diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4986 @@
+{
+  "best_metric": 0.22935771942138672,
+  "best_model_checkpoint": "output/output__lora/checkpoint-600",
+  "epoch": 1.7010935601458081,
+  "eval_steps": 100,
+  "global_step": 700,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002430133657351154,
+      "grad_norm": 2.0297858715057373,
+      "learning_rate": 0.0,
+      "loss": 0.3878,
+      "step": 1
+    },
+    {
+      "epoch": 0.004860267314702308,
+      "grad_norm": 2.0333988666534424,
+      "learning_rate": 8.859191006777897e-06,
+      "loss": 0.4074,
+      "step": 2
+    },
+    {
+      "epoch": 0.007290400972053463,
+      "grad_norm": 2.1134026050567627,
+      "learning_rate": 1.4041485532469073e-05,
+      "loss": 0.3097,
+      "step": 3
+    },
+    {
+      "epoch": 0.009720534629404616,
+      "grad_norm": 2.1784684658050537,
+      "learning_rate": 1.7718382013555794e-05,
+      "loss": 0.4578,
+      "step": 4
+    },
+    {
+      "epoch": 0.012150668286755772,
+      "grad_norm": 1.0290217399597168,
+      "learning_rate": 2.0570404496611053e-05,
+      "loss": 0.3389,
+      "step": 5
+    },
+    {
+      "epoch": 0.014580801944106925,
+      "grad_norm": 3.486976385116577,
+      "learning_rate": 2.2900676539246968e-05,
+      "loss": 0.2617,
+      "step": 6
+    },
+    {
+      "epoch": 0.01701093560145808,
+      "grad_norm": 1.7177977561950684,
+      "learning_rate": 2.4870893478326387e-05,
+      "loss": 0.5574,
+      "step": 7
+    },
+    {
+      "epoch": 0.019441069258809233,
+      "grad_norm": 0.8118329048156738,
+      "learning_rate": 2.6577573020333684e-05,
+      "loss": 0.4099,
+      "step": 8
+    },
+    {
+      "epoch": 0.02187120291616039,
+      "grad_norm": 2.2652881145477295,
+      "learning_rate": 2.8082971064938146e-05,
+      "loss": 0.3207,
+      "step": 9
+    },
+    {
+      "epoch": 0.024301336573511544,
+      "grad_norm": 1.2344919443130493,
+      "learning_rate": 2.9429595503388953e-05,
+      "loss": 0.2961,
+      "step": 10
+    },
+    {
+      "epoch": 0.026731470230862697,
+      "grad_norm": 1.2169722318649292,
+      "learning_rate": 3.064776548439465e-05,
+      "loss": 0.3689,
+      "step": 11
+    },
+    {
+      "epoch": 0.02916160388821385,
+      "grad_norm": 1.162546992301941,
+      "learning_rate": 3.1759867546024865e-05,
+      "loss": 0.351,
+      "step": 12
+    },
+    {
+      "epoch": 0.031591737545565005,
+      "grad_norm": 2.5353000164031982,
+      "learning_rate": 3.2782902272079295e-05,
+      "loss": 0.2424,
+      "step": 13
+    },
+    {
+      "epoch": 0.03402187120291616,
+      "grad_norm": 1.6772927045822144,
+      "learning_rate": 3.373008448510428e-05,
+      "loss": 0.2366,
+      "step": 14
+    },
+    {
+      "epoch": 0.03645200486026731,
+      "grad_norm": 1.5499287843704224,
+      "learning_rate": 3.4611890029080124e-05,
+      "loss": 0.3358,
+      "step": 15
+    },
+    {
+      "epoch": 0.038882138517618466,
+      "grad_norm": 1.383988618850708,
+      "learning_rate": 3.543676402711159e-05,
+      "loss": 0.3495,
+      "step": 16
+    },
+    {
+      "epoch": 0.041312272174969626,
+      "grad_norm": 1.4687221050262451,
+      "learning_rate": 3.621161404374383e-05,
+      "loss": 0.3973,
+      "step": 17
+    },
+    {
+      "epoch": 0.04374240583232078,
+      "grad_norm": 1.1256321668624878,
+      "learning_rate": 3.694216207171603e-05,
+      "loss": 0.2556,
+      "step": 18
+    },
+    {
+      "epoch": 0.046172539489671933,
+      "grad_norm": 0.9001079797744751,
+      "learning_rate": 3.76332012245438e-05,
+      "loss": 0.2283,
+      "step": 19
+    },
+    {
+      "epoch": 0.04860267314702309,
+      "grad_norm": 2.164158582687378,
+      "learning_rate": 3.8288786510166846e-05,
+      "loss": 0.2065,
+      "step": 20
+    },
+    {
+      "epoch": 0.05103280680437424,
+      "grad_norm": 0.6704838871955872,
+      "learning_rate": 3.8912379010795455e-05,
+      "loss": 0.3622,
+      "step": 21
+    },
+    {
+      "epoch": 0.053462940461725394,
+      "grad_norm": 1.380202293395996,
+      "learning_rate": 3.9506956491172545e-05,
+      "loss": 0.2949,
+      "step": 22
+    },
+    {
+      "epoch": 0.05589307411907655,
+      "grad_norm": 1.281702995300293,
+      "learning_rate": 4.007509939970292e-05,
+      "loss": 0.1766,
+      "step": 23
+    },
+    {
+      "epoch": 0.0583232077764277,
+      "grad_norm": 1.581787347793579,
+      "learning_rate": 4.061905855280276e-05,
+      "loss": 0.2676,
+      "step": 24
+    },
+    {
+      "epoch": 0.060753341433778855,
+      "grad_norm": 1.8878554105758667,
+      "learning_rate": 4.1140808993222106e-05,
+      "loss": 0.2572,
+      "step": 25
+    },
+    {
+      "epoch": 0.06318347509113001,
+      "grad_norm": 0.6384598612785339,
+      "learning_rate": 4.164209327885719e-05,
+      "loss": 0.231,
+      "step": 26
+    },
+    {
+      "epoch": 0.06561360874848117,
+      "grad_norm": 0.9673507809638977,
+      "learning_rate": 4.2124456597407214e-05,
+      "loss": 0.1772,
+      "step": 27
+    },
+    {
+      "epoch": 0.06804374240583232,
+      "grad_norm": 1.0678174495697021,
+      "learning_rate": 4.258927549188218e-05,
+      "loss": 0.2025,
+      "step": 28
+    },
+    {
+      "epoch": 0.07047387606318348,
+      "grad_norm": 0.9189561009407043,
+      "learning_rate": 4.303778154313212e-05,
+      "loss": 0.2345,
+      "step": 29
+    },
+    {
+      "epoch": 0.07290400972053462,
+      "grad_norm": 1.3609440326690674,
+      "learning_rate": 4.347108103585803e-05,
+      "loss": 0.2201,
+      "step": 30
+    },
+    {
+      "epoch": 0.07533414337788578,
+      "grad_norm": 1.3813296556472778,
+      "learning_rate": 4.389017139879164e-05,
+      "loss": 0.2438,
+      "step": 31
+    },
+    {
+      "epoch": 0.07776427703523693,
+      "grad_norm": 1.1185611486434937,
+      "learning_rate": 4.429595503388948e-05,
+      "loss": 0.1972,
+      "step": 32
+    },
+    {
+      "epoch": 0.08019441069258809,
+      "grad_norm": 0.876745343208313,
+      "learning_rate": 4.468925101686371e-05,
+      "loss": 0.2754,
+      "step": 33
+    },
+    {
+      "epoch": 0.08262454434993925,
+      "grad_norm": 0.7372068762779236,
+      "learning_rate": 4.507080505052173e-05,
+      "loss": 0.2719,
+      "step": 34
+    },
+    {
+      "epoch": 0.0850546780072904,
+      "grad_norm": 1.4229792356491089,
+      "learning_rate": 4.544129797493744e-05,
+      "loss": 0.3219,
+      "step": 35
+    },
+    {
+      "epoch": 0.08748481166464156,
+      "grad_norm": 1.4566582441329956,
+      "learning_rate": 4.5801353078493936e-05,
+      "loss": 0.1257,
+      "step": 36
+    },
+    {
+      "epoch": 0.0899149453219927,
+      "grad_norm": 1.217944860458374,
+      "learning_rate": 4.615154240700883e-05,
+      "loss": 0.3138,
+      "step": 37
+    },
+    {
+      "epoch": 0.09234507897934387,
+      "grad_norm": 1.882925033569336,
+      "learning_rate": 4.6492392231321696e-05,
+      "loss": 0.219,
+      "step": 38
+    },
+    {
+      "epoch": 0.09477521263669501,
+      "grad_norm": 0.8372556567192078,
+      "learning_rate": 4.682438780454837e-05,
+      "loss": 0.27,
+      "step": 39
+    },
+    {
+      "epoch": 0.09720534629404617,
+      "grad_norm": 0.9709153771400452,
+      "learning_rate": 4.714797751694474e-05,
+      "loss": 0.2759,
+      "step": 40
+    },
+    {
+      "epoch": 0.09963547995139732,
+      "grad_norm": 1.1802537441253662,
+      "learning_rate": 4.7463576537657414e-05,
+      "loss": 0.2411,
+      "step": 41
+    },
+    {
+      "epoch": 0.10206561360874848,
+      "grad_norm": 0.7289960980415344,
+      "learning_rate": 4.777157001757336e-05,
+      "loss": 0.2755,
+      "step": 42
+    },
+    {
+      "epoch": 0.10449574726609964,
+      "grad_norm": 0.8493523001670837,
+      "learning_rate": 4.8072315915252694e-05,
+      "loss": 0.3177,
+      "step": 43
+    },
+    {
+      "epoch": 0.10692588092345079,
+      "grad_norm": 1.0613563060760498,
+      "learning_rate": 4.8366147497950435e-05,
+      "loss": 0.2056,
+      "step": 44
+    },
+    {
+      "epoch": 0.10935601458080195,
+      "grad_norm": 1.1137558221817017,
+      "learning_rate": 4.8653375561549195e-05,
+      "loss": 0.293,
+      "step": 45
+    },
+    {
+      "epoch": 0.1117861482381531,
+      "grad_norm": 0.6552504301071167,
+      "learning_rate": 4.8934290406480814e-05,
+      "loss": 0.2478,
+      "step": 46
+    },
+    {
+      "epoch": 0.11421628189550426,
+      "grad_norm": 0.9905387759208679,
+      "learning_rate": 4.920916360113129e-05,
+      "loss": 0.2488,
+      "step": 47
+    },
+    {
+      "epoch": 0.1166464155528554,
+      "grad_norm": 1.101691484451294,
+      "learning_rate": 4.947824955958066e-05,
+      "loss": 0.1583,
+      "step": 48
+    },
+    {
+      "epoch": 0.11907654921020656,
+      "grad_norm": 1.3675787448883057,
+      "learning_rate": 4.9741786956652774e-05,
+      "loss": 0.3629,
+      "step": 49
+    },
+    {
+      "epoch": 0.12150668286755771,
+      "grad_norm": 2.6356711387634277,
+      "learning_rate": 5e-05,
+      "loss": 0.285,
+      "step": 50
+    },
+    {
+      "epoch": 0.12393681652490887,
+      "grad_norm": 1.5550183057785034,
+      "learning_rate": 5e-05,
+      "loss": 0.1688,
+      "step": 51
+    },
+    {
+      "epoch": 0.12636695018226002,
+      "grad_norm": 1.7116848230361938,
+      "learning_rate": 5e-05,
+      "loss": 0.2495,
+      "step": 52
+    },
+    {
+      "epoch": 0.12879708383961117,
+      "grad_norm": 1.3964793682098389,
+      "learning_rate": 5e-05,
+      "loss": 0.276,
+      "step": 53
+    },
+    {
+      "epoch": 0.13122721749696234,
+      "grad_norm": 0.8001569509506226,
+      "learning_rate": 5e-05,
+      "loss": 0.2976,
+      "step": 54
+    },
+    {
+      "epoch": 0.1336573511543135,
+      "grad_norm": 1.331112027168274,
+      "learning_rate": 5e-05,
+      "loss": 0.2725,
+      "step": 55
+    },
+    {
+      "epoch": 0.13608748481166463,
+      "grad_norm": 1.0783005952835083,
+      "learning_rate": 5e-05,
+      "loss": 0.188,
+      "step": 56
+    },
+    {
+      "epoch": 0.1385176184690158,
+      "grad_norm": 0.9507465958595276,
+      "learning_rate": 5e-05,
+      "loss": 0.2917,
+      "step": 57
+    },
+    {
+      "epoch": 0.14094775212636695,
+      "grad_norm": 0.8285686373710632,
+      "learning_rate": 5e-05,
+      "loss": 0.3322,
+      "step": 58
+    },
+    {
+      "epoch": 0.1433778857837181,
+      "grad_norm": 0.8201387524604797,
+      "learning_rate": 5e-05,
+      "loss": 0.2296,
+      "step": 59
+    },
+    {
+      "epoch": 0.14580801944106925,
+      "grad_norm": 0.743626594543457,
+      "learning_rate": 5e-05,
+      "loss": 0.1785,
+      "step": 60
+    },
+    {
+      "epoch": 0.14823815309842042,
+      "grad_norm": 0.9876251816749573,
+      "learning_rate": 5e-05,
+      "loss": 0.3416,
+      "step": 61
+    },
+    {
+      "epoch": 0.15066828675577157,
+      "grad_norm": 1.6088169813156128,
+      "learning_rate": 5e-05,
+      "loss": 0.2513,
+      "step": 62
+    },
+    {
+      "epoch": 0.15309842041312272,
+      "grad_norm": 1.2406659126281738,
+      "learning_rate": 5e-05,
+      "loss": 0.2856,
+      "step": 63
+    },
+    {
+      "epoch": 0.15552855407047386,
+      "grad_norm": 1.0257083177566528,
+      "learning_rate": 5e-05,
+      "loss": 0.1392,
+      "step": 64
+    },
+    {
+      "epoch": 0.15795868772782504,
+      "grad_norm": 0.7829582691192627,
+      "learning_rate": 5e-05,
+      "loss": 0.248,
+      "step": 65
+    },
+    {
+      "epoch": 0.16038882138517618,
+      "grad_norm": 0.7451558113098145,
+      "learning_rate": 5e-05,
+      "loss": 0.2826,
+      "step": 66
+    },
+    {
+      "epoch": 0.16281895504252733,
+      "grad_norm": 0.8115066885948181,
+      "learning_rate": 5e-05,
+      "loss": 0.3322,
+      "step": 67
+    },
+    {
+      "epoch": 0.1652490886998785,
+      "grad_norm": 0.8748694658279419,
+      "learning_rate": 5e-05,
+      "loss": 0.2057,
+      "step": 68
+    },
+    {
+      "epoch": 0.16767922235722965,
+      "grad_norm": 0.6844836473464966,
+      "learning_rate": 5e-05,
+      "loss": 0.2062,
+      "step": 69
+    },
+    {
+      "epoch": 0.1701093560145808,
+      "grad_norm": 0.9419387578964233,
+      "learning_rate": 5e-05,
+      "loss": 0.169,
+      "step": 70
+    },
+    {
+      "epoch": 0.17253948967193194,
+      "grad_norm": 0.8529036045074463,
+      "learning_rate": 5e-05,
+      "loss": 0.28,
+      "step": 71
+    },
+    {
+      "epoch": 0.17496962332928312,
+      "grad_norm": 0.8096457719802856,
+      "learning_rate": 5e-05,
+      "loss": 0.1699,
+      "step": 72
+    },
+    {
+      "epoch": 0.17739975698663427,
+      "grad_norm": 0.7595834136009216,
+      "learning_rate": 5e-05,
+      "loss": 0.2321,
+      "step": 73
+    },
+    {
+      "epoch": 0.1798298906439854,
+      "grad_norm": 0.6918485164642334,
+      "learning_rate": 5e-05,
+      "loss": 0.12,
+      "step": 74
+    },
+    {
+      "epoch": 0.1822600243013366,
+      "grad_norm": 0.8465882539749146,
+      "learning_rate": 5e-05,
+      "loss": 0.2344,
+      "step": 75
+    },
+    {
+      "epoch": 0.18469015795868773,
+      "grad_norm": 1.27568519115448,
+      "learning_rate": 5e-05,
+      "loss": 0.3018,
+      "step": 76
+    },
+    {
+      "epoch": 0.18712029161603888,
+      "grad_norm": 0.907692015171051,
+      "learning_rate": 5e-05,
+      "loss": 0.2154,
+      "step": 77
+    },
+    {
+      "epoch": 0.18955042527339003,
+      "grad_norm": 0.9362420439720154,
+      "learning_rate": 5e-05,
+      "loss": 0.2024,
+      "step": 78
+    },
+    {
+      "epoch": 0.1919805589307412,
+      "grad_norm": 1.1268069744110107,
+      "learning_rate": 5e-05,
+      "loss": 0.1509,
+      "step": 79
+    },
+    {
+      "epoch": 0.19441069258809235,
+      "grad_norm": 2.944066047668457,
+      "learning_rate": 5e-05,
+      "loss": 0.1669,
+      "step": 80
+    },
+    {
+      "epoch": 0.1968408262454435,
+      "grad_norm": 0.8719034790992737,
+      "learning_rate": 5e-05,
+      "loss": 0.2229,
+      "step": 81
+    },
+    {
+      "epoch": 0.19927095990279464,
+      "grad_norm": 0.7963365912437439,
+      "learning_rate": 5e-05,
+      "loss": 0.1485,
+      "step": 82
+    },
+    {
+      "epoch": 0.20170109356014582,
+      "grad_norm": 1.1861976385116577,
+      "learning_rate": 5e-05,
+      "loss": 0.1774,
+      "step": 83
+    },
+    {
+      "epoch": 0.20413122721749696,
+      "grad_norm": 1.029300570487976,
