diff --git "a/checkpoint-734/trainer_state.json" "b/checkpoint-734/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-734/trainer_state.json"
@@ -0,0 +1,5223 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9809394145677333,
+  "eval_steps": 92,
+  "global_step": 734,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.17789718508720398,
+      "learning_rate": 2e-05,
+      "loss": 2.3216,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 2.256117105484009,
+      "eval_runtime": 52.5552,
+      "eval_samples_per_second": 95.081,
+      "eval_steps_per_second": 23.785,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.1491609364748001,
+      "learning_rate": 4e-05,
+      "loss": 2.2689,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.16114787757396698,
+      "learning_rate": 6e-05,
+      "loss": 2.1799,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.17484058439731598,
+      "learning_rate": 8e-05,
+      "loss": 2.2857,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.16328556835651398,
+      "learning_rate": 0.0001,
+      "loss": 2.1805,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.2006424516439438,
+      "learning_rate": 0.00012,
+      "loss": 2.3171,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.15052957832813263,
+      "learning_rate": 0.00014,
+      "loss": 2.2048,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1444414258003235,
+      "learning_rate": 0.00016,
+      "loss": 2.181,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.174555703997612,
+      "learning_rate": 0.00018,
+      "loss": 2.0824,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.20063771307468414,
+      "learning_rate": 0.0002,
+      "loss": 2.1274,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.19133329391479492,
+      "learning_rate": 0.00019999958540892524,
+      "loss": 2.2436,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.19068902730941772,
+      "learning_rate": 0.00019999834163913867,
+      "loss": 2.0865,
+      "step": 12
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.12937726080417633,
+      "learning_rate": 0.00019999626870095333,
+      "loss": 1.9834,
+      "step": 13
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.12377699464559555,
+      "learning_rate": 0.0001999933666115578,
+      "loss": 2.0084,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.11349006742238998,
+      "learning_rate": 0.00019998963539501557,
+      "loss": 2.0472,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.10255803167819977,
+      "learning_rate": 0.00019998507508226524,
+      "loss": 2.0017,
+      "step": 16
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.1226801723241806,
+      "learning_rate": 0.00019997968571112013,
+      "loss": 2.0241,
+      "step": 17
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.09784600138664246,
+      "learning_rate": 0.00019997346732626795,
+      "loss": 1.9264,
+      "step": 18
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.0987139344215393,
+      "learning_rate": 0.00019996641997927043,
+      "loss": 2.0021,
+      "step": 19
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.1008625328540802,
+      "learning_rate": 0.0001999585437285629,
+      "loss": 1.9401,
+      "step": 20
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.11675934493541718,
+      "learning_rate": 0.00019994983863945388,
+      "loss": 2.0533,
+      "step": 21
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.1078399121761322,
+      "learning_rate": 0.0001999403047841243,
+      "loss": 2.067,
+      "step": 22
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.10346037149429321,
+      "learning_rate": 0.00019992994224162728,
+      "loss": 1.9767,
+      "step": 23
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.11017748713493347,
+      "learning_rate": 0.0001999187510978872,
+      "loss": 1.9501,
+      "step": 24
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.11317062377929688,
+      "learning_rate": 0.00019990673144569892,
+      "loss": 1.9425,
+      "step": 25
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.10313500463962555,
+      "learning_rate": 0.0001998938833847273,
+      "loss": 1.9594,
+      "step": 26
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.1095932349562645,
+      "learning_rate": 0.00019988020702150618,
+      "loss": 1.951,
+      "step": 27
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.11742599308490753,
+      "learning_rate": 0.00019986570246943754,
+      "loss": 2.0445,
+      "step": 28
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.14208655059337616,
+      "learning_rate": 0.00019985036984879052,
+      "loss": 1.9932,
+      "step": 29
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.10696502774953842,
+      "learning_rate": 0.00019983420928670044,
+      "loss": 1.9076,
+      "step": 30
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.10935875028371811,
+      "learning_rate": 0.00019981722091716783,
+      "loss": 1.8718,
+      "step": 31
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.11814530938863754,
+      "learning_rate": 0.00019979940488105722,
+      "loss": 1.8591,
+      "step": 32
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.12232942134141922,
+      "learning_rate": 0.00019978076132609599,
+      "loss": 1.9379,
+      "step": 33
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.11652842909097672,
+      "learning_rate": 0.00019976129040687318,
+      "loss": 1.9731,
+      "step": 34
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.1147683709859848,
+      "learning_rate": 0.0001997409922848381,
+      "loss": 1.8994,
+      "step": 35
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.1180400475859642,
+      "learning_rate": 0.00019971986712829932,
+      "loss": 1.8678,
+      "step": 36
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.13028429448604584,
+      "learning_rate": 0.0001996979151124227,
+      "loss": 1.9853,
+      "step": 37
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.11380694806575775,
+      "learning_rate": 0.00019967513641923056,
+      "loss": 1.8388,
+      "step": 38
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.13089893758296967,
+      "learning_rate": 0.00019965153123759966,
+      "loss": 1.8802,
+      "step": 39
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.12069398909807205,
+      "learning_rate": 0.00019962709976326,
+      "loss": 1.9847,
+      "step": 40
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.12704235315322876,
+      "learning_rate": 0.00019960184219879303,
+      "loss": 2.0103,
+      "step": 41
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.15785923600196838,
+      "learning_rate": 0.00019957575875362993,
+      "loss": 1.9266,
+      "step": 42
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.137678325176239,
+      "learning_rate": 0.00019954884964404996,
+      "loss": 1.827,
+      "step": 43
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.12916652858257294,
+      "learning_rate": 0.0001995211150931787,
+      "loss": 1.8881,
+      "step": 44
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.14117199182510376,
+      "learning_rate": 0.00019949255533098604,
+      "loss": 1.9003,
+      "step": 45
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.122989721596241,
+      "learning_rate": 0.00019946317059428448,
+      "loss": 1.7859,
+      "step": 46
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.14118267595767975,
+      "learning_rate": 0.00019943296112672696,
+      "loss": 1.9206,
+      "step": 47
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.14580337703227997,
+      "learning_rate": 0.00019940192717880502,
+      "loss": 2.0004,
+      "step": 48
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.1319318562746048,
+      "learning_rate": 0.00019937006900784665,
+      "loss": 1.9288,
+      "step": 49
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.13234923779964447,
+      "learning_rate": 0.00019933738687801403,
+      "loss": 1.8757,
+      "step": 50
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.13480357825756073,
+      "learning_rate": 0.00019930388106030166,
+      "loss": 1.773,
+      "step": 51
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.15075266361236572,
+      "learning_rate": 0.00019926955183253374,
+      "loss": 1.9003,
+      "step": 52
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2747078537940979,
+      "learning_rate": 0.00019923439947936204,
+      "loss": 1.8076,
+      "step": 53
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.14992934465408325,
+      "learning_rate": 0.00019919842429226372,
+      "loss": 1.9598,
+      "step": 54
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.15926502645015717,
+      "learning_rate": 0.0001991616265695385,
+      "loss": 1.8815,
+      "step": 55
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.18423113226890564,
+      "learning_rate": 0.00019912400661630658,
+      "loss": 2.0042,
+      "step": 56
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.15437527000904083,
+      "learning_rate": 0.00019908556474450593,
+      "loss": 1.8524,
+      "step": 57
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.15860721468925476,
+      "learning_rate": 0.0001990463012728896,
+      "loss": 1.9301,
+      "step": 58
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.1519562005996704,
+      "learning_rate": 0.00019900621652702336,
+      "loss": 1.8769,
+      "step": 59
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.13514740765094757,
+      "learning_rate": 0.00019896531083928273,
+      "loss": 1.7548,
+      "step": 60
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.15928471088409424,
+      "learning_rate": 0.00019892358454885042,
+      "loss": 1.8667,
+      "step": 61
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.15786142647266388,
+      "learning_rate": 0.00019888103800171333,
+      "loss": 1.8641,
+      "step": 62
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.1386164277791977,
+      "learning_rate": 0.00019883767155065986,
+      "loss": 1.8872,
+      "step": 63
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.14870542287826538,
+      "learning_rate": 0.00019879348555527687,
+      "loss": 1.8602,
+      "step": 64
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.14343470335006714,
+      "learning_rate": 0.00019874848038194673,
+      "loss": 1.8312,
+      "step": 65
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.15904268622398376,
+      "learning_rate": 0.00019870265640384435,
+      "loss": 1.9288,
+      "step": 66
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.1546896994113922,
+      "learning_rate": 0.00019865601400093395,
+      "loss": 1.9167,
+      "step": 67
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.15036776661872864,
+      "learning_rate": 0.000198608553559966,
+      "loss": 1.8157,
+      "step": 68
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.155603289604187,
+      "learning_rate": 0.000198560275474474,
+      "loss": 1.8193,
+      "step": 69
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.14846371114253998,
+      "learning_rate": 0.00019851118014477126,
+      "loss": 1.7431,
+      "step": 70
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.15555807948112488,
+      "learning_rate": 0.00019846126797794743,
+      "loss": 1.9226,
+      "step": 71
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.14305201172828674,
+      "learning_rate": 0.00019841053938786534,
+      "loss": 1.8991,
+      "step": 72
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.14723321795463562,
+      "learning_rate": 0.00019835899479515737,
+      "loss": 1.881,
+      "step": 73
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.14168229699134827,
+      "learning_rate": 0.00019830663462722207,
+      "loss": 1.9365,
+      "step": 74
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.16648411750793457,
+      "learning_rate": 0.00019825345931822068,
+      "loss": 1.764,
+      "step": 75
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.13298211991786957,
+      "learning_rate": 0.00019819946930907332,
+      "loss": 1.8581,
+      "step": 76
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.15933053195476532,
+      "learning_rate": 0.00019814466504745548,
+      "loss": 1.8043,
+      "step": 77
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.15350531041622162,
+      "learning_rate": 0.00019808904698779433,
+      "loss": 1.7964,
+      "step": 78
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.17690815031528473,
+      "learning_rate": 0.00019803261559126492,
+      "loss": 1.8173,
+      "step": 79
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.14556676149368286,
+      "learning_rate": 0.0001979753713257863,
+      "loss": 1.9111,
+      "step": 80
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.14608439803123474,
+      "learning_rate": 0.00019791731466601773,
+      "loss": 1.8583,
+      "step": 81
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.15660099685192108,
+      "learning_rate": 0.0001978584460933546,
+      "loss": 1.8686,
+      "step": 82
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2378729283809662,
+      "learning_rate": 0.00019779876609592471,
+      "loss": 1.7445,
+      "step": 83
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.17274336516857147,
+      "learning_rate": 0.00019773827516858386,
+      "loss": 1.9353,
+      "step": 84
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.15144622325897217,
+      "learning_rate": 0.00019767697381291202,
+      "loss": 1.8179,
+      "step": 85
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.1504279226064682,
+      "learning_rate": 0.00019761486253720915,
+      "loss": 1.787,
+      "step": 86
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.14772076904773712,
+      "learning_rate": 0.0001975519418564908,
+      "loss": 1.8928,
+      "step": 87
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.16531221568584442,
+      "learning_rate": 0.00019748821229248405,
+      "loss": 1.7052,
+      "step": 88
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.15720781683921814,
+      "learning_rate": 0.00019742367437362306,
+      "loss": 1.8116,
+      "step": 89
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.15642230212688446,
+      "learning_rate": 0.00019735832863504476,
+      "loss": 1.7233,
+      "step": 90
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.13698022067546844,
+      "learning_rate": 0.00019729217561858433,
+      "loss": 1.7699,
+      "step": 91
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.16550354659557343,
+      "learning_rate": 0.00019722521587277076,
+      "loss": 1.7379,
+      "step": 92
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 1.785460114479065,
+      "eval_runtime": 54.2174,
+      "eval_samples_per_second": 92.166,
+      "eval_steps_per_second": 23.055,
