diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5025 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 15.318627450980392,
+  "global_step": 250000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 0.8848,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 0.7411,
+      "step": 1000
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 0.9070383906364441,
+      "eval_runtime": 1.3223,
+      "eval_samples_per_second": 756.258,
+      "eval_steps_per_second": 12.1,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 0.7398,
+      "step": 1500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 0.7395,
+      "step": 2000
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 0.9064154028892517,
+      "eval_runtime": 1.194,
+      "eval_samples_per_second": 837.495,
+      "eval_steps_per_second": 13.4,
+      "step": 2000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 0.7392,
+      "step": 2500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 0.7387,
+      "step": 3000
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 0.9047173857688904,
+      "eval_runtime": 1.3077,
+      "eval_samples_per_second": 764.675,
+      "eval_steps_per_second": 12.235,
+      "step": 3000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000168,
+      "loss": 0.7384,
+      "step": 3500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019199999999999998,
+      "loss": 0.7382,
+      "step": 4000
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 0.9014638662338257,
+      "eval_runtime": 1.2645,
+      "eval_samples_per_second": 790.849,
+      "eval_steps_per_second": 12.654,
+      "step": 4000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 0.7382,
+      "step": 4500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 0.7381,
+      "step": 5000
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 0.9044438600540161,
+      "eval_runtime": 1.2394,
+      "eval_samples_per_second": 806.828,
+      "eval_steps_per_second": 12.909,
+      "step": 5000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 0.7379,
+      "step": 5500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 0.7379,
+      "step": 6000
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 0.9041813015937805,
+      "eval_runtime": 1.2368,
+      "eval_samples_per_second": 808.566,
+      "eval_steps_per_second": 12.937,
+      "step": 6000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000312,
+      "loss": 0.7381,
+      "step": 6500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000336,
+      "loss": 0.7379,
+      "step": 7000
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 0.9053561687469482,
+      "eval_runtime": 1.2638,
+      "eval_samples_per_second": 791.247,
+      "eval_steps_per_second": 12.66,
+      "step": 7000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 0.7378,
+      "step": 7500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00038399999999999996,
+      "loss": 0.7378,
+      "step": 8000
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 0.9035025835037231,
+      "eval_runtime": 1.2079,
+      "eval_samples_per_second": 827.911,
+      "eval_steps_per_second": 13.247,
+      "step": 8000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000408,
+      "loss": 0.7378,
+      "step": 8500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043199999999999993,
+      "loss": 0.7378,
+      "step": 9000
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 0.9025757312774658,
+      "eval_runtime": 1.2745,
+      "eval_samples_per_second": 784.598,
+      "eval_steps_per_second": 12.554,
+      "step": 9000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00045599999999999997,
+      "loss": 0.7375,
+      "step": 9500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00047999999999999996,
+      "loss": 0.7371,
+      "step": 10000
+    },
+    {
+      "epoch": 0.61,
+      "eval_loss": 0.903812050819397,
+      "eval_runtime": 1.2376,
+      "eval_samples_per_second": 807.983,
+      "eval_steps_per_second": 12.928,
+      "step": 10000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005039999999999999,
+      "loss": 0.7369,
+      "step": 10500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005279999999999999,
+      "loss": 0.7369,
+      "step": 11000
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 0.9027426838874817,
+      "eval_runtime": 1.2459,
+      "eval_samples_per_second": 802.657,
+      "eval_steps_per_second": 12.843,
+      "step": 11000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000552,
+      "loss": 0.7369,
+      "step": 11500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005759999999999999,
+      "loss": 0.7368,
+      "step": 12000
+    },
+    {
+      "epoch": 0.74,
+      "eval_loss": 0.9022310972213745,
+      "eval_runtime": 1.2295,
+      "eval_samples_per_second": 813.368,
+      "eval_steps_per_second": 13.014,
+      "step": 12000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0006,
+      "loss": 0.7367,
+      "step": 12500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0005999935478721662,
+      "loss": 0.7368,
+      "step": 13000
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 0.8987133502960205,
+      "eval_runtime": 1.2643,
+      "eval_samples_per_second": 790.939,
+      "eval_steps_per_second": 12.655,
+      "step": 13000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000599974191770902,
+      "loss": 0.7375,
+      "step": 13500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0005999419325429058,
+      "loss": 0.7374,
+      "step": 14000
+    },
+    {
+      "epoch": 0.86,
+      "eval_loss": 0.9013972282409668,
+      "eval_runtime": 1.236,
+      "eval_samples_per_second": 809.042,
+      "eval_steps_per_second": 12.945,
+      "step": 14000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005998967715993009,
+      "loss": 0.7369,
+      "step": 14500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005998387109155732,
+      "loss": 0.7369,
+      "step": 15000
+    },
+    {
+      "epoch": 0.92,
+      "eval_loss": 0.9001522660255432,
+      "eval_runtime": 1.235,
+      "eval_samples_per_second": 809.697,
+      "eval_steps_per_second": 12.955,
+      "step": 15000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000599767753031485,
+      "loss": 0.737,
+      "step": 15500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005996839010509641,
+      "loss": 0.7369,
+      "step": 16000
+    },
+    {
+      "epoch": 0.98,
+      "eval_loss": 0.9001864790916443,
+      "eval_runtime": 1.1685,
+      "eval_samples_per_second": 855.784,
+      "eval_steps_per_second": 13.693,
+      "step": 16000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005995871586419678,
+      "loss": 0.7369,
+      "step": 16500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0005994775300363225,
+      "loss": 0.7372,
+      "step": 17000
+    },
+    {
+      "epoch": 1.04,
+      "eval_loss": 0.9018534421920776,
+      "eval_runtime": 1.1865,
+      "eval_samples_per_second": 842.819,
+      "eval_steps_per_second": 13.485,
+      "step": 17000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0005993550200295384,
+      "loss": 0.7371,
+      "step": 17500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0005992196339806002,
+      "loss": 0.737,
+      "step": 18000
+    },
+    {
+      "epoch": 1.1,
+      "eval_loss": 0.9001176953315735,
+      "eval_runtime": 1.2219,
+      "eval_samples_per_second": 818.375,
+      "eval_steps_per_second": 13.094,
+      "step": 18000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0005990713778117324,
+      "loss": 0.7369,
+      "step": 18500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0005989102580081398,
+      "loss": 0.737,
+      "step": 19000
+    },
+    {
+      "epoch": 1.16,
+      "eval_loss": 0.9006143808364868,
+      "eval_runtime": 1.2544,
+      "eval_samples_per_second": 797.163,
+      "eval_steps_per_second": 12.755,
+      "step": 19000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0005987362816177249,
+      "loss": 0.7369,
+      "step": 19500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0005985494562507783,
+      "loss": 0.7369,
+      "step": 20000
+    },
+    {
+      "epoch": 1.23,
+      "eval_loss": 0.9006676077842712,
+      "eval_runtime": 1.2815,
+      "eval_samples_per_second": 780.312,
+      "eval_steps_per_second": 12.485,
+      "step": 20000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000598349790079647,
+      "loss": 0.737,
+      "step": 20500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000598137291838376,
+      "loss": 0.7365,
+      "step": 21000
+    },
+    {
+      "epoch": 1.29,
+      "eval_loss": 0.8698467016220093,
+      "eval_runtime": 1.2584,
+      "eval_samples_per_second": 794.654,
+      "eval_steps_per_second": 12.714,
+      "step": 21000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000597911970822327,
+      "loss": 0.7364,
+      "step": 21500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000597673836887771,
+      "loss": 0.7363,
+      "step": 22000
+    },
+    {
+      "epoch": 1.35,
+      "eval_loss": 0.870046079158783,
+      "eval_runtime": 1.2567,
+      "eval_samples_per_second": 795.734,
+      "eval_steps_per_second": 12.732,
+      "step": 22000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005974229004514577,
+      "loss": 0.7363,
+      "step": 22500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005971591724901598,
+      "loss": 0.7366,
+      "step": 23000
+    },
+    {
+      "epoch": 1.41,
+      "eval_loss": 0.9021453857421875,
+      "eval_runtime": 1.277,
+      "eval_samples_per_second": 783.107,
+      "eval_steps_per_second": 12.53,
+      "step": 23000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005968826645401927,
+      "loss": 0.7368,
+      "step": 23500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00059659338869691,
+      "loss": 0.7362,
+      "step": 24000
+    },
+    {
+      "epoch": 1.47,
+      "eval_loss": 0.8762597441673279,
+      "eval_runtime": 1.2918,
+      "eval_samples_per_second": 774.091,
+      "eval_steps_per_second": 12.385,
+      "step": 24000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005962913576141742,
+      "loss": 0.7354,
+      "step": 24500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.000595976584503803,
+      "loss": 0.7082,
+      "step": 25000
+    },
+    {
+      "epoch": 1.53,
+      "eval_loss": 0.8719052672386169,
+      "eval_runtime": 1.2648,
+      "eval_samples_per_second": 790.61,
+      "eval_steps_per_second": 12.65,
+      "step": 25000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005956490831349923,
+      "loss": 0.6914,
+      "step": 25500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005953088678337129,
+      "loss": 0.6774,
+      "step": 26000
+    },
+    {
+      "epoch": 1.59,
+      "eval_loss": 0.8876385688781738,
+      "eval_runtime": 1.3046,
+      "eval_samples_per_second": 766.511,
+      "eval_steps_per_second": 12.264,
+      "step": 26000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005949559534820841,
+      "loss": 0.6675,
+      "step": 26500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0005945903555177229,
+      "loss": 0.6525,
+      "step": 27000
+    },
+    {
+      "epoch": 1.65,
+      "eval_loss": 0.8904880285263062,
+      "eval_runtime": 1.319,
+      "eval_samples_per_second": 758.176,
+      "eval_steps_per_second": 12.131,
+      "step": 27000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0005942120899330687,
+      "loss": 0.6186,
+      "step": 27500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0005938211732746836,
+      "loss": 0.6022,
+      "step": 28000
+    },
+    {
+      "epoch": 1.72,
+      "eval_loss": 0.8856265544891357,
+      "eval_runtime": 1.2877,
+      "eval_samples_per_second": 776.587,
+      "eval_steps_per_second": 12.425,
+      "step": 28000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0005934176226425286,
+      "loss": 0.593,
+      "step": 28500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0005930014556892158,
+      "loss": 0.5874,
+      "step": 29000
+    },
+    {
+      "epoch": 1.78,
+      "eval_loss": 0.8793675303459167,
+      "eval_runtime": 1.2489,
+      "eval_samples_per_second": 800.707,
+      "eval_steps_per_second": 12.811,
+      "step": 29000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0005925726906192357,
+      "loss": 0.5837,
+      "step": 29500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0005921313461881617,
+      "loss": 0.5765,
+      "step": 30000
+    },
+    {
+      "epoch": 1.84,
+      "eval_loss": 0.8805580735206604,
+      "eval_runtime": 1.2758,
+      "eval_samples_per_second": 783.816,
+      "eval_steps_per_second": 12.541,
+      "step": 30000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0005916774417018287,
+      "loss": 0.5714,
+      "step": 30500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0005912109970154897,
+      "loss": 0.5685,
+      "step": 31000
+    },
+    {
+      "epoch": 1.9,
+      "eval_loss": 0.8747313022613525,
+      "eval_runtime": 1.3003,
+      "eval_samples_per_second": 769.027,
+      "eval_steps_per_second": 12.304,
+      "step": 31000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0005907320325329461,
+      "loss": 0.566,
+      "step": 31500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0005902405692056561,
+      "loss": 0.564,
+      "step": 32000
+    },
+    {
+      "epoch": 1.96,
+      "eval_loss": 0.8779122233390808,
+      "eval_runtime": 1.293,
+      "eval_samples_per_second": 773.405,
+      "eval_steps_per_second": 12.374,
+      "step": 32000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0005897366285318178,
+      "loss": 0.5617,
+      "step": 32500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0005892202325554288,
+      "loss": 0.5606,
+      "step": 33000
+    },
+    {
+      "epoch": 2.02,
+      "eval_loss": 0.8761873245239258,