+      "learning_rate": 5e-05,
+      "loss": 0.2762,
+      "step": 84
+    },
+    {
+      "epoch": 0.2065613608748481,
+      "grad_norm": 0.7859634757041931,
+      "learning_rate": 5e-05,
+      "loss": 0.263,
+      "step": 85
+    },
+    {
+      "epoch": 0.20899149453219928,
+      "grad_norm": 1.2088333368301392,
+      "learning_rate": 5e-05,
+      "loss": 0.2248,
+      "step": 86
+    },
+    {
+      "epoch": 0.21142162818955043,
+      "grad_norm": 0.9491986036300659,
+      "learning_rate": 5e-05,
+      "loss": 0.1388,
+      "step": 87
+    },
+    {
+      "epoch": 0.21385176184690158,
+      "grad_norm": 1.3867732286453247,
+      "learning_rate": 5e-05,
+      "loss": 0.1932,
+      "step": 88
+    },
+    {
+      "epoch": 0.21628189550425272,
+      "grad_norm": 1.2418168783187866,
+      "learning_rate": 5e-05,
+      "loss": 0.2703,
+      "step": 89
+    },
+    {
+      "epoch": 0.2187120291616039,
+      "grad_norm": 1.514169692993164,
+      "learning_rate": 5e-05,
+      "loss": 0.3437,
+      "step": 90
+    },
+    {
+      "epoch": 0.22114216281895505,
+      "grad_norm": 1.1384562253952026,
+      "learning_rate": 5e-05,
+      "loss": 0.3157,
+      "step": 91
+    },
+    {
+      "epoch": 0.2235722964763062,
+      "grad_norm": 1.0910252332687378,
+      "learning_rate": 5e-05,
+      "loss": 0.3296,
+      "step": 92
+    },
+    {
+      "epoch": 0.22600243013365734,
+      "grad_norm": 0.8576286435127258,
+      "learning_rate": 5e-05,
+      "loss": 0.2572,
+      "step": 93
+    },
+    {
+      "epoch": 0.2284325637910085,
+      "grad_norm": 1.0798954963684082,
+      "learning_rate": 5e-05,
+      "loss": 0.2669,
+      "step": 94
+    },
+    {
+      "epoch": 0.23086269744835966,
+      "grad_norm": 1.0142358541488647,
+      "learning_rate": 5e-05,
+      "loss": 0.2607,
+      "step": 95
+    },
+    {
+      "epoch": 0.2332928311057108,
+      "grad_norm": 0.8707209825515747,
+      "learning_rate": 5e-05,
+      "loss": 0.2206,
+      "step": 96
+    },
+    {
+      "epoch": 0.23572296476306198,
+      "grad_norm": 1.038823127746582,
+      "learning_rate": 5e-05,
+      "loss": 0.1648,
+      "step": 97
+    },
+    {
+      "epoch": 0.23815309842041313,
+      "grad_norm": 1.3246694803237915,
+      "learning_rate": 5e-05,
+      "loss": 0.317,
+      "step": 98
+    },
+    {
+      "epoch": 0.24058323207776428,
+      "grad_norm": 1.0180795192718506,
+      "learning_rate": 5e-05,
+      "loss": 0.2653,
+      "step": 99
+    },
+    {
+      "epoch": 0.24301336573511542,
+      "grad_norm": 1.803346037864685,
+      "learning_rate": 5e-05,
+      "loss": 0.1221,
+      "step": 100
+    },
+    {
+      "epoch": 0.24301336573511542,
+      "eval_loss": 0.2808912396430969,
+      "eval_runtime": 508.6685,
+      "eval_samples_per_second": 5.402,
+      "eval_steps_per_second": 0.676,
+      "step": 100
+    },
+    {
+      "epoch": 0.2454434993924666,
+      "grad_norm": 1.4270009994506836,
+      "learning_rate": 5e-05,
+      "loss": 0.1513,
+      "step": 101
+    },
+    {
+      "epoch": 0.24787363304981774,
+      "grad_norm": 1.5230388641357422,
+      "learning_rate": 5e-05,
+      "loss": 0.2563,
+      "step": 102
+    },
+    {
+      "epoch": 0.2503037667071689,
+      "grad_norm": 1.0142149925231934,
+      "learning_rate": 5e-05,
+      "loss": 0.2021,
+      "step": 103
+    },
+    {
+      "epoch": 0.25273390036452004,
+      "grad_norm": 1.224928617477417,
+      "learning_rate": 5e-05,
+      "loss": 0.2227,
+      "step": 104
+    },
+    {
+      "epoch": 0.2551640340218712,
+      "grad_norm": 0.8966079950332642,
+      "learning_rate": 5e-05,
+      "loss": 0.1416,
+      "step": 105
+    },
+    {
+      "epoch": 0.25759416767922233,
+      "grad_norm": 4.392244815826416,
+      "learning_rate": 5e-05,
+      "loss": 0.1297,
+      "step": 106
+    },
+    {
+      "epoch": 0.2600243013365735,
+      "grad_norm": 0.7403039932250977,
+      "learning_rate": 5e-05,
+      "loss": 0.1383,
+      "step": 107
+    },
+    {
+      "epoch": 0.2624544349939247,
+      "grad_norm": 0.7826500535011292,
+      "learning_rate": 5e-05,
+      "loss": 0.1933,
+      "step": 108
+    },
+    {
+      "epoch": 0.2648845686512758,
+      "grad_norm": 1.2112810611724854,
+      "learning_rate": 5e-05,
+      "loss": 0.2118,
+      "step": 109
+    },
+    {
+      "epoch": 0.267314702308627,
+      "grad_norm": 1.2826062440872192,
+      "learning_rate": 5e-05,
+      "loss": 0.2028,
+      "step": 110
+    },
+    {
+      "epoch": 0.26974483596597815,
+      "grad_norm": 0.9675102829933167,
+      "learning_rate": 5e-05,
+      "loss": 0.1481,
+      "step": 111
+    },
+    {
+      "epoch": 0.27217496962332927,
+      "grad_norm": 1.2551299333572388,
+      "learning_rate": 5e-05,
+      "loss": 0.2172,
+      "step": 112
+    },
+    {
+      "epoch": 0.27460510328068044,
+      "grad_norm": 1.2526607513427734,
+      "learning_rate": 5e-05,
+      "loss": 0.213,
+      "step": 113
+    },
+    {
+      "epoch": 0.2770352369380316,
+      "grad_norm": 1.102669358253479,
+      "learning_rate": 5e-05,
+      "loss": 0.2387,
+      "step": 114
+    },
+    {
+      "epoch": 0.27946537059538273,
+      "grad_norm": 1.1452562808990479,
+      "learning_rate": 5e-05,
+      "loss": 0.2224,
+      "step": 115
+    },
+    {
+      "epoch": 0.2818955042527339,
+      "grad_norm": 1.4839582443237305,
+      "learning_rate": 5e-05,
+      "loss": 0.3572,
+      "step": 116
+    },
+    {
+      "epoch": 0.284325637910085,
+      "grad_norm": 0.7019845247268677,
+      "learning_rate": 5e-05,
+      "loss": 0.1531,
+      "step": 117
+    },
+    {
+      "epoch": 0.2867557715674362,
+      "grad_norm": 1.1344879865646362,
+      "learning_rate": 5e-05,
+      "loss": 0.1234,
+      "step": 118
+    },
+    {
+      "epoch": 0.2891859052247874,
+      "grad_norm": 0.7753707766532898,
+      "learning_rate": 5e-05,
+      "loss": 0.3054,
+      "step": 119
+    },
+    {
+      "epoch": 0.2916160388821385,
+      "grad_norm": 1.2467390298843384,
+      "learning_rate": 5e-05,
+      "loss": 0.0979,
+      "step": 120
+    },
+    {
+      "epoch": 0.29404617253948967,
+      "grad_norm": 0.8829241991043091,
+      "learning_rate": 5e-05,
+      "loss": 0.1251,
+      "step": 121
+    },
+    {
+      "epoch": 0.29647630619684084,
+      "grad_norm": 0.9316434860229492,
+      "learning_rate": 5e-05,
+      "loss": 0.1723,
+      "step": 122
+    },
+    {
+      "epoch": 0.29890643985419196,
+      "grad_norm": 0.9735895395278931,
+      "learning_rate": 5e-05,
+      "loss": 0.1992,
+      "step": 123
+    },
+    {
+      "epoch": 0.30133657351154314,
+      "grad_norm": 1.0356009006500244,
+      "learning_rate": 5e-05,
+      "loss": 0.2912,
+      "step": 124
+    },
+    {
+      "epoch": 0.3037667071688943,
+      "grad_norm": 0.8671007752418518,
+      "learning_rate": 5e-05,
+      "loss": 0.2204,
+      "step": 125
+    },
+    {
+      "epoch": 0.30619684082624543,
+      "grad_norm": 0.9986138939857483,
+      "learning_rate": 5e-05,
+      "loss": 0.1615,
+      "step": 126
+    },
+    {
+      "epoch": 0.3086269744835966,
+      "grad_norm": 0.8567970991134644,
+      "learning_rate": 5e-05,
+      "loss": 0.3127,
+      "step": 127
+    },
+    {
+      "epoch": 0.3110571081409477,
+      "grad_norm": 1.497078537940979,
+      "learning_rate": 5e-05,
+      "loss": 0.1461,
+      "step": 128
+    },
+    {
+      "epoch": 0.3134872417982989,
+      "grad_norm": 1.9854991436004639,
+      "learning_rate": 5e-05,
+      "loss": 0.1588,
+      "step": 129
+    },
+    {
+      "epoch": 0.3159173754556501,
+      "grad_norm": 0.9955502152442932,
+      "learning_rate": 5e-05,
+      "loss": 0.2703,
+      "step": 130
+    },
+    {
+      "epoch": 0.3183475091130012,
+      "grad_norm": 0.8443557620048523,
+      "learning_rate": 5e-05,
+      "loss": 0.1432,
+      "step": 131
+    },
+    {
+      "epoch": 0.32077764277035237,
+      "grad_norm": 0.8821367025375366,
+      "learning_rate": 5e-05,
+      "loss": 0.1113,
+      "step": 132
+    },
+    {
+      "epoch": 0.32320777642770354,
+      "grad_norm": 0.997931957244873,
+      "learning_rate": 5e-05,
+      "loss": 0.1363,
+      "step": 133
+    },
+    {
+      "epoch": 0.32563791008505466,
+      "grad_norm": 1.2628774642944336,
+      "learning_rate": 5e-05,
+      "loss": 0.134,
+      "step": 134
+    },
+    {
+      "epoch": 0.32806804374240583,
+      "grad_norm": 0.7694957852363586,
+      "learning_rate": 5e-05,
+      "loss": 0.1425,
+      "step": 135
+    },
+    {
+      "epoch": 0.330498177399757,
+      "grad_norm": 1.5881726741790771,
+      "learning_rate": 5e-05,
+      "loss": 0.17,
+      "step": 136
+    },
+    {
+      "epoch": 0.33292831105710813,
+      "grad_norm": 1.2904351949691772,
+      "learning_rate": 5e-05,
+      "loss": 0.1735,
+      "step": 137
+    },
+    {
+      "epoch": 0.3353584447144593,
+      "grad_norm": 1.3187663555145264,
+      "learning_rate": 5e-05,
+      "loss": 0.2035,
+      "step": 138
+    },
+    {
+      "epoch": 0.3377885783718105,
+      "grad_norm": 1.1753528118133545,
+      "learning_rate": 5e-05,
+      "loss": 0.2147,
+      "step": 139
+    },
+    {
+      "epoch": 0.3402187120291616,
+      "grad_norm": 1.6974895000457764,
+      "learning_rate": 5e-05,
+      "loss": 0.1303,
+      "step": 140
+    },
+    {
+      "epoch": 0.34264884568651277,
+      "grad_norm": 0.9945745468139648,
+      "learning_rate": 5e-05,
+      "loss": 0.1169,
+      "step": 141
+    },
+    {
+      "epoch": 0.3450789793438639,
+      "grad_norm": 1.0099682807922363,
+      "learning_rate": 5e-05,
+      "loss": 0.1446,
+      "step": 142
+    },
+    {
+      "epoch": 0.34750911300121506,
+      "grad_norm": 1.223264217376709,
+      "learning_rate": 5e-05,
+      "loss": 0.2118,
+      "step": 143
+    },
+    {
+      "epoch": 0.34993924665856624,
+      "grad_norm": 0.8575140833854675,
+      "learning_rate": 5e-05,
+      "loss": 0.1109,
+      "step": 144
+    },
+    {
+      "epoch": 0.35236938031591736,
+      "grad_norm": 1.0145286321640015,
+      "learning_rate": 5e-05,
+      "loss": 0.1664,
+      "step": 145
+    },
+    {
+      "epoch": 0.35479951397326853,
+      "grad_norm": 0.670891284942627,
+      "learning_rate": 5e-05,
+      "loss": 0.1166,
+      "step": 146
+    },
+    {
+      "epoch": 0.3572296476306197,
+      "grad_norm": 1.4332047700881958,
+      "learning_rate": 5e-05,
+      "loss": 0.1381,
+      "step": 147
+    },
+    {
+      "epoch": 0.3596597812879708,
+      "grad_norm": 1.0511322021484375,
+      "learning_rate": 5e-05,
+      "loss": 0.1279,
+      "step": 148
+    },
+    {
+      "epoch": 0.362089914945322,
+      "grad_norm": 0.751565158367157,
+      "learning_rate": 5e-05,
+      "loss": 0.1533,
+      "step": 149
+    },
+    {
+      "epoch": 0.3645200486026732,
+      "grad_norm": 1.612855076789856,
+      "learning_rate": 5e-05,
+      "loss": 0.1633,
+      "step": 150
+    },
+    {
+      "epoch": 0.3669501822600243,
+      "grad_norm": 1.2449802160263062,
+      "learning_rate": 5e-05,
+      "loss": 0.2037,
+      "step": 151
+    },
+    {
+      "epoch": 0.36938031591737547,
+      "grad_norm": 1.0060147047042847,
+      "learning_rate": 5e-05,
+      "loss": 0.0902,
+      "step": 152
+    },
+    {
+      "epoch": 0.3718104495747266,
+      "grad_norm": 0.9183961153030396,
+      "learning_rate": 5e-05,
+      "loss": 0.1748,
+      "step": 153
+    },
+    {
+      "epoch": 0.37424058323207776,
+      "grad_norm": 1.2824058532714844,
+      "learning_rate": 5e-05,
+      "loss": 0.1903,
+      "step": 154
+    },
+    {
+      "epoch": 0.37667071688942894,
+      "grad_norm": 0.524626612663269,
+      "learning_rate": 5e-05,
+      "loss": 0.1014,
+      "step": 155
+    },
+    {
+      "epoch": 0.37910085054678005,
+      "grad_norm": 0.946535587310791,
+      "learning_rate": 5e-05,
+      "loss": 0.2512,
+      "step": 156
+    },
+    {
+      "epoch": 0.38153098420413123,
+      "grad_norm": 0.8985329270362854,
+      "learning_rate": 5e-05,
+      "loss": 0.1247,
+      "step": 157
+    },
+    {
+      "epoch": 0.3839611178614824,
+      "grad_norm": 0.6025483012199402,
+      "learning_rate": 5e-05,
+      "loss": 0.0677,
+      "step": 158
+    },
+    {
+      "epoch": 0.3863912515188335,
+      "grad_norm": 0.9210860729217529,
+      "learning_rate": 5e-05,
+      "loss": 0.1361,
+      "step": 159
+    },
+    {
+      "epoch": 0.3888213851761847,
+      "grad_norm": 1.7067134380340576,
+      "learning_rate": 5e-05,
+      "loss": 0.2008,
+      "step": 160
+    },
+    {
+      "epoch": 0.39125151883353587,
+      "grad_norm": 0.8162620663642883,
+      "learning_rate": 5e-05,
+      "loss": 0.1108,
+      "step": 161
+    },
+    {
+      "epoch": 0.393681652490887,
+      "grad_norm": 0.535029947757721,
+      "learning_rate": 5e-05,
+      "loss": 0.1578,
+      "step": 162
+    },
+    {
+      "epoch": 0.39611178614823817,
+      "grad_norm": 0.7219657897949219,
+      "learning_rate": 5e-05,
+      "loss": 0.1012,
+      "step": 163
+    },
+    {
+      "epoch": 0.3985419198055893,
+      "grad_norm": 1.0636670589447021,
+      "learning_rate": 5e-05,
+      "loss": 0.1648,
+      "step": 164
+    },
+    {
+      "epoch": 0.40097205346294046,
+      "grad_norm": 1.1116470098495483,
+      "learning_rate": 5e-05,
+      "loss": 0.1191,
+      "step": 165
+    },
+    {
+      "epoch": 0.40340218712029163,
+      "grad_norm": 0.7233144044876099,
+      "learning_rate": 5e-05,
+      "loss": 0.1206,
+      "step": 166
+    },
+    {
+      "epoch": 0.40583232077764275,
+      "grad_norm": 1.7054411172866821,
+      "learning_rate": 5e-05,
+      "loss": 0.1466,
+      "step": 167
+    },
+    {
+      "epoch": 0.4082624544349939,
+      "grad_norm": 0.9031124114990234,
+      "learning_rate": 5e-05,
+      "loss": 0.1107,
+      "step": 168
+    },
+    {
+      "epoch": 0.4106925880923451,
+      "grad_norm": 0.8140673637390137,
+      "learning_rate": 5e-05,
+      "loss": 0.1503,
+      "step": 169
+    },
+    {
+      "epoch": 0.4131227217496962,
+      "grad_norm": 1.2189490795135498,
+      "learning_rate": 5e-05,
+      "loss": 0.1081,
+      "step": 170
+    },
+    {
+      "epoch": 0.4155528554070474,
+      "grad_norm": 0.8403185606002808,
+      "learning_rate": 5e-05,
+      "loss": 0.098,
+      "step": 171
+    },
+    {
+      "epoch": 0.41798298906439857,
+      "grad_norm": 0.8939706683158875,
+      "learning_rate": 5e-05,