+      "step": 92
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.1539517045021057,
+      "learning_rate": 0.00019715744995282234,
+      "loss": 1.7921,
+      "step": 93
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.1603965014219284,
+      "learning_rate": 0.00019708887842064194,
+      "loss": 1.7856,
+      "step": 94
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.14334900677204132,
+      "learning_rate": 0.0001970195018448125,
+      "loss": 1.8048,
+      "step": 95
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.17852160334587097,
+      "learning_rate": 0.00019694932080059217,
+      "loss": 1.7513,
+      "step": 96
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.15032339096069336,
+      "learning_rate": 0.00019687833586990973,
+      "loss": 1.7011,
+      "step": 97
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.1456947922706604,
+      "learning_rate": 0.00019680654764135945,
+      "loss": 1.7992,
+      "step": 98
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.14242622256278992,
+      "learning_rate": 0.00019673395671019654,
+      "loss": 1.8617,
+      "step": 99
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.15148206055164337,
+      "learning_rate": 0.00019666056367833204,
+      "loss": 1.8464,
+      "step": 100
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.15412554144859314,
+      "learning_rate": 0.00019658636915432788,
+      "loss": 1.8757,
+      "step": 101
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.1881706863641739,
+      "learning_rate": 0.0001965113737533918,
+      "loss": 1.7233,
+      "step": 102
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.16812562942504883,
+      "learning_rate": 0.0001964355780973723,
+      "loss": 1.769,
+      "step": 103
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.14640142023563385,
+      "learning_rate": 0.00019635898281475346,
+      "loss": 1.8155,
+      "step": 104
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.16034264862537384,
+      "learning_rate": 0.00019628158854064956,
+      "loss": 1.7034,
+      "step": 105
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.14522109925746918,
+      "learning_rate": 0.00019620339591680023,
+      "loss": 1.8194,
+      "step": 106
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.1690577268600464,
+      "learning_rate": 0.0001961244055915647,
+      "loss": 1.7467,
+      "step": 107
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.17564715445041656,
+      "learning_rate": 0.00019604461821991667,
+      "loss": 1.8696,
+      "step": 108
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.1871030628681183,
+      "learning_rate": 0.00019596403446343877,
+      "loss": 1.8975,
+      "step": 109
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.18129503726959229,
+      "learning_rate": 0.0001958826549903171,
+      "loss": 1.8295,
+      "step": 110
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.16357164084911346,
+      "learning_rate": 0.00019580048047533578,
+      "loss": 1.813,
+      "step": 111
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.15929792821407318,
+      "learning_rate": 0.00019571751159987115,
+      "loss": 1.7072,
+      "step": 112
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.14020949602127075,
+      "learning_rate": 0.00019563374905188637,
+      "loss": 1.8245,
+      "step": 113
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.14771193265914917,
+      "learning_rate": 0.0001955491935259255,
+      "loss": 1.7532,
+      "step": 114
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.1625107079744339,
+      "learning_rate": 0.0001954638457231079,
+      "loss": 1.7653,
+      "step": 115
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.15289072692394257,
+      "learning_rate": 0.0001953777063511223,
+      "loss": 1.7301,
+      "step": 116
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.15066611766815186,
+      "learning_rate": 0.00019529077612422103,
+      "loss": 1.7909,
+      "step": 117
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.16926345229148865,
+      "learning_rate": 0.00019520305576321395,
+      "loss": 1.7426,
+      "step": 118
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.16127844154834747,
+      "learning_rate": 0.00019511454599546268,
+      "loss": 1.7841,
+      "step": 119
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.14246778190135956,
+      "learning_rate": 0.0001950252475548744,
+      "loss": 1.7545,
+      "step": 120
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.15499025583267212,
+      "learning_rate": 0.00019493516118189582,
+      "loss": 1.7749,
+      "step": 121
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.16287535429000854,
+      "learning_rate": 0.00019484428762350708,
+      "loss": 1.7878,
+      "step": 122
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.14317578077316284,
+      "learning_rate": 0.00019475262763321552,
+      "loss": 1.8787,
+      "step": 123
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.15122784674167633,
+      "learning_rate": 0.00019466018197104946,
+      "loss": 1.8259,
+      "step": 124
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1579144448041916,
+      "learning_rate": 0.00019456695140355172,
+      "loss": 1.8568,
+      "step": 125
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.15908168256282806,
+      "learning_rate": 0.0001944729367037736,
+      "loss": 1.7898,
+      "step": 126
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.17988181114196777,
+      "learning_rate": 0.0001943781386512682,
+      "loss": 1.8017,
+      "step": 127
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.16047564148902893,
+      "learning_rate": 0.000194282558032084,
+      "loss": 1.6842,
+      "step": 128
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.2199389785528183,
+      "learning_rate": 0.0001941861956387585,
+      "loss": 1.7777,
+      "step": 129
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.17500443756580353,
+      "learning_rate": 0.00019408905227031142,
+      "loss": 1.817,
+      "step": 130
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.16257035732269287,
+      "learning_rate": 0.00019399112873223824,
+      "loss": 1.7822,
+      "step": 131
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.1655006855726242,
+      "learning_rate": 0.0001938924258365035,
+      "loss": 1.7835,
+      "step": 132
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.1625986397266388,
+      "learning_rate": 0.00019379294440153392,
+      "loss": 1.739,
+      "step": 133
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.1542993187904358,
+      "learning_rate": 0.00019369268525221185,
+      "loss": 1.8484,
+      "step": 134
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.16408205032348633,
+      "learning_rate": 0.00019359164921986825,
+      "loss": 1.7889,
+      "step": 135
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.1816944181919098,
+      "learning_rate": 0.00019348983714227583,
+      "loss": 1.8592,
+      "step": 136
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.1641003042459488,
+      "learning_rate": 0.00019338724986364222,
+      "loss": 1.7293,
+      "step": 137
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.14509302377700806,
+      "learning_rate": 0.00019328388823460283,
+      "loss": 1.653,
+      "step": 138
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.14293956756591797,
+      "learning_rate": 0.00019317975311221373,
+      "loss": 1.8303,
+      "step": 139
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.17679671943187714,
+      "learning_rate": 0.00019307484535994492,
+      "loss": 1.7262,
+      "step": 140
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.142549067735672,
+      "learning_rate": 0.00019296916584767262,
+      "loss": 1.8038,
+      "step": 141
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.17163245379924774,
+      "learning_rate": 0.00019286271545167254,
+      "loss": 1.7815,
+      "step": 142
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.1682298481464386,
+      "learning_rate": 0.0001927554950546124,
+      "loss": 1.7159,
+      "step": 143
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.17430084943771362,
+      "learning_rate": 0.00019264750554554453,
+      "loss": 1.8462,
+      "step": 144
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.1589604616165161,
+      "learning_rate": 0.00019253874781989864,
+      "loss": 1.7369,
+      "step": 145
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.15902858972549438,
+      "learning_rate": 0.00019242922277947448,
+      "loss": 1.8005,
+      "step": 146
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.15692317485809326,
+      "learning_rate": 0.00019231893133243405,
+      "loss": 1.7759,
+      "step": 147
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.15179891884326935,
+      "learning_rate": 0.00019220787439329438,
+      "loss": 1.8517,
+      "step": 148
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.14860700070858002,
+      "learning_rate": 0.00019209605288291978,
+      "loss": 1.7751,
+      "step": 149
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.1454017162322998,
+      "learning_rate": 0.0001919834677285142,
+      "loss": 1.6579,
+      "step": 150
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.16802151501178741,
+      "learning_rate": 0.00019187011986361374,
+      "loss": 1.7853,
+      "step": 151
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.16814345121383667,
+      "learning_rate": 0.0001917560102280786,
+      "loss": 1.7162,
+      "step": 152
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.15314815938472748,
+      "learning_rate": 0.0001916411397680855,
+      "loss": 1.7066,
+      "step": 153
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.16011942923069,
+      "learning_rate": 0.00019152550943611987,
+      "loss": 1.7655,
+      "step": 154
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.15423813462257385,
+      "learning_rate": 0.00019140912019096766,
+      "loss": 1.8106,
+      "step": 155
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.1538185477256775,
+      "learning_rate": 0.0001912919729977078,
+      "loss": 1.7952,
+      "step": 156
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.16051077842712402,
+      "learning_rate": 0.00019117406882770388,
+      "loss": 1.7182,
+      "step": 157
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.15189512073993683,
+      "learning_rate": 0.00019105540865859623,
+      "loss": 1.6282,
+      "step": 158
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.16353580355644226,
+      "learning_rate": 0.0001909359934742938,
+      "loss": 1.7374,
+      "step": 159
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.16629937291145325,
+      "learning_rate": 0.00019081582426496598,
+      "loss": 1.6122,
+      "step": 160
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.16899485886096954,
+      "learning_rate": 0.00019069490202703438,
+      "loss": 1.6817,
+      "step": 161
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.1827431470155716,
+      "learning_rate": 0.00019057322776316467,
+      "loss": 1.7633,
+      "step": 162
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.17833983898162842,
+      "learning_rate": 0.00019045080248225805,
+      "loss": 1.6124,
+      "step": 163
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.1423504501581192,
+      "learning_rate": 0.00019032762719944316,
+      "loss": 1.7344,
+      "step": 164
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.1658603847026825,
+      "learning_rate": 0.0001902037029360674,
+      "loss": 1.6874,
+      "step": 165
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.16833803057670593,
+      "learning_rate": 0.00019007903071968868,
+      "loss": 1.7291,
+      "step": 166
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.16885894536972046,
+      "learning_rate": 0.00018995361158406676,
+      "loss": 1.77,
+      "step": 167
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.16659806668758392,
+      "learning_rate": 0.00018982744656915475,
+      "loss": 1.8441,
+      "step": 168
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.18617461621761322,
+      "learning_rate": 0.00018970053672109037,
+      "loss": 1.8355,
+      "step": 169
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.17858782410621643,
+      "learning_rate": 0.00018957288309218745,
+      "loss": 1.7093,
+      "step": 170
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.1546701192855835,
+      "learning_rate": 0.00018944448674092714,
+      "loss": 1.6651,
+      "step": 171
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.16935370862483978,
+      "learning_rate": 0.000189315348731949,
+      "loss": 1.7196,
+      "step": 172
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.17772410809993744,
+      "learning_rate": 0.0001891854701360424,
+      "loss": 1.7215,
+      "step": 173
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.16293296217918396,
+      "learning_rate": 0.00018905485203013744,
+      "loss": 1.7094,
+      "step": 174
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.1712435483932495,
+      "learning_rate": 0.00018892349549729615,
+      "loss": 1.7627,
+      "step": 175
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.19540230929851532,
+      "learning_rate": 0.00018879140162670347,
+      "loss": 1.7626,
+      "step": 176
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.16615939140319824,
+      "learning_rate": 0.00018865857151365814,
+      "loss": 1.6614,
+      "step": 177
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.14994904398918152,
+      "learning_rate": 0.0001885250062595638,
+      "loss": 1.8007,
+      "step": 178
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.17429371178150177,
+      "learning_rate": 0.0001883907069719197,
+      "loss": 1.5822,
+      "step": 179
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.1600300520658493,
+      "learning_rate": 0.0001882556747643115,
+      "loss": 1.7315,
+      "step": 180
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.16705308854579926,
+      "learning_rate": 0.00018811991075640223,
+      "loss": 1.6875,
+      "step": 181
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.1432919055223465,
+      "learning_rate": 0.00018798341607392276,
+      "loss": 1.662,
+      "step": 182
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.1693781316280365,
+      "learning_rate": 0.00018784619184866267,
+      "loss": 1.6294,
+      "step": 183
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.16618549823760986,
+      "learning_rate": 0.0001877082392184607,
+      "loss": 1.6935,
+      "step": 184
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 1.707476258277893,
+      "eval_runtime": 53.476,
+      "eval_samples_per_second": 93.444,