+      "eval_runtime": 1.3816,
+      "eval_samples_per_second": 723.824,
+      "eval_steps_per_second": 11.581,
+      "step": 33000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005886914038653217,
+      "loss": 0.5583,
+      "step": 33500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0005881501655941771,
+      "loss": 0.5574,
+      "step": 34000
+    },
+    {
+      "epoch": 2.08,
+      "eval_loss": 0.8702684044837952,
+      "eval_runtime": 1.3086,
+      "eval_samples_per_second": 764.158,
+      "eval_steps_per_second": 12.227,
+      "step": 34000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00058759654141751,
+      "loss": 0.5548,
+      "step": 34500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005870305555526355,
+      "loss": 0.5528,
+      "step": 35000
+    },
+    {
+      "epoch": 2.14,
+      "eval_loss": 0.8663867115974426,
+      "eval_runtime": 1.2619,
+      "eval_samples_per_second": 792.469,
+      "eval_steps_per_second": 12.679,
+      "step": 35000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0005864522327576088,
+      "loss": 0.5509,
+      "step": 35500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005858615983301424,
+      "loss": 0.5494,
+      "step": 36000
+    },
+    {
+      "epoch": 2.21,
+      "eval_loss": 0.8717171549797058,
+      "eval_runtime": 1.2691,
+      "eval_samples_per_second": 787.953,
+      "eval_steps_per_second": 12.607,
+      "step": 36000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0005852586781064997,
+      "loss": 0.5471,
+      "step": 36500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0005846434984603645,
+      "loss": 0.5448,
+      "step": 37000
+    },
+    {
+      "epoch": 2.27,
+      "eval_loss": 0.8672583103179932,
+      "eval_runtime": 1.268,
+      "eval_samples_per_second": 788.629,
+      "eval_steps_per_second": 12.618,
+      "step": 37000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005840160863016872,
+      "loss": 0.5433,
+      "step": 37500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0005833764690755083,
+      "loss": 0.5419,
+      "step": 38000
+    },
+    {
+      "epoch": 2.33,
+      "eval_loss": 0.8636866211891174,
+      "eval_runtime": 1.3526,
+      "eval_samples_per_second": 739.295,
+      "eval_steps_per_second": 11.829,
+      "step": 38000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005827246747607574,
+      "loss": 0.5398,
+      "step": 38500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0005820607318690293,
+      "loss": 0.5385,
+      "step": 39000
+    },
+    {
+      "epoch": 2.39,
+      "eval_loss": 0.8634054064750671,
+      "eval_runtime": 1.2705,
+      "eval_samples_per_second": 787.122,
+      "eval_steps_per_second": 12.594,
+      "step": 39000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0005813846694433368,
+      "loss": 0.5374,
+      "step": 39500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005806965170568409,
+      "loss": 0.536,
+      "step": 40000
+    },
+    {
+      "epoch": 2.45,
+      "eval_loss": 0.8661102652549744,
+      "eval_runtime": 1.3401,
+      "eval_samples_per_second": 746.204,
+      "eval_steps_per_second": 11.939,
+      "step": 40000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005799963048115559,
+      "loss": 0.5353,
+      "step": 40500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0005792840633370341,
+      "loss": 0.5336,
+      "step": 41000
+    },
+    {
+      "epoch": 2.51,
+      "eval_loss": 0.8631040453910828,
+      "eval_runtime": 1.3088,
+      "eval_samples_per_second": 764.037,
+      "eval_steps_per_second": 12.225,
+      "step": 41000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0005785598237890247,
+      "loss": 0.5327,
+      "step": 41500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005778236178481119,
+      "loss": 0.5316,
+      "step": 42000
+    },
+    {
+      "epoch": 2.57,
+      "eval_loss": 0.8605585098266602,
+      "eval_runtime": 1.2836,
+      "eval_samples_per_second": 779.077,
+      "eval_steps_per_second": 12.465,
+      "step": 42000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0005770754777183285,
+      "loss": 0.5306,
+      "step": 42500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0005763154361257473,
+      "loss": 0.5297,
+      "step": 43000
+    },
+    {
+      "epoch": 2.63,
+      "eval_loss": 0.8589205145835876,
+      "eval_runtime": 1.2763,
+      "eval_samples_per_second": 783.485,
+      "eval_steps_per_second": 12.536,
+      "step": 43000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0005755435263170498,
+      "loss": 0.5287,
+      "step": 43500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005747597820580717,
+      "loss": 0.5305,
+      "step": 44000
+    },
+    {
+      "epoch": 2.7,
+      "eval_loss": 0.8569635152816772,
+      "eval_runtime": 1.2713,
+      "eval_samples_per_second": 786.624,
+      "eval_steps_per_second": 12.586,
+      "step": 44000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.000573964237632326,
+      "loss": 0.527,
+      "step": 44500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0005731569278395029,
+      "loss": 0.5262,
+      "step": 45000
+    },
+    {
+      "epoch": 2.76,
+      "eval_loss": 0.8558768033981323,
+      "eval_runtime": 1.2738,
+      "eval_samples_per_second": 785.051,
+      "eval_steps_per_second": 12.561,
+      "step": 45000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005723378879939481,
+      "loss": 0.5254,
+      "step": 45500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0005715071539231178,
+      "loss": 0.5247,
+      "step": 46000
+    },
+    {
+      "epoch": 2.82,
+      "eval_loss": 0.8633874654769897,
+      "eval_runtime": 1.2747,
+      "eval_samples_per_second": 784.504,
+      "eval_steps_per_second": 12.552,
+      "step": 46000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0005706647619660116,
+      "loss": 0.5243,
+      "step": 46500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005698107489715823,
+      "loss": 0.5235,
+      "step": 47000
+    },
+    {
+      "epoch": 2.88,
+      "eval_loss": 0.8606237769126892,
+      "eval_runtime": 1.3283,
+      "eval_samples_per_second": 752.838,
+      "eval_steps_per_second": 12.045,
+      "step": 47000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0005689451522971252,
+      "loss": 0.5228,
+      "step": 47500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005680680098066429,
+      "loss": 0.5227,
+      "step": 48000
+    },
+    {
+      "epoch": 2.94,
+      "eval_loss": 0.8610469698905945,
+      "eval_runtime": 1.2783,
+      "eval_samples_per_second": 782.274,
+      "eval_steps_per_second": 12.516,
+      "step": 48000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005671793598691895,
+      "loss": 0.5215,
+      "step": 48500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0005662792413571921,
+      "loss": 0.5206,
+      "step": 49000
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 0.8610268235206604,
+      "eval_runtime": 1.3555,
+      "eval_samples_per_second": 737.72,
+      "eval_steps_per_second": 11.804,
+      "step": 49000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0005653676936447504,
+      "loss": 0.5201,
+      "step": 49500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0005644447566059142,
+      "loss": 0.5194,
+      "step": 50000
+    },
+    {
+      "epoch": 3.06,
+      "eval_loss": 0.8611247539520264,
+      "eval_runtime": 1.2902,
+      "eval_samples_per_second": 775.099,
+      "eval_steps_per_second": 12.402,
+      "step": 50000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0005635104706129397,
+      "loss": 0.5189,
+      "step": 50500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0005625648765345228,
+      "loss": 0.5183,
+      "step": 51000
+    },
+    {
+      "epoch": 3.12,
+      "eval_loss": 0.8579334616661072,
+      "eval_runtime": 1.3546,
+      "eval_samples_per_second": 738.246,
+      "eval_steps_per_second": 11.812,
+      "step": 51000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0005616080157340118,
+      "loss": 0.5178,
+      "step": 51500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0005606399300675978,
+      "loss": 0.5175,
+      "step": 52000
+    },
+    {
+      "epoch": 3.19,
+      "eval_loss": 0.8597957491874695,
+      "eval_runtime": 1.3415,
+      "eval_samples_per_second": 745.449,
+      "eval_steps_per_second": 11.927,
+      "step": 52000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005596606618824843,
+      "loss": 0.5186,
+      "step": 52500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005586702540150338,
+      "loss": 0.5163,
+      "step": 53000
+    },
+    {
+      "epoch": 3.25,
+      "eval_loss": 0.8520950078964233,
+      "eval_runtime": 1.3505,
+      "eval_samples_per_second": 740.484,
+      "eval_steps_per_second": 11.848,
+      "step": 53000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.000557668749788895,
+      "loss": 0.5159,
+      "step": 53500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005566561930131072,
+      "loss": 0.5156,
+      "step": 54000
+    },
+    {
+      "epoch": 3.31,
+      "eval_loss": 0.8550420999526978,
+      "eval_runtime": 1.3983,
+      "eval_samples_per_second": 715.143,
+      "eval_steps_per_second": 11.442,
+      "step": 54000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.000555632627980184,
+      "loss": 0.5153,
+      "step": 54500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005545980994641758,
+      "loss": 0.5148,
+      "step": 55000
+    },
+    {
+      "epoch": 3.37,
+      "eval_loss": 0.8503537178039551,
+      "eval_runtime": 1.2794,
+      "eval_samples_per_second": 781.636,
+      "eval_steps_per_second": 12.506,
+      "step": 55000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0005535526527187115,
+      "loss": 0.5144,
+      "step": 55500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0005524963334750183,
+      "loss": 0.5139,
+      "step": 56000
+    },
+    {
+      "epoch": 3.43,
+      "eval_loss": 0.8530069589614868,
+      "eval_runtime": 1.2989,
+      "eval_samples_per_second": 769.886,
+      "eval_steps_per_second": 12.318,
+      "step": 56000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0005514291879399219,
+      "loss": 0.5136,
+      "step": 56500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.000550351262793825,
+      "loss": 0.5133,
+      "step": 57000
+    },
+    {
+      "epoch": 3.49,
+      "eval_loss": 0.8588612675666809,
+      "eval_runtime": 1.2931,
+      "eval_samples_per_second": 773.351,
+      "eval_steps_per_second": 12.374,
+      "step": 57000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0005492626051886659,
+      "loss": 0.5129,
+      "step": 57500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0005481632627458546,
+      "loss": 0.5126,
+      "step": 58000
+    },
+    {
+      "epoch": 3.55,
+      "eval_loss": 0.8561201095581055,
+      "eval_runtime": 1.244,
+      "eval_samples_per_second": 803.841,
+      "eval_steps_per_second": 12.861,
+      "step": 58000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0005470532835541911,
+      "loss": 0.5122,
+      "step": 58500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.000545932716167761,
+      "loss": 0.5119,
+      "step": 59000
+    },
+    {
+      "epoch": 3.62,
+      "eval_loss": 0.8574157953262329,
+      "eval_runtime": 1.3446,
+      "eval_samples_per_second": 743.721,
+      "eval_steps_per_second": 11.9,
+      "step": 59000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.000544801609603812,
+      "loss": 0.5115,
+      "step": 59500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0005436600133406095,
+      "loss": 0.5127,
+      "step": 60000
+    },
+    {
+      "epoch": 3.68,
+      "eval_loss": 0.8624024987220764,
+      "eval_runtime": 1.324,
+      "eval_samples_per_second": 755.286,
+      "eval_steps_per_second": 12.085,
+      "step": 60000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0005425079773152721,
+      "loss": 0.5113,
+      "step": 60500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0005413455519215879,
+      "loss": 0.5105,
+      "step": 61000
+    },
+    {
+      "epoch": 3.74,
+      "eval_loss": 0.8522316813468933,
+      "eval_runtime": 1.2934,
+      "eval_samples_per_second": 773.185,
+      "eval_steps_per_second": 12.371,
+      "step": 61000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0005401727880078093,
+      "loss": 0.5102,
+      "step": 61500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0005389897368744289,
+      "loss": 0.5099,
+      "step": 62000
+    },
+    {
+      "epoch": 3.8,
+      "eval_loss": 0.8549780249595642,
+      "eval_runtime": 1.3356,
+      "eval_samples_per_second": 748.732,
+      "eval_steps_per_second": 11.98,
+      "step": 62000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0005377964502719361,
+      "loss": 0.5099,
+      "step": 62500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0005365929803985524,
+      "loss": 0.5094,
+      "step": 63000
+    },
+    {
+      "epoch": 3.86,
+      "eval_loss": 0.8536927700042725,
+      "eval_runtime": 1.2799,
+      "eval_samples_per_second": 781.291,
+      "eval_steps_per_second": 12.501,
+      "step": 63000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0005353793798979489,
+      "loss": 0.5093,
+      "step": 63500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.000534155701856943,
+      "loss": 0.509,
+      "step": 64000
+    },
+    {
+      "epoch": 3.92,
+      "eval_loss": 0.8535122871398926,
+      "eval_runtime": 1.2782,
+      "eval_samples_per_second": 782.329,
+      "eval_steps_per_second": 12.517,
+      "step": 64000
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0005329219998031763,
+      "loss": 0.5087,
+      "step": 64500
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0005316783277027734,
