diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5677 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.9891808346213296,
+  "eval_steps": 500,
+  "global_step": 805,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0061823802163833074,
+      "grad_norm": 10.693047278702467,
+      "learning_rate": 9.876543209876544e-07,
+      "loss": 1.6775,
+      "step": 1
+    },
+    {
+      "epoch": 0.012364760432766615,
+      "grad_norm": 10.815416954217403,
+      "learning_rate": 1.9753086419753087e-06,
+      "loss": 1.6824,
+      "step": 2
+    },
+    {
+      "epoch": 0.01854714064914992,
+      "grad_norm": 10.616338898894972,
+      "learning_rate": 2.962962962962963e-06,
+      "loss": 1.6745,
+      "step": 3
+    },
+    {
+      "epoch": 0.02472952086553323,
+      "grad_norm": 9.861459123224664,
+      "learning_rate": 3.9506172839506175e-06,
+      "loss": 1.6421,
+      "step": 4
+    },
+    {
+      "epoch": 0.030911901081916538,
+      "grad_norm": 7.614357546616699,
+      "learning_rate": 4.938271604938272e-06,
+      "loss": 1.5691,
+      "step": 5
+    },
+    {
+      "epoch": 0.03709428129829984,
+      "grad_norm": 4.203625583627038,
+      "learning_rate": 5.925925925925926e-06,
+      "loss": 1.5004,
+      "step": 6
+    },
+    {
+      "epoch": 0.04327666151468315,
+      "grad_norm": 3.5662282715378195,
+      "learning_rate": 6.913580246913581e-06,
+      "loss": 1.4777,
+      "step": 7
+    },
+    {
+      "epoch": 0.04945904173106646,
+      "grad_norm": 5.739743938641111,
+      "learning_rate": 7.901234567901235e-06,
+      "loss": 1.4747,
+      "step": 8
+    },
+    {
+      "epoch": 0.05564142194744977,
+      "grad_norm": 6.180000116352135,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 1.4418,
+      "step": 9
+    },
+    {
+      "epoch": 0.061823802163833076,
+      "grad_norm": 6.739405843412405,
+      "learning_rate": 9.876543209876543e-06,
+      "loss": 1.3952,
+      "step": 10
+    },
+    {
+      "epoch": 0.06800618238021638,
+      "grad_norm": 7.52997521103117,
+      "learning_rate": 1.0864197530864198e-05,
+      "loss": 1.3959,
+      "step": 11
+    },
+    {
+      "epoch": 0.07418856259659969,
+      "grad_norm": 6.385085000826414,
+      "learning_rate": 1.1851851851851852e-05,
+      "loss": 1.363,
+      "step": 12
+    },
+    {
+      "epoch": 0.080370942812983,
+      "grad_norm": 5.007200781174352,
+      "learning_rate": 1.2839506172839507e-05,
+      "loss": 1.3392,
+      "step": 13
+    },
+    {
+      "epoch": 0.0865533230293663,
+      "grad_norm": 4.058377749470976,
+      "learning_rate": 1.3827160493827162e-05,
+      "loss": 1.3045,
+      "step": 14
+    },
+    {
+      "epoch": 0.09273570324574962,
+      "grad_norm": 3.5316422870540793,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 1.2877,
+      "step": 15
+    },
+    {
+      "epoch": 0.09891808346213292,
+      "grad_norm": 3.842638477890943,
+      "learning_rate": 1.580246913580247e-05,
+      "loss": 1.2683,
+      "step": 16
+    },
+    {
+      "epoch": 0.10510046367851623,
+      "grad_norm": 3.1304998105929815,
+      "learning_rate": 1.6790123456790123e-05,
+      "loss": 1.2722,
+      "step": 17
+    },
+    {
+      "epoch": 0.11128284389489954,
+      "grad_norm": 1.831350110427916,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 1.2427,
+      "step": 18
+    },
+    {
+      "epoch": 0.11746522411128284,
+      "grad_norm": 2.1448411883286593,
+      "learning_rate": 1.8765432098765433e-05,
+      "loss": 1.238,
+      "step": 19
+    },
+    {
+      "epoch": 0.12364760432766615,
+      "grad_norm": 2.0898956083682263,
+      "learning_rate": 1.9753086419753087e-05,
+      "loss": 1.2214,
+      "step": 20
+    },
+    {
+      "epoch": 0.12982998454404945,
+      "grad_norm": 1.5783782113338631,
+      "learning_rate": 2.074074074074074e-05,
+      "loss": 1.21,
+      "step": 21
+    },
+    {
+      "epoch": 0.13601236476043277,
+      "grad_norm": 1.7643935663724537,
+      "learning_rate": 2.1728395061728397e-05,
+      "loss": 1.1865,
+      "step": 22
+    },
+    {
+      "epoch": 0.14219474497681608,
+      "grad_norm": 2.0831771497504783,
+      "learning_rate": 2.271604938271605e-05,
+      "loss": 1.195,
+      "step": 23
+    },
+    {
+      "epoch": 0.14837712519319937,
+      "grad_norm": 1.7488536795684462,
+      "learning_rate": 2.3703703703703703e-05,
+      "loss": 1.1696,
+      "step": 24
+    },
+    {
+      "epoch": 0.1545595054095827,
+      "grad_norm": 1.5931909595273852,
+      "learning_rate": 2.469135802469136e-05,
+      "loss": 1.1892,
+      "step": 25
+    },
+    {
+      "epoch": 0.160741885625966,
+      "grad_norm": 1.396445568089242,
+      "learning_rate": 2.5679012345679013e-05,
+      "loss": 1.1793,
+      "step": 26
+    },
+    {
+      "epoch": 0.16692426584234932,
+      "grad_norm": 1.0145848862012221,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 1.1664,
+      "step": 27
+    },
+    {
+      "epoch": 0.1731066460587326,
+      "grad_norm": 2.4324574107146226,
+      "learning_rate": 2.7654320987654323e-05,
+      "loss": 1.1573,
+      "step": 28
+    },
+    {
+      "epoch": 0.17928902627511592,
+      "grad_norm": 1.4236022981017393,
+      "learning_rate": 2.8641975308641977e-05,
+      "loss": 1.1501,
+      "step": 29
+    },
+    {
+      "epoch": 0.18547140649149924,
+      "grad_norm": 2.1562503786156153,
+      "learning_rate": 2.962962962962963e-05,
+      "loss": 1.1608,
+      "step": 30
+    },
+    {
+      "epoch": 0.19165378670788252,
+      "grad_norm": 1.9018824527792042,
+      "learning_rate": 3.061728395061729e-05,
+      "loss": 1.1451,
+      "step": 31
+    },
+    {
+      "epoch": 0.19783616692426584,
+      "grad_norm": 1.5703835911829096,
+      "learning_rate": 3.160493827160494e-05,
+      "loss": 1.1408,
+      "step": 32
+    },
+    {
+      "epoch": 0.20401854714064915,
+      "grad_norm": 8.652718407803869,
+      "learning_rate": 3.259259259259259e-05,
+      "loss": 1.1346,
+      "step": 33
+    },
+    {
+      "epoch": 0.21020092735703247,
+      "grad_norm": 6.435439968019908,
+      "learning_rate": 3.3580246913580247e-05,
+      "loss": 1.1582,
+      "step": 34
+    },
+    {
+      "epoch": 0.21638330757341576,
+      "grad_norm": 105.3495618548623,
+      "learning_rate": 3.45679012345679e-05,
+      "loss": 1.1558,
+      "step": 35
+    },
+    {
+      "epoch": 0.22256568778979907,
+      "grad_norm": 2.5364224403663056,
+      "learning_rate": 3.555555555555555e-05,
+      "loss": 1.1431,
+      "step": 36
+    },
+    {
+      "epoch": 0.2287480680061824,
+      "grad_norm": 1.9024835350187577,
+      "learning_rate": 3.654320987654321e-05,
+      "loss": 1.1463,
+      "step": 37
+    },
+    {
+      "epoch": 0.23493044822256567,
+      "grad_norm": 2.0426678600887254,
+      "learning_rate": 3.7530864197530867e-05,
+      "loss": 1.1307,
+      "step": 38
+    },
+    {
+      "epoch": 0.241112828438949,
+      "grad_norm": 1.8582429580354745,
+      "learning_rate": 3.851851851851852e-05,
+      "loss": 1.1272,
+      "step": 39
+    },
+    {
+      "epoch": 0.2472952086553323,
+      "grad_norm": 1.860203917219744,
+      "learning_rate": 3.950617283950617e-05,
+      "loss": 1.1137,
+      "step": 40
+    },
+    {
+      "epoch": 0.2534775888717156,
+      "grad_norm": 10.27861275542122,
+      "learning_rate": 4.049382716049383e-05,
+      "loss": 1.1247,
+      "step": 41
+    },
+    {
+      "epoch": 0.2596599690880989,
+      "grad_norm": 5.066950295263267,
+      "learning_rate": 4.148148148148148e-05,
+      "loss": 1.1334,
+      "step": 42
+    },
+    {
+      "epoch": 0.26584234930448225,
+      "grad_norm": 3.502140270278529,
+      "learning_rate": 4.246913580246914e-05,
+      "loss": 1.1114,
+      "step": 43
+    },
+    {
+      "epoch": 0.27202472952086554,
+      "grad_norm": 3.203814615327836,
+      "learning_rate": 4.345679012345679e-05,
+      "loss": 1.1182,
+      "step": 44
+    },
+    {
+      "epoch": 0.2782071097372488,
+      "grad_norm": 2.4989709311447954,
+      "learning_rate": 4.444444444444445e-05,
+      "loss": 1.1157,
+      "step": 45
+    },
+    {
+      "epoch": 0.28438948995363217,
+      "grad_norm": 3.2014585800746813,
+      "learning_rate": 4.54320987654321e-05,
+      "loss": 1.1158,
+      "step": 46
+    },
+    {
+      "epoch": 0.29057187017001546,
+      "grad_norm": 2.3255133597202073,
+      "learning_rate": 4.641975308641976e-05,
+      "loss": 1.1214,
+      "step": 47
+    },
+    {
+      "epoch": 0.29675425038639874,
+      "grad_norm": 3.340427993360326,
+      "learning_rate": 4.7407407407407407e-05,
+      "loss": 1.1072,
+      "step": 48
+    },
+    {
+      "epoch": 0.3029366306027821,
+      "grad_norm": 2.6053156370494137,
+      "learning_rate": 4.8395061728395067e-05,
+      "loss": 1.1125,
+      "step": 49
+    },
+    {
+      "epoch": 0.3091190108191654,
+      "grad_norm": 3.416491408495564,
+      "learning_rate": 4.938271604938272e-05,
+      "loss": 1.1076,
+      "step": 50
+    },
+    {
+      "epoch": 0.31530139103554866,
+      "grad_norm": 3.0240291240770016,
+      "learning_rate": 5.037037037037037e-05,
+      "loss": 1.1039,
+      "step": 51
+    },
+    {
+      "epoch": 0.321483771251932,
+      "grad_norm": 2.6974546966855173,
+      "learning_rate": 5.1358024691358027e-05,
+      "loss": 1.1083,
+      "step": 52
+    },
+    {
+      "epoch": 0.3276661514683153,
+      "grad_norm": 2.1712600314640174,
+      "learning_rate": 5.234567901234569e-05,
+      "loss": 1.0969,
+      "step": 53
+    },
+    {
+      "epoch": 0.33384853168469864,
+      "grad_norm": 2.43061865655575,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 1.1009,
+      "step": 54
+    },
+    {
+      "epoch": 0.3400309119010819,
+      "grad_norm": 1.4299217962960098,
+      "learning_rate": 5.432098765432099e-05,
+      "loss": 1.1071,
+      "step": 55
+    },
+    {
+      "epoch": 0.3462132921174652,
+      "grad_norm": 3.4472743528033942,
+      "learning_rate": 5.5308641975308647e-05,
+      "loss": 1.1034,
+      "step": 56
+    },
+    {
+      "epoch": 0.35239567233384855,
+      "grad_norm": 2.2533692308911486,
+      "learning_rate": 5.62962962962963e-05,
+      "loss": 1.0926,
+      "step": 57
+    },
+    {
+      "epoch": 0.35857805255023184,
+      "grad_norm": 3.124786132702191,
+      "learning_rate": 5.728395061728395e-05,
+      "loss": 1.0998,
+      "step": 58
+    },
+    {
+      "epoch": 0.36476043276661513,
+      "grad_norm": 2.9945765774059847,
+      "learning_rate": 5.827160493827161e-05,
+      "loss": 1.0936,
+      "step": 59
+    },
+    {
+      "epoch": 0.37094281298299847,
+      "grad_norm": 1.9029030448259223,
+      "learning_rate": 5.925925925925926e-05,
+      "loss": 1.0865,
+      "step": 60
+    },
+    {
+      "epoch": 0.37712519319938176,
+      "grad_norm": 2.5191915533993434,
+      "learning_rate": 6.024691358024692e-05,
+      "loss": 1.094,
+      "step": 61
+    },
+    {
+      "epoch": 0.38330757341576505,
+      "grad_norm": 2.377743288984034,
+      "learning_rate": 6.123456790123457e-05,
+      "loss": 1.0724,
+      "step": 62
+    },
+    {
+      "epoch": 0.3894899536321484,
+      "grad_norm": 3.418329883653554,
+      "learning_rate": 6.222222222222223e-05,
+      "loss": 1.1097,
+      "step": 63
+    },
+    {
+      "epoch": 0.3956723338485317,
+      "grad_norm": 1.8243621455646457,
+      "learning_rate": 6.320987654320988e-05,
+      "loss": 1.0972,
+      "step": 64
+    },
+    {
+      "epoch": 0.401854714064915,
+      "grad_norm": 3.3653254310211804,
+      "learning_rate": 6.419753086419754e-05,
+      "loss": 1.0954,
+      "step": 65
+    },
+    {
+      "epoch": 0.4080370942812983,
+      "grad_norm": 2.859591151453805,
+      "learning_rate": 6.518518518518519e-05,
+      "loss": 1.0887,
+      "step": 66
+    },
+    {
+      "epoch": 0.4142194744976816,
+      "grad_norm": 2.132914874530937,
+      "learning_rate": 6.617283950617285e-05,
+      "loss": 1.0847,
+      "step": 67
+    },
+    {
+      "epoch": 0.42040185471406494,
+      "grad_norm": 3.0306427822925386,
+      "learning_rate": 6.716049382716049e-05,
+      "loss": 1.1009,
+      "step": 68
+    },
+    {
+      "epoch": 0.4265842349304482,
+      "grad_norm": 2.0609891283339477,
+      "learning_rate": 6.814814814814815e-05,
+      "loss": 1.092,
+      "step": 69
+    },
+    {
+      "epoch": 0.4327666151468315,
+      "grad_norm": 2.9169289764667603,
+      "learning_rate": 6.91358024691358e-05,
+      "loss": 1.0721,
+      "step": 70
+    },
+    {
+      "epoch": 0.43894899536321486,
+      "grad_norm": 2.170819034455064,
+      "learning_rate": 7.012345679012346e-05,
+      "loss": 1.0752,
+      "step": 71
+    },
+    {
+      "epoch": 0.44513137557959814,
+      "grad_norm": 2.6561925596391256,
+      "learning_rate": 7.11111111111111e-05,
+      "loss": 1.0763,
+      "step": 72
+    },
+    {
+      "epoch": 0.45131375579598143,
+      "grad_norm": 2.0412946622837453,
+      "learning_rate": 7.209876543209877e-05,
+      "loss": 1.0855,
+      "step": 73
+    },
+    {
+      "epoch": 0.4574961360123648,
+      "grad_norm": 1.7224050848012167,
+      "learning_rate": 7.308641975308643e-05,
+      "loss": 1.0781,
+      "step": 74
+    },
+    {
+      "epoch": 0.46367851622874806,
+      "grad_norm": 2.0509259917647564,
+      "learning_rate": 7.407407407407409e-05,
+      "loss": 1.0864,
+      "step": 75
+    },
+    {
+      "epoch": 0.46986089644513135,
+      "grad_norm": 3.9723221221808225,
+      "learning_rate": 7.506172839506173e-05,
+      "loss": 1.077,
+      "step": 76
+    },
+    {
+      "epoch": 0.4760432766615147,
+      "grad_norm": 2.5313197271865464,
+      "learning_rate": 7.60493827160494e-05,
+      "loss": 1.0873,
+      "step": 77
+    },
+    {
+      "epoch": 0.482225656877898,
+      "grad_norm": 2.6058497426158342,
+      "learning_rate": 7.703703703703704e-05,
+      "loss": 1.0912,
+      "step": 78
+    },
+    {
+      "epoch": 0.4884080370942813,
+      "grad_norm": 4.045410961666709,
+      "learning_rate": 7.80246913580247e-05,
+      "loss": 1.0905,
+      "step": 79
+    },
+    {
+      "epoch": 0.4945904173106646,
+      "grad_norm": 2.085038720402762,
+      "learning_rate": 7.901234567901235e-05,
+      "loss": 1.0796,
+      "step": 80
+    },
+    {
+      "epoch": 0.500772797527048,
+      "grad_norm": 4.492244112618374,
+      "learning_rate": 8e-05,
+      "loss": 1.094,
+      "step": 81
+    },
+    {
+      "epoch": 0.5069551777434312,
+      "grad_norm": 3.6362050618093122,
+      "learning_rate": 7.999962342461635e-05,
+      "loss": 1.1031,
+      "step": 82
+    },
+    {
+      "epoch": 0.5131375579598145,
+      "grad_norm": 3.1038756568216317,
+      "learning_rate": 7.999849370555585e-05,
+      "loss": 1.0726,
+      "step": 83
+    },
+    {
+      "epoch": 0.5193199381761978,
+      "grad_norm": 2.526971760311998,
+      "learning_rate": 7.999661086408972e-05,
+      "loss": 1.0824,
+      "step": 84
+    },
+    {
+      "epoch": 0.5255023183925811,
+      "grad_norm": 1.8821681779156758,
+      "learning_rate": 7.999397493566954e-05,
+      "loss": 1.0752,
+      "step": 85
+    },
+    {
+      "epoch": 0.5316846986089645,
+      "grad_norm": 2.3448656117874913,
+      "learning_rate": 7.99905859699266e-05,
+      "loss": 1.0908,
+      "step": 86
+    },
+    {
+      "epoch": 0.5378670788253478,
+      "grad_norm": 2.0611523416179542,
+      "learning_rate": 7.998644403067097e-05,
+      "loss": 1.0721,
+      "step": 87
+    },
+    {
+      "epoch": 0.5440494590417311,
+      "grad_norm": 2.9983162968300983,
+      "learning_rate": 7.998154919589024e-05,
+      "loss": 1.0776,
+      "step": 88
+    },
+    {
+      "epoch": 0.5502318392581144,
+      "grad_norm": 2.1806569564777987,
+      "learning_rate": 7.997590155774815e-05,
+      "loss": 1.0733,
+      "step": 89
+    },
+    {
+      "epoch": 0.5564142194744977,
+      "grad_norm": 2.1571411866423533,
+      "learning_rate": 7.996950122258277e-05,
+      "loss": 1.0577,
+      "step": 90
+    },
+    {
+      "epoch": 0.5625965996908809,
+      "grad_norm": 3.999074777477232,
+      "learning_rate": 7.996234831090451e-05,
+      "loss": 1.0793,
+      "step": 91
+    },
+    {
+      "epoch": 0.5687789799072643,
+      "grad_norm": 2.2312489091174172,
+      "learning_rate": 7.995444295739394e-05,
+      "loss": 1.0775,
+      "step": 92
+    },
+    {
+      "epoch": 0.5749613601236476,
+      "grad_norm": 6.134653577370118,
+      "learning_rate": 7.994578531089908e-05,
+      "loss": 1.0927,
+      "step": 93
+    },
+    {
+      "epoch": 0.5811437403400309,
+      "grad_norm": 6.274513062714145,
+      "learning_rate": 7.99363755344328e-05,
+      "loss": 1.1068,
+      "step": 94
+    },
+    {
+      "epoch": 0.5873261205564142,
+      "grad_norm": 2.3500100638359287,
+      "learning_rate": 7.99262138051696e-05,
+      "loss": 1.0885,
+      "step": 95
+    },
+    {
+      "epoch": 0.5935085007727975,
+      "grad_norm": 4.48097545076724,
+      "learning_rate": 7.991530031444231e-05,
+      "loss": 1.0838,
+      "step": 96
+    },
+    {
+      "epoch": 0.5996908809891809,
+      "grad_norm": 3.749050135435733,
+      "learning_rate": 7.990363526773856e-05,
+      "loss": 1.0947,
+      "step": 97
+    },
+    {
+      "epoch": 0.6058732612055642,
+      "grad_norm": 2.7226697465582674,
+      "learning_rate": 7.98912188846968e-05,
+      "loss": 1.0835,
+      "step": 98
+    },
+    {
+      "epoch": 0.6120556414219475,
+      "grad_norm": 2.7748343898198073,
+      "learning_rate": 7.987805139910226e-05,
+      "loss": 1.0601,
+      "step": 99
+    },
+    {
+      "epoch": 0.6182380216383307,
+      "grad_norm": 2.0786601854908087,
+      "learning_rate": 7.986413305888247e-05,
+      "loss": 1.0628,
+      "step": 100
+    },
+    {
+      "epoch": 0.624420401854714,
+      "grad_norm": 3.0450749679460842,
