diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14869 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9964614295824488,
+  "eval_steps": 500,
+  "global_step": 2118,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0014154281670205238,
+      "grad_norm": 52.03026133766843,
+      "learning_rate": 0.0,
+      "loss": 11.3104,
+      "step": 1
+    },
+    {
+      "epoch": 0.0028308563340410475,
+      "grad_norm": 44.502051688198414,
+      "learning_rate": 2.3584905660377358e-07,
+      "loss": 11.4795,
+      "step": 2
+    },
+    {
+      "epoch": 0.004246284501061571,
+      "grad_norm": 45.159850822187344,
+      "learning_rate": 4.7169811320754717e-07,
+      "loss": 11.6089,
+      "step": 3
+    },
+    {
+      "epoch": 0.005661712668082095,
+      "grad_norm": 43.100232328671524,
+      "learning_rate": 7.075471698113208e-07,
+      "loss": 11.5825,
+      "step": 4
+    },
+    {
+      "epoch": 0.007077140835102618,
+      "grad_norm": 42.5821886975642,
+      "learning_rate": 9.433962264150943e-07,
+      "loss": 11.6043,
+      "step": 5
+    },
+    {
+      "epoch": 0.008492569002123142,
+      "grad_norm": 47.15198695289155,
+      "learning_rate": 1.179245283018868e-06,
+      "loss": 11.4062,
+      "step": 6
+    },
+    {
+      "epoch": 0.009907997169143666,
+      "grad_norm": 47.21018798250648,
+      "learning_rate": 1.4150943396226415e-06,
+      "loss": 11.2894,
+      "step": 7
+    },
+    {
+      "epoch": 0.01132342533616419,
+      "grad_norm": 52.16499613461483,
+      "learning_rate": 1.650943396226415e-06,
+      "loss": 11.1135,
+      "step": 8
+    },
+    {
+      "epoch": 0.012738853503184714,
+      "grad_norm": 55.63144370327757,
+      "learning_rate": 1.8867924528301887e-06,
+      "loss": 10.9942,
+      "step": 9
+    },
+    {
+      "epoch": 0.014154281670205236,
+      "grad_norm": 64.12730354425412,
+      "learning_rate": 2.1226415094339624e-06,
+      "loss": 10.3487,
+      "step": 10
+    },
+    {
+      "epoch": 0.01556970983722576,
+      "grad_norm": 71.34125577863145,
+      "learning_rate": 2.358490566037736e-06,
+      "loss": 10.3151,
+      "step": 11
+    },
+    {
+      "epoch": 0.016985138004246284,
+      "grad_norm": 74.0270309921208,
+      "learning_rate": 2.5943396226415095e-06,
+      "loss": 9.8093,
+      "step": 12
+    },
+    {
+      "epoch": 0.01840056617126681,
+      "grad_norm": 103.30745846664922,
+      "learning_rate": 2.830188679245283e-06,
+      "loss": 6.1753,
+      "step": 13
+    },
+    {
+      "epoch": 0.019815994338287332,
+      "grad_norm": 77.09348349918515,
+      "learning_rate": 3.0660377358490567e-06,
+      "loss": 4.7991,
+      "step": 14
+    },
+    {
+      "epoch": 0.021231422505307854,
+      "grad_norm": 71.52287449239502,
+      "learning_rate": 3.30188679245283e-06,
+      "loss": 4.6793,
+      "step": 15
+    },
+    {
+      "epoch": 0.02264685067232838,
+      "grad_norm": 46.74022190241013,
+      "learning_rate": 3.5377358490566038e-06,
+      "loss": 3.1627,
+      "step": 16
+    },
+    {
+      "epoch": 0.024062278839348902,
+      "grad_norm": 40.45919537581584,
+      "learning_rate": 3.7735849056603773e-06,
+      "loss": 3.0045,
+      "step": 17
+    },
+    {
+      "epoch": 0.025477707006369428,
+      "grad_norm": 8.445011797465854,
+      "learning_rate": 4.009433962264151e-06,
+      "loss": 1.9597,
+      "step": 18
+    },
+    {
+      "epoch": 0.02689313517338995,
+      "grad_norm": 5.479013054164001,
+      "learning_rate": 4.245283018867925e-06,
+      "loss": 1.4691,
+      "step": 19
+    },
+    {
+      "epoch": 0.028308563340410473,
+      "grad_norm": 5.440314496315961,
+      "learning_rate": 4.481132075471698e-06,
+      "loss": 1.7585,
+      "step": 20
+    },
+    {
+      "epoch": 0.029723991507430998,
+      "grad_norm": 4.73587778702694,
+      "learning_rate": 4.716981132075472e-06,
+      "loss": 1.7664,
+      "step": 21
+    },
+    {
+      "epoch": 0.03113941967445152,
+      "grad_norm": 3.391157126674541,
+      "learning_rate": 4.952830188679246e-06,
+      "loss": 1.5264,
+      "step": 22
+    },
+    {
+      "epoch": 0.03255484784147204,
+      "grad_norm": 2.7846104538504726,
+      "learning_rate": 5.188679245283019e-06,
+      "loss": 1.6084,
+      "step": 23
+    },
+    {
+      "epoch": 0.03397027600849257,
+      "grad_norm": 3.368442099627268,
+      "learning_rate": 5.424528301886793e-06,
+      "loss": 1.6023,
+      "step": 24
+    },
+    {
+      "epoch": 0.035385704175513094,
+      "grad_norm": 1.6613773212139946,
+      "learning_rate": 5.660377358490566e-06,
+      "loss": 1.3373,
+      "step": 25
+    },
+    {
+      "epoch": 0.03680113234253362,
+      "grad_norm": 2.4022375902835065,
+      "learning_rate": 5.89622641509434e-06,
+      "loss": 1.3476,
+      "step": 26
+    },
+    {
+      "epoch": 0.03821656050955414,
+      "grad_norm": 1.8303381406131283,
+      "learning_rate": 6.132075471698113e-06,
+      "loss": 1.3573,
+      "step": 27
+    },
+    {
+      "epoch": 0.039631988676574664,
+      "grad_norm": 1.0492891809918425,
+      "learning_rate": 6.367924528301887e-06,
+      "loss": 0.9964,
+      "step": 28
+    },
+    {
+      "epoch": 0.04104741684359519,
+      "grad_norm": 1.0568449472146275,
+      "learning_rate": 6.60377358490566e-06,
+      "loss": 1.2754,
+      "step": 29
+    },
+    {
+      "epoch": 0.04246284501061571,
+      "grad_norm": 0.8939826961932031,
+      "learning_rate": 6.839622641509434e-06,
+      "loss": 1.0906,
+      "step": 30
+    },
+    {
+      "epoch": 0.043878273177636234,
+      "grad_norm": 0.924706231772569,
+      "learning_rate": 7.0754716981132075e-06,
+      "loss": 1.1488,
+      "step": 31
+    },
+    {
+      "epoch": 0.04529370134465676,
+      "grad_norm": 0.9874613327771626,
+      "learning_rate": 7.3113207547169815e-06,
+      "loss": 0.9942,
+      "step": 32
+    },
+    {
+      "epoch": 0.04670912951167728,
+      "grad_norm": 1.1039357027762495,
+      "learning_rate": 7.547169811320755e-06,
+      "loss": 1.2901,
+      "step": 33
+    },
+    {
+      "epoch": 0.048124557678697805,
+      "grad_norm": 0.6732249168818489,
+      "learning_rate": 7.783018867924528e-06,
+      "loss": 1.1364,
+      "step": 34
+    },
+    {
+      "epoch": 0.04953998584571833,
+      "grad_norm": 0.7389830991594677,
+      "learning_rate": 8.018867924528302e-06,
+      "loss": 1.0965,
+      "step": 35
+    },
+    {
+      "epoch": 0.050955414012738856,
+      "grad_norm": 1.361919734305708,
+      "learning_rate": 8.254716981132076e-06,
+      "loss": 1.069,
+      "step": 36
+    },
+    {
+      "epoch": 0.052370842179759375,
+      "grad_norm": 0.6946076673297682,
+      "learning_rate": 8.49056603773585e-06,
+      "loss": 0.8868,
+      "step": 37
+    },
+    {
+      "epoch": 0.0537862703467799,
+      "grad_norm": 0.678223157015022,
+      "learning_rate": 8.726415094339622e-06,
+      "loss": 1.1229,
+      "step": 38
+    },
+    {
+      "epoch": 0.055201698513800426,
+      "grad_norm": 0.6733857464549065,
+      "learning_rate": 8.962264150943396e-06,
+      "loss": 1.0492,
+      "step": 39
+    },
+    {
+      "epoch": 0.056617126680820945,
+      "grad_norm": 0.5835930567416066,
+      "learning_rate": 9.19811320754717e-06,
+      "loss": 0.9794,
+      "step": 40
+    },
+    {
+      "epoch": 0.05803255484784147,
+      "grad_norm": 0.5657763577994933,
+      "learning_rate": 9.433962264150944e-06,
+      "loss": 1.0282,
+      "step": 41
+    },
+    {
+      "epoch": 0.059447983014861996,
+      "grad_norm": 0.5916133305317316,
+      "learning_rate": 9.669811320754718e-06,
+      "loss": 0.8039,
+      "step": 42
+    },
+    {
+      "epoch": 0.06086341118188252,
+      "grad_norm": 0.6176031326803815,
+      "learning_rate": 9.905660377358492e-06,
+      "loss": 0.8717,
+      "step": 43
+    },
+    {
+      "epoch": 0.06227883934890304,
+      "grad_norm": 0.5047080085145694,
+      "learning_rate": 1.0141509433962266e-05,
+      "loss": 0.9476,
+      "step": 44
+    },
+    {
+      "epoch": 0.06369426751592357,
+      "grad_norm": 0.5057501627696119,
+      "learning_rate": 1.0377358490566038e-05,
+      "loss": 1.0082,
+      "step": 45
+    },
+    {
+      "epoch": 0.06510969568294409,
+      "grad_norm": 0.4844409794803433,
+      "learning_rate": 1.0613207547169812e-05,
+      "loss": 0.9998,
+      "step": 46
+    },
+    {
+      "epoch": 0.06652512384996462,
+      "grad_norm": 0.439835805052479,
+      "learning_rate": 1.0849056603773586e-05,
+      "loss": 0.7494,
+      "step": 47
+    },
+    {
+      "epoch": 0.06794055201698514,
+      "grad_norm": 0.47748592314903204,
+      "learning_rate": 1.108490566037736e-05,
+      "loss": 0.9679,
+      "step": 48
+    },
+    {
+      "epoch": 0.06935598018400566,
+      "grad_norm": 0.41694798317782045,
+      "learning_rate": 1.1320754716981132e-05,
+      "loss": 0.9175,
+      "step": 49
+    },
+    {
+      "epoch": 0.07077140835102619,
+      "grad_norm": 0.4131031923420625,
+      "learning_rate": 1.1556603773584906e-05,
+      "loss": 0.8525,
+      "step": 50
+    },
+    {
+      "epoch": 0.07218683651804671,
+      "grad_norm": 0.39877873003968667,
+      "learning_rate": 1.179245283018868e-05,
+      "loss": 0.9432,
+      "step": 51
+    },
+    {
+      "epoch": 0.07360226468506724,
+      "grad_norm": 0.38778633507310184,
+      "learning_rate": 1.2028301886792454e-05,
+      "loss": 0.9381,
+      "step": 52
+    },
+    {
+      "epoch": 0.07501769285208776,
+      "grad_norm": 0.35074905382025257,
+      "learning_rate": 1.2264150943396227e-05,
+      "loss": 0.7945,
+      "step": 53
+    },
+    {
+      "epoch": 0.07643312101910828,
+      "grad_norm": 0.4164483097493037,
+      "learning_rate": 1.25e-05,
+      "loss": 0.9996,
+      "step": 54
+    },
+    {
+      "epoch": 0.07784854918612881,
+      "grad_norm": 0.4634371429875181,
+      "learning_rate": 1.2735849056603775e-05,
+      "loss": 0.9571,
+      "step": 55
+    },
+    {
+      "epoch": 0.07926397735314933,
+      "grad_norm": 0.3872343128801052,
+      "learning_rate": 1.2971698113207547e-05,
+      "loss": 0.8564,
+      "step": 56
+    },
+    {
+      "epoch": 0.08067940552016985,
+      "grad_norm": 0.3918275587660996,
+      "learning_rate": 1.320754716981132e-05,
+      "loss": 0.9868,
+      "step": 57
+    },
+    {
+      "epoch": 0.08209483368719038,
+      "grad_norm": 0.3454015784380163,
+      "learning_rate": 1.3443396226415095e-05,
+      "loss": 0.8604,
+      "step": 58
+    },
+    {
+      "epoch": 0.0835102618542109,
+      "grad_norm": 0.4087333001843292,
+      "learning_rate": 1.3679245283018869e-05,
+      "loss": 0.9851,
+      "step": 59
+    },
+    {
+      "epoch": 0.08492569002123142,
+      "grad_norm": 0.7108784390602116,
+      "learning_rate": 1.3915094339622641e-05,
+      "loss": 0.9191,
+      "step": 60
+    },
+    {
+      "epoch": 0.08634111818825195,
+      "grad_norm": 0.3245781305656771,
+      "learning_rate": 1.4150943396226415e-05,
+      "loss": 0.8334,
+      "step": 61
+    },
+    {
+      "epoch": 0.08775654635527247,
+      "grad_norm": 0.362401119658312,
+      "learning_rate": 1.4386792452830189e-05,
+      "loss": 0.9544,
+      "step": 62
+    },
+    {
+      "epoch": 0.08917197452229299,
+      "grad_norm": 0.3153301984897242,
+      "learning_rate": 1.4622641509433963e-05,
+      "loss": 0.7678,
+      "step": 63
+    },
+    {
+      "epoch": 0.09058740268931352,
+      "grad_norm": 0.3410193369934483,
+      "learning_rate": 1.4858490566037735e-05,
+      "loss": 0.6978,
+      "step": 64
+    },
+    {
+      "epoch": 0.09200283085633404,
+      "grad_norm": 0.3327538499445616,
+      "learning_rate": 1.509433962264151e-05,
+      "loss": 0.8188,
+      "step": 65
+    },
+    {
+      "epoch": 0.09341825902335456,
+      "grad_norm": 0.32595848738844885,
+      "learning_rate": 1.5330188679245283e-05,
+      "loss": 0.7869,
+      "step": 66
+    },
+    {
+      "epoch": 0.09483368719037509,
+      "grad_norm": 0.3147147961761771,
+      "learning_rate": 1.5566037735849056e-05,
+      "loss": 0.7442,
+      "step": 67
+    },
+    {
+      "epoch": 0.09624911535739561,
+      "grad_norm": 1.3369548940206364,
+      "learning_rate": 1.580188679245283e-05,
+      "loss": 0.8758,
+      "step": 68
+    },
+    {
+      "epoch": 0.09766454352441614,
+      "grad_norm": 0.419616964657023,
+      "learning_rate": 1.6037735849056604e-05,
+      "loss": 0.864,
+      "step": 69
+    },
+    {
+      "epoch": 0.09907997169143666,
+      "grad_norm": 0.3995185185455742,
+      "learning_rate": 1.6273584905660376e-05,
+      "loss": 0.837,
+      "step": 70
+    },
+    {
+      "epoch": 0.10049539985845718,
+      "grad_norm": 0.3332916810605707,
+      "learning_rate": 1.650943396226415e-05,
+      "loss": 0.7015,
+      "step": 71
+    },
+    {
+      "epoch": 0.10191082802547771,
+      "grad_norm": 0.34754244002488843,
+      "learning_rate": 1.6745283018867924e-05,
+      "loss": 0.7274,
+      "step": 72
+    },
+    {
+      "epoch": 0.10332625619249823,
+      "grad_norm": 0.33973651313585607,
+      "learning_rate": 1.69811320754717e-05,
+      "loss": 0.7914,
+      "step": 73
+    },
+    {
+      "epoch": 0.10474168435951875,
+      "grad_norm": 0.38442548640761615,
+      "learning_rate": 1.7216981132075472e-05,
+      "loss": 1.0713,
+      "step": 74
+    },
+    {
+      "epoch": 0.10615711252653928,
+      "grad_norm": 0.3748707398787612,
+      "learning_rate": 1.7452830188679244e-05,
+      "loss": 0.729,
+      "step": 75
+    },
+    {
+      "epoch": 0.1075725406935598,
+      "grad_norm": 0.33799485157775017,
+      "learning_rate": 1.768867924528302e-05,
+      "loss": 0.9651,
+      "step": 76
+    },
+    {
+      "epoch": 0.10898796886058032,
+      "grad_norm": 0.32834750548367253,
+      "learning_rate": 1.7924528301886792e-05,
+      "loss": 0.7978,
+      "step": 77
+    },
+    {
+      "epoch": 0.11040339702760085,
+      "grad_norm": 0.3502524463921579,
+      "learning_rate": 1.8160377358490564e-05,
+      "loss": 0.8221,
+      "step": 78
+    },
+    {
+      "epoch": 0.11181882519462137,
+      "grad_norm": 0.3331270390715312,
+      "learning_rate": 1.839622641509434e-05,
+      "loss": 0.8731,
+      "step": 79
+    },
+    {
+      "epoch": 0.11323425336164189,
+      "grad_norm": 0.3346584437841966,
+      "learning_rate": 1.8632075471698112e-05,
+      "loss": 0.8424,
+      "step": 80
+    },
+    {
+      "epoch": 0.11464968152866242,
+      "grad_norm": 0.3305752792911303,
+      "learning_rate": 1.8867924528301888e-05,
+      "loss": 0.6816,
+      "step": 81
+    },
+    {
+      "epoch": 0.11606510969568294,
+      "grad_norm": 0.30323282202609836,
+      "learning_rate": 1.9103773584905664e-05,
+      "loss": 0.739,
+      "step": 82
+    },
+    {
+      "epoch": 0.11748053786270347,
+      "grad_norm": 0.2932445274255076,
+      "learning_rate": 1.9339622641509436e-05,
+      "loss": 0.7707,
+      "step": 83
+    },
+    {
+      "epoch": 0.11889596602972399,
+      "grad_norm": 0.35360411323590746,
+      "learning_rate": 1.9575471698113208e-05,
+      "loss": 0.9276,
+      "step": 84
+    },
+    {
+      "epoch": 0.12031139419674451,
+      "grad_norm": 0.31784927590295037,
+      "learning_rate": 1.9811320754716984e-05,
+      "loss": 0.6502,
+      "step": 85
+    },
+    {
+      "epoch": 0.12172682236376504,
+      "grad_norm": 0.2930105538589784,
+      "learning_rate": 2.0047169811320756e-05,
+      "loss": 0.5902,
+      "step": 86
+    },
+    {
+      "epoch": 0.12314225053078556,
+      "grad_norm": 0.4298747913122774,
+      "learning_rate": 2.0283018867924532e-05,
+      "loss": 0.8808,
+      "step": 87
+    },
+    {
+      "epoch": 0.12455767869780608,
+      "grad_norm": 0.3227619460942626,
+      "learning_rate": 2.0518867924528304e-05,
+      "loss": 0.6742,
+      "step": 88
+    },
+    {
+      "epoch": 0.1259731068648266,
+      "grad_norm": 0.32830224756849186,
+      "learning_rate": 2.0754716981132076e-05,
+      "loss": 0.7305,
+      "step": 89
+    },
+    {
+      "epoch": 0.12738853503184713,
+      "grad_norm": 0.3508475206918051,
+      "learning_rate": 2.0990566037735852e-05,
+      "loss": 0.8298,
+      "step": 90
+    },
+    {
+      "epoch": 0.12880396319886767,
+      "grad_norm": 0.31955241801604534,
+      "learning_rate": 2.1226415094339624e-05,
+      "loss": 0.7769,
+      "step": 91
+    },
+    {
+      "epoch": 0.13021939136588817,
+      "grad_norm": 0.3443616665430657,
+      "learning_rate": 2.1462264150943397e-05,
+      "loss": 0.8659,
+      "step": 92
+    },
+    {
+      "epoch": 0.1316348195329087,
+      "grad_norm": 0.36179953262070746,
+      "learning_rate": 2.1698113207547172e-05,
+      "loss": 0.8947,
+      "step": 93
+    },
+    {
+      "epoch": 0.13305024769992924,
+      "grad_norm": 0.3362790434923789,
+      "learning_rate": 2.1933962264150945e-05,
+      "loss": 0.8423,
+      "step": 94
+    },
+    {
+      "epoch": 0.13446567586694974,
+      "grad_norm": 0.3083875968981802,
+      "learning_rate": 2.216981132075472e-05,
+      "loss": 0.7886,
+      "step": 95
+    },
+    {
+      "epoch": 0.13588110403397027,
+      "grad_norm": 0.32344656361127455,
+      "learning_rate": 2.2405660377358493e-05,
+      "loss": 0.7875,
+      "step": 96
+    },
+    {
+      "epoch": 0.1372965322009908,
+      "grad_norm": 0.3165439663454565,
+      "learning_rate": 2.2641509433962265e-05,
+      "loss": 0.8044,
+      "step": 97
+    },
+    {
+      "epoch": 0.1387119603680113,
+      "grad_norm": 0.3558619447859139,
+      "learning_rate": 2.287735849056604e-05,
+      "loss": 0.8377,
+      "step": 98
+    },
+    {
+      "epoch": 0.14012738853503184,
+      "grad_norm": 0.3230854098787612,
+      "learning_rate": 2.3113207547169813e-05,
+      "loss": 0.8527,
+      "step": 99
+    },
+    {
+      "epoch": 0.14154281670205238,
+      "grad_norm": 0.292154547470411,
+      "learning_rate": 2.3349056603773585e-05,
+      "loss": 0.6429,
+      "step": 100
+    },
+    {
+      "epoch": 0.14295824486907288,
+      "grad_norm": 0.29618483220184183,
+      "learning_rate": 2.358490566037736e-05,
+      "loss": 0.6592,
+      "step": 101
+    },
+    {
+      "epoch": 0.14437367303609341,
+      "grad_norm": 0.40512023312218853,
+      "learning_rate": 2.3820754716981133e-05,
+      "loss": 0.9447,
+      "step": 102
+    },
+    {
+      "epoch": 0.14578910120311395,
+      "grad_norm": 0.33243783635285035,
+      "learning_rate": 2.405660377358491e-05,
+      "loss": 0.6862,
+      "step": 103
+    },
+    {
+      "epoch": 0.14720452937013448,
+      "grad_norm": 0.2989122340388981,
+      "learning_rate": 2.429245283018868e-05,
+      "loss": 0.6963,
+      "step": 104
+    },
+    {
+      "epoch": 0.14861995753715498,
+      "grad_norm": 0.36430675740062585,
+      "learning_rate": 2.4528301886792453e-05,
+      "loss": 0.8468,
+      "step": 105
+    },
+    {
+      "epoch": 0.15003538570417552,
+      "grad_norm": 0.35021308710293897,
+      "learning_rate": 2.476415094339623e-05,
+      "loss": 0.7834,
+      "step": 106
+    },
+    {
+      "epoch": 0.15145081387119605,
+      "grad_norm": 0.34224361021249144,
+      "learning_rate": 2.5e-05,
+      "loss": 0.7443,
+      "step": 107
+    },
+    {
+      "epoch": 0.15286624203821655,
+      "grad_norm": 0.29948735447722763,
+      "learning_rate": 2.5235849056603777e-05,
+      "loss": 0.6854,
+      "step": 108
+    },
+    {
+      "epoch": 0.1542816702052371,
+      "grad_norm": 0.3508933289889198,
+      "learning_rate": 2.547169811320755e-05,
+      "loss": 0.7311,
+      "step": 109
+    },
+    {
+      "epoch": 0.15569709837225762,
+      "grad_norm": 0.3296823380224191,
+      "learning_rate": 2.5707547169811325e-05,
+      "loss": 0.7032,
+      "step": 110
+    },
+    {
+      "epoch": 0.15711252653927812,
+      "grad_norm": 0.37614064574065786,
+      "learning_rate": 2.5943396226415094e-05,
+      "loss": 0.7134,
+      "step": 111
+    },
+    {
+      "epoch": 0.15852795470629866,
+      "grad_norm": 0.31973855712990845,
+      "learning_rate": 2.6179245283018873e-05,
+      "loss": 0.6965,
+      "step": 112
+    },
+    {
+      "epoch": 0.1599433828733192,
+      "grad_norm": 0.3251856086872114,
+      "learning_rate": 2.641509433962264e-05,
+      "loss": 0.7277,
+      "step": 113
+    },
+    {
+      "epoch": 0.1613588110403397,
+      "grad_norm": 0.3531176610400526,
+      "learning_rate": 2.6650943396226417e-05,
+      "loss": 0.8766,
+      "step": 114
+    },
+    {
+      "epoch": 0.16277423920736023,
+      "grad_norm": 0.35022388084220063,
+      "learning_rate": 2.688679245283019e-05,
+      "loss": 0.8223,
+      "step": 115
+    },
+    {
+      "epoch": 0.16418966737438076,
+      "grad_norm": 0.3311722208850899,
+      "learning_rate": 2.7122641509433965e-05,
+      "loss": 0.6044,
+      "step": 116
+    },
+    {
+      "epoch": 0.16560509554140126,
+      "grad_norm": 0.3196689616157804,
+      "learning_rate": 2.7358490566037738e-05,
+      "loss": 0.6474,
+      "step": 117
+    },
+    {
+      "epoch": 0.1670205237084218,
+      "grad_norm": 0.3391104637918063,
+      "learning_rate": 2.7594339622641513e-05,
+      "loss": 0.8039,
+      "step": 118
+    },
+    {
+      "epoch": 0.16843595187544233,
+      "grad_norm": 0.35406309196254343,
+      "learning_rate": 2.7830188679245282e-05,
+      "loss": 0.6403,
+      "step": 119
+    },
+    {
+      "epoch": 0.16985138004246284,
+      "grad_norm": 0.3440331198300765,
+      "learning_rate": 2.806603773584906e-05,
+      "loss": 0.673,
+      "step": 120
+    },
+    {
+      "epoch": 0.17126680820948337,
+      "grad_norm": 0.36841471396963354,
+      "learning_rate": 2.830188679245283e-05,
+      "loss": 0.7944,
+      "step": 121
+    },
+    {
+      "epoch": 0.1726822363765039,
+      "grad_norm": 0.3410997940493091,
+      "learning_rate": 2.8537735849056606e-05,
+      "loss": 0.6465,
+      "step": 122
+    },
+    {
+      "epoch": 0.1740976645435244,
+      "grad_norm": 0.3424241485506368,
+      "learning_rate": 2.8773584905660378e-05,
+      "loss": 0.7005,
+      "step": 123
+    },
+    {
+      "epoch": 0.17551309271054494,
+      "grad_norm": 0.3087363234800392,
+      "learning_rate": 2.9009433962264154e-05,
+      "loss": 0.7323,
+      "step": 124
+    },
+    {
+      "epoch": 0.17692852087756547,
+      "grad_norm": 0.3208379612118477,
+      "learning_rate": 2.9245283018867926e-05,
+      "loss": 0.715,
+      "step": 125
+    },
+    {
+      "epoch": 0.17834394904458598,
+      "grad_norm": 0.3252623056179373,
+      "learning_rate": 2.9481132075471702e-05,
+      "loss": 0.7182,
+      "step": 126
+    },
+    {
+      "epoch": 0.1797593772116065,
+      "grad_norm": 0.3260606346469109,
+      "learning_rate": 2.971698113207547e-05,
+      "loss": 0.6292,
+      "step": 127
+    },
+    {
+      "epoch": 0.18117480537862704,
+      "grad_norm": 0.33507755461970656,
+      "learning_rate": 2.995283018867925e-05,
+      "loss": 0.6781,
+      "step": 128
+    },
+    {
+      "epoch": 0.18259023354564755,
+      "grad_norm": 0.344152749953347,
+      "learning_rate": 3.018867924528302e-05,
+      "loss": 0.731,
+      "step": 129
+    },
+    {
+      "epoch": 0.18400566171266808,
+      "grad_norm": 0.3790502984860035,
+      "learning_rate": 3.0424528301886794e-05,
+      "loss": 0.7032,
+      "step": 130
+    },
+    {
+      "epoch": 0.1854210898796886,
+      "grad_norm": 0.4060176596426599,
+      "learning_rate": 3.0660377358490567e-05,
+      "loss": 0.7258,
+      "step": 131
+    },
+    {
+      "epoch": 0.18683651804670912,
+      "grad_norm": 0.29752681279467874,
+      "learning_rate": 3.0896226415094346e-05,
+      "loss": 0.5724,
+      "step": 132
+    },
+    {
+      "epoch": 0.18825194621372965,
+      "grad_norm": 0.4490897392705167,
+      "learning_rate": 3.113207547169811e-05,
+      "loss": 0.7565,
+      "step": 133
+    },
+    {
+      "epoch": 0.18966737438075018,
+      "grad_norm": 0.3748295546904542,
+      "learning_rate": 3.136792452830189e-05,
+      "loss": 0.7491,
+      "step": 134
+    },
+    {
+      "epoch": 0.1910828025477707,
+      "grad_norm": 0.4096360150169126,
+      "learning_rate": 3.160377358490566e-05,
+      "loss": 0.7355,
+      "step": 135
+    },
+    {
+      "epoch": 0.19249823071479122,
+      "grad_norm": 0.35372872386670523,
+      "learning_rate": 3.1839622641509435e-05,
+      "loss": 0.6501,
+      "step": 136
+    },
+    {
+      "epoch": 0.19391365888181175,
+      "grad_norm": 0.41983196647145415,
+      "learning_rate": 3.207547169811321e-05,
+      "loss": 0.7145,
+      "step": 137
+    },
+    {
+      "epoch": 0.19532908704883228,
+      "grad_norm": 0.4100283221259522,
+      "learning_rate": 3.2311320754716986e-05,
+      "loss": 0.7778,
+      "step": 138
+    },
+    {
+      "epoch": 0.1967445152158528,
+      "grad_norm": 0.4279192449535737,
+      "learning_rate": 3.254716981132075e-05,
+      "loss": 0.6512,
+      "step": 139
+    },
+    {
+      "epoch": 0.19815994338287332,
+      "grad_norm": 0.37722937839286214,
+      "learning_rate": 3.278301886792453e-05,
+      "loss": 0.6462,
+      "step": 140
+    },
+    {
+      "epoch": 0.19957537154989385,
+      "grad_norm": 0.5177263030721986,
+      "learning_rate": 3.30188679245283e-05,
+      "loss": 0.7195,
+      "step": 141
+    },
+    {
+      "epoch": 0.20099079971691436,
+      "grad_norm": 0.3516292254214785,
+      "learning_rate": 3.3254716981132075e-05,
+      "loss": 0.6927,
+      "step": 142
+    },
+    {
+      "epoch": 0.2024062278839349,
+      "grad_norm": 0.4369868522037781,
+      "learning_rate": 3.349056603773585e-05,
+      "loss": 0.6543,
+      "step": 143
+    },
+    {
+      "epoch": 0.20382165605095542,
+      "grad_norm": 0.351580146475787,
+      "learning_rate": 3.3726415094339627e-05,
+      "loss": 0.6476,
+      "step": 144
+    },
+    {
+      "epoch": 0.20523708421797593,
+      "grad_norm": 0.382859664648938,
+      "learning_rate": 3.39622641509434e-05,
+      "loss": 0.6577,
+      "step": 145
+    },
+    {
+      "epoch": 0.20665251238499646,
+      "grad_norm": 0.389040016792054,
+      "learning_rate": 3.419811320754717e-05,
+      "loss": 0.6965,
+      "step": 146
+    },
+    {
+      "epoch": 0.208067940552017,
+      "grad_norm": 0.30073874667213135,
+      "learning_rate": 3.4433962264150943e-05,
+      "loss": 0.5977,
+      "step": 147
+    },
+    {
+      "epoch": 0.2094833687190375,
+      "grad_norm": 0.38732380059479576,
+      "learning_rate": 3.466981132075472e-05,
+      "loss": 0.6837,
+      "step": 148
+    },
+    {
+      "epoch": 0.21089879688605803,
+      "grad_norm": 0.34655294600757985,
+      "learning_rate": 3.490566037735849e-05,
+      "loss": 0.7161,
+      "step": 149
+    },
+    {
+      "epoch": 0.21231422505307856,
+      "grad_norm": 0.3474622218711454,
+      "learning_rate": 3.514150943396227e-05,
+      "loss": 0.6879,
+      "step": 150
+    },
+    {
+      "epoch": 0.21372965322009907,
+      "grad_norm": 0.3292597346994816,
+      "learning_rate": 3.537735849056604e-05,
+      "loss": 0.6955,
+      "step": 151
+    },
+    {
+      "epoch": 0.2151450813871196,
+      "grad_norm": 0.3383153262618983,
+      "learning_rate": 3.561320754716981e-05,
+      "loss": 0.7442,
+      "step": 152
+    },
+    {
+      "epoch": 0.21656050955414013,
+      "grad_norm": 0.3373136910347554,
+      "learning_rate": 3.5849056603773584e-05,
+      "loss": 0.6368,
+      "step": 153
+    },
+    {
+      "epoch": 0.21797593772116064,
+      "grad_norm": 0.32338504092281495,
+      "learning_rate": 3.608490566037736e-05,
+      "loss": 0.6678,
+      "step": 154
+    },
+    {
+      "epoch": 0.21939136588818117,
+      "grad_norm": 0.327291405252717,
+      "learning_rate": 3.632075471698113e-05,
+      "loss": 0.7245,
+      "step": 155
+    },
+    {
+      "epoch": 0.2208067940552017,
+      "grad_norm": 0.4079198103017352,
+      "learning_rate": 3.655660377358491e-05,
+      "loss": 0.758,
+      "step": 156
+    },
+    {
+      "epoch": 0.2222222222222222,
+      "grad_norm": 0.3431479182310241,
+      "learning_rate": 3.679245283018868e-05,
+      "loss": 0.6949,
+      "step": 157
+    },
+    {
+      "epoch": 0.22363765038924274,
+      "grad_norm": 0.32740527265233493,
+      "learning_rate": 3.702830188679245e-05,
+      "loss": 0.6967,
+      "step": 158
+    },
+    {
+      "epoch": 0.22505307855626328,
+      "grad_norm": 0.355081084230637,
+      "learning_rate": 3.7264150943396224e-05,
+      "loss": 0.6636,
+      "step": 159
+    },
+    {
+      "epoch": 0.22646850672328378,
+      "grad_norm": 0.3377921193259222,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.6152,
+      "step": 160
+    },
+    {
+      "epoch": 0.2278839348903043,
+      "grad_norm": 0.35705251222543827,
+      "learning_rate": 3.7735849056603776e-05,
+      "loss": 0.7044,
+      "step": 161
+    },
+    {
+      "epoch": 0.22929936305732485,
+      "grad_norm": 0.2990541370982297,
+      "learning_rate": 3.797169811320755e-05,
+      "loss": 0.6421,
+      "step": 162
+    },
+    {
+      "epoch": 0.23071479122434538,
+      "grad_norm": 0.37680162093247666,
+      "learning_rate": 3.820754716981133e-05,
+      "loss": 0.7236,
+      "step": 163
+    },
+    {
+      "epoch": 0.23213021939136588,
+      "grad_norm": 0.3316872340797433,
+      "learning_rate": 3.844339622641509e-05,
+      "loss": 0.592,
+      "step": 164
+    },
+    {
+      "epoch": 0.23354564755838642,
+      "grad_norm": 0.3814426738691517,
+      "learning_rate": 3.867924528301887e-05,
+      "loss": 0.642,
+      "step": 165
+    },
+    {
+      "epoch": 0.23496107572540695,
+      "grad_norm": 0.3267572380840606,
+      "learning_rate": 3.8915094339622644e-05,
+      "loss": 0.7495,
+      "step": 166
+    },
+    {
+      "epoch": 0.23637650389242745,
+      "grad_norm": 0.33871837867570975,
+      "learning_rate": 3.9150943396226416e-05,
+      "loss": 0.6928,
+      "step": 167
+    },
+    {
+      "epoch": 0.23779193205944799,
+      "grad_norm": 0.323902169015648,
+      "learning_rate": 3.938679245283019e-05,
+      "loss": 0.8054,
+      "step": 168
+    },
+    {
+      "epoch": 0.23920736022646852,
+      "grad_norm": 0.3105694760719953,
+      "learning_rate": 3.962264150943397e-05,
+      "loss": 0.6868,
+      "step": 169
+    },
+    {
+      "epoch": 0.24062278839348902,
+      "grad_norm": 0.3282634870080322,
+      "learning_rate": 3.985849056603774e-05,
+      "loss": 0.6943,
+      "step": 170
+    },
+    {
+      "epoch": 0.24203821656050956,
+      "grad_norm": 0.3496477577080085,
+      "learning_rate": 4.009433962264151e-05,
+      "loss": 0.7847,
+      "step": 171
+    },
+    {
+      "epoch": 0.2434536447275301,
+      "grad_norm": 0.28731112540621906,
+      "learning_rate": 4.0330188679245284e-05,
+      "loss": 0.6661,
+      "step": 172
+    },
+    {
+      "epoch": 0.2448690728945506,
+      "grad_norm": 0.3546905432002279,
+      "learning_rate": 4.0566037735849064e-05,
+      "loss": 0.7017,
+      "step": 173
+    },
+    {
+      "epoch": 0.24628450106157113,
+      "grad_norm": 0.30690572266461835,
+      "learning_rate": 4.080188679245283e-05,
+      "loss": 0.6241,
+      "step": 174
+    },
+    {
+      "epoch": 0.24769992922859166,
+      "grad_norm": 0.3264453307239247,
+      "learning_rate": 4.103773584905661e-05,
+      "loss": 0.7084,
+      "step": 175
+    },
+    {
+      "epoch": 0.24911535739561216,
+      "grad_norm": 0.356494327898739,
+      "learning_rate": 4.127358490566038e-05,
+      "loss": 0.6766,
+      "step": 176
+    },
+    {
+      "epoch": 0.2505307855626327,
+      "grad_norm": 0.37845684989010364,
+      "learning_rate": 4.150943396226415e-05,
+      "loss": 0.6764,
+      "step": 177
+    },
+    {
+      "epoch": 0.2519462137296532,
+      "grad_norm": 0.3822237153839302,
+      "learning_rate": 4.1745283018867925e-05,
+      "loss": 0.7572,
+      "step": 178
+    },
+    {
+      "epoch": 0.25336164189667376,
+      "grad_norm": 0.33180805872343516,
+      "learning_rate": 4.1981132075471704e-05,
+      "loss": 0.5645,
+      "step": 179
+    },
+    {
+      "epoch": 0.25477707006369427,
+      "grad_norm": 0.3776125949939335,
+      "learning_rate": 4.221698113207547e-05,
+      "loss": 0.6428,
+      "step": 180
+    },
+    {
+      "epoch": 0.25619249823071477,
+      "grad_norm": 0.34307054586410674,
+      "learning_rate": 4.245283018867925e-05,
+      "loss": 0.7138,
+      "step": 181
+    },
+    {
+      "epoch": 0.25760792639773533,
+      "grad_norm": 0.3421864387970046,
+      "learning_rate": 4.268867924528302e-05,
+      "loss": 0.6872,
+      "step": 182
+    },
+    {
+      "epoch": 0.25902335456475584,
+      "grad_norm": 0.3558737974385253,
+      "learning_rate": 4.292452830188679e-05,
+      "loss": 0.6201,
+      "step": 183
+    },
+    {
+      "epoch": 0.26043878273177634,
+      "grad_norm": 0.34802732619093407,
+      "learning_rate": 4.3160377358490565e-05,
+      "loss": 0.7949,
+      "step": 184
+    },
+    {
+      "epoch": 0.2618542108987969,
+      "grad_norm": 0.47243969103960226,
+      "learning_rate": 4.3396226415094345e-05,
+      "loss": 0.7231,
+      "step": 185
+    },
+    {
+      "epoch": 0.2632696390658174,
+      "grad_norm": 0.48027533389765475,
+      "learning_rate": 4.363207547169812e-05,
+      "loss": 0.6296,
+      "step": 186
+    },
+    {
+      "epoch": 0.2646850672328379,
+      "grad_norm": 0.29575230575318046,
+      "learning_rate": 4.386792452830189e-05,
+      "loss": 0.5118,
+      "step": 187
+    },
+    {
+      "epoch": 0.26610049539985847,
+      "grad_norm": 0.45243075941702265,
+      "learning_rate": 4.410377358490566e-05,
+      "loss": 0.8435,
+      "step": 188
+    },
+    {
+      "epoch": 0.267515923566879,
+      "grad_norm": 0.4446508772321007,
+      "learning_rate": 4.433962264150944e-05,
+      "loss": 0.5955,
+      "step": 189
+    },
+    {
+      "epoch": 0.2689313517338995,
+      "grad_norm": 0.39098158765333574,
+      "learning_rate": 4.4575471698113206e-05,
+      "loss": 0.7381,
+      "step": 190
+    },
+    {
+      "epoch": 0.27034677990092004,
+      "grad_norm": 0.47812821613700407,
+      "learning_rate": 4.4811320754716985e-05,
+      "loss": 0.686,
+      "step": 191
+    },
+    {
+      "epoch": 0.27176220806794055,
+      "grad_norm": 0.4255378884479202,
+      "learning_rate": 4.504716981132076e-05,
+      "loss": 0.6267,
+      "step": 192
+    },
+    {
+      "epoch": 0.27317763623496105,
+      "grad_norm": 0.41920925771608414,
+      "learning_rate": 4.528301886792453e-05,
+      "loss": 0.7695,
+      "step": 193
+    },
+    {
+      "epoch": 0.2745930644019816,
+      "grad_norm": 0.4901895726580499,
+      "learning_rate": 4.55188679245283e-05,
+      "loss": 0.6972,
+      "step": 194
+    },
+    {
+      "epoch": 0.2760084925690021,
+      "grad_norm": 0.5172747736553965,
+      "learning_rate": 4.575471698113208e-05,
+      "loss": 0.6524,
+      "step": 195
+    },
+    {
+      "epoch": 0.2774239207360226,
+      "grad_norm": 0.32758712427375986,
+      "learning_rate": 4.5990566037735846e-05,
+      "loss": 0.6214,
+      "step": 196
+    },
+    {
+      "epoch": 0.2788393489030432,
+      "grad_norm": 0.6629263683616758,
+      "learning_rate": 4.6226415094339625e-05,
+      "loss": 0.6382,
+      "step": 197
+    },
+    {
+      "epoch": 0.2802547770700637,
+      "grad_norm": 0.5255514509389033,
+      "learning_rate": 4.64622641509434e-05,
+      "loss": 0.6253,
+      "step": 198
+    },
+    {
+      "epoch": 0.2816702052370842,
+      "grad_norm": 0.4199105090028493,
+      "learning_rate": 4.669811320754717e-05,
+      "loss": 0.6534,
+      "step": 199
+    },
+    {
+      "epoch": 0.28308563340410475,
+      "grad_norm": 0.5133025788360546,
+      "learning_rate": 4.693396226415094e-05,
+      "loss": 0.5992,
+      "step": 200
+    },
+    {
+      "epoch": 0.28450106157112526,
+      "grad_norm": 0.3515641457954602,
+      "learning_rate": 4.716981132075472e-05,
+      "loss": 0.5936,
+      "step": 201
+    },
+    {
+      "epoch": 0.28591648973814576,
+      "grad_norm": 0.46382284062883583,
+      "learning_rate": 4.7405660377358494e-05,
+      "loss": 0.683,
+      "step": 202
+    },
+    {
+      "epoch": 0.2873319179051663,
+      "grad_norm": 0.5789901497369425,
+      "learning_rate": 4.7641509433962266e-05,
+      "loss": 0.6319,
+      "step": 203
+    },
+    {
+      "epoch": 0.28874734607218683,
+      "grad_norm": 0.3643121215092036,
+      "learning_rate": 4.787735849056604e-05,
+      "loss": 0.5249,
+      "step": 204
+    },
+    {
+      "epoch": 0.29016277423920733,
+      "grad_norm": 0.47331340234321256,
+      "learning_rate": 4.811320754716982e-05,
+      "loss": 0.6546,
+      "step": 205
+    },
+    {
+      "epoch": 0.2915782024062279,
+      "grad_norm": 0.40955091988638687,
+      "learning_rate": 4.834905660377358e-05,
+      "loss": 0.7085,
+      "step": 206
+    },
+    {
+      "epoch": 0.2929936305732484,
+      "grad_norm": 0.35402965766051764,
+      "learning_rate": 4.858490566037736e-05,
+      "loss": 0.7199,
+      "step": 207
+    },
+    {
+      "epoch": 0.29440905874026896,
+      "grad_norm": 0.33550518291489584,
+      "learning_rate": 4.8820754716981134e-05,
+      "loss": 0.6114,
+      "step": 208
+    },
+    {
+      "epoch": 0.29582448690728946,
+      "grad_norm": 0.36970155693332196,
+      "learning_rate": 4.9056603773584906e-05,
+      "loss": 0.5843,
+      "step": 209
+    },
+    {
+      "epoch": 0.29723991507430997,
+      "grad_norm": 0.3697693128093627,
+      "learning_rate": 4.929245283018868e-05,
+      "loss": 0.7422,
+      "step": 210
+    },
+    {
+      "epoch": 0.29865534324133053,
+      "grad_norm": 0.41500202886671717,
+      "learning_rate": 4.952830188679246e-05,
+      "loss": 0.639,
+      "step": 211
+    },
+    {
+      "epoch": 0.30007077140835103,
+      "grad_norm": 0.2856425084141532,
+      "learning_rate": 4.976415094339622e-05,
+      "loss": 0.6182,
+      "step": 212
+    },
+    {
+      "epoch": 0.30148619957537154,
+      "grad_norm": 0.46521888562093133,
+      "learning_rate": 5e-05,
+      "loss": 0.722,
+      "step": 213
+    },
+    {
+      "epoch": 0.3029016277423921,
+      "grad_norm": 0.3845414037620586,
+      "learning_rate": 4.997376705141658e-05,
+      "loss": 0.6922,
+      "step": 214
+    },
+    {
+      "epoch": 0.3043170559094126,
+      "grad_norm": 0.45351901517196663,
+      "learning_rate": 4.994753410283316e-05,
+      "loss": 0.6222,
+      "step": 215
+    },
+    {
+      "epoch": 0.3057324840764331,
+      "grad_norm": 0.38318818305297797,
+      "learning_rate": 4.9921301154249736e-05,
+      "loss": 0.6177,
+      "step": 216
+    },
+    {
+      "epoch": 0.30714791224345367,
+      "grad_norm": 0.4406176060172244,
+      "learning_rate": 4.989506820566632e-05,
+      "loss": 0.6203,
+      "step": 217
+    },
+    {
+      "epoch": 0.3085633404104742,
+      "grad_norm": 0.3798385623910793,
+      "learning_rate": 4.98688352570829e-05,
+      "loss": 0.6833,
+      "step": 218
+    },
+    {
+      "epoch": 0.3099787685774947,
+      "grad_norm": 0.4117758569188405,
+      "learning_rate": 4.984260230849948e-05,
+      "loss": 0.5478,
+      "step": 219
+    },
+    {
+      "epoch": 0.31139419674451524,
+      "grad_norm": 0.3651670593181467,
+      "learning_rate": 4.981636935991606e-05,
+      "loss": 0.6085,
+      "step": 220
+    },
+    {
+      "epoch": 0.31280962491153574,
+      "grad_norm": 0.3985470259616175,
+      "learning_rate": 4.979013641133264e-05,
+      "loss": 0.6244,
+      "step": 221
+    },
+    {
+      "epoch": 0.31422505307855625,
+      "grad_norm": 0.33366417189852593,
+      "learning_rate": 4.976390346274922e-05,
+      "loss": 0.6859,
+      "step": 222
+    },
+    {
+      "epoch": 0.3156404812455768,
+      "grad_norm": 0.40063208383608917,
+      "learning_rate": 4.97376705141658e-05,
+      "loss": 0.6583,
+      "step": 223
+    },
+    {
+      "epoch": 0.3170559094125973,
+      "grad_norm": 0.3406033500089553,
+      "learning_rate": 4.971143756558237e-05,
+      "loss": 0.75,
+      "step": 224
+    },
+    {
+      "epoch": 0.3184713375796178,
+      "grad_norm": 0.3167388846013373,
+      "learning_rate": 4.968520461699895e-05,
+      "loss": 0.6368,
+      "step": 225
+    },
+    {
+      "epoch": 0.3198867657466384,
+      "grad_norm": 0.30183382248828816,
+      "learning_rate": 4.965897166841553e-05,
+      "loss": 0.5887,
+      "step": 226
+    },
+    {
+      "epoch": 0.3213021939136589,
+      "grad_norm": 0.3410080956406333,
+      "learning_rate": 4.963273871983211e-05,
+      "loss": 0.6944,
+      "step": 227
+    },
+    {
+      "epoch": 0.3227176220806794,
+      "grad_norm": 0.3016417065921014,
+      "learning_rate": 4.960650577124869e-05,
+      "loss": 0.7025,
+      "step": 228
+    },
+    {
+      "epoch": 0.32413305024769995,
+      "grad_norm": 0.39533283501293887,
+      "learning_rate": 4.958027282266527e-05,
+      "loss": 0.7395,
+      "step": 229
+    },
+    {
+      "epoch": 0.32554847841472045,
+      "grad_norm": 0.317776843840245,
+      "learning_rate": 4.955403987408185e-05,
+      "loss": 0.6477,
+      "step": 230
+    },
+    {
+      "epoch": 0.32696390658174096,
+      "grad_norm": 2.6162825696857506,
+      "learning_rate": 4.952780692549843e-05,
+      "loss": 0.7805,
+      "step": 231
+    },
+    {
+      "epoch": 0.3283793347487615,
+      "grad_norm": 0.38246518870773766,
+      "learning_rate": 4.950157397691501e-05,
+      "loss": 0.5729,
+      "step": 232
+    },
+    {
+      "epoch": 0.329794762915782,
+      "grad_norm": 0.35451653588292686,
+      "learning_rate": 4.947534102833158e-05,
+      "loss": 0.6034,
+      "step": 233
+    },
+    {
+      "epoch": 0.33121019108280253,
+      "grad_norm": 0.35793206668480076,
+      "learning_rate": 4.944910807974817e-05,
+      "loss": 0.6168,
+      "step": 234
+    },
+    {
+      "epoch": 0.3326256192498231,
+      "grad_norm": 0.3959020643773459,
+      "learning_rate": 4.942287513116475e-05,
+      "loss": 0.5585,
+      "step": 235
+    },
+    {
+      "epoch": 0.3340410474168436,
+      "grad_norm": 0.3805393159052667,
+      "learning_rate": 4.939664218258133e-05,
+      "loss": 0.6529,
+      "step": 236
+    },
+    {
+      "epoch": 0.3354564755838641,
+      "grad_norm": 0.34245689486771175,
+      "learning_rate": 4.93704092339979e-05,
+      "loss": 0.5962,
+      "step": 237
+    },
+    {
+      "epoch": 0.33687190375088466,
+      "grad_norm": 0.46119035852206347,
+      "learning_rate": 4.934417628541448e-05,
+      "loss": 0.6898,
+      "step": 238
+    },
+    {
+      "epoch": 0.33828733191790517,
+      "grad_norm": 0.2760463725152546,
+      "learning_rate": 4.931794333683106e-05,
+      "loss": 0.5918,
+      "step": 239
+    },
+    {
+      "epoch": 0.33970276008492567,
+      "grad_norm": 0.4212781419762082,
+      "learning_rate": 4.929171038824764e-05,
+      "loss": 0.8071,
+      "step": 240
+    },
+    {
+      "epoch": 0.34111818825194623,
+      "grad_norm": 0.3336146067216131,
+      "learning_rate": 4.926547743966422e-05,
+      "loss": 0.6629,
+      "step": 241
+    },
+    {
+      "epoch": 0.34253361641896674,
+      "grad_norm": 0.34986136867361556,
+      "learning_rate": 4.92392444910808e-05,
+      "loss": 0.6263,
+      "step": 242
+    },
+    {
+      "epoch": 0.34394904458598724,
+      "grad_norm": 0.3191579021579259,
+      "learning_rate": 4.921301154249738e-05,
+      "loss": 0.6556,
+      "step": 243
+    },
+    {
+      "epoch": 0.3453644727530078,
+      "grad_norm": 0.35217932985524025,
+      "learning_rate": 4.918677859391396e-05,
+      "loss": 0.6628,
+      "step": 244
+    },
+    {
+      "epoch": 0.3467799009200283,
+      "grad_norm": 0.3382215322275221,
+      "learning_rate": 4.916054564533054e-05,
+      "loss": 0.6878,
+      "step": 245
+    },
+    {
+      "epoch": 0.3481953290870488,
+      "grad_norm": 0.3192853417565939,
+      "learning_rate": 4.913431269674712e-05,
+      "loss": 0.6305,
+      "step": 246
+    },
+    {
+      "epoch": 0.34961075725406937,
+      "grad_norm": 0.39592306170954034,
+      "learning_rate": 4.910807974816369e-05,
+      "loss": 0.6313,
+      "step": 247
+    },
+    {
+      "epoch": 0.3510261854210899,
+      "grad_norm": 0.3717422268582204,
+      "learning_rate": 4.908184679958028e-05,
+      "loss": 0.6787,
+      "step": 248
+    },
+    {
+      "epoch": 0.3524416135881104,
+      "grad_norm": 0.4697879198033925,
+      "learning_rate": 4.905561385099686e-05,
+      "loss": 0.6593,
+      "step": 249
+    },
+    {
+      "epoch": 0.35385704175513094,
+      "grad_norm": 0.3145225755984823,
+      "learning_rate": 4.902938090241343e-05,
+      "loss": 0.559,
+      "step": 250
+    },
+    {
+      "epoch": 0.35527246992215145,
+      "grad_norm": 0.4055147305953671,
+      "learning_rate": 4.900314795383001e-05,
+      "loss": 0.6411,
+      "step": 251
+    },
+    {
+      "epoch": 0.35668789808917195,
+      "grad_norm": 0.3296941362737166,
+      "learning_rate": 4.897691500524659e-05,
+      "loss": 0.6256,
+      "step": 252
+    },
+    {
+      "epoch": 0.3581033262561925,
+      "grad_norm": 0.4307575396162028,
+      "learning_rate": 4.895068205666317e-05,
+      "loss": 0.5943,
+      "step": 253
+    },
+    {
+      "epoch": 0.359518754423213,
+      "grad_norm": 0.3602007816421482,
+      "learning_rate": 4.8924449108079753e-05,
+      "loss": 0.6412,
+      "step": 254
+    },
+    {
+      "epoch": 0.3609341825902335,
+      "grad_norm": 0.32293453159004193,
+      "learning_rate": 4.889821615949633e-05,
+      "loss": 0.5814,
+      "step": 255
+    },
+    {
+      "epoch": 0.3623496107572541,
+      "grad_norm": 0.4395377605765523,
+      "learning_rate": 4.887198321091291e-05,
+      "loss": 0.6761,
+      "step": 256
+    },
+    {
+      "epoch": 0.3637650389242746,
+      "grad_norm": 0.30063842475465097,
+      "learning_rate": 4.884575026232949e-05,
+      "loss": 0.6224,
+      "step": 257
+    },
+    {
+      "epoch": 0.3651804670912951,
+      "grad_norm": 0.4466468804355786,
+      "learning_rate": 4.881951731374607e-05,
+      "loss": 0.6249,
+      "step": 258
+    },
+    {
+      "epoch": 0.36659589525831565,
+      "grad_norm": 0.4402685935470065,
+      "learning_rate": 4.879328436516265e-05,
+      "loss": 0.6335,
+      "step": 259
+    },
+    {
+      "epoch": 0.36801132342533616,
+      "grad_norm": 0.4103451796715846,
+      "learning_rate": 4.876705141657922e-05,
+      "loss": 0.694,
+      "step": 260
+    },
+    {
+      "epoch": 0.36942675159235666,
+      "grad_norm": 0.45194536777403443,
+      "learning_rate": 4.87408184679958e-05,
+      "loss": 0.64,
+      "step": 261
+    },
+    {
+      "epoch": 0.3708421797593772,
+      "grad_norm": 0.3933446116945307,
+      "learning_rate": 4.871458551941239e-05,
+      "loss": 0.6434,
+      "step": 262
+    },
+    {
+      "epoch": 0.3722576079263977,
+      "grad_norm": 0.48408491103735724,
+      "learning_rate": 4.868835257082896e-05,
+      "loss": 0.668,
+      "step": 263
+    },
+    {
+      "epoch": 0.37367303609341823,
+      "grad_norm": 0.3422922106651585,
+      "learning_rate": 4.866211962224554e-05,
+      "loss": 0.6148,
+      "step": 264
+    },
+    {
+      "epoch": 0.3750884642604388,
+      "grad_norm": 0.47594899577183286,
+      "learning_rate": 4.863588667366212e-05,