+      "loss": 0.5091,
+      "step": 65000
+    },
+    {
+      "epoch": 3.98,
+      "eval_loss": 0.8591586351394653,
+      "eval_runtime": 1.3143,
+      "eval_samples_per_second": 760.882,
+      "eval_steps_per_second": 12.174,
+      "step": 65000
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0005304247399579808,
+      "loss": 0.508,
+      "step": 65500
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0005291612914047876,
+      "loss": 0.5079,
+      "step": 66000
+    },
+    {
+      "epoch": 4.04,
+      "eval_loss": 0.8554427027702332,
+      "eval_runtime": 1.3483,
+      "eval_samples_per_second": 741.678,
+      "eval_steps_per_second": 11.867,
+      "step": 66000
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0005278880373105263,
+      "loss": 0.5077,
+      "step": 66500
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0005266050333714561,
+      "loss": 0.5074,
+      "step": 67000
+    },
+    {
+      "epoch": 4.11,
+      "eval_loss": 0.8515585660934448,
+      "eval_runtime": 1.2858,
+      "eval_samples_per_second": 777.712,
+      "eval_steps_per_second": 12.443,
+      "step": 67000
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0005253123357103253,
+      "loss": 0.5074,
+      "step": 67500
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0005240100008739177,
+      "loss": 0.5069,
+      "step": 68000
+    },
+    {
+      "epoch": 4.17,
+      "eval_loss": 0.8490995168685913,
+      "eval_runtime": 1.2974,
+      "eval_samples_per_second": 770.751,
+      "eval_steps_per_second": 12.332,
+      "step": 68000
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0005226980858305778,
+      "loss": 0.5067,
+      "step": 68500
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0005213766479677197,
+      "loss": 0.5066,
+      "step": 69000
+    },
+    {
+      "epoch": 4.23,
+      "eval_loss": 0.8570588231086731,
+      "eval_runtime": 1.2957,
+      "eval_samples_per_second": 771.773,
+      "eval_steps_per_second": 12.348,
+      "step": 69000
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0005200457450893163,
+      "loss": 0.5063,
+      "step": 69500
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0005187054354133712,
+      "loss": 0.5068,
+      "step": 70000
+    },
+    {
+      "epoch": 4.29,
+      "eval_loss": 0.8535866141319275,
+      "eval_runtime": 1.3217,
+      "eval_samples_per_second": 756.603,
+      "eval_steps_per_second": 12.106,
+      "step": 70000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0005173557775693715,
+      "loss": 0.5058,
+      "step": 70500
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0005159968305957235,
+      "loss": 0.5066,
+      "step": 71000
+    },
+    {
+      "epoch": 4.35,
+      "eval_loss": 0.9288005232810974,
+      "eval_runtime": 1.2769,
+      "eval_samples_per_second": 783.148,
+      "eval_steps_per_second": 12.53,
+      "step": 71000
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0005146286539371703,
+      "loss": 0.5084,
+      "step": 71500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0005132513074421913,
+      "loss": 0.5051,
+      "step": 72000
+    },
+    {
+      "epoch": 4.41,
+      "eval_loss": 0.8597245812416077,
+      "eval_runtime": 1.3157,
+      "eval_samples_per_second": 760.048,
+      "eval_steps_per_second": 12.161,
+      "step": 72000
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0005118648513603841,
+      "loss": 0.5051,
+      "step": 72500
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0005104693463398293,
+      "loss": 0.5045,
+      "step": 73000
+    },
+    {
+      "epoch": 4.47,
+      "eval_loss": 0.8554754257202148,
+      "eval_runtime": 1.3269,
+      "eval_samples_per_second": 753.64,
+      "eval_steps_per_second": 12.058,
+      "step": 73000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0005090648534244371,
+      "loss": 0.5045,
+      "step": 73500
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0005076514340512776,
+      "loss": 0.5043,
+      "step": 74000
+    },
+    {
+      "epoch": 4.53,
+      "eval_loss": 0.8547331094741821,
+      "eval_runtime": 1.2736,
+      "eval_samples_per_second": 785.148,
+      "eval_steps_per_second": 12.562,
+      "step": 74000
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0005062291500478931,
+      "loss": 0.5039,
+      "step": 74500
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0005047980636295937,
+      "loss": 0.5039,
+      "step": 75000
+    },
+    {
+      "epoch": 4.6,
+      "eval_loss": 0.8560617566108704,
+      "eval_runtime": 1.2429,
+      "eval_samples_per_second": 804.574,
+      "eval_steps_per_second": 12.873,
+      "step": 75000
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0005033582373967348,
+      "loss": 0.5036,
+      "step": 75500
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0005019097343319809,
+      "loss": 0.504,
+      "step": 76000
+    },
+    {
+      "epoch": 4.66,
+      "eval_loss": 0.854142427444458,
+      "eval_runtime": 1.3399,
+      "eval_samples_per_second": 746.336,
+      "eval_steps_per_second": 11.941,
+      "step": 76000
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0005004526177975481,
+      "loss": 0.503,
+      "step": 76500
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0004989869515324342,
+      "loss": 0.5026,
+      "step": 77000
+    },
+    {
+      "epoch": 4.72,
+      "eval_loss": 0.8489631414413452,
+      "eval_runtime": 1.2942,
+      "eval_samples_per_second": 772.659,
+      "eval_steps_per_second": 12.363,
+      "step": 77000
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0004975127996496297,
+      "loss": 0.5024,
+      "step": 77500
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0004960302266333135,
+      "loss": 0.5024,
+      "step": 78000
+    },
+    {
+      "epoch": 4.78,
+      "eval_loss": 0.8499003648757935,
+      "eval_runtime": 1.3145,
+      "eval_samples_per_second": 760.727,
+      "eval_steps_per_second": 12.172,
+      "step": 78000
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0004945392973360323,
+      "loss": 0.502,
+      "step": 78500
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0004930400769758634,
+      "loss": 0.5019,
+      "step": 79000
+    },
+    {
+      "epoch": 4.84,
+      "eval_loss": 0.8521845936775208,
+      "eval_runtime": 1.3201,
+      "eval_samples_per_second": 757.527,
+      "eval_steps_per_second": 12.12,
+      "step": 79000
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0004915326311335622,
+      "loss": 0.5017,
+      "step": 79500
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0004900170257496933,
+      "loss": 0.5014,
+      "step": 80000
+    },
+    {
+      "epoch": 4.9,
+      "eval_loss": 0.8508275747299194,
+      "eval_runtime": 1.3001,
+      "eval_samples_per_second": 769.194,
+      "eval_steps_per_second": 12.307,
+      "step": 80000
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0004884933271217461,
+      "loss": 0.501,
+      "step": 80500
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0004869616019012347,
+      "loss": 0.5008,
+      "step": 81000
+    },
+    {
+      "epoch": 4.96,
+      "eval_loss": 0.8512425422668457,
+      "eval_runtime": 1.2521,
+      "eval_samples_per_second": 798.677,
+      "eval_steps_per_second": 12.779,
+      "step": 81000
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00048542191709078226,
+      "loss": 0.5004,
+      "step": 81500
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00048387434004119027,
+      "loss": 0.5002,
+      "step": 82000
+    },
+    {
+      "epoch": 5.02,
+      "eval_loss": 0.8470181822776794,
+      "eval_runtime": 1.2894,
+      "eval_samples_per_second": 775.545,
+      "eval_steps_per_second": 12.409,
+      "step": 82000
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0004823189384484924,
+      "loss": 0.5007,
+      "step": 82500
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.000480755780350993,
+      "loss": 0.4995,
+      "step": 83000
+    },
+    {
+      "epoch": 5.09,
+      "eval_loss": 0.846192479133606,
+      "eval_runtime": 1.2808,
+      "eval_samples_per_second": 780.79,
+      "eval_steps_per_second": 12.493,
+      "step": 83000
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0004791849341262914,
+      "loss": 0.4993,
+      "step": 83500
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0004776064684882901,
+      "loss": 0.4991,
+      "step": 84000
+    },
+    {
+      "epoch": 5.15,
+      "eval_loss": 0.8454980254173279,
+      "eval_runtime": 1.3356,
+      "eval_samples_per_second": 748.729,
+      "eval_steps_per_second": 11.98,
+      "step": 84000
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0004760204524841897,
+      "loss": 0.4985,
+      "step": 84500
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0004744269554914683,
+      "loss": 0.4982,
+      "step": 85000
+    },
+    {
+      "epoch": 5.21,
+      "eval_loss": 0.8464594483375549,
+      "eval_runtime": 1.368,
+      "eval_samples_per_second": 730.977,
+      "eval_steps_per_second": 11.696,
+      "step": 85000
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00047282604721484643,
+      "loss": 0.4979,
+      "step": 85500
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0004712177976832385,
+      "loss": 0.4978,
+      "step": 86000
+    },
+    {
+      "epoch": 5.27,
+      "eval_loss": 0.8433752655982971,
+      "eval_runtime": 1.3279,
+      "eval_samples_per_second": 753.055,
+      "eval_steps_per_second": 12.049,
+      "step": 86000
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0004696022772466888,
+      "loss": 0.4974,
+      "step": 86500
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00046797955657329487,
+      "loss": 0.4969,
+      "step": 87000
+    },
+    {
+      "epoch": 5.33,
+      "eval_loss": 0.843237042427063,
+      "eval_runtime": 1.3349,
+      "eval_samples_per_second": 749.108,
+      "eval_steps_per_second": 11.986,
+      "step": 87000
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.000466349706646116,
+      "loss": 0.4967,
+      "step": 87500
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00046471279876006763,
+      "loss": 0.4964,
+      "step": 88000
+    },
+    {
+      "epoch": 5.39,
+      "eval_loss": 0.8417282104492188,
+      "eval_runtime": 1.3291,
+      "eval_samples_per_second": 752.394,
+      "eval_steps_per_second": 12.038,
+      "step": 88000
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00046306890451880395,
+      "loss": 0.4964,
+      "step": 88500
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0004614180958315844,
+      "loss": 0.4957,
+      "step": 89000
+    },
+    {
+      "epoch": 5.45,
+      "eval_loss": 0.8362923860549927,
+      "eval_runtime": 1.2708,
+      "eval_samples_per_second": 786.921,
+      "eval_steps_per_second": 12.591,
+      "step": 89000
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00045976044491012884,
+      "loss": 0.4954,
+      "step": 89500
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00045809602426545847,
+      "loss": 0.495,
+      "step": 90000
+    },
+    {
+      "epoch": 5.51,
+      "eval_loss": 0.8392152190208435,
+      "eval_runtime": 1.26,
+      "eval_samples_per_second": 793.671,
+      "eval_steps_per_second": 12.699,
+      "step": 90000
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00045642490670472436,
+      "loss": 0.4947,
+      "step": 90500
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0004547471653280225,
+      "loss": 0.4946,
+      "step": 91000
+    },
+    {
+      "epoch": 5.58,
+      "eval_loss": 0.8400516510009766,
+      "eval_runtime": 1.2358,
+      "eval_samples_per_second": 809.182,
+      "eval_steps_per_second": 12.947,
+      "step": 91000
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00045306287352519543,
+      "loss": 0.4939,
+      "step": 91500
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00045137210497262333,
+      "loss": 0.4935,
+      "step": 92000
+    },
+    {
+      "epoch": 5.64,
+      "eval_loss": 0.8373098969459534,
+      "eval_runtime": 1.3142,
+      "eval_samples_per_second": 760.929,
+      "eval_steps_per_second": 12.175,
+      "step": 92000
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0004496749336299999,
+      "loss": 0.4931,
+      "step": 92500
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0004479714337370977,
+      "loss": 0.4929,
+      "step": 93000
+    },
+    {
+      "epoch": 5.7,
+      "eval_loss": 0.840123176574707,
+      "eval_runtime": 1.2896,
+      "eval_samples_per_second": 775.451,
+      "eval_steps_per_second": 12.407,
+      "step": 93000
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00044626167981052036,
+      "loss": 0.4924,
+      "step": 93500
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00044454574664044404,
+      "loss": 0.492,
+      "step": 94000
+    },
+    {
+      "epoch": 5.76,
+      "eval_loss": 0.8355880379676819,
+      "eval_runtime": 1.2671,
+      "eval_samples_per_second": 789.19,
+      "eval_steps_per_second": 12.627,
+      "step": 94000
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.000442823709287344,
+      "loss": 0.4916,
+      "step": 94500
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0004410956430787129,
+      "loss": 0.4912,
+      "step": 95000
+    },
+    {
+      "epoch": 5.82,
+      "eval_loss": 0.8333644866943359,
+      "eval_runtime": 1.299,
+      "eval_samples_per_second": 769.834,
+      "eval_steps_per_second": 12.317,
+      "step": 95000
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0004393616236057647,
+      "loss": 0.4912,
+      "step": 95500
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00043762172672012875,
+      "loss": 0.4904,
+      "step": 96000
+    },
+    {
+      "epoch": 5.88,
+      "eval_loss": 0.8280515074729919,
+      "eval_runtime": 1.2803,
+      "eval_samples_per_second": 781.038,
+      "eval_steps_per_second": 12.497,
+      "step": 96000