+      "learning_rate": 7.984946412610265e-05,
+      "loss": 1.0746,
+      "step": 101
+    },
+    {
+      "epoch": 0.6306027820710973,
+      "grad_norm": 2.0666897869515273,
+      "learning_rate": 7.983404487696076e-05,
+      "loss": 1.0712,
+      "step": 102
+    },
+    {
+      "epoch": 0.6367851622874807,
+      "grad_norm": 3.3981180649215212,
+      "learning_rate": 7.981787560178227e-05,
+      "loss": 1.0555,
+      "step": 103
+    },
+    {
+      "epoch": 0.642967542503864,
+      "grad_norm": 2.793840699546338,
+      "learning_rate": 7.980095660501473e-05,
+      "loss": 1.0705,
+      "step": 104
+    },
+    {
+      "epoch": 0.6491499227202473,
+      "grad_norm": 3.0934299956993754,
+      "learning_rate": 7.978328820522204e-05,
+      "loss": 1.0593,
+      "step": 105
+    },
+    {
+      "epoch": 0.6553323029366306,
+      "grad_norm": 2.2997166417283186,
+      "learning_rate": 7.976487073507839e-05,
+      "loss": 1.0485,
+      "step": 106
+    },
+    {
+      "epoch": 0.6615146831530139,
+      "grad_norm": 3.3687080136243903,
+      "learning_rate": 7.974570454136212e-05,
+      "loss": 1.0623,
+      "step": 107
+    },
+    {
+      "epoch": 0.6676970633693973,
+      "grad_norm": 2.668689204618569,
+      "learning_rate": 7.972578998494903e-05,
+      "loss": 1.06,
+      "step": 108
+    },
+    {
+      "epoch": 0.6738794435857806,
+      "grad_norm": 3.259938106278256,
+      "learning_rate": 7.970512744080571e-05,
+      "loss": 1.0646,
+      "step": 109
+    },
+    {
+      "epoch": 0.6800618238021638,
+      "grad_norm": 2.539712415638497,
+      "learning_rate": 7.968371729798246e-05,
+      "loss": 1.0578,
+      "step": 110
+    },
+    {
+      "epoch": 0.6862442040185471,
+      "grad_norm": 2.875087869638967,
+      "learning_rate": 7.96615599596059e-05,
+      "loss": 1.0495,
+      "step": 111
+    },
+    {
+      "epoch": 0.6924265842349304,
+      "grad_norm": 2.171964418083895,
+      "learning_rate": 7.963865584287142e-05,
+      "loss": 1.0482,
+      "step": 112
+    },
+    {
+      "epoch": 0.6986089644513137,
+      "grad_norm": 2.4445515677858776,
+      "learning_rate": 7.96150053790354e-05,
+      "loss": 1.0478,
+      "step": 113
+    },
+    {
+      "epoch": 0.7047913446676971,
+      "grad_norm": 1.4736364847794041,
+      "learning_rate": 7.959060901340691e-05,
+      "loss": 1.0644,
+      "step": 114
+    },
+    {
+      "epoch": 0.7109737248840804,
+      "grad_norm": 3.2793219071142574,
+      "learning_rate": 7.956546720533952e-05,
+      "loss": 1.0581,
+      "step": 115
+    },
+    {
+      "epoch": 0.7171561051004637,
+      "grad_norm": 2.1935733687249157,
+      "learning_rate": 7.953958042822252e-05,
+      "loss": 1.0577,
+      "step": 116
+    },
+    {
+      "epoch": 0.723338485316847,
+      "grad_norm": 1.9736309619709513,
+      "learning_rate": 7.951294916947206e-05,
+      "loss": 1.0562,
+      "step": 117
+    },
+    {
+      "epoch": 0.7295208655332303,
+      "grad_norm": 2.1690441232660724,
+      "learning_rate": 7.948557393052195e-05,
+      "loss": 1.052,
+      "step": 118
+    },
+    {
+      "epoch": 0.7357032457496137,
+      "grad_norm": 2.703798771440302,
+      "learning_rate": 7.945745522681429e-05,
+      "loss": 1.0645,
+      "step": 119
+    },
+    {
+      "epoch": 0.7418856259659969,
+      "grad_norm": 2.6860226290594427,
+      "learning_rate": 7.942859358778961e-05,
+      "loss": 1.0469,
+      "step": 120
+    },
+    {
+      "epoch": 0.7480680061823802,
+      "grad_norm": 1.2516089611393217,
+      "learning_rate": 7.939898955687709e-05,
+      "loss": 1.0442,
+      "step": 121
+    },
+    {
+      "epoch": 0.7542503863987635,
+      "grad_norm": 3.4267893638027878,
+      "learning_rate": 7.936864369148418e-05,
+      "loss": 1.0608,
+      "step": 122
+    },
+    {
+      "epoch": 0.7604327666151468,
+      "grad_norm": 2.225834600604942,
+      "learning_rate": 7.933755656298616e-05,
+      "loss": 1.0668,
+      "step": 123
+    },
+    {
+      "epoch": 0.7666151468315301,
+      "grad_norm": 1.6048920414465118,
+      "learning_rate": 7.930572875671542e-05,
+      "loss": 1.046,
+      "step": 124
+    },
+    {
+      "epoch": 0.7727975270479135,
+      "grad_norm": 3.4705093098033957,
+      "learning_rate": 7.927316087195035e-05,
+      "loss": 1.0703,
+      "step": 125
+    },
+    {
+      "epoch": 0.7789799072642968,
+      "grad_norm": 2.280254489129258,
+      "learning_rate": 7.923985352190416e-05,
+      "loss": 1.0632,
+      "step": 126
+    },
+    {
+      "epoch": 0.7851622874806801,
+      "grad_norm": 2.8077251065232884,
+      "learning_rate": 7.920580733371325e-05,
+      "loss": 1.0637,
+      "step": 127
+    },
+    {
+      "epoch": 0.7913446676970634,
+      "grad_norm": 2.79897007061372,
+      "learning_rate": 7.917102294842545e-05,
+      "loss": 1.0511,
+      "step": 128
+    },
+    {
+      "epoch": 0.7975270479134466,
+      "grad_norm": 2.356217446775361,
+      "learning_rate": 7.913550102098789e-05,
+      "loss": 1.0621,
+      "step": 129
+    },
+    {
+      "epoch": 0.80370942812983,
+      "grad_norm": 2.837553637508471,
+      "learning_rate": 7.909924222023478e-05,
+      "loss": 1.0498,
+      "step": 130
+    },
+    {
+      "epoch": 0.8098918083462133,
+      "grad_norm": 2.074863492575953,
+      "learning_rate": 7.906224722887468e-05,
+      "loss": 1.0555,
+      "step": 131
+    },
+    {
+      "epoch": 0.8160741885625966,
+      "grad_norm": 3.906632921125056,
+      "learning_rate": 7.902451674347778e-05,
+      "loss": 1.0547,
+      "step": 132
+    },
+    {
+      "epoch": 0.8222565687789799,
+      "grad_norm": 3.098700348501345,
+      "learning_rate": 7.898605147446262e-05,
+      "loss": 1.0459,
+      "step": 133
+    },
+    {
+      "epoch": 0.8284389489953632,
+      "grad_norm": 2.847790553655261,
+      "learning_rate": 7.894685214608293e-05,
+      "loss": 1.0442,
+      "step": 134
+    },
+    {
+      "epoch": 0.8346213292117465,
+      "grad_norm": 2.908201865083222,
+      "learning_rate": 7.89069194964138e-05,
+      "loss": 1.0553,
+      "step": 135
+    },
+    {
+      "epoch": 0.8408037094281299,
+      "grad_norm": 2.1751989062888146,
+      "learning_rate": 7.886625427733786e-05,
+      "loss": 1.0479,
+      "step": 136
+    },
+    {
+      "epoch": 0.8469860896445132,
+      "grad_norm": 1.729338903416391,
+      "learning_rate": 7.882485725453114e-05,
+      "loss": 1.0315,
+      "step": 137
+    },
+    {
+      "epoch": 0.8531684698608965,
+      "grad_norm": 2.2703882009096827,
+      "learning_rate": 7.878272920744862e-05,
+      "loss": 1.0496,
+      "step": 138
+    },
+    {
+      "epoch": 0.8593508500772797,
+      "grad_norm": 1.3426280628545364,
+      "learning_rate": 7.87398709293096e-05,
+      "loss": 1.0407,
+      "step": 139
+    },
+    {
+      "epoch": 0.865533230293663,
+      "grad_norm": 2.68728781786679,
+      "learning_rate": 7.869628322708269e-05,
+      "loss": 1.0551,
+      "step": 140
+    },
+    {
+      "epoch": 0.8717156105100463,
+      "grad_norm": 1.7262218207414903,
+      "learning_rate": 7.865196692147066e-05,
+      "loss": 1.0483,
+      "step": 141
+    },
+    {
+      "epoch": 0.8778979907264297,
+      "grad_norm": 2.5755795419575156,
+      "learning_rate": 7.860692284689502e-05,
+      "loss": 1.0657,
+      "step": 142
+    },
+    {
+      "epoch": 0.884080370942813,
+      "grad_norm": 2.1034039247877856,
+      "learning_rate": 7.856115185148025e-05,
+      "loss": 1.056,
+      "step": 143
+    },
+    {
+      "epoch": 0.8902627511591963,
+      "grad_norm": 1.9719606878351836,
+      "learning_rate": 7.851465479703785e-05,
+      "loss": 1.0549,
+      "step": 144
+    },
+    {
+      "epoch": 0.8964451313755796,
+      "grad_norm": 2.066417220460305,
+      "learning_rate": 7.846743255905014e-05,
+      "loss": 1.0483,
+      "step": 145
+    },
+    {
+      "epoch": 0.9026275115919629,
+      "grad_norm": 1.7106893433354307,
+      "learning_rate": 7.841948602665373e-05,
+      "loss": 1.0423,
+      "step": 146
+    },
+    {
+      "epoch": 0.9088098918083463,
+      "grad_norm": 2.4989235217550676,
+      "learning_rate": 7.83708161026228e-05,
+      "loss": 1.0486,
+      "step": 147
+    },
+    {
+      "epoch": 0.9149922720247295,
+      "grad_norm": 2.354839754481779,
+      "learning_rate": 7.832142370335215e-05,
+      "loss": 1.0445,
+      "step": 148
+    },
+    {
+      "epoch": 0.9211746522411128,
+      "grad_norm": 1.702601699332555,
+      "learning_rate": 7.827130975883982e-05,
+      "loss": 1.0206,
+      "step": 149
+    },
+    {
+      "epoch": 0.9273570324574961,
+      "grad_norm": 3.284158052097331,
+      "learning_rate": 7.822047521266973e-05,
+      "loss": 1.028,
+      "step": 150
+    },
+    {
+      "epoch": 0.9335394126738794,
+      "grad_norm": 2.5496411598803648,
+      "learning_rate": 7.816892102199382e-05,
+      "loss": 1.0367,
+      "step": 151
+    },
+    {
+      "epoch": 0.9397217928902627,
+      "grad_norm": 1.842181874139785,
+      "learning_rate": 7.811664815751404e-05,
+      "loss": 1.0415,
+      "step": 152
+    },
+    {
+      "epoch": 0.9459041731066461,
+      "grad_norm": 3.2108389564617887,
+      "learning_rate": 7.806365760346409e-05,
+      "loss": 1.0379,
+      "step": 153
+    },
+    {
+      "epoch": 0.9520865533230294,
+      "grad_norm": 1.8227204697842778,
+      "learning_rate": 7.800995035759087e-05,
+      "loss": 1.0498,
+      "step": 154
+    },
+    {
+      "epoch": 0.9582689335394127,
+      "grad_norm": 3.4947702832902476,
+      "learning_rate": 7.795552743113573e-05,
+      "loss": 1.0479,
+      "step": 155
+    },
+    {
+      "epoch": 0.964451313755796,
+      "grad_norm": 1.8114431864984537,
+      "learning_rate": 7.79003898488154e-05,
+      "loss": 1.0313,
+      "step": 156
+    },
+    {
+      "epoch": 0.9706336939721792,
+      "grad_norm": 3.535447837519456,
+      "learning_rate": 7.784453864880267e-05,
+      "loss": 1.0474,
+      "step": 157
+    },
+    {
+      "epoch": 0.9768160741885626,
+      "grad_norm": 1.7365198353975386,
+      "learning_rate": 7.77879748827069e-05,
+      "loss": 1.0511,
+      "step": 158
+    },
+    {
+      "epoch": 0.9829984544049459,
+      "grad_norm": 3.594055596899371,
+      "learning_rate": 7.77306996155542e-05,
+      "loss": 1.0603,
+      "step": 159
+    },
+    {
+      "epoch": 0.9891808346213292,
+      "grad_norm": 2.5313017940876086,
+      "learning_rate": 7.767271392576732e-05,
+      "loss": 1.0626,
+      "step": 160
+    },
+    {
+      "epoch": 0.9953632148377125,
+      "grad_norm": 3.40567680255142,
+      "learning_rate": 7.761401890514547e-05,
+      "loss": 1.0638,
+      "step": 161
+    },
+    {
+      "epoch": 1.0046367851622875,
+      "grad_norm": 4.5174674178938945,
+      "learning_rate": 7.755461565884362e-05,
+      "loss": 1.8415,
+      "step": 162
+    },
+    {
+      "epoch": 1.010819165378671,
+      "grad_norm": 3.0038968679186957,
+      "learning_rate": 7.749450530535179e-05,
+      "loss": 1.0416,
+      "step": 163
+    },
+    {
+      "epoch": 1.017001545595054,
+      "grad_norm": 2.217689466421883,
+      "learning_rate": 7.743368897647395e-05,
+      "loss": 1.0335,
+      "step": 164
+    },
+    {
+      "epoch": 1.0231839258114375,
+      "grad_norm": 3.387468691972088,
+      "learning_rate": 7.737216781730673e-05,
+      "loss": 1.0285,
+      "step": 165
+    },
+    {
+      "epoch": 1.0293663060278206,
+      "grad_norm": 2.9834708542560433,
+      "learning_rate": 7.730994298621781e-05,
+      "loss": 1.0462,
+      "step": 166
+    },
+    {
+      "epoch": 1.035548686244204,
+      "grad_norm": 2.33967613172349,
+      "learning_rate": 7.72470156548242e-05,
+      "loss": 1.0315,
+      "step": 167
+    },
+    {
+      "epoch": 1.0417310664605872,
+      "grad_norm": 1.98032140135135,
+      "learning_rate": 7.718338700797007e-05,
+      "loss": 1.021,
+      "step": 168
+    },
+    {
+      "epoch": 1.0479134466769706,
+      "grad_norm": 1.8602974186471763,
+      "learning_rate": 7.711905824370457e-05,
+      "loss": 1.0253,
+      "step": 169
+    },
+    {
+      "epoch": 1.054095826893354,
+      "grad_norm": 1.571944862888458,
+      "learning_rate": 7.70540305732591e-05,
+      "loss": 1.0096,
+      "step": 170
+    },
+    {
+      "epoch": 1.0602782071097372,
+      "grad_norm": 1.2148625487373597,
+      "learning_rate": 7.69883052210247e-05,
+      "loss": 1.0156,
+      "step": 171
+    },
+    {
+      "epoch": 1.0664605873261206,
+      "grad_norm": 2.1277306224703865,
+      "learning_rate": 7.692188342452885e-05,
+      "loss": 1.0201,
+      "step": 172
+    },
+    {
+      "epoch": 1.0726429675425038,
+      "grad_norm": 1.92583372296243,
+      "learning_rate": 7.685476643441219e-05,
+      "loss": 1.0207,
+      "step": 173
+    },
+    {
+      "epoch": 1.0788253477588872,
+      "grad_norm": 1.8133100096935075,
+      "learning_rate": 7.678695551440506e-05,
+      "loss": 1.0181,
+      "step": 174
+    },
+    {
+      "epoch": 1.0850077279752706,
+      "grad_norm": 1.4080085200242758,
+      "learning_rate": 7.671845194130363e-05,
+      "loss": 1.0219,
+      "step": 175
+    },
+    {
+      "epoch": 1.0911901081916537,
+      "grad_norm": 2.2832079127892326,
+      "learning_rate": 7.664925700494585e-05,
+      "loss": 1.0199,
+      "step": 176
+    },
+    {
+      "epoch": 1.0973724884080371,
+      "grad_norm": 2.150797505157169,
+      "learning_rate": 7.657937200818722e-05,
+      "loss": 1.0173,
+      "step": 177
+    },
+    {
+      "epoch": 1.1035548686244203,
+      "grad_norm": 0.8849916762716471,
+      "learning_rate": 7.65087982668762e-05,
+      "loss": 1.005,
+      "step": 178
+    },
+    {
+      "epoch": 1.1097372488408037,
+      "grad_norm": 3.229134861969822,
+      "learning_rate": 7.643753710982949e-05,
+      "loss": 1.0133,
+      "step": 179
+    },
+    {
+      "epoch": 1.1159196290571871,
+      "grad_norm": 1.7293989048925642,
+      "learning_rate": 7.636558987880694e-05,
+      "loss": 1.0353,
+      "step": 180
+    },
+    {
+      "epoch": 1.1221020092735703,
+      "grad_norm": 2.7325244810866747,
+      "learning_rate": 7.629295792848639e-05,
+      "loss": 1.0188,
+      "step": 181
+    },
+    {
+      "epoch": 1.1282843894899537,
+      "grad_norm": 1.9813373605063669,
+      "learning_rate": 7.621964262643805e-05,
+      "loss": 1.0331,
+      "step": 182
+    },
+    {
+      "epoch": 1.1344667697063369,
+      "grad_norm": 1.9157662111283729,
+      "learning_rate": 7.614564535309882e-05,
+      "loss": 0.9959,
+      "step": 183
+    },
+    {
+      "epoch": 1.1406491499227203,
+      "grad_norm": 2.0892926569694685,
+      "learning_rate": 7.607096750174629e-05,
+      "loss": 1.0097,
+      "step": 184
+    },
+    {
+      "epoch": 1.1468315301391034,
+      "grad_norm": 1.821781934019597,
+      "learning_rate": 7.599561047847247e-05,
+      "loss": 1.0213,
+      "step": 185
+    },
+    {
+      "epoch": 1.1530139103554868,
+      "grad_norm": 1.8556200827944644,
+      "learning_rate": 7.591957570215735e-05,
+      "loss": 1.0084,
+      "step": 186
+    },
+    {
+      "epoch": 1.1591962905718702,
+      "grad_norm": 1.3594082674263888,
+      "learning_rate": 7.584286460444222e-05,
+      "loss": 1.0044,
+      "step": 187
+    },
+    {
+      "epoch": 1.1653786707882534,
+      "grad_norm": 2.2744478173451594,
+      "learning_rate": 7.576547862970261e-05,
+      "loss": 0.993,
+      "step": 188
+    },
+    {
+      "epoch": 1.1715610510046368,
+      "grad_norm": 1.3757650290844639,
+      "learning_rate": 7.568741923502118e-05,
+      "loss": 1.0055,
+      "step": 189
+    },
+    {
+      "epoch": 1.1777434312210202,
+      "grad_norm": 1.854325739268096,
+      "learning_rate": 7.560868789016024e-05,
+      "loss": 1.0125,
+      "step": 190
+    },
+    {
+      "epoch": 1.1839258114374034,
+      "grad_norm": 1.5111560222547549,
+      "learning_rate": 7.552928607753414e-05,
+      "loss": 1.0005,
+      "step": 191
+    },
+    {
+      "epoch": 1.1901081916537868,
+      "grad_norm": 1.811472101825589,
+      "learning_rate": 7.544921529218125e-05,
+      "loss": 1.004,
+      "step": 192
+    },
+    {
+      "epoch": 1.19629057187017,
+      "grad_norm": 1.4223817808791495,
+      "learning_rate": 7.536847704173593e-05,
+      "loss": 1.0014,
+      "step": 193
+    },
+    {
+      "epoch": 1.2024729520865534,
+      "grad_norm": 1.9843008102793314,
+      "learning_rate": 7.528707284640004e-05,
+      "loss": 1.0142,
+      "step": 194
+    },
+    {
+      "epoch": 1.2086553323029365,
+      "grad_norm": 1.5650426744555666,
+      "learning_rate": 7.520500423891442e-05,
+      "loss": 1.0053,
+      "step": 195
+    },
+    {
+      "epoch": 1.21483771251932,
+      "grad_norm": 1.5008249395490572,
+      "learning_rate": 7.512227276452989e-05,
+      "loss": 1.0001,
+      "step": 196
+    },
+    {
+      "epoch": 1.2210200927357033,
+      "grad_norm": 1.8158622037471164,
+      "learning_rate": 7.503887998097833e-05,
+      "loss": 1.0046,
+      "step": 197
+    },
+    {
+      "epoch": 1.2272024729520865,
+      "grad_norm": 1.594408802837267,
+      "learning_rate": 7.495482745844317e-05,
+      "loss": 0.9926,
+      "step": 198
+    },
+    {
+      "epoch": 1.23338485316847,
+      "grad_norm": 1.6151725290749885,
+      "learning_rate": 7.487011677953e-05,
+      "loss": 1.0135,
+      "step": 199
+    },
+    {
+      "epoch": 1.239567233384853,
+      "grad_norm": 1.6326603710438912,
+      "learning_rate": 7.478474953923662e-05,
+      "loss": 1.0011,
+      "step": 200
+    },
+    {
+      "epoch": 1.2457496136012365,
+      "grad_norm": 1.8016191114443714,
+      "learning_rate": 7.469872734492308e-05,
+      "loss": 0.9952,
+      "step": 201
+    },
+    {
+      "epoch": 1.2519319938176197,
+      "grad_norm": 1.1900057306211553,
+      "learning_rate": 7.461205181628143e-05,
+      "loss": 0.9903,
