diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5272 @@
+{
+  "best_global_step": 744,
+  "best_metric": 0.9021978021978022,
+  "best_model_checkpoint": "./albert_multilabel_base\\checkpoint-744",
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 744,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004032258064516129,
+      "grad_norm": 4.463273525238037,
+      "learning_rate": 1.9973118279569895e-05,
+      "loss": 0.6697,
+      "step": 1
+    },
+    {
+      "epoch": 0.008064516129032258,
+      "grad_norm": 4.188074111938477,
+      "learning_rate": 1.9946236559139788e-05,
+      "loss": 0.618,
+      "step": 2
+    },
+    {
+      "epoch": 0.012096774193548387,
+      "grad_norm": 4.5477166175842285,
+      "learning_rate": 1.9919354838709678e-05,
+      "loss": 0.5686,
+      "step": 3
+    },
+    {
+      "epoch": 0.016129032258064516,
+      "grad_norm": 3.3298499584198,
+      "learning_rate": 1.989247311827957e-05,
+      "loss": 0.5691,
+      "step": 4
+    },
+    {
+      "epoch": 0.020161290322580645,
+      "grad_norm": 3.3022494316101074,
+      "learning_rate": 1.9865591397849465e-05,
+      "loss": 0.5157,
+      "step": 5
+    },
+    {
+      "epoch": 0.024193548387096774,
+      "grad_norm": 3.095808744430542,
+      "learning_rate": 1.9838709677419358e-05,
+      "loss": 0.4709,
+      "step": 6
+    },
+    {
+      "epoch": 0.028225806451612902,
+      "grad_norm": 2.8690195083618164,
+      "learning_rate": 1.9811827956989248e-05,
+      "loss": 0.4679,
+      "step": 7
+    },
+    {
+      "epoch": 0.03225806451612903,
+      "grad_norm": 2.376504898071289,
+      "learning_rate": 1.978494623655914e-05,
+      "loss": 0.4801,
+      "step": 8
+    },
+    {
+      "epoch": 0.036290322580645164,
+      "grad_norm": 2.947565793991089,
+      "learning_rate": 1.9758064516129035e-05,
+      "loss": 0.4196,
+      "step": 9
+    },
+    {
+      "epoch": 0.04032258064516129,
+      "grad_norm": 2.624013662338257,
+      "learning_rate": 1.9731182795698928e-05,
+      "loss": 0.3948,
+      "step": 10
+    },
+    {
+      "epoch": 0.04435483870967742,
+      "grad_norm": 2.218142032623291,
+      "learning_rate": 1.9704301075268818e-05,
+      "loss": 0.3588,
+      "step": 11
+    },
+    {
+      "epoch": 0.04838709677419355,
+      "grad_norm": 2.0879335403442383,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.3732,
+      "step": 12
+    },
+    {
+      "epoch": 0.05241935483870968,
+      "grad_norm": 1.8436014652252197,
+      "learning_rate": 1.9650537634408605e-05,
+      "loss": 0.3549,
+      "step": 13
+    },
+    {
+      "epoch": 0.056451612903225805,
+      "grad_norm": 2.7105393409729004,
+      "learning_rate": 1.9623655913978498e-05,
+      "loss": 0.3588,
+      "step": 14
+    },
+    {
+      "epoch": 0.06048387096774194,
+      "grad_norm": 2.038672685623169,
+      "learning_rate": 1.9596774193548388e-05,
+      "loss": 0.3431,
+      "step": 15
+    },
+    {
+      "epoch": 0.06451612903225806,
+      "grad_norm": 1.9189316034317017,
+      "learning_rate": 1.956989247311828e-05,
+      "loss": 0.3334,
+      "step": 16
+    },
+    {
+      "epoch": 0.06854838709677419,
+      "grad_norm": 1.5137898921966553,
+      "learning_rate": 1.9543010752688174e-05,
+      "loss": 0.3142,
+      "step": 17
+    },
+    {
+      "epoch": 0.07258064516129033,
+      "grad_norm": 1.6630574464797974,
+      "learning_rate": 1.9516129032258068e-05,
+      "loss": 0.3702,
+      "step": 18
+    },
+    {
+      "epoch": 0.07661290322580645,
+      "grad_norm": 1.7377407550811768,
+      "learning_rate": 1.9489247311827958e-05,
+      "loss": 0.3059,
+      "step": 19
+    },
+    {
+      "epoch": 0.08064516129032258,
+      "grad_norm": 1.9414572715759277,
+      "learning_rate": 1.946236559139785e-05,
+      "loss": 0.3146,
+      "step": 20
+    },
+    {
+      "epoch": 0.0846774193548387,
+      "grad_norm": 1.5869086980819702,
+      "learning_rate": 1.9435483870967744e-05,
+      "loss": 0.3053,
+      "step": 21
+    },
+    {
+      "epoch": 0.08870967741935484,
+      "grad_norm": 1.5989874601364136,
+      "learning_rate": 1.9408602150537638e-05,
+      "loss": 0.3022,
+      "step": 22
+    },
+    {
+      "epoch": 0.09274193548387097,
+      "grad_norm": 2.2120656967163086,
+      "learning_rate": 1.9381720430107528e-05,
+      "loss": 0.3372,
+      "step": 23
+    },
+    {
+      "epoch": 0.0967741935483871,
+      "grad_norm": 1.8420207500457764,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.2947,
+      "step": 24
+    },
+    {
+      "epoch": 0.10080645161290322,
+      "grad_norm": 1.324240803718567,
+      "learning_rate": 1.9327956989247314e-05,
+      "loss": 0.3117,
+      "step": 25
+    },
+    {
+      "epoch": 0.10483870967741936,
+      "grad_norm": 1.3763035535812378,
+      "learning_rate": 1.9301075268817207e-05,
+      "loss": 0.3097,
+      "step": 26
+    },
+    {
+      "epoch": 0.10887096774193548,
+      "grad_norm": 1.13789963722229,
+      "learning_rate": 1.9274193548387097e-05,
+      "loss": 0.3073,
+      "step": 27
+    },
+    {
+      "epoch": 0.11290322580645161,
+      "grad_norm": 1.8375061750411987,
+      "learning_rate": 1.924731182795699e-05,
+      "loss": 0.2729,
+      "step": 28
+    },
+    {
+      "epoch": 0.11693548387096774,
+      "grad_norm": 1.4714664220809937,
+      "learning_rate": 1.9220430107526884e-05,
+      "loss": 0.2884,
+      "step": 29
+    },
+    {
+      "epoch": 0.12096774193548387,
+      "grad_norm": 1.7859275341033936,
+      "learning_rate": 1.9193548387096777e-05,
+      "loss": 0.2921,
+      "step": 30
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 1.1975992918014526,
+      "learning_rate": 1.916666666666667e-05,
+      "loss": 0.2999,
+      "step": 31
+    },
+    {
+      "epoch": 0.12903225806451613,
+      "grad_norm": 1.1491625308990479,
+      "learning_rate": 1.913978494623656e-05,
+      "loss": 0.3037,
+      "step": 32
+    },
+    {
+      "epoch": 0.13306451612903225,
+      "grad_norm": 1.7719403505325317,
+      "learning_rate": 1.9112903225806454e-05,
+      "loss": 0.3097,
+      "step": 33
+    },
+    {
+      "epoch": 0.13709677419354838,
+      "grad_norm": 2.003831148147583,
+      "learning_rate": 1.9086021505376347e-05,
+      "loss": 0.3419,
+      "step": 34
+    },
+    {
+      "epoch": 0.14112903225806453,
+      "grad_norm": 1.2528795003890991,
+      "learning_rate": 1.905913978494624e-05,
+      "loss": 0.3104,
+      "step": 35
+    },
+    {
+      "epoch": 0.14516129032258066,
+      "grad_norm": 1.9549521207809448,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.2948,
+      "step": 36
+    },
+    {
+      "epoch": 0.14919354838709678,
+      "grad_norm": 3.2783548831939697,
+      "learning_rate": 1.9005376344086024e-05,
+      "loss": 0.2442,
+      "step": 37
+    },
+    {
+      "epoch": 0.1532258064516129,
+      "grad_norm": 2.185920476913452,
+      "learning_rate": 1.8978494623655917e-05,
+      "loss": 0.2714,
+      "step": 38
+    },
+    {
+      "epoch": 0.15725806451612903,
+      "grad_norm": 1.0794386863708496,
+      "learning_rate": 1.895161290322581e-05,
+      "loss": 0.27,
+      "step": 39
+    },
+    {
+      "epoch": 0.16129032258064516,
+      "grad_norm": 1.1620965003967285,
+      "learning_rate": 1.89247311827957e-05,
+      "loss": 0.2259,
+      "step": 40
+    },
+    {
+      "epoch": 0.16532258064516128,
+      "grad_norm": 1.599996566772461,
+      "learning_rate": 1.8897849462365594e-05,
+      "loss": 0.2035,
+      "step": 41
+    },
+    {
+      "epoch": 0.1693548387096774,
+      "grad_norm": 1.2780784368515015,
+      "learning_rate": 1.8870967741935487e-05,
+      "loss": 0.22,
+      "step": 42
+    },
+    {
+      "epoch": 0.17338709677419356,
+      "grad_norm": 1.4363526105880737,
+      "learning_rate": 1.884408602150538e-05,
+      "loss": 0.2368,
+      "step": 43
+    },
+    {
+      "epoch": 0.1774193548387097,
+      "grad_norm": 2.520078420639038,
+      "learning_rate": 1.881720430107527e-05,
+      "loss": 0.1892,
+      "step": 44
+    },
+    {
+      "epoch": 0.1814516129032258,
+      "grad_norm": 1.2548242807388306,
+      "learning_rate": 1.8790322580645163e-05,
+      "loss": 0.2666,
+      "step": 45
+    },
+    {
+      "epoch": 0.18548387096774194,
+      "grad_norm": 3.150402545928955,
+      "learning_rate": 1.8763440860215057e-05,
+      "loss": 0.2119,
+      "step": 46
+    },
+    {
+      "epoch": 0.18951612903225806,
+      "grad_norm": 2.226469039916992,
+      "learning_rate": 1.8736559139784947e-05,
+      "loss": 0.2952,
+      "step": 47
+    },
+    {
+      "epoch": 0.1935483870967742,
+      "grad_norm": 2.2046310901641846,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.2089,
+      "step": 48
+    },
+    {
+      "epoch": 0.1975806451612903,
+      "grad_norm": 2.4914636611938477,
+      "learning_rate": 1.8682795698924733e-05,
+      "loss": 0.2622,
+      "step": 49
+    },
+    {
+      "epoch": 0.20161290322580644,
+      "grad_norm": 1.6519418954849243,
+      "learning_rate": 1.8655913978494623e-05,
+      "loss": 0.2268,
+      "step": 50
+    },
+    {
+      "epoch": 0.2056451612903226,
+      "grad_norm": 2.803372383117676,
+      "learning_rate": 1.8629032258064517e-05,
+      "loss": 0.2628,
+      "step": 51
+    },
+    {
+      "epoch": 0.20967741935483872,
+      "grad_norm": 1.6514664888381958,
+      "learning_rate": 1.860215053763441e-05,
+      "loss": 0.1742,
+      "step": 52
+    },
+    {
+      "epoch": 0.21370967741935484,
+      "grad_norm": 4.286014556884766,
+      "learning_rate": 1.8575268817204303e-05,
+      "loss": 0.3353,
+      "step": 53
+    },
+    {
+      "epoch": 0.21774193548387097,
+      "grad_norm": 1.3388841152191162,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.193,
+      "step": 54
+    },
+    {
+      "epoch": 0.2217741935483871,
+      "grad_norm": 3.4638161659240723,
+      "learning_rate": 1.8521505376344086e-05,
+      "loss": 0.243,
+      "step": 55
+    },
+    {
+      "epoch": 0.22580645161290322,
+      "grad_norm": 2.002242088317871,
+      "learning_rate": 1.849462365591398e-05,
+      "loss": 0.2098,
+      "step": 56
+    },
+    {
+      "epoch": 0.22983870967741934,
+      "grad_norm": 2.865968942642212,
+      "learning_rate": 1.8467741935483873e-05,
+      "loss": 0.2875,
+      "step": 57
+    },
+    {
+      "epoch": 0.23387096774193547,
+      "grad_norm": 4.366675853729248,
+      "learning_rate": 1.8440860215053763e-05,
+      "loss": 0.1991,
+      "step": 58
+    },
+    {
+      "epoch": 0.23790322580645162,
+      "grad_norm": 3.3366456031799316,
+      "learning_rate": 1.8413978494623656e-05,
+      "loss": 0.2024,
+      "step": 59
+    },
+    {
+      "epoch": 0.24193548387096775,
+      "grad_norm": 3.0101261138916016,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.1529,
+      "step": 60
+    },
+    {
+      "epoch": 0.24596774193548387,
+      "grad_norm": 1.7896872758865356,
+      "learning_rate": 1.8360215053763443e-05,
+      "loss": 0.1969,
+      "step": 61
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.260681390762329,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.2426,
+      "step": 62
+    },
+    {
+      "epoch": 0.2540322580645161,
+      "grad_norm": 2.9987573623657227,
+      "learning_rate": 1.8306451612903226e-05,
+      "loss": 0.2763,
+      "step": 63
+    },
+    {
+      "epoch": 0.25806451612903225,
+      "grad_norm": 2.370321273803711,
+      "learning_rate": 1.827956989247312e-05,
+      "loss": 0.1721,
+      "step": 64
+    },
+    {
+      "epoch": 0.2620967741935484,
+      "grad_norm": 2.941296339035034,
+      "learning_rate": 1.8252688172043013e-05,
+      "loss": 0.2495,
+      "step": 65
+    },
+    {
+      "epoch": 0.2661290322580645,
+      "grad_norm": 2.7063724994659424,
+      "learning_rate": 1.8225806451612903e-05,
+      "loss": 0.1995,
+      "step": 66
+    },
+    {
+      "epoch": 0.2701612903225806,
+      "grad_norm": 2.6740007400512695,
+      "learning_rate": 1.8198924731182796e-05,
+      "loss": 0.1484,
+      "step": 67
+    },
+    {
+      "epoch": 0.27419354838709675,
+      "grad_norm": 4.054372310638428,
+      "learning_rate": 1.817204301075269e-05,
+      "loss": 0.1958,
+      "step": 68
+    },
+    {
+      "epoch": 0.2782258064516129,
+      "grad_norm": 3.995595932006836,
+      "learning_rate": 1.8145161290322583e-05,
+      "loss": 0.1581,
+      "step": 69
+    },
+    {
+      "epoch": 0.28225806451612906,
+      "grad_norm": 4.279732704162598,
+      "learning_rate": 1.8118279569892473e-05,
+      "loss": 0.2051,
+      "step": 70
+    },
+    {
+      "epoch": 0.2862903225806452,
+      "grad_norm": 2.194711446762085,
+      "learning_rate": 1.8091397849462366e-05,
+      "loss": 0.2699,
+      "step": 71
+    },
+    {
+      "epoch": 0.2903225806451613,
+      "grad_norm": 2.1309289932250977,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.202,
+      "step": 72
+    },
+    {
+      "epoch": 0.29435483870967744,
+      "grad_norm": 4.026139259338379,
+      "learning_rate": 1.8037634408602153e-05,
+      "loss": 0.2346,
+      "step": 73
+    },
+    {
+      "epoch": 0.29838709677419356,
+      "grad_norm": 2.3642139434814453,
+      "learning_rate": 1.8010752688172042e-05,
+      "loss": 0.2095,
+      "step": 74
+    },
+    {
+      "epoch": 0.3024193548387097,
+      "grad_norm": 1.9105987548828125,
+      "learning_rate": 1.7983870967741936e-05,
+      "loss": 0.1839,
+      "step": 75
+    },
+    {
+      "epoch": 0.3064516129032258,
+      "grad_norm": 1.926434874534607,
+      "learning_rate": 1.795698924731183e-05,
+      "loss": 0.1984,
+      "step": 76
+    },
+    {
+      "epoch": 0.31048387096774194,
+      "grad_norm": 1.3975627422332764,
+      "learning_rate": 1.7930107526881722e-05,
+      "loss": 0.2008,
+      "step": 77
+    },
+    {
+      "epoch": 0.31451612903225806,
+      "grad_norm": 2.820605993270874,
+      "learning_rate": 1.7903225806451612e-05,
+      "loss": 0.1451,
+      "step": 78
+    },
+    {
+      "epoch": 0.3185483870967742,
+      "grad_norm": 2.23899245262146,
+      "learning_rate": 1.7876344086021506e-05,
+      "loss": 0.1839,
+      "step": 79
+    },
+    {
+      "epoch": 0.3225806451612903,
+      "grad_norm": 1.4833577871322632,
+      "learning_rate": 1.78494623655914e-05,
+      "loss": 0.1874,
+      "step": 80
+    },
+    {
+      "epoch": 0.32661290322580644,
+      "grad_norm": 3.754470109939575,
+      "learning_rate": 1.7822580645161292e-05,
+      "loss": 0.2299,
+      "step": 81
+    },
+    {
+      "epoch": 0.33064516129032256,
+      "grad_norm": 3.65791654586792,
+      "learning_rate": 1.7795698924731186e-05,
+      "loss": 0.2075,
+      "step": 82
+    },
+    {
+      "epoch": 0.3346774193548387,
+      "grad_norm": 2.704558849334717,
+      "learning_rate": 1.7768817204301075e-05,
+      "loss": 0.1688,
+      "step": 83
+    },
+    {
+      "epoch": 0.3387096774193548,
+      "grad_norm": 1.7909516096115112,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.2472,
+      "step": 84
+    },
+    {
+      "epoch": 0.34274193548387094,
+      "grad_norm": 1.412300705909729,
+      "learning_rate": 1.7715053763440862e-05,
+      "loss": 0.1858,
+      "step": 85
+    },
+    {
+      "epoch": 0.3467741935483871,
+      "grad_norm": 2.0086007118225098,
+      "learning_rate": 1.7688172043010755e-05,
+      "loss": 0.2088,
+      "step": 86
+    },
+    {
+      "epoch": 0.35080645161290325,
+      "grad_norm": 1.4397780895233154,
+      "learning_rate": 1.7661290322580645e-05,
+      "loss": 0.1816,
+      "step": 87
+    },
+    {
+      "epoch": 0.3548387096774194,
+      "grad_norm": 2.947721242904663,
+      "learning_rate": 1.763440860215054e-05,
+      "loss": 0.1629,
+      "step": 88
+    },
+    {
+      "epoch": 0.3588709677419355,
+      "grad_norm": 1.9372519254684448,
+      "learning_rate": 1.7607526881720432e-05,
+      "loss": 0.1636,
+      "step": 89
+    },
+    {
+      "epoch": 0.3629032258064516,
+      "grad_norm": 1.4998315572738647,
+      "learning_rate": 1.7580645161290325e-05,
+      "loss": 0.2162,
+      "step": 90
+    },
+    {
+      "epoch": 0.36693548387096775,
+      "grad_norm": 3.8654611110687256,
+      "learning_rate": 1.7553763440860215e-05,
+      "loss": 0.2311,
+      "step": 91
+    },
+    {
+      "epoch": 0.3709677419354839,
+      "grad_norm": 2.416931390762329,
+      "learning_rate": 1.752688172043011e-05,
+      "loss": 0.161,
+      "step": 92
