diff --git "a/checkpoint-1015/trainer_state.json" "b/checkpoint-1015/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1015/trainer_state.json"
@@ -0,0 +1,7210 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.995079950799508,
+  "eval_steps": 102,
+  "global_step": 1015,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004920049200492005,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.2408,
+      "step": 1
+    },
+    {
+      "epoch": 0.00984009840098401,
+      "grad_norm": 2.8125,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.2626,
+      "step": 2
+    },
+    {
+      "epoch": 0.014760147601476014,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-06,
+      "loss": 0.2438,
+      "step": 3
+    },
+    {
+      "epoch": 0.01968019680196802,
+      "grad_norm": 2.375,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.2335,
+      "step": 4
+    },
+    {
+      "epoch": 0.024600246002460024,
+      "grad_norm": 2.453125,
+      "learning_rate": 5e-06,
+      "loss": 0.229,
+      "step": 5
+    },
+    {
+      "epoch": 0.02952029520295203,
+      "grad_norm": 1.78125,
+      "learning_rate": 6e-06,
+      "loss": 0.2125,
+      "step": 6
+    },
+    {
+      "epoch": 0.03444034440344403,
+      "grad_norm": 1.4765625,
+      "learning_rate": 7e-06,
+      "loss": 0.1421,
+      "step": 7
+    },
+    {
+      "epoch": 0.03936039360393604,
+      "grad_norm": 1.53125,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.1727,
+      "step": 8
+    },
+    {
+      "epoch": 0.04428044280442804,
+      "grad_norm": 1.4296875,
+      "learning_rate": 9e-06,
+      "loss": 0.1599,
+      "step": 9
+    },
+    {
+      "epoch": 0.04920049200492005,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1e-05,
+      "loss": 0.2048,
+      "step": 10
+    },
+    {
+      "epoch": 0.05412054120541206,
+      "grad_norm": 2.125,
+      "learning_rate": 9.99997557091071e-06,
+      "loss": 0.1458,
+      "step": 11
+    },
+    {
+      "epoch": 0.05904059040590406,
+      "grad_norm": 1.6015625,
+      "learning_rate": 9.99990228388155e-06,
+      "loss": 0.1435,
+      "step": 12
+    },
+    {
+      "epoch": 0.06396063960639606,
+      "grad_norm": 1.2578125,
+      "learning_rate": 9.999780139628657e-06,
+      "loss": 0.1791,
+      "step": 13
+    },
+    {
+      "epoch": 0.06888068880688807,
+      "grad_norm": 1.375,
+      "learning_rate": 9.999609139345578e-06,
+      "loss": 0.1412,
+      "step": 14
+    },
+    {
+      "epoch": 0.07380073800738007,
+      "grad_norm": 1.0859375,
+      "learning_rate": 9.999389284703265e-06,
+      "loss": 0.1416,
+      "step": 15
+    },
+    {
+      "epoch": 0.07872078720787208,
+      "grad_norm": 1.21875,
+      "learning_rate": 9.99912057785006e-06,
+      "loss": 0.1255,
+      "step": 16
+    },
+    {
+      "epoch": 0.08364083640836409,
+      "grad_norm": 0.9140625,
+      "learning_rate": 9.998803021411668e-06,
+      "loss": 0.1201,
+      "step": 17
+    },
+    {
+      "epoch": 0.08856088560885608,
+      "grad_norm": 1.6875,
+      "learning_rate": 9.99843661849113e-06,
+      "loss": 0.2039,
+      "step": 18
+    },
+    {
+      "epoch": 0.09348093480934809,
+      "grad_norm": 0.96875,
+      "learning_rate": 9.998021372668809e-06,
+      "loss": 0.1352,
+      "step": 19
+    },
+    {
+      "epoch": 0.0984009840098401,
+      "grad_norm": 0.9296875,
+      "learning_rate": 9.99755728800233e-06,
+      "loss": 0.1448,
+      "step": 20
+    },
+    {
+      "epoch": 0.1033210332103321,
+      "grad_norm": 1.109375,
+      "learning_rate": 9.997044369026563e-06,
+      "loss": 0.1537,
+      "step": 21
+    },
+    {
+      "epoch": 0.10824108241082411,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.996482620753565e-06,
+      "loss": 0.1506,
+      "step": 22
+    },
+    {
+      "epoch": 0.11316113161131611,
+      "grad_norm": 1.109375,
+      "learning_rate": 9.995872048672535e-06,
+      "loss": 0.1474,
+      "step": 23
+    },
+    {
+      "epoch": 0.11808118081180811,
+      "grad_norm": 0.8671875,
+      "learning_rate": 9.99521265874976e-06,
+      "loss": 0.1215,
+      "step": 24
+    },
+    {
+      "epoch": 0.12300123001230012,
+      "grad_norm": 1.0859375,
+      "learning_rate": 9.994504457428557e-06,
+      "loss": 0.126,
+      "step": 25
+    },
+    {
+      "epoch": 0.12792127921279212,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.993747451629217e-06,
+      "loss": 0.1417,
+      "step": 26
+    },
+    {
+      "epoch": 0.13284132841328414,
+      "grad_norm": 1.15625,
+      "learning_rate": 9.992941648748917e-06,
+      "loss": 0.1163,
+      "step": 27
+    },
+    {
+      "epoch": 0.13776137761377613,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.992087056661676e-06,
+      "loss": 0.1332,
+      "step": 28
+    },
+    {
+      "epoch": 0.14268142681426815,
+      "grad_norm": 0.8671875,
+      "learning_rate": 9.991183683718253e-06,
+      "loss": 0.1269,
+      "step": 29
+    },
+    {
+      "epoch": 0.14760147601476015,
+      "grad_norm": 1.0,
+      "learning_rate": 9.99023153874608e-06,
+      "loss": 0.1071,
+      "step": 30
+    },
+    {
+      "epoch": 0.15252152521525214,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.989230631049171e-06,
+      "loss": 0.1415,
+      "step": 31
+    },
+    {
+      "epoch": 0.15744157441574416,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.988180970408032e-06,
+      "loss": 0.1439,
+      "step": 32
+    },
+    {
+      "epoch": 0.16236162361623616,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.987082567079563e-06,
+      "loss": 0.1333,
+      "step": 33
+    },
+    {
+      "epoch": 0.16728167281672818,
+      "grad_norm": 0.9296875,
+      "learning_rate": 9.985935431796962e-06,
+      "loss": 0.134,
+      "step": 34
+    },
+    {
+      "epoch": 0.17220172201722017,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.984739575769619e-06,
+      "loss": 0.1315,
+      "step": 35
+    },
+    {
+      "epoch": 0.17712177121771217,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.983495010683e-06,
+      "loss": 0.1258,
+      "step": 36
+    },
+    {
+      "epoch": 0.1820418204182042,
+      "grad_norm": 0.87890625,
+      "learning_rate": 9.982201748698543e-06,
+      "loss": 0.1141,
+      "step": 37
+    },
+    {
+      "epoch": 0.18696186961869618,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.980859802453533e-06,
+      "loss": 0.1437,
+      "step": 38
+    },
+    {
+      "epoch": 0.1918819188191882,
+      "grad_norm": 0.95703125,
+      "learning_rate": 9.97946918506098e-06,
+      "loss": 0.1326,
+      "step": 39
+    },
+    {
+      "epoch": 0.1968019680196802,
+      "grad_norm": 0.87109375,
+      "learning_rate": 9.978029910109491e-06,
+      "loss": 0.0979,
+      "step": 40
+    },
+    {
+      "epoch": 0.2017220172201722,
+      "grad_norm": 0.8203125,
+      "learning_rate": 9.976541991663136e-06,
+      "loss": 0.1086,
+      "step": 41
+    },
+    {
+      "epoch": 0.2066420664206642,
+      "grad_norm": 1.2265625,
+      "learning_rate": 9.975005444261312e-06,
+      "loss": 0.1504,
+      "step": 42
+    },
+    {
+      "epoch": 0.2115621156211562,
+      "grad_norm": 0.88671875,
+      "learning_rate": 9.973420282918603e-06,
+      "loss": 0.1282,
+      "step": 43
+    },
+    {
+      "epoch": 0.21648216482164823,
+      "grad_norm": 0.75,
+      "learning_rate": 9.971786523124623e-06,
+      "loss": 0.0807,
+      "step": 44
+    },
+    {
+      "epoch": 0.22140221402214022,
+      "grad_norm": 0.77734375,
+      "learning_rate": 9.970104180843881e-06,
+      "loss": 0.1084,
+      "step": 45
+    },
+    {
+      "epoch": 0.22632226322263221,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.968373272515612e-06,
+      "loss": 0.1385,
+      "step": 46
+    },
+    {
+      "epoch": 0.23124231242312424,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.966593815053623e-06,
+      "loss": 0.1538,
+      "step": 47
+    },
+    {
+      "epoch": 0.23616236162361623,
+      "grad_norm": 0.82421875,
+      "learning_rate": 9.964765825846123e-06,
+      "loss": 0.1035,
+      "step": 48
+    },
+    {
+      "epoch": 0.24108241082410825,
+      "grad_norm": 0.96484375,
+      "learning_rate": 9.962889322755556e-06,
+      "loss": 0.1055,
+      "step": 49
+    },
+    {
+      "epoch": 0.24600246002460024,
+      "grad_norm": 0.7734375,
+      "learning_rate": 9.960964324118428e-06,
+      "loss": 0.108,
+      "step": 50
+    },
+    {
+      "epoch": 0.25092250922509224,
+      "grad_norm": 0.94921875,
+      "learning_rate": 9.958990848745123e-06,
+      "loss": 0.1407,
+      "step": 51
+    },
+    {
+      "epoch": 0.25584255842558423,
+      "grad_norm": 0.890625,
+      "learning_rate": 9.956968915919725e-06,
+      "loss": 0.1281,
+      "step": 52
+    },
+    {
+      "epoch": 0.2607626076260763,
+      "grad_norm": 1.2109375,
+      "learning_rate": 9.954898545399822e-06,
+      "loss": 0.1094,
+      "step": 53
+    },
+    {
+      "epoch": 0.2656826568265683,
+      "grad_norm": 0.8984375,
+      "learning_rate": 9.952779757416326e-06,
+      "loss": 0.1341,
+      "step": 54
+    },
+    {
+      "epoch": 0.27060270602706027,
+      "grad_norm": 1.1484375,
+      "learning_rate": 9.950612572673255e-06,
+      "loss": 0.1497,
+      "step": 55
+    },
+    {
+      "epoch": 0.27552275522755226,
+      "grad_norm": 1.0859375,
+      "learning_rate": 9.948397012347553e-06,
+      "loss": 0.146,
+      "step": 56
+    },
+    {
+      "epoch": 0.28044280442804426,
+      "grad_norm": 0.94140625,
+      "learning_rate": 9.946133098088868e-06,
+      "loss": 0.1294,
+      "step": 57
+    },
+    {
+      "epoch": 0.2853628536285363,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.943820852019345e-06,
+      "loss": 0.1228,
+      "step": 58
+    },
+    {
+      "epoch": 0.2902829028290283,
+      "grad_norm": 0.890625,
+      "learning_rate": 9.941460296733409e-06,
+      "loss": 0.0948,
+      "step": 59
+    },
+    {
+      "epoch": 0.2952029520295203,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.939051455297548e-06,
+      "loss": 0.1337,
+      "step": 60
+    },
+    {
+      "epoch": 0.3001230012300123,
+      "grad_norm": 0.90234375,
+      "learning_rate": 9.936594351250082e-06,
+      "loss": 0.1267,
+      "step": 61
+    },
+    {
+      "epoch": 0.3050430504305043,
+      "grad_norm": 0.6640625,
+      "learning_rate": 9.934089008600937e-06,
+      "loss": 0.0852,
+      "step": 62
+    },
+    {
+      "epoch": 0.30996309963099633,
+      "grad_norm": 0.89453125,
+      "learning_rate": 9.931535451831408e-06,
+      "loss": 0.0853,
+      "step": 63
+    },
+    {
+      "epoch": 0.3148831488314883,
+      "grad_norm": 1.2578125,
+      "learning_rate": 9.928933705893924e-06,
+      "loss": 0.104,
+      "step": 64
+    },
+    {
+      "epoch": 0.3198031980319803,
+      "grad_norm": 0.96484375,
+      "learning_rate": 9.926283796211796e-06,
+      "loss": 0.0999,
+      "step": 65
+    },
+    {
+      "epoch": 0.3247232472324723,
+      "grad_norm": 0.83203125,
+      "learning_rate": 9.923585748678976e-06,
+      "loss": 0.1169,
+      "step": 66
+    },
+    {
+      "epoch": 0.3296432964329643,
+      "grad_norm": 0.984375,
+      "learning_rate": 9.920839589659803e-06,
+      "loss": 0.1413,
+      "step": 67
+    },
+    {
+      "epoch": 0.33456334563345635,
+      "grad_norm": 0.90625,
+      "learning_rate": 9.918045345988743e-06,
+      "loss": 0.0995,
+      "step": 68
+    },
+    {
+      "epoch": 0.33948339483394835,
+      "grad_norm": 0.95703125,
+      "learning_rate": 9.915203044970125e-06,
+      "loss": 0.1242,
+      "step": 69
+    },
+    {
+      "epoch": 0.34440344403444034,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.91231271437788e-06,
+      "loss": 0.0816,
+      "step": 70
+    },
+    {
+      "epoch": 0.34932349323493234,
+      "grad_norm": 0.96875,
+      "learning_rate": 9.909374382455268e-06,
+      "loss": 0.1365,
+      "step": 71
+    },
+    {
+      "epoch": 0.35424354243542433,
+      "grad_norm": 0.91796875,
+      "learning_rate": 9.906388077914598e-06,
+      "loss": 0.1388,
+      "step": 72
+    },
+    {
+      "epoch": 0.3591635916359164,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.903353829936943e-06,
+      "loss": 0.1291,
+      "step": 73
+    },
+    {
+      "epoch": 0.3640836408364084,
+      "grad_norm": 0.8828125,
+      "learning_rate": 9.900271668171879e-06,
+      "loss": 0.1014,
+      "step": 74
+    },
+    {
+      "epoch": 0.36900369003690037,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.89714162273716e-06,
+      "loss": 0.1653,
+      "step": 75
+    },
+    {
+      "epoch": 0.37392373923739236,
+      "grad_norm": 0.76953125,
+      "learning_rate": 9.893963724218456e-06,
+      "loss": 0.1097,
+      "step": 76
+    },
+    {
+      "epoch": 0.37884378843788435,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.890738003669029e-06,
+      "loss": 0.1509,
+      "step": 77
+    },
+    {
+      "epoch": 0.3837638376383764,
+      "grad_norm": 1.15625,
+      "learning_rate": 9.887464492609447e-06,
+      "loss": 0.119,
+      "step": 78
+    },
+    {
+      "epoch": 0.3886838868388684,
+      "grad_norm": 0.90234375,
+      "learning_rate": 9.884143223027267e-06,
+      "loss": 0.1354,
+      "step": 79
+    },
+    {
+      "epoch": 0.3936039360393604,
+      "grad_norm": 0.84375,
+      "learning_rate": 9.880774227376727e-06,
+      "loss": 0.0938,
+      "step": 80
+    },
+    {
+      "epoch": 0.3985239852398524,
+      "grad_norm": 0.890625,
+      "learning_rate": 9.877357538578426e-06,
+      "loss": 0.1072,
+      "step": 81
+    },
+    {
+      "epoch": 0.4034440344403444,
+      "grad_norm": 0.8125,
+      "learning_rate": 9.873893190018995e-06,
+      "loss": 0.099,
+      "step": 82
+    },
+    {
+      "epoch": 0.40836408364083643,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.870381215550796e-06,
+      "loss": 0.1036,
+      "step": 83
+    },
+    {
+      "epoch": 0.4132841328413284,
+      "grad_norm": 0.7734375,
+      "learning_rate": 9.866821649491558e-06,
+      "loss": 0.0779,
+      "step": 84
+    },
+    {
+      "epoch": 0.4182041820418204,
+      "grad_norm": 1.140625,
+      "learning_rate": 9.863214526624065e-06,
+      "loss": 0.1597,
+      "step": 85
+    },
+    {
+      "epoch": 0.4231242312423124,
+      "grad_norm": 0.96875,
+      "learning_rate": 9.85955988219581e-06,
+      "loss": 0.1283,
+      "step": 86
+    },
+    {
+      "epoch": 0.4280442804428044,
+      "grad_norm": 1.1171875,
+      "learning_rate": 9.855857751918644e-06,
+      "loss": 0.1327,
+      "step": 87
+    },
+    {
+      "epoch": 0.43296432964329645,
+      "grad_norm": 0.80859375,
+      "learning_rate": 9.852108171968438e-06,
+      "loss": 0.1,
+      "step": 88
+    },
+    {
+      "epoch": 0.43788437884378845,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.848311178984718e-06,
+      "loss": 0.1653,
+      "step": 89
+    },
+    {
+      "epoch": 0.44280442804428044,
+      "grad_norm": 0.69921875,
+      "learning_rate": 9.844466810070319e-06,
+      "loss": 0.0841,
+      "step": 90
+    },
+    {
+      "epoch": 0.44772447724477243,
+      "grad_norm": 0.8828125,
+      "learning_rate": 9.840575102791013e-06,
+      "loss": 0.1193,
+      "step": 91
+    },
+    {
+      "epoch": 0.45264452644526443,
+      "grad_norm": 0.9921875,
+      "learning_rate": 9.836636095175148e-06,
+      "loss": 0.1187,
+      "step": 92
+    },
+    {
+      "epoch": 0.4575645756457565,
+      "grad_norm": 0.95703125,
+      "learning_rate": 9.832649825713266e-06,
+      "loss": 0.1198,
+      "step": 93
+    },
+    {
+      "epoch": 0.46248462484624847,
+      "grad_norm": 0.96875,
+      "learning_rate": 9.828616333357743e-06,
+      "loss": 0.1148,
+      "step": 94
+    },
+    {
+      "epoch": 0.46740467404674046,
+      "grad_norm": 0.79296875,
+      "learning_rate": 9.8245356575224e-06,
+      "loss": 0.1074,
+      "step": 95
+    },
+    {
+      "epoch": 0.47232472324723246,
+      "grad_norm": 0.7109375,
+      "learning_rate": 9.820407838082109e-06,
+      "loss": 0.1193,
+      "step": 96
+    },
+    {
+      "epoch": 0.47724477244772445,
+      "grad_norm": 0.66796875,
+      "learning_rate": 9.816232915372423e-06,
+      "loss": 0.0722,
+      "step": 97
+    },
+    {
+      "epoch": 0.4821648216482165,
+      "grad_norm": 0.9765625,
+      "learning_rate": 9.812010930189163e-06,
+      "loss": 0.0929,
+      "step": 98
+    },
+    {
+      "epoch": 0.4870848708487085,
+      "grad_norm": 0.89453125,
+      "learning_rate": 9.80774192378803e-06,
+      "loss": 0.0973,
+      "step": 99
+    },
+    {
+      "epoch": 0.4920049200492005,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.803425937884202e-06,
+      "loss": 0.1335,
+      "step": 100
+    },
+    {
+      "epoch": 0.4969249692496925,
+      "grad_norm": 0.77734375,
+      "learning_rate": 9.799063014651918e-06,
+      "loss": 0.1169,
+      "step": 101
+    },
+    {
+      "epoch": 0.5018450184501845,
+      "grad_norm": 0.93359375,
+      "learning_rate": 9.794653196724077e-06,
+      "loss": 0.1159,
+      "step": 102
+    },
+    {
+      "epoch": 0.5018450184501845,
+      "eval_loss": 0.12004748731851578,
+      "eval_runtime": 203.4298,
+      "eval_samples_per_second": 8.042,
+      "eval_steps_per_second": 1.342,
+      "step": 102
+    },
+    {
+      "epoch": 0.5067650676506765,
+      "grad_norm": 0.9375,
+      "learning_rate": 9.790196527191811e-06,
+      "loss": 0.1142,
+      "step": 103
+    },
+    {
+      "epoch": 0.5116851168511685,
+      "grad_norm": 0.8203125,
+      "learning_rate": 9.785693049604073e-06,
+      "loss": 0.1113,
+      "step": 104
+    },
+    {
+      "epoch": 0.5166051660516605,
+      "grad_norm": 0.80859375,
+      "learning_rate": 9.781142807967205e-06,
+      "loss": 0.1035,
+      "step": 105
+    },
+    {
+      "epoch": 0.5215252152521526,
+      "grad_norm": 0.91796875,
+      "learning_rate": 9.776545846744509e-06,
+      "loss": 0.1238,
+      "step": 106
+    },
+    {
+      "epoch": 0.5264452644526445,
+      "grad_norm": 1.4765625,
+      "learning_rate": 9.771902210855819e-06,
+      "loss": 0.0921,
+      "step": 107
+    },
+    {
+      "epoch": 0.5313653136531366,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.767211945677051e-06,
+      "loss": 0.1459,
+      "step": 108
+    },
+    {
+      "epoch": 0.5362853628536285,
+      "grad_norm": 1.0,
+      "learning_rate": 9.762475097039767e-06,
+      "loss": 0.101,
+      "step": 109
+    },
+    {
+      "epoch": 0.5412054120541205,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.757691711230728e-06,
+      "loss": 0.1259,
+      "step": 110
+    },
+    {
+      "epoch": 0.5461254612546126,
+      "grad_norm": 0.8828125,
+      "learning_rate": 9.752861834991436e-06,
+      "loss": 0.1146,
+      "step": 111
+    },
+    {
+      "epoch": 0.5510455104551045,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.747985515517683e-06,
+      "loss": 0.1609,
+      "step": 112
+    },
+    {
+      "epoch": 0.5559655596555966,
+      "grad_norm": 0.94921875,
+      "learning_rate": 9.743062800459089e-06,
+      "loss": 0.1242,
+      "step": 113
+    },
+    {
+      "epoch": 0.5608856088560885,
+      "grad_norm": 0.9765625,
+      "learning_rate": 9.738093737918627e-06,
+      "loss": 0.1095,
+      "step": 114
+    },
+    {
+      "epoch": 0.5658056580565806,
+      "grad_norm": 1.0,
+      "learning_rate": 9.733078376452172e-06,
+      "loss": 0.1083,
+      "step": 115
+    },
+    {
+      "epoch": 0.5707257072570726,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.728016765068004e-06,
+      "loss": 0.1428,
+      "step": 116
+    },
+    {
+      "epoch": 0.5756457564575646,
+      "grad_norm": 0.859375,
+      "learning_rate": 9.722908953226349e-06,
+      "loss": 0.1073,
+      "step": 117
+    },
+    {
+      "epoch": 0.5805658056580566,
+      "grad_norm": 0.8125,
+      "learning_rate": 9.717754990838882e-06,
+      "loss": 0.1049,
+      "step": 118
+    },
+    {
+      "epoch": 0.5854858548585485,
+      "grad_norm": 1.140625,
+      "learning_rate": 9.712554928268245e-06,
+      "loss": 0.1479,
+      "step": 119
+    },
+    {
+      "epoch": 0.5904059040590406,
+      "grad_norm": 0.90234375,
+      "learning_rate": 9.707308816327557e-06,
+      "loss": 0.1195,
+      "step": 120
+    },
+    {
+      "epoch": 0.5953259532595326,
+      "grad_norm": 0.82421875,
+      "learning_rate": 9.702016706279914e-06,
+      "loss": 0.1048,
+      "step": 121
+    },
+    {
+      "epoch": 0.6002460024600246,
+      "grad_norm": 0.984375,
+      "learning_rate": 9.696678649837883e-06,
+      "loss": 0.1073,
+      "step": 122
+    },
+    {
+      "epoch": 0.6051660516605166,
+      "grad_norm": 1.0859375,
+      "learning_rate": 9.691294699163013e-06,
+      "loss": 0.1308,
+      "step": 123
+    },
+    {
+      "epoch": 0.6100861008610086,
+      "grad_norm": 0.86328125,
+      "learning_rate": 9.685864906865303e-06,
+      "loss": 0.1248,
+      "step": 124
+    },
+    {
+      "epoch": 0.6150061500615006,
+      "grad_norm": 0.8515625,
+      "learning_rate": 9.680389326002708e-06,
+      "loss": 0.1312,
+      "step": 125
+    },
+    {
+      "epoch": 0.6199261992619927,
+      "grad_norm": 0.71875,
+      "learning_rate": 9.67486801008061e-06,
+      "loss": 0.0868,
+      "step": 126
+    },
+    {
+      "epoch": 0.6248462484624846,
+      "grad_norm": 0.76953125,
