diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7821 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 1000,
+  "global_step": 100000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.8100409507751465,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 1.816,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0648651123046875,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 1.5968,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.9017549753189087,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 1.56,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5334885120391846,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 1.587,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3036648035049438,
+      "learning_rate": 0.0003,
+      "loss": 1.6182,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.709660530090332,
+      "learning_rate": 0.00029969849246231153,
+      "loss": 1.6102,
+      "step": 600
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.5684775114059448,
+      "learning_rate": 0.0002993969849246231,
+      "loss": 1.6094,
+      "step": 700
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.3330438137054443,
+      "learning_rate": 0.00029909547738693465,
+      "loss": 1.6118,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1563549041748047,
+      "learning_rate": 0.0002987939698492462,
+      "loss": 1.6596,
+      "step": 900
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4043567180633545,
+      "learning_rate": 0.00029849547738693464,
+      "loss": 1.6071,
+      "step": 1000
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 1.585342288017273,
+      "eval_runtime": 37.6462,
+      "eval_samples_per_second": 26.563,
+      "eval_steps_per_second": 3.32,
+      "step": 1000
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3647234439849854,
+      "learning_rate": 0.0002981939698492462,
+      "loss": 1.611,
+      "step": 1100
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3917016983032227,
+      "learning_rate": 0.00029789246231155776,
+      "loss": 1.6003,
+      "step": 1200
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.7931370735168457,
+      "learning_rate": 0.0002975909547738693,
+      "loss": 1.5789,
+      "step": 1300
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.542971611022949,
+      "learning_rate": 0.0002972894472361809,
+      "loss": 1.5435,
+      "step": 1400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8555421829223633,
+      "learning_rate": 0.00029698793969849243,
+      "loss": 1.5513,
+      "step": 1500
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.9988830089569092,
+      "learning_rate": 0.000296686432160804,
+      "loss": 1.5763,
+      "step": 1600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.5328696966171265,
+      "learning_rate": 0.00029638492462311555,
+      "loss": 1.5529,
+      "step": 1700
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.442533254623413,
+      "learning_rate": 0.0002960834170854271,
+      "loss": 1.5581,
+      "step": 1800
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.4188216924667358,
+      "learning_rate": 0.00029578190954773867,
+      "loss": 1.5598,
+      "step": 1900
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.700873851776123,
+      "learning_rate": 0.00029548040201005023,
+      "loss": 1.6091,
+      "step": 2000
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 1.5680323839187622,
+      "eval_runtime": 37.9632,
+      "eval_samples_per_second": 26.341,
+      "eval_steps_per_second": 3.293,
+      "step": 2000
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.415462493896484,
+      "learning_rate": 0.0002951788944723618,
+      "loss": 1.5435,
+      "step": 2100
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5002624988555908,
+      "learning_rate": 0.00029487738693467335,
+      "loss": 1.5485,
+      "step": 2200
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8552610874176025,
+      "learning_rate": 0.0002945758793969849,
+      "loss": 1.5687,
+      "step": 2300
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.6914422512054443,
+      "learning_rate": 0.00029427437185929647,
+      "loss": 1.5549,
+      "step": 2400
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.5994210243225098,
+      "learning_rate": 0.00029397286432160803,
+      "loss": 1.5541,
+      "step": 2500
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9448769092559814,
+      "learning_rate": 0.0002936713567839196,
+      "loss": 1.5348,
+      "step": 2600
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3909597396850586,
+      "learning_rate": 0.00029336984924623115,
+      "loss": 1.5629,
+      "step": 2700
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4517822265625,
+      "learning_rate": 0.0002930683417085427,
+      "loss": 1.4946,
+      "step": 2800
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7407867908477783,
+      "learning_rate": 0.0002927668341708542,
+      "loss": 1.568,
+      "step": 2900
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.3732205629348755,
+      "learning_rate": 0.0002924653266331658,
+      "loss": 1.4928,
+      "step": 3000
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 1.5172981023788452,
+      "eval_runtime": 37.8358,
+      "eval_samples_per_second": 26.43,
+      "eval_steps_per_second": 3.304,
+      "step": 3000
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9255911111831665,
+      "learning_rate": 0.0002921638190954774,
+      "loss": 1.5208,
+      "step": 3100
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7328695058822632,
+      "learning_rate": 0.00029186231155778895,
+      "loss": 1.5442,
+      "step": 3200
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.286285400390625,
+      "learning_rate": 0.00029156080402010045,
+      "loss": 1.5071,
+      "step": 3300
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.426595687866211,
+      "learning_rate": 0.000291259296482412,
+      "loss": 1.5424,
+      "step": 3400
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8213595151901245,
+      "learning_rate": 0.0002909577889447236,
+      "loss": 1.487,
+      "step": 3500
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4181461334228516,
+      "learning_rate": 0.000290659296482412,
+      "loss": 1.5083,
+      "step": 3600
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.4696974754333496,
+      "learning_rate": 0.0002903577889447236,
+      "loss": 1.5204,
+      "step": 3700
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.285097360610962,
+      "learning_rate": 0.00029005628140703517,
+      "loss": 1.515,
+      "step": 3800
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.7307722568511963,
+      "learning_rate": 0.00028975477386934673,
+      "loss": 1.5283,
+      "step": 3900
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.5405428409576416,
+      "learning_rate": 0.00028945326633165823,
+      "loss": 1.4657,
+      "step": 4000
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 1.4836663007736206,
+      "eval_runtime": 37.7733,
+      "eval_samples_per_second": 26.474,
+      "eval_steps_per_second": 3.309,
+      "step": 4000
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2221779823303223,
+      "learning_rate": 0.00028915175879396985,
+      "loss": 1.4936,
+      "step": 4100
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.700119733810425,
+      "learning_rate": 0.0002888502512562814,
+      "loss": 1.446,
+      "step": 4200
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.11588716506958,
+      "learning_rate": 0.0002885487437185929,
+      "loss": 1.4789,
+      "step": 4300
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.144611358642578,
+      "learning_rate": 0.00028824723618090447,
+      "loss": 1.4913,
+      "step": 4400
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7891815900802612,
+      "learning_rate": 0.0002879457286432161,
+      "loss": 1.4693,
+      "step": 4500
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2549595832824707,
+      "learning_rate": 0.0002876442211055276,
+      "loss": 1.4957,
+      "step": 4600
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4034409523010254,
+      "learning_rate": 0.00028734271356783915,
+      "loss": 1.4909,
+      "step": 4700
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.4686906337738037,
+      "learning_rate": 0.0002870412060301507,
+      "loss": 1.4989,
+      "step": 4800
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1314849853515625,
+      "learning_rate": 0.0002867396984924623,
+      "loss": 1.4899,
+      "step": 4900
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.703493595123291,
+      "learning_rate": 0.00028643819095477383,
+      "loss": 1.4897,
+      "step": 5000
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 1.5144654512405396,
+      "eval_runtime": 38.0015,
+      "eval_samples_per_second": 26.315,
+      "eval_steps_per_second": 3.289,
+      "step": 5000
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.8537943363189697,
+      "learning_rate": 0.0002861366834170854,
+      "loss": 1.4702,
+      "step": 5100
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.885312557220459,
+      "learning_rate": 0.00028583517587939695,
+      "loss": 1.4918,
+      "step": 5200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.6149489879608154,
+      "learning_rate": 0.0002855336683417085,
+      "loss": 1.4867,
+      "step": 5300
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8222806453704834,
+      "learning_rate": 0.00028523216080402007,
+      "loss": 1.4894,
+      "step": 5400
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.105160713195801,
+      "learning_rate": 0.0002849306532663316,
+      "loss": 1.4865,
+      "step": 5500
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9180357456207275,
+      "learning_rate": 0.0002846291457286432,
+      "loss": 1.4365,
+      "step": 5600
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.4675670862197876,
+      "learning_rate": 0.00028432763819095474,
+      "loss": 1.4323,
+      "step": 5700
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.664919376373291,
+      "learning_rate": 0.0002840261306532663,
+      "loss": 1.4605,
+      "step": 5800
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5559368133544922,
+      "learning_rate": 0.00028372462311557786,
+      "loss": 1.4799,
+      "step": 5900
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0738680362701416,
+      "learning_rate": 0.0002834261306532663,
+      "loss": 1.4923,
+      "step": 6000
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 1.4727822542190552,
+      "eval_runtime": 38.2425,
+      "eval_samples_per_second": 26.149,
+      "eval_steps_per_second": 3.269,
+      "step": 6000
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9228754043579102,
+      "learning_rate": 0.00028312462311557785,
+      "loss": 1.4127,
+      "step": 6100
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0438356399536133,
+      "learning_rate": 0.0002828231155778894,
+      "loss": 1.4835,
+      "step": 6200
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.734626293182373,
+      "learning_rate": 0.00028252160804020097,
+      "loss": 1.4489,
+      "step": 6300
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1490132808685303,
+      "learning_rate": 0.0002822201005025125,
+      "loss": 1.4684,
+      "step": 6400
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1819868087768555,
+      "learning_rate": 0.0002819185929648241,
+      "loss": 1.4416,
+      "step": 6500
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.5763262510299683,
+      "learning_rate": 0.00028161708542713565,
+      "loss": 1.4532,
+      "step": 6600
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9584680795669556,
+      "learning_rate": 0.0002813155778894472,
+      "loss": 1.4558,
+      "step": 6700
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6148059368133545,
+      "learning_rate": 0.00028101407035175876,
+      "loss": 1.4588,
+      "step": 6800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5689460039138794,
+      "learning_rate": 0.0002807125628140703,
+      "loss": 1.4352,
+      "step": 6900
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.145756483078003,
+      "learning_rate": 0.0002804110552763819,
+      "loss": 1.4207,
+      "step": 7000
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 1.4386738538742065,
+      "eval_runtime": 38.107,
+      "eval_samples_per_second": 26.242,
+      "eval_steps_per_second": 3.28,
+      "step": 7000
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.316162586212158,
+      "learning_rate": 0.00028010954773869344,
+      "loss": 1.4085,
+      "step": 7100
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.0866541862487793,
+      "learning_rate": 0.000279808040201005,
+      "loss": 1.4634,
+      "step": 7200
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.0577406883239746,
+      "learning_rate": 0.00027950653266331656,
+      "loss": 1.4515,
+      "step": 7300
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.723168969154358,
+      "learning_rate": 0.0002792050251256281,
+      "loss": 1.4372,
+      "step": 7400
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.8033313751220703,
+      "learning_rate": 0.0002789035175879397,
+      "loss": 1.4844,
+      "step": 7500
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.051619529724121,
+      "learning_rate": 0.00027860201005025124,
+      "loss": 1.4352,
+      "step": 7600
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4199312925338745,
+      "learning_rate": 0.0002783005025125628,
+      "loss": 1.4641,
+      "step": 7700
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3949058055877686,
+      "learning_rate": 0.00027799899497487436,
+      "loss": 1.4592,
+      "step": 7800
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.8449528217315674,
+      "learning_rate": 0.0002776974874371859,
+      "loss": 1.4196,
+      "step": 7900
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.709972858428955,
+      "learning_rate": 0.0002773959798994975,
+      "loss": 1.4375,
+      "step": 8000
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 1.4270827770233154,
+      "eval_runtime": 38.3346,
+      "eval_samples_per_second": 26.086,
+      "eval_steps_per_second": 3.261,
+      "step": 8000
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7984100580215454,
+      "learning_rate": 0.00027709447236180904,
+      "loss": 1.3943,
+      "step": 8100
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.1693639755249023,
+      "learning_rate": 0.00027679597989949746,
+      "loss": 1.4636,
+      "step": 8200
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8211654424667358,
+      "learning_rate": 0.000276494472361809,
+      "loss": 1.4539,
+      "step": 8300
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.11051869392395,
+      "learning_rate": 0.0002761929648241206,
+      "loss": 1.4214,
+      "step": 8400
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5553231239318848,
+      "learning_rate": 0.00027589145728643214,
+      "loss": 1.4475,
+      "step": 8500
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0080809593200684,
+      "learning_rate": 0.0002755899497487437,
+      "loss": 1.4024,
+      "step": 8600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.6698598861694336,
+      "learning_rate": 0.00027528844221105526,
+      "loss": 1.4159,
+      "step": 8700
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2336277961730957,
+      "learning_rate": 0.0002749869346733668,
+      "loss": 1.437,
+      "step": 8800
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7006186246871948,
+      "learning_rate": 0.0002746854271356784,
+      "loss": 1.4465,
+      "step": 8900
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.934051513671875,
+      "learning_rate": 0.0002743839195979899,
+      "loss": 1.4319,
+      "step": 9000
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 1.4331704378128052,
+      "eval_runtime": 37.9595,
+      "eval_samples_per_second": 26.344,
+      "eval_steps_per_second": 3.293,
+      "step": 9000
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.549532890319824,
+      "learning_rate": 0.0002740824120603015,
+      "loss": 1.4018,
+      "step": 9100
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9921625852584839,
+      "learning_rate": 0.00027378090452261306,
+      "loss": 1.4354,
+      "step": 9200
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.5784940719604492,
+      "learning_rate": 0.0002734793969849246,
+      "loss": 1.4515,
+      "step": 9300
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9822384119033813,
+      "learning_rate": 0.0002731778894472361,
+      "loss": 1.4784,
+      "step": 9400
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.0514814853668213,
+      "learning_rate": 0.00027287638190954774,
+      "loss": 1.4235,
+      "step": 9500
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.5947296619415283,
+      "learning_rate": 0.0002725748743718593,
+      "loss": 1.4325,
+      "step": 9600
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.838723659515381,
+      "learning_rate": 0.0002722733668341708,
+      "loss": 1.4318,
+      "step": 9700
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.7525815963745117,
+      "learning_rate": 0.00027197185929648236,
+      "loss": 1.4323,
+      "step": 9800
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.186182975769043,
+      "learning_rate": 0.000271670351758794,
+      "loss": 1.4122,
+      "step": 9900
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.5111092329025269,
+      "learning_rate": 0.00027136884422110553,
+      "loss": 1.4278,
+      "step": 10000
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 1.4226535558700562,
+      "eval_runtime": 37.925,
+      "eval_samples_per_second": 26.368,
+      "eval_steps_per_second": 3.296,
+      "step": 10000
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.4402307271957397,
+      "learning_rate": 0.00027106733668341704,
+      "loss": 1.4775,
+      "step": 10100
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.803475379943848,
+      "learning_rate": 0.0002707658291457286,
+      "loss": 1.4434,
+      "step": 10200
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.159541606903076,
+      "learning_rate": 0.0002704643216080402,
+      "loss": 1.4505,
+      "step": 10300
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.613765835762024,
+      "learning_rate": 0.0002701658291457286,
+      "loss": 1.4336,
+      "step": 10400
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.0653555393218994,
+      "learning_rate": 0.0002698643216080402,
+      "loss": 1.4238,
+      "step": 10500
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0688183307647705,
+      "learning_rate": 0.00026956281407035176,
+      "loss": 1.4048,
+      "step": 10600
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.271068572998047,
+      "learning_rate": 0.0002692613065326633,
+      "loss": 1.4412,
+      "step": 10700
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7365072965621948,
+      "learning_rate": 0.0002689597989949748,
+      "loss": 1.3864,
+      "step": 10800
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7095474004745483,
+      "learning_rate": 0.00026865829145728643,
+      "loss": 1.4509,
+      "step": 10900
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.595015287399292,
+      "learning_rate": 0.000268356783919598,
+      "loss": 1.4068,
+      "step": 11000
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 1.4620698690414429,
+      "eval_runtime": 37.8254,
+      "eval_samples_per_second": 26.437,
+      "eval_steps_per_second": 3.305,
+      "step": 11000
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.6796025037765503,
+      "learning_rate": 0.0002680552763819095,
+      "loss": 1.4059,
+      "step": 11100
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.259477376937866,
+      "learning_rate": 0.00026775376884422106,
+      "loss": 1.4112,
+      "step": 11200
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.8005051612854,
+      "learning_rate": 0.00026745226130653267,
+      "loss": 1.367,
+      "step": 11300
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.824021577835083,
+      "learning_rate": 0.00026715075376884423,
+      "loss": 1.4156,
+      "step": 11400
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4818904399871826,
+      "learning_rate": 0.00026684924623115574,
+      "loss": 1.3846,
+      "step": 11500
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6064958572387695,
+      "learning_rate": 0.0002665477386934673,
+      "loss": 1.4062,
+      "step": 11600
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8354562520980835,
+      "learning_rate": 0.00026624623115577886,
+      "loss": 1.3761,
+      "step": 11700
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.094172477722168,
+      "learning_rate": 0.0002659447236180904,
+      "loss": 1.3576,
+      "step": 11800
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.000718832015991,
+      "learning_rate": 0.000265643216080402,
+      "loss": 1.401,
+      "step": 11900
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.301866054534912,
+      "learning_rate": 0.00026534170854271353,
+      "loss": 1.4267,
+      "step": 12000
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 1.4072773456573486,
+      "eval_runtime": 37.8474,
+      "eval_samples_per_second": 26.422,
+      "eval_steps_per_second": 3.303,
+      "step": 12000
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8116004467010498,
+      "learning_rate": 0.0002650402010050251,
+      "loss": 1.4141,
+      "step": 12100
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7951298952102661,
+      "learning_rate": 0.00026473869346733665,
+      "loss": 1.4006,
+      "step": 12200
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9248169660568237,
+      "learning_rate": 0.0002644371859296482,
+      "loss": 1.4143,
+      "step": 12300
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.0492172241210938,
+      "learning_rate": 0.00026413567839195977,
+      "loss": 1.3808,
+      "step": 12400
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.3698550462722778,
+      "learning_rate": 0.00026383417085427133,
+      "loss": 1.339,
+      "step": 12500
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.8333966732025146,
+      "learning_rate": 0.0002635326633165829,
+      "loss": 1.3977,
+      "step": 12600
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5511767864227295,
+      "learning_rate": 0.0002632341708542713,
+      "loss": 1.4027,
+      "step": 12700
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.912987470626831,
+      "learning_rate": 0.0002629326633165829,
+      "loss": 1.4062,
+      "step": 12800
