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": 11.152748037116346,
+  "global_step": 250000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 0.8845,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 0.689,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 0.6793121099472046,
+      "eval_runtime": 2.3939,
+      "eval_samples_per_second": 959.513,
+      "eval_steps_per_second": 15.038,
+      "step": 1000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 0.6805,
+      "step": 1500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 0.6802,
+      "step": 2000
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 0.6786620616912842,
+      "eval_runtime": 2.1637,
+      "eval_samples_per_second": 1061.605,
+      "eval_steps_per_second": 16.638,
+      "step": 2000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 0.6799,
+      "step": 2500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 0.6795,
+      "step": 3000
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 0.6788004040718079,
+      "eval_runtime": 2.1379,
+      "eval_samples_per_second": 1074.407,
+      "eval_steps_per_second": 16.839,
+      "step": 3000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000168,
+      "loss": 0.6791,
+      "step": 3500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019199999999999998,
+      "loss": 0.679,
+      "step": 4000
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 0.6781774163246155,
+      "eval_runtime": 2.1652,
+      "eval_samples_per_second": 1060.868,
+      "eval_steps_per_second": 16.627,
+      "step": 4000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 0.6788,
+      "step": 4500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 0.6787,
+      "step": 5000
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 0.6782493591308594,
+      "eval_runtime": 2.1305,
+      "eval_samples_per_second": 1078.133,
+      "eval_steps_per_second": 16.897,
+      "step": 5000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 0.6786,
+      "step": 5500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 0.6786,
+      "step": 6000
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 0.6780672073364258,
+      "eval_runtime": 2.1865,
+      "eval_samples_per_second": 1050.561,
+      "eval_steps_per_second": 16.465,
+      "step": 6000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000312,
+      "loss": 0.6785,
+      "step": 6500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000336,
+      "loss": 0.6784,
+      "step": 7000
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 0.6781116724014282,
+      "eval_runtime": 2.1436,
+      "eval_samples_per_second": 1071.581,
+      "eval_steps_per_second": 16.794,
+      "step": 7000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 0.6783,
+      "step": 7500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00038399999999999996,
+      "loss": 0.6783,
+      "step": 8000
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 0.6780590415000916,
+      "eval_runtime": 2.0716,
+      "eval_samples_per_second": 1108.795,
+      "eval_steps_per_second": 17.378,
+      "step": 8000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000408,
+      "loss": 0.6783,
+      "step": 8500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00043199999999999993,
+      "loss": 0.6781,
+      "step": 9000
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 0.677251398563385,
+      "eval_runtime": 2.0351,
+      "eval_samples_per_second": 1128.706,
+      "eval_steps_per_second": 17.69,
+      "step": 9000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00045599999999999997,
+      "loss": 0.6776,
+      "step": 9500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00047999999999999996,
+      "loss": 0.6775,
+      "step": 10000
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 0.6778020262718201,
+      "eval_runtime": 2.2347,
+      "eval_samples_per_second": 1027.899,
+      "eval_steps_per_second": 16.11,
+      "step": 10000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005039999999999999,
+      "loss": 0.6775,
+      "step": 10500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005279999999999999,
+      "loss": 0.6775,
+      "step": 11000
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 0.6769479513168335,
+      "eval_runtime": 2.177,
+      "eval_samples_per_second": 1055.099,
+      "eval_steps_per_second": 16.536,
+      "step": 11000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000552,
+      "loss": 0.6773,
+      "step": 11500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0005759999999999999,
+      "loss": 0.6773,
+      "step": 12000
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 0.6773238182067871,
+      "eval_runtime": 2.1281,
+      "eval_samples_per_second": 1079.366,
+      "eval_steps_per_second": 16.916,
+      "step": 12000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0006,
+      "loss": 0.6773,
+      "step": 12500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0005999935478721662,
+      "loss": 0.6774,
+      "step": 13000
+    },
+    {
+      "epoch": 0.58,
+      "eval_loss": 0.677127480506897,
+      "eval_runtime": 2.1773,
+      "eval_samples_per_second": 1054.986,
+      "eval_steps_per_second": 16.534,
+      "step": 13000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000599974191770902,
+      "loss": 0.6773,
+      "step": 13500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0005999419325429058,
+      "loss": 0.6773,
+      "step": 14000
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": 0.6771531105041504,
+      "eval_runtime": 2.1173,
+      "eval_samples_per_second": 1084.86,
+      "eval_steps_per_second": 17.003,
+      "step": 14000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005998967715993009,
+      "loss": 0.6773,
+      "step": 14500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005998387109155732,
+      "loss": 0.6773,
+      "step": 15000
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 0.6771678924560547,
+      "eval_runtime": 2.1919,
+      "eval_samples_per_second": 1047.963,
+      "eval_steps_per_second": 16.424,
+      "step": 15000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000599767753031485,
+      "loss": 0.6773,
+      "step": 15500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0005996839010509641,
+      "loss": 0.6772,
+      "step": 16000
+    },
+    {
+      "epoch": 0.71,
+      "eval_loss": 0.6776318550109863,
+      "eval_runtime": 2.199,
+      "eval_samples_per_second": 1044.559,
+      "eval_steps_per_second": 16.371,
+      "step": 16000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0005995871586419678,
+      "loss": 0.6773,
+      "step": 16500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0005994775300363225,
+      "loss": 0.6773,
+      "step": 17000
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 0.676984429359436,
+      "eval_runtime": 2.1946,
+      "eval_samples_per_second": 1046.652,
+      "eval_steps_per_second": 16.404,
+      "step": 17000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0005993550200295384,
+      "loss": 0.6772,
+      "step": 17500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0005992196339806002,
+      "loss": 0.6772,
+      "step": 18000
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 0.6774880290031433,
+      "eval_runtime": 2.1027,
+      "eval_samples_per_second": 1092.415,
+      "eval_steps_per_second": 17.121,
+      "step": 18000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0005990713778117324,
+      "loss": 0.6773,
+      "step": 18500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0005989102580081398,
+      "loss": 0.6772,
+      "step": 19000
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.676984965801239,
+      "eval_runtime": 2.1686,
+      "eval_samples_per_second": 1059.223,
+      "eval_steps_per_second": 16.601,
+      "step": 19000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0005987362816177249,
+      "loss": 0.6773,
+      "step": 19500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0005985494562507783,
+      "loss": 0.6774,
+      "step": 20000
+    },
+    {
+      "epoch": 0.89,
+      "eval_loss": 0.6769698262214661,
+      "eval_runtime": 2.1456,
+      "eval_samples_per_second": 1070.586,
+      "eval_steps_per_second": 16.779,
+      "step": 20000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000598349790079647,
+      "loss": 0.6773,
+      "step": 20500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000598137291838376,
+      "loss": 0.6772,
+      "step": 21000
+    },
+    {
+      "epoch": 0.94,
+      "eval_loss": 0.6761835217475891,
+      "eval_runtime": 2.1367,
+      "eval_samples_per_second": 1075.027,
+      "eval_steps_per_second": 16.848,
+      "step": 21000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000597911970822327,
+      "loss": 0.6772,
+      "step": 21500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000597673836887771,
+      "loss": 0.6773,
+      "step": 22000
+    },
+    {
+      "epoch": 0.98,
+      "eval_loss": 0.6774830222129822,
+      "eval_runtime": 2.1107,
+      "eval_samples_per_second": 1088.282,
+      "eval_steps_per_second": 17.056,
+      "step": 22000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005974229004514577,
+      "loss": 0.6773,
+      "step": 22500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005971591724901598,
+      "loss": 0.6773,
+      "step": 23000
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 0.6764113903045654,
+      "eval_runtime": 2.206,
+      "eval_samples_per_second": 1041.236,
+      "eval_steps_per_second": 16.319,
+      "step": 23000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0005968826645401927,
+      "loss": 0.6772,
+      "step": 23500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00059659338869691,
+      "loss": 0.6772,
+      "step": 24000
+    },
+    {
+      "epoch": 1.07,
+      "eval_loss": 0.6768463253974915,
+      "eval_runtime": 2.2173,
+      "eval_samples_per_second": 1035.939,
+      "eval_steps_per_second": 16.236,
+      "step": 24000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0005962913576141742,
+      "loss": 0.6773,
+      "step": 24500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000595976584503803,
+      "loss": 0.6772,
+      "step": 25000
+    },
+    {
+      "epoch": 1.12,
+      "eval_loss": 0.6769193410873413,
+      "eval_runtime": 2.2077,
+      "eval_samples_per_second": 1040.465,
+      "eval_steps_per_second": 16.307,
+      "step": 25000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0005956490831349923,
+      "loss": 0.6773,
+      "step": 25500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0005953088678337129,
+      "loss": 0.6772,
+      "step": 26000
+    },
+    {
+      "epoch": 1.16,
+      "eval_loss": 0.6775402426719666,
+      "eval_runtime": 2.2094,
+      "eval_samples_per_second": 1039.671,
+      "eval_steps_per_second": 16.294,
+      "step": 26000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0005949559534820841,
+      "loss": 0.6772,
+      "step": 26500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0005945903555177229,
+      "loss": 0.6772,
+      "step": 27000
+    },
+    {
+      "epoch": 1.2,
+      "eval_loss": 0.6775718331336975,
+      "eval_runtime": 2.1372,
+      "eval_samples_per_second": 1074.76,
+      "eval_steps_per_second": 16.844,
+      "step": 27000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0005942120899330687,
+      "loss": 0.6772,
+      "step": 27500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0005938211732746836,
+      "loss": 0.6772,
+      "step": 28000
+    },
+    {
+      "epoch": 1.25,
+      "eval_loss": 0.6771659255027771,
+      "eval_runtime": 2.191,
+      "eval_samples_per_second": 1048.378,
+      "eval_steps_per_second": 16.431,
+      "step": 28000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0005934176226425286,
+      "loss": 0.6772,
+      "step": 28500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0005930014556892158,
+      "loss": 0.6772,
+      "step": 29000
+    },
+    {
+      "epoch": 1.29,
+      "eval_loss": 0.6768732666969299,
+      "eval_runtime": 2.2912,
+      "eval_samples_per_second": 1002.553,
+      "eval_steps_per_second": 15.713,
+      "step": 29000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0005925726906192357,
+      "loss": 0.6772,
+      "step": 29500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0005921313461881617,
+      "loss": 0.6773,
+      "step": 30000
+    },
+    {
+      "epoch": 1.34,
+      "eval_loss": 0.6772350072860718,
+      "eval_runtime": 2.2465,
+      "eval_samples_per_second": 1022.498,
+      "eval_steps_per_second": 16.025,
+      "step": 30000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0005916774417018287,
+      "loss": 0.6772,
+      "step": 30500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005912109970154897,
+      "loss": 0.6772,
+      "step": 31000
+    },
+    {
+      "epoch": 1.38,
+      "eval_loss": 0.6777171492576599,
+      "eval_runtime": 2.1569,
+      "eval_samples_per_second": 1064.935,
+      "eval_steps_per_second": 16.69,
+      "step": 31000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005907320325329461,
+      "loss": 0.6772,
+      "step": 31500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005902405692056561,
+      "loss": 0.6772,
+      "step": 32000
+    },
+    {
+      "epoch": 1.43,
+      "eval_loss": 0.676947832107544,
+      "eval_runtime": 2.3119,
+      "eval_samples_per_second": 993.573,
+      "eval_steps_per_second": 15.572,
+      "step": 32000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005897366285318178,
+      "loss": 0.6772,
+      "step": 32500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005892202325554288,
+      "loss": 0.6773,
+      "step": 33000
+    },
+    {
+      "epoch": 1.47,
+      "eval_loss": 0.6766595840454102,