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 1.4065924882888794,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.1919,
+      "step": 93
+    },
+    {
+      "epoch": 0.3790322580645161,
+      "grad_norm": 3.5286269187927246,
+      "learning_rate": 1.7473118279569895e-05,
+      "loss": 0.2338,
+      "step": 94
+    },
+    {
+      "epoch": 0.38306451612903225,
+      "grad_norm": 3.0253779888153076,
+      "learning_rate": 1.7446236559139785e-05,
+      "loss": 0.1824,
+      "step": 95
+    },
+    {
+      "epoch": 0.3870967741935484,
+      "grad_norm": 1.31352961063385,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.1203,
+      "step": 96
+    },
+    {
+      "epoch": 0.3911290322580645,
+      "grad_norm": 2.3616368770599365,
+      "learning_rate": 1.7392473118279572e-05,
+      "loss": 0.2555,
+      "step": 97
+    },
+    {
+      "epoch": 0.3951612903225806,
+      "grad_norm": 1.9040699005126953,
+      "learning_rate": 1.7365591397849465e-05,
+      "loss": 0.1857,
+      "step": 98
+    },
+    {
+      "epoch": 0.39919354838709675,
+      "grad_norm": 1.4389543533325195,
+      "learning_rate": 1.7338709677419355e-05,
+      "loss": 0.1379,
+      "step": 99
+    },
+    {
+      "epoch": 0.4032258064516129,
+      "grad_norm": 1.4597176313400269,
+      "learning_rate": 1.7311827956989248e-05,
+      "loss": 0.1258,
+      "step": 100
+    },
+    {
+      "epoch": 0.40725806451612906,
+      "grad_norm": 2.730079174041748,
+      "learning_rate": 1.728494623655914e-05,
+      "loss": 0.1936,
+      "step": 101
+    },
+    {
+      "epoch": 0.4112903225806452,
+      "grad_norm": 1.6900081634521484,
+      "learning_rate": 1.7258064516129035e-05,
+      "loss": 0.1782,
+      "step": 102
+    },
+    {
+      "epoch": 0.4153225806451613,
+      "grad_norm": 3.9283804893493652,
+      "learning_rate": 1.7231182795698925e-05,
+      "loss": 0.1447,
+      "step": 103
+    },
+    {
+      "epoch": 0.41935483870967744,
+      "grad_norm": 2.625030755996704,
+      "learning_rate": 1.7204301075268818e-05,
+      "loss": 0.1361,
+      "step": 104
+    },
+    {
+      "epoch": 0.42338709677419356,
+      "grad_norm": 6.293011665344238,
+      "learning_rate": 1.717741935483871e-05,
+      "loss": 0.2563,
+      "step": 105
+    },
+    {
+      "epoch": 0.4274193548387097,
+      "grad_norm": 3.8414859771728516,
+      "learning_rate": 1.7150537634408605e-05,
+      "loss": 0.2118,
+      "step": 106
+    },
+    {
+      "epoch": 0.4314516129032258,
+      "grad_norm": 1.7090542316436768,
+      "learning_rate": 1.7123655913978495e-05,
+      "loss": 0.2919,
+      "step": 107
+    },
+    {
+      "epoch": 0.43548387096774194,
+      "grad_norm": 1.4093631505966187,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.2151,
+      "step": 108
+    },
+    {
+      "epoch": 0.43951612903225806,
+      "grad_norm": 1.6816186904907227,
+      "learning_rate": 1.706989247311828e-05,
+      "loss": 0.2001,
+      "step": 109
+    },
+    {
+      "epoch": 0.4435483870967742,
+      "grad_norm": 2.5388357639312744,
+      "learning_rate": 1.7043010752688175e-05,
+      "loss": 0.1839,
+      "step": 110
+    },
+    {
+      "epoch": 0.4475806451612903,
+      "grad_norm": 4.355051040649414,
+      "learning_rate": 1.7016129032258068e-05,
+      "loss": 0.1517,
+      "step": 111
+    },
+    {
+      "epoch": 0.45161290322580644,
+      "grad_norm": 5.043978691101074,
+      "learning_rate": 1.6989247311827958e-05,
+      "loss": 0.1148,
+      "step": 112
+    },
+    {
+      "epoch": 0.45564516129032256,
+      "grad_norm": 4.623153209686279,
+      "learning_rate": 1.696236559139785e-05,
+      "loss": 0.2277,
+      "step": 113
+    },
+    {
+      "epoch": 0.4596774193548387,
+      "grad_norm": 2.165748119354248,
+      "learning_rate": 1.6935483870967744e-05,
+      "loss": 0.175,
+      "step": 114
+    },
+    {
+      "epoch": 0.4637096774193548,
+      "grad_norm": 2.074892520904541,
+      "learning_rate": 1.6908602150537638e-05,
+      "loss": 0.107,
+      "step": 115
+    },
+    {
+      "epoch": 0.46774193548387094,
+      "grad_norm": 4.597195625305176,
+      "learning_rate": 1.6881720430107528e-05,
+      "loss": 0.201,
+      "step": 116
+    },
+    {
+      "epoch": 0.4717741935483871,
+      "grad_norm": 1.589660406112671,
+      "learning_rate": 1.685483870967742e-05,
+      "loss": 0.1904,
+      "step": 117
+    },
+    {
+      "epoch": 0.47580645161290325,
+      "grad_norm": 3.892699718475342,
+      "learning_rate": 1.6827956989247314e-05,
+      "loss": 0.1536,
+      "step": 118
+    },
+    {
+      "epoch": 0.4798387096774194,
+      "grad_norm": 1.4085686206817627,
+      "learning_rate": 1.6801075268817208e-05,
+      "loss": 0.1762,
+      "step": 119
+    },
+    {
+      "epoch": 0.4838709677419355,
+      "grad_norm": 2.0159590244293213,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.1433,
+      "step": 120
+    },
+    {
+      "epoch": 0.4879032258064516,
+      "grad_norm": 3.211463451385498,
+      "learning_rate": 1.674731182795699e-05,
+      "loss": 0.2131,
+      "step": 121
+    },
+    {
+      "epoch": 0.49193548387096775,
+      "grad_norm": 1.9942578077316284,
+      "learning_rate": 1.6720430107526884e-05,
+      "loss": 0.1962,
+      "step": 122
+    },
+    {
+      "epoch": 0.4959677419354839,
+      "grad_norm": 2.7773678302764893,
+      "learning_rate": 1.6693548387096778e-05,
+      "loss": 0.1516,
+      "step": 123
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.387947082519531,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.2402,
+      "step": 124
+    },
+    {
+      "epoch": 0.5040322580645161,
+      "grad_norm": 3.206733465194702,
+      "learning_rate": 1.663978494623656e-05,
+      "loss": 0.1615,
+      "step": 125
+    },
+    {
+      "epoch": 0.5080645161290323,
+      "grad_norm": 4.810270309448242,
+      "learning_rate": 1.6612903225806454e-05,
+      "loss": 0.1587,
+      "step": 126
+    },
+    {
+      "epoch": 0.5120967741935484,
+      "grad_norm": 3.2066237926483154,
+      "learning_rate": 1.6586021505376347e-05,
+      "loss": 0.1747,
+      "step": 127
+    },
+    {
+      "epoch": 0.5161290322580645,
+      "grad_norm": 1.2533971071243286,
+      "learning_rate": 1.6559139784946237e-05,
+      "loss": 0.1662,
+      "step": 128
+    },
+    {
+      "epoch": 0.5201612903225806,
+      "grad_norm": 4.853026390075684,
+      "learning_rate": 1.653225806451613e-05,
+      "loss": 0.1895,
+      "step": 129
+    },
+    {
+      "epoch": 0.5241935483870968,
+      "grad_norm": 1.5984219312667847,
+      "learning_rate": 1.6505376344086024e-05,
+      "loss": 0.1867,
+      "step": 130
+    },
+    {
+      "epoch": 0.5282258064516129,
+      "grad_norm": 1.9208190441131592,
+      "learning_rate": 1.6478494623655917e-05,
+      "loss": 0.1661,
+      "step": 131
+    },
+    {
+      "epoch": 0.532258064516129,
+      "grad_norm": 4.714879035949707,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.1978,
+      "step": 132
+    },
+    {
+      "epoch": 0.5362903225806451,
+      "grad_norm": 1.9932862520217896,
+      "learning_rate": 1.64247311827957e-05,
+      "loss": 0.1894,
+      "step": 133
+    },
+    {
+      "epoch": 0.5403225806451613,
+      "grad_norm": 1.8594914674758911,
+      "learning_rate": 1.6397849462365594e-05,
+      "loss": 0.1819,
+      "step": 134
+    },
+    {
+      "epoch": 0.5443548387096774,
+      "grad_norm": 3.666376829147339,
+      "learning_rate": 1.6370967741935487e-05,
+      "loss": 0.1594,
+      "step": 135
+    },
+    {
+      "epoch": 0.5483870967741935,
+      "grad_norm": 5.887596607208252,
+      "learning_rate": 1.6344086021505377e-05,
+      "loss": 0.1807,
+      "step": 136
+    },
+    {
+      "epoch": 0.5524193548387096,
+      "grad_norm": 1.3112871646881104,
+      "learning_rate": 1.631720430107527e-05,
+      "loss": 0.1411,
+      "step": 137
+    },
+    {
+      "epoch": 0.5564516129032258,
+      "grad_norm": 2.456446409225464,
+      "learning_rate": 1.6290322580645164e-05,
+      "loss": 0.1413,
+      "step": 138
+    },
+    {
+      "epoch": 0.5604838709677419,
+      "grad_norm": 4.192088603973389,
+      "learning_rate": 1.6263440860215057e-05,
+      "loss": 0.1422,
+      "step": 139
+    },
+    {
+      "epoch": 0.5645161290322581,
+      "grad_norm": 2.659003734588623,
+      "learning_rate": 1.6236559139784947e-05,
+      "loss": 0.1812,
+      "step": 140
+    },
+    {
+      "epoch": 0.5685483870967742,
+      "grad_norm": 1.699617624282837,
+      "learning_rate": 1.620967741935484e-05,
+      "loss": 0.1771,
+      "step": 141
+    },
+    {
+      "epoch": 0.5725806451612904,
+      "grad_norm": 2.221937656402588,
+      "learning_rate": 1.618279569892473e-05,
+      "loss": 0.1815,
+      "step": 142
+    },
+    {
+      "epoch": 0.5766129032258065,
+      "grad_norm": 1.5366272926330566,
+      "learning_rate": 1.6155913978494623e-05,
+      "loss": 0.184,
+      "step": 143
+    },
+    {
+      "epoch": 0.5806451612903226,
+      "grad_norm": 2.0756564140319824,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.145,
+      "step": 144
+    },
+    {
+      "epoch": 0.5846774193548387,
+      "grad_norm": 1.1063413619995117,
+      "learning_rate": 1.610215053763441e-05,
+      "loss": 0.0964,
+      "step": 145
+    },
+    {
+      "epoch": 0.5887096774193549,
+      "grad_norm": 1.9002684354782104,
+      "learning_rate": 1.60752688172043e-05,
+      "loss": 0.2313,
+      "step": 146
+    },
+    {
+      "epoch": 0.592741935483871,
+      "grad_norm": 3.5199642181396484,
+      "learning_rate": 1.6048387096774193e-05,
+      "loss": 0.1803,
+      "step": 147
+    },
+    {
+      "epoch": 0.5967741935483871,
+      "grad_norm": 2.263772487640381,
+      "learning_rate": 1.6021505376344087e-05,
+      "loss": 0.1661,
+      "step": 148
+    },
+    {
+      "epoch": 0.6008064516129032,
+      "grad_norm": 3.434790849685669,
+      "learning_rate": 1.599462365591398e-05,
+      "loss": 0.1729,
+      "step": 149
+    },
+    {
+      "epoch": 0.6048387096774194,
+      "grad_norm": 3.1659963130950928,
+      "learning_rate": 1.596774193548387e-05,
+      "loss": 0.2271,
+      "step": 150
+    },
+    {
+      "epoch": 0.6088709677419355,
+      "grad_norm": 3.9298930168151855,
+      "learning_rate": 1.5940860215053763e-05,
+      "loss": 0.1787,
+      "step": 151
+    },
+    {
+      "epoch": 0.6129032258064516,
+      "grad_norm": 5.221301078796387,
+      "learning_rate": 1.5913978494623657e-05,
+      "loss": 0.1432,
+      "step": 152
+    },
+    {
+      "epoch": 0.6169354838709677,
+      "grad_norm": 0.9517404437065125,
+      "learning_rate": 1.588709677419355e-05,
+      "loss": 0.143,
+      "step": 153
+    },
+    {
+      "epoch": 0.6209677419354839,
+      "grad_norm": 1.685404658317566,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.1446,
+      "step": 154
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 4.428375720977783,
+      "learning_rate": 1.5833333333333333e-05,
+      "loss": 0.2276,
+      "step": 155
+    },
+    {
+      "epoch": 0.6290322580645161,
+      "grad_norm": 1.976226568222046,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.1822,
+      "step": 156
+    },
+    {
+      "epoch": 0.6330645161290323,
+      "grad_norm": 2.600104570388794,
+      "learning_rate": 1.577956989247312e-05,
+      "loss": 0.2352,
+      "step": 157
+    },
+    {
+      "epoch": 0.6370967741935484,
+      "grad_norm": 2.413137197494507,
+      "learning_rate": 1.5752688172043013e-05,
+      "loss": 0.258,
+      "step": 158
+    },
+    {
+      "epoch": 0.6411290322580645,
+      "grad_norm": 2.801959753036499,
+      "learning_rate": 1.5725806451612903e-05,
+      "loss": 0.1796,
+      "step": 159
+    },
+    {
+      "epoch": 0.6451612903225806,
+      "grad_norm": 2.2109487056732178,
+      "learning_rate": 1.5698924731182796e-05,
+      "loss": 0.1749,
+      "step": 160
+    },
+    {
+      "epoch": 0.6491935483870968,
+      "grad_norm": 3.4560952186584473,
+      "learning_rate": 1.567204301075269e-05,
+      "loss": 0.1315,
+      "step": 161
+    },
+    {
+      "epoch": 0.6532258064516129,
+      "grad_norm": 3.611140251159668,
+      "learning_rate": 1.5645161290322583e-05,
+      "loss": 0.2518,
+      "step": 162
+    },
+    {
+      "epoch": 0.657258064516129,
+      "grad_norm": 3.5987987518310547,
+      "learning_rate": 1.5618279569892473e-05,
+      "loss": 0.2134,
+      "step": 163
+    },
+    {
+      "epoch": 0.6612903225806451,
+      "grad_norm": 3.6609513759613037,
+      "learning_rate": 1.5591397849462366e-05,
+      "loss": 0.1134,
+      "step": 164
+    },
+    {
+      "epoch": 0.6653225806451613,
+      "grad_norm": 4.708600044250488,
+      "learning_rate": 1.556451612903226e-05,
+      "loss": 0.1581,
+      "step": 165
+    },
+    {
+      "epoch": 0.6693548387096774,
+      "grad_norm": 2.0652990341186523,
+      "learning_rate": 1.5537634408602153e-05,
+      "loss": 0.114,
+      "step": 166
+    },
+    {
+      "epoch": 0.6733870967741935,
+      "grad_norm": 2.3634023666381836,
+      "learning_rate": 1.5510752688172043e-05,
+      "loss": 0.1383,
+      "step": 167
+    },
+    {
+      "epoch": 0.6774193548387096,
+      "grad_norm": 1.8686503171920776,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.1071,
+      "step": 168
+    },
+    {
+      "epoch": 0.6814516129032258,
+      "grad_norm": 2.7749221324920654,
+      "learning_rate": 1.545698924731183e-05,
+      "loss": 0.1394,
+      "step": 169
+    },
+    {
+      "epoch": 0.6854838709677419,
+      "grad_norm": 5.210952281951904,
+      "learning_rate": 1.5430107526881723e-05,
+      "loss": 0.1698,
+      "step": 170
+    },
+    {
+      "epoch": 0.6895161290322581,
+      "grad_norm": 1.9079982042312622,
+      "learning_rate": 1.5403225806451613e-05,
+      "loss": 0.149,
+      "step": 171
+    },
+    {
+      "epoch": 0.6935483870967742,
+      "grad_norm": 3.1015419960021973,
+      "learning_rate": 1.5376344086021506e-05,
+      "loss": 0.1821,
+      "step": 172
+    },
+    {
+      "epoch": 0.6975806451612904,
+      "grad_norm": 2.9040095806121826,
+      "learning_rate": 1.53494623655914e-05,
+      "loss": 0.1624,
+      "step": 173
+    },
+    {
+      "epoch": 0.7016129032258065,
+      "grad_norm": 3.451782703399658,
+      "learning_rate": 1.5322580645161292e-05,
+      "loss": 0.2349,
+      "step": 174
+    },
+    {
+      "epoch": 0.7056451612903226,
+      "grad_norm": 2.0777103900909424,
+      "learning_rate": 1.5295698924731182e-05,
+      "loss": 0.1575,
+      "step": 175
+    },
+    {
+      "epoch": 0.7096774193548387,
+      "grad_norm": 2.8455822467803955,
+      "learning_rate": 1.5268817204301076e-05,
+      "loss": 0.1402,
+      "step": 176
+    },
+    {
+      "epoch": 0.7137096774193549,
+      "grad_norm": 2.1864373683929443,
+      "learning_rate": 1.5241935483870969e-05,
+      "loss": 0.2501,
+      "step": 177
+    },
+    {
+      "epoch": 0.717741935483871,
+      "grad_norm": 2.945974349975586,
+      "learning_rate": 1.521505376344086e-05,
+      "loss": 0.1758,
+      "step": 178
+    },
+    {
+      "epoch": 0.7217741935483871,
+      "grad_norm": 2.071998357772827,
+      "learning_rate": 1.5188172043010754e-05,
+      "loss": 0.1185,
+      "step": 179
+    },
+    {
+      "epoch": 0.7258064516129032,
+      "grad_norm": 2.1226487159729004,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.1809,
+      "step": 180
+    },
+    {
+      "epoch": 0.7298387096774194,
+      "grad_norm": 3.2159790992736816,
+      "learning_rate": 1.5134408602150539e-05,
+      "loss": 0.1583,
+      "step": 181
+    },
+    {
+      "epoch": 0.7338709677419355,
+      "grad_norm": 1.7255215644836426,
+      "learning_rate": 1.510752688172043e-05,
+      "loss": 0.1706,
+      "step": 182
+    },
+    {
+      "epoch": 0.7379032258064516,
+      "grad_norm": 1.9457521438598633,
+      "learning_rate": 1.5080645161290324e-05,
+      "loss": 0.1658,
+      "step": 183
+    },
+    {
+      "epoch": 0.7419354838709677,
+      "grad_norm": 7.032689571380615,
+      "learning_rate": 1.5053763440860215e-05,
+      "loss": 0.173,
+      "step": 184
+    },
+    {
+      "epoch": 0.7459677419354839,
+      "grad_norm": 2.1762492656707764,
+      "learning_rate": 1.5026881720430109e-05,
+      "loss": 0.144,
+      "step": 185
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.8494467735290527,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.1933,