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0004358760285305312,
+      "loss": 0.4901,
+      "step": 96500
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0004341246053994663,
+      "loss": 0.4898,
+      "step": 97000
+    },
+    {
+      "epoch": 5.94,
+      "eval_loss": 0.8338386416435242,
+      "eval_runtime": 1.3216,
+      "eval_samples_per_second": 756.685,
+      "eval_steps_per_second": 12.107,
+      "step": 97000
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00043236753393985534,
+      "loss": 0.4892,
+      "step": 97500
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0004306048910116964,
+      "loss": 0.4891,
+      "step": 98000
+    },
+    {
+      "epoch": 6.0,
+      "eval_loss": 0.8300430774688721,
+      "eval_runtime": 1.316,
+      "eval_samples_per_second": 759.879,
+      "eval_steps_per_second": 12.158,
+      "step": 98000
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0004288367537187012,
+      "loss": 0.4887,
+      "step": 98500
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00042706319940492284,
+      "loss": 0.4882,
+      "step": 99000
+    },
+    {
+      "epoch": 6.07,
+      "eval_loss": 0.8262238502502441,
+      "eval_runtime": 1.3745,
+      "eval_samples_per_second": 727.548,
+      "eval_steps_per_second": 11.641,
+      "step": 99000
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00042528430565137254,
+      "loss": 0.488,
+      "step": 99500
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00042350015027262593,
+      "loss": 0.4876,
+      "step": 100000
+    },
+    {
+      "epoch": 6.13,
+      "eval_loss": 0.8171582221984863,
+      "eval_runtime": 1.3024,
+      "eval_samples_per_second": 767.786,
+      "eval_steps_per_second": 12.285,
+      "step": 100000
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00042171081131341917,
+      "loss": 0.4877,
+      "step": 100500
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00041991636704523497,
+      "loss": 0.4868,
+      "step": 101000
+    },
+    {
+      "epoch": 6.19,
+      "eval_loss": 0.8239555954933167,
+      "eval_runtime": 1.2567,
+      "eval_samples_per_second": 795.735,
+      "eval_steps_per_second": 12.732,
+      "step": 101000
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00041811689596287893,
+      "loss": 0.4864,
+      "step": 101500
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0004163124767810454,
+      "loss": 0.4861,
+      "step": 102000
+    },
+    {
+      "epoch": 6.25,
+      "eval_loss": 0.8212010264396667,
+      "eval_runtime": 1.2247,
+      "eval_samples_per_second": 816.535,
+      "eval_steps_per_second": 13.065,
+      "step": 102000
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00041450318843087506,
+      "loss": 0.4858,
+      "step": 102500
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00041268911005650166,
+      "loss": 0.4854,
+      "step": 103000
+    },
+    {
+      "epoch": 6.31,
+      "eval_loss": 0.8242572546005249,
+      "eval_runtime": 1.357,
+      "eval_samples_per_second": 736.942,
+      "eval_steps_per_second": 11.791,
+      "step": 103000
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00041087032101159006,
+      "loss": 0.4851,
+      "step": 103500
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00040904690085586515,
+      "loss": 0.4847,
+      "step": 104000
+    },
+    {
+      "epoch": 6.37,
+      "eval_loss": 0.8227641582489014,
+      "eval_runtime": 1.3326,
+      "eval_samples_per_second": 750.437,
+      "eval_steps_per_second": 12.007,
+      "step": 104000
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0004072189293516316,
+      "loss": 0.4844,
+      "step": 104500
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0004053864864602847,
+      "loss": 0.4841,
+      "step": 105000
+    },
+    {
+      "epoch": 6.43,
+      "eval_loss": 0.8184976577758789,
+      "eval_runtime": 1.3395,
+      "eval_samples_per_second": 746.538,
+      "eval_steps_per_second": 11.945,
+      "step": 105000
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00040354965233881297,
+      "loss": 0.4846,
+      "step": 105500
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0004017085073362913,
+      "loss": 0.4837,
+      "step": 106000
+    },
+    {
+      "epoch": 6.5,
+      "eval_loss": 0.8177208304405212,
+      "eval_runtime": 1.2838,
+      "eval_samples_per_second": 778.932,
+      "eval_steps_per_second": 12.463,
+      "step": 106000
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00039986313199036664,
+      "loss": 0.4828,
+      "step": 106500
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00039801360702373484,
+      "loss": 0.4827,
+      "step": 107000
+    },
+    {
+      "epoch": 6.56,
+      "eval_loss": 0.8140051364898682,
+      "eval_runtime": 1.3554,
+      "eval_samples_per_second": 737.763,
+      "eval_steps_per_second": 11.804,
+      "step": 107000
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00039616001334060954,
+      "loss": 0.4824,
+      "step": 107500
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00039430243202318314,
+      "loss": 0.4819,
+      "step": 108000
+    },
+    {
+      "epoch": 6.62,
+      "eval_loss": 0.8147432208061218,
+      "eval_runtime": 1.3502,
+      "eval_samples_per_second": 740.632,
+      "eval_steps_per_second": 11.85,
+      "step": 108000
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00039244094432808034,
+      "loss": 0.4816,
+      "step": 108500
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0003905756316828033,
+      "loss": 0.4813,
+      "step": 109000
+    },
+    {
+      "epoch": 6.68,
+      "eval_loss": 0.8172094225883484,
+      "eval_runtime": 1.4168,
+      "eval_samples_per_second": 705.811,
+      "eval_steps_per_second": 11.293,
+      "step": 109000
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00038870657568216963,
+      "loss": 0.4808,
+      "step": 109500
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00038683385808474416,
+      "loss": 0.4807,
+      "step": 110000
+    },
+    {
+      "epoch": 6.74,
+      "eval_loss": 0.8148666024208069,
+      "eval_runtime": 1.2865,
+      "eval_samples_per_second": 777.303,
+      "eval_steps_per_second": 12.437,
+      "step": 110000
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00038495756080926107,
+      "loss": 0.4802,
+      "step": 110500
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0003830777659310416,
+      "loss": 0.4801,
+      "step": 111000
+    },
+    {
+      "epoch": 6.8,
+      "eval_loss": 0.8152140974998474,
+      "eval_runtime": 1.3248,
+      "eval_samples_per_second": 754.82,
+      "eval_steps_per_second": 12.077,
+      "step": 111000
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0003811945556784033,
+      "loss": 0.4795,
+      "step": 111500
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00037930801242906366,
+      "loss": 0.4792,
+      "step": 112000
+    },
+    {
+      "epoch": 6.86,
+      "eval_loss": 0.8089267015457153,
+      "eval_runtime": 1.2928,
+      "eval_samples_per_second": 773.538,
+      "eval_steps_per_second": 12.377,
+      "step": 112000
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00037741821870653576,
+      "loss": 0.4789,
+      "step": 112500
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00037552525717651905,
+      "loss": 0.4785,
+      "step": 113000
+    },
+    {
+      "epoch": 6.92,
+      "eval_loss": 0.808442234992981,
+      "eval_runtime": 1.3221,
+      "eval_samples_per_second": 756.389,
+      "eval_steps_per_second": 12.102,
+      "step": 113000
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0003736292106432834,
+      "loss": 0.4782,
+      "step": 113500
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00037173016204604647,
+      "loss": 0.4777,
+      "step": 114000
+    },
+    {
+      "epoch": 6.99,
+      "eval_loss": 0.8103413581848145,
+      "eval_runtime": 1.3239,
+      "eval_samples_per_second": 755.361,
+      "eval_steps_per_second": 12.086,
+      "step": 114000
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.0003698281944553456,
+      "loss": 0.4774,
+      "step": 114500
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.0003679233910694053,
+      "loss": 0.477,
+      "step": 115000
+    },
+    {
+      "epoch": 7.05,
+      "eval_loss": 0.8103800415992737,
+      "eval_runtime": 1.3524,
+      "eval_samples_per_second": 739.44,
+      "eval_steps_per_second": 11.831,
+      "step": 115000
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.000366015835210496,
+      "loss": 0.4767,
+      "step": 115500
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.0003641056103212908,
+      "loss": 0.4772,
+      "step": 116000
+    },
+    {
+      "epoch": 7.11,
+      "eval_loss": 0.8142406344413757,
+      "eval_runtime": 1.3149,
+      "eval_samples_per_second": 760.539,
+      "eval_steps_per_second": 12.169,
+      "step": 116000
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.00036219279996121446,
+      "loss": 0.4758,
+      "step": 116500
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.0003602774878027888,
+      "loss": 0.4754,
+      "step": 117000
+    },
+    {
+      "epoch": 7.17,
+      "eval_loss": 0.8158754706382751,
+      "eval_runtime": 1.2794,
+      "eval_samples_per_second": 781.632,
+      "eval_steps_per_second": 12.506,
+      "step": 117000
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.00035835975762797245,
+      "loss": 0.4749,
+      "step": 117500
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.0003564396933244957,
+      "loss": 0.4748,
+      "step": 118000
+    },
+    {
+      "epoch": 7.23,
+      "eval_loss": 0.8092423677444458,
+      "eval_runtime": 1.319,
+      "eval_samples_per_second": 758.146,
+      "eval_steps_per_second": 12.13,
+      "step": 118000
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.0003545173788821915,
+      "loss": 0.4742,
+      "step": 118500
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.00035259289838932104,
+      "loss": 0.4738,
+      "step": 119000
+    },
+    {
+      "epoch": 7.29,
+      "eval_loss": 0.8036469221115112,
+      "eval_runtime": 1.3425,
+      "eval_samples_per_second": 744.902,
+      "eval_steps_per_second": 11.918,
+      "step": 119000
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.0003506663360288954,
+      "loss": 0.4736,
+      "step": 119500
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.00034873777607499334,
+      "loss": 0.473,
+      "step": 120000
+    },
+    {
+      "epoch": 7.35,
+      "eval_loss": 0.8085124492645264,
+      "eval_runtime": 1.3135,
+      "eval_samples_per_second": 761.353,
+      "eval_steps_per_second": 12.182,
+      "step": 120000
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.0003468073028890751,
+      "loss": 0.4725,
+      "step": 120500
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.0003448750009162916,
+      "loss": 0.4724,
+      "step": 121000
+    },
+    {
+      "epoch": 7.41,
+      "eval_loss": 0.8083821535110474,
+      "eval_runtime": 1.2822,
+      "eval_samples_per_second": 779.917,
+      "eval_steps_per_second": 12.479,
+      "step": 121000
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.00034294095468179094,
+      "loss": 0.4721,
+      "step": 121500
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.00034100524878702073,
+      "loss": 0.4714,
+      "step": 122000
+    },
+    {
+      "epoch": 7.48,
+      "eval_loss": 0.8066253662109375,
+      "eval_runtime": 1.3256,
+      "eval_samples_per_second": 754.393,
+      "eval_steps_per_second": 12.07,
+      "step": 122000
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.000339067967906028,
+      "loss": 0.4711,
+      "step": 122500
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.0003371291967817539,
+      "loss": 0.4705,
+      "step": 123000
+    },
+    {
+      "epoch": 7.54,
+      "eval_loss": 0.8094301819801331,
+      "eval_runtime": 1.2682,
+      "eval_samples_per_second": 788.537,
+      "eval_steps_per_second": 12.617,
+      "step": 123000
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 0.0003351890202223285,
+      "loss": 0.4704,
+      "step": 123500
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 0.0003332475230973597,
+      "loss": 0.4699,
+      "step": 124000
+    },
+    {
+      "epoch": 7.6,
+      "eval_loss": 0.8094833493232727,
+      "eval_runtime": 1.3119,
+      "eval_samples_per_second": 762.242,
+      "eval_steps_per_second": 12.196,
+      "step": 124000
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.00033130479033422134,
+      "loss": 0.4697,
+      "step": 124500
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 0.0003293609069143381,
+      "loss": 0.4693,
+      "step": 125000
+    },
+    {
+      "epoch": 7.66,
+      "eval_loss": 0.8100947141647339,
+      "eval_runtime": 1.3107,
+      "eval_samples_per_second": 762.972,
+      "eval_steps_per_second": 12.208,
+      "step": 125000
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.00032741595786946783,
+      "loss": 0.4688,
+      "step": 125500
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.000325470028277983,
+      "loss": 0.4685,
+      "step": 126000
+    },
+    {
+      "epoch": 7.72,
+      "eval_loss": 0.8091694116592407,
+      "eval_runtime": 1.2961,
+      "eval_samples_per_second": 771.538,
+      "eval_steps_per_second": 12.345,
+      "step": 126000
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.00032352320326114754,
+      "loss": 0.4686,
+      "step": 126500
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 0.00032157556797939436,
+      "loss": 0.4679,
+      "step": 127000
+    },
+    {
+      "epoch": 7.78,
+      "eval_loss": 0.8025205135345459,
+      "eval_runtime": 1.2855,
+      "eval_samples_per_second": 777.911,
+      "eval_steps_per_second": 12.447,
+      "step": 127000
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 0.00031962720762860057,
+      "loss": 0.4676,
+      "step": 127500
