
+    {
+      "epoch": 2.954407294832827,
+      "grad_norm": 0.2771537138202783,
+      "learning_rate": 1.7087323695506994e-05,
+      "loss": 0.4574,
+      "step": 728
+    },
+    {
+      "epoch": 2.9584599797365754,
+      "grad_norm": 0.3106087990541143,
+      "learning_rate": 1.7031181954215194e-05,
+      "loss": 0.4539,
+      "step": 729
+    },
+    {
+      "epoch": 2.9625126646403244,
+      "grad_norm": 0.31238413735800574,
+      "learning_rate": 1.6975064123375788e-05,
+      "loss": 0.444,
+      "step": 730
+    },
+    {
+      "epoch": 2.966565349544073,
+      "grad_norm": 0.26556593177038745,
+      "learning_rate": 1.6918970654954084e-05,
+      "loss": 0.434,
+      "step": 731
+    },
+    {
+      "epoch": 2.9706180344478215,
+      "grad_norm": 0.29900316896919654,
+      "learning_rate": 1.686290200071915e-05,
+      "loss": 0.4497,
+      "step": 732
+    },
+    {
+      "epoch": 2.9746707193515705,
+      "grad_norm": 0.2914092049199421,
+      "learning_rate": 1.6806858612240234e-05,
+      "loss": 0.436,
+      "step": 733
+    },
+    {
+      "epoch": 2.978723404255319,
+      "grad_norm": 0.2665007332908628,
+      "learning_rate": 1.6750840940883078e-05,
+      "loss": 0.4468,
+      "step": 734
+    },
+    {
+      "epoch": 2.982776089159068,
+      "grad_norm": 0.2578489529933866,
+      "learning_rate": 1.6694849437806305e-05,
+      "loss": 0.4583,
+      "step": 735
+    },
+    {
+      "epoch": 2.9868287740628165,
+      "grad_norm": 0.2949833436060018,
+      "learning_rate": 1.663888455395778e-05,
+      "loss": 0.4309,
+      "step": 736
+    },
+    {
+      "epoch": 2.990881458966565,
+      "grad_norm": 0.29864411839978644,
+      "learning_rate": 1.6582946740070995e-05,
+      "loss": 0.4378,
+      "step": 737
+    },
+    {
+      "epoch": 2.994934143870314,
+      "grad_norm": 0.30509975133692285,
+      "learning_rate": 1.6527036446661396e-05,
+      "loss": 0.4507,
+      "step": 738
+    },
+    {
+      "epoch": 3.0010131712259374,
+      "grad_norm": 0.2904486926820399,
+      "learning_rate": 1.6471154124022818e-05,
+      "loss": 0.4434,
+      "step": 739
+    },
+    {
+      "epoch": 3.005065856129686,
+      "grad_norm": 0.424593221842747,
+      "learning_rate": 1.6415300222223788e-05,
+      "loss": 0.3968,
+      "step": 740
+    },
+    {
+      "epoch": 3.0091185410334345,
+      "grad_norm": 0.3516938590175839,
+      "learning_rate": 1.6359475191103958e-05,
+      "loss": 0.3781,
+      "step": 741
+    },
+    {
+      "epoch": 3.0131712259371835,
+      "grad_norm": 0.6150739562559858,
+      "learning_rate": 1.6303679480270466e-05,
+      "loss": 0.3683,
+      "step": 742
+    },
+    {
+      "epoch": 3.017223910840932,
+      "grad_norm": 0.4200938480292203,
+      "learning_rate": 1.624791353909428e-05,
+      "loss": 0.3862,
+      "step": 743
+    },
+    {
+      "epoch": 3.021276595744681,
+      "grad_norm": 0.4530342634633421,
+      "learning_rate": 1.619217781670663e-05,
+      "loss": 0.4055,
+      "step": 744
+    },
+    {
+      "epoch": 3.0253292806484295,
+      "grad_norm": 0.4177319884095146,
+      "learning_rate": 1.6136472761995373e-05,
+      "loss": 0.3678,
+      "step": 745
+    },
+    {
+      "epoch": 3.0293819655521785,
+      "grad_norm": 0.4379591988524175,
+      "learning_rate": 1.608079882360134e-05,
+      "loss": 0.368,
+      "step": 746
+    },
+    {
+      "epoch": 3.033434650455927,
+      "grad_norm": 0.44141483682509264,
+      "learning_rate": 1.60251564499148e-05,
+      "loss": 0.3793,
+      "step": 747
+    },
+    {
+      "epoch": 3.0374873353596756,
+      "grad_norm": 0.44547407014593793,
+      "learning_rate": 1.596954608907176e-05,
+      "loss": 0.3757,
+      "step": 748
+    },
+    {
+      "epoch": 3.0415400202634246,
+      "grad_norm": 0.34663492303163934,
+      "learning_rate": 1.591396818895043e-05,
+      "loss": 0.377,
+      "step": 749
+    },
+    {
+      "epoch": 3.045592705167173,
+      "grad_norm": 0.4130705246369375,
+      "learning_rate": 1.585842319716759e-05,
+      "loss": 0.3754,
+      "step": 750
+    },
+    {
+      "epoch": 3.049645390070922,
+      "grad_norm": 0.3542988434115936,
+      "learning_rate": 1.5802911561074944e-05,
+      "loss": 0.3772,
+      "step": 751
+    },
+    {
+      "epoch": 3.0536980749746707,
+      "grad_norm": 0.3840931272267641,
+      "learning_rate": 1.5747433727755595e-05,
+      "loss": 0.3782,
+      "step": 752
+    },
+    {
+      "epoch": 3.0577507598784193,
+      "grad_norm": 0.3640354547040424,
+      "learning_rate": 1.5691990144020376e-05,
+      "loss": 0.3765,
+      "step": 753
+    },
+    {
+      "epoch": 3.0618034447821683,
+      "grad_norm": 0.31522179710902176,
+      "learning_rate": 1.5636581256404297e-05,
+      "loss": 0.3665,
+      "step": 754
+    },
+    {
+      "epoch": 3.065856129685917,
+      "grad_norm": 0.3707685956522802,
+      "learning_rate": 1.558120751116291e-05,
+      "loss": 0.3742,
+      "step": 755
+    },
+    {
+      "epoch": 3.069908814589666,
+      "grad_norm": 0.3217415553060016,
+      "learning_rate": 1.552586935426876e-05,
+      "loss": 0.3841,
+      "step": 756
+    },
+    {
+      "epoch": 3.0739614994934144,
+      "grad_norm": 0.3258575564881046,
+      "learning_rate": 1.547056723140774e-05,
+      "loss": 0.3825,
+      "step": 757
+    },
+    {
+      "epoch": 3.078014184397163,
+      "grad_norm": 0.3478935300480006,
+      "learning_rate": 1.5415301587975565e-05,
+      "loss": 0.3673,
+      "step": 758
+    },
+    {
+      "epoch": 3.082066869300912,
+      "grad_norm": 0.30221546863403326,
+      "learning_rate": 1.536007286907411e-05,
+      "loss": 0.376,
+      "step": 759
+    },
+    {
+      "epoch": 3.0861195542046604,
+      "grad_norm": 0.3780320676399285,
+      "learning_rate": 1.5304881519507896e-05,
+      "loss": 0.3925,
+      "step": 760
+    },
+    {
+      "epoch": 3.0901722391084094,
+      "grad_norm": 0.2804951652470931,
+      "learning_rate": 1.5249727983780453e-05,
+      "loss": 0.3765,
+      "step": 761
+    },
+    {
+      "epoch": 3.094224924012158,
+      "grad_norm": 0.345845021631185,
+      "learning_rate": 1.5194612706090786e-05,
+      "loss": 0.3787,
+      "step": 762
+    },
+    {
+      "epoch": 3.098277608915907,
+      "grad_norm": 0.3049223900579112,
+      "learning_rate": 1.5139536130329771e-05,
+      "loss": 0.37,
+      "step": 763
+    },
+    {
+      "epoch": 3.1023302938196555,
+      "grad_norm": 0.29705894965763635,
+      "learning_rate": 1.508449870007656e-05,
+      "loss": 0.387,
+      "step": 764
+    },
+    {
+      "epoch": 3.106382978723404,
+      "grad_norm": 0.30195231262870714,
+      "learning_rate": 1.5029500858595056e-05,
+      "loss": 0.3799,
+      "step": 765
+    },
+    {
+      "epoch": 3.110435663627153,
+      "grad_norm": 0.30220918166568445,
+      "learning_rate": 1.4974543048830328e-05,
+      "loss": 0.3702,
+      "step": 766
+    },
+    {
+      "epoch": 3.1144883485309016,
+      "grad_norm": 0.3268285877005793,
+      "learning_rate": 1.4919625713405e-05,
+      "loss": 0.3589,
+      "step": 767
+    },
+    {
+      "epoch": 3.1185410334346506,
+      "grad_norm": 0.3002920237627172,
+      "learning_rate": 1.4864749294615756e-05,
+      "loss": 0.3962,
+      "step": 768
+    },
+    {
+      "epoch": 3.122593718338399,
+      "grad_norm": 0.3107827545607655,
+      "learning_rate": 1.4809914234429716e-05,
+      "loss": 0.3799,
+      "step": 769
+    },
+    {
+      "epoch": 3.126646403242148,
+      "grad_norm": 0.3055912508616054,
+      "learning_rate": 1.4755120974480923e-05,
+      "loss": 0.3793,
+      "step": 770
+    },
+    {
+      "epoch": 3.1306990881458967,
+      "grad_norm": 0.3152214387272536,
+      "learning_rate": 1.4700369956066771e-05,
+      "loss": 0.3773,
+      "step": 771
+    },
+    {
+      "epoch": 3.1347517730496453,
+      "grad_norm": 0.3073789019687832,
+      "learning_rate": 1.4645661620144413e-05,
+      "loss": 0.3867,
+      "step": 772
+    },
+    {
+      "epoch": 3.1388044579533942,
+      "grad_norm": 0.31435177793268887,
+      "learning_rate": 1.4590996407327284e-05,
+      "loss": 0.3736,
+      "step": 773
+    },
+    {
+      "epoch": 3.142857142857143,
+      "grad_norm": 0.29461137758717176,
+      "learning_rate": 1.4536374757881487e-05,
+      "loss": 0.3877,
+      "step": 774
+    },
+    {
+      "epoch": 3.146909827760892,
+      "grad_norm": 0.28198099001639415,
+      "learning_rate": 1.4481797111722271e-05,
+      "loss": 0.4069,
+      "step": 775
+    },
+    {
+      "epoch": 3.1509625126646403,
+      "grad_norm": 0.293567171721376,
+      "learning_rate": 1.4427263908410507e-05,
+      "loss": 0.3772,
+      "step": 776
+    },
+    {
+      "epoch": 3.155015197568389,
+      "grad_norm": 0.2986529484683073,
+      "learning_rate": 1.4372775587149108e-05,
+      "loss": 0.3918,
+      "step": 777
+    },
+    {
+      "epoch": 3.159067882472138,
+      "grad_norm": 0.2818015213233193,
+      "learning_rate": 1.4318332586779522e-05,
+      "loss": 0.3964,
+      "step": 778
+    },
+    {
+      "epoch": 3.1631205673758864,
+      "grad_norm": 0.3074096888569812,
+      "learning_rate": 1.4263935345778202e-05,
+      "loss": 0.4049,
+      "step": 779
+    },
+    {
+      "epoch": 3.1671732522796354,
+      "grad_norm": 0.25953109238577915,
+      "learning_rate": 1.420958430225303e-05,
+      "loss": 0.3794,
+      "step": 780
+    },
+    {
+      "epoch": 3.171225937183384,
+      "grad_norm": 0.31294619245811045,
+      "learning_rate": 1.415527989393985e-05,
+      "loss": 0.3548,
+      "step": 781
+    },
+    {
+      "epoch": 3.1752786220871325,
+      "grad_norm": 0.27614100810599057,
+      "learning_rate": 1.410102255819891e-05,
+      "loss": 0.3736,
+      "step": 782
+    },
+    {
+      "epoch": 3.1793313069908815,
+      "grad_norm": 0.28708403168863017,
+      "learning_rate": 1.404681273201131e-05,
+      "loss": 0.3757,
+      "step": 783
+    },
+    {
+      "epoch": 3.18338399189463,
+      "grad_norm": 0.3075666016417243,
+      "learning_rate": 1.399265085197556e-05,
+      "loss": 0.3912,
+      "step": 784
+    },
+    {
+      "epoch": 3.187436676798379,
+      "grad_norm": 0.2955733847675601,
+      "learning_rate": 1.393853735430398e-05,
+      "loss": 0.3641,
+      "step": 785
+    },
+    {
+      "epoch": 3.1914893617021276,
+      "grad_norm": 0.29322215407182345,
+      "learning_rate": 1.3884472674819246e-05,
+      "loss": 0.3766,
+      "step": 786
+    },
+    {
+      "epoch": 3.1955420466058766,
+      "grad_norm": 0.2849763688385754,
+      "learning_rate": 1.3830457248950864e-05,
+      "loss": 0.3843,
+      "step": 787
+    },
+    {
+      "epoch": 3.199594731509625,
+      "grad_norm": 0.2974370891380051,
+      "learning_rate": 1.377649151173163e-05,
+      "loss": 0.3871,
+      "step": 788
+    },
+    {
+      "epoch": 3.2036474164133737,
+      "grad_norm": 0.29433329535090247,
+      "learning_rate": 1.3722575897794181e-05,
+      "loss": 0.3956,
+      "step": 789
+    },
+    {
+      "epoch": 3.2077001013171227,
+      "grad_norm": 0.2907393464166778,
+      "learning_rate": 1.3668710841367472e-05,
+      "loss": 0.3816,
+      "step": 790
+    },
+    {
+      "epoch": 3.2117527862208712,