+      "eval_steps_per_second": 23.375,
+      "step": 184
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.19029615819454193,
+      "learning_rate": 0.00018756955932719546,
+      "loss": 1.6409,
+      "step": 185
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.15953810513019562,
+      "learning_rate": 0.00018743015332477588,
+      "loss": 1.7592,
+      "step": 186
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1445915549993515,
+      "learning_rate": 0.0001872900223671316,
+      "loss": 1.7382,
+      "step": 187
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1905539631843567,
+      "learning_rate": 0.0001871491676162035,
+      "loss": 1.7496,
+      "step": 188
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.19065023958683014,
+      "learning_rate": 0.0001870075902399341,
+      "loss": 1.7459,
+      "step": 189
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.15250128507614136,
+      "learning_rate": 0.00018686529141225767,
+      "loss": 1.7847,
+      "step": 190
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.16181302070617676,
+      "learning_rate": 0.00018672227231309068,
+      "loss": 1.6404,
+      "step": 191
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.15402203798294067,
+      "learning_rate": 0.00018657853412832202,
+      "loss": 1.7612,
+      "step": 192
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.16924121975898743,
+      "learning_rate": 0.00018643407804980303,
+      "loss": 1.7262,
+      "step": 193
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.1564388871192932,
+      "learning_rate": 0.00018628890527533777,
+      "loss": 1.7721,
+      "step": 194
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.17039404809474945,
+      "learning_rate": 0.00018614301700867287,
+      "loss": 1.6788,
+      "step": 195
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.1548219621181488,
+      "learning_rate": 0.0001859964144594879,
+      "loss": 1.7128,
+      "step": 196
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.16957992315292358,
+      "learning_rate": 0.000185849098843385,
+      "loss": 1.8554,
+      "step": 197
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.15887251496315002,
+      "learning_rate": 0.00018570107138187893,
+      "loss": 1.691,
+      "step": 198
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.1746351420879364,
+      "learning_rate": 0.000185552333302387,
+      "loss": 1.6902,
+      "step": 199
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.15400560200214386,
+      "learning_rate": 0.00018540288583821882,
+      "loss": 1.7756,
+      "step": 200
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.163401260972023,
+      "learning_rate": 0.00018525273022856607,
+      "loss": 1.6391,
+      "step": 201
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.14885050058364868,
+      "learning_rate": 0.0001851018677184923,
+      "loss": 1.7449,
+      "step": 202
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.15566030144691467,
+      "learning_rate": 0.00018495029955892248,
+      "loss": 1.7574,
+      "step": 203
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.16466747224330902,
+      "learning_rate": 0.00018479802700663268,
+      "loss": 1.7116,
+      "step": 204
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.15742923319339752,
+      "learning_rate": 0.00018464505132423983,
+      "loss": 1.6491,
+      "step": 205
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.15772201120853424,
+      "learning_rate": 0.00018449137378019094,
+      "loss": 1.7971,
+      "step": 206
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.17034031450748444,
+      "learning_rate": 0.00018433699564875274,
+      "loss": 1.7036,
+      "step": 207
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.1637440025806427,
+      "learning_rate": 0.0001841819182100012,
+      "loss": 1.5544,
+      "step": 208
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.18623477220535278,
+      "learning_rate": 0.00018402614274981073,
+      "loss": 1.632,
+      "step": 209
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.16553111374378204,
+      "learning_rate": 0.0001838696705598436,
+      "loss": 1.7081,
+      "step": 210
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.15178020298480988,
+      "learning_rate": 0.0001837125029375393,
+      "loss": 1.679,
+      "step": 211
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.16419006884098053,
+      "learning_rate": 0.00018355464118610372,
+      "loss": 1.7082,
+      "step": 212
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.1607130765914917,
+      "learning_rate": 0.00018339608661449832,
+      "loss": 1.6493,
+      "step": 213
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.17754729092121124,
+      "learning_rate": 0.00018323684053742932,
+      "loss": 1.7558,
+      "step": 214
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.16623522341251373,
+      "learning_rate": 0.00018307690427533672,
+      "loss": 1.7677,
+      "step": 215
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.16194763779640198,
+      "learning_rate": 0.00018291627915438348,
+      "loss": 1.7749,
+      "step": 216
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.17232032120227814,
+      "learning_rate": 0.00018275496650644444,
+      "loss": 1.8273,
+      "step": 217
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.16804206371307373,
+      "learning_rate": 0.00018259296766909527,
+      "loss": 1.7504,
+      "step": 218
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.1571667343378067,
+      "learning_rate": 0.00018243028398560142,
+      "loss": 1.7331,
+      "step": 219
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.16855494678020477,
+      "learning_rate": 0.00018226691680490694,
+      "loss": 1.694,
+      "step": 220
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.18118086457252502,
+      "learning_rate": 0.00018210286748162336,
+      "loss": 1.7338,
+      "step": 221
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.16491609811782837,
+      "learning_rate": 0.00018193813737601834,
+      "loss": 1.5984,
+      "step": 222
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.15478603541851044,
+      "learning_rate": 0.00018177272785400455,
+      "loss": 1.7077,
+      "step": 223
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.14907459914684296,
+      "learning_rate": 0.0001816066402871282,
+      "loss": 1.6867,
+      "step": 224
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.16465577483177185,
+      "learning_rate": 0.00018143987605255774,
+      "loss": 1.7904,
+      "step": 225
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1661553680896759,
+      "learning_rate": 0.00018127243653307248,
+      "loss": 1.7439,
+      "step": 226
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.15185391902923584,
+      "learning_rate": 0.00018110432311705096,
+      "loss": 1.6711,
+      "step": 227
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.15746577084064484,
+      "learning_rate": 0.00018093553719845967,
+      "loss": 1.7139,
+      "step": 228
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.16057291626930237,
+      "learning_rate": 0.00018076608017684128,
+      "loss": 1.6659,
+      "step": 229
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.17637218534946442,
+      "learning_rate": 0.00018059595345730318,
+      "loss": 1.7832,
+      "step": 230
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.16048584878444672,
+      "learning_rate": 0.00018042515845050576,
+      "loss": 1.8092,
+      "step": 231
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.1690172702074051,
+      "learning_rate": 0.00018025369657265075,
+      "loss": 1.6495,
+      "step": 232
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.1493247002363205,
+      "learning_rate": 0.0001800815692454694,
+      "loss": 1.6867,
+      "step": 233
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1496494710445404,
+      "learning_rate": 0.00017990877789621082,
+      "loss": 1.6808,
+      "step": 234
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1575721800327301,
+      "learning_rate": 0.00017973532395763,
+      "loss": 1.7341,
+      "step": 235
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.16140134632587433,
+      "learning_rate": 0.00017956120886797604,
+      "loss": 1.7025,
+      "step": 236
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.1708807647228241,
+      "learning_rate": 0.0001793864340709802,
+      "loss": 1.7168,
+      "step": 237
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.1652187556028366,
+      "learning_rate": 0.00017921100101584388,
+      "loss": 1.6333,
+      "step": 238
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.15541191399097443,
+      "learning_rate": 0.00017903491115722666,
+      "loss": 1.6765,
+      "step": 239
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.15625408291816711,
+      "learning_rate": 0.00017885816595523423,
+      "loss": 1.7528,
+      "step": 240
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.171896293759346,
+      "learning_rate": 0.00017868076687540624,
+      "loss": 1.6741,
+      "step": 241
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1624470204114914,
+      "learning_rate": 0.00017850271538870423,
+      "loss": 1.5913,
+      "step": 242
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.163376584649086,
+      "learning_rate": 0.00017832401297149928,
+      "loss": 1.6242,
+      "step": 243
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1690063625574112,
+      "learning_rate": 0.00017814466110555998,
+      "loss": 1.6997,
+      "step": 244
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.18232744932174683,
+      "learning_rate": 0.00017796466127804,
+      "loss": 1.7033,
+      "step": 245
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.16556936502456665,
+      "learning_rate": 0.0001777840149814657,
+      "loss": 1.8036,
+      "step": 246
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.1530187577009201,
+      "learning_rate": 0.00017760272371372402,
+      "loss": 1.7847,
+      "step": 247
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.17229686677455902,
+      "learning_rate": 0.00017742078897804974,
+      "loss": 1.6999,
+      "step": 248
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.1453051120042801,
+      "learning_rate": 0.00017723821228301322,
+      "loss": 1.7255,
+      "step": 249
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.1597098410129547,
+      "learning_rate": 0.00017705499514250784,
+      "loss": 1.7228,
+      "step": 250
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.20154576003551483,
+      "learning_rate": 0.0001768711390757374,
+      "loss": 1.6099,
+      "step": 251
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.17710669338703156,
+      "learning_rate": 0.0001766866456072036,
+      "loss": 1.6167,
+      "step": 252
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.15468710660934448,
+      "learning_rate": 0.00017650151626669337,
+      "loss": 1.6875,
+      "step": 253
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.18249762058258057,
+      "learning_rate": 0.0001763157525892661,
+      "loss": 1.5819,
+      "step": 254
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.16782326996326447,
+      "learning_rate": 0.0001761293561152411,
+      "loss": 1.626,
+      "step": 255
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.16319702565670013,
+      "learning_rate": 0.0001759423283901846,
+      "loss": 1.6173,
+      "step": 256
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.15339410305023193,
+      "learning_rate": 0.00017575467096489717,
+      "loss": 1.7897,
+      "step": 257
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.15668465197086334,
+      "learning_rate": 0.00017556638539540067,
+      "loss": 1.6749,
+      "step": 258
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.1564616560935974,
+      "learning_rate": 0.00017537747324292542,
+      "loss": 1.6263,
+      "step": 259
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.15310418605804443,
+      "learning_rate": 0.00017518793607389723,
+      "loss": 1.7237,
+      "step": 260
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.1633821427822113,
+      "learning_rate": 0.00017499777545992452,
+      "loss": 1.7156,
+      "step": 261
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.18414661288261414,
+      "learning_rate": 0.00017480699297778518,
+      "loss": 1.6333,
+      "step": 262
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.18508172035217285,
+      "learning_rate": 0.00017461559020941341,
+      "loss": 1.681,
+      "step": 263
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.17592370510101318,
+      "learning_rate": 0.0001744235687418869,
+      "loss": 1.7787,
+      "step": 264
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.22177156805992126,
+      "learning_rate": 0.00017423093016741333,
+      "loss": 1.6066,
+      "step": 265
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.1583690643310547,
+      "learning_rate": 0.00017403767608331733,
+      "loss": 1.7423,
+      "step": 266
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.17558494210243225,
+      "learning_rate": 0.00017384380809202737,
+      "loss": 1.7306,
+      "step": 267
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.1737305372953415,
+      "learning_rate": 0.00017364932780106212,
+      "loss": 1.8139,
+      "step": 268
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.15237008035182953,
+      "learning_rate": 0.00017345423682301755,
+      "loss": 1.7415,
+      "step": 269
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.1741713285446167,
+      "learning_rate": 0.00017325853677555312,
+      "loss": 1.73,
+      "step": 270
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.1688138246536255,
+      "learning_rate": 0.00017306222928137875,
+      "loss": 1.774,
+      "step": 271
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.16588321328163147,
+      "learning_rate": 0.00017286531596824112,
+      "loss": 1.758,
+      "step": 272
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.17456404864788055,
+      "learning_rate": 0.00017266779846891029,
+      "loss": 1.7528,
+      "step": 273
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.15763048827648163,
+      "learning_rate": 0.00017246967842116605,
+      "loss": 1.661,
+      "step": 274
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.1601736843585968,
+      "learning_rate": 0.00017227095746778456,
+      "loss": 1.7317,
+      "step": 275
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.17678223550319672,
+      "learning_rate": 0.00017207163725652445,
+      "loss": 1.7016,
+      "step": 276
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 1.666338324546814,
