diff --git "a/checkpoint-29000/trainer_state.json" "b/checkpoint-29000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-29000/trainer_state.json"
@@ -0,0 +1,4383 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 446.15384615384613,
+  "eval_steps": 100,
+  "global_step": 29000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 47.11796188354492,
+      "learning_rate": 9.990900000000001e-06,
+      "loss": 3.6644,
+      "step": 100
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "eval_loss": 2.4919605255126953,
+      "eval_runtime": 12.5517,
+      "eval_samples_per_second": 10.437,
+      "eval_steps_per_second": 1.354,
+      "step": 100
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 31.28130531311035,
+      "learning_rate": 9.980900000000001e-06,
+      "loss": 2.2347,
+      "step": 200
+    },
+    {
+      "epoch": 3.076923076923077,
+      "eval_loss": 2.156316041946411,
+      "eval_runtime": 11.1792,
+      "eval_samples_per_second": 11.718,
+      "eval_steps_per_second": 1.521,
+      "step": 200
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 39.616390228271484,
+      "learning_rate": 9.970900000000001e-06,
+      "loss": 2.0254,
+      "step": 300
+    },
+    {
+      "epoch": 4.615384615384615,
+      "eval_loss": 2.024153709411621,
+      "eval_runtime": 11.1934,
+      "eval_samples_per_second": 11.703,
+      "eval_steps_per_second": 1.519,
+      "step": 300
+    },
+    {
+      "epoch": 6.153846153846154,
+      "grad_norm": 28.285825729370117,
+      "learning_rate": 9.960900000000001e-06,
+      "loss": 1.9361,
+      "step": 400
+    },
+    {
+      "epoch": 6.153846153846154,
+      "eval_loss": 1.9094743728637695,
+      "eval_runtime": 11.3855,
+      "eval_samples_per_second": 11.506,
+      "eval_steps_per_second": 1.493,
+      "step": 400
+    },
+    {
+      "epoch": 7.6923076923076925,
+      "grad_norm": 34.14302062988281,
+      "learning_rate": 9.950900000000002e-06,
+      "loss": 1.8531,
+      "step": 500
+    },
+    {
+      "epoch": 7.6923076923076925,
+      "eval_loss": 1.8729331493377686,
+      "eval_runtime": 11.2935,
+      "eval_samples_per_second": 11.6,
+      "eval_steps_per_second": 1.505,
+      "step": 500
+    },
+    {
+      "epoch": 9.23076923076923,
+      "grad_norm": 39.09531784057617,
+      "learning_rate": 9.940900000000002e-06,
+      "loss": 1.7669,
+      "step": 600
+    },
+    {
+      "epoch": 9.23076923076923,
+      "eval_loss": 1.831756830215454,
+      "eval_runtime": 11.0535,
+      "eval_samples_per_second": 11.851,
+      "eval_steps_per_second": 1.538,
+      "step": 600
+    },
+    {
+      "epoch": 10.76923076923077,
+      "grad_norm": 93.24444580078125,
+      "learning_rate": 9.930900000000002e-06,
+      "loss": 1.7518,
+      "step": 700
+    },
+    {
+      "epoch": 10.76923076923077,
+      "eval_loss": 1.7832175493240356,
+      "eval_runtime": 11.1684,
+      "eval_samples_per_second": 11.729,
+      "eval_steps_per_second": 1.522,
+      "step": 700
+    },
+    {
+      "epoch": 12.307692307692308,
+      "grad_norm": 32.21013641357422,
+      "learning_rate": 9.920900000000002e-06,
+      "loss": 1.7149,
+      "step": 800
+    },
+    {
+      "epoch": 12.307692307692308,
+      "eval_loss": 1.7581098079681396,
+      "eval_runtime": 11.101,
+      "eval_samples_per_second": 11.801,
+      "eval_steps_per_second": 1.531,
+      "step": 800
+    },
+    {
+      "epoch": 13.846153846153847,
+      "grad_norm": 59.90657043457031,
+      "learning_rate": 9.9109e-06,
+      "loss": 1.6734,
+      "step": 900
+    },
+    {
+      "epoch": 13.846153846153847,
+      "eval_loss": 1.7163844108581543,
+      "eval_runtime": 11.1167,
+      "eval_samples_per_second": 11.784,
+      "eval_steps_per_second": 1.529,
+      "step": 900
+    },
+    {
+      "epoch": 15.384615384615385,
+      "grad_norm": 20.61592674255371,
+      "learning_rate": 9.9009e-06,
+      "loss": 1.6612,
+      "step": 1000
+    },
+    {
+      "epoch": 15.384615384615385,
+      "eval_loss": 1.6949567794799805,
+      "eval_runtime": 11.0663,
+      "eval_samples_per_second": 11.838,
+      "eval_steps_per_second": 1.536,
+      "step": 1000
+    },
+    {
+      "epoch": 16.923076923076923,
+      "grad_norm": 17.60099220275879,
+      "learning_rate": 9.8909e-06,
+      "loss": 1.6199,
+      "step": 1100
+    },
+    {
+      "epoch": 16.923076923076923,
+      "eval_loss": 1.6769332885742188,
+      "eval_runtime": 11.0531,
+      "eval_samples_per_second": 11.852,
+      "eval_steps_per_second": 1.538,
+      "step": 1100
+    },
+    {
+      "epoch": 18.46153846153846,
+      "grad_norm": 20.802692413330078,
+      "learning_rate": 9.8809e-06,
+      "loss": 1.6008,
+      "step": 1200
+    },
+    {
+      "epoch": 18.46153846153846,
+      "eval_loss": 1.6524990797042847,
+      "eval_runtime": 11.0831,
+      "eval_samples_per_second": 11.82,
+      "eval_steps_per_second": 1.534,
+      "step": 1200
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 21.809823989868164,
+      "learning_rate": 9.8709e-06,
+      "loss": 1.5812,
+      "step": 1300
+    },
+    {
+      "epoch": 20.0,
+      "eval_loss": 1.6428295373916626,
+      "eval_runtime": 11.1093,
+      "eval_samples_per_second": 11.792,
+      "eval_steps_per_second": 1.53,
+      "step": 1300
+    },
+    {
+      "epoch": 21.53846153846154,
+      "grad_norm": 46.8908576965332,
+      "learning_rate": 9.8609e-06,
+      "loss": 1.5419,
+      "step": 1400
+    },
+    {
+      "epoch": 21.53846153846154,
+      "eval_loss": 1.6006404161453247,
+      "eval_runtime": 11.2393,
+      "eval_samples_per_second": 11.655,
+      "eval_steps_per_second": 1.513,
+      "step": 1400
+    },
+    {
+      "epoch": 23.076923076923077,
+      "grad_norm": 27.15238380432129,
+      "learning_rate": 9.8509e-06,
+      "loss": 1.5374,
+      "step": 1500
+    },
+    {
+      "epoch": 23.076923076923077,
+      "eval_loss": 1.5862094163894653,
+      "eval_runtime": 11.1815,
+      "eval_samples_per_second": 11.716,
+      "eval_steps_per_second": 1.52,
+      "step": 1500
+    },
+    {
+      "epoch": 24.615384615384617,
+      "grad_norm": 40.26778030395508,
+      "learning_rate": 9.840900000000001e-06,
+      "loss": 1.4923,
+      "step": 1600
+    },
+    {
+      "epoch": 24.615384615384617,
+      "eval_loss": 1.576373815536499,
+      "eval_runtime": 11.1215,
+      "eval_samples_per_second": 11.779,
+      "eval_steps_per_second": 1.529,
+      "step": 1600
+    },
+    {
+      "epoch": 26.153846153846153,
+      "grad_norm": 35.266971588134766,
+      "learning_rate": 9.830900000000001e-06,
+      "loss": 1.4989,
+      "step": 1700
+    },
+    {
+      "epoch": 26.153846153846153,
+      "eval_loss": 1.5671430826187134,
+      "eval_runtime": 11.1873,
+      "eval_samples_per_second": 11.71,
+      "eval_steps_per_second": 1.52,
+      "step": 1700
+    },
+    {
+      "epoch": 27.692307692307693,
+      "grad_norm": 26.813480377197266,
+      "learning_rate": 9.820900000000001e-06,
+      "loss": 1.4711,
+      "step": 1800
+    },
+    {
+      "epoch": 27.692307692307693,
+      "eval_loss": 1.522908329963684,
+      "eval_runtime": 11.2106,
+      "eval_samples_per_second": 11.685,
+      "eval_steps_per_second": 1.516,
+      "step": 1800
+    },
+    {
+      "epoch": 29.23076923076923,
+      "grad_norm": 24.576723098754883,
+      "learning_rate": 9.810900000000001e-06,
+      "loss": 1.4421,
+      "step": 1900
+    },
+    {
+      "epoch": 29.23076923076923,
+      "eval_loss": 1.5039104223251343,
+      "eval_runtime": 11.257,
+      "eval_samples_per_second": 11.637,
+      "eval_steps_per_second": 1.51,
+      "step": 1900
+    },
+    {
+      "epoch": 30.76923076923077,
+      "grad_norm": 28.480438232421875,
+      "learning_rate": 9.800900000000001e-06,
+      "loss": 1.4347,
+      "step": 2000
+    },
+    {
+      "epoch": 30.76923076923077,
+      "eval_loss": 1.5123459100723267,
+      "eval_runtime": 11.187,
+      "eval_samples_per_second": 11.71,
+      "eval_steps_per_second": 1.52,
+      "step": 2000
+    },
+    {
+      "epoch": 32.30769230769231,
+      "grad_norm": 56.582088470458984,
+      "learning_rate": 9.790900000000001e-06,
+      "loss": 1.4212,
+      "step": 2100
+    },
+    {
+      "epoch": 32.30769230769231,
+      "eval_loss": 1.481844425201416,
+      "eval_runtime": 11.2075,
+      "eval_samples_per_second": 11.689,
+      "eval_steps_per_second": 1.517,
+      "step": 2100
+    },
+    {
+      "epoch": 33.84615384615385,
+      "grad_norm": 38.5254020690918,
+      "learning_rate": 9.780900000000002e-06,
+      "loss": 1.3908,
+      "step": 2200
+    },
+    {
+      "epoch": 33.84615384615385,
+      "eval_loss": 1.4529048204421997,
+      "eval_runtime": 11.197,
+      "eval_samples_per_second": 11.7,
+      "eval_steps_per_second": 1.518,
+      "step": 2200
+    },
+    {
+      "epoch": 35.38461538461539,
+      "grad_norm": 44.74857711791992,
+      "learning_rate": 9.770900000000002e-06,
+      "loss": 1.3734,
+      "step": 2300
+    },
+    {
+      "epoch": 35.38461538461539,
+      "eval_loss": 1.4617427587509155,
+      "eval_runtime": 11.1235,
+      "eval_samples_per_second": 11.777,
+      "eval_steps_per_second": 1.528,
+      "step": 2300
+    },
+    {
+      "epoch": 36.92307692307692,
+      "grad_norm": 28.926782608032227,
+      "learning_rate": 9.760900000000002e-06,
+      "loss": 1.365,
+      "step": 2400
+    },
+    {
+      "epoch": 36.92307692307692,
+      "eval_loss": 1.4297789335250854,
+      "eval_runtime": 11.3007,
+      "eval_samples_per_second": 11.592,
+      "eval_steps_per_second": 1.504,
+      "step": 2400
+    },
+    {
+      "epoch": 38.46153846153846,
+      "grad_norm": 27.750213623046875,
+      "learning_rate": 9.7509e-06,
+      "loss": 1.3306,
+      "step": 2500
+    },
+    {
+      "epoch": 38.46153846153846,
+      "eval_loss": 1.4345914125442505,
+      "eval_runtime": 11.2795,
+      "eval_samples_per_second": 11.614,
+      "eval_steps_per_second": 1.507,
+      "step": 2500
+    },
+    {
+      "epoch": 40.0,
+      "grad_norm": 35.288352966308594,
+      "learning_rate": 9.7409e-06,
+      "loss": 1.3677,
+      "step": 2600
+    },
+    {
+      "epoch": 40.0,
+      "eval_loss": 1.447089433670044,
+      "eval_runtime": 11.1366,
+      "eval_samples_per_second": 11.763,
+      "eval_steps_per_second": 1.527,
+      "step": 2600
+    },
+    {
+      "epoch": 41.53846153846154,
+      "grad_norm": 65.49736022949219,
+      "learning_rate": 9.7309e-06,
+      "loss": 1.3453,
+      "step": 2700
+    },
+    {
+      "epoch": 41.53846153846154,
+      "eval_loss": 1.405300259590149,
+      "eval_runtime": 11.1549,
+      "eval_samples_per_second": 11.744,
+      "eval_steps_per_second": 1.524,
+      "step": 2700
+    },
+    {
+      "epoch": 43.07692307692308,
+      "grad_norm": 24.333518981933594,
+      "learning_rate": 9.7209e-06,
+      "loss": 1.3206,
+      "step": 2800
+    },
+    {
+      "epoch": 43.07692307692308,
+      "eval_loss": 1.4218717813491821,
+      "eval_runtime": 11.3113,
+      "eval_samples_per_second": 11.581,
+      "eval_steps_per_second": 1.503,
+      "step": 2800
+    },
+    {
+      "epoch": 44.61538461538461,
+      "grad_norm": 45.50777816772461,
+      "learning_rate": 9.7109e-06,
+      "loss": 1.3363,
+      "step": 2900
+    },
+    {
+      "epoch": 44.61538461538461,
+      "eval_loss": 1.4220006465911865,
+      "eval_runtime": 11.156,
+      "eval_samples_per_second": 11.743,
+      "eval_steps_per_second": 1.524,
+      "step": 2900
+    },
+    {
+      "epoch": 46.15384615384615,
+      "grad_norm": 25.898344039916992,
+      "learning_rate": 9.7009e-06,
+      "loss": 1.2995,
+      "step": 3000
+    },
+    {
+      "epoch": 46.15384615384615,
+      "eval_loss": 1.3942431211471558,
+      "eval_runtime": 11.2282,
+      "eval_samples_per_second": 11.667,
+      "eval_steps_per_second": 1.514,
+      "step": 3000
+    },
+    {
+      "epoch": 47.69230769230769,
+      "grad_norm": 56.0889892578125,
+      "learning_rate": 9.6909e-06,
+      "loss": 1.2994,
+      "step": 3100
+    },
+    {
+      "epoch": 47.69230769230769,
+      "eval_loss": 1.3970586061477661,
+      "eval_runtime": 11.1682,
+      "eval_samples_per_second": 11.73,
+      "eval_steps_per_second": 1.522,
+      "step": 3100
+    },
+    {
+      "epoch": 49.23076923076923,
+      "grad_norm": 58.10311508178711,
+      "learning_rate": 9.6809e-06,
+      "loss": 1.2761,
+      "step": 3200
+    },
+    {
+      "epoch": 49.23076923076923,
+      "eval_loss": 1.390371561050415,
+      "eval_runtime": 11.2406,
+      "eval_samples_per_second": 11.654,
+      "eval_steps_per_second": 1.512,
+      "step": 3200
+    },
+    {
+      "epoch": 50.76923076923077,
+      "grad_norm": 17.050870895385742,
+      "learning_rate": 9.670900000000001e-06,
+      "loss": 1.2712,
+      "step": 3300
+    },
+    {
+      "epoch": 50.76923076923077,
+      "eval_loss": 1.3936753273010254,
+      "eval_runtime": 11.2364,
+      "eval_samples_per_second": 11.659,
+      "eval_steps_per_second": 1.513,
+      "step": 3300
+    },
+    {
+      "epoch": 52.30769230769231,
+      "grad_norm": 232.21804809570312,
+      "learning_rate": 9.660900000000001e-06,
+      "loss": 1.262,
+      "step": 3400
+    },
+    {
+      "epoch": 52.30769230769231,
+      "eval_loss": 1.4037091732025146,
+      "eval_runtime": 11.2231,
+      "eval_samples_per_second": 11.672,
+      "eval_steps_per_second": 1.515,
+      "step": 3400
+    },
+    {
+      "epoch": 53.84615384615385,
+      "grad_norm": 35.11832046508789,
+      "learning_rate": 9.650900000000001e-06,
+      "loss": 1.2788,
+      "step": 3500
+    },
+    {
+      "epoch": 53.84615384615385,
+      "eval_loss": 1.3545280694961548,
+      "eval_runtime": 11.1609,
+      "eval_samples_per_second": 11.737,
+      "eval_steps_per_second": 1.523,
+      "step": 3500
+    },
+    {
+      "epoch": 55.38461538461539,
+      "grad_norm": 27.077022552490234,
+      "learning_rate": 9.640900000000001e-06,
+      "loss": 1.2711,
+      "step": 3600
+    },
+    {
+      "epoch": 55.38461538461539,
+      "eval_loss": 1.3528518676757812,
+      "eval_runtime": 11.2852,
+      "eval_samples_per_second": 11.608,