+      "grad_norm": 0.30027682817655954,
+      "learning_rate": 1.361489677627324e-05,
+      "loss": 0.3756,
+      "step": 791
+    },
+    {
+      "epoch": 3.2158054711246202,
+      "grad_norm": 0.2847557925785226,
+      "learning_rate": 1.3561134135922585e-05,
+      "loss": 0.3791,
+      "step": 792
+    },
+    {
+      "epoch": 3.219858156028369,
+      "grad_norm": 0.28216159575831395,
+      "learning_rate": 1.350742335331241e-05,
+      "loss": 0.386,
+      "step": 793
+    },
+    {
+      "epoch": 3.2239108409321173,
+      "grad_norm": 0.2710375038479823,
+      "learning_rate": 1.345376486102198e-05,
+      "loss": 0.3735,
+      "step": 794
+    },
+    {
+      "epoch": 3.2279635258358663,
+      "grad_norm": 0.28059097426665736,
+      "learning_rate": 1.3400159091209414e-05,
+      "loss": 0.3768,
+      "step": 795
+    },
+    {
+      "epoch": 3.232016210739615,
+      "grad_norm": 0.2723536413524358,
+      "learning_rate": 1.3346606475608216e-05,
+      "loss": 0.3842,
+      "step": 796
+    },
+    {
+      "epoch": 3.236068895643364,
+      "grad_norm": 0.2853816952220937,
+      "learning_rate": 1.3293107445523781e-05,
+      "loss": 0.3787,
+      "step": 797
+    },
+    {
+      "epoch": 3.2401215805471124,
+      "grad_norm": 0.2959084383207551,
+      "learning_rate": 1.3239662431829949e-05,
+      "loss": 0.3918,
+      "step": 798
+    },
+    {
+      "epoch": 3.2441742654508614,
+      "grad_norm": 0.2766603098804154,
+      "learning_rate": 1.3186271864965509e-05,
+      "loss": 0.3624,
+      "step": 799
+    },
+    {
+      "epoch": 3.24822695035461,
+      "grad_norm": 0.3026042150857366,
+      "learning_rate": 1.3132936174930756e-05,
+      "loss": 0.3621,
+      "step": 800
+    },
+    {
+      "epoch": 3.2522796352583585,
+      "grad_norm": 0.2742561487743748,
+      "learning_rate": 1.3079655791283995e-05,
+      "loss": 0.3893,
+      "step": 801
+    },
+    {
+      "epoch": 3.2563323201621075,
+      "grad_norm": 0.2760375628958661,
+      "learning_rate": 1.3026431143138108e-05,
+      "loss": 0.3922,
+      "step": 802
+    },
+    {
+      "epoch": 3.260385005065856,
+      "grad_norm": 0.28813637507981776,
+      "learning_rate": 1.2973262659157114e-05,
+      "loss": 0.4042,
+      "step": 803
+    },
+    {
+      "epoch": 3.264437689969605,
+      "grad_norm": 0.27877541883218393,
+      "learning_rate": 1.2920150767552651e-05,
+      "loss": 0.3561,
+      "step": 804
+    },
+    {
+      "epoch": 3.2684903748733536,
+      "grad_norm": 0.2731620905857762,
+      "learning_rate": 1.2867095896080607e-05,
+      "loss": 0.3785,
+      "step": 805
+    },
+    {
+      "epoch": 3.272543059777102,
+      "grad_norm": 0.2961690620105766,
+      "learning_rate": 1.2814098472037612e-05,
+      "loss": 0.3767,
+      "step": 806
+    },
+    {
+      "epoch": 3.276595744680851,
+      "grad_norm": 0.29509104047589546,
+      "learning_rate": 1.276115892225764e-05,
+      "loss": 0.3667,
+      "step": 807
+    },
+    {
+      "epoch": 3.2806484295845997,
+      "grad_norm": 0.27336537556697477,
+      "learning_rate": 1.2708277673108555e-05,
+      "loss": 0.387,
+      "step": 808
+    },
+    {
+      "epoch": 3.2847011144883487,
+      "grad_norm": 0.30747567351227856,
+      "learning_rate": 1.2655455150488649e-05,
+      "loss": 0.3674,
+      "step": 809
+    },
+    {
+      "epoch": 3.288753799392097,
+      "grad_norm": 0.29696171125542814,
+      "learning_rate": 1.2602691779823272e-05,
+      "loss": 0.3996,
+      "step": 810
+    },
+    {
+      "epoch": 3.2928064842958458,
+      "grad_norm": 0.2778289306346103,
+      "learning_rate": 1.2549987986061355e-05,
+      "loss": 0.3777,
+      "step": 811
+    },
+    {
+      "epoch": 3.2968591691995948,
+      "grad_norm": 0.2871112908730644,
+      "learning_rate": 1.2497344193672005e-05,
+      "loss": 0.3735,
+      "step": 812
+    },
+    {
+      "epoch": 3.3009118541033433,
+      "grad_norm": 0.2703737925153029,
+      "learning_rate": 1.2444760826641092e-05,
+      "loss": 0.4039,
+      "step": 813
+    },
+    {
+      "epoch": 3.3049645390070923,
+      "grad_norm": 0.29268238257235124,
+      "learning_rate": 1.2392238308467817e-05,
+      "loss": 0.3783,
+      "step": 814
+    },
+    {
+      "epoch": 3.309017223910841,
+      "grad_norm": 0.25914824607343756,
+      "learning_rate": 1.2339777062161326e-05,
+      "loss": 0.3925,
+      "step": 815
+    },
+    {
+      "epoch": 3.31306990881459,
+      "grad_norm": 0.279073011138421,
+      "learning_rate": 1.2287377510237293e-05,
+      "loss": 0.3873,
+      "step": 816
+    },
+    {
+      "epoch": 3.3171225937183384,
+      "grad_norm": 0.28560425005596807,
+      "learning_rate": 1.2235040074714488e-05,
+      "loss": 0.3985,
+      "step": 817
+    },
+    {
+      "epoch": 3.321175278622087,
+      "grad_norm": 0.2434895695464389,
+      "learning_rate": 1.2182765177111434e-05,
+      "loss": 0.3672,
+      "step": 818
+    },
+    {
+      "epoch": 3.325227963525836,
+      "grad_norm": 0.3157947536896033,
+      "learning_rate": 1.213055323844297e-05,
+      "loss": 0.3688,
+      "step": 819
+    },
+    {
+      "epoch": 3.3292806484295845,
+      "grad_norm": 0.25622059523143437,
+      "learning_rate": 1.2078404679216864e-05,
+      "loss": 0.3856,
+      "step": 820
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 0.2549363724525438,
+      "learning_rate": 1.2026319919430458e-05,
+      "loss": 0.3818,
+      "step": 821
+    },
+    {
+      "epoch": 3.337386018237082,
+      "grad_norm": 0.27402175080321245,
+      "learning_rate": 1.1974299378567227e-05,
+      "loss": 0.377,
+      "step": 822
+    },
+    {
+      "epoch": 3.341438703140831,
+      "grad_norm": 0.2606916214581092,
+      "learning_rate": 1.1922343475593462e-05,
+      "loss": 0.3859,
+      "step": 823
+    },
+    {
+      "epoch": 3.3454913880445796,
+      "grad_norm": 0.26546023109038186,
+      "learning_rate": 1.187045262895488e-05,
+      "loss": 0.3857,
+      "step": 824
+    },
+    {
+      "epoch": 3.349544072948328,
+      "grad_norm": 0.28538081996767695,
+      "learning_rate": 1.1818627256573203e-05,
+      "loss": 0.3805,
+      "step": 825
+    },
+    {
+      "epoch": 3.353596757852077,
+      "grad_norm": 0.2625457689722491,
+      "learning_rate": 1.1766867775842864e-05,
+      "loss": 0.3957,
+      "step": 826
+    },
+    {
+      "epoch": 3.3576494427558257,
+      "grad_norm": 0.27508901977596273,
+      "learning_rate": 1.1715174603627615e-05,
+      "loss": 0.3839,
+      "step": 827
+    },
+    {
+      "epoch": 3.3617021276595747,
+      "grad_norm": 0.2986378797415174,
+      "learning_rate": 1.1663548156257147e-05,
+      "loss": 0.3789,
+      "step": 828
+    },
+    {
+      "epoch": 3.365754812563323,
+      "grad_norm": 0.23867793076966923,
+      "learning_rate": 1.161198884952377e-05,
+      "loss": 0.3996,
+      "step": 829
+    },
+    {
+      "epoch": 3.3698074974670718,
+      "grad_norm": 0.26597500572465627,
+      "learning_rate": 1.1560497098679056e-05,
+      "loss": 0.3785,
+      "step": 830
+    },
+    {
+      "epoch": 3.3738601823708207,
+      "grad_norm": 0.27059348209968437,
+      "learning_rate": 1.1509073318430479e-05,
+      "loss": 0.3949,
+      "step": 831
+    },
+    {
+      "epoch": 3.3779128672745693,
+      "grad_norm": 0.26741435023113813,
+      "learning_rate": 1.1457717922938116e-05,
+      "loss": 0.387,
+      "step": 832
+    },
+    {
+      "epoch": 3.3819655521783183,
+      "grad_norm": 0.28979849854333145,
+      "learning_rate": 1.1406431325811233e-05,
+      "loss": 0.4014,
+      "step": 833
+    },
+    {
+      "epoch": 3.386018237082067,
+      "grad_norm": 0.2804573125543681,
+      "learning_rate": 1.135521394010506e-05,
+      "loss": 0.369,
+      "step": 834
+    },
+    {
+      "epoch": 3.3900709219858154,
+      "grad_norm": 0.25260685176343123,
+      "learning_rate": 1.1304066178317367e-05,
+      "loss": 0.378,
+      "step": 835
+    },
+    {
+      "epoch": 3.3941236068895644,
+      "grad_norm": 0.29897722846356234,
+      "learning_rate": 1.1252988452385199e-05,
+      "loss": 0.3719,
+      "step": 836
+    },
+    {
+      "epoch": 3.398176291793313,
+      "grad_norm": 0.2657487645919887,
+      "learning_rate": 1.1201981173681536e-05,
+      "loss": 0.3698,
+      "step": 837
+    },
+    {
+      "epoch": 3.402228976697062,
+      "grad_norm": 0.26029149813444563,
+      "learning_rate": 1.1151044753011991e-05,
+      "loss": 0.3904,
+      "step": 838
+    },
+    {
+      "epoch": 3.4062816616008105,
+      "grad_norm": 0.2734679281200348,
+      "learning_rate": 1.1100179600611491e-05,
+      "loss": 0.3848,
+      "step": 839
+    },
+    {
+      "epoch": 3.410334346504559,
+      "grad_norm": 0.2716137131398029,
+      "learning_rate": 1.1049386126140985e-05,
+      "loss": 0.3741,
+      "step": 840
+    },
+    {
+      "epoch": 3.414387031408308,
+      "grad_norm": 0.2634184139064893,
+      "learning_rate": 1.0998664738684128e-05,
+      "loss": 0.3765,
+      "step": 841
+    },
+    {
+      "epoch": 3.4184397163120566,
+      "grad_norm": 0.27904435880000683,
+      "learning_rate": 1.0948015846744e-05,
+      "loss": 0.3849,
+      "step": 842
+    },
+    {
+      "epoch": 3.4224924012158056,
+      "grad_norm": 0.25797630616007766,
+      "learning_rate": 1.0897439858239832e-05,
+      "loss": 0.3801,
+      "step": 843
+    },
+    {
+      "epoch": 3.426545086119554,
+      "grad_norm": 0.2844606074488202,
+      "learning_rate": 1.0846937180503652e-05,
+      "loss": 0.37,
+      "step": 844
+    },
+    {
+      "epoch": 3.430597771023303,
+      "grad_norm": 0.26661373746076855,
+      "learning_rate": 1.0796508220277117e-05,
+      "loss": 0.399,
+      "step": 845
+    },
+    {
+      "epoch": 3.4346504559270516,
+      "grad_norm": 0.26920641761077396,
+      "learning_rate": 1.0746153383708107e-05,
+      "loss": 0.383,
+      "step": 846
+    },
+    {
+      "epoch": 3.4387031408308006,
+      "grad_norm": 0.2542666129754366,
+      "learning_rate": 1.0695873076347579e-05,
+      "loss": 0.3935,
+      "step": 847
+    },
+    {
+      "epoch": 3.442755825734549,
+      "grad_norm": 0.25541270782132497,
+      "learning_rate": 1.0645667703146205e-05,
+      "loss": 0.3712,
+      "step": 848
+    },
+    {
+      "epoch": 3.4468085106382977,
+      "grad_norm": 0.2914645649576862,
+      "learning_rate": 1.0595537668451161e-05,
+      "loss": 0.3908,
+      "step": 849
+    },
+    {
+      "epoch": 3.4508611955420467,
+      "grad_norm": 0.2596787124867935,
+      "learning_rate": 1.0545483376002854e-05,
+      "loss": 0.378,
+      "step": 850
+    },
+    {
+      "epoch": 3.4549138804457953,
+      "grad_norm": 0.28695781466548614,
+      "learning_rate": 1.0495505228931676e-05,
+      "loss": 0.373,
+      "step": 851
+    },
+    {
+      "epoch": 3.4589665653495443,
+      "grad_norm": 0.27698444545983414,
+      "learning_rate": 1.044560362975474e-05,
+      "loss": 0.3769,
+      "step": 852
+    },
+    {
+      "epoch": 3.463019250253293,
+      "grad_norm": 0.25443344921029964,
+      "learning_rate": 1.0395778980372695e-05,
+      "loss": 0.366,
+      "step": 853
+    },
+    {
+      "epoch": 3.4670719351570414,
+      "grad_norm": 0.2701842417622334,