+      "eval_runtime": 53.3064,
+      "eval_samples_per_second": 93.741,
+      "eval_steps_per_second": 23.449,
+      "step": 276
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.1592448651790619,
+      "learning_rate": 0.00017187171944011329,
+      "loss": 1.6313,
+      "step": 277
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.1765425056219101,
+      "learning_rate": 0.000171671205676234,
+      "loss": 1.5646,
+      "step": 278
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.14587025344371796,
+      "learning_rate": 0.00017147009762751085,
+      "loss": 1.7655,
+      "step": 279
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.15663780272006989,
+      "learning_rate": 0.00017126839696149596,
+      "loss": 1.6468,
+      "step": 280
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.15041159093379974,
+      "learning_rate": 0.00017106610535065517,
+      "loss": 1.6655,
+      "step": 281
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.18636269867420197,
+      "learning_rate": 0.0001708632244723545,
+      "loss": 1.7274,
+      "step": 282
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.2026071399450302,
+      "learning_rate": 0.00017065975600884584,
+      "loss": 1.7669,
+      "step": 283
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.16321098804473877,
+      "learning_rate": 0.00017045570164725348,
+      "loss": 1.5634,
+      "step": 284
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.17123541235923767,
+      "learning_rate": 0.00017025106307955973,
+      "loss": 1.6881,
+      "step": 285
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.17135639488697052,
+      "learning_rate": 0.00017004584200259107,
+      "loss": 1.5511,
+      "step": 286
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.149318128824234,
+      "learning_rate": 0.00016984004011800404,
+      "loss": 1.7526,
+      "step": 287
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.17461049556732178,
+      "learning_rate": 0.0001696336591322711,
+      "loss": 1.7297,
+      "step": 288
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.22167545557022095,
+      "learning_rate": 0.00016942670075666656,
+      "loss": 1.5982,
+      "step": 289
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.17257270216941833,
+      "learning_rate": 0.00016921916670725233,
+      "loss": 1.6988,
+      "step": 290
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.16062453389167786,
+      "learning_rate": 0.00016901105870486372,
+      "loss": 1.621,
+      "step": 291
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.18726034462451935,
+      "learning_rate": 0.00016880237847509514,
+      "loss": 1.6037,
+      "step": 292
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.1890726536512375,
+      "learning_rate": 0.00016859312774828578,
+      "loss": 1.6991,
+      "step": 293
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.1621987223625183,
+      "learning_rate": 0.0001683833082595053,
+      "loss": 1.6365,
+      "step": 294
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.15142592787742615,
+      "learning_rate": 0.0001681729217485395,
+      "loss": 1.5566,
+      "step": 295
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.15878473222255707,
+      "learning_rate": 0.0001679619699598757,
+      "loss": 1.6339,
+      "step": 296
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.19840176403522491,
+      "learning_rate": 0.00016775045464268855,
+      "loss": 1.7045,
+      "step": 297
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.15840516984462738,
+      "learning_rate": 0.00016753837755082527,
+      "loss": 1.7066,
+      "step": 298
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.14891847968101501,
+      "learning_rate": 0.00016732574044279122,
+      "loss": 1.6095,
+      "step": 299
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.1543874889612198,
+      "learning_rate": 0.0001671125450817354,
+      "loss": 1.6635,
+      "step": 300
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.1737508326768875,
+      "learning_rate": 0.00016689879323543566,
+      "loss": 1.7243,
+      "step": 301
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.1600562036037445,
+      "learning_rate": 0.00016668448667628418,
+      "loss": 1.6657,
+      "step": 302
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.16577617824077606,
+      "learning_rate": 0.00016646962718127264,
+      "loss": 1.6408,
+      "step": 303
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1584424525499344,
+      "learning_rate": 0.00016625421653197766,
+      "loss": 1.7327,
+      "step": 304
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.16025269031524658,
+      "learning_rate": 0.00016603825651454588,
+      "loss": 1.7496,
+      "step": 305
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1708802878856659,
+      "learning_rate": 0.0001658217489196792,
+      "loss": 1.6233,
+      "step": 306
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.17108985781669617,
+      "learning_rate": 0.00016560469554262,
+      "loss": 1.675,
+      "step": 307
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.17331109941005707,
+      "learning_rate": 0.00016538709818313604,
+      "loss": 1.6796,
+      "step": 308
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.15554888546466827,
+      "learning_rate": 0.0001651689586455059,
+      "loss": 1.6977,
+      "step": 309
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.16161341965198517,
+      "learning_rate": 0.0001649502787385036,
+      "loss": 1.5421,
+      "step": 310
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.16354724764823914,
+      "learning_rate": 0.00016473106027538393,
+      "loss": 1.6105,
+      "step": 311
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.16034872829914093,
+      "learning_rate": 0.0001645113050738673,
+      "loss": 1.7455,
+      "step": 312
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.15332373976707458,
+      "learning_rate": 0.00016429101495612453,
+      "loss": 1.5715,
+      "step": 313
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.18118220567703247,
+      "learning_rate": 0.00016407019174876196,
+      "loss": 1.6783,
+      "step": 314
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.16754361987113953,
+      "learning_rate": 0.00016384883728280626,
+      "loss": 1.596,
+      "step": 315
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.1616910696029663,
+      "learning_rate": 0.00016362695339368913,
+      "loss": 1.699,
+      "step": 316
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.1674581617116928,
+      "learning_rate": 0.00016340454192123217,
+      "loss": 1.6238,
+      "step": 317
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.1575167030096054,
+      "learning_rate": 0.00016318160470963158,
+      "loss": 1.7563,
+      "step": 318
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.15855339169502258,
+      "learning_rate": 0.00016295814360744296,
+      "loss": 1.6587,
+      "step": 319
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.16675756871700287,
+      "learning_rate": 0.00016273416046756585,
+      "loss": 1.5694,
+      "step": 320
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.17339494824409485,
+      "learning_rate": 0.0001625096571472285,
+      "loss": 1.6924,
+      "step": 321
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.16791300475597382,
+      "learning_rate": 0.00016228463550797234,
+      "loss": 1.6103,
+      "step": 322
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.1806572824716568,
+      "learning_rate": 0.00016205909741563663,
+      "loss": 1.7016,
+      "step": 323
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.18338143825531006,
+      "learning_rate": 0.00016183304474034303,
+      "loss": 1.6367,
+      "step": 324
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.17175601422786713,
+      "learning_rate": 0.00016160647935647987,
+      "loss": 1.6221,
+      "step": 325
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1569746434688568,
+      "learning_rate": 0.00016137940314268695,
+      "loss": 1.6587,
+      "step": 326
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.16621094942092896,
+      "learning_rate": 0.00016115181798183968,
+      "loss": 1.7476,
+      "step": 327
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1589597761631012,
+      "learning_rate": 0.00016092372576103362,
+      "loss": 1.608,
+      "step": 328
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.1526154726743698,
+      "learning_rate": 0.00016069512837156867,
+      "loss": 1.6524,
+      "step": 329
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.16909892857074738,
+      "learning_rate": 0.00016046602770893365,
+      "loss": 1.7135,
+      "step": 330
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.16316530108451843,
+      "learning_rate": 0.00016023642567279033,
+      "loss": 1.645,
+      "step": 331
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.15762114524841309,
+      "learning_rate": 0.00016000632416695782,
+      "loss": 1.6342,
+      "step": 332
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.15512891113758087,
+      "learning_rate": 0.0001597757250993967,
+      "loss": 1.6937,
+      "step": 333
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.1490001082420349,
+      "learning_rate": 0.0001595446303821933,
+      "loss": 1.6558,
+      "step": 334
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.167209193110466,
+      "learning_rate": 0.00015931304193154375,
+      "loss": 1.7354,
+      "step": 335
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.15756173431873322,
+      "learning_rate": 0.00015908096166773817,
+      "loss": 1.6624,
+      "step": 336
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.35730743408203125,
+      "learning_rate": 0.0001588483915151447,
+      "loss": 1.7502,
+      "step": 337
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.17207658290863037,
+      "learning_rate": 0.00015861533340219347,
+      "loss": 1.7045,
+      "step": 338
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.1688750982284546,
+      "learning_rate": 0.00015838178926136078,
+      "loss": 1.7443,
+      "step": 339
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.15904007852077484,
+      "learning_rate": 0.00015814776102915295,
+      "loss": 1.6744,
+      "step": 340
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.15581969916820526,
+      "learning_rate": 0.0001579132506460903,
+      "loss": 1.6926,
+      "step": 341
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1648828089237213,
+      "learning_rate": 0.0001576782600566911,
+      "loss": 1.6929,
+      "step": 342
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.16498371958732605,
+      "learning_rate": 0.00015744279120945534,
+      "loss": 1.7607,
+      "step": 343
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.1721397489309311,
+      "learning_rate": 0.00015720684605684864,
+      "loss": 1.7413,
+      "step": 344
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.20110780000686646,
+      "learning_rate": 0.00015697042655528617,
+      "loss": 1.7579,
+      "step": 345
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.17458587884902954,
+      "learning_rate": 0.00015673353466511618,
+      "loss": 1.557,
+      "step": 346
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.1674317717552185,
+      "learning_rate": 0.00015649617235060395,
+      "loss": 1.7079,
+      "step": 347
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.17642170190811157,
+      "learning_rate": 0.0001562583415799154,
+      "loss": 1.7137,
+      "step": 348
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.17334245145320892,
+      "learning_rate": 0.00015602004432510084,
+      "loss": 1.6618,
+      "step": 349
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.19081345200538635,
+      "learning_rate": 0.00015578128256207857,
+      "loss": 1.6069,
+      "step": 350
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.1641138792037964,
+      "learning_rate": 0.00015554205827061855,
+      "loss": 1.7307,
+      "step": 351
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.18359099328517914,
+      "learning_rate": 0.0001553023734343258,
+      "loss": 1.6646,
+      "step": 352
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.15749670565128326,
+      "learning_rate": 0.00015506223004062427,
+      "loss": 1.6696,
+      "step": 353
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.15676377713680267,
+      "learning_rate": 0.00015482163008074016,
+      "loss": 1.6257,
+      "step": 354
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.1613236665725708,
+      "learning_rate": 0.00015458057554968533,
+      "loss": 1.6906,
+      "step": 355
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.15893089771270752,
+      "learning_rate": 0.0001543390684462409,
+      "loss": 1.7515,
+      "step": 356
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.16093188524246216,
+      "learning_rate": 0.00015409711077294066,
+      "loss": 1.7285,
+      "step": 357
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.1732104867696762,
+      "learning_rate": 0.00015385470453605456,
+      "loss": 1.728,
+      "step": 358
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.18478652834892273,
+      "learning_rate": 0.0001536118517455717,
+      "loss": 1.6798,
+      "step": 359
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.1561679095029831,
+      "learning_rate": 0.00015336855441518414,
+      "loss": 1.4988,
+      "step": 360
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.16175530850887299,
+      "learning_rate": 0.00015312481456226986,
+      "loss": 1.5661,
+      "step": 361
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.15828688442707062,
+      "learning_rate": 0.0001528806342078763,
+      "loss": 1.6894,
+      "step": 362
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.1670212745666504,
+      "learning_rate": 0.00015263601537670332,
+      "loss": 1.6946,
+      "step": 363
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.15261001884937286,
+      "learning_rate": 0.00015239096009708665,
+      "loss": 1.647,
+      "step": 364
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.15754546225070953,
+      "learning_rate": 0.00015214547040098082,
+      "loss": 1.65,
+      "step": 365
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.15401124954223633,
+      "learning_rate": 0.00015189954832394266,
+      "loss": 1.6826,
+      "step": 366
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.15601174533367157,
+      "learning_rate": 0.00015165319590511412,
+      "loss": 1.7028,
+      "step": 367
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.17138823866844177,
+      "learning_rate": 0.00015140641518720545,
+      "loss": 1.5761,
+      "step": 368