+      "eval_steps_per_second": 1.506,
+      "step": 3600
+    },
+    {
+      "epoch": 56.92307692307692,
+      "grad_norm": 46.97712326049805,
+      "learning_rate": 9.630900000000001e-06,
+      "loss": 1.2492,
+      "step": 3700
+    },
+    {
+      "epoch": 56.92307692307692,
+      "eval_loss": 1.3534098863601685,
+      "eval_runtime": 11.2368,
+      "eval_samples_per_second": 11.658,
+      "eval_steps_per_second": 1.513,
+      "step": 3700
+    },
+    {
+      "epoch": 58.46153846153846,
+      "grad_norm": 42.06857681274414,
+      "learning_rate": 9.620900000000001e-06,
+      "loss": 1.2506,
+      "step": 3800
+    },
+    {
+      "epoch": 58.46153846153846,
+      "eval_loss": 1.3613977432250977,
+      "eval_runtime": 11.2206,
+      "eval_samples_per_second": 11.675,
+      "eval_steps_per_second": 1.515,
+      "step": 3800
+    },
+    {
+      "epoch": 60.0,
+      "grad_norm": 19.298952102661133,
+      "learning_rate": 9.610900000000001e-06,
+      "loss": 1.2201,
+      "step": 3900
+    },
+    {
+      "epoch": 60.0,
+      "eval_loss": 1.3586474657058716,
+      "eval_runtime": 11.2045,
+      "eval_samples_per_second": 11.692,
+      "eval_steps_per_second": 1.517,
+      "step": 3900
+    },
+    {
+      "epoch": 61.53846153846154,
+      "grad_norm": 30.0198974609375,
+      "learning_rate": 9.600900000000002e-06,
+      "loss": 1.2086,
+      "step": 4000
+    },
+    {
+      "epoch": 61.53846153846154,
+      "eval_loss": 1.3304755687713623,
+      "eval_runtime": 11.193,
+      "eval_samples_per_second": 11.704,
+      "eval_steps_per_second": 1.519,
+      "step": 4000
+    },
+    {
+      "epoch": 63.07692307692308,
+      "grad_norm": 37.59902572631836,
+      "learning_rate": 9.5909e-06,
+      "loss": 1.2375,
+      "step": 4100
+    },
+    {
+      "epoch": 63.07692307692308,
+      "eval_loss": 1.331407904624939,
+      "eval_runtime": 10.6714,
+      "eval_samples_per_second": 12.276,
+      "eval_steps_per_second": 1.593,
+      "step": 4100
+    },
+    {
+      "epoch": 64.61538461538461,
+      "grad_norm": 36.82079315185547,
+      "learning_rate": 9.5809e-06,
+      "loss": 1.2148,
+      "step": 4200
+    },
+    {
+      "epoch": 64.61538461538461,
+      "eval_loss": 1.3441548347473145,
+      "eval_runtime": 10.7472,
+      "eval_samples_per_second": 12.189,
+      "eval_steps_per_second": 1.582,
+      "step": 4200
+    },
+    {
+      "epoch": 66.15384615384616,
+      "grad_norm": 30.974130630493164,
+      "learning_rate": 9.5709e-06,
+      "loss": 1.197,
+      "step": 4300
+    },
+    {
+      "epoch": 66.15384615384616,
+      "eval_loss": 1.34512197971344,
+      "eval_runtime": 10.7398,
+      "eval_samples_per_second": 12.198,
+      "eval_steps_per_second": 1.583,
+      "step": 4300
+    },
+    {
+      "epoch": 67.6923076923077,
+      "grad_norm": 20.45345115661621,
+      "learning_rate": 9.5609e-06,
+      "loss": 1.2361,
+      "step": 4400
+    },
+    {
+      "epoch": 67.6923076923077,
+      "eval_loss": 1.3371080160140991,
+      "eval_runtime": 10.654,
+      "eval_samples_per_second": 12.296,
+      "eval_steps_per_second": 1.596,
+      "step": 4400
+    },
+    {
+      "epoch": 69.23076923076923,
+      "grad_norm": 19.758630752563477,
+      "learning_rate": 9.5509e-06,
+      "loss": 1.2001,
+      "step": 4500
+    },
+    {
+      "epoch": 69.23076923076923,
+      "eval_loss": 1.3270760774612427,
+      "eval_runtime": 10.6928,
+      "eval_samples_per_second": 12.251,
+      "eval_steps_per_second": 1.59,
+      "step": 4500
+    },
+    {
+      "epoch": 70.76923076923077,
+      "grad_norm": 45.2899055480957,
+      "learning_rate": 9.5409e-06,
+      "loss": 1.192,
+      "step": 4600
+    },
+    {
+      "epoch": 70.76923076923077,
+      "eval_loss": 1.3184590339660645,
+      "eval_runtime": 10.7154,
+      "eval_samples_per_second": 12.225,
+      "eval_steps_per_second": 1.586,
+      "step": 4600
+    },
+    {
+      "epoch": 72.3076923076923,
+      "grad_norm": 45.60734939575195,
+      "learning_rate": 9.5309e-06,
+      "loss": 1.2081,
+      "step": 4700
+    },
+    {
+      "epoch": 72.3076923076923,
+      "eval_loss": 1.3107666969299316,
+      "eval_runtime": 10.7029,
+      "eval_samples_per_second": 12.24,
+      "eval_steps_per_second": 1.588,
+      "step": 4700
+    },
+    {
+      "epoch": 73.84615384615384,
+      "grad_norm": 26.859603881835938,
+      "learning_rate": 9.5209e-06,
+      "loss": 1.1729,
+      "step": 4800
+    },
+    {
+      "epoch": 73.84615384615384,
+      "eval_loss": 1.310062289237976,
+      "eval_runtime": 10.7544,
+      "eval_samples_per_second": 12.181,
+      "eval_steps_per_second": 1.581,
+      "step": 4800
+    },
+    {
+      "epoch": 75.38461538461539,
+      "grad_norm": 19.90914535522461,
+      "learning_rate": 9.5109e-06,
+      "loss": 1.1899,
+      "step": 4900
+    },
+    {
+      "epoch": 75.38461538461539,
+      "eval_loss": 1.3038017749786377,
+      "eval_runtime": 10.6532,
+      "eval_samples_per_second": 12.297,
+      "eval_steps_per_second": 1.596,
+      "step": 4900
+    },
+    {
+      "epoch": 76.92307692307692,
+      "grad_norm": 22.629934310913086,
+      "learning_rate": 9.5009e-06,
+      "loss": 1.1875,
+      "step": 5000
+    },
+    {
+      "epoch": 76.92307692307692,
+      "eval_loss": 1.2785409688949585,
+      "eval_runtime": 10.6388,
+      "eval_samples_per_second": 12.313,
+      "eval_steps_per_second": 1.598,
+      "step": 5000
+    },
+    {
+      "epoch": 78.46153846153847,
+      "grad_norm": 45.462059020996094,
+      "learning_rate": 9.490900000000001e-06,
+      "loss": 1.1717,
+      "step": 5100
+    },
+    {
+      "epoch": 78.46153846153847,
+      "eval_loss": 1.278078317642212,
+      "eval_runtime": 10.9666,
+      "eval_samples_per_second": 11.945,
+      "eval_steps_per_second": 1.55,
+      "step": 5100
+    },
+    {
+      "epoch": 80.0,
+      "grad_norm": 34.85255432128906,
+      "learning_rate": 9.480900000000001e-06,
+      "loss": 1.1657,
+      "step": 5200
+    },
+    {
+      "epoch": 80.0,
+      "eval_loss": 1.2711185216903687,
+      "eval_runtime": 11.0211,
+      "eval_samples_per_second": 11.886,
+      "eval_steps_per_second": 1.543,
+      "step": 5200
+    },
+    {
+      "epoch": 81.53846153846153,
+      "grad_norm": 19.078449249267578,
+      "learning_rate": 9.470900000000001e-06,
+      "loss": 1.1814,
+      "step": 5300
+    },
+    {
+      "epoch": 81.53846153846153,
+      "eval_loss": 1.2781996726989746,
+      "eval_runtime": 11.0663,
+      "eval_samples_per_second": 11.838,
+      "eval_steps_per_second": 1.536,
+      "step": 5300
+    },
+    {
+      "epoch": 83.07692307692308,
+      "grad_norm": 31.05898094177246,
+      "learning_rate": 9.460900000000001e-06,
+      "loss": 1.1452,
+      "step": 5400
+    },
+    {
+      "epoch": 83.07692307692308,
+      "eval_loss": 1.2848775386810303,
+      "eval_runtime": 11.01,
+      "eval_samples_per_second": 11.898,
+      "eval_steps_per_second": 1.544,
+      "step": 5400
+    },
+    {
+      "epoch": 84.61538461538461,
+      "grad_norm": 28.712461471557617,
+      "learning_rate": 9.450900000000001e-06,
+      "loss": 1.1465,
+      "step": 5500
+    },
+    {
+      "epoch": 84.61538461538461,
+      "eval_loss": 1.2928494215011597,
+      "eval_runtime": 10.9253,
+      "eval_samples_per_second": 11.991,
+      "eval_steps_per_second": 1.556,
+      "step": 5500
+    },
+    {
+      "epoch": 86.15384615384616,
+      "grad_norm": 19.871828079223633,
+      "learning_rate": 9.440900000000001e-06,
+      "loss": 1.1736,
+      "step": 5600
+    },
+    {
+      "epoch": 86.15384615384616,
+      "eval_loss": 1.2648124694824219,
+      "eval_runtime": 11.0314,
+      "eval_samples_per_second": 11.875,
+      "eval_steps_per_second": 1.541,
+      "step": 5600
+    },
+    {
+      "epoch": 87.6923076923077,
+      "grad_norm": 22.47665023803711,
+      "learning_rate": 9.4309e-06,
+      "loss": 1.1184,
+      "step": 5700
+    },
+    {
+      "epoch": 87.6923076923077,
+      "eval_loss": 1.2936598062515259,
+      "eval_runtime": 10.9322,
+      "eval_samples_per_second": 11.983,
+      "eval_steps_per_second": 1.555,
+      "step": 5700
+    },
+    {
+      "epoch": 89.23076923076923,
+      "grad_norm": 38.79877471923828,
+      "learning_rate": 9.4209e-06,
+      "loss": 1.1616,
+      "step": 5800
+    },
+    {
+      "epoch": 89.23076923076923,
+      "eval_loss": 1.2650004625320435,
+      "eval_runtime": 10.9434,
+      "eval_samples_per_second": 11.971,
+      "eval_steps_per_second": 1.553,
+      "step": 5800
+    },
+    {
+      "epoch": 90.76923076923077,
+      "grad_norm": 40.851097106933594,
+      "learning_rate": 9.4109e-06,
+      "loss": 1.1469,
+      "step": 5900
+    },
+    {
+      "epoch": 90.76923076923077,
+      "eval_loss": 1.252148151397705,
+      "eval_runtime": 10.9867,
+      "eval_samples_per_second": 11.923,
+      "eval_steps_per_second": 1.547,
+      "step": 5900
+    },
+    {
+      "epoch": 92.3076923076923,
+      "grad_norm": 21.35544204711914,
+      "learning_rate": 9.4009e-06,
+      "loss": 1.1489,
+      "step": 6000
+    },
+    {
+      "epoch": 92.3076923076923,
+      "eval_loss": 1.259344220161438,
+      "eval_runtime": 10.9649,
+      "eval_samples_per_second": 11.947,
+      "eval_steps_per_second": 1.55,
+      "step": 6000
+    },
+    {
+      "epoch": 93.84615384615384,
+      "grad_norm": 33.265132904052734,
+      "learning_rate": 9.3909e-06,
+      "loss": 1.1315,
+      "step": 6100
+    },
+    {
+      "epoch": 93.84615384615384,
+      "eval_loss": 1.252693772315979,
+      "eval_runtime": 10.9852,
+      "eval_samples_per_second": 11.925,
+      "eval_steps_per_second": 1.548,
+      "step": 6100
+    },
+    {
+      "epoch": 95.38461538461539,
+      "grad_norm": 23.43667221069336,
+      "learning_rate": 9.381e-06,
+      "loss": 1.119,
+      "step": 6200
+    },
+    {
+      "epoch": 95.38461538461539,
+      "eval_loss": 1.254772424697876,
+      "eval_runtime": 11.0937,
+      "eval_samples_per_second": 11.808,
+      "eval_steps_per_second": 1.532,
+      "step": 6200
+    },
+    {
+      "epoch": 96.92307692307692,
+      "grad_norm": 51.93602752685547,
+      "learning_rate": 9.371e-06,
+      "loss": 1.1333,
+      "step": 6300
+    },
+    {
+      "epoch": 96.92307692307692,
+      "eval_loss": 1.249408483505249,
+      "eval_runtime": 11.0388,
+      "eval_samples_per_second": 11.867,
+      "eval_steps_per_second": 1.54,
+      "step": 6300
+    },
+    {
+      "epoch": 98.46153846153847,
+      "grad_norm": 23.473421096801758,
+      "learning_rate": 9.361e-06,
+      "loss": 1.1164,
+      "step": 6400
+    },
+    {
+      "epoch": 98.46153846153847,
+      "eval_loss": 1.2438522577285767,
+      "eval_runtime": 11.1581,
+      "eval_samples_per_second": 11.74,
+      "eval_steps_per_second": 1.524,
+      "step": 6400
+    },
+    {
+      "epoch": 100.0,
+      "grad_norm": 24.228403091430664,
+      "learning_rate": 9.351e-06,
+      "loss": 1.1333,
+      "step": 6500
+    },
+    {
+      "epoch": 100.0,
+      "eval_loss": 1.248632788658142,
+      "eval_runtime": 10.9675,
+      "eval_samples_per_second": 11.944,
+      "eval_steps_per_second": 1.55,
+      "step": 6500
+    },
+    {
+      "epoch": 101.53846153846153,
+      "grad_norm": 18.29631996154785,
+      "learning_rate": 9.341000000000001e-06,
+      "loss": 1.1082,
+      "step": 6600
+    },
+    {
+      "epoch": 101.53846153846153,
+      "eval_loss": 1.2509865760803223,
+      "eval_runtime": 11.0693,
+      "eval_samples_per_second": 11.834,
+      "eval_steps_per_second": 1.536,
+      "step": 6600
+    },
+    {
+      "epoch": 103.07692307692308,
+      "grad_norm": 42.855491638183594,
+      "learning_rate": 9.331000000000001e-06,
+      "loss": 1.1178,
+      "step": 6700
+    },
+    {
+      "epoch": 103.07692307692308,
+      "eval_loss": 1.2890292406082153,
+      "eval_runtime": 11.0366,
+      "eval_samples_per_second": 11.87,
+      "eval_steps_per_second": 1.54,
+      "step": 6700
+    },
+    {
+      "epoch": 104.61538461538461,
+      "grad_norm": 46.675655364990234,
+      "learning_rate": 9.321000000000001e-06,
+      "loss": 1.1106,
+      "step": 6800
+    },
+    {
+      "epoch": 104.61538461538461,
+      "eval_loss": 1.2719863653182983,
+      "eval_runtime": 11.1266,
+      "eval_samples_per_second": 11.774,
+      "eval_steps_per_second": 1.528,
+      "step": 6800
+    },
+    {
+      "epoch": 106.15384615384616,
+      "grad_norm": 26.414846420288086,
+      "learning_rate": 9.311000000000001e-06,
+      "loss": 1.1216,
+      "step": 6900
+    },
+    {
+      "epoch": 106.15384615384616,
+      "eval_loss": 1.2423394918441772,
+      "eval_runtime": 11.0197,
+      "eval_samples_per_second": 11.888,
+      "eval_steps_per_second": 1.543,
+      "step": 6900
+    },
+    {
+      "epoch": 107.6923076923077,
+      "grad_norm": 31.4022274017334,
+      "learning_rate": 9.301000000000001e-06,
+      "loss": 1.1052,
+      "step": 7000
+    },
+    {
+      "epoch": 107.6923076923077,
+      "eval_loss": 1.2372961044311523,
+      "eval_runtime": 11.1127,
+      "eval_samples_per_second": 11.788,
+      "eval_steps_per_second": 1.53,
+      "step": 7000
+    },
+    {
+      "epoch": 109.23076923076923,
+      "grad_norm": 23.16703987121582,
+      "learning_rate": 9.291000000000001e-06,
+      "loss": 1.0911,
+      "step": 7100
+    },
+    {
+      "epoch": 109.23076923076923,
+      "eval_loss": 1.2309863567352295,
+      "eval_runtime": 10.9572,
+      "eval_samples_per_second": 11.956,
+      "eval_steps_per_second": 1.551,
+      "step": 7100
+    },
+    {
+      "epoch": 110.76923076923077,
+      "grad_norm": 21.648773193359375,
+      "learning_rate": 9.281000000000001e-06,
+      "loss": 1.0956,
+      "step": 7200
+    },
+    {
+      "epoch": 110.76923076923077,
+      "eval_loss": 1.2261079549789429,
+      "eval_runtime": 10.9351,
+      "eval_samples_per_second": 11.98,
+      "eval_steps_per_second": 1.555,
+      "step": 7200
+    },
+    {
+      "epoch": 112.3076923076923,
+      "grad_norm": 24.5791072845459,
+      "learning_rate": 9.271000000000002e-06,
+      "loss": 1.0751,
+      "step": 7300
+    },
+    {
+      "epoch": 112.3076923076923,
