diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8330 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 1184,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0033783783783783786,
+      "grad_norm": 134.4763946533203,
+      "learning_rate": 0.0,
+      "loss": 4.6817,
+      "step": 1
+    },
+    {
+      "epoch": 0.006756756756756757,
+      "grad_norm": 125.12445068359375,
+      "learning_rate": 5.017166594399687e-06,
+      "loss": 4.5642,
+      "step": 2
+    },
+    {
+      "epoch": 0.010135135135135136,
+      "grad_norm": 133.296142578125,
+      "learning_rate": 7.952020911994375e-06,
+      "loss": 4.787,
+      "step": 3
+    },
+    {
+      "epoch": 0.013513513513513514,
+      "grad_norm": 72.2229995727539,
+      "learning_rate": 1.0034333188799373e-05,
+      "loss": 3.1422,
+      "step": 4
+    },
+    {
+      "epoch": 0.016891891891891893,
+      "grad_norm": 83.57042694091797,
+      "learning_rate": 1.164950007226698e-05,
+      "loss": 1.9827,
+      "step": 5
+    },
+    {
+      "epoch": 0.02027027027027027,
+      "grad_norm": 35.78062057495117,
+      "learning_rate": 1.2969187506394062e-05,
+      "loss": 1.0212,
+      "step": 6
+    },
+    {
+      "epoch": 0.02364864864864865,
+      "grad_norm": 48.31964111328125,
+      "learning_rate": 1.4084967333570947e-05,
+      "loss": 1.1823,
+      "step": 7
+    },
+    {
+      "epoch": 0.02702702702702703,
+      "grad_norm": 10.594072341918945,
+      "learning_rate": 1.505149978319906e-05,
+      "loss": 0.7368,
+      "step": 8
+    },
+    {
+      "epoch": 0.030405405405405407,
+      "grad_norm": 7.897088050842285,
+      "learning_rate": 1.590404182398875e-05,
+      "loss": 0.6276,
+      "step": 9
+    },
+    {
+      "epoch": 0.033783783783783786,
+      "grad_norm": 3.8260443210601807,
+      "learning_rate": 1.666666666666667e-05,
+      "loss": 0.5934,
+      "step": 10
+    },
+    {
+      "epoch": 0.037162162162162164,
+      "grad_norm": 11.113093376159668,
+      "learning_rate": 1.7356544752637084e-05,
+      "loss": 0.6987,
+      "step": 11
+    },
+    {
+      "epoch": 0.04054054054054054,
+      "grad_norm": 3.2817893028259277,
+      "learning_rate": 1.7986354100793748e-05,
+      "loss": 0.5722,
+      "step": 12
+    },
+    {
+      "epoch": 0.04391891891891892,
+      "grad_norm": 1.4319814443588257,
+      "learning_rate": 1.8565722538447282e-05,
+      "loss": 0.5199,
+      "step": 13
+    },
+    {
+      "epoch": 0.0472972972972973,
+      "grad_norm": 1.1898471117019653,
+      "learning_rate": 1.9102133927970633e-05,
+      "loss": 0.5063,
+      "step": 14
+    },
+    {
+      "epoch": 0.05067567567567568,
+      "grad_norm": 2.1912143230438232,
+      "learning_rate": 1.9601520984261358e-05,
+      "loss": 0.5368,
+      "step": 15
+    },
+    {
+      "epoch": 0.05405405405405406,
+      "grad_norm": 1.0015403032302856,
+      "learning_rate": 2.0068666377598747e-05,
+      "loss": 0.4682,
+      "step": 16
+    },
+    {
+      "epoch": 0.057432432432432436,
+      "grad_norm": 1.114931583404541,
+      "learning_rate": 2.0507482022971233e-05,
+      "loss": 0.4997,
+      "step": 17
+    },
+    {
+      "epoch": 0.060810810810810814,
+      "grad_norm": 1.0931981801986694,
+      "learning_rate": 2.0921208418388435e-05,
+      "loss": 0.4659,
+      "step": 18
+    },
+    {
+      "epoch": 0.06418918918918919,
+      "grad_norm": 0.8113773465156555,
+      "learning_rate": 2.1312560015880482e-05,
+      "loss": 0.4697,
+      "step": 19
+    },
+    {
+      "epoch": 0.06756756756756757,
+      "grad_norm": 1.0267772674560547,
+      "learning_rate": 2.1683833261066357e-05,
+      "loss": 0.4851,
+      "step": 20
+    },
+    {
+      "epoch": 0.07094594594594594,
+      "grad_norm": 0.8238614201545715,
+      "learning_rate": 2.2036988245565324e-05,
+      "loss": 0.4517,
+      "step": 21
+    },
+    {
+      "epoch": 0.07432432432432433,
+      "grad_norm": 0.9366945624351501,
+      "learning_rate": 2.2373711347036773e-05,
+      "loss": 0.4982,
+      "step": 22
+    },
+    {
+      "epoch": 0.0777027027027027,
+      "grad_norm": 0.8443475365638733,
+      "learning_rate": 2.269546393362655e-05,
+      "loss": 0.4128,
+      "step": 23
+    },
+    {
+      "epoch": 0.08108108108108109,
+      "grad_norm": 0.8733354806900024,
+      "learning_rate": 2.3003520695193437e-05,
+      "loss": 0.4281,
+      "step": 24
+    },
+    {
+      "epoch": 0.08445945945945946,
+      "grad_norm": 0.7766995429992676,
+      "learning_rate": 2.329900014453396e-05,
+      "loss": 0.4407,
+      "step": 25
+    },
+    {
+      "epoch": 0.08783783783783784,
+      "grad_norm": 0.8224064111709595,
+      "learning_rate": 2.3582889132846968e-05,
+      "loss": 0.4155,
+      "step": 26
+    },
+    {
+      "epoch": 0.09121621621621621,
+      "grad_norm": 0.7931010127067566,
+      "learning_rate": 2.3856062735983123e-05,
+      "loss": 0.4191,
+      "step": 27
+    },
+    {
+      "epoch": 0.0945945945945946,
+      "grad_norm": 0.6803894639015198,
+      "learning_rate": 2.4119300522370322e-05,
+      "loss": 0.4055,
+      "step": 28
+    },
+    {
+      "epoch": 0.09797297297297297,
+      "grad_norm": 0.748294472694397,
+      "learning_rate": 2.4373299964982603e-05,
+      "loss": 0.4443,
+      "step": 29
+    },
+    {
+      "epoch": 0.10135135135135136,
+      "grad_norm": 0.706881582736969,
+      "learning_rate": 2.4618687578661044e-05,
+      "loss": 0.4093,
+      "step": 30
+    },
+    {
+      "epoch": 0.10472972972972973,
+      "grad_norm": 0.7517805695533752,
+      "learning_rate": 2.4856028230571212e-05,
+      "loss": 0.4288,
+      "step": 31
+    },
+    {
+      "epoch": 0.10810810810810811,
+      "grad_norm": 0.6549198031425476,
+      "learning_rate": 2.5085832971998436e-05,
+      "loss": 0.4222,
+      "step": 32
+    },
+    {
+      "epoch": 0.11148648648648649,
+      "grad_norm": 0.9364942908287048,
+      "learning_rate": 2.530856566463146e-05,
+      "loss": 0.416,
+      "step": 33
+    },
+    {
+      "epoch": 0.11486486486486487,
+      "grad_norm": 0.624920129776001,
+      "learning_rate": 2.552464861737092e-05,
+      "loss": 0.3735,
+      "step": 34
+    },
+    {
+      "epoch": 0.11824324324324324,
+      "grad_norm": 0.6083908081054688,
+      "learning_rate": 2.5734467405837933e-05,
+      "loss": 0.3315,
+      "step": 35
+    },
+    {
+      "epoch": 0.12162162162162163,
+      "grad_norm": 0.6106983423233032,
+      "learning_rate": 2.5938375012788124e-05,
+      "loss": 0.3652,
+      "step": 36
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 0.7572540044784546,
+      "learning_rate": 2.6136695401116585e-05,
+      "loss": 0.452,
+      "step": 37
+    },
+    {
+      "epoch": 0.12837837837837837,
+      "grad_norm": 0.6654285788536072,
+      "learning_rate": 2.6329726610280168e-05,
+      "loss": 0.3328,
+      "step": 38
+    },
+    {
+      "epoch": 0.13175675675675674,
+      "grad_norm": 0.676705002784729,
+      "learning_rate": 2.651774345044166e-05,
+      "loss": 0.3987,
+      "step": 39
+    },
+    {
+      "epoch": 0.13513513513513514,
+      "grad_norm": 0.6630686521530151,
+      "learning_rate": 2.6700999855466042e-05,
+      "loss": 0.3794,
+      "step": 40
+    },
+    {
+      "epoch": 0.13851351351351351,
+      "grad_norm": 0.8196331858634949,
+      "learning_rate": 2.687973094532893e-05,
+      "loss": 0.4218,
+      "step": 41
+    },
+    {
+      "epoch": 0.14189189189189189,
+      "grad_norm": 0.6006856560707092,
+      "learning_rate": 2.7054154839965013e-05,
+      "loss": 0.3788,
+      "step": 42
+    },
+    {
+      "epoch": 0.14527027027027026,
+      "grad_norm": 0.6180748343467712,
+      "learning_rate": 2.722447425965978e-05,
+      "loss": 0.3946,
+      "step": 43
+    },
+    {
+      "epoch": 0.14864864864864866,
+      "grad_norm": 0.7010207772254944,
+      "learning_rate": 2.739087794143646e-05,
+      "loss": 0.389,
+      "step": 44
+    },
+    {
+      "epoch": 0.15202702702702703,
+      "grad_norm": 0.681907594203949,
+      "learning_rate": 2.755354189625573e-05,
+      "loss": 0.399,
+      "step": 45
+    },
+    {
+      "epoch": 0.1554054054054054,
+      "grad_norm": 0.6506620645523071,
+      "learning_rate": 2.771263052802624e-05,
+      "loss": 0.3924,
+      "step": 46
+    },
+    {
+      "epoch": 0.15878378378378377,
+      "grad_norm": 0.6619815230369568,
+      "learning_rate": 2.7868297632261957e-05,
+      "loss": 0.4111,
+      "step": 47
+    },
+    {
+      "epoch": 0.16216216216216217,
+      "grad_norm": 0.6953612565994263,
+      "learning_rate": 2.8020687289593123e-05,
+      "loss": 0.3958,
+      "step": 48
+    },
+    {
+      "epoch": 0.16554054054054054,
+      "grad_norm": 0.761862576007843,
+      "learning_rate": 2.8169934667141895e-05,
+      "loss": 0.3844,
+      "step": 49
+    },
+    {
+      "epoch": 0.16891891891891891,
+      "grad_norm": 0.7193901538848877,
+      "learning_rate": 2.8316166738933646e-05,
+      "loss": 0.3847,
+      "step": 50
+    },
+    {
+      "epoch": 0.17229729729729729,
+      "grad_norm": 0.6792317032814026,
+      "learning_rate": 2.845950293496561e-05,
+      "loss": 0.4043,
+      "step": 51
+    },
+    {
+      "epoch": 0.17567567567567569,
+      "grad_norm": 0.6307753920555115,
+      "learning_rate": 2.8600055727246657e-05,
+      "loss": 0.3806,
+      "step": 52
+    },
+    {
+      "epoch": 0.17905405405405406,
+      "grad_norm": 0.7366315722465515,
+      "learning_rate": 2.8737931160013153e-05,
+      "loss": 0.416,
+      "step": 53
+    },
+    {
+      "epoch": 0.18243243243243243,
+      "grad_norm": 0.603863537311554,
+      "learning_rate": 2.8873229330382812e-05,
+      "loss": 0.3548,
+      "step": 54
+    },
+    {
+      "epoch": 0.1858108108108108,
+      "grad_norm": 0.7424588799476624,
+      "learning_rate": 2.9006044824904066e-05,
+      "loss": 0.3982,
+      "step": 55
+    },
+    {
+      "epoch": 0.1891891891891892,
+      "grad_norm": 0.5944585204124451,
+      "learning_rate": 2.913646711677001e-05,
+      "loss": 0.3536,
+      "step": 56
+    },
+    {
+      "epoch": 0.19256756756756757,
+      "grad_norm": 0.6465046405792236,
+      "learning_rate": 2.926458092787486e-05,
+      "loss": 0.3594,
+      "step": 57
+    },
+    {
+      "epoch": 0.19594594594594594,
+      "grad_norm": 0.7178723812103271,
+      "learning_rate": 2.939046655938229e-05,
+      "loss": 0.3725,
+      "step": 58
+    },
+    {
+      "epoch": 0.19932432432432431,
+      "grad_norm": 0.7240431308746338,
+      "learning_rate": 2.951420019403574e-05,
+      "loss": 0.3896,
+      "step": 59
+    },
+    {
+      "epoch": 0.20270270270270271,
+      "grad_norm": 0.7555009126663208,
+      "learning_rate": 2.963585417306073e-05,
+      "loss": 0.3633,
+      "step": 60
+    },
+    {
+      "epoch": 0.20608108108108109,
+      "grad_norm": 0.6279475092887878,
+      "learning_rate": 2.9755497250179453e-05,
+      "loss": 0.3447,
+      "step": 61
+    },
+    {
+      "epoch": 0.20945945945945946,
+      "grad_norm": 0.6943121552467346,
+      "learning_rate": 2.98731948249709e-05,
+      "loss": 0.3851,
+      "step": 62
+    },
+    {
+      "epoch": 0.21283783783783783,
+      "grad_norm": 0.6565203070640564,
+      "learning_rate": 2.9989009157559694e-05,
+      "loss": 0.3896,
+      "step": 63
+    },
+    {
+      "epoch": 0.21621621621621623,
+      "grad_norm": 4.912997722625732,
+      "learning_rate": 3.010299956639812e-05,
+      "loss": 0.4528,
+      "step": 64
+    },
+    {
+      "epoch": 0.2195945945945946,
+      "grad_norm": 0.7776851058006287,
+      "learning_rate": 3.021522261071426e-05,
+      "loss": 0.3151,
+      "step": 65
+    },
+    {
+      "epoch": 0.22297297297297297,
+      "grad_norm": 0.5963460206985474,
+      "learning_rate": 3.0325732259031143e-05,
+      "loss": 0.3501,
+      "step": 66
+    },
+    {
+      "epoch": 0.22635135135135134,
+      "grad_norm": 0.8160498142242432,
+      "learning_rate": 3.043458004501377e-05,
+      "loss": 0.3383,
+      "step": 67
+    },
+    {
+      "epoch": 0.22972972972972974,
+      "grad_norm": 0.6507856249809265,
+      "learning_rate": 3.054181521177061e-05,
+      "loss": 0.3559,
+      "step": 68
+    },
+    {
+      "epoch": 0.23310810810810811,
+      "grad_norm": 0.6282461881637573,
+      "learning_rate": 3.064748484562093e-05,
+      "loss": 0.3951,
+      "step": 69
+    },
+    {
+      "epoch": 0.23648648648648649,
+      "grad_norm": 0.6464638710021973,
+      "learning_rate": 3.0751634000237615e-05,
+      "loss": 0.3759,
+      "step": 70
+    },
+    {
+      "epoch": 0.23986486486486486,
+      "grad_norm": 0.621934175491333,
+      "learning_rate": 3.085430581198459e-05,
+      "loss": 0.409,
+      "step": 71
+    },
+    {
+      "epoch": 0.24324324324324326,
+      "grad_norm": 0.6526879668235779,
+      "learning_rate": 3.095554160718781e-05,
+      "loss": 0.3938,
+      "step": 72
+    },
+    {
+      "epoch": 0.24662162162162163,
+      "grad_norm": 0.6536738276481628,
+      "learning_rate": 3.10553810020076e-05,
+      "loss": 0.3657,
+      "step": 73
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6381665468215942,
+      "learning_rate": 3.115386199551628e-05,
+      "loss": 0.3743,
+      "step": 74
+    },
+    {
+      "epoch": 0.2533783783783784,
+      "grad_norm": 0.6818379759788513,
+      "learning_rate": 3.1251021056528336e-05,
+      "loss": 0.3188,
+      "step": 75
+    },
+    {
+      "epoch": 0.25675675675675674,
+      "grad_norm": 0.5829269886016846,
+      "learning_rate": 3.134689320467986e-05,
+      "loss": 0.3622,
+      "step": 76
+    },
+    {
+      "epoch": 0.26013513513513514,
+      "grad_norm": 0.6988232731819153,
+      "learning_rate": 3.144151208620804e-05,
+      "loss": 0.3768,
+      "step": 77
+    },
+    {
+      "epoch": 0.2635135135135135,
+      "grad_norm": 0.5981537699699402,
+      "learning_rate": 3.1534910044841344e-05,
+      "loss": 0.3655,
+      "step": 78
+    },
+    {
+      "epoch": 0.2668918918918919,
+      "grad_norm": 0.5821400284767151,
+      "learning_rate": 3.1627118188174024e-05,
+      "loss": 0.358,
+      "step": 79
+    },
+    {
+      "epoch": 0.2702702702702703,
+      "grad_norm": 0.620126485824585,
+      "learning_rate": 3.171816644986573e-05,
+      "loss": 0.3579,
+      "step": 80
+    },
+    {
+      "epoch": 0.27364864864864863,
+      "grad_norm": 0.6327139139175415,
+      "learning_rate": 3.18080836479775e-05,
+      "loss": 0.3916,
+      "step": 81
+    },
+    {
+      "epoch": 0.27702702702702703,
+      "grad_norm": 0.5900242328643799,
+      "learning_rate": 3.1896897539728616e-05,
+      "loss": 0.3768,
+      "step": 82
+    },
+    {
+      "epoch": 0.28040540540540543,
+      "grad_norm": 0.5920627117156982,
+      "learning_rate": 3.198463487293457e-05,
+      "loss": 0.3709,
+      "step": 83
+    },
+    {
+      "epoch": 0.28378378378378377,
+      "grad_norm": 0.5604241490364075,
+      "learning_rate": 3.207132143436469e-05,
+      "loss": 0.3871,
+      "step": 84
+    },
+    {
+      "epoch": 0.28716216216216217,
+      "grad_norm": 0.6093663573265076,
+      "learning_rate": 3.215698209523821e-05,
+      "loss": 0.3632,
+      "step": 85
+    },
+    {
+      "epoch": 0.2905405405405405,
+      "grad_norm": 0.568435549736023,
+      "learning_rate": 3.224164085405946e-05,
+      "loss": 0.3788,
+      "step": 86
+    },
+    {
+      "epoch": 0.2939189189189189,
+      "grad_norm": 0.5885617136955261,
+      "learning_rate": 3.232532087697698e-05,
+      "loss": 0.3472,
+      "step": 87
+    },
+    {
+      "epoch": 0.2972972972972973,
+      "grad_norm": 0.5474864840507507,
+      "learning_rate": 3.240804453583615e-05,
+      "loss": 0.3815,
+      "step": 88
+    },
+    {
+      "epoch": 0.30067567567567566,
+      "grad_norm": 0.6116411089897156,
+      "learning_rate": 3.248983344408188e-05,
+      "loss": 0.3374,
+      "step": 89
+    },
+    {
+      "epoch": 0.30405405405405406,
+      "grad_norm": 0.5523970127105713,
+      "learning_rate": 3.2570708490655414e-05,
+      "loss": 0.3503,
+      "step": 90
+    },
+    {
+      "epoch": 0.30743243243243246,
+      "grad_norm": 0.6035043597221375,
+      "learning_rate": 3.265068987201822e-05,
+      "loss": 0.3699,
+      "step": 91
+    },
+    {
+      "epoch": 0.3108108108108108,
+      "grad_norm": 0.4965035617351532,
+      "learning_rate": 3.2729797122425925e-05,
+      "loss": 0.3702,
+      "step": 92
+    },
+    {
+      "epoch": 0.3141891891891892,
+      "grad_norm": 0.5582573413848877,
+      "learning_rate": 3.280804914256559e-05,
+      "loss": 0.333,
+      "step": 93
+    },
+    {
+      "epoch": 0.31756756756756754,
+      "grad_norm": 0.5313113331794739,
+      "learning_rate": 3.288546422666164e-05,
+      "loss": 0.329,
+      "step": 94
+    },
+    {
+      "epoch": 0.32094594594594594,
+      "grad_norm": 0.6049978733062744,
+      "learning_rate": 3.2962060088147464e-05,
+      "loss": 0.3931,
+      "step": 95
+    },
+    {
+      "epoch": 0.32432432432432434,
+      "grad_norm": 0.5387487411499023,
+      "learning_rate": 3.3037853883992805e-05,
+      "loss": 0.3432,
+      "step": 96
+    },
+    {
+      "epoch": 0.3277027027027027,
+      "grad_norm": 0.588671088218689,
+      "learning_rate": 3.3112862237770756e-05,
+      "loss": 0.3508,
+      "step": 97
+    },
+    {
+      "epoch": 0.3310810810810811,
+      "grad_norm": 0.536973774433136,
+      "learning_rate": 3.3187101261541584e-05,
+      "loss": 0.372,
+      "step": 98
+    },
+    {
+      "epoch": 0.3344594594594595,
+      "grad_norm": 0.676276445388794,
+      "learning_rate": 3.326058657662584e-05,
+      "loss": 0.3832,
+      "step": 99
+    },
+    {
+      "epoch": 0.33783783783783783,
+      "grad_norm": 0.46296200156211853,
+      "learning_rate": 3.333333333333334e-05,
+      "loss": 0.3553,
+      "step": 100
+    },
+    {
+      "epoch": 0.34121621621621623,
+      "grad_norm": 0.674472451210022,
+      "learning_rate": 3.340535622971072e-05,
+      "loss": 0.3448,
+      "step": 101
+    },
+    {
+      "epoch": 0.34459459459459457,
+      "grad_norm": 0.5100315809249878,
+      "learning_rate": 3.3476669529365295e-05,
+      "loss": 0.3367,
+      "step": 102
+    },
+    {
+      "epoch": 0.34797297297297297,
+      "grad_norm": 0.5637315511703491,
+      "learning_rate": 3.3547287078419544e-05,
+      "loss": 0.3786,
+      "step": 103
+    },
+    {
+      "epoch": 0.35135135135135137,
+      "grad_norm": 0.5659502744674683,
+      "learning_rate": 3.361722232164634e-05,
+      "loss": 0.352,
+      "step": 104
+    },
+    {
+      "epoch": 0.3547297297297297,
+      "grad_norm": 0.5390239357948303,
+      "learning_rate": 3.3686488317832306e-05,
+      "loss": 0.3473,
+      "step": 105
+    },
+    {
+      "epoch": 0.3581081081081081,
+      "grad_norm": 0.5552096366882324,
+      "learning_rate": 3.375509775441284e-05,
+      "loss": 0.2945,
+      "step": 106
+    },
+    {
+      "epoch": 0.3614864864864865,
+      "grad_norm": 0.5016259551048279,
+      "learning_rate": 3.382306296142016e-05,
+      "loss": 0.313,
+      "step": 107
+    },
+    {
+      "epoch": 0.36486486486486486,
+      "grad_norm": 0.5755091905593872,
+      "learning_rate": 3.38903959247825e-05,
+      "loss": 0.3367,
+      "step": 108
+    },
+    {
+      "epoch": 0.36824324324324326,
+      "grad_norm": 0.5590441823005676,
+      "learning_rate": 3.395710829901039e-05,
+      "loss": 0.3639,
+      "step": 109
+    },
+    {
+      "epoch": 0.3716216216216216,
+      "grad_norm": 0.5077652335166931,
+      "learning_rate": 3.402321141930376e-05,
+      "loss": 0.3523,
+      "step": 110
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 0.5614023804664612,
+      "learning_rate": 3.4088716313110955e-05,
+      "loss": 0.3418,
+      "step": 111
+    },
+    {
+      "epoch": 0.3783783783783784,
+      "grad_norm": 0.5325702428817749,
+      "learning_rate": 3.415363371116969e-05,
+      "loss": 0.3208,
+      "step": 112
+    },
+    {
+      "epoch": 0.38175675675675674,
+      "grad_norm": 0.5361766219139099,
+      "learning_rate": 3.4217974058057e-05,
+      "loss": 0.3412,
+      "step": 113
+    },
+    {
+      "epoch": 0.38513513513513514,
+      "grad_norm": 0.5173611044883728,
+      "learning_rate": 3.428174752227455e-05,
+      "loss": 0.3615,
+      "step": 114
+    },
+    {
+      "epoch": 0.3885135135135135,
+      "grad_norm": 0.5852875709533691,
+      "learning_rate": 3.434496400589353e-05,
+      "loss": 0.3171,
+      "step": 115
+    },
+    {
+      "epoch": 0.3918918918918919,
+      "grad_norm": 0.48835834860801697,
+      "learning_rate": 3.440763315378198e-05,
+      "loss": 0.315,
+      "step": 116
+    },
+    {
+      "epoch": 0.3952702702702703,
+      "grad_norm": 0.5958307385444641,
+      "learning_rate": 3.446976436243603e-05,
+      "loss": 0.3117,
+      "step": 117
+    },
+    {
+      "epoch": 0.39864864864864863,
+      "grad_norm": 0.5645899772644043,
+      "learning_rate": 3.4531366788435425e-05,
+      "loss": 0.3192,
+      "step": 118
+    },
+    {
+      "epoch": 0.40202702702702703,
+      "grad_norm": 0.5764517784118652,
+      "learning_rate": 3.459244935654219e-05,
+      "loss": 0.3353,
+      "step": 119
+    },
+    {
+      "epoch": 0.40540540540540543,
+      "grad_norm": 0.5075375437736511,
+      "learning_rate": 3.465302076746041e-05,
+      "loss": 0.3572,
+      "step": 120
+    },
+    {
+      "epoch": 0.40878378378378377,
+      "grad_norm": 0.48014962673187256,
+      "learning_rate": 3.471308950527417e-05,
+      "loss": 0.3217,
+      "step": 121
+    },
+    {
+      "epoch": 0.41216216216216217,
+      "grad_norm": 0.533892035484314,
+      "learning_rate": 3.477266384457914e-05,
+      "loss": 0.3214,
+      "step": 122
+    },
+    {
+      "epoch": 0.4155405405405405,
+      "grad_norm": 0.4845181405544281,
+      "learning_rate": 3.48317518573233e-05,
+      "loss": 0.3332,
+      "step": 123
+    },
+    {
+      "epoch": 0.4189189189189189,
+      "grad_norm": 0.5133293271064758,
+      "learning_rate": 3.489036141937059e-05,
+      "loss": 0.354,
+      "step": 124
+    },
+    {
+      "epoch": 0.4222972972972973,
+      "grad_norm": 0.5352755188941956,
+      "learning_rate": 3.494850021680094e-05,
+      "loss": 0.353,
+      "step": 125
+    },
+    {
+      "epoch": 0.42567567567567566,
+      "grad_norm": 0.5208103060722351,
+      "learning_rate": 3.500617575195938e-05,
+      "loss": 0.3065,
+      "step": 126
+    },
+    {
+      "epoch": 0.42905405405405406,
+      "grad_norm": 0.5982815623283386,
+      "learning_rate": 3.5063395349265945e-05,
+      "loss": 0.3329,
+      "step": 127
+    },
+    {
+      "epoch": 0.43243243243243246,
+      "grad_norm": 0.5241547226905823,
+      "learning_rate": 3.5120166160797804e-05,
+      "loss": 0.3476,
+      "step": 128
+    },
+    {
+      "epoch": 0.4358108108108108,
+      "grad_norm": 0.5543828010559082,
+      "learning_rate": 3.517649517165415e-05,
+      "loss": 0.3517,
+      "step": 129
+    },
+    {
+      "epoch": 0.4391891891891892,
+      "grad_norm": 0.5167660117149353,
+      "learning_rate": 3.523238920511395e-05,
+      "loss": 0.3417,
+      "step": 130
+    },
+    {
+      "epoch": 0.44256756756756754,
+      "grad_norm": 0.5201629400253296,
+      "learning_rate": 3.528785492759607e-05,
+      "loss": 0.3543,
+      "step": 131
+    },
+    {
+      "epoch": 0.44594594594594594,
+      "grad_norm": 0.5214123129844666,
+      "learning_rate": 3.5342898853430836e-05,
+      "loss": 0.3552,
+      "step": 132
+    },
+    {
+      "epoch": 0.44932432432432434,
+      "grad_norm": 0.526899516582489,
+      "learning_rate": 3.539752734945143e-05,
+      "loss": 0.3469,
+      "step": 133
+    },
+    {
+      "epoch": 0.4527027027027027,
+      "grad_norm": 0.55535888671875,
+      "learning_rate": 3.5451746639413466e-05,
+      "loss": 0.294,
+      "step": 134
+    },
+    {
+      "epoch": 0.4560810810810811,
+      "grad_norm": 0.46593329310417175,
+      "learning_rate": 3.550556280825011e-05,
+      "loss": 0.2557,
+      "step": 135
+    },
+    {
+      "epoch": 0.4594594594594595,
+      "grad_norm": 0.5066341757774353,
+      "learning_rate": 3.55589818061703e-05,
+      "loss": 0.2885,
+      "step": 136
+    },
+    {
+      "epoch": 0.46283783783783783,
+      "grad_norm": 0.4943198263645172,
+      "learning_rate": 3.561200945260678e-05,
+      "loss": 0.3436,
+      "step": 137
+    },
+    {
+      "epoch": 0.46621621621621623,
+      "grad_norm": 0.5013923048973083,
+      "learning_rate": 3.5664651440020616e-05,
+      "loss": 0.2865,
+      "step": 138
+    },
+    {
+      "epoch": 0.46959459459459457,
+      "grad_norm": 0.5165452361106873,
+      "learning_rate": 3.571691333756825e-05,
+      "loss": 0.3394,
+      "step": 139
+    },
+    {
+      "epoch": 0.47297297297297297,
+      "grad_norm": 0.5002908706665039,
+      "learning_rate": 3.5768800594637304e-05,
+      "loss": 0.3278,
+      "step": 140
+    },
+    {
+      "epoch": 0.47635135135135137,
+      "grad_norm": 0.5236304402351379,
+      "learning_rate": 3.582031854425634e-05,
+      "loss": 0.3343,
+      "step": 141
+    },
+    {
+      "epoch": 0.4797297297297297,
+      "grad_norm": 0.49403268098831177,
+      "learning_rate": 3.587147240638428e-05,
+      "loss": 0.3079,
+      "step": 142
+    },
+    {
+      "epoch": 0.4831081081081081,
+      "grad_norm": 0.4847567677497864,
+      "learning_rate": 3.5922267291084366e-05,
+      "loss": 0.3305,
+      "step": 143
+    },
+    {
+      "epoch": 0.4864864864864865,
+      "grad_norm": 0.5461480617523193,
+      "learning_rate": 3.5972708201587496e-05,
+      "loss": 0.3242,
+      "step": 144
+    },
+    {
+      "epoch": 0.48986486486486486,
+      "grad_norm": 0.5380986332893372,
+      "learning_rate": 3.6022800037249585e-05,
+      "loss": 0.3341,
+      "step": 145
+    },
+    {
+      "epoch": 0.49324324324324326,
+      "grad_norm": 0.49580562114715576,
+      "learning_rate": 3.607254759640729e-05,
+      "loss": 0.31,
+      "step": 146
+    },
+    {
+      "epoch": 0.4966216216216216,
+      "grad_norm": 0.5374506115913391,
+      "learning_rate": 3.612195557913627e-05,
+      "loss": 0.3048,
+      "step": 147
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.535591721534729,
+      "learning_rate": 3.6171028589915954e-05,