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 1.637104868888855,
+      "eval_runtime": 53.887,
+      "eval_samples_per_second": 92.731,
+      "eval_steps_per_second": 23.197,
+      "step": 368
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.16486236453056335,
+      "learning_rate": 0.00015115920821647834,
+      "loss": 1.6967,
+      "step": 369
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.16460780799388885,
+      "learning_rate": 0.00015091157704272886,
+      "loss": 1.6389,
+      "step": 370
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.16463211178779602,
+      "learning_rate": 0.00015066352371927047,
+      "loss": 1.7642,
+      "step": 371
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.15675517916679382,
+      "learning_rate": 0.0001504150503029171,
+      "loss": 1.6816,
+      "step": 372
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.1745896190404892,
+      "learning_rate": 0.0001501661588539659,
+      "loss": 1.7393,
+      "step": 373
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.16300994157791138,
+      "learning_rate": 0.00014991685143618042,
+      "loss": 1.6244,
+      "step": 374
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.17243793606758118,
+      "learning_rate": 0.00014966713011677318,
+      "loss": 1.664,
+      "step": 375
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.1658598929643631,
+      "learning_rate": 0.00014941699696638887,
+      "loss": 1.5184,
+      "step": 376
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.16609761118888855,
+      "learning_rate": 0.00014916645405908683,
+      "loss": 1.6516,
+      "step": 377
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.17976997792720795,
+      "learning_rate": 0.0001489155034723242,
+      "loss": 1.619,
+      "step": 378
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.1680922657251358,
+      "learning_rate": 0.00014866414728693838,
+      "loss": 1.6772,
+      "step": 379
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.18461324274539948,
+      "learning_rate": 0.00014841238758713006,
+      "loss": 1.5051,
+      "step": 380
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.20495639741420746,
+      "learning_rate": 0.0001481602264604457,
+      "loss": 1.6497,
+      "step": 381
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.16210030019283295,
+      "learning_rate": 0.00014790766599776038,
+      "loss": 1.376,
+      "step": 382
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.16862326860427856,
+      "learning_rate": 0.00014765470829326033,
+      "loss": 1.5394,
+      "step": 383
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.16202732920646667,
+      "learning_rate": 0.0001474013554444257,
+      "loss": 1.6899,
+      "step": 384
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.1672043651342392,
+      "learning_rate": 0.00014714760955201306,
+      "loss": 1.6943,
+      "step": 385
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.1694658100605011,
+      "learning_rate": 0.00014689347272003813,
+      "loss": 1.6172,
+      "step": 386
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.2547871470451355,
+      "learning_rate": 0.0001466389470557581,
+      "loss": 1.6476,
+      "step": 387
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.16380763053894043,
+      "learning_rate": 0.0001463840346696543,
+      "loss": 1.6212,
+      "step": 388
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.16450123488903046,
+      "learning_rate": 0.00014612873767541485,
+      "loss": 1.7284,
+      "step": 389
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.18062160909175873,
+      "learning_rate": 0.00014587305818991673,
+      "loss": 1.5628,
+      "step": 390
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.16628506779670715,
+      "learning_rate": 0.0001456169983332087,
+      "loss": 1.5655,
+      "step": 391
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.17104601860046387,
+      "learning_rate": 0.00014536056022849327,
+      "loss": 1.6204,
+      "step": 392
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.17847618460655212,
+      "learning_rate": 0.00014510374600210953,
+      "loss": 1.626,
+      "step": 393
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.17856758832931519,
+      "learning_rate": 0.0001448465577835152,
+      "loss": 1.6905,
+      "step": 394
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.15889039635658264,
+      "learning_rate": 0.000144588997705269,
+      "loss": 1.5162,
+      "step": 395
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.17019660770893097,
+      "learning_rate": 0.0001443310679030132,
+      "loss": 1.6263,
+      "step": 396
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.1884152591228485,
+      "learning_rate": 0.00014407277051545564,
+      "loss": 1.5887,
+      "step": 397
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.16880416870117188,
+      "learning_rate": 0.00014381410768435216,
+      "loss": 1.6554,
+      "step": 398
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.172772616147995,
+      "learning_rate": 0.0001435550815544888,
+      "loss": 1.6194,
+      "step": 399
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.1651185303926468,
+      "learning_rate": 0.00014329569427366394,
+      "loss": 1.6202,
+      "step": 400
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.1783447414636612,
+      "learning_rate": 0.00014303594799267065,
+      "loss": 1.5183,
+      "step": 401
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.18523593246936798,
+      "learning_rate": 0.00014277584486527872,
+      "loss": 1.5517,
+      "step": 402
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.17499391734600067,
+      "learning_rate": 0.0001425153870482169,
+      "loss": 1.6889,
+      "step": 403
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.16847828030586243,
+      "learning_rate": 0.00014225457670115482,
+      "loss": 1.6499,
+      "step": 404
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.187130406498909,
+      "learning_rate": 0.00014199341598668537,
+      "loss": 1.6138,
+      "step": 405
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.17778280377388,
+      "learning_rate": 0.0001417319070703066,
+      "loss": 1.5702,
+      "step": 406
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.18552978336811066,
+      "learning_rate": 0.00014147005212040374,
+      "loss": 1.4837,
+      "step": 407
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.19115741550922394,
+      "learning_rate": 0.00014120785330823128,
+      "loss": 1.5809,
+      "step": 408
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.19195811450481415,
+      "learning_rate": 0.000140945312807895,
+      "loss": 1.6841,
+      "step": 409
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.1775340586900711,
+      "learning_rate": 0.0001406824327963338,
+      "loss": 1.4879,
+      "step": 410
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.17470397055149078,
+      "learning_rate": 0.00014041921545330193,
+      "loss": 1.7223,
+      "step": 411
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.18359576165676117,
+      "learning_rate": 0.00014015566296135047,
+      "loss": 1.6814,
+      "step": 412
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.18946854770183563,
+      "learning_rate": 0.00013989177750580974,
+      "loss": 1.5659,
+      "step": 413
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.17491303384304047,
+      "learning_rate": 0.00013962756127477077,
+      "loss": 1.5633,
+      "step": 414
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.19263826310634613,
+      "learning_rate": 0.00013936301645906744,
+      "loss": 1.4806,
+      "step": 415
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.20540061593055725,
+      "learning_rate": 0.0001390981452522581,
+      "loss": 1.6358,
+      "step": 416
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.1703554391860962,
+      "learning_rate": 0.00013883294985060754,
+      "loss": 1.5266,
+      "step": 417
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.1858687400817871,
+      "learning_rate": 0.00013856743245306867,
+      "loss": 1.5421,
+      "step": 418
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.19925490021705627,
+      "learning_rate": 0.0001383015952612644,
+      "loss": 1.5334,
+      "step": 419
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.18249362707138062,
+      "learning_rate": 0.00013803544047946917,
+      "loss": 1.6678,
+      "step": 420
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.1895124316215515,
+      "learning_rate": 0.00013776897031459104,
+      "loss": 1.5825,
+      "step": 421
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.18998436629772186,
+      "learning_rate": 0.00013750218697615297,
+      "loss": 1.4063,
+      "step": 422
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.18359903991222382,
+      "learning_rate": 0.00013723509267627479,
+      "loss": 1.5474,
+      "step": 423
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.1771814525127411,
+      "learning_rate": 0.0001369676896296548,
+      "loss": 1.6217,
+      "step": 424
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.2542651891708374,
+      "learning_rate": 0.0001366999800535513,
+      "loss": 1.5389,
+      "step": 425
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.16922856867313385,
+      "learning_rate": 0.00013643196616776432,
+      "loss": 1.6299,
+      "step": 426
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.19103561341762543,
+      "learning_rate": 0.00013616365019461716,
+      "loss": 1.6539,
+      "step": 427
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.17129819095134735,
+      "learning_rate": 0.00013589503435893792,
+      "loss": 1.5554,
+      "step": 428
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.17848622798919678,
+      "learning_rate": 0.00013562612088804127,
+      "loss": 1.6609,
+      "step": 429
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.1793471872806549,
+      "learning_rate": 0.00013535691201170958,
+      "loss": 1.4906,
+      "step": 430
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.17249123752117157,
+      "learning_rate": 0.00013508740996217493,
+      "loss": 1.5901,
+      "step": 431
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.17167752981185913,
+      "learning_rate": 0.00013481761697410009,
+      "loss": 1.5892,
+      "step": 432
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.18118228018283844,
+      "learning_rate": 0.00013454753528456038,
+      "loss": 1.6449,
+      "step": 433
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.19412502646446228,
+      "learning_rate": 0.00013427716713302506,
+      "loss": 1.4823,
+      "step": 434
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.16546295583248138,
+      "learning_rate": 0.00013400651476133844,
+      "loss": 1.5447,
+      "step": 435
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.17301104962825775,
+      "learning_rate": 0.00013373558041370178,
+      "loss": 1.5071,
+      "step": 436
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.19761618971824646,
+      "learning_rate": 0.00013346436633665428,
+      "loss": 1.6934,
+      "step": 437
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.17933383584022522,
+      "learning_rate": 0.00013319287477905464,
+      "loss": 1.514,
+      "step": 438
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.20005792379379272,
+      "learning_rate": 0.00013292110799206243,
+      "loss": 1.4735,
+      "step": 439
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.1816921830177307,
+      "learning_rate": 0.0001326490682291193,
+      "loss": 1.6867,
+      "step": 440
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.19630655646324158,
+      "learning_rate": 0.00013237675774593045,
+      "loss": 1.3871,
+      "step": 441
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.17641153931617737,
+      "learning_rate": 0.0001321041788004458,
+      "loss": 1.5511,
+      "step": 442
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.18330830335617065,
+      "learning_rate": 0.00013183133365284123,
+      "loss": 1.432,
+      "step": 443
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.20252883434295654,
+      "learning_rate": 0.00013155822456550006,
+      "loss": 1.6277,
+      "step": 444
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.17543216049671173,
+      "learning_rate": 0.00013128485380299408,
+      "loss": 1.6337,
+      "step": 445
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.1888299435377121,
+      "learning_rate": 0.00013101122363206488,
+      "loss": 1.6009,
+      "step": 446
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.20228877663612366,
+      "learning_rate": 0.00013073733632160494,
+      "loss": 1.5939,
+      "step": 447
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.19537915289402008,
+      "learning_rate": 0.00013046319414263902,
+      "loss": 1.5944,
+      "step": 448
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.18867214024066925,
+      "learning_rate": 0.00013018879936830505,
+      "loss": 1.5614,
+      "step": 449
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.18890655040740967,
+      "learning_rate": 0.00012991415427383556,
+      "loss": 1.6572,
+      "step": 450
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.19154617190361023,
+      "learning_rate": 0.00012963926113653863,
+      "loss": 1.4916,
+      "step": 451
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.18493875861167908,
+      "learning_rate": 0.00012936412223577915,
+      "loss": 1.6289,
+      "step": 452
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.18440605700016022,
+      "learning_rate": 0.00012908873985295968,
+      "loss": 1.5943,
+      "step": 453
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.1951192021369934,
+      "learning_rate": 0.00012881311627150187,
+      "loss": 1.6152,
+      "step": 454
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.18237607181072235,
+      "learning_rate": 0.00012853725377682718,
+      "loss": 1.7562,
+      "step": 455
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.17004793882369995,
+      "learning_rate": 0.0001282611546563382,
+      "loss": 1.5222,
+      "step": 456
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.19362376630306244,
+      "learning_rate": 0.00012798482119939956,
+      "loss": 1.5306,
+      "step": 457
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.1877073347568512,
+      "learning_rate": 0.00012770825569731895,
+      "loss": 1.4202,
+      "step": 458
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.177567720413208,
+      "learning_rate": 0.00012743146044332817,
+      "loss": 1.5566,
+      "step": 459
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.20578047633171082,
+      "learning_rate": 0.00012715443773256402,