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 0.0003176782074363595,
+      "loss": 0.4672,
+      "step": 128000
+    },
+    {
+      "epoch": 7.84,
+      "eval_loss": 0.800028920173645,
+      "eval_runtime": 1.3752,
+      "eval_samples_per_second": 727.176,
+      "eval_steps_per_second": 11.635,
+      "step": 128000
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.0003157286526582535,
+      "loss": 0.4669,
+      "step": 128500
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.0003137786285741241,
+      "loss": 0.4665,
+      "step": 129000
+    },
+    {
+      "epoch": 7.9,
+      "eval_loss": 0.802001416683197,
+      "eval_runtime": 1.3448,
+      "eval_samples_per_second": 743.623,
+      "eval_steps_per_second": 11.898,
+      "step": 129000
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.0003118282204843421,
+      "loss": 0.4663,
+      "step": 129500
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.0003098775137060758,
+      "loss": 0.4659,
+      "step": 130000
+    },
+    {
+      "epoch": 7.97,
+      "eval_loss": 0.8022414445877075,
+      "eval_runtime": 1.3715,
+      "eval_samples_per_second": 729.142,
+      "eval_steps_per_second": 11.666,
+      "step": 130000
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 0.00030792659356955893,
+      "loss": 0.4657,
+      "step": 130500
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 0.0003059755454143586,
+      "loss": 0.4653,
+      "step": 131000
+    },
+    {
+      "epoch": 8.03,
+      "eval_loss": 0.8070600628852844,
+      "eval_runtime": 1.2854,
+      "eval_samples_per_second": 777.995,
+      "eval_steps_per_second": 12.448,
+      "step": 131000
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 0.00030402445458564144,
+      "loss": 0.4649,
+      "step": 131500
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 0.0003020734064304411,
+      "loss": 0.4647,
+      "step": 132000
+    },
+    {
+      "epoch": 8.09,
+      "eval_loss": 0.799366295337677,
+      "eval_runtime": 1.2985,
+      "eval_samples_per_second": 770.128,
+      "eval_steps_per_second": 12.322,
+      "step": 132000
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 0.00030012248629392423,
+      "loss": 0.4647,
+      "step": 132500
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 0.00029817177951565793,
+      "loss": 0.4639,
+      "step": 133000
+    },
+    {
+      "epoch": 8.15,
+      "eval_loss": 0.8033633232116699,
+      "eval_runtime": 1.2955,
+      "eval_samples_per_second": 771.897,
+      "eval_steps_per_second": 12.35,
+      "step": 133000
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 0.00029622137142587594,
+      "loss": 0.4637,
+      "step": 133500
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 0.0002942713473417466,
+      "loss": 0.4634,
+      "step": 134000
+    },
+    {
+      "epoch": 8.21,
+      "eval_loss": 0.8022355437278748,
+      "eval_runtime": 1.3019,
+      "eval_samples_per_second": 768.12,
+      "eval_steps_per_second": 12.29,
+      "step": 134000
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 0.00029232179256364054,
+      "loss": 0.4631,
+      "step": 134500
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 0.0002903727923713994,
+      "loss": 0.4656,
+      "step": 135000
+    },
+    {
+      "epoch": 8.27,
+      "eval_loss": 0.8051571249961853,
+      "eval_runtime": 1.3053,
+      "eval_samples_per_second": 766.083,
+      "eval_steps_per_second": 12.257,
+      "step": 135000
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 0.00028842443202060556,
+      "loss": 0.4625,
+      "step": 135500
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 0.00028647679673885255,
+      "loss": 0.4623,
+      "step": 136000
+    },
+    {
+      "epoch": 8.33,
+      "eval_loss": 0.7988797426223755,
+      "eval_runtime": 1.3231,
+      "eval_samples_per_second": 755.795,
+      "eval_steps_per_second": 12.093,
+      "step": 136000
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 0.000284529971722017,
+      "loss": 0.462,
+      "step": 136500
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 0.0002825840421305321,
+      "loss": 0.4617,
+      "step": 137000
+    },
+    {
+      "epoch": 8.39,
+      "eval_loss": 0.7993477582931519,
+      "eval_runtime": 1.2892,
+      "eval_samples_per_second": 775.645,
+      "eval_steps_per_second": 12.41,
+      "step": 137000
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 0.00028063909308566196,
+      "loss": 0.4616,
+      "step": 137500
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 0.00027869520966577874,
+      "loss": 0.4612,
+      "step": 138000
+    },
+    {
+      "epoch": 8.46,
+      "eval_loss": 0.8003228902816772,
+      "eval_runtime": 1.2968,
+      "eval_samples_per_second": 771.107,
+      "eval_steps_per_second": 12.338,
+      "step": 138000
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 0.00027675247690264027,
+      "loss": 0.461,
+      "step": 138500
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 0.0002748109797776715,
+      "loss": 0.4608,
+      "step": 139000
+    },
+    {
+      "epoch": 8.52,
+      "eval_loss": 0.7989851236343384,
+      "eval_runtime": 1.2801,
+      "eval_samples_per_second": 781.164,
+      "eval_steps_per_second": 12.499,
+      "step": 139000
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 0.0002728708032182461,
+      "loss": 0.4603,
+      "step": 139500
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 0.0002709320320939721,
+      "loss": 0.4603,
+      "step": 140000
+    },
+    {
+      "epoch": 8.58,
+      "eval_loss": 0.8073873519897461,
+      "eval_runtime": 1.2866,
+      "eval_samples_per_second": 777.265,
+      "eval_steps_per_second": 12.436,
+      "step": 140000
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 0.00026899475121297924,
+      "loss": 0.46,
+      "step": 140500
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 0.00026705904531820914,
+      "loss": 0.4597,
+      "step": 141000
+    },
+    {
+      "epoch": 8.64,
+      "eval_loss": 0.8088939189910889,
+      "eval_runtime": 1.2732,
+      "eval_samples_per_second": 785.451,
+      "eval_steps_per_second": 12.567,
+      "step": 141000
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 0.0002651249990837085,
+      "loss": 0.4596,
+      "step": 141500
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 0.00026319269711092485,
+      "loss": 0.4591,
+      "step": 142000
+    },
+    {
+      "epoch": 8.7,
+      "eval_loss": 0.8040044903755188,
+      "eval_runtime": 1.2887,
+      "eval_samples_per_second": 775.97,
+      "eval_steps_per_second": 12.416,
+      "step": 142000
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 0.0002612622239250066,
+      "loss": 0.459,
+      "step": 142500
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 0.0002593336639711046,
+      "loss": 0.4586,
+      "step": 143000
+    },
+    {
+      "epoch": 8.76,
+      "eval_loss": 0.7993264198303223,
+      "eval_runtime": 1.3493,
+      "eval_samples_per_second": 741.132,
+      "eval_steps_per_second": 11.858,
+      "step": 143000
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 0.000257407101610679,
+      "loss": 0.4583,
+      "step": 143500
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 0.00025548262111780846,
+      "loss": 0.4584,
+      "step": 144000
+    },
+    {
+      "epoch": 8.82,
+      "eval_loss": 0.8003845810890198,
+      "eval_runtime": 1.306,
+      "eval_samples_per_second": 765.701,
+      "eval_steps_per_second": 12.251,
+      "step": 144000
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 0.0002535603066755043,
+      "loss": 0.4579,
+      "step": 144500
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 0.00025164024237202764,
+      "loss": 0.4594,
+      "step": 145000
+    },
+    {
+      "epoch": 8.88,
+      "eval_loss": 0.7990729808807373,
+      "eval_runtime": 1.2616,
+      "eval_samples_per_second": 792.645,
+      "eval_steps_per_second": 12.682,
+      "step": 145000
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 0.00024972251219721115,
+      "loss": 0.4573,
+      "step": 145500
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 0.00024780720003878557,
+      "loss": 0.4574,
+      "step": 146000
+    },
+    {
+      "epoch": 8.95,
+      "eval_loss": 0.7956343293190002,
+      "eval_runtime": 1.2847,
+      "eval_samples_per_second": 778.391,
+      "eval_steps_per_second": 12.454,
+      "step": 146000
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 0.00024589438967870925,
+      "loss": 0.4571,
+      "step": 146500
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 0.00024398416478950394,
+      "loss": 0.4571,
+      "step": 147000
+    },
+    {
+      "epoch": 9.01,
+      "eval_loss": 0.7948459386825562,
+      "eval_runtime": 1.3733,
+      "eval_samples_per_second": 728.166,
+      "eval_steps_per_second": 11.651,
+      "step": 147000
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 0.00024207660893059467,
+      "loss": 0.4565,
+      "step": 147500
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 0.0002401718055446543,
+      "loss": 0.4565,
+      "step": 148000
+    },
+    {
+      "epoch": 9.07,
+      "eval_loss": 0.7982079982757568,
+      "eval_runtime": 1.249,
+      "eval_samples_per_second": 800.631,
+      "eval_steps_per_second": 12.81,
+      "step": 148000
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 0.00023826983795395364,
+      "loss": 0.4561,
+      "step": 148500
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 0.00023637078935671656,
+      "loss": 0.4563,
+      "step": 149000
+    },
+    {
+      "epoch": 9.13,
+      "eval_loss": 0.7960088849067688,
+      "eval_runtime": 1.2502,
+      "eval_samples_per_second": 799.9,
+      "eval_steps_per_second": 12.798,
+      "step": 149000
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 0.00023447474282348085,
+      "loss": 0.4558,
+      "step": 149500
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 0.00023258178129346424,
+      "loss": 0.4555,
+      "step": 150000
+    },
+    {
+      "epoch": 9.19,
+      "eval_loss": 0.8043127655982971,
+      "eval_runtime": 1.3326,
+      "eval_samples_per_second": 750.39,
+      "eval_steps_per_second": 12.006,
+      "step": 150000
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 0.00023069198757093631,
+      "loss": 0.4555,
+      "step": 150500
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 0.00022880544432159663,
+      "loss": 0.4551,
+      "step": 151000
+    },
+    {
+      "epoch": 9.25,
+      "eval_loss": 0.8021363615989685,
+      "eval_runtime": 1.3155,
+      "eval_samples_per_second": 760.176,
+      "eval_steps_per_second": 12.163,
+      "step": 151000
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 0.00022692223406895848,
+      "loss": 0.4554,
+      "step": 151500
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 0.000225042439190739,
+      "loss": 0.4549,
+      "step": 152000
+    },
+    {
+      "epoch": 9.31,
+      "eval_loss": 0.7971563339233398,
+      "eval_runtime": 1.2684,
+      "eval_samples_per_second": 788.389,
+      "eval_steps_per_second": 12.614,
+      "step": 152000
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 0.00022316614191525587,
+      "loss": 0.4546,
+      "step": 152500
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 0.00022129342431783026,
+      "loss": 0.4545,
+      "step": 153000
+    },
+    {
+      "epoch": 9.38,
+      "eval_loss": 0.8002565503120422,
+      "eval_runtime": 1.2924,
+      "eval_samples_per_second": 773.739,
+      "eval_steps_per_second": 12.38,
+      "step": 153000
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 0.00021942436831719677,
+      "loss": 0.4542,
+      "step": 153500
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 0.00021755905567191967,
+      "loss": 0.4542,
+      "step": 154000
+    },
+    {
+      "epoch": 9.44,
+      "eval_loss": 0.7999687790870667,
+      "eval_runtime": 1.2876,
+      "eval_samples_per_second": 776.658,
+      "eval_steps_per_second": 12.427,
+      "step": 154000
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 0.00021569756797681686,
+      "loss": 0.4538,
+      "step": 154500
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 0.00021383998665939054,
+      "loss": 0.4539,
+      "step": 155000
+    },
+    {
+      "epoch": 9.5,
+      "eval_loss": 0.7960466146469116,
+      "eval_runtime": 1.3325,
+      "eval_samples_per_second": 750.454,
+      "eval_steps_per_second": 12.007,
+      "step": 155000
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 0.00021198639297626516,
+      "loss": 0.4537,
+      "step": 155500
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 0.0002101368680096334,
+      "loss": 0.4533,
+      "step": 156000
+    },
+    {
+      "epoch": 9.56,
+      "eval_loss": 0.8035251498222351,
+      "eval_runtime": 1.3033,
+      "eval_samples_per_second": 767.279,
+      "eval_steps_per_second": 12.276,
+      "step": 156000
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 0.00020829149266370862,
+      "loss": 0.4532,
+      "step": 156500
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 0.00020645034766118703,
+      "loss": 0.453,
+      "step": 157000
+    },
+    {
+      "epoch": 9.62,
+      "eval_loss": 0.7953096628189087,
+      "eval_runtime": 1.3024,
+      "eval_samples_per_second": 767.789,
+      "eval_steps_per_second": 12.285,
+      "step": 157000
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 0.00020461351353971526,
+      "loss": 0.4527,
+      "step": 157500
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 0.00020278107064836847,
+      "loss": 0.4527,
+      "step": 158000
+    },
+    {
+      "epoch": 9.68,
+      "eval_loss": 0.7937498688697815,
+      "eval_runtime": 1.2999,
+      "eval_samples_per_second": 769.267,
+      "eval_steps_per_second": 12.308,
+      "step": 158000
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 0.00020095309914413485,