+      "loss": 0.3266,
+      "step": 148
+    },
+    {
+      "epoch": 0.5033783783783784,
+      "grad_norm": 0.5777239799499512,
+      "learning_rate": 3.6219771140204575e-05,
+      "loss": 0.3509,
+      "step": 149
+    },
+    {
+      "epoch": 0.5067567567567568,
+      "grad_norm": 0.5570788383483887,
+      "learning_rate": 3.626818765092802e-05,
+      "loss": 0.3348,
+      "step": 150
+    },
+    {
+      "epoch": 0.5101351351351351,
+      "grad_norm": 0.5780688524246216,
+      "learning_rate": 3.6316282454886157e-05,
+      "loss": 0.3218,
+      "step": 151
+    },
+    {
+      "epoch": 0.5135135135135135,
+      "grad_norm": 0.5464823842048645,
+      "learning_rate": 3.636405979907955e-05,
+      "loss": 0.347,
+      "step": 152
+    },
+    {
+      "epoch": 0.5168918918918919,
+      "grad_norm": 0.46229949593544006,
+      "learning_rate": 3.6411523846959985e-05,
+      "loss": 0.2618,
+      "step": 153
+    },
+    {
+      "epoch": 0.5202702702702703,
+      "grad_norm": 0.5127717852592468,
+      "learning_rate": 3.645867868060772e-05,
+      "loss": 0.3364,
+      "step": 154
+    },
+    {
+      "epoch": 0.5236486486486487,
+      "grad_norm": 0.4915751814842224,
+      "learning_rate": 3.6505528302838193e-05,
+      "loss": 0.327,
+      "step": 155
+    },
+    {
+      "epoch": 0.527027027027027,
+      "grad_norm": 0.4658984839916229,
+      "learning_rate": 3.6552076639241027e-05,
+      "loss": 0.3008,
+      "step": 156
+    },
+    {
+      "epoch": 0.5304054054054054,
+      "grad_norm": 0.4983694553375244,
+      "learning_rate": 3.65983275401539e-05,
+      "loss": 0.3153,
+      "step": 157
+    },
+    {
+      "epoch": 0.5337837837837838,
+      "grad_norm": 0.49560025334358215,
+      "learning_rate": 3.664428478257371e-05,
+      "loss": 0.3409,
+      "step": 158
+    },
+    {
+      "epoch": 0.5371621621621622,
+      "grad_norm": 0.5617703199386597,
+      "learning_rate": 3.668995207200753e-05,
+      "loss": 0.3295,
+      "step": 159
+    },
+    {
+      "epoch": 0.5405405405405406,
+      "grad_norm": 0.5226999521255493,
+      "learning_rate": 3.673533304426541e-05,
+      "loss": 0.3522,
+      "step": 160
+    },
+    {
+      "epoch": 0.543918918918919,
+      "grad_norm": 0.49685564637184143,
+      "learning_rate": 3.67804312671975e-05,
+      "loss": 0.3322,
+      "step": 161
+    },
+    {
+      "epoch": 0.5472972972972973,
+      "grad_norm": 0.543743371963501,
+      "learning_rate": 3.682525024237719e-05,
+      "loss": 0.3283,
+      "step": 162
+    },
+    {
+      "epoch": 0.5506756756756757,
+      "grad_norm": 0.5247477889060974,
+      "learning_rate": 3.6869793406732636e-05,
+      "loss": 0.3104,
+      "step": 163
+    },
+    {
+      "epoch": 0.5540540540540541,
+      "grad_norm": 0.5228151679039001,
+      "learning_rate": 3.69140641341283e-05,
+      "loss": 0.3292,
+      "step": 164
+    },
+    {
+      "epoch": 0.5574324324324325,
+      "grad_norm": 0.48610207438468933,
+      "learning_rate": 3.695806573689844e-05,
+      "loss": 0.3183,
+      "step": 165
+    },
+    {
+      "epoch": 0.5608108108108109,
+      "grad_norm": 0.4703191816806793,
+      "learning_rate": 3.700180146733426e-05,
+      "loss": 0.2935,
+      "step": 166
+    },
+    {
+      "epoch": 0.5641891891891891,
+      "grad_norm": 0.5501482486724854,
+      "learning_rate": 3.704527451912639e-05,
+      "loss": 0.3249,
+      "step": 167
+    },
+    {
+      "epoch": 0.5675675675675675,
+      "grad_norm": 0.45025497674942017,
+      "learning_rate": 3.708848802876438e-05,
+      "loss": 0.3128,
+      "step": 168
+    },
+    {
+      "epoch": 0.5709459459459459,
+      "grad_norm": 0.5103681087493896,
+      "learning_rate": 3.7131445076894564e-05,
+      "loss": 0.2882,
+      "step": 169
+    },
+    {
+      "epoch": 0.5743243243243243,
+      "grad_norm": 0.4952690601348877,
+      "learning_rate": 3.717414868963791e-05,
+      "loss": 0.3268,
+      "step": 170
+    },
+    {
+      "epoch": 0.5777027027027027,
+      "grad_norm": 0.4887889325618744,
+      "learning_rate": 3.721660183986924e-05,
+      "loss": 0.3161,
+      "step": 171
+    },
+    {
+      "epoch": 0.581081081081081,
+      "grad_norm": 0.5142275094985962,
+      "learning_rate": 3.725880744845915e-05,
+      "loss": 0.2942,
+      "step": 172
+    },
+    {
+      "epoch": 0.5844594594594594,
+      "grad_norm": 0.518671452999115,
+      "learning_rate": 3.730076838547993e-05,
+      "loss": 0.3232,
+      "step": 173
+    },
+    {
+      "epoch": 0.5878378378378378,
+      "grad_norm": 0.5324185490608215,
+      "learning_rate": 3.734248747137666e-05,
+      "loss": 0.3115,
+      "step": 174
+    },
+    {
+      "epoch": 0.5912162162162162,
+      "grad_norm": 0.5307193398475647,
+      "learning_rate": 3.738396747810492e-05,
+      "loss": 0.3245,
+      "step": 175
+    },
+    {
+      "epoch": 0.5945945945945946,
+      "grad_norm": 0.47318169474601746,
+      "learning_rate": 3.7425211130235834e-05,
+      "loss": 0.3152,
+      "step": 176
+    },
+    {
+      "epoch": 0.597972972972973,
+      "grad_norm": 0.5233657956123352,
+      "learning_rate": 3.7466221106030115e-05,
+      "loss": 0.3115,
+      "step": 177
+    },
+    {
+      "epoch": 0.6013513513513513,
+      "grad_norm": 0.4699764847755432,
+      "learning_rate": 3.750700003848157e-05,
+      "loss": 0.2948,
+      "step": 178
+    },
+    {
+      "epoch": 0.6047297297297297,
+      "grad_norm": 0.5511758327484131,
+      "learning_rate": 3.7547550516331555e-05,
+      "loss": 0.337,
+      "step": 179
+    },
+    {
+      "epoch": 0.6081081081081081,
+      "grad_norm": 0.5334018468856812,
+      "learning_rate": 3.75878750850551e-05,
+      "loss": 0.3244,
+      "step": 180
+    },
+    {
+      "epoch": 0.6114864864864865,
+      "grad_norm": 0.5040500164031982,
+      "learning_rate": 3.7627976247819744e-05,
+      "loss": 0.3173,
+      "step": 181
+    },
+    {
+      "epoch": 0.6148648648648649,
+      "grad_norm": 0.5094459652900696,
+      "learning_rate": 3.766785646641792e-05,
+      "loss": 0.3087,
+      "step": 182
+    },
+    {
+      "epoch": 0.6182432432432432,
+      "grad_norm": 0.4679316282272339,
+      "learning_rate": 3.770751816217383e-05,
+      "loss": 0.3261,
+      "step": 183
+    },
+    {
+      "epoch": 0.6216216216216216,
+      "grad_norm": 0.5099210143089294,
+      "learning_rate": 3.7746963716825615e-05,
+      "loss": 0.3239,
+      "step": 184
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 0.4242151379585266,
+      "learning_rate": 3.778619547338356e-05,
+      "loss": 0.3082,
+      "step": 185
+    },
+    {
+      "epoch": 0.6283783783783784,
+      "grad_norm": 0.4796642065048218,
+      "learning_rate": 3.782521573696528e-05,
+      "loss": 0.2753,
+      "step": 186
+    },
+    {
+      "epoch": 0.6317567567567568,
+      "grad_norm": 0.43798592686653137,
+      "learning_rate": 3.786402677560832e-05,
+      "loss": 0.3033,
+      "step": 187
+    },
+    {
+      "epoch": 0.6351351351351351,
+      "grad_norm": 0.4947024881839752,
+      "learning_rate": 3.790263082106134e-05,
+      "loss": 0.3171,
+      "step": 188
+    },
+    {
+      "epoch": 0.6385135135135135,
+      "grad_norm": 0.40923503041267395,
+      "learning_rate": 3.794103006955407e-05,
+      "loss": 0.2927,
+      "step": 189
+    },
+    {
+      "epoch": 0.6418918918918919,
+      "grad_norm": 0.6707382202148438,
+      "learning_rate": 3.797922668254715e-05,
+      "loss": 0.2885,
+      "step": 190
+    },
+    {
+      "epoch": 0.6452702702702703,
+      "grad_norm": 0.4996836185455322,
+      "learning_rate": 3.801722278746213e-05,
+      "loss": 0.3379,
+      "step": 191
+    },
+    {
+      "epoch": 0.6486486486486487,
+      "grad_norm": 0.4873179495334625,
+      "learning_rate": 3.8055020478392495e-05,
+      "loss": 0.3091,
+      "step": 192
+    },
+    {
+      "epoch": 0.652027027027027,
+      "grad_norm": 0.4989306330680847,
+      "learning_rate": 3.809262181679623e-05,
+      "loss": 0.3758,
+      "step": 193
+    },
+    {
+      "epoch": 0.6554054054054054,
+      "grad_norm": 0.5524582862854004,
+      "learning_rate": 3.813002883217044e-05,
+      "loss": 0.3044,
+      "step": 194
+    },
+    {
+      "epoch": 0.6587837837837838,
+      "grad_norm": 0.5253859162330627,
+      "learning_rate": 3.816724352270863e-05,
+      "loss": 0.3556,
+      "step": 195
+    },
+    {
+      "epoch": 0.6621621621621622,
+      "grad_norm": 0.5060839653015137,
+      "learning_rate": 3.8204267855941266e-05,
+      "loss": 0.3352,
+      "step": 196
+    },
+    {
+      "epoch": 0.6655405405405406,
+      "grad_norm": 0.5375229716300964,
+      "learning_rate": 3.824110376935989e-05,
+      "loss": 0.3328,
+      "step": 197
+    },
+    {
+      "epoch": 0.668918918918919,
+      "grad_norm": 0.4941028654575348,
+      "learning_rate": 3.827775317102552e-05,
+      "loss": 0.296,
+      "step": 198
+    },
+    {
+      "epoch": 0.6722972972972973,
+      "grad_norm": 0.5323918461799622,
+      "learning_rate": 3.831421794016178e-05,
+      "loss": 0.3049,
+      "step": 199
+    },
+    {
+      "epoch": 0.6756756756756757,
+      "grad_norm": 0.5118021965026855,
+      "learning_rate": 3.835049992773302e-05,
+      "loss": 0.3133,
+      "step": 200
+    },
+    {
+      "epoch": 0.6790540540540541,
+      "grad_norm": 0.46183663606643677,
+      "learning_rate": 3.838660095700815e-05,
+      "loss": 0.3185,
+      "step": 201
+    },
+    {
+      "epoch": 0.6824324324324325,
+      "grad_norm": 0.47574761509895325,
+      "learning_rate": 3.84225228241104e-05,
+      "loss": 0.302,
+      "step": 202
+    },
+    {
+      "epoch": 0.6858108108108109,
+      "grad_norm": 0.5129667520523071,
+      "learning_rate": 3.8458267298553554e-05,
+      "loss": 0.3407,
+      "step": 203
+    },
+    {
+      "epoch": 0.6891891891891891,
+      "grad_norm": 1.532895803451538,
+      "learning_rate": 3.8493836123764984e-05,
+      "loss": 0.2858,
+      "step": 204
+    },
+    {
+      "epoch": 0.6925675675675675,
+      "grad_norm": 0.5879459381103516,
+      "learning_rate": 3.852923101759591e-05,
+      "loss": 0.3638,
+      "step": 205
+    },
+    {
+      "epoch": 0.6959459459459459,
+      "grad_norm": 0.589292585849762,
+      "learning_rate": 3.856445367281923e-05,
+      "loss": 0.3077,
+      "step": 206
+    },
+    {
+      "epoch": 0.6993243243243243,
+      "grad_norm": 0.44842028617858887,
+      "learning_rate": 3.859950575761529e-05,
+      "loss": 0.2791,
+      "step": 207
+    },
+    {
+      "epoch": 0.7027027027027027,
+      "grad_norm": 0.6291372179985046,
+      "learning_rate": 3.8634388916046025e-05,
+      "loss": 0.3404,
+      "step": 208
+    },
+    {
+      "epoch": 0.706081081081081,
+      "grad_norm": 0.4508068561553955,
+      "learning_rate": 3.866910476851757e-05,
+      "loss": 0.2859,
+      "step": 209
+    },
+    {
+      "epoch": 0.7094594594594594,
+      "grad_norm": 0.7987334132194519,
+      "learning_rate": 3.870365491223199e-05,
+      "loss": 0.3224,
+      "step": 210
+    },
+    {
+      "epoch": 0.7128378378378378,
+      "grad_norm": 0.5030388236045837,
+      "learning_rate": 3.8738040921628215e-05,
+      "loss": 0.3225,
+      "step": 211
+    },
+    {
+      "epoch": 0.7162162162162162,
+      "grad_norm": 0.5433597564697266,
+      "learning_rate": 3.877226434881253e-05,
+      "loss": 0.3294,
+      "step": 212
+    },
+    {
+      "epoch": 0.7195945945945946,
+      "grad_norm": 0.48354920744895935,
+      "learning_rate": 3.880632672397897e-05,
+      "loss": 0.3035,
+      "step": 213
+    },
+    {
+      "epoch": 0.722972972972973,
+      "grad_norm": 0.5904508829116821,
+      "learning_rate": 3.884022955581985e-05,
+      "loss": 0.3261,
+      "step": 214
+    },
+    {
+      "epoch": 0.7263513513513513,
+      "grad_norm": 0.5152267813682556,
+      "learning_rate": 3.887397433192676e-05,
+      "loss": 0.2926,
+      "step": 215
+    },
+    {
+      "epoch": 0.7297297297297297,
+      "grad_norm": 0.46898239850997925,
+      "learning_rate": 3.890756251918219e-05,
+      "loss": 0.2808,
+      "step": 216
+    },
+    {
+      "epoch": 0.7331081081081081,
+      "grad_norm": 0.4848116338253021,
+      "learning_rate": 3.894099556414216e-05,
+      "loss": 0.2995,
+      "step": 217
+    },
+    {
+      "epoch": 0.7364864864864865,
+      "grad_norm": 0.48101919889450073,
+      "learning_rate": 3.897427489341009e-05,
+      "loss": 0.3054,
+      "step": 218
+    },
+    {
+      "epoch": 0.7398648648648649,
+      "grad_norm": 0.49954646825790405,
+      "learning_rate": 3.900740191400198e-05,
+      "loss": 0.3239,
+      "step": 219
+    },
+    {
+      "epoch": 0.7432432432432432,
+      "grad_norm": 0.46193623542785645,
+      "learning_rate": 3.904037801370344e-05,
+      "loss": 0.3215,
+      "step": 220
+    },
+    {
+      "epoch": 0.7466216216216216,
+      "grad_norm": 0.4544968605041504,
+      "learning_rate": 3.9073204561418514e-05,
+      "loss": 0.2829,
+      "step": 221
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4560447335243225,
+      "learning_rate": 3.9105882907510644e-05,
+      "loss": 0.3052,
+      "step": 222
+    },
+    {
+      "epoch": 0.7533783783783784,
+      "grad_norm": 0.4681329131126404,
+      "learning_rate": 3.913841438413601e-05,
+      "loss": 0.3259,
+      "step": 223
+    },
+    {
+      "epoch": 0.7567567567567568,
+      "grad_norm": 0.4784477949142456,
+      "learning_rate": 3.917080030556938e-05,
+      "loss": 0.3252,
+      "step": 224
+    },
+    {
+      "epoch": 0.7601351351351351,
+      "grad_norm": 0.4931364953517914,
+      "learning_rate": 3.9203041968522716e-05,
+      "loss": 0.3252,
+      "step": 225
+    },
+    {
+      "epoch": 0.7635135135135135,
+      "grad_norm": 0.4844168424606323,
+      "learning_rate": 3.923514065245669e-05,
+      "loss": 0.3185,
+      "step": 226
+    },
+    {
+      "epoch": 0.7668918918918919,
+      "grad_norm": 0.5114946961402893,
+      "learning_rate": 3.926709761988538e-05,
+      "loss": 0.3136,
+      "step": 227
+    },
+    {
+      "epoch": 0.7702702702702703,
+      "grad_norm": 0.4893558621406555,
+      "learning_rate": 3.929891411667424e-05,
+      "loss": 0.2977,
+      "step": 228
+    },
+    {
+      "epoch": 0.7736486486486487,
+      "grad_norm": 0.4515063762664795,
+      "learning_rate": 3.933059137233147e-05,
+      "loss": 0.3031,
+      "step": 229
+    },
+    {
+      "epoch": 0.777027027027027,
+      "grad_norm": 0.48304086923599243,
+      "learning_rate": 3.9362130600293214e-05,
+      "loss": 0.3292,
+      "step": 230
+    },
+    {
+      "epoch": 0.7804054054054054,
+      "grad_norm": 0.4545653760433197,
+      "learning_rate": 3.9393532998202405e-05,
+      "loss": 0.3014,
+      "step": 231
+    },
+    {
+      "epoch": 0.7837837837837838,
+      "grad_norm": 0.4236096739768982,
+      "learning_rate": 3.942479974818166e-05,
+      "loss": 0.2804,
+      "step": 232
+    },
+    {
+      "epoch": 0.7871621621621622,
+      "grad_norm": 0.4648077189922333,
+      "learning_rate": 3.945593201710032e-05,
+      "loss": 0.3243,
+      "step": 233
+    },
+    {
+      "epoch": 0.7905405405405406,
+      "grad_norm": 0.47247567772865295,
+      "learning_rate": 3.9486930956835724e-05,
+      "loss": 0.2907,
+      "step": 234
+    },
+    {
+      "epoch": 0.793918918918919,
+      "grad_norm": 0.4848794639110565,
+      "learning_rate": 3.951779770452894e-05,
+      "loss": 0.3265,
+      "step": 235
+    },
+    {
+      "epoch": 0.7972972972972973,
+      "grad_norm": 0.5114079713821411,
+      "learning_rate": 3.954853338283512e-05,
+      "loss": 0.3194,
+      "step": 236
+    },
+    {
+      "epoch": 0.8006756756756757,
+      "grad_norm": 0.4802190959453583,
+      "learning_rate": 3.9579139100168404e-05,
+      "loss": 0.2857,
+      "step": 237
+    },
+    {
+      "epoch": 0.8040540540540541,
+      "grad_norm": 0.522281289100647,
+      "learning_rate": 3.960961595094187e-05,
+      "loss": 0.2843,
+      "step": 238
+    },
+    {
+      "epoch": 0.8074324324324325,
+      "grad_norm": 0.4458360970020294,
+      "learning_rate": 3.96399650158023e-05,
+      "loss": 0.2755,
+      "step": 239
+    },
+    {
+      "epoch": 0.8108108108108109,
+      "grad_norm": 0.53693687915802,
+      "learning_rate": 3.96701873618601e-05,
+      "loss": 0.3015,
+      "step": 240
+    },
+    {
+      "epoch": 0.8141891891891891,
+      "grad_norm": 0.46697714924812317,
+      "learning_rate": 3.970028404291448e-05,
+      "loss": 0.2911,
+      "step": 241
+    },
+    {
+      "epoch": 0.8175675675675675,
+      "grad_norm": 0.5392897725105286,
+      "learning_rate": 3.9730256099673865e-05,
+      "loss": 0.3093,
+      "step": 242
+    },
+    {
+      "epoch": 0.8209459459459459,
+      "grad_norm": 0.5555176734924316,
+      "learning_rate": 3.976010455997187e-05,
+      "loss": 0.3078,
+      "step": 243
+    },
+    {
+      "epoch": 0.8243243243243243,
+      "grad_norm": 0.42975664138793945,
+      "learning_rate": 3.978983043897883e-05,
+      "loss": 0.2723,
+      "step": 244
+    },
+    {
+      "epoch": 0.8277027027027027,
+      "grad_norm": 0.6645532250404358,
+      "learning_rate": 3.981943473940888e-05,
+      "loss": 0.3161,
+      "step": 245
+    },
+    {
+      "epoch": 0.831081081081081,
+      "grad_norm": 0.47302716970443726,
+      "learning_rate": 3.984891845172299e-05,
+      "loss": 0.3002,
+      "step": 246
+    },
+    {
+      "epoch": 0.8344594594594594,
+      "grad_norm": 0.5395675301551819,
+      "learning_rate": 3.987828255432777e-05,
+      "loss": 0.3614,
+      "step": 247
+    },
+    {
+      "epoch": 0.8378378378378378,
+      "grad_norm": 0.5418782234191895,
+      "learning_rate": 3.9907528013770276e-05,
+      "loss": 0.3226,
+      "step": 248
+    },
+    {
+      "epoch": 0.8412162162162162,
+      "grad_norm": 0.4323344826698303,
+      "learning_rate": 3.993665578492894e-05,
+      "loss": 0.289,
+      "step": 249
+    },
+    {
+      "epoch": 0.8445945945945946,
+      "grad_norm": 0.5098227858543396,
+      "learning_rate": 3.9965666811200624e-05,
+      "loss": 0.3346,
+      "step": 250
+    },
+    {
+      "epoch": 0.847972972972973,
+      "grad_norm": 0.4577200412750244,
+      "learning_rate": 3.999456202468397e-05,
+      "loss": 0.3526,
+      "step": 251
+    },
+    {
+      "epoch": 0.8513513513513513,
+      "grad_norm": 0.48974189162254333,
+      "learning_rate": 4.002334234635907e-05,
+      "loss": 0.3086,
+      "step": 252
+    },
+    {
+      "epoch": 0.8547297297297297,
+      "grad_norm": 0.47080233693122864,
+      "learning_rate": 4.005200868626364e-05,
+      "loss": 0.3578,
+      "step": 253
+    },
+    {
+      "epoch": 0.8581081081081081,
+      "grad_norm": 0.5034478902816772,
+      "learning_rate": 4.008056194366564e-05,
+      "loss": 0.3166,
+      "step": 254
+    },
+    {
+      "epoch": 0.8614864864864865,
+      "grad_norm": 0.44603192806243896,
+      "learning_rate": 4.010900300723259e-05,
+      "loss": 0.2969,
+      "step": 255
+    },
+    {
+      "epoch": 0.8648648648648649,
+      "grad_norm": 0.5267269611358643,
+      "learning_rate": 4.013733275519749e-05,
+      "loss": 0.3341,
+      "step": 256
+    },
+    {
+      "epoch": 0.8682432432432432,
+      "grad_norm": 0.424925297498703,
+      "learning_rate": 4.016555205552158e-05,
+      "loss": 0.2942,
+      "step": 257
+    },
+    {
+      "epoch": 0.8716216216216216,
+      "grad_norm": 0.5034767389297485,
+      "learning_rate": 4.0193661766053834e-05,
+      "loss": 0.2668,
+      "step": 258
+    },
+    {
+      "epoch": 0.875,
+      "grad_norm": 0.5288417935371399,
+      "learning_rate": 4.022166273468753e-05,
+      "loss": 0.3424,
+      "step": 259
+    },
+    {
+      "epoch": 0.8783783783783784,
+      "grad_norm": 0.4726288616657257,
+      "learning_rate": 4.024955579951363e-05,
+      "loss": 0.2906,
+      "step": 260
+    },
+    {
+      "epoch": 0.8817567567567568,
+      "grad_norm": 0.5311090350151062,
+      "learning_rate": 4.027734178897136e-05,
+      "loss": 0.3307,
+      "step": 261
+    },
+    {
+      "epoch": 0.8851351351351351,
+      "grad_norm": 0.427002489566803,
+      "learning_rate": 4.030502152199576e-05,
+      "loss": 0.2569,
+      "step": 262
+    },
+    {
+      "epoch": 0.8885135135135135,
+      "grad_norm": 0.543899416923523,
+      "learning_rate": 4.033259580816264e-05,
+      "loss": 0.2925,
+      "step": 263
+    },
+    {
+      "epoch": 0.8918918918918919,
+      "grad_norm": 0.4932996928691864,
+      "learning_rate": 4.036006544783052e-05,
+      "loss": 0.3058,
+      "step": 264
+    },
+    {
+      "epoch": 0.8952702702702703,
+      "grad_norm": 0.5138298273086548,
+      "learning_rate": 4.0387431232280135e-05,
+      "loss": 0.2952,
+      "step": 265
+    },
+    {
+      "epoch": 0.8986486486486487,
+      "grad_norm": 0.4801797568798065,
+      "learning_rate": 4.041469394385112e-05,
+      "loss": 0.2873,
+      "step": 266
+    },
+    {
+      "epoch": 0.902027027027027,
+      "grad_norm": 0.453140527009964,
+      "learning_rate": 4.0441854356076257e-05,
+      "loss": 0.2747,
+      "step": 267
+    },
+    {
+      "epoch": 0.9054054054054054,
+      "grad_norm": 0.4941580593585968,
+      "learning_rate": 4.046891323381315e-05,
+      "loss": 0.3157,
+      "step": 268
+    },
+    {
+      "epoch": 0.9087837837837838,
+      "grad_norm": 0.6151431798934937,
+      "learning_rate": 4.049587133337347e-05,
+      "loss": 0.3139,
+      "step": 269
+    },
+    {
+      "epoch": 0.9121621621621622,
+      "grad_norm": 0.4674379229545593,
+      "learning_rate": 4.0522729402649793e-05,
+      "loss": 0.3122,
+      "step": 270
+    },
+    {
+      "epoch": 0.9155405405405406,
+      "grad_norm": 0.4719599187374115,
+      "learning_rate": 4.0549488181240096e-05,
+      "loss": 0.3101,
+      "step": 271
+    },
+    {
+      "epoch": 0.918918918918919,
+      "grad_norm": 0.483374685049057,
+      "learning_rate": 4.057614840056998e-05,
+      "loss": 0.3087,
+      "step": 272
+    },
+    {
+      "epoch": 0.9222972972972973,
+      "grad_norm": 0.48100754618644714,
+      "learning_rate": 4.06027107840126e-05,
+      "loss": 0.3044,
+      "step": 273
+    },
+    {
+      "epoch": 0.9256756756756757,
+      "grad_norm": 0.44335874915122986,
+      "learning_rate": 4.0629176047006474e-05,
+      "loss": 0.2929,
+      "step": 274
+    },
+    {
+      "epoch": 0.9290540540540541,
+      "grad_norm": 0.4475744962692261,
+      "learning_rate": 4.065554489717105e-05,
+      "loss": 0.321,
+      "step": 275
+    },
+    {
+      "epoch": 0.9324324324324325,
+      "grad_norm": 0.43009960651397705,
+      "learning_rate": 4.068181803442029e-05,
+      "loss": 0.3026,
+      "step": 276
+    },
+    {
+      "epoch": 0.9358108108108109,
+      "grad_norm": 0.48129919171333313,
+      "learning_rate": 4.0707996151074147e-05,
+      "loss": 0.2876,
+      "step": 277
+    },
+    {
+      "epoch": 0.9391891891891891,
+      "grad_norm": 0.423658549785614,
+      "learning_rate": 4.073407993196794e-05,
+      "loss": 0.2968,
+      "step": 278
+    },
+    {
+      "epoch": 0.9425675675675675,
+      "grad_norm": 0.485857218503952,
+      "learning_rate": 4.076007005455996e-05,
+      "loss": 0.3144,
+      "step": 279
+    },
+    {
+      "epoch": 0.9459459459459459,
+      "grad_norm": 0.4936545193195343,
+      "learning_rate": 4.0785967189036986e-05,
+      "loss": 0.3103,
+      "step": 280
+    },
+    {
+      "epoch": 0.9493243243243243,
+      "grad_norm": 0.47265681624412537,
+      "learning_rate": 4.0811771998418e-05,
+      "loss": 0.3136,
+      "step": 281
+    },
+    {
+      "epoch": 0.9527027027027027,
+      "grad_norm": 0.4872439503669739,
+      "learning_rate": 4.083748513865602e-05,
+      "loss": 0.335,
+      "step": 282
+    },
+    {
+      "epoch": 0.956081081081081,
+      "grad_norm": 0.4380621910095215,
+      "learning_rate": 4.086310725873818e-05,
+      "loss": 0.3036,
+      "step": 283
+    },
+    {
+      "epoch": 0.9594594594594594,
+      "grad_norm": 0.4759541451931,
+      "learning_rate": 4.0888639000783966e-05,
+      "loss": 0.2827,
+      "step": 284
+    },
+    {
+      "epoch": 0.9628378378378378,
+      "grad_norm": 0.995196521282196,
+      "learning_rate": 4.0914081000141844e-05,
+      "loss": 0.3162,
+      "step": 285
+    },
+    {
+      "epoch": 0.9662162162162162,
+      "grad_norm": 0.5686355829238892,
+      "learning_rate": 4.0939433885484055e-05,
+      "loss": 0.2992,
+      "step": 286
+    },
+    {
+      "epoch": 0.9695945945945946,
+      "grad_norm": 0.43086209893226624,
+      "learning_rate": 4.0964698278899874e-05,
+      "loss": 0.278,
+      "step": 287
+    },
+    {
+      "epoch": 0.972972972972973,
+      "grad_norm": 0.5099641680717468,
+      "learning_rate": 4.0989874795987185e-05,
+      "loss": 0.3023,
+      "step": 288
+    },
+    {
+      "epoch": 0.9763513513513513,
+      "grad_norm": 0.39634063839912415,
+      "learning_rate": 4.1014964045942465e-05,
+      "loss": 0.2806,
+      "step": 289
+    },
+    {
+      "epoch": 0.9797297297297297,
+      "grad_norm": 0.4685455858707428,
+      "learning_rate": 4.103996663164927e-05,
+      "loss": 0.3078,
+      "step": 290
+    },
+    {
+      "epoch": 0.9831081081081081,
+      "grad_norm": 0.4501771330833435,
+      "learning_rate": 4.106488314976513e-05,
+      "loss": 0.3048,
+      "step": 291
+    },
+    {
+      "epoch": 0.9864864864864865,
+      "grad_norm": 0.4352227747440338,
+      "learning_rate": 4.108971419080698e-05,
+      "loss": 0.305,
+      "step": 292
+    },
+    {
+      "epoch": 0.9898648648648649,
+      "grad_norm": 0.4946969151496887,
+      "learning_rate": 4.111446033923516e-05,
+      "loss": 0.3027,
+      "step": 293
+    },
+    {
+      "epoch": 0.9932432432432432,
+      "grad_norm": 0.463313490152359,
+      "learning_rate": 4.113912217353596e-05,
+      "loss": 0.3148,
+      "step": 294
+    },
+    {
+      "epoch": 0.9966216216216216,
+      "grad_norm": 0.42501333355903625,
+      "learning_rate": 4.116370026630272e-05,
+      "loss": 0.303,
+      "step": 295
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4174898862838745,
+      "learning_rate": 4.118819518431564e-05,
+      "loss": 0.2768,
+      "step": 296
+    },
+    {
+      "epoch": 1.0033783783783783,