+      "learning_rate": 1.0346031682066381e-05,
+      "loss": 0.3878,
+      "step": 854
+    },
+    {
+      "epoch": 3.4711246200607904,
+      "grad_norm": 0.2636816194464245,
+      "learning_rate": 1.0296362135493724e-05,
+      "loss": 0.3912,
+      "step": 855
+    },
+    {
+      "epoch": 3.475177304964539,
+      "grad_norm": 0.27153472589611827,
+      "learning_rate": 1.0246770740686422e-05,
+      "loss": 0.3831,
+      "step": 856
+    },
+    {
+      "epoch": 3.479229989868288,
+      "grad_norm": 0.27422607972872015,
+      "learning_rate": 1.0197257897046743e-05,
+      "loss": 0.3817,
+      "step": 857
+    },
+    {
+      "epoch": 3.4832826747720365,
+      "grad_norm": 0.26624394577954064,
+      "learning_rate": 1.014782400334433e-05,
+      "loss": 0.3951,
+      "step": 858
+    },
+    {
+      "epoch": 3.487335359675785,
+      "grad_norm": 0.2632487391095868,
+      "learning_rate": 1.009846945771296e-05,
+      "loss": 0.3748,
+      "step": 859
+    },
+    {
+      "epoch": 3.491388044579534,
+      "grad_norm": 0.2633740274445684,
+      "learning_rate": 1.0049194657647363e-05,
+      "loss": 0.3924,
+      "step": 860
+    },
+    {
+      "epoch": 3.4954407294832825,
+      "grad_norm": 0.25876546737194156,
+      "learning_rate": 1.0000000000000006e-05,
+      "loss": 0.3857,
+      "step": 861
+    },
+    {
+      "epoch": 3.4994934143870315,
+      "grad_norm": 0.2678369579513087,
+      "learning_rate": 9.950885880977891e-06,
+      "loss": 0.3701,
+      "step": 862
+    },
+    {
+      "epoch": 3.50354609929078,
+      "grad_norm": 0.2730054374452717,
+      "learning_rate": 9.901852696139382e-06,
+      "loss": 0.3713,
+      "step": 863
+    },
+    {
+      "epoch": 3.5075987841945286,
+      "grad_norm": 0.26020743387710277,
+      "learning_rate": 9.852900840391027e-06,
+      "loss": 0.3934,
+      "step": 864
+    },
+    {
+      "epoch": 3.5116514690982776,
+      "grad_norm": 0.2958818361826814,
+      "learning_rate": 9.804030707984313e-06,
+      "loss": 0.3774,
+      "step": 865
+    },
+    {
+      "epoch": 3.515704154002026,
+      "grad_norm": 0.2805318368422625,
+      "learning_rate": 9.755242692512599e-06,
+      "loss": 0.3669,
+      "step": 866
+    },
+    {
+      "epoch": 3.519756838905775,
+      "grad_norm": 0.2480031469657075,
+      "learning_rate": 9.70653718690782e-06,
+      "loss": 0.3898,
+      "step": 867
+    },
+    {
+      "epoch": 3.5238095238095237,
+      "grad_norm": 0.28233674872525927,
+      "learning_rate": 9.657914583437454e-06,
+      "loss": 0.3837,
+      "step": 868
+    },
+    {
+      "epoch": 3.5278622087132723,
+      "grad_norm": 0.2670526331881823,
+      "learning_rate": 9.609375273701246e-06,
+      "loss": 0.3735,
+      "step": 869
+    },
+    {
+      "epoch": 3.5319148936170213,
+      "grad_norm": 0.24365645103486785,
+      "learning_rate": 9.560919648628133e-06,
+      "loss": 0.3988,
+      "step": 870
+    },
+    {
+      "epoch": 3.5359675785207703,
+      "grad_norm": 0.27796258992061873,
+      "learning_rate": 9.512548098473047e-06,
+      "loss": 0.3958,
+      "step": 871
+    },
+    {
+      "epoch": 3.540020263424519,
+      "grad_norm": 0.7051140925998627,
+      "learning_rate": 9.464261012813825e-06,
+      "loss": 0.4069,
+      "step": 872
+    },
+    {
+      "epoch": 3.5440729483282674,
+      "grad_norm": 0.2746925444918877,
+      "learning_rate": 9.416058780547987e-06,
+      "loss": 0.3714,
+      "step": 873
+    },
+    {
+      "epoch": 3.5481256332320164,
+      "grad_norm": 0.25279237447573455,
+      "learning_rate": 9.367941789889714e-06,
+      "loss": 0.3909,
+      "step": 874
+    },
+    {
+      "epoch": 3.552178318135765,
+      "grad_norm": 0.24768085262923958,
+      "learning_rate": 9.319910428366607e-06,
+      "loss": 0.3755,
+      "step": 875
+    },
+    {
+      "epoch": 3.556231003039514,
+      "grad_norm": 0.2747154223771578,
+      "learning_rate": 9.271965082816667e-06,
+      "loss": 0.3875,
+      "step": 876
+    },
+    {
+      "epoch": 3.5602836879432624,
+      "grad_norm": 0.28027387227942985,
+      "learning_rate": 9.224106139385111e-06,
+      "loss": 0.3775,
+      "step": 877
+    },
+    {
+      "epoch": 3.564336372847011,
+      "grad_norm": 0.24608580364638544,
+      "learning_rate": 9.176333983521291e-06,
+      "loss": 0.3836,
+      "step": 878
+    },
+    {
+      "epoch": 3.56838905775076,
+      "grad_norm": 0.26904768871843926,
+      "learning_rate": 9.12864899997558e-06,
+      "loss": 0.3985,
+      "step": 879
+    },
+    {
+      "epoch": 3.5724417426545085,
+      "grad_norm": 0.27128606048802917,
+      "learning_rate": 9.08105157279628e-06,
+      "loss": 0.3891,
+      "step": 880
+    },
+    {
+      "epoch": 3.5764944275582575,
+      "grad_norm": 0.2464368499810563,
+      "learning_rate": 9.03354208532653e-06,
+      "loss": 0.3944,
+      "step": 881
+    },
+    {
+      "epoch": 3.580547112462006,
+      "grad_norm": 0.26687798754210473,
+      "learning_rate": 8.986120920201205e-06,
+      "loss": 0.3631,
+      "step": 882
+    },
+    {
+      "epoch": 3.5845997973657546,
+      "grad_norm": 0.2868137283380602,
+      "learning_rate": 8.938788459343852e-06,
+      "loss": 0.3779,
+      "step": 883
+    },
+    {
+      "epoch": 3.5886524822695036,
+      "grad_norm": 0.2642451697752491,
+      "learning_rate": 8.8915450839636e-06,
+      "loss": 0.3791,
+      "step": 884
+    },
+    {
+      "epoch": 3.592705167173252,
+      "grad_norm": 0.27812817513330057,
+      "learning_rate": 8.844391174552116e-06,
+      "loss": 0.4046,
+      "step": 885
+    },
+    {
+      "epoch": 3.596757852077001,
+      "grad_norm": 0.2893191344552168,
+      "learning_rate": 8.797327110880479e-06,
+      "loss": 0.3684,
+      "step": 886
+    },
+    {
+      "epoch": 3.6008105369807497,
+      "grad_norm": 0.2641902188587831,
+      "learning_rate": 8.750353271996206e-06,
+      "loss": 0.3761,
+      "step": 887
+    },
+    {
+      "epoch": 3.6048632218844983,
+      "grad_norm": 0.27965262140526126,
+      "learning_rate": 8.703470036220132e-06,
+      "loss": 0.3831,
+      "step": 888
+    },
+    {
+      "epoch": 3.6089159067882473,
+      "grad_norm": 0.2551322147085001,
+      "learning_rate": 8.656677781143394e-06,
+      "loss": 0.3865,
+      "step": 889
+    },
+    {
+      "epoch": 3.612968591691996,
+      "grad_norm": 0.25594419535685065,
+      "learning_rate": 8.609976883624377e-06,
+      "loss": 0.3824,
+      "step": 890
+    },
+    {
+      "epoch": 3.617021276595745,
+      "grad_norm": 0.2668346673433008,
+      "learning_rate": 8.563367719785698e-06,
+      "loss": 0.3564,
+      "step": 891
+    },
+    {
+      "epoch": 3.6210739614994933,
+      "grad_norm": 0.26069517385786256,
+      "learning_rate": 8.516850665011138e-06,
+      "loss": 0.3855,
+      "step": 892
+    },
+    {
+      "epoch": 3.625126646403242,
+      "grad_norm": 0.25569320363985437,
+      "learning_rate": 8.47042609394269e-06,
+      "loss": 0.3772,
+      "step": 893
+    },
+    {
+      "epoch": 3.629179331306991,
+      "grad_norm": 0.258067276079996,
+      "learning_rate": 8.424094380477432e-06,
+      "loss": 0.3823,
+      "step": 894
+    },
+    {
+      "epoch": 3.63323201621074,
+      "grad_norm": 0.2620702666969095,
+      "learning_rate": 8.37785589776465e-06,
+      "loss": 0.3783,
+      "step": 895
+    },
+    {
+      "epoch": 3.6372847011144884,
+      "grad_norm": 0.2726967173128483,
+      "learning_rate": 8.331711018202694e-06,
+      "loss": 0.381,
+      "step": 896
+    },
+    {
+      "epoch": 3.641337386018237,
+      "grad_norm": 0.259983151457623,
+      "learning_rate": 8.285660113436104e-06,
+      "loss": 0.3681,
+      "step": 897
+    },
+    {
+      "epoch": 3.645390070921986,
+      "grad_norm": 0.2622312817717958,
+      "learning_rate": 8.239703554352527e-06,
+      "loss": 0.3651,
+      "step": 898
+    },
+    {
+      "epoch": 3.6494427558257345,
+      "grad_norm": 0.25590741385747956,
+      "learning_rate": 8.193841711079775e-06,
+      "loss": 0.387,
+      "step": 899
+    },
+    {
+      "epoch": 3.6534954407294835,
+      "grad_norm": 0.27101692134935057,
+      "learning_rate": 8.148074952982828e-06,
+      "loss": 0.3671,
+      "step": 900
+    },
+    {
+      "epoch": 3.657548125633232,
+      "grad_norm": 0.2560869396492808,
+      "learning_rate": 8.102403648660859e-06,
+      "loss": 0.3858,
+      "step": 901
+    },
+    {
+      "epoch": 3.6616008105369806,
+      "grad_norm": 0.23509852953511742,
+      "learning_rate": 8.056828165944282e-06,
+      "loss": 0.3926,
+      "step": 902
+    },
+    {
+      "epoch": 3.6656534954407296,
+      "grad_norm": 0.24745891186561988,
+      "learning_rate": 8.011348871891762e-06,
+      "loss": 0.3971,
+      "step": 903
+    },
+    {
+      "epoch": 3.669706180344478,
+      "grad_norm": 0.2581252689769421,
+      "learning_rate": 7.965966132787287e-06,
+      "loss": 0.3928,
+      "step": 904
+    },
+    {
+      "epoch": 3.673758865248227,
+      "grad_norm": 0.25265013418438403,
+      "learning_rate": 7.920680314137189e-06,
+      "loss": 0.3862,
+      "step": 905
+    },
+    {
+      "epoch": 3.6778115501519757,
+      "grad_norm": 0.23624572949864894,
+      "learning_rate": 7.875491780667246e-06,
+      "loss": 0.3991,
+      "step": 906
+    },
+    {
+      "epoch": 3.6818642350557242,
+      "grad_norm": 0.27279624118736484,
+      "learning_rate": 7.830400896319667e-06,
+      "loss": 0.3829,
+      "step": 907
+    },
+    {
+      "epoch": 3.6859169199594732,
+      "grad_norm": 0.25203918482922044,
+      "learning_rate": 7.785408024250259e-06,
+      "loss": 0.3906,
+      "step": 908
+    },
+    {
+      "epoch": 3.689969604863222,
+      "grad_norm": 0.2293814254120697,
+      "learning_rate": 7.74051352682542e-06,
+      "loss": 0.4032,
+      "step": 909
+    },
+    {
+      "epoch": 3.694022289766971,
+      "grad_norm": 0.23877643679843694,
+      "learning_rate": 7.695717765619257e-06,
+      "loss": 0.391,
+      "step": 910
+    },
+    {
+      "epoch": 3.6980749746707193,
+      "grad_norm": 0.26103122293520664,
+      "learning_rate": 7.651021101410673e-06,
+      "loss": 0.3932,
+      "step": 911
+    },
+    {
+      "epoch": 3.702127659574468,
+      "grad_norm": 0.2545919331953898,
+      "learning_rate": 7.606423894180464e-06,
+      "loss": 0.3661,
+      "step": 912
+    },
+    {
+      "epoch": 3.706180344478217,
+      "grad_norm": 0.25485257631189995,
+      "learning_rate": 7.56192650310839e-06,
+      "loss": 0.3705,
+      "step": 913
+    },
+    {
+      "epoch": 3.7102330293819654,
+      "grad_norm": 0.2791055521595466,
+      "learning_rate": 7.517529286570349e-06,
+      "loss": 0.3785,
+      "step": 914
+    },
+    {
+      "epoch": 3.7142857142857144,
+      "grad_norm": 0.26615395865428915,
+      "learning_rate": 7.473232602135387e-06,
+      "loss": 0.3818,
+      "step": 915
+    },
+    {
+      "epoch": 3.718338399189463,
+      "grad_norm": 0.25261794890827705,
+      "learning_rate": 7.429036806562935e-06,
+      "loss": 0.3696,
+      "step": 916
+    },
+    {
+      "epoch": 3.7223910840932115,
+      "grad_norm": 0.24877986054887413,
+      "learning_rate": 7.3849422557998455e-06,
+      "loss": 0.396,