+      "loss": 0.4526,
+      "step": 158500
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 0.00019912967898840997,
+      "loss": 0.4524,
+      "step": 159000
+    },
+    {
+      "epoch": 9.74,
+      "eval_loss": 0.8021422624588013,
+      "eval_runtime": 1.3181,
+      "eval_samples_per_second": 758.64,
+      "eval_steps_per_second": 12.138,
+      "step": 159000
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 0.00019731088994349834,
+      "loss": 0.4524,
+      "step": 159500
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 0.0001954968115691248,
+      "loss": 0.4519,
+      "step": 160000
+    },
+    {
+      "epoch": 9.8,
+      "eval_loss": 0.8028143644332886,
+      "eval_runtime": 1.282,
+      "eval_samples_per_second": 780.028,
+      "eval_steps_per_second": 12.48,
+      "step": 160000
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 0.00019368752321895452,
+      "loss": 0.4519,
+      "step": 160500
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 0.00019188310403712105,
+      "loss": 0.4517,
+      "step": 161000
+    },
+    {
+      "epoch": 9.87,
+      "eval_loss": 0.8005779385566711,
+      "eval_runtime": 1.3237,
+      "eval_samples_per_second": 755.47,
+      "eval_steps_per_second": 12.088,
+      "step": 161000
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 0.00019008363295476495,
+      "loss": 0.4516,
+      "step": 161500
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 0.0001882891886865808,
+      "loss": 0.4514,
+      "step": 162000
+    },
+    {
+      "epoch": 9.93,
+      "eval_loss": 0.8066567182540894,
+      "eval_runtime": 1.307,
+      "eval_samples_per_second": 765.087,
+      "eval_steps_per_second": 12.241,
+      "step": 162000
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 0.00018649984972737404,
+      "loss": 0.4513,
+      "step": 162500
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 0.00018471569434862749,
+      "loss": 0.4512,
+      "step": 163000
+    },
+    {
+      "epoch": 9.99,
+      "eval_loss": 0.7989670038223267,
+      "eval_runtime": 1.3388,
+      "eval_samples_per_second": 746.958,
+      "eval_steps_per_second": 11.951,
+      "step": 163000
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 0.00018293680059507713,
+      "loss": 0.4512,
+      "step": 163500
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 0.00018116324628129882,
+      "loss": 0.4508,
+      "step": 164000
+    },
+    {
+      "epoch": 10.05,
+      "eval_loss": 0.804061770439148,
+      "eval_runtime": 1.3751,
+      "eval_samples_per_second": 727.204,
+      "eval_steps_per_second": 11.635,
+      "step": 164000
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 0.00017939510898830357,
+      "loss": 0.4505,
+      "step": 164500
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 0.0001776324660601446,
+      "loss": 0.4504,
+      "step": 165000
+    },
+    {
+      "epoch": 10.11,
+      "eval_loss": 0.7995474934577942,
+      "eval_runtime": 1.3093,
+      "eval_samples_per_second": 763.796,
+      "eval_steps_per_second": 12.221,
+      "step": 165000
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 0.00017587539460053368,
+      "loss": 0.4505,
+      "step": 165500
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 0.0001741239714694688,
+      "loss": 0.4501,
+      "step": 166000
+    },
+    {
+      "epoch": 10.17,
+      "eval_loss": 0.7978888154029846,
+      "eval_runtime": 1.2913,
+      "eval_samples_per_second": 774.39,
+      "eval_steps_per_second": 12.39,
+      "step": 166000
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 0.0001723782732798713,
+      "loss": 0.4501,
+      "step": 166500
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 0.00017063837639423517,
+      "loss": 0.4499,
+      "step": 167000
+    },
+    {
+      "epoch": 10.23,
+      "eval_loss": 0.796922504901886,
+      "eval_runtime": 1.3094,
+      "eval_samples_per_second": 763.7,
+      "eval_steps_per_second": 12.219,
+      "step": 167000
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 0.00016890435692128712,
+      "loss": 0.4498,
+      "step": 167500
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 0.0001671762907126559,
+      "loss": 0.4497,
+      "step": 168000
+    },
+    {
+      "epoch": 10.29,
+      "eval_loss": 0.8040737509727478,
+      "eval_runtime": 1.3701,
+      "eval_samples_per_second": 729.885,
+      "eval_steps_per_second": 11.678,
+      "step": 168000
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 0.00016545425335955596,
+      "loss": 0.4498,
+      "step": 168500
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 0.00016373832018947945,
+      "loss": 0.4495,
+      "step": 169000
+    },
+    {
+      "epoch": 10.36,
+      "eval_loss": 0.8050036430358887,
+      "eval_runtime": 1.3107,
+      "eval_samples_per_second": 762.923,
+      "eval_steps_per_second": 12.207,
+      "step": 169000
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 0.0001620285662629024,
+      "loss": 0.4492,
+      "step": 169500
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 0.0001603250663700002,
+      "loss": 0.4492,
+      "step": 170000
+    },
+    {
+      "epoch": 10.42,
+      "eval_loss": 0.7998891472816467,
+      "eval_runtime": 1.3642,
+      "eval_samples_per_second": 733.005,
+      "eval_steps_per_second": 11.728,
+      "step": 170000
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 0.00015862789502737648,
+      "loss": 0.4491,
+      "step": 170500
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 0.00015693712647480446,
+      "loss": 0.4494,
+      "step": 171000
+    },
+    {
+      "epoch": 10.48,
+      "eval_loss": 0.7991927266120911,
+      "eval_runtime": 1.3282,
+      "eval_samples_per_second": 752.902,
+      "eval_steps_per_second": 12.046,
+      "step": 171000
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 0.00015525283467197743,
+      "loss": 0.4487,
+      "step": 171500
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 0.00015357509329527556,
+      "loss": 0.4486,
+      "step": 172000
+    },
+    {
+      "epoch": 10.54,
+      "eval_loss": 0.8018996715545654,
+      "eval_runtime": 1.3294,
+      "eval_samples_per_second": 752.227,
+      "eval_steps_per_second": 12.036,
+      "step": 172000
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 0.00015190397573454158,
+      "loss": 0.4488,
+      "step": 172500
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 0.00015023955508987127,
+      "loss": 0.4485,
+      "step": 173000
+    },
+    {
+      "epoch": 10.6,
+      "eval_loss": 0.8025578260421753,
+      "eval_runtime": 1.3279,
+      "eval_samples_per_second": 753.043,
+      "eval_steps_per_second": 12.049,
+      "step": 173000
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 0.00014858190416841565,
+      "loss": 0.4483,
+      "step": 173500
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 0.00014693109548119591,
+      "loss": 0.4483,
+      "step": 174000
+    },
+    {
+      "epoch": 10.66,
+      "eval_loss": 0.8008602261543274,
+      "eval_runtime": 1.2568,
+      "eval_samples_per_second": 795.696,
+      "eval_steps_per_second": 12.731,
+      "step": 174000
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 0.00014528720123993226,
+      "loss": 0.448,
+      "step": 174500
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 0.0001436502933538841,
+      "loss": 0.448,
+      "step": 175000
+    },
+    {
+      "epoch": 10.72,
+      "eval_loss": 0.8021511435508728,
+      "eval_runtime": 1.2829,
+      "eval_samples_per_second": 779.473,
+      "eval_steps_per_second": 12.472,
+      "step": 175000
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 0.00014202044342670508,
+      "loss": 0.448,
+      "step": 175500
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 0.00014039772275331125,
+      "loss": 0.4479,
+      "step": 176000
+    },
+    {
+      "epoch": 10.78,
+      "eval_loss": 0.8016372323036194,
+      "eval_runtime": 1.3768,
+      "eval_samples_per_second": 726.333,
+      "eval_steps_per_second": 11.621,
+      "step": 176000
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 0.00013878220231676152,
+      "loss": 0.4475,
+      "step": 176500
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 0.00013717395278515355,
+      "loss": 0.4476,
+      "step": 177000
+    },
+    {
+      "epoch": 10.85,
+      "eval_loss": 0.7988106608390808,
+      "eval_runtime": 1.2962,
+      "eval_samples_per_second": 771.516,
+      "eval_steps_per_second": 12.344,
+      "step": 177000
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 0.00013557304450853162,
+      "loss": 0.4472,
+      "step": 177500
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 0.00013397954751581014,
+      "loss": 0.4474,
+      "step": 178000
+    },
+    {
+      "epoch": 10.91,
+      "eval_loss": 0.8025058507919312,
+      "eval_runtime": 1.3447,
+      "eval_samples_per_second": 743.664,
+      "eval_steps_per_second": 11.899,
+      "step": 178000
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 0.00013239353151170983,
+      "loss": 0.4471,
+      "step": 178500
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 0.00013081506587370853,
+      "loss": 0.4471,
+      "step": 179000
+    },
+    {
+      "epoch": 10.97,
+      "eval_loss": 0.8035358190536499,
+      "eval_runtime": 1.362,
+      "eval_samples_per_second": 734.202,
+      "eval_steps_per_second": 11.747,
+      "step": 179000
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 0.00012924421964900695,
+      "loss": 0.447,
+      "step": 179500
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 0.00012768106155150758,
+      "loss": 0.4471,
+      "step": 180000
+    },
+    {
+      "epoch": 11.03,
+      "eval_loss": 0.7982646226882935,
+      "eval_runtime": 1.3411,
+      "eval_samples_per_second": 745.664,
+      "eval_steps_per_second": 11.931,
+      "step": 180000
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 0.00012612565995880976,
+      "loss": 0.4467,
+      "step": 180500
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 0.00012457808290921774,
+      "loss": 0.4467,
+      "step": 181000
+    },
+    {
+      "epoch": 11.09,
+      "eval_loss": 0.8010080456733704,
+      "eval_runtime": 1.2919,
+      "eval_samples_per_second": 774.08,
+      "eval_steps_per_second": 12.385,
+      "step": 181000
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 0.00012303839809876525,
+      "loss": 0.4466,
+      "step": 181500
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 0.00012150667287825382,
+      "loss": 0.4463,
+      "step": 182000
+    },
+    {
+      "epoch": 11.15,
+      "eval_loss": 0.8034773468971252,
+      "eval_runtime": 1.2819,
+      "eval_samples_per_second": 780.103,
+      "eval_steps_per_second": 12.482,
+      "step": 182000
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 0.00011998297425030656,
+      "loss": 0.4464,
+      "step": 182500
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 0.00011846736886643775,
+      "loss": 0.4463,
+      "step": 183000
+    },
+    {
+      "epoch": 11.21,
+      "eval_loss": 0.8048831820487976,
+      "eval_runtime": 1.3614,
+      "eval_samples_per_second": 734.528,
+      "eval_steps_per_second": 11.752,
+      "step": 183000
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 0.00011695992302413651,
+      "loss": 0.4462,
+      "step": 183500
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 0.00011546070266396771,
+      "loss": 0.4462,
+      "step": 184000
+    },
+    {
+      "epoch": 11.27,
+      "eval_loss": 0.7998443841934204,
+      "eval_runtime": 1.3341,
+      "eval_samples_per_second": 749.594,
+      "eval_steps_per_second": 11.994,
+      "step": 184000
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 0.00011396977336668645,
+      "loss": 0.4459,
+      "step": 184500
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 0.00011248720035037021,
+      "loss": 0.4459,
+      "step": 185000
+    },
+    {
+      "epoch": 11.34,
+      "eval_loss": 0.7987710237503052,
+      "eval_runtime": 1.3437,
+      "eval_samples_per_second": 744.203,
+      "eval_steps_per_second": 11.907,
+      "step": 185000
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 0.00011101304846756577,
+      "loss": 0.4458,
+      "step": 185500
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 0.00010954738220245183,
+      "loss": 0.4457,
+      "step": 186000
+    },
+    {
+      "epoch": 11.4,
+      "eval_loss": 0.8063639402389526,
+      "eval_runtime": 1.2501,
+      "eval_samples_per_second": 799.917,
+      "eval_steps_per_second": 12.799,
+      "step": 186000
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 0.00010809026566801912,
+      "loss": 0.4457,
+      "step": 186500
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 0.00010664176260326507,
+      "loss": 0.4456,
+      "step": 187000
+    },
+    {
+      "epoch": 11.46,
+      "eval_loss": 0.8042049407958984,
+      "eval_runtime": 1.3155,
+      "eval_samples_per_second": 760.186,
+      "eval_steps_per_second": 12.163,
+      "step": 187000
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 0.00010520193637040641,
+      "loss": 0.4454,
+      "step": 187500
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 0.00010377084995210682,
+      "loss": 0.4454,
+      "step": 188000
+    },
+    {
+      "epoch": 11.52,
+      "eval_loss": 0.7998358607292175,
+      "eval_runtime": 1.3257,
+      "eval_samples_per_second": 754.322,
+      "eval_steps_per_second": 12.069,
+      "step": 188000
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 0.00010234856594872234,
+      "loss": 0.4452,
+      "step": 188500
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 0.00010093514657556295,
+      "loss": 0.4453,
+      "step": 189000
+    },
+    {
+      "epoch": 11.58,
+      "eval_loss": 0.8026143908500671,
+      "eval_runtime": 1.2929,
+      "eval_samples_per_second": 773.468,
+      "eval_steps_per_second": 12.375,