+      "loss": 0.6714,
+      "step": 265
+    },
+    {
+      "epoch": 0.3765038924274593,
+      "grad_norm": 0.40685951284647204,
+      "learning_rate": 4.86096537250787e-05,
+      "loss": 0.636,
+      "step": 266
+    },
+    {
+      "epoch": 0.37791932059447986,
+      "grad_norm": 0.332440654967862,
+      "learning_rate": 4.8583420776495284e-05,
+      "loss": 0.6513,
+      "step": 267
+    },
+    {
+      "epoch": 0.37933474876150036,
+      "grad_norm": 0.652696489413694,
+      "learning_rate": 4.855718782791186e-05,
+      "loss": 0.6459,
+      "step": 268
+    },
+    {
+      "epoch": 0.38075017692852087,
+      "grad_norm": 0.3123837110864311,
+      "learning_rate": 4.853095487932844e-05,
+      "loss": 0.6556,
+      "step": 269
+    },
+    {
+      "epoch": 0.3821656050955414,
+      "grad_norm": 0.4288519383694323,
+      "learning_rate": 4.850472193074502e-05,
+      "loss": 0.6684,
+      "step": 270
+    },
+    {
+      "epoch": 0.38358103326256193,
+      "grad_norm": 0.48238162709886434,
+      "learning_rate": 4.84784889821616e-05,
+      "loss": 0.7107,
+      "step": 271
+    },
+    {
+      "epoch": 0.38499646142958244,
+      "grad_norm": 0.38376189241758835,
+      "learning_rate": 4.845225603357818e-05,
+      "loss": 0.6473,
+      "step": 272
+    },
+    {
+      "epoch": 0.386411889596603,
+      "grad_norm": 0.7127505698452409,
+      "learning_rate": 4.842602308499475e-05,
+      "loss": 0.6139,
+      "step": 273
+    },
+    {
+      "epoch": 0.3878273177636235,
+      "grad_norm": 0.3692518181743927,
+      "learning_rate": 4.839979013641133e-05,
+      "loss": 0.677,
+      "step": 274
+    },
+    {
+      "epoch": 0.389242745930644,
+      "grad_norm": 0.38701071771235995,
+      "learning_rate": 4.837355718782791e-05,
+      "loss": 0.6924,
+      "step": 275
+    },
+    {
+      "epoch": 0.39065817409766457,
+      "grad_norm": 1.725951353723549,
+      "learning_rate": 4.834732423924449e-05,
+      "loss": 0.6087,
+      "step": 276
+    },
+    {
+      "epoch": 0.3920736022646851,
+      "grad_norm": 1.942438208713734,
+      "learning_rate": 4.832109129066107e-05,
+      "loss": 0.7028,
+      "step": 277
+    },
+    {
+      "epoch": 0.3934890304317056,
+      "grad_norm": 1.3706349023564903,
+      "learning_rate": 4.829485834207765e-05,
+      "loss": 0.6555,
+      "step": 278
+    },
+    {
+      "epoch": 0.39490445859872614,
+      "grad_norm": 0.6987371799804422,
+      "learning_rate": 4.8268625393494233e-05,
+      "loss": 0.6366,
+      "step": 279
+    },
+    {
+      "epoch": 0.39631988676574664,
+      "grad_norm": 0.48038293423554523,
+      "learning_rate": 4.8242392444910814e-05,
+      "loss": 0.6256,
+      "step": 280
+    },
+    {
+      "epoch": 0.39773531493276715,
+      "grad_norm": 0.41106735186436566,
+      "learning_rate": 4.821615949632739e-05,
+      "loss": 0.6777,
+      "step": 281
+    },
+    {
+      "epoch": 0.3991507430997877,
+      "grad_norm": 0.5165628841938925,
+      "learning_rate": 4.818992654774397e-05,
+      "loss": 0.6373,
+      "step": 282
+    },
+    {
+      "epoch": 0.4005661712668082,
+      "grad_norm": 2.4701073204434203,
+      "learning_rate": 4.816369359916055e-05,
+      "loss": 0.656,
+      "step": 283
+    },
+    {
+      "epoch": 0.4019815994338287,
+      "grad_norm": 0.5737260562005873,
+      "learning_rate": 4.813746065057713e-05,
+      "loss": 0.6091,
+      "step": 284
+    },
+    {
+      "epoch": 0.4033970276008493,
+      "grad_norm": 0.3502815963568765,
+      "learning_rate": 4.811122770199371e-05,
+      "loss": 0.571,
+      "step": 285
+    },
+    {
+      "epoch": 0.4048124557678698,
+      "grad_norm": 6.354647535609901,
+      "learning_rate": 4.808499475341028e-05,
+      "loss": 0.6529,
+      "step": 286
+    },
+    {
+      "epoch": 0.4062278839348903,
+      "grad_norm": 0.6715967469992172,
+      "learning_rate": 4.805876180482686e-05,
+      "loss": 0.6772,
+      "step": 287
+    },
+    {
+      "epoch": 0.40764331210191085,
+      "grad_norm": 0.4037300470693345,
+      "learning_rate": 4.803252885624344e-05,
+      "loss": 0.5825,
+      "step": 288
+    },
+    {
+      "epoch": 0.40905874026893135,
+      "grad_norm": 0.6251534477877752,
+      "learning_rate": 4.800629590766002e-05,
+      "loss": 0.6897,
+      "step": 289
+    },
+    {
+      "epoch": 0.41047416843595186,
+      "grad_norm": 0.5295316359819032,
+      "learning_rate": 4.79800629590766e-05,
+      "loss": 0.6591,
+      "step": 290
+    },
+    {
+      "epoch": 0.4118895966029724,
+      "grad_norm": 0.43307943598387333,
+      "learning_rate": 4.795383001049318e-05,
+      "loss": 0.6226,
+      "step": 291
+    },
+    {
+      "epoch": 0.4133050247699929,
+      "grad_norm": 1.2645789414674928,
+      "learning_rate": 4.7927597061909764e-05,
+      "loss": 0.5863,
+      "step": 292
+    },
+    {
+      "epoch": 0.41472045293701343,
+      "grad_norm": 0.43201269228704425,
+      "learning_rate": 4.7901364113326344e-05,
+      "loss": 0.5843,
+      "step": 293
+    },
+    {
+      "epoch": 0.416135881104034,
+      "grad_norm": 0.40349127042570915,
+      "learning_rate": 4.787513116474292e-05,
+      "loss": 0.549,
+      "step": 294
+    },
+    {
+      "epoch": 0.4175513092710545,
+      "grad_norm": 0.6492196187297071,
+      "learning_rate": 4.78488982161595e-05,
+      "loss": 0.6521,
+      "step": 295
+    },
+    {
+      "epoch": 0.418966737438075,
+      "grad_norm": 0.5361960066185653,
+      "learning_rate": 4.782266526757608e-05,
+      "loss": 0.6529,
+      "step": 296
+    },
+    {
+      "epoch": 0.42038216560509556,
+      "grad_norm": 1.9127877440573802,
+      "learning_rate": 4.779643231899266e-05,
+      "loss": 0.6502,
+      "step": 297
+    },
+    {
+      "epoch": 0.42179759377211606,
+      "grad_norm": 0.4987202952653217,
+      "learning_rate": 4.777019937040924e-05,
+      "loss": 0.5868,
+      "step": 298
+    },
+    {
+      "epoch": 0.42321302193913657,
+      "grad_norm": 0.42834190595019034,
+      "learning_rate": 4.774396642182581e-05,
+      "loss": 0.588,
+      "step": 299
+    },
+    {
+      "epoch": 0.42462845010615713,
+      "grad_norm": 0.37598238951488705,
+      "learning_rate": 4.771773347324239e-05,
+      "loss": 0.63,
+      "step": 300
+    },
+    {
+      "epoch": 0.42604387827317763,
+      "grad_norm": 0.4023043136893649,
+      "learning_rate": 4.769150052465897e-05,
+      "loss": 0.563,
+      "step": 301
+    },
+    {
+      "epoch": 0.42745930644019814,
+      "grad_norm": 1.8869422299730276,
+      "learning_rate": 4.766526757607555e-05,
+      "loss": 0.5939,
+      "step": 302
+    },
+    {
+      "epoch": 0.4288747346072187,
+      "grad_norm": 0.4570024354376153,
+      "learning_rate": 4.763903462749213e-05,
+      "loss": 0.5991,
+      "step": 303
+    },
+    {
+      "epoch": 0.4302901627742392,
+      "grad_norm": 0.33510204236751395,
+      "learning_rate": 4.7612801678908713e-05,
+      "loss": 0.5857,
+      "step": 304
+    },
+    {
+      "epoch": 0.4317055909412597,
+      "grad_norm": 4.13866712753077,
+      "learning_rate": 4.7586568730325294e-05,
+      "loss": 0.6918,
+      "step": 305
+    },
+    {
+      "epoch": 0.43312101910828027,
+      "grad_norm": 0.566238838007585,
+      "learning_rate": 4.7560335781741874e-05,
+      "loss": 0.6613,
+      "step": 306
+    },
+    {
+      "epoch": 0.4345364472753008,
+      "grad_norm": 0.4204911407406948,
+      "learning_rate": 4.753410283315845e-05,
+      "loss": 0.6573,
+      "step": 307
+    },
+    {
+      "epoch": 0.4359518754423213,
+      "grad_norm": 0.4687083421971458,
+      "learning_rate": 4.750786988457503e-05,
+      "loss": 0.6571,
+      "step": 308
+    },
+    {
+      "epoch": 0.43736730360934184,
+      "grad_norm": 5.055944341640214,
+      "learning_rate": 4.748163693599161e-05,
+      "loss": 0.6597,
+      "step": 309
+    },
+    {
+      "epoch": 0.43878273177636234,
+      "grad_norm": 0.5275330144202897,
+      "learning_rate": 4.745540398740819e-05,
+      "loss": 0.7025,
+      "step": 310
+    },
+    {
+      "epoch": 0.44019815994338285,
+      "grad_norm": 0.38739607411426685,
+      "learning_rate": 4.742917103882477e-05,
+      "loss": 0.6323,
+      "step": 311
+    },
+    {
+      "epoch": 0.4416135881104034,
+      "grad_norm": 1.5067287407388463,
+      "learning_rate": 4.740293809024134e-05,
+      "loss": 0.6731,
+      "step": 312
+    },
+    {
+      "epoch": 0.4430290162774239,
+      "grad_norm": 0.4590664765317656,
+      "learning_rate": 4.737670514165792e-05,
+      "loss": 0.6059,
+      "step": 313
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 0.38594803494368446,
+      "learning_rate": 4.73504721930745e-05,
+      "loss": 0.6469,
+      "step": 314
+    },
+    {
+      "epoch": 0.445859872611465,
+      "grad_norm": 1.6681766507590514,
+      "learning_rate": 4.732423924449108e-05,
+      "loss": 0.6949,
+      "step": 315
+    },
+    {
+      "epoch": 0.4472753007784855,
+      "grad_norm": 2.2024847217066563,
+      "learning_rate": 4.7298006295907657e-05,
+      "loss": 0.6696,
+      "step": 316
+    },
+    {
+      "epoch": 0.448690728945506,
+      "grad_norm": 0.6485924362900419,
+      "learning_rate": 4.7271773347324244e-05,
+      "loss": 0.6551,
+      "step": 317
+    },
+    {
+      "epoch": 0.45010615711252655,
+      "grad_norm": 0.32656948375873524,
+      "learning_rate": 4.7245540398740824e-05,
+      "loss": 0.5412,
+      "step": 318
+    },
+    {
+      "epoch": 0.45152158527954706,
+      "grad_norm": 0.525034119738257,
+      "learning_rate": 4.7219307450157404e-05,
+      "loss": 0.7114,
+      "step": 319
+    },
+    {
+      "epoch": 0.45293701344656756,
+      "grad_norm": 0.3910634973163426,
+      "learning_rate": 4.719307450157398e-05,
+      "loss": 0.6183,
+      "step": 320
+    },
+    {
+      "epoch": 0.4543524416135881,
+      "grad_norm": 0.6158827475228706,
+      "learning_rate": 4.716684155299056e-05,
+      "loss": 0.7741,
+      "step": 321
+    },
+    {
+      "epoch": 0.4557678697806086,
+      "grad_norm": 0.4390670439291706,
+      "learning_rate": 4.714060860440714e-05,
+      "loss": 0.6353,
+      "step": 322
+    },
+    {
+      "epoch": 0.45718329794762913,
+      "grad_norm": 0.3634832232181815,
+      "learning_rate": 4.711437565582372e-05,
+      "loss": 0.6317,
+      "step": 323
+    },
+    {
+      "epoch": 0.4585987261146497,
+      "grad_norm": 0.33099299386363096,
+      "learning_rate": 4.70881427072403e-05,
+      "loss": 0.5928,
+      "step": 324
+    },
+    {
+      "epoch": 0.4600141542816702,
+      "grad_norm": 0.7283175564058771,
+      "learning_rate": 4.706190975865687e-05,
+      "loss": 0.5132,
+      "step": 325
+    },
+    {
+      "epoch": 0.46142958244869076,
+      "grad_norm": 0.39663831734790206,
+      "learning_rate": 4.703567681007345e-05,
+      "loss": 0.5949,
+      "step": 326
+    },
+    {
+      "epoch": 0.46284501061571126,
+      "grad_norm": 0.33617445280552166,
+      "learning_rate": 4.700944386149003e-05,
+      "loss": 0.67,
+      "step": 327
+    },
+    {
+      "epoch": 0.46426043878273177,
+      "grad_norm": 0.3790314173603713,
+      "learning_rate": 4.698321091290661e-05,
+      "loss": 0.6469,
+      "step": 328
+    },
+    {
+      "epoch": 0.4656758669497523,
+      "grad_norm": 0.3641500603632123,
+      "learning_rate": 4.695697796432319e-05,
+      "loss": 0.6129,
+      "step": 329
+    },
+    {
+      "epoch": 0.46709129511677283,
+      "grad_norm": 0.38523640099062106,
+      "learning_rate": 4.693074501573977e-05,
+      "loss": 0.6352,
+      "step": 330
+    },
+    {
+      "epoch": 0.46850672328379334,
+      "grad_norm": 0.31533765441818223,
+      "learning_rate": 4.6904512067156354e-05,
+      "loss": 0.5071,
+      "step": 331
+    },
+    {
+      "epoch": 0.4699221514508139,
+      "grad_norm": 0.3607326185265803,
+      "learning_rate": 4.6878279118572934e-05,
+      "loss": 0.5643,
+      "step": 332
+    },
+    {
+      "epoch": 0.4713375796178344,
+      "grad_norm": 0.3340538630754626,
+      "learning_rate": 4.685204616998951e-05,
+      "loss": 0.6602,
+      "step": 333
+    },
+    {
+      "epoch": 0.4727530077848549,
+      "grad_norm": 0.3653458517506279,
+      "learning_rate": 4.682581322140609e-05,
+      "loss": 0.5338,
+      "step": 334
+    },
+    {
+      "epoch": 0.47416843595187547,
+      "grad_norm": 1.0688271736428423,
+      "learning_rate": 4.679958027282267e-05,
+      "loss": 0.6009,
+      "step": 335
+    },
+    {
+      "epoch": 0.47558386411889597,
+      "grad_norm": 0.3809122272722356,
+      "learning_rate": 4.677334732423925e-05,
+      "loss": 0.6423,
+      "step": 336
+    },
+    {
+      "epoch": 0.4769992922859165,
+      "grad_norm": 0.39627182381703874,
+      "learning_rate": 4.674711437565583e-05,
+      "loss": 0.6301,
+      "step": 337
+    },
+    {
+      "epoch": 0.47841472045293704,
+      "grad_norm": 0.38449241857885996,
+      "learning_rate": 4.67208814270724e-05,
+      "loss": 0.7192,
+      "step": 338
+    },
+    {
+      "epoch": 0.47983014861995754,
+      "grad_norm": 0.35204634256821776,
+      "learning_rate": 4.669464847848898e-05,
+      "loss": 0.5858,
+      "step": 339
+    },
+    {
+      "epoch": 0.48124557678697805,
+      "grad_norm": 0.3608851500498845,
+      "learning_rate": 4.666841552990556e-05,
+      "loss": 0.6616,
+      "step": 340
+    },
+    {
+      "epoch": 0.4826610049539986,
+      "grad_norm": 2.044217701010847,
+      "learning_rate": 4.664218258132214e-05,
+      "loss": 0.6232,
+      "step": 341
+    },
+    {
+      "epoch": 0.4840764331210191,
+      "grad_norm": 0.48159872669244885,
+      "learning_rate": 4.661594963273872e-05,
+      "loss": 0.6532,
+      "step": 342
+    },
+    {
+      "epoch": 0.4854918612880396,
+      "grad_norm": 0.3124760248273583,
+      "learning_rate": 4.65897166841553e-05,
+      "loss": 0.6304,
+      "step": 343
+    },
+    {
+      "epoch": 0.4869072894550602,
+      "grad_norm": 1.1543392767277225,
+      "learning_rate": 4.656348373557188e-05,
+      "loss": 0.5875,
+      "step": 344
+    },
+    {
+      "epoch": 0.4883227176220807,
+      "grad_norm": 0.3944218685357484,
+      "learning_rate": 4.6537250786988465e-05,
+      "loss": 0.6112,
+      "step": 345
+    },
+    {
+      "epoch": 0.4897381457891012,
+      "grad_norm": 0.3257422007112123,
+      "learning_rate": 4.651101783840504e-05,
+      "loss": 0.5728,
+      "step": 346
+    },
+    {
+      "epoch": 0.49115357395612175,
+      "grad_norm": 1.1660986725076619,
+      "learning_rate": 4.648478488982162e-05,
+      "loss": 0.668,
+      "step": 347
+    },
+    {
+      "epoch": 0.49256900212314225,
+      "grad_norm": 0.37555157866666583,
+      "learning_rate": 4.64585519412382e-05,
+      "loss": 0.6376,
+      "step": 348
+    },
+    {
+      "epoch": 0.49398443029016276,
+      "grad_norm": 0.37256619809919633,
+      "learning_rate": 4.643231899265478e-05,
+      "loss": 0.6579,
+      "step": 349
+    },
+    {
+      "epoch": 0.4953998584571833,
+      "grad_norm": 0.36137278553859004,
+      "learning_rate": 4.640608604407136e-05,
+      "loss": 0.6346,
+      "step": 350
+    },
+    {
+      "epoch": 0.4968152866242038,
+      "grad_norm": 0.3116835508059875,
+      "learning_rate": 4.637985309548793e-05,
+      "loss": 0.6401,
+      "step": 351
+    },
+    {
+      "epoch": 0.4982307147912243,
+      "grad_norm": 0.3651551236969446,
+      "learning_rate": 4.635362014690451e-05,
+      "loss": 0.5816,
+      "step": 352
+    },
+    {
+      "epoch": 0.4996461429582449,
+      "grad_norm": 0.3207905146349543,
+      "learning_rate": 4.632738719832109e-05,
+      "loss": 0.576,
+      "step": 353
+    },
+    {
+      "epoch": 0.5010615711252654,
+      "grad_norm": 0.3633417389596548,
+      "learning_rate": 4.6301154249737674e-05,
+      "loss": 0.5614,
+      "step": 354
+    },
+    {
+      "epoch": 0.502476999292286,
+      "grad_norm": 0.34577140815932816,
+      "learning_rate": 4.627492130115425e-05,
+      "loss": 0.6194,
+      "step": 355
+    },
+    {
+      "epoch": 0.5038924274593064,
+      "grad_norm": 0.6817177742810275,
+      "learning_rate": 4.624868835257083e-05,
+      "loss": 0.5554,
+      "step": 356
+    },
+    {
+      "epoch": 0.505307855626327,
+      "grad_norm": 0.33098759185554166,
+      "learning_rate": 4.622245540398741e-05,
+      "loss": 0.6662,
+      "step": 357
+    },
+    {
+      "epoch": 0.5067232837933475,
+      "grad_norm": 0.3091638749967989,
+      "learning_rate": 4.619622245540399e-05,
+      "loss": 0.5937,
+      "step": 358
+    },
+    {
+      "epoch": 0.508138711960368,
+      "grad_norm": 0.33686096205760013,
+      "learning_rate": 4.616998950682057e-05,
+      "loss": 0.6425,
+      "step": 359
+    },
+    {
+      "epoch": 0.5095541401273885,
+      "grad_norm": 0.4555661086258636,
+      "learning_rate": 4.614375655823715e-05,
+      "loss": 0.6441,
+      "step": 360
+    },
+    {
+      "epoch": 0.5109695682944091,
+      "grad_norm": 0.3597460163123963,
+      "learning_rate": 4.611752360965373e-05,
+      "loss": 0.62,
+      "step": 361
+    },
+    {
+      "epoch": 0.5123849964614295,
+      "grad_norm": 0.3252882049035588,
+      "learning_rate": 4.609129066107031e-05,
+      "loss": 0.6298,
+      "step": 362
+    },
+    {
+      "epoch": 0.5138004246284501,
+      "grad_norm": 0.27357662277889927,
+      "learning_rate": 4.606505771248689e-05,
+      "loss": 0.6079,
+      "step": 363
+    },
+    {
+      "epoch": 0.5152158527954707,
+      "grad_norm": 0.34345308801248725,
+      "learning_rate": 4.603882476390346e-05,
+      "loss": 0.6715,
+      "step": 364
+    },
+    {
+      "epoch": 0.5166312809624911,
+      "grad_norm": 1.1812651580481537,
+      "learning_rate": 4.601259181532004e-05,
+      "loss": 0.6276,
+      "step": 365
+    },
+    {
+      "epoch": 0.5180467091295117,
+      "grad_norm": 0.3433024108167062,
+      "learning_rate": 4.5986358866736623e-05,
+      "loss": 0.5972,
+      "step": 366
+    },
+    {
+      "epoch": 0.5194621372965322,
+      "grad_norm": 0.7489416791007744,
+      "learning_rate": 4.5960125918153204e-05,
+      "loss": 0.5834,
+      "step": 367
+    },
+    {
+      "epoch": 0.5208775654635527,
+      "grad_norm": 1.7901825570043197,
+      "learning_rate": 4.5933892969569784e-05,
+      "loss": 0.7262,
+      "step": 368
+    },
+    {
+      "epoch": 0.5222929936305732,
+      "grad_norm": 0.35535801236874504,
+      "learning_rate": 4.590766002098636e-05,
+      "loss": 0.6818,
+      "step": 369
+    },
+    {
+      "epoch": 0.5237084217975938,
+      "grad_norm": 0.32103386113441407,
+      "learning_rate": 4.588142707240294e-05,
+      "loss": 0.5779,
+      "step": 370
+    },
+    {
+      "epoch": 0.5251238499646143,
+      "grad_norm": 0.3412242384849581,
+      "learning_rate": 4.585519412381952e-05,
+      "loss": 0.6121,
+      "step": 371
+    },
+    {
+      "epoch": 0.5265392781316348,
+      "grad_norm": 0.3032846386207085,
+      "learning_rate": 4.58289611752361e-05,
+      "loss": 0.5789,
+      "step": 372
+    },
+    {
+      "epoch": 0.5279547062986554,
+      "grad_norm": 0.35028746399522914,
+      "learning_rate": 4.580272822665268e-05,
+      "loss": 0.6587,
+      "step": 373
+    },
+    {
+      "epoch": 0.5293701344656758,
+      "grad_norm": 0.3475580934754737,
+      "learning_rate": 4.577649527806926e-05,
+      "loss": 0.6348,
+      "step": 374
+    },
+    {
+      "epoch": 0.5307855626326964,
+      "grad_norm": 0.2858824540747321,
+      "learning_rate": 4.575026232948584e-05,
+      "loss": 0.5532,
+      "step": 375
+    },
+    {
+      "epoch": 0.5322009907997169,
+      "grad_norm": 0.3775123937334319,
+      "learning_rate": 4.572402938090242e-05,
+      "loss": 0.7839,
+      "step": 376
+    },
+    {
+      "epoch": 0.5336164189667374,
+      "grad_norm": 1.821003071814801,
+      "learning_rate": 4.569779643231899e-05,
+      "loss": 0.7244,
+      "step": 377
+    },
+    {
+      "epoch": 0.535031847133758,
+      "grad_norm": 0.36372118605519194,
+      "learning_rate": 4.567156348373557e-05,
+      "loss": 0.5677,
+      "step": 378
+    },
+    {
+      "epoch": 0.5364472753007785,
+      "grad_norm": 0.3392417180576534,
+      "learning_rate": 4.5645330535152154e-05,
+      "loss": 0.6502,
+      "step": 379
+    },
+    {
+      "epoch": 0.537862703467799,
+      "grad_norm": 0.32558658253497313,
+      "learning_rate": 4.5619097586568734e-05,
+      "loss": 0.607,
+      "step": 380
+    },
+    {
+      "epoch": 0.5392781316348195,
+      "grad_norm": 0.3057510804204223,
+      "learning_rate": 4.5592864637985314e-05,
+      "loss": 0.5875,
+      "step": 381
+    },
+    {
+      "epoch": 0.5406935598018401,
+      "grad_norm": 1.4205434623418252,
+      "learning_rate": 4.556663168940189e-05,
+      "loss": 0.6474,
+      "step": 382
+    },
+    {
+      "epoch": 0.5421089879688605,
+      "grad_norm": 0.43334996565829886,
+      "learning_rate": 4.554039874081847e-05,
+      "loss": 0.6331,
+      "step": 383
+    },
+    {
+      "epoch": 0.5435244161358811,
+      "grad_norm": 0.2893938355609154,
+      "learning_rate": 4.551416579223505e-05,
+      "loss": 0.6066,
+      "step": 384
+    },
+    {
+      "epoch": 0.5449398443029017,
+      "grad_norm": 0.385254742728004,
+      "learning_rate": 4.548793284365163e-05,
+      "loss": 0.561,
+      "step": 385
+    },
+    {
+      "epoch": 0.5463552724699221,
+      "grad_norm": 0.4067804940921093,
+      "learning_rate": 4.546169989506821e-05,
+      "loss": 0.6397,
+      "step": 386
+    },
+    {
+      "epoch": 0.5477707006369427,
+      "grad_norm": 0.3634827637755242,
+      "learning_rate": 4.543546694648479e-05,
+      "loss": 0.5953,
+      "step": 387
+    },
+    {
+      "epoch": 0.5491861288039632,
+      "grad_norm": 0.4063791990642425,
+      "learning_rate": 4.540923399790137e-05,
+      "loss": 0.6715,
+      "step": 388
+    },
+    {
+      "epoch": 0.5506015569709837,
+      "grad_norm": 0.36506068471085257,
+      "learning_rate": 4.538300104931795e-05,
+      "loss": 0.5811,
+      "step": 389
+    },
+    {
+      "epoch": 0.5520169851380042,
+      "grad_norm": 0.3857615944899794,
+      "learning_rate": 4.535676810073452e-05,
+      "loss": 0.5902,
+      "step": 390
+    },
+    {
+      "epoch": 0.5534324133050248,
+      "grad_norm": 0.3591246927876168,
+      "learning_rate": 4.5330535152151103e-05,
+      "loss": 0.5877,
+      "step": 391
+    },
+    {
+      "epoch": 0.5548478414720452,
+      "grad_norm": 0.32793131919774665,
+      "learning_rate": 4.5304302203567684e-05,
+      "loss": 0.5179,
+      "step": 392
+    },
+    {
+      "epoch": 0.5562632696390658,
+      "grad_norm": 0.41862081925025835,
+      "learning_rate": 4.5278069254984264e-05,
+      "loss": 0.5812,
+      "step": 393
+    },
+    {
+      "epoch": 0.5576786978060864,
+      "grad_norm": 0.37153219913252544,
+      "learning_rate": 4.5251836306400844e-05,
+      "loss": 0.6806,
+      "step": 394
+    },
+    {
+      "epoch": 0.5590941259731068,
+      "grad_norm": 0.3899375923227849,
+      "learning_rate": 4.522560335781742e-05,
+      "loss": 0.6399,
+      "step": 395
+    },
+    {
+      "epoch": 0.5605095541401274,
+      "grad_norm": 0.2931470563734505,
+      "learning_rate": 4.5199370409234e-05,
+      "loss": 0.6244,
+      "step": 396
+    },
+    {
+      "epoch": 0.5619249823071479,
+      "grad_norm": 3.92179635800638,
+      "learning_rate": 4.517313746065058e-05,
+      "loss": 0.7752,
+      "step": 397
+    },
+    {
+      "epoch": 0.5633404104741684,
+      "grad_norm": 0.5167593678135474,
+      "learning_rate": 4.514690451206716e-05,
+      "loss": 0.7345,
+      "step": 398
+    },
+    {
+      "epoch": 0.564755838641189,
+      "grad_norm": 0.3106438677475814,
+      "learning_rate": 4.512067156348373e-05,
+      "loss": 0.5603,
+      "step": 399
+    },
+    {
+      "epoch": 0.5661712668082095,
+      "grad_norm": 4.58393490326734,
+      "learning_rate": 4.509443861490032e-05,
+      "loss": 0.6884,
+      "step": 400
+    },
+    {
+      "epoch": 0.56758669497523,
+      "grad_norm": 0.6102958460821443,
+      "learning_rate": 4.50682056663169e-05,
+      "loss": 0.6524,
+      "step": 401
+    },
+    {
+      "epoch": 0.5690021231422505,
+      "grad_norm": 0.3886188068845869,
+      "learning_rate": 4.504197271773348e-05,
+      "loss": 0.5546,
+      "step": 402
+    },
+    {
+      "epoch": 0.5704175513092711,
+      "grad_norm": 0.4812854857575041,
+      "learning_rate": 4.501573976915005e-05,
+      "loss": 0.6143,
+      "step": 403
+    },
+    {
+      "epoch": 0.5718329794762915,
+      "grad_norm": 0.5122348712154553,
+      "learning_rate": 4.4989506820566634e-05,
+      "loss": 0.6373,
+      "step": 404
+    },
+    {
+      "epoch": 0.5732484076433121,
+      "grad_norm": 0.5776634035946314,
+      "learning_rate": 4.4963273871983214e-05,
+      "loss": 0.7066,
+      "step": 405
+    },
+    {
+      "epoch": 0.5746638358103326,
+      "grad_norm": 0.32137764365447785,
+      "learning_rate": 4.4937040923399794e-05,
+      "loss": 0.5542,
+      "step": 406
+    },
+    {
+      "epoch": 0.5760792639773531,
+      "grad_norm": 0.5479916180781855,
+      "learning_rate": 4.4910807974816374e-05,
+      "loss": 0.6616,
+      "step": 407
+    },
+    {
+      "epoch": 0.5774946921443737,
+      "grad_norm": 0.4386841207404295,
+      "learning_rate": 4.488457502623295e-05,
+      "loss": 0.5806,
+      "step": 408
+    },
+    {
+      "epoch": 0.5789101203113942,
+      "grad_norm": 0.2996291147474398,
+      "learning_rate": 4.485834207764953e-05,
+      "loss": 0.6317,
+      "step": 409
+    },
+    {
+      "epoch": 0.5803255484784147,
+      "grad_norm": 0.423626648128835,
+      "learning_rate": 4.483210912906611e-05,
+      "loss": 0.5113,
+      "step": 410
+    },
+    {
+      "epoch": 0.5817409766454352,
+      "grad_norm": 0.37350221595148414,
+      "learning_rate": 4.480587618048269e-05,
+      "loss": 0.5596,
+      "step": 411
+    },
+    {
+      "epoch": 0.5831564048124558,
+      "grad_norm": 0.31258204461663813,
+      "learning_rate": 4.477964323189926e-05,
+      "loss": 0.561,
+      "step": 412
+    },
+    {
+      "epoch": 0.5845718329794763,
+      "grad_norm": 0.42453118149870256,
+      "learning_rate": 4.475341028331584e-05,
+      "loss": 0.6642,
+      "step": 413
+    },
+    {
+      "epoch": 0.5859872611464968,
+      "grad_norm": 0.3307633002928578,
+      "learning_rate": 4.472717733473243e-05,
+      "loss": 0.6141,
+      "step": 414
+    },
+    {
+      "epoch": 0.5874026893135174,
+      "grad_norm": 0.32803157792209847,
+      "learning_rate": 4.470094438614901e-05,
+      "loss": 0.4943,
+      "step": 415
+    },
+    {
+      "epoch": 0.5888181174805379,
+      "grad_norm": 0.2952618534431419,
+      "learning_rate": 4.4674711437565583e-05,
+      "loss": 0.5624,
+      "step": 416
+    },
+    {
+      "epoch": 0.5902335456475584,
+      "grad_norm": 0.32571174942874237,
+      "learning_rate": 4.4648478488982164e-05,
+      "loss": 0.5794,
+      "step": 417
+    },
+    {
+      "epoch": 0.5916489738145789,
+      "grad_norm": 0.3121943905539143,
+      "learning_rate": 4.4622245540398744e-05,
+      "loss": 0.5624,
+      "step": 418
+    },
+    {
+      "epoch": 0.5930644019815995,
+      "grad_norm": 0.31577278636640205,
+      "learning_rate": 4.4596012591815324e-05,
+      "loss": 0.6004,
+      "step": 419
+    },
+    {
+      "epoch": 0.5944798301486199,
+      "grad_norm": 0.4489394982468287,
+      "learning_rate": 4.4569779643231905e-05,
+      "loss": 0.6909,
+      "step": 420
+    },
+    {
+      "epoch": 0.5958952583156405,
+      "grad_norm": 1.6471017152087755,
+      "learning_rate": 4.454354669464848e-05,
+      "loss": 0.7002,
+      "step": 421
+    },
+    {
+      "epoch": 0.5973106864826611,
+      "grad_norm": 0.3439433119149739,
+      "learning_rate": 4.451731374606506e-05,
+      "loss": 0.492,
+      "step": 422
+    },
+    {
+      "epoch": 0.5987261146496815,
+      "grad_norm": 0.3266215336524973,
+      "learning_rate": 4.449108079748164e-05,
+      "loss": 0.6881,
+      "step": 423
+    },
+    {
+      "epoch": 0.6001415428167021,
+      "grad_norm": 0.2864170199424571,
+      "learning_rate": 4.446484784889822e-05,
+      "loss": 0.6268,
+      "step": 424
+    },
+    {
+      "epoch": 0.6015569709837226,
+      "grad_norm": 0.3395106503408427,
+      "learning_rate": 4.443861490031479e-05,
+      "loss": 0.5892,
+      "step": 425
+    },
+    {
+      "epoch": 0.6029723991507431,
+      "grad_norm": 0.3257927040748985,
+      "learning_rate": 4.441238195173137e-05,
+      "loss": 0.5565,
+      "step": 426
+    },
+    {
+      "epoch": 0.6043878273177636,
+      "grad_norm": 0.26864176496804787,
+      "learning_rate": 4.438614900314795e-05,
+      "loss": 0.5998,
+      "step": 427
+    },
+    {
+      "epoch": 0.6058032554847842,
+      "grad_norm": 0.298872589737985,
+      "learning_rate": 4.435991605456454e-05,
+      "loss": 0.5388,
+      "step": 428
+    },
+    {
+      "epoch": 0.6072186836518046,
+      "grad_norm": 0.3285567835993635,
+      "learning_rate": 4.4333683105981114e-05,
+      "loss": 0.5981,
+      "step": 429
+    },
+    {
+      "epoch": 0.6086341118188252,
+      "grad_norm": 0.2935869880085779,
+      "learning_rate": 4.4307450157397694e-05,
+      "loss": 0.6392,
+      "step": 430
+    },
+    {
+      "epoch": 0.6100495399858458,
+      "grad_norm": 0.311961309731244,
+      "learning_rate": 4.4281217208814274e-05,
+      "loss": 0.5346,
+      "step": 431
+    },
+    {
+      "epoch": 0.6114649681528662,
+      "grad_norm": 0.2937715983397586,
+      "learning_rate": 4.4254984260230854e-05,
+      "loss": 0.513,
+      "step": 432
+    },
+    {
+      "epoch": 0.6128803963198868,
+      "grad_norm": 0.3086091462694762,
+      "learning_rate": 4.4228751311647435e-05,
+      "loss": 0.6774,
+      "step": 433
+    },
+    {
+      "epoch": 0.6142958244869073,
+      "grad_norm": 0.3137400042943637,
+      "learning_rate": 4.420251836306401e-05,
+      "loss": 0.6156,
+      "step": 434
+    },
+    {
+      "epoch": 0.6157112526539278,
+      "grad_norm": 0.301249268154309,
+      "learning_rate": 4.417628541448059e-05,
+      "loss": 0.6278,
+      "step": 435
+    },
+    {
+      "epoch": 0.6171266808209483,
+      "grad_norm": 1.2953455962994163,
+      "learning_rate": 4.415005246589717e-05,
+      "loss": 0.6142,
+      "step": 436
+    },
+    {
+      "epoch": 0.6185421089879689,
+      "grad_norm": 0.3825945056357451,
+      "learning_rate": 4.412381951731375e-05,
+      "loss": 0.5328,
+      "step": 437
+    },
+    {
+      "epoch": 0.6199575371549894,
+      "grad_norm": 0.34362025288757764,
+      "learning_rate": 4.409758656873032e-05,
+      "loss": 0.5534,
+      "step": 438
+    },
+    {
+      "epoch": 0.6213729653220099,
+      "grad_norm": 0.3020577660807664,
+      "learning_rate": 4.40713536201469e-05,
+      "loss": 0.6801,
+      "step": 439
+    },
+    {
+      "epoch": 0.6227883934890305,
+      "grad_norm": 0.3773050366602719,
+      "learning_rate": 4.404512067156348e-05,
+      "loss": 0.5966,
+      "step": 440
+    },
+    {
+      "epoch": 0.6242038216560509,
+      "grad_norm": 3.7089677265124594,
+      "learning_rate": 4.401888772298007e-05,
+      "loss": 0.8189,
+      "step": 441
+    },
+    {
+      "epoch": 0.6256192498230715,
+      "grad_norm": 0.4274436279207135,
+      "learning_rate": 4.3992654774396644e-05,
+      "loss": 0.514,
+      "step": 442
+    },
+    {
+      "epoch": 0.627034677990092,
+      "grad_norm": 0.36160488766019305,
+      "learning_rate": 4.3966421825813224e-05,
+      "loss": 0.5859,
+      "step": 443
+    },
+    {
+      "epoch": 0.6284501061571125,
+      "grad_norm": 0.3227055099211922,
+      "learning_rate": 4.3940188877229804e-05,
+      "loss": 0.5393,
+      "step": 444
+    },
+    {
+      "epoch": 0.6298655343241331,
+      "grad_norm": 0.6724594497481464,
+      "learning_rate": 4.3913955928646385e-05,
+      "loss": 0.6873,
+      "step": 445
+    },
+    {
+      "epoch": 0.6312809624911536,
+      "grad_norm": 0.3723175424196791,
+      "learning_rate": 4.3887722980062965e-05,
+      "loss": 0.5801,
+      "step": 446
+    },
+    {
+      "epoch": 0.6326963906581741,
+      "grad_norm": 0.30694921528390495,
+      "learning_rate": 4.386149003147954e-05,
+      "loss": 0.6157,
+      "step": 447
+    },
+    {
+      "epoch": 0.6341118188251946,
+      "grad_norm": 0.4225504956814101,
+      "learning_rate": 4.383525708289612e-05,
+      "loss": 0.5958,
+      "step": 448
+    },
+    {
+      "epoch": 0.6355272469922152,
+      "grad_norm": 0.3464379091821917,
+      "learning_rate": 4.38090241343127e-05,
+      "loss": 0.6272,
+      "step": 449
+    },
+    {
+      "epoch": 0.6369426751592356,
+      "grad_norm": 0.29965769968222034,
+      "learning_rate": 4.378279118572928e-05,
+      "loss": 0.6077,
+      "step": 450
+    },
+    {
+      "epoch": 0.6383581033262562,
+      "grad_norm": 0.3576403182854974,
+      "learning_rate": 4.375655823714585e-05,
+      "loss": 0.6889,
+      "step": 451
+    },
+    {
+      "epoch": 0.6397735314932768,
+      "grad_norm": 0.3809053842975852,
+      "learning_rate": 4.373032528856243e-05,
+      "loss": 0.6196,
+      "step": 452
+    },
+    {
+      "epoch": 0.6411889596602972,
+      "grad_norm": 0.2609993967316393,
+      "learning_rate": 4.370409233997901e-05,
+      "loss": 0.5372,
+      "step": 453
+    },
+    {
+      "epoch": 0.6426043878273178,
+      "grad_norm": 0.33683959705079014,
+      "learning_rate": 4.3677859391395594e-05,
+      "loss": 0.6254,
+      "step": 454
+    },
+    {
+      "epoch": 0.6440198159943383,
+      "grad_norm": 0.33712328791082974,
+      "learning_rate": 4.3651626442812174e-05,
+      "loss": 0.6412,
+      "step": 455
+    },
+    {
+      "epoch": 0.6454352441613588,
+      "grad_norm": 0.29273668833002314,
+      "learning_rate": 4.3625393494228754e-05,
+      "loss": 0.6813,
+      "step": 456
+    },
+    {
+      "epoch": 0.6468506723283793,
+      "grad_norm": 0.2884106844348947,
+      "learning_rate": 4.3599160545645335e-05,
+      "loss": 0.6326,
+      "step": 457
+    },
+    {
+      "epoch": 0.6482661004953999,
+      "grad_norm": 0.3008543789946977,
+      "learning_rate": 4.3572927597061915e-05,
+      "loss": 0.5453,
+      "step": 458
+    },
+    {
+      "epoch": 0.6496815286624203,
+      "grad_norm": 10.35738969499521,
+      "learning_rate": 4.3546694648478495e-05,
+      "loss": 1.025,
+      "step": 459
+    },
+    {
+      "epoch": 0.6510969568294409,
+      "grad_norm": 0.3714989087251277,
+      "learning_rate": 4.352046169989507e-05,
+      "loss": 0.6139,
+      "step": 460
+    },
+    {
+      "epoch": 0.6525123849964615,
+      "grad_norm": 0.2980789729774163,
+      "learning_rate": 4.349422875131165e-05,
+      "loss": 0.537,
+      "step": 461
+    },
+    {
+      "epoch": 0.6539278131634819,
+      "grad_norm": 0.3303567136210746,
+      "learning_rate": 4.346799580272823e-05,
+      "loss": 0.6871,
+      "step": 462
+    },
+    {
+      "epoch": 0.6553432413305025,
+      "grad_norm": 0.30196589524687256,
+      "learning_rate": 4.344176285414481e-05,
+      "loss": 0.5793,
+      "step": 463
+    },
+    {
+      "epoch": 0.656758669497523,
+      "grad_norm": 0.45188015017130356,
+      "learning_rate": 4.341552990556138e-05,
+      "loss": 0.482,
+      "step": 464
+    },
+    {
+      "epoch": 0.6581740976645435,
+      "grad_norm": 0.3025850006337929,
+      "learning_rate": 4.338929695697796e-05,
+      "loss": 0.6142,
+      "step": 465
+    },
+    {
+      "epoch": 0.659589525831564,
+      "grad_norm": 0.5015898326623891,
+      "learning_rate": 4.3363064008394544e-05,
+      "loss": 0.6674,
+      "step": 466
+    },
+    {
+      "epoch": 0.6610049539985846,
+      "grad_norm": 0.3145802398243886,
+      "learning_rate": 4.3336831059811124e-05,
+      "loss": 0.581,
+      "step": 467
+    },
+    {
+      "epoch": 0.6624203821656051,
+      "grad_norm": 0.27586547369610387,
+      "learning_rate": 4.3310598111227704e-05,
+      "loss": 0.6641,
+      "step": 468
+    },
+    {
+      "epoch": 0.6638358103326256,
+      "grad_norm": 0.33843075842140036,
+      "learning_rate": 4.3284365162644284e-05,
+      "loss": 0.6853,
+      "step": 469
+    },
+    {
+      "epoch": 0.6652512384996462,
+      "grad_norm": 0.2765663953850667,
+      "learning_rate": 4.3258132214060865e-05,
+      "loss": 0.6237,
+      "step": 470
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.2898526449010812,
+      "learning_rate": 4.3231899265477445e-05,
+      "loss": 0.5708,
+      "step": 471
+    },
+    {
+      "epoch": 0.6680820948336872,
+      "grad_norm": 0.27564496200589295,
+      "learning_rate": 4.3205666316894025e-05,
+      "loss": 0.5657,
+      "step": 472
+    },
+    {
+      "epoch": 0.6694975230007078,
+      "grad_norm": 0.25175800000637805,
+      "learning_rate": 4.31794333683106e-05,
+      "loss": 0.6057,
+      "step": 473
+    },
+    {
+      "epoch": 0.6709129511677282,
+      "grad_norm": 0.322554869858225,
+      "learning_rate": 4.315320041972718e-05,
+      "loss": 0.6661,
+      "step": 474
+    },
+    {
+      "epoch": 0.6723283793347488,
+      "grad_norm": 0.28101804468399305,
+      "learning_rate": 4.312696747114376e-05,
+      "loss": 0.5861,
+      "step": 475
+    },
+    {
+      "epoch": 0.6737438075017693,
+      "grad_norm": 0.29168395505013195,
+      "learning_rate": 4.310073452256034e-05,
+      "loss": 0.6473,
+      "step": 476
+    },
+    {
+      "epoch": 0.6751592356687898,
+      "grad_norm": 0.26749351847887926,
+      "learning_rate": 4.307450157397691e-05,
+      "loss": 0.5958,
+      "step": 477
+    },
+    {
+      "epoch": 0.6765746638358103,
+      "grad_norm": 0.3135420128834056,
+      "learning_rate": 4.304826862539349e-05,
+      "loss": 0.5514,
+      "step": 478
+    },
+    {
+      "epoch": 0.6779900920028309,
+      "grad_norm": 0.25355647122491026,
+      "learning_rate": 4.3022035676810074e-05,
+      "loss": 0.5944,
+      "step": 479
+    },
+    {
+      "epoch": 0.6794055201698513,
+      "grad_norm": 0.27688273093419147,
+      "learning_rate": 4.2995802728226654e-05,
+      "loss": 0.589,
+      "step": 480
+    },
+    {
+      "epoch": 0.6808209483368719,
+      "grad_norm": 0.2964986987389644,
+      "learning_rate": 4.2969569779643234e-05,
+      "loss": 0.5189,
+      "step": 481
+    },
+    {
+      "epoch": 0.6822363765038925,
+      "grad_norm": 0.3004168997919821,
+      "learning_rate": 4.294333683105981e-05,
+      "loss": 0.6814,
+      "step": 482
+    },
+    {
+      "epoch": 0.6836518046709129,
+      "grad_norm": 0.2717886113590949,
+      "learning_rate": 4.2917103882476395e-05,
+      "loss": 0.5767,
+      "step": 483
+    },
+    {
+      "epoch": 0.6850672328379335,
+      "grad_norm": 0.2539767960016265,
+      "learning_rate": 4.2890870933892975e-05,
+      "loss": 0.6011,
+      "step": 484
+    },
+    {
+      "epoch": 0.686482661004954,
+      "grad_norm": 0.7440050555777202,
+      "learning_rate": 4.2864637985309555e-05,
+      "loss": 0.5521,
+      "step": 485
+    },
+    {
+      "epoch": 0.6878980891719745,
+      "grad_norm": 0.29647620149339093,
+      "learning_rate": 4.283840503672613e-05,
+      "loss": 0.6721,
+      "step": 486
+    },
+    {
+      "epoch": 0.689313517338995,
+      "grad_norm": 0.2963743861972195,
+      "learning_rate": 4.281217208814271e-05,
+      "loss": 0.6564,
+      "step": 487
+    },
+    {
+      "epoch": 0.6907289455060156,
+      "grad_norm": 0.30060259079097623,
+      "learning_rate": 4.278593913955929e-05,
+      "loss": 0.6195,
+      "step": 488
+    },
+    {
+      "epoch": 0.692144373673036,
+      "grad_norm": 0.25412311784703495,
+      "learning_rate": 4.275970619097587e-05,
+      "loss": 0.5647,
+      "step": 489
+    },
+    {
+      "epoch": 0.6935598018400566,
+      "grad_norm": 0.35278688480809256,
+      "learning_rate": 4.273347324239245e-05,
+      "loss": 0.5898,
+      "step": 490
+    },
+    {
+      "epoch": 0.6949752300070772,
+      "grad_norm": 0.2862876117817953,
+      "learning_rate": 4.2707240293809024e-05,
+      "loss": 0.591,
+      "step": 491
+    },
+    {
+      "epoch": 0.6963906581740976,
+      "grad_norm": 0.2818896648532148,
+      "learning_rate": 4.2681007345225604e-05,
+      "loss": 0.5724,
+      "step": 492
+    },
+    {
+      "epoch": 0.6978060863411182,
+      "grad_norm": 0.28125410229184444,
+      "learning_rate": 4.2654774396642184e-05,
+      "loss": 0.6174,
+      "step": 493
+    },
+    {
+      "epoch": 0.6992215145081387,
+      "grad_norm": 0.2744705379358727,
+      "learning_rate": 4.2628541448058764e-05,
+      "loss": 0.5892,
+      "step": 494
+    },
+    {
+      "epoch": 0.7006369426751592,
+      "grad_norm": 0.2741434669225602,
+      "learning_rate": 4.260230849947534e-05,
+      "loss": 0.5604,
+      "step": 495
+    },
+    {
+      "epoch": 0.7020523708421798,
+      "grad_norm": 0.32668510179495774,
+      "learning_rate": 4.257607555089192e-05,
+      "loss": 0.5718,
+      "step": 496
+    },
+    {
+      "epoch": 0.7034677990092003,
+      "grad_norm": 0.3047620022933404,
+      "learning_rate": 4.2549842602308505e-05,
+      "loss": 0.5691,
+      "step": 497
+    },
+    {
+      "epoch": 0.7048832271762208,
+      "grad_norm": 0.28241567675152757,
+      "learning_rate": 4.2523609653725086e-05,
+      "loss": 0.6505,
+      "step": 498
+    },
+    {
+      "epoch": 0.7062986553432413,
+      "grad_norm": 0.3014000437347817,
+      "learning_rate": 4.249737670514166e-05,
+      "loss": 0.5782,
+      "step": 499
+    },
+    {
+      "epoch": 0.7077140835102619,
+      "grad_norm": 0.23099688650678965,
+      "learning_rate": 4.247114375655824e-05,
+      "loss": 0.5292,
+      "step": 500
+    },
+    {
+      "epoch": 0.7091295116772823,
+      "grad_norm": 0.288512470003529,
+      "learning_rate": 4.244491080797482e-05,
+      "loss": 0.5417,
+      "step": 501
+    },
+    {
+      "epoch": 0.7105449398443029,
+      "grad_norm": 0.2890211026356785,
+      "learning_rate": 4.24186778593914e-05,
+      "loss": 0.6717,
+      "step": 502
+    },
+    {
+      "epoch": 0.7119603680113235,
+      "grad_norm": 0.2603112959698144,
+      "learning_rate": 4.239244491080798e-05,
+      "loss": 0.5458,
+      "step": 503
+    },
+    {
+      "epoch": 0.7133757961783439,
+      "grad_norm": 0.2755082295336734,
+      "learning_rate": 4.2366211962224554e-05,
+      "loss": 0.5793,
+      "step": 504
+    },
+    {
+      "epoch": 0.7147912243453645,
+      "grad_norm": 0.2754622975959251,
+      "learning_rate": 4.2339979013641134e-05,
+      "loss": 0.5343,
+      "step": 505
+    },
+    {
+      "epoch": 0.716206652512385,
+      "grad_norm": 0.27011894670220055,
+      "learning_rate": 4.2313746065057714e-05,
+      "loss": 0.6084,
+      "step": 506
+    },
+    {
+      "epoch": 0.7176220806794055,
+      "grad_norm": 0.2696120852707313,
+      "learning_rate": 4.2287513116474295e-05,
+      "loss": 0.6192,
+      "step": 507
+    },
+    {
+      "epoch": 0.719037508846426,
+      "grad_norm": 0.24868838929558423,
+      "learning_rate": 4.226128016789087e-05,
+      "loss": 0.5834,
+      "step": 508
+    },
+    {
+      "epoch": 0.7204529370134466,
+      "grad_norm": 0.2799818580148587,
+      "learning_rate": 4.223504721930745e-05,
+      "loss": 0.619,
+      "step": 509
+    },
+    {
+      "epoch": 0.721868365180467,
+      "grad_norm": 0.24896768145718445,
+      "learning_rate": 4.2208814270724035e-05,
+      "loss": 0.4383,
+      "step": 510
+    },
+    {
+      "epoch": 0.7232837933474876,
+      "grad_norm": 2.784964837059265,
+      "learning_rate": 4.2182581322140616e-05,
+      "loss": 0.6158,
+      "step": 511
+    },
+    {
+      "epoch": 0.7246992215145082,
+      "grad_norm": 0.3288428985349652,
+      "learning_rate": 4.215634837355719e-05,
+      "loss": 0.5824,
+      "step": 512
+    },
+    {
+      "epoch": 0.7261146496815286,
+      "grad_norm": 0.270381517889225,
+      "learning_rate": 4.213011542497377e-05,
+      "loss": 0.6237,
+      "step": 513
+    },
+    {
+      "epoch": 0.7275300778485492,
+      "grad_norm": 0.2887367100459008,
+      "learning_rate": 4.210388247639035e-05,
+      "loss": 0.5857,
+      "step": 514
+    },
+    {
+      "epoch": 0.7289455060155697,
+      "grad_norm": 0.31003369966092637,
+      "learning_rate": 4.207764952780693e-05,
+      "loss": 0.5976,
+      "step": 515
+    },
+    {
+      "epoch": 0.7303609341825902,
+      "grad_norm": 0.28351506632352835,
+      "learning_rate": 4.205141657922351e-05,
+      "loss": 0.52,
+      "step": 516
+    },
+    {
+      "epoch": 0.7317763623496107,
+      "grad_norm": 0.3356356489905326,
+      "learning_rate": 4.2025183630640084e-05,
+      "loss": 0.6782,
+      "step": 517
+    },
+    {
+      "epoch": 0.7331917905166313,
+      "grad_norm": 0.2765878293984698,
+      "learning_rate": 4.1998950682056664e-05,
+      "loss": 0.6152,
+      "step": 518
+    },
+    {
+      "epoch": 0.7346072186836518,
+      "grad_norm": 0.26499313822264126,
+      "learning_rate": 4.1972717733473244e-05,
+      "loss": 0.5279,
+      "step": 519
+    },
+    {
+      "epoch": 0.7360226468506723,
+      "grad_norm": 0.27460211792497674,
+      "learning_rate": 4.1946484784889825e-05,
+      "loss": 0.6255,
+      "step": 520
+    },
+    {
+      "epoch": 0.7374380750176929,
+      "grad_norm": 0.2554505704777475,
+      "learning_rate": 4.19202518363064e-05,
+      "loss": 0.5178,
+      "step": 521
+    },
+    {
+      "epoch": 0.7388535031847133,
+      "grad_norm": 0.27446639448887195,
+      "learning_rate": 4.189401888772298e-05,
+      "loss": 0.5982,
+      "step": 522
+    },
+    {
+      "epoch": 0.7402689313517339,
+      "grad_norm": 0.2753277870564774,
+      "learning_rate": 4.186778593913956e-05,
+      "loss": 0.5425,
+      "step": 523
+    },
+    {
+      "epoch": 0.7416843595187544,
+      "grad_norm": 0.30388507743331217,
+      "learning_rate": 4.1841552990556146e-05,
+      "loss": 0.6095,
+      "step": 524
+    },
+    {
+      "epoch": 0.7430997876857749,
+      "grad_norm": 0.2674547695054725,
+      "learning_rate": 4.181532004197272e-05,
+      "loss": 0.6144,
+      "step": 525
+    },
+    {
+      "epoch": 0.7445152158527955,
+      "grad_norm": 0.25908756166992924,
+      "learning_rate": 4.17890870933893e-05,
+      "loss": 0.5348,
+      "step": 526
+    },
+    {
+      "epoch": 0.745930644019816,
+      "grad_norm": 0.25834605700842117,
+      "learning_rate": 4.176285414480588e-05,
+      "loss": 0.4905,
+      "step": 527
+    },
+    {
+      "epoch": 0.7473460721868365,
+      "grad_norm": 0.2604563358719648,
+      "learning_rate": 4.173662119622246e-05,
+      "loss": 0.6299,
+      "step": 528
+    },
+    {
+      "epoch": 0.748761500353857,
+      "grad_norm": 0.2884185006269154,
+      "learning_rate": 4.171038824763904e-05,
+      "loss": 0.6399,
+      "step": 529
+    },
+    {
+      "epoch": 0.7501769285208776,
+      "grad_norm": 0.27330464240404045,
+      "learning_rate": 4.1684155299055614e-05,
+      "loss": 0.5452,
+      "step": 530
+    },
+    {
+      "epoch": 0.7515923566878981,
+      "grad_norm": 1.3715142138570997,
+      "learning_rate": 4.1657922350472194e-05,
+      "loss": 0.6225,
+      "step": 531
+    },
+    {
