diff --git "a/checkpoints/checkpoint-34495/trainer_state.json" "b/checkpoints/checkpoint-34495/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoints/checkpoint-34495/trainer_state.json"
@@ -0,0 +1,5128 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.999818820886328,
+  "eval_steps": 1000,
+  "global_step": 34495,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.007247164546871037,
+      "grad_norm": 3.8405699729919434,
+      "learning_rate": 4.9928975213799106e-05,
+      "loss": 1.4877,
+      "step": 50
+    },
+    {
+      "epoch": 0.014494329093742073,
+      "grad_norm": 4.489353179931641,
+      "learning_rate": 4.9856500942165534e-05,
+      "loss": 1.1021,
+      "step": 100
+    },
+    {
+      "epoch": 0.021741493640613112,
+      "grad_norm": 5.1422295570373535,
+      "learning_rate": 4.978402667053196e-05,
+      "loss": 1.0659,
+      "step": 150
+    },
+    {
+      "epoch": 0.028988658187484147,
+      "grad_norm": 4.703744411468506,
+      "learning_rate": 4.971155239889839e-05,
+      "loss": 1.0636,
+      "step": 200
+    },
+    {
+      "epoch": 0.03623582273435518,
+      "grad_norm": 3.330397605895996,
+      "learning_rate": 4.963907812726482e-05,
+      "loss": 1.0025,
+      "step": 250
+    },
+    {
+      "epoch": 0.043482987281226224,
+      "grad_norm": 3.511565685272217,
+      "learning_rate": 4.9566603855631256e-05,
+      "loss": 0.9945,
+      "step": 300
+    },
+    {
+      "epoch": 0.05073015182809726,
+      "grad_norm": 3.470792055130005,
+      "learning_rate": 4.9494129583997684e-05,
+      "loss": 1.0228,
+      "step": 350
+    },
+    {
+      "epoch": 0.057977316374968293,
+      "grad_norm": 3.1396453380584717,
+      "learning_rate": 4.942165531236411e-05,
+      "loss": 0.9935,
+      "step": 400
+    },
+    {
+      "epoch": 0.06522448092183933,
+      "grad_norm": 3.4228012561798096,
+      "learning_rate": 4.934918104073054e-05,
+      "loss": 0.9904,
+      "step": 450
+    },
+    {
+      "epoch": 0.07247164546871036,
+      "grad_norm": 3.423574447631836,
+      "learning_rate": 4.927670676909698e-05,
+      "loss": 0.9884,
+      "step": 500
+    },
+    {
+      "epoch": 0.0797188100155814,
+      "grad_norm": 2.9020352363586426,
+      "learning_rate": 4.9204232497463405e-05,
+      "loss": 0.9093,
+      "step": 550
+    },
+    {
+      "epoch": 0.08696597456245245,
+      "grad_norm": 3.0173707008361816,
+      "learning_rate": 4.9131758225829834e-05,
+      "loss": 0.9374,
+      "step": 600
+    },
+    {
+      "epoch": 0.09421313910932348,
+      "grad_norm": 3.4516496658325195,
+      "learning_rate": 4.905928395419626e-05,
+      "loss": 0.9798,
+      "step": 650
+    },
+    {
+      "epoch": 0.10146030365619452,
+      "grad_norm": 2.8202695846557617,
+      "learning_rate": 4.898680968256269e-05,
+      "loss": 0.9293,
+      "step": 700
+    },
+    {
+      "epoch": 0.10870746820306555,
+      "grad_norm": 3.3415849208831787,
+      "learning_rate": 4.8914335410929126e-05,
+      "loss": 0.9622,
+      "step": 750
+    },
+    {
+      "epoch": 0.11595463274993659,
+      "grad_norm": 3.4417173862457275,
+      "learning_rate": 4.8841861139295555e-05,
+      "loss": 0.9363,
+      "step": 800
+    },
+    {
+      "epoch": 0.12320179729680762,
+      "grad_norm": 3.051495313644409,
+      "learning_rate": 4.8769386867661983e-05,
+      "loss": 0.9086,
+      "step": 850
+    },
+    {
+      "epoch": 0.13044896184367866,
+      "grad_norm": 2.5990054607391357,
+      "learning_rate": 4.869691259602841e-05,
+      "loss": 0.8976,
+      "step": 900
+    },
+    {
+      "epoch": 0.1376961263905497,
+      "grad_norm": 3.3123867511749268,
+      "learning_rate": 4.862443832439485e-05,
+      "loss": 0.9568,
+      "step": 950
+    },
+    {
+      "epoch": 0.14494329093742073,
+      "grad_norm": 2.7175092697143555,
+      "learning_rate": 4.8551964052761276e-05,
+      "loss": 0.8665,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14494329093742073,
+      "eval_loss": 0.855056643486023,
+      "eval_runtime": 245.3157,
+      "eval_samples_per_second": 140.615,
+      "eval_steps_per_second": 14.064,
+      "step": 1000
+    },
+    {
+      "epoch": 0.15219045548429178,
+      "grad_norm": 2.541044235229492,
+      "learning_rate": 4.848093926656037e-05,
+      "loss": 0.887,
+      "step": 1050
+    },
+    {
+      "epoch": 0.1594376200311628,
+      "grad_norm": 3.0265707969665527,
+      "learning_rate": 4.840846499492681e-05,
+      "loss": 0.8987,
+      "step": 1100
+    },
+    {
+      "epoch": 0.16668478457803385,
+      "grad_norm": 3.479360818862915,
+      "learning_rate": 4.8335990723293237e-05,
+      "loss": 0.8744,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1739319491249049,
+      "grad_norm": 3.381410598754883,
+      "learning_rate": 4.8263516451659665e-05,
+      "loss": 0.869,
+      "step": 1200
+    },
+    {
+      "epoch": 0.18117911367177592,
+      "grad_norm": 3.526550769805908,
+      "learning_rate": 4.8191042180026094e-05,
+      "loss": 0.8753,
+      "step": 1250
+    },
+    {
+      "epoch": 0.18842627821864696,
+      "grad_norm": 3.5922884941101074,
+      "learning_rate": 4.811856790839252e-05,
+      "loss": 0.8661,
+      "step": 1300
+    },
+    {
+      "epoch": 0.19567344276551799,
+      "grad_norm": 3.390925168991089,
+      "learning_rate": 4.804609363675896e-05,
+      "loss": 0.865,
+      "step": 1350
+    },
+    {
+      "epoch": 0.20292060731238903,
+      "grad_norm": 3.6313884258270264,
+      "learning_rate": 4.7973619365125386e-05,
+      "loss": 0.9077,
+      "step": 1400
+    },
+    {
+      "epoch": 0.21016777185926006,
+      "grad_norm": 3.6675050258636475,
+      "learning_rate": 4.7901145093491815e-05,
+      "loss": 0.8724,
+      "step": 1450
+    },
+    {
+      "epoch": 0.2174149364061311,
+      "grad_norm": 2.9119815826416016,
+      "learning_rate": 4.782867082185824e-05,
+      "loss": 0.887,
+      "step": 1500
+    },
+    {
+      "epoch": 0.22466210095300213,
+      "grad_norm": 3.978513479232788,
+      "learning_rate": 4.775619655022467e-05,
+      "loss": 0.909,
+      "step": 1550
+    },
+    {
+      "epoch": 0.23190926549987317,
+      "grad_norm": 2.6962685585021973,
+      "learning_rate": 4.76837222785911e-05,
+      "loss": 0.9013,
+      "step": 1600
+    },
+    {
+      "epoch": 0.23915643004674422,
+      "grad_norm": 3.688558578491211,
+      "learning_rate": 4.761124800695753e-05,
+      "loss": 0.8777,
+      "step": 1650
+    },
+    {
+      "epoch": 0.24640359459361524,
+      "grad_norm": 2.6610682010650635,
+      "learning_rate": 4.753877373532396e-05,
+      "loss": 0.854,
+      "step": 1700
+    },
+    {
+      "epoch": 0.2536507591404863,
+      "grad_norm": 2.8136966228485107,
+      "learning_rate": 4.746629946369039e-05,
+      "loss": 0.8523,
+      "step": 1750
+    },
+    {
+      "epoch": 0.2608979236873573,
+      "grad_norm": 2.660381555557251,
+      "learning_rate": 4.739382519205682e-05,
+      "loss": 0.8764,
+      "step": 1800
+    },
+    {
+      "epoch": 0.26814508823422833,
+      "grad_norm": 2.8120713233947754,
+      "learning_rate": 4.732135092042325e-05,
+      "loss": 0.8401,
+      "step": 1850
+    },
+    {
+      "epoch": 0.2753922527810994,
+      "grad_norm": 3.686016321182251,
+      "learning_rate": 4.724887664878968e-05,
+      "loss": 0.8891,
+      "step": 1900
+    },
+    {
+      "epoch": 0.28263941732797043,
+      "grad_norm": 3.8460748195648193,
+      "learning_rate": 4.717640237715611e-05,
+      "loss": 0.8909,
+      "step": 1950
+    },
+    {
+      "epoch": 0.28988658187484145,
+      "grad_norm": 3.1707077026367188,
+      "learning_rate": 4.710392810552254e-05,
+      "loss": 0.8564,
+      "step": 2000
+    },
+    {
+      "epoch": 0.28988658187484145,
+      "eval_loss": 0.8084499835968018,
+      "eval_runtime": 245.4746,
+      "eval_samples_per_second": 140.524,
+      "eval_steps_per_second": 14.054,
+      "step": 2000
+    },
+    {
+      "epoch": 0.29713374642171253,
+      "grad_norm": 2.892038106918335,
+      "learning_rate": 4.703145383388897e-05,
+      "loss": 0.8496,
+      "step": 2050
+    },
+    {
+      "epoch": 0.30438091096858355,
+      "grad_norm": 3.043569803237915,
+      "learning_rate": 4.69589795622554e-05,
+      "loss": 0.9013,
+      "step": 2100
+    },
+    {
+      "epoch": 0.3116280755154546,
+      "grad_norm": 2.6177148818969727,
+      "learning_rate": 4.688650529062183e-05,
+      "loss": 0.8438,
+      "step": 2150
+    },
+    {
+      "epoch": 0.3188752400623256,
+      "grad_norm": 3.0475521087646484,
+      "learning_rate": 4.6814031018988264e-05,
+      "loss": 0.8892,
+      "step": 2200
+    },
+    {
+      "epoch": 0.32612240460919667,
+      "grad_norm": 3.0972824096679688,
+      "learning_rate": 4.674155674735469e-05,
+      "loss": 0.8019,
+      "step": 2250
+    },
+    {
+      "epoch": 0.3333695691560677,
+      "grad_norm": 3.916921854019165,
+      "learning_rate": 4.666908247572112e-05,
+      "loss": 0.8278,
+      "step": 2300
+    },
+    {
+      "epoch": 0.3406167337029387,
+      "grad_norm": 2.8206372261047363,
+      "learning_rate": 4.659660820408755e-05,
+      "loss": 0.8532,
+      "step": 2350
+    },
+    {
+      "epoch": 0.3478638982498098,
+      "grad_norm": 3.05037260055542,
+      "learning_rate": 4.6524133932453985e-05,
+      "loss": 0.8516,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3551110627966808,
+      "grad_norm": 2.940920829772949,
+      "learning_rate": 4.645165966082041e-05,
+      "loss": 0.8326,
+      "step": 2450
+    },
+    {
+      "epoch": 0.36235822734355183,
+      "grad_norm": 2.752655267715454,
+      "learning_rate": 4.637918538918684e-05,
+      "loss": 0.8705,
+      "step": 2500
+    },
+    {
+      "epoch": 0.36960539189042285,
+      "grad_norm": 2.369166851043701,
+      "learning_rate": 4.630671111755327e-05,
+      "loss": 0.8111,
+      "step": 2550
+    },
+    {
+      "epoch": 0.37685255643729393,
+      "grad_norm": 2.991446018218994,
+      "learning_rate": 4.62342368459197e-05,
+      "loss": 0.8379,
+      "step": 2600
+    },
+    {
+      "epoch": 0.38409972098416495,
+      "grad_norm": 2.6964468955993652,
+      "learning_rate": 4.6161762574286134e-05,
+      "loss": 0.8459,
+      "step": 2650
+    },
+    {
+      "epoch": 0.39134688553103597,
+      "grad_norm": 2.8385136127471924,
+      "learning_rate": 4.608928830265256e-05,
+      "loss": 0.8301,
+      "step": 2700
+    },
+    {
+      "epoch": 0.398594050077907,
+      "grad_norm": 2.3763439655303955,
+      "learning_rate": 4.601681403101899e-05,
+      "loss": 0.7899,
+      "step": 2750
+    },
+    {
+      "epoch": 0.40584121462477807,
+      "grad_norm": 3.157857894897461,
+      "learning_rate": 4.594433975938542e-05,
+      "loss": 0.8118,
+      "step": 2800
+    },
+    {
+      "epoch": 0.4130883791716491,
+      "grad_norm": 3.2475242614746094,
+      "learning_rate": 4.5871865487751855e-05,
+      "loss": 0.8096,
+      "step": 2850
+    },
+    {
+      "epoch": 0.4203355437185201,
+      "grad_norm": 2.474079132080078,
+      "learning_rate": 4.5799391216118284e-05,
+      "loss": 0.8164,
+      "step": 2900
+    },
+    {
+      "epoch": 0.4275827082653912,
+      "grad_norm": 2.8297600746154785,
+      "learning_rate": 4.572691694448471e-05,
+      "loss": 0.7803,
+      "step": 2950
+    },
+    {
+      "epoch": 0.4348298728122622,
+      "grad_norm": 3.616157054901123,
+      "learning_rate": 4.565444267285114e-05,
+      "loss": 0.7867,
+      "step": 3000
+    },
+    {
+      "epoch": 0.4348298728122622,
+      "eval_loss": 0.7722809910774231,
+      "eval_runtime": 245.4414,
+      "eval_samples_per_second": 140.543,
+      "eval_steps_per_second": 14.056,
+      "step": 3000
+    },
+    {
+      "epoch": 0.44207703735913323,
+      "grad_norm": 2.828176736831665,
+      "learning_rate": 4.558196840121757e-05,
+      "loss": 0.8496,
+      "step": 3050
+    },
+    {
+      "epoch": 0.44932420190600425,
+      "grad_norm": 3.1643614768981934,
+      "learning_rate": 4.5509494129584e-05,
+      "loss": 0.7921,
+      "step": 3100
+    },
+    {
+      "epoch": 0.4565713664528753,
+      "grad_norm": 3.158106803894043,
+      "learning_rate": 4.543701985795043e-05,
+      "loss": 0.8128,
+      "step": 3150
+    },
+    {
+      "epoch": 0.46381853099974635,
+      "grad_norm": 3.462926149368286,
+      "learning_rate": 4.536454558631686e-05,
+      "loss": 0.7772,
+      "step": 3200
+    },
+    {
+      "epoch": 0.47106569554661737,
+      "grad_norm": 2.819894552230835,
+      "learning_rate": 4.529207131468329e-05,
+      "loss": 0.7911,
+      "step": 3250
+    },
+    {
+      "epoch": 0.47831286009348845,
+      "grad_norm": 2.3228747844696045,
+      "learning_rate": 4.521959704304972e-05,
+      "loss": 0.7974,
+      "step": 3300
+    },
+    {
+      "epoch": 0.48556002464035947,
+      "grad_norm": 2.4810431003570557,
+      "learning_rate": 4.514712277141615e-05,
+      "loss": 0.8064,
+      "step": 3350
+    },
+    {
+      "epoch": 0.4928071891872305,
+      "grad_norm": 2.91050124168396,
+      "learning_rate": 4.5074648499782577e-05,
+      "loss": 0.8269,
+      "step": 3400
+    },
+    {
+      "epoch": 0.5000543537341016,
+      "grad_norm": 2.8636369705200195,
+      "learning_rate": 4.5002174228149005e-05,
+      "loss": 0.8139,
+      "step": 3450
+    },
+    {
+      "epoch": 0.5073015182809726,
+      "grad_norm": 2.4558308124542236,
+      "learning_rate": 4.4929699956515434e-05,
+      "loss": 0.7364,
+      "step": 3500
+    },
+    {
+      "epoch": 0.5145486828278436,
+      "grad_norm": 2.5896005630493164,
+      "learning_rate": 4.485722568488187e-05,
+      "loss": 0.7858,
+      "step": 3550
+    },
+    {
+      "epoch": 0.5217958473747146,
+      "grad_norm": 2.745985746383667,
+      "learning_rate": 4.47847514132483e-05,
+      "loss": 0.8279,
+      "step": 3600
+    },
+    {
+      "epoch": 0.5290430119215856,
+      "grad_norm": 3.3217363357543945,
+      "learning_rate": 4.4712277141614726e-05,
+      "loss": 0.7929,
+      "step": 3650
+    },
+    {
+      "epoch": 0.5362901764684567,
+      "grad_norm": 2.981140375137329,
+      "learning_rate": 4.4639802869981155e-05,
+      "loss": 0.8282,
+      "step": 3700
+    },
+    {
+      "epoch": 0.5435373410153278,
+      "grad_norm": 3.112213611602783,
+      "learning_rate": 4.456732859834759e-05,
+      "loss": 0.7823,
+      "step": 3750
+    },
+    {
+      "epoch": 0.5507845055621988,
+      "grad_norm": 2.5669238567352295,
+      "learning_rate": 4.449485432671402e-05,
+      "loss": 0.8118,
+      "step": 3800
+    },
+    {
+      "epoch": 0.5580316701090698,
+      "grad_norm": 2.7261078357696533,
+      "learning_rate": 4.442238005508045e-05,
+      "loss": 0.8098,
+      "step": 3850
+    },
+    {
+      "epoch": 0.5652788346559409,
+      "grad_norm": 2.5646941661834717,
+      "learning_rate": 4.4349905783446876e-05,
+      "loss": 0.8,
+      "step": 3900
+    },
+    {
+      "epoch": 0.5725259992028119,
+      "grad_norm": 2.627681016921997,
+      "learning_rate": 4.427743151181331e-05,
+      "loss": 0.7732,
+      "step": 3950
+    },
+    {
+      "epoch": 0.5797731637496829,
+      "grad_norm": 2.443345069885254,
+      "learning_rate": 4.420495724017974e-05,
+      "loss": 0.7949,
+      "step": 4000
+    },
+    {
+      "epoch": 0.5797731637496829,
+      "eval_loss": 0.7400076389312744,
+      "eval_runtime": 245.51,
+      "eval_samples_per_second": 140.503,
+      "eval_steps_per_second": 14.052,
+      "step": 4000
+    },
+    {
+      "epoch": 0.5870203282965539,
+      "grad_norm": 2.263333559036255,
+      "learning_rate": 4.413248296854617e-05,
+      "loss": 0.7523,
+      "step": 4050
+    },
+    {
+      "epoch": 0.5942674928434251,
+      "grad_norm": 2.655808210372925,
+      "learning_rate": 4.40600086969126e-05,
+      "loss": 0.7489,
+      "step": 4100
+    },
+    {
+      "epoch": 0.6015146573902961,
+      "grad_norm": 3.2610058784484863,
+      "learning_rate": 4.3987534425279025e-05,
+      "loss": 0.7853,
+      "step": 4150
+    },
+    {
+      "epoch": 0.6087618219371671,
+      "grad_norm": 2.82106876373291,
+      "learning_rate": 4.391506015364546e-05,
+      "loss": 0.7688,
+      "step": 4200
+    },
+    {
+      "epoch": 0.6160089864840381,
+      "grad_norm": 1.9377267360687256,
+      "learning_rate": 4.384258588201189e-05,
+      "loss": 0.7887,
+      "step": 4250