+      "loss": 0.0961,
+      "step": 172
+    },
+    {
+      "epoch": 0.4204131227217497,
+      "grad_norm": 0.8737717270851135,
+      "learning_rate": 5e-05,
+      "loss": 0.1165,
+      "step": 173
+    },
+    {
+      "epoch": 0.42284325637910086,
+      "grad_norm": 0.9247780442237854,
+      "learning_rate": 5e-05,
+      "loss": 0.1787,
+      "step": 174
+    },
+    {
+      "epoch": 0.425273390036452,
+      "grad_norm": 0.9532423615455627,
+      "learning_rate": 5e-05,
+      "loss": 0.1676,
+      "step": 175
+    },
+    {
+      "epoch": 0.42770352369380316,
+      "grad_norm": 0.5811410546302795,
+      "learning_rate": 5e-05,
+      "loss": 0.1221,
+      "step": 176
+    },
+    {
+      "epoch": 0.43013365735115433,
+      "grad_norm": 0.6600822806358337,
+      "learning_rate": 5e-05,
+      "loss": 0.1809,
+      "step": 177
+    },
+    {
+      "epoch": 0.43256379100850545,
+      "grad_norm": 0.5638179183006287,
+      "learning_rate": 5e-05,
+      "loss": 0.0668,
+      "step": 178
+    },
+    {
+      "epoch": 0.4349939246658566,
+      "grad_norm": 0.5698412656784058,
+      "learning_rate": 5e-05,
+      "loss": 0.164,
+      "step": 179
+    },
+    {
+      "epoch": 0.4374240583232078,
+      "grad_norm": 1.2823748588562012,
+      "learning_rate": 5e-05,
+      "loss": 0.2111,
+      "step": 180
+    },
+    {
+      "epoch": 0.4398541919805589,
+      "grad_norm": 1.1044129133224487,
+      "learning_rate": 5e-05,
+      "loss": 0.1188,
+      "step": 181
+    },
+    {
+      "epoch": 0.4422843256379101,
+      "grad_norm": 0.9815372824668884,
+      "learning_rate": 5e-05,
+      "loss": 0.186,
+      "step": 182
+    },
+    {
+      "epoch": 0.44471445929526127,
+      "grad_norm": 0.8543856739997864,
+      "learning_rate": 5e-05,
+      "loss": 0.1237,
+      "step": 183
+    },
+    {
+      "epoch": 0.4471445929526124,
+      "grad_norm": 1.0749317407608032,
+      "learning_rate": 5e-05,
+      "loss": 0.1385,
+      "step": 184
+    },
+    {
+      "epoch": 0.44957472660996356,
+      "grad_norm": 1.4067716598510742,
+      "learning_rate": 5e-05,
+      "loss": 0.1269,
+      "step": 185
+    },
+    {
+      "epoch": 0.4520048602673147,
+      "grad_norm": 0.933139443397522,
+      "learning_rate": 5e-05,
+      "loss": 0.0476,
+      "step": 186
+    },
+    {
+      "epoch": 0.45443499392466585,
+      "grad_norm": 1.2160463333129883,
+      "learning_rate": 5e-05,
+      "loss": 0.0823,
+      "step": 187
+    },
+    {
+      "epoch": 0.456865127582017,
+      "grad_norm": 0.8305476903915405,
+      "learning_rate": 5e-05,
+      "loss": 0.2482,
+      "step": 188
+    },
+    {
+      "epoch": 0.45929526123936815,
+      "grad_norm": 0.8701503872871399,
+      "learning_rate": 5e-05,
+      "loss": 0.2007,
+      "step": 189
+    },
+    {
+      "epoch": 0.4617253948967193,
+      "grad_norm": 1.1040858030319214,
+      "learning_rate": 5e-05,
+      "loss": 0.1031,
+      "step": 190
+    },
+    {
+      "epoch": 0.4641555285540705,
+      "grad_norm": 1.149579644203186,
+      "learning_rate": 5e-05,
+      "loss": 0.1384,
+      "step": 191
+    },
+    {
+      "epoch": 0.4665856622114216,
+      "grad_norm": 1.0616861581802368,
+      "learning_rate": 5e-05,
+      "loss": 0.1961,
+      "step": 192
+    },
+    {
+      "epoch": 0.4690157958687728,
+      "grad_norm": 1.3416138887405396,
+      "learning_rate": 5e-05,
+      "loss": 0.2223,
+      "step": 193
+    },
+    {
+      "epoch": 0.47144592952612396,
+      "grad_norm": 1.2578601837158203,
+      "learning_rate": 5e-05,
+      "loss": 0.0833,
+      "step": 194
+    },
+    {
+      "epoch": 0.4738760631834751,
+      "grad_norm": 0.9381983876228333,
+      "learning_rate": 5e-05,
+      "loss": 0.135,
+      "step": 195
+    },
+    {
+      "epoch": 0.47630619684082626,
+      "grad_norm": 1.0591320991516113,
+      "learning_rate": 5e-05,
+      "loss": 0.1658,
+      "step": 196
+    },
+    {
+      "epoch": 0.4787363304981774,
+      "grad_norm": 0.9011105895042419,
+      "learning_rate": 5e-05,
+      "loss": 0.189,
+      "step": 197
+    },
+    {
+      "epoch": 0.48116646415552855,
+      "grad_norm": 0.9011105895042419,
+      "learning_rate": 5e-05,
+      "loss": 0.1729,
+      "step": 198
+    },
+    {
+      "epoch": 0.4835965978128797,
+      "grad_norm": 1.0387153625488281,
+      "learning_rate": 5e-05,
+      "loss": 0.1534,
+      "step": 199
+    },
+    {
+      "epoch": 0.48602673147023084,
+      "grad_norm": 1.7026396989822388,
+      "learning_rate": 5e-05,
+      "loss": 0.1534,
+      "step": 200
+    },
+    {
+      "epoch": 0.48602673147023084,
+      "eval_loss": 0.2694580554962158,
+      "eval_runtime": 510.8356,
+      "eval_samples_per_second": 5.379,
+      "eval_steps_per_second": 0.673,
+      "step": 200
+    },
+    {
+      "epoch": 0.488456865127582,
+      "grad_norm": 0.7090218663215637,
+      "learning_rate": 5e-05,
+      "loss": 0.1022,
+      "step": 201
+    },
+    {
+      "epoch": 0.4908869987849332,
+      "grad_norm": 0.6188836693763733,
+      "learning_rate": 5e-05,
+      "loss": 0.1839,
+      "step": 202
+    },
+    {
+      "epoch": 0.4933171324422843,
+      "grad_norm": 0.8001598715782166,
+      "learning_rate": 5e-05,
+      "loss": 0.1525,
+      "step": 203
+    },
+    {
+      "epoch": 0.4957472660996355,
+      "grad_norm": 1.006425142288208,
+      "learning_rate": 5e-05,
+      "loss": 0.1417,
+      "step": 204
+    },
+    {
+      "epoch": 0.49817739975698666,
+      "grad_norm": 0.9688583612442017,
+      "learning_rate": 5e-05,
+      "loss": 0.1504,
+      "step": 205
+    },
+    {
+      "epoch": 0.5006075334143378,
+      "grad_norm": 0.9707381129264832,
+      "learning_rate": 5e-05,
+      "loss": 0.0538,
+      "step": 206
+    },
+    {
+      "epoch": 0.503037667071689,
+      "grad_norm": 0.9181280136108398,
+      "learning_rate": 5e-05,
+      "loss": 0.2319,
+      "step": 207
+    },
+    {
+      "epoch": 0.5054678007290401,
+      "grad_norm": 1.059377670288086,
+      "learning_rate": 5e-05,
+      "loss": 0.2679,
+      "step": 208
+    },
+    {
+      "epoch": 0.5078979343863913,
+      "grad_norm": 0.6514914035797119,
+      "learning_rate": 5e-05,
+      "loss": 0.0623,
+      "step": 209
+    },
+    {
+      "epoch": 0.5103280680437424,
+      "grad_norm": 0.7265043258666992,
+      "learning_rate": 5e-05,
+      "loss": 0.1034,
+      "step": 210
+    },
+    {
+      "epoch": 0.5127582017010935,
+      "grad_norm": 1.5795567035675049,
+      "learning_rate": 5e-05,
+      "loss": 0.1236,
+      "step": 211
+    },
+    {
+      "epoch": 0.5151883353584447,
+      "grad_norm": 0.8533775806427002,
+      "learning_rate": 5e-05,
+      "loss": 0.0702,
+      "step": 212
+    },
+    {
+      "epoch": 0.5176184690157959,
+      "grad_norm": 0.7225982546806335,
+      "learning_rate": 5e-05,
+      "loss": 0.2005,
+      "step": 213
+    },
+    {
+      "epoch": 0.520048602673147,
+      "grad_norm": 0.9016891121864319,
+      "learning_rate": 5e-05,
+      "loss": 0.1329,
+      "step": 214
+    },
+    {
+      "epoch": 0.5224787363304981,
+      "grad_norm": 1.1623356342315674,
+      "learning_rate": 5e-05,
+      "loss": 0.1728,
+      "step": 215
+    },
+    {
+      "epoch": 0.5249088699878494,
+      "grad_norm": 1.1869186162948608,
+      "learning_rate": 5e-05,
+      "loss": 0.2517,
+      "step": 216
+    },
+    {
+      "epoch": 0.5273390036452005,
+      "grad_norm": 1.2271298170089722,
+      "learning_rate": 5e-05,
+      "loss": 0.1402,
+      "step": 217
+    },
+    {
+      "epoch": 0.5297691373025516,
+      "grad_norm": 1.182732105255127,
+      "learning_rate": 5e-05,
+      "loss": 0.1341,
+      "step": 218
+    },
+    {
+      "epoch": 0.5321992709599028,
+      "grad_norm": 0.8633968234062195,
+      "learning_rate": 5e-05,
+      "loss": 0.1469,
+      "step": 219
+    },
+    {
+      "epoch": 0.534629404617254,
+      "grad_norm": 0.9293995499610901,
+      "learning_rate": 5e-05,
+      "loss": 0.1121,
+      "step": 220
+    },
+    {
+      "epoch": 0.5370595382746051,
+      "grad_norm": 1.3438748121261597,
+      "learning_rate": 5e-05,
+      "loss": 0.1718,
+      "step": 221
+    },
+    {
+      "epoch": 0.5394896719319563,
+      "grad_norm": 1.2411259412765503,
+      "learning_rate": 5e-05,
+      "loss": 0.1304,
+      "step": 222
+    },
+    {
+      "epoch": 0.5419198055893074,
+      "grad_norm": 0.8153496384620667,
+      "learning_rate": 5e-05,
+      "loss": 0.1334,
+      "step": 223
+    },
+    {
+      "epoch": 0.5443499392466585,
+      "grad_norm": 0.7595478892326355,
+      "learning_rate": 5e-05,
+      "loss": 0.0994,
+      "step": 224
+    },
+    {
+      "epoch": 0.5467800729040098,
+      "grad_norm": 1.0710370540618896,
+      "learning_rate": 5e-05,
+      "loss": 0.0351,
+      "step": 225
+    },
+    {
+      "epoch": 0.5492102065613609,
+      "grad_norm": 1.2209466695785522,
+      "learning_rate": 5e-05,
+      "loss": 0.1049,
+      "step": 226
+    },
+    {
+      "epoch": 0.551640340218712,
+      "grad_norm": 0.5865522623062134,
+      "learning_rate": 5e-05,
+      "loss": 0.0948,
+      "step": 227
+    },
+    {
+      "epoch": 0.5540704738760632,
+      "grad_norm": 0.8379694223403931,
+      "learning_rate": 5e-05,
+      "loss": 0.1269,
+      "step": 228
+    },
+    {
+      "epoch": 0.5565006075334143,
+      "grad_norm": 0.7321844100952148,
+      "learning_rate": 5e-05,
+      "loss": 0.0912,
+      "step": 229
+    },
+    {
+      "epoch": 0.5589307411907655,
+      "grad_norm": 1.0253808498382568,
+      "learning_rate": 5e-05,
+      "loss": 0.0811,
+      "step": 230
+    },
+    {
+      "epoch": 0.5613608748481167,
+      "grad_norm": 0.9015201330184937,
+      "learning_rate": 5e-05,
+      "loss": 0.0725,
+      "step": 231
+    },
+    {
+      "epoch": 0.5637910085054678,
+      "grad_norm": 0.8760166168212891,
+      "learning_rate": 5e-05,
+      "loss": 0.1727,
+      "step": 232
+    },
+    {
+      "epoch": 0.5662211421628189,
+      "grad_norm": 0.892677903175354,
+      "learning_rate": 5e-05,
+      "loss": 0.084,
+      "step": 233
+    },
+    {
+      "epoch": 0.56865127582017,
+      "grad_norm": 0.6722489595413208,
+      "learning_rate": 5e-05,
+      "loss": 0.0845,
+      "step": 234
+    },
+    {
+      "epoch": 0.5710814094775213,
+      "grad_norm": 0.6187300682067871,
+      "learning_rate": 5e-05,
+      "loss": 0.1269,
+      "step": 235
+    },
+    {
+      "epoch": 0.5735115431348724,
+      "grad_norm": 0.8674693703651428,
+      "learning_rate": 5e-05,
+      "loss": 0.09,
+      "step": 236
+    },
+    {
+      "epoch": 0.5759416767922235,
+      "grad_norm": 0.9376964569091797,
+      "learning_rate": 5e-05,
+      "loss": 0.1916,
+      "step": 237
+    },
+    {
+      "epoch": 0.5783718104495748,
+      "grad_norm": 0.6777618527412415,
+      "learning_rate": 5e-05,
+      "loss": 0.1321,
+      "step": 238
+    },
+    {
+      "epoch": 0.5808019441069259,
+      "grad_norm": 1.1223134994506836,
+      "learning_rate": 5e-05,
+      "loss": 0.1811,
+      "step": 239
+    },
+    {
+      "epoch": 0.583232077764277,
+      "grad_norm": 0.8348031044006348,
+      "learning_rate": 5e-05,
+      "loss": 0.1508,
+      "step": 240
+    },
+    {
+      "epoch": 0.5856622114216282,
+      "grad_norm": 0.7387763857841492,
+      "learning_rate": 5e-05,
+      "loss": 0.058,
+      "step": 241
+    },
+    {
+      "epoch": 0.5880923450789793,
+      "grad_norm": 0.6327199339866638,
+      "learning_rate": 5e-05,
+      "loss": 0.1096,
+      "step": 242
+    },
+    {
+      "epoch": 0.5905224787363305,
+      "grad_norm": 0.7040870785713196,
+      "learning_rate": 5e-05,
+      "loss": 0.1281,
+      "step": 243
+    },
+    {
+      "epoch": 0.5929526123936817,
+      "grad_norm": 1.2110347747802734,
+      "learning_rate": 5e-05,
+      "loss": 0.1959,
+      "step": 244
+    },
+    {
+      "epoch": 0.5953827460510328,
+      "grad_norm": 0.8480271100997925,
+      "learning_rate": 5e-05,
+      "loss": 0.1234,
+      "step": 245
+    },
+    {
+      "epoch": 0.5978128797083839,
+      "grad_norm": 1.0009665489196777,
+      "learning_rate": 5e-05,
+      "loss": 0.1223,
+      "step": 246
+    },
+    {
+      "epoch": 0.6002430133657352,
+      "grad_norm": 0.7057176828384399,
+      "learning_rate": 5e-05,
+      "loss": 0.1747,
+      "step": 247
+    },
+    {
+      "epoch": 0.6026731470230863,
+      "grad_norm": 0.9945976138114929,
+      "learning_rate": 5e-05,
+      "loss": 0.1766,
+      "step": 248
+    },
+    {
+      "epoch": 0.6051032806804374,
+      "grad_norm": 0.7461872696876526,
+      "learning_rate": 5e-05,
+      "loss": 0.1224,
+      "step": 249
+    },
+    {
+      "epoch": 0.6075334143377886,
+      "grad_norm": 0.5109842419624329,
+      "learning_rate": 5e-05,
+      "loss": 0.1105,
+      "step": 250
+    },
+    {
+      "epoch": 0.6099635479951397,
+      "grad_norm": 0.7398750185966492,
+      "learning_rate": 5e-05,
+      "loss": 0.139,
+      "step": 251
+    },
+    {
+      "epoch": 0.6123936816524909,
+      "grad_norm": 0.7645841836929321,
+      "learning_rate": 5e-05,
+      "loss": 0.11,
+      "step": 252
+    },
+    {
+      "epoch": 0.6148238153098421,
+      "grad_norm": 0.8627939224243164,
+      "learning_rate": 5e-05,
+      "loss": 0.1277,
+      "step": 253
+    },
+    {
+      "epoch": 0.6172539489671932,
+      "grad_norm": 0.6933022737503052,
+      "learning_rate": 5e-05,
+      "loss": 0.1283,
+      "step": 254
+    },
+    {
+      "epoch": 0.6196840826245443,
+      "grad_norm": 0.8254914283752441,
+      "learning_rate": 5e-05,
+      "loss": 0.1087,
+      "step": 255
+    },
+    {
+      "epoch": 0.6221142162818954,
+      "grad_norm": 0.9561761617660522,
+      "learning_rate": 5e-05,
+      "loss": 0.1019,
+      "step": 256
+    },
+    {
+      "epoch": 0.6245443499392467,
+      "grad_norm": 1.0569480657577515,
+      "learning_rate": 5e-05,
+      "loss": 0.1235,
+      "step": 257
+    },
+    {
+      "epoch": 0.6269744835965978,
+      "grad_norm": 1.0708096027374268,
+      "learning_rate": 5e-05,
+      "loss": 0.1797,
+      "step": 258
+    },
+    {
+      "epoch": 0.6294046172539489,
+      "grad_norm": 0.5817265510559082,
+      "learning_rate": 5e-05,
+      "loss": 0.0787,
+      "step": 259
+    },
+    {
+      "epoch": 0.6318347509113001,
+      "grad_norm": 0.8255194425582886,
+      "learning_rate": 5e-05,
+      "loss": 0.1754,
+      "step": 260
+    },
+    {
