diff --git "a/checkpoint-565/trainer_state.json" "b/checkpoint-565/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-565/trainer_state.json"
@@ -0,0 +1,3988 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.988962472406181,
+  "eval_steps": 500,
+  "global_step": 565,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.008830022075055188,
+      "grad_norm": 25.06905818289659,
+      "learning_rate": 2.9411764705882356e-07,
+      "loss": 1.9387,
+      "step": 1
+    },
+    {
+      "epoch": 0.017660044150110375,
+      "grad_norm": 25.68568630022093,
+      "learning_rate": 5.882352941176471e-07,
+      "loss": 1.928,
+      "step": 2
+    },
+    {
+      "epoch": 0.026490066225165563,
+      "grad_norm": 26.502734598525898,
+      "learning_rate": 8.823529411764707e-07,
+      "loss": 1.9155,
+      "step": 3
+    },
+    {
+      "epoch": 0.03532008830022075,
+      "grad_norm": 25.310906985150506,
+      "learning_rate": 1.1764705882352942e-06,
+      "loss": 1.9235,
+      "step": 4
+    },
+    {
+      "epoch": 0.04415011037527594,
+      "grad_norm": 24.72651604573988,
+      "learning_rate": 1.4705882352941177e-06,
+      "loss": 1.8924,
+      "step": 5
+    },
+    {
+      "epoch": 0.052980132450331126,
+      "grad_norm": 24.205412937319906,
+      "learning_rate": 1.7647058823529414e-06,
+      "loss": 1.8026,
+      "step": 6
+    },
+    {
+      "epoch": 0.06181015452538632,
+      "grad_norm": 22.15412876974167,
+      "learning_rate": 2.058823529411765e-06,
+      "loss": 1.5564,
+      "step": 7
+    },
+    {
+      "epoch": 0.0706401766004415,
+      "grad_norm": 19.764774939212682,
+      "learning_rate": 2.3529411764705885e-06,
+      "loss": 1.4926,
+      "step": 8
+    },
+    {
+      "epoch": 0.07947019867549669,
+      "grad_norm": 18.085419039016024,
+      "learning_rate": 2.647058823529412e-06,
+      "loss": 1.036,
+      "step": 9
+    },
+    {
+      "epoch": 0.08830022075055188,
+      "grad_norm": 15.607291712181967,
+      "learning_rate": 2.9411764705882355e-06,
+      "loss": 0.9977,
+      "step": 10
+    },
+    {
+      "epoch": 0.09713024282560706,
+      "grad_norm": 14.945489252455502,
+      "learning_rate": 3.2352941176470594e-06,
+      "loss": 0.9075,
+      "step": 11
+    },
+    {
+      "epoch": 0.10596026490066225,
+      "grad_norm": 27.120860454783713,
+      "learning_rate": 3.529411764705883e-06,
+      "loss": 0.503,
+      "step": 12
+    },
+    {
+      "epoch": 0.11479028697571744,
+      "grad_norm": 13.65940891061477,
+      "learning_rate": 3.8235294117647055e-06,
+      "loss": 0.4615,
+      "step": 13
+    },
+    {
+      "epoch": 0.12362030905077263,
+      "grad_norm": 13.175755458095395,
+      "learning_rate": 4.11764705882353e-06,
+      "loss": 0.4129,
+      "step": 14
+    },
+    {
+      "epoch": 0.13245033112582782,
+      "grad_norm": 11.148898315396657,
+      "learning_rate": 4.411764705882353e-06,
+      "loss": 0.3516,
+      "step": 15
+    },
+    {
+      "epoch": 0.141280353200883,
+      "grad_norm": 7.428424652374598,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 0.3313,
+      "step": 16
+    },
+    {
+      "epoch": 0.15011037527593818,
+      "grad_norm": 11.107291096031085,
+      "learning_rate": 5e-06,
+      "loss": 0.322,
+      "step": 17
+    },
+    {
+      "epoch": 0.15894039735099338,
+      "grad_norm": 8.331582263755678,
+      "learning_rate": 5.294117647058824e-06,
+      "loss": 0.2905,
+      "step": 18
+    },
+    {
+      "epoch": 0.16777041942604856,
+      "grad_norm": 7.123373742560772,
+      "learning_rate": 5.588235294117647e-06,
+      "loss": 0.2977,
+      "step": 19
+    },
+    {
+      "epoch": 0.17660044150110377,
+      "grad_norm": 5.2783172446030395,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 0.2857,
+      "step": 20
+    },
+    {
+      "epoch": 0.18543046357615894,
+      "grad_norm": 4.180778519803163,
+      "learning_rate": 6.176470588235295e-06,
+      "loss": 0.285,
+      "step": 21
+    },
+    {
+      "epoch": 0.19426048565121412,
+      "grad_norm": 4.943481049285879,
+      "learning_rate": 6.470588235294119e-06,
+      "loss": 0.2707,
+      "step": 22
+    },
+    {
+      "epoch": 0.20309050772626933,
+      "grad_norm": 6.261139315191096,
+      "learning_rate": 6.764705882352942e-06,
+      "loss": 0.2713,
+      "step": 23
+    },
+    {
+      "epoch": 0.2119205298013245,
+      "grad_norm": 5.205543023717728,
+      "learning_rate": 7.058823529411766e-06,
+      "loss": 0.3015,
+      "step": 24
+    },
+    {
+      "epoch": 0.22075055187637968,
+      "grad_norm": 7.1297538914315925,
+      "learning_rate": 7.352941176470589e-06,
+      "loss": 0.2687,
+      "step": 25
+    },
+    {
+      "epoch": 0.22958057395143489,
+      "grad_norm": 9.380313333856593,
+      "learning_rate": 7.647058823529411e-06,
+      "loss": 0.2752,
+      "step": 26
+    },
+    {
+      "epoch": 0.23841059602649006,
+      "grad_norm": 5.461368987186966,
+      "learning_rate": 7.941176470588236e-06,
+      "loss": 0.268,
+      "step": 27
+    },
+    {
+      "epoch": 0.24724061810154527,
+      "grad_norm": 6.130043348495507,
+      "learning_rate": 8.23529411764706e-06,
+      "loss": 0.2691,
+      "step": 28
+    },
+    {
+      "epoch": 0.2560706401766004,
+      "grad_norm": 4.667685505341346,
+      "learning_rate": 8.529411764705883e-06,
+      "loss": 0.2696,
+      "step": 29
+    },
+    {
+      "epoch": 0.26490066225165565,
+      "grad_norm": 4.33954165829827,
+      "learning_rate": 8.823529411764707e-06,
+      "loss": 0.2782,
+      "step": 30
+    },
+    {
+      "epoch": 0.2737306843267108,
+      "grad_norm": 4.142057835286045,
+      "learning_rate": 9.11764705882353e-06,
+      "loss": 0.2661,
+      "step": 31
+    },
+    {
+      "epoch": 0.282560706401766,
+      "grad_norm": 5.125587426107246,
+      "learning_rate": 9.411764705882354e-06,
+      "loss": 0.2658,
+      "step": 32
+    },
+    {
+      "epoch": 0.2913907284768212,
+      "grad_norm": 6.160516391015107,
+      "learning_rate": 9.705882352941177e-06,
+      "loss": 0.2724,
+      "step": 33
+    },
+    {
+      "epoch": 0.30022075055187636,
+      "grad_norm": 7.3062949667119215,
+      "learning_rate": 1e-05,
+      "loss": 0.2795,
+      "step": 34
+    },
+    {
+      "epoch": 0.3090507726269316,
+      "grad_norm": 4.75477126908679,
+      "learning_rate": 9.99997945915297e-06,
+      "loss": 0.2771,
+      "step": 35
+    },
+    {
+      "epoch": 0.31788079470198677,
+      "grad_norm": 6.469930861735197,
+      "learning_rate": 9.999917836780642e-06,
+      "loss": 0.2746,
+      "step": 36
+    },
+    {
+      "epoch": 0.32671081677704195,
+      "grad_norm": 6.774915528699888,
+      "learning_rate": 9.999815133389334e-06,
+      "loss": 0.2625,
+      "step": 37
+    },
+    {
+      "epoch": 0.3355408388520971,
+      "grad_norm": 3.532349122431978,
+      "learning_rate": 9.999671349822887e-06,
+      "loss": 0.2574,
+      "step": 38
+    },
+    {
+      "epoch": 0.3443708609271523,
+      "grad_norm": 4.296282052033279,
+      "learning_rate": 9.999486487262677e-06,
+      "loss": 0.2878,
+      "step": 39
+    },
+    {
+      "epoch": 0.35320088300220753,
+      "grad_norm": 5.6368589176942665,
+      "learning_rate": 9.999260547227599e-06,
+      "loss": 0.2695,
+      "step": 40
+    },
+    {
+      "epoch": 0.3620309050772627,
+      "grad_norm": 5.904653496281388,
+      "learning_rate": 9.998993531574048e-06,
+      "loss": 0.26,
+      "step": 41
+    },
+    {
+      "epoch": 0.3708609271523179,
+      "grad_norm": 6.201716230094972,
+      "learning_rate": 9.998685442495921e-06,
+      "loss": 0.277,
+      "step": 42
+    },
+    {
+      "epoch": 0.37969094922737306,
+      "grad_norm": 3.9609425619162084,
+      "learning_rate": 9.998336282524579e-06,
+      "loss": 0.2445,
+      "step": 43
+    },
+    {
+      "epoch": 0.38852097130242824,
+      "grad_norm": 4.548164086756688,
+      "learning_rate": 9.997946054528837e-06,
+      "loss": 0.249,
+      "step": 44
+    },
+    {
+      "epoch": 0.3973509933774834,
+      "grad_norm": 5.446596078550373,
+      "learning_rate": 9.997514761714946e-06,
+      "loss": 0.2644,
+      "step": 45
+    },
+    {
+      "epoch": 0.40618101545253865,
+      "grad_norm": 6.853964332896531,
+      "learning_rate": 9.99704240762655e-06,
+      "loss": 0.2407,
+      "step": 46
+    },
+    {
+      "epoch": 0.41501103752759383,
+      "grad_norm": 5.248615557746856,
+      "learning_rate": 9.996528996144668e-06,
+      "loss": 0.2343,
+      "step": 47
+    },
+    {
+      "epoch": 0.423841059602649,
+      "grad_norm": 3.851621358245188,
+      "learning_rate": 9.995974531487668e-06,
+      "loss": 0.2456,
+      "step": 48
+    },
+    {
+      "epoch": 0.4326710816777042,
+      "grad_norm": 7.360655763951351,
+      "learning_rate": 9.995379018211215e-06,
+      "loss": 0.2591,
+      "step": 49
+    },
+    {
+      "epoch": 0.44150110375275936,
+      "grad_norm": 3.8007967239727547,
+      "learning_rate": 9.994742461208251e-06,
+      "loss": 0.2464,
+      "step": 50
+    },
+    {
+      "epoch": 0.4503311258278146,
+      "grad_norm": 6.54251380380296,
+      "learning_rate": 9.994064865708944e-06,
+      "loss": 0.2474,
+      "step": 51
+    },
+    {
+      "epoch": 0.45916114790286977,
+      "grad_norm": 4.345236663462358,
+      "learning_rate": 9.993346237280646e-06,
+      "loss": 0.241,
+      "step": 52
+    },
+    {
+      "epoch": 0.46799116997792495,
+      "grad_norm": 4.340752142993626,
+      "learning_rate": 9.992586581827853e-06,
+      "loss": 0.2445,
+      "step": 53
+    },
+    {
+      "epoch": 0.4768211920529801,
+      "grad_norm": 8.680644031503022,
+      "learning_rate": 9.991785905592149e-06,
+      "loss": 0.2382,
+      "step": 54
+    },
+    {
+      "epoch": 0.4856512141280353,
+      "grad_norm": 3.9551260449633228,
+      "learning_rate": 9.990944215152166e-06,
+      "loss": 0.2415,
+      "step": 55
+    },
+    {
+      "epoch": 0.49448123620309054,
+      "grad_norm": 4.393232827634922,
+      "learning_rate": 9.990061517423513e-06,
+      "loss": 0.2348,
+      "step": 56
+    },
+    {
+      "epoch": 0.5033112582781457,
+      "grad_norm": 4.711307408468823,
+      "learning_rate": 9.989137819658738e-06,
+      "loss": 0.2453,
+      "step": 57
+    },
+    {
+      "epoch": 0.5121412803532008,
+      "grad_norm": 3.562557321457758,
+      "learning_rate": 9.988173129447251e-06,
+      "loss": 0.2544,
+      "step": 58
+    },
+    {
+      "epoch": 0.5209713024282561,
+      "grad_norm": 3.1473783838712714,
+      "learning_rate": 9.987167454715277e-06,
+      "loss": 0.2496,
+      "step": 59
+    },
+    {
+      "epoch": 0.5298013245033113,
+      "grad_norm": 4.258342746195014,
+      "learning_rate": 9.986120803725776e-06,
+      "loss": 0.2417,
+      "step": 60
+    },
+    {
+      "epoch": 0.5386313465783664,
+      "grad_norm": 3.0661639870207944,
+      "learning_rate": 9.985033185078392e-06,
+      "loss": 0.2416,
+      "step": 61
+    },
+    {
+      "epoch": 0.5474613686534217,
+      "grad_norm": 6.350094923609783,
+      "learning_rate": 9.983904607709365e-06,
+      "loss": 0.2364,
+      "step": 62
+    },
+    {
+      "epoch": 0.5562913907284768,
+      "grad_norm": 3.218801211881891,
+      "learning_rate": 9.982735080891471e-06,
+      "loss": 0.2427,
+      "step": 63
+    },
+    {
+      "epoch": 0.565121412803532,
+      "grad_norm": 4.189964625089664,
+      "learning_rate": 9.981524614233938e-06,
+      "loss": 0.2414,
+      "step": 64
+    },
+    {
+      "epoch": 0.5739514348785872,
+      "grad_norm": 4.295839759826754,
+      "learning_rate": 9.98027321768237e-06,
+      "loss": 0.2334,
+      "step": 65
+    },
+    {
+      "epoch": 0.5827814569536424,
+      "grad_norm": 5.620588539030716,
+      "learning_rate": 9.978980901518663e-06,
+      "loss": 0.2594,
+      "step": 66
+    },
+    {
+      "epoch": 0.5916114790286976,
+      "grad_norm": 3.0744693491498944,
+      "learning_rate": 9.977647676360927e-06,
+      "loss": 0.2394,
+      "step": 67
+    },
+    {
+      "epoch": 0.6004415011037527,
+      "grad_norm": 6.359908476906296,
+      "learning_rate": 9.976273553163393e-06,
+      "loss": 0.231,
+      "step": 68
+    },
+    {
+      "epoch": 0.609271523178808,
+      "grad_norm": 6.316442880887049,
+      "learning_rate": 9.974858543216319e-06,
+      "loss": 0.2274,
+      "step": 69
+    },
+    {
+      "epoch": 0.6181015452538632,
+      "grad_norm": 5.324156655672741,