+    },
+    {
+      "epoch": 0.6232561510309091,
+      "grad_norm": 2.4087536334991455,
+      "learning_rate": 4.377011161037832e-05,
+      "loss": 0.8098,
+      "step": 4300
+    },
+    {
+      "epoch": 0.6305033155777802,
+      "grad_norm": 5.600491046905518,
+      "learning_rate": 4.3697637338744747e-05,
+      "loss": 0.7683,
+      "step": 4350
+    },
+    {
+      "epoch": 0.6377504801246512,
+      "grad_norm": 3.0500433444976807,
+      "learning_rate": 4.362516306711118e-05,
+      "loss": 0.7995,
+      "step": 4400
+    },
+    {
+      "epoch": 0.6449976446715223,
+      "grad_norm": 1.7417362928390503,
+      "learning_rate": 4.355268879547761e-05,
+      "loss": 0.7822,
+      "step": 4450
+    },
+    {
+      "epoch": 0.6522448092183933,
+      "grad_norm": 2.1057236194610596,
+      "learning_rate": 4.348021452384404e-05,
+      "loss": 0.7812,
+      "step": 4500
+    },
+    {
+      "epoch": 0.6594919737652644,
+      "grad_norm": 2.780358076095581,
+      "learning_rate": 4.341063922307581e-05,
+      "loss": 0.849,
+      "step": 4550
+    },
+    {
+      "epoch": 0.6667391383121354,
+      "grad_norm": 2.8940329551696777,
+      "learning_rate": 4.333816495144224e-05,
+      "loss": 0.7883,
+      "step": 4600
+    },
+    {
+      "epoch": 0.6739863028590064,
+      "grad_norm": 2.65120530128479,
+      "learning_rate": 4.326569067980867e-05,
+      "loss": 0.8011,
+      "step": 4650
+    },
+    {
+      "epoch": 0.6812334674058774,
+      "grad_norm": 3.0340774059295654,
+      "learning_rate": 4.31932164081751e-05,
+      "loss": 0.7784,
+      "step": 4700
+    },
+    {
+      "epoch": 0.6884806319527484,
+      "grad_norm": 3.227555990219116,
+      "learning_rate": 4.312074213654153e-05,
+      "loss": 0.7798,
+      "step": 4750
+    },
+    {
+      "epoch": 0.6957277964996196,
+      "grad_norm": 2.9701457023620605,
+      "learning_rate": 4.304826786490796e-05,
+      "loss": 0.7623,
+      "step": 4800
+    },
+    {
+      "epoch": 0.7029749610464906,
+      "grad_norm": 3.183608293533325,
+      "learning_rate": 4.297579359327439e-05,
+      "loss": 0.7591,
+      "step": 4850
+    },
+    {
+      "epoch": 0.7102221255933616,
+      "grad_norm": 3.034790277481079,
+      "learning_rate": 4.290331932164082e-05,
+      "loss": 0.75,
+      "step": 4900
+    },
+    {
+      "epoch": 0.7174692901402326,
+      "grad_norm": 2.8872621059417725,
+      "learning_rate": 4.283084505000725e-05,
+      "loss": 0.7672,
+      "step": 4950
+    },
+    {
+      "epoch": 0.7247164546871037,
+      "grad_norm": 2.583475112915039,
+      "learning_rate": 4.275837077837368e-05,
+      "loss": 0.7396,
+      "step": 5000
+    },
+    {
+      "epoch": 0.7247164546871037,
+      "eval_loss": 0.7259831428527832,
+      "eval_runtime": 245.9324,
+      "eval_samples_per_second": 140.262,
+      "eval_steps_per_second": 14.028,
+      "step": 5000
+    },
+    {
+      "epoch": 0.7319636192339747,
+      "grad_norm": 2.8117451667785645,
+      "learning_rate": 4.268589650674011e-05,
+      "loss": 0.7471,
+      "step": 5050
+    },
+    {
+      "epoch": 0.7392107837808457,
+      "grad_norm": 2.556171178817749,
+      "learning_rate": 4.261342223510654e-05,
+      "loss": 0.734,
+      "step": 5100
+    },
+    {
+      "epoch": 0.7464579483277168,
+      "grad_norm": 2.494264602661133,
+      "learning_rate": 4.2540947963472974e-05,
+      "loss": 0.7657,
+      "step": 5150
+    },
+    {
+      "epoch": 0.7537051128745879,
+      "grad_norm": 2.5196897983551025,
+      "learning_rate": 4.24684736918394e-05,
+      "loss": 0.7373,
+      "step": 5200
+    },
+    {
+      "epoch": 0.7609522774214589,
+      "grad_norm": 1.9920554161071777,
+      "learning_rate": 4.239599942020583e-05,
+      "loss": 0.7603,
+      "step": 5250
+    },
+    {
+      "epoch": 0.7681994419683299,
+      "grad_norm": 3.2272536754608154,
+      "learning_rate": 4.232352514857226e-05,
+      "loss": 0.798,
+      "step": 5300
+    },
+    {
+      "epoch": 0.7754466065152009,
+      "grad_norm": 2.447430372238159,
+      "learning_rate": 4.225105087693869e-05,
+      "loss": 0.7918,
+      "step": 5350
+    },
+    {
+      "epoch": 0.7826937710620719,
+      "grad_norm": 2.375603199005127,
+      "learning_rate": 4.217857660530512e-05,
+      "loss": 0.7578,
+      "step": 5400
+    },
+    {
+      "epoch": 0.789940935608943,
+      "grad_norm": 3.0463624000549316,
+      "learning_rate": 4.210610233367155e-05,
+      "loss": 0.749,
+      "step": 5450
+    },
+    {
+      "epoch": 0.797188100155814,
+      "grad_norm": 1.8498198986053467,
+      "learning_rate": 4.203362806203798e-05,
+      "loss": 0.7619,
+      "step": 5500
+    },
+    {
+      "epoch": 0.8044352647026851,
+      "grad_norm": 3.3921091556549072,
+      "learning_rate": 4.196115379040441e-05,
+      "loss": 0.7655,
+      "step": 5550
+    },
+    {
+      "epoch": 0.8116824292495561,
+      "grad_norm": 3.1778969764709473,
+      "learning_rate": 4.188867951877084e-05,
+      "loss": 0.7119,
+      "step": 5600
+    },
+    {
+      "epoch": 0.8189295937964272,
+      "grad_norm": 2.292695999145508,
+      "learning_rate": 4.1816205247137266e-05,
+      "loss": 0.7378,
+      "step": 5650
+    },
+    {
+      "epoch": 0.8261767583432982,
+      "grad_norm": 2.388732671737671,
+      "learning_rate": 4.1743730975503695e-05,
+      "loss": 0.7376,
+      "step": 5700
+    },
+    {
+      "epoch": 0.8334239228901692,
+      "grad_norm": 2.4461729526519775,
+      "learning_rate": 4.1671256703870124e-05,
+      "loss": 0.7656,
+      "step": 5750
+    },
+    {
+      "epoch": 0.8406710874370402,
+      "grad_norm": 3.55210280418396,
+      "learning_rate": 4.159878243223656e-05,
+      "loss": 0.7814,
+      "step": 5800
+    },
+    {
+      "epoch": 0.8479182519839112,
+      "grad_norm": 2.640709400177002,
+      "learning_rate": 4.152630816060299e-05,
+      "loss": 0.7541,
+      "step": 5850
+    },
+    {
+      "epoch": 0.8551654165307824,
+      "grad_norm": 2.837186336517334,
+      "learning_rate": 4.1453833888969416e-05,
+      "loss": 0.7492,
+      "step": 5900
+    },
+    {
+      "epoch": 0.8624125810776534,
+      "grad_norm": 3.748387336730957,
+      "learning_rate": 4.1381359617335845e-05,
+      "loss": 0.7317,
+      "step": 5950
+    },
+    {
+      "epoch": 0.8696597456245244,
+      "grad_norm": 2.7219724655151367,
+      "learning_rate": 4.130888534570227e-05,
+      "loss": 0.7546,
+      "step": 6000
+    },
+    {
+      "epoch": 0.8696597456245244,
+      "eval_loss": 0.7048457860946655,
+      "eval_runtime": 245.6596,
+      "eval_samples_per_second": 140.418,
+      "eval_steps_per_second": 14.044,
+      "step": 6000
+    },
+    {
+      "epoch": 0.8769069101713954,
+      "grad_norm": 3.017869234085083,
+      "learning_rate": 4.123641107406871e-05,
+      "loss": 0.7354,
+      "step": 6050
+    },
+    {
+      "epoch": 0.8841540747182665,
+      "grad_norm": 2.677828550338745,
+      "learning_rate": 4.116393680243514e-05,
+      "loss": 0.746,
+      "step": 6100
+    },
+    {
+      "epoch": 0.8914012392651375,
+      "grad_norm": 2.0534462928771973,
+      "learning_rate": 4.1091462530801566e-05,
+      "loss": 0.7614,
+      "step": 6150
+    },
+    {
+      "epoch": 0.8986484038120085,
+      "grad_norm": 3.0294041633605957,
+      "learning_rate": 4.1018988259167994e-05,
+      "loss": 0.7256,
+      "step": 6200
+    },
+    {
+      "epoch": 0.9058955683588796,
+      "grad_norm": 2.640902042388916,
+      "learning_rate": 4.094651398753443e-05,
+      "loss": 0.7276,
+      "step": 6250
+    },
+    {
+      "epoch": 0.9131427329057507,
+      "grad_norm": 2.773237466812134,
+      "learning_rate": 4.087403971590086e-05,
+      "loss": 0.7404,
+      "step": 6300
+    },
+    {
+      "epoch": 0.9203898974526217,
+      "grad_norm": 2.8102800846099854,
+      "learning_rate": 4.080156544426729e-05,
+      "loss": 0.7631,
+      "step": 6350
+    },
+    {
+      "epoch": 0.9276370619994927,
+      "grad_norm": 2.767054796218872,
+      "learning_rate": 4.0729091172633715e-05,
+      "loss": 0.764,
+      "step": 6400
+    },
+    {
+      "epoch": 0.9348842265463637,
+      "grad_norm": 2.9610891342163086,
+      "learning_rate": 4.065661690100015e-05,
+      "loss": 0.7536,
+      "step": 6450
+    },
+    {
+      "epoch": 0.9421313910932347,
+      "grad_norm": 3.0401785373687744,
+      "learning_rate": 4.058414262936658e-05,
+      "loss": 0.7824,
+      "step": 6500
+    },
+    {
+      "epoch": 0.9493785556401058,
+      "grad_norm": 2.3925185203552246,
+      "learning_rate": 4.051166835773301e-05,
+      "loss": 0.7277,
+      "step": 6550
+    },
+    {
+      "epoch": 0.9566257201869769,
+      "grad_norm": 2.707064390182495,
+      "learning_rate": 4.0439194086099436e-05,
+      "loss": 0.7133,
+      "step": 6600
+    },
+    {
+      "epoch": 0.9638728847338479,
+      "grad_norm": 2.639535665512085,
+      "learning_rate": 4.0366719814465865e-05,
+      "loss": 0.7198,
+      "step": 6650
+    },
+    {
+      "epoch": 0.9711200492807189,
+      "grad_norm": 2.1093649864196777,
+      "learning_rate": 4.02942455428323e-05,
+      "loss": 0.7417,
+      "step": 6700
+    },
+    {
+      "epoch": 0.97836721382759,
+      "grad_norm": 4.515178203582764,
+      "learning_rate": 4.022177127119873e-05,
+      "loss": 0.7299,
+      "step": 6750
+    },
+    {
+      "epoch": 0.985614378374461,
+      "grad_norm": 2.626970052719116,
+      "learning_rate": 4.014929699956516e-05,
+      "loss": 0.741,
+      "step": 6800
+    },
+    {
+      "epoch": 0.992861542921332,
+      "grad_norm": 3.5896389484405518,
+      "learning_rate": 4.0076822727931586e-05,
+      "loss": 0.7568,
+      "step": 6850
+    },
+    {
+      "epoch": 1.0001087074682031,
+      "grad_norm": 5.62574577331543,
+      "learning_rate": 4.000434845629802e-05,
+      "loss": 0.7374,
+      "step": 6900
+    },
+    {
+      "epoch": 1.0073558720150741,
+      "grad_norm": 3.0508015155792236,
+      "learning_rate": 3.993187418466445e-05,
+      "loss": 0.5983,
+      "step": 6950
+    },
+    {
+      "epoch": 1.0146030365619452,
+      "grad_norm": 1.7448738813400269,
+      "learning_rate": 3.985939991303088e-05,
+      "loss": 0.5915,
+      "step": 7000
+    },
+    {
+      "epoch": 1.0146030365619452,
+      "eval_loss": 0.696858823299408,
+      "eval_runtime": 245.4484,
+      "eval_samples_per_second": 140.539,
+      "eval_steps_per_second": 14.056,
+      "step": 7000
+    },
+    {
+      "epoch": 1.0218502011088162,
+      "grad_norm": 3.0310137271881104,
+      "learning_rate": 3.978692564139731e-05,
+      "loss": 0.5752,
+      "step": 7050
+    },
+    {
+      "epoch": 1.0290973656556872,
+      "grad_norm": 2.422286033630371,
+      "learning_rate": 3.9714451369763736e-05,
+      "loss": 0.6085,
+      "step": 7100
+    },
+    {
+      "epoch": 1.0363445302025582,
+      "grad_norm": 2.385861396789551,
+      "learning_rate": 3.9641977098130164e-05,
+      "loss": 0.5922,
+      "step": 7150
+    },
+    {
+      "epoch": 1.0435916947494293,
+      "grad_norm": 2.8933980464935303,
+      "learning_rate": 3.956950282649659e-05,
+      "loss": 0.6146,
+      "step": 7200
+    },
+    {
+      "epoch": 1.0508388592963003,
+      "grad_norm": 2.2987074851989746,
+      "learning_rate": 3.949702855486303e-05,
+      "loss": 0.6061,
+      "step": 7250
+    },
+    {
+      "epoch": 1.0580860238431713,
+      "grad_norm": 2.174790859222412,
+      "learning_rate": 3.942455428322946e-05,
+      "loss": 0.6033,
+      "step": 7300
+    },
+    {
+      "epoch": 1.0653331883900423,
+      "grad_norm": 2.604837417602539,
+      "learning_rate": 3.9352080011595885e-05,
+      "loss": 0.6027,
+      "step": 7350
+    },
+    {
+      "epoch": 1.0725803529369133,
+      "grad_norm": 3.116417646408081,
+      "learning_rate": 3.9279605739962314e-05,
+      "loss": 0.5783,
+      "step": 7400
+    },
+    {
+      "epoch": 1.0798275174837846,
+      "grad_norm": 2.4891812801361084,
+      "learning_rate": 3.920713146832874e-05,
+      "loss": 0.6108,
+      "step": 7450
+    },
+    {
+      "epoch": 1.0870746820306556,
+      "grad_norm": 2.293124198913574,
+      "learning_rate": 3.913465719669517e-05,
+      "loss": 0.5853,
+      "step": 7500
+    },
+    {
+      "epoch": 1.0943218465775266,
+      "grad_norm": 2.6642067432403564,
+      "learning_rate": 3.90621829250616e-05,
+      "loss": 0.569,
+      "step": 7550
+    },
+    {
+      "epoch": 1.1015690111243976,
+      "grad_norm": 2.5504302978515625,
+      "learning_rate": 3.8989708653428035e-05,
+      "loss": 0.6019,
+      "step": 7600
+    },
+    {
+      "epoch": 1.1088161756712687,
+      "grad_norm": 2.672874927520752,
+      "learning_rate": 3.8917234381794463e-05,
+      "loss": 0.5799,
+      "step": 7650
+    },
+    {
+      "epoch": 1.1160633402181397,
+      "grad_norm": 2.6319384574890137,
+      "learning_rate": 3.884476011016089e-05,
+      "loss": 0.6357,
+      "step": 7700
+    },
+    {
+      "epoch": 1.1233105047650107,
+      "grad_norm": 3.3530545234680176,
+      "learning_rate": 3.877228583852732e-05,
+      "loss": 0.5737,
+      "step": 7750
+    },
+    {
+      "epoch": 1.1305576693118817,
+      "grad_norm": 2.6204335689544678,
+      "learning_rate": 3.8699811566893756e-05,
+      "loss": 0.6139,
+      "step": 7800
+    },
+    {
+      "epoch": 1.1378048338587528,
+      "grad_norm": 2.570143938064575,
+      "learning_rate": 3.8627337295260185e-05,
+      "loss": 0.6038,
+      "step": 7850
+    },
+    {
+      "epoch": 1.1450519984056238,
+      "grad_norm": 2.826798439025879,
+      "learning_rate": 3.855486302362661e-05,
+      "loss": 0.5985,
+      "step": 7900
+    },
+    {
+      "epoch": 1.1522991629524948,
+      "grad_norm": 1.774380087852478,
+      "learning_rate": 3.848238875199304e-05,
+      "loss": 0.6042,
+      "step": 7950
+    },
+    {
+      "epoch": 1.1595463274993658,
+      "grad_norm": 2.9775238037109375,
+      "learning_rate": 3.840991448035948e-05,
+      "loss": 0.5887,
+      "step": 8000
+    },
+    {
+      "epoch": 1.1595463274993658,
+      "eval_loss": 0.69095379114151,
+      "eval_runtime": 245.6192,
+      "eval_samples_per_second": 140.441,
+      "eval_steps_per_second": 14.046,
+      "step": 8000
+    },
+    {
+      "epoch": 1.1667934920462368,
+      "grad_norm": 2.4235680103302,
+      "learning_rate": 3.8337440208725906e-05,
+      "loss": 0.6397,
+      "step": 8050
+    },
+    {
+      "epoch": 1.1740406565931079,
+      "grad_norm": 2.372230291366577,
+      "learning_rate": 3.8264965937092334e-05,
+      "loss": 0.6083,
+      "step": 8100
+    },
+    {
+      "epoch": 1.1812878211399789,
+      "grad_norm": 2.4329166412353516,
+      "learning_rate": 3.819249166545876e-05,
+      "loss": 0.5956,
+      "step": 8150
+    },
+    {
+      "epoch": 1.1885349856868501,
+      "grad_norm": 3.0749635696411133,
+      "learning_rate": 3.812001739382519e-05,
+      "loss": 0.6162,
+      "step": 8200
+    },
+    {
+      "epoch": 1.1957821502337211,
+      "grad_norm": 2.3049559593200684,
+      "learning_rate": 3.804754312219163e-05,
+      "loss": 0.593,
+      "step": 8250
+    },
+    {
+      "epoch": 1.2030293147805922,
+      "grad_norm": 2.3846399784088135,
+      "learning_rate": 3.7975068850558055e-05,
+      "loss": 0.6,
+      "step": 8300
+    },
+    {
+      "epoch": 1.2102764793274632,
+      "grad_norm": 2.4246721267700195,
+      "learning_rate": 3.7902594578924484e-05,
+      "loss": 0.6299,
+      "step": 8350
+    },
+    {
+      "epoch": 1.2175236438743342,
+      "grad_norm": 3.0294981002807617,
+      "learning_rate": 3.783012030729091e-05,
+      "loss": 0.5957,
+      "step": 8400
+    },
+    {
+      "epoch": 1.2247708084212052,
+      "grad_norm": 2.184633255004883,
+      "learning_rate": 3.775764603565735e-05,
+      "loss": 0.5781,
+      "step": 8450
+    },
+    {
+      "epoch": 1.2320179729680762,
+      "grad_norm": 2.423351287841797,
+      "learning_rate": 3.7685171764023776e-05,
+      "loss": 0.571,
+      "step": 8500
+    },
+    {
+      "epoch": 1.2392651375149473,
+      "grad_norm": 3.334449529647827,
+      "learning_rate": 3.7612697492390205e-05,
+      "loss": 0.579,
+      "step": 8550
+    },
+    {
+      "epoch": 1.2465123020618183,
+      "grad_norm": 2.2588014602661133,
+      "learning_rate": 3.7540223220756633e-05,
+      "loss": 0.5874,
+      "step": 8600
+    },
+    {