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8692814111709595,
+      "learning_rate": 0.00026263115577889444,
+      "loss": 1.3901,
+      "step": 12900
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.620612859725952,
+      "learning_rate": 0.000262329648241206,
+      "loss": 1.3992,
+      "step": 13000
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 1.3693994283676147,
+      "eval_runtime": 38.004,
+      "eval_samples_per_second": 26.313,
+      "eval_steps_per_second": 3.289,
+      "step": 13000
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.1771810054779053,
+      "learning_rate": 0.00026202814070351756,
+      "loss": 1.3733,
+      "step": 13100
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4650421142578125,
+      "learning_rate": 0.0002617266331658291,
+      "loss": 1.399,
+      "step": 13200
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.9789535999298096,
+      "learning_rate": 0.0002614251256281407,
+      "loss": 1.4291,
+      "step": 13300
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.4404784440994263,
+      "learning_rate": 0.00026112361809045223,
+      "loss": 1.3833,
+      "step": 13400
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0667450428009033,
+      "learning_rate": 0.0002608221105527638,
+      "loss": 1.3884,
+      "step": 13500
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.014460563659668,
+      "learning_rate": 0.00026052060301507535,
+      "loss": 1.3819,
+      "step": 13600
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.360121965408325,
+      "learning_rate": 0.0002602190954773869,
+      "loss": 1.3695,
+      "step": 13700
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.6982303857803345,
+      "learning_rate": 0.00025991758793969847,
+      "loss": 1.3864,
+      "step": 13800
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2350399494171143,
+      "learning_rate": 0.00025961608040201003,
+      "loss": 1.4096,
+      "step": 13900
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.4647042751312256,
+      "learning_rate": 0.0002593145728643216,
+      "loss": 1.3915,
+      "step": 14000
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 1.3878337144851685,
+      "eval_runtime": 37.7254,
+      "eval_samples_per_second": 26.507,
+      "eval_steps_per_second": 3.313,
+      "step": 14000
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.002542734146118,
+      "learning_rate": 0.00025901306532663315,
+      "loss": 1.4214,
+      "step": 14100
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9857007265090942,
+      "learning_rate": 0.0002587115577889447,
+      "loss": 1.3636,
+      "step": 14200
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4016737937927246,
+      "learning_rate": 0.00025841005025125627,
+      "loss": 1.4259,
+      "step": 14300
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.929931879043579,
+      "learning_rate": 0.0002581085427135678,
+      "loss": 1.3937,
+      "step": 14400
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.6266632080078125,
+      "learning_rate": 0.0002578070351758794,
+      "loss": 1.3678,
+      "step": 14500
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.905378580093384,
+      "learning_rate": 0.00025750552763819095,
+      "loss": 1.3526,
+      "step": 14600
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.535842180252075,
+      "learning_rate": 0.0002572040201005025,
+      "loss": 1.4062,
+      "step": 14700
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.5988209247589111,
+      "learning_rate": 0.000256902512562814,
+      "loss": 1.3915,
+      "step": 14800
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.5643303394317627,
+      "learning_rate": 0.0002566010050251256,
+      "loss": 1.3783,
+      "step": 14900
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.4297415018081665,
+      "learning_rate": 0.0002562994974874372,
+      "loss": 1.3782,
+      "step": 15000
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 1.405114769935608,
+      "eval_runtime": 37.9898,
+      "eval_samples_per_second": 26.323,
+      "eval_steps_per_second": 3.29,
+      "step": 15000
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.6650172472000122,
+      "learning_rate": 0.0002559979899497487,
+      "loss": 1.3387,
+      "step": 15100
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.118579864501953,
+      "learning_rate": 0.00025569648241206025,
+      "loss": 1.393,
+      "step": 15200
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.74748694896698,
+      "learning_rate": 0.00025539497487437186,
+      "loss": 1.3353,
+      "step": 15300
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.794631004333496,
+      "learning_rate": 0.0002550934673366834,
+      "loss": 1.3942,
+      "step": 15400
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7065675258636475,
+      "learning_rate": 0.00025479195979899493,
+      "loss": 1.3962,
+      "step": 15500
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.389014720916748,
+      "learning_rate": 0.0002544904522613065,
+      "loss": 1.3758,
+      "step": 15600
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.534252405166626,
+      "learning_rate": 0.0002541889447236181,
+      "loss": 1.3526,
+      "step": 15700
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.7374197244644165,
+      "learning_rate": 0.0002538874371859296,
+      "loss": 1.3577,
+      "step": 15800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.1230342388153076,
+      "learning_rate": 0.00025358592964824117,
+      "loss": 1.3548,
+      "step": 15900
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.261570692062378,
+      "learning_rate": 0.0002532844221105527,
+      "loss": 1.3932,
+      "step": 16000
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 1.3275749683380127,
+      "eval_runtime": 37.9493,
+      "eval_samples_per_second": 26.351,
+      "eval_steps_per_second": 3.294,
+      "step": 16000
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0108933448791504,
+      "learning_rate": 0.00025298291457286434,
+      "loss": 1.3445,
+      "step": 16100
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.536722421646118,
+      "learning_rate": 0.00025268140703517584,
+      "loss": 1.364,
+      "step": 16200
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.637465238571167,
+      "learning_rate": 0.0002523829145728643,
+      "loss": 1.376,
+      "step": 16300
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.8907904624938965,
+      "learning_rate": 0.0002520814070351759,
+      "loss": 1.3623,
+      "step": 16400
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4385364055633545,
+      "learning_rate": 0.0002517798994974874,
+      "loss": 1.318,
+      "step": 16500
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.9113733768463135,
+      "learning_rate": 0.00025147839195979895,
+      "loss": 1.3906,
+      "step": 16600
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.8118414878845215,
+      "learning_rate": 0.00025117688442211056,
+      "loss": 1.3336,
+      "step": 16700
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.9629586935043335,
+      "learning_rate": 0.0002508753768844221,
+      "loss": 1.3959,
+      "step": 16800
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.0420243740081787,
+      "learning_rate": 0.0002505738693467336,
+      "loss": 1.3523,
+      "step": 16900
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.0758414268493652,
+      "learning_rate": 0.0002502723618090452,
+      "loss": 1.3747,
+      "step": 17000
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 1.3606867790222168,
+      "eval_runtime": 37.9681,
+      "eval_samples_per_second": 26.338,
+      "eval_steps_per_second": 3.292,
+      "step": 17000
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.486980438232422,
+      "learning_rate": 0.00024997085427135675,
+      "loss": 1.3402,
+      "step": 17100
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.211982250213623,
+      "learning_rate": 0.0002496693467336683,
+      "loss": 1.3419,
+      "step": 17200
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3362228870391846,
+      "learning_rate": 0.00024936783919597986,
+      "loss": 1.3748,
+      "step": 17300
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.515100121498108,
+      "learning_rate": 0.0002490663316582914,
+      "loss": 1.3747,
+      "step": 17400
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.1747968196868896,
+      "learning_rate": 0.000248764824120603,
+      "loss": 1.3458,
+      "step": 17500
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.6045758724212646,
+      "learning_rate": 0.00024846331658291454,
+      "loss": 1.3623,
+      "step": 17600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.5456433296203613,
+      "learning_rate": 0.0002481618090452261,
+      "loss": 1.3107,
+      "step": 17700
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.5310312509536743,
+      "learning_rate": 0.00024786030150753766,
+      "loss": 1.3541,
+      "step": 17800
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2094223499298096,
+      "learning_rate": 0.0002475587939698492,
+      "loss": 1.3445,
+      "step": 17900
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7595880031585693,
+      "learning_rate": 0.0002472572864321608,
+      "loss": 1.3537,
+      "step": 18000
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 1.3503804206848145,
+      "eval_runtime": 37.8049,
+      "eval_samples_per_second": 26.452,
+      "eval_steps_per_second": 3.306,
+      "step": 18000
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.4382781982421875,
+      "learning_rate": 0.00024695577889447234,
+      "loss": 1.3584,
+      "step": 18100
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7903175354003906,
+      "learning_rate": 0.0002466542713567839,
+      "loss": 1.3272,
+      "step": 18200
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.6171114444732666,
+      "learning_rate": 0.00024635276381909546,
+      "loss": 1.3601,
+      "step": 18300
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9426279067993164,
+      "learning_rate": 0.000246051256281407,
+      "loss": 1.3782,
+      "step": 18400
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.36596941947937,
+      "learning_rate": 0.0002457497487437186,
+      "loss": 1.3307,
+      "step": 18500
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.3205448389053345,
+      "learning_rate": 0.00024544824120603014,
+      "loss": 1.3929,
+      "step": 18600
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.9464951753616333,
+      "learning_rate": 0.0002451467336683417,
+      "loss": 1.3415,
+      "step": 18700
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.7700294256210327,
+      "learning_rate": 0.00024484522613065326,
+      "loss": 1.3473,
+      "step": 18800
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.687060832977295,
+      "learning_rate": 0.0002445437185929648,
+      "loss": 1.3606,
+      "step": 18900
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.02754282951355,
+      "learning_rate": 0.0002442422110552764,
+      "loss": 1.3799,
+      "step": 19000
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 1.365315556526184,
+      "eval_runtime": 37.6707,
+      "eval_samples_per_second": 26.546,
+      "eval_steps_per_second": 3.318,
+      "step": 19000
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.187087059020996,
+      "learning_rate": 0.0002439407035175879,
+      "loss": 1.3585,
+      "step": 19100
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.8181040287017822,
+      "learning_rate": 0.00024363919597989947,
+      "loss": 1.3723,
+      "step": 19200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.6949020624160767,
+      "learning_rate": 0.00024333768844221105,
+      "loss": 1.3074,
+      "step": 19300
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.716754913330078,
+      "learning_rate": 0.00024303618090452259,
+      "loss": 1.3589,
+      "step": 19400
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.5216838121414185,
+      "learning_rate": 0.00024273467336683415,
+      "loss": 1.3398,
+      "step": 19500
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.7370058298110962,
+      "learning_rate": 0.0002424331658291457,
+      "loss": 1.3546,
+      "step": 19600
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.0907745361328125,
+      "learning_rate": 0.00024213165829145726,
+      "loss": 1.3161,
+      "step": 19700
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9564626216888428,
+      "learning_rate": 0.00024183015075376882,
+      "loss": 1.3623,
+      "step": 19800
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6082723140716553,
+      "learning_rate": 0.00024152864321608038,
+      "loss": 1.3158,
+      "step": 19900
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.0046592950820923,
+      "learning_rate": 0.00024122713567839192,
+      "loss": 1.3366,
+      "step": 20000
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 1.3484834432601929,
+      "eval_runtime": 37.9475,
+      "eval_samples_per_second": 26.352,
+      "eval_steps_per_second": 3.294,
+      "step": 20000
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.5935070514678955,
+      "learning_rate": 0.0002409256281407035,
+      "loss": 1.3512,
+      "step": 20100
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 3.790050506591797,
+      "learning_rate": 0.00024062412060301506,
+      "loss": 1.3272,
+      "step": 20200
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.3440461158752441,
+      "learning_rate": 0.00024032562814070351,
+      "loss": 1.333,
+      "step": 20300
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 6.51857852935791,
+      "learning_rate": 0.00024002412060301505,
+      "loss": 1.3334,
+      "step": 20400
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.882919192314148,
+      "learning_rate": 0.0002397226130653266,
+      "loss": 1.3241,
+      "step": 20500
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.361558198928833,
+      "learning_rate": 0.00023942110552763817,
+      "loss": 1.3207,
+      "step": 20600
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 2.0967071056365967,
+      "learning_rate": 0.00023911959798994975,
+      "loss": 1.2993,
+      "step": 20700
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.2517688274383545,
+      "learning_rate": 0.00023881809045226128,
+      "loss": 1.3353,
+      "step": 20800
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 7.7647480964660645,
+      "learning_rate": 0.00023851658291457284,
+      "loss": 1.3326,
+      "step": 20900
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.0270638465881348,
+      "learning_rate": 0.0002382180904522613,
+      "loss": 1.3046,
+      "step": 21000
+    },
+    {
+      "epoch": 1.05,
+      "eval_loss": 1.3456777334213257,
+      "eval_runtime": 38.0868,
+      "eval_samples_per_second": 26.256,
+      "eval_steps_per_second": 3.282,
+      "step": 21000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.9642785787582397,
+      "learning_rate": 0.00023791658291457283,
+      "loss": 1.3131,
+      "step": 21100
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.517357587814331,
+      "learning_rate": 0.0002376150753768844,
+      "loss": 1.3627,
+      "step": 21200
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.4660860300064087,
+      "learning_rate": 0.00023731356783919598,
+      "loss": 1.2805,
+      "step": 21300
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 3.102552652359009,
+      "learning_rate": 0.00023701206030150753,
+      "loss": 1.339,
+      "step": 21400
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.017504930496216,
+      "learning_rate": 0.00023671055276381907,
+      "loss": 1.3307,
+      "step": 21500
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.4260824918746948,
+      "learning_rate": 0.00023640904522613063,
+      "loss": 1.3216,
+      "step": 21600
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 4.0052361488342285,
+      "learning_rate": 0.0002361075376884422,
+      "loss": 1.3544,
+      "step": 21700
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 3.664625883102417,
+      "learning_rate": 0.00023580603015075375,
+      "loss": 1.3508,
+      "step": 21800
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 2.1044421195983887,
+      "learning_rate": 0.0002355045226130653,
+      "loss": 1.3205,
+      "step": 21900
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.6608549356460571,
+      "learning_rate": 0.00023520301507537686,
+      "loss": 1.3373,
+      "step": 22000
+    },
+    {
+      "epoch": 1.1,
+      "eval_loss": 1.319154977798462,
+      "eval_runtime": 37.7789,
+      "eval_samples_per_second": 26.47,
+      "eval_steps_per_second": 3.309,
+      "step": 22000
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.131612777709961,
+      "learning_rate": 0.00023490150753768845,
+      "loss": 1.3244,
+      "step": 22100
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.0854969024658203,
+      "learning_rate": 0.00023459999999999998,
+      "loss": 1.3357,
+      "step": 22200
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.3622310161590576,
+      "learning_rate": 0.0002343075376884422,
+      "loss": 1.4118,
+      "step": 22300
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.5198066234588623,
+      "learning_rate": 0.00023400603015075376,
+      "loss": 1.319,
+      "step": 22400
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.4654555320739746,
+      "learning_rate": 0.00023370452261306532,
+      "loss": 1.3055,
+      "step": 22500
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.53120756149292,
+      "learning_rate": 0.00023340301507537685,
+      "loss": 1.3763,
+      "step": 22600
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 2.199324131011963,
+      "learning_rate": 0.00023310150753768843,
+      "loss": 1.3148,
+      "step": 22700
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 2.951871633529663,
+      "learning_rate": 0.0002328,
+      "loss": 1.3234,
+      "step": 22800
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 2.5513529777526855,
+      "learning_rate": 0.00023249849246231153,
+      "loss": 1.302,
+      "step": 22900
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 5.096097469329834,
+      "learning_rate": 0.00023219698492462309,
+      "loss": 1.3102,
+      "step": 23000
+    },
+    {
+      "epoch": 1.15,
+      "eval_loss": 1.3704819679260254,
+      "eval_runtime": 37.8283,
+      "eval_samples_per_second": 26.435,
+      "eval_steps_per_second": 3.304,
+      "step": 23000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.3565678596496582,
+      "learning_rate": 0.00023189547738693467,
+      "loss": 1.3182,
+      "step": 23100
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 3.1972274780273438,
+      "learning_rate": 0.00023159396984924623,
+      "loss": 1.316,
+      "step": 23200
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.4728245735168457,
+      "learning_rate": 0.00023129246231155776,
+      "loss": 1.2934,
+      "step": 23300
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.917893648147583,
+      "learning_rate": 0.00023099095477386932,
+      "loss": 1.3241,
+      "step": 23400
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 2.30876088142395,
+      "learning_rate": 0.00023068944723618086,
+      "loss": 1.3031,
+      "step": 23500
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 2.5653178691864014,
+      "learning_rate": 0.00023038793969849244,
+      "loss": 1.2819,
+      "step": 23600
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 3.500821352005005,
+      "learning_rate": 0.000230086432160804,
+      "loss": 1.2829,
+      "step": 23700
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.6564580202102661,
+      "learning_rate": 0.00022978492462311556,
+      "loss": 1.3209,
+      "step": 23800
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.6477315425872803,
+      "learning_rate": 0.0002294834170854271,
+      "loss": 1.2991,
+      "step": 23900
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.9583780765533447,
+      "learning_rate": 0.00022918190954773868,
+      "loss": 1.3011,
+      "step": 24000
+    },
+    {
+      "epoch": 1.2,
+      "eval_loss": 1.3160556554794312,
+      "eval_runtime": 37.7643,
+      "eval_samples_per_second": 26.48,
+      "eval_steps_per_second": 3.31,
+      "step": 24000
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.3997368812561035,
+      "learning_rate": 0.00022888040201005024,
+      "loss": 1.2866,
+      "step": 24100
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.5909266471862793,
+      "learning_rate": 0.00022857889447236177,
+      "loss": 1.3133,
+      "step": 24200
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.9457557201385498,
+      "learning_rate": 0.00022827738693467333,
+      "loss": 1.2716,
+      "step": 24300
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 2.85856032371521,
+      "learning_rate": 0.00022797587939698492,
+      "loss": 1.2932,
+      "step": 24400
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 3.180671215057373,
+      "learning_rate": 0.00022767437185929648,
+      "loss": 1.317,
+      "step": 24500
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.630612850189209,
+      "learning_rate": 0.000227372864321608,
+      "loss": 1.3176,
+      "step": 24600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.159804582595825,
+      "learning_rate": 0.00022707135678391957,
+      "loss": 1.3288,
+      "step": 24700
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.314036250114441,
+      "learning_rate": 0.00022676984924623116,
+      "loss": 1.3157,
+      "step": 24800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.718198776245117,
+      "learning_rate": 0.0002264683417085427,
+      "loss": 1.2915,
+      "step": 24900
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.3423640727996826,
+      "learning_rate": 0.00022616683417085425,
+      "loss": 1.2976,
+      "step": 25000
+    },
+    {
+      "epoch": 1.25,
+      "eval_loss": 1.3594353199005127,
+      "eval_runtime": 37.7829,
+      "eval_samples_per_second": 26.467,
+      "eval_steps_per_second": 3.308,
+      "step": 25000
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.3341753482818604,
+      "learning_rate": 0.0002258653266331658,
+      "loss": 1.322,
+      "step": 25100
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.0798075199127197,
+      "learning_rate": 0.0002255638190954774,
+      "loss": 1.3182,
+      "step": 25200
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.5256847143173218,
+      "learning_rate": 0.00022526231155778893,
+      "loss": 1.3102,
+      "step": 25300
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 2.4831185340881348,