+      "step": 186
+    },
+    {
+      "epoch": 0.7540322580645161,
+      "grad_norm": 1.369672179222107,
+      "learning_rate": 1.4973118279569894e-05,
+      "loss": 0.1082,
+      "step": 187
+    },
+    {
+      "epoch": 0.7580645161290323,
+      "grad_norm": 2.0053024291992188,
+      "learning_rate": 1.4946236559139787e-05,
+      "loss": 0.1564,
+      "step": 188
+    },
+    {
+      "epoch": 0.7620967741935484,
+      "grad_norm": 3.1830568313598633,
+      "learning_rate": 1.4919354838709679e-05,
+      "loss": 0.1956,
+      "step": 189
+    },
+    {
+      "epoch": 0.7661290322580645,
+      "grad_norm": 2.129666328430176,
+      "learning_rate": 1.4892473118279572e-05,
+      "loss": 0.2594,
+      "step": 190
+    },
+    {
+      "epoch": 0.7701612903225806,
+      "grad_norm": 2.121384859085083,
+      "learning_rate": 1.4865591397849464e-05,
+      "loss": 0.1363,
+      "step": 191
+    },
+    {
+      "epoch": 0.7741935483870968,
+      "grad_norm": 2.5438694953918457,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.15,
+      "step": 192
+    },
+    {
+      "epoch": 0.7782258064516129,
+      "grad_norm": 1.4790626764297485,
+      "learning_rate": 1.4811827956989248e-05,
+      "loss": 0.1677,
+      "step": 193
+    },
+    {
+      "epoch": 0.782258064516129,
+      "grad_norm": 5.348865985870361,
+      "learning_rate": 1.4784946236559142e-05,
+      "loss": 0.2535,
+      "step": 194
+    },
+    {
+      "epoch": 0.7862903225806451,
+      "grad_norm": 7.304387092590332,
+      "learning_rate": 1.4758064516129033e-05,
+      "loss": 0.2327,
+      "step": 195
+    },
+    {
+      "epoch": 0.7903225806451613,
+      "grad_norm": 1.6960372924804688,
+      "learning_rate": 1.4731182795698927e-05,
+      "loss": 0.1453,
+      "step": 196
+    },
+    {
+      "epoch": 0.7943548387096774,
+      "grad_norm": 5.402785301208496,
+      "learning_rate": 1.4704301075268818e-05,
+      "loss": 0.2091,
+      "step": 197
+    },
+    {
+      "epoch": 0.7983870967741935,
+      "grad_norm": 6.772708415985107,
+      "learning_rate": 1.4677419354838712e-05,
+      "loss": 0.1622,
+      "step": 198
+    },
+    {
+      "epoch": 0.8024193548387096,
+      "grad_norm": 2.564171075820923,
+      "learning_rate": 1.4650537634408603e-05,
+      "loss": 0.2299,
+      "step": 199
+    },
+    {
+      "epoch": 0.8064516129032258,
+      "grad_norm": 2.6038239002227783,
+      "learning_rate": 1.4623655913978497e-05,
+      "loss": 0.1725,
+      "step": 200
+    },
+    {
+      "epoch": 0.8104838709677419,
+      "grad_norm": 2.527942180633545,
+      "learning_rate": 1.4596774193548388e-05,
+      "loss": 0.1359,
+      "step": 201
+    },
+    {
+      "epoch": 0.8145161290322581,
+      "grad_norm": 3.5944533348083496,
+      "learning_rate": 1.4569892473118282e-05,
+      "loss": 0.1268,
+      "step": 202
+    },
+    {
+      "epoch": 0.8185483870967742,
+      "grad_norm": 3.4926414489746094,
+      "learning_rate": 1.4543010752688173e-05,
+      "loss": 0.1473,
+      "step": 203
+    },
+    {
+      "epoch": 0.8225806451612904,
+      "grad_norm": 5.071582794189453,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.196,
+      "step": 204
+    },
+    {
+      "epoch": 0.8266129032258065,
+      "grad_norm": 4.80251932144165,
+      "learning_rate": 1.4489247311827958e-05,
+      "loss": 0.19,
+      "step": 205
+    },
+    {
+      "epoch": 0.8306451612903226,
+      "grad_norm": 3.378063678741455,
+      "learning_rate": 1.4462365591397851e-05,
+      "loss": 0.1293,
+      "step": 206
+    },
+    {
+      "epoch": 0.8346774193548387,
+      "grad_norm": 3.138362407684326,
+      "learning_rate": 1.4435483870967743e-05,
+      "loss": 0.1919,
+      "step": 207
+    },
+    {
+      "epoch": 0.8387096774193549,
+      "grad_norm": 3.114121675491333,
+      "learning_rate": 1.4408602150537636e-05,
+      "loss": 0.3021,
+      "step": 208
+    },
+    {
+      "epoch": 0.842741935483871,
+      "grad_norm": 2.903273582458496,
+      "learning_rate": 1.4381720430107528e-05,
+      "loss": 0.1613,
+      "step": 209
+    },
+    {
+      "epoch": 0.8467741935483871,
+      "grad_norm": 2.530947685241699,
+      "learning_rate": 1.4354838709677421e-05,
+      "loss": 0.2119,
+      "step": 210
+    },
+    {
+      "epoch": 0.8508064516129032,
+      "grad_norm": 1.9773578643798828,
+      "learning_rate": 1.4327956989247313e-05,
+      "loss": 0.1994,
+      "step": 211
+    },
+    {
+      "epoch": 0.8548387096774194,
+      "grad_norm": 3.0400474071502686,
+      "learning_rate": 1.4301075268817206e-05,
+      "loss": 0.1553,
+      "step": 212
+    },
+    {
+      "epoch": 0.8588709677419355,
+      "grad_norm": 2.43621826171875,
+      "learning_rate": 1.4274193548387098e-05,
+      "loss": 0.1592,
+      "step": 213
+    },
+    {
+      "epoch": 0.8629032258064516,
+      "grad_norm": 3.8775973320007324,
+      "learning_rate": 1.4247311827956991e-05,
+      "loss": 0.161,
+      "step": 214
+    },
+    {
+      "epoch": 0.8669354838709677,
+      "grad_norm": 2.805100202560425,
+      "learning_rate": 1.4220430107526883e-05,
+      "loss": 0.1363,
+      "step": 215
+    },
+    {
+      "epoch": 0.8709677419354839,
+      "grad_norm": 5.203761577606201,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.1403,
+      "step": 216
+    },
+    {
+      "epoch": 0.875,
+      "grad_norm": 2.3190488815307617,
+      "learning_rate": 1.416666666666667e-05,
+      "loss": 0.2115,
+      "step": 217
+    },
+    {
+      "epoch": 0.8790322580645161,
+      "grad_norm": 2.513376474380493,
+      "learning_rate": 1.4139784946236561e-05,
+      "loss": 0.227,
+      "step": 218
+    },
+    {
+      "epoch": 0.8830645161290323,
+      "grad_norm": 3.1920270919799805,
+      "learning_rate": 1.4112903225806454e-05,
+      "loss": 0.1732,
+      "step": 219
+    },
+    {
+      "epoch": 0.8870967741935484,
+      "grad_norm": 2.252589225769043,
+      "learning_rate": 1.4086021505376346e-05,
+      "loss": 0.1329,
+      "step": 220
+    },
+    {
+      "epoch": 0.8911290322580645,
+      "grad_norm": 3.943742275238037,
+      "learning_rate": 1.405913978494624e-05,
+      "loss": 0.1473,
+      "step": 221
+    },
+    {
+      "epoch": 0.8951612903225806,
+      "grad_norm": 4.05203104019165,
+      "learning_rate": 1.4032258064516131e-05,
+      "loss": 0.1647,
+      "step": 222
+    },
+    {
+      "epoch": 0.8991935483870968,
+      "grad_norm": 4.86761474609375,
+      "learning_rate": 1.4005376344086024e-05,
+      "loss": 0.118,
+      "step": 223
+    },
+    {
+      "epoch": 0.9032258064516129,
+      "grad_norm": 2.4546005725860596,
+      "learning_rate": 1.3978494623655916e-05,
+      "loss": 0.2356,
+      "step": 224
+    },
+    {
+      "epoch": 0.907258064516129,
+      "grad_norm": 1.0434294939041138,
+      "learning_rate": 1.3951612903225809e-05,
+      "loss": 0.1536,
+      "step": 225
+    },
+    {
+      "epoch": 0.9112903225806451,
+      "grad_norm": 1.4482147693634033,
+      "learning_rate": 1.39247311827957e-05,
+      "loss": 0.1479,
+      "step": 226
+    },
+    {
+      "epoch": 0.9153225806451613,
+      "grad_norm": 2.1162939071655273,
+      "learning_rate": 1.3897849462365594e-05,
+      "loss": 0.2139,
+      "step": 227
+    },
+    {
+      "epoch": 0.9193548387096774,
+      "grad_norm": 1.401563048362732,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.2074,
+      "step": 228
+    },
+    {
+      "epoch": 0.9233870967741935,
+      "grad_norm": 2.659918785095215,
+      "learning_rate": 1.3844086021505379e-05,
+      "loss": 0.1711,
+      "step": 229
+    },
+    {
+      "epoch": 0.9274193548387096,
+      "grad_norm": 2.3634390830993652,
+      "learning_rate": 1.381720430107527e-05,
+      "loss": 0.1347,
+      "step": 230
+    },
+    {
+      "epoch": 0.9314516129032258,
+      "grad_norm": 3.2036492824554443,
+      "learning_rate": 1.3790322580645164e-05,
+      "loss": 0.1348,
+      "step": 231
+    },
+    {
+      "epoch": 0.9354838709677419,
+      "grad_norm": 2.362253427505493,
+      "learning_rate": 1.3763440860215056e-05,
+      "loss": 0.1623,
+      "step": 232
+    },
+    {
+      "epoch": 0.9395161290322581,
+      "grad_norm": 4.0349860191345215,
+      "learning_rate": 1.3736559139784945e-05,
+      "loss": 0.1421,
+      "step": 233
+    },
+    {
+      "epoch": 0.9435483870967742,
+      "grad_norm": 2.3942513465881348,
+      "learning_rate": 1.3709677419354839e-05,
+      "loss": 0.1678,
+      "step": 234
+    },
+    {
+      "epoch": 0.9475806451612904,
+      "grad_norm": 2.3485310077667236,
+      "learning_rate": 1.3682795698924732e-05,
+      "loss": 0.1327,
+      "step": 235
+    },
+    {
+      "epoch": 0.9516129032258065,
+      "grad_norm": 1.8485794067382812,
+      "learning_rate": 1.3655913978494624e-05,
+      "loss": 0.0943,
+      "step": 236
+    },
+    {
+      "epoch": 0.9556451612903226,
+      "grad_norm": 2.820059299468994,
+      "learning_rate": 1.3629032258064517e-05,
+      "loss": 0.1664,
+      "step": 237
+    },
+    {
+      "epoch": 0.9596774193548387,
+      "grad_norm": 3.143728256225586,
+      "learning_rate": 1.3602150537634409e-05,
+      "loss": 0.2003,
+      "step": 238
+    },
+    {
+      "epoch": 0.9637096774193549,
+      "grad_norm": 3.496140241622925,
+      "learning_rate": 1.3575268817204302e-05,
+      "loss": 0.1833,
+      "step": 239
+    },
+    {
+      "epoch": 0.967741935483871,
+      "grad_norm": 2.736649751663208,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.1505,
+      "step": 240
+    },
+    {
+      "epoch": 0.9717741935483871,
+      "grad_norm": 2.9205009937286377,
+      "learning_rate": 1.3521505376344087e-05,
+      "loss": 0.22,
+      "step": 241
+    },
+    {
+      "epoch": 0.9758064516129032,
+      "grad_norm": 2.8891749382019043,
+      "learning_rate": 1.3494623655913978e-05,
+      "loss": 0.1415,
+      "step": 242
+    },
+    {
+      "epoch": 0.9798387096774194,
+      "grad_norm": 2.7234678268432617,
+      "learning_rate": 1.3467741935483872e-05,
+      "loss": 0.1112,
+      "step": 243
+    },
+    {
+      "epoch": 0.9838709677419355,
+      "grad_norm": 2.395036220550537,
+      "learning_rate": 1.3440860215053763e-05,
+      "loss": 0.1288,
+      "step": 244
+    },
+    {
+      "epoch": 0.9879032258064516,
+      "grad_norm": 3.214719772338867,
+      "learning_rate": 1.3413978494623657e-05,
+      "loss": 0.1637,
+      "step": 245
+    },
+    {
+      "epoch": 0.9919354838709677,
+      "grad_norm": 3.2017691135406494,
+      "learning_rate": 1.3387096774193548e-05,
+      "loss": 0.1614,
+      "step": 246
+    },
+    {
+      "epoch": 0.9959677419354839,
+      "grad_norm": 2.6992430686950684,
+      "learning_rate": 1.3360215053763442e-05,
+      "loss": 0.125,
+      "step": 247
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.5830893516540527,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.1237,
+      "step": 248
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.7306818181818182,
+      "eval_f1": 0.8891346419684821,
+      "eval_loss": 0.1686243861913681,
+      "eval_runtime": 2.375,
+      "eval_samples_per_second": 370.523,
+      "eval_steps_per_second": 11.789,
+      "step": 248
+    },
+    {
+      "epoch": 1.0040322580645162,
+      "grad_norm": 3.1516823768615723,
+      "learning_rate": 1.3306451612903227e-05,
+      "loss": 0.2192,
+      "step": 249
+    },
+    {
+      "epoch": 1.0080645161290323,
+      "grad_norm": 3.530272960662842,
+      "learning_rate": 1.3279569892473118e-05,
+      "loss": 0.2153,
+      "step": 250
+    },
+    {
+      "epoch": 1.0120967741935485,
+      "grad_norm": 1.239799976348877,
+      "learning_rate": 1.3252688172043012e-05,
+      "loss": 0.1496,
+      "step": 251
+    },
+    {
+      "epoch": 1.0161290322580645,
+      "grad_norm": 1.6457215547561646,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.1542,
+      "step": 252
+    },
+    {
+      "epoch": 1.0201612903225807,
+      "grad_norm": 3.6155178546905518,
+      "learning_rate": 1.3198924731182796e-05,
+      "loss": 0.2078,
+      "step": 253
+    },
+    {
+      "epoch": 1.0241935483870968,
+      "grad_norm": 1.6368227005004883,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.1271,
+      "step": 254
+    },
+    {
+      "epoch": 1.028225806451613,
+      "grad_norm": 3.2075986862182617,
+      "learning_rate": 1.3145161290322581e-05,
+      "loss": 0.2012,
+      "step": 255
+    },
+    {
+      "epoch": 1.032258064516129,
+      "grad_norm": 2.500032901763916,
+      "learning_rate": 1.3118279569892473e-05,
+      "loss": 0.1093,
+      "step": 256
+    },
+    {
+      "epoch": 1.0362903225806452,
+      "grad_norm": 2.440420150756836,
+      "learning_rate": 1.3091397849462366e-05,
+      "loss": 0.1072,
+      "step": 257
+    },
+    {
+      "epoch": 1.0403225806451613,
+      "grad_norm": 3.545008659362793,
+      "learning_rate": 1.3064516129032258e-05,
+      "loss": 0.1596,
+      "step": 258
+    },
+    {
+      "epoch": 1.0443548387096775,
+      "grad_norm": 3.0299694538116455,
+      "learning_rate": 1.3037634408602151e-05,
+      "loss": 0.1784,
+      "step": 259
+    },
+    {
+      "epoch": 1.0483870967741935,
+      "grad_norm": 1.5560317039489746,
+      "learning_rate": 1.3010752688172043e-05,
+      "loss": 0.1254,
+      "step": 260
+    },
+    {
+      "epoch": 1.0524193548387097,
+      "grad_norm": 3.661555528640747,
+      "learning_rate": 1.2983870967741936e-05,
+      "loss": 0.1665,
+      "step": 261
+    },
+    {
+      "epoch": 1.0564516129032258,
+      "grad_norm": 2.275700092315674,
+      "learning_rate": 1.2956989247311828e-05,
+      "loss": 0.113,
+      "step": 262
+    },
+    {
+      "epoch": 1.060483870967742,
+      "grad_norm": 0.8315228223800659,
+      "learning_rate": 1.2930107526881721e-05,
+      "loss": 0.1173,
+      "step": 263
+    },
+    {
+      "epoch": 1.064516129032258,
+      "grad_norm": 1.0392459630966187,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.0891,
+      "step": 264
+    },
+    {
+      "epoch": 1.0685483870967742,
+      "grad_norm": 1.7290191650390625,
+      "learning_rate": 1.2876344086021506e-05,
+      "loss": 0.1432,
+      "step": 265
+    },
+    {
+      "epoch": 1.0725806451612903,
+      "grad_norm": 4.252922058105469,
+      "learning_rate": 1.28494623655914e-05,
+      "loss": 0.1772,
+      "step": 266
+    },
+    {
+      "epoch": 1.0766129032258065,
+      "grad_norm": 3.656017303466797,
+      "learning_rate": 1.2822580645161291e-05,
+      "loss": 0.1832,
+      "step": 267
+    },
+    {
+      "epoch": 1.0806451612903225,
+      "grad_norm": 8.33147144317627,
+      "learning_rate": 1.2795698924731184e-05,
+      "loss": 0.219,
+      "step": 268
+    },
+    {
+      "epoch": 1.0846774193548387,
+      "grad_norm": 2.9899277687072754,
+      "learning_rate": 1.2768817204301076e-05,
+      "loss": 0.1716,
+      "step": 269
+    },
+    {
+      "epoch": 1.0887096774193548,
+      "grad_norm": 1.3610584735870361,
+      "learning_rate": 1.274193548387097e-05,
+      "loss": 0.145,
+      "step": 270
+    },
+    {
+      "epoch": 1.092741935483871,
+      "grad_norm": 3.155823230743408,
+      "learning_rate": 1.2715053763440861e-05,
+      "loss": 0.1671,
+      "step": 271
+    },
+    {
+      "epoch": 1.096774193548387,
+      "grad_norm": 2.408665657043457,
+      "learning_rate": 1.2688172043010754e-05,
+      "loss": 0.1151,
+      "step": 272
+    },
+    {
+      "epoch": 1.1008064516129032,
+      "grad_norm": 1.192688226699829,
+      "learning_rate": 1.2661290322580646e-05,
+      "loss": 0.1489,
+      "step": 273
+    },
+    {
+      "epoch": 1.1048387096774193,
+      "grad_norm": 2.6764822006225586,
+      "learning_rate": 1.2634408602150539e-05,
+      "loss": 0.1593,
+      "step": 274
+    },
+    {
+      "epoch": 1.1088709677419355,
+      "grad_norm": 4.13783073425293,
+      "learning_rate": 1.260752688172043e-05,
+      "loss": 0.1911,
+      "step": 275
+    },
+    {
+      "epoch": 1.1129032258064515,
+      "grad_norm": 1.4944156408309937,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.2192,
+      "step": 276
+    },
+    {
+      "epoch": 1.1169354838709677,
+      "grad_norm": 1.162097692489624,
+      "learning_rate": 1.2553763440860216e-05,
+      "loss": 0.1277,
+      "step": 277
+    },
+    {
+      "epoch": 1.120967741935484,
+      "grad_norm": 2.7152187824249268,
+      "learning_rate": 1.2526881720430109e-05,
+      "loss": 0.2191,