+      "step": 917
+    },
+    {
+      "epoch": 3.7264437689969605,
+      "grad_norm": 0.25119564908315795,
+      "learning_rate": 7.340949304977567e-06,
+      "loss": 0.3691,
+      "step": 918
+    },
+    {
+      "epoch": 3.7304964539007095,
+      "grad_norm": 0.23963904655600143,
+      "learning_rate": 7.297058308409282e-06,
+      "loss": 0.3853,
+      "step": 919
+    },
+    {
+      "epoch": 3.734549138804458,
+      "grad_norm": 0.23874508630645142,
+      "learning_rate": 7.25326961958704e-06,
+      "loss": 0.3979,
+      "step": 920
+    },
+    {
+      "epoch": 3.7386018237082066,
+      "grad_norm": 0.2613251252562162,
+      "learning_rate": 7.209583591178921e-06,
+      "loss": 0.361,
+      "step": 921
+    },
+    {
+      "epoch": 3.742654508611955,
+      "grad_norm": 0.247763757883156,
+      "learning_rate": 7.1660005750261925e-06,
+      "loss": 0.3905,
+      "step": 922
+    },
+    {
+      "epoch": 3.746707193515704,
+      "grad_norm": 0.25433935321193435,
+      "learning_rate": 7.1225209221404765e-06,
+      "loss": 0.3587,
+      "step": 923
+    },
+    {
+      "epoch": 3.750759878419453,
+      "grad_norm": 0.24161679896860355,
+      "learning_rate": 7.079144982700909e-06,
+      "loss": 0.3794,
+      "step": 924
+    },
+    {
+      "epoch": 3.7548125633232017,
+      "grad_norm": 0.24676946599970728,
+      "learning_rate": 7.0358731060513695e-06,
+      "loss": 0.3831,
+      "step": 925
+    },
+    {
+      "epoch": 3.7588652482269502,
+      "grad_norm": 0.24864272551566333,
+      "learning_rate": 6.99270564069757e-06,
+      "loss": 0.3559,
+      "step": 926
+    },
+    {
+      "epoch": 3.762917933130699,
+      "grad_norm": 0.2721763681367414,
+      "learning_rate": 6.949642934304375e-06,
+      "loss": 0.3794,
+      "step": 927
+    },
+    {
+      "epoch": 3.7669706180344478,
+      "grad_norm": 0.2555495781313021,
+      "learning_rate": 6.906685333692871e-06,
+      "loss": 0.3742,
+      "step": 928
+    },
+    {
+      "epoch": 3.7710233029381968,
+      "grad_norm": 0.24024176761244234,
+      "learning_rate": 6.86383318483769e-06,
+      "loss": 0.3747,
+      "step": 929
+    },
+    {
+      "epoch": 3.7750759878419453,
+      "grad_norm": 0.24739244511136232,
+      "learning_rate": 6.821086832864139e-06,
+      "loss": 0.3876,
+      "step": 930
+    },
+    {
+      "epoch": 3.779128672745694,
+      "grad_norm": 0.25229345464321506,
+      "learning_rate": 6.77844662204546e-06,
+      "loss": 0.399,
+      "step": 931
+    },
+    {
+      "epoch": 3.783181357649443,
+      "grad_norm": 0.23944201751332087,
+      "learning_rate": 6.7359128958000455e-06,
+      "loss": 0.4029,
+      "step": 932
+    },
+    {
+      "epoch": 3.7872340425531914,
+      "grad_norm": 0.25122823970264063,
+      "learning_rate": 6.693485996688695e-06,
+      "loss": 0.3698,
+      "step": 933
+    },
+    {
+      "epoch": 3.7912867274569404,
+      "grad_norm": 0.25961657067824834,
+      "learning_rate": 6.651166266411801e-06,
+      "loss": 0.3811,
+      "step": 934
+    },
+    {
+      "epoch": 3.795339412360689,
+      "grad_norm": 0.2502513717675922,
+      "learning_rate": 6.6089540458066725e-06,
+      "loss": 0.3864,
+      "step": 935
+    },
+    {
+      "epoch": 3.7993920972644375,
+      "grad_norm": 0.2610775403663525,
+      "learning_rate": 6.566849674844711e-06,
+      "loss": 0.3712,
+      "step": 936
+    },
+    {
+      "epoch": 3.8034447821681865,
+      "grad_norm": 0.247743927908705,
+      "learning_rate": 6.524853492628747e-06,
+      "loss": 0.3706,
+      "step": 937
+    },
+    {
+      "epoch": 3.807497467071935,
+      "grad_norm": 0.23974356395992896,
+      "learning_rate": 6.4829658373902536e-06,
+      "loss": 0.4033,
+      "step": 938
+    },
+    {
+      "epoch": 3.811550151975684,
+      "grad_norm": 0.25639644385453203,
+      "learning_rate": 6.441187046486648e-06,
+      "loss": 0.3845,
+      "step": 939
+    },
+    {
+      "epoch": 3.8156028368794326,
+      "grad_norm": 0.24478652216578706,
+      "learning_rate": 6.399517456398567e-06,
+      "loss": 0.384,
+      "step": 940
+    },
+    {
+      "epoch": 3.819655521783181,
+      "grad_norm": 0.24102927962772855,
+      "learning_rate": 6.357957402727164e-06,
+      "loss": 0.3852,
+      "step": 941
+    },
+    {
+      "epoch": 3.82370820668693,
+      "grad_norm": 0.24609536521299766,
+      "learning_rate": 6.316507220191395e-06,
+      "loss": 0.3911,
+      "step": 942
+    },
+    {
+      "epoch": 3.8277608915906787,
+      "grad_norm": 0.24020736597303688,
+      "learning_rate": 6.275167242625331e-06,
+      "loss": 0.3907,
+      "step": 943
+    },
+    {
+      "epoch": 3.8318135764944277,
+      "grad_norm": 0.2575315279870402,
+      "learning_rate": 6.233937802975471e-06,
+      "loss": 0.3933,
+      "step": 944
+    },
+    {
+      "epoch": 3.835866261398176,
+      "grad_norm": 0.25265995274165626,
+      "learning_rate": 6.192819233298046e-06,
+      "loss": 0.3734,
+      "step": 945
+    },
+    {
+      "epoch": 3.8399189463019248,
+      "grad_norm": 0.24215442344831176,
+      "learning_rate": 6.151811864756383e-06,
+      "loss": 0.3716,
+      "step": 946
+    },
+    {
+      "epoch": 3.8439716312056738,
+      "grad_norm": 0.253408968259172,
+      "learning_rate": 6.1109160276181655e-06,
+      "loss": 0.396,
+      "step": 947
+    },
+    {
+      "epoch": 3.8480243161094227,
+      "grad_norm": 1.3659760912520276,
+      "learning_rate": 6.070132051252868e-06,
+      "loss": 0.4565,
+      "step": 948
+    },
+    {
+      "epoch": 3.8520770010131713,
+      "grad_norm": 0.23365812349180248,
+      "learning_rate": 6.0294602641290034e-06,
+      "loss": 0.3779,
+      "step": 949
+    },
+    {
+      "epoch": 3.85612968591692,
+      "grad_norm": 0.23662609605247903,
+      "learning_rate": 5.988900993811575e-06,
+      "loss": 0.3765,
+      "step": 950
+    },
+    {
+      "epoch": 3.860182370820669,
+      "grad_norm": 0.2438744097200581,
+      "learning_rate": 5.948454566959363e-06,
+      "loss": 0.3833,
+      "step": 951
+    },
+    {
+      "epoch": 3.8642350557244174,
+      "grad_norm": 0.24384298517536226,
+      "learning_rate": 5.908121309322328e-06,
+      "loss": 0.3727,
+      "step": 952
+    },
+    {
+      "epoch": 3.8682877406281664,
+      "grad_norm": 0.23522797137541018,
+      "learning_rate": 5.867901545738976e-06,
+      "loss": 0.3895,
+      "step": 953
+    },
+    {
+      "epoch": 3.872340425531915,
+      "grad_norm": 0.2428423246864904,
+      "learning_rate": 5.827795600133774e-06,
+      "loss": 0.3717,
+      "step": 954
+    },
+    {
+      "epoch": 3.8763931104356635,
+      "grad_norm": 0.23184907736324284,
+      "learning_rate": 5.787803795514466e-06,
+      "loss": 0.3722,
+      "step": 955
+    },
+    {
+      "epoch": 3.8804457953394125,
+      "grad_norm": 0.22695505517851844,
+      "learning_rate": 5.747926453969576e-06,
+      "loss": 0.3944,
+      "step": 956
+    },
+    {
+      "epoch": 3.884498480243161,
+      "grad_norm": 0.25622634926368437,
+      "learning_rate": 5.708163896665708e-06,
+      "loss": 0.3698,
+      "step": 957
+    },
+    {
+      "epoch": 3.88855116514691,
+      "grad_norm": 0.2330703813197096,
+      "learning_rate": 5.668516443845047e-06,
+      "loss": 0.3787,
+      "step": 958
+    },
+    {
+      "epoch": 3.8926038500506586,
+      "grad_norm": 0.2453268239202265,
+      "learning_rate": 5.6289844148227225e-06,
+      "loss": 0.3632,
+      "step": 959
+    },
+    {
+      "epoch": 3.896656534954407,
+      "grad_norm": 0.26220693956586866,
+      "learning_rate": 5.5895681279842615e-06,
+      "loss": 0.377,
+      "step": 960
+    },
+    {
+      "epoch": 3.900709219858156,
+      "grad_norm": 0.24331804630697973,
+      "learning_rate": 5.550267900783019e-06,
+      "loss": 0.3718,
+      "step": 961
+    },
+    {
+      "epoch": 3.9047619047619047,
+      "grad_norm": 0.22770359856100209,
+      "learning_rate": 5.511084049737623e-06,
+      "loss": 0.3798,
+      "step": 962
+    },
+    {
+      "epoch": 3.9088145896656536,
+      "grad_norm": 0.2428561035827692,
+      "learning_rate": 5.4720168904294215e-06,
+      "loss": 0.3854,
+      "step": 963
+    },
+    {
+      "epoch": 3.912867274569402,
+      "grad_norm": 0.2611971830968991,
+      "learning_rate": 5.433066737499948e-06,
+      "loss": 0.3769,
+      "step": 964
+    },
+    {
+      "epoch": 3.9169199594731507,
+      "grad_norm": 0.2406934295793003,
+      "learning_rate": 5.394233904648376e-06,
+      "loss": 0.3654,
+      "step": 965
+    },
+    {
+      "epoch": 3.9209726443768997,
+      "grad_norm": 0.22090493999761748,
+      "learning_rate": 5.355518704628997e-06,
+      "loss": 0.3815,
+      "step": 966
+    },
+    {
+      "epoch": 3.9250253292806483,
+      "grad_norm": 0.24259711573794973,
+      "learning_rate": 5.316921449248731e-06,
+      "loss": 0.3785,
+      "step": 967
+    },
+    {
+      "epoch": 3.9290780141843973,
+      "grad_norm": 1.5430490430588688,
+      "learning_rate": 5.278442449364538e-06,
+      "loss": 0.383,
+      "step": 968
+    },
+    {
+      "epoch": 3.933130699088146,
+      "grad_norm": 0.24031238951607808,
+      "learning_rate": 5.240082014881016e-06,
+      "loss": 0.3976,
+      "step": 969
+    },
+    {
+      "epoch": 3.9371833839918944,
+      "grad_norm": 0.2437610986619112,
+      "learning_rate": 5.201840454747822e-06,
+      "loss": 0.3696,
+      "step": 970
+    },
+    {
+      "epoch": 3.9412360688956434,
+      "grad_norm": 0.22082397830342182,
+      "learning_rate": 5.163718076957223e-06,
+      "loss": 0.397,
+      "step": 971
+    },
+    {
+      "epoch": 3.9452887537993924,
+      "grad_norm": 0.300049065779442,
+      "learning_rate": 5.125715188541609e-06,
+      "loss": 0.3745,
+      "step": 972
+    },
+    {
+      "epoch": 3.949341438703141,
+      "grad_norm": 0.24960191240344437,
+      "learning_rate": 5.087832095571021e-06,
+      "loss": 0.3595,
+      "step": 973
+    },
+    {
+      "epoch": 3.9533941236068895,
+      "grad_norm": 0.23260955229232055,
+      "learning_rate": 5.0500691031506766e-06,
+      "loss": 0.3854,
+      "step": 974
+    },
+    {
+      "epoch": 3.9574468085106385,
+      "grad_norm": 0.2347783236855166,
+      "learning_rate": 5.01242651541854e-06,
+      "loss": 0.3778,
+      "step": 975
+    },
+    {
+      "epoch": 3.961499493414387,
+      "grad_norm": 0.24120605951175264,
+      "learning_rate": 4.974904635542815e-06,
+      "loss": 0.3767,
+      "step": 976
+    },
+    {
+      "epoch": 3.965552178318136,
+      "grad_norm": 0.2562211408097052,
+      "learning_rate": 4.937503765719582e-06,
+      "loss": 0.3729,
+      "step": 977
+    },
+    {
+      "epoch": 3.9696048632218845,
+      "grad_norm": 0.2291642976443571,
+      "learning_rate": 4.900224207170299e-06,
+      "loss": 0.3786,
+      "step": 978
+    },
+    {
+      "epoch": 3.973657548125633,
+      "grad_norm": 0.24761238539181818,
+      "learning_rate": 4.8630662601394065e-06,
+      "loss": 0.3981,
+      "step": 979
+    },
+    {
+      "epoch": 3.977710233029382,
+      "grad_norm": 0.2307968561015442,
+      "learning_rate": 4.8260302238918995e-06,
+      "loss": 0.3677,
+      "step": 980
+    },
+    {
+      "epoch": 3.9817629179331306,