+      "learning_rate": 0.00022496080402010049,
+      "loss": 1.3183,
+      "step": 25400
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 9.853681564331055,
+      "learning_rate": 0.00022465929648241204,
+      "loss": 1.2963,
+      "step": 25500
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.833552837371826,
+      "learning_rate": 0.00022435778894472358,
+      "loss": 1.3226,
+      "step": 25600
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.7486400604248047,
+      "learning_rate": 0.00022405628140703516,
+      "loss": 1.2742,
+      "step": 25700
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.3708908557891846,
+      "learning_rate": 0.00022375477386934672,
+      "loss": 1.2878,
+      "step": 25800
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 3.6677916049957275,
+      "learning_rate": 0.00022345326633165826,
+      "loss": 1.3113,
+      "step": 25900
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 2.7909395694732666,
+      "learning_rate": 0.00022315175879396981,
+      "loss": 1.3221,
+      "step": 26000
+    },
+    {
+      "epoch": 1.3,
+      "eval_loss": 1.313453197479248,
+      "eval_runtime": 37.7782,
+      "eval_samples_per_second": 26.47,
+      "eval_steps_per_second": 3.309,
+      "step": 26000
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 2.592221736907959,
+      "learning_rate": 0.0002228502512562814,
+      "loss": 1.2918,
+      "step": 26100
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.911118984222412,
+      "learning_rate": 0.00022254874371859296,
+      "loss": 1.3392,
+      "step": 26200
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 2.15328049659729,
+      "learning_rate": 0.0002222472361809045,
+      "loss": 1.261,
+      "step": 26300
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 3.0731029510498047,
+      "learning_rate": 0.00022194572864321605,
+      "loss": 1.289,
+      "step": 26400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 3.032560348510742,
+      "learning_rate": 0.00022164422110552764,
+      "loss": 1.3186,
+      "step": 26500
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 5.388736724853516,
+      "learning_rate": 0.00022134271356783917,
+      "loss": 1.3214,
+      "step": 26600
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 2.6400022506713867,
+      "learning_rate": 0.00022104120603015073,
+      "loss": 1.2936,
+      "step": 26700
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 3.9355711936950684,
+      "learning_rate": 0.0002207396984924623,
+      "loss": 1.3039,
+      "step": 26800
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.6818647384643555,
+      "learning_rate": 0.00022043819095477388,
+      "loss": 1.2992,
+      "step": 26900
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 2.2356157302856445,
+      "learning_rate": 0.0002201366834170854,
+      "loss": 1.3011,
+      "step": 27000
+    },
+    {
+      "epoch": 1.35,
+      "eval_loss": 1.3157364130020142,
+      "eval_runtime": 37.9238,
+      "eval_samples_per_second": 26.369,
+      "eval_steps_per_second": 3.296,
+      "step": 27000
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 2.158803701400757,
+      "learning_rate": 0.00021983517587939697,
+      "loss": 1.308,
+      "step": 27100
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.4748259782791138,
+      "learning_rate": 0.0002195336683417085,
+      "loss": 1.2873,
+      "step": 27200
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 2.382047653198242,
+      "learning_rate": 0.0002192321608040201,
+      "loss": 1.2795,
+      "step": 27300
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.8785953521728516,
+      "learning_rate": 0.00021893065326633165,
+      "loss": 1.3101,
+      "step": 27400
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 2.4842770099639893,
+      "learning_rate": 0.0002186291457286432,
+      "loss": 1.3124,
+      "step": 27500
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.7258535623550415,
+      "learning_rate": 0.00021832763819095474,
+      "loss": 1.3315,
+      "step": 27600
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.157860517501831,
+      "learning_rate": 0.00021802613065326633,
+      "loss": 1.2848,
+      "step": 27700
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 3.1965837478637695,
+      "learning_rate": 0.00021772462311557788,
+      "loss": 1.3105,
+      "step": 27800
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 3.141603708267212,
+      "learning_rate": 0.00021742311557788942,
+      "loss": 1.3197,
+      "step": 27900
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 2.0368692874908447,
+      "learning_rate": 0.00021712160804020098,
+      "loss": 1.3113,
+      "step": 28000
+    },
+    {
+      "epoch": 1.4,
+      "eval_loss": 1.3079107999801636,
+      "eval_runtime": 37.8037,
+      "eval_samples_per_second": 26.452,
+      "eval_steps_per_second": 3.307,
+      "step": 28000
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 3.013373851776123,
+      "learning_rate": 0.00021682010050251254,
+      "loss": 1.2892,
+      "step": 28100
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 2.766491651535034,
+      "learning_rate": 0.00021651859296482412,
+      "loss": 1.3414,
+      "step": 28200
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.6288301944732666,
+      "learning_rate": 0.00021621708542713566,
+      "loss": 1.3156,
+      "step": 28300
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 2.3904545307159424,
+      "learning_rate": 0.00021591557788944721,
+      "loss": 1.2905,
+      "step": 28400
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.263744831085205,
+      "learning_rate": 0.00021561407035175877,
+      "loss": 1.2961,
+      "step": 28500
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.985129714012146,
+      "learning_rate": 0.00021531256281407033,
+      "loss": 1.2703,
+      "step": 28600
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.4574270248413086,
+      "learning_rate": 0.0002150110552763819,
+      "loss": 1.2793,
+      "step": 28700
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.312525510787964,
+      "learning_rate": 0.00021470954773869345,
+      "loss": 1.2669,
+      "step": 28800
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.5253132581710815,
+      "learning_rate": 0.00021440804020100498,
+      "loss": 1.3187,
+      "step": 28900
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.7550122737884521,
+      "learning_rate": 0.00021410653266331657,
+      "loss": 1.3154,
+      "step": 29000
+    },
+    {
+      "epoch": 1.45,
+      "eval_loss": 1.2937275171279907,
+      "eval_runtime": 37.9639,
+      "eval_samples_per_second": 26.341,
+      "eval_steps_per_second": 3.293,
+      "step": 29000
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 2.492000102996826,
+      "learning_rate": 0.00021380502512562813,
+      "loss": 1.2868,
+      "step": 29100
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 4.013311862945557,
+      "learning_rate": 0.00021350351758793966,
+      "loss": 1.2578,
+      "step": 29200
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 3.991748809814453,
+      "learning_rate": 0.00021320201005025122,
+      "loss": 1.3347,
+      "step": 29300
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 4.655180931091309,
+      "learning_rate": 0.0002129005025125628,
+      "loss": 1.2935,
+      "step": 29400
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.9497921466827393,
+      "learning_rate": 0.00021259899497487437,
+      "loss": 1.248,
+      "step": 29500
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 3.372061252593994,
+      "learning_rate": 0.0002122974874371859,
+      "loss": 1.2877,
+      "step": 29600
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 2.1920547485351562,
+      "learning_rate": 0.00021199597989949746,
+      "loss": 1.2407,
+      "step": 29700
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 3.5231897830963135,
+      "learning_rate": 0.0002116974874371859,
+      "loss": 1.2296,
+      "step": 29800
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 4.537712097167969,
+      "learning_rate": 0.00021139597989949745,
+      "loss": 1.2704,
+      "step": 29900
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 3.12864351272583,
+      "learning_rate": 0.00021109447236180903,
+      "loss": 1.3093,
+      "step": 30000
+    },
+    {
+      "epoch": 1.5,
+      "eval_loss": 1.2697720527648926,
+      "eval_runtime": 37.8104,
+      "eval_samples_per_second": 26.448,
+      "eval_steps_per_second": 3.306,
+      "step": 30000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.9532142877578735,
+      "learning_rate": 0.0002107929648241206,
+      "loss": 1.2892,
+      "step": 30100
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.9121806621551514,
+      "learning_rate": 0.00021049145728643215,
+      "loss": 1.282,
+      "step": 30200
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.2597557306289673,
+      "learning_rate": 0.00021018994974874368,
+      "loss": 1.2793,
+      "step": 30300
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.7637083530426025,
+      "learning_rate": 0.00020988844221105527,
+      "loss": 1.3253,
+      "step": 30400
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 3.788984775543213,
+      "learning_rate": 0.00020958693467336683,
+      "loss": 1.249,
+      "step": 30500
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 3.1422038078308105,
+      "learning_rate": 0.00020928542713567836,
+      "loss": 1.2429,
+      "step": 30600
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.995868444442749,
+      "learning_rate": 0.00020898391959798992,
+      "loss": 1.2827,
+      "step": 30700
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.3635036945343018,
+      "learning_rate": 0.00020868241206030148,
+      "loss": 1.2653,
+      "step": 30800
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.0892832279205322,
+      "learning_rate": 0.00020838090452261307,
+      "loss": 1.2814,
+      "step": 30900
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 2.8766140937805176,
+      "learning_rate": 0.0002080793969849246,
+      "loss": 1.2809,
+      "step": 31000
+    },
+    {
+      "epoch": 1.55,
+      "eval_loss": 1.2703502178192139,
+      "eval_runtime": 37.818,
+      "eval_samples_per_second": 26.442,
+      "eval_steps_per_second": 3.305,
+      "step": 31000
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.5487587451934814,
+      "learning_rate": 0.00020777788944723616,
+      "loss": 1.2811,
+      "step": 31100
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.325295925140381,
+      "learning_rate": 0.00020747638190954772,
+      "loss": 1.2769,
+      "step": 31200
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.741773009300232,
+      "learning_rate": 0.00020717487437185928,
+      "loss": 1.2741,
+      "step": 31300
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 5.916422367095947,
+      "learning_rate": 0.00020687336683417084,
+      "loss": 1.2567,
+      "step": 31400
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 2.166018009185791,
+      "learning_rate": 0.0002065718592964824,
+      "loss": 1.2491,
+      "step": 31500
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.7622108459472656,
+      "learning_rate": 0.00020627035175879393,
+      "loss": 1.2815,
+      "step": 31600
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 2.2861111164093018,
+      "learning_rate": 0.00020596884422110552,
+      "loss": 1.2485,
+      "step": 31700
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 2.8738324642181396,
+      "learning_rate": 0.00020566733668341708,
+      "loss": 1.2747,
+      "step": 31800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.920782208442688,
+      "learning_rate": 0.00020536582914572863,
+      "loss": 1.3094,
+      "step": 31900
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 2.591792345046997,
+      "learning_rate": 0.00020506432160804017,
+      "loss": 1.3178,
+      "step": 32000
+    },
+    {
+      "epoch": 1.6,
+      "eval_loss": 1.2383744716644287,
+      "eval_runtime": 37.8786,
+      "eval_samples_per_second": 26.4,
+      "eval_steps_per_second": 3.3,
+      "step": 32000
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 3.4940438270568848,
+      "learning_rate": 0.00020476281407035175,
+      "loss": 1.2755,
+      "step": 32100
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.377112627029419,
+      "learning_rate": 0.0002044613065326633,
+      "loss": 1.2667,
+      "step": 32200
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.5229716300964355,
+      "learning_rate": 0.00020415979899497485,
+      "loss": 1.2695,
+      "step": 32300
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 2.469883441925049,
+      "learning_rate": 0.0002038582914572864,
+      "loss": 1.3089,
+      "step": 32400
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.9299498796463013,
+      "learning_rate": 0.000203556783919598,
+      "loss": 1.2835,
+      "step": 32500
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.486790895462036,
+      "learning_rate": 0.00020325527638190955,
+      "loss": 1.2531,
+      "step": 32600
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 3.485691785812378,
+      "learning_rate": 0.00020295376884422108,
+      "loss": 1.2568,
+      "step": 32700
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.674727201461792,
+      "learning_rate": 0.00020265226130653264,
+      "loss": 1.2739,
+      "step": 32800
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 4.50739049911499,
+      "learning_rate": 0.00020235075376884417,
+      "loss": 1.211,
+      "step": 32900
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 11.218056678771973,
+      "learning_rate": 0.00020204924623115576,
+      "loss": 1.2891,
+      "step": 33000
+    },
+    {
+      "epoch": 1.65,
+      "eval_loss": 1.2705625295639038,
+      "eval_runtime": 37.8291,
+      "eval_samples_per_second": 26.435,
+      "eval_steps_per_second": 3.304,
+      "step": 33000
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.9991952180862427,
+      "learning_rate": 0.00020174773869346732,
+      "loss": 1.2636,
+      "step": 33100
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 3.0366969108581543,
+      "learning_rate": 0.00020144623115577888,
+      "loss": 1.2903,
+      "step": 33200
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.7985395193099976,
+      "learning_rate": 0.0002011447236180904,
+      "loss": 1.2437,
+      "step": 33300
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 3.8208954334259033,
+      "learning_rate": 0.000200843216080402,
+      "loss": 1.244,
+      "step": 33400
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 3.2836215496063232,
+      "learning_rate": 0.00020054170854271356,
+      "loss": 1.2837,
+      "step": 33500
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 3.15663480758667,
+      "learning_rate": 0.0002002402010050251,
+      "loss": 1.2253,
+      "step": 33600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.6871391534805298,
+      "learning_rate": 0.00019993869346733665,
+      "loss": 1.2564,
+      "step": 33700
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 2.3701913356781006,
+      "learning_rate": 0.00019963718592964824,
+      "loss": 1.2925,
+      "step": 33800
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 2.9534804821014404,
+      "learning_rate": 0.0001993356783919598,
+      "loss": 1.2613,
+      "step": 33900
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 2.273113489151001,
+      "learning_rate": 0.00019903417085427133,
+      "loss": 1.29,
+      "step": 34000
+    },
+    {
+      "epoch": 1.7,
+      "eval_loss": 1.2713490724563599,
+      "eval_runtime": 37.9786,
+      "eval_samples_per_second": 26.331,
+      "eval_steps_per_second": 3.291,
+      "step": 34000
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 2.1708054542541504,
+      "learning_rate": 0.0001987326633165829,
+      "loss": 1.2775,
+      "step": 34100
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 2.242708683013916,
+      "learning_rate": 0.00019843115577889447,
+      "loss": 1.2561,
+      "step": 34200
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 2.0170931816101074,
+      "learning_rate": 0.000198129648241206,
+      "loss": 1.2168,
+      "step": 34300
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.094848871231079,
+      "learning_rate": 0.00019782814070351757,
+      "loss": 1.2588,
+      "step": 34400
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 2.1762752532958984,
+      "learning_rate": 0.00019752663316582913,
+      "loss": 1.1837,
+      "step": 34500
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 3.1318016052246094,
+      "learning_rate": 0.0001972251256281407,
+      "loss": 1.2196,
+      "step": 34600
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 3.2971861362457275,
+      "learning_rate": 0.00019692361809045225,
+      "loss": 1.2778,
+      "step": 34700
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 3.452091693878174,
+      "learning_rate": 0.0001966221105527638,
+      "loss": 1.2385,
+      "step": 34800
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.7514299154281616,
+      "learning_rate": 0.00019632060301507536,
+      "loss": 1.2769,
+      "step": 34900
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 2.3494088649749756,
+      "learning_rate": 0.00019601909547738692,
+      "loss": 1.2689,
+      "step": 35000
+    },
+    {
+      "epoch": 1.75,
+      "eval_loss": 1.2675199508666992,
+      "eval_runtime": 37.8879,
+      "eval_samples_per_second": 26.394,
+      "eval_steps_per_second": 3.299,
+      "step": 35000
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.5741009712219238,
+      "learning_rate": 0.00019571758793969848,
+      "loss": 1.2352,
+      "step": 35100
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 2.652435302734375,
+      "learning_rate": 0.00019541608040201004,
+      "loss": 1.2824,
+      "step": 35200
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 2.9557676315307617,
+      "learning_rate": 0.00019511457286432157,
+      "loss": 1.2453,
+      "step": 35300
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 2.8758041858673096,
+      "learning_rate": 0.00019481306532663313,
+      "loss": 1.2507,
+      "step": 35400
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 2.5828402042388916,
+      "learning_rate": 0.0001945145728643216,
+      "loss": 1.2201,
+      "step": 35500
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 2.887206554412842,
+      "learning_rate": 0.00019421306532663312,
+      "loss": 1.2754,
+      "step": 35600
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 2.5521140098571777,
+      "learning_rate": 0.0001939115577889447,
+      "loss": 1.234,
+      "step": 35700
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.9570846557617188,
+      "learning_rate": 0.00019361005025125627,
+      "loss": 1.2708,
+      "step": 35800
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 2.89273738861084,
+      "learning_rate": 0.00019330854271356782,
+      "loss": 1.2343,
+      "step": 35900
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 3.624706506729126,
+      "learning_rate": 0.00019300703517587936,
+      "loss": 1.2576,
+      "step": 36000
+    },
+    {
+      "epoch": 1.8,
+      "eval_loss": 1.2644726037979126,
+      "eval_runtime": 37.8527,
+      "eval_samples_per_second": 26.418,
+      "eval_steps_per_second": 3.302,
+      "step": 36000
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 2.5976133346557617,
+      "learning_rate": 0.00019270552763819094,
+      "loss": 1.2812,
+      "step": 36100
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 2.899306297302246,
+      "learning_rate": 0.0001924040201005025,
+      "loss": 1.2541,
+      "step": 36200
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 3.964782476425171,
+      "learning_rate": 0.00019210251256281404,
+      "loss": 1.2639,
+      "step": 36300
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 2.4634933471679688,
+      "learning_rate": 0.0001918010050251256,
+      "loss": 1.2089,
+      "step": 36400
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 2.6023619174957275,
+      "learning_rate": 0.00019149949748743718,
+      "loss": 1.2612,
+      "step": 36500
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 3.0462849140167236,
+      "learning_rate": 0.00019119798994974874,
+      "loss": 1.2204,
+      "step": 36600
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 2.1344144344329834,
+      "learning_rate": 0.00019089648241206027,
+      "loss": 1.2142,
+      "step": 36700
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.5994189977645874,
+      "learning_rate": 0.00019059497487437183,
+      "loss": 1.2586,
+      "step": 36800
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.357469916343689,
+      "learning_rate": 0.00019029346733668342,
+      "loss": 1.2705,
+      "step": 36900
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 2.4201526641845703,
+      "learning_rate": 0.00018999195979899495,
+      "loss": 1.2409,
+      "step": 37000
+    },
+    {
+      "epoch": 1.85,
+      "eval_loss": 1.2103183269500732,
+      "eval_runtime": 37.8707,
+      "eval_samples_per_second": 26.406,
+      "eval_steps_per_second": 3.301,
+      "step": 37000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 3.1790504455566406,
+      "learning_rate": 0.0001896904522613065,
+      "loss": 1.2639,
+      "step": 37100
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 2.565474033355713,
+      "learning_rate": 0.00018938894472361807,
+      "loss": 1.2853,
+      "step": 37200
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 2.6977927684783936,
+      "learning_rate": 0.00018908743718592966,
+      "loss": 1.2178,
+      "step": 37300
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 2.588975191116333,
+      "learning_rate": 0.0001887859296482412,
+      "loss": 1.2492,
+      "step": 37400
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 2.23592209815979,
+      "learning_rate": 0.00018848442211055275,
+      "loss": 1.2273,
+      "step": 37500
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 2.0961692333221436,
+      "learning_rate": 0.0001881859296482412,
+      "loss": 1.2375,
+      "step": 37600
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 2.4870264530181885,
+      "learning_rate": 0.00018788442211055273,
+      "loss": 1.2564,
+      "step": 37700
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.9144058227539062,
+      "learning_rate": 0.0001875829145728643,
+      "loss": 1.2403,
+      "step": 37800
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 2.209117889404297,
+      "learning_rate": 0.00018728140703517588,
+      "loss": 1.2168,