+      "learning_rate": 9.973402658145908e-06,
+      "loss": 0.2288,
+      "step": 70
+    },
+    {
+      "epoch": 0.6269315673289183,
+      "grad_norm": 3.9538127249812374,
+      "learning_rate": 9.971905909914206e-06,
+      "loss": 0.2395,
+      "step": 71
+    },
+    {
+      "epoch": 0.6357615894039735,
+      "grad_norm": 4.015855350861166,
+      "learning_rate": 9.970368310819e-06,
+      "loss": 0.2078,
+      "step": 72
+    },
+    {
+      "epoch": 0.6445916114790287,
+      "grad_norm": 4.719041520529964,
+      "learning_rate": 9.96878987349373e-06,
+      "loss": 0.2311,
+      "step": 73
+    },
+    {
+      "epoch": 0.6534216335540839,
+      "grad_norm": 4.1070724317700975,
+      "learning_rate": 9.96717061090737e-06,
+      "loss": 0.2262,
+      "step": 74
+    },
+    {
+      "epoch": 0.6622516556291391,
+      "grad_norm": 6.469366294468439,
+      "learning_rate": 9.965510536364329e-06,
+      "loss": 0.2151,
+      "step": 75
+    },
+    {
+      "epoch": 0.6710816777041942,
+      "grad_norm": 4.930762891482967,
+      "learning_rate": 9.96380966350434e-06,
+      "loss": 0.2335,
+      "step": 76
+    },
+    {
+      "epoch": 0.6799116997792495,
+      "grad_norm": 3.6682691137381753,
+      "learning_rate": 9.962068006302357e-06,
+      "loss": 0.2228,
+      "step": 77
+    },
+    {
+      "epoch": 0.6887417218543046,
+      "grad_norm": 3.8312756603355096,
+      "learning_rate": 9.960285579068419e-06,
+      "loss": 0.2376,
+      "step": 78
+    },
+    {
+      "epoch": 0.6975717439293598,
+      "grad_norm": 5.690370014674548,
+      "learning_rate": 9.958462396447556e-06,
+      "loss": 0.2254,
+      "step": 79
+    },
+    {
+      "epoch": 0.7064017660044151,
+      "grad_norm": 7.10575289465107,
+      "learning_rate": 9.956598473419652e-06,
+      "loss": 0.2367,
+      "step": 80
+    },
+    {
+      "epoch": 0.7152317880794702,
+      "grad_norm": 3.786799193333354,
+      "learning_rate": 9.954693825299333e-06,
+      "loss": 0.2507,
+      "step": 81
+    },
+    {
+      "epoch": 0.7240618101545254,
+      "grad_norm": 4.164532105917261,
+      "learning_rate": 9.95274846773583e-06,
+      "loss": 0.2304,
+      "step": 82
+    },
+    {
+      "epoch": 0.7328918322295805,
+      "grad_norm": 2.957167457981859,
+      "learning_rate": 9.950762416712862e-06,
+      "loss": 0.2309,
+      "step": 83
+    },
+    {
+      "epoch": 0.7417218543046358,
+      "grad_norm": 3.950243607284656,
+      "learning_rate": 9.948735688548496e-06,
+      "loss": 0.2394,
+      "step": 84
+    },
+    {
+      "epoch": 0.7505518763796909,
+      "grad_norm": 4.199393340080547,
+      "learning_rate": 9.946668299895017e-06,
+      "loss": 0.2232,
+      "step": 85
+    },
+    {
+      "epoch": 0.7593818984547461,
+      "grad_norm": 5.353432006990455,
+      "learning_rate": 9.944560267738792e-06,
+      "loss": 0.2295,
+      "step": 86
+    },
+    {
+      "epoch": 0.7682119205298014,
+      "grad_norm": 5.187956574955682,
+      "learning_rate": 9.942411609400127e-06,
+      "loss": 0.2366,
+      "step": 87
+    },
+    {
+      "epoch": 0.7770419426048565,
+      "grad_norm": 3.857122403820624,
+      "learning_rate": 9.940222342533126e-06,
+      "loss": 0.2302,
+      "step": 88
+    },
+    {
+      "epoch": 0.7858719646799117,
+      "grad_norm": 3.8235009485898135,
+      "learning_rate": 9.937992485125547e-06,
+      "loss": 0.2221,
+      "step": 89
+    },
+    {
+      "epoch": 0.7947019867549668,
+      "grad_norm": 4.994710419417956,
+      "learning_rate": 9.935722055498655e-06,
+      "loss": 0.2441,
+      "step": 90
+    },
+    {
+      "epoch": 0.8035320088300221,
+      "grad_norm": 3.6274639946148177,
+      "learning_rate": 9.933411072307071e-06,
+      "loss": 0.2126,
+      "step": 91
+    },
+    {
+      "epoch": 0.8123620309050773,
+      "grad_norm": 3.494744001694175,
+      "learning_rate": 9.931059554538613e-06,
+      "loss": 0.2329,
+      "step": 92
+    },
+    {
+      "epoch": 0.8211920529801324,
+      "grad_norm": 3.78348592809509,
+      "learning_rate": 9.928667521514149e-06,
+      "loss": 0.2127,
+      "step": 93
+    },
+    {
+      "epoch": 0.8300220750551877,
+      "grad_norm": 5.340365599979518,
+      "learning_rate": 9.92623499288743e-06,
+      "loss": 0.2222,
+      "step": 94
+    },
+    {
+      "epoch": 0.8388520971302428,
+      "grad_norm": 6.5579314449709445,
+      "learning_rate": 9.92376198864494e-06,
+      "loss": 0.2183,
+      "step": 95
+    },
+    {
+      "epoch": 0.847682119205298,
+      "grad_norm": 4.959651982342057,
+      "learning_rate": 9.921248529105716e-06,
+      "loss": 0.223,
+      "step": 96
+    },
+    {
+      "epoch": 0.8565121412803532,
+      "grad_norm": 3.4308915999621528,
+      "learning_rate": 9.918694634921195e-06,
+      "loss": 0.2263,
+      "step": 97
+    },
+    {
+      "epoch": 0.8653421633554084,
+      "grad_norm": 4.279324835441828,
+      "learning_rate": 9.916100327075038e-06,
+      "loss": 0.2106,
+      "step": 98
+    },
+    {
+      "epoch": 0.8741721854304636,
+      "grad_norm": 5.118504467469523,
+      "learning_rate": 9.913465626882954e-06,
+      "loss": 0.2313,
+      "step": 99
+    },
+    {
+      "epoch": 0.8830022075055187,
+      "grad_norm": 5.611920388270936,
+      "learning_rate": 9.910790555992536e-06,
+      "loss": 0.2241,
+      "step": 100
+    },
+    {
+      "epoch": 0.891832229580574,
+      "grad_norm": 3.201069614707369,
+      "learning_rate": 9.908075136383068e-06,
+      "loss": 0.2045,
+      "step": 101
+    },
+    {
+      "epoch": 0.9006622516556292,
+      "grad_norm": 6.023860874306649,
+      "learning_rate": 9.905319390365364e-06,
+      "loss": 0.2361,
+      "step": 102
+    },
+    {
+      "epoch": 0.9094922737306843,
+      "grad_norm": 4.541003622505249,
+      "learning_rate": 9.902523340581562e-06,
+      "loss": 0.2231,
+      "step": 103
+    },
+    {
+      "epoch": 0.9183222958057395,
+      "grad_norm": 3.892822194098379,
+      "learning_rate": 9.899687010004956e-06,
+      "loss": 0.2346,
+      "step": 104
+    },
+    {
+      "epoch": 0.9271523178807947,
+      "grad_norm": 4.550335472806411,
+      "learning_rate": 9.896810421939797e-06,
+      "loss": 0.2332,
+      "step": 105
+    },
+    {
+      "epoch": 0.9359823399558499,
+      "grad_norm": 3.85828415445978,
+      "learning_rate": 9.893893600021112e-06,
+      "loss": 0.2159,
+      "step": 106
+    },
+    {
+      "epoch": 0.9448123620309051,
+      "grad_norm": 5.446836037485189,
+      "learning_rate": 9.890936568214493e-06,
+      "loss": 0.2336,
+      "step": 107
+    },
+    {
+      "epoch": 0.9536423841059603,
+      "grad_norm": 5.648084319092979,
+      "learning_rate": 9.88793935081592e-06,
+      "loss": 0.219,
+      "step": 108
+    },
+    {
+      "epoch": 0.9624724061810155,
+      "grad_norm": 10.700799326300302,
+      "learning_rate": 9.884901972451542e-06,
+      "loss": 0.2257,
+      "step": 109
+    },
+    {
+      "epoch": 0.9713024282560706,
+      "grad_norm": 6.281034440727802,
+      "learning_rate": 9.881824458077491e-06,
+      "loss": 0.2295,
+      "step": 110
+    },
+    {
+      "epoch": 0.9801324503311258,
+      "grad_norm": 4.701514002631264,
+      "learning_rate": 9.878706832979668e-06,
+      "loss": 0.2129,
+      "step": 111
+    },
+    {
+      "epoch": 0.9889624724061811,
+      "grad_norm": 4.383624556135786,
+      "learning_rate": 9.875549122773536e-06,
+      "loss": 0.2312,
+      "step": 112
+    },
+    {
+      "epoch": 0.9977924944812362,
+      "grad_norm": 3.576068647188603,
+      "learning_rate": 9.872351353403912e-06,
+      "loss": 0.2141,
+      "step": 113
+    },
+    {
+      "epoch": 1.0066225165562914,
+      "grad_norm": 6.0830817761682985,
+      "learning_rate": 9.869113551144754e-06,
+      "loss": 0.2014,
+      "step": 114
+    },
+    {
+      "epoch": 1.0154525386313467,
+      "grad_norm": 4.7531813854125655,
+      "learning_rate": 9.865835742598942e-06,
+      "loss": 0.2022,
+      "step": 115
+    },
+    {
+      "epoch": 1.0242825607064017,
+      "grad_norm": 3.2824105375326478,
+      "learning_rate": 9.86251795469806e-06,
+      "loss": 0.1814,
+      "step": 116
+    },
+    {
+      "epoch": 1.033112582781457,
+      "grad_norm": 2.786895822375094,
+      "learning_rate": 9.85916021470218e-06,
+      "loss": 0.1898,
+      "step": 117
+    },
+    {
+      "epoch": 1.0419426048565121,
+      "grad_norm": 4.098036626468437,
+      "learning_rate": 9.85576255019963e-06,
+      "loss": 0.1923,
+      "step": 118
+    },
+    {
+      "epoch": 1.0507726269315674,
+      "grad_norm": 3.916652138771565,
+      "learning_rate": 9.852324989106772e-06,
+      "loss": 0.1877,
+      "step": 119
+    },
+    {
+      "epoch": 1.0596026490066226,
+      "grad_norm": 5.455813349247477,
+      "learning_rate": 9.848847559667774e-06,
+      "loss": 0.1846,
+      "step": 120
+    },
+    {
+      "epoch": 1.0684326710816776,
+      "grad_norm": 10.542661969730387,
+      "learning_rate": 9.845330290454373e-06,
+      "loss": 0.1718,
+      "step": 121
+    },
+    {
+      "epoch": 1.0772626931567328,
+      "grad_norm": 4.02767225020254,
+      "learning_rate": 9.841773210365646e-06,
+      "loss": 0.19,
+      "step": 122
+    },
+    {
+      "epoch": 1.086092715231788,
+      "grad_norm": 6.839436084957277,
+      "learning_rate": 9.838176348627768e-06,
+      "loss": 0.1708,
+      "step": 123
+    },
+    {
+      "epoch": 1.0949227373068433,
+      "grad_norm": 4.798409291630464,
+      "learning_rate": 9.834539734793774e-06,
+      "loss": 0.1846,
+      "step": 124
+    },
+    {
+      "epoch": 1.1037527593818985,
+      "grad_norm": 4.137394904622889,
+      "learning_rate": 9.830863398743313e-06,
+      "loss": 0.1789,
+      "step": 125
+    },
+    {
+      "epoch": 1.1125827814569536,
+      "grad_norm": 3.88794974694025,
+      "learning_rate": 9.82714737068241e-06,
+      "loss": 0.1737,
+      "step": 126
+    },
+    {
+      "epoch": 1.1214128035320088,
+      "grad_norm": 3.4537997696122362,
+      "learning_rate": 9.82339168114321e-06,
+      "loss": 0.1814,
+      "step": 127
+    },
+    {
+      "epoch": 1.130242825607064,
+      "grad_norm": 3.743862822832632,
+      "learning_rate": 9.81959636098373e-06,
+      "loss": 0.1776,
+      "step": 128
+    },
+    {
+      "epoch": 1.1390728476821192,
+      "grad_norm": 3.64783731609154,
+      "learning_rate": 9.815761441387609e-06,
+      "loss": 0.1863,
+      "step": 129
+    },
+    {
+      "epoch": 1.1479028697571745,
+      "grad_norm": 4.945406809993386,
+      "learning_rate": 9.811886953863841e-06,
+      "loss": 0.1948,
+      "step": 130
+    },
+    {
+      "epoch": 1.1567328918322295,
+      "grad_norm": 3.995572235306309,
+      "learning_rate": 9.807972930246531e-06,
+      "loss": 0.189,
+      "step": 131
+    },
+    {
+      "epoch": 1.1655629139072847,
+      "grad_norm": 5.315570871040949,
+      "learning_rate": 9.804019402694627e-06,
+      "loss": 0.1878,
+      "step": 132
+    },
+    {
+      "epoch": 1.17439293598234,
+      "grad_norm": 7.942697593904259,
+      "learning_rate": 9.800026403691643e-06,
+      "loss": 0.1822,
+      "step": 133
+    },
+    {
+      "epoch": 1.1832229580573952,
+      "grad_norm": 3.7100381437064605,
+      "learning_rate": 9.795993966045418e-06,
+      "loss": 0.184,
+      "step": 134
+    },
+    {
+      "epoch": 1.1920529801324504,
+      "grad_norm": 3.515806231432155,
+      "learning_rate": 9.791922122887823e-06,
+      "loss": 0.1557,
+      "step": 135
+    },
+    {
+      "epoch": 1.2008830022075054,
+      "grad_norm": 11.905142017862433,
+      "learning_rate": 9.7878109076745e-06,
+      "loss": 0.1886,
+      "step": 136
+    },
+    {
+      "epoch": 1.2097130242825607,
+      "grad_norm": 5.549791017896183,
+      "learning_rate": 9.783660354184589e-06,
+      "loss": 0.1911,
+      "step": 137
+    },
+    {
+      "epoch": 1.218543046357616,
+      "grad_norm": 3.1831986903773575,
+      "learning_rate": 9.779470496520442e-06,
+      "loss": 0.1845,
+      "step": 138
+    },
+    {
+      "epoch": 1.2273730684326711,
+      "grad_norm": 7.241858960668914,
+      "learning_rate": 9.77524136910735e-06,
+      "loss": 0.1746,
+      "step": 139
+    },
+    {
+      "epoch": 1.2362030905077264,
+      "grad_norm": 3.2346972578304034,
+      "learning_rate": 9.770973006693256e-06,
+      "loss": 0.1519,
+      "step": 140
+    },
+    {
+      "epoch": 1.2450331125827814,
+      "grad_norm": 6.403826372889126,
