diff --git "a/checkpoint-11082/trainer_state.json" "b/checkpoint-11082/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-11082/trainer_state.json"
@@ -0,0 +1,8669 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 100,
+  "global_step": 11082,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0027070925825663237,
+      "grad_norm": 6.21875,
+      "learning_rate": 5e-05,
+      "loss": 1.6293,
+      "step": 10
+    },
+    {
+      "epoch": 0.005414185165132647,
+      "grad_norm": 4.875,
+      "learning_rate": 0.0001,
+      "loss": 1.3134,
+      "step": 20
+    },
+    {
+      "epoch": 0.008121277747698972,
+      "grad_norm": 2.046875,
+      "learning_rate": 0.00015,
+      "loss": 1.0506,
+      "step": 30
+    },
+    {
+      "epoch": 0.010828370330265295,
+      "grad_norm": 3.875,
+      "learning_rate": 0.0002,
+      "loss": 0.8961,
+      "step": 40
+    },
+    {
+      "epoch": 0.01353546291283162,
+      "grad_norm": 6.1875,
+      "learning_rate": 0.00025,
+      "loss": 0.7928,
+      "step": 50
+    },
+    {
+      "epoch": 0.016242555495397944,
+      "grad_norm": 2.078125,
+      "learning_rate": 0.0003,
+      "loss": 0.7576,
+      "step": 60
+    },
+    {
+      "epoch": 0.018949648077964266,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.00035,
+      "loss": 0.7346,
+      "step": 70
+    },
+    {
+      "epoch": 0.02165674066053059,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0004,
+      "loss": 0.7174,
+      "step": 80
+    },
+    {
+      "epoch": 0.024363833243096916,
+      "grad_norm": 2.28125,
+      "learning_rate": 0.00045000000000000004,
+      "loss": 0.7025,
+      "step": 90
+    },
+    {
+      "epoch": 0.02707092582566324,
+      "grad_norm": 1.9609375,
+      "learning_rate": 0.0005,
+      "loss": 0.6959,
+      "step": 100
+    },
+    {
+      "epoch": 0.02707092582566324,
+      "eval_loss": 0.6594426035881042,
+      "eval_runtime": 42.3806,
+      "eval_samples_per_second": 11.798,
+      "eval_steps_per_second": 0.755,
+      "step": 100
+    },
+    {
+      "epoch": 0.02977801840822956,
+      "grad_norm": 1.6640625,
+      "learning_rate": 0.0004995447095246767,
+      "loss": 0.6783,
+      "step": 110
+    },
+    {
+      "epoch": 0.03248511099079589,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.0004990894190493535,
+      "loss": 0.628,
+      "step": 120
+    },
+    {
+      "epoch": 0.03519220357336221,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.0004986341285740302,
+      "loss": 0.5667,
+      "step": 130
+    },
+    {
+      "epoch": 0.03789929615592853,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.000498178838098707,
+      "loss": 0.4724,
+      "step": 140
+    },
+    {
+      "epoch": 0.040606388738494856,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0004977235476233838,
+      "loss": 0.3979,
+      "step": 150
+    },
+    {
+      "epoch": 0.04331348132106118,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.0004972682571480605,
+      "loss": 0.3501,
+      "step": 160
+    },
+    {
+      "epoch": 0.0460205739036275,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.0004968129666727372,
+      "loss": 0.2907,
+      "step": 170
+    },
+    {
+      "epoch": 0.04872766648619383,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.0004963576761974139,
+      "loss": 0.2785,
+      "step": 180
+    },
+    {
+      "epoch": 0.051434759068760154,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.0004959023857220907,
+      "loss": 0.2925,
+      "step": 190
+    },
+    {
+      "epoch": 0.05414185165132648,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.0004954470952467675,
+      "loss": 0.2558,
+      "step": 200
+    },
+    {
+      "epoch": 0.05414185165132648,
+      "eval_loss": 0.23236435651779175,
+      "eval_runtime": 37.6221,
+      "eval_samples_per_second": 13.29,
+      "eval_steps_per_second": 0.851,
+      "step": 200
+    },
+    {
+      "epoch": 0.0568489442338928,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0004949918047714442,
+      "loss": 0.3145,
+      "step": 210
+    },
+    {
+      "epoch": 0.05955603681645912,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0004945365142961209,
+      "loss": 0.2717,
+      "step": 220
+    },
+    {
+      "epoch": 0.062263129399025445,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0004940812238207977,
+      "loss": 0.2709,
+      "step": 230
+    },
+    {
+      "epoch": 0.06497022198159177,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0004936259333454744,
+      "loss": 0.2667,
+      "step": 240
+    },
+    {
+      "epoch": 0.0676773145641581,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0004931706428701511,
+      "loss": 0.2234,
+      "step": 250
+    },
+    {
+      "epoch": 0.07038440714672442,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.0004927153523948279,
+      "loss": 0.2305,
+      "step": 260
+    },
+    {
+      "epoch": 0.07309149972929074,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0004922600619195047,
+      "loss": 0.2321,
+      "step": 270
+    },
+    {
+      "epoch": 0.07579859231185707,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0004918047714441814,
+      "loss": 0.2085,
+      "step": 280
+    },
+    {
+      "epoch": 0.07850568489442339,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0004913494809688581,
+      "loss": 0.1982,
+      "step": 290
+    },
+    {
+      "epoch": 0.08121277747698971,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0004908941904935349,
+      "loss": 0.2223,
+      "step": 300
+    },
+    {
+      "epoch": 0.08121277747698971,
+      "eval_loss": 0.18759433925151825,
+      "eval_runtime": 37.8667,
+      "eval_samples_per_second": 13.204,
+      "eval_steps_per_second": 0.845,
+      "step": 300
+    },
+    {
+      "epoch": 0.08391987005955603,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0004904389000182116,
+      "loss": 0.2137,
+      "step": 310
+    },
+    {
+      "epoch": 0.08662696264212236,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0004899836095428884,
+      "loss": 0.217,
+      "step": 320
+    },
+    {
+      "epoch": 0.08933405522468868,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.0004895283190675652,
+      "loss": 0.2492,
+      "step": 330
+    },
+    {
+      "epoch": 0.092041147807255,
+      "grad_norm": 0.375,
+      "learning_rate": 0.0004890730285922419,
+      "loss": 0.2386,
+      "step": 340
+    },
+    {
+      "epoch": 0.09474824038982133,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0004886177381169186,
+      "loss": 0.2434,
+      "step": 350
+    },
+    {
+      "epoch": 0.09745533297238766,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00048816244764159533,
+      "loss": 0.238,
+      "step": 360
+    },
+    {
+      "epoch": 0.10016242555495398,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00048770715716627206,
+      "loss": 0.2441,
+      "step": 370
+    },
+    {
+      "epoch": 0.10286951813752031,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.00048725186669094884,
+      "loss": 0.2009,
+      "step": 380
+    },
+    {
+      "epoch": 0.10557661072008663,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.00048679657621562557,
+      "loss": 0.1767,
+      "step": 390
+    },
+    {
+      "epoch": 0.10828370330265295,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.00048634128574030235,
+      "loss": 0.2211,
+      "step": 400
+    },
+    {
+      "epoch": 0.10828370330265295,
+      "eval_loss": 0.17902880907058716,
+      "eval_runtime": 37.9544,
+      "eval_samples_per_second": 13.174,
+      "eval_steps_per_second": 0.843,
+      "step": 400
+    },
+    {
+      "epoch": 0.11099079588521928,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0004858859952649791,
+      "loss": 0.1893,
+      "step": 410
+    },
+    {
+      "epoch": 0.1136978884677856,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.0004854307047896558,
+      "loss": 0.1928,
+      "step": 420
+    },
+    {
+      "epoch": 0.11640498105035192,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.0004849754143143326,
+      "loss": 0.1993,
+      "step": 430
+    },
+    {
+      "epoch": 0.11911207363291824,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0004845201238390093,
+      "loss": 0.1894,
+      "step": 440
+    },
+    {
+      "epoch": 0.12181916621548457,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0004840648333636861,
+      "loss": 0.2047,
+      "step": 450
+    },
+    {
+      "epoch": 0.12452625879805089,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00048360954288836276,
+      "loss": 0.184,
+      "step": 460
+    },
+    {
+      "epoch": 0.12723335138061723,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0004831542524130395,
+      "loss": 0.1882,
+      "step": 470
+    },
+    {
+      "epoch": 0.12994044396318355,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.00048269896193771627,
+      "loss": 0.2025,
+      "step": 480
+    },
+    {
+      "epoch": 0.13264753654574987,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.000482243671462393,
+      "loss": 0.181,
+      "step": 490
+    },
+    {
+      "epoch": 0.1353546291283162,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0004817883809870698,
+      "loss": 0.2364,
+      "step": 500
+    },
+    {
+      "epoch": 0.1353546291283162,
+      "eval_loss": 0.17341379821300507,
+      "eval_runtime": 39.1063,
+      "eval_samples_per_second": 12.786,
+      "eval_steps_per_second": 0.818,
+      "step": 500
+    },
+    {
+      "epoch": 0.13806172171088252,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0004813330905117465,
+      "loss": 0.1816,
+      "step": 510
+    },
+    {
+      "epoch": 0.14076881429344884,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00048087780003642323,
+      "loss": 0.2083,
+      "step": 520
+    },
+    {
+      "epoch": 0.14347590687601516,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0004804225095611,
+      "loss": 0.1921,
+      "step": 530
+    },
+    {
+      "epoch": 0.1461829994585815,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00047996721908577674,
+      "loss": 0.1706,
+      "step": 540
+    },
+    {
+      "epoch": 0.1488900920411478,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00047951192861045346,
+      "loss": 0.1907,
+      "step": 550
+    },
+    {
+      "epoch": 0.15159718462371413,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00047905663813513024,
+      "loss": 0.1994,
+      "step": 560
+    },
+    {
+      "epoch": 0.15430427720628045,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00047860134765980697,
+      "loss": 0.1865,
+      "step": 570
+    },
+    {
+      "epoch": 0.15701136978884678,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00047814605718448375,
+      "loss": 0.2122,
+      "step": 580
+    },
+    {
+      "epoch": 0.1597184623714131,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0004776907667091605,
+      "loss": 0.2284,
+      "step": 590
+    },
+    {
+      "epoch": 0.16242555495397942,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00047723547623383715,
+      "loss": 0.1636,
+      "step": 600
+    },
+    {
+      "epoch": 0.16242555495397942,
+      "eval_loss": 0.17351534962654114,
+      "eval_runtime": 39.4102,
+      "eval_samples_per_second": 12.687,
+      "eval_steps_per_second": 0.812,
+      "step": 600
+    },
+    {
+      "epoch": 0.16513264753654575,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00047678018575851393,
+      "loss": 0.1869,
+      "step": 610
+    },
+    {
+      "epoch": 0.16783974011911207,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.00047632489528319066,
+      "loss": 0.2092,
+      "step": 620
+    },
+    {
+      "epoch": 0.1705468327016784,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.00047586960480786744,
+      "loss": 0.1843,
+      "step": 630
+    },
+    {
+      "epoch": 0.17325392528424471,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.00047541431433254416,
+      "loss": 0.1945,
+      "step": 640
+    },
+    {
+      "epoch": 0.17596101786681104,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.0004749590238572209,
+      "loss": 0.1903,
+      "step": 650
+    },
+    {
+      "epoch": 0.17866811044937736,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00047450373338189767,
+      "loss": 0.1847,
+      "step": 660
+    },
+    {
+      "epoch": 0.18137520303194368,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0004740484429065744,
+      "loss": 0.1642,
+      "step": 670
+    },
+    {
+      "epoch": 0.18408229561451,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.0004735931524312512,
+      "loss": 0.2171,
+      "step": 680
+    },
+    {
+      "epoch": 0.18678938819707633,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.0004731378619559279,
+      "loss": 0.1856,
+      "step": 690
+    },
+    {
+      "epoch": 0.18949648077964265,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.00047268257148060463,
+      "loss": 0.1867,
+      "step": 700
+    },
+    {
+      "epoch": 0.18949648077964265,
+      "eval_loss": 0.1605198085308075,
+      "eval_runtime": 41.5486,
+      "eval_samples_per_second": 12.034,
+      "eval_steps_per_second": 0.77,
+      "step": 700
+    },
+    {
+      "epoch": 0.19220357336220897,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.0004722272810052814,
+      "loss": 0.2102,
+      "step": 710
+    },
+    {
+      "epoch": 0.19491066594477532,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00047177199052995814,
+      "loss": 0.1767,
+      "step": 720
+    },
+    {
+      "epoch": 0.19761775852734165,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00047131670005463487,
+      "loss": 0.1774,
+      "step": 730
+    },
+    {
+      "epoch": 0.20032485110990797,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0004708614095793116,
+      "loss": 0.139,
+      "step": 740
+    },
+    {
+      "epoch": 0.2030319436924743,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.0004704061191039883,
+      "loss": 0.1481,
+      "step": 750
+    },
+    {
+      "epoch": 0.20573903627504062,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.0004699508286286651,
+      "loss": 0.1865,
+      "step": 760
+    },
+    {
+      "epoch": 0.20844612885760694,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.0004694955381533418,
+      "loss": 0.144,
+      "step": 770
+    },
+    {
+      "epoch": 0.21115322144017326,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00046904024767801855,
+      "loss": 0.1715,
+      "step": 780
+    },
+    {
+      "epoch": 0.21386031402273958,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.00046858495720269533,
+      "loss": 0.2009,
+      "step": 790
+    },
+    {
+      "epoch": 0.2165674066053059,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00046812966672737206,
+      "loss": 0.1736,
+      "step": 800
+    },
+    {
+      "epoch": 0.2165674066053059,
+      "eval_loss": 0.15155640244483948,
+      "eval_runtime": 37.7255,
+      "eval_samples_per_second": 13.254,
+      "eval_steps_per_second": 0.848,
+      "step": 800
+    },
+    {
+      "epoch": 0.21927449918787223,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00046767437625204884,
+      "loss": 0.1267,
+      "step": 810
+    },
+    {
+      "epoch": 0.22198159177043855,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00046721908577672557,
+      "loss": 0.1506,
+      "step": 820
+    },
+    {
+      "epoch": 0.22468868435300487,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.0004667637953014023,
+      "loss": 0.1306,
+      "step": 830
+    },
+    {
+      "epoch": 0.2273957769355712,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.0004663085048260791,
+      "loss": 0.1666,
+      "step": 840
+    },
+    {
+      "epoch": 0.23010286951813752,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0004658532143507558,
+      "loss": 0.2141,
+      "step": 850
+    },
+    {
+      "epoch": 0.23280996210070384,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.0004653979238754326,
+      "loss": 0.1866,
+      "step": 860
+    },
+    {
+      "epoch": 0.23551705468327017,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.0004649426334001093,
+      "loss": 0.1656,
+      "step": 870
+    },
+    {
+      "epoch": 0.2382241472658365,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.000464487342924786,
+      "loss": 0.121,
+      "step": 880
+    },
+    {
+      "epoch": 0.2409312398484028,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00046403205244946276,
+      "loss": 0.1475,
+      "step": 890
+    },
+    {
+      "epoch": 0.24363833243096913,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0004635767619741395,
+      "loss": 0.1437,
+      "step": 900
+    },
+    {
+      "epoch": 0.24363833243096913,
+      "eval_loss": 0.1451132893562317,
+      "eval_runtime": 39.5516,
+      "eval_samples_per_second": 12.642,
+      "eval_steps_per_second": 0.809,
+      "step": 900
+    },
+    {
+      "epoch": 0.24634542501353546,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.0004631214714988162,
+      "loss": 0.1376,
+      "step": 910
+    },
+    {
+      "epoch": 0.24905251759610178,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.000462666181023493,
+      "loss": 0.1732,
+      "step": 920
+    },
+    {
+      "epoch": 0.2517596101786681,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0004622108905481697,
+      "loss": 0.1889,
+      "step": 930
+    },
+    {
+      "epoch": 0.25446670276123445,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0004617556000728465,
+      "loss": 0.1697,
+      "step": 940
+    },
+    {
+      "epoch": 0.25717379534380075,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.00046130030959752323,
+      "loss": 0.13,
+      "step": 950
+    },
+    {
+      "epoch": 0.2598808879263671,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00046084501912219995,
+      "loss": 0.1368,
+      "step": 960
+    },
+    {
+      "epoch": 0.2625879805089334,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00046038972864687674,
+      "loss": 0.1706,
+      "step": 970
+    },
+    {
+      "epoch": 0.26529507309149974,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.00045993443817155346,
+      "loss": 0.1569,
+      "step": 980
+    },
+    {
+      "epoch": 0.26800216567406604,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.00045947914769623024,
+      "loss": 0.1439,
+      "step": 990
+    },
+    {
+      "epoch": 0.2707092582566324,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00045902385722090697,
+      "loss": 0.1286,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2707092582566324,
+      "eval_loss": 0.15346874296665192,
+      "eval_runtime": 39.3332,
+      "eval_samples_per_second": 12.712,
+      "eval_steps_per_second": 0.814,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2734163508391987,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.0004585685667455837,
+      "loss": 0.155,
+      "step": 1010
+    },
+    {
+      "epoch": 0.27612344342176504,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0004581132762702604,
+      "loss": 0.1152,
+      "step": 1020
+    },
+    {
+      "epoch": 0.27883053600433133,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00045765798579493715,
+      "loss": 0.1296,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2815376285868977,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00045720269531961393,
+      "loss": 0.1436,
+      "step": 1040
+    },
+    {
+      "epoch": 0.284244721169464,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00045674740484429066,
+      "loss": 0.1526,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2869518137520303,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0004562921143689674,
+      "loss": 0.1374,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2896589063345966,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.00045583682389364416,
+      "loss": 0.1408,
+      "step": 1070
+    },
+    {
+      "epoch": 0.292365998917163,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.0004553815334183209,
+      "loss": 0.1447,
+      "step": 1080
+    },
+    {
+      "epoch": 0.29507309149972927,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0004549262429429976,
+      "loss": 0.1485,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2977801840822956,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0004544709524676744,
+      "loss": 0.1571,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2977801840822956,
+      "eval_loss": 0.15628516674041748,
+      "eval_runtime": 37.9759,
+      "eval_samples_per_second": 13.166,
+      "eval_steps_per_second": 0.843,
+      "step": 1100
+    },
+    {
+      "epoch": 0.3004872766648619,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.0004540156619923511,
+      "loss": 0.18,
+      "step": 1110
+    },
+    {
+      "epoch": 0.30319436924742826,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.0004535603715170279,
+      "loss": 0.1677,
+      "step": 1120
+    },
+    {
+      "epoch": 0.30590146182999456,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00045310508104170463,
+      "loss": 0.1368,
+      "step": 1130
+    },
+    {
+      "epoch": 0.3086085544125609,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00045264979056638136,
+      "loss": 0.1485,
+      "step": 1140
+    },
+    {
+      "epoch": 0.31131564699512726,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00045219450009105814,
+      "loss": 0.1261,
+      "step": 1150
+    },
+    {
+      "epoch": 0.31402273957769355,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.0004517392096157348,
+      "loss": 0.1353,
+      "step": 1160
+    },
+    {
+      "epoch": 0.3167298321602599,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.0004512839191404116,
+      "loss": 0.1721,
+      "step": 1170
+    },
+    {
+      "epoch": 0.3194369247428262,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.0004508286286650883,
+      "loss": 0.1788,
+      "step": 1180
+    },
+    {
+      "epoch": 0.32214401732539255,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00045037333818976504,
+      "loss": 0.1373,
+      "step": 1190
+    },
+    {
+      "epoch": 0.32485110990795885,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.0004499180477144418,
+      "loss": 0.1345,
+      "step": 1200
+    },
+    {
+      "epoch": 0.32485110990795885,
+      "eval_loss": 0.14543119072914124,
+      "eval_runtime": 38.5424,
+      "eval_samples_per_second": 12.973,
+      "eval_steps_per_second": 0.83,
+      "step": 1200
+    },
+    {
+      "epoch": 0.3275582024905252,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.00044946275723911855,
+      "loss": 0.1346,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3302652950730915,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00044900746676379533,
+      "loss": 0.1426,
+      "step": 1220
+    },
+    {
+      "epoch": 0.33297238765565784,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00044855217628847206,
+      "loss": 0.179,
+      "step": 1230
+    },
+    {
+      "epoch": 0.33567948023822414,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0004480968858131488,
+      "loss": 0.1287,
+      "step": 1240
+    },
+    {
+      "epoch": 0.3383865728207905,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00044764159533782557,
+      "loss": 0.1443,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3410936654033568,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0004471863048625023,
+      "loss": 0.1381,
+      "step": 1260
+    },
+    {
+      "epoch": 0.34380075798592313,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.000446731014387179,
+      "loss": 0.1481,
+      "step": 1270
+    },
+    {
+      "epoch": 0.34650785056848943,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0004462757239118558,
+      "loss": 0.1793,
+      "step": 1280
+    },
+    {
+      "epoch": 0.3492149431510558,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.0004458204334365325,
+      "loss": 0.1527,
+      "step": 1290
+    },
+    {
+      "epoch": 0.3519220357336221,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0004453651429612093,
+      "loss": 0.1738,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3519220357336221,
+      "eval_loss": 0.14374633133411407,
+      "eval_runtime": 41.3232,
+      "eval_samples_per_second": 12.1,
+      "eval_steps_per_second": 0.774,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3546291283161884,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.000444909852485886,
+      "loss": 0.1587,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3573362208987547,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.0004444545620105627,
+      "loss": 0.1587,
+      "step": 1320
+    },
+    {
+      "epoch": 0.36004331348132107,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.0004439992715352395,
+      "loss": 0.1905,
+      "step": 1330
+    },
+    {
+      "epoch": 0.36275040606388737,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0004435439810599162,
+      "loss": 0.1775,
+      "step": 1340
+    },
+    {
+      "epoch": 0.3654574986464537,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.000443088690584593,
+      "loss": 0.1517,
+      "step": 1350
+    },
+    {
+      "epoch": 0.36816459122902,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0004426334001092697,
+      "loss": 0.113,
+      "step": 1360
+    },
+    {
+      "epoch": 0.37087168381158636,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.00044217810963394645,
+      "loss": 0.1438,
+      "step": 1370
+    },
+    {
+      "epoch": 0.37357877639415266,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0004417228191586232,
+      "loss": 0.1407,
+      "step": 1380
+    },
+    {
+      "epoch": 0.376285868976719,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00044126752868329995,
+      "loss": 0.15,
+      "step": 1390
+    },
+    {
+      "epoch": 0.3789929615592853,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00044081223820797673,
+      "loss": 0.1412,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3789929615592853,
+      "eval_loss": 0.1404724419116974,
+      "eval_runtime": 38.9598,
+      "eval_samples_per_second": 12.834,
+      "eval_steps_per_second": 0.821,
+      "step": 1400
+    },
+    {
+      "epoch": 0.38170005414185165,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00044035694773265346,
+      "loss": 0.1291,
+      "step": 1410
+    },
+    {