+      "epoch": 1.2537594666086893,
+      "grad_norm": 3.5975542068481445,
+      "learning_rate": 3.746774894912306e-05,
+      "loss": 0.5633,
+      "step": 8650
+    },
+    {
+      "epoch": 1.2610066311555603,
+      "grad_norm": 2.7754578590393066,
+      "learning_rate": 3.73952746774895e-05,
+      "loss": 0.6276,
+      "step": 8700
+    },
+    {
+      "epoch": 1.2682537957024314,
+      "grad_norm": 2.410757064819336,
+      "learning_rate": 3.7322800405855926e-05,
+      "loss": 0.6118,
+      "step": 8750
+    },
+    {
+      "epoch": 1.2755009602493024,
+      "grad_norm": 2.5323429107666016,
+      "learning_rate": 3.7250326134222355e-05,
+      "loss": 0.5945,
+      "step": 8800
+    },
+    {
+      "epoch": 1.2827481247961736,
+      "grad_norm": 2.5905425548553467,
+      "learning_rate": 3.717785186258878e-05,
+      "loss": 0.5903,
+      "step": 8850
+    },
+    {
+      "epoch": 1.2899952893430444,
+      "grad_norm": 2.042097568511963,
+      "learning_rate": 3.710537759095521e-05,
+      "loss": 0.5997,
+      "step": 8900
+    },
+    {
+      "epoch": 1.2972424538899157,
+      "grad_norm": 2.1564393043518066,
+      "learning_rate": 3.703290331932164e-05,
+      "loss": 0.5697,
+      "step": 8950
+    },
+    {
+      "epoch": 1.3044896184367867,
+      "grad_norm": 2.8595190048217773,
+      "learning_rate": 3.696042904768807e-05,
+      "loss": 0.6026,
+      "step": 9000
+    },
+    {
+      "epoch": 1.3044896184367867,
+      "eval_loss": 0.6820585131645203,
+      "eval_runtime": 245.5719,
+      "eval_samples_per_second": 140.468,
+      "eval_steps_per_second": 14.049,
+      "step": 9000
+    },
+    {
+      "epoch": 1.3117367829836577,
+      "grad_norm": 2.401057243347168,
+      "learning_rate": 3.6887954776054504e-05,
+      "loss": 0.615,
+      "step": 9050
+    },
+    {
+      "epoch": 1.3189839475305287,
+      "grad_norm": 2.1161983013153076,
+      "learning_rate": 3.681548050442093e-05,
+      "loss": 0.6468,
+      "step": 9100
+    },
+    {
+      "epoch": 1.3262311120773997,
+      "grad_norm": 2.9642062187194824,
+      "learning_rate": 3.674300623278736e-05,
+      "loss": 0.5908,
+      "step": 9150
+    },
+    {
+      "epoch": 1.3334782766242708,
+      "grad_norm": 2.200223445892334,
+      "learning_rate": 3.667053196115379e-05,
+      "loss": 0.5493,
+      "step": 9200
+    },
+    {
+      "epoch": 1.3407254411711418,
+      "grad_norm": 2.8368406295776367,
+      "learning_rate": 3.659805768952022e-05,
+      "loss": 0.6157,
+      "step": 9250
+    },
+    {
+      "epoch": 1.3479726057180128,
+      "grad_norm": 2.500457286834717,
+      "learning_rate": 3.652558341788665e-05,
+      "loss": 0.6217,
+      "step": 9300
+    },
+    {
+      "epoch": 1.3552197702648838,
+      "grad_norm": 2.435392379760742,
+      "learning_rate": 3.645310914625308e-05,
+      "loss": 0.5988,
+      "step": 9350
+    },
+    {
+      "epoch": 1.3624669348117548,
+      "grad_norm": 2.5148136615753174,
+      "learning_rate": 3.638063487461951e-05,
+      "loss": 0.6043,
+      "step": 9400
+    },
+    {
+      "epoch": 1.3697140993586259,
+      "grad_norm": 2.5917887687683105,
+      "learning_rate": 3.630816060298594e-05,
+      "loss": 0.6048,
+      "step": 9450
+    },
+    {
+      "epoch": 1.3769612639054971,
+      "grad_norm": 3.429553747177124,
+      "learning_rate": 3.623568633135237e-05,
+      "loss": 0.5958,
+      "step": 9500
+    },
+    {
+      "epoch": 1.384208428452368,
+      "grad_norm": 3.326967239379883,
+      "learning_rate": 3.6163212059718803e-05,
+      "loss": 0.5629,
+      "step": 9550
+    },
+    {
+      "epoch": 1.3914555929992392,
+      "grad_norm": 2.0327517986297607,
+      "learning_rate": 3.609073778808523e-05,
+      "loss": 0.6238,
+      "step": 9600
+    },
+    {
+      "epoch": 1.39870275754611,
+      "grad_norm": 2.4792556762695312,
+      "learning_rate": 3.601826351645166e-05,
+      "loss": 0.5662,
+      "step": 9650
+    },
+    {
+      "epoch": 1.4059499220929812,
+      "grad_norm": 2.692080497741699,
+      "learning_rate": 3.594578924481809e-05,
+      "loss": 0.6257,
+      "step": 9700
+    },
+    {
+      "epoch": 1.4131970866398522,
+      "grad_norm": 3.3779165744781494,
+      "learning_rate": 3.587331497318452e-05,
+      "loss": 0.5654,
+      "step": 9750
+    },
+    {
+      "epoch": 1.4204442511867232,
+      "grad_norm": 2.2963967323303223,
+      "learning_rate": 3.580084070155095e-05,
+      "loss": 0.6,
+      "step": 9800
+    },
+    {
+      "epoch": 1.4276914157335943,
+      "grad_norm": 2.7266321182250977,
+      "learning_rate": 3.572836642991738e-05,
+      "loss": 0.6333,
+      "step": 9850
+    },
+    {
+      "epoch": 1.4349385802804653,
+      "grad_norm": 1.9801080226898193,
+      "learning_rate": 3.565589215828381e-05,
+      "loss": 0.583,
+      "step": 9900
+    },
+    {
+      "epoch": 1.4421857448273363,
+      "grad_norm": 2.2797017097473145,
+      "learning_rate": 3.558341788665024e-05,
+      "loss": 0.5918,
+      "step": 9950
+    },
+    {
+      "epoch": 1.4494329093742073,
+      "grad_norm": 2.325143575668335,
+      "learning_rate": 3.5510943615016674e-05,
+      "loss": 0.601,
+      "step": 10000
+    },
+    {
+      "epoch": 1.4494329093742073,
+      "eval_loss": 0.6719304919242859,
+      "eval_runtime": 245.6786,
+      "eval_samples_per_second": 140.407,
+      "eval_steps_per_second": 14.043,
+      "step": 10000
+    },
+    {
+      "epoch": 1.4566800739210783,
+      "grad_norm": 2.4414122104644775,
+      "learning_rate": 3.54384693433831e-05,
+      "loss": 0.5557,
+      "step": 10050
+    },
+    {
+      "epoch": 1.4639272384679494,
+      "grad_norm": 3.4254276752471924,
+      "learning_rate": 3.536599507174953e-05,
+      "loss": 0.5961,
+      "step": 10100
+    },
+    {
+      "epoch": 1.4711744030148204,
+      "grad_norm": 1.9034006595611572,
+      "learning_rate": 3.529352080011596e-05,
+      "loss": 0.5716,
+      "step": 10150
+    },
+    {
+      "epoch": 1.4784215675616914,
+      "grad_norm": 1.9632625579833984,
+      "learning_rate": 3.5221046528482395e-05,
+      "loss": 0.6656,
+      "step": 10200
+    },
+    {
+      "epoch": 1.4856687321085627,
+      "grad_norm": 2.634131669998169,
+      "learning_rate": 3.5148572256848824e-05,
+      "loss": 0.605,
+      "step": 10250
+    },
+    {
+      "epoch": 1.4929158966554334,
+      "grad_norm": 2.171637535095215,
+      "learning_rate": 3.507609798521525e-05,
+      "loss": 0.5942,
+      "step": 10300
+    },
+    {
+      "epoch": 1.5001630612023047,
+      "grad_norm": 2.2301058769226074,
+      "learning_rate": 3.500362371358168e-05,
+      "loss": 0.5776,
+      "step": 10350
+    },
+    {
+      "epoch": 1.5074102257491755,
+      "grad_norm": 2.1996707916259766,
+      "learning_rate": 3.493114944194811e-05,
+      "loss": 0.6082,
+      "step": 10400
+    },
+    {
+      "epoch": 1.5146573902960467,
+      "grad_norm": 2.6578609943389893,
+      "learning_rate": 3.4858675170314545e-05,
+      "loss": 0.637,
+      "step": 10450
+    },
+    {
+      "epoch": 1.5219045548429178,
+      "grad_norm": 2.902642011642456,
+      "learning_rate": 3.4786200898680973e-05,
+      "loss": 0.5898,
+      "step": 10500
+    },
+    {
+      "epoch": 1.5291517193897888,
+      "grad_norm": 2.331738233566284,
+      "learning_rate": 3.47137266270474e-05,
+      "loss": 0.6314,
+      "step": 10550
+    },
+    {
+      "epoch": 1.5363988839366598,
+      "grad_norm": 2.1168956756591797,
+      "learning_rate": 3.464125235541383e-05,
+      "loss": 0.6306,
+      "step": 10600
+    },
+    {
+      "epoch": 1.5436460484835308,
+      "grad_norm": 3.4085164070129395,
+      "learning_rate": 3.456877808378026e-05,
+      "loss": 0.6065,
+      "step": 10650
+    },
+    {
+      "epoch": 1.5508932130304018,
+      "grad_norm": 3.533252000808716,
+      "learning_rate": 3.449630381214669e-05,
+      "loss": 0.5818,
+      "step": 10700
+    },
+    {
+      "epoch": 1.5581403775772729,
+      "grad_norm": 2.3345284461975098,
+      "learning_rate": 3.4423829540513116e-05,
+      "loss": 0.5762,
+      "step": 10750
+    },
+    {
+      "epoch": 1.5653875421241439,
+      "grad_norm": 2.5388126373291016,
+      "learning_rate": 3.4351355268879545e-05,
+      "loss": 0.5967,
+      "step": 10800
+    },
+    {
+      "epoch": 1.572634706671015,
+      "grad_norm": 2.583822250366211,
+      "learning_rate": 3.4278880997245973e-05,
+      "loss": 0.6388,
+      "step": 10850
+    },
+    {
+      "epoch": 1.5798818712178861,
+      "grad_norm": 2.8442914485931396,
+      "learning_rate": 3.420640672561241e-05,
+      "loss": 0.6049,
+      "step": 10900
+    },
+    {
+      "epoch": 1.587129035764757,
+      "grad_norm": 2.080157518386841,
+      "learning_rate": 3.413393245397884e-05,
+      "loss": 0.5681,
+      "step": 10950
+    },
+    {
+      "epoch": 1.5943762003116282,
+      "grad_norm": 3.178879976272583,
+      "learning_rate": 3.4061458182345266e-05,
+      "loss": 0.5877,
+      "step": 11000
+    },
+    {
+      "epoch": 1.5943762003116282,
+      "eval_loss": 0.6599454283714294,
+      "eval_runtime": 245.7578,
+      "eval_samples_per_second": 140.362,
+      "eval_steps_per_second": 14.038,
+      "step": 11000
+    },
+    {
+      "epoch": 1.601623364858499,
+      "grad_norm": 2.6149051189422607,
+      "learning_rate": 3.3988983910711694e-05,
+      "loss": 0.5706,
+      "step": 11050
+    },
+    {
+      "epoch": 1.6088705294053702,
+      "grad_norm": 2.2884621620178223,
+      "learning_rate": 3.391650963907813e-05,
+      "loss": 0.5929,
+      "step": 11100
+    },
+    {
+      "epoch": 1.616117693952241,
+      "grad_norm": 2.5482795238494873,
+      "learning_rate": 3.384403536744456e-05,
+      "loss": 0.5772,
+      "step": 11150
+    },
+    {
+      "epoch": 1.6233648584991123,
+      "grad_norm": 2.499694347381592,
+      "learning_rate": 3.377156109581099e-05,
+      "loss": 0.589,
+      "step": 11200
+    },
+    {
+      "epoch": 1.6306120230459833,
+      "grad_norm": 3.6115667819976807,
+      "learning_rate": 3.3699086824177416e-05,
+      "loss": 0.5824,
+      "step": 11250
+    },
+    {
+      "epoch": 1.6378591875928543,
+      "grad_norm": 3.2503881454467773,
+      "learning_rate": 3.3626612552543844e-05,
+      "loss": 0.5768,
+      "step": 11300
+    },
+    {
+      "epoch": 1.6451063521397253,
+      "grad_norm": 2.821199893951416,
+      "learning_rate": 3.355413828091028e-05,
+      "loss": 0.5685,
+      "step": 11350
+    },
+    {
+      "epoch": 1.6523535166865964,
+      "grad_norm": 2.602804660797119,
+      "learning_rate": 3.348166400927671e-05,
+      "loss": 0.595,
+      "step": 11400
+    },
+    {
+      "epoch": 1.6596006812334674,
+      "grad_norm": 3.2898738384246826,
+      "learning_rate": 3.340918973764314e-05,
+      "loss": 0.5737,
+      "step": 11450
+    },
+    {
+      "epoch": 1.6668478457803384,
+      "grad_norm": 3.5453619956970215,
+      "learning_rate": 3.3336715466009565e-05,
+      "loss": 0.6154,
+      "step": 11500
+    },
+    {
+      "epoch": 1.6740950103272096,
+      "grad_norm": 2.276144027709961,
+      "learning_rate": 3.3264241194376e-05,
+      "loss": 0.5962,
+      "step": 11550
+    },
+    {
+      "epoch": 1.6813421748740804,
+      "grad_norm": 2.5161609649658203,
+      "learning_rate": 3.319176692274243e-05,
+      "loss": 0.5486,
+      "step": 11600
+    },
+    {
+      "epoch": 1.6885893394209517,
+      "grad_norm": 2.3692619800567627,
+      "learning_rate": 3.311929265110886e-05,
+      "loss": 0.5959,
+      "step": 11650
+    },
+    {
+      "epoch": 1.6958365039678225,
+      "grad_norm": 2.2731165885925293,
+      "learning_rate": 3.3046818379475286e-05,
+      "loss": 0.5484,
+      "step": 11700
+    },
+    {
+      "epoch": 1.7030836685146937,
+      "grad_norm": 2.9670755863189697,
+      "learning_rate": 3.297434410784172e-05,
+      "loss": 0.5648,
+      "step": 11750
+    },
+    {
+      "epoch": 1.7103308330615645,
+      "grad_norm": 2.7299599647521973,
+      "learning_rate": 3.290186983620815e-05,
+      "loss": 0.6107,
+      "step": 11800
+    },
+    {
+      "epoch": 1.7175779976084358,
+      "grad_norm": 2.5043752193450928,
+      "learning_rate": 3.282939556457458e-05,
+      "loss": 0.5998,
+      "step": 11850
+    },
+    {
+      "epoch": 1.7248251621553068,
+      "grad_norm": 2.849438428878784,
+      "learning_rate": 3.275692129294101e-05,
+      "loss": 0.5729,
+      "step": 11900
+    },
+    {
+      "epoch": 1.7320723267021778,
+      "grad_norm": 2.4421026706695557,
+      "learning_rate": 3.2684447021307436e-05,
+      "loss": 0.5821,
+      "step": 11950
+    },
+    {
+      "epoch": 1.7393194912490488,
+      "grad_norm": 2.8539223670959473,
+      "learning_rate": 3.261197274967387e-05,
+      "loss": 0.566,
+      "step": 12000
+    },
+    {
+      "epoch": 1.7393194912490488,
+      "eval_loss": 0.6502553820610046,
+      "eval_runtime": 247.5286,
+      "eval_samples_per_second": 139.358,
+      "eval_steps_per_second": 13.938,
+      "step": 12000
+    },
+    {
+      "epoch": 1.7465666557959199,
+      "grad_norm": 2.71627140045166,
+      "learning_rate": 3.25394984780403e-05,
+      "loss": 0.6029,
+      "step": 12050
+    },
+    {
+      "epoch": 1.7538138203427909,
+      "grad_norm": 2.3385589122772217,
+      "learning_rate": 3.246702420640673e-05,
+      "loss": 0.5514,
+      "step": 12100
+    },
+    {
+      "epoch": 1.761060984889662,
+      "grad_norm": 2.6175739765167236,
+      "learning_rate": 3.239454993477316e-05,
+      "loss": 0.6304,
+      "step": 12150
+    },
+    {
+      "epoch": 1.768308149436533,
+      "grad_norm": 2.545201301574707,
+      "learning_rate": 3.232207566313959e-05,
+      "loss": 0.5833,
+      "step": 12200
+    },
+    {
+      "epoch": 1.775555313983404,
+      "grad_norm": 3.0050761699676514,
+      "learning_rate": 3.225105087693869e-05,
+      "loss": 0.5914,
+      "step": 12250
+    },
+    {
+      "epoch": 1.7828024785302752,
+      "grad_norm": 2.7314724922180176,
+      "learning_rate": 3.217857660530512e-05,
+      "loss": 0.5652,
+      "step": 12300
+    },
+    {
+      "epoch": 1.790049643077146,
+      "grad_norm": 2.4256742000579834,
+      "learning_rate": 3.210610233367155e-05,
+      "loss": 0.5891,
+      "step": 12350
+    },
+    {
+      "epoch": 1.7972968076240172,
+      "grad_norm": 3.057068109512329,
+      "learning_rate": 3.203362806203798e-05,
+      "loss": 0.5873,
+      "step": 12400
+    },
+    {
+      "epoch": 1.804543972170888,
+      "grad_norm": 2.393101215362549,
+      "learning_rate": 3.196115379040441e-05,
+      "loss": 0.5735,
+      "step": 12450
+    },
+    {
+      "epoch": 1.8117911367177593,
+      "grad_norm": 2.644343614578247,
+      "learning_rate": 3.188867951877084e-05,
+      "loss": 0.569,
+      "step": 12500
+    },
+    {
+      "epoch": 1.81903830126463,
+      "grad_norm": 2.401324987411499,
+      "learning_rate": 3.181620524713727e-05,
+      "loss": 0.5358,
+      "step": 12550
+    },
+    {
+      "epoch": 1.8262854658115013,
+      "grad_norm": 2.5138838291168213,
+      "learning_rate": 3.17437309755037e-05,
+      "loss": 0.5348,
+      "step": 12600
+    },
+    {
+      "epoch": 1.8335326303583723,
+      "grad_norm": 1.8828959465026855,
+      "learning_rate": 3.167125670387013e-05,
+      "loss": 0.5899,
+      "step": 12650
+    },
+    {
+      "epoch": 1.8407797949052433,
+      "grad_norm": 2.3968441486358643,
+      "learning_rate": 3.159878243223656e-05,
+      "loss": 0.5385,
+      "step": 12700
+    },
+    {
+      "epoch": 1.8480269594521144,
+      "grad_norm": 2.0079128742218018,
+      "learning_rate": 3.152630816060299e-05,
+      "loss": 0.5341,
+      "step": 12750
+    },
+    {
+      "epoch": 1.8552741239989854,
+      "grad_norm": 1.722406029701233,
+      "learning_rate": 3.145383388896942e-05,
+      "loss": 0.5592,
+      "step": 12800
+    },
+    {
+      "epoch": 1.8625212885458564,
+      "grad_norm": 2.9170126914978027,
+      "learning_rate": 3.1381359617335845e-05,
+      "loss": 0.5637,
+      "step": 12850
+    },
+    {
+      "epoch": 1.8697684530927274,
+      "grad_norm": 3.159693479537964,
+      "learning_rate": 3.1308885345702274e-05,
+      "loss": 0.6119,
+      "step": 12900
+    },
+    {
+      "epoch": 1.8770156176395985,
+      "grad_norm": 1.9590739011764526,
+      "learning_rate": 3.12364110740687e-05,
+      "loss": 0.6167,
+      "step": 12950