+      "learning_rate": 9.766665444348472e-06,
+      "loss": 0.1758,
+      "step": 141
+    },
+    {
+      "epoch": 1.2538631346578366,
+      "grad_norm": 4.4524344269491625,
+      "learning_rate": 9.76231871746539e-06,
+      "loss": 0.1979,
+      "step": 142
+    },
+    {
+      "epoch": 1.2626931567328918,
+      "grad_norm": 3.862760163174027,
+      "learning_rate": 9.757932861758188e-06,
+      "loss": 0.2054,
+      "step": 143
+    },
+    {
+      "epoch": 1.271523178807947,
+      "grad_norm": 3.2208460551997553,
+      "learning_rate": 9.753507913262548e-06,
+      "loss": 0.1711,
+      "step": 144
+    },
+    {
+      "epoch": 1.280353200883002,
+      "grad_norm": 3.9685569501042948,
+      "learning_rate": 9.749043908335343e-06,
+      "loss": 0.154,
+      "step": 145
+    },
+    {
+      "epoch": 1.2891832229580573,
+      "grad_norm": 4.700535965091022,
+      "learning_rate": 9.744540883654348e-06,
+      "loss": 0.1803,
+      "step": 146
+    },
+    {
+      "epoch": 1.2980132450331126,
+      "grad_norm": 3.290798769098629,
+      "learning_rate": 9.739998876217943e-06,
+      "loss": 0.1904,
+      "step": 147
+    },
+    {
+      "epoch": 1.3068432671081678,
+      "grad_norm": 4.379650402165434,
+      "learning_rate": 9.735417923344798e-06,
+      "loss": 0.1828,
+      "step": 148
+    },
+    {
+      "epoch": 1.315673289183223,
+      "grad_norm": 4.4512235264983335,
+      "learning_rate": 9.730798062673575e-06,
+      "loss": 0.1963,
+      "step": 149
+    },
+    {
+      "epoch": 1.3245033112582782,
+      "grad_norm": 4.928447363077229,
+      "learning_rate": 9.726139332162613e-06,
+      "loss": 0.1986,
+      "step": 150
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 3.167430281144184,
+      "learning_rate": 9.721441770089621e-06,
+      "loss": 0.1783,
+      "step": 151
+    },
+    {
+      "epoch": 1.3421633554083885,
+      "grad_norm": 5.16299387787309,
+      "learning_rate": 9.716705415051362e-06,
+      "loss": 0.1817,
+      "step": 152
+    },
+    {
+      "epoch": 1.3509933774834437,
+      "grad_norm": 5.22565296604379,
+      "learning_rate": 9.711930305963333e-06,
+      "loss": 0.1606,
+      "step": 153
+    },
+    {
+      "epoch": 1.359823399558499,
+      "grad_norm": 6.989669414948377,
+      "learning_rate": 9.707116482059447e-06,
+      "loss": 0.1924,
+      "step": 154
+    },
+    {
+      "epoch": 1.368653421633554,
+      "grad_norm": 4.288151185944291,
+      "learning_rate": 9.702263982891712e-06,
+      "loss": 0.1651,
+      "step": 155
+    },
+    {
+      "epoch": 1.3774834437086092,
+      "grad_norm": 4.567897047089374,
+      "learning_rate": 9.697372848329905e-06,
+      "loss": 0.1842,
+      "step": 156
+    },
+    {
+      "epoch": 1.3863134657836644,
+      "grad_norm": 4.29424871998317,
+      "learning_rate": 9.692443118561248e-06,
+      "loss": 0.173,
+      "step": 157
+    },
+    {
+      "epoch": 1.3951434878587197,
+      "grad_norm": 7.122508056007231,
+      "learning_rate": 9.68747483409007e-06,
+      "loss": 0.1813,
+      "step": 158
+    },
+    {
+      "epoch": 1.403973509933775,
+      "grad_norm": 20.80552868339202,
+      "learning_rate": 9.682468035737475e-06,
+      "loss": 0.1948,
+      "step": 159
+    },
+    {
+      "epoch": 1.4128035320088301,
+      "grad_norm": 7.039552426013288,
+      "learning_rate": 9.677422764641021e-06,
+      "loss": 0.1782,
+      "step": 160
+    },
+    {
+      "epoch": 1.4216335540838851,
+      "grad_norm": 5.217781974573682,
+      "learning_rate": 9.672339062254359e-06,
+      "loss": 0.1813,
+      "step": 161
+    },
+    {
+      "epoch": 1.4304635761589404,
+      "grad_norm": 3.4648301225351474,
+      "learning_rate": 9.667216970346916e-06,
+      "loss": 0.1789,
+      "step": 162
+    },
+    {
+      "epoch": 1.4392935982339956,
+      "grad_norm": 23.4115100661157,
+      "learning_rate": 9.662056531003528e-06,
+      "loss": 0.1783,
+      "step": 163
+    },
+    {
+      "epoch": 1.4481236203090508,
+      "grad_norm": 4.149843420090105,
+      "learning_rate": 9.656857786624119e-06,
+      "loss": 0.1895,
+      "step": 164
+    },
+    {
+      "epoch": 1.4569536423841059,
+      "grad_norm": 6.867974570202788,
+      "learning_rate": 9.651620779923332e-06,
+      "loss": 0.1766,
+      "step": 165
+    },
+    {
+      "epoch": 1.465783664459161,
+      "grad_norm": 5.828193293904613,
+      "learning_rate": 9.646345553930187e-06,
+      "loss": 0.175,
+      "step": 166
+    },
+    {
+      "epoch": 1.4746136865342163,
+      "grad_norm": 6.2210187622807265,
+      "learning_rate": 9.64103215198773e-06,
+      "loss": 0.1891,
+      "step": 167
+    },
+    {
+      "epoch": 1.4834437086092715,
+      "grad_norm": 9.016484689070838,
+      "learning_rate": 9.63568061775267e-06,
+      "loss": 0.189,
+      "step": 168
+    },
+    {
+      "epoch": 1.4922737306843268,
+      "grad_norm": 3.531196386481693,
+      "learning_rate": 9.630290995195028e-06,
+      "loss": 0.1937,
+      "step": 169
+    },
+    {
+      "epoch": 1.501103752759382,
+      "grad_norm": 4.641638553308808,
+      "learning_rate": 9.624863328597767e-06,
+      "loss": 0.1763,
+      "step": 170
+    },
+    {
+      "epoch": 1.5099337748344372,
+      "grad_norm": 4.68576860426708,
+      "learning_rate": 9.619397662556434e-06,
+      "loss": 0.1859,
+      "step": 171
+    },
+    {
+      "epoch": 1.5187637969094923,
+      "grad_norm": 5.383190718687304,
+      "learning_rate": 9.613894041978795e-06,
+      "loss": 0.1931,
+      "step": 172
+    },
+    {
+      "epoch": 1.5275938189845475,
+      "grad_norm": 4.33787132489928,
+      "learning_rate": 9.60835251208446e-06,
+      "loss": 0.181,
+      "step": 173
+    },
+    {
+      "epoch": 1.5364238410596025,
+      "grad_norm": 5.0471966768108585,
+      "learning_rate": 9.602773118404518e-06,
+      "loss": 0.1828,
+      "step": 174
+    },
+    {
+      "epoch": 1.5452538631346577,
+      "grad_norm": 4.52992338383335,
+      "learning_rate": 9.597155906781154e-06,
+      "loss": 0.1805,
+      "step": 175
+    },
+    {
+      "epoch": 1.554083885209713,
+      "grad_norm": 4.631621416482625,
+      "learning_rate": 9.591500923367287e-06,
+      "loss": 0.1764,
+      "step": 176
+    },
+    {
+      "epoch": 1.5629139072847682,
+      "grad_norm": 4.130494606770307,
+      "learning_rate": 9.585808214626173e-06,
+      "loss": 0.1811,
+      "step": 177
+    },
+    {
+      "epoch": 1.5717439293598234,
+      "grad_norm": 9.27347846169718,
+      "learning_rate": 9.580077827331038e-06,
+      "loss": 0.1794,
+      "step": 178
+    },
+    {
+      "epoch": 1.5805739514348787,
+      "grad_norm": 4.156796060089488,
+      "learning_rate": 9.574309808564682e-06,
+      "loss": 0.1791,
+      "step": 179
+    },
+    {
+      "epoch": 1.589403973509934,
+      "grad_norm": 5.7164314447515885,
+      "learning_rate": 9.568504205719106e-06,
+      "loss": 0.1583,
+      "step": 180
+    },
+    {
+      "epoch": 1.598233995584989,
+      "grad_norm": 4.1934597232319035,
+      "learning_rate": 9.562661066495108e-06,
+      "loss": 0.1632,
+      "step": 181
+    },
+    {
+      "epoch": 1.6070640176600441,
+      "grad_norm": 3.7973289921753484,
+      "learning_rate": 9.556780438901899e-06,
+      "loss": 0.1832,
+      "step": 182
+    },
+    {
+      "epoch": 1.6158940397350994,
+      "grad_norm": 6.698619995777142,
+      "learning_rate": 9.550862371256705e-06,
+      "loss": 0.1686,
+      "step": 183
+    },
+    {
+      "epoch": 1.6247240618101544,
+      "grad_norm": 5.1689172402316075,
+      "learning_rate": 9.544906912184383e-06,
+      "loss": 0.1565,
+      "step": 184
+    },
+    {
+      "epoch": 1.6335540838852096,
+      "grad_norm": 3.654331376556239,
+      "learning_rate": 9.538914110616995e-06,
+      "loss": 0.1708,
+      "step": 185
+    },
+    {
+      "epoch": 1.6423841059602649,
+      "grad_norm": 6.304736886487628,
+      "learning_rate": 9.532884015793432e-06,
+      "loss": 0.1825,
+      "step": 186
+    },
+    {
+      "epoch": 1.65121412803532,
+      "grad_norm": 3.859565209636668,
+      "learning_rate": 9.526816677258995e-06,
+      "loss": 0.1735,
+      "step": 187
+    },
+    {
+      "epoch": 1.6600441501103753,
+      "grad_norm": 3.5711915158076932,
+      "learning_rate": 9.520712144864997e-06,
+      "loss": 0.1681,
+      "step": 188
+    },
+    {
+      "epoch": 1.6688741721854305,
+      "grad_norm": 4.435977147295718,
+      "learning_rate": 9.514570468768338e-06,
+      "loss": 0.1789,
+      "step": 189
+    },
+    {
+      "epoch": 1.6777041942604858,
+      "grad_norm": 4.0501977712489765,
+      "learning_rate": 9.508391699431114e-06,
+      "loss": 0.1528,
+      "step": 190
+    },
+    {
+      "epoch": 1.6865342163355408,
+      "grad_norm": 4.071633625203301,
+      "learning_rate": 9.502175887620188e-06,
+      "loss": 0.1729,
+      "step": 191
+    },
+    {
+      "epoch": 1.695364238410596,
+      "grad_norm": 3.6268611641097297,
+      "learning_rate": 9.495923084406773e-06,
+      "loss": 0.1661,
+      "step": 192
+    },
+    {
+      "epoch": 1.7041942604856513,
+      "grad_norm": 16.20474671557156,
+      "learning_rate": 9.48963334116602e-06,
+      "loss": 0.1748,
+      "step": 193
+    },
+    {
+      "epoch": 1.7130242825607063,
+      "grad_norm": 4.423640904498312,
+      "learning_rate": 9.48330670957659e-06,
+      "loss": 0.1836,
+      "step": 194
+    },
+    {
+      "epoch": 1.7218543046357615,
+      "grad_norm": 6.3298241198543135,
+      "learning_rate": 9.476943241620233e-06,
+      "loss": 0.1748,
+      "step": 195
+    },
+    {
+      "epoch": 1.7306843267108167,
+      "grad_norm": 4.032022414185224,
+      "learning_rate": 9.470542989581357e-06,
+      "loss": 0.1801,
+      "step": 196
+    },
+    {
+      "epoch": 1.739514348785872,
+      "grad_norm": 5.580442910747046,
+      "learning_rate": 9.464106006046602e-06,
+      "loss": 0.18,
+      "step": 197
+    },
+    {
+      "epoch": 1.7483443708609272,
+      "grad_norm": 4.9625941262808935,
+      "learning_rate": 9.457632343904404e-06,
+      "loss": 0.1641,
+      "step": 198
+    },
+    {
+      "epoch": 1.7571743929359824,
+      "grad_norm": 4.036510413716825,
+      "learning_rate": 9.451122056344564e-06,
+      "loss": 0.1623,
+      "step": 199
+    },
+    {
+      "epoch": 1.7660044150110377,
+      "grad_norm": 4.351769673457381,
+      "learning_rate": 9.444575196857814e-06,
+      "loss": 0.1667,
+      "step": 200
+    },
+    {
+      "epoch": 1.7748344370860927,
+      "grad_norm": 3.503055068180116,
+      "learning_rate": 9.437991819235366e-06,
+      "loss": 0.1761,
+      "step": 201
+    },
+    {
+      "epoch": 1.783664459161148,
+      "grad_norm": 4.040453399695261,
+      "learning_rate": 9.431371977568483e-06,
+      "loss": 0.1845,
+      "step": 202
+    },
+    {
+      "epoch": 1.7924944812362031,
+      "grad_norm": 4.543045949052899,
+      "learning_rate": 9.424715726248027e-06,
+      "loss": 0.1763,
+      "step": 203
+    },
+    {
+      "epoch": 1.8013245033112582,
+      "grad_norm": 4.994252724100901,
+      "learning_rate": 9.418023119964012e-06,
+      "loss": 0.19,
+      "step": 204
+    },
+    {
+      "epoch": 1.8101545253863134,
+      "grad_norm": 5.806173853040929,
+      "learning_rate": 9.411294213705162e-06,
+      "loss": 0.1947,
+      "step": 205
+    },
+    {
+      "epoch": 1.8189845474613686,
+      "grad_norm": 5.481669048806609,
+      "learning_rate": 9.404529062758447e-06,
+      "loss": 0.16,
+      "step": 206
+    },
+    {
+      "epoch": 1.8278145695364238,
+      "grad_norm": 3.2196597657326893,
+      "learning_rate": 9.397727722708643e-06,
+      "loss": 0.1881,
+      "step": 207
+    },
+    {
+      "epoch": 1.836644591611479,
+      "grad_norm": 4.75501651474976,
+      "learning_rate": 9.390890249437863e-06,
+      "loss": 0.1652,
+      "step": 208
+    },
+    {
+      "epoch": 1.8454746136865343,
+      "grad_norm": 6.469771054964556,
+      "learning_rate": 9.384016699125102e-06,
+      "loss": 0.1548,
+      "step": 209
+    },
+    {
+      "epoch": 1.8543046357615895,
+      "grad_norm": 3.7820316395600924,
+      "learning_rate": 9.377107128245782e-06,
+      "loss": 0.172,
+      "step": 210
+    },
+    {
+      "epoch": 1.8631346578366446,
+      "grad_norm": 4.667257391932444,
+      "learning_rate": 9.370161593571274e-06,
+      "loss": 0.1751,
+      "step": 211
+    },
+    {
+      "epoch": 1.8719646799116998,
+      "grad_norm": 5.767881384215297,