+      "learning_rate": 9.669301013051297e-06,
+      "loss": 0.0859,
+      "step": 127
+    },
+    {
+      "epoch": 0.6297662976629766,
+      "grad_norm": 0.734375,
+      "learning_rate": 9.663688389313435e-06,
+      "loss": 0.0878,
+      "step": 128
+    },
+    {
+      "epoch": 0.6346863468634686,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.658030193711538e-06,
+      "loss": 0.1229,
+      "step": 129
+    },
+    {
+      "epoch": 0.6396063960639606,
+      "grad_norm": 0.97265625,
+      "learning_rate": 9.652326481535434e-06,
+      "loss": 0.1439,
+      "step": 130
+    },
+    {
+      "epoch": 0.6445264452644527,
+      "grad_norm": 0.703125,
+      "learning_rate": 9.646577308519719e-06,
+      "loss": 0.1063,
+      "step": 131
+    },
+    {
+      "epoch": 0.6494464944649446,
+      "grad_norm": 0.9609375,
+      "learning_rate": 9.640782730843219e-06,
+      "loss": 0.1023,
+      "step": 132
+    },
+    {
+      "epoch": 0.6543665436654367,
+      "grad_norm": 0.984375,
+      "learning_rate": 9.634942805128433e-06,
+      "loss": 0.1002,
+      "step": 133
+    },
+    {
+      "epoch": 0.6592865928659286,
+      "grad_norm": 0.76953125,
+      "learning_rate": 9.629057588440993e-06,
+      "loss": 0.0978,
+      "step": 134
+    },
+    {
+      "epoch": 0.6642066420664207,
+      "grad_norm": 0.8671875,
+      "learning_rate": 9.623127138289087e-06,
+      "loss": 0.1042,
+      "step": 135
+    },
+    {
+      "epoch": 0.6691266912669127,
+      "grad_norm": 0.8515625,
+      "learning_rate": 9.617151512622918e-06,
+      "loss": 0.0873,
+      "step": 136
+    },
+    {
+      "epoch": 0.6740467404674046,
+      "grad_norm": 0.91796875,
+      "learning_rate": 9.611130769834121e-06,
+      "loss": 0.1118,
+      "step": 137
+    },
+    {
+      "epoch": 0.6789667896678967,
+      "grad_norm": 0.9375,
+      "learning_rate": 9.6050649687552e-06,
+      "loss": 0.13,
+      "step": 138
+    },
+    {
+      "epoch": 0.6838868388683886,
+      "grad_norm": 0.828125,
+      "learning_rate": 9.598954168658956e-06,
+      "loss": 0.1096,
+      "step": 139
+    },
+    {
+      "epoch": 0.6888068880688807,
+      "grad_norm": 0.86328125,
+      "learning_rate": 9.592798429257899e-06,
+      "loss": 0.1219,
+      "step": 140
+    },
+    {
+      "epoch": 0.6937269372693727,
+      "grad_norm": 0.95703125,
+      "learning_rate": 9.586597810703674e-06,
+      "loss": 0.1002,
+      "step": 141
+    },
+    {
+      "epoch": 0.6986469864698647,
+      "grad_norm": 0.765625,
+      "learning_rate": 9.580352373586468e-06,
+      "loss": 0.0765,
+      "step": 142
+    },
+    {
+      "epoch": 0.7035670356703567,
+      "grad_norm": 0.90234375,
+      "learning_rate": 9.574062178934414e-06,
+      "loss": 0.1227,
+      "step": 143
+    },
+    {
+      "epoch": 0.7084870848708487,
+      "grad_norm": 0.82421875,
+      "learning_rate": 9.567727288213005e-06,
+      "loss": 0.1052,
+      "step": 144
+    },
+    {
+      "epoch": 0.7134071340713407,
+      "grad_norm": 0.87109375,
+      "learning_rate": 9.561347763324484e-06,
+      "loss": 0.0909,
+      "step": 145
+    },
+    {
+      "epoch": 0.7183271832718328,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.554923666607245e-06,
+      "loss": 0.1219,
+      "step": 146
+    },
+    {
+      "epoch": 0.7232472324723247,
+      "grad_norm": 0.98046875,
+      "learning_rate": 9.548455060835223e-06,
+      "loss": 0.1077,
+      "step": 147
+    },
+    {
+      "epoch": 0.7281672816728167,
+      "grad_norm": 0.94921875,
+      "learning_rate": 9.541942009217273e-06,
+      "loss": 0.132,
+      "step": 148
+    },
+    {
+      "epoch": 0.7330873308733087,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.535384575396565e-06,
+      "loss": 0.1462,
+      "step": 149
+    },
+    {
+      "epoch": 0.7380073800738007,
+      "grad_norm": 0.79296875,
+      "learning_rate": 9.528782823449954e-06,
+      "loss": 0.0951,
+      "step": 150
+    },
+    {
+      "epoch": 0.7429274292742928,
+      "grad_norm": 0.99609375,
+      "learning_rate": 9.522136817887353e-06,
+      "loss": 0.1451,
+      "step": 151
+    },
+    {
+      "epoch": 0.7478474784747847,
+      "grad_norm": 0.81640625,
+      "learning_rate": 9.51544662365111e-06,
+      "loss": 0.072,
+      "step": 152
+    },
+    {
+      "epoch": 0.7527675276752768,
+      "grad_norm": 0.80078125,
+      "learning_rate": 9.508712306115366e-06,
+      "loss": 0.1202,
+      "step": 153
+    },
+    {
+      "epoch": 0.7576875768757687,
+      "grad_norm": 0.890625,
+      "learning_rate": 9.501933931085416e-06,
+      "loss": 0.1068,
+      "step": 154
+    },
+    {
+      "epoch": 0.7626076260762608,
+      "grad_norm": 0.8984375,
+      "learning_rate": 9.495111564797073e-06,
+      "loss": 0.1091,
+      "step": 155
+    },
+    {
+      "epoch": 0.7675276752767528,
+      "grad_norm": 0.84375,
+      "learning_rate": 9.488245273916016e-06,
+      "loss": 0.0982,
+      "step": 156
+    },
+    {
+      "epoch": 0.7724477244772447,
+      "grad_norm": 0.828125,
+      "learning_rate": 9.481335125537138e-06,
+      "loss": 0.1064,
+      "step": 157
+    },
+    {
+      "epoch": 0.7773677736777368,
+      "grad_norm": 0.8828125,
+      "learning_rate": 9.47438118718389e-06,
+      "loss": 0.1027,
+      "step": 158
+    },
+    {
+      "epoch": 0.7822878228782287,
+      "grad_norm": 0.765625,
+      "learning_rate": 9.467383526807626e-06,
+      "loss": 0.0917,
+      "step": 159
+    },
+    {
+      "epoch": 0.7872078720787208,
+      "grad_norm": 0.875,
+      "learning_rate": 9.460342212786933e-06,
+      "loss": 0.0975,
+      "step": 160
+    },
+    {
+      "epoch": 0.7921279212792128,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.453257313926969e-06,
+      "loss": 0.0837,
+      "step": 161
+    },
+    {
+      "epoch": 0.7970479704797048,
+      "grad_norm": 0.68359375,
+      "learning_rate": 9.44612889945878e-06,
+      "loss": 0.093,
+      "step": 162
+    },
+    {
+      "epoch": 0.8019680196801968,
+      "grad_norm": 0.7578125,
+      "learning_rate": 9.43895703903864e-06,
+      "loss": 0.1094,
+      "step": 163
+    },
+    {
+      "epoch": 0.8068880688806888,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.431741802747355e-06,
+      "loss": 0.1081,
+      "step": 164
+    },
+    {
+      "epoch": 0.8118081180811808,
+      "grad_norm": 0.96875,
+      "learning_rate": 9.424483261089584e-06,
+      "loss": 0.1207,
+      "step": 165
+    },
+    {
+      "epoch": 0.8167281672816729,
+      "grad_norm": 0.67578125,
+      "learning_rate": 9.417181484993154e-06,
+      "loss": 0.0717,
+      "step": 166
+    },
+    {
+      "epoch": 0.8216482164821648,
+      "grad_norm": 0.99609375,
+      "learning_rate": 9.40983654580836e-06,
+      "loss": 0.1013,
+      "step": 167
+    },
+    {
+      "epoch": 0.8265682656826568,
+      "grad_norm": 0.87890625,
+      "learning_rate": 9.402448515307274e-06,
+      "loss": 0.1122,
+      "step": 168
+    },
+    {
+      "epoch": 0.8314883148831488,
+      "grad_norm": 0.99609375,
+      "learning_rate": 9.395017465683036e-06,
+      "loss": 0.1234,
+      "step": 169
+    },
+    {
+      "epoch": 0.8364083640836408,
+      "grad_norm": 0.85546875,
+      "learning_rate": 9.387543469549156e-06,
+      "loss": 0.1259,
+      "step": 170
+    },
+    {
+      "epoch": 0.8413284132841329,
+      "grad_norm": 0.78515625,
+      "learning_rate": 9.380026599938804e-06,
+      "loss": 0.1007,
+      "step": 171
+    },
+    {
+      "epoch": 0.8462484624846248,
+      "grad_norm": 0.95703125,
+      "learning_rate": 9.372466930304091e-06,
+      "loss": 0.1148,
+      "step": 172
+    },
+    {
+      "epoch": 0.8511685116851169,
+      "grad_norm": 0.765625,
+      "learning_rate": 9.364864534515353e-06,
+      "loss": 0.11,
+      "step": 173
+    },
+    {
+      "epoch": 0.8560885608856088,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.357219486860434e-06,
+      "loss": 0.1255,
+      "step": 174
+    },
+    {
+      "epoch": 0.8610086100861009,
+      "grad_norm": 0.8671875,
+      "learning_rate": 9.349531862043952e-06,
+      "loss": 0.0973,
+      "step": 175
+    },
+    {
+      "epoch": 0.8659286592865929,
+      "grad_norm": 0.8671875,
+      "learning_rate": 9.34180173518658e-06,
+      "loss": 0.0872,
+      "step": 176
+    },
+    {
+      "epoch": 0.8708487084870848,
+      "grad_norm": 0.796875,
+      "learning_rate": 9.3340291818243e-06,
+      "loss": 0.089,
+      "step": 177
+    },
+    {
+      "epoch": 0.8757687576875769,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.32621427790767e-06,
+      "loss": 0.1148,
+      "step": 178
+    },
+    {
+      "epoch": 0.8806888068880688,
+      "grad_norm": 0.78515625,
+      "learning_rate": 9.318357099801087e-06,
+      "loss": 0.0896,
+      "step": 179
+    },
+    {
+      "epoch": 0.8856088560885609,
+      "grad_norm": 0.9296875,
+      "learning_rate": 9.310457724282034e-06,
+      "loss": 0.1104,
+      "step": 180
+    },
+    {
+      "epoch": 0.8905289052890529,
+      "grad_norm": 0.859375,
+      "learning_rate": 9.302516228540328e-06,
+      "loss": 0.1115,
+      "step": 181
+    },
+    {
+      "epoch": 0.8954489544895449,
+      "grad_norm": 0.921875,
+      "learning_rate": 9.294532690177373e-06,
+      "loss": 0.12,
+      "step": 182
+    },
+    {
+      "epoch": 0.9003690036900369,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.286507187205399e-06,
+      "loss": 0.137,
+      "step": 183
+    },
+    {
+      "epoch": 0.9052890528905289,
+      "grad_norm": 0.80859375,
+      "learning_rate": 9.278439798046697e-06,
+      "loss": 0.1273,
+      "step": 184
+    },
+    {
+      "epoch": 0.9102091020910209,
+      "grad_norm": 0.96484375,
+      "learning_rate": 9.270330601532855e-06,
+      "loss": 0.1339,
+      "step": 185
+    },
+    {
+      "epoch": 0.915129151291513,
+      "grad_norm": 0.90234375,
+      "learning_rate": 9.262179676903986e-06,
+      "loss": 0.1011,
+      "step": 186
+    },
+    {
+      "epoch": 0.9200492004920049,
+      "grad_norm": 0.90234375,
+      "learning_rate": 9.25398710380796e-06,
+      "loss": 0.1096,
+      "step": 187
+    },
+    {
+      "epoch": 0.9249692496924969,
+      "grad_norm": 0.8125,
+      "learning_rate": 9.245752962299612e-06,
+      "loss": 0.1316,
+      "step": 188
+    },
+    {
+      "epoch": 0.9298892988929889,
+      "grad_norm": 0.93359375,
+      "learning_rate": 9.237477332839975e-06,
+      "loss": 0.0908,
+      "step": 189
+    },
+    {
+      "epoch": 0.9348093480934809,
+      "grad_norm": 0.97265625,
+      "learning_rate": 9.229160296295488e-06,
+      "loss": 0.138,
+      "step": 190
+    },
+    {
+      "epoch": 0.939729397293973,
+      "grad_norm": 0.81640625,
+      "learning_rate": 9.220801933937199e-06,
+      "loss": 0.1259,
+      "step": 191
+    },
+    {
+      "epoch": 0.9446494464944649,
+      "grad_norm": 0.92578125,
+      "learning_rate": 9.212402327439982e-06,
+      "loss": 0.117,
+      "step": 192
+    },
+    {
+      "epoch": 0.949569495694957,
+      "grad_norm": 1.0859375,
+      "learning_rate": 9.20396155888173e-06,
+      "loss": 0.1405,
+      "step": 193
+    },
+    {
+      "epoch": 0.9544895448954489,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.19547971074256e-06,
+      "loss": 0.1154,
+      "step": 194
+    },
+    {
+      "epoch": 0.959409594095941,
+      "grad_norm": 0.8828125,
+      "learning_rate": 9.186956865904004e-06,
+      "loss": 0.1203,
+      "step": 195
+    },
+    {
+      "epoch": 0.964329643296433,
+      "grad_norm": 0.71875,
+      "learning_rate": 9.178393107648193e-06,
+      "loss": 0.1009,
+      "step": 196
+    },
+    {
+      "epoch": 0.9692496924969249,
+      "grad_norm": 0.859375,
+      "learning_rate": 9.169788519657056e-06,
+      "loss": 0.1023,
+      "step": 197
+    },
+    {
+      "epoch": 0.974169741697417,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.161143186011492e-06,
+      "loss": 0.1402,
+      "step": 198
+    },
+    {
+      "epoch": 0.9790897908979089,
+      "grad_norm": 0.828125,
+      "learning_rate": 9.15245719119055e-06,
+      "loss": 0.1065,
+      "step": 199
+    },
+    {
+      "epoch": 0.984009840098401,
+      "grad_norm": 0.8515625,
+      "learning_rate": 9.143730620070609e-06,
+      "loss": 0.097,
+      "step": 200
+    },
+    {
+      "epoch": 0.988929889298893,
+      "grad_norm": 0.87109375,
+      "learning_rate": 9.134963557924543e-06,
+      "loss": 0.0979,
+      "step": 201
+    },
+    {
+      "epoch": 0.993849938499385,
+      "grad_norm": 0.87109375,
+      "learning_rate": 9.12615609042089e-06,
+      "loss": 0.1012,
+      "step": 202
+    },
+    {
+      "epoch": 0.998769987699877,
+      "grad_norm": 0.7578125,
+      "learning_rate": 9.11730830362301e-06,
+      "loss": 0.0764,
+      "step": 203
+    },
+    {
+      "epoch": 1.0030750307503076,
+      "grad_norm": 1.9296875,
+      "learning_rate": 9.10842028398826e-06,
+      "loss": 0.1883,
+      "step": 204
+    },
+    {
+      "epoch": 1.0030750307503076,
+      "eval_loss": 0.11158129572868347,
+      "eval_runtime": 204.933,
+      "eval_samples_per_second": 7.983,
+      "eval_steps_per_second": 1.332,
+      "step": 204
+    },
+    {
+      "epoch": 1.0079950799507995,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.099492118367123e-06,
+      "loss": 0.1227,
+      "step": 205
+    },
+    {
+      "epoch": 1.0129151291512914,
+      "grad_norm": 0.80078125,
+      "learning_rate": 9.090523894002386e-06,
+      "loss": 0.0759,
+      "step": 206
+    },
+    {
+      "epoch": 1.0178351783517836,
+      "grad_norm": 1.234375,
+      "learning_rate": 9.081515698528267e-06,
+      "loss": 0.1026,
+      "step": 207
+    },
+    {
+      "epoch": 1.0227552275522755,
+      "grad_norm": 1.296875,
+      "learning_rate": 9.072467619969574e-06,
+      "loss": 0.1144,
+      "step": 208
+    },
+    {
+      "epoch": 1.0276752767527675,
+      "grad_norm": 1.2578125,
+      "learning_rate": 9.06337974674083e-06,
+      "loss": 0.1129,
+      "step": 209
+    },
+    {
+      "epoch": 1.0325953259532596,
+      "grad_norm": 1.5078125,
+      "learning_rate": 9.054252167645426e-06,
+      "loss": 0.081,
+      "step": 210
+    },
+    {
+      "epoch": 1.0375153751537516,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.045084971874738e-06,
+      "loss": 0.1211,
+      "step": 211
+    },
+    {
+      "epoch": 1.0424354243542435,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.035878249007264e-06,
+      "loss": 0.0857,
+      "step": 212
+    },
+    {
+      "epoch": 1.0473554735547355,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.026632089007745e-06,
+      "loss": 0.1407,
+      "step": 213
+    },
+    {
+      "epoch": 1.0522755227552276,
+      "grad_norm": 1.1484375,
+      "learning_rate": 9.017346582226289e-06,
+      "loss": 0.0835,
+      "step": 214
+    },
+    {
+      "epoch": 1.0571955719557196,
+      "grad_norm": 1.21875,
+      "learning_rate": 9.008021819397488e-06,
+      "loss": 0.1263,
+      "step": 215
+    },
+    {
+      "epoch": 1.0621156211562115,
+      "grad_norm": 1.2109375,
+      "learning_rate": 8.998657891639523e-06,
+      "loss": 0.0747,
+      "step": 216
+    },
+    {
+      "epoch": 1.0670356703567037,
+      "grad_norm": 0.89453125,
+      "learning_rate": 8.989254890453289e-06,
+      "loss": 0.09,
+      "step": 217
+    },
+    {
+      "epoch": 1.0719557195571956,
+      "grad_norm": 1.0,
+      "learning_rate": 8.979812907721485e-06,
+      "loss": 0.0924,
+      "step": 218
+    },
+    {
+      "epoch": 1.0768757687576875,
+      "grad_norm": 1.1171875,
+      "learning_rate": 8.97033203570773e-06,
+      "loss": 0.1017,
+      "step": 219
+    },
+    {
+      "epoch": 1.0817958179581795,
+      "grad_norm": 0.9921875,
+      "learning_rate": 8.960812367055646e-06,
+      "loss": 0.1033,
+      "step": 220
+    },
+    {
+      "epoch": 1.0867158671586716,
+      "grad_norm": 1.3515625,
+      "learning_rate": 8.951253994787975e-06,
+      "loss": 0.1425,
+      "step": 221
+    },
+    {
+      "epoch": 1.0916359163591636,
+      "grad_norm": 1.140625,
+      "learning_rate": 8.941657012305644e-06,
+      "loss": 0.1076,
+      "step": 222
+    },
+    {
+      "epoch": 1.0965559655596555,
+      "grad_norm": 0.9296875,
+      "learning_rate": 8.93202151338687e-06,
+      "loss": 0.081,
+      "step": 223
+    },
+    {
+      "epoch": 1.1014760147601477,
+      "grad_norm": 0.90234375,
+      "learning_rate": 8.922347592186237e-06,
+      "loss": 0.096,
+      "step": 224
+    },
+    {
+      "epoch": 1.1063960639606396,
+      "grad_norm": 0.79296875,
+      "learning_rate": 8.912635343233784e-06,
+      "loss": 0.0681,
+      "step": 225
+    },
+    {
+      "epoch": 1.1113161131611315,
+      "grad_norm": 1.28125,
+      "learning_rate": 8.902884861434066e-06,
+      "loss": 0.117,
+      "step": 226
+    },
+    {
+      "epoch": 1.1162361623616237,
+      "grad_norm": 1.2734375,
+      "learning_rate": 8.89309624206524e-06,
+      "loss": 0.1133,
+      "step": 227
+    },
+    {
+      "epoch": 1.1211562115621156,
+      "grad_norm": 0.96484375,
+      "learning_rate": 8.883269580778127e-06,
+      "loss": 0.0933,
+      "step": 228
+    },
+    {
+      "epoch": 1.1260762607626076,
+      "grad_norm": 1.2265625,
+      "learning_rate": 8.873404973595284e-06,
+      "loss": 0.1243,
+      "step": 229
+    },
+    {
+      "epoch": 1.1309963099630997,
+      "grad_norm": 1.6796875,
+      "learning_rate": 8.863502516910058e-06,
+      "loss": 0.0959,
+      "step": 230
+    },
+    {
+      "epoch": 1.1359163591635917,
+      "grad_norm": 1.0390625,
+      "learning_rate": 8.853562307485649e-06,
+      "loss": 0.1095,
+      "step": 231
+    },
+    {
+      "epoch": 1.1408364083640836,
+      "grad_norm": 1.3984375,
+      "learning_rate": 8.84358444245416e-06,
+      "loss": 0.1128,
+      "step": 232
+    },
+    {
+      "epoch": 1.1457564575645756,
+      "grad_norm": 1.046875,
+      "learning_rate": 8.833569019315654e-06,
+      "loss": 0.098,
+      "step": 233
+    },
+    {
+      "epoch": 1.1506765067650677,
+      "grad_norm": 1.6328125,
+      "learning_rate": 8.8235161359372e-06,
+      "loss": 0.1001,
+      "step": 234
+    },
+    {
+      "epoch": 1.1555965559655597,
+      "grad_norm": 1.1796875,
+      "learning_rate": 8.81342589055191e-06,
+      "loss": 0.1213,
+      "step": 235
+    },
+    {
+      "epoch": 1.1605166051660516,
+      "grad_norm": 1.078125,
+      "learning_rate": 8.803298381757987e-06,
+      "loss": 0.0891,
+      "step": 236
+    },
+    {
+      "epoch": 1.1654366543665438,
+      "grad_norm": 0.9375,
+      "learning_rate": 8.793133708517759e-06,
+      "loss": 0.0878,
+      "step": 237
+    },
+    {
+      "epoch": 1.1703567035670357,
+      "grad_norm": 1.0625,
+      "learning_rate": 8.782931970156708e-06,
+      "loss": 0.0967,
+      "step": 238
+    },
+    {
+      "epoch": 1.1752767527675276,
+      "grad_norm": 1.171875,
+      "learning_rate": 8.772693266362504e-06,
+      "loss": 0.0976,
+      "step": 239
+    },
+    {
+      "epoch": 1.1801968019680196,
+      "grad_norm": 0.875,
+      "learning_rate": 8.762417697184034e-06,
+      "loss": 0.0737,
+      "step": 240
+    },
+    {
+      "epoch": 1.1851168511685117,
+      "grad_norm": 1.109375,
+      "learning_rate": 8.752105363030414e-06,
+      "loss": 0.1093,
+      "step": 241
+    },
+    {
+      "epoch": 1.1900369003690037,
+      "grad_norm": 1.0703125,
+      "learning_rate": 8.741756364670018e-06,
+      "loss": 0.0853,
+      "step": 242
+    },
+    {
+      "epoch": 1.1949569495694956,
+      "grad_norm": 1.1640625,
+      "learning_rate": 8.731370803229488e-06,
+      "loss": 0.0941,
+      "step": 243
+    },
+    {
+      "epoch": 1.1998769987699878,
+      "grad_norm": 1.125,
+      "learning_rate": 8.720948780192747e-06,
+      "loss": 0.0935,
+      "step": 244
+    },
+    {
+      "epoch": 1.2047970479704797,
+      "grad_norm": 1.109375,
+      "learning_rate": 8.710490397400007e-06,
+      "loss": 0.1011,
+      "step": 245
+    },
+    {
+      "epoch": 1.2097170971709716,
+      "grad_norm": 1.171875,
+      "learning_rate": 8.699995757046773e-06,
+      "loss": 0.1141,
+      "step": 246
+    },
+    {
+      "epoch": 1.2146371463714638,
+      "grad_norm": 1.421875,
+      "learning_rate": 8.689464961682853e-06,
+      "loss": 0.1281,
+      "step": 247
+    },
+    {
+      "epoch": 1.2195571955719557,
+      "grad_norm": 1.1953125,
+      "learning_rate": 8.678898114211338e-06,
+      "loss": 0.1005,
+      "step": 248
+    },
+    {
+      "epoch": 1.2244772447724477,
+      "grad_norm": 1.1171875,
+      "learning_rate": 8.668295317887615e-06,
+      "loss": 0.1019,
+      "step": 249
+    },
+    {
+      "epoch": 1.2293972939729398,
+      "grad_norm": 1.0703125,
+      "learning_rate": 8.657656676318346e-06,
+      "loss": 0.1101,
+      "step": 250
+    },
+    {
+      "epoch": 1.2343173431734318,
+      "grad_norm": 1.2265625,
+      "learning_rate": 8.646982293460461e-06,
+      "loss": 0.1366,
+      "step": 251
+    },
+    {
+      "epoch": 1.2392373923739237,
+      "grad_norm": 1.21875,
+      "learning_rate": 8.63627227362014e-06,
+      "loss": 0.0937,
+      "step": 252
+    },
+    {
+      "epoch": 1.2441574415744157,
+      "grad_norm": 1.2265625,
+      "learning_rate": 8.625526721451799e-06,
+      "loss": 0.1273,
+      "step": 253
+    },
+    {
