diff --git "a/train_cosine_ckpt600/trainer_state.json" "b/train_cosine_ckpt600/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/train_cosine_ckpt600/trainer_state.json"
@@ -0,0 +1,4834 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.134946176894906,
+  "eval_steps": 500,
+  "global_step": 600,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0035512151814449007,
+      "grad_norm": 3.746668577194214,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 0.4128,
+      "mean_token_accuracy": 0.03184996646086802,
+      "step": 1
+    },
+    {
+      "epoch": 0.007102430362889801,
+      "grad_norm": 4.603147029876709,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.4483,
+      "mean_token_accuracy": 0.03573758571292274,
+      "step": 2
+    },
+    {
+      "epoch": 0.010653645544334702,
+      "grad_norm": 3.065110206604004,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 0.3794,
+      "mean_token_accuracy": 0.031494880182435736,
+      "step": 3
+    },
+    {
+      "epoch": 0.014204860725779603,
+      "grad_norm": 2.540454864501953,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.3385,
+      "mean_token_accuracy": 0.033872520692966646,
+      "step": 4
+    },
+    {
+      "epoch": 0.017756075907224503,
+      "grad_norm": 3.2157845497131348,
+      "learning_rate": 7.1428571428571436e-06,
+      "loss": 0.3214,
+      "mean_token_accuracy": 0.03215573047782527,
+      "step": 5
+    },
+    {
+      "epoch": 0.021307291088669404,
+      "grad_norm": 2.591473340988159,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.3133,
+      "mean_token_accuracy": 0.031042124777741265,
+      "step": 6
+    },
+    {
+      "epoch": 0.024858506270114305,
+      "grad_norm": 5.040935516357422,
+      "learning_rate": 1e-05,
+      "loss": 0.3018,
+      "mean_token_accuracy": 0.029873713359847898,
+      "step": 7
+    },
+    {
+      "epoch": 0.028409721451559206,
+      "grad_norm": 3.1976966857910156,
+      "learning_rate": 9.999980365120307e-06,
+      "loss": 0.3003,
+      "mean_token_accuracy": 0.029250323150336044,
+      "step": 8
+    },
+    {
+      "epoch": 0.0319609366330041,
+      "grad_norm": 2.1849372386932373,
+      "learning_rate": 9.999921460635436e-06,
+      "loss": 0.2999,
+      "mean_token_accuracy": 0.02684653406322468,
+      "step": 9
+    },
+    {
+      "epoch": 0.03551215181444901,
+      "grad_norm": 1.1641992330551147,
+      "learning_rate": 9.999823287008022e-06,
+      "loss": 0.2781,
+      "mean_token_accuracy": 0.030458620822173543,
+      "step": 10
+    },
+    {
+      "epoch": 0.039063366995893904,
+      "grad_norm": 1.8289661407470703,
+      "learning_rate": 9.999685845009114e-06,
+      "loss": 0.282,
+      "mean_token_accuracy": 0.03673582868941594,
+      "step": 11
+    },
+    {
+      "epoch": 0.04261458217733881,
+      "grad_norm": 2.047220468521118,
+      "learning_rate": 9.999509135718176e-06,
+      "loss": 0.285,
+      "mean_token_accuracy": 0.03312260610982776,
+      "step": 12
+    },
+    {
+      "epoch": 0.046165797358783706,
+      "grad_norm": 1.0920840501785278,
+      "learning_rate": 9.999293160523074e-06,
+      "loss": 0.2865,
+      "mean_token_accuracy": 0.03169899153363076,
+      "step": 13
+    },
+    {
+      "epoch": 0.04971701254022861,
+      "grad_norm": 2.3721137046813965,
+      "learning_rate": 9.999037921120068e-06,
+      "loss": 0.2752,
+      "mean_token_accuracy": 0.02908840416057501,
+      "step": 14
+    },
+    {
+      "epoch": 0.05326822772167351,
+      "grad_norm": 1.869300127029419,
+      "learning_rate": 9.998743419513795e-06,
+      "loss": 0.2727,
+      "mean_token_accuracy": 0.029675625406525796,
+      "step": 15
+    },
+    {
+      "epoch": 0.05681944290311841,
+      "grad_norm": 1.656387448310852,
+      "learning_rate": 9.998409658017256e-06,
+      "loss": 0.2803,
+      "mean_token_accuracy": 0.028696611592749832,
+      "step": 16
+    },
+    {
+      "epoch": 0.06037065808456331,
+      "grad_norm": 1.5127822160720825,
+      "learning_rate": 9.998036639251798e-06,
+      "loss": 0.2854,
+      "mean_token_accuracy": 0.03376526683859993,
+      "step": 17
+    },
+    {
+      "epoch": 0.0639218732660082,
+      "grad_norm": 1.3357048034667969,
+      "learning_rate": 9.997624366147094e-06,
+      "loss": 0.2692,
+      "mean_token_accuracy": 0.03486147220610292,
+      "step": 18
+    },
+    {
+      "epoch": 0.06747308844745312,
+      "grad_norm": 1.4938886165618896,
+      "learning_rate": 9.997172841941114e-06,
+      "loss": 0.2681,
+      "mean_token_accuracy": 0.03322860744447098,
+      "step": 19
+    },
+    {
+      "epoch": 0.07102430362889801,
+      "grad_norm": 0.8506525158882141,
+      "learning_rate": 9.99668207018011e-06,
+      "loss": 0.2694,
+      "mean_token_accuracy": 0.032776252537587425,
+      "step": 20
+    },
+    {
+      "epoch": 0.07457551881034291,
+      "grad_norm": 0.9477233290672302,
+      "learning_rate": 9.996152054718579e-06,
+      "loss": 0.263,
+      "mean_token_accuracy": 0.03463000854389975,
+      "step": 21
+    },
+    {
+      "epoch": 0.07812673399178781,
+      "grad_norm": 1.0164031982421875,
+      "learning_rate": 9.995582799719237e-06,
+      "loss": 0.2696,
+      "mean_token_accuracy": 0.03465155349476845,
+      "step": 22
+    },
+    {
+      "epoch": 0.08167794917323272,
+      "grad_norm": 1.2138930559158325,
+      "learning_rate": 9.994974309652984e-06,
+      "loss": 0.2676,
+      "mean_token_accuracy": 0.03345586970317527,
+      "step": 23
+    },
+    {
+      "epoch": 0.08522916435467762,
+      "grad_norm": 1.0064730644226074,
+      "learning_rate": 9.994326589298875e-06,
+      "loss": 0.2538,
+      "mean_token_accuracy": 0.03691909779263369,
+      "step": 24
+    },
+    {
+      "epoch": 0.08878037953612251,
+      "grad_norm": 1.5932868719100952,
+      "learning_rate": 9.993639643744071e-06,
+      "loss": 0.2772,
+      "mean_token_accuracy": 0.03495999022561591,
+      "step": 25
+    },
+    {
+      "epoch": 0.09233159471756741,
+      "grad_norm": 1.5711798667907715,
+      "learning_rate": 9.99291347838381e-06,
+      "loss": 0.2632,
+      "mean_token_accuracy": 0.03424531156633748,
+      "step": 26
+    },
+    {
+      "epoch": 0.09588280989901232,
+      "grad_norm": 1.2252583503723145,
+      "learning_rate": 9.992148098921361e-06,
+      "loss": 0.2445,
+      "mean_token_accuracy": 0.03593750138679752,
+      "step": 27
+    },
+    {
+      "epoch": 0.09943402508045722,
+      "grad_norm": 0.9910848140716553,
+      "learning_rate": 9.99134351136798e-06,
+      "loss": 0.2582,
+      "mean_token_accuracy": 0.038970169473032,
+      "step": 28
+    },
+    {
+      "epoch": 0.10298524026190212,
+      "grad_norm": 0.8660365343093872,
+      "learning_rate": 9.990499722042852e-06,
+      "loss": 0.2622,
+      "mean_token_accuracy": 0.03539915586588904,
+      "step": 29
+    },
+    {
+      "epoch": 0.10653645544334701,
+      "grad_norm": 1.146954894065857,
+      "learning_rate": 9.989616737573064e-06,
+      "loss": 0.266,
+      "mean_token_accuracy": 0.03540461422744556,
+      "step": 30
+    },
+    {
+      "epoch": 0.11008767062479193,
+      "grad_norm": 1.0442326068878174,
+      "learning_rate": 9.98869456489353e-06,
+      "loss": 0.263,
+      "mean_token_accuracy": 0.03115065989732102,
+      "step": 31
+    },
+    {
+      "epoch": 0.11363888580623682,
+      "grad_norm": 1.205544114112854,
+      "learning_rate": 9.987733211246952e-06,
+      "loss": 0.2488,
+      "mean_token_accuracy": 0.0319663421723817,
+      "step": 32
+    },
+    {
+      "epoch": 0.11719010098768172,
+      "grad_norm": 1.2089142799377441,
+      "learning_rate": 9.986732684183753e-06,
+      "loss": 0.2522,
+      "mean_token_accuracy": 0.03767831435834523,
+      "step": 33
+    },
+    {
+      "epoch": 0.12074131616912662,
+      "grad_norm": 1.5163798332214355,
+      "learning_rate": 9.985692991562026e-06,
+      "loss": 0.2507,
+      "mean_token_accuracy": 0.039727197097818134,
+      "step": 34
+    },
+    {
+      "epoch": 0.12429253135057153,
+      "grad_norm": 1.222414255142212,
+      "learning_rate": 9.984614141547468e-06,
+      "loss": 0.2449,
+      "mean_token_accuracy": 0.04050094210106181,
+      "step": 35
+    },
+    {
+      "epoch": 0.1278437465320164,
+      "grad_norm": 1.3800718784332275,
+      "learning_rate": 9.983496142613314e-06,
+      "loss": 0.254,
+      "mean_token_accuracy": 0.03262984177126782,
+      "step": 36
+    },
+    {
+      "epoch": 0.13139496171346132,
+      "grad_norm": 1.6306910514831543,
+      "learning_rate": 9.982339003540272e-06,
+      "loss": 0.2583,
+      "mean_token_accuracy": 0.03810747154057026,
+      "step": 37
+    },
+    {
+      "epoch": 0.13494617689490623,
+      "grad_norm": 1.1124504804611206,
+      "learning_rate": 9.981142733416457e-06,
+      "loss": 0.258,
+      "mean_token_accuracy": 0.03563006369222421,
+      "step": 38
+    },
+    {
+      "epoch": 0.13849739207635112,
+      "grad_norm": 1.0881694555282593,
+      "learning_rate": 9.97990734163732e-06,
+      "loss": 0.2464,
+      "mean_token_accuracy": 0.0357487567125645,
+      "step": 39
+    },
+    {
+      "epoch": 0.14204860725779603,
+      "grad_norm": 1.3406707048416138,
+      "learning_rate": 9.978632837905566e-06,
+      "loss": 0.2492,
+      "mean_token_accuracy": 0.03550324517709669,
+      "step": 40
+    },
+    {
+      "epoch": 0.14559982243924094,
+      "grad_norm": 1.7592324018478394,
+      "learning_rate": 9.977319232231088e-06,
+      "loss": 0.2534,
+      "mean_token_accuracy": 0.03265573883254547,
+      "step": 41
+    },
+    {
+      "epoch": 0.14915103762068582,
+      "grad_norm": 0.9613096117973328,
+      "learning_rate": 9.975966534930879e-06,
+      "loss": 0.2496,
+      "mean_token_accuracy": 0.03335131827407167,
+      "step": 42
+    },
+    {
+      "epoch": 0.15270225280213073,
+      "grad_norm": 1.0681848526000977,
+      "learning_rate": 9.974574756628961e-06,
+      "loss": 0.2521,
+      "mean_token_accuracy": 0.033948773343581706,
+      "step": 43
+    },
+    {
+      "epoch": 0.15625346798357562,
+      "grad_norm": 2.593073844909668,
+      "learning_rate": 9.973143908256291e-06,
+      "loss": 0.2533,
+      "mean_token_accuracy": 0.03512008564939606,
+      "step": 44
+    },
+    {
+      "epoch": 0.15980468316502053,
+      "grad_norm": 1.112453579902649,
+      "learning_rate": 9.971674001050687e-06,
+      "loss": 0.2485,
+      "mean_token_accuracy": 0.038668419925670605,
+      "step": 45
+    },
+    {
+      "epoch": 0.16335589834646544,
+      "grad_norm": 1.1808503866195679,
+      "learning_rate": 9.970165046556726e-06,
+      "loss": 0.2422,
+      "mean_token_accuracy": 0.042593366157234414,
+      "step": 46
+    },
+    {
+      "epoch": 0.16690711352791032,
+      "grad_norm": 1.1615327596664429,
+      "learning_rate": 9.968617056625665e-06,
+      "loss": 0.2469,
+      "mean_token_accuracy": 0.03583648353378521,
+      "step": 47
+    },
+    {
+      "epoch": 0.17045832870935523,
+      "grad_norm": 0.9273613691329956,
+      "learning_rate": 9.967030043415345e-06,
+      "loss": 0.2494,
+      "mean_token_accuracy": 0.03705584820272634,
+      "step": 48
+    },
+    {
+      "epoch": 0.17400954389080014,
+      "grad_norm": 1.2233788967132568,
+      "learning_rate": 9.965404019390087e-06,
+      "loss": 0.2617,
+      "mean_token_accuracy": 0.03263758916364168,
+      "step": 49
+    },
+    {
+      "epoch": 0.17756075907224503,
+      "grad_norm": 1.7746872901916504,
+      "learning_rate": 9.963738997320609e-06,
+      "loss": 0.2565,
+      "mean_token_accuracy": 0.029688541861105477,
+      "step": 50
+    },
+    {
+      "epoch": 0.18111197425368994,
+      "grad_norm": 1.6324982643127441,
+      "learning_rate": 9.962034990283912e-06,
+      "loss": 0.2711,
+      "mean_token_accuracy": 0.031018283694720594,
+      "step": 51
+    },
+    {
+      "epoch": 0.18466318943513482,
+      "grad_norm": 1.072430968284607,
+      "learning_rate": 9.960292011663186e-06,
+      "loss": 0.2416,
+      "mean_token_accuracy": 0.03291421085668844,
+      "step": 52
+    },
+    {
+      "epoch": 0.18821440461657973,
+      "grad_norm": 1.6084487438201904,
+      "learning_rate": 9.958510075147703e-06,
+      "loss": 0.2537,
+      "mean_token_accuracy": 0.03283511256086058,
+      "step": 53
+    },
+    {
+      "epoch": 0.19176561979802464,
+      "grad_norm": 3.659834623336792,
+      "learning_rate": 9.956689194732702e-06,
+      "loss": 0.2606,
+      "mean_token_accuracy": 0.03444362949267088,
+      "step": 54
+    },
+    {
+      "epoch": 0.19531683497946953,
+      "grad_norm": 1.3310655355453491,
+      "learning_rate": 9.954829384719296e-06,
+      "loss": 0.2436,
+      "mean_token_accuracy": 0.0340039685870579,
+      "step": 55
+    },
+    {
+      "epoch": 0.19886805016091444,
+      "grad_norm": 1.2658661603927612,
+      "learning_rate": 9.95293065971434e-06,
+      "loss": 0.2523,
+      "mean_token_accuracy": 0.03461971837168676,
+      "step": 56
+    },
+    {
+      "epoch": 0.20241926534235935,
+      "grad_norm": 0.9548496007919312,
+      "learning_rate": 9.950993034630328e-06,
+      "loss": 0.2563,
+      "mean_token_accuracy": 0.03424741265553166,
+      "step": 57
+    },
+    {
+      "epoch": 0.20597048052380423,
+      "grad_norm": 1.5720734596252441,
+      "learning_rate": 9.949016524685277e-06,
+      "loss": 0.246,
+      "mean_token_accuracy": 0.03490980070273508,
+      "step": 58
+    },
+    {
+      "epoch": 0.20952169570524914,
+      "grad_norm": 1.0101721286773682,
+      "learning_rate": 9.947001145402598e-06,
+      "loss": 0.2424,
+      "mean_token_accuracy": 0.033455477951065404,
+      "step": 59
+    },
+    {
+      "epoch": 0.21307291088669403,
+      "grad_norm": 0.8857386112213135,
+      "learning_rate": 9.944946912610986e-06,
+      "loss": 0.2361,
+      "mean_token_accuracy": 0.03596452736383071,
+      "step": 60
+    },
+    {
+      "epoch": 0.21662412606813894,
+      "grad_norm": 1.0570611953735352,
+      "learning_rate": 9.942853842444283e-06,
+      "loss": 0.256,
+      "mean_token_accuracy": 0.03371396200964227,
+      "step": 61
+    },
+    {
+      "epoch": 0.22017534124958385,
+      "grad_norm": 0.9776428937911987,
+      "learning_rate": 9.940721951341365e-06,
+      "loss": 0.2467,
+      "mean_token_accuracy": 0.037202409857854946,
+      "step": 62
+    },
+    {
+      "epoch": 0.22372655643102873,
+      "grad_norm": 1.1859705448150635,
+      "learning_rate": 9.938551256046e-06,
+      "loss": 0.2418,
+      "mean_token_accuracy": 0.038250373654591385,
+      "step": 63
+    },
+    {
+      "epoch": 0.22727777161247364,
+      "grad_norm": 1.2014352083206177,
+      "learning_rate": 9.936341773606723e-06,
+      "loss": 0.2399,
+      "mean_token_accuracy": 0.03623262169639929,
+      "step": 64
+    },
+    {
+      "epoch": 0.23082898679391856,
+      "grad_norm": 1.0376017093658447,
+      "learning_rate": 9.934093521376707e-06,
+      "loss": 0.2344,
+      "mean_token_accuracy": 0.038730696996935876,
+      "step": 65
+    },
+    {
+      "epoch": 0.23438020197536344,
+      "grad_norm": 1.1257619857788086,
+      "learning_rate": 9.931806517013612e-06,
+      "loss": 0.2415,
+      "mean_token_accuracy": 0.03321881674128235,
+      "step": 66
+    },
+    {
+      "epoch": 0.23793141715680835,
+      "grad_norm": 1.2318155765533447,
+      "learning_rate": 9.929480778479465e-06,
+      "loss": 0.2437,
+      "mean_token_accuracy": 0.037453887685842346,
+      "step": 67
+    },
+    {
+      "epoch": 0.24148263233825323,
+      "grad_norm": 2.6596481800079346,
+      "learning_rate": 9.9271163240405e-06,
+      "loss": 0.2399,
+      "mean_token_accuracy": 0.03390174064406892,
+      "step": 68
+    },
+    {
+      "epoch": 0.24503384751969814,
+      "grad_norm": 1.1814781427383423,
+      "learning_rate": 9.92471317226703e-06,
+      "loss": 0.2411,
+      "mean_token_accuracy": 0.03444516302442935,
+      "step": 69
+    },
+    {
+      "epoch": 0.24858506270114306,
+      "grad_norm": 1.9260839223861694,
+      "learning_rate": 9.922271342033295e-06,
+      "loss": 0.232,
+      "mean_token_accuracy": 0.037160640149522806,
+      "step": 70
+    },
+    {
+      "epoch": 0.25213627788258797,
+      "grad_norm": 0.9885060787200928,
+      "learning_rate": 9.919790852517313e-06,
+      "loss": 0.2329,
+      "mean_token_accuracy": 0.03519195963963284,
+      "step": 71
+    },
+    {
+      "epoch": 0.2556874930640328,
+      "grad_norm": 1.4856972694396973,
+      "learning_rate": 9.917271723200725e-06,
+      "loss": 0.2477,
+      "mean_token_accuracy": 0.03523147122905357,
+      "step": 72
+    },
+    {
+      "epoch": 0.25923870824547773,
+      "grad_norm": 1.6875951290130615,
+      "learning_rate": 9.914713973868654e-06,
+      "loss": 0.2531,
+      "mean_token_accuracy": 0.034952133555634646,
+      "step": 73
+    },
+    {
+      "epoch": 0.26278992342692264,
+      "grad_norm": 1.6447116136550903,
+      "learning_rate": 9.91211762460954e-06,
+      "loss": 0.2591,
+      "mean_token_accuracy": 0.031627864758775104,
+      "step": 74
+    },
+    {
