diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
--- "a/last-checkpoint/trainer_state.json"
+++ "b/last-checkpoint/trainer_state.json"
@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.14742739200943536,
+  "epoch": 0.2948547840188707,
   "eval_steps": 500,
-  "global_step": 5000,
+  "global_step": 10000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -7094,6 +7094,7086 @@
       "eval_samples_per_second": 83.94,
       "eval_steps_per_second": 2.756,
       "step": 5000
+    },
+    {
+      "epoch": 0.1475748194014448,
+      "grad_norm": 23.375,
+      "learning_rate": 1.98625051877691e-06,
+      "loss": 2.3127,
+      "step": 5005
+    },
+    {
+      "epoch": 0.14772224679345422,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.9861653423891667e-06,
+      "loss": 2.3383,
+      "step": 5010
+    },
+    {
+      "epoch": 0.14786967418546365,
+      "grad_norm": 15.75,
+      "learning_rate": 1.9860799048243535e-06,
+      "loss": 2.3119,
+      "step": 5015
+    },
+    {
+      "epoch": 0.14801710157747308,
+      "grad_norm": 17.125,
+      "learning_rate": 1.9859942061050965e-06,
+      "loss": 2.3144,
+      "step": 5020
+    },
+    {
+      "epoch": 0.14816452896948254,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.985908246254093e-06,
+      "loss": 2.166,
+      "step": 5025
+    },
+    {
+      "epoch": 0.14831195636149197,
+      "grad_norm": 15.625,
+      "learning_rate": 1.985822025294108e-06,
+      "loss": 2.2456,
+      "step": 5030
+    },
+    {
+      "epoch": 0.1484593837535014,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.9857355432479763e-06,
+      "loss": 2.3008,
+      "step": 5035
+    },
+    {
+      "epoch": 0.14860681114551083,
+      "grad_norm": 16.25,
+      "learning_rate": 1.9856488001386026e-06,
+      "loss": 2.3406,
+      "step": 5040
+    },
+    {
+      "epoch": 0.14875423853752026,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.9855617959889598e-06,
+      "loss": 2.3188,
+      "step": 5045
+    },
+    {
+      "epoch": 0.14890166592952972,
+      "grad_norm": 19.25,
+      "learning_rate": 1.9854745308220895e-06,
+      "loss": 2.3037,
+      "step": 5050
+    },
+    {
+      "epoch": 0.14904909332153915,
+      "grad_norm": 16.875,
+      "learning_rate": 1.985387004661104e-06,
+      "loss": 2.2972,
+      "step": 5055
+    },
+    {
+      "epoch": 0.14919652071354858,
+      "grad_norm": 15.25,
+      "learning_rate": 1.9852992175291837e-06,
+      "loss": 2.3653,
+      "step": 5060
+    },
+    {
+      "epoch": 0.149343948105558,
+      "grad_norm": 18.375,
+      "learning_rate": 1.985211169449578e-06,
+      "loss": 2.3043,
+      "step": 5065
+    },
+    {
+      "epoch": 0.14949137549756744,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.9851228604456056e-06,
+      "loss": 2.3394,
+      "step": 5070
+    },
+    {
+      "epoch": 0.14963880288957687,
+      "grad_norm": 15.125,
+      "learning_rate": 1.985034290540654e-06,
+      "loss": 2.2842,
+      "step": 5075
+    },
+    {
+      "epoch": 0.14978623028158633,
+      "grad_norm": 13.5,
+      "learning_rate": 1.984945459758181e-06,
+      "loss": 2.2723,
+      "step": 5080
+    },
+    {
+      "epoch": 0.14993365767359576,
+      "grad_norm": 15.25,
+      "learning_rate": 1.984856368121712e-06,
+      "loss": 2.2761,
+      "step": 5085
+    },
+    {
+      "epoch": 0.1500810850656052,
+      "grad_norm": 13.875,
+      "learning_rate": 1.9847670156548424e-06,
+      "loss": 2.2788,
+      "step": 5090
+    },
+    {
+      "epoch": 0.15022851245761462,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.984677402381236e-06,
+      "loss": 2.3517,
+      "step": 5095
+    },
+    {
+      "epoch": 0.15037593984962405,
+      "grad_norm": 14.125,
+      "learning_rate": 1.9845875283246267e-06,
+      "loss": 2.2871,
+      "step": 5100
+    },
+    {
+      "epoch": 0.15052336724163348,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.984497393508817e-06,
+      "loss": 2.3584,
+      "step": 5105
+    },
+    {
+      "epoch": 0.15067079463364294,
+      "grad_norm": 15.625,
+      "learning_rate": 1.984406997957677e-06,
+      "loss": 2.4279,
+      "step": 5110
+    },
+    {
+      "epoch": 0.15081822202565237,
+      "grad_norm": 16.0,
+      "learning_rate": 1.984316341695148e-06,
+      "loss": 2.3569,
+      "step": 5115
+    },
+    {
+      "epoch": 0.1509656494176618,
+      "grad_norm": 26.0,
+      "learning_rate": 1.9842254247452402e-06,
+      "loss": 2.2929,
+      "step": 5120
+    },
+    {
+      "epoch": 0.15111307680967123,
+      "grad_norm": 12.625,
+      "learning_rate": 1.984134247132031e-06,
+      "loss": 2.2725,
+      "step": 5125
+    },
+    {
+      "epoch": 0.15126050420168066,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.984042808879668e-06,
+      "loss": 2.2815,
+      "step": 5130
+    },
+    {
+      "epoch": 0.15140793159369012,
+      "grad_norm": 17.5,
+      "learning_rate": 1.983951110012369e-06,
+      "loss": 2.2957,
+      "step": 5135
+    },
+    {
+      "epoch": 0.15155535898569955,
+      "grad_norm": 12.375,
+      "learning_rate": 1.9838591505544182e-06,
+      "loss": 2.2805,
+      "step": 5140
+    },
+    {
+      "epoch": 0.15170278637770898,
+      "grad_norm": 16.75,
+      "learning_rate": 1.983766930530171e-06,
+      "loss": 2.2924,
+      "step": 5145
+    },
+    {
+      "epoch": 0.1518502137697184,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.9836744499640515e-06,
+      "loss": 2.2116,
+      "step": 5150
+    },
+    {
+      "epoch": 0.15199764116172784,
+      "grad_norm": 15.125,
+      "learning_rate": 1.983581708880551e-06,
+      "loss": 2.3455,
+      "step": 5155
+    },
+    {
+      "epoch": 0.15214506855373727,
+      "grad_norm": 17.5,
+      "learning_rate": 1.983488707304232e-06,
+      "loss": 2.35,
+      "step": 5160
+    },
+    {
+      "epoch": 0.15229249594574673,
+      "grad_norm": 14.625,
+      "learning_rate": 1.9833954452597255e-06,
+      "loss": 2.2939,
+      "step": 5165
+    },
+    {
+      "epoch": 0.15243992333775616,
+      "grad_norm": 14.75,
+      "learning_rate": 1.9833019227717306e-06,
+      "loss": 2.3036,
+      "step": 5170
+    },
+    {
+      "epoch": 0.1525873507297656,
+      "grad_norm": 14.75,
+      "learning_rate": 1.9832081398650158e-06,
+      "loss": 2.353,
+      "step": 5175
+    },
+    {
+      "epoch": 0.15273477812177502,
+      "grad_norm": 13.375,
+      "learning_rate": 1.9831140965644187e-06,
+      "loss": 2.1931,
+      "step": 5180
+    },
+    {
+      "epoch": 0.15288220551378445,
+      "grad_norm": 16.0,
+      "learning_rate": 1.9830197928948464e-06,
+      "loss": 2.3679,
+      "step": 5185
+    },
+    {
+      "epoch": 0.15302963290579388,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.9829252288812735e-06,
+      "loss": 2.1672,
+      "step": 5190
+    },
+    {
+      "epoch": 0.15317706029780334,
+      "grad_norm": 14.5,
+      "learning_rate": 1.982830404548745e-06,
+      "loss": 2.1524,
+      "step": 5195
+    },
+    {
+      "epoch": 0.15332448768981277,
+      "grad_norm": 17.5,
+      "learning_rate": 1.9827353199223744e-06,
+      "loss": 2.4355,
+      "step": 5200
+    },
+    {
+      "epoch": 0.1534719150818222,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.9826399750273432e-06,
+      "loss": 2.3482,
+      "step": 5205
+    },
+    {
+      "epoch": 0.15361934247383163,
+      "grad_norm": 16.875,
+      "learning_rate": 1.9825443698889035e-06,
+      "loss": 2.3487,
+      "step": 5210
+    },
+    {
+      "epoch": 0.15376676986584106,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.982448504532375e-06,
+      "loss": 2.2663,
+      "step": 5215
+    },
+    {
+      "epoch": 0.15391419725785052,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.9823523789831474e-06,
+      "loss": 2.3294,
+      "step": 5220
+    },
+    {
+      "epoch": 0.15406162464985995,
+      "grad_norm": 16.25,
+      "learning_rate": 1.982255993266678e-06,
+      "loss": 2.2712,
+      "step": 5225
+    },
+    {
+      "epoch": 0.15420905204186938,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.9821593474084938e-06,
+      "loss": 2.3677,
+      "step": 5230
+    },
+    {
+      "epoch": 0.1543564794338788,
+      "grad_norm": 14.5,
+      "learning_rate": 1.982062441434191e-06,
+      "loss": 2.2574,
+      "step": 5235
+    },
+    {
+      "epoch": 0.15450390682588824,
+      "grad_norm": 49.0,
+      "learning_rate": 1.9819652753694336e-06,
+      "loss": 2.209,
+      "step": 5240
+    },
+    {
+      "epoch": 0.15465133421789767,
+      "grad_norm": 12.25,
+      "learning_rate": 1.9818678492399557e-06,
+      "loss": 2.2998,
+      "step": 5245
+    },
+    {
+      "epoch": 0.15479876160990713,
+      "grad_norm": 17.125,
+      "learning_rate": 1.98177016307156e-06,
+      "loss": 2.1183,
+      "step": 5250
+    },
+    {
+      "epoch": 0.15494618900191656,
+      "grad_norm": 15.625,
+      "learning_rate": 1.981672216890117e-06,
+      "loss": 2.2239,
+      "step": 5255
+    },
+    {
+      "epoch": 0.155093616393926,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9815740107215676e-06,
+      "loss": 2.2356,
+      "step": 5260
+    },
+    {
+      "epoch": 0.15524104378593542,
+      "grad_norm": 16.875,
+      "learning_rate": 1.9814755445919204e-06,
+      "loss": 2.3464,
+      "step": 5265
+    },
+    {
+      "epoch": 0.15538847117794485,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.9813768185272536e-06,
+      "loss": 2.3359,
+      "step": 5270
+    },
+    {
+      "epoch": 0.15553589856995428,
+      "grad_norm": 15.25,
+      "learning_rate": 1.981277832553713e-06,
+      "loss": 2.3663,
+      "step": 5275
+    },
+    {
+      "epoch": 0.15568332596196374,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.9811785866975153e-06,
+      "loss": 2.2262,
+      "step": 5280
+    },
+    {
+      "epoch": 0.15583075335397317,
+      "grad_norm": 15.375,
+      "learning_rate": 1.9810790809849446e-06,
+      "loss": 2.2454,
+      "step": 5285
+    },
+    {
+      "epoch": 0.1559781807459826,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.980979315442354e-06,
+      "loss": 2.289,
+      "step": 5290
+    },
+    {
+      "epoch": 0.15612560813799203,
+      "grad_norm": 15.375,
+      "learning_rate": 1.980879290096165e-06,
+      "loss": 2.2416,
+      "step": 5295
+    },
+    {
+      "epoch": 0.15627303553000146,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.9807790049728692e-06,
+      "loss": 2.2994,
+      "step": 5300
+    },
+    {
+      "epoch": 0.15642046292201092,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.9806784600990255e-06,
+      "loss": 2.3597,
+      "step": 5305
+    },
+    {
+      "epoch": 0.15656789031402035,
+      "grad_norm": 21.625,
+      "learning_rate": 1.980577655501263e-06,
+      "loss": 2.3614,
+      "step": 5310
+    },
+    {
+      "epoch": 0.15671531770602978,
+      "grad_norm": 12.375,
+      "learning_rate": 1.9804765912062786e-06,
+      "loss": 2.2174,
+      "step": 5315
+    },
+    {
+      "epoch": 0.1568627450980392,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.9803752672408385e-06,
+      "loss": 2.2548,
+      "step": 5320
+    },
+    {
+      "epoch": 0.15701017249004864,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.9802736836317767e-06,
+      "loss": 2.3221,
+      "step": 5325
+    },
+    {
+      "epoch": 0.15715759988205807,
+      "grad_norm": 19.25,
+      "learning_rate": 1.9801718404059973e-06,
+      "loss": 2.4205,
+      "step": 5330
+    },
+    {
+      "epoch": 0.15730502727406753,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.9800697375904727e-06,
+      "loss": 2.3176,
+      "step": 5335
+    },
+    {
+      "epoch": 0.15745245466607696,
+      "grad_norm": 15.0,
+      "learning_rate": 1.9799673752122436e-06,
+      "loss": 2.3065,
+      "step": 5340
+    },
+    {
+      "epoch": 0.1575998820580864,
+      "grad_norm": 37.5,
+      "learning_rate": 1.9798647532984197e-06,
+      "loss": 2.3744,
+      "step": 5345
+    },
+    {
+      "epoch": 0.15774730945009582,
+      "grad_norm": 15.875,
+      "learning_rate": 1.97976187187618e-06,
+      "loss": 2.222,
+      "step": 5350
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 22.0,
+      "learning_rate": 1.979658730972771e-06,
+      "loss": 2.2452,
+      "step": 5355
+    },
+    {
+      "epoch": 0.15804216423411468,
+      "grad_norm": 16.5,
+      "learning_rate": 1.9795553306155096e-06,
+      "loss": 2.3767,
+      "step": 5360
+    },
+    {
+      "epoch": 0.15818959162612414,
+      "grad_norm": 15.875,
+      "learning_rate": 1.9794516708317792e-06,
+      "loss": 2.2898,
+      "step": 5365
+    },
+    {
+      "epoch": 0.15833701901813357,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.9793477516490343e-06,
+      "loss": 2.3616,
+      "step": 5370
+    },
+    {
+      "epoch": 0.158484446410143,
+      "grad_norm": 15.375,
+      "learning_rate": 1.979243573094796e-06,
+      "loss": 2.2351,
+      "step": 5375
+    },
+    {
+      "epoch": 0.15863187380215243,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.979139135196656e-06,
+      "loss": 2.3049,
+      "step": 5380
+    },
+    {
+      "epoch": 0.15877930119416186,
+      "grad_norm": 16.625,
+      "learning_rate": 1.9790344379822735e-06,
+      "loss": 2.3058,
+      "step": 5385
+    },
+    {
+      "epoch": 0.15892672858617132,
+      "grad_norm": 19.75,
+      "learning_rate": 1.978929481479376e-06,
+      "loss": 2.4101,
+      "step": 5390
+    },
+    {
+      "epoch": 0.15907415597818075,
+      "grad_norm": 17.25,
+      "learning_rate": 1.9788242657157613e-06,
+      "loss": 2.2751,
+      "step": 5395
+    },
+    {
+      "epoch": 0.15922158337019018,
+      "grad_norm": 16.0,
+      "learning_rate": 1.9787187907192936e-06,
+      "loss": 2.253,
+      "step": 5400
+    },
+    {
+      "epoch": 0.15936901076219961,
+      "grad_norm": 15.5,
+      "learning_rate": 1.978613056517908e-06,
+      "loss": 2.3013,
+      "step": 5405
+    },
+    {
+      "epoch": 0.15951643815420904,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.9785070631396072e-06,
+      "loss": 2.2712,
+      "step": 5410
+    },
+    {
+      "epoch": 0.15966386554621848,
+      "grad_norm": 31.0,
+      "learning_rate": 1.978400810612462e-06,
+      "loss": 2.3801,
+      "step": 5415
+    },
+    {
+      "epoch": 0.15981129293822793,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.978294298964613e-06,
+      "loss": 2.1809,
+      "step": 5420
+    },
+    {
+      "epoch": 0.15995872033023736,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.978187528224269e-06,
+      "loss": 2.1765,
+      "step": 5425
+    },
+    {
+      "epoch": 0.1601061477222468,
+      "grad_norm": 13.875,
+      "learning_rate": 1.978080498419706e-06,
+      "loss": 2.2463,
+      "step": 5430
+    },
+    {
+      "epoch": 0.16025357511425622,
+      "grad_norm": 16.75,
+      "learning_rate": 1.9779732095792715e-06,
+      "loss": 2.1921,
+      "step": 5435
+    },
+    {
+      "epoch": 0.16040100250626566,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.977865661731379e-06,
+      "loss": 2.3557,
+      "step": 5440
+    },
+    {
+      "epoch": 0.16054842989827509,
+      "grad_norm": 12.0,
+      "learning_rate": 1.977757854904512e-06,
+      "loss": 2.2406,
+      "step": 5445
+    },
+    {
+      "epoch": 0.16069585729028454,
+      "grad_norm": 16.5,
+      "learning_rate": 1.9776497891272222e-06,
+      "loss": 2.2514,
+      "step": 5450
+    },
+    {
+      "epoch": 0.16084328468229397,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.9775414644281296e-06,
+      "loss": 2.3495,
+      "step": 5455
+    },
+    {
+      "epoch": 0.1609907120743034,
+      "grad_norm": 17.5,
+      "learning_rate": 1.977432880835923e-06,
+      "loss": 2.3667,
+      "step": 5460
+    },
+    {
+      "epoch": 0.16113813946631284,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.97732403837936e-06,
+      "loss": 2.2498,
+      "step": 5465
+    },
+    {
+      "epoch": 0.16128556685832227,
+      "grad_norm": 15.5,
+      "learning_rate": 1.9772149370872666e-06,
+      "loss": 2.2269,
+      "step": 5470
+    },
+    {
+      "epoch": 0.16143299425033172,
+      "grad_norm": 16.125,
+      "learning_rate": 1.9771055769885366e-06,
+      "loss": 2.3518,
+      "step": 5475
+    },
+    {
+      "epoch": 0.16158042164234115,
+      "grad_norm": 14.625,
+      "learning_rate": 1.9769959581121343e-06,
+      "loss": 2.1878,
+      "step": 5480
+    },
+    {
+      "epoch": 0.16172784903435058,
+      "grad_norm": 24.0,
+      "learning_rate": 1.9768860804870905e-06,
+      "loss": 2.3007,
+      "step": 5485
+    },
+    {
+      "epoch": 0.16187527642636002,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.9767759441425055e-06,
+      "loss": 2.3047,
+      "step": 5490
+    },
+    {
+      "epoch": 0.16202270381836945,
+      "grad_norm": 25.25,
+      "learning_rate": 1.9766655491075473e-06,
+      "loss": 2.2359,
+      "step": 5495
+    },
+    {
+      "epoch": 0.16217013121037888,
+      "grad_norm": 16.375,
+      "learning_rate": 1.976554895411454e-06,
+      "loss": 2.3287,
+      "step": 5500
+    },
+    {
+      "epoch": 0.16217013121037888,
+      "eval_loss": 2.275634765625,
+      "eval_runtime": 4.7203,
+      "eval_samples_per_second": 83.894,
+      "eval_steps_per_second": 2.754,
+      "step": 5500
+    },
+    {
+      "epoch": 0.16231755860238833,
+      "grad_norm": 16.375,
+      "learning_rate": 1.976443983083531e-06,
+      "loss": 2.1762,
+      "step": 5505
+    },
+    {
+      "epoch": 0.16246498599439776,
+      "grad_norm": 15.625,
+      "learning_rate": 1.9763328121531517e-06,
+      "loss": 2.3403,
+      "step": 5510
+    },
+    {
+      "epoch": 0.1626124133864072,
+      "grad_norm": 29.125,
+      "learning_rate": 1.9762213826497595e-06,
+      "loss": 2.2859,
+      "step": 5515
+    },
+    {
+      "epoch": 0.16275984077841663,
+      "grad_norm": 20.25,
+      "learning_rate": 1.9761096946028654e-06,
+      "loss": 2.2856,
+      "step": 5520
+    },
+    {
+      "epoch": 0.16290726817042606,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.9759977480420485e-06,
+      "loss": 2.282,
+      "step": 5525
+    },
+    {
+      "epoch": 0.16305469556243551,
+      "grad_norm": 17.25,
+      "learning_rate": 1.975885542996958e-06,
+      "loss": 2.3487,
+      "step": 5530
+    },
+    {
+      "epoch": 0.16320212295444494,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.9757730794973088e-06,
+      "loss": 2.2834,
+      "step": 5535
+    },
+    {
+      "epoch": 0.16334955034645438,
+      "grad_norm": 15.875,
+      "learning_rate": 1.975660357572887e-06,
+      "loss": 2.3061,
+      "step": 5540
+    },
+    {
+      "epoch": 0.1634969777384638,
+      "grad_norm": 14.875,
+      "learning_rate": 1.975547377253546e-06,
+      "loss": 2.3091,
+      "step": 5545
+    },
+    {
+      "epoch": 0.16364440513047324,
+      "grad_norm": 18.75,
+      "learning_rate": 1.9754341385692067e-06,
+      "loss": 2.3004,
+      "step": 5550
+    },
+    {
+      "epoch": 0.16379183252248267,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.97532064154986e-06,
+      "loss": 2.2878,
+      "step": 5555
+    },
+    {
+      "epoch": 0.16393925991449212,
+      "grad_norm": 14.125,
+      "learning_rate": 1.975206886225565e-06,
+      "loss": 2.2775,
+      "step": 5560
+    },
+    {
+      "epoch": 0.16408668730650156,
+      "grad_norm": 16.0,
+      "learning_rate": 1.975092872626448e-06,
+      "loss": 2.273,
+      "step": 5565
+    },
+    {
+      "epoch": 0.16423411469851099,
+      "grad_norm": 14.75,
+      "learning_rate": 1.9749786007827046e-06,
+      "loss": 2.2756,
+      "step": 5570
+    },
+    {
+      "epoch": 0.16438154209052042,
+      "grad_norm": 20.625,
+      "learning_rate": 1.9748640707245992e-06,
+      "loss": 2.2598,
+      "step": 5575
+    },
+    {
+      "epoch": 0.16452896948252985,
+      "grad_norm": 24.25,
+      "learning_rate": 1.9747492824824638e-06,
+      "loss": 2.2965,
+      "step": 5580
+    },
+    {
+      "epoch": 0.16467639687453928,
+      "grad_norm": 25.375,
+      "learning_rate": 1.9746342360866987e-06,
+      "loss": 2.2723,
+      "step": 5585
+    },
+    {
+      "epoch": 0.16482382426654874,
+      "grad_norm": 17.625,
+      "learning_rate": 1.974518931567773e-06,
+      "loss": 2.4182,
+      "step": 5590
+    },
+    {
+      "epoch": 0.16497125165855817,
+      "grad_norm": 13.25,
+      "learning_rate": 1.9744033689562244e-06,
+      "loss": 2.3571,
+      "step": 5595
+    },
+    {
+      "epoch": 0.1651186790505676,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.9742875482826583e-06,
+      "loss": 2.2016,
+      "step": 5600
+    },
+    {
+      "epoch": 0.16526610644257703,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.974171469577749e-06,
+      "loss": 2.1913,
+      "step": 5605
+    },
+    {
+      "epoch": 0.16541353383458646,
+      "grad_norm": 15.125,
+      "learning_rate": 1.9740551328722387e-06,
+      "loss": 2.3939,
+      "step": 5610
+    },
+    {
+      "epoch": 0.16556096122659592,
+      "grad_norm": 16.125,
+      "learning_rate": 1.9739385381969387e-06,
+      "loss": 2.31,
+      "step": 5615