+      "grad_norm": 0.22523116284897532,
+      "learning_rate": 4.789116396710924e-06,
+      "loss": 0.3849,
+      "step": 981
+    },
+    {
+      "epoch": 3.9858156028368796,
+      "grad_norm": 0.23670569985444856,
+      "learning_rate": 4.752325075895368e-06,
+      "loss": 0.3701,
+      "step": 982
+    },
+    {
+      "epoch": 3.989868287740628,
+      "grad_norm": 0.2470731180414348,
+      "learning_rate": 4.715656557757473e-06,
+      "loss": 0.3654,
+      "step": 983
+    },
+    {
+      "epoch": 3.9939209726443767,
+      "grad_norm": 0.6229189248354251,
+      "learning_rate": 4.679111137620442e-06,
+      "loss": 0.397,
+      "step": 984
+    },
+    {
+      "epoch": 3.9979736575481257,
+      "grad_norm": 0.2460543297899704,
+      "learning_rate": 4.6426891098160585e-06,
+      "loss": 0.3777,
+      "step": 985
+    },
+    {
+      "epoch": 4.0040526849037485,
+      "grad_norm": 0.43441965883150363,
+      "learning_rate": 4.6063907676823474e-06,
+      "loss": 0.3207,
+      "step": 986
+    },
+    {
+      "epoch": 4.008105369807497,
+      "grad_norm": 0.3172185792272455,
+      "learning_rate": 4.570216403561141e-06,
+      "loss": 0.3441,
+      "step": 987
+    },
+    {
+      "epoch": 4.0121580547112465,
+      "grad_norm": 0.2559676560011084,
+      "learning_rate": 4.534166308795815e-06,
+      "loss": 0.361,
+      "step": 988
+    },
+    {
+      "epoch": 4.016210739614995,
+      "grad_norm": 0.5003179698215972,
+      "learning_rate": 4.498240773728859e-06,
+      "loss": 0.3313,
+      "step": 989
+    },
+    {
+      "epoch": 4.020263424518744,
+      "grad_norm": 0.39127556056111495,
+      "learning_rate": 4.462440087699609e-06,
+      "loss": 0.3437,
+      "step": 990
+    },
+    {
+      "epoch": 4.024316109422492,
+      "grad_norm": 0.2923205642374672,
+      "learning_rate": 4.426764539041861e-06,
+      "loss": 0.3312,
+      "step": 991
+    },
+    {
+      "epoch": 4.028368794326241,
+      "grad_norm": 0.34303442581156296,
+      "learning_rate": 4.391214415081582e-06,
+      "loss": 0.3339,
+      "step": 992
+    },
+    {
+      "epoch": 4.03242147922999,
+      "grad_norm": 0.37236892902762114,
+      "learning_rate": 4.355790002134579e-06,
+      "loss": 0.3317,
+      "step": 993
+    },
+    {
+      "epoch": 4.036474164133739,
+      "grad_norm": 0.29381660233429446,
+      "learning_rate": 4.320491585504207e-06,
+      "loss": 0.347,
+      "step": 994
+    },
+    {
+      "epoch": 4.040526849037487,
+      "grad_norm": 0.2828414471885096,
+      "learning_rate": 4.2853194494790615e-06,
+      "loss": 0.3428,
+      "step": 995
+    },
+    {
+      "epoch": 4.044579533941236,
+      "grad_norm": 0.3059883256410596,
+      "learning_rate": 4.250273877330691e-06,
+      "loss": 0.3205,
+      "step": 996
+    },
+    {
+      "epoch": 4.048632218844984,
+      "grad_norm": 0.35630036602168547,
+      "learning_rate": 4.215355151311313e-06,
+      "loss": 0.3457,
+      "step": 997
+    },
+    {
+      "epoch": 4.052684903748734,
+      "grad_norm": 0.3391856526917267,
+      "learning_rate": 4.180563552651542e-06,
+      "loss": 0.3428,
+      "step": 998
+    },
+    {
+      "epoch": 4.056737588652482,
+      "grad_norm": 0.26573757236833684,
+      "learning_rate": 4.145899361558147e-06,
+      "loss": 0.339,
+      "step": 999
+    },
+    {
+      "epoch": 4.060790273556231,
+      "grad_norm": 0.2637561449078847,
+      "learning_rate": 4.111362857211738e-06,
+      "loss": 0.3496,
+      "step": 1000
+    },
+    {
+      "epoch": 4.0648429584599794,
+      "grad_norm": 0.3202643730230884,
+      "learning_rate": 4.076954317764592e-06,
+      "loss": 0.3259,
+      "step": 1001
+    },
+    {
+      "epoch": 4.068895643363729,
+      "grad_norm": 0.2859231904760011,
+      "learning_rate": 4.042674020338335e-06,
+      "loss": 0.3594,
+      "step": 1002
+    },
+    {
+      "epoch": 4.072948328267477,
+      "grad_norm": 0.2655163295044421,
+      "learning_rate": 4.0085222410217835e-06,
+      "loss": 0.3356,
+      "step": 1003
+    },
+    {
+      "epoch": 4.077001013171226,
+      "grad_norm": 0.2567605286085258,
+      "learning_rate": 3.974499254868674e-06,
+      "loss": 0.3365,
+      "step": 1004
+    },
+    {
+      "epoch": 4.0810536980749745,
+      "grad_norm": 0.29881391994396056,
+      "learning_rate": 3.940605335895451e-06,
+      "loss": 0.3191,
+      "step": 1005
+    },
+    {
+      "epoch": 4.085106382978723,
+      "grad_norm": 0.28317244235385675,
+      "learning_rate": 3.90684075707908e-06,
+      "loss": 0.3338,
+      "step": 1006
+    },
+    {
+      "epoch": 4.0891590678824725,
+      "grad_norm": 0.2727772100652722,
+      "learning_rate": 3.8732057903548505e-06,
+      "loss": 0.337,
+      "step": 1007
+    },
+    {
+      "epoch": 4.093211752786221,
+      "grad_norm": 0.2411091975837039,
+      "learning_rate": 3.8397007066141375e-06,
+      "loss": 0.3317,
+      "step": 1008
+    },
+    {
+      "epoch": 4.09726443768997,
+      "grad_norm": 0.39773060601877774,
+      "learning_rate": 3.806325775702304e-06,
+      "loss": 0.3216,
+      "step": 1009
+    },
+    {
+      "epoch": 4.101317122593718,
+      "grad_norm": 0.23294668790624837,
+      "learning_rate": 3.773081266416434e-06,
+      "loss": 0.3432,
+      "step": 1010
+    },
+    {
+      "epoch": 4.105369807497467,
+      "grad_norm": 0.25048276745175535,
+      "learning_rate": 3.739967446503245e-06,
+      "loss": 0.3537,
+      "step": 1011
+    },
+    {
+      "epoch": 4.109422492401216,
+      "grad_norm": 0.23963348940540824,
+      "learning_rate": 3.706984582656894e-06,
+      "loss": 0.3502,
+      "step": 1012
+    },
+    {
+      "epoch": 4.113475177304965,
+      "grad_norm": 0.23162128480111757,
+      "learning_rate": 3.6741329405168237e-06,
+      "loss": 0.3424,
+      "step": 1013
+    },
+    {
+      "epoch": 4.117527862208713,
+      "grad_norm": 0.2714004979790431,
+      "learning_rate": 3.641412784665648e-06,
+      "loss": 0.3214,
+      "step": 1014
+    },
+    {
+      "epoch": 4.121580547112462,
+      "grad_norm": 0.2446202152678492,
+      "learning_rate": 3.608824378627005e-06,
+      "loss": 0.3632,
+      "step": 1015
+    },
+    {
+      "epoch": 4.12563323201621,
+      "grad_norm": 0.23320102692982828,
+      "learning_rate": 3.5763679848634337e-06,
+      "loss": 0.3268,
+      "step": 1016
+    },
+    {
+      "epoch": 4.12968591691996,
+      "grad_norm": 0.22971961010867067,
+      "learning_rate": 3.544043864774269e-06,
+      "loss": 0.3146,
+      "step": 1017
+    },
+    {
+      "epoch": 4.133738601823708,
+      "grad_norm": 0.2393940157111149,
+      "learning_rate": 3.5118522786935282e-06,
+      "loss": 0.3218,
+      "step": 1018
+    },
+    {
+      "epoch": 4.137791286727457,
+      "grad_norm": 0.2366905046775755,
+      "learning_rate": 3.479793485887819e-06,
+      "loss": 0.3285,
+      "step": 1019
+    },
+    {
+      "epoch": 4.141843971631205,
+      "grad_norm": 0.23327188316325834,
+      "learning_rate": 3.4478677445542653e-06,
+      "loss": 0.346,
+      "step": 1020
+    },
+    {
+      "epoch": 4.145896656534954,
+      "grad_norm": 0.23805694262971164,
+      "learning_rate": 3.4160753118183767e-06,
+      "loss": 0.3201,
+      "step": 1021
+    },
+    {
+      "epoch": 4.149949341438703,
+      "grad_norm": 0.21952183222157157,
+      "learning_rate": 3.3844164437320527e-06,
+      "loss": 0.3558,
+      "step": 1022
+    },
+    {
+      "epoch": 4.154002026342452,
+      "grad_norm": 0.2367399283120628,
+      "learning_rate": 3.3528913952714558e-06,
+      "loss": 0.3375,
+      "step": 1023
+    },
+    {
+      "epoch": 4.1580547112462005,
+      "grad_norm": 0.24076807751844387,
+      "learning_rate": 3.321500420335e-06,
+      "loss": 0.3581,
+      "step": 1024
+    },
+    {
+      "epoch": 4.162107396149949,
+      "grad_norm": 0.2361755515108673,
+      "learning_rate": 3.290243771741275e-06,
+      "loss": 0.3234,
+      "step": 1025
+    },
+    {
+      "epoch": 4.1661600810536985,
+      "grad_norm": 0.22473654665646817,
+      "learning_rate": 3.2591217012270325e-06,
+      "loss": 0.3413,
+      "step": 1026
+    },
+    {
+      "epoch": 4.170212765957447,
+      "grad_norm": 0.22618701949927464,
+      "learning_rate": 3.228134459445149e-06,
+      "loss": 0.3442,
+      "step": 1027
+    },
+    {
+      "epoch": 4.174265450861196,
+      "grad_norm": 0.229825739318835,
+      "learning_rate": 3.1972822959626205e-06,
+      "loss": 0.3377,
+      "step": 1028
+    },
+    {
+      "epoch": 4.178318135764944,
+      "grad_norm": 0.238151569893642,
+      "learning_rate": 3.166565459258513e-06,
+      "loss": 0.33,
+      "step": 1029
+    },
+    {
+      "epoch": 4.182370820668693,
+      "grad_norm": 0.22754261204629825,
+      "learning_rate": 3.1359841967220193e-06,
+      "loss": 0.3252,
+      "step": 1030
+    },
+    {
+      "epoch": 4.186423505572442,
+      "grad_norm": 0.21644352401658654,
+      "learning_rate": 3.105538754650419e-06,
+      "loss": 0.3467,
+      "step": 1031
+    },
+    {
+      "epoch": 4.190476190476191,
+      "grad_norm": 0.2222016041322117,
+      "learning_rate": 3.07522937824712e-06,
+      "loss": 0.3338,
+      "step": 1032
+    },
+    {
+      "epoch": 4.194528875379939,
+      "grad_norm": 0.23395716092818578,
+      "learning_rate": 3.0450563116196697e-06,
+      "loss": 0.3178,
+      "step": 1033
+    },
+    {
+      "epoch": 4.198581560283688,
+      "grad_norm": 0.22676626549455461,
+      "learning_rate": 3.0150197977778008e-06,
+      "loss": 0.3508,
+      "step": 1034
+    },
+    {
+      "epoch": 4.202634245187436,
+      "grad_norm": 0.22253048027652203,
+      "learning_rate": 2.985120078631465e-06,
+      "loss": 0.3212,
+      "step": 1035
+    },
+    {
+      "epoch": 4.206686930091186,
+      "grad_norm": 0.22580895403178536,
+      "learning_rate": 2.9553573949888893e-06,
+      "loss": 0.3391,
+      "step": 1036
+    },
+    {
+      "epoch": 4.210739614994934,
+      "grad_norm": 0.22623805753396767,
+      "learning_rate": 2.9257319865546384e-06,
+      "loss": 0.3543,
+      "step": 1037
+    },
+    {
+      "epoch": 4.214792299898683,
+      "grad_norm": 0.23771752218328732,
+      "learning_rate": 2.896244091927678e-06,
+      "loss": 0.3334,
+      "step": 1038
+    },
+    {
+      "epoch": 4.218844984802431,
+      "grad_norm": 0.2175923691488866,
+      "learning_rate": 2.8668939485994584e-06,
+      "loss": 0.3413,
+      "step": 1039
+    },
+    {
+      "epoch": 4.22289766970618,
+      "grad_norm": 0.22294735532990967,
+      "learning_rate": 2.837681792951994e-06,
+      "loss": 0.3498,
+      "step": 1040
+    },
+    {
+      "epoch": 4.226950354609929,
+      "grad_norm": 0.22575799265879024,
+      "learning_rate": 2.808607860255981e-06,
+      "loss": 0.356,
+      "step": 1041
+    },
+    {
+      "epoch": 4.231003039513678,
+      "grad_norm": 0.22390186791509342,
+      "learning_rate": 2.7796723846688634e-06,
+      "loss": 0.3464,
+      "step": 1042
+    },
+    {
+      "epoch": 4.2350557244174265,
+      "grad_norm": 0.2372253372859002,
+      "learning_rate": 2.7508755992329937e-06,
+      "loss": 0.333,
+      "step": 1043
+    },
+    {
+      "epoch": 4.239108409321175,
+      "grad_norm": 0.23263498450222925,