+      "epoch": 1.2490774907749078,
+      "grad_norm": 1.125,
+      "learning_rate": 8.614745741957054e-06,
+      "loss": 0.097,
+      "step": 254
+    },
+    {
+      "epoch": 1.2539975399753998,
+      "grad_norm": 1.4453125,
+      "learning_rate": 8.603929440483714e-06,
+      "loss": 0.1064,
+      "step": 255
+    },
+    {
+      "epoch": 1.2589175891758917,
+      "grad_norm": 1.1953125,
+      "learning_rate": 8.593077922724733e-06,
+      "loss": 0.118,
+      "step": 256
+    },
+    {
+      "epoch": 1.2638376383763839,
+      "grad_norm": 1.0625,
+      "learning_rate": 8.582191294717192e-06,
+      "loss": 0.0993,
+      "step": 257
+    },
+    {
+      "epoch": 1.2687576875768758,
+      "grad_norm": 1.34375,
+      "learning_rate": 8.571269662841253e-06,
+      "loss": 0.0659,
+      "step": 258
+    },
+    {
+      "epoch": 1.2736777367773677,
+      "grad_norm": 1.3828125,
+      "learning_rate": 8.560313133819124e-06,
+      "loss": 0.12,
+      "step": 259
+    },
+    {
+      "epoch": 1.2785977859778597,
+      "grad_norm": 1.390625,
+      "learning_rate": 8.549321814714018e-06,
+      "loss": 0.0868,
+      "step": 260
+    },
+    {
+      "epoch": 1.2835178351783518,
+      "grad_norm": 1.2734375,
+      "learning_rate": 8.538295812929096e-06,
+      "loss": 0.0902,
+      "step": 261
+    },
+    {
+      "epoch": 1.2884378843788438,
+      "grad_norm": 1.09375,
+      "learning_rate": 8.527235236206438e-06,
+      "loss": 0.0966,
+      "step": 262
+    },
+    {
+      "epoch": 1.293357933579336,
+      "grad_norm": 0.984375,
+      "learning_rate": 8.516140192625962e-06,
+      "loss": 0.0993,
+      "step": 263
+    },
+    {
+      "epoch": 1.2982779827798279,
+      "grad_norm": 0.97265625,
+      "learning_rate": 8.5050107906044e-06,
+      "loss": 0.0865,
+      "step": 264
+    },
+    {
+      "epoch": 1.3031980319803198,
+      "grad_norm": 1.1640625,
+      "learning_rate": 8.49384713889421e-06,
+      "loss": 0.089,
+      "step": 265
+    },
+    {
+      "epoch": 1.3081180811808117,
+      "grad_norm": 1.046875,
+      "learning_rate": 8.482649346582529e-06,
+      "loss": 0.0956,
+      "step": 266
+    },
+    {
+      "epoch": 1.3130381303813037,
+      "grad_norm": 1.1953125,
+      "learning_rate": 8.471417523090109e-06,
+      "loss": 0.0985,
+      "step": 267
+    },
+    {
+      "epoch": 1.3179581795817958,
+      "grad_norm": 0.875,
+      "learning_rate": 8.460151778170231e-06,
+      "loss": 0.0789,
+      "step": 268
+    },
+    {
+      "epoch": 1.3228782287822878,
+      "grad_norm": 1.3515625,
+      "learning_rate": 8.448852221907657e-06,
+      "loss": 0.0995,
+      "step": 269
+    },
+    {
+      "epoch": 1.32779827798278,
+      "grad_norm": 1.1328125,
+      "learning_rate": 8.43751896471753e-06,
+      "loss": 0.1006,
+      "step": 270
+    },
+    {
+      "epoch": 1.3327183271832719,
+      "grad_norm": 1.1328125,
+      "learning_rate": 8.426152117344314e-06,
+      "loss": 0.1119,
+      "step": 271
+    },
+    {
+      "epoch": 1.3376383763837638,
+      "grad_norm": 1.3359375,
+      "learning_rate": 8.414751790860697e-06,
+      "loss": 0.1329,
+      "step": 272
+    },
+    {
+      "epoch": 1.3425584255842558,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.40331809666652e-06,
+      "loss": 0.0863,
+      "step": 273
+    },
+    {
+      "epoch": 1.347478474784748,
+      "grad_norm": 1.265625,
+      "learning_rate": 8.391851146487675e-06,
+      "loss": 0.1255,
+      "step": 274
+    },
+    {
+      "epoch": 1.3523985239852399,
+      "grad_norm": 1.453125,
+      "learning_rate": 8.380351052375023e-06,
+      "loss": 0.1031,
+      "step": 275
+    },
+    {
+      "epoch": 1.3573185731857318,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.368817926703296e-06,
+      "loss": 0.0798,
+      "step": 276
+    },
+    {
+      "epoch": 1.362238622386224,
+      "grad_norm": 1.28125,
+      "learning_rate": 8.357251882169993e-06,
+      "loss": 0.0923,
+      "step": 277
+    },
+    {
+      "epoch": 1.367158671586716,
+      "grad_norm": 1.0859375,
+      "learning_rate": 8.345653031794292e-06,
+      "loss": 0.0831,
+      "step": 278
+    },
+    {
+      "epoch": 1.3720787207872078,
+      "grad_norm": 1.09375,
+      "learning_rate": 8.33402148891593e-06,
+      "loss": 0.071,
+      "step": 279
+    },
+    {
+      "epoch": 1.3769987699876998,
+      "grad_norm": 1.1015625,
+      "learning_rate": 8.32235736719411e-06,
+      "loss": 0.0697,
+      "step": 280
+    },
+    {
+      "epoch": 1.381918819188192,
+      "grad_norm": 0.97265625,
+      "learning_rate": 8.310660780606376e-06,
+      "loss": 0.0704,
+      "step": 281
+    },
+    {
+      "epoch": 1.3868388683886839,
+      "grad_norm": 1.34375,
+      "learning_rate": 8.298931843447517e-06,
+      "loss": 0.1128,
+      "step": 282
+    },
+    {
+      "epoch": 1.391758917589176,
+      "grad_norm": 1.2421875,
+      "learning_rate": 8.28717067032843e-06,
+      "loss": 0.1112,
+      "step": 283
+    },
+    {
+      "epoch": 1.396678966789668,
+      "grad_norm": 0.90625,
+      "learning_rate": 8.275377376175015e-06,
+      "loss": 0.0798,
+      "step": 284
+    },
+    {
+      "epoch": 1.40159901599016,
+      "grad_norm": 1.1484375,
+      "learning_rate": 8.263552076227048e-06,
+      "loss": 0.1072,
+      "step": 285
+    },
+    {
+      "epoch": 1.4065190651906518,
+      "grad_norm": 1.0390625,
+      "learning_rate": 8.251694886037052e-06,
+      "loss": 0.0953,
+      "step": 286
+    },
+    {
+      "epoch": 1.4114391143911438,
+      "grad_norm": 1.21875,
+      "learning_rate": 8.239805921469168e-06,
+      "loss": 0.1145,
+      "step": 287
+    },
+    {
+      "epoch": 1.416359163591636,
+      "grad_norm": 1.4453125,
+      "learning_rate": 8.227885298698029e-06,
+      "loss": 0.1047,
+      "step": 288
+    },
+    {
+      "epoch": 1.4212792127921279,
+      "grad_norm": 1.046875,
+      "learning_rate": 8.21593313420762e-06,
+      "loss": 0.0905,
+      "step": 289
+    },
+    {
+      "epoch": 1.42619926199262,
+      "grad_norm": 0.9609375,
+      "learning_rate": 8.203949544790131e-06,
+      "loss": 0.1013,
+      "step": 290
+    },
+    {
+      "epoch": 1.431119311193112,
+      "grad_norm": 1.2109375,
+      "learning_rate": 8.19193464754484e-06,
+      "loss": 0.085,
+      "step": 291
+    },
+    {
+      "epoch": 1.436039360393604,
+      "grad_norm": 1.2421875,
+      "learning_rate": 8.179888559876943e-06,
+      "loss": 0.1029,
+      "step": 292
+    },
+    {
+      "epoch": 1.4409594095940959,
+      "grad_norm": 1.1796875,
+      "learning_rate": 8.16781139949642e-06,
+      "loss": 0.1262,
+      "step": 293
+    },
+    {
+      "epoch": 1.445879458794588,
+      "grad_norm": 1.421875,
+      "learning_rate": 8.155703284416884e-06,
+      "loss": 0.125,
+      "step": 294
+    },
+    {
+      "epoch": 1.45079950799508,
+      "grad_norm": 0.8984375,
+      "learning_rate": 8.143564332954426e-06,
+      "loss": 0.0796,
+      "step": 295
+    },
+    {
+      "epoch": 1.455719557195572,
+      "grad_norm": 1.4453125,
+      "learning_rate": 8.131394663726452e-06,
+      "loss": 0.1024,
+      "step": 296
+    },
+    {
+      "epoch": 1.460639606396064,
+      "grad_norm": 1.109375,
+      "learning_rate": 8.119194395650545e-06,
+      "loss": 0.0994,
+      "step": 297
+    },
+    {
+      "epoch": 1.465559655596556,
+      "grad_norm": 1.15625,
+      "learning_rate": 8.106963647943273e-06,
+      "loss": 0.1104,
+      "step": 298
+    },
+    {
+      "epoch": 1.470479704797048,
+      "grad_norm": 1.359375,
+      "learning_rate": 8.09470254011905e-06,
+      "loss": 0.1446,
+      "step": 299
+    },
+    {
+      "epoch": 1.4753997539975399,
+      "grad_norm": 1.0078125,
+      "learning_rate": 8.082411191988956e-06,
+      "loss": 0.1061,
+      "step": 300
+    },
+    {
+      "epoch": 1.480319803198032,
+      "grad_norm": 0.953125,
+      "learning_rate": 8.070089723659567e-06,
+      "loss": 0.074,
+      "step": 301
+    },
+    {
+      "epoch": 1.485239852398524,
+      "grad_norm": 1.296875,
+      "learning_rate": 8.057738255531781e-06,
+      "loss": 0.0852,
+      "step": 302
+    },
+    {
+      "epoch": 1.4901599015990161,
+      "grad_norm": 1.015625,
+      "learning_rate": 8.045356908299647e-06,
+      "loss": 0.091,
+      "step": 303
+    },
+    {
+      "epoch": 1.495079950799508,
+      "grad_norm": 1.0,
+      "learning_rate": 8.032945802949179e-06,
+      "loss": 0.0849,
+      "step": 304
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.98828125,
+      "learning_rate": 8.02050506075718e-06,
+      "loss": 0.1125,
+      "step": 305
+    },
+    {
+      "epoch": 1.504920049200492,
+      "grad_norm": 0.8984375,
+      "learning_rate": 8.008034803290045e-06,
+      "loss": 0.086,
+      "step": 306
+    },
+    {
+      "epoch": 1.504920049200492,
+      "eval_loss": 0.10929346084594727,
+      "eval_runtime": 203.7138,
+      "eval_samples_per_second": 8.031,
+      "eval_steps_per_second": 1.34,
+      "step": 306
+    },
+    {
+      "epoch": 1.5098400984009839,
+      "grad_norm": 1.0390625,
+      "learning_rate": 7.995535152402592e-06,
+      "loss": 0.1118,
+      "step": 307
+    },
+    {
+      "epoch": 1.514760147601476,
+      "grad_norm": 1.203125,
+      "learning_rate": 7.983006230236855e-06,
+      "loss": 0.0959,
+      "step": 308
+    },
+    {
+      "epoch": 1.519680196801968,
+      "grad_norm": 1.234375,
+      "learning_rate": 7.970448159220897e-06,
+      "loss": 0.0916,
+      "step": 309
+    },
+    {
+      "epoch": 1.5246002460024601,
+      "grad_norm": 1.21875,
+      "learning_rate": 7.957861062067614e-06,
+      "loss": 0.1117,
+      "step": 310
+    },
+    {
+      "epoch": 1.529520295202952,
+      "grad_norm": 0.87109375,
+      "learning_rate": 7.945245061773531e-06,
+      "loss": 0.073,
+      "step": 311
+    },
+    {
+      "epoch": 1.534440344403444,
+      "grad_norm": 0.96875,
+      "learning_rate": 7.93260028161761e-06,
+      "loss": 0.072,
+      "step": 312
+    },
+    {
+      "epoch": 1.539360393603936,
+      "grad_norm": 1.046875,
+      "learning_rate": 7.919926845160037e-06,
+      "loss": 0.1082,
+      "step": 313
+    },
+    {
+      "epoch": 1.5442804428044279,
+      "grad_norm": 0.9375,
+      "learning_rate": 7.907224876241015e-06,
+      "loss": 0.0828,
+      "step": 314
+    },
+    {
+      "epoch": 1.54920049200492,
+      "grad_norm": 0.9296875,
+      "learning_rate": 7.894494498979558e-06,
+      "loss": 0.0924,
+      "step": 315
+    },
+    {
+      "epoch": 1.5541205412054122,
+      "grad_norm": 1.203125,
+      "learning_rate": 7.881735837772274e-06,
+      "loss": 0.1099,
+      "step": 316
+    },
+    {
+      "epoch": 1.5590405904059041,
+      "grad_norm": 0.984375,
+      "learning_rate": 7.868949017292153e-06,
+      "loss": 0.0909,
+      "step": 317
+    },
+    {
+      "epoch": 1.563960639606396,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.856134162487346e-06,
+      "loss": 0.0986,
+      "step": 318
+    },
+    {
+      "epoch": 1.568880688806888,
+      "grad_norm": 0.94140625,
+      "learning_rate": 7.843291398579946e-06,
+      "loss": 0.1045,
+      "step": 319
+    },
+    {
+      "epoch": 1.57380073800738,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.830420851064767e-06,
+      "loss": 0.0971,
+      "step": 320
+    },
+    {
+      "epoch": 1.5787207872078721,
+      "grad_norm": 1.03125,
+      "learning_rate": 7.817522645708106e-06,
+      "loss": 0.0974,
+      "step": 321
+    },
+    {
+      "epoch": 1.583640836408364,
+      "grad_norm": 1.015625,
+      "learning_rate": 7.804596908546529e-06,
+      "loss": 0.1178,
+      "step": 322
+    },
+    {
+      "epoch": 1.5885608856088562,
+      "grad_norm": 1.1015625,
+      "learning_rate": 7.791643765885632e-06,
+      "loss": 0.093,
+      "step": 323
+    },
+    {
+      "epoch": 1.5934809348093482,
+      "grad_norm": 1.015625,
+      "learning_rate": 7.778663344298804e-06,
+      "loss": 0.0798,
+      "step": 324
+    },
+    {
+      "epoch": 1.59840098400984,
+      "grad_norm": 0.8828125,
+      "learning_rate": 7.765655770625997e-06,
+      "loss": 0.0991,
+      "step": 325
+    },
+    {
+      "epoch": 1.603321033210332,
+      "grad_norm": 0.93359375,
+      "learning_rate": 7.752621171972486e-06,
+      "loss": 0.0673,
+      "step": 326
+    },
+    {
+      "epoch": 1.608241082410824,
+      "grad_norm": 0.94140625,
+      "learning_rate": 7.739559675707615e-06,
+      "loss": 0.094,
+      "step": 327
+    },
+    {
+      "epoch": 1.6131611316113161,
+      "grad_norm": 1.15625,
+      "learning_rate": 7.726471409463572e-06,
+      "loss": 0.1286,
+      "step": 328
+    },
+    {
+      "epoch": 1.618081180811808,
+      "grad_norm": 1.3125,
+      "learning_rate": 7.713356501134124e-06,
+      "loss": 0.0938,
+      "step": 329
+    },
+    {
+      "epoch": 1.6230012300123002,
+      "grad_norm": 1.203125,
+      "learning_rate": 7.70021507887338e-06,
+      "loss": 0.1141,
+      "step": 330
+    },
+    {
+      "epoch": 1.6279212792127922,
+      "grad_norm": 0.71875,
+      "learning_rate": 7.687047271094528e-06,
+      "loss": 0.0708,
+      "step": 331
+    },
+    {
+      "epoch": 1.632841328413284,
+      "grad_norm": 0.96875,
+      "learning_rate": 7.673853206468593e-06,
+      "loss": 0.0816,
+      "step": 332
+    },
+    {
+      "epoch": 1.637761377613776,
+      "grad_norm": 1.3515625,
+      "learning_rate": 7.660633013923164e-06,
+      "loss": 0.1224,
+      "step": 333
+    },
+    {
+      "epoch": 1.642681426814268,
+      "grad_norm": 1.265625,
+      "learning_rate": 7.64738682264115e-06,
+      "loss": 0.1012,
+      "step": 334
+    },
+    {
+      "epoch": 1.6476014760147601,
+      "grad_norm": 0.8984375,
+      "learning_rate": 7.634114762059504e-06,
+      "loss": 0.103,
+      "step": 335
+    },
+    {
+      "epoch": 1.6525215252152523,
+      "grad_norm": 0.9375,
+      "learning_rate": 7.62081696186797e-06,
+      "loss": 0.0773,
+      "step": 336
+    },
+    {
+      "epoch": 1.6574415744157442,
+      "grad_norm": 0.98828125,
+      "learning_rate": 7.607493552007805e-06,
+      "loss": 0.1176,
+      "step": 337
+    },
+    {
+      "epoch": 1.6623616236162362,
+      "grad_norm": 0.9765625,
+      "learning_rate": 7.5941446626705175e-06,
+      "loss": 0.0996,
+      "step": 338
+    },
+    {
+      "epoch": 1.6672816728167281,
+      "grad_norm": 0.96484375,
+      "learning_rate": 7.580770424296591e-06,
+      "loss": 0.1001,
+      "step": 339
+    },
+    {
+      "epoch": 1.67220172201722,
+      "grad_norm": 0.8359375,
+      "learning_rate": 7.56737096757421e-06,
+      "loss": 0.1052,
+      "step": 340
+    },
+    {
+      "epoch": 1.6771217712177122,
+      "grad_norm": 1.28125,
+      "learning_rate": 7.553946423437988e-06,
+      "loss": 0.1104,
+      "step": 341
+    },
+    {
+      "epoch": 1.6820418204182042,
+      "grad_norm": 1.078125,
+      "learning_rate": 7.540496923067675e-06,
+      "loss": 0.1092,
+      "step": 342
+    },
+    {
+      "epoch": 1.6869618696186963,
+      "grad_norm": 1.1640625,
+      "learning_rate": 7.527022597886895e-06,
+      "loss": 0.0994,
+      "step": 343
+    },
+    {
+      "epoch": 1.6918819188191883,
+      "grad_norm": 1.484375,
+      "learning_rate": 7.513523579561839e-06,
+      "loss": 0.0787,
+      "step": 344
+    },
+    {
+      "epoch": 1.6968019680196802,
+      "grad_norm": 1.125,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.0986,
+      "step": 345
+    },
+    {
+      "epoch": 1.7017220172201721,
+      "grad_norm": 1.0546875,
+      "learning_rate": 7.486451991348872e-06,
+      "loss": 0.0806,
+      "step": 346
+    },
+    {
+      "epoch": 1.706642066420664,
+      "grad_norm": 1.2890625,
+      "learning_rate": 7.472879685994658e-06,
+      "loss": 0.1059,
+      "step": 347
+    },
+    {
+      "epoch": 1.7115621156211562,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.459283216560982e-06,
+      "loss": 0.0947,
+      "step": 348
+    },
+    {
+      "epoch": 1.7164821648216482,
+      "grad_norm": 1.015625,
+      "learning_rate": 7.445662715907591e-06,
+      "loss": 0.0897,
+      "step": 349
+    },
+    {
+      "epoch": 1.7214022140221403,
+      "grad_norm": 1.046875,
+      "learning_rate": 7.432018317129056e-06,
+      "loss": 0.1083,
+      "step": 350
+    },
+    {
+      "epoch": 1.7263222632226323,
+      "grad_norm": 0.8984375,
+      "learning_rate": 7.418350153553471e-06,
+      "loss": 0.07,
+      "step": 351
+    },
+    {
+      "epoch": 1.7312423124231242,
+      "grad_norm": 1.0703125,
+      "learning_rate": 7.40465835874115e-06,
+      "loss": 0.0942,
+      "step": 352
+    },
+    {
+      "epoch": 1.7361623616236161,
+      "grad_norm": 0.9765625,
+      "learning_rate": 7.390943066483327e-06,
+      "loss": 0.0881,
+      "step": 353
+    },
+    {
+      "epoch": 1.741082410824108,
+      "grad_norm": 1.046875,
+      "learning_rate": 7.377204410800839e-06,
+      "loss": 0.1029,
+      "step": 354
+    },
+    {
+      "epoch": 1.7460024600246002,
+      "grad_norm": 0.96484375,
+      "learning_rate": 7.363442525942827e-06,
+      "loss": 0.093,
+      "step": 355
+    },
+    {
+      "epoch": 1.7509225092250924,
+      "grad_norm": 1.1328125,
+      "learning_rate": 7.349657546385414e-06,
+      "loss": 0.0857,
+      "step": 356
+    },
+    {
+      "epoch": 1.7558425584255843,
+      "grad_norm": 0.9375,
+      "learning_rate": 7.335849606830402e-06,
+      "loss": 0.1016,
+      "step": 357
+    },
+    {
+      "epoch": 1.7607626076260763,
+      "grad_norm": 1.015625,
+      "learning_rate": 7.322018842203942e-06,
+      "loss": 0.1107,
+      "step": 358
+    },
+    {
+      "epoch": 1.7656826568265682,
+      "grad_norm": 1.15625,
+      "learning_rate": 7.308165387655231e-06,
+      "loss": 0.1411,
+      "step": 359
+    },
+    {
+      "epoch": 1.7706027060270602,
+      "grad_norm": 0.89453125,
+      "learning_rate": 7.294289378555179e-06,
+      "loss": 0.1037,
+      "step": 360
+    },
+    {
+      "epoch": 1.7755227552275523,
+      "grad_norm": 0.87109375,
+      "learning_rate": 7.2803909504950935e-06,
+      "loss": 0.0838,
+      "step": 361
+    },
+    {
+      "epoch": 1.7804428044280443,
+      "grad_norm": 0.99609375,
+      "learning_rate": 7.266470239285347e-06,
+      "loss": 0.0947,
+      "step": 362
+    },
+    {
+      "epoch": 1.7853628536285364,
+      "grad_norm": 1.0625,
+      "learning_rate": 7.252527380954062e-06,
+      "loss": 0.0904,
+      "step": 363
+    },
+    {
+      "epoch": 1.7902829028290284,
+      "grad_norm": 1.1640625,
+      "learning_rate": 7.238562511745768e-06,
+      "loss": 0.0877,
+      "step": 364
+    },
+    {
+      "epoch": 1.7952029520295203,
+      "grad_norm": 1.015625,
+      "learning_rate": 7.2245757681200835e-06,
+      "loss": 0.0871,
+      "step": 365
+    },
+    {
+      "epoch": 1.8001230012300122,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.210567286750368e-06,
+      "loss": 0.0898,
+      "step": 366
+    },
+    {
+      "epoch": 1.8050430504305042,
+      "grad_norm": 1.140625,
+      "learning_rate": 7.196537204522401e-06,
+      "loss": 0.0826,
+      "step": 367
+    },
+    {
+      "epoch": 1.8099630996309963,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.182485658533036e-06,
+      "loss": 0.0991,
+      "step": 368
+    },
+    {
+      "epoch": 1.8148831488314883,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.168412786088857e-06,
+      "loss": 0.0775,
+      "step": 369
+    },
+    {
+      "epoch": 1.8198031980319804,
+      "grad_norm": 0.8984375,
+      "learning_rate": 7.1543187247048525e-06,
+      "loss": 0.0915,
+      "step": 370
+    },
+    {
+      "epoch": 1.8247232472324724,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.140203612103052e-06,
+      "loss": 0.0983,
+      "step": 371
+    },
+    {
+      "epoch": 1.8296432964329643,
+      "grad_norm": 0.89453125,
+      "learning_rate": 7.1260675862111986e-06,
+      "loss": 0.086,
+      "step": 372
+    },
+    {
+      "epoch": 1.8345633456334562,
+      "grad_norm": 1.34375,
+      "learning_rate": 7.111910785161381e-06,
+      "loss": 0.1281,
+      "step": 373
+    },
+    {
+      "epoch": 1.8394833948339482,
+      "grad_norm": 1.3203125,
+      "learning_rate": 7.0977333472887076e-06,
+      "loss": 0.0902,
+      "step": 374
+    },
+    {
+      "epoch": 1.8444034440344403,
+      "grad_norm": 1.234375,
+      "learning_rate": 7.083535411129934e-06,
+      "loss": 0.0996,
+      "step": 375
+    },
+    {
+      "epoch": 1.8493234932349325,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.069317115422121e-06,
+      "loss": 0.0887,
+      "step": 376
+    },
+    {
+      "epoch": 1.8542435424354244,
+      "grad_norm": 1.15625,
+      "learning_rate": 7.055078599101275e-06,
+      "loss": 0.1046,
+      "step": 377
+    },
+    {
+      "epoch": 1.8591635916359164,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.040820001300992e-06,
+      "loss": 0.1243,
+      "step": 378
+    },
+    {
+      "epoch": 1.8640836408364083,
+      "grad_norm": 1.1328125,
+      "learning_rate": 7.026541461351092e-06,
+      "loss": 0.0664,
+      "step": 379
+    },
+    {
+      "epoch": 1.8690036900369003,
+      "grad_norm": 1.0859375,
+      "learning_rate": 7.01224311877627e-06,
+      "loss": 0.1314,
+      "step": 380
+    },
+    {
+      "epoch": 1.8739237392373924,
+      "grad_norm": 0.82421875,