+      "eval_loss": 1.2161471843719482,
+      "eval_runtime": 11.1052,
+      "eval_samples_per_second": 11.796,
+      "eval_steps_per_second": 1.531,
+      "step": 7300
+    },
+    {
+      "epoch": 113.84615384615384,
+      "grad_norm": 35.867801666259766,
+      "learning_rate": 9.261000000000002e-06,
+      "loss": 1.086,
+      "step": 7400
+    },
+    {
+      "epoch": 113.84615384615384,
+      "eval_loss": 1.2092362642288208,
+      "eval_runtime": 11.2048,
+      "eval_samples_per_second": 11.691,
+      "eval_steps_per_second": 1.517,
+      "step": 7400
+    },
+    {
+      "epoch": 115.38461538461539,
+      "grad_norm": 67.91041564941406,
+      "learning_rate": 9.251000000000002e-06,
+      "loss": 1.092,
+      "step": 7500
+    },
+    {
+      "epoch": 115.38461538461539,
+      "eval_loss": 1.241829514503479,
+      "eval_runtime": 10.9937,
+      "eval_samples_per_second": 11.916,
+      "eval_steps_per_second": 1.546,
+      "step": 7500
+    },
+    {
+      "epoch": 116.92307692307692,
+      "grad_norm": 128.73751831054688,
+      "learning_rate": 9.241000000000002e-06,
+      "loss": 1.0764,
+      "step": 7600
+    },
+    {
+      "epoch": 116.92307692307692,
+      "eval_loss": 1.2462713718414307,
+      "eval_runtime": 10.9625,
+      "eval_samples_per_second": 11.95,
+      "eval_steps_per_second": 1.551,
+      "step": 7600
+    },
+    {
+      "epoch": 118.46153846153847,
+      "grad_norm": 86.5144271850586,
+      "learning_rate": 9.231000000000002e-06,
+      "loss": 1.0643,
+      "step": 7700
+    },
+    {
+      "epoch": 118.46153846153847,
+      "eval_loss": 1.2187525033950806,
+      "eval_runtime": 10.9492,
+      "eval_samples_per_second": 11.964,
+      "eval_steps_per_second": 1.553,
+      "step": 7700
+    },
+    {
+      "epoch": 120.0,
+      "grad_norm": 19.4710750579834,
+      "learning_rate": 9.221e-06,
+      "loss": 1.0966,
+      "step": 7800
+    },
+    {
+      "epoch": 120.0,
+      "eval_loss": 1.2282384634017944,
+      "eval_runtime": 10.9078,
+      "eval_samples_per_second": 12.01,
+      "eval_steps_per_second": 1.559,
+      "step": 7800
+    },
+    {
+      "epoch": 121.53846153846153,
+      "grad_norm": 37.673744201660156,
+      "learning_rate": 9.211e-06,
+      "loss": 1.0632,
+      "step": 7900
+    },
+    {
+      "epoch": 121.53846153846153,
+      "eval_loss": 1.2206230163574219,
+      "eval_runtime": 11.0018,
+      "eval_samples_per_second": 11.907,
+      "eval_steps_per_second": 1.545,
+      "step": 7900
+    },
+    {
+      "epoch": 123.07692307692308,
+      "grad_norm": 25.10326385498047,
+      "learning_rate": 9.201e-06,
+      "loss": 1.0873,
+      "step": 8000
+    },
+    {
+      "epoch": 123.07692307692308,
+      "eval_loss": 1.2137339115142822,
+      "eval_runtime": 10.9232,
+      "eval_samples_per_second": 11.993,
+      "eval_steps_per_second": 1.556,
+      "step": 8000
+    },
+    {
+      "epoch": 124.61538461538461,
+      "grad_norm": 32.02176284790039,
+      "learning_rate": 9.191e-06,
+      "loss": 1.0568,
+      "step": 8100
+    },
+    {
+      "epoch": 124.61538461538461,
+      "eval_loss": 1.2065187692642212,
+      "eval_runtime": 10.9614,
+      "eval_samples_per_second": 11.951,
+      "eval_steps_per_second": 1.551,
+      "step": 8100
+    },
+    {
+      "epoch": 126.15384615384616,
+      "grad_norm": 19.97406005859375,
+      "learning_rate": 9.181e-06,
+      "loss": 1.065,
+      "step": 8200
+    },
+    {
+      "epoch": 126.15384615384616,
+      "eval_loss": 1.2094841003417969,
+      "eval_runtime": 11.0274,
+      "eval_samples_per_second": 11.879,
+      "eval_steps_per_second": 1.542,
+      "step": 8200
+    },
+    {
+      "epoch": 127.6923076923077,
+      "grad_norm": 31.624399185180664,
+      "learning_rate": 9.171e-06,
+      "loss": 1.0805,
+      "step": 8300
+    },
+    {
+      "epoch": 127.6923076923077,
+      "eval_loss": 1.2149733304977417,
+      "eval_runtime": 11.2014,
+      "eval_samples_per_second": 11.695,
+      "eval_steps_per_second": 1.518,
+      "step": 8300
+    },
+    {
+      "epoch": 129.23076923076923,
+      "grad_norm": 29.24848747253418,
+      "learning_rate": 9.161000000000001e-06,
+      "loss": 1.0463,
+      "step": 8400
+    },
+    {
+      "epoch": 129.23076923076923,
+      "eval_loss": 1.2077099084854126,
+      "eval_runtime": 10.9432,
+      "eval_samples_per_second": 11.971,
+      "eval_steps_per_second": 1.553,
+      "step": 8400
+    },
+    {
+      "epoch": 130.76923076923077,
+      "grad_norm": 27.14787483215332,
+      "learning_rate": 9.151000000000001e-06,
+      "loss": 1.0607,
+      "step": 8500
+    },
+    {
+      "epoch": 130.76923076923077,
+      "eval_loss": 1.2046644687652588,
+      "eval_runtime": 11.001,
+      "eval_samples_per_second": 11.908,
+      "eval_steps_per_second": 1.545,
+      "step": 8500
+    },
+    {
+      "epoch": 132.30769230769232,
+      "grad_norm": 32.416194915771484,
+      "learning_rate": 9.141000000000001e-06,
+      "loss": 1.0365,
+      "step": 8600
+    },
+    {
+      "epoch": 132.30769230769232,
+      "eval_loss": 1.195080041885376,
+      "eval_runtime": 11.0205,
+      "eval_samples_per_second": 11.887,
+      "eval_steps_per_second": 1.543,
+      "step": 8600
+    },
+    {
+      "epoch": 133.84615384615384,
+      "grad_norm": 58.863582611083984,
+      "learning_rate": 9.131000000000001e-06,
+      "loss": 1.0564,
+      "step": 8700
+    },
+    {
+      "epoch": 133.84615384615384,
+      "eval_loss": 1.1966980695724487,
+      "eval_runtime": 10.9043,
+      "eval_samples_per_second": 12.014,
+      "eval_steps_per_second": 1.559,
+      "step": 8700
+    },
+    {
+      "epoch": 135.3846153846154,
+      "grad_norm": 26.08232307434082,
+      "learning_rate": 9.121000000000001e-06,
+      "loss": 1.0507,
+      "step": 8800
+    },
+    {
+      "epoch": 135.3846153846154,
+      "eval_loss": 1.201367735862732,
+      "eval_runtime": 10.9714,
+      "eval_samples_per_second": 11.94,
+      "eval_steps_per_second": 1.549,
+      "step": 8800
+    },
+    {
+      "epoch": 136.92307692307693,
+      "grad_norm": 59.368019104003906,
+      "learning_rate": 9.111000000000001e-06,
+      "loss": 1.0508,
+      "step": 8900
+    },
+    {
+      "epoch": 136.92307692307693,
+      "eval_loss": 1.208795428276062,
+      "eval_runtime": 11.0529,
+      "eval_samples_per_second": 11.852,
+      "eval_steps_per_second": 1.538,
+      "step": 8900
+    },
+    {
+      "epoch": 138.46153846153845,
+      "grad_norm": 43.07460021972656,
+      "learning_rate": 9.101000000000001e-06,
+      "loss": 1.0359,
+      "step": 9000
+    },
+    {
+      "epoch": 138.46153846153845,
+      "eval_loss": 1.1782392263412476,
+      "eval_runtime": 11.1519,
+      "eval_samples_per_second": 11.747,
+      "eval_steps_per_second": 1.524,
+      "step": 9000
+    },
+    {
+      "epoch": 140.0,
+      "grad_norm": 22.39567756652832,
+      "learning_rate": 9.091000000000002e-06,
+      "loss": 1.0584,
+      "step": 9100
+    },
+    {
+      "epoch": 140.0,
+      "eval_loss": 1.1873483657836914,
+      "eval_runtime": 10.8732,
+      "eval_samples_per_second": 12.048,
+      "eval_steps_per_second": 1.563,
+      "step": 9100
+    },
+    {
+      "epoch": 141.53846153846155,
+      "grad_norm": 22.813621520996094,
+      "learning_rate": 9.081000000000002e-06,
+      "loss": 1.0354,
+      "step": 9200
+    },
+    {
+      "epoch": 141.53846153846155,
+      "eval_loss": 1.1731117963790894,
+      "eval_runtime": 11.0847,
+      "eval_samples_per_second": 11.818,
+      "eval_steps_per_second": 1.534,
+      "step": 9200
+    },
+    {
+      "epoch": 143.07692307692307,
+      "grad_norm": 31.44073486328125,
+      "learning_rate": 9.071000000000002e-06,
+      "loss": 1.0457,
+      "step": 9300
+    },
+    {
+      "epoch": 143.07692307692307,
+      "eval_loss": 1.1962807178497314,
+      "eval_runtime": 11.0668,
+      "eval_samples_per_second": 11.837,
+      "eval_steps_per_second": 1.536,
+      "step": 9300
+    },
+    {
+      "epoch": 144.6153846153846,
+      "grad_norm": 18.18711280822754,
+      "learning_rate": 9.061e-06,
+      "loss": 1.0481,
+      "step": 9400
+    },
+    {
+      "epoch": 144.6153846153846,
+      "eval_loss": 1.185339093208313,
+      "eval_runtime": 10.9916,
+      "eval_samples_per_second": 11.918,
+      "eval_steps_per_second": 1.547,
+      "step": 9400
+    },
+    {
+      "epoch": 146.15384615384616,
+      "grad_norm": 38.05665969848633,
+      "learning_rate": 9.051e-06,
+      "loss": 1.0391,
+      "step": 9500
+    },
+    {
+      "epoch": 146.15384615384616,
+      "eval_loss": 1.1856777667999268,
+      "eval_runtime": 11.0391,
+      "eval_samples_per_second": 11.867,
+      "eval_steps_per_second": 1.54,
+      "step": 9500
+    },
+    {
+      "epoch": 147.69230769230768,
+      "grad_norm": 19.963260650634766,
+      "learning_rate": 9.041e-06,
+      "loss": 1.0322,
+      "step": 9600
+    },
+    {
+      "epoch": 147.69230769230768,
+      "eval_loss": 1.1843148469924927,
+      "eval_runtime": 11.0382,
+      "eval_samples_per_second": 11.868,
+      "eval_steps_per_second": 1.54,
+      "step": 9600
+    },
+    {
+      "epoch": 149.23076923076923,
+      "grad_norm": 28.58108901977539,
+      "learning_rate": 9.031e-06,
+      "loss": 1.0369,
+      "step": 9700
+    },
+    {
+      "epoch": 149.23076923076923,
+      "eval_loss": 1.1637182235717773,
+      "eval_runtime": 11.2766,
+      "eval_samples_per_second": 11.617,
+      "eval_steps_per_second": 1.508,
+      "step": 9700
+    },
+    {
+      "epoch": 150.76923076923077,
+      "grad_norm": 64.3956527709961,
+      "learning_rate": 9.021e-06,
+      "loss": 1.0519,
+      "step": 9800
+    },
+    {
+      "epoch": 150.76923076923077,
+      "eval_loss": 1.1796802282333374,
+      "eval_runtime": 11.3418,
+      "eval_samples_per_second": 11.55,
+      "eval_steps_per_second": 1.499,
+      "step": 9800
+    },
+    {
+      "epoch": 152.30769230769232,
+      "grad_norm": 18.857580184936523,
+      "learning_rate": 9.011e-06,
+      "loss": 1.0272,
+      "step": 9900
+    },
+    {
+      "epoch": 152.30769230769232,
+      "eval_loss": 1.1634279489517212,
+      "eval_runtime": 11.6749,
+      "eval_samples_per_second": 11.221,
+      "eval_steps_per_second": 1.456,
+      "step": 9900
+    },
+    {
+      "epoch": 153.84615384615384,
+      "grad_norm": 42.541683197021484,
+      "learning_rate": 9.001e-06,
+      "loss": 1.0287,
+      "step": 10000
+    },
+    {
+      "epoch": 153.84615384615384,
+      "eval_loss": 1.1698088645935059,
+      "eval_runtime": 11.068,
+      "eval_samples_per_second": 11.836,
+      "eval_steps_per_second": 1.536,
+      "step": 10000
+    },
+    {
+      "epoch": 155.3846153846154,
+      "grad_norm": 30.52286720275879,
+      "learning_rate": 8.991e-06,
+      "loss": 1.0237,
+      "step": 10100
+    },
+    {
+      "epoch": 155.3846153846154,
+      "eval_loss": 1.178311824798584,
+      "eval_runtime": 11.1107,
+      "eval_samples_per_second": 11.79,
+      "eval_steps_per_second": 1.53,
+      "step": 10100
+    },
+    {
+      "epoch": 156.92307692307693,
+      "grad_norm": 32.60612487792969,
+      "learning_rate": 8.981000000000001e-06,
+      "loss": 1.0362,
+      "step": 10200
+    },
+    {
+      "epoch": 156.92307692307693,
+      "eval_loss": 1.157893180847168,
+      "eval_runtime": 11.0484,
+      "eval_samples_per_second": 11.857,
+      "eval_steps_per_second": 1.539,
+      "step": 10200
+    },
+    {
+      "epoch": 158.46153846153845,
+      "grad_norm": 26.15647315979004,
+      "learning_rate": 8.971000000000001e-06,
+      "loss": 0.998,
+      "step": 10300
+    },
+    {
+      "epoch": 158.46153846153845,
+      "eval_loss": 1.1724605560302734,
+      "eval_runtime": 11.2323,
+      "eval_samples_per_second": 11.663,
+      "eval_steps_per_second": 1.513,
+      "step": 10300
+    },
+    {
+      "epoch": 160.0,
+      "grad_norm": 45.76993942260742,
+      "learning_rate": 8.961000000000001e-06,
+      "loss": 1.0432,
+      "step": 10400
+    },
+    {
+      "epoch": 160.0,
+      "eval_loss": 1.1688075065612793,
+      "eval_runtime": 11.0984,
+      "eval_samples_per_second": 11.803,
+      "eval_steps_per_second": 1.532,
+      "step": 10400
+    },
+    {
+      "epoch": 161.53846153846155,
+      "grad_norm": 40.384578704833984,
+      "learning_rate": 8.951000000000001e-06,
+      "loss": 1.0421,
+      "step": 10500
+    },
+    {
+      "epoch": 161.53846153846155,
+      "eval_loss": 1.1621845960617065,
+      "eval_runtime": 11.1624,
+      "eval_samples_per_second": 11.736,
+      "eval_steps_per_second": 1.523,
+      "step": 10500
+    },
+    {
+      "epoch": 163.07692307692307,
+      "grad_norm": 22.886058807373047,
+      "learning_rate": 8.941000000000001e-06,
+      "loss": 0.9912,
+      "step": 10600
+    },
+    {
+      "epoch": 163.07692307692307,
+      "eval_loss": 1.1695055961608887,
+      "eval_runtime": 10.9842,
+      "eval_samples_per_second": 11.926,
+      "eval_steps_per_second": 1.548,
+      "step": 10600
+    },
+    {
+      "epoch": 164.6153846153846,
+      "grad_norm": 26.940736770629883,
+      "learning_rate": 8.931000000000001e-06,
+      "loss": 1.011,
+      "step": 10700
+    },
+    {
+      "epoch": 164.6153846153846,
+      "eval_loss": 1.1458157300949097,
+      "eval_runtime": 11.025,
+      "eval_samples_per_second": 11.882,
+      "eval_steps_per_second": 1.542,
+      "step": 10700
+    },
+    {
+      "epoch": 166.15384615384616,
+      "grad_norm": 26.5013484954834,
+      "learning_rate": 8.921000000000001e-06,
+      "loss": 0.9876,
+      "step": 10800
+    },
+    {
+      "epoch": 166.15384615384616,
+      "eval_loss": 1.1569631099700928,
+      "eval_runtime": 11.101,
+      "eval_samples_per_second": 11.801,
+      "eval_steps_per_second": 1.531,
+      "step": 10800
+    },
+    {
+      "epoch": 167.69230769230768,
+      "grad_norm": 29.859987258911133,
+      "learning_rate": 8.911000000000002e-06,
+      "loss": 1.0374,
+      "step": 10900
+    },
+    {
+      "epoch": 167.69230769230768,
+      "eval_loss": 1.149316668510437,
+      "eval_runtime": 11.1242,
+      "eval_samples_per_second": 11.776,