+      "eval_runtime": 2.1462,
+      "eval_samples_per_second": 1070.248,
+      "eval_steps_per_second": 16.774,
+      "step": 33000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005886914038653217,
+      "loss": 0.6772,
+      "step": 33500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005881501655941771,
+      "loss": 0.677,
+      "step": 34000
+    },
+    {
+      "epoch": 1.52,
+      "eval_loss": 0.6765649914741516,
+      "eval_runtime": 2.1369,
+      "eval_samples_per_second": 1074.929,
+      "eval_steps_per_second": 16.847,
+      "step": 34000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00058759654141751,
+      "loss": 0.6768,
+      "step": 34500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005870305555526355,
+      "loss": 0.6765,
+      "step": 35000
+    },
+    {
+      "epoch": 1.56,
+      "eval_loss": 0.676632821559906,
+      "eval_runtime": 2.2745,
+      "eval_samples_per_second": 1009.888,
+      "eval_steps_per_second": 15.828,
+      "step": 35000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005864522327576088,
+      "loss": 0.6764,
+      "step": 35500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005858615983301424,
+      "loss": 0.6763,
+      "step": 36000
+    },
+    {
+      "epoch": 1.61,
+      "eval_loss": 0.6765820980072021,
+      "eval_runtime": 2.1766,
+      "eval_samples_per_second": 1055.304,
+      "eval_steps_per_second": 16.539,
+      "step": 36000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005852586781064997,
+      "loss": 0.6763,
+      "step": 36500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0005846434984603645,
+      "loss": 0.6764,
+      "step": 37000
+    },
+    {
+      "epoch": 1.65,
+      "eval_loss": 0.6758345365524292,
+      "eval_runtime": 2.237,
+      "eval_samples_per_second": 1026.821,
+      "eval_steps_per_second": 16.093,
+      "step": 37000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0005840160863016872,
+      "loss": 0.6763,
+      "step": 37500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0005833764690755083,
+      "loss": 0.6764,
+      "step": 38000
+    },
+    {
+      "epoch": 1.7,
+      "eval_loss": 0.6762025952339172,
+      "eval_runtime": 2.1722,
+      "eval_samples_per_second": 1057.45,
+      "eval_steps_per_second": 16.573,
+      "step": 38000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0005827246747607574,
+      "loss": 0.6765,
+      "step": 38500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0005820607318690293,
+      "loss": 0.6758,
+      "step": 39000
+    },
+    {
+      "epoch": 1.74,
+      "eval_loss": 0.6771443486213684,
+      "eval_runtime": 2.2147,
+      "eval_samples_per_second": 1037.153,
+      "eval_steps_per_second": 16.255,
+      "step": 39000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0005813846694433368,
+      "loss": 0.6773,
+      "step": 39500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0005806965170568409,
+      "loss": 0.6772,
+      "step": 40000
+    },
+    {
+      "epoch": 1.78,
+      "eval_loss": 0.6770426630973816,
+      "eval_runtime": 2.1711,
+      "eval_samples_per_second": 1057.979,
+      "eval_steps_per_second": 16.581,
+      "step": 40000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0005799963048115559,
+      "loss": 0.6757,
+      "step": 40500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0005792840633370341,
+      "loss": 0.6575,
+      "step": 41000
+    },
+    {
+      "epoch": 1.83,
+      "eval_loss": 0.6465174555778503,
+      "eval_runtime": 2.2243,
+      "eval_samples_per_second": 1032.694,
+      "eval_steps_per_second": 16.185,
+      "step": 41000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0005785598237890247,
+      "loss": 0.6461,
+      "step": 41500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0005778236178481119,
+      "loss": 0.6373,
+      "step": 42000
+    },
+    {
+      "epoch": 1.87,
+      "eval_loss": 0.631793737411499,
+      "eval_runtime": 2.22,
+      "eval_samples_per_second": 1034.694,
+      "eval_steps_per_second": 16.216,
+      "step": 42000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0005770754777183285,
+      "loss": 0.6309,
+      "step": 42500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0005763154361257473,
+      "loss": 0.6257,
+      "step": 43000
+    },
+    {
+      "epoch": 1.92,
+      "eval_loss": 0.6184197664260864,
+      "eval_runtime": 2.2848,
+      "eval_samples_per_second": 1005.32,
+      "eval_steps_per_second": 15.756,
+      "step": 43000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0005755435263170498,
+      "loss": 0.623,
+      "step": 43500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0005747597820580717,
+      "loss": 0.621,
+      "step": 44000
+    },
+    {
+      "epoch": 1.96,
+      "eval_loss": 0.6136025786399841,
+      "eval_runtime": 2.2706,
+      "eval_samples_per_second": 1011.621,
+      "eval_steps_per_second": 15.855,
+      "step": 44000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.000573964237632326,
+      "loss": 0.6195,
+      "step": 44500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0005731569278395029,
+      "loss": 0.6183,
+      "step": 45000
+    },
+    {
+      "epoch": 2.01,
+      "eval_loss": 0.6127315163612366,
+      "eval_runtime": 2.1317,
+      "eval_samples_per_second": 1077.566,
+      "eval_steps_per_second": 16.888,
+      "step": 45000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0005723378879939481,
+      "loss": 0.6172,
+      "step": 45500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0005715071539231178,
+      "loss": 0.6165,
+      "step": 46000
+    },
+    {
+      "epoch": 2.05,
+      "eval_loss": 0.6103396415710449,
+      "eval_runtime": 2.1687,
+      "eval_samples_per_second": 1059.145,
+      "eval_steps_per_second": 16.6,
+      "step": 46000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0005706647619660116,
+      "loss": 0.6155,
+      "step": 46500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0005698107489715823,
+      "loss": 0.612,
+      "step": 47000
+    },
+    {
+      "epoch": 2.1,
+      "eval_loss": 0.6013069748878479,
+      "eval_runtime": 2.2612,
+      "eval_samples_per_second": 1015.821,
+      "eval_steps_per_second": 15.921,
+      "step": 47000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0005689451522971252,
+      "loss": 0.6068,
+      "step": 47500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0005680680098066429,
+      "loss": 0.6037,
+      "step": 48000
+    },
+    {
+      "epoch": 2.14,
+      "eval_loss": 0.5942632555961609,
+      "eval_runtime": 2.2238,
+      "eval_samples_per_second": 1032.912,
+      "eval_steps_per_second": 16.188,
+      "step": 48000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0005671793598691895,
+      "loss": 0.6015,
+      "step": 48500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0005662792413571921,
+      "loss": 0.6,
+      "step": 49000
+    },
+    {
+      "epoch": 2.19,
+      "eval_loss": 0.5915176868438721,
+      "eval_runtime": 2.2389,
+      "eval_samples_per_second": 1025.951,
+      "eval_steps_per_second": 16.079,
+      "step": 49000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0005653676936447504,
+      "loss": 0.5986,
+      "step": 49500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0005644447566059142,
+      "loss": 0.5973,
+      "step": 50000
+    },
+    {
+      "epoch": 2.23,
+      "eval_loss": 0.5880685448646545,
+      "eval_runtime": 2.2056,
+      "eval_samples_per_second": 1041.459,
+      "eval_steps_per_second": 16.322,
+      "step": 50000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0005635104706129397,
+      "loss": 0.5958,
+      "step": 50500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0005625648765345228,
+      "loss": 0.5924,
+      "step": 51000
+    },
+    {
+      "epoch": 2.28,
+      "eval_loss": 0.5799189209938049,
+      "eval_runtime": 2.2525,
+      "eval_samples_per_second": 1019.737,
+      "eval_steps_per_second": 15.982,
+      "step": 51000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0005616080157340118,
+      "loss": 0.5872,
+      "step": 51500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0005606399300675978,
+      "loss": 0.5817,
+      "step": 52000
+    },
+    {
+      "epoch": 2.32,
+      "eval_loss": 0.5670354962348938,
+      "eval_runtime": 2.3437,
+      "eval_samples_per_second": 980.093,
+      "eval_steps_per_second": 15.361,
+      "step": 52000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0005596606618824843,
+      "loss": 0.5766,
+      "step": 52500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0005586702540150338,
+      "loss": 0.5719,
+      "step": 53000
+    },
+    {
+      "epoch": 2.36,
+      "eval_loss": 0.5556795597076416,
+      "eval_runtime": 2.1734,
+      "eval_samples_per_second": 1056.846,
+      "eval_steps_per_second": 16.564,
+      "step": 53000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.000557668749788895,
+      "loss": 0.5681,
+      "step": 53500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0005566561930131072,
+      "loss": 0.5651,
+      "step": 54000
+    },
+    {
+      "epoch": 2.41,
+      "eval_loss": 0.5476705431938171,
+      "eval_runtime": 2.2374,
+      "eval_samples_per_second": 1026.619,
+      "eval_steps_per_second": 16.09,
+      "step": 54000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.000555632627980184,
+      "loss": 0.562,
+      "step": 54500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0005545980994641758,
+      "loss": 0.5592,
+      "step": 55000
+    },
+    {
+      "epoch": 2.45,
+      "eval_loss": 0.5407888889312744,
+      "eval_runtime": 2.2932,
+      "eval_samples_per_second": 1001.641,
+      "eval_steps_per_second": 15.698,
+      "step": 55000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0005535526527187115,
+      "loss": 0.5563,
+      "step": 55500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0005524963334750183,
+      "loss": 0.5534,
+      "step": 56000
+    },
+    {
+      "epoch": 2.5,
+      "eval_loss": 0.536229133605957,
+      "eval_runtime": 2.1818,
+      "eval_samples_per_second": 1052.81,
+      "eval_steps_per_second": 16.5,
+      "step": 56000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0005514291879399219,
+      "loss": 0.5496,
+      "step": 56500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.000550351262793825,
+      "loss": 0.5446,
+      "step": 57000
+    },
+    {
+      "epoch": 2.54,
+      "eval_loss": 0.5251114964485168,
+      "eval_runtime": 2.298,
+      "eval_samples_per_second": 999.556,
+      "eval_steps_per_second": 15.666,
+      "step": 57000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0005492626051886659,
+      "loss": 0.5392,
+      "step": 57500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0005481632627458546,
+      "loss": 0.5342,
+      "step": 58000
+    },
+    {
+      "epoch": 2.59,
+      "eval_loss": 0.5129896402359009,
+      "eval_runtime": 2.2194,
+      "eval_samples_per_second": 1034.968,
+      "eval_steps_per_second": 16.221,
+      "step": 58000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0005470532835541911,
+      "loss": 0.529,
+      "step": 58500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.000545932716167761,
+      "loss": 0.5239,
+      "step": 59000
+    },
+    {
+      "epoch": 2.63,
+      "eval_loss": 0.5024312734603882,
+      "eval_runtime": 2.205,
+      "eval_samples_per_second": 1041.743,
+      "eval_steps_per_second": 16.327,
+      "step": 59000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.000544801609603812,
+      "loss": 0.5193,
+      "step": 59500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0005436600133406095,
+      "loss": 0.5147,
+      "step": 60000
+    },
+    {
+      "epoch": 2.68,
+      "eval_loss": 0.49465516209602356,
+      "eval_runtime": 2.2427,
+      "eval_samples_per_second": 1024.23,
+      "eval_steps_per_second": 16.052,
+      "step": 60000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0005425079773152721,
+      "loss": 0.5102,
+      "step": 60500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0005413455519215879,
+      "loss": 0.5061,
+      "step": 61000
+    },
+    {
+      "epoch": 2.72,
+      "eval_loss": 0.4848201274871826,
+      "eval_runtime": 2.1727,
+      "eval_samples_per_second": 1057.234,
+      "eval_steps_per_second": 16.57,
+      "step": 61000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0005401727880078093,
+      "loss": 0.502,
+      "step": 61500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0005389897368744289,
+      "loss": 0.4981,
+      "step": 62000
+    },
+    {
+      "epoch": 2.77,
+      "eval_loss": 0.4745917320251465,
+      "eval_runtime": 2.1553,
+      "eval_samples_per_second": 1065.724,
+      "eval_steps_per_second": 16.703,
+      "step": 62000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0005377964502719361,
+      "loss": 0.4972,
+      "step": 62500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0005365929803985524,
+      "loss": 0.4912,
+      "step": 63000
+    },
+    {
+      "epoch": 2.81,
+      "eval_loss": 0.46808210015296936,
+      "eval_runtime": 2.2693,
+      "eval_samples_per_second": 1012.199,
+      "eval_steps_per_second": 15.864,
+      "step": 63000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0005353793798979489,
+      "loss": 0.4882,
+      "step": 63500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.000534155701856943,
+      "loss": 0.4847,
+      "step": 64000
+    },
+    {
+      "epoch": 2.86,
+      "eval_loss": 0.45994389057159424,
+      "eval_runtime": 2.2357,
+      "eval_samples_per_second": 1027.41,
+      "eval_steps_per_second": 16.102,
+      "step": 64000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0005329219998031763,
+      "loss": 0.4818,
+      "step": 64500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0005316783277027734,