+      "epoch": 0.6342648845686513,
+      "grad_norm": 0.8279481530189514,
+      "learning_rate": 5e-05,
+      "loss": 0.1668,
+      "step": 261
+    },
+    {
+      "epoch": 0.6366950182260024,
+      "grad_norm": 0.6393010020256042,
+      "learning_rate": 5e-05,
+      "loss": 0.0756,
+      "step": 262
+    },
+    {
+      "epoch": 0.6391251518833536,
+      "grad_norm": 0.49387434124946594,
+      "learning_rate": 5e-05,
+      "loss": 0.0757,
+      "step": 263
+    },
+    {
+      "epoch": 0.6415552855407047,
+      "grad_norm": 0.5705033540725708,
+      "learning_rate": 5e-05,
+      "loss": 0.0949,
+      "step": 264
+    },
+    {
+      "epoch": 0.6439854191980559,
+      "grad_norm": 0.9022333025932312,
+      "learning_rate": 5e-05,
+      "loss": 0.1716,
+      "step": 265
+    },
+    {
+      "epoch": 0.6464155528554071,
+      "grad_norm": 0.5299041271209717,
+      "learning_rate": 5e-05,
+      "loss": 0.0799,
+      "step": 266
+    },
+    {
+      "epoch": 0.6488456865127582,
+      "grad_norm": 0.7919138669967651,
+      "learning_rate": 5e-05,
+      "loss": 0.1476,
+      "step": 267
+    },
+    {
+      "epoch": 0.6512758201701093,
+      "grad_norm": 1.394972324371338,
+      "learning_rate": 5e-05,
+      "loss": 0.112,
+      "step": 268
+    },
+    {
+      "epoch": 0.6537059538274606,
+      "grad_norm": 0.9046385884284973,
+      "learning_rate": 5e-05,
+      "loss": 0.1683,
+      "step": 269
+    },
+    {
+      "epoch": 0.6561360874848117,
+      "grad_norm": 0.7190728783607483,
+      "learning_rate": 5e-05,
+      "loss": 0.1434,
+      "step": 270
+    },
+    {
+      "epoch": 0.6585662211421628,
+      "grad_norm": 1.0634692907333374,
+      "learning_rate": 5e-05,
+      "loss": 0.1244,
+      "step": 271
+    },
+    {
+      "epoch": 0.660996354799514,
+      "grad_norm": 0.569398045539856,
+      "learning_rate": 5e-05,
+      "loss": 0.0728,
+      "step": 272
+    },
+    {
+      "epoch": 0.6634264884568651,
+      "grad_norm": 0.9083871841430664,
+      "learning_rate": 5e-05,
+      "loss": 0.1467,
+      "step": 273
+    },
+    {
+      "epoch": 0.6658566221142163,
+      "grad_norm": 0.7927613854408264,
+      "learning_rate": 5e-05,
+      "loss": 0.1832,
+      "step": 274
+    },
+    {
+      "epoch": 0.6682867557715675,
+      "grad_norm": 1.0088517665863037,
+      "learning_rate": 5e-05,
+      "loss": 0.1422,
+      "step": 275
+    },
+    {
+      "epoch": 0.6707168894289186,
+      "grad_norm": 0.5374720692634583,
+      "learning_rate": 5e-05,
+      "loss": 0.0971,
+      "step": 276
+    },
+    {
+      "epoch": 0.6731470230862697,
+      "grad_norm": 0.7200155854225159,
+      "learning_rate": 5e-05,
+      "loss": 0.0722,
+      "step": 277
+    },
+    {
+      "epoch": 0.675577156743621,
+      "grad_norm": 0.7518504858016968,
+      "learning_rate": 5e-05,
+      "loss": 0.126,
+      "step": 278
+    },
+    {
+      "epoch": 0.6780072904009721,
+      "grad_norm": 0.8544089794158936,
+      "learning_rate": 5e-05,
+      "loss": 0.113,
+      "step": 279
+    },
+    {
+      "epoch": 0.6804374240583232,
+      "grad_norm": 0.7537454962730408,
+      "learning_rate": 5e-05,
+      "loss": 0.0615,
+      "step": 280
+    },
+    {
+      "epoch": 0.6828675577156743,
+      "grad_norm": 0.8440731763839722,
+      "learning_rate": 5e-05,
+      "loss": 0.0933,
+      "step": 281
+    },
+    {
+      "epoch": 0.6852976913730255,
+      "grad_norm": 1.7643004655838013,
+      "learning_rate": 5e-05,
+      "loss": 0.1807,
+      "step": 282
+    },
+    {
+      "epoch": 0.6877278250303767,
+      "grad_norm": 0.6502506732940674,
+      "learning_rate": 5e-05,
+      "loss": 0.1189,
+      "step": 283
+    },
+    {
+      "epoch": 0.6901579586877278,
+      "grad_norm": 0.8678569197654724,
+      "learning_rate": 5e-05,
+      "loss": 0.094,
+      "step": 284
+    },
+    {
+      "epoch": 0.692588092345079,
+      "grad_norm": 0.8592473864555359,
+      "learning_rate": 5e-05,
+      "loss": 0.1197,
+      "step": 285
+    },
+    {
+      "epoch": 0.6950182260024301,
+      "grad_norm": 0.8664891719818115,
+      "learning_rate": 5e-05,
+      "loss": 0.0575,
+      "step": 286
+    },
+    {
+      "epoch": 0.6974483596597812,
+      "grad_norm": 0.5715927481651306,
+      "learning_rate": 5e-05,
+      "loss": 0.0668,
+      "step": 287
+    },
+    {
+      "epoch": 0.6998784933171325,
+      "grad_norm": 1.0751949548721313,
+      "learning_rate": 5e-05,
+      "loss": 0.1457,
+      "step": 288
+    },
+    {
+      "epoch": 0.7023086269744836,
+      "grad_norm": 0.8039306402206421,
+      "learning_rate": 5e-05,
+      "loss": 0.1206,
+      "step": 289
+    },
+    {
+      "epoch": 0.7047387606318347,
+      "grad_norm": 1.5916086435317993,
+      "learning_rate": 5e-05,
+      "loss": 0.1471,
+      "step": 290
+    },
+    {
+      "epoch": 0.707168894289186,
+      "grad_norm": 0.7829633355140686,
+      "learning_rate": 5e-05,
+      "loss": 0.106,
+      "step": 291
+    },
+    {
+      "epoch": 0.7095990279465371,
+      "grad_norm": 0.8277871608734131,
+      "learning_rate": 5e-05,
+      "loss": 0.0347,
+      "step": 292
+    },
+    {
+      "epoch": 0.7120291616038882,
+      "grad_norm": 0.9798584580421448,
+      "learning_rate": 5e-05,
+      "loss": 0.1231,
+      "step": 293
+    },
+    {
+      "epoch": 0.7144592952612394,
+      "grad_norm": 0.6273385286331177,
+      "learning_rate": 5e-05,
+      "loss": 0.0838,
+      "step": 294
+    },
+    {
+      "epoch": 0.7168894289185905,
+      "grad_norm": 1.2111278772354126,
+      "learning_rate": 5e-05,
+      "loss": 0.1971,
+      "step": 295
+    },
+    {
+      "epoch": 0.7193195625759417,
+      "grad_norm": 0.5927327275276184,
+      "learning_rate": 5e-05,
+      "loss": 0.1198,
+      "step": 296
+    },
+    {
+      "epoch": 0.7217496962332929,
+      "grad_norm": 0.672064483165741,
+      "learning_rate": 5e-05,
+      "loss": 0.1295,
+      "step": 297
+    },
+    {
+      "epoch": 0.724179829890644,
+      "grad_norm": 0.6398962736129761,
+      "learning_rate": 5e-05,
+      "loss": 0.0933,
+      "step": 298
+    },
+    {
+      "epoch": 0.7266099635479951,
+      "grad_norm": 0.7458367943763733,
+      "learning_rate": 5e-05,
+      "loss": 0.1407,
+      "step": 299
+    },
+    {
+      "epoch": 0.7290400972053463,
+      "grad_norm": 0.6593539118766785,
+      "learning_rate": 5e-05,
+      "loss": 0.1055,
+      "step": 300
+    },
+    {
+      "epoch": 0.7290400972053463,
+      "eval_loss": 0.24557192623615265,
+      "eval_runtime": 513.4098,
+      "eval_samples_per_second": 5.352,
+      "eval_steps_per_second": 0.67,
+      "step": 300
+    },
+    {
+      "epoch": 0.7314702308626975,
+      "grad_norm": 0.5555610656738281,
+      "learning_rate": 5e-05,
+      "loss": 0.1641,
+      "step": 301
+    },
+    {
+      "epoch": 0.7339003645200486,
+      "grad_norm": 0.49084270000457764,
+      "learning_rate": 5e-05,
+      "loss": 0.0786,
+      "step": 302
+    },
+    {
+      "epoch": 0.7363304981773997,
+      "grad_norm": 0.9267096519470215,
+      "learning_rate": 5e-05,
+      "loss": 0.2053,
+      "step": 303
+    },
+    {
+      "epoch": 0.7387606318347509,
+      "grad_norm": 0.440594345331192,
+      "learning_rate": 5e-05,
+      "loss": 0.1224,
+      "step": 304
+    },
+    {
+      "epoch": 0.741190765492102,
+      "grad_norm": 1.0237218141555786,
+      "learning_rate": 5e-05,
+      "loss": 0.1263,
+      "step": 305
+    },
+    {
+      "epoch": 0.7436208991494532,
+      "grad_norm": 1.0087871551513672,
+      "learning_rate": 5e-05,
+      "loss": 0.1218,
+      "step": 306
+    },
+    {
+      "epoch": 0.7460510328068044,
+      "grad_norm": 0.8280074000358582,
+      "learning_rate": 5e-05,
+      "loss": 0.1851,
+      "step": 307
+    },
+    {
+      "epoch": 0.7484811664641555,
+      "grad_norm": 0.681713342666626,
+      "learning_rate": 5e-05,
+      "loss": 0.1757,
+      "step": 308
+    },
+    {
+      "epoch": 0.7509113001215066,
+      "grad_norm": 4.372740268707275,
+      "learning_rate": 5e-05,
+      "loss": 0.104,
+      "step": 309
+    },
+    {
+      "epoch": 0.7533414337788579,
+      "grad_norm": 0.8002400398254395,
+      "learning_rate": 5e-05,
+      "loss": 0.0922,
+      "step": 310
+    },
+    {
+      "epoch": 0.755771567436209,
+      "grad_norm": 0.7685486674308777,
+      "learning_rate": 5e-05,
+      "loss": 0.1391,
+      "step": 311
+    },
+    {
+      "epoch": 0.7582017010935601,
+      "grad_norm": 1.282363772392273,
+      "learning_rate": 5e-05,
+      "loss": 0.1141,
+      "step": 312
+    },
+    {
+      "epoch": 0.7606318347509113,
+      "grad_norm": 0.9099876284599304,
+      "learning_rate": 5e-05,
+      "loss": 0.1396,
+      "step": 313
+    },
+    {
+      "epoch": 0.7630619684082625,
+      "grad_norm": 0.5911856293678284,
+      "learning_rate": 5e-05,
+      "loss": 0.0631,
+      "step": 314
+    },
+    {
+      "epoch": 0.7654921020656136,
+      "grad_norm": 0.7148822546005249,
+      "learning_rate": 5e-05,
+      "loss": 0.1714,
+      "step": 315
+    },
+    {
+      "epoch": 0.7679222357229648,
+      "grad_norm": 0.6756410002708435,
+      "learning_rate": 5e-05,
+      "loss": 0.0894,
+      "step": 316
+    },
+    {
+      "epoch": 0.7703523693803159,
+      "grad_norm": 0.8938581347465515,
+      "learning_rate": 5e-05,
+      "loss": 0.1538,
+      "step": 317
+    },
+    {
+      "epoch": 0.772782503037667,
+      "grad_norm": 1.061012625694275,
+      "learning_rate": 5e-05,
+      "loss": 0.1069,
+      "step": 318
+    },
+    {
+      "epoch": 0.7752126366950183,
+      "grad_norm": 0.8262349963188171,
+      "learning_rate": 5e-05,
+      "loss": 0.0745,
+      "step": 319
+    },
+    {
+      "epoch": 0.7776427703523694,
+      "grad_norm": 1.3609825372695923,
+      "learning_rate": 5e-05,
+      "loss": 0.0689,
+      "step": 320
+    },
+    {
+      "epoch": 0.7800729040097205,
+      "grad_norm": 1.2151660919189453,
+      "learning_rate": 5e-05,
+      "loss": 0.0863,
+      "step": 321
+    },
+    {
+      "epoch": 0.7825030376670717,
+      "grad_norm": 0.7165310978889465,
+      "learning_rate": 5e-05,
+      "loss": 0.084,
+      "step": 322
+    },
+    {
+      "epoch": 0.7849331713244229,
+      "grad_norm": 1.2151451110839844,
+      "learning_rate": 5e-05,
+      "loss": 0.1229,
+      "step": 323
+    },
+    {
+      "epoch": 0.787363304981774,
+      "grad_norm": 0.9600043296813965,
+      "learning_rate": 5e-05,
+      "loss": 0.0524,
+      "step": 324
+    },
+    {
+      "epoch": 0.7897934386391251,
+      "grad_norm": 1.1785893440246582,
+      "learning_rate": 5e-05,
+      "loss": 0.1635,
+      "step": 325
+    },
+    {
+      "epoch": 0.7922235722964763,
+      "grad_norm": 0.6502353549003601,
+      "learning_rate": 5e-05,
+      "loss": 0.103,
+      "step": 326
+    },
+    {
+      "epoch": 0.7946537059538274,
+      "grad_norm": 0.6894869208335876,
+      "learning_rate": 5e-05,
+      "loss": 0.0824,
+      "step": 327
+    },
+    {
+      "epoch": 0.7970838396111786,
+      "grad_norm": 0.8827742338180542,
+      "learning_rate": 5e-05,
+      "loss": 0.0899,
+      "step": 328
+    },
+    {
+      "epoch": 0.7995139732685298,
+      "grad_norm": 0.82513827085495,
+      "learning_rate": 5e-05,
+      "loss": 0.0642,
+      "step": 329
+    },
+    {
+      "epoch": 0.8019441069258809,
+      "grad_norm": 0.7896872758865356,
+      "learning_rate": 5e-05,
+      "loss": 0.1869,
+      "step": 330
+    },
+    {
+      "epoch": 0.804374240583232,
+      "grad_norm": 1.421380639076233,
+      "learning_rate": 5e-05,
+      "loss": 0.0668,
+      "step": 331
+    },
+    {
+      "epoch": 0.8068043742405833,
+      "grad_norm": 0.826987087726593,
+      "learning_rate": 5e-05,
+      "loss": 0.1543,
+      "step": 332
+    },
+    {
+      "epoch": 0.8092345078979344,
+      "grad_norm": 0.7504023909568787,
+      "learning_rate": 5e-05,
+      "loss": 0.1167,
+      "step": 333
+    },
+    {
+      "epoch": 0.8116646415552855,
+      "grad_norm": 0.9217634797096252,
+      "learning_rate": 5e-05,
+      "loss": 0.0801,
+      "step": 334
+    },
+    {
+      "epoch": 0.8140947752126367,
+      "grad_norm": 0.6423681974411011,
+      "learning_rate": 5e-05,
+      "loss": 0.0445,
+      "step": 335
+    },
+    {
+      "epoch": 0.8165249088699879,
+      "grad_norm": 0.3945442736148834,
+      "learning_rate": 5e-05,
+      "loss": 0.0737,
+      "step": 336
+    },
+    {
+      "epoch": 0.818955042527339,
+      "grad_norm": 0.6396763324737549,
+      "learning_rate": 5e-05,
+      "loss": 0.1118,
+      "step": 337
+    },
+    {
+      "epoch": 0.8213851761846902,
+      "grad_norm": 1.31553053855896,
+      "learning_rate": 5e-05,
+      "loss": 0.0989,
+      "step": 338
+    },
+    {
+      "epoch": 0.8238153098420413,
+      "grad_norm": 0.962189793586731,
+      "learning_rate": 5e-05,
+      "loss": 0.1043,
+      "step": 339
+    },
+    {
+      "epoch": 0.8262454434993924,
+      "grad_norm": 0.9305171370506287,
+      "learning_rate": 5e-05,
+      "loss": 0.1333,
+      "step": 340
+    },
+    {
+      "epoch": 0.8286755771567437,
+      "grad_norm": 0.6102299690246582,
+      "learning_rate": 5e-05,
+      "loss": 0.0743,
+      "step": 341
+    },
+    {
+      "epoch": 0.8311057108140948,
+      "grad_norm": 0.9117159247398376,
+      "learning_rate": 5e-05,
+      "loss": 0.0399,
+      "step": 342
+    },
+    {
+      "epoch": 0.8335358444714459,
+      "grad_norm": 0.614378035068512,
+      "learning_rate": 5e-05,
+      "loss": 0.0767,
+      "step": 343
+    },
+    {
+      "epoch": 0.8359659781287971,
+      "grad_norm": 0.934684157371521,
+      "learning_rate": 5e-05,
+      "loss": 0.0814,
+      "step": 344
+    },
+    {
+      "epoch": 0.8383961117861483,
+      "grad_norm": 0.8567538261413574,
+      "learning_rate": 5e-05,
+      "loss": 0.0866,
+      "step": 345
+    },
+    {
+      "epoch": 0.8408262454434994,
+      "grad_norm": 1.083308219909668,
+      "learning_rate": 5e-05,
+      "loss": 0.0534,
+      "step": 346
+    },
+    {
+      "epoch": 0.8432563791008505,
+      "grad_norm": 0.8664528727531433,
+      "learning_rate": 5e-05,
+      "loss": 0.1348,
+      "step": 347
+    },
+    {
+      "epoch": 0.8456865127582017,
+      "grad_norm": 0.7151762843132019,
+      "learning_rate": 5e-05,
+      "loss": 0.0615,
+      "step": 348
+    },
+    {
+      "epoch": 0.8481166464155528,
+      "grad_norm": 0.784515917301178,