+      "epoch": 0.38440714672441795,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0004399016572573302,
+      "loss": 0.1533,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3871142393069843,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00043944636678200697,
+      "loss": 0.1526,
+      "step": 1430
+    },
+    {
+      "epoch": 0.38982133188955065,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0004389910763066837,
+      "loss": 0.1439,
+      "step": 1440
+    },
+    {
+      "epoch": 0.39252842447211694,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00043853578583136037,
+      "loss": 0.1741,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3952355170546833,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00043808049535603715,
+      "loss": 0.15,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3979426096372496,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0004376252048807139,
+      "loss": 0.1329,
+      "step": 1470
+    },
+    {
+      "epoch": 0.40064970221981594,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.00043716991440539065,
+      "loss": 0.1707,
+      "step": 1480
+    },
+    {
+      "epoch": 0.40335679480238223,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.0004367146239300674,
+      "loss": 0.1453,
+      "step": 1490
+    },
+    {
+      "epoch": 0.4060638873849486,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0004362593334547441,
+      "loss": 0.1476,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4060638873849486,
+      "eval_loss": 0.13253003358840942,
+      "eval_runtime": 37.5351,
+      "eval_samples_per_second": 13.321,
+      "eval_steps_per_second": 0.853,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4087709799675149,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.0004358040429794209,
+      "loss": 0.1376,
+      "step": 1510
+    },
+    {
+      "epoch": 0.41147807255008123,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0004353487525040976,
+      "loss": 0.1508,
+      "step": 1520
+    },
+    {
+      "epoch": 0.4141851651326475,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.0004348934620287744,
+      "loss": 0.1933,
+      "step": 1530
+    },
+    {
+      "epoch": 0.4168922577152139,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0004344381715534511,
+      "loss": 0.1309,
+      "step": 1540
+    },
+    {
+      "epoch": 0.41959935029778017,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00043398288107812785,
+      "loss": 0.1369,
+      "step": 1550
+    },
+    {
+      "epoch": 0.4223064428803465,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00043352759060280463,
+      "loss": 0.1147,
+      "step": 1560
+    },
+    {
+      "epoch": 0.4250135354629128,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00043307230012748136,
+      "loss": 0.1351,
+      "step": 1570
+    },
+    {
+      "epoch": 0.42772062804547917,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00043261700965215814,
+      "loss": 0.1702,
+      "step": 1580
+    },
+    {
+      "epoch": 0.43042772062804546,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.0004321617191768348,
+      "loss": 0.1418,
+      "step": 1590
+    },
+    {
+      "epoch": 0.4331348132106118,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.00043170642870151154,
+      "loss": 0.156,
+      "step": 1600
+    },
+    {
+      "epoch": 0.4331348132106118,
+      "eval_loss": 0.13202697038650513,
+      "eval_runtime": 38.05,
+      "eval_samples_per_second": 13.141,
+      "eval_steps_per_second": 0.841,
+      "step": 1600
+    },
+    {
+      "epoch": 0.4358419057931781,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0004312511382261883,
+      "loss": 0.1278,
+      "step": 1610
+    },
+    {
+      "epoch": 0.43854899837574446,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.00043079584775086504,
+      "loss": 0.1584,
+      "step": 1620
+    },
+    {
+      "epoch": 0.44125609095831075,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.00043034055727554177,
+      "loss": 0.1308,
+      "step": 1630
+    },
+    {
+      "epoch": 0.4439631835408771,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00042988526680021855,
+      "loss": 0.124,
+      "step": 1640
+    },
+    {
+      "epoch": 0.4466702761234434,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0004294299763248953,
+      "loss": 0.162,
+      "step": 1650
+    },
+    {
+      "epoch": 0.44937736870600975,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00042897468584957206,
+      "loss": 0.1747,
+      "step": 1660
+    },
+    {
+      "epoch": 0.45208446128857604,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0004285193953742488,
+      "loss": 0.1515,
+      "step": 1670
+    },
+    {
+      "epoch": 0.4547915538711424,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0004280641048989255,
+      "loss": 0.1623,
+      "step": 1680
+    },
+    {
+      "epoch": 0.4574986464537087,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0004276088144236023,
+      "loss": 0.1262,
+      "step": 1690
+    },
+    {
+      "epoch": 0.46020573903627504,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.000427153523948279,
+      "loss": 0.1287,
+      "step": 1700
+    },
+    {
+      "epoch": 0.46020573903627504,
+      "eval_loss": 0.14365537464618683,
+      "eval_runtime": 37.7236,
+      "eval_samples_per_second": 13.254,
+      "eval_steps_per_second": 0.848,
+      "step": 1700
+    },
+    {
+      "epoch": 0.4629128316188414,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0004266982334729558,
+      "loss": 0.1561,
+      "step": 1710
+    },
+    {
+      "epoch": 0.4656199242014077,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0004262429429976325,
+      "loss": 0.1426,
+      "step": 1720
+    },
+    {
+      "epoch": 0.46832701678397404,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0004257876525223092,
+      "loss": 0.1772,
+      "step": 1730
+    },
+    {
+      "epoch": 0.47103410936654033,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.000425332362046986,
+      "loss": 0.1398,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4737412019491067,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.0004248770715716627,
+      "loss": 0.1499,
+      "step": 1750
+    },
+    {
+      "epoch": 0.476448294531673,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0004244217810963395,
+      "loss": 0.14,
+      "step": 1760
+    },
+    {
+      "epoch": 0.47915538711423933,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0004239664906210162,
+      "loss": 0.1685,
+      "step": 1770
+    },
+    {
+      "epoch": 0.4818624796968056,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00042351120014569294,
+      "loss": 0.1627,
+      "step": 1780
+    },
+    {
+      "epoch": 0.484569572279372,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0004230559096703697,
+      "loss": 0.1185,
+      "step": 1790
+    },
+    {
+      "epoch": 0.48727666486193827,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00042260061919504645,
+      "loss": 0.1505,
+      "step": 1800
+    },
+    {
+      "epoch": 0.48727666486193827,
+      "eval_loss": 0.13145561516284943,
+      "eval_runtime": 36.3693,
+      "eval_samples_per_second": 13.748,
+      "eval_steps_per_second": 0.88,
+      "step": 1800
+    },
+    {
+      "epoch": 0.4899837574445046,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.00042214532871972317,
+      "loss": 0.1285,
+      "step": 1810
+    },
+    {
+      "epoch": 0.4926908500270709,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.00042169003824439995,
+      "loss": 0.1554,
+      "step": 1820
+    },
+    {
+      "epoch": 0.49539794260963727,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0004212347477690767,
+      "loss": 0.1204,
+      "step": 1830
+    },
+    {
+      "epoch": 0.49810503519220356,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.00042077945729375346,
+      "loss": 0.134,
+      "step": 1840
+    },
+    {
+      "epoch": 0.5008121277747699,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0004203241668184302,
+      "loss": 0.1609,
+      "step": 1850
+    },
+    {
+      "epoch": 0.5035192203573362,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0004198688763431069,
+      "loss": 0.1293,
+      "step": 1860
+    },
+    {
+      "epoch": 0.5062263129399025,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00041941358586778364,
+      "loss": 0.1088,
+      "step": 1870
+    },
+    {
+      "epoch": 0.5089334055224689,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00041895829539246037,
+      "loss": 0.1177,
+      "step": 1880
+    },
+    {
+      "epoch": 0.5116404981050352,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00041850300491713715,
+      "loss": 0.1231,
+      "step": 1890
+    },
+    {
+      "epoch": 0.5143475906876015,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0004180477144418139,
+      "loss": 0.1359,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5143475906876015,
+      "eval_loss": 0.13306322693824768,
+      "eval_runtime": 38.9149,
+      "eval_samples_per_second": 12.849,
+      "eval_steps_per_second": 0.822,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5170546832701678,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0004175924239664906,
+      "loss": 0.1426,
+      "step": 1910
+    },
+    {
+      "epoch": 0.5197617758527342,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0004171371334911674,
+      "loss": 0.1418,
+      "step": 1920
+    },
+    {
+      "epoch": 0.5224688684353005,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0004166818430158441,
+      "loss": 0.1343,
+      "step": 1930
+    },
+    {
+      "epoch": 0.5251759610178668,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0004162265525405209,
+      "loss": 0.1505,
+      "step": 1940
+    },
+    {
+      "epoch": 0.5278830536004331,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0004157712620651976,
+      "loss": 0.1764,
+      "step": 1950
+    },
+    {
+      "epoch": 0.5305901461829995,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00041531597158987434,
+      "loss": 0.1161,
+      "step": 1960
+    },
+    {
+      "epoch": 0.5332972387655658,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.0004148606811145511,
+      "loss": 0.1666,
+      "step": 1970
+    },
+    {
+      "epoch": 0.5360043313481321,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.00041440539063922785,
+      "loss": 0.1415,
+      "step": 1980
+    },
+    {
+      "epoch": 0.5387114239306985,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0004139501001639046,
+      "loss": 0.1414,
+      "step": 1990
+    },
+    {
+      "epoch": 0.5414185165132648,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.00041349480968858136,
+      "loss": 0.1433,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5414185165132648,
+      "eval_loss": 0.13291554152965546,
+      "eval_runtime": 37.9452,
+      "eval_samples_per_second": 13.177,
+      "eval_steps_per_second": 0.843,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5441256090958311,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.00041303951921325803,
+      "loss": 0.1205,
+      "step": 2010
+    },
+    {
+      "epoch": 0.5468327016783974,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0004125842287379348,
+      "loss": 0.1108,
+      "step": 2020
+    },
+    {
+      "epoch": 0.5495397942609638,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.00041212893826261153,
+      "loss": 0.1539,
+      "step": 2030
+    },
+    {
+      "epoch": 0.5522468868435301,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00041167364778728826,
+      "loss": 0.1161,
+      "step": 2040
+    },
+    {
+      "epoch": 0.5549539794260964,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.00041121835731196504,
+      "loss": 0.1131,
+      "step": 2050
+    },
+    {
+      "epoch": 0.5576610720086627,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00041076306683664177,
+      "loss": 0.1205,
+      "step": 2060
+    },
+    {
+      "epoch": 0.5603681645912291,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00041030777636131855,
+      "loss": 0.142,
+      "step": 2070
+    },
+    {
+      "epoch": 0.5630752571737954,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0004098524858859953,
+      "loss": 0.144,
+      "step": 2080
+    },
+    {
+      "epoch": 0.5657823497563617,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.000409397195410672,
+      "loss": 0.1397,
+      "step": 2090
+    },
+    {
+      "epoch": 0.568489442338928,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.0004089419049353488,
+      "loss": 0.1394,
+      "step": 2100
+    },
+    {
+      "epoch": 0.568489442338928,
+      "eval_loss": 0.1250193864107132,
+      "eval_runtime": 38.6869,
+      "eval_samples_per_second": 12.924,
+      "eval_steps_per_second": 0.827,
+      "step": 2100
+    },
+    {
+      "epoch": 0.5711965349214944,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.0004084866144600255,
+      "loss": 0.1083,
+      "step": 2110
+    },
+    {
+      "epoch": 0.5739036275040607,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.0004080313239847023,
+      "loss": 0.1016,
+      "step": 2120
+    },
+    {
+      "epoch": 0.576610720086627,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.000407576033509379,
+      "loss": 0.1371,
+      "step": 2130
+    },
+    {
+      "epoch": 0.5793178126691932,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00040712074303405574,
+      "loss": 0.1305,
+      "step": 2140
+    },
+    {
+      "epoch": 0.5820249052517596,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00040666545255873247,
+      "loss": 0.1282,
+      "step": 2150
+    },
+    {
+      "epoch": 0.584731997834326,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0004062101620834092,
+      "loss": 0.1282,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5874390904168922,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.0004057548716080859,
+      "loss": 0.1408,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5901461829994585,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.0004052995811327627,
+      "loss": 0.1233,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5928532755820249,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00040484429065743943,
+      "loss": 0.1248,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5955603681645912,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0004043890001821162,
+      "loss": 0.1307,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5955603681645912,
+      "eval_loss": 0.12977245450019836,
+      "eval_runtime": 37.863,
+      "eval_samples_per_second": 13.206,
+      "eval_steps_per_second": 0.845,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5982674607471575,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.00040393370970679294,
+      "loss": 0.1684,
+      "step": 2210
+    },
+    {
+      "epoch": 0.6009745533297238,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.00040347841923146966,
+      "loss": 0.0936,
+      "step": 2220
+    },
+    {
+      "epoch": 0.6036816459122902,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.00040302312875614644,
+      "loss": 0.17,
+      "step": 2230
+    },
+    {
+      "epoch": 0.6063887384948565,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00040256783828082317,
+      "loss": 0.1312,
+      "step": 2240
+    },
+    {
+      "epoch": 0.6090958310774228,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.00040211254780549995,
+      "loss": 0.1081,
+      "step": 2250
+    },
+    {
+      "epoch": 0.6118029236599891,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.0004016572573301767,
+      "loss": 0.1195,
+      "step": 2260
+    },
+    {
+      "epoch": 0.6145100162425555,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0004012019668548534,
+      "loss": 0.1451,
+      "step": 2270
+    },
+    {
+      "epoch": 0.6172171088251218,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0004007466763795302,
+      "loss": 0.1348,
+      "step": 2280
+    },
+    {
+      "epoch": 0.6199242014076881,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00040029138590420686,
+      "loss": 0.1238,
+      "step": 2290
+    },
+    {
+      "epoch": 0.6226312939902545,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00039983609542888364,
+      "loss": 0.1424,
+      "step": 2300
+    },
+    {
+      "epoch": 0.6226312939902545,
+      "eval_loss": 0.13612085580825806,
+      "eval_runtime": 39.3274,
+      "eval_samples_per_second": 12.714,
+      "eval_steps_per_second": 0.814,
+      "step": 2300
+    },
+    {
+      "epoch": 0.6253383865728208,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00039938080495356037,
+      "loss": 0.1156,
+      "step": 2310
+    },
+    {
+      "epoch": 0.6280454791553871,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.0003989255144782371,
+      "loss": 0.1526,
+      "step": 2320
+    },
+    {
+      "epoch": 0.6307525717379534,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00039847022400291387,
+      "loss": 0.114,
+      "step": 2330
+    },
+    {
+      "epoch": 0.6334596643205198,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0003980149335275906,
+      "loss": 0.1156,
+      "step": 2340
+    },
+    {
+      "epoch": 0.6361667569030861,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0003975596430522673,
+      "loss": 0.1046,
+      "step": 2350
+    },
+    {
+      "epoch": 0.6388738494856524,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.0003971043525769441,
+      "loss": 0.1142,
+      "step": 2360
+    },
+    {
+      "epoch": 0.6415809420682187,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00039664906210162083,
+      "loss": 0.1438,
+      "step": 2370
+    },
+    {
+      "epoch": 0.6442880346507851,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0003961937716262976,
+      "loss": 0.1531,
+      "step": 2380
+    },
+    {
+      "epoch": 0.6469951272333514,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.00039573848115097434,
+      "loss": 0.1419,
+      "step": 2390
+    },
+    {
+      "epoch": 0.6497022198159177,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00039528319067565107,
+      "loss": 0.1308,
+      "step": 2400
+    },
+    {
+      "epoch": 0.6497022198159177,
+      "eval_loss": 0.13486501574516296,
+      "eval_runtime": 39.8762,
+      "eval_samples_per_second": 12.539,
+      "eval_steps_per_second": 0.802,
+      "step": 2400
+    },
+    {
+      "epoch": 0.652409312398484,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00039482790020032785,
+      "loss": 0.1405,
+      "step": 2410
+    },
+    {
+      "epoch": 0.6551164049810504,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.0003943726097250046,
+      "loss": 0.1523,
+      "step": 2420
+    },
+    {
+      "epoch": 0.6578234975636167,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.00039391731924968135,
+      "loss": 0.1504,
+      "step": 2430
+    },
+    {
+      "epoch": 0.660530590146183,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.000393462028774358,
+      "loss": 0.1139,
+      "step": 2440
+    },
+    {
+      "epoch": 0.6632376827287493,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.00039300673829903475,
+      "loss": 0.1321,
+      "step": 2450
+    },
+    {
+      "epoch": 0.6659447753113157,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00039255144782371153,
+      "loss": 0.1049,
+      "step": 2460
+    },
+    {
+      "epoch": 0.668651867893882,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.00039209615734838826,
+      "loss": 0.1105,
+      "step": 2470
+    },
+    {
+      "epoch": 0.6713589604764483,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.00039164086687306504,
+      "loss": 0.1293,
+      "step": 2480
+    },
+    {
+      "epoch": 0.6740660530590146,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00039118557639774177,
+      "loss": 0.1171,
+      "step": 2490
+    },
+    {
+      "epoch": 0.676773145641581,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.0003907302859224185,
+      "loss": 0.1185,
+      "step": 2500
+    },
+    {
+      "epoch": 0.676773145641581,
+      "eval_loss": 0.13772137463092804,
+      "eval_runtime": 37.959,
+      "eval_samples_per_second": 13.172,
+      "eval_steps_per_second": 0.843,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6794802382241473,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0003902749954470953,
+      "loss": 0.1517,
+      "step": 2510
+    },
+    {
+      "epoch": 0.6821873308067136,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.000389819704971772,
+      "loss": 0.126,
+      "step": 2520
+    },
+    {
+      "epoch": 0.6848944233892799,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00038936441449644873,
+      "loss": 0.1425,
+      "step": 2530
+    },
+    {
+      "epoch": 0.6876015159718463,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0003889091240211255,
+      "loss": 0.0997,
+      "step": 2540
+    },
+    {
+      "epoch": 0.6903086085544126,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00038845383354580224,
+      "loss": 0.1443,
+      "step": 2550
+    },
+    {
+      "epoch": 0.6930157011369789,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.000387998543070479,
+      "loss": 0.1133,
+      "step": 2560
+    },
+    {
+      "epoch": 0.6957227937195453,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.00038754325259515574,
+      "loss": 0.1083,
+      "step": 2570
+    },
+    {
+      "epoch": 0.6984298863021116,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.0003870879621198324,
+      "loss": 0.1003,
+      "step": 2580
+    },
+    {
+      "epoch": 0.7011369788846779,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0003866326716445092,
+      "loss": 0.125,
+      "step": 2590
+    },
+    {
+      "epoch": 0.7038440714672441,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0003861773811691859,
+      "loss": 0.1104,
+      "step": 2600
+    },
+    {
+      "epoch": 0.7038440714672441,
+      "eval_loss": 0.1279294639825821,
+      "eval_runtime": 38.5949,
+      "eval_samples_per_second": 12.955,
+      "eval_steps_per_second": 0.829,
+      "step": 2600
+    },
+    {
+      "epoch": 0.7065511640498106,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0003857220906938627,
+      "loss": 0.1436,
+      "step": 2610
+    },
+    {
+      "epoch": 0.7092582566323768,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00038526680021853943,
+      "loss": 0.1387,
+      "step": 2620
+    },
+    {
+      "epoch": 0.7119653492149431,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00038481150974321616,
+      "loss": 0.183,
+      "step": 2630
+    },
+    {
+      "epoch": 0.7146724417975094,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.00038435621926789294,
+      "loss": 0.118,
+      "step": 2640
+    },
+    {
+      "epoch": 0.7173795343800758,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00038390092879256966,
+      "loss": 0.1324,
+      "step": 2650
+    },
+    {
+      "epoch": 0.7200866269626421,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.00038344563831724644,
+      "loss": 0.1202,
+      "step": 2660
+    },
+    {
+      "epoch": 0.7227937195452084,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00038299034784192317,
+      "loss": 0.1387,
+      "step": 2670
+    },
+    {
+      "epoch": 0.7255008121277747,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0003825350573665999,
+      "loss": 0.1481,
+      "step": 2680
+    },
+    {
+      "epoch": 0.7282079047103411,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0003820797668912767,
+      "loss": 0.1421,
+      "step": 2690
+    },
+    {
+      "epoch": 0.7309149972929074,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0003816244764159534,
+      "loss": 0.0967,
+      "step": 2700
+    },
+    {
+      "epoch": 0.7309149972929074,
+      "eval_loss": 0.1377170830965042,
+      "eval_runtime": 37.9358,
+      "eval_samples_per_second": 13.18,
+      "eval_steps_per_second": 0.844,
+      "step": 2700
+    },
+    {
+      "epoch": 0.7336220898754737,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0003811691859406301,
+      "loss": 0.1121,
+      "step": 2710
+    },
+    {
+      "epoch": 0.73632918245804,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.00038071389546530686,
+      "loss": 0.1426,
+      "step": 2720
+    },
+    {
+      "epoch": 0.7390362750406064,
+      "grad_norm": 0.375,
+      "learning_rate": 0.0003802586049899836,
+      "loss": 0.1319,
+      "step": 2730
+    },
+    {
+      "epoch": 0.7417433676231727,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00037980331451466036,
+      "loss": 0.123,
+      "step": 2740
+    },
+    {
+      "epoch": 0.744450460205739,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0003793480240393371,
+      "loss": 0.1048,
+      "step": 2750
+    },
+    {
+      "epoch": 0.7471575527883053,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0003788927335640138,
+      "loss": 0.1228,
+      "step": 2760
+    },
+    {
+      "epoch": 0.7498646453708717,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.0003784374430886906,
+      "loss": 0.144,
+      "step": 2770
+    },
+    {
+      "epoch": 0.752571737953438,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0003779821526133673,
+      "loss": 0.1292,
+      "step": 2780
+    },
+    {
+      "epoch": 0.7552788305360043,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0003775268621380441,
+      "loss": 0.1234,
+      "step": 2790
+    },
+    {
+      "epoch": 0.7579859231185706,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00037707157166272083,
+      "loss": 0.1363,
+      "step": 2800
+    },
+    {
+      "epoch": 0.7579859231185706,
+      "eval_loss": 0.1276787519454956,
+      "eval_runtime": 38.997,
+      "eval_samples_per_second": 12.822,
+      "eval_steps_per_second": 0.821,
+      "step": 2800
+    },
+    {
+      "epoch": 0.760693015701137,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00037661628118739756,
+      "loss": 0.1212,
+      "step": 2810
+    },
+    {
+      "epoch": 0.7634001082837033,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.00037616099071207434,
+      "loss": 0.1129,
+      "step": 2820
+    },
+    {
+      "epoch": 0.7661072008662696,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.00037570570023675107,
+      "loss": 0.103,
+      "step": 2830
+    },
+    {
+      "epoch": 0.7688142934488359,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00037525040976142785,