+      "grad_norm": 0.5591414570808411,
+      "learning_rate": 4.121260748862021e-05,
+      "loss": 0.2319,
+      "step": 297
+    },
+    {
+      "epoch": 1.0067567567567568,
+      "grad_norm": 0.4628024101257324,
+      "learning_rate": 4.123693773460426e-05,
+      "loss": 0.2332,
+      "step": 298
+    },
+    {
+      "epoch": 1.010135135135135,
+      "grad_norm": 0.9493624567985535,
+      "learning_rate": 4.126118647207383e-05,
+      "loss": 0.2113,
+      "step": 299
+    },
+    {
+      "epoch": 1.0135135135135136,
+      "grad_norm": 0.6151244044303894,
+      "learning_rate": 4.1285354245327715e-05,
+      "loss": 0.2428,
+      "step": 300
+    },
+    {
+      "epoch": 1.0168918918918919,
+      "grad_norm": 0.4934402406215668,
+      "learning_rate": 4.1309441593230726e-05,
+      "loss": 0.2394,
+      "step": 301
+    },
+    {
+      "epoch": 1.0202702702702702,
+      "grad_norm": 0.5464348793029785,
+      "learning_rate": 4.133344904928585e-05,
+      "loss": 0.2584,
+      "step": 302
+    },
+    {
+      "epoch": 1.0236486486486487,
+      "grad_norm": 0.754717230796814,
+      "learning_rate": 4.1357377141705084e-05,
+      "loss": 0.2191,
+      "step": 303
+    },
+    {
+      "epoch": 1.027027027027027,
+      "grad_norm": 0.48623141646385193,
+      "learning_rate": 4.1381226393479236e-05,
+      "loss": 0.2185,
+      "step": 304
+    },
+    {
+      "epoch": 1.0304054054054055,
+      "grad_norm": 32.5340461730957,
+      "learning_rate": 4.1404997322446435e-05,
+      "loss": 0.2328,
+      "step": 305
+    },
+    {
+      "epoch": 1.0337837837837838,
+      "grad_norm": 0.7727927565574646,
+      "learning_rate": 4.142869044135967e-05,
+      "loss": 0.2197,
+      "step": 306
+    },
+    {
+      "epoch": 1.037162162162162,
+      "grad_norm": 0.47851279377937317,
+      "learning_rate": 4.145230625795311e-05,
+      "loss": 0.2093,
+      "step": 307
+    },
+    {
+      "epoch": 1.0405405405405406,
+      "grad_norm": 0.684687077999115,
+      "learning_rate": 4.14758452750074e-05,
+      "loss": 0.2145,
+      "step": 308
+    },
+    {
+      "epoch": 1.0439189189189189,
+      "grad_norm": 0.4834594428539276,
+      "learning_rate": 4.149930799041392e-05,
+      "loss": 0.222,
+      "step": 309
+    },
+    {
+      "epoch": 1.0472972972972974,
+      "grad_norm": 0.5479368567466736,
+      "learning_rate": 4.152269489723788e-05,
+      "loss": 0.2245,
+      "step": 310
+    },
+    {
+      "epoch": 1.0506756756756757,
+      "grad_norm": 0.48793625831604004,
+      "learning_rate": 4.1546006483780626e-05,
+      "loss": 0.2428,
+      "step": 311
+    },
+    {
+      "epoch": 1.054054054054054,
+      "grad_norm": 0.5712085962295532,
+      "learning_rate": 4.156924323364072e-05,
+      "loss": 0.2311,
+      "step": 312
+    },
+    {
+      "epoch": 1.0574324324324325,
+      "grad_norm": 0.46200135350227356,
+      "learning_rate": 4.1592405625774144e-05,
+      "loss": 0.2315,
+      "step": 313
+    },
+    {
+      "epoch": 1.0608108108108107,
+      "grad_norm": 0.5134051442146301,
+      "learning_rate": 4.161549413455358e-05,
+      "loss": 0.2205,
+      "step": 314
+    },
+    {
+      "epoch": 1.0641891891891893,
+      "grad_norm": 0.4155457019805908,
+      "learning_rate": 4.163850922982668e-05,
+      "loss": 0.1969,
+      "step": 315
+    },
+    {
+      "epoch": 1.0675675675675675,
+      "grad_norm": 0.47662997245788574,
+      "learning_rate": 4.16614513769734e-05,
+      "loss": 0.2183,
+      "step": 316
+    },
+    {
+      "epoch": 1.070945945945946,
+      "grad_norm": 0.4776044189929962,
+      "learning_rate": 4.1684321036962526e-05,
+      "loss": 0.2581,
+      "step": 317
+    },
+    {
+      "epoch": 1.0743243243243243,
+      "grad_norm": 0.5491393804550171,
+      "learning_rate": 4.170711866640721e-05,
+      "loss": 0.2454,
+      "step": 318
+    },
+    {
+      "epoch": 1.0777027027027026,
+      "grad_norm": 0.4745365381240845,
+      "learning_rate": 4.1729844717619684e-05,
+      "loss": 0.2303,
+      "step": 319
+    },
+    {
+      "epoch": 1.0810810810810811,
+      "grad_norm": 0.49116480350494385,
+      "learning_rate": 4.17524996386651e-05,
+      "loss": 0.2205,
+      "step": 320
+    },
+    {
+      "epoch": 1.0844594594594594,
+      "grad_norm": 0.47651857137680054,
+      "learning_rate": 4.177508387341454e-05,
+      "loss": 0.2538,
+      "step": 321
+    },
+    {
+      "epoch": 1.087837837837838,
+      "grad_norm": 24.205638885498047,
+      "learning_rate": 4.179759786159719e-05,
+      "loss": 0.5427,
+      "step": 322
+    },
+    {
+      "epoch": 1.0912162162162162,
+      "grad_norm": 0.7813121676445007,
+      "learning_rate": 4.182004203885172e-05,
+      "loss": 0.2445,
+      "step": 323
+    },
+    {
+      "epoch": 1.0945945945945945,
+      "grad_norm": 52.29485321044922,
+      "learning_rate": 4.184241683677687e-05,
+      "loss": 0.2577,
+      "step": 324
+    },
+    {
+      "epoch": 1.097972972972973,
+      "grad_norm": 0.8282439112663269,
+      "learning_rate": 4.1864722682981245e-05,
+      "loss": 0.2421,
+      "step": 325
+    },
+    {
+      "epoch": 1.1013513513513513,
+      "grad_norm": 0.4729321300983429,
+      "learning_rate": 4.188696000113232e-05,
+      "loss": 0.2122,
+      "step": 326
+    },
+    {
+      "epoch": 1.1047297297297298,
+      "grad_norm": 0.6732134819030762,
+      "learning_rate": 4.190912921100477e-05,
+      "loss": 0.2295,
+      "step": 327
+    },
+    {
+      "epoch": 1.1081081081081081,
+      "grad_norm": 0.567088782787323,
+      "learning_rate": 4.1931230728527994e-05,
+      "loss": 0.2352,
+      "step": 328
+    },
+    {
+      "epoch": 1.1114864864864864,
+      "grad_norm": 8.066313743591309,
+      "learning_rate": 4.195326496583291e-05,
+      "loss": 0.7099,
+      "step": 329
+    },
+    {
+      "epoch": 1.114864864864865,
+      "grad_norm": 1.0606003999710083,
+      "learning_rate": 4.1975232331298125e-05,
+      "loss": 0.2105,
+      "step": 330
+    },
+    {
+      "epoch": 1.1182432432432432,
+      "grad_norm": 0.4454852342605591,
+      "learning_rate": 4.1997133229595316e-05,
+      "loss": 0.2063,
+      "step": 331
+    },
+    {
+      "epoch": 1.1216216216216217,
+      "grad_norm": 0.873603880405426,
+      "learning_rate": 4.201896806173394e-05,
+      "loss": 0.2447,
+      "step": 332
+    },
+    {
+      "epoch": 1.125,
+      "grad_norm": 0.6670548319816589,
+      "learning_rate": 4.2040737225105335e-05,
+      "loss": 0.2298,
+      "step": 333
+    },
+    {
+      "epoch": 1.1283783783783783,
+      "grad_norm": 0.5834859609603882,
+      "learning_rate": 4.206244111352608e-05,
+      "loss": 0.1799,
+      "step": 334
+    },
+    {
+      "epoch": 1.1317567567567568,
+      "grad_norm": 0.8503464460372925,
+      "learning_rate": 4.2084080117280756e-05,
+      "loss": 0.2254,
+      "step": 335
+    },
+    {
+      "epoch": 1.135135135135135,
+      "grad_norm": 0.5728862285614014,
+      "learning_rate": 4.210565462316407e-05,
+      "loss": 0.2342,
+      "step": 336
+    },
+    {
+      "epoch": 1.1385135135135136,
+      "grad_norm": 0.7374505400657654,
+      "learning_rate": 4.2127165014522315e-05,
+      "loss": 0.2384,
+      "step": 337
+    },
+    {
+      "epoch": 1.1418918918918919,
+      "grad_norm": 0.5631051659584045,
+      "learning_rate": 4.214861167129425e-05,
+      "loss": 0.2213,
+      "step": 338
+    },
+    {
+      "epoch": 1.1452702702702702,
+      "grad_norm": 26.428640365600586,
+      "learning_rate": 4.2169994970051365e-05,
+      "loss": 0.8302,
+      "step": 339
+    },
+    {
+      "epoch": 1.1486486486486487,
+      "grad_norm": 1.1665905714035034,
+      "learning_rate": 4.219131528403759e-05,
+      "loss": 0.2568,
+      "step": 340
+    },
+    {
+      "epoch": 1.152027027027027,
+      "grad_norm": 0.6159250140190125,
+      "learning_rate": 4.22125729832083e-05,
+      "loss": 0.2574,
+      "step": 341
+    },
+    {
+      "epoch": 1.1554054054054055,
+      "grad_norm": 182.47439575195312,
+      "learning_rate": 4.2233768434268914e-05,
+      "loss": 0.8118,
+      "step": 342
+    },
+    {
+      "epoch": 1.1587837837837838,
+      "grad_norm": 1.685766339302063,
+      "learning_rate": 4.225490200071284e-05,
+      "loss": 0.2347,
+      "step": 343
+    },
+    {
+      "epoch": 1.1621621621621623,
+      "grad_norm": 0.9619600176811218,
+      "learning_rate": 4.227597404285883e-05,
+      "loss": 0.2453,
+      "step": 344
+    },
+    {
+      "epoch": 1.1655405405405406,
+      "grad_norm": 0.49926599860191345,
+      "learning_rate": 4.229698491788791e-05,
+      "loss": 0.2313,
+      "step": 345
+    },
+    {
+      "epoch": 1.1689189189189189,
+      "grad_norm": 0.9699143767356873,
+      "learning_rate": 4.231793497987961e-05,
+      "loss": 0.256,
+      "step": 346
+    },
+    {
+      "epoch": 1.1722972972972974,
+      "grad_norm": 0.7428460121154785,
+      "learning_rate": 4.2338824579847904e-05,
+      "loss": 0.2655,
+      "step": 347
+    },
+    {
+      "epoch": 1.1756756756756757,
+      "grad_norm": 0.6480849385261536,
+      "learning_rate": 4.235965406577636e-05,
+      "loss": 0.2385,
+      "step": 348
+    },
+    {
+      "epoch": 1.179054054054054,
+      "grad_norm": 0.6347863078117371,
+      "learning_rate": 4.2380423782653e-05,
+      "loss": 0.2196,
+      "step": 349
+    },
+    {
+      "epoch": 1.1824324324324325,
+      "grad_norm": 0.6471324563026428,
+      "learning_rate": 4.240113407250459e-05,
+      "loss": 0.2337,
+      "step": 350
+    },
+    {
+      "epoch": 1.1858108108108107,
+      "grad_norm": 0.5807657241821289,
+      "learning_rate": 4.24217852744304e-05,
+      "loss": 0.2438,
+      "step": 351
+    },
+    {
+      "epoch": 1.1891891891891893,
+      "grad_norm": 0.5200501084327698,
+      "learning_rate": 4.244237772463552e-05,
+      "loss": 0.2458,
+      "step": 352
+    },
+    {
+      "epoch": 1.1925675675675675,
+      "grad_norm": 0.5445655584335327,
+      "learning_rate": 4.246291175646371e-05,
+      "loss": 0.2391,
+      "step": 353
+    },
+    {
+      "epoch": 1.195945945945946,
+      "grad_norm": 0.6283414363861084,
+      "learning_rate": 4.24833877004298e-05,
+      "loss": 0.2299,
+      "step": 354
+    },
+    {
+      "epoch": 1.1993243243243243,
+      "grad_norm": 0.5581656694412231,
+      "learning_rate": 4.250380588425157e-05,
+      "loss": 0.2272,
+      "step": 355
+    },
+    {
+      "epoch": 1.2027027027027026,
+      "grad_norm": 0.5711120367050171,
+      "learning_rate": 4.2524166632881255e-05,
+      "loss": 0.2737,
+      "step": 356
+    },
+    {
+      "epoch": 1.2060810810810811,
+      "grad_norm": 64.4117660522461,
+      "learning_rate": 4.254447026853656e-05,
+      "loss": 0.2648,
+      "step": 357
+    },
+    {
+      "epoch": 1.2094594594594594,
+      "grad_norm": 0.84456467628479,
+      "learning_rate": 4.2564717110731244e-05,
+      "loss": 0.2643,
+      "step": 358
+    },
+    {
+      "epoch": 1.2128378378378377,
+      "grad_norm": 1.7171592712402344,
+      "learning_rate": 4.258490747630532e-05,
+      "loss": 0.6041,
+      "step": 359
+    },
+    {
+      "epoch": 1.2162162162162162,
+      "grad_norm": 0.6733882427215576,
+      "learning_rate": 4.260504167945479e-05,
+      "loss": 0.2226,
+      "step": 360
+    },
+    {
+      "epoch": 1.2195945945945945,
+      "grad_norm": 0.5079653859138489,
+      "learning_rate": 4.2625120031760965e-05,
+      "loss": 0.2205,
+      "step": 361
+    },
+    {
+      "epoch": 1.222972972972973,
+      "grad_norm": 0.5646266341209412,
+      "learning_rate": 4.264514284221944e-05,
+      "loss": 0.2272,
+      "step": 362
+    },
+    {
+      "epoch": 1.2263513513513513,
+      "grad_norm": 0.5336301922798157,
+      "learning_rate": 4.266511041726854e-05,
+      "loss": 0.2536,
+      "step": 363
+    },
+    {
+      "epoch": 1.2297297297297298,
+      "grad_norm": 0.5218775272369385,
+      "learning_rate": 4.26850230608176e-05,
+      "loss": 0.2582,
+      "step": 364
+    },
+    {
+      "epoch": 1.2331081081081081,
+      "grad_norm": 0.47051167488098145,
+      "learning_rate": 4.2704881074274584e-05,
+      "loss": 0.2288,
+      "step": 365
+    },
+    {
+      "epoch": 1.2364864864864864,
+      "grad_norm": 0.4820377826690674,
+      "learning_rate": 4.272468475657351e-05,
+      "loss": 0.2396,
+      "step": 366
+    },
+    {
+      "epoch": 1.239864864864865,
+      "grad_norm": 1219.8048095703125,
+      "learning_rate": 4.2744434404201497e-05,
+      "loss": 1.9213,
+      "step": 367
+    },
+    {
+      "epoch": 1.2432432432432432,
+      "grad_norm": 1.3034559488296509,
+      "learning_rate": 4.27641303112253e-05,
+      "loss": 0.2528,
+      "step": 368
+    },
+    {
+      "epoch": 1.2466216216216217,
+      "grad_norm": 0.6443231105804443,
+      "learning_rate": 4.278377276931767e-05,
+      "loss": 0.2235,
+      "step": 369
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.4247282147407532,
+      "learning_rate": 4.2803362067783256e-05,
+      "loss": 0.2111,
+      "step": 370
+    },
+    {
+      "epoch": 1.2533783783783785,
+      "grad_norm": 0.5400434136390686,
+      "learning_rate": 4.2822898493584104e-05,
+      "loss": 0.2256,
+      "step": 371
+    },
+    {
+      "epoch": 1.2567567567567568,
+      "grad_norm": 0.43781599402427673,
+      "learning_rate": 4.284238233136496e-05,
+      "loss": 0.2295,
+      "step": 372
+    },
+    {
+      "epoch": 1.260135135135135,
+      "grad_norm": 0.5443878173828125,
+      "learning_rate": 4.286181386347813e-05,
+      "loss": 0.2394,
+      "step": 373
+    },
+    {
+      "epoch": 1.2635135135135136,
+      "grad_norm": 0.4616173207759857,
+      "learning_rate": 4.288119337000801e-05,
+      "loss": 0.2108,
+      "step": 374
+    },
+    {
+      "epoch": 1.2668918918918919,
+      "grad_norm": 0.4537034332752228,
+      "learning_rate": 4.2900521128795315e-05,
+      "loss": 0.2192,
+      "step": 375
+    },
+    {
+      "epoch": 1.2702702702702702,
+      "grad_norm": 0.5639699697494507,
+      "learning_rate": 4.291979741546102e-05,
+      "loss": 0.2403,
+      "step": 376
+    },
+    {
+      "epoch": 1.2736486486486487,
+      "grad_norm": 0.4133036434650421,
+      "learning_rate": 4.293902250342989e-05,
+      "loss": 0.2086,
+      "step": 377
+    },
+    {
+      "epoch": 1.277027027027027,
+      "grad_norm": 0.45922860503196716,
+      "learning_rate": 4.295819666395376e-05,
+      "loss": 0.236,
+      "step": 378
+    },
+    {
+      "epoch": 1.2804054054054055,
+      "grad_norm": 0.4423050284385681,
+      "learning_rate": 4.297732016613454e-05,
+      "loss": 0.2271,
+      "step": 379
+    },
+    {
+      "epoch": 1.2837837837837838,
+      "grad_norm": 0.4588952362537384,
+      "learning_rate": 4.299639327694684e-05,
+      "loss": 0.2356,
+      "step": 380
+    },
+    {
+      "epoch": 1.2871621621621623,
+      "grad_norm": 26.979991912841797,
+      "learning_rate": 4.3015416261260325e-05,
+      "loss": 0.2247,
+      "step": 381
+    },
+    {
+      "epoch": 1.2905405405405406,
+      "grad_norm": 0.6023054718971252,
+      "learning_rate": 4.303438938186182e-05,
+      "loss": 0.2477,
+      "step": 382
+    },
+    {
+      "epoch": 1.2939189189189189,
+      "grad_norm": 0.48854556679725647,
+      "learning_rate": 4.305331289947705e-05,
+      "loss": 0.221,
+      "step": 383
+    },
+    {
+      "epoch": 1.2972972972972974,
+      "grad_norm": 0.5875428915023804,
+      "learning_rate": 4.3072187072792184e-05,
+      "loss": 0.2443,
+      "step": 384
+    },
+    {
+      "epoch": 1.3006756756756757,
+      "grad_norm": 0.42068448662757874,
+      "learning_rate": 4.309101215847502e-05,
+      "loss": 0.2043,
+      "step": 385
+    },
+    {
+      "epoch": 1.304054054054054,
+      "grad_norm": 0.492660254240036,
+      "learning_rate": 4.3109788411195924e-05,
+      "loss": 0.246,
+      "step": 386
+    },
+    {
+      "epoch": 1.3074324324324325,
+      "grad_norm": 0.4522532820701599,
+      "learning_rate": 4.312851608364853e-05,
+      "loss": 0.2305,
+      "step": 387
+    },
+    {
+      "epoch": 1.3108108108108107,
+      "grad_norm": 0.4410349130630493,
+      "learning_rate": 4.314719542657013e-05,
+      "loss": 0.2146,
+      "step": 388
+    },
+    {
+      "epoch": 1.3141891891891893,
+      "grad_norm": 38.59849548339844,
+      "learning_rate": 4.3165826688761796e-05,
+      "loss": 1.1445,
+      "step": 389
+    },
+    {
+      "epoch": 1.3175675675675675,
+      "grad_norm": 0.6608863472938538,
+      "learning_rate": 4.318441011710833e-05,
+      "loss": 0.2591,
+      "step": 390
+    },
+    {
+      "epoch": 1.320945945945946,
+      "grad_norm": 0.7261571884155273,
+      "learning_rate": 4.3202945956597786e-05,
+      "loss": 0.2409,
+      "step": 391
+    },
+    {
+      "epoch": 1.3243243243243243,
+      "grad_norm": 13.75738525390625,
+      "learning_rate": 4.3221434450340956e-05,
+      "loss": 1.022,
+      "step": 392
+    },
+    {
+      "epoch": 1.3277027027027026,
+      "grad_norm": 1.4299050569534302,
+      "learning_rate": 4.323987583959045e-05,
+      "loss": 0.2287,
+      "step": 393
+    },
+    {
+      "epoch": 1.3310810810810811,
+      "grad_norm": 0.6660119295120239,
+      "learning_rate": 4.325827036375957e-05,
+      "loss": 0.2148,
+      "step": 394
+    },
+    {
+      "epoch": 1.3344594594594594,
+      "grad_norm": 3.13312029838562,
+      "learning_rate": 4.327661826044101e-05,
+      "loss": 0.6968,
+      "step": 395
+    },
+    {
+      "epoch": 1.3378378378378377,
+      "grad_norm": 29.60978126525879,
+      "learning_rate": 4.329491976542521e-05,
+      "loss": 0.6915,
+      "step": 396
+    },
+    {
+      "epoch": 1.3412162162162162,
+      "grad_norm": 0.8736880421638489,
+      "learning_rate": 4.331317511271859e-05,
+      "loss": 0.2388,
+      "step": 397
+    },
+    {
+      "epoch": 1.3445945945945945,
+      "grad_norm": 0.5873726606369019,
+      "learning_rate": 4.333138453456147e-05,
+      "loss": 0.2655,
+      "step": 398
+    },
+    {
+      "epoch": 1.347972972972973,
+      "grad_norm": 2.68512225151062,
+      "learning_rate": 4.334954826144581e-05,
+      "loss": 0.6129,
+      "step": 399
+    },
+    {
+      "epoch": 1.3513513513513513,
+      "grad_norm": 0.8919135928153992,
+      "learning_rate": 4.336766652213271e-05,
+      "loss": 0.2492,
+      "step": 400
+    },
+    {
+      "epoch": 1.3547297297297298,
+      "grad_norm": 2.113346576690674,
+      "learning_rate": 4.338573954366971e-05,
+      "loss": 0.2578,
+      "step": 401
+    },
+    {
+      "epoch": 1.3581081081081081,
+      "grad_norm": 2.666167736053467,
+      "learning_rate": 4.340376755140784e-05,
+      "loss": 0.6334,
+      "step": 402
+    },
+    {
+      "epoch": 1.3614864864864864,
+      "grad_norm": 0.8454524278640747,
+      "learning_rate": 4.342175076901849e-05,
+      "loss": 0.2426,
+      "step": 403
+    },
+    {
+      "epoch": 1.364864864864865,
+      "grad_norm": 0.7469472885131836,
+      "learning_rate": 4.343968941851009e-05,
+      "loss": 0.2715,
+      "step": 404
+    },
+    {
+      "epoch": 1.3682432432432432,
+      "grad_norm": 0.6965801119804382,
+      "learning_rate": 4.345758372024448e-05,
+      "loss": 0.2462,
+      "step": 405
+    },
+    {
+      "epoch": 1.3716216216216215,
+      "grad_norm": 1.5828707218170166,
+      "learning_rate": 4.347543389295324e-05,
+      "loss": 0.5967,
+      "step": 406
+    },
+    {
+      "epoch": 1.375,
+      "grad_norm": 0.5627582669258118,
+      "learning_rate": 4.3493240153753666e-05,
+      "loss": 0.226,
+      "step": 407
+    },
+    {
+      "epoch": 1.3783783783783785,
+      "grad_norm": 0.5101889967918396,
+      "learning_rate": 4.3511002718164666e-05,
+      "loss": 0.2266,
+      "step": 408
+    },
+    {
+      "epoch": 1.3817567567567568,
+      "grad_norm": 0.6242619156837463,
+      "learning_rate": 4.352872180012237e-05,
+      "loss": 0.2028,
+      "step": 409
+    },
+    {
+      "epoch": 1.385135135135135,
+      "grad_norm": 0.5768154263496399,
+      "learning_rate": 4.35463976119956e-05,
+      "loss": 0.2291,
+      "step": 410
+    },
+    {
+      "epoch": 1.3885135135135136,
+      "grad_norm": 0.5170788764953613,
+      "learning_rate": 4.356403036460115e-05,
+      "loss": 0.2124,
+      "step": 411
+    },
+    {
+      "epoch": 1.3918918918918919,
+      "grad_norm": 0.530846357345581,
+      "learning_rate": 4.3581620267218916e-05,
+      "loss": 0.2013,
+      "step": 412
+    },
+    {
+      "epoch": 1.3952702702702702,
+      "grad_norm": 0.5346778631210327,
+      "learning_rate": 4.359916752760669e-05,
+      "loss": 0.2439,
+      "step": 413
+    },
+    {
+      "epoch": 1.3986486486486487,
+      "grad_norm": 0.5345392227172852,
+      "learning_rate": 4.361667235201499e-05,
+      "loss": 0.2321,
+      "step": 414
+    },
+    {
+      "epoch": 1.402027027027027,
+      "grad_norm": 0.5348331928253174,
+      "learning_rate": 4.363413494520154e-05,
+      "loss": 0.2418,
+      "step": 415
+    },
+    {
+      "epoch": 1.4054054054054055,
+      "grad_norm": 0.9726234078407288,
+      "learning_rate": 4.365155551044572e-05,
+      "loss": 0.213,
+      "step": 416
+    },
+    {
+      "epoch": 1.4087837837837838,
+      "grad_norm": 2.116424083709717,
+      "learning_rate": 4.366893424956263e-05,
+      "loss": 0.5716,
+      "step": 417
+    },
+    {
+      "epoch": 1.4121621621621623,
+      "grad_norm": 0.6004624962806702,
+      "learning_rate": 4.368627136291726e-05,
+      "loss": 0.2404,
+      "step": 418
+    },
+    {
+      "epoch": 1.4155405405405406,
+      "grad_norm": 0.44696101546287537,
+      "learning_rate": 4.370356704943825e-05,
+      "loss": 0.2349,
+      "step": 419
+    },
+    {
+      "epoch": 1.4189189189189189,
+      "grad_norm": 0.4704833924770355,
+      "learning_rate": 4.372082150663168e-05,
+      "loss": 0.2365,
+      "step": 420
+    },
+    {
+      "epoch": 1.4222972972972974,
+      "grad_norm": 0.48295876383781433,
+      "learning_rate": 4.3738034930594475e-05,
+      "loss": 0.2256,
+      "step": 421
+    },
+    {
+      "epoch": 1.4256756756756757,
+      "grad_norm": 0.4726882576942444,
+      "learning_rate": 4.3755207516027904e-05,
+      "loss": 0.2283,
+      "step": 422
+    },
+    {
+      "epoch": 1.429054054054054,
+      "grad_norm": 0.5251411199569702,
+      "learning_rate": 4.377233945625071e-05,
+      "loss": 0.2374,
+      "step": 423
+    },
+    {
+      "epoch": 1.4324324324324325,
+      "grad_norm": 94.52925109863281,
+      "learning_rate": 4.378943094321221e-05,
+      "loss": 0.4887,
+      "step": 424
+    },
+    {
+      "epoch": 1.4358108108108107,
+      "grad_norm": 0.5848643183708191,
+      "learning_rate": 4.3806482167505196e-05,
+      "loss": 0.2373,
+      "step": 425
+    },
+    {
+      "epoch": 1.4391891891891893,
+      "grad_norm": 0.870020866394043,
+      "learning_rate": 4.382349331837866e-05,
+      "loss": 0.5312,
+      "step": 426
+    },
+    {
+      "epoch": 1.4425675675675675,
+      "grad_norm": 0.58022540807724,
+      "learning_rate": 4.3840464583750404e-05,
+      "loss": 0.2347,
+      "step": 427
+    },
+    {
+      "epoch": 1.445945945945946,
+      "grad_norm": 0.4780952036380768,
+      "learning_rate": 4.385739615021954e-05,
+      "loss": 0.2392,
+      "step": 428
+    },
+    {
+      "epoch": 1.4493243243243243,
+      "grad_norm": 0.4691276252269745,
+      "learning_rate": 4.387428820307874e-05,
+      "loss": 0.2346,
+      "step": 429
+    },
+    {
+      "epoch": 1.4527027027027026,
+      "grad_norm": 0.4635365903377533,
+      "learning_rate": 4.3891140926326446e-05,
+      "loss": 0.2242,
+      "step": 430
+    },
+    {
+      "epoch": 1.4560810810810811,
+      "grad_norm": 0.7928866147994995,
+      "learning_rate": 4.390795450267886e-05,
+      "loss": 0.2267,
+      "step": 431
+    },
+    {
+      "epoch": 1.4594594594594594,
+      "grad_norm": 0.5844922661781311,
+      "learning_rate": 4.3924729113581876e-05,
+      "loss": 0.237,
+      "step": 432
+    },
+    {
+      "epoch": 1.4628378378378377,
+      "grad_norm": 0.4539477229118347,
+      "learning_rate": 4.394146493922276e-05,
+      "loss": 0.2624,
+      "step": 433
+    },
+    {
+      "epoch": 1.4662162162162162,
+      "grad_norm": 0.4881402254104614,
+      "learning_rate": 4.395816215854185e-05,
+      "loss": 0.2076,
+      "step": 434
+    },
+    {
+      "epoch": 1.4695945945945945,
+      "grad_norm": 0.42912018299102783,
+      "learning_rate": 4.397482094924396e-05,
+      "loss": 0.224,
+      "step": 435
+    },
+    {
+      "epoch": 1.472972972972973,
+      "grad_norm": 8.578511238098145,
+      "learning_rate": 4.399144148780977e-05,
+      "loss": 0.5361,
+      "step": 436
+    },
+    {
+      "epoch": 1.4763513513513513,
+      "grad_norm": 0.4810888469219208,
+      "learning_rate": 4.400802394950703e-05,
+      "loss": 0.2274,
+      "step": 437
+    },
+    {
+      "epoch": 1.4797297297297298,
+      "grad_norm": 0.5107089281082153,
+      "learning_rate": 4.402456850840166e-05,
+      "loss": 0.2295,
+      "step": 438
+    },
+    {
+      "epoch": 1.4831081081081081,
+      "grad_norm": 0.4564654529094696,
+      "learning_rate": 4.4041075337368695e-05,
+      "loss": 0.2447,
+      "step": 439
+    },
+    {
+      "epoch": 1.4864864864864864,
+      "grad_norm": 1.4533860683441162,
+      "learning_rate": 4.405754460810312e-05,
+      "loss": 0.2522,
+      "step": 440
+    },
+    {
+      "epoch": 1.489864864864865,
+      "grad_norm": 2.172696828842163,
+      "learning_rate": 4.407397649113065e-05,
+      "loss": 0.8104,
+      "step": 441
+    },
+    {
+      "epoch": 1.4932432432432432,
+      "grad_norm": 0.4864687919616699,
+      "learning_rate": 4.40903711558182e-05,
+      "loss": 0.2657,
+      "step": 442
+    },
+    {
+      "epoch": 1.4966216216216215,
+      "grad_norm": 0.4211529791355133,
+      "learning_rate": 4.41067287703845e-05,
+      "loss": 0.246,
+      "step": 443
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.5068204998970032,
+      "learning_rate": 4.412304950191033e-05,
+      "loss": 0.219,
+      "step": 444
+    },
+    {
+      "epoch": 1.5033783783783785,
+      "grad_norm": 0.52825927734375,
+      "learning_rate": 4.413933351634886e-05,
+      "loss": 0.2293,