+      "loss": 0.4792,
+      "step": 65000
+    },
+    {
+      "epoch": 2.9,
+      "eval_loss": 0.4537006914615631,
+      "eval_runtime": 2.1952,
+      "eval_samples_per_second": 1046.372,
+      "eval_steps_per_second": 16.399,
+      "step": 65000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0005304247399579808,
+      "loss": 0.4766,
+      "step": 65500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0005291612914047876,
+      "loss": 0.474,
+      "step": 66000
+    },
+    {
+      "epoch": 2.94,
+      "eval_loss": 0.44910311698913574,
+      "eval_runtime": 2.304,
+      "eval_samples_per_second": 996.96,
+      "eval_steps_per_second": 15.625,
+      "step": 66000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0005278880373105263,
+      "loss": 0.4713,
+      "step": 66500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0005266050333714561,
+      "loss": 0.4688,
+      "step": 67000
+    },
+    {
+      "epoch": 2.99,
+      "eval_loss": 0.4437292516231537,
+      "eval_runtime": 2.2404,
+      "eval_samples_per_second": 1025.255,
+      "eval_steps_per_second": 16.068,
+      "step": 67000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0005253123357103253,
+      "loss": 0.4665,
+      "step": 67500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0005240100008739177,
+      "loss": 0.464,
+      "step": 68000
+    },
+    {
+      "epoch": 3.03,
+      "eval_loss": 0.4391787052154541,
+      "eval_runtime": 2.1873,
+      "eval_samples_per_second": 1050.17,
+      "eval_steps_per_second": 16.459,
+      "step": 68000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0005226980858305778,
+      "loss": 0.4616,
+      "step": 68500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0005213766479677197,
+      "loss": 0.4592,
+      "step": 69000
+    },
+    {
+      "epoch": 3.08,
+      "eval_loss": 0.43239352107048035,
+      "eval_runtime": 2.1819,
+      "eval_samples_per_second": 1052.771,
+      "eval_steps_per_second": 16.5,
+      "step": 69000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0005200457450893163,
+      "loss": 0.457,
+      "step": 69500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0005187054354133712,
+      "loss": 0.4547,
+      "step": 70000
+    },
+    {
+      "epoch": 3.12,
+      "eval_loss": 0.4284396469593048,
+      "eval_runtime": 2.1519,
+      "eval_samples_per_second": 1067.405,
+      "eval_steps_per_second": 16.729,
+      "step": 70000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0005173557775693715,
+      "loss": 0.4528,
+      "step": 70500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0005159968305957235,
+      "loss": 0.4507,
+      "step": 71000
+    },
+    {
+      "epoch": 3.17,
+      "eval_loss": 0.42597973346710205,
+      "eval_runtime": 2.1768,
+      "eval_samples_per_second": 1055.242,
+      "eval_steps_per_second": 16.538,
+      "step": 71000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0005146286539371703,
+      "loss": 0.4489,
+      "step": 71500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005132513074421913,
+      "loss": 0.4468,
+      "step": 72000
+    },
+    {
+      "epoch": 3.21,
+      "eval_loss": 0.4192256033420563,
+      "eval_runtime": 2.2165,
+      "eval_samples_per_second": 1036.301,
+      "eval_steps_per_second": 16.242,
+      "step": 72000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005118648513603841,
+      "loss": 0.445,
+      "step": 72500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005104693463398293,
+      "loss": 0.4432,
+      "step": 73000
+    },
+    {
+      "epoch": 3.26,
+      "eval_loss": 0.4161074459552765,
+      "eval_runtime": 2.2735,
+      "eval_samples_per_second": 1010.341,
+      "eval_steps_per_second": 15.835,
+      "step": 73000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005090648534244371,
+      "loss": 0.4415,
+      "step": 73500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005076514340512776,
+      "loss": 0.44,
+      "step": 74000
+    },
+    {
+      "epoch": 3.3,
+      "eval_loss": 0.41528481245040894,
+      "eval_runtime": 2.1341,
+      "eval_samples_per_second": 1076.31,
+      "eval_steps_per_second": 16.869,
+      "step": 74000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005062291500478931,
+      "loss": 0.4386,
+      "step": 74500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005047980636295937,
+      "loss": 0.4367,
+      "step": 75000
+    },
+    {
+      "epoch": 3.35,
+      "eval_loss": 0.4101610779762268,
+      "eval_runtime": 2.2225,
+      "eval_samples_per_second": 1033.52,
+      "eval_steps_per_second": 16.198,
+      "step": 75000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005033582373967348,
+      "loss": 0.4351,
+      "step": 75500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0005019097343319809,
+      "loss": 0.4337,
+      "step": 76000
+    },
+    {
+      "epoch": 3.39,
+      "eval_loss": 0.40619799494743347,
+      "eval_runtime": 2.1631,
+      "eval_samples_per_second": 1061.919,
+      "eval_steps_per_second": 16.643,
+      "step": 76000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0005004526177975481,
+      "loss": 0.4322,
+      "step": 76500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004989869515324342,
+      "loss": 0.4311,
+      "step": 77000
+    },
+    {
+      "epoch": 3.44,
+      "eval_loss": 0.4019022583961487,
+      "eval_runtime": 2.2694,
+      "eval_samples_per_second": 1012.161,
+      "eval_steps_per_second": 15.863,
+      "step": 77000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004975127996496297,
+      "loss": 0.4294,
+      "step": 77500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004960302266333135,
+      "loss": 0.4286,
+      "step": 78000
+    },
+    {
+      "epoch": 3.48,
+      "eval_loss": 0.40067020058631897,
+      "eval_runtime": 2.31,
+      "eval_samples_per_second": 994.37,
+      "eval_steps_per_second": 15.584,
+      "step": 78000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004945392973360323,
+      "loss": 0.427,
+      "step": 78500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004930400769758634,
+      "loss": 0.4259,
+      "step": 79000
+    },
+    {
+      "epoch": 3.52,
+      "eval_loss": 0.3996644616127014,
+      "eval_runtime": 2.1788,
+      "eval_samples_per_second": 1054.255,
+      "eval_steps_per_second": 16.523,
+      "step": 79000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004915326311335622,
+      "loss": 0.425,
+      "step": 79500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004900170257496933,
+      "loss": 0.4239,
+      "step": 80000
+    },
+    {
+      "epoch": 3.57,
+      "eval_loss": 0.3968483507633209,
+      "eval_runtime": 2.2257,
+      "eval_samples_per_second": 1032.02,
+      "eval_steps_per_second": 16.174,
+      "step": 80000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0004884933271217461,
+      "loss": 0.4229,
+      "step": 80500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0004869616019012347,
+      "loss": 0.4218,
+      "step": 81000
+    },
+    {
+      "epoch": 3.61,
+      "eval_loss": 0.3948546051979065,
+      "eval_runtime": 2.2856,
+      "eval_samples_per_second": 1005.007,
+      "eval_steps_per_second": 15.751,
+      "step": 81000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00048542191709078226,
+      "loss": 0.4209,
+      "step": 81500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00048387434004119027,
+      "loss": 0.4201,
+      "step": 82000
+    },
+    {
+      "epoch": 3.66,
+      "eval_loss": 0.39345651865005493,
+      "eval_runtime": 2.2871,
+      "eval_samples_per_second": 1004.335,
+      "eval_steps_per_second": 15.741,
+      "step": 82000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0004823189384484924,
+      "loss": 0.4206,
+      "step": 82500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.000480755780350993,
+      "loss": 0.4182,
+      "step": 83000
+    },
+    {
+      "epoch": 3.7,
+      "eval_loss": 0.39259859919548035,
+      "eval_runtime": 2.2778,
+      "eval_samples_per_second": 1008.432,
+      "eval_steps_per_second": 15.805,
+      "step": 83000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0004791849341262914,
+      "loss": 0.4176,
+      "step": 83500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0004776064684882901,
+      "loss": 0.4168,
+      "step": 84000
+    },
+    {
+      "epoch": 3.75,
+      "eval_loss": 0.38789093494415283,
+      "eval_runtime": 2.2568,
+      "eval_samples_per_second": 1017.832,
+      "eval_steps_per_second": 15.952,
+      "step": 84000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0004760204524841897,
+      "loss": 0.416,
+      "step": 84500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0004744269554914683,
+      "loss": 0.4155,
+      "step": 85000
+    },
+    {
+      "epoch": 3.79,
+      "eval_loss": 0.3885124921798706,
+      "eval_runtime": 2.2513,
+      "eval_samples_per_second": 1020.282,
+      "eval_steps_per_second": 15.99,
+      "step": 85000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00047282604721484643,
+      "loss": 0.4144,
+      "step": 85500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0004712177976832385,
+      "loss": 0.4136,
+      "step": 86000
+    },
+    {
+      "epoch": 3.84,
+      "eval_loss": 0.38439127802848816,
+      "eval_runtime": 2.2526,
+      "eval_samples_per_second": 1019.704,
+      "eval_steps_per_second": 15.981,
+      "step": 86000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0004696022772466888,
+      "loss": 0.413,
+      "step": 86500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00046797955657329487,
+      "loss": 0.4124,
+      "step": 87000
+    },
+    {
+      "epoch": 3.88,
+      "eval_loss": 0.3855268955230713,
+      "eval_runtime": 2.2218,
+      "eval_samples_per_second": 1033.851,
+      "eval_steps_per_second": 16.203,
+      "step": 87000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.000466349706646116,
+      "loss": 0.4118,
+      "step": 87500
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00046471279876006763,
+      "loss": 0.4116,
+      "step": 88000
+    },
+    {
+      "epoch": 3.93,
+      "eval_loss": 0.3830411732196808,
+      "eval_runtime": 2.2947,
+      "eval_samples_per_second": 1001.012,
+      "eval_steps_per_second": 15.688,
+      "step": 88000
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00046306890451880395,
+      "loss": 0.4107,
+      "step": 88500
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0004614180958315844,
+      "loss": 0.4098,
+      "step": 89000
+    },
+    {
+      "epoch": 3.97,
+      "eval_loss": 0.38374021649360657,
+      "eval_runtime": 2.1424,
+      "eval_samples_per_second": 1072.163,
+      "eval_steps_per_second": 16.804,
+      "step": 89000
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00045976044491012884,
+      "loss": 0.4094,
+      "step": 89500
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00045809602426545847,
+      "loss": 0.4087,
+      "step": 90000
+    },
+    {
+      "epoch": 4.01,
+      "eval_loss": 0.3802391290664673,
+      "eval_runtime": 2.1802,
+      "eval_samples_per_second": 1053.586,
+      "eval_steps_per_second": 16.512,
+      "step": 90000
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00045642490670472436,
+      "loss": 0.4084,
+      "step": 90500
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0004547471653280225,
+      "loss": 0.4078,
+      "step": 91000
+    },
+    {
+      "epoch": 4.06,
+      "eval_loss": 0.37994059920310974,
+      "eval_runtime": 2.2046,
+      "eval_samples_per_second": 1041.911,
+      "eval_steps_per_second": 16.329,
+      "step": 91000
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00045306287352519543,
+      "loss": 0.407,
+      "step": 91500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00045137210497262333,
+      "loss": 0.4068,
+      "step": 92000
+    },
+    {
+      "epoch": 4.1,
+      "eval_loss": 0.3793868124485016,
+      "eval_runtime": 2.2708,
+      "eval_samples_per_second": 1011.534,
+      "eval_steps_per_second": 15.853,
+      "step": 92000
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0004496749336299999,
+      "loss": 0.4063,
+      "step": 92500
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0004479714337370977,
+      "loss": 0.4057,
+      "step": 93000
+    },
+    {
+      "epoch": 4.15,
+      "eval_loss": 0.37835967540740967,
+      "eval_runtime": 2.237,
+      "eval_samples_per_second": 1026.834,
+      "eval_steps_per_second": 16.093,
+      "step": 93000
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00044626167981052036,
+      "loss": 0.4052,
+      "step": 93500
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00044454574664044404,
+      "loss": 0.4047,
+      "step": 94000
+    },
+    {
+      "epoch": 4.19,
+      "eval_loss": 0.37884432077407837,
+      "eval_runtime": 2.2678,
+      "eval_samples_per_second": 1012.88,
+      "eval_steps_per_second": 15.874,
+      "step": 94000
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.000442823709287344,
+      "loss": 0.4044,
+      "step": 94500
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0004410956430787129,
+      "loss": 0.4047,
+      "step": 95000
+    },
+    {
+      "epoch": 4.24,
+      "eval_loss": 0.37695789337158203,
+      "eval_runtime": 2.1458,
+      "eval_samples_per_second": 1070.466,
+      "eval_steps_per_second": 16.777,
+      "step": 95000
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0004393616236057647,
+      "loss": 0.4036,
+      "step": 95500
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00043762172672012875,
+      "loss": 0.4029,
+      "step": 96000
+    },
+    {
+      "epoch": 4.28,
+      "eval_loss": 0.37500157952308655,
+      "eval_runtime": 2.2301,
+      "eval_samples_per_second": 1029.999,
+      "eval_steps_per_second": 16.143,