+      "loss": 0.1346,
+      "step": 2840
+    },
+    {
+      "epoch": 0.7715213860314023,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00037479511928610457,
+      "loss": 0.1365,
+      "step": 2850
+    },
+    {
+      "epoch": 0.7742284786139686,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00037433982881078124,
+      "loss": 0.1226,
+      "step": 2860
+    },
+    {
+      "epoch": 0.7769355711965349,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.000373884538335458,
+      "loss": 0.1205,
+      "step": 2870
+    },
+    {
+      "epoch": 0.7796426637791013,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00037342924786013475,
+      "loss": 0.137,
+      "step": 2880
+    },
+    {
+      "epoch": 0.7823497563616676,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0003729739573848115,
+      "loss": 0.1306,
+      "step": 2890
+    },
+    {
+      "epoch": 0.7850568489442339,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00037251866690948826,
+      "loss": 0.0963,
+      "step": 2900
+    },
+    {
+      "epoch": 0.7850568489442339,
+      "eval_loss": 0.12620532512664795,
+      "eval_runtime": 36.9199,
+      "eval_samples_per_second": 13.543,
+      "eval_steps_per_second": 0.867,
+      "step": 2900
+    },
+    {
+      "epoch": 0.7877639415268002,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.000372063376434165,
+      "loss": 0.1394,
+      "step": 2910
+    },
+    {
+      "epoch": 0.7904710341093666,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.00037160808595884177,
+      "loss": 0.1223,
+      "step": 2920
+    },
+    {
+      "epoch": 0.7931781266919329,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.0003711527954835185,
+      "loss": 0.1065,
+      "step": 2930
+    },
+    {
+      "epoch": 0.7958852192744992,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0003706975050081952,
+      "loss": 0.141,
+      "step": 2940
+    },
+    {
+      "epoch": 0.7985923118570655,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.000370242214532872,
+      "loss": 0.1221,
+      "step": 2950
+    },
+    {
+      "epoch": 0.8012994044396319,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0003697869240575487,
+      "loss": 0.1298,
+      "step": 2960
+    },
+    {
+      "epoch": 0.8040064970221982,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.0003693316335822255,
+      "loss": 0.1327,
+      "step": 2970
+    },
+    {
+      "epoch": 0.8067135896047645,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.00036887634310690223,
+      "loss": 0.1205,
+      "step": 2980
+    },
+    {
+      "epoch": 0.8094206821873308,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00036842105263157896,
+      "loss": 0.1298,
+      "step": 2990
+    },
+    {
+      "epoch": 0.8121277747698972,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0003679657621562557,
+      "loss": 0.1137,
+      "step": 3000
+    },
+    {
+      "epoch": 0.8121277747698972,
+      "eval_loss": 0.11737874150276184,
+      "eval_runtime": 40.0752,
+      "eval_samples_per_second": 12.477,
+      "eval_steps_per_second": 0.798,
+      "step": 3000
+    },
+    {
+      "epoch": 0.8148348673524635,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0003675104716809324,
+      "loss": 0.1192,
+      "step": 3010
+    },
+    {
+      "epoch": 0.8175419599350298,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0003670551812056092,
+      "loss": 0.1022,
+      "step": 3020
+    },
+    {
+      "epoch": 0.8202490525175961,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.0003665998907302859,
+      "loss": 0.1407,
+      "step": 3030
+    },
+    {
+      "epoch": 0.8229561451001625,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.00036614460025496265,
+      "loss": 0.1215,
+      "step": 3040
+    },
+    {
+      "epoch": 0.8256632376827288,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.00036568930977963943,
+      "loss": 0.115,
+      "step": 3050
+    },
+    {
+      "epoch": 0.828370330265295,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00036523401930431615,
+      "loss": 0.1423,
+      "step": 3060
+    },
+    {
+      "epoch": 0.8310774228478613,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0003647787288289929,
+      "loss": 0.1523,
+      "step": 3070
+    },
+    {
+      "epoch": 0.8337845154304278,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00036432343835366966,
+      "loss": 0.1138,
+      "step": 3080
+    },
+    {
+      "epoch": 0.836491608012994,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0003638681478783464,
+      "loss": 0.1224,
+      "step": 3090
+    },
+    {
+      "epoch": 0.8391987005955603,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.00036341285740302317,
+      "loss": 0.1245,
+      "step": 3100
+    },
+    {
+      "epoch": 0.8391987005955603,
+      "eval_loss": 0.11910546571016312,
+      "eval_runtime": 38.1778,
+      "eval_samples_per_second": 13.097,
+      "eval_steps_per_second": 0.838,
+      "step": 3100
+    },
+    {
+      "epoch": 0.8419057931781266,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.0003629575669276999,
+      "loss": 0.149,
+      "step": 3110
+    },
+    {
+      "epoch": 0.844612885760693,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.0003625022764523766,
+      "loss": 0.1449,
+      "step": 3120
+    },
+    {
+      "epoch": 0.8473199783432593,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0003620469859770534,
+      "loss": 0.1249,
+      "step": 3130
+    },
+    {
+      "epoch": 0.8500270709258256,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.0003615916955017301,
+      "loss": 0.1543,
+      "step": 3140
+    },
+    {
+      "epoch": 0.852734163508392,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00036113640502640686,
+      "loss": 0.1739,
+      "step": 3150
+    },
+    {
+      "epoch": 0.8554412560909583,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.0003606811145510836,
+      "loss": 0.1178,
+      "step": 3160
+    },
+    {
+      "epoch": 0.8581483486735246,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0003602258240757603,
+      "loss": 0.1503,
+      "step": 3170
+    },
+    {
+      "epoch": 0.8608554412560909,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.0003597705336004371,
+      "loss": 0.1313,
+      "step": 3180
+    },
+    {
+      "epoch": 0.8635625338386573,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0003593152431251138,
+      "loss": 0.1582,
+      "step": 3190
+    },
+    {
+      "epoch": 0.8662696264212236,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0003588599526497906,
+      "loss": 0.109,
+      "step": 3200
+    },
+    {
+      "epoch": 0.8662696264212236,
+      "eval_loss": 0.11779695004224777,
+      "eval_runtime": 38.9774,
+      "eval_samples_per_second": 12.828,
+      "eval_steps_per_second": 0.821,
+      "step": 3200
+    },
+    {
+      "epoch": 0.8689767190037899,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0003584046621744673,
+      "loss": 0.1341,
+      "step": 3210
+    },
+    {
+      "epoch": 0.8716838115863562,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00035794937169914405,
+      "loss": 0.1225,
+      "step": 3220
+    },
+    {
+      "epoch": 0.8743909041689226,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.00035749408122382083,
+      "loss": 0.1134,
+      "step": 3230
+    },
+    {
+      "epoch": 0.8770979967514889,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.00035703879074849756,
+      "loss": 0.1545,
+      "step": 3240
+    },
+    {
+      "epoch": 0.8798050893340552,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0003565835002731743,
+      "loss": 0.1192,
+      "step": 3250
+    },
+    {
+      "epoch": 0.8825121819166215,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00035612820979785106,
+      "loss": 0.1194,
+      "step": 3260
+    },
+    {
+      "epoch": 0.8852192744991879,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.0003556729193225278,
+      "loss": 0.1288,
+      "step": 3270
+    },
+    {
+      "epoch": 0.8879263670817542,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0003552176288472045,
+      "loss": 0.1278,
+      "step": 3280
+    },
+    {
+      "epoch": 0.8906334596643205,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00035476233837188124,
+      "loss": 0.1197,
+      "step": 3290
+    },
+    {
+      "epoch": 0.8933405522468868,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00035430704789655797,
+      "loss": 0.1553,
+      "step": 3300
+    },
+    {
+      "epoch": 0.8933405522468868,
+      "eval_loss": 0.121039979159832,
+      "eval_runtime": 38.445,
+      "eval_samples_per_second": 13.006,
+      "eval_steps_per_second": 0.832,
+      "step": 3300
+    },
+    {
+      "epoch": 0.8960476448294532,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00035385175742123475,
+      "loss": 0.111,
+      "step": 3310
+    },
+    {
+      "epoch": 0.8987547374120195,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0003533964669459115,
+      "loss": 0.0833,
+      "step": 3320
+    },
+    {
+      "epoch": 0.9014618299945858,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00035294117647058826,
+      "loss": 0.121,
+      "step": 3330
+    },
+    {
+      "epoch": 0.9041689225771521,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.000352485885995265,
+      "loss": 0.0713,
+      "step": 3340
+    },
+    {
+      "epoch": 0.9068760151597185,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0003520305955199417,
+      "loss": 0.1103,
+      "step": 3350
+    },
+    {
+      "epoch": 0.9095831077422848,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0003515753050446185,
+      "loss": 0.1191,
+      "step": 3360
+    },
+    {
+      "epoch": 0.9122902003248511,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.0003511200145692952,
+      "loss": 0.1558,
+      "step": 3370
+    },
+    {
+      "epoch": 0.9149972929074174,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.000350664724093972,
+      "loss": 0.1417,
+      "step": 3380
+    },
+    {
+      "epoch": 0.9177043854899838,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0003502094336186487,
+      "loss": 0.0834,
+      "step": 3390
+    },
+    {
+      "epoch": 0.9204114780725501,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.00034975414314332545,
+      "loss": 0.1322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.9204114780725501,
+      "eval_loss": 0.12038355320692062,
+      "eval_runtime": 37.7423,
+      "eval_samples_per_second": 13.248,
+      "eval_steps_per_second": 0.848,
+      "step": 3400
+    },
+    {
+      "epoch": 0.9231185706551164,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00034929885266800223,
+      "loss": 0.1402,
+      "step": 3410
+    },
+    {
+      "epoch": 0.9258256632376828,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0003488435621926789,
+      "loss": 0.1149,
+      "step": 3420
+    },
+    {
+      "epoch": 0.9285327558202491,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00034838827171735563,
+      "loss": 0.1371,
+      "step": 3430
+    },
+    {
+      "epoch": 0.9312398484028154,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0003479329812420324,
+      "loss": 0.1265,
+      "step": 3440
+    },
+    {
+      "epoch": 0.9339469409853817,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00034747769076670914,
+      "loss": 0.1262,
+      "step": 3450
+    },
+    {
+      "epoch": 0.9366540335679481,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0003470224002913859,
+      "loss": 0.1449,
+      "step": 3460
+    },
+    {
+      "epoch": 0.9393611261505144,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.00034656710981606265,
+      "loss": 0.107,
+      "step": 3470
+    },
+    {
+      "epoch": 0.9420682187330807,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.0003461118193407394,
+      "loss": 0.1311,
+      "step": 3480
+    },
+    {
+      "epoch": 0.944775311315647,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00034565652886541615,
+      "loss": 0.1397,
+      "step": 3490
+    },
+    {
+      "epoch": 0.9474824038982134,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0003452012383900929,
+      "loss": 0.1094,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9474824038982134,
+      "eval_loss": 0.11934860050678253,
+      "eval_runtime": 42.0038,
+      "eval_samples_per_second": 11.904,
+      "eval_steps_per_second": 0.762,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9501894964807797,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00034474594791476966,
+      "loss": 0.0849,
+      "step": 3510
+    },
+    {
+      "epoch": 0.952896589063346,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.0003442906574394464,
+      "loss": 0.1449,
+      "step": 3520
+    },
+    {
+      "epoch": 0.9556036816459123,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.0003438353669641231,
+      "loss": 0.09,
+      "step": 3530
+    },
+    {
+      "epoch": 0.9583107742284787,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.0003433800764887999,
+      "loss": 0.114,
+      "step": 3540
+    },
+    {
+      "epoch": 0.961017866811045,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.0003429247860134766,
+      "loss": 0.1139,
+      "step": 3550
+    },
+    {
+      "epoch": 0.9637249593936112,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.00034246949553815335,
+      "loss": 0.1504,
+      "step": 3560
+    },
+    {
+      "epoch": 0.9664320519761775,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0003420142050628301,
+      "loss": 0.1239,
+      "step": 3570
+    },
+    {
+      "epoch": 0.969139144558744,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.0003415589145875068,
+      "loss": 0.1371,
+      "step": 3580
+    },
+    {
+      "epoch": 0.9718462371413102,
+      "grad_norm": 0.04541015625,
+      "learning_rate": 0.0003411036241121836,
+      "loss": 0.1405,
+      "step": 3590
+    },
+    {
+      "epoch": 0.9745533297238765,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0003406483336368603,
+      "loss": 0.1161,
+      "step": 3600
+    },
+    {
+      "epoch": 0.9745533297238765,
+      "eval_loss": 0.11894183605909348,
+      "eval_runtime": 38.5178,
+      "eval_samples_per_second": 12.981,
+      "eval_steps_per_second": 0.831,
+      "step": 3600
+    },
+    {
+      "epoch": 0.9772604223064428,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00034019304316153703,
+      "loss": 0.1261,
+      "step": 3610
+    },
+    {
+      "epoch": 0.9799675148890092,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0003397377526862138,
+      "loss": 0.1114,
+      "step": 3620
+    },
+    {
+      "epoch": 0.9826746074715755,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.00033928246221089054,
+      "loss": 0.121,
+      "step": 3630
+    },
+    {
+      "epoch": 0.9853817000541418,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0003388271717355673,
+      "loss": 0.123,
+      "step": 3640
+    },
+    {
+      "epoch": 0.9880887926367081,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.00033837188126024405,
+      "loss": 0.1275,
+      "step": 3650
+    },
+    {
+      "epoch": 0.9907958852192745,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.0003379165907849208,
+      "loss": 0.1321,
+      "step": 3660
+    },
+    {
+      "epoch": 0.9935029778018408,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00033746130030959756,
+      "loss": 0.1401,
+      "step": 3670
+    },
+    {
+      "epoch": 0.9962100703844071,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.0003370060098342743,
+      "loss": 0.1018,
+      "step": 3680
+    },
+    {
+      "epoch": 0.9989171629669734,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00033655071935895106,
+      "loss": 0.1132,
+      "step": 3690
+    },
+    {
+      "epoch": 1.0016242555495398,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00033609542888362774,
+      "loss": 0.1288,
+      "step": 3700
+    },
+    {
+      "epoch": 1.0016242555495398,
+      "eval_loss": 0.1131463572382927,
+      "eval_runtime": 40.8237,
+      "eval_samples_per_second": 12.248,
+      "eval_steps_per_second": 0.784,
+      "step": 3700
+    },
+    {
+      "epoch": 1.0043313481321061,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00033564013840830446,
+      "loss": 0.1046,
+      "step": 3710
+    },
+    {
+      "epoch": 1.0070384407146724,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00033518484793298124,
+      "loss": 0.107,
+      "step": 3720
+    },
+    {
+      "epoch": 1.0097455332972387,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.00033472955745765797,
+      "loss": 0.0998,
+      "step": 3730
+    },
+    {
+      "epoch": 1.012452625879805,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.00033427426698233475,
+      "loss": 0.1289,
+      "step": 3740
+    },
+    {
+      "epoch": 1.0151597184623715,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0003338189765070115,
+      "loss": 0.1021,
+      "step": 3750
+    },
+    {
+      "epoch": 1.0178668110449378,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0003333636860316882,
+      "loss": 0.1126,
+      "step": 3760
+    },
+    {
+      "epoch": 1.020573903627504,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.000332908395556365,
+      "loss": 0.1091,
+      "step": 3770
+    },
+    {
+      "epoch": 1.0232809962100704,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0003324531050810417,
+      "loss": 0.1305,
+      "step": 3780
+    },
+    {
+      "epoch": 1.0259880887926367,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00033199781460571844,
+      "loss": 0.1031,
+      "step": 3790
+    },
+    {
+      "epoch": 1.028695181375203,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.0003315425241303952,
+      "loss": 0.1155,
+      "step": 3800
+    },
+    {
+      "epoch": 1.028695181375203,
+      "eval_loss": 0.12730546295642853,
+      "eval_runtime": 41.3126,
+      "eval_samples_per_second": 12.103,
+      "eval_steps_per_second": 0.775,
+      "step": 3800
+    },
+    {
+      "epoch": 1.0314022739577693,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00033108723365507194,
+      "loss": 0.128,
+      "step": 3810
+    },
+    {
+      "epoch": 1.0341093665403356,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0003306319431797487,
+      "loss": 0.1377,
+      "step": 3820
+    },
+    {
+      "epoch": 1.036816459122902,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.00033017665270442545,
+      "loss": 0.0985,
+      "step": 3830
+    },
+    {
+      "epoch": 1.0395235517054684,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0003297213622291021,
+      "loss": 0.133,
+      "step": 3840
+    },
+    {
+      "epoch": 1.0422306442880347,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.0003292660717537789,
+      "loss": 0.1003,
+      "step": 3850
+    },
+    {
+      "epoch": 1.044937736870601,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00032881078127845563,
+      "loss": 0.1267,
+      "step": 3860
+    },
+    {
+      "epoch": 1.0476448294531673,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.0003283554908031324,
+      "loss": 0.1278,
+      "step": 3870
+    },
+    {
+      "epoch": 1.0503519220357336,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00032790020032780914,
+      "loss": 0.1297,
+      "step": 3880
+    },
+    {
+      "epoch": 1.0530590146182999,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00032744490985248587,
+      "loss": 0.1042,
+      "step": 3890
+    },
+    {
+      "epoch": 1.0557661072008662,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.00032698961937716265,
+      "loss": 0.1137,
+      "step": 3900
+    },
+    {
+      "epoch": 1.0557661072008662,
+      "eval_loss": 0.13213278353214264,
+      "eval_runtime": 40.0183,
+      "eval_samples_per_second": 12.494,
+      "eval_steps_per_second": 0.8,
+      "step": 3900
+    },
+    {
+      "epoch": 1.0584731997834327,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00032653432890183937,
+      "loss": 0.1327,
+      "step": 3910
+    },
+    {
+      "epoch": 1.061180292365999,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00032607903842651615,
+      "loss": 0.0866,
+      "step": 3920
+    },
+    {
+      "epoch": 1.0638873849485653,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0003256237479511929,
+      "loss": 0.1132,
+      "step": 3930
+    },
+    {
+      "epoch": 1.0665944775311316,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.0003251684574758696,
+      "loss": 0.1088,
+      "step": 3940
+    },
+    {
+      "epoch": 1.0693015701136979,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0003247131670005464,
+      "loss": 0.119,
+      "step": 3950
+    },
+    {
+      "epoch": 1.0720086626962642,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.0003242578765252231,
+      "loss": 0.122,
+      "step": 3960
+    },
+    {
+      "epoch": 1.0747157552788305,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.00032380258604989984,
+      "loss": 0.1392,
+      "step": 3970
+    },
+    {
+      "epoch": 1.077422847861397,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.0003233472955745766,
+      "loss": 0.1316,
+      "step": 3980
+    },
+    {
+      "epoch": 1.0801299404439633,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0003228920050992533,
+      "loss": 0.1043,
+      "step": 3990
+    },
+    {
+      "epoch": 1.0828370330265296,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.0003224367146239301,
+      "loss": 0.0902,
+      "step": 4000
+    },
+    {
+      "epoch": 1.0828370330265296,
+      "eval_loss": 0.1267096847295761,
+      "eval_runtime": 39.5817,
+      "eval_samples_per_second": 12.632,
+      "eval_steps_per_second": 0.808,
+      "step": 4000
+    },
+    {
+      "epoch": 1.0855441256090959,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0003219814241486068,
+      "loss": 0.0885,
+      "step": 4010
+    },
+    {
+      "epoch": 1.0882512181916622,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0003215261336732835,
+      "loss": 0.1009,
+      "step": 4020
+    },
+    {
+      "epoch": 1.0909583107742284,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0003210708431979603,
+      "loss": 0.1196,
+      "step": 4030
+    },
+    {
+      "epoch": 1.0936654033567947,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.00032061555272263703,
+      "loss": 0.09,
+      "step": 4040
+    },
+    {
+      "epoch": 1.096372495939361,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.0003201602622473138,
+      "loss": 0.0979,
+      "step": 4050
+    },
+    {
+      "epoch": 1.0990795885219276,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00031970497177199054,
+      "loss": 0.0712,
+      "step": 4060
+    },
+    {
+      "epoch": 1.1017866811044938,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.00031924968129666727,
+      "loss": 0.0862,
+      "step": 4070
+    },
+    {
+      "epoch": 1.1044937736870601,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.00031879439082134405,
+      "loss": 0.1119,
+      "step": 4080
+    },
+    {
+      "epoch": 1.1072008662696264,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.0003183391003460208,
+      "loss": 0.0858,
+      "step": 4090
+    },
+    {
+      "epoch": 1.1099079588521927,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00031788380987069756,
+      "loss": 0.1125,
+      "step": 4100
+    },
+    {
+      "epoch": 1.1099079588521927,
+      "eval_loss": 0.11800329387187958,
+      "eval_runtime": 39.7373,
+      "eval_samples_per_second": 12.583,
+      "eval_steps_per_second": 0.805,
+      "step": 4100
+    },
+    {
+      "epoch": 1.112615051434759,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.0003174285193953743,
+      "loss": 0.112,
+      "step": 4110
+    },
+    {
+      "epoch": 1.1153221440173253,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.000316973228920051,
+      "loss": 0.0926,
+      "step": 4120
+    },
+    {
+      "epoch": 1.1180292365998916,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.00031651793844472774,
+      "loss": 0.0932,
+      "step": 4130
+    },
+    {
+      "epoch": 1.1207363291824581,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00031606264796940446,
+      "loss": 0.1129,
+      "step": 4140
+    },
+    {
+      "epoch": 1.1234434217650244,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.0003156073574940812,
+      "loss": 0.1132,
+      "step": 4150
+    },
+    {
+      "epoch": 1.1261505143475907,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.00031515206701875797,
+      "loss": 0.104,
+      "step": 4160
+    },
+    {
+      "epoch": 1.128857606930157,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0003146967765434347,
+      "loss": 0.1011,
+      "step": 4170
+    },
+    {
+      "epoch": 1.1315646995127233,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0003142414860681115,
+      "loss": 0.0928,
+      "step": 4180
+    },
+    {
+      "epoch": 1.1342717920952896,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0003137861955927882,
+      "loss": 0.1058,
+      "step": 4190
+    },
+    {
+      "epoch": 1.136978884677856,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.00031333090511746493,
+      "loss": 0.1273,
+      "step": 4200
+    },
+    {
+      "epoch": 1.136978884677856,
+      "eval_loss": 0.12301485985517502,
+      "eval_runtime": 39.6523,
+      "eval_samples_per_second": 12.61,
+      "eval_steps_per_second": 0.807,
+      "step": 4200
+    },
+    {
+      "epoch": 1.1396859772604224,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0003128756146421417,
+      "loss": 0.1019,
+      "step": 4210
+    },
+    {
+      "epoch": 1.1423930698429887,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00031242032416681844,
+      "loss": 0.0924,
+      "step": 4220
+    },
+    {
+      "epoch": 1.145100162425555,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0003119650336914952,
+      "loss": 0.1101,
+      "step": 4230
+    },
+    {
+      "epoch": 1.1478072550081213,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.00031150974321617194,
+      "loss": 0.0659,
+      "step": 4240
+    },
+    {
+      "epoch": 1.1505143475906876,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00031105445274084867,
+      "loss": 0.1175,
+      "step": 4250
+    },
+    {
+      "epoch": 1.153221440173254,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.00031059916226552545,
+      "loss": 0.1113,