+      "step": 202
+    },
+    {
+      "epoch": 1.258114374034003,
+      "grad_norm": 2.4753152288862763,
+      "learning_rate": 7.452472458530522e-05,
+      "loss": 1.0031,
+      "step": 203
+    },
+    {
+      "epoch": 1.2642967542503865,
+      "grad_norm": 1.156802595501098,
+      "learning_rate": 7.443674729625869e-05,
+      "loss": 1.0059,
+      "step": 204
+    },
+    {
+      "epoch": 1.2704791344667696,
+      "grad_norm": 1.41497847335521,
+      "learning_rate": 7.434812160564592e-05,
+      "loss": 1.0144,
+      "step": 205
+    },
+    {
+      "epoch": 1.276661514683153,
+      "grad_norm": 2.3844147534681155,
+      "learning_rate": 7.425884918217958e-05,
+      "loss": 1.0058,
+      "step": 206
+    },
+    {
+      "epoch": 1.2828438948995364,
+      "grad_norm": 1.3271435145317263,
+      "learning_rate": 7.416893170674954e-05,
+      "loss": 1.0165,
+      "step": 207
+    },
+    {
+      "epoch": 1.2890262751159196,
+      "grad_norm": 1.6975589849591926,
+      "learning_rate": 7.407837087239117e-05,
+      "loss": 0.99,
+      "step": 208
+    },
+    {
+      "epoch": 1.295208655332303,
+      "grad_norm": 1.8519388740645097,
+      "learning_rate": 7.398716838425352e-05,
+      "loss": 1.0089,
+      "step": 209
+    },
+    {
+      "epoch": 1.3013910355486862,
+      "grad_norm": 1.6947179785568054,
+      "learning_rate": 7.38953259595672e-05,
+      "loss": 0.9879,
+      "step": 210
+    },
+    {
+      "epoch": 1.3075734157650696,
+      "grad_norm": 1.3528548666192206,
+      "learning_rate": 7.380284532761202e-05,
+      "loss": 1.0024,
+      "step": 211
+    },
+    {
+      "epoch": 1.3137557959814528,
+      "grad_norm": 1.6098695630590318,
+      "learning_rate": 7.370972822968446e-05,
+      "loss": 0.9804,
+      "step": 212
+    },
+    {
+      "epoch": 1.3199381761978362,
+      "grad_norm": 1.5753154362328263,
+      "learning_rate": 7.361597641906486e-05,
+      "loss": 0.9905,
+      "step": 213
+    },
+    {
+      "epoch": 1.3261205564142196,
+      "grad_norm": 1.4448197507546439,
+      "learning_rate": 7.352159166098441e-05,
+      "loss": 0.9847,
+      "step": 214
+    },
+    {
+      "epoch": 1.3323029366306027,
+      "grad_norm": 2.188314280055232,
+      "learning_rate": 7.342657573259194e-05,
+      "loss": 0.9992,
+      "step": 215
+    },
+    {
+      "epoch": 1.3384853168469861,
+      "grad_norm": 1.2657997745237497,
+      "learning_rate": 7.333093042292044e-05,
+      "loss": 0.9904,
+      "step": 216
+    },
+    {
+      "epoch": 1.3446676970633695,
+      "grad_norm": 1.2714200260276027,
+      "learning_rate": 7.323465753285337e-05,
+      "loss": 1.0057,
+      "step": 217
+    },
+    {
+      "epoch": 1.3508500772797527,
+      "grad_norm": 1.3287688093801202,
+      "learning_rate": 7.313775887509075e-05,
+      "loss": 1.0102,
+      "step": 218
+    },
+    {
+      "epoch": 1.3570324574961359,
+      "grad_norm": 3.081273929769149,
+      "learning_rate": 7.304023627411505e-05,
+      "loss": 1.0116,
+      "step": 219
+    },
+    {
+      "epoch": 1.3632148377125193,
+      "grad_norm": 1.767890752318478,
+      "learning_rate": 7.29420915661568e-05,
+      "loss": 0.9999,
+      "step": 220
+    },
+    {
+      "epoch": 1.3693972179289027,
+      "grad_norm": 3.3832241258249143,
+      "learning_rate": 7.284332659916006e-05,
+      "loss": 1.0146,
+      "step": 221
+    },
+    {
+      "epoch": 1.3755795981452859,
+      "grad_norm": 2.792347707620742,
+      "learning_rate": 7.274394323274761e-05,
+      "loss": 1.038,
+      "step": 222
+    },
+    {
+      "epoch": 1.3817619783616693,
+      "grad_norm": 2.717608779519406,
+      "learning_rate": 7.26439433381859e-05,
+      "loss": 0.9968,
+      "step": 223
+    },
+    {
+      "epoch": 1.3879443585780527,
+      "grad_norm": 2.6449000513172023,
+      "learning_rate": 7.254332879834986e-05,
+      "loss": 1.0202,
+      "step": 224
+    },
+    {
+      "epoch": 1.3941267387944358,
+      "grad_norm": 2.5816179353247386,
+      "learning_rate": 7.244210150768745e-05,
+      "loss": 1.0126,
+      "step": 225
+    },
+    {
+      "epoch": 1.4003091190108192,
+      "grad_norm": 1.8819265089556003,
+      "learning_rate": 7.234026337218395e-05,
+      "loss": 1.0044,
+      "step": 226
+    },
+    {
+      "epoch": 1.4064914992272024,
+      "grad_norm": 2.778434728345896,
+      "learning_rate": 7.223781630932612e-05,
+      "loss": 1.0112,
+      "step": 227
+    },
+    {
+      "epoch": 1.4126738794435858,
+      "grad_norm": 2.3953953560776164,
+      "learning_rate": 7.213476224806604e-05,
+      "loss": 1.007,
+      "step": 228
+    },
+    {
+      "epoch": 1.418856259659969,
+      "grad_norm": 2.1884861738889554,
+      "learning_rate": 7.203110312878487e-05,
+      "loss": 0.9997,
+      "step": 229
+    },
+    {
+      "epoch": 1.4250386398763524,
+      "grad_norm": 1.9108092994214398,
+      "learning_rate": 7.192684090325621e-05,
+      "loss": 0.9961,
+      "step": 230
+    },
+    {
+      "epoch": 1.4312210200927358,
+      "grad_norm": 2.070150415301729,
+      "learning_rate": 7.182197753460947e-05,
+      "loss": 0.9928,
+      "step": 231
+    },
+    {
+      "epoch": 1.437403400309119,
+      "grad_norm": 1.2896742775847363,
+      "learning_rate": 7.171651499729278e-05,
+      "loss": 0.9843,
+      "step": 232
+    },
+    {
+      "epoch": 1.4435857805255023,
+      "grad_norm": 2.4892589479787914,
+      "learning_rate": 7.161045527703593e-05,
+      "loss": 0.997,
+      "step": 233
+    },
+    {
+      "epoch": 1.4497681607418857,
+      "grad_norm": 1.9172236222391295,
+      "learning_rate": 7.150380037081293e-05,
+      "loss": 0.9986,
+      "step": 234
+    },
+    {
+      "epoch": 1.455950540958269,
+      "grad_norm": 2.2862715410270282,
+      "learning_rate": 7.139655228680438e-05,
+      "loss": 1.0075,
+      "step": 235
+    },
+    {
+      "epoch": 1.4621329211746523,
+      "grad_norm": 1.8213511562680735,
+      "learning_rate": 7.128871304435968e-05,
+      "loss": 1.0109,
+      "step": 236
+    },
+    {
+      "epoch": 1.4683153013910355,
+      "grad_norm": 2.1357271598290186,
+      "learning_rate": 7.118028467395904e-05,
+      "loss": 0.9899,
+      "step": 237
+    },
+    {
+      "epoch": 1.474497681607419,
+      "grad_norm": 1.51723941617764,
+      "learning_rate": 7.107126921717526e-05,
+      "loss": 0.999,
+      "step": 238
+    },
+    {
+      "epoch": 1.480680061823802,
+      "grad_norm": 2.216906523864941,
+      "learning_rate": 7.096166872663516e-05,
+      "loss": 0.9908,
+      "step": 239
+    },
+    {
+      "epoch": 1.4868624420401855,
+      "grad_norm": 1.6475567811924245,
+      "learning_rate": 7.085148526598109e-05,
+      "loss": 0.9846,
+      "step": 240
+    },
+    {
+      "epoch": 1.4930448222565689,
+      "grad_norm": 1.8561898420792513,
+      "learning_rate": 7.074072090983201e-05,
+      "loss": 0.9885,
+      "step": 241
+    },
+    {
+      "epoch": 1.499227202472952,
+      "grad_norm": 1.4076817648604656,
+      "learning_rate": 7.062937774374444e-05,
+      "loss": 0.9827,
+      "step": 242
+    },
+    {
+      "epoch": 1.5054095826893354,
+      "grad_norm": 2.043899369455665,
+      "learning_rate": 7.051745786417309e-05,
+      "loss": 1.0022,
+      "step": 243
+    },
+    {
+      "epoch": 1.5115919629057188,
+      "grad_norm": 1.7025584839030323,
+      "learning_rate": 7.04049633784316e-05,
+      "loss": 0.9935,
+      "step": 244
+    },
+    {
+      "epoch": 1.517774343122102,
+      "grad_norm": 1.8180249630171987,
+      "learning_rate": 7.029189640465266e-05,
+      "loss": 0.9948,
+      "step": 245
+    },
+    {
+      "epoch": 1.5239567233384852,
+      "grad_norm": 1.617153539919805,
+      "learning_rate": 7.017825907174819e-05,
+      "loss": 0.9946,
+      "step": 246
+    },
+    {
+      "epoch": 1.5301391035548686,
+      "grad_norm": 1.9584695599096857,
+      "learning_rate": 7.006405351936935e-05,
+      "loss": 0.981,
+      "step": 247
+    },
+    {
+      "epoch": 1.536321483771252,
+      "grad_norm": 1.5120141733299006,
+      "learning_rate": 6.99492818978661e-05,
+      "loss": 0.9957,
+      "step": 248
+    },
+    {
+      "epoch": 1.5425038639876352,
+      "grad_norm": 1.7224427233243673,
+      "learning_rate": 6.983394636824681e-05,
+      "loss": 0.985,
+      "step": 249
+    },
+    {
+      "epoch": 1.5486862442040186,
+      "grad_norm": 1.352522335955625,
+      "learning_rate": 6.971804910213753e-05,
+      "loss": 0.9947,
+      "step": 250
+    },
+    {
+      "epoch": 1.554868624420402,
+      "grad_norm": 1.9649504513247107,
+      "learning_rate": 6.960159228174118e-05,
+      "loss": 0.9802,
+      "step": 251
+    },
+    {
+      "epoch": 1.5610510046367851,
+      "grad_norm": 1.400091397431256,
+      "learning_rate": 6.94845780997963e-05,
+      "loss": 0.9869,
+      "step": 252
+    },
+    {
+      "epoch": 1.5672333848531683,
+      "grad_norm": 1.809258707764459,
+      "learning_rate": 6.936700875953593e-05,
+      "loss": 0.9951,
+      "step": 253
+    },
+    {
+      "epoch": 1.573415765069552,
+      "grad_norm": 1.544914399926865,
+      "learning_rate": 6.924888647464606e-05,
+      "loss": 0.9866,
+      "step": 254
+    },
+    {
+      "epoch": 1.5795981452859351,
+      "grad_norm": 1.6678917783004827,
+      "learning_rate": 6.91302134692239e-05,
+      "loss": 0.9839,
+      "step": 255
+    },
+    {
+      "epoch": 1.5857805255023183,
+      "grad_norm": 1.4327533793298342,
+      "learning_rate": 6.90109919777361e-05,
+      "loss": 0.9842,
+      "step": 256
+    },
+    {
+      "epoch": 1.5919629057187017,
+      "grad_norm": 1.4415951253324433,
+      "learning_rate": 6.889122424497659e-05,
+      "loss": 0.9888,
+      "step": 257
+    },
+    {
+      "epoch": 1.598145285935085,
+      "grad_norm": 1.305272535568148,
+      "learning_rate": 6.877091252602437e-05,
+      "loss": 0.9925,
+      "step": 258
+    },
+    {
+      "epoch": 1.6043276661514683,
+      "grad_norm": 2.4748168628602016,
+      "learning_rate": 6.865005908620102e-05,
+      "loss": 0.9862,
+      "step": 259
+    },
+    {
+      "epoch": 1.6105100463678517,
+      "grad_norm": 1.5958180723817172,
+      "learning_rate": 6.852866620102808e-05,
+      "loss": 1.0002,
+      "step": 260
+    },
+    {
+      "epoch": 1.616692426584235,
+      "grad_norm": 1.6536900940347805,
+      "learning_rate": 6.840673615618416e-05,
+      "loss": 0.9708,
+      "step": 261
+    },
+    {
+      "epoch": 1.6228748068006182,
+      "grad_norm": 2.6056238946624335,
+      "learning_rate": 6.828427124746191e-05,
+      "loss": 0.9896,
+      "step": 262
+    },
+    {
+      "epoch": 1.6290571870170014,
+      "grad_norm": 1.090301990739562,
+      "learning_rate": 6.816127378072484e-05,
+      "loss": 0.9909,
+      "step": 263
+    },
+    {
+      "epoch": 1.6352395672333848,
+      "grad_norm": 2.9921825238289768,
+      "learning_rate": 6.803774607186389e-05,
+      "loss": 1.014,
+      "step": 264
+    },
+    {
+      "epoch": 1.6414219474497682,
+      "grad_norm": 2.2996006999305822,
+      "learning_rate": 6.791369044675372e-05,
+      "loss": 1.0089,
+      "step": 265
+    },
+    {
+      "epoch": 1.6476043276661514,
+      "grad_norm": 2.472497102419325,
+      "learning_rate": 6.778910924120913e-05,
+      "loss": 0.9898,
+      "step": 266
+    },
+    {
+      "epoch": 1.6537867078825348,
+      "grad_norm": 1.4712820809315457,
+      "learning_rate": 6.766400480094084e-05,
+      "loss": 0.9982,
+      "step": 267
+    },
+    {
+      "epoch": 1.6599690880989182,
+      "grad_norm": 1.7900944541125188,
+      "learning_rate": 6.753837948151148e-05,
+      "loss": 1.0049,
+      "step": 268
+    },
+    {
+      "epoch": 1.6661514683153014,
+      "grad_norm": 1.5422419041247095,
+      "learning_rate": 6.74122356482912e-05,
+      "loss": 0.984,
+      "step": 269
+    },
+    {
+      "epoch": 1.6723338485316845,
+      "grad_norm": 1.7666863054997888,
+      "learning_rate": 6.728557567641313e-05,
+      "loss": 0.9982,
+      "step": 270
+    },
+    {
+      "epoch": 1.6785162287480682,
+      "grad_norm": 1.0188448221447397,
+      "learning_rate": 6.715840195072862e-05,
+      "loss": 0.9927,
+      "step": 271
+    },
+    {
+      "epoch": 1.6846986089644513,
+      "grad_norm": 1.4323614508510398,
+      "learning_rate": 6.703071686576243e-05,
+      "loss": 0.9933,
+      "step": 272
+    },
+    {
+      "epoch": 1.6908809891808345,
+      "grad_norm": 1.3733086668334178,
+      "learning_rate": 6.690252282566753e-05,
+      "loss": 0.98,
+      "step": 273
+    },
+    {
+      "epoch": 1.697063369397218,
+      "grad_norm": 1.7979930738646992,
+      "learning_rate": 6.677382224417991e-05,
+      "loss": 0.9915,
+      "step": 274
+    },
+    {
+      "epoch": 1.7032457496136013,
+      "grad_norm": 1.5018213587724738,
+      "learning_rate": 6.664461754457313e-05,
+      "loss": 0.9789,
+      "step": 275
+    },
+    {
+      "epoch": 1.7094281298299845,
+      "grad_norm": 0.8313817964266842,
+      "learning_rate": 6.651491115961264e-05,
+      "loss": 0.9898,
+      "step": 276
+    },
+    {
+      "epoch": 1.7156105100463679,
+      "grad_norm": 1.563766264104193,
+      "learning_rate": 6.638470553151003e-05,
+      "loss": 0.9839,
+      "step": 277
+    },
+    {
+      "epoch": 1.7217928902627513,
+      "grad_norm": 1.3394469302914662,
+      "learning_rate": 6.625400311187701e-05,
+      "loss": 0.9862,
+      "step": 278
+    },
+    {
+      "epoch": 1.7279752704791345,
+      "grad_norm": 0.9069310750596258,
+      "learning_rate": 6.61228063616793e-05,
+      "loss": 0.9772,
+      "step": 279
+    },
+    {
+      "epoch": 1.7341576506955176,
+      "grad_norm": 1.1937761300056076,
+      "learning_rate": 6.599111775119019e-05,
+      "loss": 0.9834,
+      "step": 280
+    },
+    {
+      "epoch": 1.740340030911901,
+      "grad_norm": 2.0959839523048722,
+      "learning_rate": 6.585893975994416e-05,
+      "loss": 0.9908,
+      "step": 281
+    },
+    {
+      "epoch": 1.7465224111282844,
+      "grad_norm": 1.4068229617185941,
+      "learning_rate": 6.572627487669005e-05,
+      "loss": 0.9857,
+      "step": 282
+    },
+    {
+      "epoch": 1.7527047913446676,
+      "grad_norm": 1.497838436021524,
+      "learning_rate": 6.55931255993444e-05,
+      "loss": 0.9982,
+      "step": 283
+    },
+    {
+      "epoch": 1.758887171561051,
+      "grad_norm": 1.2049062127509398,
+      "learning_rate": 6.545949443494416e-05,
+      "loss": 0.9816,
+      "step": 284
+    },
+    {
+      "epoch": 1.7650695517774344,
+      "grad_norm": 2.293459748858112,
+      "learning_rate": 6.53253838995997e-05,
+      "loss": 0.9939,
+      "step": 285
+    },
+    {
+      "epoch": 1.7712519319938176,
+      "grad_norm": 1.196771206420521,
+      "learning_rate": 6.519079651844733e-05,
+      "loss": 0.9917,
+      "step": 286
+    },
+    {
+      "epoch": 1.7774343122102008,
+      "grad_norm": 2.6719101899588464,
+      "learning_rate": 6.505573482560181e-05,
+      "loss": 1.0024,
+      "step": 287
+    },
+    {
+      "epoch": 1.7836166924265844,
+      "grad_norm": 2.151184032156713,
+      "learning_rate": 6.492020136410854e-05,
+      "loss": 1.0023,
+      "step": 288
+    },
+    {
+      "epoch": 1.7897990726429676,
+      "grad_norm": 2.049945153207459,
+      "learning_rate": 6.478419868589582e-05,
+      "loss": 0.9942,
+      "step": 289
+    },
+    {
+      "epoch": 1.7959814528593507,
+      "grad_norm": 1.7030760467943136,
+      "learning_rate": 6.464772935172666e-05,
+      "loss": 0.9972,
+      "step": 290
+    },
+    {
+      "epoch": 1.8021638330757341,
+      "grad_norm": 1.6124042099877869,
+      "learning_rate": 6.451079593115065e-05,
+      "loss": 0.9859,
+      "step": 291
+    },
+    {
+      "epoch": 1.8083462132921175,
+      "grad_norm": 1.7086055086974385,
+      "learning_rate": 6.437340100245558e-05,
+      "loss": 0.9875,
+      "step": 292
+    },
+    {
+      "epoch": 1.8145285935085007,
+      "grad_norm": 1.1804574569777302,
+      "learning_rate": 6.423554715261885e-05,
+      "loss": 0.9863,
+      "step": 293
+    },
+    {
+      "epoch": 1.820710973724884,
+      "grad_norm": 2.072000983521134,
+      "learning_rate": 6.409723697725876e-05,
+      "loss": 0.984,
+      "step": 294
+    },
+    {
+      "epoch": 1.8268933539412675,
+      "grad_norm": 1.5605771368628958,
+      "learning_rate": 6.395847308058569e-05,
+      "loss": 1.0016,
+      "step": 295
+    },
+    {
+      "epoch": 1.8330757341576507,
+      "grad_norm": 2.0470560137227944,
+      "learning_rate": 6.381925807535302e-05,
+      "loss": 0.9797,
+      "step": 296
+    },
+    {
+      "epoch": 1.8392581143740339,
+      "grad_norm": 1.5269600972745945,
+      "learning_rate": 6.367959458280795e-05,
+      "loss": 0.9952,
+      "step": 297
+    },
+    {
+      "epoch": 1.8454404945904173,
+      "grad_norm": 1.6911790008170018,
+      "learning_rate": 6.353948523264216e-05,
+      "loss": 0.9948,
+      "step": 298
+    },
+    {
+      "epoch": 1.8516228748068007,
+      "grad_norm": 1.5053272401335374,
+      "learning_rate": 6.339893266294223e-05,
+      "loss": 0.9851,
+      "step": 299
+    },
+    {
+      "epoch": 1.8578052550231838,
+      "grad_norm": 1.3443219469962477,
+      "learning_rate": 6.325793952014007e-05,
+      "loss": 0.9819,
+      "step": 300
+    },
+    {
+      "epoch": 1.8639876352395672,
+      "grad_norm": 1.3516905581796814,
+      "learning_rate": 6.311650845896303e-05,
+      "loss": 0.9957,
+      "step": 301
+    },
+    {
+      "epoch": 1.8701700154559506,
+      "grad_norm": 1.2720069275023291,
+      "learning_rate": 6.297464214238391e-05,
+      "loss": 0.9773,
+      "step": 302
+    },
+    {
+      "epoch": 1.8763523956723338,
+      "grad_norm": 1.281381485288164,
+      "learning_rate": 6.283234324157084e-05,
+      "loss": 0.9751,
+      "step": 303
+    },
+    {