+      "step": 278
+    },
+    {
+      "epoch": 1.125,
+      "grad_norm": 2.895779609680176,
+      "learning_rate": 1.25e-05,
+      "loss": 0.1696,
+      "step": 279
+    },
+    {
+      "epoch": 1.129032258064516,
+      "grad_norm": 2.1827454566955566,
+      "learning_rate": 1.2473118279569894e-05,
+      "loss": 0.0868,
+      "step": 280
+    },
+    {
+      "epoch": 1.1330645161290323,
+      "grad_norm": 2.273141860961914,
+      "learning_rate": 1.2446236559139786e-05,
+      "loss": 0.214,
+      "step": 281
+    },
+    {
+      "epoch": 1.1370967741935485,
+      "grad_norm": 1.9680695533752441,
+      "learning_rate": 1.2419354838709679e-05,
+      "loss": 0.1716,
+      "step": 282
+    },
+    {
+      "epoch": 1.1411290322580645,
+      "grad_norm": 1.9214951992034912,
+      "learning_rate": 1.239247311827957e-05,
+      "loss": 0.1965,
+      "step": 283
+    },
+    {
+      "epoch": 1.1451612903225807,
+      "grad_norm": 5.198123931884766,
+      "learning_rate": 1.2365591397849464e-05,
+      "loss": 0.1683,
+      "step": 284
+    },
+    {
+      "epoch": 1.1491935483870968,
+      "grad_norm": 1.6851317882537842,
+      "learning_rate": 1.2338709677419355e-05,
+      "loss": 0.1698,
+      "step": 285
+    },
+    {
+      "epoch": 1.153225806451613,
+      "grad_norm": 2.037721633911133,
+      "learning_rate": 1.2311827956989249e-05,
+      "loss": 0.1484,
+      "step": 286
+    },
+    {
+      "epoch": 1.157258064516129,
+      "grad_norm": 1.9201178550720215,
+      "learning_rate": 1.228494623655914e-05,
+      "loss": 0.1128,
+      "step": 287
+    },
+    {
+      "epoch": 1.1612903225806452,
+      "grad_norm": 2.517728090286255,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.1865,
+      "step": 288
+    },
+    {
+      "epoch": 1.1653225806451613,
+      "grad_norm": 2.0093541145324707,
+      "learning_rate": 1.2231182795698925e-05,
+      "loss": 0.1765,
+      "step": 289
+    },
+    {
+      "epoch": 1.1693548387096775,
+      "grad_norm": 2.9235808849334717,
+      "learning_rate": 1.2204301075268819e-05,
+      "loss": 0.2289,
+      "step": 290
+    },
+    {
+      "epoch": 1.1733870967741935,
+      "grad_norm": 2.3604586124420166,
+      "learning_rate": 1.217741935483871e-05,
+      "loss": 0.1056,
+      "step": 291
+    },
+    {
+      "epoch": 1.1774193548387097,
+      "grad_norm": 2.574960947036743,
+      "learning_rate": 1.2150537634408604e-05,
+      "loss": 0.1599,
+      "step": 292
+    },
+    {
+      "epoch": 1.1814516129032258,
+      "grad_norm": 3.23299241065979,
+      "learning_rate": 1.2123655913978495e-05,
+      "loss": 0.0828,
+      "step": 293
+    },
+    {
+      "epoch": 1.185483870967742,
+      "grad_norm": 3.568268299102783,
+      "learning_rate": 1.2096774193548388e-05,
+      "loss": 0.168,
+      "step": 294
+    },
+    {
+      "epoch": 1.189516129032258,
+      "grad_norm": 1.8116379976272583,
+      "learning_rate": 1.2069892473118282e-05,
+      "loss": 0.1233,
+      "step": 295
+    },
+    {
+      "epoch": 1.1935483870967742,
+      "grad_norm": 2.839566469192505,
+      "learning_rate": 1.2043010752688173e-05,
+      "loss": 0.1286,
+      "step": 296
+    },
+    {
+      "epoch": 1.1975806451612903,
+      "grad_norm": 1.3624751567840576,
+      "learning_rate": 1.2016129032258067e-05,
+      "loss": 0.1815,
+      "step": 297
+    },
+    {
+      "epoch": 1.2016129032258065,
+      "grad_norm": 1.082029938697815,
+      "learning_rate": 1.1989247311827958e-05,
+      "loss": 0.1486,
+      "step": 298
+    },
+    {
+      "epoch": 1.2056451612903225,
+      "grad_norm": 1.6093990802764893,
+      "learning_rate": 1.1962365591397852e-05,
+      "loss": 0.176,
+      "step": 299
+    },
+    {
+      "epoch": 1.2096774193548387,
+      "grad_norm": 1.5934983491897583,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.1167,
+      "step": 300
+    },
+    {
+      "epoch": 1.2137096774193548,
+      "grad_norm": 1.5063905715942383,
+      "learning_rate": 1.1908602150537637e-05,
+      "loss": 0.1091,
+      "step": 301
+    },
+    {
+      "epoch": 1.217741935483871,
+      "grad_norm": 2.482288122177124,
+      "learning_rate": 1.1881720430107528e-05,
+      "loss": 0.171,
+      "step": 302
+    },
+    {
+      "epoch": 1.221774193548387,
+      "grad_norm": 2.164362907409668,
+      "learning_rate": 1.1854838709677421e-05,
+      "loss": 0.1996,
+      "step": 303
+    },
+    {
+      "epoch": 1.2258064516129032,
+      "grad_norm": 1.5960315465927124,
+      "learning_rate": 1.1827956989247313e-05,
+      "loss": 0.1211,
+      "step": 304
+    },
+    {
+      "epoch": 1.2298387096774193,
+      "grad_norm": 3.244338274002075,
+      "learning_rate": 1.1801075268817206e-05,
+      "loss": 0.1378,
+      "step": 305
+    },
+    {
+      "epoch": 1.2338709677419355,
+      "grad_norm": 1.7096904516220093,
+      "learning_rate": 1.1774193548387098e-05,
+      "loss": 0.1706,
+      "step": 306
+    },
+    {
+      "epoch": 1.2379032258064515,
+      "grad_norm": 1.980499505996704,
+      "learning_rate": 1.1747311827956991e-05,
+      "loss": 0.1233,
+      "step": 307
+    },
+    {
+      "epoch": 1.2419354838709677,
+      "grad_norm": 1.3346567153930664,
+      "learning_rate": 1.1720430107526883e-05,
+      "loss": 0.1496,
+      "step": 308
+    },
+    {
+      "epoch": 1.245967741935484,
+      "grad_norm": 1.437386393547058,
+      "learning_rate": 1.1693548387096776e-05,
+      "loss": 0.1306,
+      "step": 309
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.8189259767532349,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.1579,
+      "step": 310
+    },
+    {
+      "epoch": 1.254032258064516,
+      "grad_norm": 2.060695171356201,
+      "learning_rate": 1.1639784946236561e-05,
+      "loss": 0.1289,
+      "step": 311
+    },
+    {
+      "epoch": 1.2580645161290323,
+      "grad_norm": 2.4348952770233154,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.1002,
+      "step": 312
+    },
+    {
+      "epoch": 1.2620967741935485,
+      "grad_norm": 1.843311071395874,
+      "learning_rate": 1.1586021505376346e-05,
+      "loss": 0.1457,
+      "step": 313
+    },
+    {
+      "epoch": 1.2661290322580645,
+      "grad_norm": 2.4911584854125977,
+      "learning_rate": 1.1559139784946238e-05,
+      "loss": 0.1832,
+      "step": 314
+    },
+    {
+      "epoch": 1.2701612903225805,
+      "grad_norm": 4.787015438079834,
+      "learning_rate": 1.1532258064516131e-05,
+      "loss": 0.1762,
+      "step": 315
+    },
+    {
+      "epoch": 1.2741935483870968,
+      "grad_norm": 2.0192394256591797,
+      "learning_rate": 1.1505376344086023e-05,
+      "loss": 0.1286,
+      "step": 316
+    },
+    {
+      "epoch": 1.278225806451613,
+      "grad_norm": 0.9655114412307739,
+      "learning_rate": 1.1478494623655916e-05,
+      "loss": 0.0727,
+      "step": 317
+    },
+    {
+      "epoch": 1.282258064516129,
+      "grad_norm": 1.3966825008392334,
+      "learning_rate": 1.1451612903225808e-05,
+      "loss": 0.1365,
+      "step": 318
+    },
+    {
+      "epoch": 1.2862903225806452,
+      "grad_norm": 1.7151025533676147,
+      "learning_rate": 1.1424731182795701e-05,
+      "loss": 0.1583,
+      "step": 319
+    },
+    {
+      "epoch": 1.2903225806451613,
+      "grad_norm": 2.77651309967041,
+      "learning_rate": 1.1397849462365593e-05,
+      "loss": 0.0976,
+      "step": 320
+    },
+    {
+      "epoch": 1.2943548387096775,
+      "grad_norm": 1.6842812299728394,
+      "learning_rate": 1.1370967741935486e-05,
+      "loss": 0.0929,
+      "step": 321
+    },
+    {
+      "epoch": 1.2983870967741935,
+      "grad_norm": 1.3794842958450317,
+      "learning_rate": 1.1344086021505377e-05,
+      "loss": 0.2264,
+      "step": 322
+    },
+    {
+      "epoch": 1.3024193548387097,
+      "grad_norm": 1.75082266330719,
+      "learning_rate": 1.131720430107527e-05,
+      "loss": 0.1209,
+      "step": 323
+    },
+    {
+      "epoch": 1.3064516129032258,
+      "grad_norm": 3.0138330459594727,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.1394,
+      "step": 324
+    },
+    {
+      "epoch": 1.310483870967742,
+      "grad_norm": 1.3630236387252808,
+      "learning_rate": 1.1263440860215056e-05,
+      "loss": 0.0678,
+      "step": 325
+    },
+    {
+      "epoch": 1.314516129032258,
+      "grad_norm": 3.041550636291504,
+      "learning_rate": 1.1236559139784946e-05,
+      "loss": 0.2057,
+      "step": 326
+    },
+    {
+      "epoch": 1.3185483870967742,
+      "grad_norm": 1.457026720046997,
+      "learning_rate": 1.1209677419354839e-05,
+      "loss": 0.1104,
+      "step": 327
+    },
+    {
+      "epoch": 1.3225806451612903,
+      "grad_norm": 2.0328333377838135,
+      "learning_rate": 1.118279569892473e-05,
+      "loss": 0.13,
+      "step": 328
+    },
+    {
+      "epoch": 1.3266129032258065,
+      "grad_norm": 1.102391242980957,
+      "learning_rate": 1.1155913978494624e-05,
+      "loss": 0.1015,
+      "step": 329
+    },
+    {
+      "epoch": 1.3306451612903225,
+      "grad_norm": 1.9321506023406982,
+      "learning_rate": 1.1129032258064516e-05,
+      "loss": 0.085,
+      "step": 330
+    },
+    {
+      "epoch": 1.3346774193548387,
+      "grad_norm": 1.9749724864959717,
+      "learning_rate": 1.1102150537634409e-05,
+      "loss": 0.1343,
+      "step": 331
+    },
+    {
+      "epoch": 1.3387096774193548,
+      "grad_norm": 2.8448033332824707,
+      "learning_rate": 1.10752688172043e-05,
+      "loss": 0.1639,
+      "step": 332
+    },
+    {
+      "epoch": 1.342741935483871,
+      "grad_norm": 1.1829770803451538,
+      "learning_rate": 1.1048387096774194e-05,
+      "loss": 0.0986,
+      "step": 333
+    },
+    {
+      "epoch": 1.346774193548387,
+      "grad_norm": 3.9288711547851562,
+      "learning_rate": 1.1021505376344085e-05,
+      "loss": 0.1465,
+      "step": 334
+    },
+    {
+      "epoch": 1.3508064516129032,
+      "grad_norm": 2.407360553741455,
+      "learning_rate": 1.0994623655913979e-05,
+      "loss": 0.1261,
+      "step": 335
+    },
+    {
+      "epoch": 1.3548387096774195,
+      "grad_norm": 3.4650471210479736,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.1828,
+      "step": 336
+    },
+    {
+      "epoch": 1.3588709677419355,
+      "grad_norm": 2.9928531646728516,
+      "learning_rate": 1.0940860215053764e-05,
+      "loss": 0.1139,
+      "step": 337
+    },
+    {
+      "epoch": 1.3629032258064515,
+      "grad_norm": 4.049187183380127,
+      "learning_rate": 1.0913978494623655e-05,
+      "loss": 0.1622,
+      "step": 338
+    },
+    {
+      "epoch": 1.3669354838709677,
+      "grad_norm": 3.758409023284912,
+      "learning_rate": 1.0887096774193549e-05,
+      "loss": 0.1058,
+      "step": 339
+    },
+    {
+      "epoch": 1.370967741935484,
+      "grad_norm": 3.3022146224975586,
+      "learning_rate": 1.086021505376344e-05,
+      "loss": 0.1414,
+      "step": 340
+    },
+    {
+      "epoch": 1.375,
+      "grad_norm": 6.569581985473633,
+      "learning_rate": 1.0833333333333334e-05,
+      "loss": 0.2535,
+      "step": 341
+    },
+    {
+      "epoch": 1.379032258064516,
+      "grad_norm": 1.4087936878204346,
+      "learning_rate": 1.0806451612903225e-05,
+      "loss": 0.1063,
+      "step": 342
+    },
+    {
+      "epoch": 1.3830645161290323,
+      "grad_norm": 1.7777235507965088,
+      "learning_rate": 1.0779569892473118e-05,
+      "loss": 0.1408,
+      "step": 343
+    },
+    {
+      "epoch": 1.3870967741935485,
+      "grad_norm": 1.5745316743850708,
+      "learning_rate": 1.0752688172043012e-05,
+      "loss": 0.0839,
+      "step": 344
+    },
+    {
+      "epoch": 1.3911290322580645,
+      "grad_norm": 3.317399024963379,
+      "learning_rate": 1.0725806451612903e-05,
+      "loss": 0.1597,
+      "step": 345
+    },
+    {
+      "epoch": 1.3951612903225805,
+      "grad_norm": 2.1819188594818115,
+      "learning_rate": 1.0698924731182797e-05,
+      "loss": 0.0801,
+      "step": 346
+    },
+    {
+      "epoch": 1.3991935483870968,
+      "grad_norm": 3.0169897079467773,
+      "learning_rate": 1.0672043010752688e-05,
+      "loss": 0.1419,
+      "step": 347
+    },
+    {
+      "epoch": 1.403225806451613,
+      "grad_norm": 2.023202419281006,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.1038,
+      "step": 348
+    },
+    {
+      "epoch": 1.407258064516129,
+      "grad_norm": 1.4076378345489502,
+      "learning_rate": 1.0618279569892473e-05,
+      "loss": 0.1048,
+      "step": 349
+    },
+    {
+      "epoch": 1.4112903225806452,
+      "grad_norm": 2.562121629714966,
+      "learning_rate": 1.0591397849462367e-05,
+      "loss": 0.2181,
+      "step": 350
+    },
+    {
+      "epoch": 1.4153225806451613,
+      "grad_norm": 2.308161973953247,
+      "learning_rate": 1.0564516129032258e-05,
+      "loss": 0.1669,
+      "step": 351
+    },
+    {
+      "epoch": 1.4193548387096775,
+      "grad_norm": 1.25361168384552,
+      "learning_rate": 1.0537634408602151e-05,
+      "loss": 0.0966,
+      "step": 352
+    },
+    {
+      "epoch": 1.4233870967741935,
+      "grad_norm": 4.17478609085083,
+      "learning_rate": 1.0510752688172043e-05,
+      "loss": 0.169,
+      "step": 353
+    },
+    {
+      "epoch": 1.4274193548387097,
+      "grad_norm": 3.2306907176971436,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.1613,
+      "step": 354
+    },
+    {
+      "epoch": 1.4314516129032258,
+      "grad_norm": 4.039083957672119,
+      "learning_rate": 1.0456989247311828e-05,
+      "loss": 0.334,
+      "step": 355
+    },
+    {
+      "epoch": 1.435483870967742,
+      "grad_norm": 3.141918182373047,
+      "learning_rate": 1.0430107526881721e-05,
+      "loss": 0.1727,
+      "step": 356
+    },
+    {
+      "epoch": 1.439516129032258,
+      "grad_norm": 4.222794532775879,
+      "learning_rate": 1.0403225806451613e-05,
+      "loss": 0.1147,
+      "step": 357
+    },
+    {
+      "epoch": 1.4435483870967742,
+      "grad_norm": 1.3107784986495972,
+      "learning_rate": 1.0376344086021506e-05,
+      "loss": 0.0728,
+      "step": 358
+    },
+    {
+      "epoch": 1.4475806451612903,
+      "grad_norm": 5.24909782409668,
+      "learning_rate": 1.0349462365591398e-05,
+      "loss": 0.2182,
+      "step": 359
+    },
+    {
+      "epoch": 1.4516129032258065,
+      "grad_norm": 4.10903787612915,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.1699,
+      "step": 360
+    },
+    {
+      "epoch": 1.4556451612903225,
+      "grad_norm": 2.6097729206085205,
+      "learning_rate": 1.0295698924731183e-05,
+      "loss": 0.1172,
+      "step": 361
+    },
+    {
+      "epoch": 1.4596774193548387,
+      "grad_norm": 2.7892961502075195,
+      "learning_rate": 1.0268817204301076e-05,
+      "loss": 0.1382,
+      "step": 362
+    },
+    {
+      "epoch": 1.4637096774193548,
+      "grad_norm": 4.481143474578857,
+      "learning_rate": 1.0241935483870968e-05,
+      "loss": 0.171,
+      "step": 363
+    },
+    {
+      "epoch": 1.467741935483871,
+      "grad_norm": 2.8739893436431885,
+      "learning_rate": 1.0215053763440861e-05,
+      "loss": 0.0895,
+      "step": 364
+    },
+    {
+      "epoch": 1.471774193548387,
+      "grad_norm": 2.5068511962890625,
+      "learning_rate": 1.0188172043010753e-05,
+      "loss": 0.1144,
+      "step": 365
+    },
+    {
+      "epoch": 1.4758064516129032,
+      "grad_norm": 2.39939284324646,
+      "learning_rate": 1.0161290322580646e-05,
+      "loss": 0.1195,
+      "step": 366
+    },
+    {
+      "epoch": 1.4798387096774195,
+      "grad_norm": 2.4179749488830566,
+      "learning_rate": 1.0134408602150538e-05,
+      "loss": 0.1791,
+      "step": 367
+    },
+    {
+      "epoch": 1.4838709677419355,
+      "grad_norm": 1.944332480430603,
+      "learning_rate": 1.0107526881720431e-05,
+      "loss": 0.1656,
+      "step": 368
+    },
+    {
+      "epoch": 1.4879032258064515,
+      "grad_norm": 4.383450031280518,
+      "learning_rate": 1.0080645161290323e-05,
+      "loss": 0.1545,
+      "step": 369
+    },
+    {
+      "epoch": 1.4919354838709677,
+      "grad_norm": 1.622109293937683,
+      "learning_rate": 1.0053763440860216e-05,
+      "loss": 0.0967,
+      "step": 370
+    },
+    {
+      "epoch": 1.495967741935484,
+      "grad_norm": 1.5308200120925903,
+      "learning_rate": 1.0026881720430108e-05,
+      "loss": 0.0721,
+      "step": 371
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.9791924953460693,
+      "learning_rate": 1e-05,