+    },
+    {
+      "epoch": 0.16570838861860535,
+      "grad_norm": 14.375,
+      "learning_rate": 1.9738216855827276e-06,
+      "loss": 2.3633,
+      "step": 5620
+    },
+    {
+      "epoch": 0.16585581601061478,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.9737045750605523e-06,
+      "loss": 2.297,
+      "step": 5625
+    },
+    {
+      "epoch": 0.1660032434026242,
+      "grad_norm": 17.125,
+      "learning_rate": 1.9735872066614294e-06,
+      "loss": 2.3581,
+      "step": 5630
+    },
+    {
+      "epoch": 0.16615067079463364,
+      "grad_norm": 17.0,
+      "learning_rate": 1.973469580416442e-06,
+      "loss": 2.2865,
+      "step": 5635
+    },
+    {
+      "epoch": 0.16629809818664307,
+      "grad_norm": 13.375,
+      "learning_rate": 1.9733516963567433e-06,
+      "loss": 2.245,
+      "step": 5640
+    },
+    {
+      "epoch": 0.16644552557865253,
+      "grad_norm": 17.625,
+      "learning_rate": 1.973233554513553e-06,
+      "loss": 2.3806,
+      "step": 5645
+    },
+    {
+      "epoch": 0.16659295297066196,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.9731151549181603e-06,
+      "loss": 2.204,
+      "step": 5650
+    },
+    {
+      "epoch": 0.1667403803626714,
+      "grad_norm": 15.375,
+      "learning_rate": 1.9729964976019223e-06,
+      "loss": 2.1747,
+      "step": 5655
+    },
+    {
+      "epoch": 0.16688780775468082,
+      "grad_norm": 15.25,
+      "learning_rate": 1.972877582596264e-06,
+      "loss": 2.134,
+      "step": 5660
+    },
+    {
+      "epoch": 0.16703523514669025,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.9727584099326796e-06,
+      "loss": 2.3692,
+      "step": 5665
+    },
+    {
+      "epoch": 0.16718266253869968,
+      "grad_norm": 18.375,
+      "learning_rate": 1.9726389796427303e-06,
+      "loss": 2.3848,
+      "step": 5670
+    },
+    {
+      "epoch": 0.16733008993070914,
+      "grad_norm": 17.875,
+      "learning_rate": 1.9725192917580466e-06,
+      "loss": 2.2684,
+      "step": 5675
+    },
+    {
+      "epoch": 0.16747751732271857,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.9723993463103265e-06,
+      "loss": 2.3341,
+      "step": 5680
+    },
+    {
+      "epoch": 0.167624944714728,
+      "grad_norm": 15.25,
+      "learning_rate": 1.9722791433313364e-06,
+      "loss": 2.3528,
+      "step": 5685
+    },
+    {
+      "epoch": 0.16777237210673743,
+      "grad_norm": 16.75,
+      "learning_rate": 1.972158682852911e-06,
+      "loss": 2.2617,
+      "step": 5690
+    },
+    {
+      "epoch": 0.16791979949874686,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.9720379649069537e-06,
+      "loss": 2.2348,
+      "step": 5695
+    },
+    {
+      "epoch": 0.16806722689075632,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.9719169895254347e-06,
+      "loss": 2.2609,
+      "step": 5700
+    },
+    {
+      "epoch": 0.16821465428276575,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.971795756740394e-06,
+      "loss": 2.3166,
+      "step": 5705
+    },
+    {
+      "epoch": 0.16836208167477518,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.9716742665839387e-06,
+      "loss": 2.1999,
+      "step": 5710
+    },
+    {
+      "epoch": 0.1685095090667846,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.9715525190882444e-06,
+      "loss": 2.2247,
+      "step": 5715
+    },
+    {
+      "epoch": 0.16865693645879404,
+      "grad_norm": 15.625,
+      "learning_rate": 1.9714305142855545e-06,
+      "loss": 2.2155,
+      "step": 5720
+    },
+    {
+      "epoch": 0.16880436385080347,
+      "grad_norm": 18.125,
+      "learning_rate": 1.971308252208182e-06,
+      "loss": 2.2774,
+      "step": 5725
+    },
+    {
+      "epoch": 0.16895179124281293,
+      "grad_norm": 17.5,
+      "learning_rate": 1.9711857328885056e-06,
+      "loss": 2.3131,
+      "step": 5730
+    },
+    {
+      "epoch": 0.16909921863482236,
+      "grad_norm": 14.75,
+      "learning_rate": 1.971062956358974e-06,
+      "loss": 2.1557,
+      "step": 5735
+    },
+    {
+      "epoch": 0.1692466460268318,
+      "grad_norm": 19.0,
+      "learning_rate": 1.9709399226521034e-06,
+      "loss": 2.3339,
+      "step": 5740
+    },
+    {
+      "epoch": 0.16939407341884122,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.9708166318004785e-06,
+      "loss": 2.2882,
+      "step": 5745
+    },
+    {
+      "epoch": 0.16954150081085065,
+      "grad_norm": 20.375,
+      "learning_rate": 1.9706930838367513e-06,
+      "loss": 2.2205,
+      "step": 5750
+    },
+    {
+      "epoch": 0.16968892820286008,
+      "grad_norm": 13.625,
+      "learning_rate": 1.9705692787936427e-06,
+      "loss": 2.2539,
+      "step": 5755
+    },
+    {
+      "epoch": 0.16983635559486954,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.970445216703941e-06,
+      "loss": 2.2197,
+      "step": 5760
+    },
+    {
+      "epoch": 0.16998378298687897,
+      "grad_norm": 15.375,
+      "learning_rate": 1.9703208976005035e-06,
+      "loss": 2.3474,
+      "step": 5765
+    },
+    {
+      "epoch": 0.1701312103788884,
+      "grad_norm": 15.125,
+      "learning_rate": 1.9701963215162546e-06,
+      "loss": 2.2846,
+      "step": 5770
+    },
+    {
+      "epoch": 0.17027863777089783,
+      "grad_norm": 13.875,
+      "learning_rate": 1.9700714884841872e-06,
+      "loss": 2.2496,
+      "step": 5775
+    },
+    {
+      "epoch": 0.17042606516290726,
+      "grad_norm": 14.125,
+      "learning_rate": 1.9699463985373623e-06,
+      "loss": 2.2557,
+      "step": 5780
+    },
+    {
+      "epoch": 0.17057349255491672,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.9698210517089085e-06,
+      "loss": 2.3008,
+      "step": 5785
+    },
+    {
+      "epoch": 0.17072091994692615,
+      "grad_norm": 17.125,
+      "learning_rate": 1.9696954480320237e-06,
+      "loss": 2.3636,
+      "step": 5790
+    },
+    {
+      "epoch": 0.17086834733893558,
+      "grad_norm": 16.0,
+      "learning_rate": 1.9695695875399717e-06,
+      "loss": 2.3082,
+      "step": 5795
+    },
+    {
+      "epoch": 0.171015774730945,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.9694434702660866e-06,
+      "loss": 2.1362,
+      "step": 5800
+    },
+    {
+      "epoch": 0.17116320212295444,
+      "grad_norm": 18.625,
+      "learning_rate": 1.9693170962437686e-06,
+      "loss": 2.196,
+      "step": 5805
+    },
+    {
+      "epoch": 0.17131062951496387,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.9691904655064873e-06,
+      "loss": 2.2226,
+      "step": 5810
+    },
+    {
+      "epoch": 0.17145805690697333,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.9690635780877794e-06,
+      "loss": 2.2897,
+      "step": 5815
+    },
+    {
+      "epoch": 0.17160548429898276,
+      "grad_norm": 15.5,
+      "learning_rate": 1.96893643402125e-06,
+      "loss": 2.3065,
+      "step": 5820
+    },
+    {
+      "epoch": 0.1717529116909922,
+      "grad_norm": 16.5,
+      "learning_rate": 1.968809033340572e-06,
+      "loss": 2.2296,
+      "step": 5825
+    },
+    {
+      "epoch": 0.17190033908300162,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.9686813760794865e-06,
+      "loss": 2.3175,
+      "step": 5830
+    },
+    {
+      "epoch": 0.17204776647501105,
+      "grad_norm": 16.125,
+      "learning_rate": 1.9685534622718023e-06,
+      "loss": 2.2041,
+      "step": 5835
+    },
+    {
+      "epoch": 0.17219519386702048,
+      "grad_norm": 18.5,
+      "learning_rate": 1.9684252919513963e-06,
+      "loss": 2.2254,
+      "step": 5840
+    },
+    {
+      "epoch": 0.17234262125902994,
+      "grad_norm": 16.5,
+      "learning_rate": 1.9682968651522133e-06,
+      "loss": 2.3361,
+      "step": 5845
+    },
+    {
+      "epoch": 0.17249004865103937,
+      "grad_norm": 13.625,
+      "learning_rate": 1.9681681819082655e-06,
+      "loss": 2.275,
+      "step": 5850
+    },
+    {
+      "epoch": 0.1726374760430488,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.968039242253634e-06,
+      "loss": 2.199,
+      "step": 5855
+    },
+    {
+      "epoch": 0.17278490343505823,
+      "grad_norm": 20.5,
+      "learning_rate": 1.9679100462224673e-06,
+      "loss": 2.2255,
+      "step": 5860
+    },
+    {
+      "epoch": 0.17293233082706766,
+      "grad_norm": 19.75,
+      "learning_rate": 1.967780593848982e-06,
+      "loss": 2.3403,
+      "step": 5865
+    },
+    {
+      "epoch": 0.17307975821907712,
+      "grad_norm": 18.75,
+      "learning_rate": 1.9676508851674616e-06,
+      "loss": 2.2646,
+      "step": 5870
+    },
+    {
+      "epoch": 0.17322718561108655,
+      "grad_norm": 15.625,
+      "learning_rate": 1.9675209202122587e-06,
+      "loss": 2.3766,
+      "step": 5875
+    },
+    {
+      "epoch": 0.17337461300309598,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.967390699017794e-06,
+      "loss": 2.266,
+      "step": 5880
+    },
+    {
+      "epoch": 0.1735220403951054,
+      "grad_norm": 32.5,
+      "learning_rate": 1.9672602216185545e-06,
+      "loss": 2.1764,
+      "step": 5885
+    },
+    {
+      "epoch": 0.17366946778711484,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.9671294880490966e-06,
+      "loss": 2.1586,
+      "step": 5890
+    },
+    {
+      "epoch": 0.17381689517912427,
+      "grad_norm": 15.0,
+      "learning_rate": 1.9669984983440434e-06,
+      "loss": 2.2782,
+      "step": 5895
+    },
+    {
+      "epoch": 0.17396432257113373,
+      "grad_norm": 19.125,
+      "learning_rate": 1.9668672525380865e-06,
+      "loss": 2.1917,
+      "step": 5900
+    },
+    {
+      "epoch": 0.17411174996314316,
+      "grad_norm": 14.625,
+      "learning_rate": 1.9667357506659856e-06,
+      "loss": 2.29,
+      "step": 5905
+    },
+    {
+      "epoch": 0.1742591773551526,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.9666039927625673e-06,
+      "loss": 2.2455,
+      "step": 5910
+    },
+    {
+      "epoch": 0.17440660474716202,
+      "grad_norm": 16.5,
+      "learning_rate": 1.9664719788627267e-06,
+      "loss": 2.1826,
+      "step": 5915
+    },
+    {
+      "epoch": 0.17455403213917145,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.9663397090014265e-06,
+      "loss": 2.2362,
+      "step": 5920
+    },
+    {
+      "epoch": 0.17470145953118088,
+      "grad_norm": 20.75,
+      "learning_rate": 1.9662071832136973e-06,
+      "loss": 2.2673,
+      "step": 5925
+    },
+    {
+      "epoch": 0.17484888692319034,
+      "grad_norm": 17.625,
+      "learning_rate": 1.966074401534637e-06,
+      "loss": 2.2222,
+      "step": 5930
+    },
+    {
+      "epoch": 0.17499631431519977,
+      "grad_norm": 18.875,
+      "learning_rate": 1.9659413639994124e-06,
+      "loss": 2.2023,
+      "step": 5935
+    },
+    {
+      "epoch": 0.1751437417072092,
+      "grad_norm": 16.125,
+      "learning_rate": 1.965808070643256e-06,
+      "loss": 2.2542,
+      "step": 5940
+    },
+    {
+      "epoch": 0.17529116909921863,
+      "grad_norm": 17.0,
+      "learning_rate": 1.965674521501471e-06,
+      "loss": 2.2869,
+      "step": 5945
+    },
+    {
+      "epoch": 0.17543859649122806,
+      "grad_norm": 16.75,
+      "learning_rate": 1.965540716609425e-06,
+      "loss": 2.2464,
+      "step": 5950
+    },
+    {
+      "epoch": 0.17558602388323752,
+      "grad_norm": 14.75,
+      "learning_rate": 1.965406656002556e-06,
+      "loss": 2.242,
+      "step": 5955
+    },
+    {
+      "epoch": 0.17573345127524695,
+      "grad_norm": 12.5,
+      "learning_rate": 1.965272339716369e-06,
+      "loss": 2.2244,
+      "step": 5960
+    },
+    {
+      "epoch": 0.17588087866725638,
+      "grad_norm": 17.75,
+      "learning_rate": 1.965137767786436e-06,
+      "loss": 2.2567,
+      "step": 5965
+    },
+    {
+      "epoch": 0.1760283060592658,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.9650029402483974e-06,
+      "loss": 2.258,
+      "step": 5970
+    },
+    {
+      "epoch": 0.17617573345127524,
+      "grad_norm": 18.375,
+      "learning_rate": 1.9648678571379603e-06,
+      "loss": 2.268,
+      "step": 5975
+    },
+    {
+      "epoch": 0.17632316084328467,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.9647325184909014e-06,
+      "loss": 2.2409,
+      "step": 5980
+    },
+    {
+      "epoch": 0.17647058823529413,
+      "grad_norm": 15.75,
+      "learning_rate": 1.9645969243430632e-06,
+      "loss": 2.3359,
+      "step": 5985
+    },
+    {
+      "epoch": 0.17661801562730356,
+      "grad_norm": 17.625,
+      "learning_rate": 1.9644610747303567e-06,
+      "loss": 2.3641,
+      "step": 5990
+    },
+    {
+      "epoch": 0.176765443019313,
+      "grad_norm": 17.125,
+      "learning_rate": 1.9643249696887613e-06,
+      "loss": 2.2575,
+      "step": 5995
+    },
+    {
+      "epoch": 0.17691287041132242,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.9641886092543215e-06,
+      "loss": 2.3008,
+      "step": 6000
+    },
+    {
+      "epoch": 0.17691287041132242,
+      "eval_loss": 2.2403786182403564,
+      "eval_runtime": 4.7252,
+      "eval_samples_per_second": 83.807,
+      "eval_steps_per_second": 2.751,
+      "step": 6000
+    },
+    {
+      "epoch": 0.17706029780333185,
+      "grad_norm": 13.25,
+      "learning_rate": 1.9640519934631527e-06,
+      "loss": 2.1941,
+      "step": 6005
+    },
+    {
+      "epoch": 0.17720772519534128,
+      "grad_norm": 14.375,
+      "learning_rate": 1.9639151223514356e-06,
+      "loss": 2.3091,
+      "step": 6010
+    },
+    {
+      "epoch": 0.17735515258735074,
+      "grad_norm": 19.375,
+      "learning_rate": 1.9637779959554193e-06,
+      "loss": 2.3316,
+      "step": 6015
+    },
+    {
+      "epoch": 0.17750257997936017,
+      "grad_norm": 14.75,
+      "learning_rate": 1.963640614311421e-06,
+      "loss": 2.3345,
+      "step": 6020
+    },
+    {
+      "epoch": 0.1776500073713696,
+      "grad_norm": 17.125,
+      "learning_rate": 1.9635029774558245e-06,
+      "loss": 2.1962,
+      "step": 6025
+    },
+    {
+      "epoch": 0.17779743476337903,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.9633650854250818e-06,
+      "loss": 2.2806,
+      "step": 6030
+    },
+    {
+      "epoch": 0.17794486215538846,
+      "grad_norm": 17.25,
+      "learning_rate": 1.9632269382557123e-06,
+      "loss": 2.2432,
+      "step": 6035
+    },
+    {
+      "epoch": 0.17809228954739792,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.9630885359843034e-06,
+      "loss": 2.2453,
+      "step": 6040
+    },
+    {
+      "epoch": 0.17823971693940735,
+      "grad_norm": 13.25,
+      "learning_rate": 1.9629498786475094e-06,
+      "loss": 2.328,
+      "step": 6045
+    },
+    {
+      "epoch": 0.17838714433141678,
+      "grad_norm": 17.0,
+      "learning_rate": 1.9628109662820525e-06,
+      "loss": 2.2208,
+      "step": 6050
+    },
+    {
+      "epoch": 0.1785345717234262,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.9626717989247222e-06,
+      "loss": 2.2077,
+      "step": 6055
+    },
+    {
+      "epoch": 0.17868199911543564,
+      "grad_norm": 14.75,
+      "learning_rate": 1.9625323766123764e-06,
+      "loss": 2.2507,
+      "step": 6060
+    },
+    {
+      "epoch": 0.17882942650744507,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.9623926993819394e-06,
+      "loss": 2.2847,
+      "step": 6065
+    },
+    {
+      "epoch": 0.17897685389945453,
+      "grad_norm": 16.0,
+      "learning_rate": 1.962252767270403e-06,
+      "loss": 2.2177,
+      "step": 6070
+    },
+    {
+      "epoch": 0.17912428129146396,
+      "grad_norm": 17.875,
+      "learning_rate": 1.9621125803148275e-06,
+      "loss": 2.2478,
+      "step": 6075
+    },
+    {
+      "epoch": 0.1792717086834734,
+      "grad_norm": 14.875,
+      "learning_rate": 1.9619721385523404e-06,
+      "loss": 2.1835,
+      "step": 6080
+    },
+    {
+      "epoch": 0.17941913607548282,
+      "grad_norm": 18.125,
+      "learning_rate": 1.961831442020136e-06,
+      "loss": 2.3067,
+      "step": 6085
+    },
+    {
+      "epoch": 0.17956656346749225,
+      "grad_norm": 17.125,
+      "learning_rate": 1.961690490755477e-06,
+      "loss": 2.2633,
+      "step": 6090
+    },
+    {
+      "epoch": 0.17971399085950168,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.961549284795692e-06,
+      "loss": 2.2044,
+      "step": 6095
+    },
+    {
+      "epoch": 0.17986141825151114,
+      "grad_norm": 13.0,
+      "learning_rate": 1.9614078241781797e-06,
+      "loss": 2.1931,
+      "step": 6100
+    },
+    {
+      "epoch": 0.18000884564352057,
+      "grad_norm": 17.125,
+      "learning_rate": 1.961266108940403e-06,
+      "loss": 2.2065,
+      "step": 6105
+    },
+    {
+      "epoch": 0.18015627303553,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.9611241391198956e-06,
+      "loss": 2.2238,
+      "step": 6110
+    },
+    {
+      "epoch": 0.18030370042753943,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.9609819147542555e-06,
+      "loss": 2.2079,
+      "step": 6115
+    },
+    {
+      "epoch": 0.18045112781954886,
+      "grad_norm": 13.875,
+      "learning_rate": 1.9608394358811505e-06,
+      "loss": 2.176,
+      "step": 6120
+    },
+    {
+      "epoch": 0.18059855521155832,
+      "grad_norm": 15.5,
+      "learning_rate": 1.9606967025383147e-06,
+      "loss": 2.3294,
+      "step": 6125
+    },
+    {
+      "epoch": 0.18074598260356775,
+      "grad_norm": 13.375,
+      "learning_rate": 1.9605537147635493e-06,
+      "loss": 2.1568,
+      "step": 6130
+    },
+    {
+      "epoch": 0.18089340999557718,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.960410472594723e-06,
+      "loss": 2.1856,
+      "step": 6135
+    },
+    {
+      "epoch": 0.1810408373875866,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.9602669760697737e-06,
+      "loss": 2.3059,
+      "step": 6140
+    },
+    {
+      "epoch": 0.18118826477959604,
+      "grad_norm": 15.875,
+      "learning_rate": 1.960123225226704e-06,
+      "loss": 2.305,
+      "step": 6145
+    },
+    {
+      "epoch": 0.18133569217160547,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9599792201035852e-06,
+      "loss": 2.3145,
+      "step": 6150
+    },
+    {
+      "epoch": 0.18148311956361493,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.959834960738556e-06,
+      "loss": 2.2752,
+      "step": 6155
+    },
+    {
+      "epoch": 0.18163054695562436,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.9596904471698223e-06,
+      "loss": 2.2703,
+      "step": 6160
+    },
+    {
+      "epoch": 0.1817779743476338,
+      "grad_norm": 19.75,
+      "learning_rate": 1.9595456794356564e-06,
+      "loss": 2.1955,
+      "step": 6165
+    },
+    {
+      "epoch": 0.18192540173964322,
+      "grad_norm": 19.5,
+      "learning_rate": 1.9594006575743997e-06,
+      "loss": 2.3101,
+      "step": 6170
+    },
+    {
+      "epoch": 0.18207282913165265,
+      "grad_norm": 17.0,
+      "learning_rate": 1.9592553816244596e-06,
+      "loss": 2.153,
+      "step": 6175
+    },
+    {
+      "epoch": 0.18222025652366208,
+      "grad_norm": 15.375,
+      "learning_rate": 1.959109851624311e-06,
+      "loss": 2.2548,
+      "step": 6180
+    },
+    {
+      "epoch": 0.18236768391567154,
+      "grad_norm": 16.25,
+      "learning_rate": 1.9589640676124963e-06,
+      "loss": 2.1377,
+      "step": 6185
+    },
+    {
+      "epoch": 0.18251511130768097,
+      "grad_norm": 15.875,
+      "learning_rate": 1.9588180296276254e-06,
+      "loss": 2.2501,
+      "step": 6190
+    },
+    {
+      "epoch": 0.1826625386996904,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.9586717377083748e-06,
+      "loss": 2.277,
+      "step": 6195
+    },
+    {
+      "epoch": 0.18280996609169983,
+      "grad_norm": 12.0,
+      "learning_rate": 1.9585251918934884e-06,
+      "loss": 2.1835,
+      "step": 6200
+    },
+    {
+      "epoch": 0.18295739348370926,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.958378392221778e-06,
+      "loss": 2.0964,
+      "step": 6205
+    },
+    {
+      "epoch": 0.18310482087571872,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.958231338732122e-06,
+      "loss": 2.1945,
+      "step": 6210
+    },
+    {
+      "epoch": 0.18325224826772815,
+      "grad_norm": 15.375,
+      "learning_rate": 1.9580840314634665e-06,
+      "loss": 2.2639,
+      "step": 6215
+    },
+    {
+      "epoch": 0.18339967565973758,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.957936470454824e-06,
+      "loss": 2.2378,
+      "step": 6220
+    },
+    {
+      "epoch": 0.183547103051747,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.957788655745275e-06,
+      "loss": 2.2162,
+      "step": 6225
+    },
+    {
+      "epoch": 0.18369453044375644,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.9576405873739664e-06,
+      "loss": 2.2687,
+      "step": 6230
+    },
+    {
+      "epoch": 0.18384195783576587,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.9574922653801138e-06,
+      "loss": 2.2551,
+      "step": 6235
+    },
+    {
+      "epoch": 0.18398938522777533,
+      "grad_norm": 15.875,
+      "learning_rate": 1.957343689802998e-06,
+      "loss": 2.2938,
+      "step": 6240
+    },
+    {
+      "epoch": 0.18413681261978476,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.9571948606819687e-06,