+      "step": 445
+    },
+    {
+      "epoch": 1.5067567567567568,
+      "grad_norm": 0.45864129066467285,
+      "learning_rate": 4.4155580978535707e-05,
+      "loss": 0.218,
+      "step": 446
+    },
+    {
+      "epoch": 1.510135135135135,
+      "grad_norm": 1.0356566905975342,
+      "learning_rate": 4.417179205219895e-05,
+      "loss": 0.2278,
+      "step": 447
+    },
+    {
+      "epoch": 1.5135135135135136,
+      "grad_norm": 0.4032718241214752,
+      "learning_rate": 4.418796689996907e-05,
+      "loss": 0.2276,
+      "step": 448
+    },
+    {
+      "epoch": 1.5168918918918919,
+      "grad_norm": 0.4162694811820984,
+      "learning_rate": 4.420410568338872e-05,
+      "loss": 0.23,
+      "step": 449
+    },
+    {
+      "epoch": 1.5202702702702702,
+      "grad_norm": 0.4550395607948303,
+      "learning_rate": 4.42202085629224e-05,
+      "loss": 0.2445,
+      "step": 450
+    },
+    {
+      "epoch": 1.5236486486486487,
+      "grad_norm": 0.5222985148429871,
+      "learning_rate": 4.423627569796601e-05,
+      "loss": 0.2405,
+      "step": 451
+    },
+    {
+      "epoch": 1.527027027027027,
+      "grad_norm": 0.4746388792991638,
+      "learning_rate": 4.425230724685638e-05,
+      "loss": 0.2299,
+      "step": 452
+    },
+    {
+      "epoch": 1.5304054054054053,
+      "grad_norm": 0.47175803780555725,
+      "learning_rate": 4.4268303366880536e-05,
+      "loss": 0.234,
+      "step": 453
+    },
+    {
+      "epoch": 1.5337837837837838,
+      "grad_norm": 0.46304166316986084,
+      "learning_rate": 4.428426421428507e-05,
+      "loss": 0.2165,
+      "step": 454
+    },
+    {
+      "epoch": 1.5371621621621623,
+      "grad_norm": 0.8317951560020447,
+      "learning_rate": 4.430018994428521e-05,
+      "loss": 0.2648,
+      "step": 455
+    },
+    {
+      "epoch": 1.5405405405405406,
+      "grad_norm": 0.4764087498188019,
+      "learning_rate": 4.431608071107392e-05,
+      "loss": 0.2526,
+      "step": 456
+    },
+    {
+      "epoch": 1.5439189189189189,
+      "grad_norm": 0.39065808057785034,
+      "learning_rate": 4.433193666783084e-05,
+      "loss": 0.2428,
+      "step": 457
+    },
+    {
+      "epoch": 1.5472972972972974,
+      "grad_norm": 0.40975654125213623,
+      "learning_rate": 4.4347757966731156e-05,
+      "loss": 0.2437,
+      "step": 458
+    },
+    {
+      "epoch": 1.5506756756756757,
+      "grad_norm": 0.4318806231021881,
+      "learning_rate": 4.436354475895436e-05,
+      "loss": 0.233,
+      "step": 459
+    },
+    {
+      "epoch": 1.554054054054054,
+      "grad_norm": 0.4574219882488251,
+      "learning_rate": 4.437929719469291e-05,
+      "loss": 0.2441,
+      "step": 460
+    },
+    {
+      "epoch": 1.5574324324324325,
+      "grad_norm": 0.39155369997024536,
+      "learning_rate": 4.4395015423160807e-05,
+      "loss": 0.2233,
+      "step": 461
+    },
+    {
+      "epoch": 1.560810810810811,
+      "grad_norm": 0.4589376747608185,
+      "learning_rate": 4.4410699592602094e-05,
+      "loss": 0.225,
+      "step": 462
+    },
+    {
+      "epoch": 1.564189189189189,
+      "grad_norm": 0.8952370882034302,
+      "learning_rate": 4.442634985029922e-05,
+      "loss": 0.4926,
+      "step": 463
+    },
+    {
+      "epoch": 1.5675675675675675,
+      "grad_norm": 0.43023157119750977,
+      "learning_rate": 4.444196634258136e-05,
+      "loss": 0.2333,
+      "step": 464
+    },
+    {
+      "epoch": 1.570945945945946,
+      "grad_norm": 1.1993354558944702,
+      "learning_rate": 4.4457549214832566e-05,
+      "loss": 0.2457,
+      "step": 465
+    },
+    {
+      "epoch": 1.5743243243243243,
+      "grad_norm": 0.3993515074253082,
+      "learning_rate": 4.44730986115e-05,
+      "loss": 0.2457,
+      "step": 466
+    },
+    {
+      "epoch": 1.5777027027027026,
+      "grad_norm": 0.4565802812576294,
+      "learning_rate": 4.448861467610187e-05,
+      "loss": 0.2335,
+      "step": 467
+    },
+    {
+      "epoch": 1.5810810810810811,
+      "grad_norm": 0.4085776209831238,
+      "learning_rate": 4.4504097551235406e-05,
+      "loss": 0.237,
+      "step": 468
+    },
+    {
+      "epoch": 1.5844594594594594,
+      "grad_norm": 0.4212399125099182,
+      "learning_rate": 4.4519547378584725e-05,
+      "loss": 0.24,
+      "step": 469
+    },
+    {
+      "epoch": 1.5878378378378377,
+      "grad_norm": 0.408805251121521,
+      "learning_rate": 4.453496429892863e-05,
+      "loss": 0.2252,
+      "step": 470
+    },
+    {
+      "epoch": 1.5912162162162162,
+      "grad_norm": 0.42911630868911743,
+      "learning_rate": 4.455034845214827e-05,
+      "loss": 0.2099,
+      "step": 471
+    },
+    {
+      "epoch": 1.5945945945945947,
+      "grad_norm": 0.47564277052879333,
+      "learning_rate": 4.4565699977234796e-05,
+      "loss": 0.2135,
+      "step": 472
+    },
+    {
+      "epoch": 1.597972972972973,
+      "grad_norm": 0.3962784707546234,
+      "learning_rate": 4.458101901229686e-05,
+      "loss": 0.2346,
+      "step": 473
+    },
+    {
+      "epoch": 1.6013513513513513,
+      "grad_norm": 0.3911672532558441,
+      "learning_rate": 4.459630569456809e-05,
+      "loss": 0.2198,
+      "step": 474
+    },
+    {
+      "epoch": 1.6047297297297298,
+      "grad_norm": 1093.1121826171875,
+      "learning_rate": 4.461156016041444e-05,
+      "loss": 0.2989,
+      "step": 475
+    },
+    {
+      "epoch": 1.6081081081081081,
+      "grad_norm": 3.732886552810669,
+      "learning_rate": 4.462678254534156e-05,
+      "loss": 0.5155,
+      "step": 476
+    },
+    {
+      "epoch": 1.6114864864864864,
+      "grad_norm": 0.5420750975608826,
+      "learning_rate": 4.464197298400191e-05,
+      "loss": 0.2645,
+      "step": 477
+    },
+    {
+      "epoch": 1.614864864864865,
+      "grad_norm": 4.157422065734863,
+      "learning_rate": 4.4657131610201994e-05,
+      "loss": 0.5192,
+      "step": 478
+    },
+    {
+      "epoch": 1.6182432432432432,
+      "grad_norm": 0.4917280972003937,
+      "learning_rate": 4.467225855690939e-05,
+      "loss": 0.2187,
+      "step": 479
+    },
+    {
+      "epoch": 1.6216216216216215,
+      "grad_norm": 0.4480583071708679,
+      "learning_rate": 4.468735395625979e-05,
+      "loss": 0.2214,
+      "step": 480
+    },
+    {
+      "epoch": 1.625,
+      "grad_norm": 0.4467374384403229,
+      "learning_rate": 4.470241793956387e-05,
+      "loss": 0.2414,
+      "step": 481
+    },
+    {
+      "epoch": 1.6283783783783785,
+      "grad_norm": 0.48159006237983704,
+      "learning_rate": 4.471745063731416e-05,
+      "loss": 0.2386,
+      "step": 482
+    },
+    {
+      "epoch": 1.6317567567567568,
+      "grad_norm": 0.3858594000339508,
+      "learning_rate": 4.473245217919187e-05,
+      "loss": 0.1981,
+      "step": 483
+    },
+    {
+      "epoch": 1.635135135135135,
+      "grad_norm": 0.4672364592552185,
+      "learning_rate": 4.474742269407355e-05,
+      "loss": 0.2212,
+      "step": 484
+    },
+    {
+      "epoch": 1.6385135135135136,
+      "grad_norm": 0.5253187417984009,
+      "learning_rate": 4.476236231003773e-05,
+      "loss": 0.2753,
+      "step": 485
+    },
+    {
+      "epoch": 1.6418918918918919,
+      "grad_norm": 0.488766610622406,
+      "learning_rate": 4.477727115437156e-05,
+      "loss": 0.2571,
+      "step": 486
+    },
+    {
+      "epoch": 1.6452702702702702,
+      "grad_norm": 0.4526199996471405,
+      "learning_rate": 4.479214935357724e-05,
+      "loss": 0.2341,
+      "step": 487
+    },
+    {
+      "epoch": 1.6486486486486487,
+      "grad_norm": 0.4283704459667206,
+      "learning_rate": 4.480699703337852e-05,
+      "loss": 0.2533,
+      "step": 488
+    },
+    {
+      "epoch": 1.652027027027027,
+      "grad_norm": 0.45583993196487427,
+      "learning_rate": 4.4821814318727016e-05,
+      "loss": 0.2394,
+      "step": 489
+    },
+    {
+      "epoch": 1.6554054054054053,
+      "grad_norm": 0.3872677683830261,
+      "learning_rate": 4.483660133380856e-05,
+      "loss": 0.2005,
+      "step": 490
+    },
+    {
+      "epoch": 1.6587837837837838,
+      "grad_norm": 0.5054983496665955,
+      "learning_rate": 4.485135820204948e-05,
+      "loss": 0.231,
+      "step": 491
+    },
+    {
+      "epoch": 1.6621621621621623,
+      "grad_norm": 0.40404146909713745,
+      "learning_rate": 4.486608504612267e-05,
+      "loss": 0.2529,
+      "step": 492
+    },
+    {
+      "epoch": 1.6655405405405406,
+      "grad_norm": 0.4361649453639984,
+      "learning_rate": 4.488078198795383e-05,
+      "loss": 0.2485,
+      "step": 493
+    },
+    {
+      "epoch": 1.6689189189189189,
+      "grad_norm": 0.4425160586833954,
+      "learning_rate": 4.489544914872745e-05,
+      "loss": 0.2375,
+      "step": 494
+    },
+    {
+      "epoch": 1.6722972972972974,
+      "grad_norm": 0.4194391965866089,
+      "learning_rate": 4.4910086648892815e-05,
+      "loss": 0.2461,
+      "step": 495
+    },
+    {
+      "epoch": 1.6756756756756757,
+      "grad_norm": 0.42155027389526367,
+      "learning_rate": 4.4924694608169965e-05,
+      "loss": 0.2301,
+      "step": 496
+    },
+    {
+      "epoch": 1.679054054054054,
+      "grad_norm": 0.412643164396286,
+      "learning_rate": 4.4939273145555536e-05,
+      "loss": 0.235,
+      "step": 497
+    },
+    {
+      "epoch": 1.6824324324324325,
+      "grad_norm": 0.3827148973941803,
+      "learning_rate": 4.495382237932863e-05,
+      "loss": 0.23,
+      "step": 498
+    },
+    {
+      "epoch": 1.685810810810811,
+      "grad_norm": 0.4173565208911896,
+      "learning_rate": 4.4968342427056505e-05,
+      "loss": 0.2374,
+      "step": 499
+    },
+    {
+      "epoch": 1.689189189189189,
+      "grad_norm": 1055.430908203125,
+      "learning_rate": 4.498283340560031e-05,
+      "loss": 0.3647,
+      "step": 500
+    },
+    {
+      "epoch": 1.6925675675675675,
+      "grad_norm": 0.5496036410331726,
+      "learning_rate": 4.499729543112076e-05,
+      "loss": 0.2686,
+      "step": 501
+    },
+    {
+      "epoch": 1.695945945945946,
+      "grad_norm": 0.437640905380249,
+      "learning_rate": 4.501172861908366e-05,
+      "loss": 0.2521,
+      "step": 502
+    },
+    {
+      "epoch": 1.6993243243243243,
+      "grad_norm": 0.4216068685054779,
+      "learning_rate": 4.502613308426546e-05,
+      "loss": 0.2447,
+      "step": 503
+    },
+    {
+      "epoch": 1.7027027027027026,
+      "grad_norm": 0.4138765037059784,
+      "learning_rate": 4.504050894075876e-05,
+      "loss": 0.2379,
+      "step": 504
+    },
+    {
+      "epoch": 1.7060810810810811,
+      "grad_norm": 0.8677256107330322,
+      "learning_rate": 4.5054856301977696e-05,
+      "loss": 0.2399,
+      "step": 505
+    },
+    {
+      "epoch": 1.7094594594594594,
+      "grad_norm": 0.4616876542568207,
+      "learning_rate": 4.506917528066332e-05,
+      "loss": 0.229,
+      "step": 506
+    },
+    {
+      "epoch": 1.7128378378378377,
+      "grad_norm": 0.6452389359474182,
+      "learning_rate": 4.508346598888894e-05,
+      "loss": 0.2237,
+      "step": 507
+    },
+    {
+      "epoch": 1.7162162162162162,
+      "grad_norm": 0.4205520749092102,
+      "learning_rate": 4.509772853806532e-05,
+      "loss": 0.2175,
+      "step": 508
+    },
+    {
+      "epoch": 1.7195945945945947,
+      "grad_norm": 0.456882506608963,
+      "learning_rate": 4.511196303894598e-05,
+      "loss": 0.2444,
+      "step": 509
+    },
+    {
+      "epoch": 1.722972972972973,
+      "grad_norm": 0.43970227241516113,
+      "learning_rate": 4.512616960163227e-05,
+      "loss": 0.2357,
+      "step": 510
+    },
+    {
+      "epoch": 1.7263513513513513,
+      "grad_norm": 0.4025394320487976,
+      "learning_rate": 4.5140348335578547e-05,
+      "loss": 0.2375,
+      "step": 511
+    },
+    {
+      "epoch": 1.7297297297297298,
+      "grad_norm": 0.39442214369773865,
+      "learning_rate": 4.515449934959718e-05,
+      "loss": 0.2444,
+      "step": 512
+    },
+    {
+      "epoch": 1.7331081081081081,
+      "grad_norm": 0.44028565287590027,
+      "learning_rate": 4.516862275186361e-05,
+      "loss": 0.2324,
+      "step": 513
+    },
+    {
+      "epoch": 1.7364864864864864,
+      "grad_norm": 0.4466850161552429,
+      "learning_rate": 4.518271864992127e-05,
+      "loss": 0.2359,
+      "step": 514
+    },
+    {
+      "epoch": 1.739864864864865,
+      "grad_norm": 0.4365015923976898,
+      "learning_rate": 4.519678715068652e-05,
+      "loss": 0.2708,
+      "step": 515
+    },
+    {
+      "epoch": 1.7432432432432432,
+      "grad_norm": 0.4398203194141388,
+      "learning_rate": 4.521082836045353e-05,
+      "loss": 0.2452,
+      "step": 516
+    },
+    {
+      "epoch": 1.7466216216216215,
+      "grad_norm": 0.44056662917137146,
+      "learning_rate": 4.5224842384899045e-05,
+      "loss": 0.2344,
+      "step": 517
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.43141037225723267,
+      "learning_rate": 4.523882932908722e-05,
+      "loss": 0.2511,
+      "step": 518
+    },
+    {
+      "epoch": 1.7533783783783785,
+      "grad_norm": 0.3922167122364044,
+      "learning_rate": 4.52527892974743e-05,
+      "loss": 0.2162,
+      "step": 519
+    },
+    {
+      "epoch": 1.7567567567567568,
+      "grad_norm": 0.40995466709136963,
+      "learning_rate": 4.526672239391333e-05,
+      "loss": 0.2281,
+      "step": 520
+    },
+    {
+      "epoch": 1.760135135135135,
+      "grad_norm": 0.4008403718471527,
+      "learning_rate": 4.528062872165875e-05,
+      "loss": 0.2257,
+      "step": 521
+    },
+    {
+      "epoch": 1.7635135135135136,
+      "grad_norm": 0.5048766136169434,
+      "learning_rate": 4.529450838337104e-05,
+      "loss": 0.2304,
+      "step": 522
+    },
+    {
+      "epoch": 1.7668918918918919,
+      "grad_norm": 0.40119826793670654,
+      "learning_rate": 4.530836148112124e-05,
+      "loss": 0.2341,
+      "step": 523
+    },
+    {
+      "epoch": 1.7702702702702702,
+      "grad_norm": 0.41344285011291504,
+      "learning_rate": 4.532218811639545e-05,
+      "loss": 0.272,
+      "step": 524
+    },
+    {
+      "epoch": 1.7736486486486487,
+      "grad_norm": 0.38902515172958374,
+      "learning_rate": 4.5335988390099284e-05,
+      "loss": 0.2195,
+      "step": 525
+    },
+    {
+      "epoch": 1.777027027027027,
+      "grad_norm": 0.5754374861717224,
+      "learning_rate": 4.534976240256232e-05,
+      "loss": 0.2335,
+      "step": 526
+    },
+    {
+      "epoch": 1.7804054054054053,
+      "grad_norm": 31.7464656829834,
+      "learning_rate": 4.536351025354245e-05,
+      "loss": 0.4532,
+      "step": 527
+    },
+    {
+      "epoch": 1.7837837837837838,
+      "grad_norm": 0.46500369906425476,
+      "learning_rate": 4.537723204223021e-05,
+      "loss": 0.2541,
+      "step": 528
+    },
+    {
+      "epoch": 1.7871621621621623,
+      "grad_norm": 0.41807517409324646,
+      "learning_rate": 4.53909278672531e-05,
+      "loss": 0.2287,
+      "step": 529
+    },
+    {
+      "epoch": 1.7905405405405406,
+      "grad_norm": 0.40613386034965515,
+      "learning_rate": 4.5404597826679824e-05,
+      "loss": 0.2196,
+      "step": 530
+    },
+    {
+      "epoch": 1.7939189189189189,
+      "grad_norm": 0.4200097322463989,
+      "learning_rate": 4.541824201802449e-05,
+      "loss": 0.2393,
+      "step": 531
+    },
+    {
+      "epoch": 1.7972972972972974,
+      "grad_norm": 0.48739388585090637,
+      "learning_rate": 4.543186053825081e-05,
+      "loss": 0.2469,
+      "step": 532
+    },
+    {
+      "epoch": 1.8006756756756757,
+      "grad_norm": 0.5199280381202698,
+      "learning_rate": 4.544545348377621e-05,
+      "loss": 0.2652,
+      "step": 533
+    },
+    {
+      "epoch": 1.804054054054054,
+      "grad_norm": 0.44481709599494934,
+      "learning_rate": 4.5459020950475946e-05,
+      "loss": 0.2386,
+      "step": 534
+    },
+    {
+      "epoch": 1.8074324324324325,
+      "grad_norm": 1.6978758573532104,
+      "learning_rate": 4.5472563033687145e-05,
+      "loss": 0.5287,
+      "step": 535
+    },
+    {
+      "epoch": 1.810810810810811,
+      "grad_norm": 0.3990303575992584,
+      "learning_rate": 4.548607982821284e-05,
+      "loss": 0.2102,
+      "step": 536
+    },
+    {
+      "epoch": 1.814189189189189,
+      "grad_norm": 0.4208585321903229,
+      "learning_rate": 4.5499571428325935e-05,
+      "loss": 0.2058,
+      "step": 537
+    },
+    {
+      "epoch": 1.8175675675675675,
+      "grad_norm": 0.41925048828125,
+      "learning_rate": 4.5513037927773155e-05,
+      "loss": 0.2349,
+      "step": 538
+    },
+    {
+      "epoch": 1.820945945945946,
+      "grad_norm": 0.43288394808769226,
+      "learning_rate": 4.5526479419778986e-05,
+      "loss": 0.2519,
+      "step": 539
+    },
+    {
+      "epoch": 1.8243243243243243,
+      "grad_norm": 0.4215989112854004,
+      "learning_rate": 4.553989599704948e-05,
+      "loss": 0.2518,
+      "step": 540
+    },
+    {
+      "epoch": 1.8277027027027026,
+      "grad_norm": 0.4134393334388733,
+      "learning_rate": 4.555328775177616e-05,
+      "loss": 0.2341,
+      "step": 541
+    },
+    {
+      "epoch": 1.8310810810810811,
+      "grad_norm": 0.4243868887424469,
+      "learning_rate": 4.5566654775639785e-05,
+      "loss": 0.2553,
+      "step": 542
+    },
+    {
+      "epoch": 1.8344594594594594,
+      "grad_norm": 0.40011635422706604,
+      "learning_rate": 4.5579997159814117e-05,
+      "loss": 0.2447,
+      "step": 543
+    },
+    {
+      "epoch": 1.8378378378378377,
+      "grad_norm": 0.38058069348335266,
+      "learning_rate": 4.5593314994969665e-05,
+      "loss": 0.2307,
+      "step": 544
+    },
+    {
+      "epoch": 1.8412162162162162,
+      "grad_norm": 0.47068503499031067,
+      "learning_rate": 4.560660837127738e-05,
+      "loss": 0.1971,
+      "step": 545
+    },
+    {
+      "epoch": 1.8445945945945947,
+      "grad_norm": 0.4493515193462372,
+      "learning_rate": 4.561987737841229e-05,
+      "loss": 0.2552,
+      "step": 546
+    },
+    {
+      "epoch": 1.847972972972973,
+      "grad_norm": 0.43196865916252136,
+      "learning_rate": 4.563312210555719e-05,
+      "loss": 0.2373,
+      "step": 547
+    },
+    {
+      "epoch": 1.8513513513513513,
+      "grad_norm": 0.4201470911502838,
+      "learning_rate": 4.564634264140616e-05,
+      "loss": 0.2304,
+      "step": 548
+    },
+    {
+      "epoch": 1.8547297297297298,
+      "grad_norm": 0.3680441975593567,
+      "learning_rate": 4.56595390741682e-05,
+      "loss": 0.2263,
+      "step": 549
+    },
+    {
+      "epoch": 1.8581081081081081,
+      "grad_norm": 0.3840002119541168,
+      "learning_rate": 4.567271149157073e-05,
+      "loss": 0.218,
+      "step": 550
+    },
+    {
+      "epoch": 1.8614864864864864,
+      "grad_norm": 0.39841052889823914,
+      "learning_rate": 4.5685859980863086e-05,
+      "loss": 0.2497,
+      "step": 551
+    },
+    {
+      "epoch": 1.864864864864865,
+      "grad_norm": 0.3956766128540039,
+      "learning_rate": 4.569898462881999e-05,
+      "loss": 0.2514,
+      "step": 552
+    },
+    {
+      "epoch": 1.8682432432432432,
+      "grad_norm": 0.4630524218082428,
+      "learning_rate": 4.571208552174497e-05,
+      "loss": 0.2744,
+      "step": 553
+    },
+    {
+      "epoch": 1.8716216216216215,
+      "grad_norm": 0.4056280851364136,
+      "learning_rate": 4.572516274547383e-05,
+      "loss": 0.2712,
+      "step": 554
+    },
+    {
+      "epoch": 1.875,
+      "grad_norm": 0.4127073585987091,
+      "learning_rate": 4.573821638537794e-05,
+      "loss": 0.2319,
+      "step": 555
+    },
+    {
+      "epoch": 1.8783783783783785,
+      "grad_norm": 0.38253894448280334,
+      "learning_rate": 4.575124652636763e-05,
+      "loss": 0.2302,
+      "step": 556
+    },
+    {
+      "epoch": 1.8817567567567568,
+      "grad_norm": 0.405472993850708,
+      "learning_rate": 4.5764253252895486e-05,
+      "loss": 0.2413,
+      "step": 557
+    },
+    {
+      "epoch": 1.885135135135135,
+      "grad_norm": 0.41931191086769104,
+      "learning_rate": 4.577723664895965e-05,
+      "loss": 0.2477,
+      "step": 558
+    },
+    {
+      "epoch": 1.8885135135135136,
+      "grad_norm": 0.3967287540435791,
+      "learning_rate": 4.579019679810706e-05,
+      "loss": 0.2525,
+      "step": 559
+    },
+    {
+      "epoch": 1.8918918918918919,
+      "grad_norm": 0.3867218792438507,
+      "learning_rate": 4.5803133783436676e-05,
+      "loss": 0.2226,
+      "step": 560
+    },
+    {
+      "epoch": 1.8952702702702702,
+      "grad_norm": 0.4048525094985962,
+      "learning_rate": 4.581604768760269e-05,
+      "loss": 0.2434,
+      "step": 561
+    },
+    {
+      "epoch": 1.8986486486486487,
+      "grad_norm": 0.38650572299957275,
+      "learning_rate": 4.582893859281769e-05,
+      "loss": 0.2287,
+      "step": 562
+    },
+    {
+      "epoch": 1.902027027027027,
+      "grad_norm": 0.3860650062561035,
+      "learning_rate": 4.584180658085578e-05,
+      "loss": 0.2285,
+      "step": 563
+    },
+    {
+      "epoch": 1.9054054054054053,
+      "grad_norm": 0.3902740478515625,
+      "learning_rate": 4.585465173305571e-05,
+      "loss": 0.2174,
+      "step": 564
+    },
+    {
+      "epoch": 1.9087837837837838,
+      "grad_norm": 0.4053342342376709,
+      "learning_rate": 4.5867474130323984e-05,
+      "loss": 0.2482,
+      "step": 565
+    },
+    {
+      "epoch": 1.9121621621621623,
+      "grad_norm": 0.35971924662590027,
+      "learning_rate": 4.588027385313786e-05,
+      "loss": 0.1858,
+      "step": 566
+    },
+    {
+      "epoch": 1.9155405405405406,
+      "grad_norm": 0.4442739486694336,
+      "learning_rate": 4.5893050981548446e-05,
+      "loss": 0.2847,
+      "step": 567
+    },
+    {
+      "epoch": 1.9189189189189189,
+      "grad_norm": 0.4430129826068878,
+      "learning_rate": 4.5905805595183656e-05,
+      "loss": 0.2265,
+      "step": 568
+    },
+    {
+      "epoch": 1.9222972972972974,
+      "grad_norm": 0.39476659893989563,
+      "learning_rate": 4.591853777325119e-05,
+      "loss": 0.258,
+      "step": 569
+    },
+    {
+      "epoch": 1.9256756756756757,
+      "grad_norm": 0.4573532044887543,
+      "learning_rate": 4.593124759454153e-05,
+      "loss": 0.2398,
+      "step": 570
+    },
+    {
+      "epoch": 1.929054054054054,
+      "grad_norm": 0.3871706426143646,
+      "learning_rate": 4.5943935137430806e-05,
+      "loss": 0.2292,
+      "step": 571
+    },
+    {
+      "epoch": 1.9324324324324325,
+      "grad_norm": 0.38287678360939026,
+      "learning_rate": 4.595660047988374e-05,
+      "loss": 0.2313,
+      "step": 572
+    },
+    {
+      "epoch": 1.935810810810811,
+      "grad_norm": 0.41778552532196045,
+      "learning_rate": 4.59692436994565e-05,
+      "loss": 0.2501,
+      "step": 573
+    },
+    {
+      "epoch": 1.939189189189189,
+      "grad_norm": 0.3646586537361145,
+      "learning_rate": 4.5981864873299563e-05,
+      "loss": 0.2099,
+      "step": 574
+    },
+    {
+      "epoch": 1.9425675675675675,
+      "grad_norm": 0.3954225778579712,
+      "learning_rate": 4.599446407816052e-05,
+      "loss": 0.2272,
+      "step": 575
+    },
+    {
+      "epoch": 1.945945945945946,
+      "grad_norm": 0.39867308735847473,
+      "learning_rate": 4.6007041390386874e-05,
+      "loss": 0.2352,
+      "step": 576
+    },
+    {
+      "epoch": 1.9493243243243243,
+      "grad_norm": 0.3616117238998413,
+      "learning_rate": 4.601959688592886e-05,
+      "loss": 0.2207,
+      "step": 577
+    },
+    {
+      "epoch": 1.9527027027027026,
+      "grad_norm": 0.42957910895347595,
+      "learning_rate": 4.603213064034216e-05,
+      "loss": 0.2437,
+      "step": 578
+    },
+    {
+      "epoch": 1.9560810810810811,
+      "grad_norm": 0.3899705708026886,
+      "learning_rate": 4.604464272879061e-05,
+      "loss": 0.2436,
+      "step": 579
+    },
+    {
+      "epoch": 1.9594594594594594,
+      "grad_norm": 0.3997238278388977,
+      "learning_rate": 4.605713322604896e-05,
+      "loss": 0.2065,
+      "step": 580
+    },
+    {
+      "epoch": 1.9628378378378377,
+      "grad_norm": 0.4001472592353821,
+      "learning_rate": 4.606960220650551e-05,
+      "loss": 0.2435,
+      "step": 581
+    },
+    {
+      "epoch": 1.9662162162162162,
+      "grad_norm": 0.4067547023296356,
+      "learning_rate": 4.608204974416481e-05,
+      "loss": 0.2456,
+      "step": 582
+    },
+    {
+      "epoch": 1.9695945945945947,
+      "grad_norm": 0.3919241428375244,
+      "learning_rate": 4.6094475912650234e-05,
+      "loss": 0.2214,
+      "step": 583
+    },
+    {
+      "epoch": 1.972972972972973,
+      "grad_norm": 0.3752463757991791,
+      "learning_rate": 4.610688078520666e-05,
+      "loss": 0.2148,
+      "step": 584
+    },
+    {
+      "epoch": 1.9763513513513513,
+      "grad_norm": 0.42536041140556335,
+      "learning_rate": 4.611926443470301e-05,
+      "loss": 0.2219,
+      "step": 585
+    },
+    {
+      "epoch": 1.9797297297297298,
+      "grad_norm": 0.3676905930042267,
+      "learning_rate": 4.6131626933634844e-05,
+      "loss": 0.2153,
+      "step": 586
+    },
+    {
+      "epoch": 1.9831081081081081,
+      "grad_norm": 0.3774767816066742,
+      "learning_rate": 4.6143968354126914e-05,
+      "loss": 0.2459,
+      "step": 587
+    },
+    {
+      "epoch": 1.9864864864864864,
+      "grad_norm": 0.38765907287597656,
+      "learning_rate": 4.6156288767935646e-05,
+      "loss": 0.2339,
+      "step": 588
+    },
+    {
+      "epoch": 1.989864864864865,
+      "grad_norm": 0.3474208116531372,
+      "learning_rate": 4.61685882464517e-05,
+      "loss": 0.2109,
+      "step": 589
+    },
+    {
+      "epoch": 1.9932432432432432,
+      "grad_norm": 0.39578577876091003,
+      "learning_rate": 4.61808668607024e-05,
+      "loss": 0.2232,
+      "step": 590
+    },
+    {
+      "epoch": 1.9966216216216215,
+      "grad_norm": 0.3756243884563446,
+      "learning_rate": 4.619312468135426e-05,
+      "loss": 0.2286,
+      "step": 591
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.40366870164871216,
+      "learning_rate": 4.620536177871533e-05,
+      "loss": 0.1979,
+      "step": 592
+    },
+    {
+      "epoch": 2.0033783783783785,
+      "grad_norm": 0.5792023539543152,
+      "learning_rate": 4.621757822273772e-05,
+      "loss": 0.1664,
+      "step": 593
+    },
+    {
+      "epoch": 2.0067567567567566,