+      "step": 96000
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0004358760285305312,
+      "loss": 0.4025,
+      "step": 96500
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0004341246053994663,
+      "loss": 0.4022,
+      "step": 97000
+    },
+    {
+      "epoch": 4.33,
+      "eval_loss": 0.37471264600753784,
+      "eval_runtime": 2.2149,
+      "eval_samples_per_second": 1037.069,
+      "eval_steps_per_second": 16.254,
+      "step": 97000
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00043236753393985534,
+      "loss": 0.4019,
+      "step": 97500
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0004306048910116964,
+      "loss": 0.4015,
+      "step": 98000
+    },
+    {
+      "epoch": 4.37,
+      "eval_loss": 0.3735538125038147,
+      "eval_runtime": 2.2361,
+      "eval_samples_per_second": 1027.233,
+      "eval_steps_per_second": 16.099,
+      "step": 98000
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0004288367537187012,
+      "loss": 0.4011,
+      "step": 98500
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00042706319940492284,
+      "loss": 0.4007,
+      "step": 99000
+    },
+    {
+      "epoch": 4.42,
+      "eval_loss": 0.3751888573169708,
+      "eval_runtime": 2.2595,
+      "eval_samples_per_second": 1016.613,
+      "eval_steps_per_second": 15.933,
+      "step": 99000
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00042528430565137254,
+      "loss": 0.4003,
+      "step": 99500
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00042350015027262593,
+      "loss": 0.4,
+      "step": 100000
+    },
+    {
+      "epoch": 4.46,
+      "eval_loss": 0.37434616684913635,
+      "eval_runtime": 2.2061,
+      "eval_samples_per_second": 1041.22,
+      "eval_steps_per_second": 16.319,
+      "step": 100000
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00042171081131341917,
+      "loss": 0.3996,
+      "step": 100500
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00041991636704523497,
+      "loss": 0.3995,
+      "step": 101000
+    },
+    {
+      "epoch": 4.51,
+      "eval_loss": 0.37412840127944946,
+      "eval_runtime": 2.2118,
+      "eval_samples_per_second": 1038.514,
+      "eval_steps_per_second": 16.276,
+      "step": 101000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00041811689596287893,
+      "loss": 0.3989,
+      "step": 101500
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0004163124767810454,
+      "loss": 0.3985,
+      "step": 102000
+    },
+    {
+      "epoch": 4.55,
+      "eval_loss": 0.37019699811935425,
+      "eval_runtime": 2.207,
+      "eval_samples_per_second": 1040.78,
+      "eval_steps_per_second": 16.312,
+      "step": 102000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00041450318843087506,
+      "loss": 0.3983,
+      "step": 102500
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00041268911005650166,
+      "loss": 0.3981,
+      "step": 103000
+    },
+    {
+      "epoch": 4.59,
+      "eval_loss": 0.3799527883529663,
+      "eval_runtime": 2.7041,
+      "eval_samples_per_second": 849.448,
+      "eval_steps_per_second": 13.313,
+      "step": 103000
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00041087032101159006,
+      "loss": 0.3978,
+      "step": 103500
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00040904690085586515,
+      "loss": 0.3986,
+      "step": 104000
+    },
+    {
+      "epoch": 4.64,
+      "eval_loss": 0.3734039068222046,
+      "eval_runtime": 2.2569,
+      "eval_samples_per_second": 1017.749,
+      "eval_steps_per_second": 15.951,
+      "step": 104000
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0004072189293516316,
+      "loss": 0.3969,
+      "step": 104500
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0004053864864602847,
+      "loss": 0.3966,
+      "step": 105000
+    },
+    {
+      "epoch": 4.68,
+      "eval_loss": 0.37050846219062805,
+      "eval_runtime": 2.2224,
+      "eval_samples_per_second": 1033.581,
+      "eval_steps_per_second": 16.199,
+      "step": 105000
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00040354965233881297,
+      "loss": 0.3961,
+      "step": 105500
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0004017085073362913,
+      "loss": 0.3957,
+      "step": 106000
+    },
+    {
+      "epoch": 4.73,
+      "eval_loss": 0.36802881956100464,
+      "eval_runtime": 2.2556,
+      "eval_samples_per_second": 1018.364,
+      "eval_steps_per_second": 15.96,
+      "step": 106000
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00039986313199036664,
+      "loss": 0.3956,
+      "step": 106500
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00039801360702373484,
+      "loss": 0.3957,
+      "step": 107000
+    },
+    {
+      "epoch": 4.77,
+      "eval_loss": 0.3662741482257843,
+      "eval_runtime": 2.5144,
+      "eval_samples_per_second": 913.525,
+      "eval_steps_per_second": 14.317,
+      "step": 107000
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00039616001334060954,
+      "loss": 0.3951,
+      "step": 107500
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00039430243202318314,
+      "loss": 0.3948,
+      "step": 108000
+    },
+    {
+      "epoch": 4.82,
+      "eval_loss": 0.36831599473953247,
+      "eval_runtime": 2.2146,
+      "eval_samples_per_second": 1037.206,
+      "eval_steps_per_second": 16.256,
+      "step": 108000
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00039244094432808034,
+      "loss": 0.3945,
+      "step": 108500
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0003905756316828033,
+      "loss": 0.3943,
+      "step": 109000
+    },
+    {
+      "epoch": 4.86,
+      "eval_loss": 0.36974549293518066,
+      "eval_runtime": 2.1759,
+      "eval_samples_per_second": 1055.667,
+      "eval_steps_per_second": 16.545,
+      "step": 109000
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00038870657568216963,
+      "loss": 0.3939,
+      "step": 109500
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00038683385808474416,
+      "loss": 0.3936,
+      "step": 110000
+    },
+    {
+      "epoch": 4.91,
+      "eval_loss": 0.3671797811985016,
+      "eval_runtime": 2.2064,
+      "eval_samples_per_second": 1041.083,
+      "eval_steps_per_second": 16.316,
+      "step": 110000
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00038495756080926107,
+      "loss": 0.3933,
+      "step": 110500
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0003830777659310416,
+      "loss": 0.3932,
+      "step": 111000
+    },
+    {
+      "epoch": 4.95,
+      "eval_loss": 0.36492955684661865,
+      "eval_runtime": 2.2207,
+      "eval_samples_per_second": 1034.354,
+      "eval_steps_per_second": 16.211,
+      "step": 111000
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0003811945556784033,
+      "loss": 0.3931,
+      "step": 111500
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00037930801242906366,
+      "loss": 0.3925,
+      "step": 112000
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 0.36507272720336914,
+      "eval_runtime": 2.3005,
+      "eval_samples_per_second": 998.488,
+      "eval_steps_per_second": 15.649,
+      "step": 112000
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00037741821870653576,
+      "loss": 0.3924,
+      "step": 112500
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00037552525717651905,
+      "loss": 0.3919,
+      "step": 113000
+    },
+    {
+      "epoch": 5.04,
+      "eval_loss": 0.3649584650993347,
+      "eval_runtime": 2.1025,
+      "eval_samples_per_second": 1092.533,
+      "eval_steps_per_second": 17.123,
+      "step": 113000
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0003736292106432834,
+      "loss": 0.3918,
+      "step": 113500
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00037173016204604647,
+      "loss": 0.3915,
+      "step": 114000
+    },
+    {
+      "epoch": 5.09,
+      "eval_loss": 0.36355069279670715,
+      "eval_runtime": 2.1224,
+      "eval_samples_per_second": 1082.26,
+      "eval_steps_per_second": 16.962,
+      "step": 114000
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0003698281944553456,
+      "loss": 0.3915,
+      "step": 114500
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0003679233910694053,
+      "loss": 0.3911,
+      "step": 115000
+    },
+    {
+      "epoch": 5.13,
+      "eval_loss": 0.36546093225479126,
+      "eval_runtime": 2.202,
+      "eval_samples_per_second": 1043.148,
+      "eval_steps_per_second": 16.349,
+      "step": 115000
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.000366015835210496,
+      "loss": 0.3911,
+      "step": 115500
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0003641056103212908,
+      "loss": 0.3905,
+      "step": 116000
+    },
+    {
+      "epoch": 5.17,
+      "eval_loss": 0.36497965455055237,
+      "eval_runtime": 2.1541,
+      "eval_samples_per_second": 1066.328,
+      "eval_steps_per_second": 16.712,
+      "step": 116000
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00036219279996121446,
+      "loss": 0.3904,
+      "step": 116500
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0003602774878027888,
+      "loss": 0.3905,
+      "step": 117000
+    },
+    {
+      "epoch": 5.22,
+      "eval_loss": 0.40538668632507324,
+      "eval_runtime": 2.1607,
+      "eval_samples_per_second": 1063.067,
+      "eval_steps_per_second": 16.661,
+      "step": 117000
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00035835975762797245,
+      "loss": 0.3917,
+      "step": 117500
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0003564396933244957,
+      "loss": 0.3894,
+      "step": 118000
+    },
+    {
+      "epoch": 5.26,
+      "eval_loss": 0.36090952157974243,
+      "eval_runtime": 2.2235,
+      "eval_samples_per_second": 1033.034,
+      "eval_steps_per_second": 16.19,
+      "step": 118000
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0003545173788821915,
+      "loss": 0.3893,
+      "step": 118500
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00035259289838932104,
+      "loss": 0.3889,
+      "step": 119000
+    },
+    {
+      "epoch": 5.31,
+      "eval_loss": 0.3599041998386383,
+      "eval_runtime": 2.1374,
+      "eval_samples_per_second": 1074.669,
+      "eval_steps_per_second": 16.843,
+      "step": 119000
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0003506663360288954,
+      "loss": 0.3889,
+      "step": 119500
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00034873777607499334,
+      "loss": 0.3888,
+      "step": 120000
+    },
+    {
+      "epoch": 5.35,
+      "eval_loss": 0.3593074679374695,
+      "eval_runtime": 2.1457,
+      "eval_samples_per_second": 1070.506,
+      "eval_steps_per_second": 16.778,
+      "step": 120000
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0003468073028890751,
+      "loss": 0.3886,
+      "step": 120500
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0003448750009162916,
+      "loss": 0.3887,
+      "step": 121000
+    },
+    {
+      "epoch": 5.4,
+      "eval_loss": 0.3600543737411499,
+      "eval_runtime": 2.1567,
+      "eval_samples_per_second": 1065.071,
+      "eval_steps_per_second": 16.692,
+      "step": 121000
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00034294095468179094,
+      "loss": 0.3881,
+      "step": 121500
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00034100524878702073,
+      "loss": 0.3883,
+      "step": 122000
+    },
+    {
+      "epoch": 5.44,
+      "eval_loss": 0.3610630929470062,
+      "eval_runtime": 2.1629,
+      "eval_samples_per_second": 1061.977,
+      "eval_steps_per_second": 16.644,
+      "step": 122000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.000339067967906028,
+      "loss": 0.4695,
+      "step": 122500
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0003371291967817539,
+      "loss": 0.6776,
+      "step": 123000
+    },
+    {
+      "epoch": 5.49,
+      "eval_loss": 0.6768860220909119,
+      "eval_runtime": 2.1959,
+      "eval_samples_per_second": 1046.056,
+      "eval_steps_per_second": 16.394,
+      "step": 123000
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0003351890202223285,
+      "loss": 0.6171,
+      "step": 123500
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0003332475230973597,
+      "loss": 0.3917,
+      "step": 124000
+    },
+    {
+      "epoch": 5.53,
+      "eval_loss": 0.36259856820106506,
+      "eval_runtime": 2.1732,
+      "eval_samples_per_second": 1056.948,
+      "eval_steps_per_second": 16.565,
+      "step": 124000
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00033130479033422134,
+      "loss": 0.3889,
+      "step": 124500
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0003293609069143381,
+      "loss": 0.3897,
+      "step": 125000
+    },
+    {
+      "epoch": 5.58,
+      "eval_loss": 0.3617185652256012,
+      "eval_runtime": 2.1669,
+      "eval_samples_per_second": 1060.024,
+      "eval_steps_per_second": 16.613,
+      "step": 125000
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00032741595786946783,
+      "loss": 0.3879,
+      "step": 125500
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.000325470028277983,
+      "loss": 0.3869,
+      "step": 126000
+    },
+    {
+      "epoch": 5.62,
+      "eval_loss": 0.35780513286590576,
+      "eval_runtime": 2.1966,
+      "eval_samples_per_second": 1045.71,
+      "eval_steps_per_second": 16.389,
+      "step": 126000
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00032352320326114754,
+      "loss": 0.3868,
+      "step": 126500
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00032157556797939436,
+      "loss": 0.3864,
+      "step": 127000
+    },
+    {
+      "epoch": 5.67,
+      "eval_loss": 0.35783183574676514,
+      "eval_runtime": 2.2195,
+      "eval_samples_per_second": 1034.94,
+      "eval_steps_per_second": 16.22,
+      "step": 127000