+      "step": 4260
+    },
+    {
+      "epoch": 1.1559285327558202,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.0003101438717902021,
+      "loss": 0.0947,
+      "step": 4270
+    },
+    {
+      "epoch": 1.1586356253383865,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.0003096885813148789,
+      "loss": 0.1187,
+      "step": 4280
+    },
+    {
+      "epoch": 1.1613427179209528,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00030923329083955563,
+      "loss": 0.1121,
+      "step": 4290
+    },
+    {
+      "epoch": 1.1640498105035193,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00030877800036423236,
+      "loss": 0.107,
+      "step": 4300
+    },
+    {
+      "epoch": 1.1640498105035193,
+      "eval_loss": 0.11996840685606003,
+      "eval_runtime": 40.6861,
+      "eval_samples_per_second": 12.289,
+      "eval_steps_per_second": 0.787,
+      "step": 4300
+    },
+    {
+      "epoch": 1.1667569030860856,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00030832270988890914,
+      "loss": 0.1103,
+      "step": 4310
+    },
+    {
+      "epoch": 1.169463995668652,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.00030786741941358586,
+      "loss": 0.099,
+      "step": 4320
+    },
+    {
+      "epoch": 1.1721710882512182,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0003074121289382626,
+      "loss": 0.1028,
+      "step": 4330
+    },
+    {
+      "epoch": 1.1748781808337845,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00030695683846293937,
+      "loss": 0.1539,
+      "step": 4340
+    },
+    {
+      "epoch": 1.1775852734163508,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0003065015479876161,
+      "loss": 0.0825,
+      "step": 4350
+    },
+    {
+      "epoch": 1.180292365998917,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0003060462575122929,
+      "loss": 0.0821,
+      "step": 4360
+    },
+    {
+      "epoch": 1.1829994585814836,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.0003055909670369696,
+      "loss": 0.1258,
+      "step": 4370
+    },
+    {
+      "epoch": 1.1857065511640499,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00030513567656164633,
+      "loss": 0.1169,
+      "step": 4380
+    },
+    {
+      "epoch": 1.1884136437466162,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.0003046803860863231,
+      "loss": 0.103,
+      "step": 4390
+    },
+    {
+      "epoch": 1.1911207363291825,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.00030422509561099984,
+      "loss": 0.1101,
+      "step": 4400
+    },
+    {
+      "epoch": 1.1911207363291825,
+      "eval_loss": 0.11814282089471817,
+      "eval_runtime": 39.7722,
+      "eval_samples_per_second": 12.572,
+      "eval_steps_per_second": 0.805,
+      "step": 4400
+    },
+    {
+      "epoch": 1.1938278289117488,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.00030376980513567657,
+      "loss": 0.0832,
+      "step": 4410
+    },
+    {
+      "epoch": 1.196534921494315,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0003033145146603533,
+      "loss": 0.079,
+      "step": 4420
+    },
+    {
+      "epoch": 1.1992420140768814,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00030285922418503,
+      "loss": 0.1129,
+      "step": 4430
+    },
+    {
+      "epoch": 1.2019491066594479,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.0003024039337097068,
+      "loss": 0.1219,
+      "step": 4440
+    },
+    {
+      "epoch": 1.2046561992420142,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0003019486432343835,
+      "loss": 0.1056,
+      "step": 4450
+    },
+    {
+      "epoch": 1.2073632918245805,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0003014933527590603,
+      "loss": 0.1318,
+      "step": 4460
+    },
+    {
+      "epoch": 1.2100703844071468,
+      "grad_norm": 0.03759765625,
+      "learning_rate": 0.00030103806228373703,
+      "loss": 0.0913,
+      "step": 4470
+    },
+    {
+      "epoch": 1.212777476989713,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.00030058277180841376,
+      "loss": 0.1169,
+      "step": 4480
+    },
+    {
+      "epoch": 1.2154845695722793,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.00030012748133309054,
+      "loss": 0.1011,
+      "step": 4490
+    },
+    {
+      "epoch": 1.2181916621548456,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.00029967219085776727,
+      "loss": 0.0834,
+      "step": 4500
+    },
+    {
+      "epoch": 1.2181916621548456,
+      "eval_loss": 0.11971238255500793,
+      "eval_runtime": 40.4082,
+      "eval_samples_per_second": 12.374,
+      "eval_steps_per_second": 0.792,
+      "step": 4500
+    },
+    {
+      "epoch": 1.220898754737412,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.000299216900382444,
+      "loss": 0.1424,
+      "step": 4510
+    },
+    {
+      "epoch": 1.2236058473199782,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.0002987616099071208,
+      "loss": 0.1215,
+      "step": 4520
+    },
+    {
+      "epoch": 1.2263129399025448,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.0002983063194317975,
+      "loss": 0.1197,
+      "step": 4530
+    },
+    {
+      "epoch": 1.229020032485111,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0002978510289564743,
+      "loss": 0.0992,
+      "step": 4540
+    },
+    {
+      "epoch": 1.2317271250676773,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.00029739573848115095,
+      "loss": 0.1004,
+      "step": 4550
+    },
+    {
+      "epoch": 1.2344342176502436,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0002969404480058277,
+      "loss": 0.1156,
+      "step": 4560
+    },
+    {
+      "epoch": 1.23714131023281,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00029648515753050446,
+      "loss": 0.0987,
+      "step": 4570
+    },
+    {
+      "epoch": 1.2398484028153762,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0002960298670551812,
+      "loss": 0.0976,
+      "step": 4580
+    },
+    {
+      "epoch": 1.2425554953979425,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00029557457657985797,
+      "loss": 0.1162,
+      "step": 4590
+    },
+    {
+      "epoch": 1.245262587980509,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0002951192861045347,
+      "loss": 0.0946,
+      "step": 4600
+    },
+    {
+      "epoch": 1.245262587980509,
+      "eval_loss": 0.11577141284942627,
+      "eval_runtime": 40.3235,
+      "eval_samples_per_second": 12.4,
+      "eval_steps_per_second": 0.794,
+      "step": 4600
+    },
+    {
+      "epoch": 1.2479696805630753,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.0002946639956292114,
+      "loss": 0.0847,
+      "step": 4610
+    },
+    {
+      "epoch": 1.2506767731456416,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0002942087051538882,
+      "loss": 0.1086,
+      "step": 4620
+    },
+    {
+      "epoch": 1.253383865728208,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00029375341467856493,
+      "loss": 0.0946,
+      "step": 4630
+    },
+    {
+      "epoch": 1.2560909583107742,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0002932981242032417,
+      "loss": 0.1022,
+      "step": 4640
+    },
+    {
+      "epoch": 1.2587980508933405,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.00029284283372791844,
+      "loss": 0.0824,
+      "step": 4650
+    },
+    {
+      "epoch": 1.2615051434759068,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.00029238754325259516,
+      "loss": 0.1017,
+      "step": 4660
+    },
+    {
+      "epoch": 1.2642122360584733,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00029193225277727194,
+      "loss": 0.0725,
+      "step": 4670
+    },
+    {
+      "epoch": 1.2669193286410394,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.00029147696230194867,
+      "loss": 0.0964,
+      "step": 4680
+    },
+    {
+      "epoch": 1.269626421223606,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.00029102167182662534,
+      "loss": 0.0863,
+      "step": 4690
+    },
+    {
+      "epoch": 1.2723335138061722,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002905663813513021,
+      "loss": 0.0912,
+      "step": 4700
+    },
+    {
+      "epoch": 1.2723335138061722,
+      "eval_loss": 0.12092173099517822,
+      "eval_runtime": 37.4322,
+      "eval_samples_per_second": 13.357,
+      "eval_steps_per_second": 0.855,
+      "step": 4700
+    },
+    {
+      "epoch": 1.2750406063887385,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00029011109087597885,
+      "loss": 0.1198,
+      "step": 4710
+    },
+    {
+      "epoch": 1.2777476989713048,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00028965580040065563,
+      "loss": 0.1071,
+      "step": 4720
+    },
+    {
+      "epoch": 1.280454791553871,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.00028920050992533236,
+      "loss": 0.145,
+      "step": 4730
+    },
+    {
+      "epoch": 1.2831618841364374,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0002887452194500091,
+      "loss": 0.0872,
+      "step": 4740
+    },
+    {
+      "epoch": 1.2858689767190037,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.00028828992897468586,
+      "loss": 0.0945,
+      "step": 4750
+    },
+    {
+      "epoch": 1.2885760693015702,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0002878346384993626,
+      "loss": 0.1104,
+      "step": 4760
+    },
+    {
+      "epoch": 1.2912831618841365,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.00028737934802403937,
+      "loss": 0.0873,
+      "step": 4770
+    },
+    {
+      "epoch": 1.2939902544667028,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002869240575487161,
+      "loss": 0.0925,
+      "step": 4780
+    },
+    {
+      "epoch": 1.296697347049269,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0002864687670733928,
+      "loss": 0.1177,
+      "step": 4790
+    },
+    {
+      "epoch": 1.2994044396318354,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0002860134765980696,
+      "loss": 0.1131,
+      "step": 4800
+    },
+    {
+      "epoch": 1.2994044396318354,
+      "eval_loss": 0.11782730370759964,
+      "eval_runtime": 41.533,
+      "eval_samples_per_second": 12.039,
+      "eval_steps_per_second": 0.77,
+      "step": 4800
+    },
+    {
+      "epoch": 1.3021115322144017,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.00028555818612274633,
+      "loss": 0.0798,
+      "step": 4810
+    },
+    {
+      "epoch": 1.304818624796968,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0002851028956474231,
+      "loss": 0.1115,
+      "step": 4820
+    },
+    {
+      "epoch": 1.3075257173795345,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0002846476051720998,
+      "loss": 0.1045,
+      "step": 4830
+    },
+    {
+      "epoch": 1.3102328099621008,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.0002841923146967765,
+      "loss": 0.091,
+      "step": 4840
+    },
+    {
+      "epoch": 1.312939902544667,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.0002837370242214533,
+      "loss": 0.101,
+      "step": 4850
+    },
+    {
+      "epoch": 1.3156469951272334,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.00028328173374613,
+      "loss": 0.1312,
+      "step": 4860
+    },
+    {
+      "epoch": 1.3183540877097997,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00028282644327080674,
+      "loss": 0.0798,
+      "step": 4870
+    },
+    {
+      "epoch": 1.321061180292366,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0002823711527954835,
+      "loss": 0.1023,
+      "step": 4880
+    },
+    {
+      "epoch": 1.3237682728749323,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00028191586232016025,
+      "loss": 0.1415,
+      "step": 4890
+    },
+    {
+      "epoch": 1.3264753654574988,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.00028146057184483703,
+      "loss": 0.1495,
+      "step": 4900
+    },
+    {
+      "epoch": 1.3264753654574988,
+      "eval_loss": 0.11481285095214844,
+      "eval_runtime": 39.4863,
+      "eval_samples_per_second": 12.663,
+      "eval_steps_per_second": 0.81,
+      "step": 4900
+    },
+    {
+      "epoch": 1.3291824580400649,
+      "grad_norm": 0.033447265625,
+      "learning_rate": 0.00028100528136951376,
+      "loss": 0.1077,
+      "step": 4910
+    },
+    {
+      "epoch": 1.3318895506226314,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.0002805499908941905,
+      "loss": 0.1228,
+      "step": 4920
+    },
+    {
+      "epoch": 1.3345966432051977,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00028009470041886727,
+      "loss": 0.1077,
+      "step": 4930
+    },
+    {
+      "epoch": 1.337303735787764,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.000279639409943544,
+      "loss": 0.1179,
+      "step": 4940
+    },
+    {
+      "epoch": 1.3400108283703303,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0002791841194682208,
+      "loss": 0.1034,
+      "step": 4950
+    },
+    {
+      "epoch": 1.3427179209528965,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0002787288289928975,
+      "loss": 0.1298,
+      "step": 4960
+    },
+    {
+      "epoch": 1.3454250135354628,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.00027827353851757417,
+      "loss": 0.0908,
+      "step": 4970
+    },
+    {
+      "epoch": 1.3481321061180291,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.00027781824804225095,
+      "loss": 0.109,
+      "step": 4980
+    },
+    {
+      "epoch": 1.3508391987005957,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0002773629575669277,
+      "loss": 0.1044,
+      "step": 4990
+    },
+    {
+      "epoch": 1.353546291283162,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00027690766709160446,
+      "loss": 0.1165,
+      "step": 5000
+    },
+    {
+      "epoch": 1.353546291283162,
+      "eval_loss": 0.11778061091899872,
+      "eval_runtime": 39.2971,
+      "eval_samples_per_second": 12.724,
+      "eval_steps_per_second": 0.814,
+      "step": 5000
+    },
+    {
+      "epoch": 1.3562533838657282,
+      "grad_norm": 0.0537109375,
+      "learning_rate": 0.0002764523766162812,
+      "loss": 0.1143,
+      "step": 5010
+    },
+    {
+      "epoch": 1.3589604764482945,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0002759970861409579,
+      "loss": 0.095,
+      "step": 5020
+    },
+    {
+      "epoch": 1.3616675690308608,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.0002755417956656347,
+      "loss": 0.0966,
+      "step": 5030
+    },
+    {
+      "epoch": 1.3643746616134271,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0002750865051903114,
+      "loss": 0.0914,
+      "step": 5040
+    },
+    {
+      "epoch": 1.3670817541959934,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.00027463121471498815,
+      "loss": 0.0967,
+      "step": 5050
+    },
+    {
+      "epoch": 1.36978884677856,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.00027417592423966493,
+      "loss": 0.1003,
+      "step": 5060
+    },
+    {
+      "epoch": 1.372495939361126,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00027372063376434165,
+      "loss": 0.1457,
+      "step": 5070
+    },
+    {
+      "epoch": 1.3752030319436925,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00027326534328901843,
+      "loss": 0.0941,
+      "step": 5080
+    },
+    {
+      "epoch": 1.3779101245262588,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00027281005281369516,
+      "loss": 0.1045,
+      "step": 5090
+    },
+    {
+      "epoch": 1.3806172171088251,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.0002723547623383719,
+      "loss": 0.113,
+      "step": 5100
+    },
+    {
+      "epoch": 1.3806172171088251,
+      "eval_loss": 0.11332318186759949,
+      "eval_runtime": 39.6613,
+      "eval_samples_per_second": 12.607,
+      "eval_steps_per_second": 0.807,
+      "step": 5100
+    },
+    {
+      "epoch": 1.3833243096913914,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00027189947186304867,
+      "loss": 0.1225,
+      "step": 5110
+    },
+    {
+      "epoch": 1.3860314022739577,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.00027144418138772534,
+      "loss": 0.11,
+      "step": 5120
+    },
+    {
+      "epoch": 1.3887384948565242,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0002709888909124021,
+      "loss": 0.1042,
+      "step": 5130
+    },
+    {
+      "epoch": 1.3914455874390903,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.00027053360043707885,
+      "loss": 0.1091,
+      "step": 5140
+    },
+    {
+      "epoch": 1.3941526800216568,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0002700783099617556,
+      "loss": 0.1222,
+      "step": 5150
+    },
+    {
+      "epoch": 1.3968597726042231,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00026962301948643236,
+      "loss": 0.0904,
+      "step": 5160
+    },
+    {
+      "epoch": 1.3995668651867894,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0002691677290111091,
+      "loss": 0.1117,
+      "step": 5170
+    },
+    {
+      "epoch": 1.4022739577693557,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.00026871243853578586,
+      "loss": 0.1238,
+      "step": 5180
+    },
+    {
+      "epoch": 1.404981050351922,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0002682571480604626,
+      "loss": 0.1132,
+      "step": 5190
+    },
+    {
+      "epoch": 1.4076881429344883,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.0002678018575851393,
+      "loss": 0.1079,
+      "step": 5200
+    },
+    {
+      "epoch": 1.4076881429344883,
+      "eval_loss": 0.10996056348085403,
+      "eval_runtime": 39.438,
+      "eval_samples_per_second": 12.678,
+      "eval_steps_per_second": 0.811,
+      "step": 5200
+    },
+    {
+      "epoch": 1.4103952355170546,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0002673465671098161,
+      "loss": 0.1106,
+      "step": 5210
+    },
+    {
+      "epoch": 1.413102328099621,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.0002668912766344928,
+      "loss": 0.1174,
+      "step": 5220
+    },
+    {
+      "epoch": 1.4158094206821874,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.00026643598615916955,
+      "loss": 0.1096,
+      "step": 5230
+    },
+    {
+      "epoch": 1.4185165132647537,
+      "grad_norm": 0.05517578125,
+      "learning_rate": 0.00026598069568384633,
+      "loss": 0.103,
+      "step": 5240
+    },
+    {
+      "epoch": 1.42122360584732,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.00026552540520852306,
+      "loss": 0.1042,
+      "step": 5250
+    },
+    {
+      "epoch": 1.4239306984298863,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0002650701147331998,
+      "loss": 0.0832,
+      "step": 5260
+    },
+    {
+      "epoch": 1.4266377910124526,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0002646148242578765,
+      "loss": 0.0908,
+      "step": 5270
+    },
+    {
+      "epoch": 1.4293448835950189,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00026415953378255324,
+      "loss": 0.0812,
+      "step": 5280
+    },
+    {
+      "epoch": 1.4320519761775854,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00026370424330723,
+      "loss": 0.0986,
+      "step": 5290
+    },
+    {
+      "epoch": 1.4347590687601515,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00026324895283190674,
+      "loss": 0.095,
+      "step": 5300
+    },
+    {
+      "epoch": 1.4347590687601515,
+      "eval_loss": 0.1117488220334053,
+      "eval_runtime": 38.1549,
+      "eval_samples_per_second": 13.104,
+      "eval_steps_per_second": 0.839,
+      "step": 5300
+    },
+    {
+      "epoch": 1.437466161342718,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.0002627936623565835,
+      "loss": 0.0827,
+      "step": 5310
+    },
+    {
+      "epoch": 1.4401732539252843,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.00026233837188126025,
+      "loss": 0.1282,
+      "step": 5320
+    },
+    {
+      "epoch": 1.4428803465078506,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.000261883081405937,
+      "loss": 0.1364,
+      "step": 5330
+    },
+    {
+      "epoch": 1.4455874390904169,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.00026142779093061376,
+      "loss": 0.097,
+      "step": 5340
+    },
+    {
+      "epoch": 1.4482945316729832,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.0002609725004552905,
+      "loss": 0.0879,
+      "step": 5350
+    },
+    {
+      "epoch": 1.4510016242555495,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00026051720997996726,
+      "loss": 0.1077,
+      "step": 5360
+    },
+    {
+      "epoch": 1.4537087168381158,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.000260061919504644,
+      "loss": 0.0768,
+      "step": 5370
+    },
+    {
+      "epoch": 1.4564158094206823,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.0002596066290293207,
+      "loss": 0.1017,
+      "step": 5380
+    },
+    {
+      "epoch": 1.4591229020032486,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.0002591513385539975,
+      "loss": 0.1392,
+      "step": 5390
+    },
+    {
+      "epoch": 1.4618299945858149,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00025869604807867417,
+      "loss": 0.1141,
+      "step": 5400
+    },
+    {
+      "epoch": 1.4618299945858149,
+      "eval_loss": 0.11587672680616379,
+      "eval_runtime": 37.6874,
+      "eval_samples_per_second": 13.267,
+      "eval_steps_per_second": 0.849,
+      "step": 5400
+    },
+    {
+      "epoch": 1.4645370871683812,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0002582407576033509,
+      "loss": 0.116,
+      "step": 5410
+    },
+    {
+      "epoch": 1.4672441797509475,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0002577854671280277,
+      "loss": 0.0827,
+      "step": 5420
+    },
+    {
+      "epoch": 1.4699512723335137,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.0002573301766527044,
+      "loss": 0.0996,
+      "step": 5430
+    },
+    {
+      "epoch": 1.47265836491608,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.0002568748861773812,
+      "loss": 0.0986,
+      "step": 5440
+    },
+    {
+      "epoch": 1.4753654574986466,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0002564195957020579,
+      "loss": 0.0899,
+      "step": 5450
+    },
+    {
+      "epoch": 1.4780725500812129,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00025596430522673464,
+      "loss": 0.1163,
+      "step": 5460
+    },
+    {
+      "epoch": 1.4807796426637791,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0002555090147514114,
+      "loss": 0.1429,
+      "step": 5470
+    },
+    {
+      "epoch": 1.4834867352463454,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00025505372427608815,
+      "loss": 0.1164,
+      "step": 5480
+    },
+    {
+      "epoch": 1.4861938278289117,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.0002545984338007649,
+      "loss": 0.0843,
+      "step": 5490
+    },
+    {
+      "epoch": 1.488900920411478,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.00025414314332544165,
+      "loss": 0.1051,
+      "step": 5500
+    },
+    {
+      "epoch": 1.488900920411478,
+      "eval_loss": 0.11653291434049606,
+      "eval_runtime": 40.0288,
+      "eval_samples_per_second": 12.491,
+      "eval_steps_per_second": 0.799,
+      "step": 5500
+    },
+    {
+      "epoch": 1.4916080129940443,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0002536878528501184,
+      "loss": 0.0868,
+      "step": 5510
+    },
+    {
+      "epoch": 1.4943151055766108,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.00025323256237479516,
+      "loss": 0.0935,
+      "step": 5520
+    },
+    {
+      "epoch": 1.497022198159177,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0002527772718994719,
+      "loss": 0.1148,
+      "step": 5530
+    },
+    {
+      "epoch": 1.4997292907417434,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0002523219814241486,
+      "loss": 0.118,
+      "step": 5540
+    },
+    {
+      "epoch": 1.5024363833243097,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00025186669094882534,
+      "loss": 0.1216,
+      "step": 5550
+    },
+    {
+      "epoch": 1.505143475906876,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00025141140047350207,
+      "loss": 0.0875,
+      "step": 5560
+    },
+    {
+      "epoch": 1.5078505684894423,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00025095610999817885,
+      "loss": 0.1039,
+      "step": 5570
+    },
+    {
+      "epoch": 1.5105576610720086,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.0002505008195228556,
+      "loss": 0.0991,
+      "step": 5580
+    },
+    {
+      "epoch": 1.5132647536545751,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0002500455290475323,
+      "loss": 0.125,
+      "step": 5590
+    },
+    {
+      "epoch": 1.5159718462371412,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.0002495902385722091,
+      "loss": 0.105,
+      "step": 5600
+    },
+    {
+      "epoch": 1.5159718462371412,
+      "eval_loss": 0.11451131105422974,
+      "eval_runtime": 38.7843,
+      "eval_samples_per_second": 12.892,
+      "eval_steps_per_second": 0.825,
+      "step": 5600
+    },
+    {
+      "epoch": 1.5186789388197077,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.0002491349480968858,
+      "loss": 0.1141,
+      "step": 5610
+    },
+    {
+      "epoch": 1.5213860314022738,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.00024867965762156253,
+      "loss": 0.1081,
+      "step": 5620
+    },
+    {
+      "epoch": 1.5240931239848403,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.0002482243671462393,
+      "loss": 0.0963,
+      "step": 5630
+    },
+    {
+      "epoch": 1.5268002165674066,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00024776907667091604,
+      "loss": 0.1038,
+      "step": 5640
+    },
+    {
+      "epoch": 1.529507309149973,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0002473137861955928,
+      "loss": 0.0928,
+      "step": 5650
+    },
+    {
+      "epoch": 1.5322144017325392,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00024685849572026955,
+      "loss": 0.0973,
+      "step": 5660
+    },
+    {
+      "epoch": 1.5349214943151055,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.0002464032052449463,
+      "loss": 0.125,
+      "step": 5670
+    },
+    {
+      "epoch": 1.537628586897672,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.000245947914769623,
+      "loss": 0.0932,
+      "step": 5680