+      "epoch": 1.8825347758887172,
+      "grad_norm": 1.109592460056133,
+      "learning_rate": 6.268961443583696e-05,
+      "loss": 0.9908,
+      "step": 304
+    },
+    {
+      "epoch": 1.8887171561051006,
+      "grad_norm": 2.4545190255394225,
+      "learning_rate": 6.254645841259005e-05,
+      "loss": 0.9889,
+      "step": 305
+    },
+    {
+      "epoch": 1.8948995363214838,
+      "grad_norm": 1.4595972385019333,
+      "learning_rate": 6.24028778672818e-05,
+      "loss": 0.987,
+      "step": 306
+    },
+    {
+      "epoch": 1.901081916537867,
+      "grad_norm": 3.0482536802039535,
+      "learning_rate": 6.225887550335715e-05,
+      "loss": 0.9898,
+      "step": 307
+    },
+    {
+      "epoch": 1.9072642967542504,
+      "grad_norm": 2.1361756083703503,
+      "learning_rate": 6.21144540322034e-05,
+      "loss": 0.9899,
+      "step": 308
+    },
+    {
+      "epoch": 1.9134466769706338,
+      "grad_norm": 2.820913172877751,
+      "learning_rate": 6.196961617309908e-05,
+      "loss": 0.9906,
+      "step": 309
+    },
+    {
+      "epoch": 1.919629057187017,
+      "grad_norm": 2.1277865542014536,
+      "learning_rate": 6.182436465316279e-05,
+      "loss": 0.9755,
+      "step": 310
+    },
+    {
+      "epoch": 1.9258114374034003,
+      "grad_norm": 2.3636934602350372,
+      "learning_rate": 6.16787022073019e-05,
+      "loss": 0.981,
+      "step": 311
+    },
+    {
+      "epoch": 1.9319938176197837,
+      "grad_norm": 1.5489010186147583,
+      "learning_rate": 6.153263157816098e-05,
+      "loss": 0.9714,
+      "step": 312
+    },
+    {
+      "epoch": 1.938176197836167,
+      "grad_norm": 2.4665556380164366,
+      "learning_rate": 6.138615551607017e-05,
+      "loss": 0.9852,
+      "step": 313
+    },
+    {
+      "epoch": 1.94435857805255,
+      "grad_norm": 1.769365451752886,
+      "learning_rate": 6.123927677899344e-05,
+      "loss": 0.9817,
+      "step": 314
+    },
+    {
+      "epoch": 1.9505409582689337,
+      "grad_norm": 2.276315268584059,
+      "learning_rate": 6.109199813247666e-05,
+      "loss": 0.98,
+      "step": 315
+    },
+    {
+      "epoch": 1.9567233384853169,
+      "grad_norm": 1.7457681230603967,
+      "learning_rate": 6.0944322349595426e-05,
+      "loss": 0.9943,
+      "step": 316
+    },
+    {
+      "epoch": 1.9629057187017,
+      "grad_norm": 1.897451826245397,
+      "learning_rate": 6.0796252210902986e-05,
+      "loss": 0.9707,
+      "step": 317
+    },
+    {
+      "epoch": 1.9690880989180835,
+      "grad_norm": 1.7758674602809486,
+      "learning_rate": 6.06477905043778e-05,
+      "loss": 0.986,
+      "step": 318
+    },
+    {
+      "epoch": 1.9752704791344669,
+      "grad_norm": 1.4801297268108475,
+      "learning_rate": 6.049894002537109e-05,
+      "loss": 0.9624,
+      "step": 319
+    },
+    {
+      "epoch": 1.98145285935085,
+      "grad_norm": 1.1171987537741563,
+      "learning_rate": 6.034970357655415e-05,
+      "loss": 0.9712,
+      "step": 320
+    },
+    {
+      "epoch": 1.9876352395672334,
+      "grad_norm": 1.5190947129294494,
+      "learning_rate": 6.020008396786562e-05,
+      "loss": 0.9823,
+      "step": 321
+    },
+    {
+      "epoch": 1.9938176197836168,
+      "grad_norm": 1.006947536865789,
+      "learning_rate": 6.00500840164586e-05,
+      "loss": 0.9771,
+      "step": 322
+    },
+    {
+      "epoch": 2.003091190108192,
+      "grad_norm": 4.006886737490512,
+      "learning_rate": 5.989970654664756e-05,
+      "loss": 1.6791,
+      "step": 323
+    },
+    {
+      "epoch": 2.009273570324575,
+      "grad_norm": 1.8558698221909793,
+      "learning_rate": 5.974895438985514e-05,
+      "loss": 0.952,
+      "step": 324
+    },
+    {
+      "epoch": 2.015455950540958,
+      "grad_norm": 1.455258316310302,
+      "learning_rate": 5.959783038455891e-05,
+      "loss": 0.9402,
+      "step": 325
+    },
+    {
+      "epoch": 2.021638330757342,
+      "grad_norm": 1.7859640511983568,
+      "learning_rate": 5.944633737623791e-05,
+      "loss": 0.9346,
+      "step": 326
+    },
+    {
+      "epoch": 2.027820710973725,
+      "grad_norm": 0.9891580259760341,
+      "learning_rate": 5.929447821731899e-05,
+      "loss": 0.9467,
+      "step": 327
+    },
+    {
+      "epoch": 2.034003091190108,
+      "grad_norm": 1.2780787887464713,
+      "learning_rate": 5.914225576712323e-05,
+      "loss": 0.9364,
+      "step": 328
+    },
+    {
+      "epoch": 2.0401854714064913,
+      "grad_norm": 1.747509422985778,
+      "learning_rate": 5.8989672891811987e-05,
+      "loss": 0.9337,
+      "step": 329
+    },
+    {
+      "epoch": 2.046367851622875,
+      "grad_norm": 1.2249047782739366,
+      "learning_rate": 5.883673246433302e-05,
+      "loss": 0.9498,
+      "step": 330
+    },
+    {
+      "epoch": 2.052550231839258,
+      "grad_norm": 2.003614583053326,
+      "learning_rate": 5.8683437364366316e-05,
+      "loss": 0.9535,
+      "step": 331
+    },
+    {
+      "epoch": 2.0587326120556413,
+      "grad_norm": 1.226972934974554,
+      "learning_rate": 5.852979047826996e-05,
+      "loss": 0.9329,
+      "step": 332
+    },
+    {
+      "epoch": 2.064914992272025,
+      "grad_norm": 2.172547322337205,
+      "learning_rate": 5.837579469902567e-05,
+      "loss": 0.9479,
+      "step": 333
+    },
+    {
+      "epoch": 2.071097372488408,
+      "grad_norm": 1.6402327078514485,
+      "learning_rate": 5.822145292618444e-05,
+      "loss": 0.9462,
+      "step": 334
+    },
+    {
+      "epoch": 2.0772797527047913,
+      "grad_norm": 1.6699665609409802,
+      "learning_rate": 5.806676806581192e-05,
+      "loss": 0.9431,
+      "step": 335
+    },
+    {
+      "epoch": 2.0834621329211744,
+      "grad_norm": 1.3197835509782132,
+      "learning_rate": 5.7911743030433596e-05,
+      "loss": 0.9557,
+      "step": 336
+    },
+    {
+      "epoch": 2.089644513137558,
+      "grad_norm": 1.1660288371337755,
+      "learning_rate": 5.77563807389801e-05,
+      "loss": 0.9356,
+      "step": 337
+    },
+    {
+      "epoch": 2.0958268933539412,
+      "grad_norm": 1.1228481962160324,
+      "learning_rate": 5.760068411673215e-05,
+      "loss": 0.9417,
+      "step": 338
+    },
+    {
+      "epoch": 2.1020092735703244,
+      "grad_norm": 1.372197582933058,
+      "learning_rate": 5.7444656095265506e-05,
+      "loss": 0.9269,
+      "step": 339
+    },
+    {
+      "epoch": 2.108191653786708,
+      "grad_norm": 1.0621673720020575,
+      "learning_rate": 5.7288299612395784e-05,
+      "loss": 0.9376,
+      "step": 340
+    },
+    {
+      "epoch": 2.114374034003091,
+      "grad_norm": 1.290738782393939,
+      "learning_rate": 5.713161761212309e-05,
+      "loss": 0.9484,
+      "step": 341
+    },
+    {
+      "epoch": 2.1205564142194744,
+      "grad_norm": 1.5545485725757495,
+      "learning_rate": 5.697461304457667e-05,
+      "loss": 0.9451,
+      "step": 342
+    },
+    {
+      "epoch": 2.126738794435858,
+      "grad_norm": 0.9984947527180189,
+      "learning_rate": 5.681728886595926e-05,
+      "loss": 0.9264,
+      "step": 343
+    },
+    {
+      "epoch": 2.132921174652241,
+      "grad_norm": 1.435440933520763,
+      "learning_rate": 5.665964803849152e-05,
+      "loss": 0.9576,
+      "step": 344
+    },
+    {
+      "epoch": 2.1391035548686244,
+      "grad_norm": 1.0914760893879398,
+      "learning_rate": 5.65016935303562e-05,
+      "loss": 0.9341,
+      "step": 345
+    },
+    {
+      "epoch": 2.1452859350850075,
+      "grad_norm": 0.9611538837750422,
+      "learning_rate": 5.634342831564228e-05,
+      "loss": 0.9408,
+      "step": 346
+    },
+    {
+      "epoch": 2.151468315301391,
+      "grad_norm": 1.1965180402340152,
+      "learning_rate": 5.6184855374288946e-05,
+      "loss": 0.9415,
+      "step": 347
+    },
+    {
+      "epoch": 2.1576506955177743,
+      "grad_norm": 1.0359352454914963,
+      "learning_rate": 5.602597769202953e-05,
+      "loss": 0.9506,
+      "step": 348
+    },
+    {
+      "epoch": 2.1638330757341575,
+      "grad_norm": 1.8769371801728494,
+      "learning_rate": 5.586679826033521e-05,
+      "loss": 0.9369,
+      "step": 349
+    },
+    {
+      "epoch": 2.170015455950541,
+      "grad_norm": 0.9836827672447646,
+      "learning_rate": 5.5707320076358776e-05,
+      "loss": 0.9452,
+      "step": 350
+    },
+    {
+      "epoch": 2.1761978361669243,
+      "grad_norm": 1.32097187927076,
+      "learning_rate": 5.5547546142878164e-05,
+      "loss": 0.9464,
+      "step": 351
+    },
+    {
+      "epoch": 2.1823802163833075,
+      "grad_norm": 1.3411663056222534,
+      "learning_rate": 5.538747946823987e-05,
+      "loss": 0.9321,
+      "step": 352
+    },
+    {
+      "epoch": 2.1885625965996907,
+      "grad_norm": 1.3570047609173206,
+      "learning_rate": 5.522712306630236e-05,
+      "loss": 0.9373,
+      "step": 353
+    },
+    {
+      "epoch": 2.1947449768160743,
+      "grad_norm": 0.9761232794419572,
+      "learning_rate": 5.506647995637932e-05,
+      "loss": 0.9462,
+      "step": 354
+    },
+    {
+      "epoch": 2.2009273570324575,
+      "grad_norm": 1.430074471656494,
+      "learning_rate": 5.490555316318279e-05,
+      "loss": 0.9485,
+      "step": 355
+    },
+    {
+      "epoch": 2.2071097372488406,
+      "grad_norm": 1.059127482296428,
+      "learning_rate": 5.474434571676622e-05,
+      "loss": 0.9422,
+      "step": 356
+    },
+    {
+      "epoch": 2.2132921174652243,
+      "grad_norm": 1.8097837664747738,
+      "learning_rate": 5.4582860652467385e-05,
+      "loss": 0.938,
+      "step": 357
+    },
+    {
+      "epoch": 2.2194744976816074,
+      "grad_norm": 1.125902108018908,
+      "learning_rate": 5.4421101010851315e-05,
+      "loss": 0.9334,
+      "step": 358
+    },
+    {
+      "epoch": 2.2256568778979906,
+      "grad_norm": 1.8628860336453976,
+      "learning_rate": 5.425906983765295e-05,
+      "loss": 0.9462,
+      "step": 359
+    },
+    {
+      "epoch": 2.2318392581143742,
+      "grad_norm": 1.5709465886819862,
+      "learning_rate": 5.409677018371986e-05,
+      "loss": 0.9563,
+      "step": 360
+    },
+    {
+      "epoch": 2.2380216383307574,
+      "grad_norm": 1.5452591189158589,
+      "learning_rate": 5.393420510495476e-05,
+      "loss": 0.9542,
+      "step": 361
+    },
+    {
+      "epoch": 2.2442040185471406,
+      "grad_norm": 1.215273856313518,
+      "learning_rate": 5.377137766225799e-05,
+      "loss": 0.9401,
+      "step": 362
+    },
+    {
+      "epoch": 2.250386398763524,
+      "grad_norm": 1.2233276485983613,
+      "learning_rate": 5.36082909214699e-05,
+      "loss": 0.9448,
+      "step": 363
+    },
+    {
+      "epoch": 2.2565687789799074,
+      "grad_norm": 1.3147561966681955,
+      "learning_rate": 5.344494795331308e-05,
+      "loss": 0.9465,
+      "step": 364
+    },
+    {
+      "epoch": 2.2627511591962906,
+      "grad_norm": 1.4155564384468768,
+      "learning_rate": 5.328135183333456e-05,
+      "loss": 0.951,
+      "step": 365
+    },
+    {
+      "epoch": 2.2689335394126737,
+      "grad_norm": 0.6114114300419576,
+      "learning_rate": 5.311750564184795e-05,
+      "loss": 0.9396,
+      "step": 366
+    },
+    {
+      "epoch": 2.2751159196290573,
+      "grad_norm": 1.5798139980221138,
+      "learning_rate": 5.2953412463875364e-05,
+      "loss": 0.9506,
+      "step": 367
+    },
+    {
+      "epoch": 2.2812982998454405,
+      "grad_norm": 0.9988799351566259,
+      "learning_rate": 5.278907538908936e-05,
+      "loss": 0.9526,
+      "step": 368
+    },
+    {
+      "epoch": 2.2874806800618237,
+      "grad_norm": 1.05377706000738,
+      "learning_rate": 5.26244975117548e-05,
+      "loss": 0.942,
+      "step": 369
+    },
+    {
+      "epoch": 2.293663060278207,
+      "grad_norm": 1.5699703997946644,
+      "learning_rate": 5.2459681930670536e-05,
+      "loss": 0.9511,
+      "step": 370
+    },
+    {
+      "epoch": 2.2998454404945905,
+      "grad_norm": 0.9947848025404337,
+      "learning_rate": 5.2294631749111125e-05,
+      "loss": 0.9151,
+      "step": 371
+    },
+    {
+      "epoch": 2.3060278207109737,
+      "grad_norm": 1.1765237704089029,
+      "learning_rate": 5.2129350074768316e-05,
+      "loss": 0.9467,
+      "step": 372
+    },
+    {
+      "epoch": 2.312210200927357,
+      "grad_norm": 1.5204036418963962,
+      "learning_rate": 5.1963840019692616e-05,
+      "loss": 0.953,
+      "step": 373
+    },
+    {
+      "epoch": 2.3183925811437405,
+      "grad_norm": 0.9961157368195384,
+      "learning_rate": 5.1798104700234655e-05,
+      "loss": 0.9353,
+      "step": 374
+    },
+    {
+      "epoch": 2.3245749613601236,
+      "grad_norm": 1.2662517756577527,
+      "learning_rate": 5.16321472369865e-05,
+      "loss": 0.9313,
+      "step": 375
+    },
+    {
+      "epoch": 2.330757341576507,
+      "grad_norm": 1.0077333440409275,
+      "learning_rate": 5.146597075472293e-05,
+      "loss": 0.949,
+      "step": 376
+    },
+    {
+      "epoch": 2.3369397217928904,
+      "grad_norm": 1.1569892416058336,
+      "learning_rate": 5.129957838234255e-05,
+      "loss": 0.9421,
+      "step": 377
+    },
+    {
+      "epoch": 2.3431221020092736,
+      "grad_norm": 1.0327382337486777,
+      "learning_rate": 5.113297325280896e-05,
+      "loss": 0.9359,
+      "step": 378
+    },
+    {
+      "epoch": 2.349304482225657,
+      "grad_norm": 1.5413059311272268,
+      "learning_rate": 5.0966158503091673e-05,
+      "loss": 0.9308,
+      "step": 379
+    },
+    {
+      "epoch": 2.3554868624420404,
+      "grad_norm": 1.0041025469926124,
+      "learning_rate": 5.079913727410712e-05,
+      "loss": 0.9472,
+      "step": 380
+    },
+    {
+      "epoch": 2.3616692426584236,
+      "grad_norm": 1.0449355058196719,
+      "learning_rate": 5.063191271065945e-05,
+      "loss": 0.9507,
+      "step": 381
+    },
+    {
+      "epoch": 2.3678516228748068,
+      "grad_norm": 1.4231290368942768,
+      "learning_rate": 5.046448796138138e-05,
+      "loss": 0.9617,
+      "step": 382
+    },
+    {
+      "epoch": 2.37403400309119,
+      "grad_norm": 1.4686945371549769,
+      "learning_rate": 5.029686617867488e-05,
+      "loss": 0.9487,
+      "step": 383
+    },
+    {
+      "epoch": 2.3802163833075736,
+      "grad_norm": 0.7321343928729324,
+      "learning_rate": 5.012905051865179e-05,
+      "loss": 0.9134,
+      "step": 384
+    },
+    {
+      "epoch": 2.3863987635239567,
+      "grad_norm": 1.7871707923975175,
+      "learning_rate": 4.9961044141074454e-05,
+      "loss": 0.9454,
+      "step": 385
+    },
+    {
+      "epoch": 2.39258114374034,
+      "grad_norm": 0.8554156480675892,
+      "learning_rate": 4.979285020929617e-05,
+      "loss": 0.958,
+      "step": 386
+    },
+    {
+      "epoch": 2.398763523956723,
+      "grad_norm": 2.0222183736349177,
+      "learning_rate": 4.9624471890201656e-05,
+      "loss": 0.9398,
+      "step": 387
+    },
+    {
+      "epoch": 2.4049459041731067,
+      "grad_norm": 1.1083392384698756,
+      "learning_rate": 4.945591235414741e-05,
+      "loss": 0.9353,
+      "step": 388
+    },
+    {
+      "epoch": 2.41112828438949,
+      "grad_norm": 2.249237349775883,
+      "learning_rate": 4.9287174774902034e-05,
+      "loss": 0.9299,
+      "step": 389
+    },
+    {
+      "epoch": 2.417310664605873,
+      "grad_norm": 1.7547197652519158,
+      "learning_rate": 4.911826232958647e-05,
+      "loss": 0.9465,
+      "step": 390
+    },
+    {
+      "epoch": 2.4234930448222567,
+      "grad_norm": 1.816970919376788,
+      "learning_rate": 4.894917819861416e-05,
+      "loss": 0.9425,
+      "step": 391
+    },
+    {
+      "epoch": 2.42967542503864,
+      "grad_norm": 1.5766998688721199,
+      "learning_rate": 4.8779925565631166e-05,
+      "loss": 0.9572,
+      "step": 392
+    },
+    {
+      "epoch": 2.435857805255023,
+      "grad_norm": 1.6988141319269177,
+      "learning_rate": 4.8610507617456265e-05,
+      "loss": 0.9401,
+      "step": 393
+    },
+    {
+      "epoch": 2.4420401854714067,
+      "grad_norm": 1.4658219077550854,
+      "learning_rate": 4.8440927544020896e-05,
+      "loss": 0.9429,
+      "step": 394
+    },
+    {
+      "epoch": 2.44822256568779,
+      "grad_norm": 1.4942830281619395,
+      "learning_rate": 4.82711885383091e-05,
+      "loss": 0.9305,
+      "step": 395
+    },
+    {
+      "epoch": 2.454404945904173,
+      "grad_norm": 1.4911047290042807,
+      "learning_rate": 4.810129379629745e-05,
+      "loss": 0.9509,
+      "step": 396
+    },
+    {
+      "epoch": 2.4605873261205566,
+      "grad_norm": 1.0945238549084377,
+      "learning_rate": 4.793124651689484e-05,
+      "loss": 0.9319,
+      "step": 397
+    },
+    {
+      "epoch": 2.46676970633694,
+      "grad_norm": 1.6671980576283978,
+      "learning_rate": 4.7761049901882227e-05,
+      "loss": 0.939,
+      "step": 398
+    },
+    {
+      "epoch": 2.472952086553323,
+      "grad_norm": 0.92852786359375,
+      "learning_rate": 4.7590707155852406e-05,
+      "loss": 0.954,
+      "step": 399
+    },
+    {
+      "epoch": 2.479134466769706,
+      "grad_norm": 1.7875978184704922,
+      "learning_rate": 4.74202214861496e-05,
+      "loss": 0.9381,
+      "step": 400
+    },
+    {
+      "epoch": 2.48531684698609,
+      "grad_norm": 1.212284622798296,
+      "learning_rate": 4.7249596102809164e-05,
+      "loss": 0.9467,
+      "step": 401
+    },
+    {
+      "epoch": 2.491499227202473,
+      "grad_norm": 1.7168328470834615,
+      "learning_rate": 4.707883421849703e-05,
+      "loss": 0.9496,
+      "step": 402
+    },
+    {
+      "epoch": 2.497681607418856,
+      "grad_norm": 1.3407496449046934,
+      "learning_rate": 4.6907939048449314e-05,
+      "loss": 0.9486,
+      "step": 403
+    },
+    {
+      "epoch": 2.5038639876352393,
+      "grad_norm": 1.4996984373050763,
+      "learning_rate": 4.673691381041173e-05,
+      "loss": 0.9298,
+      "step": 404
+    },
+    {
+      "epoch": 2.510046367851623,
+      "grad_norm": 1.4309169783530684,