+      "loss": 0.1107,
+      "step": 372
+    },
+    {
+      "epoch": 1.504032258064516,
+      "grad_norm": 2.284703254699707,
+      "learning_rate": 9.973118279569894e-06,
+      "loss": 0.1647,
+      "step": 373
+    },
+    {
+      "epoch": 1.5080645161290323,
+      "grad_norm": 3.273378849029541,
+      "learning_rate": 9.946236559139786e-06,
+      "loss": 0.1238,
+      "step": 374
+    },
+    {
+      "epoch": 1.5120967741935485,
+      "grad_norm": 1.6005984544754028,
+      "learning_rate": 9.919354838709679e-06,
+      "loss": 0.1222,
+      "step": 375
+    },
+    {
+      "epoch": 1.5161290322580645,
+      "grad_norm": 3.7144405841827393,
+      "learning_rate": 9.89247311827957e-06,
+      "loss": 0.1352,
+      "step": 376
+    },
+    {
+      "epoch": 1.5201612903225805,
+      "grad_norm": 2.2433135509490967,
+      "learning_rate": 9.865591397849464e-06,
+      "loss": 0.1353,
+      "step": 377
+    },
+    {
+      "epoch": 1.5241935483870968,
+      "grad_norm": 1.9060373306274414,
+      "learning_rate": 9.838709677419356e-06,
+      "loss": 0.0938,
+      "step": 378
+    },
+    {
+      "epoch": 1.528225806451613,
+      "grad_norm": 2.5664989948272705,
+      "learning_rate": 9.811827956989249e-06,
+      "loss": 0.1925,
+      "step": 379
+    },
+    {
+      "epoch": 1.532258064516129,
+      "grad_norm": 3.4527804851531982,
+      "learning_rate": 9.78494623655914e-06,
+      "loss": 0.17,
+      "step": 380
+    },
+    {
+      "epoch": 1.536290322580645,
+      "grad_norm": 1.0636487007141113,
+      "learning_rate": 9.758064516129034e-06,
+      "loss": 0.0896,
+      "step": 381
+    },
+    {
+      "epoch": 1.5403225806451613,
+      "grad_norm": 2.5372161865234375,
+      "learning_rate": 9.731182795698925e-06,
+      "loss": 0.2065,
+      "step": 382
+    },
+    {
+      "epoch": 1.5443548387096775,
+      "grad_norm": 3.2576494216918945,
+      "learning_rate": 9.704301075268819e-06,
+      "loss": 0.1855,
+      "step": 383
+    },
+    {
+      "epoch": 1.5483870967741935,
+      "grad_norm": 3.3559024333953857,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.2192,
+      "step": 384
+    },
+    {
+      "epoch": 1.5524193548387095,
+      "grad_norm": 3.2494945526123047,
+      "learning_rate": 9.650537634408604e-06,
+      "loss": 0.1617,
+      "step": 385
+    },
+    {
+      "epoch": 1.5564516129032258,
+      "grad_norm": 2.113792896270752,
+      "learning_rate": 9.623655913978495e-06,
+      "loss": 0.187,
+      "step": 386
+    },
+    {
+      "epoch": 1.560483870967742,
+      "grad_norm": 1.530720829963684,
+      "learning_rate": 9.596774193548389e-06,
+      "loss": 0.0767,
+      "step": 387
+    },
+    {
+      "epoch": 1.564516129032258,
+      "grad_norm": 1.9626264572143555,
+      "learning_rate": 9.56989247311828e-06,
+      "loss": 0.1725,
+      "step": 388
+    },
+    {
+      "epoch": 1.5685483870967742,
+      "grad_norm": 1.6587375402450562,
+      "learning_rate": 9.543010752688174e-06,
+      "loss": 0.0998,
+      "step": 389
+    },
+    {
+      "epoch": 1.5725806451612905,
+      "grad_norm": 1.3230797052383423,
+      "learning_rate": 9.516129032258065e-06,
+      "loss": 0.1522,
+      "step": 390
+    },
+    {
+      "epoch": 1.5766129032258065,
+      "grad_norm": 2.594186305999756,
+      "learning_rate": 9.489247311827959e-06,
+      "loss": 0.2221,
+      "step": 391
+    },
+    {
+      "epoch": 1.5806451612903225,
+      "grad_norm": 3.0997118949890137,
+      "learning_rate": 9.46236559139785e-06,
+      "loss": 0.1009,
+      "step": 392
+    },
+    {
+      "epoch": 1.5846774193548387,
+      "grad_norm": 4.0840325355529785,
+      "learning_rate": 9.435483870967743e-06,
+      "loss": 0.1231,
+      "step": 393
+    },
+    {
+      "epoch": 1.588709677419355,
+      "grad_norm": 3.59895658493042,
+      "learning_rate": 9.408602150537635e-06,
+      "loss": 0.1296,
+      "step": 394
+    },
+    {
+      "epoch": 1.592741935483871,
+      "grad_norm": 1.21365487575531,
+      "learning_rate": 9.381720430107528e-06,
+      "loss": 0.1062,
+      "step": 395
+    },
+    {
+      "epoch": 1.596774193548387,
+      "grad_norm": 2.8088643550872803,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.1877,
+      "step": 396
+    },
+    {
+      "epoch": 1.6008064516129032,
+      "grad_norm": 2.218878746032715,
+      "learning_rate": 9.327956989247312e-06,
+      "loss": 0.1258,
+      "step": 397
+    },
+    {
+      "epoch": 1.6048387096774195,
+      "grad_norm": 1.3823351860046387,
+      "learning_rate": 9.301075268817205e-06,
+      "loss": 0.1048,
+      "step": 398
+    },
+    {
+      "epoch": 1.6088709677419355,
+      "grad_norm": 2.248091220855713,
+      "learning_rate": 9.274193548387097e-06,
+      "loss": 0.2004,
+      "step": 399
+    },
+    {
+      "epoch": 1.6129032258064515,
+      "grad_norm": 2.475921392440796,
+      "learning_rate": 9.24731182795699e-06,
+      "loss": 0.1216,
+      "step": 400
+    },
+    {
+      "epoch": 1.6169354838709677,
+      "grad_norm": 1.0884982347488403,
+      "learning_rate": 9.220430107526881e-06,
+      "loss": 0.099,
+      "step": 401
+    },
+    {
+      "epoch": 1.620967741935484,
+      "grad_norm": 2.633883237838745,
+      "learning_rate": 9.193548387096775e-06,
+      "loss": 0.1015,
+      "step": 402
+    },
+    {
+      "epoch": 1.625,
+      "grad_norm": 1.0618445873260498,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 0.1017,
+      "step": 403
+    },
+    {
+      "epoch": 1.629032258064516,
+      "grad_norm": 6.406484127044678,
+      "learning_rate": 9.13978494623656e-06,
+      "loss": 0.159,
+      "step": 404
+    },
+    {
+      "epoch": 1.6330645161290323,
+      "grad_norm": 1.9261949062347412,
+      "learning_rate": 9.112903225806451e-06,
+      "loss": 0.1234,
+      "step": 405
+    },
+    {
+      "epoch": 1.6370967741935485,
+      "grad_norm": 4.344938278198242,
+      "learning_rate": 9.086021505376345e-06,
+      "loss": 0.1519,
+      "step": 406
+    },
+    {
+      "epoch": 1.6411290322580645,
+      "grad_norm": 1.627624273300171,
+      "learning_rate": 9.059139784946236e-06,
+      "loss": 0.1154,
+      "step": 407
+    },
+    {
+      "epoch": 1.6451612903225805,
+      "grad_norm": 1.8751394748687744,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.1693,
+      "step": 408
+    },
+    {
+      "epoch": 1.6491935483870968,
+      "grad_norm": 2.318906784057617,
+      "learning_rate": 9.005376344086021e-06,
+      "loss": 0.1491,
+      "step": 409
+    },
+    {
+      "epoch": 1.653225806451613,
+      "grad_norm": 5.5316853523254395,
+      "learning_rate": 8.978494623655915e-06,
+      "loss": 0.2173,
+      "step": 410
+    },
+    {
+      "epoch": 1.657258064516129,
+      "grad_norm": 2.1360862255096436,
+      "learning_rate": 8.951612903225806e-06,
+      "loss": 0.1482,
+      "step": 411
+    },
+    {
+      "epoch": 1.661290322580645,
+      "grad_norm": 1.136660099029541,
+      "learning_rate": 8.9247311827957e-06,
+      "loss": 0.1142,
+      "step": 412
+    },
+    {
+      "epoch": 1.6653225806451613,
+      "grad_norm": 5.797788143157959,
+      "learning_rate": 8.897849462365593e-06,
+      "loss": 0.1382,
+      "step": 413
+    },
+    {
+      "epoch": 1.6693548387096775,
+      "grad_norm": 4.6974287033081055,
+      "learning_rate": 8.870967741935484e-06,
+      "loss": 0.1369,
+      "step": 414
+    },
+    {
+      "epoch": 1.6733870967741935,
+      "grad_norm": 7.289107799530029,
+      "learning_rate": 8.844086021505378e-06,
+      "loss": 0.2441,
+      "step": 415
+    },
+    {
+      "epoch": 1.6774193548387095,
+      "grad_norm": 4.214983940124512,
+      "learning_rate": 8.81720430107527e-06,
+      "loss": 0.213,
+      "step": 416
+    },
+    {
+      "epoch": 1.6814516129032258,
+      "grad_norm": 3.4454896450042725,
+      "learning_rate": 8.790322580645163e-06,
+      "loss": 0.0803,
+      "step": 417
+    },
+    {
+      "epoch": 1.685483870967742,
+      "grad_norm": 2.8119094371795654,
+      "learning_rate": 8.763440860215054e-06,
+      "loss": 0.1307,
+      "step": 418
+    },
+    {
+      "epoch": 1.689516129032258,
+      "grad_norm": 3.920039653778076,
+      "learning_rate": 8.736559139784948e-06,
+      "loss": 0.1729,
+      "step": 419
+    },
+    {
+      "epoch": 1.6935483870967742,
+      "grad_norm": 4.052835941314697,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.0988,
+      "step": 420
+    },
+    {
+      "epoch": 1.6975806451612905,
+      "grad_norm": 2.297703742980957,
+      "learning_rate": 8.682795698924733e-06,
+      "loss": 0.1176,
+      "step": 421
+    },
+    {
+      "epoch": 1.7016129032258065,
+      "grad_norm": 2.6884663105010986,
+      "learning_rate": 8.655913978494624e-06,
+      "loss": 0.1113,
+      "step": 422
+    },
+    {
+      "epoch": 1.7056451612903225,
+      "grad_norm": 2.510374069213867,
+      "learning_rate": 8.629032258064517e-06,
+      "loss": 0.1428,
+      "step": 423
+    },
+    {
+      "epoch": 1.7096774193548387,
+      "grad_norm": 1.734438419342041,
+      "learning_rate": 8.602150537634409e-06,
+      "loss": 0.0651,
+      "step": 424
+    },
+    {
+      "epoch": 1.713709677419355,
+      "grad_norm": 2.6852872371673584,
+      "learning_rate": 8.575268817204302e-06,
+      "loss": 0.1801,
+      "step": 425
+    },
+    {
+      "epoch": 1.717741935483871,
+      "grad_norm": 1.1802809238433838,
+      "learning_rate": 8.548387096774194e-06,
+      "loss": 0.1295,
+      "step": 426
+    },
+    {
+      "epoch": 1.721774193548387,
+      "grad_norm": 3.5888192653656006,
+      "learning_rate": 8.521505376344087e-06,
+      "loss": 0.1373,
+      "step": 427
+    },
+    {
+      "epoch": 1.7258064516129032,
+      "grad_norm": 3.463402271270752,
+      "learning_rate": 8.494623655913979e-06,
+      "loss": 0.104,
+      "step": 428
+    },
+    {
+      "epoch": 1.7298387096774195,
+      "grad_norm": 2.838108777999878,
+      "learning_rate": 8.467741935483872e-06,
+      "loss": 0.1541,
+      "step": 429
+    },
+    {
+      "epoch": 1.7338709677419355,
+      "grad_norm": 3.127835512161255,
+      "learning_rate": 8.440860215053764e-06,
+      "loss": 0.1103,
+      "step": 430
+    },
+    {
+      "epoch": 1.7379032258064515,
+      "grad_norm": 2.3887135982513428,
+      "learning_rate": 8.413978494623657e-06,
+      "loss": 0.1686,
+      "step": 431
+    },
+    {
+      "epoch": 1.7419354838709677,
+      "grad_norm": 2.5920352935791016,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.1187,
+      "step": 432
+    },
+    {
+      "epoch": 1.745967741935484,
+      "grad_norm": 4.120633125305176,
+      "learning_rate": 8.360215053763442e-06,
+      "loss": 0.1966,
+      "step": 433
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.4372769594192505,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.1419,
+      "step": 434
+    },
+    {
+      "epoch": 1.754032258064516,
+      "grad_norm": 2.421602964401245,
+      "learning_rate": 8.306451612903227e-06,
+      "loss": 0.145,
+      "step": 435
+    },
+    {
+      "epoch": 1.7580645161290323,
+      "grad_norm": 2.807343006134033,
+      "learning_rate": 8.279569892473119e-06,
+      "loss": 0.0802,
+      "step": 436
+    },
+    {
+      "epoch": 1.7620967741935485,
+      "grad_norm": 2.304760217666626,
+      "learning_rate": 8.252688172043012e-06,
+      "loss": 0.1192,
+      "step": 437
+    },
+    {
+      "epoch": 1.7661290322580645,
+      "grad_norm": 2.381269693374634,
+      "learning_rate": 8.225806451612904e-06,
+      "loss": 0.1027,
+      "step": 438
+    },
+    {
+      "epoch": 1.7701612903225805,
+      "grad_norm": 2.1986842155456543,
+      "learning_rate": 8.198924731182797e-06,
+      "loss": 0.1012,
+      "step": 439
+    },
+    {
+      "epoch": 1.7741935483870968,
+      "grad_norm": 2.638580560684204,
+      "learning_rate": 8.172043010752689e-06,
+      "loss": 0.1746,
+      "step": 440
+    },
+    {
+      "epoch": 1.778225806451613,
+      "grad_norm": 4.079723358154297,
+      "learning_rate": 8.145161290322582e-06,
+      "loss": 0.1684,
+      "step": 441
+    },
+    {
+      "epoch": 1.782258064516129,
+      "grad_norm": 1.7967092990875244,
+      "learning_rate": 8.118279569892473e-06,
+      "loss": 0.2538,
+      "step": 442
+    },
+    {
+      "epoch": 1.786290322580645,
+      "grad_norm": 2.0538618564605713,
+      "learning_rate": 8.091397849462365e-06,
+      "loss": 0.1666,
+      "step": 443
+    },
+    {
+      "epoch": 1.7903225806451613,
+      "grad_norm": 3.8617451190948486,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.1725,
+      "step": 444
+    },
+    {
+      "epoch": 1.7943548387096775,
+      "grad_norm": 3.2343311309814453,
+      "learning_rate": 8.03763440860215e-06,
+      "loss": 0.1595,
+      "step": 445
+    },
+    {
+      "epoch": 1.7983870967741935,
+      "grad_norm": 2.797788619995117,
+      "learning_rate": 8.010752688172043e-06,
+      "loss": 0.0903,
+      "step": 446
+    },
+    {
+      "epoch": 1.8024193548387095,
+      "grad_norm": 3.4088759422302246,
+      "learning_rate": 7.983870967741935e-06,
+      "loss": 0.1113,
+      "step": 447
+    },
+    {
+      "epoch": 1.8064516129032258,
+      "grad_norm": 3.5541694164276123,
+      "learning_rate": 7.956989247311828e-06,
+      "loss": 0.1433,
+      "step": 448
+    },
+    {
+      "epoch": 1.810483870967742,
+      "grad_norm": 1.529510736465454,
+      "learning_rate": 7.93010752688172e-06,
+      "loss": 0.0728,
+      "step": 449
+    },
+    {
+      "epoch": 1.814516129032258,
+      "grad_norm": 1.9386626482009888,
+      "learning_rate": 7.903225806451613e-06,
+      "loss": 0.1112,
+      "step": 450
+    },
+    {
+      "epoch": 1.8185483870967742,
+      "grad_norm": 2.3068759441375732,
+      "learning_rate": 7.876344086021507e-06,
+      "loss": 0.0967,
+      "step": 451
+    },
+    {
+      "epoch": 1.8225806451612905,
+      "grad_norm": 2.4969186782836914,
+      "learning_rate": 7.849462365591398e-06,
+      "loss": 0.0896,
+      "step": 452
+    },
+    {
+      "epoch": 1.8266129032258065,
+      "grad_norm": 2.4641716480255127,
+      "learning_rate": 7.822580645161291e-06,
+      "loss": 0.1365,
+      "step": 453
+    },
+    {
+      "epoch": 1.8306451612903225,
+      "grad_norm": 4.361322402954102,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.1544,
+      "step": 454
+    },
+    {
+      "epoch": 1.8346774193548387,
+      "grad_norm": 1.1930088996887207,
+      "learning_rate": 7.768817204301076e-06,
+      "loss": 0.0941,
+      "step": 455
+    },
+    {
+      "epoch": 1.838709677419355,
+      "grad_norm": 1.9687577486038208,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.114,
+      "step": 456
+    },
+    {
+      "epoch": 1.842741935483871,
+      "grad_norm": 4.404200077056885,
+      "learning_rate": 7.715053763440861e-06,
+      "loss": 0.153,
+      "step": 457
+    },
+    {
+      "epoch": 1.846774193548387,
+      "grad_norm": 0.9312377572059631,
+      "learning_rate": 7.688172043010753e-06,
+      "loss": 0.1008,
+      "step": 458
+    },
+    {
+      "epoch": 1.8508064516129032,
+      "grad_norm": 2.732656717300415,
+      "learning_rate": 7.661290322580646e-06,
+      "loss": 0.0629,
+      "step": 459
+    },
+    {
+      "epoch": 1.8548387096774195,
+      "grad_norm": 1.351778268814087,
+      "learning_rate": 7.634408602150538e-06,
+      "loss": 0.0842,
+      "step": 460
+    },
+    {
+      "epoch": 1.8588709677419355,
+      "grad_norm": 1.4621678590774536,
+      "learning_rate": 7.60752688172043e-06,
+      "loss": 0.0929,
+      "step": 461
+    },
+    {
+      "epoch": 1.8629032258064515,
+      "grad_norm": 2.530639171600342,
+      "learning_rate": 7.580645161290323e-06,
+      "loss": 0.1128,
+      "step": 462
+    },
+    {
+      "epoch": 1.8669354838709677,
+      "grad_norm": 1.704332709312439,
+      "learning_rate": 7.553763440860215e-06,
+      "loss": 0.1268,
+      "step": 463
+    },
+    {
+      "epoch": 1.870967741935484,
+      "grad_norm": 1.6047230958938599,
+      "learning_rate": 7.526881720430108e-06,
+      "loss": 0.1598,
+      "step": 464
+    },
+    {
+      "epoch": 1.875,
+      "grad_norm": 1.620526909828186,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.0815,
+      "step": 465
+    },
+    {
+      "epoch": 1.879032258064516,
+      "grad_norm": 3.030583620071411,
+      "learning_rate": 7.4731182795698935e-06,
+      "loss": 0.1197,