+      "step": 189000
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 9.953065366017073e-05,
+      "loss": 0.4451,
+      "step": 189500
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 9.813514863961586e-05,
+      "loss": 0.4449,
+      "step": 190000
+    },
+    {
+      "epoch": 11.64,
+      "eval_loss": 0.7992528080940247,
+      "eval_runtime": 1.2891,
+      "eval_samples_per_second": 775.752,
+      "eval_steps_per_second": 12.412,
+      "step": 190000
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 9.67486925578087e-05,
+      "loss": 0.4449,
+      "step": 190500
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 9.537134606282964e-05,
+      "loss": 0.4448,
+      "step": 191000
+    },
+    {
+      "epoch": 11.7,
+      "eval_loss": 0.8036643266677856,
+      "eval_runtime": 1.3143,
+      "eval_samples_per_second": 760.834,
+      "eval_steps_per_second": 12.173,
+      "step": 191000
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 9.400316940427652e-05,
+      "loss": 0.4447,
+      "step": 191500
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 9.264422243062844e-05,
+      "loss": 0.4448,
+      "step": 192000
+    },
+    {
+      "epoch": 11.76,
+      "eval_loss": 0.8037863969802856,
+      "eval_runtime": 1.3425,
+      "eval_samples_per_second": 744.863,
+      "eval_steps_per_second": 11.918,
+      "step": 192000
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 9.129456458662876e-05,
+      "loss": 0.4445,
+      "step": 192500
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 8.995425491068365e-05,
+      "loss": 0.4445,
+      "step": 193000
+    },
+    {
+      "epoch": 11.83,
+      "eval_loss": 0.8010460734367371,
+      "eval_runtime": 1.3665,
+      "eval_samples_per_second": 731.803,
+      "eval_steps_per_second": 11.709,
+      "step": 193000
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 8.862335203228025e-05,
+      "loss": 0.4444,
+      "step": 193500
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 8.73019141694222e-05,
+      "loss": 0.4442,
+      "step": 194000
+    },
+    {
+      "epoch": 11.89,
+      "eval_loss": 0.7977059483528137,
+      "eval_runtime": 1.3628,
+      "eval_samples_per_second": 733.81,
+      "eval_steps_per_second": 11.741,
+      "step": 194000
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 8.598999912608229e-05,
+      "loss": 0.4442,
+      "step": 194500
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 8.468766428967468e-05,
+      "loss": 0.4443,
+      "step": 195000
+    },
+    {
+      "epoch": 11.95,
+      "eval_loss": 0.80078125,
+      "eval_runtime": 1.316,
+      "eval_samples_per_second": 759.905,
+      "eval_steps_per_second": 12.158,
+      "step": 195000
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 8.339496662854397e-05,
+      "loss": 0.444,
+      "step": 195500
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.211196268947367e-05,
+      "loss": 0.4441,
+      "step": 196000
+    },
+    {
+      "epoch": 12.01,
+      "eval_loss": 0.8048492670059204,
+      "eval_runtime": 1.3312,
+      "eval_samples_per_second": 751.193,
+      "eval_steps_per_second": 12.019,
+      "step": 196000
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.083870859521251e-05,
+      "loss": 0.4441,
+      "step": 196500
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 7.95752600420192e-05,
+      "loss": 0.4439,
+      "step": 197000
+    },
+    {
+      "epoch": 12.07,
+      "eval_loss": 0.8033810257911682,
+      "eval_runtime": 1.3277,
+      "eval_samples_per_second": 753.203,
+      "eval_steps_per_second": 12.051,
+      "step": 197000
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 7.832167229722666e-05,
+      "loss": 0.4438,
+      "step": 197500
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 7.707800019682362e-05,
+      "loss": 0.4438,
+      "step": 198000
+    },
+    {
+      "epoch": 12.13,
+      "eval_loss": 0.8051833510398865,
+      "eval_runtime": 1.2928,
+      "eval_samples_per_second": 773.488,
+      "eval_steps_per_second": 12.376,
+      "step": 198000
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 7.5844298143057e-05,
+      "loss": 0.4437,
+      "step": 198500
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 7.462062010205106e-05,
+      "loss": 0.4437,
+      "step": 199000
+    },
+    {
+      "epoch": 12.19,
+      "eval_loss": 0.8041102290153503,
+      "eval_runtime": 1.3104,
+      "eval_samples_per_second": 763.133,
+      "eval_steps_per_second": 12.21,
+      "step": 199000
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 7.340701960144751e-05,
+      "loss": 0.4437,
+      "step": 199500
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 7.220354972806392e-05,
+      "loss": 0.4434,
+      "step": 200000
+    },
+    {
+      "epoch": 12.25,
+      "eval_loss": 0.8000948429107666,
+      "eval_runtime": 1.2692,
+      "eval_samples_per_second": 787.882,
+      "eval_steps_per_second": 12.606,
+      "step": 200000
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 7.101026312557103e-05,
+      "loss": 0.4433,
+      "step": 200500
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 6.982721199219075e-05,
+      "loss": 0.4434,
+      "step": 201000
+    },
+    {
+      "epoch": 12.32,
+      "eval_loss": 0.8013458251953125,
+      "eval_runtime": 1.4369,
+      "eval_samples_per_second": 695.945,
+      "eval_steps_per_second": 11.135,
+      "step": 201000
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 6.865444807841203e-05,
+      "loss": 0.4434,
+      "step": 201500
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 6.749202268472787e-05,
+      "loss": 0.4432,
+      "step": 202000
+    },
+    {
+      "epoch": 12.38,
+      "eval_loss": 0.7986797094345093,
+      "eval_runtime": 1.2992,
+      "eval_samples_per_second": 769.697,
+      "eval_steps_per_second": 12.315,
+      "step": 202000
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 6.633998665939053e-05,
+      "loss": 0.4433,
+      "step": 202500
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 6.519839039618793e-05,
+      "loss": 0.443,
+      "step": 203000
+    },
+    {
+      "epoch": 12.44,
+      "eval_loss": 0.7961978316307068,
+      "eval_runtime": 1.3561,
+      "eval_samples_per_second": 737.396,
+      "eval_steps_per_second": 11.798,
+      "step": 203000
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 6.406728383223897e-05,
+      "loss": 0.4431,
+      "step": 203500
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 6.294671644580888e-05,
+      "loss": 0.443,
+      "step": 204000
+    },
+    {
+      "epoch": 12.5,
+      "eval_loss": 0.8016523122787476,
+      "eval_runtime": 1.3833,
+      "eval_samples_per_second": 722.928,
+      "eval_steps_per_second": 11.567,
+      "step": 204000
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 6.18367372541454e-05,
+      "loss": 0.443,
+      "step": 204500
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 6.073739481133408e-05,
+      "loss": 0.4429,
+      "step": 205000
+    },
+    {
+      "epoch": 12.56,
+      "eval_loss": 0.7996025681495667,
+      "eval_runtime": 1.274,
+      "eval_samples_per_second": 784.9,
+      "eval_steps_per_second": 12.558,
+      "step": 205000
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 5.96487372061749e-05,
+      "loss": 0.4428,
+      "step": 205500
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 5.857081206007811e-05,
+      "loss": 0.4428,
+      "step": 206000
+    },
+    {
+      "epoch": 12.62,
+      "eval_loss": 0.7997360229492188,
+      "eval_runtime": 1.3318,
+      "eval_samples_per_second": 750.873,
+      "eval_steps_per_second": 12.014,
+      "step": 206000
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 5.750366652498169e-05,
+      "loss": 0.4426,
+      "step": 206500
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 5.6447347281288474e-05,
+      "loss": 0.4425,
+      "step": 207000
+    },
+    {
+      "epoch": 12.68,
+      "eval_loss": 0.801701009273529,
+      "eval_runtime": 1.3022,
+      "eval_samples_per_second": 767.911,
+      "eval_steps_per_second": 12.287,
+      "step": 207000
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 5.540190053582401e-05,
+      "loss": 0.4426,
+      "step": 207500
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 5.43673720198159e-05,
+      "loss": 0.4424,
+      "step": 208000
+    },
+    {
+      "epoch": 12.75,
+      "eval_loss": 0.8007999062538147,
+      "eval_runtime": 1.3254,
+      "eval_samples_per_second": 754.474,
+      "eval_steps_per_second": 12.072,
+      "step": 208000
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 5.33438069868928e-05,
+      "loss": 0.4425,
+      "step": 208500
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 5.2331250211105e-05,
+      "loss": 0.4424,
+      "step": 209000
+    },
+    {
+      "epoch": 12.81,
+      "eval_loss": 0.8051698207855225,
+      "eval_runtime": 1.3459,
+      "eval_samples_per_second": 742.979,
+      "eval_steps_per_second": 11.888,
+      "step": 209000
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 5.13297459849662e-05,
+      "loss": 0.4424,
+      "step": 209500
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 5.0339338117515696e-05,
+      "loss": 0.4422,
+      "step": 210000
+    },
+    {
+      "epoch": 12.87,
+      "eval_loss": 0.8003759980201721,
+      "eval_runtime": 1.333,
+      "eval_samples_per_second": 750.181,
+      "eval_steps_per_second": 12.003,
+      "step": 210000
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.9360069932402115e-05,
+      "loss": 0.4423,
+      "step": 210500
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.839198426598824e-05,
+      "loss": 0.4421,
+      "step": 211000
+    },
+    {
+      "epoch": 12.93,
+      "eval_loss": 0.8022798895835876,
+      "eval_runtime": 1.2766,
+      "eval_samples_per_second": 783.354,
+      "eval_steps_per_second": 12.534,
+      "step": 211000
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.7435123465477156e-05,
+      "loss": 0.4422,
+      "step": 211500
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.6489529387060306e-05,
+      "loss": 0.4421,
+      "step": 212000
+    },
+    {
+      "epoch": 12.99,
+      "eval_loss": 0.8013637661933899,
+      "eval_runtime": 1.3245,
+      "eval_samples_per_second": 754.989,
+      "eval_steps_per_second": 12.08,
+      "step": 212000
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.555524339408575e-05,
+      "loss": 0.4421,
+      "step": 212500
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.46323063552496e-05,
+      "loss": 0.442,
+      "step": 213000
+    },
+    {
+      "epoch": 13.05,
+      "eval_loss": 0.7998749017715454,
+      "eval_runtime": 1.3047,
+      "eval_samples_per_second": 766.488,
+      "eval_steps_per_second": 12.264,
+      "step": 213000
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.3720758642807817e-05,
+      "loss": 0.4419,
+      "step": 213500
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.2820640130810446e-05,
+      "loss": 0.4418,
+      "step": 214000
+    },
+    {
+      "epoch": 13.11,
+      "eval_loss": 0.8018962144851685,
+      "eval_runtime": 1.2527,
+      "eval_samples_per_second": 798.295,
+      "eval_steps_per_second": 12.773,
+      "step": 214000
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.193199019335709e-05,
+      "loss": 0.4418,
+      "step": 214500
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.105484770287477e-05,
+      "loss": 0.4417,
+      "step": 215000
+    },
+    {
+      "epoch": 13.17,
+      "eval_loss": 0.7995709180831909,
+      "eval_runtime": 1.3104,
+      "eval_samples_per_second": 763.154,
+      "eval_steps_per_second": 12.21,
+      "step": 215000
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.018925102841773e-05,
+      "loss": 0.4416,
+      "step": 215500
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 3.9335238033988484e-05,
+      "loss": 0.4416,
+      "step": 216000
+    },
+    {
+      "epoch": 13.24,
+      "eval_loss": 0.8006547689437866,
+      "eval_runtime": 1.3204,
+      "eval_samples_per_second": 757.326,
+      "eval_steps_per_second": 12.117,
+      "step": 216000
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 3.8492846076882115e-05,
+      "loss": 0.4414,
+      "step": 216500
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 3.766211200605186e-05,
+      "loss": 0.4414,
+      "step": 217000
+    },
+    {
+      "epoch": 13.3,
+      "eval_loss": 0.8029482960700989,
+      "eval_runtime": 1.3397,
+      "eval_samples_per_second": 746.459,
+      "eval_steps_per_second": 11.943,
+      "step": 217000
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 3.684307216049706e-05,
+      "loss": 0.4415,
+      "step": 217500
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 3.6035762367673984e-05,
+      "loss": 0.4415,
+      "step": 218000
+    },
+    {
+      "epoch": 13.36,
+      "eval_loss": 0.7989561557769775,
+      "eval_runtime": 1.2937,
+      "eval_samples_per_second": 773.007,
+      "eval_steps_per_second": 12.368,
+      "step": 218000
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 3.52402179419282e-05,
+      "loss": 0.4414,
+      "step": 218500
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 3.4456473682950194e-05,
+      "loss": 0.4413,
+      "step": 219000
+    },
+    {
+      "epoch": 13.42,
+      "eval_loss": 0.7997331619262695,
+      "eval_runtime": 1.2967,
+      "eval_samples_per_second": 771.217,
+      "eval_steps_per_second": 12.339,
+      "step": 219000
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 3.3684563874252695e-05,
+      "loss": 0.4412,
+      "step": 219500
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 3.2924522281671496e-05,
+      "loss": 0.4413,
+      "step": 220000
+    },
+    {
+      "epoch": 13.48,
+      "eval_loss": 0.7997393608093262,
+      "eval_runtime": 1.3064,
+      "eval_samples_per_second": 765.469,
+      "eval_steps_per_second": 12.247,