+      "epoch": 0.7530077848549186,
+      "grad_norm": 0.3543237791515518,
+      "learning_rate": 4.1631689401888775e-05,
+      "loss": 0.6815,
+      "step": 532
+    },
+    {
+      "epoch": 0.7544232130219392,
+      "grad_norm": 0.3221700313412834,
+      "learning_rate": 4.1605456453305355e-05,
+      "loss": 0.629,
+      "step": 533
+    },
+    {
+      "epoch": 0.7558386411889597,
+      "grad_norm": 0.30392541264587125,
+      "learning_rate": 4.157922350472193e-05,
+      "loss": 0.5199,
+      "step": 534
+    },
+    {
+      "epoch": 0.7572540693559802,
+      "grad_norm": 0.3029339931387272,
+      "learning_rate": 4.155299055613851e-05,
+      "loss": 0.5197,
+      "step": 535
+    },
+    {
+      "epoch": 0.7586694975230007,
+      "grad_norm": 0.31621698004884125,
+      "learning_rate": 4.152675760755509e-05,
+      "loss": 0.5867,
+      "step": 536
+    },
+    {
+      "epoch": 0.7600849256900213,
+      "grad_norm": 0.28814019849199607,
+      "learning_rate": 4.150052465897167e-05,
+      "loss": 0.6364,
+      "step": 537
+    },
+    {
+      "epoch": 0.7615003538570417,
+      "grad_norm": 0.29237179383094253,
+      "learning_rate": 4.147429171038825e-05,
+      "loss": 0.5636,
+      "step": 538
+    },
+    {
+      "epoch": 0.7629157820240623,
+      "grad_norm": 0.3642930281551312,
+      "learning_rate": 4.144805876180483e-05,
+      "loss": 0.692,
+      "step": 539
+    },
+    {
+      "epoch": 0.7643312101910829,
+      "grad_norm": 0.2816381389682949,
+      "learning_rate": 4.142182581322141e-05,
+      "loss": 0.6313,
+      "step": 540
+    },
+    {
+      "epoch": 0.7657466383581033,
+      "grad_norm": 0.31808906679687576,
+      "learning_rate": 4.139559286463799e-05,
+      "loss": 0.5918,
+      "step": 541
+    },
+    {
+      "epoch": 0.7671620665251239,
+      "grad_norm": 0.27256916513618773,
+      "learning_rate": 4.136935991605457e-05,
+      "loss": 0.5621,
+      "step": 542
+    },
+    {
+      "epoch": 0.7685774946921444,
+      "grad_norm": 0.2607254484094606,
+      "learning_rate": 4.1343126967471144e-05,
+      "loss": 0.5182,
+      "step": 543
+    },
+    {
+      "epoch": 0.7699929228591649,
+      "grad_norm": 0.2829216229335715,
+      "learning_rate": 4.1316894018887724e-05,
+      "loss": 0.6379,
+      "step": 544
+    },
+    {
+      "epoch": 0.7714083510261854,
+      "grad_norm": 0.2880790487930277,
+      "learning_rate": 4.1290661070304305e-05,
+      "loss": 0.6227,
+      "step": 545
+    },
+    {
+      "epoch": 0.772823779193206,
+      "grad_norm": 0.2996132822785243,
+      "learning_rate": 4.1264428121720885e-05,
+      "loss": 0.6055,
+      "step": 546
+    },
+    {
+      "epoch": 0.7742392073602264,
+      "grad_norm": 0.2538777986121411,
+      "learning_rate": 4.123819517313746e-05,
+      "loss": 0.4933,
+      "step": 547
+    },
+    {
+      "epoch": 0.775654635527247,
+      "grad_norm": 0.2946301022847751,
+      "learning_rate": 4.121196222455404e-05,
+      "loss": 0.5704,
+      "step": 548
+    },
+    {
+      "epoch": 0.7770700636942676,
+      "grad_norm": 0.2601702246669579,
+      "learning_rate": 4.118572927597062e-05,
+      "loss": 0.5303,
+      "step": 549
+    },
+    {
+      "epoch": 0.778485491861288,
+      "grad_norm": 0.28467177701497776,
+      "learning_rate": 4.11594963273872e-05,
+      "loss": 0.5815,
+      "step": 550
+    },
+    {
+      "epoch": 0.7799009200283086,
+      "grad_norm": 0.30475496699597565,
+      "learning_rate": 4.113326337880378e-05,
+      "loss": 0.623,
+      "step": 551
+    },
+    {
+      "epoch": 0.7813163481953291,
+      "grad_norm": 0.29736546055647334,
+      "learning_rate": 4.110703043022036e-05,
+      "loss": 0.6027,
+      "step": 552
+    },
+    {
+      "epoch": 0.7827317763623496,
+      "grad_norm": 0.25963842574091467,
+      "learning_rate": 4.108079748163694e-05,
+      "loss": 0.544,
+      "step": 553
+    },
+    {
+      "epoch": 0.7841472045293701,
+      "grad_norm": 0.2882644562548918,
+      "learning_rate": 4.105456453305352e-05,
+      "loss": 0.6175,
+      "step": 554
+    },
+    {
+      "epoch": 0.7855626326963907,
+      "grad_norm": 0.6580202281062054,
+      "learning_rate": 4.10283315844701e-05,
+      "loss": 0.6881,
+      "step": 555
+    },
+    {
+      "epoch": 0.7869780608634112,
+      "grad_norm": 0.2510772787710439,
+      "learning_rate": 4.1002098635886674e-05,
+      "loss": 0.5427,
+      "step": 556
+    },
+    {
+      "epoch": 0.7883934890304317,
+      "grad_norm": 0.25333209936456746,
+      "learning_rate": 4.0975865687303255e-05,
+      "loss": 0.5314,
+      "step": 557
+    },
+    {
+      "epoch": 0.7898089171974523,
+      "grad_norm": 0.3399760693466158,
+      "learning_rate": 4.0949632738719835e-05,
+      "loss": 0.5766,
+      "step": 558
+    },
+    {
+      "epoch": 0.7912243453644727,
+      "grad_norm": 0.2545404542414338,
+      "learning_rate": 4.0923399790136415e-05,
+      "loss": 0.605,
+      "step": 559
+    },
+    {
+      "epoch": 0.7926397735314933,
+      "grad_norm": 0.2689807888410843,
+      "learning_rate": 4.089716684155299e-05,
+      "loss": 0.6298,
+      "step": 560
+    },
+    {
+      "epoch": 0.7940552016985138,
+      "grad_norm": 0.26979275459836133,
+      "learning_rate": 4.087093389296957e-05,
+      "loss": 0.6411,
+      "step": 561
+    },
+    {
+      "epoch": 0.7954706298655343,
+      "grad_norm": 0.26477102023358506,
+      "learning_rate": 4.084470094438615e-05,
+      "loss": 0.6496,
+      "step": 562
+    },
+    {
+      "epoch": 0.7968860580325549,
+      "grad_norm": 0.2842035921406642,
+      "learning_rate": 4.081846799580273e-05,
+      "loss": 0.6734,
+      "step": 563
+    },
+    {
+      "epoch": 0.7983014861995754,
+      "grad_norm": 0.2759384254763282,
+      "learning_rate": 4.079223504721931e-05,
+      "loss": 0.5948,
+      "step": 564
+    },
+    {
+      "epoch": 0.7997169143665959,
+      "grad_norm": 0.29312620721071925,
+      "learning_rate": 4.076600209863588e-05,
+      "loss": 0.6588,
+      "step": 565
+    },
+    {
+      "epoch": 0.8011323425336164,
+      "grad_norm": 0.279881273844713,
+      "learning_rate": 4.073976915005247e-05,
+      "loss": 0.5988,
+      "step": 566
+    },
+    {
+      "epoch": 0.802547770700637,
+      "grad_norm": 0.24491092162063974,
+      "learning_rate": 4.071353620146905e-05,
+      "loss": 0.5586,
+      "step": 567
+    },
+    {
+      "epoch": 0.8039631988676574,
+      "grad_norm": 0.24392190118225973,
+      "learning_rate": 4.068730325288563e-05,
+      "loss": 0.6342,
+      "step": 568
+    },
+    {
+      "epoch": 0.805378627034678,
+      "grad_norm": 0.2582997869107178,
+      "learning_rate": 4.0661070304302204e-05,
+      "loss": 0.6197,
+      "step": 569
+    },
+    {
+      "epoch": 0.8067940552016986,
+      "grad_norm": 0.24443958952182887,
+      "learning_rate": 4.0634837355718785e-05,
+      "loss": 0.5583,
+      "step": 570
+    },
+    {
+      "epoch": 0.808209483368719,
+      "grad_norm": 0.25103295210737114,
+      "learning_rate": 4.0608604407135365e-05,
+      "loss": 0.5813,
+      "step": 571
+    },
+    {
+      "epoch": 0.8096249115357396,
+      "grad_norm": 0.2587091249984754,
+      "learning_rate": 4.0582371458551945e-05,
+      "loss": 0.6203,
+      "step": 572
+    },
+    {
+      "epoch": 0.8110403397027601,
+      "grad_norm": 5.077427791117972,
+      "learning_rate": 4.055613850996852e-05,
+      "loss": 0.571,
+      "step": 573
+    },
+    {
+      "epoch": 0.8124557678697806,
+      "grad_norm": 0.267569113524592,
+      "learning_rate": 4.05299055613851e-05,
+      "loss": 0.5514,
+      "step": 574
+    },
+    {
+      "epoch": 0.8138711960368011,
+      "grad_norm": 0.25344246892144145,
+      "learning_rate": 4.050367261280168e-05,
+      "loss": 0.5189,
+      "step": 575
+    },
+    {
+      "epoch": 0.8152866242038217,
+      "grad_norm": 0.26036664546568666,
+      "learning_rate": 4.047743966421826e-05,
+      "loss": 0.6321,
+      "step": 576
+    },
+    {
+      "epoch": 0.8167020523708421,
+      "grad_norm": 0.30472024610966936,
+      "learning_rate": 4.045120671563484e-05,
+      "loss": 0.6286,
+      "step": 577
+    },
+    {
+      "epoch": 0.8181174805378627,
+      "grad_norm": 0.3009657072585352,
+      "learning_rate": 4.0424973767051413e-05,
+      "loss": 0.5791,
+      "step": 578
+    },
+    {
+      "epoch": 0.8195329087048833,
+      "grad_norm": 0.30448223273011926,
+      "learning_rate": 4.0398740818467994e-05,
+      "loss": 0.6528,
+      "step": 579
+    },
+    {
+      "epoch": 0.8209483368719037,
+      "grad_norm": 0.2947969249489438,
+      "learning_rate": 4.037250786988458e-05,
+      "loss": 0.6334,
+      "step": 580
+    },
+    {
+      "epoch": 0.8223637650389243,
+      "grad_norm": 0.24462892435405953,
+      "learning_rate": 4.034627492130116e-05,
+      "loss": 0.5516,
+      "step": 581
+    },
+    {
+      "epoch": 0.8237791932059448,
+      "grad_norm": 0.3055832851870209,
+      "learning_rate": 4.0320041972717735e-05,
+      "loss": 0.6258,
+      "step": 582
+    },
+    {
+      "epoch": 0.8251946213729653,
+      "grad_norm": 0.24441620573166245,
+      "learning_rate": 4.0293809024134315e-05,
+      "loss": 0.5825,
+      "step": 583
+    },
+    {
+      "epoch": 0.8266100495399858,
+      "grad_norm": 0.24212270024095364,
+      "learning_rate": 4.0267576075550895e-05,
+      "loss": 0.5374,
+      "step": 584
+    },
+    {
+      "epoch": 0.8280254777070064,
+      "grad_norm": 0.2892849365436293,
+      "learning_rate": 4.0241343126967475e-05,
+      "loss": 0.5259,
+      "step": 585
+    },
+    {
+      "epoch": 0.8294409058740269,
+      "grad_norm": 0.26206893040947127,
+      "learning_rate": 4.021511017838405e-05,
+      "loss": 0.5325,
+      "step": 586
+    },
+    {
+      "epoch": 0.8308563340410474,
+      "grad_norm": 0.2574461087293804,
+      "learning_rate": 4.018887722980063e-05,
+      "loss": 0.5218,
+      "step": 587
+    },
+    {
+      "epoch": 0.832271762208068,
+      "grad_norm": 0.28681708948939577,
+      "learning_rate": 4.016264428121721e-05,
+      "loss": 0.6362,
+      "step": 588
+    },
+    {
+      "epoch": 0.8336871903750884,
+      "grad_norm": 1.0318753511372296,
+      "learning_rate": 4.013641133263379e-05,
+      "loss": 0.6171,
+      "step": 589
+    },
+    {
+      "epoch": 0.835102618542109,
+      "grad_norm": 0.34847852929225803,
+      "learning_rate": 4.011017838405037e-05,
+      "loss": 0.6329,
+      "step": 590
+    },
+    {
+      "epoch": 0.8365180467091295,
+      "grad_norm": 0.24148552486467043,
+      "learning_rate": 4.0083945435466944e-05,
+      "loss": 0.5311,
+      "step": 591
+    },
+    {
+      "epoch": 0.83793347487615,
+      "grad_norm": 0.3498402864926876,
+      "learning_rate": 4.0057712486883524e-05,
+      "loss": 0.709,
+      "step": 592
+    },
+    {
+      "epoch": 0.8393489030431706,
+      "grad_norm": 0.252828599776942,
+      "learning_rate": 4.003147953830011e-05,
+      "loss": 0.5973,
+      "step": 593
+    },
+    {
+      "epoch": 0.8407643312101911,
+      "grad_norm": 0.2708337319292238,
+      "learning_rate": 4.000524658971669e-05,
+      "loss": 0.5746,
+      "step": 594
+    },
+    {
+      "epoch": 0.8421797593772116,
+      "grad_norm": 0.26006848312181075,
+      "learning_rate": 3.9979013641133265e-05,
+      "loss": 0.6047,
+      "step": 595
+    },
+    {
+      "epoch": 0.8435951875442321,
+      "grad_norm": 0.25011185374558303,
+      "learning_rate": 3.9952780692549845e-05,
+      "loss": 0.5338,
+      "step": 596
+    },
+    {
+      "epoch": 0.8450106157112527,
+      "grad_norm": 0.32906968439296075,
+      "learning_rate": 3.9926547743966425e-05,
+      "loss": 0.6514,
+      "step": 597
+    },
+    {
+      "epoch": 0.8464260438782731,
+      "grad_norm": 0.25228064747265583,
+      "learning_rate": 3.9900314795383006e-05,
+      "loss": 0.5676,
+      "step": 598
+    },
+    {
+      "epoch": 0.8478414720452937,
+      "grad_norm": 0.31326927533470245,
+      "learning_rate": 3.987408184679958e-05,
+      "loss": 0.5229,
+      "step": 599
+    },
+    {
+      "epoch": 0.8492569002123143,
+      "grad_norm": 0.2435322870011402,
+      "learning_rate": 3.984784889821616e-05,
+      "loss": 0.5208,
+      "step": 600
+    },
+    {
+      "epoch": 0.8506723283793347,
+      "grad_norm": 0.33182164176895124,
+      "learning_rate": 3.982161594963274e-05,
+      "loss": 0.6113,
+      "step": 601
+    },
+    {
+      "epoch": 0.8520877565463553,
+      "grad_norm": 0.28453071974121313,
+      "learning_rate": 3.979538300104932e-05,
+      "loss": 0.5691,
+      "step": 602
+    },
+    {
+      "epoch": 0.8535031847133758,
+      "grad_norm": 0.3218593621452372,
+      "learning_rate": 3.97691500524659e-05,
+      "loss": 0.5606,
+      "step": 603
+    },
+    {
+      "epoch": 0.8549186128803963,
+      "grad_norm": 0.3040945875286931,
+      "learning_rate": 3.9742917103882474e-05,
+      "loss": 0.5704,
+      "step": 604
+    },
+    {
+      "epoch": 0.8563340410474168,
+      "grad_norm": 0.2683282837210314,
+      "learning_rate": 3.9716684155299054e-05,
+      "loss": 0.5483,
+      "step": 605
+    },
+    {
+      "epoch": 0.8577494692144374,
+      "grad_norm": 0.25839227614777754,
+      "learning_rate": 3.9690451206715634e-05,
+      "loss": 0.4989,
+      "step": 606
+    },
+    {
+      "epoch": 0.8591648973814578,
+      "grad_norm": 0.26956974149348745,
+      "learning_rate": 3.966421825813222e-05,
+      "loss": 0.64,
+      "step": 607
+    },
+    {
+      "epoch": 0.8605803255484784,
+      "grad_norm": 0.2672956498419619,
+      "learning_rate": 3.9637985309548795e-05,
+      "loss": 0.5982,
+      "step": 608
+    },
+    {
+      "epoch": 0.861995753715499,
+      "grad_norm": 0.25016570426383566,
+      "learning_rate": 3.9611752360965375e-05,
+      "loss": 0.6098,
+      "step": 609
+    },
+    {
+      "epoch": 0.8634111818825194,
+      "grad_norm": 0.27107103773859054,
+      "learning_rate": 3.9585519412381956e-05,
+      "loss": 0.6166,
+      "step": 610
+    },
+    {
+      "epoch": 0.86482661004954,
+      "grad_norm": 0.4151753982663599,
+      "learning_rate": 3.9559286463798536e-05,
+      "loss": 0.5807,
+      "step": 611
+    },
+    {
+      "epoch": 0.8662420382165605,
+      "grad_norm": 0.23976821200260148,
+      "learning_rate": 3.953305351521511e-05,
+      "loss": 0.5509,
+      "step": 612
+    },
+    {
+      "epoch": 0.867657466383581,
+      "grad_norm": 0.3520835351925965,
+      "learning_rate": 3.950682056663169e-05,
+      "loss": 0.5085,
+      "step": 613
+    },
+    {
+      "epoch": 0.8690728945506015,
+      "grad_norm": 0.2696270539363782,
+      "learning_rate": 3.948058761804827e-05,
+      "loss": 0.6259,
+      "step": 614
+    },
+    {
+      "epoch": 0.8704883227176221,
+      "grad_norm": 0.2566798072335374,
+      "learning_rate": 3.945435466946485e-05,
+      "loss": 0.5034,
+      "step": 615
+    },
+    {
+      "epoch": 0.8719037508846426,
+      "grad_norm": 0.27860917796336493,
+      "learning_rate": 3.942812172088143e-05,
+      "loss": 0.5738,
+      "step": 616
+    },
+    {
+      "epoch": 0.8733191790516631,
+      "grad_norm": 0.26826393817331345,
+      "learning_rate": 3.9401888772298004e-05,
+      "loss": 0.6201,
+      "step": 617
+    },
+    {
+      "epoch": 0.8747346072186837,
+      "grad_norm": 0.27074655141154563,
+      "learning_rate": 3.9375655823714584e-05,
+      "loss": 0.5748,
+      "step": 618
+    },
+    {
+      "epoch": 0.8761500353857041,
+      "grad_norm": 0.25848566297252273,
+      "learning_rate": 3.9349422875131165e-05,
+      "loss": 0.567,
+      "step": 619
+    },
+    {
+      "epoch": 0.8775654635527247,
+      "grad_norm": 0.2538167120734919,
+      "learning_rate": 3.9323189926547745e-05,
+      "loss": 0.645,
+      "step": 620
+    },
+    {
+      "epoch": 0.8789808917197452,
+      "grad_norm": 0.25591843624105864,
+      "learning_rate": 3.9296956977964325e-05,
+      "loss": 0.5171,
+      "step": 621
+    },
+    {
+      "epoch": 0.8803963198867657,
+      "grad_norm": 0.2711546089590064,
+      "learning_rate": 3.9270724029380905e-05,
+      "loss": 0.6277,
+      "step": 622
+    },
+    {
+      "epoch": 0.8818117480537863,
+      "grad_norm": 0.2566505626483008,
+      "learning_rate": 3.9244491080797486e-05,
+      "loss": 0.5329,
+      "step": 623
+    },
+    {
+      "epoch": 0.8832271762208068,
+      "grad_norm": 0.27195831622640293,
+      "learning_rate": 3.9218258132214066e-05,
+      "loss": 0.6002,
+      "step": 624
+    },
+    {
+      "epoch": 0.8846426043878273,
+      "grad_norm": 0.239594292957346,
+      "learning_rate": 3.9192025183630646e-05,
+      "loss": 0.6301,
+      "step": 625
+    },
+    {
+      "epoch": 0.8860580325548478,
+      "grad_norm": 0.21543925331090563,
+      "learning_rate": 3.916579223504722e-05,
+      "loss": 0.4855,
+      "step": 626
+    },
+    {
+      "epoch": 0.8874734607218684,
+      "grad_norm": 0.2627020572877223,
+      "learning_rate": 3.91395592864638e-05,
+      "loss": 0.536,
+      "step": 627
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.25994843629610076,
+      "learning_rate": 3.911332633788038e-05,
+      "loss": 0.5694,
+      "step": 628
+    },
+    {
+      "epoch": 0.8903043170559094,
+      "grad_norm": 0.2596772324229409,
+      "learning_rate": 3.908709338929696e-05,
+      "loss": 0.5503,
+      "step": 629
+    },
+    {
+      "epoch": 0.89171974522293,
+      "grad_norm": 0.24003235792566835,
+      "learning_rate": 3.9060860440713534e-05,
+      "loss": 0.5838,
+      "step": 630
+    },
+    {
+      "epoch": 0.8931351733899504,
+      "grad_norm": 0.22872137738038825,
+      "learning_rate": 3.9034627492130114e-05,
+      "loss": 0.4768,
+      "step": 631
+    },
+    {
+      "epoch": 0.894550601556971,
+      "grad_norm": 0.26240114327499175,
+      "learning_rate": 3.9008394543546695e-05,
+      "loss": 0.6221,
+      "step": 632
+    },
+    {
+      "epoch": 0.8959660297239915,
+      "grad_norm": 0.2736984168600782,
+      "learning_rate": 3.8982161594963275e-05,
+      "loss": 0.646,
+      "step": 633
+    },
+    {
+      "epoch": 0.897381457891012,
+      "grad_norm": 0.24510404040546305,
+      "learning_rate": 3.8955928646379855e-05,
+      "loss": 0.6028,
+      "step": 634
+    },
+    {
+      "epoch": 0.8987968860580325,
+      "grad_norm": 0.2487820797852633,
+      "learning_rate": 3.8929695697796436e-05,
+      "loss": 0.4868,
+      "step": 635
+    },
+    {
+      "epoch": 0.9002123142250531,
+      "grad_norm": 0.3011312076480095,
+      "learning_rate": 3.8903462749213016e-05,
+      "loss": 0.6963,
+      "step": 636
+    },
+    {
+      "epoch": 0.9016277423920736,
+      "grad_norm": 0.2703438446586986,
+      "learning_rate": 3.8877229800629596e-05,
+      "loss": 0.6708,
+      "step": 637
+    },
+    {
+      "epoch": 0.9030431705590941,
+      "grad_norm": 0.27901463098331925,
+      "learning_rate": 3.8850996852046176e-05,
+      "loss": 0.6576,
+      "step": 638
+    },
+    {
+      "epoch": 0.9044585987261147,
+      "grad_norm": 0.30297040426188737,
+      "learning_rate": 3.882476390346275e-05,
+      "loss": 0.6201,
+      "step": 639
+    },
+    {
+      "epoch": 0.9058740268931351,
+      "grad_norm": 0.2410564866480961,
+      "learning_rate": 3.879853095487933e-05,
+      "loss": 0.5523,
+      "step": 640
+    },
+    {
+      "epoch": 0.9072894550601557,
+      "grad_norm": 0.30255449339120893,
+      "learning_rate": 3.877229800629591e-05,
+      "loss": 0.5466,
+      "step": 641
+    },
+    {
+      "epoch": 0.9087048832271762,
+      "grad_norm": 0.6032084870416943,
+      "learning_rate": 3.874606505771249e-05,
+      "loss": 0.5904,
+      "step": 642
+    },
+    {
+      "epoch": 0.9101203113941967,
+      "grad_norm": 0.256190445638519,
+      "learning_rate": 3.8719832109129064e-05,
+      "loss": 0.5878,
+      "step": 643
+    },
+    {
+      "epoch": 0.9115357395612173,
+      "grad_norm": 0.3102660753676702,
+      "learning_rate": 3.8693599160545645e-05,
+      "loss": 0.6893,
+      "step": 644
+    },
+    {
+      "epoch": 0.9129511677282378,
+      "grad_norm": 0.28928475943481113,
+      "learning_rate": 3.8667366211962225e-05,
+      "loss": 0.5205,
+      "step": 645
+    },
+    {
+      "epoch": 0.9143665958952583,
+      "grad_norm": 2.5782261650324085,
+      "learning_rate": 3.8641133263378805e-05,
+      "loss": 0.8237,
+      "step": 646
+    },
+    {
+      "epoch": 0.9157820240622788,
+      "grad_norm": 0.3375425332462325,
+      "learning_rate": 3.8614900314795385e-05,
+      "loss": 0.6284,
+      "step": 647
+    },
+    {
+      "epoch": 0.9171974522292994,
+      "grad_norm": 0.5450089954920853,
+      "learning_rate": 3.858866736621196e-05,
+      "loss": 0.6052,
+      "step": 648
+    },
+    {
+      "epoch": 0.9186128803963199,
+      "grad_norm": 0.29482791588888185,
+      "learning_rate": 3.8562434417628546e-05,
+      "loss": 0.5513,
+      "step": 649
+    },
+    {
+      "epoch": 0.9200283085633404,
+      "grad_norm": 0.2790147345264433,
+      "learning_rate": 3.8536201469045126e-05,
+      "loss": 0.5789,
+      "step": 650
+    },
+    {
+      "epoch": 0.921443736730361,
+      "grad_norm": 0.2799735113368812,
+      "learning_rate": 3.8509968520461707e-05,
+      "loss": 0.5998,
+      "step": 651
+    },
+    {
+      "epoch": 0.9228591648973815,
+      "grad_norm": 0.29273992863088466,
+      "learning_rate": 3.848373557187828e-05,
+      "loss": 0.5695,
+      "step": 652
+    },
+    {
+      "epoch": 0.924274593064402,
+      "grad_norm": 1.3412545528101212,
+      "learning_rate": 3.845750262329486e-05,
+      "loss": 0.6271,
+      "step": 653
+    },
+    {
+      "epoch": 0.9256900212314225,
+      "grad_norm": 0.2960936597898938,
+      "learning_rate": 3.843126967471144e-05,
+      "loss": 0.5124,
+      "step": 654
+    },
+    {
+      "epoch": 0.9271054493984431,
+      "grad_norm": 0.24973177405132,
+      "learning_rate": 3.840503672612802e-05,
+      "loss": 0.5026,
+      "step": 655
+    },
+    {
+      "epoch": 0.9285208775654635,
+      "grad_norm": 0.31183159855056186,
+      "learning_rate": 3.8378803777544594e-05,
+      "loss": 0.5358,
+      "step": 656
+    },
+    {
+      "epoch": 0.9299363057324841,
+      "grad_norm": 0.32824518111338075,
+      "learning_rate": 3.8352570828961175e-05,
+      "loss": 0.5609,
+      "step": 657
+    },
+    {
+      "epoch": 0.9313517338995047,
+      "grad_norm": 0.2714867436973302,
+      "learning_rate": 3.8326337880377755e-05,
+      "loss": 0.499,
+      "step": 658
+    },
+    {
+      "epoch": 0.9327671620665251,
+      "grad_norm": 0.28085284253709075,
+      "learning_rate": 3.8300104931794335e-05,
+      "loss": 0.5869,
+      "step": 659
+    },
+    {
+      "epoch": 0.9341825902335457,
+      "grad_norm": 0.2870910451588877,
+      "learning_rate": 3.8273871983210916e-05,
+      "loss": 0.5582,
+      "step": 660
+    },
+    {
+      "epoch": 0.9355980184005662,
+      "grad_norm": 0.2688685777279877,
+      "learning_rate": 3.824763903462749e-05,
+      "loss": 0.5481,
+      "step": 661
+    },
+    {
+      "epoch": 0.9370134465675867,
+      "grad_norm": 0.2633867881477559,
+      "learning_rate": 3.8221406086044076e-05,
+      "loss": 0.4945,
+      "step": 662
+    },
+    {
+      "epoch": 0.9384288747346072,
+      "grad_norm": 0.24924008904882905,
+      "learning_rate": 3.8195173137460656e-05,
+      "loss": 0.5321,
+      "step": 663
+    },
+    {
+      "epoch": 0.9398443029016278,
+      "grad_norm": 0.26365210458344096,
+      "learning_rate": 3.816894018887724e-05,
+      "loss": 0.6119,
+      "step": 664
+    },
+    {
+      "epoch": 0.9412597310686482,
+      "grad_norm": 0.25688659153777077,
+      "learning_rate": 3.814270724029381e-05,
+      "loss": 0.5305,
+      "step": 665
+    },
+    {
+      "epoch": 0.9426751592356688,
+      "grad_norm": 0.2689108075584822,
+      "learning_rate": 3.811647429171039e-05,
+      "loss": 0.4616,
+      "step": 666
+    },
+    {
+      "epoch": 0.9440905874026894,
+      "grad_norm": 0.2753729809225918,
+      "learning_rate": 3.809024134312697e-05,
+      "loss": 0.538,
+      "step": 667
+    },
+    {
+      "epoch": 0.9455060155697098,
+      "grad_norm": 0.24398271598476046,
+      "learning_rate": 3.806400839454355e-05,
+      "loss": 0.5212,
+      "step": 668
+    },
+    {
+      "epoch": 0.9469214437367304,
+      "grad_norm": 0.5451156284839687,
+      "learning_rate": 3.8037775445960125e-05,
+      "loss": 0.5724,
+      "step": 669
+    },
+    {
+      "epoch": 0.9483368719037509,
+      "grad_norm": 0.25713924672797955,
+      "learning_rate": 3.8011542497376705e-05,
+      "loss": 0.5829,
+      "step": 670
+    },
+    {
+      "epoch": 0.9497523000707714,
+      "grad_norm": 0.30912466280078726,
+      "learning_rate": 3.7985309548793285e-05,
+      "loss": 0.515,
+      "step": 671
+    },
+    {
+      "epoch": 0.9511677282377919,
+      "grad_norm": 0.28775548891760677,
+      "learning_rate": 3.7959076600209865e-05,
+      "loss": 0.6289,
+      "step": 672
+    },
+    {
+      "epoch": 0.9525831564048125,
+      "grad_norm": 0.2498263173619407,
+      "learning_rate": 3.7932843651626446e-05,
+      "loss": 0.6028,
+      "step": 673
+    },
+    {
+      "epoch": 0.953998584571833,
+      "grad_norm": 0.25277311684116194,
+      "learning_rate": 3.790661070304302e-05,
+      "loss": 0.5237,
+      "step": 674
+    },
+    {
+      "epoch": 0.9554140127388535,
+      "grad_norm": 0.2739873998144345,
+      "learning_rate": 3.78803777544596e-05,
+      "loss": 0.5346,
+      "step": 675
+    },
+    {
+      "epoch": 0.9568294409058741,
+      "grad_norm": 0.29695038936503027,
+      "learning_rate": 3.7854144805876187e-05,
+      "loss": 0.6129,
+      "step": 676
+    },
+    {
+      "epoch": 0.9582448690728945,
+      "grad_norm": 0.2592036420138258,
+      "learning_rate": 3.782791185729277e-05,
+      "loss": 0.5731,
+      "step": 677
+    },
+    {
+      "epoch": 0.9596602972399151,
+      "grad_norm": 0.2558871446271218,
+      "learning_rate": 3.780167890870934e-05,
+      "loss": 0.5979,
+      "step": 678
+    },
+    {
+      "epoch": 0.9610757254069356,
+      "grad_norm": 0.2982843354022497,
+      "learning_rate": 3.777544596012592e-05,
+      "loss": 0.5924,
+      "step": 679
+    },
+    {
+      "epoch": 0.9624911535739561,
+      "grad_norm": 0.2488352289399196,
+      "learning_rate": 3.77492130115425e-05,
+      "loss": 0.5526,
+      "step": 680
+    },
+    {
+      "epoch": 0.9639065817409767,
+      "grad_norm": 0.25611812584464694,
+      "learning_rate": 3.772298006295908e-05,
+      "loss": 0.6042,
+      "step": 681
+    },
+    {
+      "epoch": 0.9653220099079972,
+      "grad_norm": 0.29555532650943567,
+      "learning_rate": 3.7696747114375655e-05,
+      "loss": 0.5963,
+      "step": 682
+    },
+    {
+      "epoch": 0.9667374380750177,
+      "grad_norm": 0.25428893954509546,
+      "learning_rate": 3.7670514165792235e-05,
+      "loss": 0.5021,
+      "step": 683
+    },
+    {
+      "epoch": 0.9681528662420382,
+      "grad_norm": 0.2244747782441335,
+      "learning_rate": 3.7644281217208815e-05,
+      "loss": 0.5129,
+      "step": 684
+    },
+    {
+      "epoch": 0.9695682944090588,
+      "grad_norm": 0.2539476280706618,
+      "learning_rate": 3.7618048268625396e-05,
+      "loss": 0.6208,
+      "step": 685
+    },
+    {
+      "epoch": 0.9709837225760792,
+      "grad_norm": 0.26689201349616526,
+      "learning_rate": 3.7591815320041976e-05,
+      "loss": 0.4655,
+      "step": 686
+    },
+    {
+      "epoch": 0.9723991507430998,
+      "grad_norm": 0.2665073304122392,
+      "learning_rate": 3.756558237145855e-05,
+      "loss": 0.5653,
+      "step": 687
+    },
+    {
+      "epoch": 0.9738145789101204,
+      "grad_norm": 0.2857822474503496,
+      "learning_rate": 3.753934942287513e-05,
+      "loss": 0.5386,
+      "step": 688
+    },
+    {
+      "epoch": 0.9752300070771408,
+      "grad_norm": 0.2924353589789488,
+      "learning_rate": 3.751311647429171e-05,
+      "loss": 0.5909,
+      "step": 689
+    },
+    {
+      "epoch": 0.9766454352441614,
+      "grad_norm": 0.2873977701123475,
+      "learning_rate": 3.74868835257083e-05,
+      "loss": 0.6711,
+      "step": 690
+    },
+    {
+      "epoch": 0.9780608634111819,
+      "grad_norm": 0.2960710414329434,
+      "learning_rate": 3.746065057712487e-05,
+      "loss": 0.5039,
+      "step": 691
+    },
+    {
+      "epoch": 0.9794762915782024,
+      "grad_norm": 0.2807847179526359,
+      "learning_rate": 3.743441762854145e-05,
+      "loss": 0.5993,
+      "step": 692
+    },
+    {
+      "epoch": 0.9808917197452229,
+      "grad_norm": 0.23634958473891998,
+      "learning_rate": 3.740818467995803e-05,
+      "loss": 0.4978,
+      "step": 693
+    },
+    {
+      "epoch": 0.9823071479122435,
+      "grad_norm": 0.2765985645444167,
+      "learning_rate": 3.738195173137461e-05,
+      "loss": 0.669,
+      "step": 694
+    },
+    {
+      "epoch": 0.9837225760792639,
+      "grad_norm": 0.26167974900207414,
+      "learning_rate": 3.7355718782791185e-05,
+      "loss": 0.504,
+      "step": 695
+    },
+    {
+      "epoch": 0.9851380042462845,
+      "grad_norm": 0.31832168867659044,
+      "learning_rate": 3.7329485834207765e-05,
+      "loss": 0.6157,
+      "step": 696
+    },
+    {
+      "epoch": 0.9865534324133051,
+      "grad_norm": 0.27502165919598376,
+      "learning_rate": 3.7303252885624345e-05,
+      "loss": 0.6101,
+      "step": 697
+    },
+    {
+      "epoch": 0.9879688605803255,
+      "grad_norm": 0.30620856960200665,
+      "learning_rate": 3.7277019937040926e-05,
+      "loss": 0.6019,
+      "step": 698
+    },
+    {
+      "epoch": 0.9893842887473461,
+      "grad_norm": 0.24607762964278576,
+      "learning_rate": 3.7250786988457506e-05,
+      "loss": 0.5015,
+      "step": 699
+    },
+    {
+      "epoch": 0.9907997169143666,
+      "grad_norm": 0.2911017746963043,
+      "learning_rate": 3.722455403987408e-05,
+      "loss": 0.5343,
+      "step": 700
+    },
+    {
+      "epoch": 0.9922151450813871,
+      "grad_norm": 0.31548858358379145,
+      "learning_rate": 3.719832109129066e-05,
+      "loss": 0.5551,
+      "step": 701
+    },
+    {
+      "epoch": 0.9936305732484076,
+      "grad_norm": 0.3039085464002624,
+      "learning_rate": 3.717208814270724e-05,
+      "loss": 0.6619,
+      "step": 702
+    },
+    {
+      "epoch": 0.9950460014154282,
+      "grad_norm": 0.24692888851369393,
+      "learning_rate": 3.714585519412382e-05,
+      "loss": 0.5174,
+      "step": 703
+    },
+    {
+      "epoch": 0.9964614295824487,
+      "grad_norm": 0.29659026256453436,
+      "learning_rate": 3.71196222455404e-05,
+      "loss": 0.4911,
+      "step": 704
+    },
+    {
+      "epoch": 0.9978768577494692,
+      "grad_norm": 0.302397588200181,
+      "learning_rate": 3.709338929695698e-05,
+      "loss": 0.5383,
+      "step": 705
+    },
+    {
+      "epoch": 0.9992922859164898,
+      "grad_norm": 0.2613748044843372,
+      "learning_rate": 3.706715634837356e-05,
+      "loss": 0.5833,
+      "step": 706
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.2613748044843372,
+      "learning_rate": 3.704092339979014e-05,
+      "loss": 0.5796,
+      "step": 707
+    },
+    {
+      "epoch": 1.0014154281670204,
+      "grad_norm": 0.5528101354566592,
+      "learning_rate": 3.7014690451206715e-05,
+      "loss": 0.4508,
+      "step": 708
+    },
+    {
+      "epoch": 1.0028308563340411,
+      "grad_norm": 0.288786666348198,
+      "learning_rate": 3.6988457502623295e-05,
+      "loss": 0.485,
+      "step": 709
+    },
+    {
+      "epoch": 1.0042462845010616,
+      "grad_norm": 0.7619106987569315,
+      "learning_rate": 3.6962224554039876e-05,
+      "loss": 0.4902,
+      "step": 710
+    },
+    {
+      "epoch": 1.005661712668082,
+      "grad_norm": 0.2996892202706279,
+      "learning_rate": 3.6935991605456456e-05,
+      "loss": 0.3895,
+      "step": 711
+    },
+    {
+      "epoch": 1.0070771408351027,
+      "grad_norm": 0.29613902550473165,
+      "learning_rate": 3.6909758656873036e-05,
+      "loss": 0.4934,
+      "step": 712
+    },
+    {
+      "epoch": 1.0084925690021231,
+      "grad_norm": 0.31978838237376206,
+      "learning_rate": 3.688352570828961e-05,
+      "loss": 0.4632,
+      "step": 713
+    },
+    {
+      "epoch": 1.0099079971691436,
+      "grad_norm": 0.3409684419101744,
+      "learning_rate": 3.685729275970619e-05,
+      "loss": 0.5109,
+      "step": 714
+    },
+    {
+      "epoch": 1.0113234253361643,
+      "grad_norm": 0.31456340994022647,
+      "learning_rate": 3.683105981112277e-05,
+      "loss": 0.5287,
+      "step": 715
+    },
+    {
+      "epoch": 1.0127388535031847,
+      "grad_norm": 0.32103586566695375,
+      "learning_rate": 3.680482686253935e-05,
+      "loss": 0.5167,
+      "step": 716
+    },
+    {
+      "epoch": 1.0141542816702052,
+      "grad_norm": 0.3320968062517569,
+      "learning_rate": 3.677859391395593e-05,
+      "loss": 0.5082,
+      "step": 717
+    },
+    {
+      "epoch": 1.0155697098372258,
+      "grad_norm": 0.30584543091711014,
+      "learning_rate": 3.675236096537251e-05,
+      "loss": 0.524,
+      "step": 718
+    },
+    {
+      "epoch": 1.0169851380042463,
+      "grad_norm": 0.3138045748342782,
+      "learning_rate": 3.672612801678909e-05,
+      "loss": 0.4462,
+      "step": 719
+    },
+    {
+      "epoch": 1.0184005661712667,
+      "grad_norm": 0.2745307135839983,
+      "learning_rate": 3.669989506820567e-05,
+      "loss": 0.5087,
+      "step": 720
+    },
+    {
+      "epoch": 1.0198159943382874,
+      "grad_norm": 5.258650257167126,
+      "learning_rate": 3.6673662119622245e-05,
+      "loss": 0.4828,
+      "step": 721
+    },
+    {
+      "epoch": 1.0212314225053079,
+      "grad_norm": 0.3287645006651341,
+      "learning_rate": 3.6647429171038825e-05,
+      "loss": 0.4432,
+      "step": 722
+    },
+    {
+      "epoch": 1.0226468506723283,
+      "grad_norm": 0.31292657982725264,
+      "learning_rate": 3.6621196222455406e-05,
+      "loss": 0.4675,
+      "step": 723
+    },
+    {
+      "epoch": 1.024062278839349,
+      "grad_norm": 0.2847961461690093,
+      "learning_rate": 3.6594963273871986e-05,
+      "loss": 0.4833,
+      "step": 724
+    },
+    {
+      "epoch": 1.0254777070063694,
+      "grad_norm": 0.31120964842943327,
+      "learning_rate": 3.6568730325288566e-05,
+      "loss": 0.5254,
+      "step": 725
+    },
+    {
+      "epoch": 1.0268931351733899,
+      "grad_norm": 0.258099839637945,
+      "learning_rate": 3.654249737670514e-05,
+      "loss": 0.4168,
+      "step": 726
+    },
+    {
+      "epoch": 1.0283085633404105,
+      "grad_norm": 0.28589238757128593,
+      "learning_rate": 3.651626442812172e-05,
+      "loss": 0.3825,
+      "step": 727
+    },
+    {
+      "epoch": 1.029723991507431,
+      "grad_norm": 0.2575331936520694,
+      "learning_rate": 3.64900314795383e-05,
+      "loss": 0.4914,
+      "step": 728
+    },
+    {
+      "epoch": 1.0311394196744514,
+      "grad_norm": 0.30838547918171205,
+      "learning_rate": 3.646379853095488e-05,
+      "loss": 0.42,
+      "step": 729
+    },
+    {
+      "epoch": 1.0325548478414721,
+      "grad_norm": 0.31046089073138644,
+      "learning_rate": 3.643756558237146e-05,
+      "loss": 0.4155,
+      "step": 730
+    },
+    {
+      "epoch": 1.0339702760084926,
+      "grad_norm": 0.26747997987601974,
+      "learning_rate": 3.641133263378804e-05,
+      "loss": 0.5073,
+      "step": 731
+    },
+    {
+      "epoch": 1.035385704175513,
+      "grad_norm": 0.30602021395282875,
+      "learning_rate": 3.638509968520462e-05,
+      "loss": 0.5294,
+      "step": 732
+    },
+    {
+      "epoch": 1.0368011323425337,
+      "grad_norm": 0.2916135518120182,
+      "learning_rate": 3.63588667366212e-05,
+      "loss": 0.4692,
+      "step": 733
+    },
+    {
+      "epoch": 1.0382165605095541,
+      "grad_norm": 0.26730632113660724,
+      "learning_rate": 3.6332633788037775e-05,
+      "loss": 0.441,
+      "step": 734
+    },
+    {
+      "epoch": 1.0396319886765746,
+      "grad_norm": 0.25696358589681917,
+      "learning_rate": 3.6306400839454356e-05,
+      "loss": 0.4457,
+      "step": 735
+    },
+    {
+      "epoch": 1.0410474168435953,
+      "grad_norm": 0.2990631968890318,
+      "learning_rate": 3.6280167890870936e-05,
+      "loss": 0.4836,
+      "step": 736
+    },
+    {
+      "epoch": 1.0424628450106157,
+      "grad_norm": 0.28660601960737864,
+      "learning_rate": 3.6253934942287516e-05,
+      "loss": 0.4691,
+      "step": 737
+    },
+    {
+      "epoch": 1.0438782731776362,
+      "grad_norm": 0.24380734366700835,
+      "learning_rate": 3.6227701993704097e-05,
+      "loss": 0.4592,
+      "step": 738
+    },
+    {
+      "epoch": 1.0452937013446568,
+      "grad_norm": 0.25241122343037414,
+      "learning_rate": 3.620146904512067e-05,
+      "loss": 0.4531,
+      "step": 739
+    },
+    {
+      "epoch": 1.0467091295116773,
+      "grad_norm": 0.2535564968999766,
+      "learning_rate": 3.617523609653725e-05,
+      "loss": 0.5335,
+      "step": 740
+    },
+    {
+      "epoch": 1.0481245576786977,
+      "grad_norm": 0.25979506943880126,
+      "learning_rate": 3.614900314795383e-05,
+      "loss": 0.4713,
+      "step": 741
+    },
+    {
+      "epoch": 1.0495399858457184,
+      "grad_norm": 0.2664669724250493,
+      "learning_rate": 3.612277019937041e-05,
+      "loss": 0.4686,
+      "step": 742
+    },
+    {
+      "epoch": 1.0509554140127388,
+      "grad_norm": 0.2509227853228027,
+      "learning_rate": 3.609653725078699e-05,
+      "loss": 0.5015,
+      "step": 743
+    },
+    {
+      "epoch": 1.0523708421797593,
+      "grad_norm": 0.24602973422666952,
+      "learning_rate": 3.6070304302203565e-05,
+      "loss": 0.5268,
+      "step": 744
+    },
+    {
+      "epoch": 1.05378627034678,
+      "grad_norm": 0.2441351012508443,
+      "learning_rate": 3.604407135362015e-05,
+      "loss": 0.4722,
+      "step": 745
+    },
+    {
+      "epoch": 1.0552016985138004,
+      "grad_norm": 0.2618929433054825,
+      "learning_rate": 3.601783840503673e-05,
+      "loss": 0.5075,
+      "step": 746
+    },
+    {
+      "epoch": 1.0566171266808209,
+      "grad_norm": 0.23628916616707335,
+      "learning_rate": 3.599160545645331e-05,
+      "loss": 0.4584,
+      "step": 747
+    },
+    {
+      "epoch": 1.0580325548478415,
+      "grad_norm": 0.22777912078546816,
+      "learning_rate": 3.5965372507869886e-05,
+      "loss": 0.4912,
+      "step": 748
+    },
+    {
+      "epoch": 1.059447983014862,
+      "grad_norm": 0.22856290218831565,
+      "learning_rate": 3.5939139559286466e-05,
+      "loss": 0.4486,
+      "step": 749
+    },
+    {
+      "epoch": 1.0608634111818824,
+      "grad_norm": 0.23825245216080632,
+      "learning_rate": 3.5912906610703046e-05,
+      "loss": 0.4592,
+      "step": 750
+    },
+    {
+      "epoch": 1.062278839348903,
+      "grad_norm": 0.2408459349082231,
+      "learning_rate": 3.588667366211963e-05,
+      "loss": 0.4928,
+      "step": 751
+    },
+    {
+      "epoch": 1.0636942675159236,
+      "grad_norm": 0.24863045510811296,
+      "learning_rate": 3.58604407135362e-05,
+      "loss": 0.4711,
+      "step": 752
+    },
+    {
+      "epoch": 1.065109695682944,
+      "grad_norm": 0.235408066450741,
+      "learning_rate": 3.583420776495278e-05,
+      "loss": 0.4264,
+      "step": 753
+    },
+    {
+      "epoch": 1.0665251238499647,
+      "grad_norm": 0.22142286921184834,
+      "learning_rate": 3.580797481636936e-05,
+      "loss": 0.4651,
+      "step": 754
+    },
+    {
+      "epoch": 1.0679405520169851,
+      "grad_norm": 0.2446926882206817,
+      "learning_rate": 3.578174186778594e-05,
+      "loss": 0.4873,
+      "step": 755
+    },
+    {
+      "epoch": 1.0693559801840056,
+      "grad_norm": 0.24042796503093813,
+      "learning_rate": 3.575550891920252e-05,
+      "loss": 0.4189,
+      "step": 756
+    },
+    {
+      "epoch": 1.0707714083510262,
+      "grad_norm": 0.22711997762409242,
+      "learning_rate": 3.5729275970619095e-05,
+      "loss": 0.4473,
+      "step": 757
+    },
+    {
+      "epoch": 1.0721868365180467,
+      "grad_norm": 0.23545924865106554,
+      "learning_rate": 3.5703043022035675e-05,
+      "loss": 0.5082,
+      "step": 758
+    },
+    {
+      "epoch": 1.0736022646850671,
+      "grad_norm": 0.23555821624394915,
+      "learning_rate": 3.567681007345226e-05,
+      "loss": 0.5034,
+      "step": 759
+    },
+    {
+      "epoch": 1.0750176928520878,
+      "grad_norm": 0.31042957046084124,
+      "learning_rate": 3.565057712486884e-05,
+      "loss": 0.5632,
+      "step": 760
+    },
+    {
+      "epoch": 1.0764331210191083,
+      "grad_norm": 0.23561315854681714,
+      "learning_rate": 3.5624344176285416e-05,
+      "loss": 0.4329,
+      "step": 761
+    },
+    {
+      "epoch": 1.0778485491861287,
+      "grad_norm": 4.634358749996599,
+      "learning_rate": 3.5598111227701996e-05,
+      "loss": 0.8723,
+      "step": 762
+    },
+    {
+      "epoch": 1.0792639773531494,
+      "grad_norm": 0.24825196084343662,
+      "learning_rate": 3.5571878279118577e-05,
+      "loss": 0.4512,
+      "step": 763
+    },
+    {
+      "epoch": 1.0806794055201698,
+      "grad_norm": 0.23154206128644247,
+      "learning_rate": 3.554564533053516e-05,
+      "loss": 0.5075,
+      "step": 764
+    },
+    {
+      "epoch": 1.0820948336871903,
+      "grad_norm": 0.22375260713189787,
+      "learning_rate": 3.551941238195173e-05,
+      "loss": 0.5031,
+      "step": 765
+    },
+    {
+      "epoch": 1.083510261854211,
+      "grad_norm": 0.2658012519094841,
+      "learning_rate": 3.549317943336831e-05,
+      "loss": 0.4712,
+      "step": 766
+    },
+    {
+      "epoch": 1.0849256900212314,
+      "grad_norm": 0.21384773833507273,
+      "learning_rate": 3.546694648478489e-05,
+      "loss": 0.4179,
+      "step": 767
+    },
+    {
+      "epoch": 1.0863411181882519,
+      "grad_norm": 0.25518402808963636,
+      "learning_rate": 3.544071353620147e-05,
+      "loss": 0.5349,
+      "step": 768
+    },
+    {
+      "epoch": 1.0877565463552725,
+      "grad_norm": 0.23346259858577614,
+      "learning_rate": 3.541448058761805e-05,
+      "loss": 0.4919,
+      "step": 769
+    },
+    {
+      "epoch": 1.089171974522293,
+      "grad_norm": 0.2654448042799127,
+      "learning_rate": 3.5388247639034625e-05,
+      "loss": 0.4578,
+      "step": 770
+    },
+    {
+      "epoch": 1.0905874026893134,
+      "grad_norm": 0.23663475395121936,
+      "learning_rate": 3.5362014690451205e-05,
+      "loss": 0.4876,
+      "step": 771
+    },
+    {
+      "epoch": 1.092002830856334,
+      "grad_norm": 0.28682779756077115,
+      "learning_rate": 3.5335781741867786e-05,
+      "loss": 0.4672,
+      "step": 772
+    },
+    {
+      "epoch": 1.0934182590233545,
+      "grad_norm": 0.23900495511199596,
+      "learning_rate": 3.530954879328437e-05,
+      "loss": 0.4966,
+      "step": 773
+    },
+    {
+      "epoch": 1.094833687190375,
+      "grad_norm": 0.22906300377082478,
+      "learning_rate": 3.5283315844700946e-05,
+      "loss": 0.4555,
+      "step": 774
+    },
+    {
+      "epoch": 1.0962491153573957,
+      "grad_norm": 0.28952521918138424,
+      "learning_rate": 3.5257082896117526e-05,
+      "loss": 0.4936,
+      "step": 775
+    },
+    {
+      "epoch": 1.0976645435244161,
+      "grad_norm": 0.25159046633454313,
+      "learning_rate": 3.523084994753411e-05,
+      "loss": 0.5253,
+      "step": 776
+    },
+    {
+      "epoch": 1.0990799716914366,
+      "grad_norm": 0.22215431109988767,
+      "learning_rate": 3.520461699895069e-05,
+      "loss": 0.4271,
+      "step": 777
+    },
+    {
+      "epoch": 1.1004953998584572,
+      "grad_norm": 0.23733069585816344,
+      "learning_rate": 3.517838405036726e-05,
+      "loss": 0.4476,
+      "step": 778
+    },
+    {
+      "epoch": 1.1019108280254777,
+      "grad_norm": 0.22833326061820136,
+      "learning_rate": 3.515215110178384e-05,
+      "loss": 0.4446,
+      "step": 779
+    },
+    {
+      "epoch": 1.1033262561924981,
+      "grad_norm": 0.9783242275899222,
+      "learning_rate": 3.512591815320042e-05,
+      "loss": 0.5131,
+      "step": 780
+    },
+    {
+      "epoch": 1.1047416843595188,
+      "grad_norm": 0.2520699915905762,
+      "learning_rate": 3.5099685204617e-05,
+      "loss": 0.5245,
+      "step": 781
+    },
+    {
+      "epoch": 1.1061571125265393,
+      "grad_norm": 0.2540029796902151,
+      "learning_rate": 3.507345225603358e-05,
+      "loss": 0.4551,
+      "step": 782
+    },
+    {
+      "epoch": 1.1075725406935597,
+      "grad_norm": 0.24500481650416778,
+      "learning_rate": 3.5047219307450155e-05,
+      "loss": 0.4209,
+      "step": 783
+    },
+    {
+      "epoch": 1.1089879688605804,
+      "grad_norm": 0.24689125325205152,
+      "learning_rate": 3.5020986358866735e-05,
+      "loss": 0.4828,
+      "step": 784
+    },
+    {
+      "epoch": 1.1104033970276008,
+      "grad_norm": 0.26383620785800466,
+      "learning_rate": 3.4994753410283316e-05,
+      "loss": 0.4429,
+      "step": 785
+    },
+    {
+      "epoch": 1.1118188251946213,
+      "grad_norm": 0.24930702957956494,
+      "learning_rate": 3.4968520461699896e-05,
+      "loss": 0.5144,
+      "step": 786
+    },
+    {
+      "epoch": 1.113234253361642,
+      "grad_norm": 0.323610419343545,
+      "learning_rate": 3.4942287513116476e-05,
+      "loss": 0.4868,
+      "step": 787
+    },
+    {
+      "epoch": 1.1146496815286624,
+      "grad_norm": 0.24451259912991474,
+      "learning_rate": 3.4916054564533057e-05,
+      "loss": 0.5213,
+      "step": 788
+    },
+    {
+      "epoch": 1.1160651096956828,
+      "grad_norm": 26.067316043931317,
+      "learning_rate": 3.488982161594964e-05,
+      "loss": 0.6876,
+      "step": 789
+    },
+    {
+      "epoch": 1.1174805378627035,
+      "grad_norm": 0.28236163541057463,
+      "learning_rate": 3.486358866736622e-05,
+      "loss": 0.51,
+      "step": 790
+    },
+    {
+      "epoch": 1.118895966029724,
+      "grad_norm": 2.266655130223754,
+      "learning_rate": 3.483735571878279e-05,
+      "loss": 0.446,
+      "step": 791
+    },
+    {
+      "epoch": 1.1203113941967444,
+      "grad_norm": 0.2963042982811455,
+      "learning_rate": 3.481112277019937e-05,
+      "loss": 0.4626,
+      "step": 792
+    },
+    {
+      "epoch": 1.121726822363765,
+      "grad_norm": 0.23579236543255203,
+      "learning_rate": 3.478488982161595e-05,
+      "loss": 0.4657,
+      "step": 793
+    },
+    {
+      "epoch": 1.1231422505307855,
+      "grad_norm": 0.3116708551094102,
+      "learning_rate": 3.475865687303253e-05,
+      "loss": 0.5268,
+      "step": 794
+    },
+    {
+      "epoch": 1.124557678697806,
+      "grad_norm": 0.25446445484842806,
+      "learning_rate": 3.473242392444911e-05,
+      "loss": 0.4926,
+      "step": 795
+    },
+    {
+      "epoch": 1.1259731068648267,
+      "grad_norm": 0.27850892606942623,
+      "learning_rate": 3.4706190975865685e-05,
+      "loss": 0.5049,
+      "step": 796
+    },
+    {
+      "epoch": 1.127388535031847,
+      "grad_norm": 0.25055938561632096,
+      "learning_rate": 3.4679958027282266e-05,
+      "loss": 0.4825,
+      "step": 797
+    },
+    {