+      "step": 466
+    },
+    {
+      "epoch": 1.8830645161290323,
+      "grad_norm": 2.4432520866394043,
+      "learning_rate": 7.446236559139786e-06,
+      "loss": 0.2071,
+      "step": 467
+    },
+    {
+      "epoch": 1.8870967741935485,
+      "grad_norm": 1.968050241470337,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.0968,
+      "step": 468
+    },
+    {
+      "epoch": 1.8911290322580645,
+      "grad_norm": 1.4884843826293945,
+      "learning_rate": 7.392473118279571e-06,
+      "loss": 0.0825,
+      "step": 469
+    },
+    {
+      "epoch": 1.8951612903225805,
+      "grad_norm": 1.9214873313903809,
+      "learning_rate": 7.365591397849463e-06,
+      "loss": 0.1782,
+      "step": 470
+    },
+    {
+      "epoch": 1.8991935483870968,
+      "grad_norm": 2.7788476943969727,
+      "learning_rate": 7.338709677419356e-06,
+      "loss": 0.1283,
+      "step": 471
+    },
+    {
+      "epoch": 1.903225806451613,
+      "grad_norm": 5.547091484069824,
+      "learning_rate": 7.311827956989248e-06,
+      "loss": 0.1202,
+      "step": 472
+    },
+    {
+      "epoch": 1.907258064516129,
+      "grad_norm": 1.0326168537139893,
+      "learning_rate": 7.284946236559141e-06,
+      "loss": 0.1248,
+      "step": 473
+    },
+    {
+      "epoch": 1.911290322580645,
+      "grad_norm": 2.0643792152404785,
+      "learning_rate": 7.258064516129033e-06,
+      "loss": 0.1846,
+      "step": 474
+    },
+    {
+      "epoch": 1.9153225806451613,
+      "grad_norm": 3.634772300720215,
+      "learning_rate": 7.231182795698926e-06,
+      "loss": 0.1574,
+      "step": 475
+    },
+    {
+      "epoch": 1.9193548387096775,
+      "grad_norm": 2.878852605819702,
+      "learning_rate": 7.204301075268818e-06,
+      "loss": 0.1066,
+      "step": 476
+    },
+    {
+      "epoch": 1.9233870967741935,
+      "grad_norm": 2.5109782218933105,
+      "learning_rate": 7.177419354838711e-06,
+      "loss": 0.1653,
+      "step": 477
+    },
+    {
+      "epoch": 1.9274193548387095,
+      "grad_norm": 3.136270761489868,
+      "learning_rate": 7.150537634408603e-06,
+      "loss": 0.1387,
+      "step": 478
+    },
+    {
+      "epoch": 1.9314516129032258,
+      "grad_norm": 1.636473536491394,
+      "learning_rate": 7.1236559139784956e-06,
+      "loss": 0.2813,
+      "step": 479
+    },
+    {
+      "epoch": 1.935483870967742,
+      "grad_norm": 1.7131333351135254,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.0882,
+      "step": 480
+    },
+    {
+      "epoch": 1.939516129032258,
+      "grad_norm": 1.6337999105453491,
+      "learning_rate": 7.0698924731182805e-06,
+      "loss": 0.182,
+      "step": 481
+    },
+    {
+      "epoch": 1.9435483870967742,
+      "grad_norm": 3.1437745094299316,
+      "learning_rate": 7.043010752688173e-06,
+      "loss": 0.1109,
+      "step": 482
+    },
+    {
+      "epoch": 1.9475806451612905,
+      "grad_norm": 2.556929588317871,
+      "learning_rate": 7.0161290322580654e-06,
+      "loss": 0.1183,
+      "step": 483
+    },
+    {
+      "epoch": 1.9516129032258065,
+      "grad_norm": 1.5759491920471191,
+      "learning_rate": 6.989247311827958e-06,
+      "loss": 0.1137,
+      "step": 484
+    },
+    {
+      "epoch": 1.9556451612903225,
+      "grad_norm": 3.2312610149383545,
+      "learning_rate": 6.96236559139785e-06,
+      "loss": 0.1508,
+      "step": 485
+    },
+    {
+      "epoch": 1.9596774193548387,
+      "grad_norm": 1.5898686647415161,
+      "learning_rate": 6.935483870967743e-06,
+      "loss": 0.1276,
+      "step": 486
+    },
+    {
+      "epoch": 1.963709677419355,
+      "grad_norm": 2.8799052238464355,
+      "learning_rate": 6.908602150537635e-06,
+      "loss": 0.118,
+      "step": 487
+    },
+    {
+      "epoch": 1.967741935483871,
+      "grad_norm": 3.8616013526916504,
+      "learning_rate": 6.881720430107528e-06,
+      "loss": 0.1884,
+      "step": 488
+    },
+    {
+      "epoch": 1.971774193548387,
+      "grad_norm": 1.7562378644943237,
+      "learning_rate": 6.854838709677419e-06,
+      "loss": 0.1048,
+      "step": 489
+    },
+    {
+      "epoch": 1.9758064516129032,
+      "grad_norm": 1.8255165815353394,
+      "learning_rate": 6.827956989247312e-06,
+      "loss": 0.1632,
+      "step": 490
+    },
+    {
+      "epoch": 1.9798387096774195,
+      "grad_norm": 1.126791000366211,
+      "learning_rate": 6.801075268817204e-06,
+      "loss": 0.1529,
+      "step": 491
+    },
+    {
+      "epoch": 1.9838709677419355,
+      "grad_norm": 3.804797649383545,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.1882,
+      "step": 492
+    },
+    {
+      "epoch": 1.9879032258064515,
+      "grad_norm": 2.6595122814178467,
+      "learning_rate": 6.747311827956989e-06,
+      "loss": 0.1564,
+      "step": 493
+    },
+    {
+      "epoch": 1.9919354838709677,
+      "grad_norm": 1.5479234457015991,
+      "learning_rate": 6.720430107526882e-06,
+      "loss": 0.1166,
+      "step": 494
+    },
+    {
+      "epoch": 1.995967741935484,
+      "grad_norm": 2.059999704360962,
+      "learning_rate": 6.693548387096774e-06,
+      "loss": 0.1269,
+      "step": 495
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 3.3951058387756348,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.153,
+      "step": 496
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.7477272727272727,
+      "eval_f1": 0.8979367262723521,
+      "eval_loss": 0.1496548354625702,
+      "eval_runtime": 2.3658,
+      "eval_samples_per_second": 371.972,
+      "eval_steps_per_second": 11.835,
+      "step": 496
+    },
+    {
+      "epoch": 2.004032258064516,
+      "grad_norm": 1.5380765199661255,
+      "learning_rate": 6.639784946236559e-06,
+      "loss": 0.09,
+      "step": 497
+    },
+    {
+      "epoch": 2.0080645161290325,
+      "grad_norm": 2.340376615524292,
+      "learning_rate": 6.612903225806452e-06,
+      "loss": 0.1141,
+      "step": 498
+    },
+    {
+      "epoch": 2.0120967741935485,
+      "grad_norm": 1.2891623973846436,
+      "learning_rate": 6.586021505376344e-06,
+      "loss": 0.1402,
+      "step": 499
+    },
+    {
+      "epoch": 2.0161290322580645,
+      "grad_norm": 2.0554511547088623,
+      "learning_rate": 6.5591397849462365e-06,
+      "loss": 0.1812,
+      "step": 500
+    },
+    {
+      "epoch": 2.0201612903225805,
+      "grad_norm": 3.6273601055145264,
+      "learning_rate": 6.532258064516129e-06,
+      "loss": 0.1321,
+      "step": 501
+    },
+    {
+      "epoch": 2.024193548387097,
+      "grad_norm": 2.3132970333099365,
+      "learning_rate": 6.5053763440860214e-06,
+      "loss": 0.1039,
+      "step": 502
+    },
+    {
+      "epoch": 2.028225806451613,
+      "grad_norm": 1.4552321434020996,
+      "learning_rate": 6.478494623655914e-06,
+      "loss": 0.1449,
+      "step": 503
+    },
+    {
+      "epoch": 2.032258064516129,
+      "grad_norm": 2.3734288215637207,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.1196,
+      "step": 504
+    },
+    {
+      "epoch": 2.036290322580645,
+      "grad_norm": 3.308845043182373,
+      "learning_rate": 6.4247311827957e-06,
+      "loss": 0.1678,
+      "step": 505
+    },
+    {
+      "epoch": 2.0403225806451615,
+      "grad_norm": 2.2413101196289062,
+      "learning_rate": 6.397849462365592e-06,
+      "loss": 0.1195,
+      "step": 506
+    },
+    {
+      "epoch": 2.0443548387096775,
+      "grad_norm": 2.4938807487487793,
+      "learning_rate": 6.370967741935485e-06,
+      "loss": 0.1344,
+      "step": 507
+    },
+    {
+      "epoch": 2.0483870967741935,
+      "grad_norm": 1.7536572217941284,
+      "learning_rate": 6.344086021505377e-06,
+      "loss": 0.1037,
+      "step": 508
+    },
+    {
+      "epoch": 2.0524193548387095,
+      "grad_norm": 5.4296345710754395,
+      "learning_rate": 6.3172043010752696e-06,
+      "loss": 0.1651,
+      "step": 509
+    },
+    {
+      "epoch": 2.056451612903226,
+      "grad_norm": 1.4947242736816406,
+      "learning_rate": 6.290322580645162e-06,
+      "loss": 0.1779,
+      "step": 510
+    },
+    {
+      "epoch": 2.060483870967742,
+      "grad_norm": 2.460106372833252,
+      "learning_rate": 6.2634408602150545e-06,
+      "loss": 0.1335,
+      "step": 511
+    },
+    {
+      "epoch": 2.064516129032258,
+      "grad_norm": 2.0023834705352783,
+      "learning_rate": 6.236559139784947e-06,
+      "loss": 0.1134,
+      "step": 512
+    },
+    {
+      "epoch": 2.068548387096774,
+      "grad_norm": 9.138616561889648,
+      "learning_rate": 6.209677419354839e-06,
+      "loss": 0.1426,
+      "step": 513
+    },
+    {
+      "epoch": 2.0725806451612905,
+      "grad_norm": 2.8430464267730713,
+      "learning_rate": 6.182795698924732e-06,
+      "loss": 0.1099,
+      "step": 514
+    },
+    {
+      "epoch": 2.0766129032258065,
+      "grad_norm": 2.494035482406616,
+      "learning_rate": 6.155913978494624e-06,
+      "loss": 0.1609,
+      "step": 515
+    },
+    {
+      "epoch": 2.0806451612903225,
+      "grad_norm": 0.7258223295211792,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.1089,
+      "step": 516
+    },
+    {
+      "epoch": 2.0846774193548385,
+      "grad_norm": 2.798238754272461,
+      "learning_rate": 6.102150537634409e-06,
+      "loss": 0.1433,
+      "step": 517
+    },
+    {
+      "epoch": 2.088709677419355,
+      "grad_norm": 1.2602635622024536,
+      "learning_rate": 6.075268817204302e-06,
+      "loss": 0.0947,
+      "step": 518
+    },
+    {
+      "epoch": 2.092741935483871,
+      "grad_norm": 2.4262948036193848,
+      "learning_rate": 6.048387096774194e-06,
+      "loss": 0.1297,
+      "step": 519
+    },
+    {
+      "epoch": 2.096774193548387,
+      "grad_norm": 1.0531463623046875,
+      "learning_rate": 6.021505376344087e-06,
+      "loss": 0.0798,
+      "step": 520
+    },
+    {
+      "epoch": 2.100806451612903,
+      "grad_norm": 2.968100070953369,
+      "learning_rate": 5.994623655913979e-06,
+      "loss": 0.1638,
+      "step": 521
+    },
+    {
+      "epoch": 2.1048387096774195,
+      "grad_norm": 1.3490394353866577,
+      "learning_rate": 5.967741935483872e-06,
+      "loss": 0.1011,
+      "step": 522
+    },
+    {
+      "epoch": 2.1088709677419355,
+      "grad_norm": 1.9310318231582642,
+      "learning_rate": 5.940860215053764e-06,
+      "loss": 0.1442,
+      "step": 523
+    },
+    {
+      "epoch": 2.1129032258064515,
+      "grad_norm": 5.254608631134033,
+      "learning_rate": 5.9139784946236566e-06,
+      "loss": 0.1855,
+      "step": 524
+    },
+    {
+      "epoch": 2.1169354838709675,
+      "grad_norm": 1.3291866779327393,
+      "learning_rate": 5.887096774193549e-06,
+      "loss": 0.1166,
+      "step": 525
+    },
+    {
+      "epoch": 2.120967741935484,
+      "grad_norm": 1.8239635229110718,
+      "learning_rate": 5.8602150537634415e-06,
+      "loss": 0.1574,
+      "step": 526
+    },
+    {
+      "epoch": 2.125,
+      "grad_norm": 1.7408251762390137,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 0.073,
+      "step": 527
+    },
+    {
+      "epoch": 2.129032258064516,
+      "grad_norm": 1.5561959743499756,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.0942,
+      "step": 528
+    },
+    {
+      "epoch": 2.133064516129032,
+      "grad_norm": 1.163049340248108,
+      "learning_rate": 5.779569892473119e-06,
+      "loss": 0.1472,
+      "step": 529
+    },
+    {
+      "epoch": 2.1370967741935485,
+      "grad_norm": 2.868626356124878,
+      "learning_rate": 5.752688172043011e-06,
+      "loss": 0.1565,
+      "step": 530
+    },
+    {
+      "epoch": 2.1411290322580645,
+      "grad_norm": 4.834112644195557,
+      "learning_rate": 5.725806451612904e-06,
+      "loss": 0.1909,
+      "step": 531
+    },
+    {
+      "epoch": 2.1451612903225805,
+      "grad_norm": 3.0024447441101074,
+      "learning_rate": 5.698924731182796e-06,
+      "loss": 0.1108,
+      "step": 532
+    },
+    {
+      "epoch": 2.149193548387097,
+      "grad_norm": 5.311757564544678,
+      "learning_rate": 5.672043010752689e-06,
+      "loss": 0.1366,
+      "step": 533
+    },
+    {
+      "epoch": 2.153225806451613,
+      "grad_norm": 3.452758312225342,
+      "learning_rate": 5.645161290322582e-06,
+      "loss": 0.2077,
+      "step": 534
+    },
+    {
+      "epoch": 2.157258064516129,
+      "grad_norm": 1.520835280418396,
+      "learning_rate": 5.618279569892473e-06,
+      "loss": 0.0959,
+      "step": 535
+    },
+    {
+      "epoch": 2.161290322580645,
+      "grad_norm": 3.092453718185425,
+      "learning_rate": 5.591397849462365e-06,
+      "loss": 0.1078,
+      "step": 536
+    },
+    {
+      "epoch": 2.1653225806451615,
+      "grad_norm": 3.7799909114837646,
+      "learning_rate": 5.564516129032258e-06,
+      "loss": 0.0856,
+      "step": 537
+    },
+    {
+      "epoch": 2.1693548387096775,
+      "grad_norm": 1.4819005727767944,
+      "learning_rate": 5.53763440860215e-06,
+      "loss": 0.1015,
+      "step": 538
+    },
+    {
+      "epoch": 2.1733870967741935,
+      "grad_norm": 2.0473012924194336,
+      "learning_rate": 5.510752688172043e-06,
+      "loss": 0.1372,
+      "step": 539
+    },
+    {
+      "epoch": 2.1774193548387095,
+      "grad_norm": 3.29624080657959,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.1395,
+      "step": 540
+    },
+    {
+      "epoch": 2.181451612903226,
+      "grad_norm": 1.8019927740097046,
+      "learning_rate": 5.456989247311828e-06,
+      "loss": 0.1426,
+      "step": 541
+    },
+    {
+      "epoch": 2.185483870967742,
+      "grad_norm": 1.848576307296753,
+      "learning_rate": 5.43010752688172e-06,
+      "loss": 0.1854,
+      "step": 542
+    },
+    {
+      "epoch": 2.189516129032258,
+      "grad_norm": 1.5595825910568237,
+      "learning_rate": 5.4032258064516126e-06,
+      "loss": 0.0969,
+      "step": 543
+    },
+    {
+      "epoch": 2.193548387096774,
+      "grad_norm": 1.914633870124817,
+      "learning_rate": 5.376344086021506e-06,
+      "loss": 0.1466,
+      "step": 544
+    },
+    {
+      "epoch": 2.1975806451612905,
+      "grad_norm": 1.185180902481079,
+      "learning_rate": 5.349462365591398e-06,
+      "loss": 0.1457,
+      "step": 545
+    },
+    {
+      "epoch": 2.2016129032258065,
+      "grad_norm": 4.04098653793335,
+      "learning_rate": 5.322580645161291e-06,
+      "loss": 0.0852,
+      "step": 546
+    },
+    {
+      "epoch": 2.2056451612903225,
+      "grad_norm": 1.7851275205612183,
+      "learning_rate": 5.295698924731183e-06,
+      "loss": 0.1771,
+      "step": 547
+    },
+    {
+      "epoch": 2.2096774193548385,
+      "grad_norm": 2.551523208618164,
+      "learning_rate": 5.268817204301076e-06,
+      "loss": 0.1229,
+      "step": 548
+    },
+    {
+      "epoch": 2.213709677419355,
+      "grad_norm": 1.8646845817565918,
+      "learning_rate": 5.241935483870968e-06,
+      "loss": 0.0997,
+      "step": 549
+    },
+    {
+      "epoch": 2.217741935483871,
+      "grad_norm": 3.173009157180786,
+      "learning_rate": 5.215053763440861e-06,
+      "loss": 0.1498,
+      "step": 550
+    },
+    {
+      "epoch": 2.221774193548387,
+      "grad_norm": 2.938751220703125,
+      "learning_rate": 5.188172043010753e-06,
+      "loss": 0.1432,
+      "step": 551
+    },
+    {
+      "epoch": 2.225806451612903,
+      "grad_norm": 1.863869071006775,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.1246,
+      "step": 552
+    },
+    {
+      "epoch": 2.2298387096774195,
+      "grad_norm": 2.414386034011841,
+      "learning_rate": 5.134408602150538e-06,
+      "loss": 0.137,
+      "step": 553
+    },
+    {
+      "epoch": 2.2338709677419355,
+      "grad_norm": 1.476468563079834,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.1444,
+      "step": 554
+    },
+    {
+      "epoch": 2.2379032258064515,
+      "grad_norm": 1.2723708152770996,
+      "learning_rate": 5.080645161290323e-06,
+      "loss": 0.0898,
+      "step": 555
+    },
+    {
+      "epoch": 2.241935483870968,
+      "grad_norm": 2.391453742980957,
+      "learning_rate": 5.0537634408602155e-06,
+      "loss": 0.0788,
+      "step": 556
+    },
+    {
+      "epoch": 2.245967741935484,
+      "grad_norm": 1.704126000404358,
+      "learning_rate": 5.026881720430108e-06,
+      "loss": 0.1598,
+      "step": 557
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 2.0079946517944336,
+      "learning_rate": 5e-06,
+      "loss": 0.115,
+      "step": 558
+    },
+    {
+      "epoch": 2.254032258064516,
+      "grad_norm": 2.3702168464660645,