+      "learning_rate": 4.6565761724579e-05,
+      "loss": 0.9452,
+      "step": 405
+    },
+    {
+      "epoch": 2.516228748068006,
+      "grad_norm": 1.4369198232445761,
+      "learning_rate": 4.6394486013534265e-05,
+      "loss": 0.9274,
+      "step": 406
+    },
+    {
+      "epoch": 2.5224111282843893,
+      "grad_norm": 1.3376962293350674,
+      "learning_rate": 4.6223089902188336e-05,
+      "loss": 0.9321,
+      "step": 407
+    },
+    {
+      "epoch": 2.528593508500773,
+      "grad_norm": 1.3175000668496903,
+      "learning_rate": 4.605157661771904e-05,
+      "loss": 0.9385,
+      "step": 408
+    },
+    {
+      "epoch": 2.534775888717156,
+      "grad_norm": 1.6617422309422225,
+      "learning_rate": 4.5879949389510406e-05,
+      "loss": 0.9301,
+      "step": 409
+    },
+    {
+      "epoch": 2.5409582689335393,
+      "grad_norm": 1.3172794288702236,
+      "learning_rate": 4.570821144909193e-05,
+      "loss": 0.9423,
+      "step": 410
+    },
+    {
+      "epoch": 2.547140649149923,
+      "grad_norm": 1.6218116986054838,
+      "learning_rate": 4.553636603007761e-05,
+      "loss": 0.9309,
+      "step": 411
+    },
+    {
+      "epoch": 2.553323029366306,
+      "grad_norm": 1.282648230450059,
+      "learning_rate": 4.536441636810521e-05,
+      "loss": 0.9393,
+      "step": 412
+    },
+    {
+      "epoch": 2.5595054095826892,
+      "grad_norm": 1.7085653102331866,
+      "learning_rate": 4.519236570077523e-05,
+      "loss": 0.9387,
+      "step": 413
+    },
+    {
+      "epoch": 2.565687789799073,
+      "grad_norm": 1.181853985248212,
+      "learning_rate": 4.502021726758994e-05,
+      "loss": 0.9329,
+      "step": 414
+    },
+    {
+      "epoch": 2.571870170015456,
+      "grad_norm": 1.7534552716011547,
+      "learning_rate": 4.484797430989248e-05,
+      "loss": 0.9412,
+      "step": 415
+    },
+    {
+      "epoch": 2.578052550231839,
+      "grad_norm": 1.3292585743678849,
+      "learning_rate": 4.467564007080574e-05,
+      "loss": 0.9399,
+      "step": 416
+    },
+    {
+      "epoch": 2.584234930448223,
+      "grad_norm": 1.7630942395608926,
+      "learning_rate": 4.4503217795171326e-05,
+      "loss": 0.9303,
+      "step": 417
+    },
+    {
+      "epoch": 2.590417310664606,
+      "grad_norm": 1.5547379842247044,
+      "learning_rate": 4.433071072948846e-05,
+      "loss": 0.9404,
+      "step": 418
+    },
+    {
+      "epoch": 2.596599690880989,
+      "grad_norm": 1.55358461949987,
+      "learning_rate": 4.415812212185287e-05,
+      "loss": 0.9479,
+      "step": 419
+    },
+    {
+      "epoch": 2.6027820710973724,
+      "grad_norm": 1.3519799459619244,
+      "learning_rate": 4.398545522189562e-05,
+      "loss": 0.9381,
+      "step": 420
+    },
+    {
+      "epoch": 2.6089644513137555,
+      "grad_norm": 1.4063429925697228,
+      "learning_rate": 4.38127132807219e-05,
+      "loss": 0.9308,
+      "step": 421
+    },
+    {
+      "epoch": 2.615146831530139,
+      "grad_norm": 1.1100048648257177,
+      "learning_rate": 4.363989955084988e-05,
+      "loss": 0.9316,
+      "step": 422
+    },
+    {
+      "epoch": 2.6213292117465223,
+      "grad_norm": 1.5473918787261554,
+      "learning_rate": 4.3467017286149355e-05,
+      "loss": 0.9288,
+      "step": 423
+    },
+    {
+      "epoch": 2.6275115919629055,
+      "grad_norm": 1.1413947086858753,
+      "learning_rate": 4.32940697417806e-05,
+      "loss": 0.9281,
+      "step": 424
+    },
+    {
+      "epoch": 2.633693972179289,
+      "grad_norm": 1.4822907840670647,
+      "learning_rate": 4.3121060174133006e-05,
+      "loss": 0.9391,
+      "step": 425
+    },
+    {
+      "epoch": 2.6398763523956723,
+      "grad_norm": 1.1893158134844697,
+      "learning_rate": 4.294799184076379e-05,
+      "loss": 0.9358,
+      "step": 426
+    },
+    {
+      "epoch": 2.6460587326120555,
+      "grad_norm": 1.395540540534361,
+      "learning_rate": 4.277486800033664e-05,
+      "loss": 0.9344,
+      "step": 427
+    },
+    {
+      "epoch": 2.652241112828439,
+      "grad_norm": 1.1389047154561869,
+      "learning_rate": 4.260169191256041e-05,
+      "loss": 0.9121,
+      "step": 428
+    },
+    {
+      "epoch": 2.6584234930448223,
+      "grad_norm": 1.349515321074191,
+      "learning_rate": 4.242846683812768e-05,
+      "loss": 0.9424,
+      "step": 429
+    },
+    {
+      "epoch": 2.6646058732612055,
+      "grad_norm": 1.2170924594039827,
+      "learning_rate": 4.225519603865338e-05,
+      "loss": 0.9285,
+      "step": 430
+    },
+    {
+      "epoch": 2.670788253477589,
+      "grad_norm": 1.2637766040441283,
+      "learning_rate": 4.208188277661341e-05,
+      "loss": 0.9489,
+      "step": 431
+    },
+    {
+      "epoch": 2.6769706336939723,
+      "grad_norm": 0.9722140941647698,
+      "learning_rate": 4.190853031528317e-05,
+      "loss": 0.9042,
+      "step": 432
+    },
+    {
+      "epoch": 2.6831530139103554,
+      "grad_norm": 1.3212458332351424,
+      "learning_rate": 4.173514191867615e-05,
+      "loss": 0.9294,
+      "step": 433
+    },
+    {
+      "epoch": 2.689335394126739,
+      "grad_norm": 1.0216097339005061,
+      "learning_rate": 4.156172085148244e-05,
+      "loss": 0.9365,
+      "step": 434
+    },
+    {
+      "epoch": 2.6955177743431222,
+      "grad_norm": 1.4931636832422879,
+      "learning_rate": 4.138827037900728e-05,
+      "loss": 0.9453,
+      "step": 435
+    },
+    {
+      "epoch": 2.7017001545595054,
+      "grad_norm": 1.5106660907983307,
+      "learning_rate": 4.1214793767109607e-05,
+      "loss": 0.9488,
+      "step": 436
+    },
+    {
+      "epoch": 2.7078825347758886,
+      "grad_norm": 0.8429680654036087,
+      "learning_rate": 4.104129428214048e-05,
+      "loss": 0.9286,
+      "step": 437
+    },
+    {
+      "epoch": 2.7140649149922718,
+      "grad_norm": 1.1888732405027307,
+      "learning_rate": 4.0867775190881673e-05,
+      "loss": 0.9389,
+      "step": 438
+    },
+    {
+      "epoch": 2.7202472952086554,
+      "grad_norm": 0.8249720034944886,
+      "learning_rate": 4.0694239760484084e-05,
+      "loss": 0.9236,
+      "step": 439
+    },
+    {
+      "epoch": 2.7264296754250386,
+      "grad_norm": 0.8112023719901648,
+      "learning_rate": 4.05206912584063e-05,
+      "loss": 0.93,
+      "step": 440
+    },
+    {
+      "epoch": 2.7326120556414217,
+      "grad_norm": 0.8700099921141801,
+      "learning_rate": 4.0347132952352976e-05,
+      "loss": 0.9274,
+      "step": 441
+    },
+    {
+      "epoch": 2.7387944358578054,
+      "grad_norm": 0.7451536808284058,
+      "learning_rate": 4.0173568110213444e-05,
+      "loss": 0.9444,
+      "step": 442
+    },
+    {
+      "epoch": 2.7449768160741885,
+      "grad_norm": 0.7730600029414318,
+      "learning_rate": 4e-05,
+      "loss": 0.9264,
+      "step": 443
+    },
+    {
+      "epoch": 2.7511591962905717,
+      "grad_norm": 0.64992428377988,
+      "learning_rate": 3.9826431889786576e-05,
+      "loss": 0.9228,
+      "step": 444
+    },
+    {
+      "epoch": 2.7573415765069553,
+      "grad_norm": 0.6092471954250116,
+      "learning_rate": 3.965286704764702e-05,
+      "loss": 0.9386,
+      "step": 445
+    },
+    {
+      "epoch": 2.7635239567233385,
+      "grad_norm": 0.5829783913867378,
+      "learning_rate": 3.947930874159371e-05,
+      "loss": 0.9466,
+      "step": 446
+    },
+    {
+      "epoch": 2.7697063369397217,
+      "grad_norm": 0.5791113943292933,
+      "learning_rate": 3.930576023951593e-05,
+      "loss": 0.9315,
+      "step": 447
+    },
+    {
+      "epoch": 2.7758887171561053,
+      "grad_norm": 0.496929341909277,
+      "learning_rate": 3.9132224809118347e-05,
+      "loss": 0.9433,
+      "step": 448
+    },
+    {
+      "epoch": 2.7820710973724885,
+      "grad_norm": 0.5814134399140919,
+      "learning_rate": 3.8958705717859526e-05,
+      "loss": 0.94,
+      "step": 449
+    },
+    {
+      "epoch": 2.7882534775888717,
+      "grad_norm": 0.49572568964279784,
+      "learning_rate": 3.878520623289041e-05,
+      "loss": 0.9288,
+      "step": 450
+    },
+    {
+      "epoch": 2.7944358578052553,
+      "grad_norm": 0.5284855150274684,
+      "learning_rate": 3.8611729620992726e-05,
+      "loss": 0.9333,
+      "step": 451
+    },
+    {
+      "epoch": 2.8006182380216385,
+      "grad_norm": 0.516369342375662,
+      "learning_rate": 3.8438279148517575e-05,
+      "loss": 0.9484,
+      "step": 452
+    },
+    {
+      "epoch": 2.8068006182380216,
+      "grad_norm": 0.46332381283153445,
+      "learning_rate": 3.826485808132386e-05,
+      "loss": 0.9394,
+      "step": 453
+    },
+    {
+      "epoch": 2.812982998454405,
+      "grad_norm": 0.5353899030659641,
+      "learning_rate": 3.8091469684716845e-05,
+      "loss": 0.93,
+      "step": 454
+    },
+    {
+      "epoch": 2.819165378670788,
+      "grad_norm": 0.3932685579553986,
+      "learning_rate": 3.7918117223386615e-05,
+      "loss": 0.9368,
+      "step": 455
+    },
+    {
+      "epoch": 2.8253477588871716,
+      "grad_norm": 0.43586697192948876,
+      "learning_rate": 3.774480396134663e-05,
+      "loss": 0.9397,
+      "step": 456
+    },
+    {
+      "epoch": 2.8315301391035548,
+      "grad_norm": 0.40361521367835107,
+      "learning_rate": 3.7571533161872334e-05,
+      "loss": 0.9213,
+      "step": 457
+    },
+    {
+      "epoch": 2.837712519319938,
+      "grad_norm": 0.4604217029229173,
+      "learning_rate": 3.7398308087439603e-05,
+      "loss": 0.9268,
+      "step": 458
+    },
+    {
+      "epoch": 2.8438948995363216,
+      "grad_norm": 0.538514910957252,
+      "learning_rate": 3.722513199966336e-05,
+      "loss": 0.9455,
+      "step": 459
+    },
+    {
+      "epoch": 2.8500772797527048,
+      "grad_norm": 0.6470702786089004,
+      "learning_rate": 3.705200815923622e-05,
+      "loss": 0.9123,
+      "step": 460
+    },
+    {
+      "epoch": 2.856259659969088,
+      "grad_norm": 0.5008393569354235,
+      "learning_rate": 3.6878939825867014e-05,
+      "loss": 0.9382,
+      "step": 461
+    },
+    {
+      "epoch": 2.8624420401854715,
+      "grad_norm": 0.41610970566343775,
+      "learning_rate": 3.670593025821942e-05,
+      "loss": 0.9387,
+      "step": 462
+    },
+    {
+      "epoch": 2.8686244204018547,
+      "grad_norm": 0.3582328984320768,
+      "learning_rate": 3.653298271385065e-05,
+      "loss": 0.9365,
+      "step": 463
+    },
+    {
+      "epoch": 2.874806800618238,
+      "grad_norm": 0.36268739794043653,
+      "learning_rate": 3.6360100449150135e-05,
+      "loss": 0.9398,
+      "step": 464
+    },
+    {
+      "epoch": 2.8809891808346215,
+      "grad_norm": 0.31637193895013355,
+      "learning_rate": 3.618728671927811e-05,
+      "loss": 0.9468,
+      "step": 465
+    },
+    {
+      "epoch": 2.8871715610510047,
+      "grad_norm": 0.37689814936512106,
+      "learning_rate": 3.6014544778104394e-05,
+      "loss": 0.9499,
+      "step": 466
+    },
+    {
+      "epoch": 2.893353941267388,
+      "grad_norm": 0.4779224408619651,
+      "learning_rate": 3.584187787814714e-05,
+      "loss": 0.9396,
+      "step": 467
+    },
+    {
+      "epoch": 2.8995363214837715,
+      "grad_norm": 0.509229566930791,
+      "learning_rate": 3.566928927051155e-05,
+      "loss": 0.9291,
+      "step": 468
+    },
+    {
+      "epoch": 2.9057187017001547,
+      "grad_norm": 0.48474577075424746,
+      "learning_rate": 3.5496782204828694e-05,
+      "loss": 0.9203,
+      "step": 469
+    },
+    {
+      "epoch": 2.911901081916538,
+      "grad_norm": 0.3971864637298685,
+      "learning_rate": 3.5324359929194274e-05,
+      "loss": 0.9394,
+      "step": 470
+    },
+    {
+      "epoch": 2.918083462132921,
+      "grad_norm": 0.3053259719312419,
+      "learning_rate": 3.5152025690107524e-05,
+      "loss": 0.9268,
+      "step": 471
+    },
+    {
+      "epoch": 2.9242658423493046,
+      "grad_norm": 0.29216270961639657,
+      "learning_rate": 3.497978273241007e-05,
+      "loss": 0.9148,
+      "step": 472
+    },
+    {
+      "epoch": 2.930448222565688,
+      "grad_norm": 0.2654490613569246,
+      "learning_rate": 3.480763429922479e-05,
+      "loss": 0.9379,
+      "step": 473
+    },
+    {
+      "epoch": 2.936630602782071,
+      "grad_norm": 0.38722618089775157,
+      "learning_rate": 3.463558363189479e-05,
+      "loss": 0.9277,
+      "step": 474
+    },
+    {
+      "epoch": 2.942812982998454,
+      "grad_norm": 0.3877069119339608,
+      "learning_rate": 3.4463633969922395e-05,
+      "loss": 0.9165,
+      "step": 475
+    },
+    {
+      "epoch": 2.948995363214838,
+      "grad_norm": 0.49582780013936295,
+      "learning_rate": 3.429178855090809e-05,
+      "loss": 0.9395,
+      "step": 476
+    },
+    {
+      "epoch": 2.955177743431221,
+      "grad_norm": 0.48453229307728,
+      "learning_rate": 3.412005061048959e-05,
+      "loss": 0.9184,
+      "step": 477
+    },
+    {
+      "epoch": 2.961360123647604,
+      "grad_norm": 0.372292891013131,
+      "learning_rate": 3.394842338228097e-05,
+      "loss": 0.9257,
+      "step": 478
+    },
+    {
+      "epoch": 2.9675425038639878,
+      "grad_norm": 0.21481675893348937,
+      "learning_rate": 3.377691009781168e-05,
+      "loss": 0.9342,
+      "step": 479
+    },
+    {
+      "epoch": 2.973724884080371,
+      "grad_norm": 0.2791079849765998,
+      "learning_rate": 3.360551398646575e-05,
+      "loss": 0.927,
+      "step": 480
+    },
+    {
+      "epoch": 2.979907264296754,
+      "grad_norm": 0.31864441303786156,
+      "learning_rate": 3.3434238275421e-05,
+      "loss": 0.9364,
+      "step": 481
+    },
+    {
+      "epoch": 2.9860896445131377,
+      "grad_norm": 0.3426078238120316,
+      "learning_rate": 3.326308618958828e-05,
+      "loss": 0.9237,
+      "step": 482
+    },
+    {
+      "epoch": 2.992272024729521,
+      "grad_norm": 0.33872000941997643,
+      "learning_rate": 3.309206095155069e-05,
+      "loss": 0.9331,
+      "step": 483
+    },
+    {
+      "epoch": 3.001545595054096,
+      "grad_norm": 0.41746066648889707,
+      "learning_rate": 3.292116578150298e-05,
+      "loss": 1.632,
+      "step": 484
+    },
+    {
+      "epoch": 3.007727975270479,
+      "grad_norm": 0.479690534350614,
+      "learning_rate": 3.2750403897190856e-05,
+      "loss": 0.8917,
+      "step": 485
+    },
+    {
+      "epoch": 3.0139103554868623,
+      "grad_norm": 0.37954910983497075,
+      "learning_rate": 3.2579778513850405e-05,
+      "loss": 0.8802,
+      "step": 486
+    },
+    {
+      "epoch": 3.020092735703246,
+      "grad_norm": 0.43731886585463775,
+      "learning_rate": 3.240929284414762e-05,
+      "loss": 0.9013,
+      "step": 487
+    },
+    {
+      "epoch": 3.026275115919629,
+      "grad_norm": 0.3851011426541529,
+      "learning_rate": 3.223895009811777e-05,
+      "loss": 0.876,
+      "step": 488
+    },
+    {
+      "epoch": 3.0324574961360122,
+      "grad_norm": 0.321671225557479,
+      "learning_rate": 3.2068753483105165e-05,
+      "loss": 0.8888,
+      "step": 489
+    },
+    {
+      "epoch": 3.038639876352396,
+      "grad_norm": 0.4076120056603458,
+      "learning_rate": 3.189870620370256e-05,
+      "loss": 0.9015,
+      "step": 490
+    },
+    {
+      "epoch": 3.044822256568779,
+      "grad_norm": 0.32597913024373476,
+      "learning_rate": 3.1728811461690904e-05,
+      "loss": 0.8858,
+      "step": 491
+    },
+    {
+      "epoch": 3.051004636785162,
+      "grad_norm": 0.39721200533205675,
+      "learning_rate": 3.155907245597912e-05,
+      "loss": 0.8971,
+      "step": 492
+    },
+    {
+      "epoch": 3.0571870170015454,
+      "grad_norm": 0.33689618227170504,
+      "learning_rate": 3.138949238254375e-05,
+      "loss": 0.8913,
+      "step": 493
+    },
+    {
+      "epoch": 3.063369397217929,
+      "grad_norm": 0.31095628858697094,
+      "learning_rate": 3.122007443436885e-05,
+      "loss": 0.8996,
+      "step": 494
+    },
+    {
+      "epoch": 3.069551777434312,
+      "grad_norm": 0.4028486236419519,
+      "learning_rate": 3.105082180138585e-05,
+      "loss": 0.8889,
+      "step": 495
+    },
+    {
+      "epoch": 3.0757341576506954,
+      "grad_norm": 0.484359324518245,
+      "learning_rate": 3.0881737670413534e-05,
+      "loss": 0.8964,
+      "step": 496
+    },
+    {
+      "epoch": 3.081916537867079,
+      "grad_norm": 0.38584852659317825,
+      "learning_rate": 3.071282522509798e-05,
+      "loss": 0.8794,
+      "step": 497
+    },
+    {
+      "epoch": 3.088098918083462,
+      "grad_norm": 0.3397498441090358,
+      "learning_rate": 3.054408764585261e-05,
+      "loss": 0.8957,
+      "step": 498
+    },
+    {
+      "epoch": 3.0942812982998453,
+      "grad_norm": 0.31333063472846795,
+      "learning_rate": 3.037552810979836e-05,
+      "loss": 0.8786,
+      "step": 499
+    },
+    {
+      "epoch": 3.100463678516229,
+      "grad_norm": 0.3649292765310615,
+      "learning_rate": 3.020714979070385e-05,
+      "loss": 0.8884,
+      "step": 500
+    },
+    {
+      "epoch": 3.106646058732612,
+      "grad_norm": 0.4200346160735661,
+      "learning_rate": 3.0038955858925556e-05,
+      "loss": 0.8846,
+      "step": 501
+    },
+    {
+      "epoch": 3.1128284389489953,
+      "grad_norm": 0.28488029439459195,
+      "learning_rate": 2.9870949481348214e-05,
+      "loss": 0.8942,
+      "step": 502
+    },
+    {
+      "epoch": 3.1190108191653785,
+      "grad_norm": 0.2697608832955852,
+      "learning_rate": 2.9703133821325125e-05,
+      "loss": 0.8915,
+      "step": 503
+    },
+    {
+      "epoch": 3.125193199381762,
+      "grad_norm": 0.3357046563951315,
+      "learning_rate": 2.953551203861863e-05,
+      "loss": 0.8885,
+      "step": 504
+    },
+    {
+      "epoch": 3.1313755795981453,
+      "grad_norm": 0.3200692987355479,
+      "learning_rate": 2.9368087289340575e-05,
+      "loss": 0.8844,
+      "step": 505
+    },
+    {
+      "epoch": 3.1375579598145285,
+      "grad_norm": 0.24036855461462353,