+      "step": 37900
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 2.7400968074798584,
+      "learning_rate": 0.00018697989949748744,
+      "loss": 1.1786,
+      "step": 38000
+    },
+    {
+      "epoch": 1.9,
+      "eval_loss": 1.2550157308578491,
+      "eval_runtime": 37.907,
+      "eval_samples_per_second": 26.38,
+      "eval_steps_per_second": 3.298,
+      "step": 38000
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 2.392390251159668,
+      "learning_rate": 0.00018667839195979897,
+      "loss": 1.2294,
+      "step": 38100
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 3.434168577194214,
+      "learning_rate": 0.00018637688442211053,
+      "loss": 1.2491,
+      "step": 38200
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 2.082618236541748,
+      "learning_rate": 0.0001860753768844221,
+      "loss": 1.2602,
+      "step": 38300
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.6049084663391113,
+      "learning_rate": 0.00018577386934673365,
+      "loss": 1.2067,
+      "step": 38400
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 2.1953368186950684,
+      "learning_rate": 0.0001854723618090452,
+      "loss": 1.2292,
+      "step": 38500
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 2.6085190773010254,
+      "learning_rate": 0.00018517085427135677,
+      "loss": 1.2269,
+      "step": 38600
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 2.9110639095306396,
+      "learning_rate": 0.0001848693467336683,
+      "loss": 1.1898,
+      "step": 38700
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.514410138130188,
+      "learning_rate": 0.0001845678391959799,
+      "loss": 1.199,
+      "step": 38800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 4.6756134033203125,
+      "learning_rate": 0.00018426633165829145,
+      "loss": 1.183,
+      "step": 38900
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 2.704317808151245,
+      "learning_rate": 0.000183964824120603,
+      "loss": 1.1999,
+      "step": 39000
+    },
+    {
+      "epoch": 1.95,
+      "eval_loss": 1.2309662103652954,
+      "eval_runtime": 37.8598,
+      "eval_samples_per_second": 26.413,
+      "eval_steps_per_second": 3.302,
+      "step": 39000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 2.5975565910339355,
+      "learning_rate": 0.00018366331658291454,
+      "loss": 1.2576,
+      "step": 39100
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 3.3112730979919434,
+      "learning_rate": 0.00018336180904522613,
+      "loss": 1.2128,
+      "step": 39200
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 2.5991640090942383,
+      "learning_rate": 0.00018306030150753769,
+      "loss": 1.2294,
+      "step": 39300
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 4.411704063415527,
+      "learning_rate": 0.00018275879396984922,
+      "loss": 1.1977,
+      "step": 39400
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.509308099746704,
+      "learning_rate": 0.00018245728643216078,
+      "loss": 1.2712,
+      "step": 39500
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 2.136350631713867,
+      "learning_rate": 0.00018215577889447236,
+      "loss": 1.2359,
+      "step": 39600
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 2.1651546955108643,
+      "learning_rate": 0.0001818542713567839,
+      "loss": 1.2448,
+      "step": 39700
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 2.9962761402130127,
+      "learning_rate": 0.00018155577889447235,
+      "loss": 1.218,
+      "step": 39800
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 2.8525376319885254,
+      "learning_rate": 0.0001812542713567839,
+      "loss": 1.2564,
+      "step": 39900
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 2.120208740234375,
+      "learning_rate": 0.00018095276381909547,
+      "loss": 1.2287,
+      "step": 40000
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 1.2058476209640503,
+      "eval_runtime": 38.0203,
+      "eval_samples_per_second": 26.302,
+      "eval_steps_per_second": 3.288,
+      "step": 40000
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 3.9785573482513428,
+      "learning_rate": 0.000180651256281407,
+      "loss": 1.2161,
+      "step": 40100
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 2.7897050380706787,
+      "learning_rate": 0.0001803497487437186,
+      "loss": 1.2525,
+      "step": 40200
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 2.042492389678955,
+      "learning_rate": 0.00018004824120603015,
+      "loss": 1.2087,
+      "step": 40300
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.8287073373794556,
+      "learning_rate": 0.00017974673366834168,
+      "loss": 1.2404,
+      "step": 40400
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.6399390697479248,
+      "learning_rate": 0.00017944522613065324,
+      "loss": 1.174,
+      "step": 40500
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 3.9909472465515137,
+      "learning_rate": 0.00017914371859296482,
+      "loss": 1.1869,
+      "step": 40600
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 2.9356400966644287,
+      "learning_rate": 0.00017884221105527638,
+      "loss": 1.2271,
+      "step": 40700
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 2.205498218536377,
+      "learning_rate": 0.00017854070351758792,
+      "loss": 1.2505,
+      "step": 40800
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 2.2801437377929688,
+      "learning_rate": 0.00017823919597989948,
+      "loss": 1.2232,
+      "step": 40900
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 4.001745223999023,
+      "learning_rate": 0.00017793768844221104,
+      "loss": 1.257,
+      "step": 41000
+    },
+    {
+      "epoch": 2.05,
+      "eval_loss": 1.1965339183807373,
+      "eval_runtime": 37.9045,
+      "eval_samples_per_second": 26.382,
+      "eval_steps_per_second": 3.298,
+      "step": 41000
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 3.484135150909424,
+      "learning_rate": 0.0001776361809045226,
+      "loss": 1.2232,
+      "step": 41100
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 2.7462897300720215,
+      "learning_rate": 0.00017733467336683415,
+      "loss": 1.22,
+      "step": 41200
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 2.9418435096740723,
+      "learning_rate": 0.00017703316582914571,
+      "loss": 1.2141,
+      "step": 41300
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 2.188680410385132,
+      "learning_rate": 0.00017673165829145725,
+      "loss": 1.1909,
+      "step": 41400
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 3.728938579559326,
+      "learning_rate": 0.00017643015075376883,
+      "loss": 1.2146,
+      "step": 41500
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 2.8790736198425293,
+      "learning_rate": 0.0001761286432160804,
+      "loss": 1.2305,
+      "step": 41600
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 3.6593847274780273,
+      "learning_rate": 0.00017582713567839195,
+      "loss": 1.1753,
+      "step": 41700
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 2.408237934112549,
+      "learning_rate": 0.00017552562814070348,
+      "loss": 1.2229,
+      "step": 41800
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 2.574580669403076,
+      "learning_rate": 0.00017522412060301507,
+      "loss": 1.2173,
+      "step": 41900
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 2.2249817848205566,
+      "learning_rate": 0.00017492261306532663,
+      "loss": 1.2112,
+      "step": 42000
+    },
+    {
+      "epoch": 2.1,
+      "eval_loss": 1.2255558967590332,
+      "eval_runtime": 37.9009,
+      "eval_samples_per_second": 26.385,
+      "eval_steps_per_second": 3.298,
+      "step": 42000
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 2.2712411880493164,
+      "learning_rate": 0.00017462110552763816,
+      "loss": 1.1862,
+      "step": 42100
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.646330714225769,
+      "learning_rate": 0.00017431959798994972,
+      "loss": 1.1812,
+      "step": 42200
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 2.9691689014434814,
+      "learning_rate": 0.0001740180904522613,
+      "loss": 1.2055,
+      "step": 42300
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 5.179681777954102,
+      "learning_rate": 0.00017371658291457287,
+      "loss": 1.1625,
+      "step": 42400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 2.634462833404541,
+      "learning_rate": 0.0001734150753768844,
+      "loss": 1.2257,
+      "step": 42500
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 8.693337440490723,
+      "learning_rate": 0.00017311356783919596,
+      "loss": 1.2447,
+      "step": 42600
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 3.228513240814209,
+      "learning_rate": 0.00017281206030150755,
+      "loss": 1.1993,
+      "step": 42700
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 7.938237190246582,
+      "learning_rate": 0.00017251055276381908,
+      "loss": 1.2084,
+      "step": 42800
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 3.0843794345855713,
+      "learning_rate": 0.00017220904522613064,
+      "loss": 1.2017,
+      "step": 42900
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 2.86205792427063,
+      "learning_rate": 0.0001719075376884422,
+      "loss": 1.1706,
+      "step": 43000
+    },
+    {
+      "epoch": 2.15,
+      "eval_loss": 1.2179350852966309,
+      "eval_runtime": 37.9173,
+      "eval_samples_per_second": 26.373,
+      "eval_steps_per_second": 3.297,
+      "step": 43000
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 2.137380361557007,
+      "learning_rate": 0.00017160904522613062,
+      "loss": 1.2066,
+      "step": 43100
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 2.250091075897217,
+      "learning_rate": 0.00017130753768844218,
+      "loss": 1.211,
+      "step": 43200
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 2.008875608444214,
+      "learning_rate": 0.00017100603015075377,
+      "loss": 1.2116,
+      "step": 43300
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 2.6691529750823975,
+      "learning_rate": 0.00017070452261306533,
+      "loss": 1.1844,
+      "step": 43400
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.8802026510238647,
+      "learning_rate": 0.00017040301507537686,
+      "loss": 1.1849,
+      "step": 43500
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 2.4100139141082764,
+      "learning_rate": 0.00017010150753768842,
+      "loss": 1.1887,
+      "step": 43600
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 3.3384740352630615,
+      "learning_rate": 0.00016979999999999998,
+      "loss": 1.2338,
+      "step": 43700
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 2.349433183670044,
+      "learning_rate": 0.00016949849246231154,
+      "loss": 1.1633,
+      "step": 43800
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 3.019296884536743,
+      "learning_rate": 0.0001691969849246231,
+      "loss": 1.2456,
+      "step": 43900
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 2.497424364089966,
+      "learning_rate": 0.00016889547738693466,
+      "loss": 1.1671,
+      "step": 44000
+    },
+    {
+      "epoch": 2.2,
+      "eval_loss": 1.2000114917755127,
+      "eval_runtime": 40.4714,
+      "eval_samples_per_second": 24.709,
+      "eval_steps_per_second": 3.089,
+      "step": 44000
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.6698800325393677,
+      "learning_rate": 0.0001685939698492462,
+      "loss": 1.2105,
+      "step": 44100
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 2.3846988677978516,
+      "learning_rate": 0.00016829246231155778,
+      "loss": 1.2229,
+      "step": 44200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 5.891537189483643,
+      "learning_rate": 0.00016799095477386934,
+      "loss": 1.1848,
+      "step": 44300
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.4433008432388306,
+      "learning_rate": 0.0001676894472361809,
+      "loss": 1.1905,
+      "step": 44400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 2.5641889572143555,
+      "learning_rate": 0.00016738793969849243,
+      "loss": 1.2219,
+      "step": 44500
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 3.052948474884033,
+      "learning_rate": 0.00016708643216080402,
+      "loss": 1.1887,
+      "step": 44600
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 2.8185369968414307,
+      "learning_rate": 0.00016678492462311557,
+      "loss": 1.2107,
+      "step": 44700
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 2.9409399032592773,
+      "learning_rate": 0.0001664834170854271,
+      "loss": 1.2222,
+      "step": 44800
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 2.728256940841675,
+      "learning_rate": 0.00016618190954773867,
+      "loss": 1.1767,
+      "step": 44900
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 2.4744584560394287,
+      "learning_rate": 0.00016588040201005025,
+      "loss": 1.1663,
+      "step": 45000
+    },
+    {
+      "epoch": 2.25,
+      "eval_loss": 1.2085031270980835,
+      "eval_runtime": 41.0345,
+      "eval_samples_per_second": 24.37,
+      "eval_steps_per_second": 3.046,
+      "step": 45000
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 3.215564250946045,
+      "learning_rate": 0.00016558190954773868,
+      "loss": 1.173,
+      "step": 45100
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.7013347148895264,
+      "learning_rate": 0.00016528040201005024,
+      "loss": 1.1637,
+      "step": 45200
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 3.1096675395965576,
+      "learning_rate": 0.0001649788944723618,
+      "loss": 1.1702,
+      "step": 45300
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 2.5975756645202637,
+      "learning_rate": 0.00016467738693467336,
+      "loss": 1.1763,
+      "step": 45400
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 2.7020699977874756,
+      "learning_rate": 0.0001643758793969849,
+      "loss": 1.1761,
+      "step": 45500
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.7007007598876953,
+      "learning_rate": 0.00016407437185929648,
+      "loss": 1.2064,
+      "step": 45600
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 3.6038424968719482,
+      "learning_rate": 0.00016377286432160804,
+      "loss": 1.1716,
+      "step": 45700
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 2.3656082153320312,
+      "learning_rate": 0.0001634713567839196,
+      "loss": 1.1954,
+      "step": 45800
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 2.390509605407715,
+      "learning_rate": 0.00016316984924623113,
+      "loss": 1.1664,
+      "step": 45900
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.8767670392990112,
+      "learning_rate": 0.00016286834170854271,
+      "loss": 1.1784,
+      "step": 46000
+    },
+    {
+      "epoch": 2.3,
+      "eval_loss": 1.1809154748916626,
+      "eval_runtime": 43.7304,
+      "eval_samples_per_second": 22.867,
+      "eval_steps_per_second": 2.858,
+      "step": 46000
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 3.4367122650146484,
+      "learning_rate": 0.00016256683417085427,
+      "loss": 1.2055,
+      "step": 46100
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.672525405883789,
+      "learning_rate": 0.0001622653266331658,
+      "loss": 1.1954,
+      "step": 46200
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 3.2755866050720215,
+      "learning_rate": 0.00016196381909547737,
+      "loss": 1.1801,
+      "step": 46300
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 2.347280979156494,
+      "learning_rate": 0.00016166231155778892,
+      "loss": 1.1651,
+      "step": 46400
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.9565701484680176,
+      "learning_rate": 0.0001613608040201005,
+      "loss": 1.2142,
+      "step": 46500
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 2.317847728729248,
+      "learning_rate": 0.00016105929648241204,
+      "loss": 1.188,
+      "step": 46600
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.812322974205017,
+      "learning_rate": 0.0001607577889447236,
+      "loss": 1.1425,
+      "step": 46700
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 2.5393502712249756,
+      "learning_rate": 0.00016045628140703514,
+      "loss": 1.1854,
+      "step": 46800
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 6.562712669372559,
+      "learning_rate": 0.00016015477386934672,
+      "loss": 1.1517,
+      "step": 46900
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 2.2086706161499023,
+      "learning_rate": 0.00015985326633165828,
+      "loss": 1.1634,
+      "step": 47000
+    },
+    {
+      "epoch": 2.35,
+      "eval_loss": 1.1972031593322754,
+      "eval_runtime": 43.2883,
+      "eval_samples_per_second": 23.101,
+      "eval_steps_per_second": 2.888,
+      "step": 47000
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 2.061951160430908,
+      "learning_rate": 0.00015955175879396984,
+      "loss": 1.2409,
+      "step": 47100
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 2.0312881469726562,
+      "learning_rate": 0.00015925025125628137,
+      "loss": 1.1731,
+      "step": 47200
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 4.90245246887207,
+      "learning_rate": 0.00015894874371859296,
+      "loss": 1.1849,
+      "step": 47300
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 2.4970901012420654,
+      "learning_rate": 0.00015864723618090452,
+      "loss": 1.1684,
+      "step": 47400
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 2.4406049251556396,
+      "learning_rate": 0.00015834572864321605,
+      "loss": 1.1855,
+      "step": 47500
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 2.8650543689727783,
+      "learning_rate": 0.0001580442211055276,
+      "loss": 1.1586,
+      "step": 47600
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 2.4787731170654297,
+      "learning_rate": 0.0001577427135678392,
+      "loss": 1.1913,
+      "step": 47700
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 2.5188841819763184,
+      "learning_rate": 0.00015744120603015076,
+      "loss": 1.1938,
+      "step": 47800
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 3.8095650672912598,
+      "learning_rate": 0.0001571396984924623,
+      "loss": 1.1858,
+      "step": 47900
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 2.147993564605713,
+      "learning_rate": 0.00015683819095477385,
+      "loss": 1.1703,
+      "step": 48000
+    },
+    {
+      "epoch": 2.4,
+      "eval_loss": 1.1952226161956787,
+      "eval_runtime": 42.4811,
+      "eval_samples_per_second": 23.54,
+      "eval_steps_per_second": 2.942,
+      "step": 48000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 3.050976514816284,
+      "learning_rate": 0.00015653668341708544,
+      "loss": 1.1868,
+      "step": 48100
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 2.6880428791046143,
+      "learning_rate": 0.00015623517587939697,
+      "loss": 1.1486,
+      "step": 48200
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 2.169895648956299,
+      "learning_rate": 0.00015593366834170853,
+      "loss": 1.1646,
+      "step": 48300
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 9.948437690734863,
+      "learning_rate": 0.0001556321608040201,
+      "loss": 1.1625,
+      "step": 48400
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 2.1219215393066406,
+      "learning_rate": 0.00015533065326633162,
+      "loss": 1.1854,
+      "step": 48500
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 3.2466542720794678,
+      "learning_rate": 0.0001550291457286432,
+      "loss": 1.1556,
+      "step": 48600
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.8362162113189697,
+      "learning_rate": 0.00015472763819095477,
+      "loss": 1.177,
+      "step": 48700
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 3.579221725463867,
+      "learning_rate": 0.00015442613065326632,
+      "loss": 1.1671,
+      "step": 48800
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 2.256967782974243,
+      "learning_rate": 0.00015412462311557786,
+      "loss": 1.1807,
+      "step": 48900
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 2.107179641723633,
+      "learning_rate": 0.00015382311557788944,
+      "loss": 1.186,
+      "step": 49000
+    },
+    {
+      "epoch": 2.45,
+      "eval_loss": 1.1811304092407227,
+      "eval_runtime": 43.1582,
+      "eval_samples_per_second": 23.171,
+      "eval_steps_per_second": 2.896,
+      "step": 49000
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 2.615290880203247,
+      "learning_rate": 0.000153521608040201,
+      "loss": 1.1828,
+      "step": 49100
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.600845217704773,
+      "learning_rate": 0.00015322010050251254,
+      "loss": 1.1438,
+      "step": 49200
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 2.272726058959961,
+      "learning_rate": 0.0001529185929648241,
+      "loss": 1.1802,
+      "step": 49300
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.9845112562179565,
+      "learning_rate": 0.00015261708542713568,
+      "loss": 1.1828,
+      "step": 49400
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.4725877046585083,
+      "learning_rate": 0.00015231859296482408,
+      "loss": 1.1938,
+      "step": 49500
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 2.4453134536743164,
+      "learning_rate": 0.00015201708542713567,
+      "loss": 1.1928,
+      "step": 49600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 2.9869000911712646,
+      "learning_rate": 0.00015171557788944723,
+      "loss": 1.1982,
+      "step": 49700
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 2.633794069290161,
+      "learning_rate": 0.00015141407035175879,
+      "loss": 1.1287,
+      "step": 49800
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.8146005868911743,
+      "learning_rate": 0.00015111256281407032,
+      "loss": 1.1747,
+      "step": 49900
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 6.4758405685424805,
+      "learning_rate": 0.0001508110552763819,
+      "loss": 1.1548,
+      "step": 50000
+    },
+    {
+      "epoch": 2.5,
+      "eval_loss": 1.1896699666976929,
+      "eval_runtime": 43.2315,
+      "eval_samples_per_second": 23.131,
+      "eval_steps_per_second": 2.891,
+      "step": 50000
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.5688796043395996,
+      "learning_rate": 0.00015050954773869346,
+      "loss": 1.168,
+      "step": 50100
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 1.4024161100387573,
+      "learning_rate": 0.000150208040201005,
+      "loss": 1.1796,
+      "step": 50200
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 2.066570997238159,
+      "learning_rate": 0.00014990653266331658,