+      "epoch": 1.1288039631988678,
+      "grad_norm": 0.22638791452627927,
+      "learning_rate": 3.4653725078698846e-05,
+      "loss": 0.4704,
+      "step": 798
+    },
+    {
+      "epoch": 1.1302193913658882,
+      "grad_norm": 0.26913458804456786,
+      "learning_rate": 3.4627492130115426e-05,
+      "loss": 0.5425,
+      "step": 799
+    },
+    {
+      "epoch": 1.1316348195329087,
+      "grad_norm": 0.2580916567814921,
+      "learning_rate": 3.4601259181532006e-05,
+      "loss": 0.4642,
+      "step": 800
+    },
+    {
+      "epoch": 1.1330502476999293,
+      "grad_norm": 0.23641384169641477,
+      "learning_rate": 3.457502623294859e-05,
+      "loss": 0.4672,
+      "step": 801
+    },
+    {
+      "epoch": 1.1344656758669498,
+      "grad_norm": 0.2494168779349966,
+      "learning_rate": 3.454879328436517e-05,
+      "loss": 0.4502,
+      "step": 802
+    },
+    {
+      "epoch": 1.1358811040339702,
+      "grad_norm": 0.22721047026479216,
+      "learning_rate": 3.452256033578175e-05,
+      "loss": 0.4378,
+      "step": 803
+    },
+    {
+      "epoch": 1.137296532200991,
+      "grad_norm": 0.285321833786431,
+      "learning_rate": 3.449632738719832e-05,
+      "loss": 0.4712,
+      "step": 804
+    },
+    {
+      "epoch": 1.1387119603680114,
+      "grad_norm": 0.2676136691460945,
+      "learning_rate": 3.44700944386149e-05,
+      "loss": 0.4391,
+      "step": 805
+    },
+    {
+      "epoch": 1.1401273885350318,
+      "grad_norm": 0.26345274892177956,
+      "learning_rate": 3.444386149003148e-05,
+      "loss": 0.574,
+      "step": 806
+    },
+    {
+      "epoch": 1.1415428167020525,
+      "grad_norm": 0.25711450287192117,
+      "learning_rate": 3.441762854144806e-05,
+      "loss": 0.4732,
+      "step": 807
+    },
+    {
+      "epoch": 1.142958244869073,
+      "grad_norm": 3.871966915539433,
+      "learning_rate": 3.439139559286464e-05,
+      "loss": 0.5585,
+      "step": 808
+    },
+    {
+      "epoch": 1.1443736730360934,
+      "grad_norm": 0.2885361898507915,
+      "learning_rate": 3.4365162644281215e-05,
+      "loss": 0.4955,
+      "step": 809
+    },
+    {
+      "epoch": 1.145789101203114,
+      "grad_norm": 0.27866300020903967,
+      "learning_rate": 3.4338929695697796e-05,
+      "loss": 0.4305,
+      "step": 810
+    },
+    {
+      "epoch": 1.1472045293701345,
+      "grad_norm": 0.23500417618489086,
+      "learning_rate": 3.4312696747114376e-05,
+      "loss": 0.8119,
+      "step": 811
+    },
+    {
+      "epoch": 1.148619957537155,
+      "grad_norm": 6.094213492209398,
+      "learning_rate": 3.4286463798530956e-05,
+      "loss": 0.4879,
+      "step": 812
+    },
+    {
+      "epoch": 1.1500353857041756,
+      "grad_norm": 0.2876658250108991,
+      "learning_rate": 3.4260230849947537e-05,
+      "loss": 0.4099,
+      "step": 813
+    },
+    {
+      "epoch": 1.151450813871196,
+      "grad_norm": 0.27228640641162283,
+      "learning_rate": 3.423399790136412e-05,
+      "loss": 0.5456,
+      "step": 814
+    },
+    {
+      "epoch": 1.1528662420382165,
+      "grad_norm": 0.23578573565103855,
+      "learning_rate": 3.42077649527807e-05,
+      "loss": 0.5166,
+      "step": 815
+    },
+    {
+      "epoch": 1.1542816702052372,
+      "grad_norm": 0.29346102651122435,
+      "learning_rate": 3.418153200419728e-05,
+      "loss": 0.5555,
+      "step": 816
+    },
+    {
+      "epoch": 1.1556970983722576,
+      "grad_norm": 0.3214253003176066,
+      "learning_rate": 3.415529905561385e-05,
+      "loss": 0.5404,
+      "step": 817
+    },
+    {
+      "epoch": 1.157112526539278,
+      "grad_norm": 0.23422729928452454,
+      "learning_rate": 3.412906610703043e-05,
+      "loss": 0.4641,
+      "step": 818
+    },
+    {
+      "epoch": 1.1585279547062988,
+      "grad_norm": 0.2511446359869489,
+      "learning_rate": 3.410283315844701e-05,
+      "loss": 0.5066,
+      "step": 819
+    },
+    {
+      "epoch": 1.1599433828733192,
+      "grad_norm": 0.27588936308706186,
+      "learning_rate": 3.407660020986359e-05,
+      "loss": 0.5084,
+      "step": 820
+    },
+    {
+      "epoch": 1.1613588110403397,
+      "grad_norm": 0.23047802744875695,
+      "learning_rate": 3.405036726128017e-05,
+      "loss": 0.4677,
+      "step": 821
+    },
+    {
+      "epoch": 1.1627742392073603,
+      "grad_norm": 0.2830873368697821,
+      "learning_rate": 3.4024134312696746e-05,
+      "loss": 0.4899,
+      "step": 822
+    },
+    {
+      "epoch": 1.1641896673743808,
+      "grad_norm": 0.2464140175950677,
+      "learning_rate": 3.3997901364113326e-05,
+      "loss": 0.4852,
+      "step": 823
+    },
+    {
+      "epoch": 1.1656050955414012,
+      "grad_norm": 0.28234016547070406,
+      "learning_rate": 3.3971668415529906e-05,
+      "loss": 0.5077,
+      "step": 824
+    },
+    {
+      "epoch": 1.167020523708422,
+      "grad_norm": 0.2865470126601044,
+      "learning_rate": 3.3945435466946486e-05,
+      "loss": 0.4079,
+      "step": 825
+    },
+    {
+      "epoch": 1.1684359518754424,
+      "grad_norm": 0.24870979695020073,
+      "learning_rate": 3.391920251836307e-05,
+      "loss": 0.5259,
+      "step": 826
+    },
+    {
+      "epoch": 1.1698513800424628,
+      "grad_norm": 0.2655929806781905,
+      "learning_rate": 3.389296956977964e-05,
+      "loss": 0.4316,
+      "step": 827
+    },
+    {
+      "epoch": 1.1712668082094835,
+      "grad_norm": 0.2465051182015438,
+      "learning_rate": 3.386673662119623e-05,
+      "loss": 0.504,
+      "step": 828
+    },
+    {
+      "epoch": 1.172682236376504,
+      "grad_norm": 0.25728893232852484,
+      "learning_rate": 3.384050367261281e-05,
+      "loss": 0.4545,
+      "step": 829
+    },
+    {
+      "epoch": 1.1740976645435244,
+      "grad_norm": 0.2534788672603512,
+      "learning_rate": 3.381427072402938e-05,
+      "loss": 0.4468,
+      "step": 830
+    },
+    {
+      "epoch": 1.175513092710545,
+      "grad_norm": 0.2702950684859484,
+      "learning_rate": 3.378803777544596e-05,
+      "loss": 0.4836,
+      "step": 831
+    },
+    {
+      "epoch": 1.1769285208775655,
+      "grad_norm": 2.1650581544740706,
+      "learning_rate": 3.376180482686254e-05,
+      "loss": 0.5522,
+      "step": 832
+    },
+    {
+      "epoch": 1.178343949044586,
+      "grad_norm": 0.26212934230121615,
+      "learning_rate": 3.373557187827912e-05,
+      "loss": 0.4916,
+      "step": 833
+    },
+    {
+      "epoch": 1.1797593772116066,
+      "grad_norm": 0.23992354272423078,
+      "learning_rate": 3.37093389296957e-05,
+      "loss": 0.4496,
+      "step": 834
+    },
+    {
+      "epoch": 1.181174805378627,
+      "grad_norm": 0.24347908114752023,
+      "learning_rate": 3.3683105981112276e-05,
+      "loss": 0.4423,
+      "step": 835
+    },
+    {
+      "epoch": 1.1825902335456475,
+      "grad_norm": 0.2481739432773745,
+      "learning_rate": 3.3656873032528856e-05,
+      "loss": 0.4634,
+      "step": 836
+    },
+    {
+      "epoch": 1.1840056617126682,
+      "grad_norm": 0.23592862661803066,
+      "learning_rate": 3.3630640083945436e-05,
+      "loss": 0.5092,
+      "step": 837
+    },
+    {
+      "epoch": 1.1854210898796886,
+      "grad_norm": 0.26981634557028705,
+      "learning_rate": 3.3604407135362017e-05,
+      "loss": 0.5233,
+      "step": 838
+    },
+    {
+      "epoch": 1.186836518046709,
+      "grad_norm": 0.27743377500140626,
+      "learning_rate": 3.35781741867786e-05,
+      "loss": 0.5167,
+      "step": 839
+    },
+    {
+      "epoch": 1.1882519462137298,
+      "grad_norm": 0.8530392452926353,
+      "learning_rate": 3.355194123819517e-05,
+      "loss": 0.4858,
+      "step": 840
+    },
+    {
+      "epoch": 1.1896673743807502,
+      "grad_norm": 0.2478597033408157,
+      "learning_rate": 3.352570828961175e-05,
+      "loss": 0.4535,
+      "step": 841
+    },
+    {
+      "epoch": 1.1910828025477707,
+      "grad_norm": 0.22109567674510855,
+      "learning_rate": 3.349947534102834e-05,
+      "loss": 0.4773,
+      "step": 842
+    },
+    {
+      "epoch": 1.1924982307147913,
+      "grad_norm": 0.24990807485377547,
+      "learning_rate": 3.347324239244491e-05,
+      "loss": 0.5202,
+      "step": 843
+    },
+    {
+      "epoch": 1.1939136588818118,
+      "grad_norm": 0.2233711639059363,
+      "learning_rate": 3.344700944386149e-05,
+      "loss": 0.4579,
+      "step": 844
+    },
+    {
+      "epoch": 1.1953290870488322,
+      "grad_norm": 0.219961031890485,
+      "learning_rate": 3.342077649527807e-05,
+      "loss": 0.4246,
+      "step": 845
+    },
+    {
+      "epoch": 1.196744515215853,
+      "grad_norm": 0.5028905087093073,
+      "learning_rate": 3.339454354669465e-05,
+      "loss": 0.5419,
+      "step": 846
+    },
+    {
+      "epoch": 1.1981599433828733,
+      "grad_norm": 0.2578553709131576,
+      "learning_rate": 3.336831059811123e-05,
+      "loss": 0.4634,
+      "step": 847
+    },
+    {
+      "epoch": 1.1995753715498938,
+      "grad_norm": 0.2312399763446513,
+      "learning_rate": 3.3342077649527806e-05,
+      "loss": 0.478,
+      "step": 848
+    },
+    {
+      "epoch": 1.2009907997169145,
+      "grad_norm": 0.2637322321499473,
+      "learning_rate": 3.3315844700944386e-05,
+      "loss": 0.57,
+      "step": 849
+    },
+    {
+      "epoch": 1.202406227883935,
+      "grad_norm": 0.24732656083357885,
+      "learning_rate": 3.3289611752360966e-05,
+      "loss": 0.4796,
+      "step": 850
+    },
+    {
+      "epoch": 1.2038216560509554,
+      "grad_norm": 0.21999558816702827,
+      "learning_rate": 3.326337880377755e-05,
+      "loss": 0.4194,
+      "step": 851
+    },
+    {
+      "epoch": 1.205237084217976,
+      "grad_norm": 0.22634734957293223,
+      "learning_rate": 3.323714585519413e-05,
+      "loss": 0.4247,
+      "step": 852
+    },
+    {
+      "epoch": 1.2066525123849965,
+      "grad_norm": 0.21455915933962133,
+      "learning_rate": 3.32109129066107e-05,
+      "loss": 0.4582,
+      "step": 853
+    },
+    {
+      "epoch": 1.208067940552017,
+      "grad_norm": 0.23973577008088373,
+      "learning_rate": 3.318467995802728e-05,
+      "loss": 0.529,
+      "step": 854
+    },
+    {
+      "epoch": 1.2094833687190376,
+      "grad_norm": 1.4575856656096844,
+      "learning_rate": 3.315844700944386e-05,
+      "loss": 0.5281,
+      "step": 855
+    },
+    {
+      "epoch": 1.210898796886058,
+      "grad_norm": 0.22286056311078778,
+      "learning_rate": 3.313221406086044e-05,
+      "loss": 0.5208,
+      "step": 856
+    },
+    {
+      "epoch": 1.2123142250530785,
+      "grad_norm": 0.24242130053002622,
+      "learning_rate": 3.310598111227702e-05,
+      "loss": 0.4632,
+      "step": 857
+    },
+    {
+      "epoch": 1.2137296532200992,
+      "grad_norm": 0.2191878726682969,
+      "learning_rate": 3.30797481636936e-05,
+      "loss": 0.4902,
+      "step": 858
+    },
+    {
+      "epoch": 1.2151450813871196,
+      "grad_norm": 0.22845347926773774,
+      "learning_rate": 3.305351521511018e-05,
+      "loss": 0.4516,
+      "step": 859
+    },
+    {
+      "epoch": 1.21656050955414,
+      "grad_norm": 0.22736406821613941,
+      "learning_rate": 3.302728226652676e-05,
+      "loss": 0.5536,
+      "step": 860
+    },
+    {
+      "epoch": 1.2179759377211608,
+      "grad_norm": 0.27494329647314425,
+      "learning_rate": 3.3001049317943336e-05,
+      "loss": 0.4804,
+      "step": 861
+    },
+    {
+      "epoch": 1.2193913658881812,
+      "grad_norm": 0.22810142235797856,
+      "learning_rate": 3.2974816369359916e-05,
+      "loss": 0.428,
+      "step": 862
+    },
+    {
+      "epoch": 1.2208067940552016,
+      "grad_norm": 0.23307785071101525,
+      "learning_rate": 3.29485834207765e-05,
+      "loss": 0.4578,
+      "step": 863
+    },
+    {
+      "epoch": 1.2222222222222223,
+      "grad_norm": 0.2452984529709648,
+      "learning_rate": 3.292235047219308e-05,
+      "loss": 0.5095,
+      "step": 864
+    },
+    {
+      "epoch": 1.2236376503892428,
+      "grad_norm": 0.23636401383941374,
+      "learning_rate": 3.289611752360966e-05,
+      "loss": 0.4137,
+      "step": 865
+    },
+    {
+      "epoch": 1.2250530785562632,
+      "grad_norm": 0.22493101535805382,
+      "learning_rate": 3.286988457502623e-05,
+      "loss": 0.4343,
+      "step": 866
+    },
+    {
+      "epoch": 1.226468506723284,
+      "grad_norm": 0.2615979917519968,
+      "learning_rate": 3.284365162644281e-05,
+      "loss": 0.4999,
+      "step": 867
+    },
+    {
+      "epoch": 1.2278839348903043,
+      "grad_norm": 0.26507045747378327,
+      "learning_rate": 3.281741867785939e-05,
+      "loss": 0.4466,
+      "step": 868
+    },
+    {
+      "epoch": 1.2292993630573248,
+      "grad_norm": 0.2570384535246848,
+      "learning_rate": 3.279118572927597e-05,
+      "loss": 0.5331,
+      "step": 869
+    },
+    {
+      "epoch": 1.2307147912243455,
+      "grad_norm": 0.29062338170307533,
+      "learning_rate": 3.276495278069255e-05,
+      "loss": 0.505,
+      "step": 870
+    },
+    {
+      "epoch": 1.232130219391366,
+      "grad_norm": 0.26476425087429617,
+      "learning_rate": 3.273871983210913e-05,
+      "loss": 0.425,
+      "step": 871
+    },
+    {
+      "epoch": 1.2335456475583864,
+      "grad_norm": 0.2477692770292864,
+      "learning_rate": 3.271248688352571e-05,
+      "loss": 0.5162,
+      "step": 872
+    },
+    {
+      "epoch": 1.234961075725407,
+      "grad_norm": 0.27488613410288,
+      "learning_rate": 3.268625393494229e-05,
+      "loss": 0.4596,
+      "step": 873
+    },
+    {
+      "epoch": 1.2363765038924275,
+      "grad_norm": 0.23168934163955254,
+      "learning_rate": 3.2660020986358866e-05,
+      "loss": 0.4323,
+      "step": 874
+    },
+    {
+      "epoch": 1.237791932059448,
+      "grad_norm": 0.24431079242225726,
+      "learning_rate": 3.2633788037775447e-05,
+      "loss": 0.4508,
+      "step": 875
+    },
+    {
+      "epoch": 1.2392073602264686,
+      "grad_norm": 0.25823573038102476,
+      "learning_rate": 3.260755508919203e-05,
+      "loss": 0.4996,
+      "step": 876
+    },
+    {
+      "epoch": 1.240622788393489,
+      "grad_norm": 0.27358168249697484,
+      "learning_rate": 3.258132214060861e-05,
+      "loss": 0.4859,
+      "step": 877
+    },
+    {
+      "epoch": 1.2420382165605095,
+      "grad_norm": 0.2626978965439729,
+      "learning_rate": 3.255508919202519e-05,
+      "loss": 0.4217,
+      "step": 878
+    },
+    {
+      "epoch": 1.2434536447275302,
+      "grad_norm": 0.26935438304307635,
+      "learning_rate": 3.252885624344176e-05,
+      "loss": 0.4972,
+      "step": 879
+    },
+    {
+      "epoch": 1.2448690728945506,
+      "grad_norm": 0.29258459138262827,
+      "learning_rate": 3.250262329485834e-05,
+      "loss": 0.4704,
+      "step": 880
+    },
+    {
+      "epoch": 1.246284501061571,
+      "grad_norm": 0.279869756393473,
+      "learning_rate": 3.247639034627492e-05,
+      "loss": 0.5122,
+      "step": 881
+    },
+    {
+      "epoch": 1.2476999292285917,
+      "grad_norm": 0.24949248682073935,
+      "learning_rate": 3.24501573976915e-05,
+      "loss": 0.4552,
+      "step": 882
+    },
+    {
+      "epoch": 1.2491153573956122,
+      "grad_norm": 0.2592993293922595,
+      "learning_rate": 3.242392444910808e-05,
+      "loss": 0.4835,
+      "step": 883
+    },
+    {
+      "epoch": 1.2505307855626326,
+      "grad_norm": 0.2531961962401023,
+      "learning_rate": 3.239769150052466e-05,
+      "loss": 0.4653,
+      "step": 884
+    },
+    {
+      "epoch": 1.2519462137296533,
+      "grad_norm": 0.24363061117308296,
+      "learning_rate": 3.237145855194124e-05,
+      "loss": 0.8861,
+      "step": 885
+    },
+    {
+      "epoch": 1.2533616418966738,
+      "grad_norm": 5.495245862133001,
+      "learning_rate": 3.234522560335782e-05,
+      "loss": 0.489,
+      "step": 886
+    },
+    {
+      "epoch": 1.2547770700636942,
+      "grad_norm": 0.2795802637952873,
+      "learning_rate": 3.2318992654774396e-05,
+      "loss": 0.4847,
+      "step": 887
+    },
+    {
+      "epoch": 1.2561924982307149,
+      "grad_norm": 0.27057315035050855,
+      "learning_rate": 3.229275970619098e-05,
+      "loss": 0.4453,
+      "step": 888
+    },
+    {
+      "epoch": 1.2576079263977353,
+      "grad_norm": 0.23049854015007026,
+      "learning_rate": 3.226652675760756e-05,
+      "loss": 0.4268,
+      "step": 889
+    },
+    {
+      "epoch": 1.2590233545647558,
+      "grad_norm": 0.286146313618306,
+      "learning_rate": 3.224029380902414e-05,
+      "loss": 0.4644,
+      "step": 890
+    },
+    {
+      "epoch": 1.2604387827317765,
+      "grad_norm": 0.24410307044469992,
+      "learning_rate": 3.221406086044072e-05,
+      "loss": 0.4784,
+      "step": 891
+    },
+    {
+      "epoch": 1.261854210898797,
+      "grad_norm": 0.23806816934844402,
+      "learning_rate": 3.218782791185729e-05,
+      "loss": 0.5156,
+      "step": 892
+    },
+    {
+      "epoch": 1.2632696390658174,
+      "grad_norm": 0.23357027315842244,
+      "learning_rate": 3.216159496327387e-05,
+      "loss": 0.4456,
+      "step": 893
+    },
+    {
+      "epoch": 1.264685067232838,
+      "grad_norm": 0.2566041334719668,
+      "learning_rate": 3.213536201469045e-05,
+      "loss": 0.4236,
+      "step": 894
+    },
+    {
+      "epoch": 1.2661004953998585,
+      "grad_norm": 0.24594304892172075,
+      "learning_rate": 3.210912906610703e-05,
+      "loss": 0.4315,
+      "step": 895
+    },
+    {
+      "epoch": 1.267515923566879,
+      "grad_norm": 0.23237072851985374,
+      "learning_rate": 3.2082896117523605e-05,
+      "loss": 0.4882,
+      "step": 896
+    },
+    {
+      "epoch": 1.2689313517338996,
+      "grad_norm": 0.2716965951301799,
+      "learning_rate": 3.205666316894019e-05,
+      "loss": 0.4854,
+      "step": 897
+    },
+    {
+      "epoch": 1.27034677990092,
+      "grad_norm": 0.24686150409198165,
+      "learning_rate": 3.203043022035677e-05,
+      "loss": 0.5459,
+      "step": 898
+    },
+    {
+      "epoch": 1.2717622080679405,
+      "grad_norm": 0.23380918880880155,
+      "learning_rate": 3.200419727177335e-05,
+      "loss": 0.4041,
+      "step": 899
+    },
+    {
+      "epoch": 1.2731776362349612,
+      "grad_norm": 0.22071068762301357,
+      "learning_rate": 3.1977964323189927e-05,
+      "loss": 0.434,
+      "step": 900
+    },
+    {
+      "epoch": 1.2745930644019816,
+      "grad_norm": 0.2585145407637671,
+      "learning_rate": 3.195173137460651e-05,
+      "loss": 0.3878,
+      "step": 901
+    },
+    {
+      "epoch": 1.276008492569002,
+      "grad_norm": 0.23289141730231533,
+      "learning_rate": 3.192549842602309e-05,
+      "loss": 0.5144,
+      "step": 902
+    },
+    {
+      "epoch": 1.2774239207360227,
+      "grad_norm": 0.23222966956027472,
+      "learning_rate": 3.189926547743967e-05,
+      "loss": 0.4388,
+      "step": 903
+    },
+    {
+      "epoch": 1.2788393489030432,
+      "grad_norm": 0.22261881527619265,
+      "learning_rate": 3.187303252885625e-05,
+      "loss": 0.4779,
+      "step": 904
+    },
+    {
+      "epoch": 1.2802547770700636,
+      "grad_norm": 0.23633660841854076,
+      "learning_rate": 3.184679958027282e-05,
+      "loss": 0.5065,
+      "step": 905
+    },
+    {
+      "epoch": 1.2816702052370843,
+      "grad_norm": 0.2563369762371776,
+      "learning_rate": 3.18205666316894e-05,
+      "loss": 0.4675,
+      "step": 906
+    },
+    {
+      "epoch": 1.2830856334041048,
+      "grad_norm": 0.24241008063485786,
+      "learning_rate": 3.179433368310598e-05,
+      "loss": 0.526,
+      "step": 907
+    },
+    {
+      "epoch": 1.2845010615711252,
+      "grad_norm": 0.2438974850078751,
+      "learning_rate": 3.176810073452256e-05,
+      "loss": 0.4372,
+      "step": 908
+    },
+    {
+      "epoch": 1.2859164897381459,
+      "grad_norm": 0.2484499708071917,
+      "learning_rate": 3.1741867785939136e-05,
+      "loss": 0.4471,
+      "step": 909
+    },
+    {
+      "epoch": 1.2873319179051663,
+      "grad_norm": 0.22020624012488788,
+      "learning_rate": 3.1715634837355716e-05,
+      "loss": 0.4817,
+      "step": 910
+    },
+    {
+      "epoch": 1.2887473460721868,
+      "grad_norm": 0.2079438907389061,
+      "learning_rate": 3.16894018887723e-05,
+      "loss": 0.4036,
+      "step": 911
+    },
+    {
+      "epoch": 1.2901627742392074,
+      "grad_norm": 0.23809027919871859,
+      "learning_rate": 3.166316894018888e-05,
+      "loss": 0.4757,
+      "step": 912
+    },
+    {
+      "epoch": 1.291578202406228,
+      "grad_norm": 0.2534537581446485,
+      "learning_rate": 3.163693599160546e-05,
+      "loss": 0.4632,
+      "step": 913
+    },
+    {
+      "epoch": 1.2929936305732483,
+      "grad_norm": 0.2508591501784022,
+      "learning_rate": 3.161070304302204e-05,
+      "loss": 0.5236,
+      "step": 914
+    },
+    {
+      "epoch": 1.294409058740269,
+      "grad_norm": 0.2692799394554751,
+      "learning_rate": 3.158447009443862e-05,
+      "loss": 0.5129,
+      "step": 915
+    },
+    {
+      "epoch": 1.2958244869072895,
+      "grad_norm": 0.26762393032776954,
+      "learning_rate": 3.15582371458552e-05,
+      "loss": 0.4533,
+      "step": 916
+    },
+    {
+      "epoch": 1.29723991507431,
+      "grad_norm": 0.22352913686126197,
+      "learning_rate": 3.153200419727178e-05,
+      "loss": 0.5367,
+      "step": 917
+    },
+    {
+      "epoch": 1.2986553432413306,
+      "grad_norm": 0.2702914668199128,
+      "learning_rate": 3.150577124868835e-05,
+      "loss": 0.4394,
+      "step": 918
+    },
+    {
+      "epoch": 1.300070771408351,
+      "grad_norm": 0.26063754811266,
+      "learning_rate": 3.147953830010493e-05,
+      "loss": 0.4127,
+      "step": 919
+    },
+    {
+      "epoch": 1.3014861995753715,
+      "grad_norm": 0.24585300821822167,
+      "learning_rate": 3.145330535152151e-05,
+      "loss": 0.4451,
+      "step": 920
+    },
+    {
+      "epoch": 1.3029016277423922,
+      "grad_norm": 0.268034212777459,
+      "learning_rate": 3.142707240293809e-05,
+      "loss": 0.467,
+      "step": 921
+    },
+    {
+      "epoch": 1.3043170559094126,
+      "grad_norm": 0.2409662370057274,
+      "learning_rate": 3.140083945435467e-05,
+      "loss": 0.4499,
+      "step": 922
+    },
+    {
+      "epoch": 1.305732484076433,
+      "grad_norm": 0.28243546373141387,
+      "learning_rate": 3.1374606505771246e-05,
+      "loss": 0.5331,
+      "step": 923
+    },
+    {
+      "epoch": 1.3071479122434537,
+      "grad_norm": 0.25855346685649155,
+      "learning_rate": 3.1348373557187826e-05,
+      "loss": 0.5168,
+      "step": 924
+    },
+    {
+      "epoch": 1.3085633404104742,
+      "grad_norm": 0.28182218038496243,
+      "learning_rate": 3.132214060860441e-05,
+      "loss": 0.4565,
+      "step": 925
+    },
+    {
+      "epoch": 1.3099787685774946,
+      "grad_norm": 0.2403385267177625,
+      "learning_rate": 3.129590766002099e-05,
+      "loss": 0.5043,
+      "step": 926
+    },
+    {
+      "epoch": 1.3113941967445153,
+      "grad_norm": 0.27408089069596053,
+      "learning_rate": 3.126967471143757e-05,
+      "loss": 0.5451,
+      "step": 927
+    },
+    {
+      "epoch": 1.3128096249115357,
+      "grad_norm": 0.2646909795264679,
+      "learning_rate": 3.124344176285415e-05,
+      "loss": 0.479,
+      "step": 928
+    },
+    {
+      "epoch": 1.3142250530785562,
+      "grad_norm": 0.2305069678068314,
+      "learning_rate": 3.121720881427073e-05,
+      "loss": 0.5296,
+      "step": 929
+    },
+    {
+      "epoch": 1.3156404812455769,
+      "grad_norm": 0.2744177117718372,
+      "learning_rate": 3.119097586568731e-05,
+      "loss": 0.4754,
+      "step": 930
+    },
+    {
+      "epoch": 1.3170559094125973,
+      "grad_norm": 0.6499533802326067,
+      "learning_rate": 3.116474291710388e-05,
+      "loss": 0.5306,
+      "step": 931
+    },
+    {
+      "epoch": 1.3184713375796178,
+      "grad_norm": 0.2303504782797021,
+      "learning_rate": 3.113850996852046e-05,
+      "loss": 0.4832,
+      "step": 932
+    },
+    {
+      "epoch": 1.3198867657466384,
+      "grad_norm": 0.2615164107491973,
+      "learning_rate": 3.111227701993704e-05,
+      "loss": 0.5244,
+      "step": 933
+    },
+    {
+      "epoch": 1.3213021939136589,
+      "grad_norm": 0.2959878077888004,
+      "learning_rate": 3.108604407135362e-05,
+      "loss": 0.4774,
+      "step": 934
+    },
+    {
+      "epoch": 1.3227176220806793,
+      "grad_norm": 0.23964789845458753,
+      "learning_rate": 3.10598111227702e-05,
+      "loss": 0.4766,
+      "step": 935
+    },
+    {
+      "epoch": 1.3241330502477,
+      "grad_norm": 0.21690181913877962,
+      "learning_rate": 3.1033578174186776e-05,
+      "loss": 0.419,
+      "step": 936
+    },
+    {
+      "epoch": 1.3255484784147205,
+      "grad_norm": 0.6426389076756095,
+      "learning_rate": 3.1007345225603356e-05,
+      "loss": 0.4507,
+      "step": 937
+    },
+    {
+      "epoch": 1.326963906581741,
+      "grad_norm": 0.263147646977239,
+      "learning_rate": 3.098111227701994e-05,
+      "loss": 0.4573,
+      "step": 938
+    },
+    {
+      "epoch": 1.3283793347487616,
+      "grad_norm": 0.24757143586337804,
+      "learning_rate": 3.095487932843652e-05,
+      "loss": 0.5759,
+      "step": 939
+    },
+    {
+      "epoch": 1.329794762915782,
+      "grad_norm": 0.25257042616743186,
+      "learning_rate": 3.09286463798531e-05,
+      "loss": 0.4862,
+      "step": 940
+    },
+    {
+      "epoch": 1.3312101910828025,
+      "grad_norm": 0.24251280132899075,
+      "learning_rate": 3.090241343126968e-05,
+      "loss": 0.4896,
+      "step": 941
+    },
+    {
+      "epoch": 1.3326256192498231,
+      "grad_norm": 0.25064062263912584,
+      "learning_rate": 3.087618048268626e-05,
+      "loss": 0.489,
+      "step": 942
+    },
+    {
+      "epoch": 1.3340410474168436,
+      "grad_norm": 0.24516768410214754,
+      "learning_rate": 3.084994753410284e-05,
+      "loss": 0.4174,
+      "step": 943
+    },
+    {
+      "epoch": 1.335456475583864,
+      "grad_norm": 0.25861482096183286,
+      "learning_rate": 3.082371458551941e-05,
+      "loss": 0.4574,
+      "step": 944
+    },
+    {
+      "epoch": 1.3368719037508847,
+      "grad_norm": 6.68702234250407,
+      "learning_rate": 3.079748163693599e-05,
+      "loss": 0.4732,
+      "step": 945
+    },
+    {
+      "epoch": 1.3382873319179052,
+      "grad_norm": 0.25068169445356625,
+      "learning_rate": 3.077124868835257e-05,
+      "loss": 0.4789,
+      "step": 946
+    },
+    {
+      "epoch": 1.3397027600849256,
+      "grad_norm": 0.25093075289474387,
+      "learning_rate": 3.074501573976915e-05,
+      "loss": 0.4568,
+      "step": 947
+    },
+    {
+      "epoch": 1.3411181882519463,
+      "grad_norm": 0.28246974745034786,
+      "learning_rate": 3.071878279118573e-05,
+      "loss": 0.4825,
+      "step": 948
+    },
+    {
+      "epoch": 1.3425336164189667,
+      "grad_norm": 0.21166319183056573,
+      "learning_rate": 3.0692549842602306e-05,
+      "loss": 0.439,
+      "step": 949
+    },
+    {
+      "epoch": 1.3439490445859872,
+      "grad_norm": 1.0385631184142639,
+      "learning_rate": 3.0666316894018887e-05,
+      "loss": 0.4454,
+      "step": 950
+    },
+    {
+      "epoch": 1.3453644727530079,
+      "grad_norm": 0.25777032972003616,
+      "learning_rate": 3.064008394543547e-05,
+      "loss": 0.5057,
+      "step": 951
+    },
+    {
+      "epoch": 1.3467799009200283,
+      "grad_norm": 0.26692860775734006,
+      "learning_rate": 3.061385099685205e-05,
+      "loss": 0.481,
+      "step": 952
+    },
+    {
+      "epoch": 1.3481953290870488,
+      "grad_norm": 0.24088223472029782,
+      "learning_rate": 3.058761804826863e-05,
+      "loss": 0.5184,
+      "step": 953
+    },
+    {
+      "epoch": 1.3496107572540694,
+      "grad_norm": 0.2247620489296305,
+      "learning_rate": 3.056138509968521e-05,
+      "loss": 0.5462,
+      "step": 954
+    },
+    {
+      "epoch": 1.3510261854210899,
+      "grad_norm": 0.23526403150601521,
+      "learning_rate": 3.053515215110179e-05,
+      "loss": 0.472,
+      "step": 955
+    },
+    {
+      "epoch": 1.3524416135881103,
+      "grad_norm": 0.2318641777138758,
+      "learning_rate": 3.0508919202518365e-05,
+      "loss": 0.446,
+      "step": 956
+    },
+    {
+      "epoch": 1.353857041755131,
+      "grad_norm": 0.24399478839109545,
+      "learning_rate": 3.0482686253934945e-05,
+      "loss": 0.45,
+      "step": 957
+    },
+    {
+      "epoch": 1.3552724699221514,
+      "grad_norm": 0.2467593178269623,
+      "learning_rate": 3.0456453305351522e-05,
+      "loss": 0.5054,
+      "step": 958
+    },
+    {
+      "epoch": 1.356687898089172,
+      "grad_norm": 0.2417281084040149,
+      "learning_rate": 3.0430220356768102e-05,
+      "loss": 0.4866,
+      "step": 959
+    },
+    {
+      "epoch": 1.3581033262561926,
+      "grad_norm": 0.24492624076347055,
+      "learning_rate": 3.040398740818468e-05,
+      "loss": 0.5484,
+      "step": 960
+    },
+    {
+      "epoch": 1.359518754423213,
+      "grad_norm": 0.2666390176436146,
+      "learning_rate": 3.037775445960126e-05,
+      "loss": 0.513,
+      "step": 961
+    },
+    {
+      "epoch": 1.3609341825902335,
+      "grad_norm": 0.2619780519186626,
+      "learning_rate": 3.035152151101784e-05,
+      "loss": 0.4835,
+      "step": 962
+    },
+    {
+      "epoch": 1.3623496107572541,
+      "grad_norm": 0.2753506150378985,
+      "learning_rate": 3.0325288562434417e-05,
+      "loss": 0.4276,
+      "step": 963
+    },
+    {
+      "epoch": 1.3637650389242746,
+      "grad_norm": 0.23838022166368394,
+      "learning_rate": 3.0299055613850997e-05,
+      "loss": 0.4849,
+      "step": 964
+    },
+    {
+      "epoch": 1.365180467091295,
+      "grad_norm": 0.24285680424504225,
+      "learning_rate": 3.0272822665267574e-05,
+      "loss": 0.4538,
+      "step": 965
+    },
+    {
+      "epoch": 1.3665958952583157,
+      "grad_norm": 4.7723592644763535,
+      "learning_rate": 3.024658971668416e-05,
+      "loss": 0.4491,
+      "step": 966
+    },
+    {
+      "epoch": 1.3680113234253362,
+      "grad_norm": 0.24476982720764057,
+      "learning_rate": 3.0220356768100738e-05,
+      "loss": 0.4339,
+      "step": 967
+    },
+    {
+      "epoch": 1.3694267515923566,
+      "grad_norm": 0.2959167527473618,
+      "learning_rate": 3.0194123819517318e-05,
+      "loss": 0.4118,
+      "step": 968
+    },
+    {
+      "epoch": 1.3708421797593773,
+      "grad_norm": 0.2408274889934171,
+      "learning_rate": 3.0167890870933895e-05,
+      "loss": 0.405,
+      "step": 969
+    },
+    {
+      "epoch": 1.3722576079263977,
+      "grad_norm": 0.23977225860357196,
+      "learning_rate": 3.0141657922350475e-05,
+      "loss": 0.4566,
+      "step": 970
+    },
+    {
+      "epoch": 1.3736730360934182,
+      "grad_norm": 0.258233874390108,
+      "learning_rate": 3.0115424973767052e-05,
+      "loss": 0.4709,
+      "step": 971
+    },
+    {
+      "epoch": 1.3750884642604388,
+      "grad_norm": 0.23298288179195542,
+      "learning_rate": 3.0089192025183633e-05,
+      "loss": 0.4883,
+      "step": 972
+    },
+    {
+      "epoch": 1.3765038924274593,
+      "grad_norm": 0.25732371520926234,
+      "learning_rate": 3.0062959076600213e-05,
+      "loss": 0.4587,
+      "step": 973
+    },
+    {
+      "epoch": 1.3779193205944797,
+      "grad_norm": 0.2770166948552172,
+      "learning_rate": 3.003672612801679e-05,
+      "loss": 0.4446,
+      "step": 974
+    },
+    {
+      "epoch": 1.3793347487615004,
+      "grad_norm": 0.25686220374398416,
+      "learning_rate": 3.001049317943337e-05,
+      "loss": 0.4659,
+      "step": 975
+    },
+    {
+      "epoch": 1.3807501769285209,
+      "grad_norm": 0.2748242222470758,
+      "learning_rate": 2.9984260230849947e-05,
+      "loss": 0.5211,
+      "step": 976
+    },
+    {
+      "epoch": 1.3821656050955413,
+      "grad_norm": 0.24194184131657517,
+      "learning_rate": 2.9958027282266527e-05,
+      "loss": 0.4752,
+      "step": 977
+    },
+    {
+      "epoch": 1.383581033262562,
+      "grad_norm": 0.2725631825825404,
+      "learning_rate": 2.9931794333683104e-05,
+      "loss": 0.4234,
+      "step": 978
+    },
+    {
+      "epoch": 1.3849964614295824,
+      "grad_norm": 0.23748217806062075,
+      "learning_rate": 2.9905561385099684e-05,
+      "loss": 0.464,
+      "step": 979
+    },
+    {
+      "epoch": 1.3864118895966029,
+      "grad_norm": 0.2309228037376953,
+      "learning_rate": 2.9879328436516268e-05,
+      "loss": 0.4827,
+      "step": 980
+    },
+    {
+      "epoch": 1.3878273177636236,
+      "grad_norm": 0.2891815389720354,
+      "learning_rate": 2.985309548793285e-05,
+      "loss": 0.4465,
+      "step": 981
+    },
+    {
+      "epoch": 1.389242745930644,
+      "grad_norm": 0.22958479608149038,
+      "learning_rate": 2.9826862539349425e-05,
+      "loss": 0.5058,
+      "step": 982
+    },
+    {
+      "epoch": 1.3906581740976645,
+      "grad_norm": 0.24514929972489777,
+      "learning_rate": 2.9800629590766006e-05,
+      "loss": 0.5549,
+      "step": 983
+    },
+    {
+      "epoch": 1.3920736022646851,
+      "grad_norm": 1.069141050088208,
+      "learning_rate": 2.9774396642182582e-05,
+      "loss": 0.5151,
+      "step": 984
+    },
+    {
+      "epoch": 1.3934890304317056,
+      "grad_norm": 0.24065132318214158,
+      "learning_rate": 2.9748163693599163e-05,
+      "loss": 0.5095,
+      "step": 985
+    },
+    {
+      "epoch": 1.394904458598726,
+      "grad_norm": 0.29294351959468184,
+      "learning_rate": 2.9721930745015743e-05,
+      "loss": 0.4606,
+      "step": 986
+    },
+    {
+      "epoch": 1.3963198867657467,
+      "grad_norm": 0.2700738054926797,
+      "learning_rate": 2.969569779643232e-05,
+      "loss": 0.4666,
+      "step": 987
+    },
+    {
+      "epoch": 1.3977353149327671,
+      "grad_norm": 0.24780694955297938,
+      "learning_rate": 2.96694648478489e-05,
+      "loss": 0.4336,
+      "step": 988
+    },
+    {
+      "epoch": 1.3991507430997876,
+      "grad_norm": 0.2825839661544209,
+      "learning_rate": 2.9643231899265477e-05,
+      "loss": 0.511,
+      "step": 989
+    },
+    {
+      "epoch": 1.4005661712668083,
+      "grad_norm": 0.31985767764921325,
+      "learning_rate": 2.9616998950682057e-05,
+      "loss": 0.4769,
+      "step": 990
+    },
+    {
+      "epoch": 1.4019815994338287,
+      "grad_norm": 0.21515488179858627,
+      "learning_rate": 2.9590766002098634e-05,
+      "loss": 0.4841,
+      "step": 991
+    },
+    {
+      "epoch": 1.4033970276008492,
+      "grad_norm": 0.2999937500375315,
+      "learning_rate": 2.9564533053515215e-05,
+      "loss": 0.5267,
+      "step": 992
+    },
+    {
+      "epoch": 1.4048124557678698,
+      "grad_norm": 0.27110098463243326,
+      "learning_rate": 2.9538300104931795e-05,
+      "loss": 0.4391,
+      "step": 993
+    },
+    {
+      "epoch": 1.4062278839348903,
+      "grad_norm": 0.2632120517796395,
+      "learning_rate": 2.951206715634838e-05,
+      "loss": 0.453,
+      "step": 994
+    },
+    {
+      "epoch": 1.4076433121019107,
+      "grad_norm": 0.2738208188659536,
+      "learning_rate": 2.9485834207764955e-05,
+      "loss": 0.4775,
+      "step": 995
+    },
+    {
+      "epoch": 1.4090587402689314,
+      "grad_norm": 0.258769719684058,
+      "learning_rate": 2.9459601259181536e-05,
+      "loss": 0.5675,
+      "step": 996
+    },
+    {
+      "epoch": 1.4104741684359519,
+      "grad_norm": 0.2513014251809082,
+      "learning_rate": 2.9433368310598113e-05,
+      "loss": 0.5778,
+      "step": 997
+    },
+    {
+      "epoch": 1.4118895966029723,
+      "grad_norm": 0.2615702887534967,
+      "learning_rate": 2.9407135362014693e-05,
+      "loss": 0.4508,
+      "step": 998
+    },
+    {
+      "epoch": 1.413305024769993,
+      "grad_norm": 0.2630569521513353,
+      "learning_rate": 2.9380902413431273e-05,
+      "loss": 0.4531,
+      "step": 999
+    },
+    {
+      "epoch": 1.4147204529370134,
+      "grad_norm": 0.24297232603525354,
+      "learning_rate": 2.935466946484785e-05,
+      "loss": 0.4656,
+      "step": 1000
+    },
+    {
+      "epoch": 1.4161358811040339,
+      "grad_norm": 0.23366469373399976,
+      "learning_rate": 2.932843651626443e-05,
+      "loss": 0.4308,
+      "step": 1001
+    },
+    {
+      "epoch": 1.4175513092710545,
+      "grad_norm": 0.23892561947726176,
+      "learning_rate": 2.9302203567681007e-05,
+      "loss": 0.4375,
+      "step": 1002
+    },
+    {
+      "epoch": 1.418966737438075,
+      "grad_norm": 0.23195340218843138,
+      "learning_rate": 2.9275970619097587e-05,
+      "loss": 0.444,
+      "step": 1003
+    },
+    {
+      "epoch": 1.4203821656050954,
+      "grad_norm": 0.22948701966121285,
+      "learning_rate": 2.9249737670514164e-05,
+      "loss": 0.4848,
+      "step": 1004
+    },
+    {
+      "epoch": 1.4217975937721161,
+      "grad_norm": 0.24785624514905413,
+      "learning_rate": 2.9223504721930745e-05,
+      "loss": 0.4634,
+      "step": 1005
+    },
+    {
+      "epoch": 1.4232130219391366,
+      "grad_norm": 0.2796816145500579,
+      "learning_rate": 2.9197271773347325e-05,
+      "loss": 0.5248,
+      "step": 1006
+    },
+    {
+      "epoch": 1.424628450106157,
+      "grad_norm": 0.21395483176765387,
+      "learning_rate": 2.9171038824763902e-05,
+      "loss": 0.4028,
+      "step": 1007
+    },
+    {
+      "epoch": 1.4260438782731777,
+      "grad_norm": 0.23564260855313562,
+      "learning_rate": 2.9144805876180486e-05,
+      "loss": 0.5343,
+      "step": 1008
+    },
+    {
+      "epoch": 1.4274593064401981,
+      "grad_norm": 0.26481931465222386,
+      "learning_rate": 2.9118572927597066e-05,
+      "loss": 0.4962,
+      "step": 1009
+    },
+    {
+      "epoch": 1.4288747346072186,
+      "grad_norm": 0.26619192924819485,
+      "learning_rate": 2.9092339979013643e-05,
+      "loss": 0.5139,
+      "step": 1010
+    },
+    {
+      "epoch": 1.4302901627742393,
+      "grad_norm": 0.25387021436623847,
+      "learning_rate": 2.9066107030430223e-05,
+      "loss": 0.5136,
+      "step": 1011
+    },
+    {
+      "epoch": 1.4317055909412597,
+      "grad_norm": 0.24429440511933043,
+      "learning_rate": 2.9039874081846803e-05,
+      "loss": 0.4366,
+      "step": 1012
+    },
+    {
+      "epoch": 1.4331210191082802,
+      "grad_norm": 0.22053476848614068,
+      "learning_rate": 2.901364113326338e-05,
+      "loss": 0.4985,
+      "step": 1013
+    },
+    {
+      "epoch": 1.4345364472753008,
+      "grad_norm": 0.44885317266562963,
+      "learning_rate": 2.898740818467996e-05,
+      "loss": 0.4722,
+      "step": 1014
+    },
+    {
+      "epoch": 1.4359518754423213,
+      "grad_norm": 0.2647577301470883,
+      "learning_rate": 2.8961175236096537e-05,
+      "loss": 0.5957,
+      "step": 1015
+    },
+    {
+      "epoch": 1.4373673036093417,
+      "grad_norm": 0.25300983944384264,
+      "learning_rate": 2.8934942287513118e-05,
+      "loss": 0.5029,
+      "step": 1016
+    },
+    {
+      "epoch": 1.4387827317763624,
+      "grad_norm": 0.3069886239763433,
+      "learning_rate": 2.8908709338929695e-05,
+      "loss": 0.4781,
+      "step": 1017
+    },
+    {
+      "epoch": 1.4401981599433828,
+      "grad_norm": 0.24029825627591023,
+      "learning_rate": 2.8882476390346275e-05,
+      "loss": 0.4628,
+      "step": 1018
+    },
+    {
+      "epoch": 1.4416135881104033,
+      "grad_norm": 0.2542260271787419,
+      "learning_rate": 2.8856243441762855e-05,
+      "loss": 0.5957,
+      "step": 1019
+    },
+    {
+      "epoch": 1.443029016277424,
+      "grad_norm": 0.2659087195888366,
+      "learning_rate": 2.8830010493179432e-05,
+      "loss": 0.4687,
+      "step": 1020
+    },
+    {
+      "epoch": 1.4444444444444444,
+      "grad_norm": 0.22703447924437756,
+      "learning_rate": 2.8803777544596016e-05,
+      "loss": 0.4701,
+      "step": 1021
+    },
+    {
+      "epoch": 1.4458598726114649,
+      "grad_norm": 0.23932999856039933,
+      "learning_rate": 2.8777544596012596e-05,
+      "loss": 0.4962,
+      "step": 1022
+    },
+    {
+      "epoch": 1.4472753007784855,
+      "grad_norm": 0.23705026394966283,
+      "learning_rate": 2.8751311647429173e-05,
+      "loss": 0.4994,
+      "step": 1023
+    },
+    {
+      "epoch": 1.448690728945506,
+      "grad_norm": 0.23924281655893787,
+      "learning_rate": 2.8725078698845753e-05,
+      "loss": 0.4897,
+      "step": 1024
+    },
+    {
+      "epoch": 1.4501061571125264,
+      "grad_norm": 0.24534844390507993,
+      "learning_rate": 2.8698845750262333e-05,
+      "loss": 0.4633,
+      "step": 1025
+    },
+    {
+      "epoch": 1.451521585279547,
+      "grad_norm": 0.24647975510130424,
+      "learning_rate": 2.867261280167891e-05,
+      "loss": 0.5761,
+      "step": 1026
+    },
+    {
+      "epoch": 1.4529370134465676,
+      "grad_norm": 0.25902949327635205,
+      "learning_rate": 2.864637985309549e-05,
+      "loss": 0.5124,
+      "step": 1027
+    },
+    {
+      "epoch": 1.454352441613588,
+      "grad_norm": 0.24227778031818434,
+      "learning_rate": 2.8620146904512068e-05,
+      "loss": 0.4764,
+      "step": 1028
+    },
+    {
+      "epoch": 1.4557678697806087,
+      "grad_norm": 0.23574028432437744,
+      "learning_rate": 2.8593913955928648e-05,
+      "loss": 0.4763,
+      "step": 1029
+    },
+    {
+      "epoch": 1.4571832979476291,
+      "grad_norm": 0.255082791321367,
+      "learning_rate": 2.8567681007345225e-05,
+      "loss": 0.4469,
+      "step": 1030
+    },
+    {
+      "epoch": 1.4585987261146496,
+      "grad_norm": 0.2337134120806889,
+      "learning_rate": 2.8541448058761805e-05,
+      "loss": 0.4831,
+      "step": 1031
+    },
+    {
+      "epoch": 1.4600141542816703,
+      "grad_norm": 0.2373703793885165,
+      "learning_rate": 2.8515215110178385e-05,
+      "loss": 0.5018,
+      "step": 1032
+    },
+    {
+      "epoch": 1.4614295824486907,
+      "grad_norm": 0.23276457945691267,
+      "learning_rate": 2.8488982161594962e-05,
+      "loss": 0.4405,
+      "step": 1033
+    },
+    {
+      "epoch": 1.4628450106157111,
+      "grad_norm": 0.25045501124926023,
+      "learning_rate": 2.8462749213011542e-05,
+      "loss": 0.4732,
+      "step": 1034
+    },
+    {
+      "epoch": 1.4642604387827318,
+      "grad_norm": 0.25464013273882563,
+      "learning_rate": 2.8436516264428126e-05,
+      "loss": 0.4849,
+      "step": 1035
+    },
+    {
+      "epoch": 1.4656758669497523,
+      "grad_norm": 0.22818989277388887,
+      "learning_rate": 2.8410283315844703e-05,
+      "loss": 0.4076,
+      "step": 1036
+    },
+    {
+      "epoch": 1.4670912951167727,
+      "grad_norm": 0.22555792664130817,
+      "learning_rate": 2.8384050367261283e-05,
+      "loss": 0.4359,
+      "step": 1037
+    },
+    {
+      "epoch": 1.4685067232837934,
+      "grad_norm": 0.21960333308218966,
+      "learning_rate": 2.8357817418677864e-05,
+      "loss": 0.47,
+      "step": 1038
+    },
+    {
+      "epoch": 1.4699221514508138,
+      "grad_norm": 0.22431300388479583,
+      "learning_rate": 2.833158447009444e-05,
+      "loss": 0.4458,
+      "step": 1039
+    },
+    {
+      "epoch": 1.4713375796178343,
+      "grad_norm": 0.22822883270470412,
+      "learning_rate": 2.830535152151102e-05,
+      "loss": 0.5535,
+      "step": 1040
+    },
+    {
+      "epoch": 1.472753007784855,
+      "grad_norm": 0.25040042431079407,
+      "learning_rate": 2.8279118572927598e-05,
+      "loss": 0.4972,
+      "step": 1041
+    },
+    {
+      "epoch": 1.4741684359518754,
+      "grad_norm": 0.24527285015339936,
+      "learning_rate": 2.8252885624344178e-05,
+      "loss": 0.4816,
+      "step": 1042
+    },
+    {
+      "epoch": 1.4755838641188959,
+      "grad_norm": 0.21545250579454714,
+      "learning_rate": 2.8226652675760755e-05,
+      "loss": 0.4822,
+      "step": 1043
+    },
+    {
+      "epoch": 1.4769992922859165,
+      "grad_norm": 0.21160614593382832,
+      "learning_rate": 2.8200419727177335e-05,
+      "loss": 0.4696,
+      "step": 1044
+    },
+    {
+      "epoch": 1.478414720452937,
+      "grad_norm": 6.954437350346113,
+      "learning_rate": 2.8174186778593915e-05,
+      "loss": 0.4696,
+      "step": 1045
+    },
+    {
+      "epoch": 1.4798301486199574,
+      "grad_norm": 0.2446422497343303,
+      "learning_rate": 2.8147953830010492e-05,
+      "loss": 0.4642,
+      "step": 1046
+    },
+    {
+      "epoch": 1.481245576786978,
+      "grad_norm": 0.23710815310012034,
+      "learning_rate": 2.8121720881427073e-05,
+      "loss": 0.4564,
+      "step": 1047
+    },
+    {
+      "epoch": 1.4826610049539986,
+      "grad_norm": 0.21642519714321273,
+      "learning_rate": 2.809548793284365e-05,
+      "loss": 0.467,
+      "step": 1048
+    },
+    {
+      "epoch": 1.484076433121019,
+      "grad_norm": 0.24533444903258622,
+      "learning_rate": 2.8069254984260233e-05,
+      "loss": 0.4857,
+      "step": 1049
+    },
+    {
+      "epoch": 1.4854918612880397,
+      "grad_norm": 0.24166298624425117,
+      "learning_rate": 2.8043022035676813e-05,
+      "loss": 0.4672,
+      "step": 1050
+    },
+    {
+      "epoch": 1.4869072894550601,
+      "grad_norm": 0.21749550059484396,
+      "learning_rate": 2.8016789087093394e-05,
+      "loss": 0.4719,
+      "step": 1051
+    },
+    {
+      "epoch": 1.4883227176220806,
+      "grad_norm": 0.26678528308771254,
+      "learning_rate": 2.799055613850997e-05,
+      "loss": 0.4649,
+      "step": 1052
+    },
+    {
+      "epoch": 1.4897381457891012,
+      "grad_norm": 0.21661033809145605,
+      "learning_rate": 2.796432318992655e-05,
+      "loss": 0.4224,
+      "step": 1053
+    },
+    {
+      "epoch": 1.4911535739561217,
+      "grad_norm": 0.2313532912014695,
+      "learning_rate": 2.7938090241343128e-05,
+      "loss": 0.5097,
+      "step": 1054
+    },
+    {
+      "epoch": 1.4925690021231421,
+      "grad_norm": 0.26205168168462356,
+      "learning_rate": 2.7911857292759708e-05,
+      "loss": 0.5469,
+      "step": 1055
+    },
+    {
+      "epoch": 1.4939844302901628,
+      "grad_norm": 0.4863580049870588,
+      "learning_rate": 2.7885624344176285e-05,
+      "loss": 0.429,
+      "step": 1056
+    },
+    {
+      "epoch": 1.4953998584571833,
+      "grad_norm": 0.23631813461640477,
+      "learning_rate": 2.7859391395592865e-05,
+      "loss": 0.4717,
+      "step": 1057
+    },
+    {
+      "epoch": 1.4968152866242037,
+      "grad_norm": 0.2629048815627727,
+      "learning_rate": 2.7833158447009446e-05,
+      "loss": 0.4474,
+      "step": 1058
+    },
+    {
+      "epoch": 1.4982307147912244,
+      "grad_norm": 0.23677824354152793,
+      "learning_rate": 2.7806925498426022e-05,
+      "loss": 0.4815,
+      "step": 1059
+    },
+    {
+      "epoch": 1.4996461429582448,
+      "grad_norm": 0.24762130191336798,
+      "learning_rate": 2.7780692549842603e-05,
+      "loss": 0.44,
+      "step": 1060
+    },
+    {
+      "epoch": 1.5010615711252653,
+      "grad_norm": 0.24988799078497606,
+      "learning_rate": 2.775445960125918e-05,
+      "loss": 0.3898,
+      "step": 1061
+    },
+    {
+      "epoch": 1.502476999292286,
+      "grad_norm": 0.2368243984498134,
+      "learning_rate": 2.772822665267576e-05,
+      "loss": 0.448,
+      "step": 1062
+    },
+    {
+      "epoch": 1.5038924274593064,
+      "grad_norm": 0.2725939583573723,
+      "learning_rate": 2.7701993704092344e-05,