+    },
+    {
+      "epoch": 1.540335679480238,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0002454926242942998,
+      "loss": 0.1075,
+      "step": 5690
+    },
+    {
+      "epoch": 1.5430427720628046,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.0002450373338189765,
+      "loss": 0.0948,
+      "step": 5700
+    },
+    {
+      "epoch": 1.5430427720628046,
+      "eval_loss": 0.11705286055803299,
+      "eval_runtime": 39.936,
+      "eval_samples_per_second": 12.52,
+      "eval_steps_per_second": 0.801,
+      "step": 5700
+    },
+    {
+      "epoch": 1.545749864645371,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.00024458204334365324,
+      "loss": 0.1026,
+      "step": 5710
+    },
+    {
+      "epoch": 1.5484569572279372,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00024412675286833,
+      "loss": 0.0738,
+      "step": 5720
+    },
+    {
+      "epoch": 1.5511640498105035,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.00024367146239300674,
+      "loss": 0.0988,
+      "step": 5730
+    },
+    {
+      "epoch": 1.5538711423930698,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0002432161719176835,
+      "loss": 0.116,
+      "step": 5740
+    },
+    {
+      "epoch": 1.5565782349756363,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00024276088144236022,
+      "loss": 0.0986,
+      "step": 5750
+    },
+    {
+      "epoch": 1.5592853275582024,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00024230559096703698,
+      "loss": 0.1294,
+      "step": 5760
+    },
+    {
+      "epoch": 1.561992420140769,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00024185030049171373,
+      "loss": 0.1016,
+      "step": 5770
+    },
+    {
+      "epoch": 1.5646995127233352,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00024139501001639048,
+      "loss": 0.118,
+      "step": 5780
+    },
+    {
+      "epoch": 1.5674066053059015,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00024093971954106718,
+      "loss": 0.1331,
+      "step": 5790
+    },
+    {
+      "epoch": 1.5701136978884678,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00024048442906574394,
+      "loss": 0.1233,
+      "step": 5800
+    },
+    {
+      "epoch": 1.5701136978884678,
+      "eval_loss": 0.11803543567657471,
+      "eval_runtime": 37.22,
+      "eval_samples_per_second": 13.434,
+      "eval_steps_per_second": 0.86,
+      "step": 5800
+    },
+    {
+      "epoch": 1.572820790471034,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0002400291385904207,
+      "loss": 0.0962,
+      "step": 5810
+    },
+    {
+      "epoch": 1.5755278830536006,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00023957384811509744,
+      "loss": 0.1116,
+      "step": 5820
+    },
+    {
+      "epoch": 1.5782349756361667,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.0002391185576397742,
+      "loss": 0.1198,
+      "step": 5830
+    },
+    {
+      "epoch": 1.5809420682187332,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00023866326716445092,
+      "loss": 0.1113,
+      "step": 5840
+    },
+    {
+      "epoch": 1.5836491608012992,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00023820797668912768,
+      "loss": 0.0988,
+      "step": 5850
+    },
+    {
+      "epoch": 1.5863562533838658,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0002377526862138044,
+      "loss": 0.1199,
+      "step": 5860
+    },
+    {
+      "epoch": 1.589063345966432,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.00023729739573848116,
+      "loss": 0.0896,
+      "step": 5870
+    },
+    {
+      "epoch": 1.5917704385489984,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00023684210526315788,
+      "loss": 0.1092,
+      "step": 5880
+    },
+    {
+      "epoch": 1.5944775311315647,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.00023638681478783464,
+      "loss": 0.0751,
+      "step": 5890
+    },
+    {
+      "epoch": 1.597184623714131,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002359315243125114,
+      "loss": 0.1026,
+      "step": 5900
+    },
+    {
+      "epoch": 1.597184623714131,
+      "eval_loss": 0.1152162030339241,
+      "eval_runtime": 38.8315,
+      "eval_samples_per_second": 12.876,
+      "eval_steps_per_second": 0.824,
+      "step": 5900
+    },
+    {
+      "epoch": 1.5998917162966975,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.00023547623383718814,
+      "loss": 0.1131,
+      "step": 5910
+    },
+    {
+      "epoch": 1.6025988088792635,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0002350209433618649,
+      "loss": 0.1205,
+      "step": 5920
+    },
+    {
+      "epoch": 1.60530590146183,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0002345656528865416,
+      "loss": 0.1041,
+      "step": 5930
+    },
+    {
+      "epoch": 1.6080129940443963,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00023411036241121835,
+      "loss": 0.1163,
+      "step": 5940
+    },
+    {
+      "epoch": 1.6107200866269626,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.0002336550719358951,
+      "loss": 0.0879,
+      "step": 5950
+    },
+    {
+      "epoch": 1.613427179209529,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00023319978146057186,
+      "loss": 0.1224,
+      "step": 5960
+    },
+    {
+      "epoch": 1.6161342717920952,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00023274449098524859,
+      "loss": 0.1078,
+      "step": 5970
+    },
+    {
+      "epoch": 1.6188413643746618,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00023228920050992534,
+      "loss": 0.1166,
+      "step": 5980
+    },
+    {
+      "epoch": 1.6215484569572278,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0002318339100346021,
+      "loss": 0.086,
+      "step": 5990
+    },
+    {
+      "epoch": 1.6242555495397943,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00023137861955927882,
+      "loss": 0.1043,
+      "step": 6000
+    },
+    {
+      "epoch": 1.6242555495397943,
+      "eval_loss": 0.11764785647392273,
+      "eval_runtime": 38.4731,
+      "eval_samples_per_second": 12.996,
+      "eval_steps_per_second": 0.832,
+      "step": 6000
+    },
+    {
+      "epoch": 1.6269626421223606,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.00023092332908395557,
+      "loss": 0.0901,
+      "step": 6010
+    },
+    {
+      "epoch": 1.629669734704927,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0002304680386086323,
+      "loss": 0.0803,
+      "step": 6020
+    },
+    {
+      "epoch": 1.6323768272874932,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00023001274813330905,
+      "loss": 0.1129,
+      "step": 6030
+    },
+    {
+      "epoch": 1.6350839198700595,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0002295574576579858,
+      "loss": 0.0876,
+      "step": 6040
+    },
+    {
+      "epoch": 1.637791012452626,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00022910216718266256,
+      "loss": 0.1302,
+      "step": 6050
+    },
+    {
+      "epoch": 1.6404981050351921,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0002286468767073393,
+      "loss": 0.1103,
+      "step": 6060
+    },
+    {
+      "epoch": 1.6432051976177586,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.000228191586232016,
+      "loss": 0.0802,
+      "step": 6070
+    },
+    {
+      "epoch": 1.6459122902003247,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.00022773629575669277,
+      "loss": 0.0909,
+      "step": 6080
+    },
+    {
+      "epoch": 1.6486193827828912,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.00022728100528136952,
+      "loss": 0.1132,
+      "step": 6090
+    },
+    {
+      "epoch": 1.6513264753654575,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00022682571480604627,
+      "loss": 0.1084,
+      "step": 6100
+    },
+    {
+      "epoch": 1.6513264753654575,
+      "eval_loss": 0.11631617695093155,
+      "eval_runtime": 38.9235,
+      "eval_samples_per_second": 12.846,
+      "eval_steps_per_second": 0.822,
+      "step": 6100
+    },
+    {
+      "epoch": 1.6540335679480238,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.000226370424330723,
+      "loss": 0.1098,
+      "step": 6110
+    },
+    {
+      "epoch": 1.65674066053059,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.00022591513385539975,
+      "loss": 0.1091,
+      "step": 6120
+    },
+    {
+      "epoch": 1.6594477531131564,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0002254598433800765,
+      "loss": 0.0971,
+      "step": 6130
+    },
+    {
+      "epoch": 1.662154845695723,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00022500455290475323,
+      "loss": 0.0926,
+      "step": 6140
+    },
+    {
+      "epoch": 1.664861938278289,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00022454926242942996,
+      "loss": 0.1007,
+      "step": 6150
+    },
+    {
+      "epoch": 1.6675690308608555,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.00022409397195410671,
+      "loss": 0.1084,
+      "step": 6160
+    },
+    {
+      "epoch": 1.6702761234434218,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.00022363868147878347,
+      "loss": 0.1088,
+      "step": 6170
+    },
+    {
+      "epoch": 1.672983216025988,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.00022318339100346022,
+      "loss": 0.0886,
+      "step": 6180
+    },
+    {
+      "epoch": 1.6756903086085544,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00022272810052813698,
+      "loss": 0.1062,
+      "step": 6190
+    },
+    {
+      "epoch": 1.6783974011911207,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0002222728100528137,
+      "loss": 0.0901,
+      "step": 6200
+    },
+    {
+      "epoch": 1.6783974011911207,
+      "eval_loss": 0.11608723551034927,
+      "eval_runtime": 37.4319,
+      "eval_samples_per_second": 13.358,
+      "eval_steps_per_second": 0.855,
+      "step": 6200
+    },
+    {
+      "epoch": 1.6811044937736872,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.00022181751957749043,
+      "loss": 0.0942,
+      "step": 6210
+    },
+    {
+      "epoch": 1.6838115863562533,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00022136222910216718,
+      "loss": 0.0826,
+      "step": 6220
+    },
+    {
+      "epoch": 1.6865186789388198,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.00022090693862684394,
+      "loss": 0.1088,
+      "step": 6230
+    },
+    {
+      "epoch": 1.6892257715213859,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00022045164815152066,
+      "loss": 0.115,
+      "step": 6240
+    },
+    {
+      "epoch": 1.6919328641039524,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00021999635767619742,
+      "loss": 0.0821,
+      "step": 6250
+    },
+    {
+      "epoch": 1.6946399566865187,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00021954106720087417,
+      "loss": 0.0991,
+      "step": 6260
+    },
+    {
+      "epoch": 1.697347049269085,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00021908577672555092,
+      "loss": 0.1201,
+      "step": 6270
+    },
+    {
+      "epoch": 1.7000541418516515,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00021863048625022765,
+      "loss": 0.1007,
+      "step": 6280
+    },
+    {
+      "epoch": 1.7027612344342176,
+      "grad_norm": 0.04150390625,
+      "learning_rate": 0.00021817519577490438,
+      "loss": 0.0985,
+      "step": 6290
+    },
+    {
+      "epoch": 1.705468327016784,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00021771990529958113,
+      "loss": 0.0929,
+      "step": 6300
+    },
+    {
+      "epoch": 1.705468327016784,
+      "eval_loss": 0.11211559921503067,
+      "eval_runtime": 38.6493,
+      "eval_samples_per_second": 12.937,
+      "eval_steps_per_second": 0.828,
+      "step": 6300
+    },
+    {
+      "epoch": 1.7081754195993502,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00021726461482425788,
+      "loss": 0.0933,
+      "step": 6310
+    },
+    {
+      "epoch": 1.7108825121819167,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00021680932434893464,
+      "loss": 0.1198,
+      "step": 6320
+    },
+    {
+      "epoch": 1.713589604764483,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.00021635403387361136,
+      "loss": 0.1113,
+      "step": 6330
+    },
+    {
+      "epoch": 1.7162966973470493,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00021589874339828812,
+      "loss": 0.0899,
+      "step": 6340
+    },
+    {
+      "epoch": 1.7190037899296156,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00021544345292296484,
+      "loss": 0.0959,
+      "step": 6350
+    },
+    {
+      "epoch": 1.7217108825121819,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0002149881624476416,
+      "loss": 0.0943,
+      "step": 6360
+    },
+    {
+      "epoch": 1.7244179750947484,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.00021453287197231835,
+      "loss": 0.1053,
+      "step": 6370
+    },
+    {
+      "epoch": 1.7271250676773144,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.00021407758149699508,
+      "loss": 0.1009,
+      "step": 6380
+    },
+    {
+      "epoch": 1.729832160259881,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.00021362229102167183,
+      "loss": 0.0786,
+      "step": 6390
+    },
+    {
+      "epoch": 1.7325392528424473,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.00021316700054634858,
+      "loss": 0.1006,
+      "step": 6400
+    },
+    {
+      "epoch": 1.7325392528424473,
+      "eval_loss": 0.114571712911129,
+      "eval_runtime": 39.6409,
+      "eval_samples_per_second": 12.613,
+      "eval_steps_per_second": 0.807,
+      "step": 6400
+    },
+    {
+      "epoch": 1.7352463454250135,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.00021271171007102534,
+      "loss": 0.1241,
+      "step": 6410
+    },
+    {
+      "epoch": 1.7379534380075798,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00021225641959570204,
+      "loss": 0.1104,
+      "step": 6420
+    },
+    {
+      "epoch": 1.7406605305901461,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0002118011291203788,
+      "loss": 0.0793,
+      "step": 6430
+    },
+    {
+      "epoch": 1.7433676231727127,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00021134583864505554,
+      "loss": 0.0885,
+      "step": 6440
+    },
+    {
+      "epoch": 1.7460747157552787,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0002108905481697323,
+      "loss": 0.0945,
+      "step": 6450
+    },
+    {
+      "epoch": 1.7487818083378452,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00021043525769440905,
+      "loss": 0.0576,
+      "step": 6460
+    },
+    {
+      "epoch": 1.7514889009204113,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00020997996721908578,
+      "loss": 0.1148,
+      "step": 6470
+    },
+    {
+      "epoch": 1.7541959935029778,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00020952467674376253,
+      "loss": 0.0779,
+      "step": 6480
+    },
+    {
+      "epoch": 1.7569030860855441,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.00020906938626843926,
+      "loss": 0.1052,
+      "step": 6490
+    },
+    {
+      "epoch": 1.7596101786681104,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.000208614095793116,
+      "loss": 0.0875,
+      "step": 6500
+    },
+    {
+      "epoch": 1.7596101786681104,
+      "eval_loss": 0.11314798146486282,
+      "eval_runtime": 39.9256,
+      "eval_samples_per_second": 12.523,
+      "eval_steps_per_second": 0.801,
+      "step": 6500
+    },
+    {
+      "epoch": 1.7623172712506767,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.00020815880531779274,
+      "loss": 0.0939,
+      "step": 6510
+    },
+    {
+      "epoch": 1.765024363833243,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0002077035148424695,
+      "loss": 0.1089,
+      "step": 6520
+    },
+    {
+      "epoch": 1.7677314564158095,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.00020724822436714625,
+      "loss": 0.101,
+      "step": 6530
+    },
+    {
+      "epoch": 1.7704385489983756,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.000206792933891823,
+      "loss": 0.0958,
+      "step": 6540
+    },
+    {
+      "epoch": 1.7731456415809421,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00020633764341649975,
+      "loss": 0.1273,
+      "step": 6550
+    },
+    {
+      "epoch": 1.7758527341635084,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.00020588235294117645,
+      "loss": 0.0887,
+      "step": 6560
+    },
+    {
+      "epoch": 1.7785598267460747,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.0002054270624658532,
+      "loss": 0.1078,
+      "step": 6570
+    },
+    {
+      "epoch": 1.781266919328641,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00020497177199052996,
+      "loss": 0.1294,
+      "step": 6580
+    },
+    {
+      "epoch": 1.7839740119112073,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0002045164815152067,
+      "loss": 0.1114,
+      "step": 6590
+    },
+    {
+      "epoch": 1.7866811044937738,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.00020406119103988344,
+      "loss": 0.0863,
+      "step": 6600
+    },
+    {
+      "epoch": 1.7866811044937738,
+      "eval_loss": 0.11271684616804123,
+      "eval_runtime": 39.8605,
+      "eval_samples_per_second": 12.544,
+      "eval_steps_per_second": 0.803,
+      "step": 6600
+    },
+    {
+      "epoch": 1.78938819707634,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.0002036059005645602,
+      "loss": 0.1203,
+      "step": 6610
+    },
+    {
+      "epoch": 1.7920952896589064,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00020315061008923695,
+      "loss": 0.0963,
+      "step": 6620
+    },
+    {
+      "epoch": 1.7948023822414727,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.00020269531961391367,
+      "loss": 0.131,
+      "step": 6630
+    },
+    {
+      "epoch": 1.797509474824039,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.00020224002913859043,
+      "loss": 0.098,
+      "step": 6640
+    },
+    {
+      "epoch": 1.8002165674066053,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00020178473866326715,
+      "loss": 0.1427,
+      "step": 6650
+    },
+    {
+      "epoch": 1.8029236599891716,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0002013294481879439,
+      "loss": 0.0952,
+      "step": 6660
+    },
+    {
+      "epoch": 1.805630752571738,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.00020087415771262066,
+      "loss": 0.0862,
+      "step": 6670
+    },
+    {
+      "epoch": 1.8083378451543042,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.00020041886723729741,
+      "loss": 0.0947,
+      "step": 6680
+    },
+    {
+      "epoch": 1.8110449377368707,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00019996357676197414,
+      "loss": 0.1191,
+      "step": 6690
+    },
+    {
+      "epoch": 1.8137520303194368,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.00019950828628665087,
+      "loss": 0.107,
+      "step": 6700
+    },
+    {
+      "epoch": 1.8137520303194368,
+      "eval_loss": 0.11586461216211319,
+      "eval_runtime": 37.8187,
+      "eval_samples_per_second": 13.221,
+      "eval_steps_per_second": 0.846,
+      "step": 6700
+    },
+    {
+      "epoch": 1.8164591229020033,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00019905299581132762,
+      "loss": 0.0974,
+      "step": 6710
+    },
+    {
+      "epoch": 1.8191662154845696,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00019859770533600437,
+      "loss": 0.1047,
+      "step": 6720
+    },
+    {
+      "epoch": 1.8218733080671359,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.00019814241486068113,
+      "loss": 0.1047,
+      "step": 6730
+    },
+    {
+      "epoch": 1.8245804006497022,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.00019768712438535786,
+      "loss": 0.0842,
+      "step": 6740
+    },
+    {
+      "epoch": 1.8272874932322685,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.0001972318339100346,
+      "loss": 0.0691,
+      "step": 6750
+    },
+    {
+      "epoch": 1.829994585814835,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.00019677654343471136,
+      "loss": 0.0659,
+      "step": 6760
+    },
+    {
+      "epoch": 1.832701678397401,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.00019632125295938812,
+      "loss": 0.091,
+      "step": 6770
+    },
+    {
+      "epoch": 1.8354087709799676,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00019586596248406482,
+      "loss": 0.0737,
+      "step": 6780
+    },
+    {
+      "epoch": 1.8381158635625339,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.00019541067200874157,
+      "loss": 0.0682,
+      "step": 6790
+    },
+    {
+      "epoch": 1.8408229561451002,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.00019495538153341832,
+      "loss": 0.1117,
+      "step": 6800
+    },
+    {
+      "epoch": 1.8408229561451002,
+      "eval_loss": 0.11802293360233307,
+      "eval_runtime": 37.8456,
+      "eval_samples_per_second": 13.212,
+      "eval_steps_per_second": 0.846,
+      "step": 6800
+    },
+    {
+      "epoch": 1.8435300487276665,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.00019450009105809508,
+      "loss": 0.0857,
+      "step": 6810
+    },
+    {
+      "epoch": 1.8462371413102328,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.00019404480058277183,
+      "loss": 0.0934,
+      "step": 6820
+    },
+    {
+      "epoch": 1.8489442338927993,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.00019358951010744856,
+      "loss": 0.0846,
+      "step": 6830
+    },
+    {
+      "epoch": 1.8516513264753653,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0001931342196321253,
+      "loss": 0.1062,
+      "step": 6840
+    },
+    {
+      "epoch": 1.8543584190579319,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00019267892915680204,
+      "loss": 0.1304,
+      "step": 6850
+    },
+    {
+      "epoch": 1.8570655116404982,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0001922236386814788,
+      "loss": 0.1257,
+      "step": 6860
+    },
+    {
+      "epoch": 1.8597726042230645,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00019176834820615552,
+      "loss": 0.1089,
+      "step": 6870
+    },
+    {
+      "epoch": 1.8624796968056307,
+      "grad_norm": 0.044677734375,
+      "learning_rate": 0.00019131305773083227,
+      "loss": 0.0975,
+      "step": 6880
+    },
+    {
+      "epoch": 1.865186789388197,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.00019085776725550902,
+      "loss": 0.0867,
+      "step": 6890
+    },
+    {
+      "epoch": 1.8678938819707636,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019040247678018578,
+      "loss": 0.1158,
+      "step": 6900
+    },
+    {
+      "epoch": 1.8678938819707636,
+      "eval_loss": 0.11736615002155304,
+      "eval_runtime": 40.0892,
+      "eval_samples_per_second": 12.472,
+      "eval_steps_per_second": 0.798,
+      "step": 6900
+    },
+    {
+      "epoch": 1.8706009745533296,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00018994718630486253,
+      "loss": 0.0951,
+      "step": 6910
+    },
+    {
+      "epoch": 1.8733080671358961,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00018949189582953923,
+      "loss": 0.0823,
+      "step": 6920
+    },
+    {
+      "epoch": 1.8760151597184622,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00018903660535421598,
+      "loss": 0.0952,
+      "step": 6930
+    },
+    {
+      "epoch": 1.8787222523010287,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00018858131487889274,
+      "loss": 0.0927,
+      "step": 6940
+    },
+    {
+      "epoch": 1.881429344883595,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0001881260244035695,
+      "loss": 0.1272,
+      "step": 6950
+    },
+    {
+      "epoch": 1.8841364374661613,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.00018767073392824622,
+      "loss": 0.1149,
+      "step": 6960
+    },
+    {
+      "epoch": 1.8868435300487276,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.00018721544345292297,
+      "loss": 0.1241,
+      "step": 6970
+    },
+    {
+      "epoch": 1.889550622631294,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.00018676015297759973,
+      "loss": 0.0877,
+      "step": 6980
+    },
+    {
+      "epoch": 1.8922577152138604,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.00018630486250227645,
+      "loss": 0.0987,
+      "step": 6990
+    },
+    {
+      "epoch": 1.8949648077964265,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0001858495720269532,
+      "loss": 0.0896,
+      "step": 7000
+    },
+    {
+      "epoch": 1.8949648077964265,
+      "eval_loss": 0.11874046176671982,
+      "eval_runtime": 38.0336,
+      "eval_samples_per_second": 13.146,
+      "eval_steps_per_second": 0.841,
+      "step": 7000
+    },
+    {
+      "epoch": 1.897671900378993,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.00018539428155162993,
+      "loss": 0.0852,
+      "step": 7010
+    },
+    {
+      "epoch": 1.9003789929615593,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00018493899107630669,
+      "loss": 0.0763,
+      "step": 7020
+    },
+    {
+      "epoch": 1.9030860855441256,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.00018448370060098344,
+      "loss": 0.1101,
+      "step": 7030
+    },
+    {
+      "epoch": 1.905793178126692,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.0001840284101256602,
+      "loss": 0.113,
+      "step": 7040
+    },
+    {
+      "epoch": 1.9085002707092582,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.00018357311965033692,
+      "loss": 0.1065,
+      "step": 7050
+    },
+    {
+      "epoch": 1.9112073632918247,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00018311782917501365,
+      "loss": 0.1083,
+      "step": 7060
+    },
+    {
+      "epoch": 1.9139144558743908,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0001826625386996904,
+      "loss": 0.0848,
+      "step": 7070
+    },
+    {
+      "epoch": 1.9166215484569573,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.00018220724822436715,
+      "loss": 0.112,
+      "step": 7080
+    },
+    {
+      "epoch": 1.9193286410395234,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0001817519577490439,
+      "loss": 0.1068,
+      "step": 7090
+    },
+    {
+      "epoch": 1.92203573362209,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.00018129666727372063,
+      "loss": 0.101,