+      "learning_rate": 5e-05,
+      "loss": 0.1014,
+      "step": 349
+    },
+    {
+      "epoch": 0.850546780072904,
+      "grad_norm": 0.7713536620140076,
+      "learning_rate": 5e-05,
+      "loss": 0.0684,
+      "step": 350
+    },
+    {
+      "epoch": 0.8529769137302552,
+      "grad_norm": 0.8839780688285828,
+      "learning_rate": 5e-05,
+      "loss": 0.2673,
+      "step": 351
+    },
+    {
+      "epoch": 0.8554070473876063,
+      "grad_norm": 0.6095230579376221,
+      "learning_rate": 5e-05,
+      "loss": 0.1667,
+      "step": 352
+    },
+    {
+      "epoch": 0.8578371810449574,
+      "grad_norm": 1.1133002042770386,
+      "learning_rate": 5e-05,
+      "loss": 0.1746,
+      "step": 353
+    },
+    {
+      "epoch": 0.8602673147023087,
+      "grad_norm": 0.5206813812255859,
+      "learning_rate": 5e-05,
+      "loss": 0.1507,
+      "step": 354
+    },
+    {
+      "epoch": 0.8626974483596598,
+      "grad_norm": 0.6527222990989685,
+      "learning_rate": 5e-05,
+      "loss": 0.1045,
+      "step": 355
+    },
+    {
+      "epoch": 0.8651275820170109,
+      "grad_norm": 0.6029911637306213,
+      "learning_rate": 5e-05,
+      "loss": 0.1157,
+      "step": 356
+    },
+    {
+      "epoch": 0.8675577156743621,
+      "grad_norm": 0.7916355729103088,
+      "learning_rate": 5e-05,
+      "loss": 0.0817,
+      "step": 357
+    },
+    {
+      "epoch": 0.8699878493317132,
+      "grad_norm": 0.6410179734230042,
+      "learning_rate": 5e-05,
+      "loss": 0.0707,
+      "step": 358
+    },
+    {
+      "epoch": 0.8724179829890644,
+      "grad_norm": 1.0566637516021729,
+      "learning_rate": 5e-05,
+      "loss": 0.0774,
+      "step": 359
+    },
+    {
+      "epoch": 0.8748481166464156,
+      "grad_norm": 0.6340796947479248,
+      "learning_rate": 5e-05,
+      "loss": 0.1003,
+      "step": 360
+    },
+    {
+      "epoch": 0.8772782503037667,
+      "grad_norm": 0.6086530089378357,
+      "learning_rate": 5e-05,
+      "loss": 0.0441,
+      "step": 361
+    },
+    {
+      "epoch": 0.8797083839611178,
+      "grad_norm": 0.8955773115158081,
+      "learning_rate": 5e-05,
+      "loss": 0.1389,
+      "step": 362
+    },
+    {
+      "epoch": 0.8821385176184691,
+      "grad_norm": 0.4679882228374481,
+      "learning_rate": 5e-05,
+      "loss": 0.0708,
+      "step": 363
+    },
+    {
+      "epoch": 0.8845686512758202,
+      "grad_norm": 0.6539987921714783,
+      "learning_rate": 5e-05,
+      "loss": 0.1078,
+      "step": 364
+    },
+    {
+      "epoch": 0.8869987849331713,
+      "grad_norm": 0.8322189450263977,
+      "learning_rate": 5e-05,
+      "loss": 0.0477,
+      "step": 365
+    },
+    {
+      "epoch": 0.8894289185905225,
+      "grad_norm": 0.5329148769378662,
+      "learning_rate": 5e-05,
+      "loss": 0.0638,
+      "step": 366
+    },
+    {
+      "epoch": 0.8918590522478737,
+      "grad_norm": 0.7909948825836182,
+      "learning_rate": 5e-05,
+      "loss": 0.0941,
+      "step": 367
+    },
+    {
+      "epoch": 0.8942891859052248,
+      "grad_norm": 0.6703280806541443,
+      "learning_rate": 5e-05,
+      "loss": 0.0989,
+      "step": 368
+    },
+    {
+      "epoch": 0.8967193195625759,
+      "grad_norm": 0.6218695640563965,
+      "learning_rate": 5e-05,
+      "loss": 0.0712,
+      "step": 369
+    },
+    {
+      "epoch": 0.8991494532199271,
+      "grad_norm": 1.6482758522033691,
+      "learning_rate": 5e-05,
+      "loss": 0.1537,
+      "step": 370
+    },
+    {
+      "epoch": 0.9015795868772782,
+      "grad_norm": 0.7616936564445496,
+      "learning_rate": 5e-05,
+      "loss": 0.1234,
+      "step": 371
+    },
+    {
+      "epoch": 0.9040097205346294,
+      "grad_norm": 0.7617467045783997,
+      "learning_rate": 5e-05,
+      "loss": 0.0895,
+      "step": 372
+    },
+    {
+      "epoch": 0.9064398541919806,
+      "grad_norm": 0.7406112551689148,
+      "learning_rate": 5e-05,
+      "loss": 0.0559,
+      "step": 373
+    },
+    {
+      "epoch": 0.9088699878493317,
+      "grad_norm": 1.1627577543258667,
+      "learning_rate": 5e-05,
+      "loss": 0.0807,
+      "step": 374
+    },
+    {
+      "epoch": 0.9113001215066828,
+      "grad_norm": 0.6091077923774719,
+      "learning_rate": 5e-05,
+      "loss": 0.0649,
+      "step": 375
+    },
+    {
+      "epoch": 0.913730255164034,
+      "grad_norm": 0.8935027122497559,
+      "learning_rate": 5e-05,
+      "loss": 0.1563,
+      "step": 376
+    },
+    {
+      "epoch": 0.9161603888213852,
+      "grad_norm": 0.8287168741226196,
+      "learning_rate": 5e-05,
+      "loss": 0.0725,
+      "step": 377
+    },
+    {
+      "epoch": 0.9185905224787363,
+      "grad_norm": 0.5430625677108765,
+      "learning_rate": 5e-05,
+      "loss": 0.0984,
+      "step": 378
+    },
+    {
+      "epoch": 0.9210206561360875,
+      "grad_norm": 0.7538096308708191,
+      "learning_rate": 5e-05,
+      "loss": 0.1231,
+      "step": 379
+    },
+    {
+      "epoch": 0.9234507897934386,
+      "grad_norm": 0.9483656287193298,
+      "learning_rate": 5e-05,
+      "loss": 0.1341,
+      "step": 380
+    },
+    {
+      "epoch": 0.9258809234507898,
+      "grad_norm": 0.5689654350280762,
+      "learning_rate": 5e-05,
+      "loss": 0.1116,
+      "step": 381
+    },
+    {
+      "epoch": 0.928311057108141,
+      "grad_norm": 0.987369179725647,
+      "learning_rate": 5e-05,
+      "loss": 0.1815,
+      "step": 382
+    },
+    {
+      "epoch": 0.9307411907654921,
+      "grad_norm": 0.872460126876831,
+      "learning_rate": 5e-05,
+      "loss": 0.0994,
+      "step": 383
+    },
+    {
+      "epoch": 0.9331713244228432,
+      "grad_norm": 0.785040020942688,
+      "learning_rate": 5e-05,
+      "loss": 0.1006,
+      "step": 384
+    },
+    {
+      "epoch": 0.9356014580801945,
+      "grad_norm": 0.6228112578392029,
+      "learning_rate": 5e-05,
+      "loss": 0.0805,
+      "step": 385
+    },
+    {
+      "epoch": 0.9380315917375456,
+      "grad_norm": 0.746306836605072,
+      "learning_rate": 5e-05,
+      "loss": 0.0672,
+      "step": 386
+    },
+    {
+      "epoch": 0.9404617253948967,
+      "grad_norm": 0.44906482100486755,
+      "learning_rate": 5e-05,
+      "loss": 0.0831,
+      "step": 387
+    },
+    {
+      "epoch": 0.9428918590522479,
+      "grad_norm": 0.7880069613456726,
+      "learning_rate": 5e-05,
+      "loss": 0.0871,
+      "step": 388
+    },
+    {
+      "epoch": 0.945321992709599,
+      "grad_norm": 0.7579517960548401,
+      "learning_rate": 5e-05,
+      "loss": 0.1306,
+      "step": 389
+    },
+    {
+      "epoch": 0.9477521263669502,
+      "grad_norm": 0.5668421387672424,
+      "learning_rate": 5e-05,
+      "loss": 0.0738,
+      "step": 390
+    },
+    {
+      "epoch": 0.9501822600243013,
+      "grad_norm": 0.8249059319496155,
+      "learning_rate": 5e-05,
+      "loss": 0.069,
+      "step": 391
+    },
+    {
+      "epoch": 0.9526123936816525,
+      "grad_norm": 0.7063718438148499,
+      "learning_rate": 5e-05,
+      "loss": 0.0488,
+      "step": 392
+    },
+    {
+      "epoch": 0.9550425273390036,
+      "grad_norm": 0.9006249904632568,
+      "learning_rate": 5e-05,
+      "loss": 0.029,
+      "step": 393
+    },
+    {
+      "epoch": 0.9574726609963548,
+      "grad_norm": 0.5778633952140808,
+      "learning_rate": 5e-05,
+      "loss": 0.0859,
+      "step": 394
+    },
+    {
+      "epoch": 0.959902794653706,
+      "grad_norm": 0.9541341066360474,
+      "learning_rate": 5e-05,
+      "loss": 0.0698,
+      "step": 395
+    },
+    {
+      "epoch": 0.9623329283110571,
+      "grad_norm": 0.7825549244880676,
+      "learning_rate": 5e-05,
+      "loss": 0.0714,
+      "step": 396
+    },
+    {
+      "epoch": 0.9647630619684082,
+      "grad_norm": 0.729721188545227,
+      "learning_rate": 5e-05,
+      "loss": 0.0651,
+      "step": 397
+    },
+    {
+      "epoch": 0.9671931956257594,
+      "grad_norm": 0.7052625417709351,
+      "learning_rate": 5e-05,
+      "loss": 0.1518,
+      "step": 398
+    },
+    {
+      "epoch": 0.9696233292831106,
+      "grad_norm": 2.3800599575042725,
+      "learning_rate": 5e-05,
+      "loss": 0.0919,
+      "step": 399
+    },
+    {
+      "epoch": 0.9720534629404617,
+      "grad_norm": 0.5317090749740601,
+      "learning_rate": 5e-05,
+      "loss": 0.0828,
+      "step": 400
+    },
+    {
+      "epoch": 0.9720534629404617,
+      "eval_loss": 0.2335556447505951,
+      "eval_runtime": 513.3338,
+      "eval_samples_per_second": 5.353,
+      "eval_steps_per_second": 0.67,
+      "step": 400
+    },
+    {
+      "epoch": 0.9744835965978129,
+      "grad_norm": 0.6768686771392822,
+      "learning_rate": 5e-05,
+      "loss": 0.1675,
+      "step": 401
+    },
+    {
+      "epoch": 0.976913730255164,
+      "grad_norm": 0.5037873983383179,
+      "learning_rate": 5e-05,
+      "loss": 0.0199,
+      "step": 402
+    },
+    {
+      "epoch": 0.9793438639125152,
+      "grad_norm": 1.7766000032424927,
+      "learning_rate": 5e-05,
+      "loss": 0.0846,
+      "step": 403
+    },
+    {
+      "epoch": 0.9817739975698664,
+      "grad_norm": 0.978345513343811,
+      "learning_rate": 5e-05,
+      "loss": 0.0593,
+      "step": 404
+    },
+    {
+      "epoch": 0.9842041312272175,
+      "grad_norm": 1.4286214113235474,
+      "learning_rate": 5e-05,
+      "loss": 0.1963,
+      "step": 405
+    },
+    {
+      "epoch": 0.9866342648845686,
+      "grad_norm": 1.1499220132827759,
+      "learning_rate": 5e-05,
+      "loss": 0.0786,
+      "step": 406
+    },
+    {
+      "epoch": 0.9890643985419199,
+      "grad_norm": 1.4914474487304688,
+      "learning_rate": 5e-05,
+      "loss": 0.1152,
+      "step": 407
+    },
+    {
+      "epoch": 0.991494532199271,
+      "grad_norm": 0.7232803106307983,
+      "learning_rate": 5e-05,
+      "loss": 0.1338,
+      "step": 408
+    },
+    {
+      "epoch": 0.9939246658566221,
+      "grad_norm": 0.8456845283508301,
+      "learning_rate": 5e-05,
+      "loss": 0.0631,
+      "step": 409
+    },
+    {
+      "epoch": 0.9963547995139733,
+      "grad_norm": 0.5165804028511047,
+      "learning_rate": 5e-05,
+      "loss": 0.0725,
+      "step": 410
+    },
+    {
+      "epoch": 0.9987849331713244,
+      "grad_norm": 0.8364010453224182,
+      "learning_rate": 5e-05,
+      "loss": 0.1215,
+      "step": 411
+    },
+    {
+      "epoch": 1.0012150668286757,
+      "grad_norm": 0.6961699724197388,
+      "learning_rate": 5e-05,
+      "loss": 0.0327,
+      "step": 412
+    },
+    {
+      "epoch": 1.0036452004860268,
+      "grad_norm": 0.9358980655670166,
+      "learning_rate": 5e-05,
+      "loss": 0.0724,
+      "step": 413
+    },
+    {
+      "epoch": 1.006075334143378,
+      "grad_norm": 0.641076922416687,
+      "learning_rate": 5e-05,
+      "loss": 0.0824,
+      "step": 414
+    },
+    {
+      "epoch": 1.008505467800729,
+      "grad_norm": 0.6918608546257019,
+      "learning_rate": 5e-05,
+      "loss": 0.083,
+      "step": 415
+    },
+    {
+      "epoch": 1.0109356014580801,
+      "grad_norm": 0.7481106519699097,
+      "learning_rate": 5e-05,
+      "loss": 0.0887,
+      "step": 416
+    },
+    {
+      "epoch": 1.0133657351154313,
+      "grad_norm": 0.7577409148216248,
+      "learning_rate": 5e-05,
+      "loss": 0.0684,
+      "step": 417
+    },
+    {
+      "epoch": 1.0157958687727826,
+      "grad_norm": 1.0815515518188477,
+      "learning_rate": 5e-05,
+      "loss": 0.1167,
+      "step": 418
+    },
+    {
+      "epoch": 1.0182260024301337,
+      "grad_norm": 1.248225450515747,
+      "learning_rate": 5e-05,
+      "loss": 0.0644,
+      "step": 419
+    },
+    {
+      "epoch": 1.0206561360874848,
+      "grad_norm": 0.7970739006996155,
+      "learning_rate": 5e-05,
+      "loss": 0.1712,
+      "step": 420
+    },
+    {
+      "epoch": 1.023086269744836,
+      "grad_norm": 1.343601942062378,
+      "learning_rate": 5e-05,
+      "loss": 0.1359,
+      "step": 421
+    },
+    {
+      "epoch": 1.025516403402187,
+      "grad_norm": 0.705987274646759,
+      "learning_rate": 5e-05,
+      "loss": 0.0961,
+      "step": 422
+    },
+    {
+      "epoch": 1.0279465370595382,
+      "grad_norm": 0.5520442724227905,
+      "learning_rate": 5e-05,
+      "loss": 0.0839,
+      "step": 423
+    },
+    {
+      "epoch": 1.0303766707168893,
+      "grad_norm": 0.5799084305763245,
+      "learning_rate": 5e-05,
+      "loss": 0.0759,
+      "step": 424
+    },
+    {
+      "epoch": 1.0328068043742407,
+      "grad_norm": 0.610801637172699,
+      "learning_rate": 5e-05,
+      "loss": 0.0593,
+      "step": 425
+    },
+    {
+      "epoch": 1.0352369380315918,
+      "grad_norm": 1.035436987876892,
+      "learning_rate": 5e-05,
+      "loss": 0.194,
+      "step": 426
+    },
+    {
+      "epoch": 1.037667071688943,
+      "grad_norm": 0.9973961114883423,
+      "learning_rate": 5e-05,
+      "loss": 0.0954,
+      "step": 427
+    },
+    {
+      "epoch": 1.040097205346294,
+      "grad_norm": 0.8617413640022278,
+      "learning_rate": 5e-05,
+      "loss": 0.0917,
+      "step": 428
+    },
+    {
+      "epoch": 1.0425273390036451,
+      "grad_norm": 0.9682510495185852,
+      "learning_rate": 5e-05,
+      "loss": 0.1391,
+      "step": 429
+    },
+    {
+      "epoch": 1.0449574726609963,
+      "grad_norm": 0.3044584393501282,
+      "learning_rate": 5e-05,
+      "loss": 0.025,
+      "step": 430
+    },
+    {
+      "epoch": 1.0473876063183476,
+      "grad_norm": 0.6183155179023743,
+      "learning_rate": 5e-05,
+      "loss": 0.0347,
+      "step": 431
+    },
+    {
+      "epoch": 1.0498177399756987,
+      "grad_norm": 0.37442171573638916,
+      "learning_rate": 5e-05,
+      "loss": 0.0452,
+      "step": 432
+    },
+    {
+      "epoch": 1.0522478736330498,
+      "grad_norm": 0.565359890460968,
+      "learning_rate": 5e-05,
+      "loss": 0.0943,
+      "step": 433
+    },
+    {
+      "epoch": 1.054678007290401,
+      "grad_norm": 0.6165122389793396,
+      "learning_rate": 5e-05,
+      "loss": 0.0851,
+      "step": 434
+    },
+    {
+      "epoch": 1.057108140947752,
+      "grad_norm": 0.7150765061378479,
+      "learning_rate": 5e-05,
+      "loss": 0.0704,
+      "step": 435
+    },
+    {
+      "epoch": 1.0595382746051032,
+      "grad_norm": 0.7983779311180115,
+      "learning_rate": 5e-05,
+      "loss": 0.121,
+      "step": 436
+    },
+    {
+      "epoch": 1.0619684082624545,
+      "grad_norm": 0.5900336503982544,
+      "learning_rate": 5e-05,
+      "loss": 0.0938,
+      "step": 437
+    },
+    {