+      "loss": 0.4815,
+      "step": 1063
+    },
+    {
+      "epoch": 1.5053078556263269,
+      "grad_norm": 0.22491600948340684,
+      "learning_rate": 2.7675760755508924e-05,
+      "loss": 0.4645,
+      "step": 1064
+    },
+    {
+      "epoch": 1.5067232837933475,
+      "grad_norm": 0.24749866709382495,
+      "learning_rate": 2.76495278069255e-05,
+      "loss": 0.4113,
+      "step": 1065
+    },
+    {
+      "epoch": 1.508138711960368,
+      "grad_norm": 0.2665066221199876,
+      "learning_rate": 2.762329485834208e-05,
+      "loss": 0.5052,
+      "step": 1066
+    },
+    {
+      "epoch": 1.5095541401273884,
+      "grad_norm": 0.24199054657747365,
+      "learning_rate": 2.7597061909758658e-05,
+      "loss": 0.4706,
+      "step": 1067
+    },
+    {
+      "epoch": 1.510969568294409,
+      "grad_norm": 0.21771253763658477,
+      "learning_rate": 2.7570828961175238e-05,
+      "loss": 0.4613,
+      "step": 1068
+    },
+    {
+      "epoch": 1.5123849964614295,
+      "grad_norm": 0.2237890153752989,
+      "learning_rate": 2.7544596012591815e-05,
+      "loss": 0.436,
+      "step": 1069
+    },
+    {
+      "epoch": 1.51380042462845,
+      "grad_norm": 0.22977916156772293,
+      "learning_rate": 2.7518363064008395e-05,
+      "loss": 0.4764,
+      "step": 1070
+    },
+    {
+      "epoch": 1.5152158527954707,
+      "grad_norm": 0.2436197119328285,
+      "learning_rate": 2.7492130115424976e-05,
+      "loss": 0.4191,
+      "step": 1071
+    },
+    {
+      "epoch": 1.5166312809624911,
+      "grad_norm": 0.2814116794689585,
+      "learning_rate": 2.7465897166841553e-05,
+      "loss": 0.4331,
+      "step": 1072
+    },
+    {
+      "epoch": 1.5180467091295116,
+      "grad_norm": 0.25998996932832125,
+      "learning_rate": 2.7439664218258133e-05,
+      "loss": 0.4788,
+      "step": 1073
+    },
+    {
+      "epoch": 1.5194621372965322,
+      "grad_norm": 0.23739549641148952,
+      "learning_rate": 2.741343126967471e-05,
+      "loss": 0.4872,
+      "step": 1074
+    },
+    {
+      "epoch": 1.5208775654635527,
+      "grad_norm": 0.28943485657274004,
+      "learning_rate": 2.738719832109129e-05,
+      "loss": 0.4178,
+      "step": 1075
+    },
+    {
+      "epoch": 1.5222929936305731,
+      "grad_norm": 0.20996237549839084,
+      "learning_rate": 2.7360965372507867e-05,
+      "loss": 0.4111,
+      "step": 1076
+    },
+    {
+      "epoch": 1.5237084217975938,
+      "grad_norm": 0.23327937342431862,
+      "learning_rate": 2.7334732423924454e-05,
+      "loss": 0.5006,
+      "step": 1077
+    },
+    {
+      "epoch": 1.5251238499646143,
+      "grad_norm": 0.26823642539957526,
+      "learning_rate": 2.730849947534103e-05,
+      "loss": 0.4652,
+      "step": 1078
+    },
+    {
+      "epoch": 1.5265392781316347,
+      "grad_norm": 0.21072129577399698,
+      "learning_rate": 2.728226652675761e-05,
+      "loss": 0.4129,
+      "step": 1079
+    },
+    {
+      "epoch": 1.5279547062986554,
+      "grad_norm": 0.22274933741818195,
+      "learning_rate": 2.7256033578174188e-05,
+      "loss": 0.4531,
+      "step": 1080
+    },
+    {
+      "epoch": 1.5293701344656758,
+      "grad_norm": 0.4986936028511287,
+      "learning_rate": 2.722980062959077e-05,
+      "loss": 0.4597,
+      "step": 1081
+    },
+    {
+      "epoch": 1.5307855626326963,
+      "grad_norm": 0.25019227363506485,
+      "learning_rate": 2.7203567681007345e-05,
+      "loss": 0.4921,
+      "step": 1082
+    },
+    {
+      "epoch": 1.532200990799717,
+      "grad_norm": 0.21779980593438472,
+      "learning_rate": 2.7177334732423926e-05,
+      "loss": 0.4393,
+      "step": 1083
+    },
+    {
+      "epoch": 1.5336164189667374,
+      "grad_norm": 0.24912965629233305,
+      "learning_rate": 2.7151101783840506e-05,
+      "loss": 0.433,
+      "step": 1084
+    },
+    {
+      "epoch": 1.5350318471337578,
+      "grad_norm": 0.41623656096204564,
+      "learning_rate": 2.7124868835257083e-05,
+      "loss": 0.516,
+      "step": 1085
+    },
+    {
+      "epoch": 1.5364472753007785,
+      "grad_norm": 0.2687253995199058,
+      "learning_rate": 2.7098635886673663e-05,
+      "loss": 0.5026,
+      "step": 1086
+    },
+    {
+      "epoch": 1.537862703467799,
+      "grad_norm": 0.22337167247678047,
+      "learning_rate": 2.707240293809024e-05,
+      "loss": 0.4452,
+      "step": 1087
+    },
+    {
+      "epoch": 1.5392781316348194,
+      "grad_norm": 0.24670114978050278,
+      "learning_rate": 2.704616998950682e-05,
+      "loss": 0.544,
+      "step": 1088
+    },
+    {
+      "epoch": 1.54069355980184,
+      "grad_norm": 0.23666084857328787,
+      "learning_rate": 2.7019937040923397e-05,
+      "loss": 0.4499,
+      "step": 1089
+    },
+    {
+      "epoch": 1.5421089879688605,
+      "grad_norm": 0.21971236961761928,
+      "learning_rate": 2.6993704092339977e-05,
+      "loss": 0.4844,
+      "step": 1090
+    },
+    {
+      "epoch": 1.543524416135881,
+      "grad_norm": 0.23803059606252222,
+      "learning_rate": 2.696747114375656e-05,
+      "loss": 0.4869,
+      "step": 1091
+    },
+    {
+      "epoch": 1.5449398443029017,
+      "grad_norm": 0.24184454986545373,
+      "learning_rate": 2.694123819517314e-05,
+      "loss": 0.4538,
+      "step": 1092
+    },
+    {
+      "epoch": 1.546355272469922,
+      "grad_norm": 0.2432655023422465,
+      "learning_rate": 2.6915005246589718e-05,
+      "loss": 0.5182,
+      "step": 1093
+    },
+    {
+      "epoch": 1.5477707006369426,
+      "grad_norm": 0.21884280908716727,
+      "learning_rate": 2.68887722980063e-05,
+      "loss": 0.4508,
+      "step": 1094
+    },
+    {
+      "epoch": 1.5491861288039632,
+      "grad_norm": 0.23837889034609053,
+      "learning_rate": 2.686253934942288e-05,
+      "loss": 0.4026,
+      "step": 1095
+    },
+    {
+      "epoch": 1.5506015569709837,
+      "grad_norm": 2.186577418971484,
+      "learning_rate": 2.6836306400839456e-05,
+      "loss": 0.4374,
+      "step": 1096
+    },
+    {
+      "epoch": 1.5520169851380041,
+      "grad_norm": 0.2301787401222238,
+      "learning_rate": 2.6810073452256036e-05,
+      "loss": 0.4811,
+      "step": 1097
+    },
+    {
+      "epoch": 1.5534324133050248,
+      "grad_norm": 0.26506009481521925,
+      "learning_rate": 2.6783840503672613e-05,
+      "loss": 0.5027,
+      "step": 1098
+    },
+    {
+      "epoch": 1.5548478414720452,
+      "grad_norm": 0.22334070900295735,
+      "learning_rate": 2.6757607555089193e-05,
+      "loss": 0.4562,
+      "step": 1099
+    },
+    {
+      "epoch": 1.5562632696390657,
+      "grad_norm": 0.23509770488005835,
+      "learning_rate": 2.673137460650577e-05,
+      "loss": 0.5018,
+      "step": 1100
+    },
+    {
+      "epoch": 1.5576786978060864,
+      "grad_norm": 0.25822797097500144,
+      "learning_rate": 2.670514165792235e-05,
+      "loss": 0.4649,
+      "step": 1101
+    },
+    {
+      "epoch": 1.5590941259731068,
+      "grad_norm": 0.22020847666976612,
+      "learning_rate": 2.6678908709338927e-05,
+      "loss": 0.5247,
+      "step": 1102
+    },
+    {
+      "epoch": 1.5605095541401273,
+      "grad_norm": 0.2352135371428812,
+      "learning_rate": 2.6652675760755508e-05,
+      "loss": 0.5146,
+      "step": 1103
+    },
+    {
+      "epoch": 1.561924982307148,
+      "grad_norm": 0.2489630883187179,
+      "learning_rate": 2.662644281217209e-05,
+      "loss": 0.481,
+      "step": 1104
+    },
+    {
+      "epoch": 1.5633404104741684,
+      "grad_norm": 0.25105003951133575,
+      "learning_rate": 2.660020986358867e-05,
+      "loss": 0.4849,
+      "step": 1105
+    },
+    {
+      "epoch": 1.5647558386411888,
+      "grad_norm": 0.22146346229154182,
+      "learning_rate": 2.657397691500525e-05,
+      "loss": 0.514,
+      "step": 1106
+    },
+    {
+      "epoch": 1.5661712668082095,
+      "grad_norm": 0.24361463581563225,
+      "learning_rate": 2.654774396642183e-05,
+      "loss": 0.5323,
+      "step": 1107
+    },
+    {
+      "epoch": 1.56758669497523,
+      "grad_norm": 0.269911412108188,
+      "learning_rate": 2.652151101783841e-05,
+      "loss": 0.4628,
+      "step": 1108
+    },
+    {
+      "epoch": 1.5690021231422504,
+      "grad_norm": 0.2537121060671178,
+      "learning_rate": 2.6495278069254986e-05,
+      "loss": 0.4818,
+      "step": 1109
+    },
+    {
+      "epoch": 1.570417551309271,
+      "grad_norm": 0.25966850377325174,
+      "learning_rate": 2.6469045120671566e-05,
+      "loss": 0.5097,
+      "step": 1110
+    },
+    {
+      "epoch": 1.5718329794762915,
+      "grad_norm": 0.265144191902071,
+      "learning_rate": 2.6442812172088143e-05,
+      "loss": 0.4998,
+      "step": 1111
+    },
+    {
+      "epoch": 1.573248407643312,
+      "grad_norm": 0.26052675948490334,
+      "learning_rate": 2.6416579223504723e-05,
+      "loss": 0.469,
+      "step": 1112
+    },
+    {
+      "epoch": 1.5746638358103326,
+      "grad_norm": 1.5639646666796965,
+      "learning_rate": 2.63903462749213e-05,
+      "loss": 0.4874,
+      "step": 1113
+    },
+    {
+      "epoch": 1.576079263977353,
+      "grad_norm": 0.34737557088415505,
+      "learning_rate": 2.636411332633788e-05,
+      "loss": 0.4283,
+      "step": 1114
+    },
+    {
+      "epoch": 1.5774946921443735,
+      "grad_norm": 0.25563683675844123,
+      "learning_rate": 2.6337880377754457e-05,
+      "loss": 0.4203,
+      "step": 1115
+    },
+    {
+      "epoch": 1.5789101203113942,
+      "grad_norm": 0.28441359753020323,
+      "learning_rate": 2.6311647429171038e-05,
+      "loss": 0.468,
+      "step": 1116
+    },
+    {
+      "epoch": 1.5803255484784147,
+      "grad_norm": 0.372909035094879,
+      "learning_rate": 2.6285414480587618e-05,
+      "loss": 0.5886,
+      "step": 1117
+    },
+    {
+      "epoch": 1.5817409766454351,
+      "grad_norm": 1.5462598794149647,
+      "learning_rate": 2.6259181532004202e-05,
+      "loss": 0.5186,
+      "step": 1118
+    },
+    {
+      "epoch": 1.5831564048124558,
+      "grad_norm": 0.2600177084809281,
+      "learning_rate": 2.623294858342078e-05,
+      "loss": 0.3898,
+      "step": 1119
+    },
+    {
+      "epoch": 1.5845718329794765,
+      "grad_norm": 0.29616234684449666,
+      "learning_rate": 2.620671563483736e-05,
+      "loss": 0.4848,
+      "step": 1120
+    },
+    {
+      "epoch": 1.5859872611464967,
+      "grad_norm": 0.25246015830438095,
+      "learning_rate": 2.618048268625394e-05,
+      "loss": 0.5138,
+      "step": 1121
+    },
+    {
+      "epoch": 1.5874026893135174,
+      "grad_norm": 0.2592928418343624,
+      "learning_rate": 2.6154249737670516e-05,
+      "loss": 0.5177,
+      "step": 1122
+    },
+    {
+      "epoch": 1.588818117480538,
+      "grad_norm": 1.343675437936197,
+      "learning_rate": 2.6128016789087096e-05,
+      "loss": 0.4652,
+      "step": 1123
+    },
+    {
+      "epoch": 1.5902335456475583,
+      "grad_norm": 0.3378412630785343,
+      "learning_rate": 2.6101783840503673e-05,
+      "loss": 0.4741,
+      "step": 1124
+    },
+    {
+      "epoch": 1.591648973814579,
+      "grad_norm": 0.23591041247632422,
+      "learning_rate": 2.6075550891920254e-05,
+      "loss": 0.4027,
+      "step": 1125
+    },
+    {
+      "epoch": 1.5930644019815996,
+      "grad_norm": 0.27025424322925645,
+      "learning_rate": 2.604931794333683e-05,
+      "loss": 0.5317,
+      "step": 1126
+    },
+    {
+      "epoch": 1.5944798301486198,
+      "grad_norm": 0.3032366021165955,
+      "learning_rate": 2.602308499475341e-05,
+      "loss": 0.4353,
+      "step": 1127
+    },
+    {
+      "epoch": 1.5958952583156405,
+      "grad_norm": 0.2380386156268189,
+      "learning_rate": 2.599685204616999e-05,
+      "loss": 0.4725,
+      "step": 1128
+    },
+    {
+      "epoch": 1.5973106864826612,
+      "grad_norm": 0.2861736326134302,
+      "learning_rate": 2.5970619097586568e-05,
+      "loss": 0.4417,
+      "step": 1129
+    },
+    {
+      "epoch": 1.5987261146496814,
+      "grad_norm": 0.26368405652585286,
+      "learning_rate": 2.5944386149003148e-05,
+      "loss": 0.4391,
+      "step": 1130
+    },
+    {
+      "epoch": 1.600141542816702,
+      "grad_norm": 0.2276153908320552,
+      "learning_rate": 2.5918153200419725e-05,
+      "loss": 0.4493,
+      "step": 1131
+    },
+    {
+      "epoch": 1.6015569709837227,
+      "grad_norm": 0.22999463566162967,
+      "learning_rate": 2.589192025183631e-05,
+      "loss": 0.5438,
+      "step": 1132
+    },
+    {
+      "epoch": 1.602972399150743,
+      "grad_norm": 0.2809859072713602,
+      "learning_rate": 2.586568730325289e-05,
+      "loss": 0.5141,
+      "step": 1133
+    },
+    {
+      "epoch": 1.6043878273177636,
+      "grad_norm": 0.2396152170770102,
+      "learning_rate": 2.583945435466947e-05,
+      "loss": 0.5947,
+      "step": 1134
+    },
+    {
+      "epoch": 1.6058032554847843,
+      "grad_norm": 0.2931573292499133,
+      "learning_rate": 2.5813221406086046e-05,
+      "loss": 0.4498,
+      "step": 1135
+    },
+    {
+      "epoch": 1.6072186836518045,
+      "grad_norm": 0.2492431057897274,
+      "learning_rate": 2.5786988457502627e-05,
+      "loss": 0.4594,
+      "step": 1136
+    },
+    {
+      "epoch": 1.6086341118188252,
+      "grad_norm": 0.22849622509328704,
+      "learning_rate": 2.5760755508919203e-05,
+      "loss": 0.4161,
+      "step": 1137
+    },
+    {
+      "epoch": 1.6100495399858459,
+      "grad_norm": 0.2276611441960024,
+      "learning_rate": 2.5734522560335784e-05,
+      "loss": 0.4717,
+      "step": 1138
+    },
+    {
+      "epoch": 1.611464968152866,
+      "grad_norm": 0.24910642514475684,
+      "learning_rate": 2.570828961175236e-05,
+      "loss": 0.5203,
+      "step": 1139
+    },
+    {
+      "epoch": 1.6128803963198868,
+      "grad_norm": 0.2607001585160038,
+      "learning_rate": 2.568205666316894e-05,
+      "loss": 0.5007,
+      "step": 1140
+    },
+    {
+      "epoch": 1.6142958244869074,
+      "grad_norm": 0.22409171708653328,
+      "learning_rate": 2.565582371458552e-05,
+      "loss": 0.4343,
+      "step": 1141
+    },
+    {
+      "epoch": 1.6157112526539277,
+      "grad_norm": 0.22361825047458853,
+      "learning_rate": 2.5629590766002098e-05,
+      "loss": 0.518,
+      "step": 1142
+    },
+    {
+      "epoch": 1.6171266808209483,
+      "grad_norm": 0.23193326257443647,
+      "learning_rate": 2.560335781741868e-05,
+      "loss": 0.4865,
+      "step": 1143
+    },
+    {
+      "epoch": 1.618542108987969,
+      "grad_norm": 0.23639779385117354,
+      "learning_rate": 2.5577124868835255e-05,
+      "loss": 0.4892,
+      "step": 1144
+    },
+    {
+      "epoch": 1.6199575371549892,
+      "grad_norm": 0.2439972587338885,
+      "learning_rate": 2.5550891920251836e-05,
+      "loss": 0.5131,
+      "step": 1145
+    },
+    {
+      "epoch": 1.62137296532201,
+      "grad_norm": 0.23805120948621156,
+      "learning_rate": 2.552465897166842e-05,
+      "loss": 0.5121,
+      "step": 1146
+    },
+    {
+      "epoch": 1.6227883934890306,
+      "grad_norm": 0.21467323810324876,
+      "learning_rate": 2.5498426023085e-05,
+      "loss": 0.4598,
+      "step": 1147
+    },
+    {
+      "epoch": 1.6242038216560508,
+      "grad_norm": 0.22227022281283504,
+      "learning_rate": 2.5472193074501576e-05,
+      "loss": 0.4353,
+      "step": 1148
+    },
+    {
+      "epoch": 1.6256192498230715,
+      "grad_norm": 0.22915765706021562,
+      "learning_rate": 2.5445960125918157e-05,
+      "loss": 0.5105,
+      "step": 1149
+    },
+    {
+      "epoch": 1.6270346779900922,
+      "grad_norm": 0.25397114301762364,
+      "learning_rate": 2.5419727177334734e-05,
+      "loss": 0.4626,
+      "step": 1150
+    },
+    {
+      "epoch": 1.6284501061571124,
+      "grad_norm": 0.22184273484318087,
+      "learning_rate": 2.5393494228751314e-05,
+      "loss": 0.4462,
+      "step": 1151
+    },
+    {
+      "epoch": 1.629865534324133,
+      "grad_norm": 0.2089350680438182,
+      "learning_rate": 2.536726128016789e-05,
+      "loss": 0.3759,
+      "step": 1152
+    },
+    {
+      "epoch": 1.6312809624911537,
+      "grad_norm": 0.24870576324892082,
+      "learning_rate": 2.534102833158447e-05,
+      "loss": 0.4802,
+      "step": 1153
+    },
+    {
+      "epoch": 1.632696390658174,
+      "grad_norm": 0.21126461355901588,
+      "learning_rate": 2.531479538300105e-05,
+      "loss": 0.4249,
+      "step": 1154
+    },
+    {
+      "epoch": 1.6341118188251946,
+      "grad_norm": 0.2486402703677725,
+      "learning_rate": 2.5288562434417628e-05,
+      "loss": 0.4362,
+      "step": 1155
+    },
+    {
+      "epoch": 1.6355272469922153,
+      "grad_norm": 0.2152020707436181,
+      "learning_rate": 2.526232948583421e-05,
+      "loss": 0.4393,
+      "step": 1156
+    },
+    {
+      "epoch": 1.6369426751592355,
+      "grad_norm": 0.22050316835622147,
+      "learning_rate": 2.5236096537250785e-05,
+      "loss": 0.4635,
+      "step": 1157
+    },
+    {
+      "epoch": 1.6383581033262562,
+      "grad_norm": 0.23009695935320004,
+      "learning_rate": 2.5209863588667366e-05,
+      "loss": 0.4453,
+      "step": 1158
+    },
+    {
+      "epoch": 1.6397735314932769,
+      "grad_norm": 0.25731684547250727,
+      "learning_rate": 2.5183630640083943e-05,
+      "loss": 0.4021,
+      "step": 1159
+    },
+    {
+      "epoch": 1.641188959660297,
+      "grad_norm": 0.2206489665591597,
+      "learning_rate": 2.515739769150053e-05,
+      "loss": 0.5104,
+      "step": 1160
+    },
+    {
+      "epoch": 1.6426043878273178,
+      "grad_norm": 0.252288856938944,
+      "learning_rate": 2.5131164742917107e-05,
+      "loss": 0.4646,
+      "step": 1161
+    },
+    {
+      "epoch": 1.6440198159943384,
+      "grad_norm": 0.25452492357715273,
+      "learning_rate": 2.5104931794333687e-05,
+      "loss": 0.449,
+      "step": 1162
+    },
+    {
+      "epoch": 1.6454352441613587,
+      "grad_norm": 0.24120366995017042,
+      "learning_rate": 2.5078698845750264e-05,
+      "loss": 0.4649,
+      "step": 1163
+    },
+    {
+      "epoch": 1.6468506723283793,
+      "grad_norm": 0.2279786516010407,
+      "learning_rate": 2.5052465897166844e-05,
+      "loss": 0.4609,
+      "step": 1164
+    },
+    {
+      "epoch": 1.6482661004954,
+      "grad_norm": 0.19460609034904436,
+      "learning_rate": 2.502623294858342e-05,
+      "loss": 0.4021,
+      "step": 1165
+    },
+    {
+      "epoch": 1.6496815286624202,
+      "grad_norm": 0.23745682422534709,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4729,
+      "step": 1166
+    },
+    {
+      "epoch": 1.651096956829441,
+      "grad_norm": 0.22632870172314187,
+      "learning_rate": 2.497376705141658e-05,
+      "loss": 0.4038,
+      "step": 1167
+    },
+    {
+      "epoch": 1.6525123849964616,
+      "grad_norm": 0.21060671269418166,
+      "learning_rate": 2.494753410283316e-05,
+      "loss": 0.4152,
+      "step": 1168
+    },
+    {
+      "epoch": 1.6539278131634818,
+      "grad_norm": 0.2144278010114158,
+      "learning_rate": 2.492130115424974e-05,
+      "loss": 0.491,
+      "step": 1169
+    },
+    {
+      "epoch": 1.6553432413305025,
+      "grad_norm": 0.23930362218821835,
+      "learning_rate": 2.489506820566632e-05,
+      "loss": 0.5087,
+      "step": 1170
+    },
+    {
+      "epoch": 1.6567586694975232,
+      "grad_norm": 0.23480237512635063,
+      "learning_rate": 2.48688352570829e-05,
+      "loss": 0.5187,
+      "step": 1171
+    },
+    {
+      "epoch": 1.6581740976645434,
+      "grad_norm": 0.24535990295367607,
+      "learning_rate": 2.4842602308499476e-05,
+      "loss": 0.4854,
+      "step": 1172
+    },
+    {
+      "epoch": 1.659589525831564,
+      "grad_norm": 0.21706667220912929,
+      "learning_rate": 2.4816369359916056e-05,
+      "loss": 0.4305,
+      "step": 1173
+    },
+    {
+      "epoch": 1.6610049539985847,
+      "grad_norm": 0.21570842708240046,
+      "learning_rate": 2.4790136411332633e-05,
+      "loss": 0.4588,
+      "step": 1174
+    },
+    {
+      "epoch": 1.662420382165605,
+      "grad_norm": 0.24620836293827908,
+      "learning_rate": 2.4763903462749214e-05,
+      "loss": 0.5459,
+      "step": 1175
+    },
+    {
+      "epoch": 1.6638358103326256,
+      "grad_norm": 0.23227318833395602,
+      "learning_rate": 2.473767051416579e-05,
+      "loss": 0.4533,
+      "step": 1176
+    },
+    {
+      "epoch": 1.6652512384996463,
+      "grad_norm": 0.2420577256811438,
+      "learning_rate": 2.4711437565582374e-05,
+      "loss": 0.5202,
+      "step": 1177
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.24918517663693857,
+      "learning_rate": 2.468520461699895e-05,
+      "loss": 0.4424,
+      "step": 1178
+    },
+    {
+      "epoch": 1.6680820948336872,
+      "grad_norm": 0.22057124481546828,
+      "learning_rate": 2.465897166841553e-05,
+      "loss": 0.4245,
+      "step": 1179
+    },
+    {
+      "epoch": 1.6694975230007079,
+      "grad_norm": 0.2399440923022933,
+      "learning_rate": 2.463273871983211e-05,
+      "loss": 0.4362,
+      "step": 1180
+    },
+    {
+      "epoch": 1.670912951167728,
+      "grad_norm": 4.334965775067043,
+      "learning_rate": 2.460650577124869e-05,
+      "loss": 0.4193,
+      "step": 1181
+    },
+    {
+      "epoch": 1.6723283793347488,
+      "grad_norm": 0.2672420532047751,
+      "learning_rate": 2.458027282266527e-05,
+      "loss": 0.4602,
+      "step": 1182
+    },
+    {
+      "epoch": 1.6737438075017694,
+      "grad_norm": 0.45886783224845246,
+      "learning_rate": 2.4554039874081846e-05,
+      "loss": 0.4929,
+      "step": 1183
+    },
+    {
+      "epoch": 1.6751592356687897,
+      "grad_norm": 0.23253089832864093,
+      "learning_rate": 2.452780692549843e-05,
+      "loss": 0.5029,
+      "step": 1184
+    },
+    {
+      "epoch": 1.6765746638358103,
+      "grad_norm": 0.2754881453715846,
+      "learning_rate": 2.4501573976915006e-05,
+      "loss": 0.4858,
+      "step": 1185
+    },
+    {
+      "epoch": 1.677990092002831,
+      "grad_norm": 0.24718432903788826,
+      "learning_rate": 2.4475341028331587e-05,
+      "loss": 0.4406,
+      "step": 1186
+    },
+    {
+      "epoch": 1.6794055201698512,
+      "grad_norm": 0.24486533742290445,
+      "learning_rate": 2.4449108079748163e-05,
+      "loss": 0.4639,
+      "step": 1187
+    },
+    {
+      "epoch": 1.680820948336872,
+      "grad_norm": 0.27303563278355925,
+      "learning_rate": 2.4422875131164744e-05,
+      "loss": 0.4587,
+      "step": 1188
+    },
+    {
+      "epoch": 1.6822363765038926,
+      "grad_norm": 0.23682103443569977,
+      "learning_rate": 2.4396642182581324e-05,
+      "loss": 0.4478,
+      "step": 1189
+    },
+    {
+      "epoch": 1.6836518046709128,
+      "grad_norm": 0.22594744221435362,
+      "learning_rate": 2.43704092339979e-05,
+      "loss": 0.4854,
+      "step": 1190
+    },
+    {
+      "epoch": 1.6850672328379335,
+      "grad_norm": 0.2734834603871489,
+      "learning_rate": 2.434417628541448e-05,
+      "loss": 0.4582,
+      "step": 1191
+    },
+    {
+      "epoch": 1.6864826610049541,
+      "grad_norm": 0.21788822338250274,
+      "learning_rate": 2.431794333683106e-05,
+      "loss": 0.3986,
+      "step": 1192
+    },
+    {
+      "epoch": 1.6878980891719744,
+      "grad_norm": 0.238195100497809,
+      "learning_rate": 2.4291710388247642e-05,
+      "loss": 0.4712,
+      "step": 1193
+    },
+    {
+      "epoch": 1.689313517338995,
+      "grad_norm": 0.2787782727643181,
+      "learning_rate": 2.426547743966422e-05,
+      "loss": 0.4726,
+      "step": 1194
+    },
+    {
+      "epoch": 1.6907289455060157,
+      "grad_norm": 0.4416723084952574,
+      "learning_rate": 2.42392444910808e-05,
+      "loss": 0.4401,
+      "step": 1195
+    },
+    {
+      "epoch": 1.692144373673036,
+      "grad_norm": 0.2419495654272071,
+      "learning_rate": 2.4213011542497376e-05,
+      "loss": 0.4695,
+      "step": 1196
+    },
+    {
+      "epoch": 1.6935598018400566,
+      "grad_norm": 0.26693116473266787,
+      "learning_rate": 2.4186778593913956e-05,
+      "loss": 0.4934,
+      "step": 1197
+    },
+    {
+      "epoch": 1.6949752300070773,
+      "grad_norm": 0.22799379548924234,
+      "learning_rate": 2.4160545645330536e-05,
+      "loss": 0.4356,
+      "step": 1198
+    },
+    {
+      "epoch": 1.6963906581740975,
+      "grad_norm": 0.9900211172083577,
+      "learning_rate": 2.4134312696747117e-05,
+      "loss": 0.4798,
+      "step": 1199
+    },
+    {
+      "epoch": 1.6978060863411182,
+      "grad_norm": 0.23148916684473117,
+      "learning_rate": 2.4108079748163694e-05,
+      "loss": 0.4515,
+      "step": 1200
+    },
+    {
+      "epoch": 1.6992215145081389,
+      "grad_norm": 0.22874042461258706,
+      "learning_rate": 2.4081846799580274e-05,
+      "loss": 0.4527,
+      "step": 1201
+    },
+    {
+      "epoch": 1.700636942675159,
+      "grad_norm": 0.24211185463412316,
+      "learning_rate": 2.4055613850996854e-05,
+      "loss": 0.445,
+      "step": 1202
+    },
+    {
+      "epoch": 1.7020523708421798,
+      "grad_norm": 0.30431255476741226,
+      "learning_rate": 2.402938090241343e-05,
+      "loss": 0.5107,
+      "step": 1203
+    },
+    {
+      "epoch": 1.7034677990092004,
+      "grad_norm": 0.22113554817941933,
+      "learning_rate": 2.400314795383001e-05,
+      "loss": 0.45,
+      "step": 1204
+    },
+    {
+      "epoch": 1.7048832271762207,
+      "grad_norm": 0.24032698117750861,
+      "learning_rate": 2.397691500524659e-05,
+      "loss": 0.5043,
+      "step": 1205
+    },
+    {
+      "epoch": 1.7062986553432413,
+      "grad_norm": 0.23870364857077842,
+      "learning_rate": 2.3950682056663172e-05,
+      "loss": 0.5209,
+      "step": 1206
+    },
+    {
+      "epoch": 1.707714083510262,
+      "grad_norm": 0.2302688871180308,
+      "learning_rate": 2.392444910807975e-05,
+      "loss": 0.4242,
+      "step": 1207
+    },
+    {
+      "epoch": 1.7091295116772822,
+      "grad_norm": 0.2436155876922659,
+      "learning_rate": 2.389821615949633e-05,
+      "loss": 0.4704,
+      "step": 1208
+    },
+    {
+      "epoch": 1.710544939844303,
+      "grad_norm": 0.4273326272781385,
+      "learning_rate": 2.3871983210912906e-05,
+      "loss": 0.4401,
+      "step": 1209
+    },
+    {
+      "epoch": 1.7119603680113236,
+      "grad_norm": 0.268965312167453,
+      "learning_rate": 2.3845750262329486e-05,
+      "loss": 0.4998,
+      "step": 1210
+    },
+    {
+      "epoch": 1.7133757961783438,
+      "grad_norm": 0.21968584742302194,
+      "learning_rate": 2.3819517313746067e-05,
+      "loss": 0.4093,
+      "step": 1211
+    },
+    {
+      "epoch": 1.7147912243453645,
+      "grad_norm": 0.23577769866318082,
+      "learning_rate": 2.3793284365162647e-05,
+      "loss": 0.4447,
+      "step": 1212
+    },
+    {
+      "epoch": 1.7162066525123851,
+      "grad_norm": 0.22990195035734406,
+      "learning_rate": 2.3767051416579224e-05,
+      "loss": 0.5292,
+      "step": 1213
+    },
+    {
+      "epoch": 1.7176220806794054,
+      "grad_norm": 0.21474090925988665,
+      "learning_rate": 2.3740818467995804e-05,
+      "loss": 0.503,
+      "step": 1214
+    },
+    {
+      "epoch": 1.719037508846426,
+      "grad_norm": 0.24605939907412877,
+      "learning_rate": 2.3714585519412384e-05,
+      "loss": 0.4326,
+      "step": 1215
+    },
+    {
+      "epoch": 1.7204529370134467,
+      "grad_norm": 0.24308945674870652,
+      "learning_rate": 2.368835257082896e-05,
+      "loss": 0.3943,
+      "step": 1216
+    },
+    {
+      "epoch": 1.721868365180467,
+      "grad_norm": 0.21001902737119993,
+      "learning_rate": 2.366211962224554e-05,
+      "loss": 0.5185,
+      "step": 1217
+    },
+    {
+      "epoch": 1.7232837933474876,
+      "grad_norm": 0.22899520312784166,
+      "learning_rate": 2.3635886673662122e-05,
+      "loss": 0.4642,
+      "step": 1218
+    },
+    {
+      "epoch": 1.7246992215145083,
+      "grad_norm": 0.23724020514959368,
+      "learning_rate": 2.3609653725078702e-05,
+      "loss": 0.4431,
+      "step": 1219
+    },
+    {
+      "epoch": 1.7261146496815285,
+      "grad_norm": 0.2263807462898294,
+      "learning_rate": 2.358342077649528e-05,
+      "loss": 0.5352,
+      "step": 1220
+    },
+    {
+      "epoch": 1.7275300778485492,
+      "grad_norm": 0.2230790085454061,
+      "learning_rate": 2.355718782791186e-05,
+      "loss": 0.481,
+      "step": 1221
+    },
+    {
+      "epoch": 1.7289455060155698,
+      "grad_norm": 0.23095257603790728,
+      "learning_rate": 2.3530954879328436e-05,
+      "loss": 0.4367,
+      "step": 1222
+    },
+    {
+      "epoch": 1.73036093418259,
+      "grad_norm": 0.22096408952247884,
+      "learning_rate": 2.3504721930745016e-05,
+      "loss": 0.4688,
+      "step": 1223
+    },
+    {
+      "epoch": 1.7317763623496107,
+      "grad_norm": 0.21323951749813636,
+      "learning_rate": 2.3478488982161593e-05,
+      "loss": 0.4884,
+      "step": 1224
+    },
+    {
+      "epoch": 1.7331917905166314,
+      "grad_norm": 0.21045169434522537,
+      "learning_rate": 2.3452256033578177e-05,
+      "loss": 0.4512,
+      "step": 1225
+    },
+    {
+      "epoch": 1.7346072186836516,
+      "grad_norm": 0.22474956769797635,
+      "learning_rate": 2.3426023084994754e-05,
+      "loss": 0.531,
+      "step": 1226
+    },
+    {
+      "epoch": 1.7360226468506723,
+      "grad_norm": 0.24318820018692644,
+      "learning_rate": 2.3399790136411334e-05,
+      "loss": 0.4697,
+      "step": 1227
+    },
+    {
+      "epoch": 1.737438075017693,
+      "grad_norm": 0.22215833754925915,
+      "learning_rate": 2.3373557187827915e-05,
+      "loss": 0.4926,
+      "step": 1228
+    },
+    {
+      "epoch": 1.7388535031847132,
+      "grad_norm": 0.21453240070070215,
+      "learning_rate": 2.334732423924449e-05,
+      "loss": 0.4158,
+      "step": 1229
+    },
+    {
+      "epoch": 1.7402689313517339,
+      "grad_norm": 0.2586416867995017,
+      "learning_rate": 2.332109129066107e-05,
+      "loss": 0.4979,
+      "step": 1230
+    },
+    {
+      "epoch": 1.7416843595187546,
+      "grad_norm": 0.24659631712596117,
+      "learning_rate": 2.329485834207765e-05,
+      "loss": 0.5521,
+      "step": 1231
+    },
+    {
+      "epoch": 1.7430997876857748,
+      "grad_norm": 0.32417069581572344,
+      "learning_rate": 2.3268625393494232e-05,
+      "loss": 0.4195,
+      "step": 1232
+    },
+    {
+      "epoch": 1.7445152158527955,
+      "grad_norm": 0.23039000450013308,
+      "learning_rate": 2.324239244491081e-05,
+      "loss": 0.5154,
+      "step": 1233
+    },
+    {
+      "epoch": 1.7459306440198161,
+      "grad_norm": 0.23344012631501052,
+      "learning_rate": 2.321615949632739e-05,
+      "loss": 0.4377,
+      "step": 1234
+    },
+    {
+      "epoch": 1.7473460721868364,
+      "grad_norm": 0.21715813767080286,
+      "learning_rate": 2.3189926547743966e-05,
+      "loss": 0.4044,
+      "step": 1235
+    },
+    {
+      "epoch": 1.748761500353857,
+      "grad_norm": 0.2725802380164536,
+      "learning_rate": 2.3163693599160547e-05,
+      "loss": 0.4771,
+      "step": 1236
+    },
+    {
+      "epoch": 1.7501769285208777,
+      "grad_norm": 0.23461308933319772,
+      "learning_rate": 2.3137460650577124e-05,
+      "loss": 0.5712,
+      "step": 1237
+    },
+    {
+      "epoch": 1.7515923566878981,
+      "grad_norm": 0.21422574478415968,
+      "learning_rate": 2.3111227701993704e-05,
+      "loss": 0.4413,
+      "step": 1238
+    },
+    {
+      "epoch": 1.7530077848549186,
+      "grad_norm": 0.27760237803106597,
+      "learning_rate": 2.3084994753410284e-05,
+      "loss": 0.4901,
+      "step": 1239
+    },
+    {
+      "epoch": 1.7544232130219393,
+      "grad_norm": 4.188952617257135,
+      "learning_rate": 2.3058761804826864e-05,
+      "loss": 0.5375,
+      "step": 1240
+    },
+    {
+      "epoch": 1.7558386411889597,
+      "grad_norm": 0.2480906746200484,
+      "learning_rate": 2.3032528856243445e-05,
+      "loss": 0.483,
+      "step": 1241
+    },
+    {
+      "epoch": 1.7572540693559802,
+      "grad_norm": 0.22531099081716,
+      "learning_rate": 2.300629590766002e-05,
+      "loss": 0.5066,
+      "step": 1242
+    },
+    {
+      "epoch": 1.7586694975230008,
+      "grad_norm": 0.21940727426318465,
+      "learning_rate": 2.2980062959076602e-05,
+      "loss": 0.4087,
+      "step": 1243
+    },
+    {
+      "epoch": 1.7600849256900213,
+      "grad_norm": 0.19554804756344746,
+      "learning_rate": 2.295383001049318e-05,
+      "loss": 0.3825,
+      "step": 1244
+    },
+    {
+      "epoch": 1.7615003538570417,
+      "grad_norm": 0.22562606588810402,
+      "learning_rate": 2.292759706190976e-05,
+      "loss": 0.4439,
+      "step": 1245
+    },
+    {
+      "epoch": 1.7629157820240624,
+      "grad_norm": 0.2271682488662497,
+      "learning_rate": 2.290136411332634e-05,
+      "loss": 0.4196,
+      "step": 1246
+    },
+    {
+      "epoch": 1.7643312101910829,
+      "grad_norm": 0.21455820735860226,
+      "learning_rate": 2.287513116474292e-05,
+      "loss": 0.4322,
+      "step": 1247
+    },
+    {
+      "epoch": 1.7657466383581033,
+      "grad_norm": 0.22651953039208592,
+      "learning_rate": 2.2848898216159496e-05,
+      "loss": 0.4682,
+      "step": 1248
+    },
+    {
+      "epoch": 1.767162066525124,
+      "grad_norm": 0.24921888820516835,
+      "learning_rate": 2.2822665267576077e-05,
+      "loss": 0.4786,
+      "step": 1249
+    },
+    {
+      "epoch": 1.7685774946921444,
+      "grad_norm": 0.2351290162709047,
+      "learning_rate": 2.2796432318992657e-05,
+      "loss": 0.4688,
+      "step": 1250
+    },
+    {
+      "epoch": 1.7699929228591649,
+      "grad_norm": 0.22097668014081087,
+      "learning_rate": 2.2770199370409234e-05,
+      "loss": 0.4237,
+      "step": 1251
+    },
+    {
+      "epoch": 1.7714083510261855,
+      "grad_norm": 0.24303531282271149,
+      "learning_rate": 2.2743966421825814e-05,
+      "loss": 0.4993,
+      "step": 1252
+    },
+    {
+      "epoch": 1.772823779193206,
+      "grad_norm": 0.999217635929671,
+      "learning_rate": 2.2717733473242395e-05,
+      "loss": 0.4566,
+      "step": 1253
+    },
+    {
+      "epoch": 1.7742392073602264,
+      "grad_norm": 0.23661072479165118,
+      "learning_rate": 2.2691500524658975e-05,
+      "loss": 0.4344,
+      "step": 1254
+    },
+    {
+      "epoch": 1.7756546355272471,
+      "grad_norm": 0.2238140247175545,
+      "learning_rate": 2.2665267576075552e-05,
+      "loss": 0.4854,
+      "step": 1255
+    },
+    {
+      "epoch": 1.7770700636942676,
+      "grad_norm": 0.23846918971265546,
+      "learning_rate": 2.2639034627492132e-05,
+      "loss": 0.4918,
+      "step": 1256
+    },
+    {
+      "epoch": 1.778485491861288,
+      "grad_norm": 0.27010281272058134,
+      "learning_rate": 2.261280167890871e-05,
+      "loss": 0.4618,
+      "step": 1257
+    },
+    {
+      "epoch": 1.7799009200283087,
+      "grad_norm": 0.2402855254748168,
+      "learning_rate": 2.258656873032529e-05,
+      "loss": 0.4816,
+      "step": 1258
+    },
+    {
+      "epoch": 1.7813163481953291,
+      "grad_norm": 0.284861020752327,
+      "learning_rate": 2.2560335781741866e-05,
+      "loss": 0.4537,
+      "step": 1259
+    },
+    {
+      "epoch": 1.7827317763623496,
+      "grad_norm": 0.23162852747166598,
+      "learning_rate": 2.253410283315845e-05,
+      "loss": 0.5254,
+      "step": 1260
+    },
+    {
+      "epoch": 1.7841472045293703,
+      "grad_norm": 0.23010356275368837,
+      "learning_rate": 2.2507869884575027e-05,
+      "loss": 0.4629,
+      "step": 1261
+    },
+    {
+      "epoch": 1.7855626326963907,
+      "grad_norm": 0.26477440200537117,
+      "learning_rate": 2.2481636935991607e-05,
+      "loss": 0.431,
+      "step": 1262
+    },
+    {
+      "epoch": 1.7869780608634112,
+      "grad_norm": 0.2469472463475338,
+      "learning_rate": 2.2455403987408187e-05,
+      "loss": 0.5095,
+      "step": 1263
+    },
+    {
+      "epoch": 1.7883934890304318,
+      "grad_norm": 0.250854568912954,
+      "learning_rate": 2.2429171038824764e-05,
+      "loss": 0.4695,
+      "step": 1264
+    },
+    {
+      "epoch": 1.7898089171974523,
+      "grad_norm": 0.2358553336427738,
+      "learning_rate": 2.2402938090241344e-05,
+      "loss": 0.4718,
+      "step": 1265
+    },
+    {
+      "epoch": 1.7912243453644727,
+      "grad_norm": 0.26092941743814313,
+      "learning_rate": 2.237670514165792e-05,
+      "loss": 0.4844,
+      "step": 1266
+    },
+    {
+      "epoch": 1.7926397735314934,
+      "grad_norm": 0.25364777428866075,
+      "learning_rate": 2.2350472193074505e-05,
+      "loss": 0.5053,
+      "step": 1267
+    },
+    {
+      "epoch": 1.7940552016985138,
+      "grad_norm": 0.22195077907811386,
+      "learning_rate": 2.2324239244491082e-05,
+      "loss": 0.4421,
+      "step": 1268
+    },
+    {
+      "epoch": 1.7954706298655343,
+      "grad_norm": 1.4410343829585255,
+      "learning_rate": 2.2298006295907662e-05,
+      "loss": 0.4146,
+      "step": 1269
+    },
+    {
+      "epoch": 1.796886058032555,
+      "grad_norm": 0.25129972037159265,
+      "learning_rate": 2.227177334732424e-05,
+      "loss": 0.483,
+      "step": 1270
+    },
+    {
+      "epoch": 1.7983014861995754,
+      "grad_norm": 0.26142130013103143,
+      "learning_rate": 2.224554039874082e-05,
+      "loss": 0.4594,
+      "step": 1271
+    },
+    {
+      "epoch": 1.7997169143665959,
+      "grad_norm": 0.23266793598351823,
+      "learning_rate": 2.2219307450157396e-05,
+      "loss": 0.491,
+      "step": 1272
+    },
+    {
+      "epoch": 1.8011323425336165,
+      "grad_norm": 0.2303010389050869,
+      "learning_rate": 2.2193074501573977e-05,
+      "loss": 0.4874,
+      "step": 1273
+    },
+    {
+      "epoch": 1.802547770700637,
+      "grad_norm": 0.23118206006279696,
+      "learning_rate": 2.2166841552990557e-05,
+      "loss": 0.3986,
+      "step": 1274
+    },
+    {
+      "epoch": 1.8039631988676574,
+      "grad_norm": 0.2143231238326485,
+      "learning_rate": 2.2140608604407137e-05,
+      "loss": 0.3811,
+      "step": 1275
+    },
+    {
+      "epoch": 1.805378627034678,
+      "grad_norm": 0.22724689078735588,
+      "learning_rate": 2.2114375655823717e-05,
+      "loss": 0.4208,
+      "step": 1276
+    },
+    {
+      "epoch": 1.8067940552016986,
+      "grad_norm": 2.0654374438327734,
+      "learning_rate": 2.2088142707240294e-05,
+      "loss": 0.5874,
+      "step": 1277
+    },
+    {
+      "epoch": 1.808209483368719,
+      "grad_norm": 0.2409280392188967,
+      "learning_rate": 2.2061909758656875e-05,
+      "loss": 0.5285,
+      "step": 1278
+    },
+    {
+      "epoch": 1.8096249115357397,
+      "grad_norm": 0.23334400572166863,
+      "learning_rate": 2.203567681007345e-05,
+      "loss": 0.4063,
+      "step": 1279
+    },
+    {
+      "epoch": 1.8110403397027601,
+      "grad_norm": 0.22984388021071223,
+      "learning_rate": 2.2009443861490035e-05,
+      "loss": 0.4671,
+      "step": 1280
+    },
+    {
+      "epoch": 1.8124557678697806,
+      "grad_norm": 0.2301151153313693,
+      "learning_rate": 2.1983210912906612e-05,
+      "loss": 0.5002,
+      "step": 1281
+    },
+    {
+      "epoch": 1.8138711960368012,
+      "grad_norm": 0.24246653508800442,
+      "learning_rate": 2.1956977964323192e-05,
+      "loss": 0.4151,
+      "step": 1282
+    },
+    {
+      "epoch": 1.8152866242038217,
+      "grad_norm": 0.2525452240180046,
+      "learning_rate": 2.193074501573977e-05,
+      "loss": 0.4274,
+      "step": 1283
+    },
+    {
+      "epoch": 1.8167020523708421,
+      "grad_norm": 0.20804041175775848,
+      "learning_rate": 2.190451206715635e-05,
+      "loss": 0.4412,
+      "step": 1284
+    },
+    {
+      "epoch": 1.8181174805378628,
+      "grad_norm": 0.23209003341106552,
+      "learning_rate": 2.1878279118572926e-05,
+      "loss": 0.4785,
+      "step": 1285
+    },
+    {
+      "epoch": 1.8195329087048833,
+      "grad_norm": 0.21506235966409293,
+      "learning_rate": 2.1852046169989507e-05,
+      "loss": 0.4901,
+      "step": 1286
+    },
+    {
+      "epoch": 1.8209483368719037,
+      "grad_norm": 0.24201084469050718,
+      "learning_rate": 2.1825813221406087e-05,
+      "loss": 0.4653,
+      "step": 1287
+    },
+    {
+      "epoch": 1.8223637650389244,
+      "grad_norm": 0.23237298583987862,
+      "learning_rate": 2.1799580272822667e-05,
+      "loss": 0.4521,
+      "step": 1288
+    },
+    {
+      "epoch": 1.8237791932059448,
+      "grad_norm": 0.23184894299524061,
+      "learning_rate": 2.1773347324239248e-05,
+      "loss": 0.4141,
+      "step": 1289
+    },
+    {
+      "epoch": 1.8251946213729653,
+      "grad_norm": 0.21915908624117594,
+      "learning_rate": 2.1747114375655824e-05,
+      "loss": 0.4833,
+      "step": 1290
+    },
+    {
+      "epoch": 1.826610049539986,
+      "grad_norm": 0.2204582698480696,
+      "learning_rate": 2.1720881427072405e-05,
+      "loss": 0.4646,
+      "step": 1291
+    },
+    {
+      "epoch": 1.8280254777070064,
+      "grad_norm": 0.23857430239630453,
+      "learning_rate": 2.169464847848898e-05,
+      "loss": 0.4334,
+      "step": 1292
+    },
+    {
+      "epoch": 1.8294409058740269,
+      "grad_norm": 0.2626478642231881,
+      "learning_rate": 2.1668415529905562e-05,
+      "loss": 0.4823,
+      "step": 1293
+    },
+    {
+      "epoch": 1.8308563340410475,
+      "grad_norm": 0.22467020564400583,
+      "learning_rate": 2.1642182581322142e-05,
+      "loss": 0.5265,
+      "step": 1294
+    },
+    {
+      "epoch": 1.832271762208068,
+      "grad_norm": 0.24570130352348415,
+      "learning_rate": 2.1615949632738722e-05,
+      "loss": 0.4235,
+      "step": 1295
+    },
+    {
+      "epoch": 1.8336871903750884,
+      "grad_norm": 0.23356731544612377,
+      "learning_rate": 2.15897166841553e-05,
+      "loss": 0.503,
+      "step": 1296
+    },
+    {
+      "epoch": 1.835102618542109,
+      "grad_norm": 0.25095911275324184,
+      "learning_rate": 2.156348373557188e-05,
+      "loss": 0.5299,
+      "step": 1297
+    },
+    {
+      "epoch": 1.8365180467091295,
+      "grad_norm": 0.31733420723462563,
+      "learning_rate": 2.1537250786988457e-05,
+      "loss": 0.5588,
+      "step": 1298
+    },
+    {
+      "epoch": 1.83793347487615,
+      "grad_norm": 0.23939961643806795,
+      "learning_rate": 2.1511017838405037e-05,
+      "loss": 0.4863,
+      "step": 1299
+    },
+    {
+      "epoch": 1.8393489030431707,
+      "grad_norm": 0.2229451447526651,
+      "learning_rate": 2.1484784889821617e-05,
+      "loss": 0.4864,
+      "step": 1300
+    },
+    {
+      "epoch": 1.8407643312101911,
+      "grad_norm": 0.28005231708013956,
+      "learning_rate": 2.1458551941238197e-05,
+      "loss": 0.4793,
+      "step": 1301
+    },
+    {
+      "epoch": 1.8421797593772116,
+      "grad_norm": 0.2416939594691491,
+      "learning_rate": 2.1432318992654778e-05,
+      "loss": 0.4641,
+      "step": 1302
+    },
+    {
+      "epoch": 1.8435951875442322,
+      "grad_norm": 0.22685599373710225,
+      "learning_rate": 2.1406086044071355e-05,
+      "loss": 0.4525,
+      "step": 1303
+    },
+    {
+      "epoch": 1.8450106157112527,
+      "grad_norm": 0.22150306708586515,
+      "learning_rate": 2.1379853095487935e-05,
+      "loss": 0.4576,
+      "step": 1304
+    },
+    {
+      "epoch": 1.8464260438782731,
+      "grad_norm": 0.23965364622580573,
+      "learning_rate": 2.1353620146904512e-05,
+      "loss": 0.4441,
+      "step": 1305
+    },
+    {
+      "epoch": 1.8478414720452938,
+      "grad_norm": 0.24247715310259696,
+      "learning_rate": 2.1327387198321092e-05,
+      "loss": 0.4975,
+      "step": 1306
+    },
+    {
+      "epoch": 1.8492569002123143,
+      "grad_norm": 0.22511524026533378,
+      "learning_rate": 2.130115424973767e-05,
+      "loss": 0.5094,
+      "step": 1307
+    },
+    {
+      "epoch": 1.8506723283793347,
+      "grad_norm": 0.24407240925469395,
+      "learning_rate": 2.1274921301154253e-05,
+      "loss": 0.4551,
+      "step": 1308
+    },
+    {
+      "epoch": 1.8520877565463554,
+      "grad_norm": 0.24306659852597814,
+      "learning_rate": 2.124868835257083e-05,
+      "loss": 0.4386,
+      "step": 1309
+    },
+    {
+      "epoch": 1.8535031847133758,
+      "grad_norm": 0.2241048925204749,
+      "learning_rate": 2.122245540398741e-05,
+      "loss": 0.471,
+      "step": 1310
+    },
+    {
+      "epoch": 1.8549186128803963,
+      "grad_norm": 0.22335342460079466,
+      "learning_rate": 2.119622245540399e-05,
+      "loss": 0.4297,
+      "step": 1311
+    },
+    {
+      "epoch": 1.856334041047417,
+      "grad_norm": 0.22466883866462156,
+      "learning_rate": 2.1169989506820567e-05,
+      "loss": 0.4116,
+      "step": 1312
+    },
+    {
+      "epoch": 1.8577494692144374,
+      "grad_norm": 0.22804012366044188,
+      "learning_rate": 2.1143756558237147e-05,
+      "loss": 0.4921,
+      "step": 1313
+    },
+    {
+      "epoch": 1.8591648973814578,
+      "grad_norm": 0.2513473537546202,
+      "learning_rate": 2.1117523609653724e-05,
+      "loss": 0.4047,
+      "step": 1314
+    },
+    {
+      "epoch": 1.8605803255484785,
+      "grad_norm": 0.2084112605966311,
+      "learning_rate": 2.1091290661070308e-05,
+      "loss": 0.4821,
+      "step": 1315
+    },
+    {
+      "epoch": 1.861995753715499,
+      "grad_norm": 0.23646819837085292,
+      "learning_rate": 2.1065057712486885e-05,
+      "loss": 0.3978,
+      "step": 1316
+    },
+    {
+      "epoch": 1.8634111818825194,
+      "grad_norm": 0.22261204824879074,
+      "learning_rate": 2.1038824763903465e-05,
+      "loss": 0.4947,
+      "step": 1317
+    },
+    {
+      "epoch": 1.86482661004954,
+      "grad_norm": 0.2437405611108718,
+      "learning_rate": 2.1012591815320042e-05,
+      "loss": 0.439,
+      "step": 1318
+    },
+    {
+      "epoch": 1.8662420382165605,
+      "grad_norm": 0.23344188200332183,
+      "learning_rate": 2.0986358866736622e-05,
+      "loss": 0.4871,
+      "step": 1319
+    },
+    {
+      "epoch": 1.867657466383581,
+      "grad_norm": 0.23982921750771513,
+      "learning_rate": 2.09601259181532e-05,
+      "loss": 0.482,
+      "step": 1320
+    },
+    {
+      "epoch": 1.8690728945506017,
+      "grad_norm": 0.23790242563138025,
+      "learning_rate": 2.093389296956978e-05,
+      "loss": 0.4202,
+      "step": 1321
+    },
+    {
+      "epoch": 1.870488322717622,
+      "grad_norm": 0.23424229708006536,
+      "learning_rate": 2.090766002098636e-05,
+      "loss": 0.4865,
+      "step": 1322
+    },
+    {
+      "epoch": 1.8719037508846426,
+      "grad_norm": 0.22239898962936439,
+      "learning_rate": 2.088142707240294e-05,
+      "loss": 0.4049,
+      "step": 1323
+    },
+    {
+      "epoch": 1.8733191790516632,
+      "grad_norm": 0.19757565268413543,
+      "learning_rate": 2.085519412381952e-05,
+      "loss": 0.4318,
+      "step": 1324
+    },
+    {
+      "epoch": 1.8747346072186837,
+      "grad_norm": 0.24461105569164185,
+      "learning_rate": 2.0828961175236097e-05,
+      "loss": 0.4631,
+      "step": 1325
+    },
+    {
+      "epoch": 1.8761500353857041,
+      "grad_norm": 0.5947746440626941,
+      "learning_rate": 2.0802728226652677e-05,
+      "loss": 0.4549,
+      "step": 1326
+    },
+    {
+      "epoch": 1.8775654635527248,
+      "grad_norm": 0.22133962167412313,
+      "learning_rate": 2.0776495278069254e-05,
+      "loss": 0.5018,