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00031962720762860057,
+      "loss": 0.3865,
+      "step": 127500
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0003176782074363595,
+      "loss": 0.3862,
+      "step": 128000
+    },
+    {
+      "epoch": 5.71,
+      "eval_loss": 0.35733312368392944,
+      "eval_runtime": 2.1604,
+      "eval_samples_per_second": 1063.224,
+      "eval_steps_per_second": 16.664,
+      "step": 128000
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0003157286526582535,
+      "loss": 0.3858,
+      "step": 128500
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0003137786285741241,
+      "loss": 0.3855,
+      "step": 129000
+    },
+    {
+      "epoch": 5.75,
+      "eval_loss": 0.3577572703361511,
+      "eval_runtime": 2.1724,
+      "eval_samples_per_second": 1057.343,
+      "eval_steps_per_second": 16.571,
+      "step": 129000
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0003118282204843421,
+      "loss": 0.3854,
+      "step": 129500
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0003098775137060758,
+      "loss": 0.3854,
+      "step": 130000
+    },
+    {
+      "epoch": 5.8,
+      "eval_loss": 0.357120543718338,
+      "eval_runtime": 2.158,
+      "eval_samples_per_second": 1064.4,
+      "eval_steps_per_second": 16.682,
+      "step": 130000
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00030792659356955893,
+      "loss": 0.3853,
+      "step": 130500
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0003059755454143586,
+      "loss": 0.3849,
+      "step": 131000
+    },
+    {
+      "epoch": 5.84,
+      "eval_loss": 0.3566107451915741,
+      "eval_runtime": 2.2772,
+      "eval_samples_per_second": 1008.7,
+      "eval_steps_per_second": 15.809,
+      "step": 131000
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00030402445458564144,
+      "loss": 0.3845,
+      "step": 131500
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0003020734064304411,
+      "loss": 0.3845,
+      "step": 132000
+    },
+    {
+      "epoch": 5.89,
+      "eval_loss": 0.35691574215888977,
+      "eval_runtime": 2.2557,
+      "eval_samples_per_second": 1018.294,
+      "eval_steps_per_second": 15.959,
+      "step": 132000
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00030012248629392423,
+      "loss": 0.3842,
+      "step": 132500
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00029817177951565793,
+      "loss": 0.384,
+      "step": 133000
+    },
+    {
+      "epoch": 5.93,
+      "eval_loss": 0.35665038228034973,
+      "eval_runtime": 2.2062,
+      "eval_samples_per_second": 1041.164,
+      "eval_steps_per_second": 16.318,
+      "step": 133000
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00029622137142587594,
+      "loss": 0.3837,
+      "step": 133500
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0002942713473417466,
+      "loss": 0.3921,
+      "step": 134000
+    },
+    {
+      "epoch": 5.98,
+      "eval_loss": 0.3628464341163635,
+      "eval_runtime": 2.2578,
+      "eval_samples_per_second": 1017.356,
+      "eval_steps_per_second": 15.945,
+      "step": 134000
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00029232179256364054,
+      "loss": 0.3854,
+      "step": 134500
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0002903727923713994,
+      "loss": 0.3844,
+      "step": 135000
+    },
+    {
+      "epoch": 6.02,
+      "eval_loss": 0.3565491735935211,
+      "eval_runtime": 2.2392,
+      "eval_samples_per_second": 1025.813,
+      "eval_steps_per_second": 16.077,
+      "step": 135000
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00028842443202060556,
+      "loss": 0.383,
+      "step": 135500
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00028647679673885255,
+      "loss": 0.383,
+      "step": 136000
+    },
+    {
+      "epoch": 6.07,
+      "eval_loss": 0.35466697812080383,
+      "eval_runtime": 2.2168,
+      "eval_samples_per_second": 1036.2,
+      "eval_steps_per_second": 16.24,
+      "step": 136000
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.000284529971722017,
+      "loss": 0.3829,
+      "step": 136500
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0002825840421305321,
+      "loss": 0.3828,
+      "step": 137000
+    },
+    {
+      "epoch": 6.11,
+      "eval_loss": 0.3585571050643921,
+      "eval_runtime": 2.2164,
+      "eval_samples_per_second": 1036.381,
+      "eval_steps_per_second": 16.243,
+      "step": 137000
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00028063909308566196,
+      "loss": 0.3826,
+      "step": 137500
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00027869520966577874,
+      "loss": 0.3824,
+      "step": 138000
+    },
+    {
+      "epoch": 6.16,
+      "eval_loss": 0.35530510544776917,
+      "eval_runtime": 2.2219,
+      "eval_samples_per_second": 1033.801,
+      "eval_steps_per_second": 16.202,
+      "step": 138000
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00027675247690264027,
+      "loss": 0.3823,
+      "step": 138500
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0002748109797776715,
+      "loss": 0.3825,
+      "step": 139000
+    },
+    {
+      "epoch": 6.2,
+      "eval_loss": 0.3549206256866455,
+      "eval_runtime": 2.1769,
+      "eval_samples_per_second": 1055.166,
+      "eval_steps_per_second": 16.537,
+      "step": 139000
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0002728708032182461,
+      "loss": 0.382,
+      "step": 139500
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0002709320320939721,
+      "loss": 0.3818,
+      "step": 140000
+    },
+    {
+      "epoch": 6.25,
+      "eval_loss": 0.3537048399448395,
+      "eval_runtime": 2.2266,
+      "eval_samples_per_second": 1031.628,
+      "eval_steps_per_second": 16.168,
+      "step": 140000
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00026899475121297924,
+      "loss": 0.3815,
+      "step": 140500
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00026705904531820914,
+      "loss": 0.3815,
+      "step": 141000
+    },
+    {
+      "epoch": 6.29,
+      "eval_loss": 0.3550131618976593,
+      "eval_runtime": 2.1731,
+      "eval_samples_per_second": 1057.017,
+      "eval_steps_per_second": 16.566,
+      "step": 141000
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0002651249990837085,
+      "loss": 0.3813,
+      "step": 141500
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00026319269711092485,
+      "loss": 0.3812,
+      "step": 142000
+    },
+    {
+      "epoch": 6.33,
+      "eval_loss": 0.3538789749145508,
+      "eval_runtime": 2.1483,
+      "eval_samples_per_second": 1069.198,
+      "eval_steps_per_second": 16.757,
+      "step": 142000
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0002612622239250066,
+      "loss": 0.3809,
+      "step": 142500
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0002593336639711046,
+      "loss": 0.3806,
+      "step": 143000
+    },
+    {
+      "epoch": 6.38,
+      "eval_loss": 0.35349225997924805,
+      "eval_runtime": 2.1799,
+      "eval_samples_per_second": 1053.718,
+      "eval_steps_per_second": 16.515,
+      "step": 143000
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.000257407101610679,
+      "loss": 0.3805,
+      "step": 143500
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00025548262111780846,
+      "loss": 0.3804,
+      "step": 144000
+    },
+    {
+      "epoch": 6.42,
+      "eval_loss": 0.3532736003398895,
+      "eval_runtime": 2.2355,
+      "eval_samples_per_second": 1027.515,
+      "eval_steps_per_second": 16.104,
+      "step": 144000
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0002535603066755043,
+      "loss": 0.3803,
+      "step": 144500
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00025164024237202764,
+      "loss": 0.3799,
+      "step": 145000
+    },
+    {
+      "epoch": 6.47,
+      "eval_loss": 0.3539290130138397,
+      "eval_runtime": 2.1927,
+      "eval_samples_per_second": 1047.568,
+      "eval_steps_per_second": 16.418,
+      "step": 145000
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00024972251219721115,
+      "loss": 0.38,
+      "step": 145500
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00024780720003878557,
+      "loss": 0.3799,
+      "step": 146000
+    },
+    {
+      "epoch": 6.51,
+      "eval_loss": 0.3527817726135254,
+      "eval_runtime": 2.1772,
+      "eval_samples_per_second": 1055.009,
+      "eval_steps_per_second": 16.535,
+      "step": 146000
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00024589438967870925,
+      "loss": 0.3795,
+      "step": 146500
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00024398416478950394,
+      "loss": 0.3794,
+      "step": 147000
+    },
+    {
+      "epoch": 6.56,
+      "eval_loss": 0.3519304692745209,
+      "eval_runtime": 2.1956,
+      "eval_samples_per_second": 1046.2,
+      "eval_steps_per_second": 16.397,
+      "step": 147000
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00024207660893059467,
+      "loss": 0.3793,
+      "step": 147500
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0002401718055446543,
+      "loss": 0.3792,
+      "step": 148000
+    },
+    {
+      "epoch": 6.6,
+      "eval_loss": 0.3501195013523102,
+      "eval_runtime": 2.1687,
+      "eval_samples_per_second": 1059.172,
+      "eval_steps_per_second": 16.6,
+      "step": 148000
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00023826983795395364,
+      "loss": 0.379,
+      "step": 148500
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00023637078935671656,
+      "loss": 0.3791,
+      "step": 149000
+    },
+    {
+      "epoch": 6.65,
+      "eval_loss": 0.35130566358566284,
+      "eval_runtime": 2.1713,
+      "eval_samples_per_second": 1057.877,
+      "eval_steps_per_second": 16.58,
+      "step": 149000
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00023447474282348085,
+      "loss": 0.3787,
+      "step": 149500
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00023258178129346424,
+      "loss": 0.3784,
+      "step": 150000
+    },
+    {
+      "epoch": 6.69,
+      "eval_loss": 0.35110124945640564,
+      "eval_runtime": 2.1647,
+      "eval_samples_per_second": 1061.094,
+      "eval_steps_per_second": 16.63,
+      "step": 150000
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00023069198757093631,
+      "loss": 0.379,
+      "step": 150500
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00022880544432159663,
+      "loss": 0.3833,
+      "step": 151000
+    },
+    {
+      "epoch": 6.74,
+      "eval_loss": 0.3517898917198181,
+      "eval_runtime": 2.2,
+      "eval_samples_per_second": 1044.076,
+      "eval_steps_per_second": 16.363,
+      "step": 151000
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00022692223406895848,
+      "loss": 0.3782,
+      "step": 151500
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.000225042439190739,
+      "loss": 0.3805,
+      "step": 152000
+    },
+    {
+      "epoch": 6.78,
+      "eval_loss": 0.35134270787239075,
+      "eval_runtime": 2.2005,
+      "eval_samples_per_second": 1043.87,
+      "eval_steps_per_second": 16.36,
+      "step": 152000
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00022316614191525587,
+      "loss": 0.3788,
+      "step": 152500
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00022129342431783026,
+      "loss": 0.3785,
+      "step": 153000
+    },
+    {
+      "epoch": 6.83,
+      "eval_loss": 0.35215121507644653,
+      "eval_runtime": 2.2421,
+      "eval_samples_per_second": 1024.496,
+      "eval_steps_per_second": 16.057,
+      "step": 153000
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00021942436831719677,
+      "loss": 0.3776,
+      "step": 153500
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00021755905567191967,
+      "loss": 0.3772,
+      "step": 154000
+    },
+    {
+      "epoch": 6.87,
+      "eval_loss": 0.34930697083473206,
+      "eval_runtime": 2.254,
+      "eval_samples_per_second": 1019.092,
+      "eval_steps_per_second": 15.972,
+      "step": 154000
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00021569756797681686,
+      "loss": 0.3775,
+      "step": 154500
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00021383998665939054,
+      "loss": 0.3772,
+      "step": 155000
+    },
+    {
+      "epoch": 6.91,
+      "eval_loss": 0.35032403469085693,
+      "eval_runtime": 2.2005,
+      "eval_samples_per_second": 1043.877,
+      "eval_steps_per_second": 16.36,
+      "step": 155000
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00021198639297626516,
+      "loss": 0.3769,
+      "step": 155500
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.0002101368680096334,
+      "loss": 0.3771,
+      "step": 156000
+    },
+    {
+      "epoch": 6.96,
+      "eval_loss": 0.35130995512008667,
+      "eval_runtime": 2.2636,
+      "eval_samples_per_second": 1014.77,
+      "eval_steps_per_second": 15.904,
+      "step": 156000
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00020829149266370862,
+      "loss": 0.377,
+      "step": 156500
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00020645034766118703,
+      "loss": 0.3769,
+      "step": 157000
+    },
+    {
+      "epoch": 7.0,
+      "eval_loss": 0.35048824548721313,
+      "eval_runtime": 2.2749,
+      "eval_samples_per_second": 1009.702,
+      "eval_steps_per_second": 15.825,
+      "step": 157000
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.00020461351353971526,
+      "loss": 0.3766,
+      "step": 157500
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.00020278107064836847,
+      "loss": 0.3766,
+      "step": 158000
+    },
+    {
+      "epoch": 7.05,
+      "eval_loss": 0.3498741090297699,
+      "eval_runtime": 2.2625,
+      "eval_samples_per_second": 1015.239,
+      "eval_steps_per_second": 15.911,
+      "step": 158000
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.00020095309914413485,
+      "loss": 0.3764,