+      "epoch": 0.26634113860836756,
+      "grad_norm": 1.2278292179107666,
+      "learning_rate": 9.909482695814986e-06,
+      "loss": 0.2348,
+      "mean_token_accuracy": 0.034353334242041456,
+      "step": 75
+    },
+    {
+      "epoch": 0.26989235378981247,
+      "grad_norm": 1.049460530281067,
+      "learning_rate": 9.906809208179593e-06,
+      "loss": 0.2457,
+      "mean_token_accuracy": 0.038979591363386135,
+      "step": 76
+    },
+    {
+      "epoch": 0.2734435689712574,
+      "grad_norm": 1.2099536657333374,
+      "learning_rate": 9.904097182700806e-06,
+      "loss": 0.2524,
+      "mean_token_accuracy": 0.0339672717054782,
+      "step": 77
+    },
+    {
+      "epoch": 0.27699478415270223,
+      "grad_norm": 1.4347492456436157,
+      "learning_rate": 9.901346640678744e-06,
+      "loss": 0.2311,
+      "mean_token_accuracy": 0.03691132989479229,
+      "step": 78
+    },
+    {
+      "epoch": 0.28054599933414714,
+      "grad_norm": 1.3881317377090454,
+      "learning_rate": 9.898557603716031e-06,
+      "loss": 0.2574,
+      "mean_token_accuracy": 0.03597194645408308,
+      "step": 79
+    },
+    {
+      "epoch": 0.28409721451559206,
+      "grad_norm": 0.9869890809059143,
+      "learning_rate": 9.895730093717629e-06,
+      "loss": 0.2413,
+      "mean_token_accuracy": 0.034210870442620944,
+      "step": 80
+    },
+    {
+      "epoch": 0.28764842969703697,
+      "grad_norm": 1.0517185926437378,
+      "learning_rate": 9.892864132890663e-06,
+      "loss": 0.2441,
+      "mean_token_accuracy": 0.036712895660457434,
+      "step": 81
+    },
+    {
+      "epoch": 0.2911996448784819,
+      "grad_norm": 1.3158183097839355,
+      "learning_rate": 9.889959743744253e-06,
+      "loss": 0.2381,
+      "mean_token_accuracy": 0.035375316572753945,
+      "step": 82
+    },
+    {
+      "epoch": 0.29475086005992673,
+      "grad_norm": 1.3365075588226318,
+      "learning_rate": 9.887016949089334e-06,
+      "loss": 0.2514,
+      "mean_token_accuracy": 0.03394350171038241,
+      "step": 83
+    },
+    {
+      "epoch": 0.29830207524137164,
+      "grad_norm": 0.9649895429611206,
+      "learning_rate": 9.884035772038471e-06,
+      "loss": 0.236,
+      "mean_token_accuracy": 0.042694962285168,
+      "step": 84
+    },
+    {
+      "epoch": 0.30185329042281656,
+      "grad_norm": 1.5863405466079712,
+      "learning_rate": 9.881016236005686e-06,
+      "loss": 0.2345,
+      "mean_token_accuracy": 0.035104533468256705,
+      "step": 85
+    },
+    {
+      "epoch": 0.30540450560426147,
+      "grad_norm": 0.9601231217384338,
+      "learning_rate": 9.877958364706269e-06,
+      "loss": 0.2415,
+      "mean_token_accuracy": 0.03544433661591029,
+      "step": 86
+    },
+    {
+      "epoch": 0.3089557207857064,
+      "grad_norm": 1.2249877452850342,
+      "learning_rate": 9.874862182156596e-06,
+      "loss": 0.2467,
+      "mean_token_accuracy": 0.029483627313311445,
+      "step": 87
+    },
+    {
+      "epoch": 0.31250693596715123,
+      "grad_norm": 0.9038888216018677,
+      "learning_rate": 9.871727712673931e-06,
+      "loss": 0.2406,
+      "mean_token_accuracy": 0.03652138033430674,
+      "step": 88
+    },
+    {
+      "epoch": 0.31605815114859614,
+      "grad_norm": 1.1572030782699585,
+      "learning_rate": 9.868554980876253e-06,
+      "loss": 0.2352,
+      "mean_token_accuracy": 0.03487317850886029,
+      "step": 89
+    },
+    {
+      "epoch": 0.31960936633004106,
+      "grad_norm": 1.7031879425048828,
+      "learning_rate": 9.865344011682038e-06,
+      "loss": 0.2489,
+      "mean_token_accuracy": 0.03696917567867786,
+      "step": 90
+    },
+    {
+      "epoch": 0.32316058151148597,
+      "grad_norm": 1.187495470046997,
+      "learning_rate": 9.86209483031009e-06,
+      "loss": 0.244,
+      "mean_token_accuracy": 0.03272304959318717,
+      "step": 91
+    },
+    {
+      "epoch": 0.3267117966929309,
+      "grad_norm": 1.3152642250061035,
+      "learning_rate": 9.858807462279319e-06,
+      "loss": 0.2464,
+      "mean_token_accuracy": 0.03474508359613537,
+      "step": 92
+    },
+    {
+      "epoch": 0.3302630118743758,
+      "grad_norm": 0.9680661559104919,
+      "learning_rate": 9.855481933408557e-06,
+      "loss": 0.2378,
+      "mean_token_accuracy": 0.035624928095785435,
+      "step": 93
+    },
+    {
+      "epoch": 0.33381422705582064,
+      "grad_norm": 1.0342296361923218,
+      "learning_rate": 9.852118269816348e-06,
+      "loss": 0.242,
+      "mean_token_accuracy": 0.03397064344426326,
+      "step": 94
+    },
+    {
+      "epoch": 0.33736544223726556,
+      "grad_norm": 1.5446665287017822,
+      "learning_rate": 9.848716497920742e-06,
+      "loss": 0.2406,
+      "mean_token_accuracy": 0.03505690078418411,
+      "step": 95
+    },
+    {
+      "epoch": 0.34091665741871047,
+      "grad_norm": 1.498883843421936,
+      "learning_rate": 9.845276644439093e-06,
+      "loss": 0.2364,
+      "mean_token_accuracy": 0.036183488669848884,
+      "step": 96
+    },
+    {
+      "epoch": 0.3444678726001554,
+      "grad_norm": 1.212774395942688,
+      "learning_rate": 9.841798736387846e-06,
+      "loss": 0.2499,
+      "mean_token_accuracy": 0.03257646691054106,
+      "step": 97
+    },
+    {
+      "epoch": 0.3480190877816003,
+      "grad_norm": 0.9920580983161926,
+      "learning_rate": 9.838282801082322e-06,
+      "loss": 0.2363,
+      "mean_token_accuracy": 0.033475329148132005,
+      "step": 98
+    },
+    {
+      "epoch": 0.35157030296304514,
+      "grad_norm": 1.1953113079071045,
+      "learning_rate": 9.834728866136506e-06,
+      "loss": 0.2534,
+      "mean_token_accuracy": 0.03439514616547967,
+      "step": 99
+    },
+    {
+      "epoch": 0.35512151814449006,
+      "grad_norm": 1.2545762062072754,
+      "learning_rate": 9.831136959462835e-06,
+      "loss": 0.2405,
+      "mean_token_accuracy": 0.035495126408932265,
+      "step": 100
+    },
+    {
+      "epoch": 0.35867273332593497,
+      "grad_norm": 1.419901728630066,
+      "learning_rate": 9.82750710927197e-06,
+      "loss": 0.2283,
+      "mean_token_accuracy": 0.03381325537702651,
+      "step": 101
+    },
+    {
+      "epoch": 0.3622239485073799,
+      "grad_norm": 0.9885729551315308,
+      "learning_rate": 9.823839344072582e-06,
+      "loss": 0.2403,
+      "mean_token_accuracy": 0.03792576077466947,
+      "step": 102
+    },
+    {
+      "epoch": 0.3657751636888248,
+      "grad_norm": 0.9770134687423706,
+      "learning_rate": 9.820133692671116e-06,
+      "loss": 0.2382,
+      "mean_token_accuracy": 0.03301898992503993,
+      "step": 103
+    },
+    {
+      "epoch": 0.36932637887026964,
+      "grad_norm": 1.1737765073776245,
+      "learning_rate": 9.816390184171587e-06,
+      "loss": 0.2399,
+      "mean_token_accuracy": 0.03295345153310336,
+      "step": 104
+    },
+    {
+      "epoch": 0.37287759405171456,
+      "grad_norm": 1.1783053874969482,
+      "learning_rate": 9.812608847975327e-06,
+      "loss": 0.2271,
+      "mean_token_accuracy": 0.04089054596624919,
+      "step": 105
+    },
+    {
+      "epoch": 0.37642880923315947,
+      "grad_norm": 1.124003291130066,
+      "learning_rate": 9.808789713780768e-06,
+      "loss": 0.2473,
+      "mean_token_accuracy": 0.038133521022245986,
+      "step": 106
+    },
+    {
+      "epoch": 0.3799800244146044,
+      "grad_norm": 1.0809928178787231,
+      "learning_rate": 9.804932811583208e-06,
+      "loss": 0.2311,
+      "mean_token_accuracy": 0.037309447398001794,
+      "step": 107
+    },
+    {
+      "epoch": 0.3835312395960493,
+      "grad_norm": 1.0467100143432617,
+      "learning_rate": 9.801038171674571e-06,
+      "loss": 0.2353,
+      "mean_token_accuracy": 0.03867107529367786,
+      "step": 108
+    },
+    {
+      "epoch": 0.3870824547774942,
+      "grad_norm": 1.0050996541976929,
+      "learning_rate": 9.797105824643171e-06,
+      "loss": 0.2551,
+      "mean_token_accuracy": 0.030720586502866354,
+      "step": 109
+    },
+    {
+      "epoch": 0.39063366995893906,
+      "grad_norm": 0.9074941277503967,
+      "learning_rate": 9.793135801373472e-06,
+      "loss": 0.2436,
+      "mean_token_accuracy": 0.034242983878357336,
+      "step": 110
+    },
+    {
+      "epoch": 0.39418488514038397,
+      "grad_norm": 1.3338314294815063,
+      "learning_rate": 9.789128133045846e-06,
+      "loss": 0.2319,
+      "mean_token_accuracy": 0.04001064739713911,
+      "step": 111
+    },
+    {
+      "epoch": 0.3977361003218289,
+      "grad_norm": 0.9372747540473938,
+      "learning_rate": 9.785082851136327e-06,
+      "loss": 0.23,
+      "mean_token_accuracy": 0.0371717325233476,
+      "step": 112
+    },
+    {
+      "epoch": 0.4012873155032738,
+      "grad_norm": 1.2198783159255981,
+      "learning_rate": 9.780999987416363e-06,
+      "loss": 0.247,
+      "mean_token_accuracy": 0.03400729462009622,
+      "step": 113
+    },
+    {
+      "epoch": 0.4048385306847187,
+      "grad_norm": 1.3612855672836304,
+      "learning_rate": 9.776879573952573e-06,
+      "loss": 0.2483,
+      "mean_token_accuracy": 0.037493000188987935,
+      "step": 114
+    },
+    {
+      "epoch": 0.40838974586616356,
+      "grad_norm": 1.3870880603790283,
+      "learning_rate": 9.772721643106483e-06,
+      "loss": 0.2424,
+      "mean_token_accuracy": 0.03547792824974749,
+      "step": 115
+    },
+    {
+      "epoch": 0.41194096104760847,
+      "grad_norm": 0.9454783201217651,
+      "learning_rate": 9.768526227534286e-06,
+      "loss": 0.2369,
+      "mean_token_accuracy": 0.034673961708904244,
+      "step": 116
+    },
+    {
+      "epoch": 0.4154921762290534,
+      "grad_norm": 1.094822645187378,
+      "learning_rate": 9.764293360186568e-06,
+      "loss": 0.2376,
+      "mean_token_accuracy": 0.03994236302605714,
+      "step": 117
+    },
+    {
+      "epoch": 0.4190433914104983,
+      "grad_norm": 1.2195683717727661,
+      "learning_rate": 9.760023074308067e-06,
+      "loss": 0.2481,
+      "mean_token_accuracy": 0.035971136057924014,
+      "step": 118
+    },
+    {
+      "epoch": 0.4225946065919432,
+      "grad_norm": 1.001287579536438,
+      "learning_rate": 9.755715403437405e-06,
+      "loss": 0.2396,
+      "mean_token_accuracy": 0.030299254751298577,
+      "step": 119
+    },
+    {
+      "epoch": 0.42614582177338806,
+      "grad_norm": 1.384993314743042,
+      "learning_rate": 9.75137038140682e-06,
+      "loss": 0.2349,
+      "mean_token_accuracy": 0.034674650321903755,
+      "step": 120
+    },
+    {
+      "epoch": 0.42969703695483297,
+      "grad_norm": 1.7163217067718506,
+      "learning_rate": 9.746988042341907e-06,
+      "loss": 0.2347,
+      "mean_token_accuracy": 0.0343936083263543,
+      "step": 121
+    },
+    {
+      "epoch": 0.4332482521362779,
+      "grad_norm": 1.5516914129257202,
+      "learning_rate": 9.742568420661347e-06,
+      "loss": 0.2314,
+      "mean_token_accuracy": 0.03660766014945693,
+      "step": 122
+    },
+    {
+      "epoch": 0.4367994673177228,
+      "grad_norm": 1.278739094734192,
+      "learning_rate": 9.738111551076633e-06,
+      "loss": 0.2428,
+      "mean_token_accuracy": 0.0372478184508509,
+      "step": 123
+    },
+    {
+      "epoch": 0.4403506824991677,
+      "grad_norm": 1.1330451965332031,
+      "learning_rate": 9.733617468591806e-06,
+      "loss": 0.2318,
+      "mean_token_accuracy": 0.03080255860913894,
+      "step": 124
+    },
+    {
+      "epoch": 0.4439018976806126,
+      "grad_norm": 1.3162201642990112,
+      "learning_rate": 9.729086208503174e-06,
+      "loss": 0.2396,
+      "mean_token_accuracy": 0.037999619937181706,
+      "step": 125
+    },
+    {
+      "epoch": 0.44745311286205747,
+      "grad_norm": 0.9050405621528625,
+      "learning_rate": 9.724517806399035e-06,
+      "loss": 0.2318,
+      "mean_token_accuracy": 0.03370005947726895,
+      "step": 126
+    },
+    {
+      "epoch": 0.4510043280435024,
+      "grad_norm": 0.9520599842071533,
+      "learning_rate": 9.7199122981594e-06,
+      "loss": 0.2492,
+      "mean_token_accuracy": 0.03589029877548455,
+      "step": 127
+    },
+    {
+      "epoch": 0.4545555432249473,
+      "grad_norm": 1.1406325101852417,
+      "learning_rate": 9.715269719955708e-06,
+      "loss": 0.2527,
+      "mean_token_accuracy": 0.033397240742488066,
+      "step": 128
+    },
+    {
+      "epoch": 0.4581067584063922,
+      "grad_norm": 1.3370436429977417,
+      "learning_rate": 9.710590108250546e-06,
+      "loss": 0.2448,
+      "mean_token_accuracy": 0.03260400308499811,
+      "step": 129
+    },
+    {
+      "epoch": 0.4616579735878371,
+      "grad_norm": 1.1314901113510132,
+      "learning_rate": 9.705873499797358e-06,
+      "loss": 0.2361,
+      "mean_token_accuracy": 0.03638589083129773,
+      "step": 130
+    },
+    {
+      "epoch": 0.46520918876928197,
+      "grad_norm": 0.8222520351409912,
+      "learning_rate": 9.701119931640161e-06,
+      "loss": 0.2369,
+      "mean_token_accuracy": 0.039288555715756956,
+      "step": 131
+    },
+    {
+      "epoch": 0.4687604039507269,
+      "grad_norm": 1.8080474138259888,
+      "learning_rate": 9.69632944111325e-06,
+      "loss": 0.2457,
+      "mean_token_accuracy": 0.03426650948313181,
+      "step": 132
+    },
+    {
+      "epoch": 0.4723116191321718,
+      "grad_norm": 1.0059517621994019,
+      "learning_rate": 9.691502065840905e-06,
+      "loss": 0.2318,
+      "mean_token_accuracy": 0.041799060640187236,
+      "step": 133
+    },
+    {
+      "epoch": 0.4758628343136167,
+      "grad_norm": 1.5800079107284546,
+      "learning_rate": 9.686637843737104e-06,
+      "loss": 0.2453,
+      "mean_token_accuracy": 0.035510726302163675,
+      "step": 134
+    },
+    {
+      "epoch": 0.4794140494950616,
+      "grad_norm": 1.1752355098724365,
+      "learning_rate": 9.681736813005207e-06,
+      "loss": 0.243,
+      "mean_token_accuracy": 0.03367971369516454,
+      "step": 135
+    },
+    {
+      "epoch": 0.48296526467650647,
+      "grad_norm": 1.1899182796478271,
+      "learning_rate": 9.676799012137678e-06,
+      "loss": 0.2333,
+      "mean_token_accuracy": 0.03657108947663801,
+      "step": 136
+    },
+    {
+      "epoch": 0.4865164798579514,
+      "grad_norm": 0.9200431704521179,
+      "learning_rate": 9.671824479915768e-06,
+      "loss": 0.2228,
+      "mean_token_accuracy": 0.04016808481901535,
+      "step": 137
+    },
+    {
+      "epoch": 0.4900676950393963,
+      "grad_norm": 1.828884243965149,
+      "learning_rate": 9.666813255409212e-06,
+      "loss": 0.249,
+      "mean_token_accuracy": 0.03349095129669877,
+      "step": 138
+    },
+    {
+      "epoch": 0.4936189102208412,
+      "grad_norm": 0.9927622079849243,
+      "learning_rate": 9.661765377975924e-06,
+      "loss": 0.2457,
+      "mean_token_accuracy": 0.035240638426330406,
+      "step": 139
+    },
+    {
+      "epoch": 0.4971701254022861,
+      "grad_norm": 1.5192276239395142,
+      "learning_rate": 9.656680887261693e-06,
+      "loss": 0.2209,
+      "mean_token_accuracy": 0.034234478349389974,
+      "step": 140
+    },
+    {
+      "epoch": 0.500721340583731,
+      "grad_norm": 1.0839775800704956,
+      "learning_rate": 9.651559823199865e-06,
+      "loss": 0.2449,
+      "mean_token_accuracy": 0.03251597065900569,
+      "step": 141
+    },
+    {
+      "epoch": 0.5042725557651759,
+      "grad_norm": 1.23935866355896,
+      "learning_rate": 9.646402226011028e-06,
+      "loss": 0.2435,
+      "mean_token_accuracy": 0.03836617184606439,
+      "step": 142
+    },
+    {
+      "epoch": 0.5078237709466208,
+      "grad_norm": 1.1805026531219482,
+      "learning_rate": 9.641208136202705e-06,
+      "loss": 0.2317,
+      "mean_token_accuracy": 0.037522376484048436,
+      "step": 143
+    },
+    {
+      "epoch": 0.5113749861280656,
+      "grad_norm": 1.4856340885162354,
+      "learning_rate": 9.635977594569025e-06,
+      "loss": 0.2464,
+      "mean_token_accuracy": 0.03469312738525332,
+      "step": 144
+    },
+    {
+      "epoch": 0.5149262013095106,
+      "grad_norm": 1.1366444826126099,
+      "learning_rate": 9.630710642190412e-06,
+      "loss": 0.2318,
+      "mean_token_accuracy": 0.03400819407397648,
+      "step": 145
+    },
+    {
+      "epoch": 0.5184774164909555,
+      "grad_norm": 1.1727653741836548,
+      "learning_rate": 9.625407320433257e-06,
+      "loss": 0.2378,
+      "mean_token_accuracy": 0.0386261100647971,
+      "step": 146
+    },
+    {
+      "epoch": 0.5220286316724004,
+      "grad_norm": 1.0548739433288574,
+      "learning_rate": 9.620067670949593e-06,
+      "loss": 0.2375,
+      "mean_token_accuracy": 0.03571181180086569,
+      "step": 147
+    },
+    {
+      "epoch": 0.5255798468538453,
+      "grad_norm": 1.2281569242477417,
+      "learning_rate": 9.614691735676768e-06,
+      "loss": 0.2358,
+      "mean_token_accuracy": 0.034898955855169334,
+      "step": 148
+    },
+    {
+      "epoch": 0.5291310620352903,
+      "grad_norm": 1.2215083837509155,
+      "learning_rate": 9.609279556837122e-06,
+      "loss": 0.2364,
+      "mean_token_accuracy": 0.03552666449104436,
+      "step": 149
+    },
+    {
+      "epoch": 0.5326822772167351,