+      "step": 7100
+    },
+    {
+      "epoch": 1.92203573362209,
+      "eval_loss": 0.11733894050121307,
+      "eval_runtime": 39.2535,
+      "eval_samples_per_second": 12.738,
+      "eval_steps_per_second": 0.815,
+      "step": 7100
+    },
+    {
+      "epoch": 1.9247428262046562,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.0001808413767983974,
+      "loss": 0.082,
+      "step": 7110
+    },
+    {
+      "epoch": 1.9274499187872225,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00018038608632307414,
+      "loss": 0.0724,
+      "step": 7120
+    },
+    {
+      "epoch": 1.930157011369789,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.00017993079584775087,
+      "loss": 0.1028,
+      "step": 7130
+    },
+    {
+      "epoch": 1.932864103952355,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0001794755053724276,
+      "loss": 0.1051,
+      "step": 7140
+    },
+    {
+      "epoch": 1.9355711965349216,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.00017902021489710435,
+      "loss": 0.0853,
+      "step": 7150
+    },
+    {
+      "epoch": 1.9382782891174877,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001785649244217811,
+      "loss": 0.0961,
+      "step": 7160
+    },
+    {
+      "epoch": 1.9409853817000542,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00017810963394645785,
+      "loss": 0.1029,
+      "step": 7170
+    },
+    {
+      "epoch": 1.9436924742826205,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.0001776543434711346,
+      "loss": 0.1002,
+      "step": 7180
+    },
+    {
+      "epoch": 1.9463995668651868,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.00017719905299581133,
+      "loss": 0.1077,
+      "step": 7190
+    },
+    {
+      "epoch": 1.949106659447753,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.00017674376252048806,
+      "loss": 0.0833,
+      "step": 7200
+    },
+    {
+      "epoch": 1.949106659447753,
+      "eval_loss": 0.11724207550287247,
+      "eval_runtime": 40.2923,
+      "eval_samples_per_second": 12.409,
+      "eval_steps_per_second": 0.794,
+      "step": 7200
+    },
+    {
+      "epoch": 1.9518137520303194,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.00017628847204516481,
+      "loss": 0.093,
+      "step": 7210
+    },
+    {
+      "epoch": 1.9545208446128859,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00017583318156984157,
+      "loss": 0.1411,
+      "step": 7220
+    },
+    {
+      "epoch": 1.957227937195452,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001753778910945183,
+      "loss": 0.1111,
+      "step": 7230
+    },
+    {
+      "epoch": 1.9599350297780185,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00017492260061919505,
+      "loss": 0.0973,
+      "step": 7240
+    },
+    {
+      "epoch": 1.9626421223605848,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0001744673101438718,
+      "loss": 0.0927,
+      "step": 7250
+    },
+    {
+      "epoch": 1.965349214943151,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00017401201966854856,
+      "loss": 0.0803,
+      "step": 7260
+    },
+    {
+      "epoch": 1.9680563075257174,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.00017355672919322528,
+      "loss": 0.1199,
+      "step": 7270
+    },
+    {
+      "epoch": 1.9707634001082837,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.000173101438717902,
+      "loss": 0.0934,
+      "step": 7280
+    },
+    {
+      "epoch": 1.9734704926908502,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00017264614824257876,
+      "loss": 0.14,
+      "step": 7290
+    },
+    {
+      "epoch": 1.9761775852734162,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.00017219085776725552,
+      "loss": 0.0883,
+      "step": 7300
+    },
+    {
+      "epoch": 1.9761775852734162,
+      "eval_loss": 0.11611966043710709,
+      "eval_runtime": 38.0167,
+      "eval_samples_per_second": 13.152,
+      "eval_steps_per_second": 0.842,
+      "step": 7300
+    },
+    {
+      "epoch": 1.9788846778559828,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.00017173556729193227,
+      "loss": 0.0851,
+      "step": 7310
+    },
+    {
+      "epoch": 1.9815917704385488,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.000171280276816609,
+      "loss": 0.0915,
+      "step": 7320
+    },
+    {
+      "epoch": 1.9842988630211154,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00017082498634128575,
+      "loss": 0.1046,
+      "step": 7330
+    },
+    {
+      "epoch": 1.9870059556036817,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00017036969586596248,
+      "loss": 0.0847,
+      "step": 7340
+    },
+    {
+      "epoch": 1.989713048186248,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.00016991440539063923,
+      "loss": 0.0995,
+      "step": 7350
+    },
+    {
+      "epoch": 1.9924201407688142,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.00016945911491531598,
+      "loss": 0.1084,
+      "step": 7360
+    },
+    {
+      "epoch": 1.9951272333513805,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.0001690038244399927,
+      "loss": 0.1002,
+      "step": 7370
+    },
+    {
+      "epoch": 1.997834325933947,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00016854853396466946,
+      "loss": 0.0775,
+      "step": 7380
+    },
+    {
+      "epoch": 2.000541418516513,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.00016809324348934622,
+      "loss": 0.1,
+      "step": 7390
+    },
+    {
+      "epoch": 2.0032485110990796,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.00016763795301402297,
+      "loss": 0.0885,
+      "step": 7400
+    },
+    {
+      "epoch": 2.0032485110990796,
+      "eval_loss": 0.11766168475151062,
+      "eval_runtime": 39.2016,
+      "eval_samples_per_second": 12.755,
+      "eval_steps_per_second": 0.816,
+      "step": 7400
+    },
+    {
+      "epoch": 2.0059556036816457,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.00016718266253869967,
+      "loss": 0.0724,
+      "step": 7410
+    },
+    {
+      "epoch": 2.0086626962642122,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00016672737206337642,
+      "loss": 0.0709,
+      "step": 7420
+    },
+    {
+      "epoch": 2.0113697888467788,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00016627208158805318,
+      "loss": 0.1251,
+      "step": 7430
+    },
+    {
+      "epoch": 2.014076881429345,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.00016581679111272993,
+      "loss": 0.073,
+      "step": 7440
+    },
+    {
+      "epoch": 2.0167839740119113,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00016536150063740668,
+      "loss": 0.094,
+      "step": 7450
+    },
+    {
+      "epoch": 2.0194910665944774,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0001649062101620834,
+      "loss": 0.0676,
+      "step": 7460
+    },
+    {
+      "epoch": 2.022198159177044,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00016445091968676016,
+      "loss": 0.1292,
+      "step": 7470
+    },
+    {
+      "epoch": 2.02490525175961,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0001639956292114369,
+      "loss": 0.0929,
+      "step": 7480
+    },
+    {
+      "epoch": 2.0276123443421765,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.00016354033873611364,
+      "loss": 0.1,
+      "step": 7490
+    },
+    {
+      "epoch": 2.030319436924743,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00016308504826079037,
+      "loss": 0.1094,
+      "step": 7500
+    },
+    {
+      "epoch": 2.030319436924743,
+      "eval_loss": 0.11815401911735535,
+      "eval_runtime": 39.042,
+      "eval_samples_per_second": 12.807,
+      "eval_steps_per_second": 0.82,
+      "step": 7500
+    },
+    {
+      "epoch": 2.033026529507309,
+      "grad_norm": 0.18556566536426544,
+      "learning_rate": 0.00016262975778546712,
+      "loss": 0.114,
+      "step": 7510
+    },
+    {
+      "epoch": 2.0357336220898756,
+      "grad_norm": 0.203460693359375,
+      "learning_rate": 0.00016217446731014388,
+      "loss": 0.0862,
+      "step": 7520
+    },
+    {
+      "epoch": 2.0384407146724417,
+      "grad_norm": 0.2872123420238495,
+      "learning_rate": 0.00016171917683482063,
+      "loss": 0.1207,
+      "step": 7530
+    },
+    {
+      "epoch": 2.041147807255008,
+      "grad_norm": 0.3013206422328949,
+      "learning_rate": 0.00016126388635949739,
+      "loss": 0.1044,
+      "step": 7540
+    },
+    {
+      "epoch": 2.0438548998375743,
+      "grad_norm": 0.377379447221756,
+      "learning_rate": 0.00016080859588417409,
+      "loss": 0.1324,
+      "step": 7550
+    },
+    {
+      "epoch": 2.046561992420141,
+      "grad_norm": 0.10991441458463669,
+      "learning_rate": 0.00016035330540885084,
+      "loss": 0.0851,
+      "step": 7560
+    },
+    {
+      "epoch": 2.049269085002707,
+      "grad_norm": 0.4809432923793793,
+      "learning_rate": 0.0001598980149335276,
+      "loss": 0.1081,
+      "step": 7570
+    },
+    {
+      "epoch": 2.0519761775852734,
+      "grad_norm": 0.12087661027908325,
+      "learning_rate": 0.00015944272445820435,
+      "loss": 0.0859,
+      "step": 7580
+    },
+    {
+      "epoch": 2.05468327016784,
+      "grad_norm": 0.38357388973236084,
+      "learning_rate": 0.00015898743398288107,
+      "loss": 0.0946,
+      "step": 7590
+    },
+    {
+      "epoch": 2.057390362750406,
+      "grad_norm": 0.26211583614349365,
+      "learning_rate": 0.00015853214350755783,
+      "loss": 0.0893,
+      "step": 7600
+    },
+    {
+      "epoch": 2.057390362750406,
+      "eval_loss": 0.12073516845703125,
+      "eval_runtime": 46.4442,
+      "eval_samples_per_second": 10.766,
+      "eval_steps_per_second": 0.689,
+      "step": 7600
+    },
+    {
+      "epoch": 2.0600974553329725,
+      "grad_norm": 0.22928234934806824,
+      "learning_rate": 0.00015807685303223458,
+      "loss": 0.116,
+      "step": 7610
+    },
+    {
+      "epoch": 2.0628045479155386,
+      "grad_norm": 0.30186566710472107,
+      "learning_rate": 0.0001576215625569113,
+      "loss": 0.0963,
+      "step": 7620
+    },
+    {
+      "epoch": 2.065511640498105,
+      "grad_norm": 0.14704805612564087,
+      "learning_rate": 0.00015716627208158806,
+      "loss": 0.0848,
+      "step": 7630
+    },
+    {
+      "epoch": 2.068218733080671,
+      "grad_norm": 0.17083260416984558,
+      "learning_rate": 0.0001567109816062648,
+      "loss": 0.1225,
+      "step": 7640
+    },
+    {
+      "epoch": 2.0709258256632377,
+      "grad_norm": 0.42746874690055847,
+      "learning_rate": 0.00015625569113094154,
+      "loss": 0.1049,
+      "step": 7650
+    },
+    {
+      "epoch": 2.073632918245804,
+      "grad_norm": 0.3558850586414337,
+      "learning_rate": 0.0001558004006556183,
+      "loss": 0.1165,
+      "step": 7660
+    },
+    {
+      "epoch": 2.0763400108283703,
+      "grad_norm": 0.24918563663959503,
+      "learning_rate": 0.00015534511018029505,
+      "loss": 0.1008,
+      "step": 7670
+    },
+    {
+      "epoch": 2.079047103410937,
+      "grad_norm": 0.28519049286842346,
+      "learning_rate": 0.00015488981970497177,
+      "loss": 0.1068,
+      "step": 7680
+    },
+    {
+      "epoch": 2.081754195993503,
+      "grad_norm": 0.16140486299991608,
+      "learning_rate": 0.0001544345292296485,
+      "loss": 0.0974,
+      "step": 7690
+    },
+    {
+      "epoch": 2.0844612885760694,
+      "grad_norm": 0.4082144498825073,
+      "learning_rate": 0.00015397923875432525,
+      "loss": 0.0931,
+      "step": 7700
+    },
+    {
+      "epoch": 2.0844612885760694,
+      "eval_loss": 0.118402861058712,
+      "eval_runtime": 41.3665,
+      "eval_samples_per_second": 12.087,
+      "eval_steps_per_second": 0.774,
+      "step": 7700
+    },
+    {
+      "epoch": 2.0871683811586355,
+      "grad_norm": 0.11690416932106018,
+      "learning_rate": 0.000153523948279002,
+      "loss": 0.0659,
+      "step": 7710
+    },
+    {
+      "epoch": 2.089875473741202,
+      "grad_norm": 0.2224791944026947,
+      "learning_rate": 0.00015306865780367876,
+      "loss": 0.0878,
+      "step": 7720
+    },
+    {
+      "epoch": 2.0925825663237685,
+      "grad_norm": 0.48507657647132874,
+      "learning_rate": 0.0001526133673283555,
+      "loss": 0.1014,
+      "step": 7730
+    },
+    {
+      "epoch": 2.0952896589063346,
+      "grad_norm": 0.13743431866168976,
+      "learning_rate": 0.00015215807685303224,
+      "loss": 0.0803,
+      "step": 7740
+    },
+    {
+      "epoch": 2.097996751488901,
+      "grad_norm": 0.09833896160125732,
+      "learning_rate": 0.000151702786377709,
+      "loss": 0.1008,
+      "step": 7750
+    },
+    {
+      "epoch": 2.100703844071467,
+      "grad_norm": 0.5322971343994141,
+      "learning_rate": 0.00015124749590238572,
+      "loss": 0.0926,
+      "step": 7760
+    },
+    {
+      "epoch": 2.1034109366540337,
+      "grad_norm": 0.24878880381584167,
+      "learning_rate": 0.00015079220542706245,
+      "loss": 0.0986,
+      "step": 7770
+    },
+    {
+      "epoch": 2.1061180292365997,
+      "grad_norm": 0.3437240421772003,
+      "learning_rate": 0.0001503369149517392,
+      "loss": 0.1103,
+      "step": 7780
+    },
+    {
+      "epoch": 2.1088251218191663,
+      "grad_norm": 0.19925567507743835,
+      "learning_rate": 0.00014988162447641596,
+      "loss": 0.0948,
+      "step": 7790
+    },
+    {
+      "epoch": 2.1115322144017323,
+      "grad_norm": 0.22366894781589508,
+      "learning_rate": 0.0001494263340010927,
+      "loss": 0.1182,
+      "step": 7800
+    },
+    {
+      "epoch": 2.1115322144017323,
+      "eval_loss": 0.11602531373500824,
+      "eval_runtime": 40.5737,
+      "eval_samples_per_second": 12.323,
+      "eval_steps_per_second": 0.789,
+      "step": 7800
+    },
+    {
+      "epoch": 2.114239306984299,
+      "grad_norm": 0.2947825789451599,
+      "learning_rate": 0.00014897104352576946,
+      "loss": 0.105,
+      "step": 7810
+    },
+    {
+      "epoch": 2.1169463995668654,
+      "grad_norm": 0.3085196614265442,
+      "learning_rate": 0.0001485157530504462,
+      "loss": 0.0998,
+      "step": 7820
+    },
+    {
+      "epoch": 2.1196534921494314,
+      "grad_norm": 0.2723862826824188,
+      "learning_rate": 0.00014806046257512292,
+      "loss": 0.1043,
+      "step": 7830
+    },
+    {
+      "epoch": 2.122360584731998,
+      "grad_norm": 0.27626705169677734,
+      "learning_rate": 0.00014760517209979967,
+      "loss": 0.0904,
+      "step": 7840
+    },
+    {
+      "epoch": 2.125067677314564,
+      "grad_norm": 0.3315790891647339,
+      "learning_rate": 0.00014714988162447642,
+      "loss": 0.0899,
+      "step": 7850
+    },
+    {
+      "epoch": 2.1277747698971305,
+      "grad_norm": 0.3768678903579712,
+      "learning_rate": 0.00014669459114915315,
+      "loss": 0.1176,
+      "step": 7860
+    },
+    {
+      "epoch": 2.1304818624796966,
+      "grad_norm": 0.1696341633796692,
+      "learning_rate": 0.0001462393006738299,
+      "loss": 0.0698,
+      "step": 7870
+    },
+    {
+      "epoch": 2.133188955062263,
+      "grad_norm": 0.15817686915397644,
+      "learning_rate": 0.00014578401019850666,
+      "loss": 0.0875,
+      "step": 7880
+    },
+    {
+      "epoch": 2.1358960476448297,
+      "grad_norm": 0.12926264107227325,
+      "learning_rate": 0.0001453287197231834,
+      "loss": 0.09,
+      "step": 7890
+    },
+    {
+      "epoch": 2.1386031402273957,
+      "grad_norm": 0.18673348426818848,
+      "learning_rate": 0.00014487342924786014,
+      "loss": 0.1062,
+      "step": 7900
+    },
+    {
+      "epoch": 2.1386031402273957,
+      "eval_loss": 0.1177026778459549,
+      "eval_runtime": 41.823,
+      "eval_samples_per_second": 11.955,
+      "eval_steps_per_second": 0.765,
+      "step": 7900
+    },
+    {
+      "epoch": 2.1413102328099622,
+      "grad_norm": 0.16135463118553162,
+      "learning_rate": 0.00014441813877253686,
+      "loss": 0.1048,
+      "step": 7910
+    },
+    {
+      "epoch": 2.1440173253925283,
+      "grad_norm": 0.19764307141304016,
+      "learning_rate": 0.00014396284829721362,
+      "loss": 0.0868,
+      "step": 7920
+    },
+    {
+      "epoch": 2.146724417975095,
+      "grad_norm": 0.21126599609851837,
+      "learning_rate": 0.00014350755782189037,
+      "loss": 0.104,
+      "step": 7930
+    },
+    {
+      "epoch": 2.149431510557661,
+      "grad_norm": 0.279085636138916,
+      "learning_rate": 0.00014305226734656712,
+      "loss": 0.0664,
+      "step": 7940
+    },
+    {
+      "epoch": 2.1521386031402274,
+      "grad_norm": 0.12439724802970886,
+      "learning_rate": 0.00014259697687124385,
+      "loss": 0.1273,
+      "step": 7950
+    },
+    {
+      "epoch": 2.154845695722794,
+      "grad_norm": 0.23388315737247467,
+      "learning_rate": 0.0001421416863959206,
+      "loss": 0.0838,
+      "step": 7960
+    },
+    {
+      "epoch": 2.15755278830536,
+      "grad_norm": 0.36762672662734985,
+      "learning_rate": 0.00014168639592059733,
+      "loss": 0.0764,
+      "step": 7970
+    },
+    {
+      "epoch": 2.1602598808879265,
+      "grad_norm": 0.39345329999923706,
+      "learning_rate": 0.00014123110544527408,
+      "loss": 0.0908,
+      "step": 7980
+    },
+    {
+      "epoch": 2.1629669734704926,
+      "grad_norm": 0.4398602545261383,
+      "learning_rate": 0.00014077581496995084,
+      "loss": 0.1017,
+      "step": 7990
+    },
+    {
+      "epoch": 2.165674066053059,
+      "grad_norm": 0.06859143823385239,
+      "learning_rate": 0.00014032052449462756,
+      "loss": 0.1048,
+      "step": 8000
+    },
+    {
+      "epoch": 2.165674066053059,
+      "eval_loss": 0.1204909160733223,
+      "eval_runtime": 40.7006,
+      "eval_samples_per_second": 12.285,
+      "eval_steps_per_second": 0.786,
+      "step": 8000
+    },
+    {
+      "epoch": 2.168381158635625,
+      "grad_norm": 0.09467579424381256,
+      "learning_rate": 0.00013986523401930432,
+      "loss": 0.0925,
+      "step": 8010
+    },
+    {
+      "epoch": 2.1710882512181917,
+      "grad_norm": 0.09981294721364975,
+      "learning_rate": 0.00013940994354398107,
+      "loss": 0.091,
+      "step": 8020
+    },
+    {
+      "epoch": 2.173795343800758,
+      "grad_norm": 0.11407893151044846,
+      "learning_rate": 0.00013895465306865783,
+      "loss": 0.0766,
+      "step": 8030
+    },
+    {
+      "epoch": 2.1765024363833243,
+      "grad_norm": 0.37597501277923584,
+      "learning_rate": 0.00013849936259333452,
+      "loss": 0.0888,
+      "step": 8040
+    },
+    {
+      "epoch": 2.179209528965891,
+      "grad_norm": 0.1445028930902481,
+      "learning_rate": 0.00013804407211801128,
+      "loss": 0.0856,
+      "step": 8050
+    },
+    {
+      "epoch": 2.181916621548457,
+      "grad_norm": 0.2679959833621979,
+      "learning_rate": 0.00013758878164268803,
+      "loss": 0.0818,
+      "step": 8060
+    },
+    {
+      "epoch": 2.1846237141310234,
+      "grad_norm": 0.5526220798492432,
+      "learning_rate": 0.00013713349116736479,
+      "loss": 0.0947,
+      "step": 8070
+    },
+    {
+      "epoch": 2.1873308067135895,
+      "grad_norm": 0.059446126222610474,
+      "learning_rate": 0.00013667820069204154,
+      "loss": 0.077,
+      "step": 8080
+    },
+    {
+      "epoch": 2.190037899296156,
+      "grad_norm": 0.15422673523426056,
+      "learning_rate": 0.00013622291021671827,
+      "loss": 0.0907,
+      "step": 8090
+    },
+    {
+      "epoch": 2.192744991878722,
+      "grad_norm": 0.30202794075012207,
+      "learning_rate": 0.00013576761974139502,
+      "loss": 0.0948,
+      "step": 8100
+    },
+    {
+      "epoch": 2.192744991878722,
+      "eval_loss": 0.12145506590604782,
+      "eval_runtime": 40.459,
+      "eval_samples_per_second": 12.358,
+      "eval_steps_per_second": 0.791,
+      "step": 8100
+    },
+    {
+      "epoch": 2.1954520844612886,
+      "grad_norm": 0.2053290456533432,
+      "learning_rate": 0.00013531232926607177,
+      "loss": 0.098,
+      "step": 8110
+    },
+    {
+      "epoch": 2.198159177043855,
+      "grad_norm": 0.2384454607963562,
+      "learning_rate": 0.0001348570387907485,
+      "loss": 0.1054,
+      "step": 8120
+    },
+    {
+      "epoch": 2.200866269626421,
+      "grad_norm": 0.294878751039505,
+      "learning_rate": 0.00013440174831542523,
+      "loss": 0.0816,
+      "step": 8130
+    },
+    {
+      "epoch": 2.2035733622089877,
+      "grad_norm": 0.13888828456401825,
+      "learning_rate": 0.00013394645784010198,
+      "loss": 0.1259,
+      "step": 8140
+    },
+    {
+      "epoch": 2.2062804547915538,
+      "grad_norm": 0.28140896558761597,
+      "learning_rate": 0.00013349116736477873,
+      "loss": 0.0927,
+      "step": 8150
+    },
+    {
+      "epoch": 2.2089875473741203,
+      "grad_norm": 0.5658301711082458,
+      "learning_rate": 0.0001330358768894555,
+      "loss": 0.0921,
+      "step": 8160
+    },
+    {
+      "epoch": 2.2116946399566864,
+      "grad_norm": 0.16429400444030762,
+      "learning_rate": 0.00013258058641413224,
+      "loss": 0.0741,
+      "step": 8170
+    },
+    {
+      "epoch": 2.214401732539253,
+      "grad_norm": 0.3106441795825958,
+      "learning_rate": 0.00013212529593880897,
+      "loss": 0.1015,
+      "step": 8180
+    },
+    {
+      "epoch": 2.2171088251218194,
+      "grad_norm": 0.32946762442588806,
+      "learning_rate": 0.0001316700054634857,
+      "loss": 0.1171,
+      "step": 8190
+    },
+    {
+      "epoch": 2.2198159177043855,
+      "grad_norm": 0.39804428815841675,
+      "learning_rate": 0.00013121471498816245,
+      "loss": 0.1027,
+      "step": 8200
+    },
+    {
+      "epoch": 2.2198159177043855,
+      "eval_loss": 0.11924636363983154,
+      "eval_runtime": 43.4596,
+      "eval_samples_per_second": 11.505,
+      "eval_steps_per_second": 0.736,
+      "step": 8200
+    },
+    {
+      "epoch": 2.222523010286952,
+      "grad_norm": 0.4865691363811493,
+      "learning_rate": 0.0001307594245128392,
+      "loss": 0.0931,
+      "step": 8210
+    },
+    {
+      "epoch": 2.225230102869518,
+      "grad_norm": 0.25576359033584595,
+      "learning_rate": 0.00013030413403751593,
+      "loss": 0.1292,
+      "step": 8220
+    },
+    {
+      "epoch": 2.2279371954520846,
+      "grad_norm": 0.21919693052768707,
+      "learning_rate": 0.00012984884356219268,
+      "loss": 0.0951,
+      "step": 8230
+    },
+    {
+      "epoch": 2.2306442880346506,
+      "grad_norm": 0.3063872456550598,
+      "learning_rate": 0.00012939355308686943,
+      "loss": 0.1107,
+      "step": 8240
+    },
+    {
+      "epoch": 2.233351380617217,
+      "grad_norm": 0.4055786728858948,
+      "learning_rate": 0.0001289382626115462,
+      "loss": 0.1015,
+      "step": 8250
+    },
+    {
+      "epoch": 2.2360584731997832,
+      "grad_norm": 0.17080678045749664,
+      "learning_rate": 0.00012848297213622291,
+      "loss": 0.1126,
+      "step": 8260
+    },
+    {
+      "epoch": 2.2387655657823498,
+      "grad_norm": 0.08725818991661072,
+      "learning_rate": 0.00012802768166089964,
+      "loss": 0.0907,
+      "step": 8270
+    },
+    {
+      "epoch": 2.2414726583649163,
+      "grad_norm": 0.07784614711999893,
+      "learning_rate": 0.0001275723911855764,
+      "loss": 0.0817,
+      "step": 8280
+    },
+    {
+      "epoch": 2.2441797509474823,
+      "grad_norm": 0.13232044875621796,
+      "learning_rate": 0.00012711710071025315,
+      "loss": 0.0754,
+      "step": 8290
+    },
+    {
+      "epoch": 2.246886843530049,
+      "grad_norm": 0.27317237854003906,
+      "learning_rate": 0.0001266618102349299,
+      "loss": 0.0868,
+      "step": 8300
+    },
+    {
+      "epoch": 2.246886843530049,
+      "eval_loss": 0.1212162896990776,
+      "eval_runtime": 41.0638,
+      "eval_samples_per_second": 12.176,
+      "eval_steps_per_second": 0.779,
+      "step": 8300
+    },
+    {
+      "epoch": 2.249593936112615,
+      "grad_norm": 0.3514440357685089,
+      "learning_rate": 0.00012620651975960663,
+      "loss": 0.0922,
+      "step": 8310
+    },
+    {
+      "epoch": 2.2523010286951815,
+      "grad_norm": 0.18054944276809692,
+      "learning_rate": 0.00012575122928428338,
+      "loss": 0.1005,
+      "step": 8320
+    },
+    {
+      "epoch": 2.2550081212777475,
+      "grad_norm": 0.20426245033740997,
+      "learning_rate": 0.0001252959388089601,
+      "loss": 0.0919,
+      "step": 8330
+    },
+    {
+      "epoch": 2.257715213860314,
+      "grad_norm": 0.14870111644268036,
+      "learning_rate": 0.00012484064833363686,
+      "loss": 0.0819,
+      "step": 8340
+    },
+    {
+      "epoch": 2.26042230644288,
+      "grad_norm": 0.09837736189365387,
+      "learning_rate": 0.00012438535785831362,
+      "loss": 0.0633,
+      "step": 8350
+    },
+    {
+      "epoch": 2.2631293990254466,
+      "grad_norm": 0.20747441053390503,
+      "learning_rate": 0.00012393006738299034,
+      "loss": 0.0698,
+      "step": 8360
+    },
+    {
+      "epoch": 2.265836491608013,
+      "grad_norm": 0.29765015840530396,
+      "learning_rate": 0.0001234747769076671,
+      "loss": 0.1034,
+      "step": 8370
+    },
+    {
+      "epoch": 2.268543584190579,
+      "grad_norm": 0.22020211815834045,
+      "learning_rate": 0.00012301948643234385,
+      "loss": 0.1182,
+      "step": 8380
+    },
+    {
+      "epoch": 2.2712506767731457,
+      "grad_norm": 0.25610819458961487,
+      "learning_rate": 0.00012256419595702058,
+      "loss": 0.1114,
+      "step": 8390
+    },
+    {
+      "epoch": 2.273957769355712,
+      "grad_norm": 0.09575118869543076,
+      "learning_rate": 0.00012210890548169733,
+      "loss": 0.112,
+      "step": 8400
+    },
+    {
+      "epoch": 2.273957769355712,
+      "eval_loss": 0.12145300209522247,
+      "eval_runtime": 41.8368,
+      "eval_samples_per_second": 11.951,
+      "eval_steps_per_second": 0.765,
+      "step": 8400
+    },
+    {
+      "epoch": 2.2766648619382783,
+      "grad_norm": 0.14825108647346497,
+      "learning_rate": 0.00012165361500637407,
+      "loss": 0.089,
+      "step": 8410
+    },
+    {
+      "epoch": 2.279371954520845,
+      "grad_norm": 0.1572876274585724,
+      "learning_rate": 0.00012119832453105081,
+      "loss": 0.0963,
+      "step": 8420
+    },
+    {
+      "epoch": 2.282079047103411,
+      "grad_norm": 0.17988793551921844,
+      "learning_rate": 0.00012074303405572756,
+      "loss": 0.0855,
+      "step": 8430
+    },
+    {
+      "epoch": 2.2847861396859774,
+      "grad_norm": 0.21360033750534058,
+      "learning_rate": 0.0001202877435804043,
+      "loss": 0.1171,
+      "step": 8440
+    },
+    {
+      "epoch": 2.2874932322685435,