+      "learning_rate": 6.997925113294717e-06,
+      "loss": 0.0811,
+      "step": 381
+    },
+    {
+      "epoch": 1.8788437884378844,
+      "grad_norm": 0.9921875,
+      "learning_rate": 6.983587584816769e-06,
+      "loss": 0.0776,
+      "step": 382
+    },
+    {
+      "epoch": 1.8837638376383765,
+      "grad_norm": 1.359375,
+      "learning_rate": 6.969230673443531e-06,
+      "loss": 0.116,
+      "step": 383
+    },
+    {
+      "epoch": 1.8886838868388685,
+      "grad_norm": 0.9765625,
+      "learning_rate": 6.9548545194655115e-06,
+      "loss": 0.0983,
+      "step": 384
+    },
+    {
+      "epoch": 1.8936039360393604,
+      "grad_norm": 1.28125,
+      "learning_rate": 6.9404592633612486e-06,
+      "loss": 0.1323,
+      "step": 385
+    },
+    {
+      "epoch": 1.8985239852398523,
+      "grad_norm": 1.328125,
+      "learning_rate": 6.926045045795944e-06,
+      "loss": 0.1106,
+      "step": 386
+    },
+    {
+      "epoch": 1.9034440344403443,
+      "grad_norm": 1.3046875,
+      "learning_rate": 6.911612007620077e-06,
+      "loss": 0.0884,
+      "step": 387
+    },
+    {
+      "epoch": 1.9083640836408364,
+      "grad_norm": 0.953125,
+      "learning_rate": 6.897160289868042e-06,
+      "loss": 0.101,
+      "step": 388
+    },
+    {
+      "epoch": 1.9132841328413284,
+      "grad_norm": 0.81640625,
+      "learning_rate": 6.8826900337567595e-06,
+      "loss": 0.0907,
+      "step": 389
+    },
+    {
+      "epoch": 1.9182041820418205,
+      "grad_norm": 1.0078125,
+      "learning_rate": 6.8682013806842985e-06,
+      "loss": 0.083,
+      "step": 390
+    },
+    {
+      "epoch": 1.9231242312423125,
+      "grad_norm": 0.9765625,
+      "learning_rate": 6.853694472228504e-06,
+      "loss": 0.0821,
+      "step": 391
+    },
+    {
+      "epoch": 1.9280442804428044,
+      "grad_norm": 1.046875,
+      "learning_rate": 6.839169450145595e-06,
+      "loss": 0.0605,
+      "step": 392
+    },
+    {
+      "epoch": 1.9329643296432963,
+      "grad_norm": 1.0234375,
+      "learning_rate": 6.824626456368802e-06,
+      "loss": 0.1001,
+      "step": 393
+    },
+    {
+      "epoch": 1.9378843788437883,
+      "grad_norm": 1.203125,
+      "learning_rate": 6.810065633006956e-06,
+      "loss": 0.0936,
+      "step": 394
+    },
+    {
+      "epoch": 1.9428044280442804,
+      "grad_norm": 1.046875,
+      "learning_rate": 6.795487122343124e-06,
+      "loss": 0.0866,
+      "step": 395
+    },
+    {
+      "epoch": 1.9477244772447726,
+      "grad_norm": 1.3125,
+      "learning_rate": 6.7808910668331985e-06,
+      "loss": 0.1078,
+      "step": 396
+    },
+    {
+      "epoch": 1.9526445264452645,
+      "grad_norm": 1.109375,
+      "learning_rate": 6.766277609104518e-06,
+      "loss": 0.0854,
+      "step": 397
+    },
+    {
+      "epoch": 1.9575645756457565,
+      "grad_norm": 0.9453125,
+      "learning_rate": 6.751646891954466e-06,
+      "loss": 0.0848,
+      "step": 398
+    },
+    {
+      "epoch": 1.9624846248462484,
+      "grad_norm": 1.0234375,
+      "learning_rate": 6.736999058349085e-06,
+      "loss": 0.0861,
+      "step": 399
+    },
+    {
+      "epoch": 1.9674046740467404,
+      "grad_norm": 1.4296875,
+      "learning_rate": 6.722334251421665e-06,
+      "loss": 0.108,
+      "step": 400
+    },
+    {
+      "epoch": 1.9723247232472325,
+      "grad_norm": 0.76171875,
+      "learning_rate": 6.707652614471359e-06,
+      "loss": 0.0691,
+      "step": 401
+    },
+    {
+      "epoch": 1.9772447724477245,
+      "grad_norm": 1.4765625,
+      "learning_rate": 6.692954290961774e-06,
+      "loss": 0.1234,
+      "step": 402
+    },
+    {
+      "epoch": 1.9821648216482166,
+      "grad_norm": 0.9140625,
+      "learning_rate": 6.678239424519575e-06,
+      "loss": 0.0785,
+      "step": 403
+    },
+    {
+      "epoch": 1.9870848708487086,
+      "grad_norm": 1.328125,
+      "learning_rate": 6.6635081589330745e-06,
+      "loss": 0.096,
+      "step": 404
+    },
+    {
+      "epoch": 1.9920049200492005,
+      "grad_norm": 1.25,
+      "learning_rate": 6.648760638150833e-06,
+      "loss": 0.1109,
+      "step": 405
+    },
+    {
+      "epoch": 1.9969249692496924,
+      "grad_norm": 1.125,
+      "learning_rate": 6.6339970062802526e-06,
+      "loss": 0.0892,
+      "step": 406
+    },
+    {
+      "epoch": 2.0030750307503076,
+      "grad_norm": 0.83984375,
+      "learning_rate": 6.619217407586167e-06,
+      "loss": 0.065,
+      "step": 407
+    },
+    {
+      "epoch": 2.0079950799507995,
+      "grad_norm": 1.2578125,
+      "learning_rate": 6.604421986489428e-06,
+      "loss": 0.1044,
+      "step": 408
+    },
+    {
+      "epoch": 2.0079950799507995,
+      "eval_loss": 0.10599377751350403,
+      "eval_runtime": 203.102,
+      "eval_samples_per_second": 8.055,
+      "eval_steps_per_second": 1.344,
+      "step": 408
+    },
+    {
+      "epoch": 2.0129151291512914,
+      "grad_norm": 0.8828125,
+      "learning_rate": 6.589610887565503e-06,
+      "loss": 0.0784,
+      "step": 409
+    },
+    {
+      "epoch": 2.0178351783517834,
+      "grad_norm": 0.9609375,
+      "learning_rate": 6.574784255543052e-06,
+      "loss": 0.0788,
+      "step": 410
+    },
+    {
+      "epoch": 2.0227552275522753,
+      "grad_norm": 0.81640625,
+      "learning_rate": 6.559942235302527e-06,
+      "loss": 0.0876,
+      "step": 411
+    },
+    {
+      "epoch": 2.0276752767527677,
+      "grad_norm": 1.125,
+      "learning_rate": 6.545084971874738e-06,
+      "loss": 0.0702,
+      "step": 412
+    },
+    {
+      "epoch": 2.0325953259532596,
+      "grad_norm": 1.703125,
+      "learning_rate": 6.530212610439455e-06,
+      "loss": 0.1014,
+      "step": 413
+    },
+    {
+      "epoch": 2.0375153751537516,
+      "grad_norm": 0.87890625,
+      "learning_rate": 6.5153252963239736e-06,
+      "loss": 0.0846,
+      "step": 414
+    },
+    {
+      "epoch": 2.0424354243542435,
+      "grad_norm": 1.34375,
+      "learning_rate": 6.500423175001705e-06,
+      "loss": 0.1422,
+      "step": 415
+    },
+    {
+      "epoch": 2.0473554735547355,
+      "grad_norm": 1.1796875,
+      "learning_rate": 6.485506392090749e-06,
+      "loss": 0.1003,
+      "step": 416
+    },
+    {
+      "epoch": 2.0522755227552274,
+      "grad_norm": 1.1328125,
+      "learning_rate": 6.470575093352477e-06,
+      "loss": 0.1012,
+      "step": 417
+    },
+    {
+      "epoch": 2.0571955719557193,
+      "grad_norm": 1.1484375,
+      "learning_rate": 6.4556294246900994e-06,
+      "loss": 0.0778,
+      "step": 418
+    },
+    {
+      "epoch": 2.0621156211562117,
+      "grad_norm": 1.109375,
+      "learning_rate": 6.440669532147246e-06,
+      "loss": 0.0975,
+      "step": 419
+    },
+    {
+      "epoch": 2.0670356703567037,
+      "grad_norm": 0.91015625,
+      "learning_rate": 6.4256955619065375e-06,
+      "loss": 0.0772,
+      "step": 420
+    },
+    {
+      "epoch": 2.0719557195571956,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.410707660288154e-06,
+      "loss": 0.1044,
+      "step": 421
+    },
+    {
+      "epoch": 2.0768757687576875,
+      "grad_norm": 1.1796875,
+      "learning_rate": 6.395705973748414e-06,
+      "loss": 0.0868,
+      "step": 422
+    },
+    {
+      "epoch": 2.0817958179581795,
+      "grad_norm": 1.234375,
+      "learning_rate": 6.3806906488783305e-06,
+      "loss": 0.1051,
+      "step": 423
+    },
+    {
+      "epoch": 2.0867158671586714,
+      "grad_norm": 1.1171875,
+      "learning_rate": 6.36566183240219e-06,
+      "loss": 0.0572,
+      "step": 424
+    },
+    {
+      "epoch": 2.091635916359164,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.350619671176111e-06,
+      "loss": 0.1055,
+      "step": 425
+    },
+    {
+      "epoch": 2.0965559655596557,
+      "grad_norm": 0.9140625,
+      "learning_rate": 6.335564312186615e-06,
+      "loss": 0.0956,
+      "step": 426
+    },
+    {
+      "epoch": 2.1014760147601477,
+      "grad_norm": 1.75,
+      "learning_rate": 6.3204959025491844e-06,
+      "loss": 0.0727,
+      "step": 427
+    },
+    {
+      "epoch": 2.1063960639606396,
+      "grad_norm": 1.1015625,
+      "learning_rate": 6.30541458950683e-06,
+      "loss": 0.1014,
+      "step": 428
+    },
+    {
+      "epoch": 2.1113161131611315,
+      "grad_norm": 1.1953125,
+      "learning_rate": 6.2903205204286474e-06,
+      "loss": 0.0968,
+      "step": 429
+    },
+    {
+      "epoch": 2.1162361623616235,
+      "grad_norm": 1.21875,
+      "learning_rate": 6.275213842808383e-06,
+      "loss": 0.0856,
+      "step": 430
+    },
+    {
+      "epoch": 2.1211562115621154,
+      "grad_norm": 1.359375,
+      "learning_rate": 6.260094704262986e-06,
+      "loss": 0.0773,
+      "step": 431
+    },
+    {
+      "epoch": 2.126076260762608,
+      "grad_norm": 1.125,
+      "learning_rate": 6.244963252531171e-06,
+      "loss": 0.0802,
+      "step": 432
+    },
+    {
+      "epoch": 2.1309963099630997,
+      "grad_norm": 1.21875,
+      "learning_rate": 6.229819635471972e-06,
+      "loss": 0.0809,
+      "step": 433
+    },
+    {
+      "epoch": 2.1359163591635917,
+      "grad_norm": 1.1171875,
+      "learning_rate": 6.2146640010633e-06,
+      "loss": 0.0833,
+      "step": 434
+    },
+    {
+      "epoch": 2.1408364083640836,
+      "grad_norm": 0.984375,
+      "learning_rate": 6.19949649740049e-06,
+      "loss": 0.0784,
+      "step": 435
+    },
+    {
+      "epoch": 2.1457564575645756,
+      "grad_norm": 0.9453125,
+      "learning_rate": 6.184317272694866e-06,
+      "loss": 0.0885,
+      "step": 436
+    },
+    {
+      "epoch": 2.1506765067650675,
+      "grad_norm": 1.140625,
+      "learning_rate": 6.16912647527228e-06,
+      "loss": 0.0763,
+      "step": 437
+    },
+    {
+      "epoch": 2.15559655596556,
+      "grad_norm": 1.1484375,
+      "learning_rate": 6.1539242535716704e-06,
+      "loss": 0.0805,
+      "step": 438
+    },
+    {
+      "epoch": 2.160516605166052,
+      "grad_norm": 1.2421875,
+      "learning_rate": 6.138710756143613e-06,
+      "loss": 0.1051,
+      "step": 439
+    },
+    {
+      "epoch": 2.1654366543665438,
+      "grad_norm": 0.96875,
+      "learning_rate": 6.123486131648859e-06,
+      "loss": 0.0741,
+      "step": 440
+    },
+    {
+      "epoch": 2.1703567035670357,
+      "grad_norm": 1.171875,
+      "learning_rate": 6.108250528856895e-06,
+      "loss": 0.0725,
+      "step": 441
+    },
+    {
+      "epoch": 2.1752767527675276,
+      "grad_norm": 1.0859375,
+      "learning_rate": 6.0930040966444815e-06,
+      "loss": 0.0957,
+      "step": 442
+    },
+    {
+      "epoch": 2.1801968019680196,
+      "grad_norm": 1.421875,
+      "learning_rate": 6.077746983994198e-06,
+      "loss": 0.1151,
+      "step": 443
+    },
+    {
+      "epoch": 2.1851168511685115,
+      "grad_norm": 1.1328125,
+      "learning_rate": 6.062479339992993e-06,
+      "loss": 0.0939,
+      "step": 444
+    },
+    {
+      "epoch": 2.190036900369004,
+      "grad_norm": 1.0234375,
+      "learning_rate": 6.047201313830724e-06,
+      "loss": 0.0944,
+      "step": 445
+    },
+    {
+      "epoch": 2.194956949569496,
+      "grad_norm": 1.0859375,
+      "learning_rate": 6.031913054798692e-06,
+      "loss": 0.0825,
+      "step": 446
+    },
+    {
+      "epoch": 2.1998769987699878,
+      "grad_norm": 0.94921875,
+      "learning_rate": 6.016614712288198e-06,
+      "loss": 0.0732,
+      "step": 447
+    },
+    {
+      "epoch": 2.2047970479704797,
+      "grad_norm": 1.2421875,
+      "learning_rate": 6.0013064357890715e-06,
+      "loss": 0.0749,
+      "step": 448
+    },
+    {
+      "epoch": 2.2097170971709716,
+      "grad_norm": 1.3828125,
+      "learning_rate": 5.985988374888216e-06,
+      "loss": 0.0935,
+      "step": 449
+    },
+    {
+      "epoch": 2.2146371463714636,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.970660679268139e-06,
+      "loss": 0.0762,
+      "step": 450
+    },
+    {
+      "epoch": 2.2195571955719555,
+      "grad_norm": 1.3671875,
+      "learning_rate": 5.955323498705501e-06,
+      "loss": 0.0929,
+      "step": 451
+    },
+    {
+      "epoch": 2.224477244772448,
+      "grad_norm": 1.3046875,
+      "learning_rate": 5.9399769830696404e-06,
+      "loss": 0.1067,
+      "step": 452
+    },
+    {
+      "epoch": 2.22939729397294,
+      "grad_norm": 0.9609375,
+      "learning_rate": 5.924621282321123e-06,
+      "loss": 0.0736,
+      "step": 453
+    },
+    {
+      "epoch": 2.234317343173432,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5.909256546510257e-06,
+      "loss": 0.0699,
+      "step": 454
+    },
+    {
+      "epoch": 2.2392373923739237,
+      "grad_norm": 1.0,
+      "learning_rate": 5.893882925775648e-06,
+      "loss": 0.0792,
+      "step": 455
+    },
+    {
+      "epoch": 2.2441574415744157,
+      "grad_norm": 1.15625,
+      "learning_rate": 5.878500570342714e-06,
+      "loss": 0.0874,
+      "step": 456
+    },
+    {
+      "epoch": 2.2490774907749076,
+      "grad_norm": 1.296875,
+      "learning_rate": 5.86310963052223e-06,
+      "loss": 0.1197,
+      "step": 457
+    },
+    {
+      "epoch": 2.2539975399753995,
+      "grad_norm": 1.21875,
+      "learning_rate": 5.847710256708854e-06,
+      "loss": 0.1147,
+      "step": 458
+    },
+    {
+      "epoch": 2.258917589175892,
+      "grad_norm": 0.9453125,
+      "learning_rate": 5.832302599379657e-06,
+      "loss": 0.0647,
+      "step": 459
+    },
+    {
+      "epoch": 2.263837638376384,
+      "grad_norm": 0.86328125,
+      "learning_rate": 5.816886809092651e-06,
+      "loss": 0.0794,
+      "step": 460
+    },
+    {
+      "epoch": 2.268757687576876,
+      "grad_norm": 1.1171875,
+      "learning_rate": 5.8014630364853275e-06,
+      "loss": 0.0787,
+      "step": 461
+    },
+    {
+      "epoch": 2.2736777367773677,
+      "grad_norm": 1.375,
+      "learning_rate": 5.7860314322731705e-06,
+      "loss": 0.1395,
+      "step": 462
+    },
+    {
+      "epoch": 2.2785977859778597,
+      "grad_norm": 1.6640625,
+      "learning_rate": 5.770592147248197e-06,
+      "loss": 0.083,
+      "step": 463
+    },
+    {
+      "epoch": 2.2835178351783516,
+      "grad_norm": 1.390625,
+      "learning_rate": 5.755145332277472e-06,
+      "loss": 0.1035,
+      "step": 464
+    },
+    {
+      "epoch": 2.288437884378844,
+      "grad_norm": 1.2265625,
+      "learning_rate": 5.73969113830165e-06,
+      "loss": 0.1229,
+      "step": 465
+    },
+    {
+      "epoch": 2.293357933579336,
+      "grad_norm": 0.96875,
+      "learning_rate": 5.72422971633348e-06,
+      "loss": 0.0847,
+      "step": 466
+    },
+    {
+      "epoch": 2.298277982779828,
+      "grad_norm": 1.2734375,
+      "learning_rate": 5.70876121745635e-06,
+      "loss": 0.0944,
+      "step": 467
+    },
+    {
+      "epoch": 2.30319803198032,
+      "grad_norm": 0.8828125,
+      "learning_rate": 5.69328579282279e-06,
+      "loss": 0.0729,
+      "step": 468
+    },
+    {
+      "epoch": 2.3081180811808117,
+      "grad_norm": 1.2890625,
+      "learning_rate": 5.677803593653018e-06,
+      "loss": 0.076,
+      "step": 469
+    },
+    {
+      "epoch": 2.3130381303813037,
+      "grad_norm": 1.3515625,
+      "learning_rate": 5.66231477123344e-06,
+      "loss": 0.0779,
+      "step": 470
+    },
+    {
+      "epoch": 2.3179581795817956,
+      "grad_norm": 1.09375,
+      "learning_rate": 5.646819476915189e-06,
+      "loss": 0.0596,
+      "step": 471
+    },
+    {
+      "epoch": 2.322878228782288,
+      "grad_norm": 0.98046875,
+      "learning_rate": 5.631317862112636e-06,
+      "loss": 0.1077,
+      "step": 472
+    },
+    {
+      "epoch": 2.32779827798278,
+      "grad_norm": 1.3125,
+      "learning_rate": 5.615810078301912e-06,
+      "loss": 0.1024,
+      "step": 473
+    },
+    {
+      "epoch": 2.332718327183272,
+      "grad_norm": 1.3984375,
+      "learning_rate": 5.600296277019434e-06,
+      "loss": 0.1357,
+      "step": 474
+    },
+    {
+      "epoch": 2.337638376383764,
+      "grad_norm": 1.09375,
+      "learning_rate": 5.584776609860414e-06,
+      "loss": 0.0891,
+      "step": 475
+    },
+    {
+      "epoch": 2.3425584255842558,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5.569251228477386e-06,
+      "loss": 0.0744,
+      "step": 476
+    },
+    {
+      "epoch": 2.3474784747847477,
+      "grad_norm": 0.984375,
+      "learning_rate": 5.553720284578723e-06,
+      "loss": 0.0669,
+      "step": 477
+    },
+    {
+      "epoch": 2.35239852398524,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.538183929927152e-06,
+      "loss": 0.0819,
+      "step": 478
+    },
+    {
+      "epoch": 2.357318573185732,
+      "grad_norm": 1.140625,
+      "learning_rate": 5.522642316338268e-06,
+      "loss": 0.0794,
+      "step": 479
+    },
+    {
+      "epoch": 2.362238622386224,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.507095595679059e-06,
+      "loss": 0.0887,
+      "step": 480
+    },
+    {
+      "epoch": 2.367158671586716,
+      "grad_norm": 1.4296875,
+      "learning_rate": 5.491543919866417e-06,
+      "loss": 0.0863,
+      "step": 481
+    },
+    {
+      "epoch": 2.372078720787208,
+      "grad_norm": 0.9140625,
+      "learning_rate": 5.47598744086565e-06,
+      "loss": 0.0844,
+      "step": 482
+    },
+    {
+      "epoch": 2.3769987699876998,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.460426310689006e-06,
+      "loss": 0.0712,
+      "step": 483
+    },
+    {
+      "epoch": 2.3819188191881917,
+      "grad_norm": 1.359375,
+      "learning_rate": 5.4448606813941805e-06,
+      "loss": 0.0768,
+      "step": 484
+    },
+    {
+      "epoch": 2.3868388683886836,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.42929070508283e-06,
+      "loss": 0.079,
+      "step": 485
+    },
+    {
+      "epoch": 2.391758917589176,
+      "grad_norm": 0.96875,
+      "learning_rate": 5.413716533899096e-06,
+      "loss": 0.0853,
+      "step": 486
+    },
+    {
+      "epoch": 2.396678966789668,
+      "grad_norm": 1.1171875,
+      "learning_rate": 5.3981383200281004e-06,
+      "loss": 0.1153,
+      "step": 487
+    },
+    {
+      "epoch": 2.40159901599016,
+      "grad_norm": 1.1171875,
+      "learning_rate": 5.382556215694478e-06,
+      "loss": 0.0533,
+      "step": 488
+    },
+    {
+      "epoch": 2.406519065190652,
+      "grad_norm": 1.234375,
+      "learning_rate": 5.366970373160873e-06,
+      "loss": 0.0905,
+      "step": 489
+    },
+    {
+      "epoch": 2.411439114391144,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.351380944726465e-06,
+      "loss": 0.0787,
+      "step": 490
+    },
+    {
+      "epoch": 2.416359163591636,
+      "grad_norm": 1.21875,
+      "learning_rate": 5.335788082725467e-06,
+      "loss": 0.0781,
+      "step": 491
+    },
+    {
+      "epoch": 2.421279212792128,
+      "grad_norm": 0.90234375,
+      "learning_rate": 5.3201919395256475e-06,
+      "loss": 0.0689,
+      "step": 492
+    },
+    {
+      "epoch": 2.42619926199262,
+      "grad_norm": 1.2265625,
+      "learning_rate": 5.304592667526835e-06,
+      "loss": 0.0849,
+      "step": 493
+    },
+    {
+      "epoch": 2.431119311193112,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.288990419159433e-06,
+      "loss": 0.1141,
+      "step": 494
+    },
+    {
+      "epoch": 2.436039360393604,
+      "grad_norm": 0.85546875,
+      "learning_rate": 5.2733853468829295e-06,
+      "loss": 0.0842,
+      "step": 495
+    },
+    {
+      "epoch": 2.440959409594096,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5.257777603184408e-06,
+      "loss": 0.0599,
+      "step": 496
+    },
+    {
+      "epoch": 2.445879458794588,
+      "grad_norm": 0.859375,
+      "learning_rate": 5.24216734057705e-06,
+      "loss": 0.0906,
+      "step": 497
+    },
+    {
+      "epoch": 2.4507995079950797,
+      "grad_norm": 1.84375,
+      "learning_rate": 5.226554711598659e-06,
+      "loss": 0.0934,
+      "step": 498
+    },
+    {
+      "epoch": 2.455719557195572,
+      "grad_norm": 0.75390625,
+      "learning_rate": 5.210939868810156e-06,
+      "loss": 0.0783,
+      "step": 499
+    },
+    {
+      "epoch": 2.460639606396064,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.195322964794098e-06,
+      "loss": 0.0944,
+      "step": 500
+    },
+    {
+      "epoch": 2.465559655596556,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.1797041521531795e-06,
+      "loss": 0.0831,
+      "step": 501
+    },
+    {
+      "epoch": 2.470479704797048,
+      "grad_norm": 1.3828125,
+      "learning_rate": 5.16408358350875e-06,
+      "loss": 0.1166,
+      "step": 502
+    },
+    {
+      "epoch": 2.47539975399754,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.1484614114993156e-06,
+      "loss": 0.089,
+      "step": 503
+    },
+    {
+      "epoch": 2.480319803198032,
+      "grad_norm": 1.546875,
+      "learning_rate": 5.132837788779049e-06,
+      "loss": 0.1219,
+      "step": 504
+    },
+    {
+      "epoch": 2.485239852398524,
+      "grad_norm": 0.95703125,
+      "learning_rate": 5.117212868016303e-06,
+      "loss": 0.0802,
+      "step": 505
+    },
+    {
+      "epoch": 2.490159901599016,
+      "grad_norm": 0.90625,
+      "learning_rate": 5.101586801892109e-06,
+      "loss": 0.0862,
+      "step": 506
+    },
+    {
+      "epoch": 2.495079950799508,
+      "grad_norm": 1.46875,
+      "learning_rate": 5.085959743098693e-06,
+      "loss": 0.0834,
+      "step": 507
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.125,