+      "grad_norm": 1.1264266967773438,
+      "learning_rate": 9.603831176937645e-06,
+      "loss": 0.2429,
+      "mean_token_accuracy": 0.03438397626450751,
+      "step": 150
+    },
+    {
+      "epoch": 0.53623349239818,
+      "grad_norm": 0.9536157846450806,
+      "learning_rate": 9.598346638769653e-06,
+      "loss": 0.2457,
+      "mean_token_accuracy": 0.03396081876780954,
+      "step": 151
+    },
+    {
+      "epoch": 0.5397847075796249,
+      "grad_norm": 1.115391731262207,
+      "learning_rate": 9.592825985408443e-06,
+      "loss": 0.2324,
+      "mean_token_accuracy": 0.0359127135889139,
+      "step": 152
+    },
+    {
+      "epoch": 0.5433359227610698,
+      "grad_norm": 1.0178639888763428,
+      "learning_rate": 9.58726926021296e-06,
+      "loss": 0.2312,
+      "mean_token_accuracy": 0.0330092023978068,
+      "step": 153
+    },
+    {
+      "epoch": 0.5468871379425148,
+      "grad_norm": 1.2052620649337769,
+      "learning_rate": 9.581676506825458e-06,
+      "loss": 0.2263,
+      "mean_token_accuracy": 0.038123431153508136,
+      "step": 154
+    },
+    {
+      "epoch": 0.5504383531239596,
+      "grad_norm": 1.4705978631973267,
+      "learning_rate": 9.576047769171154e-06,
+      "loss": 0.2442,
+      "mean_token_accuracy": 0.038363178242434515,
+      "step": 155
+    },
+    {
+      "epoch": 0.5539895683054045,
+      "grad_norm": 1.043751835823059,
+      "learning_rate": 9.57038309145788e-06,
+      "loss": 0.2371,
+      "mean_token_accuracy": 0.03430595702957362,
+      "step": 156
+    },
+    {
+      "epoch": 0.5575407834868494,
+      "grad_norm": 1.1250089406967163,
+      "learning_rate": 9.564682518175745e-06,
+      "loss": 0.2289,
+      "mean_token_accuracy": 0.034604653304995736,
+      "step": 157
+    },
+    {
+      "epoch": 0.5610919986682943,
+      "grad_norm": 1.0135129690170288,
+      "learning_rate": 9.558946094096773e-06,
+      "loss": 0.2437,
+      "mean_token_accuracy": 0.03395201343300869,
+      "step": 158
+    },
+    {
+      "epoch": 0.5646432138497393,
+      "grad_norm": 1.4492766857147217,
+      "learning_rate": 9.553173864274567e-06,
+      "loss": 0.2324,
+      "mean_token_accuracy": 0.03644484080723487,
+      "step": 159
+    },
+    {
+      "epoch": 0.5681944290311841,
+      "grad_norm": 1.1282693147659302,
+      "learning_rate": 9.547365874043939e-06,
+      "loss": 0.2241,
+      "mean_token_accuracy": 0.03566149080506875,
+      "step": 160
+    },
+    {
+      "epoch": 0.571745644212629,
+      "grad_norm": 1.1843982934951782,
+      "learning_rate": 9.541522169020568e-06,
+      "loss": 0.2326,
+      "mean_token_accuracy": 0.03567118446153472,
+      "step": 161
+    },
+    {
+      "epoch": 0.5752968593940739,
+      "grad_norm": 0.8801372647285461,
+      "learning_rate": 9.535642795100628e-06,
+      "loss": 0.227,
+      "mean_token_accuracy": 0.03749654847706552,
+      "step": 162
+    },
+    {
+      "epoch": 0.5788480745755188,
+      "grad_norm": 1.0453648567199707,
+      "learning_rate": 9.529727798460443e-06,
+      "loss": 0.2282,
+      "mean_token_accuracy": 0.032709133800381096,
+      "step": 163
+    },
+    {
+      "epoch": 0.5823992897569638,
+      "grad_norm": 1.3325905799865723,
+      "learning_rate": 9.52377722555611e-06,
+      "loss": 0.2356,
+      "mean_token_accuracy": 0.035057135524766636,
+      "step": 164
+    },
+    {
+      "epoch": 0.5859505049384086,
+      "grad_norm": 1.281227707862854,
+      "learning_rate": 9.517791123123141e-06,
+      "loss": 0.2365,
+      "mean_token_accuracy": 0.036725746911542956,
+      "step": 165
+    },
+    {
+      "epoch": 0.5895017201198535,
+      "grad_norm": 1.3783758878707886,
+      "learning_rate": 9.5117695381761e-06,
+      "loss": 0.2363,
+      "mean_token_accuracy": 0.03276101666051545,
+      "step": 166
+    },
+    {
+      "epoch": 0.5930529353012984,
+      "grad_norm": 1.5050840377807617,
+      "learning_rate": 9.50571251800822e-06,
+      "loss": 0.2369,
+      "mean_token_accuracy": 0.03802061551141378,
+      "step": 167
+    },
+    {
+      "epoch": 0.5966041504827433,
+      "grad_norm": 0.8776186108589172,
+      "learning_rate": 9.49962011019105e-06,
+      "loss": 0.2455,
+      "mean_token_accuracy": 0.030048082109715324,
+      "step": 168
+    },
+    {
+      "epoch": 0.6001553656641883,
+      "grad_norm": 1.001824975013733,
+      "learning_rate": 9.493492362574069e-06,
+      "loss": 0.2132,
+      "mean_token_accuracy": 0.04021573403588263,
+      "step": 169
+    },
+    {
+      "epoch": 0.6037065808456331,
+      "grad_norm": 0.9855964779853821,
+      "learning_rate": 9.487329323284306e-06,
+      "loss": 0.2409,
+      "mean_token_accuracy": 0.03264193019276718,
+      "step": 170
+    },
+    {
+      "epoch": 0.607257796027078,
+      "grad_norm": 2.301527976989746,
+      "learning_rate": 9.481131040725982e-06,
+      "loss": 0.2431,
+      "mean_token_accuracy": 0.03716234745661495,
+      "step": 171
+    },
+    {
+      "epoch": 0.6108090112085229,
+      "grad_norm": 1.8789873123168945,
+      "learning_rate": 9.474897563580105e-06,
+      "loss": 0.2187,
+      "mean_token_accuracy": 0.03926722923642956,
+      "step": 172
+    },
+    {
+      "epoch": 0.6143602263899678,
+      "grad_norm": 1.0674188137054443,
+      "learning_rate": 9.468628940804109e-06,
+      "loss": 0.2238,
+      "mean_token_accuracy": 0.039616512553038774,
+      "step": 173
+    },
+    {
+      "epoch": 0.6179114415714128,
+      "grad_norm": 1.457808256149292,
+      "learning_rate": 9.46232522163145e-06,
+      "loss": 0.2439,
+      "mean_token_accuracy": 0.03321230758956517,
+      "step": 174
+    },
+    {
+      "epoch": 0.6214626567528576,
+      "grad_norm": 0.8303464651107788,
+      "learning_rate": 9.45598645557124e-06,
+      "loss": 0.2333,
+      "mean_token_accuracy": 0.035840667918819236,
+      "step": 175
+    },
+    {
+      "epoch": 0.6250138719343025,
+      "grad_norm": 3.895867347717285,
+      "learning_rate": 9.44961269240784e-06,
+      "loss": 0.2339,
+      "mean_token_accuracy": 0.03322799965098966,
+      "step": 176
+    },
+    {
+      "epoch": 0.6285650871157474,
+      "grad_norm": 1.2655450105667114,
+      "learning_rate": 9.443203982200479e-06,
+      "loss": 0.2463,
+      "mean_token_accuracy": 0.03559355030665756,
+      "step": 177
+    },
+    {
+      "epoch": 0.6321163022971923,
+      "grad_norm": 1.4873794317245483,
+      "learning_rate": 9.436760375282858e-06,
+      "loss": 0.222,
+      "mean_token_accuracy": 0.04104303981876001,
+      "step": 178
+    },
+    {
+      "epoch": 0.6356675174786373,
+      "grad_norm": 1.4663066864013672,
+      "learning_rate": 9.430281922262758e-06,
+      "loss": 0.2258,
+      "mean_token_accuracy": 0.03942336593900109,
+      "step": 179
+    },
+    {
+      "epoch": 0.6392187326600821,
+      "grad_norm": 0.9148339629173279,
+      "learning_rate": 9.423768674021638e-06,
+      "loss": 0.238,
+      "mean_token_accuracy": 0.03264097062674409,
+      "step": 180
+    },
+    {
+      "epoch": 0.642769947841527,
+      "grad_norm": 1.171152114868164,
+      "learning_rate": 9.417220681714232e-06,
+      "loss": 0.227,
+      "mean_token_accuracy": 0.034916177624836564,
+      "step": 181
+    },
+    {
+      "epoch": 0.6463211630229719,
+      "grad_norm": 1.0754197835922241,
+      "learning_rate": 9.410637996768161e-06,
+      "loss": 0.2239,
+      "mean_token_accuracy": 0.036934319799911464,
+      "step": 182
+    },
+    {
+      "epoch": 0.6498723782044168,
+      "grad_norm": 1.3153674602508545,
+      "learning_rate": 9.404020670883511e-06,
+      "loss": 0.2346,
+      "mean_token_accuracy": 0.03758588138043706,
+      "step": 183
+    },
+    {
+      "epoch": 0.6534235933858618,
+      "grad_norm": 1.2649421691894531,
+      "learning_rate": 9.397368756032445e-06,
+      "loss": 0.2282,
+      "mean_token_accuracy": 0.03252386118401773,
+      "step": 184
+    },
+    {
+      "epoch": 0.6569748085673066,
+      "grad_norm": 1.4130192995071411,
+      "learning_rate": 9.390682304458782e-06,
+      "loss": 0.2199,
+      "mean_token_accuracy": 0.03749604838958476,
+      "step": 185
+    },
+    {
+      "epoch": 0.6605260237487516,
+      "grad_norm": 1.3785449266433716,
+      "learning_rate": 9.38396136867759e-06,
+      "loss": 0.2242,
+      "mean_token_accuracy": 0.037102947433595546,
+      "step": 186
+    },
+    {
+      "epoch": 0.6640772389301964,
+      "grad_norm": 1.6444662809371948,
+      "learning_rate": 9.377206001474773e-06,
+      "loss": 0.2333,
+      "mean_token_accuracy": 0.03256806965509895,
+      "step": 187
+    },
+    {
+      "epoch": 0.6676284541116413,
+      "grad_norm": 1.3745763301849365,
+      "learning_rate": 9.370416255906663e-06,
+      "loss": 0.2367,
+      "mean_token_accuracy": 0.034505165538575966,
+      "step": 188
+    },
+    {
+      "epoch": 0.6711796692930863,
+      "grad_norm": 0.9810280203819275,
+      "learning_rate": 9.363592185299593e-06,
+      "loss": 0.2341,
+      "mean_token_accuracy": 0.03995652306730335,
+      "step": 189
+    },
+    {
+      "epoch": 0.6747308844745311,
+      "grad_norm": 1.3432363271713257,
+      "learning_rate": 9.356733843249487e-06,
+      "loss": 0.2309,
+      "mean_token_accuracy": 0.030092548069660552,
+      "step": 190
+    },
+    {
+      "epoch": 0.6782820996559761,
+      "grad_norm": 2.273892641067505,
+      "learning_rate": 9.349841283621432e-06,
+      "loss": 0.241,
+      "mean_token_accuracy": 0.031039203957334394,
+      "step": 191
+    },
+    {
+      "epoch": 0.6818333148374209,
+      "grad_norm": 1.2541855573654175,
+      "learning_rate": 9.34291456054926e-06,
+      "loss": 0.2188,
+      "mean_token_accuracy": 0.039437778086721664,
+      "step": 192
+    },
+    {
+      "epoch": 0.6853845300188658,
+      "grad_norm": 1.4611351490020752,
+      "learning_rate": 9.33595372843512e-06,
+      "loss": 0.2245,
+      "mean_token_accuracy": 0.03319703893066617,
+      "step": 193
+    },
+    {
+      "epoch": 0.6889357452003108,
+      "grad_norm": 1.4521944522857666,
+      "learning_rate": 9.328958841949056e-06,
+      "loss": 0.2427,
+      "mean_token_accuracy": 0.03371610334215802,
+      "step": 194
+    },
+    {
+      "epoch": 0.6924869603817556,
+      "grad_norm": 1.4048359394073486,
+      "learning_rate": 9.321929956028565e-06,
+      "loss": 0.2379,
+      "mean_token_accuracy": 0.039065972556272754,
+      "step": 195
+    },
+    {
+      "epoch": 0.6960381755632006,
+      "grad_norm": 1.262329339981079,
+      "learning_rate": 9.31486712587818e-06,
+      "loss": 0.2395,
+      "mean_token_accuracy": 0.03516424228109827,
+      "step": 196
+    },
+    {
+      "epoch": 0.6995893907446454,
+      "grad_norm": 1.9205256700515747,
+      "learning_rate": 9.307770406969032e-06,
+      "loss": 0.235,
+      "mean_token_accuracy": 0.03766121077205753,
+      "step": 197
+    },
+    {
+      "epoch": 0.7031406059260903,
+      "grad_norm": 0.861643373966217,
+      "learning_rate": 9.300639855038405e-06,
+      "loss": 0.2333,
+      "mean_token_accuracy": 0.03458166824930231,
+      "step": 198
+    },
+    {
+      "epoch": 0.7066918211075353,
+      "grad_norm": 1.0324631929397583,
+      "learning_rate": 9.293475526089316e-06,
+      "loss": 0.2266,
+      "mean_token_accuracy": 0.0378209915907064,
+      "step": 199
+    },
+    {
+      "epoch": 0.7102430362889801,
+      "grad_norm": 1.0572841167449951,
+      "learning_rate": 9.286277476390056e-06,
+      "loss": 0.2228,
+      "mean_token_accuracy": 0.033504162300232565,
+      "step": 200
+    },
+    {
+      "epoch": 0.7137942514704251,
+      "grad_norm": 0.977691113948822,
+      "learning_rate": 9.279045762473764e-06,
+      "loss": 0.2319,
+      "mean_token_accuracy": 0.03699978294389439,
+      "step": 201
+    },
+    {
+      "epoch": 0.7173454666518699,
+      "grad_norm": 1.1246163845062256,
+      "learning_rate": 9.27178044113797e-06,
+      "loss": 0.2289,
+      "mean_token_accuracy": 0.03518677899228351,
+      "step": 202
+    },
+    {
+      "epoch": 0.7208966818333148,
+      "grad_norm": 1.03395676612854,
+      "learning_rate": 9.264481569444157e-06,
+      "loss": 0.2276,
+      "mean_token_accuracy": 0.03365413462961442,
+      "step": 203
+    },
+    {
+      "epoch": 0.7244478970147598,
+      "grad_norm": 1.1664369106292725,
+      "learning_rate": 9.257149204717317e-06,
+      "loss": 0.2322,
+      "mean_token_accuracy": 0.03404544836485002,
+      "step": 204
+    },
+    {
+      "epoch": 0.7279991121962046,
+      "grad_norm": 1.9012341499328613,
+      "learning_rate": 9.249783404545488e-06,
+      "loss": 0.2295,
+      "mean_token_accuracy": 0.0391799282260763,
+      "step": 205
+    },
+    {
+      "epoch": 0.7315503273776496,
+      "grad_norm": 1.7467440366744995,
+      "learning_rate": 9.242384226779308e-06,
+      "loss": 0.2284,
+      "mean_token_accuracy": 0.03427470392125542,
+      "step": 206
+    },
+    {
+      "epoch": 0.7351015425590944,
+      "grad_norm": 1.0553667545318604,
+      "learning_rate": 9.234951729531564e-06,
+      "loss": 0.2225,
+      "mean_token_accuracy": 0.039189494265883695,
+      "step": 207
+    },
+    {
+      "epoch": 0.7386527577405393,
+      "grad_norm": 1.2300719022750854,
+      "learning_rate": 9.227485971176734e-06,
+      "loss": 0.2304,
+      "mean_token_accuracy": 0.04073228745619417,
+      "step": 208
+    },
+    {
+      "epoch": 0.7422039729219843,
+      "grad_norm": 1.1945102214813232,
+      "learning_rate": 9.219987010350522e-06,
+      "loss": 0.221,
+      "mean_token_accuracy": 0.038925422908505425,
+      "step": 209
+    },
+    {
+      "epoch": 0.7457551881034291,
+      "grad_norm": 0.7722426056861877,
+      "learning_rate": 9.212454905949406e-06,
+      "loss": 0.2224,
+      "mean_token_accuracy": 0.033903196033861605,
+      "step": 210
+    },
+    {
+      "epoch": 0.7493064032848741,
+      "grad_norm": 1.1328966617584229,
+      "learning_rate": 9.204889717130172e-06,
+      "loss": 0.2335,
+      "mean_token_accuracy": 0.03668181393004488,
+      "step": 211
+    },
+    {
+      "epoch": 0.7528576184663189,
+      "grad_norm": 1.0629628896713257,
+      "learning_rate": 9.197291503309448e-06,
+      "loss": 0.2243,
+      "mean_token_accuracy": 0.03883554994536098,
+      "step": 212
+    },
+    {
+      "epoch": 0.7564088336477638,
+      "grad_norm": 1.6294198036193848,
+      "learning_rate": 9.189660324163243e-06,
+      "loss": 0.2295,
+      "mean_token_accuracy": 0.03799285244895145,
+      "step": 213
+    },
+    {
+      "epoch": 0.7599600488292088,
+      "grad_norm": 0.9800763726234436,
+      "learning_rate": 9.181996239626468e-06,
+      "loss": 0.218,
+      "mean_token_accuracy": 0.03557873377940268,
+      "step": 214
+    },
+    {
+      "epoch": 0.7635112640106536,
+      "grad_norm": 1.125218391418457,
+      "learning_rate": 9.174299309892474e-06,
+      "loss": 0.2264,
+      "mean_token_accuracy": 0.037647852554073324,
+      "step": 215
+    },
+    {
+      "epoch": 0.7670624791920986,
+      "grad_norm": 1.0833550691604614,
+      "learning_rate": 9.166569595412576e-06,
+      "loss": 0.2169,
+      "mean_token_accuracy": 0.03527776500050095,
+      "step": 216
+    },
+    {
+      "epoch": 0.7706136943735434,
+      "grad_norm": 1.386415719985962,
+      "learning_rate": 9.158807156895581e-06,
+      "loss": 0.223,
+      "mean_token_accuracy": 0.03334747435292229,
+      "step": 217
+    },
+    {
+      "epoch": 0.7741649095549884,
+      "grad_norm": 1.3321865797042847,
+      "learning_rate": 9.151012055307308e-06,
+      "loss": 0.2239,
+      "mean_token_accuracy": 0.036874980345601216,
+      "step": 218
+    },
+    {
+      "epoch": 0.7777161247364333,
+      "grad_norm": 1.1025806665420532,
+      "learning_rate": 9.14318435187011e-06,
+      "loss": 0.2261,
+      "mean_token_accuracy": 0.03307490728911944,
+      "step": 219
+    },
+    {
+      "epoch": 0.7812673399178781,
+      "grad_norm": 1.6245715618133545,
+      "learning_rate": 9.135324108062391e-06,
+      "loss": 0.2085,
+      "mean_token_accuracy": 0.04187627647843328,
+      "step": 220
+    },
+    {
+      "epoch": 0.7848185550993231,
+      "grad_norm": 1.4246594905853271,
+      "learning_rate": 9.127431385618129e-06,
+      "loss": 0.2189,
+      "mean_token_accuracy": 0.040660028342244914,
+      "step": 221
+    },
+    {
+      "epoch": 0.7883697702807679,
+      "grad_norm": 1.2004705667495728,
+      "learning_rate": 9.119506246526386e-06,
+      "loss": 0.2248,
+      "mean_token_accuracy": 0.034995593878193176,
+      "step": 222
+    },
+    {
+      "epoch": 0.7919209854622129,
+      "grad_norm": 1.6502361297607422,
+      "learning_rate": 9.111548753030824e-06,
+      "loss": 0.223,
+      "mean_token_accuracy": 0.03552522065365338,
+      "step": 223
+    },
+    {
+      "epoch": 0.7954722006436578,
+      "grad_norm": 1.5170484781265259,
+      "learning_rate": 9.103558967629211e-06,
+      "loss": 0.2267,
+      "mean_token_accuracy": 0.037829035503818886,
+      "step": 224
+    },
+    {
+      "epoch": 0.7990234158251026,
+      "grad_norm": 1.460752010345459,
+      "learning_rate": 9.09553695307294e-06,