+      "eval_steps_per_second": 1.528,
+      "step": 10900
+    },
+    {
+      "epoch": 169.23076923076923,
+      "grad_norm": 27.777738571166992,
+      "learning_rate": 8.901e-06,
+      "loss": 0.985,
+      "step": 11000
+    },
+    {
+      "epoch": 169.23076923076923,
+      "eval_loss": 1.1608328819274902,
+      "eval_runtime": 11.2215,
+      "eval_samples_per_second": 11.674,
+      "eval_steps_per_second": 1.515,
+      "step": 11000
+    },
+    {
+      "epoch": 170.76923076923077,
+      "grad_norm": 39.21344757080078,
+      "learning_rate": 8.891e-06,
+      "loss": 1.0049,
+      "step": 11100
+    },
+    {
+      "epoch": 170.76923076923077,
+      "eval_loss": 1.1642228364944458,
+      "eval_runtime": 11.1947,
+      "eval_samples_per_second": 11.702,
+      "eval_steps_per_second": 1.519,
+      "step": 11100
+    },
+    {
+      "epoch": 172.30769230769232,
+      "grad_norm": 29.880149841308594,
+      "learning_rate": 8.881e-06,
+      "loss": 0.9843,
+      "step": 11200
+    },
+    {
+      "epoch": 172.30769230769232,
+      "eval_loss": 1.1574000120162964,
+      "eval_runtime": 11.2634,
+      "eval_samples_per_second": 11.631,
+      "eval_steps_per_second": 1.509,
+      "step": 11200
+    },
+    {
+      "epoch": 173.84615384615384,
+      "grad_norm": 63.53031539916992,
+      "learning_rate": 8.871e-06,
+      "loss": 1.0354,
+      "step": 11300
+    },
+    {
+      "epoch": 173.84615384615384,
+      "eval_loss": 1.1575734615325928,
+      "eval_runtime": 11.0265,
+      "eval_samples_per_second": 11.88,
+      "eval_steps_per_second": 1.542,
+      "step": 11300
+    },
+    {
+      "epoch": 175.3846153846154,
+      "grad_norm": 26.937786102294922,
+      "learning_rate": 8.861e-06,
+      "loss": 0.9964,
+      "step": 11400
+    },
+    {
+      "epoch": 175.3846153846154,
+      "eval_loss": 1.1552445888519287,
+      "eval_runtime": 11.0325,
+      "eval_samples_per_second": 11.874,
+      "eval_steps_per_second": 1.541,
+      "step": 11400
+    },
+    {
+      "epoch": 176.92307692307693,
+      "grad_norm": 87.28536987304688,
+      "learning_rate": 8.851e-06,
+      "loss": 0.9932,
+      "step": 11500
+    },
+    {
+      "epoch": 176.92307692307693,
+      "eval_loss": 1.1411677598953247,
+      "eval_runtime": 11.1527,
+      "eval_samples_per_second": 11.746,
+      "eval_steps_per_second": 1.524,
+      "step": 11500
+    },
+    {
+      "epoch": 178.46153846153845,
+      "grad_norm": 25.903568267822266,
+      "learning_rate": 8.841e-06,
+      "loss": 0.9768,
+      "step": 11600
+    },
+    {
+      "epoch": 178.46153846153845,
+      "eval_loss": 1.1635726690292358,
+      "eval_runtime": 11.123,
+      "eval_samples_per_second": 11.777,
+      "eval_steps_per_second": 1.528,
+      "step": 11600
+    },
+    {
+      "epoch": 180.0,
+      "grad_norm": 24.315654754638672,
+      "learning_rate": 8.831e-06,
+      "loss": 0.9984,
+      "step": 11700
+    },
+    {
+      "epoch": 180.0,
+      "eval_loss": 1.1710366010665894,
+      "eval_runtime": 11.1375,
+      "eval_samples_per_second": 11.762,
+      "eval_steps_per_second": 1.526,
+      "step": 11700
+    },
+    {
+      "epoch": 181.53846153846155,
+      "grad_norm": 60.17182540893555,
+      "learning_rate": 8.821e-06,
+      "loss": 0.9703,
+      "step": 11800
+    },
+    {
+      "epoch": 181.53846153846155,
+      "eval_loss": 1.1556936502456665,
+      "eval_runtime": 11.0387,
+      "eval_samples_per_second": 11.867,
+      "eval_steps_per_second": 1.54,
+      "step": 11800
+    },
+    {
+      "epoch": 183.07692307692307,
+      "grad_norm": 32.373477935791016,
+      "learning_rate": 8.811000000000001e-06,
+      "loss": 0.9996,
+      "step": 11900
+    },
+    {
+      "epoch": 183.07692307692307,
+      "eval_loss": 1.146790623664856,
+      "eval_runtime": 11.1805,
+      "eval_samples_per_second": 11.717,
+      "eval_steps_per_second": 1.521,
+      "step": 11900
+    },
+    {
+      "epoch": 184.6153846153846,
+      "grad_norm": 42.578575134277344,
+      "learning_rate": 8.801000000000001e-06,
+      "loss": 0.9795,
+      "step": 12000
+    },
+    {
+      "epoch": 184.6153846153846,
+      "eval_loss": 1.144544005393982,
+      "eval_runtime": 11.0704,
+      "eval_samples_per_second": 11.833,
+      "eval_steps_per_second": 1.536,
+      "step": 12000
+    },
+    {
+      "epoch": 186.15384615384616,
+      "grad_norm": 22.79789161682129,
+      "learning_rate": 8.791000000000001e-06,
+      "loss": 0.9905,
+      "step": 12100
+    },
+    {
+      "epoch": 186.15384615384616,
+      "eval_loss": 1.1581685543060303,
+      "eval_runtime": 11.272,
+      "eval_samples_per_second": 11.622,
+      "eval_steps_per_second": 1.508,
+      "step": 12100
+    },
+    {
+      "epoch": 187.69230769230768,
+      "grad_norm": 52.061012268066406,
+      "learning_rate": 8.781200000000002e-06,
+      "loss": 0.9817,
+      "step": 12200
+    },
+    {
+      "epoch": 187.69230769230768,
+      "eval_loss": 1.159809947013855,
+      "eval_runtime": 11.2021,
+      "eval_samples_per_second": 11.694,
+      "eval_steps_per_second": 1.518,
+      "step": 12200
+    },
+    {
+      "epoch": 189.23076923076923,
+      "grad_norm": 35.95882034301758,
+      "learning_rate": 8.7712e-06,
+      "loss": 1.0071,
+      "step": 12300
+    },
+    {
+      "epoch": 189.23076923076923,
+      "eval_loss": 1.1944890022277832,
+      "eval_runtime": 10.9263,
+      "eval_samples_per_second": 11.989,
+      "eval_steps_per_second": 1.556,
+      "step": 12300
+    },
+    {
+      "epoch": 190.76923076923077,
+      "grad_norm": 213.48587036132812,
+      "learning_rate": 8.7612e-06,
+      "loss": 0.9997,
+      "step": 12400
+    },
+    {
+      "epoch": 190.76923076923077,
+      "eval_loss": 1.191455602645874,
+      "eval_runtime": 10.917,
+      "eval_samples_per_second": 12.0,
+      "eval_steps_per_second": 1.557,
+      "step": 12400
+    },
+    {
+      "epoch": 192.30769230769232,
+      "grad_norm": 19.97510528564453,
+      "learning_rate": 8.7512e-06,
+      "loss": 1.001,
+      "step": 12500
+    },
+    {
+      "epoch": 192.30769230769232,
+      "eval_loss": 1.167776346206665,
+      "eval_runtime": 11.033,
+      "eval_samples_per_second": 11.873,
+      "eval_steps_per_second": 1.541,
+      "step": 12500
+    },
+    {
+      "epoch": 193.84615384615384,
+      "grad_norm": 30.815828323364258,
+      "learning_rate": 8.7412e-06,
+      "loss": 0.9719,
+      "step": 12600
+    },
+    {
+      "epoch": 193.84615384615384,
+      "eval_loss": 1.150451898574829,
+      "eval_runtime": 11.2471,
+      "eval_samples_per_second": 11.647,
+      "eval_steps_per_second": 1.512,
+      "step": 12600
+    },
+    {
+      "epoch": 195.3846153846154,
+      "grad_norm": 40.32701110839844,
+      "learning_rate": 8.7312e-06,
+      "loss": 0.9658,
+      "step": 12700
+    },
+    {
+      "epoch": 195.3846153846154,
+      "eval_loss": 1.1517494916915894,
+      "eval_runtime": 11.1893,
+      "eval_samples_per_second": 11.708,
+      "eval_steps_per_second": 1.519,
+      "step": 12700
+    },
+    {
+      "epoch": 196.92307692307693,
+      "grad_norm": 42.11077117919922,
+      "learning_rate": 8.7212e-06,
+      "loss": 0.9744,
+      "step": 12800
+    },
+    {
+      "epoch": 196.92307692307693,
+      "eval_loss": 1.1507395505905151,
+      "eval_runtime": 11.2196,
+      "eval_samples_per_second": 11.676,
+      "eval_steps_per_second": 1.515,
+      "step": 12800
+    },
+    {
+      "epoch": 198.46153846153845,
+      "grad_norm": 20.991779327392578,
+      "learning_rate": 8.7112e-06,
+      "loss": 0.9695,
+      "step": 12900
+    },
+    {
+      "epoch": 198.46153846153845,
+      "eval_loss": 1.1557880640029907,
+      "eval_runtime": 11.2158,
+      "eval_samples_per_second": 11.68,
+      "eval_steps_per_second": 1.516,
+      "step": 12900
+    },
+    {
+      "epoch": 200.0,
+      "grad_norm": 22.79688835144043,
+      "learning_rate": 8.7012e-06,
+      "loss": 0.9652,
+      "step": 13000
+    },
+    {
+      "epoch": 200.0,
+      "eval_loss": 1.1554670333862305,
+      "eval_runtime": 11.1246,
+      "eval_samples_per_second": 11.776,
+      "eval_steps_per_second": 1.528,
+      "step": 13000
+    },
+    {
+      "epoch": 201.53846153846155,
+      "grad_norm": 180.34512329101562,
+      "learning_rate": 8.6912e-06,
+      "loss": 0.9664,
+      "step": 13100
+    },
+    {
+      "epoch": 201.53846153846155,
+      "eval_loss": 1.1403967142105103,
+      "eval_runtime": 10.9895,
+      "eval_samples_per_second": 11.921,
+      "eval_steps_per_second": 1.547,
+      "step": 13100
+    },
+    {
+      "epoch": 203.07692307692307,
+      "grad_norm": 31.583358764648438,
+      "learning_rate": 8.6812e-06,
+      "loss": 0.9672,
+      "step": 13200
+    },
+    {
+      "epoch": 203.07692307692307,
+      "eval_loss": 1.1569470167160034,
+      "eval_runtime": 11.2695,
+      "eval_samples_per_second": 11.624,
+      "eval_steps_per_second": 1.508,
+      "step": 13200
+    },
+    {
+      "epoch": 204.6153846153846,
+      "grad_norm": 34.05722427368164,
+      "learning_rate": 8.671200000000001e-06,
+      "loss": 0.9531,
+      "step": 13300
+    },
+    {
+      "epoch": 204.6153846153846,
+      "eval_loss": 1.1408321857452393,
+      "eval_runtime": 11.1074,
+      "eval_samples_per_second": 11.794,
+      "eval_steps_per_second": 1.531,
+      "step": 13300
+    },
+    {
+      "epoch": 206.15384615384616,
+      "grad_norm": 26.748388290405273,
+      "learning_rate": 8.661200000000001e-06,
+      "loss": 0.9484,
+      "step": 13400
+    },
+    {
+      "epoch": 206.15384615384616,
+      "eval_loss": 1.151693344116211,
+      "eval_runtime": 10.9585,
+      "eval_samples_per_second": 11.954,
+      "eval_steps_per_second": 1.551,
+      "step": 13400
+    },
+    {
+      "epoch": 207.69230769230768,
+      "grad_norm": 19.531770706176758,
+      "learning_rate": 8.651200000000001e-06,
+      "loss": 0.971,
+      "step": 13500
+    },
+    {
+      "epoch": 207.69230769230768,
+      "eval_loss": 1.138724684715271,
+      "eval_runtime": 11.0296,
+      "eval_samples_per_second": 11.877,
+      "eval_steps_per_second": 1.541,
+      "step": 13500
+    },
+    {
+      "epoch": 209.23076923076923,
+      "grad_norm": 23.87537384033203,
+      "learning_rate": 8.641200000000001e-06,
+      "loss": 0.944,
+      "step": 13600
+    },
+    {
+      "epoch": 209.23076923076923,
+      "eval_loss": 1.1402664184570312,
+      "eval_runtime": 11.1505,
+      "eval_samples_per_second": 11.748,
+      "eval_steps_per_second": 1.525,
+      "step": 13600
+    },
+    {
+      "epoch": 210.76923076923077,
+      "grad_norm": 25.069852828979492,
+      "learning_rate": 8.631200000000001e-06,
+      "loss": 0.9581,
+      "step": 13700
+    },
+    {
+      "epoch": 210.76923076923077,
+      "eval_loss": 1.1348073482513428,
+      "eval_runtime": 11.063,
+      "eval_samples_per_second": 11.841,
+      "eval_steps_per_second": 1.537,
+      "step": 13700
+    },
+    {
+      "epoch": 212.30769230769232,
+      "grad_norm": 23.794719696044922,
+      "learning_rate": 8.621200000000001e-06,
+      "loss": 0.957,
+      "step": 13800
+    },
+    {
+      "epoch": 212.30769230769232,
+      "eval_loss": 1.143198013305664,
+      "eval_runtime": 11.1422,
+      "eval_samples_per_second": 11.757,
+      "eval_steps_per_second": 1.526,
+      "step": 13800
+    },
+    {
+      "epoch": 213.84615384615384,
+      "grad_norm": 26.059829711914062,
+      "learning_rate": 8.611200000000002e-06,
+      "loss": 0.9554,
+      "step": 13900
+    },
+    {
+      "epoch": 213.84615384615384,
+      "eval_loss": 1.1333541870117188,
+      "eval_runtime": 10.948,
+      "eval_samples_per_second": 11.966,
+      "eval_steps_per_second": 1.553,
+      "step": 13900
+    },
+    {
+      "epoch": 215.3846153846154,
+      "grad_norm": 49.8937873840332,
+      "learning_rate": 8.6012e-06,
+      "loss": 0.9607,
+      "step": 14000
+    },
+    {
+      "epoch": 215.3846153846154,
+      "eval_loss": 1.1584446430206299,
+      "eval_runtime": 11.2488,
+      "eval_samples_per_second": 11.646,
+      "eval_steps_per_second": 1.511,
+      "step": 14000
+    },
+    {
+      "epoch": 216.92307692307693,
+      "grad_norm": 17.9267520904541,
+      "learning_rate": 8.5912e-06,
+      "loss": 0.9444,
+      "step": 14100
+    },
+    {
+      "epoch": 216.92307692307693,
+      "eval_loss": 1.1573532819747925,
+      "eval_runtime": 11.143,
+      "eval_samples_per_second": 11.756,
+      "eval_steps_per_second": 1.526,
+      "step": 14100
+    },
+    {
+      "epoch": 218.46153846153845,
+      "grad_norm": 27.38156509399414,
+      "learning_rate": 8.5812e-06,
+      "loss": 0.928,
+      "step": 14200
+    },
+    {
+      "epoch": 218.46153846153845,
+      "eval_loss": 1.1540145874023438,
+      "eval_runtime": 11.0475,
+      "eval_samples_per_second": 11.858,
+      "eval_steps_per_second": 1.539,
+      "step": 14200
+    },
+    {
+      "epoch": 220.0,
+      "grad_norm": 42.785037994384766,
+      "learning_rate": 8.5712e-06,
+      "loss": 0.9548,
+      "step": 14300
+    },
+    {
+      "epoch": 220.0,
+      "eval_loss": 1.1379021406173706,
+      "eval_runtime": 10.9412,
+      "eval_samples_per_second": 11.973,
+      "eval_steps_per_second": 1.554,
+      "step": 14300
+    },
+    {
+      "epoch": 221.53846153846155,
+      "grad_norm": 39.50480270385742,
+      "learning_rate": 8.5612e-06,
+      "loss": 0.9583,
+      "step": 14400
+    },
+    {
+      "epoch": 221.53846153846155,
+      "eval_loss": 1.1666078567504883,
+      "eval_runtime": 11.2066,
+      "eval_samples_per_second": 11.69,
+      "eval_steps_per_second": 1.517,
+      "step": 14400
+    },
+    {
+      "epoch": 223.07692307692307,
+      "grad_norm": 15.560932159423828,
+      "learning_rate": 8.5512e-06,
+      "loss": 0.9306,
+      "step": 14500
+    },
+    {
+      "epoch": 223.07692307692307,
+      "eval_loss": 1.151904582977295,
+      "eval_runtime": 11.2376,
+      "eval_samples_per_second": 11.657,
+      "eval_steps_per_second": 1.513,
+      "step": 14500
+    },
+    {
+      "epoch": 224.6153846153846,
+      "grad_norm": 36.12020492553711,
+      "learning_rate": 8.541400000000001e-06,