+      "grad_norm": 0.48093676567077637,
+      "learning_rate": 4.62297740830199e-05,
+      "loss": 0.1489,
+      "step": 594
+    },
+    {
+      "epoch": 2.010135135135135,
+      "grad_norm": 0.7113845348358154,
+      "learning_rate": 4.6241949428809165e-05,
+      "loss": 0.1666,
+      "step": 595
+    },
+    {
+      "epoch": 2.0135135135135136,
+      "grad_norm": 0.4888218343257904,
+      "learning_rate": 4.625410432900395e-05,
+      "loss": 0.1667,
+      "step": 596
+    },
+    {
+      "epoch": 2.016891891891892,
+      "grad_norm": 0.43272024393081665,
+      "learning_rate": 4.626623885215616e-05,
+      "loss": 0.1611,
+      "step": 597
+    },
+    {
+      "epoch": 2.02027027027027,
+      "grad_norm": 0.4639625549316406,
+      "learning_rate": 4.627835306647352e-05,
+      "loss": 0.183,
+      "step": 598
+    },
+    {
+      "epoch": 2.0236486486486487,
+      "grad_norm": 0.4807380735874176,
+      "learning_rate": 4.629044703982186e-05,
+      "loss": 0.1592,
+      "step": 599
+    },
+    {
+      "epoch": 2.027027027027027,
+      "grad_norm": 0.4514772891998291,
+      "learning_rate": 4.63025208397274e-05,
+      "loss": 0.1459,
+      "step": 600
+    },
+    {
+      "epoch": 2.0304054054054053,
+      "grad_norm": 0.4227406978607178,
+      "learning_rate": 4.6314574533379e-05,
+      "loss": 0.1632,
+      "step": 601
+    },
+    {
+      "epoch": 2.0337837837837838,
+      "grad_norm": 0.42751410603523254,
+      "learning_rate": 4.632660818763041e-05,
+      "loss": 0.1699,
+      "step": 602
+    },
+    {
+      "epoch": 2.0371621621621623,
+      "grad_norm": 0.45877569913864136,
+      "learning_rate": 4.633862186900253e-05,
+      "loss": 0.1725,
+      "step": 603
+    },
+    {
+      "epoch": 2.0405405405405403,
+      "grad_norm": 0.41468122601509094,
+      "learning_rate": 4.6350615643685535e-05,
+      "loss": 0.172,
+      "step": 604
+    },
+    {
+      "epoch": 2.043918918918919,
+      "grad_norm": 0.42624783515930176,
+      "learning_rate": 4.6362589577541154e-05,
+      "loss": 0.1591,
+      "step": 605
+    },
+    {
+      "epoch": 2.0472972972972974,
+      "grad_norm": 0.3854513466358185,
+      "learning_rate": 4.637454373610477e-05,
+      "loss": 0.1486,
+      "step": 606
+    },
+    {
+      "epoch": 2.050675675675676,
+      "grad_norm": 0.40797704458236694,
+      "learning_rate": 4.638647818458763e-05,
+      "loss": 0.1521,
+      "step": 607
+    },
+    {
+      "epoch": 2.054054054054054,
+      "grad_norm": 0.4326710104942322,
+      "learning_rate": 4.639839298787892e-05,
+      "loss": 0.1441,
+      "step": 608
+    },
+    {
+      "epoch": 2.0574324324324325,
+      "grad_norm": 0.39950937032699585,
+      "learning_rate": 4.641028821054793e-05,
+      "loss": 0.165,
+      "step": 609
+    },
+    {
+      "epoch": 2.060810810810811,
+      "grad_norm": 0.38399842381477356,
+      "learning_rate": 4.6422163916846124e-05,
+      "loss": 0.1464,
+      "step": 610
+    },
+    {
+      "epoch": 2.064189189189189,
+      "grad_norm": 0.40971097350120544,
+      "learning_rate": 4.643402017070924e-05,
+      "loss": 0.1604,
+      "step": 611
+    },
+    {
+      "epoch": 2.0675675675675675,
+      "grad_norm": 0.4519285261631012,
+      "learning_rate": 4.644585703575936e-05,
+      "loss": 0.1739,
+      "step": 612
+    },
+    {
+      "epoch": 2.070945945945946,
+      "grad_norm": 0.40968844294548035,
+      "learning_rate": 4.645767457530692e-05,
+      "loss": 0.1592,
+      "step": 613
+    },
+    {
+      "epoch": 2.074324324324324,
+      "grad_norm": 0.4271821975708008,
+      "learning_rate": 4.64694728523528e-05,
+      "loss": 0.1591,
+      "step": 614
+    },
+    {
+      "epoch": 2.0777027027027026,
+      "grad_norm": 0.44706177711486816,
+      "learning_rate": 4.648125192959028e-05,
+      "loss": 0.1745,
+      "step": 615
+    },
+    {
+      "epoch": 2.081081081081081,
+      "grad_norm": 0.3895583152770996,
+      "learning_rate": 4.649301186940709e-05,
+      "loss": 0.1541,
+      "step": 616
+    },
+    {
+      "epoch": 2.0844594594594597,
+      "grad_norm": 0.5112977623939514,
+      "learning_rate": 4.650475273388737e-05,
+      "loss": 0.1851,
+      "step": 617
+    },
+    {
+      "epoch": 2.0878378378378377,
+      "grad_norm": 0.4201546609401703,
+      "learning_rate": 4.651647458481359e-05,
+      "loss": 0.1502,
+      "step": 618
+    },
+    {
+      "epoch": 2.0912162162162162,
+      "grad_norm": 0.4547381103038788,
+      "learning_rate": 4.652817748366864e-05,
+      "loss": 0.1533,
+      "step": 619
+    },
+    {
+      "epoch": 2.0945945945945947,
+      "grad_norm": 0.48063594102859497,
+      "learning_rate": 4.653986149163757e-05,
+      "loss": 0.1704,
+      "step": 620
+    },
+    {
+      "epoch": 2.097972972972973,
+      "grad_norm": 0.3941943943500519,
+      "learning_rate": 4.655152666960967e-05,
+      "loss": 0.1338,
+      "step": 621
+    },
+    {
+      "epoch": 2.1013513513513513,
+      "grad_norm": 0.43712350726127625,
+      "learning_rate": 4.6563173078180315e-05,
+      "loss": 0.1573,
+      "step": 622
+    },
+    {
+      "epoch": 2.10472972972973,
+      "grad_norm": 0.43442657589912415,
+      "learning_rate": 4.657480077765283e-05,
+      "loss": 0.1672,
+      "step": 623
+    },
+    {
+      "epoch": 2.108108108108108,
+      "grad_norm": 0.45270854234695435,
+      "learning_rate": 4.6586409828040405e-05,
+      "loss": 0.1851,
+      "step": 624
+    },
+    {
+      "epoch": 2.1114864864864864,
+      "grad_norm": 0.5774021744728088,
+      "learning_rate": 4.659800028906792e-05,
+      "loss": 0.1451,
+      "step": 625
+    },
+    {
+      "epoch": 2.114864864864865,
+      "grad_norm": 0.4964922368526459,
+      "learning_rate": 4.660957222017383e-05,
+      "loss": 0.1656,
+      "step": 626
+    },
+    {
+      "epoch": 2.1182432432432434,
+      "grad_norm": 0.417864054441452,
+      "learning_rate": 4.662112568051194e-05,
+      "loss": 0.1672,
+      "step": 627
+    },
+    {
+      "epoch": 2.1216216216216215,
+      "grad_norm": 0.5328717231750488,
+      "learning_rate": 4.663266072895327e-05,
+      "loss": 0.1761,
+      "step": 628
+    },
+    {
+      "epoch": 2.125,
+      "grad_norm": 0.4234524369239807,
+      "learning_rate": 4.664417742408782e-05,
+      "loss": 0.1679,
+      "step": 629
+    },
+    {
+      "epoch": 2.1283783783783785,
+      "grad_norm": 0.5051260590553284,
+      "learning_rate": 4.665567582422637e-05,
+      "loss": 0.1763,
+      "step": 630
+    },
+    {
+      "epoch": 2.1317567567567566,
+      "grad_norm": 0.44952192902565,
+      "learning_rate": 4.666715598740224e-05,
+      "loss": 0.1535,
+      "step": 631
+    },
+    {
+      "epoch": 2.135135135135135,
+      "grad_norm": 0.44377925992012024,
+      "learning_rate": 4.667861797137309e-05,
+      "loss": 0.1456,
+      "step": 632
+    },
+    {
+      "epoch": 2.1385135135135136,
+      "grad_norm": 0.3786265254020691,
+      "learning_rate": 4.669006183362258e-05,
+      "loss": 0.1478,
+      "step": 633
+    },
+    {
+      "epoch": 2.141891891891892,
+      "grad_norm": 0.4277481436729431,
+      "learning_rate": 4.670148763136221e-05,
+      "loss": 0.157,
+      "step": 634
+    },
+    {
+      "epoch": 2.14527027027027,
+      "grad_norm": 0.43035444617271423,
+      "learning_rate": 4.671289542153293e-05,
+      "loss": 0.1552,
+      "step": 635
+    },
+    {
+      "epoch": 2.1486486486486487,
+      "grad_norm": 0.45837166905403137,
+      "learning_rate": 4.672428526080691e-05,
+      "loss": 0.1848,
+      "step": 636
+    },
+    {
+      "epoch": 2.152027027027027,
+      "grad_norm": 0.41413986682891846,
+      "learning_rate": 4.673565720558918e-05,
+      "loss": 0.1803,
+      "step": 637
+    },
+    {
+      "epoch": 2.1554054054054053,
+      "grad_norm": 0.4246644377708435,
+      "learning_rate": 4.6747011312019374e-05,
+      "loss": 0.177,
+      "step": 638
+    },
+    {
+      "epoch": 2.1587837837837838,
+      "grad_norm": 0.4299694001674652,
+      "learning_rate": 4.6758347635973334e-05,
+      "loss": 0.164,
+      "step": 639
+    },
+    {
+      "epoch": 2.1621621621621623,
+      "grad_norm": 0.4426731765270233,
+      "learning_rate": 4.676966623306479e-05,
+      "loss": 0.164,
+      "step": 640
+    },
+    {
+      "epoch": 2.1655405405405403,
+      "grad_norm": 0.4473170340061188,
+      "learning_rate": 4.678096715864696e-05,
+      "loss": 0.1642,
+      "step": 641
+    },
+    {
+      "epoch": 2.168918918918919,
+      "grad_norm": 0.44445449113845825,
+      "learning_rate": 4.679225046781422e-05,
+      "loss": 0.155,
+      "step": 642
+    },
+    {
+      "epoch": 2.1722972972972974,
+      "grad_norm": 0.42584556341171265,
+      "learning_rate": 4.68035162154037e-05,
+      "loss": 0.171,
+      "step": 643
+    },
+    {
+      "epoch": 2.175675675675676,
+      "grad_norm": 0.39437368512153625,
+      "learning_rate": 4.681476445599687e-05,
+      "loss": 0.1692,
+      "step": 644
+    },
+    {
+      "epoch": 2.179054054054054,
+      "grad_norm": 0.40476444363594055,
+      "learning_rate": 4.6825995243921137e-05,
+      "loss": 0.162,
+      "step": 645
+    },
+    {
+      "epoch": 2.1824324324324325,
+      "grad_norm": 0.42337456345558167,
+      "learning_rate": 4.683720863325141e-05,
+      "loss": 0.184,
+      "step": 646
+    },
+    {
+      "epoch": 2.185810810810811,
+      "grad_norm": 0.4256432056427002,
+      "learning_rate": 4.684840467781168e-05,
+      "loss": 0.175,
+      "step": 647
+    },
+    {
+      "epoch": 2.189189189189189,
+      "grad_norm": 0.40138664841651917,
+      "learning_rate": 4.685958343117656e-05,
+      "loss": 0.1577,
+      "step": 648
+    },
+    {
+      "epoch": 2.1925675675675675,
+      "grad_norm": 0.4015848636627197,
+      "learning_rate": 4.6870744946672826e-05,
+      "loss": 0.1521,
+      "step": 649
+    },
+    {
+      "epoch": 2.195945945945946,
+      "grad_norm": 0.4208281934261322,
+      "learning_rate": 4.688188927738093e-05,
+      "loss": 0.1685,
+      "step": 650
+    },
+    {
+      "epoch": 2.199324324324324,
+      "grad_norm": 0.4102751910686493,
+      "learning_rate": 4.689301647613653e-05,
+      "loss": 0.1837,
+      "step": 651
+    },
+    {
+      "epoch": 2.2027027027027026,
+      "grad_norm": 0.43685656785964966,
+      "learning_rate": 4.6904126595532014e-05,
+      "loss": 0.174,
+      "step": 652
+    },
+    {
+      "epoch": 2.206081081081081,
+      "grad_norm": 0.392456978559494,
+      "learning_rate": 4.69152196879179e-05,
+      "loss": 0.1741,
+      "step": 653
+    },
+    {
+      "epoch": 2.2094594594594597,
+      "grad_norm": 0.4086831212043762,
+      "learning_rate": 4.692629580540446e-05,
+      "loss": 0.1559,
+      "step": 654
+    },
+    {
+      "epoch": 2.2128378378378377,
+      "grad_norm": 0.38833364844322205,
+      "learning_rate": 4.693735499986305e-05,
+      "loss": 0.1618,
+      "step": 655
+    },
+    {
+      "epoch": 2.2162162162162162,
+      "grad_norm": 0.36838966608047485,
+      "learning_rate": 4.694839732292767e-05,
+      "loss": 0.1478,
+      "step": 656
+    },
+    {
+      "epoch": 2.2195945945945947,
+      "grad_norm": 0.40487056970596313,
+      "learning_rate": 4.6959422825996345e-05,
+      "loss": 0.1738,
+      "step": 657
+    },
+    {
+      "epoch": 2.222972972972973,
+      "grad_norm": 0.3815852999687195,
+      "learning_rate": 4.69704315602326e-05,
+      "loss": 0.1665,
+      "step": 658
+    },
+    {
+      "epoch": 2.2263513513513513,
+      "grad_norm": 0.400643914937973,
+      "learning_rate": 4.698142357656684e-05,
+      "loss": 0.1771,
+      "step": 659
+    },
+    {
+      "epoch": 2.22972972972973,
+      "grad_norm": 0.3691990077495575,
+      "learning_rate": 4.6992398925697814e-05,
+      "loss": 0.1591,
+      "step": 660
+    },
+    {
+      "epoch": 2.233108108108108,
+      "grad_norm": 0.4398171305656433,
+      "learning_rate": 4.7003357658094e-05,
+      "loss": 0.2062,
+      "step": 661
+    },
+    {
+      "epoch": 2.2364864864864864,
+      "grad_norm": 0.3774857521057129,
+      "learning_rate": 4.7014299823995005e-05,
+      "loss": 0.1502,
+      "step": 662
+    },
+    {
+      "epoch": 2.239864864864865,
+      "grad_norm": 0.4315558075904846,
+      "learning_rate": 4.702522547341289e-05,
+      "loss": 0.1636,
+      "step": 663
+    },
+    {
+      "epoch": 2.2432432432432434,
+      "grad_norm": 0.4040674865245819,
+      "learning_rate": 4.703613465613363e-05,
+      "loss": 0.1625,
+      "step": 664
+    },
+    {
+      "epoch": 2.2466216216216215,
+      "grad_norm": 0.46972906589508057,
+      "learning_rate": 4.704702742171841e-05,
+      "loss": 0.1833,
+      "step": 665
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.41066527366638184,
+      "learning_rate": 4.7057903819505024e-05,
+      "loss": 0.1707,
+      "step": 666
+    },
+    {
+      "epoch": 2.2533783783783785,
+      "grad_norm": 0.3562461733818054,
+      "learning_rate": 4.7068763898609154e-05,
+      "loss": 0.1508,
+      "step": 667
+    },
+    {
+      "epoch": 2.2567567567567566,
+      "grad_norm": 0.4074662923812866,
+      "learning_rate": 4.707960770792576e-05,
+      "loss": 0.1786,
+      "step": 668
+    },
+    {
+      "epoch": 2.260135135135135,
+      "grad_norm": 0.3978296220302582,
+      "learning_rate": 4.709043529613039e-05,
+      "loss": 0.1797,
+      "step": 669
+    },
+    {
+      "epoch": 2.2635135135135136,
+      "grad_norm": 0.42612069845199585,
+      "learning_rate": 4.710124671168044e-05,
+      "loss": 0.1874,
+      "step": 670
+    },
+    {
+      "epoch": 2.266891891891892,
+      "grad_norm": 0.4137380123138428,
+      "learning_rate": 4.711204200281654e-05,
+      "loss": 0.1897,
+      "step": 671
+    },
+    {
+      "epoch": 2.27027027027027,
+      "grad_norm": 0.37814861536026,
+      "learning_rate": 4.712282121756376e-05,
+      "loss": 0.1617,
+      "step": 672
+    },
+    {
+      "epoch": 2.2736486486486487,
+      "grad_norm": 0.42515134811401367,
+      "learning_rate": 4.713358440373295e-05,
+      "loss": 0.1652,
+      "step": 673
+    },
+    {
+      "epoch": 2.277027027027027,
+      "grad_norm": 0.42217129468917847,
+      "learning_rate": 4.7144331608922e-05,
+      "loss": 0.1872,
+      "step": 674
+    },
+    {
+      "epoch": 2.2804054054054053,
+      "grad_norm": 0.3996788263320923,
+      "learning_rate": 4.715506288051709e-05,
+      "loss": 0.1603,
+      "step": 675
+    },
+    {
+      "epoch": 2.2837837837837838,
+      "grad_norm": 0.4035404622554779,
+      "learning_rate": 4.7165778265693935e-05,
+      "loss": 0.1689,
+      "step": 676
+    },
+    {
+      "epoch": 2.2871621621621623,
+      "grad_norm": 0.39791160821914673,
+      "learning_rate": 4.7176477811419076e-05,
+      "loss": 0.1804,
+      "step": 677
+    },
+    {
+      "epoch": 2.2905405405405403,
+      "grad_norm": 0.4175347685813904,
+      "learning_rate": 4.718716156445106e-05,
+      "loss": 0.1811,
+      "step": 678
+    },
+    {
+      "epoch": 2.293918918918919,
+      "grad_norm": 0.39992383122444153,
+      "learning_rate": 4.7197829571341704e-05,
+      "loss": 0.1805,
+      "step": 679
+    },
+    {
+      "epoch": 2.2972972972972974,
+      "grad_norm": 0.40433621406555176,
+      "learning_rate": 4.720848187843727e-05,
+      "loss": 0.1726,
+      "step": 680
+    },
+    {
+      "epoch": 2.3006756756756754,
+      "grad_norm": 0.38958102464675903,
+      "learning_rate": 4.721911853187975e-05,
+      "loss": 0.1663,
+      "step": 681
+    },
+    {
+      "epoch": 2.304054054054054,
+      "grad_norm": 0.40906044840812683,
+      "learning_rate": 4.722973957760799e-05,
+      "loss": 0.1637,
+      "step": 682
+    },
+    {
+      "epoch": 2.3074324324324325,
+      "grad_norm": 0.4296117424964905,
+      "learning_rate": 4.724034506135888e-05,
+      "loss": 0.178,
+      "step": 683
+    },
+    {
+      "epoch": 2.310810810810811,
+      "grad_norm": 0.40840208530426025,
+      "learning_rate": 4.725093502866861e-05,
+      "loss": 0.1725,
+      "step": 684
+    },
+    {
+      "epoch": 2.314189189189189,
+      "grad_norm": 0.3669198751449585,
+      "learning_rate": 4.7261509524873764e-05,
+      "loss": 0.161,
+      "step": 685
+    },
+    {
+      "epoch": 2.3175675675675675,
+      "grad_norm": 0.40982183814048767,
+      "learning_rate": 4.727206859511253e-05,
+      "loss": 0.1757,
+      "step": 686
+    },
+    {
+      "epoch": 2.320945945945946,
+      "grad_norm": 0.4490487575531006,
+      "learning_rate": 4.7282612284325846e-05,
+      "loss": 0.1975,
+      "step": 687
+    },
+    {
+      "epoch": 2.3243243243243246,
+      "grad_norm": 0.4303133487701416,
+      "learning_rate": 4.729314063725853e-05,
+      "loss": 0.1753,
+      "step": 688
+    },
+    {
+      "epoch": 2.3277027027027026,
+      "grad_norm": 0.5020397901535034,
+      "learning_rate": 4.730365369846044e-05,
+      "loss": 0.1959,
+      "step": 689
+    },
+    {
+      "epoch": 2.331081081081081,
+      "grad_norm": 0.4409542679786682,
+      "learning_rate": 4.7314151512287594e-05,
+      "loss": 0.1779,
+      "step": 690
+    },
+    {
+      "epoch": 2.3344594594594597,
+      "grad_norm": 0.41255807876586914,
+      "learning_rate": 4.732463412290331e-05,
+      "loss": 0.1647,
+      "step": 691
+    },
+    {
+      "epoch": 2.3378378378378377,
+      "grad_norm": 0.41047075390815735,
+      "learning_rate": 4.73351015742793e-05,
+      "loss": 0.1699,
+      "step": 692
+    },
+    {
+      "epoch": 2.3412162162162162,
+      "grad_norm": 0.4248691201210022,
+      "learning_rate": 4.7345553910196785e-05,
+      "loss": 0.1785,
+      "step": 693
+    },
+    {
+      "epoch": 2.3445945945945947,
+      "grad_norm": 0.41576990485191345,
+      "learning_rate": 4.735599117424759e-05,
+      "loss": 0.1806,
+      "step": 694
+    },
+    {
+      "epoch": 2.347972972972973,
+      "grad_norm": 0.399854838848114,
+      "learning_rate": 4.7366413409835235e-05,
+      "loss": 0.1689,
+      "step": 695
+    },
+    {
+      "epoch": 2.3513513513513513,
+      "grad_norm": 0.39361926913261414,
+      "learning_rate": 4.737682066017604e-05,
+      "loss": 0.1757,
+      "step": 696
+    },
+    {
+      "epoch": 2.35472972972973,
+      "grad_norm": 0.36152175068855286,
+      "learning_rate": 4.7387212968300166e-05,
+      "loss": 0.1421,
+      "step": 697
+    },
+    {
+      "epoch": 2.358108108108108,
+      "grad_norm": 0.4112738370895386,
+      "learning_rate": 4.7397590377052686e-05,
+      "loss": 0.156,
+      "step": 698
+    },
+    {
+      "epoch": 2.3614864864864864,
+      "grad_norm": 1.494449496269226,
+      "learning_rate": 4.74079529290947e-05,
+      "loss": 0.433,
+      "step": 699
+    },
+    {
+      "epoch": 2.364864864864865,
+      "grad_norm": 0.44618678092956543,
+      "learning_rate": 4.741830066690428e-05,
+      "loss": 0.1738,
+      "step": 700
+    },
+    {
+      "epoch": 2.3682432432432434,
+      "grad_norm": 0.4271227717399597,
+      "learning_rate": 4.742863363277765e-05,
+      "loss": 0.1859,
+      "step": 701
+    },
+    {
+      "epoch": 2.3716216216216215,
+      "grad_norm": 0.38577789068222046,
+      "learning_rate": 4.743895186883009e-05,
+      "loss": 0.1699,
+      "step": 702
+    },
+    {
+      "epoch": 2.375,
+      "grad_norm": 0.45399177074432373,
+      "learning_rate": 4.7449255416997075e-05,
+      "loss": 0.2042,
+      "step": 703
+    },
+    {
+      "epoch": 2.3783783783783785,
+      "grad_norm": 0.4084223806858063,
+      "learning_rate": 4.7459544319035206e-05,
+      "loss": 0.1626,
+      "step": 704
+    },
+    {
+      "epoch": 2.3817567567567566,
+      "grad_norm": 0.4444795846939087,
+      "learning_rate": 4.746981861652332e-05,
+      "loss": 0.1524,
+      "step": 705
+    },
+    {
+      "epoch": 2.385135135135135,
+      "grad_norm": 0.41611728072166443,
+      "learning_rate": 4.74800783508634e-05,
+      "loss": 0.1965,
+      "step": 706
+    },
+    {
+      "epoch": 2.3885135135135136,
+      "grad_norm": 0.4463081657886505,
+      "learning_rate": 4.7490323563281665e-05,
+      "loss": 0.1703,
+      "step": 707
+    },
+    {
+      "epoch": 2.391891891891892,
+      "grad_norm": 0.4274112582206726,
+      "learning_rate": 4.750055429482949e-05,
+      "loss": 0.1468,
+      "step": 708
+    },
+    {
+      "epoch": 2.39527027027027,
+      "grad_norm": 0.5068261623382568,
+      "learning_rate": 4.751077058638445e-05,
+      "loss": 0.1959,
+      "step": 709
+    },
+    {
+      "epoch": 2.3986486486486487,
+      "grad_norm": 0.46266356110572815,
+      "learning_rate": 4.752097247865126e-05,
+      "loss": 0.1876,
+      "step": 710
+    },
+    {
+      "epoch": 2.402027027027027,
+      "grad_norm": 0.4017108082771301,
+      "learning_rate": 4.753116001216277e-05,
+      "loss": 0.1567,
+      "step": 711
+    },
+    {
+      "epoch": 2.4054054054054053,
+      "grad_norm": 0.4414288103580475,
+      "learning_rate": 4.7541333227280944e-05,
+      "loss": 0.171,
+      "step": 712
+    },
+    {
+      "epoch": 2.4087837837837838,
+      "grad_norm": 0.47343650460243225,
+      "learning_rate": 4.755149216419776e-05,
+      "loss": 0.1973,
+      "step": 713
+    },
+    {
+      "epoch": 2.4121621621621623,
+      "grad_norm": 0.4446316361427307,
+      "learning_rate": 4.756163686293624e-05,
+      "loss": 0.1804,
+      "step": 714
+    },
+    {
+      "epoch": 2.4155405405405403,
+      "grad_norm": 0.4418063163757324,
+      "learning_rate": 4.7571767363351344e-05,
+      "loss": 0.1913,
+      "step": 715
+    },
+    {
+      "epoch": 2.418918918918919,
+      "grad_norm": 0.4041652977466583,
+      "learning_rate": 4.758188370513093e-05,
+      "loss": 0.1852,
+      "step": 716
+    },
+    {
+      "epoch": 2.4222972972972974,
+      "grad_norm": 0.41937291622161865,
+      "learning_rate": 4.759198592779667e-05,
+      "loss": 0.1925,
+      "step": 717
+    },
+    {
+      "epoch": 2.4256756756756754,
+      "grad_norm": 0.4121197462081909,
+      "learning_rate": 4.760207407070501e-05,
+      "loss": 0.1821,
+      "step": 718
+    },
+    {
+      "epoch": 2.429054054054054,
+      "grad_norm": 0.3996151387691498,
+      "learning_rate": 4.761214817304805e-05,
+      "loss": 0.1837,
+      "step": 719
+    },
+    {
+      "epoch": 2.4324324324324325,
+      "grad_norm": 0.38412001729011536,
+      "learning_rate": 4.762220827385448e-05,
+      "loss": 0.1685,
+      "step": 720
+    },
+    {
+      "epoch": 2.435810810810811,
+      "grad_norm": 1.5490686893463135,
+      "learning_rate": 4.763225441199049e-05,
+      "loss": 0.432,
+      "step": 721
+    },
+    {
+      "epoch": 2.439189189189189,
+      "grad_norm": 0.4263227880001068,
+      "learning_rate": 4.7642286626160654e-05,
+      "loss": 0.1847,
+      "step": 722
+    },
+    {
+      "epoch": 2.4425675675675675,
+      "grad_norm": 0.3754761815071106,
+      "learning_rate": 4.765230495490885e-05,
+      "loss": 0.1786,
+      "step": 723
+    },
+    {
+      "epoch": 2.445945945945946,
+      "grad_norm": 0.43871861696243286,
+      "learning_rate": 4.7662309436619115e-05,
+      "loss": 0.189,
+      "step": 724
+    },
+    {
+      "epoch": 2.4493243243243246,
+      "grad_norm": 0.4133760333061218,
+      "learning_rate": 4.7672300109516563e-05,
+      "loss": 0.1794,
+      "step": 725
+    },
+    {
+      "epoch": 2.4527027027027026,
+      "grad_norm": 0.38480374217033386,
+      "learning_rate": 4.768227701166823e-05,
+      "loss": 0.1666,
+      "step": 726
+    },
+    {
+      "epoch": 2.456081081081081,
+      "grad_norm": 0.4138960540294647,
+      "learning_rate": 4.7692240180983964e-05,
+      "loss": 0.1695,
+      "step": 727
+    },
+    {
+      "epoch": 2.4594594594594597,
+      "grad_norm": 0.3594178557395935,
+      "learning_rate": 4.770218965521729e-05,
+      "loss": 0.1519,
+      "step": 728
+    },
+    {
+      "epoch": 2.4628378378378377,
+      "grad_norm": 0.4005281627178192,
+      "learning_rate": 4.7712125471966245e-05,
+      "loss": 0.1696,
+      "step": 729
+    },
+    {
+      "epoch": 2.4662162162162162,
+      "grad_norm": 0.3780256509780884,
+      "learning_rate": 4.7722047668674267e-05,
+      "loss": 0.174,
+      "step": 730
+    },
+    {
+      "epoch": 2.4695945945945947,
+      "grad_norm": 0.4074977934360504,
+      "learning_rate": 4.7731956282631004e-05,
+      "loss": 0.1696,
+      "step": 731
+    },
+    {
+      "epoch": 2.472972972972973,
+      "grad_norm": 0.4063095152378082,
+      "learning_rate": 4.77418513509732e-05,
+      "loss": 0.1731,
+      "step": 732
+    },
+    {
+      "epoch": 2.4763513513513513,
+      "grad_norm": 0.40110018849372864,
+      "learning_rate": 4.775173291068547e-05,
+      "loss": 0.1787,
+      "step": 733
+    },
+    {
+      "epoch": 2.47972972972973,
+      "grad_norm": 0.3782990574836731,
+      "learning_rate": 4.776160099860117e-05,
+      "loss": 0.1753,
+      "step": 734
+    },
+    {
+      "epoch": 2.483108108108108,
+      "grad_norm": 0.3815288543701172,
+      "learning_rate": 4.777145565140325e-05,
+      "loss": 0.1766,
+      "step": 735
+    },
+    {
+      "epoch": 2.4864864864864864,
+      "grad_norm": 0.3779781758785248,
+      "learning_rate": 4.7781296905624986e-05,
+      "loss": 0.1857,
+      "step": 736
+    },
+    {
+      "epoch": 2.489864864864865,
+      "grad_norm": 0.33802899718284607,
+      "learning_rate": 4.779112479765086e-05,
+      "loss": 0.1515,
+      "step": 737
+    },
+    {
+      "epoch": 2.4932432432432434,
+      "grad_norm": 0.4050140082836151,
+      "learning_rate": 4.780093936371736e-05,
+      "loss": 0.1908,
+      "step": 738
+    },
+    {
+      "epoch": 2.4966216216216215,
+      "grad_norm": 0.4146076738834381,
+      "learning_rate": 4.781074063991376e-05,
+      "loss": 0.1812,
+      "step": 739
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.34979405999183655,
+      "learning_rate": 4.782052866218294e-05,
+      "loss": 0.1505,
+      "step": 740
+    },
+    {
+      "epoch": 2.5033783783783785,
+      "grad_norm": 0.3686615526676178,
+      "learning_rate": 4.783030346632214e-05,
+      "loss": 0.1656,
+      "step": 741
+    },
+    {
+      "epoch": 2.506756756756757,
+      "grad_norm": 0.38285037875175476,
+      "learning_rate": 4.7840065087983786e-05,
+      "loss": 0.181,
+      "step": 742