+      "loss": 0.234,
+      "mean_token_accuracy": 0.028729848068905994,
+      "step": 225
+    },
+    {
+      "epoch": 0.8025746310065476,
+      "grad_norm": 1.1285916566848755,
+      "learning_rate": 9.087482772366529e-06,
+      "loss": 0.2304,
+      "mean_token_accuracy": 0.033455276032327674,
+      "step": 226
+    },
+    {
+      "epoch": 0.8061258461879924,
+      "grad_norm": 0.9813740849494934,
+      "learning_rate": 9.07939648876712e-06,
+      "loss": 0.2199,
+      "mean_token_accuracy": 0.03443802075344138,
+      "step": 227
+    },
+    {
+      "epoch": 0.8096770613694374,
+      "grad_norm": 0.9483611583709717,
+      "learning_rate": 9.071278165784001e-06,
+      "loss": 0.2283,
+      "mean_token_accuracy": 0.03545017533178907,
+      "step": 228
+    },
+    {
+      "epoch": 0.8132282765508823,
+      "grad_norm": 1.1032304763793945,
+      "learning_rate": 9.063127867178085e-06,
+      "loss": 0.2182,
+      "mean_token_accuracy": 0.04028806747373892,
+      "step": 229
+    },
+    {
+      "epoch": 0.8167794917323271,
+      "grad_norm": 2.941054344177246,
+      "learning_rate": 9.054945656961429e-06,
+      "loss": 0.2166,
+      "mean_token_accuracy": 0.04179423921959824,
+      "step": 230
+    },
+    {
+      "epoch": 0.8203307069137721,
+      "grad_norm": 1.0875635147094727,
+      "learning_rate": 9.046731599396716e-06,
+      "loss": 0.2295,
+      "mean_token_accuracy": 0.034964146889251424,
+      "step": 231
+    },
+    {
+      "epoch": 0.8238819220952169,
+      "grad_norm": 1.1555942296981812,
+      "learning_rate": 9.03848575899676e-06,
+      "loss": 0.2269,
+      "mean_token_accuracy": 0.04241397403529845,
+      "step": 232
+    },
+    {
+      "epoch": 0.8274331372766619,
+      "grad_norm": 0.948806881904602,
+      "learning_rate": 9.030208200523994e-06,
+      "loss": 0.2202,
+      "mean_token_accuracy": 0.03350217042680015,
+      "step": 233
+    },
+    {
+      "epoch": 0.8309843524581068,
+      "grad_norm": 1.2038196325302124,
+      "learning_rate": 9.021898988989966e-06,
+      "loss": 0.2247,
+      "mean_token_accuracy": 0.03134763890193426,
+      "step": 234
+    },
+    {
+      "epoch": 0.8345355676395516,
+      "grad_norm": 1.6858941316604614,
+      "learning_rate": 9.013558189654819e-06,
+      "loss": 0.218,
+      "mean_token_accuracy": 0.03856555292077246,
+      "step": 235
+    },
+    {
+      "epoch": 0.8380867828209966,
+      "grad_norm": 1.1609561443328857,
+      "learning_rate": 9.005185868026793e-06,
+      "loss": 0.2169,
+      "mean_token_accuracy": 0.036304176566773094,
+      "step": 236
+    },
+    {
+      "epoch": 0.8416379980024414,
+      "grad_norm": 1.1071784496307373,
+      "learning_rate": 8.996782089861699e-06,
+      "loss": 0.2419,
+      "mean_token_accuracy": 0.032351685425965115,
+      "step": 237
+    },
+    {
+      "epoch": 0.8451892131838864,
+      "grad_norm": 1.0100809335708618,
+      "learning_rate": 8.988346921162407e-06,
+      "loss": 0.2319,
+      "mean_token_accuracy": 0.03175320574155194,
+      "step": 238
+    },
+    {
+      "epoch": 0.8487404283653313,
+      "grad_norm": 1.4512320756912231,
+      "learning_rate": 8.979880428178323e-06,
+      "loss": 0.2249,
+      "mean_token_accuracy": 0.03225310727066244,
+      "step": 239
+    },
+    {
+      "epoch": 0.8522916435467761,
+      "grad_norm": 1.0108217000961304,
+      "learning_rate": 8.971382677404878e-06,
+      "loss": 0.233,
+      "mean_token_accuracy": 0.03236227970774053,
+      "step": 240
+    },
+    {
+      "epoch": 0.8558428587282211,
+      "grad_norm": 1.3838858604431152,
+      "learning_rate": 8.962853735582996e-06,
+      "loss": 0.2155,
+      "mean_token_accuracy": 0.036608970935049,
+      "step": 241
+    },
+    {
+      "epoch": 0.8593940739096659,
+      "grad_norm": 1.138848900794983,
+      "learning_rate": 8.95429366969858e-06,
+      "loss": 0.2345,
+      "mean_token_accuracy": 0.0361562248053815,
+      "step": 242
+    },
+    {
+      "epoch": 0.8629452890911109,
+      "grad_norm": 0.8619561791419983,
+      "learning_rate": 8.94570254698197e-06,
+      "loss": 0.2309,
+      "mean_token_accuracy": 0.04240272083552554,
+      "step": 243
+    },
+    {
+      "epoch": 0.8664965042725558,
+      "grad_norm": 1.1279878616333008,
+      "learning_rate": 8.93708043490743e-06,
+      "loss": 0.2304,
+      "mean_token_accuracy": 0.03780577134239138,
+      "step": 244
+    },
+    {
+      "epoch": 0.8700477194540006,
+      "grad_norm": 1.1739652156829834,
+      "learning_rate": 8.928427401192618e-06,
+      "loss": 0.2265,
+      "mean_token_accuracy": 0.037726832346379524,
+      "step": 245
+    },
+    {
+      "epoch": 0.8735989346354456,
+      "grad_norm": 0.7752535343170166,
+      "learning_rate": 8.919743513798044e-06,
+      "loss": 0.2236,
+      "mean_token_accuracy": 0.03597815223474754,
+      "step": 246
+    },
+    {
+      "epoch": 0.8771501498168904,
+      "grad_norm": 0.7733471989631653,
+      "learning_rate": 8.911028840926537e-06,
+      "loss": 0.2221,
+      "mean_token_accuracy": 0.04170313518625335,
+      "step": 247
+    },
+    {
+      "epoch": 0.8807013649983354,
+      "grad_norm": 0.7916595935821533,
+      "learning_rate": 8.902283451022725e-06,
+      "loss": 0.2138,
+      "mean_token_accuracy": 0.03986302729026647,
+      "step": 248
+    },
+    {
+      "epoch": 0.8842525801797803,
+      "grad_norm": 1.2641276121139526,
+      "learning_rate": 8.89350741277247e-06,
+      "loss": 0.2362,
+      "mean_token_accuracy": 0.03598766483992222,
+      "step": 249
+    },
+    {
+      "epoch": 0.8878037953612252,
+      "grad_norm": 1.6378071308135986,
+      "learning_rate": 8.884700795102365e-06,
+      "loss": 0.2299,
+      "mean_token_accuracy": 0.03333858217229135,
+      "step": 250
+    },
+    {
+      "epoch": 0.8913550105426701,
+      "grad_norm": 1.092628836631775,
+      "learning_rate": 8.875863667179155e-06,
+      "loss": 0.2276,
+      "mean_token_accuracy": 0.03222806086341734,
+      "step": 251
+    },
+    {
+      "epoch": 0.8949062257241149,
+      "grad_norm": 1.4616332054138184,
+      "learning_rate": 8.866996098409217e-06,
+      "loss": 0.2212,
+      "mean_token_accuracy": 0.0341962569837051,
+      "step": 252
+    },
+    {
+      "epoch": 0.8984574409055599,
+      "grad_norm": 1.8922557830810547,
+      "learning_rate": 8.858098158438013e-06,
+      "loss": 0.2308,
+      "mean_token_accuracy": 0.03757270462301676,
+      "step": 253
+    },
+    {
+      "epoch": 0.9020086560870048,
+      "grad_norm": 1.3642209768295288,
+      "learning_rate": 8.849169917149532e-06,
+      "loss": 0.2239,
+      "mean_token_accuracy": 0.03927645156363724,
+      "step": 254
+    },
+    {
+      "epoch": 0.9055598712684497,
+      "grad_norm": 1.0666165351867676,
+      "learning_rate": 8.840211444665754e-06,
+      "loss": 0.2302,
+      "mean_token_accuracy": 0.0400322419773147,
+      "step": 255
+    },
+    {
+      "epoch": 0.9091110864498946,
+      "grad_norm": 1.0763819217681885,
+      "learning_rate": 8.831222811346088e-06,
+      "loss": 0.2227,
+      "mean_token_accuracy": 0.038276460734778084,
+      "step": 256
+    },
+    {
+      "epoch": 0.9126623016313394,
+      "grad_norm": 1.3471585512161255,
+      "learning_rate": 8.822204087786831e-06,
+      "loss": 0.2316,
+      "mean_token_accuracy": 0.03097587551746983,
+      "step": 257
+    },
+    {
+      "epoch": 0.9162135168127844,
+      "grad_norm": 1.2004120349884033,
+      "learning_rate": 8.813155344820602e-06,
+      "loss": 0.2179,
+      "mean_token_accuracy": 0.03698654064646689,
+      "step": 258
+    },
+    {
+      "epoch": 0.9197647319942293,
+      "grad_norm": 1.486273169517517,
+      "learning_rate": 8.804076653515792e-06,
+      "loss": 0.2206,
+      "mean_token_accuracy": 0.03723590575464186,
+      "step": 259
+    },
+    {
+      "epoch": 0.9233159471756742,
+      "grad_norm": 0.8779009580612183,
+      "learning_rate": 8.794968085176006e-06,
+      "loss": 0.2253,
+      "mean_token_accuracy": 0.03663652328577882,
+      "step": 260
+    },
+    {
+      "epoch": 0.9268671623571191,
+      "grad_norm": 0.961213231086731,
+      "learning_rate": 8.785829711339502e-06,
+      "loss": 0.2193,
+      "mean_token_accuracy": 0.03452028890387737,
+      "step": 261
+    },
+    {
+      "epoch": 0.9304183775385639,
+      "grad_norm": 1.5822545289993286,
+      "learning_rate": 8.776661603778629e-06,
+      "loss": 0.2283,
+      "mean_token_accuracy": 0.03859327645113808,
+      "step": 262
+    },
+    {
+      "epoch": 0.9339695927200089,
+      "grad_norm": 1.3098887205123901,
+      "learning_rate": 8.767463834499261e-06,
+      "loss": 0.2355,
+      "mean_token_accuracy": 0.03154927755895187,
+      "step": 263
+    },
+    {
+      "epoch": 0.9375208079014538,
+      "grad_norm": 1.379447102546692,
+      "learning_rate": 8.758236475740236e-06,
+      "loss": 0.2152,
+      "mean_token_accuracy": 0.03622293883745442,
+      "step": 264
+    },
+    {
+      "epoch": 0.9410720230828987,
+      "grad_norm": 1.2279735803604126,
+      "learning_rate": 8.748979599972787e-06,
+      "loss": 0.2245,
+      "mean_token_accuracy": 0.033197632066730876,
+      "step": 265
+    },
+    {
+      "epoch": 0.9446232382643436,
+      "grad_norm": 1.2326911687850952,
+      "learning_rate": 8.739693279899969e-06,
+      "loss": 0.2176,
+      "mean_token_accuracy": 0.03476043541013496,
+      "step": 266
+    },
+    {
+      "epoch": 0.9481744534457884,
+      "grad_norm": 0.9703882336616516,
+      "learning_rate": 8.730377588456092e-06,
+      "loss": 0.2202,
+      "mean_token_accuracy": 0.0382119021815015,
+      "step": 267
+    },
+    {
+      "epoch": 0.9517256686272334,
+      "grad_norm": 1.0774365663528442,
+      "learning_rate": 8.72103259880615e-06,
+      "loss": 0.2243,
+      "mean_token_accuracy": 0.03840548314838088,
+      "step": 268
+    },
+    {
+      "epoch": 0.9552768838086783,
+      "grad_norm": 1.117008924484253,
+      "learning_rate": 8.711658384345244e-06,
+      "loss": 0.225,
+      "mean_token_accuracy": 0.035315932909725234,
+      "step": 269
+    },
+    {
+      "epoch": 0.9588280989901232,
+      "grad_norm": 1.1434279680252075,
+      "learning_rate": 8.702255018698e-06,
+      "loss": 0.21,
+      "mean_token_accuracy": 0.03513298414691235,
+      "step": 270
+    },
+    {
+      "epoch": 0.9623793141715681,
+      "grad_norm": 0.7816430330276489,
+      "learning_rate": 8.692822575718e-06,
+      "loss": 0.2157,
+      "mean_token_accuracy": 0.03654791028384352,
+      "step": 271
+    },
+    {
+      "epoch": 0.9659305293530129,
+      "grad_norm": 1.0198558568954468,
+      "learning_rate": 8.683361129487198e-06,
+      "loss": 0.2165,
+      "mean_token_accuracy": 0.04054846076542162,
+      "step": 272
+    },
+    {
+      "epoch": 0.9694817445344579,
+      "grad_norm": 1.140480637550354,
+      "learning_rate": 8.673870754315336e-06,
+      "loss": 0.2168,
+      "mean_token_accuracy": 0.04233230032332358,
+      "step": 273
+    },
+    {
+      "epoch": 0.9730329597159028,
+      "grad_norm": 0.9938759803771973,
+      "learning_rate": 8.664351524739368e-06,
+      "loss": 0.2304,
+      "mean_token_accuracy": 0.03541681009664899,
+      "step": 274
+    },
+    {
+      "epoch": 0.9765841748973477,
+      "grad_norm": 0.8925390243530273,
+      "learning_rate": 8.65480351552286e-06,
+      "loss": 0.2295,
+      "mean_token_accuracy": 0.03779280128037499,
+      "step": 275
+    },
+    {
+      "epoch": 0.9801353900787926,
+      "grad_norm": 1.2885398864746094,
+      "learning_rate": 8.645226801655418e-06,
+      "loss": 0.2223,
+      "mean_token_accuracy": 0.03629426244151546,
+      "step": 276
+    },
+    {
+      "epoch": 0.9836866052602374,
+      "grad_norm": 0.9392145872116089,
+      "learning_rate": 8.635621458352094e-06,
+      "loss": 0.2313,
+      "mean_token_accuracy": 0.0358071523696708,
+      "step": 277
+    },
+    {
+      "epoch": 0.9872378204416824,
+      "grad_norm": 0.9267784357070923,
+      "learning_rate": 8.625987561052789e-06,
+      "loss": 0.2288,
+      "mean_token_accuracy": 0.034536936505901394,
+      "step": 278
+    },
+    {
+      "epoch": 0.9907890356231273,
+      "grad_norm": 1.0564697980880737,
+      "learning_rate": 8.616325185421673e-06,
+      "loss": 0.2313,
+      "mean_token_accuracy": 0.032076333489385433,
+      "step": 279
+    },
+    {
+      "epoch": 0.9943402508045722,
+      "grad_norm": 1.2704871892929077,
+      "learning_rate": 8.606634407346575e-06,
+      "loss": 0.2254,
+      "mean_token_accuracy": 0.03875074074676377,
+      "step": 280
+    },
+    {
+      "epoch": 0.9978914659860171,
+      "grad_norm": 0.9733898043632507,
+      "learning_rate": 8.596915302938403e-06,
+      "loss": 0.2252,
+      "mean_token_accuracy": 0.032296190700435545,
+      "step": 281
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6648575663566589,
+      "learning_rate": 8.587167948530533e-06,
+      "loss": 0.1356,
+      "mean_token_accuracy": 0.03019784249083482,
+      "step": 282
+    },
+    {
+      "epoch": 1.003551215181445,
+      "grad_norm": 1.3055585622787476,
+      "learning_rate": 8.577392420678217e-06,
+      "loss": 0.2124,
+      "mean_token_accuracy": 0.03929255892217043,
+      "step": 283
+    },
+    {
+      "epoch": 1.0071024303628897,
+      "grad_norm": 1.0825082063674927,
+      "learning_rate": 8.567588796157983e-06,
+      "loss": 0.2265,
+      "mean_token_accuracy": 0.03449270429700846,
+      "step": 284
+    },
+    {
+      "epoch": 1.0106536455443347,
+      "grad_norm": 0.9831315279006958,
+      "learning_rate": 8.557757151967025e-06,
+      "loss": 0.2257,
+      "mean_token_accuracy": 0.03714761855371762,
+      "step": 285
+    },
+    {
+      "epoch": 1.0142048607257796,
+      "grad_norm": 0.7693557739257812,
+      "learning_rate": 8.547897565322601e-06,
+      "loss": 0.2208,
+      "mean_token_accuracy": 0.039464723708078964,
+      "step": 286
+    },
+    {
+      "epoch": 1.0177560759072246,
+      "grad_norm": 0.9719224572181702,
+      "learning_rate": 8.538010113661434e-06,
+      "loss": 0.215,
+      "mean_token_accuracy": 0.03782110561223817,
+      "step": 287
+    },
+    {
+      "epoch": 1.0213072910886694,
+      "grad_norm": 1.2199909687042236,
+      "learning_rate": 8.528094874639092e-06,
+      "loss": 0.2298,
+      "mean_token_accuracy": 0.03888447284771246,
+      "step": 288
+    },
+    {
+      "epoch": 1.0248585062701143,
+      "grad_norm": 1.0379016399383545,
+      "learning_rate": 8.518151926129384e-06,
+      "loss": 0.2048,
+      "mean_token_accuracy": 0.04360219682166644,
+      "step": 289
+    },
+    {
+      "epoch": 1.0284097214515593,
+      "grad_norm": 0.8994417786598206,
+      "learning_rate": 8.508181346223749e-06,
+      "loss": 0.227,
+      "mean_token_accuracy": 0.03488920487325231,
+      "step": 290
+    },
+    {
+      "epoch": 1.031960936633004,
+      "grad_norm": 0.9707528948783875,
+      "learning_rate": 8.498183213230646e-06,
+      "loss": 0.2199,
+      "mean_token_accuracy": 0.04115383749740431,
+      "step": 291
+    },
+    {
+      "epoch": 1.035512151814449,
+      "grad_norm": 1.017004370689392,
+      "learning_rate": 8.488157605674924e-06,
+      "loss": 0.217,
+      "mean_token_accuracy": 0.035543966083423584,
+      "step": 292
+    },
+    {
+      "epoch": 1.039063366995894,
+      "grad_norm": 1.0191696882247925,
+      "learning_rate": 8.478104602297226e-06,
+      "loss": 0.2168,
+      "mean_token_accuracy": 0.03651970579812769,
+      "step": 293
+    },
+    {
+      "epoch": 1.0426145821773387,
+      "grad_norm": 1.1888155937194824,
+      "learning_rate": 8.468024282053357e-06,
+      "loss": 0.2232,
+      "mean_token_accuracy": 0.03674552895972738,
+      "step": 294
+    },
+    {
+      "epoch": 1.0461657973587837,
+      "grad_norm": 1.1544668674468994,
+      "learning_rate": 8.457916724113667e-06,
+      "loss": 0.2193,
+      "mean_token_accuracy": 0.035183957385015674,
+      "step": 295
+    },
+    {
+      "epoch": 1.0497170125402286,
+      "grad_norm": 1.1487054824829102,
+      "learning_rate": 8.447782007862427e-06,
+      "loss": 0.2127,
+      "mean_token_accuracy": 0.03971767070652277,
+      "step": 296
+    },
+    {
+      "epoch": 1.0532682277216736,
+      "grad_norm": 1.1387826204299927,
+      "learning_rate": 8.437620212897213e-06,
+      "loss": 0.2139,
+      "mean_token_accuracy": 0.03873823723915848,
+      "step": 297
+    },
+    {
+      "epoch": 1.0568194429031184,
+      "grad_norm": 1.2510346174240112,
+      "learning_rate": 8.427431419028273e-06,
+      "loss": 0.2265,
+      "mean_token_accuracy": 0.03466822681366466,
+      "step": 298
+    },
+    {
+      "epoch": 1.0603706580845633,
+      "grad_norm": 1.0624276399612427,
+      "learning_rate": 8.417215706277905e-06,
+      "loss": 0.2138,
+      "mean_token_accuracy": 0.0331218589690252,
+      "step": 299
+    },
+    {
+      "epoch": 1.0639218732660083,
+      "grad_norm": 1.1180115938186646,
+      "learning_rate": 8.406973154879826e-06,
+      "loss": 0.2237,
+      "mean_token_accuracy": 0.03226224229001673,