+      "loss": 2.2451,
+      "step": 6245
+    },
+    {
+      "epoch": 0.1842842400117942,
+      "grad_norm": 17.75,
+      "learning_rate": 1.9570457780564415e-06,
+      "loss": 2.3438,
+      "step": 6250
+    },
+    {
+      "epoch": 0.18443166740380362,
+      "grad_norm": 14.375,
+      "learning_rate": 1.956896441965899e-06,
+      "loss": 2.2574,
+      "step": 6255
+    },
+    {
+      "epoch": 0.18457909479581305,
+      "grad_norm": 18.625,
+      "learning_rate": 1.956746852449893e-06,
+      "loss": 2.259,
+      "step": 6260
+    },
+    {
+      "epoch": 0.18472652218782248,
+      "grad_norm": 16.125,
+      "learning_rate": 1.95659700954804e-06,
+      "loss": 2.2494,
+      "step": 6265
+    },
+    {
+      "epoch": 0.18487394957983194,
+      "grad_norm": 21.375,
+      "learning_rate": 1.9564469133000244e-06,
+      "loss": 2.373,
+      "step": 6270
+    },
+    {
+      "epoch": 0.18502137697184137,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.9562965637455984e-06,
+      "loss": 2.2845,
+      "step": 6275
+    },
+    {
+      "epoch": 0.1851688043638508,
+      "grad_norm": 19.125,
+      "learning_rate": 1.95614596092458e-06,
+      "loss": 2.1535,
+      "step": 6280
+    },
+    {
+      "epoch": 0.18531623175586023,
+      "grad_norm": 15.5,
+      "learning_rate": 1.955995104876856e-06,
+      "loss": 2.2633,
+      "step": 6285
+    },
+    {
+      "epoch": 0.18546365914786966,
+      "grad_norm": 15.5,
+      "learning_rate": 1.9558439956423788e-06,
+      "loss": 2.3352,
+      "step": 6290
+    },
+    {
+      "epoch": 0.18561108653987912,
+      "grad_norm": 19.25,
+      "learning_rate": 1.955692633261168e-06,
+      "loss": 2.3492,
+      "step": 6295
+    },
+    {
+      "epoch": 0.18575851393188855,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.9555410177733108e-06,
+      "loss": 2.2918,
+      "step": 6300
+    },
+    {
+      "epoch": 0.18590594132389798,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.9553891492189613e-06,
+      "loss": 2.3091,
+      "step": 6305
+    },
+    {
+      "epoch": 0.1860533687159074,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.9552370276383406e-06,
+      "loss": 2.2345,
+      "step": 6310
+    },
+    {
+      "epoch": 0.18620079610791684,
+      "grad_norm": 15.25,
+      "learning_rate": 1.9550846530717368e-06,
+      "loss": 2.1788,
+      "step": 6315
+    },
+    {
+      "epoch": 0.18634822349992627,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.9549320255595044e-06,
+      "loss": 2.2549,
+      "step": 6320
+    },
+    {
+      "epoch": 0.18649565089193573,
+      "grad_norm": 13.75,
+      "learning_rate": 1.954779145142066e-06,
+      "loss": 2.2416,
+      "step": 6325
+    },
+    {
+      "epoch": 0.18664307828394516,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.9546260118599103e-06,
+      "loss": 2.1621,
+      "step": 6330
+    },
+    {
+      "epoch": 0.1867905056759546,
+      "grad_norm": 21.0,
+      "learning_rate": 1.9544726257535936e-06,
+      "loss": 2.3335,
+      "step": 6335
+    },
+    {
+      "epoch": 0.18693793306796402,
+      "grad_norm": 16.625,
+      "learning_rate": 1.9543189868637383e-06,
+      "loss": 2.2599,
+      "step": 6340
+    },
+    {
+      "epoch": 0.18708536045997345,
+      "grad_norm": 15.25,
+      "learning_rate": 1.954165095231035e-06,
+      "loss": 2.1533,
+      "step": 6345
+    },
+    {
+      "epoch": 0.18723278785198288,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.95401095089624e-06,
+      "loss": 2.1882,
+      "step": 6350
+    },
+    {
+      "epoch": 0.18738021524399234,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.9538565539001774e-06,
+      "loss": 2.2502,
+      "step": 6355
+    },
+    {
+      "epoch": 0.18752764263600177,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.953701904283737e-06,
+      "loss": 2.2742,
+      "step": 6360
+    },
+    {
+      "epoch": 0.1876750700280112,
+      "grad_norm": 14.375,
+      "learning_rate": 1.9535470020878776e-06,
+      "loss": 2.2428,
+      "step": 6365
+    },
+    {
+      "epoch": 0.18782249742002063,
+      "grad_norm": 13.125,
+      "learning_rate": 1.953391847353623e-06,
+      "loss": 2.1909,
+      "step": 6370
+    },
+    {
+      "epoch": 0.18796992481203006,
+      "grad_norm": 20.125,
+      "learning_rate": 1.9532364401220645e-06,
+      "loss": 2.3489,
+      "step": 6375
+    },
+    {
+      "epoch": 0.18811735220403952,
+      "grad_norm": 18.0,
+      "learning_rate": 1.9530807804343603e-06,
+      "loss": 2.224,
+      "step": 6380
+    },
+    {
+      "epoch": 0.18826477959604895,
+      "grad_norm": 15.0,
+      "learning_rate": 1.952924868331736e-06,
+      "loss": 2.2268,
+      "step": 6385
+    },
+    {
+      "epoch": 0.18841220698805838,
+      "grad_norm": 13.625,
+      "learning_rate": 1.9527687038554828e-06,
+      "loss": 2.1472,
+      "step": 6390
+    },
+    {
+      "epoch": 0.18855963438006781,
+      "grad_norm": 17.75,
+      "learning_rate": 1.9526122870469603e-06,
+      "loss": 2.3077,
+      "step": 6395
+    },
+    {
+      "epoch": 0.18870706177207724,
+      "grad_norm": 14.0,
+      "learning_rate": 1.952455617947593e-06,
+      "loss": 2.3211,
+      "step": 6400
+    },
+    {
+      "epoch": 0.18885448916408668,
+      "grad_norm": 16.375,
+      "learning_rate": 1.952298696598874e-06,
+      "loss": 2.2684,
+      "step": 6405
+    },
+    {
+      "epoch": 0.18900191655609613,
+      "grad_norm": 18.75,
+      "learning_rate": 1.952141523042363e-06,
+      "loss": 2.3083,
+      "step": 6410
+    },
+    {
+      "epoch": 0.18914934394810556,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.951984097319685e-06,
+      "loss": 2.3116,
+      "step": 6415
+    },
+    {
+      "epoch": 0.189296771340115,
+      "grad_norm": 13.5,
+      "learning_rate": 1.9518264194725333e-06,
+      "loss": 2.2399,
+      "step": 6420
+    },
+    {
+      "epoch": 0.18944419873212442,
+      "grad_norm": 15.5,
+      "learning_rate": 1.9516684895426676e-06,
+      "loss": 2.2946,
+      "step": 6425
+    },
+    {
+      "epoch": 0.18959162612413386,
+      "grad_norm": 14.0,
+      "learning_rate": 1.9515103075719133e-06,
+      "loss": 2.2819,
+      "step": 6430
+    },
+    {
+      "epoch": 0.18973905351614329,
+      "grad_norm": 17.125,
+      "learning_rate": 1.951351873602165e-06,
+      "loss": 2.3172,
+      "step": 6435
+    },
+    {
+      "epoch": 0.18988648090815274,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.9511931876753813e-06,
+      "loss": 2.1765,
+      "step": 6440
+    },
+    {
+      "epoch": 0.19003390830016217,
+      "grad_norm": 13.75,
+      "learning_rate": 1.9510342498335893e-06,
+      "loss": 2.1027,
+      "step": 6445
+    },
+    {
+      "epoch": 0.1901813356921716,
+      "grad_norm": 13.875,
+      "learning_rate": 1.9508750601188823e-06,
+      "loss": 2.1639,
+      "step": 6450
+    },
+    {
+      "epoch": 0.19032876308418104,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.95071561857342e-06,
+      "loss": 2.1655,
+      "step": 6455
+    },
+    {
+      "epoch": 0.19047619047619047,
+      "grad_norm": 18.375,
+      "learning_rate": 1.9505559252394292e-06,
+      "loss": 2.2215,
+      "step": 6460
+    },
+    {
+      "epoch": 0.19062361786819992,
+      "grad_norm": 15.75,
+      "learning_rate": 1.9503959801592035e-06,
+      "loss": 2.3045,
+      "step": 6465
+    },
+    {
+      "epoch": 0.19077104526020935,
+      "grad_norm": 22.25,
+      "learning_rate": 1.950235783375102e-06,
+      "loss": 2.3103,
+      "step": 6470
+    },
+    {
+      "epoch": 0.19091847265221878,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.9500753349295524e-06,
+      "loss": 2.2239,
+      "step": 6475
+    },
+    {
+      "epoch": 0.19106590004422822,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.9499146348650477e-06,
+      "loss": 2.2765,
+      "step": 6480
+    },
+    {
+      "epoch": 0.19121332743623765,
+      "grad_norm": 13.25,
+      "learning_rate": 1.949753683224148e-06,
+      "loss": 2.1889,
+      "step": 6485
+    },
+    {
+      "epoch": 0.19136075482824708,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9495924800494796e-06,
+      "loss": 2.1884,
+      "step": 6490
+    },
+    {
+      "epoch": 0.19150818222025653,
+      "grad_norm": 16.875,
+      "learning_rate": 1.9494310253837357e-06,
+      "loss": 2.2143,
+      "step": 6495
+    },
+    {
+      "epoch": 0.19165560961226596,
+      "grad_norm": 15.625,
+      "learning_rate": 1.9492693192696766e-06,
+      "loss": 2.2323,
+      "step": 6500
+    },
+    {
+      "epoch": 0.19165560961226596,
+      "eval_loss": 2.214259147644043,
+      "eval_runtime": 4.7188,
+      "eval_samples_per_second": 83.92,
+      "eval_steps_per_second": 2.755,
+      "step": 6500
+    },
+    {
+      "epoch": 0.1918030370042754,
+      "grad_norm": 19.75,
+      "learning_rate": 1.949107361750128e-06,
+      "loss": 2.2991,
+      "step": 6505
+    },
+    {
+      "epoch": 0.19195046439628483,
+      "grad_norm": 16.125,
+      "learning_rate": 1.948945152867984e-06,
+      "loss": 2.3435,
+      "step": 6510
+    },
+    {
+      "epoch": 0.19209789178829426,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.948782692666203e-06,
+      "loss": 2.2187,
+      "step": 6515
+    },
+    {
+      "epoch": 0.19224531918030371,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.9486199811878116e-06,
+      "loss": 2.2188,
+      "step": 6520
+    },
+    {
+      "epoch": 0.19239274657231314,
+      "grad_norm": 12.875,
+      "learning_rate": 1.9484570184759027e-06,
+      "loss": 2.2358,
+      "step": 6525
+    },
+    {
+      "epoch": 0.19254017396432258,
+      "grad_norm": 18.0,
+      "learning_rate": 1.9482938045736353e-06,
+      "loss": 2.2281,
+      "step": 6530
+    },
+    {
+      "epoch": 0.192687601356332,
+      "grad_norm": 13.875,
+      "learning_rate": 1.948130339524235e-06,
+      "loss": 2.2475,
+      "step": 6535
+    },
+    {
+      "epoch": 0.19283502874834144,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.9479666233709945e-06,
+      "loss": 2.1642,
+      "step": 6540
+    },
+    {
+      "epoch": 0.19298245614035087,
+      "grad_norm": 16.125,
+      "learning_rate": 1.947802656157272e-06,
+      "loss": 2.2591,
+      "step": 6545
+    },
+    {
+      "epoch": 0.19312988353236032,
+      "grad_norm": 17.375,
+      "learning_rate": 1.9476384379264933e-06,
+      "loss": 2.2818,
+      "step": 6550
+    },
+    {
+      "epoch": 0.19327731092436976,
+      "grad_norm": 21.875,
+      "learning_rate": 1.9474739687221494e-06,
+      "loss": 2.1193,
+      "step": 6555
+    },
+    {
+      "epoch": 0.19342473831637919,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.9473092485877994e-06,
+      "loss": 2.1854,
+      "step": 6560
+    },
+    {
+      "epoch": 0.19357216570838862,
+      "grad_norm": 14.625,
+      "learning_rate": 1.9471442775670673e-06,
+      "loss": 2.2582,
+      "step": 6565
+    },
+    {
+      "epoch": 0.19371959310039805,
+      "grad_norm": 16.75,
+      "learning_rate": 1.9469790557036443e-06,
+      "loss": 2.2225,
+      "step": 6570
+    },
+    {
+      "epoch": 0.19386702049240748,
+      "grad_norm": 17.0,
+      "learning_rate": 1.9468135830412886e-06,
+      "loss": 2.2606,
+      "step": 6575
+    },
+    {
+      "epoch": 0.19401444788441694,
+      "grad_norm": 16.875,
+      "learning_rate": 1.946647859623823e-06,
+      "loss": 2.1635,
+      "step": 6580
+    },
+    {
+      "epoch": 0.19416187527642637,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.9464818854951388e-06,
+      "loss": 2.3414,
+      "step": 6585
+    },
+    {
+      "epoch": 0.1943093026684358,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.9463156606991918e-06,
+      "loss": 2.4052,
+      "step": 6590
+    },
+    {
+      "epoch": 0.19445673006044523,
+      "grad_norm": 16.5,
+      "learning_rate": 1.9461491852800065e-06,
+      "loss": 2.129,
+      "step": 6595
+    },
+    {
+      "epoch": 0.19460415745245466,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.9459824592816716e-06,
+      "loss": 2.3136,
+      "step": 6600
+    },
+    {
+      "epoch": 0.19475158484446412,
+      "grad_norm": 15.125,
+      "learning_rate": 1.9458154827483427e-06,
+      "loss": 2.1992,
+      "step": 6605
+    },
+    {
+      "epoch": 0.19489901223647355,
+      "grad_norm": 17.375,
+      "learning_rate": 1.9456482557242427e-06,
+      "loss": 2.2089,
+      "step": 6610
+    },
+    {
+      "epoch": 0.19504643962848298,
+      "grad_norm": 17.875,
+      "learning_rate": 1.94548077825366e-06,
+      "loss": 2.1713,
+      "step": 6615
+    },
+    {
+      "epoch": 0.1951938670204924,
+      "grad_norm": 14.5,
+      "learning_rate": 1.945313050380949e-06,
+      "loss": 2.2417,
+      "step": 6620
+    },
+    {
+      "epoch": 0.19534129441250184,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.945145072150532e-06,
+      "loss": 2.3048,
+      "step": 6625
+    },
+    {
+      "epoch": 0.19548872180451127,
+      "grad_norm": 16.75,
+      "learning_rate": 1.9449768436068953e-06,
+      "loss": 2.1964,
+      "step": 6630
+    },
+    {
+      "epoch": 0.19563614919652073,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.944808364794594e-06,
+      "loss": 2.2846,
+      "step": 6635
+    },
+    {
+      "epoch": 0.19578357658853016,
+      "grad_norm": 17.75,
+      "learning_rate": 1.944639635758247e-06,
+      "loss": 2.2495,
+      "step": 6640
+    },
+    {
+      "epoch": 0.1959310039805396,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.944470656542541e-06,
+      "loss": 2.1223,
+      "step": 6645
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 17.5,
+      "learning_rate": 1.944301427192229e-06,
+      "loss": 2.2621,
+      "step": 6650
+    },
+    {
+      "epoch": 0.19622585876455845,
+      "grad_norm": 104.0,
+      "learning_rate": 1.94413194775213e-06,
+      "loss": 2.0909,
+      "step": 6655
+    },
+    {
+      "epoch": 0.19637328615656788,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.9439622182671282e-06,
+      "loss": 2.2395,
+      "step": 6660
+    },
+    {
+      "epoch": 0.19652071354857734,
+      "grad_norm": 14.125,
+      "learning_rate": 1.943792238782176e-06,
+      "loss": 2.2649,
+      "step": 6665
+    },
+    {
+      "epoch": 0.19666814094058677,
+      "grad_norm": 14.75,
+      "learning_rate": 1.9436220093422907e-06,
+      "loss": 2.3661,
+      "step": 6670
+    },
+    {
+      "epoch": 0.1968155683325962,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9434515299925557e-06,
+      "loss": 2.349,
+      "step": 6675
+    },
+    {
+      "epoch": 0.19696299572460563,
+      "grad_norm": 16.625,
+      "learning_rate": 1.943280800778121e-06,
+      "loss": 2.2771,
+      "step": 6680
+    },
+    {
+      "epoch": 0.19711042311661506,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.9431098217442027e-06,
+      "loss": 2.2262,
+      "step": 6685
+    },
+    {
+      "epoch": 0.19725785050862452,
+      "grad_norm": 14.375,
+      "learning_rate": 1.942938592936083e-06,
+      "loss": 2.2435,
+      "step": 6690
+    },
+    {
+      "epoch": 0.19740527790063395,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.9427671143991103e-06,
+      "loss": 2.1591,
+      "step": 6695
+    },
+    {
+      "epoch": 0.19755270529264338,
+      "grad_norm": 14.75,
+      "learning_rate": 1.9425953861787e-06,
+      "loss": 2.296,
+      "step": 6700
+    },
+    {
+      "epoch": 0.1977001326846528,
+      "grad_norm": 17.875,
+      "learning_rate": 1.942423408320332e-06,
+      "loss": 2.2397,
+      "step": 6705
+    },
+    {
+      "epoch": 0.19784756007666224,
+      "grad_norm": 12.25,
+      "learning_rate": 1.9422511808695525e-06,
+      "loss": 2.2015,
+      "step": 6710
+    },
+    {
+      "epoch": 0.19799498746867167,
+      "grad_norm": 15.25,
+      "learning_rate": 1.942078703871976e-06,
+      "loss": 2.2699,
+      "step": 6715
+    },
+    {
+      "epoch": 0.19814241486068113,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.94190597737328e-06,
+      "loss": 2.0769,
+      "step": 6720
+    },
+    {
+      "epoch": 0.19828984225269056,
+      "grad_norm": 14.375,
+      "learning_rate": 1.9417330014192103e-06,
+      "loss": 2.0982,
+      "step": 6725
+    },
+    {
+      "epoch": 0.1984372696447,
+      "grad_norm": 15.875,
+      "learning_rate": 1.941559776055578e-06,
+      "loss": 2.2998,
+      "step": 6730
+    },
+    {
+      "epoch": 0.19858469703670942,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.94138630132826e-06,
+      "loss": 2.1314,
+      "step": 6735
+    },
+    {
+      "epoch": 0.19873212442871885,
+      "grad_norm": 18.5,
+      "learning_rate": 1.9412125772832e-06,
+      "loss": 2.2776,
+      "step": 6740
+    },
+    {
+      "epoch": 0.19887955182072828,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.9410386039664067e-06,
+      "loss": 2.1857,
+      "step": 6745
+    },
+    {
+      "epoch": 0.19902697921273774,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.940864381423956e-06,
+      "loss": 2.2383,
+      "step": 6750
+    },
+    {
+      "epoch": 0.19917440660474717,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.9406899097019883e-06,
+      "loss": 2.2486,
+      "step": 6755
+    },
+    {
+      "epoch": 0.1993218339967566,
+      "grad_norm": 16.875,
+      "learning_rate": 1.940515188846712e-06,
+      "loss": 2.3064,
+      "step": 6760
+    },
+    {
+      "epoch": 0.19946926138876603,
+      "grad_norm": 14.0,
+      "learning_rate": 1.9403402189043994e-06,
+      "loss": 2.1471,
+      "step": 6765
+    },
+    {
+      "epoch": 0.19961668878077546,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.9401649999213904e-06,
+      "loss": 2.1606,
+      "step": 6770
+    },
+    {
+      "epoch": 0.19976411617278492,
+      "grad_norm": 19.0,
+      "learning_rate": 1.9399895319440893e-06,
+      "loss": 2.2425,
+      "step": 6775
+    },
+    {
+      "epoch": 0.19991154356479435,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9398138150189683e-06,
+      "loss": 2.1473,
+      "step": 6780
+    },
+    {
+      "epoch": 0.20005897095680378,
+      "grad_norm": 13.375,
+      "learning_rate": 1.9396378491925636e-06,
+      "loss": 2.1876,
+      "step": 6785
+    },
+    {
+      "epoch": 0.2002063983488132,
+      "grad_norm": 16.125,
+      "learning_rate": 1.939461634511479e-06,
+      "loss": 2.2899,
+      "step": 6790
+    },
+    {
+      "epoch": 0.20035382574082264,
+      "grad_norm": 12.625,
+      "learning_rate": 1.9392851710223823e-06,
+      "loss": 2.1244,
+      "step": 6795
+    },
+    {
+      "epoch": 0.20050125313283207,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.9391084587720093e-06,
+      "loss": 2.257,
+      "step": 6800
+    },
+    {
+      "epoch": 0.20064868052484153,
+      "grad_norm": 28.875,
+      "learning_rate": 1.93893149780716e-06,
+      "loss": 2.2221,
+      "step": 6805
+    },
+    {
+      "epoch": 0.20079610791685096,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.9387542881747016e-06,
+      "loss": 2.1064,
+      "step": 6810
+    },
+    {
+      "epoch": 0.2009435353088604,
+      "grad_norm": 13.125,
+      "learning_rate": 1.9385768299215656e-06,
+      "loss": 2.1955,
+      "step": 6815
+    },
+    {
+      "epoch": 0.20109096270086982,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.938399123094751e-06,
+      "loss": 2.2424,
+      "step": 6820
+    },
+    {
+      "epoch": 0.20123839009287925,
+      "grad_norm": 14.25,
+      "learning_rate": 1.9382211677413213e-06,
+      "loss": 2.1846,
+      "step": 6825
+    },
+    {
+      "epoch": 0.20138581748488868,
+      "grad_norm": 14.375,
+      "learning_rate": 1.938042963908407e-06,
+      "loss": 2.2189,
+      "step": 6830
+    },
+    {
+      "epoch": 0.20153324487689814,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.937864511643203e-06,
+      "loss": 2.3278,
+      "step": 6835
+    },
+    {
+      "epoch": 0.20168067226890757,
+      "grad_norm": 15.25,
+      "learning_rate": 1.9376858109929713e-06,
+      "loss": 2.2339,
+      "step": 6840
+    },
+    {
+      "epoch": 0.201828099660917,
+      "grad_norm": 16.375,
+      "learning_rate": 1.937506862005039e-06,
+      "loss": 2.2715,
+      "step": 6845
+    },
+    {
+      "epoch": 0.20197552705292643,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.9373276647267996e-06,
+      "loss": 2.0827,
+      "step": 6850
+    },
+    {
+      "epoch": 0.20212295444493586,
+      "grad_norm": 14.125,
+      "learning_rate": 1.9371482192057114e-06,
+      "loss": 2.2222,
+      "step": 6855
+    },
+    {
+      "epoch": 0.20227038183694532,
+      "grad_norm": 14.5,
+      "learning_rate": 1.936968525489299e-06,
+      "loss": 2.2901,
+      "step": 6860
+    },
+    {
+      "epoch": 0.20241780922895475,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.936788583625153e-06,
+      "loss": 2.0957,
+      "step": 6865
+    },
+    {
+      "epoch": 0.20256523662096418,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.936608393660929e-06,
+      "loss": 2.1918,
+      "step": 6870
+    },
+    {
+      "epoch": 0.2027126640129736,
+      "grad_norm": 19.625,