+      "loss": 1.1419,
+      "step": 50300
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 3.7978389263153076,
+      "learning_rate": 0.00014960502512562812,
+      "loss": 1.1497,
+      "step": 50400
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 2.2129733562469482,
+      "learning_rate": 0.0001493035175879397,
+      "loss": 1.1371,
+      "step": 50500
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 3.0140724182128906,
+      "learning_rate": 0.00014900201005025123,
+      "loss": 1.1778,
+      "step": 50600
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 2.457521915435791,
+      "learning_rate": 0.00014870050251256282,
+      "loss": 1.1266,
+      "step": 50700
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 2.1066813468933105,
+      "learning_rate": 0.00014839899497487435,
+      "loss": 1.1635,
+      "step": 50800
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 2.801196336746216,
+      "learning_rate": 0.0001480974874371859,
+      "loss": 1.1842,
+      "step": 50900
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 4.693379878997803,
+      "learning_rate": 0.00014779597989949747,
+      "loss": 1.1449,
+      "step": 51000
+    },
+    {
+      "epoch": 2.55,
+      "eval_loss": 1.1495003700256348,
+      "eval_runtime": 37.9097,
+      "eval_samples_per_second": 26.378,
+      "eval_steps_per_second": 3.297,
+      "step": 51000
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.917925477027893,
+      "learning_rate": 0.00014749447236180903,
+      "loss": 1.1303,
+      "step": 51100
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 2.6460864543914795,
+      "learning_rate": 0.0001471929648241206,
+      "loss": 1.1638,
+      "step": 51200
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 2.5040736198425293,
+      "learning_rate": 0.00014689145728643215,
+      "loss": 1.1382,
+      "step": 51300
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 2.7533071041107178,
+      "learning_rate": 0.0001465899497487437,
+      "loss": 1.1803,
+      "step": 51400
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 2.220345973968506,
+      "learning_rate": 0.00014629145728643214,
+      "loss": 1.1506,
+      "step": 51500
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.3668216466903687,
+      "learning_rate": 0.0001459899497487437,
+      "loss": 1.1538,
+      "step": 51600
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 2.26232647895813,
+      "learning_rate": 0.00014568844221105525,
+      "loss": 1.2085,
+      "step": 51700
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 5.508904933929443,
+      "learning_rate": 0.00014538693467336681,
+      "loss": 1.1528,
+      "step": 51800
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 2.9169905185699463,
+      "learning_rate": 0.00014508542713567837,
+      "loss": 1.1632,
+      "step": 51900
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 2.5156240463256836,
+      "learning_rate": 0.00014478391959798993,
+      "loss": 1.1677,
+      "step": 52000
+    },
+    {
+      "epoch": 2.6,
+      "eval_loss": 1.174816370010376,
+      "eval_runtime": 42.0784,
+      "eval_samples_per_second": 23.765,
+      "eval_steps_per_second": 2.971,
+      "step": 52000
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.622004747390747,
+      "learning_rate": 0.0001444824120603015,
+      "loss": 1.1174,
+      "step": 52100
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 2.5255143642425537,
+      "learning_rate": 0.00014418090452261305,
+      "loss": 1.1415,
+      "step": 52200
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 1.7780824899673462,
+      "learning_rate": 0.0001438793969849246,
+      "loss": 1.1871,
+      "step": 52300
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 2.320028305053711,
+      "learning_rate": 0.00014357788944723617,
+      "loss": 1.1841,
+      "step": 52400
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 2.6219685077667236,
+      "learning_rate": 0.00014327638190954773,
+      "loss": 1.1349,
+      "step": 52500
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 3.0288233757019043,
+      "learning_rate": 0.0001429748743718593,
+      "loss": 1.1753,
+      "step": 52600
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 2.3062517642974854,
+      "learning_rate": 0.00014267336683417085,
+      "loss": 1.1836,
+      "step": 52700
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.8819166421890259,
+      "learning_rate": 0.0001423718592964824,
+      "loss": 1.1491,
+      "step": 52800
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.7771334648132324,
+      "learning_rate": 0.00014207035175879397,
+      "loss": 1.1311,
+      "step": 52900
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.9495539665222168,
+      "learning_rate": 0.00014176884422110553,
+      "loss": 1.1757,
+      "step": 53000
+    },
+    {
+      "epoch": 2.65,
+      "eval_loss": 1.161841869354248,
+      "eval_runtime": 41.6597,
+      "eval_samples_per_second": 24.004,
+      "eval_steps_per_second": 3.0,
+      "step": 53000
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 2.317021131515503,
+      "learning_rate": 0.00014146733668341706,
+      "loss": 1.145,
+      "step": 53100
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 1.4079538583755493,
+      "learning_rate": 0.00014116582914572865,
+      "loss": 1.0893,
+      "step": 53200
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 6.593141555786133,
+      "learning_rate": 0.00014086432160804018,
+      "loss": 1.1357,
+      "step": 53300
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 2.657529830932617,
+      "learning_rate": 0.00014056281407035177,
+      "loss": 1.1651,
+      "step": 53400
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 3.312056541442871,
+      "learning_rate": 0.0001402613065326633,
+      "loss": 1.165,
+      "step": 53500
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 2.1961281299591064,
+      "learning_rate": 0.00013995979899497486,
+      "loss": 1.1584,
+      "step": 53600
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 1.933409571647644,
+      "learning_rate": 0.00013965829145728642,
+      "loss": 1.1382,
+      "step": 53700
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 2.6763832569122314,
+      "learning_rate": 0.00013935678391959798,
+      "loss": 1.1238,
+      "step": 53800
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 3.3957033157348633,
+      "learning_rate": 0.00013905527638190954,
+      "loss": 1.154,
+      "step": 53900
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 3.526700019836426,
+      "learning_rate": 0.0001387537688442211,
+      "loss": 1.1325,
+      "step": 54000
+    },
+    {
+      "epoch": 2.7,
+      "eval_loss": 1.141178011894226,
+      "eval_runtime": 37.9667,
+      "eval_samples_per_second": 26.339,
+      "eval_steps_per_second": 3.292,
+      "step": 54000
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 3.3937137126922607,
+      "learning_rate": 0.00013845226130653265,
+      "loss": 1.141,
+      "step": 54100
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 1.9187488555908203,
+      "learning_rate": 0.00013815075376884421,
+      "loss": 1.1253,
+      "step": 54200
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 2.2351136207580566,
+      "learning_rate": 0.00013784924623115577,
+      "loss": 1.2008,
+      "step": 54300
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 3.97955584526062,
+      "learning_rate": 0.0001375477386934673,
+      "loss": 1.1609,
+      "step": 54400
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 3.5734050273895264,
+      "learning_rate": 0.0001372462311557789,
+      "loss": 1.1584,
+      "step": 54500
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 2.3804807662963867,
+      "learning_rate": 0.00013694472361809042,
+      "loss": 1.1343,
+      "step": 54600
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 2.0606038570404053,
+      "learning_rate": 0.000136643216080402,
+      "loss": 1.1555,
+      "step": 54700
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 4.046571731567383,
+      "learning_rate": 0.00013634170854271354,
+      "loss": 1.1543,
+      "step": 54800
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 2.470393180847168,
+      "learning_rate": 0.00013604020100502513,
+      "loss": 1.1651,
+      "step": 54900
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.4677540063858032,
+      "learning_rate": 0.00013573869346733666,
+      "loss": 1.1366,
+      "step": 55000
+    },
+    {
+      "epoch": 2.75,
+      "eval_loss": 1.1223907470703125,
+      "eval_runtime": 43.6458,
+      "eval_samples_per_second": 22.912,
+      "eval_steps_per_second": 2.864,
+      "step": 55000
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 2.5567593574523926,
+      "learning_rate": 0.00013543718592964822,
+      "loss": 1.1348,
+      "step": 55100
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 4.812506675720215,
+      "learning_rate": 0.00013513567839195978,
+      "loss": 1.1675,
+      "step": 55200
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 2.5467748641967773,
+      "learning_rate": 0.00013483417085427134,
+      "loss": 1.1238,
+      "step": 55300
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 4.469081878662109,
+      "learning_rate": 0.0001345326633165829,
+      "loss": 1.102,
+      "step": 55400
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 3.878526449203491,
+      "learning_rate": 0.00013423115577889446,
+      "loss": 1.131,
+      "step": 55500
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 2.0142953395843506,
+      "learning_rate": 0.00013392964824120602,
+      "loss": 1.1349,
+      "step": 55600
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 2.600478410720825,
+      "learning_rate": 0.00013362814070351758,
+      "loss": 1.1363,
+      "step": 55700
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 2.58322811126709,
+      "learning_rate": 0.00013332663316582914,
+      "loss": 1.1426,
+      "step": 55800
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 2.2471609115600586,
+      "learning_rate": 0.0001330251256281407,
+      "loss": 1.1446,
+      "step": 55900
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 1.8442782163619995,
+      "learning_rate": 0.00013272361809045226,
+      "loss": 1.1315,
+      "step": 56000
+    },
+    {
+      "epoch": 2.8,
+      "eval_loss": 1.1661006212234497,
+      "eval_runtime": 47.0159,
+      "eval_samples_per_second": 21.269,
+      "eval_steps_per_second": 2.659,
+      "step": 56000
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 2.2928128242492676,
+      "learning_rate": 0.0001324221105527638,
+      "loss": 1.122,
+      "step": 56100
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 2.192915201187134,
+      "learning_rate": 0.00013212361809045224,
+      "loss": 1.1251,
+      "step": 56200
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 2.334547519683838,
+      "learning_rate": 0.00013182211055276383,
+      "loss": 1.1408,
+      "step": 56300
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 1.832930088043213,
+      "learning_rate": 0.00013152060301507536,
+      "loss": 1.1146,
+      "step": 56400
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 4.524071216583252,
+      "learning_rate": 0.00013121909547738692,
+      "loss": 1.1661,
+      "step": 56500
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.4990063905715942,
+      "learning_rate": 0.00013091758793969848,
+      "loss": 1.1247,
+      "step": 56600
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 3.572678804397583,
+      "learning_rate": 0.00013061608040201004,
+      "loss": 1.1251,
+      "step": 56700
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 2.0090138912200928,
+      "learning_rate": 0.0001303145728643216,
+      "loss": 1.1267,
+      "step": 56800
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 2.0328962802886963,
+      "learning_rate": 0.00013001306532663316,
+      "loss": 1.1343,
+      "step": 56900
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.5744613409042358,
+      "learning_rate": 0.00012971155778894472,
+      "loss": 1.1208,
+      "step": 57000
+    },
+    {
+      "epoch": 2.85,
+      "eval_loss": 1.1388169527053833,
+      "eval_runtime": 65.7696,
+      "eval_samples_per_second": 15.205,
+      "eval_steps_per_second": 1.901,
+      "step": 57000
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.2835485935211182,
+      "learning_rate": 0.00012941005025125628,
+      "loss": 1.1561,
+      "step": 57100
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 3.413334846496582,
+      "learning_rate": 0.00012910854271356784,
+      "loss": 1.126,
+      "step": 57200
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 2.6612489223480225,
+      "learning_rate": 0.00012880703517587937,
+      "loss": 1.1705,
+      "step": 57300
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 2.0389411449432373,
+      "learning_rate": 0.00012850552763819096,
+      "loss": 1.1322,
+      "step": 57400
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 2.203789710998535,
+      "learning_rate": 0.0001282040201005025,
+      "loss": 1.1437,
+      "step": 57500
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 5.272101879119873,
+      "learning_rate": 0.00012790251256281407,
+      "loss": 1.1333,
+      "step": 57600
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 3.0776541233062744,
+      "learning_rate": 0.0001276010050251256,
+      "loss": 1.1235,
+      "step": 57700
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 3.8333828449249268,
+      "learning_rate": 0.0001272994974874372,
+      "loss": 1.1141,
+      "step": 57800
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 3.3916189670562744,
+      "learning_rate": 0.00012699798994974873,
+      "loss": 1.1084,
+      "step": 57900
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 1.6035398244857788,
+      "learning_rate": 0.00012669648241206029,
+      "loss": 1.1057,
+      "step": 58000
+    },
+    {
+      "epoch": 2.9,
+      "eval_loss": 1.1182321310043335,
+      "eval_runtime": 60.567,
+      "eval_samples_per_second": 16.511,
+      "eval_steps_per_second": 2.064,
+      "step": 58000
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 2.41086745262146,
+      "learning_rate": 0.00012639497487437184,
+      "loss": 1.1424,
+      "step": 58100
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.8278477191925049,
+      "learning_rate": 0.0001260934673366834,
+      "loss": 1.1126,
+      "step": 58200
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 2.7294256687164307,
+      "learning_rate": 0.00012579195979899496,
+      "loss": 1.1207,
+      "step": 58300
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 2.813084602355957,
+      "learning_rate": 0.00012549045226130652,
+      "loss": 1.1498,
+      "step": 58400
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 2.6869473457336426,
+      "learning_rate": 0.00012519195979899495,
+      "loss": 1.1198,
+      "step": 58500
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 1.8101871013641357,
+      "learning_rate": 0.00012489045226130654,
+      "loss": 1.1725,
+      "step": 58600
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 4.7469305992126465,
+      "learning_rate": 0.00012458894472361807,
+      "loss": 1.1382,
+      "step": 58700
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 1.8046541213989258,
+      "learning_rate": 0.00012428743718592965,
+      "loss": 1.082,
+      "step": 58800
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 2.176015615463257,
+      "learning_rate": 0.0001239859296482412,
+      "loss": 1.1304,
+      "step": 58900
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 1.8910236358642578,
+      "learning_rate": 0.00012368442211055277,
+      "loss": 1.1638,
+      "step": 59000
+    },
+    {
+      "epoch": 2.95,
+      "eval_loss": 1.1192156076431274,
+      "eval_runtime": 41.8257,
+      "eval_samples_per_second": 23.909,
+      "eval_steps_per_second": 2.989,
+      "step": 59000
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 2.288358211517334,
+      "learning_rate": 0.0001233829145728643,
+      "loss": 1.1203,
+      "step": 59100
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 1.9389914274215698,
+      "learning_rate": 0.00012308140703517586,
+      "loss": 1.0892,
+      "step": 59200
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 2.1551334857940674,
+      "learning_rate": 0.00012277989949748742,
+      "loss": 1.1046,
+      "step": 59300
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 1.5200018882751465,
+      "learning_rate": 0.00012247839195979898,
+      "loss": 1.1373,
+      "step": 59400
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 2.45053768157959,
+      "learning_rate": 0.00012217688442211054,
+      "loss": 1.1403,
+      "step": 59500
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 2.767160177230835,
+      "learning_rate": 0.00012187537688442209,
+      "loss": 1.0693,
+      "step": 59600
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 2.3581674098968506,
+      "learning_rate": 0.00012157386934673366,
+      "loss": 1.125,
+      "step": 59700
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 1.4579651355743408,
+      "learning_rate": 0.00012127236180904521,
+      "loss": 1.127,
+      "step": 59800
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 4.08085298538208,
+      "learning_rate": 0.00012097085427135678,
+      "loss": 1.1539,
+      "step": 59900
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.5620448589324951,
+      "learning_rate": 0.00012066934673366833,
+      "loss": 1.1372,
+      "step": 60000
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 1.130272626876831,
+      "eval_runtime": 37.9665,
+      "eval_samples_per_second": 26.339,
+      "eval_steps_per_second": 3.292,
+      "step": 60000
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 3.270860433578491,
+      "learning_rate": 0.00012036783919597989,
+      "loss": 1.0761,
+      "step": 60100
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 2.5301287174224854,
+      "learning_rate": 0.00012006633165829145,
+      "loss": 1.0881,
+      "step": 60200
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 2.5292015075683594,
+      "learning_rate": 0.000119764824120603,
+      "loss": 1.046,
+      "step": 60300
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 2.8234751224517822,
+      "learning_rate": 0.00011946331658291456,
+      "loss": 1.0802,
+      "step": 60400
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 2.536975860595703,
+      "learning_rate": 0.00011916180904522612,
+      "loss": 1.0993,
+      "step": 60500
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 3.510464906692505,
+      "learning_rate": 0.00011886030150753767,
+      "loss": 1.1108,
+      "step": 60600
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 1.9273101091384888,
+      "learning_rate": 0.00011855879396984924,
+      "loss": 1.1081,
+      "step": 60700
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 2.1979687213897705,
+      "learning_rate": 0.00011825728643216079,
+      "loss": 1.1059,
+      "step": 60800
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 2.097529172897339,
+      "learning_rate": 0.00011795577889447236,
+      "loss": 1.1098,
+      "step": 60900
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 2.970689296722412,
+      "learning_rate": 0.00011765427135678391,
+      "loss": 1.0915,
+      "step": 61000
+    },
+    {
+      "epoch": 3.05,
+      "eval_loss": 1.0778993368148804,
+      "eval_runtime": 37.9552,
+      "eval_samples_per_second": 26.347,
+      "eval_steps_per_second": 3.293,
+      "step": 61000
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 2.3489325046539307,
+      "learning_rate": 0.00011735577889447236,
+      "loss": 1.1174,
+      "step": 61100
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 2.9280216693878174,
+      "learning_rate": 0.00011705427135678391,
+      "loss": 1.0977,
+      "step": 61200
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 2.231684446334839,
+      "learning_rate": 0.00011675276381909548,
+      "loss": 1.1004,
+      "step": 61300
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 1.8373113870620728,
+      "learning_rate": 0.00011645125628140703,
+      "loss": 1.109,
+      "step": 61400
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 3.446971893310547,
+      "learning_rate": 0.00011614974874371859,
+      "loss": 1.092,
+      "step": 61500
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 2.2681097984313965,
+      "learning_rate": 0.00011584824120603014,
+      "loss": 1.0901,
+      "step": 61600
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 2.173755407333374,
+      "learning_rate": 0.0001155467336683417,
+      "loss": 1.0638,
+      "step": 61700
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 3.3374030590057373,
+      "learning_rate": 0.00011524522613065325,
+      "loss": 1.1036,
+      "step": 61800
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 2.082169771194458,
+      "learning_rate": 0.00011494371859296481,
+      "loss": 1.0737,
+      "step": 61900
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 2.741830587387085,
+      "learning_rate": 0.00011464221105527637,
+      "loss": 1.0705,
+      "step": 62000
+    },
+    {
+      "epoch": 3.1,
+      "eval_loss": 1.079288125038147,
+      "eval_runtime": 37.9824,
+      "eval_samples_per_second": 26.328,
+      "eval_steps_per_second": 3.291,
+      "step": 62000
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 2.128262996673584,
+      "learning_rate": 0.00011434070351758793,
+      "loss": 1.0964,
+      "step": 62100
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 2.100025177001953,
+      "learning_rate": 0.00011403919597989949,
+      "loss": 1.0951,
+      "step": 62200
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 7.355963706970215,
+      "learning_rate": 0.00011373768844221103,
+      "loss": 1.128,
+      "step": 62300
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 2.6374123096466064,
+      "learning_rate": 0.0001134361809045226,
+      "loss": 1.0928,
+      "step": 62400
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 2.6389834880828857,
+      "learning_rate": 0.00011313467336683415,
+      "loss": 1.1067,
+      "step": 62500
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 3.367866277694702,
+      "learning_rate": 0.00011283316582914573,
+      "loss": 1.0719,
+      "step": 62600
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 2.0250422954559326,
+      "learning_rate": 0.00011253165829145727,
+      "loss": 1.0967,
+      "step": 62700
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 3.8763527870178223,
+      "learning_rate": 0.00011223015075376884,