+      "grad_norm": 0.18448184430599213,
+      "learning_rate": 0.00011983245310508104,
+      "loss": 0.0884,
+      "step": 8450
+    },
+    {
+      "epoch": 2.29020032485111,
+      "grad_norm": 0.20717447996139526,
+      "learning_rate": 0.00011937716262975778,
+      "loss": 0.0872,
+      "step": 8460
+    },
+    {
+      "epoch": 2.292907417433676,
+      "grad_norm": 0.36214086413383484,
+      "learning_rate": 0.00011892187215443454,
+      "loss": 0.0964,
+      "step": 8470
+    },
+    {
+      "epoch": 2.2956145100162426,
+      "grad_norm": 0.0883878692984581,
+      "learning_rate": 0.00011846658167911128,
+      "loss": 0.0698,
+      "step": 8480
+    },
+    {
+      "epoch": 2.2983216025988087,
+      "grad_norm": 0.1264089047908783,
+      "learning_rate": 0.00011801129120378802,
+      "loss": 0.089,
+      "step": 8490
+    },
+    {
+      "epoch": 2.301028695181375,
+      "grad_norm": 0.18756982684135437,
+      "learning_rate": 0.00011755600072846476,
+      "loss": 0.0987,
+      "step": 8500
+    },
+    {
+      "epoch": 2.301028695181375,
+      "eval_loss": 0.1207582876086235,
+      "eval_runtime": 40.3933,
+      "eval_samples_per_second": 12.378,
+      "eval_steps_per_second": 0.792,
+      "step": 8500
+    },
+    {
+      "epoch": 2.3037357877639417,
+      "grad_norm": 0.32923880219459534,
+      "learning_rate": 0.00011710071025314151,
+      "loss": 0.0973,
+      "step": 8510
+    },
+    {
+      "epoch": 2.306442880346508,
+      "grad_norm": 0.28304001688957214,
+      "learning_rate": 0.00011664541977781825,
+      "loss": 0.0942,
+      "step": 8520
+    },
+    {
+      "epoch": 2.3091499729290743,
+      "grad_norm": 0.19546718895435333,
+      "learning_rate": 0.00011619012930249499,
+      "loss": 0.1055,
+      "step": 8530
+    },
+    {
+      "epoch": 2.3118570655116404,
+      "grad_norm": 0.20252230763435364,
+      "learning_rate": 0.00011573483882717174,
+      "loss": 0.1207,
+      "step": 8540
+    },
+    {
+      "epoch": 2.314564158094207,
+      "grad_norm": 0.2110469788312912,
+      "learning_rate": 0.00011527954835184849,
+      "loss": 0.0849,
+      "step": 8550
+    },
+    {
+      "epoch": 2.317271250676773,
+      "grad_norm": 0.33697932958602905,
+      "learning_rate": 0.00011482425787652523,
+      "loss": 0.1018,
+      "step": 8560
+    },
+    {
+      "epoch": 2.3199783432593395,
+      "grad_norm": 0.22342148423194885,
+      "learning_rate": 0.00011436896740120197,
+      "loss": 0.0983,
+      "step": 8570
+    },
+    {
+      "epoch": 2.3226854358419056,
+      "grad_norm": 0.1446394920349121,
+      "learning_rate": 0.00011391367692587872,
+      "loss": 0.1008,
+      "step": 8580
+    },
+    {
+      "epoch": 2.325392528424472,
+      "grad_norm": 0.46597152948379517,
+      "learning_rate": 0.00011345838645055545,
+      "loss": 0.1355,
+      "step": 8590
+    },
+    {
+      "epoch": 2.3280996210070386,
+      "grad_norm": 0.1944790929555893,
+      "learning_rate": 0.0001130030959752322,
+      "loss": 0.0744,
+      "step": 8600
+    },
+    {
+      "epoch": 2.3280996210070386,
+      "eval_loss": 0.11873700469732285,
+      "eval_runtime": 42.559,
+      "eval_samples_per_second": 11.748,
+      "eval_steps_per_second": 0.752,
+      "step": 8600
+    },
+    {
+      "epoch": 2.3308067135896047,
+      "grad_norm": 0.277667760848999,
+      "learning_rate": 0.00011254780549990895,
+      "loss": 0.1042,
+      "step": 8610
+    },
+    {
+      "epoch": 2.333513806172171,
+      "grad_norm": 0.1340201199054718,
+      "learning_rate": 0.00011209251502458569,
+      "loss": 0.071,
+      "step": 8620
+    },
+    {
+      "epoch": 2.3362208987547373,
+      "grad_norm": 0.15275384485721588,
+      "learning_rate": 0.00011163722454926243,
+      "loss": 0.0851,
+      "step": 8630
+    },
+    {
+      "epoch": 2.338927991337304,
+      "grad_norm": 0.571696937084198,
+      "learning_rate": 0.00011118193407393917,
+      "loss": 0.0629,
+      "step": 8640
+    },
+    {
+      "epoch": 2.3416350839198703,
+      "grad_norm": 0.16484041512012482,
+      "learning_rate": 0.00011072664359861593,
+      "loss": 0.1282,
+      "step": 8650
+    },
+    {
+      "epoch": 2.3443421765024364,
+      "grad_norm": 0.424176424741745,
+      "learning_rate": 0.00011027135312329265,
+      "loss": 0.0751,
+      "step": 8660
+    },
+    {
+      "epoch": 2.347049269085003,
+      "grad_norm": 0.2215598225593567,
+      "learning_rate": 0.0001098160626479694,
+      "loss": 0.0842,
+      "step": 8670
+    },
+    {
+      "epoch": 2.349756361667569,
+      "grad_norm": 0.13979794085025787,
+      "learning_rate": 0.00010936077217264615,
+      "loss": 0.076,
+      "step": 8680
+    },
+    {
+      "epoch": 2.3524634542501355,
+      "grad_norm": 0.19357985258102417,
+      "learning_rate": 0.0001089054816973229,
+      "loss": 0.0887,
+      "step": 8690
+    },
+    {
+      "epoch": 2.3551705468327016,
+      "grad_norm": 0.12454704940319061,
+      "learning_rate": 0.00010845019122199964,
+      "loss": 0.0934,
+      "step": 8700
+    },
+    {
+      "epoch": 2.3551705468327016,
+      "eval_loss": 0.1178118884563446,
+      "eval_runtime": 42.1306,
+      "eval_samples_per_second": 11.868,
+      "eval_steps_per_second": 0.76,
+      "step": 8700
+    },
+    {
+      "epoch": 2.357877639415268,
+      "grad_norm": 0.14147062599658966,
+      "learning_rate": 0.00010799490074667638,
+      "loss": 0.0818,
+      "step": 8710
+    },
+    {
+      "epoch": 2.360584731997834,
+      "grad_norm": 0.3022850453853607,
+      "learning_rate": 0.00010753961027135313,
+      "loss": 0.085,
+      "step": 8720
+    },
+    {
+      "epoch": 2.3632918245804007,
+      "grad_norm": 0.2855629324913025,
+      "learning_rate": 0.00010708431979602986,
+      "loss": 0.0764,
+      "step": 8730
+    },
+    {
+      "epoch": 2.365998917162967,
+      "grad_norm": 0.2926426827907562,
+      "learning_rate": 0.00010662902932070661,
+      "loss": 0.1003,
+      "step": 8740
+    },
+    {
+      "epoch": 2.3687060097455332,
+      "grad_norm": 0.3509204089641571,
+      "learning_rate": 0.00010617373884538335,
+      "loss": 0.0839,
+      "step": 8750
+    },
+    {
+      "epoch": 2.3714131023280998,
+      "grad_norm": 0.2791828215122223,
+      "learning_rate": 0.00010571844837006011,
+      "loss": 0.1086,
+      "step": 8760
+    },
+    {
+      "epoch": 2.374120194910666,
+      "grad_norm": 0.12237832695245743,
+      "learning_rate": 0.00010526315789473683,
+      "loss": 0.0878,
+      "step": 8770
+    },
+    {
+      "epoch": 2.3768272874932324,
+      "grad_norm": 0.3214811384677887,
+      "learning_rate": 0.00010480786741941359,
+      "loss": 0.0824,
+      "step": 8780
+    },
+    {
+      "epoch": 2.3795343800757984,
+      "grad_norm": 0.3366682827472687,
+      "learning_rate": 0.00010435257694409034,
+      "loss": 0.0798,
+      "step": 8790
+    },
+    {
+      "epoch": 2.382241472658365,
+      "grad_norm": 0.45946112275123596,
+      "learning_rate": 0.00010389728646876707,
+      "loss": 0.0854,
+      "step": 8800
+    },
+    {
+      "epoch": 2.382241472658365,
+      "eval_loss": 0.11880424618721008,
+      "eval_runtime": 40.1429,
+      "eval_samples_per_second": 12.456,
+      "eval_steps_per_second": 0.797,
+      "step": 8800
+    },
+    {
+      "epoch": 2.384948565240931,
+      "grad_norm": 0.16509723663330078,
+      "learning_rate": 0.00010344199599344382,
+      "loss": 0.0926,
+      "step": 8810
+    },
+    {
+      "epoch": 2.3876556578234975,
+      "grad_norm": 0.4739864766597748,
+      "learning_rate": 0.00010298670551812056,
+      "loss": 0.0749,
+      "step": 8820
+    },
+    {
+      "epoch": 2.390362750406064,
+      "grad_norm": 0.1993105560541153,
+      "learning_rate": 0.00010253141504279732,
+      "loss": 0.0846,
+      "step": 8830
+    },
+    {
+      "epoch": 2.39306984298863,
+      "grad_norm": 0.24470995366573334,
+      "learning_rate": 0.00010207612456747404,
+      "loss": 0.1196,
+      "step": 8840
+    },
+    {
+      "epoch": 2.3957769355711966,
+      "grad_norm": 0.07803446054458618,
+      "learning_rate": 0.0001016208340921508,
+      "loss": 0.1189,
+      "step": 8850
+    },
+    {
+      "epoch": 2.3984840281537627,
+      "grad_norm": 0.15416881442070007,
+      "learning_rate": 0.00010116554361682754,
+      "loss": 0.0886,
+      "step": 8860
+    },
+    {
+      "epoch": 2.4011911207363292,
+      "grad_norm": 0.14576180279254913,
+      "learning_rate": 0.00010071025314150428,
+      "loss": 0.0948,
+      "step": 8870
+    },
+    {
+      "epoch": 2.4038982133188957,
+      "grad_norm": 0.1478489637374878,
+      "learning_rate": 0.00010025496266618103,
+      "loss": 0.0672,
+      "step": 8880
+    },
+    {
+      "epoch": 2.406605305901462,
+      "grad_norm": 0.34242862462997437,
+      "learning_rate": 9.979967219085777e-05,
+      "loss": 0.125,
+      "step": 8890
+    },
+    {
+      "epoch": 2.4093123984840283,
+      "grad_norm": 0.0876445397734642,
+      "learning_rate": 9.934438171553452e-05,
+      "loss": 0.1304,
+      "step": 8900
+    },
+    {
+      "epoch": 2.4093123984840283,
+      "eval_loss": 0.11949001252651215,
+      "eval_runtime": 41.6295,
+      "eval_samples_per_second": 12.011,
+      "eval_steps_per_second": 0.769,
+      "step": 8900
+    },
+    {
+      "epoch": 2.4120194910665944,
+      "grad_norm": 0.053039271384477615,
+      "learning_rate": 9.888909124021125e-05,
+      "loss": 0.09,
+      "step": 8910
+    },
+    {
+      "epoch": 2.414726583649161,
+      "grad_norm": 0.145661860704422,
+      "learning_rate": 9.8433800764888e-05,
+      "loss": 0.0907,
+      "step": 8920
+    },
+    {
+      "epoch": 2.417433676231727,
+      "grad_norm": 0.21455125510692596,
+      "learning_rate": 9.797851028956474e-05,
+      "loss": 0.0771,
+      "step": 8930
+    },
+    {
+      "epoch": 2.4201407688142935,
+      "grad_norm": 0.2978738248348236,
+      "learning_rate": 9.75232198142415e-05,
+      "loss": 0.1215,
+      "step": 8940
+    },
+    {
+      "epoch": 2.4228478613968596,
+      "grad_norm": 0.25670331716537476,
+      "learning_rate": 9.706792933891822e-05,
+      "loss": 0.1097,
+      "step": 8950
+    },
+    {
+      "epoch": 2.425554953979426,
+      "grad_norm": 0.14389146864414215,
+      "learning_rate": 9.661263886359498e-05,
+      "loss": 0.0839,
+      "step": 8960
+    },
+    {
+      "epoch": 2.4282620465619926,
+      "grad_norm": 0.018924234434962273,
+      "learning_rate": 9.615734838827173e-05,
+      "loss": 0.0792,
+      "step": 8970
+    },
+    {
+      "epoch": 2.4309691391445587,
+      "grad_norm": 0.2112540900707245,
+      "learning_rate": 9.570205791294846e-05,
+      "loss": 0.1021,
+      "step": 8980
+    },
+    {
+      "epoch": 2.433676231727125,
+      "grad_norm": 0.18614734709262848,
+      "learning_rate": 9.524676743762521e-05,
+      "loss": 0.0825,
+      "step": 8990
+    },
+    {
+      "epoch": 2.4363833243096913,
+      "grad_norm": 0.17315484583377838,
+      "learning_rate": 9.479147696230195e-05,
+      "loss": 0.0936,
+      "step": 9000
+    },
+    {
+      "epoch": 2.4363833243096913,
+      "eval_loss": 0.12042009085416794,
+      "eval_runtime": 40.5282,
+      "eval_samples_per_second": 12.337,
+      "eval_steps_per_second": 0.79,
+      "step": 9000
+    },
+    {
+      "epoch": 2.439090416892258,
+      "grad_norm": 0.3776973485946655,
+      "learning_rate": 9.43361864869787e-05,
+      "loss": 0.1001,
+      "step": 9010
+    },
+    {
+      "epoch": 2.441797509474824,
+      "grad_norm": 0.08049853146076202,
+      "learning_rate": 9.388089601165543e-05,
+      "loss": 0.0884,
+      "step": 9020
+    },
+    {
+      "epoch": 2.4445046020573904,
+      "grad_norm": 0.08153152465820312,
+      "learning_rate": 9.342560553633218e-05,
+      "loss": 0.0955,
+      "step": 9030
+    },
+    {
+      "epoch": 2.4472116946399565,
+      "grad_norm": 0.2704846262931824,
+      "learning_rate": 9.297031506100892e-05,
+      "loss": 0.0813,
+      "step": 9040
+    },
+    {
+      "epoch": 2.449918787222523,
+      "grad_norm": 0.41619887948036194,
+      "learning_rate": 9.251502458568566e-05,
+      "loss": 0.1172,
+      "step": 9050
+    },
+    {
+      "epoch": 2.4526258798050895,
+      "grad_norm": 0.19113139808177948,
+      "learning_rate": 9.205973411036242e-05,
+      "loss": 0.0873,
+      "step": 9060
+    },
+    {
+      "epoch": 2.4553329723876556,
+      "grad_norm": 0.32321926951408386,
+      "learning_rate": 9.160444363503916e-05,
+      "loss": 0.1091,
+      "step": 9070
+    },
+    {
+      "epoch": 2.458040064970222,
+      "grad_norm": 0.612158477306366,
+      "learning_rate": 9.114915315971591e-05,
+      "loss": 0.1414,
+      "step": 9080
+    },
+    {
+      "epoch": 2.460747157552788,
+      "grad_norm": 0.2993355393409729,
+      "learning_rate": 9.069386268439264e-05,
+      "loss": 0.1011,
+      "step": 9090
+    },
+    {
+      "epoch": 2.4634542501353547,
+      "grad_norm": 0.3718532621860504,
+      "learning_rate": 9.023857220906939e-05,
+      "loss": 0.0863,
+      "step": 9100
+    },
+    {
+      "epoch": 2.4634542501353547,
+      "eval_loss": 0.11713527143001556,
+      "eval_runtime": 38.7457,
+      "eval_samples_per_second": 12.905,
+      "eval_steps_per_second": 0.826,
+      "step": 9100
+    },
+    {
+      "epoch": 2.466161342717921,
+      "grad_norm": 0.35147398710250854,
+      "learning_rate": 8.978328173374613e-05,
+      "loss": 0.0951,
+      "step": 9110
+    },
+    {
+      "epoch": 2.4688684353004873,
+      "grad_norm": 0.3393324911594391,
+      "learning_rate": 8.932799125842287e-05,
+      "loss": 0.0818,
+      "step": 9120
+    },
+    {
+      "epoch": 2.471575527883054,
+      "grad_norm": 0.06653381884098053,
+      "learning_rate": 8.887270078309961e-05,
+      "loss": 0.0885,
+      "step": 9130
+    },
+    {
+      "epoch": 2.47428262046562,
+      "grad_norm": 0.1788959950208664,
+      "learning_rate": 8.841741030777637e-05,
+      "loss": 0.1032,
+      "step": 9140
+    },
+    {
+      "epoch": 2.4769897130481864,
+      "grad_norm": 0.24008777737617493,
+      "learning_rate": 8.796211983245312e-05,
+      "loss": 0.1034,
+      "step": 9150
+    },
+    {
+      "epoch": 2.4796968056307525,
+      "grad_norm": 0.07579132169485092,
+      "learning_rate": 8.750682935712985e-05,
+      "loss": 0.1162,
+      "step": 9160
+    },
+    {
+      "epoch": 2.482403898213319,
+      "grad_norm": 0.295665979385376,
+      "learning_rate": 8.70515388818066e-05,
+      "loss": 0.1095,
+      "step": 9170
+    },
+    {
+      "epoch": 2.485110990795885,
+      "grad_norm": 0.11401191353797913,
+      "learning_rate": 8.659624840648334e-05,
+      "loss": 0.0812,
+      "step": 9180
+    },
+    {
+      "epoch": 2.4878180833784516,
+      "grad_norm": 0.15605595707893372,
+      "learning_rate": 8.614095793116008e-05,
+      "loss": 0.081,
+      "step": 9190
+    },
+    {
+      "epoch": 2.490525175961018,
+      "grad_norm": 0.20551259815692902,
+      "learning_rate": 8.568566745583682e-05,
+      "loss": 0.0884,
+      "step": 9200
+    },
+    {
+      "epoch": 2.490525175961018,
+      "eval_loss": 0.11728712916374207,
+      "eval_runtime": 41.2745,
+      "eval_samples_per_second": 12.114,
+      "eval_steps_per_second": 0.775,
+      "step": 9200
+    },
+    {
+      "epoch": 2.493232268543584,
+      "grad_norm": 0.3178587555885315,
+      "learning_rate": 8.523037698051357e-05,
+      "loss": 0.0867,
+      "step": 9210
+    },
+    {
+      "epoch": 2.4959393611261507,
+      "grad_norm": 0.21806322038173676,
+      "learning_rate": 8.477508650519031e-05,
+      "loss": 0.0981,
+      "step": 9220
+    },
+    {
+      "epoch": 2.4986464537087167,
+      "grad_norm": 0.11352866142988205,
+      "learning_rate": 8.431979602986705e-05,
+      "loss": 0.1255,
+      "step": 9230
+    },
+    {
+      "epoch": 2.5013535462912833,
+      "grad_norm": 0.39240339398384094,
+      "learning_rate": 8.386450555454381e-05,
+      "loss": 0.1052,
+      "step": 9240
+    },
+    {
+      "epoch": 2.5040606388738493,
+      "grad_norm": 0.27294594049453735,
+      "learning_rate": 8.340921507922055e-05,
+      "loss": 0.1039,
+      "step": 9250
+    },
+    {
+      "epoch": 2.506767731456416,
+      "grad_norm": 0.38210707902908325,
+      "learning_rate": 8.295392460389729e-05,
+      "loss": 0.0951,
+      "step": 9260
+    },
+    {
+      "epoch": 2.509474824038982,
+      "grad_norm": 0.23839713633060455,
+      "learning_rate": 8.249863412857403e-05,
+      "loss": 0.0843,
+      "step": 9270
+    },
+    {
+      "epoch": 2.5121819166215484,
+      "grad_norm": 0.21028132736682892,
+      "learning_rate": 8.204334365325078e-05,
+      "loss": 0.0828,
+      "step": 9280
+    },
+    {
+      "epoch": 2.514889009204115,
+      "grad_norm": 0.24762462079524994,
+      "learning_rate": 8.158805317792752e-05,
+      "loss": 0.0761,
+      "step": 9290
+    },
+    {
+      "epoch": 2.517596101786681,
+      "grad_norm": 0.24827738106250763,
+      "learning_rate": 8.113276270260426e-05,
+      "loss": 0.1043,
+      "step": 9300
+    },
+    {
+      "epoch": 2.517596101786681,
+      "eval_loss": 0.11610767990350723,
+      "eval_runtime": 42.4786,
+      "eval_samples_per_second": 11.771,
+      "eval_steps_per_second": 0.753,
+      "step": 9300
+    },
+    {
+      "epoch": 2.5203031943692475,
+      "grad_norm": 0.3143943250179291,
+      "learning_rate": 8.0677472227281e-05,
+      "loss": 0.0884,
+      "step": 9310
+    },
+    {
+      "epoch": 2.5230102869518136,
+      "grad_norm": 0.21991610527038574,
+      "learning_rate": 8.022218175195775e-05,
+      "loss": 0.0915,
+      "step": 9320
+    },
+    {
+      "epoch": 2.52571737953438,
+      "grad_norm": 0.19134922325611115,
+      "learning_rate": 7.97668912766345e-05,
+      "loss": 0.1121,
+      "step": 9330
+    },
+    {
+      "epoch": 2.5284244721169467,
+      "grad_norm": 0.37185224890708923,
+      "learning_rate": 7.931160080131124e-05,
+      "loss": 0.0734,
+      "step": 9340
+    },
+    {
+      "epoch": 2.5311315646995127,
+      "grad_norm": 0.08370956033468246,
+      "learning_rate": 7.885631032598799e-05,
+      "loss": 0.0841,
+      "step": 9350
+    },
+    {
+      "epoch": 2.533838657282079,
+      "grad_norm": 0.1807362586259842,
+      "learning_rate": 7.840101985066473e-05,
+      "loss": 0.0912,
+      "step": 9360
+    },
+    {
+      "epoch": 2.5365457498646453,
+      "grad_norm": 0.24182716012001038,
+      "learning_rate": 7.794572937534147e-05,
+      "loss": 0.0818,
+      "step": 9370
+    },
+    {
+      "epoch": 2.539252842447212,
+      "grad_norm": 0.07888434082269669,
+      "learning_rate": 7.749043890001821e-05,
+      "loss": 0.0791,
+      "step": 9380
+    },
+    {
+      "epoch": 2.541959935029778,
+      "grad_norm": 0.5377534031867981,
+      "learning_rate": 7.703514842469496e-05,
+      "loss": 0.0928,
+      "step": 9390
+    },
+    {
+      "epoch": 2.5446670276123444,
+      "grad_norm": 0.18035459518432617,
+      "learning_rate": 7.657985794937169e-05,
+      "loss": 0.0945,
+      "step": 9400
+    },
+    {
+      "epoch": 2.5446670276123444,
+      "eval_loss": 0.11670617759227753,
+      "eval_runtime": 43.1703,
+      "eval_samples_per_second": 11.582,
+      "eval_steps_per_second": 0.741,
+      "step": 9400
+    },
+    {
+      "epoch": 2.5473741201949105,
+      "grad_norm": 0.09470925480127335,
+      "learning_rate": 7.612456747404844e-05,
+      "loss": 0.0827,
+      "step": 9410
+    },
+    {
+      "epoch": 2.550081212777477,
+      "grad_norm": 0.2242165207862854,
+      "learning_rate": 7.56692769987252e-05,
+      "loss": 0.0933,
+      "step": 9420
+    },
+    {
+      "epoch": 2.5527883053600435,
+      "grad_norm": 0.12018734216690063,
+      "learning_rate": 7.521398652340194e-05,
+      "loss": 0.0773,
+      "step": 9430
+    },
+    {
+      "epoch": 2.5554953979426096,
+      "grad_norm": 0.32595720887184143,
+      "learning_rate": 7.475869604807868e-05,
+      "loss": 0.0986,
+      "step": 9440
+    },
+    {
+      "epoch": 2.558202490525176,
+      "grad_norm": 0.15024429559707642,
+      "learning_rate": 7.430340557275542e-05,
+      "loss": 0.0749,
+      "step": 9450
+    },
+    {
+      "epoch": 2.560909583107742,
+      "grad_norm": 0.13837756216526031,
+      "learning_rate": 7.384811509743217e-05,
+      "loss": 0.0972,
+      "step": 9460
+    },
+    {
+      "epoch": 2.5636166756903087,
+      "grad_norm": 0.18516859412193298,
+      "learning_rate": 7.33928246221089e-05,
+      "loss": 0.0841,
+      "step": 9470
+    },
+    {
+      "epoch": 2.566323768272875,
+      "grad_norm": 0.4080735146999359,
+      "learning_rate": 7.293753414678565e-05,
+      "loss": 0.0906,
+      "step": 9480
+    },
+    {
+      "epoch": 2.5690308608554413,
+      "grad_norm": 0.4364292621612549,
+      "learning_rate": 7.248224367146239e-05,
+      "loss": 0.1048,
+      "step": 9490
+    },
+    {
+      "epoch": 2.5717379534380074,
+      "grad_norm": 0.26743844151496887,
+      "learning_rate": 7.202695319613914e-05,
+      "loss": 0.0863,
+      "step": 9500
+    },
+    {
+      "epoch": 2.5717379534380074,
+      "eval_loss": 0.11930057406425476,
+      "eval_runtime": 42.0052,
+      "eval_samples_per_second": 11.903,
+      "eval_steps_per_second": 0.762,
+      "step": 9500
+    },
+    {
+      "epoch": 2.574445046020574,
+      "grad_norm": 0.2927606403827667,
+      "learning_rate": 7.157166272081588e-05,
+      "loss": 0.1026,
+      "step": 9510
+    },
+    {
+      "epoch": 2.5771521386031404,
+      "grad_norm": 0.1961708813905716,
+      "learning_rate": 7.111637224549262e-05,
+      "loss": 0.0796,
+      "step": 9520
+    },
+    {
+      "epoch": 2.5798592311857065,
+      "grad_norm": 0.4174273610115051,
+      "learning_rate": 7.066108177016938e-05,
+      "loss": 0.0961,
+      "step": 9530
+    },
+    {
+      "epoch": 2.582566323768273,
+      "grad_norm": 0.383043497800827,
+      "learning_rate": 7.02057912948461e-05,
+      "loss": 0.1027,
+      "step": 9540
+    },
+    {
+      "epoch": 2.585273416350839,
+      "grad_norm": 0.39172255992889404,
+      "learning_rate": 6.975050081952286e-05,
+      "loss": 0.1017,
+      "step": 9550
+    },
+    {
+      "epoch": 2.5879805089334056,
+      "grad_norm": 0.292972207069397,
+      "learning_rate": 6.92952103441996e-05,
+      "loss": 0.1229,
+      "step": 9560
+    },
+    {
+      "epoch": 2.590687601515972,
+      "grad_norm": 0.14154380559921265,
+      "learning_rate": 6.883991986887635e-05,
+      "loss": 0.112,
+      "step": 9570
+    },
+    {
+      "epoch": 2.593394694098538,
+      "grad_norm": 0.31920763850212097,
+      "learning_rate": 6.838462939355308e-05,
+      "loss": 0.0925,
+      "step": 9580
+    },
+    {
+      "epoch": 2.5961017866811043,
+      "grad_norm": 0.7873738408088684,
+      "learning_rate": 6.792933891822983e-05,
+      "loss": 0.146,
+      "step": 9590
+    },
+    {
+      "epoch": 2.5988088792636708,
+      "grad_norm": 0.18883179128170013,
+      "learning_rate": 6.747404844290659e-05,
+      "loss": 0.1135,
+      "step": 9600
+    },
+    {
+      "epoch": 2.5988088792636708,
+      "eval_loss": 0.11629679054021835,
+      "eval_runtime": 42.3772,
+      "eval_samples_per_second": 11.799,
+      "eval_steps_per_second": 0.755,
+      "step": 9600
+    },
+    {
+      "epoch": 2.6015159718462373,
+      "grad_norm": 0.31590747833251953,
+      "learning_rate": 6.701875796758333e-05,
+      "loss": 0.0989,
+      "step": 9610
+    },
+    {
+      "epoch": 2.6042230644288034,
+      "grad_norm": 0.1910378634929657,
+      "learning_rate": 6.656346749226007e-05,
+      "loss": 0.0759,
+      "step": 9620
+    },
+    {
+      "epoch": 2.60693015701137,
+      "grad_norm": 0.0984259769320488,
+      "learning_rate": 6.61081770169368e-05,
+      "loss": 0.078,
+      "step": 9630
+    },
+    {
+      "epoch": 2.609637249593936,
+      "grad_norm": 0.23022520542144775,
+      "learning_rate": 6.565288654161356e-05,
+      "loss": 0.0956,
+      "step": 9640
+    },
+    {
+      "epoch": 2.6123443421765025,
+      "grad_norm": 0.21245041489601135,
+      "learning_rate": 6.519759606629029e-05,
+      "loss": 0.0959,
+      "step": 9650
+    },
+    {
+      "epoch": 2.615051434759069,
+      "grad_norm": 0.12750999629497528,
+      "learning_rate": 6.474230559096704e-05,
+      "loss": 0.0684,
+      "step": 9660
+    },
+    {
+      "epoch": 2.617758527341635,
+      "grad_norm": 0.07224147766828537,
+      "learning_rate": 6.428701511564378e-05,
+      "loss": 0.0853,
+      "step": 9670
+    },
+    {
+      "epoch": 2.6204656199242016,
+      "grad_norm": 0.2618503272533417,
+      "learning_rate": 6.383172464032053e-05,
+      "loss": 0.0761,
+      "step": 9680
+    },
+    {
+      "epoch": 2.6231727125067676,
+      "grad_norm": 0.12275584042072296,
+      "learning_rate": 6.337643416499727e-05,
+      "loss": 0.0687,
+      "step": 9690
+    },
+    {
+      "epoch": 2.625879805089334,
+      "grad_norm": 0.344372421503067,
+      "learning_rate": 6.292114368967401e-05,
+      "loss": 0.0941,
+      "step": 9700
+    },
+    {
+      "epoch": 2.625879805089334,
+      "eval_loss": 0.11677241325378418,
+      "eval_runtime": 40.6561,
+      "eval_samples_per_second": 12.298,
+      "eval_steps_per_second": 0.787,
+      "step": 9700
+    },
+    {
+      "epoch": 2.6285868976719002,
+      "grad_norm": 0.27887827157974243,
+      "learning_rate": 6.246585321435075e-05,
+      "loss": 0.102,
+      "step": 9710
+    },
+    {
+      "epoch": 2.6312939902544668,
+      "grad_norm": 0.11722956597805023,
+      "learning_rate": 6.201056273902749e-05,
+      "loss": 0.0863,
+      "step": 9720
+    },
+    {
+      "epoch": 2.634001082837033,
+      "grad_norm": 0.24264639616012573,
+      "learning_rate": 6.155527226370425e-05,
+      "loss": 0.1036,
+      "step": 9730
+    },
+    {
+      "epoch": 2.6367081754195993,
+      "grad_norm": 0.5664383172988892,
+      "learning_rate": 6.109998178838099e-05,
+      "loss": 0.0991,
+      "step": 9740
+    },
+    {
+      "epoch": 2.639415268002166,
+      "grad_norm": 0.1473313570022583,
+      "learning_rate": 6.0644691313057734e-05,
+      "loss": 0.0996,
+      "step": 9750
+    },
+    {
+      "epoch": 2.642122360584732,
+      "grad_norm": 0.1233234703540802,
+      "learning_rate": 6.018940083773448e-05,
+      "loss": 0.0807,
+      "step": 9760
+    },
+    {
+      "epoch": 2.6448294531672984,
+      "grad_norm": 0.39607110619544983,
+      "learning_rate": 5.973411036241122e-05,
+      "loss": 0.0879,
+      "step": 9770