+      "learning_rate": 5.07033184433798e-06,
+      "loss": 0.0764,
+      "step": 508
+    },
+    {
+      "epoch": 2.504920049200492,
+      "grad_norm": 1.21875,
+      "learning_rate": 5.054703258320107e-06,
+      "loss": 0.0789,
+      "step": 509
+    },
+    {
+      "epoch": 2.509840098400984,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.03907413776192e-06,
+      "loss": 0.0909,
+      "step": 510
+    },
+    {
+      "epoch": 2.509840098400984,
+      "eval_loss": 0.10507026314735413,
+      "eval_runtime": 204.2755,
+      "eval_samples_per_second": 8.009,
+      "eval_steps_per_second": 1.336,
+      "step": 510
+    },
+    {
+      "epoch": 2.514760147601476,
+      "grad_norm": 0.88671875,
+      "learning_rate": 5.0234446353854934e-06,
+      "loss": 0.0751,
+      "step": 511
+    },
+    {
+      "epoch": 2.5196801968019678,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5.00781490391663e-06,
+      "loss": 0.1011,
+      "step": 512
+    },
+    {
+      "epoch": 2.52460024600246,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.992185096083372e-06,
+      "loss": 0.1007,
+      "step": 513
+    },
+    {
+      "epoch": 2.529520295202952,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.976555364614509e-06,
+      "loss": 0.0879,
+      "step": 514
+    },
+    {
+      "epoch": 2.534440344403444,
+      "grad_norm": 1.3203125,
+      "learning_rate": 4.96092586223808e-06,
+      "loss": 0.0806,
+      "step": 515
+    },
+    {
+      "epoch": 2.539360393603936,
+      "grad_norm": 1.1875,
+      "learning_rate": 4.9452967416798945e-06,
+      "loss": 0.0916,
+      "step": 516
+    },
+    {
+      "epoch": 2.544280442804428,
+      "grad_norm": 1.4296875,
+      "learning_rate": 4.929668155662021e-06,
+      "loss": 0.0825,
+      "step": 517
+    },
+    {
+      "epoch": 2.5492004920049203,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.914040256901309e-06,
+      "loss": 0.0761,
+      "step": 518
+    },
+    {
+      "epoch": 2.554120541205412,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.898413198107892e-06,
+      "loss": 0.0657,
+      "step": 519
+    },
+    {
+      "epoch": 2.559040590405904,
+      "grad_norm": 1.625,
+      "learning_rate": 4.882787131983698e-06,
+      "loss": 0.1164,
+      "step": 520
+    },
+    {
+      "epoch": 2.563960639606396,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.867162211220952e-06,
+      "loss": 0.0794,
+      "step": 521
+    },
+    {
+      "epoch": 2.568880688806888,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.851538588500687e-06,
+      "loss": 0.0943,
+      "step": 522
+    },
+    {
+      "epoch": 2.57380073800738,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.835916416491251e-06,
+      "loss": 0.0911,
+      "step": 523
+    },
+    {
+      "epoch": 2.578720787207872,
+      "grad_norm": 0.9140625,
+      "learning_rate": 4.820295847846822e-06,
+      "loss": 0.0579,
+      "step": 524
+    },
+    {
+      "epoch": 2.583640836408364,
+      "grad_norm": 1.125,
+      "learning_rate": 4.804677035205903e-06,
+      "loss": 0.0955,
+      "step": 525
+    },
+    {
+      "epoch": 2.588560885608856,
+      "grad_norm": 1.484375,
+      "learning_rate": 4.789060131189845e-06,
+      "loss": 0.0879,
+      "step": 526
+    },
+    {
+      "epoch": 2.593480934809348,
+      "grad_norm": 1.140625,
+      "learning_rate": 4.773445288401343e-06,
+      "loss": 0.089,
+      "step": 527
+    },
+    {
+      "epoch": 2.59840098400984,
+      "grad_norm": 0.9375,
+      "learning_rate": 4.75783265942295e-06,
+      "loss": 0.0871,
+      "step": 528
+    },
+    {
+      "epoch": 2.603321033210332,
+      "grad_norm": 1.1640625,
+      "learning_rate": 4.742222396815593e-06,
+      "loss": 0.1021,
+      "step": 529
+    },
+    {
+      "epoch": 2.608241082410824,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.726614653117071e-06,
+      "loss": 0.0909,
+      "step": 530
+    },
+    {
+      "epoch": 2.6131611316113164,
+      "grad_norm": 1.234375,
+      "learning_rate": 4.711009580840569e-06,
+      "loss": 0.1082,
+      "step": 531
+    },
+    {
+      "epoch": 2.6180811808118083,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.695407332473166e-06,
+      "loss": 0.083,
+      "step": 532
+    },
+    {
+      "epoch": 2.6230012300123002,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.679808060474354e-06,
+      "loss": 0.0745,
+      "step": 533
+    },
+    {
+      "epoch": 2.627921279212792,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.6642119172745345e-06,
+      "loss": 0.0776,
+      "step": 534
+    },
+    {
+      "epoch": 2.632841328413284,
+      "grad_norm": 0.9453125,
+      "learning_rate": 4.6486190552735375e-06,
+      "loss": 0.0676,
+      "step": 535
+    },
+    {
+      "epoch": 2.637761377613776,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.633029626839128e-06,
+      "loss": 0.0794,
+      "step": 536
+    },
+    {
+      "epoch": 2.642681426814268,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.617443784305524e-06,
+      "loss": 0.1136,
+      "step": 537
+    },
+    {
+      "epoch": 2.64760147601476,
+      "grad_norm": 1.1875,
+      "learning_rate": 4.601861679971901e-06,
+      "loss": 0.0747,
+      "step": 538
+    },
+    {
+      "epoch": 2.6525215252152523,
+      "grad_norm": 1.625,
+      "learning_rate": 4.5862834661009074e-06,
+      "loss": 0.0982,
+      "step": 539
+    },
+    {
+      "epoch": 2.6574415744157442,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.57070929491717e-06,
+      "loss": 0.0788,
+      "step": 540
+    },
+    {
+      "epoch": 2.662361623616236,
+      "grad_norm": 1.4765625,
+      "learning_rate": 4.555139318605821e-06,
+      "loss": 0.0857,
+      "step": 541
+    },
+    {
+      "epoch": 2.667281672816728,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.539573689310995e-06,
+      "loss": 0.09,
+      "step": 542
+    },
+    {
+      "epoch": 2.67220172201722,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.524012559134352e-06,
+      "loss": 0.0934,
+      "step": 543
+    },
+    {
+      "epoch": 2.6771217712177124,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.508456080133584e-06,
+      "loss": 0.0742,
+      "step": 544
+    },
+    {
+      "epoch": 2.6820418204182044,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.492904404320942e-06,
+      "loss": 0.0677,
+      "step": 545
+    },
+    {
+      "epoch": 2.6869618696186963,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.477357683661734e-06,
+      "loss": 0.072,
+      "step": 546
+    },
+    {
+      "epoch": 2.6918819188191883,
+      "grad_norm": 1.2578125,
+      "learning_rate": 4.461816070072851e-06,
+      "loss": 0.0948,
+      "step": 547
+    },
+    {
+      "epoch": 2.69680196801968,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.446279715421277e-06,
+      "loss": 0.0846,
+      "step": 548
+    },
+    {
+      "epoch": 2.701722017220172,
+      "grad_norm": 1.3828125,
+      "learning_rate": 4.430748771522615e-06,
+      "loss": 0.0952,
+      "step": 549
+    },
+    {
+      "epoch": 2.706642066420664,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.415223390139588e-06,
+      "loss": 0.0797,
+      "step": 550
+    },
+    {
+      "epoch": 2.711562115621156,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.399703722980569e-06,
+      "loss": 0.0822,
+      "step": 551
+    },
+    {
+      "epoch": 2.716482164821648,
+      "grad_norm": 1.328125,
+      "learning_rate": 4.3841899216980895e-06,
+      "loss": 0.1041,
+      "step": 552
+    },
+    {
+      "epoch": 2.7214022140221403,
+      "grad_norm": 1.21875,
+      "learning_rate": 4.368682137887365e-06,
+      "loss": 0.0884,
+      "step": 553
+    },
+    {
+      "epoch": 2.7263222632226323,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.3531805230848116e-06,
+      "loss": 0.0799,
+      "step": 554
+    },
+    {
+      "epoch": 2.731242312423124,
+      "grad_norm": 1.375,
+      "learning_rate": 4.337685228766561e-06,
+      "loss": 0.1055,
+      "step": 555
+    },
+    {
+      "epoch": 2.736162361623616,
+      "grad_norm": 1.15625,
+      "learning_rate": 4.322196406346984e-06,
+      "loss": 0.1206,
+      "step": 556
+    },
+    {
+      "epoch": 2.741082410824108,
+      "grad_norm": 1.2265625,
+      "learning_rate": 4.30671420717721e-06,
+      "loss": 0.085,
+      "step": 557
+    },
+    {
+      "epoch": 2.7460024600246005,
+      "grad_norm": 1.4765625,
+      "learning_rate": 4.291238782543652e-06,
+      "loss": 0.1146,
+      "step": 558
+    },
+    {
+      "epoch": 2.7509225092250924,
+      "grad_norm": 1.125,
+      "learning_rate": 4.275770283666521e-06,
+      "loss": 0.0742,
+      "step": 559
+    },
+    {
+      "epoch": 2.7558425584255843,
+      "grad_norm": 1.2109375,
+      "learning_rate": 4.260308861698351e-06,
+      "loss": 0.091,
+      "step": 560
+    },
+    {
+      "epoch": 2.7607626076260763,
+      "grad_norm": 1.171875,
+      "learning_rate": 4.244854667722527e-06,
+      "loss": 0.0741,
+      "step": 561
+    },
+    {
+      "epoch": 2.765682656826568,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.229407852751806e-06,
+      "loss": 0.0738,
+      "step": 562
+    },
+    {
+      "epoch": 2.77060270602706,
+      "grad_norm": 1.3359375,
+      "learning_rate": 4.21396856772683e-06,
+      "loss": 0.0999,
+      "step": 563
+    },
+    {
+      "epoch": 2.775522755227552,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.198536963514674e-06,
+      "loss": 0.0619,
+      "step": 564
+    },
+    {
+      "epoch": 2.780442804428044,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.183113190907349e-06,
+      "loss": 0.0841,
+      "step": 565
+    },
+    {
+      "epoch": 2.7853628536285364,
+      "grad_norm": 0.96484375,
+      "learning_rate": 4.1676974006203456e-06,
+      "loss": 0.0738,
+      "step": 566
+    },
+    {
+      "epoch": 2.7902829028290284,
+      "grad_norm": 1.21875,
+      "learning_rate": 4.152289743291148e-06,
+      "loss": 0.1018,
+      "step": 567
+    },
+    {
+      "epoch": 2.7952029520295203,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.136890369477773e-06,
+      "loss": 0.0664,
+      "step": 568
+    },
+    {
+      "epoch": 2.8001230012300122,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.121499429657287e-06,
+      "loss": 0.0769,
+      "step": 569
+    },
+    {
+      "epoch": 2.805043050430504,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.106117074224354e-06,
+      "loss": 0.0779,
+      "step": 570
+    },
+    {
+      "epoch": 2.8099630996309966,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.090743453489744e-06,
+      "loss": 0.0958,
+      "step": 571
+    },
+    {
+      "epoch": 2.8148831488314885,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.07537871767888e-06,
+      "loss": 0.1121,
+      "step": 572
+    },
+    {
+      "epoch": 2.8198031980319804,
+      "grad_norm": 1.4375,
+      "learning_rate": 4.060023016930359e-06,
+      "loss": 0.1435,
+      "step": 573
+    },
+    {
+      "epoch": 2.8247232472324724,
+      "grad_norm": 1.2265625,
+      "learning_rate": 4.044676501294501e-06,
+      "loss": 0.1366,
+      "step": 574
+    },
+    {
+      "epoch": 2.8296432964329643,
+      "grad_norm": 1.171875,
+      "learning_rate": 4.029339320731862e-06,
+      "loss": 0.0755,
+      "step": 575
+    },
+    {
+      "epoch": 2.8345633456334562,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.0140116251117865e-06,
+      "loss": 0.0837,
+      "step": 576
+    },
+    {
+      "epoch": 2.839483394833948,
+      "grad_norm": 1.25,
+      "learning_rate": 3.998693564210929e-06,
+      "loss": 0.1097,
+      "step": 577
+    },
+    {
+      "epoch": 2.84440344403444,
+      "grad_norm": 1.0078125,
+      "learning_rate": 3.983385287711803e-06,
+      "loss": 0.0652,
+      "step": 578
+    },
+    {
+      "epoch": 2.8493234932349325,
+      "grad_norm": 0.94921875,
+      "learning_rate": 3.96808694520131e-06,
+      "loss": 0.0785,
+      "step": 579
+    },
+    {
+      "epoch": 2.8542435424354244,
+      "grad_norm": 1.3125,
+      "learning_rate": 3.952798686169279e-06,
+      "loss": 0.0947,
+      "step": 580
+    },
+    {
+      "epoch": 2.8591635916359164,
+      "grad_norm": 1.1796875,
+      "learning_rate": 3.937520660007008e-06,
+      "loss": 0.0803,
+      "step": 581
+    },
+    {
+      "epoch": 2.8640836408364083,
+      "grad_norm": 0.8203125,
+      "learning_rate": 3.9222530160058025e-06,
+      "loss": 0.0664,
+      "step": 582
+    },
+    {
+      "epoch": 2.8690036900369003,
+      "grad_norm": 1.0703125,
+      "learning_rate": 3.90699590335552e-06,
+      "loss": 0.0737,
+      "step": 583
+    },
+    {
+      "epoch": 2.8739237392373926,
+      "grad_norm": 1.6875,
+      "learning_rate": 3.891749471143106e-06,
+      "loss": 0.1152,
+      "step": 584
+    },
+    {
+      "epoch": 2.8788437884378846,
+      "grad_norm": 1.0,
+      "learning_rate": 3.876513868351142e-06,
+      "loss": 0.0932,
+      "step": 585
+    },
+    {
+      "epoch": 2.8837638376383765,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.861289243856388e-06,
+      "loss": 0.1052,
+      "step": 586
+    },
+    {
+      "epoch": 2.8886838868388685,
+      "grad_norm": 1.5234375,
+      "learning_rate": 3.84607574642833e-06,
+      "loss": 0.1152,
+      "step": 587
+    },
+    {
+      "epoch": 2.8936039360393604,
+      "grad_norm": 1.1796875,
+      "learning_rate": 3.830873524727722e-06,
+      "loss": 0.083,
+      "step": 588
+    },
+    {
+      "epoch": 2.8985239852398523,
+      "grad_norm": 1.2109375,
+      "learning_rate": 3.815682727305136e-06,
+      "loss": 0.0824,
+      "step": 589
+    },
+    {
+      "epoch": 2.9034440344403443,
+      "grad_norm": 1.171875,
+      "learning_rate": 3.800503502599511e-06,
+      "loss": 0.0933,
+      "step": 590
+    },
+    {
+      "epoch": 2.908364083640836,
+      "grad_norm": 0.91015625,
+      "learning_rate": 3.7853359989367023e-06,
+      "loss": 0.0831,
+      "step": 591
+    },
+    {
+      "epoch": 2.913284132841328,
+      "grad_norm": 1.1484375,
+      "learning_rate": 3.7701803645280296e-06,
+      "loss": 0.0954,
+      "step": 592
+    },
+    {
+      "epoch": 2.9182041820418205,
+      "grad_norm": 1.046875,
+      "learning_rate": 3.7550367474688315e-06,
+      "loss": 0.0829,
+      "step": 593
+    },
+    {
+      "epoch": 2.9231242312423125,
+      "grad_norm": 1.703125,
+      "learning_rate": 3.739905295737015e-06,
+      "loss": 0.0905,
+      "step": 594
+    },
+    {
+      "epoch": 2.9280442804428044,
+      "grad_norm": 1.3125,
+      "learning_rate": 3.7247861571916183e-06,
+      "loss": 0.0838,
+      "step": 595
+    },
+    {
+      "epoch": 2.9329643296432963,
+      "grad_norm": 1.2109375,
+      "learning_rate": 3.7096794795713542e-06,
+      "loss": 0.1098,
+      "step": 596
+    },
+    {
+      "epoch": 2.9378843788437883,
+      "grad_norm": 1.140625,
+      "learning_rate": 3.6945854104931726e-06,
+      "loss": 0.0776,
+      "step": 597
+    },
+    {
+      "epoch": 2.9428044280442807,
+      "grad_norm": 1.203125,
+      "learning_rate": 3.6795040974508164e-06,
+      "loss": 0.0694,
+      "step": 598
+    },
+    {
+      "epoch": 2.9477244772447726,
+      "grad_norm": 1.2890625,
+      "learning_rate": 3.6644356878133862e-06,
+      "loss": 0.0996,
+      "step": 599
+    },
+    {
+      "epoch": 2.9526445264452645,
+      "grad_norm": 1.1953125,
+      "learning_rate": 3.6493803288238894e-06,
+      "loss": 0.0801,
+      "step": 600
+    },
+    {
+      "epoch": 2.9575645756457565,
+      "grad_norm": 1.265625,
+      "learning_rate": 3.634338167597812e-06,
+      "loss": 0.0904,
+      "step": 601
+    },
+    {
+      "epoch": 2.9624846248462484,
+      "grad_norm": 1.3125,
+      "learning_rate": 3.61930935112167e-06,
+      "loss": 0.0676,
+      "step": 602
+    },
+    {
+      "epoch": 2.9674046740467404,
+      "grad_norm": 1.265625,
+      "learning_rate": 3.6042940262515867e-06,
+      "loss": 0.0985,
+      "step": 603
+    },
+    {
+      "epoch": 2.9723247232472323,
+      "grad_norm": 1.109375,
+      "learning_rate": 3.5892923397118473e-06,
+      "loss": 0.0816,
+      "step": 604
+    },
+    {
+      "epoch": 2.9772447724477242,
+      "grad_norm": 1.1328125,
+      "learning_rate": 3.5743044380934655e-06,
+      "loss": 0.076,
+      "step": 605
+    },
+    {
+      "epoch": 2.9821648216482166,
+      "grad_norm": 0.97265625,
+      "learning_rate": 3.5593304678527547e-06,
+      "loss": 0.0932,
+      "step": 606
+    },
+    {
+      "epoch": 2.9870848708487086,
+      "grad_norm": 1.140625,
+      "learning_rate": 3.544370575309902e-06,
+      "loss": 0.0982,
+      "step": 607
+    },
+    {
+      "epoch": 2.9920049200492005,
+      "grad_norm": 1.15625,
+      "learning_rate": 3.5294249066475245e-06,
+      "loss": 0.0997,
+      "step": 608
+    },
+    {
+      "epoch": 2.9969249692496924,
+      "grad_norm": 1.1953125,
+      "learning_rate": 3.5144936079092528e-06,
+      "loss": 0.0907,
+      "step": 609
+    },
+    {
+      "epoch": 3.002460024600246,
+      "grad_norm": 7.90625,
+      "learning_rate": 3.4995768249982975e-06,
+      "loss": 0.1999,
+      "step": 610
+    },
+    {
+      "epoch": 3.007380073800738,
+      "grad_norm": 1.0546875,
+      "learning_rate": 3.4846747036760285e-06,
+      "loss": 0.0815,
+      "step": 611
+    },
+    {
+      "epoch": 3.01230012300123,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.4697873895605466e-06,
+      "loss": 0.0754,
+      "step": 612
+    },
+    {
+      "epoch": 3.01230012300123,
+      "eval_loss": 0.1035788357257843,
+      "eval_runtime": 204.6775,
+      "eval_samples_per_second": 7.993,
+      "eval_steps_per_second": 1.334,
+      "step": 612
+    },
+    {
+      "epoch": 3.0172201722017222,
+      "grad_norm": 1.1796875,
+      "learning_rate": 3.4549150281252635e-06,
+      "loss": 0.1022,
+      "step": 613
+    },
+    {
+      "epoch": 3.022140221402214,
+      "grad_norm": 0.953125,
+      "learning_rate": 3.4400577646974766e-06,
+      "loss": 0.074,
+      "step": 614
+    },
+    {
+      "epoch": 3.027060270602706,
+      "grad_norm": 0.98046875,
+      "learning_rate": 3.4252157444569478e-06,
+      "loss": 0.0619,
+      "step": 615
+    },
+    {
+      "epoch": 3.031980319803198,
+      "grad_norm": 1.2265625,
+      "learning_rate": 3.410389112434499e-06,
+      "loss": 0.0887,
+      "step": 616
+    },
+    {
+      "epoch": 3.03690036900369,
+      "grad_norm": 0.87109375,
+      "learning_rate": 3.3955780135105736e-06,
+      "loss": 0.0831,
+      "step": 617
+    },
+    {
+      "epoch": 3.041820418204182,
+      "grad_norm": 1.0625,
+      "learning_rate": 3.3807825924138356e-06,
+      "loss": 0.1026,
+      "step": 618
+    },
+    {
+      "epoch": 3.046740467404674,
+      "grad_norm": 1.1953125,
+      "learning_rate": 3.366002993719747e-06,
+      "loss": 0.0814,
+      "step": 619
+    },
+    {
+      "epoch": 3.0516605166051662,
+      "grad_norm": 1.1015625,
+      "learning_rate": 3.351239361849168e-06,
+      "loss": 0.1059,
+      "step": 620
+    },
+    {
+      "epoch": 3.056580565805658,
+      "grad_norm": 0.82421875,
+      "learning_rate": 3.336491841066928e-06,
+      "loss": 0.083,
+      "step": 621
+    },
+    {
+      "epoch": 3.06150061500615,
+      "grad_norm": 1.109375,
+      "learning_rate": 3.3217605754804273e-06,
+      "loss": 0.0612,
+      "step": 622
+    },
+    {
+      "epoch": 3.066420664206642,
+      "grad_norm": 1.1015625,
+      "learning_rate": 3.307045709038226e-06,
+      "loss": 0.0833,
+      "step": 623
+    },
+    {
+      "epoch": 3.071340713407134,
+      "grad_norm": 1.0078125,
+      "learning_rate": 3.2923473855286426e-06,
+      "loss": 0.0717,
+      "step": 624
+    },
+    {
+      "epoch": 3.076260762607626,
+      "grad_norm": 0.8671875,
+      "learning_rate": 3.2776657485783357e-06,
+      "loss": 0.0752,
+      "step": 625
+    },
+    {
+      "epoch": 3.081180811808118,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.2630009416509167e-06,
+      "loss": 0.0891,
+      "step": 626
+    },
+    {
+      "epoch": 3.0861008610086103,
+      "grad_norm": 1.2109375,
+      "learning_rate": 3.2483531080455334e-06,
+      "loss": 0.1059,
+      "step": 627
+    },
+    {
+      "epoch": 3.091020910209102,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.2337223908954834e-06,
+      "loss": 0.0874,
+      "step": 628
+    },
+    {
+      "epoch": 3.095940959409594,
+      "grad_norm": 1.0078125,
+      "learning_rate": 3.2191089331668036e-06,
+      "loss": 0.0814,
+      "step": 629
+    },
+    {
+      "epoch": 3.100861008610086,
+      "grad_norm": 1.1875,
+      "learning_rate": 3.2045128776568783e-06,
+      "loss": 0.0982,
+      "step": 630
+    },
+    {
+      "epoch": 3.105781057810578,
+      "grad_norm": 0.7265625,
+      "learning_rate": 3.1899343669930446e-06,
+      "loss": 0.0647,
+      "step": 631
+    },
+    {
+      "epoch": 3.11070110701107,
+      "grad_norm": 1.2734375,
+      "learning_rate": 3.1753735436312005e-06,
+      "loss": 0.1091,
+      "step": 632
+    },
+    {
+      "epoch": 3.1156211562115623,
+      "grad_norm": 1.046875,
+      "learning_rate": 3.1608305498544056e-06,
+      "loss": 0.0939,
+      "step": 633
+    },
+    {
+      "epoch": 3.1205412054120543,
+      "grad_norm": 1.0390625,
+      "learning_rate": 3.146305527771499e-06,
+      "loss": 0.074,