+    },
+    {
+      "epoch": 1.8842627821864695,
+      "grad_norm": 2.080077886581421,
+      "learning_rate": 3.116393680243514e-05,
+      "loss": 0.5713,
+      "step": 13000
+    },
+    {
+      "epoch": 1.8842627821864695,
+      "eval_loss": 0.6442924737930298,
+      "eval_runtime": 246.2742,
+      "eval_samples_per_second": 140.067,
+      "eval_steps_per_second": 14.009,
+      "step": 13000
+    },
+    {
+      "epoch": 1.8915099467333407,
+      "grad_norm": 3.022768974304199,
+      "learning_rate": 3.1091462530801567e-05,
+      "loss": 0.5889,
+      "step": 13050
+    },
+    {
+      "epoch": 1.8987571112802115,
+      "grad_norm": 2.6845195293426514,
+      "learning_rate": 3.1018988259167995e-05,
+      "loss": 0.5536,
+      "step": 13100
+    },
+    {
+      "epoch": 1.9060042758270828,
+      "grad_norm": 3.178252696990967,
+      "learning_rate": 3.0946513987534424e-05,
+      "loss": 0.6397,
+      "step": 13150
+    },
+    {
+      "epoch": 1.9132514403739536,
+      "grad_norm": 3.4886860847473145,
+      "learning_rate": 3.087403971590085e-05,
+      "loss": 0.5945,
+      "step": 13200
+    },
+    {
+      "epoch": 1.9204986049208248,
+      "grad_norm": 2.4418983459472656,
+      "learning_rate": 3.080156544426729e-05,
+      "loss": 0.6039,
+      "step": 13250
+    },
+    {
+      "epoch": 1.9277457694676956,
+      "grad_norm": 2.7599799633026123,
+      "learning_rate": 3.0729091172633716e-05,
+      "loss": 0.5496,
+      "step": 13300
+    },
+    {
+      "epoch": 1.9349929340145668,
+      "grad_norm": 2.276477098464966,
+      "learning_rate": 3.0656616901000145e-05,
+      "loss": 0.5836,
+      "step": 13350
+    },
+    {
+      "epoch": 1.9422400985614379,
+      "grad_norm": 2.3959388732910156,
+      "learning_rate": 3.058414262936657e-05,
+      "loss": 0.603,
+      "step": 13400
+    },
+    {
+      "epoch": 1.9494872631083089,
+      "grad_norm": 2.58145809173584,
+      "learning_rate": 3.051166835773301e-05,
+      "loss": 0.5671,
+      "step": 13450
+    },
+    {
+      "epoch": 1.95673442765518,
+      "grad_norm": 1.9884346723556519,
+      "learning_rate": 3.0439194086099437e-05,
+      "loss": 0.5766,
+      "step": 13500
+    },
+    {
+      "epoch": 1.963981592202051,
+      "grad_norm": 2.8933162689208984,
+      "learning_rate": 3.0366719814465866e-05,
+      "loss": 0.5529,
+      "step": 13550
+    },
+    {
+      "epoch": 1.971228756748922,
+      "grad_norm": 3.837768316268921,
+      "learning_rate": 3.0294245542832294e-05,
+      "loss": 0.5698,
+      "step": 13600
+    },
+    {
+      "epoch": 1.978475921295793,
+      "grad_norm": 5.475804805755615,
+      "learning_rate": 3.0221771271198723e-05,
+      "loss": 0.5887,
+      "step": 13650
+    },
+    {
+      "epoch": 1.9857230858426642,
+      "grad_norm": 2.9813613891601562,
+      "learning_rate": 3.014929699956516e-05,
+      "loss": 0.5685,
+      "step": 13700
+    },
+    {
+      "epoch": 1.992970250389535,
+      "grad_norm": 2.6293349266052246,
+      "learning_rate": 3.0076822727931587e-05,
+      "loss": 0.5773,
+      "step": 13750
+    },
+    {
+      "epoch": 2.0002174149364063,
+      "grad_norm": 2.593838691711426,
+      "learning_rate": 3.0004348456298015e-05,
+      "loss": 0.5628,
+      "step": 13800
+    },
+    {
+      "epoch": 2.007464579483277,
+      "grad_norm": 1.6289088726043701,
+      "learning_rate": 2.9931874184664444e-05,
+      "loss": 0.4418,
+      "step": 13850
+    },
+    {
+      "epoch": 2.0147117440301483,
+      "grad_norm": 2.8888840675354004,
+      "learning_rate": 2.985939991303088e-05,
+      "loss": 0.4652,
+      "step": 13900
+    },
+    {
+      "epoch": 2.021958908577019,
+      "grad_norm": 2.572951316833496,
+      "learning_rate": 2.9786925641397308e-05,
+      "loss": 0.4685,
+      "step": 13950
+    },
+    {
+      "epoch": 2.0292060731238903,
+      "grad_norm": 3.1881916522979736,
+      "learning_rate": 2.9714451369763737e-05,
+      "loss": 0.4583,
+      "step": 14000
+    },
+    {
+      "epoch": 2.0292060731238903,
+      "eval_loss": 0.6516901254653931,
+      "eval_runtime": 247.3517,
+      "eval_samples_per_second": 139.457,
+      "eval_steps_per_second": 13.948,
+      "step": 14000
+    },
+    {
+      "epoch": 2.036453237670761,
+      "grad_norm": 2.9465830326080322,
+      "learning_rate": 2.9641977098130165e-05,
+      "loss": 0.4494,
+      "step": 14050
+    },
+    {
+      "epoch": 2.0437004022176324,
+      "grad_norm": 1.7231141328811646,
+      "learning_rate": 2.9569502826496597e-05,
+      "loss": 0.4183,
+      "step": 14100
+    },
+    {
+      "epoch": 2.050947566764503,
+      "grad_norm": 3.4768612384796143,
+      "learning_rate": 2.9497028554863026e-05,
+      "loss": 0.4223,
+      "step": 14150
+    },
+    {
+      "epoch": 2.0581947313113744,
+      "grad_norm": 2.902104139328003,
+      "learning_rate": 2.9424554283229454e-05,
+      "loss": 0.4569,
+      "step": 14200
+    },
+    {
+      "epoch": 2.0654418958582457,
+      "grad_norm": 2.661472797393799,
+      "learning_rate": 2.9352080011595883e-05,
+      "loss": 0.4551,
+      "step": 14250
+    },
+    {
+      "epoch": 2.0726890604051165,
+      "grad_norm": 2.9054114818573,
+      "learning_rate": 2.927960573996231e-05,
+      "loss": 0.4279,
+      "step": 14300
+    },
+    {
+      "epoch": 2.0799362249519877,
+      "grad_norm": 2.9954373836517334,
+      "learning_rate": 2.9207131468328747e-05,
+      "loss": 0.4291,
+      "step": 14350
+    },
+    {
+      "epoch": 2.0871833894988585,
+      "grad_norm": 2.9396731853485107,
+      "learning_rate": 2.9134657196695175e-05,
+      "loss": 0.4392,
+      "step": 14400
+    },
+    {
+      "epoch": 2.0944305540457298,
+      "grad_norm": 2.4928243160247803,
+      "learning_rate": 2.9062182925061604e-05,
+      "loss": 0.402,
+      "step": 14450
+    },
+    {
+      "epoch": 2.1016777185926006,
+      "grad_norm": 3.2848997116088867,
+      "learning_rate": 2.8989708653428032e-05,
+      "loss": 0.4189,
+      "step": 14500
+    },
+    {
+      "epoch": 2.108924883139472,
+      "grad_norm": 3.1870994567871094,
+      "learning_rate": 2.8917234381794468e-05,
+      "loss": 0.4421,
+      "step": 14550
+    },
+    {
+      "epoch": 2.1161720476863426,
+      "grad_norm": 2.7032647132873535,
+      "learning_rate": 2.8844760110160896e-05,
+      "loss": 0.4415,
+      "step": 14600
+    },
+    {
+      "epoch": 2.123419212233214,
+      "grad_norm": 3.0945403575897217,
+      "learning_rate": 2.8772285838527325e-05,
+      "loss": 0.4515,
+      "step": 14650
+    },
+    {
+      "epoch": 2.1306663767800846,
+      "grad_norm": 2.7250170707702637,
+      "learning_rate": 2.8699811566893753e-05,
+      "loss": 0.4374,
+      "step": 14700
+    },
+    {
+      "epoch": 2.137913541326956,
+      "grad_norm": 3.7444469928741455,
+      "learning_rate": 2.8627337295260182e-05,
+      "loss": 0.4738,
+      "step": 14750
+    },
+    {
+      "epoch": 2.1451607058738267,
+      "grad_norm": 3.042750597000122,
+      "learning_rate": 2.8554863023626614e-05,
+      "loss": 0.444,
+      "step": 14800
+    },
+    {
+      "epoch": 2.152407870420698,
+      "grad_norm": 2.8017637729644775,
+      "learning_rate": 2.8482388751993046e-05,
+      "loss": 0.4425,
+      "step": 14850
+    },
+    {
+      "epoch": 2.159655034967569,
+      "grad_norm": 2.268974781036377,
+      "learning_rate": 2.8409914480359475e-05,
+      "loss": 0.4594,
+      "step": 14900
+    },
+    {
+      "epoch": 2.16690219951444,
+      "grad_norm": 2.553354263305664,
+      "learning_rate": 2.8337440208725903e-05,
+      "loss": 0.451,
+      "step": 14950
+    },
+    {
+      "epoch": 2.174149364061311,
+      "grad_norm": 3.928313970565796,
+      "learning_rate": 2.8264965937092335e-05,
+      "loss": 0.4609,
+      "step": 15000
+    },
+    {
+      "epoch": 2.174149364061311,
+      "eval_loss": 0.6510897278785706,
+      "eval_runtime": 245.6483,
+      "eval_samples_per_second": 140.424,
+      "eval_steps_per_second": 14.044,
+      "step": 15000
+    },
+    {
+      "epoch": 2.181396528608182,
+      "grad_norm": 2.9727916717529297,
+      "learning_rate": 2.8192491665458764e-05,
+      "loss": 0.445,
+      "step": 15050
+    },
+    {
+      "epoch": 2.1886436931550532,
+      "grad_norm": 2.9330101013183594,
+      "learning_rate": 2.8120017393825192e-05,
+      "loss": 0.4395,
+      "step": 15100
+    },
+    {
+      "epoch": 2.195890857701924,
+      "grad_norm": 2.5062899589538574,
+      "learning_rate": 2.804754312219162e-05,
+      "loss": 0.4293,
+      "step": 15150
+    },
+    {
+      "epoch": 2.2031380222487953,
+      "grad_norm": 3.488398313522339,
+      "learning_rate": 2.797506885055805e-05,
+      "loss": 0.4382,
+      "step": 15200
+    },
+    {
+      "epoch": 2.210385186795666,
+      "grad_norm": 2.6354928016662598,
+      "learning_rate": 2.7902594578924485e-05,
+      "loss": 0.4348,
+      "step": 15250
+    },
+    {
+      "epoch": 2.2176323513425373,
+      "grad_norm": 2.0676591396331787,
+      "learning_rate": 2.7830120307290913e-05,
+      "loss": 0.4506,
+      "step": 15300
+    },
+    {
+      "epoch": 2.224879515889408,
+      "grad_norm": 3.5719449520111084,
+      "learning_rate": 2.7757646035657342e-05,
+      "loss": 0.4718,
+      "step": 15350
+    },
+    {
+      "epoch": 2.2321266804362794,
+      "grad_norm": 2.755336284637451,
+      "learning_rate": 2.768517176402377e-05,
+      "loss": 0.4934,
+      "step": 15400
+    },
+    {
+      "epoch": 2.23937384498315,
+      "grad_norm": 2.0585694313049316,
+      "learning_rate": 2.7612697492390206e-05,
+      "loss": 0.4199,
+      "step": 15450
+    },
+    {
+      "epoch": 2.2466210095300214,
+      "grad_norm": 3.553250551223755,
+      "learning_rate": 2.7540223220756634e-05,
+      "loss": 0.4414,
+      "step": 15500
+    },
+    {
+      "epoch": 2.253868174076892,
+      "grad_norm": 2.6371939182281494,
+      "learning_rate": 2.7467748949123063e-05,
+      "loss": 0.4238,
+      "step": 15550
+    },
+    {
+      "epoch": 2.2611153386237635,
+      "grad_norm": 2.935482978820801,
+      "learning_rate": 2.739527467748949e-05,
+      "loss": 0.4421,
+      "step": 15600
+    },
+    {
+      "epoch": 2.2683625031706347,
+      "grad_norm": 2.8056271076202393,
+      "learning_rate": 2.7322800405855923e-05,
+      "loss": 0.4838,
+      "step": 15650
+    },
+    {
+      "epoch": 2.2756096677175055,
+      "grad_norm": 1.6445329189300537,
+      "learning_rate": 2.7250326134222352e-05,
+      "loss": 0.4727,
+      "step": 15700
+    },
+    {
+      "epoch": 2.2828568322643767,
+      "grad_norm": 2.901073694229126,
+      "learning_rate": 2.7177851862588784e-05,
+      "loss": 0.451,
+      "step": 15750
+    },
+    {
+      "epoch": 2.2901039968112475,
+      "grad_norm": 3.6521453857421875,
+      "learning_rate": 2.7105377590955213e-05,
+      "loss": 0.4441,
+      "step": 15800
+    },
+    {
+      "epoch": 2.297351161358119,
+      "grad_norm": 3.1472737789154053,
+      "learning_rate": 2.703290331932164e-05,
+      "loss": 0.4436,
+      "step": 15850
+    },
+    {
+      "epoch": 2.3045983259049896,
+      "grad_norm": 2.873993396759033,
+      "learning_rate": 2.6960429047688073e-05,
+      "loss": 0.4555,
+      "step": 15900
+    },
+    {
+      "epoch": 2.311845490451861,
+      "grad_norm": 2.5647237300872803,
+      "learning_rate": 2.68879547760545e-05,
+      "loss": 0.4621,
+      "step": 15950
+    },
+    {
+      "epoch": 2.3190926549987316,
+      "grad_norm": 2.7584807872772217,
+      "learning_rate": 2.681548050442093e-05,
+      "loss": 0.4497,
+      "step": 16000
+    },
+    {
+      "epoch": 2.3190926549987316,
+      "eval_loss": 0.644283652305603,
+      "eval_runtime": 245.6009,
+      "eval_samples_per_second": 140.451,
+      "eval_steps_per_second": 14.047,
+      "step": 16000
+    },
+    {
+      "epoch": 2.326339819545603,
+      "grad_norm": 3.315335750579834,
+      "learning_rate": 2.674300623278736e-05,
+      "loss": 0.4505,
+      "step": 16050
+    },
+    {
+      "epoch": 2.3335869840924737,
+      "grad_norm": 2.8071439266204834,
+      "learning_rate": 2.6670531961153794e-05,
+      "loss": 0.4457,
+      "step": 16100
+    },
+    {
+      "epoch": 2.340834148639345,
+      "grad_norm": 2.4625017642974854,
+      "learning_rate": 2.6598057689520223e-05,
+      "loss": 0.418,
+      "step": 16150
+    },
+    {
+      "epoch": 2.3480813131862157,
+      "grad_norm": 3.5972659587860107,
+      "learning_rate": 2.652558341788665e-05,
+      "loss": 0.4188,
+      "step": 16200
+    },
+    {
+      "epoch": 2.355328477733087,
+      "grad_norm": 2.557579517364502,
+      "learning_rate": 2.645310914625308e-05,
+      "loss": 0.4161,
+      "step": 16250
+    },
+    {
+      "epoch": 2.3625756422799578,
+      "grad_norm": 2.0252676010131836,
+      "learning_rate": 2.638063487461951e-05,
+      "loss": 0.4504,
+      "step": 16300
+    },
+    {
+      "epoch": 2.369822806826829,
+      "grad_norm": 2.324404001235962,
+      "learning_rate": 2.6308160602985944e-05,
+      "loss": 0.454,
+      "step": 16350
+    },
+    {
+      "epoch": 2.3770699713737002,
+      "grad_norm": 2.3498804569244385,
+      "learning_rate": 2.6235686331352372e-05,
+      "loss": 0.4782,
+      "step": 16400
+    },
+    {
+      "epoch": 2.384317135920571,
+      "grad_norm": 2.642273187637329,
+      "learning_rate": 2.61632120597188e-05,
+      "loss": 0.4531,
+      "step": 16450
+    },
+    {
+      "epoch": 2.3915643004674423,
+      "grad_norm": 3.5636932849884033,
+      "learning_rate": 2.609073778808523e-05,
+      "loss": 0.4545,
+      "step": 16500
+    },
+    {
+      "epoch": 2.398811465014313,
+      "grad_norm": 3.15792179107666,
+      "learning_rate": 2.601826351645166e-05,
+      "loss": 0.4827,
+      "step": 16550
+    },
+    {
+      "epoch": 2.4060586295611843,
+      "grad_norm": 2.4860501289367676,
+      "learning_rate": 2.594578924481809e-05,
+      "loss": 0.4629,
+      "step": 16600
+    },
+    {
+      "epoch": 2.413305794108055,
+      "grad_norm": 5.471420764923096,
+      "learning_rate": 2.5873314973184522e-05,
+      "loss": 0.4376,
+      "step": 16650
+    },
+    {
+      "epoch": 2.4205529586549264,
+      "grad_norm": 3.098515033721924,
+      "learning_rate": 2.580084070155095e-05,
+      "loss": 0.4371,
+      "step": 16700
+    },
+    {
+      "epoch": 2.427800123201797,
+      "grad_norm": 2.5739009380340576,
+      "learning_rate": 2.5728366429917383e-05,
+      "loss": 0.4339,
+      "step": 16750
+    },
+    {
+      "epoch": 2.4350472877486684,
+      "grad_norm": 3.0826826095581055,
+      "learning_rate": 2.565589215828381e-05,
+      "loss": 0.4541,
+      "step": 16800
+    },
+    {
+      "epoch": 2.442294452295539,
+      "grad_norm": 3.0049545764923096,
+      "learning_rate": 2.558341788665024e-05,
+      "loss": 0.4492,
+      "step": 16850
+    },
+    {
+      "epoch": 2.4495416168424105,
+      "grad_norm": 3.194600820541382,
+      "learning_rate": 2.5510943615016668e-05,
+      "loss": 0.4509,
+      "step": 16900
+    },
+    {
+      "epoch": 2.4567887813892813,
+      "grad_norm": 2.9486989974975586,
+      "learning_rate": 2.5438469343383097e-05,
+      "loss": 0.4104,
+      "step": 16950
+    },
+    {
+      "epoch": 2.4640359459361525,
+      "grad_norm": 2.8794190883636475,
+      "learning_rate": 2.5365995071749532e-05,
+      "loss": 0.4308,
+      "step": 17000
+    },
+    {
+      "epoch": 2.4640359459361525,
+      "eval_loss": 0.6368168592453003,
+      "eval_runtime": 245.6078,
+      "eval_samples_per_second": 140.447,
+      "eval_steps_per_second": 14.047,
+      "step": 17000
+    },
+    {
+      "epoch": 2.4712831104830233,
+      "grad_norm": 2.2457711696624756,
+      "learning_rate": 2.529352080011596e-05,
+      "loss": 0.4467,
+      "step": 17050
+    },
+    {
+      "epoch": 2.4785302750298945,
+      "grad_norm": 3.0438215732574463,
+      "learning_rate": 2.522104652848239e-05,
+      "loss": 0.4748,
+      "step": 17100
+    },
+    {
+      "epoch": 2.485777439576766,
+      "grad_norm": 3.1453700065612793,
+      "learning_rate": 2.5148572256848818e-05,
+      "loss": 0.4594,
+      "step": 17150
+    },
+    {
+      "epoch": 2.4930246041236366,
+      "grad_norm": 2.476498603820801,
+      "learning_rate": 2.5076097985215253e-05,
+      "loss": 0.4637,
+      "step": 17200
+    },
+    {
+      "epoch": 2.500271768670508,