+    },
+    {
+      "epoch": 2.510135135135135,
+      "grad_norm": 0.39826491475105286,
+      "learning_rate": 4.784981356267626e-05,
+      "loss": 0.201,
+      "step": 743
+    },
+    {
+      "epoch": 2.5135135135135136,
+      "grad_norm": 0.40031111240386963,
+      "learning_rate": 4.785954892576465e-05,
+      "loss": 0.1676,
+      "step": 744
+    },
+    {
+      "epoch": 2.516891891891892,
+      "grad_norm": 0.412266343832016,
+      "learning_rate": 4.7869271212471554e-05,
+      "loss": 0.1807,
+      "step": 745
+    },
+    {
+      "epoch": 2.52027027027027,
+      "grad_norm": 0.38458314538002014,
+      "learning_rate": 4.7878980457877814e-05,
+      "loss": 0.1639,
+      "step": 746
+    },
+    {
+      "epoch": 2.5236486486486487,
+      "grad_norm": 0.45593783259391785,
+      "learning_rate": 4.7888676696923315e-05,
+      "loss": 0.1844,
+      "step": 747
+    },
+    {
+      "epoch": 2.527027027027027,
+      "grad_norm": 0.36613929271698,
+      "learning_rate": 4.7898359964407695e-05,
+      "loss": 0.1536,
+      "step": 748
+    },
+    {
+      "epoch": 2.5304054054054053,
+      "grad_norm": 0.4267789125442505,
+      "learning_rate": 4.790803029499111e-05,
+      "loss": 0.1706,
+      "step": 749
+    },
+    {
+      "epoch": 2.5337837837837838,
+      "grad_norm": 1.5388500690460205,
+      "learning_rate": 4.7917687723195004e-05,
+      "loss": 0.4571,
+      "step": 750
+    },
+    {
+      "epoch": 2.5371621621621623,
+      "grad_norm": 0.42774277925491333,
+      "learning_rate": 4.792733228340281e-05,
+      "loss": 0.1907,
+      "step": 751
+    },
+    {
+      "epoch": 2.5405405405405403,
+      "grad_norm": 0.38267603516578674,
+      "learning_rate": 4.793696400986071e-05,
+      "loss": 0.163,
+      "step": 752
+    },
+    {
+      "epoch": 2.543918918918919,
+      "grad_norm": 0.4370862543582916,
+      "learning_rate": 4.7946582936678344e-05,
+      "loss": 0.2019,
+      "step": 753
+    },
+    {
+      "epoch": 2.5472972972972974,
+      "grad_norm": 0.3963480293750763,
+      "learning_rate": 4.795618909782957e-05,
+      "loss": 0.158,
+      "step": 754
+    },
+    {
+      "epoch": 2.5506756756756754,
+      "grad_norm": 0.4094291031360626,
+      "learning_rate": 4.796578252715314e-05,
+      "loss": 0.1769,
+      "step": 755
+    },
+    {
+      "epoch": 2.554054054054054,
+      "grad_norm": 0.41850391030311584,
+      "learning_rate": 4.797536325835345e-05,
+      "loss": 0.17,
+      "step": 756
+    },
+    {
+      "epoch": 2.5574324324324325,
+      "grad_norm": 0.4879460036754608,
+      "learning_rate": 4.7984931325001216e-05,
+      "loss": 0.2163,
+      "step": 757
+    },
+    {
+      "epoch": 2.560810810810811,
+      "grad_norm": 0.4456654489040375,
+      "learning_rate": 4.799448676053423e-05,
+      "loss": 0.2086,
+      "step": 758
+    },
+    {
+      "epoch": 2.564189189189189,
+      "grad_norm": 0.41925883293151855,
+      "learning_rate": 4.800402959825802e-05,
+      "loss": 0.1884,
+      "step": 759
+    },
+    {
+      "epoch": 2.5675675675675675,
+      "grad_norm": 0.4378669857978821,
+      "learning_rate": 4.801355987134653e-05,
+      "loss": 0.2024,
+      "step": 760
+    },
+    {
+      "epoch": 2.570945945945946,
+      "grad_norm": 0.5047132968902588,
+      "learning_rate": 4.802307761284289e-05,
+      "loss": 0.1961,
+      "step": 761
+    },
+    {
+      "epoch": 2.5743243243243246,
+      "grad_norm": 0.41319629549980164,
+      "learning_rate": 4.8032582855660014e-05,
+      "loss": 0.1957,
+      "step": 762
+    },
+    {
+      "epoch": 2.5777027027027026,
+      "grad_norm": 0.46079912781715393,
+      "learning_rate": 4.8042075632581346e-05,
+      "loss": 0.1931,
+      "step": 763
+    },
+    {
+      "epoch": 2.581081081081081,
+      "grad_norm": 0.431325763463974,
+      "learning_rate": 4.80515559762615e-05,
+      "loss": 0.1788,
+      "step": 764
+    },
+    {
+      "epoch": 2.5844594594594597,
+      "grad_norm": 0.40309104323387146,
+      "learning_rate": 4.8061023919226964e-05,
+      "loss": 0.1817,
+      "step": 765
+    },
+    {
+      "epoch": 2.5878378378378377,
+      "grad_norm": 0.434621661901474,
+      "learning_rate": 4.807047949387674e-05,
+      "loss": 0.1831,
+      "step": 766
+    },
+    {
+      "epoch": 2.5912162162162162,
+      "grad_norm": 0.35631585121154785,
+      "learning_rate": 4.807992273248302e-05,
+      "loss": 0.1683,
+      "step": 767
+    },
+    {
+      "epoch": 2.5945945945945947,
+      "grad_norm": 0.41151365637779236,
+      "learning_rate": 4.808935366719187e-05,
+      "loss": 0.1904,
+      "step": 768
+    },
+    {
+      "epoch": 2.597972972972973,
+      "grad_norm": 0.4089500904083252,
+      "learning_rate": 4.8098772330023855e-05,
+      "loss": 0.1742,
+      "step": 769
+    },
+    {
+      "epoch": 2.6013513513513513,
+      "grad_norm": 0.38123077154159546,
+      "learning_rate": 4.81081787528747e-05,
+      "loss": 0.1727,
+      "step": 770
+    },
+    {
+      "epoch": 2.60472972972973,
+      "grad_norm": 0.37730783224105835,
+      "learning_rate": 4.811757296751595e-05,
+      "loss": 0.1904,
+      "step": 771
+    },
+    {
+      "epoch": 2.608108108108108,
+      "grad_norm": 0.38781440258026123,
+      "learning_rate": 4.812695500559561e-05,
+      "loss": 0.1931,
+      "step": 772
+    },
+    {
+      "epoch": 2.6114864864864864,
+      "grad_norm": 0.4114435315132141,
+      "learning_rate": 4.8136324898638756e-05,
+      "loss": 0.2105,
+      "step": 773
+    },
+    {
+      "epoch": 2.614864864864865,
+      "grad_norm": 0.39372387528419495,
+      "learning_rate": 4.8145682678048214e-05,
+      "loss": 0.1831,
+      "step": 774
+    },
+    {
+      "epoch": 2.618243243243243,
+      "grad_norm": 0.40174078941345215,
+      "learning_rate": 4.815502837510518e-05,
+      "loss": 0.1975,
+      "step": 775
+    },
+    {
+      "epoch": 2.6216216216216215,
+      "grad_norm": 0.4061259627342224,
+      "learning_rate": 4.816436202096981e-05,
+      "loss": 0.1744,
+      "step": 776
+    },
+    {
+      "epoch": 2.625,
+      "grad_norm": 0.3997017741203308,
+      "learning_rate": 4.81736836466819e-05,
+      "loss": 0.1782,
+      "step": 777
+    },
+    {
+      "epoch": 2.6283783783783785,
+      "grad_norm": 0.4183482229709625,
+      "learning_rate": 4.8182993283161485e-05,
+      "loss": 0.1941,
+      "step": 778
+    },
+    {
+      "epoch": 2.631756756756757,
+      "grad_norm": 0.3742780387401581,
+      "learning_rate": 4.819229096120941e-05,
+      "loss": 0.1681,
+      "step": 779
+    },
+    {
+      "epoch": 2.635135135135135,
+      "grad_norm": 0.3982739746570587,
+      "learning_rate": 4.820157671150801e-05,
+      "loss": 0.1941,
+      "step": 780
+    },
+    {
+      "epoch": 2.6385135135135136,
+      "grad_norm": 0.3757505714893341,
+      "learning_rate": 4.821085056462168e-05,
+      "loss": 0.1744,
+      "step": 781
+    },
+    {
+      "epoch": 2.641891891891892,
+      "grad_norm": 0.39277997612953186,
+      "learning_rate": 4.822011255099747e-05,
+      "loss": 0.1803,
+      "step": 782
+    },
+    {
+      "epoch": 2.64527027027027,
+      "grad_norm": 0.35470277070999146,
+      "learning_rate": 4.8229362700965726e-05,
+      "loss": 0.1651,
+      "step": 783
+    },
+    {
+      "epoch": 2.6486486486486487,
+      "grad_norm": 0.381610631942749,
+      "learning_rate": 4.8238601044740645e-05,
+      "loss": 0.183,
+      "step": 784
+    },
+    {
+      "epoch": 2.652027027027027,
+      "grad_norm": 0.37985488772392273,
+      "learning_rate": 4.824782761242088e-05,
+      "loss": 0.1637,
+      "step": 785
+    },
+    {
+      "epoch": 2.6554054054054053,
+      "grad_norm": 0.4003821909427643,
+      "learning_rate": 4.8257042433990135e-05,
+      "loss": 0.18,
+      "step": 786
+    },
+    {
+      "epoch": 2.6587837837837838,
+      "grad_norm": 0.37772974371910095,
+      "learning_rate": 4.826624553931775e-05,
+      "loss": 0.1886,
+      "step": 787
+    },
+    {
+      "epoch": 2.6621621621621623,
+      "grad_norm": 0.3801021873950958,
+      "learning_rate": 4.827543695815926e-05,
+      "loss": 0.1718,
+      "step": 788
+    },
+    {
+      "epoch": 2.6655405405405403,
+      "grad_norm": 0.3759152591228485,
+      "learning_rate": 4.8284616720157006e-05,
+      "loss": 0.1698,
+      "step": 789
+    },
+    {
+      "epoch": 2.668918918918919,
+      "grad_norm": 0.38654980063438416,
+      "learning_rate": 4.82937848548407e-05,
+      "loss": 0.1885,
+      "step": 790
+    },
+    {
+      "epoch": 2.6722972972972974,
+      "grad_norm": 0.4393473267555237,
+      "learning_rate": 4.8302941391627947e-05,
+      "loss": 0.1891,
+      "step": 791
+    },
+    {
+      "epoch": 2.6756756756756754,
+      "grad_norm": 0.3831874132156372,
+      "learning_rate": 4.83120863598249e-05,
+      "loss": 0.1888,
+      "step": 792
+    },
+    {
+      "epoch": 2.679054054054054,
+      "grad_norm": 0.38641834259033203,
+      "learning_rate": 4.832121978862673e-05,
+      "loss": 0.1842,
+      "step": 793
+    },
+    {
+      "epoch": 2.6824324324324325,
+      "grad_norm": 0.39984583854675293,
+      "learning_rate": 4.8330341707118276e-05,
+      "loss": 0.1681,
+      "step": 794
+    },
+    {
+      "epoch": 2.685810810810811,
+      "grad_norm": 0.40252962708473206,
+      "learning_rate": 4.833945214427451e-05,
+      "loss": 0.1756,
+      "step": 795
+    },
+    {
+      "epoch": 2.689189189189189,
+      "grad_norm": 0.40013495087623596,
+      "learning_rate": 4.834855112896116e-05,
+      "loss": 0.2146,
+      "step": 796
+    },
+    {
+      "epoch": 2.6925675675675675,
+      "grad_norm": 0.385640412569046,
+      "learning_rate": 4.835763868993521e-05,
+      "loss": 0.1851,
+      "step": 797
+    },
+    {
+      "epoch": 2.695945945945946,
+      "grad_norm": 0.3977515697479248,
+      "learning_rate": 4.8366714855845496e-05,
+      "loss": 0.1809,
+      "step": 798
+    },
+    {
+      "epoch": 2.6993243243243246,
+      "grad_norm": 0.40790390968322754,
+      "learning_rate": 4.837577965523319e-05,
+      "loss": 0.1887,
+      "step": 799
+    },
+    {
+      "epoch": 2.7027027027027026,
+      "grad_norm": 0.3771408498287201,
+      "learning_rate": 4.8384833116532396e-05,
+      "loss": 0.1732,
+      "step": 800
+    },
+    {
+      "epoch": 2.706081081081081,
+      "grad_norm": 0.42990434169769287,
+      "learning_rate": 4.8393875268070636e-05,
+      "loss": 0.1952,
+      "step": 801
+    },
+    {
+      "epoch": 2.7094594594594597,
+      "grad_norm": 0.38725292682647705,
+      "learning_rate": 4.84029061380694e-05,
+      "loss": 0.1903,
+      "step": 802
+    },
+    {
+      "epoch": 2.7128378378378377,
+      "grad_norm": 0.4196310341358185,
+      "learning_rate": 4.841192575464469e-05,
+      "loss": 0.2035,
+      "step": 803
+    },
+    {
+      "epoch": 2.7162162162162162,
+      "grad_norm": 0.4179129898548126,
+      "learning_rate": 4.842093414580753e-05,
+      "loss": 0.2002,
+      "step": 804
+    },
+    {
+      "epoch": 2.7195945945945947,
+      "grad_norm": 0.44000861048698425,
+      "learning_rate": 4.842993133946448e-05,
+      "loss": 0.2299,
+      "step": 805
+    },
+    {
+      "epoch": 2.722972972972973,
+      "grad_norm": 0.4462367296218872,
+      "learning_rate": 4.843891736341818e-05,
+      "loss": 0.1931,
+      "step": 806
+    },
+    {
+      "epoch": 2.7263513513513513,
+      "grad_norm": 0.42056065797805786,
+      "learning_rate": 4.8447892245367846e-05,
+      "loss": 0.2012,
+      "step": 807
+    },
+    {
+      "epoch": 2.72972972972973,
+      "grad_norm": 0.48042038083076477,
+      "learning_rate": 4.845685601290977e-05,
+      "loss": 0.1988,
+      "step": 808
+    },
+    {
+      "epoch": 2.733108108108108,
+      "grad_norm": 0.39594507217407227,
+      "learning_rate": 4.846580869353787e-05,
+      "loss": 0.1966,
+      "step": 809
+    },
+    {
+      "epoch": 2.7364864864864864,
+      "grad_norm": 0.4334581792354584,
+      "learning_rate": 4.847475031464416e-05,
+      "loss": 0.1861,
+      "step": 810
+    },
+    {
+      "epoch": 2.739864864864865,
+      "grad_norm": 0.40071550011634827,
+      "learning_rate": 4.8483680903519274e-05,
+      "loss": 0.2009,
+      "step": 811
+    },
+    {
+      "epoch": 2.743243243243243,
+      "grad_norm": 0.40345826745033264,
+      "learning_rate": 4.8492600487352926e-05,
+      "loss": 0.1692,
+      "step": 812
+    },
+    {
+      "epoch": 2.7466216216216215,
+      "grad_norm": 0.41826799511909485,
+      "learning_rate": 4.850150909323447e-05,
+      "loss": 0.1904,
+      "step": 813
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.3715426027774811,
+      "learning_rate": 4.8510406748153355e-05,
+      "loss": 0.1782,
+      "step": 814
+    },
+    {
+      "epoch": 2.7533783783783785,
+      "grad_norm": 0.39606815576553345,
+      "learning_rate": 4.8519293478999614e-05,
+      "loss": 0.1824,
+      "step": 815
+    },
+    {
+      "epoch": 2.756756756756757,
+      "grad_norm": 0.3894452452659607,
+      "learning_rate": 4.8528169312564355e-05,
+      "loss": 0.1799,
+      "step": 816
+    },
+    {
+      "epoch": 2.760135135135135,
+      "grad_norm": 0.38379618525505066,
+      "learning_rate": 4.8537034275540264e-05,
+      "loss": 0.1792,
+      "step": 817
+    },
+    {
+      "epoch": 2.7635135135135136,
+      "grad_norm": 0.4054020643234253,
+      "learning_rate": 4.854588839452205e-05,
+      "loss": 0.1842,
+      "step": 818
+    },
+    {
+      "epoch": 2.766891891891892,
+      "grad_norm": 0.424482524394989,
+      "learning_rate": 4.855473169600698e-05,
+      "loss": 0.1781,
+      "step": 819
+    },
+    {
+      "epoch": 2.77027027027027,
+      "grad_norm": 0.40162649750709534,
+      "learning_rate": 4.856356420639528e-05,
+      "loss": 0.2129,
+      "step": 820
+    },
+    {
+      "epoch": 2.7736486486486487,
+      "grad_norm": 0.4196532666683197,
+      "learning_rate": 4.857238595199068e-05,
+      "loss": 0.1844,
+      "step": 821
+    },
+    {
+      "epoch": 2.777027027027027,
+      "grad_norm": 0.39450812339782715,
+      "learning_rate": 4.858119695900084e-05,
+      "loss": 0.1917,
+      "step": 822
+    },
+    {
+      "epoch": 2.7804054054054053,
+      "grad_norm": 0.37791892886161804,
+      "learning_rate": 4.858999725353783e-05,
+      "loss": 0.1562,
+      "step": 823
+    },
+    {
+      "epoch": 2.7837837837837838,
+      "grad_norm": 0.4025024473667145,
+      "learning_rate": 4.8598786861618605e-05,
+      "loss": 0.1963,
+      "step": 824
+    },
+    {
+      "epoch": 2.7871621621621623,
+      "grad_norm": 0.382782518863678,
+      "learning_rate": 4.860756580916542e-05,
+      "loss": 0.1811,
+      "step": 825
+    },
+    {
+      "epoch": 2.7905405405405403,
+      "grad_norm": 0.40827038884162903,
+      "learning_rate": 4.861633412200637e-05,
+      "loss": 0.1914,
+      "step": 826
+    },
+    {
+      "epoch": 2.793918918918919,
+      "grad_norm": 0.35224664211273193,
+      "learning_rate": 4.862509182587578e-05,
+      "loss": 0.1587,
+      "step": 827
+    },
+    {
+      "epoch": 2.7972972972972974,
+      "grad_norm": 0.40827634930610657,
+      "learning_rate": 4.863383894641467e-05,
+      "loss": 0.1648,
+      "step": 828
+    },
+    {
+      "epoch": 2.8006756756756754,
+      "grad_norm": 0.3934107720851898,
+      "learning_rate": 4.864257550917123e-05,
+      "loss": 0.1867,
+      "step": 829
+    },
+    {
+      "epoch": 2.804054054054054,
+      "grad_norm": 0.4319005012512207,
+      "learning_rate": 4.865130153960124e-05,
+      "loss": 0.1789,
+      "step": 830
+    },
+    {
+      "epoch": 2.8074324324324325,
+      "grad_norm": 0.4071970582008362,
+      "learning_rate": 4.8660017063068526e-05,
+      "loss": 0.1718,
+      "step": 831
+    },
+    {
+      "epoch": 2.810810810810811,
+      "grad_norm": 0.4246816635131836,
+      "learning_rate": 4.8668722104845403e-05,
+      "loss": 0.1911,
+      "step": 832
+    },
+    {
+      "epoch": 2.814189189189189,
+      "grad_norm": 0.41370970010757446,
+      "learning_rate": 4.8677416690113134e-05,
+      "loss": 0.1968,
+      "step": 833
+    },
+    {
+      "epoch": 2.8175675675675675,
+      "grad_norm": 0.3547184467315674,
+      "learning_rate": 4.868610084396232e-05,
+      "loss": 0.1612,
+      "step": 834
+    },
+    {
+      "epoch": 2.820945945945946,
+      "grad_norm": 0.4511527121067047,
+      "learning_rate": 4.869477459139337e-05,
+      "loss": 0.1608,
+      "step": 835
+    },
+    {
+      "epoch": 2.8243243243243246,
+      "grad_norm": 0.394466757774353,
+      "learning_rate": 4.870343795731694e-05,
+      "loss": 0.1846,
+      "step": 836
+    },
+    {
+      "epoch": 2.8277027027027026,
+      "grad_norm": 0.3896351158618927,
+      "learning_rate": 4.8712090966554334e-05,
+      "loss": 0.1954,
+      "step": 837
+    },
+    {
+      "epoch": 2.831081081081081,
+      "grad_norm": 0.4012662172317505,
+      "learning_rate": 4.872073364383795e-05,
+      "loss": 0.1875,
+      "step": 838
+    },
+    {
+      "epoch": 2.8344594594594597,
+      "grad_norm": 0.39143961668014526,
+      "learning_rate": 4.8729366013811674e-05,
+      "loss": 0.2066,
+      "step": 839
+    },
+    {
+      "epoch": 2.8378378378378377,
+      "grad_norm": 0.4024941325187683,
+      "learning_rate": 4.8737988101031366e-05,
+      "loss": 0.1836,
+      "step": 840
+    },
+    {
+      "epoch": 2.8412162162162162,
+      "grad_norm": 0.3429911732673645,
+      "learning_rate": 4.874659992996521e-05,
+      "loss": 0.1687,
+      "step": 841
+    },
+    {
+      "epoch": 2.8445945945945947,
+      "grad_norm": 0.3772415816783905,
+      "learning_rate": 4.875520152499416e-05,
+      "loss": 0.2006,
+      "step": 842
+    },
+    {
+      "epoch": 2.847972972972973,
+      "grad_norm": 0.41397014260292053,
+      "learning_rate": 4.876379291041238e-05,
+      "loss": 0.2175,
+      "step": 843
+    },
+    {
+      "epoch": 2.8513513513513513,
+      "grad_norm": 0.41761255264282227,
+      "learning_rate": 4.8772374110427594e-05,
+      "loss": 0.1779,
+      "step": 844
+    },
+    {
+      "epoch": 2.85472972972973,
+      "grad_norm": 0.4116886854171753,
+      "learning_rate": 4.878094514916154e-05,
+      "loss": 0.2,
+      "step": 845
+    },
+    {
+      "epoch": 2.858108108108108,
+      "grad_norm": 0.3655768930912018,
+      "learning_rate": 4.8789506050650396e-05,
+      "loss": 0.1818,
+      "step": 846
+    },
+    {
+      "epoch": 2.8614864864864864,
+      "grad_norm": 0.43548357486724854,
+      "learning_rate": 4.879805683884512e-05,
+      "loss": 0.2029,
+      "step": 847
+    },
+    {
+      "epoch": 2.864864864864865,
+      "grad_norm": 0.3908335566520691,
+      "learning_rate": 4.8806597537611906e-05,
+      "loss": 0.1936,
+      "step": 848
+    },
+    {
+      "epoch": 2.868243243243243,
+      "grad_norm": 0.3648586869239807,
+      "learning_rate": 4.881512817073255e-05,
+      "loss": 0.1765,
+      "step": 849
+    },
+    {
+      "epoch": 2.8716216216216215,
+      "grad_norm": 0.39892107248306274,
+      "learning_rate": 4.882364876190489e-05,
+      "loss": 0.1594,
+      "step": 850
+    },
+    {
+      "epoch": 2.875,
+      "grad_norm": 0.3832027316093445,
+      "learning_rate": 4.8832159334743136e-05,
+      "loss": 0.1773,
+      "step": 851
+    },
+    {
+      "epoch": 2.8783783783783785,
+      "grad_norm": 0.4372403621673584,
+      "learning_rate": 4.884065991277833e-05,
+      "loss": 0.2057,
+      "step": 852
+    },
+    {
+      "epoch": 2.881756756756757,
+      "grad_norm": 0.36279168725013733,
+      "learning_rate": 4.8849150519458726e-05,
+      "loss": 0.1726,
+      "step": 853
+    },
+    {
+      "epoch": 2.885135135135135,
+      "grad_norm": 0.37340691685676575,
+      "learning_rate": 4.885763117815009e-05,
+      "loss": 0.178,
+      "step": 854
+    },
+    {
+      "epoch": 2.8885135135135136,
+      "grad_norm": 0.3998451828956604,
+      "learning_rate": 4.886610191213622e-05,
+      "loss": 0.1816,
+      "step": 855
+    },
+    {
+      "epoch": 2.891891891891892,
+      "grad_norm": 0.3779732584953308,
+      "learning_rate": 4.887456274461922e-05,
+      "loss": 0.2068,
+      "step": 856
+    },
+    {
+      "epoch": 2.89527027027027,
+      "grad_norm": 0.4435446262359619,
+      "learning_rate": 4.8883013698719973e-05,
+      "loss": 0.215,
+      "step": 857
+    },
+    {
+      "epoch": 2.8986486486486487,
+      "grad_norm": 0.37998583912849426,
+      "learning_rate": 4.889145479747843e-05,
+      "loss": 0.181,
+      "step": 858
+    },
+    {
+      "epoch": 2.902027027027027,
+      "grad_norm": 0.4032045900821686,
+      "learning_rate": 4.889988606385404e-05,
+      "loss": 0.1878,
+      "step": 859
+    },
+    {
+      "epoch": 2.9054054054054053,
+      "grad_norm": 0.4383464753627777,
+      "learning_rate": 4.8908307520726135e-05,
+      "loss": 0.1948,
+      "step": 860
+    },
+    {
+      "epoch": 2.9087837837837838,
+      "grad_norm": 0.38737547397613525,
+      "learning_rate": 4.891671919089425e-05,
+      "loss": 0.2007,
+      "step": 861
+    },
+    {
+      "epoch": 2.9121621621621623,
+      "grad_norm": 0.39179664850234985,
+      "learning_rate": 4.892512109707855e-05,
+      "loss": 0.1865,
+      "step": 862
+    },
+    {
+      "epoch": 2.9155405405405403,
+      "grad_norm": 0.3944191038608551,
+      "learning_rate": 4.893351326192016e-05,
+      "loss": 0.1831,
+      "step": 863
+    },
+    {
+      "epoch": 2.918918918918919,
+      "grad_norm": 0.3212383985519409,
+      "learning_rate": 4.894189570798156e-05,
+      "loss": 0.1494,
+      "step": 864
+    },
+    {
+      "epoch": 2.9222972972972974,
+      "grad_norm": 0.3861388564109802,
+      "learning_rate": 4.895026845774691e-05,
+      "loss": 0.2045,
+      "step": 865
+    },
+    {
+      "epoch": 2.9256756756756754,
+      "grad_norm": 0.37900376319885254,
+      "learning_rate": 4.895863153362244e-05,
+      "loss": 0.1818,
+      "step": 866
+    },
+    {
+      "epoch": 2.929054054054054,
+      "grad_norm": 0.4495367109775543,
+      "learning_rate": 4.896698495793684e-05,
+      "loss": 0.1936,
+      "step": 867
+    },
+    {
+      "epoch": 2.9324324324324325,
+      "grad_norm": 0.40020546317100525,
+      "learning_rate": 4.897532875294154e-05,
+      "loss": 0.1964,
+      "step": 868
+    },
+    {
+      "epoch": 2.935810810810811,
+      "grad_norm": 0.3625187277793884,
+      "learning_rate": 4.8983662940811115e-05,
+      "loss": 0.1737,
+      "step": 869
+    },
+    {
+      "epoch": 2.939189189189189,
+      "grad_norm": 0.3938811719417572,
+      "learning_rate": 4.899198754364365e-05,
+      "loss": 0.1488,
+      "step": 870
+    },
+    {
+      "epoch": 2.9425675675675675,
+      "grad_norm": 0.36418697237968445,
+      "learning_rate": 4.900030258346106e-05,
+      "loss": 0.1822,
+      "step": 871
+    },
+    {
+      "epoch": 2.945945945945946,
+      "grad_norm": 0.3703506588935852,
+      "learning_rate": 4.900860808220946e-05,
+      "loss": 0.1846,
+      "step": 872
+    },
+    {
+      "epoch": 2.9493243243243246,
+      "grad_norm": 0.4332616925239563,
+      "learning_rate": 4.90169040617595e-05,
+      "loss": 0.2105,
+      "step": 873
+    },
+    {
+      "epoch": 2.9527027027027026,
+      "grad_norm": 0.3770284354686737,
+      "learning_rate": 4.9025190543906715e-05,
+      "loss": 0.1815,
+      "step": 874
+    },
+    {
+      "epoch": 2.956081081081081,
+      "grad_norm": 0.3703818619251251,
+      "learning_rate": 4.903346755037189e-05,
+      "loss": 0.1713,
+      "step": 875
+    },
+    {
+      "epoch": 2.9594594594594597,
+      "grad_norm": 0.37928834557533264,
+      "learning_rate": 4.904173510280135e-05,
+      "loss": 0.1915,
+      "step": 876
+    },
+    {
+      "epoch": 2.9628378378378377,
+      "grad_norm": 0.4267534911632538,
+      "learning_rate": 4.904999322276735e-05,
+      "loss": 0.2157,
+      "step": 877
+    },
+    {
+      "epoch": 2.9662162162162162,
+      "grad_norm": 0.3936428725719452,
+      "learning_rate": 4.9058241931768385e-05,
+      "loss": 0.1947,
+      "step": 878
+    },
+    {
+      "epoch": 2.9695945945945947,
+      "grad_norm": 0.3503859043121338,
+      "learning_rate": 4.9066481251229535e-05,
+      "loss": 0.1836,
+      "step": 879
+    },
+    {
+      "epoch": 2.972972972972973,
+      "grad_norm": 1.8326294422149658,
+      "learning_rate": 4.907471120250281e-05,
+      "loss": 0.4757,
+      "step": 880
+    },
+    {
+      "epoch": 2.9763513513513513,
+      "grad_norm": 0.4133761525154114,
+      "learning_rate": 4.9082931806867474e-05,
+      "loss": 0.2003,
+      "step": 881
+    },
+    {
+      "epoch": 2.97972972972973,
+      "grad_norm": 0.3848228454589844,
+      "learning_rate": 4.909114308553033e-05,
+      "loss": 0.19,
+      "step": 882
+    },
+    {
+      "epoch": 2.983108108108108,
+      "grad_norm": 0.38853901624679565,
+      "learning_rate": 4.909934505962615e-05,
+      "loss": 0.2071,
+      "step": 883
+    },
+    {
+      "epoch": 2.9864864864864864,
+      "grad_norm": 0.35326841473579407,
+      "learning_rate": 4.9107537750217886e-05,
+      "loss": 0.1742,
+      "step": 884
+    },
+    {
+      "epoch": 2.989864864864865,
+      "grad_norm": 0.33690720796585083,
+      "learning_rate": 4.9115721178297093e-05,
+      "loss": 0.1734,
+      "step": 885
+    },
+    {
+      "epoch": 2.993243243243243,
+      "grad_norm": 0.35074329376220703,
+      "learning_rate": 4.9123895364784184e-05,
+      "loss": 0.1623,
+      "step": 886
+    },
+    {
+      "epoch": 2.9966216216216215,
+      "grad_norm": 0.35647299885749817,
+      "learning_rate": 4.913206033052877e-05,
+      "loss": 0.1647,
+      "step": 887
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.4068041145801544,
+      "learning_rate": 4.914021609631002e-05,
+      "loss": 0.1831,
+      "step": 888
+    },
+    {
+      "epoch": 3.0033783783783785,
+      "grad_norm": 0.4528217017650604,
+      "learning_rate": 4.91483626828369e-05,
+      "loss": 0.128,
+      "step": 889
+    },
+    {
+      "epoch": 3.0067567567567566,
+      "grad_norm": 0.3695124089717865,
+      "learning_rate": 4.915650011074855e-05,
+      "loss": 0.1077,
+      "step": 890
+    },
+    {
+      "epoch": 3.010135135135135,
+      "grad_norm": 0.570489227771759,