+      "learning_rate": 2.722217735873718e-06,
+      "loss": 0.3196,
+      "step": 1044
+    },
+    {
+      "epoch": 4.243161094224924,
+      "grad_norm": 0.22082644683840363,
+      "learning_rate": 2.6936990253975315e-06,
+      "loss": 0.3429,
+      "step": 1045
+    },
+    {
+      "epoch": 4.247213779128673,
+      "grad_norm": 0.21254303248638728,
+      "learning_rate": 2.665319697490205e-06,
+      "loss": 0.3398,
+      "step": 1046
+    },
+    {
+      "epoch": 4.251266464032422,
+      "grad_norm": 0.2243713817826788,
+      "learning_rate": 2.637079980714945e-06,
+      "loss": 0.3514,
+      "step": 1047
+    },
+    {
+      "epoch": 4.25531914893617,
+      "grad_norm": 0.23194174383328658,
+      "learning_rate": 2.6089801025105453e-06,
+      "loss": 0.3139,
+      "step": 1048
+    },
+    {
+      "epoch": 4.259371833839919,
+      "grad_norm": 0.21504063264920145,
+      "learning_rate": 2.581020289189571e-06,
+      "loss": 0.3592,
+      "step": 1049
+    },
+    {
+      "epoch": 4.263424518743667,
+      "grad_norm": 0.22028653603149462,
+      "learning_rate": 2.553200765936501e-06,
+      "loss": 0.3487,
+      "step": 1050
+    },
+    {
+      "epoch": 4.267477203647417,
+      "grad_norm": 0.21704198667745012,
+      "learning_rate": 2.525521756805962e-06,
+      "loss": 0.3417,
+      "step": 1051
+    },
+    {
+      "epoch": 4.271529888551165,
+      "grad_norm": 0.23369822909954296,
+      "learning_rate": 2.497983484720885e-06,
+      "loss": 0.3195,
+      "step": 1052
+    },
+    {
+      "epoch": 4.275582573454914,
+      "grad_norm": 0.22063242924242107,
+      "learning_rate": 2.470586171470728e-06,
+      "loss": 0.3279,
+      "step": 1053
+    },
+    {
+      "epoch": 4.279635258358662,
+      "grad_norm": 0.22141494083722552,
+      "learning_rate": 2.4433300377096836e-06,
+      "loss": 0.3405,
+      "step": 1054
+    },
+    {
+      "epoch": 4.283687943262412,
+      "grad_norm": 0.219540341440935,
+      "learning_rate": 2.4162153029549073e-06,
+      "loss": 0.3408,
+      "step": 1055
+    },
+    {
+      "epoch": 4.28774062816616,
+      "grad_norm": 0.23861957890867846,
+      "learning_rate": 2.3892421855847458e-06,
+      "loss": 0.3333,
+      "step": 1056
+    },
+    {
+      "epoch": 4.291793313069909,
+      "grad_norm": 0.21548248088429095,
+      "learning_rate": 2.362410902836978e-06,
+      "loss": 0.3426,
+      "step": 1057
+    },
+    {
+      "epoch": 4.295845997973657,
+      "grad_norm": 0.24508055932875364,
+      "learning_rate": 2.3357216708070653e-06,
+      "loss": 0.3153,
+      "step": 1058
+    },
+    {
+      "epoch": 4.299898682877406,
+      "grad_norm": 0.2306722897699338,
+      "learning_rate": 2.309174704446411e-06,
+      "loss": 0.3208,
+      "step": 1059
+    },
+    {
+      "epoch": 4.303951367781155,
+      "grad_norm": 0.23510473888692848,
+      "learning_rate": 2.2827702175606437e-06,
+      "loss": 0.3336,
+      "step": 1060
+    },
+    {
+      "epoch": 4.308004052684904,
+      "grad_norm": 0.22997236852861927,
+      "learning_rate": 2.256508422807855e-06,
+      "loss": 0.3411,
+      "step": 1061
+    },
+    {
+      "epoch": 4.3120567375886525,
+      "grad_norm": 0.2253232394674051,
+      "learning_rate": 2.230389531696946e-06,
+      "loss": 0.3369,
+      "step": 1062
+    },
+    {
+      "epoch": 4.316109422492401,
+      "grad_norm": 0.2143512913097352,
+      "learning_rate": 2.204413754585857e-06,
+      "loss": 0.3418,
+      "step": 1063
+    },
+    {
+      "epoch": 4.32016210739615,
+      "grad_norm": 0.22817923766265974,
+      "learning_rate": 2.1785813006799406e-06,
+      "loss": 0.3207,
+      "step": 1064
+    },
+    {
+      "epoch": 4.324214792299899,
+      "grad_norm": 0.23139168272181188,
+      "learning_rate": 2.1528923780302224e-06,
+      "loss": 0.3216,
+      "step": 1065
+    },
+    {
+      "epoch": 4.328267477203648,
+      "grad_norm": 0.23507424193237603,
+      "learning_rate": 2.127347193531757e-06,
+      "loss": 0.3515,
+      "step": 1066
+    },
+    {
+      "epoch": 4.332320162107396,
+      "grad_norm": 0.23575923810648702,
+      "learning_rate": 2.101945952921942e-06,
+      "loss": 0.3274,
+      "step": 1067
+    },
+    {
+      "epoch": 4.336372847011145,
+      "grad_norm": 0.22545259489057787,
+      "learning_rate": 2.0766888607788906e-06,
+      "loss": 0.312,
+      "step": 1068
+    },
+    {
+      "epoch": 4.340425531914893,
+      "grad_norm": 0.23495567380127616,
+      "learning_rate": 2.0515761205197337e-06,
+      "loss": 0.3323,
+      "step": 1069
+    },
+    {
+      "epoch": 4.344478216818643,
+      "grad_norm": 0.22098817493937745,
+      "learning_rate": 2.0266079343990453e-06,
+      "loss": 0.3464,
+      "step": 1070
+    },
+    {
+      "epoch": 4.348530901722391,
+      "grad_norm": 0.21753324625510542,
+      "learning_rate": 2.0017845035071494e-06,
+      "loss": 0.3274,
+      "step": 1071
+    },
+    {
+      "epoch": 4.35258358662614,
+      "grad_norm": 0.22625446678673608,
+      "learning_rate": 1.9771060277685537e-06,
+      "loss": 0.3154,
+      "step": 1072
+    },
+    {
+      "epoch": 4.356636271529888,
+      "grad_norm": 0.23500122248011943,
+      "learning_rate": 1.95257270594031e-06,
+      "loss": 0.3405,
+      "step": 1073
+    },
+    {
+      "epoch": 4.360688956433638,
+      "grad_norm": 0.2386139808225573,
+      "learning_rate": 1.9281847356104188e-06,
+      "loss": 0.33,
+      "step": 1074
+    },
+    {
+      "epoch": 4.364741641337386,
+      "grad_norm": 0.23769347511234643,
+      "learning_rate": 1.9039423131962365e-06,
+      "loss": 0.3398,
+      "step": 1075
+    },
+    {
+      "epoch": 4.368794326241135,
+      "grad_norm": 0.22501807224528134,
+      "learning_rate": 1.8798456339429027e-06,
+      "loss": 0.3356,
+      "step": 1076
+    },
+    {
+      "epoch": 4.372847011144883,
+      "grad_norm": 0.22216700448089727,
+      "learning_rate": 1.8558948919217612e-06,
+      "loss": 0.3441,
+      "step": 1077
+    },
+    {
+      "epoch": 4.376899696048632,
+      "grad_norm": 0.22077620576922963,
+      "learning_rate": 1.8320902800287954e-06,
+      "loss": 0.3368,
+      "step": 1078
+    },
+    {
+      "epoch": 4.380952380952381,
+      "grad_norm": 0.21975774334133966,
+      "learning_rate": 1.8084319899830726e-06,
+      "loss": 0.3422,
+      "step": 1079
+    },
+    {
+      "epoch": 4.38500506585613,
+      "grad_norm": 0.2427476217633319,
+      "learning_rate": 1.7849202123252097e-06,
+      "loss": 0.3472,
+      "step": 1080
+    },
+    {
+      "epoch": 4.3890577507598785,
+      "grad_norm": 0.2238976923066399,
+      "learning_rate": 1.7615551364158401e-06,
+      "loss": 0.3311,
+      "step": 1081
+    },
+    {
+      "epoch": 4.393110435663627,
+      "grad_norm": 0.22245634423159233,
+      "learning_rate": 1.738336950434061e-06,
+      "loss": 0.3425,
+      "step": 1082
+    },
+    {
+      "epoch": 4.397163120567376,
+      "grad_norm": 0.2291645134335145,
+      "learning_rate": 1.715265841375957e-06,
+      "loss": 0.3182,
+      "step": 1083
+    },
+    {
+      "epoch": 4.401215805471125,
+      "grad_norm": 0.2207581523272618,
+      "learning_rate": 1.6923419950530684e-06,
+      "loss": 0.3351,
+      "step": 1084
+    },
+    {
+      "epoch": 4.405268490374874,
+      "grad_norm": 0.23169198562237595,
+      "learning_rate": 1.6695655960909008e-06,
+      "loss": 0.3038,
+      "step": 1085
+    },
+    {
+      "epoch": 4.409321175278622,
+      "grad_norm": 0.22495310274089256,
+      "learning_rate": 1.646936827927441e-06,
+      "loss": 0.3411,
+      "step": 1086
+    },
+    {
+      "epoch": 4.413373860182371,
+      "grad_norm": 0.22847221895158423,
+      "learning_rate": 1.6244558728116766e-06,
+      "loss": 0.333,
+      "step": 1087
+    },
+    {
+      "epoch": 4.417426545086119,
+      "grad_norm": 0.22898838199855842,
+      "learning_rate": 1.6021229118021265e-06,
+      "loss": 0.3323,
+      "step": 1088
+    },
+    {
+      "epoch": 4.421479229989869,
+      "grad_norm": 0.2216867202534966,
+      "learning_rate": 1.5799381247653967e-06,
+      "loss": 0.3372,
+      "step": 1089
+    },
+    {
+      "epoch": 4.425531914893617,
+      "grad_norm": 0.22599597212969008,
+      "learning_rate": 1.5579016903747013e-06,
+      "loss": 0.346,
+      "step": 1090
+    },
+    {
+      "epoch": 4.429584599797366,
+      "grad_norm": 0.23297686867498105,
+      "learning_rate": 1.5360137861084656e-06,
+      "loss": 0.3392,
+      "step": 1091
+    },
+    {
+      "epoch": 4.433637284701114,
+      "grad_norm": 0.23207674307321688,
+      "learning_rate": 1.5142745882488475e-06,
+      "loss": 0.3423,
+      "step": 1092
+    },
+    {
+      "epoch": 4.437689969604863,
+      "grad_norm": 0.2332743630255867,
+      "learning_rate": 1.4926842718803691e-06,
+      "loss": 0.3412,
+      "step": 1093
+    },
+    {
+      "epoch": 4.441742654508612,
+      "grad_norm": 0.2168993664129981,
+      "learning_rate": 1.4712430108884657e-06,
+      "loss": 0.3533,
+      "step": 1094
+    },
+    {
+      "epoch": 4.445795339412361,
+      "grad_norm": 0.2101188856783912,
+      "learning_rate": 1.4499509779581078e-06,
+      "loss": 0.3349,
+      "step": 1095
+    },
+    {
+      "epoch": 4.449848024316109,
+      "grad_norm": 0.2194823348902552,
+      "learning_rate": 1.4288083445723988e-06,
+      "loss": 0.3084,
+      "step": 1096
+    },
+    {
+      "epoch": 4.453900709219858,
+      "grad_norm": 0.23080580647339557,
+      "learning_rate": 1.4078152810112045e-06,
+      "loss": 0.3266,
+      "step": 1097
+    },
+    {
+      "epoch": 4.4579533941236065,
+      "grad_norm": 0.21833585919702747,
+      "learning_rate": 1.3869719563497697e-06,
+      "loss": 0.3424,
+      "step": 1098
+    },
+    {
+      "epoch": 4.462006079027356,
+      "grad_norm": 0.22387352091239793,
+      "learning_rate": 1.3662785384573663e-06,
+      "loss": 0.3425,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4660587639311045,
+      "grad_norm": 0.2249058767071061,
+      "learning_rate": 1.3457351939959383e-06,
+      "loss": 0.3396,
+      "step": 1100
+    },
+    {
+      "epoch": 4.470111448834853,
+      "grad_norm": 0.23870232680622352,
+      "learning_rate": 1.3253420884187551e-06,
+      "loss": 0.3405,
+      "step": 1101
+    },
+    {
+      "epoch": 4.474164133738602,
+      "grad_norm": 0.21607513672934625,
+      "learning_rate": 1.3050993859690953e-06,
+      "loss": 0.3328,
+      "step": 1102
+    },
+    {
+      "epoch": 4.47821681864235,
+      "grad_norm": 0.22535998908174948,
+      "learning_rate": 1.2850072496788869e-06,
+      "loss": 0.3554,
+      "step": 1103
+    },
+    {
+      "epoch": 4.4822695035460995,
+      "grad_norm": 0.21671387099938133,
+      "learning_rate": 1.2650658413674434e-06,
+      "loss": 0.35,
+      "step": 1104
+    },
+    {
+      "epoch": 4.486322188449848,
+      "grad_norm": 0.22130453377452314,
+      "learning_rate": 1.2452753216401226e-06,
+      "loss": 0.3334,
+      "step": 1105
+    },
+    {
+      "epoch": 4.490374873353597,
+      "grad_norm": 0.2404688922288609,
+      "learning_rate": 1.2256358498870503e-06,
+      "loss": 0.3433,
+      "step": 1106
+    },
+    {
+      "epoch": 4.494427558257345,