+      "grad_norm": 2.5618162155151367,
+      "learning_rate": 2.5003623713581682e-05,
+      "loss": 0.4272,
+      "step": 17250
+    },
+    {
+      "epoch": 2.5075189332173786,
+      "grad_norm": 2.719830274581909,
+      "learning_rate": 2.4932598927380782e-05,
+      "loss": 0.4436,
+      "step": 17300
+    },
+    {
+      "epoch": 2.51476609776425,
+      "grad_norm": 2.393423557281494,
+      "learning_rate": 2.486012465574721e-05,
+      "loss": 0.5006,
+      "step": 17350
+    },
+    {
+      "epoch": 2.5220132623111207,
+      "grad_norm": 2.518490791320801,
+      "learning_rate": 2.4787650384113642e-05,
+      "loss": 0.4378,
+      "step": 17400
+    },
+    {
+      "epoch": 2.529260426857992,
+      "grad_norm": 3.152761697769165,
+      "learning_rate": 2.471517611248007e-05,
+      "loss": 0.4415,
+      "step": 17450
+    },
+    {
+      "epoch": 2.5365075914048627,
+      "grad_norm": 2.689821243286133,
+      "learning_rate": 2.46427018408465e-05,
+      "loss": 0.4736,
+      "step": 17500
+    },
+    {
+      "epoch": 2.543754755951734,
+      "grad_norm": 2.7041850090026855,
+      "learning_rate": 2.457022756921293e-05,
+      "loss": 0.4619,
+      "step": 17550
+    },
+    {
+      "epoch": 2.5510019204986047,
+      "grad_norm": 2.8072993755340576,
+      "learning_rate": 2.449775329757936e-05,
+      "loss": 0.443,
+      "step": 17600
+    },
+    {
+      "epoch": 2.558249085045476,
+      "grad_norm": 2.700951099395752,
+      "learning_rate": 2.4425279025945792e-05,
+      "loss": 0.4338,
+      "step": 17650
+    },
+    {
+      "epoch": 2.5654962495923472,
+      "grad_norm": 2.2559311389923096,
+      "learning_rate": 2.435280475431222e-05,
+      "loss": 0.4371,
+      "step": 17700
+    },
+    {
+      "epoch": 2.572743414139218,
+      "grad_norm": 2.7183778285980225,
+      "learning_rate": 2.4280330482678653e-05,
+      "loss": 0.4479,
+      "step": 17750
+    },
+    {
+      "epoch": 2.579990578686089,
+      "grad_norm": 2.337385654449463,
+      "learning_rate": 2.420785621104508e-05,
+      "loss": 0.4703,
+      "step": 17800
+    },
+    {
+      "epoch": 2.58723774323296,
+      "grad_norm": 2.985521078109741,
+      "learning_rate": 2.413538193941151e-05,
+      "loss": 0.44,
+      "step": 17850
+    },
+    {
+      "epoch": 2.5944849077798313,
+      "grad_norm": 2.26230788230896,
+      "learning_rate": 2.406290766777794e-05,
+      "loss": 0.4485,
+      "step": 17900
+    },
+    {
+      "epoch": 2.601732072326702,
+      "grad_norm": 3.0831480026245117,
+      "learning_rate": 2.399043339614437e-05,
+      "loss": 0.439,
+      "step": 17950
+    },
+    {
+      "epoch": 2.6089792368735734,
+      "grad_norm": 3.325925827026367,
+      "learning_rate": 2.39179591245108e-05,
+      "loss": 0.4809,
+      "step": 18000
+    },
+    {
+      "epoch": 2.6089792368735734,
+      "eval_loss": 0.6339168548583984,
+      "eval_runtime": 245.517,
+      "eval_samples_per_second": 140.499,
+      "eval_steps_per_second": 14.052,
+      "step": 18000
+    },
+    {
+      "epoch": 2.616226401420444,
+      "grad_norm": 3.043121099472046,
+      "learning_rate": 2.3845484852877227e-05,
+      "loss": 0.4382,
+      "step": 18050
+    },
+    {
+      "epoch": 2.6234735659673154,
+      "grad_norm": 2.225372314453125,
+      "learning_rate": 2.377301058124366e-05,
+      "loss": 0.4379,
+      "step": 18100
+    },
+    {
+      "epoch": 2.630720730514186,
+      "grad_norm": 3.059220790863037,
+      "learning_rate": 2.3700536309610088e-05,
+      "loss": 0.4475,
+      "step": 18150
+    },
+    {
+      "epoch": 2.6379678950610574,
+      "grad_norm": 3.8372066020965576,
+      "learning_rate": 2.362806203797652e-05,
+      "loss": 0.4272,
+      "step": 18200
+    },
+    {
+      "epoch": 2.6452150596079282,
+      "grad_norm": 1.4763438701629639,
+      "learning_rate": 2.355558776634295e-05,
+      "loss": 0.4629,
+      "step": 18250
+    },
+    {
+      "epoch": 2.6524622241547995,
+      "grad_norm": 3.0369021892547607,
+      "learning_rate": 2.348311349470938e-05,
+      "loss": 0.438,
+      "step": 18300
+    },
+    {
+      "epoch": 2.6597093887016703,
+      "grad_norm": 2.6003599166870117,
+      "learning_rate": 2.341063922307581e-05,
+      "loss": 0.4205,
+      "step": 18350
+    },
+    {
+      "epoch": 2.6669565532485415,
+      "grad_norm": 2.9726734161376953,
+      "learning_rate": 2.333816495144224e-05,
+      "loss": 0.4453,
+      "step": 18400
+    },
+    {
+      "epoch": 2.6742037177954128,
+      "grad_norm": 3.3538806438446045,
+      "learning_rate": 2.326569067980867e-05,
+      "loss": 0.4237,
+      "step": 18450
+    },
+    {
+      "epoch": 2.6814508823422836,
+      "grad_norm": 2.160491704940796,
+      "learning_rate": 2.31932164081751e-05,
+      "loss": 0.4433,
+      "step": 18500
+    },
+    {
+      "epoch": 2.6886980468891544,
+      "grad_norm": 3.117530345916748,
+      "learning_rate": 2.312074213654153e-05,
+      "loss": 0.4219,
+      "step": 18550
+    },
+    {
+      "epoch": 2.6959452114360256,
+      "grad_norm": 2.787057638168335,
+      "learning_rate": 2.304826786490796e-05,
+      "loss": 0.4391,
+      "step": 18600
+    },
+    {
+      "epoch": 2.703192375982897,
+      "grad_norm": 3.635530471801758,
+      "learning_rate": 2.2975793593274387e-05,
+      "loss": 0.461,
+      "step": 18650
+    },
+    {
+      "epoch": 2.7104395405297677,
+      "grad_norm": 2.3657073974609375,
+      "learning_rate": 2.290331932164082e-05,
+      "loss": 0.4692,
+      "step": 18700
+    },
+    {
+      "epoch": 2.717686705076639,
+      "grad_norm": 2.611757516860962,
+      "learning_rate": 2.2830845050007248e-05,
+      "loss": 0.4479,
+      "step": 18750
+    },
+    {
+      "epoch": 2.7249338696235097,
+      "grad_norm": 2.967528820037842,
+      "learning_rate": 2.2758370778373676e-05,
+      "loss": 0.424,
+      "step": 18800
+    },
+    {
+      "epoch": 2.732181034170381,
+      "grad_norm": 5.77213191986084,
+      "learning_rate": 2.268589650674011e-05,
+      "loss": 0.4389,
+      "step": 18850
+    },
+    {
+      "epoch": 2.7394281987172517,
+      "grad_norm": 2.8429954051971436,
+      "learning_rate": 2.2613422235106537e-05,
+      "loss": 0.4829,
+      "step": 18900
+    },
+    {
+      "epoch": 2.746675363264123,
+      "grad_norm": 2.9622368812561035,
+      "learning_rate": 2.254094796347297e-05,
+      "loss": 0.4781,
+      "step": 18950
+    },
+    {
+      "epoch": 2.7539225278109942,
+      "grad_norm": 3.898066997528076,
+      "learning_rate": 2.2468473691839397e-05,
+      "loss": 0.4847,
+      "step": 19000
+    },
+    {
+      "epoch": 2.7539225278109942,
+      "eval_loss": 0.6265138983726501,
+      "eval_runtime": 245.5072,
+      "eval_samples_per_second": 140.505,
+      "eval_steps_per_second": 14.053,
+      "step": 19000
+    },
+    {
+      "epoch": 2.761169692357865,
+      "grad_norm": 2.6696014404296875,
+      "learning_rate": 2.239599942020583e-05,
+      "loss": 0.4233,
+      "step": 19050
+    },
+    {
+      "epoch": 2.768416856904736,
+      "grad_norm": 2.9069409370422363,
+      "learning_rate": 2.2323525148572258e-05,
+      "loss": 0.4632,
+      "step": 19100
+    },
+    {
+      "epoch": 2.775664021451607,
+      "grad_norm": 3.3168020248413086,
+      "learning_rate": 2.2251050876938687e-05,
+      "loss": 0.4483,
+      "step": 19150
+    },
+    {
+      "epoch": 2.7829111859984783,
+      "grad_norm": 2.685267686843872,
+      "learning_rate": 2.217857660530512e-05,
+      "loss": 0.4321,
+      "step": 19200
+    },
+    {
+      "epoch": 2.790158350545349,
+      "grad_norm": 3.203169584274292,
+      "learning_rate": 2.2106102333671547e-05,
+      "loss": 0.4187,
+      "step": 19250
+    },
+    {
+      "epoch": 2.79740551509222,
+      "grad_norm": 2.244285821914673,
+      "learning_rate": 2.203362806203798e-05,
+      "loss": 0.4336,
+      "step": 19300
+    },
+    {
+      "epoch": 2.804652679639091,
+      "grad_norm": 2.8344221115112305,
+      "learning_rate": 2.1961153790404408e-05,
+      "loss": 0.4469,
+      "step": 19350
+    },
+    {
+      "epoch": 2.8118998441859624,
+      "grad_norm": 2.193204879760742,
+      "learning_rate": 2.188867951877084e-05,
+      "loss": 0.4394,
+      "step": 19400
+    },
+    {
+      "epoch": 2.819147008732833,
+      "grad_norm": 2.7660672664642334,
+      "learning_rate": 2.1816205247137268e-05,
+      "loss": 0.4646,
+      "step": 19450
+    },
+    {
+      "epoch": 2.8263941732797044,
+      "grad_norm": 1.7778669595718384,
+      "learning_rate": 2.1743730975503697e-05,
+      "loss": 0.4176,
+      "step": 19500
+    },
+    {
+      "epoch": 2.8336413378265752,
+      "grad_norm": 2.1182384490966797,
+      "learning_rate": 2.1671256703870125e-05,
+      "loss": 0.4266,
+      "step": 19550
+    },
+    {
+      "epoch": 2.8408885023734465,
+      "grad_norm": 3.023648262023926,
+      "learning_rate": 2.1598782432236557e-05,
+      "loss": 0.4378,
+      "step": 19600
+    },
+    {
+      "epoch": 2.8481356669203173,
+      "grad_norm": 3.217515707015991,
+      "learning_rate": 2.1526308160602986e-05,
+      "loss": 0.4328,
+      "step": 19650
+    },
+    {
+      "epoch": 2.8553828314671885,
+      "grad_norm": 2.5275774002075195,
+      "learning_rate": 2.1453833888969414e-05,
+      "loss": 0.3978,
+      "step": 19700
+    },
+    {
+      "epoch": 2.8626299960140598,
+      "grad_norm": 3.3163788318634033,
+      "learning_rate": 2.1381359617335846e-05,
+      "loss": 0.4594,
+      "step": 19750
+    },
+    {
+      "epoch": 2.8698771605609306,
+      "grad_norm": 3.3472440242767334,
+      "learning_rate": 2.1308885345702275e-05,
+      "loss": 0.4368,
+      "step": 19800
+    },
+    {
+      "epoch": 2.8771243251078014,
+      "grad_norm": 3.1864593029022217,
+      "learning_rate": 2.1236411074068707e-05,
+      "loss": 0.4715,
+      "step": 19850
+    },
+    {
+      "epoch": 2.8843714896546726,
+      "grad_norm": 2.73544979095459,
+      "learning_rate": 2.1163936802435135e-05,
+      "loss": 0.443,
+      "step": 19900
+    },
+    {
+      "epoch": 2.891618654201544,
+      "grad_norm": 3.4767727851867676,
+      "learning_rate": 2.1091462530801567e-05,
+      "loss": 0.4429,
+      "step": 19950
+    },
+    {
+      "epoch": 2.8988658187484146,
+      "grad_norm": 2.7174811363220215,
+      "learning_rate": 2.1018988259167996e-05,
+      "loss": 0.4067,
+      "step": 20000
+    },
+    {
+      "epoch": 2.8988658187484146,
+      "eval_loss": 0.622748076915741,
+      "eval_runtime": 245.5244,
+      "eval_samples_per_second": 140.495,
+      "eval_steps_per_second": 14.052,
+      "step": 20000
+    },
+    {
+      "epoch": 2.9061129832952854,
+      "grad_norm": 2.311414957046509,
+      "learning_rate": 2.0947963472967096e-05,
+      "loss": 0.451,
+      "step": 20050
+    },
+    {
+      "epoch": 2.9133601478421567,
+      "grad_norm": 2.9489965438842773,
+      "learning_rate": 2.0875489201333528e-05,
+      "loss": 0.4621,
+      "step": 20100
+    },
+    {
+      "epoch": 2.920607312389028,
+      "grad_norm": 3.1284987926483154,
+      "learning_rate": 2.0803014929699957e-05,
+      "loss": 0.4455,
+      "step": 20150
+    },
+    {
+      "epoch": 2.9278544769358987,
+      "grad_norm": 2.579033613204956,
+      "learning_rate": 2.073054065806639e-05,
+      "loss": 0.4555,
+      "step": 20200
+    },
+    {
+      "epoch": 2.93510164148277,
+      "grad_norm": 2.4020378589630127,
+      "learning_rate": 2.0658066386432817e-05,
+      "loss": 0.4531,
+      "step": 20250
+    },
+    {
+      "epoch": 2.9423488060296408,
+      "grad_norm": 2.3254435062408447,
+      "learning_rate": 2.0585592114799246e-05,
+      "loss": 0.4434,
+      "step": 20300
+    },
+    {
+      "epoch": 2.949595970576512,
+      "grad_norm": 2.742219924926758,
+      "learning_rate": 2.0513117843165678e-05,
+      "loss": 0.444,
+      "step": 20350
+    },
+    {
+      "epoch": 2.956843135123383,
+      "grad_norm": 2.3544771671295166,
+      "learning_rate": 2.0440643571532106e-05,
+      "loss": 0.4576,
+      "step": 20400
+    },
+    {
+      "epoch": 2.964090299670254,
+      "grad_norm": 2.842970609664917,
+      "learning_rate": 2.0368169299898538e-05,
+      "loss": 0.4416,
+      "step": 20450
+    },
+    {
+      "epoch": 2.9713374642171253,
+      "grad_norm": 2.2534501552581787,
+      "learning_rate": 2.0295695028264967e-05,
+      "loss": 0.4348,
+      "step": 20500
+    },
+    {
+      "epoch": 2.978584628763996,
+      "grad_norm": 2.976383924484253,
+      "learning_rate": 2.02232207566314e-05,
+      "loss": 0.4237,
+      "step": 20550
+    },
+    {
+      "epoch": 2.985831793310867,
+      "grad_norm": 3.3762621879577637,
+      "learning_rate": 2.0150746484997827e-05,
+      "loss": 0.4506,
+      "step": 20600
+    },
+    {
+      "epoch": 2.993078957857738,
+      "grad_norm": 3.0444388389587402,
+      "learning_rate": 2.0078272213364256e-05,
+      "loss": 0.4357,
+      "step": 20650
+    },
+    {
+      "epoch": 3.0003261224046094,
+      "grad_norm": 3.294370412826538,
+      "learning_rate": 2.0005797941730688e-05,
+      "loss": 0.4358,
+      "step": 20700
+    },
+    {
+      "epoch": 3.00757328695148,
+      "grad_norm": 2.1435303688049316,
+      "learning_rate": 1.9933323670097116e-05,
+      "loss": 0.343,
+      "step": 20750
+    },
+    {
+      "epoch": 3.0148204514983514,
+      "grad_norm": 2.5070137977600098,
+      "learning_rate": 1.9860849398463545e-05,
+      "loss": 0.3397,
+      "step": 20800
+    },
+    {
+      "epoch": 3.0220676160452222,
+      "grad_norm": 2.5345394611358643,
+      "learning_rate": 1.9788375126829974e-05,
+      "loss": 0.3221,
+      "step": 20850
+    },
+    {
+      "epoch": 3.0293147805920935,
+      "grad_norm": 2.4468677043914795,
+      "learning_rate": 1.9715900855196406e-05,
+      "loss": 0.3404,
+      "step": 20900
+    },
+    {
+      "epoch": 3.0365619451389643,
+      "grad_norm": 3.5916829109191895,
+      "learning_rate": 1.964487606899551e-05,
+      "loss": 0.3412,
+      "step": 20950
+    },
+    {
+      "epoch": 3.0438091096858355,
+      "grad_norm": 1.7320780754089355,
+      "learning_rate": 1.9572401797361938e-05,
+      "loss": 0.3573,
+      "step": 21000
+    },
+    {
+      "epoch": 3.0438091096858355,
+      "eval_loss": 0.6396881937980652,
+      "eval_runtime": 245.0839,
+      "eval_samples_per_second": 140.748,
+      "eval_steps_per_second": 14.077,
+      "step": 21000
+    },
+    {
+      "epoch": 3.0510562742327063,
+      "grad_norm": 2.4702775478363037,
+      "learning_rate": 1.9499927525728366e-05,
+      "loss": 0.3485,
+      "step": 21050
+    },
+    {
+      "epoch": 3.0583034387795776,
+      "grad_norm": 3.04402232170105,
+      "learning_rate": 1.9427453254094795e-05,
+      "loss": 0.3575,
+      "step": 21100
+    },
+    {
+      "epoch": 3.0655506033264484,
+      "grad_norm": 2.487696886062622,
+      "learning_rate": 1.9354978982461227e-05,
+      "loss": 0.3178,
+      "step": 21150
+    },
+    {
+      "epoch": 3.0727977678733196,
+      "grad_norm": 2.565079689025879,
+      "learning_rate": 1.9282504710827655e-05,
+      "loss": 0.3311,
+      "step": 21200
+    },
+    {
+      "epoch": 3.0800449324201904,
+      "grad_norm": 3.043264865875244,
+      "learning_rate": 1.9210030439194087e-05,
+      "loss": 0.3201,
+      "step": 21250
+    },
+    {
+      "epoch": 3.0872920969670616,
+      "grad_norm": 3.086071729660034,
+      "learning_rate": 1.9137556167560516e-05,
+      "loss": 0.3385,
+      "step": 21300
+    },
+    {
+      "epoch": 3.094539261513933,
+      "grad_norm": 1.5752780437469482,
+      "learning_rate": 1.9065081895926948e-05,
+      "loss": 0.305,
+      "step": 21350
+    },
+    {
+      "epoch": 3.1017864260608037,
+      "grad_norm": 2.464972496032715,
+      "learning_rate": 1.8992607624293376e-05,
+      "loss": 0.3421,
+      "step": 21400
+    },
+    {
+      "epoch": 3.109033590607675,
+      "grad_norm": 2.98641037940979,
+      "learning_rate": 1.892013335265981e-05,
+      "loss": 0.3581,
+      "step": 21450
+    },
+    {
+      "epoch": 3.1162807551545457,
+      "grad_norm": 2.293949842453003,
+      "learning_rate": 1.8847659081026237e-05,
+      "loss": 0.3241,
+      "step": 21500
+    },
+    {
+      "epoch": 3.123527919701417,
+      "grad_norm": 2.361656427383423,
+      "learning_rate": 1.8775184809392666e-05,
+      "loss": 0.3154,