+      "epoch": 1.0643985419198057,
+      "grad_norm": 0.6962295174598694,
+      "learning_rate": 5e-05,
+      "loss": 0.1227,
+      "step": 438
+    },
+    {
+      "epoch": 1.0668286755771568,
+      "grad_norm": 1.0119210481643677,
+      "learning_rate": 5e-05,
+      "loss": 0.0737,
+      "step": 439
+    },
+    {
+      "epoch": 1.069258809234508,
+      "grad_norm": 0.4835350811481476,
+      "learning_rate": 5e-05,
+      "loss": 0.0644,
+      "step": 440
+    },
+    {
+      "epoch": 1.071688942891859,
+      "grad_norm": 0.6343538761138916,
+      "learning_rate": 5e-05,
+      "loss": 0.0796,
+      "step": 441
+    },
+    {
+      "epoch": 1.0741190765492101,
+      "grad_norm": 0.41634368896484375,
+      "learning_rate": 5e-05,
+      "loss": 0.087,
+      "step": 442
+    },
+    {
+      "epoch": 1.0765492102065615,
+      "grad_norm": 0.7903019189834595,
+      "learning_rate": 5e-05,
+      "loss": 0.0564,
+      "step": 443
+    },
+    {
+      "epoch": 1.0789793438639126,
+      "grad_norm": 0.45615118741989136,
+      "learning_rate": 5e-05,
+      "loss": 0.0928,
+      "step": 444
+    },
+    {
+      "epoch": 1.0814094775212637,
+      "grad_norm": 0.5224602222442627,
+      "learning_rate": 5e-05,
+      "loss": 0.1134,
+      "step": 445
+    },
+    {
+      "epoch": 1.0838396111786148,
+      "grad_norm": 0.7974500060081482,
+      "learning_rate": 5e-05,
+      "loss": 0.1592,
+      "step": 446
+    },
+    {
+      "epoch": 1.086269744835966,
+      "grad_norm": 0.545687735080719,
+      "learning_rate": 5e-05,
+      "loss": 0.0892,
+      "step": 447
+    },
+    {
+      "epoch": 1.088699878493317,
+      "grad_norm": 0.88418048620224,
+      "learning_rate": 5e-05,
+      "loss": 0.0406,
+      "step": 448
+    },
+    {
+      "epoch": 1.0911300121506682,
+      "grad_norm": 0.6175111532211304,
+      "learning_rate": 5e-05,
+      "loss": 0.0536,
+      "step": 449
+    },
+    {
+      "epoch": 1.0935601458080195,
+      "grad_norm": 0.5780444741249084,
+      "learning_rate": 5e-05,
+      "loss": 0.0806,
+      "step": 450
+    },
+    {
+      "epoch": 1.0959902794653706,
+      "grad_norm": 2.7306559085845947,
+      "learning_rate": 5e-05,
+      "loss": 0.102,
+      "step": 451
+    },
+    {
+      "epoch": 1.0984204131227218,
+      "grad_norm": 0.8727555871009827,
+      "learning_rate": 5e-05,
+      "loss": 0.0723,
+      "step": 452
+    },
+    {
+      "epoch": 1.1008505467800729,
+      "grad_norm": 0.49243301153182983,
+      "learning_rate": 5e-05,
+      "loss": 0.0589,
+      "step": 453
+    },
+    {
+      "epoch": 1.103280680437424,
+      "grad_norm": 1.0444567203521729,
+      "learning_rate": 5e-05,
+      "loss": 0.1075,
+      "step": 454
+    },
+    {
+      "epoch": 1.1057108140947751,
+      "grad_norm": 0.5982443690299988,
+      "learning_rate": 5e-05,
+      "loss": 0.0646,
+      "step": 455
+    },
+    {
+      "epoch": 1.1081409477521265,
+      "grad_norm": 0.4428943395614624,
+      "learning_rate": 5e-05,
+      "loss": 0.0421,
+      "step": 456
+    },
+    {
+      "epoch": 1.1105710814094776,
+      "grad_norm": 0.612250566482544,
+      "learning_rate": 5e-05,
+      "loss": 0.0694,
+      "step": 457
+    },
+    {
+      "epoch": 1.1130012150668287,
+      "grad_norm": 0.8445409536361694,
+      "learning_rate": 5e-05,
+      "loss": 0.1089,
+      "step": 458
+    },
+    {
+      "epoch": 1.1154313487241798,
+      "grad_norm": 0.5859739184379578,
+      "learning_rate": 5e-05,
+      "loss": 0.1143,
+      "step": 459
+    },
+    {
+      "epoch": 1.117861482381531,
+      "grad_norm": 0.38835424184799194,
+      "learning_rate": 5e-05,
+      "loss": 0.0866,
+      "step": 460
+    },
+    {
+      "epoch": 1.120291616038882,
+      "grad_norm": 0.7463281750679016,
+      "learning_rate": 5e-05,
+      "loss": 0.0815,
+      "step": 461
+    },
+    {
+      "epoch": 1.1227217496962334,
+      "grad_norm": 0.6839048266410828,
+      "learning_rate": 5e-05,
+      "loss": 0.1349,
+      "step": 462
+    },
+    {
+      "epoch": 1.1251518833535845,
+      "grad_norm": 1.4645401239395142,
+      "learning_rate": 5e-05,
+      "loss": 0.0799,
+      "step": 463
+    },
+    {
+      "epoch": 1.1275820170109356,
+      "grad_norm": 0.7507435083389282,
+      "learning_rate": 5e-05,
+      "loss": 0.0598,
+      "step": 464
+    },
+    {
+      "epoch": 1.1300121506682868,
+      "grad_norm": 0.7818620800971985,
+      "learning_rate": 5e-05,
+      "loss": 0.0938,
+      "step": 465
+    },
+    {
+      "epoch": 1.1324422843256379,
+      "grad_norm": 0.6816830635070801,
+      "learning_rate": 5e-05,
+      "loss": 0.0787,
+      "step": 466
+    },
+    {
+      "epoch": 1.134872417982989,
+      "grad_norm": 0.575626015663147,
+      "learning_rate": 5e-05,
+      "loss": 0.1461,
+      "step": 467
+    },
+    {
+      "epoch": 1.13730255164034,
+      "grad_norm": 0.73714679479599,
+      "learning_rate": 5e-05,
+      "loss": 0.0761,
+      "step": 468
+    },
+    {
+      "epoch": 1.1397326852976915,
+      "grad_norm": 0.7288960814476013,
+      "learning_rate": 5e-05,
+      "loss": 0.074,
+      "step": 469
+    },
+    {
+      "epoch": 1.1421628189550426,
+      "grad_norm": 0.5384907126426697,
+      "learning_rate": 5e-05,
+      "loss": 0.0732,
+      "step": 470
+    },
+    {
+      "epoch": 1.1445929526123937,
+      "grad_norm": 0.6071632504463196,
+      "learning_rate": 5e-05,
+      "loss": 0.03,
+      "step": 471
+    },
+    {
+      "epoch": 1.1470230862697448,
+      "grad_norm": 0.5744608044624329,
+      "learning_rate": 5e-05,
+      "loss": 0.0912,
+      "step": 472
+    },
+    {
+      "epoch": 1.149453219927096,
+      "grad_norm": 0.9392772912979126,
+      "learning_rate": 5e-05,
+      "loss": 0.1024,
+      "step": 473
+    },
+    {
+      "epoch": 1.1518833535844473,
+      "grad_norm": 1.170997142791748,
+      "learning_rate": 5e-05,
+      "loss": 0.1027,
+      "step": 474
+    },
+    {
+      "epoch": 1.1543134872417984,
+      "grad_norm": 0.791221559047699,
+      "learning_rate": 5e-05,
+      "loss": 0.1305,
+      "step": 475
+    },
+    {
+      "epoch": 1.1567436208991495,
+      "grad_norm": 0.5443055033683777,
+      "learning_rate": 5e-05,
+      "loss": 0.0749,
+      "step": 476
+    },
+    {
+      "epoch": 1.1591737545565006,
+      "grad_norm": 0.8006246089935303,
+      "learning_rate": 5e-05,
+      "loss": 0.0976,
+      "step": 477
+    },
+    {
+      "epoch": 1.1616038882138517,
+      "grad_norm": 0.47016945481300354,
+      "learning_rate": 5e-05,
+      "loss": 0.0386,
+      "step": 478
+    },
+    {
+      "epoch": 1.1640340218712029,
+      "grad_norm": 0.7536148428916931,
+      "learning_rate": 5e-05,
+      "loss": 0.1385,
+      "step": 479
+    },
+    {
+      "epoch": 1.166464155528554,
+      "grad_norm": 0.6575655341148376,
+      "learning_rate": 5e-05,
+      "loss": 0.0851,
+      "step": 480
+    },
+    {
+      "epoch": 1.1688942891859053,
+      "grad_norm": 0.484667032957077,
+      "learning_rate": 5e-05,
+      "loss": 0.0301,
+      "step": 481
+    },
+    {
+      "epoch": 1.1713244228432564,
+      "grad_norm": 0.8194088339805603,
+      "learning_rate": 5e-05,
+      "loss": 0.129,
+      "step": 482
+    },
+    {
+      "epoch": 1.1737545565006076,
+      "grad_norm": 0.6559669375419617,
+      "learning_rate": 5e-05,
+      "loss": 0.0667,
+      "step": 483
+    },
+    {
+      "epoch": 1.1761846901579587,
+      "grad_norm": 0.8752015829086304,
+      "learning_rate": 5e-05,
+      "loss": 0.0637,
+      "step": 484
+    },
+    {
+      "epoch": 1.1786148238153098,
+      "grad_norm": 0.6029524207115173,
+      "learning_rate": 5e-05,
+      "loss": 0.0557,
+      "step": 485
+    },
+    {
+      "epoch": 1.181044957472661,
+      "grad_norm": 0.5431787371635437,
+      "learning_rate": 5e-05,
+      "loss": 0.0338,
+      "step": 486
+    },
+    {
+      "epoch": 1.183475091130012,
+      "grad_norm": 0.5199941992759705,
+      "learning_rate": 5e-05,
+      "loss": 0.0603,
+      "step": 487
+    },
+    {
+      "epoch": 1.1859052247873634,
+      "grad_norm": 0.520455539226532,
+      "learning_rate": 5e-05,
+      "loss": 0.0221,
+      "step": 488
+    },
+    {
+      "epoch": 1.1883353584447145,
+      "grad_norm": 0.5124850273132324,
+      "learning_rate": 5e-05,
+      "loss": 0.0306,
+      "step": 489
+    },
+    {
+      "epoch": 1.1907654921020656,
+      "grad_norm": 0.6515238881111145,
+      "learning_rate": 5e-05,
+      "loss": 0.0612,
+      "step": 490
+    },
+    {
+      "epoch": 1.1931956257594167,
+      "grad_norm": 0.816235363483429,
+      "learning_rate": 5e-05,
+      "loss": 0.1585,
+      "step": 491
+    },
+    {
+      "epoch": 1.1956257594167679,
+      "grad_norm": 0.6307040452957153,
+      "learning_rate": 5e-05,
+      "loss": 0.0551,
+      "step": 492
+    },
+    {
+      "epoch": 1.1980558930741192,
+      "grad_norm": 0.46716415882110596,
+      "learning_rate": 5e-05,
+      "loss": 0.0457,
+      "step": 493
+    },
+    {
+      "epoch": 1.2004860267314703,
+      "grad_norm": 0.5497251749038696,
+      "learning_rate": 5e-05,
+      "loss": 0.0874,
+      "step": 494
+    },
+    {
+      "epoch": 1.2029161603888214,
+      "grad_norm": 1.0956586599349976,
+      "learning_rate": 5e-05,
+      "loss": 0.0796,
+      "step": 495
+    },
+    {
+      "epoch": 1.2053462940461726,
+      "grad_norm": 0.6908737421035767,
+      "learning_rate": 5e-05,
+      "loss": 0.0406,
+      "step": 496
+    },
+    {
+      "epoch": 1.2077764277035237,
+      "grad_norm": 0.8598071932792664,
+      "learning_rate": 5e-05,
+      "loss": 0.1157,
+      "step": 497
+    },
+    {
+      "epoch": 1.2102065613608748,
+      "grad_norm": 0.6752752065658569,
+      "learning_rate": 5e-05,
+      "loss": 0.068,
+      "step": 498
+    },
+    {
+      "epoch": 1.212636695018226,
+      "grad_norm": 0.5425324440002441,
+      "learning_rate": 5e-05,
+      "loss": 0.0976,
+      "step": 499
+    },
+    {
+      "epoch": 1.2150668286755772,
+      "grad_norm": 0.7367557883262634,
+      "learning_rate": 5e-05,
+      "loss": 0.1028,
+      "step": 500
+    },
+    {
+      "epoch": 1.2150668286755772,
+      "eval_loss": 0.23548956215381622,
+      "eval_runtime": 512.2983,
+      "eval_samples_per_second": 5.364,
+      "eval_steps_per_second": 0.671,
+      "step": 500
+    },
+    {
+      "epoch": 1.2174969623329284,
+      "grad_norm": 0.8351901769638062,
+      "learning_rate": 5e-05,
+      "loss": 0.0927,
+      "step": 501
+    },
+    {
+      "epoch": 1.2199270959902795,
+      "grad_norm": 0.5388638377189636,
+      "learning_rate": 5e-05,
+      "loss": 0.0956,
+      "step": 502
+    },
+    {
+      "epoch": 1.2223572296476306,
+      "grad_norm": 0.9194818139076233,
+      "learning_rate": 5e-05,
+      "loss": 0.0533,
+      "step": 503
+    },
+    {
+      "epoch": 1.2247873633049817,
+      "grad_norm": 0.7008382081985474,
+      "learning_rate": 5e-05,
+      "loss": 0.0554,
+      "step": 504
+    },
+    {
+      "epoch": 1.2272174969623328,
+      "grad_norm": 0.5537183284759521,
+      "learning_rate": 5e-05,
+      "loss": 0.1277,
+      "step": 505
+    },
+    {
+      "epoch": 1.229647630619684,
+      "grad_norm": 0.570819079875946,
+      "learning_rate": 5e-05,
+      "loss": 0.1037,
+      "step": 506
+    },
+    {
+      "epoch": 1.2320777642770353,
+      "grad_norm": 0.5867807269096375,
+      "learning_rate": 5e-05,
+      "loss": 0.0578,
+      "step": 507
+    },
+    {
+      "epoch": 1.2345078979343864,
+      "grad_norm": 0.7660566568374634,
+      "learning_rate": 5e-05,
+      "loss": 0.0856,
+      "step": 508
+    },
+    {
+      "epoch": 1.2369380315917375,
+      "grad_norm": 0.7493465542793274,
+      "learning_rate": 5e-05,
+      "loss": 0.0567,
+      "step": 509
+    },
+    {
+      "epoch": 1.2393681652490887,
+      "grad_norm": 0.6718665957450867,
+      "learning_rate": 5e-05,
+      "loss": 0.1113,
+      "step": 510
+    },
+    {
+      "epoch": 1.2417982989064398,
+      "grad_norm": 0.7018153667449951,
+      "learning_rate": 5e-05,
+      "loss": 0.0966,
+      "step": 511
+    },
+    {
+      "epoch": 1.2442284325637911,
+      "grad_norm": 1.3940482139587402,
+      "learning_rate": 5e-05,
+      "loss": 0.1352,
+      "step": 512
+    },
+    {
+      "epoch": 1.2466585662211422,
+      "grad_norm": 0.663314938545227,
+      "learning_rate": 5e-05,
+      "loss": 0.0556,
+      "step": 513
+    },
+    {
+      "epoch": 1.2490886998784934,
+      "grad_norm": 1.0090134143829346,
+      "learning_rate": 5e-05,
+      "loss": 0.0352,
+      "step": 514
+    },
+    {
+      "epoch": 1.2515188335358445,
+      "grad_norm": 0.5742064714431763,
+      "learning_rate": 5e-05,
+      "loss": 0.0669,
+      "step": 515
+    },
+    {
+      "epoch": 1.2539489671931956,
+      "grad_norm": 0.726700484752655,
+      "learning_rate": 5e-05,
+      "loss": 0.0533,
+      "step": 516
+    },
+    {
+      "epoch": 1.2563791008505467,
+      "grad_norm": 0.6788687109947205,
+      "learning_rate": 5e-05,
+      "loss": 0.0567,
+      "step": 517
+    },
+    {
+      "epoch": 1.2588092345078978,
+      "grad_norm": 0.48673132061958313,
+      "learning_rate": 5e-05,
+      "loss": 0.0406,
+      "step": 518
+    },
+    {
+      "epoch": 1.2612393681652492,
+      "grad_norm": 1.1016216278076172,
+      "learning_rate": 5e-05,
+      "loss": 0.0668,
+      "step": 519
+    },
+    {
+      "epoch": 1.2636695018226003,
+      "grad_norm": 1.2345398664474487,
+      "learning_rate": 5e-05,
+      "loss": 0.029,
+      "step": 520
+    },
+    {
+      "epoch": 1.2660996354799514,
+      "grad_norm": 0.7507830858230591,
+      "learning_rate": 5e-05,
+      "loss": 0.0991,
+      "step": 521
+    },
+    {
+      "epoch": 1.2685297691373025,
+      "grad_norm": 0.819266140460968,
+      "learning_rate": 5e-05,
+      "loss": 0.0723,
+      "step": 522
+    },
+    {
+      "epoch": 1.2709599027946537,
+      "grad_norm": 0.9876227974891663,
+      "learning_rate": 5e-05,
+      "loss": 0.1555,
+      "step": 523
+    },
+    {
+      "epoch": 1.273390036452005,
+      "grad_norm": 0.7902403473854065,
+      "learning_rate": 5e-05,
+      "loss": 0.156,
+      "step": 524
+    },
+    {
+      "epoch": 1.2758201701093559,
+      "grad_norm": 0.6342129111289978,
+      "learning_rate": 5e-05,
+      "loss": 0.1161,
+      "step": 525
+    },
+    {
+      "epoch": 1.2782503037667072,
+      "grad_norm": 0.9039121866226196,