+      "loss": 1.4785,
+      "step": 460
+    },
+    {
+      "epoch": 1.23,
+      "eval_loss": 1.6219624280929565,
+      "eval_runtime": 53.6944,
+      "eval_samples_per_second": 93.064,
+      "eval_steps_per_second": 23.28,
+      "step": 460
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.180220827460289,
+      "learning_rate": 0.0001268771898620494,
+      "loss": 1.6456,
+      "step": 461
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.20168617367744446,
+      "learning_rate": 0.00012659971913067414,
+      "loss": 1.4979,
+      "step": 462
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.1822381317615509,
+      "learning_rate": 0.000126322027839176,
+      "loss": 1.5769,
+      "step": 463
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.18748806416988373,
+      "learning_rate": 0.00012604411829012166,
+      "loss": 1.6408,
+      "step": 464
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.17949888110160828,
+      "learning_rate": 0.00012576599278788742,
+      "loss": 1.6341,
+      "step": 465
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.19509534537792206,
+      "learning_rate": 0.00012548765363864036,
+      "loss": 1.5458,
+      "step": 466
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.1767253279685974,
+      "learning_rate": 0.00012520910315031895,
+      "loss": 1.577,
+      "step": 467
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.17574870586395264,
+      "learning_rate": 0.0001249303436326142,
+      "loss": 1.6089,
+      "step": 468
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.17545011639595032,
+      "learning_rate": 0.0001246513773969502,
+      "loss": 1.6331,
+      "step": 469
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.18224777281284332,
+      "learning_rate": 0.00012437220675646523,
+      "loss": 1.5737,
+      "step": 470
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.20681160688400269,
+      "learning_rate": 0.00012409283402599238,
+      "loss": 1.5503,
+      "step": 471
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.196259543299675,
+      "learning_rate": 0.0001238132615220405,
+      "loss": 1.6594,
+      "step": 472
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.18110986053943634,
+      "learning_rate": 0.0001235334915627748,
+      "loss": 1.6193,
+      "step": 473
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.18576951324939728,
+      "learning_rate": 0.00012325352646799795,
+      "loss": 1.6723,
+      "step": 474
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.18934178352355957,
+      "learning_rate": 0.00012297336855913047,
+      "loss": 1.6192,
+      "step": 475
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.185868501663208,
+      "learning_rate": 0.00012269302015919172,
+      "loss": 1.6163,
+      "step": 476
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.16594283282756805,
+      "learning_rate": 0.00012241248359278064,
+      "loss": 1.6334,
+      "step": 477
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.17133650183677673,
+      "learning_rate": 0.00012213176118605637,
+      "loss": 1.5889,
+      "step": 478
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.1712963730096817,
+      "learning_rate": 0.00012185085526671893,
+      "loss": 1.5003,
+      "step": 479
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.1756075769662857,
+      "learning_rate": 0.00012156976816399013,
+      "loss": 1.6512,
+      "step": 480
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.17380216717720032,
+      "learning_rate": 0.00012128850220859397,
+      "loss": 1.4197,
+      "step": 481
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.18200692534446716,
+      "learning_rate": 0.00012100705973273765,
+      "loss": 1.5587,
+      "step": 482
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.17649267613887787,
+      "learning_rate": 0.00012072544307009184,
+      "loss": 1.662,
+      "step": 483
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.20866695046424866,
+      "learning_rate": 0.00012044365455577171,
+      "loss": 1.5038,
+      "step": 484
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.18661659955978394,
+      "learning_rate": 0.00012016169652631726,
+      "loss": 1.4872,
+      "step": 485
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.19411514699459076,
+      "learning_rate": 0.00011987957131967418,
+      "loss": 1.5808,
+      "step": 486
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.1833687424659729,
+      "learning_rate": 0.00011959728127517434,
+      "loss": 1.6578,
+      "step": 487
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.2166387289762497,
+      "learning_rate": 0.00011931482873351632,
+      "loss": 1.6438,
+      "step": 488
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.17330534756183624,
+      "learning_rate": 0.00011903221603674628,
+      "loss": 1.5918,
+      "step": 489
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.17018438875675201,
+      "learning_rate": 0.00011874944552823817,
+      "loss": 1.5898,
+      "step": 490
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.1763826310634613,
+      "learning_rate": 0.00011846651955267463,
+      "loss": 1.5376,
+      "step": 491
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.18420416116714478,
+      "learning_rate": 0.00011818344045602727,
+      "loss": 1.5467,
+      "step": 492
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.19673533737659454,
+      "learning_rate": 0.00011790021058553751,
+      "loss": 1.582,
+      "step": 493
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.1835012137889862,
+      "learning_rate": 0.00011761683228969682,
+      "loss": 1.6558,
+      "step": 494
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.18565072119235992,
+      "learning_rate": 0.00011733330791822749,
+      "loss": 1.5635,
+      "step": 495
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.18372558057308197,
+      "learning_rate": 0.00011704963982206299,
+      "loss": 1.5643,
+      "step": 496
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.17450691759586334,
+      "learning_rate": 0.00011676583035332853,
+      "loss": 1.5644,
+      "step": 497
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.1876295804977417,
+      "learning_rate": 0.00011648188186532154,
+      "loss": 1.5997,
+      "step": 498
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.19220595061779022,
+      "learning_rate": 0.00011619779671249223,
+      "loss": 1.6757,
+      "step": 499
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.1927752047777176,
+      "learning_rate": 0.00011591357725042393,
+      "loss": 1.5843,
+      "step": 500
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.18117199838161469,
+      "learning_rate": 0.00011562922583581375,
+      "loss": 1.6051,
+      "step": 501
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.18213218450546265,
+      "learning_rate": 0.0001153447448264528,
+      "loss": 1.5653,
+      "step": 502
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.18642397224903107,
+      "learning_rate": 0.00011506013658120687,
+      "loss": 1.5057,
+      "step": 503
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.18502803146839142,
+      "learning_rate": 0.00011477540345999669,
+      "loss": 1.5657,
+      "step": 504
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.1808471977710724,
+      "learning_rate": 0.00011449054782377855,
+      "loss": 1.6401,
+      "step": 505
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.18637515604496002,
+      "learning_rate": 0.00011420557203452444,
+      "loss": 1.5472,
+      "step": 506
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.19144316017627716,
+      "learning_rate": 0.00011392047845520282,
+      "loss": 1.5232,
+      "step": 507
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.17375171184539795,
+      "learning_rate": 0.00011363526944975867,
+      "loss": 1.5129,
+      "step": 508
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.18536758422851562,
+      "learning_rate": 0.00011334994738309423,
+      "loss": 1.5233,
+      "step": 509
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.1873825639486313,
+      "learning_rate": 0.00011306451462104912,
+      "loss": 1.6106,
+      "step": 510
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.18517866730690002,
+      "learning_rate": 0.00011277897353038085,
+      "loss": 1.5758,
+      "step": 511
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.18197666108608246,
+      "learning_rate": 0.00011249332647874513,
+      "loss": 1.5425,
+      "step": 512
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.17675837874412537,
+      "learning_rate": 0.00011220757583467643,
+      "loss": 1.548,
+      "step": 513
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.18274517357349396,
+      "learning_rate": 0.00011192172396756797,
+      "loss": 1.5736,
+      "step": 514
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.24651746451854706,
+      "learning_rate": 0.00011163577324765248,
+      "loss": 1.5953,
+      "step": 515
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.18200458586215973,
+      "learning_rate": 0.00011134972604598224,
+      "loss": 1.5957,
+      "step": 516
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.19455678761005402,
+      "learning_rate": 0.00011106358473440963,
+      "loss": 1.659,
+      "step": 517
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.19950920343399048,
+      "learning_rate": 0.00011077735168556729,
+      "loss": 1.5277,
+      "step": 518
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.19501787424087524,
+      "learning_rate": 0.00011049102927284857,
+      "loss": 1.612,
+      "step": 519
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.1821345090866089,
+      "learning_rate": 0.00011020461987038781,
+      "loss": 1.5435,
+      "step": 520
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.17476962506771088,
+      "learning_rate": 0.00010991812585304069,
+      "loss": 1.5899,
+      "step": 521
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.1734679490327835,
+      "learning_rate": 0.00010963154959636438,
+      "loss": 1.675,
+      "step": 522
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.19216276705265045,
+      "learning_rate": 0.00010934489347659816,
+      "loss": 1.5913,
+      "step": 523
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.18504251539707184,
+      "learning_rate": 0.00010905815987064328,
+      "loss": 1.5604,
+      "step": 524
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.1940964013338089,
+      "learning_rate": 0.00010877135115604372,
+      "loss": 1.5155,
+      "step": 525
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.19056075811386108,
+      "learning_rate": 0.00010848446971096606,
+      "loss": 1.5704,
+      "step": 526
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.18702279031276703,
+      "learning_rate": 0.00010819751791418011,
+      "loss": 1.5397,
+      "step": 527
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.1916041225194931,
+      "learning_rate": 0.00010791049814503888,
+      "loss": 1.5917,
+      "step": 528
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.18013226985931396,
+      "learning_rate": 0.00010762341278345909,
+      "loss": 1.5779,
+      "step": 529
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.1704178899526596,
+      "learning_rate": 0.00010733626420990134,
+      "loss": 1.5725,
+      "step": 530
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.19062383472919464,
+      "learning_rate": 0.0001070490548053503,
+      "loss": 1.6347,
+      "step": 531
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.18978549540042877,
+      "learning_rate": 0.00010676178695129513,
+      "loss": 1.5762,
+      "step": 532
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.18251344561576843,
+      "learning_rate": 0.00010647446302970954,
+      "loss": 1.5688,
+      "step": 533
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.17656119167804718,
+      "learning_rate": 0.00010618708542303226,
+      "loss": 1.5179,
+      "step": 534
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.19226515293121338,
+      "learning_rate": 0.00010589965651414704,
+      "loss": 1.6262,
+      "step": 535
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.19419853389263153,
+      "learning_rate": 0.00010561217868636313,
+      "loss": 1.5603,
+      "step": 536
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.1746627241373062,
+      "learning_rate": 0.00010532465432339538,
+      "loss": 1.5841,
+      "step": 537
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.17498454451560974,
+      "learning_rate": 0.00010503708580934442,
+      "loss": 1.6575,
+      "step": 538
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.17727667093276978,
+      "learning_rate": 0.00010474947552867706,
+      "loss": 1.5722,
+      "step": 539
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.19356906414031982,
+      "learning_rate": 0.00010446182586620648,
+      "loss": 1.5017,
+      "step": 540
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.18533338606357574,
+      "learning_rate": 0.00010417413920707222,
+      "loss": 1.5829,
+      "step": 541
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.18073119223117828,
+      "learning_rate": 0.00010388641793672078,
+      "loss": 1.6755,
+      "step": 542
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.1906929314136505,
+      "learning_rate": 0.00010359866444088555,
+      "loss": 1.5658,
+      "step": 543
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.186784565448761,
+      "learning_rate": 0.00010331088110556717,
+      "loss": 1.5248,
+      "step": 544
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.20720143616199493,
+      "learning_rate": 0.00010302307031701364,
+      "loss": 1.628,
+      "step": 545
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.16866104304790497,
+      "learning_rate": 0.0001027352344617007,
+      "loss": 1.5764,
+      "step": 546
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.19025076925754547,
+      "learning_rate": 0.00010244737592631181,
+      "loss": 1.7834,
+      "step": 547
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.19599542021751404,
+      "learning_rate": 0.00010215949709771866,
+      "loss": 1.5223,
+      "step": 548
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.2109784334897995,
+      "learning_rate": 0.00010187160036296103,
+      "loss": 1.5431,
+      "step": 549
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.2028791755437851,
+      "learning_rate": 0.00010158368810922729,
+      "loss": 1.5948,
+      "step": 550
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.20557260513305664,
+      "learning_rate": 0.00010129576272383445,
+      "loss": 1.6074,
+      "step": 551
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.20252369344234467,