+      "learning_rate": 4.973118279569893e-06,
+      "loss": 0.1229,
+      "step": 559
+    },
+    {
+      "epoch": 2.258064516129032,
+      "grad_norm": 4.38910436630249,
+      "learning_rate": 4.946236559139785e-06,
+      "loss": 0.1102,
+      "step": 560
+    },
+    {
+      "epoch": 2.2620967741935485,
+      "grad_norm": 4.0716776847839355,
+      "learning_rate": 4.919354838709678e-06,
+      "loss": 0.2517,
+      "step": 561
+    },
+    {
+      "epoch": 2.2661290322580645,
+      "grad_norm": 0.9846423268318176,
+      "learning_rate": 4.89247311827957e-06,
+      "loss": 0.1007,
+      "step": 562
+    },
+    {
+      "epoch": 2.2701612903225805,
+      "grad_norm": 4.429399490356445,
+      "learning_rate": 4.865591397849463e-06,
+      "loss": 0.0691,
+      "step": 563
+    },
+    {
+      "epoch": 2.274193548387097,
+      "grad_norm": 2.5109846591949463,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.1717,
+      "step": 564
+    },
+    {
+      "epoch": 2.278225806451613,
+      "grad_norm": 2.3060779571533203,
+      "learning_rate": 4.811827956989248e-06,
+      "loss": 0.0871,
+      "step": 565
+    },
+    {
+      "epoch": 2.282258064516129,
+      "grad_norm": 2.165689706802368,
+      "learning_rate": 4.78494623655914e-06,
+      "loss": 0.1648,
+      "step": 566
+    },
+    {
+      "epoch": 2.286290322580645,
+      "grad_norm": 3.6668314933776855,
+      "learning_rate": 4.758064516129033e-06,
+      "loss": 0.093,
+      "step": 567
+    },
+    {
+      "epoch": 2.2903225806451615,
+      "grad_norm": 3.3539183139801025,
+      "learning_rate": 4.731182795698925e-06,
+      "loss": 0.1632,
+      "step": 568
+    },
+    {
+      "epoch": 2.2943548387096775,
+      "grad_norm": 2.386240243911743,
+      "learning_rate": 4.7043010752688175e-06,
+      "loss": 0.2024,
+      "step": 569
+    },
+    {
+      "epoch": 2.2983870967741935,
+      "grad_norm": 1.5170916318893433,
+      "learning_rate": 4.67741935483871e-06,
+      "loss": 0.1433,
+      "step": 570
+    },
+    {
+      "epoch": 2.3024193548387095,
+      "grad_norm": 1.2362958192825317,
+      "learning_rate": 4.6505376344086025e-06,
+      "loss": 0.0842,
+      "step": 571
+    },
+    {
+      "epoch": 2.306451612903226,
+      "grad_norm": 2.1873104572296143,
+      "learning_rate": 4.623655913978495e-06,
+      "loss": 0.1048,
+      "step": 572
+    },
+    {
+      "epoch": 2.310483870967742,
+      "grad_norm": 1.4580453634262085,
+      "learning_rate": 4.596774193548387e-06,
+      "loss": 0.0954,
+      "step": 573
+    },
+    {
+      "epoch": 2.314516129032258,
+      "grad_norm": 1.951701283454895,
+      "learning_rate": 4.56989247311828e-06,
+      "loss": 0.2442,
+      "step": 574
+    },
+    {
+      "epoch": 2.318548387096774,
+      "grad_norm": 1.1734662055969238,
+      "learning_rate": 4.543010752688172e-06,
+      "loss": 0.1161,
+      "step": 575
+    },
+    {
+      "epoch": 2.3225806451612905,
+      "grad_norm": 2.285795211791992,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.1039,
+      "step": 576
+    },
+    {
+      "epoch": 2.3266129032258065,
+      "grad_norm": 1.4747077226638794,
+      "learning_rate": 4.489247311827957e-06,
+      "loss": 0.1783,
+      "step": 577
+    },
+    {
+      "epoch": 2.3306451612903225,
+      "grad_norm": 1.9864970445632935,
+      "learning_rate": 4.46236559139785e-06,
+      "loss": 0.1308,
+      "step": 578
+    },
+    {
+      "epoch": 2.3346774193548385,
+      "grad_norm": 1.8854461908340454,
+      "learning_rate": 4.435483870967742e-06,
+      "loss": 0.1359,
+      "step": 579
+    },
+    {
+      "epoch": 2.338709677419355,
+      "grad_norm": 1.6464263200759888,
+      "learning_rate": 4.408602150537635e-06,
+      "loss": 0.1447,
+      "step": 580
+    },
+    {
+      "epoch": 2.342741935483871,
+      "grad_norm": 1.9216904640197754,
+      "learning_rate": 4.381720430107527e-06,
+      "loss": 0.0523,
+      "step": 581
+    },
+    {
+      "epoch": 2.346774193548387,
+      "grad_norm": 1.3401274681091309,
+      "learning_rate": 4.35483870967742e-06,
+      "loss": 0.0718,
+      "step": 582
+    },
+    {
+      "epoch": 2.350806451612903,
+      "grad_norm": 1.0903416872024536,
+      "learning_rate": 4.327956989247312e-06,
+      "loss": 0.1176,
+      "step": 583
+    },
+    {
+      "epoch": 2.3548387096774195,
+      "grad_norm": 2.7714807987213135,
+      "learning_rate": 4.3010752688172045e-06,
+      "loss": 0.1075,
+      "step": 584
+    },
+    {
+      "epoch": 2.3588709677419355,
+      "grad_norm": 0.9703463315963745,
+      "learning_rate": 4.274193548387097e-06,
+      "loss": 0.0649,
+      "step": 585
+    },
+    {
+      "epoch": 2.3629032258064515,
+      "grad_norm": 5.25622034072876,
+      "learning_rate": 4.2473118279569895e-06,
+      "loss": 0.1146,
+      "step": 586
+    },
+    {
+      "epoch": 2.366935483870968,
+      "grad_norm": 1.6004530191421509,
+      "learning_rate": 4.220430107526882e-06,
+      "loss": 0.1173,
+      "step": 587
+    },
+    {
+      "epoch": 2.370967741935484,
+      "grad_norm": 1.7906506061553955,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.1305,
+      "step": 588
+    },
+    {
+      "epoch": 2.375,
+      "grad_norm": 5.008229732513428,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 0.1907,
+      "step": 589
+    },
+    {
+      "epoch": 2.379032258064516,
+      "grad_norm": 2.5902585983276367,
+      "learning_rate": 4.139784946236559e-06,
+      "loss": 0.1836,
+      "step": 590
+    },
+    {
+      "epoch": 2.383064516129032,
+      "grad_norm": 2.13139009475708,
+      "learning_rate": 4.112903225806452e-06,
+      "loss": 0.1012,
+      "step": 591
+    },
+    {
+      "epoch": 2.3870967741935485,
+      "grad_norm": 1.3434728384017944,
+      "learning_rate": 4.086021505376344e-06,
+      "loss": 0.1011,
+      "step": 592
+    },
+    {
+      "epoch": 2.3911290322580645,
+      "grad_norm": 1.74350106716156,
+      "learning_rate": 4.059139784946237e-06,
+      "loss": 0.1008,
+      "step": 593
+    },
+    {
+      "epoch": 2.3951612903225805,
+      "grad_norm": 1.2974146604537964,
+      "learning_rate": 4.032258064516129e-06,
+      "loss": 0.0692,
+      "step": 594
+    },
+    {
+      "epoch": 2.399193548387097,
+      "grad_norm": 2.330141544342041,
+      "learning_rate": 4.005376344086022e-06,
+      "loss": 0.1496,
+      "step": 595
+    },
+    {
+      "epoch": 2.403225806451613,
+      "grad_norm": 4.154720306396484,
+      "learning_rate": 3.978494623655914e-06,
+      "loss": 0.1431,
+      "step": 596
+    },
+    {
+      "epoch": 2.407258064516129,
+      "grad_norm": 1.0211693048477173,
+      "learning_rate": 3.951612903225807e-06,
+      "loss": 0.1015,
+      "step": 597
+    },
+    {
+      "epoch": 2.411290322580645,
+      "grad_norm": 4.42000675201416,
+      "learning_rate": 3.924731182795699e-06,
+      "loss": 0.1123,
+      "step": 598
+    },
+    {
+      "epoch": 2.4153225806451615,
+      "grad_norm": 4.67465353012085,
+      "learning_rate": 3.8978494623655915e-06,
+      "loss": 0.2117,
+      "step": 599
+    },
+    {
+      "epoch": 2.4193548387096775,
+      "grad_norm": 2.3731729984283447,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.1847,
+      "step": 600
+    },
+    {
+      "epoch": 2.4233870967741935,
+      "grad_norm": 2.7813737392425537,
+      "learning_rate": 3.8440860215053765e-06,
+      "loss": 0.1696,
+      "step": 601
+    },
+    {
+      "epoch": 2.4274193548387095,
+      "grad_norm": 1.7436891794204712,
+      "learning_rate": 3.817204301075269e-06,
+      "loss": 0.1275,
+      "step": 602
+    },
+    {
+      "epoch": 2.431451612903226,
+      "grad_norm": 2.3506593704223633,
+      "learning_rate": 3.7903225806451614e-06,
+      "loss": 0.1749,
+      "step": 603
+    },
+    {
+      "epoch": 2.435483870967742,
+      "grad_norm": 3.1209230422973633,
+      "learning_rate": 3.763440860215054e-06,
+      "loss": 0.1307,
+      "step": 604
+    },
+    {
+      "epoch": 2.439516129032258,
+      "grad_norm": 2.593312978744507,
+      "learning_rate": 3.7365591397849468e-06,
+      "loss": 0.1168,
+      "step": 605
+    },
+    {
+      "epoch": 2.443548387096774,
+      "grad_norm": 2.2633056640625,
+      "learning_rate": 3.7096774193548392e-06,
+      "loss": 0.1355,
+      "step": 606
+    },
+    {
+      "epoch": 2.4475806451612905,
+      "grad_norm": 2.898655891418457,
+      "learning_rate": 3.6827956989247317e-06,
+      "loss": 0.1298,
+      "step": 607
+    },
+    {
+      "epoch": 2.4516129032258065,
+      "grad_norm": 1.7566426992416382,
+      "learning_rate": 3.655913978494624e-06,
+      "loss": 0.1159,
+      "step": 608
+    },
+    {
+      "epoch": 2.4556451612903225,
+      "grad_norm": 6.452301025390625,
+      "learning_rate": 3.6290322580645166e-06,
+      "loss": 0.1349,
+      "step": 609
+    },
+    {
+      "epoch": 2.4596774193548385,
+      "grad_norm": 2.342876434326172,
+      "learning_rate": 3.602150537634409e-06,
+      "loss": 0.0856,
+      "step": 610
+    },
+    {
+      "epoch": 2.463709677419355,
+      "grad_norm": 2.236654281616211,
+      "learning_rate": 3.5752688172043015e-06,
+      "loss": 0.1418,
+      "step": 611
+    },
+    {
+      "epoch": 2.467741935483871,
+      "grad_norm": 3.464710235595703,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.097,
+      "step": 612
+    },
+    {
+      "epoch": 2.471774193548387,
+      "grad_norm": 1.936618447303772,
+      "learning_rate": 3.5215053763440865e-06,
+      "loss": 0.0876,
+      "step": 613
+    },
+    {
+      "epoch": 2.475806451612903,
+      "grad_norm": 3.4211089611053467,
+      "learning_rate": 3.494623655913979e-06,
+      "loss": 0.1502,
+      "step": 614
+    },
+    {
+      "epoch": 2.4798387096774195,
+      "grad_norm": 2.9869165420532227,
+      "learning_rate": 3.4677419354838714e-06,
+      "loss": 0.2398,
+      "step": 615
+    },
+    {
+      "epoch": 2.4838709677419355,
+      "grad_norm": 3.0065438747406006,
+      "learning_rate": 3.440860215053764e-06,
+      "loss": 0.1115,
+      "step": 616
+    },
+    {
+      "epoch": 2.4879032258064515,
+      "grad_norm": 1.830977439880371,
+      "learning_rate": 3.413978494623656e-06,
+      "loss": 0.1161,
+      "step": 617
+    },
+    {
+      "epoch": 2.491935483870968,
+      "grad_norm": 1.9489115476608276,
+      "learning_rate": 3.3870967741935484e-06,
+      "loss": 0.1224,
+      "step": 618
+    },
+    {
+      "epoch": 2.495967741935484,
+      "grad_norm": 2.2884771823883057,
+      "learning_rate": 3.360215053763441e-06,
+      "loss": 0.1092,
+      "step": 619
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 2.671285629272461,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.1711,
+      "step": 620
+    },
+    {
+      "epoch": 2.504032258064516,
+      "grad_norm": 1.9497143030166626,
+      "learning_rate": 3.306451612903226e-06,
+      "loss": 0.1145,
+      "step": 621
+    },
+    {
+      "epoch": 2.508064516129032,
+      "grad_norm": 1.3452483415603638,
+      "learning_rate": 3.2795698924731183e-06,
+      "loss": 0.1217,
+      "step": 622
+    },
+    {
+      "epoch": 2.5120967741935485,
+      "grad_norm": 1.2507896423339844,
+      "learning_rate": 3.2526881720430107e-06,
+      "loss": 0.0931,
+      "step": 623
+    },
+    {
+      "epoch": 2.5161290322580645,
+      "grad_norm": 1.3425334692001343,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.1145,
+      "step": 624
+    },
+    {
+      "epoch": 2.5201612903225805,
+      "grad_norm": 3.7666163444519043,
+      "learning_rate": 3.198924731182796e-06,
+      "loss": 0.1044,
+      "step": 625
+    },
+    {
+      "epoch": 2.524193548387097,
+      "grad_norm": 2.4426252841949463,
+      "learning_rate": 3.1720430107526885e-06,
+      "loss": 0.0856,
+      "step": 626
+    },
+    {
+      "epoch": 2.528225806451613,
+      "grad_norm": 3.1804256439208984,
+      "learning_rate": 3.145161290322581e-06,
+      "loss": 0.0867,
+      "step": 627
+    },
+    {
+      "epoch": 2.532258064516129,
+      "grad_norm": 2.037343740463257,
+      "learning_rate": 3.1182795698924735e-06,
+      "loss": 0.1085,
+      "step": 628
+    },
+    {
+      "epoch": 2.536290322580645,
+      "grad_norm": 4.272365093231201,
+      "learning_rate": 3.091397849462366e-06,
+      "loss": 0.174,
+      "step": 629
+    },
+    {
+      "epoch": 2.540322580645161,
+      "grad_norm": 2.1623380184173584,
+      "learning_rate": 3.0645161290322584e-06,
+      "loss": 0.0866,
+      "step": 630
+    },
+    {
+      "epoch": 2.5443548387096775,
+      "grad_norm": 3.3426926136016846,
+      "learning_rate": 3.037634408602151e-06,
+      "loss": 0.1142,
+      "step": 631
+    },
+    {
+      "epoch": 2.5483870967741935,
+      "grad_norm": 2.2241578102111816,
+      "learning_rate": 3.0107526881720433e-06,
+      "loss": 0.0995,
+      "step": 632
+    },
+    {
+      "epoch": 2.5524193548387095,
+      "grad_norm": 3.2953665256500244,
+      "learning_rate": 2.983870967741936e-06,
+      "loss": 0.1487,
+      "step": 633
+    },
+    {
+      "epoch": 2.556451612903226,
+      "grad_norm": 1.5125126838684082,
+      "learning_rate": 2.9569892473118283e-06,
+      "loss": 0.0567,
+      "step": 634
+    },
+    {
+      "epoch": 2.560483870967742,
+      "grad_norm": 2.612718343734741,
+      "learning_rate": 2.9301075268817207e-06,
+      "loss": 0.1033,
+      "step": 635
+    },
+    {
+      "epoch": 2.564516129032258,
+      "grad_norm": 1.220692753791809,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.0999,
+      "step": 636
+    },
+    {
+      "epoch": 2.568548387096774,
+      "grad_norm": 1.8477866649627686,
+      "learning_rate": 2.8763440860215057e-06,
+      "loss": 0.2083,
+      "step": 637
+    },
+    {
+      "epoch": 2.5725806451612905,
+      "grad_norm": 4.778720855712891,
+      "learning_rate": 2.849462365591398e-06,
+      "loss": 0.1335,
+      "step": 638
+    },
+    {
+      "epoch": 2.5766129032258065,
+      "grad_norm": 1.4977765083312988,
+      "learning_rate": 2.822580645161291e-06,
+      "loss": 0.147,
+      "step": 639
+    },
+    {
+      "epoch": 2.5806451612903225,
+      "grad_norm": 1.5701426267623901,
+      "learning_rate": 2.7956989247311827e-06,
+      "loss": 0.1333,
+      "step": 640
+    },
+    {
+      "epoch": 2.584677419354839,
+      "grad_norm": 1.051927924156189,
+      "learning_rate": 2.768817204301075e-06,
+      "loss": 0.0888,
+      "step": 641
+    },
+    {
+      "epoch": 2.588709677419355,
+      "grad_norm": 4.721953868865967,
+      "learning_rate": 2.7419354838709676e-06,
+      "loss": 0.131,
+      "step": 642
+    },
+    {
+      "epoch": 2.592741935483871,
+      "grad_norm": 2.7012953758239746,
+      "learning_rate": 2.71505376344086e-06,
+      "loss": 0.0714,
+      "step": 643
+    },
+    {
+      "epoch": 2.596774193548387,
+      "grad_norm": 5.694790363311768,
+      "learning_rate": 2.688172043010753e-06,
+      "loss": 0.1403,
+      "step": 644
+    },
+    {
+      "epoch": 2.600806451612903,
+      "grad_norm": 4.28272008895874,
+      "learning_rate": 2.6612903225806454e-06,
+      "loss": 0.1169,
+      "step": 645
+    },
+    {
+      "epoch": 2.6048387096774195,
+      "grad_norm": 1.0393054485321045,
+      "learning_rate": 2.634408602150538e-06,
+      "loss": 0.0731,
+      "step": 646
+    },
+    {
+      "epoch": 2.6088709677419355,
+      "grad_norm": 0.9973479509353638,
+      "learning_rate": 2.6075268817204303e-06,
+      "loss": 0.1072,
+      "step": 647
+    },
+    {
+      "epoch": 2.6129032258064515,
+      "grad_norm": 1.4460006952285767,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.1372,
+      "step": 648
+    },
+    {
+      "epoch": 2.616935483870968,
+      "grad_norm": 2.2750110626220703,
+      "learning_rate": 2.5537634408602153e-06,
+      "loss": 0.1131,
+      "step": 649
+    },
+    {
+      "epoch": 2.620967741935484,
+      "grad_norm": 2.0815539360046387,
+      "learning_rate": 2.5268817204301077e-06,
+      "loss": 0.1101,
+      "step": 650
+    },
+    {
+      "epoch": 2.625,
+      "grad_norm": 1.0647339820861816,
+      "learning_rate": 2.5e-06,
+      "loss": 0.1106,
+      "step": 651
+    },
+    {
+      "epoch": 2.629032258064516,
+      "grad_norm": 1.694512963294983,
+      "learning_rate": 2.4731182795698927e-06,
+      "loss": 0.1207,
+      "step": 652
+    },
+    {
+      "epoch": 2.633064516129032,
+      "grad_norm": 1.1872451305389404,