+      "learning_rate": 9.363180152168448e-06,
+      "loss": 0.1753,
+      "step": 212
+    },
+    {
+      "epoch": 1.8807947019867548,
+      "grad_norm": 3.62378559765123,
+      "learning_rate": 9.356162861399188e-06,
+      "loss": 0.1595,
+      "step": 213
+    },
+    {
+      "epoch": 1.88962472406181,
+      "grad_norm": 5.406688558269064,
+      "learning_rate": 9.349109778919938e-06,
+      "loss": 0.1757,
+      "step": 214
+    },
+    {
+      "epoch": 1.8984547461368653,
+      "grad_norm": 4.5151455175909,
+      "learning_rate": 9.342020962681206e-06,
+      "loss": 0.1591,
+      "step": 215
+    },
+    {
+      "epoch": 1.9072847682119205,
+      "grad_norm": 7.874317090048882,
+      "learning_rate": 9.334896470927115e-06,
+      "loss": 0.177,
+      "step": 216
+    },
+    {
+      "epoch": 1.9161147902869757,
+      "grad_norm": 4.916816313808601,
+      "learning_rate": 9.327736362194899e-06,
+      "loss": 0.1708,
+      "step": 217
+    },
+    {
+      "epoch": 1.924944812362031,
+      "grad_norm": 4.778734339774051,
+      "learning_rate": 9.32054069531444e-06,
+      "loss": 0.1749,
+      "step": 218
+    },
+    {
+      "epoch": 1.9337748344370862,
+      "grad_norm": 6.049032923004645,
+      "learning_rate": 9.313309529407773e-06,
+      "loss": 0.1697,
+      "step": 219
+    },
+    {
+      "epoch": 1.9426048565121414,
+      "grad_norm": 4.146956823403823,
+      "learning_rate": 9.306042923888607e-06,
+      "loss": 0.177,
+      "step": 220
+    },
+    {
+      "epoch": 1.9514348785871964,
+      "grad_norm": 6.77533831621212,
+      "learning_rate": 9.298740938461835e-06,
+      "loss": 0.1758,
+      "step": 221
+    },
+    {
+      "epoch": 1.9602649006622517,
+      "grad_norm": 3.206891132737421,
+      "learning_rate": 9.291403633123046e-06,
+      "loss": 0.152,
+      "step": 222
+    },
+    {
+      "epoch": 1.9690949227373067,
+      "grad_norm": 4.306726275574873,
+      "learning_rate": 9.284031068158023e-06,
+      "loss": 0.1691,
+      "step": 223
+    },
+    {
+      "epoch": 1.977924944812362,
+      "grad_norm": 4.0544766400174765,
+      "learning_rate": 9.27662330414226e-06,
+      "loss": 0.1724,
+      "step": 224
+    },
+    {
+      "epoch": 1.9867549668874172,
+      "grad_norm": 7.120274337729496,
+      "learning_rate": 9.269180401940455e-06,
+      "loss": 0.1743,
+      "step": 225
+    },
+    {
+      "epoch": 1.9955849889624724,
+      "grad_norm": 3.4193567491490855,
+      "learning_rate": 9.261702422706014e-06,
+      "loss": 0.1597,
+      "step": 226
+    },
+    {
+      "epoch": 2.0044150110375276,
+      "grad_norm": 4.106288088301316,
+      "learning_rate": 9.254189427880548e-06,
+      "loss": 0.141,
+      "step": 227
+    },
+    {
+      "epoch": 2.013245033112583,
+      "grad_norm": 2.882770025798958,
+      "learning_rate": 9.24664147919337e-06,
+      "loss": 0.1305,
+      "step": 228
+    },
+    {
+      "epoch": 2.022075055187638,
+      "grad_norm": 3.511207421364191,
+      "learning_rate": 9.239058638660983e-06,
+      "loss": 0.123,
+      "step": 229
+    },
+    {
+      "epoch": 2.0309050772626933,
+      "grad_norm": 5.765668062012182,
+      "learning_rate": 9.231440968586572e-06,
+      "loss": 0.1284,
+      "step": 230
+    },
+    {
+      "epoch": 2.0397350993377485,
+      "grad_norm": 6.6705151722336895,
+      "learning_rate": 9.223788531559495e-06,
+      "loss": 0.1166,
+      "step": 231
+    },
+    {
+      "epoch": 2.0485651214128033,
+      "grad_norm": 5.678256758749395,
+      "learning_rate": 9.216101390454771e-06,
+      "loss": 0.1467,
+      "step": 232
+    },
+    {
+      "epoch": 2.0573951434878586,
+      "grad_norm": 4.597940897671489,
+      "learning_rate": 9.208379608432552e-06,
+      "loss": 0.1406,
+      "step": 233
+    },
+    {
+      "epoch": 2.066225165562914,
+      "grad_norm": 5.145951489387972,
+      "learning_rate": 9.200623248937619e-06,
+      "loss": 0.1201,
+      "step": 234
+    },
+    {
+      "epoch": 2.075055187637969,
+      "grad_norm": 6.588636998795935,
+      "learning_rate": 9.192832375698845e-06,
+      "loss": 0.1285,
+      "step": 235
+    },
+    {
+      "epoch": 2.0838852097130243,
+      "grad_norm": 5.135646392206335,
+      "learning_rate": 9.185007052728689e-06,
+      "loss": 0.116,
+      "step": 236
+    },
+    {
+      "epoch": 2.0927152317880795,
+      "grad_norm": 4.220282717321963,
+      "learning_rate": 9.177147344322651e-06,
+      "loss": 0.1326,
+      "step": 237
+    },
+    {
+      "epoch": 2.1015452538631347,
+      "grad_norm": 5.711014478089566,
+      "learning_rate": 9.169253315058764e-06,
+      "loss": 0.1352,
+      "step": 238
+    },
+    {
+      "epoch": 2.11037527593819,
+      "grad_norm": 4.054495215920748,
+      "learning_rate": 9.161325029797044e-06,
+      "loss": 0.1145,
+      "step": 239
+    },
+    {
+      "epoch": 2.119205298013245,
+      "grad_norm": 4.521366476282431,
+      "learning_rate": 9.153362553678967e-06,
+      "loss": 0.1226,
+      "step": 240
+    },
+    {
+      "epoch": 2.1280353200883004,
+      "grad_norm": 4.722112139773477,
+      "learning_rate": 9.145365952126937e-06,
+      "loss": 0.1185,
+      "step": 241
+    },
+    {
+      "epoch": 2.136865342163355,
+      "grad_norm": 7.469499081939174,
+      "learning_rate": 9.13733529084374e-06,
+      "loss": 0.1279,
+      "step": 242
+    },
+    {
+      "epoch": 2.1456953642384105,
+      "grad_norm": 5.488709858524371,
+      "learning_rate": 9.129270635812013e-06,
+      "loss": 0.1371,
+      "step": 243
+    },
+    {
+      "epoch": 2.1545253863134657,
+      "grad_norm": 4.729490430275487,
+      "learning_rate": 9.12117205329369e-06,
+      "loss": 0.1232,
+      "step": 244
+    },
+    {
+      "epoch": 2.163355408388521,
+      "grad_norm": 3.697980948897901,
+      "learning_rate": 9.113039609829472e-06,
+      "loss": 0.1262,
+      "step": 245
+    },
+    {
+      "epoch": 2.172185430463576,
+      "grad_norm": 5.729648089011882,
+      "learning_rate": 9.104873372238269e-06,
+      "loss": 0.1253,
+      "step": 246
+    },
+    {
+      "epoch": 2.1810154525386314,
+      "grad_norm": 4.416650629008491,
+      "learning_rate": 9.096673407616656e-06,
+      "loss": 0.1255,
+      "step": 247
+    },
+    {
+      "epoch": 2.1898454746136866,
+      "grad_norm": 3.6240003911621166,
+      "learning_rate": 9.08843978333832e-06,
+      "loss": 0.1326,
+      "step": 248
+    },
+    {
+      "epoch": 2.198675496688742,
+      "grad_norm": 5.710954987113158,
+      "learning_rate": 9.08017256705351e-06,
+      "loss": 0.1209,
+      "step": 249
+    },
+    {
+      "epoch": 2.207505518763797,
+      "grad_norm": 3.4673951330264674,
+      "learning_rate": 9.071871826688472e-06,
+      "loss": 0.1335,
+      "step": 250
+    },
+    {
+      "epoch": 2.216335540838852,
+      "grad_norm": 6.636636329580233,
+      "learning_rate": 9.063537630444903e-06,
+      "loss": 0.1039,
+      "step": 251
+    },
+    {
+      "epoch": 2.225165562913907,
+      "grad_norm": 4.560755557071873,
+      "learning_rate": 9.055170046799386e-06,
+      "loss": 0.1227,
+      "step": 252
+    },
+    {
+      "epoch": 2.2339955849889623,
+      "grad_norm": 4.623925302817527,
+      "learning_rate": 9.046769144502818e-06,
+      "loss": 0.1261,
+      "step": 253
+    },
+    {
+      "epoch": 2.2428256070640176,
+      "grad_norm": 4.233240951948466,
+      "learning_rate": 9.038334992579863e-06,
+      "loss": 0.1211,
+      "step": 254
+    },
+    {
+      "epoch": 2.251655629139073,
+      "grad_norm": 4.332831951430898,
+      "learning_rate": 9.029867660328369e-06,
+      "loss": 0.1281,
+      "step": 255
+    },
+    {
+      "epoch": 2.260485651214128,
+      "grad_norm": 5.35275306470717,
+      "learning_rate": 9.021367217318808e-06,
+      "loss": 0.1266,
+      "step": 256
+    },
+    {
+      "epoch": 2.2693156732891833,
+      "grad_norm": 5.683503750461502,
+      "learning_rate": 9.012833733393697e-06,
+      "loss": 0.1308,
+      "step": 257
+    },
+    {
+      "epoch": 2.2781456953642385,
+      "grad_norm": 16.76988877462198,
+      "learning_rate": 9.004267278667032e-06,
+      "loss": 0.1281,
+      "step": 258
+    },
+    {
+      "epoch": 2.2869757174392937,
+      "grad_norm": 8.92628471689616,
+      "learning_rate": 8.99566792352371e-06,
+      "loss": 0.1386,
+      "step": 259
+    },
+    {
+      "epoch": 2.295805739514349,
+      "grad_norm": 3.6085083460551646,
+      "learning_rate": 8.987035738618943e-06,
+      "loss": 0.122,
+      "step": 260
+    },
+    {
+      "epoch": 2.304635761589404,
+      "grad_norm": 5.51707997275494,
+      "learning_rate": 8.978370794877691e-06,
+      "loss": 0.1302,
+      "step": 261
+    },
+    {
+      "epoch": 2.313465783664459,
+      "grad_norm": 4.397721834795367,
+      "learning_rate": 8.969673163494063e-06,
+      "loss": 0.1199,
+      "step": 262
+    },
+    {
+      "epoch": 2.322295805739514,
+      "grad_norm": 3.3276046272638395,
+      "learning_rate": 8.960942915930749e-06,
+      "loss": 0.1374,
+      "step": 263
+    },
+    {
+      "epoch": 2.3311258278145695,
+      "grad_norm": 4.481239495813595,
+      "learning_rate": 8.952180123918419e-06,
+      "loss": 0.1155,
+      "step": 264
+    },
+    {
+      "epoch": 2.3399558498896247,
+      "grad_norm": 3.3371182566992528,
+      "learning_rate": 8.94338485945514e-06,
+      "loss": 0.1272,
+      "step": 265
+    },
+    {
+      "epoch": 2.34878587196468,
+      "grad_norm": 9.526212734949459,
+      "learning_rate": 8.934557194805787e-06,
+      "loss": 0.1375,
+      "step": 266
+    },
+    {
+      "epoch": 2.357615894039735,
+      "grad_norm": 3.2449386868810666,
+      "learning_rate": 8.925697202501442e-06,
+      "loss": 0.1288,
+      "step": 267
+    },
+    {
+      "epoch": 2.3664459161147904,
+      "grad_norm": 10.65763509588136,
+      "learning_rate": 8.916804955338807e-06,
+      "loss": 0.1257,
+      "step": 268
+    },
+    {
+      "epoch": 2.3752759381898456,
+      "grad_norm": 4.285910268021899,
+      "learning_rate": 8.907880526379594e-06,
+      "loss": 0.1197,
+      "step": 269
+    },
+    {
+      "epoch": 2.384105960264901,
+      "grad_norm": 4.138365874339128,
+      "learning_rate": 8.898923988949936e-06,
+      "loss": 0.1189,
+      "step": 270
+    },
+    {
+      "epoch": 2.3929359823399556,
+      "grad_norm": 4.922900685358142,
+      "learning_rate": 8.88993541663978e-06,
+      "loss": 0.1211,
+      "step": 271
+    },
+    {
+      "epoch": 2.401766004415011,
+      "grad_norm": 7.231740869831,
+      "learning_rate": 8.880914883302278e-06,
+      "loss": 0.1215,
+      "step": 272
+    },
+    {
+      "epoch": 2.410596026490066,
+      "grad_norm": 5.019824722823812,
+      "learning_rate": 8.871862463053193e-06,
+      "loss": 0.1186,
+      "step": 273
+    },
+    {
+      "epoch": 2.4194260485651213,
+      "grad_norm": 13.177984589352702,
+      "learning_rate": 8.862778230270276e-06,
+      "loss": 0.1295,
+      "step": 274
+    },
+    {
+      "epoch": 2.4282560706401766,
+      "grad_norm": 5.092803874683864,
+      "learning_rate": 8.85366225959266e-06,
+      "loss": 0.1193,
+      "step": 275
+    },
+    {
+      "epoch": 2.437086092715232,
+      "grad_norm": 4.645436040352056,
+      "learning_rate": 8.844514625920246e-06,
+      "loss": 0.1195,
+      "step": 276
+    },
+    {
+      "epoch": 2.445916114790287,
+      "grad_norm": 4.4792099516559345,
+      "learning_rate": 8.835335404413096e-06,
+      "loss": 0.1315,
+      "step": 277
+    },
+    {
+      "epoch": 2.4547461368653423,
+      "grad_norm": 4.376327551414001,
+      "learning_rate": 8.826124670490804e-06,
+      "loss": 0.1265,
+      "step": 278
+    },
+    {
+      "epoch": 2.4635761589403975,
+      "grad_norm": 5.143967207001637,
+      "learning_rate": 8.816882499831877e-06,
+      "loss": 0.1385,
+      "step": 279
+    },
+    {
+      "epoch": 2.4724061810154527,
+      "grad_norm": 4.727297780865657,
+      "learning_rate": 8.807608968373123e-06,
+      "loss": 0.1213,
+      "step": 280
+    },
+    {
+      "epoch": 2.481236203090508,
+      "grad_norm": 4.348943546633772,
+      "learning_rate": 8.798304152309019e-06,
+      "loss": 0.1194,
+      "step": 281
+    },
+    {
+      "epoch": 2.4900662251655628,
+      "grad_norm": 3.5448573901775275,
+      "learning_rate": 8.788968128091084e-06,
+      "loss": 0.1455,
+      "step": 282
+    },
+    {
+      "epoch": 2.498896247240618,
+      "grad_norm": 4.0224898315004065,