+      "learning_rate": 2.9200862725892897e-05,
+      "loss": 0.8929,
+      "step": 506
+    },
+    {
+      "epoch": 3.143740340030912,
+      "grad_norm": 0.238395091168019,
+      "learning_rate": 2.903384149690834e-05,
+      "loss": 0.8874,
+      "step": 507
+    },
+    {
+      "epoch": 3.1499227202472952,
+      "grad_norm": 0.2401017173847325,
+      "learning_rate": 2.8867026747191054e-05,
+      "loss": 0.8876,
+      "step": 508
+    },
+    {
+      "epoch": 3.1561051004636784,
+      "grad_norm": 0.2799086754124938,
+      "learning_rate": 2.8700421617657446e-05,
+      "loss": 0.8958,
+      "step": 509
+    },
+    {
+      "epoch": 3.162287480680062,
+      "grad_norm": 0.2780648545074759,
+      "learning_rate": 2.853402924527708e-05,
+      "loss": 0.8952,
+      "step": 510
+    },
+    {
+      "epoch": 3.1684698608964452,
+      "grad_norm": 0.23416871362717123,
+      "learning_rate": 2.8367852763013508e-05,
+      "loss": 0.8783,
+      "step": 511
+    },
+    {
+      "epoch": 3.1746522411128284,
+      "grad_norm": 0.19517664014067915,
+      "learning_rate": 2.8201895299765358e-05,
+      "loss": 0.8885,
+      "step": 512
+    },
+    {
+      "epoch": 3.1808346213292116,
+      "grad_norm": 0.24055029383355403,
+      "learning_rate": 2.803615998030738e-05,
+      "loss": 0.8781,
+      "step": 513
+    },
+    {
+      "epoch": 3.187017001545595,
+      "grad_norm": 0.21251629293077243,
+      "learning_rate": 2.787064992523169e-05,
+      "loss": 0.896,
+      "step": 514
+    },
+    {
+      "epoch": 3.1931993817619784,
+      "grad_norm": 0.2205317155524029,
+      "learning_rate": 2.7705368250888892e-05,
+      "loss": 0.8977,
+      "step": 515
+    },
+    {
+      "epoch": 3.1993817619783615,
+      "grad_norm": 0.2710843453756844,
+      "learning_rate": 2.7540318069329468e-05,
+      "loss": 0.8809,
+      "step": 516
+    },
+    {
+      "epoch": 3.205564142194745,
+      "grad_norm": 0.2134694304897921,
+      "learning_rate": 2.7375502488245218e-05,
+      "loss": 0.8852,
+      "step": 517
+    },
+    {
+      "epoch": 3.2117465224111283,
+      "grad_norm": 0.21525338899378574,
+      "learning_rate": 2.7210924610910658e-05,
+      "loss": 0.8843,
+      "step": 518
+    },
+    {
+      "epoch": 3.2179289026275115,
+      "grad_norm": 0.2257874439917892,
+      "learning_rate": 2.7046587536124656e-05,
+      "loss": 0.8912,
+      "step": 519
+    },
+    {
+      "epoch": 3.2241112828438947,
+      "grad_norm": 0.2146891709570046,
+      "learning_rate": 2.6882494358152055e-05,
+      "loss": 0.8902,
+      "step": 520
+    },
+    {
+      "epoch": 3.2302936630602783,
+      "grad_norm": 0.19521290168739072,
+      "learning_rate": 2.6718648166665445e-05,
+      "loss": 0.8931,
+      "step": 521
+    },
+    {
+      "epoch": 3.2364760432766615,
+      "grad_norm": 0.2375023715085163,
+      "learning_rate": 2.6555052046686943e-05,
+      "loss": 0.8799,
+      "step": 522
+    },
+    {
+      "epoch": 3.2426584234930447,
+      "grad_norm": 0.2440992341795168,
+      "learning_rate": 2.6391709078530106e-05,
+      "loss": 0.9015,
+      "step": 523
+    },
+    {
+      "epoch": 3.2488408037094283,
+      "grad_norm": 0.2660799604690821,
+      "learning_rate": 2.6228622337742016e-05,
+      "loss": 0.8869,
+      "step": 524
+    },
+    {
+      "epoch": 3.2550231839258115,
+      "grad_norm": 0.1898820986247372,
+      "learning_rate": 2.606579489504526e-05,
+      "loss": 0.8921,
+      "step": 525
+    },
+    {
+      "epoch": 3.2612055641421946,
+      "grad_norm": 0.21206019174841687,
+      "learning_rate": 2.590322981628016e-05,
+      "loss": 0.8969,
+      "step": 526
+    },
+    {
+      "epoch": 3.2673879443585783,
+      "grad_norm": 0.21516060193400413,
+      "learning_rate": 2.5740930162347053e-05,
+      "loss": 0.8839,
+      "step": 527
+    },
+    {
+      "epoch": 3.2735703245749614,
+      "grad_norm": 0.19044808648502842,
+      "learning_rate": 2.5578898989148692e-05,
+      "loss": 0.8927,
+      "step": 528
+    },
+    {
+      "epoch": 3.2797527047913446,
+      "grad_norm": 0.22486834698093985,
+      "learning_rate": 2.5417139347532625e-05,
+      "loss": 0.8889,
+      "step": 529
+    },
+    {
+      "epoch": 3.285935085007728,
+      "grad_norm": 0.19216048198447702,
+      "learning_rate": 2.525565428323379e-05,
+      "loss": 0.8893,
+      "step": 530
+    },
+    {
+      "epoch": 3.2921174652241114,
+      "grad_norm": 0.20965862072042166,
+      "learning_rate": 2.509444683681722e-05,
+      "loss": 0.8989,
+      "step": 531
+    },
+    {
+      "epoch": 3.2982998454404946,
+      "grad_norm": 0.1815566372511265,
+      "learning_rate": 2.49335200436207e-05,
+      "loss": 0.8795,
+      "step": 532
+    },
+    {
+      "epoch": 3.3044822256568778,
+      "grad_norm": 0.23072409231554872,
+      "learning_rate": 2.4772876933697658e-05,
+      "loss": 0.909,
+      "step": 533
+    },
+    {
+      "epoch": 3.3106646058732614,
+      "grad_norm": 0.2299841285498698,
+      "learning_rate": 2.461252053176015e-05,
+      "loss": 0.8966,
+      "step": 534
+    },
+    {
+      "epoch": 3.3168469860896446,
+      "grad_norm": 0.23678203939185868,
+      "learning_rate": 2.4452453857121846e-05,
+      "loss": 0.904,
+      "step": 535
+    },
+    {
+      "epoch": 3.3230293663060277,
+      "grad_norm": 0.23345278141855083,
+      "learning_rate": 2.429267992364123e-05,
+      "loss": 0.8709,
+      "step": 536
+    },
+    {
+      "epoch": 3.329211746522411,
+      "grad_norm": 0.21351104494670123,
+      "learning_rate": 2.413320173966481e-05,
+      "loss": 0.886,
+      "step": 537
+    },
+    {
+      "epoch": 3.3353941267387945,
+      "grad_norm": 0.24843505340058486,
+      "learning_rate": 2.3974022307970488e-05,
+      "loss": 0.8994,
+      "step": 538
+    },
+    {
+      "epoch": 3.3415765069551777,
+      "grad_norm": 0.2144602081955269,
+      "learning_rate": 2.3815144625711063e-05,
+      "loss": 0.8928,
+      "step": 539
+    },
+    {
+      "epoch": 3.347758887171561,
+      "grad_norm": 0.214440559424067,
+      "learning_rate": 2.3656571684357736e-05,
+      "loss": 0.8952,
+      "step": 540
+    },
+    {
+      "epoch": 3.3539412673879445,
+      "grad_norm": 0.1870653320452973,
+      "learning_rate": 2.34983064696438e-05,
+      "loss": 0.8789,
+      "step": 541
+    },
+    {
+      "epoch": 3.3601236476043277,
+      "grad_norm": 0.1816470397576087,
+      "learning_rate": 2.3340351961508495e-05,
+      "loss": 0.8904,
+      "step": 542
+    },
+    {
+      "epoch": 3.366306027820711,
+      "grad_norm": 0.23629374681914123,
+      "learning_rate": 2.318271113404075e-05,
+      "loss": 0.8853,
+      "step": 543
+    },
+    {
+      "epoch": 3.3724884080370945,
+      "grad_norm": 0.2136915907110259,
+      "learning_rate": 2.3025386955423342e-05,
+      "loss": 0.9049,
+      "step": 544
+    },
+    {
+      "epoch": 3.3786707882534777,
+      "grad_norm": 0.22758131014283642,
+      "learning_rate": 2.2868382387876905e-05,
+      "loss": 0.8876,
+      "step": 545
+    },
+    {
+      "epoch": 3.384853168469861,
+      "grad_norm": 0.2365037683934222,
+      "learning_rate": 2.2711700387604223e-05,
+      "loss": 0.9025,
+      "step": 546
+    },
+    {
+      "epoch": 3.391035548686244,
+      "grad_norm": 0.21359052590913488,
+      "learning_rate": 2.255534390473451e-05,
+      "loss": 0.8939,
+      "step": 547
+    },
+    {
+      "epoch": 3.3972179289026276,
+      "grad_norm": 0.21656702849393653,
+      "learning_rate": 2.2399315883267853e-05,
+      "loss": 0.8987,
+      "step": 548
+    },
+    {
+      "epoch": 3.403400309119011,
+      "grad_norm": 0.225661122723241,
+      "learning_rate": 2.2243619261019918e-05,
+      "loss": 0.8893,
+      "step": 549
+    },
+    {
+      "epoch": 3.409582689335394,
+      "grad_norm": 0.18874144934647666,
+      "learning_rate": 2.2088256969566413e-05,
+      "loss": 0.9041,
+      "step": 550
+    },
+    {
+      "epoch": 3.4157650695517776,
+      "grad_norm": 0.28668375357552045,
+      "learning_rate": 2.1933231934188104e-05,
+      "loss": 0.8855,
+      "step": 551
+    },
+    {
+      "epoch": 3.421947449768161,
+      "grad_norm": 0.2531027429361992,
+      "learning_rate": 2.1778547073815554e-05,
+      "loss": 0.8954,
+      "step": 552
+    },
+    {
+      "epoch": 3.428129829984544,
+      "grad_norm": 0.21767635407476482,
+      "learning_rate": 2.1624205300974346e-05,
+      "loss": 0.8951,
+      "step": 553
+    },
+    {
+      "epoch": 3.434312210200927,
+      "grad_norm": 0.18527077734942415,
+      "learning_rate": 2.1470209521730064e-05,
+      "loss": 0.891,
+      "step": 554
+    },
+    {
+      "epoch": 3.4404945904173108,
+      "grad_norm": 0.26428285410878966,
+      "learning_rate": 2.131656263563369e-05,
+      "loss": 0.9102,
+      "step": 555
+    },
+    {
+      "epoch": 3.446676970633694,
+      "grad_norm": 0.17077454909439538,
+      "learning_rate": 2.1163267535666994e-05,
+      "loss": 0.9022,
+      "step": 556
+    },
+    {
+      "epoch": 3.452859350850077,
+      "grad_norm": 0.2136026248917033,
+      "learning_rate": 2.1010327108188037e-05,
+      "loss": 0.8797,
+      "step": 557
+    },
+    {
+      "epoch": 3.4590417310664607,
+      "grad_norm": 0.23170046494717933,
+      "learning_rate": 2.0857744232876797e-05,
+      "loss": 0.9014,
+      "step": 558
+    },
+    {
+      "epoch": 3.465224111282844,
+      "grad_norm": 0.18466452017435814,
+      "learning_rate": 2.070552178268102e-05,
+      "loss": 0.893,
+      "step": 559
+    },
+    {
+      "epoch": 3.471406491499227,
+      "grad_norm": 0.2938322911655651,
+      "learning_rate": 2.0553662623762105e-05,
+      "loss": 0.8872,
+      "step": 560
+    },
+    {
+      "epoch": 3.4775888717156107,
+      "grad_norm": 0.28795550547667914,
+      "learning_rate": 2.040216961544109e-05,
+      "loss": 0.8975,
+      "step": 561
+    },
+    {
+      "epoch": 3.483771251931994,
+      "grad_norm": 0.22847154183739538,
+      "learning_rate": 2.0251045610144865e-05,
+      "loss": 0.8918,
+      "step": 562
+    },
+    {
+      "epoch": 3.489953632148377,
+      "grad_norm": 0.28080426895854366,
+      "learning_rate": 2.0100293453352446e-05,
+      "loss": 0.8703,
+      "step": 563
+    },
+    {
+      "epoch": 3.4961360123647607,
+      "grad_norm": 0.23972543857227632,
+      "learning_rate": 1.9949915983541405e-05,
+      "loss": 0.8794,
+      "step": 564
+    },
+    {
+      "epoch": 3.502318392581144,
+      "grad_norm": 0.2159429902592839,
+      "learning_rate": 1.9799916032134384e-05,
+      "loss": 0.8844,
+      "step": 565
+    },
+    {
+      "epoch": 3.508500772797527,
+      "grad_norm": 0.2126223807781731,
+      "learning_rate": 1.9650296423445865e-05,
+      "loss": 0.8934,
+      "step": 566
+    },
+    {
+      "epoch": 3.51468315301391,
+      "grad_norm": 0.25136395675418194,
+      "learning_rate": 1.9501059974628923e-05,
+      "loss": 0.8951,
+      "step": 567
+    },
+    {
+      "epoch": 3.5208655332302934,
+      "grad_norm": 0.2422901671285048,
+      "learning_rate": 1.9352209495622215e-05,
+      "loss": 0.8931,
+      "step": 568
+    },
+    {
+      "epoch": 3.527047913446677,
+      "grad_norm": 0.22399150155898304,
+      "learning_rate": 1.920374778909702e-05,
+      "loss": 0.894,
+      "step": 569
+    },
+    {
+      "epoch": 3.53323029366306,
+      "grad_norm": 0.17997096245249658,
+      "learning_rate": 1.9055677650404598e-05,
+      "loss": 0.8854,
+      "step": 570
+    },
+    {
+      "epoch": 3.5394126738794434,
+      "grad_norm": 0.20091228641905334,
+      "learning_rate": 1.890800186752336e-05,
+      "loss": 0.8912,
+      "step": 571
+    },
+    {
+      "epoch": 3.545595054095827,
+      "grad_norm": 0.1713636230002705,
+      "learning_rate": 1.8760723221006577e-05,
+      "loss": 0.8984,
+      "step": 572
+    },
+    {
+      "epoch": 3.55177743431221,
+      "grad_norm": 0.22444993402898378,
+      "learning_rate": 1.861384448392984e-05,
+      "loss": 0.8957,
+      "step": 573
+    },
+    {
+      "epoch": 3.5579598145285933,
+      "grad_norm": 0.1622926494483365,
+      "learning_rate": 1.8467368421839045e-05,
+      "loss": 0.8882,
+      "step": 574
+    },
+    {
+      "epoch": 3.564142194744977,
+      "grad_norm": 0.22561783403087973,
+      "learning_rate": 1.832129779269811e-05,
+      "loss": 0.8837,
+      "step": 575
+    },
+    {
+      "epoch": 3.57032457496136,
+      "grad_norm": 0.1776156141854532,
+      "learning_rate": 1.817563534683722e-05,
+      "loss": 0.8705,
+      "step": 576
+    },
+    {
+      "epoch": 3.5765069551777433,
+      "grad_norm": 0.1905281137121667,
+      "learning_rate": 1.8030383826900935e-05,
+      "loss": 0.8814,
+      "step": 577
+    },
+    {
+      "epoch": 3.582689335394127,
+      "grad_norm": 0.2044502851986332,
+      "learning_rate": 1.7885545967796606e-05,
+      "loss": 0.8981,
+      "step": 578
+    },
+    {
+      "epoch": 3.58887171561051,
+      "grad_norm": 0.1784191323277939,
+      "learning_rate": 1.7741124496642862e-05,
+      "loss": 0.8901,
+      "step": 579
+    },
+    {
+      "epoch": 3.5950540958268933,
+      "grad_norm": 0.18594801839576822,
+      "learning_rate": 1.7597122132718202e-05,
+      "loss": 0.8666,
+      "step": 580
+    },
+    {
+      "epoch": 3.601236476043277,
+      "grad_norm": 0.18365311459244033,
+      "learning_rate": 1.7453541587409963e-05,
+      "loss": 0.8936,
+      "step": 581
+    },
+    {
+      "epoch": 3.60741885625966,
+      "grad_norm": 0.20713153169279955,
+      "learning_rate": 1.731038556416304e-05,
+      "loss": 0.8898,
+      "step": 582
+    },
+    {
+      "epoch": 3.6136012364760433,
+      "grad_norm": 0.20905582740873777,
+      "learning_rate": 1.716765675842919e-05,
+      "loss": 0.8905,
+      "step": 583
+    },
+    {
+      "epoch": 3.6197836166924264,
+      "grad_norm": 0.18415135918656184,
+      "learning_rate": 1.70253578576161e-05,
+      "loss": 0.8945,
+      "step": 584
+    },
+    {
+      "epoch": 3.62596599690881,
+      "grad_norm": 0.2042255922116097,
+      "learning_rate": 1.688349154103699e-05,
+      "loss": 0.8839,
+      "step": 585
+    },
+    {
+      "epoch": 3.6321483771251932,
+      "grad_norm": 0.17189580999683968,
+      "learning_rate": 1.6742060479859944e-05,
+      "loss": 0.8747,
+      "step": 586
+    },
+    {
+      "epoch": 3.6383307573415764,
+      "grad_norm": 0.21696578872996222,
+      "learning_rate": 1.6601067337057783e-05,
+      "loss": 0.8863,
+      "step": 587
+    },
+    {
+      "epoch": 3.6445131375579596,
+      "grad_norm": 0.20566434805583705,
+      "learning_rate": 1.6460514767357855e-05,
+      "loss": 0.8978,
+      "step": 588
+    },
+    {
+      "epoch": 3.650695517774343,
+      "grad_norm": 0.212105435407208,
+      "learning_rate": 1.6320405417192066e-05,
+      "loss": 0.8796,
+      "step": 589
+    },
+    {
+      "epoch": 3.6568778979907264,
+      "grad_norm": 0.21704798077180454,
+      "learning_rate": 1.618074192464699e-05,
+      "loss": 0.8962,
+      "step": 590
+    },
+    {
+      "epoch": 3.6630602782071096,
+      "grad_norm": 0.26219775112073423,
+      "learning_rate": 1.6041526919414318e-05,
+      "loss": 0.8965,
+      "step": 591
+    },
+    {
+      "epoch": 3.669242658423493,
+      "grad_norm": 0.18361447307055298,
+      "learning_rate": 1.5902763022741247e-05,
+      "loss": 0.8912,
+      "step": 592
+    },
+    {
+      "epoch": 3.6754250386398764,
+      "grad_norm": 0.2767924709588361,
+      "learning_rate": 1.576445284738116e-05,
+      "loss": 0.8974,
+      "step": 593
+    },
+    {
+      "epoch": 3.6816074188562595,
+      "grad_norm": 0.21044903902705353,
+      "learning_rate": 1.5626598997544422e-05,
+      "loss": 0.8932,
+      "step": 594
+    },
+    {
+      "epoch": 3.687789799072643,
+      "grad_norm": 0.22141335816275728,
+      "learning_rate": 1.548920406884935e-05,
+      "loss": 0.8966,
+      "step": 595
+    },
+    {
+      "epoch": 3.6939721792890263,
+      "grad_norm": 0.19199822889292517,
+      "learning_rate": 1.5352270648273363e-05,
+      "loss": 0.8927,
+      "step": 596
+    },
+    {
+      "epoch": 3.7001545595054095,
+      "grad_norm": 0.2256809667548217,
+      "learning_rate": 1.5215801314104202e-05,
+      "loss": 0.8917,
+      "step": 597
+    },
+    {
+      "epoch": 3.706336939721793,
+      "grad_norm": 0.23515679935616451,
+      "learning_rate": 1.5079798635891468e-05,
+      "loss": 0.8907,
+      "step": 598
+    },
+    {
+      "epoch": 3.7125193199381763,
+      "grad_norm": 0.1640102854982691,
+      "learning_rate": 1.4944265174398203e-05,
+      "loss": 0.9016,
+      "step": 599
+    },
+    {
+      "epoch": 3.7187017001545595,
+      "grad_norm": 0.19789240249181025,
+      "learning_rate": 1.4809203481552681e-05,
+      "loss": 0.8811,
+      "step": 600
+    },
+    {
+      "epoch": 3.7248840803709427,
+      "grad_norm": 0.17774590981378566,
+      "learning_rate": 1.4674616100400303e-05,
+      "loss": 0.8811,
+      "step": 601
+    },
+    {
+      "epoch": 3.7310664605873263,
+      "grad_norm": 0.19736844331190076,
+      "learning_rate": 1.4540505565055858e-05,
+      "loss": 0.8996,
+      "step": 602
+    },
+    {
+      "epoch": 3.7372488408037094,
+      "grad_norm": 0.17893905096693596,
+      "learning_rate": 1.4406874400655615e-05,
+      "loss": 0.9037,
+      "step": 603
+    },
+    {
+      "epoch": 3.7434312210200926,
+      "grad_norm": 0.19684792739449095,
+      "learning_rate": 1.427372512330996e-05,
+      "loss": 0.8851,
+      "step": 604
+    },
+    {
+      "epoch": 3.749613601236476,
+      "grad_norm": 0.2512744015419969,
+      "learning_rate": 1.4141060240055859e-05,
+      "loss": 0.8954,
+      "step": 605
+    },
+    {
+      "epoch": 3.7557959814528594,
+      "grad_norm": 0.17423327717089102,
+      "learning_rate": 1.400888224880983e-05,
+      "loss": 0.8808,