+      "learning_rate": 5e-05,
+      "loss": 0.1078,
+      "step": 526
+    },
+    {
+      "epoch": 1.2806804374240583,
+      "grad_norm": 0.9820278882980347,
+      "learning_rate": 5e-05,
+      "loss": 0.1182,
+      "step": 527
+    },
+    {
+      "epoch": 1.2831105710814095,
+      "grad_norm": 0.45999497175216675,
+      "learning_rate": 5e-05,
+      "loss": 0.0469,
+      "step": 528
+    },
+    {
+      "epoch": 1.2855407047387606,
+      "grad_norm": 0.6754646301269531,
+      "learning_rate": 5e-05,
+      "loss": 0.0875,
+      "step": 529
+    },
+    {
+      "epoch": 1.2879708383961117,
+      "grad_norm": 0.7348721027374268,
+      "learning_rate": 5e-05,
+      "loss": 0.0639,
+      "step": 530
+    },
+    {
+      "epoch": 1.290400972053463,
+      "grad_norm": 0.7348721027374268,
+      "learning_rate": 5e-05,
+      "loss": 0.0764,
+      "step": 531
+    },
+    {
+      "epoch": 1.2928311057108142,
+      "grad_norm": 0.4331410229206085,
+      "learning_rate": 5e-05,
+      "loss": 0.0278,
+      "step": 532
+    },
+    {
+      "epoch": 1.2952612393681653,
+      "grad_norm": 0.7986178994178772,
+      "learning_rate": 5e-05,
+      "loss": 0.0893,
+      "step": 533
+    },
+    {
+      "epoch": 1.2976913730255164,
+      "grad_norm": 0.7255426049232483,
+      "learning_rate": 5e-05,
+      "loss": 0.0802,
+      "step": 534
+    },
+    {
+      "epoch": 1.3001215066828675,
+      "grad_norm": 0.590578556060791,
+      "learning_rate": 5e-05,
+      "loss": 0.0853,
+      "step": 535
+    },
+    {
+      "epoch": 1.3025516403402186,
+      "grad_norm": 0.6147257685661316,
+      "learning_rate": 5e-05,
+      "loss": 0.0462,
+      "step": 536
+    },
+    {
+      "epoch": 1.3049817739975698,
+      "grad_norm": 0.6174589991569519,
+      "learning_rate": 5e-05,
+      "loss": 0.0758,
+      "step": 537
+    },
+    {
+      "epoch": 1.307411907654921,
+      "grad_norm": 0.5428439378738403,
+      "learning_rate": 5e-05,
+      "loss": 0.1452,
+      "step": 538
+    },
+    {
+      "epoch": 1.3098420413122722,
+      "grad_norm": 0.554800271987915,
+      "learning_rate": 5e-05,
+      "loss": 0.0266,
+      "step": 539
+    },
+    {
+      "epoch": 1.3122721749696233,
+      "grad_norm": 0.627275824546814,
+      "learning_rate": 5e-05,
+      "loss": 0.0604,
+      "step": 540
+    },
+    {
+      "epoch": 1.3147023086269745,
+      "grad_norm": 0.5268173813819885,
+      "learning_rate": 5e-05,
+      "loss": 0.036,
+      "step": 541
+    },
+    {
+      "epoch": 1.3171324422843256,
+      "grad_norm": 0.4525201916694641,
+      "learning_rate": 5e-05,
+      "loss": 0.0746,
+      "step": 542
+    },
+    {
+      "epoch": 1.319562575941677,
+      "grad_norm": 1.299209475517273,
+      "learning_rate": 5e-05,
+      "loss": 0.0573,
+      "step": 543
+    },
+    {
+      "epoch": 1.3219927095990278,
+      "grad_norm": 0.5899752378463745,
+      "learning_rate": 5e-05,
+      "loss": 0.0373,
+      "step": 544
+    },
+    {
+      "epoch": 1.3244228432563792,
+      "grad_norm": 0.5044320225715637,
+      "learning_rate": 5e-05,
+      "loss": 0.0415,
+      "step": 545
+    },
+    {
+      "epoch": 1.3268529769137303,
+      "grad_norm": 0.4812248945236206,
+      "learning_rate": 5e-05,
+      "loss": 0.0898,
+      "step": 546
+    },
+    {
+      "epoch": 1.3292831105710814,
+      "grad_norm": 0.47461068630218506,
+      "learning_rate": 5e-05,
+      "loss": 0.0248,
+      "step": 547
+    },
+    {
+      "epoch": 1.3317132442284325,
+      "grad_norm": 0.5620236396789551,
+      "learning_rate": 5e-05,
+      "loss": 0.0418,
+      "step": 548
+    },
+    {
+      "epoch": 1.3341433778857836,
+      "grad_norm": 0.8658341765403748,
+      "learning_rate": 5e-05,
+      "loss": 0.0208,
+      "step": 549
+    },
+    {
+      "epoch": 1.336573511543135,
+      "grad_norm": 0.6640876531600952,
+      "learning_rate": 5e-05,
+      "loss": 0.0652,
+      "step": 550
+    },
+    {
+      "epoch": 1.339003645200486,
+      "grad_norm": 0.3873005509376526,
+      "learning_rate": 5e-05,
+      "loss": 0.018,
+      "step": 551
+    },
+    {
+      "epoch": 1.3414337788578372,
+      "grad_norm": 0.4546399414539337,
+      "learning_rate": 5e-05,
+      "loss": 0.0823,
+      "step": 552
+    },
+    {
+      "epoch": 1.3438639125151883,
+      "grad_norm": 0.5188842415809631,
+      "learning_rate": 5e-05,
+      "loss": 0.0604,
+      "step": 553
+    },
+    {
+      "epoch": 1.3462940461725394,
+      "grad_norm": 0.41030353307724,
+      "learning_rate": 5e-05,
+      "loss": 0.0112,
+      "step": 554
+    },
+    {
+      "epoch": 1.3487241798298906,
+      "grad_norm": 0.7154520750045776,
+      "learning_rate": 5e-05,
+      "loss": 0.0687,
+      "step": 555
+    },
+    {
+      "epoch": 1.3511543134872417,
+      "grad_norm": 0.6743410229682922,
+      "learning_rate": 5e-05,
+      "loss": 0.1262,
+      "step": 556
+    },
+    {
+      "epoch": 1.353584447144593,
+      "grad_norm": 0.38846057653427124,
+      "learning_rate": 5e-05,
+      "loss": 0.0371,
+      "step": 557
+    },
+    {
+      "epoch": 1.3560145808019441,
+      "grad_norm": 0.5022516846656799,
+      "learning_rate": 5e-05,
+      "loss": 0.0159,
+      "step": 558
+    },
+    {
+      "epoch": 1.3584447144592953,
+      "grad_norm": 0.5445972084999084,
+      "learning_rate": 5e-05,
+      "loss": 0.0652,
+      "step": 559
+    },
+    {
+      "epoch": 1.3608748481166464,
+      "grad_norm": 0.6346146464347839,
+      "learning_rate": 5e-05,
+      "loss": 0.1128,
+      "step": 560
+    },
+    {
+      "epoch": 1.3633049817739975,
+      "grad_norm": 0.7129392623901367,
+      "learning_rate": 5e-05,
+      "loss": 0.059,
+      "step": 561
+    },
+    {
+      "epoch": 1.3657351154313488,
+      "grad_norm": 0.4905102252960205,
+      "learning_rate": 5e-05,
+      "loss": 0.1114,
+      "step": 562
+    },
+    {
+      "epoch": 1.3681652490886997,
+      "grad_norm": 0.7290562391281128,
+      "learning_rate": 5e-05,
+      "loss": 0.1029,
+      "step": 563
+    },
+    {
+      "epoch": 1.370595382746051,
+      "grad_norm": 0.5576648116111755,
+      "learning_rate": 5e-05,
+      "loss": 0.0383,
+      "step": 564
+    },
+    {
+      "epoch": 1.3730255164034022,
+      "grad_norm": 0.595310389995575,
+      "learning_rate": 5e-05,
+      "loss": 0.0906,
+      "step": 565
+    },
+    {
+      "epoch": 1.3754556500607533,
+      "grad_norm": 0.7409107089042664,
+      "learning_rate": 5e-05,
+      "loss": 0.0239,
+      "step": 566
+    },
+    {
+      "epoch": 1.3778857837181044,
+      "grad_norm": 1.0220900774002075,
+      "learning_rate": 5e-05,
+      "loss": 0.0537,
+      "step": 567
+    },
+    {
+      "epoch": 1.3803159173754556,
+      "grad_norm": 1.0970425605773926,
+      "learning_rate": 5e-05,
+      "loss": 0.0615,
+      "step": 568
+    },
+    {
+      "epoch": 1.382746051032807,
+      "grad_norm": 0.45453736186027527,
+      "learning_rate": 5e-05,
+      "loss": 0.1,
+      "step": 569
+    },
+    {
+      "epoch": 1.385176184690158,
+      "grad_norm": 0.61972576379776,
+      "learning_rate": 5e-05,
+      "loss": 0.1029,
+      "step": 570
+    },
+    {
+      "epoch": 1.3876063183475091,
+      "grad_norm": 0.47734493017196655,
+      "learning_rate": 5e-05,
+      "loss": 0.0683,
+      "step": 571
+    },
+    {
+      "epoch": 1.3900364520048603,
+      "grad_norm": 0.3732990026473999,
+      "learning_rate": 5e-05,
+      "loss": 0.0706,
+      "step": 572
+    },
+    {
+      "epoch": 1.3924665856622114,
+      "grad_norm": 0.49241572618484497,
+      "learning_rate": 5e-05,
+      "loss": 0.0667,
+      "step": 573
+    },
+    {
+      "epoch": 1.3948967193195625,
+      "grad_norm": 0.6907554268836975,
+      "learning_rate": 5e-05,
+      "loss": 0.0602,
+      "step": 574
+    },
+    {
+      "epoch": 1.3973268529769136,
+      "grad_norm": 0.8935093879699707,
+      "learning_rate": 5e-05,
+      "loss": 0.2114,
+      "step": 575
+    },
+    {
+      "epoch": 1.399756986634265,
+      "grad_norm": 0.2839941680431366,
+      "learning_rate": 5e-05,
+      "loss": 0.0257,
+      "step": 576
+    },
+    {
+      "epoch": 1.402187120291616,
+      "grad_norm": 0.7202839255332947,
+      "learning_rate": 5e-05,
+      "loss": 0.0878,
+      "step": 577
+    },
+    {
+      "epoch": 1.4046172539489672,
+      "grad_norm": 0.6520135402679443,
+      "learning_rate": 5e-05,
+      "loss": 0.0295,
+      "step": 578
+    },
+    {
+      "epoch": 1.4070473876063183,
+      "grad_norm": 0.7360886335372925,
+      "learning_rate": 5e-05,
+      "loss": 0.065,
+      "step": 579
+    },
+    {
+      "epoch": 1.4094775212636694,
+      "grad_norm": 0.5855164527893066,
+      "learning_rate": 5e-05,
+      "loss": 0.0752,
+      "step": 580
+    },
+    {
+      "epoch": 1.4119076549210208,
+      "grad_norm": 0.38474756479263306,
+      "learning_rate": 5e-05,
+      "loss": 0.0541,
+      "step": 581
+    },
+    {
+      "epoch": 1.414337788578372,
+      "grad_norm": 0.5105370283126831,
+      "learning_rate": 5e-05,
+      "loss": 0.0749,
+      "step": 582
+    },
+    {
+      "epoch": 1.416767922235723,
+      "grad_norm": 0.5076673030853271,
+      "learning_rate": 5e-05,
+      "loss": 0.0528,
+      "step": 583
+    },
+    {
+      "epoch": 1.4191980558930741,
+      "grad_norm": 0.5034826397895813,
+      "learning_rate": 5e-05,
+      "loss": 0.0505,
+      "step": 584
+    },
+    {
+      "epoch": 1.4216281895504252,
+      "grad_norm": 0.6378771066665649,
+      "learning_rate": 5e-05,
+      "loss": 0.0676,
+      "step": 585
+    },
+    {
+      "epoch": 1.4240583232077764,
+      "grad_norm": 0.5122814774513245,
+      "learning_rate": 5e-05,
+      "loss": 0.0515,
+      "step": 586
+    },
+    {
+      "epoch": 1.4264884568651275,
+      "grad_norm": 0.5807358026504517,
+      "learning_rate": 5e-05,
+      "loss": 0.0507,
+      "step": 587
+    },
+    {
+      "epoch": 1.4289185905224788,
+      "grad_norm": 0.5156981348991394,
+      "learning_rate": 5e-05,
+      "loss": 0.0503,
+      "step": 588
+    },
+    {
+      "epoch": 1.43134872417983,
+      "grad_norm": 0.9181635975837708,
+      "learning_rate": 5e-05,
+      "loss": 0.0964,
+      "step": 589
+    },
+    {
+      "epoch": 1.433778857837181,
+      "grad_norm": 0.38104763627052307,
+      "learning_rate": 5e-05,
+      "loss": 0.055,
+      "step": 590
+    },
+    {
+      "epoch": 1.4362089914945322,
+      "grad_norm": 0.4987882673740387,
+      "learning_rate": 5e-05,
+      "loss": 0.0347,
+      "step": 591
+    },
+    {
+      "epoch": 1.4386391251518833,
+      "grad_norm": 0.5201525688171387,
+      "learning_rate": 5e-05,
+      "loss": 0.0904,
+      "step": 592
+    },
+    {
+      "epoch": 1.4410692588092346,
+      "grad_norm": 0.5238099694252014,
+      "learning_rate": 5e-05,
+      "loss": 0.077,
+      "step": 593
+    },
+    {
+      "epoch": 1.4434993924665855,
+      "grad_norm": 0.924893319606781,
+      "learning_rate": 5e-05,
+      "loss": 0.1248,
+      "step": 594
+    },
+    {
+      "epoch": 1.4459295261239369,
+      "grad_norm": 0.6318195462226868,
+      "learning_rate": 5e-05,
+      "loss": 0.0267,
+      "step": 595
+    },
+    {
+      "epoch": 1.448359659781288,
+      "grad_norm": 0.9913945198059082,
+      "learning_rate": 5e-05,
+      "loss": 0.0917,
+      "step": 596
+    },
+    {
+      "epoch": 1.4507897934386391,
+      "grad_norm": 0.4909703731536865,
+      "learning_rate": 5e-05,
+      "loss": 0.036,
+      "step": 597
+    },
+    {
+      "epoch": 1.4532199270959902,
+      "grad_norm": 0.5913828015327454,
+      "learning_rate": 5e-05,
+      "loss": 0.0353,
+      "step": 598
+    },
+    {
+      "epoch": 1.4556500607533414,
+      "grad_norm": 0.3615020215511322,
+      "learning_rate": 5e-05,
+      "loss": 0.0517,
+      "step": 599
+    },
+    {
+      "epoch": 1.4580801944106927,
+      "grad_norm": 0.5106006264686584,
+      "learning_rate": 5e-05,
+      "loss": 0.0444,
+      "step": 600
+    },
+    {
+      "epoch": 1.4580801944106927,
+      "eval_loss": 0.22935771942138672,
+      "eval_runtime": 507.4275,
+      "eval_samples_per_second": 5.416,
+      "eval_steps_per_second": 0.678,
+      "step": 600
+    },
+    {
+      "epoch": 1.4605103280680438,
+      "grad_norm": 0.7134807705879211,
+      "learning_rate": 5e-05,
+      "loss": 0.056,
+      "step": 601
+    },
+    {
+      "epoch": 1.462940461725395,
+      "grad_norm": 0.2919784188270569,
+      "learning_rate": 5e-05,
+      "loss": 0.0146,
+      "step": 602
+    },
+    {
+      "epoch": 1.465370595382746,
+      "grad_norm": 0.6526296734809875,
+      "learning_rate": 5e-05,
+      "loss": 0.1207,
+      "step": 603
+    },
+    {
+      "epoch": 1.4678007290400972,
+      "grad_norm": 0.5833859443664551,
+      "learning_rate": 5e-05,
+      "loss": 0.0657,
+      "step": 604
+    },
+    {
+      "epoch": 1.4702308626974483,
+      "grad_norm": 0.6107626557350159,
+      "learning_rate": 5e-05,
+      "loss": 0.0655,
+      "step": 605
+    },
+    {
+      "epoch": 1.4726609963547994,
+      "grad_norm": 0.5227603912353516,
+      "learning_rate": 5e-05,
+      "loss": 0.0504,
+      "step": 606
+    },
+    {
+      "epoch": 1.4750911300121508,
+      "grad_norm": 0.6677300333976746,
+      "learning_rate": 5e-05,
+      "loss": 0.0465,
+      "step": 607
+    },
+    {
+      "epoch": 1.4775212636695019,
+      "grad_norm": 0.7898095846176147,
+      "learning_rate": 5e-05,
+      "loss": 0.0932,
+      "step": 608
+    },
+    {
+      "epoch": 1.479951397326853,
+      "grad_norm": 0.9366335868835449,
+      "learning_rate": 5e-05,
+      "loss": 0.0819,
+      "step": 609
+    },
+    {
+      "epoch": 1.482381530984204,
+      "grad_norm": 0.6056090593338013,
+      "learning_rate": 5e-05,
+      "loss": 0.0638,
+      "step": 610
+    },
+    {
+      "epoch": 1.4848116646415552,
+      "grad_norm": 0.7640931606292725,
+      "learning_rate": 5e-05,
+      "loss": 0.1152,
+      "step": 611
+    },
+    {
+      "epoch": 1.4872417982989066,
+      "grad_norm": 0.7496662735939026,
+      "learning_rate": 5e-05,
+      "loss": 0.0461,
+      "step": 612
+    },
+    {
+      "epoch": 1.4896719319562575,
+      "grad_norm": 0.6870647072792053,
+      "learning_rate": 5e-05,
+      "loss": 0.0812,
+      "step": 613
+    },
+    {
+      "epoch": 1.4921020656136088,
+      "grad_norm": 0.7589767575263977,
+      "learning_rate": 5e-05,
+      "loss": 0.0846,
+      "step": 614
+    },
+    {