+      "learning_rate": 8.779600972427257e-06,
+      "loss": 0.1375,
+      "step": 283
+    },
+    {
+      "epoch": 2.507726269315673,
+      "grad_norm": 6.783149660122743,
+      "learning_rate": 8.770202762281267e-06,
+      "loss": 0.1232,
+      "step": 284
+    },
+    {
+      "epoch": 2.5165562913907285,
+      "grad_norm": 3.097212078028231,
+      "learning_rate": 8.760773574871985e-06,
+      "loss": 0.115,
+      "step": 285
+    },
+    {
+      "epoch": 2.5253863134657837,
+      "grad_norm": 7.372711877322491,
+      "learning_rate": 8.751313487672815e-06,
+      "loss": 0.134,
+      "step": 286
+    },
+    {
+      "epoch": 2.534216335540839,
+      "grad_norm": 3.8347010004386894,
+      "learning_rate": 8.741822578411036e-06,
+      "loss": 0.1376,
+      "step": 287
+    },
+    {
+      "epoch": 2.543046357615894,
+      "grad_norm": 3.2301374236057434,
+      "learning_rate": 8.732300925067177e-06,
+      "loss": 0.1315,
+      "step": 288
+    },
+    {
+      "epoch": 2.5518763796909494,
+      "grad_norm": 6.855372976047249,
+      "learning_rate": 8.722748605874365e-06,
+      "loss": 0.1355,
+      "step": 289
+    },
+    {
+      "epoch": 2.560706401766004,
+      "grad_norm": 6.1942078029864405,
+      "learning_rate": 8.71316569931769e-06,
+      "loss": 0.132,
+      "step": 290
+    },
+    {
+      "epoch": 2.5695364238410594,
+      "grad_norm": 3.7145168130345305,
+      "learning_rate": 8.703552284133565e-06,
+      "loss": 0.1336,
+      "step": 291
+    },
+    {
+      "epoch": 2.5783664459161146,
+      "grad_norm": 5.010946900888269,
+      "learning_rate": 8.69390843930906e-06,
+      "loss": 0.1308,
+      "step": 292
+    },
+    {
+      "epoch": 2.58719646799117,
+      "grad_norm": 4.886631546645197,
+      "learning_rate": 8.684234244081274e-06,
+      "loss": 0.1259,
+      "step": 293
+    },
+    {
+      "epoch": 2.596026490066225,
+      "grad_norm": 2.940069775298397,
+      "learning_rate": 8.674529777936674e-06,
+      "loss": 0.1138,
+      "step": 294
+    },
+    {
+      "epoch": 2.6048565121412803,
+      "grad_norm": 5.251924590305838,
+      "learning_rate": 8.66479512061044e-06,
+      "loss": 0.1417,
+      "step": 295
+    },
+    {
+      "epoch": 2.6136865342163356,
+      "grad_norm": 3.7733208885863867,
+      "learning_rate": 8.655030352085816e-06,
+      "loss": 0.1374,
+      "step": 296
+    },
+    {
+      "epoch": 2.622516556291391,
+      "grad_norm": 3.9170101048822246,
+      "learning_rate": 8.645235552593447e-06,
+      "loss": 0.1253,
+      "step": 297
+    },
+    {
+      "epoch": 2.631346578366446,
+      "grad_norm": 4.150269659617099,
+      "learning_rate": 8.635410802610724e-06,
+      "loss": 0.1276,
+      "step": 298
+    },
+    {
+      "epoch": 2.6401766004415013,
+      "grad_norm": 3.5267586217995732,
+      "learning_rate": 8.625556182861126e-06,
+      "loss": 0.132,
+      "step": 299
+    },
+    {
+      "epoch": 2.6490066225165565,
+      "grad_norm": 3.9458309081426877,
+      "learning_rate": 8.615671774313543e-06,
+      "loss": 0.1444,
+      "step": 300
+    },
+    {
+      "epoch": 2.6578366445916117,
+      "grad_norm": 4.0021555334427195,
+      "learning_rate": 8.605757658181626e-06,
+      "loss": 0.1249,
+      "step": 301
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 3.4783433195038156,
+      "learning_rate": 8.595813915923113e-06,
+      "loss": 0.1368,
+      "step": 302
+    },
+    {
+      "epoch": 2.6754966887417218,
+      "grad_norm": 3.9613428910630617,
+      "learning_rate": 8.585840629239158e-06,
+      "loss": 0.1149,
+      "step": 303
+    },
+    {
+      "epoch": 2.684326710816777,
+      "grad_norm": 3.8749406279119936,
+      "learning_rate": 8.575837880073663e-06,
+      "loss": 0.1251,
+      "step": 304
+    },
+    {
+      "epoch": 2.693156732891832,
+      "grad_norm": 7.232470603311793,
+      "learning_rate": 8.565805750612607e-06,
+      "loss": 0.1424,
+      "step": 305
+    },
+    {
+      "epoch": 2.7019867549668874,
+      "grad_norm": 3.6550319288550157,
+      "learning_rate": 8.555744323283364e-06,
+      "loss": 0.1277,
+      "step": 306
+    },
+    {
+      "epoch": 2.7108167770419427,
+      "grad_norm": 6.097422404372455,
+      "learning_rate": 8.545653680754029e-06,
+      "loss": 0.1315,
+      "step": 307
+    },
+    {
+      "epoch": 2.719646799116998,
+      "grad_norm": 3.473890295262665,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 0.1207,
+      "step": 308
+    },
+    {
+      "epoch": 2.7284768211920527,
+      "grad_norm": 3.1376859437887052,
+      "learning_rate": 8.525385081966992e-06,
+      "loss": 0.1188,
+      "step": 309
+    },
+    {
+      "epoch": 2.737306843267108,
+      "grad_norm": 3.729731832819541,
+      "learning_rate": 8.515207292242969e-06,
+      "loss": 0.1281,
+      "step": 310
+    },
+    {
+      "epoch": 2.746136865342163,
+      "grad_norm": 7.591563114540939,
+      "learning_rate": 8.505000620384834e-06,
+      "loss": 0.14,
+      "step": 311
+    },
+    {
+      "epoch": 2.7549668874172184,
+      "grad_norm": 6.158725034789448,
+      "learning_rate": 8.494765150254063e-06,
+      "loss": 0.1284,
+      "step": 312
+    },
+    {
+      "epoch": 2.7637969094922736,
+      "grad_norm": 3.3869145111066437,
+      "learning_rate": 8.484500965948746e-06,
+      "loss": 0.1252,
+      "step": 313
+    },
+    {
+      "epoch": 2.772626931567329,
+      "grad_norm": 9.32989798703237,
+      "learning_rate": 8.474208151802898e-06,
+      "loss": 0.136,
+      "step": 314
+    },
+    {
+      "epoch": 2.781456953642384,
+      "grad_norm": 3.964154637019035,
+      "learning_rate": 8.46388679238577e-06,
+      "loss": 0.1269,
+      "step": 315
+    },
+    {
+      "epoch": 2.7902869757174393,
+      "grad_norm": 3.8668966734142876,
+      "learning_rate": 8.453536972501146e-06,
+      "loss": 0.1319,
+      "step": 316
+    },
+    {
+      "epoch": 2.7991169977924946,
+      "grad_norm": 4.800270912440046,
+      "learning_rate": 8.443158777186652e-06,
+      "loss": 0.125,
+      "step": 317
+    },
+    {
+      "epoch": 2.80794701986755,
+      "grad_norm": 8.17005200004958,
+      "learning_rate": 8.432752291713058e-06,
+      "loss": 0.1407,
+      "step": 318
+    },
+    {
+      "epoch": 2.816777041942605,
+      "grad_norm": 4.195643135069415,
+      "learning_rate": 8.422317601583576e-06,
+      "loss": 0.1259,
+      "step": 319
+    },
+    {
+      "epoch": 2.8256070640176603,
+      "grad_norm": 5.061888356796782,
+      "learning_rate": 8.411854792533154e-06,
+      "loss": 0.151,
+      "step": 320
+    },
+    {
+      "epoch": 2.8344370860927155,
+      "grad_norm": 4.503620333584586,
+      "learning_rate": 8.401363950527777e-06,
+      "loss": 0.1152,
+      "step": 321
+    },
+    {
+      "epoch": 2.8432671081677703,
+      "grad_norm": 5.13276069257213,
+      "learning_rate": 8.390845161763756e-06,
+      "loss": 0.1334,
+      "step": 322
+    },
+    {
+      "epoch": 2.8520971302428255,
+      "grad_norm": 4.374037184205516,
+      "learning_rate": 8.380298512667023e-06,
+      "loss": 0.1323,
+      "step": 323
+    },
+    {
+      "epoch": 2.8609271523178808,
+      "grad_norm": 3.0639753366300453,
+      "learning_rate": 8.369724089892423e-06,
+      "loss": 0.1252,
+      "step": 324
+    },
+    {
+      "epoch": 2.869757174392936,
+      "grad_norm": 4.19388003319346,
+      "learning_rate": 8.359121980322992e-06,
+      "loss": 0.1193,
+      "step": 325
+    },
+    {
+      "epoch": 2.878587196467991,
+      "grad_norm": 15.214151684387948,
+      "learning_rate": 8.34849227106926e-06,
+      "loss": 0.1336,
+      "step": 326
+    },
+    {
+      "epoch": 2.8874172185430464,
+      "grad_norm": 3.7275946590748967,
+      "learning_rate": 8.337835049468517e-06,
+      "loss": 0.1251,
+      "step": 327
+    },
+    {
+      "epoch": 2.8962472406181017,
+      "grad_norm": 4.290520924678558,
+      "learning_rate": 8.327150403084105e-06,
+      "loss": 0.1133,
+      "step": 328
+    },
+    {
+      "epoch": 2.9050772626931565,
+      "grad_norm": 8.977560548421655,
+      "learning_rate": 8.3164384197047e-06,
+      "loss": 0.1228,
+      "step": 329
+    },
+    {
+      "epoch": 2.9139072847682117,
+      "grad_norm": 3.655985008304991,
+      "learning_rate": 8.305699187343586e-06,
+      "loss": 0.1252,
+      "step": 330
+    },
+    {
+      "epoch": 2.922737306843267,
+      "grad_norm": 3.1746423345857178,
+      "learning_rate": 8.294932794237936e-06,
+      "loss": 0.1269,
+      "step": 331
+    },
+    {
+      "epoch": 2.931567328918322,
+      "grad_norm": 5.169537592761684,
+      "learning_rate": 8.284139328848083e-06,
+      "loss": 0.1287,
+      "step": 332
+    },
+    {
+      "epoch": 2.9403973509933774,
+      "grad_norm": 6.020170962551394,
+      "learning_rate": 8.273318879856794e-06,
+      "loss": 0.125,
+      "step": 333
+    },
+    {
+      "epoch": 2.9492273730684326,
+      "grad_norm": 3.9184909634093397,
+      "learning_rate": 8.262471536168547e-06,
+      "loss": 0.1386,
+      "step": 334
+    },
+    {
+      "epoch": 2.958057395143488,
+      "grad_norm": 3.6631131424597103,
+      "learning_rate": 8.251597386908791e-06,
+      "loss": 0.127,
+      "step": 335
+    },
+    {
+      "epoch": 2.966887417218543,
+      "grad_norm": 3.70785572766733,
+      "learning_rate": 8.240696521423221e-06,
+      "loss": 0.1508,
+      "step": 336
+    },
+    {
+      "epoch": 2.9757174392935983,
+      "grad_norm": 3.1503977758349584,
+      "learning_rate": 8.229769029277044e-06,
+      "loss": 0.1286,
+      "step": 337
+    },
+    {
+      "epoch": 2.9845474613686536,
+      "grad_norm": 6.870504017240963,
+      "learning_rate": 8.218815000254233e-06,
+      "loss": 0.1427,
+      "step": 338
+    },
+    {
+      "epoch": 2.993377483443709,
+      "grad_norm": 2.768532609030032,
+      "learning_rate": 8.207834524356804e-06,
+      "loss": 0.1138,
+      "step": 339
+    },
+    {
+      "epoch": 3.0022075055187636,
+      "grad_norm": 2.8207046730318344,
+      "learning_rate": 8.196827691804066e-06,
+      "loss": 0.1205,
+      "step": 340
+    },
+    {
+      "epoch": 3.011037527593819,
+      "grad_norm": 4.402177665136975,
+      "learning_rate": 8.185794593031889e-06,
+      "loss": 0.0963,
+      "step": 341
+    },
+    {
+      "epoch": 3.019867549668874,
+      "grad_norm": 2.874087994226284,
+      "learning_rate": 8.174735318691946e-06,
+      "loss": 0.0985,
+      "step": 342
+    },
+    {
+      "epoch": 3.0286975717439293,
+      "grad_norm": 2.773091790681681,
+      "learning_rate": 8.163649959650983e-06,
+      "loss": 0.095,
+      "step": 343
+    },
+    {
+      "epoch": 3.0375275938189845,
+      "grad_norm": 3.1852082070922703,
+      "learning_rate": 8.152538606990065e-06,
+      "loss": 0.0919,
+      "step": 344
+    },
+    {
+      "epoch": 3.0463576158940397,
+      "grad_norm": 4.97982244086657,
+      "learning_rate": 8.141401352003834e-06,
+      "loss": 0.0997,
+      "step": 345
+    },
+    {
+      "epoch": 3.055187637969095,
+      "grad_norm": 8.541612200165412,
+      "learning_rate": 8.130238286199747e-06,
+      "loss": 0.103,
+      "step": 346
+    },
+    {
+      "epoch": 3.06401766004415,
+      "grad_norm": 4.486284012587778,
+      "learning_rate": 8.119049501297336e-06,
+      "loss": 0.1161,
+      "step": 347
+    },
+    {
+      "epoch": 3.0728476821192054,
+      "grad_norm": 3.3537490721423926,
+      "learning_rate": 8.107835089227446e-06,
+      "loss": 0.0997,
+      "step": 348
+    },
+    {
+      "epoch": 3.0816777041942607,
+      "grad_norm": 5.17366168346374,
+      "learning_rate": 8.096595142131491e-06,
+      "loss": 0.0941,
+      "step": 349
+    },
+    {
+      "epoch": 3.0905077262693155,
+      "grad_norm": 4.052922245836476,
+      "learning_rate": 8.085329752360683e-06,
+      "loss": 0.1003,
+      "step": 350
+    },
+    {
+      "epoch": 3.0993377483443707,
+      "grad_norm": 10.466691824028937,
+      "learning_rate": 8.074039012475277e-06,
+      "loss": 0.0972,
+      "step": 351
+    },
+    {
+      "epoch": 3.108167770419426,
+      "grad_norm": 6.899695990615844,
+      "learning_rate": 8.062723015243821e-06,
+      "loss": 0.1018,
+      "step": 352
+    },
+    {
+      "epoch": 3.116997792494481,
+      "grad_norm": 6.935623054251069,
+      "learning_rate": 8.051381853642385e-06,
+      "loss": 0.0959,
+      "step": 353
+    },
+    {
+      "epoch": 3.1258278145695364,
+      "grad_norm": 5.380240033624375,