+      "loss": 0.9668,
+      "step": 14600
+    },
+    {
+      "epoch": 224.6153846153846,
+      "eval_loss": 1.139450192451477,
+      "eval_runtime": 11.1643,
+      "eval_samples_per_second": 11.734,
+      "eval_steps_per_second": 1.523,
+      "step": 14600
+    },
+    {
+      "epoch": 226.15384615384616,
+      "grad_norm": 31.29511070251465,
+      "learning_rate": 8.531400000000001e-06,
+      "loss": 0.9646,
+      "step": 14700
+    },
+    {
+      "epoch": 226.15384615384616,
+      "eval_loss": 1.1311490535736084,
+      "eval_runtime": 11.0,
+      "eval_samples_per_second": 11.909,
+      "eval_steps_per_second": 1.545,
+      "step": 14700
+    },
+    {
+      "epoch": 227.69230769230768,
+      "grad_norm": 22.08748435974121,
+      "learning_rate": 8.521400000000001e-06,
+      "loss": 0.922,
+      "step": 14800
+    },
+    {
+      "epoch": 227.69230769230768,
+      "eval_loss": 1.1504499912261963,
+      "eval_runtime": 11.135,
+      "eval_samples_per_second": 11.765,
+      "eval_steps_per_second": 1.527,
+      "step": 14800
+    },
+    {
+      "epoch": 229.23076923076923,
+      "grad_norm": 26.33457374572754,
+      "learning_rate": 8.511400000000001e-06,
+      "loss": 0.9306,
+      "step": 14900
+    },
+    {
+      "epoch": 229.23076923076923,
+      "eval_loss": 1.136217713356018,
+      "eval_runtime": 11.1465,
+      "eval_samples_per_second": 11.753,
+      "eval_steps_per_second": 1.525,
+      "step": 14900
+    },
+    {
+      "epoch": 230.76923076923077,
+      "grad_norm": 49.193206787109375,
+      "learning_rate": 8.501400000000001e-06,
+      "loss": 0.938,
+      "step": 15000
+    },
+    {
+      "epoch": 230.76923076923077,
+      "eval_loss": 1.1409271955490112,
+      "eval_runtime": 11.0858,
+      "eval_samples_per_second": 11.817,
+      "eval_steps_per_second": 1.533,
+      "step": 15000
+    },
+    {
+      "epoch": 232.30769230769232,
+      "grad_norm": 22.36850357055664,
+      "learning_rate": 8.491400000000001e-06,
+      "loss": 0.9218,
+      "step": 15100
+    },
+    {
+      "epoch": 232.30769230769232,
+      "eval_loss": 1.131103515625,
+      "eval_runtime": 11.0691,
+      "eval_samples_per_second": 11.835,
+      "eval_steps_per_second": 1.536,
+      "step": 15100
+    },
+    {
+      "epoch": 233.84615384615384,
+      "grad_norm": 26.34011459350586,
+      "learning_rate": 8.481400000000002e-06,
+      "loss": 0.9617,
+      "step": 15200
+    },
+    {
+      "epoch": 233.84615384615384,
+      "eval_loss": 1.1415542364120483,
+      "eval_runtime": 10.9887,
+      "eval_samples_per_second": 11.921,
+      "eval_steps_per_second": 1.547,
+      "step": 15200
+    },
+    {
+      "epoch": 235.3846153846154,
+      "grad_norm": 29.583358764648438,
+      "learning_rate": 8.4714e-06,
+      "loss": 0.9272,
+      "step": 15300
+    },
+    {
+      "epoch": 235.3846153846154,
+      "eval_loss": 1.144914150238037,
+      "eval_runtime": 11.2646,
+      "eval_samples_per_second": 11.629,
+      "eval_steps_per_second": 1.509,
+      "step": 15300
+    },
+    {
+      "epoch": 236.92307692307693,
+      "grad_norm": 31.824247360229492,
+      "learning_rate": 8.4614e-06,
+      "loss": 0.9207,
+      "step": 15400
+    },
+    {
+      "epoch": 236.92307692307693,
+      "eval_loss": 1.1387474536895752,
+      "eval_runtime": 11.0721,
+      "eval_samples_per_second": 11.832,
+      "eval_steps_per_second": 1.535,
+      "step": 15400
+    },
+    {
+      "epoch": 238.46153846153845,
+      "grad_norm": 41.94277572631836,
+      "learning_rate": 8.4514e-06,
+      "loss": 0.9454,
+      "step": 15500
+    },
+    {
+      "epoch": 238.46153846153845,
+      "eval_loss": 1.1316168308258057,
+      "eval_runtime": 11.1831,
+      "eval_samples_per_second": 11.714,
+      "eval_steps_per_second": 1.52,
+      "step": 15500
+    },
+    {
+      "epoch": 240.0,
+      "grad_norm": 21.150598526000977,
+      "learning_rate": 8.4414e-06,
+      "loss": 0.9249,
+      "step": 15600
+    },
+    {
+      "epoch": 240.0,
+      "eval_loss": 1.1368097066879272,
+      "eval_runtime": 11.0887,
+      "eval_samples_per_second": 11.814,
+      "eval_steps_per_second": 1.533,
+      "step": 15600
+    },
+    {
+      "epoch": 241.53846153846155,
+      "grad_norm": 47.432212829589844,
+      "learning_rate": 8.4314e-06,
+      "loss": 0.9212,
+      "step": 15700
+    },
+    {
+      "epoch": 241.53846153846155,
+      "eval_loss": 1.125348448753357,
+      "eval_runtime": 11.2434,
+      "eval_samples_per_second": 11.651,
+      "eval_steps_per_second": 1.512,
+      "step": 15700
+    },
+    {
+      "epoch": 243.07692307692307,
+      "grad_norm": 28.406036376953125,
+      "learning_rate": 8.4214e-06,
+      "loss": 0.9272,
+      "step": 15800
+    },
+    {
+      "epoch": 243.07692307692307,
+      "eval_loss": 1.1328097581863403,
+      "eval_runtime": 11.1097,
+      "eval_samples_per_second": 11.791,
+      "eval_steps_per_second": 1.53,
+      "step": 15800
+    },
+    {
+      "epoch": 244.6153846153846,
+      "grad_norm": 53.369564056396484,
+      "learning_rate": 8.4114e-06,
+      "loss": 0.9174,
+      "step": 15900
+    },
+    {
+      "epoch": 244.6153846153846,
+      "eval_loss": 1.1235861778259277,
+      "eval_runtime": 10.9581,
+      "eval_samples_per_second": 11.955,
+      "eval_steps_per_second": 1.551,
+      "step": 15900
+    },
+    {
+      "epoch": 246.15384615384616,
+      "grad_norm": 31.435935974121094,
+      "learning_rate": 8.4014e-06,
+      "loss": 0.9041,
+      "step": 16000
+    },
+    {
+      "epoch": 246.15384615384616,
+      "eval_loss": 1.1266686916351318,
+      "eval_runtime": 11.1276,
+      "eval_samples_per_second": 11.773,
+      "eval_steps_per_second": 1.528,
+      "step": 16000
+    },
+    {
+      "epoch": 247.69230769230768,
+      "grad_norm": 32.799991607666016,
+      "learning_rate": 8.3914e-06,
+      "loss": 0.9062,
+      "step": 16100
+    },
+    {
+      "epoch": 247.69230769230768,
+      "eval_loss": 1.1481693983078003,
+      "eval_runtime": 11.0435,
+      "eval_samples_per_second": 11.862,
+      "eval_steps_per_second": 1.539,
+      "step": 16100
+    },
+    {
+      "epoch": 249.23076923076923,
+      "grad_norm": 36.49935531616211,
+      "learning_rate": 8.3814e-06,
+      "loss": 0.9163,
+      "step": 16200
+    },
+    {
+      "epoch": 249.23076923076923,
+      "eval_loss": 1.139769434928894,
+      "eval_runtime": 11.1922,
+      "eval_samples_per_second": 11.705,
+      "eval_steps_per_second": 1.519,
+      "step": 16200
+    },
+    {
+      "epoch": 250.76923076923077,
+      "grad_norm": 35.4781379699707,
+      "learning_rate": 8.371400000000001e-06,
+      "loss": 0.9219,
+      "step": 16300
+    },
+    {
+      "epoch": 250.76923076923077,
+      "eval_loss": 1.1498539447784424,
+      "eval_runtime": 11.1805,
+      "eval_samples_per_second": 11.717,
+      "eval_steps_per_second": 1.52,
+      "step": 16300
+    },
+    {
+      "epoch": 252.30769230769232,
+      "grad_norm": 17.645612716674805,
+      "learning_rate": 8.361400000000001e-06,
+      "loss": 0.9278,
+      "step": 16400
+    },
+    {
+      "epoch": 252.30769230769232,
+      "eval_loss": 1.1338902711868286,
+      "eval_runtime": 11.1146,
+      "eval_samples_per_second": 11.786,
+      "eval_steps_per_second": 1.53,
+      "step": 16400
+    },
+    {
+      "epoch": 253.84615384615384,
+      "grad_norm": 32.81660079956055,
+      "learning_rate": 8.351400000000001e-06,
+      "loss": 0.9108,
+      "step": 16500
+    },
+    {
+      "epoch": 253.84615384615384,
+      "eval_loss": 1.1279574632644653,
+      "eval_runtime": 11.1151,
+      "eval_samples_per_second": 11.786,
+      "eval_steps_per_second": 1.529,
+      "step": 16500
+    },
+    {
+      "epoch": 255.3846153846154,
+      "grad_norm": 22.3878116607666,
+      "learning_rate": 8.341400000000001e-06,
+      "loss": 0.9011,
+      "step": 16600
+    },
+    {
+      "epoch": 255.3846153846154,
+      "eval_loss": 1.1570419073104858,
+      "eval_runtime": 11.0784,
+      "eval_samples_per_second": 11.825,
+      "eval_steps_per_second": 1.535,
+      "step": 16600
+    },
+    {
+      "epoch": 256.9230769230769,
+      "grad_norm": 29.845205307006836,
+      "learning_rate": 8.331400000000001e-06,
+      "loss": 0.9314,
+      "step": 16700
+    },
+    {
+      "epoch": 256.9230769230769,
+      "eval_loss": 1.1365561485290527,
+      "eval_runtime": 11.2405,
+      "eval_samples_per_second": 11.654,
+      "eval_steps_per_second": 1.512,
+      "step": 16700
+    },
+    {
+      "epoch": 258.46153846153845,
+      "grad_norm": 21.02674102783203,
+      "learning_rate": 8.321400000000001e-06,
+      "loss": 0.9021,
+      "step": 16800
+    },
+    {
+      "epoch": 258.46153846153845,
+      "eval_loss": 1.164974570274353,
+      "eval_runtime": 11.0643,
+      "eval_samples_per_second": 11.84,
+      "eval_steps_per_second": 1.536,
+      "step": 16800
+    },
+    {
+      "epoch": 260.0,
+      "grad_norm": 22.380117416381836,
+      "learning_rate": 8.3114e-06,
+      "loss": 0.912,
+      "step": 16900
+    },
+    {
+      "epoch": 260.0,
+      "eval_loss": 1.1483317613601685,
+      "eval_runtime": 11.1852,
+      "eval_samples_per_second": 11.712,
+      "eval_steps_per_second": 1.52,
+      "step": 16900
+    },
+    {
+      "epoch": 261.53846153846155,
+      "grad_norm": 39.20146560668945,
+      "learning_rate": 8.3014e-06,
+      "loss": 0.9165,
+      "step": 17000
+    },
+    {
+      "epoch": 261.53846153846155,
+      "eval_loss": 1.159449577331543,
+      "eval_runtime": 11.4058,
+      "eval_samples_per_second": 11.485,
+      "eval_steps_per_second": 1.49,
+      "step": 17000
+    },
+    {
+      "epoch": 263.0769230769231,
+      "grad_norm": 46.305389404296875,
+      "learning_rate": 8.2914e-06,
+      "loss": 0.916,
+      "step": 17100
+    },
+    {
+      "epoch": 263.0769230769231,
+      "eval_loss": 1.146033525466919,
+      "eval_runtime": 11.3638,
+      "eval_samples_per_second": 11.528,
+      "eval_steps_per_second": 1.496,
+      "step": 17100
+    },
+    {
+      "epoch": 264.61538461538464,
+      "grad_norm": 33.07489776611328,
+      "learning_rate": 8.2814e-06,
+      "loss": 0.9147,
+      "step": 17200
+    },
+    {
+      "epoch": 264.61538461538464,
+      "eval_loss": 1.143062710762024,
+      "eval_runtime": 11.3544,
+      "eval_samples_per_second": 11.537,
+      "eval_steps_per_second": 1.497,
+      "step": 17200
+    },
+    {
+      "epoch": 266.15384615384613,
+      "grad_norm": 35.233131408691406,
+      "learning_rate": 8.2714e-06,
+      "loss": 0.9151,
+      "step": 17300
+    },
+    {
+      "epoch": 266.15384615384613,
+      "eval_loss": 1.126172661781311,
+      "eval_runtime": 11.1185,
+      "eval_samples_per_second": 11.782,
+      "eval_steps_per_second": 1.529,
+      "step": 17300
+    },
+    {
+      "epoch": 267.6923076923077,
+      "grad_norm": 32.72975540161133,
+      "learning_rate": 8.2614e-06,
+      "loss": 0.8881,
+      "step": 17400
+    },
+    {
+      "epoch": 267.6923076923077,
+      "eval_loss": 1.1455607414245605,
+      "eval_runtime": 11.0568,
+      "eval_samples_per_second": 11.848,
+      "eval_steps_per_second": 1.538,
+      "step": 17400
+    },
+    {
+      "epoch": 269.2307692307692,
+      "grad_norm": 16.41983985900879,
+      "learning_rate": 8.2514e-06,
+      "loss": 0.9027,
+      "step": 17500
+    },
+    {
+      "epoch": 269.2307692307692,
+      "eval_loss": 1.1283539533615112,
+      "eval_runtime": 11.3233,
+      "eval_samples_per_second": 11.569,
+      "eval_steps_per_second": 1.501,
+      "step": 17500
+    },
+    {
+      "epoch": 270.7692307692308,
+      "grad_norm": 20.32726287841797,
+      "learning_rate": 8.2414e-06,
+      "loss": 0.9391,
+      "step": 17600
+    },
+    {
+      "epoch": 270.7692307692308,
+      "eval_loss": 1.124210000038147,
+      "eval_runtime": 11.3345,
+      "eval_samples_per_second": 11.558,
+      "eval_steps_per_second": 1.5,
+      "step": 17600
+    },
+    {
+      "epoch": 272.3076923076923,
+      "grad_norm": 23.14797019958496,
+      "learning_rate": 8.2314e-06,
+      "loss": 0.8899,
+      "step": 17700
+    },
+    {
+      "epoch": 272.3076923076923,
+      "eval_loss": 1.1297597885131836,
+      "eval_runtime": 11.1272,
+      "eval_samples_per_second": 11.773,
+      "eval_steps_per_second": 1.528,
+      "step": 17700
+    },
+    {
+      "epoch": 273.84615384615387,
+      "grad_norm": 18.778406143188477,
+      "learning_rate": 8.2214e-06,
+      "loss": 0.9074,
+      "step": 17800
+    },
+    {
+      "epoch": 273.84615384615387,
+      "eval_loss": 1.135562777519226,
+      "eval_runtime": 11.2964,
+      "eval_samples_per_second": 11.597,
+      "eval_steps_per_second": 1.505,
+      "step": 17800
+    },
+    {
+      "epoch": 275.38461538461536,
+      "grad_norm": 27.574323654174805,
+      "learning_rate": 8.2114e-06,
+      "loss": 0.8931,
+      "step": 17900
+    },
+    {
+      "epoch": 275.38461538461536,
+      "eval_loss": 1.1423242092132568,
+      "eval_runtime": 11.0992,
+      "eval_samples_per_second": 11.803,
+      "eval_steps_per_second": 1.532,
+      "step": 17900
+    },
+    {
+      "epoch": 276.9230769230769,
+      "grad_norm": 26.559467315673828,
+      "learning_rate": 8.2014e-06,
+      "loss": 0.8913,
+      "step": 18000
+    },
+    {
+      "epoch": 276.9230769230769,
+      "eval_loss": 1.1252741813659668,
+      "eval_runtime": 11.2765,
+      "eval_samples_per_second": 11.617,
+      "eval_steps_per_second": 1.508,
+      "step": 18000
+    },
+    {
+      "epoch": 278.46153846153845,
+      "grad_norm": 24.442596435546875,
+      "learning_rate": 8.191400000000001e-06,
+      "loss": 0.8993,
+      "step": 18100
+    },
+    {
+      "epoch": 278.46153846153845,
+      "eval_loss": 1.1197646856307983,
+      "eval_runtime": 11.0479,
+      "eval_samples_per_second": 11.857,
+      "eval_steps_per_second": 1.539,
+      "step": 18100
+    },
+    {
+      "epoch": 280.0,
+      "grad_norm": 42.99458694458008,
+      "learning_rate": 8.181400000000001e-06,
+      "loss": 0.8925,
+      "step": 18200
+    },
+    {
+      "epoch": 280.0,
+      "eval_loss": 1.129381775856018,
+      "eval_runtime": 11.1979,