+      "step": 300
+    },
+    {
+      "epoch": 1.071024303628898,
+      "grad_norm": 0.9965167045593262,
+      "learning_rate": 8.396703845278537e-06,
+      "loss": 0.221,
+      "mean_token_accuracy": 0.03691619080200326,
+      "step": 301
+    },
+    {
+      "epoch": 1.074575518810343,
+      "grad_norm": 1.0591256618499756,
+      "learning_rate": 8.386407858128707e-06,
+      "loss": 0.2196,
+      "mean_token_accuracy": 0.03875752529347665,
+      "step": 302
+    },
+    {
+      "epoch": 1.0781267339917877,
+      "grad_norm": 1.100635051727295,
+      "learning_rate": 8.376085274294518e-06,
+      "loss": 0.2219,
+      "mean_token_accuracy": 0.037956459491397254,
+      "step": 303
+    },
+    {
+      "epoch": 1.0816779491732327,
+      "grad_norm": 0.9553704261779785,
+      "learning_rate": 8.365736174849053e-06,
+      "loss": 0.2171,
+      "mean_token_accuracy": 0.03450688702832849,
+      "step": 304
+    },
+    {
+      "epoch": 1.0852291643546776,
+      "grad_norm": 0.9865755438804626,
+      "learning_rate": 8.355360641073637e-06,
+      "loss": 0.2168,
+      "mean_token_accuracy": 0.036030067367391894,
+      "step": 305
+    },
+    {
+      "epoch": 1.0887803795361226,
+      "grad_norm": 1.0858834981918335,
+      "learning_rate": 8.344958754457214e-06,
+      "loss": 0.2256,
+      "mean_token_accuracy": 0.03380810684393509,
+      "step": 306
+    },
+    {
+      "epoch": 1.0923315947175674,
+      "grad_norm": 1.026892066001892,
+      "learning_rate": 8.3345305966957e-06,
+      "loss": 0.222,
+      "mean_token_accuracy": 0.03506634270888753,
+      "step": 307
+    },
+    {
+      "epoch": 1.0958828098990123,
+      "grad_norm": 0.98395836353302,
+      "learning_rate": 8.324076249691347e-06,
+      "loss": 0.2131,
+      "mean_token_accuracy": 0.037066298340505455,
+      "step": 308
+    },
+    {
+      "epoch": 1.0994340250804573,
+      "grad_norm": 0.8688275814056396,
+      "learning_rate": 8.31359579555209e-06,
+      "loss": 0.2087,
+      "mean_token_accuracy": 0.035992500863358146,
+      "step": 309
+    },
+    {
+      "epoch": 1.102985240261902,
+      "grad_norm": 7.995417594909668,
+      "learning_rate": 8.30308931659091e-06,
+      "loss": 0.2316,
+      "mean_token_accuracy": 0.03655194863677025,
+      "step": 310
+    },
+    {
+      "epoch": 1.106536455443347,
+      "grad_norm": 1.0539637804031372,
+      "learning_rate": 8.292556895325195e-06,
+      "loss": 0.227,
+      "mean_token_accuracy": 0.03512826969381422,
+      "step": 311
+    },
+    {
+      "epoch": 1.110087670624792,
+      "grad_norm": 0.8968519568443298,
+      "learning_rate": 8.281998614476066e-06,
+      "loss": 0.226,
+      "mean_token_accuracy": 0.037603993361699395,
+      "step": 312
+    },
+    {
+      "epoch": 1.1136388858062367,
+      "grad_norm": 0.9943398237228394,
+      "learning_rate": 8.271414556967758e-06,
+      "loss": 0.2156,
+      "mean_token_accuracy": 0.03420978486246895,
+      "step": 313
+    },
+    {
+      "epoch": 1.1171901009876817,
+      "grad_norm": 1.1365629434585571,
+      "learning_rate": 8.260804805926948e-06,
+      "loss": 0.2233,
+      "mean_token_accuracy": 0.03783213975839317,
+      "step": 314
+    },
+    {
+      "epoch": 1.1207413161691266,
+      "grad_norm": 1.178529977798462,
+      "learning_rate": 8.250169444682109e-06,
+      "loss": 0.2198,
+      "mean_token_accuracy": 0.03976841686562693,
+      "step": 315
+    },
+    {
+      "epoch": 1.1242925313505716,
+      "grad_norm": 1.0319217443466187,
+      "learning_rate": 8.239508556762857e-06,
+      "loss": 0.2241,
+      "mean_token_accuracy": 0.031951177821611054,
+      "step": 316
+    },
+    {
+      "epoch": 1.1278437465320164,
+      "grad_norm": 1.2369407415390015,
+      "learning_rate": 8.228822225899294e-06,
+      "loss": 0.212,
+      "mean_token_accuracy": 0.04304049965503509,
+      "step": 317
+    },
+    {
+      "epoch": 1.1313949617134613,
+      "grad_norm": 1.0376719236373901,
+      "learning_rate": 8.218110536021347e-06,
+      "loss": 0.2085,
+      "mean_token_accuracy": 0.040561975689342944,
+      "step": 318
+    },
+    {
+      "epoch": 1.1349461768949063,
+      "grad_norm": 1.4454623460769653,
+      "learning_rate": 8.207373571258113e-06,
+      "loss": 0.2171,
+      "mean_token_accuracy": 0.03455675131408498,
+      "step": 319
+    },
+    {
+      "epoch": 1.138497392076351,
+      "grad_norm": 1.2810287475585938,
+      "learning_rate": 8.196611415937196e-06,
+      "loss": 0.2207,
+      "mean_token_accuracy": 0.03659132699976908,
+      "step": 320
+    },
+    {
+      "epoch": 1.142048607257796,
+      "grad_norm": 0.9189555048942566,
+      "learning_rate": 8.18582415458405e-06,
+      "loss": 0.2257,
+      "mean_token_accuracy": 0.03751322147945757,
+      "step": 321
+    },
+    {
+      "epoch": 1.145599822439241,
+      "grad_norm": 1.264859914779663,
+      "learning_rate": 8.1750118719213e-06,
+      "loss": 0.2195,
+      "mean_token_accuracy": 0.037348673798987875,
+      "step": 322
+    },
+    {
+      "epoch": 1.149151037620686,
+      "grad_norm": 0.9888535737991333,
+      "learning_rate": 8.164174652868097e-06,
+      "loss": 0.2255,
+      "mean_token_accuracy": 0.03571068402743549,
+      "step": 323
+    },
+    {
+      "epoch": 1.1527022528021307,
+      "grad_norm": 2.5238475799560547,
+      "learning_rate": 8.153312582539438e-06,
+      "loss": 0.2119,
+      "mean_token_accuracy": 0.038603254355621175,
+      "step": 324
+    },
+    {
+      "epoch": 1.1562534679835756,
+      "grad_norm": 0.9326993227005005,
+      "learning_rate": 8.142425746245503e-06,
+      "loss": 0.2127,
+      "mean_token_accuracy": 0.03710804141519475,
+      "step": 325
+    },
+    {
+      "epoch": 1.1598046831650206,
+      "grad_norm": 0.9224873185157776,
+      "learning_rate": 8.131514229490975e-06,
+      "loss": 0.2231,
+      "mean_token_accuracy": 0.041035250716959126,
+      "step": 326
+    },
+    {
+      "epoch": 1.1633558983464654,
+      "grad_norm": 1.1489871740341187,
+      "learning_rate": 8.120578117974388e-06,
+      "loss": 0.207,
+      "mean_token_accuracy": 0.035616049925010884,
+      "step": 327
+    },
+    {
+      "epoch": 1.1669071135279103,
+      "grad_norm": 1.3951830863952637,
+      "learning_rate": 8.109617497587429e-06,
+      "loss": 0.2138,
+      "mean_token_accuracy": 0.03988678594760131,
+      "step": 328
+    },
+    {
+      "epoch": 1.1704583287093553,
+      "grad_norm": 0.8216147422790527,
+      "learning_rate": 8.098632454414286e-06,
+      "loss": 0.2159,
+      "mean_token_accuracy": 0.03136701822586474,
+      "step": 329
+    },
+    {
+      "epoch": 1.1740095438908003,
+      "grad_norm": 0.9959815144538879,
+      "learning_rate": 8.08762307473096e-06,
+      "loss": 0.2064,
+      "mean_token_accuracy": 0.04014438956437516,
+      "step": 330
+    },
+    {
+      "epoch": 1.177560759072245,
+      "grad_norm": 1.0166196823120117,
+      "learning_rate": 8.07658944500459e-06,
+      "loss": 0.2289,
+      "mean_token_accuracy": 0.03964185667791753,
+      "step": 331
+    },
+    {
+      "epoch": 1.18111197425369,
+      "grad_norm": 1.4372248649597168,
+      "learning_rate": 8.065531651892771e-06,
+      "loss": 0.2184,
+      "mean_token_accuracy": 0.03303642533865059,
+      "step": 332
+    },
+    {
+      "epoch": 1.1846631894351347,
+      "grad_norm": 1.0342096090316772,
+      "learning_rate": 8.054449782242876e-06,
+      "loss": 0.2117,
+      "mean_token_accuracy": 0.03587262858854956,
+      "step": 333
+    },
+    {
+      "epoch": 1.1882144046165797,
+      "grad_norm": 1.0142632722854614,
+      "learning_rate": 8.043343923091382e-06,
+      "loss": 0.2144,
+      "mean_token_accuracy": 0.03768015895911958,
+      "step": 334
+    },
+    {
+      "epoch": 1.1917656197980246,
+      "grad_norm": 1.0247020721435547,
+      "learning_rate": 8.03221416166317e-06,
+      "loss": 0.2094,
+      "mean_token_accuracy": 0.04144499241374433,
+      "step": 335
+    },
+    {
+      "epoch": 1.1953168349794696,
+      "grad_norm": 1.1292544603347778,
+      "learning_rate": 8.021060585370845e-06,
+      "loss": 0.2211,
+      "mean_token_accuracy": 0.03466017847676994,
+      "step": 336
+    },
+    {
+      "epoch": 1.1988680501609144,
+      "grad_norm": 1.0083832740783691,
+      "learning_rate": 8.009883281814066e-06,
+      "loss": 0.2099,
+      "mean_token_accuracy": 0.04081952280466794,
+      "step": 337
+    },
+    {
+      "epoch": 1.2024192653423593,
+      "grad_norm": 1.3860927820205688,
+      "learning_rate": 7.998682338778834e-06,
+      "loss": 0.2076,
+      "mean_token_accuracy": 0.034925906780699734,
+      "step": 338
+    },
+    {
+      "epoch": 1.2059704805238043,
+      "grad_norm": 0.8742907643318176,
+      "learning_rate": 7.987457844236817e-06,
+      "loss": 0.2178,
+      "mean_token_accuracy": 0.03645704350856249,
+      "step": 339
+    },
+    {
+      "epoch": 1.209521695705249,
+      "grad_norm": 1.479430913925171,
+      "learning_rate": 7.976209886344654e-06,
+      "loss": 0.2147,
+      "mean_token_accuracy": 0.03608882729895413,
+      "step": 340
+    },
+    {
+      "epoch": 1.213072910886694,
+      "grad_norm": 1.5018517971038818,
+      "learning_rate": 7.964938553443267e-06,
+      "loss": 0.2275,
+      "mean_token_accuracy": 0.03744373080553487,
+      "step": 341
+    },
+    {
+      "epoch": 1.216624126068139,
+      "grad_norm": 1.1072347164154053,
+      "learning_rate": 7.953643934057162e-06,
+      "loss": 0.2148,
+      "mean_token_accuracy": 0.038897628408449236,
+      "step": 342
+    },
+    {
+      "epoch": 1.220175341249584,
+      "grad_norm": 0.8880918622016907,
+      "learning_rate": 7.942326116893733e-06,
+      "loss": 0.2166,
+      "mean_token_accuracy": 0.03384126560558798,
+      "step": 343
+    },
+    {
+      "epoch": 1.2237265564310287,
+      "grad_norm": 1.4236310720443726,
+      "learning_rate": 7.930985190842576e-06,
+      "loss": 0.2247,
+      "mean_token_accuracy": 0.038816192594822496,
+      "step": 344
+    },
+    {
+      "epoch": 1.2272777716124736,
+      "grad_norm": 1.3533358573913574,
+      "learning_rate": 7.919621244974773e-06,
+      "loss": 0.2154,
+      "mean_token_accuracy": 0.03848464526163298,
+      "step": 345
+    },
+    {
+      "epoch": 1.2308289867939186,
+      "grad_norm": 1.5561343431472778,
+      "learning_rate": 7.908234368542214e-06,
+      "loss": 0.1965,
+      "mean_token_accuracy": 0.04135380086154328,
+      "step": 346
+    },
+    {
+      "epoch": 1.2343802019753634,
+      "grad_norm": 1.2853002548217773,
+      "learning_rate": 7.896824650976873e-06,
+      "loss": 0.2127,
+      "mean_token_accuracy": 0.04092289448453812,
+      "step": 347
+    },
+    {
+      "epoch": 1.2379314171568083,
+      "grad_norm": 1.1815927028656006,
+      "learning_rate": 7.885392181890126e-06,
+      "loss": 0.2138,
+      "mean_token_accuracy": 0.03702772545875632,
+      "step": 348
+    },
+    {
+      "epoch": 1.2414826323382533,
+      "grad_norm": 2.602296829223633,
+      "learning_rate": 7.873937051072037e-06,
+      "loss": 0.2198,
+      "mean_token_accuracy": 0.03560452723831986,
+      "step": 349
+    },
+    {
+      "epoch": 1.2450338475196983,
+      "grad_norm": 1.52498197555542,
+      "learning_rate": 7.862459348490645e-06,
+      "loss": 0.2185,
+      "mean_token_accuracy": 0.0324377975302923,
+      "step": 350
+    },
+    {
+      "epoch": 1.248585062701143,
+      "grad_norm": 1.2384191751480103,
+      "learning_rate": 7.85095916429128e-06,
+      "loss": 0.2171,
+      "mean_token_accuracy": 0.03811432257498382,
+      "step": 351
+    },
+    {
+      "epoch": 1.252136277882588,
+      "grad_norm": 1.357321858406067,
+      "learning_rate": 7.839436588795834e-06,
+      "loss": 0.2034,
+      "mean_token_accuracy": 0.04051168984005926,
+      "step": 352
+    },
+    {
+      "epoch": 1.2556874930640327,
+      "grad_norm": 1.1218904256820679,
+      "learning_rate": 7.82789171250206e-06,
+      "loss": 0.2218,
+      "mean_token_accuracy": 0.032163290336029604,
+      "step": 353
+    },
+    {
+      "epoch": 1.2592387082454777,
+      "grad_norm": 1.573211669921875,
+      "learning_rate": 7.816324626082864e-06,
+      "loss": 0.2227,
+      "mean_token_accuracy": 0.04222477459552465,
+      "step": 354
+    },
+    {
+      "epoch": 1.2627899234269226,
+      "grad_norm": 1.5673032999038696,
+      "learning_rate": 7.804735420385578e-06,
+      "loss": 0.2163,
+      "mean_token_accuracy": 0.035729966661165236,
+      "step": 355
+    },
+    {
+      "epoch": 1.2663411386083676,
+      "grad_norm": 3.8074378967285156,
+      "learning_rate": 7.793124186431271e-06,
+      "loss": 0.2082,
+      "mean_token_accuracy": 0.04122993563032651,
+      "step": 356
+    },
+    {
+      "epoch": 1.2698923537898126,
+      "grad_norm": 0.7542282938957214,
+      "learning_rate": 7.781491015414018e-06,
+      "loss": 0.2151,
+      "mean_token_accuracy": 0.03684526444703806,
+      "step": 357
+    },
+    {
+      "epoch": 1.2734435689712573,
+      "grad_norm": 1.1057549715042114,
+      "learning_rate": 7.769835998700182e-06,
+      "loss": 0.216,
+      "mean_token_accuracy": 0.03879151448927587,
+      "step": 358
+    },
+    {
+      "epoch": 1.2769947841527023,
+      "grad_norm": 1.0620713233947754,
+      "learning_rate": 7.758159227827701e-06,
+      "loss": 0.2275,
+      "mean_token_accuracy": 0.0345176875598554,
+      "step": 359
+    },
+    {
+      "epoch": 1.280545999334147,
+      "grad_norm": 0.9399253129959106,
+      "learning_rate": 7.746460794505375e-06,
+      "loss": 0.2181,
+      "mean_token_accuracy": 0.03365673908410827,
+      "step": 360
+    },
+    {
+      "epoch": 1.284097214515592,
+      "grad_norm": 0.9033060669898987,
+      "learning_rate": 7.734740790612137e-06,
+      "loss": 0.2096,
+      "mean_token_accuracy": 0.038896608664799714,
+      "step": 361
+    },
+    {
+      "epoch": 1.287648429697037,
+      "grad_norm": 1.473177433013916,
+      "learning_rate": 7.722999308196329e-06,
+      "loss": 0.2176,
+      "mean_token_accuracy": 0.036279349031246966,
+      "step": 362
+    },
+    {
+      "epoch": 1.291199644878482,
+      "grad_norm": 1.2070612907409668,
+      "learning_rate": 7.711236439474991e-06,
+      "loss": 0.2179,
+      "mean_token_accuracy": 0.04047916533454554,
+      "step": 363
+    },
+    {
+      "epoch": 1.2947508600599267,
+      "grad_norm": 1.7395224571228027,
+      "learning_rate": 7.69945227683313e-06,
+      "loss": 0.2305,
+      "mean_token_accuracy": 0.03877663587627467,
+      "step": 364
+    },
+    {
+      "epoch": 1.2983020752413716,
+      "grad_norm": 0.9686111211776733,
+      "learning_rate": 7.68764691282299e-06,
+      "loss": 0.2117,
+      "mean_token_accuracy": 0.04020940974805853,
+      "step": 365
+    },
+    {
+      "epoch": 1.3018532904228166,
+      "grad_norm": 5.303558349609375,
+      "learning_rate": 7.675820440163334e-06,
+      "loss": 0.214,
+      "mean_token_accuracy": 0.03582368233401212,
+      "step": 366
+    },
+    {
+      "epoch": 1.3054045056042614,
+      "grad_norm": 1.0394208431243896,
+      "learning_rate": 7.663972951738708e-06,
+      "loss": 0.2109,
+      "mean_token_accuracy": 0.03519305539157358,
+      "step": 367
+    },
+    {
+      "epoch": 1.3089557207857063,
+      "grad_norm": 1.7033623456954956,
+      "learning_rate": 7.652104540598712e-06,
+      "loss": 0.2245,
+      "mean_token_accuracy": 0.032953111793176504,
+      "step": 368
+    },
+    {
+      "epoch": 1.3125069359671513,
+      "grad_norm": 1.0205411911010742,
+      "learning_rate": 7.640215299957283e-06,
+      "loss": 0.2089,
+      "mean_token_accuracy": 0.033462781550042564,
+      "step": 369
+    },
+    {
+      "epoch": 1.3160581511485963,
+      "grad_norm": 1.0846375226974487,
+      "learning_rate": 7.628305323191942e-06,
+      "loss": 0.2231,
+      "mean_token_accuracy": 0.03299743057868909,
+      "step": 370
+    },
+    {
+      "epoch": 1.319609366330041,
+      "grad_norm": 1.0049022436141968,
+      "learning_rate": 7.616374703843071e-06,
+      "loss": 0.2095,
+      "mean_token_accuracy": 0.03856489048484946,
+      "step": 371
+    },
+    {
+      "epoch": 1.323160581511486,
+      "grad_norm": 1.0001732110977173,
+      "learning_rate": 7.604423535613183e-06,
+      "loss": 0.2082,
+      "mean_token_accuracy": 0.038502528364915634,
+      "step": 372
+    },
+    {
+      "epoch": 1.326711796692931,
+      "grad_norm": 0.8959653377532959,
+      "learning_rate": 7.592451912366176e-06,
+      "loss": 0.2196,
+      "mean_token_accuracy": 0.03337086039027781,
+      "step": 373
+    },
+    {
+      "epoch": 1.3302630118743757,
+      "grad_norm": 0.7602843046188354,
+      "learning_rate": 7.580459928126607e-06,
+      "loss": 0.2158,
+      "mean_token_accuracy": 0.0357423342011316,
+      "step": 374
+    },
+    {
+      "epoch": 1.3338142270558206,
+      "grad_norm": 0.9946548342704773,
+      "learning_rate": 7.568447677078937e-06,
+      "loss": 0.2133,
+      "mean_token_accuracy": 0.04242287410306744,
+      "step": 375
+    },
+    {
+      "epoch": 1.3373654422372656,