+      "step": 634
+    },
+    {
+      "epoch": 3.125461254612546,
+      "grad_norm": 1.03125,
+      "learning_rate": 3.1317986193157023e-06,
+      "loss": 0.0734,
+      "step": 635
+    },
+    {
+      "epoch": 3.130381303813038,
+      "grad_norm": 1.3203125,
+      "learning_rate": 3.1173099662432426e-06,
+      "loss": 0.1068,
+      "step": 636
+    },
+    {
+      "epoch": 3.13530135301353,
+      "grad_norm": 1.484375,
+      "learning_rate": 3.1028397101319584e-06,
+      "loss": 0.0917,
+      "step": 637
+    },
+    {
+      "epoch": 3.140221402214022,
+      "grad_norm": 1.2265625,
+      "learning_rate": 3.0883879923799244e-06,
+      "loss": 0.0743,
+      "step": 638
+    },
+    {
+      "epoch": 3.145141451414514,
+      "grad_norm": 1.2734375,
+      "learning_rate": 3.0739549542040583e-06,
+      "loss": 0.0971,
+      "step": 639
+    },
+    {
+      "epoch": 3.1500615006150063,
+      "grad_norm": 1.0625,
+      "learning_rate": 3.059540736638751e-06,
+      "loss": 0.1033,
+      "step": 640
+    },
+    {
+      "epoch": 3.1549815498154983,
+      "grad_norm": 0.94921875,
+      "learning_rate": 3.0451454805344893e-06,
+      "loss": 0.0899,
+      "step": 641
+    },
+    {
+      "epoch": 3.15990159901599,
+      "grad_norm": 1.0859375,
+      "learning_rate": 3.0307693265564708e-06,
+      "loss": 0.0845,
+      "step": 642
+    },
+    {
+      "epoch": 3.164821648216482,
+      "grad_norm": 1.1484375,
+      "learning_rate": 3.016412415183233e-06,
+      "loss": 0.0914,
+      "step": 643
+    },
+    {
+      "epoch": 3.169741697416974,
+      "grad_norm": 1.1796875,
+      "learning_rate": 3.002074886705284e-06,
+      "loss": 0.0802,
+      "step": 644
+    },
+    {
+      "epoch": 3.174661746617466,
+      "grad_norm": 1.3125,
+      "learning_rate": 2.9877568812237325e-06,
+      "loss": 0.0946,
+      "step": 645
+    },
+    {
+      "epoch": 3.179581795817958,
+      "grad_norm": 1.2265625,
+      "learning_rate": 2.9734585386489095e-06,
+      "loss": 0.0782,
+      "step": 646
+    },
+    {
+      "epoch": 3.1845018450184504,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.9591799986990098e-06,
+      "loss": 0.0594,
+      "step": 647
+    },
+    {
+      "epoch": 3.1894218942189423,
+      "grad_norm": 1.3515625,
+      "learning_rate": 2.9449214008987253e-06,
+      "loss": 0.0794,
+      "step": 648
+    },
+    {
+      "epoch": 3.1943419434194342,
+      "grad_norm": 1.1640625,
+      "learning_rate": 2.93068288457788e-06,
+      "loss": 0.0818,
+      "step": 649
+    },
+    {
+      "epoch": 3.199261992619926,
+      "grad_norm": 1.1796875,
+      "learning_rate": 2.916464588870067e-06,
+      "loss": 0.1144,
+      "step": 650
+    },
+    {
+      "epoch": 3.204182041820418,
+      "grad_norm": 1.265625,
+      "learning_rate": 2.9022666527112954e-06,
+      "loss": 0.093,
+      "step": 651
+    },
+    {
+      "epoch": 3.20910209102091,
+      "grad_norm": 0.98828125,
+      "learning_rate": 2.8880892148386198e-06,
+      "loss": 0.0725,
+      "step": 652
+    },
+    {
+      "epoch": 3.2140221402214024,
+      "grad_norm": 1.1796875,
+      "learning_rate": 2.873932413788805e-06,
+      "loss": 0.068,
+      "step": 653
+    },
+    {
+      "epoch": 3.2189421894218944,
+      "grad_norm": 0.96484375,
+      "learning_rate": 2.859796387896949e-06,
+      "loss": 0.0583,
+      "step": 654
+    },
+    {
+      "epoch": 3.2238622386223863,
+      "grad_norm": 1.078125,
+      "learning_rate": 2.8456812752951483e-06,
+      "loss": 0.0748,
+      "step": 655
+    },
+    {
+      "epoch": 3.2287822878228782,
+      "grad_norm": 1.3984375,
+      "learning_rate": 2.831587213911142e-06,
+      "loss": 0.1128,
+      "step": 656
+    },
+    {
+      "epoch": 3.23370233702337,
+      "grad_norm": 1.328125,
+      "learning_rate": 2.817514341466965e-06,
+      "loss": 0.0812,
+      "step": 657
+    },
+    {
+      "epoch": 3.238622386223862,
+      "grad_norm": 1.1875,
+      "learning_rate": 2.8034627954775993e-06,
+      "loss": 0.0995,
+      "step": 658
+    },
+    {
+      "epoch": 3.243542435424354,
+      "grad_norm": 1.0546875,
+      "learning_rate": 2.7894327132496324e-06,
+      "loss": 0.0865,
+      "step": 659
+    },
+    {
+      "epoch": 3.2484624846248464,
+      "grad_norm": 1.171875,
+      "learning_rate": 2.7754242318799174e-06,
+      "loss": 0.0807,
+      "step": 660
+    },
+    {
+      "epoch": 3.2533825338253384,
+      "grad_norm": 1.015625,
+      "learning_rate": 2.761437488254232e-06,
+      "loss": 0.1041,
+      "step": 661
+    },
+    {
+      "epoch": 3.2583025830258303,
+      "grad_norm": 1.0703125,
+      "learning_rate": 2.74747261904594e-06,
+      "loss": 0.074,
+      "step": 662
+    },
+    {
+      "epoch": 3.2632226322263223,
+      "grad_norm": 1.203125,
+      "learning_rate": 2.733529760714655e-06,
+      "loss": 0.0852,
+      "step": 663
+    },
+    {
+      "epoch": 3.268142681426814,
+      "grad_norm": 1.3515625,
+      "learning_rate": 2.719609049504911e-06,
+      "loss": 0.1049,
+      "step": 664
+    },
+    {
+      "epoch": 3.273062730627306,
+      "grad_norm": 1.2265625,
+      "learning_rate": 2.7057106214448216e-06,
+      "loss": 0.0682,
+      "step": 665
+    },
+    {
+      "epoch": 3.2779827798277985,
+      "grad_norm": 1.3828125,
+      "learning_rate": 2.6918346123447708e-06,
+      "loss": 0.1297,
+      "step": 666
+    },
+    {
+      "epoch": 3.2829028290282904,
+      "grad_norm": 1.1328125,
+      "learning_rate": 2.677981157796059e-06,
+      "loss": 0.0748,
+      "step": 667
+    },
+    {
+      "epoch": 3.2878228782287824,
+      "grad_norm": 1.125,
+      "learning_rate": 2.6641503931696e-06,
+      "loss": 0.1069,
+      "step": 668
+    },
+    {
+      "epoch": 3.2927429274292743,
+      "grad_norm": 1.0703125,
+      "learning_rate": 2.650342453614586e-06,
+      "loss": 0.0802,
+      "step": 669
+    },
+    {
+      "epoch": 3.2976629766297663,
+      "grad_norm": 1.1796875,
+      "learning_rate": 2.636557474057173e-06,
+      "loss": 0.0908,
+      "step": 670
+    },
+    {
+      "epoch": 3.302583025830258,
+      "grad_norm": 0.91015625,
+      "learning_rate": 2.6227955891991617e-06,
+      "loss": 0.0663,
+      "step": 671
+    },
+    {
+      "epoch": 3.30750307503075,
+      "grad_norm": 1.390625,
+      "learning_rate": 2.609056933516675e-06,
+      "loss": 0.0686,
+      "step": 672
+    },
+    {
+      "epoch": 3.312423124231242,
+      "grad_norm": 1.0390625,
+      "learning_rate": 2.5953416412588504e-06,
+      "loss": 0.0833,
+      "step": 673
+    },
+    {
+      "epoch": 3.3173431734317345,
+      "grad_norm": 1.1328125,
+      "learning_rate": 2.58164984644653e-06,
+      "loss": 0.0934,
+      "step": 674
+    },
+    {
+      "epoch": 3.3222632226322264,
+      "grad_norm": 1.1796875,
+      "learning_rate": 2.567981682870946e-06,
+      "loss": 0.0953,
+      "step": 675
+    },
+    {
+      "epoch": 3.3271832718327183,
+      "grad_norm": 1.1328125,
+      "learning_rate": 2.554337284092411e-06,
+      "loss": 0.1022,
+      "step": 676
+    },
+    {
+      "epoch": 3.3321033210332103,
+      "grad_norm": 1.1953125,
+      "learning_rate": 2.540716783439019e-06,
+      "loss": 0.0674,
+      "step": 677
+    },
+    {
+      "epoch": 3.337023370233702,
+      "grad_norm": 1.28125,
+      "learning_rate": 2.5271203140053436e-06,
+      "loss": 0.1064,
+      "step": 678
+    },
+    {
+      "epoch": 3.341943419434194,
+      "grad_norm": 1.1796875,
+      "learning_rate": 2.5135480086511306e-06,
+      "loss": 0.0989,
+      "step": 679
+    },
+    {
+      "epoch": 3.3468634686346865,
+      "grad_norm": 1.25,
+      "learning_rate": 2.5000000000000015e-06,
+      "loss": 0.0908,
+      "step": 680
+    },
+    {
+      "epoch": 3.3517835178351785,
+      "grad_norm": 1.2421875,
+      "learning_rate": 2.4864764204381624e-06,
+      "loss": 0.0834,
+      "step": 681
+    },
+    {
+      "epoch": 3.3567035670356704,
+      "grad_norm": 1.4921875,
+      "learning_rate": 2.472977402113107e-06,
+      "loss": 0.0902,
+      "step": 682
+    },
+    {
+      "epoch": 3.3616236162361623,
+      "grad_norm": 1.1875,
+      "learning_rate": 2.4595030769323246e-06,
+      "loss": 0.0718,
+      "step": 683
+    },
+    {
+      "epoch": 3.3665436654366543,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.4460535765620147e-06,
+      "loss": 0.1142,
+      "step": 684
+    },
+    {
+      "epoch": 3.3714637146371462,
+      "grad_norm": 1.1953125,
+      "learning_rate": 2.4326290324257896e-06,
+      "loss": 0.0951,
+      "step": 685
+    },
+    {
+      "epoch": 3.376383763837638,
+      "grad_norm": 1.328125,
+      "learning_rate": 2.419229575703411e-06,
+      "loss": 0.0985,
+      "step": 686
+    },
+    {
+      "epoch": 3.3813038130381305,
+      "grad_norm": 1.109375,
+      "learning_rate": 2.4058553373294846e-06,
+      "loss": 0.0713,
+      "step": 687
+    },
+    {
+      "epoch": 3.3862238622386225,
+      "grad_norm": 1.3359375,
+      "learning_rate": 2.3925064479921985e-06,
+      "loss": 0.0882,
+      "step": 688
+    },
+    {
+      "epoch": 3.3911439114391144,
+      "grad_norm": 1.2109375,
+      "learning_rate": 2.379183038132031e-06,
+      "loss": 0.1156,
+      "step": 689
+    },
+    {
+      "epoch": 3.3960639606396064,
+      "grad_norm": 1.171875,
+      "learning_rate": 2.3658852379404973e-06,
+      "loss": 0.0914,
+      "step": 690
+    },
+    {
+      "epoch": 3.4009840098400983,
+      "grad_norm": 1.1953125,
+      "learning_rate": 2.352613177358852e-06,
+      "loss": 0.0794,
+      "step": 691
+    },
+    {
+      "epoch": 3.4059040590405902,
+      "grad_norm": 1.0859375,
+      "learning_rate": 2.3393669860768364e-06,
+      "loss": 0.0941,
+      "step": 692
+    },
+    {
+      "epoch": 3.4108241082410826,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.3261467935314097e-06,
+      "loss": 0.0648,
+      "step": 693
+    },
+    {
+      "epoch": 3.4157441574415746,
+      "grad_norm": 1.25,
+      "learning_rate": 2.3129527289054716e-06,
+      "loss": 0.0912,
+      "step": 694
+    },
+    {
+      "epoch": 3.4206642066420665,
+      "grad_norm": 1.0390625,
+      "learning_rate": 2.299784921126622e-06,
+      "loss": 0.0801,
+      "step": 695
+    },
+    {
+      "epoch": 3.4255842558425584,
+      "grad_norm": 1.078125,
+      "learning_rate": 2.286643498865877e-06,
+      "loss": 0.072,
+      "step": 696
+    },
+    {
+      "epoch": 3.4305043050430504,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.2735285905364307e-06,
+      "loss": 0.0599,
+      "step": 697
+    },
+    {
+      "epoch": 3.4354243542435423,
+      "grad_norm": 1.015625,
+      "learning_rate": 2.260440324292385e-06,
+      "loss": 0.0748,
+      "step": 698
+    },
+    {
+      "epoch": 3.4403444034440342,
+      "grad_norm": 0.95703125,
+      "learning_rate": 2.2473788280275164e-06,
+      "loss": 0.0613,
+      "step": 699
+    },
+    {
+      "epoch": 3.4452644526445266,
+      "grad_norm": 1.390625,
+      "learning_rate": 2.234344229374003e-06,
+      "loss": 0.107,
+      "step": 700
+    },
+    {
+      "epoch": 3.4501845018450186,
+      "grad_norm": 1.109375,
+      "learning_rate": 2.2213366557011988e-06,
+      "loss": 0.083,
+      "step": 701
+    },
+    {
+      "epoch": 3.4551045510455105,
+      "grad_norm": 0.953125,
+      "learning_rate": 2.2083562341143695e-06,
+      "loss": 0.0823,
+      "step": 702
+    },
+    {
+      "epoch": 3.4600246002460024,
+      "grad_norm": 1.4609375,
+      "learning_rate": 2.195403091453473e-06,
+      "loss": 0.1011,
+      "step": 703
+    },
+    {
+      "epoch": 3.4649446494464944,
+      "grad_norm": 1.171875,
+      "learning_rate": 2.1824773542918957e-06,
+      "loss": 0.0832,
+      "step": 704
+    },
+    {
+      "epoch": 3.4698646986469863,
+      "grad_norm": 1.5859375,
+      "learning_rate": 2.1695791489352346e-06,
+      "loss": 0.1078,
+      "step": 705
+    },
+    {
+      "epoch": 3.4747847478474787,
+      "grad_norm": 0.8984375,
+      "learning_rate": 2.156708601420053e-06,
+      "loss": 0.0529,
+      "step": 706
+    },
+    {
+      "epoch": 3.4797047970479706,
+      "grad_norm": 1.0546875,
+      "learning_rate": 2.1438658375126544e-06,
+      "loss": 0.0788,
+      "step": 707
+    },
+    {
+      "epoch": 3.4846248462484626,
+      "grad_norm": 1.15625,
+      "learning_rate": 2.131050982707849e-06,
+      "loss": 0.0749,
+      "step": 708
+    },
+    {
+      "epoch": 3.4895448954489545,
+      "grad_norm": 1.296875,
+      "learning_rate": 2.1182641622277273e-06,
+      "loss": 0.0666,
+      "step": 709
+    },
+    {
+      "epoch": 3.4944649446494465,
+      "grad_norm": 1.0,
+      "learning_rate": 2.1055055010204427e-06,
+      "loss": 0.0892,
+      "step": 710
+    },
+    {
+      "epoch": 3.4993849938499384,
+      "grad_norm": 1.203125,
+      "learning_rate": 2.092775123758985e-06,
+      "loss": 0.1143,
+      "step": 711
+    },
+    {
+      "epoch": 3.5043050430504303,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.080073154839964e-06,
+      "loss": 0.0703,
+      "step": 712
+    },
+    {
+      "epoch": 3.5092250922509223,
+      "grad_norm": 1.0703125,
+      "learning_rate": 2.06739971838239e-06,
+      "loss": 0.083,
+      "step": 713
+    },
+    {
+      "epoch": 3.5141451414514147,
+      "grad_norm": 1.09375,
+      "learning_rate": 2.05475493822647e-06,
+      "loss": 0.0697,
+      "step": 714
+    },
+    {
+      "epoch": 3.5141451414514147,
+      "eval_loss": 0.10337568074464798,
+      "eval_runtime": 203.8015,
+      "eval_samples_per_second": 8.027,
+      "eval_steps_per_second": 1.34,
+      "step": 714
+    },
+    {
+      "epoch": 3.5190651906519066,
+      "grad_norm": 1.375,
+      "learning_rate": 2.042138937932388e-06,
+      "loss": 0.0886,
+      "step": 715
+    },
+    {
+      "epoch": 3.5239852398523985,
+      "grad_norm": 1.21875,
+      "learning_rate": 2.0295518407791054e-06,
+      "loss": 0.1,
+      "step": 716
+    },
+    {
+      "epoch": 3.5289052890528905,
+      "grad_norm": 1.1484375,
+      "learning_rate": 2.016993769763147e-06,
+      "loss": 0.0802,
+      "step": 717
+    },
+    {
+      "epoch": 3.5338253382533824,
+      "grad_norm": 0.98828125,
+      "learning_rate": 2.0044648475974095e-06,
+      "loss": 0.058,
+      "step": 718
+    },
+    {
+      "epoch": 3.538745387453875,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.9919651967099556e-06,
+      "loss": 0.0875,
+      "step": 719
+    },
+    {
+      "epoch": 3.5436654366543667,
+      "grad_norm": 1.25,
+      "learning_rate": 1.979494939242822e-06,
+      "loss": 0.0656,
+      "step": 720
+    },
+    {
+      "epoch": 3.5485854858548587,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.9670541970508224e-06,
+      "loss": 0.0676,
+      "step": 721
+    },
+    {
+      "epoch": 3.5535055350553506,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.954643091700354e-06,
+      "loss": 0.0672,
+      "step": 722
+    },
+    {
+      "epoch": 3.5584255842558425,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.9422617444682196e-06,
+      "loss": 0.0816,
+      "step": 723
+    },
+    {
+      "epoch": 3.5633456334563345,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.9299102763404335e-06,
+      "loss": 0.0769,
+      "step": 724
+    },
+    {
+      "epoch": 3.5682656826568264,
+      "grad_norm": 0.81640625,
+      "learning_rate": 1.917588808011045e-06,
+      "loss": 0.069,
+      "step": 725
+    },
+    {
+      "epoch": 3.5731857318573184,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.9052974598809505e-06,
+      "loss": 0.075,
+      "step": 726
+    },
+    {
+      "epoch": 3.5781057810578107,
+      "grad_norm": 1.34375,
+      "learning_rate": 1.893036352056728e-06,
+      "loss": 0.1036,
+      "step": 727
+    },
+    {
+      "epoch": 3.5830258302583027,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.880805604349456e-06,
+      "loss": 0.0658,
+      "step": 728
+    },
+    {
+      "epoch": 3.5879458794587946,
+      "grad_norm": 0.8046875,
+      "learning_rate": 1.8686053362735468e-06,
+      "loss": 0.052,
+      "step": 729
+    },
+    {
+      "epoch": 3.5928659286592866,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.856435667045577e-06,
+      "loss": 0.0592,
+      "step": 730
+    },
+    {
+      "epoch": 3.5977859778597785,
+      "grad_norm": 1.2265625,
+      "learning_rate": 1.844296715583117e-06,
+      "loss": 0.0679,
+      "step": 731
+    },
+    {
+      "epoch": 3.602706027060271,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.8321886005035812e-06,
+      "loss": 0.0883,
+      "step": 732
+    },
+    {
+      "epoch": 3.607626076260763,
+      "grad_norm": 1.25,
+      "learning_rate": 1.8201114401230586e-06,
+      "loss": 0.0875,
+      "step": 733
+    },
+    {
+      "epoch": 3.6125461254612548,
+      "grad_norm": 1.1953125,
+      "learning_rate": 1.8080653524551623e-06,
+      "loss": 0.0662,
+      "step": 734
+    },
+    {
+      "epoch": 3.6174661746617467,
+      "grad_norm": 1.859375,
+      "learning_rate": 1.796050455209869e-06,
+      "loss": 0.1162,
+      "step": 735
+    },
+    {
+      "epoch": 3.6223862238622386,
+      "grad_norm": 1.21875,
+      "learning_rate": 1.7840668657923838e-06,
+      "loss": 0.0753,
+      "step": 736
+    },
+    {
+      "epoch": 3.6273062730627306,
+      "grad_norm": 1.1953125,
+      "learning_rate": 1.772114701301972e-06,
+      "loss": 0.0767,
+      "step": 737
+    },
+    {
+      "epoch": 3.6322263222632225,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.7601940785308337e-06,
+      "loss": 0.1026,
+      "step": 738
+    },
+    {
+      "epoch": 3.6371463714637144,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.7483051139629482e-06,
+      "loss": 0.0724,
+      "step": 739
+    },
+    {
+      "epoch": 3.6420664206642064,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.7364479237729526e-06,
+      "loss": 0.0915,
+      "step": 740
+    },
+    {
+      "epoch": 3.6469864698646988,
+      "grad_norm": 1.375,
+      "learning_rate": 1.7246226238249853e-06,
+      "loss": 0.057,
+      "step": 741
+    },
+    {
+      "epoch": 3.6519065190651907,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.7128293296715704e-06,
+      "loss": 0.0772,
+      "step": 742
+    },
+    {
+      "epoch": 3.6568265682656826,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.7010681565524834e-06,
+      "loss": 0.0597,
+      "step": 743
+    },
+    {
+      "epoch": 3.6617466174661746,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.6893392193936231e-06,
+      "loss": 0.1057,
+      "step": 744
+    },
+    {
+      "epoch": 3.6666666666666665,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.677642632805892e-06,
+      "loss": 0.0869,
+      "step": 745
+    },
+    {
+      "epoch": 3.671586715867159,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.6659785110840704e-06,
+      "loss": 0.0846,
+      "step": 746
+    },
+    {
+      "epoch": 3.676506765067651,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.6543469682057105e-06,
+      "loss": 0.0953,
+      "step": 747
+    },
+    {
+      "epoch": 3.6814268142681428,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.6427481178300064e-06,
+      "loss": 0.0984,
+      "step": 748
+    },
+    {
+      "epoch": 3.6863468634686347,
+      "grad_norm": 1.2265625,
+      "learning_rate": 1.631182073296706e-06,
+      "loss": 0.099,
+      "step": 749
+    },
+    {
+      "epoch": 3.6912669126691267,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.6196489476249777e-06,
+      "loss": 0.1055,
+      "step": 750
+    },
+    {
+      "epoch": 3.6961869618696186,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.6081488535123274e-06,
+      "loss": 0.1011,
+      "step": 751
+    },
+    {
+      "epoch": 3.7011070110701105,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.5966819033334807e-06,
+      "loss": 0.0697,
+      "step": 752
+    },
+    {
+      "epoch": 3.7060270602706025,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.5852482091393045e-06,
+      "loss": 0.0792,
+      "step": 753
+    },
+    {
+      "epoch": 3.710947109471095,
+      "grad_norm": 0.8671875,
+      "learning_rate": 1.5738478826556885e-06,
+      "loss": 0.0589,
+      "step": 754
+    },
+    {
+      "epoch": 3.715867158671587,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.5624810352824709e-06,
+      "loss": 0.1095,
+      "step": 755
+    },
+    {
+      "epoch": 3.7207872078720787,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.551147778092344e-06,
+      "loss": 0.0697,
+      "step": 756
+    },
+    {
+      "epoch": 3.7257072570725707,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.539848221829769e-06,
+      "loss": 0.0562,
+      "step": 757
+    },
+    {
+      "epoch": 3.7306273062730626,
+      "grad_norm": 1.125,
+      "learning_rate": 1.5285824769098938e-06,
+      "loss": 0.064,
+      "step": 758
+    },
+    {
+      "epoch": 3.735547355473555,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.517350653417472e-06,
+      "loss": 0.0805,
+      "step": 759
+    },
+    {
+      "epoch": 3.740467404674047,
+      "grad_norm": 1.34375,
+      "learning_rate": 1.5061528611057917e-06,
+      "loss": 0.0751,
+      "step": 760
+    },
+    {
+      "epoch": 3.745387453874539,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.4949892093956015e-06,
+      "loss": 0.0958,