+      "step": 1327
+    },
+    {
+      "epoch": 1.8789808917197452,
+      "grad_norm": 0.2233899656516572,
+      "learning_rate": 2.0750262329485835e-05,
+      "loss": 0.4883,
+      "step": 1328
+    },
+    {
+      "epoch": 1.8803963198867657,
+      "grad_norm": 0.2444529353766216,
+      "learning_rate": 2.0724029380902415e-05,
+      "loss": 0.5194,
+      "step": 1329
+    },
+    {
+      "epoch": 1.8818117480537864,
+      "grad_norm": 0.25965188711877873,
+      "learning_rate": 2.0697796432318995e-05,
+      "loss": 0.4585,
+      "step": 1330
+    },
+    {
+      "epoch": 1.8832271762208068,
+      "grad_norm": 0.21130420022445334,
+      "learning_rate": 2.0671563483735572e-05,
+      "loss": 0.5233,
+      "step": 1331
+    },
+    {
+      "epoch": 1.8846426043878273,
+      "grad_norm": 0.24423899208067895,
+      "learning_rate": 2.0645330535152152e-05,
+      "loss": 0.5032,
+      "step": 1332
+    },
+    {
+      "epoch": 1.886058032554848,
+      "grad_norm": 0.23473966436217122,
+      "learning_rate": 2.061909758656873e-05,
+      "loss": 0.444,
+      "step": 1333
+    },
+    {
+      "epoch": 1.8874734607218684,
+      "grad_norm": 0.23024212934980773,
+      "learning_rate": 2.059286463798531e-05,
+      "loss": 0.4864,
+      "step": 1334
+    },
+    {
+      "epoch": 1.8888888888888888,
+      "grad_norm": 0.23297536760980386,
+      "learning_rate": 2.056663168940189e-05,
+      "loss": 0.3823,
+      "step": 1335
+    },
+    {
+      "epoch": 1.8903043170559095,
+      "grad_norm": 0.2381046335808105,
+      "learning_rate": 2.054039874081847e-05,
+      "loss": 0.4893,
+      "step": 1336
+    },
+    {
+      "epoch": 1.89171974522293,
+      "grad_norm": 0.24897499927700323,
+      "learning_rate": 2.051416579223505e-05,
+      "loss": 0.5544,
+      "step": 1337
+    },
+    {
+      "epoch": 1.8931351733899504,
+      "grad_norm": 0.25742996674526,
+      "learning_rate": 2.0487932843651627e-05,
+      "loss": 0.4581,
+      "step": 1338
+    },
+    {
+      "epoch": 1.894550601556971,
+      "grad_norm": 0.22376579236901967,
+      "learning_rate": 2.0461699895068208e-05,
+      "loss": 0.4679,
+      "step": 1339
+    },
+    {
+      "epoch": 1.8959660297239915,
+      "grad_norm": 0.22104742077146222,
+      "learning_rate": 2.0435466946484784e-05,
+      "loss": 0.5172,
+      "step": 1340
+    },
+    {
+      "epoch": 1.897381457891012,
+      "grad_norm": 0.2800835936243694,
+      "learning_rate": 2.0409233997901365e-05,
+      "loss": 0.4541,
+      "step": 1341
+    },
+    {
+      "epoch": 1.8987968860580327,
+      "grad_norm": 0.23180196456243227,
+      "learning_rate": 2.038300104931794e-05,
+      "loss": 0.4398,
+      "step": 1342
+    },
+    {
+      "epoch": 1.900212314225053,
+      "grad_norm": 0.21488819834194242,
+      "learning_rate": 2.0356768100734525e-05,
+      "loss": 0.4557,
+      "step": 1343
+    },
+    {
+      "epoch": 1.9016277423920736,
+      "grad_norm": 0.24372869965856336,
+      "learning_rate": 2.0330535152151102e-05,
+      "loss": 0.4794,
+      "step": 1344
+    },
+    {
+      "epoch": 1.9030431705590942,
+      "grad_norm": 0.2366213579699343,
+      "learning_rate": 2.0304302203567683e-05,
+      "loss": 0.3981,
+      "step": 1345
+    },
+    {
+      "epoch": 1.9044585987261147,
+      "grad_norm": 0.3648752073142994,
+      "learning_rate": 2.027806925498426e-05,
+      "loss": 0.534,
+      "step": 1346
+    },
+    {
+      "epoch": 1.9058740268931351,
+      "grad_norm": 0.2197325729465737,
+      "learning_rate": 2.025183630640084e-05,
+      "loss": 0.3927,
+      "step": 1347
+    },
+    {
+      "epoch": 1.9072894550601558,
+      "grad_norm": 0.231081310750387,
+      "learning_rate": 2.022560335781742e-05,
+      "loss": 0.4873,
+      "step": 1348
+    },
+    {
+      "epoch": 1.9087048832271762,
+      "grad_norm": 0.24082762445235403,
+      "learning_rate": 2.0199370409233997e-05,
+      "loss": 0.4328,
+      "step": 1349
+    },
+    {
+      "epoch": 1.9101203113941967,
+      "grad_norm": 0.20825514018587932,
+      "learning_rate": 2.017313746065058e-05,
+      "loss": 0.4976,
+      "step": 1350
+    },
+    {
+      "epoch": 1.9115357395612174,
+      "grad_norm": 0.21894111983154066,
+      "learning_rate": 2.0146904512067157e-05,
+      "loss": 0.4424,
+      "step": 1351
+    },
+    {
+      "epoch": 1.9129511677282378,
+      "grad_norm": 0.2286478651129217,
+      "learning_rate": 2.0120671563483738e-05,
+      "loss": 0.4343,
+      "step": 1352
+    },
+    {
+      "epoch": 1.9143665958952583,
+      "grad_norm": 0.2204646971643326,
+      "learning_rate": 2.0094438614900315e-05,
+      "loss": 0.4452,
+      "step": 1353
+    },
+    {
+      "epoch": 1.915782024062279,
+      "grad_norm": 0.23638163886103097,
+      "learning_rate": 2.0068205666316895e-05,
+      "loss": 0.482,
+      "step": 1354
+    },
+    {
+      "epoch": 1.9171974522292994,
+      "grad_norm": 0.20878654613258196,
+      "learning_rate": 2.0041972717733472e-05,
+      "loss": 0.424,
+      "step": 1355
+    },
+    {
+      "epoch": 1.9186128803963198,
+      "grad_norm": 0.2098932184443533,
+      "learning_rate": 2.0015739769150055e-05,
+      "loss": 0.4047,
+      "step": 1356
+    },
+    {
+      "epoch": 1.9200283085633405,
+      "grad_norm": 0.20399735990780962,
+      "learning_rate": 1.9989506820566632e-05,
+      "loss": 0.4147,
+      "step": 1357
+    },
+    {
+      "epoch": 1.921443736730361,
+      "grad_norm": 0.22943335349547167,
+      "learning_rate": 1.9963273871983213e-05,
+      "loss": 0.477,
+      "step": 1358
+    },
+    {
+      "epoch": 1.9228591648973814,
+      "grad_norm": 0.2228970758057448,
+      "learning_rate": 1.993704092339979e-05,
+      "loss": 0.4254,
+      "step": 1359
+    },
+    {
+      "epoch": 1.924274593064402,
+      "grad_norm": 0.2107497813441486,
+      "learning_rate": 1.991080797481637e-05,
+      "loss": 0.4092,
+      "step": 1360
+    },
+    {
+      "epoch": 1.9256900212314225,
+      "grad_norm": 0.19967419194846606,
+      "learning_rate": 1.988457502623295e-05,
+      "loss": 0.4836,
+      "step": 1361
+    },
+    {
+      "epoch": 1.927105449398443,
+      "grad_norm": 0.23592377920831498,
+      "learning_rate": 1.9858342077649527e-05,
+      "loss": 0.4821,
+      "step": 1362
+    },
+    {
+      "epoch": 1.9285208775654636,
+      "grad_norm": 0.23650322697474802,
+      "learning_rate": 1.983210912906611e-05,
+      "loss": 0.4715,
+      "step": 1363
+    },
+    {
+      "epoch": 1.929936305732484,
+      "grad_norm": 0.2185083326059402,
+      "learning_rate": 1.9805876180482688e-05,
+      "loss": 0.5157,
+      "step": 1364
+    },
+    {
+      "epoch": 1.9313517338995045,
+      "grad_norm": 0.2296974244832122,
+      "learning_rate": 1.9779643231899268e-05,
+      "loss": 0.4627,
+      "step": 1365
+    },
+    {
+      "epoch": 1.9327671620665252,
+      "grad_norm": 0.21126190043837798,
+      "learning_rate": 1.9753410283315845e-05,
+      "loss": 0.4603,
+      "step": 1366
+    },
+    {
+      "epoch": 1.9341825902335457,
+      "grad_norm": 0.21423948829296396,
+      "learning_rate": 1.9727177334732425e-05,
+      "loss": 0.4744,
+      "step": 1367
+    },
+    {
+      "epoch": 1.9355980184005661,
+      "grad_norm": 0.21810060321266686,
+      "learning_rate": 1.9700944386149002e-05,
+      "loss": 0.4797,
+      "step": 1368
+    },
+    {
+      "epoch": 1.9370134465675868,
+      "grad_norm": 0.2828545367941199,
+      "learning_rate": 1.9674711437565582e-05,
+      "loss": 0.4612,
+      "step": 1369
+    },
+    {
+      "epoch": 1.9384288747346072,
+      "grad_norm": 0.2165297001551884,
+      "learning_rate": 1.9648478488982163e-05,
+      "loss": 0.4409,
+      "step": 1370
+    },
+    {
+      "epoch": 1.9398443029016277,
+      "grad_norm": 0.24431415598675452,
+      "learning_rate": 1.9622245540398743e-05,
+      "loss": 0.4572,
+      "step": 1371
+    },
+    {
+      "epoch": 1.9412597310686484,
+      "grad_norm": 0.4086000321654116,
+      "learning_rate": 1.9596012591815323e-05,
+      "loss": 0.4953,
+      "step": 1372
+    },
+    {
+      "epoch": 1.9426751592356688,
+      "grad_norm": 0.22847856039481462,
+      "learning_rate": 1.95697796432319e-05,
+      "loss": 0.3886,
+      "step": 1373
+    },
+    {
+      "epoch": 1.9440905874026893,
+      "grad_norm": 0.20902762981225703,
+      "learning_rate": 1.954354669464848e-05,
+      "loss": 0.4709,
+      "step": 1374
+    },
+    {
+      "epoch": 1.94550601556971,
+      "grad_norm": 0.2424786257148823,
+      "learning_rate": 1.9517313746065057e-05,
+      "loss": 0.5191,
+      "step": 1375
+    },
+    {
+      "epoch": 1.9469214437367304,
+      "grad_norm": 0.2159429193506879,
+      "learning_rate": 1.9491080797481637e-05,
+      "loss": 0.4453,
+      "step": 1376
+    },
+    {
+      "epoch": 1.9483368719037508,
+      "grad_norm": 0.23477092809891625,
+      "learning_rate": 1.9464847848898218e-05,
+      "loss": 0.4891,
+      "step": 1377
+    },
+    {
+      "epoch": 1.9497523000707715,
+      "grad_norm": 0.22425815180240935,
+      "learning_rate": 1.9438614900314798e-05,
+      "loss": 0.4888,
+      "step": 1378
+    },
+    {
+      "epoch": 1.951167728237792,
+      "grad_norm": 0.45474270956547114,
+      "learning_rate": 1.9412381951731375e-05,
+      "loss": 0.5187,
+      "step": 1379
+    },
+    {
+      "epoch": 1.9525831564048124,
+      "grad_norm": 0.2538581966186155,
+      "learning_rate": 1.9386149003147955e-05,
+      "loss": 0.4714,
+      "step": 1380
+    },
+    {
+      "epoch": 1.953998584571833,
+      "grad_norm": 0.22480122774302955,
+      "learning_rate": 1.9359916054564532e-05,
+      "loss": 0.4594,
+      "step": 1381
+    },
+    {
+      "epoch": 1.9554140127388535,
+      "grad_norm": 0.22940865698452123,
+      "learning_rate": 1.9333683105981112e-05,
+      "loss": 0.4497,
+      "step": 1382
+    },
+    {
+      "epoch": 1.956829440905874,
+      "grad_norm": 0.2297486104164606,
+      "learning_rate": 1.9307450157397693e-05,
+      "loss": 0.4814,
+      "step": 1383
+    },
+    {
+      "epoch": 1.9582448690728946,
+      "grad_norm": 0.2374799661012413,
+      "learning_rate": 1.9281217208814273e-05,
+      "loss": 0.4425,
+      "step": 1384
+    },
+    {
+      "epoch": 1.959660297239915,
+      "grad_norm": 0.2517317396047779,
+      "learning_rate": 1.9254984260230853e-05,
+      "loss": 0.4487,
+      "step": 1385
+    },
+    {
+      "epoch": 1.9610757254069355,
+      "grad_norm": 0.21629554380819382,
+      "learning_rate": 1.922875131164743e-05,
+      "loss": 0.4222,
+      "step": 1386
+    },
+    {
+      "epoch": 1.9624911535739562,
+      "grad_norm": 0.21894295567177233,
+      "learning_rate": 1.920251836306401e-05,
+      "loss": 0.5109,
+      "step": 1387
+    },
+    {
+      "epoch": 1.9639065817409767,
+      "grad_norm": 0.25551379680235153,
+      "learning_rate": 1.9176285414480587e-05,
+      "loss": 0.5169,
+      "step": 1388
+    },
+    {
+      "epoch": 1.965322009907997,
+      "grad_norm": 0.22701226384061032,
+      "learning_rate": 1.9150052465897168e-05,
+      "loss": 0.4238,
+      "step": 1389
+    },
+    {
+      "epoch": 1.9667374380750178,
+      "grad_norm": 0.21764512077540232,
+      "learning_rate": 1.9123819517313745e-05,
+      "loss": 0.483,
+      "step": 1390
+    },
+    {
+      "epoch": 1.9681528662420382,
+      "grad_norm": 0.2379062720137337,
+      "learning_rate": 1.9097586568730328e-05,
+      "loss": 0.4301,
+      "step": 1391
+    },
+    {
+      "epoch": 1.9695682944090587,
+      "grad_norm": 0.21706551907158597,
+      "learning_rate": 1.9071353620146905e-05,
+      "loss": 0.4435,
+      "step": 1392
+    },
+    {
+      "epoch": 1.9709837225760793,
+      "grad_norm": 0.2681923527504643,
+      "learning_rate": 1.9045120671563485e-05,
+      "loss": 0.5055,
+      "step": 1393
+    },
+    {
+      "epoch": 1.9723991507430998,
+      "grad_norm": 0.24764910859697023,
+      "learning_rate": 1.9018887722980062e-05,
+      "loss": 0.5379,
+      "step": 1394
+    },
+    {
+      "epoch": 1.9738145789101202,
+      "grad_norm": 0.24028803996244996,
+      "learning_rate": 1.8992654774396643e-05,
+      "loss": 0.4422,
+      "step": 1395
+    },
+    {
+      "epoch": 1.975230007077141,
+      "grad_norm": 0.2667908175923755,
+      "learning_rate": 1.8966421825813223e-05,
+      "loss": 0.4601,
+      "step": 1396
+    },
+    {
+      "epoch": 1.9766454352441614,
+      "grad_norm": 0.1983331591989112,
+      "learning_rate": 1.89401888772298e-05,
+      "loss": 0.3583,
+      "step": 1397
+    },
+    {
+      "epoch": 1.9780608634111818,
+      "grad_norm": 0.2129299178458175,
+      "learning_rate": 1.8913955928646383e-05,
+      "loss": 0.5322,
+      "step": 1398
+    },
+    {
+      "epoch": 1.9794762915782025,
+      "grad_norm": 0.24494279276071143,
+      "learning_rate": 1.888772298006296e-05,
+      "loss": 0.4284,
+      "step": 1399
+    },
+    {
+      "epoch": 1.980891719745223,
+      "grad_norm": 0.2346234778429022,
+      "learning_rate": 1.886149003147954e-05,
+      "loss": 0.5095,
+      "step": 1400
+    },
+    {
+      "epoch": 1.9823071479122434,
+      "grad_norm": 0.22148149677042864,
+      "learning_rate": 1.8835257082896118e-05,
+      "loss": 0.4191,
+      "step": 1401
+    },
+    {
+      "epoch": 1.983722576079264,
+      "grad_norm": 0.20471789202971866,
+      "learning_rate": 1.8809024134312698e-05,
+      "loss": 0.4225,
+      "step": 1402
+    },
+    {
+      "epoch": 1.9851380042462845,
+      "grad_norm": 0.24568010662632564,
+      "learning_rate": 1.8782791185729275e-05,
+      "loss": 0.4981,
+      "step": 1403
+    },
+    {
+      "epoch": 1.986553432413305,
+      "grad_norm": 0.21930415253021862,
+      "learning_rate": 1.8756558237145855e-05,
+      "loss": 0.4752,
+      "step": 1404
+    },
+    {
+      "epoch": 1.9879688605803256,
+      "grad_norm": 0.251898694328717,
+      "learning_rate": 1.8730325288562435e-05,
+      "loss": 0.4316,
+      "step": 1405
+    },
+    {
+      "epoch": 1.989384288747346,
+      "grad_norm": 0.2176564270865631,
+      "learning_rate": 1.8704092339979016e-05,
+      "loss": 0.5356,
+      "step": 1406
+    },
+    {
+      "epoch": 1.9907997169143665,
+      "grad_norm": 0.2330020823912981,
+      "learning_rate": 1.8677859391395592e-05,
+      "loss": 0.4502,
+      "step": 1407
+    },
+    {
+      "epoch": 1.9922151450813872,
+      "grad_norm": 0.2357008471649503,
+      "learning_rate": 1.8651626442812173e-05,
+      "loss": 0.463,
+      "step": 1408
+    },
+    {
+      "epoch": 1.9936305732484076,
+      "grad_norm": 0.2167498961975368,
+      "learning_rate": 1.8625393494228753e-05,
+      "loss": 0.4825,
+      "step": 1409
+    },
+    {
+      "epoch": 1.995046001415428,
+      "grad_norm": 0.3680365703435767,
+      "learning_rate": 1.859916054564533e-05,
+      "loss": 0.4899,
+      "step": 1410
+    },
+    {
+      "epoch": 1.9964614295824488,
+      "grad_norm": 0.2261187086530332,
+      "learning_rate": 1.857292759706191e-05,
+      "loss": 0.5036,
+      "step": 1411
+    },
+    {
+      "epoch": 1.9978768577494692,
+      "grad_norm": 0.22055517468425753,
+      "learning_rate": 1.854669464847849e-05,
+      "loss": 0.5127,
+      "step": 1412
+    },
+    {
+      "epoch": 1.9992922859164897,
+      "grad_norm": 0.24038986520047975,
+      "learning_rate": 1.852046169989507e-05,
+      "loss": 0.4774,
+      "step": 1413
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.3649991056816497,
+      "learning_rate": 1.8494228751311648e-05,
+      "loss": 0.4991,
+      "step": 1414
+    },
+    {
+      "epoch": 2.0014154281670207,
+      "grad_norm": 0.4166693991610525,
+      "learning_rate": 1.8467995802728228e-05,
+      "loss": 0.4337,
+      "step": 1415
+    },
+    {
+      "epoch": 2.002830856334041,
+      "grad_norm": 0.260666322974337,
+      "learning_rate": 1.8441762854144805e-05,
+      "loss": 0.3922,
+      "step": 1416
+    },
+    {
+      "epoch": 2.0042462845010616,
+      "grad_norm": 0.29238039510242636,
+      "learning_rate": 1.8415529905561385e-05,
+      "loss": 0.3747,
+      "step": 1417
+    },
+    {
+      "epoch": 2.0056617126680822,
+      "grad_norm": 0.3564915371981773,
+      "learning_rate": 1.8389296956977965e-05,
+      "loss": 0.382,
+      "step": 1418
+    },
+    {
+      "epoch": 2.0070771408351025,
+      "grad_norm": 0.26783850124835984,
+      "learning_rate": 1.8363064008394546e-05,
+      "loss": 0.4175,
+      "step": 1419
+    },
+    {
+      "epoch": 2.008492569002123,
+      "grad_norm": 0.25248301392487915,
+      "learning_rate": 1.8336831059811123e-05,
+      "loss": 0.3667,
+      "step": 1420
+    },
+    {
+      "epoch": 2.009907997169144,
+      "grad_norm": 0.2970158100432694,
+      "learning_rate": 1.8310598111227703e-05,
+      "loss": 0.4218,
+      "step": 1421
+    },
+    {
+      "epoch": 2.011323425336164,
+      "grad_norm": 0.32058082497326085,
+      "learning_rate": 1.8284365162644283e-05,
+      "loss": 0.3682,
+      "step": 1422
+    },
+    {
+      "epoch": 2.0127388535031847,
+      "grad_norm": 0.258438347594369,
+      "learning_rate": 1.825813221406086e-05,
+      "loss": 0.3814,
+      "step": 1423
+    },
+    {
+      "epoch": 2.0141542816702054,
+      "grad_norm": 0.2309440049259677,
+      "learning_rate": 1.823189926547744e-05,
+      "loss": 0.3613,
+      "step": 1424
+    },
+    {
+      "epoch": 2.0155697098372256,
+      "grad_norm": 0.23711843903773477,
+      "learning_rate": 1.820566631689402e-05,
+      "loss": 0.3638,
+      "step": 1425
+    },
+    {
+      "epoch": 2.0169851380042463,
+      "grad_norm": 0.30154935042770303,
+      "learning_rate": 1.81794333683106e-05,
+      "loss": 0.3911,
+      "step": 1426
+    },
+    {
+      "epoch": 2.018400566171267,
+      "grad_norm": 0.24985753759283674,
+      "learning_rate": 1.8153200419727178e-05,
+      "loss": 0.3859,
+      "step": 1427
+    },
+    {
+      "epoch": 2.019815994338287,
+      "grad_norm": 0.21822783879615154,
+      "learning_rate": 1.8126967471143758e-05,
+      "loss": 0.3692,
+      "step": 1428
+    },
+    {
+      "epoch": 2.021231422505308,
+      "grad_norm": 0.23574623808166204,
+      "learning_rate": 1.8100734522560335e-05,
+      "loss": 0.3694,
+      "step": 1429
+    },
+    {
+      "epoch": 2.0226468506723285,
+      "grad_norm": 0.2656054636640407,
+      "learning_rate": 1.8074501573976915e-05,
+      "loss": 0.3394,
+      "step": 1430
+    },
+    {
+      "epoch": 2.0240622788393487,
+      "grad_norm": 0.2603011508150877,
+      "learning_rate": 1.8048268625393496e-05,
+      "loss": 0.3267,
+      "step": 1431
+    },
+    {
+      "epoch": 2.0254777070063694,
+      "grad_norm": 0.23364246069398123,
+      "learning_rate": 1.8022035676810076e-05,
+      "loss": 0.3429,
+      "step": 1432
+    },
+    {
+      "epoch": 2.02689313517339,
+      "grad_norm": 0.2491023446129911,
+      "learning_rate": 1.7995802728226656e-05,
+      "loss": 0.4053,
+      "step": 1433
+    },
+    {
+      "epoch": 2.0283085633404103,
+      "grad_norm": 0.24128596635121036,
+      "learning_rate": 1.7969569779643233e-05,
+      "loss": 0.3407,
+      "step": 1434
+    },
+    {
+      "epoch": 2.029723991507431,
+      "grad_norm": 0.2803188728012592,
+      "learning_rate": 1.7943336831059813e-05,
+      "loss": 0.3737,
+      "step": 1435
+    },
+    {
+      "epoch": 2.0311394196744517,
+      "grad_norm": 0.27015953132865195,
+      "learning_rate": 1.791710388247639e-05,
+      "loss": 0.3921,
+      "step": 1436
+    },
+    {
+      "epoch": 2.032554847841472,
+      "grad_norm": 0.25020458517438576,
+      "learning_rate": 1.789087093389297e-05,
+      "loss": 0.3706,
+      "step": 1437
+    },
+    {
+      "epoch": 2.0339702760084926,
+      "grad_norm": 0.2640785254176496,
+      "learning_rate": 1.7864637985309547e-05,
+      "loss": 0.3618,
+      "step": 1438
+    },
+    {
+      "epoch": 2.0353857041755132,
+      "grad_norm": 0.240523144913415,
+      "learning_rate": 1.783840503672613e-05,
+      "loss": 0.332,
+      "step": 1439
+    },
+    {
+      "epoch": 2.0368011323425335,
+      "grad_norm": 0.23611315437615582,
+      "learning_rate": 1.7812172088142708e-05,
+      "loss": 0.2785,
+      "step": 1440
+    },
+    {
+      "epoch": 2.038216560509554,
+      "grad_norm": 0.2252806592035128,
+      "learning_rate": 1.7785939139559288e-05,
+      "loss": 0.3577,
+      "step": 1441
+    },
+    {
+      "epoch": 2.039631988676575,
+      "grad_norm": 0.24882838329271678,
+      "learning_rate": 1.7759706190975865e-05,
+      "loss": 0.4204,
+      "step": 1442
+    },
+    {
+      "epoch": 2.041047416843595,
+      "grad_norm": 0.2107192080934422,
+      "learning_rate": 1.7733473242392445e-05,
+      "loss": 0.3904,
+      "step": 1443
+    },
+    {
+      "epoch": 2.0424628450106157,
+      "grad_norm": 0.21057384801951756,
+      "learning_rate": 1.7707240293809026e-05,
+      "loss": 0.3695,
+      "step": 1444
+    },
+    {
+      "epoch": 2.0438782731776364,
+      "grad_norm": 0.2165214812310043,
+      "learning_rate": 1.7681007345225603e-05,
+      "loss": 0.3465,
+      "step": 1445
+    },
+    {
+      "epoch": 2.0452937013446566,
+      "grad_norm": 0.21451815510889052,
+      "learning_rate": 1.7654774396642186e-05,
+      "loss": 0.3473,
+      "step": 1446
+    },
+    {
+      "epoch": 2.0467091295116773,
+      "grad_norm": 0.2255237074033376,
+      "learning_rate": 1.7628541448058763e-05,
+      "loss": 0.3915,
+      "step": 1447
+    },
+    {
+      "epoch": 2.048124557678698,
+      "grad_norm": 0.21102539596402395,
+      "learning_rate": 1.7602308499475343e-05,
+      "loss": 0.4007,
+      "step": 1448
+    },
+    {
+      "epoch": 2.049539985845718,
+      "grad_norm": 0.23126500561761867,
+      "learning_rate": 1.757607555089192e-05,
+      "loss": 0.3449,
+      "step": 1449
+    },
+    {
+      "epoch": 2.050955414012739,
+      "grad_norm": 0.21701265916378112,
+      "learning_rate": 1.75498426023085e-05,
+      "loss": 0.3613,
+      "step": 1450
+    },
+    {
+      "epoch": 2.0523708421797595,
+      "grad_norm": 0.22711923189993327,
+      "learning_rate": 1.7523609653725078e-05,
+      "loss": 0.401,
+      "step": 1451
+    },
+    {
+      "epoch": 2.0537862703467797,
+      "grad_norm": 0.20861442180382758,
+      "learning_rate": 1.7497376705141658e-05,
+      "loss": 0.3655,
+      "step": 1452
+    },
+    {
+      "epoch": 2.0552016985138004,
+      "grad_norm": 0.2270859623252876,
+      "learning_rate": 1.7471143756558238e-05,
+      "loss": 0.3595,
+      "step": 1453
+    },
+    {
+      "epoch": 2.056617126680821,
+      "grad_norm": 0.23364629764481326,
+      "learning_rate": 1.744491080797482e-05,
+      "loss": 0.3295,
+      "step": 1454
+    },
+    {
+      "epoch": 2.0580325548478413,
+      "grad_norm": 0.22567027139214516,
+      "learning_rate": 1.7418677859391395e-05,
+      "loss": 0.3831,
+      "step": 1455
+    },
+    {
+      "epoch": 2.059447983014862,
+      "grad_norm": 0.21839431235112447,
+      "learning_rate": 1.7392444910807976e-05,
+      "loss": 0.3556,
+      "step": 1456
+    },
+    {
+      "epoch": 2.0608634111818827,
+      "grad_norm": 0.22426774563431578,
+      "learning_rate": 1.7366211962224556e-05,
+      "loss": 0.3484,
+      "step": 1457
+    },
+    {
+      "epoch": 2.062278839348903,
+      "grad_norm": 0.2233566251331184,
+      "learning_rate": 1.7339979013641133e-05,
+      "loss": 0.3736,
+      "step": 1458
+    },
+    {
+      "epoch": 2.0636942675159236,
+      "grad_norm": 0.19773703937830703,
+      "learning_rate": 1.7313746065057713e-05,
+      "loss": 0.3069,
+      "step": 1459
+    },
+    {
+      "epoch": 2.0651096956829442,
+      "grad_norm": 0.23091198480431166,
+      "learning_rate": 1.7287513116474293e-05,
+      "loss": 0.38,
+      "step": 1460
+    },
+    {
+      "epoch": 2.0665251238499645,
+      "grad_norm": 0.21888763266422998,
+      "learning_rate": 1.7261280167890874e-05,
+      "loss": 0.3814,
+      "step": 1461
+    },
+    {
+      "epoch": 2.067940552016985,
+      "grad_norm": 0.2118377785821229,
+      "learning_rate": 1.723504721930745e-05,
+      "loss": 0.3675,
+      "step": 1462
+    },
+    {
+      "epoch": 2.069355980184006,
+      "grad_norm": 0.20564189159820773,
+      "learning_rate": 1.720881427072403e-05,
+      "loss": 0.3421,
+      "step": 1463
+    },
+    {
+      "epoch": 2.070771408351026,
+      "grad_norm": 0.2169277793958252,
+      "learning_rate": 1.7182581322140608e-05,
+      "loss": 0.3781,
+      "step": 1464
+    },
+    {
+      "epoch": 2.0721868365180467,
+      "grad_norm": 0.20777975555156794,
+      "learning_rate": 1.7156348373557188e-05,
+      "loss": 0.3421,
+      "step": 1465
+    },
+    {
+      "epoch": 2.0736022646850674,
+      "grad_norm": 0.23434517928918236,
+      "learning_rate": 1.7130115424973768e-05,
+      "loss": 0.3926,
+      "step": 1466
+    },
+    {
+      "epoch": 2.0750176928520876,
+      "grad_norm": 0.2237156164837064,
+      "learning_rate": 1.710388247639035e-05,
+      "loss": 0.3952,
+      "step": 1467
+    },
+    {
+      "epoch": 2.0764331210191083,
+      "grad_norm": 0.20852775525749587,
+      "learning_rate": 1.7077649527806925e-05,
+      "loss": 0.3471,
+      "step": 1468
+    },
+    {
+      "epoch": 2.077848549186129,
+      "grad_norm": 0.21165243525823502,
+      "learning_rate": 1.7051416579223506e-05,
+      "loss": 0.3677,
+      "step": 1469
+    },
+    {
+      "epoch": 2.079263977353149,
+      "grad_norm": 1.1822920981729839,
+      "learning_rate": 1.7025183630640086e-05,
+      "loss": 0.4679,
+      "step": 1470
+    },
+    {
+      "epoch": 2.08067940552017,
+      "grad_norm": 0.2479312294370373,
+      "learning_rate": 1.6998950682056663e-05,
+      "loss": 0.3967,
+      "step": 1471
+    },
+    {
+      "epoch": 2.0820948336871905,
+      "grad_norm": 0.25982595855025337,
+      "learning_rate": 1.6972717733473243e-05,
+      "loss": 0.3706,
+      "step": 1472
+    },
+    {
+      "epoch": 2.0835102618542107,
+      "grad_norm": 0.21858931940249068,
+      "learning_rate": 1.694648478488982e-05,
+      "loss": 0.3613,
+      "step": 1473
+    },
+    {
+      "epoch": 2.0849256900212314,
+      "grad_norm": 0.22549270661935017,
+      "learning_rate": 1.6920251836306404e-05,
+      "loss": 0.3655,
+      "step": 1474
+    },
+    {
+      "epoch": 2.086341118188252,
+      "grad_norm": 0.2121034477698962,
+      "learning_rate": 1.689401888772298e-05,
+      "loss": 0.3416,
+      "step": 1475
+    },
+    {
+      "epoch": 2.0877565463552723,
+      "grad_norm": 0.2034045361151294,
+      "learning_rate": 1.686778593913956e-05,
+      "loss": 0.3352,
+      "step": 1476
+    },
+    {
+      "epoch": 2.089171974522293,
+      "grad_norm": 0.21723699645973873,
+      "learning_rate": 1.6841552990556138e-05,
+      "loss": 0.3432,
+      "step": 1477
+    },
+    {
+      "epoch": 2.0905874026893136,
+      "grad_norm": 0.23426304130031295,
+      "learning_rate": 1.6815320041972718e-05,
+      "loss": 0.3608,
+      "step": 1478
+    },
+    {
+      "epoch": 2.092002830856334,
+      "grad_norm": 0.20725051222778673,
+      "learning_rate": 1.67890870933893e-05,
+      "loss": 0.347,
+      "step": 1479
+    },
+    {
+      "epoch": 2.0934182590233545,
+      "grad_norm": 0.23093322390994894,
+      "learning_rate": 1.6762854144805875e-05,
+      "loss": 0.4253,
+      "step": 1480
+    },
+    {
+      "epoch": 2.094833687190375,
+      "grad_norm": 0.22746696959851062,
+      "learning_rate": 1.6736621196222456e-05,
+      "loss": 0.4098,
+      "step": 1481
+    },
+    {
+      "epoch": 2.0962491153573954,
+      "grad_norm": 0.223823439118962,
+      "learning_rate": 1.6710388247639036e-05,
+      "loss": 0.3938,
+      "step": 1482
+    },
+    {
+      "epoch": 2.097664543524416,
+      "grad_norm": 0.21937232678259852,
+      "learning_rate": 1.6684155299055616e-05,
+      "loss": 0.3975,
+      "step": 1483
+    },
+    {
+      "epoch": 2.099079971691437,
+      "grad_norm": 0.21873413992633153,
+      "learning_rate": 1.6657922350472193e-05,
+      "loss": 0.3965,
+      "step": 1484
+    },
+    {
+      "epoch": 2.100495399858457,
+      "grad_norm": 0.22559007031341452,
+      "learning_rate": 1.6631689401888773e-05,
+      "loss": 0.3667,
+      "step": 1485
+    },
+    {
+      "epoch": 2.1019108280254777,
+      "grad_norm": 0.18962001559298866,
+      "learning_rate": 1.660545645330535e-05,
+      "loss": 0.2863,
+      "step": 1486
+    },
+    {
+      "epoch": 2.1033262561924984,
+      "grad_norm": 0.2256149343155719,
+      "learning_rate": 1.657922350472193e-05,
+      "loss": 0.3789,
+      "step": 1487
+    },
+    {
+      "epoch": 2.1047416843595186,
+      "grad_norm": 0.24432634254170046,
+      "learning_rate": 1.655299055613851e-05,
+      "loss": 0.4192,
+      "step": 1488
+    },
+    {
+      "epoch": 2.1061571125265393,
+      "grad_norm": 0.20331866520876218,
+      "learning_rate": 1.652675760755509e-05,
+      "loss": 0.3157,
+      "step": 1489
+    },
+    {
+      "epoch": 2.10757254069356,
+      "grad_norm": 0.2218135446867955,
+      "learning_rate": 1.6500524658971668e-05,
+      "loss": 0.4136,
+      "step": 1490
+    },
+    {
+      "epoch": 2.10898796886058,
+      "grad_norm": 0.20788675568869244,
+      "learning_rate": 1.647429171038825e-05,
+      "loss": 0.3828,
+      "step": 1491
+    },
+    {
+      "epoch": 2.110403397027601,
+      "grad_norm": 0.2106124475328733,
+      "learning_rate": 1.644805876180483e-05,
+      "loss": 0.385,
+      "step": 1492
+    },
+    {
+      "epoch": 2.1118188251946215,
+      "grad_norm": 0.22319116024778723,
+      "learning_rate": 1.6421825813221406e-05,
+      "loss": 0.3942,
+      "step": 1493
+    },
+    {
+      "epoch": 2.1132342533616417,
+      "grad_norm": 0.2045669215686749,
+      "learning_rate": 1.6395592864637986e-05,
+      "loss": 0.3559,
+      "step": 1494
+    },
+    {
+      "epoch": 2.1146496815286624,
+      "grad_norm": 0.20167320873533356,
+      "learning_rate": 1.6369359916054566e-05,
+      "loss": 0.3492,
+      "step": 1495
+    },
+    {
+      "epoch": 2.116065109695683,
+      "grad_norm": 0.563539207714512,
+      "learning_rate": 1.6343126967471146e-05,
+      "loss": 0.3628,
+      "step": 1496
+    },
+    {
+      "epoch": 2.1174805378627033,
+      "grad_norm": 0.21048626662633552,
+      "learning_rate": 1.6316894018887723e-05,
+      "loss": 0.3927,
+      "step": 1497
+    },
+    {
+      "epoch": 2.118895966029724,
+      "grad_norm": 0.21525926446490412,
+      "learning_rate": 1.6290661070304304e-05,
+      "loss": 0.3693,
+      "step": 1498
+    },
+    {
+      "epoch": 2.1203113941967446,
+      "grad_norm": 0.2278308053047193,
+      "learning_rate": 1.626442812172088e-05,
+      "loss": 0.4186,
+      "step": 1499
+    },
+    {
+      "epoch": 2.121726822363765,
+      "grad_norm": 0.20026421206561354,
+      "learning_rate": 1.623819517313746e-05,
+      "loss": 0.3624,
+      "step": 1500
+    },
+    {
+      "epoch": 2.1231422505307855,
+      "grad_norm": 0.21648706832363435,
+      "learning_rate": 1.621196222455404e-05,
+      "loss": 0.3781,
+      "step": 1501
+    },
+    {
+      "epoch": 2.124557678697806,
+      "grad_norm": 0.19515781163709509,
+      "learning_rate": 1.618572927597062e-05,
+      "loss": 0.336,
+      "step": 1502
+    },
+    {
+      "epoch": 2.1259731068648264,
+      "grad_norm": 0.24081153314019926,
+      "learning_rate": 1.6159496327387198e-05,
+      "loss": 0.4276,
+      "step": 1503
+    },
+    {
+      "epoch": 2.127388535031847,
+      "grad_norm": 0.2156910140075714,
+      "learning_rate": 1.613326337880378e-05,
+      "loss": 0.3482,
+      "step": 1504
+    },
+    {
+      "epoch": 2.1288039631988678,
+      "grad_norm": 0.20631513186015674,
+      "learning_rate": 1.610703043022036e-05,
+      "loss": 0.3488,
+      "step": 1505
+    },
+    {
+      "epoch": 2.130219391365888,
+      "grad_norm": 0.2413240933427328,
+      "learning_rate": 1.6080797481636936e-05,
+      "loss": 0.4308,
+      "step": 1506
+    },
+    {
+      "epoch": 2.1316348195329087,
+      "grad_norm": 0.19686510455803716,
+      "learning_rate": 1.6054564533053516e-05,
+      "loss": 0.3236,
+      "step": 1507
+    },
+    {
+      "epoch": 2.1330502476999293,
+      "grad_norm": 0.21937196597825326,
+      "learning_rate": 1.6028331584470096e-05,
+      "loss": 0.3522,
+      "step": 1508
+    },
+    {
+      "epoch": 2.1344656758669496,
+      "grad_norm": 0.2004049675013169,
+      "learning_rate": 1.6002098635886677e-05,
+      "loss": 0.3719,
+      "step": 1509
+    },
+    {
+      "epoch": 2.1358811040339702,
+      "grad_norm": 0.21926244774280124,
+      "learning_rate": 1.5975865687303253e-05,
+      "loss": 0.4035,
+      "step": 1510
+    },
+    {
+      "epoch": 2.137296532200991,
+      "grad_norm": 0.20670203803808587,
+      "learning_rate": 1.5949632738719834e-05,
+      "loss": 0.3807,
+      "step": 1511
+    },
+    {
+      "epoch": 2.138711960368011,
+      "grad_norm": 0.2090052811208187,
+      "learning_rate": 1.592339979013641e-05,
+      "loss": 0.331,
+      "step": 1512
+    },
+    {
+      "epoch": 2.140127388535032,
+      "grad_norm": 0.2218332758157514,
+      "learning_rate": 1.589716684155299e-05,
+      "loss": 0.3351,
+      "step": 1513
+    },
+    {
+      "epoch": 2.1415428167020525,
+      "grad_norm": 0.20286719665033282,
+      "learning_rate": 1.5870933892969568e-05,
+      "loss": 0.3508,
+      "step": 1514
+    },
+    {
+      "epoch": 2.1429582448690727,
+      "grad_norm": 0.2157182790914069,
+      "learning_rate": 1.584470094438615e-05,
+      "loss": 0.4096,
+      "step": 1515
+    },
+    {
+      "epoch": 2.1443736730360934,
+      "grad_norm": 0.23149191498509575,
+      "learning_rate": 1.581846799580273e-05,
+      "loss": 0.3795,
+      "step": 1516
+    },
+    {
+      "epoch": 2.145789101203114,
+      "grad_norm": 0.2179465608537574,
+      "learning_rate": 1.579223504721931e-05,
+      "loss": 0.3652,
+      "step": 1517
+    },
+    {
+      "epoch": 2.1472045293701343,
+      "grad_norm": 0.23025993168424932,
+      "learning_rate": 1.576600209863589e-05,
+      "loss": 0.4007,
+      "step": 1518
+    },
+    {
+      "epoch": 2.148619957537155,
+      "grad_norm": 0.22000849317187712,
+      "learning_rate": 1.5739769150052466e-05,
+      "loss": 0.3652,
+      "step": 1519
+    },
+    {
+      "epoch": 2.1500353857041756,
+      "grad_norm": 0.24530162425275553,
+      "learning_rate": 1.5713536201469046e-05,
+      "loss": 0.3768,
+      "step": 1520
+    },
+    {
+      "epoch": 2.151450813871196,
+      "grad_norm": 0.2085214000510065,
+      "learning_rate": 1.5687303252885623e-05,
+      "loss": 0.3877,
+      "step": 1521
+    },
+    {
+      "epoch": 2.1528662420382165,
+      "grad_norm": 0.23192817616235128,
+      "learning_rate": 1.5661070304302207e-05,
+      "loss": 0.3871,
+      "step": 1522
+    },
+    {
+      "epoch": 2.154281670205237,
+      "grad_norm": 0.26726756540877716,
+      "learning_rate": 1.5634837355718784e-05,
+      "loss": 0.3914,
+      "step": 1523
+    },
+    {
+      "epoch": 2.1556970983722574,
+      "grad_norm": 0.2286927722327317,
+      "learning_rate": 1.5608604407135364e-05,
+      "loss": 0.3612,
+      "step": 1524
+    },
+    {
+      "epoch": 2.157112526539278,
+      "grad_norm": 0.20785526014793734,
+      "learning_rate": 1.558237145855194e-05,
+      "loss": 0.3915,
+      "step": 1525
+    },
+    {
+      "epoch": 2.1585279547062988,
+      "grad_norm": 0.22692806461653298,
+      "learning_rate": 1.555613850996852e-05,
+      "loss": 0.3322,
+      "step": 1526
+    },
+    {
+      "epoch": 2.159943382873319,
+      "grad_norm": 0.21694441189573116,
+      "learning_rate": 1.55299055613851e-05,
+      "loss": 0.3668,
+      "step": 1527
+    },
+    {
+      "epoch": 2.1613588110403397,
+      "grad_norm": 0.2178270572564254,
+      "learning_rate": 1.5503672612801678e-05,
+      "loss": 0.3506,
+      "step": 1528
+    },
+    {
+      "epoch": 2.1627742392073603,
+      "grad_norm": 0.19684157868907207,
+      "learning_rate": 1.547743966421826e-05,
+      "loss": 0.3425,
+      "step": 1529
+    },
+    {
+      "epoch": 2.1641896673743806,
+      "grad_norm": 0.2185293631247771,
+      "learning_rate": 1.545120671563484e-05,
+      "loss": 0.373,
+      "step": 1530
+    },
+    {
+      "epoch": 2.1656050955414012,
+      "grad_norm": 0.24937859931810394,
+      "learning_rate": 1.542497376705142e-05,
+      "loss": 0.4453,
+      "step": 1531
+    },
+    {
+      "epoch": 2.167020523708422,
+      "grad_norm": 0.20250770374257762,
+      "learning_rate": 1.5398740818467996e-05,
+      "loss": 0.3603,
+      "step": 1532
+    },
+    {
+      "epoch": 2.168435951875442,
+      "grad_norm": 0.2229953291778268,
+      "learning_rate": 1.5372507869884576e-05,
+      "loss": 0.3546,
+      "step": 1533
+    },
+    {
+      "epoch": 2.169851380042463,
+      "grad_norm": 0.22147353462647917,
+      "learning_rate": 1.5346274921301153e-05,
+      "loss": 0.4118,
+      "step": 1534
+    },
+    {
+      "epoch": 2.1712668082094835,
+      "grad_norm": 0.21956103582223985,
+      "learning_rate": 1.5320041972717733e-05,
+      "loss": 0.3779,
+      "step": 1535
+    },
+    {
+      "epoch": 2.1726822363765037,
+      "grad_norm": 0.22705905713793162,
+      "learning_rate": 1.5293809024134314e-05,
+      "loss": 0.3873,
+      "step": 1536
+    },
+    {
+      "epoch": 2.1740976645435244,
+      "grad_norm": 0.20964973432427997,
+      "learning_rate": 1.5267576075550894e-05,
+      "loss": 0.3526,
+      "step": 1537
+    },
+    {
+      "epoch": 2.175513092710545,
+      "grad_norm": 0.21249751481689355,
+      "learning_rate": 1.5241343126967473e-05,
+      "loss": 0.3628,
+      "step": 1538
+    },
+    {
+      "epoch": 2.1769285208775653,
+      "grad_norm": 0.23622172484407175,
+      "learning_rate": 1.5215110178384051e-05,
+      "loss": 0.3835,
+      "step": 1539
+    },
+    {
+      "epoch": 2.178343949044586,
+      "grad_norm": 0.23211487603406503,
+      "learning_rate": 1.518887722980063e-05,
+      "loss": 0.3913,
+      "step": 1540
+    },
+    {
+      "epoch": 2.1797593772116066,
+      "grad_norm": 0.20060668542207974,
+      "learning_rate": 1.5162644281217208e-05,
+      "loss": 0.3122,
+      "step": 1541
+    },
+    {
+      "epoch": 2.181174805378627,
+      "grad_norm": 0.22790100973419974,
+      "learning_rate": 1.5136411332633787e-05,
+      "loss": 0.3871,
+      "step": 1542
+    },
+    {
+      "epoch": 2.1825902335456475,
+      "grad_norm": 0.20199828675191117,
+      "learning_rate": 1.5110178384050369e-05,
+      "loss": 0.3644,
+      "step": 1543
+    },
+    {
+      "epoch": 2.184005661712668,
+      "grad_norm": 0.2189262776259267,
+      "learning_rate": 1.5083945435466948e-05,
+      "loss": 0.4043,
+      "step": 1544
+    },
+    {
+      "epoch": 2.1854210898796884,
+      "grad_norm": 0.22346416372410774,
+      "learning_rate": 1.5057712486883526e-05,
+      "loss": 0.3789,
+      "step": 1545
+    },
+    {
+      "epoch": 2.186836518046709,
+      "grad_norm": 0.20526825039948007,
+      "learning_rate": 1.5031479538300106e-05,
+      "loss": 0.3364,
+      "step": 1546
+    },
+    {
+      "epoch": 2.1882519462137298,
+      "grad_norm": 0.2078165417942967,
+      "learning_rate": 1.5005246589716685e-05,
+      "loss": 0.334,
+      "step": 1547
+    },
+    {
+      "epoch": 2.18966737438075,
+      "grad_norm": 0.21094891720826406,
+      "learning_rate": 1.4979013641133264e-05,
+      "loss": 0.373,
+      "step": 1548
+    },
+    {
+      "epoch": 2.1910828025477707,
+      "grad_norm": 0.22362768217164414,
+      "learning_rate": 1.4952780692549842e-05,
+      "loss": 0.4026,
+      "step": 1549
+    },
+    {
+      "epoch": 2.1924982307147913,
+      "grad_norm": 0.203456004686505,
+      "learning_rate": 1.4926547743966424e-05,
+      "loss": 0.3633,
+      "step": 1550
+    },
+    {
+      "epoch": 2.1939136588818116,
+      "grad_norm": 0.23121854991500074,
+      "learning_rate": 1.4900314795383003e-05,
+      "loss": 0.3626,
+      "step": 1551
+    },
+    {
+      "epoch": 2.1953290870488322,
+      "grad_norm": 0.2138959069426217,
+      "learning_rate": 1.4874081846799581e-05,
+      "loss": 0.3576,
+      "step": 1552
+    },
+    {
+      "epoch": 2.196744515215853,
+      "grad_norm": 0.24699037645814892,
+      "learning_rate": 1.484784889821616e-05,
+      "loss": 0.3856,
+      "step": 1553
+    },
+    {
+      "epoch": 2.198159943382873,
+      "grad_norm": 0.19854082616956784,
+      "learning_rate": 1.4821615949632739e-05,
+      "loss": 0.3418,
+      "step": 1554
+    },
+    {
+      "epoch": 2.199575371549894,
+      "grad_norm": 0.21266454735849155,
+      "learning_rate": 1.4795383001049317e-05,
+      "loss": 0.3587,
+      "step": 1555
+    },
+    {
+      "epoch": 2.2009907997169145,
+      "grad_norm": 0.21339582475218097,
+      "learning_rate": 1.4769150052465897e-05,
+      "loss": 0.3761,
+      "step": 1556
+    },
+    {
+      "epoch": 2.2024062278839347,
+      "grad_norm": 0.20939402295306628,
+      "learning_rate": 1.4742917103882478e-05,
+      "loss": 0.3781,
+      "step": 1557
+    },
+    {
+      "epoch": 2.2038216560509554,
+      "grad_norm": 0.2322645157397984,
+      "learning_rate": 1.4716684155299056e-05,
+      "loss": 0.3948,
+      "step": 1558
+    },
+    {
+      "epoch": 2.205237084217976,
+      "grad_norm": 0.22585337716531292,
+      "learning_rate": 1.4690451206715637e-05,
+      "loss": 0.3426,
+      "step": 1559
+    },
+    {
+      "epoch": 2.2066525123849963,
+      "grad_norm": 0.2148503491337666,
+      "learning_rate": 1.4664218258132215e-05,
+      "loss": 0.3525,
+      "step": 1560
+    },
+    {
+      "epoch": 2.208067940552017,
+      "grad_norm": 0.22665694917883897,
+      "learning_rate": 1.4637985309548794e-05,
+      "loss": 0.3798,
+      "step": 1561
+    },
+    {
+      "epoch": 2.2094833687190376,
+      "grad_norm": 0.20861364422501297,
+      "learning_rate": 1.4611752360965372e-05,
+      "loss": 0.3757,
+      "step": 1562
+    },
+    {
+      "epoch": 2.210898796886058,
+      "grad_norm": 0.2068838790900965,
+      "learning_rate": 1.4585519412381951e-05,
+      "loss": 0.3345,
+      "step": 1563
+    },
+    {
+      "epoch": 2.2123142250530785,
+      "grad_norm": 0.21314165921334982,
+      "learning_rate": 1.4559286463798533e-05,
+      "loss": 0.3743,
+      "step": 1564
+    },
+    {
+      "epoch": 2.213729653220099,
+      "grad_norm": 0.20606173894877292,
+      "learning_rate": 1.4533053515215112e-05,
+      "loss": 0.3515,
+      "step": 1565
+    },
+    {
+      "epoch": 2.2151450813871194,
+      "grad_norm": 0.21170676741754832,
+      "learning_rate": 1.450682056663169e-05,
+      "loss": 0.3532,
+      "step": 1566
+    },
+    {
+      "epoch": 2.21656050955414,
+      "grad_norm": 0.21890755771438766,
+      "learning_rate": 1.4480587618048269e-05,
+      "loss": 0.3606,
+      "step": 1567
+    },
+    {
+      "epoch": 2.2179759377211608,
+      "grad_norm": 0.21049879748213451,
+      "learning_rate": 1.4454354669464847e-05,
+      "loss": 0.3722,
+      "step": 1568
+    },
+    {
+      "epoch": 2.219391365888181,
+      "grad_norm": 0.19689279358689524,
+      "learning_rate": 1.4428121720881428e-05,
+      "loss": 0.3332,
+      "step": 1569
+    },
+    {
+      "epoch": 2.2208067940552016,
+      "grad_norm": 0.221244953805384,
+      "learning_rate": 1.4401888772298008e-05,
+      "loss": 0.3691,
+      "step": 1570
+    },
+    {
+      "epoch": 2.2222222222222223,
+      "grad_norm": 0.21019982504697307,
+      "learning_rate": 1.4375655823714586e-05,
+      "loss": 0.3514,
+      "step": 1571
+    },
+    {
+      "epoch": 2.2236376503892425,
+      "grad_norm": 0.22327263464343836,
+      "learning_rate": 1.4349422875131167e-05,
+      "loss": 0.3736,
+      "step": 1572
+    },
+    {
+      "epoch": 2.225053078556263,
+      "grad_norm": 0.2337778954093035,
+      "learning_rate": 1.4323189926547745e-05,
+      "loss": 0.3802,
+      "step": 1573
+    },
+    {
+      "epoch": 2.226468506723284,
+      "grad_norm": 0.22491847887006883,
+      "learning_rate": 1.4296956977964324e-05,
+      "loss": 0.3761,
+      "step": 1574
+    },
+    {
+      "epoch": 2.227883934890304,
+      "grad_norm": 0.2055020512327927,
+      "learning_rate": 1.4270724029380902e-05,
+      "loss": 0.3534,
+      "step": 1575
+    },
+    {
+      "epoch": 2.229299363057325,
+      "grad_norm": 0.22421411423810259,
+      "learning_rate": 1.4244491080797481e-05,
+      "loss": 0.3637,
+      "step": 1576
+    },
+    {
+      "epoch": 2.2307147912243455,
+      "grad_norm": 0.19551123110438778,
+      "learning_rate": 1.4218258132214063e-05,
+      "loss": 0.3291,
+      "step": 1577
+    },
+    {
+      "epoch": 2.2321302193913657,
+      "grad_norm": 0.2290439051675896,
+      "learning_rate": 1.4192025183630642e-05,
+      "loss": 0.3396,
+      "step": 1578
+    },
+    {
+      "epoch": 2.2335456475583864,
+      "grad_norm": 0.20583543421027845,
+      "learning_rate": 1.416579223504722e-05,
+      "loss": 0.3544,
+      "step": 1579
+    },
+    {
+      "epoch": 2.234961075725407,
+      "grad_norm": 0.2342349710976409,
+      "learning_rate": 1.4139559286463799e-05,
+      "loss": 0.4077,
+      "step": 1580
+    },
+    {
+      "epoch": 2.2363765038924273,
+      "grad_norm": 0.2198729597030276,
+      "learning_rate": 1.4113326337880377e-05,
+      "loss": 0.3705,
+      "step": 1581
+    },
+    {
+      "epoch": 2.237791932059448,
+      "grad_norm": 0.22235081348422966,
+      "learning_rate": 1.4087093389296958e-05,
+      "loss": 0.3551,
+      "step": 1582
+    },
+    {
+      "epoch": 2.2392073602264686,
+      "grad_norm": 0.22345356674467037,
+      "learning_rate": 1.4060860440713536e-05,
+      "loss": 0.3462,
+      "step": 1583
+    },
+    {
+      "epoch": 2.240622788393489,
+      "grad_norm": 0.20517643442250938,
+      "learning_rate": 1.4034627492130117e-05,
+      "loss": 0.3265,
+      "step": 1584
+    },
+    {
+      "epoch": 2.2420382165605095,
+      "grad_norm": 0.21959226573979157,
+      "learning_rate": 1.4008394543546697e-05,
+      "loss": 0.3859,
+      "step": 1585
+    },
+    {
+      "epoch": 2.24345364472753,
+      "grad_norm": 0.24355893909564938,
+      "learning_rate": 1.3982161594963275e-05,
+      "loss": 0.4022,
+      "step": 1586
+    },
+    {
+      "epoch": 2.2448690728945504,
+      "grad_norm": 0.23377288359241896,
+      "learning_rate": 1.3955928646379854e-05,
+      "loss": 0.4207,
+      "step": 1587
+    },
+    {
+      "epoch": 2.246284501061571,
+      "grad_norm": 0.2199085047090607,
+      "learning_rate": 1.3929695697796433e-05,
+      "loss": 0.379,
+      "step": 1588
+    },
+    {
+      "epoch": 2.2476999292285917,
+      "grad_norm": 0.22062658898222495,
+      "learning_rate": 1.3903462749213011e-05,
+      "loss": 0.36,
+      "step": 1589
+    },
+    {
+      "epoch": 2.249115357395612,
+      "grad_norm": 0.2491554525684039,
+      "learning_rate": 1.387722980062959e-05,
+      "loss": 0.4043,
+      "step": 1590
+    },
+    {
+      "epoch": 2.2505307855626326,
+      "grad_norm": 0.2213422309822871,
+      "learning_rate": 1.3850996852046172e-05,
+      "loss": 0.3459,
+      "step": 1591
+    },
+    {