+      "step": 606
+    },
+    {
+      "epoch": 3.7619783616692426,
+      "grad_norm": 0.1819843673558297,
+      "learning_rate": 1.3877193638320718e-05,
+      "loss": 0.8982,
+      "step": 607
+    },
+    {
+      "epoch": 3.7681607418856258,
+      "grad_norm": 0.1693796580052696,
+      "learning_rate": 1.3745996888122992e-05,
+      "loss": 0.8701,
+      "step": 608
+    },
+    {
+      "epoch": 3.7743431221020094,
+      "grad_norm": 0.1821658263170253,
+      "learning_rate": 1.361529446848998e-05,
+      "loss": 0.8843,
+      "step": 609
+    },
+    {
+      "epoch": 3.7805255023183926,
+      "grad_norm": 0.24650364015406534,
+      "learning_rate": 1.3485088840387369e-05,
+      "loss": 0.8966,
+      "step": 610
+    },
+    {
+      "epoch": 3.7867078825347757,
+      "grad_norm": 0.18771305026658164,
+      "learning_rate": 1.3355382455426892e-05,
+      "loss": 0.8846,
+      "step": 611
+    },
+    {
+      "epoch": 3.7928902627511594,
+      "grad_norm": 0.18610630863899502,
+      "learning_rate": 1.3226177755820087e-05,
+      "loss": 0.8846,
+      "step": 612
+    },
+    {
+      "epoch": 3.7990726429675425,
+      "grad_norm": 0.17549326513432761,
+      "learning_rate": 1.3097477174332482e-05,
+      "loss": 0.8855,
+      "step": 613
+    },
+    {
+      "epoch": 3.8052550231839257,
+      "grad_norm": 0.17546252086320724,
+      "learning_rate": 1.296928313423758e-05,
+      "loss": 0.894,
+      "step": 614
+    },
+    {
+      "epoch": 3.8114374034003093,
+      "grad_norm": 0.1904786805615447,
+      "learning_rate": 1.2841598049271395e-05,
+      "loss": 0.8847,
+      "step": 615
+    },
+    {
+      "epoch": 3.8176197836166925,
+      "grad_norm": 0.16824287711277486,
+      "learning_rate": 1.271442432358688e-05,
+      "loss": 0.882,
+      "step": 616
+    },
+    {
+      "epoch": 3.8238021638330757,
+      "grad_norm": 0.15191088188497087,
+      "learning_rate": 1.2587764351708813e-05,
+      "loss": 0.897,
+      "step": 617
+    },
+    {
+      "epoch": 3.8299845440494593,
+      "grad_norm": 0.17144785228852993,
+      "learning_rate": 1.2461620518488533e-05,
+      "loss": 0.8947,
+      "step": 618
+    },
+    {
+      "epoch": 3.8361669242658425,
+      "grad_norm": 0.1981987256042829,
+      "learning_rate": 1.2335995199059175e-05,
+      "loss": 0.8888,
+      "step": 619
+    },
+    {
+      "epoch": 3.8423493044822257,
+      "grad_norm": 0.1664686018609899,
+      "learning_rate": 1.2210890758790877e-05,
+      "loss": 0.8865,
+      "step": 620
+    },
+    {
+      "epoch": 3.848531684698609,
+      "grad_norm": 0.17952380029164736,
+      "learning_rate": 1.2086309553246282e-05,
+      "loss": 0.8735,
+      "step": 621
+    },
+    {
+      "epoch": 3.854714064914992,
+      "grad_norm": 0.1654528501349644,
+      "learning_rate": 1.1962253928136129e-05,
+      "loss": 0.8975,
+      "step": 622
+    },
+    {
+      "epoch": 3.8608964451313756,
+      "grad_norm": 0.19932349446317263,
+      "learning_rate": 1.1838726219275163e-05,
+      "loss": 0.8994,
+      "step": 623
+    },
+    {
+      "epoch": 3.867078825347759,
+      "grad_norm": 0.16515103220193778,
+      "learning_rate": 1.1715728752538103e-05,
+      "loss": 0.8795,
+      "step": 624
+    },
+    {
+      "epoch": 3.873261205564142,
+      "grad_norm": 0.15802093965187888,
+      "learning_rate": 1.1593263843815854e-05,
+      "loss": 0.8846,
+      "step": 625
+    },
+    {
+      "epoch": 3.8794435857805256,
+      "grad_norm": 0.18002633059902745,
+      "learning_rate": 1.1471333798971922e-05,
+      "loss": 0.8813,
+      "step": 626
+    },
+    {
+      "epoch": 3.885625965996909,
+      "grad_norm": 0.16043447191960278,
+      "learning_rate": 1.1349940913798978e-05,
+      "loss": 0.88,
+      "step": 627
+    },
+    {
+      "epoch": 3.891808346213292,
+      "grad_norm": 0.21327907033825158,
+      "learning_rate": 1.1229087473975641e-05,
+      "loss": 0.8961,
+      "step": 628
+    },
+    {
+      "epoch": 3.8979907264296756,
+      "grad_norm": 0.15650552928778205,
+      "learning_rate": 1.1108775755023422e-05,
+      "loss": 0.8748,
+      "step": 629
+    },
+    {
+      "epoch": 3.9041731066460588,
+      "grad_norm": 0.18123273261524603,
+      "learning_rate": 1.0989008022263906e-05,
+      "loss": 0.8935,
+      "step": 630
+    },
+    {
+      "epoch": 3.910355486862442,
+      "grad_norm": 0.15260920437276415,
+      "learning_rate": 1.08697865307761e-05,
+      "loss": 0.8852,
+      "step": 631
+    },
+    {
+      "epoch": 3.9165378670788256,
+      "grad_norm": 0.165453780873266,
+      "learning_rate": 1.0751113525353957e-05,
+      "loss": 0.9049,
+      "step": 632
+    },
+    {
+      "epoch": 3.9227202472952087,
+      "grad_norm": 0.13033209339725224,
+      "learning_rate": 1.0632991240464068e-05,
+      "loss": 0.8931,
+      "step": 633
+    },
+    {
+      "epoch": 3.928902627511592,
+      "grad_norm": 0.16631041812626945,
+      "learning_rate": 1.0515421900203714e-05,
+      "loss": 0.888,
+      "step": 634
+    },
+    {
+      "epoch": 3.9350850077279755,
+      "grad_norm": 0.16953390165836668,
+      "learning_rate": 1.0398407718258836e-05,
+      "loss": 0.8984,
+      "step": 635
+    },
+    {
+      "epoch": 3.9412673879443587,
+      "grad_norm": 0.14284684056392535,
+      "learning_rate": 1.028195089786248e-05,
+      "loss": 0.9026,
+      "step": 636
+    },
+    {
+      "epoch": 3.947449768160742,
+      "grad_norm": 0.16855244368923294,
+      "learning_rate": 1.0166053631753204e-05,
+      "loss": 0.8875,
+      "step": 637
+    },
+    {
+      "epoch": 3.953632148377125,
+      "grad_norm": 0.14590639079902812,
+      "learning_rate": 1.0050718102133916e-05,
+      "loss": 0.8929,
+      "step": 638
+    },
+    {
+      "epoch": 3.9598145285935082,
+      "grad_norm": 0.14384779730638278,
+      "learning_rate": 9.935946480630658e-06,
+      "loss": 0.8825,
+      "step": 639
+    },
+    {
+      "epoch": 3.965996908809892,
+      "grad_norm": 0.1534372289046035,
+      "learning_rate": 9.82174092825181e-06,
+      "loss": 0.8982,
+      "step": 640
+    },
+    {
+      "epoch": 3.972179289026275,
+      "grad_norm": 0.15609090465249761,
+      "learning_rate": 9.708103595347352e-06,
+      "loss": 0.8939,
+      "step": 641
+    },
+    {
+      "epoch": 3.978361669242658,
+      "grad_norm": 0.1403869008769113,
+      "learning_rate": 9.595036621568398e-06,
+      "loss": 0.8843,
+      "step": 642
+    },
+    {
+      "epoch": 3.984544049459042,
+      "grad_norm": 0.14678540378586016,
+      "learning_rate": 9.482542135826916e-06,
+      "loss": 0.8987,
+      "step": 643
+    },
+    {
+      "epoch": 3.990726429675425,
+      "grad_norm": 0.1875416299280276,
+      "learning_rate": 9.370622256255571e-06,
+      "loss": 0.8774,
+      "step": 644
+    },
+    {
+      "epoch": 3.996908809891808,
+      "grad_norm": 0.27186472008456647,
+      "learning_rate": 9.259279090167994e-06,
+      "loss": 1.5557,
+      "step": 645
+    },
+    {
+      "epoch": 4.006182380216384,
+      "grad_norm": 0.20982320372706692,
+      "learning_rate": 9.148514734018917e-06,
+      "loss": 0.87,
+      "step": 646
+    },
+    {
+      "epoch": 4.012364760432766,
+      "grad_norm": 0.17703070468066032,
+      "learning_rate": 9.038331273364869e-06,
+      "loss": 0.8649,
+      "step": 647
+    },
+    {
+      "epoch": 4.01854714064915,
+      "grad_norm": 0.16674753883298887,
+      "learning_rate": 8.928730782824755e-06,
+      "loss": 0.8577,
+      "step": 648
+    },
+    {
+      "epoch": 4.024729520865534,
+      "grad_norm": 0.16839923962797768,
+      "learning_rate": 8.819715326040965e-06,
+      "loss": 0.8571,
+      "step": 649
+    },
+    {
+      "epoch": 4.030911901081916,
+      "grad_norm": 0.220038624506249,
+      "learning_rate": 8.711286955640332e-06,
+      "loss": 0.8609,
+      "step": 650
+    },
+    {
+      "epoch": 4.0370942812983,
+      "grad_norm": 0.16790422765840488,
+      "learning_rate": 8.603447713195634e-06,
+      "loss": 0.8566,
+      "step": 651
+    },
+    {
+      "epoch": 4.043276661514684,
+      "grad_norm": 0.18920607343960402,
+      "learning_rate": 8.496199629187072e-06,
+      "loss": 0.8608,
+      "step": 652
+    },
+    {
+      "epoch": 4.049459041731066,
+      "grad_norm": 0.18186173348032736,
+      "learning_rate": 8.389544722964076e-06,
+      "loss": 0.8675,
+      "step": 653
+    },
+    {
+      "epoch": 4.05564142194745,
+      "grad_norm": 0.17348169700628485,
+      "learning_rate": 8.283485002707233e-06,
+      "loss": 0.8689,
+      "step": 654
+    },
+    {
+      "epoch": 4.061823802163833,
+      "grad_norm": 0.17798870193571564,
+      "learning_rate": 8.178022465390549e-06,
+      "loss": 0.853,
+      "step": 655
+    },
+    {
+      "epoch": 4.068006182380216,
+      "grad_norm": 0.18943450530940084,
+      "learning_rate": 8.073159096743799e-06,
+      "loss": 0.8645,
+      "step": 656
+    },
+    {
+      "epoch": 4.0741885625966,
+      "grad_norm": 0.1602196742298431,
+      "learning_rate": 7.968896871215147e-06,
+      "loss": 0.8517,
+      "step": 657
+    },
+    {
+      "epoch": 4.080370942812983,
+      "grad_norm": 0.16728654932655415,
+      "learning_rate": 7.865237751933965e-06,
+      "loss": 0.858,
+      "step": 658
+    },
+    {
+      "epoch": 4.086553323029366,
+      "grad_norm": 0.17345900037449719,
+      "learning_rate": 7.762183690673887e-06,
+      "loss": 0.8552,
+      "step": 659
+    },
+    {
+      "epoch": 4.09273570324575,
+      "grad_norm": 0.16849302168626767,
+      "learning_rate": 7.659736627816064e-06,
+      "loss": 0.8559,
+      "step": 660
+    },
+    {
+      "epoch": 4.098918083462133,
+      "grad_norm": 0.15814956982575118,
+      "learning_rate": 7.557898492312561e-06,
+      "loss": 0.8527,
+      "step": 661
+    },
+    {
+      "epoch": 4.105100463678516,
+      "grad_norm": 0.15201462004812408,
+      "learning_rate": 7.456671201650145e-06,
+      "loss": 0.8551,
+      "step": 662
+    },
+    {
+      "epoch": 4.1112828438949,
+      "grad_norm": 0.13755795678813534,
+      "learning_rate": 7.356056661814106e-06,
+      "loss": 0.8557,
+      "step": 663
+    },
+    {
+      "epoch": 4.117465224111283,
+      "grad_norm": 0.15959755820070046,
+      "learning_rate": 7.256056767252402e-06,
+      "loss": 0.8539,
+      "step": 664
+    },
+    {
+      "epoch": 4.123647604327666,
+      "grad_norm": 0.1511791942647596,
+      "learning_rate": 7.156673400839933e-06,
+      "loss": 0.853,
+      "step": 665
+    },
+    {
+      "epoch": 4.12982998454405,
+      "grad_norm": 0.14354560331309524,
+      "learning_rate": 7.057908433843205e-06,
+      "loss": 0.8464,
+      "step": 666
+    },
+    {
+      "epoch": 4.1360123647604325,
+      "grad_norm": 0.1365948998014824,
+      "learning_rate": 6.959763725884956e-06,
+      "loss": 0.86,
+      "step": 667
+    },
+    {
+      "epoch": 4.142194744976816,
+      "grad_norm": 0.1745107521168772,
+      "learning_rate": 6.862241124909262e-06,
+      "loss": 0.8559,
+      "step": 668
+    },
+    {
+      "epoch": 4.1483771251932,
+      "grad_norm": 0.16007671239764104,
+      "learning_rate": 6.765342467146632e-06,
+      "loss": 0.8547,
+      "step": 669
+    },
+    {
+      "epoch": 4.1545595054095825,
+      "grad_norm": 0.14588280210721616,
+      "learning_rate": 6.66906957707957e-06,
+      "loss": 0.8563,
+      "step": 670
+    },
+    {
+      "epoch": 4.160741885625966,
+      "grad_norm": 0.13534918455746356,
+      "learning_rate": 6.5734242674080705e-06,
+      "loss": 0.8536,
+      "step": 671
+    },
+    {
+      "epoch": 4.166924265842349,
+      "grad_norm": 0.16010646794448433,
+      "learning_rate": 6.478408339015603e-06,
+      "loss": 0.8624,
+      "step": 672
+    },
+    {
+      "epoch": 4.1731066460587325,
+      "grad_norm": 0.22075894460772627,
+      "learning_rate": 6.384023580935155e-06,
+      "loss": 0.8588,
+      "step": 673
+    },
+    {
+      "epoch": 4.179289026275116,
+      "grad_norm": 0.17179022848077144,
+      "learning_rate": 6.290271770315541e-06,
+      "loss": 0.8497,
+      "step": 674
+    },
+    {
+      "epoch": 4.185471406491499,
+      "grad_norm": 0.14586862929974034,
+      "learning_rate": 6.1971546723879864e-06,
+      "loss": 0.859,
+      "step": 675
+    },
+    {
+      "epoch": 4.1916537867078825,
+      "grad_norm": 0.1629793349005496,
+      "learning_rate": 6.104674040432797e-06,
+      "loss": 0.8649,
+      "step": 676
+    },
+    {
+      "epoch": 4.197836166924266,
+      "grad_norm": 0.15123853435524942,
+      "learning_rate": 6.012831615746484e-06,
+      "loss": 0.8536,
+      "step": 677
+    },
+    {
+      "epoch": 4.204018547140649,
+      "grad_norm": 0.14020729925440908,
+      "learning_rate": 5.9216291276088435e-06,
+      "loss": 0.8633,
+      "step": 678
+    },
+    {
+      "epoch": 4.210200927357032,
+      "grad_norm": 0.1342216269142737,
+      "learning_rate": 5.831068293250481e-06,
+      "loss": 0.863,
+      "step": 679
+    },
+    {
+      "epoch": 4.216383307573416,
+      "grad_norm": 0.16252430886540703,
+      "learning_rate": 5.74115081782042e-06,
+      "loss": 0.8623,
+      "step": 680
+    },
+    {
+      "epoch": 4.222565687789799,
+      "grad_norm": 0.16349108022579686,
+      "learning_rate": 5.651878394354091e-06,
+      "loss": 0.8582,
+      "step": 681
+    },
+    {
+      "epoch": 4.228748068006182,
+      "grad_norm": 0.14338160366251965,
+      "learning_rate": 5.563252703741322e-06,
+      "loss": 0.8565,
+      "step": 682
+    },
+    {
+      "epoch": 4.234930448222566,
+      "grad_norm": 0.12199876601554722,
+      "learning_rate": 5.475275414694787e-06,
+      "loss": 0.8553,
+      "step": 683
+    },
+    {
+      "epoch": 4.241112828438949,
+      "grad_norm": 0.13894410880991992,
+      "learning_rate": 5.387948183718563e-06,
+      "loss": 0.8586,
+      "step": 684
+    },
+    {
+      "epoch": 4.247295208655332,
+      "grad_norm": 0.14581402965942716,
+      "learning_rate": 5.30127265507693e-06,
+      "loss": 0.8639,
+      "step": 685
+    },
+    {
+      "epoch": 4.253477588871716,
+      "grad_norm": 0.12813655780527142,
+      "learning_rate": 5.215250460763397e-06,
+      "loss": 0.845,
+      "step": 686
+    },
+    {
+      "epoch": 4.259659969088099,
+      "grad_norm": 0.1303992841804699,
+      "learning_rate": 5.129883220470007e-06,
+      "loss": 0.8509,
+      "step": 687
+    },
+    {
+      "epoch": 4.265842349304482,
+      "grad_norm": 0.1334156527459534,
+      "learning_rate": 5.045172541556831e-06,
+      "loss": 0.8522,
+      "step": 688
+    },
+    {
+      "epoch": 4.272024729520865,
+      "grad_norm": 0.12398532810622111,
+      "learning_rate": 4.961120019021684e-06,
+      "loss": 0.8647,
+      "step": 689
+    },
+    {
+      "epoch": 4.278207109737249,
+      "grad_norm": 0.12909526271671973,
+      "learning_rate": 4.877727235470113e-06,
+      "loss": 0.8586,
+      "step": 690
+    },
+    {
+      "epoch": 4.284389489953632,
+      "grad_norm": 0.128229338131557,
+      "learning_rate": 4.794995761085593e-06,
+      "loss": 0.8606,
+      "step": 691
+    },
+    {
+      "epoch": 4.290571870170015,
+      "grad_norm": 0.12394373173993793,
+      "learning_rate": 4.712927153599967e-06,
+      "loss": 0.8561,
+      "step": 692
+    },
+    {
+      "epoch": 4.296754250386399,
+      "grad_norm": 0.12226322639491397,
+      "learning_rate": 4.631522958264083e-06,
+      "loss": 0.8526,
+      "step": 693
+    },
+    {
+      "epoch": 4.302936630602782,
+      "grad_norm": 0.11913527209349921,
+      "learning_rate": 4.550784707818756e-06,
+      "loss": 0.8556,
+      "step": 694
+    },
+    {
+      "epoch": 4.309119010819165,
+      "grad_norm": 0.1459462503792828,
+      "learning_rate": 4.47071392246587e-06,
+      "loss": 0.8616,
+      "step": 695
+    },
+    {
+      "epoch": 4.315301391035549,
+      "grad_norm": 0.13037715558952065,
+      "learning_rate": 4.3913121098397675e-06,
+      "loss": 0.8595,
+      "step": 696
+    },
+    {
+      "epoch": 4.321483771251932,
+      "grad_norm": 0.11600080031698429,
+      "learning_rate": 4.312580764978825e-06,
+      "loss": 0.8472,
+      "step": 697
+    },
+    {
+      "epoch": 4.327666151468315,
+      "grad_norm": 0.12978365860295593,
+      "learning_rate": 4.234521370297398e-06,
+      "loss": 0.8652,
+      "step": 698
+    },
+    {
+      "epoch": 4.333848531684699,
+      "grad_norm": 0.13293173492859417,
+      "learning_rate": 4.157135395557786e-06,
+      "loss": 0.862,
+      "step": 699
+    },
+    {
+      "epoch": 4.340030911901082,
+      "grad_norm": 0.1159570605469207,
+      "learning_rate": 4.080424297842656e-06,
+      "loss": 0.8626,
+      "step": 700
+    },
+    {
+      "epoch": 4.346213292117465,
+      "grad_norm": 0.12754643596592866,
+      "learning_rate": 4.004389521527543e-06,
+      "loss": 0.8599,
+      "step": 701
+    },
+    {
+      "epoch": 4.352395672333849,
+      "grad_norm": 0.12442724249604767,
+      "learning_rate": 3.929032498253729e-06,
+      "loss": 0.8643,
+      "step": 702
+    },
+    {
+      "epoch": 4.358578052550232,
+      "grad_norm": 0.12301901657925675,
+      "learning_rate": 3.8543546469011904e-06,
+      "loss": 0.8684,
+      "step": 703
+    },
+    {
+      "epoch": 4.364760432766615,
+      "grad_norm": 0.11793079664110967,
+      "learning_rate": 3.780357373561958e-06,
+      "loss": 0.8554,
+      "step": 704
+    },
+    {
+      "epoch": 4.370942812982999,
+      "grad_norm": 0.11575437385032925,
+      "learning_rate": 3.7070420715136133e-06,
+      "loss": 0.8649,
+      "step": 705
+    },
+    {
+      "epoch": 4.377125193199381,
+      "grad_norm": 0.12884152614107805,
+      "learning_rate": 3.634410121193059e-06,
+      "loss": 0.8549,
+      "step": 706
+    },
+    {
+      "epoch": 4.383307573415765,
+      "grad_norm": 0.13882711090422645,
+      "learning_rate": 3.562462890170526e-06,