+      "learning_rate": 1.9364279556443486e-06,
+      "loss": 2.2613,
+      "step": 6875
+    },
+    {
+      "epoch": 0.20286009140498304,
+      "grad_norm": 14.75,
+      "learning_rate": 1.9362472696231994e-06,
+      "loss": 2.2336,
+      "step": 6880
+    },
+    {
+      "epoch": 0.20300751879699247,
+      "grad_norm": 12.75,
+      "learning_rate": 1.9360663356453344e-06,
+      "loss": 2.2267,
+      "step": 6885
+    },
+    {
+      "epoch": 0.20315494618900193,
+      "grad_norm": 16.375,
+      "learning_rate": 1.935885153758673e-06,
+      "loss": 2.311,
+      "step": 6890
+    },
+    {
+      "epoch": 0.20330237358101136,
+      "grad_norm": 14.625,
+      "learning_rate": 1.9357037240111985e-06,
+      "loss": 2.2432,
+      "step": 6895
+    },
+    {
+      "epoch": 0.2034498009730208,
+      "grad_norm": 11.75,
+      "learning_rate": 1.9355220464509617e-06,
+      "loss": 2.17,
+      "step": 6900
+    },
+    {
+      "epoch": 0.20359722836503022,
+      "grad_norm": 14.875,
+      "learning_rate": 1.935340121126078e-06,
+      "loss": 2.1693,
+      "step": 6905
+    },
+    {
+      "epoch": 0.20374465575703965,
+      "grad_norm": 14.875,
+      "learning_rate": 1.9351579480847288e-06,
+      "loss": 2.2669,
+      "step": 6910
+    },
+    {
+      "epoch": 0.20389208314904908,
+      "grad_norm": 17.25,
+      "learning_rate": 1.934975527375161e-06,
+      "loss": 2.2202,
+      "step": 6915
+    },
+    {
+      "epoch": 0.20403951054105854,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.9347928590456874e-06,
+      "loss": 2.2297,
+      "step": 6920
+    },
+    {
+      "epoch": 0.20418693793306797,
+      "grad_norm": 15.625,
+      "learning_rate": 1.9346099431446853e-06,
+      "loss": 2.1914,
+      "step": 6925
+    },
+    {
+      "epoch": 0.2043343653250774,
+      "grad_norm": 13.625,
+      "learning_rate": 1.9344267797205988e-06,
+      "loss": 2.2208,
+      "step": 6930
+    },
+    {
+      "epoch": 0.20448179271708683,
+      "grad_norm": 13.25,
+      "learning_rate": 1.934243368821937e-06,
+      "loss": 2.1804,
+      "step": 6935
+    },
+    {
+      "epoch": 0.20462922010909626,
+      "grad_norm": 17.0,
+      "learning_rate": 1.934059710497275e-06,
+      "loss": 2.2179,
+      "step": 6940
+    },
+    {
+      "epoch": 0.20477664750110572,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.9338758047952527e-06,
+      "loss": 2.1055,
+      "step": 6945
+    },
+    {
+      "epoch": 0.20492407489311515,
+      "grad_norm": 14.625,
+      "learning_rate": 1.9336916517645757e-06,
+      "loss": 2.2436,
+      "step": 6950
+    },
+    {
+      "epoch": 0.20507150228512458,
+      "grad_norm": 16.625,
+      "learning_rate": 1.933507251454016e-06,
+      "loss": 2.1477,
+      "step": 6955
+    },
+    {
+      "epoch": 0.205218929677134,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.933322603912409e-06,
+      "loss": 2.1234,
+      "step": 6960
+    },
+    {
+      "epoch": 0.20536635706914344,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.933137709188659e-06,
+      "loss": 2.257,
+      "step": 6965
+    },
+    {
+      "epoch": 0.20551378446115287,
+      "grad_norm": 13.875,
+      "learning_rate": 1.932952567331732e-06,
+      "loss": 2.1056,
+      "step": 6970
+    },
+    {
+      "epoch": 0.20566121185316233,
+      "grad_norm": 17.125,
+      "learning_rate": 1.9327671783906614e-06,
+      "loss": 2.161,
+      "step": 6975
+    },
+    {
+      "epoch": 0.20580863924517176,
+      "grad_norm": 25.875,
+      "learning_rate": 1.9325815424145465e-06,
+      "loss": 2.1589,
+      "step": 6980
+    },
+    {
+      "epoch": 0.2059560666371812,
+      "grad_norm": 16.625,
+      "learning_rate": 1.9323956594525514e-06,
+      "loss": 2.2228,
+      "step": 6985
+    },
+    {
+      "epoch": 0.20610349402919062,
+      "grad_norm": 18.0,
+      "learning_rate": 1.9322095295539045e-06,
+      "loss": 2.2412,
+      "step": 6990
+    },
+    {
+      "epoch": 0.20625092142120005,
+      "grad_norm": 15.125,
+      "learning_rate": 1.9320231527679014e-06,
+      "loss": 2.2868,
+      "step": 6995
+    },
+    {
+      "epoch": 0.20639834881320948,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.931836529143902e-06,
+      "loss": 2.1774,
+      "step": 7000
+    },
+    {
+      "epoch": 0.20639834881320948,
+      "eval_loss": 2.193103075027466,
+      "eval_runtime": 4.7184,
+      "eval_samples_per_second": 83.927,
+      "eval_steps_per_second": 2.755,
+      "step": 7000
+    },
+    {
+      "epoch": 0.20654577620521894,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.9316496587313323e-06,
+      "loss": 2.3368,
+      "step": 7005
+    },
+    {
+      "epoch": 0.20669320359722837,
+      "grad_norm": 13.625,
+      "learning_rate": 1.931462541579683e-06,
+      "loss": 2.3225,
+      "step": 7010
+    },
+    {
+      "epoch": 0.2068406309892378,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.9312751777385103e-06,
+      "loss": 2.2313,
+      "step": 7015
+    },
+    {
+      "epoch": 0.20698805838124723,
+      "grad_norm": 15.25,
+      "learning_rate": 1.931087567257436e-06,
+      "loss": 2.2718,
+      "step": 7020
+    },
+    {
+      "epoch": 0.20713548577325666,
+      "grad_norm": 13.375,
+      "learning_rate": 1.9308997101861474e-06,
+      "loss": 2.1488,
+      "step": 7025
+    },
+    {
+      "epoch": 0.20728291316526612,
+      "grad_norm": 14.25,
+      "learning_rate": 1.930711606574396e-06,
+      "loss": 2.2197,
+      "step": 7030
+    },
+    {
+      "epoch": 0.20743034055727555,
+      "grad_norm": 16.125,
+      "learning_rate": 1.930523256472e-06,
+      "loss": 2.1245,
+      "step": 7035
+    },
+    {
+      "epoch": 0.20757776794928498,
+      "grad_norm": 21.625,
+      "learning_rate": 1.9303346599288415e-06,
+      "loss": 2.2834,
+      "step": 7040
+    },
+    {
+      "epoch": 0.2077251953412944,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.9301458169948695e-06,
+      "loss": 2.145,
+      "step": 7045
+    },
+    {
+      "epoch": 0.20787262273330384,
+      "grad_norm": 13.875,
+      "learning_rate": 1.929956727720097e-06,
+      "loss": 2.206,
+      "step": 7050
+    },
+    {
+      "epoch": 0.20802005012531327,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.9297673921546026e-06,
+      "loss": 2.1265,
+      "step": 7055
+    },
+    {
+      "epoch": 0.20816747751732273,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.92957781034853e-06,
+      "loss": 2.283,
+      "step": 7060
+    },
+    {
+      "epoch": 0.20831490490933216,
+      "grad_norm": 15.125,
+      "learning_rate": 1.929387982352088e-06,
+      "loss": 2.2967,
+      "step": 7065
+    },
+    {
+      "epoch": 0.2084623323013416,
+      "grad_norm": 17.0,
+      "learning_rate": 1.9291979082155514e-06,
+      "loss": 2.2077,
+      "step": 7070
+    },
+    {
+      "epoch": 0.20860975969335102,
+      "grad_norm": 11.75,
+      "learning_rate": 1.9290075879892593e-06,
+      "loss": 2.1595,
+      "step": 7075
+    },
+    {
+      "epoch": 0.20875718708536045,
+      "grad_norm": 15.0,
+      "learning_rate": 1.9288170217236167e-06,
+      "loss": 2.272,
+      "step": 7080
+    },
+    {
+      "epoch": 0.20890461447736988,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.928626209469093e-06,
+      "loss": 2.2495,
+      "step": 7085
+    },
+    {
+      "epoch": 0.20905204186937934,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.9284351512762235e-06,
+      "loss": 2.202,
+      "step": 7090
+    },
+    {
+      "epoch": 0.20919946926138877,
+      "grad_norm": 17.5,
+      "learning_rate": 1.9282438471956074e-06,
+      "loss": 2.238,
+      "step": 7095
+    },
+    {
+      "epoch": 0.2093468966533982,
+      "grad_norm": 19.5,
+      "learning_rate": 1.9280522972779105e-06,
+      "loss": 2.2096,
+      "step": 7100
+    },
+    {
+      "epoch": 0.20949432404540763,
+      "grad_norm": 15.5,
+      "learning_rate": 1.9278605015738635e-06,
+      "loss": 2.3443,
+      "step": 7105
+    },
+    {
+      "epoch": 0.20964175143741706,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.927668460134261e-06,
+      "loss": 2.0769,
+      "step": 7110
+    },
+    {
+      "epoch": 0.20978917882942652,
+      "grad_norm": 17.125,
+      "learning_rate": 1.927476173009964e-06,
+      "loss": 2.1903,
+      "step": 7115
+    },
+    {
+      "epoch": 0.20993660622143595,
+      "grad_norm": 18.75,
+      "learning_rate": 1.9272836402518975e-06,
+      "loss": 2.3107,
+      "step": 7120
+    },
+    {
+      "epoch": 0.21008403361344538,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.927090861911053e-06,
+      "loss": 2.3182,
+      "step": 7125
+    },
+    {
+      "epoch": 0.2102314610054548,
+      "grad_norm": 16.75,
+      "learning_rate": 1.9268978380384846e-06,
+      "loss": 2.2175,
+      "step": 7130
+    },
+    {
+      "epoch": 0.21037888839746424,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.926704568685314e-06,
+      "loss": 2.219,
+      "step": 7135
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.9265110539027273e-06,
+      "loss": 2.1772,
+      "step": 7140
+    },
+    {
+      "epoch": 0.21067374318148313,
+      "grad_norm": 16.5,
+      "learning_rate": 1.9263172937419742e-06,
+      "loss": 2.2305,
+      "step": 7145
+    },
+    {
+      "epoch": 0.21082117057349256,
+      "grad_norm": 25.5,
+      "learning_rate": 1.9261232882543706e-06,
+      "loss": 2.3339,
+      "step": 7150
+    },
+    {
+      "epoch": 0.210968597965502,
+      "grad_norm": 13.375,
+      "learning_rate": 1.9259290374912976e-06,
+      "loss": 2.0968,
+      "step": 7155
+    },
+    {
+      "epoch": 0.21111602535751142,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.9257345415042e-06,
+      "loss": 2.1461,
+      "step": 7160
+    },
+    {
+      "epoch": 0.21126345274952085,
+      "grad_norm": 13.75,
+      "learning_rate": 1.9255398003445887e-06,
+      "loss": 2.2249,
+      "step": 7165
+    },
+    {
+      "epoch": 0.21141088014153028,
+      "grad_norm": 16.75,
+      "learning_rate": 1.9253448140640392e-06,
+      "loss": 2.3032,
+      "step": 7170
+    },
+    {
+      "epoch": 0.21155830753353974,
+      "grad_norm": 14.625,
+      "learning_rate": 1.925149582714192e-06,
+      "loss": 2.1457,
+      "step": 7175
+    },
+    {
+      "epoch": 0.21170573492554917,
+      "grad_norm": 14.75,
+      "learning_rate": 1.9249541063467524e-06,
+      "loss": 2.1607,
+      "step": 7180
+    },
+    {
+      "epoch": 0.2118531623175586,
+      "grad_norm": 16.375,
+      "learning_rate": 1.92475838501349e-06,
+      "loss": 2.1474,
+      "step": 7185
+    },
+    {
+      "epoch": 0.21200058970956803,
+      "grad_norm": 14.875,
+      "learning_rate": 1.92456241876624e-06,
+      "loss": 2.2473,
+      "step": 7190
+    },
+    {
+      "epoch": 0.21214801710157746,
+      "grad_norm": 13.75,
+      "learning_rate": 1.9243662076569034e-06,
+      "loss": 2.3273,
+      "step": 7195
+    },
+    {
+      "epoch": 0.21229544449358692,
+      "grad_norm": 26.5,
+      "learning_rate": 1.924169751737443e-06,
+      "loss": 2.253,
+      "step": 7200
+    },
+    {
+      "epoch": 0.21244287188559635,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.9239730510598906e-06,
+      "loss": 2.1251,
+      "step": 7205
+    },
+    {
+      "epoch": 0.21259029927760578,
+      "grad_norm": 14.25,
+      "learning_rate": 1.923776105676339e-06,
+      "loss": 2.1633,
+      "step": 7210
+    },
+    {
+      "epoch": 0.2127377266696152,
+      "grad_norm": 14.875,
+      "learning_rate": 1.923578915638948e-06,
+      "loss": 2.2504,
+      "step": 7215
+    },
+    {
+      "epoch": 0.21288515406162464,
+      "grad_norm": 14.375,
+      "learning_rate": 1.9233814809999417e-06,
+      "loss": 2.0792,
+      "step": 7220
+    },
+    {
+      "epoch": 0.21303258145363407,
+      "grad_norm": 13.25,
+      "learning_rate": 1.9231838018116084e-06,
+      "loss": 2.0842,
+      "step": 7225
+    },
+    {
+      "epoch": 0.21318000884564353,
+      "grad_norm": 15.875,
+      "learning_rate": 1.922985878126302e-06,
+      "loss": 2.373,
+      "step": 7230
+    },
+    {
+      "epoch": 0.21332743623765296,
+      "grad_norm": 14.0,
+      "learning_rate": 1.9227877099964413e-06,
+      "loss": 2.2181,
+      "step": 7235
+    },
+    {
+      "epoch": 0.2134748636296624,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.9225892974745083e-06,
+      "loss": 2.2031,
+      "step": 7240
+    },
+    {
+      "epoch": 0.21362229102167182,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.9223906406130515e-06,
+      "loss": 2.0738,
+      "step": 7245
+    },
+    {
+      "epoch": 0.21376971841368125,
+      "grad_norm": 13.375,
+      "learning_rate": 1.9221917394646833e-06,
+      "loss": 2.1385,
+      "step": 7250
+    },
+    {
+      "epoch": 0.21391714580569068,
+      "grad_norm": 13.875,
+      "learning_rate": 1.9219925940820813e-06,
+      "loss": 2.2051,
+      "step": 7255
+    },
+    {
+      "epoch": 0.21406457319770014,
+      "grad_norm": 13.875,
+      "learning_rate": 1.9217932045179864e-06,
+      "loss": 2.206,
+      "step": 7260
+    },
+    {
+      "epoch": 0.21421200058970957,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.921593570825206e-06,
+      "loss": 2.2147,
+      "step": 7265
+    },
+    {
+      "epoch": 0.214359427981719,
+      "grad_norm": 16.125,
+      "learning_rate": 1.921393693056611e-06,
+      "loss": 2.1987,
+      "step": 7270
+    },
+    {
+      "epoch": 0.21450685537372843,
+      "grad_norm": 19.375,
+      "learning_rate": 1.921193571265137e-06,
+      "loss": 2.268,
+      "step": 7275
+    },
+    {
+      "epoch": 0.21465428276573786,
+      "grad_norm": 13.875,
+      "learning_rate": 1.9209932055037844e-06,
+      "loss": 2.2881,
+      "step": 7280
+    },
+    {
+      "epoch": 0.21480171015774732,
+      "grad_norm": 16.125,
+      "learning_rate": 1.920792595825619e-06,
+      "loss": 2.1479,
+      "step": 7285
+    },
+    {
+      "epoch": 0.21494913754975675,
+      "grad_norm": 16.0,
+      "learning_rate": 1.92059174228377e-06,
+      "loss": 2.2282,
+      "step": 7290
+    },
+    {
+      "epoch": 0.21509656494176618,
+      "grad_norm": 14.5,
+      "learning_rate": 1.9203906449314315e-06,
+      "loss": 2.1761,
+      "step": 7295
+    },
+    {
+      "epoch": 0.2152439923337756,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.920189303821862e-06,
+      "loss": 2.2016,
+      "step": 7300
+    },
+    {
+      "epoch": 0.21539141972578504,
+      "grad_norm": 17.25,
+      "learning_rate": 1.9199877190083863e-06,
+      "loss": 2.0975,
+      "step": 7305
+    },
+    {
+      "epoch": 0.21553884711779447,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9197858905443916e-06,
+      "loss": 2.2305,
+      "step": 7310
+    },
+    {
+      "epoch": 0.21568627450980393,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.91958381848333e-06,
+      "loss": 2.1979,
+      "step": 7315
+    },
+    {
+      "epoch": 0.21583370190181336,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.919381502878718e-06,
+      "loss": 2.2607,
+      "step": 7320
+    },
+    {
+      "epoch": 0.2159811292938228,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.9191789437841384e-06,
+      "loss": 2.1914,
+      "step": 7325
+    },
+    {
+      "epoch": 0.21612855668583222,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.9189761412532365e-06,
+      "loss": 2.2348,
+      "step": 7330
+    },
+    {
+      "epoch": 0.21627598407784165,
+      "grad_norm": 14.0,
+      "learning_rate": 1.9187730953397225e-06,
+      "loss": 2.2635,
+      "step": 7335
+    },
+    {
+      "epoch": 0.21642341146985108,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.918569806097372e-06,
+      "loss": 2.2494,
+      "step": 7340
+    },
+    {
+      "epoch": 0.21657083886186054,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9183662735800237e-06,
+      "loss": 2.2639,
+      "step": 7345
+    },
+    {
+      "epoch": 0.21671826625386997,
+      "grad_norm": 16.125,
+      "learning_rate": 1.9181624978415814e-06,
+      "loss": 2.2761,
+      "step": 7350
+    },
+    {
+      "epoch": 0.2168656936458794,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.917958478936014e-06,
+      "loss": 2.3698,
+      "step": 7355
+    },
+    {
+      "epoch": 0.21701312103788883,
+      "grad_norm": 15.5,
+      "learning_rate": 1.917754216917353e-06,
+      "loss": 2.2325,
+      "step": 7360
+    },
+    {
+      "epoch": 0.21716054842989826,
+      "grad_norm": 19.375,
+      "learning_rate": 1.917549711839696e-06,
+      "loss": 2.2518,
+      "step": 7365
+    },
+    {
+      "epoch": 0.21730797582190772,
+      "grad_norm": 14.75,
+      "learning_rate": 1.9173449637572042e-06,
+      "loss": 2.2386,
+      "step": 7370
+    },
+    {
+      "epoch": 0.21745540321391715,
+      "grad_norm": 13.25,
+      "learning_rate": 1.9171399727241035e-06,
+      "loss": 2.2323,
+      "step": 7375
+    },
+    {
+      "epoch": 0.21760283060592658,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.9169347387946836e-06,
+      "loss": 2.207,
+      "step": 7380
+    },
+    {
+      "epoch": 0.21775025799793601,
+      "grad_norm": 19.0,
+      "learning_rate": 1.916729262023299e-06,
+      "loss": 2.2985,
+      "step": 7385
+    },
+    {
+      "epoch": 0.21789768538994544,
+      "grad_norm": 15.875,
+      "learning_rate": 1.916523542464369e-06,
+      "loss": 2.156,
+      "step": 7390
+    },
+    {
+      "epoch": 0.21804511278195488,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.916317580172376e-06,
+      "loss": 2.1725,
+      "step": 7395
+    },
+    {
+      "epoch": 0.21819254017396433,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.9161113752018666e-06,
+      "loss": 2.2422,
+      "step": 7400
+    },
+    {
+      "epoch": 0.21833996756597376,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.915904927607454e-06,
+      "loss": 2.2404,
+      "step": 7405
+    },
+    {
+      "epoch": 0.2184873949579832,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.915698237443812e-06,
+      "loss": 2.3074,
+      "step": 7410
+    },
+    {
+      "epoch": 0.21863482234999262,
+      "grad_norm": 15.25,
+      "learning_rate": 1.915491304765682e-06,
+      "loss": 2.1644,
+      "step": 7415
+    },
+    {
+      "epoch": 0.21878224974200206,
+      "grad_norm": 20.625,
+      "learning_rate": 1.9152841296278687e-06,
+      "loss": 2.2966,
+      "step": 7420
+    },
+    {
+      "epoch": 0.21892967713401149,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.915076712085239e-06,
+      "loss": 2.223,
+      "step": 7425
+    },
+    {
+      "epoch": 0.21907710452602094,
+      "grad_norm": 17.625,
+      "learning_rate": 1.9148690521927267e-06,
+      "loss": 2.2195,
+      "step": 7430
+    },
+    {
+      "epoch": 0.21922453191803037,
+      "grad_norm": 15.0,
+      "learning_rate": 1.9146611500053282e-06,
+      "loss": 2.2043,
+      "step": 7435
+    },
+    {
+      "epoch": 0.2193719593100398,
+      "grad_norm": 15.0,
+      "learning_rate": 1.914453005578105e-06,
+      "loss": 2.1418,
+      "step": 7440
+    },
+    {
+      "epoch": 0.21951938670204924,
+      "grad_norm": 18.0,
+      "learning_rate": 1.9142446189661818e-06,
+      "loss": 2.1438,
+      "step": 7445
+    },
+    {
+      "epoch": 0.21966681409405867,
+      "grad_norm": 15.25,
+      "learning_rate": 1.9140359902247485e-06,
+      "loss": 2.3102,
+      "step": 7450
+    },
+    {
+      "epoch": 0.21981424148606812,
+      "grad_norm": 13.75,
+      "learning_rate": 1.9138271194090576e-06,
+      "loss": 2.1613,
+      "step": 7455
+    },
+    {
+      "epoch": 0.21996166887807755,
+      "grad_norm": 16.5,
+      "learning_rate": 1.9136180065744278e-06,
+      "loss": 2.2573,
+      "step": 7460
+    },
+    {
+      "epoch": 0.22010909627008698,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.91340865177624e-06,
+      "loss": 2.1977,
+      "step": 7465
+    },
+    {
+      "epoch": 0.22025652366209642,
+      "grad_norm": 13.75,
+      "learning_rate": 1.91319905506994e-06,
+      "loss": 2.2246,
+      "step": 7470
+    },
+    {
+      "epoch": 0.22040395105410585,
+      "grad_norm": 14.125,
+      "learning_rate": 1.9129892165110383e-06,
+      "loss": 2.2511,
+      "step": 7475
+    },
+    {
+      "epoch": 0.22055137844611528,
+      "grad_norm": 13.375,
+      "learning_rate": 1.9127791361551077e-06,
+      "loss": 2.1218,
+      "step": 7480
+    },
+    {
+      "epoch": 0.22069880583812473,
+      "grad_norm": 14.0,
+      "learning_rate": 1.912568814057787e-06,
+      "loss": 2.1278,
+      "step": 7485
+    },
+    {
+      "epoch": 0.22084623323013416,
+      "grad_norm": 14.875,
+      "learning_rate": 1.9123582502747776e-06,
+      "loss": 2.177,
+      "step": 7490
+    },
+    {
+      "epoch": 0.2209936606221436,
+      "grad_norm": 18.0,
+      "learning_rate": 1.9121474448618455e-06,
+      "loss": 2.1431,
+      "step": 7495
+    },
+    {
+      "epoch": 0.22114108801415303,
+      "grad_norm": 16.0,
+      "learning_rate": 1.911936397874821e-06,
+      "loss": 2.2285,
+      "step": 7500
+    },
+    {
+      "epoch": 0.22114108801415303,
+      "eval_loss": 2.1763086318969727,
+      "eval_runtime": 4.7087,