+      "learning_rate": 8.04001562085379e-06,
+      "loss": 0.1009,
+      "step": 354
+    },
+    {
+      "epoch": 3.1346578366445916,
+      "grad_norm": 5.204167174531336,
+      "learning_rate": 8.028624410266856e-06,
+      "loss": 0.1044,
+      "step": 355
+    },
+    {
+      "epoch": 3.143487858719647,
+      "grad_norm": 5.267521866348581,
+      "learning_rate": 8.017208315475633e-06,
+      "loss": 0.0925,
+      "step": 356
+    },
+    {
+      "epoch": 3.152317880794702,
+      "grad_norm": 6.366799712282832,
+      "learning_rate": 8.005767430278619e-06,
+      "loss": 0.1012,
+      "step": 357
+    },
+    {
+      "epoch": 3.1611479028697573,
+      "grad_norm": 3.9450113952746904,
+      "learning_rate": 7.994301848678006e-06,
+      "loss": 0.0997,
+      "step": 358
+    },
+    {
+      "epoch": 3.1699779249448126,
+      "grad_norm": 3.1063823479949377,
+      "learning_rate": 7.982811664878897e-06,
+      "loss": 0.1008,
+      "step": 359
+    },
+    {
+      "epoch": 3.1788079470198674,
+      "grad_norm": 4.66629360897943,
+      "learning_rate": 7.971296973288534e-06,
+      "loss": 0.0905,
+      "step": 360
+    },
+    {
+      "epoch": 3.1876379690949226,
+      "grad_norm": 4.158029256356261,
+      "learning_rate": 7.959757868515526e-06,
+      "loss": 0.1065,
+      "step": 361
+    },
+    {
+      "epoch": 3.196467991169978,
+      "grad_norm": 4.761932913825092,
+      "learning_rate": 7.948194445369065e-06,
+      "loss": 0.0925,
+      "step": 362
+    },
+    {
+      "epoch": 3.205298013245033,
+      "grad_norm": 3.8759763859212755,
+      "learning_rate": 7.936606798858154e-06,
+      "loss": 0.0925,
+      "step": 363
+    },
+    {
+      "epoch": 3.2141280353200883,
+      "grad_norm": 4.532153769879546,
+      "learning_rate": 7.924995024190825e-06,
+      "loss": 0.0927,
+      "step": 364
+    },
+    {
+      "epoch": 3.2229580573951435,
+      "grad_norm": 2.8255168649376614,
+      "learning_rate": 7.91335921677335e-06,
+      "loss": 0.0979,
+      "step": 365
+    },
+    {
+      "epoch": 3.2317880794701987,
+      "grad_norm": 6.973702691843859,
+      "learning_rate": 7.901699472209467e-06,
+      "loss": 0.0883,
+      "step": 366
+    },
+    {
+      "epoch": 3.240618101545254,
+      "grad_norm": 4.809908522267683,
+      "learning_rate": 7.890015886299587e-06,
+      "loss": 0.1006,
+      "step": 367
+    },
+    {
+      "epoch": 3.249448123620309,
+      "grad_norm": 4.7958905830128895,
+      "learning_rate": 7.878308555040012e-06,
+      "loss": 0.0928,
+      "step": 368
+    },
+    {
+      "epoch": 3.258278145695364,
+      "grad_norm": 2.86896545751829,
+      "learning_rate": 7.86657757462214e-06,
+      "loss": 0.0988,
+      "step": 369
+    },
+    {
+      "epoch": 3.2671081677704192,
+      "grad_norm": 3.6073305272465515,
+      "learning_rate": 7.85482304143168e-06,
+      "loss": 0.0879,
+      "step": 370
+    },
+    {
+      "epoch": 3.2759381898454745,
+      "grad_norm": 6.14779480895514,
+      "learning_rate": 7.843045052047863e-06,
+      "loss": 0.0968,
+      "step": 371
+    },
+    {
+      "epoch": 3.2847682119205297,
+      "grad_norm": 3.576211782488104,
+      "learning_rate": 7.831243703242636e-06,
+      "loss": 0.1012,
+      "step": 372
+    },
+    {
+      "epoch": 3.293598233995585,
+      "grad_norm": 4.598562775247985,
+      "learning_rate": 7.819419091979884e-06,
+      "loss": 0.0894,
+      "step": 373
+    },
+    {
+      "epoch": 3.30242825607064,
+      "grad_norm": 10.407962558531288,
+      "learning_rate": 7.807571315414616e-06,
+      "loss": 0.0998,
+      "step": 374
+    },
+    {
+      "epoch": 3.3112582781456954,
+      "grad_norm": 4.0838619374407354,
+      "learning_rate": 7.795700470892177e-06,
+      "loss": 0.1108,
+      "step": 375
+    },
+    {
+      "epoch": 3.3200883002207506,
+      "grad_norm": 3.390233963930005,
+      "learning_rate": 7.783806655947454e-06,
+      "loss": 0.0996,
+      "step": 376
+    },
+    {
+      "epoch": 3.328918322295806,
+      "grad_norm": 3.4343376189192925,
+      "learning_rate": 7.771889968304054e-06,
+      "loss": 0.1022,
+      "step": 377
+    },
+    {
+      "epoch": 3.337748344370861,
+      "grad_norm": 3.9967675391286464,
+      "learning_rate": 7.759950505873523e-06,
+      "loss": 0.0966,
+      "step": 378
+    },
+    {
+      "epoch": 3.3465783664459163,
+      "grad_norm": 3.931184294041522,
+      "learning_rate": 7.747988366754529e-06,
+      "loss": 0.1019,
+      "step": 379
+    },
+    {
+      "epoch": 3.355408388520971,
+      "grad_norm": 3.6434067114246695,
+      "learning_rate": 7.736003649232058e-06,
+      "loss": 0.101,
+      "step": 380
+    },
+    {
+      "epoch": 3.3642384105960264,
+      "grad_norm": 3.0415486494416197,
+      "learning_rate": 7.723996451776615e-06,
+      "loss": 0.095,
+      "step": 381
+    },
+    {
+      "epoch": 3.3730684326710816,
+      "grad_norm": 3.0405884839128516,
+      "learning_rate": 7.711966873043396e-06,
+      "loss": 0.0989,
+      "step": 382
+    },
+    {
+      "epoch": 3.381898454746137,
+      "grad_norm": 3.8412175131705473,
+      "learning_rate": 7.699915011871502e-06,
+      "loss": 0.1193,
+      "step": 383
+    },
+    {
+      "epoch": 3.390728476821192,
+      "grad_norm": 2.993701017330564,
+      "learning_rate": 7.687840967283102e-06,
+      "loss": 0.0947,
+      "step": 384
+    },
+    {
+      "epoch": 3.3995584988962473,
+      "grad_norm": 5.643942446934809,
+      "learning_rate": 7.675744838482641e-06,
+      "loss": 0.1089,
+      "step": 385
+    },
+    {
+      "epoch": 3.4083885209713025,
+      "grad_norm": 3.798921749626721,
+      "learning_rate": 7.66362672485601e-06,
+      "loss": 0.1111,
+      "step": 386
+    },
+    {
+      "epoch": 3.4172185430463577,
+      "grad_norm": 4.07519615976915,
+      "learning_rate": 7.651486725969736e-06,
+      "loss": 0.0995,
+      "step": 387
+    },
+    {
+      "epoch": 3.426048565121413,
+      "grad_norm": 3.109263108741034,
+      "learning_rate": 7.639324941570165e-06,
+      "loss": 0.0957,
+      "step": 388
+    },
+    {
+      "epoch": 3.4348785871964678,
+      "grad_norm": 3.966120155328651,
+      "learning_rate": 7.627141471582635e-06,
+      "loss": 0.0898,
+      "step": 389
+    },
+    {
+      "epoch": 3.443708609271523,
+      "grad_norm": 5.637058048676252,
+      "learning_rate": 7.614936416110668e-06,
+      "loss": 0.093,
+      "step": 390
+    },
+    {
+      "epoch": 3.4525386313465782,
+      "grad_norm": 2.741187454133543,
+      "learning_rate": 7.6027098754351306e-06,
+      "loss": 0.1018,
+      "step": 391
+    },
+    {
+      "epoch": 3.4613686534216335,
+      "grad_norm": 3.8393555875989978,
+      "learning_rate": 7.590461950013424e-06,
+      "loss": 0.0919,
+      "step": 392
+    },
+    {
+      "epoch": 3.4701986754966887,
+      "grad_norm": 2.919287435009137,
+      "learning_rate": 7.578192740478656e-06,
+      "loss": 0.1025,
+      "step": 393
+    },
+    {
+      "epoch": 3.479028697571744,
+      "grad_norm": 3.3130879815381604,
+      "learning_rate": 7.565902347638806e-06,
+      "loss": 0.0869,
+      "step": 394
+    },
+    {
+      "epoch": 3.487858719646799,
+      "grad_norm": 3.0285191115691843,
+      "learning_rate": 7.553590872475909e-06,
+      "loss": 0.1024,
+      "step": 395
+    },
+    {
+      "epoch": 3.4966887417218544,
+      "grad_norm": 2.6155203207930295,
+      "learning_rate": 7.541258416145212e-06,
+      "loss": 0.0897,
+      "step": 396
+    },
+    {
+      "epoch": 3.5055187637969096,
+      "grad_norm": 5.716915177287957,
+      "learning_rate": 7.528905079974358e-06,
+      "loss": 0.0958,
+      "step": 397
+    },
+    {
+      "epoch": 3.514348785871965,
+      "grad_norm": 3.3529943708033945,
+      "learning_rate": 7.5165309654625405e-06,
+      "loss": 0.0917,
+      "step": 398
+    },
+    {
+      "epoch": 3.52317880794702,
+      "grad_norm": 4.407394855256224,
+      "learning_rate": 7.504136174279679e-06,
+      "loss": 0.104,
+      "step": 399
+    },
+    {
+      "epoch": 3.5320088300220753,
+      "grad_norm": 3.141064942786671,
+      "learning_rate": 7.491720808265576e-06,
+      "loss": 0.0963,
+      "step": 400
+    },
+    {
+      "epoch": 3.54083885209713,
+      "grad_norm": 3.2198136199963856,
+      "learning_rate": 7.4792849694290846e-06,
+      "loss": 0.1089,
+      "step": 401
+    },
+    {
+      "epoch": 3.5496688741721854,
+      "grad_norm": 3.9253303059749833,
+      "learning_rate": 7.466828759947271e-06,
+      "loss": 0.0929,
+      "step": 402
+    },
+    {
+      "epoch": 3.5584988962472406,
+      "grad_norm": 4.280334098281072,
+      "learning_rate": 7.454352282164572e-06,
+      "loss": 0.0973,
+      "step": 403
+    },
+    {
+      "epoch": 3.567328918322296,
+      "grad_norm": 3.969696029648232,
+      "learning_rate": 7.441855638591958e-06,
+      "loss": 0.1006,
+      "step": 404
+    },
+    {
+      "epoch": 3.576158940397351,
+      "grad_norm": 19.883828843473825,
+      "learning_rate": 7.429338931906085e-06,
+      "loss": 0.1017,
+      "step": 405
+    },
+    {
+      "epoch": 3.5849889624724063,
+      "grad_norm": 4.198510642360416,
+      "learning_rate": 7.416802264948455e-06,
+      "loss": 0.1006,
+      "step": 406
+    },
+    {
+      "epoch": 3.5938189845474615,
+      "grad_norm": 2.822853105650156,
+      "learning_rate": 7.404245740724573e-06,
+      "loss": 0.1079,
+      "step": 407
+    },
+    {
+      "epoch": 3.6026490066225163,
+      "grad_norm": 3.353228799417008,
+      "learning_rate": 7.391669462403096e-06,
+      "loss": 0.0903,
+      "step": 408
+    },
+    {
+      "epoch": 3.6114790286975715,
+      "grad_norm": 5.576752865540119,
+      "learning_rate": 7.379073533314988e-06,
+      "loss": 0.1009,
+      "step": 409
+    },
+    {
+      "epoch": 3.6203090507726268,
+      "grad_norm": 4.017600918158823,
+      "learning_rate": 7.366458056952668e-06,
+      "loss": 0.1022,
+      "step": 410
+    },
+    {
+      "epoch": 3.629139072847682,
+      "grad_norm": 3.0289707352387065,
+      "learning_rate": 7.353823136969167e-06,
+      "loss": 0.1118,
+      "step": 411
+    },
+    {
+      "epoch": 3.6379690949227372,
+      "grad_norm": 3.2030363014750387,
+      "learning_rate": 7.341168877177267e-06,
+      "loss": 0.111,
+      "step": 412
+    },
+    {
+      "epoch": 3.6467991169977925,
+      "grad_norm": 2.8373469235474413,
+      "learning_rate": 7.328495381548655e-06,
+      "loss": 0.0984,
+      "step": 413
+    },
+    {
+      "epoch": 3.6556291390728477,
+      "grad_norm": 4.237589768108583,
+      "learning_rate": 7.315802754213062e-06,
+      "loss": 0.0928,
+      "step": 414
+    },
+    {
+      "epoch": 3.664459161147903,
+      "grad_norm": 3.293617760716342,
+      "learning_rate": 7.303091099457418e-06,
+      "loss": 0.1035,
+      "step": 415
+    },
+    {
+      "epoch": 3.673289183222958,
+      "grad_norm": 3.695166960242193,
+      "learning_rate": 7.290360521724984e-06,
+      "loss": 0.1051,
+      "step": 416
+    },
+    {
+      "epoch": 3.6821192052980134,
+      "grad_norm": 3.2546124145541944,
+      "learning_rate": 7.277611125614499e-06,
+      "loss": 0.0999,
+      "step": 417
+    },
+    {
+      "epoch": 3.6909492273730686,
+      "grad_norm": 4.98428989250065,
+      "learning_rate": 7.264843015879321e-06,
+      "loss": 0.1002,
+      "step": 418
+    },
+    {
+      "epoch": 3.699779249448124,
+      "grad_norm": 6.408155086873857,
+      "learning_rate": 7.25205629742657e-06,
+      "loss": 0.1004,
+      "step": 419
+    },
+    {
+      "epoch": 3.7086092715231787,
+      "grad_norm": 11.152466179896583,
+      "learning_rate": 7.2392510753162516e-06,
+      "loss": 0.0996,
+      "step": 420
+    },
+    {
+      "epoch": 3.717439293598234,
+      "grad_norm": 3.558825070857552,
+      "learning_rate": 7.226427454760412e-06,
+      "loss": 0.1001,
+      "step": 421
+    },
+    {
+      "epoch": 3.726269315673289,
+      "grad_norm": 6.4326311484879,
+      "learning_rate": 7.213585541122261e-06,
+      "loss": 0.0899,
+      "step": 422
+    },
+    {
+      "epoch": 3.7350993377483444,
+      "grad_norm": 3.107269880751236,
+      "learning_rate": 7.200725439915314e-06,
+      "loss": 0.1054,
+      "step": 423
+    },
+    {
+      "epoch": 3.7439293598233996,
+      "grad_norm": 3.496564737590626,
+      "learning_rate": 7.187847256802518e-06,
+      "loss": 0.0861,
+      "step": 424
+    },
+    {
+      "epoch": 3.752759381898455,
+      "grad_norm": 3.888379002270393,