+      "step": 21550
+    },
+    {
+      "epoch": 3.1307750842482878,
+      "grad_norm": 3.095930576324463,
+      "learning_rate": 1.8702710537759097e-05,
+      "loss": 0.3043,
+      "step": 21600
+    },
+    {
+      "epoch": 3.138022248795159,
+      "grad_norm": 2.254836320877075,
+      "learning_rate": 1.8630236266125526e-05,
+      "loss": 0.3381,
+      "step": 21650
+    },
+    {
+      "epoch": 3.14526941334203,
+      "grad_norm": 3.281912088394165,
+      "learning_rate": 1.8557761994491958e-05,
+      "loss": 0.3292,
+      "step": 21700
+    },
+    {
+      "epoch": 3.152516577888901,
+      "grad_norm": 2.4811136722564697,
+      "learning_rate": 1.8485287722858387e-05,
+      "loss": 0.3221,
+      "step": 21750
+    },
+    {
+      "epoch": 3.159763742435772,
+      "grad_norm": 2.5498745441436768,
+      "learning_rate": 1.8412813451224815e-05,
+      "loss": 0.3548,
+      "step": 21800
+    },
+    {
+      "epoch": 3.167010906982643,
+      "grad_norm": 2.9773364067077637,
+      "learning_rate": 1.8340339179591247e-05,
+      "loss": 0.3199,
+      "step": 21850
+    },
+    {
+      "epoch": 3.174258071529514,
+      "grad_norm": 2.239015579223633,
+      "learning_rate": 1.8267864907957676e-05,
+      "loss": 0.3258,
+      "step": 21900
+    },
+    {
+      "epoch": 3.181505236076385,
+      "grad_norm": 2.470496416091919,
+      "learning_rate": 1.8195390636324104e-05,
+      "loss": 0.3199,
+      "step": 21950
+    },
+    {
+      "epoch": 3.1887524006232564,
+      "grad_norm": 2.567301034927368,
+      "learning_rate": 1.8122916364690536e-05,
+      "loss": 0.3859,
+      "step": 22000
+    },
+    {
+      "epoch": 3.1887524006232564,
+      "eval_loss": 0.6378082036972046,
+      "eval_runtime": 245.7126,
+      "eval_samples_per_second": 140.388,
+      "eval_steps_per_second": 14.041,
+      "step": 22000
+    },
+    {
+      "epoch": 3.195999565170127,
+      "grad_norm": 3.044546127319336,
+      "learning_rate": 1.8050442093056965e-05,
+      "loss": 0.3116,
+      "step": 22050
+    },
+    {
+      "epoch": 3.2032467297169984,
+      "grad_norm": 2.832991600036621,
+      "learning_rate": 1.7977967821423393e-05,
+      "loss": 0.3343,
+      "step": 22100
+    },
+    {
+      "epoch": 3.210493894263869,
+      "grad_norm": 2.9658920764923096,
+      "learning_rate": 1.7905493549789825e-05,
+      "loss": 0.3576,
+      "step": 22150
+    },
+    {
+      "epoch": 3.2177410588107405,
+      "grad_norm": 3.258549213409424,
+      "learning_rate": 1.7833019278156254e-05,
+      "loss": 0.3396,
+      "step": 22200
+    },
+    {
+      "epoch": 3.2249882233576113,
+      "grad_norm": 3.3627419471740723,
+      "learning_rate": 1.7760545006522686e-05,
+      "loss": 0.3206,
+      "step": 22250
+    },
+    {
+      "epoch": 3.2322353879044825,
+      "grad_norm": 2.9110984802246094,
+      "learning_rate": 1.7688070734889114e-05,
+      "loss": 0.3612,
+      "step": 22300
+    },
+    {
+      "epoch": 3.2394825524513533,
+      "grad_norm": 2.8267645835876465,
+      "learning_rate": 1.7615596463255546e-05,
+      "loss": 0.3477,
+      "step": 22350
+    },
+    {
+      "epoch": 3.2467297169982245,
+      "grad_norm": 2.8990402221679688,
+      "learning_rate": 1.7543122191621975e-05,
+      "loss": 0.3301,
+      "step": 22400
+    },
+    {
+      "epoch": 3.2539768815450953,
+      "grad_norm": 2.5321147441864014,
+      "learning_rate": 1.7470647919988407e-05,
+      "loss": 0.3212,
+      "step": 22450
+    },
+    {
+      "epoch": 3.2612240460919666,
+      "grad_norm": 3.9064178466796875,
+      "learning_rate": 1.7398173648354835e-05,
+      "loss": 0.3752,
+      "step": 22500
+    },
+    {
+      "epoch": 3.2684712106388374,
+      "grad_norm": 2.5248515605926514,
+      "learning_rate": 1.7325699376721267e-05,
+      "loss": 0.3467,
+      "step": 22550
+    },
+    {
+      "epoch": 3.2757183751857086,
+      "grad_norm": 2.442370653152466,
+      "learning_rate": 1.7253225105087696e-05,
+      "loss": 0.3171,
+      "step": 22600
+    },
+    {
+      "epoch": 3.2829655397325794,
+      "grad_norm": 2.523216485977173,
+      "learning_rate": 1.7180750833454125e-05,
+      "loss": 0.3422,
+      "step": 22650
+    },
+    {
+      "epoch": 3.2902127042794507,
+      "grad_norm": 3.633876323699951,
+      "learning_rate": 1.7108276561820553e-05,
+      "loss": 0.3236,
+      "step": 22700
+    },
+    {
+      "epoch": 3.297459868826322,
+      "grad_norm": 3.5989325046539307,
+      "learning_rate": 1.7035802290186985e-05,
+      "loss": 0.3562,
+      "step": 22750
+    },
+    {
+      "epoch": 3.3047070333731927,
+      "grad_norm": 2.8764047622680664,
+      "learning_rate": 1.6963328018553414e-05,
+      "loss": 0.3432,
+      "step": 22800
+    },
+    {
+      "epoch": 3.311954197920064,
+      "grad_norm": 2.8960604667663574,
+      "learning_rate": 1.6890853746919842e-05,
+      "loss": 0.339,
+      "step": 22850
+    },
+    {
+      "epoch": 3.3192013624669348,
+      "grad_norm": 3.1542296409606934,
+      "learning_rate": 1.6818379475286274e-05,
+      "loss": 0.3578,
+      "step": 22900
+    },
+    {
+      "epoch": 3.326448527013806,
+      "grad_norm": 3.966387987136841,
+      "learning_rate": 1.6745905203652703e-05,
+      "loss": 0.3437,
+      "step": 22950
+    },
+    {
+      "epoch": 3.333695691560677,
+      "grad_norm": 2.1080222129821777,
+      "learning_rate": 1.6673430932019135e-05,
+      "loss": 0.3476,
+      "step": 23000
+    },
+    {
+      "epoch": 3.333695691560677,
+      "eval_loss": 0.6372683644294739,
+      "eval_runtime": 245.8192,
+      "eval_samples_per_second": 140.327,
+      "eval_steps_per_second": 14.035,
+      "step": 23000
+    },
+    {
+      "epoch": 3.340942856107548,
+      "grad_norm": 2.4654605388641357,
+      "learning_rate": 1.6600956660385563e-05,
+      "loss": 0.3582,
+      "step": 23050
+    },
+    {
+      "epoch": 3.348190020654419,
+      "grad_norm": 3.2088425159454346,
+      "learning_rate": 1.6528482388751995e-05,
+      "loss": 0.3121,
+      "step": 23100
+    },
+    {
+      "epoch": 3.35543718520129,
+      "grad_norm": 2.5280325412750244,
+      "learning_rate": 1.6456008117118424e-05,
+      "loss": 0.3233,
+      "step": 23150
+    },
+    {
+      "epoch": 3.362684349748161,
+      "grad_norm": 2.5772511959075928,
+      "learning_rate": 1.6383533845484852e-05,
+      "loss": 0.3277,
+      "step": 23200
+    },
+    {
+      "epoch": 3.369931514295032,
+      "grad_norm": 2.4769599437713623,
+      "learning_rate": 1.6311059573851284e-05,
+      "loss": 0.3349,
+      "step": 23250
+    },
+    {
+      "epoch": 3.377178678841903,
+      "grad_norm": 3.1656038761138916,
+      "learning_rate": 1.6238585302217713e-05,
+      "loss": 0.3601,
+      "step": 23300
+    },
+    {
+      "epoch": 3.384425843388774,
+      "grad_norm": 3.1911141872406006,
+      "learning_rate": 1.6166111030584145e-05,
+      "loss": 0.3664,
+      "step": 23350
+    },
+    {
+      "epoch": 3.391673007935645,
+      "grad_norm": 3.001246213912964,
+      "learning_rate": 1.6093636758950574e-05,
+      "loss": 0.3547,
+      "step": 23400
+    },
+    {
+      "epoch": 3.398920172482516,
+      "grad_norm": 2.2357707023620605,
+      "learning_rate": 1.6021162487317005e-05,
+      "loss": 0.3581,
+      "step": 23450
+    },
+    {
+      "epoch": 3.4061673370293875,
+      "grad_norm": 2.813751459121704,
+      "learning_rate": 1.5948688215683434e-05,
+      "loss": 0.3436,
+      "step": 23500
+    },
+    {
+      "epoch": 3.4134145015762583,
+      "grad_norm": 2.3340563774108887,
+      "learning_rate": 1.5876213944049863e-05,
+      "loss": 0.3424,
+      "step": 23550
+    },
+    {
+      "epoch": 3.4206616661231295,
+      "grad_norm": 3.3509624004364014,
+      "learning_rate": 1.580373967241629e-05,
+      "loss": 0.365,
+      "step": 23600
+    },
+    {
+      "epoch": 3.4279088306700003,
+      "grad_norm": 2.6918370723724365,
+      "learning_rate": 1.5731265400782723e-05,
+      "loss": 0.3396,
+      "step": 23650
+    },
+    {
+      "epoch": 3.4351559952168715,
+      "grad_norm": 3.110868215560913,
+      "learning_rate": 1.5658791129149152e-05,
+      "loss": 0.3303,
+      "step": 23700
+    },
+    {
+      "epoch": 3.4424031597637423,
+      "grad_norm": 3.6813771724700928,
+      "learning_rate": 1.558631685751558e-05,
+      "loss": 0.3726,
+      "step": 23750
+    },
+    {
+      "epoch": 3.4496503243106136,
+      "grad_norm": 2.564406633377075,
+      "learning_rate": 1.5513842585882012e-05,
+      "loss": 0.3507,
+      "step": 23800
+    },
+    {
+      "epoch": 3.4568974888574844,
+      "grad_norm": 2.496525526046753,
+      "learning_rate": 1.544136831424844e-05,
+      "loss": 0.3348,
+      "step": 23850
+    },
+    {
+      "epoch": 3.4641446534043556,
+      "grad_norm": 3.00034761428833,
+      "learning_rate": 1.5368894042614873e-05,
+      "loss": 0.35,
+      "step": 23900
+    },
+    {
+      "epoch": 3.4713918179512264,
+      "grad_norm": 2.688913106918335,
+      "learning_rate": 1.52964197709813e-05,
+      "loss": 0.3123,
+      "step": 23950
+    },
+    {
+      "epoch": 3.4786389824980977,
+      "grad_norm": 3.100461721420288,
+      "learning_rate": 1.5223945499347733e-05,
+      "loss": 0.3559,
+      "step": 24000
+    },
+    {
+      "epoch": 3.4786389824980977,
+      "eval_loss": 0.6312422156333923,
+      "eval_runtime": 246.9159,
+      "eval_samples_per_second": 139.703,
+      "eval_steps_per_second": 13.972,
+      "step": 24000
+    },
+    {
+      "epoch": 3.4858861470449685,
+      "grad_norm": 2.221304178237915,
+      "learning_rate": 1.5151471227714162e-05,
+      "loss": 0.3437,
+      "step": 24050
+    },
+    {
+      "epoch": 3.4931333115918397,
+      "grad_norm": 2.807159662246704,
+      "learning_rate": 1.5078996956080594e-05,
+      "loss": 0.3284,
+      "step": 24100
+    },
+    {
+      "epoch": 3.5003804761387105,
+      "grad_norm": 2.910870313644409,
+      "learning_rate": 1.5006522684447022e-05,
+      "loss": 0.3436,
+      "step": 24150
+    },
+    {
+      "epoch": 3.5076276406855817,
+      "grad_norm": 2.7148754596710205,
+      "learning_rate": 1.4934048412813451e-05,
+      "loss": 0.364,
+      "step": 24200
+    },
+    {
+      "epoch": 3.514874805232453,
+      "grad_norm": 3.629567861557007,
+      "learning_rate": 1.4861574141179881e-05,
+      "loss": 0.3293,
+      "step": 24250
+    },
+    {
+      "epoch": 3.522121969779324,
+      "grad_norm": 1.6957030296325684,
+      "learning_rate": 1.478909986954631e-05,
+      "loss": 0.3418,
+      "step": 24300
+    },
+    {
+      "epoch": 3.529369134326195,
+      "grad_norm": 2.671588659286499,
+      "learning_rate": 1.4716625597912742e-05,
+      "loss": 0.3054,
+      "step": 24350
+    },
+    {
+      "epoch": 3.536616298873066,
+      "grad_norm": 2.832435369491577,
+      "learning_rate": 1.464415132627917e-05,
+      "loss": 0.3457,
+      "step": 24400
+    },
+    {
+      "epoch": 3.543863463419937,
+      "grad_norm": 3.806084156036377,
+      "learning_rate": 1.4571677054645602e-05,
+      "loss": 0.3366,
+      "step": 24450
+    },
+    {
+      "epoch": 3.551110627966808,
+      "grad_norm": 3.169780731201172,
+      "learning_rate": 1.4499202783012031e-05,
+      "loss": 0.3337,
+      "step": 24500
+    },
+    {
+      "epoch": 3.558357792513679,
+      "grad_norm": 2.461219310760498,
+      "learning_rate": 1.4426728511378463e-05,
+      "loss": 0.3394,
+      "step": 24550
+    },
+    {
+      "epoch": 3.56560495706055,
+      "grad_norm": 2.458402633666992,
+      "learning_rate": 1.4354254239744891e-05,
+      "loss": 0.3245,
+      "step": 24600
+    },
+    {
+      "epoch": 3.572852121607421,
+      "grad_norm": 2.172034740447998,
+      "learning_rate": 1.4281779968111322e-05,
+      "loss": 0.3335,
+      "step": 24650
+    },
+    {
+      "epoch": 3.580099286154292,
+      "grad_norm": 2.7269339561462402,
+      "learning_rate": 1.420930569647775e-05,
+      "loss": 0.3268,
+      "step": 24700
+    },
+    {
+      "epoch": 3.587346450701163,
+      "grad_norm": 3.2520856857299805,
+      "learning_rate": 1.4136831424844179e-05,
+      "loss": 0.3282,
+      "step": 24750
+    },
+    {
+      "epoch": 3.5945936152480344,
+      "grad_norm": 3.6039845943450928,
+      "learning_rate": 1.406435715321061e-05,
+      "loss": 0.3375,
+      "step": 24800
+    },
+    {
+      "epoch": 3.6018407797949052,
+      "grad_norm": 2.7368907928466797,
+      "learning_rate": 1.399188288157704e-05,
+      "loss": 0.3368,
+      "step": 24850
+    },
+    {
+      "epoch": 3.609087944341776,
+      "grad_norm": 3.1287124156951904,
+      "learning_rate": 1.3919408609943471e-05,
+      "loss": 0.3517,
+      "step": 24900
+    },
+    {
+      "epoch": 3.6163351088886473,
+      "grad_norm": 3.3379523754119873,
+      "learning_rate": 1.38469343383099e-05,
+      "loss": 0.3334,
+      "step": 24950
+    },
+    {
+      "epoch": 3.6235822734355185,
+      "grad_norm": 2.828714609146118,
+      "learning_rate": 1.3774460066676332e-05,
+      "loss": 0.3148,
+      "step": 25000
+    },
+    {
+      "epoch": 3.6235822734355185,
+      "eval_loss": 0.6326374411582947,
+      "eval_runtime": 245.5715,
+      "eval_samples_per_second": 140.468,
+      "eval_steps_per_second": 14.049,
+      "step": 25000
+    },
+    {
+      "epoch": 3.6308294379823893,
+      "grad_norm": 2.7624423503875732,
+      "learning_rate": 1.370198579504276e-05,
+      "loss": 0.3333,
+      "step": 25050
+    },
+    {
+      "epoch": 3.6380766025292606,
+      "grad_norm": 2.4403984546661377,
+      "learning_rate": 1.362951152340919e-05,
+      "loss": 0.3306,
+      "step": 25100
+    },
+    {
+      "epoch": 3.6453237670761314,
+      "grad_norm": 3.4315109252929688,
+      "learning_rate": 1.355703725177562e-05,
+      "loss": 0.3428,
+      "step": 25150
+    },
+    {
+      "epoch": 3.6525709316230026,
+      "grad_norm": 4.232142925262451,
+      "learning_rate": 1.3484562980142051e-05,
+      "loss": 0.3437,
+      "step": 25200
+    },
+    {
+      "epoch": 3.6598180961698734,
+      "grad_norm": 2.562215805053711,
+      "learning_rate": 1.341208870850848e-05,
+      "loss": 0.3437,
+      "step": 25250
+    },
+    {
+      "epoch": 3.6670652607167447,
+      "grad_norm": 2.4503726959228516,
+      "learning_rate": 1.3339614436874908e-05,
+      "loss": 0.3018,
+      "step": 25300
+    },
+    {
+      "epoch": 3.6743124252636155,
+      "grad_norm": 2.709066390991211,
+      "learning_rate": 1.3268589650674012e-05,
+      "loss": 0.3582,
+      "step": 25350
+    },
+    {
+      "epoch": 3.6815595898104867,
+      "grad_norm": 2.3442864418029785,
+      "learning_rate": 1.319611537904044e-05,
+      "loss": 0.3224,
+      "step": 25400
+    },
+    {
+      "epoch": 3.6888067543573575,
+      "grad_norm": 4.138051509857178,
+      "learning_rate": 1.312364110740687e-05,
+      "loss": 0.3791,
+      "step": 25450
+    },
+    {
+      "epoch": 3.6960539189042287,
+      "grad_norm": 3.238833427429199,
+      "learning_rate": 1.3051166835773301e-05,
+      "loss": 0.3191,
+      "step": 25500
+    },
+    {
+      "epoch": 3.7033010834511,
+      "grad_norm": 3.0697717666625977,
+      "learning_rate": 1.297869256413973e-05,
+      "loss": 0.32,
+      "step": 25550
+    },
+    {
+      "epoch": 3.710548247997971,
+      "grad_norm": 2.4563581943511963,
+      "learning_rate": 1.2906218292506162e-05,
+      "loss": 0.3269,
+      "step": 25600
+    },
+    {
+      "epoch": 3.7177954125448416,
+      "grad_norm": 2.1714043617248535,
+      "learning_rate": 1.283374402087259e-05,
+      "loss": 0.3085,
+      "step": 25650
+    },
+    {
+      "epoch": 3.725042577091713,
+      "grad_norm": 2.205698013305664,
+      "learning_rate": 1.2761269749239022e-05,
+      "loss": 0.3558,
+      "step": 25700
+    },
+    {
+      "epoch": 3.732289741638584,
+      "grad_norm": 2.4091830253601074,
+      "learning_rate": 1.268879547760545e-05,
+      "loss": 0.3631,
+      "step": 25750
+    },
+    {
+      "epoch": 3.739536906185455,
+      "grad_norm": 1.7875028848648071,
+      "learning_rate": 1.2616321205971881e-05,
+      "loss": 0.3241,
+      "step": 25800
+    },
+    {
+      "epoch": 3.746784070732326,
+      "grad_norm": 3.206444501876831,
+      "learning_rate": 1.2545296419770983e-05,
+      "loss": 0.3237,
+      "step": 25850
+    },
+    {
+      "epoch": 3.754031235279197,
+      "grad_norm": 2.4100027084350586,