+      "grad_norm": 0.9074912071228027,
+      "learning_rate": 7.556415253566814e-06,
+      "loss": 0.2127,
+      "mean_token_accuracy": 0.03988895907787082,
+      "step": 376
+    },
+    {
+      "epoch": 1.3409166574187106,
+      "grad_norm": 0.9580668807029724,
+      "learning_rate": 7.544362752092309e-06,
+      "loss": 0.2192,
+      "mean_token_accuracy": 0.037165690009715036,
+      "step": 377
+    },
+    {
+      "epoch": 1.3444678726001553,
+      "grad_norm": 1.1636556386947632,
+      "learning_rate": 7.532290267315189e-06,
+      "loss": 0.2139,
+      "mean_token_accuracy": 0.03949245312833227,
+      "step": 378
+    },
+    {
+      "epoch": 1.3480190877816003,
+      "grad_norm": 1.179194450378418,
+      "learning_rate": 7.52019789405217e-06,
+      "loss": 0.2205,
+      "mean_token_accuracy": 0.04195636221629684,
+      "step": 379
+    },
+    {
+      "epoch": 1.351570302963045,
+      "grad_norm": 1.0651406049728394,
+      "learning_rate": 7.508085727276169e-06,
+      "loss": 0.2106,
+      "mean_token_accuracy": 0.035512080634362064,
+      "step": 380
+    },
+    {
+      "epoch": 1.35512151814449,
+      "grad_norm": 1.2458323240280151,
+      "learning_rate": 7.495953862115561e-06,
+      "loss": 0.202,
+      "mean_token_accuracy": 0.041346150457684416,
+      "step": 381
+    },
+    {
+      "epoch": 1.358672733325935,
+      "grad_norm": 1.0106011629104614,
+      "learning_rate": 7.483802393853431e-06,
+      "loss": 0.2243,
+      "mean_token_accuracy": 0.03703319465057575,
+      "step": 382
+    },
+    {
+      "epoch": 1.36222394850738,
+      "grad_norm": 0.9817686676979065,
+      "learning_rate": 7.471631417926826e-06,
+      "loss": 0.2095,
+      "mean_token_accuracy": 0.04350194331345847,
+      "step": 383
+    },
+    {
+      "epoch": 1.365775163688825,
+      "grad_norm": 0.8153394460678101,
+      "learning_rate": 7.459441029926006e-06,
+      "loss": 0.2142,
+      "mean_token_accuracy": 0.036771614526514895,
+      "step": 384
+    },
+    {
+      "epoch": 1.3693263788702696,
+      "grad_norm": 1.0360249280929565,
+      "learning_rate": 7.447231325593689e-06,
+      "loss": 0.2179,
+      "mean_token_accuracy": 0.04122182544961106,
+      "step": 385
+    },
+    {
+      "epoch": 1.3728775940517146,
+      "grad_norm": 1.5009338855743408,
+      "learning_rate": 7.435002400824309e-06,
+      "loss": 0.2055,
+      "mean_token_accuracy": 0.03624642355134711,
+      "step": 386
+    },
+    {
+      "epoch": 1.3764288092331594,
+      "grad_norm": 1.2901028394699097,
+      "learning_rate": 7.422754351663252e-06,
+      "loss": 0.2141,
+      "mean_token_accuracy": 0.03373757504050445,
+      "step": 387
+    },
+    {
+      "epoch": 1.3799800244146043,
+      "grad_norm": 0.9240864515304565,
+      "learning_rate": 7.410487274306104e-06,
+      "loss": 0.2137,
+      "mean_token_accuracy": 0.037080507419887,
+      "step": 388
+    },
+    {
+      "epoch": 1.3835312395960493,
+      "grad_norm": 0.9881526827812195,
+      "learning_rate": 7.398201265097902e-06,
+      "loss": 0.2093,
+      "mean_token_accuracy": 0.038321658816130366,
+      "step": 389
+    },
+    {
+      "epoch": 1.3870824547774943,
+      "grad_norm": 1.0277929306030273,
+      "learning_rate": 7.385896420532372e-06,
+      "loss": 0.2107,
+      "mean_token_accuracy": 0.03264967034010624,
+      "step": 390
+    },
+    {
+      "epoch": 1.390633669958939,
+      "grad_norm": 1.092461347579956,
+      "learning_rate": 7.37357283725117e-06,
+      "loss": 0.2179,
+      "mean_token_accuracy": 0.041954617630835855,
+      "step": 391
+    },
+    {
+      "epoch": 1.394184885140384,
+      "grad_norm": 0.9751242399215698,
+      "learning_rate": 7.361230612043125e-06,
+      "loss": 0.212,
+      "mean_token_accuracy": 0.03894812406542769,
+      "step": 392
+    },
+    {
+      "epoch": 1.397736100321829,
+      "grad_norm": 1.890830159187317,
+      "learning_rate": 7.3488698418434824e-06,
+      "loss": 0.2172,
+      "mean_token_accuracy": 0.03717777330894023,
+      "step": 393
+    },
+    {
+      "epoch": 1.4012873155032737,
+      "grad_norm": 0.8804880380630493,
+      "learning_rate": 7.3364906237331345e-06,
+      "loss": 0.2204,
+      "mean_token_accuracy": 0.04008191364482627,
+      "step": 394
+    },
+    {
+      "epoch": 1.4048385306847186,
+      "grad_norm": 0.8936333060264587,
+      "learning_rate": 7.324093054937864e-06,
+      "loss": 0.2257,
+      "mean_token_accuracy": 0.03301443805685267,
+      "step": 395
+    },
+    {
+      "epoch": 1.4083897458661636,
+      "grad_norm": 1.1294957399368286,
+      "learning_rate": 7.311677232827583e-06,
+      "loss": 0.2233,
+      "mean_token_accuracy": 0.04345220190225518,
+      "step": 396
+    },
+    {
+      "epoch": 1.4119409610476086,
+      "grad_norm": 0.895491898059845,
+      "learning_rate": 7.299243254915558e-06,
+      "loss": 0.2083,
+      "mean_token_accuracy": 0.04362610646057874,
+      "step": 397
+    },
+    {
+      "epoch": 1.4154921762290533,
+      "grad_norm": 0.995071291923523,
+      "learning_rate": 7.286791218857654e-06,
+      "loss": 0.2061,
+      "mean_token_accuracy": 0.04004020981301437,
+      "step": 398
+    },
+    {
+      "epoch": 1.4190433914104983,
+      "grad_norm": 1.0868768692016602,
+      "learning_rate": 7.274321222451561e-06,
+      "loss": 0.2202,
+      "mean_token_accuracy": 0.03820264393652906,
+      "step": 399
+    },
+    {
+      "epoch": 1.4225946065919433,
+      "grad_norm": 0.8654125928878784,
+      "learning_rate": 7.261833363636036e-06,
+      "loss": 0.212,
+      "mean_token_accuracy": 0.03712031566828955,
+      "step": 400
+    },
+    {
+      "epoch": 1.426145821773388,
+      "grad_norm": 0.8807328939437866,
+      "learning_rate": 7.249327740490114e-06,
+      "loss": 0.2177,
+      "mean_token_accuracy": 0.03679542864483665,
+      "step": 401
+    },
+    {
+      "epoch": 1.429697036954833,
+      "grad_norm": 1.5379787683486938,
+      "learning_rate": 7.236804451232364e-06,
+      "loss": 0.2246,
+      "mean_token_accuracy": 0.03669572240141861,
+      "step": 402
+    },
+    {
+      "epoch": 1.433248252136278,
+      "grad_norm": 1.0445082187652588,
+      "learning_rate": 7.224263594220093e-06,
+      "loss": 0.2067,
+      "mean_token_accuracy": 0.03605292076827027,
+      "step": 403
+    },
+    {
+      "epoch": 1.436799467317723,
+      "grad_norm": 1.1058682203292847,
+      "learning_rate": 7.211705267948592e-06,
+      "loss": 0.2292,
+      "mean_token_accuracy": 0.036163140674034366,
+      "step": 404
+    },
+    {
+      "epoch": 1.4403506824991676,
+      "grad_norm": 0.9819757342338562,
+      "learning_rate": 7.199129571050345e-06,
+      "loss": 0.2037,
+      "mean_token_accuracy": 0.0389055151536013,
+      "step": 405
+    },
+    {
+      "epoch": 1.4439018976806126,
+      "grad_norm": 1.1496120691299438,
+      "learning_rate": 7.186536602294278e-06,
+      "loss": 0.2129,
+      "mean_token_accuracy": 0.03939257287129294,
+      "step": 406
+    },
+    {
+      "epoch": 1.4474531128620574,
+      "grad_norm": 0.9700692892074585,
+      "learning_rate": 7.173926460584956e-06,
+      "loss": 0.2229,
+      "mean_token_accuracy": 0.034446998419298325,
+      "step": 407
+    },
+    {
+      "epoch": 1.4510043280435023,
+      "grad_norm": 1.0145217180252075,
+      "learning_rate": 7.161299244961828e-06,
+      "loss": 0.1973,
+      "mean_token_accuracy": 0.03743944912821462,
+      "step": 408
+    },
+    {
+      "epoch": 1.4545555432249473,
+      "grad_norm": 4.707770347595215,
+      "learning_rate": 7.148655054598436e-06,
+      "loss": 0.2118,
+      "mean_token_accuracy": 0.03483109064109158,
+      "step": 409
+    },
+    {
+      "epoch": 1.4581067584063923,
+      "grad_norm": 1.4786440134048462,
+      "learning_rate": 7.135993988801644e-06,
+      "loss": 0.2141,
+      "mean_token_accuracy": 0.038917855665204115,
+      "step": 410
+    },
+    {
+      "epoch": 1.4616579735878372,
+      "grad_norm": 1.013339638710022,
+      "learning_rate": 7.1233161470108525e-06,
+      "loss": 0.2192,
+      "mean_token_accuracy": 0.033355345083691645,
+      "step": 411
+    },
+    {
+      "epoch": 1.465209188769282,
+      "grad_norm": 1.132614016532898,
+      "learning_rate": 7.110621628797222e-06,
+      "loss": 0.2125,
+      "mean_token_accuracy": 0.03544226154917851,
+      "step": 412
+    },
+    {
+      "epoch": 1.468760403950727,
+      "grad_norm": 0.8791604042053223,
+      "learning_rate": 7.097910533862886e-06,
+      "loss": 0.2229,
+      "mean_token_accuracy": 0.038792295250459574,
+      "step": 413
+    },
+    {
+      "epoch": 1.4723116191321717,
+      "grad_norm": 0.8531237840652466,
+      "learning_rate": 7.085182962040173e-06,
+      "loss": 0.2097,
+      "mean_token_accuracy": 0.04108742300013546,
+      "step": 414
+    },
+    {
+      "epoch": 1.4758628343136166,
+      "grad_norm": 0.8082413673400879,
+      "learning_rate": 7.072439013290824e-06,
+      "loss": 0.2204,
+      "mean_token_accuracy": 0.03489942911255639,
+      "step": 415
+    },
+    {
+      "epoch": 1.4794140494950616,
+      "grad_norm": 0.8210154175758362,
+      "learning_rate": 7.059678787705191e-06,
+      "loss": 0.2165,
+      "mean_token_accuracy": 0.03975592031565611,
+      "step": 416
+    },
+    {
+      "epoch": 1.4829652646765066,
+      "grad_norm": 0.9179903268814087,
+      "learning_rate": 7.046902385501477e-06,
+      "loss": 0.2115,
+      "mean_token_accuracy": 0.03542604565154761,
+      "step": 417
+    },
+    {
+      "epoch": 1.4865164798579513,
+      "grad_norm": 0.9118829965591431,
+      "learning_rate": 7.03410990702493e-06,
+      "loss": 0.2156,
+      "mean_token_accuracy": 0.036973950522224186,
+      "step": 418
+    },
+    {
+      "epoch": 1.4900676950393963,
+      "grad_norm": 0.9617339372634888,
+      "learning_rate": 7.02130145274706e-06,
+      "loss": 0.2196,
+      "mean_token_accuracy": 0.03498687346291263,
+      "step": 419
+    },
+    {
+      "epoch": 1.4936189102208413,
+      "grad_norm": 1.2348254919052124,
+      "learning_rate": 7.008477123264849e-06,
+      "loss": 0.2227,
+      "mean_token_accuracy": 0.035096731524390634,
+      "step": 420
+    },
+    {
+      "epoch": 1.497170125402286,
+      "grad_norm": 1.189163327217102,
+      "learning_rate": 6.995637019299963e-06,
+      "loss": 0.219,
+      "mean_token_accuracy": 0.039271164114325074,
+      "step": 421
+    },
+    {
+      "epoch": 1.500721340583731,
+      "grad_norm": 1.7504018545150757,
+      "learning_rate": 6.982781241697963e-06,
+      "loss": 0.2074,
+      "mean_token_accuracy": 0.03581648114777636,
+      "step": 422
+    },
+    {
+      "epoch": 1.504272555765176,
+      "grad_norm": 0.9267095327377319,
+      "learning_rate": 6.969909891427509e-06,
+      "loss": 0.2035,
+      "mean_token_accuracy": 0.04157684100209735,
+      "step": 423
+    },
+    {
+      "epoch": 1.507823770946621,
+      "grad_norm": 0.8455734848976135,
+      "learning_rate": 6.957023069579561e-06,
+      "loss": 0.2127,
+      "mean_token_accuracy": 0.04015160134440521,
+      "step": 424
+    },
+    {
+      "epoch": 1.5113749861280656,
+      "grad_norm": 0.744219958782196,
+      "learning_rate": 6.944120877366605e-06,
+      "loss": 0.1999,
+      "mean_token_accuracy": 0.040050789906672435,
+      "step": 425
+    },
+    {
+      "epoch": 1.5149262013095106,
+      "grad_norm": 1.2124446630477905,
+      "learning_rate": 6.931203416121831e-06,
+      "loss": 0.2172,
+      "mean_token_accuracy": 0.03541620170290116,
+      "step": 426
+    },
+    {
+      "epoch": 1.5184774164909554,
+      "grad_norm": 0.9840066432952881,
+      "learning_rate": 6.918270787298361e-06,
+      "loss": 0.2214,
+      "mean_token_accuracy": 0.03850063406935078,
+      "step": 427
+    },
+    {
+      "epoch": 1.5220286316724003,
+      "grad_norm": 0.9824987649917603,
+      "learning_rate": 6.90532309246844e-06,
+      "loss": 0.1999,
+      "mean_token_accuracy": 0.04028791991004255,
+      "step": 428
+    },
+    {
+      "epoch": 1.5255798468538453,
+      "grad_norm": 0.895193874835968,
+      "learning_rate": 6.89236043332264e-06,
+      "loss": 0.2085,
+      "mean_token_accuracy": 0.03992676907000714,
+      "step": 429
+    },
+    {
+      "epoch": 1.5291310620352903,
+      "grad_norm": 0.7865180969238281,
+      "learning_rate": 6.87938291166906e-06,
+      "loss": 0.2103,
+      "mean_token_accuracy": 0.04152852020342834,
+      "step": 430
+    },
+    {
+      "epoch": 1.5326822772167352,
+      "grad_norm": 1.021116018295288,
+      "learning_rate": 6.866390629432533e-06,
+      "loss": 0.227,
+      "mean_token_accuracy": 0.035348424105904996,
+      "step": 431
+    },
+    {
+      "epoch": 1.53623349239818,
+      "grad_norm": 1.3936994075775146,
+      "learning_rate": 6.8533836886538175e-06,
+      "loss": 0.2196,
+      "mean_token_accuracy": 0.038212617593671894,
+      "step": 432
+    },
+    {
+      "epoch": 1.539784707579625,
+      "grad_norm": 0.9515417814254761,
+      "learning_rate": 6.840362191488801e-06,
+      "loss": 0.2171,
+      "mean_token_accuracy": 0.037786570228490746,
+      "step": 433
+    },
+    {
+      "epoch": 1.5433359227610697,
+      "grad_norm": 1.0766609907150269,
+      "learning_rate": 6.8273262402076935e-06,
+      "loss": 0.2167,
+      "mean_token_accuracy": 0.033810261673352215,
+      "step": 434
+    },
+    {
+      "epoch": 1.5468871379425146,
+      "grad_norm": 0.9607550501823425,
+      "learning_rate": 6.814275937194233e-06,
+      "loss": 0.2182,
+      "mean_token_accuracy": 0.03374743430686067,
+      "step": 435
+    },
+    {
+      "epoch": 1.5504383531239596,
+      "grad_norm": 0.8036266565322876,
+      "learning_rate": 6.801211384944867e-06,
+      "loss": 0.2058,
+      "mean_token_accuracy": 0.04227837919461308,
+      "step": 436
+    },
+    {
+      "epoch": 1.5539895683054046,
+      "grad_norm": 1.0911484956741333,
+      "learning_rate": 6.788132686067963e-06,
+      "loss": 0.2087,
+      "mean_token_accuracy": 0.03773209520295495,
+      "step": 437
+    },
+    {
+      "epoch": 1.5575407834868495,
+      "grad_norm": 0.9918379187583923,
+      "learning_rate": 6.77503994328299e-06,
+      "loss": 0.2111,
+      "mean_token_accuracy": 0.03738878494914388,
+      "step": 438
+    },
+    {
+      "epoch": 1.5610919986682943,
+      "grad_norm": 0.8218954205513,
+      "learning_rate": 6.761933259419725e-06,
+      "loss": 0.2112,
+      "mean_token_accuracy": 0.03775092940122704,
+      "step": 439
+    },
+    {
+      "epoch": 1.5646432138497393,
+      "grad_norm": 0.9518681168556213,
+      "learning_rate": 6.748812737417428e-06,
+      "loss": 0.2242,
+      "mean_token_accuracy": 0.03522239228550461,
+      "step": 440
+    },
+    {
+      "epoch": 1.568194429031184,
+      "grad_norm": 1.0347586870193481,
+      "learning_rate": 6.7356784803240464e-06,
+      "loss": 0.2118,
+      "mean_token_accuracy": 0.03939944650483085,
+      "step": 441
+    },
+    {
+      "epoch": 1.571745644212629,
+      "grad_norm": 0.9376158714294434,
+      "learning_rate": 6.722530591295406e-06,
+      "loss": 0.2172,
+      "mean_token_accuracy": 0.03799405520840082,
+      "step": 442
+    },
+    {
+      "epoch": 1.575296859394074,
+      "grad_norm": 0.8542515635490417,
+      "learning_rate": 6.709369173594396e-06,
+      "loss": 0.2127,
+      "mean_token_accuracy": 0.04009221462183632,
+      "step": 443
+    },
+    {
+      "epoch": 1.578848074575519,
+      "grad_norm": 0.9119886159896851,
+      "learning_rate": 6.6961943305901515e-06,
+      "loss": 0.2111,
+      "mean_token_accuracy": 0.035265431808511494,
+      "step": 444
+    },
+    {
+      "epoch": 1.5823992897569639,
+      "grad_norm": 1.0846859216690063,
+      "learning_rate": 6.683006165757262e-06,
+      "loss": 0.2119,
+      "mean_token_accuracy": 0.03483350280839659,
+      "step": 445
+    },
+    {
+      "epoch": 1.5859505049384086,
+      "grad_norm": 0.9828306436538696,
+      "learning_rate": 6.669804782674937e-06,
+      "loss": 0.209,
+      "mean_token_accuracy": 0.039095990352507215,
+      "step": 446
+    },
+    {
+      "epoch": 1.5895017201198534,
+      "grad_norm": 0.9361124634742737,
+      "learning_rate": 6.656590285026203e-06,
+      "loss": 0.23,
+      "mean_token_accuracy": 0.03466227107855957,
+      "step": 447
+    },
+    {
+      "epoch": 1.5930529353012983,
+      "grad_norm": 1.067156434059143,
+      "learning_rate": 6.643362776597089e-06,
+      "loss": 0.2166,
+      "mean_token_accuracy": 0.03962546743787243,
+      "step": 448
+    },
+    {
+      "epoch": 1.5966041504827433,
+      "grad_norm": 0.9628080725669861,
+      "learning_rate": 6.630122361275811e-06,
+      "loss": 0.2206,
+      "mean_token_accuracy": 0.031071114382939413,
+      "step": 449
+    },
+    {
+      "epoch": 1.6001553656641883,
+      "grad_norm": 0.9671528935432434,
+      "learning_rate": 6.6168691430519524e-06,
+      "loss": 0.2121,
+      "mean_token_accuracy": 0.04189558584039332,
+      "step": 450
+    },
+    {
+      "epoch": 1.6037065808456332,
+      "grad_norm": 1.0035713911056519,
+      "learning_rate": 6.6036032260156526e-06,