+      "grad_norm": 0.22385795363554434,
+      "learning_rate": 1.2061475842818337e-06,
+      "loss": 0.3461,
+      "step": 1107
+    },
+    {
+      "epoch": 4.498480243161095,
+      "grad_norm": 0.22413303912779692,
+      "learning_rate": 1.1868106817802816e-06,
+      "loss": 0.3436,
+      "step": 1108
+    },
+    {
+      "epoch": 4.502532928064843,
+      "grad_norm": 0.22122645891361573,
+      "learning_rate": 1.1676252981191482e-06,
+      "loss": 0.3338,
+      "step": 1109
+    },
+    {
+      "epoch": 4.506585612968592,
+      "grad_norm": 0.21376741367814048,
+      "learning_rate": 1.1485915878148823e-06,
+      "loss": 0.3635,
+      "step": 1110
+    },
+    {
+      "epoch": 4.51063829787234,
+      "grad_norm": 0.22603242927238207,
+      "learning_rate": 1.1297097041623584e-06,
+      "loss": 0.3223,
+      "step": 1111
+    },
+    {
+      "epoch": 4.514690982776089,
+      "grad_norm": 0.2331149933292654,
+      "learning_rate": 1.1109797992336847e-06,
+      "loss": 0.3258,
+      "step": 1112
+    },
+    {
+      "epoch": 4.518743667679838,
+      "grad_norm": 0.23959865419619877,
+      "learning_rate": 1.092402023876933e-06,
+      "loss": 0.3197,
+      "step": 1113
+    },
+    {
+      "epoch": 4.522796352583587,
+      "grad_norm": 0.22749193255505235,
+      "learning_rate": 1.0739765277149527e-06,
+      "loss": 0.3426,
+      "step": 1114
+    },
+    {
+      "epoch": 4.526849037487335,
+      "grad_norm": 0.22796993048828815,
+      "learning_rate": 1.0557034591441596e-06,
+      "loss": 0.3506,
+      "step": 1115
+    },
+    {
+      "epoch": 4.530901722391084,
+      "grad_norm": 0.22831565877977422,
+      "learning_rate": 1.0375829653333324e-06,
+      "loss": 0.3277,
+      "step": 1116
+    },
+    {
+      "epoch": 4.5349544072948325,
+      "grad_norm": 0.2203320741275949,
+      "learning_rate": 1.0196151922224385e-06,
+      "loss": 0.3494,
+      "step": 1117
+    },
+    {
+      "epoch": 4.539007092198582,
+      "grad_norm": 0.21633254005013025,
+      "learning_rate": 1.0018002845214526e-06,
+      "loss": 0.3275,
+      "step": 1118
+    },
+    {
+      "epoch": 4.54305977710233,
+      "grad_norm": 0.22961674015653205,
+      "learning_rate": 9.841383857091947e-07,
+      "loss": 0.3251,
+      "step": 1119
+    },
+    {
+      "epoch": 4.547112462006079,
+      "grad_norm": 0.21671390434912122,
+      "learning_rate": 9.666296380321616e-07,
+      "loss": 0.343,
+      "step": 1120
+    },
+    {
+      "epoch": 4.5511651469098275,
+      "grad_norm": 0.22603352068624985,
+      "learning_rate": 9.492741825034124e-07,
+      "loss": 0.3398,
+      "step": 1121
+    },
+    {
+      "epoch": 4.555217831813577,
+      "grad_norm": 0.22069987043277506,
+      "learning_rate": 9.320721589013892e-07,
+      "loss": 0.3458,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5592705167173255,
+      "grad_norm": 0.23010691388882049,
+      "learning_rate": 9.150237057688339e-07,
+      "loss": 0.3203,
+      "step": 1123
+    },
+    {
+      "epoch": 4.563323201621074,
+      "grad_norm": 0.2157397913050929,
+      "learning_rate": 8.981289604116328e-07,
+      "loss": 0.3397,
+      "step": 1124
+    },
+    {
+      "epoch": 4.567375886524823,
+      "grad_norm": 0.21786115695116354,
+      "learning_rate": 8.813880588977542e-07,
+      "loss": 0.3319,
+      "step": 1125
+    },
+    {
+      "epoch": 4.571428571428571,
+      "grad_norm": 0.20464643646357325,
+      "learning_rate": 8.648011360561126e-07,
+      "loss": 0.3439,
+      "step": 1126
+    },
+    {
+      "epoch": 4.575481256332321,
+      "grad_norm": 0.20755239544430837,
+      "learning_rate": 8.483683254755037e-07,
+      "loss": 0.344,
+      "step": 1127
+    },
+    {
+      "epoch": 4.579533941236069,
+      "grad_norm": 0.2089781380363568,
+      "learning_rate": 8.320897595035227e-07,
+      "loss": 0.35,
+      "step": 1128
+    },
+    {
+      "epoch": 4.583586626139818,
+      "grad_norm": 0.22082440420011618,
+      "learning_rate": 8.159655692455093e-07,
+      "loss": 0.3095,
+      "step": 1129
+    },
+    {
+      "epoch": 4.587639311043566,
+      "grad_norm": 0.2413705648165135,
+      "learning_rate": 7.999958845634648e-07,
+      "loss": 0.3247,
+      "step": 1130
+    },
+    {
+      "epoch": 4.591691995947315,
+      "grad_norm": 0.24045589166324063,
+      "learning_rate": 7.841808340750478e-07,
+      "loss": 0.3071,
+      "step": 1131
+    },
+    {
+      "epoch": 4.595744680851064,
+      "grad_norm": 0.22571645579981925,
+      "learning_rate": 7.685205451524869e-07,
+      "loss": 0.3435,
+      "step": 1132
+    },
+    {
+      "epoch": 4.599797365754813,
+      "grad_norm": 0.20648467045148045,
+      "learning_rate": 7.530151439216027e-07,
+      "loss": 0.3521,
+      "step": 1133
+    },
+    {
+      "epoch": 4.603850050658561,
+      "grad_norm": 0.21051011053555221,
+      "learning_rate": 7.376647552607675e-07,
+      "loss": 0.3427,
+      "step": 1134
+    },
+    {
+      "epoch": 4.60790273556231,
+      "grad_norm": 0.217544136467324,
+      "learning_rate": 7.224695027998963e-07,
+      "loss": 0.3341,
+      "step": 1135
+    },
+    {
+      "epoch": 4.611955420466058,
+      "grad_norm": 0.20640277762515988,
+      "learning_rate": 7.07429508919466e-07,
+      "loss": 0.3455,
+      "step": 1136
+    },
+    {
+      "epoch": 4.616008105369808,
+      "grad_norm": 0.21797673582665247,
+      "learning_rate": 6.925448947495206e-07,
+      "loss": 0.3434,
+      "step": 1137
+    },
+    {
+      "epoch": 4.620060790273556,
+      "grad_norm": 0.22022447802223266,
+      "learning_rate": 6.778157801686936e-07,
+      "loss": 0.3293,
+      "step": 1138
+    },
+    {
+      "epoch": 4.624113475177305,
+      "grad_norm": 0.21263282274197012,
+      "learning_rate": 6.632422838032515e-07,
+      "loss": 0.3373,
+      "step": 1139
+    },
+    {
+      "epoch": 4.6281661600810535,
+      "grad_norm": 0.214396956832954,
+      "learning_rate": 6.488245230261281e-07,
+      "loss": 0.3544,
+      "step": 1140
+    },
+    {
+      "epoch": 4.632218844984802,
+      "grad_norm": 0.2148090104291169,
+      "learning_rate": 6.345626139559868e-07,
+      "loss": 0.3518,
+      "step": 1141
+    },
+    {
+      "epoch": 4.6362715298885515,
+      "grad_norm": 0.22783424145149683,
+      "learning_rate": 6.204566714562866e-07,
+      "loss": 0.333,
+      "step": 1142
+    },
+    {
+      "epoch": 4.6403242147923,
+      "grad_norm": 0.21532478922335205,
+      "learning_rate": 6.06506809134344e-07,
+      "loss": 0.3493,
+      "step": 1143
+    },
+    {
+      "epoch": 4.644376899696049,
+      "grad_norm": 0.22297481495250232,
+      "learning_rate": 5.927131393404373e-07,
+      "loss": 0.3396,
+      "step": 1144
+    },
+    {
+      "epoch": 4.648429584599797,
+      "grad_norm": 0.22299241012230464,
+      "learning_rate": 5.790757731668817e-07,
+      "loss": 0.3402,
+      "step": 1145
+    },
+    {
+      "epoch": 4.652482269503546,
+      "grad_norm": 0.22405458848750587,
+      "learning_rate": 5.655948204471507e-07,
+      "loss": 0.3389,
+      "step": 1146
+    },
+    {
+      "epoch": 4.656534954407295,
+      "grad_norm": 0.2059481832033599,
+      "learning_rate": 5.522703897549875e-07,
+      "loss": 0.363,
+      "step": 1147
+    },
+    {
+      "epoch": 4.660587639311044,
+      "grad_norm": 0.22406832732174456,
+      "learning_rate": 5.391025884035239e-07,
+      "loss": 0.3192,
+      "step": 1148
+    },
+    {
+      "epoch": 4.664640324214792,
+      "grad_norm": 0.2123157382455703,
+      "learning_rate": 5.260915224444207e-07,
+      "loss": 0.3177,
+      "step": 1149
+    },
+    {
+      "epoch": 4.668693009118541,
+      "grad_norm": 0.21214183161863634,
+      "learning_rate": 5.132372966670129e-07,
+      "loss": 0.3428,
+      "step": 1150
+    },
+    {
+      "epoch": 4.672745694022289,
+      "grad_norm": 0.22214217040509063,
+      "learning_rate": 5.005400145974704e-07,
+      "loss": 0.3228,
+      "step": 1151
+    },
+    {
+      "epoch": 4.676798378926039,
+      "grad_norm": 0.22206843636459048,
+      "learning_rate": 4.879997784979562e-07,
+      "loss": 0.3128,
+      "step": 1152
+    },
+    {
+      "epoch": 4.680851063829787,
+      "grad_norm": 0.22502942651648752,
+      "learning_rate": 4.7561668936580984e-07,
+      "loss": 0.2986,
+      "step": 1153
+    },
+    {
+      "epoch": 4.684903748733536,
+      "grad_norm": 0.20286793667111946,
+      "learning_rate": 4.6339084693272306e-07,
+      "loss": 0.3457,
+      "step": 1154
+    },
+    {
+      "epoch": 4.688956433637284,
+      "grad_norm": 0.21521927202873165,
+      "learning_rate": 4.5132234966395847e-07,
+      "loss": 0.3333,
+      "step": 1155
+    },
+    {
+      "epoch": 4.693009118541033,
+      "grad_norm": 0.22798314042818463,
+      "learning_rate": 4.3941129475752795e-07,
+      "loss": 0.3297,
+      "step": 1156
+    },
+    {
+      "epoch": 4.697061803444782,
+      "grad_norm": 0.21904215857665196,
+      "learning_rate": 4.27657778143431e-07,
+      "loss": 0.3436,
+      "step": 1157
+    },
+    {
+      "epoch": 4.701114488348531,
+      "grad_norm": 0.22182735929955658,
+      "learning_rate": 4.1606189448287757e-07,
+      "loss": 0.3314,
+      "step": 1158
+    },
+    {
+      "epoch": 4.7051671732522795,
+      "grad_norm": 0.2095186956955109,
+      "learning_rate": 4.046237371675177e-07,
+      "loss": 0.3347,
+      "step": 1159
+    },
+    {
+      "epoch": 4.709219858156028,
+      "grad_norm": 0.22442127204518006,
+      "learning_rate": 3.9334339831869963e-07,
+      "loss": 0.3359,
+      "step": 1160
+    },
+    {
+      "epoch": 4.713272543059777,
+      "grad_norm": 0.21684324027495205,
+      "learning_rate": 3.8222096878671955e-07,
+      "loss": 0.3373,
+      "step": 1161
+    },
+    {
+      "epoch": 4.717325227963526,
+      "grad_norm": 0.2107650959337461,
+      "learning_rate": 3.7125653815009545e-07,
+      "loss": 0.3294,
+      "step": 1162
+    },
+    {
+      "epoch": 4.721377912867275,
+      "grad_norm": 0.20655868278548511,
+      "learning_rate": 3.6045019471484974e-07,
+      "loss": 0.3589,
+      "step": 1163
+    },
+    {
+      "epoch": 4.725430597771023,
+      "grad_norm": 0.21931130102701069,
+      "learning_rate": 3.498020255137813e-07,
+      "loss": 0.3214,
+      "step": 1164
+    },
+    {
+      "epoch": 4.729483282674772,
+      "grad_norm": 0.22416747292732309,
+      "learning_rate": 3.393121163057811e-07,
+      "loss": 0.3153,
+      "step": 1165
+    },
+    {
+      "epoch": 4.733535967578521,
+      "grad_norm": 0.21689993421909323,
+      "learning_rate": 3.289805515751399e-07,
+      "loss": 0.3353,
+      "step": 1166
+    },
+    {
+      "epoch": 4.73758865248227,
+      "grad_norm": 0.22352148295492932,
+      "learning_rate": 3.188074145308573e-07,
+      "loss": 0.3325,
+      "step": 1167
+    },
+    {
+      "epoch": 4.741641337386018,
+      "grad_norm": 0.21260809903189817,
+      "learning_rate": 3.087927871059804e-07,
+      "loss": 0.3392,
+      "step": 1168
+    },
+    {
+      "epoch": 4.745694022289767,
+      "grad_norm": 0.21638797232716142,
+      "learning_rate": 2.989367499569418e-07,
+      "loss": 0.3459,
+      "step": 1169
+    },
+    {
+      "epoch": 4.749746707193515,
+      "grad_norm": 0.22534975566915338,