+      "learning_rate": 1.2472822148137411e-05,
+      "loss": 0.3116,
+      "step": 25900
+    },
+    {
+      "epoch": 3.761278399826068,
+      "grad_norm": 3.0889265537261963,
+      "learning_rate": 1.2400347876503842e-05,
+      "loss": 0.3599,
+      "step": 25950
+    },
+    {
+      "epoch": 3.768525564372939,
+      "grad_norm": 2.965827703475952,
+      "learning_rate": 1.2327873604870272e-05,
+      "loss": 0.3477,
+      "step": 26000
+    },
+    {
+      "epoch": 3.768525564372939,
+      "eval_loss": 0.6264123320579529,
+      "eval_runtime": 245.5751,
+      "eval_samples_per_second": 140.466,
+      "eval_steps_per_second": 14.049,
+      "step": 26000
+    },
+    {
+      "epoch": 3.77577272891981,
+      "grad_norm": 3.572783946990967,
+      "learning_rate": 1.2255399333236702e-05,
+      "loss": 0.3519,
+      "step": 26050
+    },
+    {
+      "epoch": 3.7830198934666814,
+      "grad_norm": 3.2291600704193115,
+      "learning_rate": 1.218292506160313e-05,
+      "loss": 0.3377,
+      "step": 26100
+    },
+    {
+      "epoch": 3.7902670580135522,
+      "grad_norm": 3.1608381271362305,
+      "learning_rate": 1.2110450789969561e-05,
+      "loss": 0.3306,
+      "step": 26150
+    },
+    {
+      "epoch": 3.797514222560423,
+      "grad_norm": 2.326995611190796,
+      "learning_rate": 1.2037976518335991e-05,
+      "loss": 0.3425,
+      "step": 26200
+    },
+    {
+      "epoch": 3.8047613871072943,
+      "grad_norm": 2.580730438232422,
+      "learning_rate": 1.1965502246702421e-05,
+      "loss": 0.3544,
+      "step": 26250
+    },
+    {
+      "epoch": 3.8120085516541655,
+      "grad_norm": 3.1434969902038574,
+      "learning_rate": 1.1893027975068852e-05,
+      "loss": 0.3475,
+      "step": 26300
+    },
+    {
+      "epoch": 3.8192557162010363,
+      "grad_norm": 2.08758282661438,
+      "learning_rate": 1.1820553703435282e-05,
+      "loss": 0.3287,
+      "step": 26350
+    },
+    {
+      "epoch": 3.826502880747907,
+      "grad_norm": 2.8469362258911133,
+      "learning_rate": 1.174807943180171e-05,
+      "loss": 0.3236,
+      "step": 26400
+    },
+    {
+      "epoch": 3.8337500452947784,
+      "grad_norm": 3.5601906776428223,
+      "learning_rate": 1.1675605160168141e-05,
+      "loss": 0.3263,
+      "step": 26450
+    },
+    {
+      "epoch": 3.8409972098416496,
+      "grad_norm": 3.756640911102295,
+      "learning_rate": 1.1603130888534571e-05,
+      "loss": 0.3382,
+      "step": 26500
+    },
+    {
+      "epoch": 3.8482443743885204,
+      "grad_norm": 2.394885778427124,
+      "learning_rate": 1.1530656616901e-05,
+      "loss": 0.3373,
+      "step": 26550
+    },
+    {
+      "epoch": 3.8554915389353916,
+      "grad_norm": 2.798363208770752,
+      "learning_rate": 1.145818234526743e-05,
+      "loss": 0.3259,
+      "step": 26600
+    },
+    {
+      "epoch": 3.8627387034822624,
+      "grad_norm": 2.411869764328003,
+      "learning_rate": 1.138570807363386e-05,
+      "loss": 0.3418,
+      "step": 26650
+    },
+    {
+      "epoch": 3.8699858680291337,
+      "grad_norm": 3.126814603805542,
+      "learning_rate": 1.131323380200029e-05,
+      "loss": 0.3334,
+      "step": 26700
+    },
+    {
+      "epoch": 3.8772330325760045,
+      "grad_norm": 3.4210116863250732,
+      "learning_rate": 1.124075953036672e-05,
+      "loss": 0.3249,
+      "step": 26750
+    },
+    {
+      "epoch": 3.8844801971228757,
+      "grad_norm": 2.846679925918579,
+      "learning_rate": 1.1168285258733151e-05,
+      "loss": 0.3377,
+      "step": 26800
+    },
+    {
+      "epoch": 3.891727361669747,
+      "grad_norm": 3.338003635406494,
+      "learning_rate": 1.1095810987099581e-05,
+      "loss": 0.3626,
+      "step": 26850
+    },
+    {
+      "epoch": 3.8989745262166178,
+      "grad_norm": 3.7777626514434814,
+      "learning_rate": 1.102333671546601e-05,
+      "loss": 0.3385,
+      "step": 26900
+    },
+    {
+      "epoch": 3.9062216907634886,
+      "grad_norm": 2.5645010471343994,
+      "learning_rate": 1.095086244383244e-05,
+      "loss": 0.3485,
+      "step": 26950
+    },
+    {
+      "epoch": 3.91346885531036,
+      "grad_norm": 2.8242435455322266,
+      "learning_rate": 1.0878388172198869e-05,
+      "loss": 0.353,
+      "step": 27000
+    },
+    {
+      "epoch": 3.91346885531036,
+      "eval_loss": 0.6270226836204529,
+      "eval_runtime": 245.4756,
+      "eval_samples_per_second": 140.523,
+      "eval_steps_per_second": 14.054,
+      "step": 27000
+    },
+    {
+      "epoch": 3.920716019857231,
+      "grad_norm": 3.0427374839782715,
+      "learning_rate": 1.0805913900565299e-05,
+      "loss": 0.3625,
+      "step": 27050
+    },
+    {
+      "epoch": 3.927963184404102,
+      "grad_norm": 3.5145153999328613,
+      "learning_rate": 1.073343962893173e-05,
+      "loss": 0.3377,
+      "step": 27100
+    },
+    {
+      "epoch": 3.9352103489509727,
+      "grad_norm": 3.0157294273376465,
+      "learning_rate": 1.066096535729816e-05,
+      "loss": 0.3231,
+      "step": 27150
+    },
+    {
+      "epoch": 3.942457513497844,
+      "grad_norm": 3.3504700660705566,
+      "learning_rate": 1.058849108566459e-05,
+      "loss": 0.3407,
+      "step": 27200
+    },
+    {
+      "epoch": 3.949704678044715,
+      "grad_norm": 2.313544988632202,
+      "learning_rate": 1.051601681403102e-05,
+      "loss": 0.3252,
+      "step": 27250
+    },
+    {
+      "epoch": 3.956951842591586,
+      "grad_norm": 3.034682035446167,
+      "learning_rate": 1.044354254239745e-05,
+      "loss": 0.319,
+      "step": 27300
+    },
+    {
+      "epoch": 3.964199007138457,
+      "grad_norm": 2.8075735569000244,
+      "learning_rate": 1.037106827076388e-05,
+      "loss": 0.3371,
+      "step": 27350
+    },
+    {
+      "epoch": 3.971446171685328,
+      "grad_norm": 2.394465446472168,
+      "learning_rate": 1.0298593999130309e-05,
+      "loss": 0.3286,
+      "step": 27400
+    },
+    {
+      "epoch": 3.9786933362321992,
+      "grad_norm": 2.6049180030822754,
+      "learning_rate": 1.0226119727496738e-05,
+      "loss": 0.3295,
+      "step": 27450
+    },
+    {
+      "epoch": 3.98594050077907,
+      "grad_norm": 2.6557512283325195,
+      "learning_rate": 1.0153645455863168e-05,
+      "loss": 0.3566,
+      "step": 27500
+    },
+    {
+      "epoch": 3.9931876653259413,
+      "grad_norm": 2.997840166091919,
+      "learning_rate": 1.0081171184229598e-05,
+      "loss": 0.326,
+      "step": 27550
+    },
+    {
+      "epoch": 4.0004348298728125,
+      "grad_norm": 2.4651620388031006,
+      "learning_rate": 1.0008696912596029e-05,
+      "loss": 0.3328,
+      "step": 27600
+    },
+    {
+      "epoch": 4.007681994419683,
+      "grad_norm": 2.93027400970459,
+      "learning_rate": 9.936222640962459e-06,
+      "loss": 0.2705,
+      "step": 27650
+    },
+    {
+      "epoch": 4.014929158966554,
+      "grad_norm": 3.154695987701416,
+      "learning_rate": 9.863748369328889e-06,
+      "loss": 0.251,
+      "step": 27700
+    },
+    {
+      "epoch": 4.022176323513426,
+      "grad_norm": 2.877485990524292,
+      "learning_rate": 9.79127409769532e-06,
+      "loss": 0.2557,
+      "step": 27750
+    },
+    {
+      "epoch": 4.029423488060297,
+      "grad_norm": 2.5868325233459473,
+      "learning_rate": 9.71879982606175e-06,
+      "loss": 0.2481,
+      "step": 27800
+    },
+    {
+      "epoch": 4.036670652607167,
+      "grad_norm": 4.68599271774292,
+      "learning_rate": 9.646325554428178e-06,
+      "loss": 0.2704,
+      "step": 27850
+    },
+    {
+      "epoch": 4.043917817154038,
+      "grad_norm": 2.302772045135498,
+      "learning_rate": 9.573851282794608e-06,
+      "loss": 0.2537,
+      "step": 27900
+    },
+    {
+      "epoch": 4.05116498170091,
+      "grad_norm": 2.2476446628570557,
+      "learning_rate": 9.501377011161039e-06,
+      "loss": 0.2488,
+      "step": 27950
+    },
+    {
+      "epoch": 4.058412146247781,
+      "grad_norm": 1.8352832794189453,
+      "learning_rate": 9.428902739527467e-06,
+      "loss": 0.2693,
+      "step": 28000
+    },
+    {
+      "epoch": 4.058412146247781,
+      "eval_loss": 0.6433804631233215,
+      "eval_runtime": 245.0452,
+      "eval_samples_per_second": 140.77,
+      "eval_steps_per_second": 14.079,
+      "step": 28000
+    },
+    {
+      "epoch": 4.0656593107946515,
+      "grad_norm": 2.861711263656616,
+      "learning_rate": 9.356428467893898e-06,
+      "loss": 0.2552,
+      "step": 28050
+    },
+    {
+      "epoch": 4.072906475341522,
+      "grad_norm": 2.669003963470459,
+      "learning_rate": 9.283954196260328e-06,
+      "loss": 0.2737,
+      "step": 28100
+    },
+    {
+      "epoch": 4.080153639888394,
+      "grad_norm": 3.106980323791504,
+      "learning_rate": 9.211479924626758e-06,
+      "loss": 0.265,
+      "step": 28150
+    },
+    {
+      "epoch": 4.087400804435265,
+      "grad_norm": 2.7630670070648193,
+      "learning_rate": 9.139005652993188e-06,
+      "loss": 0.2457,
+      "step": 28200
+    },
+    {
+      "epoch": 4.094647968982136,
+      "grad_norm": 2.5765066146850586,
+      "learning_rate": 9.066531381359619e-06,
+      "loss": 0.2815,
+      "step": 28250
+    },
+    {
+      "epoch": 4.101895133529006,
+      "grad_norm": 2.980583667755127,
+      "learning_rate": 8.994057109726047e-06,
+      "loss": 0.2763,
+      "step": 28300
+    },
+    {
+      "epoch": 4.109142298075878,
+      "grad_norm": 2.6509013175964355,
+      "learning_rate": 8.921582838092477e-06,
+      "loss": 0.2454,
+      "step": 28350
+    },
+    {
+      "epoch": 4.116389462622749,
+      "grad_norm": 2.8553245067596436,
+      "learning_rate": 8.849108566458908e-06,
+      "loss": 0.2749,
+      "step": 28400
+    },
+    {
+      "epoch": 4.12363662716962,
+      "grad_norm": 2.524636745452881,
+      "learning_rate": 8.776634294825338e-06,
+      "loss": 0.2537,
+      "step": 28450
+    },
+    {
+      "epoch": 4.130883791716491,
+      "grad_norm": 3.848393440246582,
+      "learning_rate": 8.704160023191767e-06,
+      "loss": 0.2635,
+      "step": 28500
+    },
+    {
+      "epoch": 4.138130956263362,
+      "grad_norm": 3.296485662460327,
+      "learning_rate": 8.631685751558197e-06,
+      "loss": 0.2499,
+      "step": 28550
+    },
+    {
+      "epoch": 4.145378120810233,
+      "grad_norm": 3.0012335777282715,
+      "learning_rate": 8.559211479924627e-06,
+      "loss": 0.2698,
+      "step": 28600
+    },
+    {
+      "epoch": 4.152625285357104,
+      "grad_norm": 1.961544156074524,
+      "learning_rate": 8.486737208291057e-06,
+      "loss": 0.23,
+      "step": 28650
+    },
+    {
+      "epoch": 4.159872449903975,
+      "grad_norm": 3.157874822616577,
+      "learning_rate": 8.414262936657488e-06,
+      "loss": 0.2493,
+      "step": 28700
+    },
+    {
+      "epoch": 4.167119614450846,
+      "grad_norm": 2.372300624847412,
+      "learning_rate": 8.341788665023916e-06,
+      "loss": 0.2874,
+      "step": 28750
+    },
+    {
+      "epoch": 4.174366778997717,
+      "grad_norm": 1.9763847589492798,
+      "learning_rate": 8.269314393390346e-06,
+      "loss": 0.2508,
+      "step": 28800
+    },
+    {
+      "epoch": 4.181613943544588,
+      "grad_norm": 3.00111985206604,
+      "learning_rate": 8.196840121756777e-06,
+      "loss": 0.2732,
+      "step": 28850
+    },
+    {
+      "epoch": 4.1888611080914595,
+      "grad_norm": 3.2600185871124268,
+      "learning_rate": 8.124365850123207e-06,
+      "loss": 0.2718,
+      "step": 28900
+    },
+    {
+      "epoch": 4.19610827263833,
+      "grad_norm": 2.755221366882324,
+      "learning_rate": 8.051891578489637e-06,
+      "loss": 0.2401,
+      "step": 28950
+    },
+    {
+      "epoch": 4.203355437185201,
+      "grad_norm": 3.3103065490722656,
+      "learning_rate": 7.979417306856067e-06,
+      "loss": 0.2898,
+      "step": 29000
+    },
+    {
+      "epoch": 4.203355437185201,
+      "eval_loss": 0.6491243243217468,
+      "eval_runtime": 245.4343,
+      "eval_samples_per_second": 140.547,
+      "eval_steps_per_second": 14.057,
+      "step": 29000
+    },
+    {
+      "epoch": 4.210602601732072,
+      "grad_norm": 2.783529281616211,
+      "learning_rate": 7.906943035222496e-06,
+      "loss": 0.2774,
+      "step": 29050
+    },
+    {
+      "epoch": 4.217849766278944,
+      "grad_norm": 2.3917746543884277,
+      "learning_rate": 7.834468763588926e-06,
+      "loss": 0.2896,
+      "step": 29100
+    },
+    {
+      "epoch": 4.225096930825814,
+      "grad_norm": 3.132794141769409,
+      "learning_rate": 7.761994491955357e-06,
+      "loss": 0.2704,
+      "step": 29150
+    },
+    {
+      "epoch": 4.232344095372685,
+      "grad_norm": 2.8275017738342285,
+      "learning_rate": 7.689520220321785e-06,
+      "loss": 0.2695,
+      "step": 29200
+    },
+    {
+      "epoch": 4.239591259919557,
+      "grad_norm": 3.1233084201812744,
+      "learning_rate": 7.617045948688216e-06,
+      "loss": 0.2677,
+      "step": 29250
+    },
+    {
+      "epoch": 4.246838424466428,
+      "grad_norm": 3.1787428855895996,
+      "learning_rate": 7.544571677054646e-06,
+      "loss": 0.2611,
+      "step": 29300
+    },
+    {
+      "epoch": 4.2540855890132985,
+      "grad_norm": 3.4065091609954834,
+      "learning_rate": 7.472097405421076e-06,
+      "loss": 0.285,
+      "step": 29350
+    },
+    {
+      "epoch": 4.261332753560169,
+      "grad_norm": 2.7599704265594482,
+      "learning_rate": 7.399623133787506e-06,
+      "loss": 0.2586,
+      "step": 29400
+    },
+    {
+      "epoch": 4.268579918107041,
+      "grad_norm": 2.2776358127593994,
+      "learning_rate": 7.3271488621539365e-06,
+      "loss": 0.2532,
+      "step": 29450
+    },
+    {
+      "epoch": 4.275827082653912,
+      "grad_norm": 3.391362428665161,
+      "learning_rate": 7.254674590520366e-06,
+      "loss": 0.257,
+      "step": 29500
+    },
+    {
+      "epoch": 4.283074247200783,
+      "grad_norm": 4.152310371398926,
+      "learning_rate": 7.182200318886796e-06,
+      "loss": 0.2655,
+      "step": 29550
+    },
+    {
+      "epoch": 4.290321411747653,
+      "grad_norm": 1.8384218215942383,
+      "learning_rate": 7.109726047253225e-06,
+      "loss": 0.2556,
+      "step": 29600
+    },
+    {
+      "epoch": 4.297568576294525,
+      "grad_norm": 2.8006668090820312,
+      "learning_rate": 7.037251775619655e-06,
+      "loss": 0.252,
+      "step": 29650
+    },
+    {
+      "epoch": 4.304815740841396,
+      "grad_norm": 2.0686655044555664,
+      "learning_rate": 6.964777503986085e-06,
+      "loss": 0.2626,
+      "step": 29700
+    },
+    {
+      "epoch": 4.312062905388267,
+      "grad_norm": 4.304172515869141,
+      "learning_rate": 6.892303232352515e-06,
+      "loss": 0.2566,
+      "step": 29750
+    },
+    {
+      "epoch": 4.319310069935138,
+      "grad_norm": 3.3372154235839844,
+      "learning_rate": 6.819828960718945e-06,
+      "loss": 0.256,
+      "step": 29800
+    },
+    {
+      "epoch": 4.326557234482009,
+      "grad_norm": 2.2065439224243164,
+      "learning_rate": 6.747354689085375e-06,
+      "loss": 0.251,
+      "step": 29850
+    },
+    {
+      "epoch": 4.33380439902888,
+      "grad_norm": 3.629650354385376,
+      "learning_rate": 6.6748804174518055e-06,
+      "loss": 0.2696,
+      "step": 29900
+    },
+    {
+      "epoch": 4.341051563575751,
+      "grad_norm": 2.2397236824035645,
+      "learning_rate": 6.602406145818235e-06,
+      "loss": 0.2359,
+      "step": 29950
+    },
+    {
+      "epoch": 4.348298728122622,
+      "grad_norm": 3.494893789291382,
+      "learning_rate": 6.529931874184665e-06,
+      "loss": 0.2714,
+      "step": 30000
+    },
+    {
+      "epoch": 4.348298728122622,
+      "eval_loss": 0.6459131836891174,
+      "eval_runtime": 245.3544,
+      "eval_samples_per_second": 140.593,
+      "eval_steps_per_second": 14.061,
+      "step": 30000
+    },
+    {
+      "epoch": 4.355545892669493,
+      "grad_norm": 2.9454917907714844,
+      "learning_rate": 6.4574576025510955e-06,
+      "loss": 0.2539,
+      "step": 30050
+    },
+    {
+      "epoch": 4.362793057216364,
+      "grad_norm": 2.9951882362365723,
+      "learning_rate": 6.384983330917525e-06,
+      "loss": 0.2687,
+      "step": 30100
+    },
+    {
+      "epoch": 4.370040221763235,
+      "grad_norm": 3.583976984024048,
+      "learning_rate": 6.312509059283954e-06,
+      "loss": 0.2879,
+      "step": 30150
+    },
+    {
+      "epoch": 4.3772873863101065,