+      "loss": 0.205,
+      "mean_token_accuracy": 0.03662525209074374,
+      "step": 451
+    },
+    {
+      "epoch": 1.607257796027078,
+      "grad_norm": 1.223681092262268,
+      "learning_rate": 6.590324714356784e-06,
+      "loss": 0.2028,
+      "mean_token_accuracy": 0.03729518407271826,
+      "step": 452
+    },
+    {
+      "epoch": 1.610809011208523,
+      "grad_norm": 1.0575910806655884,
+      "learning_rate": 6.5770337123641405e-06,
+      "loss": 0.2196,
+      "mean_token_accuracy": 0.035925558919188916,
+      "step": 453
+    },
+    {
+      "epoch": 1.6143602263899677,
+      "grad_norm": 0.9104030132293701,
+      "learning_rate": 6.563730324424609e-06,
+      "loss": 0.2034,
+      "mean_token_accuracy": 0.03999634759748005,
+      "step": 454
+    },
+    {
+      "epoch": 1.6179114415714126,
+      "grad_norm": 0.9227558970451355,
+      "learning_rate": 6.55041465502236e-06,
+      "loss": 0.2169,
+      "mean_token_accuracy": 0.04184209147206275,
+      "step": 455
+    },
+    {
+      "epoch": 1.6214626567528576,
+      "grad_norm": 1.1232300996780396,
+      "learning_rate": 6.53708680873802e-06,
+      "loss": 0.2108,
+      "mean_token_accuracy": 0.03624327100624214,
+      "step": 456
+    },
+    {
+      "epoch": 1.6250138719343026,
+      "grad_norm": 1.1040129661560059,
+      "learning_rate": 6.523746890247853e-06,
+      "loss": 0.2165,
+      "mean_token_accuracy": 0.033005929173668846,
+      "step": 457
+    },
+    {
+      "epoch": 1.6285650871157475,
+      "grad_norm": 1.0858967304229736,
+      "learning_rate": 6.510395004322937e-06,
+      "loss": 0.2156,
+      "mean_token_accuracy": 0.03949285501221311,
+      "step": 458
+    },
+    {
+      "epoch": 1.6321163022971923,
+      "grad_norm": 1.2183141708374023,
+      "learning_rate": 6.49703125582834e-06,
+      "loss": 0.2179,
+      "mean_token_accuracy": 0.03773052282485878,
+      "step": 459
+    },
+    {
+      "epoch": 1.6356675174786373,
+      "grad_norm": 0.9809972643852234,
+      "learning_rate": 6.4836557497222995e-06,
+      "loss": 0.2185,
+      "mean_token_accuracy": 0.03755065660698165,
+      "step": 460
+    },
+    {
+      "epoch": 1.639218732660082,
+      "grad_norm": 0.8669877648353577,
+      "learning_rate": 6.470268591055398e-06,
+      "loss": 0.1969,
+      "mean_token_accuracy": 0.03949119582102867,
+      "step": 461
+    },
+    {
+      "epoch": 1.642769947841527,
+      "grad_norm": 0.9765094518661499,
+      "learning_rate": 6.456869884969738e-06,
+      "loss": 0.2146,
+      "mean_token_accuracy": 0.03167624427806004,
+      "step": 462
+    },
+    {
+      "epoch": 1.646321163022972,
+      "grad_norm": 1.4784200191497803,
+      "learning_rate": 6.443459736698106e-06,
+      "loss": 0.2109,
+      "mean_token_accuracy": 0.03601810210602707,
+      "step": 463
+    },
+    {
+      "epoch": 1.649872378204417,
+      "grad_norm": 1.0272235870361328,
+      "learning_rate": 6.430038251563166e-06,
+      "loss": 0.2069,
+      "mean_token_accuracy": 0.039396893360390095,
+      "step": 464
+    },
+    {
+      "epoch": 1.6534235933858619,
+      "grad_norm": 0.9422529339790344,
+      "learning_rate": 6.416605534976614e-06,
+      "loss": 0.2046,
+      "mean_token_accuracy": 0.03725221472996054,
+      "step": 465
+    },
+    {
+      "epoch": 1.6569748085673066,
+      "grad_norm": 1.115491271018982,
+      "learning_rate": 6.403161692438364e-06,
+      "loss": 0.2054,
+      "mean_token_accuracy": 0.03913160233787494,
+      "step": 466
+    },
+    {
+      "epoch": 1.6605260237487516,
+      "grad_norm": 0.8288925886154175,
+      "learning_rate": 6.3897068295357e-06,
+      "loss": 0.2092,
+      "mean_token_accuracy": 0.03602888582099695,
+      "step": 467
+    },
+    {
+      "epoch": 1.6640772389301963,
+      "grad_norm": 1.186328649520874,
+      "learning_rate": 6.376241051942477e-06,
+      "loss": 0.2133,
+      "mean_token_accuracy": 0.037299348470696714,
+      "step": 468
+    },
+    {
+      "epoch": 1.6676284541116413,
+      "grad_norm": 0.938909113407135,
+      "learning_rate": 6.362764465418258e-06,
+      "loss": 0.2099,
+      "mean_token_accuracy": 0.03910366345371585,
+      "step": 469
+    },
+    {
+      "epoch": 1.6711796692930863,
+      "grad_norm": 1.1985607147216797,
+      "learning_rate": 6.349277175807506e-06,
+      "loss": 0.2213,
+      "mean_token_accuracy": 0.029635498765856028,
+      "step": 470
+    },
+    {
+      "epoch": 1.6747308844745312,
+      "grad_norm": 0.8005155324935913,
+      "learning_rate": 6.3357792890387485e-06,
+      "loss": 0.2098,
+      "mean_token_accuracy": 0.03648436323783244,
+      "step": 471
+    },
+    {
+      "epoch": 1.6782820996559762,
+      "grad_norm": 1.2143996953964233,
+      "learning_rate": 6.322270911123734e-06,
+      "loss": 0.2019,
+      "mean_token_accuracy": 0.033510430232126964,
+      "step": 472
+    },
+    {
+      "epoch": 1.681833314837421,
+      "grad_norm": 1.0368294715881348,
+      "learning_rate": 6.308752148156614e-06,
+      "loss": 0.2147,
+      "mean_token_accuracy": 0.03938718199060531,
+      "step": 473
+    },
+    {
+      "epoch": 1.6853845300188657,
+      "grad_norm": 1.0544377565383911,
+      "learning_rate": 6.295223106313104e-06,
+      "loss": 0.2091,
+      "mean_token_accuracy": 0.03419090985698858,
+      "step": 474
+    },
+    {
+      "epoch": 1.6889357452003106,
+      "grad_norm": 0.8890665173530579,
+      "learning_rate": 6.281683891849645e-06,
+      "loss": 0.2069,
+      "mean_token_accuracy": 0.041956931356253335,
+      "step": 475
+    },
+    {
+      "epoch": 1.6924869603817556,
+      "grad_norm": 0.9658064842224121,
+      "learning_rate": 6.268134611102578e-06,
+      "loss": 0.2068,
+      "mean_token_accuracy": 0.03853438159421785,
+      "step": 476
+    },
+    {
+      "epoch": 1.6960381755632006,
+      "grad_norm": 0.9596953392028809,
+      "learning_rate": 6.254575370487299e-06,
+      "loss": 0.2093,
+      "mean_token_accuracy": 0.03829575631971238,
+      "step": 477
+    },
+    {
+      "epoch": 1.6995893907446455,
+      "grad_norm": 1.0915920734405518,
+      "learning_rate": 6.2410062764974366e-06,
+      "loss": 0.2098,
+      "mean_token_accuracy": 0.03425979395979084,
+      "step": 478
+    },
+    {
+      "epoch": 1.7031406059260903,
+      "grad_norm": 1.2812224626541138,
+      "learning_rate": 6.227427435703997e-06,
+      "loss": 0.218,
+      "mean_token_accuracy": 0.04161245372961275,
+      "step": 479
+    },
+    {
+      "epoch": 1.7066918211075353,
+      "grad_norm": 1.0697816610336304,
+      "learning_rate": 6.213838954754543e-06,
+      "loss": 0.2111,
+      "mean_token_accuracy": 0.037880031057284214,
+      "step": 480
+    },
+    {
+      "epoch": 1.71024303628898,
+      "grad_norm": 1.0685772895812988,
+      "learning_rate": 6.2002409403723525e-06,
+      "loss": 0.2126,
+      "mean_token_accuracy": 0.03928147720944253,
+      "step": 481
+    },
+    {
+      "epoch": 1.713794251470425,
+      "grad_norm": 1.0521827936172485,
+      "learning_rate": 6.186633499355576e-06,
+      "loss": 0.2215,
+      "mean_token_accuracy": 0.03902989218295261,
+      "step": 482
+    },
+    {
+      "epoch": 1.71734546665187,
+      "grad_norm": 1.095388650894165,
+      "learning_rate": 6.173016738576396e-06,
+      "loss": 0.2178,
+      "mean_token_accuracy": 0.03534844119349145,
+      "step": 483
+    },
+    {
+      "epoch": 1.720896681833315,
+      "grad_norm": 1.2036548852920532,
+      "learning_rate": 6.159390764980202e-06,
+      "loss": 0.2042,
+      "mean_token_accuracy": 0.039945425454789074,
+      "step": 484
+    },
+    {
+      "epoch": 1.7244478970147599,
+      "grad_norm": 1.3349320888519287,
+      "learning_rate": 6.145755685584731e-06,
+      "loss": 0.2095,
+      "mean_token_accuracy": 0.040973992294311756,
+      "step": 485
+    },
+    {
+      "epoch": 1.7279991121962046,
+      "grad_norm": 0.9357543587684631,
+      "learning_rate": 6.132111607479243e-06,
+      "loss": 0.209,
+      "mean_token_accuracy": 0.035786267388175474,
+      "step": 486
+    },
+    {
+      "epoch": 1.7315503273776496,
+      "grad_norm": 1.2930482625961304,
+      "learning_rate": 6.118458637823669e-06,
+      "loss": 0.213,
+      "mean_token_accuracy": 0.03360192391482997,
+      "step": 487
+    },
+    {
+      "epoch": 1.7351015425590943,
+      "grad_norm": 1.242834210395813,
+      "learning_rate": 6.104796883847777e-06,
+      "loss": 0.1924,
+      "mean_token_accuracy": 0.04279121259241947,
+      "step": 488
+    },
+    {
+      "epoch": 1.7386527577405393,
+      "grad_norm": 0.919654130935669,
+      "learning_rate": 6.091126452850324e-06,
+      "loss": 0.2134,
+      "mean_token_accuracy": 0.03625394479422539,
+      "step": 489
+    },
+    {
+      "epoch": 1.7422039729219843,
+      "grad_norm": 1.0187784433364868,
+      "learning_rate": 6.077447452198219e-06,
+      "loss": 0.204,
+      "mean_token_accuracy": 0.046423808718827786,
+      "step": 490
+    },
+    {
+      "epoch": 1.7457551881034292,
+      "grad_norm": 1.249862790107727,
+      "learning_rate": 6.063759989325673e-06,
+      "loss": 0.2159,
+      "mean_token_accuracy": 0.03750733687047614,
+      "step": 491
+    },
+    {
+      "epoch": 1.7493064032848742,
+      "grad_norm": 1.0927352905273438,
+      "learning_rate": 6.050064171733362e-06,
+      "loss": 0.2136,
+      "mean_token_accuracy": 0.040005916329391766,
+      "step": 492
+    },
+    {
+      "epoch": 1.752857618466319,
+      "grad_norm": 0.8594483137130737,
+      "learning_rate": 6.0363601069875755e-06,
+      "loss": 0.2241,
+      "mean_token_accuracy": 0.035941219015512615,
+      "step": 493
+    },
+    {
+      "epoch": 1.7564088336477637,
+      "grad_norm": 0.9354445934295654,
+      "learning_rate": 6.022647902719384e-06,
+      "loss": 0.2037,
+      "mean_token_accuracy": 0.037646496271918295,
+      "step": 494
+    },
+    {
+      "epoch": 1.7599600488292086,
+      "grad_norm": 0.9080971479415894,
+      "learning_rate": 6.008927666623775e-06,
+      "loss": 0.2043,
+      "mean_token_accuracy": 0.04433829447225435,
+      "step": 495
+    },
+    {
+      "epoch": 1.7635112640106536,
+      "grad_norm": 1.0300090312957764,
+      "learning_rate": 5.9951995064588245e-06,
+      "loss": 0.1977,
+      "mean_token_accuracy": 0.0387411278388754,
+      "step": 496
+    },
+    {
+      "epoch": 1.7670624791920986,
+      "grad_norm": 1.3653299808502197,
+      "learning_rate": 5.981463530044841e-06,
+      "loss": 0.2134,
+      "mean_token_accuracy": 0.03764768938708585,
+      "step": 497
+    },
+    {
+      "epoch": 1.7706136943735435,
+      "grad_norm": 1.3612983226776123,
+      "learning_rate": 5.967719845263524e-06,
+      "loss": 0.2152,
+      "mean_token_accuracy": 0.03451220032366109,
+      "step": 498
+    },
+    {
+      "epoch": 1.7741649095549885,
+      "grad_norm": 0.9325419068336487,
+      "learning_rate": 5.953968560057112e-06,
+      "loss": 0.2147,
+      "mean_token_accuracy": 0.0384930808286299,
+      "step": 499
+    },
+    {
+      "epoch": 1.7777161247364333,
+      "grad_norm": 0.928591787815094,
+      "learning_rate": 5.940209782427535e-06,
+      "loss": 0.214,
+      "mean_token_accuracy": 0.03583947871811688,
+      "step": 500
+    },
+    {
+      "epoch": 1.781267339917878,
+      "grad_norm": 0.897335946559906,
+      "learning_rate": 5.926443620435572e-06,
+      "loss": 0.2157,
+      "mean_token_accuracy": 0.038455418245575856,
+      "step": 501
+    },
+    {
+      "epoch": 1.784818555099323,
+      "grad_norm": 0.91599440574646,
+      "learning_rate": 5.912670182199998e-06,
+      "loss": 0.2073,
+      "mean_token_accuracy": 0.03296034881896048,
+      "step": 502
+    },
+    {
+      "epoch": 1.788369770280768,
+      "grad_norm": 1.1746525764465332,
+      "learning_rate": 5.898889575896731e-06,
+      "loss": 0.2048,
+      "mean_token_accuracy": 0.03975271817398607,
+      "step": 503
+    },
+    {
+      "epoch": 1.791920985462213,
+      "grad_norm": 0.8332899808883667,
+      "learning_rate": 5.8851019097579935e-06,
+      "loss": 0.2103,
+      "mean_token_accuracy": 0.037684340750274714,
+      "step": 504
+    },
+    {
+      "epoch": 1.7954722006436579,
+      "grad_norm": 0.9919717907905579,
+      "learning_rate": 5.871307292071449e-06,
+      "loss": 0.2195,
+      "mean_token_accuracy": 0.03749480012265849,
+      "step": 505
+    },
+    {
+      "epoch": 1.7990234158251026,
+      "grad_norm": 0.9827485680580139,
+      "learning_rate": 5.857505831179361e-06,
+      "loss": 0.2189,
+      "mean_token_accuracy": 0.03539751814241754,
+      "step": 506
+    },
+    {
+      "epoch": 1.8025746310065476,
+      "grad_norm": 1.1644009351730347,
+      "learning_rate": 5.843697635477742e-06,
+      "loss": 0.2006,
+      "mean_token_accuracy": 0.03862465297243034,
+      "step": 507
+    },
+    {
+      "epoch": 1.8061258461879923,
+      "grad_norm": 0.9732457399368286,
+      "learning_rate": 5.8298828134154935e-06,
+      "loss": 0.2134,
+      "mean_token_accuracy": 0.037525917465245584,
+      "step": 508
+    },
+    {
+      "epoch": 1.8096770613694373,
+      "grad_norm": 0.8709985017776489,
+      "learning_rate": 5.816061473493565e-06,
+      "loss": 0.2217,
+      "mean_token_accuracy": 0.034015703642580775,
+      "step": 509
+    },
+    {
+      "epoch": 1.8132282765508823,
+      "grad_norm": 1.3914642333984375,
+      "learning_rate": 5.802233724264094e-06,
+      "loss": 0.2185,
+      "mean_token_accuracy": 0.03763900963895139,
+      "step": 510
+    },
+    {
+      "epoch": 1.8167794917323272,
+      "grad_norm": 0.8272927403450012,
+      "learning_rate": 5.788399674329559e-06,
+      "loss": 0.2144,
+      "mean_token_accuracy": 0.03993615026047337,
+      "step": 511
+    },
+    {
+      "epoch": 1.8203307069137722,
+      "grad_norm": 0.843360185623169,
+      "learning_rate": 5.774559432341918e-06,
+      "loss": 0.2125,
+      "mean_token_accuracy": 0.04127829002754879,
+      "step": 512
+    },
+    {
+      "epoch": 1.823881922095217,
+      "grad_norm": 0.801755964756012,
+      "learning_rate": 5.760713107001773e-06,
+      "loss": 0.2045,
+      "mean_token_accuracy": 0.03806999635708053,
+      "step": 513
+    },
+    {
+      "epoch": 1.827433137276662,
+      "grad_norm": 1.2582050561904907,
+      "learning_rate": 5.746860807057491e-06,
+      "loss": 0.2072,
+      "mean_token_accuracy": 0.04005840528043336,
+      "step": 514
+    },
+    {
+      "epoch": 1.8309843524581066,
+      "grad_norm": 0.9559704661369324,
+      "learning_rate": 5.7330026413043726e-06,
+      "loss": 0.2228,
+      "mean_token_accuracy": 0.034368609467492206,
+      "step": 515
+    },
+    {
+      "epoch": 1.8345355676395516,
+      "grad_norm": 0.9817132353782654,
+      "learning_rate": 5.719138718583781e-06,
+      "loss": 0.2036,
+      "mean_token_accuracy": 0.03506710561487125,
+      "step": 516
+    },
+    {
+      "epoch": 1.8380867828209966,
+      "grad_norm": 0.9805539846420288,
+      "learning_rate": 5.705269147782303e-06,
+      "loss": 0.2147,
+      "mean_token_accuracy": 0.03908530145417899,
+      "step": 517
+    },
+    {
+      "epoch": 1.8416379980024415,
+      "grad_norm": 1.0621707439422607,
+      "learning_rate": 5.6913940378308755e-06,
+      "loss": 0.2108,
+      "mean_token_accuracy": 0.04440686923044268,
+      "step": 518
+    },
+    {
+      "epoch": 1.8451892131838865,
+      "grad_norm": 1.127663254737854,
+      "learning_rate": 5.677513497703947e-06,
+      "loss": 0.2088,
+      "mean_token_accuracy": 0.03553861051477725,
+      "step": 519
+    },
+    {
+      "epoch": 1.8487404283653313,
+      "grad_norm": 0.9272313117980957,
+      "learning_rate": 5.663627636418611e-06,
+      "loss": 0.2182,
+      "mean_token_accuracy": 0.03748767764045624,
+      "step": 520
+    },
+    {
+      "epoch": 1.852291643546776,
+      "grad_norm": 0.8673235177993774,
+      "learning_rate": 5.649736563033754e-06,
+      "loss": 0.2166,
+      "mean_token_accuracy": 0.03570136218331754,
+      "step": 521
+    },
+    {
+      "epoch": 1.855842858728221,
+      "grad_norm": 1.1679906845092773,
+      "learning_rate": 5.635840386649197e-06,
+      "loss": 0.2268,
+      "mean_token_accuracy": 0.03190056296989496,
+      "step": 522
+    },
+    {
+      "epoch": 1.859394073909666,
+      "grad_norm": 1.715111494064331,
+      "learning_rate": 5.621939216404842e-06,
+      "loss": 0.1987,
+      "mean_token_accuracy": 0.04647469030533102,
+      "step": 523
+    },
+    {
+      "epoch": 1.862945289091111,
+      "grad_norm": 0.9395443201065063,
+      "learning_rate": 5.608033161479811e-06,
+      "loss": 0.2078,
+      "mean_token_accuracy": 0.037822478254383896,
+      "step": 524
+    },
+    {
+      "epoch": 1.8664965042725559,
+      "grad_norm": 1.0644373893737793,
+      "learning_rate": 5.594122331091591e-06,
+      "loss": 0.2138,
+      "mean_token_accuracy": 0.03309350729978178,
+      "step": 525
+    },
+    {
+      "epoch": 1.8700477194540006,
+      "grad_norm": 1.1506586074829102,
+      "learning_rate": 5.580206834495169e-06,
+      "loss": 0.2106,