+      "eval_samples_per_second": 84.099,
+      "eval_steps_per_second": 2.761,
+      "step": 7500
+    },
+    {
+      "epoch": 0.22128851540616246,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.911725109369598e-06,
+      "loss": 2.2166,
+      "step": 7505
+    },
+    {
+      "epoch": 0.22143594279817191,
+      "grad_norm": 27.875,
+      "learning_rate": 1.9115135794021336e-06,
+      "loss": 2.2402,
+      "step": 7510
+    },
+    {
+      "epoch": 0.22158337019018134,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.91130180802845e-06,
+      "loss": 2.204,
+      "step": 7515
+    },
+    {
+      "epoch": 0.22173079758219078,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.911089795304634e-06,
+      "loss": 2.1249,
+      "step": 7520
+    },
+    {
+      "epoch": 0.2218782249742002,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.9108775412868333e-06,
+      "loss": 2.1547,
+      "step": 7525
+    },
+    {
+      "epoch": 0.22202565236620964,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.9106650460312634e-06,
+      "loss": 2.2753,
+      "step": 7530
+    },
+    {
+      "epoch": 0.22217307975821907,
+      "grad_norm": 14.75,
+      "learning_rate": 1.9104523095942003e-06,
+      "loss": 2.1348,
+      "step": 7535
+    },
+    {
+      "epoch": 0.22232050715022852,
+      "grad_norm": 15.25,
+      "learning_rate": 1.910239332031986e-06,
+      "loss": 2.2014,
+      "step": 7540
+    },
+    {
+      "epoch": 0.22246793454223796,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.910026113401026e-06,
+      "loss": 2.2529,
+      "step": 7545
+    },
+    {
+      "epoch": 0.22261536193424739,
+      "grad_norm": 25.5,
+      "learning_rate": 1.909812653757789e-06,
+      "loss": 2.2115,
+      "step": 7550
+    },
+    {
+      "epoch": 0.22276278932625682,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.909598953158808e-06,
+      "loss": 2.228,
+      "step": 7555
+    },
+    {
+      "epoch": 0.22291021671826625,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.909385011660679e-06,
+      "loss": 2.2131,
+      "step": 7560
+    },
+    {
+      "epoch": 0.22305764411027568,
+      "grad_norm": 14.25,
+      "learning_rate": 1.9091708293200635e-06,
+      "loss": 2.1549,
+      "step": 7565
+    },
+    {
+      "epoch": 0.22320507150228514,
+      "grad_norm": 13.875,
+      "learning_rate": 1.908956406193685e-06,
+      "loss": 2.1435,
+      "step": 7570
+    },
+    {
+      "epoch": 0.22335249889429457,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.908741742338332e-06,
+      "loss": 2.2743,
+      "step": 7575
+    },
+    {
+      "epoch": 0.223499926286304,
+      "grad_norm": 23.625,
+      "learning_rate": 1.908526837810857e-06,
+      "loss": 2.2482,
+      "step": 7580
+    },
+    {
+      "epoch": 0.22364735367831343,
+      "grad_norm": 15.375,
+      "learning_rate": 1.9083116926681735e-06,
+      "loss": 2.348,
+      "step": 7585
+    },
+    {
+      "epoch": 0.22379478107032286,
+      "grad_norm": 15.625,
+      "learning_rate": 1.908096306967263e-06,
+      "loss": 2.1831,
+      "step": 7590
+    },
+    {
+      "epoch": 0.22394220846233232,
+      "grad_norm": 14.0,
+      "learning_rate": 1.907880680765167e-06,
+      "loss": 2.2471,
+      "step": 7595
+    },
+    {
+      "epoch": 0.22408963585434175,
+      "grad_norm": 15.5,
+      "learning_rate": 1.9076648141189925e-06,
+      "loss": 2.1452,
+      "step": 7600
+    },
+    {
+      "epoch": 0.22423706324635118,
+      "grad_norm": 14.375,
+      "learning_rate": 1.9074487070859102e-06,
+      "loss": 2.2139,
+      "step": 7605
+    },
+    {
+      "epoch": 0.2243844906383606,
+      "grad_norm": 13.375,
+      "learning_rate": 1.907232359723154e-06,
+      "loss": 2.0523,
+      "step": 7610
+    },
+    {
+      "epoch": 0.22453191803037004,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.9070157720880213e-06,
+      "loss": 2.2727,
+      "step": 7615
+    },
+    {
+      "epoch": 0.22467934542237947,
+      "grad_norm": 15.625,
+      "learning_rate": 1.9067989442378738e-06,
+      "loss": 2.2582,
+      "step": 7620
+    },
+    {
+      "epoch": 0.22482677281438893,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.9065818762301363e-06,
+      "loss": 2.1734,
+      "step": 7625
+    },
+    {
+      "epoch": 0.22497420020639836,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.906364568122297e-06,
+      "loss": 2.1759,
+      "step": 7630
+    },
+    {
+      "epoch": 0.2251216275984078,
+      "grad_norm": 14.625,
+      "learning_rate": 1.9061470199719083e-06,
+      "loss": 2.1728,
+      "step": 7635
+    },
+    {
+      "epoch": 0.22526905499041722,
+      "grad_norm": 16.625,
+      "learning_rate": 1.9059292318365855e-06,
+      "loss": 2.3187,
+      "step": 7640
+    },
+    {
+      "epoch": 0.22541648238242665,
+      "grad_norm": 15.0,
+      "learning_rate": 1.9057112037740084e-06,
+      "loss": 2.1401,
+      "step": 7645
+    },
+    {
+      "epoch": 0.22556390977443608,
+      "grad_norm": 15.0,
+      "learning_rate": 1.9054929358419195e-06,
+      "loss": 2.1785,
+      "step": 7650
+    },
+    {
+      "epoch": 0.22571133716644554,
+      "grad_norm": 15.875,
+      "learning_rate": 1.9052744280981251e-06,
+      "loss": 2.3024,
+      "step": 7655
+    },
+    {
+      "epoch": 0.22585876455845497,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.9050556806004955e-06,
+      "loss": 2.2835,
+      "step": 7660
+    },
+    {
+      "epoch": 0.2260061919504644,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.904836693406963e-06,
+      "loss": 2.2379,
+      "step": 7665
+    },
+    {
+      "epoch": 0.22615361934247383,
+      "grad_norm": 13.5,
+      "learning_rate": 1.9046174665755252e-06,
+      "loss": 2.1806,
+      "step": 7670
+    },
+    {
+      "epoch": 0.22630104673448326,
+      "grad_norm": 15.0,
+      "learning_rate": 1.904398000164242e-06,
+      "loss": 2.2357,
+      "step": 7675
+    },
+    {
+      "epoch": 0.22644847412649272,
+      "grad_norm": 19.125,
+      "learning_rate": 1.9041782942312374e-06,
+      "loss": 2.1981,
+      "step": 7680
+    },
+    {
+      "epoch": 0.22659590151850215,
+      "grad_norm": 15.0,
+      "learning_rate": 1.9039583488346987e-06,
+      "loss": 2.2131,
+      "step": 7685
+    },
+    {
+      "epoch": 0.22674332891051158,
+      "grad_norm": 15.375,
+      "learning_rate": 1.9037381640328757e-06,
+      "loss": 2.1719,
+      "step": 7690
+    },
+    {
+      "epoch": 0.226890756302521,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.9035177398840832e-06,
+      "loss": 2.308,
+      "step": 7695
+    },
+    {
+      "epoch": 0.22703818369453044,
+      "grad_norm": 10.875,
+      "learning_rate": 1.9032970764466981e-06,
+      "loss": 2.139,
+      "step": 7700
+    },
+    {
+      "epoch": 0.22718561108653987,
+      "grad_norm": 16.625,
+      "learning_rate": 1.9030761737791612e-06,
+      "loss": 2.1771,
+      "step": 7705
+    },
+    {
+      "epoch": 0.22733303847854933,
+      "grad_norm": 16.0,
+      "learning_rate": 1.9028550319399765e-06,
+      "loss": 2.1578,
+      "step": 7710
+    },
+    {
+      "epoch": 0.22748046587055876,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.9026336509877119e-06,
+      "loss": 2.174,
+      "step": 7715
+    },
+    {
+      "epoch": 0.2276278932625682,
+      "grad_norm": 13.625,
+      "learning_rate": 1.9024120309809978e-06,
+      "loss": 2.1622,
+      "step": 7720
+    },
+    {
+      "epoch": 0.22777532065457762,
+      "grad_norm": 19.375,
+      "learning_rate": 1.9021901719785282e-06,
+      "loss": 2.2129,
+      "step": 7725
+    },
+    {
+      "epoch": 0.22792274804658705,
+      "grad_norm": 17.0,
+      "learning_rate": 1.9019680740390607e-06,
+      "loss": 2.2368,
+      "step": 7730
+    },
+    {
+      "epoch": 0.22807017543859648,
+      "grad_norm": 13.25,
+      "learning_rate": 1.9017457372214158e-06,
+      "loss": 2.1988,
+      "step": 7735
+    },
+    {
+      "epoch": 0.22821760283060594,
+      "grad_norm": 12.625,
+      "learning_rate": 1.9015231615844773e-06,
+      "loss": 2.1546,
+      "step": 7740
+    },
+    {
+      "epoch": 0.22836503022261537,
+      "grad_norm": 16.5,
+      "learning_rate": 1.9013003471871932e-06,
+      "loss": 2.2133,
+      "step": 7745
+    },
+    {
+      "epoch": 0.2285124576146248,
+      "grad_norm": 13.875,
+      "learning_rate": 1.9010772940885727e-06,
+      "loss": 2.1367,
+      "step": 7750
+    },
+    {
+      "epoch": 0.22865988500663423,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.90085400234769e-06,
+      "loss": 2.2353,
+      "step": 7755
+    },
+    {
+      "epoch": 0.22880731239864366,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.9006304720236818e-06,
+      "loss": 2.1319,
+      "step": 7760
+    },
+    {
+      "epoch": 0.22895473979065312,
+      "grad_norm": 15.625,
+      "learning_rate": 1.9004067031757484e-06,
+      "loss": 2.1314,
+      "step": 7765
+    },
+    {
+      "epoch": 0.22910216718266255,
+      "grad_norm": 18.5,
+      "learning_rate": 1.9001826958631529e-06,
+      "loss": 2.1489,
+      "step": 7770
+    },
+    {
+      "epoch": 0.22924959457467198,
+      "grad_norm": 15.5,
+      "learning_rate": 1.8999584501452213e-06,
+      "loss": 2.2059,
+      "step": 7775
+    },
+    {
+      "epoch": 0.2293970219666814,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.8997339660813433e-06,
+      "loss": 2.0898,
+      "step": 7780
+    },
+    {
+      "epoch": 0.22954444935869084,
+      "grad_norm": 13.375,
+      "learning_rate": 1.8995092437309714e-06,
+      "loss": 2.1954,
+      "step": 7785
+    },
+    {
+      "epoch": 0.22969187675070027,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.8992842831536215e-06,
+      "loss": 2.2219,
+      "step": 7790
+    },
+    {
+      "epoch": 0.22983930414270973,
+      "grad_norm": 20.5,
+      "learning_rate": 1.8990590844088723e-06,
+      "loss": 2.333,
+      "step": 7795
+    },
+    {
+      "epoch": 0.22998673153471916,
+      "grad_norm": 31.5,
+      "learning_rate": 1.8988336475563654e-06,
+      "loss": 2.0667,
+      "step": 7800
+    },
+    {
+      "epoch": 0.2301341589267286,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.8986079726558064e-06,
+      "loss": 2.2898,
+      "step": 7805
+    },
+    {
+      "epoch": 0.23028158631873802,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.8983820597669626e-06,
+      "loss": 2.2644,
+      "step": 7810
+    },
+    {
+      "epoch": 0.23042901371074745,
+      "grad_norm": 13.625,
+      "learning_rate": 1.8981559089496652e-06,
+      "loss": 2.2789,
+      "step": 7815
+    },
+    {
+      "epoch": 0.23057644110275688,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.8979295202638086e-06,
+      "loss": 2.1742,
+      "step": 7820
+    },
+    {
+      "epoch": 0.23072386849476634,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.8977028937693493e-06,
+      "loss": 2.2069,
+      "step": 7825
+    },
+    {
+      "epoch": 0.23087129588677577,
+      "grad_norm": 15.625,
+      "learning_rate": 1.8974760295263075e-06,
+      "loss": 2.1841,
+      "step": 7830
+    },
+    {
+      "epoch": 0.2310187232787852,
+      "grad_norm": 17.875,
+      "learning_rate": 1.8972489275947662e-06,
+      "loss": 2.2387,
+      "step": 7835
+    },
+    {
+      "epoch": 0.23116615067079463,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.8970215880348713e-06,
+      "loss": 2.2129,
+      "step": 7840
+    },
+    {
+      "epoch": 0.23131357806280406,
+      "grad_norm": 17.75,
+      "learning_rate": 1.8967940109068316e-06,
+      "loss": 2.1558,
+      "step": 7845
+    },
+    {
+      "epoch": 0.23146100545481352,
+      "grad_norm": 14.625,
+      "learning_rate": 1.896566196270919e-06,
+      "loss": 2.2171,
+      "step": 7850
+    },
+    {
+      "epoch": 0.23160843284682295,
+      "grad_norm": 15.875,
+      "learning_rate": 1.896338144187468e-06,
+      "loss": 2.2006,
+      "step": 7855
+    },
+    {
+      "epoch": 0.23175586023883238,
+      "grad_norm": 16.125,
+      "learning_rate": 1.896109854716876e-06,
+      "loss": 2.1893,
+      "step": 7860
+    },
+    {
+      "epoch": 0.2319032876308418,
+      "grad_norm": 16.125,
+      "learning_rate": 1.895881327919604e-06,
+      "loss": 2.1611,
+      "step": 7865
+    },
+    {
+      "epoch": 0.23205071502285124,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.8956525638561749e-06,
+      "loss": 2.1714,
+      "step": 7870
+    },
+    {
+      "epoch": 0.23219814241486067,
+      "grad_norm": 16.25,
+      "learning_rate": 1.8954235625871747e-06,
+      "loss": 2.0627,
+      "step": 7875
+    },
+    {
+      "epoch": 0.23234556980687013,
+      "grad_norm": 16.625,
+      "learning_rate": 1.8951943241732526e-06,
+      "loss": 2.1204,
+      "step": 7880
+    },
+    {
+      "epoch": 0.23249299719887956,
+      "grad_norm": 13.5,
+      "learning_rate": 1.8949648486751202e-06,
+      "loss": 2.2194,
+      "step": 7885
+    },
+    {
+      "epoch": 0.232640424590889,
+      "grad_norm": 13.5,
+      "learning_rate": 1.8947351361535523e-06,
+      "loss": 2.1443,
+      "step": 7890
+    },
+    {
+      "epoch": 0.23278785198289842,
+      "grad_norm": 13.875,
+      "learning_rate": 1.8945051866693856e-06,
+      "loss": 2.244,
+      "step": 7895
+    },
+    {
+      "epoch": 0.23293527937490785,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.8942750002835205e-06,
+      "loss": 2.1937,
+      "step": 7900
+    },
+    {
+      "epoch": 0.23308270676691728,
+      "grad_norm": 17.375,
+      "learning_rate": 1.89404457705692e-06,
+      "loss": 2.3146,
+      "step": 7905
+    },
+    {
+      "epoch": 0.23323013415892674,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.8938139170506095e-06,
+      "loss": 2.1182,
+      "step": 7910
+    },
+    {
+      "epoch": 0.23337756155093617,
+      "grad_norm": 16.625,
+      "learning_rate": 1.8935830203256772e-06,
+      "loss": 2.272,
+      "step": 7915
+    },
+    {
+      "epoch": 0.2335249889429456,
+      "grad_norm": 16.25,
+      "learning_rate": 1.893351886943274e-06,
+      "loss": 2.1908,
+      "step": 7920
+    },
+    {
+      "epoch": 0.23367241633495503,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.8931205169646136e-06,
+      "loss": 2.2338,
+      "step": 7925
+    },
+    {
+      "epoch": 0.23381984372696446,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.8928889104509721e-06,
+      "loss": 2.1454,
+      "step": 7930
+    },
+    {
+      "epoch": 0.23396727111897392,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.8926570674636888e-06,
+      "loss": 2.2695,
+      "step": 7935
+    },
+    {
+      "epoch": 0.23411469851098335,
+      "grad_norm": 22.25,
+      "learning_rate": 1.8924249880641647e-06,
+      "loss": 2.1593,
+      "step": 7940
+    },
+    {
+      "epoch": 0.23426212590299278,
+      "grad_norm": 14.125,
+      "learning_rate": 1.8921926723138644e-06,
+      "loss": 2.1392,
+      "step": 7945
+    },
+    {
+      "epoch": 0.2344095532950022,
+      "grad_norm": 22.875,
+      "learning_rate": 1.8919601202743146e-06,
+      "loss": 2.1933,
+      "step": 7950
+    },
+    {
+      "epoch": 0.23455698068701164,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.8917273320071044e-06,
+      "loss": 2.1975,
+      "step": 7955
+    },
+    {
+      "epoch": 0.23470440807902107,
+      "grad_norm": 16.0,
+      "learning_rate": 1.8914943075738856e-06,
+      "loss": 2.1972,
+      "step": 7960
+    },
+    {
+      "epoch": 0.23485183547103053,
+      "grad_norm": 15.75,
+      "learning_rate": 1.891261047036373e-06,
+      "loss": 2.2886,
+      "step": 7965
+    },
+    {
+      "epoch": 0.23499926286303996,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.8910275504563434e-06,
+      "loss": 2.2055,
+      "step": 7970
+    },
+    {
+      "epoch": 0.2351466902550494,
+      "grad_norm": 16.625,
+      "learning_rate": 1.890793817895636e-06,
+      "loss": 2.1969,
+      "step": 7975
+    },
+    {
+      "epoch": 0.23529411764705882,
+      "grad_norm": 16.0,
+      "learning_rate": 1.8905598494161535e-06,
+      "loss": 2.1731,
+      "step": 7980
+    },
+    {
+      "epoch": 0.23544154503906825,
+      "grad_norm": 14.625,
+      "learning_rate": 1.8903256450798594e-06,
+      "loss": 2.2552,
+      "step": 7985
+    },
+    {
+      "epoch": 0.23558897243107768,
+      "grad_norm": 15.75,
+      "learning_rate": 1.890091204948782e-06,
+      "loss": 2.2014,
+      "step": 7990
+    },
+    {
+      "epoch": 0.23573639982308714,
+      "grad_norm": 16.375,
+      "learning_rate": 1.889856529085009e-06,
+      "loss": 2.2227,
+      "step": 7995
+    },
+    {
+      "epoch": 0.23588382721509657,
+      "grad_norm": 13.875,
+      "learning_rate": 1.8896216175506932e-06,
+      "loss": 2.1592,
+      "step": 8000
+    },
+    {
+      "epoch": 0.23588382721509657,
+      "eval_loss": 2.1633877754211426,
+      "eval_runtime": 4.717,
+      "eval_samples_per_second": 83.952,
+      "eval_steps_per_second": 2.756,
+      "step": 8000
+    },
+    {
+      "epoch": 0.236031254607106,
+      "grad_norm": 11.625,
+      "learning_rate": 1.889386470408049e-06,
+      "loss": 2.1509,
+      "step": 8005
+    },
+    {
+      "epoch": 0.23617868199911543,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.8891510877193522e-06,
+      "loss": 2.2325,
+      "step": 8010
+    },
+    {
+      "epoch": 0.23632610939112486,
+      "grad_norm": 14.25,
+      "learning_rate": 1.8889154695469424e-06,
+      "loss": 2.2458,
+      "step": 8015
+    },
+    {
+      "epoch": 0.23647353678313432,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.888679615953221e-06,
+      "loss": 2.2179,
+      "step": 8020
+    },
+    {
+      "epoch": 0.23662096417514375,
+      "grad_norm": 13.5,
+      "learning_rate": 1.8884435270006516e-06,
+      "loss": 2.1758,
+      "step": 8025
+    },
+    {
+      "epoch": 0.23676839156715318,
+      "grad_norm": 15.875,
+      "learning_rate": 1.8882072027517598e-06,
+      "loss": 2.1682,
+      "step": 8030
+    },
+    {
+      "epoch": 0.2369158189591626,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.8879706432691344e-06,
+      "loss": 2.1355,
+      "step": 8035
+    },
+    {
+      "epoch": 0.23706324635117204,
+      "grad_norm": 18.875,
+      "learning_rate": 1.8877338486154263e-06,
+      "loss": 2.1882,
+      "step": 8040
+    },
+    {
+      "epoch": 0.23721067374318147,
+      "grad_norm": 17.125,
+      "learning_rate": 1.8874968188533482e-06,
+      "loss": 2.2153,
+      "step": 8045
+    },
+    {
+      "epoch": 0.23735810113519093,
+      "grad_norm": 14.625,
+      "learning_rate": 1.887259554045675e-06,
+      "loss": 2.1607,
+      "step": 8050
+    },
+    {
+      "epoch": 0.23750552852720036,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.8870220542552445e-06,
+      "loss": 2.1877,
+      "step": 8055
+    },
+    {
+      "epoch": 0.2376529559192098,
+      "grad_norm": 53.5,
+      "learning_rate": 1.8867843195449563e-06,
+      "loss": 2.1403,
+      "step": 8060
+    },
+    {
+      "epoch": 0.23780038331121922,
+      "grad_norm": 12.125,
+      "learning_rate": 1.8865463499777724e-06,
+      "loss": 2.1845,
+      "step": 8065
+    },
+    {
+      "epoch": 0.23794781070322865,
+      "grad_norm": 14.75,
+      "learning_rate": 1.886308145616717e-06,
+      "loss": 2.1896,
+      "step": 8070
+    },
+    {
+      "epoch": 0.23809523809523808,
+      "grad_norm": 16.25,
+      "learning_rate": 1.8860697065248759e-06,
+      "loss": 2.2754,
+      "step": 8075
+    },
+    {
+      "epoch": 0.23824266548724754,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.8858310327653982e-06,
+      "loss": 2.178,
+      "step": 8080
+    },
+    {
+      "epoch": 0.23839009287925697,
+      "grad_norm": 14.625,
+      "learning_rate": 1.885592124401494e-06,
+      "loss": 2.2494,
+      "step": 8085
+    },
+    {
+      "epoch": 0.2385375202712664,
+      "grad_norm": 16.0,
+      "learning_rate": 1.8853529814964365e-06,
+      "loss": 2.1538,
+      "step": 8090
+    },
+    {
+      "epoch": 0.23868494766327583,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.88511360411356e-06,
+      "loss": 2.2637,
+      "step": 8095
+    },
+    {
+      "epoch": 0.23883237505528526,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.884873992316262e-06,
+      "loss": 2.2347,
+      "step": 8100
+    },
+    {
+      "epoch": 0.23897980244729472,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.8846341461680013e-06,
+      "loss": 2.2894,
+      "step": 8105
+    },
+    {
+      "epoch": 0.23912722983930415,
+      "grad_norm": 15.375,
+      "learning_rate": 1.884394065732299e-06,
+      "loss": 2.2071,
+      "step": 8110
+    },
+    {
+      "epoch": 0.23927465723131358,
+      "grad_norm": 16.625,
+      "learning_rate": 1.8841537510727383e-06,
+      "loss": 2.2701,
+      "step": 8115
+    },
+    {
+      "epoch": 0.239422084623323,
+      "grad_norm": 27.75,
+      "learning_rate": 1.8839132022529642e-06,
+      "loss": 2.1806,
+      "step": 8120
+    },
+    {
+      "epoch": 0.23956951201533244,
+      "grad_norm": 15.25,
+      "learning_rate": 1.8836724193366844e-06,
+      "loss": 2.1222,
+      "step": 8125