+      "step": 158500
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.00019912967898840997,
+      "loss": 0.3762,
+      "step": 159000
+    },
+    {
+      "epoch": 7.09,
+      "eval_loss": 0.34895479679107666,
+      "eval_runtime": 2.1885,
+      "eval_samples_per_second": 1049.579,
+      "eval_steps_per_second": 16.45,
+      "step": 159000
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.00019731088994349834,
+      "loss": 0.3763,
+      "step": 159500
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.0001954968115691248,
+      "loss": 0.376,
+      "step": 160000
+    },
+    {
+      "epoch": 7.14,
+      "eval_loss": 0.3464561402797699,
+      "eval_runtime": 2.2955,
+      "eval_samples_per_second": 1000.637,
+      "eval_steps_per_second": 15.683,
+      "step": 160000
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.00019368752321895452,
+      "loss": 0.3758,
+      "step": 160500
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.00019188310403712105,
+      "loss": 0.3756,
+      "step": 161000
+    },
+    {
+      "epoch": 7.18,
+      "eval_loss": 0.3490326702594757,
+      "eval_runtime": 2.2155,
+      "eval_samples_per_second": 1036.763,
+      "eval_steps_per_second": 16.249,
+      "step": 161000
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.00019008363295476495,
+      "loss": 0.3756,
+      "step": 161500
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.0001882891886865808,
+      "loss": 0.3753,
+      "step": 162000
+    },
+    {
+      "epoch": 7.23,
+      "eval_loss": 0.3482523560523987,
+      "eval_runtime": 2.2246,
+      "eval_samples_per_second": 1032.556,
+      "eval_steps_per_second": 16.183,
+      "step": 162000
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.00018649984972737404,
+      "loss": 0.3757,
+      "step": 162500
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.00018471569434862749,
+      "loss": 0.3749,
+      "step": 163000
+    },
+    {
+      "epoch": 7.27,
+      "eval_loss": 0.3480832874774933,
+      "eval_runtime": 2.1706,
+      "eval_samples_per_second": 1058.217,
+      "eval_steps_per_second": 16.585,
+      "step": 163000
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.00018293680059507713,
+      "loss": 0.3753,
+      "step": 163500
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.00018116324628129882,
+      "loss": 0.3747,
+      "step": 164000
+    },
+    {
+      "epoch": 7.32,
+      "eval_loss": 0.3469808101654053,
+      "eval_runtime": 2.1769,
+      "eval_samples_per_second": 1055.189,
+      "eval_steps_per_second": 16.538,
+      "step": 164000
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.00017939510898830357,
+      "loss": 0.3746,
+      "step": 164500
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.0001776324660601446,
+      "loss": 0.375,
+      "step": 165000
+    },
+    {
+      "epoch": 7.36,
+      "eval_loss": 0.3475760519504547,
+      "eval_runtime": 2.1442,
+      "eval_samples_per_second": 1071.239,
+      "eval_steps_per_second": 16.789,
+      "step": 165000
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.00017587539460053368,
+      "loss": 0.3743,
+      "step": 165500
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.0001741239714694688,
+      "loss": 0.3742,
+      "step": 166000
+    },
+    {
+      "epoch": 7.41,
+      "eval_loss": 0.34710729122161865,
+      "eval_runtime": 2.2456,
+      "eval_samples_per_second": 1022.894,
+      "eval_steps_per_second": 16.031,
+      "step": 166000
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.0001723782732798713,
+      "loss": 0.3741,
+      "step": 166500
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.00017063837639423517,
+      "loss": 0.3741,
+      "step": 167000
+    },
+    {
+      "epoch": 7.45,
+      "eval_loss": 0.3461616337299347,
+      "eval_runtime": 2.2634,
+      "eval_samples_per_second": 1014.851,
+      "eval_steps_per_second": 15.905,
+      "step": 167000
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00016890435692128712,
+      "loss": 0.3738,
+      "step": 167500
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.0001671762907126559,
+      "loss": 0.3738,
+      "step": 168000
+    },
+    {
+      "epoch": 7.49,
+      "eval_loss": 0.3469841480255127,
+      "eval_runtime": 2.2202,
+      "eval_samples_per_second": 1034.588,
+      "eval_steps_per_second": 16.215,
+      "step": 168000
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.00016545425335955596,
+      "loss": 0.3735,
+      "step": 168500
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.00016373832018947945,
+      "loss": 0.3735,
+      "step": 169000
+    },
+    {
+      "epoch": 7.54,
+      "eval_loss": 0.3462463617324829,
+      "eval_runtime": 2.2185,
+      "eval_samples_per_second": 1035.373,
+      "eval_steps_per_second": 16.227,
+      "step": 169000
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.0001620285662629024,
+      "loss": 0.3735,
+      "step": 169500
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.0001603250663700002,
+      "loss": 0.3736,
+      "step": 170000
+    },
+    {
+      "epoch": 7.58,
+      "eval_loss": 0.3466815948486328,
+      "eval_runtime": 2.223,
+      "eval_samples_per_second": 1033.305,
+      "eval_steps_per_second": 16.195,
+      "step": 170000
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 0.00015862789502737648,
+      "loss": 0.3728,
+      "step": 170500
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.00015693712647480446,
+      "loss": 0.3731,
+      "step": 171000
+    },
+    {
+      "epoch": 7.63,
+      "eval_loss": 0.3457169234752655,
+      "eval_runtime": 2.2902,
+      "eval_samples_per_second": 1002.962,
+      "eval_steps_per_second": 15.719,
+      "step": 171000
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.00015525283467197743,
+      "loss": 0.3727,
+      "step": 171500
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 0.00015357509329527556,
+      "loss": 0.3726,
+      "step": 172000
+    },
+    {
+      "epoch": 7.67,
+      "eval_loss": 0.3477800190448761,
+      "eval_runtime": 2.185,
+      "eval_samples_per_second": 1051.256,
+      "eval_steps_per_second": 16.476,
+      "step": 172000
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.00015190397573454158,
+      "loss": 0.3727,
+      "step": 172500
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.00015023955508987127,
+      "loss": 0.3725,
+      "step": 173000
+    },
+    {
+      "epoch": 7.72,
+      "eval_loss": 0.3447197675704956,
+      "eval_runtime": 2.1651,
+      "eval_samples_per_second": 1060.935,
+      "eval_steps_per_second": 16.628,
+      "step": 173000
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 0.00014858190416841565,
+      "loss": 0.3724,
+      "step": 173500
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.00014693109548119591,
+      "loss": 0.3722,
+      "step": 174000
+    },
+    {
+      "epoch": 7.76,
+      "eval_loss": 0.3459009826183319,
+      "eval_runtime": 2.2671,
+      "eval_samples_per_second": 1013.172,
+      "eval_steps_per_second": 15.879,
+      "step": 174000
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 0.00014528720123993226,
+      "loss": 0.3721,
+      "step": 174500
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 0.0001436502933538841,
+      "loss": 0.3723,
+      "step": 175000
+    },
+    {
+      "epoch": 7.81,
+      "eval_loss": 0.3462165296077728,
+      "eval_runtime": 2.2195,
+      "eval_samples_per_second": 1034.906,
+      "eval_steps_per_second": 16.22,
+      "step": 175000
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.00014202044342670508,
+      "loss": 0.372,
+      "step": 175500
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 0.00014039772275331125,
+      "loss": 0.3718,
+      "step": 176000
+    },
+    {
+      "epoch": 7.85,
+      "eval_loss": 0.3463585674762726,
+      "eval_runtime": 2.2454,
+      "eval_samples_per_second": 1022.987,
+      "eval_steps_per_second": 16.033,
+      "step": 176000
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.00013878220231676152,
+      "loss": 0.3716,
+      "step": 176500
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.00013717395278515355,
+      "loss": 0.3716,
+      "step": 177000
+    },
+    {
+      "epoch": 7.9,
+      "eval_loss": 0.34527209401130676,
+      "eval_runtime": 2.2682,
+      "eval_samples_per_second": 1012.694,
+      "eval_steps_per_second": 15.872,
+      "step": 177000
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 0.00013557304450853162,
+      "loss": 0.3714,
+      "step": 177500
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.00013397954751581014,
+      "loss": 0.3712,
+      "step": 178000
+    },
+    {
+      "epoch": 7.94,
+      "eval_loss": 0.34656643867492676,
+      "eval_runtime": 2.2489,
+      "eval_samples_per_second": 1021.386,
+      "eval_steps_per_second": 16.008,
+      "step": 178000
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 0.00013239353151170983,
+      "loss": 0.371,
+      "step": 178500
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 0.00013081506587370853,
+      "loss": 0.3712,
+      "step": 179000
+    },
+    {
+      "epoch": 7.99,
+      "eval_loss": 0.34555310010910034,
+      "eval_runtime": 2.1617,
+      "eval_samples_per_second": 1062.607,
+      "eval_steps_per_second": 16.654,
+      "step": 179000
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 0.00012924421964900695,
+      "loss": 0.371,
+      "step": 179500
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 0.00012768106155150758,
+      "loss": 0.3709,
+      "step": 180000
+    },
+    {
+      "epoch": 8.03,
+      "eval_loss": 0.34523507952690125,
+      "eval_runtime": 2.3269,
+      "eval_samples_per_second": 987.138,
+      "eval_steps_per_second": 15.471,
+      "step": 180000
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 0.00012612565995880976,
+      "loss": 0.3725,
+      "step": 180500
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 0.00012457808290921774,
+      "loss": 0.3709,
+      "step": 181000
+    },
+    {
+      "epoch": 8.07,
+      "eval_loss": 0.3426642119884491,
+      "eval_runtime": 2.2304,
+      "eval_samples_per_second": 1029.848,
+      "eval_steps_per_second": 16.14,
+      "step": 181000
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 0.00012303839809876525,
+      "loss": 0.3705,
+      "step": 181500
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 0.00012150667287825382,
+      "loss": 0.3707,
+      "step": 182000
+    },
+    {
+      "epoch": 8.12,
+      "eval_loss": 0.3444533050060272,
+      "eval_runtime": 2.1579,
+      "eval_samples_per_second": 1064.458,
+      "eval_steps_per_second": 16.683,
+      "step": 182000
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 0.00011998297425030656,
+      "loss": 0.3705,
+      "step": 182500
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 0.00011846736886643775,
+      "loss": 0.3703,
+      "step": 183000
+    },
+    {
+      "epoch": 8.16,
+      "eval_loss": 0.34523776173591614,
+      "eval_runtime": 2.2575,
+      "eval_samples_per_second": 1017.476,
+      "eval_steps_per_second": 15.947,
+      "step": 183000
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 0.00011695992302413651,
+      "loss": 0.3702,
+      "step": 183500
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 0.00011546070266396771,
+      "loss": 0.3701,
+      "step": 184000
+    },
+    {
+      "epoch": 8.21,
+      "eval_loss": 0.34200483560562134,
+      "eval_runtime": 2.2775,
+      "eval_samples_per_second": 1008.549,
+      "eval_steps_per_second": 15.807,
+      "step": 184000
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 0.00011396977336668645,
+      "loss": 0.3701,
+      "step": 184500
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 0.00011248720035037021,
+      "loss": 0.3699,
+      "step": 185000
+    },
+    {
+      "epoch": 8.25,
+      "eval_loss": 0.34288346767425537,
+      "eval_runtime": 2.2113,
+      "eval_samples_per_second": 1038.766,
+      "eval_steps_per_second": 16.28,
+      "step": 185000
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 0.00011101304846756577,
+      "loss": 0.3699,
+      "step": 185500
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 0.00010954738220245183,
+      "loss": 0.3697,
+      "step": 186000
+    },
+    {
+      "epoch": 8.3,
+      "eval_loss": 0.3431689739227295,
+      "eval_runtime": 2.222,
+      "eval_samples_per_second": 1033.767,
+      "eval_steps_per_second": 16.202,
+      "step": 186000
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 0.00010809026566801912,
+      "loss": 0.3695,
+      "step": 186500
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 0.00010664176260326507,
+      "loss": 0.3696,
+      "step": 187000
+    },
+    {
+      "epoch": 8.34,
+      "eval_loss": 0.3424961268901825,
+      "eval_runtime": 2.2738,
+      "eval_samples_per_second": 1010.209,
+      "eval_steps_per_second": 15.833,
+      "step": 187000
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 0.00010520193637040641,
+      "loss": 0.3697,
+      "step": 187500
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 0.00010377084995210682,
+      "loss": 0.3696,
+      "step": 188000
+    },
+    {
+      "epoch": 8.39,
+      "eval_loss": 0.34366557002067566,
+      "eval_runtime": 2.2531,
+      "eval_samples_per_second": 1019.467,
+      "eval_steps_per_second": 15.978,
+      "step": 188000
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 0.00010234856594872234,
+      "loss": 0.3692,
+      "step": 188500
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 0.00010093514657556295,
+      "loss": 0.3694,
+      "step": 189000
+    },
+    {
+      "epoch": 8.43,
+      "eval_loss": 0.34253063797950745,
+      "eval_runtime": 2.2641,
+      "eval_samples_per_second": 1014.519,
+      "eval_steps_per_second": 15.9,
+      "step": 189000