+      "epoch": 1.49453219927096,
+      "grad_norm": 0.681777834892273,
+      "learning_rate": 5e-05,
+      "loss": 0.0407,
+      "step": 615
+    },
+    {
+      "epoch": 1.496962332928311,
+      "grad_norm": 0.7369627952575684,
+      "learning_rate": 5e-05,
+      "loss": 0.0955,
+      "step": 616
+    },
+    {
+      "epoch": 1.4993924665856622,
+      "grad_norm": 0.7186546921730042,
+      "learning_rate": 5e-05,
+      "loss": 0.0745,
+      "step": 617
+    },
+    {
+      "epoch": 1.5018226002430133,
+      "grad_norm": 0.6271875500679016,
+      "learning_rate": 5e-05,
+      "loss": 0.0533,
+      "step": 618
+    },
+    {
+      "epoch": 1.5042527339003646,
+      "grad_norm": 0.5378838777542114,
+      "learning_rate": 5e-05,
+      "loss": 0.0688,
+      "step": 619
+    },
+    {
+      "epoch": 1.5066828675577155,
+      "grad_norm": 0.7019214630126953,
+      "learning_rate": 5e-05,
+      "loss": 0.0369,
+      "step": 620
+    },
+    {
+      "epoch": 1.5091130012150669,
+      "grad_norm": 0.9618707299232483,
+      "learning_rate": 5e-05,
+      "loss": 0.0469,
+      "step": 621
+    },
+    {
+      "epoch": 1.511543134872418,
+      "grad_norm": 0.4820462167263031,
+      "learning_rate": 5e-05,
+      "loss": 0.074,
+      "step": 622
+    },
+    {
+      "epoch": 1.513973268529769,
+      "grad_norm": 0.411123126745224,
+      "learning_rate": 5e-05,
+      "loss": 0.0307,
+      "step": 623
+    },
+    {
+      "epoch": 1.5164034021871204,
+      "grad_norm": 0.5718781352043152,
+      "learning_rate": 5e-05,
+      "loss": 0.0276,
+      "step": 624
+    },
+    {
+      "epoch": 1.5188335358444713,
+      "grad_norm": 0.4341799020767212,
+      "learning_rate": 5e-05,
+      "loss": 0.0651,
+      "step": 625
+    },
+    {
+      "epoch": 1.5212636695018227,
+      "grad_norm": 0.515855073928833,
+      "learning_rate": 5e-05,
+      "loss": 0.0413,
+      "step": 626
+    },
+    {
+      "epoch": 1.5236938031591738,
+      "grad_norm": 0.6174566149711609,
+      "learning_rate": 5e-05,
+      "loss": 0.0307,
+      "step": 627
+    },
+    {
+      "epoch": 1.526123936816525,
+      "grad_norm": 0.5931458473205566,
+      "learning_rate": 5e-05,
+      "loss": 0.0852,
+      "step": 628
+    },
+    {
+      "epoch": 1.528554070473876,
+      "grad_norm": 0.46777036786079407,
+      "learning_rate": 5e-05,
+      "loss": 0.0438,
+      "step": 629
+    },
+    {
+      "epoch": 1.5309842041312272,
+      "grad_norm": 0.4254392087459564,
+      "learning_rate": 5e-05,
+      "loss": 0.0606,
+      "step": 630
+    },
+    {
+      "epoch": 1.5334143377885785,
+      "grad_norm": 0.5670008063316345,
+      "learning_rate": 5e-05,
+      "loss": 0.0514,
+      "step": 631
+    },
+    {
+      "epoch": 1.5358444714459294,
+      "grad_norm": 1.1566509008407593,
+      "learning_rate": 5e-05,
+      "loss": 0.1133,
+      "step": 632
+    },
+    {
+      "epoch": 1.5382746051032807,
+      "grad_norm": 0.5660587549209595,
+      "learning_rate": 5e-05,
+      "loss": 0.0544,
+      "step": 633
+    },
+    {
+      "epoch": 1.5407047387606319,
+      "grad_norm": 0.5406026840209961,
+      "learning_rate": 5e-05,
+      "loss": 0.046,
+      "step": 634
+    },
+    {
+      "epoch": 1.543134872417983,
+      "grad_norm": 0.49723613262176514,
+      "learning_rate": 5e-05,
+      "loss": 0.0177,
+      "step": 635
+    },
+    {
+      "epoch": 1.545565006075334,
+      "grad_norm": 1.263848900794983,
+      "learning_rate": 5e-05,
+      "loss": 0.0659,
+      "step": 636
+    },
+    {
+      "epoch": 1.5479951397326852,
+      "grad_norm": 0.6971784830093384,
+      "learning_rate": 5e-05,
+      "loss": 0.0747,
+      "step": 637
+    },
+    {
+      "epoch": 1.5504252733900366,
+      "grad_norm": 0.5171895623207092,
+      "learning_rate": 5e-05,
+      "loss": 0.0512,
+      "step": 638
+    },
+    {
+      "epoch": 1.5528554070473874,
+      "grad_norm": 0.7318109273910522,
+      "learning_rate": 5e-05,
+      "loss": 0.0655,
+      "step": 639
+    },
+    {
+      "epoch": 1.5552855407047388,
+      "grad_norm": 0.46023720502853394,
+      "learning_rate": 5e-05,
+      "loss": 0.0568,
+      "step": 640
+    },
+    {
+      "epoch": 1.55771567436209,
+      "grad_norm": 0.7200696468353271,
+      "learning_rate": 5e-05,
+      "loss": 0.0511,
+      "step": 641
+    },
+    {
+      "epoch": 1.560145808019441,
+      "grad_norm": 1.0338624715805054,
+      "learning_rate": 5e-05,
+      "loss": 0.0822,
+      "step": 642
+    },
+    {
+      "epoch": 1.5625759416767924,
+      "grad_norm": 0.7976377010345459,
+      "learning_rate": 5e-05,
+      "loss": 0.0623,
+      "step": 643
+    },
+    {
+      "epoch": 1.5650060753341433,
+      "grad_norm": 0.6757969260215759,
+      "learning_rate": 5e-05,
+      "loss": 0.0737,
+      "step": 644
+    },
+    {
+      "epoch": 1.5674362089914946,
+      "grad_norm": 1.4083917140960693,
+      "learning_rate": 5e-05,
+      "loss": 0.0374,
+      "step": 645
+    },
+    {
+      "epoch": 1.5698663426488457,
+      "grad_norm": 0.348634272813797,
+      "learning_rate": 5e-05,
+      "loss": 0.0249,
+      "step": 646
+    },
+    {
+      "epoch": 1.5722964763061968,
+      "grad_norm": 0.7224584817886353,
+      "learning_rate": 5e-05,
+      "loss": 0.1013,
+      "step": 647
+    },
+    {
+      "epoch": 1.574726609963548,
+      "grad_norm": 0.919998288154602,
+      "learning_rate": 5e-05,
+      "loss": 0.1093,
+      "step": 648
+    },
+    {
+      "epoch": 1.577156743620899,
+      "grad_norm": 0.7341464161872864,
+      "learning_rate": 5e-05,
+      "loss": 0.1234,
+      "step": 649
+    },
+    {
+      "epoch": 1.5795868772782504,
+      "grad_norm": 0.8548154234886169,
+      "learning_rate": 5e-05,
+      "loss": 0.1628,
+      "step": 650
+    },
+    {
+      "epoch": 1.5820170109356013,
+      "grad_norm": 0.7931060194969177,
+      "learning_rate": 5e-05,
+      "loss": 0.1439,
+      "step": 651
+    },
+    {
+      "epoch": 1.5844471445929527,
+      "grad_norm": 0.9627839922904968,
+      "learning_rate": 5e-05,
+      "loss": 0.0327,
+      "step": 652
+    },
+    {
+      "epoch": 1.5868772782503038,
+      "grad_norm": 0.8406124114990234,
+      "learning_rate": 5e-05,
+      "loss": 0.0574,
+      "step": 653
+    },
+    {
+      "epoch": 1.589307411907655,
+      "grad_norm": 0.7667601108551025,
+      "learning_rate": 5e-05,
+      "loss": 0.0551,
+      "step": 654
+    },
+    {
+      "epoch": 1.5917375455650062,
+      "grad_norm": 0.4747026264667511,
+      "learning_rate": 5e-05,
+      "loss": 0.052,
+      "step": 655
+    },
+    {
+      "epoch": 1.5941676792223571,
+      "grad_norm": 0.632870078086853,
+      "learning_rate": 5e-05,
+      "loss": 0.0587,
+      "step": 656
+    },
+    {
+      "epoch": 1.5965978128797085,
+      "grad_norm": 0.7700726985931396,
+      "learning_rate": 5e-05,
+      "loss": 0.0986,
+      "step": 657
+    },
+    {
+      "epoch": 1.5990279465370594,
+      "grad_norm": 1.121201992034912,
+      "learning_rate": 5e-05,
+      "loss": 0.0318,
+      "step": 658
+    },
+    {
+      "epoch": 1.6014580801944107,
+      "grad_norm": 0.6213607788085938,
+      "learning_rate": 5e-05,
+      "loss": 0.0932,
+      "step": 659
+    },
+    {
+      "epoch": 1.6038882138517618,
+      "grad_norm": 0.30946460366249084,
+      "learning_rate": 5e-05,
+      "loss": 0.0333,
+      "step": 660
+    },
+    {
+      "epoch": 1.606318347509113,
+      "grad_norm": 0.5171956419944763,
+      "learning_rate": 5e-05,
+      "loss": 0.0617,
+      "step": 661
+    },
+    {
+      "epoch": 1.6087484811664643,
+      "grad_norm": 0.6288155317306519,
+      "learning_rate": 5e-05,
+      "loss": 0.0506,
+      "step": 662
+    },
+    {
+      "epoch": 1.6111786148238152,
+      "grad_norm": 0.6943643689155579,
+      "learning_rate": 5e-05,
+      "loss": 0.0285,
+      "step": 663
+    },
+    {
+      "epoch": 1.6136087484811665,
+      "grad_norm": 0.8144752383232117,
+      "learning_rate": 5e-05,
+      "loss": 0.0484,
+      "step": 664
+    },
+    {
+      "epoch": 1.6160388821385177,
+      "grad_norm": 0.4773041307926178,
+      "learning_rate": 5e-05,
+      "loss": 0.0993,
+      "step": 665
+    },
+    {
+      "epoch": 1.6184690157958688,
+      "grad_norm": 0.7703022956848145,
+      "learning_rate": 5e-05,
+      "loss": 0.0735,
+      "step": 666
+    },
+    {
+      "epoch": 1.62089914945322,
+      "grad_norm": 0.6951743960380554,
+      "learning_rate": 5e-05,
+      "loss": 0.0446,
+      "step": 667
+    },
+    {
+      "epoch": 1.623329283110571,
+      "grad_norm": 0.4781995415687561,
+      "learning_rate": 5e-05,
+      "loss": 0.0178,
+      "step": 668
+    },
+    {
+      "epoch": 1.6257594167679223,
+      "grad_norm": 0.5066447854042053,
+      "learning_rate": 5e-05,
+      "loss": 0.0969,
+      "step": 669
+    },
+    {
+      "epoch": 1.6281895504252732,
+      "grad_norm": 0.4288381040096283,
+      "learning_rate": 5e-05,
+      "loss": 0.0317,
+      "step": 670
+    },
+    {
+      "epoch": 1.6306196840826246,
+      "grad_norm": 0.5544601678848267,
+      "learning_rate": 5e-05,
+      "loss": 0.035,
+      "step": 671
+    },
+    {
+      "epoch": 1.6330498177399757,
+      "grad_norm": 0.48424339294433594,
+      "learning_rate": 5e-05,
+      "loss": 0.0231,
+      "step": 672
+    },
+    {
+      "epoch": 1.6354799513973268,
+      "grad_norm": 0.770346462726593,
+      "learning_rate": 5e-05,
+      "loss": 0.0767,
+      "step": 673
+    },
+    {
+      "epoch": 1.6379100850546782,
+      "grad_norm": 0.7101414203643799,
+      "learning_rate": 5e-05,
+      "loss": 0.0247,
+      "step": 674
+    },
+    {
+      "epoch": 1.640340218712029,
+      "grad_norm": 0.9238234758377075,
+      "learning_rate": 5e-05,
+      "loss": 0.0528,
+      "step": 675
+    },
+    {
+      "epoch": 1.6427703523693804,
+      "grad_norm": 0.7192332148551941,
+      "learning_rate": 5e-05,
+      "loss": 0.0738,
+      "step": 676
+    },
+    {
+      "epoch": 1.6452004860267315,
+      "grad_norm": 0.5393381118774414,
+      "learning_rate": 5e-05,
+      "loss": 0.0337,
+      "step": 677
+    },
+    {
+      "epoch": 1.6476306196840826,
+      "grad_norm": 0.2713688015937805,
+      "learning_rate": 5e-05,
+      "loss": 0.0159,
+      "step": 678
+    },
+    {
+      "epoch": 1.6500607533414338,
+      "grad_norm": 0.3296976089477539,
+      "learning_rate": 5e-05,
+      "loss": 0.0608,
+      "step": 679
+    },
+    {
+      "epoch": 1.6524908869987849,
+      "grad_norm": 0.4995911717414856,
+      "learning_rate": 5e-05,
+      "loss": 0.0119,
+      "step": 680
+    },
+    {
+      "epoch": 1.6549210206561362,
+      "grad_norm": 0.5983487367630005,
+      "learning_rate": 5e-05,
+      "loss": 0.0188,
+      "step": 681
+    },
+    {
+      "epoch": 1.6573511543134871,
+      "grad_norm": 0.4184721112251282,
+      "learning_rate": 5e-05,
+      "loss": 0.0207,
+      "step": 682
+    },
+    {
+      "epoch": 1.6597812879708385,
+      "grad_norm": 0.6056463122367859,
+      "learning_rate": 5e-05,
+      "loss": 0.0321,
+      "step": 683
+    },
+    {
+      "epoch": 1.6622114216281896,
+      "grad_norm": 0.8600037693977356,
+      "learning_rate": 5e-05,
+      "loss": 0.0716,
+      "step": 684
+    },
+    {
+      "epoch": 1.6646415552855407,
+      "grad_norm": 0.7863733768463135,
+      "learning_rate": 5e-05,
+      "loss": 0.1251,
+      "step": 685
+    },
+    {
+      "epoch": 1.6670716889428918,
+      "grad_norm": 0.4304426908493042,
+      "learning_rate": 5e-05,
+      "loss": 0.0824,
+      "step": 686
+    },
+    {
+      "epoch": 1.669501822600243,
+      "grad_norm": 0.6984259486198425,
+      "learning_rate": 5e-05,
+      "loss": 0.0332,
+      "step": 687
+    },
+    {
+      "epoch": 1.6719319562575943,
+      "grad_norm": 0.400698184967041,
+      "learning_rate": 5e-05,
+      "loss": 0.0444,
+      "step": 688
+    },
+    {
+      "epoch": 1.6743620899149452,
+      "grad_norm": 0.5730535387992859,
+      "learning_rate": 5e-05,
+      "loss": 0.0488,
+      "step": 689
+    },
+    {
+      "epoch": 1.6767922235722965,
+      "grad_norm": 0.4522067606449127,
+      "learning_rate": 5e-05,
+      "loss": 0.0596,
+      "step": 690
+    },
+    {
+      "epoch": 1.6792223572296476,
+      "grad_norm": 0.44992417097091675,
+      "learning_rate": 5e-05,
+      "loss": 0.0311,
+      "step": 691
+    },
+    {
+      "epoch": 1.6816524908869988,
+      "grad_norm": 0.8305267095565796,
+      "learning_rate": 5e-05,
+      "loss": 0.0933,
+      "step": 692
+    },
+    {
+      "epoch": 1.68408262454435,
+      "grad_norm": 0.7188434600830078,
+      "learning_rate": 5e-05,
+      "loss": 0.0443,
+      "step": 693
+    },
+    {
+      "epoch": 1.686512758201701,
+      "grad_norm": 0.5052474141120911,
+      "learning_rate": 5e-05,
+      "loss": 0.1067,
+      "step": 694
+    },
+    {
+      "epoch": 1.6889428918590523,
+      "grad_norm": 0.686128556728363,
+      "learning_rate": 5e-05,
+      "loss": 0.055,
+      "step": 695
+    },
+    {
+      "epoch": 1.6913730255164034,
+      "grad_norm": 0.4116796851158142,
+      "learning_rate": 5e-05,
+      "loss": 0.0728,
+      "step": 696
+    },
+    {
+      "epoch": 1.6938031591737546,
+      "grad_norm": 0.44450122117996216,
+      "learning_rate": 5e-05,
+      "loss": 0.0359,
+      "step": 697
+    },
+    {
+      "epoch": 1.6962332928311057,
+      "grad_norm": 0.3920440375804901,
+      "learning_rate": 5e-05,
+      "loss": 0.0654,
+      "step": 698
+    },
+    {
+      "epoch": 1.6986634264884568,
+      "grad_norm": 0.7238947749137878,
+      "learning_rate": 5e-05,
+      "loss": 0.0593,
+      "step": 699
+    },
+    {
+      "epoch": 1.7010935601458081,
+      "grad_norm": 0.3644774854183197,
+      "learning_rate": 5e-05,
+      "loss": 0.054,
+      "step": 700
+    },
+    {
+      "epoch": 1.7010935601458081,
+      "eval_loss": 0.23474913835525513,
+      "eval_runtime": 505.0402,
+      "eval_samples_per_second": 5.441,
+      "eval_steps_per_second": 0.681,
+      "step": 700
+    },
+    {
+      "epoch": 1.7010935601458081,
+      "step": 700,
+      "total_flos": 2.1333404857820774e+18,
+      "train_loss": 0.12394342508699213,
+      "train_runtime": 14191.792,
+      "train_samples_per_second": 2.255,
+      "train_steps_per_second": 0.07
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 100,
+  "total_flos": 2.1333404857820774e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}