+      "loss": 1.0819,
+      "step": 62800
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 2.7926995754241943,
+      "learning_rate": 0.00011192864321608039,
+      "loss": 1.1123,
+      "step": 62900
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 2.5031745433807373,
+      "learning_rate": 0.00011162713567839195,
+      "loss": 1.0725,
+      "step": 63000
+    },
+    {
+      "epoch": 3.15,
+      "eval_loss": 1.117138147354126,
+      "eval_runtime": 37.9757,
+      "eval_samples_per_second": 26.333,
+      "eval_steps_per_second": 3.292,
+      "step": 63000
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 2.086465835571289,
+      "learning_rate": 0.00011132562814070351,
+      "loss": 1.0588,
+      "step": 63100
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 3.295759439468384,
+      "learning_rate": 0.00011102412060301507,
+      "loss": 1.1175,
+      "step": 63200
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 2.666032075881958,
+      "learning_rate": 0.00011072261306532661,
+      "loss": 1.0963,
+      "step": 63300
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 1.8267697095870972,
+      "learning_rate": 0.00011042110552763819,
+      "loss": 1.0691,
+      "step": 63400
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 2.682745933532715,
+      "learning_rate": 0.00011011959798994973,
+      "loss": 1.0671,
+      "step": 63500
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 2.914111375808716,
+      "learning_rate": 0.00010982110552763819,
+      "loss": 1.0809,
+      "step": 63600
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 2.7258005142211914,
+      "learning_rate": 0.00010951959798994973,
+      "loss": 1.0527,
+      "step": 63700
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 2.646939992904663,
+      "learning_rate": 0.0001092180904522613,
+      "loss": 1.0523,
+      "step": 63800
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 2.107849359512329,
+      "learning_rate": 0.00010891658291457285,
+      "loss": 1.0629,
+      "step": 63900
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 1.9583218097686768,
+      "learning_rate": 0.00010861507537688442,
+      "loss": 1.065,
+      "step": 64000
+    },
+    {
+      "epoch": 3.2,
+      "eval_loss": 1.1121866703033447,
+      "eval_runtime": 37.9368,
+      "eval_samples_per_second": 26.36,
+      "eval_steps_per_second": 3.295,
+      "step": 64000
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 2.384493589401245,
+      "learning_rate": 0.00010831356783919597,
+      "loss": 1.0664,
+      "step": 64100
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 2.060441732406616,
+      "learning_rate": 0.00010801206030150753,
+      "loss": 1.0762,
+      "step": 64200
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 6.751837253570557,
+      "learning_rate": 0.00010771055276381909,
+      "loss": 1.0553,
+      "step": 64300
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 2.9765820503234863,
+      "learning_rate": 0.00010740904522613064,
+      "loss": 1.0636,
+      "step": 64400
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 2.2694509029388428,
+      "learning_rate": 0.00010710753768844221,
+      "loss": 1.1031,
+      "step": 64500
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 3.272937536239624,
+      "learning_rate": 0.00010680603015075375,
+      "loss": 1.1053,
+      "step": 64600
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 3.242722988128662,
+      "learning_rate": 0.00010650452261306531,
+      "loss": 1.1013,
+      "step": 64700
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 2.7234878540039062,
+      "learning_rate": 0.00010620301507537687,
+      "loss": 1.0428,
+      "step": 64800
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 2.30928373336792,
+      "learning_rate": 0.00010590150753768843,
+      "loss": 1.067,
+      "step": 64900
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 4.809457302093506,
+      "learning_rate": 0.00010559999999999998,
+      "loss": 1.053,
+      "step": 65000
+    },
+    {
+      "epoch": 3.25,
+      "eval_loss": 1.1082242727279663,
+      "eval_runtime": 37.9286,
+      "eval_samples_per_second": 26.365,
+      "eval_steps_per_second": 3.296,
+      "step": 65000
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 2.282684087753296,
+      "learning_rate": 0.00010529849246231155,
+      "loss": 1.0547,
+      "step": 65100
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 3.756114959716797,
+      "learning_rate": 0.0001049969849246231,
+      "loss": 1.0435,
+      "step": 65200
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 3.709932565689087,
+      "learning_rate": 0.00010469547738693467,
+      "loss": 1.0678,
+      "step": 65300
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 1.6080820560455322,
+      "learning_rate": 0.00010439396984924622,
+      "loss": 1.101,
+      "step": 65400
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 2.2617008686065674,
+      "learning_rate": 0.00010409246231155779,
+      "loss": 1.0729,
+      "step": 65500
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 3.1394824981689453,
+      "learning_rate": 0.00010379095477386933,
+      "loss": 1.0861,
+      "step": 65600
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 2.8208096027374268,
+      "learning_rate": 0.0001034894472361809,
+      "loss": 1.0535,
+      "step": 65700
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 2.7133829593658447,
+      "learning_rate": 0.00010318793969849245,
+      "loss": 1.0498,
+      "step": 65800
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 2.2674591541290283,
+      "learning_rate": 0.00010288643216080401,
+      "loss": 1.0861,
+      "step": 65900
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 2.238206386566162,
+      "learning_rate": 0.00010258492462311557,
+      "loss": 1.0557,
+      "step": 66000
+    },
+    {
+      "epoch": 3.3,
+      "eval_loss": 1.0877478122711182,
+      "eval_runtime": 37.9734,
+      "eval_samples_per_second": 26.334,
+      "eval_steps_per_second": 3.292,
+      "step": 66000
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 1.8776639699935913,
+      "learning_rate": 0.00010228643216080401,
+      "loss": 1.0898,
+      "step": 66100
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 2.540071725845337,
+      "learning_rate": 0.00010198492462311557,
+      "loss": 1.0437,
+      "step": 66200
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 3.616443157196045,
+      "learning_rate": 0.00010168341708542713,
+      "loss": 1.0698,
+      "step": 66300
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 2.866360902786255,
+      "learning_rate": 0.00010138190954773868,
+      "loss": 1.0666,
+      "step": 66400
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 3.1752941608428955,
+      "learning_rate": 0.00010108040201005025,
+      "loss": 1.0723,
+      "step": 66500
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 4.475529193878174,
+      "learning_rate": 0.0001007788944723618,
+      "loss": 1.105,
+      "step": 66600
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 2.9230782985687256,
+      "learning_rate": 0.00010047738693467337,
+      "loss": 1.0674,
+      "step": 66700
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 4.472579479217529,
+      "learning_rate": 0.00010017587939698491,
+      "loss": 1.0798,
+      "step": 66800
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 2.9080252647399902,
+      "learning_rate": 9.987437185929649e-05,
+      "loss": 1.0789,
+      "step": 66900
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 2.728170394897461,
+      "learning_rate": 9.957286432160803e-05,
+      "loss": 1.0771,
+      "step": 67000
+    },
+    {
+      "epoch": 3.35,
+      "eval_loss": 1.0558359622955322,
+      "eval_runtime": 37.9887,
+      "eval_samples_per_second": 26.324,
+      "eval_steps_per_second": 3.29,
+      "step": 67000
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 2.227384328842163,
+      "learning_rate": 9.927135678391958e-05,
+      "loss": 1.0336,
+      "step": 67100
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 2.5888235569000244,
+      "learning_rate": 9.896984924623115e-05,
+      "loss": 1.0525,
+      "step": 67200
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 1.9375131130218506,
+      "learning_rate": 9.86683417085427e-05,
+      "loss": 1.1218,
+      "step": 67300
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 1.8543367385864258,
+      "learning_rate": 9.836683417085426e-05,
+      "loss": 1.0761,
+      "step": 67400
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 3.050717353820801,
+      "learning_rate": 9.806532663316582e-05,
+      "loss": 1.07,
+      "step": 67500
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 3.321708917617798,
+      "learning_rate": 9.776381909547738e-05,
+      "loss": 1.0606,
+      "step": 67600
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 2.958376407623291,
+      "learning_rate": 9.746231155778894e-05,
+      "loss": 1.0608,
+      "step": 67700
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 2.215822219848633,
+      "learning_rate": 9.71608040201005e-05,
+      "loss": 1.0605,
+      "step": 67800
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 2.430649518966675,
+      "learning_rate": 9.685929648241204e-05,
+      "loss": 1.0783,
+      "step": 67900
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 2.4160895347595215,
+      "learning_rate": 9.655778894472361e-05,
+      "loss": 1.0783,
+      "step": 68000
+    },
+    {
+      "epoch": 3.4,
+      "eval_loss": 1.1083147525787354,
+      "eval_runtime": 37.9578,
+      "eval_samples_per_second": 26.345,
+      "eval_steps_per_second": 3.293,
+      "step": 68000
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 3.5485310554504395,
+      "learning_rate": 9.625628140703516e-05,
+      "loss": 1.0299,
+      "step": 68100
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 2.0450522899627686,
+      "learning_rate": 9.595477386934673e-05,
+      "loss": 1.0662,
+      "step": 68200
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 2.339768171310425,
+      "learning_rate": 9.565326633165828e-05,
+      "loss": 1.0781,
+      "step": 68300
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 2.055027484893799,
+      "learning_rate": 9.535477386934673e-05,
+      "loss": 1.0586,
+      "step": 68400
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 3.186723232269287,
+      "learning_rate": 9.505326633165828e-05,
+      "loss": 1.071,
+      "step": 68500
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 2.934070587158203,
+      "learning_rate": 9.475175879396985e-05,
+      "loss": 1.0474,
+      "step": 68600
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 4.080368995666504,
+      "learning_rate": 9.44502512562814e-05,
+      "loss": 1.0376,
+      "step": 68700
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 9.1796236038208,
+      "learning_rate": 9.415175879396985e-05,
+      "loss": 1.0362,
+      "step": 68800
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 2.9005532264709473,
+      "learning_rate": 9.38502512562814e-05,
+      "loss": 1.0581,
+      "step": 68900
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 2.2525532245635986,
+      "learning_rate": 9.354874371859296e-05,
+      "loss": 1.0664,
+      "step": 69000
+    },
+    {
+      "epoch": 3.45,
+      "eval_loss": 1.0977917909622192,
+      "eval_runtime": 37.9301,
+      "eval_samples_per_second": 26.364,
+      "eval_steps_per_second": 3.296,
+      "step": 69000
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 4.754021644592285,
+      "learning_rate": 9.324723618090452e-05,
+      "loss": 1.0512,
+      "step": 69100
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 2.1440653800964355,
+      "learning_rate": 9.294572864321607e-05,
+      "loss": 1.0653,
+      "step": 69200
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 2.278679609298706,
+      "learning_rate": 9.264422110552762e-05,
+      "loss": 1.0466,
+      "step": 69300
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 2.176259994506836,
+      "learning_rate": 9.23427135678392e-05,
+      "loss": 1.0664,
+      "step": 69400
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 2.2514779567718506,
+      "learning_rate": 9.204120603015074e-05,
+      "loss": 1.0597,
+      "step": 69500
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 3.136343002319336,
+      "learning_rate": 9.173969849246231e-05,
+      "loss": 1.0742,
+      "step": 69600
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 1.6031814813613892,
+      "learning_rate": 9.143819095477386e-05,
+      "loss": 1.0435,
+      "step": 69700
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 5.727216720581055,
+      "learning_rate": 9.113668341708543e-05,
+      "loss": 1.0837,
+      "step": 69800
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 2.909613609313965,
+      "learning_rate": 9.083517587939698e-05,
+      "loss": 1.0292,
+      "step": 69900
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 2.8508193492889404,
+      "learning_rate": 9.053366834170854e-05,
+      "loss": 1.0643,
+      "step": 70000
+    },
+    {
+      "epoch": 3.5,
+      "eval_loss": 1.0314569473266602,
+      "eval_runtime": 45.3565,
+      "eval_samples_per_second": 22.048,
+      "eval_steps_per_second": 2.756,
+      "step": 70000
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 1.3868812322616577,
+      "learning_rate": 9.02321608040201e-05,
+      "loss": 1.0719,
+      "step": 70100
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 2.059966564178467,
+      "learning_rate": 8.993065326633164e-05,
+      "loss": 1.0496,
+      "step": 70200
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 2.371212959289551,
+      "learning_rate": 8.962914572864322e-05,
+      "loss": 1.0416,
+      "step": 70300
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 5.051455497741699,
+      "learning_rate": 8.932763819095476e-05,
+      "loss": 1.0817,
+      "step": 70400
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 2.4436607360839844,
+      "learning_rate": 8.902613065326632e-05,
+      "loss": 1.0434,
+      "step": 70500
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 2.097843885421753,
+      "learning_rate": 8.872462311557788e-05,
+      "loss": 1.06,
+      "step": 70600
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 3.9826953411102295,
+      "learning_rate": 8.842311557788944e-05,
+      "loss": 1.0921,
+      "step": 70700
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 3.572988748550415,
+      "learning_rate": 8.812160804020099e-05,
+      "loss": 1.0503,
+      "step": 70800
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 3.2607603073120117,
+      "learning_rate": 8.782010050251256e-05,
+      "loss": 1.0308,
+      "step": 70900
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 2.152568817138672,
+      "learning_rate": 8.75185929648241e-05,
+      "loss": 1.0508,
+      "step": 71000
+    },
+    {
+      "epoch": 3.55,
+      "eval_loss": 1.035280704498291,
+      "eval_runtime": 44.3432,
+      "eval_samples_per_second": 22.551,
+      "eval_steps_per_second": 2.819,
+      "step": 71000
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 1.5636742115020752,
+      "learning_rate": 8.721708542713568e-05,
+      "loss": 1.0177,
+      "step": 71100
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 1.9526029825210571,
+      "learning_rate": 8.691557788944722e-05,
+      "loss": 1.0516,
+      "step": 71200
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 2.2071800231933594,
+      "learning_rate": 8.66140703517588e-05,
+      "loss": 1.034,
+      "step": 71300
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 2.6768360137939453,
+      "learning_rate": 8.631256281407034e-05,
+      "loss": 1.0642,
+      "step": 71400
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 1.6602065563201904,
+      "learning_rate": 8.60110552763819e-05,
+      "loss": 1.0389,
+      "step": 71500
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 2.439145565032959,
+      "learning_rate": 8.570954773869346e-05,
+      "loss": 1.0536,
+      "step": 71600
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 6.254899978637695,
+      "learning_rate": 8.54110552763819e-05,
+      "loss": 1.0141,
+      "step": 71700
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 1.8221715688705444,
+      "learning_rate": 8.510954773869346e-05,
+      "loss": 1.044,
+      "step": 71800
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 4.5664849281311035,
+      "learning_rate": 8.480804020100502e-05,
+      "loss": 1.0665,
+      "step": 71900
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 2.4576423168182373,
+      "learning_rate": 8.450653266331658e-05,
+      "loss": 1.0615,
+      "step": 72000
+    },
+    {
+      "epoch": 3.6,
+      "eval_loss": 1.031246542930603,
+      "eval_runtime": 42.8264,
+      "eval_samples_per_second": 23.35,
+      "eval_steps_per_second": 2.919,
+      "step": 72000
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 2.763627290725708,
+      "learning_rate": 8.420502512562814e-05,
+      "loss": 1.0333,
+      "step": 72100
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 1.6231377124786377,
+      "learning_rate": 8.390351758793968e-05,
+      "loss": 1.0572,
+      "step": 72200
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 1.9768860340118408,
+      "learning_rate": 8.360201005025126e-05,
+      "loss": 1.0423,
+      "step": 72300
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 2.292513132095337,
+      "learning_rate": 8.33005025125628e-05,
+      "loss": 1.0655,
+      "step": 72400
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 2.1181390285491943,
+      "learning_rate": 8.299899497487438e-05,
+      "loss": 1.0216,
+      "step": 72500
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 2.3944106101989746,
+      "learning_rate": 8.269748743718592e-05,
+      "loss": 1.0585,
+      "step": 72600
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 1.5745407342910767,
+      "learning_rate": 8.23959798994975e-05,
+      "loss": 1.0629,
+      "step": 72700
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 2.130709648132324,
+      "learning_rate": 8.209447236180904e-05,
+      "loss": 1.0027,
+      "step": 72800
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 3.202035427093506,
+      "learning_rate": 8.179296482412059e-05,
+      "loss": 1.0385,
+      "step": 72900
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 2.009536027908325,
+      "learning_rate": 8.149145728643216e-05,
+      "loss": 1.0471,
+      "step": 73000
+    },
+    {
+      "epoch": 3.65,
+      "eval_loss": 1.047244668006897,
+      "eval_runtime": 38.0986,
+      "eval_samples_per_second": 26.248,
+      "eval_steps_per_second": 3.281,
+      "step": 73000
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 5.239896774291992,
+      "learning_rate": 8.11899497487437e-05,
+      "loss": 1.0527,
+      "step": 73100
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 3.438692808151245,
+      "learning_rate": 8.088844221105527e-05,
+      "loss": 1.0198,
+      "step": 73200
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 2.0132901668548584,
+      "learning_rate": 8.058693467336682e-05,
+      "loss": 0.989,
+      "step": 73300
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 2.9494431018829346,
+      "learning_rate": 8.028542713567838e-05,
+      "loss": 1.0329,
+      "step": 73400
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 2.8393380641937256,
+      "learning_rate": 7.998391959798994e-05,
+      "loss": 1.043,
+      "step": 73500
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 3.039391279220581,
+      "learning_rate": 7.96824120603015e-05,
+      "loss": 1.0035,
+      "step": 73600
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 3.696676731109619,
+      "learning_rate": 7.938090452261305e-05,
+      "loss": 1.0472,
+      "step": 73700
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 2.8557331562042236,
+      "learning_rate": 7.907939698492462e-05,
+      "loss": 1.0665,
+      "step": 73800
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 3.7987170219421387,
+      "learning_rate": 7.877788944723617e-05,
+      "loss": 1.0233,
+      "step": 73900
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 1.9759894609451294,
+      "learning_rate": 7.847638190954774e-05,
+      "loss": 1.0303,
+      "step": 74000
+    },
+    {
+      "epoch": 3.7,
+      "eval_loss": 1.0124469995498657,
+      "eval_runtime": 38.8346,
+      "eval_samples_per_second": 25.75,
+      "eval_steps_per_second": 3.219,
+      "step": 74000
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 1.9311368465423584,
+      "learning_rate": 7.817487437185929e-05,
+      "loss": 1.0479,
+      "step": 74100
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 4.948327541351318,
+      "learning_rate": 7.787336683417086e-05,
+      "loss": 1.0197,
+      "step": 74200
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 2.6867167949676514,
+      "learning_rate": 7.75718592964824e-05,
+      "loss": 1.0209,
+      "step": 74300
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 1.8292616605758667,
+      "learning_rate": 7.727035175879396e-05,
+      "loss": 1.0257,
+      "step": 74400
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 3.2925384044647217,
+      "learning_rate": 7.696884422110552e-05,
+      "loss": 1.0635,
+      "step": 74500
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 2.2040624618530273,
+      "learning_rate": 7.666733668341708e-05,
+      "loss": 1.0285,
+      "step": 74600
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 2.1025142669677734,
+      "learning_rate": 7.636582914572863e-05,
+      "loss": 1.05,
+      "step": 74700
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 2.409148693084717,
+      "learning_rate": 7.60643216080402e-05,
+      "loss": 1.0638,
+      "step": 74800
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 3.284660577774048,
+      "learning_rate": 7.576281407035175e-05,
+      "loss": 1.0203,
+      "step": 74900
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 2.3454208374023438,
+      "learning_rate": 7.546130653266332e-05,
+      "loss": 1.0425,
+      "step": 75000
+    },
+    {
+      "epoch": 3.75,
+      "eval_loss": 1.0414044857025146,
+      "eval_runtime": 38.2892,
+      "eval_samples_per_second": 26.117,
+      "eval_steps_per_second": 3.265,
+      "step": 75000
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 2.6853275299072266,
+      "learning_rate": 7.515979899497487e-05,
+      "loss": 0.9762,
+      "step": 75100
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 1.439287543296814,
+      "learning_rate": 7.485829145728643e-05,
+      "loss": 0.9955,