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 9.953065366017073e-05,
+      "loss": 0.369,
+      "step": 189500
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 9.813514863961586e-05,
+      "loss": 0.369,
+      "step": 190000
+    },
+    {
+      "epoch": 8.48,
+      "eval_loss": 0.34287169575691223,
+      "eval_runtime": 2.2268,
+      "eval_samples_per_second": 1031.502,
+      "eval_steps_per_second": 16.166,
+      "step": 190000
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 9.67486925578087e-05,
+      "loss": 0.3686,
+      "step": 190500
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 9.537134606282964e-05,
+      "loss": 0.369,
+      "step": 191000
+    },
+    {
+      "epoch": 8.52,
+      "eval_loss": 0.3414945900440216,
+      "eval_runtime": 2.186,
+      "eval_samples_per_second": 1050.801,
+      "eval_steps_per_second": 16.469,
+      "step": 191000
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 9.400316940427652e-05,
+      "loss": 0.3688,
+      "step": 191500
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 9.264422243062844e-05,
+      "loss": 0.3685,
+      "step": 192000
+    },
+    {
+      "epoch": 8.57,
+      "eval_loss": 0.34313786029815674,
+      "eval_runtime": 2.2429,
+      "eval_samples_per_second": 1024.133,
+      "eval_steps_per_second": 16.051,
+      "step": 192000
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 9.129456458662876e-05,
+      "loss": 0.3684,
+      "step": 192500
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 8.995425491068365e-05,
+      "loss": 0.3684,
+      "step": 193000
+    },
+    {
+      "epoch": 8.61,
+      "eval_loss": 0.341512531042099,
+      "eval_runtime": 2.1692,
+      "eval_samples_per_second": 1058.928,
+      "eval_steps_per_second": 16.596,
+      "step": 193000
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 8.862335203228025e-05,
+      "loss": 0.3683,
+      "step": 193500
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 8.73019141694222e-05,
+      "loss": 0.3683,
+      "step": 194000
+    },
+    {
+      "epoch": 8.65,
+      "eval_loss": 0.3420671820640564,
+      "eval_runtime": 2.2415,
+      "eval_samples_per_second": 1024.765,
+      "eval_steps_per_second": 16.061,
+      "step": 194000
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 8.598999912608229e-05,
+      "loss": 0.3681,
+      "step": 194500
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 8.468766428967468e-05,
+      "loss": 0.368,
+      "step": 195000
+    },
+    {
+      "epoch": 8.7,
+      "eval_loss": 0.3422170877456665,
+      "eval_runtime": 2.2216,
+      "eval_samples_per_second": 1033.959,
+      "eval_steps_per_second": 16.205,
+      "step": 195000
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 8.339496662854397e-05,
+      "loss": 0.3682,
+      "step": 195500
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 8.211196268947367e-05,
+      "loss": 0.3719,
+      "step": 196000
+    },
+    {
+      "epoch": 8.74,
+      "eval_loss": 0.3433374762535095,
+      "eval_runtime": 2.2305,
+      "eval_samples_per_second": 1029.81,
+      "eval_steps_per_second": 16.14,
+      "step": 196000
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 8.083870859521251e-05,
+      "loss": 0.368,
+      "step": 196500
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 7.95752600420192e-05,
+      "loss": 0.3678,
+      "step": 197000
+    },
+    {
+      "epoch": 8.79,
+      "eval_loss": 0.33997857570648193,
+      "eval_runtime": 2.2361,
+      "eval_samples_per_second": 1027.215,
+      "eval_steps_per_second": 16.099,
+      "step": 197000
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 7.832167229722666e-05,
+      "loss": 0.3677,
+      "step": 197500
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 7.707800019682362e-05,
+      "loss": 0.3675,
+      "step": 198000
+    },
+    {
+      "epoch": 8.83,
+      "eval_loss": 0.3419722616672516,
+      "eval_runtime": 2.2466,
+      "eval_samples_per_second": 1022.412,
+      "eval_steps_per_second": 16.024,
+      "step": 198000
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 7.5844298143057e-05,
+      "loss": 0.3676,
+      "step": 198500
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 7.462062010205106e-05,
+      "loss": 0.3676,
+      "step": 199000
+    },
+    {
+      "epoch": 8.88,
+      "eval_loss": 0.3425526022911072,
+      "eval_runtime": 2.2092,
+      "eval_samples_per_second": 1039.756,
+      "eval_steps_per_second": 16.296,
+      "step": 199000
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 7.340701960144751e-05,
+      "loss": 0.3676,
+      "step": 199500
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 7.220354972806392e-05,
+      "loss": 0.3674,
+      "step": 200000
+    },
+    {
+      "epoch": 8.92,
+      "eval_loss": 0.33957439661026,
+      "eval_runtime": 2.28,
+      "eval_samples_per_second": 1007.437,
+      "eval_steps_per_second": 15.789,
+      "step": 200000
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 7.101026312557103e-05,
+      "loss": 0.3672,
+      "step": 200500
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 6.982721199219075e-05,
+      "loss": 0.3673,
+      "step": 201000
+    },
+    {
+      "epoch": 8.97,
+      "eval_loss": 0.3403836786746979,
+      "eval_runtime": 2.153,
+      "eval_samples_per_second": 1066.9,
+      "eval_steps_per_second": 16.721,
+      "step": 201000
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 6.865444807841203e-05,
+      "loss": 0.3671,
+      "step": 201500
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 6.749202268472787e-05,
+      "loss": 0.3671,
+      "step": 202000
+    },
+    {
+      "epoch": 9.01,
+      "eval_loss": 0.3397007882595062,
+      "eval_runtime": 2.2258,
+      "eval_samples_per_second": 1032.002,
+      "eval_steps_per_second": 16.174,
+      "step": 202000
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 6.633998665939053e-05,
+      "loss": 0.3672,
+      "step": 202500
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 6.519839039618793e-05,
+      "loss": 0.3669,
+      "step": 203000
+    },
+    {
+      "epoch": 9.06,
+      "eval_loss": 0.3417140245437622,
+      "eval_runtime": 2.2137,
+      "eval_samples_per_second": 1037.631,
+      "eval_steps_per_second": 16.262,
+      "step": 203000
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 6.406728383223897e-05,
+      "loss": 0.3669,
+      "step": 203500
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 6.294671644580888e-05,
+      "loss": 0.3669,
+      "step": 204000
+    },
+    {
+      "epoch": 9.1,
+      "eval_loss": 0.3413463830947876,
+      "eval_runtime": 2.179,
+      "eval_samples_per_second": 1054.147,
+      "eval_steps_per_second": 16.521,
+      "step": 204000
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 6.18367372541454e-05,
+      "loss": 0.3668,
+      "step": 204500
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 6.073739481133408e-05,
+      "loss": 0.3666,
+      "step": 205000
+    },
+    {
+      "epoch": 9.15,
+      "eval_loss": 0.33860939741134644,
+      "eval_runtime": 2.2027,
+      "eval_samples_per_second": 1042.794,
+      "eval_steps_per_second": 16.343,
+      "step": 205000
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.96487372061749e-05,
+      "loss": 0.3667,
+      "step": 205500
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.857081206007811e-05,
+      "loss": 0.3666,
+      "step": 206000
+    },
+    {
+      "epoch": 9.19,
+      "eval_loss": 0.3413560390472412,
+      "eval_runtime": 2.2133,
+      "eval_samples_per_second": 1037.836,
+      "eval_steps_per_second": 16.266,
+      "step": 206000
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.750366652498169e-05,
+      "loss": 0.3663,
+      "step": 206500
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.6447347281288474e-05,
+      "loss": 0.3664,
+      "step": 207000
+    },
+    {
+      "epoch": 9.23,
+      "eval_loss": 0.3407277464866638,
+      "eval_runtime": 2.2449,
+      "eval_samples_per_second": 1023.218,
+      "eval_steps_per_second": 16.037,
+      "step": 207000
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.540190053582401e-05,
+      "loss": 0.3665,
+      "step": 207500
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.43673720198159e-05,
+      "loss": 0.3662,
+      "step": 208000
+    },
+    {
+      "epoch": 9.28,
+      "eval_loss": 0.3401471674442291,
+      "eval_runtime": 2.2056,
+      "eval_samples_per_second": 1041.45,
+      "eval_steps_per_second": 16.322,
+      "step": 208000
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.33438069868928e-05,
+      "loss": 0.3662,
+      "step": 208500
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.2331250211105e-05,
+      "loss": 0.3661,
+      "step": 209000
+    },
+    {
+      "epoch": 9.32,
+      "eval_loss": 0.34120145440101624,
+      "eval_runtime": 2.1294,
+      "eval_samples_per_second": 1078.699,
+      "eval_steps_per_second": 16.906,
+      "step": 209000
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.13297459849662e-05,
+      "loss": 0.366,
+      "step": 209500
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.0339338117515696e-05,
+      "loss": 0.366,
+      "step": 210000
+    },
+    {
+      "epoch": 9.37,
+      "eval_loss": 0.3374158442020416,
+      "eval_runtime": 2.1871,
+      "eval_samples_per_second": 1050.236,
+      "eval_steps_per_second": 16.46,
+      "step": 210000
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.9360069932402115e-05,
+      "loss": 0.3659,
+      "step": 210500
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.839198426598824e-05,
+      "loss": 0.3659,
+      "step": 211000
+    },
+    {
+      "epoch": 9.41,
+      "eval_loss": 0.3399938941001892,
+      "eval_runtime": 2.2613,
+      "eval_samples_per_second": 1015.803,
+      "eval_steps_per_second": 15.92,
+      "step": 211000
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.7435123465477156e-05,
+      "loss": 0.3659,
+      "step": 211500
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.6489529387060306e-05,
+      "loss": 0.3658,
+      "step": 212000
+    },
+    {
+      "epoch": 9.46,
+      "eval_loss": 0.34058016538619995,
+      "eval_runtime": 2.2213,
+      "eval_samples_per_second": 1034.088,
+      "eval_steps_per_second": 16.207,
+      "step": 212000
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 4.555524339408575e-05,
+      "loss": 0.3658,
+      "step": 212500
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.46323063552496e-05,
+      "loss": 0.3658,
+      "step": 213000
+    },
+    {
+      "epoch": 9.5,
+      "eval_loss": 0.33832791447639465,
+      "eval_runtime": 2.1746,
+      "eval_samples_per_second": 1056.275,
+      "eval_steps_per_second": 16.555,
+      "step": 213000
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.3720758642807817e-05,
+      "loss": 0.3655,
+      "step": 213500
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.2820640130810446e-05,
+      "loss": 0.3656,
+      "step": 214000
+    },
+    {
+      "epoch": 9.55,
+      "eval_loss": 0.33994585275650024,
+      "eval_runtime": 2.2047,
+      "eval_samples_per_second": 1041.878,
+      "eval_steps_per_second": 16.329,
+      "step": 214000
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.193199019335709e-05,
+      "loss": 0.3654,
+      "step": 214500
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.105484770287477e-05,
+      "loss": 0.3655,
+      "step": 215000
+    },
+    {
+      "epoch": 9.59,
+      "eval_loss": 0.3385065793991089,
+      "eval_runtime": 2.1784,
+      "eval_samples_per_second": 1054.433,
+      "eval_steps_per_second": 16.526,
+      "step": 215000
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 4.018925102841773e-05,
+      "loss": 0.3652,
+      "step": 215500
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.9335238033988484e-05,
+      "loss": 0.3653,
+      "step": 216000
+    },
+    {
+      "epoch": 9.64,
+      "eval_loss": 0.3406268358230591,
+      "eval_runtime": 2.2102,
+      "eval_samples_per_second": 1039.279,
+      "eval_steps_per_second": 16.288,
+      "step": 216000
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.8492846076882115e-05,
+      "loss": 0.3652,
+      "step": 216500
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.766211200605186e-05,
+      "loss": 0.3652,
+      "step": 217000
+    },
+    {
+      "epoch": 9.68,
+      "eval_loss": 0.33878666162490845,
+      "eval_runtime": 2.1865,
+      "eval_samples_per_second": 1050.535,
+      "eval_steps_per_second": 16.465,
+      "step": 217000
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.684307216049706e-05,
+      "loss": 0.3653,
+      "step": 217500
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.6035762367673984e-05,
+      "loss": 0.3674,
+      "step": 218000
+    },
+    {
+      "epoch": 9.73,
+      "eval_loss": 0.33809101581573486,
+      "eval_runtime": 2.1832,
+      "eval_samples_per_second": 1052.106,
+      "eval_steps_per_second": 16.489,
+      "step": 218000
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.52402179419282e-05,
+      "loss": 0.3649,
+      "step": 218500
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.4456473682950194e-05,
+      "loss": 0.365,
+      "step": 219000
+    },
+    {
+      "epoch": 9.77,
+      "eval_loss": 0.33866041898727417,
+      "eval_runtime": 2.2773,
+      "eval_samples_per_second": 1008.633,
+      "eval_steps_per_second": 15.808,
+      "step": 219000
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 3.3684563874252695e-05,
+      "loss": 0.365,
+      "step": 219500
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.2924522281671496e-05,
+      "loss": 0.3648,
+      "step": 220000
+    },
+    {
+      "epoch": 9.81,
+      "eval_loss": 0.33735984563827515,
+      "eval_runtime": 2.208,
+      "eval_samples_per_second": 1040.301,
+      "eval_steps_per_second": 16.304,
+      "step": 220000