+      "learning_rate": 7.174951097595389e-06,
+      "loss": 0.1022,
+      "step": 425
+    },
+    {
+      "epoch": 3.76158940397351,
+      "grad_norm": 5.99226737328626,
+      "learning_rate": 7.162037068253141e-06,
+      "loss": 0.1046,
+      "step": 426
+    },
+    {
+      "epoch": 3.7704194260485653,
+      "grad_norm": 3.7128788164666875,
+      "learning_rate": 7.149105274881815e-06,
+      "loss": 0.103,
+      "step": 427
+    },
+    {
+      "epoch": 3.77924944812362,
+      "grad_norm": 5.407802305998047,
+      "learning_rate": 7.136155823733405e-06,
+      "loss": 0.0967,
+      "step": 428
+    },
+    {
+      "epoch": 3.7880794701986753,
+      "grad_norm": 3.817443194172372,
+      "learning_rate": 7.12318882120499e-06,
+      "loss": 0.108,
+      "step": 429
+    },
+    {
+      "epoch": 3.7969094922737305,
+      "grad_norm": 3.5868651536922886,
+      "learning_rate": 7.110204373837857e-06,
+      "loss": 0.1057,
+      "step": 430
+    },
+    {
+      "epoch": 3.8057395143487858,
+      "grad_norm": 3.2056096939347043,
+      "learning_rate": 7.097202588316625e-06,
+      "loss": 0.1156,
+      "step": 431
+    },
+    {
+      "epoch": 3.814569536423841,
+      "grad_norm": 2.7349141183729357,
+      "learning_rate": 7.084183571468368e-06,
+      "loss": 0.0983,
+      "step": 432
+    },
+    {
+      "epoch": 3.8233995584988962,
+      "grad_norm": 4.246208306345596,
+      "learning_rate": 7.071147430261738e-06,
+      "loss": 0.0978,
+      "step": 433
+    },
+    {
+      "epoch": 3.8322295805739515,
+      "grad_norm": 5.8984888612881585,
+      "learning_rate": 7.058094271806091e-06,
+      "loss": 0.1004,
+      "step": 434
+    },
+    {
+      "epoch": 3.8410596026490067,
+      "grad_norm": 3.3105818478273696,
+      "learning_rate": 7.045024203350598e-06,
+      "loss": 0.0973,
+      "step": 435
+    },
+    {
+      "epoch": 3.849889624724062,
+      "grad_norm": 4.809955878560633,
+      "learning_rate": 7.031937332283367e-06,
+      "loss": 0.0982,
+      "step": 436
+    },
+    {
+      "epoch": 3.858719646799117,
+      "grad_norm": 3.4599118483793365,
+      "learning_rate": 7.018833766130571e-06,
+      "loss": 0.1121,
+      "step": 437
+    },
+    {
+      "epoch": 3.8675496688741724,
+      "grad_norm": 3.962277461297394,
+      "learning_rate": 7.0057136125555456e-06,
+      "loss": 0.0954,
+      "step": 438
+    },
+    {
+      "epoch": 3.8763796909492276,
+      "grad_norm": 2.521073351235719,
+      "learning_rate": 6.9925769793579165e-06,
+      "loss": 0.082,
+      "step": 439
+    },
+    {
+      "epoch": 3.8852097130242824,
+      "grad_norm": 3.669379634426741,
+      "learning_rate": 6.979423974472714e-06,
+      "loss": 0.0916,
+      "step": 440
+    },
+    {
+      "epoch": 3.8940397350993377,
+      "grad_norm": 2.810339070177215,
+      "learning_rate": 6.966254705969484e-06,
+      "loss": 0.0971,
+      "step": 441
+    },
+    {
+      "epoch": 3.902869757174393,
+      "grad_norm": 3.3113163980254345,
+      "learning_rate": 6.953069282051397e-06,
+      "loss": 0.0984,
+      "step": 442
+    },
+    {
+      "epoch": 3.911699779249448,
+      "grad_norm": 2.873394765233237,
+      "learning_rate": 6.939867811054365e-06,
+      "loss": 0.0835,
+      "step": 443
+    },
+    {
+      "epoch": 3.9205298013245033,
+      "grad_norm": 6.471682111167088,
+      "learning_rate": 6.9266504014461425e-06,
+      "loss": 0.0947,
+      "step": 444
+    },
+    {
+      "epoch": 3.9293598233995586,
+      "grad_norm": 4.5132940160523125,
+      "learning_rate": 6.913417161825449e-06,
+      "loss": 0.1179,
+      "step": 445
+    },
+    {
+      "epoch": 3.938189845474614,
+      "grad_norm": 4.273369420328601,
+      "learning_rate": 6.900168200921065e-06,
+      "loss": 0.0941,
+      "step": 446
+    },
+    {
+      "epoch": 3.9470198675496686,
+      "grad_norm": 3.136852814569383,
+      "learning_rate": 6.886903627590938e-06,
+      "loss": 0.0903,
+      "step": 447
+    },
+    {
+      "epoch": 3.955849889624724,
+      "grad_norm": 24.428120897543273,
+      "learning_rate": 6.8736235508213024e-06,
+      "loss": 0.0996,
+      "step": 448
+    },
+    {
+      "epoch": 3.964679911699779,
+      "grad_norm": 2.5560553413107474,
+      "learning_rate": 6.860328079725764e-06,
+      "loss": 0.0988,
+      "step": 449
+    },
+    {
+      "epoch": 3.9735099337748343,
+      "grad_norm": 5.943985947216547,
+      "learning_rate": 6.84701732354442e-06,
+      "loss": 0.0984,
+      "step": 450
+    },
+    {
+      "epoch": 3.9823399558498895,
+      "grad_norm": 3.765915925509269,
+      "learning_rate": 6.8336913916429515e-06,
+      "loss": 0.0944,
+      "step": 451
+    },
+    {
+      "epoch": 3.9911699779249448,
+      "grad_norm": 3.0838860947525606,
+      "learning_rate": 6.820350393511732e-06,
+      "loss": 0.0919,
+      "step": 452
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 5.258262477534644,
+      "learning_rate": 6.806994438764922e-06,
+      "loss": 0.1015,
+      "step": 453
+    },
+    {
+      "epoch": 4.008830022075055,
+      "grad_norm": 2.487600464960827,
+      "learning_rate": 6.79362363713957e-06,
+      "loss": 0.0811,
+      "step": 454
+    },
+    {
+      "epoch": 4.0176600441501105,
+      "grad_norm": 2.704929925231827,
+      "learning_rate": 6.780238098494711e-06,
+      "loss": 0.0747,
+      "step": 455
+    },
+    {
+      "epoch": 4.026490066225166,
+      "grad_norm": 4.590794377813223,
+      "learning_rate": 6.766837932810468e-06,
+      "loss": 0.0773,
+      "step": 456
+    },
+    {
+      "epoch": 4.035320088300221,
+      "grad_norm": 3.694952643706577,
+      "learning_rate": 6.7534232501871425e-06,
+      "loss": 0.0839,
+      "step": 457
+    },
+    {
+      "epoch": 4.044150110375276,
+      "grad_norm": 6.619965935098318,
+      "learning_rate": 6.7399941608443096e-06,
+      "loss": 0.0849,
+      "step": 458
+    },
+    {
+      "epoch": 4.052980132450331,
+      "grad_norm": 3.2744763933991017,
+      "learning_rate": 6.72655077511992e-06,
+      "loss": 0.0814,
+      "step": 459
+    },
+    {
+      "epoch": 4.061810154525387,
+      "grad_norm": 2.2788263144306162,
+      "learning_rate": 6.713093203469384e-06,
+      "loss": 0.073,
+      "step": 460
+    },
+    {
+      "epoch": 4.070640176600442,
+      "grad_norm": 2.7293401871470646,
+      "learning_rate": 6.6996215564646705e-06,
+      "loss": 0.0833,
+      "step": 461
+    },
+    {
+      "epoch": 4.079470198675497,
+      "grad_norm": 2.7878106813417296,
+      "learning_rate": 6.686135944793395e-06,
+      "loss": 0.0764,
+      "step": 462
+    },
+    {
+      "epoch": 4.088300220750552,
+      "grad_norm": 3.4508666937690347,
+      "learning_rate": 6.672636479257912e-06,
+      "loss": 0.0685,
+      "step": 463
+    },
+    {
+      "epoch": 4.097130242825607,
+      "grad_norm": 3.2939345109347338,
+      "learning_rate": 6.659123270774406e-06,
+      "loss": 0.0849,
+      "step": 464
+    },
+    {
+      "epoch": 4.105960264900662,
+      "grad_norm": 3.760887373221355,
+      "learning_rate": 6.645596430371976e-06,
+      "loss": 0.0808,
+      "step": 465
+    },
+    {
+      "epoch": 4.114790286975717,
+      "grad_norm": 3.2597140513855005,
+      "learning_rate": 6.632056069191723e-06,
+      "loss": 0.0797,
+      "step": 466
+    },
+    {
+      "epoch": 4.123620309050772,
+      "grad_norm": 6.450410996460324,
+      "learning_rate": 6.618502298485844e-06,
+      "loss": 0.0861,
+      "step": 467
+    },
+    {
+      "epoch": 4.132450331125828,
+      "grad_norm": 3.3573492020873656,
+      "learning_rate": 6.604935229616711e-06,
+      "loss": 0.0782,
+      "step": 468
+    },
+    {
+      "epoch": 4.141280353200883,
+      "grad_norm": 6.126384157468954,
+      "learning_rate": 6.5913549740559606e-06,
+      "loss": 0.0821,
+      "step": 469
+    },
+    {
+      "epoch": 4.150110375275938,
+      "grad_norm": 3.438150104347235,
+      "learning_rate": 6.57776164338357e-06,
+      "loss": 0.082,
+      "step": 470
+    },
+    {
+      "epoch": 4.158940397350993,
+      "grad_norm": 2.9135640923053896,
+      "learning_rate": 6.564155349286952e-06,
+      "loss": 0.0776,
+      "step": 471
+    },
+    {
+      "epoch": 4.1677704194260485,
+      "grad_norm": 2.9999523172662523,
+      "learning_rate": 6.550536203560028e-06,
+      "loss": 0.0699,
+      "step": 472
+    },
+    {
+      "epoch": 4.176600441501104,
+      "grad_norm": 3.746386815870089,
+      "learning_rate": 6.536904318102314e-06,
+      "loss": 0.0769,
+      "step": 473
+    },
+    {
+      "epoch": 4.185430463576159,
+      "grad_norm": 2.9718529127994286,
+      "learning_rate": 6.523259804918001e-06,
+      "loss": 0.0837,
+      "step": 474
+    },
+    {
+      "epoch": 4.194260485651214,
+      "grad_norm": 3.474007331109214,
+      "learning_rate": 6.509602776115029e-06,
+      "loss": 0.0755,
+      "step": 475
+    },
+    {
+      "epoch": 4.2030905077262695,
+      "grad_norm": 3.8851000736196526,
+      "learning_rate": 6.4959333439041775e-06,
+      "loss": 0.0774,
+      "step": 476
+    },
+    {
+      "epoch": 4.211920529801325,
+      "grad_norm": 3.3949435150028924,
+      "learning_rate": 6.482251620598129e-06,
+      "loss": 0.0694,
+      "step": 477
+    },
+    {
+      "epoch": 4.22075055187638,
+      "grad_norm": 4.26273289010399,
+      "learning_rate": 6.4685577186105595e-06,
+      "loss": 0.0778,
+      "step": 478
+    },
+    {
+      "epoch": 4.229580573951435,
+      "grad_norm": 3.1626694349683224,
+      "learning_rate": 6.45485175045521e-06,
+      "loss": 0.0799,
+      "step": 479
+    },
+    {
+      "epoch": 4.23841059602649,
+      "grad_norm": 3.2387645523490214,
+      "learning_rate": 6.441133828744954e-06,
+      "loss": 0.0817,
+      "step": 480
+    },
+    {
+      "epoch": 4.247240618101546,
+      "grad_norm": 3.6360260747998914,
+      "learning_rate": 6.427404066190889e-06,
+      "loss": 0.0841,
+      "step": 481
+    },
+    {
+      "epoch": 4.256070640176601,
+      "grad_norm": 2.5866840173292402,
+      "learning_rate": 6.413662575601391e-06,
+      "loss": 0.079,
+      "step": 482
+    },
+    {
+      "epoch": 4.264900662251655,
+      "grad_norm": 2.7677550181684603,
+      "learning_rate": 6.3999094698812055e-06,
+      "loss": 0.0815,
+      "step": 483
+    },
+    {
+      "epoch": 4.27373068432671,
+      "grad_norm": 3.16309495676388,
+      "learning_rate": 6.386144862030508e-06,
+      "loss": 0.0647,
+      "step": 484
+    },
+    {
+      "epoch": 4.282560706401766,
+      "grad_norm": 2.4645179267970736,
+      "learning_rate": 6.3723688651439806e-06,
+      "loss": 0.0846,
+      "step": 485
+    },
+    {
+      "epoch": 4.291390728476821,
+      "grad_norm": 6.38903880108792,
+      "learning_rate": 6.358581592409881e-06,
+      "loss": 0.0731,
+      "step": 486
+    },
+    {
+      "epoch": 4.300220750551876,
+      "grad_norm": 3.0558384340208686,
+      "learning_rate": 6.344783157109114e-06,
+      "loss": 0.0747,
+      "step": 487
+    },
+    {
+      "epoch": 4.309050772626931,
+      "grad_norm": 2.9983291623446764,
+      "learning_rate": 6.3309736726142965e-06,
+      "loss": 0.075,
+      "step": 488
+    },
+    {
+      "epoch": 4.317880794701987,
+      "grad_norm": 3.2690342947361493,
+      "learning_rate": 6.317153252388834e-06,
+      "loss": 0.0867,
+      "step": 489
+    },
+    {
+      "epoch": 4.326710816777042,
+      "grad_norm": 3.610248860150095,
+      "learning_rate": 6.303322009985984e-06,
+      "loss": 0.0724,
+      "step": 490
+    },
+    {
+      "epoch": 4.335540838852097,
+      "grad_norm": 3.8688332246602752,
+      "learning_rate": 6.289480059047915e-06,
+      "loss": 0.0753,
+      "step": 491
+    },
+    {
+      "epoch": 4.344370860927152,
+      "grad_norm": 7.441887079342132,
+      "learning_rate": 6.27562751330479e-06,
+      "loss": 0.0842,
+      "step": 492
+    },
+    {
+      "epoch": 4.3532008830022075,
+      "grad_norm": 2.9839561092138718,
+      "learning_rate": 6.261764486573816e-06,
+      "loss": 0.079,
+      "step": 493
+    },
+    {
+      "epoch": 4.362030905077263,
+      "grad_norm": 4.200459499442376,
+      "learning_rate": 6.247891092758319e-06,
+      "loss": 0.0736,
+      "step": 494
+    },
+    {
+      "epoch": 4.370860927152318,
+      "grad_norm": 3.4070365441270467,
+      "learning_rate": 6.2340074458468014e-06,
+      "loss": 0.0806,
+      "step": 495
+    },
+    {
+      "epoch": 4.379690949227373,