+    },
+    {
+      "epoch": 0.23971693940734187,
+      "grad_norm": 21.0,
+      "learning_rate": 1.8834314023876675e-06,
+      "loss": 2.2144,
+      "step": 8130
+    },
+    {
+      "epoch": 0.23986436679935133,
+      "grad_norm": 16.5,
+      "learning_rate": 1.8831901514697452e-06,
+      "loss": 2.1489,
+      "step": 8135
+    },
+    {
+      "epoch": 0.24001179419136076,
+      "grad_norm": 13.75,
+      "learning_rate": 1.8829486666468104e-06,
+      "loss": 2.1968,
+      "step": 8140
+    },
+    {
+      "epoch": 0.2401592215833702,
+      "grad_norm": 16.0,
+      "learning_rate": 1.882706947982818e-06,
+      "loss": 2.148,
+      "step": 8145
+    },
+    {
+      "epoch": 0.24030664897537962,
+      "grad_norm": 15.625,
+      "learning_rate": 1.8824649955417853e-06,
+      "loss": 2.0118,
+      "step": 8150
+    },
+    {
+      "epoch": 0.24045407636738905,
+      "grad_norm": 14.5,
+      "learning_rate": 1.8822228093877912e-06,
+      "loss": 2.2747,
+      "step": 8155
+    },
+    {
+      "epoch": 0.24060150375939848,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.8819803895849767e-06,
+      "loss": 2.2399,
+      "step": 8160
+    },
+    {
+      "epoch": 0.24074893115140794,
+      "grad_norm": 16.75,
+      "learning_rate": 1.881737736197544e-06,
+      "loss": 2.2886,
+      "step": 8165
+    },
+    {
+      "epoch": 0.24089635854341737,
+      "grad_norm": 22.625,
+      "learning_rate": 1.881494849289758e-06,
+      "loss": 2.1599,
+      "step": 8170
+    },
+    {
+      "epoch": 0.2410437859354268,
+      "grad_norm": 13.875,
+      "learning_rate": 1.8812517289259454e-06,
+      "loss": 2.1261,
+      "step": 8175
+    },
+    {
+      "epoch": 0.24119121332743623,
+      "grad_norm": 14.75,
+      "learning_rate": 1.881008375170494e-06,
+      "loss": 2.2204,
+      "step": 8180
+    },
+    {
+      "epoch": 0.24133864071944566,
+      "grad_norm": 14.0,
+      "learning_rate": 1.880764788087854e-06,
+      "loss": 2.1926,
+      "step": 8185
+    },
+    {
+      "epoch": 0.24148606811145512,
+      "grad_norm": 15.75,
+      "learning_rate": 1.8805209677425374e-06,
+      "loss": 2.1834,
+      "step": 8190
+    },
+    {
+      "epoch": 0.24163349550346455,
+      "grad_norm": 15.125,
+      "learning_rate": 1.8802769141991177e-06,
+      "loss": 2.2266,
+      "step": 8195
+    },
+    {
+      "epoch": 0.24178092289547398,
+      "grad_norm": 15.625,
+      "learning_rate": 1.8800326275222304e-06,
+      "loss": 2.3082,
+      "step": 8200
+    },
+    {
+      "epoch": 0.2419283502874834,
+      "grad_norm": 15.5,
+      "learning_rate": 1.8797881077765724e-06,
+      "loss": 2.2098,
+      "step": 8205
+    },
+    {
+      "epoch": 0.24207577767949284,
+      "grad_norm": 14.5,
+      "learning_rate": 1.8795433550269028e-06,
+      "loss": 2.139,
+      "step": 8210
+    },
+    {
+      "epoch": 0.24222320507150227,
+      "grad_norm": 25.625,
+      "learning_rate": 1.8792983693380424e-06,
+      "loss": 2.2472,
+      "step": 8215
+    },
+    {
+      "epoch": 0.24237063246351173,
+      "grad_norm": 15.5,
+      "learning_rate": 1.879053150774873e-06,
+      "loss": 2.2395,
+      "step": 8220
+    },
+    {
+      "epoch": 0.24251805985552116,
+      "grad_norm": 14.25,
+      "learning_rate": 1.8788076994023387e-06,
+      "loss": 2.1807,
+      "step": 8225
+    },
+    {
+      "epoch": 0.2426654872475306,
+      "grad_norm": 14.625,
+      "learning_rate": 1.8785620152854453e-06,
+      "loss": 2.2557,
+      "step": 8230
+    },
+    {
+      "epoch": 0.24281291463954002,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.8783160984892598e-06,
+      "loss": 2.1958,
+      "step": 8235
+    },
+    {
+      "epoch": 0.24296034203154945,
+      "grad_norm": 14.5,
+      "learning_rate": 1.878069949078911e-06,
+      "loss": 2.2767,
+      "step": 8240
+    },
+    {
+      "epoch": 0.24310776942355888,
+      "grad_norm": 14.25,
+      "learning_rate": 1.8778235671195897e-06,
+      "loss": 2.3034,
+      "step": 8245
+    },
+    {
+      "epoch": 0.24325519681556834,
+      "grad_norm": 15.375,
+      "learning_rate": 1.877576952676548e-06,
+      "loss": 2.2439,
+      "step": 8250
+    },
+    {
+      "epoch": 0.24340262420757777,
+      "grad_norm": 16.5,
+      "learning_rate": 1.877330105815099e-06,
+      "loss": 2.1794,
+      "step": 8255
+    },
+    {
+      "epoch": 0.2435500515995872,
+      "grad_norm": 16.0,
+      "learning_rate": 1.8770830266006182e-06,
+      "loss": 2.212,
+      "step": 8260
+    },
+    {
+      "epoch": 0.24369747899159663,
+      "grad_norm": 15.5,
+      "learning_rate": 1.8768357150985424e-06,
+      "loss": 2.0937,
+      "step": 8265
+    },
+    {
+      "epoch": 0.24384490638360606,
+      "grad_norm": 16.375,
+      "learning_rate": 1.8765881713743696e-06,
+      "loss": 2.2065,
+      "step": 8270
+    },
+    {
+      "epoch": 0.24399233377561552,
+      "grad_norm": 13.625,
+      "learning_rate": 1.87634039549366e-06,
+      "loss": 2.3326,
+      "step": 8275
+    },
+    {
+      "epoch": 0.24413976116762495,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.876092387522034e-06,
+      "loss": 2.1759,
+      "step": 8280
+    },
+    {
+      "epoch": 0.24428718855963438,
+      "grad_norm": 17.0,
+      "learning_rate": 1.875844147525175e-06,
+      "loss": 2.2042,
+      "step": 8285
+    },
+    {
+      "epoch": 0.2444346159516438,
+      "grad_norm": 25.375,
+      "learning_rate": 1.875595675568827e-06,
+      "loss": 2.2385,
+      "step": 8290
+    },
+    {
+      "epoch": 0.24458204334365324,
+      "grad_norm": 15.125,
+      "learning_rate": 1.8753469717187956e-06,
+      "loss": 2.1559,
+      "step": 8295
+    },
+    {
+      "epoch": 0.24472947073566267,
+      "grad_norm": 15.125,
+      "learning_rate": 1.8750980360409478e-06,
+      "loss": 2.1629,
+      "step": 8300
+    },
+    {
+      "epoch": 0.24487689812767213,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.8748488686012118e-06,
+      "loss": 2.136,
+      "step": 8305
+    },
+    {
+      "epoch": 0.24502432551968156,
+      "grad_norm": 16.0,
+      "learning_rate": 1.8745994694655775e-06,
+      "loss": 2.1955,
+      "step": 8310
+    },
+    {
+      "epoch": 0.245171752911691,
+      "grad_norm": 13.375,
+      "learning_rate": 1.874349838700096e-06,
+      "loss": 2.2115,
+      "step": 8315
+    },
+    {
+      "epoch": 0.24531918030370042,
+      "grad_norm": 16.5,
+      "learning_rate": 1.8740999763708798e-06,
+      "loss": 2.1912,
+      "step": 8320
+    },
+    {
+      "epoch": 0.24546660769570985,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.8738498825441025e-06,
+      "loss": 2.247,
+      "step": 8325
+    },
+    {
+      "epoch": 0.24561403508771928,
+      "grad_norm": 37.75,
+      "learning_rate": 1.8735995572859995e-06,
+      "loss": 2.2053,
+      "step": 8330
+    },
+    {
+      "epoch": 0.24576146247972874,
+      "grad_norm": 15.5,
+      "learning_rate": 1.8733490006628672e-06,
+      "loss": 2.1562,
+      "step": 8335
+    },
+    {
+      "epoch": 0.24590888987173817,
+      "grad_norm": 13.0,
+      "learning_rate": 1.873098212741063e-06,
+      "loss": 2.1676,
+      "step": 8340
+    },
+    {
+      "epoch": 0.2460563172637476,
+      "grad_norm": 13.75,
+      "learning_rate": 1.8728471935870062e-06,
+      "loss": 2.1155,
+      "step": 8345
+    },
+    {
+      "epoch": 0.24620374465575703,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.8725959432671767e-06,
+      "loss": 2.1968,
+      "step": 8350
+    },
+    {
+      "epoch": 0.24635117204776646,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.8723444618481162e-06,
+      "loss": 2.2811,
+      "step": 8355
+    },
+    {
+      "epoch": 0.24649859943977592,
+      "grad_norm": 14.375,
+      "learning_rate": 1.872092749396427e-06,
+      "loss": 2.177,
+      "step": 8360
+    },
+    {
+      "epoch": 0.24664602683178535,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.8718408059787727e-06,
+      "loss": 2.1862,
+      "step": 8365
+    },
+    {
+      "epoch": 0.24679345422379478,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.8715886316618787e-06,
+      "loss": 2.2133,
+      "step": 8370
+    },
+    {
+      "epoch": 0.24694088161580421,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.8713362265125313e-06,
+      "loss": 2.0847,
+      "step": 8375
+    },
+    {
+      "epoch": 0.24708830900781364,
+      "grad_norm": 16.75,
+      "learning_rate": 1.871083590597577e-06,
+      "loss": 2.2498,
+      "step": 8380
+    },
+    {
+      "epoch": 0.24723573639982308,
+      "grad_norm": 18.375,
+      "learning_rate": 1.8708307239839248e-06,
+      "loss": 2.2175,
+      "step": 8385
+    },
+    {
+      "epoch": 0.24738316379183253,
+      "grad_norm": 20.5,
+      "learning_rate": 1.8705776267385436e-06,
+      "loss": 2.2745,
+      "step": 8390
+    },
+    {
+      "epoch": 0.24753059118384196,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.8703242989284647e-06,
+      "loss": 2.0961,
+      "step": 8395
+    },
+    {
+      "epoch": 0.2476780185758514,
+      "grad_norm": 17.875,
+      "learning_rate": 1.8700707406207793e-06,
+      "loss": 2.2088,
+      "step": 8400
+    },
+    {
+      "epoch": 0.24782544596786082,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.8698169518826397e-06,
+      "loss": 2.1112,
+      "step": 8405
+    },
+    {
+      "epoch": 0.24797287335987026,
+      "grad_norm": 16.75,
+      "learning_rate": 1.86956293278126e-06,
+      "loss": 2.2312,
+      "step": 8410
+    },
+    {
+      "epoch": 0.24812030075187969,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.8693086833839148e-06,
+      "loss": 2.1858,
+      "step": 8415
+    },
+    {
+      "epoch": 0.24826772814388914,
+      "grad_norm": 14.75,
+      "learning_rate": 1.86905420375794e-06,
+      "loss": 2.2127,
+      "step": 8420
+    },
+    {
+      "epoch": 0.24841515553589857,
+      "grad_norm": 14.75,
+      "learning_rate": 1.868799493970732e-06,
+      "loss": 2.1444,
+      "step": 8425
+    },
+    {
+      "epoch": 0.248562582927908,
+      "grad_norm": 14.625,
+      "learning_rate": 1.8685445540897483e-06,
+      "loss": 2.2396,
+      "step": 8430
+    },
+    {
+      "epoch": 0.24871001031991744,
+      "grad_norm": 31.125,
+      "learning_rate": 1.8682893841825074e-06,
+      "loss": 2.1866,
+      "step": 8435
+    },
+    {
+      "epoch": 0.24885743771192687,
+      "grad_norm": 14.5,
+      "learning_rate": 1.8680339843165891e-06,
+      "loss": 2.1701,
+      "step": 8440
+    },
+    {
+      "epoch": 0.24900486510393632,
+      "grad_norm": 14.5,
+      "learning_rate": 1.8677783545596338e-06,
+      "loss": 2.1813,
+      "step": 8445
+    },
+    {
+      "epoch": 0.24915229249594575,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.8675224949793424e-06,
+      "loss": 2.2112,
+      "step": 8450
+    },
+    {
+      "epoch": 0.24929971988795518,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.8672664056434773e-06,
+      "loss": 2.2627,
+      "step": 8455
+    },
+    {
+      "epoch": 0.24944714727996462,
+      "grad_norm": 14.75,
+      "learning_rate": 1.8670100866198613e-06,
+      "loss": 2.2006,
+      "step": 8460
+    },
+    {
+      "epoch": 0.24959457467197405,
+      "grad_norm": 15.0,
+      "learning_rate": 1.866753537976378e-06,
+      "loss": 2.1875,
+      "step": 8465
+    },
+    {
+      "epoch": 0.24974200206398348,
+      "grad_norm": 13.625,
+      "learning_rate": 1.8664967597809729e-06,
+      "loss": 2.3218,
+      "step": 8470
+    },
+    {
+      "epoch": 0.24988942945599293,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.8662397521016503e-06,
+      "loss": 2.2054,
+      "step": 8475
+    },
+    {
+      "epoch": 0.25003685684800236,
+      "grad_norm": 17.375,
+      "learning_rate": 1.8659825150064773e-06,
+      "loss": 2.2152,
+      "step": 8480
+    },
+    {
+      "epoch": 0.25018428424001177,
+      "grad_norm": 16.25,
+      "learning_rate": 1.86572504856358e-06,
+      "loss": 2.0906,
+      "step": 8485
+    },
+    {
+      "epoch": 0.2503317116320212,
+      "grad_norm": 14.875,
+      "learning_rate": 1.8654673528411466e-06,
+      "loss": 2.1785,
+      "step": 8490
+    },
+    {
+      "epoch": 0.2504791390240307,
+      "grad_norm": 17.375,
+      "learning_rate": 1.8652094279074255e-06,
+      "loss": 2.3136,
+      "step": 8495
+    },
+    {
+      "epoch": 0.2506265664160401,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.8649512738307258e-06,
+      "loss": 2.1993,
+      "step": 8500
+    },
+    {
+      "epoch": 0.2506265664160401,
+      "eval_loss": 2.151705265045166,
+      "eval_runtime": 4.7212,
+      "eval_samples_per_second": 83.877,
+      "eval_steps_per_second": 2.754,
+      "step": 8500
+    },
+    {
+      "epoch": 0.25077399380804954,
+      "grad_norm": 17.125,
+      "learning_rate": 1.864692890679417e-06,
+      "loss": 2.1964,
+      "step": 8505
+    },
+    {
+      "epoch": 0.25092142120005895,
+      "grad_norm": 16.75,
+      "learning_rate": 1.86443427852193e-06,
+      "loss": 2.2787,
+      "step": 8510
+    },
+    {
+      "epoch": 0.2510688485920684,
+      "grad_norm": 14.625,
+      "learning_rate": 1.8641754374267558e-06,
+      "loss": 2.1101,
+      "step": 8515
+    },
+    {
+      "epoch": 0.25121627598407786,
+      "grad_norm": 14.75,
+      "learning_rate": 1.863916367462446e-06,
+      "loss": 2.26,
+      "step": 8520
+    },
+    {
+      "epoch": 0.25136370337608727,
+      "grad_norm": 14.125,
+      "learning_rate": 1.8636570686976127e-06,
+      "loss": 2.1509,
+      "step": 8525
+    },
+    {
+      "epoch": 0.2515111307680967,
+      "grad_norm": 16.125,
+      "learning_rate": 1.8633975412009294e-06,
+      "loss": 2.2034,
+      "step": 8530
+    },
+    {
+      "epoch": 0.2516585581601061,
+      "grad_norm": 15.125,
+      "learning_rate": 1.8631377850411293e-06,
+      "loss": 2.1605,
+      "step": 8535
+    },
+    {
+      "epoch": 0.2518059855521156,
+      "grad_norm": 15.75,
+      "learning_rate": 1.8628778002870069e-06,
+      "loss": 2.1062,
+      "step": 8540
+    },
+    {
+      "epoch": 0.25195341294412504,
+      "grad_norm": 17.0,
+      "learning_rate": 1.8626175870074165e-06,
+      "loss": 2.1868,
+      "step": 8545
+    },
+    {
+      "epoch": 0.25210084033613445,
+      "grad_norm": 17.125,
+      "learning_rate": 1.862357145271273e-06,
+      "loss": 2.2456,
+      "step": 8550
+    },
+    {
+      "epoch": 0.2522482677281439,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.8620964751475525e-06,
+      "loss": 2.2008,
+      "step": 8555
+    },
+    {
+      "epoch": 0.2523956951201533,
+      "grad_norm": 21.625,
+      "learning_rate": 1.861835576705291e-06,
+      "loss": 2.1544,
+      "step": 8560
+    },
+    {
+      "epoch": 0.25254312251216277,
+      "grad_norm": 14.625,
+      "learning_rate": 1.8615744500135855e-06,
+      "loss": 2.18,
+      "step": 8565
+    },
+    {
+      "epoch": 0.25269054990417217,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.8613130951415924e-06,
+      "loss": 2.1628,
+      "step": 8570
+    },
+    {
+      "epoch": 0.2528379772961816,
+      "grad_norm": 14.625,
+      "learning_rate": 1.8610515121585296e-06,
+      "loss": 2.1433,
+      "step": 8575
+    },
+    {
+      "epoch": 0.2529854046881911,
+      "grad_norm": 12.875,
+      "learning_rate": 1.8607897011336746e-06,
+      "loss": 2.1122,
+      "step": 8580
+    },
+    {
+      "epoch": 0.2531328320802005,
+      "grad_norm": 14.25,
+      "learning_rate": 1.8605276621363664e-06,
+      "loss": 2.1748,
+      "step": 8585
+    },
+    {
+      "epoch": 0.25328025947220995,
+      "grad_norm": 19.625,
+      "learning_rate": 1.8602653952360032e-06,
+      "loss": 2.1691,
+      "step": 8590
+    },
+    {
+      "epoch": 0.25342768686421935,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.8600029005020442e-06,
+      "loss": 2.2135,
+      "step": 8595
+    },
+    {
+      "epoch": 0.2535751142562288,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.8597401780040088e-06,
+      "loss": 2.2858,
+      "step": 8600
+    },
+    {
+      "epoch": 0.25372254164823826,
+      "grad_norm": 21.0,
+      "learning_rate": 1.8594772278114764e-06,
+      "loss": 2.2727,
+      "step": 8605
+    },
+    {
+      "epoch": 0.25386996904024767,
+      "grad_norm": 16.125,
+      "learning_rate": 1.8592140499940876e-06,
+      "loss": 2.1415,
+      "step": 8610
+    },
+    {
+      "epoch": 0.2540173964322571,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.858950644621542e-06,
+      "loss": 2.1434,
+      "step": 8615
+    },
+    {
+      "epoch": 0.25416482382426653,
+      "grad_norm": 13.375,
+      "learning_rate": 1.858687011763601e-06,
+      "loss": 2.2154,
+      "step": 8620
+    },
+    {
+      "epoch": 0.254312251216276,
+      "grad_norm": 15.5,
+      "learning_rate": 1.8584231514900842e-06,
+      "loss": 2.2386,
+      "step": 8625
+    },
+    {
+      "epoch": 0.25445967860828544,
+      "grad_norm": 16.0,
+      "learning_rate": 1.8581590638708739e-06,
+      "loss": 2.2126,
+      "step": 8630
+    },
+    {
+      "epoch": 0.25460710600029485,
+      "grad_norm": 19.125,
+      "learning_rate": 1.8578947489759105e-06,
+      "loss": 2.2209,
+      "step": 8635
+    },
+    {
+      "epoch": 0.2547545333923043,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.8576302068751958e-06,
+      "loss": 2.2278,
+      "step": 8640
+    },
+    {
+      "epoch": 0.2549019607843137,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.8573654376387915e-06,
+      "loss": 2.0973,
+      "step": 8645
+    },
+    {
+      "epoch": 0.25504938817632317,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.857100441336819e-06,
+      "loss": 2.1817,
+      "step": 8650
+    },
+    {
+      "epoch": 0.25519681556833257,
+      "grad_norm": 17.125,
+      "learning_rate": 1.8568352180394603e-06,
+      "loss": 2.161,
+      "step": 8655
+    },
+    {
+      "epoch": 0.255344242960342,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.8565697678169578e-06,
+      "loss": 2.2521,
+      "step": 8660
+    },
+    {
+      "epoch": 0.2554916703523515,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.8563040907396132e-06,
+      "loss": 2.1843,
+      "step": 8665
+    },
+    {
+      "epoch": 0.2556390977443609,
+      "grad_norm": 16.75,
+      "learning_rate": 1.8560381868777886e-06,
+      "loss": 2.1212,
+      "step": 8670
+    },
+    {
+      "epoch": 0.25578652513637035,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.8557720563019068e-06,
+      "loss": 2.2056,
+      "step": 8675
+    },
+    {
+      "epoch": 0.25593395252837975,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.8555056990824496e-06,
+      "loss": 2.0907,
+      "step": 8680
+    },
+    {
+      "epoch": 0.2560813799203892,
+      "grad_norm": 16.75,
+      "learning_rate": 1.8552391152899599e-06,
+      "loss": 2.3035,
+      "step": 8685
+    },
+    {
+      "epoch": 0.25622880731239867,
+      "grad_norm": 15.375,
+      "learning_rate": 1.854972304995039e-06,
+      "loss": 2.2537,
+      "step": 8690
+    },
+    {
+      "epoch": 0.25637623470440807,
+      "grad_norm": 17.25,
+      "learning_rate": 1.8547052682683505e-06,
+      "loss": 2.2827,
+      "step": 8695
+    },
+    {
+      "epoch": 0.2565236620964175,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.8544380051806157e-06,
+      "loss": 2.1187,
+      "step": 8700
+    },
+    {
+      "epoch": 0.25667108948842693,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.8541705158026173e-06,
+      "loss": 2.1568,
+      "step": 8705
+    },
+    {
+      "epoch": 0.2568185168804364,
+      "grad_norm": 15.875,
+      "learning_rate": 1.8539028002051973e-06,
+      "loss": 2.1486,
+      "step": 8710
+    },
+    {
+      "epoch": 0.25696594427244585,
+      "grad_norm": 15.125,
+      "learning_rate": 1.8536348584592578e-06,
+      "loss": 2.2145,
+      "step": 8715
+    },
+    {
+      "epoch": 0.25711337166445525,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.8533666906357607e-06,
+      "loss": 2.1246,
+      "step": 8720
+    },
+    {
+      "epoch": 0.2572607990564647,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.8530982968057285e-06,
+      "loss": 2.2059,
+      "step": 8725
+    },
+    {
+      "epoch": 0.2574082264484741,
+      "grad_norm": 14.125,
+      "learning_rate": 1.852829677040242e-06,
+      "loss": 2.2232,
+      "step": 8730
+    },
+    {
+      "epoch": 0.25755565384048357,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.8525608314104431e-06,
+      "loss": 2.2328,
+      "step": 8735
+    },
+    {
+      "epoch": 0.25770308123249297,
+      "grad_norm": 15.5,
+      "learning_rate": 1.8522917599875334e-06,
+      "loss": 2.2464,
+      "step": 8740
+    },
+    {
+      "epoch": 0.25785050862450243,
+      "grad_norm": 15.375,
+      "learning_rate": 1.8520224628427736e-06,
+      "loss": 2.2532,
+      "step": 8745
+    },
+    {
+      "epoch": 0.2579979360165119,
+      "grad_norm": 16.125,
+      "learning_rate": 1.8517529400474848e-06,
+      "loss": 2.1911,
+      "step": 8750
+    },
+    {
+      "epoch": 0.2581453634085213,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.8514831916730482e-06,