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 3.2176382151888054e-05,
+      "loss": 0.365,
+      "step": 220500
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 3.1440176210975204e-05,
+      "loss": 0.3649,
+      "step": 221000
+    },
+    {
+      "epoch": 9.86,
+      "eval_loss": 0.3377821445465088,
+      "eval_runtime": 2.2679,
+      "eval_samples_per_second": 1012.829,
+      "eval_steps_per_second": 15.874,
+      "step": 221000
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.071593666296585e-05,
+      "loss": 0.3648,
+      "step": 221500
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.000369518844396e-05,
+      "loss": 0.3649,
+      "step": 222000
+    },
+    {
+      "epoch": 9.9,
+      "eval_loss": 0.33791235089302063,
+      "eval_runtime": 2.2027,
+      "eval_samples_per_second": 1042.832,
+      "eval_steps_per_second": 16.344,
+      "step": 222000
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.9303482943159077e-05,
+      "loss": 0.3648,
+      "step": 222500
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.861533055666306e-05,
+      "loss": 0.3646,
+      "step": 223000
+    },
+    {
+      "epoch": 9.95,
+      "eval_loss": 0.33819380402565,
+      "eval_runtime": 2.1945,
+      "eval_samples_per_second": 1046.691,
+      "eval_steps_per_second": 16.404,
+      "step": 223000
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 2.793926813097066e-05,
+      "loss": 0.3645,
+      "step": 223500
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 2.7275325239242546e-05,
+      "loss": 0.3647,
+      "step": 224000
+    },
+    {
+      "epoch": 9.99,
+      "eval_loss": 0.33774814009666443,
+      "eval_runtime": 2.2427,
+      "eval_samples_per_second": 1024.194,
+      "eval_steps_per_second": 16.052,
+      "step": 224000
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 2.6623530924491626e-05,
+      "loss": 0.3645,
+      "step": 224500
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 2.5983913698312782e-05,
+      "loss": 0.3644,
+      "step": 225000
+    },
+    {
+      "epoch": 10.04,
+      "eval_loss": 0.33505749702453613,
+      "eval_runtime": 2.2238,
+      "eval_samples_per_second": 1032.916,
+      "eval_steps_per_second": 16.188,
+      "step": 225000
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 2.5356501539635512e-05,
+      "loss": 0.3644,
+      "step": 225500
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 2.4741321893500244e-05,
+      "loss": 0.3644,
+      "step": 226000
+    },
+    {
+      "epoch": 10.08,
+      "eval_loss": 0.337401807308197,
+      "eval_runtime": 2.2527,
+      "eval_samples_per_second": 1019.654,
+      "eval_steps_per_second": 15.981,
+      "step": 226000
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 2.4138401669857587e-05,
+      "loss": 0.3644,
+      "step": 226500
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 2.3547767242391212e-05,
+      "loss": 0.3644,
+      "step": 227000
+    },
+    {
+      "epoch": 10.13,
+      "eval_loss": 0.33791208267211914,
+      "eval_runtime": 2.1965,
+      "eval_samples_per_second": 1045.741,
+      "eval_steps_per_second": 16.389,
+      "step": 227000
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 2.2969444447364498e-05,
+      "loss": 0.3642,
+      "step": 227500
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 2.240345858248992e-05,
+      "loss": 0.3651,
+      "step": 228000
+    },
+    {
+      "epoch": 10.17,
+      "eval_loss": 0.33648931980133057,
+      "eval_runtime": 2.2235,
+      "eval_samples_per_second": 1033.033,
+      "eval_steps_per_second": 16.19,
+      "step": 228000
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 2.184983440582284e-05,
+      "loss": 0.3644,
+      "step": 228500
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 2.1308596134678134e-05,
+      "loss": 0.3643,
+      "step": 229000
+    },
+    {
+      "epoch": 10.22,
+      "eval_loss": 0.3360297977924347,
+      "eval_runtime": 2.2193,
+      "eval_samples_per_second": 1035.0,
+      "eval_steps_per_second": 16.221,
+      "step": 229000
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 2.0779767444571236e-05,
+      "loss": 0.3643,
+      "step": 229500
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 2.0263371468182175e-05,
+      "loss": 0.3642,
+      "step": 230000
+    },
+    {
+      "epoch": 10.26,
+      "eval_loss": 0.33706384897232056,
+      "eval_runtime": 2.2761,
+      "eval_samples_per_second": 1009.19,
+      "eval_steps_per_second": 15.817,
+      "step": 230000
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.975943079434381e-05,
+      "loss": 0.3642,
+      "step": 230500
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.9267967467053834e-05,
+      "loss": 0.364,
+      "step": 231000
+    },
+    {
+      "epoch": 10.31,
+      "eval_loss": 0.33804792165756226,
+      "eval_runtime": 2.1938,
+      "eval_samples_per_second": 1047.056,
+      "eval_steps_per_second": 16.41,
+      "step": 231000
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.878900298451024e-05,
+      "loss": 0.3639,
+      "step": 231500
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.83225582981712e-05,
+      "loss": 0.364,
+      "step": 232000
+    },
+    {
+      "epoch": 10.35,
+      "eval_loss": 0.3375454843044281,
+      "eval_runtime": 2.2078,
+      "eval_samples_per_second": 1040.381,
+      "eval_steps_per_second": 16.305,
+      "step": 232000
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.7868653811838307e-05,
+      "loss": 0.3638,
+      "step": 232500
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.7427309380764227e-05,
+      "loss": 0.364,
+      "step": 233000
+    },
+    {
+      "epoch": 10.39,
+      "eval_loss": 0.33860087394714355,
+      "eval_runtime": 2.2,
+      "eval_samples_per_second": 1044.098,
+      "eval_steps_per_second": 16.364,
+      "step": 233000
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.6998544310784175e-05,
+      "loss": 0.3638,
+      "step": 233500
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.6582377357471285e-05,
+      "loss": 0.3639,
+      "step": 234000
+    },
+    {
+      "epoch": 10.44,
+      "eval_loss": 0.33734622597694397,
+      "eval_runtime": 2.1793,
+      "eval_samples_per_second": 1054.021,
+      "eval_steps_per_second": 16.519,
+      "step": 234000
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.617882672531633e-05,
+      "loss": 0.3639,
+      "step": 234500
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.578791006693124e-05,
+      "loss": 0.364,
+      "step": 235000
+    },
+    {
+      "epoch": 10.48,
+      "eval_loss": 0.3377144932746887,
+      "eval_runtime": 2.2337,
+      "eval_samples_per_second": 1028.33,
+      "eval_steps_per_second": 16.117,
+      "step": 235000
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.5409644482277073e-05,
+      "loss": 0.3638,
+      "step": 235500
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.504404651791591e-05,
+      "loss": 0.3636,
+      "step": 236000
+    },
+    {
+      "epoch": 10.53,
+      "eval_loss": 0.3383854627609253,
+      "eval_runtime": 2.1919,
+      "eval_samples_per_second": 1047.95,
+      "eval_steps_per_second": 16.424,
+      "step": 236000
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.4691132166287069e-05,
+      "loss": 0.3638,
+      "step": 236500
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.4350916865007609e-05,
+      "loss": 0.3636,
+      "step": 237000
+    },
+    {
+      "epoch": 10.57,
+      "eval_loss": 0.33673885464668274,
+      "eval_runtime": 2.1228,
+      "eval_samples_per_second": 1082.067,
+      "eval_steps_per_second": 16.959,
+      "step": 237000
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.4023415496196915e-05,
+      "loss": 0.3635,
+      "step": 237500
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.3708642385825806e-05,
+      "loss": 0.3638,
+      "step": 238000
+    },
+    {
+      "epoch": 10.62,
+      "eval_loss": 0.3373904228210449,
+      "eval_runtime": 2.309,
+      "eval_samples_per_second": 994.807,
+      "eval_steps_per_second": 15.591,
+      "step": 238000
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.3406611303089918e-05,
+      "loss": 0.3639,
+      "step": 238500
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.311733545980718e-05,
+      "loss": 0.3637,
+      "step": 239000
+    },
+    {
+      "epoch": 10.66,
+      "eval_loss": 0.3368372619152069,
+      "eval_runtime": 2.2218,
+      "eval_samples_per_second": 1033.83,
+      "eval_steps_per_second": 16.203,
+      "step": 239000
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.284082750984018e-05,
+      "loss": 0.3635,
+      "step": 239500
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.2577099548542319e-05,
+      "loss": 0.3635,
+      "step": 240000
+    },
+    {
+      "epoch": 10.71,
+      "eval_loss": 0.33518004417419434,
+      "eval_runtime": 2.1947,
+      "eval_samples_per_second": 1046.621,
+      "eval_steps_per_second": 16.403,
+      "step": 240000
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.232616311222906e-05,
+      "loss": 0.3635,
+      "step": 240500
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.2088029177672978e-05,
+      "loss": 0.3635,
+      "step": 241000
+    },
+    {
+      "epoch": 10.75,
+      "eval_loss": 0.3392580449581146,
+      "eval_runtime": 2.2637,
+      "eval_samples_per_second": 1014.699,
+      "eval_steps_per_second": 15.903,
+      "step": 241000
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.1862708161623889e-05,
+      "loss": 0.3634,
+      "step": 241500
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.165020992035296e-05,
+      "loss": 0.3634,
+      "step": 242000
+    },
+    {
+      "epoch": 10.8,
+      "eval_loss": 0.3343862295150757,
+      "eval_runtime": 2.2213,
+      "eval_samples_per_second": 1034.066,
+      "eval_steps_per_second": 16.207,
+      "step": 242000
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.1450543749221632e-05,
+      "loss": 0.3635,
+      "step": 242500
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.126371838227509e-05,
+      "loss": 0.3635,
+      "step": 243000
+    },
+    {
+      "epoch": 10.84,
+      "eval_loss": 0.3382641077041626,
+      "eval_runtime": 2.2515,
+      "eval_samples_per_second": 1020.218,
+      "eval_steps_per_second": 15.989,
+      "step": 243000
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.1089741991860081e-05,
+      "loss": 0.3634,
+      "step": 243500
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.0928622188267536e-05,
+      "loss": 0.3633,
+      "step": 244000
+    },
+    {
+      "epoch": 10.89,
+      "eval_loss": 0.33619487285614014,
+      "eval_runtime": 2.2703,
+      "eval_samples_per_second": 1011.748,
+      "eval_steps_per_second": 15.857,
+      "step": 244000
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.0780366019399665e-05,
+      "loss": 0.3635,
+      "step": 244500
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.0644979970461512e-05,
+      "loss": 0.3635,
+      "step": 245000
+    },
+    {
+      "epoch": 10.93,
+      "eval_loss": 0.3352552354335785,
+      "eval_runtime": 2.2392,
+      "eval_samples_per_second": 1025.803,
+      "eval_steps_per_second": 16.077,
+      "step": 245000
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.0522469963677483e-05,
+      "loss": 0.3634,
+      "step": 245500
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.0412841358032126e-05,
+      "loss": 0.3634,
+      "step": 246000
+    },
+    {
+      "epoch": 10.97,
+      "eval_loss": 0.33572423458099365,
+      "eval_runtime": 2.2818,
+      "eval_samples_per_second": 1006.659,
+      "eval_steps_per_second": 15.777,
+      "step": 246000
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.0316098949035847e-05,
+      "loss": 0.3634,
+      "step": 246500
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.0232246968514984e-05,
+      "loss": 0.3632,
+      "step": 247000
+    },
+    {
+      "epoch": 11.02,
+      "eval_loss": 0.3374698758125305,
+      "eval_runtime": 3.6783,
+      "eval_samples_per_second": 624.479,
+      "eval_steps_per_second": 9.787,
+      "step": 247000
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.0161289084426815e-05,
+      "loss": 0.3633,
+      "step": 247500
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.0103228400699063e-05,
+      "loss": 0.3633,
+      "step": 248000
+    },
+    {
+      "epoch": 11.06,
+      "eval_loss": 0.3395208418369293,
+      "eval_runtime": 2.2423,
+      "eval_samples_per_second": 1024.379,
+      "eval_steps_per_second": 16.055,
+      "step": 248000
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.0058067457094136e-05,
+      "loss": 0.3635,
+      "step": 248500
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.0025808229097982e-05,
+      "loss": 0.3635,
+      "step": 249000
+    },
+    {
+      "epoch": 11.11,
+      "eval_loss": 0.33821383118629456,
+      "eval_runtime": 2.2205,
+      "eval_samples_per_second": 1034.472,
+      "eval_steps_per_second": 16.213,
+      "step": 249000
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.0006452127833747e-05,
+      "loss": 0.3633,
+      "step": 249500
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1e-05,
+      "loss": 0.3634,
+      "step": 250000
+    },
+    {
+      "epoch": 11.15,
+      "eval_loss": 0.3380183279514313,
+      "eval_runtime": 2.1982,
+      "eval_samples_per_second": 1044.959,
+      "eval_steps_per_second": 16.377,
+      "step": 250000
+    },
+    {
+      "epoch": 11.15,
+      "step": 250000,
+      "total_flos": 4.004121619841193e+21,
+      "train_loss": 0.45158287280273435,
+      "train_runtime": 76711.7807,
+      "train_samples_per_second": 3337.167,
+      "train_steps_per_second": 3.259
+    }
+  ],
+  "max_steps": 250000,
+  "num_train_epochs": 12,
+  "total_flos": 4.004121619841193e+21,
+  "trial_name": null,
+  "trial_params": null
+}