+      "learning_rate": 4.916462840061458e-05,
+      "loss": 0.11,
+      "step": 891
+    },
+    {
+      "epoch": 3.0135135135135136,
+      "grad_norm": 0.43204137682914734,
+      "learning_rate": 4.917274757293539e-05,
+      "loss": 0.1032,
+      "step": 892
+    },
+    {
+      "epoch": 3.016891891891892,
+      "grad_norm": 0.5314778089523315,
+      "learning_rate": 4.918085764814244e-05,
+      "loss": 0.1304,
+      "step": 893
+    },
+    {
+      "epoch": 3.02027027027027,
+      "grad_norm": 0.3921876847743988,
+      "learning_rate": 4.9188958646598624e-05,
+      "loss": 0.0949,
+      "step": 894
+    },
+    {
+      "epoch": 3.0236486486486487,
+      "grad_norm": 0.35549548268318176,
+      "learning_rate": 4.919705058859854e-05,
+      "loss": 0.0989,
+      "step": 895
+    },
+    {
+      "epoch": 3.027027027027027,
+      "grad_norm": 0.4394858181476593,
+      "learning_rate": 4.920513349436875e-05,
+      "loss": 0.1152,
+      "step": 896
+    },
+    {
+      "epoch": 3.0304054054054053,
+      "grad_norm": 0.4386696219444275,
+      "learning_rate": 4.92132073840682e-05,
+      "loss": 0.1163,
+      "step": 897
+    },
+    {
+      "epoch": 3.0337837837837838,
+      "grad_norm": 0.38169318437576294,
+      "learning_rate": 4.922127227778841e-05,
+      "loss": 0.1078,
+      "step": 898
+    },
+    {
+      "epoch": 3.0371621621621623,
+      "grad_norm": 0.410858690738678,
+      "learning_rate": 4.9229328195553815e-05,
+      "loss": 0.1165,
+      "step": 899
+    },
+    {
+      "epoch": 3.0405405405405403,
+      "grad_norm": 0.4674147069454193,
+      "learning_rate": 4.923737515732209e-05,
+      "loss": 0.12,
+      "step": 900
+    },
+    {
+      "epoch": 3.043918918918919,
+      "grad_norm": 0.3922157883644104,
+      "learning_rate": 4.924541318298438e-05,
+      "loss": 0.1131,
+      "step": 901
+    },
+    {
+      "epoch": 3.0472972972972974,
+      "grad_norm": 0.3732883632183075,
+      "learning_rate": 4.92534422923657e-05,
+      "loss": 0.1025,
+      "step": 902
+    },
+    {
+      "epoch": 3.050675675675676,
+      "grad_norm": 0.4513222277164459,
+      "learning_rate": 4.9261462505225106e-05,
+      "loss": 0.1207,
+      "step": 903
+    },
+    {
+      "epoch": 3.054054054054054,
+      "grad_norm": 0.39665260910987854,
+      "learning_rate": 4.926947384125606e-05,
+      "loss": 0.1115,
+      "step": 904
+    },
+    {
+      "epoch": 3.0574324324324325,
+      "grad_norm": 0.3784550130367279,
+      "learning_rate": 4.927747632008672e-05,
+      "loss": 0.1151,
+      "step": 905
+    },
+    {
+      "epoch": 3.060810810810811,
+      "grad_norm": 0.44802817702293396,
+      "learning_rate": 4.9285469961280226e-05,
+      "loss": 0.113,
+      "step": 906
+    },
+    {
+      "epoch": 3.064189189189189,
+      "grad_norm": 0.37687918543815613,
+      "learning_rate": 4.9293454784334924e-05,
+      "loss": 0.1157,
+      "step": 907
+    },
+    {
+      "epoch": 3.0675675675675675,
+      "grad_norm": 0.3582609295845032,
+      "learning_rate": 4.9301430808684754e-05,
+      "loss": 0.1031,
+      "step": 908
+    },
+    {
+      "epoch": 3.070945945945946,
+      "grad_norm": 0.3885122239589691,
+      "learning_rate": 4.930939805369946e-05,
+      "loss": 0.1043,
+      "step": 909
+    },
+    {
+      "epoch": 3.074324324324324,
+      "grad_norm": 0.5625414252281189,
+      "learning_rate": 4.93173565386849e-05,
+      "loss": 0.1109,
+      "step": 910
+    },
+    {
+      "epoch": 3.0777027027027026,
+      "grad_norm": 0.3783611059188843,
+      "learning_rate": 4.932530628288331e-05,
+      "loss": 0.1047,
+      "step": 911
+    },
+    {
+      "epoch": 3.081081081081081,
+      "grad_norm": 0.48341429233551025,
+      "learning_rate": 4.933324730547361e-05,
+      "loss": 0.0961,
+      "step": 912
+    },
+    {
+      "epoch": 3.0844594594594597,
+      "grad_norm": 0.399813711643219,
+      "learning_rate": 4.934117962557165e-05,
+      "loss": 0.1178,
+      "step": 913
+    },
+    {
+      "epoch": 3.0878378378378377,
+      "grad_norm": 0.40703117847442627,
+      "learning_rate": 4.9349103262230524e-05,
+      "loss": 0.1179,
+      "step": 914
+    },
+    {
+      "epoch": 3.0912162162162162,
+      "grad_norm": 0.39595040678977966,
+      "learning_rate": 4.935701823444081e-05,
+      "loss": 0.1161,
+      "step": 915
+    },
+    {
+      "epoch": 3.0945945945945947,
+      "grad_norm": 0.3974362313747406,
+      "learning_rate": 4.9364924561130845e-05,
+      "loss": 0.1167,
+      "step": 916
+    },
+    {
+      "epoch": 3.097972972972973,
+      "grad_norm": 0.426070898771286,
+      "learning_rate": 4.937282226116702e-05,
+      "loss": 0.1305,
+      "step": 917
+    },
+    {
+      "epoch": 3.1013513513513513,
+      "grad_norm": 0.38482606410980225,
+      "learning_rate": 4.938071135335405e-05,
+      "loss": 0.1018,
+      "step": 918
+    },
+    {
+      "epoch": 3.10472972972973,
+      "grad_norm": 0.4263748228549957,
+      "learning_rate": 4.938859185643519e-05,
+      "loss": 0.1156,
+      "step": 919
+    },
+    {
+      "epoch": 3.108108108108108,
+      "grad_norm": 0.39638155698776245,
+      "learning_rate": 4.939646378909259e-05,
+      "loss": 0.1157,
+      "step": 920
+    },
+    {
+      "epoch": 3.1114864864864864,
+      "grad_norm": 0.40607044100761414,
+      "learning_rate": 4.940432716994748e-05,
+      "loss": 0.1197,
+      "step": 921
+    },
+    {
+      "epoch": 3.114864864864865,
+      "grad_norm": 0.4135359823703766,
+      "learning_rate": 4.9412182017560496e-05,
+      "loss": 0.1202,
+      "step": 922
+    },
+    {
+      "epoch": 3.1182432432432434,
+      "grad_norm": 0.38252270221710205,
+      "learning_rate": 4.942002835043187e-05,
+      "loss": 0.1126,
+      "step": 923
+    },
+    {
+      "epoch": 3.1216216216216215,
+      "grad_norm": 0.3727289140224457,
+      "learning_rate": 4.942786618700178e-05,
+      "loss": 0.1111,
+      "step": 924
+    },
+    {
+      "epoch": 3.125,
+      "grad_norm": 0.380953848361969,
+      "learning_rate": 4.9435695545650545e-05,
+      "loss": 0.1151,
+      "step": 925
+    },
+    {
+      "epoch": 3.1283783783783785,
+      "grad_norm": 0.37086960673332214,
+      "learning_rate": 4.944351644469891e-05,
+      "loss": 0.1106,
+      "step": 926
+    },
+    {
+      "epoch": 3.1317567567567566,
+      "grad_norm": 0.4138876497745514,
+      "learning_rate": 4.945132890240829e-05,
+      "loss": 0.1239,
+      "step": 927
+    },
+    {
+      "epoch": 3.135135135135135,
+      "grad_norm": 0.4137141704559326,
+      "learning_rate": 4.945913293698104e-05,
+      "loss": 0.1252,
+      "step": 928
+    },
+    {
+      "epoch": 3.1385135135135136,
+      "grad_norm": 0.3904976546764374,
+      "learning_rate": 4.9466928566560696e-05,
+      "loss": 0.107,
+      "step": 929
+    },
+    {
+      "epoch": 3.141891891891892,
+      "grad_norm": 0.3888641595840454,
+      "learning_rate": 4.9474715809232256e-05,
+      "loss": 0.113,
+      "step": 930
+    },
+    {
+      "epoch": 3.14527027027027,
+      "grad_norm": 0.39176711440086365,
+      "learning_rate": 4.948249468302239e-05,
+      "loss": 0.1135,
+      "step": 931
+    },
+    {
+      "epoch": 3.1486486486486487,
+      "grad_norm": 0.3809608817100525,
+      "learning_rate": 4.9490265205899697e-05,
+      "loss": 0.1051,
+      "step": 932
+    },
+    {
+      "epoch": 3.152027027027027,
+      "grad_norm": 0.4019430875778198,
+      "learning_rate": 4.9498027395775006e-05,
+      "loss": 0.121,
+      "step": 933
+    },
+    {
+      "epoch": 3.1554054054054053,
+      "grad_norm": 0.3715710937976837,
+      "learning_rate": 4.950578127050156e-05,
+      "loss": 0.1225,
+      "step": 934
+    },
+    {
+      "epoch": 3.1587837837837838,
+      "grad_norm": 0.4521862864494324,
+      "learning_rate": 4.95135268478753e-05,
+      "loss": 0.1294,
+      "step": 935
+    },
+    {
+      "epoch": 3.1621621621621623,
+      "grad_norm": 0.4279133379459381,
+      "learning_rate": 4.952126414563509e-05,
+      "loss": 0.1229,
+      "step": 936
+    },
+    {
+      "epoch": 3.1655405405405403,
+      "grad_norm": 0.4043607711791992,
+      "learning_rate": 4.952899318146297e-05,
+      "loss": 0.1117,
+      "step": 937
+    },
+    {
+      "epoch": 3.168918918918919,
+      "grad_norm": 0.34410539269447327,
+      "learning_rate": 4.9536713972984414e-05,
+      "loss": 0.0972,
+      "step": 938
+    },
+    {
+      "epoch": 3.1722972972972974,
+      "grad_norm": 0.43061432242393494,
+      "learning_rate": 4.954442653776852e-05,
+      "loss": 0.1333,
+      "step": 939
+    },
+    {
+      "epoch": 3.175675675675676,
+      "grad_norm": 0.36851924657821655,
+      "learning_rate": 4.955213089332832e-05,
+      "loss": 0.1114,
+      "step": 940
+    },
+    {
+      "epoch": 3.179054054054054,
+      "grad_norm": 0.37471747398376465,
+      "learning_rate": 4.955982705712095e-05,
+      "loss": 0.1169,
+      "step": 941
+    },
+    {
+      "epoch": 3.1824324324324325,
+      "grad_norm": 0.35380908846855164,
+      "learning_rate": 4.956751504654796e-05,
+      "loss": 0.1005,
+      "step": 942
+    },
+    {
+      "epoch": 3.185810810810811,
+      "grad_norm": 0.38561439514160156,
+      "learning_rate": 4.957519487895548e-05,
+      "loss": 0.1187,
+      "step": 943
+    },
+    {
+      "epoch": 3.189189189189189,
+      "grad_norm": 0.4120042324066162,
+      "learning_rate": 4.9582866571634485e-05,
+      "loss": 0.1273,
+      "step": 944
+    },
+    {
+      "epoch": 3.1925675675675675,
+      "grad_norm": 0.39863529801368713,
+      "learning_rate": 4.959053014182106e-05,
+      "loss": 0.1125,
+      "step": 945
+    },
+    {
+      "epoch": 3.195945945945946,
+      "grad_norm": 0.38955143094062805,
+      "learning_rate": 4.959818560669655e-05,
+      "loss": 0.1175,
+      "step": 946
+    },
+    {
+      "epoch": 3.199324324324324,
+      "grad_norm": 0.3573088049888611,
+      "learning_rate": 4.96058329833879e-05,
+      "loss": 0.1131,
+      "step": 947
+    },
+    {
+      "epoch": 3.2027027027027026,
+      "grad_norm": 0.4304436147212982,
+      "learning_rate": 4.961347228896777e-05,
+      "loss": 0.1262,
+      "step": 948
+    },
+    {
+      "epoch": 3.206081081081081,
+      "grad_norm": 0.39087799191474915,
+      "learning_rate": 4.962110354045488e-05,
+      "loss": 0.1191,
+      "step": 949
+    },
+    {
+      "epoch": 3.2094594594594597,
+      "grad_norm": 0.3944397568702698,
+      "learning_rate": 4.962872675481414e-05,
+      "loss": 0.1182,
+      "step": 950
+    },
+    {
+      "epoch": 3.2128378378378377,
+      "grad_norm": 0.40278294682502747,
+      "learning_rate": 4.9636341948956906e-05,
+      "loss": 0.1211,
+      "step": 951
+    },
+    {
+      "epoch": 3.2162162162162162,
+      "grad_norm": 0.41054508090019226,
+      "learning_rate": 4.964394913974124e-05,
+      "loss": 0.1195,
+      "step": 952
+    },
+    {
+      "epoch": 3.2195945945945947,
+      "grad_norm": 0.4175059199333191,
+      "learning_rate": 4.965154834397211e-05,
+      "loss": 0.1353,
+      "step": 953
+    },
+    {
+      "epoch": 3.222972972972973,
+      "grad_norm": 0.3865712583065033,
+      "learning_rate": 4.965913957840159e-05,
+      "loss": 0.1111,
+      "step": 954
+    },
+    {
+      "epoch": 3.2263513513513513,
+      "grad_norm": 0.39877578616142273,
+      "learning_rate": 4.966672285972911e-05,
+      "loss": 0.1256,
+      "step": 955
+    },
+    {
+      "epoch": 3.22972972972973,
+      "grad_norm": 0.34512677788734436,
+      "learning_rate": 4.967429820460167e-05,
+      "loss": 0.1078,
+      "step": 956
+    },
+    {
+      "epoch": 3.233108108108108,
+      "grad_norm": 0.4256219267845154,
+      "learning_rate": 4.9681865629614064e-05,
+      "loss": 0.1313,
+      "step": 957
+    },
+    {
+      "epoch": 3.2364864864864864,
+      "grad_norm": 0.46931833028793335,
+      "learning_rate": 4.9689425151309074e-05,
+      "loss": 0.1396,
+      "step": 958
+    },
+    {
+      "epoch": 3.239864864864865,
+      "grad_norm": 0.40905606746673584,
+      "learning_rate": 4.969697678617773e-05,
+      "loss": 0.1242,
+      "step": 959
+    },
+    {
+      "epoch": 3.2432432432432434,
+      "grad_norm": 0.36054709553718567,
+      "learning_rate": 4.970452055065948e-05,
+      "loss": 0.1149,
+      "step": 960
+    },
+    {
+      "epoch": 3.2466216216216215,
+      "grad_norm": 0.3850068747997284,
+      "learning_rate": 4.9712056461142423e-05,
+      "loss": 0.1074,
+      "step": 961
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.42875200510025024,
+      "learning_rate": 4.971958453396355e-05,
+      "loss": 0.1152,
+      "step": 962
+    },
+    {
+      "epoch": 3.2533783783783785,
+      "grad_norm": 0.413173109292984,
+      "learning_rate": 4.972710478540891e-05,
+      "loss": 0.1273,
+      "step": 963
+    },
+    {
+      "epoch": 3.2567567567567566,
+      "grad_norm": 0.3743407428264618,
+      "learning_rate": 4.973461723171385e-05,
+      "loss": 0.1149,
+      "step": 964
+    },
+    {
+      "epoch": 3.260135135135135,
+      "grad_norm": 0.44052329659461975,
+      "learning_rate": 4.9742121889063213e-05,
+      "loss": 0.1301,
+      "step": 965
+    },
+    {
+      "epoch": 3.2635135135135136,
+      "grad_norm": 0.38358744978904724,
+      "learning_rate": 4.974961877359156e-05,
+      "loss": 0.1133,
+      "step": 966
+    },
+    {
+      "epoch": 3.266891891891892,
+      "grad_norm": 0.423977792263031,
+      "learning_rate": 4.975710790138336e-05,
+      "loss": 0.1346,
+      "step": 967
+    },
+    {
+      "epoch": 3.27027027027027,
+      "grad_norm": 1.5824426412582397,
+      "learning_rate": 4.976458928847323e-05,
+      "loss": 0.3884,
+      "step": 968
+    },
+    {
+      "epoch": 3.2736486486486487,
+      "grad_norm": 0.41981515288352966,
+      "learning_rate": 4.977206295084609e-05,
+      "loss": 0.1164,
+      "step": 969
+    },
+    {
+      "epoch": 3.277027027027027,
+      "grad_norm": 0.42901647090911865,
+      "learning_rate": 4.9779528904437424e-05,
+      "loss": 0.1269,
+      "step": 970
+    },
+    {
+      "epoch": 3.2804054054054053,
+      "grad_norm": 0.3595876693725586,
+      "learning_rate": 4.978698716513342e-05,
+      "loss": 0.1121,
+      "step": 971
+    },
+    {
+      "epoch": 3.2837837837837838,
+      "grad_norm": 0.39047256112098694,
+      "learning_rate": 4.9794437748771244e-05,
+      "loss": 0.1219,
+      "step": 972
+    },
+    {
+      "epoch": 3.2871621621621623,
+      "grad_norm": 0.41640281677246094,
+      "learning_rate": 4.9801880671139204e-05,
+      "loss": 0.1303,
+      "step": 973
+    },
+    {
+      "epoch": 3.2905405405405403,
+      "grad_norm": 0.371383398771286,
+      "learning_rate": 4.980931594797693e-05,
+      "loss": 0.112,
+      "step": 974
+    },
+    {
+      "epoch": 3.293918918918919,
+      "grad_norm": 0.4297455847263336,
+      "learning_rate": 4.981674359497562e-05,
+      "loss": 0.1326,
+      "step": 975
+    },
+    {
+      "epoch": 3.2972972972972974,
+      "grad_norm": 0.39308154582977295,
+      "learning_rate": 4.98241636277782e-05,
+      "loss": 0.1244,
+      "step": 976
+    },
+    {
+      "epoch": 3.3006756756756754,
+      "grad_norm": 0.3956491947174072,
+      "learning_rate": 4.983157606197955e-05,
+      "loss": 0.1203,
+      "step": 977
+    },
+    {
+      "epoch": 3.304054054054054,
+      "grad_norm": 0.41044941544532776,
+      "learning_rate": 4.98389809131267e-05,
+      "loss": 0.1247,
+      "step": 978
+    },
+    {
+      "epoch": 3.3074324324324325,
+      "grad_norm": 0.4159061312675476,
+      "learning_rate": 4.984637819671897e-05,
+      "loss": 0.1321,
+      "step": 979
+    },
+    {
+      "epoch": 3.310810810810811,
+      "grad_norm": 0.3757896423339844,
+      "learning_rate": 4.985376792820825e-05,
+      "loss": 0.1157,
+      "step": 980
+    },
+    {
+      "epoch": 3.314189189189189,
+      "grad_norm": 0.3729749321937561,
+      "learning_rate": 4.986115012299915e-05,
+      "loss": 0.1103,
+      "step": 981
+    },
+    {
+      "epoch": 3.3175675675675675,
+      "grad_norm": 0.3994838297367096,
+      "learning_rate": 4.986852479644916e-05,
+      "loss": 0.1284,
+      "step": 982
+    },
+    {
+      "epoch": 3.320945945945946,
+      "grad_norm": 0.3987760841846466,
+      "learning_rate": 4.987589196386893e-05,
+      "loss": 0.1126,
+      "step": 983
+    },
+    {
+      "epoch": 3.3243243243243246,
+      "grad_norm": 0.3847789168357849,
+      "learning_rate": 4.988325164052236e-05,
+      "loss": 0.1199,
+      "step": 984
+    },
+    {
+      "epoch": 3.3277027027027026,
+      "grad_norm": 0.41571831703186035,
+      "learning_rate": 4.9890603841626866e-05,
+      "loss": 0.1295,
+      "step": 985
+    },
+    {
+      "epoch": 3.331081081081081,
+      "grad_norm": 0.3801005184650421,
+      "learning_rate": 4.989794858235352e-05,
+      "loss": 0.1276,
+      "step": 986
+    },
+    {
+      "epoch": 3.3344594594594597,
+      "grad_norm": 0.4302000403404236,
+      "learning_rate": 4.990528587782729e-05,
+      "loss": 0.1355,
+      "step": 987
+    },
+    {
+      "epoch": 3.3378378378378377,
+      "grad_norm": 0.3750097155570984,
+      "learning_rate": 4.9912615743127146e-05,
+      "loss": 0.1159,
+      "step": 988
+    },
+    {
+      "epoch": 3.3412162162162162,
+      "grad_norm": 0.43669402599334717,
+      "learning_rate": 4.991993819328633e-05,
+      "loss": 0.1349,
+      "step": 989
+    },
+    {
+      "epoch": 3.3445945945945947,
+      "grad_norm": 0.40606677532196045,
+      "learning_rate": 4.9927253243292505e-05,
+      "loss": 0.123,
+      "step": 990
+    },
+    {
+      "epoch": 3.347972972972973,
+      "grad_norm": 0.40521538257598877,
+      "learning_rate": 4.993456090808793e-05,
+      "loss": 0.1212,
+      "step": 991
+    },
+    {
+      "epoch": 3.3513513513513513,
+      "grad_norm": 0.38929492235183716,
+      "learning_rate": 4.994186120256965e-05,
+      "loss": 0.1204,
+      "step": 992
+    },
+    {
+      "epoch": 3.35472972972973,
+      "grad_norm": 0.3639843761920929,
+      "learning_rate": 4.9949154141589696e-05,
+      "loss": 0.1175,
+      "step": 993
+    },
+    {
+      "epoch": 3.358108108108108,
+      "grad_norm": 0.3805122971534729,
+      "learning_rate": 4.995643973995523e-05,
+      "loss": 0.1198,
+      "step": 994
+    },
+    {
+      "epoch": 3.3614864864864864,
+      "grad_norm": 0.395796000957489,
+      "learning_rate": 4.9963718012428765e-05,
+      "loss": 0.1348,
+      "step": 995
+    },
+    {
+      "epoch": 3.364864864864865,
+      "grad_norm": 0.3744940459728241,
+      "learning_rate": 4.9970988973728314e-05,
+      "loss": 0.1104,
+      "step": 996
+    },
+    {
+      "epoch": 3.3682432432432434,
+      "grad_norm": 1.3542141914367676,
+      "learning_rate": 4.99782526385276e-05,
+      "loss": 0.3864,
+      "step": 997
+    },
+    {
+      "epoch": 3.3716216216216215,
+      "grad_norm": 0.42771685123443604,
+      "learning_rate": 4.998550902145619e-05,
+      "loss": 0.1383,
+      "step": 998
+    },
+    {
+      "epoch": 3.375,
+      "grad_norm": 0.3889855444431305,
+      "learning_rate": 4.999275813709971e-05,
+      "loss": 0.1318,
+      "step": 999
+    },
+    {
+      "epoch": 3.3783783783783785,
+      "grad_norm": 0.37391164898872375,
+      "learning_rate": 5e-05,
+      "loss": 0.1285,
+      "step": 1000
+    },
+    {
+      "epoch": 3.3817567567567566,
+      "grad_norm": 0.3580280542373657,
+      "learning_rate": 4.9996356488619556e-05,
+      "loss": 0.1125,
+      "step": 1001
+    },
+    {
+      "epoch": 3.385135135135135,
+      "grad_norm": 0.3707447052001953,
+      "learning_rate": 4.9985427016598435e-05,
+      "loss": 0.1221,
+      "step": 1002
+    },
+    {
+      "epoch": 3.3885135135135136,
+      "grad_norm": 0.3772697150707245,
+      "learning_rate": 4.996721476998771e-05,
+      "loss": 0.1215,
+      "step": 1003
+    },
+    {
+      "epoch": 3.391891891891892,
+      "grad_norm": 0.40340831875801086,
+      "learning_rate": 4.9941725057840504e-05,
+      "loss": 0.1233,
+      "step": 1004
+    },
+    {
+      "epoch": 3.39527027027027,
+      "grad_norm": 0.3812180161476135,
+      "learning_rate": 4.9908965310664374e-05,
+      "loss": 0.1149,
+      "step": 1005
+    },
+    {
+      "epoch": 3.3986486486486487,
+      "grad_norm": 0.4367975890636444,
+      "learning_rate": 4.986894507825522e-05,
+      "loss": 0.1358,
+      "step": 1006
+    },
+    {
+      "epoch": 3.402027027027027,
+      "grad_norm": 0.4075244963169098,
+      "learning_rate": 4.9821676026913475e-05,
+      "loss": 0.132,
+      "step": 1007
+    },
+    {
+      "epoch": 3.4054054054054053,
+      "grad_norm": 0.392610639333725,
+      "learning_rate": 4.9767171936043175e-05,
+      "loss": 0.1169,
+      "step": 1008
+    },
+    {
+      "epoch": 3.4087837837837838,
+      "grad_norm": 0.3933659791946411,
+      "learning_rate": 4.970544869413522e-05,
+      "loss": 0.1197,
+      "step": 1009
+    },
+    {
+      "epoch": 3.4121621621621623,
+      "grad_norm": 0.3939339816570282,
+      "learning_rate": 4.963652429413563e-05,
+      "loss": 0.1315,
+      "step": 1010
+    },
+    {
+      "epoch": 3.4155405405405403,
+      "grad_norm": 0.37384697794914246,
+      "learning_rate": 4.9560418828200494e-05,
+      "loss": 0.1187,
+      "step": 1011
+    },
+    {
+      "epoch": 3.418918918918919,
+      "grad_norm": 0.3945925831794739,
+      "learning_rate": 4.9477154481838875e-05,
+      "loss": 0.1353,
+      "step": 1012
+    },
+    {
+      "epoch": 3.4222972972972974,
+      "grad_norm": 0.42826682329177856,
+      "learning_rate": 4.9386755527445475e-05,
+      "loss": 0.1459,
+      "step": 1013
+    },
+    {
+      "epoch": 3.4256756756756754,
+      "grad_norm": 0.4206187427043915,
+      "learning_rate": 4.928924831722504e-05,
+      "loss": 0.1292,
+      "step": 1014
+    },
+    {
+      "epoch": 3.429054054054054,
+      "grad_norm": 0.38580143451690674,
+      "learning_rate": 4.9184661275510446e-05,
+      "loss": 0.1153,
+      "step": 1015
+    },
+    {
+      "epoch": 3.4324324324324325,
+      "grad_norm": 0.393794983625412,
+      "learning_rate": 4.907302489047662e-05,
+      "loss": 0.1179,
+      "step": 1016
+    },
+    {
+      "epoch": 3.435810810810811,
+      "grad_norm": 0.44305288791656494,
+      "learning_rate": 4.895437170525303e-05,
+      "loss": 0.1302,
+      "step": 1017
+    },
+    {
+      "epoch": 3.439189189189189,
+      "grad_norm": 0.4056559205055237,
+      "learning_rate": 4.882873630843699e-05,
+      "loss": 0.1302,
+      "step": 1018
+    },
+    {
+      "epoch": 3.4425675675675675,
+      "grad_norm": 0.40912213921546936,
+      "learning_rate": 4.869615532401074e-05,
+      "loss": 0.1327,
+      "step": 1019
+    },
+    {
+      "epoch": 3.445945945945946,
+      "grad_norm": 0.38370949029922485,
+      "learning_rate": 4.855666740066522e-05,
+      "loss": 0.129,
+      "step": 1020
+    },
+    {
+      "epoch": 3.4493243243243246,
+      "grad_norm": 0.41348859667778015,
+      "learning_rate": 4.841031320053351e-05,
+      "loss": 0.1243,
+      "step": 1021
+    },
+    {
+      "epoch": 3.4527027027027026,
+      "grad_norm": 0.40817150473594666,
+      "learning_rate": 4.825713538733748e-05,
+      "loss": 0.1321,
+      "step": 1022
+    },
+    {
+      "epoch": 3.456081081081081,
+      "grad_norm": 31.2849063873291,
+      "learning_rate": 4.80971786139509e-05,
+      "loss": 0.186,
+      "step": 1023
+    },
+    {
+      "epoch": 3.4594594594594597,
+      "grad_norm": 8.919435501098633,
+      "learning_rate": 4.793048950938256e-05,
+      "loss": 0.6012,
+      "step": 1024
+    },
+    {
+      "epoch": 3.4628378378378377,
+      "grad_norm": 1.2597789764404297,
+      "learning_rate": 4.7757116665183614e-05,
+      "loss": 0.373,
+      "step": 1025
+    },
+    {
+      "epoch": 3.4662162162162162,
+      "grad_norm": 144.5867156982422,
+      "learning_rate": 4.757711062128251e-05,
+      "loss": 0.6334,
+      "step": 1026
+    },
+    {
+      "epoch": 3.4695945945945947,
+      "grad_norm": 42.916805267333984,
+      "learning_rate": 4.739052385125216e-05,
+      "loss": 0.6214,
+      "step": 1027
+    },
+    {
+      "epoch": 3.472972972972973,
+      "grad_norm": 0.7827467322349548,
+      "learning_rate": 4.7197410747013376e-05,
+      "loss": 0.308,
+      "step": 1028
+    },
+    {
+      "epoch": 3.4763513513513513,
+      "grad_norm": 0.7671857476234436,
+      "learning_rate": 4.6997827602979024e-05,
+      "loss": 0.3552,
+      "step": 1029
+    },
+    {
+      "epoch": 3.47972972972973,
+      "grad_norm": 3.2267792224884033,
+      "learning_rate": 4.67918325996437e-05,
+      "loss": 0.3405,
+      "step": 1030
+    },
+    {
+      "epoch": 3.483108108108108,
+      "grad_norm": 13.3894624710083,
+      "learning_rate": 4.6579485786623475e-05,
+      "loss": 0.4361,
+      "step": 1031
+    },
+    {
+      "epoch": 3.4864864864864864,
+      "grad_norm": 2.525721311569214,
+      "learning_rate": 4.636084906515085e-05,
+      "loss": 0.1359,
+      "step": 1032
+    },
+    {
+      "epoch": 3.489864864864865,
+      "grad_norm": 52.08081817626953,
+      "learning_rate": 4.6135986170029947e-05,
+      "loss": 0.5506,
+      "step": 1033
+    },
+    {
+      "epoch": 3.4932432432432434,
+      "grad_norm": 1.6132444143295288,
+      "learning_rate": 4.5904962651057134e-05,
+      "loss": 0.1839,
+      "step": 1034
+    },
+    {
+      "epoch": 3.4966216216216215,
+      "grad_norm": 1.116671085357666,
+      "learning_rate": 4.566784585391263e-05,
+      "loss": 0.1534,
+      "step": 1035
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.6348862051963806,
+      "learning_rate": 4.542470490052853e-05,
+      "loss": 0.1429,
+      "step": 1036
+    },
+    {
+      "epoch": 3.5033783783783785,
+      "grad_norm": 0.6647199392318726,
+      "learning_rate": 4.517561066893909e-05,
+      "loss": 0.1351,
+      "step": 1037
+    },
+    {
+      "epoch": 3.506756756756757,
+      "grad_norm": 0.4823889136314392,
+      "learning_rate": 4.492063577261908e-05,
+      "loss": 0.1312,
+      "step": 1038
+    },
+    {
+      "epoch": 3.510135135135135,
+      "grad_norm": 0.48596346378326416,
+      "learning_rate": 4.4659854539316174e-05,
+      "loss": 0.1418,
+      "step": 1039