+      "eval_samples_per_second": 11.699,
+      "eval_steps_per_second": 1.518,
+      "step": 18200
+    },
+    {
+      "epoch": 281.53846153846155,
+      "grad_norm": 38.08549118041992,
+      "learning_rate": 8.171400000000001e-06,
+      "loss": 0.8699,
+      "step": 18300
+    },
+    {
+      "epoch": 281.53846153846155,
+      "eval_loss": 1.1298307180404663,
+      "eval_runtime": 11.1097,
+      "eval_samples_per_second": 11.792,
+      "eval_steps_per_second": 1.53,
+      "step": 18300
+    },
+    {
+      "epoch": 283.0769230769231,
+      "grad_norm": 49.90501022338867,
+      "learning_rate": 8.161400000000001e-06,
+      "loss": 0.9207,
+      "step": 18400
+    },
+    {
+      "epoch": 283.0769230769231,
+      "eval_loss": 1.1229758262634277,
+      "eval_runtime": 11.284,
+      "eval_samples_per_second": 11.609,
+      "eval_steps_per_second": 1.507,
+      "step": 18400
+    },
+    {
+      "epoch": 284.61538461538464,
+      "grad_norm": 37.63615036010742,
+      "learning_rate": 8.1514e-06,
+      "loss": 0.9061,
+      "step": 18500
+    },
+    {
+      "epoch": 284.61538461538464,
+      "eval_loss": 1.1395354270935059,
+      "eval_runtime": 11.2087,
+      "eval_samples_per_second": 11.687,
+      "eval_steps_per_second": 1.517,
+      "step": 18500
+    },
+    {
+      "epoch": 286.15384615384613,
+      "grad_norm": 17.88991355895996,
+      "learning_rate": 8.1414e-06,
+      "loss": 0.8664,
+      "step": 18600
+    },
+    {
+      "epoch": 286.15384615384613,
+      "eval_loss": 1.1339645385742188,
+      "eval_runtime": 11.317,
+      "eval_samples_per_second": 11.576,
+      "eval_steps_per_second": 1.502,
+      "step": 18600
+    },
+    {
+      "epoch": 287.6923076923077,
+      "grad_norm": 33.13370895385742,
+      "learning_rate": 8.1314e-06,
+      "loss": 0.8759,
+      "step": 18700
+    },
+    {
+      "epoch": 287.6923076923077,
+      "eval_loss": 1.1445599794387817,
+      "eval_runtime": 11.0472,
+      "eval_samples_per_second": 11.858,
+      "eval_steps_per_second": 1.539,
+      "step": 18700
+    },
+    {
+      "epoch": 289.2307692307692,
+      "grad_norm": 22.776575088500977,
+      "learning_rate": 8.1214e-06,
+      "loss": 0.8889,
+      "step": 18800
+    },
+    {
+      "epoch": 289.2307692307692,
+      "eval_loss": 1.1401453018188477,
+      "eval_runtime": 11.2523,
+      "eval_samples_per_second": 11.642,
+      "eval_steps_per_second": 1.511,
+      "step": 18800
+    },
+    {
+      "epoch": 290.7692307692308,
+      "grad_norm": 19.893653869628906,
+      "learning_rate": 8.1114e-06,
+      "loss": 0.8945,
+      "step": 18900
+    },
+    {
+      "epoch": 290.7692307692308,
+      "eval_loss": 1.1185678243637085,
+      "eval_runtime": 10.9814,
+      "eval_samples_per_second": 11.929,
+      "eval_steps_per_second": 1.548,
+      "step": 18900
+    },
+    {
+      "epoch": 292.3076923076923,
+      "grad_norm": 35.09921646118164,
+      "learning_rate": 8.1015e-06,
+      "loss": 0.8821,
+      "step": 19000
+    },
+    {
+      "epoch": 292.3076923076923,
+      "eval_loss": 1.1313185691833496,
+      "eval_runtime": 11.2698,
+      "eval_samples_per_second": 11.624,
+      "eval_steps_per_second": 1.508,
+      "step": 19000
+    },
+    {
+      "epoch": 293.84615384615387,
+      "grad_norm": 36.43528366088867,
+      "learning_rate": 8.0915e-06,
+      "loss": 0.8794,
+      "step": 19100
+    },
+    {
+      "epoch": 293.84615384615387,
+      "eval_loss": 1.1506413221359253,
+      "eval_runtime": 10.9821,
+      "eval_samples_per_second": 11.928,
+      "eval_steps_per_second": 1.548,
+      "step": 19100
+    },
+    {
+      "epoch": 295.38461538461536,
+      "grad_norm": 24.73018455505371,
+      "learning_rate": 8.0815e-06,
+      "loss": 0.8856,
+      "step": 19200
+    },
+    {
+      "epoch": 295.38461538461536,
+      "eval_loss": 1.1280685663223267,
+      "eval_runtime": 11.2539,
+      "eval_samples_per_second": 11.64,
+      "eval_steps_per_second": 1.511,
+      "step": 19200
+    },
+    {
+      "epoch": 296.9230769230769,
+      "grad_norm": 50.01460647583008,
+      "learning_rate": 8.0715e-06,
+      "loss": 0.8532,
+      "step": 19300
+    },
+    {
+      "epoch": 296.9230769230769,
+      "eval_loss": 1.124637484550476,
+      "eval_runtime": 11.1106,
+      "eval_samples_per_second": 11.791,
+      "eval_steps_per_second": 1.53,
+      "step": 19300
+    },
+    {
+      "epoch": 298.46153846153845,
+      "grad_norm": 38.828033447265625,
+      "learning_rate": 8.0615e-06,
+      "loss": 0.8425,
+      "step": 19400
+    },
+    {
+      "epoch": 298.46153846153845,
+      "eval_loss": 1.1076736450195312,
+      "eval_runtime": 11.3645,
+      "eval_samples_per_second": 11.527,
+      "eval_steps_per_second": 1.496,
+      "step": 19400
+    },
+    {
+      "epoch": 300.0,
+      "grad_norm": 18.97437858581543,
+      "learning_rate": 8.0515e-06,
+      "loss": 0.8934,
+      "step": 19500
+    },
+    {
+      "epoch": 300.0,
+      "eval_loss": 1.110048532485962,
+      "eval_runtime": 11.1778,
+      "eval_samples_per_second": 11.72,
+      "eval_steps_per_second": 1.521,
+      "step": 19500
+    },
+    {
+      "epoch": 301.53846153846155,
+      "grad_norm": 49.44120407104492,
+      "learning_rate": 8.041500000000001e-06,
+      "loss": 0.8694,
+      "step": 19600
+    },
+    {
+      "epoch": 301.53846153846155,
+      "eval_loss": 1.1411337852478027,
+      "eval_runtime": 11.4024,
+      "eval_samples_per_second": 11.489,
+      "eval_steps_per_second": 1.491,
+      "step": 19600
+    },
+    {
+      "epoch": 303.0769230769231,
+      "grad_norm": 32.77436828613281,
+      "learning_rate": 8.031500000000001e-06,
+      "loss": 0.8876,
+      "step": 19700
+    },
+    {
+      "epoch": 303.0769230769231,
+      "eval_loss": 1.1274378299713135,
+      "eval_runtime": 11.3743,
+      "eval_samples_per_second": 11.517,
+      "eval_steps_per_second": 1.495,
+      "step": 19700
+    },
+    {
+      "epoch": 304.61538461538464,
+      "grad_norm": 33.37275695800781,
+      "learning_rate": 8.021500000000001e-06,
+      "loss": 0.839,
+      "step": 19800
+    },
+    {
+      "epoch": 304.61538461538464,
+      "eval_loss": 1.1159642934799194,
+      "eval_runtime": 11.2947,
+      "eval_samples_per_second": 11.598,
+      "eval_steps_per_second": 1.505,
+      "step": 19800
+    },
+    {
+      "epoch": 306.15384615384613,
+      "grad_norm": 15.981072425842285,
+      "learning_rate": 8.011500000000001e-06,
+      "loss": 0.8796,
+      "step": 19900
+    },
+    {
+      "epoch": 306.15384615384613,
+      "eval_loss": 1.1134443283081055,
+      "eval_runtime": 11.0923,
+      "eval_samples_per_second": 11.81,
+      "eval_steps_per_second": 1.533,
+      "step": 19900
+    },
+    {
+      "epoch": 307.6923076923077,
+      "grad_norm": 33.69294738769531,
+      "learning_rate": 8.001500000000001e-06,
+      "loss": 0.8757,
+      "step": 20000
+    },
+    {
+      "epoch": 307.6923076923077,
+      "eval_loss": 1.1174206733703613,
+      "eval_runtime": 11.1824,
+      "eval_samples_per_second": 11.715,
+      "eval_steps_per_second": 1.52,
+      "step": 20000
+    },
+    {
+      "epoch": 309.2307692307692,
+      "grad_norm": 39.100887298583984,
+      "learning_rate": 7.991500000000001e-06,
+      "loss": 0.9037,
+      "step": 20100
+    },
+    {
+      "epoch": 309.2307692307692,
+      "eval_loss": 1.1444469690322876,
+      "eval_runtime": 11.2517,
+      "eval_samples_per_second": 11.643,
+      "eval_steps_per_second": 1.511,
+      "step": 20100
+    },
+    {
+      "epoch": 310.7692307692308,
+      "grad_norm": 34.204410552978516,
+      "learning_rate": 7.981500000000001e-06,
+      "loss": 0.8714,
+      "step": 20200
+    },
+    {
+      "epoch": 310.7692307692308,
+      "eval_loss": 1.113229751586914,
+      "eval_runtime": 11.1634,
+      "eval_samples_per_second": 11.735,
+      "eval_steps_per_second": 1.523,
+      "step": 20200
+    },
+    {
+      "epoch": 312.3076923076923,
+      "grad_norm": 34.093692779541016,
+      "learning_rate": 7.971500000000002e-06,
+      "loss": 0.8952,
+      "step": 20300
+    },
+    {
+      "epoch": 312.3076923076923,
+      "eval_loss": 1.132067084312439,
+      "eval_runtime": 11.0613,
+      "eval_samples_per_second": 11.843,
+      "eval_steps_per_second": 1.537,
+      "step": 20300
+    },
+    {
+      "epoch": 313.84615384615387,
+      "grad_norm": 33.44735336303711,
+      "learning_rate": 7.961500000000002e-06,
+      "loss": 0.8545,
+      "step": 20400
+    },
+    {
+      "epoch": 313.84615384615387,
+      "eval_loss": 1.102783441543579,
+      "eval_runtime": 10.9986,
+      "eval_samples_per_second": 11.911,
+      "eval_steps_per_second": 1.546,
+      "step": 20400
+    },
+    {
+      "epoch": 315.38461538461536,
+      "grad_norm": 21.25609588623047,
+      "learning_rate": 7.9516e-06,
+      "loss": 0.896,
+      "step": 20500
+    },
+    {
+      "epoch": 315.38461538461536,
+      "eval_loss": 1.1276897192001343,
+      "eval_runtime": 11.0387,
+      "eval_samples_per_second": 11.867,
+      "eval_steps_per_second": 1.54,
+      "step": 20500
+    },
+    {
+      "epoch": 316.9230769230769,
+      "grad_norm": 42.066017150878906,
+      "learning_rate": 7.9416e-06,
+      "loss": 0.8545,
+      "step": 20600
+    },
+    {
+      "epoch": 316.9230769230769,
+      "eval_loss": 1.1395957469940186,
+      "eval_runtime": 11.2855,
+      "eval_samples_per_second": 11.608,
+      "eval_steps_per_second": 1.506,
+      "step": 20600
+    },
+    {
+      "epoch": 318.46153846153845,
+      "grad_norm": 22.6724796295166,
+      "learning_rate": 7.9316e-06,
+      "loss": 0.8838,
+      "step": 20700
+    },
+    {
+      "epoch": 318.46153846153845,
+      "eval_loss": 1.1254605054855347,
+      "eval_runtime": 11.2226,
+      "eval_samples_per_second": 11.673,
+      "eval_steps_per_second": 1.515,
+      "step": 20700
+    },
+    {
+      "epoch": 320.0,
+      "grad_norm": 31.141693115234375,
+      "learning_rate": 7.9216e-06,
+      "loss": 0.8704,
+      "step": 20800
+    },
+    {
+      "epoch": 320.0,
+      "eval_loss": 1.1148459911346436,
+      "eval_runtime": 11.3548,
+      "eval_samples_per_second": 11.537,
+      "eval_steps_per_second": 1.497,
+      "step": 20800
+    },
+    {
+      "epoch": 321.53846153846155,
+      "grad_norm": 53.709205627441406,
+      "learning_rate": 7.9116e-06,
+      "loss": 0.8571,
+      "step": 20900
+    },
+    {
+      "epoch": 321.53846153846155,
+      "eval_loss": 1.1235041618347168,
+      "eval_runtime": 11.3363,
+      "eval_samples_per_second": 11.556,
+      "eval_steps_per_second": 1.5,
+      "step": 20900
+    },
+    {
+      "epoch": 323.0769230769231,
+      "grad_norm": 36.34547805786133,
+      "learning_rate": 7.9016e-06,
+      "loss": 0.8749,
+      "step": 21000
+    },
+    {
+      "epoch": 323.0769230769231,
+      "eval_loss": 1.1102306842803955,
+      "eval_runtime": 11.2154,
+      "eval_samples_per_second": 11.68,
+      "eval_steps_per_second": 1.516,
+      "step": 21000
+    },
+    {
+      "epoch": 324.61538461538464,
+      "grad_norm": 19.974227905273438,
+      "learning_rate": 7.8916e-06,
+      "loss": 0.8572,
+      "step": 21100
+    },
+    {
+      "epoch": 324.61538461538464,
+      "eval_loss": 1.1081122159957886,
+      "eval_runtime": 11.2322,
+      "eval_samples_per_second": 11.663,
+      "eval_steps_per_second": 1.513,
+      "step": 21100
+    },
+    {
+      "epoch": 326.15384615384613,
+      "grad_norm": 31.686569213867188,
+      "learning_rate": 7.881600000000001e-06,
+      "loss": 0.8689,
+      "step": 21200
+    },
+    {
+      "epoch": 326.15384615384613,
+      "eval_loss": 1.1191128492355347,
+      "eval_runtime": 11.4469,
+      "eval_samples_per_second": 11.444,
+      "eval_steps_per_second": 1.485,
+      "step": 21200
+    },
+    {
+      "epoch": 327.6923076923077,
+      "grad_norm": 35.97456359863281,
+      "learning_rate": 7.871600000000001e-06,
+      "loss": 0.8421,
+      "step": 21300
+    },
+    {
+      "epoch": 327.6923076923077,
+      "eval_loss": 1.1023869514465332,
+      "eval_runtime": 11.2157,
+      "eval_samples_per_second": 11.68,
+      "eval_steps_per_second": 1.516,
+      "step": 21300
+    },
+    {
+      "epoch": 329.2307692307692,
+      "grad_norm": 34.61111068725586,
+      "learning_rate": 7.861600000000001e-06,
+      "loss": 0.8546,
+      "step": 21400
+    },
+    {
+      "epoch": 329.2307692307692,
+      "eval_loss": 1.1294784545898438,
+      "eval_runtime": 11.1536,
+      "eval_samples_per_second": 11.745,
+      "eval_steps_per_second": 1.524,
+      "step": 21400
+    },
+    {
+      "epoch": 330.7692307692308,
+      "grad_norm": 21.682897567749023,
+      "learning_rate": 7.8516e-06,
+      "loss": 0.828,
+      "step": 21500
+    },
+    {
+      "epoch": 330.7692307692308,
+      "eval_loss": 1.101345181465149,
+      "eval_runtime": 11.0361,
+      "eval_samples_per_second": 11.87,
+      "eval_steps_per_second": 1.54,
+      "step": 21500
+    },
+    {
+      "epoch": 332.3076923076923,
+      "grad_norm": 31.514686584472656,
+      "learning_rate": 7.8416e-06,
+      "loss": 0.8762,
+      "step": 21600
+    },
+    {
+      "epoch": 332.3076923076923,
+      "eval_loss": 1.1125372648239136,
+      "eval_runtime": 11.2868,
+      "eval_samples_per_second": 11.607,
+      "eval_steps_per_second": 1.506,
+      "step": 21600
+    },
+    {
+      "epoch": 333.84615384615387,
+      "grad_norm": 34.57918167114258,
+      "learning_rate": 7.8316e-06,
+      "loss": 0.8368,
+      "step": 21700
+    },
+    {
+      "epoch": 333.84615384615387,
+      "eval_loss": 1.1226669549942017,
+      "eval_runtime": 11.4486,
+      "eval_samples_per_second": 11.442,
+      "eval_steps_per_second": 1.485,
+      "step": 21700
+    },
+    {
+      "epoch": 335.38461538461536,
+      "grad_norm": 57.26470947265625,
+      "learning_rate": 7.8216e-06,
+      "loss": 0.8675,
+      "step": 21800
+    },
+    {
+      "epoch": 335.38461538461536,
+      "eval_loss": 1.1287323236465454,
+      "eval_runtime": 11.197,
+      "eval_samples_per_second": 11.7,
+      "eval_steps_per_second": 1.518,
+      "step": 21800
+    },
+    {
+      "epoch": 336.9230769230769,