+      "step": 761
+    },
+    {
+      "epoch": 3.750307503075031,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.4838598073740395e-06,
+      "loss": 0.1045,
+      "step": 762
+    },
+    {
+      "epoch": 3.7552275522755227,
+      "grad_norm": 0.984375,
+      "learning_rate": 1.472764763793565e-06,
+      "loss": 0.0532,
+      "step": 763
+    },
+    {
+      "epoch": 3.7601476014760147,
+      "grad_norm": 1.3046875,
+      "learning_rate": 1.4617041870709042e-06,
+      "loss": 0.0961,
+      "step": 764
+    },
+    {
+      "epoch": 3.7650676506765066,
+      "grad_norm": 1.2265625,
+      "learning_rate": 1.4506781852859836e-06,
+      "loss": 0.0696,
+      "step": 765
+    },
+    {
+      "epoch": 3.7699876998769986,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.4396868661808777e-06,
+      "loss": 0.076,
+      "step": 766
+    },
+    {
+      "epoch": 3.774907749077491,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.428730337158749e-06,
+      "loss": 0.0949,
+      "step": 767
+    },
+    {
+      "epoch": 3.779827798277983,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.4178087052828098e-06,
+      "loss": 0.0981,
+      "step": 768
+    },
+    {
+      "epoch": 3.784747847478475,
+      "grad_norm": 1.25,
+      "learning_rate": 1.4069220772752685e-06,
+      "loss": 0.1072,
+      "step": 769
+    },
+    {
+      "epoch": 3.7896678966789668,
+      "grad_norm": 0.9921875,
+      "learning_rate": 1.3960705595162876e-06,
+      "loss": 0.063,
+      "step": 770
+    },
+    {
+      "epoch": 3.7945879458794587,
+      "grad_norm": 0.86328125,
+      "learning_rate": 1.385254258042948e-06,
+      "loss": 0.0707,
+      "step": 771
+    },
+    {
+      "epoch": 3.799507995079951,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.3744732785482035e-06,
+      "loss": 0.0804,
+      "step": 772
+    },
+    {
+      "epoch": 3.804428044280443,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.3637277263798603e-06,
+      "loss": 0.1245,
+      "step": 773
+    },
+    {
+      "epoch": 3.809348093480935,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.35301770653954e-06,
+      "loss": 0.0788,
+      "step": 774
+    },
+    {
+      "epoch": 3.814268142681427,
+      "grad_norm": 0.85546875,
+      "learning_rate": 1.3423433236816563e-06,
+      "loss": 0.0559,
+      "step": 775
+    },
+    {
+      "epoch": 3.819188191881919,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.3317046821123868e-06,
+      "loss": 0.0958,
+      "step": 776
+    },
+    {
+      "epoch": 3.8241082410824108,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.3211018857886632e-06,
+      "loss": 0.0748,
+      "step": 777
+    },
+    {
+      "epoch": 3.8290282902829027,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.3105350383171484e-06,
+      "loss": 0.0673,
+      "step": 778
+    },
+    {
+      "epoch": 3.8339483394833946,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.3000042429532267e-06,
+      "loss": 0.0911,
+      "step": 779
+    },
+    {
+      "epoch": 3.8388683886838866,
+      "grad_norm": 0.90625,
+      "learning_rate": 1.289509602599996e-06,
+      "loss": 0.0776,
+      "step": 780
+    },
+    {
+      "epoch": 3.843788437884379,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.2790512198072558e-06,
+      "loss": 0.0996,
+      "step": 781
+    },
+    {
+      "epoch": 3.848708487084871,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.268629196770514e-06,
+      "loss": 0.0685,
+      "step": 782
+    },
+    {
+      "epoch": 3.853628536285363,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.2582436353299832e-06,
+      "loss": 0.0924,
+      "step": 783
+    },
+    {
+      "epoch": 3.8585485854858548,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.2478946369695882e-06,
+      "loss": 0.0901,
+      "step": 784
+    },
+    {
+      "epoch": 3.8634686346863467,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.2375823028159667e-06,
+      "loss": 0.0971,
+      "step": 785
+    },
+    {
+      "epoch": 3.868388683886839,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.2273067336374972e-06,
+      "loss": 0.087,
+      "step": 786
+    },
+    {
+      "epoch": 3.873308733087331,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.2170680298432934e-06,
+      "loss": 0.106,
+      "step": 787
+    },
+    {
+      "epoch": 3.878228782287823,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.2068662914822432e-06,
+      "loss": 0.0634,
+      "step": 788
+    },
+    {
+      "epoch": 3.883148831488315,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.1967016182420122e-06,
+      "loss": 0.0983,
+      "step": 789
+    },
+    {
+      "epoch": 3.888068880688807,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.186574109448091e-06,
+      "loss": 0.0899,
+      "step": 790
+    },
+    {
+      "epoch": 3.892988929889299,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.1764838640628011e-06,
+      "loss": 0.0799,
+      "step": 791
+    },
+    {
+      "epoch": 3.8979089790897907,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.1664309806843466e-06,
+      "loss": 0.0942,
+      "step": 792
+    },
+    {
+      "epoch": 3.9028290282902827,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.1564155575458414e-06,
+      "loss": 0.0895,
+      "step": 793
+    },
+    {
+      "epoch": 3.907749077490775,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.1464376925143528e-06,
+      "loss": 0.0802,
+      "step": 794
+    },
+    {
+      "epoch": 3.912669126691267,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.1364974830899438e-06,
+      "loss": 0.0689,
+      "step": 795
+    },
+    {
+      "epoch": 3.917589175891759,
+      "grad_norm": 0.96875,
+      "learning_rate": 1.1265950264047171e-06,
+      "loss": 0.0677,
+      "step": 796
+    },
+    {
+      "epoch": 3.922509225092251,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.1167304192218737e-06,
+      "loss": 0.1054,
+      "step": 797
+    },
+    {
+      "epoch": 3.927429274292743,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.1069037579347613e-06,
+      "loss": 0.1051,
+      "step": 798
+    },
+    {
+      "epoch": 3.932349323493235,
+      "grad_norm": 1.0,
+      "learning_rate": 1.0971151385659357e-06,
+      "loss": 0.0892,
+      "step": 799
+    },
+    {
+      "epoch": 3.937269372693727,
+      "grad_norm": 0.8671875,
+      "learning_rate": 1.0873646567662165e-06,
+      "loss": 0.0544,
+      "step": 800
+    },
+    {
+      "epoch": 3.942189421894219,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.077652407813764e-06,
+      "loss": 0.0906,
+      "step": 801
+    },
+    {
+      "epoch": 3.947109471094711,
+      "grad_norm": 0.8984375,
+      "learning_rate": 1.067978486613131e-06,
+      "loss": 0.056,
+      "step": 802
+    },
+    {
+      "epoch": 3.952029520295203,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.0583429876943585e-06,
+      "loss": 0.0857,
+      "step": 803
+    },
+    {
+      "epoch": 3.956949569495695,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.0487460052120262e-06,
+      "loss": 0.0857,
+      "step": 804
+    },
+    {
+      "epoch": 3.961869618696187,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.0391876329443534e-06,
+      "loss": 0.0975,
+      "step": 805
+    },
+    {
+      "epoch": 3.9667896678966788,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.0296679642922718e-06,
+      "loss": 0.0768,
+      "step": 806
+    },
+    {
+      "epoch": 3.971709717097171,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.0201870922785156e-06,
+      "loss": 0.1113,
+      "step": 807
+    },
+    {
+      "epoch": 3.976629766297663,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.010745109546713e-06,
+      "loss": 0.0812,
+      "step": 808
+    },
+    {
+      "epoch": 3.981549815498155,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.0013421083604779e-06,
+      "loss": 0.062,
+      "step": 809
+    },
+    {
+      "epoch": 3.986469864698647,
+      "grad_norm": 0.96875,
+      "learning_rate": 9.919781806025136e-07,
+      "loss": 0.0666,
+      "step": 810
+    },
+    {
+      "epoch": 3.991389913899139,
+      "grad_norm": 1.0859375,
+      "learning_rate": 9.826534177737106e-07,
+      "loss": 0.0616,
+      "step": 811
+    },
+    {
+      "epoch": 3.9963099630996313,
+      "grad_norm": 1.140625,
+      "learning_rate": 9.733679109922567e-07,
+      "loss": 0.0663,
+      "step": 812
+    },
+    {
+      "epoch": 4.001230012300123,
+      "grad_norm": 2.390625,
+      "learning_rate": 9.641217509927376e-07,
+      "loss": 0.0926,
+      "step": 813
+    },
+    {
+      "epoch": 4.006150061500615,
+      "grad_norm": 1.1875,
+      "learning_rate": 9.549150281252633e-07,
+      "loss": 0.0742,
+      "step": 814
+    },
+    {
+      "epoch": 4.011070110701107,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.457478323545749e-07,
+      "loss": 0.1039,
+      "step": 815
+    },
+    {
+      "epoch": 4.015990159901599,
+      "grad_norm": 1.0859375,
+      "learning_rate": 9.366202532591717e-07,
+      "loss": 0.0794,
+      "step": 816
+    },
+    {
+      "epoch": 4.015990159901599,
+      "eval_loss": 0.10322786867618561,
+      "eval_runtime": 204.3584,
+      "eval_samples_per_second": 8.006,
+      "eval_steps_per_second": 1.336,
+      "step": 816
+    },
+    {
+      "epoch": 4.020910209102091,
+      "grad_norm": 0.78515625,
+      "learning_rate": 9.275323800304287e-07,
+      "loss": 0.0628,
+      "step": 817
+    },
+    {
+      "epoch": 4.025830258302583,
+      "grad_norm": 0.9296875,
+      "learning_rate": 9.184843014717337e-07,
+      "loss": 0.0652,
+      "step": 818
+    },
+    {
+      "epoch": 4.030750307503075,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.094761059976154e-07,
+      "loss": 0.0884,
+      "step": 819
+    },
+    {
+      "epoch": 4.035670356703567,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.005078816328772e-07,
+      "loss": 0.0858,
+      "step": 820
+    },
+    {
+      "epoch": 4.040590405904059,
+      "grad_norm": 1.125,
+      "learning_rate": 8.915797160117423e-07,
+      "loss": 0.1029,
+      "step": 821
+    },
+    {
+      "epoch": 4.045510455104551,
+      "grad_norm": 1.140625,
+      "learning_rate": 8.826916963769888e-07,
+      "loss": 0.0981,
+      "step": 822
+    },
+    {
+      "epoch": 4.0504305043050435,
+      "grad_norm": 1.0078125,
+      "learning_rate": 8.738439095791123e-07,
+      "loss": 0.0688,
+      "step": 823
+    },
+    {
+      "epoch": 4.055350553505535,
+      "grad_norm": 0.98828125,
+      "learning_rate": 8.650364420754581e-07,
+      "loss": 0.0715,
+      "step": 824
+    },
+    {
+      "epoch": 4.060270602706027,
+      "grad_norm": 0.94140625,
+      "learning_rate": 8.562693799293931e-07,
+      "loss": 0.0624,
+      "step": 825
+    },
+    {
+      "epoch": 4.065190651906519,
+      "grad_norm": 1.328125,
+      "learning_rate": 8.475428088094517e-07,
+      "loss": 0.0977,
+      "step": 826
+    },
+    {
+      "epoch": 4.070110701107011,
+      "grad_norm": 1.140625,
+      "learning_rate": 8.388568139885101e-07,
+      "loss": 0.0827,
+      "step": 827
+    },
+    {
+      "epoch": 4.075030750307503,
+      "grad_norm": 1.0078125,
+      "learning_rate": 8.30211480342945e-07,
+      "loss": 0.0824,
+      "step": 828
+    },
+    {
+      "epoch": 4.079950799507995,
+      "grad_norm": 1.0390625,
+      "learning_rate": 8.216068923518072e-07,
+      "loss": 0.0713,
+      "step": 829
+    },
+    {
+      "epoch": 4.084870848708487,
+      "grad_norm": 1.171875,
+      "learning_rate": 8.130431340959982e-07,
+      "loss": 0.0905,
+      "step": 830
+    },
+    {
+      "epoch": 4.089790897908979,
+      "grad_norm": 1.0390625,
+      "learning_rate": 8.045202892574395e-07,
+      "loss": 0.0692,
+      "step": 831
+    },
+    {
+      "epoch": 4.094710947109471,
+      "grad_norm": 0.9765625,
+      "learning_rate": 7.960384411182709e-07,
+      "loss": 0.0713,
+      "step": 832
+    },
+    {
+      "epoch": 4.099630996309963,
+      "grad_norm": 1.03125,
+      "learning_rate": 7.875976725600193e-07,
+      "loss": 0.0824,
+      "step": 833
+    },
+    {
+      "epoch": 4.104551045510455,
+      "grad_norm": 0.76953125,
+      "learning_rate": 7.791980660628029e-07,
+      "loss": 0.0579,
+      "step": 834
+    },
+    {
+      "epoch": 4.109471094710947,
+      "grad_norm": 0.98828125,
+      "learning_rate": 7.708397037045129e-07,
+      "loss": 0.081,
+      "step": 835
+    },
+    {
+      "epoch": 4.114391143911439,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.625226671600256e-07,
+      "loss": 0.0884,
+      "step": 836
+    },
+    {
+      "epoch": 4.1193111931119315,
+      "grad_norm": 1.0546875,
+      "learning_rate": 7.542470377003897e-07,
+      "loss": 0.0809,
+      "step": 837
+    },
+    {
+      "epoch": 4.124231242312423,
+      "grad_norm": 1.0546875,
+      "learning_rate": 7.460128961920432e-07,
+      "loss": 0.0768,
+      "step": 838
+    },
+    {
+      "epoch": 4.129151291512915,
+      "grad_norm": 0.8203125,
+      "learning_rate": 7.37820323096014e-07,
+      "loss": 0.0809,
+      "step": 839
+    },
+    {
+      "epoch": 4.134071340713407,
+      "grad_norm": 0.8359375,
+      "learning_rate": 7.296693984671465e-07,
+      "loss": 0.0748,
+      "step": 840
+    },
+    {
+      "epoch": 4.138991389913899,
+      "grad_norm": 1.046875,
+      "learning_rate": 7.215602019533041e-07,
+      "loss": 0.1046,
+      "step": 841
+    },
+    {
+      "epoch": 4.143911439114391,
+      "grad_norm": 1.1015625,
+      "learning_rate": 7.134928127946017e-07,
+      "loss": 0.0804,
+      "step": 842
+    },
+    {
+      "epoch": 4.148831488314883,
+      "grad_norm": 0.9375,
+      "learning_rate": 7.054673098226278e-07,
+      "loss": 0.0735,
+      "step": 843
+    },
+    {
+      "epoch": 4.153751537515375,
+      "grad_norm": 1.1328125,
+      "learning_rate": 6.974837714596732e-07,
+      "loss": 0.0918,
+      "step": 844
+    },
+    {
+      "epoch": 4.158671586715867,
+      "grad_norm": 1.1328125,
+      "learning_rate": 6.895422757179682e-07,
+      "loss": 0.0926,
+      "step": 845
+    },
+    {
+      "epoch": 4.163591635916359,
+      "grad_norm": 1.03125,
+      "learning_rate": 6.816429001989133e-07,
+      "loss": 0.0887,
+      "step": 846
+    },
+    {
+      "epoch": 4.168511685116851,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.737857220923305e-07,
+      "loss": 0.0546,
+      "step": 847
+    },
+    {
+      "epoch": 4.173431734317343,
+      "grad_norm": 1.015625,
+      "learning_rate": 6.659708181757013e-07,
+      "loss": 0.0709,
+      "step": 848
+    },
+    {
+      "epoch": 4.178351783517835,
+      "grad_norm": 0.84375,
+      "learning_rate": 6.581982648134217e-07,
+      "loss": 0.0709,
+      "step": 849
+    },
+    {
+      "epoch": 4.183271832718328,
+      "grad_norm": 1.015625,
+      "learning_rate": 6.50468137956049e-07,
+      "loss": 0.0639,
+      "step": 850
+    },
+    {
+      "epoch": 4.1881918819188195,
+      "grad_norm": 0.90234375,
+      "learning_rate": 6.427805131395681e-07,
+      "loss": 0.0931,
+      "step": 851
+    },
+    {
+      "epoch": 4.1931119311193115,
+      "grad_norm": 0.87890625,
+      "learning_rate": 6.351354654846481e-07,
+      "loss": 0.0811,
+      "step": 852
+    },
+    {
+      "epoch": 4.198031980319803,
+      "grad_norm": 1.171875,
+      "learning_rate": 6.275330696959109e-07,
+      "loss": 0.0872,
+      "step": 853
+    },
+    {
+      "epoch": 4.202952029520295,
+      "grad_norm": 0.9921875,
+      "learning_rate": 6.199734000611968e-07,
+      "loss": 0.0826,
+      "step": 854
+    },
+    {
+      "epoch": 4.207872078720787,
+      "grad_norm": 0.81640625,
+      "learning_rate": 6.12456530450844e-07,
+      "loss": 0.0732,
+      "step": 855
+    },
+    {
+      "epoch": 4.212792127921279,
+      "grad_norm": 0.91796875,
+      "learning_rate": 6.049825343169652e-07,
+      "loss": 0.0697,
+      "step": 856
+    },
+    {
+      "epoch": 4.217712177121771,
+      "grad_norm": 0.99609375,
+      "learning_rate": 5.975514846927271e-07,
+      "loss": 0.0767,
+      "step": 857
+    },
+    {
+      "epoch": 4.222632226322263,
+      "grad_norm": 0.921875,
+      "learning_rate": 5.901634541916406e-07,
+      "loss": 0.0632,
+      "step": 858
+    },
+    {
+      "epoch": 4.227552275522755,
+      "grad_norm": 0.984375,
+      "learning_rate": 5.828185150068472e-07,
+      "loss": 0.0798,
+      "step": 859
+    },
+    {
+      "epoch": 4.232472324723247,
+      "grad_norm": 0.796875,
+      "learning_rate": 5.755167389104166e-07,
+      "loss": 0.0685,
+      "step": 860
+    },
+    {
+      "epoch": 4.237392373923739,
+      "grad_norm": 0.87109375,
+      "learning_rate": 5.682581972526463e-07,
+      "loss": 0.0645,
+      "step": 861
+    },
+    {
+      "epoch": 4.242312423124231,
+      "grad_norm": 0.84375,
+      "learning_rate": 5.610429609613615e-07,
+      "loss": 0.0588,
+      "step": 862
+    },
+    {
+      "epoch": 4.247232472324724,
+      "grad_norm": 0.7890625,
+      "learning_rate": 5.538711005412212e-07,
+      "loss": 0.0581,
+      "step": 863
+    },
+    {
+      "epoch": 4.252152521525216,
+      "grad_norm": 0.984375,
+      "learning_rate": 5.467426860730334e-07,
+      "loss": 0.082,
+      "step": 864
+    },
+    {
+      "epoch": 4.2570725707257075,
+      "grad_norm": 0.86328125,
+      "learning_rate": 5.396577872130676e-07,
+      "loss": 0.0584,
+      "step": 865
+    },
+    {
+      "epoch": 4.2619926199261995,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.32616473192375e-07,
+      "loss": 0.0822,
+      "step": 866
+    },
+    {
+      "epoch": 4.266912669126691,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5.256188128161116e-07,
+      "loss": 0.0834,
+      "step": 867
+    },
+    {
+      "epoch": 4.271832718327183,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.186648744628637e-07,
+      "loss": 0.0703,
+      "step": 868
+    },
+    {
+      "epoch": 4.276752767527675,
+      "grad_norm": 1.2734375,
+      "learning_rate": 5.117547260839845e-07,
+      "loss": 0.1001,
+      "step": 869
+    },
+    {
+      "epoch": 4.281672816728167,
+      "grad_norm": 1.3046875,
+      "learning_rate": 5.048884352029271e-07,
+      "loss": 0.1088,
+      "step": 870
+    },
+    {
+      "epoch": 4.286592865928659,
+      "grad_norm": 0.80859375,
+      "learning_rate": 4.980660689145855e-07,
+      "loss": 0.0635,
+      "step": 871
+    },
+    {
+      "epoch": 4.291512915129151,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.912876938846345e-07,
+      "loss": 0.0876,
+      "step": 872
+    },
+    {
+      "epoch": 4.296432964329643,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.845533763488902e-07,
+      "loss": 0.0893,
+      "step": 873
+    },
+    {
+      "epoch": 4.301353013530135,
+      "grad_norm": 1.2421875,
+      "learning_rate": 4.778631821126473e-07,
+      "loss": 0.0946,
+      "step": 874
+    },
+    {
+      "epoch": 4.306273062730627,
+      "grad_norm": 1.1875,
+      "learning_rate": 4.712171765500484e-07,
+      "loss": 0.0911,
+      "step": 875
+    },
+    {
+      "epoch": 4.31119311193112,
+      "grad_norm": 1.1171875,
+      "learning_rate": 4.6461542460343565e-07,
+      "loss": 0.1092,
+      "step": 876
+    },
+    {
+      "epoch": 4.316113161131612,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.580579907827287e-07,
+      "loss": 0.0921,
+      "step": 877
+    },
+    {
+      "epoch": 4.321033210332104,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.515449391647786e-07,
+      "loss": 0.0759,
+      "step": 878
+    },
+    {
+      "epoch": 4.325953259532596,
+      "grad_norm": 1.109375,
+      "learning_rate": 4.4507633339275494e-07,
+      "loss": 0.0808,
+      "step": 879
+    },
+    {
+      "epoch": 4.3308733087330875,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.386522366755169e-07,
+      "loss": 0.0842,
+      "step": 880
+    },
+    {
+      "epoch": 4.3357933579335795,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.322727117869951e-07,
+      "loss": 0.1048,
+      "step": 881
+    },
+    {
+      "epoch": 4.340713407134071,
+      "grad_norm": 1.15625,
+      "learning_rate": 4.2593782106558676e-07,
+      "loss": 0.0922,
+      "step": 882
+    },
+    {
+      "epoch": 4.345633456334563,
+      "grad_norm": 0.9765625,
+      "learning_rate": 4.1964762641353297e-07,
+      "loss": 0.0989,
+      "step": 883
+    },
+    {
+      "epoch": 4.350553505535055,
+      "grad_norm": 1.171875,
+      "learning_rate": 4.1340218929632636e-07,
+      "loss": 0.0922,
+      "step": 884
+    },
+    {
+      "epoch": 4.355473554735547,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.072015707421006e-07,
+      "loss": 0.0831,
+      "step": 885
+    },
+    {
+      "epoch": 4.360393603936039,
+      "grad_norm": 1.2109375,
+      "learning_rate": 4.0104583134104593e-07,
+      "loss": 0.092,
+      "step": 886
+    },
+    {
+      "epoch": 4.365313653136531,
+      "grad_norm": 0.9921875,
+      "learning_rate": 3.9493503124480135e-07,
+      "loss": 0.0722,
+      "step": 887
+    },
+    {
+      "epoch": 4.370233702337023,
+      "grad_norm": 1.1796875,
+      "learning_rate": 3.8886923016588195e-07,
+      "loss": 0.0792,
+      "step": 888
+    },
+    {
+      "epoch": 4.375153751537516,
+      "grad_norm": 1.21875,
+      "learning_rate": 3.828484873770832e-07,