+      "step": 220000
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 3.2176382151888054e-05,
+      "loss": 0.4414,
+      "step": 220500
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 3.1440176210975204e-05,
+      "loss": 0.4412,
+      "step": 221000
+    },
+    {
+      "epoch": 13.54,
+      "eval_loss": 0.7996479272842407,
+      "eval_runtime": 1.3112,
+      "eval_samples_per_second": 762.651,
+      "eval_steps_per_second": 12.202,
+      "step": 221000
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 3.071593666296585e-05,
+      "loss": 0.4411,
+      "step": 221500
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 3.000369518844396e-05,
+      "loss": 0.4411,
+      "step": 222000
+    },
+    {
+      "epoch": 13.6,
+      "eval_loss": 0.8002747297286987,
+      "eval_runtime": 1.2963,
+      "eval_samples_per_second": 771.412,
+      "eval_steps_per_second": 12.343,
+      "step": 222000
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.9303482943159077e-05,
+      "loss": 0.4411,
+      "step": 222500
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.861533055666306e-05,
+      "loss": 0.4411,
+      "step": 223000
+    },
+    {
+      "epoch": 13.66,
+      "eval_loss": 0.7992754578590393,
+      "eval_runtime": 1.2911,
+      "eval_samples_per_second": 774.504,
+      "eval_steps_per_second": 12.392,
+      "step": 223000
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.793926813097066e-05,
+      "loss": 0.4411,
+      "step": 223500
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.7275325239242546e-05,
+      "loss": 0.4411,
+      "step": 224000
+    },
+    {
+      "epoch": 13.73,
+      "eval_loss": 0.800546407699585,
+      "eval_runtime": 6.2685,
+      "eval_samples_per_second": 159.527,
+      "eval_steps_per_second": 2.552,
+      "step": 224000
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.6623530924491626e-05,
+      "loss": 0.4409,
+      "step": 224500
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.5983913698312782e-05,
+      "loss": 0.4409,
+      "step": 225000
+    },
+    {
+      "epoch": 13.79,
+      "eval_loss": 0.8013263940811157,
+      "eval_runtime": 1.3128,
+      "eval_samples_per_second": 761.734,
+      "eval_steps_per_second": 12.188,
+      "step": 225000
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.5356501539635512e-05,
+      "loss": 0.441,
+      "step": 225500
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.4741321893500244e-05,
+      "loss": 0.4409,
+      "step": 226000
+    },
+    {
+      "epoch": 13.85,
+      "eval_loss": 0.8015850782394409,
+      "eval_runtime": 1.3378,
+      "eval_samples_per_second": 747.504,
+      "eval_steps_per_second": 11.96,
+      "step": 226000
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.4138401669857587e-05,
+      "loss": 0.4408,
+      "step": 226500
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.3547767242391212e-05,
+      "loss": 0.4409,
+      "step": 227000
+    },
+    {
+      "epoch": 13.91,
+      "eval_loss": 0.7994450926780701,
+      "eval_runtime": 1.2846,
+      "eval_samples_per_second": 778.448,
+      "eval_steps_per_second": 12.455,
+      "step": 227000
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.2969444447364498e-05,
+      "loss": 0.4409,
+      "step": 227500
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.240345858248992e-05,
+      "loss": 0.4408,
+      "step": 228000
+    },
+    {
+      "epoch": 13.97,
+      "eval_loss": 0.8022862672805786,
+      "eval_runtime": 1.3268,
+      "eval_samples_per_second": 753.683,
+      "eval_steps_per_second": 12.059,
+      "step": 228000
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 2.184983440582284e-05,
+      "loss": 0.4408,
+      "step": 228500
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.1308596134678134e-05,
+      "loss": 0.4407,
+      "step": 229000
+    },
+    {
+      "epoch": 14.03,
+      "eval_loss": 0.8013246059417725,
+      "eval_runtime": 1.2694,
+      "eval_samples_per_second": 787.747,
+      "eval_steps_per_second": 12.604,
+      "step": 229000
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 2.0779767444571236e-05,
+      "loss": 0.4406,
+      "step": 229500
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 2.0263371468182175e-05,
+      "loss": 0.4406,
+      "step": 230000
+    },
+    {
+      "epoch": 14.09,
+      "eval_loss": 0.8037849068641663,
+      "eval_runtime": 1.356,
+      "eval_samples_per_second": 737.479,
+      "eval_steps_per_second": 11.8,
+      "step": 230000
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.975943079434381e-05,
+      "loss": 0.4406,
+      "step": 230500
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 1.9267967467053834e-05,
+      "loss": 0.4408,
+      "step": 231000
+    },
+    {
+      "epoch": 14.15,
+      "eval_loss": 0.799385130405426,
+      "eval_runtime": 1.2054,
+      "eval_samples_per_second": 829.62,
+      "eval_steps_per_second": 13.274,
+      "step": 231000
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.878900298451024e-05,
+      "loss": 0.4406,
+      "step": 231500
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.83225582981712e-05,
+      "loss": 0.4406,
+      "step": 232000
+    },
+    {
+      "epoch": 14.22,
+      "eval_loss": 0.800686776638031,
+      "eval_runtime": 1.3097,
+      "eval_samples_per_second": 763.545,
+      "eval_steps_per_second": 12.217,
+      "step": 232000
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 1.7868653811838307e-05,
+      "loss": 0.4406,
+      "step": 232500
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.7427309380764227e-05,
+      "loss": 0.4404,
+      "step": 233000
+    },
+    {
+      "epoch": 14.28,
+      "eval_loss": 0.8005698323249817,
+      "eval_runtime": 1.2953,
+      "eval_samples_per_second": 772.049,
+      "eval_steps_per_second": 12.353,
+      "step": 233000
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 1.6998544310784175e-05,
+      "loss": 0.4406,
+      "step": 233500
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.6582377357471285e-05,
+      "loss": 0.4403,
+      "step": 234000
+    },
+    {
+      "epoch": 14.34,
+      "eval_loss": 0.7986977696418762,
+      "eval_runtime": 1.3526,
+      "eval_samples_per_second": 739.295,
+      "eval_steps_per_second": 11.829,
+      "step": 234000
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.617882672531633e-05,
+      "loss": 0.4405,
+      "step": 234500
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 1.578791006693124e-05,
+      "loss": 0.4405,
+      "step": 235000
+    },
+    {
+      "epoch": 14.4,
+      "eval_loss": 0.8009754419326782,
+      "eval_runtime": 1.2995,
+      "eval_samples_per_second": 769.543,
+      "eval_steps_per_second": 12.313,
+      "step": 235000
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.5409644482277073e-05,
+      "loss": 0.4403,
+      "step": 235500
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.504404651791591e-05,
+      "loss": 0.4404,
+      "step": 236000
+    },
+    {
+      "epoch": 14.46,
+      "eval_loss": 0.798170268535614,
+      "eval_runtime": 1.2848,
+      "eval_samples_per_second": 778.322,
+      "eval_steps_per_second": 12.453,
+      "step": 236000
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.4691132166287069e-05,
+      "loss": 0.4403,
+      "step": 236500
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 1.4350916865007609e-05,
+      "loss": 0.4404,
+      "step": 237000
+    },
+    {
+      "epoch": 14.52,
+      "eval_loss": 0.7984534502029419,
+      "eval_runtime": 1.2843,
+      "eval_samples_per_second": 778.661,
+      "eval_steps_per_second": 12.459,
+      "step": 237000
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.4023415496196915e-05,
+      "loss": 0.4402,
+      "step": 237500
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 1.3708642385825806e-05,
+      "loss": 0.4403,
+      "step": 238000
+    },
+    {
+      "epoch": 14.58,
+      "eval_loss": 0.8015879392623901,
+      "eval_runtime": 1.3525,
+      "eval_samples_per_second": 739.353,
+      "eval_steps_per_second": 11.83,
+      "step": 238000
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.3406611303089918e-05,
+      "loss": 0.4404,
+      "step": 238500
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.311733545980718e-05,
+      "loss": 0.4402,
+      "step": 239000
+    },
+    {
+      "epoch": 14.64,
+      "eval_loss": 0.8024547696113586,
+      "eval_runtime": 1.3287,
+      "eval_samples_per_second": 752.597,
+      "eval_steps_per_second": 12.042,
+      "step": 239000
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.284082750984018e-05,
+      "loss": 0.4401,
+      "step": 239500
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.2577099548542319e-05,
+      "loss": 0.4402,
+      "step": 240000
+    },
+    {
+      "epoch": 14.71,
+      "eval_loss": 0.8019887804985046,
+      "eval_runtime": 1.3351,
+      "eval_samples_per_second": 748.984,
+      "eval_steps_per_second": 11.984,
+      "step": 240000
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.232616311222906e-05,
+      "loss": 0.4402,
+      "step": 240500
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.2088029177672978e-05,
+      "loss": 0.4401,
+      "step": 241000
+    },
+    {
+      "epoch": 14.77,
+      "eval_loss": 0.8008501529693604,
+      "eval_runtime": 1.2829,
+      "eval_samples_per_second": 779.48,
+      "eval_steps_per_second": 12.472,
+      "step": 241000
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 1.1862708161623889e-05,
+      "loss": 0.4402,
+      "step": 241500
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 1.165020992035296e-05,
+      "loss": 0.4401,
+      "step": 242000
+    },
+    {
+      "epoch": 14.83,
+      "eval_loss": 0.8015211224555969,
+      "eval_runtime": 1.2647,
+      "eval_samples_per_second": 790.696,
+      "eval_steps_per_second": 12.651,
+      "step": 242000
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.1450543749221632e-05,
+      "loss": 0.4401,
+      "step": 242500
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 1.126371838227509e-05,
+      "loss": 0.4401,
+      "step": 243000
+    },
+    {
+      "epoch": 14.89,
+      "eval_loss": 0.8009896278381348,
+      "eval_runtime": 1.3003,
+      "eval_samples_per_second": 769.043,
+      "eval_steps_per_second": 12.305,
+      "step": 243000
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 1.1089741991860081e-05,
+      "loss": 0.44,
+      "step": 243500
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.0928622188267536e-05,
+      "loss": 0.44,
+      "step": 244000
+    },
+    {
+      "epoch": 14.95,
+      "eval_loss": 0.799572765827179,
+      "eval_runtime": 1.3134,
+      "eval_samples_per_second": 761.387,
+      "eval_steps_per_second": 12.182,
+      "step": 244000
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 1.0780366019399665e-05,
+      "loss": 0.4401,
+      "step": 244500
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 1.0644979970461512e-05,
+      "loss": 0.4402,
+      "step": 245000
+    },
+    {
+      "epoch": 15.01,
+      "eval_loss": 0.8014461398124695,
+      "eval_runtime": 1.3153,
+      "eval_samples_per_second": 760.258,
+      "eval_steps_per_second": 12.164,
+      "step": 245000
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 1.0522469963677483e-05,
+      "loss": 0.4401,
+      "step": 245500
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 1.0412841358032126e-05,
+      "loss": 0.44,
+      "step": 246000
+    },
+    {
+      "epoch": 15.07,
+      "eval_loss": 0.800745964050293,
+      "eval_runtime": 1.3042,
+      "eval_samples_per_second": 766.776,
+      "eval_steps_per_second": 12.268,
+      "step": 246000
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 1.0316098949035847e-05,
+      "loss": 0.44,
+      "step": 246500
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 1.0232246968514984e-05,
+      "loss": 0.44,
+      "step": 247000
+    },
+    {
+      "epoch": 15.13,
+      "eval_loss": 0.7984172701835632,
+      "eval_runtime": 1.3184,
+      "eval_samples_per_second": 758.509,
+      "eval_steps_per_second": 12.136,
+      "step": 247000
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 1.0161289084426815e-05,
+      "loss": 0.44,
+      "step": 247500
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 1.0103228400699063e-05,
+      "loss": 0.44,
+      "step": 248000
+    },
+    {
+      "epoch": 15.2,
+      "eval_loss": 0.8008962273597717,
+      "eval_runtime": 1.3201,
+      "eval_samples_per_second": 757.502,
+      "eval_steps_per_second": 12.12,
+      "step": 248000
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 1.0058067457094136e-05,
+      "loss": 0.44,
+      "step": 248500
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 1.0025808229097982e-05,
+      "loss": 0.4399,
+      "step": 249000
+    },
+    {
+      "epoch": 15.26,
+      "eval_loss": 0.8005866408348083,
+      "eval_runtime": 1.3752,
+      "eval_samples_per_second": 727.163,
+      "eval_steps_per_second": 11.635,
+      "step": 249000
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 1.0006452127833747e-05,
+      "loss": 0.4399,
+      "step": 249500
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 1e-05,
+      "loss": 0.4399,
+      "step": 250000
+    },
+    {
+      "epoch": 15.32,
+      "eval_loss": 0.8015783429145813,
+      "eval_runtime": 1.3631,
+      "eval_samples_per_second": 733.637,
+      "eval_steps_per_second": 11.738,
+      "step": 250000
+    },
+    {
+      "epoch": 15.32,
+      "step": 250000,
+      "total_flos": 4.004049914090876e+21,
+      "train_loss": 0.08823756774902344,
+      "train_runtime": 15654.7731,
+      "train_samples_per_second": 16352.84,
+      "train_steps_per_second": 15.97
+    }
+  ],
+  "max_steps": 250000,
+  "num_train_epochs": 16,
+  "total_flos": 4.004049914090876e+21,
+  "trial_name": null,
+  "trial_params": null
+}