+      "grad_norm": 26.583576202392578,
+      "learning_rate": 7.8116e-06,
+      "loss": 0.8555,
+      "step": 21900
+    },
+    {
+      "epoch": 336.9230769230769,
+      "eval_loss": 1.1035945415496826,
+      "eval_runtime": 11.2619,
+      "eval_samples_per_second": 11.632,
+      "eval_steps_per_second": 1.51,
+      "step": 21900
+    },
+    {
+      "epoch": 338.46153846153845,
+      "grad_norm": 44.196590423583984,
+      "learning_rate": 7.8016e-06,
+      "loss": 0.8601,
+      "step": 22000
+    },
+    {
+      "epoch": 338.46153846153845,
+      "eval_loss": 1.09674870967865,
+      "eval_runtime": 11.2173,
+      "eval_samples_per_second": 11.678,
+      "eval_steps_per_second": 1.516,
+      "step": 22000
+    },
+    {
+      "epoch": 340.0,
+      "grad_norm": 34.388145446777344,
+      "learning_rate": 7.7916e-06,
+      "loss": 0.8347,
+      "step": 22100
+    },
+    {
+      "epoch": 340.0,
+      "eval_loss": 1.100284218788147,
+      "eval_runtime": 11.2123,
+      "eval_samples_per_second": 11.684,
+      "eval_steps_per_second": 1.516,
+      "step": 22100
+    },
+    {
+      "epoch": 341.53846153846155,
+      "grad_norm": 33.55098342895508,
+      "learning_rate": 7.7816e-06,
+      "loss": 0.8429,
+      "step": 22200
+    },
+    {
+      "epoch": 341.53846153846155,
+      "eval_loss": 1.1070573329925537,
+      "eval_runtime": 11.1389,
+      "eval_samples_per_second": 11.761,
+      "eval_steps_per_second": 1.526,
+      "step": 22200
+    },
+    {
+      "epoch": 343.0769230769231,
+      "grad_norm": 21.367860794067383,
+      "learning_rate": 7.7716e-06,
+      "loss": 0.827,
+      "step": 22300
+    },
+    {
+      "epoch": 343.0769230769231,
+      "eval_loss": 1.1184412240982056,
+      "eval_runtime": 11.2313,
+      "eval_samples_per_second": 11.664,
+      "eval_steps_per_second": 1.514,
+      "step": 22300
+    },
+    {
+      "epoch": 344.61538461538464,
+      "grad_norm": 32.62434387207031,
+      "learning_rate": 7.7616e-06,
+      "loss": 0.8566,
+      "step": 22400
+    },
+    {
+      "epoch": 344.61538461538464,
+      "eval_loss": 1.1258198022842407,
+      "eval_runtime": 11.1788,
+      "eval_samples_per_second": 11.719,
+      "eval_steps_per_second": 1.521,
+      "step": 22400
+    },
+    {
+      "epoch": 346.15384615384613,
+      "grad_norm": 44.411659240722656,
+      "learning_rate": 7.7516e-06,
+      "loss": 0.8666,
+      "step": 22500
+    },
+    {
+      "epoch": 346.15384615384613,
+      "eval_loss": 1.121845006942749,
+      "eval_runtime": 11.1251,
+      "eval_samples_per_second": 11.775,
+      "eval_steps_per_second": 1.528,
+      "step": 22500
+    },
+    {
+      "epoch": 347.6923076923077,
+      "grad_norm": 41.10401153564453,
+      "learning_rate": 7.7416e-06,
+      "loss": 0.8354,
+      "step": 22600
+    },
+    {
+      "epoch": 347.6923076923077,
+      "eval_loss": 1.1154077053070068,
+      "eval_runtime": 11.161,
+      "eval_samples_per_second": 11.737,
+      "eval_steps_per_second": 1.523,
+      "step": 22600
+    },
+    {
+      "epoch": 349.2307692307692,
+      "grad_norm": 31.414745330810547,
+      "learning_rate": 7.7316e-06,
+      "loss": 0.8321,
+      "step": 22700
+    },
+    {
+      "epoch": 349.2307692307692,
+      "eval_loss": 1.1233044862747192,
+      "eval_runtime": 11.0203,
+      "eval_samples_per_second": 11.887,
+      "eval_steps_per_second": 1.543,
+      "step": 22700
+    },
+    {
+      "epoch": 350.7692307692308,
+      "grad_norm": 37.39424133300781,
+      "learning_rate": 7.7216e-06,
+      "loss": 0.8384,
+      "step": 22800
+    },
+    {
+      "epoch": 350.7692307692308,
+      "eval_loss": 1.1059685945510864,
+      "eval_runtime": 10.9825,
+      "eval_samples_per_second": 11.928,
+      "eval_steps_per_second": 1.548,
+      "step": 22800
+    },
+    {
+      "epoch": 352.3076923076923,
+      "grad_norm": 24.75067901611328,
+      "learning_rate": 7.7116e-06,
+      "loss": 0.8216,
+      "step": 22900
+    },
+    {
+      "epoch": 352.3076923076923,
+      "eval_loss": 1.1147040128707886,
+      "eval_runtime": 11.1066,
+      "eval_samples_per_second": 11.795,
+      "eval_steps_per_second": 1.531,
+      "step": 22900
+    },
+    {
+      "epoch": 353.84615384615387,
+      "grad_norm": 21.775754928588867,
+      "learning_rate": 7.701600000000001e-06,
+      "loss": 0.8562,
+      "step": 23000
+    },
+    {
+      "epoch": 353.84615384615387,
+      "eval_loss": 1.1105334758758545,
+      "eval_runtime": 11.0277,
+      "eval_samples_per_second": 11.879,
+      "eval_steps_per_second": 1.542,
+      "step": 23000
+    },
+    {
+      "epoch": 355.38461538461536,
+      "grad_norm": 33.22850799560547,
+      "learning_rate": 7.6916e-06,
+      "loss": 0.813,
+      "step": 23100
+    },
+    {
+      "epoch": 355.38461538461536,
+      "eval_loss": 1.113241195678711,
+      "eval_runtime": 10.9954,
+      "eval_samples_per_second": 11.914,
+      "eval_steps_per_second": 1.546,
+      "step": 23100
+    },
+    {
+      "epoch": 356.9230769230769,
+      "grad_norm": 31.402652740478516,
+      "learning_rate": 7.6816e-06,
+      "loss": 0.8527,
+      "step": 23200
+    },
+    {
+      "epoch": 356.9230769230769,
+      "eval_loss": 1.1040674448013306,
+      "eval_runtime": 11.0415,
+      "eval_samples_per_second": 11.864,
+      "eval_steps_per_second": 1.54,
+      "step": 23200
+    },
+    {
+      "epoch": 358.46153846153845,
+      "grad_norm": 106.14508819580078,
+      "learning_rate": 7.6716e-06,
+      "loss": 0.8223,
+      "step": 23300
+    },
+    {
+      "epoch": 358.46153846153845,
+      "eval_loss": 1.1121721267700195,
+      "eval_runtime": 11.1579,
+      "eval_samples_per_second": 11.741,
+      "eval_steps_per_second": 1.524,
+      "step": 23300
+    },
+    {
+      "epoch": 360.0,
+      "grad_norm": 32.76395797729492,
+      "learning_rate": 7.6616e-06,
+      "loss": 0.8199,
+      "step": 23400
+    },
+    {
+      "epoch": 360.0,
+      "eval_loss": 1.109994649887085,
+      "eval_runtime": 11.0068,
+      "eval_samples_per_second": 11.902,
+      "eval_steps_per_second": 1.545,
+      "step": 23400
+    },
+    {
+      "epoch": 361.53846153846155,
+      "grad_norm": 44.64813232421875,
+      "learning_rate": 7.6516e-06,
+      "loss": 0.8566,
+      "step": 23500
+    },
+    {
+      "epoch": 361.53846153846155,
+      "eval_loss": 1.1383651494979858,
+      "eval_runtime": 11.1866,
+      "eval_samples_per_second": 11.71,
+      "eval_steps_per_second": 1.52,
+      "step": 23500
+    },
+    {
+      "epoch": 363.0769230769231,
+      "grad_norm": 20.096393585205078,
+      "learning_rate": 7.6416e-06,
+      "loss": 0.795,
+      "step": 23600
+    },
+    {
+      "epoch": 363.0769230769231,
+      "eval_loss": 1.1200674772262573,
+      "eval_runtime": 11.289,
+      "eval_samples_per_second": 11.604,
+      "eval_steps_per_second": 1.506,
+      "step": 23600
+    },
+    {
+      "epoch": 364.61538461538464,
+      "grad_norm": 30.219837188720703,
+      "learning_rate": 7.6316e-06,
+      "loss": 0.8413,
+      "step": 23700
+    },
+    {
+      "epoch": 364.61538461538464,
+      "eval_loss": 1.111639380455017,
+      "eval_runtime": 11.2547,
+      "eval_samples_per_second": 11.64,
+      "eval_steps_per_second": 1.51,
+      "step": 23700
+    },
+    {
+      "epoch": 366.15384615384613,
+      "grad_norm": 54.91815185546875,
+      "learning_rate": 7.6216e-06,
+      "loss": 0.8284,
+      "step": 23800
+    },
+    {
+      "epoch": 366.15384615384613,
+      "eval_loss": 1.1330946683883667,
+      "eval_runtime": 11.3054,
+      "eval_samples_per_second": 11.587,
+      "eval_steps_per_second": 1.504,
+      "step": 23800
+    },
+    {
+      "epoch": 367.6923076923077,
+      "grad_norm": 40.97207260131836,
+      "learning_rate": 7.6116e-06,
+      "loss": 0.8387,
+      "step": 23900
+    },
+    {
+      "epoch": 367.6923076923077,
+      "eval_loss": 1.1550222635269165,
+      "eval_runtime": 11.2372,
+      "eval_samples_per_second": 11.658,
+      "eval_steps_per_second": 1.513,
+      "step": 23900
+    },
+    {
+      "epoch": 369.2307692307692,
+      "grad_norm": 36.580745697021484,
+      "learning_rate": 7.6017e-06,
+      "loss": 0.829,
+      "step": 24000
+    },
+    {
+      "epoch": 369.2307692307692,
+      "eval_loss": 1.1261051893234253,
+      "eval_runtime": 11.1834,
+      "eval_samples_per_second": 11.714,
+      "eval_steps_per_second": 1.52,
+      "step": 24000
+    },
+    {
+      "epoch": 370.7692307692308,
+      "grad_norm": 46.304649353027344,
+      "learning_rate": 7.5917000000000005e-06,
+      "loss": 0.8316,
+      "step": 24100
+    },
+    {
+      "epoch": 370.7692307692308,
+      "eval_loss": 1.1410084962844849,
+      "eval_runtime": 11.3511,
+      "eval_samples_per_second": 11.541,
+      "eval_steps_per_second": 1.498,
+      "step": 24100
+    },
+    {
+      "epoch": 372.3076923076923,
+      "grad_norm": 32.472293853759766,
+      "learning_rate": 7.581700000000001e-06,
+      "loss": 0.8309,
+      "step": 24200
+    },
+    {
+      "epoch": 372.3076923076923,
+      "eval_loss": 1.1283127069473267,
+      "eval_runtime": 11.1949,
+      "eval_samples_per_second": 11.702,
+      "eval_steps_per_second": 1.519,
+      "step": 24200
+    },
+    {
+      "epoch": 373.84615384615387,
+      "grad_norm": 24.16025161743164,
+      "learning_rate": 7.571700000000001e-06,
+      "loss": 0.8154,
+      "step": 24300
+    },
+    {
+      "epoch": 373.84615384615387,
+      "eval_loss": 1.1386806964874268,
+      "eval_runtime": 11.1089,
+      "eval_samples_per_second": 11.792,
+      "eval_steps_per_second": 1.53,
+      "step": 24300
+    },
+    {
+      "epoch": 375.38461538461536,
+      "grad_norm": 16.86387825012207,
+      "learning_rate": 7.561700000000001e-06,
+      "loss": 0.8221,
+      "step": 24400
+    },
+    {
+      "epoch": 375.38461538461536,
+      "eval_loss": 1.09730863571167,
+      "eval_runtime": 11.1299,
+      "eval_samples_per_second": 11.77,
+      "eval_steps_per_second": 1.527,
+      "step": 24400
+    },
+    {
+      "epoch": 376.9230769230769,
+      "grad_norm": 34.62860107421875,
+      "learning_rate": 7.551700000000001e-06,
+      "loss": 0.8106,
+      "step": 24500
+    },
+    {
+      "epoch": 376.9230769230769,
+      "eval_loss": 1.1233881711959839,
+      "eval_runtime": 11.0352,
+      "eval_samples_per_second": 11.871,
+      "eval_steps_per_second": 1.541,
+      "step": 24500
+    },
+    {
+      "epoch": 378.46153846153845,
+      "grad_norm": 16.957393646240234,
+      "learning_rate": 7.541700000000001e-06,
+      "loss": 0.8362,
+      "step": 24600
+    },
+    {
+      "epoch": 378.46153846153845,
+      "eval_loss": 1.1209690570831299,
+      "eval_runtime": 11.0555,
+      "eval_samples_per_second": 11.849,
+      "eval_steps_per_second": 1.538,
+      "step": 24600
+    },
+    {
+      "epoch": 380.0,
+      "grad_norm": 21.519577026367188,
+      "learning_rate": 7.531700000000001e-06,
+      "loss": 0.8139,
+      "step": 24700
+    },
+    {
+      "epoch": 380.0,
+      "eval_loss": 1.1106966733932495,
+      "eval_runtime": 11.0492,
+      "eval_samples_per_second": 11.856,
+      "eval_steps_per_second": 1.539,
+      "step": 24700
+    },
+    {
+      "epoch": 381.53846153846155,
+      "grad_norm": 26.86697769165039,
+      "learning_rate": 7.5217e-06,
+      "loss": 0.8036,
+      "step": 24800
+    },
+    {
+      "epoch": 381.53846153846155,
+      "eval_loss": 1.1153205633163452,
+      "eval_runtime": 11.1284,
+      "eval_samples_per_second": 11.772,
+      "eval_steps_per_second": 1.528,
+      "step": 24800
+    },
+    {
+      "epoch": 383.0769230769231,
+      "grad_norm": 21.1047420501709,
+      "learning_rate": 7.5117000000000004e-06,
+      "loss": 0.8115,
+      "step": 24900
+    },
+    {
+      "epoch": 383.0769230769231,
+      "eval_loss": 1.116546630859375,
+      "eval_runtime": 11.4261,
+      "eval_samples_per_second": 11.465,
+      "eval_steps_per_second": 1.488,
+      "step": 24900
+    },
+    {
+      "epoch": 384.61538461538464,
+      "grad_norm": 40.87736892700195,
+      "learning_rate": 7.5017000000000005e-06,
+      "loss": 0.8146,
+      "step": 25000
+    },
+    {
+      "epoch": 384.61538461538464,
+      "eval_loss": 1.1162028312683105,
+      "eval_runtime": 11.3423,
+      "eval_samples_per_second": 11.55,
+      "eval_steps_per_second": 1.499,
+      "step": 25000
+    },
+    {
+      "epoch": 386.15384615384613,
+      "grad_norm": 19.28094482421875,
+      "learning_rate": 7.491700000000001e-06,
+      "loss": 0.8114,
+      "step": 25100
+    },
+    {
+      "epoch": 386.15384615384613,
+      "eval_loss": 1.106558918952942,
+      "eval_runtime": 11.0618,
+      "eval_samples_per_second": 11.843,
+      "eval_steps_per_second": 1.537,
+      "step": 25100
+    },
+    {
+      "epoch": 387.6923076923077,
+      "grad_norm": 19.689420700073242,
+      "learning_rate": 7.481700000000001e-06,
+      "loss": 0.8076,
+      "step": 25200
+    },
+    {
+      "epoch": 387.6923076923077,
+      "eval_loss": 1.1141376495361328,
+      "eval_runtime": 11.0491,
+      "eval_samples_per_second": 11.856,
+      "eval_steps_per_second": 1.539,
+      "step": 25200
+    },
+    {
+      "epoch": 389.2307692307692,
+      "grad_norm": 22.76107406616211,
+      "learning_rate": 7.471700000000001e-06,
+      "loss": 0.8077,
+      "step": 25300
+    },
+    {
+      "epoch": 389.2307692307692,
+      "eval_loss": 1.1186425685882568,
+      "eval_runtime": 11.1798,
+      "eval_samples_per_second": 11.718,
+      "eval_steps_per_second": 1.521,
+      "step": 25300
+    },
+    {
+      "epoch": 390.7692307692308,
+      "grad_norm": 40.23360061645508,
+      "learning_rate": 7.461700000000001e-06,
+      "loss": 0.7924,
+      "step": 25400
+    },
+    {
+      "epoch": 390.7692307692308,
+      "eval_loss": 1.1189498901367188,
+      "eval_runtime": 11.1824,
+      "eval_samples_per_second": 11.715,
+      "eval_steps_per_second": 1.52,
+      "step": 25400
+    },
+    {
+      "epoch": 392.3076923076923,
+      "grad_norm": 32.873207092285156,
+      "learning_rate": 7.451700000000001e-06,
+      "loss": 0.8335,
+      "step": 25500
+    },
+    {
+      "epoch": 392.3076923076923,