+      "loss": 0.0541,
+      "step": 889
+    },
+    {
+      "epoch": 4.380073800738008,
+      "grad_norm": 1.140625,
+      "learning_rate": 3.7687286171091355e-07,
+      "loss": 0.1026,
+      "step": 890
+    },
+    {
+      "epoch": 4.3849938499385,
+      "grad_norm": 1.0,
+      "learning_rate": 3.709424115590088e-07,
+      "loss": 0.0849,
+      "step": 891
+    },
+    {
+      "epoch": 4.389913899138992,
+      "grad_norm": 1.109375,
+      "learning_rate": 3.65057194871567e-07,
+      "loss": 0.098,
+      "step": 892
+    },
+    {
+      "epoch": 4.394833948339484,
+      "grad_norm": 0.8359375,
+      "learning_rate": 3.5921726915678247e-07,
+      "loss": 0.0659,
+      "step": 893
+    },
+    {
+      "epoch": 4.3997539975399755,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.534226914802813e-07,
+      "loss": 0.0839,
+      "step": 894
+    },
+    {
+      "epoch": 4.4046740467404675,
+      "grad_norm": 0.94921875,
+      "learning_rate": 3.4767351846456744e-07,
+      "loss": 0.1014,
+      "step": 895
+    },
+    {
+      "epoch": 4.409594095940959,
+      "grad_norm": 1.296875,
+      "learning_rate": 3.4196980628846297e-07,
+      "loss": 0.0843,
+      "step": 896
+    },
+    {
+      "epoch": 4.414514145141451,
+      "grad_norm": 0.9609375,
+      "learning_rate": 3.3631161068656604e-07,
+      "loss": 0.0727,
+      "step": 897
+    },
+    {
+      "epoch": 4.419434194341943,
+      "grad_norm": 1.03125,
+      "learning_rate": 3.3069898694870373e-07,
+      "loss": 0.0771,
+      "step": 898
+    },
+    {
+      "epoch": 4.424354243542435,
+      "grad_norm": 1.21875,
+      "learning_rate": 3.2513198991939054e-07,
+      "loss": 0.0975,
+      "step": 899
+    },
+    {
+      "epoch": 4.429274292742927,
+      "grad_norm": 0.8984375,
+      "learning_rate": 3.196106739972926e-07,
+      "loss": 0.0649,
+      "step": 900
+    },
+    {
+      "epoch": 4.434194341943419,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.1413509313469816e-07,
+      "loss": 0.0885,
+      "step": 901
+    },
+    {
+      "epoch": 4.439114391143911,
+      "grad_norm": 0.88671875,
+      "learning_rate": 3.087053008369889e-07,
+      "loss": 0.069,
+      "step": 902
+    },
+    {
+      "epoch": 4.444034440344403,
+      "grad_norm": 0.859375,
+      "learning_rate": 3.0332135016211794e-07,
+      "loss": 0.0539,
+      "step": 903
+    },
+    {
+      "epoch": 4.448954489544896,
+      "grad_norm": 0.97265625,
+      "learning_rate": 2.979832937200883e-07,
+      "loss": 0.0869,
+      "step": 904
+    },
+    {
+      "epoch": 4.453874538745388,
+      "grad_norm": 1.0,
+      "learning_rate": 2.9269118367244385e-07,
+      "loss": 0.0936,
+      "step": 905
+    },
+    {
+      "epoch": 4.45879458794588,
+      "grad_norm": 1.109375,
+      "learning_rate": 2.8744507173175564e-07,
+      "loss": 0.07,
+      "step": 906
+    },
+    {
+      "epoch": 4.463714637146372,
+      "grad_norm": 1.15625,
+      "learning_rate": 2.822450091611195e-07,
+      "loss": 0.0865,
+      "step": 907
+    },
+    {
+      "epoch": 4.468634686346864,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.770910467736532e-07,
+      "loss": 0.0797,
+      "step": 908
+    },
+    {
+      "epoch": 4.4735547355473555,
+      "grad_norm": 0.87890625,
+      "learning_rate": 2.719832349319973e-07,
+      "loss": 0.0591,
+      "step": 909
+    },
+    {
+      "epoch": 4.478474784747847,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.669216235478295e-07,
+      "loss": 0.0716,
+      "step": 910
+    },
+    {
+      "epoch": 4.483394833948339,
+      "grad_norm": 1.390625,
+      "learning_rate": 2.619062620813728e-07,
+      "loss": 0.0919,
+      "step": 911
+    },
+    {
+      "epoch": 4.488314883148831,
+      "grad_norm": 1.171875,
+      "learning_rate": 2.5693719954091257e-07,
+      "loss": 0.073,
+      "step": 912
+    },
+    {
+      "epoch": 4.493234932349323,
+      "grad_norm": 1.0,
+      "learning_rate": 2.520144844823169e-07,
+      "loss": 0.0683,
+      "step": 913
+    },
+    {
+      "epoch": 4.498154981549815,
+      "grad_norm": 1.1328125,
+      "learning_rate": 2.471381650085647e-07,
+      "loss": 0.0885,
+      "step": 914
+    },
+    {
+      "epoch": 4.503075030750307,
+      "grad_norm": 1.1640625,
+      "learning_rate": 2.4230828876927293e-07,
+      "loss": 0.0676,
+      "step": 915
+    },
+    {
+      "epoch": 4.507995079950799,
+      "grad_norm": 1.078125,
+      "learning_rate": 2.375249029602339e-07,
+      "loss": 0.0829,
+      "step": 916
+    },
+    {
+      "epoch": 4.512915129151292,
+      "grad_norm": 1.171875,
+      "learning_rate": 2.327880543229505e-07,
+      "loss": 0.0787,
+      "step": 917
+    },
+    {
+      "epoch": 4.517835178351784,
+      "grad_norm": 1.1171875,
+      "learning_rate": 2.2809778914418156e-07,
+      "loss": 0.0839,
+      "step": 918
+    },
+    {
+      "epoch": 4.517835178351784,
+      "eval_loss": 0.10323299467563629,
+      "eval_runtime": 203.6707,
+      "eval_samples_per_second": 8.033,
+      "eval_steps_per_second": 1.34,
+      "step": 918
+    },
+    {
+      "epoch": 4.522755227552276,
+      "grad_norm": 1.1875,
+      "learning_rate": 2.2345415325549125e-07,
+      "loss": 0.1202,
+      "step": 919
+    },
+    {
+      "epoch": 4.527675276752768,
+      "grad_norm": 0.90234375,
+      "learning_rate": 2.1885719203279587e-07,
+      "loss": 0.0706,
+      "step": 920
+    },
+    {
+      "epoch": 4.53259532595326,
+      "grad_norm": 0.953125,
+      "learning_rate": 2.143069503959283e-07,
+      "loss": 0.0835,
+      "step": 921
+    },
+    {
+      "epoch": 4.537515375153752,
+      "grad_norm": 1.140625,
+      "learning_rate": 2.0980347280818935e-07,
+      "loss": 0.0678,
+      "step": 922
+    },
+    {
+      "epoch": 4.5424354243542435,
+      "grad_norm": 0.98046875,
+      "learning_rate": 2.0534680327592426e-07,
+      "loss": 0.0662,
+      "step": 923
+    },
+    {
+      "epoch": 4.5473554735547355,
+      "grad_norm": 0.93359375,
+      "learning_rate": 2.009369853480825e-07,
+      "loss": 0.0582,
+      "step": 924
+    },
+    {
+      "epoch": 4.552275522755227,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.9657406211579966e-07,
+      "loss": 0.1065,
+      "step": 925
+    },
+    {
+      "epoch": 4.557195571955719,
+      "grad_norm": 0.97265625,
+      "learning_rate": 1.922580762119697e-07,
+      "loss": 0.0513,
+      "step": 926
+    },
+    {
+      "epoch": 4.562115621156211,
+      "grad_norm": 1.1796875,
+      "learning_rate": 1.8798906981083832e-07,
+      "loss": 0.0734,
+      "step": 927
+    },
+    {
+      "epoch": 4.567035670356703,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.8376708462757798e-07,
+      "loss": 0.0901,
+      "step": 928
+    },
+    {
+      "epoch": 4.571955719557195,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.7959216191789142e-07,
+      "loss": 0.0806,
+      "step": 929
+    },
+    {
+      "epoch": 4.576875768757688,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.7546434247760147e-07,
+      "loss": 0.0858,
+      "step": 930
+    },
+    {
+      "epoch": 4.58179581795818,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.713836666422569e-07,
+      "loss": 0.0808,
+      "step": 931
+    },
+    {
+      "epoch": 4.586715867158672,
+      "grad_norm": 0.984375,
+      "learning_rate": 1.673501742867356e-07,
+      "loss": 0.0951,
+      "step": 932
+    },
+    {
+      "epoch": 4.591635916359164,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.633639048248542e-07,
+      "loss": 0.0835,
+      "step": 933
+    },
+    {
+      "epoch": 4.596555965559656,
+      "grad_norm": 0.9140625,
+      "learning_rate": 1.594248972089879e-07,
+      "loss": 0.0686,
+      "step": 934
+    },
+    {
+      "epoch": 4.601476014760148,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.555331899296808e-07,
+      "loss": 0.1139,
+      "step": 935
+    },
+    {
+      "epoch": 4.60639606396064,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.5168882101528282e-07,
+      "loss": 0.0884,
+      "step": 936
+    },
+    {
+      "epoch": 4.6113161131611315,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.4789182803156333e-07,
+      "loss": 0.0855,
+      "step": 937
+    },
+    {
+      "epoch": 4.6162361623616235,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.44142248081357e-07,
+      "loss": 0.0854,
+      "step": 938
+    },
+    {
+      "epoch": 4.621156211562115,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.4044011780419032e-07,
+      "loss": 0.0947,
+      "step": 939
+    },
+    {
+      "epoch": 4.626076260762607,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.3678547337593494e-07,
+      "loss": 0.0637,
+      "step": 940
+    },
+    {
+      "epoch": 4.630996309963099,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.3317835050844275e-07,
+      "loss": 0.0743,
+      "step": 941
+    },
+    {
+      "epoch": 4.635916359163591,
+      "grad_norm": 1.2265625,
+      "learning_rate": 1.296187844492053e-07,
+      "loss": 0.0928,
+      "step": 942
+    },
+    {
+      "epoch": 4.640836408364084,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2610680998100476e-07,
+      "loss": 0.0978,
+      "step": 943
+    },
+    {
+      "epoch": 4.645756457564576,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.2264246142157656e-07,
+      "loss": 0.0776,
+      "step": 944
+    },
+    {
+      "epoch": 4.650676506765068,
+      "grad_norm": 0.91796875,
+      "learning_rate": 1.1922577262327374e-07,
+      "loss": 0.0896,
+      "step": 945
+    },
+    {
+      "epoch": 4.65559655596556,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.1585677697273312e-07,
+      "loss": 0.0723,
+      "step": 946
+    },
+    {
+      "epoch": 4.660516605166052,
+      "grad_norm": 1.1796875,
+      "learning_rate": 1.1253550739055374e-07,
+      "loss": 0.0738,
+      "step": 947
+    },
+    {
+      "epoch": 4.665436654366544,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.0926199633097156e-07,
+      "loss": 0.0792,
+      "step": 948
+    },
+    {
+      "epoch": 4.670356703567036,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.060362757815453e-07,
+      "loss": 0.0992,
+      "step": 949
+    },
+    {
+      "epoch": 4.675276752767528,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.0285837726283999e-07,
+      "loss": 0.0638,
+      "step": 950
+    },
+    {
+      "epoch": 4.68019680196802,
+      "grad_norm": 1.0,
+      "learning_rate": 9.972833182812225e-08,
+      "loss": 0.0741,
+      "step": 951
+    },
+    {
+      "epoch": 4.6851168511685115,
+      "grad_norm": 1.4140625,
+      "learning_rate": 9.664617006305665e-08,
+      "loss": 0.1135,
+      "step": 952
+    },
+    {
+      "epoch": 4.6900369003690034,
+      "grad_norm": 0.90625,
+      "learning_rate": 9.361192208540427e-08,
+      "loss": 0.0603,
+      "step": 953
+    },
+    {
+      "epoch": 4.694956949569495,
+      "grad_norm": 1.0,
+      "learning_rate": 9.062561754473231e-08,
+      "loss": 0.0811,
+      "step": 954
+    },
+    {
+      "epoch": 4.699876998769987,
+      "grad_norm": 1.8984375,
+      "learning_rate": 8.768728562211948e-08,
+      "loss": 0.1155,
+      "step": 955
+    },
+    {
+      "epoch": 4.70479704797048,
+      "grad_norm": 0.9140625,
+      "learning_rate": 8.479695502987551e-08,
+      "loss": 0.0756,
+      "step": 956
+    },
+    {
+      "epoch": 4.709717097170972,
+      "grad_norm": 1.265625,
+      "learning_rate": 8.195465401125812e-08,
+      "loss": 0.1058,
+      "step": 957
+    },
+    {
+      "epoch": 4.714637146371464,
+      "grad_norm": 0.8359375,
+      "learning_rate": 7.916041034019773e-08,
+      "loss": 0.0538,
+      "step": 958
+    },
+    {
+      "epoch": 4.719557195571956,
+      "grad_norm": 1.0859375,
+      "learning_rate": 7.64142513210242e-08,
+      "loss": 0.0974,
+      "step": 959
+    },
+    {
+      "epoch": 4.724477244772448,
+      "grad_norm": 1.1484375,
+      "learning_rate": 7.371620378820555e-08,
+      "loss": 0.0975,
+      "step": 960
+    },
+    {
+      "epoch": 4.72939729397294,
+      "grad_norm": 0.953125,
+      "learning_rate": 7.10662941060769e-08,
+      "loss": 0.0675,
+      "step": 961
+    },
+    {
+      "epoch": 4.734317343173432,
+      "grad_norm": 1.1796875,
+      "learning_rate": 6.84645481685925e-08,
+      "loss": 0.081,
+      "step": 962
+    },
+    {
+      "epoch": 4.739237392373924,
+      "grad_norm": 0.9453125,
+      "learning_rate": 6.59109913990641e-08,
+      "loss": 0.0836,
+      "step": 963
+    },
+    {
+      "epoch": 4.744157441574416,
+      "grad_norm": 1.1640625,
+      "learning_rate": 6.340564874991906e-08,
+      "loss": 0.0772,
+      "step": 964
+    },
+    {
+      "epoch": 4.749077490774908,
+      "grad_norm": 1.3515625,
+      "learning_rate": 6.094854470245326e-08,
+      "loss": 0.1065,
+      "step": 965
+    },
+    {
+      "epoch": 4.7539975399753995,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.853970326659186e-08,
+      "loss": 0.0749,
+      "step": 966
+    },
+    {
+      "epoch": 4.7589175891758915,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5.6179147980656154e-08,
+      "loss": 0.0792,
+      "step": 967
+    },
+    {
+      "epoch": 4.763837638376383,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.3866901911132086e-08,
+      "loss": 0.0924,
+      "step": 968
+    },
+    {
+      "epoch": 4.768757687576876,
+      "grad_norm": 0.9453125,
+      "learning_rate": 5.160298765244709e-08,
+      "loss": 0.0824,
+      "step": 969
+    },
+    {
+      "epoch": 4.773677736777367,
+      "grad_norm": 0.96484375,
+      "learning_rate": 4.9387427326745287e-08,
+      "loss": 0.0622,
+      "step": 970
+    },
+    {
+      "epoch": 4.77859778597786,
+      "grad_norm": 0.85546875,
+      "learning_rate": 4.722024258367597e-08,
+      "loss": 0.0668,
+      "step": 971
+    },
+    {
+      "epoch": 4.783517835178352,
+      "grad_norm": 1.359375,
+      "learning_rate": 4.5101454600177676e-08,
+      "loss": 0.1176,
+      "step": 972
+    },
+    {
+      "epoch": 4.788437884378844,
+      "grad_norm": 1.2109375,
+      "learning_rate": 4.303108408027668e-08,
+      "loss": 0.1058,
+      "step": 973
+    },
+    {
+      "epoch": 4.793357933579336,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.1009151254878254e-08,
+      "loss": 0.091,
+      "step": 974
+    },
+    {
+      "epoch": 4.798277982779828,
+      "grad_norm": 0.88671875,
+      "learning_rate": 3.903567588157353e-08,
+      "loss": 0.0783,
+      "step": 975
+    },
+    {
+      "epoch": 4.80319803198032,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.711067724444517e-08,
+      "loss": 0.0883,
+      "step": 976
+    },
+    {
+      "epoch": 4.808118081180812,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.523417415387864e-08,
+      "loss": 0.0801,
+      "step": 977
+    },
+    {
+      "epoch": 4.813038130381304,
+      "grad_norm": 1.0859375,
+      "learning_rate": 3.340618494637793e-08,
+      "loss": 0.0695,
+      "step": 978
+    },
+    {
+      "epoch": 4.817958179581796,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.162672748438844e-08,
+      "loss": 0.0959,
+      "step": 979
+    },
+    {
+      "epoch": 4.822878228782288,
+      "grad_norm": 1.21875,
+      "learning_rate": 2.989581915611994e-08,
+      "loss": 0.0944,
+      "step": 980
+    },
+    {
+      "epoch": 4.8277982779827795,
+      "grad_norm": 0.890625,
+      "learning_rate": 2.821347687537834e-08,
+      "loss": 0.0634,
+      "step": 981
+    },
+    {
+      "epoch": 4.832718327183272,
+      "grad_norm": 0.921875,
+      "learning_rate": 2.657971708139917e-08,
+      "loss": 0.0954,
+      "step": 982
+    },
+    {
+      "epoch": 4.837638376383763,
+      "grad_norm": 1.2265625,
+      "learning_rate": 2.4994555738688252e-08,
+      "loss": 0.0935,
+      "step": 983
+    },
+    {
+      "epoch": 4.842558425584256,
+      "grad_norm": 1.0859375,
+      "learning_rate": 2.3458008336864623e-08,
+      "loss": 0.0965,
+      "step": 984
+    },
+    {
+      "epoch": 4.847478474784748,
+      "grad_norm": 1.078125,
+      "learning_rate": 2.1970089890509527e-08,
+      "loss": 0.0944,
+      "step": 985
+    },
+    {
+      "epoch": 4.85239852398524,
+      "grad_norm": 1.21875,
+      "learning_rate": 2.0530814939020428e-08,
+      "loss": 0.0842,
+      "step": 986
+    },
+    {
+      "epoch": 4.857318573185732,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.9140197546467787e-08,
+      "loss": 0.0857,
+      "step": 987
+    },
+    {
+      "epoch": 4.862238622386224,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.7798251301458512e-08,
+      "loss": 0.0715,
+      "step": 988
+    },
+    {
+      "epoch": 4.867158671586716,
+      "grad_norm": 0.953125,
+      "learning_rate": 1.6504989317001618e-08,
+      "loss": 0.0868,
+      "step": 989
+    },
+    {
+      "epoch": 4.872078720787208,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.5260424230382763e-08,
+      "loss": 0.0842,
+      "step": 990
+    },
+    {
+      "epoch": 4.8769987699877,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.4064568203037699e-08,
+      "loss": 0.0964,
+      "step": 991
+    },
+    {
+      "epoch": 4.881918819188192,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.2917432920437345e-08,
+      "loss": 0.089,
+      "step": 992
+    },
+    {
+      "epoch": 4.886838868388684,
+      "grad_norm": 0.875,
+      "learning_rate": 1.1819029591968456e-08,
+      "loss": 0.0584,
+      "step": 993
+    },
+    {
+      "epoch": 4.891758917589176,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.0769368950829252e-08,
+      "loss": 0.0723,
+      "step": 994
+    },
+    {
+      "epoch": 4.8966789667896675,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.768461253920614e-09,
+      "loss": 0.1013,
+      "step": 995
+    },
+    {
+      "epoch": 4.9015990159901595,
+      "grad_norm": 0.9921875,
+      "learning_rate": 8.816316281747839e-09,
+      "loss": 0.0699,
+      "step": 996
+    },
+    {
+      "epoch": 4.906519065190652,
+      "grad_norm": 0.921875,
+      "learning_rate": 7.912943338324598e-09,
+      "loss": 0.0806,
+      "step": 997
+    },
+    {
+      "epoch": 4.911439114391144,
+      "grad_norm": 0.8359375,
+      "learning_rate": 7.058351251083007e-09,
+      "loss": 0.0705,
+      "step": 998
+    },
+    {
+      "epoch": 4.916359163591636,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.252548370784817e-09,
+      "loss": 0.072,
+      "step": 999
+    },
+    {
+      "epoch": 4.921279212792128,
+      "grad_norm": 0.921875,
+      "learning_rate": 5.495542571443135e-09,
+      "loss": 0.0715,
+      "step": 1000
+    },
+    {
+      "epoch": 4.92619926199262,
+      "grad_norm": 0.9453125,
+      "learning_rate": 4.787341250241384e-09,
+      "loss": 0.0838,
+      "step": 1001
+    },
+    {
+      "epoch": 4.931119311193112,
+      "grad_norm": 0.890625,
+      "learning_rate": 4.127951327466684e-09,
+      "loss": 0.0672,
+      "step": 1002
+    },
+    {
+      "epoch": 4.936039360393604,
+      "grad_norm": 1.0703125,
+      "learning_rate": 3.5173792464360256e-09,
+      "loss": 0.0753,
+      "step": 1003
+    },
+    {
+      "epoch": 4.940959409594096,
+      "grad_norm": 0.94140625,
+      "learning_rate": 2.955630973437429e-09,
+      "loss": 0.0858,
+      "step": 1004
+    },
+    {
+      "epoch": 4.945879458794588,
+      "grad_norm": 0.83984375,
+      "learning_rate": 2.442711997670544e-09,
+      "loss": 0.0674,
+      "step": 1005
+    },
+    {
+      "epoch": 4.95079950799508,
+      "grad_norm": 1.0,
+      "learning_rate": 1.978627331192806e-09,
+      "loss": 0.0892,
+      "step": 1006
+    },
+    {
+      "epoch": 4.955719557195572,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.5633815088705872e-09,
+      "loss": 0.074,
+      "step": 1007
+    },
+    {
+      "epoch": 4.960639606396064,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.196978588334785e-09,
+      "loss": 0.0951,
+      "step": 1008
+    },
+    {
+      "epoch": 4.9655596555965555,
+      "grad_norm": 1.1796875,
+      "learning_rate": 8.794221499408562e-10,
+      "loss": 0.0977,
+      "step": 1009
+    },
+    {
+      "epoch": 4.970479704797048,
+      "grad_norm": 1.2578125,
+      "learning_rate": 6.107152967349539e-10,
+      "loss": 0.1068,
+      "step": 1010
+    },
+    {
+      "epoch": 4.97539975399754,
+      "grad_norm": 1.0078125,
+      "learning_rate": 3.908606544228422e-10,
+      "loss": 0.0803,
+      "step": 1011
+    },
+    {
+      "epoch": 4.980319803198032,
+      "grad_norm": 0.88671875,
+      "learning_rate": 2.1986037134325012e-10,
+      "loss": 0.0667,
+      "step": 1012
+    },
+    {
+      "epoch": 4.985239852398524,
+      "grad_norm": 0.86328125,
+      "learning_rate": 9.771611844955343e-11,
+      "loss": 0.0616,
+      "step": 1013
+    },
+    {
+      "epoch": 4.990159901599016,
+      "grad_norm": 1.0546875,
+      "learning_rate": 2.4429089290345375e-11,
+      "loss": 0.088,
+      "step": 1014
+    },
+    {
+      "epoch": 4.995079950799508,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0,
+      "loss": 0.1053,
+      "step": 1015
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1015,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 51,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.1980182581321662e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}