+      "epoch": 2.2519462137296533,
+      "grad_norm": 0.20169981779020701,
+      "learning_rate": 1.382476390346275e-05,
+      "loss": 0.3562,
+      "step": 1592
+    },
+    {
+      "epoch": 2.2533616418966735,
+      "grad_norm": 0.21076797503294742,
+      "learning_rate": 1.3798530954879329e-05,
+      "loss": 0.3392,
+      "step": 1593
+    },
+    {
+      "epoch": 2.254777070063694,
+      "grad_norm": 0.21221372296515487,
+      "learning_rate": 1.3772298006295908e-05,
+      "loss": 0.3353,
+      "step": 1594
+    },
+    {
+      "epoch": 2.256192498230715,
+      "grad_norm": 0.2413103693540377,
+      "learning_rate": 1.3746065057712488e-05,
+      "loss": 0.3827,
+      "step": 1595
+    },
+    {
+      "epoch": 2.2576079263977356,
+      "grad_norm": 0.20595506827868224,
+      "learning_rate": 1.3719832109129066e-05,
+      "loss": 0.3802,
+      "step": 1596
+    },
+    {
+      "epoch": 2.259023354564756,
+      "grad_norm": 0.21085923154346145,
+      "learning_rate": 1.3693599160545645e-05,
+      "loss": 0.4063,
+      "step": 1597
+    },
+    {
+      "epoch": 2.2604387827317765,
+      "grad_norm": 0.22335746982937194,
+      "learning_rate": 1.3667366211962227e-05,
+      "loss": 0.3637,
+      "step": 1598
+    },
+    {
+      "epoch": 2.2618542108987967,
+      "grad_norm": 0.22208789549987143,
+      "learning_rate": 1.3641133263378806e-05,
+      "loss": 0.3656,
+      "step": 1599
+    },
+    {
+      "epoch": 2.2632696390658174,
+      "grad_norm": 0.21297418200769241,
+      "learning_rate": 1.3614900314795384e-05,
+      "loss": 0.3791,
+      "step": 1600
+    },
+    {
+      "epoch": 2.264685067232838,
+      "grad_norm": 0.19540588053519106,
+      "learning_rate": 1.3588667366211963e-05,
+      "loss": 0.3324,
+      "step": 1601
+    },
+    {
+      "epoch": 2.2661004953998587,
+      "grad_norm": 0.20759627984515497,
+      "learning_rate": 1.3562434417628541e-05,
+      "loss": 0.347,
+      "step": 1602
+    },
+    {
+      "epoch": 2.267515923566879,
+      "grad_norm": 0.2245939350902509,
+      "learning_rate": 1.353620146904512e-05,
+      "loss": 0.3697,
+      "step": 1603
+    },
+    {
+      "epoch": 2.2689313517338996,
+      "grad_norm": 0.22209454778408863,
+      "learning_rate": 1.3509968520461699e-05,
+      "loss": 0.356,
+      "step": 1604
+    },
+    {
+      "epoch": 2.27034677990092,
+      "grad_norm": 0.2059874222511523,
+      "learning_rate": 1.348373557187828e-05,
+      "loss": 0.3988,
+      "step": 1605
+    },
+    {
+      "epoch": 2.2717622080679405,
+      "grad_norm": 0.199393331852189,
+      "learning_rate": 1.3457502623294859e-05,
+      "loss": 0.3431,
+      "step": 1606
+    },
+    {
+      "epoch": 2.273177636234961,
+      "grad_norm": 0.2262724322623601,
+      "learning_rate": 1.343126967471144e-05,
+      "loss": 0.3459,
+      "step": 1607
+    },
+    {
+      "epoch": 2.274593064401982,
+      "grad_norm": 0.22062582548159224,
+      "learning_rate": 1.3405036726128018e-05,
+      "loss": 0.3832,
+      "step": 1608
+    },
+    {
+      "epoch": 2.276008492569002,
+      "grad_norm": 0.20922151718664564,
+      "learning_rate": 1.3378803777544597e-05,
+      "loss": 0.3678,
+      "step": 1609
+    },
+    {
+      "epoch": 2.2774239207360227,
+      "grad_norm": 0.23346011193736022,
+      "learning_rate": 1.3352570828961175e-05,
+      "loss": 0.4017,
+      "step": 1610
+    },
+    {
+      "epoch": 2.278839348903043,
+      "grad_norm": 0.20660194929082695,
+      "learning_rate": 1.3326337880377754e-05,
+      "loss": 0.3547,
+      "step": 1611
+    },
+    {
+      "epoch": 2.2802547770700636,
+      "grad_norm": 0.19175980626077407,
+      "learning_rate": 1.3300104931794336e-05,
+      "loss": 0.3275,
+      "step": 1612
+    },
+    {
+      "epoch": 2.2816702052370843,
+      "grad_norm": 0.22569077925192324,
+      "learning_rate": 1.3273871983210914e-05,
+      "loss": 0.3949,
+      "step": 1613
+    },
+    {
+      "epoch": 2.283085633404105,
+      "grad_norm": 0.22179564083723302,
+      "learning_rate": 1.3247639034627493e-05,
+      "loss": 0.3783,
+      "step": 1614
+    },
+    {
+      "epoch": 2.284501061571125,
+      "grad_norm": 0.2225183306260963,
+      "learning_rate": 1.3221406086044072e-05,
+      "loss": 0.4094,
+      "step": 1615
+    },
+    {
+      "epoch": 2.285916489738146,
+      "grad_norm": 0.20342682421399838,
+      "learning_rate": 1.319517313746065e-05,
+      "loss": 0.378,
+      "step": 1616
+    },
+    {
+      "epoch": 2.287331917905166,
+      "grad_norm": 0.21405156680424392,
+      "learning_rate": 1.3168940188877229e-05,
+      "loss": 0.334,
+      "step": 1617
+    },
+    {
+      "epoch": 2.2887473460721868,
+      "grad_norm": 0.20941510324984786,
+      "learning_rate": 1.3142707240293809e-05,
+      "loss": 0.3768,
+      "step": 1618
+    },
+    {
+      "epoch": 2.2901627742392074,
+      "grad_norm": 0.19213037411084616,
+      "learning_rate": 1.311647429171039e-05,
+      "loss": 0.337,
+      "step": 1619
+    },
+    {
+      "epoch": 2.291578202406228,
+      "grad_norm": 0.2160944787649923,
+      "learning_rate": 1.309024134312697e-05,
+      "loss": 0.3582,
+      "step": 1620
+    },
+    {
+      "epoch": 2.2929936305732483,
+      "grad_norm": 0.20595871082883357,
+      "learning_rate": 1.3064008394543548e-05,
+      "loss": 0.3367,
+      "step": 1621
+    },
+    {
+      "epoch": 2.294409058740269,
+      "grad_norm": 0.20042645092539838,
+      "learning_rate": 1.3037775445960127e-05,
+      "loss": 0.3576,
+      "step": 1622
+    },
+    {
+      "epoch": 2.2958244869072892,
+      "grad_norm": 0.2036670157662137,
+      "learning_rate": 1.3011542497376705e-05,
+      "loss": 0.3471,
+      "step": 1623
+    },
+    {
+      "epoch": 2.29723991507431,
+      "grad_norm": 0.19284023614390752,
+      "learning_rate": 1.2985309548793284e-05,
+      "loss": 0.2943,
+      "step": 1624
+    },
+    {
+      "epoch": 2.2986553432413306,
+      "grad_norm": 0.20293208468285898,
+      "learning_rate": 1.2959076600209863e-05,
+      "loss": 0.368,
+      "step": 1625
+    },
+    {
+      "epoch": 2.3000707714083513,
+      "grad_norm": 0.3737401960483002,
+      "learning_rate": 1.2932843651626445e-05,
+      "loss": 0.4036,
+      "step": 1626
+    },
+    {
+      "epoch": 2.3014861995753715,
+      "grad_norm": 0.19354635460810501,
+      "learning_rate": 1.2906610703043023e-05,
+      "loss": 0.355,
+      "step": 1627
+    },
+    {
+      "epoch": 2.302901627742392,
+      "grad_norm": 0.2094545599338365,
+      "learning_rate": 1.2880377754459602e-05,
+      "loss": 0.3914,
+      "step": 1628
+    },
+    {
+      "epoch": 2.3043170559094124,
+      "grad_norm": 2.1242623654331476,
+      "learning_rate": 1.285414480587618e-05,
+      "loss": 0.4319,
+      "step": 1629
+    },
+    {
+      "epoch": 2.305732484076433,
+      "grad_norm": 0.22344074159287974,
+      "learning_rate": 1.282791185729276e-05,
+      "loss": 0.3775,
+      "step": 1630
+    },
+    {
+      "epoch": 2.3071479122434537,
+      "grad_norm": 0.23710780067237636,
+      "learning_rate": 1.280167890870934e-05,
+      "loss": 0.3786,
+      "step": 1631
+    },
+    {
+      "epoch": 2.3085633404104744,
+      "grad_norm": 0.1940770930013411,
+      "learning_rate": 1.2775445960125918e-05,
+      "loss": 0.3128,
+      "step": 1632
+    },
+    {
+      "epoch": 2.3099787685774946,
+      "grad_norm": 0.21297071997985506,
+      "learning_rate": 1.27492130115425e-05,
+      "loss": 0.3421,
+      "step": 1633
+    },
+    {
+      "epoch": 2.3113941967445153,
+      "grad_norm": 0.20386998779645024,
+      "learning_rate": 1.2722980062959078e-05,
+      "loss": 0.3531,
+      "step": 1634
+    },
+    {
+      "epoch": 2.3128096249115355,
+      "grad_norm": 0.21035453901853524,
+      "learning_rate": 1.2696747114375657e-05,
+      "loss": 0.3247,
+      "step": 1635
+    },
+    {
+      "epoch": 2.314225053078556,
+      "grad_norm": 0.20189874562181195,
+      "learning_rate": 1.2670514165792236e-05,
+      "loss": 0.3615,
+      "step": 1636
+    },
+    {
+      "epoch": 2.315640481245577,
+      "grad_norm": 0.2194447285407467,
+      "learning_rate": 1.2644281217208814e-05,
+      "loss": 0.3011,
+      "step": 1637
+    },
+    {
+      "epoch": 2.3170559094125975,
+      "grad_norm": 0.222431639550439,
+      "learning_rate": 1.2618048268625393e-05,
+      "loss": 0.3835,
+      "step": 1638
+    },
+    {
+      "epoch": 2.3184713375796178,
+      "grad_norm": 0.21738559273360789,
+      "learning_rate": 1.2591815320041971e-05,
+      "loss": 0.3132,
+      "step": 1639
+    },
+    {
+      "epoch": 2.3198867657466384,
+      "grad_norm": 0.20223700517296947,
+      "learning_rate": 1.2565582371458553e-05,
+      "loss": 0.3552,
+      "step": 1640
+    },
+    {
+      "epoch": 2.3213021939136587,
+      "grad_norm": 0.22347595598346545,
+      "learning_rate": 1.2539349422875132e-05,
+      "loss": 0.3623,
+      "step": 1641
+    },
+    {
+      "epoch": 2.3227176220806793,
+      "grad_norm": 0.22595395152838466,
+      "learning_rate": 1.251311647429171e-05,
+      "loss": 0.3811,
+      "step": 1642
+    },
+    {
+      "epoch": 2.3241330502477,
+      "grad_norm": 0.2218415824217748,
+      "learning_rate": 1.248688352570829e-05,
+      "loss": 0.3521,
+      "step": 1643
+    },
+    {
+      "epoch": 2.3255484784147207,
+      "grad_norm": 0.2280833710565805,
+      "learning_rate": 1.246065057712487e-05,
+      "loss": 0.4319,
+      "step": 1644
+    },
+    {
+      "epoch": 2.326963906581741,
+      "grad_norm": 0.23328148577348015,
+      "learning_rate": 1.243441762854145e-05,
+      "loss": 0.3749,
+      "step": 1645
+    },
+    {
+      "epoch": 2.3283793347487616,
+      "grad_norm": 0.22987437540955832,
+      "learning_rate": 1.2408184679958028e-05,
+      "loss": 0.4365,
+      "step": 1646
+    },
+    {
+      "epoch": 2.329794762915782,
+      "grad_norm": 0.20803738520290288,
+      "learning_rate": 1.2381951731374607e-05,
+      "loss": 0.35,
+      "step": 1647
+    },
+    {
+      "epoch": 2.3312101910828025,
+      "grad_norm": 0.21298538936199324,
+      "learning_rate": 1.2355718782791187e-05,
+      "loss": 0.333,
+      "step": 1648
+    },
+    {
+      "epoch": 2.332625619249823,
+      "grad_norm": 0.25212607611317467,
+      "learning_rate": 1.2329485834207766e-05,
+      "loss": 0.3826,
+      "step": 1649
+    },
+    {
+      "epoch": 2.334041047416844,
+      "grad_norm": 0.20052310592857706,
+      "learning_rate": 1.2303252885624344e-05,
+      "loss": 0.3371,
+      "step": 1650
+    },
+    {
+      "epoch": 2.335456475583864,
+      "grad_norm": 0.21673405119635897,
+      "learning_rate": 1.2277019937040923e-05,
+      "loss": 0.3482,
+      "step": 1651
+    },
+    {
+      "epoch": 2.3368719037508847,
+      "grad_norm": 0.2138680795145363,
+      "learning_rate": 1.2250786988457503e-05,
+      "loss": 0.3636,
+      "step": 1652
+    },
+    {
+      "epoch": 2.338287331917905,
+      "grad_norm": 0.20865904901786897,
+      "learning_rate": 1.2224554039874082e-05,
+      "loss": 0.3158,
+      "step": 1653
+    },
+    {
+      "epoch": 2.3397027600849256,
+      "grad_norm": 0.2096807726320946,
+      "learning_rate": 1.2198321091290662e-05,
+      "loss": 0.3882,
+      "step": 1654
+    },
+    {
+      "epoch": 2.3411181882519463,
+      "grad_norm": 0.19808067950244185,
+      "learning_rate": 1.217208814270724e-05,
+      "loss": 0.3268,
+      "step": 1655
+    },
+    {
+      "epoch": 2.342533616418967,
+      "grad_norm": 0.22133221526818114,
+      "learning_rate": 1.2145855194123821e-05,
+      "loss": 0.3975,
+      "step": 1656
+    },
+    {
+      "epoch": 2.343949044585987,
+      "grad_norm": 0.6558735068112217,
+      "learning_rate": 1.21196222455404e-05,
+      "loss": 0.4503,
+      "step": 1657
+    },
+    {
+      "epoch": 2.345364472753008,
+      "grad_norm": 0.19935870158714386,
+      "learning_rate": 1.2093389296956978e-05,
+      "loss": 0.3395,
+      "step": 1658
+    },
+    {
+      "epoch": 2.346779900920028,
+      "grad_norm": 0.2068664621726759,
+      "learning_rate": 1.2067156348373558e-05,
+      "loss": 0.3665,
+      "step": 1659
+    },
+    {
+      "epoch": 2.3481953290870488,
+      "grad_norm": 0.20479958605036505,
+      "learning_rate": 1.2040923399790137e-05,
+      "loss": 0.3681,
+      "step": 1660
+    },
+    {
+      "epoch": 2.3496107572540694,
+      "grad_norm": 0.2065543693099381,
+      "learning_rate": 1.2014690451206716e-05,
+      "loss": 0.3404,
+      "step": 1661
+    },
+    {
+      "epoch": 2.35102618542109,
+      "grad_norm": 0.21210242048432407,
+      "learning_rate": 1.1988457502623296e-05,
+      "loss": 0.3415,
+      "step": 1662
+    },
+    {
+      "epoch": 2.3524416135881103,
+      "grad_norm": 0.7109075104703904,
+      "learning_rate": 1.1962224554039874e-05,
+      "loss": 0.4539,
+      "step": 1663
+    },
+    {
+      "epoch": 2.353857041755131,
+      "grad_norm": 0.22108341730671974,
+      "learning_rate": 1.1935991605456453e-05,
+      "loss": 0.3713,
+      "step": 1664
+    },
+    {
+      "epoch": 2.3552724699221512,
+      "grad_norm": 0.21406850500866562,
+      "learning_rate": 1.1909758656873033e-05,
+      "loss": 0.3303,
+      "step": 1665
+    },
+    {
+      "epoch": 2.356687898089172,
+      "grad_norm": 0.22414262449386538,
+      "learning_rate": 1.1883525708289612e-05,
+      "loss": 0.3564,
+      "step": 1666
+    },
+    {
+      "epoch": 2.3581033262561926,
+      "grad_norm": 0.2211106716490476,
+      "learning_rate": 1.1857292759706192e-05,
+      "loss": 0.3534,
+      "step": 1667
+    },
+    {
+      "epoch": 2.3595187544232132,
+      "grad_norm": 0.2113618894415322,
+      "learning_rate": 1.183105981112277e-05,
+      "loss": 0.3416,
+      "step": 1668
+    },
+    {
+      "epoch": 2.3609341825902335,
+      "grad_norm": 0.21893020046407727,
+      "learning_rate": 1.1804826862539351e-05,
+      "loss": 0.3801,
+      "step": 1669
+    },
+    {
+      "epoch": 2.362349610757254,
+      "grad_norm": 0.23267909424111793,
+      "learning_rate": 1.177859391395593e-05,
+      "loss": 0.3942,
+      "step": 1670
+    },
+    {
+      "epoch": 2.3637650389242744,
+      "grad_norm": 0.22233234372987792,
+      "learning_rate": 1.1752360965372508e-05,
+      "loss": 0.3664,
+      "step": 1671
+    },
+    {
+      "epoch": 2.365180467091295,
+      "grad_norm": 0.20200623342526824,
+      "learning_rate": 1.1726128016789089e-05,
+      "loss": 0.308,
+      "step": 1672
+    },
+    {
+      "epoch": 2.3665958952583157,
+      "grad_norm": 0.21336860651200418,
+      "learning_rate": 1.1699895068205667e-05,
+      "loss": 0.343,
+      "step": 1673
+    },
+    {
+      "epoch": 2.3680113234253364,
+      "grad_norm": 0.22413229666446247,
+      "learning_rate": 1.1673662119622246e-05,
+      "loss": 0.3956,
+      "step": 1674
+    },
+    {
+      "epoch": 2.3694267515923566,
+      "grad_norm": 0.21771474988307613,
+      "learning_rate": 1.1647429171038824e-05,
+      "loss": 0.3652,
+      "step": 1675
+    },
+    {
+      "epoch": 2.3708421797593773,
+      "grad_norm": 0.2141067143591576,
+      "learning_rate": 1.1621196222455405e-05,
+      "loss": 0.3405,
+      "step": 1676
+    },
+    {
+      "epoch": 2.3722576079263975,
+      "grad_norm": 0.23877770632497497,
+      "learning_rate": 1.1594963273871983e-05,
+      "loss": 0.4238,
+      "step": 1677
+    },
+    {
+      "epoch": 2.373673036093418,
+      "grad_norm": 0.22336963831812956,
+      "learning_rate": 1.1568730325288562e-05,
+      "loss": 0.4112,
+      "step": 1678
+    },
+    {
+      "epoch": 2.375088464260439,
+      "grad_norm": 0.20451117319496986,
+      "learning_rate": 1.1542497376705142e-05,
+      "loss": 0.3647,
+      "step": 1679
+    },
+    {
+      "epoch": 2.3765038924274595,
+      "grad_norm": 0.20754787330919905,
+      "learning_rate": 1.1516264428121722e-05,
+      "loss": 0.3514,
+      "step": 1680
+    },
+    {
+      "epoch": 2.3779193205944797,
+      "grad_norm": 0.2161438866209889,
+      "learning_rate": 1.1490031479538301e-05,
+      "loss": 0.3972,
+      "step": 1681
+    },
+    {
+      "epoch": 2.3793347487615004,
+      "grad_norm": 0.18714775741431736,
+      "learning_rate": 1.146379853095488e-05,
+      "loss": 0.3199,
+      "step": 1682
+    },
+    {
+      "epoch": 2.3807501769285206,
+      "grad_norm": 0.1939274793117906,
+      "learning_rate": 1.143756558237146e-05,
+      "loss": 0.3408,
+      "step": 1683
+    },
+    {
+      "epoch": 2.3821656050955413,
+      "grad_norm": 0.2088588496866585,
+      "learning_rate": 1.1411332633788038e-05,
+      "loss": 0.3597,
+      "step": 1684
+    },
+    {
+      "epoch": 2.383581033262562,
+      "grad_norm": 0.19274140483291136,
+      "learning_rate": 1.1385099685204617e-05,
+      "loss": 0.286,
+      "step": 1685
+    },
+    {
+      "epoch": 2.3849964614295827,
+      "grad_norm": 0.19465489196136257,
+      "learning_rate": 1.1358866736621197e-05,
+      "loss": 0.3407,
+      "step": 1686
+    },
+    {
+      "epoch": 2.386411889596603,
+      "grad_norm": 0.19358748526673272,
+      "learning_rate": 1.1332633788037776e-05,
+      "loss": 0.3281,
+      "step": 1687
+    },
+    {
+      "epoch": 2.3878273177636236,
+      "grad_norm": 0.22265113555366345,
+      "learning_rate": 1.1306400839454354e-05,
+      "loss": 0.4056,
+      "step": 1688
+    },
+    {
+      "epoch": 2.389242745930644,
+      "grad_norm": 0.20308464928274864,
+      "learning_rate": 1.1280167890870933e-05,
+      "loss": 0.3679,
+      "step": 1689
+    },
+    {
+      "epoch": 2.3906581740976645,
+      "grad_norm": 0.20427851173344422,
+      "learning_rate": 1.1253934942287513e-05,
+      "loss": 0.3826,
+      "step": 1690
+    },
+    {
+      "epoch": 2.392073602264685,
+      "grad_norm": 0.20932500576312751,
+      "learning_rate": 1.1227701993704094e-05,
+      "loss": 0.3877,
+      "step": 1691
+    },
+    {
+      "epoch": 2.393489030431706,
+      "grad_norm": 0.2100695235002279,
+      "learning_rate": 1.1201469045120672e-05,
+      "loss": 0.3874,
+      "step": 1692
+    },
+    {
+      "epoch": 2.394904458598726,
+      "grad_norm": 0.19773857226605634,
+      "learning_rate": 1.1175236096537252e-05,
+      "loss": 0.3473,
+      "step": 1693
+    },
+    {
+      "epoch": 2.3963198867657467,
+      "grad_norm": 0.21235501984385707,
+      "learning_rate": 1.1149003147953831e-05,
+      "loss": 0.3579,
+      "step": 1694
+    },
+    {
+      "epoch": 2.397735314932767,
+      "grad_norm": 0.2235670395689437,
+      "learning_rate": 1.112277019937041e-05,
+      "loss": 0.3766,
+      "step": 1695
+    },
+    {
+      "epoch": 2.3991507430997876,
+      "grad_norm": 0.21562860359515665,
+      "learning_rate": 1.1096537250786988e-05,
+      "loss": 0.3839,
+      "step": 1696
+    },
+    {
+      "epoch": 2.4005661712668083,
+      "grad_norm": 0.20556522359943483,
+      "learning_rate": 1.1070304302203569e-05,
+      "loss": 0.3474,
+      "step": 1697
+    },
+    {
+      "epoch": 2.401981599433829,
+      "grad_norm": 0.21632102454011515,
+      "learning_rate": 1.1044071353620147e-05,
+      "loss": 0.3759,
+      "step": 1698
+    },
+    {
+      "epoch": 2.403397027600849,
+      "grad_norm": 0.22107312323409514,
+      "learning_rate": 1.1017838405036726e-05,
+      "loss": 0.3968,
+      "step": 1699
+    },
+    {
+      "epoch": 2.40481245576787,
+      "grad_norm": 0.20713933717369834,
+      "learning_rate": 1.0991605456453306e-05,
+      "loss": 0.3699,
+      "step": 1700
+    },
+    {
+      "epoch": 2.40622788393489,
+      "grad_norm": 0.2117918962164715,
+      "learning_rate": 1.0965372507869885e-05,
+      "loss": 0.3564,
+      "step": 1701
+    },
+    {
+      "epoch": 2.4076433121019107,
+      "grad_norm": 0.19884508709349702,
+      "learning_rate": 1.0939139559286463e-05,
+      "loss": 0.3396,
+      "step": 1702
+    },
+    {
+      "epoch": 2.4090587402689314,
+      "grad_norm": 0.20445555808171254,
+      "learning_rate": 1.0912906610703043e-05,
+      "loss": 0.3331,
+      "step": 1703
+    },
+    {
+      "epoch": 2.410474168435952,
+      "grad_norm": 0.20438554619989838,
+      "learning_rate": 1.0886673662119624e-05,
+      "loss": 0.3561,
+      "step": 1704
+    },
+    {
+      "epoch": 2.4118895966029723,
+      "grad_norm": 0.21961427283137555,
+      "learning_rate": 1.0860440713536202e-05,
+      "loss": 0.4196,
+      "step": 1705
+    },
+    {
+      "epoch": 2.413305024769993,
+      "grad_norm": 0.21068372988168393,
+      "learning_rate": 1.0834207764952781e-05,
+      "loss": 0.4156,
+      "step": 1706
+    },
+    {
+      "epoch": 2.414720452937013,
+      "grad_norm": 0.20171643481038465,
+      "learning_rate": 1.0807974816369361e-05,
+      "loss": 0.3708,
+      "step": 1707
+    },
+    {
+      "epoch": 2.416135881104034,
+      "grad_norm": 0.21885197735233844,
+      "learning_rate": 1.078174186778594e-05,
+      "loss": 0.3993,
+      "step": 1708
+    },
+    {
+      "epoch": 2.4175513092710545,
+      "grad_norm": 0.21675789921133995,
+      "learning_rate": 1.0755508919202518e-05,
+      "loss": 0.4063,
+      "step": 1709
+    },
+    {
+      "epoch": 2.418966737438075,
+      "grad_norm": 3.0522202221772288,
+      "learning_rate": 1.0729275970619099e-05,
+      "loss": 0.4202,
+      "step": 1710
+    },
+    {
+      "epoch": 2.4203821656050954,
+      "grad_norm": 0.2140766759354188,
+      "learning_rate": 1.0703043022035677e-05,
+      "loss": 0.4007,
+      "step": 1711
+    },
+    {
+      "epoch": 2.421797593772116,
+      "grad_norm": 0.2084942514923414,
+      "learning_rate": 1.0676810073452256e-05,
+      "loss": 0.3396,
+      "step": 1712
+    },
+    {
+      "epoch": 2.4232130219391363,
+      "grad_norm": 0.2149211224899716,
+      "learning_rate": 1.0650577124868834e-05,
+      "loss": 0.3422,
+      "step": 1713
+    },
+    {
+      "epoch": 2.424628450106157,
+      "grad_norm": 0.2106202261748985,
+      "learning_rate": 1.0624344176285415e-05,
+      "loss": 0.3994,
+      "step": 1714
+    },
+    {
+      "epoch": 2.4260438782731777,
+      "grad_norm": 0.22730032299583158,
+      "learning_rate": 1.0598111227701995e-05,
+      "loss": 0.3648,
+      "step": 1715
+    },
+    {
+      "epoch": 2.4274593064401984,
+      "grad_norm": 0.22643649889098394,
+      "learning_rate": 1.0571878279118574e-05,
+      "loss": 0.3391,
+      "step": 1716
+    },
+    {
+      "epoch": 2.4288747346072186,
+      "grad_norm": 0.23342791214204056,
+      "learning_rate": 1.0545645330535154e-05,
+      "loss": 0.3687,
+      "step": 1717
+    },
+    {
+      "epoch": 2.4302901627742393,
+      "grad_norm": 0.20689885848622794,
+      "learning_rate": 1.0519412381951733e-05,
+      "loss": 0.3757,
+      "step": 1718
+    },
+    {
+      "epoch": 2.4317055909412595,
+      "grad_norm": 0.210687923926889,
+      "learning_rate": 1.0493179433368311e-05,
+      "loss": 0.3501,
+      "step": 1719
+    },
+    {
+      "epoch": 2.43312101910828,
+      "grad_norm": 0.22986089011271102,
+      "learning_rate": 1.046694648478489e-05,
+      "loss": 0.3836,
+      "step": 1720
+    },
+    {
+      "epoch": 2.434536447275301,
+      "grad_norm": 0.22744764307215082,
+      "learning_rate": 1.044071353620147e-05,
+      "loss": 0.3898,
+      "step": 1721
+    },
+    {
+      "epoch": 2.4359518754423215,
+      "grad_norm": 0.21194994175423806,
+      "learning_rate": 1.0414480587618049e-05,
+      "loss": 0.334,
+      "step": 1722
+    },
+    {
+      "epoch": 2.4373673036093417,
+      "grad_norm": 0.22018978375373083,
+      "learning_rate": 1.0388247639034627e-05,
+      "loss": 0.4329,
+      "step": 1723
+    },
+    {
+      "epoch": 2.4387827317763624,
+      "grad_norm": 0.19353091165168484,
+      "learning_rate": 1.0362014690451207e-05,
+      "loss": 0.3603,
+      "step": 1724
+    },
+    {
+      "epoch": 2.4401981599433826,
+      "grad_norm": 0.21586391650702627,
+      "learning_rate": 1.0335781741867786e-05,
+      "loss": 0.4187,
+      "step": 1725
+    },
+    {
+      "epoch": 2.4416135881104033,
+      "grad_norm": 0.22306618672285558,
+      "learning_rate": 1.0309548793284365e-05,
+      "loss": 0.4009,
+      "step": 1726
+    },
+    {
+      "epoch": 2.443029016277424,
+      "grad_norm": 0.2102750167761404,
+      "learning_rate": 1.0283315844700945e-05,
+      "loss": 0.3828,
+      "step": 1727
+    },
+    {
+      "epoch": 2.4444444444444446,
+      "grad_norm": 0.18881815647673936,
+      "learning_rate": 1.0257082896117525e-05,
+      "loss": 0.3189,
+      "step": 1728
+    },
+    {
+      "epoch": 2.445859872611465,
+      "grad_norm": 0.20862013453932002,
+      "learning_rate": 1.0230849947534104e-05,
+      "loss": 0.3527,
+      "step": 1729
+    },
+    {
+      "epoch": 2.4472753007784855,
+      "grad_norm": 0.21210986825118294,
+      "learning_rate": 1.0204616998950682e-05,
+      "loss": 0.3517,
+      "step": 1730
+    },
+    {
+      "epoch": 2.4486907289455058,
+      "grad_norm": 0.21429771552588056,
+      "learning_rate": 1.0178384050367263e-05,
+      "loss": 0.3887,
+      "step": 1731
+    },
+    {
+      "epoch": 2.4501061571125264,
+      "grad_norm": 0.21001121174052864,
+      "learning_rate": 1.0152151101783841e-05,
+      "loss": 0.3587,
+      "step": 1732
+    },
+    {
+      "epoch": 2.451521585279547,
+      "grad_norm": 0.22002696555298376,
+      "learning_rate": 1.012591815320042e-05,
+      "loss": 0.3975,
+      "step": 1733
+    },
+    {
+      "epoch": 2.452937013446568,
+      "grad_norm": 0.22051929953678207,
+      "learning_rate": 1.0099685204616998e-05,
+      "loss": 0.3839,
+      "step": 1734
+    },
+    {
+      "epoch": 2.454352441613588,
+      "grad_norm": 0.20142600399295918,
+      "learning_rate": 1.0073452256033579e-05,
+      "loss": 0.338,
+      "step": 1735
+    },
+    {
+      "epoch": 2.4557678697806087,
+      "grad_norm": 0.2174560492510006,
+      "learning_rate": 1.0047219307450157e-05,
+      "loss": 0.4005,
+      "step": 1736
+    },
+    {
+      "epoch": 2.457183297947629,
+      "grad_norm": 0.22565669021953805,
+      "learning_rate": 1.0020986358866736e-05,
+      "loss": 0.3642,
+      "step": 1737
+    },
+    {
+      "epoch": 2.4585987261146496,
+      "grad_norm": 0.222555475085274,
+      "learning_rate": 9.994753410283316e-06,
+      "loss": 0.3832,
+      "step": 1738
+    },
+    {
+      "epoch": 2.4600141542816703,
+      "grad_norm": 0.20558462731843344,
+      "learning_rate": 9.968520461699895e-06,
+      "loss": 0.3346,
+      "step": 1739
+    },
+    {
+      "epoch": 2.461429582448691,
+      "grad_norm": 0.2176380490810765,
+      "learning_rate": 9.942287513116475e-06,
+      "loss": 0.4208,
+      "step": 1740
+    },
+    {
+      "epoch": 2.462845010615711,
+      "grad_norm": 0.19524678242958587,
+      "learning_rate": 9.916054564533055e-06,
+      "loss": 0.3165,
+      "step": 1741
+    },
+    {
+      "epoch": 2.464260438782732,
+      "grad_norm": 0.19314190115388655,
+      "learning_rate": 9.889821615949634e-06,
+      "loss": 0.3311,
+      "step": 1742
+    },
+    {
+      "epoch": 2.4656758669497525,
+      "grad_norm": 0.20321068255373417,
+      "learning_rate": 9.863588667366213e-06,
+      "loss": 0.3792,
+      "step": 1743
+    },
+    {
+      "epoch": 2.4670912951167727,
+      "grad_norm": 0.21499480640234953,
+      "learning_rate": 9.837355718782791e-06,
+      "loss": 0.3547,
+      "step": 1744
+    },
+    {
+      "epoch": 2.4685067232837934,
+      "grad_norm": 0.20409540141795263,
+      "learning_rate": 9.811122770199371e-06,
+      "loss": 0.3324,
+      "step": 1745
+    },
+    {
+      "epoch": 2.469922151450814,
+      "grad_norm": 0.2233290959062587,
+      "learning_rate": 9.78488982161595e-06,
+      "loss": 0.3764,
+      "step": 1746
+    },
+    {
+      "epoch": 2.4713375796178343,
+      "grad_norm": 0.20236083142308806,
+      "learning_rate": 9.758656873032529e-06,
+      "loss": 0.3677,
+      "step": 1747
+    },
+    {
+      "epoch": 2.472753007784855,
+      "grad_norm": 0.20234208948922255,
+      "learning_rate": 9.732423924449109e-06,
+      "loss": 0.3755,
+      "step": 1748
+    },
+    {
+      "epoch": 2.4741684359518756,
+      "grad_norm": 0.20256254108287847,
+      "learning_rate": 9.706190975865687e-06,
+      "loss": 0.3473,
+      "step": 1749
+    },
+    {
+      "epoch": 2.475583864118896,
+      "grad_norm": 0.2188760149731594,
+      "learning_rate": 9.679958027282266e-06,
+      "loss": 0.3543,
+      "step": 1750
+    },
+    {
+      "epoch": 2.4769992922859165,
+      "grad_norm": 0.2872207946643241,
+      "learning_rate": 9.653725078698846e-06,
+      "loss": 0.3412,
+      "step": 1751
+    },
+    {
+      "epoch": 2.478414720452937,
+      "grad_norm": 0.19773186928805295,
+      "learning_rate": 9.627492130115427e-06,
+      "loss": 0.3311,
+      "step": 1752
+    },
+    {
+      "epoch": 2.4798301486199574,
+      "grad_norm": 0.2109865330811176,
+      "learning_rate": 9.601259181532005e-06,
+      "loss": 0.364,
+      "step": 1753
+    },
+    {
+      "epoch": 2.481245576786978,
+      "grad_norm": 0.20385787880377218,
+      "learning_rate": 9.575026232948584e-06,
+      "loss": 0.3547,
+      "step": 1754
+    },
+    {
+      "epoch": 2.4826610049539988,
+      "grad_norm": 0.45454058344335446,
+      "learning_rate": 9.548793284365164e-06,
+      "loss": 0.3761,
+      "step": 1755
+    },
+    {
+      "epoch": 2.484076433121019,
+      "grad_norm": 0.20069144673754255,
+      "learning_rate": 9.522560335781743e-06,
+      "loss": 0.3777,
+      "step": 1756
+    },
+    {
+      "epoch": 2.4854918612880397,
+      "grad_norm": 0.20144636318040068,
+      "learning_rate": 9.496327387198321e-06,
+      "loss": 0.362,
+      "step": 1757
+    },
+    {
+      "epoch": 2.4869072894550603,
+      "grad_norm": 0.21818576258038522,
+      "learning_rate": 9.4700944386149e-06,
+      "loss": 0.4137,
+      "step": 1758
+    },
+    {
+      "epoch": 2.4883227176220806,
+      "grad_norm": 0.23390860882528833,
+      "learning_rate": 9.44386149003148e-06,
+      "loss": 0.3677,
+      "step": 1759
+    },
+    {
+      "epoch": 2.4897381457891012,
+      "grad_norm": 0.21303608042166824,
+      "learning_rate": 9.417628541448059e-06,
+      "loss": 0.3805,
+      "step": 1760
+    },
+    {
+      "epoch": 2.491153573956122,
+      "grad_norm": 0.21648637026495848,
+      "learning_rate": 9.391395592864637e-06,
+      "loss": 0.3957,
+      "step": 1761
+    },
+    {
+      "epoch": 2.492569002123142,
+      "grad_norm": 0.21470889424752063,
+      "learning_rate": 9.365162644281218e-06,
+      "loss": 0.3958,
+      "step": 1762
+    },
+    {
+      "epoch": 2.493984430290163,
+      "grad_norm": 0.21551369414955032,
+      "learning_rate": 9.338929695697796e-06,
+      "loss": 0.4112,
+      "step": 1763
+    },
+    {
+      "epoch": 2.4953998584571835,
+      "grad_norm": 0.20447833108692218,
+      "learning_rate": 9.312696747114377e-06,
+      "loss": 0.3472,
+      "step": 1764
+    },
+    {
+      "epoch": 2.4968152866242037,
+      "grad_norm": 0.20586409741257014,
+      "learning_rate": 9.286463798530955e-06,
+      "loss": 0.3504,
+      "step": 1765
+    },
+    {
+      "epoch": 2.4982307147912244,
+      "grad_norm": 0.21323194372082052,
+      "learning_rate": 9.260230849947535e-06,
+      "loss": 0.3675,
+      "step": 1766
+    },
+    {
+      "epoch": 2.499646142958245,
+      "grad_norm": 0.2070312822291339,
+      "learning_rate": 9.233997901364114e-06,
+      "loss": 0.3825,
+      "step": 1767
+    },
+    {
+      "epoch": 2.5010615711252653,
+      "grad_norm": 0.2151895724043526,
+      "learning_rate": 9.207764952780693e-06,
+      "loss": 0.352,
+      "step": 1768
+    },
+    {
+      "epoch": 2.502476999292286,
+      "grad_norm": 0.22400699091609746,
+      "learning_rate": 9.181532004197273e-06,
+      "loss": 0.4242,
+      "step": 1769
+    },
+    {
+      "epoch": 2.5038924274593066,
+      "grad_norm": 0.20545598835648918,
+      "learning_rate": 9.155299055613851e-06,
+      "loss": 0.3579,
+      "step": 1770
+    },
+    {
+      "epoch": 2.505307855626327,
+      "grad_norm": 0.20089201585791153,
+      "learning_rate": 9.12906610703043e-06,
+      "loss": 0.3429,
+      "step": 1771
+    },
+    {
+      "epoch": 2.5067232837933475,
+      "grad_norm": 0.19904600554390434,
+      "learning_rate": 9.10283315844701e-06,
+      "loss": 0.3482,
+      "step": 1772
+    },
+    {
+      "epoch": 2.5081387119603678,
+      "grad_norm": 0.20136909906575673,
+      "learning_rate": 9.076600209863589e-06,
+      "loss": 0.3749,
+      "step": 1773
+    },
+    {
+      "epoch": 2.5095541401273884,
+      "grad_norm": 0.2039054505197896,
+      "learning_rate": 9.050367261280168e-06,
+      "loss": 0.3596,
+      "step": 1774
+    },
+    {
+      "epoch": 2.510969568294409,
+      "grad_norm": 0.20871768692305176,
+      "learning_rate": 9.024134312696748e-06,
+      "loss": 0.3513,
+      "step": 1775
+    },
+    {
+      "epoch": 2.5123849964614298,
+      "grad_norm": 0.1922506056528137,
+      "learning_rate": 8.997901364113328e-06,
+      "loss": 0.3388,
+      "step": 1776
+    },
+    {
+      "epoch": 2.51380042462845,
+      "grad_norm": 4.0876903662085144,
+      "learning_rate": 8.971668415529907e-06,
+      "loss": 0.3932,
+      "step": 1777
+    },
+    {
+      "epoch": 2.5152158527954707,
+      "grad_norm": 0.22063988754657995,
+      "learning_rate": 8.945435466946485e-06,
+      "loss": 0.3819,
+      "step": 1778
+    },
+    {
+      "epoch": 2.516631280962491,
+      "grad_norm": 0.2182736323794303,
+      "learning_rate": 8.919202518363066e-06,
+      "loss": 0.3879,
+      "step": 1779
+    },
+    {
+      "epoch": 2.5180467091295116,
+      "grad_norm": 0.20526743907609737,
+      "learning_rate": 8.892969569779644e-06,
+      "loss": 0.3797,
+      "step": 1780
+    },
+    {
+      "epoch": 2.5194621372965322,
+      "grad_norm": 0.20693527123319747,
+      "learning_rate": 8.866736621196223e-06,
+      "loss": 0.3765,
+      "step": 1781
+    },
+    {
+      "epoch": 2.520877565463553,
+      "grad_norm": 0.21418203449223625,
+      "learning_rate": 8.840503672612801e-06,
+      "loss": 0.3959,
+      "step": 1782
+    },
+    {
+      "epoch": 2.522292993630573,
+      "grad_norm": 0.199865548007296,
+      "learning_rate": 8.814270724029382e-06,
+      "loss": 0.3331,
+      "step": 1783
+    },
+    {
+      "epoch": 2.523708421797594,
+      "grad_norm": 0.19768360191047332,
+      "learning_rate": 8.78803777544596e-06,
+      "loss": 0.3387,
+      "step": 1784
+    },
+    {
+      "epoch": 2.525123849964614,
+      "grad_norm": 0.220405625409773,
+      "learning_rate": 8.761804826862539e-06,
+      "loss": 0.3986,
+      "step": 1785
+    },
+    {
+      "epoch": 2.5265392781316347,
+      "grad_norm": 0.23298794241115453,
+      "learning_rate": 8.735571878279119e-06,
+      "loss": 0.3726,
+      "step": 1786
+    },
+    {
+      "epoch": 2.5279547062986554,
+      "grad_norm": 0.20845940872746585,
+      "learning_rate": 8.709338929695698e-06,
+      "loss": 0.3686,
+      "step": 1787
+    },
+    {
+      "epoch": 2.529370134465676,
+      "grad_norm": 0.20739265555839045,
+      "learning_rate": 8.683105981112278e-06,
+      "loss": 0.341,
+      "step": 1788
+    },
+    {
+      "epoch": 2.5307855626326963,
+      "grad_norm": 0.23387833738932326,
+      "learning_rate": 8.656873032528857e-06,
+      "loss": 0.4018,
+      "step": 1789
+    },
+    {
+      "epoch": 2.532200990799717,
+      "grad_norm": 0.1991531574710069,
+      "learning_rate": 8.630640083945437e-06,
+      "loss": 0.3642,
+      "step": 1790
+    },
+    {
+      "epoch": 2.533616418966737,
+      "grad_norm": 0.21978858324149222,
+      "learning_rate": 8.604407135362015e-06,
+      "loss": 0.3793,
+      "step": 1791
+    },
+    {
+      "epoch": 2.535031847133758,
+      "grad_norm": 0.19771421985469945,
+      "learning_rate": 8.578174186778594e-06,
+      "loss": 0.3492,
+      "step": 1792
+    },
+    {
+      "epoch": 2.5364472753007785,
+      "grad_norm": 0.21186116442046274,
+      "learning_rate": 8.551941238195174e-06,
+      "loss": 0.384,
+      "step": 1793
+    },
+    {
+      "epoch": 2.537862703467799,
+      "grad_norm": 0.2159890538489202,
+      "learning_rate": 8.525708289611753e-06,
+      "loss": 0.4236,
+      "step": 1794
+    },
+    {
+      "epoch": 2.5392781316348194,
+      "grad_norm": 0.2133088128432072,
+      "learning_rate": 8.499475341028331e-06,
+      "loss": 0.3928,
+      "step": 1795
+    },
+    {
+      "epoch": 2.54069355980184,
+      "grad_norm": 0.20400675417562208,
+      "learning_rate": 8.47324239244491e-06,
+      "loss": 0.397,
+      "step": 1796
+    },
+    {
+      "epoch": 2.5421089879688603,
+      "grad_norm": 0.20894650188281783,
+      "learning_rate": 8.44700944386149e-06,
+      "loss": 0.3664,
+      "step": 1797
+    },
+    {
+      "epoch": 2.543524416135881,
+      "grad_norm": 0.20857055765655338,
+      "learning_rate": 8.420776495278069e-06,
+      "loss": 0.3877,
+      "step": 1798
+    },
+    {
+      "epoch": 2.5449398443029017,
+      "grad_norm": 0.23029162763104252,
+      "learning_rate": 8.39454354669465e-06,
+      "loss": 0.4219,
+      "step": 1799
+    },
+    {
+      "epoch": 2.5463552724699223,
+      "grad_norm": 0.21279556644149866,
+      "learning_rate": 8.368310598111228e-06,
+      "loss": 0.3522,
+      "step": 1800
+    },
+    {
+      "epoch": 2.5477707006369426,
+      "grad_norm": 0.19956081610754542,
+      "learning_rate": 8.342077649527808e-06,
+      "loss": 0.3447,
+      "step": 1801
+    },
+    {
+      "epoch": 2.5491861288039632,
+      "grad_norm": 0.21153748180846707,
+      "learning_rate": 8.315844700944387e-06,
+      "loss": 0.3659,
+      "step": 1802
+    },
+    {
+      "epoch": 2.5506015569709835,
+      "grad_norm": 0.2064243717597128,
+      "learning_rate": 8.289611752360965e-06,
+      "loss": 0.3213,
+      "step": 1803
+    },
+    {
+      "epoch": 2.552016985138004,
+      "grad_norm": 0.210979948172296,
+      "learning_rate": 8.263378803777546e-06,
+      "loss": 0.3737,
+      "step": 1804
+    },
+    {
+      "epoch": 2.553432413305025,
+      "grad_norm": 0.20281319829998443,
+      "learning_rate": 8.237145855194124e-06,
+      "loss": 0.3581,
+      "step": 1805
+    },
+    {
+      "epoch": 2.5548478414720455,
+      "grad_norm": 0.21576724260988553,
+      "learning_rate": 8.210912906610703e-06,
+      "loss": 0.3667,
+      "step": 1806
+    },
+    {
+      "epoch": 2.5562632696390657,
+      "grad_norm": 0.19132759016872786,
+      "learning_rate": 8.184679958027283e-06,
+      "loss": 0.3485,
+      "step": 1807
+    },
+    {
+      "epoch": 2.5576786978060864,
+      "grad_norm": 0.1974784858322557,
+      "learning_rate": 8.158447009443862e-06,
+      "loss": 0.3328,
+      "step": 1808
+    },
+    {
+      "epoch": 2.5590941259731066,
+      "grad_norm": 0.20640998486691128,
+      "learning_rate": 8.13221406086044e-06,
+      "loss": 0.3642,
+      "step": 1809
+    },
+    {
+      "epoch": 2.5605095541401273,
+      "grad_norm": 0.2039268310105597,
+      "learning_rate": 8.10598111227702e-06,
+      "loss": 0.3631,
+      "step": 1810
+    },
+    {
+      "epoch": 2.561924982307148,
+      "grad_norm": 0.19215765539590668,
+      "learning_rate": 8.079748163693599e-06,
+      "loss": 0.3382,
+      "step": 1811
+    },
+    {
+      "epoch": 2.5633404104741686,
+      "grad_norm": 0.26656039783281915,
+      "learning_rate": 8.05351521511018e-06,
+      "loss": 0.3783,
+      "step": 1812
+    },
+    {
+      "epoch": 2.564755838641189,
+      "grad_norm": 0.21493592090680055,
+      "learning_rate": 8.027282266526758e-06,
+      "loss": 0.3503,
+      "step": 1813
+    },
+    {
+      "epoch": 2.5661712668082095,
+      "grad_norm": 0.21040119739922322,
+      "learning_rate": 8.001049317943338e-06,
+      "loss": 0.347,
+      "step": 1814
+    },
+    {
+      "epoch": 2.5675866949752297,
+      "grad_norm": 0.21060367789203188,
+      "learning_rate": 7.974816369359917e-06,
+      "loss": 0.4028,
+      "step": 1815
+    },
+    {
+      "epoch": 2.5690021231422504,
+      "grad_norm": 0.20945122187457563,
+      "learning_rate": 7.948583420776495e-06,
+      "loss": 0.3285,
+      "step": 1816
+    },
+    {
+      "epoch": 2.570417551309271,
+      "grad_norm": 0.2087227917081225,
+      "learning_rate": 7.922350472193076e-06,
+      "loss": 0.354,
+      "step": 1817
+    },
+    {
+      "epoch": 2.5718329794762917,
+      "grad_norm": 0.2095089045850525,
+      "learning_rate": 7.896117523609654e-06,
+      "loss": 0.3623,
+      "step": 1818
+    },
+    {
+      "epoch": 2.573248407643312,
+      "grad_norm": 0.20543371792968124,
+      "learning_rate": 7.869884575026233e-06,
+      "loss": 0.3671,
+      "step": 1819
+    },
+    {
+      "epoch": 2.5746638358103326,
+      "grad_norm": 0.19734594910360248,
+      "learning_rate": 7.843651626442811e-06,
+      "loss": 0.3265,
+      "step": 1820
+    },
+    {
+      "epoch": 2.576079263977353,
+      "grad_norm": 0.2074718265878983,
+      "learning_rate": 7.817418677859392e-06,
+      "loss": 0.363,
+      "step": 1821
+    },
+    {
+      "epoch": 2.5774946921443735,
+      "grad_norm": 0.1994386092622095,
+      "learning_rate": 7.79118572927597e-06,
+      "loss": 0.378,
+      "step": 1822
+    },
+    {
+      "epoch": 2.578910120311394,
+      "grad_norm": 0.2108485509228124,
+      "learning_rate": 7.76495278069255e-06,
+      "loss": 0.4183,
+      "step": 1823
+    },
+    {
+      "epoch": 2.580325548478415,
+      "grad_norm": 0.2013484746339703,
+      "learning_rate": 7.73871983210913e-06,
+      "loss": 0.3305,
+      "step": 1824
+    },
+    {
+      "epoch": 2.581740976645435,
+      "grad_norm": 0.2054314513051612,
+      "learning_rate": 7.71248688352571e-06,
+      "loss": 0.3454,
+      "step": 1825
+    },
+    {
+      "epoch": 2.583156404812456,
+      "grad_norm": 0.20955117943330293,
+      "learning_rate": 7.686253934942288e-06,
+      "loss": 0.344,
+      "step": 1826
+    },
+    {
+      "epoch": 2.5845718329794765,
+      "grad_norm": 0.1973210553919038,
+      "learning_rate": 7.660020986358867e-06,
+      "loss": 0.3204,
+      "step": 1827
+    },
+    {
+      "epoch": 2.5859872611464967,
+      "grad_norm": 0.2171924649126436,
+      "learning_rate": 7.633788037775447e-06,
+      "loss": 0.3737,
+      "step": 1828
+    },
+    {
+      "epoch": 2.5874026893135174,
+      "grad_norm": 0.21712977233946879,
+      "learning_rate": 7.607555089192026e-06,
+      "loss": 0.3788,
+      "step": 1829
+    },
+    {
+      "epoch": 2.588818117480538,
+      "grad_norm": 0.2078317502026498,
+      "learning_rate": 7.581322140608604e-06,
+      "loss": 0.4006,
+      "step": 1830
+    },
+    {
+      "epoch": 2.5902335456475583,
+      "grad_norm": 0.19809729361386832,
+      "learning_rate": 7.5550891920251845e-06,
+      "loss": 0.3704,
+      "step": 1831
+    },
+    {
+      "epoch": 2.591648973814579,
+      "grad_norm": 0.1968896281185804,
+      "learning_rate": 7.528856243441763e-06,
+      "loss": 0.3082,
+      "step": 1832
+    },
+    {
+      "epoch": 2.5930644019815996,
+      "grad_norm": 0.20981367025717132,
+      "learning_rate": 7.5026232948583425e-06,
+      "loss": 0.3608,
+      "step": 1833
+    },
+    {
+      "epoch": 2.59447983014862,
+      "grad_norm": 0.20797161093836125,
+      "learning_rate": 7.476390346274921e-06,
+      "loss": 0.3362,
+      "step": 1834
+    },
+    {
+      "epoch": 2.5958952583156405,
+      "grad_norm": 0.21261247348837317,
+      "learning_rate": 7.450157397691501e-06,
+      "loss": 0.3814,
+      "step": 1835
+    },
+    {
+      "epoch": 2.597310686482661,
+      "grad_norm": 0.19882968356020506,
+      "learning_rate": 7.42392444910808e-06,
+      "loss": 0.343,
+      "step": 1836
+    },
+    {
+      "epoch": 2.5987261146496814,
+      "grad_norm": 0.2103031596403086,
+      "learning_rate": 7.3976915005246586e-06,
+      "loss": 0.3829,
+      "step": 1837
+    },
+    {
+      "epoch": 2.600141542816702,
+      "grad_norm": 0.19150021569844553,
+      "learning_rate": 7.371458551941239e-06,
+      "loss": 0.3375,
+      "step": 1838
+    },
+    {
+      "epoch": 2.6015569709837227,
+      "grad_norm": 0.2011174569283403,
+      "learning_rate": 7.345225603357818e-06,
+      "loss": 0.3181,
+      "step": 1839
+    },
+    {
+      "epoch": 2.602972399150743,
+      "grad_norm": 0.21977968424911656,
+      "learning_rate": 7.318992654774397e-06,
+      "loss": 0.3879,
+      "step": 1840
+    },
+    {
+      "epoch": 2.6043878273177636,
+      "grad_norm": 0.2020175254696623,
+      "learning_rate": 7.2927597061909755e-06,
+      "loss": 0.3404,
+      "step": 1841
+    },
+    {
+      "epoch": 2.6058032554847843,
+      "grad_norm": 0.19567152319659617,
+      "learning_rate": 7.266526757607556e-06,
+      "loss": 0.3191,
+      "step": 1842
+    },
+    {
+      "epoch": 2.6072186836518045,
+      "grad_norm": 0.18597323556834847,
+      "learning_rate": 7.240293809024134e-06,
+      "loss": 0.2806,
+      "step": 1843
+    },
+    {
+      "epoch": 2.608634111818825,
+      "grad_norm": 0.21321812440621554,
+      "learning_rate": 7.214060860440714e-06,
+      "loss": 0.4131,
+      "step": 1844
+    },
+    {
+      "epoch": 2.610049539985846,
+      "grad_norm": 0.21088310897402524,
+      "learning_rate": 7.187827911857293e-06,
+      "loss": 0.3515,
+      "step": 1845
+    },
+    {
+      "epoch": 2.611464968152866,
+      "grad_norm": 0.20528663135935263,
+      "learning_rate": 7.161594963273873e-06,
+      "loss": 0.3613,
+      "step": 1846
+    },
+    {
+      "epoch": 2.6128803963198868,
+      "grad_norm": 0.20929437136431123,
+      "learning_rate": 7.135362014690451e-06,
+      "loss": 0.357,
+      "step": 1847
+    },
+    {
+      "epoch": 2.6142958244869074,
+      "grad_norm": 0.22328273062400936,
+      "learning_rate": 7.1091290661070315e-06,
+      "loss": 0.4107,
+      "step": 1848
+    },
+    {
+      "epoch": 2.6157112526539277,
+      "grad_norm": 0.2037256309767581,
+      "learning_rate": 7.08289611752361e-06,
+      "loss": 0.3576,
+      "step": 1849
+    },
+    {
+      "epoch": 2.6171266808209483,
+      "grad_norm": 0.2243164183301271,
+      "learning_rate": 7.056663168940189e-06,
+      "loss": 0.4082,
+      "step": 1850
+    },
+    {
+      "epoch": 2.618542108987969,
+      "grad_norm": 0.2186639214858174,
+      "learning_rate": 7.030430220356768e-06,
+      "loss": 0.3933,
+      "step": 1851
+    },
+    {
+      "epoch": 2.6199575371549892,
+      "grad_norm": 0.2159361736800281,
+      "learning_rate": 7.0041972717733484e-06,
+      "loss": 0.3633,
+      "step": 1852
+    },
+    {
+      "epoch": 2.62137296532201,
+      "grad_norm": 1.6506203295835573,
+      "learning_rate": 6.977964323189927e-06,
+      "loss": 0.4659,
+      "step": 1853
+    },
+    {
+      "epoch": 2.6227883934890306,
+      "grad_norm": 0.19835782691780177,
+      "learning_rate": 6.951731374606506e-06,
+      "loss": 0.3411,
+      "step": 1854
+    },
+    {
+      "epoch": 2.624203821656051,
+      "grad_norm": 0.2218844143299372,
+      "learning_rate": 6.925498426023086e-06,
+      "loss": 0.415,
+      "step": 1855
+    },
+    {
+      "epoch": 2.6256192498230715,
+      "grad_norm": 0.21714669812635473,
+      "learning_rate": 6.8992654774396645e-06,