+      "grad_norm": 3.441955046739519,
+      "learning_rate": 6.220113659912012e-06,
+      "loss": 0.0731,
+      "step": 496
+    },
+    {
+      "epoch": 4.3885209713024285,
+      "grad_norm": 3.1561631964440076,
+      "learning_rate": 6.206209849110001e-06,
+      "loss": 0.0758,
+      "step": 497
+    },
+    {
+      "epoch": 4.397350993377484,
+      "grad_norm": 3.624346507958278,
+      "learning_rate": 6.1922961276791925e-06,
+      "loss": 0.0794,
+      "step": 498
+    },
+    {
+      "epoch": 4.406181015452539,
+      "grad_norm": 6.080763290404623,
+      "learning_rate": 6.1783726099394324e-06,
+      "loss": 0.0804,
+      "step": 499
+    },
+    {
+      "epoch": 4.415011037527594,
+      "grad_norm": 3.1460557464112293,
+      "learning_rate": 6.164439410291061e-06,
+      "loss": 0.0786,
+      "step": 500
+    },
+    {
+      "epoch": 4.423841059602649,
+      "grad_norm": 4.32533416371131,
+      "learning_rate": 6.150496643213969e-06,
+      "loss": 0.0813,
+      "step": 501
+    },
+    {
+      "epoch": 4.432671081677704,
+      "grad_norm": 4.384447144192398,
+      "learning_rate": 6.136544423266651e-06,
+      "loss": 0.078,
+      "step": 502
+    },
+    {
+      "epoch": 4.441501103752759,
+      "grad_norm": 3.527942957160671,
+      "learning_rate": 6.122582865085278e-06,
+      "loss": 0.0843,
+      "step": 503
+    },
+    {
+      "epoch": 4.450331125827814,
+      "grad_norm": 3.1467040214710527,
+      "learning_rate": 6.108612083382739e-06,
+      "loss": 0.0689,
+      "step": 504
+    },
+    {
+      "epoch": 4.459161147902869,
+      "grad_norm": 5.934612259061866,
+      "learning_rate": 6.094632192947711e-06,
+      "loss": 0.0806,
+      "step": 505
+    },
+    {
+      "epoch": 4.467991169977925,
+      "grad_norm": 3.195416678057104,
+      "learning_rate": 6.08064330864371e-06,
+      "loss": 0.0841,
+      "step": 506
+    },
+    {
+      "epoch": 4.47682119205298,
+      "grad_norm": 5.946520342587703,
+      "learning_rate": 6.066645545408149e-06,
+      "loss": 0.0832,
+      "step": 507
+    },
+    {
+      "epoch": 4.485651214128035,
+      "grad_norm": 2.3820570403659693,
+      "learning_rate": 6.052639018251394e-06,
+      "loss": 0.0829,
+      "step": 508
+    },
+    {
+      "epoch": 4.49448123620309,
+      "grad_norm": 3.115965495879289,
+      "learning_rate": 6.03862384225582e-06,
+      "loss": 0.0741,
+      "step": 509
+    },
+    {
+      "epoch": 4.503311258278146,
+      "grad_norm": 4.214892939498798,
+      "learning_rate": 6.024600132574855e-06,
+      "loss": 0.0742,
+      "step": 510
+    },
+    {
+      "epoch": 4.512141280353201,
+      "grad_norm": 4.138254792265556,
+      "learning_rate": 6.010568004432055e-06,
+      "loss": 0.0825,
+      "step": 511
+    },
+    {
+      "epoch": 4.520971302428256,
+      "grad_norm": 2.729416010034935,
+      "learning_rate": 5.9965275731201364e-06,
+      "loss": 0.0808,
+      "step": 512
+    },
+    {
+      "epoch": 4.529801324503311,
+      "grad_norm": 13.118641333902108,
+      "learning_rate": 5.982478954000042e-06,
+      "loss": 0.08,
+      "step": 513
+    },
+    {
+      "epoch": 4.5386313465783665,
+      "grad_norm": 2.335880462238545,
+      "learning_rate": 5.968422262499983e-06,
+      "loss": 0.0739,
+      "step": 514
+    },
+    {
+      "epoch": 4.547461368653422,
+      "grad_norm": 3.6375779968448496,
+      "learning_rate": 5.9543576141145035e-06,
+      "loss": 0.0622,
+      "step": 515
+    },
+    {
+      "epoch": 4.556291390728477,
+      "grad_norm": 3.6160696452303984,
+      "learning_rate": 5.940285124403517e-06,
+      "loss": 0.0784,
+      "step": 516
+    },
+    {
+      "epoch": 4.565121412803532,
+      "grad_norm": 3.3711175360832777,
+      "learning_rate": 5.926204908991366e-06,
+      "loss": 0.0735,
+      "step": 517
+    },
+    {
+      "epoch": 4.5739514348785875,
+      "grad_norm": 3.2269124481467104,
+      "learning_rate": 5.912117083565874e-06,
+      "loss": 0.0791,
+      "step": 518
+    },
+    {
+      "epoch": 4.582781456953643,
+      "grad_norm": 5.900071692486368,
+      "learning_rate": 5.898021763877388e-06,
+      "loss": 0.0914,
+      "step": 519
+    },
+    {
+      "epoch": 4.591611479028698,
+      "grad_norm": 4.418513990528456,
+      "learning_rate": 5.883919065737827e-06,
+      "loss": 0.081,
+      "step": 520
+    },
+    {
+      "epoch": 4.600441501103752,
+      "grad_norm": 3.3722000807665866,
+      "learning_rate": 5.869809105019738e-06,
+      "loss": 0.0744,
+      "step": 521
+    },
+    {
+      "epoch": 4.609271523178808,
+      "grad_norm": 7.469587100913114,
+      "learning_rate": 5.85569199765534e-06,
+      "loss": 0.0723,
+      "step": 522
+    },
+    {
+      "epoch": 4.618101545253863,
+      "grad_norm": 2.9175696518893486,
+      "learning_rate": 5.841567859635572e-06,
+      "loss": 0.0728,
+      "step": 523
+    },
+    {
+      "epoch": 4.626931567328918,
+      "grad_norm": 4.026258728317068,
+      "learning_rate": 5.827436807009133e-06,
+      "loss": 0.0765,
+      "step": 524
+    },
+    {
+      "epoch": 4.635761589403973,
+      "grad_norm": 4.023701049292052,
+      "learning_rate": 5.813298955881542e-06,
+      "loss": 0.0795,
+      "step": 525
+    },
+    {
+      "epoch": 4.644591611479028,
+      "grad_norm": 7.492669294902294,
+      "learning_rate": 5.799154422414174e-06,
+      "loss": 0.0832,
+      "step": 526
+    },
+    {
+      "epoch": 4.653421633554084,
+      "grad_norm": 27.135072765679574,
+      "learning_rate": 5.785003322823307e-06,
+      "loss": 0.0641,
+      "step": 527
+    },
+    {
+      "epoch": 4.662251655629139,
+      "grad_norm": 2.1533625604567357,
+      "learning_rate": 5.7708457733791715e-06,
+      "loss": 0.0773,
+      "step": 528
+    },
+    {
+      "epoch": 4.671081677704194,
+      "grad_norm": 2.9620640076509615,
+      "learning_rate": 5.756681890404987e-06,
+      "loss": 0.0853,
+      "step": 529
+    },
+    {
+      "epoch": 4.679911699779249,
+      "grad_norm": 4.678163808609413,
+      "learning_rate": 5.7425117902760195e-06,
+      "loss": 0.0713,
+      "step": 530
+    },
+    {
+      "epoch": 4.688741721854305,
+      "grad_norm": 4.296864154061136,
+      "learning_rate": 5.72833558941861e-06,
+      "loss": 0.0859,
+      "step": 531
+    },
+    {
+      "epoch": 4.69757174392936,
+      "grad_norm": 2.922605079813754,
+      "learning_rate": 5.714153404309228e-06,
+      "loss": 0.078,
+      "step": 532
+    },
+    {
+      "epoch": 4.706401766004415,
+      "grad_norm": 3.9382223946197015,
+      "learning_rate": 5.6999653514735124e-06,
+      "loss": 0.071,
+      "step": 533
+    },
+    {
+      "epoch": 4.71523178807947,
+      "grad_norm": 3.147586504010664,
+      "learning_rate": 5.685771547485312e-06,
+      "loss": 0.0813,
+      "step": 534
+    },
+    {
+      "epoch": 4.7240618101545255,
+      "grad_norm": 2.966590218091343,
+      "learning_rate": 5.671572108965729e-06,
+      "loss": 0.0751,
+      "step": 535
+    },
+    {
+      "epoch": 4.732891832229581,
+      "grad_norm": 3.2622534882340446,
+      "learning_rate": 5.65736715258216e-06,
+      "loss": 0.0753,
+      "step": 536
+    },
+    {
+      "epoch": 4.741721854304636,
+      "grad_norm": 2.3211945267996716,
+      "learning_rate": 5.643156795047343e-06,
+      "loss": 0.0769,
+      "step": 537
+    },
+    {
+      "epoch": 4.750551876379691,
+      "grad_norm": 8.119722637541681,
+      "learning_rate": 5.628941153118388e-06,
+      "loss": 0.0822,
+      "step": 538
+    },
+    {
+      "epoch": 4.7593818984547465,
+      "grad_norm": 8.441365529225875,
+      "learning_rate": 5.6147203435958246e-06,
+      "loss": 0.0829,
+      "step": 539
+    },
+    {
+      "epoch": 4.768211920529802,
+      "grad_norm": 2.451783795869697,
+      "learning_rate": 5.600494483322643e-06,
+      "loss": 0.0763,
+      "step": 540
+    },
+    {
+      "epoch": 4.777041942604857,
+      "grad_norm": 2.991753091402444,
+      "learning_rate": 5.586263689183332e-06,
+      "loss": 0.0741,
+      "step": 541
+    },
+    {
+      "epoch": 4.785871964679911,
+      "grad_norm": 3.539488911553519,
+      "learning_rate": 5.572028078102917e-06,
+      "loss": 0.0682,
+      "step": 542
+    },
+    {
+      "epoch": 4.7947019867549665,
+      "grad_norm": 5.018439865551765,
+      "learning_rate": 5.557787767046001e-06,
+      "loss": 0.084,
+      "step": 543
+    },
+    {
+      "epoch": 4.803532008830022,
+      "grad_norm": 2.8442572450341257,
+      "learning_rate": 5.543542873015806e-06,
+      "loss": 0.0769,
+      "step": 544
+    },
+    {
+      "epoch": 4.812362030905077,
+      "grad_norm": 2.7357924462195466,
+      "learning_rate": 5.529293513053207e-06,
+      "loss": 0.0661,
+      "step": 545
+    },
+    {
+      "epoch": 4.821192052980132,
+      "grad_norm": 4.7717542869998475,
+      "learning_rate": 5.515039804235772e-06,
+      "loss": 0.0797,
+      "step": 546
+    },
+    {
+      "epoch": 4.830022075055187,
+      "grad_norm": 2.361311252969642,
+      "learning_rate": 5.5007818636768055e-06,
+      "loss": 0.0756,
+      "step": 547
+    },
+    {
+      "epoch": 4.838852097130243,
+      "grad_norm": 3.6882765130368793,
+      "learning_rate": 5.486519808524374e-06,
+      "loss": 0.0798,
+      "step": 548
+    },
+    {
+      "epoch": 4.847682119205298,
+      "grad_norm": 3.9327149256112035,
+      "learning_rate": 5.472253755960358e-06,
+      "loss": 0.0808,
+      "step": 549
+    },
+    {
+      "epoch": 4.856512141280353,
+      "grad_norm": 3.5572765179325034,
+      "learning_rate": 5.457983823199475e-06,
+      "loss": 0.0772,
+      "step": 550
+    },
+    {
+      "epoch": 4.865342163355408,
+      "grad_norm": 3.0090033893509402,
+      "learning_rate": 5.443710127488331e-06,
+      "loss": 0.0887,
+      "step": 551
+    },
+    {
+      "epoch": 4.874172185430464,
+      "grad_norm": 4.076041046123783,
+      "learning_rate": 5.429432786104446e-06,
+      "loss": 0.0901,
+      "step": 552
+    },
+    {
+      "epoch": 4.883002207505519,
+      "grad_norm": 5.244647456159231,
+      "learning_rate": 5.415151916355292e-06,
+      "loss": 0.0786,
+      "step": 553
+    },
+    {
+      "epoch": 4.891832229580574,
+      "grad_norm": 2.8707489082851136,
+      "learning_rate": 5.400867635577335e-06,
+      "loss": 0.0639,
+      "step": 554
+    },
+    {
+      "epoch": 4.900662251655629,
+      "grad_norm": 6.188907034334066,
+      "learning_rate": 5.3865800611350634e-06,
+      "loss": 0.0803,
+      "step": 555
+    },
+    {
+      "epoch": 4.9094922737306845,
+      "grad_norm": 3.791444437800692,
+      "learning_rate": 5.372289310420032e-06,
+      "loss": 0.0823,
+      "step": 556
+    },
+    {
+      "epoch": 4.91832229580574,
+      "grad_norm": 3.383269845290552,
+      "learning_rate": 5.35799550084989e-06,
+      "loss": 0.0894,
+      "step": 557
+    },
+    {
+      "epoch": 4.927152317880795,
+      "grad_norm": 3.564403614663905,
+      "learning_rate": 5.343698749867421e-06,
+      "loss": 0.0702,
+      "step": 558
+    },
+    {
+      "epoch": 4.93598233995585,
+      "grad_norm": 3.2097656260996086,
+      "learning_rate": 5.329399174939572e-06,
+      "loss": 0.0725,
+      "step": 559
+    },
+    {
+      "epoch": 4.9448123620309055,
+      "grad_norm": 2.796838877718886,
+      "learning_rate": 5.315096893556497e-06,
+      "loss": 0.0829,
+      "step": 560
+    },
+    {
+      "epoch": 4.95364238410596,
+      "grad_norm": 2.6046376261906548,
+      "learning_rate": 5.300792023230587e-06,
+      "loss": 0.0813,
+      "step": 561
+    },
+    {
+      "epoch": 4.962472406181016,
+      "grad_norm": 2.5662770693138475,
+      "learning_rate": 5.2864846814955e-06,
+      "loss": 0.0676,
+      "step": 562
+    },
+    {
+      "epoch": 4.97130242825607,
+      "grad_norm": 2.5457460369187963,
+      "learning_rate": 5.272174985905207e-06,
+      "loss": 0.0815,
+      "step": 563
+    },
+    {
+      "epoch": 4.9801324503311255,
+      "grad_norm": 3.023794300410492,
+      "learning_rate": 5.257863054033012e-06,
+      "loss": 0.078,
+      "step": 564
+    },
+    {
+      "epoch": 4.988962472406181,
+      "grad_norm": 3.422355333754827,
+      "learning_rate": 5.243549003470599e-06,
+      "loss": 0.0823,
+      "step": 565
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1130,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 565,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 80916831789056.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}