+      "grad_norm": 3.201929807662964,
+      "learning_rate": 6.2400347876503846e-06,
+      "loss": 0.2531,
+      "step": 30200
+    },
+    {
+      "epoch": 4.384534550856977,
+      "grad_norm": 2.60980486869812,
+      "learning_rate": 6.167560516016814e-06,
+      "loss": 0.2681,
+      "step": 30250
+    },
+    {
+      "epoch": 4.391781715403848,
+      "grad_norm": 2.1033682823181152,
+      "learning_rate": 6.095086244383244e-06,
+      "loss": 0.2774,
+      "step": 30300
+    },
+    {
+      "epoch": 4.399028879950719,
+      "grad_norm": 2.239474058151245,
+      "learning_rate": 6.022611972749674e-06,
+      "loss": 0.2676,
+      "step": 30350
+    },
+    {
+      "epoch": 4.406276044497591,
+      "grad_norm": 3.176302671432495,
+      "learning_rate": 5.950137701116104e-06,
+      "loss": 0.2722,
+      "step": 30400
+    },
+    {
+      "epoch": 4.413523209044461,
+      "grad_norm": 2.5901739597320557,
+      "learning_rate": 5.877663429482534e-06,
+      "loss": 0.2523,
+      "step": 30450
+    },
+    {
+      "epoch": 4.420770373591332,
+      "grad_norm": 2.8084895610809326,
+      "learning_rate": 5.8051891578489645e-06,
+      "loss": 0.2487,
+      "step": 30500
+    },
+    {
+      "epoch": 4.428017538138203,
+      "grad_norm": 3.332167387008667,
+      "learning_rate": 5.732714886215394e-06,
+      "loss": 0.2523,
+      "step": 30550
+    },
+    {
+      "epoch": 4.435264702685075,
+      "grad_norm": 2.872776746749878,
+      "learning_rate": 5.660240614581823e-06,
+      "loss": 0.2739,
+      "step": 30600
+    },
+    {
+      "epoch": 4.4425118672319455,
+      "grad_norm": 1.9077197313308716,
+      "learning_rate": 5.5877663429482536e-06,
+      "loss": 0.2476,
+      "step": 30650
+    },
+    {
+      "epoch": 4.449759031778816,
+      "grad_norm": 2.676182270050049,
+      "learning_rate": 5.515292071314684e-06,
+      "loss": 0.2579,
+      "step": 30700
+    },
+    {
+      "epoch": 4.457006196325688,
+      "grad_norm": 3.4579455852508545,
+      "learning_rate": 5.442817799681114e-06,
+      "loss": 0.2686,
+      "step": 30750
+    },
+    {
+      "epoch": 4.464253360872559,
+      "grad_norm": 2.6098556518554688,
+      "learning_rate": 5.370343528047543e-06,
+      "loss": 0.2578,
+      "step": 30800
+    },
+    {
+      "epoch": 4.4715005254194296,
+      "grad_norm": 3.4452645778656006,
+      "learning_rate": 5.297869256413973e-06,
+      "loss": 0.2868,
+      "step": 30850
+    },
+    {
+      "epoch": 4.4787476899663,
+      "grad_norm": 3.4186015129089355,
+      "learning_rate": 5.225394984780403e-06,
+      "loss": 0.2676,
+      "step": 30900
+    },
+    {
+      "epoch": 4.485994854513172,
+      "grad_norm": 3.0700855255126953,
+      "learning_rate": 5.1529207131468335e-06,
+      "loss": 0.2642,
+      "step": 30950
+    },
+    {
+      "epoch": 4.493242019060043,
+      "grad_norm": 2.718798875808716,
+      "learning_rate": 5.080446441513263e-06,
+      "loss": 0.2488,
+      "step": 31000
+    },
+    {
+      "epoch": 4.493242019060043,
+      "eval_loss": 0.6449950337409973,
+      "eval_runtime": 245.4001,
+      "eval_samples_per_second": 140.566,
+      "eval_steps_per_second": 14.059,
+      "step": 31000
+    },
+    {
+      "epoch": 4.500489183606914,
+      "grad_norm": 3.136319875717163,
+      "learning_rate": 5.007972169879692e-06,
+      "loss": 0.2831,
+      "step": 31050
+    },
+    {
+      "epoch": 4.507736348153784,
+      "grad_norm": 2.020862340927124,
+      "learning_rate": 4.935497898246123e-06,
+      "loss": 0.2312,
+      "step": 31100
+    },
+    {
+      "epoch": 4.514983512700656,
+      "grad_norm": 4.0119948387146,
+      "learning_rate": 4.863023626612553e-06,
+      "loss": 0.2583,
+      "step": 31150
+    },
+    {
+      "epoch": 4.522230677247527,
+      "grad_norm": 3.341949462890625,
+      "learning_rate": 4.790549354978983e-06,
+      "loss": 0.2777,
+      "step": 31200
+    },
+    {
+      "epoch": 4.529477841794398,
+      "grad_norm": 3.6001293659210205,
+      "learning_rate": 4.7180750833454125e-06,
+      "loss": 0.251,
+      "step": 31250
+    },
+    {
+      "epoch": 4.536725006341269,
+      "grad_norm": 2.4010775089263916,
+      "learning_rate": 4.645600811711843e-06,
+      "loss": 0.2679,
+      "step": 31300
+    },
+    {
+      "epoch": 4.54397217088814,
+      "grad_norm": 2.33186674118042,
+      "learning_rate": 4.573126540078272e-06,
+      "loss": 0.2702,
+      "step": 31350
+    },
+    {
+      "epoch": 4.551219335435011,
+      "grad_norm": 3.366321325302124,
+      "learning_rate": 4.5006522684447025e-06,
+      "loss": 0.2383,
+      "step": 31400
+    },
+    {
+      "epoch": 4.558466499981882,
+      "grad_norm": 2.606224298477173,
+      "learning_rate": 4.428177996811132e-06,
+      "loss": 0.2709,
+      "step": 31450
+    },
+    {
+      "epoch": 4.5657136645287535,
+      "grad_norm": 3.850285053253174,
+      "learning_rate": 4.355703725177562e-06,
+      "loss": 0.2712,
+      "step": 31500
+    },
+    {
+      "epoch": 4.572960829075624,
+      "grad_norm": 2.4807302951812744,
+      "learning_rate": 4.2832294535439924e-06,
+      "loss": 0.267,
+      "step": 31550
+    },
+    {
+      "epoch": 4.580207993622495,
+      "grad_norm": 1.4953159093856812,
+      "learning_rate": 4.210755181910422e-06,
+      "loss": 0.2599,
+      "step": 31600
+    },
+    {
+      "epoch": 4.587455158169366,
+      "grad_norm": 2.457629680633545,
+      "learning_rate": 4.138280910276852e-06,
+      "loss": 0.2802,
+      "step": 31650
+    },
+    {
+      "epoch": 4.594702322716238,
+      "grad_norm": 2.150555372238159,
+      "learning_rate": 4.0658066386432815e-06,
+      "loss": 0.2564,
+      "step": 31700
+    },
+    {
+      "epoch": 4.601949487263108,
+      "grad_norm": 1.8131722211837769,
+      "learning_rate": 3.993332367009712e-06,
+      "loss": 0.2805,
+      "step": 31750
+    },
+    {
+      "epoch": 4.609196651809979,
+      "grad_norm": 3.3967912197113037,
+      "learning_rate": 3.920858095376142e-06,
+      "loss": 0.2812,
+      "step": 31800
+    },
+    {
+      "epoch": 4.616443816356851,
+      "grad_norm": 2.5398590564727783,
+      "learning_rate": 3.8483838237425715e-06,
+      "loss": 0.2787,
+      "step": 31850
+    },
+    {
+      "epoch": 4.623690980903722,
+      "grad_norm": 2.45865535736084,
+      "learning_rate": 3.7759095521090013e-06,
+      "loss": 0.2333,
+      "step": 31900
+    },
+    {
+      "epoch": 4.6309381454505925,
+      "grad_norm": 3.3966212272644043,
+      "learning_rate": 3.704884765908103e-06,
+      "loss": 0.2619,
+      "step": 31950
+    },
+    {
+      "epoch": 4.638185309997463,
+      "grad_norm": 2.421985149383545,
+      "learning_rate": 3.6324104942745326e-06,
+      "loss": 0.2851,
+      "step": 32000
+    },
+    {
+      "epoch": 4.638185309997463,
+      "eval_loss": 0.6421298980712891,
+      "eval_runtime": 245.3379,
+      "eval_samples_per_second": 140.602,
+      "eval_steps_per_second": 14.062,
+      "step": 32000
+    },
+    {
+      "epoch": 4.645432474544334,
+      "grad_norm": 2.5105528831481934,
+      "learning_rate": 3.5599362226409624e-06,
+      "loss": 0.2595,
+      "step": 32050
+    },
+    {
+      "epoch": 4.652679639091206,
+      "grad_norm": 2.653074026107788,
+      "learning_rate": 3.4874619510073923e-06,
+      "loss": 0.2478,
+      "step": 32100
+    },
+    {
+      "epoch": 4.6599268036380765,
+      "grad_norm": 1.6157690286636353,
+      "learning_rate": 3.4149876793738225e-06,
+      "loss": 0.2563,
+      "step": 32150
+    },
+    {
+      "epoch": 4.667173968184947,
+      "grad_norm": 2.220090866088867,
+      "learning_rate": 3.3425134077402524e-06,
+      "loss": 0.2825,
+      "step": 32200
+    },
+    {
+      "epoch": 4.674421132731819,
+      "grad_norm": 3.191338062286377,
+      "learning_rate": 3.2700391361066826e-06,
+      "loss": 0.2421,
+      "step": 32250
+    },
+    {
+      "epoch": 4.68166829727869,
+      "grad_norm": 2.6636569499969482,
+      "learning_rate": 3.1990143499057836e-06,
+      "loss": 0.274,
+      "step": 32300
+    },
+    {
+      "epoch": 4.688915461825561,
+      "grad_norm": 2.6303908824920654,
+      "learning_rate": 3.126540078272214e-06,
+      "loss": 0.2763,
+      "step": 32350
+    },
+    {
+      "epoch": 4.696162626372431,
+      "grad_norm": 2.2153165340423584,
+      "learning_rate": 3.0540658066386433e-06,
+      "loss": 0.2512,
+      "step": 32400
+    },
+    {
+      "epoch": 4.703409790919303,
+      "grad_norm": 2.291551351547241,
+      "learning_rate": 2.9815915350050736e-06,
+      "loss": 0.2739,
+      "step": 32450
+    },
+    {
+      "epoch": 4.710656955466174,
+      "grad_norm": 3.2897346019744873,
+      "learning_rate": 2.909117263371503e-06,
+      "loss": 0.239,
+      "step": 32500
+    },
+    {
+      "epoch": 4.717904120013045,
+      "grad_norm": 3.1026740074157715,
+      "learning_rate": 2.8366429917379333e-06,
+      "loss": 0.2806,
+      "step": 32550
+    },
+    {
+      "epoch": 4.7251512845599155,
+      "grad_norm": 4.166581153869629,
+      "learning_rate": 2.764168720104363e-06,
+      "loss": 0.2597,
+      "step": 32600
+    },
+    {
+      "epoch": 4.732398449106787,
+      "grad_norm": 3.2309772968292236,
+      "learning_rate": 2.691694448470793e-06,
+      "loss": 0.2748,
+      "step": 32650
+    },
+    {
+      "epoch": 4.739645613653658,
+      "grad_norm": 3.379218816757202,
+      "learning_rate": 2.619220176837223e-06,
+      "loss": 0.2665,
+      "step": 32700
+    },
+    {
+      "epoch": 4.746892778200529,
+      "grad_norm": 2.8024582862854004,
+      "learning_rate": 2.5467459052036526e-06,
+      "loss": 0.2571,
+      "step": 32750
+    },
+    {
+      "epoch": 4.7541399427474005,
+      "grad_norm": 1.905375361442566,
+      "learning_rate": 2.474271633570083e-06,
+      "loss": 0.2547,
+      "step": 32800
+    },
+    {
+      "epoch": 4.761387107294271,
+      "grad_norm": 2.8878672122955322,
+      "learning_rate": 2.4017973619365127e-06,
+      "loss": 0.2633,
+      "step": 32850
+    },
+    {
+      "epoch": 4.768634271841142,
+      "grad_norm": 2.940661907196045,
+      "learning_rate": 2.3293230903029426e-06,
+      "loss": 0.2752,
+      "step": 32900
+    },
+    {
+      "epoch": 4.775881436388013,
+      "grad_norm": 2.747434139251709,
+      "learning_rate": 2.2568488186693724e-06,
+      "loss": 0.2762,
+      "step": 32950
+    },
+    {
+      "epoch": 4.783128600934885,
+      "grad_norm": 3.6419451236724854,
+      "learning_rate": 2.1843745470358027e-06,
+      "loss": 0.2685,
+      "step": 33000
+    },
+    {
+      "epoch": 4.783128600934885,
+      "eval_loss": 0.6418930888175964,
+      "eval_runtime": 245.4849,
+      "eval_samples_per_second": 140.518,
+      "eval_steps_per_second": 14.054,
+      "step": 33000
+    },
+    {
+      "epoch": 4.790375765481755,
+      "grad_norm": 2.916613817214966,
+      "learning_rate": 2.111900275402232e-06,
+      "loss": 0.2846,
+      "step": 33050
+    },
+    {
+      "epoch": 4.797622930028626,
+      "grad_norm": 2.95576548576355,
+      "learning_rate": 2.0394260037686624e-06,
+      "loss": 0.2626,
+      "step": 33100
+    },
+    {
+      "epoch": 4.804870094575497,
+      "grad_norm": 2.053476333618164,
+      "learning_rate": 1.966951732135092e-06,
+      "loss": 0.2619,
+      "step": 33150
+    },
+    {
+      "epoch": 4.812117259122369,
+      "grad_norm": 2.7688095569610596,
+      "learning_rate": 1.894477460501522e-06,
+      "loss": 0.2974,
+      "step": 33200
+    },
+    {
+      "epoch": 4.8193644236692395,
+      "grad_norm": 2.67800235748291,
+      "learning_rate": 1.8220031888679521e-06,
+      "loss": 0.2676,
+      "step": 33250
+    },
+    {
+      "epoch": 4.82661158821611,
+      "grad_norm": 3.279421806335449,
+      "learning_rate": 1.7495289172343818e-06,
+      "loss": 0.2689,
+      "step": 33300
+    },
+    {
+      "epoch": 4.833858752762982,
+      "grad_norm": 1.616542100906372,
+      "learning_rate": 1.6770546456008118e-06,
+      "loss": 0.2555,
+      "step": 33350
+    },
+    {
+      "epoch": 4.841105917309853,
+      "grad_norm": 3.103170156478882,
+      "learning_rate": 1.6045803739672419e-06,
+      "loss": 0.2551,
+      "step": 33400
+    },
+    {
+      "epoch": 4.8483530818567235,
+      "grad_norm": 2.5930793285369873,
+      "learning_rate": 1.5321061023336715e-06,
+      "loss": 0.2618,
+      "step": 33450
+    },
+    {
+      "epoch": 4.855600246403594,
+      "grad_norm": 3.2237420082092285,
+      "learning_rate": 1.4596318307001016e-06,
+      "loss": 0.2461,
+      "step": 33500
+    },
+    {
+      "epoch": 4.862847410950465,
+      "grad_norm": 2.2981555461883545,
+      "learning_rate": 1.3871575590665314e-06,
+      "loss": 0.2707,
+      "step": 33550
+    },
+    {
+      "epoch": 4.870094575497337,
+      "grad_norm": 1.4708250761032104,
+      "learning_rate": 1.3146832874329614e-06,
+      "loss": 0.254,
+      "step": 33600
+    },
+    {
+      "epoch": 4.877341740044208,
+      "grad_norm": 2.699856758117676,
+      "learning_rate": 1.2422090157993913e-06,
+      "loss": 0.2593,
+      "step": 33650
+    },
+    {
+      "epoch": 4.884588904591078,
+      "grad_norm": 2.0948593616485596,
+      "learning_rate": 1.1697347441658213e-06,
+      "loss": 0.2583,
+      "step": 33700
+    },
+    {
+      "epoch": 4.89183606913795,
+      "grad_norm": 2.787429094314575,
+      "learning_rate": 1.0972604725322512e-06,
+      "loss": 0.2381,
+      "step": 33750
+    },
+    {
+      "epoch": 4.899083233684821,
+      "grad_norm": 2.4767441749572754,
+      "learning_rate": 1.024786200898681e-06,
+      "loss": 0.2474,
+      "step": 33800
+    },
+    {
+      "epoch": 4.906330398231692,
+      "grad_norm": 3.3083810806274414,
+      "learning_rate": 9.523119292651109e-07,
+      "loss": 0.2647,
+      "step": 33850
+    },
+    {
+      "epoch": 4.9135775627785625,
+      "grad_norm": 2.8774940967559814,
+      "learning_rate": 8.798376576315409e-07,
+      "loss": 0.2622,
+      "step": 33900
+    },
+    {
+      "epoch": 4.920824727325434,
+      "grad_norm": 2.5657265186309814,
+      "learning_rate": 8.073633859979708e-07,
+      "loss": 0.2537,
+      "step": 33950
+    },
+    {
+      "epoch": 4.928071891872305,
+      "grad_norm": 1.964735984802246,
+      "learning_rate": 7.348891143644006e-07,
+      "loss": 0.2646,
+      "step": 34000
+    },
+    {
+      "epoch": 4.928071891872305,
+      "eval_loss": 0.6408438682556152,
+      "eval_runtime": 245.4737,
+      "eval_samples_per_second": 140.524,
+      "eval_steps_per_second": 14.054,
+      "step": 34000
+    },
+    {
+      "epoch": 4.935319056419176,
+      "grad_norm": 2.6487998962402344,
+      "learning_rate": 6.624148427308306e-07,
+      "loss": 0.2799,
+      "step": 34050
+    },
+    {
+      "epoch": 4.942566220966047,
+      "grad_norm": 2.05784010887146,
+      "learning_rate": 5.899405710972605e-07,
+      "loss": 0.2657,
+      "step": 34100
+    },
+    {
+      "epoch": 4.949813385512918,
+      "grad_norm": 2.4890284538269043,
+      "learning_rate": 5.174662994636905e-07,
+      "loss": 0.2477,
+      "step": 34150
+    },
+    {
+      "epoch": 4.957060550059789,
+      "grad_norm": 2.277297258377075,
+      "learning_rate": 4.449920278301203e-07,
+      "loss": 0.2664,
+      "step": 34200
+    },
+    {
+      "epoch": 4.96430771460666,
+      "grad_norm": 3.3281660079956055,
+      "learning_rate": 3.7251775619655025e-07,
+      "loss": 0.2582,
+      "step": 34250
+    },
+    {
+      "epoch": 4.971554879153532,
+      "grad_norm": 3.9353535175323486,
+      "learning_rate": 3.0004348456298015e-07,
+      "loss": 0.2622,
+      "step": 34300
+    },
+    {
+      "epoch": 4.978802043700402,
+      "grad_norm": 2.5661704540252686,
+      "learning_rate": 2.275692129294101e-07,
+      "loss": 0.2562,
+      "step": 34350
+    },
+    {
+      "epoch": 4.986049208247273,
+      "grad_norm": 3.4376327991485596,
+      "learning_rate": 1.5509494129583997e-07,
+      "loss": 0.2647,
+      "step": 34400
+    },
+    {
+      "epoch": 4.993296372794144,
+      "grad_norm": 3.3473730087280273,
+      "learning_rate": 8.26206696622699e-08,
+      "loss": 0.2737,
+      "step": 34450
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 34495,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.5981403722758554e+18,
+  "train_batch_size": 10,
+  "trial_name": null,
+  "trial_params": null
+}