+      "eval_loss": 1.1451479196548462,
+      "eval_runtime": 11.1543,
+      "eval_samples_per_second": 11.744,
+      "eval_steps_per_second": 1.524,
+      "step": 25500
+    },
+    {
+      "epoch": 393.84615384615387,
+      "grad_norm": 27.307552337646484,
+      "learning_rate": 7.4417e-06,
+      "loss": 0.7926,
+      "step": 25600
+    },
+    {
+      "epoch": 393.84615384615387,
+      "eval_loss": 1.1189228296279907,
+      "eval_runtime": 11.0875,
+      "eval_samples_per_second": 11.815,
+      "eval_steps_per_second": 1.533,
+      "step": 25600
+    },
+    {
+      "epoch": 395.38461538461536,
+      "grad_norm": 22.984905242919922,
+      "learning_rate": 7.4317e-06,
+      "loss": 0.8039,
+      "step": 25700
+    },
+    {
+      "epoch": 395.38461538461536,
+      "eval_loss": 1.1381311416625977,
+      "eval_runtime": 11.1226,
+      "eval_samples_per_second": 11.778,
+      "eval_steps_per_second": 1.528,
+      "step": 25700
+    },
+    {
+      "epoch": 396.9230769230769,
+      "grad_norm": 43.75572967529297,
+      "learning_rate": 7.4217000000000004e-06,
+      "loss": 0.8426,
+      "step": 25800
+    },
+    {
+      "epoch": 396.9230769230769,
+      "eval_loss": 1.1005278825759888,
+      "eval_runtime": 10.9018,
+      "eval_samples_per_second": 12.016,
+      "eval_steps_per_second": 1.559,
+      "step": 25800
+    },
+    {
+      "epoch": 398.46153846153845,
+      "grad_norm": 45.551212310791016,
+      "learning_rate": 7.4117000000000005e-06,
+      "loss": 0.7918,
+      "step": 25900
+    },
+    {
+      "epoch": 398.46153846153845,
+      "eval_loss": 1.1017777919769287,
+      "eval_runtime": 10.8997,
+      "eval_samples_per_second": 12.019,
+      "eval_steps_per_second": 1.56,
+      "step": 25900
+    },
+    {
+      "epoch": 400.0,
+      "grad_norm": 27.237998962402344,
+      "learning_rate": 7.401700000000001e-06,
+      "loss": 0.825,
+      "step": 26000
+    },
+    {
+      "epoch": 400.0,
+      "eval_loss": 1.1006879806518555,
+      "eval_runtime": 10.8272,
+      "eval_samples_per_second": 12.099,
+      "eval_steps_per_second": 1.57,
+      "step": 26000
+    },
+    {
+      "epoch": 401.53846153846155,
+      "grad_norm": 20.51366424560547,
+      "learning_rate": 7.391700000000001e-06,
+      "loss": 0.8069,
+      "step": 26100
+    },
+    {
+      "epoch": 401.53846153846155,
+      "eval_loss": 1.121657371520996,
+      "eval_runtime": 10.956,
+      "eval_samples_per_second": 11.957,
+      "eval_steps_per_second": 1.552,
+      "step": 26100
+    },
+    {
+      "epoch": 403.0769230769231,
+      "grad_norm": 38.978363037109375,
+      "learning_rate": 7.381700000000001e-06,
+      "loss": 0.8105,
+      "step": 26200
+    },
+    {
+      "epoch": 403.0769230769231,
+      "eval_loss": 1.1141672134399414,
+      "eval_runtime": 10.8817,
+      "eval_samples_per_second": 12.039,
+      "eval_steps_per_second": 1.562,
+      "step": 26200
+    },
+    {
+      "epoch": 404.61538461538464,
+      "grad_norm": 40.20176315307617,
+      "learning_rate": 7.371700000000001e-06,
+      "loss": 0.7953,
+      "step": 26300
+    },
+    {
+      "epoch": 404.61538461538464,
+      "eval_loss": 1.1042087078094482,
+      "eval_runtime": 10.8826,
+      "eval_samples_per_second": 12.038,
+      "eval_steps_per_second": 1.562,
+      "step": 26300
+    },
+    {
+      "epoch": 406.15384615384613,
+      "grad_norm": 27.35833168029785,
+      "learning_rate": 7.3617e-06,
+      "loss": 0.8075,
+      "step": 26400
+    },
+    {
+      "epoch": 406.15384615384613,
+      "eval_loss": 1.1115680932998657,
+      "eval_runtime": 10.9465,
+      "eval_samples_per_second": 11.967,
+      "eval_steps_per_second": 1.553,
+      "step": 26400
+    },
+    {
+      "epoch": 407.6923076923077,
+      "grad_norm": 32.054351806640625,
+      "learning_rate": 7.3517e-06,
+      "loss": 0.7958,
+      "step": 26500
+    },
+    {
+      "epoch": 407.6923076923077,
+      "eval_loss": 1.1209728717803955,
+      "eval_runtime": 10.8226,
+      "eval_samples_per_second": 12.104,
+      "eval_steps_per_second": 1.571,
+      "step": 26500
+    },
+    {
+      "epoch": 409.2307692307692,
+      "grad_norm": 37.53184509277344,
+      "learning_rate": 7.3417e-06,
+      "loss": 0.7859,
+      "step": 26600
+    },
+    {
+      "epoch": 409.2307692307692,
+      "eval_loss": 1.0920478105545044,
+      "eval_runtime": 10.8488,
+      "eval_samples_per_second": 12.075,
+      "eval_steps_per_second": 1.567,
+      "step": 26600
+    },
+    {
+      "epoch": 410.7692307692308,
+      "grad_norm": 39.949039459228516,
+      "learning_rate": 7.3317000000000005e-06,
+      "loss": 0.8,
+      "step": 26700
+    },
+    {
+      "epoch": 410.7692307692308,
+      "eval_loss": 1.0909216403961182,
+      "eval_runtime": 10.9975,
+      "eval_samples_per_second": 11.912,
+      "eval_steps_per_second": 1.546,
+      "step": 26700
+    },
+    {
+      "epoch": 412.3076923076923,
+      "grad_norm": 28.89598846435547,
+      "learning_rate": 7.3217000000000006e-06,
+      "loss": 0.8168,
+      "step": 26800
+    },
+    {
+      "epoch": 412.3076923076923,
+      "eval_loss": 1.108217477798462,
+      "eval_runtime": 10.9623,
+      "eval_samples_per_second": 11.95,
+      "eval_steps_per_second": 1.551,
+      "step": 26800
+    },
+    {
+      "epoch": 413.84615384615387,
+      "grad_norm": 33.56279754638672,
+      "learning_rate": 7.311700000000001e-06,
+      "loss": 0.7854,
+      "step": 26900
+    },
+    {
+      "epoch": 413.84615384615387,
+      "eval_loss": 1.1230065822601318,
+      "eval_runtime": 10.8811,
+      "eval_samples_per_second": 12.039,
+      "eval_steps_per_second": 1.562,
+      "step": 26900
+    },
+    {
+      "epoch": 415.38461538461536,
+      "grad_norm": 26.578338623046875,
+      "learning_rate": 7.301700000000001e-06,
+      "loss": 0.7954,
+      "step": 27000
+    },
+    {
+      "epoch": 415.38461538461536,
+      "eval_loss": 1.1114639043807983,
+      "eval_runtime": 10.9172,
+      "eval_samples_per_second": 11.999,
+      "eval_steps_per_second": 1.557,
+      "step": 27000
+    },
+    {
+      "epoch": 416.9230769230769,
+      "grad_norm": 31.6333065032959,
+      "learning_rate": 7.291700000000001e-06,
+      "loss": 0.8057,
+      "step": 27100
+    },
+    {
+      "epoch": 416.9230769230769,
+      "eval_loss": 1.1136469841003418,
+      "eval_runtime": 10.835,
+      "eval_samples_per_second": 12.09,
+      "eval_steps_per_second": 1.569,
+      "step": 27100
+    },
+    {
+      "epoch": 418.46153846153845,
+      "grad_norm": 17.80211067199707,
+      "learning_rate": 7.2817e-06,
+      "loss": 0.8074,
+      "step": 27200
+    },
+    {
+      "epoch": 418.46153846153845,
+      "eval_loss": 1.0823259353637695,
+      "eval_runtime": 11.1165,
+      "eval_samples_per_second": 11.784,
+      "eval_steps_per_second": 1.529,
+      "step": 27200
+    },
+    {
+      "epoch": 420.0,
+      "grad_norm": 50.79414367675781,
+      "learning_rate": 7.2717e-06,
+      "loss": 0.778,
+      "step": 27300
+    },
+    {
+      "epoch": 420.0,
+      "eval_loss": 1.0840650796890259,
+      "eval_runtime": 10.9207,
+      "eval_samples_per_second": 11.996,
+      "eval_steps_per_second": 1.557,
+      "step": 27300
+    },
+    {
+      "epoch": 421.53846153846155,
+      "grad_norm": 27.64950942993164,
+      "learning_rate": 7.2617e-06,
+      "loss": 0.79,
+      "step": 27400
+    },
+    {
+      "epoch": 421.53846153846155,
+      "eval_loss": 1.1120820045471191,
+      "eval_runtime": 10.8896,
+      "eval_samples_per_second": 12.03,
+      "eval_steps_per_second": 1.561,
+      "step": 27400
+    },
+    {
+      "epoch": 423.0769230769231,
+      "grad_norm": 28.115442276000977,
+      "learning_rate": 7.2517e-06,
+      "loss": 0.7927,
+      "step": 27500
+    },
+    {
+      "epoch": 423.0769230769231,
+      "eval_loss": 1.0890986919403076,
+      "eval_runtime": 10.8766,
+      "eval_samples_per_second": 12.044,
+      "eval_steps_per_second": 1.563,
+      "step": 27500
+    },
+    {
+      "epoch": 424.61538461538464,
+      "grad_norm": 56.913578033447266,
+      "learning_rate": 7.2417000000000005e-06,
+      "loss": 0.7843,
+      "step": 27600
+    },
+    {
+      "epoch": 424.61538461538464,
+      "eval_loss": 1.0650087594985962,
+      "eval_runtime": 10.8169,
+      "eval_samples_per_second": 12.111,
+      "eval_steps_per_second": 1.572,
+      "step": 27600
+    },
+    {
+      "epoch": 426.15384615384613,
+      "grad_norm": 28.091888427734375,
+      "learning_rate": 7.231800000000001e-06,
+      "loss": 0.8113,
+      "step": 27700
+    },
+    {
+      "epoch": 426.15384615384613,
+      "eval_loss": 1.103638768196106,
+      "eval_runtime": 10.8897,
+      "eval_samples_per_second": 12.03,
+      "eval_steps_per_second": 1.561,
+      "step": 27700
+    },
+    {
+      "epoch": 427.6923076923077,
+      "grad_norm": 21.665067672729492,
+      "learning_rate": 7.221800000000001e-06,
+      "loss": 0.7803,
+      "step": 27800
+    },
+    {
+      "epoch": 427.6923076923077,
+      "eval_loss": 1.1119697093963623,
+      "eval_runtime": 10.8141,
+      "eval_samples_per_second": 12.114,
+      "eval_steps_per_second": 1.572,
+      "step": 27800
+    },
+    {
+      "epoch": 429.2307692307692,
+      "grad_norm": 65.23661804199219,
+      "learning_rate": 7.211800000000001e-06,
+      "loss": 0.7963,
+      "step": 27900
+    },
+    {
+      "epoch": 429.2307692307692,
+      "eval_loss": 1.1217514276504517,
+      "eval_runtime": 10.8793,
+      "eval_samples_per_second": 12.041,
+      "eval_steps_per_second": 1.563,
+      "step": 27900
+    },
+    {
+      "epoch": 430.7692307692308,
+      "grad_norm": 43.18936538696289,
+      "learning_rate": 7.201800000000001e-06,
+      "loss": 0.7976,
+      "step": 28000
+    },
+    {
+      "epoch": 430.7692307692308,
+      "eval_loss": 1.1384552717208862,
+      "eval_runtime": 10.8325,
+      "eval_samples_per_second": 12.093,
+      "eval_steps_per_second": 1.569,
+      "step": 28000
+    },
+    {
+      "epoch": 432.3076923076923,
+      "grad_norm": 33.29544448852539,
+      "learning_rate": 7.191800000000001e-06,
+      "loss": 0.803,
+      "step": 28100
+    },
+    {
+      "epoch": 432.3076923076923,
+      "eval_loss": 1.1162676811218262,
+      "eval_runtime": 10.8215,
+      "eval_samples_per_second": 12.106,
+      "eval_steps_per_second": 1.571,
+      "step": 28100
+    },
+    {
+      "epoch": 433.84615384615387,
+      "grad_norm": 22.86684799194336,
+      "learning_rate": 7.181800000000001e-06,
+      "loss": 0.8037,
+      "step": 28200
+    },
+    {
+      "epoch": 433.84615384615387,
+      "eval_loss": 1.1036708354949951,
+      "eval_runtime": 10.8309,
+      "eval_samples_per_second": 12.095,
+      "eval_steps_per_second": 1.57,
+      "step": 28200
+    },
+    {
+      "epoch": 435.38461538461536,
+      "grad_norm": 21.229990005493164,
+      "learning_rate": 7.171800000000001e-06,
+      "loss": 0.7713,
+      "step": 28300
+    },
+    {
+      "epoch": 435.38461538461536,
+      "eval_loss": 1.0910826921463013,
+      "eval_runtime": 10.8187,
+      "eval_samples_per_second": 12.109,
+      "eval_steps_per_second": 1.571,
+      "step": 28300
+    },
+    {
+      "epoch": 436.9230769230769,
+      "grad_norm": 33.30841064453125,
+      "learning_rate": 7.161800000000001e-06,
+      "loss": 0.7956,
+      "step": 28400
+    },
+    {
+      "epoch": 436.9230769230769,
+      "eval_loss": 1.0862958431243896,
+      "eval_runtime": 10.8992,
+      "eval_samples_per_second": 12.019,
+      "eval_steps_per_second": 1.56,
+      "step": 28400
+    },
+    {
+      "epoch": 438.46153846153845,
+      "grad_norm": 24.01555633544922,
+      "learning_rate": 7.151800000000001e-06,
+      "loss": 0.7883,
+      "step": 28500
+    },
+    {
+      "epoch": 438.46153846153845,
+      "eval_loss": 1.1144325733184814,
+      "eval_runtime": 11.0935,
+      "eval_samples_per_second": 11.809,
+      "eval_steps_per_second": 1.532,
+      "step": 28500
+    },
+    {
+      "epoch": 440.0,
+      "grad_norm": 43.260921478271484,
+      "learning_rate": 7.141800000000001e-06,
+      "loss": 0.7885,
+      "step": 28600
+    },
+    {
+      "epoch": 440.0,
+      "eval_loss": 1.0987013578414917,
+      "eval_runtime": 10.803,
+      "eval_samples_per_second": 12.126,
+      "eval_steps_per_second": 1.574,
+      "step": 28600
+    },
+    {
+      "epoch": 441.53846153846155,
+      "grad_norm": 24.575450897216797,
+      "learning_rate": 7.131800000000001e-06,
+      "loss": 0.8052,
+      "step": 28700
+    },
+    {
+      "epoch": 441.53846153846155,
+      "eval_loss": 1.1236560344696045,
+      "eval_runtime": 10.7671,
+      "eval_samples_per_second": 12.167,
+      "eval_steps_per_second": 1.579,
+      "step": 28700
+    },
+    {
+      "epoch": 443.0769230769231,
+      "grad_norm": 58.26254653930664,
+      "learning_rate": 7.121800000000001e-06,
+      "loss": 0.7856,
+      "step": 28800
+    },
+    {
+      "epoch": 443.0769230769231,
+      "eval_loss": 1.1264104843139648,
+      "eval_runtime": 11.0135,
+      "eval_samples_per_second": 11.894,
+      "eval_steps_per_second": 1.544,
+      "step": 28800
+    },
+    {
+      "epoch": 444.61538461538464,
+      "grad_norm": 41.40846633911133,
+      "learning_rate": 7.111800000000001e-06,
+      "loss": 0.7923,
+      "step": 28900
+    },
+    {
+      "epoch": 444.61538461538464,
+      "eval_loss": 1.1042475700378418,
+      "eval_runtime": 10.8929,
+      "eval_samples_per_second": 12.026,
+      "eval_steps_per_second": 1.561,
+      "step": 28900
+    },
+    {
+      "epoch": 446.15384615384613,
+      "grad_norm": 42.24570846557617,
+      "learning_rate": 7.101800000000001e-06,
+      "loss": 0.8239,
+      "step": 29000
+    },
+    {
+      "epoch": 446.15384615384613,
+      "eval_loss": 1.0964312553405762,
+      "eval_runtime": 11.2262,
+      "eval_samples_per_second": 11.669,
+      "eval_steps_per_second": 1.514,
+      "step": 29000
+    }
+  ],
+  "logging_steps": 100,
+  "max_steps": 100000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1539,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.1102635005677568e+20,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}