+      "mean_token_accuracy": 0.03856055447977269,
+      "step": 526
+    },
+    {
+      "epoch": 1.8735989346354456,
+      "grad_norm": 1.2600033283233643,
+      "learning_rate": 5.566286780982193e-06,
+      "loss": 0.2145,
+      "mean_token_accuracy": 0.03731459183472907,
+      "step": 527
+    },
+    {
+      "epoch": 1.8771501498168903,
+      "grad_norm": 0.9154390692710876,
+      "learning_rate": 5.552362279880091e-06,
+      "loss": 0.209,
+      "mean_token_accuracy": 0.04003632850435679,
+      "step": 528
+    },
+    {
+      "epoch": 1.8807013649983353,
+      "grad_norm": 1.1231424808502197,
+      "learning_rate": 5.538433440551221e-06,
+      "loss": 0.2109,
+      "mean_token_accuracy": 0.03613886435414315,
+      "step": 529
+    },
+    {
+      "epoch": 1.8842525801797803,
+      "grad_norm": 0.8639524579048157,
+      "learning_rate": 5.524500372392021e-06,
+      "loss": 0.2216,
+      "mean_token_accuracy": 0.04050558650124003,
+      "step": 530
+    },
+    {
+      "epoch": 1.8878037953612252,
+      "grad_norm": 0.93828946352005,
+      "learning_rate": 5.5105631848321375e-06,
+      "loss": 0.2131,
+      "mean_token_accuracy": 0.03937426700940705,
+      "step": 531
+    },
+    {
+      "epoch": 1.8913550105426702,
+      "grad_norm": 0.7636436223983765,
+      "learning_rate": 5.496621987333567e-06,
+      "loss": 0.2097,
+      "mean_token_accuracy": 0.0382932660941151,
+      "step": 532
+    },
+    {
+      "epoch": 1.894906225724115,
+      "grad_norm": 0.9772413969039917,
+      "learning_rate": 5.482676889389808e-06,
+      "loss": 0.2035,
+      "mean_token_accuracy": 0.0444465134969505,
+      "step": 533
+    },
+    {
+      "epoch": 1.89845744090556,
+      "grad_norm": 0.761626660823822,
+      "learning_rate": 5.468728000524987e-06,
+      "loss": 0.2022,
+      "mean_token_accuracy": 0.03659985137346666,
+      "step": 534
+    },
+    {
+      "epoch": 1.9020086560870046,
+      "grad_norm": 0.8718194961547852,
+      "learning_rate": 5.454775430293008e-06,
+      "loss": 0.2072,
+      "mean_token_accuracy": 0.04068213729078707,
+      "step": 535
+    },
+    {
+      "epoch": 1.9055598712684496,
+      "grad_norm": 1.0313729047775269,
+      "learning_rate": 5.440819288276683e-06,
+      "loss": 0.2206,
+      "mean_token_accuracy": 0.035872990021744044,
+      "step": 536
+    },
+    {
+      "epoch": 1.9091110864498946,
+      "grad_norm": 1.0423539876937866,
+      "learning_rate": 5.426859684086881e-06,
+      "loss": 0.2074,
+      "mean_token_accuracy": 0.041051392698136624,
+      "step": 537
+    },
+    {
+      "epoch": 1.9126623016313395,
+      "grad_norm": 0.9633597135543823,
+      "learning_rate": 5.412896727361663e-06,
+      "loss": 0.2048,
+      "mean_token_accuracy": 0.03679951139201876,
+      "step": 538
+    },
+    {
+      "epoch": 1.9162135168127845,
+      "grad_norm": 0.912843644618988,
+      "learning_rate": 5.398930527765416e-06,
+      "loss": 0.2101,
+      "mean_token_accuracy": 0.03617726853190106,
+      "step": 539
+    },
+    {
+      "epoch": 1.9197647319942293,
+      "grad_norm": 1.0215671062469482,
+      "learning_rate": 5.384961194988002e-06,
+      "loss": 0.2072,
+      "mean_token_accuracy": 0.03606809282791801,
+      "step": 540
+    },
+    {
+      "epoch": 1.9233159471756742,
+      "grad_norm": 1.0828865766525269,
+      "learning_rate": 5.370988838743889e-06,
+      "loss": 0.2036,
+      "mean_token_accuracy": 0.033105893973697675,
+      "step": 541
+    },
+    {
+      "epoch": 1.926867162357119,
+      "grad_norm": 0.932388961315155,
+      "learning_rate": 5.357013568771288e-06,
+      "loss": 0.1939,
+      "mean_token_accuracy": 0.04302502845530398,
+      "step": 542
+    },
+    {
+      "epoch": 1.930418377538564,
+      "grad_norm": 0.7949482798576355,
+      "learning_rate": 5.343035494831298e-06,
+      "loss": 0.2002,
+      "mean_token_accuracy": 0.03803440502088051,
+      "step": 543
+    },
+    {
+      "epoch": 1.933969592720009,
+      "grad_norm": 0.931699275970459,
+      "learning_rate": 5.32905472670704e-06,
+      "loss": 0.2212,
+      "mean_token_accuracy": 0.03603428568749223,
+      "step": 544
+    },
+    {
+      "epoch": 1.9375208079014539,
+      "grad_norm": 0.9046686887741089,
+      "learning_rate": 5.315071374202792e-06,
+      "loss": 0.2025,
+      "mean_token_accuracy": 0.0395622910837119,
+      "step": 545
+    },
+    {
+      "epoch": 1.9410720230828988,
+      "grad_norm": 0.9733835458755493,
+      "learning_rate": 5.301085547143135e-06,
+      "loss": 0.2163,
+      "mean_token_accuracy": 0.035136927999701584,
+      "step": 546
+    },
+    {
+      "epoch": 1.9446232382643436,
+      "grad_norm": 1.6521155834197998,
+      "learning_rate": 5.287097355372079e-06,
+      "loss": 0.1976,
+      "mean_token_accuracy": 0.03871955324575538,
+      "step": 547
+    },
+    {
+      "epoch": 1.9481744534457883,
+      "grad_norm": 0.8703343272209167,
+      "learning_rate": 5.273106908752211e-06,
+      "loss": 0.2158,
+      "mean_token_accuracy": 0.03640766240459925,
+      "step": 548
+    },
+    {
+      "epoch": 1.9517256686272333,
+      "grad_norm": 1.1232815980911255,
+      "learning_rate": 5.259114317163822e-06,
+      "loss": 0.217,
+      "mean_token_accuracy": 0.036017874652316095,
+      "step": 549
+    },
+    {
+      "epoch": 1.9552768838086783,
+      "grad_norm": 1.1956578493118286,
+      "learning_rate": 5.245119690504056e-06,
+      "loss": 0.2009,
+      "mean_token_accuracy": 0.04000877890211996,
+      "step": 550
+    },
+    {
+      "epoch": 1.9588280989901232,
+      "grad_norm": 0.9923151135444641,
+      "learning_rate": 5.231123138686036e-06,
+      "loss": 0.2153,
+      "mean_token_accuracy": 0.039021609194605844,
+      "step": 551
+    },
+    {
+      "epoch": 1.9623793141715682,
+      "grad_norm": 0.8768529295921326,
+      "learning_rate": 5.217124771638008e-06,
+      "loss": 0.2066,
+      "mean_token_accuracy": 0.03712507517411723,
+      "step": 552
+    },
+    {
+      "epoch": 1.965930529353013,
+      "grad_norm": 1.1248713731765747,
+      "learning_rate": 5.2031246993024705e-06,
+      "loss": 0.2071,
+      "mean_token_accuracy": 0.039892343469546176,
+      "step": 553
+    },
+    {
+      "epoch": 1.969481744534458,
+      "grad_norm": 0.9119243025779724,
+      "learning_rate": 5.1891230316353215e-06,
+      "loss": 0.1952,
+      "mean_token_accuracy": 0.04049671125903842,
+      "step": 554
+    },
+    {
+      "epoch": 1.9730329597159026,
+      "grad_norm": 0.8841214179992676,
+      "learning_rate": 5.1751198786049815e-06,
+      "loss": 0.2062,
+      "mean_token_accuracy": 0.037975409573846264,
+      "step": 555
+    },
+    {
+      "epoch": 1.9765841748973476,
+      "grad_norm": 0.9124864935874939,
+      "learning_rate": 5.161115350191543e-06,
+      "loss": 0.198,
+      "mean_token_accuracy": 0.038486146746436134,
+      "step": 556
+    },
+    {
+      "epoch": 1.9801353900787926,
+      "grad_norm": 1.0492109060287476,
+      "learning_rate": 5.147109556385898e-06,
+      "loss": 0.2181,
+      "mean_token_accuracy": 0.03872556518763304,
+      "step": 557
+    },
+    {
+      "epoch": 1.9836866052602375,
+      "grad_norm": 1.163588047027588,
+      "learning_rate": 5.133102607188875e-06,
+      "loss": 0.2046,
+      "mean_token_accuracy": 0.04012502250043326,
+      "step": 558
+    },
+    {
+      "epoch": 1.9872378204416825,
+      "grad_norm": 0.9493401646614075,
+      "learning_rate": 5.119094612610381e-06,
+      "loss": 0.2074,
+      "mean_token_accuracy": 0.03857941404930898,
+      "step": 559
+    },
+    {
+      "epoch": 1.9907890356231273,
+      "grad_norm": 1.357474446296692,
+      "learning_rate": 5.10508568266853e-06,
+      "loss": 0.2164,
+      "mean_token_accuracy": 0.037847537991183344,
+      "step": 560
+    },
+    {
+      "epoch": 1.9943402508045722,
+      "grad_norm": 1.5208334922790527,
+      "learning_rate": 5.091075927388785e-06,
+      "loss": 0.2208,
+      "mean_token_accuracy": 0.03787937617380521,
+      "step": 561
+    },
+    {
+      "epoch": 1.997891465986017,
+      "grad_norm": 1.336180329322815,
+      "learning_rate": 5.077065456803089e-06,
+      "loss": 0.2072,
+      "mean_token_accuracy": 0.036571884866134496,
+      "step": 562
+    },
+    {
+      "epoch": 2.003551215181445,
+      "grad_norm": 1.1412030458450317,
+      "learning_rate": 5.063054380949003e-06,
+      "loss": 0.3256,
+      "mean_token_accuracy": 0.04019247937043581,
+      "step": 563
+    },
+    {
+      "epoch": 2.00710243036289,
+      "grad_norm": 1.1244559288024902,
+      "learning_rate": 5.049042809868845e-06,
+      "loss": 0.1987,
+      "mean_token_accuracy": 0.042167539832007606,
+      "step": 564
+    },
+    {
+      "epoch": 2.010653645544335,
+      "grad_norm": 0.9170485138893127,
+      "learning_rate": 5.035030853608817e-06,
+      "loss": 0.2067,
+      "mean_token_accuracy": 0.034519848279160215,
+      "step": 565
+    },
+    {
+      "epoch": 2.0142048607257794,
+      "grad_norm": 1.1951957941055298,
+      "learning_rate": 5.0210186222181515e-06,
+      "loss": 0.2132,
+      "mean_token_accuracy": 0.03651535582685028,
+      "step": 566
+    },
+    {
+      "epoch": 2.0177560759072244,
+      "grad_norm": 1.1706410646438599,
+      "learning_rate": 5.007006225748238e-06,
+      "loss": 0.198,
+      "mean_token_accuracy": 0.040820111647917656,
+      "step": 567
+    },
+    {
+      "epoch": 2.0213072910886694,
+      "grad_norm": 1.603208065032959,
+      "learning_rate": 4.992993774251764e-06,
+      "loss": 0.201,
+      "mean_token_accuracy": 0.040759597257419955,
+      "step": 568
+    },
+    {
+      "epoch": 2.0248585062701143,
+      "grad_norm": 1.4895318746566772,
+      "learning_rate": 4.97898137778185e-06,
+      "loss": 0.2021,
+      "mean_token_accuracy": 0.040484553053829586,
+      "step": 569
+    },
+    {
+      "epoch": 2.0284097214515593,
+      "grad_norm": 0.868621826171875,
+      "learning_rate": 4.964969146391184e-06,
+      "loss": 0.2115,
+      "mean_token_accuracy": 0.0362443107587751,
+      "step": 570
+    },
+    {
+      "epoch": 2.0319609366330043,
+      "grad_norm": 0.8804165124893188,
+      "learning_rate": 4.950957190131157e-06,
+      "loss": 0.2073,
+      "mean_token_accuracy": 0.034275940739462385,
+      "step": 571
+    },
+    {
+      "epoch": 2.0355121518144492,
+      "grad_norm": 0.9676554203033447,
+      "learning_rate": 4.936945619050998e-06,
+      "loss": 0.2061,
+      "mean_token_accuracy": 0.03710194981977111,
+      "step": 572
+    },
+    {
+      "epoch": 2.0390633669958937,
+      "grad_norm": 0.9396111369132996,
+      "learning_rate": 4.922934543196912e-06,
+      "loss": 0.2082,
+      "mean_token_accuracy": 0.03654846304198145,
+      "step": 573
+    },
+    {
+      "epoch": 2.0426145821773387,
+      "grad_norm": 0.9149210453033447,
+      "learning_rate": 4.908924072611218e-06,
+      "loss": 0.2101,
+      "mean_token_accuracy": 0.03367238876671763,
+      "step": 574
+    },
+    {
+      "epoch": 2.0461657973587837,
+      "grad_norm": 1.1090168952941895,
+      "learning_rate": 4.894914317331471e-06,
+      "loss": 0.213,
+      "mean_token_accuracy": 0.0373724405435496,
+      "step": 575
+    },
+    {
+      "epoch": 2.0497170125402286,
+      "grad_norm": 1.4507415294647217,
+      "learning_rate": 4.88090538738962e-06,
+      "loss": 0.2124,
+      "mean_token_accuracy": 0.037985340273735346,
+      "step": 576
+    },
+    {
+      "epoch": 2.0532682277216736,
+      "grad_norm": 1.0356301069259644,
+      "learning_rate": 4.866897392811127e-06,
+      "loss": 0.2111,
+      "mean_token_accuracy": 0.03993125998385949,
+      "step": 577
+    },
+    {
+      "epoch": 2.0568194429031186,
+      "grad_norm": 0.9829997420310974,
+      "learning_rate": 4.852890443614105e-06,
+      "loss": 0.2215,
+      "mean_token_accuracy": 0.035097517848043935,
+      "step": 578
+    },
+    {
+      "epoch": 2.060370658084563,
+      "grad_norm": 1.0447118282318115,
+      "learning_rate": 4.838884649808458e-06,
+      "loss": 0.2126,
+      "mean_token_accuracy": 0.03923964609930408,
+      "step": 579
+    },
+    {
+      "epoch": 2.063921873266008,
+      "grad_norm": 0.8506307601928711,
+      "learning_rate": 4.82488012139502e-06,
+      "loss": 0.2073,
+      "mean_token_accuracy": 0.03836632040110999,
+      "step": 580
+    },
+    {
+      "epoch": 2.067473088447453,
+      "grad_norm": 0.7885401844978333,
+      "learning_rate": 4.810876968364679e-06,
+      "loss": 0.2026,
+      "mean_token_accuracy": 0.036981942706916016,
+      "step": 581
+    },
+    {
+      "epoch": 2.071024303628898,
+      "grad_norm": 1.110027551651001,
+      "learning_rate": 4.796875300697532e-06,
+      "loss": 0.2124,
+      "mean_token_accuracy": 0.03656144966953434,
+      "step": 582
+    },
+    {
+      "epoch": 2.074575518810343,
+      "grad_norm": 1.0753891468048096,
+      "learning_rate": 4.782875228361994e-06,
+      "loss": 0.201,
+      "mean_token_accuracy": 0.03901127418066608,
+      "step": 583
+    },
+    {
+      "epoch": 2.078126733991788,
+      "grad_norm": 0.9426742792129517,
+      "learning_rate": 4.7688768613139655e-06,
+      "loss": 0.2071,
+      "mean_token_accuracy": 0.038467124963062815,
+      "step": 584
+    },
+    {
+      "epoch": 2.081677949173233,
+      "grad_norm": 0.9639796614646912,
+      "learning_rate": 4.754880309495946e-06,
+      "loss": 0.204,
+      "mean_token_accuracy": 0.03624466506698809,
+      "step": 585
+    },
+    {
+      "epoch": 2.0852291643546774,
+      "grad_norm": 1.00525963306427,
+      "learning_rate": 4.74088568283618e-06,
+      "loss": 0.2041,
+      "mean_token_accuracy": 0.0362434946000576,
+      "step": 586
+    },
+    {
+      "epoch": 2.0887803795361224,
+      "grad_norm": 0.9281271696090698,
+      "learning_rate": 4.726893091247792e-06,
+      "loss": 0.192,
+      "mean_token_accuracy": 0.03966748606035253,
+      "step": 587
+    },
+    {
+      "epoch": 2.0923315947175674,
+      "grad_norm": 1.0103939771652222,
+      "learning_rate": 4.712902644627923e-06,
+      "loss": 0.1977,
+      "mean_token_accuracy": 0.04393619080656208,
+      "step": 588
+    },
+    {
+      "epoch": 2.0958828098990123,
+      "grad_norm": 0.937879204750061,
+      "learning_rate": 4.698914452856866e-06,
+      "loss": 0.201,
+      "mean_token_accuracy": 0.03757725107789156,
+      "step": 589
+    },
+    {
+      "epoch": 2.0994340250804573,
+      "grad_norm": 2.00667667388916,
+      "learning_rate": 4.684928625797208e-06,
+      "loss": 0.2183,
+      "mean_token_accuracy": 0.04129169932093646,
+      "step": 590
+    },
+    {
+      "epoch": 2.1029852402619023,
+      "grad_norm": 1.2946336269378662,
+      "learning_rate": 4.6709452732929614e-06,
+      "loss": 0.2017,
+      "mean_token_accuracy": 0.0390139233568334,
+      "step": 591
+    },
+    {
+      "epoch": 2.1065364554433472,
+      "grad_norm": 1.0413742065429688,
+      "learning_rate": 4.656964505168703e-06,
+      "loss": 0.2057,
+      "mean_token_accuracy": 0.03321857753326185,
+      "step": 592
+    },
+    {
+      "epoch": 2.1100876706247917,
+      "grad_norm": 1.1079062223434448,
+      "learning_rate": 4.642986431228713e-06,
+      "loss": 0.2084,
+      "mean_token_accuracy": 0.039933988635311835,
+      "step": 593
+    },
+    {
+      "epoch": 2.1136388858062367,
+      "grad_norm": 1.2978788614273071,
+      "learning_rate": 4.629011161256114e-06,
+      "loss": 0.2116,
+      "mean_token_accuracy": 0.03421423670442891,
+      "step": 594
+    },
+    {
+      "epoch": 2.1171901009876817,
+      "grad_norm": 0.9687265753746033,
+      "learning_rate": 4.615038805011999e-06,
+      "loss": 0.1959,
+      "mean_token_accuracy": 0.041332302451337455,
+      "step": 595
+    },
+    {
+      "epoch": 2.1207413161691266,
+      "grad_norm": 0.7898393273353577,
+      "learning_rate": 4.601069472234584e-06,
+      "loss": 0.2053,
+      "mean_token_accuracy": 0.036602654870875995,
+      "step": 596
+    },
+    {
+      "epoch": 2.1242925313505716,
+      "grad_norm": 0.8681390285491943,
+      "learning_rate": 4.587103272638339e-06,
+      "loss": 0.2209,
+      "mean_token_accuracy": 0.03851122210471658,
+      "step": 597
+    },
+    {
+      "epoch": 2.1278437465320166,
+      "grad_norm": 0.8869109153747559,
+      "learning_rate": 4.57314031591312e-06,
+      "loss": 0.2046,
+      "mean_token_accuracy": 0.0369587852510449,
+      "step": 598
+    },
+    {
+      "epoch": 2.131394961713461,
+      "grad_norm": 0.8230946063995361,
+      "learning_rate": 4.559180711723318e-06,
+      "loss": 0.1999,
+      "mean_token_accuracy": 0.03601818616698438,
+      "step": 599
+    },
+    {
+      "epoch": 2.134946176894906,
+      "grad_norm": 0.8362439870834351,
+      "learning_rate": 4.545224569706994e-06,
+      "loss": 0.2004,
+      "mean_token_accuracy": 0.04065158122466528,
+      "step": 600
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1124,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 300,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.297489199477934e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}