+      "loss": 0.3731,
+      "step": 1856
+    },
+    {
+      "epoch": 2.627034677990092,
+      "grad_norm": 0.22171926703943656,
+      "learning_rate": 6.873032528856244e-06,
+      "loss": 0.373,
+      "step": 1857
+    },
+    {
+      "epoch": 2.6284501061571124,
+      "grad_norm": 0.2099085807003911,
+      "learning_rate": 6.8467995802728225e-06,
+      "loss": 0.3741,
+      "step": 1858
+    },
+    {
+      "epoch": 2.629865534324133,
+      "grad_norm": 0.19496236614733134,
+      "learning_rate": 6.820566631689403e-06,
+      "loss": 0.3509,
+      "step": 1859
+    },
+    {
+      "epoch": 2.6312809624911537,
+      "grad_norm": 0.1939749683125769,
+      "learning_rate": 6.794333683105981e-06,
+      "loss": 0.3255,
+      "step": 1860
+    },
+    {
+      "epoch": 2.632696390658174,
+      "grad_norm": 0.21205398195214104,
+      "learning_rate": 6.76810073452256e-06,
+      "loss": 0.3659,
+      "step": 1861
+    },
+    {
+      "epoch": 2.6341118188251946,
+      "grad_norm": 0.2032255971712124,
+      "learning_rate": 6.74186778593914e-06,
+      "loss": 0.3816,
+      "step": 1862
+    },
+    {
+      "epoch": 2.6355272469922153,
+      "grad_norm": 0.2101835507755894,
+      "learning_rate": 6.71563483735572e-06,
+      "loss": 0.3715,
+      "step": 1863
+    },
+    {
+      "epoch": 2.6369426751592355,
+      "grad_norm": 0.20345481767709614,
+      "learning_rate": 6.689401888772298e-06,
+      "loss": 0.3676,
+      "step": 1864
+    },
+    {
+      "epoch": 2.638358103326256,
+      "grad_norm": 0.20906313390295878,
+      "learning_rate": 6.663168940188877e-06,
+      "loss": 0.3802,
+      "step": 1865
+    },
+    {
+      "epoch": 2.639773531493277,
+      "grad_norm": 0.18629132905947415,
+      "learning_rate": 6.636935991605457e-06,
+      "loss": 0.3209,
+      "step": 1866
+    },
+    {
+      "epoch": 2.641188959660297,
+      "grad_norm": 0.2152572803841356,
+      "learning_rate": 6.610703043022036e-06,
+      "loss": 0.3399,
+      "step": 1867
+    },
+    {
+      "epoch": 2.6426043878273178,
+      "grad_norm": 0.19539056047254483,
+      "learning_rate": 6.584470094438614e-06,
+      "loss": 0.3122,
+      "step": 1868
+    },
+    {
+      "epoch": 2.6440198159943384,
+      "grad_norm": 0.20202618202655612,
+      "learning_rate": 6.558237145855195e-06,
+      "loss": 0.3608,
+      "step": 1869
+    },
+    {
+      "epoch": 2.6454352441613587,
+      "grad_norm": 0.21368434378231554,
+      "learning_rate": 6.532004197271774e-06,
+      "loss": 0.3408,
+      "step": 1870
+    },
+    {
+      "epoch": 2.6468506723283793,
+      "grad_norm": 0.2031122859886185,
+      "learning_rate": 6.505771248688353e-06,
+      "loss": 0.3833,
+      "step": 1871
+    },
+    {
+      "epoch": 2.6482661004954,
+      "grad_norm": 0.20061405029326895,
+      "learning_rate": 6.479538300104931e-06,
+      "loss": 0.3759,
+      "step": 1872
+    },
+    {
+      "epoch": 2.6496815286624202,
+      "grad_norm": 0.20098303425728567,
+      "learning_rate": 6.4533053515215116e-06,
+      "loss": 0.3467,
+      "step": 1873
+    },
+    {
+      "epoch": 2.651096956829441,
+      "grad_norm": 0.20278629578617524,
+      "learning_rate": 6.42707240293809e-06,
+      "loss": 0.3171,
+      "step": 1874
+    },
+    {
+      "epoch": 2.6525123849964616,
+      "grad_norm": 0.20242009886724874,
+      "learning_rate": 6.40083945435467e-06,
+      "loss": 0.366,
+      "step": 1875
+    },
+    {
+      "epoch": 2.653927813163482,
+      "grad_norm": 0.20238146901560336,
+      "learning_rate": 6.37460650577125e-06,
+      "loss": 0.3786,
+      "step": 1876
+    },
+    {
+      "epoch": 2.6553432413305025,
+      "grad_norm": 0.18806823919909846,
+      "learning_rate": 6.3483735571878285e-06,
+      "loss": 0.3383,
+      "step": 1877
+    },
+    {
+      "epoch": 2.656758669497523,
+      "grad_norm": 0.20708956711483964,
+      "learning_rate": 6.322140608604407e-06,
+      "loss": 0.391,
+      "step": 1878
+    },
+    {
+      "epoch": 2.6581740976645434,
+      "grad_norm": 0.20770833502603264,
+      "learning_rate": 6.295907660020986e-06,
+      "loss": 0.3146,
+      "step": 1879
+    },
+    {
+      "epoch": 2.659589525831564,
+      "grad_norm": 0.21822289469082637,
+      "learning_rate": 6.269674711437566e-06,
+      "loss": 0.3934,
+      "step": 1880
+    },
+    {
+      "epoch": 2.6610049539985847,
+      "grad_norm": 0.2189330109283495,
+      "learning_rate": 6.243441762854145e-06,
+      "loss": 0.3635,
+      "step": 1881
+    },
+    {
+      "epoch": 2.662420382165605,
+      "grad_norm": 0.2025224096693157,
+      "learning_rate": 6.217208814270725e-06,
+      "loss": 0.3622,
+      "step": 1882
+    },
+    {
+      "epoch": 2.6638358103326256,
+      "grad_norm": 0.20391333536797676,
+      "learning_rate": 6.190975865687303e-06,
+      "loss": 0.3605,
+      "step": 1883
+    },
+    {
+      "epoch": 2.6652512384996463,
+      "grad_norm": 0.22300673855216288,
+      "learning_rate": 6.164742917103883e-06,
+      "loss": 0.4051,
+      "step": 1884
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.1949488754489201,
+      "learning_rate": 6.1385099685204614e-06,
+      "loss": 0.3318,
+      "step": 1885
+    },
+    {
+      "epoch": 2.668082094833687,
+      "grad_norm": 0.20257754211433252,
+      "learning_rate": 6.112277019937041e-06,
+      "loss": 0.3481,
+      "step": 1886
+    },
+    {
+      "epoch": 2.669497523000708,
+      "grad_norm": 0.21284586193369576,
+      "learning_rate": 6.08604407135362e-06,
+      "loss": 0.3693,
+      "step": 1887
+    },
+    {
+      "epoch": 2.670912951167728,
+      "grad_norm": 0.20975573812946513,
+      "learning_rate": 6.0598111227702e-06,
+      "loss": 0.3945,
+      "step": 1888
+    },
+    {
+      "epoch": 2.6723283793347488,
+      "grad_norm": 0.1962050177392725,
+      "learning_rate": 6.033578174186779e-06,
+      "loss": 0.3527,
+      "step": 1889
+    },
+    {
+      "epoch": 2.6737438075017694,
+      "grad_norm": 0.20700678452506255,
+      "learning_rate": 6.007345225603358e-06,
+      "loss": 0.3703,
+      "step": 1890
+    },
+    {
+      "epoch": 2.6751592356687897,
+      "grad_norm": 0.21574145342448686,
+      "learning_rate": 5.981112277019937e-06,
+      "loss": 0.3899,
+      "step": 1891
+    },
+    {
+      "epoch": 2.6765746638358103,
+      "grad_norm": 0.21797501994034846,
+      "learning_rate": 5.954879328436517e-06,
+      "loss": 0.3867,
+      "step": 1892
+    },
+    {
+      "epoch": 2.677990092002831,
+      "grad_norm": 0.20848947685547806,
+      "learning_rate": 5.928646379853096e-06,
+      "loss": 0.388,
+      "step": 1893
+    },
+    {
+      "epoch": 2.6794055201698512,
+      "grad_norm": 0.19866839410791579,
+      "learning_rate": 5.9024134312696755e-06,
+      "loss": 0.3655,
+      "step": 1894
+    },
+    {
+      "epoch": 2.680820948336872,
+      "grad_norm": 0.1981411367729309,
+      "learning_rate": 5.876180482686254e-06,
+      "loss": 0.337,
+      "step": 1895
+    },
+    {
+      "epoch": 2.6822363765038926,
+      "grad_norm": 0.195120217473148,
+      "learning_rate": 5.8499475341028336e-06,
+      "loss": 0.3337,
+      "step": 1896
+    },
+    {
+      "epoch": 2.683651804670913,
+      "grad_norm": 0.19846949669318745,
+      "learning_rate": 5.823714585519412e-06,
+      "loss": 0.339,
+      "step": 1897
+    },
+    {
+      "epoch": 2.6850672328379335,
+      "grad_norm": 0.5980099051970161,
+      "learning_rate": 5.797481636935992e-06,
+      "loss": 0.3733,
+      "step": 1898
+    },
+    {
+      "epoch": 2.686482661004954,
+      "grad_norm": 0.21711699289634304,
+      "learning_rate": 5.771248688352571e-06,
+      "loss": 0.383,
+      "step": 1899
+    },
+    {
+      "epoch": 2.6878980891719744,
+      "grad_norm": 0.2093369096634055,
+      "learning_rate": 5.7450157397691505e-06,
+      "loss": 0.363,
+      "step": 1900
+    },
+    {
+      "epoch": 2.689313517338995,
+      "grad_norm": 0.20338112097088418,
+      "learning_rate": 5.71878279118573e-06,
+      "loss": 0.3513,
+      "step": 1901
+    },
+    {
+      "epoch": 2.6907289455060157,
+      "grad_norm": 0.19880924179095463,
+      "learning_rate": 5.6925498426023085e-06,
+      "loss": 0.3974,
+      "step": 1902
+    },
+    {
+      "epoch": 2.692144373673036,
+      "grad_norm": 0.21822057055933752,
+      "learning_rate": 5.666316894018888e-06,
+      "loss": 0.4239,
+      "step": 1903
+    },
+    {
+      "epoch": 2.6935598018400566,
+      "grad_norm": 0.2076889320094873,
+      "learning_rate": 5.6400839454354665e-06,
+      "loss": 0.3578,
+      "step": 1904
+    },
+    {
+      "epoch": 2.6949752300070773,
+      "grad_norm": 0.21095552299443449,
+      "learning_rate": 5.613850996852047e-06,
+      "loss": 0.3779,
+      "step": 1905
+    },
+    {
+      "epoch": 2.6963906581740975,
+      "grad_norm": 0.20335690282279423,
+      "learning_rate": 5.587618048268626e-06,
+      "loss": 0.3592,
+      "step": 1906
+    },
+    {
+      "epoch": 2.697806086341118,
+      "grad_norm": 0.2104837140088037,
+      "learning_rate": 5.561385099685205e-06,
+      "loss": 0.3676,
+      "step": 1907
+    },
+    {
+      "epoch": 2.699221514508139,
+      "grad_norm": 0.21468127395284245,
+      "learning_rate": 5.535152151101784e-06,
+      "loss": 0.3606,
+      "step": 1908
+    },
+    {
+      "epoch": 2.700636942675159,
+      "grad_norm": 0.2046692810039167,
+      "learning_rate": 5.508919202518363e-06,
+      "loss": 0.3678,
+      "step": 1909
+    },
+    {
+      "epoch": 2.7020523708421798,
+      "grad_norm": 0.21112991054823232,
+      "learning_rate": 5.482686253934942e-06,
+      "loss": 0.372,
+      "step": 1910
+    },
+    {
+      "epoch": 2.7034677990092004,
+      "grad_norm": 0.2141438116917265,
+      "learning_rate": 5.456453305351522e-06,
+      "loss": 0.3985,
+      "step": 1911
+    },
+    {
+      "epoch": 2.7048832271762207,
+      "grad_norm": 0.21033732243828274,
+      "learning_rate": 5.430220356768101e-06,
+      "loss": 0.4051,
+      "step": 1912
+    },
+    {
+      "epoch": 2.7062986553432413,
+      "grad_norm": 0.2111440991942263,
+      "learning_rate": 5.403987408184681e-06,
+      "loss": 0.4008,
+      "step": 1913
+    },
+    {
+      "epoch": 2.707714083510262,
+      "grad_norm": 0.20268005737827657,
+      "learning_rate": 5.377754459601259e-06,
+      "loss": 0.3695,
+      "step": 1914
+    },
+    {
+      "epoch": 2.709129511677282,
+      "grad_norm": 0.1951999738009383,
+      "learning_rate": 5.351521511017839e-06,
+      "loss": 0.3442,
+      "step": 1915
+    },
+    {
+      "epoch": 2.710544939844303,
+      "grad_norm": 0.2001087039685667,
+      "learning_rate": 5.325288562434417e-06,
+      "loss": 0.3741,
+      "step": 1916
+    },
+    {
+      "epoch": 2.7119603680113236,
+      "grad_norm": 0.19783070424964766,
+      "learning_rate": 5.2990556138509975e-06,
+      "loss": 0.3622,
+      "step": 1917
+    },
+    {
+      "epoch": 2.713375796178344,
+      "grad_norm": 0.1935708112190888,
+      "learning_rate": 5.272822665267577e-06,
+      "loss": 0.3305,
+      "step": 1918
+    },
+    {
+      "epoch": 2.7147912243453645,
+      "grad_norm": 0.19804726726173558,
+      "learning_rate": 5.2465897166841556e-06,
+      "loss": 0.343,
+      "step": 1919
+    },
+    {
+      "epoch": 2.716206652512385,
+      "grad_norm": 0.2269934633201632,
+      "learning_rate": 5.220356768100735e-06,
+      "loss": 0.3996,
+      "step": 1920
+    },
+    {
+      "epoch": 2.7176220806794054,
+      "grad_norm": 0.20480407012013166,
+      "learning_rate": 5.194123819517314e-06,
+      "loss": 0.3914,
+      "step": 1921
+    },
+    {
+      "epoch": 2.719037508846426,
+      "grad_norm": 0.19401177570868808,
+      "learning_rate": 5.167890870933893e-06,
+      "loss": 0.3278,
+      "step": 1922
+    },
+    {
+      "epoch": 2.7204529370134467,
+      "grad_norm": 0.20895425205289753,
+      "learning_rate": 5.1416579223504725e-06,
+      "loss": 0.3424,
+      "step": 1923
+    },
+    {
+      "epoch": 2.721868365180467,
+      "grad_norm": 0.2438874893174407,
+      "learning_rate": 5.115424973767052e-06,
+      "loss": 0.3961,
+      "step": 1924
+    },
+    {
+      "epoch": 2.7232837933474876,
+      "grad_norm": 0.18755381495409573,
+      "learning_rate": 5.089192025183631e-06,
+      "loss": 0.3203,
+      "step": 1925
+    },
+    {
+      "epoch": 2.7246992215145083,
+      "grad_norm": 0.20504629032498756,
+      "learning_rate": 5.06295907660021e-06,
+      "loss": 0.3787,
+      "step": 1926
+    },
+    {
+      "epoch": 2.7261146496815285,
+      "grad_norm": 0.19925744836851098,
+      "learning_rate": 5.036726128016789e-06,
+      "loss": 0.3593,
+      "step": 1927
+    },
+    {
+      "epoch": 2.727530077848549,
+      "grad_norm": 0.2004589038992432,
+      "learning_rate": 5.010493179433368e-06,
+      "loss": 0.3639,
+      "step": 1928
+    },
+    {
+      "epoch": 2.72894550601557,
+      "grad_norm": 0.20743800881777924,
+      "learning_rate": 4.984260230849947e-06,
+      "loss": 0.3755,
+      "step": 1929
+    },
+    {
+      "epoch": 2.73036093418259,
+      "grad_norm": 0.19692865498896367,
+      "learning_rate": 4.958027282266528e-06,
+      "loss": 0.3399,
+      "step": 1930
+    },
+    {
+      "epoch": 2.7317763623496107,
+      "grad_norm": 0.21841243044598094,
+      "learning_rate": 4.931794333683106e-06,
+      "loss": 0.3858,
+      "step": 1931
+    },
+    {
+      "epoch": 2.7331917905166314,
+      "grad_norm": 0.21136619232889323,
+      "learning_rate": 4.905561385099686e-06,
+      "loss": 0.3921,
+      "step": 1932
+    },
+    {
+      "epoch": 2.7346072186836516,
+      "grad_norm": 0.19484573735389385,
+      "learning_rate": 4.879328436516264e-06,
+      "loss": 0.3706,
+      "step": 1933
+    },
+    {
+      "epoch": 2.7360226468506723,
+      "grad_norm": 0.2076830427543638,
+      "learning_rate": 4.853095487932844e-06,
+      "loss": 0.3456,
+      "step": 1934
+    },
+    {
+      "epoch": 2.737438075017693,
+      "grad_norm": 0.20390841761357054,
+      "learning_rate": 4.826862539349423e-06,
+      "loss": 0.3526,
+      "step": 1935
+    },
+    {
+      "epoch": 2.738853503184713,
+      "grad_norm": 0.21702033210302102,
+      "learning_rate": 4.800629590766003e-06,
+      "loss": 0.4627,
+      "step": 1936
+    },
+    {
+      "epoch": 2.740268931351734,
+      "grad_norm": 0.20061521845716984,
+      "learning_rate": 4.774396642182582e-06,
+      "loss": 0.3781,
+      "step": 1937
+    },
+    {
+      "epoch": 2.7416843595187546,
+      "grad_norm": 0.2560094803022627,
+      "learning_rate": 4.748163693599161e-06,
+      "loss": 0.4285,
+      "step": 1938
+    },
+    {
+      "epoch": 2.743099787685775,
+      "grad_norm": 0.1932520148658916,
+      "learning_rate": 4.72193074501574e-06,
+      "loss": 0.3394,
+      "step": 1939
+    },
+    {
+      "epoch": 2.7445152158527955,
+      "grad_norm": 0.20433241819246098,
+      "learning_rate": 4.695697796432319e-06,
+      "loss": 0.377,
+      "step": 1940
+    },
+    {
+      "epoch": 2.745930644019816,
+      "grad_norm": 0.195922105002889,
+      "learning_rate": 4.669464847848898e-06,
+      "loss": 0.3262,
+      "step": 1941
+    },
+    {
+      "epoch": 2.7473460721868364,
+      "grad_norm": 0.1993898781097721,
+      "learning_rate": 4.6432318992654776e-06,
+      "loss": 0.3687,
+      "step": 1942
+    },
+    {
+      "epoch": 2.748761500353857,
+      "grad_norm": 0.20177198309564906,
+      "learning_rate": 4.616998950682057e-06,
+      "loss": 0.3741,
+      "step": 1943
+    },
+    {
+      "epoch": 2.7501769285208777,
+      "grad_norm": 0.19082999793806443,
+      "learning_rate": 4.5907660020986364e-06,
+      "loss": 0.3534,
+      "step": 1944
+    },
+    {
+      "epoch": 2.7515923566878984,
+      "grad_norm": 0.1956209768009436,
+      "learning_rate": 4.564533053515215e-06,
+      "loss": 0.3439,
+      "step": 1945
+    },
+    {
+      "epoch": 2.7530077848549186,
+      "grad_norm": 0.20345812447505207,
+      "learning_rate": 4.5383001049317945e-06,
+      "loss": 0.4009,
+      "step": 1946
+    },
+    {
+      "epoch": 2.7544232130219393,
+      "grad_norm": 0.1995521377352416,
+      "learning_rate": 4.512067156348374e-06,
+      "loss": 0.3623,
+      "step": 1947
+    },
+    {
+      "epoch": 2.7558386411889595,
+      "grad_norm": 0.2170207918509561,
+      "learning_rate": 4.485834207764953e-06,
+      "loss": 0.3573,
+      "step": 1948
+    },
+    {
+      "epoch": 2.75725406935598,
+      "grad_norm": 0.20261497371921042,
+      "learning_rate": 4.459601259181533e-06,
+      "loss": 0.3715,
+      "step": 1949
+    },
+    {
+      "epoch": 2.758669497523001,
+      "grad_norm": 0.20015382284953034,
+      "learning_rate": 4.433368310598111e-06,
+      "loss": 0.327,
+      "step": 1950
+    },
+    {
+      "epoch": 2.7600849256900215,
+      "grad_norm": 0.19249028445721247,
+      "learning_rate": 4.407135362014691e-06,
+      "loss": 0.3339,
+      "step": 1951
+    },
+    {
+      "epoch": 2.7615003538570417,
+      "grad_norm": 0.19662036433893573,
+      "learning_rate": 4.380902413431269e-06,
+      "loss": 0.3566,
+      "step": 1952
+    },
+    {
+      "epoch": 2.7629157820240624,
+      "grad_norm": 0.2018722434714739,
+      "learning_rate": 4.354669464847849e-06,
+      "loss": 0.3797,
+      "step": 1953
+    },
+    {
+      "epoch": 2.7643312101910826,
+      "grad_norm": 0.19701375920428837,
+      "learning_rate": 4.328436516264428e-06,
+      "loss": 0.3813,
+      "step": 1954
+    },
+    {
+      "epoch": 2.7657466383581033,
+      "grad_norm": 0.18850648320520433,
+      "learning_rate": 4.302203567681008e-06,
+      "loss": 0.3312,
+      "step": 1955
+    },
+    {
+      "epoch": 2.767162066525124,
+      "grad_norm": 0.20634358143885162,
+      "learning_rate": 4.275970619097587e-06,
+      "loss": 0.3556,
+      "step": 1956
+    },
+    {
+      "epoch": 2.7685774946921446,
+      "grad_norm": 0.20531390532997795,
+      "learning_rate": 4.249737670514166e-06,
+      "loss": 0.3771,
+      "step": 1957
+    },
+    {
+      "epoch": 2.769992922859165,
+      "grad_norm": 0.19949037117645418,
+      "learning_rate": 4.223504721930745e-06,
+      "loss": 0.3612,
+      "step": 1958
+    },
+    {
+      "epoch": 2.7714083510261855,
+      "grad_norm": 0.20340538899445754,
+      "learning_rate": 4.197271773347325e-06,
+      "loss": 0.3953,
+      "step": 1959
+    },
+    {
+      "epoch": 2.7728237791932058,
+      "grad_norm": 0.20229010157197905,
+      "learning_rate": 4.171038824763904e-06,
+      "loss": 0.3768,
+      "step": 1960
+    },
+    {
+      "epoch": 2.7742392073602264,
+      "grad_norm": 0.21675643705520573,
+      "learning_rate": 4.144805876180483e-06,
+      "loss": 0.3985,
+      "step": 1961
+    },
+    {
+      "epoch": 2.775654635527247,
+      "grad_norm": 0.19955801340341722,
+      "learning_rate": 4.118572927597062e-06,
+      "loss": 0.3615,
+      "step": 1962
+    },
+    {
+      "epoch": 2.777070063694268,
+      "grad_norm": 0.18933990643743973,
+      "learning_rate": 4.0923399790136415e-06,
+      "loss": 0.3394,
+      "step": 1963
+    },
+    {
+      "epoch": 2.778485491861288,
+      "grad_norm": 0.19445316739805585,
+      "learning_rate": 4.06610703043022e-06,
+      "loss": 0.3661,
+      "step": 1964
+    },
+    {
+      "epoch": 2.7799009200283087,
+      "grad_norm": 0.2110069160547559,
+      "learning_rate": 4.0398740818467995e-06,
+      "loss": 0.3558,
+      "step": 1965
+    },
+    {
+      "epoch": 2.781316348195329,
+      "grad_norm": 0.2054544611098325,
+      "learning_rate": 4.013641133263379e-06,
+      "loss": 0.3855,
+      "step": 1966
+    },
+    {
+      "epoch": 2.7827317763623496,
+      "grad_norm": 0.18777084253880907,
+      "learning_rate": 3.9874081846799584e-06,
+      "loss": 0.3239,
+      "step": 1967
+    },
+    {
+      "epoch": 2.7841472045293703,
+      "grad_norm": 0.21906578216105788,
+      "learning_rate": 3.961175236096538e-06,
+      "loss": 0.4039,
+      "step": 1968
+    },
+    {
+      "epoch": 2.785562632696391,
+      "grad_norm": 0.22430861257203039,
+      "learning_rate": 3.9349422875131165e-06,
+      "loss": 0.3919,
+      "step": 1969
+    },
+    {
+      "epoch": 2.786978060863411,
+      "grad_norm": 0.2188750732836388,
+      "learning_rate": 3.908709338929696e-06,
+      "loss": 0.408,
+      "step": 1970
+    },
+    {
+      "epoch": 2.788393489030432,
+      "grad_norm": 0.20746037337483966,
+      "learning_rate": 3.882476390346275e-06,
+      "loss": 0.3902,
+      "step": 1971
+    },
+    {
+      "epoch": 2.789808917197452,
+      "grad_norm": 8.05758732383772,
+      "learning_rate": 3.856243441762855e-06,
+      "loss": 0.8494,
+      "step": 1972
+    },
+    {
+      "epoch": 2.7912243453644727,
+      "grad_norm": 0.20122927201049465,
+      "learning_rate": 3.830010493179433e-06,
+      "loss": 0.3641,
+      "step": 1973
+    },
+    {
+      "epoch": 2.7926397735314934,
+      "grad_norm": 0.2066591638924186,
+      "learning_rate": 3.803777544596013e-06,
+      "loss": 0.3927,
+      "step": 1974
+    },
+    {
+      "epoch": 2.794055201698514,
+      "grad_norm": 0.2027018850431106,
+      "learning_rate": 3.7775445960125922e-06,
+      "loss": 0.3506,
+      "step": 1975
+    },
+    {
+      "epoch": 2.7954706298655343,
+      "grad_norm": 0.21715542211666516,
+      "learning_rate": 3.7513116474291713e-06,
+      "loss": 0.3832,
+      "step": 1976
+    },
+    {
+      "epoch": 2.796886058032555,
+      "grad_norm": 0.21860625760823418,
+      "learning_rate": 3.7250786988457507e-06,
+      "loss": 0.4152,
+      "step": 1977
+    },
+    {
+      "epoch": 2.798301486199575,
+      "grad_norm": 0.18673496863635236,
+      "learning_rate": 3.6988457502623293e-06,
+      "loss": 0.2992,
+      "step": 1978
+    },
+    {
+      "epoch": 2.799716914366596,
+      "grad_norm": 0.19095726885505263,
+      "learning_rate": 3.672612801678909e-06,
+      "loss": 0.3625,
+      "step": 1979
+    },
+    {
+      "epoch": 2.8011323425336165,
+      "grad_norm": 1.321622899703763,
+      "learning_rate": 3.6463798530954877e-06,
+      "loss": 0.4226,
+      "step": 1980
+    },
+    {
+      "epoch": 2.802547770700637,
+      "grad_norm": 5.785359000134272,
+      "learning_rate": 3.620146904512067e-06,
+      "loss": 0.3849,
+      "step": 1981
+    },
+    {
+      "epoch": 2.8039631988676574,
+      "grad_norm": 0.21740520427626114,
+      "learning_rate": 3.5939139559286466e-06,
+      "loss": 0.3797,
+      "step": 1982
+    },
+    {
+      "epoch": 2.805378627034678,
+      "grad_norm": 0.20837516610784643,
+      "learning_rate": 3.5676810073452256e-06,
+      "loss": 0.386,
+      "step": 1983
+    },
+    {
+      "epoch": 2.8067940552016983,
+      "grad_norm": 0.19932361211467717,
+      "learning_rate": 3.541448058761805e-06,
+      "loss": 0.3831,
+      "step": 1984
+    },
+    {
+      "epoch": 2.808209483368719,
+      "grad_norm": 0.19986911852545236,
+      "learning_rate": 3.515215110178384e-06,
+      "loss": 0.3374,
+      "step": 1985
+    },
+    {
+      "epoch": 2.8096249115357397,
+      "grad_norm": 0.19704540448560762,
+      "learning_rate": 3.4889821615949635e-06,
+      "loss": 0.3939,
+      "step": 1986
+    },
+    {
+      "epoch": 2.8110403397027603,
+      "grad_norm": 0.20783002369532763,
+      "learning_rate": 3.462749213011543e-06,
+      "loss": 0.3692,
+      "step": 1987
+    },
+    {
+      "epoch": 2.8124557678697806,
+      "grad_norm": 0.20847088393412946,
+      "learning_rate": 3.436516264428122e-06,
+      "loss": 0.3922,
+      "step": 1988
+    },
+    {
+      "epoch": 2.8138711960368012,
+      "grad_norm": 0.19850699462906699,
+      "learning_rate": 3.4102833158447014e-06,
+      "loss": 0.3682,
+      "step": 1989
+    },
+    {
+      "epoch": 2.8152866242038215,
+      "grad_norm": 0.2076469565314169,
+      "learning_rate": 3.38405036726128e-06,
+      "loss": 0.3917,
+      "step": 1990
+    },
+    {
+      "epoch": 2.816702052370842,
+      "grad_norm": 0.20081320372031727,
+      "learning_rate": 3.35781741867786e-06,
+      "loss": 0.3491,
+      "step": 1991
+    },
+    {
+      "epoch": 2.818117480537863,
+      "grad_norm": 0.19552115470271994,
+      "learning_rate": 3.3315844700944385e-06,
+      "loss": 0.3604,
+      "step": 1992
+    },
+    {
+      "epoch": 2.8195329087048835,
+      "grad_norm": 0.20590856907312838,
+      "learning_rate": 3.305351521511018e-06,
+      "loss": 0.381,
+      "step": 1993
+    },
+    {
+      "epoch": 2.8209483368719037,
+      "grad_norm": 0.197122671871231,
+      "learning_rate": 3.2791185729275973e-06,
+      "loss": 0.3928,
+      "step": 1994
+    },
+    {
+      "epoch": 2.8223637650389244,
+      "grad_norm": 0.19390053775641367,
+      "learning_rate": 3.2528856243441763e-06,
+      "loss": 0.3472,
+      "step": 1995
+    },
+    {
+      "epoch": 2.8237791932059446,
+      "grad_norm": 0.18663152400131341,
+      "learning_rate": 3.2266526757607558e-06,
+      "loss": 0.3025,
+      "step": 1996
+    },
+    {
+      "epoch": 2.8251946213729653,
+      "grad_norm": 0.19909280820282357,
+      "learning_rate": 3.200419727177335e-06,
+      "loss": 0.3353,
+      "step": 1997
+    },
+    {
+      "epoch": 2.826610049539986,
+      "grad_norm": 0.19904235811084578,
+      "learning_rate": 3.1741867785939142e-06,
+      "loss": 0.365,
+      "step": 1998
+    },
+    {
+      "epoch": 2.8280254777070066,
+      "grad_norm": 0.20649406059841743,
+      "learning_rate": 3.147953830010493e-06,
+      "loss": 0.3678,
+      "step": 1999
+    },
+    {
+      "epoch": 2.829440905874027,
+      "grad_norm": 0.20909728780145723,
+      "learning_rate": 3.1217208814270727e-06,
+      "loss": 0.4091,
+      "step": 2000
+    },
+    {
+      "epoch": 2.8308563340410475,
+      "grad_norm": 0.19958915848640862,
+      "learning_rate": 3.0954879328436517e-06,
+      "loss": 0.3632,
+      "step": 2001
+    },
+    {
+      "epoch": 2.8322717622080678,
+      "grad_norm": 0.21457143457564132,
+      "learning_rate": 3.0692549842602307e-06,
+      "loss": 0.4022,
+      "step": 2002
+    },
+    {
+      "epoch": 2.8336871903750884,
+      "grad_norm": 0.20610291251086,
+      "learning_rate": 3.04302203567681e-06,
+      "loss": 0.378,
+      "step": 2003
+    },
+    {
+      "epoch": 2.835102618542109,
+      "grad_norm": 0.19436597825698418,
+      "learning_rate": 3.0167890870933896e-06,
+      "loss": 0.3432,
+      "step": 2004
+    },
+    {
+      "epoch": 2.8365180467091298,
+      "grad_norm": 0.20684825531711648,
+      "learning_rate": 2.9905561385099686e-06,
+      "loss": 0.3605,
+      "step": 2005
+    },
+    {
+      "epoch": 2.83793347487615,
+      "grad_norm": 0.7760432398237981,
+      "learning_rate": 2.964323189926548e-06,
+      "loss": 0.3842,
+      "step": 2006
+    },
+    {
+      "epoch": 2.8393489030431707,
+      "grad_norm": 0.21211020483305554,
+      "learning_rate": 2.938090241343127e-06,
+      "loss": 0.4131,
+      "step": 2007
+    },
+    {
+      "epoch": 2.840764331210191,
+      "grad_norm": 0.21399514331029734,
+      "learning_rate": 2.911857292759706e-06,
+      "loss": 0.4113,
+      "step": 2008
+    },
+    {
+      "epoch": 2.8421797593772116,
+      "grad_norm": 0.19378601644607968,
+      "learning_rate": 2.8856243441762855e-06,
+      "loss": 0.347,
+      "step": 2009
+    },
+    {
+      "epoch": 2.8435951875442322,
+      "grad_norm": 0.1933589337467757,
+      "learning_rate": 2.859391395592865e-06,
+      "loss": 0.3398,
+      "step": 2010
+    },
+    {
+      "epoch": 2.845010615711253,
+      "grad_norm": 0.1926902438267549,
+      "learning_rate": 2.833158447009444e-06,
+      "loss": 0.3746,
+      "step": 2011
+    },
+    {
+      "epoch": 2.846426043878273,
+      "grad_norm": 0.19736811412649616,
+      "learning_rate": 2.8069254984260234e-06,
+      "loss": 0.3603,
+      "step": 2012
+    },
+    {
+      "epoch": 2.847841472045294,
+      "grad_norm": 0.19109684975374486,
+      "learning_rate": 2.7806925498426024e-06,
+      "loss": 0.3351,
+      "step": 2013
+    },
+    {
+      "epoch": 2.849256900212314,
+      "grad_norm": 0.2216792369983889,
+      "learning_rate": 2.7544596012591814e-06,
+      "loss": 0.4175,
+      "step": 2014
+    },
+    {
+      "epoch": 2.8506723283793347,
+      "grad_norm": 0.19625843195281362,
+      "learning_rate": 2.728226652675761e-06,
+      "loss": 0.3527,
+      "step": 2015
+    },
+    {
+      "epoch": 2.8520877565463554,
+      "grad_norm": 0.20743441777460606,
+      "learning_rate": 2.7019937040923403e-06,
+      "loss": 0.3935,
+      "step": 2016
+    },
+    {
+      "epoch": 2.853503184713376,
+      "grad_norm": 0.20561747739702188,
+      "learning_rate": 2.6757607555089193e-06,
+      "loss": 0.399,
+      "step": 2017
+    },
+    {
+      "epoch": 2.8549186128803963,
+      "grad_norm": 0.1958029226290106,
+      "learning_rate": 2.6495278069254988e-06,
+      "loss": 0.3618,
+      "step": 2018
+    },
+    {
+      "epoch": 2.856334041047417,
+      "grad_norm": 0.1802716298168241,
+      "learning_rate": 2.6232948583420778e-06,
+      "loss": 0.2994,
+      "step": 2019
+    },
+    {
+      "epoch": 2.857749469214437,
+      "grad_norm": 0.22175156102935556,
+      "learning_rate": 2.597061909758657e-06,
+      "loss": 0.3841,
+      "step": 2020
+    },
+    {
+      "epoch": 2.859164897381458,
+      "grad_norm": 0.208458674583299,
+      "learning_rate": 2.5708289611752362e-06,
+      "loss": 0.3687,
+      "step": 2021
+    },
+    {
+      "epoch": 2.8605803255484785,
+      "grad_norm": 0.23313585596854977,
+      "learning_rate": 2.5445960125918157e-06,
+      "loss": 0.4213,
+      "step": 2022
+    },
+    {
+      "epoch": 2.861995753715499,
+      "grad_norm": 0.20982546110462055,
+      "learning_rate": 2.5183630640083947e-06,
+      "loss": 0.3948,
+      "step": 2023
+    },
+    {
+      "epoch": 2.8634111818825194,
+      "grad_norm": 0.22134081258828486,
+      "learning_rate": 2.4921301154249737e-06,
+      "loss": 0.432,
+      "step": 2024
+    },
+    {
+      "epoch": 2.86482661004954,
+      "grad_norm": 0.20192104723864968,
+      "learning_rate": 2.465897166841553e-06,
+      "loss": 0.3546,
+      "step": 2025
+    },
+    {
+      "epoch": 2.8662420382165603,
+      "grad_norm": 0.19324043372722652,
+      "learning_rate": 2.439664218258132e-06,
+      "loss": 0.3735,
+      "step": 2026
+    },
+    {
+      "epoch": 2.867657466383581,
+      "grad_norm": 0.18986670112243625,
+      "learning_rate": 2.4134312696747116e-06,
+      "loss": 0.3659,
+      "step": 2027
+    },
+    {
+      "epoch": 2.8690728945506017,
+      "grad_norm": 0.19523423146769045,
+      "learning_rate": 2.387198321091291e-06,
+      "loss": 0.3395,
+      "step": 2028
+    },
+    {
+      "epoch": 2.8704883227176223,
+      "grad_norm": 0.19590239410254387,
+      "learning_rate": 2.36096537250787e-06,
+      "loss": 0.3456,
+      "step": 2029
+    },
+    {
+      "epoch": 2.8719037508846426,
+      "grad_norm": 0.19940348026404991,
+      "learning_rate": 2.334732423924449e-06,
+      "loss": 0.3757,
+      "step": 2030
+    },
+    {
+      "epoch": 2.8733191790516632,
+      "grad_norm": 0.20657571108654077,
+      "learning_rate": 2.3084994753410285e-06,
+      "loss": 0.3685,
+      "step": 2031
+    },
+    {
+      "epoch": 2.8747346072186835,
+      "grad_norm": 0.24404659838605067,
+      "learning_rate": 2.2822665267576075e-06,
+      "loss": 0.3772,
+      "step": 2032
+    },
+    {
+      "epoch": 2.876150035385704,
+      "grad_norm": 0.1930235454604922,
+      "learning_rate": 2.256033578174187e-06,
+      "loss": 0.3406,
+      "step": 2033
+    },
+    {
+      "epoch": 2.877565463552725,
+      "grad_norm": 0.2113141777693592,
+      "learning_rate": 2.2298006295907664e-06,
+      "loss": 0.3722,
+      "step": 2034
+    },
+    {
+      "epoch": 2.8789808917197455,
+      "grad_norm": 0.19355370824297508,
+      "learning_rate": 2.2035676810073454e-06,
+      "loss": 0.3416,
+      "step": 2035
+    },
+    {
+      "epoch": 2.8803963198867657,
+      "grad_norm": 0.35498128723227274,
+      "learning_rate": 2.1773347324239244e-06,
+      "loss": 0.3439,
+      "step": 2036
+    },
+    {
+      "epoch": 2.8818117480537864,
+      "grad_norm": 0.21079226500885176,
+      "learning_rate": 2.151101783840504e-06,
+      "loss": 0.3999,
+      "step": 2037
+    },
+    {
+      "epoch": 2.8832271762208066,
+      "grad_norm": 0.19993629279499625,
+      "learning_rate": 2.124868835257083e-06,
+      "loss": 0.379,
+      "step": 2038
+    },
+    {
+      "epoch": 2.8846426043878273,
+      "grad_norm": 0.20265852172396523,
+      "learning_rate": 2.0986358866736623e-06,
+      "loss": 0.3845,
+      "step": 2039
+    },
+    {
+      "epoch": 2.886058032554848,
+      "grad_norm": 0.19966660066869182,
+      "learning_rate": 2.0724029380902413e-06,
+      "loss": 0.3429,
+      "step": 2040
+    },
+    {
+      "epoch": 2.8874734607218686,
+      "grad_norm": 0.19461640316662585,
+      "learning_rate": 2.0461699895068208e-06,
+      "loss": 0.3311,
+      "step": 2041
+    },
+    {
+      "epoch": 2.888888888888889,
+      "grad_norm": 0.18685404198759573,
+      "learning_rate": 2.0199370409233998e-06,
+      "loss": 0.3168,
+      "step": 2042
+    },
+    {
+      "epoch": 2.8903043170559095,
+      "grad_norm": 0.18528748528930483,
+      "learning_rate": 1.9937040923399792e-06,
+      "loss": 0.3445,
+      "step": 2043
+    },
+    {
+      "epoch": 2.8917197452229297,
+      "grad_norm": 0.20334724073798452,
+      "learning_rate": 1.9674711437565582e-06,
+      "loss": 0.387,
+      "step": 2044
+    },
+    {
+      "epoch": 2.8931351733899504,
+      "grad_norm": 0.20068169776681,
+      "learning_rate": 1.9412381951731377e-06,
+      "loss": 0.3706,
+      "step": 2045
+    },
+    {
+      "epoch": 2.894550601556971,
+      "grad_norm": 0.19848960345962377,
+      "learning_rate": 1.9150052465897167e-06,
+      "loss": 0.3848,
+      "step": 2046
+    },
+    {
+      "epoch": 2.8959660297239918,
+      "grad_norm": 0.19394867329081025,
+      "learning_rate": 1.8887722980062961e-06,
+      "loss": 0.3823,
+      "step": 2047
+    },
+    {
+      "epoch": 2.897381457891012,
+      "grad_norm": 0.20136704982699424,
+      "learning_rate": 1.8625393494228753e-06,
+      "loss": 0.3574,
+      "step": 2048
+    },
+    {
+      "epoch": 2.8987968860580327,
+      "grad_norm": 0.18054669373990248,
+      "learning_rate": 1.8363064008394546e-06,
+      "loss": 0.3125,
+      "step": 2049
+    },
+    {
+      "epoch": 2.900212314225053,
+      "grad_norm": 0.1951565386591877,
+      "learning_rate": 1.8100734522560336e-06,
+      "loss": 0.3765,
+      "step": 2050
+    },
+    {
+      "epoch": 2.9016277423920736,
+      "grad_norm": 0.20474264180107443,
+      "learning_rate": 1.7838405036726128e-06,
+      "loss": 0.4115,
+      "step": 2051
+    },
+    {
+      "epoch": 2.903043170559094,
+      "grad_norm": 0.21647592418475772,
+      "learning_rate": 1.757607555089192e-06,
+      "loss": 0.3917,
+      "step": 2052
+    },
+    {
+      "epoch": 2.904458598726115,
+      "grad_norm": 0.18118807335850776,
+      "learning_rate": 1.7313746065057715e-06,
+      "loss": 0.2871,
+      "step": 2053
+    },
+    {
+      "epoch": 2.905874026893135,
+      "grad_norm": 0.19448101636821263,
+      "learning_rate": 1.7051416579223507e-06,
+      "loss": 0.3613,
+      "step": 2054
+    },
+    {
+      "epoch": 2.907289455060156,
+      "grad_norm": 0.18341263030502164,
+      "learning_rate": 1.67890870933893e-06,
+      "loss": 0.3271,
+      "step": 2055
+    },
+    {
+      "epoch": 2.908704883227176,
+      "grad_norm": 0.21586315019415844,
+      "learning_rate": 1.652675760755509e-06,
+      "loss": 0.3914,
+      "step": 2056
+    },
+    {
+      "epoch": 2.9101203113941967,
+      "grad_norm": 0.21643695397466847,
+      "learning_rate": 1.6264428121720882e-06,
+      "loss": 0.4036,
+      "step": 2057
+    },
+    {
+      "epoch": 2.9115357395612174,
+      "grad_norm": 0.19145620703601973,
+      "learning_rate": 1.6002098635886674e-06,
+      "loss": 0.3502,
+      "step": 2058
+    },
+    {
+      "epoch": 2.912951167728238,
+      "grad_norm": 0.19768294939507053,
+      "learning_rate": 1.5739769150052464e-06,
+      "loss": 0.3657,
+      "step": 2059
+    },
+    {
+      "epoch": 2.9143665958952583,
+      "grad_norm": 0.18062247390735356,
+      "learning_rate": 1.5477439664218259e-06,
+      "loss": 0.3118,
+      "step": 2060
+    },
+    {
+      "epoch": 2.915782024062279,
+      "grad_norm": 0.20220286484002623,
+      "learning_rate": 1.521511017838405e-06,
+      "loss": 0.3525,
+      "step": 2061
+    },
+    {
+      "epoch": 2.917197452229299,
+      "grad_norm": 0.18868728285298295,
+      "learning_rate": 1.4952780692549843e-06,
+      "loss": 0.3444,
+      "step": 2062
+    },
+    {
+      "epoch": 2.91861288039632,
+      "grad_norm": 0.2750350768980409,
+      "learning_rate": 1.4690451206715635e-06,
+      "loss": 0.4072,
+      "step": 2063
+    },
+    {
+      "epoch": 2.9200283085633405,
+      "grad_norm": 0.1784673086428867,
+      "learning_rate": 1.4428121720881428e-06,
+      "loss": 0.3148,
+      "step": 2064
+    },
+    {
+      "epoch": 2.921443736730361,
+      "grad_norm": 0.19792808089582006,
+      "learning_rate": 1.416579223504722e-06,
+      "loss": 0.3966,
+      "step": 2065
+    },
+    {
+      "epoch": 2.9228591648973814,
+      "grad_norm": 0.19134347414255334,
+      "learning_rate": 1.3903462749213012e-06,
+      "loss": 0.3323,
+      "step": 2066
+    },
+    {
+      "epoch": 2.924274593064402,
+      "grad_norm": 0.2035122251198033,
+      "learning_rate": 1.3641133263378804e-06,
+      "loss": 0.3651,
+      "step": 2067
+    },
+    {
+      "epoch": 2.9256900212314223,
+      "grad_norm": 0.20033413515213966,
+      "learning_rate": 1.3378803777544597e-06,
+      "loss": 0.3983,
+      "step": 2068
+    },
+    {
+      "epoch": 2.927105449398443,
+      "grad_norm": 0.18453302874158478,
+      "learning_rate": 1.3116474291710389e-06,
+      "loss": 0.3396,
+      "step": 2069
+    },
+    {
+      "epoch": 2.9285208775654636,
+      "grad_norm": 0.1882995173465666,
+      "learning_rate": 1.2854144805876181e-06,
+      "loss": 0.3399,
+      "step": 2070
+    },
+    {
+      "epoch": 2.9299363057324843,
+      "grad_norm": 0.20725779599771735,
+      "learning_rate": 1.2591815320041973e-06,
+      "loss": 0.3987,
+      "step": 2071
+    },
+    {
+      "epoch": 2.9313517338995045,
+      "grad_norm": 0.1883440700700296,
+      "learning_rate": 1.2329485834207766e-06,
+      "loss": 0.345,
+      "step": 2072
+    },
+    {
+      "epoch": 2.932767162066525,
+      "grad_norm": 0.20368211705548797,
+      "learning_rate": 1.2067156348373558e-06,
+      "loss": 0.3824,
+      "step": 2073
+    },
+    {
+      "epoch": 2.9341825902335454,
+      "grad_norm": 0.2095686727474719,
+      "learning_rate": 1.180482686253935e-06,
+      "loss": 0.3781,
+      "step": 2074
+    },
+    {
+      "epoch": 2.935598018400566,
+      "grad_norm": 0.18957399098358954,
+      "learning_rate": 1.1542497376705142e-06,
+      "loss": 0.3661,
+      "step": 2075
+    },
+    {
+      "epoch": 2.937013446567587,
+      "grad_norm": 0.1941943711092127,
+      "learning_rate": 1.1280167890870935e-06,
+      "loss": 0.3539,
+      "step": 2076
+    },
+    {
+      "epoch": 2.9384288747346075,
+      "grad_norm": 0.19737218357988998,
+      "learning_rate": 1.1017838405036727e-06,
+      "loss": 0.3701,
+      "step": 2077
+    },
+    {
+      "epoch": 2.9398443029016277,
+      "grad_norm": 0.21773224729027893,
+      "learning_rate": 1.075550891920252e-06,
+      "loss": 0.4206,
+      "step": 2078
+    },
+    {
+      "epoch": 2.9412597310686484,
+      "grad_norm": 4.797610101801964,
+      "learning_rate": 1.0493179433368312e-06,
+      "loss": 0.7369,
+      "step": 2079
+    },
+    {
+      "epoch": 2.9426751592356686,
+      "grad_norm": 0.1935476966162268,
+      "learning_rate": 1.0230849947534104e-06,
+      "loss": 0.3838,
+      "step": 2080
+    },
+    {
+      "epoch": 2.9440905874026893,
+      "grad_norm": 0.18450472118161934,
+      "learning_rate": 9.968520461699896e-07,
+      "loss": 0.3123,
+      "step": 2081
+    },
+    {
+      "epoch": 2.94550601556971,
+      "grad_norm": 0.1900464839132083,
+      "learning_rate": 9.706190975865688e-07,
+      "loss": 0.3108,
+      "step": 2082
+    },
+    {
+      "epoch": 2.9469214437367306,
+      "grad_norm": 0.19113932244352635,
+      "learning_rate": 9.443861490031481e-07,
+      "loss": 0.3205,
+      "step": 2083
+    },
+    {
+      "epoch": 2.948336871903751,
+      "grad_norm": 0.1976782716367035,
+      "learning_rate": 9.181532004197273e-07,
+      "loss": 0.3611,
+      "step": 2084
+    },
+    {
+      "epoch": 2.9497523000707715,
+      "grad_norm": 0.18628514897246456,
+      "learning_rate": 8.919202518363064e-07,
+      "loss": 0.3401,
+      "step": 2085
+    },
+    {
+      "epoch": 2.9511677282377917,
+      "grad_norm": 0.20330699467998223,
+      "learning_rate": 8.656873032528857e-07,
+      "loss": 0.4181,
+      "step": 2086
+    },
+    {
+      "epoch": 2.9525831564048124,
+      "grad_norm": 0.1986337270652071,
+      "learning_rate": 8.39454354669465e-07,
+      "loss": 0.3437,
+      "step": 2087
+    },
+    {
+      "epoch": 2.953998584571833,
+      "grad_norm": 0.2936645275997326,
+      "learning_rate": 8.132214060860441e-07,
+      "loss": 0.3972,
+      "step": 2088
+    },
+    {
+      "epoch": 2.9554140127388537,
+      "grad_norm": 0.18414900018164665,
+      "learning_rate": 7.869884575026232e-07,
+      "loss": 0.3218,
+      "step": 2089
+    },
+    {
+      "epoch": 2.956829440905874,
+      "grad_norm": 0.18070732731376302,
+      "learning_rate": 7.607555089192025e-07,
+      "loss": 0.2919,
+      "step": 2090
+    },
+    {
+      "epoch": 2.9582448690728946,
+      "grad_norm": 0.1838248187014801,
+      "learning_rate": 7.345225603357818e-07,
+      "loss": 0.3345,
+      "step": 2091
+    },
+    {
+      "epoch": 2.959660297239915,
+      "grad_norm": 0.21380101541068397,
+      "learning_rate": 7.08289611752361e-07,
+      "loss": 0.4093,
+      "step": 2092
+    },
+    {
+      "epoch": 2.9610757254069355,
+      "grad_norm": 0.20197658636303287,
+      "learning_rate": 6.820566631689402e-07,
+      "loss": 0.3728,
+      "step": 2093
+    },
+    {
+      "epoch": 2.962491153573956,
+      "grad_norm": 0.21149661656774726,
+      "learning_rate": 6.558237145855194e-07,
+      "loss": 0.3927,
+      "step": 2094
+    },
+    {
+      "epoch": 2.963906581740977,
+      "grad_norm": 0.1936150798022257,
+      "learning_rate": 6.295907660020987e-07,
+      "loss": 0.3682,
+      "step": 2095
+    },
+    {
+      "epoch": 2.965322009907997,
+      "grad_norm": 0.2007856015130988,
+      "learning_rate": 6.033578174186779e-07,
+      "loss": 0.397,
+      "step": 2096
+    },
+    {
+      "epoch": 2.9667374380750178,
+      "grad_norm": 0.19695599331631608,
+      "learning_rate": 5.771248688352571e-07,
+      "loss": 0.3648,
+      "step": 2097
+    },
+    {
+      "epoch": 2.968152866242038,
+      "grad_norm": 0.19728088677482633,
+      "learning_rate": 5.508919202518364e-07,
+      "loss": 0.3797,
+      "step": 2098
+    },
+    {
+      "epoch": 2.9695682944090587,
+      "grad_norm": 0.1858177347450067,
+      "learning_rate": 5.246589716684156e-07,
+      "loss": 0.3296,
+      "step": 2099
+    },
+    {
+      "epoch": 2.9709837225760793,
+      "grad_norm": 0.21170677276318048,
+      "learning_rate": 4.984260230849948e-07,
+      "loss": 0.416,
+      "step": 2100
+    },
+    {
+      "epoch": 2.9723991507431,
+      "grad_norm": 0.18831289221943484,
+      "learning_rate": 4.7219307450157403e-07,
+      "loss": 0.3297,
+      "step": 2101
+    },
+    {
+      "epoch": 2.9738145789101202,
+      "grad_norm": 0.17619689265975416,
+      "learning_rate": 4.459601259181532e-07,
+      "loss": 0.3069,
+      "step": 2102
+    },
+    {
+      "epoch": 2.975230007077141,
+      "grad_norm": 0.22540108458031174,
+      "learning_rate": 4.197271773347325e-07,
+      "loss": 0.3753,
+      "step": 2103
+    },
+    {
+      "epoch": 2.976645435244161,
+      "grad_norm": 0.19400149626944135,
+      "learning_rate": 3.934942287513116e-07,
+      "loss": 0.3474,
+      "step": 2104
+    },
+    {
+      "epoch": 2.978060863411182,
+      "grad_norm": 0.18962762860314583,
+      "learning_rate": 3.672612801678909e-07,
+      "loss": 0.333,
+      "step": 2105
+    },
+    {
+      "epoch": 2.9794762915782025,
+      "grad_norm": 0.19093979940298456,
+      "learning_rate": 3.410283315844701e-07,
+      "loss": 0.3336,
+      "step": 2106
+    },
+    {
+      "epoch": 2.980891719745223,
+      "grad_norm": 0.19735335084880196,
+      "learning_rate": 3.1479538300104934e-07,
+      "loss": 0.346,
+      "step": 2107
+    },
+    {
+      "epoch": 2.9823071479122434,
+      "grad_norm": 0.19011476516224762,
+      "learning_rate": 2.8856243441762856e-07,
+      "loss": 0.3321,
+      "step": 2108
+    },
+    {
+      "epoch": 2.983722576079264,
+      "grad_norm": 0.2017245142823575,
+      "learning_rate": 2.623294858342078e-07,
+      "loss": 0.3611,
+      "step": 2109
+    },
+    {
+      "epoch": 2.9851380042462843,
+      "grad_norm": 0.1830758960092337,
+      "learning_rate": 2.3609653725078701e-07,
+      "loss": 0.3288,
+      "step": 2110
+    },
+    {
+      "epoch": 2.986553432413305,
+      "grad_norm": 0.19067503603986885,
+      "learning_rate": 2.0986358866736624e-07,
+      "loss": 0.3642,
+      "step": 2111
+    },
+    {
+      "epoch": 2.9879688605803256,
+      "grad_norm": 0.19156299414517605,
+      "learning_rate": 1.8363064008394544e-07,
+      "loss": 0.3569,
+      "step": 2112
+    },
+    {
+      "epoch": 2.9893842887473463,
+      "grad_norm": 0.2055866092880421,
+      "learning_rate": 1.5739769150052467e-07,
+      "loss": 0.4054,
+      "step": 2113
+    },
+    {
+      "epoch": 2.9907997169143665,
+      "grad_norm": 0.22750293431289148,
+      "learning_rate": 1.311647429171039e-07,
+      "loss": 0.4022,
+      "step": 2114
+    },
+    {
+      "epoch": 2.992215145081387,
+      "grad_norm": 0.19484546995446372,
+      "learning_rate": 1.0493179433368312e-07,
+      "loss": 0.35,
+      "step": 2115
+    },
+    {
+      "epoch": 2.9936305732484074,
+      "grad_norm": 0.20414099987110698,
+      "learning_rate": 7.869884575026233e-08,
+      "loss": 0.4187,
+      "step": 2116
+    },
+    {
+      "epoch": 2.995046001415428,
+      "grad_norm": 0.19780068062196055,
+      "learning_rate": 5.246589716684156e-08,
+      "loss": 0.3583,
+      "step": 2117
+    },
+    {
+      "epoch": 2.9964614295824488,
+      "grad_norm": 0.18612586736128045,
+      "learning_rate": 2.623294858342078e-08,
+      "loss": 0.3423,
+      "step": 2118
+    },
+    {
+      "epoch": 2.9964614295824488,
+      "step": 2118,
+      "total_flos": 2.3539982914407105e+19,
+      "train_loss": 0.570159648968095,
+      "train_runtime": 70788.7073,
+      "train_samples_per_second": 0.479,
+      "train_steps_per_second": 0.03
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 2118,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.3539982914407105e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}