+      "learning_rate": 2.8923938246290917e-07,
+      "loss": 0.3265,
+      "step": 1170
+    },
+    {
+      "epoch": 4.753799392097265,
+      "grad_norm": 0.21321198039942968,
+      "learning_rate": 2.7970076272514804e-07,
+      "loss": 0.3308,
+      "step": 1171
+    },
+    {
+      "epoch": 4.757852077001013,
+      "grad_norm": 0.22012878527457655,
+      "learning_rate": 2.703209675663887e-07,
+      "loss": 0.3522,
+      "step": 1172
+    },
+    {
+      "epoch": 4.761904761904762,
+      "grad_norm": 0.21029964979907614,
+      "learning_rate": 2.6110007253021374e-07,
+      "loss": 0.3366,
+      "step": 1173
+    },
+    {
+      "epoch": 4.76595744680851,
+      "grad_norm": 0.20580076966125707,
+      "learning_rate": 2.520381518804471e-07,
+      "loss": 0.3541,
+      "step": 1174
+    },
+    {
+      "epoch": 4.77001013171226,
+      "grad_norm": 0.2134745021807859,
+      "learning_rate": 2.4313527860054585e-07,
+      "loss": 0.3193,
+      "step": 1175
+    },
+    {
+      "epoch": 4.774062816616008,
+      "grad_norm": 0.22585064205156016,
+      "learning_rate": 2.343915243930317e-07,
+      "loss": 0.3252,
+      "step": 1176
+    },
+    {
+      "epoch": 4.778115501519757,
+      "grad_norm": 0.21603823512286785,
+      "learning_rate": 2.2580695967889367e-07,
+      "loss": 0.348,
+      "step": 1177
+    },
+    {
+      "epoch": 4.7821681864235055,
+      "grad_norm": 0.21653647329587522,
+      "learning_rate": 2.1738165359704189e-07,
+      "loss": 0.3204,
+      "step": 1178
+    },
+    {
+      "epoch": 4.786220871327254,
+      "grad_norm": 0.21205769892470547,
+      "learning_rate": 2.0911567400373257e-07,
+      "loss": 0.3353,
+      "step": 1179
+    },
+    {
+      "epoch": 4.7902735562310035,
+      "grad_norm": 0.21549336304459712,
+      "learning_rate": 2.0100908747202607e-07,
+      "loss": 0.346,
+      "step": 1180
+    },
+    {
+      "epoch": 4.794326241134752,
+      "grad_norm": 0.21059046847754115,
+      "learning_rate": 1.9306195929125638e-07,
+      "loss": 0.3306,
+      "step": 1181
+    },
+    {
+      "epoch": 4.798378926038501,
+      "grad_norm": 0.22735381874544033,
+      "learning_rate": 1.8527435346650247e-07,
+      "loss": 0.3157,
+      "step": 1182
+    },
+    {
+      "epoch": 4.802431610942249,
+      "grad_norm": 0.2198041799958636,
+      "learning_rate": 1.7764633271807108e-07,
+      "loss": 0.3316,
+      "step": 1183
+    },
+    {
+      "epoch": 4.806484295845998,
+      "grad_norm": 0.21349275524365252,
+      "learning_rate": 1.7017795848099262e-07,
+      "loss": 0.3382,
+      "step": 1184
+    },
+    {
+      "epoch": 4.810536980749747,
+      "grad_norm": 0.20634207760035184,
+      "learning_rate": 1.6286929090452596e-07,
+      "loss": 0.3595,
+      "step": 1185
+    },
+    {
+      "epoch": 4.814589665653496,
+      "grad_norm": 0.2149452566773372,
+      "learning_rate": 1.557203888516745e-07,
+      "loss": 0.3254,
+      "step": 1186
+    },
+    {
+      "epoch": 4.818642350557244,
+      "grad_norm": 0.2050290266477185,
+      "learning_rate": 1.487313098987131e-07,
+      "loss": 0.3486,
+      "step": 1187
+    },
+    {
+      "epoch": 4.822695035460993,
+      "grad_norm": 0.21730349274723942,
+      "learning_rate": 1.4190211033472402e-07,
+      "loss": 0.3138,
+      "step": 1188
+    },
+    {
+      "epoch": 4.826747720364741,
+      "grad_norm": 0.2138676861848332,
+      "learning_rate": 1.3523284516113955e-07,
+      "loss": 0.3143,
+      "step": 1189
+    },
+    {
+      "epoch": 4.830800405268491,
+      "grad_norm": 0.21788352088186105,
+      "learning_rate": 1.2872356809130682e-07,
+      "loss": 0.3216,
+      "step": 1190
+    },
+    {
+      "epoch": 4.834853090172239,
+      "grad_norm": 0.2265025016281195,
+      "learning_rate": 1.2237433155004807e-07,
+      "loss": 0.3134,
+      "step": 1191
+    },
+    {
+      "epoch": 4.838905775075988,
+      "grad_norm": 0.22179359558190445,
+      "learning_rate": 1.1618518667323886e-07,
+      "loss": 0.34,
+      "step": 1192
+    },
+    {
+      "epoch": 4.842958459979736,
+      "grad_norm": 0.2259526528670437,
+      "learning_rate": 1.1015618330740385e-07,
+      "loss": 0.3226,
+      "step": 1193
+    },
+    {
+      "epoch": 4.847011144883485,
+      "grad_norm": 0.22281249929212246,
+      "learning_rate": 1.042873700093061e-07,
+      "loss": 0.3367,
+      "step": 1194
+    },
+    {
+      "epoch": 4.851063829787234,
+      "grad_norm": 0.20646844314830884,
+      "learning_rate": 9.857879404556291e-08,
+      "loss": 0.3516,
+      "step": 1195
+    },
+    {
+      "epoch": 4.855116514690983,
+      "grad_norm": 0.20917766627740703,
+      "learning_rate": 9.303050139225722e-08,
+      "loss": 0.3445,
+      "step": 1196
+    },
+    {
+      "epoch": 4.8591691995947315,
+      "grad_norm": 0.21438077777876302,
+      "learning_rate": 8.76425367345779e-08,
+      "loss": 0.348,
+      "step": 1197
+    },
+    {
+      "epoch": 4.86322188449848,
+      "grad_norm": 0.21386021572345543,
+      "learning_rate": 8.241494346644897e-08,
+      "loss": 0.3438,
+      "step": 1198
+    },
+    {
+      "epoch": 4.867274569402229,
+      "grad_norm": 0.2219330645543717,
+      "learning_rate": 7.734776369019204e-08,
+      "loss": 0.3293,
+      "step": 1199
+    },
+    {
+      "epoch": 4.871327254305978,
+      "grad_norm": 0.21913258808872252,
+      "learning_rate": 7.244103821617332e-08,
+      "loss": 0.3483,
+      "step": 1200
+    },
+    {
+      "epoch": 4.875379939209727,
+      "grad_norm": 0.2046396443128444,
+      "learning_rate": 6.769480656248606e-08,
+      "loss": 0.3666,
+      "step": 1201
+    },
+    {
+      "epoch": 4.879432624113475,
+      "grad_norm": 0.2235359374333409,
+      "learning_rate": 6.310910695462635e-08,
+      "loss": 0.3498,
+      "step": 1202
+    },
+    {
+      "epoch": 4.883485309017224,
+      "grad_norm": 0.21714147168119471,
+      "learning_rate": 5.8683976325191185e-08,
+      "loss": 0.3255,
+      "step": 1203
+    },
+    {
+      "epoch": 4.887537993920972,
+      "grad_norm": 0.20965957836909663,
+      "learning_rate": 5.4419450313571984e-08,
+      "loss": 0.3388,
+      "step": 1204
+    },
+    {
+      "epoch": 4.891590678824722,
+      "grad_norm": 0.21025007654825317,
+      "learning_rate": 5.031556326567488e-08,
+      "loss": 0.3325,
+      "step": 1205
+    },
+    {
+      "epoch": 4.89564336372847,
+      "grad_norm": 0.22045886448240218,
+      "learning_rate": 4.637234823364312e-08,
+      "loss": 0.3395,
+      "step": 1206
+    },
+    {
+      "epoch": 4.899696048632219,
+      "grad_norm": 0.2216809920246643,
+      "learning_rate": 4.258983697558838e-08,
+      "loss": 0.313,
+      "step": 1207
+    },
+    {
+      "epoch": 4.903748733535967,
+      "grad_norm": 0.21919342892553204,
+      "learning_rate": 3.896805995533548e-08,
+      "loss": 0.3327,
+      "step": 1208
+    },
+    {
+      "epoch": 4.907801418439716,
+      "grad_norm": 0.22963652477786461,
+      "learning_rate": 3.550704634218028e-08,
+      "loss": 0.3243,
+      "step": 1209
+    },
+    {
+      "epoch": 4.911854103343465,
+      "grad_norm": 0.21024186093097164,
+      "learning_rate": 3.2206824010647676e-08,
+      "loss": 0.3526,
+      "step": 1210
+    },
+    {
+      "epoch": 4.915906788247214,
+      "grad_norm": 0.22336999236382776,
+      "learning_rate": 2.9067419540278476e-08,
+      "loss": 0.3265,
+      "step": 1211
+    },
+    {
+      "epoch": 4.919959473150962,
+      "grad_norm": 0.2201397513245513,
+      "learning_rate": 2.6088858215400638e-08,
+      "loss": 0.3363,
+      "step": 1212
+    },
+    {
+      "epoch": 4.924012158054711,
+      "grad_norm": 0.22111502149214282,
+      "learning_rate": 2.3271164024940564e-08,
+      "loss": 0.3179,
+      "step": 1213
+    },
+    {
+      "epoch": 4.92806484295846,
+      "grad_norm": 0.22260370419361142,
+      "learning_rate": 2.061435966221881e-08,
+      "loss": 0.333,
+      "step": 1214
+    },
+    {
+      "epoch": 4.932117527862209,
+      "grad_norm": 0.21061532746297282,
+      "learning_rate": 1.811846652477245e-08,
+      "loss": 0.3368,
+      "step": 1215
+    },
+    {
+      "epoch": 4.9361702127659575,
+      "grad_norm": 0.2324705516186909,
+      "learning_rate": 1.5783504714184106e-08,
+      "loss": 0.3124,
+      "step": 1216
+    },
+    {
+      "epoch": 4.940222897669706,
+      "grad_norm": 0.21503925446600383,
+      "learning_rate": 1.360949303591097e-08,
+      "loss": 0.3327,
+      "step": 1217
+    },
+    {
+      "epoch": 4.944275582573455,
+      "grad_norm": 0.21453766341928934,
+      "learning_rate": 1.1596448999144916e-08,
+      "loss": 0.329,
+      "step": 1218
+    },
+    {
+      "epoch": 4.948328267477204,
+      "grad_norm": 0.21274947142888392,
+      "learning_rate": 9.744388816668172e-09,
+      "loss": 0.3356,
+      "step": 1219
+    },
+    {
+      "epoch": 4.9523809523809526,
+      "grad_norm": 0.21256060764333481,
+      "learning_rate": 8.05332740472009e-09,
+      "loss": 0.3208,
+      "step": 1220
+    },
+    {
+      "epoch": 4.956433637284701,
+      "grad_norm": 0.2162259550966047,
+      "learning_rate": 6.523278382872811e-09,
+      "loss": 0.3329,
+      "step": 1221
+    },
+    {
+      "epoch": 4.96048632218845,
+      "grad_norm": 0.21256351533196707,
+      "learning_rate": 5.15425407393133e-09,
+      "loss": 0.3527,
+      "step": 1222
+    },
+    {
+      "epoch": 4.964539007092198,
+      "grad_norm": 0.2142136871122153,
+      "learning_rate": 3.94626550383137e-09,
+      "loss": 0.3372,
+      "step": 1223
+    },
+    {
+      "epoch": 4.968591691995948,
+      "grad_norm": 0.20876176461532178,
+      "learning_rate": 2.899322401546112e-09,
+      "loss": 0.3445,
+      "step": 1224
+    },
+    {
+      "epoch": 4.972644376899696,
+      "grad_norm": 0.21723609682700945,
+      "learning_rate": 2.013433199010706e-09,
+      "loss": 0.3397,
+      "step": 1225
+    },
+    {
+      "epoch": 4.976697061803445,
+      "grad_norm": 0.21289286076311653,
+      "learning_rate": 1.2886050310556563e-09,
+      "loss": 0.3297,
+      "step": 1226
+    },
+    {
+      "epoch": 4.980749746707193,
+      "grad_norm": 0.22390735712541313,
+      "learning_rate": 7.248437353468695e-10,
+      "loss": 0.3258,
+      "step": 1227
+    },
+    {
+      "epoch": 4.984802431610943,
+      "grad_norm": 0.23301239046115135,
+      "learning_rate": 3.221538523412449e-10,
+      "loss": 0.353,
+      "step": 1228
+    },
+    {
+      "epoch": 4.988855116514691,
+      "grad_norm": 0.21838384856465362,
+      "learning_rate": 8.053862524670663e-11,
+      "loss": 0.353,
+      "step": 1229
+    },
+    {
+      "epoch": 4.99290780141844,
+      "grad_norm": 0.23683621287256168,
+      "learning_rate": 0.0,
+      "loss": 0.3136,
+      "step": 1230
+    },
+    {
+      "epoch": 4.99290780141844,
+      "step": 1230,
+      "total_flos": 4.845566732934316e+18,
+      "train_loss": 0.46200237676380124,
+      "train_runtime": 53793.4177,
+      "train_samples_per_second": 2.935,
+      "train_steps_per_second": 0.023
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1230,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.845566732934316e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}