+      "loss": 2.2427,
+      "step": 8755
+    },
+    {
+      "epoch": 0.25829279080053075,
+      "grad_norm": 13.625,
+      "learning_rate": 1.8512132177909034e-06,
+      "loss": 2.1896,
+      "step": 8760
+    },
+    {
+      "epoch": 0.25844021819254015,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.8509430184725513e-06,
+      "loss": 2.0831,
+      "step": 8765
+    },
+    {
+      "epoch": 0.2585876455845496,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.8506725937895515e-06,
+      "loss": 2.1604,
+      "step": 8770
+    },
+    {
+      "epoch": 0.25873507297655907,
+      "grad_norm": 28.875,
+      "learning_rate": 1.8504019438135235e-06,
+      "loss": 2.2493,
+      "step": 8775
+    },
+    {
+      "epoch": 0.25888250036856847,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.8501310686161463e-06,
+      "loss": 2.1755,
+      "step": 8780
+    },
+    {
+      "epoch": 0.2590299277605779,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.8498599682691592e-06,
+      "loss": 2.2261,
+      "step": 8785
+    },
+    {
+      "epoch": 0.25917735515258733,
+      "grad_norm": 18.625,
+      "learning_rate": 1.8495886428443605e-06,
+      "loss": 2.1731,
+      "step": 8790
+    },
+    {
+      "epoch": 0.2593247825445968,
+      "grad_norm": 15.125,
+      "learning_rate": 1.8493170924136083e-06,
+      "loss": 2.3196,
+      "step": 8795
+    },
+    {
+      "epoch": 0.25947220993660625,
+      "grad_norm": 16.75,
+      "learning_rate": 1.8490453170488202e-06,
+      "loss": 2.2446,
+      "step": 8800
+    },
+    {
+      "epoch": 0.25961963732861565,
+      "grad_norm": 14.875,
+      "learning_rate": 1.8487733168219739e-06,
+      "loss": 2.1648,
+      "step": 8805
+    },
+    {
+      "epoch": 0.2597670647206251,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.8485010918051059e-06,
+      "loss": 2.2215,
+      "step": 8810
+    },
+    {
+      "epoch": 0.2599144921126345,
+      "grad_norm": 17.5,
+      "learning_rate": 1.8482286420703125e-06,
+      "loss": 2.3307,
+      "step": 8815
+    },
+    {
+      "epoch": 0.26006191950464397,
+      "grad_norm": 14.375,
+      "learning_rate": 1.84795596768975e-06,
+      "loss": 2.1618,
+      "step": 8820
+    },
+    {
+      "epoch": 0.26020934689665337,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.847683068735633e-06,
+      "loss": 2.2691,
+      "step": 8825
+    },
+    {
+      "epoch": 0.26035677428866283,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.8474099452802369e-06,
+      "loss": 2.117,
+      "step": 8830
+    },
+    {
+      "epoch": 0.2605042016806723,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.8471365973958962e-06,
+      "loss": 2.0912,
+      "step": 8835
+    },
+    {
+      "epoch": 0.2606516290726817,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.846863025155004e-06,
+      "loss": 2.1718,
+      "step": 8840
+    },
+    {
+      "epoch": 0.26079905646469115,
+      "grad_norm": 17.0,
+      "learning_rate": 1.8465892286300137e-06,
+      "loss": 2.0818,
+      "step": 8845
+    },
+    {
+      "epoch": 0.26094648385670055,
+      "grad_norm": 16.875,
+      "learning_rate": 1.8463152078934383e-06,
+      "loss": 2.2438,
+      "step": 8850
+    },
+    {
+      "epoch": 0.26109391124871,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.8460409630178493e-06,
+      "loss": 2.2129,
+      "step": 8855
+    },
+    {
+      "epoch": 0.26124133864071947,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.8457664940758782e-06,
+      "loss": 2.1616,
+      "step": 8860
+    },
+    {
+      "epoch": 0.26138876603272887,
+      "grad_norm": 14.5,
+      "learning_rate": 1.8454918011402155e-06,
+      "loss": 2.2635,
+      "step": 8865
+    },
+    {
+      "epoch": 0.26153619342473833,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.8452168842836114e-06,
+      "loss": 2.2376,
+      "step": 8870
+    },
+    {
+      "epoch": 0.26168362081674773,
+      "grad_norm": 13.5,
+      "learning_rate": 1.8449417435788748e-06,
+      "loss": 2.1005,
+      "step": 8875
+    },
+    {
+      "epoch": 0.2618310482087572,
+      "grad_norm": 17.5,
+      "learning_rate": 1.8446663790988742e-06,
+      "loss": 2.164,
+      "step": 8880
+    },
+    {
+      "epoch": 0.26197847560076665,
+      "grad_norm": 15.5,
+      "learning_rate": 1.844390790916538e-06,
+      "loss": 2.1707,
+      "step": 8885
+    },
+    {
+      "epoch": 0.26212590299277605,
+      "grad_norm": 20.25,
+      "learning_rate": 1.844114979104853e-06,
+      "loss": 2.2746,
+      "step": 8890
+    },
+    {
+      "epoch": 0.2622733303847855,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.843838943736865e-06,
+      "loss": 2.1086,
+      "step": 8895
+    },
+    {
+      "epoch": 0.2624207577767949,
+      "grad_norm": 16.125,
+      "learning_rate": 1.8435626848856805e-06,
+      "loss": 2.3454,
+      "step": 8900
+    },
+    {
+      "epoch": 0.26256818516880437,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.8432862026244633e-06,
+      "loss": 2.188,
+      "step": 8905
+    },
+    {
+      "epoch": 0.26271561256081377,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.8430094970264374e-06,
+      "loss": 2.1443,
+      "step": 8910
+    },
+    {
+      "epoch": 0.26286303995282323,
+      "grad_norm": 18.375,
+      "learning_rate": 1.8427325681648861e-06,
+      "loss": 2.1986,
+      "step": 8915
+    },
+    {
+      "epoch": 0.2630104673448327,
+      "grad_norm": 15.5,
+      "learning_rate": 1.8424554161131515e-06,
+      "loss": 2.1675,
+      "step": 8920
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.8421780409446347e-06,
+      "loss": 2.2025,
+      "step": 8925
+    },
+    {
+      "epoch": 0.26330532212885155,
+      "grad_norm": 14.125,
+      "learning_rate": 1.841900442732796e-06,
+      "loss": 2.1276,
+      "step": 8930
+    },
+    {
+      "epoch": 0.26345274952086095,
+      "grad_norm": 17.75,
+      "learning_rate": 1.841622621551155e-06,
+      "loss": 2.1602,
+      "step": 8935
+    },
+    {
+      "epoch": 0.2636001769128704,
+      "grad_norm": 13.875,
+      "learning_rate": 1.8413445774732901e-06,
+      "loss": 2.2945,
+      "step": 8940
+    },
+    {
+      "epoch": 0.26374760430487987,
+      "grad_norm": 15.375,
+      "learning_rate": 1.8410663105728387e-06,
+      "loss": 2.0937,
+      "step": 8945
+    },
+    {
+      "epoch": 0.26389503169688927,
+      "grad_norm": 17.75,
+      "learning_rate": 1.840787820923497e-06,
+      "loss": 2.2,
+      "step": 8950
+    },
+    {
+      "epoch": 0.26404245908889873,
+      "grad_norm": 14.625,
+      "learning_rate": 1.8405091085990213e-06,
+      "loss": 2.1271,
+      "step": 8955
+    },
+    {
+      "epoch": 0.26418988648090813,
+      "grad_norm": 25.75,
+      "learning_rate": 1.840230173673225e-06,
+      "loss": 2.1916,
+      "step": 8960
+    },
+    {
+      "epoch": 0.2643373138729176,
+      "grad_norm": 17.25,
+      "learning_rate": 1.8399510162199824e-06,
+      "loss": 2.2481,
+      "step": 8965
+    },
+    {
+      "epoch": 0.26448474126492705,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.8396716363132255e-06,
+      "loss": 2.1464,
+      "step": 8970
+    },
+    {
+      "epoch": 0.26463216865693645,
+      "grad_norm": 17.0,
+      "learning_rate": 1.8393920340269458e-06,
+      "loss": 2.1157,
+      "step": 8975
+    },
+    {
+      "epoch": 0.2647795960489459,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.8391122094351933e-06,
+      "loss": 2.0538,
+      "step": 8980
+    },
+    {
+      "epoch": 0.2649270234409553,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.8388321626120769e-06,
+      "loss": 2.2269,
+      "step": 8985
+    },
+    {
+      "epoch": 0.26507445083296477,
+      "grad_norm": 16.75,
+      "learning_rate": 1.8385518936317645e-06,
+      "loss": 2.1711,
+      "step": 8990
+    },
+    {
+      "epoch": 0.26522187822497423,
+      "grad_norm": 13.75,
+      "learning_rate": 1.838271402568483e-06,
+      "loss": 2.1552,
+      "step": 8995
+    },
+    {
+      "epoch": 0.26536930561698363,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.837990689496518e-06,
+      "loss": 2.1522,
+      "step": 9000
+    },
+    {
+      "epoch": 0.26536930561698363,
+      "eval_loss": 2.142622232437134,
+      "eval_runtime": 4.7105,
+      "eval_samples_per_second": 84.067,
+      "eval_steps_per_second": 2.76,
+      "step": 9000
+    },
+    {
+      "epoch": 0.2655167330089931,
+      "grad_norm": 17.5,
+      "learning_rate": 1.837709754490214e-06,
+      "loss": 2.1945,
+      "step": 9005
+    },
+    {
+      "epoch": 0.2656641604010025,
+      "grad_norm": 16.125,
+      "learning_rate": 1.8374285976239734e-06,
+      "loss": 2.2451,
+      "step": 9010
+    },
+    {
+      "epoch": 0.26581158779301195,
+      "grad_norm": 15.625,
+      "learning_rate": 1.8371472189722589e-06,
+      "loss": 2.2319,
+      "step": 9015
+    },
+    {
+      "epoch": 0.26595901518502135,
+      "grad_norm": 19.25,
+      "learning_rate": 1.8368656186095905e-06,
+      "loss": 2.2219,
+      "step": 9020
+    },
+    {
+      "epoch": 0.2661064425770308,
+      "grad_norm": 22.125,
+      "learning_rate": 1.8365837966105481e-06,
+      "loss": 2.1642,
+      "step": 9025
+    },
+    {
+      "epoch": 0.26625386996904027,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.8363017530497693e-06,
+      "loss": 2.2582,
+      "step": 9030
+    },
+    {
+      "epoch": 0.26640129736104967,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.836019488001951e-06,
+      "loss": 2.1031,
+      "step": 9035
+    },
+    {
+      "epoch": 0.26654872475305913,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.8357370015418488e-06,
+      "loss": 2.225,
+      "step": 9040
+    },
+    {
+      "epoch": 0.26669615214506853,
+      "grad_norm": 15.0,
+      "learning_rate": 1.835454293744276e-06,
+      "loss": 2.254,
+      "step": 9045
+    },
+    {
+      "epoch": 0.266843579537078,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.8351713646841059e-06,
+      "loss": 2.2191,
+      "step": 9050
+    },
+    {
+      "epoch": 0.26699100692908745,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.8348882144362694e-06,
+      "loss": 2.2358,
+      "step": 9055
+    },
+    {
+      "epoch": 0.26713843432109685,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.8346048430757566e-06,
+      "loss": 2.1893,
+      "step": 9060
+    },
+    {
+      "epoch": 0.2672858617131063,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.8343212506776155e-06,
+      "loss": 2.1395,
+      "step": 9065
+    },
+    {
+      "epoch": 0.2674332891051157,
+      "grad_norm": 16.375,
+      "learning_rate": 1.8340374373169533e-06,
+      "loss": 2.1742,
+      "step": 9070
+    },
+    {
+      "epoch": 0.26758071649712517,
+      "grad_norm": 18.125,
+      "learning_rate": 1.8337534030689355e-06,
+      "loss": 2.1369,
+      "step": 9075
+    },
+    {
+      "epoch": 0.26772814388913463,
+      "grad_norm": 22.75,
+      "learning_rate": 1.8334691480087856e-06,
+      "loss": 2.1244,
+      "step": 9080
+    },
+    {
+      "epoch": 0.26787557128114403,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.8331846722117864e-06,
+      "loss": 2.227,
+      "step": 9085
+    },
+    {
+      "epoch": 0.2680229986731535,
+      "grad_norm": 15.75,
+      "learning_rate": 1.8328999757532788e-06,
+      "loss": 2.1094,
+      "step": 9090
+    },
+    {
+      "epoch": 0.2681704260651629,
+      "grad_norm": 13.75,
+      "learning_rate": 1.8326150587086617e-06,
+      "loss": 2.1565,
+      "step": 9095
+    },
+    {
+      "epoch": 0.26831785345717235,
+      "grad_norm": 14.375,
+      "learning_rate": 1.832329921153393e-06,
+      "loss": 2.272,
+      "step": 9100
+    },
+    {
+      "epoch": 0.26846528084918175,
+      "grad_norm": 17.375,
+      "learning_rate": 1.8320445631629892e-06,
+      "loss": 2.2073,
+      "step": 9105
+    },
+    {
+      "epoch": 0.2686127082411912,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.8317589848130246e-06,
+      "loss": 2.052,
+      "step": 9110
+    },
+    {
+      "epoch": 0.26876013563320067,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.831473186179132e-06,
+      "loss": 2.1517,
+      "step": 9115
+    },
+    {
+      "epoch": 0.2689075630252101,
+      "grad_norm": 12.75,
+      "learning_rate": 1.8311871673370023e-06,
+      "loss": 2.0941,
+      "step": 9120
+    },
+    {
+      "epoch": 0.26905499041721953,
+      "grad_norm": 17.125,
+      "learning_rate": 1.8309009283623854e-06,
+      "loss": 2.2009,
+      "step": 9125
+    },
+    {
+      "epoch": 0.26920241780922893,
+      "grad_norm": 16.75,
+      "learning_rate": 1.8306144693310893e-06,
+      "loss": 2.0812,
+      "step": 9130
+    },
+    {
+      "epoch": 0.2693498452012384,
+      "grad_norm": 12.75,
+      "learning_rate": 1.8303277903189798e-06,
+      "loss": 2.1815,
+      "step": 9135
+    },
+    {
+      "epoch": 0.26949727259324785,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.8300408914019813e-06,
+      "loss": 2.1776,
+      "step": 9140
+    },
+    {
+      "epoch": 0.26964469998525725,
+      "grad_norm": 16.125,
+      "learning_rate": 1.8297537726560766e-06,
+      "loss": 2.2445,
+      "step": 9145
+    },
+    {
+      "epoch": 0.2697921273772667,
+      "grad_norm": 15.875,
+      "learning_rate": 1.8294664341573063e-06,
+      "loss": 2.2909,
+      "step": 9150
+    },
+    {
+      "epoch": 0.2699395547692761,
+      "grad_norm": 16.625,
+      "learning_rate": 1.8291788759817695e-06,
+      "loss": 2.2258,
+      "step": 9155
+    },
+    {
+      "epoch": 0.27008698216128557,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.8288910982056237e-06,
+      "loss": 2.1185,
+      "step": 9160
+    },
+    {
+      "epoch": 0.27023440955329503,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.8286031009050837e-06,
+      "loss": 2.1964,
+      "step": 9165
+    },
+    {
+      "epoch": 0.27038183694530443,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.8283148841564234e-06,
+      "loss": 2.1557,
+      "step": 9170
+    },
+    {
+      "epoch": 0.2705292643373139,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.8280264480359747e-06,
+      "loss": 2.1727,
+      "step": 9175
+    },
+    {
+      "epoch": 0.2706766917293233,
+      "grad_norm": 14.25,
+      "learning_rate": 1.8277377926201268e-06,
+      "loss": 2.1334,
+      "step": 9180
+    },
+    {
+      "epoch": 0.27082411912133275,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.8274489179853273e-06,
+      "loss": 2.0905,
+      "step": 9185
+    },
+    {
+      "epoch": 0.27097154651334215,
+      "grad_norm": 17.5,
+      "learning_rate": 1.827159824208083e-06,
+      "loss": 2.2571,
+      "step": 9190
+    },
+    {
+      "epoch": 0.2711189739053516,
+      "grad_norm": 15.125,
+      "learning_rate": 1.826870511364957e-06,
+      "loss": 2.1973,
+      "step": 9195
+    },
+    {
+      "epoch": 0.27126640129736107,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.8265809795325713e-06,
+      "loss": 2.0869,
+      "step": 9200
+    },
+    {
+      "epoch": 0.2714138286893705,
+      "grad_norm": 27.0,
+      "learning_rate": 1.8262912287876065e-06,
+      "loss": 2.1648,
+      "step": 9205
+    },
+    {
+      "epoch": 0.27156125608137993,
+      "grad_norm": 20.5,
+      "learning_rate": 1.8260012592067995e-06,
+      "loss": 2.1417,
+      "step": 9210
+    },
+    {
+      "epoch": 0.27170868347338933,
+      "grad_norm": 16.0,
+      "learning_rate": 1.825711070866947e-06,
+      "loss": 2.224,
+      "step": 9215
+    },
+    {
+      "epoch": 0.2718561108653988,
+      "grad_norm": 15.625,
+      "learning_rate": 1.825420663844902e-06,
+      "loss": 2.1834,
+      "step": 9220
+    },
+    {
+      "epoch": 0.27200353825740825,
+      "grad_norm": 15.25,
+      "learning_rate": 1.8251300382175767e-06,
+      "loss": 2.0962,
+      "step": 9225
+    },
+    {
+      "epoch": 0.27215096564941765,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.824839194061941e-06,
+      "loss": 2.1178,
+      "step": 9230
+    },
+    {
+      "epoch": 0.2722983930414271,
+      "grad_norm": 21.625,
+      "learning_rate": 1.824548131455022e-06,
+      "loss": 2.2477,
+      "step": 9235
+    },
+    {
+      "epoch": 0.2724458204334365,
+      "grad_norm": 17.5,
+      "learning_rate": 1.8242568504739046e-06,
+      "loss": 2.1384,
+      "step": 9240
+    },
+    {
+      "epoch": 0.272593247825446,
+      "grad_norm": 17.125,
+      "learning_rate": 1.8239653511957326e-06,
+      "loss": 2.2245,
+      "step": 9245
+    },
+    {
+      "epoch": 0.27274067521745543,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.8236736336977065e-06,
+      "loss": 2.2292,
+      "step": 9250
+    },
+    {
+      "epoch": 0.27288810260946483,
+      "grad_norm": 16.125,
+      "learning_rate": 1.8233816980570857e-06,
+      "loss": 2.2241,
+      "step": 9255
+    },
+    {
+      "epoch": 0.2730355300014743,
+      "grad_norm": 16.875,
+      "learning_rate": 1.8230895443511861e-06,
+      "loss": 2.2429,
+      "step": 9260
+    },
+    {
+      "epoch": 0.2731829573934837,
+      "grad_norm": 12.25,
+      "learning_rate": 1.8227971726573825e-06,
+      "loss": 2.2361,
+      "step": 9265
+    },
+    {
+      "epoch": 0.27333038478549315,
+      "grad_norm": 13.75,
+      "learning_rate": 1.8225045830531068e-06,
+      "loss": 2.2299,
+      "step": 9270
+    },
+    {
+      "epoch": 0.27347781217750256,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.8222117756158486e-06,
+      "loss": 2.1748,
+      "step": 9275
+    },
+    {
+      "epoch": 0.273625239569512,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.8219187504231553e-06,
+      "loss": 2.1427,
+      "step": 9280
+    },
+    {
+      "epoch": 0.27377266696152147,
+      "grad_norm": 16.5,
+      "learning_rate": 1.821625507552632e-06,
+      "loss": 2.2049,
+      "step": 9285
+    },
+    {
+      "epoch": 0.2739200943535309,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.8213320470819413e-06,
+      "loss": 2.1321,
+      "step": 9290
+    },
+    {
+      "epoch": 0.27406752174554033,
+      "grad_norm": 13.125,
+      "learning_rate": 1.821038369088804e-06,
+      "loss": 2.057,
+      "step": 9295
+    },
+    {
+      "epoch": 0.27421494913754973,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.820744473650998e-06,
+      "loss": 2.0797,
+      "step": 9300
+    },
+    {
+      "epoch": 0.2743623765295592,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.8204503608463586e-06,
+      "loss": 2.1928,
+      "step": 9305
+    },
+    {
+      "epoch": 0.27450980392156865,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.8201560307527793e-06,
+      "loss": 2.2871,
+      "step": 9310
+    },
+    {
+      "epoch": 0.27465723131357805,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.8198614834482107e-06,
+      "loss": 2.2493,
+      "step": 9315
+    },
+    {
+      "epoch": 0.2748046587055875,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.8195667190106607e-06,
+      "loss": 2.1325,
+      "step": 9320
+    },
+    {
+      "epoch": 0.2749520860975969,
+      "grad_norm": 18.5,
+      "learning_rate": 1.8192717375181954e-06,
+      "loss": 2.2074,
+      "step": 9325
+    },
+    {
+      "epoch": 0.2750995134896064,
+      "grad_norm": 14.75,
+      "learning_rate": 1.8189765390489375e-06,
+      "loss": 2.1769,
+      "step": 9330
+    },
+    {
+      "epoch": 0.27524694088161583,
+      "grad_norm": 14.75,
+      "learning_rate": 1.8186811236810686e-06,
+      "loss": 2.3256,
+      "step": 9335
+    },
+    {
+      "epoch": 0.27539436827362523,
+      "grad_norm": 13.875,
+      "learning_rate": 1.818385491492826e-06,
+      "loss": 2.1801,
+      "step": 9340
+    },
+    {
+      "epoch": 0.2755417956656347,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.8180896425625054e-06,
+      "loss": 2.1626,
+      "step": 9345
+    },
+    {
+      "epoch": 0.2756892230576441,
+      "grad_norm": 17.25,
+      "learning_rate": 1.81779357696846e-06,
+      "loss": 2.126,
+      "step": 9350
+    },
+    {
+      "epoch": 0.27583665044965355,
+      "grad_norm": 14.875,
+      "learning_rate": 1.8174972947890998e-06,
+      "loss": 2.0943,
+      "step": 9355
+    },
+    {
+      "epoch": 0.27598407784166296,
+      "grad_norm": 18.75,
+      "learning_rate": 1.8172007961028928e-06,
+      "loss": 2.2155,
+      "step": 9360
+    },
+    {
+      "epoch": 0.2761315052336724,
+      "grad_norm": 13.5,
+      "learning_rate": 1.8169040809883639e-06,
+      "loss": 2.1628,
+      "step": 9365
+    },
+    {
+      "epoch": 0.2762789326256819,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.8166071495240952e-06,
+      "loss": 2.1454,
+      "step": 9370
+    },
+    {
+      "epoch": 0.2764263600176913,
+      "grad_norm": 14.125,
+      "learning_rate": 1.816310001788727e-06,
+      "loss": 2.1551,
+      "step": 9375
+    },
+    {
+      "epoch": 0.27657378740970073,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.8160126378609553e-06,
+      "loss": 2.1669,
+      "step": 9380
+    },
+    {
+      "epoch": 0.27672121480171014,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.8157150578195348e-06,
+      "loss": 2.1653,