+    },
+    {
+      "epoch": 3.5135135135135136,
+      "grad_norm": 0.7251044511795044,
+      "learning_rate": 4.439334298938374e-05,
+      "loss": 0.1465,
+      "step": 1040
+    },
+    {
+      "epoch": 3.516891891891892,
+      "grad_norm": 0.4920099973678589,
+      "learning_rate": 4.4121178813620046e-05,
+      "loss": 0.1481,
+      "step": 1041
+    },
+    {
+      "epoch": 3.52027027027027,
+      "grad_norm": 0.42203307151794434,
+      "learning_rate": 4.384344135062071e-05,
+      "loss": 0.1219,
+      "step": 1042
+    },
+    {
+      "epoch": 3.5236486486486487,
+      "grad_norm": 0.47877076268196106,
+      "learning_rate": 4.3560211563650635e-05,
+      "loss": 0.1282,
+      "step": 1043
+    },
+    {
+      "epoch": 3.527027027027027,
+      "grad_norm": 0.4383416175842285,
+      "learning_rate": 4.327157201704241e-05,
+      "loss": 0.133,
+      "step": 1044
+    },
+    {
+      "epoch": 3.5304054054054053,
+      "grad_norm": 0.49161508679389954,
+      "learning_rate": 4.297760685212801e-05,
+      "loss": 0.1268,
+      "step": 1045
+    },
+    {
+      "epoch": 3.5337837837837838,
+      "grad_norm": 0.44632676243782043,
+      "learning_rate": 4.2678401762710726e-05,
+      "loss": 0.1304,
+      "step": 1046
+    },
+    {
+      "epoch": 3.5371621621621623,
+      "grad_norm": 0.48019739985466003,
+      "learning_rate": 4.237404397008455e-05,
+      "loss": 0.134,
+      "step": 1047
+    },
+    {
+      "epoch": 3.5405405405405403,
+      "grad_norm": 0.3879307508468628,
+      "learning_rate": 4.206462219760831e-05,
+      "loss": 0.1106,
+      "step": 1048
+    },
+    {
+      "epoch": 3.543918918918919,
+      "grad_norm": 8.76919937133789,
+      "learning_rate": 4.175022664484191e-05,
+      "loss": 0.1268,
+      "step": 1049
+    },
+    {
+      "epoch": 3.5472972972972974,
+      "grad_norm": 0.5539897680282593,
+      "learning_rate": 4.143094896125227e-05,
+      "loss": 0.134,
+      "step": 1050
+    },
+    {
+      "epoch": 3.5506756756756754,
+      "grad_norm": 0.41708147525787354,
+      "learning_rate": 4.1106882219496535e-05,
+      "loss": 0.1185,
+      "step": 1051
+    },
+    {
+      "epoch": 3.554054054054054,
+      "grad_norm": 0.7360028028488159,
+      "learning_rate": 4.077812088829051e-05,
+      "loss": 0.1251,
+      "step": 1052
+    },
+    {
+      "epoch": 3.5574324324324325,
+      "grad_norm": 0.49543508887290955,
+      "learning_rate": 4.044476080487005e-05,
+      "loss": 0.1405,
+      "step": 1053
+    },
+    {
+      "epoch": 3.560810810810811,
+      "grad_norm": 0.40071457624435425,
+      "learning_rate": 4.0106899147053564e-05,
+      "loss": 0.1162,
+      "step": 1054
+    },
+    {
+      "epoch": 3.564189189189189,
+      "grad_norm": 0.6165151596069336,
+      "learning_rate": 3.976463440491367e-05,
+      "loss": 0.123,
+      "step": 1055
+    },
+    {
+      "epoch": 3.5675675675675675,
+      "grad_norm": 0.4160919487476349,
+      "learning_rate": 3.941806635206639e-05,
+      "loss": 0.1276,
+      "step": 1056
+    },
+    {
+      "epoch": 3.570945945945946,
+      "grad_norm": 0.4072069525718689,
+      "learning_rate": 3.9067296016586105e-05,
+      "loss": 0.1336,
+      "step": 1057
+    },
+    {
+      "epoch": 3.5743243243243246,
+      "grad_norm": 0.4126531183719635,
+      "learning_rate": 3.871242565155485e-05,
+      "loss": 0.1267,
+      "step": 1058
+    },
+    {
+      "epoch": 3.5777027027027026,
+      "grad_norm": 2.380211353302002,
+      "learning_rate": 3.835355870525455e-05,
+      "loss": 0.4067,
+      "step": 1059
+    },
+    {
+      "epoch": 3.581081081081081,
+      "grad_norm": 0.3900314271450043,
+      "learning_rate": 3.799079979101075e-05,
+      "loss": 0.1231,
+      "step": 1060
+    },
+    {
+      "epoch": 3.5844594594594597,
+      "grad_norm": 0.4255238473415375,
+      "learning_rate": 3.7624254656696864e-05,
+      "loss": 0.1405,
+      "step": 1061
+    },
+    {
+      "epoch": 3.5878378378378377,
+      "grad_norm": 0.36614516377449036,
+      "learning_rate": 3.7254030153907494e-05,
+      "loss": 0.118,
+      "step": 1062
+    },
+    {
+      "epoch": 3.5912162162162162,
+      "grad_norm": 0.40764889121055603,
+      "learning_rate": 3.688023420681019e-05,
+      "loss": 0.127,
+      "step": 1063
+    },
+    {
+      "epoch": 3.5945945945945947,
+      "grad_norm": 0.35974109172821045,
+      "learning_rate": 3.650297578068448e-05,
+      "loss": 0.1238,
+      "step": 1064
+    },
+    {
+      "epoch": 3.597972972972973,
+      "grad_norm": 0.4197885990142822,
+      "learning_rate": 3.6122364850157326e-05,
+      "loss": 0.1342,
+      "step": 1065
+    },
+    {
+      "epoch": 3.6013513513513513,
+      "grad_norm": 0.3706846237182617,
+      "learning_rate": 3.573851236714447e-05,
+      "loss": 0.1108,
+      "step": 1066
+    },
+    {
+      "epoch": 3.60472972972973,
+      "grad_norm": 0.3801131844520569,
+      "learning_rate": 3.535153022850684e-05,
+      "loss": 0.1279,
+      "step": 1067
+    },
+    {
+      "epoch": 3.608108108108108,
+      "grad_norm": 0.35572677850723267,
+      "learning_rate": 3.496153124343142e-05,
+      "loss": 0.1263,
+      "step": 1068
+    },
+    {
+      "epoch": 3.6114864864864864,
+      "grad_norm": 0.3624805510044098,
+      "learning_rate": 3.4568629100546333e-05,
+      "loss": 0.1185,
+      "step": 1069
+    },
+    {
+      "epoch": 3.614864864864865,
+      "grad_norm": 0.36778637766838074,
+      "learning_rate": 3.417293833477938e-05,
+      "loss": 0.1222,
+      "step": 1070
+    },
+    {
+      "epoch": 3.618243243243243,
+      "grad_norm": 0.34212857484817505,
+      "learning_rate": 3.377457429397001e-05,
+      "loss": 0.1196,
+      "step": 1071
+    },
+    {
+      "epoch": 3.6216216216216215,
+      "grad_norm": 0.3891725540161133,
+      "learning_rate": 3.337365310524423e-05,
+      "loss": 0.1291,
+      "step": 1072
+    },
+    {
+      "epoch": 3.625,
+      "grad_norm": 0.399291068315506,
+      "learning_rate": 3.2970291641162396e-05,
+      "loss": 0.1267,
+      "step": 1073
+    },
+    {
+      "epoch": 3.6283783783783785,
+      "grad_norm": 0.346253901720047,
+      "learning_rate": 3.25646074856497e-05,
+      "loss": 0.1207,
+      "step": 1074
+    },
+    {
+      "epoch": 3.631756756756757,
+      "grad_norm": 0.7178367376327515,
+      "learning_rate": 3.2156718899719216e-05,
+      "loss": 0.372,
+      "step": 1075
+    },
+    {
+      "epoch": 3.635135135135135,
+      "grad_norm": 0.41096293926239014,
+      "learning_rate": 3.174674478699772e-05,
+      "loss": 0.1273,
+      "step": 1076
+    },
+    {
+      "epoch": 3.6385135135135136,
+      "grad_norm": 0.3635697066783905,
+      "learning_rate": 3.133480465906398e-05,
+      "loss": 0.1256,
+      "step": 1077
+    },
+    {
+      "epoch": 3.641891891891892,
+      "grad_norm": 0.3569808006286621,
+      "learning_rate": 3.092101860061e-05,
+      "loss": 0.1334,
+      "step": 1078
+    },
+    {
+      "epoch": 3.64527027027027,
+      "grad_norm": 0.339813768863678,
+      "learning_rate": 3.0505507234435122e-05,
+      "loss": 0.111,
+      "step": 1079
+    },
+    {
+      "epoch": 3.6486486486486487,
+      "grad_norm": 0.4040168225765228,
+      "learning_rate": 3.0088391686283214e-05,
+      "loss": 0.1348,
+      "step": 1080
+    },
+    {
+      "epoch": 3.652027027027027,
+      "grad_norm": 0.37235626578330994,
+      "learning_rate": 2.966979354953336e-05,
+      "loss": 0.1262,
+      "step": 1081
+    },
+    {
+      "epoch": 3.6554054054054053,
+      "grad_norm": 0.3623943030834198,
+      "learning_rate": 2.92498348497541e-05,
+      "loss": 0.1185,
+      "step": 1082
+    },
+    {
+      "epoch": 3.6587837837837838,
+      "grad_norm": 0.36152735352516174,
+      "learning_rate": 2.882863800913182e-05,
+      "loss": 0.1248,
+      "step": 1083
+    },
+    {
+      "epoch": 3.6621621621621623,
+      "grad_norm": 0.38995814323425293,
+      "learning_rate": 2.8406325810783425e-05,
+      "loss": 0.1282,
+      "step": 1084
+    },
+    {
+      "epoch": 3.6655405405405403,
+      "grad_norm": 0.35114696621894836,
+      "learning_rate": 2.798302136296379e-05,
+      "loss": 0.1098,
+      "step": 1085
+    },
+    {
+      "epoch": 3.668918918918919,
+      "grad_norm": 0.35527390241622925,
+      "learning_rate": 2.7558848063178506e-05,
+      "loss": 0.1129,
+      "step": 1086
+    },
+    {
+      "epoch": 3.6722972972972974,
+      "grad_norm": 0.3661573827266693,
+      "learning_rate": 2.713392956221225e-05,
+      "loss": 0.1187,
+      "step": 1087
+    },
+    {
+      "epoch": 3.6756756756756754,
+      "grad_norm": 0.3659035265445709,
+      "learning_rate": 2.6708389728083372e-05,
+      "loss": 0.1311,
+      "step": 1088
+    },
+    {
+      "epoch": 3.679054054054054,
+      "grad_norm": 0.36093148589134216,
+      "learning_rate": 2.6282352609935028e-05,
+      "loss": 0.1294,
+      "step": 1089
+    },
+    {
+      "epoch": 3.6824324324324325,
+      "grad_norm": 0.36720040440559387,
+      "learning_rate": 2.5855942401873734e-05,
+      "loss": 0.128,
+      "step": 1090
+    },
+    {
+      "epoch": 3.685810810810811,
+      "grad_norm": 0.40525415539741516,
+      "learning_rate": 2.5429283406765436e-05,
+      "loss": 0.1373,
+      "step": 1091
+    },
+    {
+      "epoch": 3.689189189189189,
+      "grad_norm": 0.3309617042541504,
+      "learning_rate": 2.50025e-05,
+      "loss": 0.115,
+      "step": 1092
+    },
+    {
+      "epoch": 3.6925675675675675,
+      "grad_norm": 0.3174631893634796,
+      "learning_rate": 2.4575716593234574e-05,
+      "loss": 0.105,
+      "step": 1093
+    },
+    {
+      "epoch": 3.695945945945946,
+      "grad_norm": 0.36353635787963867,
+      "learning_rate": 2.414905759812627e-05,
+      "loss": 0.1215,
+      "step": 1094
+    },
+    {
+      "epoch": 3.6993243243243246,
+      "grad_norm": 0.3367186188697815,
+      "learning_rate": 2.372264739006497e-05,
+      "loss": 0.1121,
+      "step": 1095
+    },
+    {
+      "epoch": 3.7027027027027026,
+      "grad_norm": 0.3455405831336975,
+      "learning_rate": 2.329661027191664e-05,
+      "loss": 0.1142,
+      "step": 1096
+    },
+    {
+      "epoch": 3.706081081081081,
+      "grad_norm": 0.3431136906147003,
+      "learning_rate": 2.287107043778775e-05,
+      "loss": 0.1132,
+      "step": 1097
+    },
+    {
+      "epoch": 3.7094594594594597,
+      "grad_norm": 0.34689489006996155,
+      "learning_rate": 2.2446151936821504e-05,
+      "loss": 0.1129,
+      "step": 1098
+    },
+    {
+      "epoch": 3.7128378378378377,
+      "grad_norm": 0.3749043941497803,
+      "learning_rate": 2.2021978637036214e-05,
+      "loss": 0.1268,
+      "step": 1099
+    },
+    {
+      "epoch": 3.7162162162162162,
+      "grad_norm": 0.34978923201560974,
+      "learning_rate": 2.1598674189216575e-05,
+      "loss": 0.1285,
+      "step": 1100
+    },
+    {
+      "epoch": 3.7195945945945947,
+      "grad_norm": 0.3356480300426483,
+      "learning_rate": 2.117636199086818e-05,
+      "loss": 0.116,
+      "step": 1101
+    },
+    {
+      "epoch": 3.722972972972973,
+      "grad_norm": 0.3172987699508667,
+      "learning_rate": 2.0755165150245906e-05,
+      "loss": 0.1137,
+      "step": 1102
+    },
+    {
+      "epoch": 3.7263513513513513,
+      "grad_norm": 0.34791630506515503,
+      "learning_rate": 2.0335206450466658e-05,
+      "loss": 0.1153,
+      "step": 1103
+    },
+    {
+      "epoch": 3.72972972972973,
+      "grad_norm": 0.3284004330635071,
+      "learning_rate": 1.9916608313716786e-05,
+      "loss": 0.1131,
+      "step": 1104
+    },
+    {
+      "epoch": 3.733108108108108,
+      "grad_norm": 0.3191678822040558,
+      "learning_rate": 1.9499492765564878e-05,
+      "loss": 0.1123,
+      "step": 1105
+    },
+    {
+      "epoch": 3.7364864864864864,
+      "grad_norm": 0.33098334074020386,
+      "learning_rate": 1.9083981399389997e-05,
+      "loss": 0.1156,
+      "step": 1106
+    },
+    {
+      "epoch": 3.739864864864865,
+      "grad_norm": 0.3604113459587097,
+      "learning_rate": 1.8670195340936026e-05,
+      "loss": 0.1164,
+      "step": 1107
+    },
+    {
+      "epoch": 3.743243243243243,
+      "grad_norm": 0.33711400628089905,
+      "learning_rate": 1.825825521300229e-05,
+      "loss": 0.1077,
+      "step": 1108
+    },
+    {
+      "epoch": 3.7466216216216215,
+      "grad_norm": 0.3241155743598938,
+      "learning_rate": 1.7848281100280788e-05,
+      "loss": 0.1053,
+      "step": 1109
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.33422669768333435,
+      "learning_rate": 1.7440392514350303e-05,
+      "loss": 0.1226,
+      "step": 1110
+    },
+    {
+      "epoch": 3.7533783783783785,
+      "grad_norm": 0.30073583126068115,
+      "learning_rate": 1.7034708358837607e-05,
+      "loss": 0.0983,
+      "step": 1111
+    },
+    {
+      "epoch": 3.756756756756757,
+      "grad_norm": 0.37151581048965454,
+      "learning_rate": 1.6631346894755773e-05,
+      "loss": 0.1274,
+      "step": 1112
+    },
+    {
+      "epoch": 3.760135135135135,
+      "grad_norm": 0.3346865773200989,
+      "learning_rate": 1.6230425706029996e-05,
+      "loss": 0.1203,
+      "step": 1113
+    },
+    {
+      "epoch": 3.7635135135135136,
+      "grad_norm": 0.4401615560054779,
+      "learning_rate": 1.5832061665220622e-05,
+      "loss": 0.1122,
+      "step": 1114
+    },
+    {
+      "epoch": 3.766891891891892,
+      "grad_norm": 0.3216763436794281,
+      "learning_rate": 1.5436370899453666e-05,
+      "loss": 0.1139,
+      "step": 1115
+    },
+    {
+      "epoch": 3.77027027027027,
+      "grad_norm": 0.34183549880981445,
+      "learning_rate": 1.504346875656858e-05,
+      "loss": 0.1195,
+      "step": 1116
+    },
+    {
+      "epoch": 3.7736486486486487,
+      "grad_norm": 0.3108338415622711,
+      "learning_rate": 1.4653469771493166e-05,
+      "loss": 0.1018,
+      "step": 1117
+    },
+    {
+      "epoch": 3.777027027027027,
+      "grad_norm": 0.3019558787345886,
+      "learning_rate": 1.426648763285553e-05,
+      "loss": 0.1044,
+      "step": 1118
+    },
+    {
+      "epoch": 3.7804054054054053,
+      "grad_norm": 0.30197712779045105,
+      "learning_rate": 1.3882635149842685e-05,
+      "loss": 0.0976,
+      "step": 1119
+    },
+    {
+      "epoch": 3.7837837837837838,
+      "grad_norm": 15.628427505493164,
+      "learning_rate": 1.3502024219315524e-05,
+      "loss": 0.1159,
+      "step": 1120
+    },
+    {
+      "epoch": 3.7871621621621623,
+      "grad_norm": 0.32293814420700073,
+      "learning_rate": 1.3124765793189808e-05,
+      "loss": 0.108,
+      "step": 1121
+    },
+    {
+      "epoch": 3.7905405405405403,
+      "grad_norm": 0.31840014457702637,
+      "learning_rate": 1.2750969846092514e-05,
+      "loss": 0.1011,
+      "step": 1122
+    },
+    {
+      "epoch": 3.793918918918919,
+      "grad_norm": 0.33972540497779846,
+      "learning_rate": 1.2380745343303146e-05,
+      "loss": 0.115,
+      "step": 1123
+    },
+    {
+      "epoch": 3.7972972972972974,
+      "grad_norm": 0.3430902063846588,
+      "learning_rate": 1.2014200208989255e-05,
+      "loss": 0.1136,
+      "step": 1124
+    },
+    {
+      "epoch": 3.8006756756756754,
+      "grad_norm": 0.34559592604637146,
+      "learning_rate": 1.165144129474546e-05,
+      "loss": 0.1172,
+      "step": 1125
+    },
+    {
+      "epoch": 3.804054054054054,
+      "grad_norm": 0.33134159445762634,
+      "learning_rate": 1.1292574348445151e-05,
+      "loss": 0.1107,
+      "step": 1126
+    },
+    {
+      "epoch": 3.8074324324324325,
+      "grad_norm": 0.2891589403152466,
+      "learning_rate": 1.0937703983413898e-05,
+      "loss": 0.0962,
+      "step": 1127
+    },
+    {
+      "epoch": 3.810810810810811,
+      "grad_norm": 0.34017154574394226,
+      "learning_rate": 1.058693364793361e-05,
+      "loss": 0.1168,
+      "step": 1128
+    },
+    {
+      "epoch": 3.814189189189189,
+      "grad_norm": 0.3454993963241577,
+      "learning_rate": 1.0240365595086335e-05,
+      "loss": 0.1167,
+      "step": 1129
+    },
+    {
+      "epoch": 3.8175675675675675,
+      "grad_norm": 0.33566930890083313,
+      "learning_rate": 9.898100852946441e-06,
+      "loss": 0.1209,
+      "step": 1130
+    },
+    {
+      "epoch": 3.820945945945946,
+      "grad_norm": 0.31290584802627563,
+      "learning_rate": 9.560239195129949e-06,
+      "loss": 0.1062,
+      "step": 1131
+    },
+    {
+      "epoch": 3.8243243243243246,
+      "grad_norm": 0.33595094084739685,
+      "learning_rate": 9.226879111709494e-06,
+      "loss": 0.1091,
+      "step": 1132
+    },
+    {
+      "epoch": 3.8277027027027026,
+      "grad_norm": 0.3021032512187958,
+      "learning_rate": 8.898117780503471e-06,
+      "loss": 0.1029,
+      "step": 1133
+    },
+    {
+      "epoch": 3.831081081081081,
+      "grad_norm": 0.3187675476074219,
+      "learning_rate": 8.574051038747738e-06,
+      "loss": 0.1121,
+      "step": 1134
+    },
+    {
+      "epoch": 3.8344594594594597,
+      "grad_norm": 0.2892165780067444,
+      "learning_rate": 8.254773355158088e-06,
+      "loss": 0.1007,
+      "step": 1135
+    },
+    {
+      "epoch": 3.8378378378378377,
+      "grad_norm": 0.33364662528038025,
+      "learning_rate": 7.94037780239169e-06,
+      "loss": 0.1146,
+      "step": 1136
+    },
+    {
+      "epoch": 3.8412162162162162,
+      "grad_norm": 0.32407206296920776,
+      "learning_rate": 7.630956029915455e-06,
+      "loss": 0.1016,
+      "step": 1137
+    },
+    {
+      "epoch": 3.8445945945945947,
+      "grad_norm": 0.29210153222084045,
+      "learning_rate": 7.32659823728928e-06,
+      "loss": 0.0989,
+      "step": 1138
+    },
+    {
+      "epoch": 3.847972972972973,
+      "grad_norm": 0.3027746081352234,
+      "learning_rate": 7.027393147871994e-06,
+      "loss": 0.1031,
+      "step": 1139
+    },
+    {
+      "epoch": 3.8513513513513513,
+      "grad_norm": 0.34020087122917175,
+      "learning_rate": 6.733427982957593e-06,
+      "loss": 0.1105,
+      "step": 1140
+    },
+    {
+      "epoch": 3.85472972972973,
+      "grad_norm": 0.3309917747974396,
+      "learning_rate": 6.444788436349374e-06,
+      "loss": 0.1155,
+      "step": 1141
+    },
+    {
+      "epoch": 3.858108108108108,
+      "grad_norm": 0.3191674053668976,
+      "learning_rate": 6.16155864937929e-06,
+      "loss": 0.1113,
+      "step": 1142
+    },
+    {
+      "epoch": 3.8614864864864864,
+      "grad_norm": 0.3304139971733093,
+      "learning_rate": 5.883821186379955e-06,
+      "loss": 0.1113,
+      "step": 1143
+    },
+    {
+      "epoch": 3.864864864864865,
+      "grad_norm": 0.3479062616825104,
+      "learning_rate": 5.6116570106162695e-06,
+      "loss": 0.1173,
+      "step": 1144
+    },
+    {
+      "epoch": 3.868243243243243,
+      "grad_norm": 0.3519810140132904,
+      "learning_rate": 5.345145460683825e-06,
+      "loss": 0.1076,
+      "step": 1145
+    },
+    {
+      "epoch": 3.8716216216216215,
+      "grad_norm": 0.4121153652667999,
+      "learning_rate": 5.0843642273809276e-06,
+      "loss": 0.1073,
+      "step": 1146
+    },
+    {
+      "epoch": 3.875,
+      "grad_norm": 0.3174130618572235,
+      "learning_rate": 4.82938933106091e-06,
+      "loss": 0.1016,
+      "step": 1147
+    },
+    {
+      "epoch": 3.8783783783783785,
+      "grad_norm": 0.2952731251716614,
+      "learning_rate": 4.5802950994714764e-06,
+      "loss": 0.1038,
+      "step": 1148
+    },
+    {
+      "epoch": 3.881756756756757,
+      "grad_norm": 0.304749071598053,
+      "learning_rate": 4.337154146087377e-06,
+      "loss": 0.1083,
+      "step": 1149
+    },
+    {
+      "epoch": 3.885135135135135,
+      "grad_norm": 0.3059677481651306,
+      "learning_rate": 4.100037348942866e-06,
+      "loss": 0.099,
+      "step": 1150
+    },
+    {
+      "epoch": 3.8885135135135136,
+      "grad_norm": 0.3029370605945587,
+      "learning_rate": 3.869013829970057e-06,
+      "loss": 0.1017,
+      "step": 1151
+    },
+    {
+      "epoch": 3.891891891891892,
+      "grad_norm": 0.28835681080818176,
+      "learning_rate": 3.644150934849158e-06,
+      "loss": 0.1,
+      "step": 1152
+    },
+    {
+      "epoch": 3.89527027027027,
+      "grad_norm": 0.2965335249900818,
+      "learning_rate": 3.425514213376533e-06,
+      "loss": 0.1023,
+      "step": 1153
+    },
+    {
+      "epoch": 3.8986486486486487,
+      "grad_norm": 0.2934282422065735,
+      "learning_rate": 3.213167400356303e-06,
+      "loss": 0.0937,
+      "step": 1154
+    },
+    {
+      "epoch": 3.902027027027027,
+      "grad_norm": 0.32294800877571106,
+      "learning_rate": 3.007172397020972e-06,
+      "loss": 0.1108,
+      "step": 1155
+    },
+    {
+      "epoch": 3.9054054054054053,
+      "grad_norm": 0.2884935438632965,
+      "learning_rate": 2.8075892529866257e-06,
+      "loss": 0.1025,
+      "step": 1156
+    },
+    {
+      "epoch": 3.9087837837837838,
+      "grad_norm": 0.31093308329582214,
+      "learning_rate": 2.6144761487478416e-06,
+      "loss": 0.107,
+      "step": 1157
+    },
+    {
+      "epoch": 3.9121621621621623,
+      "grad_norm": 0.3277430236339569,
+      "learning_rate": 2.4278893787174935e-06,
+      "loss": 0.1216,
+      "step": 1158
+    },
+    {
+      "epoch": 3.9155405405405403,
+      "grad_norm": 0.467430979013443,
+      "learning_rate": 2.2478833348163886e-06,
+      "loss": 0.3528,
+      "step": 1159
+    },
+    {
+      "epoch": 3.918918918918919,
+      "grad_norm": 0.2848818302154541,
+      "learning_rate": 2.074510490617438e-06,
+      "loss": 0.0981,
+      "step": 1160
+    },
+    {
+      "epoch": 3.9222972972972974,
+      "grad_norm": 0.2947328984737396,
+      "learning_rate": 1.9078213860491097e-06,
+      "loss": 0.1003,
+      "step": 1161
+    },
+    {
+      "epoch": 3.9256756756756754,
+      "grad_norm": 0.28310471773147583,
+      "learning_rate": 1.7478646126625187e-06,
+      "loss": 0.096,
+      "step": 1162
+    },
+    {
+      "epoch": 3.929054054054054,
+      "grad_norm": 0.31671369075775146,
+      "learning_rate": 1.5946867994665007e-06,
+      "loss": 0.1062,
+      "step": 1163
+    },
+    {
+      "epoch": 3.9324324324324325,
+      "grad_norm": 0.2964227795600891,
+      "learning_rate": 1.4483325993347872e-06,
+      "loss": 0.102,
+      "step": 1164
+    },
+    {
+      "epoch": 3.935810810810811,
+      "grad_norm": 0.309048593044281,
+      "learning_rate": 1.308844675989258e-06,
+      "loss": 0.1049,
+      "step": 1165
+    },
+    {
+      "epoch": 3.939189189189189,
+      "grad_norm": 0.31580665707588196,
+      "learning_rate": 1.1762636915630122e-06,
+      "loss": 0.1019,
+      "step": 1166
+    },
+    {
+      "epoch": 3.9425675675675675,
+      "grad_norm": 0.27929919958114624,
+      "learning_rate": 1.0506282947469768e-06,
+      "loss": 0.0962,
+      "step": 1167
+    },
+    {
+      "epoch": 3.945945945945946,
+      "grad_norm": 0.44457200169563293,
+      "learning_rate": 9.319751095233885e-07,
+      "loss": 0.3502,
+      "step": 1168
+    },
+    {
+      "epoch": 3.9493243243243246,
+      "grad_norm": 0.26814886927604675,
+      "learning_rate": 8.203387244895631e-07,
+      "loss": 0.0968,
+      "step": 1169
+    },
+    {
+      "epoch": 3.9527027027027026,
+      "grad_norm": 0.31180819869041443,
+      "learning_rate": 7.157516827749585e-07,
+      "loss": 0.1097,
+      "step": 1170
+    },
+    {
+      "epoch": 3.956081081081081,
+      "grad_norm": 0.30869176983833313,
+      "learning_rate": 6.182444725545342e-07,
+      "loss": 0.1067,
+      "step": 1171
+    },
+    {
+      "epoch": 3.9594594594594597,
+      "grad_norm": 0.29772305488586426,
+      "learning_rate": 5.278455181611357e-07,
+      "loss": 0.1061,
+      "step": 1172
+    },
+    {
+      "epoch": 3.9628378378378377,
+      "grad_norm": 0.3115202784538269,
+      "learning_rate": 4.445811717995063e-07,
+      "loss": 0.1081,
+      "step": 1173
+    },
+    {
+      "epoch": 3.9662162162162162,
+      "grad_norm": 0.31144705414772034,
+      "learning_rate": 3.684757058643714e-07,
+      "loss": 0.097,
+      "step": 1174
+    },
+    {
+      "epoch": 3.9695945945945947,
+      "grad_norm": 0.3056892156600952,
+      "learning_rate": 2.995513058647855e-07,
+      "loss": 0.1069,
+      "step": 1175
+    },
+    {
+      "epoch": 3.972972972972973,
+      "grad_norm": 0.327035129070282,
+      "learning_rate": 2.3782806395682474e-07,
+      "loss": 0.1168,
+      "step": 1176
+    },
+    {
+      "epoch": 3.9763513513513513,
+      "grad_norm": 0.31039026379585266,
+      "learning_rate": 1.8332397308652485e-07,
+      "loss": 0.105,
+      "step": 1177
+    },
+    {
+      "epoch": 3.97972972972973,
+      "grad_norm": 0.2924991548061371,
+      "learning_rate": 1.3605492174477425e-07,
+      "loss": 0.0984,
+      "step": 1178
+    },
+    {
+      "epoch": 3.983108108108108,
+      "grad_norm": 0.2952897250652313,
+      "learning_rate": 9.603468933562955e-08,
+      "loss": 0.1039,
+      "step": 1179
+    },
+    {
+      "epoch": 3.9864864864864864,
+      "grad_norm": 0.28377214074134827,
+      "learning_rate": 6.32749421594948e-08,
+      "loss": 0.098,
+      "step": 1180
+    },
+    {
+      "epoch": 3.989864864864865,
+      "grad_norm": 0.3225543200969696,
+      "learning_rate": 3.778523001229054e-08,
+      "loss": 0.1196,
+      "step": 1181
+    },
+    {
+      "epoch": 3.993243243243243,
+      "grad_norm": 0.3456767201423645,
+      "learning_rate": 1.957298340156484e-08,
+      "loss": 0.1071,
+      "step": 1182
+    },
+    {
+      "epoch": 3.9966216216216215,
+      "grad_norm": 0.30064505338668823,
+      "learning_rate": 8.64351138044836e-09,
+      "loss": 0.1043,
+      "step": 1183
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.30672967433929443,
+      "learning_rate": 5e-09,
+      "loss": 0.0813,
+      "step": 1184
+    },
+    {
+      "epoch": 4.0,
+      "step": 1184,
+      "total_flos": 1.048251868267099e+19,
+      "train_loss": 0.24899632067771982,
+      "train_runtime": 10196.9918,
+      "train_samples_per_second": 2.78,
+      "train_steps_per_second": 0.116
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1184,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 250,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.048251868267099e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}