+      "learning_rate": 0.0001010078265942084,
+      "loss": 1.4492,
+      "step": 552
+    },
+    {
+      "epoch": 1.49,
+      "eval_loss": 1.6022814512252808,
+      "eval_runtime": 54.0178,
+      "eval_samples_per_second": 92.507,
+      "eval_steps_per_second": 23.141,
+      "step": 552
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.18352779746055603,
+      "learning_rate": 0.00010071988210786411,
+      "loss": 1.5881,
+      "step": 553
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.2220783829689026,
+      "learning_rate": 0.00010043193165238591,
+      "loss": 1.4926,
+      "step": 554
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.19979965686798096,
+      "learning_rate": 0.00010014397761540755,
+      "loss": 1.6466,
+      "step": 555
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.17908692359924316,
+      "learning_rate": 9.985602238459247e-05,
+      "loss": 1.6098,
+      "step": 556
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.19138608872890472,
+      "learning_rate": 9.956806834761411e-05,
+      "loss": 1.6106,
+      "step": 557
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.19390954077243805,
+      "learning_rate": 9.92801178921359e-05,
+      "loss": 1.6256,
+      "step": 558
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.17675888538360596,
+      "learning_rate": 9.899217340579164e-05,
+      "loss": 1.558,
+      "step": 559
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.18734760582447052,
+      "learning_rate": 9.870423727616558e-05,
+      "loss": 1.5597,
+      "step": 560
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.2040862888097763,
+      "learning_rate": 9.841631189077269e-05,
+      "loss": 1.5835,
+      "step": 561
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.18791481852531433,
+      "learning_rate": 9.812839963703899e-05,
+      "loss": 1.5414,
+      "step": 562
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.219825878739357,
+      "learning_rate": 9.784050290228134e-05,
+      "loss": 1.3138,
+      "step": 563
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.19013109803199768,
+      "learning_rate": 9.75526240736882e-05,
+      "loss": 1.4212,
+      "step": 564
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.17495863139629364,
+      "learning_rate": 9.726476553829932e-05,
+      "loss": 1.535,
+      "step": 565
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.19723224639892578,
+      "learning_rate": 9.69769296829864e-05,
+      "loss": 1.5841,
+      "step": 566
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.17984871566295624,
+      "learning_rate": 9.668911889443285e-05,
+      "loss": 1.5363,
+      "step": 567
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.23778770864009857,
+      "learning_rate": 9.640133555911449e-05,
+      "loss": 1.4199,
+      "step": 568
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.1896432489156723,
+      "learning_rate": 9.611358206327923e-05,
+      "loss": 1.4074,
+      "step": 569
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.18232782185077667,
+      "learning_rate": 9.582586079292778e-05,
+      "loss": 1.5881,
+      "step": 570
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.19662512838840485,
+      "learning_rate": 9.553817413379356e-05,
+      "loss": 1.563,
+      "step": 571
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.2098013013601303,
+      "learning_rate": 9.525052447132294e-05,
+      "loss": 1.5127,
+      "step": 572
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.19347095489501953,
+      "learning_rate": 9.496291419065561e-05,
+      "loss": 1.6533,
+      "step": 573
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.18512225151062012,
+      "learning_rate": 9.467534567660466e-05,
+      "loss": 1.5451,
+      "step": 574
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.21681411564350128,
+      "learning_rate": 9.43878213136369e-05,
+      "loss": 1.4848,
+      "step": 575
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.17953048646450043,
+      "learning_rate": 9.410034348585298e-05,
+      "loss": 1.6384,
+      "step": 576
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.20537644624710083,
+      "learning_rate": 9.381291457696779e-05,
+      "loss": 1.6757,
+      "step": 577
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.19091413915157318,
+      "learning_rate": 9.352553697029048e-05,
+      "loss": 1.6759,
+      "step": 578
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.19432410597801208,
+      "learning_rate": 9.323821304870489e-05,
+      "loss": 1.6414,
+      "step": 579
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.21516050398349762,
+      "learning_rate": 9.295094519464972e-05,
+      "loss": 1.468,
+      "step": 580
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.18204712867736816,
+      "learning_rate": 9.266373579009867e-05,
+      "loss": 1.5371,
+      "step": 581
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.18101070821285248,
+      "learning_rate": 9.237658721654092e-05,
+      "loss": 1.5876,
+      "step": 582
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.1908547580242157,
+      "learning_rate": 9.208950185496114e-05,
+      "loss": 1.4907,
+      "step": 583
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.20174852013587952,
+      "learning_rate": 9.180248208581994e-05,
+      "loss": 1.5133,
+      "step": 584
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.20399607717990875,
+      "learning_rate": 9.151553028903396e-05,
+      "loss": 1.6886,
+      "step": 585
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.1815037727355957,
+      "learning_rate": 9.122864884395633e-05,
+      "loss": 1.5567,
+      "step": 586
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.19035212695598602,
+      "learning_rate": 9.094184012935674e-05,
+      "loss": 1.6067,
+      "step": 587
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.18987087905406952,
+      "learning_rate": 9.065510652340185e-05,
+      "loss": 1.6233,
+      "step": 588
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.19988678395748138,
+      "learning_rate": 9.036845040363562e-05,
+      "loss": 1.494,
+      "step": 589
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.18984133005142212,
+      "learning_rate": 9.008187414695932e-05,
+      "loss": 1.5683,
+      "step": 590
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.18878312408924103,
+      "learning_rate": 8.979538012961221e-05,
+      "loss": 1.6567,
+      "step": 591
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.1855538934469223,
+      "learning_rate": 8.950897072715144e-05,
+      "loss": 1.5954,
+      "step": 592
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.24248315393924713,
+      "learning_rate": 8.922264831443274e-05,
+      "loss": 1.5471,
+      "step": 593
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.19141511619091034,
+      "learning_rate": 8.89364152655904e-05,
+      "loss": 1.4948,
+      "step": 594
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.19037242233753204,
+      "learning_rate": 8.865027395401778e-05,
+      "loss": 1.4857,
+      "step": 595
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.19110530614852905,
+      "learning_rate": 8.836422675234754e-05,
+      "loss": 1.5597,
+      "step": 596
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.24663759768009186,
+      "learning_rate": 8.807827603243204e-05,
+      "loss": 1.4354,
+      "step": 597
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.19291803240776062,
+      "learning_rate": 8.779242416532361e-05,
+      "loss": 1.5331,
+      "step": 598
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.20057077705860138,
+      "learning_rate": 8.750667352125487e-05,
+      "loss": 1.4771,
+      "step": 599
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.18224692344665527,
+      "learning_rate": 8.722102646961919e-05,
+      "loss": 1.5867,
+      "step": 600
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.20015600323677063,
+      "learning_rate": 8.69354853789509e-05,
+      "loss": 1.5603,
+      "step": 601
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.19215072691440582,
+      "learning_rate": 8.665005261690579e-05,
+      "loss": 1.5803,
+      "step": 602
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.17882663011550903,
+      "learning_rate": 8.636473055024134e-05,
+      "loss": 1.6111,
+      "step": 603
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.20356720685958862,
+      "learning_rate": 8.607952154479723e-05,
+      "loss": 1.4877,
+      "step": 604
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.1832900196313858,
+      "learning_rate": 8.579442796547558e-05,
+      "loss": 1.5668,
+      "step": 605
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.20597009360790253,
+      "learning_rate": 8.550945217622146e-05,
+      "loss": 1.5496,
+      "step": 606
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.19200527667999268,
+      "learning_rate": 8.522459654000332e-05,
+      "loss": 1.6538,
+      "step": 607
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.19875632226467133,
+      "learning_rate": 8.493986341879314e-05,
+      "loss": 1.5669,
+      "step": 608
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.1814192533493042,
+      "learning_rate": 8.465525517354724e-05,
+      "loss": 1.5273,
+      "step": 609
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.19254544377326965,
+      "learning_rate": 8.437077416418627e-05,
+      "loss": 1.5282,
+      "step": 610
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.20155031979084015,
+      "learning_rate": 8.408642274957612e-05,
+      "loss": 1.5957,
+      "step": 611
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.19522641599178314,
+      "learning_rate": 8.380220328750781e-05,
+      "loss": 1.6277,
+      "step": 612
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.18992267549037933,
+      "learning_rate": 8.351811813467851e-05,
+      "loss": 1.5687,
+      "step": 613
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.18783116340637207,
+      "learning_rate": 8.32341696466715e-05,
+      "loss": 1.5739,
+      "step": 614
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.17621149122714996,
+      "learning_rate": 8.295036017793702e-05,
+      "loss": 1.5197,
+      "step": 615
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.2008402794599533,
+      "learning_rate": 8.266669208177252e-05,
+      "loss": 1.5905,
+      "step": 616
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.212905153632164,
+      "learning_rate": 8.238316771030318e-05,
+      "loss": 1.6655,
+      "step": 617
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.17801809310913086,
+      "learning_rate": 8.209978941446252e-05,
+      "loss": 1.6393,
+      "step": 618
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.1957562416791916,
+      "learning_rate": 8.181655954397275e-05,
+      "loss": 1.4709,
+      "step": 619
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.19428160786628723,
+      "learning_rate": 8.153348044732543e-05,
+      "loss": 1.4383,
+      "step": 620
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.20604220032691956,
+      "learning_rate": 8.125055447176186e-05,
+      "loss": 1.5547,
+      "step": 621
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.18895936012268066,
+      "learning_rate": 8.096778396325377e-05,
+      "loss": 1.5027,
+      "step": 622
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.1925646960735321,
+      "learning_rate": 8.068517126648369e-05,
+      "loss": 1.6032,
+      "step": 623
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.20500631630420685,
+      "learning_rate": 8.04027187248257e-05,
+      "loss": 1.62,
+      "step": 624
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.19827888906002045,
+      "learning_rate": 8.012042868032585e-05,
+      "loss": 1.5958,
+      "step": 625
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.18163664638996124,
+      "learning_rate": 7.983830347368276e-05,
+      "loss": 1.5118,
+      "step": 626
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.20141161978244781,
+      "learning_rate": 7.955634544422834e-05,
+      "loss": 1.54,
+      "step": 627
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.1766718327999115,
+      "learning_rate": 7.927455692990818e-05,
+      "loss": 1.6012,
+      "step": 628
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.1866627186536789,
+      "learning_rate": 7.899294026726241e-05,
+      "loss": 1.529,
+      "step": 629
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.19402055442333221,
+      "learning_rate": 7.871149779140604e-05,
+      "loss": 1.4608,
+      "step": 630
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.19119343161582947,
+      "learning_rate": 7.843023183600988e-05,
+      "loss": 1.5737,
+      "step": 631
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.2043522596359253,
+      "learning_rate": 7.81491447332811e-05,
+      "loss": 1.5267,
+      "step": 632
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.20135103166103363,
+      "learning_rate": 7.786823881394364e-05,
+      "loss": 1.5529,
+      "step": 633
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.20661364495754242,
+      "learning_rate": 7.758751640721937e-05,
+      "loss": 1.6185,
+      "step": 634
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.19659817218780518,
+      "learning_rate": 7.730697984080827e-05,
+      "loss": 1.5582,
+      "step": 635
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.1995358020067215,
+      "learning_rate": 7.702663144086957e-05,
+      "loss": 1.6023,
+      "step": 636
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.23160699009895325,
+      "learning_rate": 7.674647353200208e-05,
+      "loss": 1.4295,
+      "step": 637
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.19710291922092438,
+      "learning_rate": 7.64665084372252e-05,
+      "loss": 1.5295,
+      "step": 638
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.20388691127300262,
+      "learning_rate": 7.618673847795953e-05,
+      "loss": 1.4763,
+      "step": 639
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.196650430560112,
+      "learning_rate": 7.590716597400761e-05,
+      "loss": 1.5664,
+      "step": 640
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.19940368831157684,
+      "learning_rate": 7.562779324353477e-05,
+      "loss": 1.537,
+      "step": 641
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.18506959080696106,
+      "learning_rate": 7.53486226030498e-05,
+      "loss": 1.5549,
+      "step": 642
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.19527395069599152,
+      "learning_rate": 7.506965636738583e-05,
+      "loss": 1.5862,
+      "step": 643
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.19912207126617432,