+      "step": 9385
+    },
+    {
+      "epoch": 0.2768686421937196,
+      "grad_norm": 16.875,
+      "learning_rate": 1.8154172617432772e-06,
+      "loss": 2.2091,
+      "step": 9390
+    },
+    {
+      "epoch": 0.27701606958572905,
+      "grad_norm": 14.5,
+      "learning_rate": 1.8151192497110505e-06,
+      "loss": 2.2048,
+      "step": 9395
+    },
+    {
+      "epoch": 0.27716349697773845,
+      "grad_norm": 17.25,
+      "learning_rate": 1.814821021801781e-06,
+      "loss": 2.2176,
+      "step": 9400
+    },
+    {
+      "epoch": 0.2773109243697479,
+      "grad_norm": 14.125,
+      "learning_rate": 1.8145225780944515e-06,
+      "loss": 2.2279,
+      "step": 9405
+    },
+    {
+      "epoch": 0.2774583517617573,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.8142239186681022e-06,
+      "loss": 2.1464,
+      "step": 9410
+    },
+    {
+      "epoch": 0.2776057791537668,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.8139250436018303e-06,
+      "loss": 2.2263,
+      "step": 9415
+    },
+    {
+      "epoch": 0.27775320654577623,
+      "grad_norm": 16.125,
+      "learning_rate": 1.81362595297479e-06,
+      "loss": 2.1134,
+      "step": 9420
+    },
+    {
+      "epoch": 0.27790063393778563,
+      "grad_norm": 15.25,
+      "learning_rate": 1.8133266468661934e-06,
+      "loss": 2.148,
+      "step": 9425
+    },
+    {
+      "epoch": 0.2780480613297951,
+      "grad_norm": 14.625,
+      "learning_rate": 1.8130271253553084e-06,
+      "loss": 2.1851,
+      "step": 9430
+    },
+    {
+      "epoch": 0.2781954887218045,
+      "grad_norm": 16.25,
+      "learning_rate": 1.8127273885214609e-06,
+      "loss": 2.2678,
+      "step": 9435
+    },
+    {
+      "epoch": 0.27834291611381395,
+      "grad_norm": 16.375,
+      "learning_rate": 1.8124274364440337e-06,
+      "loss": 2.1842,
+      "step": 9440
+    },
+    {
+      "epoch": 0.27849034350582336,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.8121272692024658e-06,
+      "loss": 2.2537,
+      "step": 9445
+    },
+    {
+      "epoch": 0.2786377708978328,
+      "grad_norm": 14.75,
+      "learning_rate": 1.8118268868762546e-06,
+      "loss": 2.1291,
+      "step": 9450
+    },
+    {
+      "epoch": 0.2787851982898423,
+      "grad_norm": 15.0,
+      "learning_rate": 1.811526289544953e-06,
+      "loss": 2.1512,
+      "step": 9455
+    },
+    {
+      "epoch": 0.2789326256818517,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.8112254772881717e-06,
+      "loss": 2.1319,
+      "step": 9460
+    },
+    {
+      "epoch": 0.27908005307386113,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.8109244501855782e-06,
+      "loss": 2.1444,
+      "step": 9465
+    },
+    {
+      "epoch": 0.27922748046587054,
+      "grad_norm": 13.875,
+      "learning_rate": 1.810623208316897e-06,
+      "loss": 2.1868,
+      "step": 9470
+    },
+    {
+      "epoch": 0.27937490785788,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.8103217517619094e-06,
+      "loss": 2.2162,
+      "step": 9475
+    },
+    {
+      "epoch": 0.27952233524988945,
+      "grad_norm": 16.125,
+      "learning_rate": 1.810020080600453e-06,
+      "loss": 2.2668,
+      "step": 9480
+    },
+    {
+      "epoch": 0.27966976264189886,
+      "grad_norm": 16.0,
+      "learning_rate": 1.809718194912423e-06,
+      "loss": 2.2179,
+      "step": 9485
+    },
+    {
+      "epoch": 0.2798171900339083,
+      "grad_norm": 14.875,
+      "learning_rate": 1.809416094777771e-06,
+      "loss": 2.1202,
+      "step": 9490
+    },
+    {
+      "epoch": 0.2799646174259177,
+      "grad_norm": 10.75,
+      "learning_rate": 1.8091137802765058e-06,
+      "loss": 2.0973,
+      "step": 9495
+    },
+    {
+      "epoch": 0.2801120448179272,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.8088112514886923e-06,
+      "loss": 2.0952,
+      "step": 9500
+    },
+    {
+      "epoch": 0.2801120448179272,
+      "eval_loss": 2.1350300312042236,
+      "eval_runtime": 4.7104,
+      "eval_samples_per_second": 84.069,
+      "eval_steps_per_second": 2.76,
+      "step": 9500
+    },
+    {
+      "epoch": 0.28025947220993663,
+      "grad_norm": 16.5,
+      "learning_rate": 1.8085085084944526e-06,
+      "loss": 2.1947,
+      "step": 9505
+    },
+    {
+      "epoch": 0.28040689960194604,
+      "grad_norm": 16.875,
+      "learning_rate": 1.808205551373966e-06,
+      "loss": 2.2008,
+      "step": 9510
+    },
+    {
+      "epoch": 0.2805543269939555,
+      "grad_norm": 16.125,
+      "learning_rate": 1.8079023802074674e-06,
+      "loss": 2.244,
+      "step": 9515
+    },
+    {
+      "epoch": 0.2807017543859649,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.807598995075249e-06,
+      "loss": 2.2662,
+      "step": 9520
+    },
+    {
+      "epoch": 0.28084918177797435,
+      "grad_norm": 15.75,
+      "learning_rate": 1.80729539605766e-06,
+      "loss": 2.1337,
+      "step": 9525
+    },
+    {
+      "epoch": 0.28099660916998376,
+      "grad_norm": 14.375,
+      "learning_rate": 1.8069915832351057e-06,
+      "loss": 2.1507,
+      "step": 9530
+    },
+    {
+      "epoch": 0.2811440365619932,
+      "grad_norm": 13.25,
+      "learning_rate": 1.8066875566880482e-06,
+      "loss": 2.1803,
+      "step": 9535
+    },
+    {
+      "epoch": 0.2812914639540027,
+      "grad_norm": 11.25,
+      "learning_rate": 1.8063833164970061e-06,
+      "loss": 2.0851,
+      "step": 9540
+    },
+    {
+      "epoch": 0.2814388913460121,
+      "grad_norm": 13.5,
+      "learning_rate": 1.8060788627425548e-06,
+      "loss": 2.2345,
+      "step": 9545
+    },
+    {
+      "epoch": 0.28158631873802153,
+      "grad_norm": 15.875,
+      "learning_rate": 1.8057741955053261e-06,
+      "loss": 2.3918,
+      "step": 9550
+    },
+    {
+      "epoch": 0.28173374613003094,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.8054693148660088e-06,
+      "loss": 2.1558,
+      "step": 9555
+    },
+    {
+      "epoch": 0.2818811735220404,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.805164220905347e-06,
+      "loss": 2.1571,
+      "step": 9560
+    },
+    {
+      "epoch": 0.28202860091404985,
+      "grad_norm": 15.0,
+      "learning_rate": 1.8048589137041427e-06,
+      "loss": 2.0988,
+      "step": 9565
+    },
+    {
+      "epoch": 0.28217602830605926,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.8045533933432538e-06,
+      "loss": 2.1293,
+      "step": 9570
+    },
+    {
+      "epoch": 0.2823234556980687,
+      "grad_norm": 24.25,
+      "learning_rate": 1.8042476599035944e-06,
+      "loss": 2.1763,
+      "step": 9575
+    },
+    {
+      "epoch": 0.2824708830900781,
+      "grad_norm": 14.5,
+      "learning_rate": 1.8039417134661354e-06,
+      "loss": 2.1462,
+      "step": 9580
+    },
+    {
+      "epoch": 0.2826183104820876,
+      "grad_norm": 12.125,
+      "learning_rate": 1.8036355541119038e-06,
+      "loss": 2.2079,
+      "step": 9585
+    },
+    {
+      "epoch": 0.28276573787409703,
+      "grad_norm": 14.5,
+      "learning_rate": 1.8033291819219833e-06,
+      "loss": 2.1601,
+      "step": 9590
+    },
+    {
+      "epoch": 0.28291316526610644,
+      "grad_norm": 14.375,
+      "learning_rate": 1.803022596977514e-06,
+      "loss": 2.2201,
+      "step": 9595
+    },
+    {
+      "epoch": 0.2830605926581159,
+      "grad_norm": 59.25,
+      "learning_rate": 1.802715799359692e-06,
+      "loss": 2.1743,
+      "step": 9600
+    },
+    {
+      "epoch": 0.2832080200501253,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.8024087891497697e-06,
+      "loss": 2.2809,
+      "step": 9605
+    },
+    {
+      "epoch": 0.28335544744213476,
+      "grad_norm": 18.0,
+      "learning_rate": 1.8021015664290563e-06,
+      "loss": 2.0981,
+      "step": 9610
+    },
+    {
+      "epoch": 0.28350287483414416,
+      "grad_norm": 14.25,
+      "learning_rate": 1.8017941312789172e-06,
+      "loss": 2.3601,
+      "step": 9615
+    },
+    {
+      "epoch": 0.2836503022261536,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.8014864837807732e-06,
+      "loss": 2.0582,
+      "step": 9620
+    },
+    {
+      "epoch": 0.2837977296181631,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.8011786240161025e-06,
+      "loss": 2.1897,
+      "step": 9625
+    },
+    {
+      "epoch": 0.2839451570101725,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.8008705520664388e-06,
+      "loss": 2.1527,
+      "step": 9630
+    },
+    {
+      "epoch": 0.28409258440218194,
+      "grad_norm": 17.25,
+      "learning_rate": 1.8005622680133722e-06,
+      "loss": 2.1402,
+      "step": 9635
+    },
+    {
+      "epoch": 0.28424001179419134,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.8002537719385492e-06,
+      "loss": 2.2136,
+      "step": 9640
+    },
+    {
+      "epoch": 0.2843874391862008,
+      "grad_norm": 15.375,
+      "learning_rate": 1.7999450639236716e-06,
+      "loss": 2.2145,
+      "step": 9645
+    },
+    {
+      "epoch": 0.28453486657821025,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.799636144050499e-06,
+      "loss": 2.2849,
+      "step": 9650
+    },
+    {
+      "epoch": 0.28468229397021966,
+      "grad_norm": 17.25,
+      "learning_rate": 1.7993270124008454e-06,
+      "loss": 2.1804,
+      "step": 9655
+    },
+    {
+      "epoch": 0.2848297213622291,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.7990176690565815e-06,
+      "loss": 2.1777,
+      "step": 9660
+    },
+    {
+      "epoch": 0.2849771487542385,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.798708114099634e-06,
+      "loss": 2.1158,
+      "step": 9665
+    },
+    {
+      "epoch": 0.285124576146248,
+      "grad_norm": 14.0,
+      "learning_rate": 1.7983983476119864e-06,
+      "loss": 2.1636,
+      "step": 9670
+    },
+    {
+      "epoch": 0.28527200353825743,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.798088369675677e-06,
+      "loss": 2.1837,
+      "step": 9675
+    },
+    {
+      "epoch": 0.28541943093026684,
+      "grad_norm": 15.25,
+      "learning_rate": 1.797778180372801e-06,
+      "loss": 2.2862,
+      "step": 9680
+    },
+    {
+      "epoch": 0.2855668583222763,
+      "grad_norm": 17.0,
+      "learning_rate": 1.7974677797855092e-06,
+      "loss": 2.2452,
+      "step": 9685
+    },
+    {
+      "epoch": 0.2857142857142857,
+      "grad_norm": 15.875,
+      "learning_rate": 1.7971571679960081e-06,
+      "loss": 2.1044,
+      "step": 9690
+    },
+    {
+      "epoch": 0.28586171310629516,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.7968463450865608e-06,
+      "loss": 2.1527,
+      "step": 9695
+    },
+    {
+      "epoch": 0.28600914049830456,
+      "grad_norm": 16.125,
+      "learning_rate": 1.796535311139486e-06,
+      "loss": 2.1591,
+      "step": 9700
+    },
+    {
+      "epoch": 0.286156567890314,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.796224066237158e-06,
+      "loss": 2.2229,
+      "step": 9705
+    },
+    {
+      "epoch": 0.2863039952823235,
+      "grad_norm": 16.125,
+      "learning_rate": 1.7959126104620074e-06,
+      "loss": 2.2106,
+      "step": 9710
+    },
+    {
+      "epoch": 0.2864514226743329,
+      "grad_norm": 16.625,
+      "learning_rate": 1.7956009438965204e-06,
+      "loss": 2.1492,
+      "step": 9715
+    },
+    {
+      "epoch": 0.28659885006634234,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.7952890666232391e-06,
+      "loss": 2.1762,
+      "step": 9720
+    },
+    {
+      "epoch": 0.28674627745835174,
+      "grad_norm": 15.0,
+      "learning_rate": 1.7949769787247617e-06,
+      "loss": 2.1311,
+      "step": 9725
+    },
+    {
+      "epoch": 0.2868937048503612,
+      "grad_norm": 14.625,
+      "learning_rate": 1.7946646802837416e-06,
+      "loss": 2.0461,
+      "step": 9730
+    },
+    {
+      "epoch": 0.28704113224237066,
+      "grad_norm": 18.5,
+      "learning_rate": 1.7943521713828883e-06,
+      "loss": 2.113,
+      "step": 9735
+    },
+    {
+      "epoch": 0.28718855963438006,
+      "grad_norm": 17.25,
+      "learning_rate": 1.7940394521049667e-06,
+      "loss": 2.2751,
+      "step": 9740
+    },
+    {
+      "epoch": 0.2873359870263895,
+      "grad_norm": 13.25,
+      "learning_rate": 1.7937265225327983e-06,
+      "loss": 2.1439,
+      "step": 9745
+    },
+    {
+      "epoch": 0.2874834144183989,
+      "grad_norm": 18.125,
+      "learning_rate": 1.793413382749259e-06,
+      "loss": 2.2289,
+      "step": 9750
+    },
+    {
+      "epoch": 0.2876308418104084,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.7931000328372818e-06,
+      "loss": 2.3197,
+      "step": 9755
+    },
+    {
+      "epoch": 0.28777826920241784,
+      "grad_norm": 15.0,
+      "learning_rate": 1.7927864728798543e-06,
+      "loss": 2.1508,
+      "step": 9760
+    },
+    {
+      "epoch": 0.28792569659442724,
+      "grad_norm": 12.0,
+      "learning_rate": 1.7924727029600198e-06,
+      "loss": 2.1825,
+      "step": 9765
+    },
+    {
+      "epoch": 0.2880731239864367,
+      "grad_norm": 13.125,
+      "learning_rate": 1.7921587231608777e-06,
+      "loss": 2.1113,
+      "step": 9770
+    },
+    {
+      "epoch": 0.2882205513784461,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.791844533565583e-06,
+      "loss": 2.1209,
+      "step": 9775
+    },
+    {
+      "epoch": 0.28836797877045556,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.7915301342573455e-06,
+      "loss": 2.2531,
+      "step": 9780
+    },
+    {
+      "epoch": 0.28851540616246496,
+      "grad_norm": 13.25,
+      "learning_rate": 1.7912155253194311e-06,
+      "loss": 2.1772,
+      "step": 9785
+    },
+    {
+      "epoch": 0.2886628335544744,
+      "grad_norm": 13.375,
+      "learning_rate": 1.7909007068351617e-06,
+      "loss": 1.9749,
+      "step": 9790
+    },
+    {
+      "epoch": 0.2888102609464839,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.7905856788879135e-06,
+      "loss": 2.2297,
+      "step": 9795
+    },
+    {
+      "epoch": 0.2889576883384933,
+      "grad_norm": 21.125,
+      "learning_rate": 1.7902704415611194e-06,
+      "loss": 2.1577,
+      "step": 9800
+    },
+    {
+      "epoch": 0.28910511573050274,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.7899549949382667e-06,
+      "loss": 2.1224,
+      "step": 9805
+    },
+    {
+      "epoch": 0.28925254312251214,
+      "grad_norm": 15.375,
+      "learning_rate": 1.789639339102899e-06,
+      "loss": 2.1734,
+      "step": 9810
+    },
+    {
+      "epoch": 0.2893999705145216,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.7893234741386148e-06,
+      "loss": 1.9936,
+      "step": 9815
+    },
+    {
+      "epoch": 0.28954739790653106,
+      "grad_norm": 15.625,
+      "learning_rate": 1.7890074001290678e-06,
+      "loss": 2.1629,
+      "step": 9820
+    },
+    {
+      "epoch": 0.28969482529854046,
+      "grad_norm": 15.125,
+      "learning_rate": 1.7886911171579678e-06,
+      "loss": 2.1329,
+      "step": 9825
+    },
+    {
+      "epoch": 0.2898422526905499,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.7883746253090792e-06,
+      "loss": 2.2002,
+      "step": 9830
+    },
+    {
+      "epoch": 0.2899896800825593,
+      "grad_norm": 18.5,
+      "learning_rate": 1.7880579246662225e-06,
+      "loss": 2.1516,
+      "step": 9835
+    },
+    {
+      "epoch": 0.2901371074745688,
+      "grad_norm": 17.0,
+      "learning_rate": 1.7877410153132727e-06,
+      "loss": 2.3072,
+      "step": 9840
+    },
+    {
+      "epoch": 0.29028453486657824,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.7874238973341602e-06,
+      "loss": 2.094,
+      "step": 9845
+    },
+    {
+      "epoch": 0.29043196225858764,
+      "grad_norm": 14.0,
+      "learning_rate": 1.7871065708128712e-06,
+      "loss": 2.1111,
+      "step": 9850
+    },
+    {
+      "epoch": 0.2905793896505971,
+      "grad_norm": 15.625,
+      "learning_rate": 1.7867890358334465e-06,
+      "loss": 2.2635,
+      "step": 9855
+    },
+    {
+      "epoch": 0.2907268170426065,
+      "grad_norm": 17.625,
+      "learning_rate": 1.7864712924799829e-06,
+      "loss": 2.0714,
+      "step": 9860
+    },
+    {
+      "epoch": 0.29087424443461596,
+      "grad_norm": 14.875,
+      "learning_rate": 1.7861533408366315e-06,
+      "loss": 2.1918,
+      "step": 9865
+    },
+    {
+      "epoch": 0.29102167182662536,
+      "grad_norm": 15.125,
+      "learning_rate": 1.7858351809875992e-06,
+      "loss": 2.1891,
+      "step": 9870
+    },
+    {
+      "epoch": 0.2911690992186348,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.7855168130171471e-06,
+      "loss": 2.2214,
+      "step": 9875
+    },
+    {
+      "epoch": 0.2913165266106443,
+      "grad_norm": 13.875,
+      "learning_rate": 1.785198237009593e-06,
+      "loss": 2.1657,
+      "step": 9880
+    },
+    {
+      "epoch": 0.2914639540026537,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.7848794530493083e-06,
+      "loss": 2.2377,
+      "step": 9885
+    },
+    {
+      "epoch": 0.29161138139466314,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.7845604612207206e-06,
+      "loss": 2.1852,
+      "step": 9890
+    },
+    {
+      "epoch": 0.29175880878667254,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.7842412616083116e-06,
+      "loss": 2.2394,
+      "step": 9895
+    },
+    {
+      "epoch": 0.291906236178682,
+      "grad_norm": 16.5,
+      "learning_rate": 1.7839218542966185e-06,
+      "loss": 2.2163,
+      "step": 9900
+    },
+    {
+      "epoch": 0.29205366357069146,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.7836022393702334e-06,
+      "loss": 2.1405,
+      "step": 9905
+    },
+    {
+      "epoch": 0.29220109096270086,
+      "grad_norm": 18.625,
+      "learning_rate": 1.783282416913804e-06,
+      "loss": 2.2301,
+      "step": 9910
+    },
+    {
+      "epoch": 0.2923485183547103,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.782962387012032e-06,
+      "loss": 2.0984,
+      "step": 9915
+    },
+    {
+      "epoch": 0.2924959457467197,
+      "grad_norm": 14.0,
+      "learning_rate": 1.7826421497496743e-06,
+      "loss": 2.1405,
+      "step": 9920
+    },
+    {
+      "epoch": 0.2926433731387292,
+      "grad_norm": 15.5,
+      "learning_rate": 1.7823217052115434e-06,
+      "loss": 2.2241,
+      "step": 9925
+    },
+    {
+      "epoch": 0.29279080053073864,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.7820010534825057e-06,
+      "loss": 2.1421,
+      "step": 9930
+    },
+    {
+      "epoch": 0.29293822792274804,
+      "grad_norm": 16.625,
+      "learning_rate": 1.7816801946474832e-06,
+      "loss": 2.1865,
+      "step": 9935
+    },
+    {
+      "epoch": 0.2930856553147575,
+      "grad_norm": 11.625,
+      "learning_rate": 1.781359128791452e-06,
+      "loss": 2.1026,
+      "step": 9940
+    },
+    {
+      "epoch": 0.2932330827067669,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.7810378559994442e-06,
+      "loss": 2.1224,
+      "step": 9945
+    },
+    {
+      "epoch": 0.29338051009877636,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.7807163763565457e-06,
+      "loss": 2.1026,
+      "step": 9950
+    },
+    {
+      "epoch": 0.29352793749078576,
+      "grad_norm": 17.75,
+      "learning_rate": 1.7803946899478972e-06,
+      "loss": 2.1495,
+      "step": 9955
+    },
+    {
+      "epoch": 0.2936753648827952,
+      "grad_norm": 14.5,
+      "learning_rate": 1.7800727968586952e-06,
+      "loss": 2.2219,
+      "step": 9960
+    },
+    {
+      "epoch": 0.2938227922748047,
+      "grad_norm": 14.5,
+      "learning_rate": 1.7797506971741899e-06,
+      "loss": 2.1837,
+      "step": 9965
+    },
+    {
+      "epoch": 0.2939702196668141,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.779428390979686e-06,
+      "loss": 2.0926,
+      "step": 9970
+    },
+    {
+      "epoch": 0.29411764705882354,
+      "grad_norm": 24.625,
+      "learning_rate": 1.7791058783605442e-06,
+      "loss": 2.192,
+      "step": 9975
+    },
+    {
+      "epoch": 0.29426507445083294,
+      "grad_norm": 13.75,
+      "learning_rate": 1.7787831594021787e-06,
+      "loss": 2.2558,
+      "step": 9980
+    },
+    {
+      "epoch": 0.2944125018428424,
+      "grad_norm": 15.125,
+      "learning_rate": 1.7784602341900585e-06,
+      "loss": 2.2184,
+      "step": 9985
+    },
+    {
+      "epoch": 0.29455992923485186,
+      "grad_norm": 26.125,
+      "learning_rate": 1.7781371028097079e-06,
+      "loss": 2.2509,
+      "step": 9990
+    },
+    {
+      "epoch": 0.29470735662686126,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.7778137653467052e-06,
+      "loss": 2.2607,
+      "step": 9995
+    },
+    {
+      "epoch": 0.2948547840188707,
+      "grad_norm": 13.875,
+      "learning_rate": 1.7774902218866833e-06,
+      "loss": 2.1507,
+      "step": 10000
+    },
+    {
+      "epoch": 0.2948547840188707,
+      "eval_loss": 2.1286072731018066,
+      "eval_runtime": 4.7121,
+      "eval_samples_per_second": 84.038,
+      "eval_steps_per_second": 2.759,
+      "step": 10000
     }
   ],
   "logging_steps": 5,
@@ -7113,7 +14193,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 1.3986311890783961e+19,
+  "total_flos": 2.7972623782104793e+19,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null