+      "step": 75200
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 2.0795187950134277,
+      "learning_rate": 7.455678391959799e-05,
+      "loss": 1.0148,
+      "step": 75300
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 2.318300247192383,
+      "learning_rate": 7.425527638190955e-05,
+      "loss": 1.0368,
+      "step": 75400
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 2.979464054107666,
+      "learning_rate": 7.39537688442211e-05,
+      "loss": 1.0233,
+      "step": 75500
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 2.384615421295166,
+      "learning_rate": 7.365226130653266e-05,
+      "loss": 1.0183,
+      "step": 75600
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 2.2947332859039307,
+      "learning_rate": 7.335075376884421e-05,
+      "loss": 1.046,
+      "step": 75700
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 2.707266330718994,
+      "learning_rate": 7.304924623115577e-05,
+      "loss": 1.0145,
+      "step": 75800
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 1.8125189542770386,
+      "learning_rate": 7.275075376884422e-05,
+      "loss": 1.0508,
+      "step": 75900
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 1.833924412727356,
+      "learning_rate": 7.244924623115577e-05,
+      "loss": 1.051,
+      "step": 76000
+    },
+    {
+      "epoch": 3.8,
+      "eval_loss": 1.0207512378692627,
+      "eval_runtime": 38.1696,
+      "eval_samples_per_second": 26.199,
+      "eval_steps_per_second": 3.275,
+      "step": 76000
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 2.3891940116882324,
+      "learning_rate": 7.214773869346733e-05,
+      "loss": 1.0006,
+      "step": 76100
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 2.6063296794891357,
+      "learning_rate": 7.184623115577889e-05,
+      "loss": 1.0011,
+      "step": 76200
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 1.7001017332077026,
+      "learning_rate": 7.154472361809045e-05,
+      "loss": 1.0172,
+      "step": 76300
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 2.0134339332580566,
+      "learning_rate": 7.124321608040201e-05,
+      "loss": 1.0367,
+      "step": 76400
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 2.199366807937622,
+      "learning_rate": 7.094170854271357e-05,
+      "loss": 1.044,
+      "step": 76500
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 2.8991353511810303,
+      "learning_rate": 7.064020100502511e-05,
+      "loss": 1.0121,
+      "step": 76600
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 5.798487663269043,
+      "learning_rate": 7.033869346733667e-05,
+      "loss": 0.9734,
+      "step": 76700
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 2.8960068225860596,
+      "learning_rate": 7.003718592964823e-05,
+      "loss": 1.004,
+      "step": 76800
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 2.980179786682129,
+      "learning_rate": 6.973567839195979e-05,
+      "loss": 1.0118,
+      "step": 76900
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 6.4917988777160645,
+      "learning_rate": 6.943417085427135e-05,
+      "loss": 0.9682,
+      "step": 77000
+    },
+    {
+      "epoch": 3.85,
+      "eval_loss": 1.0282562971115112,
+      "eval_runtime": 38.0717,
+      "eval_samples_per_second": 26.266,
+      "eval_steps_per_second": 3.283,
+      "step": 77000
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 2.9224038124084473,
+      "learning_rate": 6.913266331658291e-05,
+      "loss": 1.0385,
+      "step": 77100
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 4.447437763214111,
+      "learning_rate": 6.883115577889447e-05,
+      "loss": 1.0388,
+      "step": 77200
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 2.2013559341430664,
+      "learning_rate": 6.852964824120603e-05,
+      "loss": 1.034,
+      "step": 77300
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 1.3720605373382568,
+      "learning_rate": 6.822814070351757e-05,
+      "loss": 1.0512,
+      "step": 77400
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 2.4448797702789307,
+      "learning_rate": 6.792663316582913e-05,
+      "loss": 1.0012,
+      "step": 77500
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 4.061469554901123,
+      "learning_rate": 6.762512562814069e-05,
+      "loss": 1.0144,
+      "step": 77600
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 1.62380850315094,
+      "learning_rate": 6.732361809045225e-05,
+      "loss": 1.0369,
+      "step": 77700
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 1.3728336095809937,
+      "learning_rate": 6.702211055276381e-05,
+      "loss": 1.0133,
+      "step": 77800
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 7.0939435958862305,
+      "learning_rate": 6.672060301507537e-05,
+      "loss": 0.9797,
+      "step": 77900
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 2.0842604637145996,
+      "learning_rate": 6.642211055276381e-05,
+      "loss": 1.0035,
+      "step": 78000
+    },
+    {
+      "epoch": 3.9,
+      "eval_loss": 1.0243637561798096,
+      "eval_runtime": 38.1566,
+      "eval_samples_per_second": 26.208,
+      "eval_steps_per_second": 3.276,
+      "step": 78000
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 3.6360020637512207,
+      "learning_rate": 6.612060301507537e-05,
+      "loss": 0.9969,
+      "step": 78100
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 2.5551681518554688,
+      "learning_rate": 6.581909547738693e-05,
+      "loss": 1.0203,
+      "step": 78200
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 6.86871862411499,
+      "learning_rate": 6.551758793969849e-05,
+      "loss": 1.0472,
+      "step": 78300
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 2.3950083255767822,
+      "learning_rate": 6.521608040201005e-05,
+      "loss": 1.0167,
+      "step": 78400
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 1.422188401222229,
+      "learning_rate": 6.491457286432161e-05,
+      "loss": 0.9968,
+      "step": 78500
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 2.186511993408203,
+      "learning_rate": 6.461306532663317e-05,
+      "loss": 1.0113,
+      "step": 78600
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 1.764722228050232,
+      "learning_rate": 6.431155778894471e-05,
+      "loss": 0.983,
+      "step": 78700
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 4.928635597229004,
+      "learning_rate": 6.401005025125627e-05,
+      "loss": 1.0164,
+      "step": 78800
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 2.1061389446258545,
+      "learning_rate": 6.370854271356783e-05,
+      "loss": 1.0171,
+      "step": 78900
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 4.193387985229492,
+      "learning_rate": 6.340703517587939e-05,
+      "loss": 1.0072,
+      "step": 79000
+    },
+    {
+      "epoch": 3.95,
+      "eval_loss": 1.00971519947052,
+      "eval_runtime": 38.1263,
+      "eval_samples_per_second": 26.229,
+      "eval_steps_per_second": 3.279,
+      "step": 79000
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 2.4844706058502197,
+      "learning_rate": 6.310552763819095e-05,
+      "loss": 1.0064,
+      "step": 79100
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 5.7934746742248535,
+      "learning_rate": 6.280402010050251e-05,
+      "loss": 0.9509,
+      "step": 79200
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 3.7046196460723877,
+      "learning_rate": 6.250251256281406e-05,
+      "loss": 1.0139,
+      "step": 79300
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 1.9528000354766846,
+      "learning_rate": 6.220100502512562e-05,
+      "loss": 1.0214,
+      "step": 79400
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 3.4000682830810547,
+      "learning_rate": 6.189949748743718e-05,
+      "loss": 1.006,
+      "step": 79500
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 3.152561664581299,
+      "learning_rate": 6.159798994974874e-05,
+      "loss": 1.0288,
+      "step": 79600
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 3.774915933609009,
+      "learning_rate": 6.12964824120603e-05,
+      "loss": 1.022,
+      "step": 79700
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 2.291813373565674,
+      "learning_rate": 6.0994974874371854e-05,
+      "loss": 0.9845,
+      "step": 79800
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 3.019514560699463,
+      "learning_rate": 6.0693467336683413e-05,
+      "loss": 1.0246,
+      "step": 79900
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 2.4409408569335938,
+      "learning_rate": 6.0391959798994966e-05,
+      "loss": 0.9951,
+      "step": 80000
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 0.9992234110832214,
+      "eval_runtime": 39.3867,
+      "eval_samples_per_second": 25.389,
+      "eval_steps_per_second": 3.174,
+      "step": 80000
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 1.4257367849349976,
+      "learning_rate": 6.0090452261306526e-05,
+      "loss": 0.9763,
+      "step": 80100
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 4.97927713394165,
+      "learning_rate": 5.9788944723618085e-05,
+      "loss": 0.9417,
+      "step": 80200
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 2.8552098274230957,
+      "learning_rate": 5.9487437185929644e-05,
+      "loss": 0.9591,
+      "step": 80300
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": Infinity,
+      "learning_rate": 5.9188944723618084e-05,
+      "loss": 0.9783,
+      "step": 80400
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 3.83720064163208,
+      "learning_rate": 5.8887437185929643e-05,
+      "loss": 0.9607,
+      "step": 80500
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 2.607973337173462,
+      "learning_rate": 5.85859296482412e-05,
+      "loss": 0.9556,
+      "step": 80600
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 3.51914381980896,
+      "learning_rate": 5.8284422110552756e-05,
+      "loss": 0.9371,
+      "step": 80700
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 2.0518856048583984,
+      "learning_rate": 5.7982914572864315e-05,
+      "loss": 1.0154,
+      "step": 80800
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 3.5824625492095947,
+      "learning_rate": 5.7681407035175874e-05,
+      "loss": 0.9894,
+      "step": 80900
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 7.991865634918213,
+      "learning_rate": 5.7379899497487434e-05,
+      "loss": 0.9719,
+      "step": 81000
+    },
+    {
+      "epoch": 4.05,
+      "eval_loss": 1.0105689764022827,
+      "eval_runtime": 38.1347,
+      "eval_samples_per_second": 26.223,
+      "eval_steps_per_second": 3.278,
+      "step": 81000
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 1.6757104396820068,
+      "learning_rate": 5.707839195979899e-05,
+      "loss": 0.9526,
+      "step": 81100
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 3.1675045490264893,
+      "learning_rate": 5.677688442211055e-05,
+      "loss": 0.9798,
+      "step": 81200
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 2.8390209674835205,
+      "learning_rate": 5.6475376884422105e-05,
+      "loss": 0.9455,
+      "step": 81300
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 2.2900238037109375,
+      "learning_rate": 5.6173869346733665e-05,
+      "loss": 1.0016,
+      "step": 81400
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 2.4220378398895264,
+      "learning_rate": 5.5872361809045224e-05,
+      "loss": 0.9681,
+      "step": 81500
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 2.7175300121307373,
+      "learning_rate": 5.5570854271356784e-05,
+      "loss": 0.9822,
+      "step": 81600
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 3.7499475479125977,
+      "learning_rate": 5.526934673366834e-05,
+      "loss": 0.9501,
+      "step": 81700
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 2.1566553115844727,
+      "learning_rate": 5.4967839195979896e-05,
+      "loss": 0.9601,
+      "step": 81800
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 2.080754280090332,
+      "learning_rate": 5.466633165829145e-05,
+      "loss": 0.954,
+      "step": 81900
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 3.1466102600097656,
+      "learning_rate": 5.436482412060301e-05,
+      "loss": 0.9896,
+      "step": 82000
+    },
+    {
+      "epoch": 4.1,
+      "eval_loss": 1.0087724924087524,
+      "eval_runtime": 37.9931,
+      "eval_samples_per_second": 26.321,
+      "eval_steps_per_second": 3.29,
+      "step": 82000
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 4.262351989746094,
+      "learning_rate": 5.406331658291457e-05,
+      "loss": 0.9454,
+      "step": 82100
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 1.9488756656646729,
+      "learning_rate": 5.376180904522612e-05,
+      "loss": 0.9494,
+      "step": 82200
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 1.6786818504333496,
+      "learning_rate": 5.346030150753768e-05,
+      "loss": 0.9241,
+      "step": 82300
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 2.143955945968628,
+      "learning_rate": 5.315879396984924e-05,
+      "loss": 0.9958,
+      "step": 82400
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 3.6211471557617188,
+      "learning_rate": 5.286030150753768e-05,
+      "loss": 0.9641,
+      "step": 82500
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 4.066643238067627,
+      "learning_rate": 5.255879396984924e-05,
+      "loss": 0.9698,
+      "step": 82600
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 2.151590585708618,
+      "learning_rate": 5.22572864321608e-05,
+      "loss": 0.9388,
+      "step": 82700
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 4.644803524017334,
+      "learning_rate": 5.195577889447236e-05,
+      "loss": 0.9141,
+      "step": 82800
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 2.652754068374634,
+      "learning_rate": 5.1654271356783916e-05,
+      "loss": 0.9592,
+      "step": 82900
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 4.528812885284424,
+      "learning_rate": 5.135276381909547e-05,
+      "loss": 0.9778,
+      "step": 83000
+    },
+    {
+      "epoch": 4.15,
+      "eval_loss": 0.9974797368049622,
+      "eval_runtime": 38.0893,
+      "eval_samples_per_second": 26.254,
+      "eval_steps_per_second": 3.282,
+      "step": 83000
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 2.625786542892456,
+      "learning_rate": 5.105125628140703e-05,
+      "loss": 0.9594,
+      "step": 83100
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 3.7137229442596436,
+      "learning_rate": 5.074974874371859e-05,
+      "loss": 0.9462,
+      "step": 83200
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 6.682472229003906,
+      "learning_rate": 5.044824120603015e-05,
+      "loss": 0.9301,
+      "step": 83300
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 2.7188687324523926,
+      "learning_rate": 5.014673366834171e-05,
+      "loss": 0.9801,
+      "step": 83400
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 2.7037341594696045,
+      "learning_rate": 4.984522613065326e-05,
+      "loss": 0.9475,
+      "step": 83500
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 2.815229654312134,
+      "learning_rate": 4.954371859296482e-05,
+      "loss": 0.9012,
+      "step": 83600
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 2.7187130451202393,
+      "learning_rate": 4.924221105527638e-05,
+      "loss": 0.9199,
+      "step": 83700
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 1.6610496044158936,
+      "learning_rate": 4.894070351758794e-05,
+      "loss": 0.9321,
+      "step": 83800
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 2.1496291160583496,
+      "learning_rate": 4.86391959798995e-05,
+      "loss": 0.9003,
+      "step": 83900
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 2.9933974742889404,
+      "learning_rate": 4.833768844221105e-05,
+      "loss": 0.9467,
+      "step": 84000
+    },
+    {
+      "epoch": 4.2,
+      "eval_loss": 0.9802306890487671,
+      "eval_runtime": 38.0487,
+      "eval_samples_per_second": 26.282,
+      "eval_steps_per_second": 3.285,
+      "step": 84000
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 4.368553161621094,
+      "learning_rate": 4.803618090452261e-05,
+      "loss": 0.921,
+      "step": 84100
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 4.087899684906006,
+      "learning_rate": 4.773467336683417e-05,
+      "loss": 0.9413,
+      "step": 84200
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 1.8541690111160278,
+      "learning_rate": 4.743316582914573e-05,
+      "loss": 0.9657,
+      "step": 84300
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 2.6514675617218018,
+      "learning_rate": 4.713165829145729e-05,
+      "loss": 0.9645,
+      "step": 84400
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 3.2329466342926025,
+      "learning_rate": 4.683015075376885e-05,
+      "loss": 0.9465,
+      "step": 84500
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 2.358675241470337,
+      "learning_rate": 4.652864321608039e-05,
+      "loss": 0.9644,
+      "step": 84600
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 3.6738836765289307,
+      "learning_rate": 4.6230150753768846e-05,
+      "loss": 0.9357,
+      "step": 84700
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 2.8447327613830566,
+      "learning_rate": 4.59286432160804e-05,
+      "loss": 0.9308,
+      "step": 84800
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 1.6326079368591309,
+      "learning_rate": 4.562713567839195e-05,
+      "loss": 0.9068,
+      "step": 84900
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 2.3545360565185547,
+      "learning_rate": 4.532562814070351e-05,
+      "loss": 0.9436,
+      "step": 85000
+    },
+    {
+      "epoch": 4.25,
+      "eval_loss": 0.9844674468040466,
+      "eval_runtime": 38.274,
+      "eval_samples_per_second": 26.127,
+      "eval_steps_per_second": 3.266,
+      "step": 85000
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 3.2402210235595703,
+      "learning_rate": 4.502412060301507e-05,
+      "loss": 0.9313,
+      "step": 85100
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 3.3900952339172363,
+      "learning_rate": 4.472261306532662e-05,
+      "loss": 0.9385,
+      "step": 85200
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 3.8531854152679443,
+      "learning_rate": 4.442110552763818e-05,
+      "loss": 0.9292,
+      "step": 85300
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 2.3123373985290527,
+      "learning_rate": 4.411959798994974e-05,
+      "loss": 0.9544,
+      "step": 85400
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 2.5710906982421875,
+      "learning_rate": 4.38180904522613e-05,
+      "loss": 0.9591,
+      "step": 85500
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 3.4481329917907715,
+      "learning_rate": 4.351658291457286e-05,
+      "loss": 0.9281,
+      "step": 85600
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 1.7887803316116333,
+      "learning_rate": 4.321507537688442e-05,
+      "loss": 0.9371,
+      "step": 85700
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 6.177557945251465,
+      "learning_rate": 4.291356783919597e-05,
+      "loss": 0.9154,
+      "step": 85800
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 3.0554301738739014,
+      "learning_rate": 4.261206030150753e-05,
+      "loss": 0.9483,
+      "step": 85900
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 2.0133023262023926,
+      "learning_rate": 4.231055276381909e-05,
+      "loss": 0.9557,
+      "step": 86000
+    },
+    {
+      "epoch": 4.3,
+      "eval_loss": 0.9593837261199951,
+      "eval_runtime": 38.1446,
+      "eval_samples_per_second": 26.216,
+      "eval_steps_per_second": 3.277,
+      "step": 86000
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 2.1396610736846924,
+      "learning_rate": 4.200904522613065e-05,
+      "loss": 0.9643,
+      "step": 86100
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 2.709627628326416,
+      "learning_rate": 4.170753768844221e-05,
+      "loss": 0.9365,
+      "step": 86200
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 4.406678199768066,
+      "learning_rate": 4.1406030150753764e-05,
+      "loss": 0.9553,
+      "step": 86300
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 4.822593688964844,
+      "learning_rate": 4.110452261306532e-05,
+      "loss": 0.9213,
+      "step": 86400
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 4.148794651031494,
+      "learning_rate": 4.080301507537688e-05,
+      "loss": 0.9808,
+      "step": 86500
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 3.7028510570526123,
+      "learning_rate": 4.050150753768844e-05,
+      "loss": 0.9331,
+      "step": 86600
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 2.314500093460083,
+      "learning_rate": 4.02e-05,
+      "loss": 0.9551,
+      "step": 86700
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 3.741234302520752,
+      "learning_rate": 3.9898492462311554e-05,
+      "loss": 0.9053,
+      "step": 86800
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 3.7346441745758057,
+      "learning_rate": 3.9596984924623113e-05,
+      "loss": 0.9517,
+      "step": 86900
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 1.324827790260315,
+      "learning_rate": 3.929849246231156e-05,
+      "loss": 0.9764,
+      "step": 87000
+    },
+    {
+      "epoch": 4.35,
+      "eval_loss": 1.0139998197555542,
+      "eval_runtime": 38.1639,
+      "eval_samples_per_second": 26.203,
+      "eval_steps_per_second": 3.275,
+      "step": 87000
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 5.19126033782959,
+      "learning_rate": 3.899698492462311e-05,
+      "loss": 0.9366,
+      "step": 87100
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 2.899726629257202,
+      "learning_rate": 3.869547738693467e-05,
+      "loss": 0.9555,
+      "step": 87200
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 1.9099615812301636,
+      "learning_rate": 3.839396984924623e-05,
+      "loss": 0.9033,
+      "step": 87300
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 1.5814082622528076,
+      "learning_rate": 3.809246231155779e-05,
+      "loss": 0.9978,
+      "step": 87400
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 3.4520106315612793,
+      "learning_rate": 3.779095477386935e-05,
+      "loss": 0.9343,
+      "step": 87500
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 3.0876681804656982,
+      "learning_rate": 3.74894472361809e-05,
+      "loss": 0.9094,
+      "step": 87600
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 3.5139119625091553,
+      "learning_rate": 3.718793969849246e-05,
+      "loss": 0.8677,
+      "step": 87700
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 2.003330945968628,