+      "loss": 0.848,
+      "step": 707
+    },
+    {
+      "epoch": 4.3894899536321486,
+      "grad_norm": 0.13110201135261,
+      "learning_rate": 3.4912017331238057e-06,
+      "loss": 0.8634,
+      "step": 708
+    },
+    {
+      "epoch": 4.395672333848531,
+      "grad_norm": 0.1192918433930348,
+      "learning_rate": 3.420627991812788e-06,
+      "loss": 0.8595,
+      "step": 709
+    },
+    {
+      "epoch": 4.401854714064915,
+      "grad_norm": 0.12360082779438417,
+      "learning_rate": 3.3507429950541527e-06,
+      "loss": 0.8526,
+      "step": 710
+    },
+    {
+      "epoch": 4.4080370942812985,
+      "grad_norm": 0.12305496739557657,
+      "learning_rate": 3.281548058696373e-06,
+      "loss": 0.8583,
+      "step": 711
+    },
+    {
+      "epoch": 4.414219474497681,
+      "grad_norm": 0.11759210587278442,
+      "learning_rate": 3.2130444855949406e-06,
+      "loss": 0.8671,
+      "step": 712
+    },
+    {
+      "epoch": 4.420401854714065,
+      "grad_norm": 0.12180525882355446,
+      "learning_rate": 3.145233565587824e-06,
+      "loss": 0.8616,
+      "step": 713
+    },
+    {
+      "epoch": 4.4265842349304485,
+      "grad_norm": 0.11198832899126798,
+      "learning_rate": 3.078116575471173e-06,
+      "loss": 0.8566,
+      "step": 714
+    },
+    {
+      "epoch": 4.432766615146831,
+      "grad_norm": 0.11314209890394392,
+      "learning_rate": 3.0116947789753028e-06,
+      "loss": 0.8581,
+      "step": 715
+    },
+    {
+      "epoch": 4.438948995363215,
+      "grad_norm": 0.11662833912400648,
+      "learning_rate": 2.9459694267408977e-06,
+      "loss": 0.8646,
+      "step": 716
+    },
+    {
+      "epoch": 4.4451313755795985,
+      "grad_norm": 0.11975312911303863,
+      "learning_rate": 2.8809417562954435e-06,
+      "loss": 0.8636,
+      "step": 717
+    },
+    {
+      "epoch": 4.451313755795981,
+      "grad_norm": 0.11940652053609146,
+      "learning_rate": 2.8166129920299278e-06,
+      "loss": 0.8474,
+      "step": 718
+    },
+    {
+      "epoch": 4.457496136012365,
+      "grad_norm": 0.11252115918580587,
+      "learning_rate": 2.752984345175809e-06,
+      "loss": 0.848,
+      "step": 719
+    },
+    {
+      "epoch": 4.4636785162287484,
+      "grad_norm": 0.10311476898498252,
+      "learning_rate": 2.690057013782195e-06,
+      "loss": 0.8599,
+      "step": 720
+    },
+    {
+      "epoch": 4.469860896445131,
+      "grad_norm": 0.11653845815565568,
+      "learning_rate": 2.6278321826932818e-06,
+      "loss": 0.8585,
+      "step": 721
+    },
+    {
+      "epoch": 4.476043276661515,
+      "grad_norm": 0.1098739721478331,
+      "learning_rate": 2.566311023526056e-06,
+      "loss": 0.8704,
+      "step": 722
+    },
+    {
+      "epoch": 4.4822256568778975,
+      "grad_norm": 0.11465020117380761,
+      "learning_rate": 2.5054946946482208e-06,
+      "loss": 0.8619,
+      "step": 723
+    },
+    {
+      "epoch": 4.488408037094281,
+      "grad_norm": 0.11796720870498818,
+      "learning_rate": 2.445384341156389e-06,
+      "loss": 0.8498,
+      "step": 724
+    },
+    {
+      "epoch": 4.494590417310665,
+      "grad_norm": 0.1166183962860216,
+      "learning_rate": 2.3859810948545414e-06,
+      "loss": 0.8567,
+      "step": 725
+    },
+    {
+      "epoch": 4.500772797527048,
+      "grad_norm": 0.11682256683745675,
+      "learning_rate": 2.3272860742326798e-06,
+      "loss": 0.8556,
+      "step": 726
+    },
+    {
+      "epoch": 4.506955177743431,
+      "grad_norm": 0.11434906715119132,
+      "learning_rate": 2.269300384445812e-06,
+      "loss": 0.8533,
+      "step": 727
+    },
+    {
+      "epoch": 4.513137557959815,
+      "grad_norm": 0.10735947434712247,
+      "learning_rate": 2.2120251172931082e-06,
+      "loss": 0.856,
+      "step": 728
+    },
+    {
+      "epoch": 4.5193199381761975,
+      "grad_norm": 0.10434949655119075,
+      "learning_rate": 2.15546135119733e-06,
+      "loss": 0.8583,
+      "step": 729
+    },
+    {
+      "epoch": 4.525502318392581,
+      "grad_norm": 0.11530305446016027,
+      "learning_rate": 2.0996101511846056e-06,
+      "loss": 0.8522,
+      "step": 730
+    },
+    {
+      "epoch": 4.531684698608965,
+      "grad_norm": 0.11525723139251923,
+      "learning_rate": 2.0444725688642685e-06,
+      "loss": 0.8537,
+      "step": 731
+    },
+    {
+      "epoch": 4.5378670788253475,
+      "grad_norm": 0.10816720987898426,
+      "learning_rate": 1.9900496424091375e-06,
+      "loss": 0.8552,
+      "step": 732
+    },
+    {
+      "epoch": 4.544049459041731,
+      "grad_norm": 0.10384961737476203,
+      "learning_rate": 1.9363423965359195e-06,
+      "loss": 0.8562,
+      "step": 733
+    },
+    {
+      "epoch": 4.550231839258115,
+      "grad_norm": 0.11336995410635603,
+      "learning_rate": 1.883351842485972e-06,
+      "loss": 0.8506,
+      "step": 734
+    },
+    {
+      "epoch": 4.556414219474497,
+      "grad_norm": 0.1092135826099414,
+      "learning_rate": 1.8310789780061887e-06,
+      "loss": 0.8589,
+      "step": 735
+    },
+    {
+      "epoch": 4.562596599690881,
+      "grad_norm": 0.10993015362867578,
+      "learning_rate": 1.7795247873302735e-06,
+      "loss": 0.8568,
+      "step": 736
+    },
+    {
+      "epoch": 4.568778979907265,
+      "grad_norm": 0.10666335057459857,
+      "learning_rate": 1.728690241160189e-06,
+      "loss": 0.8549,
+      "step": 737
+    },
+    {
+      "epoch": 4.574961360123647,
+      "grad_norm": 0.10922147280384506,
+      "learning_rate": 1.6785762966478715e-06,
+      "loss": 0.8616,
+      "step": 738
+    },
+    {
+      "epoch": 4.581143740340031,
+      "grad_norm": 0.10394964664273364,
+      "learning_rate": 1.6291838973772068e-06,
+      "loss": 0.85,
+      "step": 739
+    },
+    {
+      "epoch": 4.587326120556414,
+      "grad_norm": 0.10207519688840014,
+      "learning_rate": 1.5805139733462827e-06,
+      "loss": 0.8542,
+      "step": 740
+    },
+    {
+      "epoch": 4.593508500772797,
+      "grad_norm": 0.11068267423445127,
+      "learning_rate": 1.532567440949868e-06,
+      "loss": 0.863,
+      "step": 741
+    },
+    {
+      "epoch": 4.599690880989181,
+      "grad_norm": 0.10290817052985828,
+      "learning_rate": 1.4853452029621518e-06,
+      "loss": 0.8582,
+      "step": 742
+    },
+    {
+      "epoch": 4.605873261205565,
+      "grad_norm": 0.10564697594520893,
+      "learning_rate": 1.4388481485197558e-06,
+      "loss": 0.8666,
+      "step": 743
+    },
+    {
+      "epoch": 4.612055641421947,
+      "grad_norm": 0.11348466915609945,
+      "learning_rate": 1.3930771531049847e-06,
+      "loss": 0.8543,
+      "step": 744
+    },
+    {
+      "epoch": 4.618238021638331,
+      "grad_norm": 0.10419274079331056,
+      "learning_rate": 1.3480330785293494e-06,
+      "loss": 0.8605,
+      "step": 745
+    },
+    {
+      "epoch": 4.624420401854714,
+      "grad_norm": 0.10717824788246437,
+      "learning_rate": 1.3037167729173273e-06,
+      "loss": 0.8602,
+      "step": 746
+    },
+    {
+      "epoch": 4.630602782071097,
+      "grad_norm": 0.11005975127499093,
+      "learning_rate": 1.2601290706904102e-06,
+      "loss": 0.8612,
+      "step": 747
+    },
+    {
+      "epoch": 4.636785162287481,
+      "grad_norm": 0.11366201823377237,
+      "learning_rate": 1.2172707925513838e-06,
+      "loss": 0.8692,
+      "step": 748
+    },
+    {
+      "epoch": 4.642967542503864,
+      "grad_norm": 0.10308899025027428,
+      "learning_rate": 1.1751427454688735e-06,
+      "loss": 0.8739,
+      "step": 749
+    },
+    {
+      "epoch": 4.649149922720247,
+      "grad_norm": 0.10280506406459125,
+      "learning_rate": 1.1337457226621518e-06,
+      "loss": 0.8507,
+      "step": 750
+    },
+    {
+      "epoch": 4.655332302936631,
+      "grad_norm": 0.10187117757230453,
+      "learning_rate": 1.0930805035862125e-06,
+      "loss": 0.8688,
+      "step": 751
+    },
+    {
+      "epoch": 4.661514683153014,
+      "grad_norm": 0.10669750138958087,
+      "learning_rate": 1.0531478539170713e-06,
+      "loss": 0.8576,
+      "step": 752
+    },
+    {
+      "epoch": 4.667697063369397,
+      "grad_norm": 0.1101955728854345,
+      "learning_rate": 1.0139485255373826e-06,
+      "loss": 0.8586,
+      "step": 753
+    },
+    {
+      "epoch": 4.673879443585781,
+      "grad_norm": 0.1012943029959148,
+      "learning_rate": 9.75483256522236e-07,
+      "loss": 0.8732,
+      "step": 754
+    },
+    {
+      "epoch": 4.680061823802164,
+      "grad_norm": 0.10388171869974447,
+      "learning_rate": 9.377527711253198e-07,
+      "loss": 0.8501,
+      "step": 755
+    },
+    {
+      "epoch": 4.686244204018547,
+      "grad_norm": 0.10442678790389989,
+      "learning_rate": 9.007577797652245e-07,
+      "loss": 0.8606,
+      "step": 756
+    },
+    {
+      "epoch": 4.69242658423493,
+      "grad_norm": 0.11370097685694985,
+      "learning_rate": 8.644989790121072e-07,
+      "loss": 0.8687,
+      "step": 757
+    },
+    {
+      "epoch": 4.698608964451314,
+      "grad_norm": 0.10057604736658246,
+      "learning_rate": 8.289770515745599e-07,
+      "loss": 0.8623,
+      "step": 758
+    },
+    {
+      "epoch": 4.704791344667697,
+      "grad_norm": 0.10924197746964438,
+      "learning_rate": 7.941926662867528e-07,
+      "loss": 0.8663,
+      "step": 759
+    },
+    {
+      "epoch": 4.710973724884081,
+      "grad_norm": 0.10139852002580828,
+      "learning_rate": 7.60146478095849e-07,
+      "loss": 0.8533,
+      "step": 760
+    },
+    {
+      "epoch": 4.717156105100464,
+      "grad_norm": 0.10196842257139534,
+      "learning_rate": 7.268391280496589e-07,
+      "loss": 0.8545,
+      "step": 761
+    },
+    {
+      "epoch": 4.723338485316847,
+      "grad_norm": 0.09982941082955533,
+      "learning_rate": 6.94271243284601e-07,
+      "loss": 0.8593,
+      "step": 762
+    },
+    {
+      "epoch": 4.72952086553323,
+      "grad_norm": 0.1057680391701117,
+      "learning_rate": 6.624434370138532e-07,
+      "loss": 0.8637,
+      "step": 763
+    },
+    {
+      "epoch": 4.7357032457496135,
+      "grad_norm": 0.09876242839351415,
+      "learning_rate": 6.313563085158425e-07,
+      "loss": 0.8527,
+      "step": 764
+    },
+    {
+      "epoch": 4.741885625965997,
+      "grad_norm": 0.1029622819478206,
+      "learning_rate": 6.010104431229202e-07,
+      "loss": 0.8512,
+      "step": 765
+    },
+    {
+      "epoch": 4.74806800618238,
+      "grad_norm": 0.1017976721996803,
+      "learning_rate": 5.714064122103935e-07,
+      "loss": 0.8535,
+      "step": 766
+    },
+    {
+      "epoch": 4.7542503863987635,
+      "grad_norm": 0.10123837826010473,
+      "learning_rate": 5.425447731857248e-07,
+      "loss": 0.8563,
+      "step": 767
+    },
+    {
+      "epoch": 4.760432766615147,
+      "grad_norm": 0.10371103096875577,
+      "learning_rate": 5.144260694780512e-07,
+      "loss": 0.8558,
+      "step": 768
+    },
+    {
+      "epoch": 4.76661514683153,
+      "grad_norm": 0.10142002208178304,
+      "learning_rate": 4.870508305279531e-07,
+      "loss": 0.8574,
+      "step": 769
+    },
+    {
+      "epoch": 4.7727975270479135,
+      "grad_norm": 0.10174156389331632,
+      "learning_rate": 4.604195717774973e-07,
+      "loss": 0.8678,
+      "step": 770
+    },
+    {
+      "epoch": 4.778979907264297,
+      "grad_norm": 0.10158475219119292,
+      "learning_rate": 4.3453279466049383e-07,
+      "loss": 0.8652,
+      "step": 771
+    },
+    {
+      "epoch": 4.78516228748068,
+      "grad_norm": 0.1037484611064643,
+      "learning_rate": 4.0939098659309895e-07,
+      "loss": 0.8489,
+      "step": 772
+    },
+    {
+      "epoch": 4.7913446676970635,
+      "grad_norm": 0.10162493935552491,
+      "learning_rate": 3.849946209646138e-07,
+      "loss": 0.87,
+      "step": 773
+    },
+    {
+      "epoch": 4.797527047913446,
+      "grad_norm": 0.10099579812699412,
+      "learning_rate": 3.6134415712857587e-07,
+      "loss": 0.861,
+      "step": 774
+    },
+    {
+      "epoch": 4.80370942812983,
+      "grad_norm": 0.10320302386148672,
+      "learning_rate": 3.384400403941124e-07,
+      "loss": 0.8506,
+      "step": 775
+    },
+    {
+      "epoch": 4.809891808346213,
+      "grad_norm": 0.10386437696890127,
+      "learning_rate": 3.1628270201754743e-07,
+      "loss": 0.8498,
+      "step": 776
+    },
+    {
+      "epoch": 4.816074188562597,
+      "grad_norm": 0.09844682031008335,
+      "learning_rate": 2.948725591942925e-07,
+      "loss": 0.8537,
+      "step": 777
+    },
+    {
+      "epoch": 4.82225656877898,
+      "grad_norm": 0.10232392208050557,
+      "learning_rate": 2.742100150509819e-07,
+      "loss": 0.8668,
+      "step": 778
+    },
+    {
+      "epoch": 4.828438948995363,
+      "grad_norm": 0.09744721925527271,
+      "learning_rate": 2.542954586378921e-07,
+      "loss": 0.8657,
+      "step": 779
+    },
+    {
+      "epoch": 4.834621329211746,
+      "grad_norm": 0.10125314166249126,
+      "learning_rate": 2.351292649216097e-07,
+      "loss": 0.8631,
+      "step": 780
+    },
+    {
+      "epoch": 4.84080370942813,
+      "grad_norm": 0.09816843886551711,
+      "learning_rate": 2.167117947779751e-07,
+      "loss": 0.8657,
+      "step": 781
+    },
+    {
+      "epoch": 4.846986089644513,
+      "grad_norm": 0.10038235663832168,
+      "learning_rate": 1.990433949852788e-07,
+      "loss": 0.8666,
+      "step": 782
+    },
+    {
+      "epoch": 4.853168469860896,
+      "grad_norm": 0.09799020116935744,
+      "learning_rate": 1.821243982177423e-07,
+      "loss": 0.8635,
+      "step": 783
+    },
+    {
+      "epoch": 4.85935085007728,
+      "grad_norm": 0.10142283764104856,
+      "learning_rate": 1.6595512303925199e-07,
+      "loss": 0.8496,
+      "step": 784
+    },
+    {
+      "epoch": 4.865533230293663,
+      "grad_norm": 0.10199837045642687,
+      "learning_rate": 1.5053587389735502e-07,
+      "loss": 0.8764,
+      "step": 785
+    },
+    {
+      "epoch": 4.871715610510046,
+      "grad_norm": 0.10337397487378153,
+      "learning_rate": 1.358669411175395e-07,
+      "loss": 0.8548,
+      "step": 786
+    },
+    {
+      "epoch": 4.87789799072643,
+      "grad_norm": 0.09793020007211001,
+      "learning_rate": 1.2194860089774995e-07,
+      "loss": 0.8653,
+      "step": 787
+    },
+    {
+      "epoch": 4.884080370942813,
+      "grad_norm": 0.09878155037542055,
+      "learning_rate": 1.0878111530320478e-07,
+      "loss": 0.8623,
+      "step": 788
+    },
+    {
+      "epoch": 4.890262751159196,
+      "grad_norm": 0.10147729772509538,
+      "learning_rate": 9.636473226144916e-08,
+      "loss": 0.8614,
+      "step": 789
+    },
+    {
+      "epoch": 4.89644513137558,
+      "grad_norm": 0.09737787354641762,
+      "learning_rate": 8.469968555769648e-08,
+      "loss": 0.8647,
+      "step": 790
+    },
+    {
+      "epoch": 4.902627511591962,
+      "grad_norm": 0.09933111257188475,
+      "learning_rate": 7.37861948304186e-08,
+      "loss": 0.8482,
+      "step": 791
+    },
+    {
+      "epoch": 4.908809891808346,
+      "grad_norm": 0.09569900244500511,
+      "learning_rate": 6.362446556720691e-08,
+      "loss": 0.851,
+      "step": 792
+    },
+    {
+      "epoch": 4.91499227202473,
+      "grad_norm": 0.1031116966509426,
+      "learning_rate": 5.421468910092209e-08,
+      "loss": 0.8663,
+      "step": 793
+    },
+    {
+      "epoch": 4.921174652241113,
+      "grad_norm": 0.11035260250280782,
+      "learning_rate": 4.555704260607474e-08,
+      "loss": 0.8488,
+      "step": 794
+    },
+    {
+      "epoch": 4.927357032457496,
+      "grad_norm": 0.09621566038424581,
+      "learning_rate": 3.765168909548589e-08,
+      "loss": 0.8606,
+      "step": 795
+    },
+    {
+      "epoch": 4.93353941267388,
+      "grad_norm": 0.10155877681943523,
+      "learning_rate": 3.049877741723606e-08,
+      "loss": 0.8639,
+      "step": 796
+    },
+    {
+      "epoch": 4.939721792890262,
+      "grad_norm": 0.10009225600988104,
+      "learning_rate": 2.4098442251849762e-08,
+      "loss": 0.8559,
+      "step": 797
+    },
+    {
+      "epoch": 4.945904173106646,
+      "grad_norm": 0.10209104156755659,
+      "learning_rate": 1.8450804109759745e-08,
+      "loss": 0.8421,
+      "step": 798
+    },
+    {
+      "epoch": 4.95208655332303,
+      "grad_norm": 0.10570200426208885,
+      "learning_rate": 1.3555969329037688e-08,
+      "loss": 0.8565,
+      "step": 799
+    },
+    {
+      "epoch": 4.958268933539412,
+      "grad_norm": 0.09810578254731156,
+      "learning_rate": 9.41403007340025e-09,
+      "loss": 0.8638,
+      "step": 800
+    },
+    {
+      "epoch": 4.964451313755796,
+      "grad_norm": 0.1002862128277217,
+      "learning_rate": 6.025064330463792e-09,
+      "loss": 0.8553,
+      "step": 801
+    },
+    {
+      "epoch": 4.97063369397218,
+      "grad_norm": 0.09978110120418406,
+      "learning_rate": 3.3891359102877773e-09,
+      "loss": 0.8633,
+      "step": 802
+    },
+    {
+      "epoch": 4.976816074188562,
+      "grad_norm": 0.09867622221883632,
+      "learning_rate": 1.506294444153511e-09,
+      "loss": 0.8764,
+      "step": 803
+    },
+    {
+      "epoch": 4.982998454404946,
+      "grad_norm": 0.10760151986299173,
+      "learning_rate": 3.7657538364932467e-10,
+      "loss": 0.8589,
+      "step": 804
+    },
+    {
+      "epoch": 4.9891808346213296,
+      "grad_norm": 0.10304725530595252,
+      "learning_rate": 0.0,
+      "loss": 0.8693,
+      "step": 805
+    },
+    {
+      "epoch": 4.9891808346213296,
+      "step": 805,
+      "total_flos": 1.657473506467224e+19,
+      "train_loss": 0.9660356465333737,
+      "train_runtime": 42578.0212,
+      "train_samples_per_second": 9.723,
+      "train_steps_per_second": 0.019
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 805,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.657473506467224e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}