+    },
+    {
+      "epoch": 2.6475365457498645,
+      "grad_norm": 0.08371660113334656,
+      "learning_rate": 5.927881988708796e-05,
+      "loss": 0.0814,
+      "step": 9780
+    },
+    {
+      "epoch": 2.650243638332431,
+      "grad_norm": 0.15465834736824036,
+      "learning_rate": 5.882352941176471e-05,
+      "loss": 0.098,
+      "step": 9790
+    },
+    {
+      "epoch": 2.6529507309149976,
+      "grad_norm": 0.16004501283168793,
+      "learning_rate": 5.836823893644145e-05,
+      "loss": 0.1037,
+      "step": 9800
+    },
+    {
+      "epoch": 2.6529507309149976,
+      "eval_loss": 0.1185322180390358,
+      "eval_runtime": 40.0044,
+      "eval_samples_per_second": 12.499,
+      "eval_steps_per_second": 0.8,
+      "step": 9800
+    },
+    {
+      "epoch": 2.6556578234975636,
+      "grad_norm": 0.15469568967819214,
+      "learning_rate": 5.7912948461118194e-05,
+      "loss": 0.1009,
+      "step": 9810
+    },
+    {
+      "epoch": 2.6583649160801297,
+      "grad_norm": 0.14093799889087677,
+      "learning_rate": 5.7457657985794934e-05,
+      "loss": 0.0729,
+      "step": 9820
+    },
+    {
+      "epoch": 2.661072008662696,
+      "grad_norm": 0.16190478205680847,
+      "learning_rate": 5.700236751047169e-05,
+      "loss": 0.0897,
+      "step": 9830
+    },
+    {
+      "epoch": 2.6637791012452627,
+      "grad_norm": 0.27476224303245544,
+      "learning_rate": 5.654707703514843e-05,
+      "loss": 0.1004,
+      "step": 9840
+    },
+    {
+      "epoch": 2.666486193827829,
+      "grad_norm": 0.2863362431526184,
+      "learning_rate": 5.609178655982517e-05,
+      "loss": 0.0924,
+      "step": 9850
+    },
+    {
+      "epoch": 2.6691932864103953,
+      "grad_norm": 0.15834559500217438,
+      "learning_rate": 5.5636496084501915e-05,
+      "loss": 0.083,
+      "step": 9860
+    },
+    {
+      "epoch": 2.6719003789929614,
+      "grad_norm": 0.1608414351940155,
+      "learning_rate": 5.5181205609178655e-05,
+      "loss": 0.0807,
+      "step": 9870
+    },
+    {
+      "epoch": 2.674607471575528,
+      "grad_norm": 0.14112305641174316,
+      "learning_rate": 5.47259151338554e-05,
+      "loss": 0.0924,
+      "step": 9880
+    },
+    {
+      "epoch": 2.6773145641580944,
+      "grad_norm": 0.5847556591033936,
+      "learning_rate": 5.427062465853214e-05,
+      "loss": 0.1086,
+      "step": 9890
+    },
+    {
+      "epoch": 2.6800216567406605,
+      "grad_norm": 0.16682197153568268,
+      "learning_rate": 5.381533418320889e-05,
+      "loss": 0.0959,
+      "step": 9900
+    },
+    {
+      "epoch": 2.6800216567406605,
+      "eval_loss": 0.11745791137218475,
+      "eval_runtime": 39.9003,
+      "eval_samples_per_second": 12.531,
+      "eval_steps_per_second": 0.802,
+      "step": 9900
+    },
+    {
+      "epoch": 2.6827287493232266,
+      "grad_norm": 0.26944684982299805,
+      "learning_rate": 5.336004370788563e-05,
+      "loss": 0.0935,
+      "step": 9910
+    },
+    {
+      "epoch": 2.685435841905793,
+      "grad_norm": 0.14497505128383636,
+      "learning_rate": 5.2904753232562376e-05,
+      "loss": 0.0674,
+      "step": 9920
+    },
+    {
+      "epoch": 2.6881429344883596,
+      "grad_norm": 0.17957589030265808,
+      "learning_rate": 5.244946275723912e-05,
+      "loss": 0.0869,
+      "step": 9930
+    },
+    {
+      "epoch": 2.6908500270709257,
+      "grad_norm": 0.10685747861862183,
+      "learning_rate": 5.199417228191586e-05,
+      "loss": 0.084,
+      "step": 9940
+    },
+    {
+      "epoch": 2.693557119653492,
+      "grad_norm": 0.3058319687843323,
+      "learning_rate": 5.153888180659261e-05,
+      "loss": 0.094,
+      "step": 9950
+    },
+    {
+      "epoch": 2.6962642122360583,
+      "grad_norm": 0.31596943736076355,
+      "learning_rate": 5.108359133126935e-05,
+      "loss": 0.0641,
+      "step": 9960
+    },
+    {
+      "epoch": 2.698971304818625,
+      "grad_norm": 0.13064707815647125,
+      "learning_rate": 5.0628300855946096e-05,
+      "loss": 0.1015,
+      "step": 9970
+    },
+    {
+      "epoch": 2.7016783974011913,
+      "grad_norm": 0.17900826036930084,
+      "learning_rate": 5.0173010380622837e-05,
+      "loss": 0.1025,
+      "step": 9980
+    },
+    {
+      "epoch": 2.7043854899837574,
+      "grad_norm": 0.6242370009422302,
+      "learning_rate": 4.971771990529958e-05,
+      "loss": 0.1105,
+      "step": 9990
+    },
+    {
+      "epoch": 2.707092582566324,
+      "grad_norm": 0.12237600237131119,
+      "learning_rate": 4.9262429429976323e-05,
+      "loss": 0.101,
+      "step": 10000
+    },
+    {
+      "epoch": 2.707092582566324,
+      "eval_loss": 0.11815793067216873,
+      "eval_runtime": 39.0482,
+      "eval_samples_per_second": 12.805,
+      "eval_steps_per_second": 0.82,
+      "step": 10000
+    },
+    {
+      "epoch": 2.70979967514889,
+      "grad_norm": 0.20948220789432526,
+      "learning_rate": 4.880713895465307e-05,
+      "loss": 0.0791,
+      "step": 10010
+    },
+    {
+      "epoch": 2.7125067677314565,
+      "grad_norm": 0.10292834043502808,
+      "learning_rate": 4.835184847932982e-05,
+      "loss": 0.0876,
+      "step": 10020
+    },
+    {
+      "epoch": 2.715213860314023,
+      "grad_norm": 0.2232556790113449,
+      "learning_rate": 4.789655800400656e-05,
+      "loss": 0.0959,
+      "step": 10030
+    },
+    {
+      "epoch": 2.717920952896589,
+      "grad_norm": 0.31674426794052124,
+      "learning_rate": 4.7441267528683304e-05,
+      "loss": 0.0832,
+      "step": 10040
+    },
+    {
+      "epoch": 2.720628045479155,
+      "grad_norm": 0.46846631169319153,
+      "learning_rate": 4.6985977053360044e-05,
+      "loss": 0.0966,
+      "step": 10050
+    },
+    {
+      "epoch": 2.7233351380617217,
+      "grad_norm": 0.27506813406944275,
+      "learning_rate": 4.653068657803679e-05,
+      "loss": 0.0938,
+      "step": 10060
+    },
+    {
+      "epoch": 2.726042230644288,
+      "grad_norm": 0.2675800025463104,
+      "learning_rate": 4.607539610271353e-05,
+      "loss": 0.082,
+      "step": 10070
+    },
+    {
+      "epoch": 2.7287493232268543,
+      "grad_norm": 0.09031280130147934,
+      "learning_rate": 4.562010562739027e-05,
+      "loss": 0.0989,
+      "step": 10080
+    },
+    {
+      "epoch": 2.731456415809421,
+      "grad_norm": 0.23713286221027374,
+      "learning_rate": 4.516481515206702e-05,
+      "loss": 0.0817,
+      "step": 10090
+    },
+    {
+      "epoch": 2.734163508391987,
+      "grad_norm": 0.2575676143169403,
+      "learning_rate": 4.4709524676743765e-05,
+      "loss": 0.077,
+      "step": 10100
+    },
+    {
+      "epoch": 2.734163508391987,
+      "eval_loss": 0.1171879693865776,
+      "eval_runtime": 41.2118,
+      "eval_samples_per_second": 12.132,
+      "eval_steps_per_second": 0.776,
+      "step": 10100
+    },
+    {
+      "epoch": 2.7368706009745534,
+      "grad_norm": 0.17856763303279877,
+      "learning_rate": 4.425423420142051e-05,
+      "loss": 0.1071,
+      "step": 10110
+    },
+    {
+      "epoch": 2.73957769355712,
+      "grad_norm": 0.3801266849040985,
+      "learning_rate": 4.379894372609725e-05,
+      "loss": 0.0872,
+      "step": 10120
+    },
+    {
+      "epoch": 2.742284786139686,
+      "grad_norm": 0.12164852768182755,
+      "learning_rate": 4.3343653250774e-05,
+      "loss": 0.0627,
+      "step": 10130
+    },
+    {
+      "epoch": 2.744991878722252,
+      "grad_norm": 0.1857558786869049,
+      "learning_rate": 4.288836277545074e-05,
+      "loss": 0.0705,
+      "step": 10140
+    },
+    {
+      "epoch": 2.7476989713048185,
+      "grad_norm": 0.1430027037858963,
+      "learning_rate": 4.243307230012748e-05,
+      "loss": 0.0885,
+      "step": 10150
+    },
+    {
+      "epoch": 2.750406063887385,
+      "grad_norm": 0.27870213985443115,
+      "learning_rate": 4.1977781824804226e-05,
+      "loss": 0.1018,
+      "step": 10160
+    },
+    {
+      "epoch": 2.753113156469951,
+      "grad_norm": 0.23771138489246368,
+      "learning_rate": 4.1522491349480966e-05,
+      "loss": 0.0694,
+      "step": 10170
+    },
+    {
+      "epoch": 2.7558202490525177,
+      "grad_norm": 0.6071820259094238,
+      "learning_rate": 4.106720087415771e-05,
+      "loss": 0.0891,
+      "step": 10180
+    },
+    {
+      "epoch": 2.7585273416350837,
+      "grad_norm": 0.33392077684402466,
+      "learning_rate": 4.061191039883446e-05,
+      "loss": 0.0982,
+      "step": 10190
+    },
+    {
+      "epoch": 2.7612344342176502,
+      "grad_norm": 0.07901237905025482,
+      "learning_rate": 4.0156619923511206e-05,
+      "loss": 0.0904,
+      "step": 10200
+    },
+    {
+      "epoch": 2.7612344342176502,
+      "eval_loss": 0.1186905950307846,
+      "eval_runtime": 39.4543,
+      "eval_samples_per_second": 12.673,
+      "eval_steps_per_second": 0.811,
+      "step": 10200
+    },
+    {
+      "epoch": 2.7639415268002168,
+      "grad_norm": 0.17425483465194702,
+      "learning_rate": 3.9701329448187946e-05,
+      "loss": 0.1105,
+      "step": 10210
+    },
+    {
+      "epoch": 2.766648619382783,
+      "grad_norm": 0.19784431159496307,
+      "learning_rate": 3.9246038972864686e-05,
+      "loss": 0.0876,
+      "step": 10220
+    },
+    {
+      "epoch": 2.7693557119653494,
+      "grad_norm": 0.5425503849983215,
+      "learning_rate": 3.879074849754143e-05,
+      "loss": 0.0788,
+      "step": 10230
+    },
+    {
+      "epoch": 2.7720628045479154,
+      "grad_norm": 0.23448367416858673,
+      "learning_rate": 3.833545802221817e-05,
+      "loss": 0.1153,
+      "step": 10240
+    },
+    {
+      "epoch": 2.774769897130482,
+      "grad_norm": 0.12311672419309616,
+      "learning_rate": 3.788016754689492e-05,
+      "loss": 0.0639,
+      "step": 10250
+    },
+    {
+      "epoch": 2.7774769897130485,
+      "grad_norm": 0.2795017659664154,
+      "learning_rate": 3.742487707157166e-05,
+      "loss": 0.0794,
+      "step": 10260
+    },
+    {
+      "epoch": 2.7801840822956145,
+      "grad_norm": 0.4258648455142975,
+      "learning_rate": 3.696958659624841e-05,
+      "loss": 0.0934,
+      "step": 10270
+    },
+    {
+      "epoch": 2.7828911748781806,
+      "grad_norm": 0.1813986748456955,
+      "learning_rate": 3.6514296120925154e-05,
+      "loss": 0.0889,
+      "step": 10280
+    },
+    {
+      "epoch": 2.785598267460747,
+      "grad_norm": 0.33424076437950134,
+      "learning_rate": 3.6059005645601894e-05,
+      "loss": 0.122,
+      "step": 10290
+    },
+    {
+      "epoch": 2.7883053600433136,
+      "grad_norm": 0.12661059200763702,
+      "learning_rate": 3.560371517027864e-05,
+      "loss": 0.0882,
+      "step": 10300
+    },
+    {
+      "epoch": 2.7883053600433136,
+      "eval_loss": 0.11712407320737839,
+      "eval_runtime": 40.6313,
+      "eval_samples_per_second": 12.306,
+      "eval_steps_per_second": 0.788,
+      "step": 10300
+    },
+    {
+      "epoch": 2.7910124526258797,
+      "grad_norm": 0.3188532292842865,
+      "learning_rate": 3.514842469495538e-05,
+      "loss": 0.1079,
+      "step": 10310
+    },
+    {
+      "epoch": 2.7937195452084462,
+      "grad_norm": 0.3484908938407898,
+      "learning_rate": 3.469313421963213e-05,
+      "loss": 0.0843,
+      "step": 10320
+    },
+    {
+      "epoch": 2.7964266377910123,
+      "grad_norm": 0.25738710165023804,
+      "learning_rate": 3.423784374430887e-05,
+      "loss": 0.084,
+      "step": 10330
+    },
+    {
+      "epoch": 2.799133730373579,
+      "grad_norm": 0.1297960728406906,
+      "learning_rate": 3.3782553268985615e-05,
+      "loss": 0.1046,
+      "step": 10340
+    },
+    {
+      "epoch": 2.8018408229561453,
+      "grad_norm": 0.413796603679657,
+      "learning_rate": 3.3327262793662355e-05,
+      "loss": 0.1023,
+      "step": 10350
+    },
+    {
+      "epoch": 2.8045479155387114,
+      "grad_norm": 0.06507105380296707,
+      "learning_rate": 3.28719723183391e-05,
+      "loss": 0.0762,
+      "step": 10360
+    },
+    {
+      "epoch": 2.8072550081212775,
+      "grad_norm": 0.1512298882007599,
+      "learning_rate": 3.241668184301585e-05,
+      "loss": 0.1036,
+      "step": 10370
+    },
+    {
+      "epoch": 2.809962100703844,
+      "grad_norm": 0.164729043841362,
+      "learning_rate": 3.196139136769259e-05,
+      "loss": 0.0859,
+      "step": 10380
+    },
+    {
+      "epoch": 2.8126691932864105,
+      "grad_norm": 0.21104639768600464,
+      "learning_rate": 3.1506100892369335e-05,
+      "loss": 0.0678,
+      "step": 10390
+    },
+    {
+      "epoch": 2.8153762858689766,
+      "grad_norm": 0.20643286406993866,
+      "learning_rate": 3.1050810417046075e-05,
+      "loss": 0.0863,
+      "step": 10400
+    },
+    {
+      "epoch": 2.8153762858689766,
+      "eval_loss": 0.11635182797908783,
+      "eval_runtime": 40.03,
+      "eval_samples_per_second": 12.491,
+      "eval_steps_per_second": 0.799,
+      "step": 10400
+    },
+    {
+      "epoch": 2.818083378451543,
+      "grad_norm": 0.34288492798805237,
+      "learning_rate": 3.059551994172282e-05,
+      "loss": 0.112,
+      "step": 10410
+    },
+    {
+      "epoch": 2.820790471034109,
+      "grad_norm": 0.16786547005176544,
+      "learning_rate": 3.0140229466399562e-05,
+      "loss": 0.0613,
+      "step": 10420
+    },
+    {
+      "epoch": 2.8234975636166757,
+      "grad_norm": 0.26734793186187744,
+      "learning_rate": 2.9684938991076306e-05,
+      "loss": 0.1045,
+      "step": 10430
+    },
+    {
+      "epoch": 2.826204656199242,
+      "grad_norm": 0.18626269698143005,
+      "learning_rate": 2.9229648515753052e-05,
+      "loss": 0.1042,
+      "step": 10440
+    },
+    {
+      "epoch": 2.8289117487818083,
+      "grad_norm": 0.27336782217025757,
+      "learning_rate": 2.8774358040429796e-05,
+      "loss": 0.0834,
+      "step": 10450
+    },
+    {
+      "epoch": 2.831618841364375,
+      "grad_norm": 0.14987875521183014,
+      "learning_rate": 2.831906756510654e-05,
+      "loss": 0.0875,
+      "step": 10460
+    },
+    {
+      "epoch": 2.834325933946941,
+      "grad_norm": 0.1719442754983902,
+      "learning_rate": 2.7863777089783283e-05,
+      "loss": 0.0878,
+      "step": 10470
+    },
+    {
+      "epoch": 2.8370330265295074,
+      "grad_norm": 0.13460715115070343,
+      "learning_rate": 2.7408486614460023e-05,
+      "loss": 0.0759,
+      "step": 10480
+    },
+    {
+      "epoch": 2.8397401191120735,
+      "grad_norm": 0.1445351541042328,
+      "learning_rate": 2.695319613913677e-05,
+      "loss": 0.0939,
+      "step": 10490
+    },
+    {
+      "epoch": 2.84244721169464,
+      "grad_norm": 0.06805498152971268,
+      "learning_rate": 2.6497905663813513e-05,
+      "loss": 0.0758,
+      "step": 10500
+    },
+    {
+      "epoch": 2.84244721169464,
+      "eval_loss": 0.1177913248538971,
+      "eval_runtime": 43.0157,
+      "eval_samples_per_second": 11.624,
+      "eval_steps_per_second": 0.744,
+      "step": 10500
+    },
+    {
+      "epoch": 2.845154304277206,
+      "grad_norm": 0.31017813086509705,
+      "learning_rate": 2.6042615188490257e-05,
+      "loss": 0.0955,
+      "step": 10510
+    },
+    {
+      "epoch": 2.8478613968597726,
+      "grad_norm": 0.2953559160232544,
+      "learning_rate": 2.5587324713167e-05,
+      "loss": 0.0955,
+      "step": 10520
+    },
+    {
+      "epoch": 2.850568489442339,
+      "grad_norm": 0.29184433817863464,
+      "learning_rate": 2.5132034237843747e-05,
+      "loss": 0.0753,
+      "step": 10530
+    },
+    {
+      "epoch": 2.853275582024905,
+      "grad_norm": 0.12063843011856079,
+      "learning_rate": 2.467674376252049e-05,
+      "loss": 0.078,
+      "step": 10540
+    },
+    {
+      "epoch": 2.8559826746074717,
+      "grad_norm": 0.14911547303199768,
+      "learning_rate": 2.4221453287197234e-05,
+      "loss": 0.0636,
+      "step": 10550
+    },
+    {
+      "epoch": 2.8586897671900378,
+      "grad_norm": 0.3377572298049927,
+      "learning_rate": 2.3766162811873974e-05,
+      "loss": 0.081,
+      "step": 10560
+    },
+    {
+      "epoch": 2.8613968597726043,
+      "grad_norm": 0.5010014772415161,
+      "learning_rate": 2.3310872336550717e-05,
+      "loss": 0.0923,
+      "step": 10570
+    },
+    {
+      "epoch": 2.864103952355171,
+      "grad_norm": 0.37955665588378906,
+      "learning_rate": 2.2855581861227464e-05,
+      "loss": 0.1156,
+      "step": 10580
+    },
+    {
+      "epoch": 2.866811044937737,
+      "grad_norm": 0.30390897393226624,
+      "learning_rate": 2.2400291385904208e-05,
+      "loss": 0.101,
+      "step": 10590
+    },
+    {
+      "epoch": 2.869518137520303,
+      "grad_norm": 0.32106971740722656,
+      "learning_rate": 2.194500091058095e-05,
+      "loss": 0.092,
+      "step": 10600
+    },
+    {
+      "epoch": 2.869518137520303,
+      "eval_loss": 0.11648274213075638,
+      "eval_runtime": 42.4431,
+      "eval_samples_per_second": 11.78,
+      "eval_steps_per_second": 0.754,
+      "step": 10600
+    },
+    {
+      "epoch": 2.8722252301028695,
+      "grad_norm": 0.14484931528568268,
+      "learning_rate": 2.1489710435257695e-05,
+      "loss": 0.0675,
+      "step": 10610
+    },
+    {
+      "epoch": 2.874932322685436,
+      "grad_norm": 0.2799387276172638,
+      "learning_rate": 2.103441995993444e-05,
+      "loss": 0.1019,
+      "step": 10620
+    },
+    {
+      "epoch": 2.877639415268002,
+      "grad_norm": 0.2622615396976471,
+      "learning_rate": 2.057912948461118e-05,
+      "loss": 0.0902,
+      "step": 10630
+    },
+    {
+      "epoch": 2.8803465078505686,
+      "grad_norm": 0.2266823649406433,
+      "learning_rate": 2.0123839009287925e-05,
+      "loss": 0.0902,
+      "step": 10640
+    },
+    {
+      "epoch": 2.8830536004331346,
+      "grad_norm": 0.191898375749588,
+      "learning_rate": 1.966854853396467e-05,
+      "loss": 0.0757,
+      "step": 10650
+    },
+    {
+      "epoch": 2.885760693015701,
+      "grad_norm": 0.22966520488262177,
+      "learning_rate": 1.9213258058641412e-05,
+      "loss": 0.0881,
+      "step": 10660
+    },
+    {
+      "epoch": 2.8884677855982677,
+      "grad_norm": 0.4418594539165497,
+      "learning_rate": 1.875796758331816e-05,
+      "loss": 0.1102,
+      "step": 10670
+    },
+    {
+      "epoch": 2.8911748781808337,
+      "grad_norm": 0.1618470698595047,
+      "learning_rate": 1.8302677107994902e-05,
+      "loss": 0.0795,
+      "step": 10680
+    },
+    {
+      "epoch": 2.8938819707634003,
+      "grad_norm": 0.1667235791683197,
+      "learning_rate": 1.7847386632671646e-05,
+      "loss": 0.0905,
+      "step": 10690
+    },
+    {
+      "epoch": 2.8965890633459663,
+      "grad_norm": 0.1314006894826889,
+      "learning_rate": 1.7392096157348386e-05,
+      "loss": 0.0969,
+      "step": 10700
+    },
+    {
+      "epoch": 2.8965890633459663,
+      "eval_loss": 0.11577699333429337,
+      "eval_runtime": 43.5168,
+      "eval_samples_per_second": 11.49,
+      "eval_steps_per_second": 0.735,
+      "step": 10700
+    },
+    {
+      "epoch": 2.899296155928533,
+      "grad_norm": 0.3928311765193939,
+      "learning_rate": 1.6936805682025133e-05,
+      "loss": 0.1017,
+      "step": 10710
+    },
+    {
+      "epoch": 2.902003248511099,
+      "grad_norm": 0.2540978491306305,
+      "learning_rate": 1.6481515206701876e-05,
+      "loss": 0.0758,
+      "step": 10720
+    },
+    {
+      "epoch": 2.9047103410936654,
+      "grad_norm": 0.3657947778701782,
+      "learning_rate": 1.602622473137862e-05,
+      "loss": 0.1291,
+      "step": 10730
+    },
+    {
+      "epoch": 2.9074174336762315,
+      "grad_norm": 0.17771603167057037,
+      "learning_rate": 1.5570934256055363e-05,
+      "loss": 0.1199,
+      "step": 10740
+    },
+    {
+      "epoch": 2.910124526258798,
+      "grad_norm": 0.4251940846443176,
+      "learning_rate": 1.5115643780732108e-05,
+      "loss": 0.1068,
+      "step": 10750
+    },
+    {
+      "epoch": 2.9128316188413645,
+      "grad_norm": 0.13539324700832367,
+      "learning_rate": 1.466035330540885e-05,
+      "loss": 0.0882,
+      "step": 10760
+    },
+    {
+      "epoch": 2.9155387114239306,
+      "grad_norm": 0.502238392829895,
+      "learning_rate": 1.4205062830085595e-05,
+      "loss": 0.0807,
+      "step": 10770
+    },
+    {
+      "epoch": 2.918245804006497,
+      "grad_norm": 0.41051560640335083,
+      "learning_rate": 1.3749772354762338e-05,
+      "loss": 0.0879,
+      "step": 10780
+    },
+    {
+      "epoch": 2.920952896589063,
+      "grad_norm": 0.2090943455696106,
+      "learning_rate": 1.3294481879439084e-05,
+      "loss": 0.1238,
+      "step": 10790
+    },
+    {
+      "epoch": 2.9236599891716297,
+      "grad_norm": 0.2868696451187134,
+      "learning_rate": 1.2839191404115825e-05,
+      "loss": 0.0974,
+      "step": 10800
+    },
+    {
+      "epoch": 2.9236599891716297,
+      "eval_loss": 0.11386918276548386,
+      "eval_runtime": 42.0058,
+      "eval_samples_per_second": 11.903,
+      "eval_steps_per_second": 0.762,
+      "step": 10800
+    },
+    {
+      "epoch": 2.9263670817541962,
+      "grad_norm": 0.3918515145778656,
+      "learning_rate": 1.238390092879257e-05,
+      "loss": 0.0752,
+      "step": 10810
+    },
+    {
+      "epoch": 2.9290741743367623,
+      "grad_norm": 0.27518269419670105,
+      "learning_rate": 1.1928610453469314e-05,
+      "loss": 0.1043,
+      "step": 10820
+    },
+    {
+      "epoch": 2.9317812669193284,
+      "grad_norm": 0.1713918149471283,
+      "learning_rate": 1.1473319978146057e-05,
+      "loss": 0.0857,
+      "step": 10830
+    },
+    {
+      "epoch": 2.934488359501895,
+      "grad_norm": 0.41794198751449585,
+      "learning_rate": 1.1018029502822801e-05,
+      "loss": 0.1127,
+      "step": 10840
+    },
+    {
+      "epoch": 2.9371954520844614,
+      "grad_norm": 0.3355647325515747,
+      "learning_rate": 1.0562739027499544e-05,
+      "loss": 0.1003,
+      "step": 10850
+    },
+    {
+      "epoch": 2.9399025446670275,
+      "grad_norm": 0.36736515164375305,
+      "learning_rate": 1.010744855217629e-05,
+      "loss": 0.1097,
+      "step": 10860
+    },
+    {
+      "epoch": 2.942609637249594,
+      "grad_norm": 0.12973126769065857,
+      "learning_rate": 9.652158076853031e-06,
+      "loss": 0.1008,
+      "step": 10870
+    },
+    {
+      "epoch": 2.94531672983216,
+      "grad_norm": 0.1610291302204132,
+      "learning_rate": 9.196867601529776e-06,
+      "loss": 0.0999,
+      "step": 10880
+    },
+    {
+      "epoch": 2.9480238224147266,
+      "grad_norm": 0.4387446343898773,
+      "learning_rate": 8.74157712620652e-06,
+      "loss": 0.0968,
+      "step": 10890
+    },
+    {
+      "epoch": 2.950730914997293,
+      "grad_norm": 0.35354000329971313,
+      "learning_rate": 8.286286650883265e-06,
+      "loss": 0.0801,
+      "step": 10900
+    },
+    {
+      "epoch": 2.950730914997293,
+      "eval_loss": 0.11372293531894684,
+      "eval_runtime": 42.3374,
+      "eval_samples_per_second": 11.81,
+      "eval_steps_per_second": 0.756,
+      "step": 10900
+    },
+    {
+      "epoch": 2.953438007579859,
+      "grad_norm": 0.48929524421691895,
+      "learning_rate": 7.830996175560007e-06,
+      "loss": 0.0807,
+      "step": 10910
+    },
+    {
+      "epoch": 2.9561451001624257,
+      "grad_norm": 0.15634261071681976,
+      "learning_rate": 7.375705700236751e-06,
+      "loss": 0.1112,
+      "step": 10920
+    },
+    {
+      "epoch": 2.958852192744992,
+      "grad_norm": 0.21457576751708984,
+      "learning_rate": 6.920415224913495e-06,
+      "loss": 0.1149,
+      "step": 10930
+    },
+    {
+      "epoch": 2.9615592853275583,
+      "grad_norm": 0.31956467032432556,
+      "learning_rate": 6.465124749590239e-06,
+      "loss": 0.0728,
+      "step": 10940
+    },
+    {
+      "epoch": 2.9642663779101244,
+      "grad_norm": 0.21584530174732208,
+      "learning_rate": 6.009834274266983e-06,
+      "loss": 0.0862,
+      "step": 10950
+    },
+    {
+      "epoch": 2.966973470492691,
+      "grad_norm": 0.28141820430755615,
+      "learning_rate": 5.554543798943726e-06,
+      "loss": 0.0994,
+      "step": 10960
+    },
+    {
+      "epoch": 2.969680563075257,
+      "grad_norm": 0.11748827993869781,
+      "learning_rate": 5.09925332362047e-06,
+      "loss": 0.1254,
+      "step": 10970
+    },
+    {
+      "epoch": 2.9723876556578235,
+      "grad_norm": 0.24274790287017822,
+      "learning_rate": 4.6439628482972135e-06,
+      "loss": 0.0928,
+      "step": 10980
+    },
+    {
+      "epoch": 2.97509474824039,
+      "grad_norm": 0.2196180820465088,
+      "learning_rate": 4.188672372973958e-06,
+      "loss": 0.0785,
+      "step": 10990
+    },
+    {
+      "epoch": 2.977801840822956,
+      "grad_norm": 0.14466066658496857,
+      "learning_rate": 3.7333818976507013e-06,
+      "loss": 0.0821,
+      "step": 11000
+    },
+    {
+      "epoch": 2.977801840822956,
+      "eval_loss": 0.11507652699947357,
+      "eval_runtime": 40.059,
+      "eval_samples_per_second": 12.482,
+      "eval_steps_per_second": 0.799,
+      "step": 11000
+    },
+    {
+      "epoch": 2.9805089334055226,
+      "grad_norm": 0.48471832275390625,
+      "learning_rate": 3.278091422327445e-06,
+      "loss": 0.0981,
+      "step": 11010
+    },
+    {
+      "epoch": 2.9832160259880887,
+      "grad_norm": 0.2775181233882904,
+      "learning_rate": 2.8228009470041886e-06,
+      "loss": 0.0903,
+      "step": 11020
+    },
+    {
+      "epoch": 2.985923118570655,
+      "grad_norm": 0.1871793568134308,
+      "learning_rate": 2.3675104716809325e-06,
+      "loss": 0.1036,
+      "step": 11030
+    },
+    {
+      "epoch": 2.9886302111532217,
+      "grad_norm": 0.39151501655578613,
+      "learning_rate": 1.9122199963576764e-06,
+      "loss": 0.1124,
+      "step": 11040
+    },
+    {
+      "epoch": 2.9913373037357878,
+      "grad_norm": 0.16927284002304077,
+      "learning_rate": 1.4569295210344198e-06,
+      "loss": 0.0991,
+      "step": 11050
+    },
+    {
+      "epoch": 2.994044396318354,
+      "grad_norm": 0.19268180429935455,
+      "learning_rate": 1.0016390457111637e-06,
+      "loss": 0.1093,
+      "step": 11060
+    },
+    {
+      "epoch": 2.9967514889009204,
+      "grad_norm": 0.2451864778995514,
+      "learning_rate": 5.463485703879075e-07,
+      "loss": 0.0745,
+      "step": 11070
+    },
+    {
+      "epoch": 2.999458581483487,
+      "grad_norm": 0.10636322945356369,
+      "learning_rate": 9.105809506465124e-08,
+      "loss": 0.0863,
+      "step": 11080
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 11082,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.041237140368352e+17,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}