+      "learning_rate": 3.6886432160804015e-05,
+      "loss": 0.9351,
+      "step": 87800
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 2.259235382080078,
+      "learning_rate": 3.6584924623115574e-05,
+      "loss": 0.9388,
+      "step": 87900
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 2.2141153812408447,
+      "learning_rate": 3.6283417085427134e-05,
+      "loss": 0.9169,
+      "step": 88000
+    },
+    {
+      "epoch": 4.4,
+      "eval_loss": 0.9528889060020447,
+      "eval_runtime": 38.0305,
+      "eval_samples_per_second": 26.295,
+      "eval_steps_per_second": 3.287,
+      "step": 88000
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 4.264975547790527,
+      "learning_rate": 3.5981909547738693e-05,
+      "loss": 0.9309,
+      "step": 88100
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 4.431647777557373,
+      "learning_rate": 3.5680402010050246e-05,
+      "loss": 0.9035,
+      "step": 88200
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 2.326883316040039,
+      "learning_rate": 3.5378894472361806e-05,
+      "loss": 0.904,
+      "step": 88300
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 2.6951944828033447,
+      "learning_rate": 3.5077386934673365e-05,
+      "loss": 0.9195,
+      "step": 88400
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 1.8017208576202393,
+      "learning_rate": 3.477587939698492e-05,
+      "loss": 0.9398,
+      "step": 88500
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 3.8392789363861084,
+      "learning_rate": 3.447437185929648e-05,
+      "loss": 0.9591,
+      "step": 88600
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 2.541273593902588,
+      "learning_rate": 3.4172864321608037e-05,
+      "loss": 0.9054,
+      "step": 88700
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 2.7736191749572754,
+      "learning_rate": 3.3874371859296476e-05,
+      "loss": 0.9473,
+      "step": 88800
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 2.660540819168091,
+      "learning_rate": 3.3572864321608036e-05,
+      "loss": 0.9582,
+      "step": 88900
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 3.161513328552246,
+      "learning_rate": 3.3271356783919595e-05,
+      "loss": 0.8943,
+      "step": 89000
+    },
+    {
+      "epoch": 4.45,
+      "eval_loss": 0.9552559852600098,
+      "eval_runtime": 38.1158,
+      "eval_samples_per_second": 26.236,
+      "eval_steps_per_second": 3.279,
+      "step": 89000
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 4.881318092346191,
+      "learning_rate": 3.2969849246231154e-05,
+      "loss": 0.9053,
+      "step": 89100
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 1.7572602033615112,
+      "learning_rate": 3.2668341708542714e-05,
+      "loss": 0.9364,
+      "step": 89200
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 3.067507743835449,
+      "learning_rate": 3.2366834170854267e-05,
+      "loss": 0.9355,
+      "step": 89300
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 3.1982858180999756,
+      "learning_rate": 3.2065326633165826e-05,
+      "loss": 0.9333,
+      "step": 89400
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 3.596789598464966,
+      "learning_rate": 3.1763819095477385e-05,
+      "loss": 0.8978,
+      "step": 89500
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 5.035818576812744,
+      "learning_rate": 3.1462311557788945e-05,
+      "loss": 0.9337,
+      "step": 89600
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 3.149653673171997,
+      "learning_rate": 3.11608040201005e-05,
+      "loss": 0.9515,
+      "step": 89700
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 3.4601404666900635,
+      "learning_rate": 3.085929648241206e-05,
+      "loss": 0.9021,
+      "step": 89800
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 2.6317124366760254,
+      "learning_rate": 3.0557788944723616e-05,
+      "loss": 0.9559,
+      "step": 89900
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 2.667861223220825,
+      "learning_rate": 3.0256281407035173e-05,
+      "loss": 0.9341,
+      "step": 90000
+    },
+    {
+      "epoch": 4.5,
+      "eval_loss": 0.9440233111381531,
+      "eval_runtime": 38.0809,
+      "eval_samples_per_second": 26.26,
+      "eval_steps_per_second": 3.282,
+      "step": 90000
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 3.903172016143799,
+      "learning_rate": 2.9954773869346732e-05,
+      "loss": 0.8857,
+      "step": 90100
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 3.9286229610443115,
+      "learning_rate": 2.9653266331658288e-05,
+      "loss": 0.9119,
+      "step": 90200
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 2.812256336212158,
+      "learning_rate": 2.9351758793969847e-05,
+      "loss": 0.9026,
+      "step": 90300
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 2.2835099697113037,
+      "learning_rate": 2.9050251256281404e-05,
+      "loss": 0.885,
+      "step": 90400
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 3.383111000061035,
+      "learning_rate": 2.8748743718592963e-05,
+      "loss": 0.8838,
+      "step": 90500
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 2.7682292461395264,
+      "learning_rate": 2.8447236180904522e-05,
+      "loss": 0.9139,
+      "step": 90600
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 6.3915019035339355,
+      "learning_rate": 2.814572864321608e-05,
+      "loss": 0.9188,
+      "step": 90700
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 5.53504753112793,
+      "learning_rate": 2.7844221105527635e-05,
+      "loss": 0.9118,
+      "step": 90800
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 2.5919177532196045,
+      "learning_rate": 2.754271356783919e-05,
+      "loss": 0.8844,
+      "step": 90900
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 1.9481797218322754,
+      "learning_rate": 2.724120603015075e-05,
+      "loss": 0.9192,
+      "step": 91000
+    },
+    {
+      "epoch": 4.55,
+      "eval_loss": 0.9217103123664856,
+      "eval_runtime": 38.1169,
+      "eval_samples_per_second": 26.235,
+      "eval_steps_per_second": 3.279,
+      "step": 91000
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 2.1429965496063232,
+      "learning_rate": 2.693969849246231e-05,
+      "loss": 0.8889,
+      "step": 91100
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 3.4818546772003174,
+      "learning_rate": 2.6638190954773866e-05,
+      "loss": 0.8932,
+      "step": 91200
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 2.3813984394073486,
+      "learning_rate": 2.6336683417085425e-05,
+      "loss": 0.9154,
+      "step": 91300
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 2.4688570499420166,
+      "learning_rate": 2.6035175879396984e-05,
+      "loss": 0.9344,
+      "step": 91400
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 4.330790996551514,
+      "learning_rate": 2.573366834170854e-05,
+      "loss": 0.9137,
+      "step": 91500
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 2.8123939037323,
+      "learning_rate": 2.54321608040201e-05,
+      "loss": 0.9041,
+      "step": 91600
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 2.1815638542175293,
+      "learning_rate": 2.5130653266331656e-05,
+      "loss": 0.8606,
+      "step": 91700
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 3.3489341735839844,
+      "learning_rate": 2.4829145728643216e-05,
+      "loss": 0.934,
+      "step": 91800
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 2.9650094509124756,
+      "learning_rate": 2.4527638190954775e-05,
+      "loss": 0.8893,
+      "step": 91900
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 3.541456460952759,
+      "learning_rate": 2.4226130653266328e-05,
+      "loss": 0.9239,
+      "step": 92000
+    },
+    {
+      "epoch": 4.6,
+      "eval_loss": 0.9656698107719421,
+      "eval_runtime": 38.5991,
+      "eval_samples_per_second": 25.907,
+      "eval_steps_per_second": 3.238,
+      "step": 92000
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 3.1648945808410645,
+      "learning_rate": 2.3924623115577887e-05,
+      "loss": 0.8777,
+      "step": 92100
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 8.632335662841797,
+      "learning_rate": 2.3623115577889443e-05,
+      "loss": 0.9047,
+      "step": 92200
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 2.9412002563476562,
+      "learning_rate": 2.3321608040201003e-05,
+      "loss": 0.8964,
+      "step": 92300
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 2.7501888275146484,
+      "learning_rate": 2.3020100502512562e-05,
+      "loss": 0.9303,
+      "step": 92400
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 3.36631178855896,
+      "learning_rate": 2.2718592964824118e-05,
+      "loss": 0.8987,
+      "step": 92500
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 2.6061251163482666,
+      "learning_rate": 2.2417085427135678e-05,
+      "loss": 0.8981,
+      "step": 92600
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 3.9636521339416504,
+      "learning_rate": 2.2115577889447234e-05,
+      "loss": 0.893,
+      "step": 92700
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 3.2085049152374268,
+      "learning_rate": 2.1814070351758793e-05,
+      "loss": 0.9298,
+      "step": 92800
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 2.590059995651245,
+      "learning_rate": 2.1512562814070353e-05,
+      "loss": 0.9118,
+      "step": 92900
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 4.868690013885498,
+      "learning_rate": 2.121105527638191e-05,
+      "loss": 0.8873,
+      "step": 93000
+    },
+    {
+      "epoch": 4.65,
+      "eval_loss": 0.918121337890625,
+      "eval_runtime": 38.3542,
+      "eval_samples_per_second": 26.073,
+      "eval_steps_per_second": 3.259,
+      "step": 93000
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 4.0143303871154785,
+      "learning_rate": 2.0909547738693465e-05,
+      "loss": 0.871,
+      "step": 93100
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 4.423349857330322,
+      "learning_rate": 2.060804020100502e-05,
+      "loss": 0.9232,
+      "step": 93200
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 3.6609606742858887,
+      "learning_rate": 2.030653266331658e-05,
+      "loss": 0.8782,
+      "step": 93300
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 3.252089738845825,
+      "learning_rate": 2.0008040201005026e-05,
+      "loss": 0.9232,
+      "step": 93400
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 2.8783979415893555,
+      "learning_rate": 1.970653266331658e-05,
+      "loss": 0.8539,
+      "step": 93500
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 5.381927967071533,
+      "learning_rate": 1.940502512562814e-05,
+      "loss": 0.9263,
+      "step": 93600
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 3.1031525135040283,
+      "learning_rate": 1.9103517587939695e-05,
+      "loss": 0.9095,
+      "step": 93700
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 2.668039321899414,
+      "learning_rate": 1.8802010050251254e-05,
+      "loss": 0.892,
+      "step": 93800
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 2.6661875247955322,
+      "learning_rate": 1.8500502512562814e-05,
+      "loss": 0.8944,
+      "step": 93900
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 3.5291526317596436,
+      "learning_rate": 1.819899497487437e-05,
+      "loss": 0.9074,
+      "step": 94000
+    },
+    {
+      "epoch": 4.7,
+      "eval_loss": 0.9208371639251709,
+      "eval_runtime": 38.4003,
+      "eval_samples_per_second": 26.041,
+      "eval_steps_per_second": 3.255,
+      "step": 94000
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 4.160482883453369,
+      "learning_rate": 1.789748743718593e-05,
+      "loss": 0.9045,
+      "step": 94100
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 3.8051962852478027,
+      "learning_rate": 1.7595979899497485e-05,
+      "loss": 0.899,
+      "step": 94200
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 3.431490898132324,
+      "learning_rate": 1.7294472361809045e-05,
+      "loss": 0.8577,
+      "step": 94300
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 2.356250524520874,
+      "learning_rate": 1.69929648241206e-05,
+      "loss": 0.9204,
+      "step": 94400
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 5.237595081329346,
+      "learning_rate": 1.669145728643216e-05,
+      "loss": 0.8973,
+      "step": 94500
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 5.023568153381348,
+      "learning_rate": 1.6389949748743716e-05,
+      "loss": 0.9064,
+      "step": 94600
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 6.610247611999512,
+      "learning_rate": 1.6088442211055276e-05,
+      "loss": 0.858,
+      "step": 94700
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 2.1937615871429443,
+      "learning_rate": 1.5786934673366835e-05,
+      "loss": 0.872,
+      "step": 94800
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 4.40328311920166,
+      "learning_rate": 1.548542713567839e-05,
+      "loss": 0.88,
+      "step": 94900
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 3.0487658977508545,
+      "learning_rate": 1.5183919597989947e-05,
+      "loss": 0.8779,
+      "step": 95000
+    },
+    {
+      "epoch": 4.75,
+      "eval_loss": 0.9459323883056641,
+      "eval_runtime": 38.1338,
+      "eval_samples_per_second": 26.223,
+      "eval_steps_per_second": 3.278,
+      "step": 95000
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 3.8922808170318604,
+      "learning_rate": 1.4882412060301507e-05,
+      "loss": 0.9075,
+      "step": 95100
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 3.232625722885132,
+      "learning_rate": 1.4580904522613064e-05,
+      "loss": 0.869,
+      "step": 95200
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 8.73833179473877,
+      "learning_rate": 1.4279396984924622e-05,
+      "loss": 0.8741,
+      "step": 95300
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 4.5711846351623535,
+      "learning_rate": 1.397788944723618e-05,
+      "loss": 0.8976,
+      "step": 95400
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 4.647241115570068,
+      "learning_rate": 1.3676381909547736e-05,
+      "loss": 0.8392,
+      "step": 95500
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 4.90078067779541,
+      "learning_rate": 1.337788944723618e-05,
+      "loss": 0.8739,
+      "step": 95600
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 3.1595067977905273,
+      "learning_rate": 1.3076381909547738e-05,
+      "loss": 0.8398,
+      "step": 95700
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 2.488835096359253,
+      "learning_rate": 1.2774874371859296e-05,
+      "loss": 0.868,
+      "step": 95800
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 4.495543003082275,
+      "learning_rate": 1.2473366834170852e-05,
+      "loss": 0.8872,
+      "step": 95900
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 3.673161268234253,
+      "learning_rate": 1.217185929648241e-05,
+      "loss": 0.8824,
+      "step": 96000
+    },
+    {
+      "epoch": 4.8,
+      "eval_loss": 0.910308301448822,
+      "eval_runtime": 38.0891,
+      "eval_samples_per_second": 26.254,
+      "eval_steps_per_second": 3.282,
+      "step": 96000
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 5.159984111785889,
+      "learning_rate": 1.187035175879397e-05,
+      "loss": 0.8672,
+      "step": 96100
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 2.706937551498413,
+      "learning_rate": 1.1568844221105527e-05,
+      "loss": 0.8914,
+      "step": 96200
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 3.727692127227783,
+      "learning_rate": 1.1267336683417085e-05,
+      "loss": 0.8485,
+      "step": 96300
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 2.665670156478882,
+      "learning_rate": 1.0965829145728641e-05,
+      "loss": 0.8695,
+      "step": 96400
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 5.077518463134766,
+      "learning_rate": 1.0664321608040199e-05,
+      "loss": 0.8767,
+      "step": 96500
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 3.4337048530578613,
+      "learning_rate": 1.0362814070351758e-05,
+      "loss": 0.8673,
+      "step": 96600
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 3.231494665145874,
+      "learning_rate": 1.0061306532663316e-05,
+      "loss": 0.8767,
+      "step": 96700
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 4.2955002784729,
+      "learning_rate": 9.759798994974874e-06,
+      "loss": 0.8645,
+      "step": 96800
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 6.2070698738098145,
+      "learning_rate": 9.458291457286431e-06,
+      "loss": 0.8683,
+      "step": 96900
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 3.6267805099487305,
+      "learning_rate": 9.159798994974874e-06,
+      "loss": 0.907,
+      "step": 97000
+    },
+    {
+      "epoch": 4.85,
+      "eval_loss": 0.9255304932594299,
+      "eval_runtime": 38.1396,
+      "eval_samples_per_second": 26.219,
+      "eval_steps_per_second": 3.277,
+      "step": 97000
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 4.985959529876709,
+      "learning_rate": 8.858291457286432e-06,
+      "loss": 0.8615,
+      "step": 97100
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 4.538032531738281,
+      "learning_rate": 8.556783919597988e-06,
+      "loss": 0.8519,
+      "step": 97200
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 6.562105178833008,
+      "learning_rate": 8.255276381909548e-06,
+      "loss": 0.8888,
+      "step": 97300
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 2.922360897064209,
+      "learning_rate": 7.953768844221105e-06,
+      "loss": 0.8784,
+      "step": 97400
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 3.8349783420562744,
+      "learning_rate": 7.652261306532663e-06,
+      "loss": 0.8962,
+      "step": 97500
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 2.096787929534912,
+      "learning_rate": 7.350753768844221e-06,
+      "loss": 0.9088,
+      "step": 97600
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 2.512312650680542,
+      "learning_rate": 7.0492462311557786e-06,
+      "loss": 0.8816,
+      "step": 97700
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 4.749015808105469,
+      "learning_rate": 6.7477386934673355e-06,
+      "loss": 0.8791,
+      "step": 97800
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 3.5753800868988037,
+      "learning_rate": 6.446231155778894e-06,
+      "loss": 0.8414,
+      "step": 97900
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 2.849839210510254,
+      "learning_rate": 6.144723618090452e-06,
+      "loss": 0.873,
+      "step": 98000
+    },
+    {
+      "epoch": 4.9,
+      "eval_loss": 0.8922821283340454,
+      "eval_runtime": 38.1228,
+      "eval_samples_per_second": 26.231,
+      "eval_steps_per_second": 3.279,
+      "step": 98000
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 4.473388195037842,
+      "learning_rate": 5.8432160804020096e-06,
+      "loss": 0.8428,
+      "step": 98100
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 2.7943496704101562,
+      "learning_rate": 5.541708542713567e-06,
+      "loss": 0.8519,
+      "step": 98200
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 2.476835012435913,
+      "learning_rate": 5.240201005025126e-06,
+      "loss": 0.8841,
+      "step": 98300
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 4.992676258087158,
+      "learning_rate": 4.938693467336683e-06,
+      "loss": 0.8409,
+      "step": 98400
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 2.4756906032562256,
+      "learning_rate": 4.637185929648241e-06,
+      "loss": 0.8527,
+      "step": 98500
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 2.157059669494629,
+      "learning_rate": 4.335678391959798e-06,
+      "loss": 0.8605,
+      "step": 98600
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 2.8840818405151367,
+      "learning_rate": 4.034170854271356e-06,
+      "loss": 0.87,
+      "step": 98700
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 4.124537944793701,
+      "learning_rate": 3.7326633165829143e-06,
+      "loss": 0.8318,
+      "step": 98800
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 4.684917449951172,
+      "learning_rate": 3.431155778894472e-06,
+      "loss": 0.8479,
+      "step": 98900
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 2.413602590560913,
+      "learning_rate": 3.12964824120603e-06,
+      "loss": 0.8452,
+      "step": 99000
+    },
+    {
+      "epoch": 4.95,
+      "eval_loss": 0.8957632780075073,
+      "eval_runtime": 38.1658,
+      "eval_samples_per_second": 26.201,
+      "eval_steps_per_second": 3.275,
+      "step": 99000
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 3.240213394165039,
+      "learning_rate": 2.828140703517588e-06,
+      "loss": 0.8303,
+      "step": 99100
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 4.0827555656433105,
+      "learning_rate": 2.5266331658291453e-06,
+      "loss": 0.8872,
+      "step": 99200
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 2.948489189147949,
+      "learning_rate": 2.2251256281407035e-06,
+      "loss": 0.8707,
+      "step": 99300
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 6.414693832397461,
+      "learning_rate": 1.9236180904522612e-06,
+      "loss": 0.837,
+      "step": 99400
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 5.013907432556152,
+      "learning_rate": 1.622110552763819e-06,
+      "loss": 0.8443,
+      "step": 99500
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 2.487205743789673,
+      "learning_rate": 1.3206030150753765e-06,
+      "loss": 0.8425,
+      "step": 99600
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 5.77063512802124,
+      "learning_rate": 1.0190954773869345e-06,
+      "loss": 0.8509,
+      "step": 99700
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 3.125368356704712,
+      "learning_rate": 7.175879396984924e-07,
+      "loss": 0.8874,
+      "step": 99800
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 8.932684898376465,
+      "learning_rate": 4.160804020100502e-07,
+      "loss": 0.858,
+      "step": 99900
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 5.0273756980896,
+      "learning_rate": 1.1457286432160803e-07,
+      "loss": 0.8394,
+      "step": 100000
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 0.9212185144424438,
+      "eval_runtime": 38.102,
+      "eval_samples_per_second": 26.245,
+      "eval_steps_per_second": 3.281,
+      "step": 100000
+    }
+  ],
+  "logging_steps": 100,
+  "max_steps": 100000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 1000,
+  "total_flos": 1.2076594495488e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}