+      "learning_rate": 7.479089684968106e-05,
+      "loss": 1.6224,
+      "step": 644
+    },
+    {
+      "epoch": 1.74,
+      "eval_loss": 1.588733434677124,
+      "eval_runtime": 53.8386,
+      "eval_samples_per_second": 92.814,
+      "eval_steps_per_second": 23.218,
+      "step": 644
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.17763713002204895,
+      "learning_rate": 7.451234636135969e-05,
+      "loss": 1.5622,
+      "step": 645
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.19509463012218475,
+      "learning_rate": 7.42340072121126e-05,
+      "loss": 1.5732,
+      "step": 646
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.19236071407794952,
+      "learning_rate": 7.395588170987839e-05,
+      "loss": 1.4118,
+      "step": 647
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.18488897383213043,
+      "learning_rate": 7.367797216082402e-05,
+      "loss": 1.5281,
+      "step": 648
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.19441011548042297,
+      "learning_rate": 7.340028086932587e-05,
+      "loss": 1.5507,
+      "step": 649
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.20152153074741364,
+      "learning_rate": 7.312281013795064e-05,
+      "loss": 1.5329,
+      "step": 650
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.20338155329227448,
+      "learning_rate": 7.284556226743598e-05,
+      "loss": 1.5696,
+      "step": 651
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.19269070029258728,
+      "learning_rate": 7.256853955667187e-05,
+      "loss": 1.6863,
+      "step": 652
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.18623311817646027,
+      "learning_rate": 7.229174430268104e-05,
+      "loss": 1.5835,
+      "step": 653
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.1871241182088852,
+      "learning_rate": 7.201517880060049e-05,
+      "loss": 1.6267,
+      "step": 654
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.18970423936843872,
+      "learning_rate": 7.173884534366182e-05,
+      "loss": 1.5929,
+      "step": 655
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.2575983703136444,
+      "learning_rate": 7.146274622317288e-05,
+      "loss": 1.5495,
+      "step": 656
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.17981579899787903,
+      "learning_rate": 7.118688372849815e-05,
+      "loss": 1.4926,
+      "step": 657
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.19219791889190674,
+      "learning_rate": 7.091126014704032e-05,
+      "loss": 1.5298,
+      "step": 658
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.2046176642179489,
+      "learning_rate": 7.063587776422088e-05,
+      "loss": 1.4501,
+      "step": 659
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.19982139766216278,
+      "learning_rate": 7.036073886346138e-05,
+      "loss": 1.5229,
+      "step": 660
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.18925295770168304,
+      "learning_rate": 7.008584572616448e-05,
+      "loss": 1.6139,
+      "step": 661
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.19633693993091583,
+      "learning_rate": 6.981120063169499e-05,
+      "loss": 1.7017,
+      "step": 662
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.21355824172496796,
+      "learning_rate": 6.953680585736105e-05,
+      "loss": 1.5256,
+      "step": 663
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.2261500209569931,
+      "learning_rate": 6.926266367839508e-05,
+      "loss": 1.423,
+      "step": 664
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.20399561524391174,
+      "learning_rate": 6.898877636793517e-05,
+      "loss": 1.5569,
+      "step": 665
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.17641954123973846,
+      "learning_rate": 6.871514619700594e-05,
+      "loss": 1.5215,
+      "step": 666
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.23922847211360931,
+      "learning_rate": 6.844177543449997e-05,
+      "loss": 1.365,
+      "step": 667
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.17896459996700287,
+      "learning_rate": 6.816866634715881e-05,
+      "loss": 1.6711,
+      "step": 668
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.1928153783082962,
+      "learning_rate": 6.789582119955424e-05,
+      "loss": 1.6874,
+      "step": 669
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.18445910513401031,
+      "learning_rate": 6.762324225406957e-05,
+      "loss": 1.6136,
+      "step": 670
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.19517773389816284,
+      "learning_rate": 6.73509317708807e-05,
+      "loss": 1.5981,
+      "step": 671
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.20011338591575623,
+      "learning_rate": 6.70788920079376e-05,
+      "loss": 1.6253,
+      "step": 672
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.19087383151054382,
+      "learning_rate": 6.680712522094537e-05,
+      "loss": 1.5274,
+      "step": 673
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.1847938448190689,
+      "learning_rate": 6.653563366334577e-05,
+      "loss": 1.4764,
+      "step": 674
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.1973254382610321,
+      "learning_rate": 6.626441958629826e-05,
+      "loss": 1.5333,
+      "step": 675
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.18505576252937317,
+      "learning_rate": 6.599348523866155e-05,
+      "loss": 1.4504,
+      "step": 676
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.2034735381603241,
+      "learning_rate": 6.572283286697496e-05,
+      "loss": 1.4913,
+      "step": 677
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.21279850602149963,
+      "learning_rate": 6.54524647154396e-05,
+      "loss": 1.566,
+      "step": 678
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.2010655552148819,
+      "learning_rate": 6.518238302589994e-05,
+      "loss": 1.5354,
+      "step": 679
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.19974732398986816,
+      "learning_rate": 6.491259003782511e-05,
+      "loss": 1.4899,
+      "step": 680
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.19252151250839233,
+      "learning_rate": 6.464308798829043e-05,
+      "loss": 1.6117,
+      "step": 681
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.18204721808433533,
+      "learning_rate": 6.437387911195875e-05,
+      "loss": 1.5365,
+      "step": 682
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.19013364613056183,
+      "learning_rate": 6.410496564106207e-05,
+      "loss": 1.6737,
+      "step": 683
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.19772353768348694,
+      "learning_rate": 6.383634980538286e-05,
+      "loss": 1.6396,
+      "step": 684
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.19503507018089294,
+      "learning_rate": 6.356803383223569e-05,
+      "loss": 1.6221,
+      "step": 685
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.18829859793186188,
+      "learning_rate": 6.33000199464487e-05,
+      "loss": 1.5017,
+      "step": 686
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.18767286837100983,
+      "learning_rate": 6.30323103703452e-05,
+      "loss": 1.6461,
+      "step": 687
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.19198834896087646,
+      "learning_rate": 6.276490732372522e-05,
+      "loss": 1.6037,
+      "step": 688
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.21538251638412476,
+      "learning_rate": 6.249781302384705e-05,
+      "loss": 1.5674,
+      "step": 689
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.20613306760787964,
+      "learning_rate": 6.2231029685409e-05,
+      "loss": 1.5887,
+      "step": 690
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.19516527652740479,
+      "learning_rate": 6.196455952053084e-05,
+      "loss": 1.5408,
+      "step": 691
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.19336941838264465,
+      "learning_rate": 6.169840473873565e-05,
+      "loss": 1.5313,
+      "step": 692
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.19075316190719604,
+      "learning_rate": 6.143256754693134e-05,
+      "loss": 1.6081,
+      "step": 693
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.1863541156053543,
+      "learning_rate": 6.116705014939246e-05,
+      "loss": 1.5864,
+      "step": 694
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.2056710124015808,
+      "learning_rate": 6.090185474774192e-05,
+      "loss": 1.5778,
+      "step": 695
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.17553465068340302,
+      "learning_rate": 6.063698354093255e-05,
+      "loss": 1.6269,
+      "step": 696
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.20064443349838257,
+      "learning_rate": 6.037243872522924e-05,
+      "loss": 1.5933,
+      "step": 697
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.19043272733688354,
+      "learning_rate": 6.010822249419027e-05,
+      "loss": 1.6484,
+      "step": 698
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.23049475252628326,
+      "learning_rate": 5.984433703864956e-05,
+      "loss": 1.4464,
+      "step": 699
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.17904847860336304,
+      "learning_rate": 5.95807845466981e-05,
+      "loss": 1.5606,
+      "step": 700
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.19818058609962463,
+      "learning_rate": 5.931756720366621e-05,
+      "loss": 1.5782,
+      "step": 701
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.24131308495998383,
+      "learning_rate": 5.9054687192105026e-05,
+      "loss": 1.4238,
+      "step": 702
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.20596317946910858,
+      "learning_rate": 5.8792146691768726e-05,
+      "loss": 1.5865,
+      "step": 703
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.19873526692390442,
+      "learning_rate": 5.852994787959628e-05,
+      "loss": 1.496,
+      "step": 704
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.2155202329158783,
+      "learning_rate": 5.8268092929693405e-05,
+      "loss": 1.4759,
+      "step": 705
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.20002122223377228,
+      "learning_rate": 5.800658401331467e-05,
+      "loss": 1.5385,
+      "step": 706
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.19707627594470978,
+      "learning_rate": 5.774542329884518e-05,
+      "loss": 1.6369,
+      "step": 707
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.2333543300628662,
+      "learning_rate": 5.748461295178315e-05,
+      "loss": 1.5858,
+      "step": 708
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.1858336627483368,
+      "learning_rate": 5.722415513472128e-05,
+      "loss": 1.5524,
+      "step": 709
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.1900344043970108,
+      "learning_rate": 5.696405200732939e-05,
+      "loss": 1.6238,
+      "step": 710
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.18405009806156158,
+      "learning_rate": 5.670430572633607e-05,
+      "loss": 1.5743,
+      "step": 711
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.22856737673282623,
+      "learning_rate": 5.644491844551121e-05,
+      "loss": 1.5763,
+      "step": 712
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.1876060515642166,
+      "learning_rate": 5.6185892315647856e-05,
+      "loss": 1.6686,
+      "step": 713
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.19633199274539948,
+      "learning_rate": 5.592722948454437e-05,
+      "loss": 1.5807,
+      "step": 714
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.18378899991512299,
+      "learning_rate": 5.56689320969868e-05,
+      "loss": 1.5507,
+      "step": 715
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.18441124260425568,
+      "learning_rate": 5.5411002294730996e-05,
+      "loss": 1.6386,
+      "step": 716
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.20815691351890564,
+      "learning_rate": 5.515344221648484e-05,
+      "loss": 1.4727,
+      "step": 717
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.19462059438228607,
+      "learning_rate": 5.489625399789048e-05,
+      "loss": 1.5376,
+      "step": 718
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.2019672840833664,
+      "learning_rate": 5.463943977150674e-05,
+      "loss": 1.5645,
+      "step": 719
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.20614993572235107,
+      "learning_rate": 5.438300166679134e-05,
+      "loss": 1.6275,
+      "step": 720
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.18458101153373718,
+      "learning_rate": 5.412694181008329e-05,
+      "loss": 1.589,
+      "step": 721
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.18971580266952515,
+      "learning_rate": 5.387126232458522e-05,
+      "loss": 1.5143,
+      "step": 722
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.18646183609962463,
+      "learning_rate": 5.3615965330345716e-05,
+      "loss": 1.5866,
+      "step": 723
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.19913147389888763,
+      "learning_rate": 5.336105294424194e-05,
+      "loss": 1.5466,
+      "step": 724
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.20623794198036194,
+      "learning_rate": 5.310652727996188e-05,
+      "loss": 1.5166,
+      "step": 725
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.19151991605758667,
+      "learning_rate": 5.285239044798695e-05,
+      "loss": 1.4791,
+      "step": 726
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.20113767683506012,
+      "learning_rate": 5.2598644555574326e-05,
+      "loss": 1.6312,
+      "step": 727
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.20970012247562408,
+      "learning_rate": 5.2345291706739684e-05,
+      "loss": 1.5929,
+      "step": 728
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.21298512816429138,
+      "learning_rate": 5.209233400223963e-05,
+      "loss": 1.5484,
+      "step": 729
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.2009279429912567,
+      "learning_rate": 5.183977353955427e-05,
+      "loss": 1.57,
+      "step": 730
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.19530166685581207,
+      "learning_rate": 5.1587612412869954e-05,
+      "loss": 1.4022,
+      "step": 731
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.19641190767288208,
+      "learning_rate": 5.1335852713061594e-05,
+      "loss": 1.6239,
+      "step": 732
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.1912887543439865,
+      "learning_rate": 5.108449652767584e-05,
+      "loss": 1.5945,
+      "step": 733
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.20099541544914246,
+      "learning_rate": 5.0833545940913183e-05,
+      "loss": 1.5148,
+      "step": 734
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1101,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 367,
+  "total_flos": 2.190286687555289e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}