+      "learning_rate": 2.446236559139785e-06,
+      "loss": 0.0823,
+      "step": 653
+    },
+    {
+      "epoch": 2.6370967741935485,
+      "grad_norm": 1.3260234594345093,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.1082,
+      "step": 654
+    },
+    {
+      "epoch": 2.6411290322580645,
+      "grad_norm": 2.025832414627075,
+      "learning_rate": 2.39247311827957e-06,
+      "loss": 0.1217,
+      "step": 655
+    },
+    {
+      "epoch": 2.6451612903225805,
+      "grad_norm": 2.340317487716675,
+      "learning_rate": 2.3655913978494625e-06,
+      "loss": 0.1507,
+      "step": 656
+    },
+    {
+      "epoch": 2.649193548387097,
+      "grad_norm": 2.595576524734497,
+      "learning_rate": 2.338709677419355e-06,
+      "loss": 0.1588,
+      "step": 657
+    },
+    {
+      "epoch": 2.653225806451613,
+      "grad_norm": 2.0549724102020264,
+      "learning_rate": 2.3118279569892475e-06,
+      "loss": 0.1111,
+      "step": 658
+    },
+    {
+      "epoch": 2.657258064516129,
+      "grad_norm": 0.7785517573356628,
+      "learning_rate": 2.28494623655914e-06,
+      "loss": 0.0675,
+      "step": 659
+    },
+    {
+      "epoch": 2.661290322580645,
+      "grad_norm": 1.5122755765914917,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.0898,
+      "step": 660
+    },
+    {
+      "epoch": 2.665322580645161,
+      "grad_norm": 1.361331582069397,
+      "learning_rate": 2.231182795698925e-06,
+      "loss": 0.12,
+      "step": 661
+    },
+    {
+      "epoch": 2.6693548387096775,
+      "grad_norm": 2.3188157081604004,
+      "learning_rate": 2.2043010752688173e-06,
+      "loss": 0.1406,
+      "step": 662
+    },
+    {
+      "epoch": 2.6733870967741935,
+      "grad_norm": 1.4102435111999512,
+      "learning_rate": 2.17741935483871e-06,
+      "loss": 0.0961,
+      "step": 663
+    },
+    {
+      "epoch": 2.6774193548387095,
+      "grad_norm": 2.144468069076538,
+      "learning_rate": 2.1505376344086023e-06,
+      "loss": 0.0633,
+      "step": 664
+    },
+    {
+      "epoch": 2.681451612903226,
+      "grad_norm": 1.7837616205215454,
+      "learning_rate": 2.1236559139784947e-06,
+      "loss": 0.1136,
+      "step": 665
+    },
+    {
+      "epoch": 2.685483870967742,
+      "grad_norm": 2.097506284713745,
+      "learning_rate": 2.096774193548387e-06,
+      "loss": 0.1284,
+      "step": 666
+    },
+    {
+      "epoch": 2.689516129032258,
+      "grad_norm": 1.8578377962112427,
+      "learning_rate": 2.0698924731182797e-06,
+      "loss": 0.0985,
+      "step": 667
+    },
+    {
+      "epoch": 2.693548387096774,
+      "grad_norm": 1.454514741897583,
+      "learning_rate": 2.043010752688172e-06,
+      "loss": 0.1499,
+      "step": 668
+    },
+    {
+      "epoch": 2.6975806451612905,
+      "grad_norm": 6.206280708312988,
+      "learning_rate": 2.0161290322580646e-06,
+      "loss": 0.1131,
+      "step": 669
+    },
+    {
+      "epoch": 2.7016129032258065,
+      "grad_norm": 1.9479646682739258,
+      "learning_rate": 1.989247311827957e-06,
+      "loss": 0.0599,
+      "step": 670
+    },
+    {
+      "epoch": 2.7056451612903225,
+      "grad_norm": 1.7119982242584229,
+      "learning_rate": 1.9623655913978495e-06,
+      "loss": 0.1553,
+      "step": 671
+    },
+    {
+      "epoch": 2.709677419354839,
+      "grad_norm": 2.411572217941284,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 0.1384,
+      "step": 672
+    },
+    {
+      "epoch": 2.713709677419355,
+      "grad_norm": 2.9151971340179443,
+      "learning_rate": 1.9086021505376345e-06,
+      "loss": 0.1513,
+      "step": 673
+    },
+    {
+      "epoch": 2.717741935483871,
+      "grad_norm": 1.6464580297470093,
+      "learning_rate": 1.881720430107527e-06,
+      "loss": 0.0747,
+      "step": 674
+    },
+    {
+      "epoch": 2.721774193548387,
+      "grad_norm": 2.2614505290985107,
+      "learning_rate": 1.8548387096774196e-06,
+      "loss": 0.1306,
+      "step": 675
+    },
+    {
+      "epoch": 2.725806451612903,
+      "grad_norm": 3.9104158878326416,
+      "learning_rate": 1.827956989247312e-06,
+      "loss": 0.1036,
+      "step": 676
+    },
+    {
+      "epoch": 2.7298387096774195,
+      "grad_norm": 2.6638762950897217,
+      "learning_rate": 1.8010752688172045e-06,
+      "loss": 0.1882,
+      "step": 677
+    },
+    {
+      "epoch": 2.7338709677419355,
+      "grad_norm": 2.4250168800354004,
+      "learning_rate": 1.774193548387097e-06,
+      "loss": 0.1356,
+      "step": 678
+    },
+    {
+      "epoch": 2.7379032258064515,
+      "grad_norm": 1.579751968383789,
+      "learning_rate": 1.7473118279569895e-06,
+      "loss": 0.0973,
+      "step": 679
+    },
+    {
+      "epoch": 2.741935483870968,
+      "grad_norm": 1.9126899242401123,
+      "learning_rate": 1.720430107526882e-06,
+      "loss": 0.1051,
+      "step": 680
+    },
+    {
+      "epoch": 2.745967741935484,
+      "grad_norm": 2.4986772537231445,
+      "learning_rate": 1.6935483870967742e-06,
+      "loss": 0.097,
+      "step": 681
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.9752743244171143,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 0.1827,
+      "step": 682
+    },
+    {
+      "epoch": 2.754032258064516,
+      "grad_norm": 5.470479488372803,
+      "learning_rate": 1.6397849462365591e-06,
+      "loss": 0.156,
+      "step": 683
+    },
+    {
+      "epoch": 2.758064516129032,
+      "grad_norm": 0.6137222647666931,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 0.0982,
+      "step": 684
+    },
+    {
+      "epoch": 2.7620967741935485,
+      "grad_norm": 1.8977144956588745,
+      "learning_rate": 1.5860215053763443e-06,
+      "loss": 0.1484,
+      "step": 685
+    },
+    {
+      "epoch": 2.7661290322580645,
+      "grad_norm": 1.982347846031189,
+      "learning_rate": 1.5591397849462367e-06,
+      "loss": 0.0805,
+      "step": 686
+    },
+    {
+      "epoch": 2.7701612903225805,
+      "grad_norm": 1.7538466453552246,
+      "learning_rate": 1.5322580645161292e-06,
+      "loss": 0.0831,
+      "step": 687
+    },
+    {
+      "epoch": 2.774193548387097,
+      "grad_norm": 1.4044026136398315,
+      "learning_rate": 1.5053763440860217e-06,
+      "loss": 0.0967,
+      "step": 688
+    },
+    {
+      "epoch": 2.778225806451613,
+      "grad_norm": 3.4833972454071045,
+      "learning_rate": 1.4784946236559141e-06,
+      "loss": 0.1445,
+      "step": 689
+    },
+    {
+      "epoch": 2.782258064516129,
+      "grad_norm": 3.1198818683624268,
+      "learning_rate": 1.4516129032258066e-06,
+      "loss": 0.1556,
+      "step": 690
+    },
+    {
+      "epoch": 2.786290322580645,
+      "grad_norm": 1.4616000652313232,
+      "learning_rate": 1.424731182795699e-06,
+      "loss": 0.1769,
+      "step": 691
+    },
+    {
+      "epoch": 2.790322580645161,
+      "grad_norm": 0.99885493516922,
+      "learning_rate": 1.3978494623655913e-06,
+      "loss": 0.0787,
+      "step": 692
+    },
+    {
+      "epoch": 2.7943548387096775,
+      "grad_norm": 1.395920991897583,
+      "learning_rate": 1.3709677419354838e-06,
+      "loss": 0.0763,
+      "step": 693
+    },
+    {
+      "epoch": 2.7983870967741935,
+      "grad_norm": 2.1815457344055176,
+      "learning_rate": 1.3440860215053765e-06,
+      "loss": 0.1605,
+      "step": 694
+    },
+    {
+      "epoch": 2.8024193548387095,
+      "grad_norm": 1.7244387865066528,
+      "learning_rate": 1.317204301075269e-06,
+      "loss": 0.0735,
+      "step": 695
+    },
+    {
+      "epoch": 2.806451612903226,
+      "grad_norm": 2.2874746322631836,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 0.1302,
+      "step": 696
+    },
+    {
+      "epoch": 2.810483870967742,
+      "grad_norm": 0.7056113481521606,
+      "learning_rate": 1.2634408602150539e-06,
+      "loss": 0.097,
+      "step": 697
+    },
+    {
+      "epoch": 2.814516129032258,
+      "grad_norm": 2.458315134048462,
+      "learning_rate": 1.2365591397849463e-06,
+      "loss": 0.1269,
+      "step": 698
+    },
+    {
+      "epoch": 2.818548387096774,
+      "grad_norm": 1.1753970384597778,
+      "learning_rate": 1.2096774193548388e-06,
+      "loss": 0.0745,
+      "step": 699
+    },
+    {
+      "epoch": 2.8225806451612905,
+      "grad_norm": 1.3138192892074585,
+      "learning_rate": 1.1827956989247313e-06,
+      "loss": 0.1289,
+      "step": 700
+    },
+    {
+      "epoch": 2.8266129032258065,
+      "grad_norm": 1.3947908878326416,
+      "learning_rate": 1.1559139784946237e-06,
+      "loss": 0.1225,
+      "step": 701
+    },
+    {
+      "epoch": 2.8306451612903225,
+      "grad_norm": 1.4919567108154297,
+      "learning_rate": 1.1290322580645162e-06,
+      "loss": 0.119,
+      "step": 702
+    },
+    {
+      "epoch": 2.834677419354839,
+      "grad_norm": 1.0990135669708252,
+      "learning_rate": 1.1021505376344087e-06,
+      "loss": 0.0763,
+      "step": 703
+    },
+    {
+      "epoch": 2.838709677419355,
+      "grad_norm": 5.159749507904053,
+      "learning_rate": 1.0752688172043011e-06,
+      "loss": 0.1223,
+      "step": 704
+    },
+    {
+      "epoch": 2.842741935483871,
+      "grad_norm": 1.0806690454483032,
+      "learning_rate": 1.0483870967741936e-06,
+      "loss": 0.1383,
+      "step": 705
+    },
+    {
+      "epoch": 2.846774193548387,
+      "grad_norm": 3.3874659538269043,
+      "learning_rate": 1.021505376344086e-06,
+      "loss": 0.1,
+      "step": 706
+    },
+    {
+      "epoch": 2.850806451612903,
+      "grad_norm": 2.235921859741211,
+      "learning_rate": 9.946236559139785e-07,
+      "loss": 0.2047,
+      "step": 707
+    },
+    {
+      "epoch": 2.8548387096774195,
+      "grad_norm": 1.7132419347763062,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 0.1235,
+      "step": 708
+    },
+    {
+      "epoch": 2.8588709677419355,
+      "grad_norm": 5.385939598083496,
+      "learning_rate": 9.408602150537635e-07,
+      "loss": 0.0994,
+      "step": 709
+    },
+    {
+      "epoch": 2.8629032258064515,
+      "grad_norm": 1.3089847564697266,
+      "learning_rate": 9.13978494623656e-07,
+      "loss": 0.0878,
+      "step": 710
+    },
+    {
+      "epoch": 2.866935483870968,
+      "grad_norm": 1.2990398406982422,
+      "learning_rate": 8.870967741935485e-07,
+      "loss": 0.1384,
+      "step": 711
+    },
+    {
+      "epoch": 2.870967741935484,
+      "grad_norm": 1.1270217895507812,
+      "learning_rate": 8.60215053763441e-07,
+      "loss": 0.0901,
+      "step": 712
+    },
+    {
+      "epoch": 2.875,
+      "grad_norm": 1.7017827033996582,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 0.1437,
+      "step": 713
+    },
+    {
+      "epoch": 2.879032258064516,
+      "grad_norm": 0.8682994246482849,
+      "learning_rate": 8.064516129032258e-07,
+      "loss": 0.0515,
+      "step": 714
+    },
+    {
+      "epoch": 2.883064516129032,
+      "grad_norm": 3.2466022968292236,
+      "learning_rate": 7.795698924731184e-07,
+      "loss": 0.1803,
+      "step": 715
+    },
+    {
+      "epoch": 2.8870967741935485,
+      "grad_norm": 2.0208983421325684,
+      "learning_rate": 7.526881720430108e-07,
+      "loss": 0.1035,
+      "step": 716
+    },
+    {
+      "epoch": 2.8911290322580645,
+      "grad_norm": 1.170736312866211,
+      "learning_rate": 7.258064516129033e-07,
+      "loss": 0.13,
+      "step": 717
+    },
+    {
+      "epoch": 2.8951612903225805,
+      "grad_norm": 1.7991589307785034,
+      "learning_rate": 6.989247311827957e-07,
+      "loss": 0.0715,
+      "step": 718
+    },
+    {
+      "epoch": 2.899193548387097,
+      "grad_norm": 1.9782204627990723,
+      "learning_rate": 6.720430107526882e-07,
+      "loss": 0.0969,
+      "step": 719
+    },
+    {
+      "epoch": 2.903225806451613,
+      "grad_norm": 2.702655792236328,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 0.0927,
+      "step": 720
+    },
+    {
+      "epoch": 2.907258064516129,
+      "grad_norm": 2.2504546642303467,
+      "learning_rate": 6.182795698924732e-07,
+      "loss": 0.146,
+      "step": 721
+    },
+    {
+      "epoch": 2.911290322580645,
+      "grad_norm": 2.546994209289551,
+      "learning_rate": 5.913978494623656e-07,
+      "loss": 0.0981,
+      "step": 722
+    },
+    {
+      "epoch": 2.915322580645161,
+      "grad_norm": 3.177995443344116,
+      "learning_rate": 5.645161290322581e-07,
+      "loss": 0.1493,
+      "step": 723
+    },
+    {
+      "epoch": 2.9193548387096775,
+      "grad_norm": 1.8446640968322754,
+      "learning_rate": 5.376344086021506e-07,
+      "loss": 0.1168,
+      "step": 724
+    },
+    {
+      "epoch": 2.9233870967741935,
+      "grad_norm": 1.462225079536438,
+      "learning_rate": 5.10752688172043e-07,
+      "loss": 0.0658,
+      "step": 725
+    },
+    {
+      "epoch": 2.9274193548387095,
+      "grad_norm": 2.062228202819824,
+      "learning_rate": 4.838709677419355e-07,
+      "loss": 0.0861,
+      "step": 726
+    },
+    {
+      "epoch": 2.931451612903226,
+      "grad_norm": 1.066767930984497,
+      "learning_rate": 4.56989247311828e-07,
+      "loss": 0.1016,
+      "step": 727
+    },
+    {
+      "epoch": 2.935483870967742,
+      "grad_norm": 1.9916908740997314,
+      "learning_rate": 4.301075268817205e-07,
+      "loss": 0.0999,
+      "step": 728
+    },
+    {
+      "epoch": 2.939516129032258,
+      "grad_norm": 2.194812536239624,
+      "learning_rate": 4.032258064516129e-07,
+      "loss": 0.1289,
+      "step": 729
+    },
+    {
+      "epoch": 2.943548387096774,
+      "grad_norm": 1.067501425743103,
+      "learning_rate": 3.763440860215054e-07,
+      "loss": 0.0866,
+      "step": 730
+    },
+    {
+      "epoch": 2.9475806451612905,
+      "grad_norm": 1.4836621284484863,
+      "learning_rate": 3.4946236559139783e-07,
+      "loss": 0.0954,
+      "step": 731
+    },
+    {
+      "epoch": 2.9516129032258065,
+      "grad_norm": 2.3710856437683105,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 0.1969,
+      "step": 732
+    },
+    {
+      "epoch": 2.9556451612903225,
+      "grad_norm": 2.83674693107605,
+      "learning_rate": 2.956989247311828e-07,
+      "loss": 0.1929,
+      "step": 733
+    },
+    {
+      "epoch": 2.959677419354839,
+      "grad_norm": 2.1235055923461914,
+      "learning_rate": 2.688172043010753e-07,
+      "loss": 0.1497,
+      "step": 734
+    },
+    {
+      "epoch": 2.963709677419355,
+      "grad_norm": 1.6132428646087646,
+      "learning_rate": 2.4193548387096775e-07,
+      "loss": 0.1602,
+      "step": 735
+    },
+    {
+      "epoch": 2.967741935483871,
+      "grad_norm": 2.551713228225708,
+      "learning_rate": 2.1505376344086024e-07,
+      "loss": 0.0789,
+      "step": 736
+    },
+    {
+      "epoch": 2.971774193548387,
+      "grad_norm": 1.6804261207580566,
+      "learning_rate": 1.881720430107527e-07,
+      "loss": 0.1895,
+      "step": 737
+    },
+    {
+      "epoch": 2.975806451612903,
+      "grad_norm": 1.8240628242492676,
+      "learning_rate": 1.6129032258064518e-07,
+      "loss": 0.1558,
+      "step": 738
+    },
+    {
+      "epoch": 2.9798387096774195,
+      "grad_norm": 1.9709798097610474,
+      "learning_rate": 1.3440860215053764e-07,
+      "loss": 0.1245,
+      "step": 739
+    },
+    {
+      "epoch": 2.9838709677419355,
+      "grad_norm": 1.7331011295318604,
+      "learning_rate": 1.0752688172043012e-07,
+      "loss": 0.0999,
+      "step": 740
+    },
+    {
+      "epoch": 2.9879032258064515,
+      "grad_norm": 1.3876549005508423,
+      "learning_rate": 8.064516129032259e-08,
+      "loss": 0.1744,
+      "step": 741
+    },
+    {
+      "epoch": 2.991935483870968,
+      "grad_norm": 1.998831033706665,
+      "learning_rate": 5.376344086021506e-08,
+      "loss": 0.0876,
+      "step": 742
+    },
+    {
+      "epoch": 2.995967741935484,
+      "grad_norm": 1.9842371940612793,
+      "learning_rate": 2.688172043010753e-08,
+      "loss": 0.1036,
+      "step": 743
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 2.3604133129119873,
+      "learning_rate": 0.0,
+      "loss": 0.0957,
+      "step": 744
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.7625,
+      "eval_f1": 0.9021978021978022,
+      "eval_loss": 0.14369329810142517,
+      "eval_runtime": 2.3693,
+      "eval_samples_per_second": 371.416,
+      "eval_steps_per_second": 11.818,
+      "step": 744
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 744,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 142052951715840.0,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}